diff options
| author | keithw <keithw> | 2002-08-26 22:16:00 +0000 |
|---|---|---|
| committer | keithw <keithw> | 2002-08-26 22:16:00 +0000 |
| commit | aab518de01935ef2757eb57067ea2cf999713ceb (patch) | |
| tree | 279b7b62a3b13b0e564c1bf7ce15d1243cb3da15 | |
| parent | 5e93eabda6a447e62c14d827ad370b84bcb89f52 (diff) | |
merged r200-0-1-branchr200-0-1-20020822-merge
73 files changed, 20030 insertions, 1283 deletions
diff --git a/xc/extras/Mesa/src/swrast/s_context.c b/xc/extras/Mesa/src/swrast/s_context.c index 094725f40..4cdb4f87f 100644 --- a/xc/extras/Mesa/src/swrast/s_context.c +++ b/xc/extras/Mesa/src/swrast/s_context.c @@ -526,6 +526,24 @@ _swrast_GetDeviceDriverReference( GLcontext *ctx ) return &swrast->Driver; } + +void +_swrast_render_start( GLcontext *ctx ) +{ + SWcontext *swrast = SWRAST_CONTEXT(ctx); + if (swrast->Driver.SpanRenderStart) + swrast->Driver.SpanRenderStart( ctx ); +} + +void +_swrast_render_finish( GLcontext *ctx ) +{ + SWcontext *swrast = SWRAST_CONTEXT(ctx); + if (swrast->Driver.SpanRenderFinish) + swrast->Driver.SpanRenderFinish( ctx ); +} + + #define SWRAST_DEBUG_VERTICES 0 void @@ -560,3 +578,5 @@ _swrast_print_vertex( GLcontext *ctx, const SWvertex *v ) fprintf(stderr, "\n"); } } + + diff --git a/xc/extras/Mesa/src/swrast/swrast.h b/xc/extras/Mesa/src/swrast/swrast.h index 6799bc309..830c1baee 100644 --- a/xc/extras/Mesa/src/swrast/swrast.h +++ b/xc/extras/Mesa/src/swrast/swrast.h @@ -152,6 +152,11 @@ _swrast_Quad( GLcontext *ctx, extern void _swrast_flush( GLcontext *ctx ); +extern void +_swrast_render_start( GLcontext *ctx ); + +extern void +_swrast_render_finish( GLcontext *ctx ); /* Tell the software rasterizer about core state changes. */ diff --git a/xc/extras/Mesa/src/swrast_setup/ss_context.c b/xc/extras/Mesa/src/swrast_setup/ss_context.c index 557d48f00..58d85faa1 100644 --- a/xc/extras/Mesa/src/swrast_setup/ss_context.c +++ b/xc/extras/Mesa/src/swrast_setup/ss_context.c @@ -48,7 +48,6 @@ #define _SWSETUP_NEW_RENDERINDEX (_NEW_POLYGON|_NEW_LIGHT) - GLboolean _swsetup_CreateContext( GLcontext *ctx ) { @@ -116,19 +115,13 @@ _swsetup_RenderStart( GLcontext *ctx ) swsetup->NewState = 0; - if (swsetup->Driver.Start) - swsetup->Driver.Start( ctx ); + _swrast_render_start( ctx ); } static void _swsetup_RenderFinish( GLcontext *ctx ) { - SScontext *swsetup = SWSETUP_CONTEXT(ctx); - - _swrast_flush( ctx ); - - if (swsetup->Driver.Finish) - swsetup->Driver.Finish( ctx ); + _swrast_render_finish( ctx ); } void diff --git a/xc/extras/Mesa/src/swrast_setup/ss_context.h b/xc/extras/Mesa/src/swrast_setup/ss_context.h index 973edd024..3174c95de 100644 --- a/xc/extras/Mesa/src/swrast_setup/ss_context.h +++ b/xc/extras/Mesa/src/swrast_setup/ss_context.h @@ -43,12 +43,6 @@ typedef struct { */ struct gl_client_array ChanColor; struct gl_client_array ChanSecondaryColor; - - - struct { - void (*Start)( GLcontext * ); - void (*Finish)( GLcontext * ); - } Driver; } SScontext; #define SWSETUP_CONTEXT(ctx) ((SScontext *)ctx->swsetup_context) diff --git a/xc/extras/Mesa/src/texstate.c b/xc/extras/Mesa/src/texstate.c index db4031fcc..89db8520c 100644 --- a/xc/extras/Mesa/src/texstate.c +++ b/xc/extras/Mesa/src/texstate.c @@ -1362,6 +1362,9 @@ _mesa_GetTexLevelParameteriv( GLenum target, GLint level, *params = 0; else if (img->TexFormat->IntensityBits > 0) *params = img->TexFormat->IntensityBits; + else if (img->TexFormat->LuminanceBits > 0) + /* intensity probably stored as luminance-alpha texture */ + *params = MIN2(img->TexFormat->LuminanceBits, img->TexFormat->AlphaBits); else /* intensity probably stored as rgb texture */ *params = MIN2(img->TexFormat->RedBits, img->TexFormat->GreenBits); return; diff --git a/xc/extras/Mesa/src/tnl_dd/t_dd_dmatmp2.h b/xc/extras/Mesa/src/tnl_dd/t_dd_dmatmp2.h index e37631e69..1d2614fbc 100644 --- a/xc/extras/Mesa/src/tnl_dd/t_dd_dmatmp2.h +++ b/xc/extras/Mesa/src/tnl_dd/t_dd_dmatmp2.h @@ -112,9 +112,11 @@ static void TAG(render_points_verts)( GLcontext *ctx, GLuint count, GLuint flags ) { - LOCAL_VARS; - if (0) fprintf(stderr, "%s\n", __FUNCTION__); - EMIT_PRIM( ctx, GL_POINTS, HW_POINTS, start, count ); + if (start < count) { + LOCAL_VARS; + if (0) fprintf(stderr, "%s\n", __FUNCTION__); + EMIT_PRIM( ctx, GL_POINTS, HW_POINTS, start, count ); + } } static void TAG(render_lines_verts)( GLcontext *ctx, @@ -156,7 +158,7 @@ static void TAG(render_line_strip_verts)( GLcontext *ctx, RESET_STIPPLE(); - if (PREFER_DISCREET_ELT_PRIM( count-start, HW_LINES )) + if (PREFER_DISCRETE_ELT_PRIM( count-start, HW_LINES )) { int dmasz = GET_SUBSEQUENT_VB_MAX_ELTS(); int currentsz; @@ -220,7 +222,7 @@ static void TAG(render_line_loop_verts)( GLcontext *ctx, if (start+1 >= count) return; - if (PREFER_DISCREET_ELT_PRIM( count-start, HW_LINES )) { + if (PREFER_DISCRETE_ELT_PRIM( count-start, HW_LINES )) { int dmasz = GET_SUBSEQUENT_VB_MAX_ELTS(); int currentsz; @@ -334,7 +336,7 @@ static void TAG(render_triangles_verts)( GLcontext *ctx, return; } - /* need a PREFER_DISCREET_ELT_PRIM here too.. + /* need a PREFER_DISCRETE_ELT_PRIM here too.. */ EMIT_PRIM( ctx, GL_TRIANGLES, HW_TRIANGLES, start, count ); } @@ -352,7 +354,7 @@ static void TAG(render_tri_strip_verts)( GLcontext *ctx, if (start + 2 >= count) return; - if (PREFER_DISCREET_ELT_PRIM( count-start, HW_TRIANGLES )) + if (PREFER_DISCRETE_ELT_PRIM( count-start, HW_TRIANGLES )) { int dmasz = GET_SUBSEQUENT_VB_MAX_ELTS(); int currentsz; @@ -434,7 +436,7 @@ static void TAG(render_tri_fan_verts)( GLcontext *ctx, if (start+2 >= count) return; - if (PREFER_DISCREET_ELT_PRIM( count-start, HW_TRIANGLES )) + if (PREFER_DISCRETE_ELT_PRIM( count-start, HW_TRIANGLES )) { int dmasz = GET_SUBSEQUENT_VB_MAX_ELTS(); int currentsz; diff --git a/xc/extras/Mesa/src/tnl_dd/t_dd_triemit.h b/xc/extras/Mesa/src/tnl_dd/t_dd_triemit.h index 39c4a20c4..2f5bcee7e 100644 --- a/xc/extras/Mesa/src/tnl_dd/t_dd_triemit.h +++ b/xc/extras/Mesa/src/tnl_dd/t_dd_triemit.h @@ -1,3 +1,10 @@ +#ifndef DO_DEBUG_VERTS +#define DO_DEBUG_VERTS 0 +#endif + +#ifndef PRINT_VERTEX +#define PRINT_VERTEX(x) +#endif #if defined(USE_X86_ASM) #define COPY_DWORDS( j, vb, vertsize, v ) \ @@ -31,6 +38,14 @@ static __inline void TAG(quad)( CTX_ARG, GLuint *vb = (GLuint *)ALLOC_VERTS( 4, vertsize); GLuint j; + if (DO_DEBUG_VERTS) { + fprintf(stderr, "%s\n", __FUNCTION__); + PRINT_VERTEX(v0); + PRINT_VERTEX(v1); + PRINT_VERTEX(v2); + PRINT_VERTEX(v3); + } + COPY_DWORDS( j, vb, vertsize, v0 ); COPY_DWORDS( j, vb, vertsize, v1 ); COPY_DWORDS( j, vb, vertsize, v2 ); @@ -47,6 +62,14 @@ static __inline void TAG(quad)( CTX_ARG, GLuint *vb = (GLuint *)ALLOC_VERTS( 6, vertsize); GLuint j; + if (DO_DEBUG_VERTS) { + fprintf(stderr, "%s\n", __FUNCTION__); + PRINT_VERTEX(v0); + PRINT_VERTEX(v1); + PRINT_VERTEX(v2); + PRINT_VERTEX(v3); + } + COPY_DWORDS( j, vb, vertsize, v0 ); COPY_DWORDS( j, vb, vertsize, v1 ); COPY_DWORDS( j, vb, vertsize, v3 ); @@ -66,6 +89,13 @@ static __inline void TAG(triangle)( CTX_ARG, GLuint *vb = (GLuint *)ALLOC_VERTS( 3, vertsize); GLuint j; + if (DO_DEBUG_VERTS) { + fprintf(stderr, "%s\n", __FUNCTION__); + PRINT_VERTEX(v0); + PRINT_VERTEX(v1); + PRINT_VERTEX(v2); + } + COPY_DWORDS( j, vb, vertsize, v0 ); COPY_DWORDS( j, vb, vertsize, v1 ); COPY_DWORDS( j, vb, vertsize, v2 ); @@ -108,7 +138,17 @@ static void TAG(fast_clipped_poly)( GLcontext *ctx, const GLuint *elts, const GLuint *start = (const GLuint *)VERT(elts[0]); int i,j; + if (DO_DEBUG_VERTS) { + fprintf(stderr, "%s\n", __FUNCTION__); + PRINT_VERTEX(VERT(elts[0])); + PRINT_VERTEX(VERT(elts[1])); + } + for (i = 2 ; i < n ; i++) { + if (DO_DEBUG_VERTS) { + PRINT_VERTEX(VERT(elts[i])); + } + COPY_DWORDS( j, vb, vertsize, VERT(elts[i-1]) ); COPY_DWORDS( j, vb, vertsize, VERT(elts[i]) ); COPY_DWORDS( j, vb, vertsize, start ); diff --git a/xc/extras/Mesa/src/tnl_dd/t_dd_vb.c b/xc/extras/Mesa/src/tnl_dd/t_dd_vb.c index 2478ebaa4..3ba8fb3d8 100644 --- a/xc/extras/Mesa/src/tnl_dd/t_dd_vb.c +++ b/xc/extras/Mesa/src/tnl_dd/t_dd_vb.c @@ -194,22 +194,78 @@ void TAG(print_vertex)( GLcontext *ctx, const VERTEX *v ) LOCALVARS GLuint format = GET_VERTEX_FORMAT(); - if (format == TINY_VERTEX_FORMAT) { - fprintf(stderr, "x %f y %f z %f\n", v->v.x, v->v.y, v->v.z); - fprintf(stderr, "r %d g %d b %d a %d\n", + fprintf(stderr, "(%x) ", format); + + switch (format) { +#if HAVE_TINY_VERTICES + case TINY_VERTEX_FORMAT: + fprintf(stderr, "xyz %.4f,%.4f,%.4f rgba %x:%x:%x:%x\n", + v->v.x, v->v.y, v->v.z, v->tv.color.red, v->tv.color.green, v->tv.color.blue, v->tv.color.alpha); - } - else { - fprintf(stderr, "x %f y %f z %f oow %f\n", - v->v.x, v->v.y, v->v.z, v->v.w); - fprintf(stderr, "r %d g %d b %d a %d\n", + break; +#endif +#if HAVE_NOTEX_VERTICES + case NOTEX_VERTEX_FORMAT: + fprintf(stderr, "xyzw %.4f,%.4f,%.4f,%.4f rgba %x:%x:%x:%x spec %x:%x:%x:%x\n", + v->v.x, v->v.y, v->v.z, v->v.w, + v->v.color.red, + v->v.color.green, + v->v.color.blue, + v->v.color.alpha, + v->v.specular.red, + v->v.specular.green, + v->v.specular.blue, + v->v.specular.alpha); + break; +#endif +#if HAVE_TEX0_VERTICES + case TEX0_VERTEX_FORMAT: + fprintf(stderr, "xyzw %.4f,%.4f,%.4f,%.4f rgba %x:%x:%x:%x st %.4f,%.4f\n", + v->v.x, v->v.y, v->v.z, v->v.w, + v->v.color.red, + v->v.color.green, + v->v.color.blue, + v->v.color.alpha, + v->v.u0, + v->v.v0); + break; +#endif +#if HAVE_TEX1_VERTICES + case TEX1_VERTEX_FORMAT: + fprintf(stderr, "xyzw %.4f,%.4f,%.4f,%.4f rgba %x:%x:%x:%x st %.4f,%.4f st %.4f,%.4f\n", + v->v.x, v->v.y, v->v.z, v->v.w, + v->v.color.red, + v->v.color.green, + v->v.color.blue, + v->v.color.alpha, + v->v.u0, + v->v.v0, + v->v.u1, + v->v.u2); + break; +#endif +#if HAVE_PTEX_VERTICES + case PROJ_TEX1_VERTEX_FORMAT: + fprintf(stderr, "xyzw %.4f,%.4f,%.4f,%.4f rgba %x:%x:%x:%x stq %.4f,%.4f,%.4f stq %.4f,%.4f,%.4f\n", + v->v.x, v->v.y, v->v.z, v->v.w, v->v.color.red, v->v.color.green, v->v.color.blue, - v->v.color.alpha); + v->v.color.alpha, + v->pv.u0, + v->pv.v0, + v->pv.q0, + v->pv.u1, + v->pv.v1, + v->pv.q1); + break; +#endif + default: + fprintf(stderr, "???\n"); + break; } fprintf(stderr, "\n"); diff --git a/xc/extras/ogl-sample/main/gfx/lib/glu/libnurbs/nurbtess/quicksort.cc b/xc/extras/ogl-sample/main/gfx/lib/glu/libnurbs/nurbtess/quicksort.cc index 02cfb906c..9d0b290b3 100755 --- a/xc/extras/ogl-sample/main/gfx/lib/glu/libnurbs/nurbtess/quicksort.cc +++ b/xc/extras/ogl-sample/main/gfx/lib/glu/libnurbs/nurbtess/quicksort.cc @@ -31,10 +31,8 @@ ** published by SGI, but has not been independently verified as being ** compliant with the OpenGL(R) version 1.2.1 Specification. ** -** $Date: 2002/07/05 08:31:06 $ $Revision: 1.3 $ */ /* -** $Header: /home/ajax/dri-backup/xc/xc/extras/ogl-sample/main/gfx/lib/glu/libnurbs/nurbtess/Attic/quicksort.cc,v 1.3 2002/07/05 08:31:06 alanh Exp $ */ #include <stdlib.h> diff --git a/xc/include/GL/glx.h b/xc/include/GL/glx.h index 0a8aa3e97..becfb864d 100644 --- a/xc/include/GL/glx.h +++ b/xc/include/GL/glx.h @@ -112,6 +112,13 @@ extern void glXGetSelectedEvent (Display *dpy, GLXDrawable draw, unsigned long * extern void (*glXGetProcAddress(const GLubyte *procname))(void); +/* AGP memory allocation */ +extern void *glXAllocateMemoryNV(GLsizei size, GLfloat readfreq, GLfloat writefreq, GLfloat priority); +extern void glXFreeMemoryNV(GLvoid *pointer); +typedef void * ( * PFNGLXALLOCATEMEMORYNVPROC) (GLsizei size, GLfloat readfreq, GLfloat writefreq, GLfloat priority); +typedef void ( * PFNGLXFREEMEMORYNVPROC) (GLvoid *pointer); + + #ifndef GLX_GLXEXT_LEGACY #include <GL/glxext.h> @@ -130,6 +137,8 @@ extern int glXQueryContextInfoEXT (Display *dpy, GLXContext ctx, int attribute, extern Display * glXGetCurrentDisplayEXT (void); extern void (*glXGetProcAddressARB(const GLubyte *procName))( void ); + + #endif /* GLX_GLXEXT_LEGACY */ /*** Should these go here, or in another header? */ diff --git a/xc/lib/GL/dri/dri_glx.c b/xc/lib/GL/dri/dri_glx.c index d962e8585..672b22690 100644 --- a/xc/lib/GL/dri/dri_glx.c +++ b/xc/lib/GL/dri/dri_glx.c @@ -52,6 +52,7 @@ SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. typedef void *(*CreateScreenFunc)(Display *dpy, int scrn, __DRIscreen *psc, int numConfigs, __GLXvisualConfig *config); +typedef void *(*RegisterExtensionsFunc)(void); #ifdef BUILT_IN_DRI_DRIVER @@ -68,6 +69,23 @@ extern void *__driCreateScreen(Display *dpy, int scrn, __DRIscreen *psc, #define DEFAULT_DRIVER_DIR "/usr/X11R6/lib/modules/dri" #endif +/* + * We keep a linked list of these structures, one per DRI device driver. + */ +typedef struct __DRIdriverRec { + const char *name; + void *handle; + CreateScreenFunc createScreenFunc; + RegisterExtensionsFunc registerExtensionsFunc; + struct __DRIdriverRec *next; +} __DRIdriver; + +static __DRIdriver *Drivers = NULL; + + +/* + * printf wrappers + */ static void InfoMessageF(const char *f, ...) { @@ -94,36 +112,6 @@ static void ErrorMessageF(const char *f, ...) } } -#if 0 -static void PrintF(const char *f, ...) -{ - va_list args; - const char *env; - - if ((env = getenv("LIBGL_DEBUG")) && strstr(env, "verbose")) { - va_start(args, f); - vfprintf(stderr, f, args); - va_end(args); - } -} -#endif - -static void ErrorMessage(const char *msg) -{ - if (getenv("LIBGL_DEBUG")) { - fprintf(stderr, "libGL error: %s\n", msg); - } -} - - -static void InfoMessage(const char *msg) -{ - const char *env = getenv("LIBGL_DEBUG"); - if (env && strstr(env, "verbose")) { - fprintf(stderr, "libGL: %s\n", msg); - } -} - /* * We'll save a pointer to this function when we couldn't find a @@ -194,7 +182,6 @@ static void ExtractDir(int index, const char *paths, int dirLen, char *dir) } - /* * Try to dlopen() the named driver. This function adds the * "_dri.so" suffix to the driver name and searches the @@ -205,10 +192,19 @@ static void ExtractDir(int index, const char *paths, int dirLen, char *dir) * Return: * handle from dlopen, or NULL if driver file not found. */ -static void *OpenDriver(const char *driverName) +static __DRIdriver *OpenDriver(const char *driverName) { char *libPaths = NULL; int i; + __DRIdriver *driver; + + /* First, search Drivers list to see if we've already opened this driver */ + for (driver = Drivers; driver; driver = driver->next) { + if (strcmp(driver->name, driverName) == 0) { + /* found it */ + return driver; + } + } if (geteuid() == getuid()) { /* don't allow setuid apps to use LIBGL_DRIVERS_PATH */ @@ -224,95 +220,104 @@ static void *OpenDriver(const char *driverName) void *handle; ExtractDir(i, libPaths, 1000, libDir); if (!libDir[0]) - return NULL; + break; /* ran out of paths to search */ snprintf(realDriverName, 200, "%s/%s_dri.so", libDir, driverName); InfoMessageF("OpenDriver: trying %s\n", realDriverName); handle = dlopen(realDriverName, RTLD_NOW | RTLD_GLOBAL); if (handle) { - return handle; - } - else { - ErrorMessageF("dlopen failed: %s\n", dlerror()); + /* allocate __DRIdriver struct */ + driver = (__DRIdriver *) Xmalloc(sizeof(__DRIdriver)); + if (!driver) + return NULL; /* out of memory! */ + /* init the struct */ + driver->name = __glXstrdup(driverName); + if (!driver->name) { + Xfree(driver); + return NULL; /* out of memory! */ + } + driver->createScreenFunc = (CreateScreenFunc) + dlsym(handle, "__driCreateScreen"); + if (!driver->createScreenFunc) { + /* If the driver doesn't have this symbol then something's + * really, really wrong. + */ + ErrorMessageF("__driCreateScreen() not defined in %s_dri.so!\n", + driverName); + Xfree(driver); + dlclose(handle); + continue; + } + driver->registerExtensionsFunc = (RegisterExtensionsFunc) + dlsym(handle, "__driRegisterExtensions"); + driver->handle = handle; + /* put at head of linked list */ + driver->next = Drivers; + Drivers = driver; + return driver; } } + ErrorMessageF("unable to find driver: %s_dri.so\n", driverName); return NULL; } - /* - * Initialize two arrays: an array of createScreen function pointers - * and an array of dlopen library handles. Arrays are indexed by - * screen number. - * We use the DRI in order to find the __driCreateScreen function - * exported by each screen on a display. + * Given a display pointer and screen number, determine the name of + * the DRI driver for the screen. (I.e. "r128", "tdfx", etc). + * Return True for success, False for failure. */ -static void Find_CreateScreenFuncs(Display *dpy, - CreateScreenFunc *createFuncs, - void **libraryHandles) +static Bool GetDriverName(Display *dpy, int scrNum, char **driverName) { - const int numScreens = ScreenCount(dpy); - int scrn; + int directCapable; + Bool b; + int driverMajor, driverMinor, driverPatch; - __glXRegisterExtensions(); + *driverName = NULL; - for (scrn = 0; scrn < numScreens; scrn++) { - int directCapable; - Bool b; - int driverMajor, driverMinor, driverPatch; - char *driverName = NULL; - void *handle; - - /* defaults */ - createFuncs[scrn] = DummyCreateScreen; - libraryHandles[scrn] = NULL; - - if (!XF86DRIQueryDirectRenderingCapable(dpy, scrn, &directCapable)) { - ErrorMessage("XF86DRIQueryDirectRenderingCapable failed"); - continue; - } - if (!directCapable) { - ErrorMessage("XF86DRIQueryDirectRenderingCapable returned false"); - continue; - } + if (!XF86DRIQueryDirectRenderingCapable(dpy, scrNum, &directCapable)) { + ErrorMessageF("XF86DRIQueryDirectRenderingCapable failed"); + return False; + } + if (!directCapable) { + ErrorMessageF("XF86DRIQueryDirectRenderingCapable returned false"); + return False; + } - /* - * Use DRI to find the device driver for use on screen number 'scrn'. - */ - b = XF86DRIGetClientDriverName(dpy, scrn, &driverMajor, &driverMinor, - &driverPatch, &driverName); - if (!b) { - ErrorMessageF("Cannot determine driver name for screen %d\n", scrn); - continue; - } + b = XF86DRIGetClientDriverName(dpy, scrNum, &driverMajor, &driverMinor, + &driverPatch, driverName); + if (!b) { + ErrorMessageF("Cannot determine driver name for screen %d\n", scrNum); + return False; + } + InfoMessageF("XF86DRIGetClientDriverName: %d.%d.%d %s (screen %d)\n", + driverMajor, driverMinor, driverPatch, *driverName, scrNum); - /* - * Open the driver module and save the pointer to its - * __driCreateScreen function. - */ - handle = OpenDriver(driverName); - if (handle) { - CreateScreenFunc createScreenFunc; - createScreenFunc = (CreateScreenFunc) dlsym(handle, "__driCreateScreen"); - if (createScreenFunc) { - /* success! */ - createFuncs[scrn] = createScreenFunc; - libraryHandles[scrn] = handle; - continue; /* onto the next screen */ - } - else { - ErrorMessage("driCreateScreen() not defined in driver!"); - dlclose(handle); - } - } - } /* for scrn */ + return True; } + +/* + * Given a display pointer and screen number, return a __DRIdriver handle. + * Return NULL if anything goes wrong. + */ +static __DRIdriver *GetDriver(Display *dpy, int scrNum) +{ + char *driverName; + + if (GetDriverName(dpy, scrNum, &driverName)) { + return OpenDriver(driverName); + } + return NULL; +} + + #endif /* BUILT_IN_DRI_DRIVER */ +/* This function isn't currently used. + */ static void driDestroyDisplay(Display *dpy, void *private) { __DRIdisplayPrivate *pdpyp = (__DRIdisplayPrivate *)private; @@ -330,12 +335,18 @@ static void driDestroyDisplay(Display *dpy, void *private) } +/* + * Allocate, initialize and return a __DRIdisplayPrivate object. + * This is called from __glXInitialize() when we are given a new + * display pointer. + */ void *driCreateDisplay(Display *dpy, __DRIdisplay *pdisp) { const int numScreens = ScreenCount(dpy); __DRIdisplayPrivate *pdpyp; int eventBase, errorBase; int major, minor, patch; + int scrn; /* Initialize these fields to NULL in case we fail. * If we don't do this we may later get segfaults trying to free random @@ -366,100 +377,49 @@ void *driCreateDisplay(Display *dpy, __DRIdisplay *pdisp) /* allocate array of pointers to createScreen funcs */ pdisp->createScreen = (CreateScreenFunc *) Xmalloc(numScreens * sizeof(void *)); - if (!pdisp->createScreen) + if (!pdisp->createScreen) { + XFree(pdpyp); return NULL; + } /* allocate array of library handles */ pdpyp->libraryHandles = (void **) Xmalloc(numScreens * sizeof(void*)); if (!pdpyp->libraryHandles) { Xfree(pdisp->createScreen); + XFree(pdpyp); return NULL; } #ifdef BUILT_IN_DRI_DRIVER - /* we'll statically bind to the __driCreateScreen function */ - { - int i; - for (i = 0; i < numScreens; i++) { - pdisp->createScreen[i] = __driCreateScreen; - pdpyp->libraryHandles[i] = NULL; - } + /* we'll statically bind to the built-in __driCreateScreen function */ + for (scrn = 0; scrn < numScreens; scrn++) { + pdisp->createScreen[scrn] = __driCreateScreen; + pdpyp->libraryHandles[scrn] = NULL; } + #else - Find_CreateScreenFuncs(dpy, pdisp->createScreen, pdpyp->libraryHandles); + /* dynamically discover DRI drivers for all screens, saving each + * driver's "__driCreateScreen" function pointer. That's the bootstrap + * entrypoint for all DRI drivers. + */ + __glXRegisterExtensions(); + for (scrn = 0; scrn < numScreens; scrn++) { + __DRIdriver *driver = GetDriver(dpy, scrn); + if (driver) { + pdisp->createScreen[scrn] = driver->createScreenFunc; + pdpyp->libraryHandles[scrn] = driver->handle; + } + else { + pdisp->createScreen[scrn] = DummyCreateScreen; + pdpyp->libraryHandles[scrn] = NULL; + } + } #endif return (void *)pdpyp; } -#ifndef BUILT_IN_DRI_DRIVER -/* - * Use the DRI and dlopen/dlsym facilities to find the GL extensions - * possible on the given display and screen. - */ -static void -register_extensions_on_screen(Display *dpy, int scrNum) -{ - int eventBase, errorBase; - Bool b, b2; - int driMajor, driMinor, driPatch; - int driverMajor, driverMinor, driverPatch; - char *driverName = NULL; - void *handle; - - /* - * Check if the DRI extension is available, check the DRI version, - * determine the 3D driver for the screen. - */ - b = XF86DRIQueryExtension(dpy, &eventBase, &errorBase); - if (!b) { - InfoMessage("XF86DRIQueryExtension failed"); - return; - } - - b = XF86DRIQueryDirectRenderingCapable(dpy, scrNum, &b2); - if (!b || !b2) { - InfoMessage("XF86DRIQueryDirectRenderingCapable failed"); - return; - } - - b = XF86DRIQueryVersion(dpy, &driMajor, &driMinor, &driPatch); - if (!b) { - InfoMessage("XF86DRIQueryVersion failed"); - } - - b = XF86DRIGetClientDriverName(dpy, scrNum, &driverMajor, &driverMinor, - &driverPatch, &driverName); - if (!b) { - InfoMessage("XF86DRIGetClientDriverName failed"); - return; - } - else { - InfoMessageF("XF86DRIGetClientDriverName: %d.%d.%d %s (screen %d)\n", - driverMajor, driverMinor, driverPatch, driverName, scrNum); - } - - /* - * OK, now we know the name of the relevant driver for this screen. - * dlopen() the driver library file, get a pointer to the driver's - * __driRegisterExtensions() function, and call it if it exists. - */ - handle = OpenDriver(driverName); - if (handle) { - typedef void *(*RegisterExtFunc)(void); - RegisterExtFunc registerExtFunc = (RegisterExtFunc) dlsym(handle, - "__driRegisterExtensions"); - if (registerExtFunc) { - (*registerExtFunc)(); - } - dlclose(handle); - return; - } -} -#endif /* !BUILT_IN_DRI_DRIVER */ - - /* ** Here we'll query the DRI driver for each screen and let each @@ -474,39 +434,45 @@ register_extensions_on_screen(Display *dpy, int scrNum) void __glXRegisterExtensions(void) { +#ifndef BUILT_IN_DRI_DRIVER static GLboolean alreadyCalled = GL_FALSE; + int displayNum, maxDisplays; - if (alreadyCalled) { + if (alreadyCalled) return; + alreadyCalled = GL_TRUE; + + if (getenv("LIBGL_MULTIHEAD")) { + /* we'd like to always take this path but doing so causes a second + * or more of delay while the XOpenDisplay() function times out. + */ + maxDisplays = 10; /* infinity, really */ + } + else { + /* just open the :0 display */ + maxDisplays = 1; } -#ifndef BUILT_IN_DRI_DRIVER - { - int displayNum, maxDisplays; - if (getenv("LIBGL_MULTIHEAD")) - maxDisplays = 10; /* infinity, really */ - else - maxDisplays = 1; - for (displayNum = 0; displayNum < maxDisplays; displayNum++) { - char displayName[200]; - Display *dpy; - snprintf(displayName, 199, ":%d.0", displayNum); - dpy = XOpenDisplay(displayName); - if (dpy) { - const int numScreens = ScreenCount(dpy); - int screenNum; - for (screenNum = 0; screenNum < numScreens; screenNum++) { - register_extensions_on_screen(dpy, screenNum); + for (displayNum = 0; displayNum < maxDisplays; displayNum++) { + char displayName[200]; + Display *dpy; + snprintf(displayName, 199, ":%d.0", displayNum); + dpy = XOpenDisplay(displayName); + if (dpy) { + const int numScreens = ScreenCount(dpy); + int screenNum; + for (screenNum = 0; screenNum < numScreens; screenNum++) { + __DRIdriver *driver = GetDriver(dpy, screenNum); + if (driver && driver->registerExtensionsFunc) { + (*driver->registerExtensionsFunc)(); } - XCloseDisplay(dpy); - } - else { - break; } + XCloseDisplay(dpy); + } + else { + break; } } - - alreadyCalled = GL_TRUE; #endif } diff --git a/xc/lib/GL/glx/glxclient.h b/xc/lib/GL/glx/glxclient.h index 1480a5854..46d2398b2 100644 --- a/xc/lib/GL/glx/glxclient.h +++ b/xc/lib/GL/glx/glxclient.h @@ -547,6 +547,10 @@ extern __GLXdisplayPrivate *__glXInitialize(Display*); /* Query drivers for dynamically registered extensions */ extern void __glXRegisterExtensions(void); +/* Functions for extending the GLX API: */ +extern void *__glXRegisterGLXFunction(const char *funcName, void *funcAddr); +extern void __glXRegisterGLXExtensionString(const char *extName); + /************************************************************************/ @@ -696,4 +700,6 @@ extern void _XSend(Display*, const void*, long); #endif +extern char *__glXstrdup(const char *str); + #endif /* !__GLX_client_h__ */ diff --git a/xc/lib/GL/glx/glxcmds.c b/xc/lib/GL/glx/glxcmds.c index f10383920..1c95befe9 100644 --- a/xc/lib/GL/glx/glxcmds.c +++ b/xc/lib/GL/glx/glxcmds.c @@ -65,12 +65,15 @@ static const char __glXGLClientExtensions[] = static const char __glXGLXClientVendorName[] = "SGI"; static const char __glXGLXClientVersion[] = "1.2"; -static const char __glXGLXClientExtensions[] = +static const char __glXGLXDefaultClientExtensions[] = "GLX_EXT_visual_info " "GLX_EXT_visual_rating " "GLX_EXT_import_context " ; +static const char *__glXGLXClientExtensions = __glXGLXDefaultClientExtensions; + + /* ** Create a new context. */ @@ -1214,7 +1217,7 @@ static char *combine_strings( const char *cext_string, const char *sext_string ) */ if ( (clen = strlen( cext_string)) > (slen = strlen( sext_string)) ) { combo_string = (char *) Xmalloc( slen + 2 ); - s1 = (char *) malloc( slen + 2 ); strcpy( s1, sext_string ); + s1 = (char *) Xmalloc( slen + 2 ); strcpy( s1, sext_string ); s2 = cext_string; } else { combo_string = (char *) Xmalloc( clen + 2 ); @@ -2020,6 +2023,21 @@ Bool GLX_PREFIX(glXSet3DfxModeMESA)( int mode ) } + +/* strdup() is actually not a standard ANSI C or POSIX routine. + * Irix will not define it if ANSI mode is in effect. + */ +char * +__glXstrdup(const char *str) +{ + char *copy; + copy = (char *) Xmalloc(strlen(str) + 1); + if (!copy) + return NULL; + strcpy(copy, str); + return copy; +} + /* ** glXGetProcAddress support */ @@ -2027,142 +2045,230 @@ Bool GLX_PREFIX(glXSet3DfxModeMESA)( int mode ) struct name_address_pair { const char *Name; GLvoid *Address; + struct name_address_pair *Next; }; static struct name_address_pair GLX_functions[] = { /*** GLX_VERSION_1_0 ***/ - { "glXChooseVisual", (GLvoid *) glXChooseVisual }, - { "glXCopyContext", (GLvoid *) glXCopyContext }, - { "glXCreateContext", (GLvoid *) glXCreateContext }, - { "glXCreateGLXPixmap", (GLvoid *) glXCreateGLXPixmap }, - { "glXDestroyContext", (GLvoid *) glXDestroyContext }, - { "glXDestroyGLXPixmap", (GLvoid *) glXDestroyGLXPixmap }, - { "glXGetConfig", (GLvoid *) glXGetConfig }, - { "glXGetCurrentContext", (GLvoid *) glXGetCurrentContext }, - { "glXGetCurrentDrawable", (GLvoid *) glXGetCurrentDrawable }, - { "glXIsDirect", (GLvoid *) glXIsDirect }, - { "glXMakeCurrent", (GLvoid *) glXMakeCurrent }, - { "glXQueryExtension", (GLvoid *) glXQueryExtension }, - { "glXQueryVersion", (GLvoid *) glXQueryVersion }, - { "glXSwapBuffers", (GLvoid *) glXSwapBuffers }, - { "glXUseXFont", (GLvoid *) glXUseXFont }, - { "glXWaitGL", (GLvoid *) glXWaitGL }, - { "glXWaitX", (GLvoid *) glXWaitX }, + { "glXChooseVisual", (GLvoid *) glXChooseVisual, NULL }, + { "glXCopyContext", (GLvoid *) glXCopyContext, NULL }, + { "glXCreateContext", (GLvoid *) glXCreateContext, NULL }, + { "glXCreateGLXPixmap", (GLvoid *) glXCreateGLXPixmap, NULL }, + { "glXDestroyContext", (GLvoid *) glXDestroyContext, NULL }, + { "glXDestroyGLXPixmap", (GLvoid *) glXDestroyGLXPixmap, NULL }, + { "glXGetConfig", (GLvoid *) glXGetConfig, NULL }, + { "glXGetCurrentContext", (GLvoid *) glXGetCurrentContext, NULL }, + { "glXGetCurrentDrawable", (GLvoid *) glXGetCurrentDrawable, NULL }, + { "glXIsDirect", (GLvoid *) glXIsDirect, NULL }, + { "glXMakeCurrent", (GLvoid *) glXMakeCurrent, NULL }, + { "glXQueryExtension", (GLvoid *) glXQueryExtension, NULL }, + { "glXQueryVersion", (GLvoid *) glXQueryVersion, NULL }, + { "glXSwapBuffers", (GLvoid *) glXSwapBuffers, NULL }, + { "glXUseXFont", (GLvoid *) glXUseXFont, NULL }, + { "glXWaitGL", (GLvoid *) glXWaitGL, NULL }, + { "glXWaitX", (GLvoid *) glXWaitX, NULL }, /*** GLX_VERSION_1_1 ***/ - { "glXGetClientString", (GLvoid *) glXGetClientString }, - { "glXQueryExtensionsString", (GLvoid *) glXQueryExtensionsString }, - { "glXQueryServerString", (GLvoid *) glXQueryServerString }, + { "glXGetClientString", (GLvoid *) glXGetClientString, NULL }, + { "glXQueryExtensionsString", (GLvoid *) glXQueryExtensionsString, NULL }, + { "glXQueryServerString", (GLvoid *) glXQueryServerString, NULL }, /*** GLX_VERSION_1_2 ***/ - { "glXGetCurrentDisplay", (GLvoid *) glXGetCurrentDisplay }, + { "glXGetCurrentDisplay", (GLvoid *) glXGetCurrentDisplay, NULL }, /*** GLX_VERSION_1_3 ***/ - { "glXChooseFBConfig", (GLvoid *) glXChooseFBConfig }, - { "glXCreateNewContext", (GLvoid *) glXCreateNewContext }, - { "glXCreatePbuffer", (GLvoid *) glXCreatePbuffer }, - { "glXCreatePixmap", (GLvoid *) glXCreatePixmap }, - { "glXCreateWindow", (GLvoid *) glXCreateWindow }, - { "glXDestroyPbuffer", (GLvoid *) glXDestroyPbuffer }, - { "glXDestroyPixmap", (GLvoid *) glXDestroyPixmap }, - { "glXDestroyWindow", (GLvoid *) glXDestroyWindow }, - { "glXGetCurrentReadDrawable", (GLvoid *) glXGetCurrentReadDrawable }, - { "glXGetFBConfigAttrib", (GLvoid *) glXGetFBConfigAttrib }, - { "glXGetFBConfigs", (GLvoid *) glXGetFBConfigs }, - { "glXGetSelectedEvent", (GLvoid *) glXGetSelectedEvent }, - { "glXGetVisualFromFBConfig", (GLvoid *) glXGetVisualFromFBConfig }, - { "glXMakeContextCurrent", (GLvoid *) glXMakeContextCurrent }, - { "glXQueryContext", (GLvoid *) glXQueryContext }, - { "glXQueryDrawable", (GLvoid *) glXQueryDrawable }, - { "glXSelectEvent", (GLvoid *) glXSelectEvent }, + { "glXChooseFBConfig", (GLvoid *) glXChooseFBConfig, NULL }, + { "glXCreateNewContext", (GLvoid *) glXCreateNewContext, NULL }, + { "glXCreatePbuffer", (GLvoid *) glXCreatePbuffer, NULL }, + { "glXCreatePixmap", (GLvoid *) glXCreatePixmap, NULL }, + { "glXCreateWindow", (GLvoid *) glXCreateWindow, NULL }, + { "glXDestroyPbuffer", (GLvoid *) glXDestroyPbuffer, NULL }, + { "glXDestroyPixmap", (GLvoid *) glXDestroyPixmap, NULL }, + { "glXDestroyWindow", (GLvoid *) glXDestroyWindow, NULL }, + { "glXGetCurrentReadDrawable", (GLvoid *) glXGetCurrentReadDrawable, NULL }, + { "glXGetFBConfigAttrib", (GLvoid *) glXGetFBConfigAttrib, NULL }, + { "glXGetFBConfigs", (GLvoid *) glXGetFBConfigs, NULL }, + { "glXGetSelectedEvent", (GLvoid *) glXGetSelectedEvent, NULL }, + { "glXGetVisualFromFBConfig", (GLvoid *) glXGetVisualFromFBConfig, NULL }, + { "glXMakeContextCurrent", (GLvoid *) glXMakeContextCurrent, NULL }, + { "glXQueryContext", (GLvoid *) glXQueryContext, NULL }, + { "glXQueryDrawable", (GLvoid *) glXQueryDrawable, NULL }, + { "glXSelectEvent", (GLvoid *) glXSelectEvent, NULL }, /*** GLX_SGI_swap_control ***/ - { "glXSwapIntervalSGI", (GLvoid *) glXSwapIntervalSGI }, + { "glXSwapIntervalSGI", (GLvoid *) glXSwapIntervalSGI, NULL }, /*** GLX_SGI_video_sync ***/ - { "glXGetVideoSyncSGI", (GLvoid *) glXGetVideoSyncSGI }, - { "glXWaitVideoSyncSGI", (GLvoid *) glXWaitVideoSyncSGI }, + { "glXGetVideoSyncSGI", (GLvoid *) glXGetVideoSyncSGI, NULL }, + { "glXWaitVideoSyncSGI", (GLvoid *) glXWaitVideoSyncSGI, NULL }, /*** GLX_SGI_make_current_read ***/ - { "glXMakeCurrentReadSGI", (GLvoid *) glXMakeCurrentReadSGI }, - { "glXGetCurrentReadDrawableSGI", (GLvoid *) glXGetCurrentReadDrawableSGI }, + { "glXMakeCurrentReadSGI", (GLvoid *) glXMakeCurrentReadSGI, NULL }, + { "glXGetCurrentReadDrawableSGI", (GLvoid *) glXGetCurrentReadDrawableSGI, NULL }, /*** GLX_SGIX_video_source ***/ #if defined(_VL_H) - { "glXCreateGLXVideoSourceSGIX", (GLvoid *) glXCreateGLXVideoSourceSGIX }, - { "glXDestroyGLXVideoSourceSGIX", (GLvoid *) glXDestroyGLXVideoSourceSGIX }, + { "glXCreateGLXVideoSourceSGIX", (GLvoid *) glXCreateGLXVideoSourceSGIX, NULL }, + { "glXDestroyGLXVideoSourceSGIX", (GLvoid *) glXDestroyGLXVideoSourceSGIX, NULL }, #endif /*** GLX_EXT_import_context ***/ - { "glXFreeContextEXT", (GLvoid *) glXFreeContextEXT }, - { "glXGetContextIDEXT", (GLvoid *) glXGetContextIDEXT }, - { "glXGetCurrentDisplayEXT", (GLvoid *) glXGetCurrentDisplayEXT }, - { "glXImportContextEXT", (GLvoid *) glXImportContextEXT }, - { "glXQueryContextInfoEXT", (GLvoid *) glXQueryContextInfoEXT }, + { "glXFreeContextEXT", (GLvoid *) glXFreeContextEXT, NULL }, + { "glXGetContextIDEXT", (GLvoid *) glXGetContextIDEXT, NULL }, + { "glXGetCurrentDisplayEXT", (GLvoid *) glXGetCurrentDisplayEXT, NULL }, + { "glXImportContextEXT", (GLvoid *) glXImportContextEXT, NULL }, + { "glXQueryContextInfoEXT", (GLvoid *) glXQueryContextInfoEXT, NULL }, /*** GLX_SGIX_fbconfig ***/ - { "glXGetFBConfigAttribSGIX", (GLvoid *) glXGetFBConfigAttribSGIX }, - { "glXChooseFBConfigSGIX", (GLvoid *) glXChooseFBConfigSGIX }, - { "glXCreateGLXPixmapWithConfigSGIX", (GLvoid *) glXCreateGLXPixmapWithConfigSGIX }, - { "glXCreateContextWithConfigSGIX", (GLvoid *) glXCreateContextWithConfigSGIX }, - { "glXGetVisualFromFBConfigSGIX", (GLvoid *) glXGetVisualFromFBConfigSGIX }, - { "glXGetFBConfigFromVisualSGIX", (GLvoid *) glXGetFBConfigFromVisualSGIX }, + { "glXGetFBConfigAttribSGIX", (GLvoid *) glXGetFBConfigAttribSGIX, NULL }, + { "glXChooseFBConfigSGIX", (GLvoid *) glXChooseFBConfigSGIX, NULL }, + { "glXCreateGLXPixmapWithConfigSGIX", (GLvoid *) glXCreateGLXPixmapWithConfigSGIX, NULL }, + { "glXCreateContextWithConfigSGIX", (GLvoid *) glXCreateContextWithConfigSGIX, NULL }, + { "glXGetVisualFromFBConfigSGIX", (GLvoid *) glXGetVisualFromFBConfigSGIX, NULL }, + { "glXGetFBConfigFromVisualSGIX", (GLvoid *) glXGetFBConfigFromVisualSGIX, NULL }, /*** GLX_SGIX_pbuffer ***/ - { "glXCreateGLXPbufferSGIX", (GLvoid *) glXCreateGLXPbufferSGIX }, - { "glXDestroyGLXPbufferSGIX", (GLvoid *) glXDestroyGLXPbufferSGIX }, - { "glXQueryGLXPbufferSGIX", (GLvoid *) glXQueryGLXPbufferSGIX }, - { "glXSelectEventSGIX", (GLvoid *) glXSelectEventSGIX }, - { "glXGetSelectedEventSGIX", (GLvoid *) glXGetSelectedEventSGIX }, + { "glXCreateGLXPbufferSGIX", (GLvoid *) glXCreateGLXPbufferSGIX, NULL }, + { "glXDestroyGLXPbufferSGIX", (GLvoid *) glXDestroyGLXPbufferSGIX, NULL }, + { "glXQueryGLXPbufferSGIX", (GLvoid *) glXQueryGLXPbufferSGIX, NULL }, + { "glXSelectEventSGIX", (GLvoid *) glXSelectEventSGIX, NULL }, + { "glXGetSelectedEventSGIX", (GLvoid *) glXGetSelectedEventSGIX, NULL }, /*** GLX_SGI_cushion ***/ - { "glXCushionSGI", (GLvoid *) glXCushionSGI }, + { "glXCushionSGI", (GLvoid *) glXCushionSGI, NULL }, /*** GLX_SGIX_video_resize ***/ - { "glXBindChannelToWindowSGIX", (GLvoid *) glXBindChannelToWindowSGIX }, - { "glXChannelRectSGIX", (GLvoid *) glXChannelRectSGIX }, - { "glXQueryChannelRectSGIX", (GLvoid *) glXQueryChannelRectSGIX }, - { "glXQueryChannelDeltasSGIX", (GLvoid *) glXQueryChannelDeltasSGIX }, - { "glXChannelRectSyncSGIX", (GLvoid *) glXChannelRectSyncSGIX }, + { "glXBindChannelToWindowSGIX", (GLvoid *) glXBindChannelToWindowSGIX, NULL }, + { "glXChannelRectSGIX", (GLvoid *) glXChannelRectSGIX, NULL }, + { "glXQueryChannelRectSGIX", (GLvoid *) glXQueryChannelRectSGIX, NULL }, + { "glXQueryChannelDeltasSGIX", (GLvoid *) glXQueryChannelDeltasSGIX, NULL }, + { "glXChannelRectSyncSGIX", (GLvoid *) glXChannelRectSyncSGIX, NULL }, /*** GLX_SGIX_dmbuffer **/ #if defined(_DM_BUFFER_H_) - { "glXAssociateDMPbufferSGIX", (GLvoid *) glXAssociateDMPbufferSGIX }, + { "glXAssociateDMPbufferSGIX", (GLvoid *) glXAssociateDMPbufferSGIX, NULL }, #endif /*** GLX_SGIX_swap_group ***/ - { "glXJoinSwapGroupSGIX", (GLvoid *) glXJoinSwapGroupSGIX }, + { "glXJoinSwapGroupSGIX", (GLvoid *) glXJoinSwapGroupSGIX, NULL }, /*** GLX_SGIX_swap_barrier ***/ - { "glXBindSwapBarrierSGIX", (GLvoid *) glXBindSwapBarrierSGIX }, - { "glXQueryMaxSwapBarriersSGIX", (GLvoid *) glXQueryMaxSwapBarriersSGIX }, + { "glXBindSwapBarrierSGIX", (GLvoid *) glXBindSwapBarrierSGIX, NULL }, + { "glXQueryMaxSwapBarriersSGIX", (GLvoid *) glXQueryMaxSwapBarriersSGIX, NULL }, /*** GLX_SUN_get_transparent_index ***/ - { "glXGetTransparentIndexSUN", (GLvoid *) glXGetTransparentIndexSUN }, + { "glXGetTransparentIndexSUN", (GLvoid *) glXGetTransparentIndexSUN, NULL }, /*** GLX_MESA_copy_sub_buffer ***/ - { "glXCopySubBufferMESA", (GLvoid *) glXCopySubBufferMESA }, + { "glXCopySubBufferMESA", (GLvoid *) glXCopySubBufferMESA, NULL }, /*** GLX_MESA_pixmap_colormap ***/ - { "glXCreateGLXPixmapMESA", (GLvoid *) glXCreateGLXPixmapMESA }, + { "glXCreateGLXPixmapMESA", (GLvoid *) glXCreateGLXPixmapMESA, NULL }, /*** GLX_MESA_release_buffers ***/ - { "glXReleaseBuffersMESA", (GLvoid *) glXReleaseBuffersMESA }, + { "glXReleaseBuffersMESA", (GLvoid *) glXReleaseBuffersMESA, NULL }, /*** GLX_MESA_set_3dfx_mode ***/ - { "glXSet3DfxModeMESA", (GLvoid *) glXSet3DfxModeMESA }, + { "glXSet3DfxModeMESA", (GLvoid *) glXSet3DfxModeMESA, NULL }, /*** GLX_ARB_get_proc_address ***/ - { "glXGetProcAddressARB", (GLvoid *) glXGetProcAddressARB }, + { "glXGetProcAddressARB", (GLvoid *) glXGetProcAddressARB, NULL }, - { NULL, NULL } /* end of list */ + /*** GLX 1.4 ***/ + { "glXGetProcAddress", (GLvoid *) glXGetProcAddress, NULL }, + + /*** GLX_???_allocate_memory ***/ + { "glXAllocateMemoryNV", (GLvoid *) glXAllocateMemoryNV, NULL }, + { "glXFreeMemoryNV", (GLvoid *) glXFreeMemoryNV, NULL }, + + { NULL, NULL, NULL } /* end of list */ }; +static struct name_address_pair *Dynamic_GLX_functions = NULL; + + +/* + * Drivers can call this function to append the name of a new GLX + * extension string to __glXGLXClientExtensions. Then, when the user + * calls glXGetClientString() they'll see it listed. + * This is a companion to __glXRegisterGLXFunction(). + */ +void +__glXRegisterGLXExtensionString(const char *extName) +{ + char *newList; + if (!extName) + return; + newList = Xmalloc(strlen(__glXGLXClientExtensions) + + strlen(extName) + 2); /* 2 for ' ' and '\0' */ + if (!newList) + return; + strcpy(newList, __glXGLXClientExtensions); + strcat(newList, " "); + strcat(newList, extName); + if (__glXGLXClientExtensions != __glXGLXDefaultClientExtensions) + Xfree((void *) __glXGLXClientExtensions); + __glXGLXClientExtensions = newList; +} + + +/* + * DRI drivers should call this function if they want to extend + * the GLX API. After registering a new GLX function, the user + * can query and use it by calling glXGetProcAddress(). + * Input: funcName - name of new GLX function + * funcAddr - pointer to the function. + * Return: address of previously registered function with this + * name, or NULL. + */ +void * +__glXRegisterGLXFunction(const char *funcName, void *funcAddr) +{ + struct name_address_pair *ext; + + /* look if the function is already registered */ + for (ext = Dynamic_GLX_functions; ext; ext = ext->Next) { + if (strcmp(ext->Name, funcName) == 0) { + /* It's up the caller to use this return value if he wants + * to chain-call or wrap the previously registered function. + */ + void *prevAddr = ext->Address; + ext->Address = funcAddr; + return prevAddr; + } + } + + /* add new function */ + ext = Xmalloc(sizeof(struct name_address_pair)); + if (!ext) + return NULL; + ext->Name = __glXstrdup(funcName); + if (!ext->Name) { + Xfree(ext); + return NULL; + } + ext->Address = funcAddr; + ext->Next = Dynamic_GLX_functions; + Dynamic_GLX_functions = ext; + return NULL; +} + + static const GLvoid * get_glx_proc_address(const char *funcName) { + const struct name_address_pair *ext; GLuint i; + + /* try dynamic functions */ + for (ext = Dynamic_GLX_functions; ext; ext = ext->Next) { + if (strcmp(ext->Name, funcName) == 0) { + return ext->Address; + } + } + + /* try static functions */ for (i = 0; GLX_functions[i].Name; i++) { if (strcmp(GLX_functions[i].Name, funcName) == 0) return GLX_functions[i].Address; @@ -2189,4 +2295,69 @@ void (*glXGetProcAddressARB(const GLubyte *procName))( void ) f = (gl_function) _glapi_get_proc_address((const char *) procName); return f; } + +/* GLX 1.4 */ +void (*glXGetProcAddress(const GLubyte *procName))( void ) +{ + return glXGetProcAddressARB(procName); +} #endif + + +/* + * AGP memory allocation + */ +void *GLX_PREFIX(glXAllocateMemoryNV)(GLsizei size, + GLfloat readFrequency, + GLfloat writeFrequency, + GLfloat priority) +{ + /* This is special - search the list of dynamically-added functions + * and call the allocator if present. + * More typically, the user will have gotten a pointer to + * glXAllocateMemoryNV() via glXGetProcAddress() so we won't be + * doing this. + */ + typedef void * (*allocFunc)(GLsizei size, GLfloat readFrequency, GLfloat writeFrequency, GLfloat priority); + const struct name_address_pair *ext; + static allocFunc f = (allocFunc) NULL; + + if (!f) { + for (ext = Dynamic_GLX_functions; ext; ext = ext->Next) { + if (strcmp(ext->Name, "glXAllocateMemoryNV") == 0) { + f = (allocFunc) ext->Address; + break; + } + } + } + if (f) + return (*f)(size, readFrequency, writeFrequency, priority); + return NULL; +} + + +void GLX_PREFIX(glXFreeMemoryNV)(GLvoid *pointer) +{ + /* This is special - search the list of dynamically-added functions + * and call the free func if present. + * More typically, the user will have gotten a pointer to + * glXFreeMemoryNV() via glXGetProcAddress() so we won't be + * doing this. + */ + typedef void * (*freeFunc)(GLvoid *pointer); + const struct name_address_pair *ext; + static freeFunc f = (freeFunc) NULL; + + if (!f) { + for (ext = Dynamic_GLX_functions; ext; ext = ext->Next) { + if (strcmp(ext->Name, "glXFreeMemoryNV") == 0) { + f = (freeFunc) ext->Address; + break; + } + } + } + if (f) + (*f)(pointer); +} + + diff --git a/xc/lib/GL/mesa/src/drv/common/mm.c b/xc/lib/GL/mesa/src/drv/common/mm.c index 89a204952..602f6d5be 100644 --- a/xc/lib/GL/mesa/src/drv/common/mm.c +++ b/xc/lib/GL/mesa/src/drv/common/mm.c @@ -29,27 +29,27 @@ #include "mm.h" #include "hwlog.h" -#define ISFREE(bptr) ((bptr)->free) -/* #define PRINTF hwMsg(1, */ -#define PRINTF fprintf(stderr, +/* KW: I don't know who the author of this code is, but it wasn't me + * despite what the copyright says... + */ void mmDumpMemInfo( memHeap_t *heap ) { - TMemBlock *p; - - PRINTF "Memory heap %p:\n", heap); - if (heap == 0) { - PRINTF " heap == 0\n"); - } else { - p = (TMemBlock *)heap; - while (p) { - PRINTF " Offset:%08x, Size:%08x, %c%c\n",p->ofs,p->size, - p->free ? '.':'U', - p->reserved ? 'R':'.'); - p = p->next; - } - } - PRINTF "End of memory blocks\n"); + TMemBlock *p; + + fprintf(stderr, "Memory heap %p:\n", heap); + if (heap == 0) { + fprintf(stderr, " heap == 0\n"); + } else { + p = (TMemBlock *)heap; + while (p) { + fprintf(stderr, " Offset:%08x, Size:%08x, %c%c\n",p->ofs,p->size, + p->free ? '.':'U', + p->reserved ? 'R':'.'); + p = p->next; + } + } + fprintf(stderr, "End of memory blocks\n"); } memHeap_t *mmInit(int ofs, @@ -70,197 +70,131 @@ memHeap_t *mmInit(int ofs, return 0; } -/* Kludgey workaround for existing i810 server. Remove soon. - */ -memHeap_t *mmAddRange( memHeap_t *heap, - int ofs, - int size ) -{ - PMemBlock blocks; - blocks = (TMemBlock *) calloc(2,sizeof(TMemBlock)); - if (blocks) { - blocks[0].size = size; - blocks[0].free = 1; - blocks[0].ofs = ofs; - blocks[0].next = &blocks[1]; - - /* Discontinuity - stops JoinBlock from trying to join non-adjacent - * ranges. - */ - blocks[1].size = 0; - blocks[1].free = 0; - blocks[1].ofs = ofs+size; - blocks[1].next = (PMemBlock) heap; - return (memHeap_t *)blocks; - } - else - return heap; -} static TMemBlock* SliceBlock(TMemBlock *p, int startofs, int size, int reserved, int alignment) { - TMemBlock *newblock; - - /* break left */ - if (startofs > p->ofs) { - newblock = (TMemBlock*) calloc(1,sizeof(TMemBlock)); - if (!newblock) - return NULL; - newblock->ofs = startofs; - newblock->size = p->size - (startofs - p->ofs); - newblock->free = 1; - newblock->next = p->next; - p->size -= newblock->size; - p->next = newblock; - p = newblock; - } + TMemBlock *newblock; + + /* break left */ + if (startofs > p->ofs) { + newblock = (TMemBlock*) calloc(1,sizeof(TMemBlock)); + if (!newblock) + return NULL; + newblock->ofs = startofs; + newblock->size = p->size - (startofs - p->ofs); + newblock->free = 1; + newblock->next = p->next; + p->size -= newblock->size; + p->next = newblock; + p = newblock; + } - /* break right */ - if (size < p->size) { - newblock = (TMemBlock*) calloc(1,sizeof(TMemBlock)); - if (!newblock) - return NULL; - newblock->ofs = startofs + size; - newblock->size = p->size - size; - newblock->free = 1; - newblock->next = p->next; - p->size = size; - p->next = newblock; - } + /* break right */ + if (size < p->size) { + newblock = (TMemBlock*) calloc(1,sizeof(TMemBlock)); + if (!newblock) + return NULL; + newblock->ofs = startofs + size; + newblock->size = p->size - size; + newblock->free = 1; + newblock->next = p->next; + p->size = size; + p->next = newblock; + } - /* p = middle block */ - p->align = alignment; - p->free = 0; - p->reserved = reserved; - return p; + /* p = middle block */ + p->align = alignment; + p->free = 0; + p->reserved = reserved; + return p; } PMemBlock mmAllocMem( memHeap_t *heap, int size, int align2, int startSearch) { - int mask,startofs,endofs; - TMemBlock *p; - - if (!heap || align2 < 0 || size <= 0) - return NULL; - mask = (1 << align2)-1; - startofs = 0; - p = (TMemBlock *)heap; - while (p) { - if (ISFREE(p)) { - startofs = (p->ofs + mask) & ~mask; - if ( startofs < startSearch ) { - startofs = startSearch; + int mask,startofs,endofs; + TMemBlock *p; + + if (!heap || align2 < 0 || size <= 0) + return NULL; + mask = (1 << align2)-1; + startofs = 0; + p = (TMemBlock *)heap; + while (p) { + if ((p)->free) { + startofs = (p->ofs + mask) & ~mask; + if ( startofs < startSearch ) { + startofs = startSearch; + } + endofs = startofs+size; + if (endofs <= (p->ofs+p->size)) + break; } - endofs = startofs+size; - if (endofs <= (p->ofs+p->size)) - break; - } - p = p->next; - } - if (!p) - return NULL; - p = SliceBlock(p,startofs,size,0,mask+1); - p->heap = heap; - return p; + p = p->next; + } + if (!p) + return NULL; + p = SliceBlock(p,startofs,size,0,mask+1); + p->heap = heap; + return p; } static __inline__ int Join2Blocks(TMemBlock *p) { - if (p->free && p->next && p->next->free) { - TMemBlock *q = p->next; - p->size += q->size; - p->next = q->next; - free(q); - return 1; - } - return 0; + if (p->free && p->next && p->next->free) { + TMemBlock *q = p->next; + p->size += q->size; + p->next = q->next; + free(q); + return 1; + } + return 0; } int mmFreeMem(PMemBlock b) { - TMemBlock *p,*prev; - - if (!b) - return 0; - if (!b->heap) { - fprintf(stderr, "no heap\n"); - return -1; - } - p = b->heap; - prev = NULL; - while (p && p != b) { - prev = p; - p = p->next; - } - if (!p || p->free || p->reserved) { - if (!p) - fprintf(stderr, "block not found in heap\n"); - else if (p->free) - fprintf(stderr, "block already free\n"); - else - fprintf(stderr, "block is reserved\n"); - return -1; - } - p->free = 1; - Join2Blocks(p); - if (prev) - Join2Blocks(prev); - return 0; -} - -int mmReserveMem(memHeap_t *heap, int offset,int size) -{ - int endofs; - TMemBlock *p; + TMemBlock *p,*prev; - if (!heap || size <= 0) - return -1; - endofs = offset+size; - p = (TMemBlock *)heap; - while (p && p->ofs <= offset) { - if (ISFREE(p) && endofs <= (p->ofs+p->size)) { - SliceBlock(p,offset,size,1,1); + if (!b) return 0; - } - p = p->next; - } - return -1; + if (!b->heap) { + fprintf(stderr, "no heap\n"); + return -1; + } + p = b->heap; + prev = NULL; + while (p && p != b) { + prev = p; + p = p->next; + } + if (!p || p->free || p->reserved) { + if (!p) + fprintf(stderr, "block not found in heap\n"); + else if (p->free) + fprintf(stderr, "block already free\n"); + else + fprintf(stderr, "block is reserved\n"); + return -1; + } + p->free = 1; + Join2Blocks(p); + if (prev) + Join2Blocks(prev); + return 0; } -int mmFreeReserved(memHeap_t *heap, int offset) -{ - TMemBlock *p,*prev; - - if (!heap) - return -1; - p = (TMemBlock *)heap; - prev = NULL; - while (p && p->ofs != offset) { - prev = p; - p = p->next; - } - if (!p || !p->reserved) - return -1; - p->free = 1; - p->reserved = 0; - Join2Blocks(p); - if (prev) - Join2Blocks(prev); - return 0; -} void mmDestroy(memHeap_t *heap) { - TMemBlock *p,*q; - - if (!heap) - return; - p = (TMemBlock *)heap; - while (p) { - q = p->next; - free(p); - p = q; - } + TMemBlock *p,*q; + + if (!heap) + return; + p = (TMemBlock *)heap; + while (p) { + q = p->next; + free(p); + p = q; + } } diff --git a/xc/lib/GL/mesa/src/drv/common/mm.h b/xc/lib/GL/mesa/src/drv/common/mm.h index 6fa6e8b69..d52871d39 100644 --- a/xc/lib/GL/mesa/src/drv/common/mm.h +++ b/xc/lib/GL/mesa/src/drv/common/mm.h @@ -45,22 +45,12 @@ static __inline__ int mmBlockSize(PMemBlock b) static __inline__ int mmOffset(PMemBlock b) { return b->ofs; } -static __inline__ void mmMarkReserved(PMemBlock b) -{ b->reserved = 1; } - /* * input: total size in bytes * return: a heap pointer if OK, NULL if error */ memHeap_t *mmInit( int ofs, int size ); - - -memHeap_t *mmAddRange( memHeap_t *heap, - int ofs, - int size ); - - /* * Allocate 'size' bytes with 2^align2 bytes alignment, * restrict the search to free memory after 'startSearch' @@ -71,7 +61,8 @@ memHeap_t *mmAddRange( memHeap_t *heap, * startSearch = linear offset from start of heap to begin search * return: pointer to the allocated block, 0 if error */ -PMemBlock mmAllocMem( memHeap_t *heap, int size, int align2, int startSearch ); +PMemBlock mmAllocMem( memHeap_t *heap, int size, int align2, + int startSearch ); /* * Free block starts at offset @@ -81,16 +72,6 @@ PMemBlock mmAllocMem( memHeap_t *heap, int size, int align2, int startSearch ); int mmFreeMem( PMemBlock b ); /* - * Reserve 'size' bytes block start at offset - * This is used to prevent allocation of memory already used - * by the X server for the front buffer, pixmaps, and cursor - * input: size, offset - * output: 0 if OK, -1 if error - */ -int mmReserveMem( memHeap_t *heap, int offset,int size ); -int mmFreeReserved( memHeap_t *heap, int offset ); - -/* * destroy MM */ void mmDestroy( memHeap_t *mmInit ); diff --git a/xc/lib/GL/mesa/src/drv/common/shared_texture_lru.c b/xc/lib/GL/mesa/src/drv/common/shared_texture_lru.c deleted file mode 100644 index 2fd80eb3e..000000000 --- a/xc/lib/GL/mesa/src/drv/common/shared_texture_lru.c +++ /dev/null @@ -1,88 +0,0 @@ -#include <stdio.h> -#include "shared_texture_lru.h" - -/* (Re)initialize the global circular LRU list. The last element - * in the array (heap->nrRegions) is the sentinal. Keeping it - * at the end of the array allows the other elements of the array - * to be addressed rationally when looking up objects at a - * particular location in texture memory. - */ -static void resetGlobalLRU( driHeapPtr heap ) -{ - driTexRegion *list = heap->shared->list; - int sz = 1 << heap->logGranularity; - int i; - - heap->localAge = ++heap->shared->texAge; - - for (i = 0 ; (i+1) * sz <= heap->size ; i++) { - list[i].prev = i-1; - list[i].next = i+1; - list[i].age = heap->shared->texAge; - } - - i--; - list[0].prev = heap->nrRegions; - list[i].prev = i-1; - list[i].next = heap->nrRegions; - list[heap->nrRegions].prev = i; - list[heap->nrRegions].next = 0; -} - -/* Called by the client whenever it touches a local texture. - */ -void driUpdateHeap( driHeapPtr heap, int start, int end ) -{ - driTexRegion *list = heap->shared->list; - int i; - - heap->localAge = ++heap->shared->globalAge; - - for (i = start ; i <= end ; i++) - { - list[i].in_use = 1; - list[i].age = heap->localAge; - - /* remove_from_list(i) - */ - list[(unsigned)list[i].next].prev = list[i].prev; - list[(unsigned)list[i].prev].next = list[i].next; - - /* insert_at_head(list, i) - */ - list[i].prev = heap->nrRegions; - list[i].next = list[heap->nrRegions].next; - list[(unsigned)list[heap->nrRegions].next].prev = i; - list[heap->nrRegions].next = i; - } -} - - -/* Called by the client on lock contention to determine whether - * textures have been stolen - */ -void driAgeTextures( driHeapPtr heap ) -{ - driTexRegion *list = heap->shared->list; - int sz = 1 << (heap->logGranularity); - int i, nr = 0; - - /* Have to go right round from the back to ensure stuff ends up - * LRU in the local list... Fix with a cursor pointer. - */ - for (i = list[heap->nrRegions].prev ; - i != heap->nrRegions && nr < heap->nrRegions ; - i = list[i].prev, nr++) - { - if (list[i].age > heap->localAge) - heap->texturesGone( heap->driverContext, heap->heapId, i * sz, sz, 1); - } - - /* Loop or uninitialized heap detected. Reset. - */ - if (nr == heap->nrRegions) { - heap->texturesGone( heap->driverContext, heap->heapId, 0, heap->size, 0); - resetGlobalLRU( heap ); - } -} - diff --git a/xc/lib/GL/mesa/src/drv/common/shared_texture_lru.h b/xc/lib/GL/mesa/src/drv/common/shared_texture_lru.h deleted file mode 100644 index e6cee567d..000000000 --- a/xc/lib/GL/mesa/src/drv/common/shared_texture_lru.h +++ /dev/null @@ -1,76 +0,0 @@ -#ifndef DRI_TEX_HEAP_H -#define DRI_TEX_HEAP_H - - -/* Private struct. - */ -typedef struct { - unsigned char next, prev; /* indices to form a circular LRU */ - unsigned char in_use; /* owned by a client, or free? */ - int age; /* tracked by clients to update local LRU's */ -} driTexRegion; - - -/* This is the global part of the shared texture mechanism. - * - * Do not use this struct directly - declare an equivalent one with a - * larger list[] array, tuned to suit your application. - * - * Your expanded struct should be placed in the driver-specific - * portion of the sarea. - */ -struct { - int globalAge; - driTexRegion list[1]; /* drivers will want to define a larger list */ -} driGlobalList; - - -/* This is the client-private part of the mechanism. - * - * Clients will place one or more of these structs in their driver - * context struct to manage one or more global texture heaps. All - * fields except print_local_lru must be filled in. - */ -struct dri_tex_heap_t { - - int heapId; /* client-supplied identifier */ - void *driverContext; /* pointer to the client's context private */ - int size; /* heap size in bytes */ - int logGranularity; /* log base 2 of size of single heap region */ - int nrRegions; /* number of elements in global list */ - driGlobalList *shared; /* pointer to sarea driGlobalList struct */ - int localAge; /* initialize to zero */ - - /* Callback to the client to let it know a region of texture - * space has changed age. The client must integrate this - * information with its local texture knowledge, in particular - * checking whether any of its own textures have been - * invalidated. - */ - void (*textures_gone)( void *driverContext, - int heapId, - int offset, - int size, - int inUse ); - - /* Optional hook for debugging. - */ - void (*print_local_lru)( void *driverContext, - int heapId ); -} driTexHeap; - - -#define DRI_AGE_TEXTURES( heap ) \ - if ((heap)->localAge > (heap)->shared->globalAge) \ - driAgeTextures( heap ); - - -/* This should be called whenever there has been contention on the - * hardware lock. Clients can shortcircuit this slightly by using - * DRI_AGE_TEXTURES, above. - */ -void driAgeTextures( driTexHeap *heap ); - -#endif - - diff --git a/xc/lib/GL/mesa/src/drv/i830/i830_render.c b/xc/lib/GL/mesa/src/drv/i830/i830_render.c index 17ad9da59..8acf24f2f 100644 --- a/xc/lib/GL/mesa/src/drv/i830/i830_render.c +++ b/xc/lib/GL/mesa/src/drv/i830/i830_render.c @@ -1,4 +1,3 @@ -/* $Id: i830_render.c,v 1.3 2002/07/05 08:31:06 alanh Exp $ */ /* * Intel i810 DRI driver for Mesa 3.5 diff --git a/xc/lib/GL/mesa/src/drv/r200/Imakefile b/xc/lib/GL/mesa/src/drv/r200/Imakefile new file mode 100644 index 000000000..2ca7cc06b --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/Imakefile @@ -0,0 +1,79 @@ +XCOMM $XFree86: xc/lib/GL/mesa/src/drv/r200/Imakefile,v 1.9 2002/02/23 00:45:50 dawes Exp $ + +#include <Threads.tmpl> + +#if GlxUseBuiltInDRIDriver +#define DoNormalLib (NormalLibGlx || SharedLibGlxWithoutPIC) +#define DoSharedLib (SharedLibGlx && !SharedLibGlxWithoutPIC) +#define DoExtraLib SharedLibGlx +#define DoDebugLib DebugLibGlx +#define DoProfileLib ProfileLibGlx +#else +#define DoNormalLib SharedLibGlxWithoutPIC +#define DoSharedLib !SharedLibGlxWithoutPIC +#define DoExtrasLib NO +#define DoDebugLib NO +#define DoProfileLib NO +#endif + +#include "../common/Imakefile.inc" +#include "Imakefile.inc" +#include "../../array_cache/Imakefile.inc" +#include "../../math/Imakefile.inc" +#include "../../swrast/Imakefile.inc" +#include "../../swrast_setup/Imakefile.inc" +#include "../../tnl/Imakefile.inc" +#include "../../tnl_dd/Imakefile.inc" +#include "../../Imakefile.inc" +#ifdef i386Architecture +#include "../../X86/Imakefile.inc" +#endif +#ifdef SparcArchitecture +#include "../../SPARC/Imakefile.inc" +#endif + + DEFINES = $(ALLOC_DEFINES) $(DRI_DEFINES) $(MESA_ASM_DEFINES) + INCLUDES = $(X_INCLUDES) $(MESA_INCLUDES) $(DRI_INCLUDES) + + DRIOBJS = $(GLXLIBSRC)/dri/dri_util.o + + DRMOBJS = $(GLXLIBSRC)/dri/drm/xf86drm.o \ + $(GLXLIBSRC)/dri/drm/xf86drmHash.o \ + $(GLXLIBSRC)/dri/drm/xf86drmRandom.o \ + $(GLXLIBSRC)/dri/drm/xf86drmSL.o + +#ifdef GlxSoProf + LOSRCS = ../../../../lowpc.c + HISRCS = ../../../../highpc.c + + LOOBJS = ../../../../lowpc.o + HIOBJS = ../../../../highpc.o +#endif + + SRCS = $(R200SRCS) + OBJS = $(LOOBJS) $(DRIOBJS) $(DRMOBJS) $(COREMESAOBJS) \ + $(MESA_ASM_OBJS) $(COMMONOBJS) $(R200OBJS) $(HIOBJS) + +REQUIREDLIBS = MathLibrary $(LDPRELIB) $(GLXLIB) + +#include <Library.tmpl> + +LibraryObjectRule() + +SubdirLibraryRule($(R200OBJS)) +NormalLintTarget($(SRCS)) + +#if !GlxUseBuiltInDRIDriver +LIBNAME = r200_dri.so +SharedDriModuleTarget($(LIBNAME),DONE $(OBJS),$(OBJS)) +InstallDynamicModule($(LIBNAME),$(MODULEDIR),dri) + +#ifdef GlxSoProf +SOPROF_LIBNAME = _r200_dri_p +NormalDepLibraryTarget($(SOPROF_LIBNAME),DONE $(OBJS),$(OBJS)) +InstallLibrary($(SOPROF_LIBNAME),$(MODULEDIR)/dri) +#endif +#endif + +DependTarget() + diff --git a/xc/lib/GL/mesa/src/drv/r200/Imakefile.inc b/xc/lib/GL/mesa/src/drv/r200/Imakefile.inc new file mode 100644 index 000000000..41236bc0f --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/Imakefile.inc @@ -0,0 +1,158 @@ +XCOMM $XFree86: xc/lib/GL/mesa/src/drv/r200/Imakefile.inc,v 1.2 2001/05/02 15:06:04 dawes Exp $ + +#ifndef MesaDrvSrcDir +#define MesaDrvSrcDir $(GLXLIBSRC)/mesa/src/drv +#endif + +MESADRVSRCDIR = MesaDrvSrcDir + +#ifndef MesaDrvR200BuildDir +#define MesaDrvR200BuildDir /**/ +#endif +MESADRVR200BUILDDIR = MesaDrvR200BuildDir + +#if Malloc0ReturnsNull +ALLOC_DEFINES = -DMALLOC_0_RETURNS_NULL +#endif + +#if BuildXF86DRI + DRI_DEFINES = GlxDefines -DX_BYTE_ORDER=ByteOrder + DRI_INCLUDES = -I$(GLXLIBSRC)/dri \ + -I$(GLXLIBSRC)/glx \ + -I$(INCLUDESRC) \ + -I$(INCLUDESRC)/GL \ + -I$(SERVERSRC)/GL/dri \ + -I$(XF86OSSRC) \ + -I$(XF86DRIVERSRC)/ati \ + -I$(XF86COMSRC) \ + -I$(GLXLIBSRC)/dri/drm \ + -I$(GLXLIBSRC)/include \ + -I$(XTOP)/include +#endif + +MESA_INCLUDES = -I$(MESASRCDIR)/src \ + -I$(MESADRVSRCDIR)/common \ + -I$(MESADRVSRCDIR)/r200 + X_INCLUDES = -I$(XINCLUDESRC) -I$(EXTINCSRC) + + R200SRCS = $(MESADRVR200BUILDDIR)r200_context.c \ + $(MESADRVR200BUILDDIR)r200_sanity.c \ + $(MESADRVR200BUILDDIR)r200_state.c \ + $(MESADRVR200BUILDDIR)r200_state_init.c \ + $(MESADRVR200BUILDDIR)r200_swtcl.c \ + $(MESADRVR200BUILDDIR)r200_ioctl.c \ + $(MESADRVR200BUILDDIR)r200_lock.c \ + $(MESADRVR200BUILDDIR)r200_maos.c \ + $(MESADRVR200BUILDDIR)r200_screen.c \ + $(MESADRVR200BUILDDIR)r200_span.c \ + $(MESADRVR200BUILDDIR)r200_tcl.c \ + $(MESADRVR200BUILDDIR)r200_tex.c \ + $(MESADRVR200BUILDDIR)r200_texmem.c \ + $(MESADRVR200BUILDDIR)r200_texstate.c \ + $(MESADRVR200BUILDDIR)r200_vtxfmt.c \ + $(MESADRVR200BUILDDIR)r200_vtxfmt_x86.c \ + $(MESADRVR200BUILDDIR)r200_vtxtmp_x86.S \ + $(MESADRVR200BUILDDIR)r200_vtxfmt_sse.c \ + $(MESADRVR200BUILDDIR)r200_vtxfmt_c.c + + R200OBJS = $(MESADRVR200BUILDDIR)r200_context.o \ + $(MESADRVR200BUILDDIR)r200_sanity.o \ + $(MESADRVR200BUILDDIR)r200_state.o \ + $(MESADRVR200BUILDDIR)r200_state_init.o \ + $(MESADRVR200BUILDDIR)r200_swtcl.o \ + $(MESADRVR200BUILDDIR)r200_ioctl.o \ + $(MESADRVR200BUILDDIR)r200_lock.o \ + $(MESADRVR200BUILDDIR)r200_maos.o \ + $(MESADRVR200BUILDDIR)r200_screen.o \ + $(MESADRVR200BUILDDIR)r200_span.o \ + $(MESADRVR200BUILDDIR)r200_tcl.o \ + $(MESADRVR200BUILDDIR)r200_tex.o \ + $(MESADRVR200BUILDDIR)r200_texmem.o \ + $(MESADRVR200BUILDDIR)r200_texstate.o \ + $(MESADRVR200BUILDDIR)r200_vtxfmt.o \ + $(MESADRVR200BUILDDIR)r200_vtxfmt_c.o \ + $(MESADRVR200BUILDDIR)r200_vtxfmt_x86.o \ + $(MESADRVR200BUILDDIR)r200_vtxtmp_x86.o \ + $(MESADRVR200BUILDDIR)r200_vtxfmt_sse.o + + R200UOBJS = $(MESADRVR200BUILDDIR)unshared/r200_context.o \ + $(MESADRVR200BUILDDIR)unshared/r200_sanity.o \ + $(MESADRVR200BUILDDIR)unshared/r200_ioctl.o \ + $(MESADRVR200BUILDDIR)unshared/r200_lock.o \ + $(MESADRVR200BUILDDIR)unshared/r200_maos.o \ + $(MESADRVR200BUILDDIR)unshared/r200_screen.o \ + $(MESADRVR200BUILDDIR)unshared/r200_span.o \ + $(MESADRVR200BUILDDIR)unshared/r200_state.o \ + $(MESADRVR200BUILDDIR)unshared/r200_state_init.o \ + $(MESADRVR200BUILDDIR)unshared/r200_swtcl.o \ + $(MESADRVR200BUILDDIR)unshared/r200_tcl.o \ + $(MESADRVR200BUILDDIR)unshared/r200_tex.o \ + $(MESADRVR200BUILDDIR)unshared/r200_texmem.o \ + $(MESADRVR200BUILDDIR)unshared/r200_texstate.o \ + $(MESADRVR200BUILDDIR)unshared/r200_vtxfmt.o \ + $(MESADRVR200BUILDDIR)unshared/r200_vtxfmt_c.o \ + $(MESADRVR200BUILDDIR)unshared/r200_vtxfmt_x86.o \ + $(MESADRVR200BUILDDIR)unshared/r200_vtxtmp_x86.o \ + $(MESADRVR200BUILDDIR)unshared/r200_vtxfmt_sse.o + + R200DOBJS = $(MESADRVR200BUILDDIR)debugger/r200_context.o \ + $(MESADRVR200BUILDDIR)debugger/r200_sanity.o \ + $(MESADRVR200BUILDDIR)debugger/r200_ioctl.o \ + $(MESADRVR200BUILDDIR)debugger/r200_lock.o \ + $(MESADRVR200BUILDDIR)debugger/r200_maos.o \ + $(MESADRVR200BUILDDIR)debugger/r200_screen.o \ + $(MESADRVR200BUILDDIR)debugger/r200_span.o \ + $(MESADRVR200BUILDDIR)debugger/r200_state.o \ + $(MESADRVR200BUILDDIR)debugger/r200_state_init.o \ + $(MESADRVR200BUILDDIR)debugger/r200_swtcl.o \ + $(MESADRVR200BUILDDIR)debugger/r200_tcl.o \ + $(MESADRVR200BUILDDIR)debugger/r200_tex.o \ + $(MESADRVR200BUILDDIR)debugger/r200_texmem.o \ + $(MESADRVR200BUILDDIR)debugger/r200_texstate.o \ + $(MESADRVR200BUILDDIR)debugger/r200_vtxfmt.o \ + $(MESADRVR200BUILDDIR)debugger/r200_vtxfmt_c.o \ + $(MESADRVR200BUILDDIR)debugger/r200_vtxfmt_x86.o \ + $(MESADRVR200BUILDDIR)debugger/r200_vtxtmp_x86.o \ + $(MESADRVR200BUILDDIR)debugger/r200_vtxfmt_sse.o + + R200POBJS = $(MESADRVR200BUILDDIR)profiled/r200_context.o \ + $(MESADRVR200BUILDDIR)profiled/r200_sanity.o \ + $(MESADRVR200BUILDDIR)profiled/r200_ioctl.o \ + $(MESADRVR200BUILDDIR)profiled/r200_lock.o \ + $(MESADRVR200BUILDDIR)profiled/r200_maos.o \ + $(MESADRVR200BUILDDIR)profiled/r200_screen.o \ + $(MESADRVR200BUILDDIR)profiled/r200_span.o \ + $(MESADRVR200BUILDDIR)profiled/r200_state.o \ + $(MESADRVR200BUILDDIR)profiled/r200_state_init.o \ + $(MESADRVR200BUILDDIR)profiled/r200_swtcl.o \ + $(MESADRVR200BUILDDIR)profiled/r200_tcl.o \ + $(MESADRVR200BUILDDIR)profiled/r200_tex.o \ + $(MESADRVR200BUILDDIR)profiled/r200_texmem.o \ + $(MESADRVR200BUILDDIR)profiled/r200_texstate.o \ + $(MESADRVR200BUILDDIR)profiled/r200_vtxfmt.o \ + $(MESADRVR200BUILDDIR)profiled/r200_vtxfmt_c.o \ + $(MESADRVR200BUILDDIR)profiled/r200_vtxfmt_x86.o \ + $(MESADRVR200BUILDDIR)profiled/r200_vtxtmp_x86.o \ + $(MESADRVR200BUILDDIR)profiled/r200_vtxfmt_sse.o + +#ifdef NeedToLinkMesaSrc +LinkSourceFile(r200_context.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_ioctl.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_lock.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_maos.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_screen.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_span.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_sanity.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_state.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_state_init.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_swtcl.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_tcl.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_tex.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_texmem.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_texstate.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_vtxfmt.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_vtxfmt_c.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_vtxfmt_x86.c, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_vtxtmp_x86.S, $(MESADRVSRCDIR)/r200) +LinkSourceFile(r200_vtxfmt_sse.c, $(MESADRVSRCDIR)/r200) +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_context.c b/xc/lib/GL/mesa/src/drv/r200/r200_context.c new file mode 100644 index 000000000..a29f597fa --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_context.c @@ -0,0 +1,759 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include <stdlib.h> +#include <dlfcn.h> + +#include "r200_context.h" +#include "r200_ioctl.h" +#include "r200_state.h" +#include "r200_span.h" +#include "r200_tex.h" +#include "r200_swtcl.h" +#include "r200_tcl.h" +#include "r200_vtxfmt.h" +#include "r200_maos.h" + +#include "swrast/swrast.h" +#include "swrast_setup/swrast_setup.h" +#include "array_cache/acache.h" + +#include "tnl/tnl.h" +#include "tnl/t_pipeline.h" + +#include "attrib.h" +#include "api_arrayelt.h" +#include "context.h" +#include "simple_list.h" +#include "mem.h" +#include "matrix.h" +#include "state.h" +#include "extensions.h" +#include "state.h" +#if defined(USE_X86_ASM) +#include "X86/common_x86_asm.h" +#endif + +#define R200_DATE "20020611" + +#ifndef R200_DEBUG +int R200_DEBUG = (0); +#endif + + + +/* Return the width and height of the given buffer. + */ +static void r200GetBufferSize( GLframebuffer *buffer, + GLuint *width, GLuint *height ) +{ + GET_CURRENT_CONTEXT(ctx); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + LOCK_HARDWARE( rmesa ); + *width = rmesa->dri.drawable->w; + *height = rmesa->dri.drawable->h; + UNLOCK_HARDWARE( rmesa ); +} + +/* Return various strings for glGetString(). + */ +static const GLubyte *r200GetString( GLcontext *ctx, GLenum name ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + static char buffer[128]; + + switch ( name ) { + case GL_VENDOR: + return (GLubyte *)"Tungsten Graphics, Inc."; + + case GL_RENDERER: + sprintf( buffer, "Mesa DRI R200 " R200_DATE); + + /* Append any chipset-specific information. None yet. + */ + + /* Append any AGP-specific information. + */ + switch ( rmesa->r200Screen->AGPMode ) { + case 1: + strncat( buffer, " AGP 1x", 7 ); + break; + case 2: + strncat( buffer, " AGP 2x", 7 ); + break; + case 4: + strncat( buffer, " AGP 4x", 7 ); + break; + } + + /* Append any CPU-specific information. + */ +#ifdef USE_X86_ASM + if ( _mesa_x86_cpu_features ) { + strncat( buffer, " x86", 4 ); + } +#ifdef USE_MMX_ASM + if ( cpu_has_mmx ) { + strncat( buffer, "/MMX", 4 ); + } +#endif +#ifdef USE_3DNOW_ASM + if ( cpu_has_3dnow ) { + strncat( buffer, "/3DNow!", 7 ); + } +#endif +#ifdef USE_SSE_ASM + if ( cpu_has_xmm ) { + strncat( buffer, "/SSE", 4 ); + } +#endif +#endif + + if ( rmesa->dri.drmMinor < R200_DRM_CURRENT ) { + strncat( buffer, " DRM-COMPAT", 11 ); + } + + if ( !(rmesa->TclFallback & R200_TCL_FALLBACK_TCL_DISABLE) ) { + strncat( buffer, " TCL", 4 ); + } + else { + strncat( buffer, " NO-TCL", 7 ); + } + + return (GLubyte *)buffer; + + default: + return NULL; + } +} + + + +/* Initialize the extensions supported by this driver. + */ +static void r200InitExtensions( GLcontext *ctx ) +{ + _mesa_enable_imaging_extensions( ctx ); + + _mesa_enable_extension( ctx, "GL_ARB_multitexture" ); + _mesa_enable_extension( ctx, "GL_ARB_texture_env_add" ); + _mesa_enable_extension( ctx, "GL_ARB_texture_env_combine" ); + _mesa_enable_extension( ctx, "GL_ARB_texture_env_dot3" ); + + _mesa_enable_extension( ctx, "GL_EXT_blend_logic_op" ); + _mesa_enable_extension( ctx, "GL_EXT_texture_env_add" ); + _mesa_enable_extension( ctx, "GL_EXT_texture_env_combine" ); + _mesa_enable_extension( ctx, "GL_EXT_texture_env_dot3" ); + _mesa_enable_extension( ctx, "GL_EXT_texture_filter_anisotropic" ); + _mesa_enable_extension( ctx, "GL_EXT_texture_lod_bias" ); + _mesa_enable_extension( ctx, "GL_EXT_secondary_color" ); + _mesa_enable_extension( ctx, "GL_EXT_blend_subtract" ); + _mesa_enable_extension( ctx, "GL_EXT_blend_minmax" ); + +/* _mesa_enable_extension( ctx, "GL_EXT_fog_coord" ); */ +} + +extern const struct gl_pipeline_stage _r200_render_stage; +extern const struct gl_pipeline_stage _r200_tcl_stage; + +static const struct gl_pipeline_stage *r200_pipeline[] = { + + /* Try and go straight to t&l + */ + &_r200_tcl_stage, + + /* Catch any t&l fallbacks + */ + &_tnl_vertex_transform_stage, + &_tnl_normal_transform_stage, + &_tnl_lighting_stage, + &_tnl_fog_coordinate_stage, + &_tnl_texgen_stage, + &_tnl_texture_transform_stage, + + /* Try again to go to tcl? + * - no good for asymmetric-twoside (do with multipass) + * - no good for asymmetric-unfilled (do with multipass) + * - good for material + * - good for texgen + * - need to manipulate a bit of state + * + * - worth it/not worth it? + */ + + /* Else do them here. + */ + &_r200_render_stage, + &_tnl_render_stage, /* FALLBACK: */ + 0, +}; + + + +/* Initialize the driver's misc functions. + */ +static void r200InitDriverFuncs( GLcontext *ctx ) +{ + ctx->Driver.GetBufferSize = r200GetBufferSize; + ctx->Driver.ResizeBuffers = _swrast_alloc_buffers; + ctx->Driver.GetString = r200GetString; + + ctx->Driver.Error = NULL; + ctx->Driver.DrawPixels = NULL; + ctx->Driver.Bitmap = NULL; +} + +static void add_debug_flags( const char *debug ) +{ + if (strstr(debug, "fall")) + R200_DEBUG |= DEBUG_FALLBACKS; + + if (strstr(debug, "tex")) + R200_DEBUG |= DEBUG_TEXTURE; + + if (strstr(debug, "ioctl")) + R200_DEBUG |= DEBUG_IOCTL; + + if (strstr(debug, "prim")) + R200_DEBUG |= DEBUG_PRIMS; + + if (strstr(debug, "vert")) + R200_DEBUG |= DEBUG_VERTS; + + if (strstr(debug, "state")) + R200_DEBUG |= DEBUG_STATE; + + if (strstr(debug, "code")) + R200_DEBUG |= DEBUG_CODEGEN; + + if (strstr(debug, "vfmt") || strstr(debug, "vtxf")) + R200_DEBUG |= DEBUG_VFMT; + + if (strstr(debug, "verb")) + R200_DEBUG |= DEBUG_VERBOSE; + + if (strstr(debug, "dri")) + R200_DEBUG |= DEBUG_DRI; + + if (strstr(debug, "dma")) + R200_DEBUG |= DEBUG_DMA; + + if (strstr(debug, "san")) + R200_DEBUG |= DEBUG_SANITY; + + if (strstr(debug, "sync")) + R200_DEBUG |= DEBUG_SYNC; +} + +/* Create the device specific context. + */ +static GLboolean +r200CreateContext( Display *dpy, const __GLcontextModes *glVisual, + __DRIcontextPrivate *driContextPriv, + void *sharedContextPrivate) +{ + __DRIscreenPrivate *sPriv = driContextPriv->driScreenPriv; + r200ScreenPtr r200Screen = (r200ScreenPtr)(sPriv->private); + r200ContextPtr rmesa; + GLcontext *ctx, *shareCtx; + int i; + + assert(dpy); + assert(glVisual); + assert(driContextPriv); + assert(r200Screen); + + /* Allocate the R200 context */ + rmesa = (r200ContextPtr) CALLOC( sizeof(*rmesa) ); + if ( !rmesa ) + return GL_FALSE; + + /* Allocate the Mesa context */ + if (sharedContextPrivate) + shareCtx = ((r200ContextPtr) sharedContextPrivate)->glCtx; + else + shareCtx = NULL; + rmesa->glCtx = _mesa_create_context(glVisual, shareCtx, rmesa, GL_TRUE); + if (!rmesa->glCtx) { + FREE(rmesa); + return GL_FALSE; + } + driContextPriv->driverPrivate = rmesa; + + /* Init r200 context data */ + rmesa->dri.display = dpy; + rmesa->dri.context = driContextPriv; + rmesa->dri.screen = sPriv; + rmesa->dri.drawable = NULL; /* Set by XMesaMakeCurrent */ + rmesa->dri.hwContext = driContextPriv->hHWContext; + rmesa->dri.hwLock = &sPriv->pSAREA->lock; + rmesa->dri.fd = sPriv->fd; + + /* If we don't have 1.3, fallback to the 1.1 interfaces. + */ + if (getenv("R200_COMPAT") || sPriv->drmMinor < R200_DRM_CURRENT) + rmesa->dri.drmMinor = 5; + else + rmesa->dri.drmMinor = sPriv->drmMinor; + + rmesa->r200Screen = r200Screen; + rmesa->sarea = (RADEONSAREAPrivPtr)((GLubyte *)sPriv->pSAREA + + r200Screen->sarea_priv_offset); + + + rmesa->dma.buf0_address = rmesa->r200Screen->buffers->list[0].address; + + for ( i = 0 ; i < r200Screen->numTexHeaps ; i++ ) { + make_empty_list( &rmesa->texture.objects[i] ); + rmesa->texture.heap[i] = mmInit( 0, r200Screen->texSize[i] ); + rmesa->texture.age[i] = -1; + } + rmesa->texture.numHeaps = r200Screen->numTexHeaps; + make_empty_list( &rmesa->texture.swapped ); + + rmesa->swtcl.RenderIndex = ~0; + rmesa->lost_context = 1; + + /* KW: Set the maximum texture size small enough that we can + * guarentee that both texture units can bind a maximal texture + * and have them both in on-card memory at once. + * Test for 2 textures * 4 bytes/texel * size * size. + */ + ctx = rmesa->glCtx; + if (r200Screen->texSize[RADEON_CARD_HEAP] >= 2 * 4 * 2048 * 2048) { + ctx->Const.MaxTextureLevels = 12; /* 2048x2048 */ + } + else if (r200Screen->texSize[RADEON_CARD_HEAP] >= 2 * 4 * 1024 * 1024) { + ctx->Const.MaxTextureLevels = 11; /* 1024x1024 */ + } + else if (r200Screen->texSize[RADEON_CARD_HEAP] >= 2 * 4 * 512 * 512) { + ctx->Const.MaxTextureLevels = 10; /* 512x512 */ + } + else { + ctx->Const.MaxTextureLevels = 9; /* 256x256 */ + } + + ctx->Const.MaxTextureUnits = 2; + ctx->Const.MaxTextureMaxAnisotropy = 16.0; + + /* No wide points. + */ + ctx->Const.MinPointSize = 1.0; + ctx->Const.MinPointSizeAA = 1.0; + ctx->Const.MaxPointSize = 1.0; + ctx->Const.MaxPointSizeAA = 1.0; + + ctx->Const.MinLineWidth = 1.0; + ctx->Const.MinLineWidthAA = 1.0; + ctx->Const.MaxLineWidth = 10.0; + ctx->Const.MaxLineWidthAA = 10.0; + ctx->Const.LineWidthGranularity = 0.0625; + + + /* Initialize the software rasterizer and helper modules. + */ + _swrast_CreateContext( ctx ); + _ac_CreateContext( ctx ); + _tnl_CreateContext( ctx ); + _swsetup_CreateContext( ctx ); + _ae_create_context( ctx ); + + /* Install the customized pipeline: + */ + _tnl_destroy_pipeline( ctx ); + _tnl_install_pipeline( ctx, r200_pipeline ); + + /* Try and keep materials and vertices separate: + */ + _tnl_isolate_materials( ctx, GL_TRUE ); + + + /* Configure swrast to match hardware characteristics: + */ + _swrast_allow_pixel_fog( ctx, GL_FALSE ); + _swrast_allow_vertex_fog( ctx, GL_TRUE ); + + + _math_matrix_ctr( &rmesa->TexGenMatrix[0] ); + _math_matrix_ctr( &rmesa->TexGenMatrix[1] ); + _math_matrix_ctr( &rmesa->tmpmat ); + _math_matrix_set_identity( &rmesa->TexGenMatrix[0] ); + _math_matrix_set_identity( &rmesa->TexGenMatrix[1] ); + _math_matrix_set_identity( &rmesa->tmpmat ); + + r200InitExtensions( ctx ); + r200InitDriverFuncs( ctx ); + r200InitIoctlFuncs( ctx ); + r200InitStateFuncs( ctx ); + r200InitSpanFuncs( ctx ); + r200InitTextureFuncs( ctx ); + r200InitState( rmesa ); + r200InitSwtcl( ctx ); + + /* Does busy waiting sleep? + */ + rmesa->do_usleeps = !getenv("R200_NO_USLEEPS"); + +#if DO_DEBUG + if (getenv("R200_DEBUG")) + add_debug_flags( getenv("R200_DEBUG") ); + if (getenv("RADEON_DEBUG")) + add_debug_flags( getenv("RADEON_DEBUG") ); +#endif + + if (getenv("R200_NO_RAST")) { + fprintf(stderr, "disabling 3D acceleration\n"); + FALLBACK(rmesa, R200_FALLBACK_DISABLE, 1); + } + else if (getenv("R200_NO_TCL")) { + fprintf(stderr, "disabling TCL support\n"); + TCL_FALLBACK(rmesa->glCtx, R200_TCL_FALLBACK_TCL_DISABLE, 1); + } + else { + if (!getenv("R200_NO_VTXFMT")) { + r200VtxfmtInit( ctx ); + } + _tnl_need_dlist_norm_lengths( ctx, GL_FALSE ); + } + return GL_TRUE; +} + + +/* Destroy the device specific context. + */ +/* Destroy the Mesa and driver specific context data. + */ +static void +r200DestroyContext( __DRIcontextPrivate *driContextPriv ) +{ + GET_CURRENT_CONTEXT(ctx); + r200ContextPtr rmesa = (r200ContextPtr) driContextPriv->driverPrivate; + r200ContextPtr current = ctx ? R200_CONTEXT(ctx) : NULL; + + fprintf(stderr, "%s\n", __FUNCTION__); + + /* check if we're deleting the currently bound context */ + if (rmesa == current) { + _mesa_make_current2(NULL, NULL, NULL); + } + + /* Free r200 context resources */ + assert(rmesa); /* should never be null */ + if ( rmesa ) { + if (rmesa->glCtx->Shared->RefCount == 1) { + /* This share group is about to go away, free our private + * texture object data. + */ + r200TexObjPtr t, next_t; + int i; + + for ( i = 0 ; i < rmesa->texture.numHeaps ; i++ ) { + foreach_s ( t, next_t, &rmesa->texture.objects[i] ) { + r200DestroyTexObj( rmesa, t ); + } + mmDestroy( rmesa->texture.heap[i] ); + } + + foreach_s ( t, next_t, &rmesa->texture.swapped ) { + r200DestroyTexObj( rmesa, t ); + } + } + + _swsetup_DestroyContext( rmesa->glCtx ); + _tnl_DestroyContext( rmesa->glCtx ); + _ac_DestroyContext( rmesa->glCtx ); + _swrast_DestroyContext( rmesa->glCtx ); + + r200DestroySwtcl( rmesa->glCtx ); + + r200ReleaseArrays( rmesa->glCtx, ~0 ); + + if (rmesa->dma.current.buf) + r200ReleaseDmaRegion( rmesa, &rmesa->dma.current, __FUNCTION__ ); + + R200_FIREVERTICES( rmesa ); + + if (!rmesa->TclFallback & R200_TCL_FALLBACK_TCL_DISABLE) + if (!getenv("R200_NO_VTXFMT")) + r200VtxfmtDestroy( rmesa->glCtx ); + + /* free the Mesa context */ + rmesa->glCtx->DriverCtx = NULL; + _mesa_destroy_context( rmesa->glCtx ); + + if (rmesa->state.scissor.pClipRects) { + FREE(rmesa->state.scissor.pClipRects); + rmesa->state.scissor.pClipRects = 0; + } + + FREE( rmesa ); + } + +#if 0 + /* Use this to force shared object profiling. */ + glx_fini_prof(); +#endif +} + + +/* Initialize the driver specific screen private data. + */ +static GLboolean +r200InitDriver( __DRIscreenPrivate *sPriv ) +{ + sPriv->private = (void *) r200CreateScreen( sPriv ); + if ( !sPriv->private ) { + r200DestroyScreen( sPriv ); + return GL_FALSE; + } + + return GL_TRUE; +} + + +/* Create and initialize the Mesa and driver specific pixmap buffer + * data. + */ +static GLboolean +r200CreateBuffer( Display *dpy, + __DRIscreenPrivate *driScrnPriv, + __DRIdrawablePrivate *driDrawPriv, + const __GLcontextModes *mesaVis, + GLboolean isPixmap ) +{ + if (isPixmap) { + return GL_FALSE; /* not implemented */ + } + else { + const GLboolean swDepth = GL_FALSE; + const GLboolean swAlpha = GL_FALSE; + const GLboolean swAccum = mesaVis->accumRedBits > 0; + const GLboolean swStencil = mesaVis->stencilBits > 0 && + mesaVis->depthBits != 24; + driDrawPriv->driverPrivate = (void *) + _mesa_create_framebuffer( mesaVis, + swDepth, + swStencil, + swAccum, + swAlpha ); + return (driDrawPriv->driverPrivate != NULL); + } +} + + +static void +r200DestroyBuffer(__DRIdrawablePrivate *driDrawPriv) +{ + _mesa_destroy_framebuffer((GLframebuffer *) (driDrawPriv->driverPrivate)); +} + + + +static void +r200SwapBuffers(Display *dpy, void *drawablePrivate) +{ + __DRIdrawablePrivate *dPriv = (__DRIdrawablePrivate *) drawablePrivate; + (void) dpy; + + if (dPriv->driContextPriv && dPriv->driContextPriv->driverPrivate) { + r200ContextPtr rmesa; + GLcontext *ctx; + rmesa = (r200ContextPtr) dPriv->driContextPriv->driverPrivate; + ctx = rmesa->glCtx; + if (ctx->Visual.doubleBufferMode) { + _mesa_swapbuffers( ctx ); /* flush pending rendering comands */ + if ( rmesa->doPageFlip ) { + r200PageFlip( dPriv ); + } + else { + r200CopyBuffer( dPriv ); + } + } + } + else { + /* XXX this shouldn't be an error but we can't handle it for now */ + _mesa_problem(NULL, "r200SwapBuffers: drawable has no context!\n"); + } +} + + +/* Force the context `c' to be the current context and associate with it + * buffer `b'. + */ +static GLboolean +r200MakeCurrent( __DRIcontextPrivate *driContextPriv, + __DRIdrawablePrivate *driDrawPriv, + __DRIdrawablePrivate *driReadPriv ) +{ + if ( driContextPriv ) { + r200ContextPtr newR200Ctx = + (r200ContextPtr) driContextPriv->driverPrivate; + + if (R200_DEBUG & DEBUG_DRI) + fprintf(stderr, "%s ctx %p\n", __FUNCTION__, newR200Ctx->glCtx); + + if ( newR200Ctx->dri.drawable != driDrawPriv ) { + newR200Ctx->dri.drawable = driDrawPriv; + r200UpdateWindow( newR200Ctx->glCtx ); + r200UpdateViewportOffset( newR200Ctx->glCtx ); + } + + _mesa_make_current2( newR200Ctx->glCtx, + (GLframebuffer *) driDrawPriv->driverPrivate, + (GLframebuffer *) driReadPriv->driverPrivate ); + + if ( !newR200Ctx->glCtx->Viewport.Width ) { + _mesa_set_viewport( newR200Ctx->glCtx, 0, 0, + driDrawPriv->w, driDrawPriv->h ); + } + + if (newR200Ctx->vb.enabled) + r200VtxfmtMakeCurrent( newR200Ctx->glCtx ); + + _mesa_update_state( newR200Ctx->glCtx ); + r200ValidateState( newR200Ctx->glCtx ); + + } else { + if (R200_DEBUG & DEBUG_DRI) + fprintf(stderr, "%s ctx is null\n", __FUNCTION__); + _mesa_make_current( 0, 0 ); + } + + if (R200_DEBUG & DEBUG_DRI) + fprintf(stderr, "End %s\n", __FUNCTION__); + return GL_TRUE; +} + +/* Force the context `c' to be unbound from its buffer. + */ +static GLboolean +r200UnbindContext( __DRIcontextPrivate *driContextPriv ) +{ + r200ContextPtr rmesa = (r200ContextPtr) driContextPriv->driverPrivate; + + if (R200_DEBUG & DEBUG_DRI) + fprintf(stderr, "%s ctx %p\n", __FUNCTION__, rmesa->glCtx); + + r200VtxfmtUnbindContext( rmesa->glCtx ); + return GL_TRUE; +} + +/* Fullscreen mode isn't used for much -- could be a way to shrink + * front/back buffers & get more texture memory if the client has + * changed the video resolution. + * + * Pageflipping is now done automatically whenever there is a single + * 3d client. + */ +static GLboolean +r200OpenCloseFullScreen( __DRIcontextPrivate *driContextPriv ) +{ + return GL_TRUE; +} + + +static void * +r200AllocateAGP(GLsizei size, GLfloat readfreq, GLfloat writefreq, GLfloat priority) +{ + printf("r200::glXAllocateMemoryNV(%d, %f, %f, %f) not implemented yet.\n", + size, readfreq, writefreq, priority); + return NULL; +} + + +static void +r200FreeAGP(GLvoid *pointer) +{ + printf("r200::glXFreeMemoryNV(%p) not implemented yet.\n", pointer); +} + + +/* This function is called by libGL.so as soon as libGL.so is loaded. + * This is where we'd register new extension functions with the dispatcher. + */ +void +__driRegisterExtensions( void ) +{ + /* dlopen ourself */ + void *dll = dlopen(NULL, RTLD_GLOBAL); + if (dll) { + /* get pointer to libGL's __glXRegisterGLXFunction, if it exists */ + typedef void *(*registerFunc)(const char *funcName, void *funcAddr); + registerFunc reg = (registerFunc) dlsym(dll, "__glXRegisterGLXFunction"); + if (reg) { + /* register our GLX extensions with libGL */ + void *p; + p = reg("glXAllocateMemoryNV", (void *) r200AllocateAGP); + if (p) + ; /* XXX already registered - what to do, wrap? */ + + p = reg("glXFreeMemoryNV", (void *) r200FreeAGP); + if (p) + ; /* XXX already registered - what to do, wrap? */ + } + dlclose(dll); + } +} + + + +static struct __DriverAPIRec r200API = { + r200InitDriver, + r200DestroyScreen, + r200CreateContext, + r200DestroyContext, + r200CreateBuffer, + r200DestroyBuffer, + r200SwapBuffers, + r200MakeCurrent, + r200UnbindContext, + r200OpenCloseFullScreen, + r200OpenCloseFullScreen +}; + + + +/* + * This is the bootstrap function for the driver. + * The __driCreateScreen name is the symbol that libGL.so fetches. + * Return: pointer to a __DRIscreenPrivate. + */ +void *__driCreateScreen(Display *dpy, int scrn, __DRIscreen *psc, + int numConfigs, __GLXvisualConfig *config) +{ + __DRIscreenPrivate *psp; + psp = __driUtilCreateScreen(dpy, scrn, psc, numConfigs, config, &r200API); + return (void *) psp; +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_context.h b/xc/lib/GL/mesa/src/drv/r200/r200_context.h new file mode 100644 index 000000000..467f803e5 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_context.h @@ -0,0 +1,891 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_CONTEXT_H__ +#define __R200_CONTEXT_H__ + +#ifdef GLX_DIRECT_RENDERING + +#include <X11/Xlibint.h> +#include "dri_util.h" +#include "xf86drm.h" +#include "radeon_common.h" + +#include "macros.h" +#include "mtypes.h" +#include "r200_reg.h" + +struct r200_context; +typedef struct r200_context r200ContextRec; +typedef struct r200_context *r200ContextPtr; + +#include "r200_lock.h" +#include "r200_screen.h" +#include "mm.h" + +/* Flags for software fallback cases */ +/* See correponding strings in r200_swtcl.c */ +#define R200_FALLBACK_TEXTURE 0x1 +#define R200_FALLBACK_DRAW_BUFFER 0x2 +#define R200_FALLBACK_STENCIL 0x4 +#define R200_FALLBACK_RENDER_MODE 0x8 +#define R200_FALLBACK_BLEND_EQ 0x10 +#define R200_FALLBACK_BLEND_FUNC 0x20 +#define R200_FALLBACK_DISABLE 0x40 + +/* Use the templated vertex format: + */ +#define COLOR_IS_RGBA +#define TAG(x) r200##x +#include "tnl_dd/t_dd_vertex.h" +#undef TAG + +typedef void (*r200_tri_func)( r200ContextPtr, + r200Vertex *, + r200Vertex *, + r200Vertex * ); + +typedef void (*r200_line_func)( r200ContextPtr, + r200Vertex *, + r200Vertex * ); + +typedef void (*r200_point_func)( r200ContextPtr, + r200Vertex * ); + + +struct r200_colorbuffer_state { + GLuint clear; + GLint drawOffset, drawPitch; +}; + + +struct r200_depthbuffer_state { + GLfloat scale; +}; + +struct r200_pixel_state { + GLint readOffset, readPitch; +}; + +struct r200_scissor_state { + XF86DRIClipRectRec rect; + GLboolean enabled; + + GLuint numClipRects; /* Cliprects active */ + GLuint numAllocedClipRects; /* Cliprects available */ + XF86DRIClipRectPtr pClipRects; +}; + +struct r200_stencilbuffer_state { + GLboolean hwBuffer; + GLuint clear; /* rb3d_stencilrefmask value */ +}; + +struct r200_stipple_state { + GLuint mask[32]; +}; + + + +#define TEX_0 0x1 +#define TEX_1 0x2 +#define TEX_ALL 0x3 + +typedef struct r200_tex_obj r200TexObj, *r200TexObjPtr; + +/* Texture object in locally shared texture space. + */ +struct r200_tex_obj { + r200TexObjPtr next, prev; + + struct gl_texture_object *tObj; /* Mesa texture object */ + + PMemBlock memBlock; /* Memory block containing texture */ + GLuint bufAddr; /* Offset to start of locally + shared texture block */ + + GLuint dirty_images; /* Flags for whether or not + images need to be uploaded to + local or AGP texture space */ + + GLuint dirty_state; /* Flags (1 per texunit) for + whether or not this texobj + has dirty hardware state + (pp_*) that needs to be + brought into the + texunit. */ + + GLint heap; /* Texture heap currently stored in */ + + drmRadeonTexImage image[RADEON_MAX_TEXTURE_LEVELS]; + + GLint totalSize; /* Total size of the texture + including all mipmap levels */ + + GLuint pp_txfilter; /* hardware register values */ + GLuint pp_txformat; + GLuint pp_txoffset; + GLuint pp_txsize; /* npot only */ + GLuint pp_txpitch; /* npot only */ + GLuint pp_border_color; + + /* texObj->Image[firstLevel] through texObj->Image[lastLevel] are the + * images to upload. + */ + GLint firstLevel; + GLint lastLevel; +}; + + +struct r200_texture_env_state { + r200TexObjPtr texobj; + GLenum format; + GLenum envMode; +}; + +#define R200_MAX_TEXTURE_UNITS 3 + +struct r200_texture_state { + struct r200_texture_env_state unit[R200_MAX_TEXTURE_UNITS]; +}; + + +struct r200_state_atom { + struct r200_state_atom *next, *prev; + const char *name; /* for debug */ + int cmd_size; /* size in bytes */ + GLuint idx; + int *cmd; /* one or more cmd's */ + int *lastcmd; /* one or more cmd's */ + GLboolean (*check)( GLcontext *, int ); /* is this state active? */ +}; + + + +/* Trying to keep these relatively short as the variables are becoming + * extravagently long. Drop the R200_ off the front of everything - + * I think we know we're in the r200 driver by now, and keep the + * prefix to 3 letters unless absolutely impossible. + */ + +#define CTX_CMD_0 0 +#define CTX_PP_MISC 1 +#define CTX_PP_FOG_COLOR 2 +#define CTX_RE_SOLID_COLOR 3 +#define CTX_RB3D_BLENDCNTL 4 +#define CTX_RB3D_DEPTHOFFSET 5 +#define CTX_RB3D_DEPTHPITCH 6 +#define CTX_RB3D_ZSTENCILCNTL 7 +#define CTX_CMD_1 8 +#define CTX_PP_CNTL 9 +#define CTX_RB3D_CNTL 10 +#define CTX_RB3D_COLOROFFSET 11 +#define CTX_CMD_2 12 /* why */ +#define CTX_RB3D_COLORPITCH 13 /* why */ +#define CTX_STATE_SIZE 14 + +#define SET_CMD_0 0 +#define SET_SE_CNTL 1 +#define SET_RE_CNTL 2 /* replace se_coord_fmt */ +#define SET_STATE_SIZE 3 + +#define VTE_CMD_0 0 +#define VTE_SE_VTE_CNTL 1 +#define VTE_STATE_SIZE 2 + +#define LIN_CMD_0 0 +#define LIN_RE_LINE_PATTERN 1 +#define LIN_RE_LINE_STATE 2 +#define LIN_CMD_1 3 +#define LIN_SE_LINE_WIDTH 4 +#define LIN_STATE_SIZE 5 + +#define MSK_CMD_0 0 +#define MSK_RB3D_STENCILREFMASK 1 +#define MSK_RB3D_ROPCNTL 2 +#define MSK_RB3D_PLANEMASK 3 +#define MSK_STATE_SIZE 4 + +#define VPT_CMD_0 0 +#define VPT_SE_VPORT_XSCALE 1 +#define VPT_SE_VPORT_XOFFSET 2 +#define VPT_SE_VPORT_YSCALE 3 +#define VPT_SE_VPORT_YOFFSET 4 +#define VPT_SE_VPORT_ZSCALE 5 +#define VPT_SE_VPORT_ZOFFSET 6 +#define VPT_STATE_SIZE 7 + +#define ZBS_CMD_0 0 +#define ZBS_SE_ZBIAS_FACTOR 1 +#define ZBS_SE_ZBIAS_CONSTANT 2 +#define ZBS_STATE_SIZE 3 + +#define MSC_CMD_0 0 +#define MSC_RE_MISC 1 +#define MSC_STATE_SIZE 2 + +#define TAM_CMD_0 0 +#define TAM_DEBUG3 1 +#define TAM_STATE_SIZE 2 + +#define TEX_CMD_0 0 +#define TEX_PP_TXFILTER 1 +#define TEX_PP_TXFORMAT 2 +#define TEX_PP_TXFORMAT_X 3 +#define TEX_PP_TXSIZE 4 +#define TEX_PP_TXPITCH 5 +#define TEX_PP_BORDER_COLOR 6 +#define TEX_CMD_1 7 +#define TEX_PP_TXOFFSET 8 +#define TEX_STATE_SIZE 9 + +#define PIX_CMD_0 0 +#define PIX_PP_TXCBLEND 1 +#define PIX_PP_TXCBLEND2 2 +#define PIX_PP_TXABLEND 3 +#define PIX_PP_TXABLEND2 4 +#define PIX_STATE_SIZE 5 + +#define TF_CMD_0 0 +#define TF_TFACTOR_0 1 +#define TF_TFACTOR_1 2 +#define TF_TFACTOR_2 3 +#define TF_TFACTOR_3 4 +#define TF_TFACTOR_4 5 +#define TF_TFACTOR_5 6 +#define TF_STATE_SIZE 7 + +#define TCL_CMD_0 0 +#define TCL_LIGHT_MODEL_CTL_0 1 +#define TCL_LIGHT_MODEL_CTL_1 2 +#define TCL_PER_LIGHT_CTL_0 3 +#define TCL_PER_LIGHT_CTL_1 4 +#define TCL_PER_LIGHT_CTL_2 5 +#define TCL_PER_LIGHT_CTL_3 6 +#define TCL_CMD_1 7 +#define TCL_UCP_VERT_BLEND_CTL 8 +#define TCL_STATE_SIZE 9 + +#define MSL_CMD_0 0 +#define MSL_MATRIX_SELECT_0 1 +#define MSL_MATRIX_SELECT_1 2 +#define MSL_MATRIX_SELECT_2 3 +#define MSL_MATRIX_SELECT_3 4 +#define MSL_MATRIX_SELECT_4 5 +#define MSL_STATE_SIZE 6 + +#define TCG_CMD_0 0 +#define TCG_TEX_PROC_CTL_2 1 +#define TCG_TEX_PROC_CTL_3 2 +#define TCG_TEX_PROC_CTL_0 3 +#define TCG_TEX_PROC_CTL_1 4 +#define TCG_TEX_CYL_WRAP_CTL 5 +#define TCG_STATE_SIZE 6 + +#define MTL_CMD_0 0 +#define MTL_EMMISSIVE_RED 1 +#define MTL_EMMISSIVE_GREEN 2 +#define MTL_EMMISSIVE_BLUE 3 +#define MTL_EMMISSIVE_ALPHA 4 +#define MTL_AMBIENT_RED 5 +#define MTL_AMBIENT_GREEN 6 +#define MTL_AMBIENT_BLUE 7 +#define MTL_AMBIENT_ALPHA 8 +#define MTL_DIFFUSE_RED 9 +#define MTL_DIFFUSE_GREEN 10 +#define MTL_DIFFUSE_BLUE 11 +#define MTL_DIFFUSE_ALPHA 12 +#define MTL_SPECULAR_RED 13 +#define MTL_SPECULAR_GREEN 14 +#define MTL_SPECULAR_BLUE 15 +#define MTL_SPECULAR_ALPHA 16 +#define MTL_CMD_1 17 +#define MTL_SHININESS 18 +#define MTL_STATE_SIZE 19 + +#define VAP_CMD_0 0 +#define VAP_SE_VAP_CNTL 1 +#define VAP_STATE_SIZE 2 + +/* Replaces a lot of packet info from radeon + */ +#define VTX_CMD_0 0 +#define VTX_VTXFMT_0 1 +#define VTX_VTXFMT_1 2 +#define VTX_TCL_OUTPUT_VTXFMT_0 3 +#define VTX_TCL_OUTPUT_VTXFMT_1 4 +#define VTX_CMD_1 5 +#define VTX_TCL_OUTPUT_COMPSEL 6 +#define VTX_CMD_2 7 +#define VTX_STATE_CNTL 8 +#define VTX_STATE_SIZE 9 + + +#define VTX_COLOR(v,n) (((v)>>(R200_VTX_COLOR_0_SHIFT+(n)*2))&\ + R200_VTX_COLOR_MASK) + +#define MAT_CMD_0 0 +#define MAT_ELT_0 1 +#define MAT_STATE_SIZE 17 + +#define GRD_CMD_0 0 +#define GRD_VERT_GUARD_CLIP_ADJ 1 +#define GRD_VERT_GUARD_DISCARD_ADJ 2 +#define GRD_HORZ_GUARD_CLIP_ADJ 3 +#define GRD_HORZ_GUARD_DISCARD_ADJ 4 +#define GRD_STATE_SIZE 5 + +/* position changes frequently when lighting in modelpos - separate + * out to new state item? + */ +#define LIT_CMD_0 0 +#define LIT_AMBIENT_RED 1 +#define LIT_AMBIENT_GREEN 2 +#define LIT_AMBIENT_BLUE 3 +#define LIT_AMBIENT_ALPHA 4 +#define LIT_DIFFUSE_RED 5 +#define LIT_DIFFUSE_GREEN 6 +#define LIT_DIFFUSE_BLUE 7 +#define LIT_DIFFUSE_ALPHA 8 +#define LIT_SPECULAR_RED 9 +#define LIT_SPECULAR_GREEN 10 +#define LIT_SPECULAR_BLUE 11 +#define LIT_SPECULAR_ALPHA 12 +#define LIT_POSITION_X 13 +#define LIT_POSITION_Y 14 +#define LIT_POSITION_Z 15 +#define LIT_POSITION_W 16 +#define LIT_DIRECTION_X 17 +#define LIT_DIRECTION_Y 18 +#define LIT_DIRECTION_Z 19 +#define LIT_DIRECTION_W 20 +#define LIT_ATTEN_CONST 21 +#define LIT_ATTEN_LINEAR 22 +#define LIT_ATTEN_QUADRATIC 23 +#define LIT_ATTEN_XXX 24 +#define LIT_CMD_1 25 +#define LIT_SPOT_DCD 26 +#define LIT_SPOT_DCM 27 +#define LIT_SPOT_EXPONENT 28 +#define LIT_SPOT_CUTOFF 29 +#define LIT_SPECULAR_THRESH 30 +#define LIT_RANGE_CUTOFF 31 /* ? */ +#define LIT_RANGE_ATTEN 32 /* ? */ +#define LIT_STATE_SIZE 33 + +/* Fog + */ +#define FOG_CMD_0 0 +#define FOG_R 1 +#define FOG_C 2 +#define FOG_D 3 +#define FOG_PAD 4 +#define FOG_STATE_SIZE 5 + +/* UCP + */ +#define UCP_CMD_0 0 +#define UCP_X 1 +#define UCP_Y 2 +#define UCP_Z 3 +#define UCP_W 4 +#define UCP_STATE_SIZE 5 + +/* GLT - Global ambient + */ +#define GLT_CMD_0 0 +#define GLT_RED 1 +#define GLT_GREEN 2 +#define GLT_BLUE 3 +#define GLT_ALPHA 4 +#define GLT_STATE_SIZE 5 + +/* EYE + */ +#define EYE_CMD_0 0 +#define EYE_X 1 +#define EYE_Y 2 +#define EYE_Z 3 +#define EYE_RESCALE_FACTOR 4 +#define EYE_STATE_SIZE 5 + +/* CST - constant state + */ +#define CST_CMD_0 0 +#define CST_PP_CNTL_X 1 +#define CST_CMD_1 2 +#define CST_RB3D_DEPTHXY_OFFSET 3 +#define CST_CMD_2 4 +#define CST_RE_AUX_SCISSOR_CNTL 5 +#define CST_CMD_3 6 +#define CST_RE_SCISSOR_TL_0 7 +#define CST_RE_SCISSOR_BR_0 8 +#define CST_CMD_4 9 +#define CST_SE_VAP_CNTL_STATUS 10 +#define CST_CMD_5 11 +#define CST_RE_POINTSIZE 12 +#define CST_CMD_6 13 +#define CST_SE_TCL_INPUT_VTX_0 14 +#define CST_SE_TCL_INPUT_VTX_1 15 +#define CST_SE_TCL_INPUT_VTX_2 16 +#define CST_SE_TCL_INPUT_VTX_3 17 +#define CST_STATE_SIZE 18 + + + + +struct r200_hw_state { + /* All state should be on one of these lists: + */ + struct r200_state_atom dirty; /* dirty list head placeholder */ + struct r200_state_atom clean; /* clean list head placeholder */ + + /* Hardware state, stored as cmdbuf commands: + * -- Need to doublebuffer for + * - reviving state after loss of context + * - eliding noop statechange loops? (except line stipple count) + */ + struct r200_state_atom ctx; + struct r200_state_atom set; + struct r200_state_atom vte; + struct r200_state_atom lin; + struct r200_state_atom msk; + struct r200_state_atom vpt; + struct r200_state_atom vap; + struct r200_state_atom vtx; + struct r200_state_atom tcl; + struct r200_state_atom msl; + struct r200_state_atom tcg; + struct r200_state_atom msc; + struct r200_state_atom cst; + struct r200_state_atom tam; + struct r200_state_atom tf; + struct r200_state_atom tex[2]; + struct r200_state_atom zbs; + struct r200_state_atom mtl[2]; + struct r200_state_atom mat[5]; + struct r200_state_atom lit[8]; /* includes vec, scl commands */ + struct r200_state_atom ucp[6]; + struct r200_state_atom pix[6]; /* pixshader stages */ + struct r200_state_atom eye; /* eye pos */ + struct r200_state_atom grd; /* guard band clipping */ + struct r200_state_atom fog; + struct r200_state_atom glt; +}; + +struct r200_state { + /* Derived state for internal purposes: + */ + struct r200_colorbuffer_state color; + struct r200_depthbuffer_state depth; + struct r200_pixel_state pixel; + struct r200_scissor_state scissor; + struct r200_stencilbuffer_state stencil; + struct r200_stipple_state stipple; + struct r200_texture_state texture; +}; + +struct r200_texture { + r200TexObj objects[R200_NR_TEX_HEAPS]; + r200TexObj swapped; + + memHeap_t *heap[R200_NR_TEX_HEAPS]; + GLint age[R200_NR_TEX_HEAPS]; + + GLint numHeaps; +}; + +/* Need refcounting on dma buffers: + */ +struct r200_dma_buffer { + int refcount; /* the number of retained regions in buf */ + drmBufPtr buf; +}; + +#define GET_START(rvb) (rmesa->dri.agp_buffer_offset + \ + (rvb)->address - rmesa->dma.buf0_address + \ + (rvb)->start) + +/* A retained region, eg vertices for indexed vertices. + */ +struct r200_dma_region { + struct r200_dma_buffer *buf; + char *address; /* == buf->address */ + int start, end, ptr; /* offsets from start of buf */ + int aos_start; + int aos_stride; + int aos_size; +}; + + +struct r200_dma { + /* Active dma region. Allocations for vertices and retained + * regions come from here. Also used for emitting random vertices, + * these may be flushed by calling flush_current(); + */ + struct r200_dma_region current; + + void (*flush)( r200ContextPtr ); + + char *buf0_address; /* start of buf[0], for index calcs */ + GLuint nr_released_bufs; /* flush after so many buffers released */ +}; + +struct r200_dri_mirror { + Display *display; /* X server display */ + + __DRIcontextPrivate *context; /* DRI context */ + __DRIscreenPrivate *screen; /* DRI screen */ + __DRIdrawablePrivate *drawable; /* DRI drawable bound to this ctx */ + + drmContext hwContext; + drmLock *hwLock; + int fd; + int drmMinor; + int agp_buffer_offset; /* offset in card memory space */ + int agp_texture_offset; /* offset in card memory space */ +}; + + +#define R200_CMD_BUF_SZ (8*1024) + +struct r200_store { + GLuint statenr; + GLuint primnr; + char cmd_buf[R200_CMD_BUF_SZ]; + int cmd_used; + int elts_start; +}; + + +/* r200_tcl.c + */ +struct r200_tcl_info { + GLuint vertex_format; + GLint last_offset; + GLuint hw_primitive; + + struct r200_dma_region *aos_components[8]; + GLuint nr_aos_components; + + GLuint *Elts; + + struct r200_dma_region indexed_verts; + struct r200_dma_region obj; + struct r200_dma_region rgba; + struct r200_dma_region spec; + struct r200_dma_region fog; + struct r200_dma_region tex[R200_MAX_TEXTURE_UNITS]; + struct r200_dma_region norm; +}; + + +/* r200_swtcl.c + */ +struct r200_swtcl_info { + GLuint SetupIndex; + GLuint SetupNewInputs; + GLuint RenderIndex; + GLuint vertex_size; + GLuint vertex_stride_shift; + GLuint vertex_format; + char *verts; + + /* Fallback rasterization functions + */ + r200_point_func draw_point; + r200_line_func draw_line; + r200_tri_func draw_tri; + + GLuint hw_primitive; + GLenum render_primitive; + GLuint numverts; + + struct r200_dma_region indexed_verts; +}; + + +struct r200_ioctl { + GLuint vertex_offset; + GLuint vertex_size; +}; + + + +#define R200_MAX_PRIMS 64 + + +/* Want to keep a cache of these around. Each is parameterized by + * only a single value which has only a small range. Only expect a + * few, so just rescan the list each time? + */ +struct dynfn { + struct dynfn *next, *prev; + int key[2]; + char *code; +}; + +struct dfn_lists { + struct dynfn Vertex2f; + struct dynfn Vertex2fv; + struct dynfn Vertex3f; + struct dynfn Vertex3fv; + struct dynfn Color4ub; + struct dynfn Color4ubv; + struct dynfn Color3ub; + struct dynfn Color3ubv; + struct dynfn Color4f; + struct dynfn Color4fv; + struct dynfn Color3f; + struct dynfn Color3fv; + struct dynfn SecondaryColor3ubEXT; + struct dynfn SecondaryColor3ubvEXT; + struct dynfn SecondaryColor3fEXT; + struct dynfn SecondaryColor3fvEXT; + struct dynfn Normal3f; + struct dynfn Normal3fv; + struct dynfn TexCoord2f; + struct dynfn TexCoord2fv; + struct dynfn TexCoord1f; + struct dynfn TexCoord1fv; + struct dynfn MultiTexCoord2fARB; + struct dynfn MultiTexCoord2fvARB; + struct dynfn MultiTexCoord1fARB; + struct dynfn MultiTexCoord1fvARB; +}; + +struct _vb; + +struct dfn_generators { + struct dynfn *(*Vertex2f)( GLcontext *, const int * ); + struct dynfn *(*Vertex2fv)( GLcontext *, const int * ); + struct dynfn *(*Vertex3f)( GLcontext *, const int * ); + struct dynfn *(*Vertex3fv)( GLcontext *, const int * ); + struct dynfn *(*Color4ub)( GLcontext *, const int * ); + struct dynfn *(*Color4ubv)( GLcontext *, const int * ); + struct dynfn *(*Color3ub)( GLcontext *, const int * ); + struct dynfn *(*Color3ubv)( GLcontext *, const int * ); + struct dynfn *(*Color4f)( GLcontext *, const int * ); + struct dynfn *(*Color4fv)( GLcontext *, const int * ); + struct dynfn *(*Color3f)( GLcontext *, const int * ); + struct dynfn *(*Color3fv)( GLcontext *, const int * ); + struct dynfn *(*SecondaryColor3ubEXT)( GLcontext *, const int * ); + struct dynfn *(*SecondaryColor3ubvEXT)( GLcontext *, const int * ); + struct dynfn *(*SecondaryColor3fEXT)( GLcontext *, const int * ); + struct dynfn *(*SecondaryColor3fvEXT)( GLcontext *, const int * ); + struct dynfn *(*Normal3f)( GLcontext *, const int * ); + struct dynfn *(*Normal3fv)( GLcontext *, const int * ); + struct dynfn *(*TexCoord2f)( GLcontext *, const int * ); + struct dynfn *(*TexCoord2fv)( GLcontext *, const int * ); + struct dynfn *(*TexCoord1f)( GLcontext *, const int * ); + struct dynfn *(*TexCoord1fv)( GLcontext *, const int * ); + struct dynfn *(*MultiTexCoord2fARB)( GLcontext *, const int * ); + struct dynfn *(*MultiTexCoord2fvARB)( GLcontext *, const int * ); + struct dynfn *(*MultiTexCoord1fARB)( GLcontext *, const int * ); + struct dynfn *(*MultiTexCoord1fvARB)( GLcontext *, const int * ); +}; + + +struct r200_vb { + /* Keep these first: referenced from codegen templates: + */ + GLint counter, initial_counter; + GLint *dmaptr; + void (*notify)( void ); + GLint vertex_size; + union { float f; int i; GLubyte ub4[4]; } vertex[15]; + + GLfloat *normalptr; + GLfloat *floatcolorptr; + GLubyte *ubytecolorptr; + GLubyte *ubytespecptr; + GLfloat *texcoordptr[2]; + + GLcontext *context; /* current context : Single thread only! */ +}; + +struct r200_prim { + GLuint start; + GLuint end; + GLuint prim; +}; + +struct r200_vbinfo { + GLenum *prim; /* &ctx->Driver.CurrentExecPrimitive */ + GLuint primflags; + GLboolean enabled; /* R200_NO_VTXFMT//R200_NO_TCL env vars */ + GLboolean installed; + GLboolean fell_back; + GLboolean recheck; + GLint initial_counter; + GLint nrverts; + GLuint vtxfmt_0, vtxfmt_1; + + GLuint installed_vertex_format; + GLuint installed_color_3f_sz; + + struct r200_prim primlist[R200_MAX_PRIMS]; + int nrprims; + + struct dfn_lists dfn_cache; + struct dfn_generators codegen; + GLvertexformat vtxfmt; +}; + + + + +struct r200_context { + GLcontext *glCtx; /* Mesa context */ + + /* Driver and hardware state management + */ + struct r200_hw_state hw; + struct r200_state state; + + /* Texture object bookkeeping + */ + struct r200_texture texture; + + + /* Rasterization and vertex state: + */ + GLuint TclFallback; + GLuint Fallback; + GLuint NewGLState; + + + /* Temporaries for translating away float colors: + */ + struct gl_client_array UbyteColor; + struct gl_client_array UbyteSecondaryColor; + + /* Vertex buffers + */ + struct r200_ioctl ioctl; + struct r200_dma dma; + struct r200_store store; + + /* Page flipping + */ + GLuint doPageFlip; + + /* Busy waiting + */ + GLuint do_usleeps; + + /* Drawable, cliprect and scissor information + */ + GLuint numClipRects; /* Cliprects for the draw buffer */ + XF86DRIClipRectPtr pClipRects; + GLuint lastStamp; + GLboolean lost_context; + r200ScreenPtr r200Screen; /* Screen private DRI data */ + RADEONSAREAPrivPtr sarea; /* Private SAREA data */ + + /* TCL stuff + */ + GLmatrix TexGenMatrix[R200_MAX_TEXTURE_UNITS]; + GLboolean recheck_texgen[R200_MAX_TEXTURE_UNITS]; + GLboolean TexGenNeedNormals[R200_MAX_TEXTURE_UNITS]; + GLuint TexMatEnabled; + GLuint TexMatCompSel; + GLuint TexGenEnabled; + GLuint TexGenInputs; + GLuint TexGenCompSel; + GLmatrix tmpmat; + + /* r200_tcl.c + */ + struct r200_tcl_info tcl; + + /* r200_swtcl.c + */ + struct r200_swtcl_info swtcl; + + /* r200_vtxfmt.c + */ + struct r200_vbinfo vb; + + /* Mirrors of some DRI state + */ + struct r200_dri_mirror dri; +}; + +#define R200_CONTEXT(ctx) ((r200ContextPtr)(ctx->DriverCtx)) + + +static __inline GLuint r200PackColor( GLuint cpp, + GLubyte r, GLubyte g, + GLubyte b, GLubyte a ) +{ + switch ( cpp ) { + case 2: + return PACK_COLOR_565( r, g, b ); + case 4: + return PACK_COLOR_8888( a, r, g, b ); + default: + return 0; + } +} + +#define R200_OLD_PACKETS 0 + +/* ================================================================ + * Debugging: + */ +#define DO_DEBUG 1 + +#if DO_DEBUG +extern int R200_DEBUG; +#else +#define R200_DEBUG 0 +#endif + +#define DEBUG_TEXTURE 0x001 +#define DEBUG_STATE 0x002 +#define DEBUG_IOCTL 0x004 +#define DEBUG_PRIMS 0x008 +#define DEBUG_VERTS 0x010 +#define DEBUG_FALLBACKS 0x020 +#define DEBUG_VFMT 0x040 +#define DEBUG_CODEGEN 0x080 +#define DEBUG_VERBOSE 0x100 +#define DEBUG_DRI 0x200 +#define DEBUG_DMA 0x400 +#define DEBUG_SANITY 0x800 +#define DEBUG_SYNC 0x1000 + +#endif +#endif /* __R200_CONTEXT_H__ */ diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_ioctl.c b/xc/lib/GL/mesa/src/drv/r200/r200_ioctl.c new file mode 100644 index 000000000..80c0a3a87 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_ioctl.c @@ -0,0 +1,958 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_state.h" +#include "r200_ioctl.h" +#include "r200_tcl.h" +#include "r200_sanity.h" +#include "radeon_reg.h" + +#include "mem.h" +#include "macros.h" +#include "swrast/swrast.h" +#include "simple_list.h" + +#define R200_TIMEOUT 512 +#define R200_IDLE_RETRY 16 + +#undef usleep + +static void do_usleep( int nr, const char *caller ) +{ + if (0) fprintf(stderr, "usleep %d in %s\n", nr, caller ); + if (1) usleep( nr ); +} + +static void r200WaitForIdle( r200ContextPtr rmesa ); + +/* ============================================================= + * Kernel command buffer handling + */ + +static void print_state_atom( struct r200_state_atom *state ) +{ + int i; + + fprintf(stderr, "emit %s/%d\n", state->name, state->cmd_size); + + if (0 & R200_DEBUG & DEBUG_VERBOSE) + for (i = 0 ; i < state->cmd_size ; i++) + fprintf(stderr, "\t%s[%d]: %x\n", state->name, i, state->cmd[i]); + +} + +static void r200_emit_state_list( r200ContextPtr rmesa, + struct r200_state_atom *list ) +{ + struct r200_state_atom *state, *tmp; + char *dest; + + foreach_s( state, tmp, list ) { + if (state->check( rmesa->glCtx, state->idx )) { + dest = r200AllocCmdBuf( rmesa, state->cmd_size * 4, __FUNCTION__); + memcpy( dest, state->cmd, state->cmd_size * 4); + move_to_head( &(rmesa->hw.clean), state ); + if (R200_DEBUG & DEBUG_STATE) + print_state_atom( state ); + } + else if (R200_DEBUG & DEBUG_STATE) + fprintf(stderr, "skip state %s\n", state->name); + } +} + + +void r200EmitState( r200ContextPtr rmesa ) +{ + struct r200_state_atom *state, *tmp; + + if (R200_DEBUG & (DEBUG_STATE|DEBUG_PRIMS)) + fprintf(stderr, "%s\n", __FUNCTION__); + + /* Somewhat overkill: + */ + if ( rmesa->lost_context) { + if (R200_DEBUG & (DEBUG_STATE|DEBUG_PRIMS|DEBUG_IOCTL)) + fprintf(stderr, "%s - lost context\n", __FUNCTION__); + + foreach_s( state, tmp, &(rmesa->hw.clean) ) + move_to_tail(&(rmesa->hw.dirty), state ); + + rmesa->lost_context = 0; + } + else { + move_to_tail( &rmesa->hw.dirty, &rmesa->hw.mtl[0] ); + /* odd bug? -- isosurf, cycle between reflect & lit */ + } + + r200_emit_state_list( rmesa, &rmesa->hw.dirty ); +} + + + +/* Fire a section of the retained (indexed_verts) buffer as a regular + * primtive. + */ +extern void r200EmitVbufPrim( r200ContextPtr rmesa, + GLuint primitive, + GLuint vertex_nr ) +{ + drmRadeonCmdHeader *cmd; + + assert(!(primitive & R200_VF_PRIM_WALK_IND)); + + r200EmitState( rmesa ); + + if (R200_DEBUG & (DEBUG_IOCTL|DEBUG_PRIMS)) + fprintf(stderr, "%s cmd_used/4: %d prim %x nr %d\n", __FUNCTION__, + rmesa->store.cmd_used/4, primitive, vertex_nr); + + cmd = (drmRadeonCmdHeader *)r200AllocCmdBuf( rmesa, 3 * sizeof(*cmd), + __FUNCTION__ ); + cmd[0].i = 0; + cmd[0].header.cmd_type = RADEON_CMD_PACKET3_CLIP; + cmd[1].i = R200_CP_CMD_3D_DRAW_VBUF_2; + cmd[2].i = (primitive | + R200_VF_PRIM_WALK_LIST | + R200_VF_COLOR_ORDER_RGBA | + (vertex_nr << R200_VF_VERTEX_NUMBER_SHIFT)); + + + if (R200_DEBUG & DEBUG_SYNC) { + fprintf(stderr, "\nSyncing\n\n"); + R200_FIREVERTICES( rmesa ); + r200Finish( rmesa->glCtx ); + } +} + + +void r200FlushElts( r200ContextPtr rmesa ) +{ + int *cmd = (int *)(rmesa->store.cmd_buf + rmesa->store.elts_start); + int dwords; + int nr = (rmesa->store.cmd_used - (rmesa->store.elts_start + 12)) / 2; + + if (R200_DEBUG & (DEBUG_IOCTL|DEBUG_PRIMS)) + fprintf(stderr, "%s\n", __FUNCTION__); + + assert( rmesa->dma.flush == r200FlushElts ); + rmesa->dma.flush = 0; + + /* Cope with odd number of elts: + */ + rmesa->store.cmd_used = (rmesa->store.cmd_used + 2) & ~2; + dwords = (rmesa->store.cmd_used - rmesa->store.elts_start) / 4; + + cmd[1] |= (dwords - 3) << 16; + cmd[2] |= nr << R200_VF_VERTEX_NUMBER_SHIFT; + + if (R200_DEBUG & DEBUG_SYNC) { + fprintf(stderr, "\nSyncing in %s\n\n", __FUNCTION__); + R200_FIREVERTICES( rmesa ); + r200Finish( rmesa->glCtx ); + } +} + + +GLushort *r200AllocEltsOpenEnded( r200ContextPtr rmesa, + GLuint primitive, + GLuint min_nr ) +{ + drmRadeonCmdHeader *cmd; + GLushort *retval; + + if (R200_DEBUG & DEBUG_IOCTL) + fprintf(stderr, "%s %d prim %x\n", __FUNCTION__, min_nr, primitive); + + assert((primitive & R200_VF_PRIM_WALK_IND)); + + r200EmitState( rmesa ); + + cmd = (drmRadeonCmdHeader *)r200AllocCmdBuf( rmesa, + 12 + min_nr*2, + __FUNCTION__ ); + cmd[0].i = 0; + cmd[0].header.cmd_type = RADEON_CMD_PACKET3_CLIP; + cmd[1].i = R200_CP_CMD_3D_DRAW_INDX_2; + cmd[2].i = (primitive | + R200_VF_PRIM_WALK_IND | + R200_VF_COLOR_ORDER_RGBA); + + + retval = (GLushort *)(cmd+3); + + if (R200_DEBUG & DEBUG_PRIMS) + fprintf(stderr, "%s: header 0x%x prim %x \n", + __FUNCTION__, + cmd[1].i, primitive); + + assert(!rmesa->dma.flush); + rmesa->dma.flush = r200FlushElts; + + rmesa->store.elts_start = ((char *)cmd) - rmesa->store.cmd_buf; + + return retval; +} + + + +void r200EmitVertexAOS( r200ContextPtr rmesa, + GLuint vertex_size, + GLuint offset ) +{ + drmRadeonCmdHeader *cmd; + + if (R200_DEBUG & (DEBUG_PRIMS|DEBUG_IOCTL)) + fprintf(stderr, "%s: vertex_size 0x%x offset 0x%x \n", + __FUNCTION__, vertex_size, offset); + + cmd = (drmRadeonCmdHeader *)r200AllocCmdBuf( rmesa, 5 * sizeof(int), + __FUNCTION__ ); + + cmd[0].header.cmd_type = RADEON_CMD_PACKET3; + cmd[1].i = R200_CP_CMD_3D_LOAD_VBPNTR | (2 << 16); + cmd[2].i = 1; + cmd[3].i = vertex_size | (vertex_size << 8); + cmd[4].i = offset; +} + + +void r200EmitAOS( r200ContextPtr rmesa, + struct r200_dma_region **component, + GLuint nr, + GLuint offset ) +{ + drmRadeonCmdHeader *cmd; + int sz = 3 + ((nr/2)*3) + ((nr&1)*2); + int i; + int *tmp; + + if (R200_DEBUG & DEBUG_IOCTL) + fprintf(stderr, "%s nr arrays: %d\n", __FUNCTION__, nr); + + cmd = (drmRadeonCmdHeader *)r200AllocCmdBuf( rmesa, sz * sizeof(int), + __FUNCTION__ ); + cmd[0].i = 0; + cmd[0].header.cmd_type = RADEON_CMD_PACKET3; + cmd[1].i = R200_CP_CMD_3D_LOAD_VBPNTR | ((sz-3) << 16); + cmd[2].i = nr; + tmp = &cmd[0].i; + cmd += 3; + + for (i = 0 ; i < nr ; i++) { + if (i & 1) { + cmd[0].i |= ((component[i]->aos_stride << 24) | + (component[i]->aos_size << 16)); + cmd[2].i = (component[i]->aos_start + + offset * component[i]->aos_stride * 4); + cmd += 3; + } + else { + cmd[0].i = ((component[i]->aos_stride << 8) | + (component[i]->aos_size << 0)); + cmd[1].i = (component[i]->aos_start + + offset * component[i]->aos_stride * 4); + } + } + + if (R200_DEBUG & DEBUG_VERTS) { + fprintf(stderr, "%s:\n", __FUNCTION__); + for (i = 0 ; i < sz ; i++) + fprintf(stderr, " %d: %x\n", i, tmp[i]); + } +} + + +static int r200FlushCmdBufLocked( r200ContextPtr rmesa, + const char * caller ) +{ + int ret, i; + drmRadeonCmdBuffer cmd; + + if (R200_DEBUG & DEBUG_IOCTL) { + fprintf(stderr, "%s from %s\n", __FUNCTION__, caller); + + if (0 & R200_DEBUG & DEBUG_VERBOSE) + for (i = 0 ; i < rmesa->store.cmd_used ; i += 4 ) + fprintf(stderr, "%d: %x\n", i/4, + *(int *)(&rmesa->store.cmd_buf[i])); + } + + if (R200_DEBUG & DEBUG_DMA) + fprintf(stderr, "%s: Releasing %d buffers\n", __FUNCTION__, + rmesa->dma.nr_released_bufs); + + + if (R200_DEBUG & DEBUG_SANITY) { + if (rmesa->state.scissor.enabled) + ret = r200SanityCmdBuffer( rmesa, + rmesa->state.scissor.numClipRects, + rmesa->state.scissor.pClipRects); + else + ret = r200SanityCmdBuffer( rmesa, + rmesa->numClipRects, + rmesa->pClipRects); + if (ret) { + fprintf(stderr, "drmSanityCommandWrite: %d\n", ret); + goto out; + } + } + + + cmd.bufsz = rmesa->store.cmd_used; + cmd.buf = rmesa->store.cmd_buf; + + if (rmesa->state.scissor.enabled) { + cmd.nbox = rmesa->state.scissor.numClipRects; + cmd.boxes = (drmClipRect *)rmesa->state.scissor.pClipRects; + } else { + cmd.nbox = rmesa->numClipRects; + cmd.boxes = (drmClipRect *)rmesa->pClipRects; + } + + ret = drmCommandWrite( rmesa->dri.fd, + DRM_RADEON_CMDBUF, + &cmd, sizeof(cmd) ); + + if (ret) + fprintf(stderr, "drmCommandWrite: %d\n", ret); + + if (R200_DEBUG & DEBUG_SYNC) { + fprintf(stderr, "\nSyncing in %s\n\n", __FUNCTION__); + r200WaitForIdleLocked( rmesa ); + } + + + out: + rmesa->store.primnr = 0; + rmesa->store.statenr = 0; + rmesa->store.cmd_used = 0; + rmesa->dma.nr_released_bufs = 0; +/* rmesa->lost_context = 0; */ + rmesa->lost_context = 1; + return ret; +} + + +/* Note: does not emit any commands to avoid recursion on + * r200AllocCmdBuf. + */ +void r200FlushCmdBuf( r200ContextPtr rmesa, const char *caller ) +{ + int ret; + + LOCK_HARDWARE( rmesa ); + + ret = r200FlushCmdBufLocked( rmesa, caller ); + + UNLOCK_HARDWARE( rmesa ); + + if (ret) { + fprintf(stderr, "drmRadeonCmdBuffer: %d\n", ret); + exit(ret); + } +} + + +/* ============================================================= + * Hardware vertex buffer handling + */ + + +void r200RefillCurrentDmaRegion( r200ContextPtr rmesa ) +{ + struct r200_dma_buffer *dmabuf; + int fd = rmesa->dri.fd; + int index = 0; + int size = 0; + drmDMAReq dma; + int ret; + + if (R200_DEBUG & (DEBUG_IOCTL|DEBUG_DMA)) + fprintf(stderr, "%s\n", __FUNCTION__); + + if (rmesa->dma.flush) { + rmesa->dma.flush( rmesa ); + } + + if (rmesa->dma.current.buf) + r200ReleaseDmaRegion( rmesa, &rmesa->dma.current, __FUNCTION__ ); + + if (rmesa->dma.nr_released_bufs > 4) + r200FlushCmdBuf( rmesa, __FUNCTION__ ); + + dma.context = rmesa->dri.hwContext; + dma.send_count = 0; + dma.send_list = NULL; + dma.send_sizes = NULL; + dma.flags = 0; + dma.request_count = 1; + dma.request_size = RADEON_BUFFER_SIZE; + dma.request_list = &index; + dma.request_sizes = &size; + dma.granted_count = 0; + + LOCK_HARDWARE(rmesa); /* no need to validate */ + + while (1) { + ret = drmDMA( fd, &dma ); + if (ret == 0) + break; + + if (rmesa->dma.nr_released_bufs) { + r200FlushCmdBufLocked( rmesa, __FUNCTION__ ); + } + + if (rmesa->do_usleeps) { + UNLOCK_HARDWARE( rmesa ); + do_usleep(1, __FUNCTION__); + LOCK_HARDWARE( rmesa ); + } + } + + UNLOCK_HARDWARE(rmesa); + + if (R200_DEBUG & DEBUG_DMA) + fprintf(stderr, "Allocated buffer %d\n", index); + + dmabuf = CALLOC_STRUCT( r200_dma_buffer ); + dmabuf->buf = &rmesa->r200Screen->buffers->list[index]; + dmabuf->refcount = 1; + + rmesa->dma.current.buf = dmabuf; + rmesa->dma.current.address = dmabuf->buf->address; + rmesa->dma.current.end = dmabuf->buf->total; + rmesa->dma.current.start = 0; + rmesa->dma.current.ptr = 0; +} + +void r200ReleaseDmaRegion( r200ContextPtr rmesa, + struct r200_dma_region *region, + const char *caller ) +{ + if (R200_DEBUG & DEBUG_IOCTL) + fprintf(stderr, "%s from %s\n", __FUNCTION__, caller); + + if (!region->buf) + return; + + if (rmesa->dma.flush) + rmesa->dma.flush( rmesa ); + + if (--region->buf->refcount == 0) { + drmRadeonCmdHeader *cmd; + + if (R200_DEBUG & (DEBUG_IOCTL|DEBUG_DMA)) + fprintf(stderr, "%s -- DISCARD BUF %d\n", __FUNCTION__, + region->buf->buf->idx); + + cmd = (drmRadeonCmdHeader *)r200AllocCmdBuf( rmesa, sizeof(*cmd), + __FUNCTION__ ); + cmd->dma.cmd_type = RADEON_CMD_DMA_DISCARD; + cmd->dma.buf_idx = region->buf->buf->idx; + FREE(region->buf); + rmesa->dma.nr_released_bufs++; + } + + region->buf = 0; + region->start = 0; +} + +/* Allocates a region from rmesa->dma.current. If there isn't enough + * space in current, grab a new buffer (and discard what was left of current) + */ +void r200AllocDmaRegion( r200ContextPtr rmesa, + struct r200_dma_region *region, + int bytes, + int alignment ) +{ + if (R200_DEBUG & DEBUG_IOCTL) + fprintf(stderr, "%s %d\n", __FUNCTION__, bytes); + + if (rmesa->dma.flush) + rmesa->dma.flush( rmesa ); + + if (region->buf) + r200ReleaseDmaRegion( rmesa, region, __FUNCTION__ ); + + alignment--; + rmesa->dma.current.start = rmesa->dma.current.ptr = + (rmesa->dma.current.ptr + alignment) & ~alignment; + + if ( rmesa->dma.current.ptr + bytes > rmesa->dma.current.end ) + r200RefillCurrentDmaRegion( rmesa ); + + region->start = rmesa->dma.current.start; + region->ptr = rmesa->dma.current.start; + region->end = rmesa->dma.current.start + bytes; + region->address = rmesa->dma.current.address; + region->buf = rmesa->dma.current.buf; + region->buf->refcount++; + + rmesa->dma.current.ptr += bytes; /* bug - if alignment > 7 */ + rmesa->dma.current.start = + rmesa->dma.current.ptr = (rmesa->dma.current.ptr + 0x7) & ~0x7; +} + +void r200AllocDmaRegionVerts( r200ContextPtr rmesa, + struct r200_dma_region *region, + int numverts, + int vertsize, + int alignment ) +{ + r200AllocDmaRegion( rmesa, region, vertsize * numverts, alignment ); +} + +/* ================================================================ + * SwapBuffers with client-side throttling + */ + +#define R200_MAX_OUTSTANDING 2 + + +static int r200WaitForFrameCompletion( r200ContextPtr rmesa ) +{ + RADEONSAREAPrivPtr sarea = rmesa->sarea; + CARD32 frame; + int wait = 0; + + while ( 1 ) { + drmRadeonGetParam gp; + int ret; + + gp.param = RADEON_PARAM_LAST_FRAME; + gp.value = (int *)&frame; + ret = drmCommandWriteRead( rmesa->dri.fd, + DRM_RADEON_GETPARAM, &gp, sizeof(gp) ); + + if ( ret ) { + fprintf( stderr, "%s: drmRadeonGetParam: %d\n", __FUNCTION__, ret ); + exit(1); + } + + if ( sarea->last_frame - frame <= R200_MAX_OUTSTANDING ) { + break; + } + wait++; + + if (rmesa->do_usleeps) { + UNLOCK_HARDWARE( rmesa ); + do_usleep(1, __FUNCTION__); + LOCK_HARDWARE( rmesa ); + } + } + + if ( R200_DEBUG & DEBUG_IOCTL ) + fprintf(stderr, "%s( done, wait=%d )\n", __FUNCTION__, wait); + + return wait; +} + + +/* Copy the back color buffer to the front color buffer. + */ +void r200CopyBuffer( const __DRIdrawablePrivate *dPriv ) +{ + r200ContextPtr rmesa; + GLint nbox, i, ret; + + assert(dPriv); + assert(dPriv->driContextPriv); + assert(dPriv->driContextPriv->driverPrivate); + + rmesa = (r200ContextPtr) dPriv->driContextPriv->driverPrivate; + + if ( R200_DEBUG & DEBUG_IOCTL ) { + fprintf( stderr, "\n%s( %p )\n\n", __FUNCTION__, rmesa->glCtx ); + } + + R200_FIREVERTICES( rmesa ); + + LOCK_HARDWARE( rmesa ); + + + /* Throttle the frame rate -- only allow one pending swap buffers + * request at a time. + */ + r200WaitForFrameCompletion( rmesa ); + + nbox = rmesa->dri.drawable->numClipRects; /* must be in locked region */ + + for ( i = 0 ; i < nbox ; ) { + GLint nr = MIN2( i + RADEON_NR_SAREA_CLIPRECTS , nbox ); + XF86DRIClipRectPtr box = rmesa->dri.drawable->pClipRects; + XF86DRIClipRectPtr b = rmesa->sarea->boxes; + GLint n = 0; + + for ( ; i < nr ; i++ ) { + *b++ = box[i]; + n++; + } + rmesa->sarea->nbox = n; + + ret = drmCommandNone( rmesa->dri.fd, DRM_RADEON_SWAP ); + + if ( ret ) { + fprintf( stderr, "DRM_R200_SWAP_BUFFERS: return = %d\n", ret ); + UNLOCK_HARDWARE( rmesa ); + exit( 1 ); + } + } + + UNLOCK_HARDWARE( rmesa ); + rmesa->lost_context = 1; +} + +void r200PageFlip( const __DRIdrawablePrivate *dPriv ) +{ + r200ContextPtr rmesa; + GLint ret; + + assert(dPriv); + assert(dPriv->driContextPriv); + assert(dPriv->driContextPriv->driverPrivate); + + rmesa = (r200ContextPtr) dPriv->driContextPriv->driverPrivate; + + if ( R200_DEBUG & DEBUG_IOCTL ) { + fprintf(stderr, "%s: pfCurrentPage: %d\n", __FUNCTION__, + rmesa->sarea->pfCurrentPage); + } + + R200_FIREVERTICES( rmesa ); + LOCK_HARDWARE( rmesa ); + + if (!rmesa->dri.drawable->numClipRects) { + UNLOCK_HARDWARE( rmesa ); + usleep( 10000 ); /* throttle invisible client 10ms */ + return; + } + + /* Need to do this for the perf box placement: + */ + { + XF86DRIClipRectPtr box = rmesa->dri.drawable->pClipRects; + XF86DRIClipRectPtr b = rmesa->sarea->boxes; + b[0] = box[0]; + rmesa->sarea->nbox = 1; + } + + /* Throttle the frame rate -- only allow a few pending swap buffers + * request at a time. + */ + r200WaitForFrameCompletion( rmesa ); + + ret = drmCommandNone( rmesa->dri.fd, DRM_RADEON_FLIP ); + + UNLOCK_HARDWARE( rmesa ); + + if ( ret ) { + fprintf( stderr, "DRM_R200_FLIP: return = %d\n", ret ); + exit( 1 ); + } + + if ( rmesa->sarea->pfCurrentPage == 1 ) { + rmesa->state.color.drawOffset = rmesa->r200Screen->frontOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->frontPitch; + } else { + rmesa->state.color.drawOffset = rmesa->r200Screen->backOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->backPitch; + } + + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_RB3D_COLOROFFSET] = rmesa->state.color.drawOffset; + rmesa->hw.ctx.cmd[CTX_RB3D_COLORPITCH] = rmesa->state.color.drawPitch; +} + + +/* ================================================================ + * Buffer clear + */ +#define R200_MAX_CLEARS 256 + +static void r200Clear( GLcontext *ctx, GLbitfield mask, GLboolean all, + GLint cx, GLint cy, GLint cw, GLint ch ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; + GLuint flags = 0; + GLuint color_mask = 0; + GLint ret, i; + + if ( R200_DEBUG & DEBUG_IOCTL ) { + fprintf( stderr, "%s: all=%d cx=%d cy=%d cw=%d ch=%d\n", + __FUNCTION__, all, cx, cy, cw, ch ); + } + + { + LOCK_HARDWARE( rmesa ); + UNLOCK_HARDWARE( rmesa ); + if ( dPriv->numClipRects == 0 ) + return; + } + + r200EmitState( rmesa ); + + /* Need to cope with lostcontext here as kernel relies on + * some residual state: + */ + R200_FIREVERTICES( rmesa ); + + if ( mask & DD_FRONT_LEFT_BIT ) { + flags |= RADEON_FRONT; + color_mask = rmesa->hw.msk.cmd[MSK_RB3D_PLANEMASK]; + mask &= ~DD_FRONT_LEFT_BIT; + } + + if ( mask & DD_BACK_LEFT_BIT ) { + flags |= RADEON_BACK; + color_mask = rmesa->hw.msk.cmd[MSK_RB3D_PLANEMASK]; + mask &= ~DD_BACK_LEFT_BIT; + } + + if ( mask & DD_DEPTH_BIT ) { + if ( ctx->Depth.Mask ) flags |= RADEON_DEPTH; /* FIXME: ??? */ + mask &= ~DD_DEPTH_BIT; + } + + if ( (mask & DD_STENCIL_BIT) && rmesa->state.stencil.hwBuffer ) { + flags |= RADEON_STENCIL; + mask &= ~DD_STENCIL_BIT; + } + + if ( mask ) + _swrast_Clear( ctx, mask, all, cx, cy, cw, ch ); + + if ( !flags ) + return; + + /* Flip top to bottom */ + cx += dPriv->x; + cy = dPriv->y + dPriv->h - cy - ch; + + LOCK_HARDWARE( rmesa ); + + /* Throttle the number of clear ioctls we do. + */ + while ( 1 ) { + drmRadeonGetParam gp; + int ret; + int clear; + + gp.param = RADEON_PARAM_LAST_CLEAR; + gp.value = (int *)&clear; + ret = drmCommandWriteRead( rmesa->dri.fd, + DRM_RADEON_GETPARAM, &gp, sizeof(gp) ); + + if ( ret ) { + fprintf( stderr, "%s: drmRadeonGetParam: %d\n", __FUNCTION__, ret ); + exit(1); + } + + if ( rmesa->sarea->last_clear - clear <= R200_MAX_OUTSTANDING ) { + break; + } + + if (rmesa->do_usleeps) { + UNLOCK_HARDWARE( rmesa ); + do_usleep(1, __FUNCTION__); + LOCK_HARDWARE( rmesa ); + } + } + + + for ( i = 0 ; i < dPriv->numClipRects ; ) { + GLint nr = MIN2( i + RADEON_NR_SAREA_CLIPRECTS, dPriv->numClipRects ); + XF86DRIClipRectPtr box = dPriv->pClipRects; + XF86DRIClipRectPtr b = rmesa->sarea->boxes; + drmRadeonClearType clear; + drmRadeonClearRect depth_boxes[RADEON_NR_SAREA_CLIPRECTS]; + GLint n = 0; + + if ( !all ) { + for ( ; i < nr ; i++ ) { + GLint x = box[i].x1; + GLint y = box[i].y1; + GLint w = box[i].x2 - x; + GLint h = box[i].y2 - y; + + if ( x < cx ) w -= cx - x, x = cx; + if ( y < cy ) h -= cy - y, y = cy; + if ( x + w > cx + cw ) w = cx + cw - x; + if ( y + h > cy + ch ) h = cy + ch - y; + if ( w <= 0 ) continue; + if ( h <= 0 ) continue; + + b->x1 = x; + b->y1 = y; + b->x2 = x + w; + b->y2 = y + h; + b++; + n++; + } + } else { + for ( ; i < nr ; i++ ) { + *b++ = box[i]; + n++; + } + } + + rmesa->sarea->nbox = n; + + clear.flags = flags; + clear.clear_color = rmesa->state.color.clear; + clear.clear_depth = 0; /* not used */ + clear.color_mask = rmesa->hw.msk.cmd[MSK_RB3D_PLANEMASK]; + clear.depth_mask = rmesa->state.stencil.clear; + clear.depth_boxes = depth_boxes; + + n--; + b = rmesa->sarea->boxes; + for ( ; n >= 0 ; n-- ) { + depth_boxes[n].f[RADEON_CLEAR_X1] = (float)b[n].x1; + depth_boxes[n].f[RADEON_CLEAR_Y1] = (float)b[n].y1; + depth_boxes[n].f[RADEON_CLEAR_X2] = (float)b[n].x2; + depth_boxes[n].f[RADEON_CLEAR_Y2] = (float)b[n].y2; + depth_boxes[n].f[RADEON_CLEAR_DEPTH] = ctx->Depth.Clear; + } + + ret = drmCommandWrite( rmesa->dri.fd, DRM_RADEON_CLEAR, + &clear, sizeof(drmRadeonClearType)); + + + if ( ret ) { + UNLOCK_HARDWARE( rmesa ); + fprintf( stderr, "DRM_RADEON_CLEAR: return = %d\n", ret ); + exit( 1 ); + } + } + + UNLOCK_HARDWARE( rmesa ); + rmesa->lost_context = 1; +} + + +void r200WaitForIdleLocked( r200ContextPtr rmesa ) +{ + int fd = rmesa->dri.fd; + int to = 0; + int ret, i = 0; + +/* fprintf(stderr, "%s\n", __FUNCTION__); */ + do { + do { + ret = drmCommandNone( fd, DRM_RADEON_CP_IDLE); + } while ( ret && errno == EBUSY && i++ < R200_IDLE_RETRY ); + if (ret == -EBUSY && rmesa->do_usleeps) { + UNLOCK_HARDWARE( rmesa ); + do_usleep(1, __FUNCTION__); + LOCK_HARDWARE( rmesa ); + } + } while ( ( ret == -EBUSY ) && ( to++ < R200_TIMEOUT ) ); + + if ( ret < 0 ) { + UNLOCK_HARDWARE( rmesa ); + fprintf( stderr, "Error: R200 timed out... exiting\n" ); + exit( -1 ); + } +} + + +static void r200WaitForIdle( r200ContextPtr rmesa ) +{ + LOCK_HARDWARE(rmesa); + r200WaitForIdleLocked( rmesa ); + UNLOCK_HARDWARE(rmesa); +} + +void r200GetAllParams( r200ContextPtr rmesa ) +{ + int ret; + drmRadeonGetParam gp; + + gp.param = RADEON_PARAM_AGP_BUFFER_OFFSET; + gp.value = &rmesa->dri.agp_buffer_offset; + + ret = drmCommandWriteRead( rmesa->dri.fd, + DRM_RADEON_GETPARAM, + &gp, sizeof(gp)); + if (ret) { + fprintf(stderr, "drmR200GetParam: %d\n", ret); + exit(1); + } + if (0) fprintf(stderr, "drmR200GetParam: agp_buffer_offset 0x%x\n", + *(gp.value)); + + rmesa->dri.agp_texture_offset = rmesa->dri.agp_buffer_offset + 2*1024*1024; + + +} + +void r200Flush( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + if (R200_DEBUG & DEBUG_IOCTL) + fprintf(stderr, "%s\n", __FUNCTION__); + + if (rmesa->dma.flush) + rmesa->dma.flush( rmesa ); + + if (!is_empty_list(&rmesa->hw.dirty)) + r200EmitState( rmesa ); + + if (rmesa->store.cmd_used) + r200FlushCmdBuf( rmesa, __FUNCTION__ ); +} + +/* Make sure all commands have been sent to the hardware and have + * completed processing. + */ +void r200Finish( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + r200Flush( ctx ); + r200WaitForIdle( rmesa ); +} + + +void r200InitIoctlFuncs( GLcontext *ctx ) +{ + ctx->Driver.Clear = r200Clear; + ctx->Driver.Finish = r200Finish; + ctx->Driver.Flush = r200Flush; + + r200GetAllParams( R200_CONTEXT( ctx ) ); +} + diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_ioctl.h b/xc/lib/GL/mesa/src/drv/r200/r200_ioctl.h new file mode 100644 index 000000000..6079b8fdf --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_ioctl.h @@ -0,0 +1,166 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_IOCTL_H__ +#define __R200_IOCTL_H__ + +#ifdef GLX_DIRECT_RENDERING + +#include "simple_list.h" +#include "radeon_dri.h" +#include "r200_lock.h" + +#include "xf86drm.h" +#include "radeon_common.h" + +extern void r200EmitState( r200ContextPtr rmesa ); +extern void r200EmitVertexAOS( r200ContextPtr rmesa, + GLuint vertex_size, + GLuint offset ); + +extern void r200EmitVbufPrim( r200ContextPtr rmesa, + GLuint primitive, + GLuint vertex_nr ); + +extern void r200FlushElts( r200ContextPtr rmesa ); + +extern GLushort *r200AllocEltsOpenEnded( r200ContextPtr rmesa, + GLuint primitive, + GLuint min_nr ); + +extern void r200EmitAOS( r200ContextPtr rmesa, + struct r200_dma_region **regions, + GLuint n, + GLuint offset ); + + + +extern void r200FlushCmdBuf( r200ContextPtr rmesa, const char * ); +extern void r200RefillCurrentDmaRegion( r200ContextPtr rmesa ); + +extern void r200AllocDmaRegion( r200ContextPtr rmesa, + struct r200_dma_region *region, + int bytes, + int alignment ); + +extern void r200AllocDmaRegionVerts( r200ContextPtr rmesa, + struct r200_dma_region *region, + int numverts, + int vertsize, + int alignment ); + +extern void r200ReleaseDmaRegion( r200ContextPtr rmesa, + struct r200_dma_region *region, + const char *caller ); + +extern void r200CopyBuffer( const __DRIdrawablePrivate *drawable ); +extern void r200PageFlip( const __DRIdrawablePrivate *drawable ); +extern void r200Flush( GLcontext *ctx ); +extern void r200Finish( GLcontext *ctx ); +extern void r200WaitForIdleLocked( r200ContextPtr rmesa ); +extern void r200InitIoctlFuncs( GLcontext *ctx ); +extern void r200GetAllParams( r200ContextPtr rmesa ); + +/* ================================================================ + * Helper macros: + */ + +/* Close off the last primitive, if it exists. + */ +#define R200_NEWPRIM( rmesa ) \ +do { \ + if ( rmesa->dma.flush ) \ + rmesa->dma.flush( rmesa ); \ +} while (0) + +/* Can accomodate several state changes and primitive changes without + * actually firing the buffer. + */ +#define R200_STATECHANGE( rmesa, ATOM ) \ +do { \ + R200_NEWPRIM( rmesa ); \ + move_to_head( &(rmesa->hw.dirty), &(rmesa->hw.ATOM)); \ +} while (0) + +#define R200_DB_STATE( ATOM ) \ + memcpy( rmesa->hw.ATOM.lastcmd, rmesa->hw.ATOM.cmd, \ + rmesa->hw.ATOM.cmd_size * 4) + +static __inline int R200_DB_STATECHANGE( + r200ContextPtr rmesa, + struct r200_state_atom *atom ) +{ + if (memcmp(atom->cmd, atom->lastcmd, atom->cmd_size*4)) { + int *tmp; + R200_NEWPRIM( rmesa ); + move_to_head( &(rmesa->hw.dirty), atom ); + tmp = atom->cmd; + atom->cmd = atom->lastcmd; + atom->lastcmd = tmp; + return 1; + } + else + return 0; +} + + +/* Fire the buffered vertices no matter what. + */ +#define R200_FIREVERTICES( rmesa ) \ +do { \ + if ( rmesa->store.cmd_used || rmesa->dma.flush ) { \ + r200Flush( rmesa->glCtx ); \ + } \ +} while (0) + +/* Alloc space in the command buffer + */ +static __inline char *r200AllocCmdBuf( r200ContextPtr rmesa, + int bytes, const char *where ) +{ + if (rmesa->store.cmd_used + bytes > R200_CMD_BUF_SZ) + r200FlushCmdBuf( rmesa, __FUNCTION__ ); + + { + char *head = rmesa->store.cmd_buf + rmesa->store.cmd_used; + rmesa->store.cmd_used += bytes; + return head; + } +} + + + + +#endif +#endif /* __R200_IOCTL_H__ */ diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_lock.c b/xc/lib/GL/mesa/src/drv/r200/r200_lock.c new file mode 100644 index 000000000..0d4db5647 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_lock.c @@ -0,0 +1,108 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_lock.h" +#include "r200_tex.h" +#include "r200_state.h" +#include "r200_ioctl.h" + +#if DEBUG_LOCKING +char *prevLockFile = NULL; +int prevLockLine = 0; +#endif + +/* Turn on/off page flipping according to the flags in the sarea: + */ +static void +r200UpdatePageFlipping( r200ContextPtr rmesa ) +{ + int use_back; + rmesa->doPageFlip = rmesa->sarea->pfAllowPageFlip; + + use_back = (rmesa->glCtx->Color.DriverDrawBuffer == GL_BACK_LEFT); + use_back ^= (rmesa->sarea->pfCurrentPage == 1); + + if (use_back) { + rmesa->state.color.drawOffset = rmesa->r200Screen->backOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->backPitch; + } else { + rmesa->state.color.drawOffset = rmesa->r200Screen->frontOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->frontPitch; + } + + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_RB3D_COLOROFFSET] = rmesa->state.color.drawOffset; + rmesa->hw.ctx.cmd[CTX_RB3D_COLORPITCH] = rmesa->state.color.drawPitch; +} + + + +/* Update the hardware state. This is called if another context has + * grabbed the hardware lock, which includes the X server. This + * function also updates the driver's window state after the X server + * moves, resizes or restacks a window -- the change will be reflected + * in the drawable position and clip rects. Since the X server grabs + * the hardware lock when it changes the window state, this routine will + * automatically be called after such a change. + */ +void r200GetLock( r200ContextPtr rmesa, GLuint flags ) +{ + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; + __DRIscreenPrivate *sPriv = rmesa->dri.screen; + RADEONSAREAPrivPtr sarea = rmesa->sarea; + + drmGetLock( rmesa->dri.fd, rmesa->dri.hwContext, flags ); + + /* The window might have moved, so we might need to get new clip + * rects. + * + * NOTE: This releases and regrabs the hw lock to allow the X server + * to respond to the DRI protocol request for new drawable info. + * Since the hardware state depends on having the latest drawable + * clip rects, all state checking must be done _after_ this call. + */ + DRI_VALIDATE_DRAWABLE_INFO( rmesa->dri.display, sPriv, dPriv ); + + if ( rmesa->lastStamp != dPriv->lastStamp ) { + r200UpdatePageFlipping( rmesa ); + r200SetCliprects( rmesa, rmesa->glCtx->Color.DriverDrawBuffer ); + r200UpdateViewportOffset( rmesa->glCtx ); + rmesa->lastStamp = dPriv->lastStamp; + } + + if ( sarea->ctxOwner != rmesa->dri.hwContext ) { + sarea->ctxOwner = rmesa->dri.hwContext; + } +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_lock.h b/xc/lib/GL/mesa/src/drv/r200/r200_lock.h new file mode 100644 index 000000000..d0b20bf8a --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_lock.h @@ -0,0 +1,111 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_LOCK_H__ +#define __R200_LOCK_H__ + +#ifdef GLX_DIRECT_RENDERING + +extern void r200GetLock( r200ContextPtr rmesa, GLuint flags ); + +/* Turn DEBUG_LOCKING on to find locking conflicts. + */ +#define DEBUG_LOCKING 0 + +#if DEBUG_LOCKING +extern char *prevLockFile; +extern int prevLockLine; + +#define DEBUG_LOCK() \ + do { \ + prevLockFile = (__FILE__); \ + prevLockLine = (__LINE__); \ + } while (0) + +#define DEBUG_RESET() \ + do { \ + prevLockFile = 0; \ + prevLockLine = 0; \ + } while (0) + +#define DEBUG_CHECK_LOCK() \ + do { \ + if ( prevLockFile ) { \ + fprintf( stderr, \ + "LOCK SET!\n\tPrevious %s:%d\n\tCurrent: %s:%d\n", \ + prevLockFile, prevLockLine, __FILE__, __LINE__ ); \ + exit( 1 ); \ + } \ + } while (0) + +#else + +#define DEBUG_LOCK() +#define DEBUG_RESET() +#define DEBUG_CHECK_LOCK() + +#endif + +/* + * !!! We may want to separate locks from locks with validation. This + * could be used to improve performance for those things commands that + * do not do any drawing !!! + */ + + +/* Lock the hardware and validate our state. + */ +#define LOCK_HARDWARE( rmesa ) \ + do { \ + char __ret = 0; \ + DEBUG_CHECK_LOCK(); \ + DRM_CAS( rmesa->dri.hwLock, rmesa->dri.hwContext, \ + (DRM_LOCK_HELD | rmesa->dri.hwContext), __ret ); \ + if ( __ret ) \ + r200GetLock( rmesa, 0 ); \ + DEBUG_LOCK(); \ + } while (0) + +/* Unlock the hardware. + */ +#define UNLOCK_HARDWARE( rmesa ) \ + do { \ + DRM_UNLOCK( rmesa->dri.fd, \ + rmesa->dri.hwLock, \ + rmesa->dri.hwContext ); \ + DEBUG_RESET(); \ + } while (0) + +#endif +#endif /* __R200_LOCK_H__ */ diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_maos.c b/xc/lib/GL/mesa/src/drv/r200/r200_maos.c new file mode 100644 index 000000000..fd2bd5102 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_maos.c @@ -0,0 +1,12 @@ + + +/* If using new packets, can choose either verts or arrays. + * Otherwise, must use verts. + */ +#include "r200_context.h" +#define R200_MAOS_VERTS 0 +#if (R200_MAOS_VERTS) || (R200_OLD_PACKETS) +#include "r200_maos_verts.c" +#else +#include "r200_maos_arrays.c" +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_maos.h b/xc/lib/GL/mesa/src/drv/r200/r200_maos.h new file mode 100644 index 000000000..1f7b1faa5 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_maos.h @@ -0,0 +1,46 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_MAOS_H__ +#define __R200_MAOS_H__ + +#ifdef GLX_DIRECT_RENDERING + +#include "r200_context.h" + +extern void r200EmitArrays( GLcontext *ctx, GLuint inputs ); +extern void r200ReleaseArrays( GLcontext *ctx, GLuint newinputs ); + +#endif +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_maos_arrays.c b/xc/lib/GL/mesa/src/drv/r200/r200_maos_arrays.c new file mode 100644 index 000000000..765a693b5 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_maos_arrays.c @@ -0,0 +1,478 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "glheader.h" +#include "mtypes.h" +#include "colormac.h" +#include "mem.h" +#include "mmath.h" +#include "macros.h" + +#include "swrast_setup/swrast_setup.h" +#include "math/m_translate.h" +#include "tnl/tnl.h" +#include "tnl/t_context.h" +#include "tnl/t_imm_debug.h" + +#include "r200_context.h" +#include "r200_ioctl.h" +#include "r200_state.h" +#include "r200_swtcl.h" +#include "r200_maos.h" + +/* Usage: + * - from r200_tcl_render + * - call r200EmitArrays to ensure uptodate arrays in dma + * - emit primitives (new type?) which reference the data + * -- need to use elts for lineloop, quads, quadstrip/flat + * -- other primitives are all well-formed (need tristrip-1,fake-poly) + * + */ +static void emit_ubyte_rgba3( GLcontext *ctx, + struct r200_dma_region *rvb, + char *data, + int stride, + int count ) +{ + int i; + char *out = (char *)(rvb->start + rvb->address); + + if (R200_DEBUG & DEBUG_VERTS) + fprintf(stderr, "%s count %d stride %d out %p\n", + __FUNCTION__, count, stride, out); + + for (i = 0; i < count; i++) { + out[0] = *data; + out[1] = *(data+1); + out[2] = *(data+2); + out[3] = 0xFF; + out += 4; + data += stride; + } +} + + +#if defined(USE_X86_ASM) +#define COPY_DWORDS( dst, src, nr ) \ +do { \ + int __tmp; \ + __asm__ __volatile__( "rep ; movsl" \ + : "=%c" (__tmp), "=D" (dst), "=S" (__tmp) \ + : "0" (nr), \ + "D" ((long)dst), \ + "S" ((long)src) ); \ +} while (0) +#else +#define COPY_DWORDS( dst, src, nr ) \ +do { \ + int j; \ + for ( j = 0 ; j < nr ; j++ ) \ + dst[j] = ((int *)v)[j]; \ + dst += nr; \ +} while (0) +#endif + + + +static void emit_ubyte_rgba4( GLcontext *ctx, + struct r200_dma_region *rvb, + char *data, + int stride, + int count ) +{ + int i; + int *out = (int *)(rvb->address + rvb->start); + + if (R200_DEBUG & DEBUG_VERTS) + fprintf(stderr, "%s count %d stride %d\n", + __FUNCTION__, count, stride); + + if (stride == 4) + COPY_DWORDS( out, data, count ); + else + for (i = 0; i < count; i++) { + *out++ = *(int *)data; + data += stride; + } +} + + +static void emit_ubyte_rgba( GLcontext *ctx, + struct r200_dma_region *rvb, + char *data, + int size, + int stride, + int count ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_VERTS) + fprintf(stderr, "%s %d/%d\n", __FUNCTION__, count, size); + + assert (!rvb->buf); + + if (stride == 0) { + r200AllocDmaRegion( rmesa, rvb, 4, 4 ); + count = 1; + rvb->aos_start = GET_START(rvb); + rvb->aos_stride = 0; + rvb->aos_size = 1; + } + else { + r200AllocDmaRegion( rmesa, rvb, 4 * count, 4 ); /* alignment? */ + rvb->aos_start = GET_START(rvb); + rvb->aos_stride = 1; + rvb->aos_size = 1; + } + + /* Emit the data + */ + switch (size) { + case 3: + emit_ubyte_rgba3( ctx, rvb, data, stride, count ); + break; + case 4: + emit_ubyte_rgba4( ctx, rvb, data, stride, count ); + break; + default: + assert(0); + exit(1); + break; + } +} + + + + +static void emit_vec8( GLcontext *ctx, + struct r200_dma_region *rvb, + char *data, + int stride, + int count ) +{ + int i; + int *out = (int *)(rvb->address + rvb->start); + + if (R200_DEBUG & DEBUG_VERTS) + fprintf(stderr, "%s count %d stride %d\n", + __FUNCTION__, count, stride); + + if (stride == 8) + COPY_DWORDS( out, data, count*2 ); + else + for (i = 0; i < count; i++) { + out[0] = *(int *)data; + out[1] = *(int *)(data+4); + out += 2; + data += stride; + } +} + +static void emit_vec12( GLcontext *ctx, + struct r200_dma_region *rvb, + char *data, + int stride, + int count ) +{ + int i; + int *out = (int *)(rvb->address + rvb->start); + + if (R200_DEBUG & DEBUG_VERTS) + fprintf(stderr, "%s count %d stride %d out %p data %p\n", + __FUNCTION__, count, stride, out, data); + + if (stride == 12) + COPY_DWORDS( out, data, count*3 ); + else + for (i = 0; i < count; i++) { + out[0] = *(int *)data; + out[1] = *(int *)(data+4); + out[2] = *(int *)(data+8); + out += 3; + data += stride; + } +} + +static void emit_vec16( GLcontext *ctx, + struct r200_dma_region *rvb, + char *data, + int stride, + int count ) +{ + int i; + int *out = (int *)(rvb->address + rvb->start); + + if (R200_DEBUG & DEBUG_VERTS) + fprintf(stderr, "%s count %d stride %d\n", + __FUNCTION__, count, stride); + + if (stride == 16) + COPY_DWORDS( out, data, count*4 ); + else + for (i = 0; i < count; i++) { + out[0] = *(int *)data; + out[1] = *(int *)(data+4); + out[2] = *(int *)(data+8); + out[3] = *(int *)(data+12); + out += 4; + data += stride; + } +} + + +static void emit_vector( GLcontext *ctx, + struct r200_dma_region *rvb, + char *data, + int size, + int stride, + int count ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_VERTS) + fprintf(stderr, "%s count %d size %d stride %d\n", + __FUNCTION__, count, size, stride); + + assert (!rvb->buf); + + if (stride == 0) { + r200AllocDmaRegion( rmesa, rvb, size * 4, 4 ); + count = 1; + rvb->aos_start = GET_START(rvb); + rvb->aos_stride = 0; + rvb->aos_size = size; + } + else { + r200AllocDmaRegion( rmesa, rvb, size * count * 4, 4 ); /* alignment? */ + rvb->aos_start = GET_START(rvb); + rvb->aos_stride = size; + rvb->aos_size = size; + } + + /* Emit the data + */ + switch (size) { + case 2: + emit_vec8( ctx, rvb, data, stride, count ); + break; + case 3: + emit_vec12( ctx, rvb, data, stride, count ); + break; + case 4: + emit_vec16( ctx, rvb, data, stride, count ); + break; + default: + assert(0); + exit(1); + break; + } + +} + + + +/* Emit any changed arrays to new agp memory, re-emit a packet to + * update the arrays. + */ +void r200EmitArrays( GLcontext *ctx, GLuint inputs ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + struct vertex_buffer *VB = &TNL_CONTEXT( ctx )->vb; + struct r200_dma_region **component = rmesa->tcl.aos_components; + GLuint nr = 0; + GLuint vfmt0 = 0, vfmt1 = 0; + GLuint count = VB->Count; + + if (R200_DEBUG & DEBUG_VERTS) + _tnl_print_vert_flags( __FUNCTION__, inputs ); + + if (1) { + if (!rmesa->tcl.obj.buf) + emit_vector( ctx, + &rmesa->tcl.obj, + (char *)VB->ObjPtr->data, + VB->ObjPtr->size, + VB->ObjPtr->stride, + count); + + switch( VB->ObjPtr->size ) { + case 4: vfmt0 |= R200_VTX_W0; + case 3: vfmt0 |= R200_VTX_Z0; + case 2: + default: + } + component[nr++] = &rmesa->tcl.obj; + } + + + if (inputs & VERT_NORM) { + if (!rmesa->tcl.norm.buf) + emit_vector( ctx, + &(rmesa->tcl.norm), + (char *)VB->NormalPtr->data, + 3, + VB->NormalPtr->stride, + count); + + vfmt0 |= R200_VTX_N0; + component[nr++] = &rmesa->tcl.norm; + } + + if (inputs & VERT_RGBA) { + if (VB->ColorPtr[0]->Type == GL_UNSIGNED_BYTE) { + if (!rmesa->tcl.rgba.buf) + emit_ubyte_rgba( ctx, + &rmesa->tcl.rgba, + (char *)VB->ColorPtr[0]->Ptr, + VB->ColorPtr[0]->Size, + VB->ColorPtr[0]->StrideB, + count); + + vfmt0 |= R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT; + } + else { + int emitsize; + + if (VB->ColorPtr[0]->Size == 4 && + (VB->ColorPtr[0]->StrideB != 0 || + ((GLfloat *)VB->ColorPtr[0]->Ptr)[3] != 1.0)) { + vfmt0 |= R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT; + emitsize = 4; + } + else { + vfmt0 |= R200_VTX_FP_RGB << R200_VTX_COLOR_0_SHIFT; + emitsize = 3; + } + + if (!rmesa->tcl.rgba.buf) + emit_vector( ctx, + &(rmesa->tcl.rgba), + (char *)VB->ColorPtr[0]->Ptr, + emitsize, + VB->ColorPtr[0]->StrideB, + count); + } + + component[nr++] = &rmesa->tcl.rgba; + } + + + if (inputs & VERT_SPEC_RGB) { + if (!rmesa->tcl.spec.buf) { + if (VB->SecondaryColorPtr[0]->Type != GL_UNSIGNED_BYTE) + r200_import_float_spec_colors( ctx ); + + emit_ubyte_rgba( ctx, + &rmesa->tcl.spec, + (char *)VB->SecondaryColorPtr[0]->Ptr, + 3, + VB->SecondaryColorPtr[0]->StrideB, + count); + } + + /* How does this work? + */ + vfmt0 |= R200_VTX_PK_RGBA << R200_VTX_COLOR_1_SHIFT; + component[nr++] = &rmesa->tcl.spec; + } + +/* vtx = (rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] & */ +/* ~(R200_TCL_VTX_Q0|R200_TCL_VTX_Q1)); */ + + if (inputs & VERT_TEX0) { + if (!rmesa->tcl.tex[0].buf) + emit_vector( ctx, + &(rmesa->tcl.tex[0]), + (char *)VB->TexCoordPtr[0]->data, + VB->TexCoordPtr[0]->size, + VB->TexCoordPtr[0]->stride, + count ); + + vfmt1 |= VB->TexCoordPtr[0]->size << R200_VTX_TEX0_COMP_CNT_SHIFT; + component[nr++] = &rmesa->tcl.tex[0]; + } + + if (inputs & VERT_TEX1) { + if (!rmesa->tcl.tex[1].buf) + emit_vector( ctx, + &(rmesa->tcl.tex[1]), + (char *)VB->TexCoordPtr[1]->data, + VB->TexCoordPtr[1]->size, + VB->TexCoordPtr[1]->stride, + count ); + + vfmt1 |= VB->TexCoordPtr[1]->size << R200_VTX_TEX1_COMP_CNT_SHIFT; + component[nr++] = &rmesa->tcl.tex[1]; + } + + if (vfmt0 != rmesa->hw.vtx.cmd[VTX_VTXFMT_0] || + vfmt1 != rmesa->hw.vtx.cmd[VTX_VTXFMT_1]) { + R200_STATECHANGE( rmesa, vtx ); + rmesa->hw.vtx.cmd[VTX_VTXFMT_0] = vfmt0; + rmesa->hw.vtx.cmd[VTX_VTXFMT_1] = vfmt1; + } + + +/* fprintf(stderr, "VTXFMT_0: %x VTXFMT_1: %x\n", vfmt0, vfmt1); */ + + rmesa->tcl.nr_aos_components = nr; + rmesa->tcl.vertex_format = vfmt0; +} + + +void r200ReleaseArrays( GLcontext *ctx, GLuint newinputs ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + if (R200_DEBUG & DEBUG_VERTS) + _tnl_print_vert_flags( __FUNCTION__, newinputs ); + + if (newinputs & VERT_OBJ) + r200ReleaseDmaRegion( rmesa, &rmesa->tcl.obj, __FUNCTION__ ); + + if (newinputs & VERT_NORM) + r200ReleaseDmaRegion( rmesa, &rmesa->tcl.norm, __FUNCTION__ ); + + if (newinputs & VERT_RGBA) + r200ReleaseDmaRegion( rmesa, &rmesa->tcl.rgba, __FUNCTION__ ); + + if (newinputs & VERT_SPEC_RGB) + r200ReleaseDmaRegion( rmesa, &rmesa->tcl.spec, __FUNCTION__ ); + + if (newinputs & VERT_TEX0) + r200ReleaseDmaRegion( rmesa, &rmesa->tcl.tex[0], __FUNCTION__ ); + + if (newinputs & VERT_TEX1) + r200ReleaseDmaRegion( rmesa, &rmesa->tcl.tex[1], __FUNCTION__ ); +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_maos_vbtmp.h b/xc/lib/GL/mesa/src/drv/r200/r200_maos_vbtmp.h new file mode 100644 index 000000000..e8edb3c1d --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_maos_vbtmp.h @@ -0,0 +1,378 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef LOCALVARS +#define LOCALVARS +#endif + +#undef TCL_DEBUG +#ifndef TCL_DEBUG +#define TCL_DEBUG 0 +#endif + +static void TAG(emit)( GLcontext *ctx, + GLuint start, GLuint end, + void *dest ) +{ + LOCALVARS + struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb; + GLuint (*tc0)[4], (*tc1)[4]; + GLfloat *fog; + GLuint (*tc2)[4], (*norm)[3]; + GLubyte (*col)[4], (*spec)[4]; + GLuint tc0_stride, tc1_stride, col_stride, spec_stride, fog_stride; + GLuint tc2_stride, norm_stride; + GLuint (*coord)[4]; + GLuint coord_stride; + GLubyte dummy[4]; + int i; + + union emit_union *v = (union emit_union *)dest; + + + if (R200_DEBUG & DEBUG_VERTS) + fprintf(stderr, "%s\n", __FUNCTION__); + + /* The vertex code expects Obj to be clean to element 3. To fix + * this, add more vertex code (for obj-2, obj-3) or preferably move + * to maos. + */ + if (VB->ObjPtr->size < 3) { + if (VB->ObjPtr->flags & VEC_NOT_WRITEABLE) { + VB->import_data( ctx, VERT_OBJ, VEC_NOT_WRITEABLE ); + } + _mesa_vector4f_clean_elem( VB->ObjPtr, VB->Count, 2 ); + } + + if (DO_W && VB->ObjPtr->size < 4) { + if (VB->ObjPtr->flags & VEC_NOT_WRITEABLE) { + VB->import_data( ctx, VERT_OBJ, VEC_NOT_WRITEABLE ); + } + _mesa_vector4f_clean_elem( VB->ObjPtr, VB->Count, 3 ); + } + + coord = (GLuint (*)[4])VB->ObjPtr->data; + coord_stride = VB->ObjPtr->stride; + + if (DO_TEX2) { + const GLuint t2 = GET_TEXSOURCE(2); + tc2 = (GLuint (*)[4])VB->TexCoordPtr[t2]->data; + tc2_stride = VB->TexCoordPtr[t2]->stride; + if (DO_PTEX && VB->TexCoordPtr[t2]->size < 4) { + if (VB->TexCoordPtr[t2]->flags & VEC_NOT_WRITEABLE) { + VB->import_data( ctx, VERT_TEX2, VEC_NOT_WRITEABLE ); + } + _mesa_vector4f_clean_elem( VB->TexCoordPtr[t2], VB->Count, 3 ); + } + } + + if (DO_TEX1) { + if (VB->TexCoordPtr[1]) { + const GLuint t1 = GET_TEXSOURCE(1); + tc1 = (GLuint (*)[4])VB->TexCoordPtr[t1]->data; + tc1_stride = VB->TexCoordPtr[t1]->stride; + if (DO_PTEX && VB->TexCoordPtr[t1]->size < 4) { + if (VB->TexCoordPtr[t1]->flags & VEC_NOT_WRITEABLE) { + VB->import_data( ctx, VERT_TEX1, VEC_NOT_WRITEABLE ); + } + _mesa_vector4f_clean_elem( VB->TexCoordPtr[t1], VB->Count, 3 ); + } + } else { + tc1 = (GLuint (*)[4])&ctx->Current.Texcoord[1]; /* could be anything, really */ + tc1_stride = 0; + } + } + + if (DO_TEX0) { + if (VB->TexCoordPtr[0]) { + const GLuint t0 = GET_TEXSOURCE(0); + tc0_stride = VB->TexCoordPtr[t0]->stride; + tc0 = (GLuint (*)[4])VB->TexCoordPtr[t0]->data; + if (DO_PTEX && VB->TexCoordPtr[t0]->size < 4) { + if (VB->TexCoordPtr[t0]->flags & VEC_NOT_WRITEABLE) { + VB->import_data( ctx, VERT_TEX0, VEC_NOT_WRITEABLE ); + } + _mesa_vector4f_clean_elem( VB->TexCoordPtr[t0], VB->Count, 3 ); + } + } else { + tc0 = (GLuint (*)[4])&ctx->Current.Texcoord[0]; /* could be anything, really */ + tc0_stride = 0; + } + + } + + if (DO_NORM) { + if (VB->NormalPtr) { + norm_stride = VB->NormalPtr->stride; + norm = (GLuint (*)[3])VB->NormalPtr->data; + } else { + norm_stride = 0; + norm = (GLuint (*)[3])&ctx->Current.Normal; + } + } + + if (DO_RGBA) { + if (VB->ColorPtr[0]) { + /* This is incorrect when colormaterial is enabled: + */ + if (VB->ColorPtr[0]->Type != GL_UNSIGNED_BYTE) { + if (0) fprintf(stderr, "IMPORTING FLOAT COLORS\n"); + IMPORT_FLOAT_COLORS( ctx ); + } + col = (GLubyte (*)[4])VB->ColorPtr[0]->Ptr; + col_stride = VB->ColorPtr[0]->StrideB; + } else { + col = &dummy; /* any old memory is fine */ + col_stride = 0; + } + + } + + if (DO_SPEC) { + if (VB->SecondaryColorPtr[0]) { + if (VB->SecondaryColorPtr[0]->Type != GL_UNSIGNED_BYTE) + IMPORT_FLOAT_SPEC_COLORS( ctx ); + spec = (GLubyte (*)[4])VB->SecondaryColorPtr[0]->Ptr; + spec_stride = VB->SecondaryColorPtr[0]->StrideB; + } else { + spec = &dummy; + spec_stride = 0; + } + + } + + if (DO_FOG) { + if (VB->FogCoordPtr) { + fog = VB->FogCoordPtr->data; + fog_stride = VB->FogCoordPtr->stride; + } else { + fog = (GLfloat *)&dummy; *fog = 0; + fog_stride = 0; + } + + } + + + if (VB->importable_data) { + if (start) { + coord = (GLuint (*)[4])((GLubyte *)coord + start * coord_stride); + if (DO_TEX0) + tc0 = (GLuint (*)[4])((GLubyte *)tc0 + start * tc0_stride); + if (DO_TEX1) + tc1 = (GLuint (*)[4])((GLubyte *)tc1 + start * tc1_stride); + if (DO_TEX2) + tc2 = (GLuint (*)[4])((GLubyte *)tc2 + start * tc2_stride); + if (DO_NORM) + norm = (GLuint (*)[3])((GLubyte *)norm + start * norm_stride); + if (DO_RGBA) + STRIDE_4UB(col, start * col_stride); + if (DO_SPEC) + STRIDE_4UB(spec, start * spec_stride); + if (DO_FOG) + STRIDE_F(fog, start * fog_stride); + } + + for (i=start; i < end; i++) { + v[0].ui = coord[0][0]; + v[1].ui = coord[0][1]; + v[2].ui = coord[0][2]; + if (TCL_DEBUG) fprintf(stderr, "%d: %.2f %.2f %.2f ", i, v[0].f, v[1].f, v[2].f); + if (DO_W) { + v[3].ui = coord[0][3]; + if (TCL_DEBUG) fprintf(stderr, "%.2f ", v[3].f); + v += 4; + } + else + v += 3; + coord = (GLuint (*)[4])((GLubyte *)coord + coord_stride); + + if (DO_NORM) { + v[0].ui = norm[0][0]; + v[1].ui = norm[0][1]; + v[2].ui = norm[0][2]; + if (TCL_DEBUG) fprintf(stderr, "norm: %.2f %.2f %.2f ", v[0].f, v[1].f, v[2].f); + v += 3; + norm = (GLuint (*)[3])((GLubyte *)norm + norm_stride); + } + if (DO_RGBA) { + v[0].ui = *(GLuint *)&col[0]; + STRIDE_4UB(col, col_stride); + if (TCL_DEBUG) fprintf(stderr, "%x ", v[0].ui); + v++; + } + if (DO_SPEC || DO_FOG) { + if (DO_SPEC) { + v[0].ub[0] = spec[0][0]; + v[0].ub[1] = spec[0][1]; + v[0].ub[2] = spec[0][2]; + STRIDE_4UB(spec, spec_stride); + } + if (DO_FOG) { + v[0].ub[3] = fog[0] * 255.0; + STRIDE_F(fog, fog_stride); + } + if (TCL_DEBUG) fprintf(stderr, "%x ", v[0].ui); + v++; + } + if (DO_TEX0) { + v[0].ui = tc0[0][0]; + v[1].ui = tc0[0][1]; + if (TCL_DEBUG) fprintf(stderr, "t0: %.2f %.2f ", v[0].f, v[1].f); + if (DO_PTEX) { + v[2].ui = tc0[0][3]; + if (TCL_DEBUG) fprintf(stderr, "%.2f ", v[2].f); + v += 3; + } + else + v += 2; + tc0 = (GLuint (*)[4])((GLubyte *)tc0 + tc0_stride); + } + if (DO_TEX1) { + v[0].ui = tc1[0][0]; + v[1].ui = tc1[0][1]; + if (TCL_DEBUG) fprintf(stderr, "t1: %.2f %.2f ", v[0].f, v[1].f); + if (DO_PTEX) { + v[2].ui = tc1[0][3]; + if (TCL_DEBUG) fprintf(stderr, "%.2f ", v[2].f); + v += 3; + } + else + v += 2; + tc1 = (GLuint (*)[4])((GLubyte *)tc1 + tc1_stride); + } + if (DO_TEX2) { + v[0].ui = tc2[0][0]; + v[1].ui = tc2[0][1]; + if (DO_PTEX) { + v[2].ui = tc2[0][3]; + v += 3; + } + else + v += 2; + tc2 = (GLuint (*)[4])((GLubyte *)tc2 + tc2_stride); + } + if (TCL_DEBUG) fprintf(stderr, "\n"); + } + } else { + for (i=start; i < end; i++) { + v[0].ui = coord[i][0]; + v[1].ui = coord[i][1]; + v[2].ui = coord[i][2]; + if (DO_W) { + v[3].ui = coord[i][3]; + v += 4; + } + else + v += 3; + + if (DO_NORM) { + v[0].ui = norm[i][0]; + v[1].ui = norm[i][1]; + v[2].ui = norm[i][2]; + v += 3; + } + if (DO_RGBA) { + v[0].ui = *(GLuint *)&col[i]; + v++; + } + if (DO_SPEC || DO_FOG) { + if (DO_SPEC) { + v[0].ub[0] = spec[i][0]; + v[0].ub[1] = spec[i][1]; + v[0].ub[2] = spec[i][2]; + } + if (DO_FOG) { + v[0].ub[3] = fog[i] * 255.0; + } + v++; + } + if (DO_TEX0) { + v[0].ui = tc0[i][0]; + v[1].ui = tc0[i][1]; + if (DO_PTEX) { + v[2].ui = tc0[i][3]; + v += 3; + } + else + v += 2; + } + if (DO_TEX1) { + v[0].ui = tc1[i][0]; + v[1].ui = tc1[i][1]; + if (DO_PTEX) { + v[2].ui = tc1[i][3]; + v += 3; + } + else + v += 2; + } + if (DO_TEX2) { + v[0].ui = tc2[i][0]; + v[1].ui = tc2[i][1]; + if (DO_PTEX) { + v[2].ui = tc2[i][3]; + v += 3; + } + else + v += 2; + } + } + } +} + + + +static void TAG(init)( void ) +{ + int sz = 3; + if (DO_W) sz++; + if (DO_NORM) sz += 3; + if (DO_RGBA) sz++; + if (DO_SPEC || DO_FOG) sz++; + if (DO_TEX0) sz += 2; + if (DO_TEX0 && DO_PTEX) sz++; + if (DO_TEX1) sz += 2; + if (DO_TEX1 && DO_PTEX) sz++; + if (DO_TEX2) sz += 2; + if (DO_TEX2 && DO_PTEX) sz++; + + setup_tab[IDX].emit = TAG(emit); + setup_tab[IDX].vertex_format = IND; + setup_tab[IDX].vertex_size = sz; +} + + +#undef IND +#undef TAG +#undef IDX diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_maos_verts.c b/xc/lib/GL/mesa/src/drv/r200/r200_maos_verts.c new file mode 100644 index 000000000..87ae7c48f --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_maos_verts.c @@ -0,0 +1,335 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_state.h" +#include "r200_ioctl.h" +#include "r200_tex.h" +#include "r200_tcl.h" +#include "r200_swtcl.h" +#include "r200_maos.h" + +#include "mmath.h" +#include "mtypes.h" +#include "enums.h" +#include "colormac.h" +#include "light.h" + +#include "array_cache/acache.h" +#include "tnl/tnl.h" +#include "tnl/t_pipeline.h" +#include "tnl/t_imm_debug.h" + +#define R200_TCL_MAX_SETUP 13 + +union emit_union { float f; GLuint ui; GLubyte ub[4]; }; + +static struct { + void (*emit)( GLcontext *, GLuint, GLuint, void * ); + GLuint vertex_size; + GLuint vertex_format; +} setup_tab[R200_TCL_MAX_SETUP]; + +#define DO_W (IND & R200_CP_VC_FRMT_W0) +#define DO_RGBA (IND & R200_CP_VC_FRMT_PKCOLOR) +#define DO_SPEC (IND & R200_CP_VC_FRMT_PKSPEC) +#define DO_FOG (IND & R200_CP_VC_FRMT_PKSPEC) +#define DO_TEX0 (IND & R200_CP_VC_FRMT_ST0) +#define DO_TEX1 (IND & R200_CP_VC_FRMT_ST1) +#define DO_PTEX (IND & R200_CP_VC_FRMT_Q0) +#define DO_NORM (IND & R200_CP_VC_FRMT_N0) + +#define DO_TEX2 0 +#define DO_TEX3 0 + +#define GET_TEXSOURCE(n) n +#define GET_UBYTE_COLOR_STORE() &R200_CONTEXT(ctx)->UbyteColor +#define GET_UBYTE_SPEC_COLOR_STORE() &R200_CONTEXT(ctx)->UbyteSecondaryColor + +#define IMPORT_FLOAT_COLORS r200_import_float_colors +#define IMPORT_FLOAT_SPEC_COLORS r200_import_float_spec_colors + +/*********************************************************************** + * Generate vertex emit functions * + ***********************************************************************/ + + +/* Defined in order of increasing vertex size: + */ +#define IDX 0 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR) +#define TAG(x) x##_rgba +#include "r200_maos_vbtmp.h" + +#define IDX 1 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_N0) +#define TAG(x) x##_n +#include "r200_maos_vbtmp.h" + +#define IDX 2 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_ST0) +#define TAG(x) x##_rgba_st +#include "r200_maos_vbtmp.h" + +#define IDX 3 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_N0) +#define TAG(x) x##_rgba_n +#include "r200_maos_vbtmp.h" + +#define IDX 4 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_N0) +#define TAG(x) x##_st_n +#include "r200_maos_vbtmp.h" + +#define IDX 5 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_ST1) +#define TAG(x) x##_rgba_st_st +#include "r200_maos_vbtmp.h" + +#define IDX 6 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_N0) +#define TAG(x) x##_rgba_st_n +#include "r200_maos_vbtmp.h" + +#define IDX 7 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_PKSPEC| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_ST1) +#define TAG(x) x##_rgba_spec_st_st +#include "r200_maos_vbtmp.h" + +#define IDX 8 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_ST1| \ + R200_CP_VC_FRMT_N0) +#define TAG(x) x##_st_st_n +#include "r200_maos_vbtmp.h" + +#define IDX 9 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_PKSPEC| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_ST1| \ + R200_CP_VC_FRMT_N0) +#define TAG(x) x##_rgpa_spec_st_st_n +#include "r200_maos_vbtmp.h" + +#define IDX 10 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_Q0) +#define TAG(x) x##_rgba_stq +#include "r200_maos_vbtmp.h" + +#define IDX 11 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_ST1| \ + R200_CP_VC_FRMT_Q1| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_Q0) +#define TAG(x) x##_rgba_stq_stq +#include "r200_maos_vbtmp.h" + +#define IDX 12 +#define IND (R200_CP_VC_FRMT_XY| \ + R200_CP_VC_FRMT_Z| \ + R200_CP_VC_FRMT_W0| \ + R200_CP_VC_FRMT_PKCOLOR| \ + R200_CP_VC_FRMT_PKSPEC| \ + R200_CP_VC_FRMT_ST0| \ + R200_CP_VC_FRMT_Q0| \ + R200_CP_VC_FRMT_ST1| \ + R200_CP_VC_FRMT_Q1| \ + R200_CP_VC_FRMT_N0) +#define TAG(x) x##_w_rgpa_spec_stq_stq_n +#include "r200_maos_vbtmp.h" + + + + + +/*********************************************************************** + * Initialization + ***********************************************************************/ + + +static void init_tcl_verts( void ) +{ + init_rgba(); + init_n(); + init_rgba_n(); + init_rgba_st(); + init_st_n(); + init_rgba_st_st(); + init_rgba_st_n(); + init_rgba_spec_st_st(); + init_st_st_n(); + init_rgpa_spec_st_st_n(); + init_rgba_stq(); + init_rgba_stq_stq(); + init_w_rgpa_spec_stq_stq_n(); +} + + +void r200EmitArrays( GLcontext *ctx, GLuint inputs ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + struct vertex_buffer *VB = &TNL_CONTEXT(ctx)->vb; + GLuint req = 0; + GLuint vtx = (rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] & + ~(R200_TCL_VTX_Q0|R200_TCL_VTX_Q1)); + int i; + static int firsttime = 1; + + if (firsttime) { + init_tcl_verts(); + firsttime = 0; + } + + if (1) { + req |= R200_CP_VC_FRMT_Z; + if (VB->ObjPtr->size == 4) { + req |= R200_CP_VC_FRMT_W0; + } + } + + if (inputs & VERT_NORM) { + req |= R200_CP_VC_FRMT_N0; + } + + if (inputs & VERT_RGBA) { + req |= R200_CP_VC_FRMT_PKCOLOR; + } + + if (inputs & VERT_SPEC_RGB) { + req |= R200_CP_VC_FRMT_PKSPEC; + } + + if (inputs & VERT_TEX0) { + req |= R200_CP_VC_FRMT_ST0; + + if (VB->TexCoordPtr[0]->size == 4) { + req |= R200_CP_VC_FRMT_Q0; + vtx |= R200_TCL_VTX_Q0; + } + } + + if (inputs & VERT_TEX1) { + req |= R200_CP_VC_FRMT_ST1; + + if (VB->TexCoordPtr[1]->size == 4) { + req |= R200_CP_VC_FRMT_Q1; + vtx |= R200_TCL_VTX_Q1; + } + } + + if (vtx != rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT]) { + R200_STATECHANGE( rmesa, tcl ); + rmesa->hw.tcl.cmd[TCL_OUTPUT_VTXFMT] = vtx; + } + + for (i = 0 ; i < R200_TCL_MAX_SETUP ; i++) + if ((setup_tab[i].vertex_format & req) == req) + break; + + if (rmesa->tcl.vertex_format == setup_tab[i].vertex_format && + rmesa->tcl.indexed_verts.buf) + return; + + if (rmesa->tcl.indexed_verts.buf) + r200ReleaseArrays( ctx, ~0 ); + + r200AllocDmaRegionVerts( rmesa, + &rmesa->tcl.indexed_verts, + VB->Count, + setup_tab[i].vertex_size * 4, + 4); + + setup_tab[i].emit( ctx, 0, VB->Count, + rmesa->tcl.indexed_verts.address + + rmesa->tcl.indexed_verts.start ); + + rmesa->tcl.vertex_format = setup_tab[i].vertex_format; + rmesa->tcl.indexed_verts.aos_start = GET_START( &rmesa->tcl.indexed_verts ); + rmesa->tcl.indexed_verts.aos_size = setup_tab[i].vertex_size; + rmesa->tcl.indexed_verts.aos_stride = setup_tab[i].vertex_size; + + rmesa->tcl.aos_components[0] = &rmesa->tcl.indexed_verts; + rmesa->tcl.nr_aos_components = 1; +} + + + +void r200ReleaseArrays( GLcontext *ctx, GLuint newinputs ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + if (R200_DEBUG & DEBUG_VERTS) + _tnl_print_vert_flags( __FUNCTION__, newinputs ); + + if (newinputs) + r200ReleaseDmaRegion( rmesa, &rmesa->tcl.indexed_verts, __FUNCTION__ ); +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_reg.h b/xc/lib/GL/mesa/src/drv/r200/r200_reg.h new file mode 100644 index 000000000..89f38dc33 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_reg.h @@ -0,0 +1,1402 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +#ifndef _R200_REG_H_ +#define _R200_REG_H_ + +#define R200_PP_MISC 0x1c14 +#define R200_REF_ALPHA_MASK 0x000000ff +#define R200_ALPHA_TEST_FAIL (0 << 8) +#define R200_ALPHA_TEST_LESS (1 << 8) +#define R200_ALPHA_TEST_LEQUAL (2 << 8) +#define R200_ALPHA_TEST_EQUAL (3 << 8) +#define R200_ALPHA_TEST_GEQUAL (4 << 8) +#define R200_ALPHA_TEST_GREATER (5 << 8) +#define R200_ALPHA_TEST_NEQUAL (6 << 8) +#define R200_ALPHA_TEST_PASS (7 << 8) +#define R200_ALPHA_TEST_OP_MASK (7 << 8) +#define R200_CHROMA_FUNC_FAIL (0 << 16) +#define R200_CHROMA_FUNC_PASS (1 << 16) +#define R200_CHROMA_FUNC_NEQUAL (2 << 16) +#define R200_CHROMA_FUNC_EQUAL (3 << 16) +#define R200_CHROMA_KEY_NEAREST (0 << 18) +#define R200_CHROMA_KEY_ZERO (1 << 18) +#define R200_RIGHT_HAND_CUBE_D3D (0 << 24) +#define R200_RIGHT_HAND_CUBE_OGL (1 << 24) +#define R200_PP_FOG_COLOR 0x1c18 +#define R200_FOG_COLOR_MASK 0x00ffffff +#define R200_FOG_VERTEX (0 << 24) +#define R200_FOG_TABLE (1 << 24) +#define R200_FOG_USE_DEPTH (0 << 25) +#define R200_FOG_USE_W (1 << 25) +#define R200_FOG_USE_DIFFUSE_ALPHA (2 << 25) +#define R200_FOG_USE_SPEC_ALPHA (3 << 25) +#define R200_FOG_USE_VTX_FOG (4 << 25) +#define R200_RE_SOLID_COLOR 0x1c1c +#define R200_RB3D_BLENDCNTL 0x1c20 +#define R200_COMB_FCN_MASK (7 << 12) +#define R200_COMB_FCN_ADD_CLAMP (0 << 12) +#define R200_COMB_FCN_ADD_NOCLAMP (1 << 12) +#define R200_COMB_FCN_SUB_CLAMP (2 << 12) +#define R200_COMB_FCN_SUB_NOCLAMP (3 << 12) +#define R200_COMB_FCN_MIN (4 << 12) +#define R200_COMB_FCN_MAX (5 << 12) +#define R200_COMB_FCN_RSUB_CLAMP (6 << 12) +#define R200_COMB_FCN_RSUB_NOCLAMP (7 << 12) +#define R200_SRC_BLEND_GL_ZERO (32 << 16) +#define R200_SRC_BLEND_GL_ONE (33 << 16) +#define R200_SRC_BLEND_GL_SRC_COLOR (34 << 16) +#define R200_SRC_BLEND_GL_ONE_MINUS_SRC_COLOR (35 << 16) +#define R200_SRC_BLEND_GL_DST_COLOR (36 << 16) +#define R200_SRC_BLEND_GL_ONE_MINUS_DST_COLOR (37 << 16) +#define R200_SRC_BLEND_GL_SRC_ALPHA (38 << 16) +#define R200_SRC_BLEND_GL_ONE_MINUS_SRC_ALPHA (39 << 16) +#define R200_SRC_BLEND_GL_DST_ALPHA (40 << 16) +#define R200_SRC_BLEND_GL_ONE_MINUS_DST_ALPHA (41 << 16) +#define R200_SRC_BLEND_GL_SRC_ALPHA_SATURATE (42 << 16) +#define R200_SRC_BLEND_GL_CONST_COLOR (43 << 16) +#define R200_SRC_BLEND_GL_ONE_MINUS_CONST_COLOR (44 << 16) +#define R200_SRC_BLEND_GL_CONST_ALPHA (45 << 16) +#define R200_SRC_BLEND_GL_ONE_MINUS_CONST_ALPHA (46 << 16) +#define R200_SRC_BLEND_MASK (63 << 16) +#define R200_DST_BLEND_GL_ZERO (32 << 24) +#define R200_DST_BLEND_GL_ONE (33 << 24) +#define R200_DST_BLEND_GL_SRC_COLOR (34 << 24) +#define R200_DST_BLEND_GL_ONE_MINUS_SRC_COLOR (35 << 24) +#define R200_DST_BLEND_GL_DST_COLOR (36 << 24) +#define R200_DST_BLEND_GL_ONE_MINUS_DST_COLOR (37 << 24) +#define R200_DST_BLEND_GL_SRC_ALPHA (38 << 24) +#define R200_DST_BLEND_GL_ONE_MINUS_SRC_ALPHA (39 << 24) +#define R200_DST_BLEND_GL_DST_ALPHA (40 << 24) +#define R200_DST_BLEND_GL_ONE_MINUS_DST_ALPHA (41 << 24) +#define R200_DST_BLEND_GL_CONST_COLOR (43 << 24) +#define R200_DST_BLEND_GL_ONE_MINUS_CONST_COLOR (44 << 24) +#define R200_DST_BLEND_GL_CONST_ALPHA (45 << 24) +#define R200_DST_BLEND_GL_ONE_MINUS_CONST_ALPHA (46 << 24) +#define R200_DST_BLEND_MASK (63 << 24) +#define R200_RB3D_DEPTHOFFSET 0x1c24 +#define R200_RB3D_DEPTHPITCH 0x1c28 +#define R200_DEPTHPITCH_MASK 0x00001ff8 +#define R200_DEPTH_ENDIAN_NO_SWAP (0 << 18) +#define R200_DEPTH_ENDIAN_WORD_SWAP (1 << 18) +#define R200_DEPTH_ENDIAN_DWORD_SWAP (2 << 18) +#define R200_RB3D_ZSTENCILCNTL 0x1c2c +#define R200_DEPTH_FORMAT_MASK (0xf << 0) +#define R200_DEPTH_FORMAT_16BIT_INT_Z (0 << 0) +#define R200_DEPTH_FORMAT_24BIT_INT_Z (2 << 0) +#define R200_DEPTH_FORMAT_24BIT_FLOAT_Z (3 << 0) +#define R200_DEPTH_FORMAT_32BIT_INT_Z (4 << 0) +#define R200_DEPTH_FORMAT_32BIT_FLOAT_Z (5 << 0) +#define R200_DEPTH_FORMAT_24BIT_FLOAT_W (9 << 0) +#define R200_DEPTH_FORMAT_32BIT_FLOAT_W (11 << 0) +#define R200_Z_TEST_NEVER (0 << 4) +#define R200_Z_TEST_LESS (1 << 4) +#define R200_Z_TEST_LEQUAL (2 << 4) +#define R200_Z_TEST_EQUAL (3 << 4) +#define R200_Z_TEST_GEQUAL (4 << 4) +#define R200_Z_TEST_GREATER (5 << 4) +#define R200_Z_TEST_NEQUAL (6 << 4) +#define R200_Z_TEST_ALWAYS (7 << 4) +#define R200_Z_TEST_MASK (7 << 4) +#define R200_STENCIL_TEST_NEVER (0 << 12) +#define R200_STENCIL_TEST_LESS (1 << 12) +#define R200_STENCIL_TEST_LEQUAL (2 << 12) +#define R200_STENCIL_TEST_EQUAL (3 << 12) +#define R200_STENCIL_TEST_GEQUAL (4 << 12) +#define R200_STENCIL_TEST_GREATER (5 << 12) +#define R200_STENCIL_TEST_NEQUAL (6 << 12) +#define R200_STENCIL_TEST_ALWAYS (7 << 12) +#define R200_STENCIL_TEST_MASK (0x7 << 12) +#define R200_STENCIL_FAIL_KEEP (0 << 16) +#define R200_STENCIL_FAIL_ZERO (1 << 16) +#define R200_STENCIL_FAIL_REPLACE (2 << 16) +#define R200_STENCIL_FAIL_INC (3 << 16) +#define R200_STENCIL_FAIL_DEC (4 << 16) +#define R200_STENCIL_FAIL_INVERT (5 << 16) +#define R200_STENCIL_FAIL_INC_WRAP (6 << 16) +#define R200_STENCIL_FAIL_DEC_WRAP (7 << 16) +#define R200_STENCIL_FAIL_MASK (0x7 << 16) +#define R200_STENCIL_ZPASS_KEEP (0 << 20) +#define R200_STENCIL_ZPASS_ZERO (1 << 20) +#define R200_STENCIL_ZPASS_REPLACE (2 << 20) +#define R200_STENCIL_ZPASS_INC (3 << 20) +#define R200_STENCIL_ZPASS_DEC (4 << 20) +#define R200_STENCIL_ZPASS_INVERT (5 << 20) +#define R200_STENCIL_ZPASS_INC_WRAP (6 << 20) +#define R200_STENCIL_ZPASS_DEC_WRAP (7 << 20) +#define R200_STENCIL_ZPASS_MASK (0x7 << 20) +#define R200_STENCIL_ZFAIL_KEEP (0 << 24) +#define R200_STENCIL_ZFAIL_ZERO (1 << 24) +#define R200_STENCIL_ZFAIL_REPLACE (2 << 24) +#define R200_STENCIL_ZFAIL_INC (3 << 24) +#define R200_STENCIL_ZFAIL_DEC (4 << 24) +#define R200_STENCIL_ZFAIL_INVERT (5 << 24) +#define R200_STENCIL_ZFAIL_INC_WRAP (6 << 24) +#define R200_STENCIL_ZFAIL_DEC_WRAP (7 << 24) +#define R200_STENCIL_ZFAIL_MASK (0x7 << 24) +#define R200_Z_WRITE_ENABLE (1 << 30) +/*gap*/ +#define R200_PP_CNTL 0x1c38 +#define R200_TEX_0_ENABLE 0x00000010 +#define R200_TEX_1_ENABLE 0x00000020 +#define R200_TEX_2_ENABLE 0x00000040 +#define R200_TEX_3_ENABLE 0x00000080 +#define R200_TEX_4_ENABLE 0x00000100 +#define R200_TEX_5_ENABLE 0x00000200 +#define R200_TEX_ENABLE_MASK 0x000003f0 +#define R200_FILTER_ROUND_MODE_MASK 0x00000400 +#define R200_TEX_BLEND_7_ENABLE 0x00000800 +#define R200_TEX_BLEND_0_ENABLE 0x00001000 +#define R200_TEX_BLEND_1_ENABLE 0x00002000 +#define R200_TEX_BLEND_2_ENABLE 0x00004000 +#define R200_TEX_BLEND_3_ENABLE 0x00008000 +#define R200_TEX_BLEND_4_ENABLE 0x00010000 +#define R200_TEX_BLEND_5_ENABLE 0x00020000 +#define R200_TEX_BLEND_6_ENABLE 0x00040000 +#define R200_MULTI_PASS_ENABLE 0x00080000 +#define R200_SPECULAR_ENABLE 0x00200000 +#define R200_FOG_ENABLE 0x00400000 +#define R200_ALPHA_TEST_ENABLE 0x00800000 +#define R200_ANTI_ALIAS_NONE 0x00000000 +#define R200_ANTI_ALIAS_LINE 0x01000000 +#define R200_ANTI_ALIAS_POLY 0x02000000 +#define R200_ANTI_ALIAS_MASK 0x03000000 +#define R200_RB3D_CNTL 0x1c3c +#define R200_ALPHA_BLEND_ENABLE (1 << 0) +#define R200_PLANE_MASK_ENABLE (1 << 1) +#define R200_DITHER_ENABLE (1 << 2) +#define R200_ROUND_ENABLE (1 << 3) +#define R200_SCALE_DITHER_ENABLE (1 << 4) +#define R200_DITHER_INIT (1 << 5) +#define R200_ROP_ENABLE (1 << 6) +#define R200_STENCIL_ENABLE (1 << 7) +#define R200_Z_ENABLE (1 << 8) +#define R200_DEPTH_XZ_OFFEST_ENABLE (1 << 9) +#define R200_COLOR_FORMAT_ARGB1555 (3 << 10) +#define R200_COLOR_FORMAT_RGB565 (4 << 10) +#define R200_COLOR_FORMAT_ARGB8888 (6 << 10) +#define R200_COLOR_FORMAT_RGB332 (7 << 10) +#define R200_COLOR_FORMAT_Y8 (8 << 10) +#define R200_COLOR_FORMAT_RGB8 (9 << 10) +#define R200_COLOR_FORMAT_YUV422_VYUY (11 << 10) +#define R200_COLOR_FORMAT_YUV422_YVYU (12 << 10) +#define R200_COLOR_FORMAT_aYUV444 (14 << 10) +#define R200_COLOR_FORMAT_ARGB4444 (15 << 10) +#define R200_CLRCMP_FLIP_ENABLE (1 << 14) +#define R200_SEPARATE_ALPHA_ENABLE (1 << 16) +#define R200_RB3D_COLOROFFSET 0x1c40 +#define R200_COLOROFFSET_MASK 0xfffffff0 +#define R200_RE_WIDTH_HEIGHT 0x1c44 +#define R200_RE_WIDTH_SHIFT 0 +#define R200_RE_HEIGHT_SHIFT 16 +#define R200_RB3D_COLORPITCH 0x1c48 +#define R200_COLORPITCH_MASK 0x000001ff8 +#define R200_COLOR_ENDIAN_NO_SWAP (0 << 18) +#define R200_COLOR_ENDIAN_WORD_SWAP (1 << 18) +#define R200_COLOR_ENDIAN_DWORD_SWAP (2 << 18) +#define R200_SE_CNTL 0x1c4c +#define R200_FFACE_CULL_CW (0 << 0) +#define R200_FFACE_CULL_CCW (1 << 0) +#define R200_FFACE_CULL_DIR_MASK (1 << 0) +#define R200_BFACE_CULL (0 << 1) +#define R200_BFACE_SOLID (3 << 1) +#define R200_FFACE_CULL (0 << 3) +#define R200_FFACE_SOLID (3 << 3) +#define R200_FFACE_CULL_MASK (3 << 3) +#define R200_FLAT_SHADE_VTX_0 (0 << 6) +#define R200_FLAT_SHADE_VTX_1 (1 << 6) +#define R200_FLAT_SHADE_VTX_2 (2 << 6) +#define R200_FLAT_SHADE_VTX_LAST (3 << 6) +#define R200_DIFFUSE_SHADE_SOLID (0 << 8) +#define R200_DIFFUSE_SHADE_FLAT (1 << 8) +#define R200_DIFFUSE_SHADE_GOURAUD (2 << 8) +#define R200_DIFFUSE_SHADE_MASK (3 << 8) +#define R200_ALPHA_SHADE_SOLID (0 << 10) +#define R200_ALPHA_SHADE_FLAT (1 << 10) +#define R200_ALPHA_SHADE_GOURAUD (2 << 10) +#define R200_ALPHA_SHADE_MASK (3 << 10) +#define R200_SPECULAR_SHADE_SOLID (0 << 12) +#define R200_SPECULAR_SHADE_FLAT (1 << 12) +#define R200_SPECULAR_SHADE_GOURAUD (2 << 12) +#define R200_SPECULAR_SHADE_MASK (3 << 12) +#define R200_FOG_SHADE_SOLID (0 << 14) +#define R200_FOG_SHADE_FLAT (1 << 14) +#define R200_FOG_SHADE_GOURAUD (2 << 14) +#define R200_FOG_SHADE_MASK (3 << 14) +#define R200_ZBIAS_ENABLE_POINT (1 << 16) +#define R200_ZBIAS_ENABLE_LINE (1 << 17) +#define R200_ZBIAS_ENABLE_TRI (1 << 18) +#define R200_WIDELINE_ENABLE (1 << 20) +#define R200_VTX_PIX_CENTER_D3D (0 << 27) +#define R200_VTX_PIX_CENTER_OGL (1 << 27) +#define R200_ROUND_MODE_TRUNC (0 << 28) +#define R200_ROUND_MODE_ROUND (1 << 28) +#define R200_ROUND_MODE_ROUND_EVEN (2 << 28) +#define R200_ROUND_MODE_ROUND_ODD (3 << 28) +#define R200_ROUND_PREC_16TH_PIX (0 << 30) +#define R200_ROUND_PREC_8TH_PIX (1 << 30) +#define R200_ROUND_PREC_4TH_PIX (2 << 30) +#define R200_ROUND_PREC_HALF_PIX (3 << 30) +#define R200_RE_CNTL 0x1c50 +#define R200_STIPPLE_ENABLE 0x1 +#define R200_SCISSOR_ENABLE 0x2 +#define R200_PATTERN_ENABLE 0x4 +#define R200_PERSPECTIVE_ENABLE 0x8 +#define R200_POINT_SMOOTH 0x20 +#define R200_VTX_STQ0_D3D 0x00010000 +#define R200_VTX_STQ1_D3D 0x00040000 +#define R200_VTX_STQ2_D3D 0x00100000 +#define R200_VTX_STQ3_D3D 0x00400000 +#define R200_VTX_STQ4_D3D 0x01000000 +#define R200_VTX_STQ5_D3D 0x04000000 +/* gap */ +#define R200_RE_STIPPLE_ADDR 0x1cc8 +#define R200_RE_STIPPLE_DATA 0x1ccc +#define R200_RE_LINE_PATTERN 0x1cd0 +#define R200_LINE_PATTERN_MASK 0x0000ffff +#define R200_LINE_REPEAT_COUNT_SHIFT 16 +#define R200_LINE_PATTERN_START_SHIFT 24 +#define R200_LINE_PATTERN_LITTLE_BIT_ORDER (0 << 28) +#define R200_LINE_PATTERN_BIG_BIT_ORDER (1 << 28) +#define R200_LINE_PATTERN_AUTO_RESET (1 << 29) +#define R200_RE_LINE_STATE 0x1cd4 +#define R200_LINE_CURRENT_PTR_SHIFT 0 +#define R200_LINE_CURRENT_COUNT_SHIFT 8 +#define R200_RE_SCISSOR_TL_0 0x1cd8 +#define R200_RE_SCISSOR_BR_0 0x1cdc +#define R200_RE_SCISSOR_TL_1 0x1ce0 +#define R200_RE_SCISSOR_BR_1 0x1ce4 +#define R200_RE_SCISSOR_TL_2 0x1ce8 +#define R200_RE_SCISSOR_BR_2 0x1cec +/* gap */ +#define R200_RB3D_DEPTHXY_OFFSET 0x1d60 +#define R200_DEPTHX_SHIFT 0 +#define R200_DEPTHY_SHIFT 16 +/* gap */ +#define R200_RB3D_STENCILREFMASK 0x1d7c +#define R200_STENCIL_REF_SHIFT 0 +#define R200_STENCIL_REF_MASK (0xff << 0) +#define R200_STENCIL_MASK_SHIFT 16 +#define R200_STENCIL_VALUE_MASK (0xff << 16) +#define R200_STENCIL_WRITEMASK_SHIFT 24 +#define R200_STENCIL_WRITE_MASK (0xff << 24) +#define R200_RB3D_ROPCNTL 0x1d80 +#define R200_ROP_MASK (15 << 8) +#define R200_ROP_CLEAR (0 << 8) +#define R200_ROP_NOR (1 << 8) +#define R200_ROP_AND_INVERTED (2 << 8) +#define R200_ROP_COPY_INVERTED (3 << 8) +#define R200_ROP_AND_REVERSE (4 << 8) +#define R200_ROP_INVERT (5 << 8) +#define R200_ROP_XOR (6 << 8) +#define R200_ROP_NAND (7 << 8) +#define R200_ROP_AND (8 << 8) +#define R200_ROP_EQUIV (9 << 8) +#define R200_ROP_NOOP (10 << 8) +#define R200_ROP_OR_INVERTED (11 << 8) +#define R200_ROP_COPY (12 << 8) +#define R200_ROP_OR_REVERSE (13 << 8) +#define R200_ROP_OR (14 << 8) +#define R200_ROP_SET (15 << 8) +#define R200_RB3D_PLANEMASK 0x1d84 +/* gap */ +#define R200_SE_VPORT_XSCALE 0x1d98 +#define R200_SE_VPORT_XOFFSET 0x1d9c +#define R200_SE_VPORT_YSCALE 0x1da0 +#define R200_SE_VPORT_YOFFSET 0x1da4 +#define R200_SE_VPORT_ZSCALE 0x1da8 +#define R200_SE_VPORT_ZOFFSET 0x1dac +#define R200_SE_ZBIAS_FACTOR 0x1db0 +#define R200_SE_ZBIAS_CONSTANT 0x1db4 +#define R200_SE_LINE_WIDTH 0x1db8 +#define R200_LINE_WIDTH_SHIFT 0x00000000 +#define R200_MINPOINTSIZE_SHIFT 0x00000010 +/* gap */ +#define R200_SE_VAP_CNTL 0x2080 +#define R200_VAP_TCL_ENABLE 0x00000001 +#define R200_VAP_SINGLE_BUF_STATE_ENABLE 0x00000010 +#define R200_VAP_FORCE_W_TO_ONE 0x00010000 +#define R200_VAP_D3D_TEX_DEFAULT 0x00020000 +#define R200_VAP_VF_MAX_VTX_NUM__SHIFT 18 +#define R200_VAP_DX_CLIP_SPACE_DEF 0x00400000 +#define R200_SE_VF_CNTL 0x2084 +#define R200_VF_PRIM_NONE 0x00000000 +#define R200_VF_PRIM_POINTS 0x00000001 +#define R200_VF_PRIM_LINES 0x00000002 +#define R200_VF_PRIM_LINE_STRIP 0x00000003 +#define R200_VF_PRIM_TRIANGLES 0x00000004 +#define R200_VF_PRIM_TRIANGLE_FAN 0x00000005 +#define R200_VF_PRIM_TRIANGLE_STRIP 0x00000006 +#define R200_VF_PRIM_RECT_LIST 0x00000008 +#define R200_VF_PRIM_3VRT_POINTS 0x00000009 +#define R200_VF_PRIM_3VRT_LINES 0x0000000a +#define R200_VF_PRIM_POINT_SPRITES 0x0000000b +#define R200_VF_PRIM_LINE_LOOP 0x0000000c +#define R200_VF_PRIM_QUADS 0x0000000d +#define R200_VF_PRIM_QUAD_STRIP 0x0000000e +#define R200_VF_PRIM_POLYGON 0x0000000f +#define R200_VF_PRIM_MASK 0x0000000f +#define R200_VF_PRIM_WALK_IND 0x00000010 +#define R200_VF_PRIM_WALK_LIST 0x00000020 +#define R200_VF_PRIM_WALK_RING 0x00000030 +#define R200_VF_PRIM_WALK_MASK 0x00000030 +#define R200_VF_COLOR_ORDER_RGBA 0x00000040 +#define R200_VF_TCL_OUTPUT_VTX_ENABLE 0x00000200 +#define R200_VF_INDEX_SZ_4 0x00000800 +#define R200_VF_VERTEX_NUMBER_MASK 0xffff0000 +#define R200_VF_VERTEX_NUMBER_SHIFT 16 +#define R200_SE_VTX_FMT_0 0x2088 +#define R200_VTX_XY 0 /* always have xy */ +#define R200_VTX_Z0 (1<<0) +#define R200_VTX_W0 (1<<1) +#define R200_VTX_WEIGHT_COUNT_SHIFT (2) +#define R200_VTX_PV_MATRIX_SEL (1<<5) +#define R200_VTX_N0 (1<<6) +#define R200_VTX_POINT_SIZE (1<<7) +#define R200_VTX_DISCRETE_FOG (1<<8) +#define R200_VTX_SHININESS_0 (1<<9) +#define R200_VTX_SHININESS_1 (1<<10) +#define R200_VTX_COLOR_NOT_PRESENT 0 +#define R200_VTX_PK_RGBA 1 +#define R200_VTX_FP_RGB 2 +#define R200_VTX_FP_RGBA 3 +#define R200_VTX_COLOR_MASK 3 +#define R200_VTX_COLOR_0_SHIFT 11 +#define R200_VTX_COLOR_1_SHIFT 13 +#define R200_VTX_COLOR_2_SHIFT 15 +#define R200_VTX_COLOR_3_SHIFT 17 +#define R200_VTX_COLOR_4_SHIFT 19 +#define R200_VTX_COLOR_5_SHIFT 21 +#define R200_VTX_COLOR_6_SHIFT 23 +#define R200_VTX_COLOR_7_SHIFT 25 +#define R200_VTX_XY1 (1<<28) +#define R200_VTX_Z1 (1<<29) +#define R200_VTX_W1 (1<<30) +#define R200_VTX_N1 (1<<31) +#define R200_SE_VTX_FMT_1 0x208c +#define R200_VTX_TEX0_COMP_CNT_SHIFT 0 +#define R200_VTX_TEX1_COMP_CNT_SHIFT 3 +#define R200_VTX_TEX2_COMP_CNT_SHIFT 6 +#define R200_VTX_TEX3_COMP_CNT_SHIFT 9 +#define R200_VTX_TEX4_COMP_CNT_SHIFT 12 +#define R200_VTX_TEX5_COMP_CNT_SHIFT 15 +#define R200_SE_TCL_OUTPUT_VTX_FMT_0 0x2090 +#define R200_SE_TCL_OUTPUT_VTX_FMT_1 0x2094 +/* gap */ +#define R200_SE_VTE_CNTL 0x20b0 +#define R200_VPORT_X_SCALE_ENA 0x00000001 +#define R200_VPORT_X_OFFSET_ENA 0x00000002 +#define R200_VPORT_Y_SCALE_ENA 0x00000004 +#define R200_VPORT_Y_OFFSET_ENA 0x00000008 +#define R200_VPORT_Z_SCALE_ENA 0x00000010 +#define R200_VPORT_Z_OFFSET_ENA 0x00000020 +#define R200_VTX_XY_FMT 0x00000100 +#define R200_VTX_Z_FMT 0x00000200 +#define R200_VTX_W0_FMT 0x00000400 +#define R200_VTX_W0_NORMALIZE 0x00000800 +#define R200_VTX_ST_DENORMALIZED 0x00001000 +/* gap */ +#define R200_SE_VTX_NUM_ARRAYS 0x20c0 +#define R200_SE_VTX_AOS_ATTR01 0x20c4 +#define R200_SE_VTX_AOS_ADDR0 0x20c8 +#define R200_SE_VTX_AOS_ADDR1 0x20cc +#define R200_SE_VTX_AOS_ATTR23 0x20d0 +#define R200_SE_VTX_AOS_ADDR2 0x20d4 +#define R200_SE_VTX_AOS_ADDR3 0x20d8 +#define R200_SE_VTX_AOS_ATTR45 0x20dc +#define R200_SE_VTX_AOS_ADDR4 0x20e0 +#define R200_SE_VTX_AOS_ADDR5 0x20e4 +#define R200_SE_VTX_AOS_ATTR67 0x20e8 +#define R200_SE_VTX_AOS_ADDR6 0x20ec +#define R200_SE_VTX_AOS_ADDR7 0x20f0 +#define R200_SE_VTX_AOS_ATTR89 0x20f4 +#define R200_SE_VTX_AOS_ADDR8 0x20f8 +#define R200_SE_VTX_AOS_ADDR9 0x20fc +#define R200_SE_VTX_AOS_ATTR1011 0x2100 +#define R200_SE_VTX_AOS_ADDR10 0x2104 +#define R200_SE_VTX_AOS_ADDR11 0x2108 +#define R200_SE_VF_MAX_VTX_INDX 0x210c +#define R200_SE_VF_MIN_VTX_INDX 0x2110 +/* gap */ +#define R200_SE_VAP_CNTL_STATUS 0x2140 +/* gap */ +#define R200_SE_VTX_STATE_CNTL 0x2180 +#define R200_VSC_COLOR_0_ASSEMBLY_CNTL_SHIFT 0x00000000 +#define R200_VSC_COLOR_1_ASSEMBLY_CNTL_SHIFT 0x00000002 +#define R200_VSC_COLOR_2_ASSEMBLY_CNTL_SHIFT 0x00000004 +#define R200_VSC_COLOR_3_ASSEMBLY_CNTL_SHIFT 0x00000006 +#define R200_VSC_COLOR_4_ASSEMBLY_CNTL_SHIFT 0x00000008 +#define R200_VSC_COLOR_5_ASSEMBLY_CNTL_SHIFT 0x0000000a +#define R200_VSC_COLOR_6_ASSEMBLY_CNTL_SHIFT 0x0000000c +#define R200_VSC_COLOR_7_ASSEMBLY_CNTL_SHIFT 0x0000000e +#define R200_VSC_UPDATE_USER_COLOR_0_ENABLE 0x00010000 +#define R200_VSC_UPDATE_USER_COLOR_1_ENABLE 0x00020000 +/* gap */ +#define R200_SE_TCL_VECTOR_INDX_REG 0x2200 +#define R200_SE_TCL_VECTOR_DATA_REG 0x2204 +#define R200_SE_TCL_SCALAR_INDX_REG 0x2208 +#define R200_SE_TCL_SCALAR_DATA_REG 0x220c +/* gap */ +#define R200_SE_TCL_MATRIX_SEL_0 0x2230 +#define R200_MODELVIEW_0_SHIFT (0) +#define R200_MODELVIEW_1_SHIFT (8) +#define R200_MODELVIEW_2_SHIFT (16) +#define R200_MODELVIEW_3_SHIFT (24) +#define R200_SE_TCL_MATRIX_SEL_1 0x2234 +#define R200_IT_MODELVIEW_0_SHIFT (0) +#define R200_IT_MODELVIEW_1_SHIFT (8) +#define R200_IT_MODELVIEW_2_SHIFT (16) +#define R200_IT_MODELVIEW_3_SHIFT (24) +#define R200_SE_TCL_MATRIX_SEL_2 0x2238 +#define R200_MODELPROJECT_0_SHIFT (0) +#define R200_MODELPROJECT_1_SHIFT (8) +#define R200_MODELPROJECT_2_SHIFT (16) +#define R200_MODELPROJECT_3_SHIFT (24) +#define R200_SE_TCL_MATRIX_SEL_3 0x223c +#define R200_TEXMAT_0_SHIFT 0 +#define R200_TEXMAT_1_SHIFT 8 +#define R200_TEXMAT_2_SHIFT 16 +#define R200_TEXMAT_3_SHIFT 24 +#define R200_SE_TCL_MATRIX_SEL_4 0x2240 +#define R200_TEXMAT_4_SHIFT 0 +#define R200_TEXMAT_5_SHIFT 8 +/* gap */ +#define R200_SE_TCL_OUTPUT_VTX_COMP_SEL 0x2250 +#define R200_OUTPUT_XYZW (1<<0) +#define R200_OUTPUT_COLOR_0 (1<<8) +#define R200_OUTPUT_COLOR_1 (1<<9) +#define R200_OUTPUT_TEX_0 (1<<16) +#define R200_OUTPUT_TEX_1 (1<<17) +#define R200_OUTPUT_TEX_2 (1<<18) +#define R200_OUTPUT_TEX_3 (1<<19) +#define R200_OUTPUT_TEX_4 (1<<20) +#define R200_OUTPUT_TEX_5 (1<<21) +#define R200_OUTPUT_TEX_MASK (0x3f<<16) +#define R200_OUTPUT_PT_SIZE (1<<25) +#define R200_FORCE_INORDER_PROC (1<<31) +#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0 0x2254 +#define R200_VERTEX_POSITION_ADDR__SHIFT 0x00000000 +#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_1 0x2258 +#define R200_VTX_COLOR_0_ADDR__SHIFT 0x00000000 +#define R200_VTX_COLOR_1_ADDR__SHIFT 0x00000008 +#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_2 0x225c +#define R200_VTX_TEX_0_ADDR__SHIFT 0x00000000 +#define R200_VTX_TEX_1_ADDR__SHIFT 0x00000008 +#define R200_VTX_TEX_2_ADDR__SHIFT 0x00000010 +#define R200_VTX_TEX_3_ADDR__SHIFT 0x00000018 +#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_3 0x2260 +#define R200_VTX_TEX_4_ADDR__SHIFT 0x00000000 +#define R200_VTX_TEX_5_ADDR__SHIFT 0x00000008 + +/* gap */ +#define R200_SE_TCL_LIGHT_MODEL_CTL_0 0x2268 +#define R200_LIGHTING_ENABLE (1<<0) +#define R200_LIGHT_IN_MODELSPACE (1<<1) +#define R200_LOCAL_VIEWER (1<<2) +#define R200_NORMALIZE_NORMALS (1<<3) +#define R200_RESCALE_NORMALS (1<<4) +#define R200_SPECULAR_LIGHTS (1<<5) +#define R200_DIFFUSE_SPECULAR_COMBINE (1<<6) +#define R200_LIGHT_ALPHA (1<<7) +#define R200_LOCAL_LIGHT_VEC_GL (1<<8) +#define R200_LIGHT_NO_NORMAL_AMBIENT_ONLY (1<<9) +#define R200_LIGHT_TWOSIDE (1<<10) +#define R200_FRONT_SHININESS_SOURCE_SHIFT (0xb) +#define R200_BACK_SHININESS_SOURCE_SHIFT (0xd) +#define R200_LM0_SOURCE_MATERIAL_0 (0) +#define R200_LM0_SOURCE_MATERIAL_1 (1) +#define R200_LM0_SOURCE_VERTEX_SHININESS_0 (2) +#define R200_LM0_SOURCE_VERTEX_SHININESS_1 (3) +#define R200_SE_TCL_LIGHT_MODEL_CTL_1 0x226c +#define R200_LM1_SOURCE_LIGHT_PREMULT (0) +#define R200_LM1_SOURCE_MATERIAL_0 (1) +#define R200_LM1_SOURCE_VERTEX_COLOR_0 (2) +#define R200_LM1_SOURCE_VERTEX_COLOR_1 (3) +#define R200_LM1_SOURCE_VERTEX_COLOR_2 (4) +#define R200_LM1_SOURCE_VERTEX_COLOR_3 (5) +#define R200_LM1_SOURCE_VERTEX_COLOR_4 (6) +#define R200_LM1_SOURCE_VERTEX_COLOR_5 (7) +#define R200_LM1_SOURCE_VERTEX_COLOR_6 (8) +#define R200_LM1_SOURCE_VERTEX_COLOR_7 (9) +#define R200_LM1_SOURCE_MATERIAL_1 (0xf) +#define R200_FRONT_EMISSIVE_SOURCE_SHIFT (0) +#define R200_FRONT_AMBIENT_SOURCE_SHIFT (4) +#define R200_FRONT_DIFFUSE_SOURCE_SHIFT (8) +#define R200_FRONT_SPECULAR_SOURCE_SHIFT (12) +#define R200_BACK_EMISSIVE_SOURCE_SHIFT (16) +#define R200_BACK_AMBIENT_SOURCE_SHIFT (20) +#define R200_BACK_DIFFUSE_SOURCE_SHIFT (24) +#define R200_BACK_SPECULAR_SOURCE_SHIFT (28) +#define R200_SE_TCL_PER_LIGHT_CTL_0 0x2270 +#define R200_LIGHT_0_ENABLE (1<<0) +#define R200_LIGHT_0_ENABLE_AMBIENT (1<<1) +#define R200_LIGHT_0_ENABLE_SPECULAR (1<<2) +#define R200_LIGHT_0_IS_LOCAL (1<<3) +#define R200_LIGHT_0_IS_SPOT (1<<4) +#define R200_LIGHT_0_DUAL_CONE (1<<5) +#define R200_LIGHT_0_ENABLE_RANGE_ATTEN (1<<6) +#define R200_LIGHT_0_CONSTANT_RANGE_ATTEN (1<<7) +#define R200_LIGHT_1_ENABLE (1<<16) +#define R200_LIGHT_1_ENABLE_AMBIENT (1<<17) +#define R200_LIGHT_1_ENABLE_SPECULAR (1<<18) +#define R200_LIGHT_1_IS_LOCAL (1<<19) +#define R200_LIGHT_1_IS_SPOT (1<<20) +#define R200_LIGHT_1_DUAL_CONE (1<<21) +#define R200_LIGHT_1_ENABLE_RANGE_ATTEN (1<<22) +#define R200_LIGHT_1_CONSTANT_RANGE_ATTEN (1<<23) +#define R200_LIGHT_0_SHIFT (0) +#define R200_LIGHT_1_SHIFT (16) +#define R200_SE_TCL_PER_LIGHT_CTL_1 0x2274 +#define R200_LIGHT_2_SHIFT (0) +#define R200_LIGHT_3_SHIFT (16) +#define R200_SE_TCL_PER_LIGHT_CTL_2 0x2278 +#define R200_LIGHT_4_SHIFT (0) +#define R200_LIGHT_5_SHIFT (16) +#define R200_SE_TCL_PER_LIGHT_CTL_3 0x227c +#define R200_LIGHT_6_SHIFT (0) +#define R200_LIGHT_7_SHIFT (16) +/* gap */ +#define R200_SE_TCL_TEX_PROC_CTL_2 0x22a8 +#define R200_TEXGEN_0_COMP_MASK_SHIFT (0) +#define R200_TEXGEN_1_COMP_MASK_SHIFT (4) +#define R200_TEXGEN_2_COMP_MASK_SHIFT (8) +#define R200_TEXGEN_3_COMP_MASK_SHIFT (12) +#define R200_TEXGEN_4_COMP_MASK_SHIFT (16) +#define R200_TEXGEN_5_COMP_MASK_SHIFT (20) +#define R200_SE_TCL_TEX_PROC_CTL_3 0x22ac +#define R200_TEXGEN_0_INPUT_TEX_SHIFT (0) +#define R200_TEXGEN_1_INPUT_TEX_SHIFT (4) +#define R200_TEXGEN_2_INPUT_TEX_SHIFT (8) +#define R200_TEXGEN_3_INPUT_TEX_SHIFT (12) +#define R200_TEXGEN_4_INPUT_TEX_SHIFT (16) +#define R200_TEXGEN_5_INPUT_TEX_SHIFT (20) +#define R200_SE_TCL_TEX_PROC_CTL_0 0x22b0 +#define R200_TEXGEN_TEXMAT_0_ENABLE (1<<0) +#define R200_TEXGEN_TEXMAT_1_ENABLE (1<<1) +#define R200_TEXGEN_TEXMAT_2_ENABLE (1<<2) +#define R200_TEXGEN_TEXMAT_3_ENABLE (1<<3) +#define R200_TEXGEN_TEXMAT_4_ENABLE (1<<4) +#define R200_TEXGEN_TEXMAT_5_ENABLE (1<<5) +#define R200_TEXMAT_0_ENABLE (1<<8) +#define R200_TEXMAT_1_ENABLE (1<<9) +#define R200_TEXMAT_2_ENABLE (1<<10) +#define R200_TEXMAT_3_ENABLE (1<<11) +#define R200_TEXMAT_4_ENABLE (1<<12) +#define R200_TEXMAT_5_ENABLE (1<<13) +#define R200_TEXGEN_FORCE_W_TO_ONE (1<<16) +#define R200_SE_TCL_TEX_PROC_CTL_1 0x22b4 +#define R200_TEXGEN_INPUT_MASK (0xf) +#define R200_TEXGEN_INPUT_TEXCOORD_0 (0) +#define R200_TEXGEN_INPUT_TEXCOORD_1 (1) +#define R200_TEXGEN_INPUT_TEXCOORD_2 (2) +#define R200_TEXGEN_INPUT_TEXCOORD_3 (3) +#define R200_TEXGEN_INPUT_TEXCOORD_4 (4) +#define R200_TEXGEN_INPUT_TEXCOORD_5 (5) +#define R200_TEXGEN_INPUT_OBJ (8) +#define R200_TEXGEN_INPUT_EYE (9) +#define R200_TEXGEN_INPUT_EYE_NORMAL (0xa) +#define R200_TEXGEN_INPUT_EYE_REFLECT (0xb) +#define R200_TEXGEN_INPUT_SPHERE (0xd) +#define R200_TEXGEN_0_INPUT_SHIFT (0) +#define R200_TEXGEN_1_INPUT_SHIFT (4) +#define R200_TEXGEN_2_INPUT_SHIFT (8) +#define R200_TEXGEN_3_INPUT_SHIFT (12) +#define R200_TEXGEN_4_INPUT_SHIFT (16) +#define R200_TEXGEN_5_INPUT_SHIFT (20) +#define R200_SE_TC_TEX_CYL_WRAP_CTL 0x22b8 +/* gap */ +#define R200_SE_TCL_UCP_VERT_BLEND_CTL 0x22c0 +#define R200_UCP_IN_CLIP_SPACE (1<<0) +#define R200_UCP_IN_MODEL_SPACE (1<<1) +#define R200_UCP_ENABLE_0 (1<<2) +#define R200_UCP_ENABLE_1 (1<<3) +#define R200_UCP_ENABLE_2 (1<<4) +#define R200_UCP_ENABLE_3 (1<<5) +#define R200_UCP_ENABLE_4 (1<<6) +#define R200_UCP_ENABLE_5 (1<<7) +#define R200_TCL_FOG_MASK (3<<8) +#define R200_TCL_FOG_DISABLE (0<<8) +#define R200_TCL_FOG_EXP (1<<8) +#define R200_TCL_FOG_EXP2 (2<<8) +#define R200_TCL_FOG_LINEAR (3<<8) +#define R200_RNG_BASED_FOG (1<<10) +#define R200_CLIP_DISABLE (1<<11) +#define R200_CULL_FRONT_IS_CW (0<<28) +#define R200_CULL_FRONT_IS_CCW (1<<28) +#define R200_CULL_FRONT (1<<29) +#define R200_CULL_BACK (1<<30) +#define R200_SE_TCL_POINT_SPRITE_CNTL 0x22c4 +/* gap */ +#define R200_SE_VTX_ST_POS_0_X_4 0x2300 +#define R200_SE_VTX_ST_POS_0_Y_4 0x2304 +#define R200_SE_VTX_ST_POS_0_Z_4 0x2308 +#define R200_SE_VTX_ST_POS_0_W_4 0x230c +#define R200_SE_VTX_ST_NORM_0_X 0x2310 +#define R200_SE_VTX_ST_NORM_0_Y 0x2314 +#define R200_SE_VTX_ST_NORM_0_Z 0x2318 +#define R200_SE_VTX_ST_PVMS 0x231c +#define R200_SE_VTX_ST_CLR_0_R 0x2320 +#define R200_SE_VTX_ST_CLR_0_G 0x2324 +#define R200_SE_VTX_ST_CLR_0_B 0x2328 +#define R200_SE_VTX_ST_CLR_0_A 0x232c +#define R200_SE_VTX_ST_CLR_1_R 0x2330 +#define R200_SE_VTX_ST_CLR_1_G 0x2334 +#define R200_SE_VTX_ST_CLR_1_B 0x2338 +#define R200_SE_VTX_ST_CLR_1_A 0x233c +#define R200_SE_VTX_ST_CLR_2_R 0x2340 +#define R200_SE_VTX_ST_CLR_2_G 0x2344 +#define R200_SE_VTX_ST_CLR_2_B 0x2348 +#define R200_SE_VTX_ST_CLR_2_A 0x234c +#define R200_SE_VTX_ST_CLR_3_R 0x2350 +#define R200_SE_VTX_ST_CLR_3_G 0x2354 +#define R200_SE_VTX_ST_CLR_3_B 0x2358 +#define R200_SE_VTX_ST_CLR_3_A 0x235c +#define R200_SE_VTX_ST_CLR_4_R 0x2360 +#define R200_SE_VTX_ST_CLR_4_G 0x2364 +#define R200_SE_VTX_ST_CLR_4_B 0x2368 +#define R200_SE_VTX_ST_CLR_4_A 0x236c +#define R200_SE_VTX_ST_CLR_5_R 0x2370 +#define R200_SE_VTX_ST_CLR_5_G 0x2374 +#define R200_SE_VTX_ST_CLR_5_B 0x2378 +#define R200_SE_VTX_ST_CLR_5_A 0x237c +#define R200_SE_VTX_ST_CLR_6_R 0x2380 +#define R200_SE_VTX_ST_CLR_6_G 0x2384 +#define R200_SE_VTX_ST_CLR_6_B 0x2388 +#define R200_SE_VTX_ST_CLR_6_A 0x238c +#define R200_SE_VTX_ST_CLR_7_R 0x2390 +#define R200_SE_VTX_ST_CLR_7_G 0x2394 +#define R200_SE_VTX_ST_CLR_7_B 0x2398 +#define R200_SE_VTX_ST_CLR_7_A 0x239c +#define R200_SE_VTX_ST_TEX_0_S 0x23a0 +#define R200_SE_VTX_ST_TEX_0_T 0x23a4 +#define R200_SE_VTX_ST_TEX_0_R 0x23a8 +#define R200_SE_VTX_ST_TEX_0_Q 0x23ac +#define R200_SE_VTX_ST_TEX_1_S 0x23b0 +#define R200_SE_VTX_ST_TEX_1_T 0x23b4 +#define R200_SE_VTX_ST_TEX_1_R 0x23b8 +#define R200_SE_VTX_ST_TEX_1_Q 0x23bc +#define R200_SE_VTX_ST_TEX_2_S 0x23c0 +#define R200_SE_VTX_ST_TEX_2_T 0x23c4 +#define R200_SE_VTX_ST_TEX_2_R 0x23c8 +#define R200_SE_VTX_ST_TEX_2_Q 0x23cc +#define R200_SE_VTX_ST_TEX_3_S 0x23d0 +#define R200_SE_VTX_ST_TEX_3_T 0x23d4 +#define R200_SE_VTX_ST_TEX_3_R 0x23d8 +#define R200_SE_VTX_ST_TEX_3_Q 0x23dc +#define R200_SE_VTX_ST_TEX_4_S 0x23e0 +#define R200_SE_VTX_ST_TEX_4_T 0x23e4 +#define R200_SE_VTX_ST_TEX_4_R 0x23e8 +#define R200_SE_VTX_ST_TEX_4_Q 0x23ec +#define R200_SE_VTX_ST_TEX_5_S 0x23f0 +#define R200_SE_VTX_ST_TEX_5_T 0x23f4 +#define R200_SE_VTX_ST_TEX_5_R 0x23f8 +#define R200_SE_VTX_ST_TEX_5_Q 0x23fc +#define R200_SE_VTX_ST_PNT_SPRT_SZ 0x2400 +#define R200_SE_VTX_ST_DISC_FOG 0x2404 +#define R200_SE_VTX_ST_SHININESS_0 0x2408 +#define R200_SE_VTX_ST_SHININESS_1 0x240c +#define R200_SE_VTX_ST_BLND_WT_0 0x2410 +#define R200_SE_VTX_ST_BLND_WT_1 0x2414 +#define R200_SE_VTX_ST_BLND_WT_2 0x2418 +#define R200_SE_VTX_ST_BLND_WT_3 0x241c +#define R200_SE_VTX_ST_POS_1_X 0x2420 +#define R200_SE_VTX_ST_POS_1_Y 0x2424 +#define R200_SE_VTX_ST_POS_1_Z 0x2428 +#define R200_SE_VTX_ST_POS_1_W 0x242c +#define R200_SE_VTX_ST_NORM_1_X 0x2430 +#define R200_SE_VTX_ST_NORM_1_Y 0x2434 +#define R200_SE_VTX_ST_NORM_1_Z 0x2438 +#define R200_SE_VTX_ST_USR_CLR_0_R 0x2440 +#define R200_SE_VTX_ST_USR_CLR_0_G 0x2444 +#define R200_SE_VTX_ST_USR_CLR_0_B 0x2448 +#define R200_SE_VTX_ST_USR_CLR_0_A 0x244c +#define R200_SE_VTX_ST_USR_CLR_1_R 0x2450 +#define R200_SE_VTX_ST_USR_CLR_1_G 0x2454 +#define R200_SE_VTX_ST_USR_CLR_1_B 0x2458 +#define R200_SE_VTX_ST_USR_CLR_1_A 0x245c +#define R200_SE_VTX_ST_CLR_0_PKD 0x2460 +#define R200_SE_VTX_ST_CLR_1_PKD 0x2464 +#define R200_SE_VTX_ST_CLR_2_PKD 0x2468 +#define R200_SE_VTX_ST_CLR_3_PKD 0x246c +#define R200_SE_VTX_ST_CLR_4_PKD 0x2470 +#define R200_SE_VTX_ST_CLR_5_PKD 0x2474 +#define R200_SE_VTX_ST_CLR_6_PKD 0x2478 +#define R200_SE_VTX_ST_CLR_7_PKD 0x247c +#define R200_SE_VTX_ST_POS_0_X_2 0x2480 +#define R200_SE_VTX_ST_POS_0_Y_2 0x2484 +#define R200_SE_VTX_ST_PAR_CLR_LD 0x2488 +#define R200_SE_VTX_ST_USR_CLR_PKD 0x248c +#define R200_SE_VTX_ST_POS_0_X_3 0x2490 +#define R200_SE_VTX_ST_POS_0_Y_3 0x2494 +#define R200_SE_VTX_ST_POS_0_Z_3 0x2498 +#define R200_SE_VTX_ST_END_OF_PKT 0x249c +/* gap */ +#define R200_RE_POINTSIZE 0x2648 +#define R200_POINTSIZE_SHIFT 0 +#define R200_MAXPOINTSIZE_SHIFT 16 +/* gap */ +#define R200_RE_TOP_LEFT 0x26c0 +#define R200_RE_LEFT_SHIFT 0 +#define R200_RE_TOP_SHIFT 16 +#define R200_RE_MISC 0x26c4 +#define R200_STIPPLE_COORD_MASK 0x1f +#define R200_STIPPLE_X_OFFSET_SHIFT 0 +#define R200_STIPPLE_X_OFFSET_MASK (0x1f << 0) +#define R200_STIPPLE_Y_OFFSET_SHIFT 8 +#define R200_STIPPLE_Y_OFFSET_MASK (0x1f << 8) +#define R200_STIPPLE_LITTLE_BIT_ORDER (0 << 16) +#define R200_STIPPLE_BIG_BIT_ORDER (1 << 16) +/* gap */ +#define R200_RE_AUX_SCISSOR_CNTL 0x26f0 +#define R200_EXCLUSIVE_SCISSOR_0 0x01000000 +#define R200_EXCLUSIVE_SCISSOR_1 0x02000000 +#define R200_EXCLUSIVE_SCISSOR_2 0x04000000 +#define R200_SCISSOR_ENABLE_0 0x10000000 +#define R200_SCISSOR_ENABLE_1 0x20000000 +#define R200_SCISSOR_ENABLE_2 0x40000000 +/* gap */ +#define R200_PP_TXFILTER_0 0x2c00 +#define R200_MAG_FILTER_NEAREST (0 << 0) +#define R200_MAG_FILTER_LINEAR (1 << 0) +#define R200_MAG_FILTER_MASK (1 << 0) +#define R200_MIN_FILTER_NEAREST (0 << 1) +#define R200_MIN_FILTER_LINEAR (1 << 1) +#define R200_MIN_FILTER_NEAREST_MIP_NEAREST (2 << 1) +#define R200_MIN_FILTER_NEAREST_MIP_LINEAR (3 << 1) +#define R200_MIN_FILTER_LINEAR_MIP_NEAREST (6 << 1) +#define R200_MIN_FILTER_LINEAR_MIP_LINEAR (7 << 1) +#define R200_MIN_FILTER_ANISO_NEAREST (8 << 1) +#define R200_MIN_FILTER_ANISO_LINEAR (9 << 1) +#define R200_MIN_FILTER_ANISO_NEAREST_MIP_NEAREST (10 << 1) +#define R200_MIN_FILTER_ANISO_NEAREST_MIP_LINEAR (11 << 1) +#define R200_MIN_FILTER_MASK (15 << 1) +#define R200_MAX_ANISO_1_TO_1 (0 << 5) +#define R200_MAX_ANISO_2_TO_1 (1 << 5) +#define R200_MAX_ANISO_4_TO_1 (2 << 5) +#define R200_MAX_ANISO_8_TO_1 (3 << 5) +#define R200_MAX_ANISO_16_TO_1 (4 << 5) +#define R200_MAX_ANISO_MASK (7 << 5) +#define R200_MAX_MIP_LEVEL_MASK (0x0f << 16) +#define R200_MAX_MIP_LEVEL_SHIFT 16 +#define R200_YUV_TO_RGB (1 << 20) +#define R200_YUV_TEMPERATURE_COOL (0 << 21) +#define R200_YUV_TEMPERATURE_HOT (1 << 21) +#define R200_YUV_TEMPERATURE_MASK (1 << 21) +#define R200_WRAPEN_S (1 << 22) +#define R200_CLAMP_S_WRAP (0 << 23) +#define R200_CLAMP_S_MIRROR (1 << 23) +#define R200_CLAMP_S_CLAMP_LAST (2 << 23) +#define R200_CLAMP_S_MIRROR_CLAMP_LAST (3 << 23) +#define R200_CLAMP_S_CLAMP_BORDER (4 << 23) +#define R200_CLAMP_S_MIRROR_CLAMP_BORDER (5 << 23) +#define R200_CLAMP_S_MASK (7 << 23) +#define R200_WRAPEN_T (1 << 26) +#define R200_CLAMP_T_WRAP (0 << 27) +#define R200_CLAMP_T_MIRROR (1 << 27) +#define R200_CLAMP_T_CLAMP_LAST (2 << 27) +#define R200_CLAMP_T_MIRROR_CLAMP_LAST (3 << 27) +#define R200_CLAMP_T_CLAMP_BORDER (4 << 27) +#define R200_CLAMP_T_MIRROR_CLAMP_BORDER (5 << 27) +#define R200_CLAMP_T_MASK (7 << 27) +#define R200_KILL_LT_ZERO (1 << 30) +#define R200_BORDER_MODE_OGL (0 << 31) +#define R200_BORDER_MODE_D3D (1 << 31) +#define R200_PP_TXFORMAT_0 0x2c04 +#define R200_TXFORMAT_I8 (0 << 0) +#define R200_TXFORMAT_AI88 (1 << 0) +#define R200_TXFORMAT_RGB332 (2 << 0) +#define R200_TXFORMAT_ARGB1555 (3 << 0) +#define R200_TXFORMAT_RGB565 (4 << 0) +#define R200_TXFORMAT_ARGB4444 (5 << 0) +#define R200_TXFORMAT_ARGB8888 (6 << 0) +#define R200_TXFORMAT_RGBA8888 (7 << 0) +#define R200_TXFORMAT_Y8 (8 << 0) +#define R200_TXFORMAT_AVYU4444 (9 << 0) +#define R200_TXFORMAT_VYUY422 (10 << 0) +#define R200_TXFORMAT_YVYU422 (11 << 0) +#define R200_TXFORMAT_DXT1 (12 << 0) +#define R200_TXFORMAT_DXT23 (14 << 0) +#define R200_TXFORMAT_DXT45 (15 << 0) +#define R200_TXFORMAT_FORMAT_MASK (31 << 0) +#define R200_TXFORMAT_FORMAT_SHIFT 0 +#define R200_TXFORMAT_ALPHA_IN_MAP (1 << 6) +#define R200_TXFORMAT_NON_POWER2 (1 << 7) +#define R200_TXFORMAT_WIDTH_MASK (15 << 8) +#define R200_TXFORMAT_WIDTH_SHIFT 8 +#define R200_TXFORMAT_HEIGHT_MASK (15 << 12) +#define R200_TXFORMAT_HEIGHT_SHIFT 12 +#define R200_TXFORMAT_F5_WIDTH_MASK (15 << 15) /* cube face 5 */ +#define R200_TXFORMAT_F5_WIDTH_SHIFT 15 +#define R200_TXFORMAT_F5_HEIGHT_MASK (15 << 20) +#define R200_TXFORMAT_F5_HEIGHT_SHIFT 20 +#define R200_TXFORMAT_ST_ROUTE_MASK (7 << 24) +#define R200_TXFORMAT_ST_ROUTE_STQ0 (0 << 24) +#define R200_TXFORMAT_ST_ROUTE_STQ1 (1 << 24) +#define R200_TXFORMAT_ST_ROUTE_STQ2 (2 << 24) +#define R200_TXFORMAT_ST_ROUTE_STQ3 (3 << 24) +#define R200_TXFORMAT_ST_ROUTE_STQ4 (4 << 24) +#define R200_TXFORMAT_ST_ROUTE_STQ5 (5 << 24) +#define R200_TXFORMAT_ALPHA_MASK_ENABLE (1 << 28) +#define R200_TXFORMAT_CHROMA_KEY_ENABLE (1 << 29) +#define R200_TXFORMAT_CUBIC_MAP_ENABLE (1 << 30) +#define R200_PP_TXFORMAT_X_0 0x2c08 +#define R200_DEPTH_LOG2_MASK (0xff << 0) +#define R200_DEPTH_LOG2_SHIFT 0 +#define R200_WRAPEN_Q (1 << 8) +#define R200_CLAMP_Q_WRAP (0 << 9) +#define R200_CLAMP_Q_MIRROR (1 << 9) +#define R200_CLAMP_Q_CLAMP_LAST (2 << 9) +#define R200_CLAMP_Q_MIRROR_CLAMP_LAST (3 << 9) +#define R200_CLAMP_Q_CLAMP_BORDER (6 << 9) +#define R200_CLAMP_Q_MIRROR_CLAMP_BORDER (7 << 9) +#define R200_CLAMP_Q_MASK (7 << 9) +#define R200_MIN_MIP_LEVEL_MASK (0xff << 12) +#define R200_MIN_MIP_LEVEL_SHIFT 12 +#define R200_TEXCOORD_NONPROJ (0 << 16) +#define R200_TEXCOORD_CUBIC_ENV (1 << 16) +#define R200_TEXCOORD_VOLUME (2 << 16) +#define R200_TEXCOORD_PROJ (3 << 16) +#define R200_TEXCOORD_DEPTH (4 << 16) +#define R200_TEXCOORD_1D_PROJ (5 << 16) +#define R200_TEXCOORD_1D (6 << 16) +#define R200_TEXCOORD_ZERO (7 << 16) +#define R200_LOD_BIAS_MASK (0xfff80000) +#define R200_LOD_BIAS_SHIFT 19 +#define R200_PP_TXSIZE_0 0x2c0c /* NPOT only */ +#define R200_PP_TXPITCH_0 0x2c10 /* NPOT only */ +#define R200_PP_BORDER_COLOR_0 0x2c14 +#define R200_PP_CUBIC_FACES_0 0x2c18 +#define R200_PP_TXFILTER_1 0x2c20 +#define R200_PP_TXFORMAT_1 0x2c24 +#define R200_PP_TXFORMAT_X_1 0x2c28 +#define R200_PP_TXSIZE_1 0x2c2c +#define R200_PP_TXPITCH_1 0x2c30 +#define R200_PP_BORDER_COLOR_1 0x2c34 +#define R200_PP_CUBIC_FACES_1 0x2c38 +#define R200_PP_TXFILTER_2 0x2c40 +#define R200_PP_TXFORMAT_2 0x2c44 +#define R200_PP_TXSIZE_2 0x2c4c +#define R200_PP_TXFORMAT_X_2 0x2c48 +#define R200_PP_TXPITCH_2 0x2c50 +#define R200_PP_BORDER_COLOR_2 0x2c54 +#define R200_PP_CUBIC_FACES_2 0x2c58 +#define R200_PP_TXFILTER_3 0x2c60 +#define R200_PP_TXFORMAT_3 0x2c64 +#define R200_PP_TXSIZE_3 0x2c6c +#define R200_PP_TXFORMAT_X_3 0x2c68 +#define R200_PP_TXPITCH_3 0x2c70 +#define R200_PP_BORDER_COLOR_3 0x2c74 +#define R200_PP_CUBIC_FACES_3 0x2c78 +#define R200_PP_TXFILTER_4 0x2c80 +#define R200_PP_TXFORMAT_4 0x2c84 +#define R200_PP_TXSIZE_4 0x2c8c +#define R200_PP_TXFORMAT_X_4 0x2c88 +#define R200_PP_TXPITCH_4 0x2c90 +#define R200_PP_BORDER_COLOR_4 0x2c94 +#define R200_PP_CUBIC_FACES_4 0x2c98 +#define R200_PP_TXFILTER_5 0x2ca0 +#define R200_PP_TXFORMAT_5 0x2ca4 +#define R200_PP_TXSIZE_5 0x2cac +#define R200_PP_TXFORMAT_X_5 0x2ca8 +#define R200_PP_TXPITCH_5 0x2cb0 +#define R200_PP_BORDER_COLOR_5 0x2cb4 +#define R200_PP_CUBIC_FACES_5 0x2cb8 +/* gap */ +#define R200_PP_CNTL_X 0x2cc4 +/* gap */ +#define R200_PP_TXOFFSET_0 0x2d00 +#define R200_TXO_ENDIAN_NO_SWAP (0 << 0) +#define R200_TXO_ENDIAN_BYTE_SWAP (1 << 0) +#define R200_TXO_ENDIAN_WORD_SWAP (2 << 0) +#define R200_TXO_ENDIAN_HALFDW_SWAP (3 << 0) +#define R200_TXO_OFFSET_MASK 0xffffffe0 +#define R200_TXO_OFFSET_SHIFT 5 +#define R200_PP_CUBIC_OFFSET_F1_0 0x2d04 +#define R200_PP_CUBIC_OFFSET_F2_0 0x2d08 +#define R200_PP_CUBIC_OFFSET_F3_0 0x2d0c +#define R200_PP_CUBIC_OFFSET_F4_0 0x2d10 +#define R200_PP_CUBIC_OFFSET_F5_0 0x2d14 +#define R200_PP_TXOFFSET_1 0x2d18 +#define R200_PP_CUBIC_OFFSET_F1_1 0x2d1c +#define R200_PP_CUBIC_OFFSET_F2_1 0x2d20 +#define R200_PP_CUBIC_OFFSET_F3_1 0x2d24 +#define R200_PP_CUBIC_OFFSET_F4_1 0x2d28 +#define R200_PP_CUBIC_OFFSET_F5_1 0x2d2c +#define R200_PP_TXOFFSET_2 0x2d30 +#define R200_PP_CUBIC_OFFSET_F1_2 0x2d34 +#define R200_PP_CUBIC_OFFSET_F2_2 0x2d38 +#define R200_PP_CUBIC_OFFSET_F3_2 0x2d3c +#define R200_PP_CUBIC_OFFSET_F4_2 0x2d40 +#define R200_PP_CUBIC_OFFSET_F5_2 0x2d44 +#define R200_PP_TXOFFSET_3 0x2d48 +#define R200_PP_CUBIC_OFFSET_F1_3 0x2d4c +#define R200_PP_CUBIC_OFFSET_F2_3 0x2d50 +#define R200_PP_CUBIC_OFFSET_F3_3 0x2d54 +#define R200_PP_CUBIC_OFFSET_F4_3 0x2d58 +#define R200_PP_CUBIC_OFFSET_F5_3 0x2d5c +#define R200_PP_TXOFFSET_4 0x2d60 +#define R200_PP_CUBIC_OFFSET_F1_4 0x2d64 +#define R200_PP_CUBIC_OFFSET_F2_4 0x2d68 +#define R200_PP_CUBIC_OFFSET_F3_4 0x2d6c +#define R200_PP_CUBIC_OFFSET_F4_4 0x2d70 +#define R200_PP_CUBIC_OFFSET_F5_4 0x2d74 +#define R200_PP_TXOFFSET_5 0x2d78 +#define R200_PP_CUBIC_OFFSET_F1_5 0x2d7c +#define R200_PP_CUBIC_OFFSET_F2_5 0x2d80 +#define R200_PP_CUBIC_OFFSET_F3_5 0x2d84 +#define R200_PP_CUBIC_OFFSET_F4_5 0x2d88 +#define R200_PP_CUBIC_OFFSET_F5_5 0x2d8c +/* gap */ +#define R200_PP_TAM_DEBUG3 0x2d9c +/* gap */ +#define R200_PP_TFACTOR_0 0x2ee0 +#define R200_PP_TFACTOR_1 0x2ee4 +#define R200_PP_TFACTOR_2 0x2ee8 +#define R200_PP_TFACTOR_3 0x2eec +#define R200_PP_TFACTOR_4 0x2ef0 +#define R200_PP_TFACTOR_5 0x2ef4 +/* gap */ +#define R200_PP_TXCBLEND_0 0x2f00 +#define R200_TXC_ARG_A_ZERO (0) +#define R200_TXC_ARG_A_CURRENT_COLOR (2) +#define R200_TXC_ARG_A_CURRENT_ALPHA (3) +#define R200_TXC_ARG_A_DIFFUSE_COLOR (4) +#define R200_TXC_ARG_A_DIFFUSE_ALPHA (5) +#define R200_TXC_ARG_A_SPECULAR_COLOR (6) +#define R200_TXC_ARG_A_SPECULAR_ALPHA (7) +#define R200_TXC_ARG_A_TFACTOR_COLOR (8) +#define R200_TXC_ARG_A_TFACTOR_ALPHA (9) +#define R200_TXC_ARG_A_R0_COLOR (10) +#define R200_TXC_ARG_A_R0_ALPHA (11) +#define R200_TXC_ARG_A_R1_COLOR (12) +#define R200_TXC_ARG_A_R1_ALPHA (13) +#define R200_TXC_ARG_A_R2_COLOR (14) +#define R200_TXC_ARG_A_R2_ALPHA (15) +#define R200_TXC_ARG_A_R3_COLOR (16) +#define R200_TXC_ARG_A_R3_ALPHA (17) +#define R200_TXC_ARG_A_R4_COLOR (18) +#define R200_TXC_ARG_A_R4_ALPHA (19) +#define R200_TXC_ARG_A_R5_COLOR (20) +#define R200_TXC_ARG_A_R5_ALPHA (21) +#define R200_TXC_ARG_A_TFACTOR1_COLOR (26) +#define R200_TXC_ARG_A_TFACTOR1_ALPHA (27) +#define R200_TXC_ARG_A_MASK (31 << 0) +#define R200_TXC_ARG_A_SHIFT 0 +#define R200_TXC_ARG_B_ZERO (0<<5) +#define R200_TXC_ARG_B_CURRENT_COLOR (2<<5) +#define R200_TXC_ARG_B_CURRENT_ALPHA (3<<5) +#define R200_TXC_ARG_B_DIFFUSE_COLOR (4<<5) +#define R200_TXC_ARG_B_DIFFUSE_ALPHA (5<<5) +#define R200_TXC_ARG_B_SPECULAR_COLOR (6<<5) +#define R200_TXC_ARG_B_SPECULAR_ALPHA (7<<5) +#define R200_TXC_ARG_B_TFACTOR_COLOR (8<<5) +#define R200_TXC_ARG_B_TFACTOR_ALPHA (9<<5) +#define R200_TXC_ARG_B_R0_COLOR (10<<5) +#define R200_TXC_ARG_B_R0_ALPHA (11<<5) +#define R200_TXC_ARG_B_R1_COLOR (12<<5) +#define R200_TXC_ARG_B_R1_ALPHA (13<<5) +#define R200_TXC_ARG_B_R2_COLOR (14<<5) +#define R200_TXC_ARG_B_R2_ALPHA (15<<5) +#define R200_TXC_ARG_B_R3_COLOR (16<<5) +#define R200_TXC_ARG_B_R3_ALPHA (17<<5) +#define R200_TXC_ARG_B_R4_COLOR (18<<5) +#define R200_TXC_ARG_B_R4_ALPHA (19<<5) +#define R200_TXC_ARG_B_R5_COLOR (20<<5) +#define R200_TXC_ARG_B_R5_ALPHA (21<<5) +#define R200_TXC_ARG_B_TFACTOR1_COLOR (26<<5) +#define R200_TXC_ARG_B_TFACTOR1_ALPHA (27<<5) +#define R200_TXC_ARG_B_MASK (31 << 5) +#define R200_TXC_ARG_B_SHIFT 5 +#define R200_TXC_ARG_C_ZERO (0<<10) +#define R200_TXC_ARG_C_CURRENT_COLOR (2<<10) +#define R200_TXC_ARG_C_CURRENT_ALPHA (3<<10) +#define R200_TXC_ARG_C_DIFFUSE_COLOR (4<<10) +#define R200_TXC_ARG_C_DIFFUSE_ALPHA (5<<10) +#define R200_TXC_ARG_C_SPECULAR_COLOR (6<<10) +#define R200_TXC_ARG_C_SPECULAR_ALPHA (7<<10) +#define R200_TXC_ARG_C_TFACTOR_COLOR (8<<10) +#define R200_TXC_ARG_C_TFACTOR_ALPHA (9<<10) +#define R200_TXC_ARG_C_R0_COLOR (10<<10) +#define R200_TXC_ARG_C_R0_ALPHA (11<<10) +#define R200_TXC_ARG_C_R1_COLOR (12<<10) +#define R200_TXC_ARG_C_R1_ALPHA (13<<10) +#define R200_TXC_ARG_C_R2_COLOR (14<<10) +#define R200_TXC_ARG_C_R2_ALPHA (15<<10) +#define R200_TXC_ARG_C_R3_COLOR (16<<10) +#define R200_TXC_ARG_C_R3_ALPHA (17<<10) +#define R200_TXC_ARG_C_R4_COLOR (18<<10) +#define R200_TXC_ARG_C_R4_ALPHA (19<<10) +#define R200_TXC_ARG_C_R5_COLOR (20<<10) +#define R200_TXC_ARG_C_R5_ALPHA (21<<10) +#define R200_TXC_ARG_C_TFACTOR1_COLOR (26<<10) +#define R200_TXC_ARG_C_TFACTOR1_ALPHA (27<<10) +#define R200_TXC_ARG_C_MASK (31 << 10) +#define R200_TXC_ARG_C_SHIFT 10 +#define R200_TXC_COMP_ARG_A (1 << 16) +#define R200_TXC_COMP_ARG_A_SHIFT (16) +#define R200_TXC_BIAS_ARG_A (1 << 17) +#define R200_TXC_SCALE_ARG_A (1 << 18) +#define R200_TXC_NEG_ARG_A (1 << 19) +#define R200_TXC_COMP_ARG_B (1 << 20) +#define R200_TXC_COMP_ARG_B_SHIFT (20) +#define R200_TXC_BIAS_ARG_B (1 << 21) +#define R200_TXC_SCALE_ARG_B (1 << 22) +#define R200_TXC_NEG_ARG_B (1 << 23) +#define R200_TXC_COMP_ARG_C (1 << 24) +#define R200_TXC_COMP_ARG_C_SHIFT (24) +#define R200_TXC_BIAS_ARG_C (1 << 25) +#define R200_TXC_SCALE_ARG_C (1 << 26) +#define R200_TXC_NEG_ARG_C (1 << 27) +#define R200_TXC_OP_MADD (0 << 28) +#define R200_TXC_OP_CND0 (2 << 28) +#define R200_TXC_OP_LERP (3 << 28) +#define R200_TXC_OP_DOT3 (4 << 28) +#define R200_TXC_OP_DOT4 (5 << 28) +#define R200_TXC_OP_CONDITIONAL (6 << 28) +#define R200_TXC_OP_DOT2_ADD (7 << 28) +#define R200_TXC_OP_MASK (7 << 28) +#define R200_PP_TXCBLEND2_0 0x2f04 +#define R200_TXC_TFACTOR_SEL_SHIFT 0 +#define R200_TXC_TFACTOR_SEL_MASK 0x7 +#define R200_TXC_TFACTOR1_SEL_SHIFT 4 +#define R200_TXC_TFACTOR1_SEL_MASK (0x7 << 4) +#define R200_TXC_SCALE_SHIFT 8 +#define R200_TXC_SCALE_MASK (7 << 8) +#define R200_TXC_SCALE_1X (0 << 8) +#define R200_TXC_SCALE_2X (1 << 8) +#define R200_TXC_SCALE_4X (2 << 8) +#define R200_TXC_SCALE_8X (3 << 8) +#define R200_TXC_SCALE_INV2 (5 << 8) +#define R200_TXC_SCALE_INV4 (6 << 8) +#define R200_TXC_SCALE_INV8 (7 << 8) +#define R200_TXC_CLAMP_SHIFT 12 +#define R200_TXC_CLAMP_MASK (3 << 12) +#define R200_TXC_CLAMP_WRAP (0 << 12) +#define R200_TXC_CLAMP_0_1 (1 << 12) +#define R200_TXC_CLAMP_8_8 (2 << 12) +#define R200_TXC_OUTPUT_REG_MASK (7 << 16) +#define R200_TXC_OUTPUT_REG_NONE (0 << 16) +#define R200_TXC_OUTPUT_REG_R0 (1 << 16) +#define R200_TXC_OUTPUT_REG_R1 (2 << 16) +#define R200_TXC_OUTPUT_REG_R2 (3 << 16) +#define R200_TXC_OUTPUT_REG_R3 (4 << 16) +#define R200_TXC_OUTPUT_REG_R4 (5 << 16) +#define R200_TXC_OUTPUT_REG_R5 (6 << 16) +#define R200_TXC_OUTPUT_MASK_MASK (7 << 20) +#define R200_TXC_OUTPUT_MASK_RGB (0 << 20) +#define R200_TXC_OUTPUT_MASK_RG (1 << 20) +#define R200_TXC_OUTPUT_MASK_RB (2 << 20) +#define R200_TXC_OUTPUT_MASK_R (3 << 20) +#define R200_TXC_OUTPUT_MASK_GB (4 << 20) +#define R200_TXC_OUTPUT_MASK_G (5 << 20) +#define R200_TXC_OUTPUT_MASK_B (6 << 20) +#define R200_TXC_OUTPUT_MASK_NONE (7 << 20) +#define R200_TXC_REPL_NORMAL 0 +#define R200_TXC_REPL_RED 1 +#define R200_TXC_REPL_GREEN 2 +#define R200_TXC_REPL_BLUE 3 +#define R200_TXC_REPL_ARG_A_SHIFT 26 +#define R200_TXC_REPL_ARG_A_MASK (3 << 26) +#define R200_TXC_REPL_ARG_B_SHIFT 28 +#define R200_TXC_REPL_ARG_B_MASK (3 << 28) +#define R200_TXC_REPL_ARG_C_SHIFT 30 +#define R200_TXC_REPL_ARG_C_MASK (3 << 30) +#define R200_PP_TXABLEND_0 0x2f08 +#define R200_TXA_ARG_A_ZERO (0) +#define R200_TXA_ARG_A_CURRENT_ALPHA (2) /* guess */ +#define R200_TXA_ARG_A_CURRENT_BLUE (3) /* guess */ +#define R200_TXA_ARG_A_DIFFUSE_ALPHA (4) +#define R200_TXA_ARG_A_DIFFUSE_BLUE (5) +#define R200_TXA_ARG_A_SPECULAR_ALPHA (6) +#define R200_TXA_ARG_A_SPECULAR_BLUE (7) +#define R200_TXA_ARG_A_TFACTOR_ALPHA (8) +#define R200_TXA_ARG_A_TFACTOR_BLUE (9) +#define R200_TXA_ARG_A_R0_ALPHA (10) +#define R200_TXA_ARG_A_R0_BLUE (11) +#define R200_TXA_ARG_A_R1_ALPHA (12) +#define R200_TXA_ARG_A_R1_BLUE (13) +#define R200_TXA_ARG_A_R2_ALPHA (14) +#define R200_TXA_ARG_A_R2_BLUE (15) +#define R200_TXA_ARG_A_R3_ALPHA (16) +#define R200_TXA_ARG_A_R3_BLUE (17) +#define R200_TXA_ARG_A_R4_ALPHA (18) +#define R200_TXA_ARG_A_R4_BLUE (19) +#define R200_TXA_ARG_A_R5_ALPHA (20) +#define R200_TXA_ARG_A_R5_BLUE (21) +#define R200_TXA_ARG_A_TFACTOR1_ALPHA (26) +#define R200_TXA_ARG_A_TFACTOR1_BLUE (27) +#define R200_TXA_ARG_A_MASK (31 << 0) +#define R200_TXA_ARG_A_SHIFT 0 +#define R200_TXA_ARG_B_ZERO (0<<5) +#define R200_TXA_ARG_B_CURRENT_ALPHA (2<<5) /* guess */ +#define R200_TXA_ARG_B_CURRENT_BLUE (3<<5) /* guess */ +#define R200_TXA_ARG_B_DIFFUSE_ALPHA (4<<5) +#define R200_TXA_ARG_B_DIFFUSE_BLUE (5<<5) +#define R200_TXA_ARG_B_SPECULAR_ALPHA (6<<5) +#define R200_TXA_ARG_B_SPECULAR_BLUE (7<<5) +#define R200_TXA_ARG_B_TFACTOR_ALPHA (8<<5) +#define R200_TXA_ARG_B_TFACTOR_BLUE (9<<5) +#define R200_TXA_ARG_B_R0_ALPHA (10<<5) +#define R200_TXA_ARG_B_R0_BLUE (11<<5) +#define R200_TXA_ARG_B_R1_ALPHA (12<<5) +#define R200_TXA_ARG_B_R1_BLUE (13<<5) +#define R200_TXA_ARG_B_R2_ALPHA (14<<5) +#define R200_TXA_ARG_B_R2_BLUE (15<<5) +#define R200_TXA_ARG_B_R3_ALPHA (16<<5) +#define R200_TXA_ARG_B_R3_BLUE (17<<5) +#define R200_TXA_ARG_B_R4_ALPHA (18<<5) +#define R200_TXA_ARG_B_R4_BLUE (19<<5) +#define R200_TXA_ARG_B_R5_ALPHA (20<<5) +#define R200_TXA_ARG_B_R5_BLUE (21<<5) +#define R200_TXA_ARG_B_TFACTOR1_ALPHA (26<<5) +#define R200_TXA_ARG_B_TFACTOR1_BLUE (27<<5) +#define R200_TXA_ARG_B_MASK (31 << 5) +#define R200_TXA_ARG_B_SHIFT 5 +#define R200_TXA_ARG_C_ZERO (0<<10) +#define R200_TXA_ARG_C_CURRENT_ALPHA (2<<10) /* guess */ +#define R200_TXA_ARG_C_CURRENT_BLUE (3<<10) /* guess */ +#define R200_TXA_ARG_C_DIFFUSE_ALPHA (4<<10) +#define R200_TXA_ARG_C_DIFFUSE_BLUE (5<<10) +#define R200_TXA_ARG_C_SPECULAR_ALPHA (6<<10) +#define R200_TXA_ARG_C_SPECULAR_BLUE (7<<10) +#define R200_TXA_ARG_C_TFACTOR_ALPHA (8<<10) +#define R200_TXA_ARG_C_TFACTOR_BLUE (9<<10) +#define R200_TXA_ARG_C_R0_ALPHA (10<<10) +#define R200_TXA_ARG_C_R0_BLUE (11<<10) +#define R200_TXA_ARG_C_R1_ALPHA (12<<10) +#define R200_TXA_ARG_C_R1_BLUE (13<<10) +#define R200_TXA_ARG_C_R2_ALPHA (14<<10) +#define R200_TXA_ARG_C_R2_BLUE (15<<10) +#define R200_TXA_ARG_C_R3_ALPHA (16<<10) +#define R200_TXA_ARG_C_R3_BLUE (17<<10) +#define R200_TXA_ARG_C_R4_ALPHA (18<<10) +#define R200_TXA_ARG_C_R4_BLUE (19<<10) +#define R200_TXA_ARG_C_R5_ALPHA (20<<10) +#define R200_TXA_ARG_C_R5_BLUE (21<<10) +#define R200_TXA_ARG_C_TFACTOR1_ALPHA (26<<10) +#define R200_TXA_ARG_C_TFACTOR1_BLUE (27<<10) +#define R200_TXA_ARG_C_MASK (31 << 10) +#define R200_TXA_ARG_C_SHIFT 10 +#define R200_TXA_COMP_ARG_A (1 << 16) +#define R200_TXA_COMP_ARG_A_SHIFT (16) +#define R200_TXA_BIAS_ARG_A (1 << 17) +#define R200_TXA_SCALE_ARG_A (1 << 18) +#define R200_TXA_NEG_ARG_A (1 << 19) +#define R200_TXA_COMP_ARG_B (1 << 20) +#define R200_TXA_COMP_ARG_B_SHIFT (20) +#define R200_TXA_BIAS_ARG_B (1 << 21) +#define R200_TXA_SCALE_ARG_B (1 << 22) +#define R200_TXA_NEG_ARG_B (1 << 23) +#define R200_TXA_COMP_ARG_C (1 << 24) +#define R200_TXA_COMP_ARG_C_SHIFT (24) +#define R200_TXA_BIAS_ARG_C (1 << 25) +#define R200_TXA_SCALE_ARG_C (1 << 26) +#define R200_TXA_NEG_ARG_C (1 << 27) +#define R200_TXA_OP_MADD (0 << 28) +#define R200_TXA_OP_CND0 (2 << 28) +#define R200_TXA_OP_LERP (3 << 28) +#define R200_TXA_OP_CONDITIONAL (6 << 28) +#define R200_TXA_OP_MASK (7 << 28) +#define R200_PP_TXABLEND2_0 0x2f0c +#define R200_TXA_TFACTOR_SEL_SHIFT 0 +#define R200_TXA_TFACTOR_SEL_MASK 0x7 +#define R200_TXA_TFACTOR1_SEL_SHIFT 4 +#define R200_TXA_TFACTOR1_SEL_MASK (0x7 << 4) +#define R200_TXA_SCALE_SHIFT 8 +#define R200_TXA_SCALE_MASK (7 << 8) +#define R200_TXA_SCALE_1X (0 << 8) +#define R200_TXA_SCALE_2X (1 << 8) +#define R200_TXA_SCALE_4X (2 << 8) +#define R200_TXA_SCALE_8X (3 << 8) +#define R200_TXA_SCALE_INV2 (5 << 8) +#define R200_TXA_SCALE_INV4 (6 << 8) +#define R200_TXA_SCALE_INV8 (7 << 8) +#define R200_TXA_CLAMP_SHIFT 12 +#define R200_TXA_CLAMP_MASK (3 << 12) +#define R200_TXA_CLAMP_WRAP (0 << 12) +#define R200_TXA_CLAMP_0_1 (1 << 12) +#define R200_TXA_CLAMP_8_8 (2 << 12) +#define R200_TXA_OUTPUT_REG_MASK (7 << 16) +#define R200_TXA_OUTPUT_REG_NONE (0 << 16) +#define R200_TXA_OUTPUT_REG_R0 (1 << 16) +#define R200_TXA_OUTPUT_REG_R1 (2 << 16) +#define R200_TXA_OUTPUT_REG_R2 (3 << 16) +#define R200_TXA_OUTPUT_REG_R3 (4 << 16) +#define R200_TXA_OUTPUT_REG_R4 (5 << 16) +#define R200_TXA_OUTPUT_REG_R5 (6 << 16) +#define R200_TXA_DOT_ALPHA (1 << 20) +#define R200_TXA_REPL_NORMAL 0 +#define R200_TXA_REPL_RED 1 +#define R200_TXA_REPL_GREEN 2 +#define R200_TXA_REPL_ARG_A_SHIFT 26 +#define R200_TXA_REPL_ARG_A_MASK (3 << 26) +#define R200_TXA_REPL_ARG_B_SHIFT 28 +#define R200_TXA_REPL_ARG_B_MASK (3 << 28) +#define R200_TXA_REPL_ARG_C_SHIFT 30 +#define R200_TXA_REPL_ARG_C_MASK (3 << 30) +#define R200_PP_TXCBLEND_1 0x2f10 +#define R200_PP_TXCBLEND2_1 0x2f14 +#define R200_PP_TXABLEND_1 0x2f18 +#define R200_PP_TXABLEND2_1 0x2f1c +#define R200_PP_TXCBLEND_2 0x2f20 +#define R200_PP_TXCBLEND2_2 0x2f24 +#define R200_PP_TXABLEND_2 0x2f28 +#define R200_PP_TXABLEND2_2 0x2f2c +#define R200_PP_TXCBLEND_3 0x2f30 +#define R200_PP_TXCBLEND2_3 0x2f34 +#define R200_PP_TXABLEND_3 0x2f38 +#define R200_PP_TXABLEND2_3 0x2f3c +#define R200_PP_TXCBLEND_4 0x2f40 +#define R200_PP_TXCBLEND2_4 0x2f44 +#define R200_PP_TXABLEND_4 0x2f48 +#define R200_PP_TXABLEND2_4 0x2f4c +#define R200_PP_TXCBLEND_5 0x2f50 +#define R200_PP_TXCBLEND2_5 0x2f54 +#define R200_PP_TXABLEND_5 0x2f58 +#define R200_PP_TXABLEND2_5 0x2f5c +#define R200_PP_TXCBLEND_6 0x2f60 +#define R200_PP_TXCBLEND2_6 0x2f64 +#define R200_PP_TXABLEND_6 0x2f68 +#define R200_PP_TXABLEND2_6 0x2f6c +#define R200_PP_TXCBLEND_7 0x2f70 +#define R200_PP_TXCBLEND2_7 0x2f74 +#define R200_PP_TXABLEND_7 0x2f78 +#define R200_PP_TXABLEND2_7 0x2f7c +/* gap */ +#define R200_RB3D_ABLENDCNTL 0x321C /* see BLENDCTL */ +#define R200_RB3D_CBLENDCNTL 0x3220 /* see BLENDCTL */ + + +/* + * Offsets in TCL vector state. NOTE: Hardwiring matrix positions. + * Multiple contexts could collaberate to eliminate state bouncing. + */ +#define R200_VS_LIGHT_AMBIENT_ADDR 0x00000028 +#define R200_VS_LIGHT_DIFFUSE_ADDR 0x00000030 +#define R200_VS_LIGHT_SPECULAR_ADDR 0x00000038 +#define R200_VS_LIGHT_DIRPOS_ADDR 0x00000040 +#define R200_VS_LIGHT_HWVSPOT_ADDR 0x00000048 +#define R200_VS_LIGHT_ATTENUATION_ADDR 0x00000050 +#define R200_VS_SPOT_DUAL_CONE 0x00000058 +#define R200_VS_GLOBAL_AMBIENT_ADDR 0x0000005C +#define R200_VS_FOG_PARAM_ADDR 0x0000005D +#define R200_VS_EYE_VECTOR_ADDR 0x0000005E +#define R200_VS_UCP_ADDR 0x00000060 +#define R200_VS_PNT_SPRITE_VPORT_SCALE 0x00000068 +#define R200_VS_MATRIX_0_MV 0x00000080 +#define R200_VS_MATRIX_1_INV_MV 0x00000084 +#define R200_VS_MATRIX_2_MVP 0x00000088 +#define R200_VS_MATRIX_3_TEX0 0x0000008C +#define R200_VS_MATRIX_4_TEX1 0x00000090 +#define R200_VS_MATRIX_5_TEX2 0x00000094 +#define R200_VS_MATRIX_6_TEX3 0x00000098 +#define R200_VS_MATRIX_7_TEX4 0x0000009C +#define R200_VS_MATRIX_8_TEX5 0x000000A0 +#define R200_VS_MAT_0_EMISS 0x000000B0 +#define R200_VS_MAT_0_AMB 0x000000B1 +#define R200_VS_MAT_0_DIF 0x000000B2 +#define R200_VS_MAT_0_SPEC 0x000000B3 +#define R200_VS_MAT_1_EMISS 0x000000B4 +#define R200_VS_MAT_1_AMB 0x000000B5 +#define R200_VS_MAT_1_DIF 0x000000B6 +#define R200_VS_MAT_1_SPEC 0x000000B7 +#define R200_VS_EYE2CLIP_MTX 0x000000B8 +#define R200_VS_PNT_SPRITE_ATT_CONST 0x000000BC +#define R200_VS_PNT_SPRITE_EYE_IN_MODEL 0x000000BD +#define R200_VS_PNT_SPRITE_CLAMP 0x000000BE +#define R200_VS_MAX 0x000001C0 + + +/* + * Offsets in TCL scalar state + */ +#define R200_SS_LIGHT_DCD_ADDR 0x00000000 +#define R200_SS_LIGHT_DCM_ADDR 0x00000008 +#define R200_SS_LIGHT_SPOT_EXPONENT_ADDR 0x00000010 +#define R200_SS_LIGHT_SPOT_CUTOFF_ADDR 0x00000018 +#define R200_SS_LIGHT_SPECULAR_THRESH_ADDR 0x00000020 +#define R200_SS_LIGHT_RANGE_CUTOFF_SQRD 0x00000028 +#define R200_SS_LIGHT_RANGE_ATT_CONST 0x00000030 +#define R200_SS_VERT_GUARD_CLIP_ADJ_ADDR 0x00000080 +#define R200_SS_VERT_GUARD_DISCARD_ADJ_ADDR 0x00000081 +#define R200_SS_HORZ_GUARD_CLIP_ADJ_ADDR 0x00000082 +#define R200_SS_HORZ_GUARD_DISCARD_ADJ_ADDR 0x00000083 +#define R200_SS_MAT_0_SHININESS 0x00000100 +#define R200_SS_MAT_1_SHININESS 0x00000101 + + +/* + * Matrix indices + */ +#define R200_MTX_MV 0 +#define R200_MTX_IMV 1 +#define R200_MTX_MVP 2 +#define R200_MTX_TEX0 3 +#define R200_MTX_TEX1 4 +#define R200_MTX_TEX2 5 +#define R200_MTX_TEX3 6 +#define R200_MTX_TEX4 7 +#define R200_MTX_TEX5 8 + +/* + * CP type-3 packets + */ +#define R200_CP_CMD_NOP 0xC0001000 +#define R200_CP_CMD_NEXT_CHAR 0xC0001900 +#define R200_CP_CMD_PLY_NEXTSCAN 0xC0001D00 +#define R200_CP_CMD_SET_SCISSORS 0xC0001E00 +#define R200_CP_CMD_LOAD_MICROCODE 0xC0002400 +#define R200_CP_CMD_WAIT_FOR_IDLE 0xC0002600 +#define R200_CP_CMD_3D_DRAW_VBUF 0xC0002800 +#define R200_CP_CMD_3D_DRAW_IMMD 0xC0002900 +#define R200_CP_CMD_3D_DRAW_INDX 0xC0002A00 +#define R200_CP_CMD_LOAD_PALETTE 0xC0002C00 +#define R200_CP_CMD_3D_LOAD_VBPNTR 0xC0002F00 +#define R200_CP_CMD_INDX_BUFFER 0xC0003300 +#define R200_CP_CMD_3D_DRAW_VBUF_2 0xC0003400 +#define R200_CP_CMD_3D_DRAW_IMMD_2 0xC0003500 +#define R200_CP_CMD_3D_DRAW_INDX_2 0xC0003600 +#define R200_CP_CMD_PAINT 0xC0009100 +#define R200_CP_CMD_BITBLT 0xC0009200 +#define R200_CP_CMD_SMALLTEXT 0xC0009300 +#define R200_CP_CMD_HOSTDATA_BLT 0xC0009400 +#define R200_CP_CMD_POLYLINE 0xC0009500 +#define R200_CP_CMD_POLYSCANLINES 0xC0009800 +#define R200_CP_CMD_PAINT_MULTI 0xC0009A00 +#define R200_CP_CMD_BITBLT_MULTI 0xC0009B00 +#define R200_CP_CMD_TRANS_BITBLT 0xC0009C00 + + +#define R200_AGP_TEX_OFFSET 0x02000000 + + + + +#endif + diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_sanity.c b/xc/lib/GL/mesa/src/drv/r200/r200_sanity.c new file mode 100644 index 000000000..0af8aa833 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_sanity.c @@ -0,0 +1,1313 @@ +/* $XFree86$ */ +/************************************************************************** + +Copyright 2002 ATI Technologies Inc., Ontario, Canada, and + Tungsten Graphics Inc, Cedar Park, TX. + +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the "Software"), +to deal in the Software without restriction, including without limitation +on the rights to use, copy, modify, merge, publish, distribute, sub +license, and/or sell copies of the Software, and to permit persons to whom +the Software is furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice (including the next +paragraph) shall be included in all copies or substantial portions of the +Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL +ATI, TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, +DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR +OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE +USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +/* + * Authors: + * Keith Whitwell <keithw@tungstengraphics.com> + * + */ + +#include "r200_context.h" +#include "r200_ioctl.h" +#include "r200_sanity.h" +#include "radeon_reg.h" +#include "r200_reg.h" + +/* Set this '1' to get more verbiage. + */ +#define MORE_VERBOSE 1 + +#if MORE_VERBOSE +#define VERBOSE (R200_DEBUG & DEBUG_VERBOSE) +#define NORMAL (1) +#else +#define VERBOSE 0 +#define NORMAL (R200_DEBUG & DEBUG_VERBOSE) +#endif + + +/* New (1.3) state mechanism. 3 commands (packet, scalar, vector) in + * 1.3 cmdbuffers allow all previous state to be updated as well as + * the tcl scalar and vector areas. + */ +static struct { + int start; + int len; + const char *name; +} packet[RADEON_MAX_STATE_PACKETS] = { + { RADEON_PP_MISC,7,"RADEON_PP_MISC" }, + { RADEON_PP_CNTL,3,"RADEON_PP_CNTL" }, + { RADEON_RB3D_COLORPITCH,1,"RADEON_RB3D_COLORPITCH" }, + { RADEON_RE_LINE_PATTERN,2,"RADEON_RE_LINE_PATTERN" }, + { RADEON_SE_LINE_WIDTH,1,"RADEON_SE_LINE_WIDTH" }, + { RADEON_PP_LUM_MATRIX,1,"RADEON_PP_LUM_MATRIX" }, + { RADEON_PP_ROT_MATRIX_0,2,"RADEON_PP_ROT_MATRIX_0" }, + { RADEON_RB3D_STENCILREFMASK,3,"RADEON_RB3D_STENCILREFMASK" }, + { RADEON_SE_VPORT_XSCALE,6,"RADEON_SE_VPORT_XSCALE" }, + { RADEON_SE_CNTL,2,"RADEON_SE_CNTL" }, + { RADEON_SE_CNTL_STATUS,1,"RADEON_SE_CNTL_STATUS" }, + { RADEON_RE_MISC,1,"RADEON_RE_MISC" }, + { RADEON_PP_TXFILTER_0,6,"RADEON_PP_TXFILTER_0" }, + { RADEON_PP_BORDER_COLOR_0,1,"RADEON_PP_BORDER_COLOR_0" }, + { RADEON_PP_TXFILTER_1,6,"RADEON_PP_TXFILTER_1" }, + { RADEON_PP_BORDER_COLOR_1,1,"RADEON_PP_BORDER_COLOR_1" }, + { RADEON_PP_TXFILTER_2,6,"RADEON_PP_TXFILTER_2" }, + { RADEON_PP_BORDER_COLOR_2,1,"RADEON_PP_BORDER_COLOR_2" }, + { RADEON_SE_ZBIAS_FACTOR,2,"RADEON_SE_ZBIAS_FACTOR" }, + { RADEON_SE_TCL_OUTPUT_VTX_FMT,11,"RADEON_SE_TCL_OUTPUT_VTX_FMT" }, + { RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED,17,"RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED" }, + { R200_PP_TXCBLEND_0, 4, "R200_EMIT_PP_TXCBLEND_0" }, + { R200_PP_TXCBLEND_1, 4, "R200_PP_TXCBLEND_1" }, + { R200_PP_TXCBLEND_2, 4, "R200_PP_TXCBLEND_2" }, + { R200_PP_TXCBLEND_3, 4, "R200_PP_TXCBLEND_3" }, + { R200_PP_TXCBLEND_4, 4, "R200_PP_TXCBLEND_4" }, + { R200_PP_TXCBLEND_5, 4, "R200_PP_TXCBLEND_5" }, + { R200_PP_TXCBLEND_6, 4, "R200_PP_TXCBLEND_6" }, + { R200_PP_TXCBLEND_7, 4, "R200_PP_TXCBLEND_7" }, + { R200_SE_TCL_LIGHT_MODEL_CTL_0, 6, "R200_SE_TCL_LIGHT_MODEL_CTL_0" }, + { R200_PP_TFACTOR_0, 6, "R200_PP_TFACTOR_0" }, + { R200_SE_VTX_FMT_0, 4, "R200_SE_VTX_FMT_0" }, + { R200_SE_VAP_CNTL, 1, "R200_SE_VAP_CNTL" }, + { R200_SE_TCL_MATRIX_SEL_0, 5, "R200_SE_TCL_MATRIX_SEL_0" }, + { R200_SE_TCL_TEX_PROC_CTL_2, 5, "R200_SE_TCL_TEX_PROC_CTL_2" }, + { R200_SE_TCL_UCP_VERT_BLEND_CTL, 1, "R200_SE_TCL_UCP_VERT_BLEND_CTL" }, + { R200_PP_TXFILTER_0, 6, "R200_PP_TXFILTER_0" }, + { R200_PP_TXFILTER_1, 6, "R200_PP_TXFILTER_1" }, + { R200_PP_TXFILTER_2, 6, "R200_PP_TXFILTER_2" }, + { R200_PP_TXFILTER_3, 6, "R200_PP_TXFILTER_3" }, + { R200_PP_TXFILTER_4, 6, "R200_PP_TXFILTER_4" }, + { R200_PP_TXFILTER_5, 6, "R200_PP_TXFILTER_5" }, + { R200_PP_TXOFFSET_0, 1, "R200_PP_TXOFFSET_0" }, + { R200_PP_TXOFFSET_1, 1, "R200_PP_TXOFFSET_1" }, + { R200_PP_TXOFFSET_2, 1, "R200_PP_TXOFFSET_2" }, + { R200_PP_TXOFFSET_3, 1, "R200_PP_TXOFFSET_3" }, + { R200_PP_TXOFFSET_4, 1, "R200_PP_TXOFFSET_4" }, + { R200_PP_TXOFFSET_5, 1, "R200_PP_TXOFFSET_5" }, + { R200_SE_VTE_CNTL, 1, "R200_SE_VTE_CNTL" }, + { R200_SE_TCL_OUTPUT_VTX_COMP_SEL, 1, "R200_SE_TCL_OUTPUT_VTX_COMP_SEL" }, + { R200_PP_TAM_DEBUG3, 1, "R200_PP_TAM_DEBUG3" }, + { R200_PP_CNTL_X, 1, "R200_PP_CNTL_X" }, + { R200_RB3D_DEPTHXY_OFFSET, 1, "R200_RB3D_DEPTHXY_OFFSET" }, + { R200_RE_AUX_SCISSOR_CNTL, 1, "R200_RE_AUX_SCISSOR_CNTL" }, + { R200_RE_SCISSOR_TL_0, 2, "R200_RE_SCISSOR_TL_0" }, + { R200_RE_SCISSOR_TL_1, 2, "R200_RE_SCISSOR_TL_1" }, + { R200_RE_SCISSOR_TL_2, 2, "R200_RE_SCISSOR_TL_2" }, + { R200_SE_VAP_CNTL_STATUS, 1, "R200_SE_VAP_CNTL_STATUS" }, + { R200_SE_VTX_STATE_CNTL, 1, "R200_SE_VTX_STATE_CNTL" }, + { R200_RE_POINTSIZE, 1, "R200_RE_POINTSIZE" }, + { R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0, 4, "R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0" }, +}; + +struct reg_names { + int idx; + const char *name; +}; + +static struct reg_names reg_names[] = { + { R200_PP_MISC, "R200_PP_MISC" }, + { R200_PP_FOG_COLOR, "R200_PP_FOG_COLOR" }, + { R200_RE_SOLID_COLOR, "R200_RE_SOLID_COLOR" }, + { R200_RB3D_BLENDCNTL, "R200_RB3D_BLENDCNTL" }, + { R200_RB3D_DEPTHOFFSET, "R200_RB3D_DEPTHOFFSET" }, + { R200_RB3D_DEPTHPITCH, "R200_RB3D_DEPTHPITCH" }, + { R200_RB3D_ZSTENCILCNTL, "R200_RB3D_ZSTENCILCNTL" }, + { R200_PP_CNTL, "R200_PP_CNTL" }, + { R200_RB3D_CNTL, "R200_RB3D_CNTL" }, + { R200_RB3D_COLOROFFSET, "R200_RB3D_COLOROFFSET" }, + { R200_RE_WIDTH_HEIGHT, "R200_RE_WIDTH_HEIGHT" }, + { R200_RB3D_COLORPITCH, "R200_RB3D_COLORPITCH" }, + { R200_SE_CNTL, "R200_SE_CNTL" }, + { R200_RE_CNTL, "R200_RE_CNTL" }, + { R200_RE_MISC, "R200_RE_MISC" }, + { R200_RE_STIPPLE_ADDR, "R200_RE_STIPPLE_ADDR" }, + { R200_RE_STIPPLE_DATA, "R200_RE_STIPPLE_DATA" }, + { R200_RE_LINE_PATTERN, "R200_RE_LINE_PATTERN" }, + { R200_RE_LINE_STATE, "R200_RE_LINE_STATE" }, + { R200_RE_SCISSOR_TL_0, "R200_RE_SCISSOR_TL_0" }, + { R200_RE_SCISSOR_BR_0, "R200_RE_SCISSOR_BR_0" }, + { R200_RE_SCISSOR_TL_1, "R200_RE_SCISSOR_TL_1" }, + { R200_RE_SCISSOR_BR_1, "R200_RE_SCISSOR_BR_1" }, + { R200_RE_SCISSOR_TL_2, "R200_RE_SCISSOR_TL_2" }, + { R200_RE_SCISSOR_BR_2, "R200_RE_SCISSOR_BR_2" }, + { R200_RB3D_DEPTHXY_OFFSET, "R200_RB3D_DEPTHXY_OFFSET" }, + { R200_RB3D_STENCILREFMASK, "R200_RB3D_STENCILREFMASK" }, + { R200_RB3D_ROPCNTL, "R200_RB3D_ROPCNTL" }, + { R200_RB3D_PLANEMASK, "R200_RB3D_PLANEMASK" }, + { R200_SE_VPORT_XSCALE, "R200_SE_VPORT_XSCALE" }, + { R200_SE_VPORT_XOFFSET, "R200_SE_VPORT_XOFFSET" }, + { R200_SE_VPORT_YSCALE, "R200_SE_VPORT_YSCALE" }, + { R200_SE_VPORT_YOFFSET, "R200_SE_VPORT_YOFFSET" }, + { R200_SE_VPORT_ZSCALE, "R200_SE_VPORT_ZSCALE" }, + { R200_SE_VPORT_ZOFFSET, "R200_SE_VPORT_ZOFFSET" }, + { R200_SE_ZBIAS_FACTOR, "R200_SE_ZBIAS_FACTOR" }, + { R200_SE_ZBIAS_CONSTANT, "R200_SE_ZBIAS_CONSTANT" }, + { R200_SE_LINE_WIDTH, "R200_SE_LINE_WIDTH" }, + { R200_SE_VAP_CNTL, "R200_SE_VAP_CNTL" }, + { R200_SE_VF_CNTL, "R200_SE_VF_CNTL" }, + { R200_SE_VTX_FMT_0, "R200_SE_VTX_FMT_0" }, + { R200_SE_VTX_FMT_1, "R200_SE_VTX_FMT_1" }, + { R200_SE_TCL_OUTPUT_VTX_FMT_0, "R200_SE_TCL_OUTPUT_VTX_FMT_0" }, + { R200_SE_TCL_OUTPUT_VTX_FMT_1, "R200_SE_TCL_OUTPUT_VTX_FMT_1" }, + { R200_SE_VTE_CNTL, "R200_SE_VTE_CNTL" }, + { R200_SE_VTX_NUM_ARRAYS, "R200_SE_VTX_NUM_ARRAYS" }, + { R200_SE_VTX_AOS_ATTR01, "R200_SE_VTX_AOS_ATTR01" }, + { R200_SE_VTX_AOS_ADDR0, "R200_SE_VTX_AOS_ADDR0" }, + { R200_SE_VTX_AOS_ADDR1, "R200_SE_VTX_AOS_ADDR1" }, + { R200_SE_VTX_AOS_ATTR23, "R200_SE_VTX_AOS_ATTR23" }, + { R200_SE_VTX_AOS_ADDR2, "R200_SE_VTX_AOS_ADDR2" }, + { R200_SE_VTX_AOS_ADDR3, "R200_SE_VTX_AOS_ADDR3" }, + { R200_SE_VTX_AOS_ATTR45, "R200_SE_VTX_AOS_ATTR45" }, + { R200_SE_VTX_AOS_ADDR4, "R200_SE_VTX_AOS_ADDR4" }, + { R200_SE_VTX_AOS_ADDR5, "R200_SE_VTX_AOS_ADDR5" }, + { R200_SE_VTX_AOS_ATTR67, "R200_SE_VTX_AOS_ATTR67" }, + { R200_SE_VTX_AOS_ADDR6, "R200_SE_VTX_AOS_ADDR6" }, + { R200_SE_VTX_AOS_ADDR7, "R200_SE_VTX_AOS_ADDR7" }, + { R200_SE_VTX_AOS_ATTR89, "R200_SE_VTX_AOS_ATTR89" }, + { R200_SE_VTX_AOS_ADDR8, "R200_SE_VTX_AOS_ADDR8" }, + { R200_SE_VTX_AOS_ADDR9, "R200_SE_VTX_AOS_ADDR9" }, + { R200_SE_VTX_AOS_ATTR1011, "R200_SE_VTX_AOS_ATTR1011" }, + { R200_SE_VTX_AOS_ADDR10, "R200_SE_VTX_AOS_ADDR10" }, + { R200_SE_VTX_AOS_ADDR11, "R200_SE_VTX_AOS_ADDR11" }, + { R200_SE_VF_MAX_VTX_INDX, "R200_SE_VF_MAX_VTX_INDX" }, + { R200_SE_VF_MIN_VTX_INDX, "R200_SE_VF_MIN_VTX_INDX" }, + { R200_SE_VTX_STATE_CNTL, "R200_SE_VTX_STATE_CNTL" }, + { R200_SE_TCL_VECTOR_INDX_REG, "R200_SE_TCL_VECTOR_INDX_REG" }, + { R200_SE_TCL_VECTOR_DATA_REG, "R200_SE_TCL_VECTOR_DATA_REG" }, + { R200_SE_TCL_SCALAR_INDX_REG, "R200_SE_TCL_SCALAR_INDX_REG" }, + { R200_SE_TCL_SCALAR_DATA_REG, "R200_SE_TCL_SCALAR_DATA_REG" }, + { R200_SE_TCL_MATRIX_SEL_0, "R200_SE_TCL_MATRIX_SEL_0" }, + { R200_SE_TCL_MATRIX_SEL_1, "R200_SE_TCL_MATRIX_SEL_1" }, + { R200_SE_TCL_MATRIX_SEL_2, "R200_SE_TCL_MATRIX_SEL_2" }, + { R200_SE_TCL_MATRIX_SEL_3, "R200_SE_TCL_MATRIX_SEL_3" }, + { R200_SE_TCL_MATRIX_SEL_4, "R200_SE_TCL_MATRIX_SEL_4" }, + { R200_SE_TCL_LIGHT_MODEL_CTL_0, "R200_SE_TCL_LIGHT_MODEL_CTL_0" }, + { R200_SE_TCL_LIGHT_MODEL_CTL_1, "R200_SE_TCL_LIGHT_MODEL_CTL_1" }, + { R200_SE_TCL_PER_LIGHT_CTL_0, "R200_SE_TCL_PER_LIGHT_CTL_0" }, + { R200_SE_TCL_PER_LIGHT_CTL_1, "R200_SE_TCL_PER_LIGHT_CTL_1" }, + { R200_SE_TCL_PER_LIGHT_CTL_2, "R200_SE_TCL_PER_LIGHT_CTL_2" }, + { R200_SE_TCL_PER_LIGHT_CTL_3, "R200_SE_TCL_PER_LIGHT_CTL_3" }, + { R200_SE_TCL_TEX_PROC_CTL_2, "R200_SE_TCL_TEX_PROC_CTL_2" }, + { R200_SE_TCL_TEX_PROC_CTL_3, "R200_SE_TCL_TEX_PROC_CTL_3" }, + { R200_SE_TCL_TEX_PROC_CTL_0, "R200_SE_TCL_TEX_PROC_CTL_0" }, + { R200_SE_TCL_TEX_PROC_CTL_1, "R200_SE_TCL_TEX_PROC_CTL_1" }, + { R200_SE_TC_TEX_CYL_WRAP_CTL, "R200_SE_TC_TEX_CYL_WRAP_CTL" }, + { R200_SE_TCL_UCP_VERT_BLEND_CTL, "R200_SE_TCL_UCP_VERT_BLEND_CTL" }, + { R200_SE_TCL_POINT_SPRITE_CNTL, "R200_SE_TCL_POINT_SPRITE_CNTL" }, + { R200_SE_VTX_ST_POS_0_X_4, "R200_SE_VTX_ST_POS_0_X_4" }, + { R200_SE_VTX_ST_POS_0_Y_4, "R200_SE_VTX_ST_POS_0_Y_4" }, + { R200_SE_VTX_ST_POS_0_Z_4, "R200_SE_VTX_ST_POS_0_Z_4" }, + { R200_SE_VTX_ST_POS_0_W_4, "R200_SE_VTX_ST_POS_0_W_4" }, + { R200_SE_VTX_ST_NORM_0_X, "R200_SE_VTX_ST_NORM_0_X" }, + { R200_SE_VTX_ST_NORM_0_Y, "R200_SE_VTX_ST_NORM_0_Y" }, + { R200_SE_VTX_ST_NORM_0_Z, "R200_SE_VTX_ST_NORM_0_Z" }, + { R200_SE_VTX_ST_PVMS, "R200_SE_VTX_ST_PVMS" }, + { R200_SE_VTX_ST_CLR_0_R, "R200_SE_VTX_ST_CLR_0_R" }, + { R200_SE_VTX_ST_CLR_0_G, "R200_SE_VTX_ST_CLR_0_G" }, + { R200_SE_VTX_ST_CLR_0_B, "R200_SE_VTX_ST_CLR_0_B" }, + { R200_SE_VTX_ST_CLR_0_A, "R200_SE_VTX_ST_CLR_0_A" }, + { R200_SE_VTX_ST_CLR_1_R, "R200_SE_VTX_ST_CLR_1_R" }, + { R200_SE_VTX_ST_CLR_1_G, "R200_SE_VTX_ST_CLR_1_G" }, + { R200_SE_VTX_ST_CLR_1_B, "R200_SE_VTX_ST_CLR_1_B" }, + { R200_SE_VTX_ST_CLR_1_A, "R200_SE_VTX_ST_CLR_1_A" }, + { R200_SE_VTX_ST_CLR_2_R, "R200_SE_VTX_ST_CLR_2_R" }, + { R200_SE_VTX_ST_CLR_2_G, "R200_SE_VTX_ST_CLR_2_G" }, + { R200_SE_VTX_ST_CLR_2_B, "R200_SE_VTX_ST_CLR_2_B" }, + { R200_SE_VTX_ST_CLR_2_A, "R200_SE_VTX_ST_CLR_2_A" }, + { R200_SE_VTX_ST_CLR_3_R, "R200_SE_VTX_ST_CLR_3_R" }, + { R200_SE_VTX_ST_CLR_3_G, "R200_SE_VTX_ST_CLR_3_G" }, + { R200_SE_VTX_ST_CLR_3_B, "R200_SE_VTX_ST_CLR_3_B" }, + { R200_SE_VTX_ST_CLR_3_A, "R200_SE_VTX_ST_CLR_3_A" }, + { R200_SE_VTX_ST_CLR_4_R, "R200_SE_VTX_ST_CLR_4_R" }, + { R200_SE_VTX_ST_CLR_4_G, "R200_SE_VTX_ST_CLR_4_G" }, + { R200_SE_VTX_ST_CLR_4_B, "R200_SE_VTX_ST_CLR_4_B" }, + { R200_SE_VTX_ST_CLR_4_A, "R200_SE_VTX_ST_CLR_4_A" }, + { R200_SE_VTX_ST_CLR_5_R, "R200_SE_VTX_ST_CLR_5_R" }, + { R200_SE_VTX_ST_CLR_5_G, "R200_SE_VTX_ST_CLR_5_G" }, + { R200_SE_VTX_ST_CLR_5_B, "R200_SE_VTX_ST_CLR_5_B" }, + { R200_SE_VTX_ST_CLR_5_A, "R200_SE_VTX_ST_CLR_5_A" }, + { R200_SE_VTX_ST_CLR_6_R, "R200_SE_VTX_ST_CLR_6_R" }, + { R200_SE_VTX_ST_CLR_6_G, "R200_SE_VTX_ST_CLR_6_G" }, + { R200_SE_VTX_ST_CLR_6_B, "R200_SE_VTX_ST_CLR_6_B" }, + { R200_SE_VTX_ST_CLR_6_A, "R200_SE_VTX_ST_CLR_6_A" }, + { R200_SE_VTX_ST_CLR_7_R, "R200_SE_VTX_ST_CLR_7_R" }, + { R200_SE_VTX_ST_CLR_7_G, "R200_SE_VTX_ST_CLR_7_G" }, + { R200_SE_VTX_ST_CLR_7_B, "R200_SE_VTX_ST_CLR_7_B" }, + { R200_SE_VTX_ST_CLR_7_A, "R200_SE_VTX_ST_CLR_7_A" }, + { R200_SE_VTX_ST_TEX_0_S, "R200_SE_VTX_ST_TEX_0_S" }, + { R200_SE_VTX_ST_TEX_0_T, "R200_SE_VTX_ST_TEX_0_T" }, + { R200_SE_VTX_ST_TEX_0_R, "R200_SE_VTX_ST_TEX_0_R" }, + { R200_SE_VTX_ST_TEX_0_Q, "R200_SE_VTX_ST_TEX_0_Q" }, + { R200_SE_VTX_ST_TEX_1_S, "R200_SE_VTX_ST_TEX_1_S" }, + { R200_SE_VTX_ST_TEX_1_T, "R200_SE_VTX_ST_TEX_1_T" }, + { R200_SE_VTX_ST_TEX_1_R, "R200_SE_VTX_ST_TEX_1_R" }, + { R200_SE_VTX_ST_TEX_1_Q, "R200_SE_VTX_ST_TEX_1_Q" }, + { R200_SE_VTX_ST_TEX_2_S, "R200_SE_VTX_ST_TEX_2_S" }, + { R200_SE_VTX_ST_TEX_2_T, "R200_SE_VTX_ST_TEX_2_T" }, + { R200_SE_VTX_ST_TEX_2_R, "R200_SE_VTX_ST_TEX_2_R" }, + { R200_SE_VTX_ST_TEX_2_Q, "R200_SE_VTX_ST_TEX_2_Q" }, + { R200_SE_VTX_ST_TEX_3_S, "R200_SE_VTX_ST_TEX_3_S" }, + { R200_SE_VTX_ST_TEX_3_T, "R200_SE_VTX_ST_TEX_3_T" }, + { R200_SE_VTX_ST_TEX_3_R, "R200_SE_VTX_ST_TEX_3_R" }, + { R200_SE_VTX_ST_TEX_3_Q, "R200_SE_VTX_ST_TEX_3_Q" }, + { R200_SE_VTX_ST_TEX_4_S, "R200_SE_VTX_ST_TEX_4_S" }, + { R200_SE_VTX_ST_TEX_4_T, "R200_SE_VTX_ST_TEX_4_T" }, + { R200_SE_VTX_ST_TEX_4_R, "R200_SE_VTX_ST_TEX_4_R" }, + { R200_SE_VTX_ST_TEX_4_Q, "R200_SE_VTX_ST_TEX_4_Q" }, + { R200_SE_VTX_ST_TEX_5_S, "R200_SE_VTX_ST_TEX_5_S" }, + { R200_SE_VTX_ST_TEX_5_T, "R200_SE_VTX_ST_TEX_5_T" }, + { R200_SE_VTX_ST_TEX_5_R, "R200_SE_VTX_ST_TEX_5_R" }, + { R200_SE_VTX_ST_TEX_5_Q, "R200_SE_VTX_ST_TEX_5_Q" }, + { R200_SE_VTX_ST_PNT_SPRT_SZ, "R200_SE_VTX_ST_PNT_SPRT_SZ" }, + { R200_SE_VTX_ST_DISC_FOG, "R200_SE_VTX_ST_DISC_FOG" }, + { R200_SE_VTX_ST_SHININESS_0, "R200_SE_VTX_ST_SHININESS_0" }, + { R200_SE_VTX_ST_SHININESS_1, "R200_SE_VTX_ST_SHININESS_1" }, + { R200_SE_VTX_ST_BLND_WT_0, "R200_SE_VTX_ST_BLND_WT_0" }, + { R200_SE_VTX_ST_BLND_WT_1, "R200_SE_VTX_ST_BLND_WT_1" }, + { R200_SE_VTX_ST_BLND_WT_2, "R200_SE_VTX_ST_BLND_WT_2" }, + { R200_SE_VTX_ST_BLND_WT_3, "R200_SE_VTX_ST_BLND_WT_3" }, + { R200_SE_VTX_ST_POS_1_X, "R200_SE_VTX_ST_POS_1_X" }, + { R200_SE_VTX_ST_POS_1_Y, "R200_SE_VTX_ST_POS_1_Y" }, + { R200_SE_VTX_ST_POS_1_Z, "R200_SE_VTX_ST_POS_1_Z" }, + { R200_SE_VTX_ST_POS_1_W, "R200_SE_VTX_ST_POS_1_W" }, + { R200_SE_VTX_ST_NORM_1_X, "R200_SE_VTX_ST_NORM_1_X" }, + { R200_SE_VTX_ST_NORM_1_Y, "R200_SE_VTX_ST_NORM_1_Y" }, + { R200_SE_VTX_ST_NORM_1_Z, "R200_SE_VTX_ST_NORM_1_Z" }, + { R200_SE_VTX_ST_USR_CLR_0_R, "R200_SE_VTX_ST_USR_CLR_0_R" }, + { R200_SE_VTX_ST_USR_CLR_0_G, "R200_SE_VTX_ST_USR_CLR_0_G" }, + { R200_SE_VTX_ST_USR_CLR_0_B, "R200_SE_VTX_ST_USR_CLR_0_B" }, + { R200_SE_VTX_ST_USR_CLR_0_A, "R200_SE_VTX_ST_USR_CLR_0_A" }, + { R200_SE_VTX_ST_USR_CLR_1_R, "R200_SE_VTX_ST_USR_CLR_1_R" }, + { R200_SE_VTX_ST_USR_CLR_1_G, "R200_SE_VTX_ST_USR_CLR_1_G" }, + { R200_SE_VTX_ST_USR_CLR_1_B, "R200_SE_VTX_ST_USR_CLR_1_B" }, + { R200_SE_VTX_ST_USR_CLR_1_A, "R200_SE_VTX_ST_USR_CLR_1_A" }, + { R200_SE_VTX_ST_CLR_0_PKD, "R200_SE_VTX_ST_CLR_0_PKD" }, + { R200_SE_VTX_ST_CLR_1_PKD, "R200_SE_VTX_ST_CLR_1_PKD" }, + { R200_SE_VTX_ST_CLR_2_PKD, "R200_SE_VTX_ST_CLR_2_PKD" }, + { R200_SE_VTX_ST_CLR_3_PKD, "R200_SE_VTX_ST_CLR_3_PKD" }, + { R200_SE_VTX_ST_CLR_4_PKD, "R200_SE_VTX_ST_CLR_4_PKD" }, + { R200_SE_VTX_ST_CLR_5_PKD, "R200_SE_VTX_ST_CLR_5_PKD" }, + { R200_SE_VTX_ST_CLR_6_PKD, "R200_SE_VTX_ST_CLR_6_PKD" }, + { R200_SE_VTX_ST_CLR_7_PKD, "R200_SE_VTX_ST_CLR_7_PKD" }, + { R200_SE_VTX_ST_POS_0_X_2, "R200_SE_VTX_ST_POS_0_X_2" }, + { R200_SE_VTX_ST_POS_0_Y_2, "R200_SE_VTX_ST_POS_0_Y_2" }, + { R200_SE_VTX_ST_PAR_CLR_LD, "R200_SE_VTX_ST_PAR_CLR_LD" }, + { R200_SE_VTX_ST_USR_CLR_PKD, "R200_SE_VTX_ST_USR_CLR_PKD" }, + { R200_SE_VTX_ST_POS_0_X_3, "R200_SE_VTX_ST_POS_0_X_3" }, + { R200_SE_VTX_ST_POS_0_Y_3, "R200_SE_VTX_ST_POS_0_Y_3" }, + { R200_SE_VTX_ST_POS_0_Z_3, "R200_SE_VTX_ST_POS_0_Z_3" }, + { R200_SE_VTX_ST_END_OF_PKT, "R200_SE_VTX_ST_END_OF_PKT" }, + { R200_RE_POINTSIZE, "R200_RE_POINTSIZE" }, + { R200_RE_TOP_LEFT, "R200_RE_TOP_LEFT" }, + { R200_RE_AUX_SCISSOR_CNTL, "R200_RE_AUX_SCISSOR_CNTL" }, + { R200_PP_TXFILTER_0, "R200_PP_TXFILTER_0" }, + { R200_PP_TXFORMAT_0, "R200_PP_TXFORMAT_0" }, + { R200_PP_TXSIZE_0, "R200_PP_TXSIZE_0" }, + { R200_PP_TXFORMAT_X_0, "R200_PP_TXFORMAT_X_0" }, + { R200_PP_TXPITCH_0, "R200_PP_TXPITCH_0" }, + { R200_PP_BORDER_COLOR_0, "R200_PP_BORDER_COLOR_0" }, + { R200_PP_CUBIC_FACES_0, "R200_PP_CUBIC_FACES_0" }, + { R200_PP_TXFILTER_1, "R200_PP_TXFILTER_1" }, + { R200_PP_TXFORMAT_1, "R200_PP_TXFORMAT_1" }, + { R200_PP_TXSIZE_1, "R200_PP_TXSIZE_1" }, + { R200_PP_TXFORMAT_X_1, "R200_PP_TXFORMAT_X_1" }, + { R200_PP_TXPITCH_1, "R200_PP_TXPITCH_1" }, + { R200_PP_BORDER_COLOR_1, "R200_PP_BORDER_COLOR_1" }, + { R200_PP_CUBIC_FACES_1, "R200_PP_CUBIC_FACES_1" }, + { R200_PP_TXFILTER_2, "R200_PP_TXFILTER_2" }, + { R200_PP_TXFORMAT_2, "R200_PP_TXFORMAT_2" }, + { R200_PP_TXSIZE_2, "R200_PP_TXSIZE_2" }, + { R200_PP_TXFORMAT_X_2, "R200_PP_TXFORMAT_X_2" }, + { R200_PP_TXPITCH_2, "R200_PP_TXPITCH_2" }, + { R200_PP_BORDER_COLOR_2, "R200_PP_BORDER_COLOR_2" }, + { R200_PP_CUBIC_FACES_2, "R200_PP_CUBIC_FACES_2" }, + { R200_PP_TXFILTER_3, "R200_PP_TXFILTER_3" }, + { R200_PP_TXFORMAT_3, "R200_PP_TXFORMAT_3" }, + { R200_PP_TXSIZE_3, "R200_PP_TXSIZE_3" }, + { R200_PP_TXFORMAT_X_3, "R200_PP_TXFORMAT_X_3" }, + { R200_PP_TXPITCH_3, "R200_PP_TXPITCH_3" }, + { R200_PP_BORDER_COLOR_3, "R200_PP_BORDER_COLOR_3" }, + { R200_PP_CUBIC_FACES_3, "R200_PP_CUBIC_FACES_3" }, + { R200_PP_TXFILTER_4, "R200_PP_TXFILTER_4" }, + { R200_PP_TXFORMAT_4, "R200_PP_TXFORMAT_4" }, + { R200_PP_TXSIZE_4, "R200_PP_TXSIZE_4" }, + { R200_PP_TXFORMAT_X_4, "R200_PP_TXFORMAT_X_4" }, + { R200_PP_TXPITCH_4, "R200_PP_TXPITCH_4" }, + { R200_PP_BORDER_COLOR_4, "R200_PP_BORDER_COLOR_4" }, + { R200_PP_CUBIC_FACES_4, "R200_PP_CUBIC_FACES_4" }, + { R200_PP_TXFILTER_5, "R200_PP_TXFILTER_5" }, + { R200_PP_TXFORMAT_5, "R200_PP_TXFORMAT_5" }, + { R200_PP_TXSIZE_5, "R200_PP_TXSIZE_5" }, + { R200_PP_TXFORMAT_X_5, "R200_PP_TXFORMAT_X_5" }, + { R200_PP_TXPITCH_5, "R200_PP_TXPITCH_5" }, + { R200_PP_BORDER_COLOR_5, "R200_PP_BORDER_COLOR_5" }, + { R200_PP_CUBIC_FACES_5, "R200_PP_CUBIC_FACES_5" }, + { R200_PP_TXOFFSET_0, "R200_PP_TXOFFSET_0" }, + { R200_PP_CUBIC_OFFSET_F1_0, "R200_PP_CUBIC_OFFSET_F1_0" }, + { R200_PP_CUBIC_OFFSET_F2_0, "R200_PP_CUBIC_OFFSET_F2_0" }, + { R200_PP_CUBIC_OFFSET_F3_0, "R200_PP_CUBIC_OFFSET_F3_0" }, + { R200_PP_CUBIC_OFFSET_F4_0, "R200_PP_CUBIC_OFFSET_F4_0" }, + { R200_PP_CUBIC_OFFSET_F5_0, "R200_PP_CUBIC_OFFSET_F5_0" }, + { R200_PP_TXOFFSET_1, "R200_PP_TXOFFSET_1" }, + { R200_PP_CUBIC_OFFSET_F1_1, "R200_PP_CUBIC_OFFSET_F1_1" }, + { R200_PP_CUBIC_OFFSET_F2_1, "R200_PP_CUBIC_OFFSET_F2_1" }, + { R200_PP_CUBIC_OFFSET_F3_1, "R200_PP_CUBIC_OFFSET_F3_1" }, + { R200_PP_CUBIC_OFFSET_F4_1, "R200_PP_CUBIC_OFFSET_F4_1" }, + { R200_PP_CUBIC_OFFSET_F5_1, "R200_PP_CUBIC_OFFSET_F5_1" }, + { R200_PP_TXOFFSET_2, "R200_PP_TXOFFSET_2" }, + { R200_PP_CUBIC_OFFSET_F1_2, "R200_PP_CUBIC_OFFSET_F1_2" }, + { R200_PP_CUBIC_OFFSET_F2_2, "R200_PP_CUBIC_OFFSET_F2_2" }, + { R200_PP_CUBIC_OFFSET_F3_2, "R200_PP_CUBIC_OFFSET_F3_2" }, + { R200_PP_CUBIC_OFFSET_F4_2, "R200_PP_CUBIC_OFFSET_F4_2" }, + { R200_PP_CUBIC_OFFSET_F5_2, "R200_PP_CUBIC_OFFSET_F5_2" }, + { R200_PP_TXOFFSET_3, "R200_PP_TXOFFSET_3" }, + { R200_PP_CUBIC_OFFSET_F1_3, "R200_PP_CUBIC_OFFSET_F1_3" }, + { R200_PP_CUBIC_OFFSET_F2_3, "R200_PP_CUBIC_OFFSET_F2_3" }, + { R200_PP_CUBIC_OFFSET_F3_3, "R200_PP_CUBIC_OFFSET_F3_3" }, + { R200_PP_CUBIC_OFFSET_F4_3, "R200_PP_CUBIC_OFFSET_F4_3" }, + { R200_PP_CUBIC_OFFSET_F5_3, "R200_PP_CUBIC_OFFSET_F5_3" }, + { R200_PP_TXOFFSET_4, "R200_PP_TXOFFSET_4" }, + { R200_PP_CUBIC_OFFSET_F1_4, "R200_PP_CUBIC_OFFSET_F1_4" }, + { R200_PP_CUBIC_OFFSET_F2_4, "R200_PP_CUBIC_OFFSET_F2_4" }, + { R200_PP_CUBIC_OFFSET_F3_4, "R200_PP_CUBIC_OFFSET_F3_4" }, + { R200_PP_CUBIC_OFFSET_F4_4, "R200_PP_CUBIC_OFFSET_F4_4" }, + { R200_PP_CUBIC_OFFSET_F5_4, "R200_PP_CUBIC_OFFSET_F5_4" }, + { R200_PP_TXOFFSET_5, "R200_PP_TXOFFSET_5" }, + { R200_PP_CUBIC_OFFSET_F1_5, "R200_PP_CUBIC_OFFSET_F1_5" }, + { R200_PP_CUBIC_OFFSET_F2_5, "R200_PP_CUBIC_OFFSET_F2_5" }, + { R200_PP_CUBIC_OFFSET_F3_5, "R200_PP_CUBIC_OFFSET_F3_5" }, + { R200_PP_CUBIC_OFFSET_F4_5, "R200_PP_CUBIC_OFFSET_F4_5" }, + { R200_PP_CUBIC_OFFSET_F5_5, "R200_PP_CUBIC_OFFSET_F5_5" }, + { R200_PP_TAM_DEBUG3, "R200_PP_TAM_DEBUG3" }, + { R200_PP_TFACTOR_0, "R200_PP_TFACTOR_0" }, + { R200_PP_TFACTOR_1, "R200_PP_TFACTOR_1" }, + { R200_PP_TFACTOR_2, "R200_PP_TFACTOR_2" }, + { R200_PP_TFACTOR_3, "R200_PP_TFACTOR_3" }, + { R200_PP_TFACTOR_4, "R200_PP_TFACTOR_4" }, + { R200_PP_TFACTOR_5, "R200_PP_TFACTOR_5" }, + { R200_PP_TXCBLEND_0, "R200_PP_TXCBLEND_0" }, + { R200_PP_TXCBLEND2_0, "R200_PP_TXCBLEND2_0" }, + { R200_PP_TXABLEND_0, "R200_PP_TXABLEND_0" }, + { R200_PP_TXABLEND2_0, "R200_PP_TXABLEND2_0" }, + { R200_PP_TXCBLEND_1, "R200_PP_TXCBLEND_1" }, + { R200_PP_TXCBLEND2_1, "R200_PP_TXCBLEND2_1" }, + { R200_PP_TXABLEND_1, "R200_PP_TXABLEND_1" }, + { R200_PP_TXABLEND2_1, "R200_PP_TXABLEND2_1" }, + { R200_PP_TXCBLEND_2, "R200_PP_TXCBLEND_2" }, + { R200_PP_TXCBLEND2_2, "R200_PP_TXCBLEND2_2" }, + { R200_PP_TXABLEND_2, "R200_PP_TXABLEND_2" }, + { R200_PP_TXABLEND2_2, "R200_PP_TXABLEND2_2" }, + { R200_PP_TXCBLEND_3, "R200_PP_TXCBLEND_3" }, + { R200_PP_TXCBLEND2_3, "R200_PP_TXCBLEND2_3" }, + { R200_PP_TXABLEND_3, "R200_PP_TXABLEND_3" }, + { R200_PP_TXABLEND2_3, "R200_PP_TXABLEND2_3" }, + { R200_PP_TXCBLEND_4, "R200_PP_TXCBLEND_4" }, + { R200_PP_TXCBLEND2_4, "R200_PP_TXCBLEND2_4" }, + { R200_PP_TXABLEND_4, "R200_PP_TXABLEND_4" }, + { R200_PP_TXABLEND2_4, "R200_PP_TXABLEND2_4" }, + { R200_PP_TXCBLEND_5, "R200_PP_TXCBLEND_5" }, + { R200_PP_TXCBLEND2_5, "R200_PP_TXCBLEND2_5" }, + { R200_PP_TXABLEND_5, "R200_PP_TXABLEND_5" }, + { R200_PP_TXABLEND2_5, "R200_PP_TXABLEND2_5" }, + { R200_PP_TXCBLEND_6, "R200_PP_TXCBLEND_6" }, + { R200_PP_TXCBLEND2_6, "R200_PP_TXCBLEND2_6" }, + { R200_PP_TXABLEND_6, "R200_PP_TXABLEND_6" }, + { R200_PP_TXABLEND2_6, "R200_PP_TXABLEND2_6" }, + { R200_PP_TXCBLEND_7, "R200_PP_TXCBLEND_7" }, + { R200_PP_TXCBLEND2_7, "R200_PP_TXCBLEND2_7" }, + { R200_PP_TXABLEND_7, "R200_PP_TXABLEND_7" }, + { R200_PP_TXABLEND2_7, "R200_PP_TXABLEND2_7" }, + { R200_RB3D_ABLENDCNTL, "R200_RB3D_ABLENDCNTL" }, + { R200_RB3D_CBLENDCNTL, "R200_RB3D_CBLENDCNTL" }, + { R200_SE_TCL_OUTPUT_VTX_COMP_SEL, "R200_SE_TCL_OUTPUT_VTX_COMP_SEL" }, + { R200_PP_CNTL_X, "R200_PP_CNTL_X" }, + { R200_SE_VAP_CNTL_STATUS, "R200_SE_VAP_CNTL_STATUS" }, + { R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0, "R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0" }, + { R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_1, "R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_1" }, + { R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_2, "R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_2" }, + { R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_3, "R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_3" }, +}; + +static struct reg_names scalar_names[] = { + { R200_SS_LIGHT_DCD_ADDR, "R200_SS_LIGHT_DCD_ADDR" }, + { R200_SS_LIGHT_DCM_ADDR, "R200_SS_LIGHT_DCM_ADDR" }, + { R200_SS_LIGHT_SPOT_EXPONENT_ADDR, "R200_SS_LIGHT_SPOT_EXPONENT_ADDR" }, + { R200_SS_LIGHT_SPOT_CUTOFF_ADDR, "R200_SS_LIGHT_SPOT_CUTOFF_ADDR" }, + { R200_SS_LIGHT_SPECULAR_THRESH_ADDR, "R200_SS_LIGHT_SPECULAR_THRESH_ADDR" }, + { R200_SS_LIGHT_RANGE_CUTOFF_SQRD, "R200_SS_LIGHT_RANGE_CUTOFF_SQRD" }, + { R200_SS_LIGHT_RANGE_ATT_CONST, "R200_SS_LIGHT_RANGE_ATT_CONST" }, + { R200_SS_VERT_GUARD_CLIP_ADJ_ADDR, "R200_SS_VERT_GUARD_CLIP_ADJ_ADDR" }, + { R200_SS_VERT_GUARD_DISCARD_ADJ_ADDR, "R200_SS_VERT_GUARD_DISCARD_ADJ_ADDR" }, + { R200_SS_HORZ_GUARD_CLIP_ADJ_ADDR, "R200_SS_HORZ_GUARD_CLIP_ADJ_ADDR" }, + { R200_SS_HORZ_GUARD_DISCARD_ADJ_ADDR, "R200_SS_HORZ_GUARD_DISCARD_ADJ_ADDR" }, + { R200_SS_MAT_0_SHININESS, "R200_SS_MAT_0_SHININESS" }, + { R200_SS_MAT_1_SHININESS, "R200_SS_MAT_1_SHININESS" }, + { 1000, "" }, +}; + +/* Puff these out to make them look like normal (dword) registers. + */ +static struct reg_names vector_names[] = { + { 0, "start" }, + { R200_VS_LIGHT_AMBIENT_ADDR, "R200_VS_LIGHT_AMBIENT_ADDR" }, + { R200_VS_LIGHT_DIFFUSE_ADDR, "R200_VS_LIGHT_DIFFUSE_ADDR" }, + { R200_VS_LIGHT_SPECULAR_ADDR, "R200_VS_LIGHT_SPECULAR_ADDR" }, + { R200_VS_LIGHT_DIRPOS_ADDR, "R200_VS_LIGHT_DIRPOS_ADDR" }, + { R200_VS_LIGHT_HWVSPOT_ADDR, "R200_VS_LIGHT_HWVSPOT_ADDR" }, + { R200_VS_LIGHT_ATTENUATION_ADDR, "R200_VS_LIGHT_ATTENUATION_ADDR" }, + { R200_VS_SPOT_DUAL_CONE, "R200_VS_SPOT_DUAL_CONE" }, + { R200_VS_GLOBAL_AMBIENT_ADDR, "R200_VS_GLOBAL_AMBIENT_ADDR" }, + { R200_VS_FOG_PARAM_ADDR, "R200_VS_FOG_PARAM_ADDR" }, + { R200_VS_EYE_VECTOR_ADDR, "R200_VS_EYE_VECTOR_ADDR" }, + { R200_VS_UCP_ADDR, "R200_VS_UCP_ADDR" }, + { R200_VS_PNT_SPRITE_VPORT_SCALE, "R200_VS_PNT_SPRITE_VPORT_SCALE" }, + { R200_VS_MATRIX_0_MV, "R200_VS_MATRIX_0_MV" }, + { R200_VS_MATRIX_1_INV_MV, "R200_VS_MATRIX_1_INV_MV" }, + { R200_VS_MATRIX_2_MVP, "R200_VS_MATRIX_2_MVP" }, + { R200_VS_MATRIX_3_TEX0, "R200_VS_MATRIX_3_TEX0" }, + { R200_VS_MATRIX_4_TEX1, "R200_VS_MATRIX_4_TEX1" }, + { R200_VS_MATRIX_5_TEX2, "R200_VS_MATRIX_5_TEX2" }, + { R200_VS_MATRIX_6_TEX3, "R200_VS_MATRIX_6_TEX3" }, + { R200_VS_MATRIX_7_TEX4, "R200_VS_MATRIX_7_TEX4" }, + { R200_VS_MATRIX_8_TEX5, "R200_VS_MATRIX_8_TEX5" }, + { R200_VS_MAT_0_EMISS, "R200_VS_MAT_0_EMISS" }, + { R200_VS_MAT_0_AMB, "R200_VS_MAT_0_AMB" }, + { R200_VS_MAT_0_DIF, "R200_VS_MAT_0_DIF" }, + { R200_VS_MAT_0_SPEC, "R200_VS_MAT_0_SPEC" }, + { R200_VS_MAT_1_EMISS, "R200_VS_MAT_1_EMISS" }, + { R200_VS_MAT_1_AMB, "R200_VS_MAT_1_AMB" }, + { R200_VS_MAT_1_DIF, "R200_VS_MAT_1_DIF" }, + { R200_VS_MAT_1_SPEC, "R200_VS_MAT_1_SPEC" }, + { R200_VS_EYE2CLIP_MTX, "R200_VS_EYE2CLIP_MTX" }, + { R200_VS_PNT_SPRITE_ATT_CONST, "R200_VS_PNT_SPRITE_ATT_CONST" }, + { R200_VS_PNT_SPRITE_EYE_IN_MODEL, "R200_VS_PNT_SPRITE_EYE_IN_MODEL" }, + { R200_VS_PNT_SPRITE_CLAMP, "R200_VS_PNT_SPRITE_CLAMP" }, + { R200_VS_MAX, "R200_VS_MAX" }, + { 1000, "" }, +}; + +union fi { float f; int i; }; + +#define ISVEC 1 +#define ISFLOAT 2 +#define TOUCHED 4 + +struct reg { + int idx; + struct reg_names *closest; + int flags; + union fi current; + union fi *values; + int nvalues; + int nalloc; + float vmin, vmax; +}; + + +static struct reg regs[Elements(reg_names)+1]; +static struct reg scalars[512+1]; +static struct reg vectors[512*4+1]; + +static int total, total_changed, bufs; + +static void init_regs( void ) +{ + struct reg_names *tmp; + int i; + + for (i = 0 ; i < Elements(regs) ; i++) { + regs[i].idx = reg_names[i].idx; + regs[i].closest = ®_names[i]; + regs[i].flags = 0; + } + + for (i = 0, tmp = scalar_names ; i < Elements(scalars) ; i++) { + if (tmp[1].idx == i) tmp++; + scalars[i].idx = i; + scalars[i].closest = tmp; + scalars[i].flags = ISFLOAT; + } + + for (i = 0, tmp = vector_names ; i < Elements(vectors) ; i++) { + if (tmp[1].idx*4 == i) tmp++; + vectors[i].idx = i; + vectors[i].closest = tmp; + vectors[i].flags = ISFLOAT|ISVEC; + } + + regs[Elements(regs)-1].idx = -1; + scalars[Elements(scalars)-1].idx = -1; + vectors[Elements(vectors)-1].idx = -1; +} + +static int find_or_add_value( struct reg *reg, int val ) +{ + int j; + + for ( j = 0 ; j < reg->nvalues ; j++) + if ( val == reg->values[j].i ) + return 1; + + if (j == reg->nalloc) { + reg->nalloc += 5; + reg->nalloc *= 2; + reg->values = (union fi *) realloc( reg->values, + reg->nalloc * sizeof(union fi) ); + } + + reg->values[reg->nvalues++].i = val; + return 0; +} + +static struct reg *lookup_reg( struct reg *tab, int reg ) +{ + int i; + + for (i = 0 ; tab[i].idx != -1 ; i++) { + if (tab[i].idx == reg) + return &tab[i]; + } + + fprintf(stderr, "*** unknown reg 0x%x\n", reg); + return 0; +} + + +static const char *get_reg_name( struct reg *reg ) +{ + static char tmp[80]; + + if (reg->idx == reg->closest->idx) + return reg->closest->name; + + + if (reg->flags & ISVEC) { + if (reg->idx/4 != reg->closest->idx) + sprintf(tmp, "%s+%d[%d]", + reg->closest->name, + (reg->idx/4) - reg->closest->idx, + reg->idx%4); + else + sprintf(tmp, "%s[%d]", reg->closest->name, reg->idx%4); + } + else { + if (reg->idx != reg->closest->idx) + sprintf(tmp, "%s+%d", reg->closest->name, reg->idx - reg->closest->idx); + else + sprintf(tmp, "%s", reg->closest->name); + } + + return tmp; +} + +static int print_int_reg_assignment( struct reg *reg, int data ) +{ + int changed = (reg->current.i != data); + int ever_seen = find_or_add_value( reg, data ); + + if (VERBOSE || (NORMAL && (changed || !ever_seen))) + fprintf(stderr, " %s <-- 0x%x", get_reg_name(reg), data); + + if (NORMAL) { + if (!ever_seen) + fprintf(stderr, " *** BRAND NEW VALUE"); + else if (changed) + fprintf(stderr, " *** CHANGED"); + } + + reg->current.i = data; + + if (VERBOSE || (NORMAL && (changed || !ever_seen))) + fprintf(stderr, "\n"); + + return changed; +} + + +static int print_float_reg_assignment( struct reg *reg, float data ) +{ + int changed = (reg->current.f != data); + int newmin = (data < reg->vmin); + int newmax = (data > reg->vmax); + + if (VERBOSE || (NORMAL && (newmin || newmax || changed))) + fprintf(stderr, " %s <-- %.3f", get_reg_name(reg), data); + + if (NORMAL) { + if (newmin) { + fprintf(stderr, " *** NEW MIN (prev %.3f)", reg->vmin); + reg->vmin = data; + } + else if (newmax) { + fprintf(stderr, " *** NEW MAX (prev %.3f)", reg->vmax); + reg->vmax = data; + } + else if (changed) { + fprintf(stderr, " *** CHANGED"); + } + } + + reg->current.f = data; + + if (VERBOSE || (NORMAL && (newmin || newmax || changed))) + fprintf(stderr, "\n"); + + return changed; +} + +static int print_reg_assignment( struct reg *reg, int data ) +{ + reg->flags |= TOUCHED; + if (reg->flags & ISFLOAT) + return print_float_reg_assignment( reg, *(float *)&data ); + else + return print_int_reg_assignment( reg, data ); +} + +static void print_reg( struct reg *reg ) +{ + if (reg->flags & TOUCHED) { + if (reg->flags & ISFLOAT) { + fprintf(stderr, " %s == %f\n", get_reg_name(reg), reg->current.f); + } else { + fprintf(stderr, " %s == 0x%x\n", get_reg_name(reg), reg->current.i); + } + } +} + + +static void dump_state( void ) +{ + int i; + + for (i = 0 ; i < Elements(regs) ; i++) + print_reg( ®s[i] ); + + for (i = 0 ; i < Elements(scalars) ; i++) + print_reg( &scalars[i] ); + + for (i = 0 ; i < Elements(vectors) ; i++) + print_reg( &vectors[i] ); +} + + + +static int radeon_emit_packets( + drmRadeonCmdHeader header, + drmRadeonCmdBuffer *cmdbuf ) +{ + int id = (int)header.packet.packet_id; + int sz = packet[id].len; + int *data = (int *)cmdbuf->buf; + int i; + + if (sz * sizeof(int) > cmdbuf->bufsz) { + fprintf(stderr, "Packet overflows cmdbuf\n"); + return -EINVAL; + } + + if (!packet[id].name) { + fprintf(stderr, "*** Unknown packet 0 nr %d\n", id ); + return -EINVAL; + } + + + if (VERBOSE) + fprintf(stderr, "Packet 0 reg %s nr %d\n", packet[id].name, sz ); + + for ( i = 0 ; i < sz ; i++) { + struct reg *reg = lookup_reg( regs, packet[id].start + i*4 ); + if (print_reg_assignment( reg, data[i] )) + total_changed++; + total++; + } + + cmdbuf->buf += sz * sizeof(int); + cmdbuf->bufsz -= sz * sizeof(int); + return 0; +} + + +static int radeon_emit_scalars( + drmRadeonCmdHeader header, + drmRadeonCmdBuffer *cmdbuf ) +{ + int sz = header.scalars.count; + int *data = (int *)cmdbuf->buf; + int start = header.scalars.offset; + int stride = header.scalars.stride; + int i; + + if (VERBOSE) + fprintf(stderr, "emit scalars, start %d stride %d nr %d (end %d)\n", + start, stride, sz, start + stride * sz); + + + for (i = 0 ; i < sz ; i++, start += stride) { + struct reg *reg = lookup_reg( scalars, start ); + if (print_reg_assignment( reg, data[i] )) + total_changed++; + total++; + } + + cmdbuf->buf += sz * sizeof(int); + cmdbuf->bufsz -= sz * sizeof(int); + return 0; +} + + +static int radeon_emit_scalars2( + drmRadeonCmdHeader header, + drmRadeonCmdBuffer *cmdbuf ) +{ + int sz = header.scalars.count; + int *data = (int *)cmdbuf->buf; + int start = header.scalars.offset + 0x100; + int stride = header.scalars.stride; + int i; + + if (VERBOSE) + fprintf(stderr, "emit scalars2, start %d stride %d nr %d (end %d)\n", + start, stride, sz, start + stride * sz); + + if (start + stride * sz > 257) { + fprintf(stderr, "emit scalars OVERFLOW %d/%d/%d\n", start, stride, sz); + return -1; + } + + for (i = 0 ; i < sz ; i++, start += stride) { + struct reg *reg = lookup_reg( scalars, start ); + if (print_reg_assignment( reg, data[i] )) + total_changed++; + total++; + } + + cmdbuf->buf += sz * sizeof(int); + cmdbuf->bufsz -= sz * sizeof(int); + return 0; +} + +/* Check: inf/nan/extreme-size? + * Check: table start, end, nr, etc. + */ +static int radeon_emit_vectors( + drmRadeonCmdHeader header, + drmRadeonCmdBuffer *cmdbuf ) +{ + int sz = header.vectors.count; + int *data = (int *)cmdbuf->buf; + int start = header.vectors.offset; + int stride = header.vectors.stride; + int i,j; + + if (VERBOSE) + fprintf(stderr, "emit vectors, start %d stride %d nr %d (end %d) (0x%x)\n", + start, stride, sz, start + stride * sz, header.i); + +/* if (start + stride * (sz/4) > 128) { */ +/* fprintf(stderr, "emit vectors OVERFLOW %d/%d/%d\n", start, stride, sz); */ +/* return -1; */ +/* } */ + + for (i = 0 ; i < sz ; start += stride) { + int changed = 0; + for (j = 0 ; j < 4 ; i++,j++) { + struct reg *reg = lookup_reg( vectors, start*4+j ); + if (print_reg_assignment( reg, data[i] )) + changed = 1; + } + if (changed) + total_changed += 4; + total += 4; + } + + + cmdbuf->buf += sz * sizeof(int); + cmdbuf->bufsz -= sz * sizeof(int); + return 0; +} + +#if 0 +static int print_vertex_format( int vfmt ) +{ + if (NORMAL) { + fprintf(stderr, " %s(%x): %s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s", + "vertex format", + vfmt, + "xy,", + (vfmt & R200_VTX_Z0) ? "z," : "", + (vfmt & R200_VTX_W0) ? "w0," : "", + (vfmt & R200_VTX_FPCOLOR) ? "fpcolor," : "", + (vfmt & R200_VTX_FPALPHA) ? "fpalpha," : "", + (vfmt & R200_VTX_PKCOLOR) ? "pkcolor," : "", + (vfmt & R200_VTX_FPSPEC) ? "fpspec," : "", + (vfmt & R200_VTX_FPFOG) ? "fpfog," : "", + (vfmt & R200_VTX_PKSPEC) ? "pkspec," : "", + (vfmt & R200_VTX_ST0) ? "st0," : "", + (vfmt & R200_VTX_ST1) ? "st1," : "", + (vfmt & R200_VTX_Q1) ? "q1," : "", + (vfmt & R200_VTX_ST2) ? "st2," : "", + (vfmt & R200_VTX_Q2) ? "q2," : "", + (vfmt & R200_VTX_ST3) ? "st3," : "", + (vfmt & R200_VTX_Q3) ? "q3," : "", + (vfmt & R200_VTX_Q0) ? "q0," : "", + (vfmt & R200_VTX_N0) ? "n0," : "", + (vfmt & R200_VTX_XY1) ? "xy1," : "", + (vfmt & R200_VTX_Z1) ? "z1," : "", + (vfmt & R200_VTX_W1) ? "w1," : "", + (vfmt & R200_VTX_N1) ? "n1," : ""); + + + if (!find_or_add_value( &others[V_VTXFMT], vfmt )) + fprintf(stderr, " *** NEW VALUE"); + + fprintf(stderr, "\n"); + } + + return 0; +} +#endif + +static char *primname[0x10] = { + "NONE", + "POINTS", + "LINES", + "LINE_STRIP", + "TRIANGLES", + "TRIANGLE_FAN", + "TRIANGLE_STRIP", + "RECT_LIST", + 0, + "3VRT_POINTS", + "3VRT_LINES", + "POINT_SPRITES", + "LINE_LOOP", + "QUADS", + "QUAD_STRIP", + "POLYGON", +}; + +static int print_prim_and_flags( int prim ) +{ + int numverts; + + if (NORMAL) + fprintf(stderr, " %s(%x): %s%s%s%s%s%s\n", + "prim flags", + prim, + ((prim & 0x30) == R200_VF_PRIM_WALK_IND) ? "IND," : "", + ((prim & 0x30) == R200_VF_PRIM_WALK_LIST) ? "LIST," : "", + ((prim & 0x30) == R200_VF_PRIM_WALK_RING) ? "RING," : "", + (prim & R200_VF_COLOR_ORDER_RGBA) ? "RGBA," : "BGRA, ", + (prim & R200_VF_INDEX_SZ_4) ? "INDX-32," : "", + (prim & R200_VF_TCL_OUTPUT_VTX_ENABLE) ? "TCL_OUT_VTX," : ""); + + numverts = prim>>16; + + if (NORMAL) + fprintf(stderr, " prim: %s numverts %d\n", primname[prim&0xf], numverts); + + switch (prim & 0xf) { + case R200_VF_PRIM_NONE: + case R200_VF_PRIM_POINTS: + if (numverts < 1) { + fprintf(stderr, "Bad nr verts for line %d\n", numverts); + return -1; + } + break; + case R200_VF_PRIM_LINES: + case R200_VF_PRIM_POINT_SPRITES: + if ((numverts & 1) || numverts == 0) { + fprintf(stderr, "Bad nr verts for line %d\n", numverts); + return -1; + } + break; + case R200_VF_PRIM_LINE_STRIP: + case R200_VF_PRIM_LINE_LOOP: + if (numverts < 2) { + fprintf(stderr, "Bad nr verts for line_strip %d\n", numverts); + return -1; + } + break; + case R200_VF_PRIM_TRIANGLES: + case R200_VF_PRIM_3VRT_POINTS: + case R200_VF_PRIM_3VRT_LINES: + case R200_VF_PRIM_RECT_LIST: + if (numverts % 3 || numverts == 0) { + fprintf(stderr, "Bad nr verts for tri %d\n", numverts); + return -1; + } + break; + case R200_VF_PRIM_TRIANGLE_FAN: + case R200_VF_PRIM_TRIANGLE_STRIP: + case R200_VF_PRIM_POLYGON: + if (numverts < 3) { + fprintf(stderr, "Bad nr verts for strip/fan %d\n", numverts); + return -1; + } + break; + case R200_VF_PRIM_QUADS: + if (numverts % 4 || numverts == 0) { + fprintf(stderr, "Bad nr verts for quad %d\n", numverts); + return -1; + } + break; + case R200_VF_PRIM_QUAD_STRIP: + if (numverts % 2 || numverts < 4) { + fprintf(stderr, "Bad nr verts for quadstrip %d\n", numverts); + return -1; + } + break; + default: + fprintf(stderr, "Bad primitive\n"); + return -1; + } + return 0; +} + +/* build in knowledge about each packet type + */ +static int radeon_emit_packet3( drmRadeonCmdBuffer *cmdbuf ) +{ + int cmdsz; + int *cmd = (int *)cmdbuf->buf; + int *tmp; + int i, stride, size, start; + + cmdsz = 2 + ((cmd[0] & RADEON_CP_PACKET_COUNT_MASK) >> 16); + + if ((cmd[0] & RADEON_CP_PACKET_MASK) != RADEON_CP_PACKET3 || + cmdsz * 4 > cmdbuf->bufsz || + cmdsz > RADEON_CP_PACKET_MAX_DWORDS) { + fprintf(stderr, "Bad packet\n"); + return -EINVAL; + } + + switch( cmd[0] & ~RADEON_CP_PACKET_COUNT_MASK ) { + case R200_CP_CMD_NOP: + if (NORMAL) + fprintf(stderr, "PACKET3_NOP, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_NEXT_CHAR: + if (NORMAL) + fprintf(stderr, "PACKET3_NEXT_CHAR, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_PLY_NEXTSCAN: + if (NORMAL) + fprintf(stderr, "PACKET3_PLY_NEXTSCAN, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_SET_SCISSORS: + if (NORMAL) + fprintf(stderr, "PACKET3_SET_SCISSORS, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_LOAD_MICROCODE: + if (NORMAL) + fprintf(stderr, "PACKET3_LOAD_MICROCODE, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_WAIT_FOR_IDLE: + if (NORMAL) + fprintf(stderr, "PACKET3_WAIT_FOR_IDLE, %d dwords\n", cmdsz); + break; + + case R200_CP_CMD_3D_DRAW_VBUF: + if (NORMAL) + fprintf(stderr, "PACKET3_3D_DRAW_VBUF, %d dwords\n", cmdsz); +/* print_vertex_format(cmd[1]); */ + if (print_prim_and_flags(cmd[2])) + return -EINVAL; + break; + + case R200_CP_CMD_3D_DRAW_IMMD: + if (NORMAL) + fprintf(stderr, "PACKET3_3D_DRAW_IMMD, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_3D_DRAW_INDX: { + int neltdwords; + if (NORMAL) + fprintf(stderr, "PACKET3_3D_DRAW_INDX, %d dwords\n", cmdsz); +/* print_vertex_format(cmd[1]); */ + if (print_prim_and_flags(cmd[2])) + return -EINVAL; + neltdwords = cmd[2]>>16; + neltdwords += neltdwords & 1; + neltdwords /= 2; + if (neltdwords + 3 != cmdsz) + fprintf(stderr, "Mismatch in DRAW_INDX, %d vs cmdsz %d\n", + neltdwords, cmdsz); + break; + } + case R200_CP_CMD_LOAD_PALETTE: + if (NORMAL) + fprintf(stderr, "PACKET3_LOAD_PALETTE, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_3D_LOAD_VBPNTR: + if (NORMAL) { + fprintf(stderr, "PACKET3_3D_LOAD_VBPNTR, %d dwords\n", cmdsz); + fprintf(stderr, " nr arrays: %d\n", cmd[1]); + } + + if (((cmd[1]/2)*3) + ((cmd[1]%2)*2) != cmdsz - 2) { + fprintf(stderr, " ****** MISMATCH %d/%d *******\n", + ((cmd[1]/2)*3) + ((cmd[1]%2)*2) + 2, cmdsz); + return -EINVAL; + } + + if (NORMAL) { + tmp = cmd+2; + for (i = 0 ; i < cmd[1] ; i++) { + if (i & 1) { + stride = (tmp[0]>>24) & 0xff; + size = (tmp[0]>>16) & 0xff; + start = tmp[2]; + tmp += 3; + } + else { + stride = (tmp[0]>>8) & 0xff; + size = (tmp[0]) & 0xff; + start = tmp[1]; + } + fprintf(stderr, " array %d: start 0x%x vsize %d vstride %d\n", + i, start, size, stride ); + } + } + break; + case R200_CP_CMD_PAINT: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_PAINT, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_BITBLT: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_BITBLT, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_SMALLTEXT: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_SMALLTEXT, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_HOSTDATA_BLT: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_HOSTDATA_BLT, %d dwords\n", + cmdsz); + break; + case R200_CP_CMD_POLYLINE: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_POLYLINE, %d dwords\n", cmdsz); + break; + case R200_CP_CMD_POLYSCANLINES: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_POLYSCANLINES, %d dwords\n", + cmdsz); + break; + case R200_CP_CMD_PAINT_MULTI: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_PAINT_MULTI, %d dwords\n", + cmdsz); + break; + case R200_CP_CMD_BITBLT_MULTI: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_BITBLT_MULTI, %d dwords\n", + cmdsz); + break; + case R200_CP_CMD_TRANS_BITBLT: + if (NORMAL) + fprintf(stderr, "PACKET3_CNTL_TRANS_BITBLT, %d dwords\n", + cmdsz); + break; + case R200_CP_CMD_3D_DRAW_VBUF_2: + if (NORMAL) + fprintf(stderr, "R200_CP_CMD_3D_DRAW_VBUF_2, %d dwords\n", + cmdsz); + if (print_prim_and_flags(cmd[1])) + return -EINVAL; + break; + case R200_CP_CMD_3D_DRAW_IMMD_2: + if (NORMAL) + fprintf(stderr, "R200_CP_CMD_3D_DRAW_IMMD_2, %d dwords\n", + cmdsz); + if (print_prim_and_flags(cmd[1])) + return -EINVAL; + break; + case R200_CP_CMD_3D_DRAW_INDX_2: + if (NORMAL) + fprintf(stderr, "R200_CP_CMD_3D_DRAW_INDX_2, %d dwords\n", + cmdsz); + if (print_prim_and_flags(cmd[1])) + return -EINVAL; + break; + default: + fprintf(stderr, "UNKNOWN PACKET, %d dwords\n", cmdsz); + break; + } + + cmdbuf->buf += cmdsz * 4; + cmdbuf->bufsz -= cmdsz * 4; + return 0; +} + + +/* Check cliprects for bounds, then pass on to above: + */ +static int radeon_emit_packet3_cliprect( drmRadeonCmdBuffer *cmdbuf ) +{ + XF86DRIClipRectRec *boxes = (XF86DRIClipRectRec *)cmdbuf->boxes; + int i = 0; + + if (VERBOSE && total_changed) { + dump_state(); + total_changed = 0; + } + + if (NORMAL) { + do { + if ( i < cmdbuf->nbox ) { + fprintf(stderr, "Emit box %d/%d %d,%d %d,%d\n", + i, cmdbuf->nbox, + boxes[i].x1, boxes[i].y1, boxes[i].x2, boxes[i].y2); + } + } while ( ++i < cmdbuf->nbox ); + } + + if (cmdbuf->nbox == 1) + cmdbuf->nbox = 0; + + return radeon_emit_packet3( cmdbuf ); +} + + +int r200SanityCmdBuffer( r200ContextPtr rmesa, + int nbox, + XF86DRIClipRectRec *boxes ) +{ + int idx; + drmRadeonCmdBuffer cmdbuf; + drmRadeonCmdHeader header; + static int inited = 0; + + if (!inited) { + init_regs(); + inited = 1; + } + + + cmdbuf.buf = rmesa->store.cmd_buf; + cmdbuf.bufsz = rmesa->store.cmd_used; + cmdbuf.boxes = (drmClipRect *)boxes; + cmdbuf.nbox = nbox; + + while ( cmdbuf.bufsz >= sizeof(header) ) { + + header.i = *(int *)cmdbuf.buf; + cmdbuf.buf += sizeof(header); + cmdbuf.bufsz -= sizeof(header); + + switch (header.header.cmd_type) { + case RADEON_CMD_PACKET: + if (radeon_emit_packets( header, &cmdbuf )) { + fprintf(stderr,"radeon_emit_packets failed\n"); + return -EINVAL; + } + break; + + case RADEON_CMD_SCALARS: + if (radeon_emit_scalars( header, &cmdbuf )) { + fprintf(stderr,"radeon_emit_scalars failed\n"); + return -EINVAL; + } + break; + + case RADEON_CMD_SCALARS2: + if (radeon_emit_scalars2( header, &cmdbuf )) { + fprintf(stderr,"radeon_emit_scalars failed\n"); + return -EINVAL; + } + break; + + case RADEON_CMD_VECTORS: + if (radeon_emit_vectors( header, &cmdbuf )) { + fprintf(stderr,"radeon_emit_vectors failed\n"); + return -EINVAL; + } + break; + + case RADEON_CMD_DMA_DISCARD: + idx = header.dma.buf_idx; + if (NORMAL) + fprintf(stderr, "RADEON_CMD_DMA_DISCARD buf %d\n", idx); + bufs++; + break; + + case RADEON_CMD_PACKET3: + if (radeon_emit_packet3( &cmdbuf )) { + fprintf(stderr,"radeon_emit_packet3 failed\n"); + return -EINVAL; + } + break; + + case RADEON_CMD_PACKET3_CLIP: + if (radeon_emit_packet3_cliprect( &cmdbuf )) { + fprintf(stderr,"radeon_emit_packet3_clip failed\n"); + return -EINVAL; + } + break; + + default: + fprintf(stderr,"bad cmd_type %d at %p\n", + header.header.cmd_type, + cmdbuf.buf - sizeof(header)); + return -EINVAL; + } + } + + if (0) + { + static int n = 0; + n++; + if (n == 10) { + fprintf(stderr, "Bufs %d Total emitted %d real changes %d (%.2f%%)\n", + bufs, + total, total_changed, + ((float)total_changed/(float)total*100.0)); + fprintf(stderr, "Total emitted per buf: %.2f\n", + (float)total/(float)bufs); + fprintf(stderr, "Real changes per buf: %.2f\n", + (float)total_changed/(float)bufs); + + bufs = n = total = total_changed = 0; + } + } + + fprintf(stderr, "leaving %s\n\n\n", __FUNCTION__); + + return 0; +} + + +/* Do the same job to a native command stream + * -- pull apart packets after they are built. + * -- understand SCALAR, VECTOR stores + * -- understand INDIRECT registers & trace down into indirect buffers. + */ diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_sanity.h b/xc/lib/GL/mesa/src/drv/r200/r200_sanity.h new file mode 100644 index 000000000..10260f211 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_sanity.h @@ -0,0 +1,8 @@ +#ifndef R200_SANITY_H +#define R200_SANITY_H + +extern int r200SanityCmdBuffer( r200ContextPtr rmesa, + int nbox, + XF86DRIClipRectRec *boxes ); + +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_screen.c b/xc/lib/GL/mesa/src/drv/r200/r200_screen.c new file mode 100644 index 000000000..63510b23c --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_screen.c @@ -0,0 +1,235 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_screen.h" +#include "r200_context.h" +#include "r200_ioctl.h" + +#include "mem.h" + +#if 1 +/* Including xf86PciInfo.h introduces a bunch of errors... + */ +#define PCI_CHIP_R200_QD 0x5144 +#define PCI_CHIP_R200_QE 0x5145 +#define PCI_CHIP_R200_QF 0x5146 +#define PCI_CHIP_R200_QG 0x5147 +#define PCI_CHIP_R200_QY 0x5159 +#define PCI_CHIP_R200_QZ 0x515A +#define PCI_CHIP_R200_LW 0x4C57 +#define PCI_CHIP_R200_LY 0x4C59 +#define PCI_CHIP_R200_LZ 0x4C5A +#define PCI_CHIP_RV200_QW 0x5157 +#endif + + +/* Create the device specific screen private data struct. + */ +r200ScreenPtr r200CreateScreen( __DRIscreenPrivate *sPriv ) +{ + r200ScreenPtr r200Screen; + RADEONDRIPtr r200DRIPriv = (RADEONDRIPtr)sPriv->pDevPriv; + + /* Check the DRI version */ + { + int major, minor, patch; + if ( XF86DRIQueryVersion( sPriv->display, &major, &minor, &patch ) ) { + if ( major != 4 || minor < 0 ) { + __driUtilMessage( "R200 DRI driver expected DRI version 4.0.x " + "but got version %d.%d.%d", + major, minor, patch ); + return NULL; + } + } + } + + /* Check that the DDX driver version is compatible */ + if ( sPriv->ddxMajor != 4 || + sPriv->ddxMinor < 0 ) { + __driUtilMessage( "R200 DRI driver expected DDX driver version 4.0.x " + "but got version %d.%d.%d", + sPriv->ddxMajor, sPriv->ddxMinor, sPriv->ddxPatch ); + return NULL; + } + + /* Check that the DRM driver version is compatible + * -- R200 support added at 1.4.0. + */ + if ( sPriv->drmMajor != 1 || + sPriv->drmMinor < 5) { + __driUtilMessage( "R200 DRI driver expected DRM driver version 1.5.x " + "but got version %d.%d.%d", + sPriv->drmMajor, sPriv->drmMinor, sPriv->drmPatch ); + return NULL; + } + + + /* Allocate the private area */ + r200Screen = (r200ScreenPtr) CALLOC( sizeof(*r200Screen) ); + if ( !r200Screen ) { + __driUtilMessage("%s: CALLOC r200Screen struct failed", + __FUNCTION__); + return NULL; + } + + + /* This is first since which regions we map depends on whether or + * not we are using a PCI card. + */ + r200Screen->IsPCI = r200DRIPriv->IsPCI; + + r200Screen->mmio.handle = r200DRIPriv->registerHandle; + r200Screen->mmio.size = r200DRIPriv->registerSize; + if ( drmMap( sPriv->fd, + r200Screen->mmio.handle, + r200Screen->mmio.size, + &r200Screen->mmio.map ) ) { + FREE( r200Screen ); + __driUtilMessage("r200CreateScreen(): drmMap failed\n"); + return NULL; + } + + r200Screen->status.handle = r200DRIPriv->statusHandle; + r200Screen->status.size = r200DRIPriv->statusSize; + if ( drmMap( sPriv->fd, + r200Screen->status.handle, + r200Screen->status.size, + &r200Screen->status.map ) ) { + drmUnmap( r200Screen->mmio.map, r200Screen->mmio.size ); + FREE( r200Screen ); + __driUtilMessage("r200CreateScreen(): drmMap (2) failed\n"); + return NULL; + } + r200Screen->scratch = (__volatile__ CARD32 *) + ((GLubyte *)r200Screen->status.map + RADEON_SCRATCH_REG_OFFSET); + + r200Screen->buffers = drmMapBufs( sPriv->fd ); + if ( !r200Screen->buffers ) { + drmUnmap( r200Screen->status.map, r200Screen->status.size ); + drmUnmap( r200Screen->mmio.map, r200Screen->mmio.size ); + FREE( r200Screen ); + __driUtilMessage("r200CreateScreen(): drmMapBufs failed\n"); + return NULL; + } + + if ( !r200Screen->IsPCI ) { + r200Screen->agpTextures.handle = r200DRIPriv->agpTexHandle; + r200Screen->agpTextures.size = r200DRIPriv->agpTexMapSize; + if ( drmMap( sPriv->fd, + r200Screen->agpTextures.handle, + r200Screen->agpTextures.size, + (drmAddressPtr)&r200Screen->agpTextures.map ) ) { + drmUnmapBufs( r200Screen->buffers ); + drmUnmap( r200Screen->status.map, r200Screen->status.size ); + drmUnmap( r200Screen->mmio.map, r200Screen->mmio.size ); + FREE( r200Screen ); + __driUtilMessage("r200CreateScreen(): IsPCI failed\n"); + return NULL; + } + } + + + switch ( r200DRIPriv->deviceID ) { + case PCI_CHIP_R200_QD: + case PCI_CHIP_R200_QE: + case PCI_CHIP_R200_QF: + case PCI_CHIP_R200_QG: + case PCI_CHIP_R200_QY: + case PCI_CHIP_R200_QZ: + case PCI_CHIP_RV200_QW: + case PCI_CHIP_R200_LW: + case PCI_CHIP_R200_LY: + case PCI_CHIP_R200_LZ: + __driUtilMessage("r200CreateScreen(): Device isn't an r200!\n"); + return NULL; + break; + default: + r200Screen->chipset = R200_CHIPSET_R200; + break; + } + + r200Screen->cpp = r200DRIPriv->bpp / 8; + r200Screen->AGPMode = r200DRIPriv->AGPMode; + + r200Screen->frontOffset = r200DRIPriv->frontOffset; + r200Screen->frontPitch = r200DRIPriv->frontPitch; + r200Screen->backOffset = r200DRIPriv->backOffset; + r200Screen->backPitch = r200DRIPriv->backPitch; + r200Screen->depthOffset = r200DRIPriv->depthOffset; + r200Screen->depthPitch = r200DRIPriv->depthPitch; + + r200Screen->texOffset[RADEON_CARD_HEAP] = r200DRIPriv->textureOffset; + r200Screen->texSize[RADEON_CARD_HEAP] = r200DRIPriv->textureSize; + r200Screen->logTexGranularity[RADEON_CARD_HEAP] = + r200DRIPriv->log2TexGran; + + if ( r200Screen->IsPCI ) { + r200Screen->numTexHeaps = RADEON_NR_TEX_HEAPS - 1; + r200Screen->texOffset[RADEON_AGP_HEAP] = 0; + r200Screen->texSize[RADEON_AGP_HEAP] = 0; + r200Screen->logTexGranularity[RADEON_AGP_HEAP] = 0; + } else { + r200Screen->numTexHeaps = RADEON_NR_TEX_HEAPS; + r200Screen->texOffset[RADEON_AGP_HEAP] = + r200DRIPriv->agpTexOffset + R200_AGP_TEX_OFFSET; + r200Screen->texSize[RADEON_AGP_HEAP] = r200DRIPriv->agpTexMapSize; + r200Screen->logTexGranularity[RADEON_AGP_HEAP] = + r200DRIPriv->log2AGPTexGran; + } + + r200Screen->driScreen = sPriv; + r200Screen->sarea_priv_offset = r200DRIPriv->sarea_priv_offset; + return r200Screen; +} + +/* Destroy the device specific screen private data struct. + */ +void r200DestroyScreen( __DRIscreenPrivate *sPriv ) +{ + r200ScreenPtr r200Screen = (r200ScreenPtr)sPriv->private; + + if (!r200Screen) + return; + + if ( !r200Screen->IsPCI ) { + drmUnmap( r200Screen->agpTextures.map, + r200Screen->agpTextures.size ); + } + drmUnmapBufs( r200Screen->buffers ); + drmUnmap( r200Screen->status.map, r200Screen->status.size ); + drmUnmap( r200Screen->mmio.map, r200Screen->mmio.size ); + + FREE( r200Screen ); + sPriv->private = NULL; +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_screen.h b/xc/lib/GL/mesa/src/drv/r200/r200_screen.h new file mode 100644 index 000000000..89148892f --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_screen.h @@ -0,0 +1,96 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_SCREEN_H__ +#define __R200_SCREEN_H__ + +#ifdef GLX_DIRECT_RENDERING + +#include <X11/Xlibint.h> +#include "dri_util.h" +#include "xf86drm.h" +#include "radeon_common.h" +#include "radeon_sarea.h" + +typedef struct { + drmHandle handle; /* Handle to the DRM region */ + drmSize size; /* Size of the DRM region */ + drmAddress map; /* Mapping of the DRM region */ +} r200RegionRec, *r200RegionPtr; + +#define R200_CHIPSET_R200 1 +#define R200_CHIPSET_MOBILITY 2 + +#define R200_DRM_CURRENT 5 /* Current drm minor version */ + +#define R200_NR_TEX_HEAPS 2 + +typedef struct { + + int chipset; + int cpp; + int IsPCI; /* Current card is a PCI card */ + int AGPMode; + + unsigned int frontOffset; + unsigned int frontPitch; + unsigned int backOffset; + unsigned int backPitch; + + unsigned int depthOffset; + unsigned int depthPitch; + + /* Shared texture data */ + int numTexHeaps; + int texOffset[R200_NR_TEX_HEAPS]; + int texSize[R200_NR_TEX_HEAPS]; + int logTexGranularity[R200_NR_TEX_HEAPS]; + + r200RegionRec mmio; + r200RegionRec status; + r200RegionRec agpTextures; + + drmBufMapPtr buffers; + + __volatile__ CARD32 *scratch; + + __DRIscreenPrivate *driScreen; + unsigned int sarea_priv_offset; +} r200ScreenRec, *r200ScreenPtr; + +extern r200ScreenPtr r200CreateScreen( __DRIscreenPrivate *sPriv ); +extern void r200DestroyScreen( __DRIscreenPrivate *sPriv ); + +#endif +#endif /* __R200_SCREEN_H__ */ diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_span.c b/xc/lib/GL/mesa/src/drv/r200/r200_span.c new file mode 100644 index 000000000..d67534794 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_span.c @@ -0,0 +1,415 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_ioctl.h" +#include "r200_state.h" +#include "r200_span.h" +#include "r200_tex.h" + +#include "swrast/swrast.h" + +#define DBG 0 + +#define LOCAL_VARS \ + r200ContextPtr rmesa = R200_CONTEXT(ctx); \ + r200ScreenPtr r200Screen = rmesa->r200Screen; \ + __DRIscreenPrivate *sPriv = rmesa->dri.screen; \ + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; \ + GLuint pitch = r200Screen->frontPitch * r200Screen->cpp; \ + GLuint height = dPriv->h; \ + char *buf = (char *)(sPriv->pFB + \ + rmesa->state.color.drawOffset + \ + (dPriv->x * r200Screen->cpp) + \ + (dPriv->y * pitch)); \ + char *read_buf = (char *)(sPriv->pFB + \ + rmesa->state.pixel.readOffset + \ + (dPriv->x * r200Screen->cpp) + \ + (dPriv->y * pitch)); \ + GLuint p; \ + (void) read_buf; (void) buf; (void) p + +#define LOCAL_DEPTH_VARS \ + r200ContextPtr rmesa = R200_CONTEXT(ctx); \ + r200ScreenPtr r200Screen = rmesa->r200Screen; \ + __DRIscreenPrivate *sPriv = rmesa->dri.screen; \ + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; \ + GLuint height = dPriv->h; \ + GLuint xo = dPriv->x; \ + GLuint yo = dPriv->y; \ + char *buf = (char *)(sPriv->pFB + r200Screen->depthOffset); \ + (void) buf + +#define LOCAL_STENCIL_VARS LOCAL_DEPTH_VARS + + +#define CLIPPIXEL( _x, _y ) \ + ((_x >= minx) && (_x < maxx) && (_y >= miny) && (_y < maxy)) + + +#define CLIPSPAN( _x, _y, _n, _x1, _n1, _i ) \ + if ( _y < miny || _y >= maxy ) { \ + _n1 = 0, _x1 = x; \ + } else { \ + _n1 = _n; \ + _x1 = _x; \ + if ( _x1 < minx ) _i += (minx-_x1), n1 -= (minx-_x1), _x1 = minx; \ + if ( _x1 + _n1 >= maxx ) n1 -= (_x1 + n1 - maxx); \ + } + +#define Y_FLIP( _y ) (height - _y - 1) + + +#define HW_LOCK() + +#define HW_CLIPLOOP() \ + do { \ + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; \ + int _nc = dPriv->numClipRects; \ + \ + while ( _nc-- ) { \ + int minx = dPriv->pClipRects[_nc].x1 - dPriv->x; \ + int miny = dPriv->pClipRects[_nc].y1 - dPriv->y; \ + int maxx = dPriv->pClipRects[_nc].x2 - dPriv->x; \ + int maxy = dPriv->pClipRects[_nc].y2 - dPriv->y; + +#define HW_ENDCLIPLOOP() \ + } \ + } while (0) + +#define HW_UNLOCK() + + + +/* ================================================================ + * Color buffer + */ + +/* 16 bit, RGB565 color spanline and pixel functions + */ +#define INIT_MONO_PIXEL(p, color) \ + p = PACK_COLOR_565( color[0], color[1], color[2] ) + +#define WRITE_RGBA( _x, _y, r, g, b, a ) \ + *(GLushort *)(buf + _x*2 + _y*pitch) = ((((int)r & 0xf8) << 8) | \ + (((int)g & 0xfc) << 3) | \ + (((int)b & 0xf8) >> 3)) + +#define WRITE_PIXEL( _x, _y, p ) \ + *(GLushort *)(buf + _x*2 + _y*pitch) = p + +#define READ_RGBA( rgba, _x, _y ) \ + do { \ + GLushort p = *(GLushort *)(read_buf + _x*2 + _y*pitch); \ + rgba[0] = ((p >> 8) & 0xf8) * 255 / 0xf8; \ + rgba[1] = ((p >> 3) & 0xfc) * 255 / 0xfc; \ + rgba[2] = ((p << 3) & 0xf8) * 255 / 0xf8; \ + rgba[3] = 0xff; \ + } while (0) + +#define TAG(x) r200##x##_RGB565 +#include "spantmp.h" + +/* 32 bit, ARGB8888 color spanline and pixel functions + */ +#undef INIT_MONO_PIXEL +#define INIT_MONO_PIXEL(p, color) \ + p = PACK_COLOR_8888( color[3], color[0], color[1], color[2] ) + +#define WRITE_RGBA( _x, _y, r, g, b, a ) \ +do { \ + *(GLuint *)(buf + _x*4 + _y*pitch) = ((b << 0) | \ + (g << 8) | \ + (r << 16) | \ + (a << 24) ); \ +} while (0) + +#define WRITE_PIXEL( _x, _y, p ) \ +do { \ + *(GLuint *)(buf + _x*4 + _y*pitch) = p; \ +} while (0) + +#define READ_RGBA( rgba, _x, _y ) \ +do { \ + volatile GLuint *ptr = (volatile GLuint *)(read_buf + _x*4 + _y*pitch); \ + GLuint p = *ptr; \ + rgba[0] = (p >> 16) & 0xff; \ + rgba[1] = (p >> 8) & 0xff; \ + rgba[2] = (p >> 0) & 0xff; \ + rgba[3] = (p >> 24) & 0xff; \ +} while (0) + +#define TAG(x) r200##x##_ARGB8888 +#include "spantmp.h" + + + +/* ================================================================ + * Depth buffer + */ + +/* The R200 has depth tiling on all the time, so we have to convert + * the x,y coordinates into the memory bus address (mba) in the same + * manner as the engine. In each case, the linear block address (ba) + * is calculated, and then wired with x and y to produce the final + * memory address. + */ + +#define BIT(x,b) ((x & (1<<b))>>b) +static GLuint r200_mba_z32( r200ContextPtr rmesa, + GLint x, GLint y ) +{ + GLuint pitch = rmesa->r200Screen->frontPitch; + GLuint b = ((y & 0x3FF) >> 4) * ((pitch & 0xFFF) >> 5) + ((x & 0x3FF) >> 5); + GLuint a = + (BIT(x,0) << 2) | + (BIT(y,0) << 3) | + (BIT(x,1) << 4) | + (BIT(y,1) << 5) | + (BIT(x,3) << 6) | + (BIT(x,4) << 7) | + (BIT(x,2) << 8) | + (BIT(y,2) << 9) | + (BIT(y,3) << 10) | + (((pitch & 0x20) ? (b & 0x01) : ((b & 0x01) ^ (BIT(y,4)))) << 11) | + ((b >> 1) << 12); + return a; +} + +static GLuint r200_mba_z16( r200ContextPtr rmesa, GLint x, GLint y ) +{ + GLuint pitch = rmesa->r200Screen->frontPitch; + GLuint b = ((y & 0x3FF) >> 4) * ((pitch & 0xFFF) >> 6) + ((x & 0x3FF) >> 6); + GLuint a = + (BIT(x,0) << 1) | + (BIT(y,0) << 2) | + (BIT(x,1) << 3) | + (BIT(y,1) << 4) | + (BIT(x,2) << 5) | + (BIT(x,4) << 6) | + (BIT(x,5) << 7) | + (BIT(x,3) << 8) | + (BIT(y,2) << 9) | + (BIT(y,3) << 10) | + (((pitch & 0x40) ? (b & 0x01) : ((b & 0x01) ^ (BIT(y,4)))) << 11) | + ((b >> 1) << 12); + return a; +} + + +/* 16-bit depth buffer functions + */ +#define WRITE_DEPTH( _x, _y, d ) \ + *(GLushort *)(buf + r200_mba_z16( rmesa, _x + xo, _y + yo )) = d; + +#define READ_DEPTH( d, _x, _y ) \ + d = *(GLushort *)(buf + r200_mba_z16( rmesa, _x + xo, _y + yo )); + +#define TAG(x) r200##x##_16 +#include "depthtmp.h" + +/* 24 bit depth, 8 bit stencil depthbuffer functions + */ +#define WRITE_DEPTH( _x, _y, d ) \ +do { \ + GLuint offset = r200_mba_z32( rmesa, _x + xo, _y + yo ); \ + GLuint tmp = *(GLuint *)(buf + offset); \ + tmp &= 0xff000000; \ + tmp |= ((d) & 0x00ffffff); \ + *(GLuint *)(buf + offset) = tmp; \ +} while (0) + +#define READ_DEPTH( d, _x, _y ) \ + d = *(GLuint *)(buf + r200_mba_z32( rmesa, _x + xo, \ + _y + yo )) & 0x00ffffff; + +#define TAG(x) r200##x##_24_8 +#include "depthtmp.h" + + +/* ================================================================ + * Stencil buffer + */ + +/* 24 bit depth, 8 bit stencil depthbuffer functions + */ +#define WRITE_STENCIL( _x, _y, d ) \ +do { \ + GLuint offset = r200_mba_z32( rmesa, _x + xo, _y + yo ); \ + GLuint tmp = *(GLuint *)(buf + offset); \ + tmp &= 0x00ffffff; \ + tmp |= (((d) & 0xff) << 24); \ + *(GLuint *)(buf + offset) = tmp; \ +} while (0) + +#define READ_STENCIL( d, _x, _y ) \ +do { \ + GLuint offset = r200_mba_z32( rmesa, _x + xo, _y + yo ); \ + GLuint tmp = *(GLuint *)(buf + offset); \ + tmp &= 0xff000000; \ + d = tmp >> 24; \ +} while (0) + +#define TAG(x) r200##x##_24_8 +#include "stenciltmp.h" + + +static void r200SetReadBuffer( GLcontext *ctx, + GLframebuffer *colorBuffer, + GLenum mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + switch ( mode ) { + case GL_FRONT_LEFT: + if ( rmesa->doPageFlip && rmesa->sarea->pfCurrentPage == 1 ) { + rmesa->state.pixel.readOffset = rmesa->r200Screen->backOffset; + rmesa->state.pixel.readPitch = rmesa->r200Screen->backPitch; + } else { + rmesa->state.pixel.readOffset = rmesa->r200Screen->frontOffset; + rmesa->state.pixel.readPitch = rmesa->r200Screen->frontPitch; + } + break; + case GL_BACK_LEFT: + if ( rmesa->doPageFlip && rmesa->sarea->pfCurrentPage == 1 ) { + rmesa->state.pixel.readOffset = rmesa->r200Screen->frontOffset; + rmesa->state.pixel.readPitch = rmesa->r200Screen->frontPitch; + } else { + rmesa->state.pixel.readOffset = rmesa->r200Screen->backOffset; + rmesa->state.pixel.readPitch = rmesa->r200Screen->backPitch; + } + break; + default: + assert(0); + break; + } +} + +/* Move locking out to get reasonable span performance (10x better + * than doing this in HW_LOCK above). WaitForIdle() is the main + * culprit. + */ + +static void r200SpanRenderStart( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + R200_FIREVERTICES( rmesa ); + LOCK_HARDWARE( rmesa ); + r200WaitForIdleLocked( rmesa ); + + /* Read & rewrite the first pixel in the frame buffer. This should + * be a noop, right? In fact without this conform fails as reading + * from the framebuffer sometimes produces old results -- the + * on-card read cache gets mixed up and doesn't notice that the + * framebuffer has been updated. + * + * In the worst case this is buggy too as p might get the wrong + * value first time, so really need a hidden pixel somewhere for this. + */ + { + int p; + volatile int *read_buf = (volatile int *)(rmesa->dri.screen->pFB + + rmesa->state.pixel.readOffset); + p = *read_buf; + *read_buf = p; + } +} + +static void r200SpanRenderFinish( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + _swrast_flush( ctx ); + UNLOCK_HARDWARE( rmesa ); +} + +void r200InitSpanFuncs( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + struct swrast_device_driver *swdd = _swrast_GetDeviceDriverReference(ctx); + + swdd->SetReadBuffer = r200SetReadBuffer; + + switch ( rmesa->r200Screen->cpp ) { + case 2: + swdd->WriteRGBASpan = r200WriteRGBASpan_RGB565; + swdd->WriteRGBSpan = r200WriteRGBSpan_RGB565; + swdd->WriteMonoRGBASpan = r200WriteMonoRGBASpan_RGB565; + swdd->WriteRGBAPixels = r200WriteRGBAPixels_RGB565; + swdd->WriteMonoRGBAPixels = r200WriteMonoRGBAPixels_RGB565; + swdd->ReadRGBASpan = r200ReadRGBASpan_RGB565; + swdd->ReadRGBAPixels = r200ReadRGBAPixels_RGB565; + break; + + case 4: + swdd->WriteRGBASpan = r200WriteRGBASpan_ARGB8888; + swdd->WriteRGBSpan = r200WriteRGBSpan_ARGB8888; + swdd->WriteMonoRGBASpan = r200WriteMonoRGBASpan_ARGB8888; + swdd->WriteRGBAPixels = r200WriteRGBAPixels_ARGB8888; + swdd->WriteMonoRGBAPixels = r200WriteMonoRGBAPixels_ARGB8888; + swdd->ReadRGBASpan = r200ReadRGBASpan_ARGB8888; + swdd->ReadRGBAPixels = r200ReadRGBAPixels_ARGB8888; + break; + + default: + break; + } + + switch ( rmesa->glCtx->Visual.depthBits ) { + case 16: + swdd->ReadDepthSpan = r200ReadDepthSpan_16; + swdd->WriteDepthSpan = r200WriteDepthSpan_16; + swdd->ReadDepthPixels = r200ReadDepthPixels_16; + swdd->WriteDepthPixels = r200WriteDepthPixels_16; + break; + + case 24: + swdd->ReadDepthSpan = r200ReadDepthSpan_24_8; + swdd->WriteDepthSpan = r200WriteDepthSpan_24_8; + swdd->ReadDepthPixels = r200ReadDepthPixels_24_8; + swdd->WriteDepthPixels = r200WriteDepthPixels_24_8; + + swdd->ReadStencilSpan = r200ReadStencilSpan_24_8; + swdd->WriteStencilSpan = r200WriteStencilSpan_24_8; + swdd->ReadStencilPixels = r200ReadStencilPixels_24_8; + swdd->WriteStencilPixels = r200WriteStencilPixels_24_8; + break; + + default: + break; + } + + swdd->SpanRenderStart = r200SpanRenderStart; + swdd->SpanRenderFinish = r200SpanRenderFinish; +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_span.h b/xc/lib/GL/mesa/src/drv/r200/r200_span.h new file mode 100644 index 000000000..4931d924f --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_span.h @@ -0,0 +1,43 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_SPAN_H__ +#define __R200_SPAN_H__ + +#ifdef GLX_DIRECT_RENDERING + +extern void r200InitSpanFuncs( GLcontext *ctx ); + +#endif +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_state.c b/xc/lib/GL/mesa/src/drv/r200/r200_state.c new file mode 100644 index 000000000..b425c5058 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_state.c @@ -0,0 +1,2132 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_ioctl.h" +#include "r200_state.h" +#include "r200_tcl.h" +#include "r200_tex.h" +#include "r200_swtcl.h" +#include "r200_vtxfmt.h" + +#include "mem.h" +#include "mmath.h" +#include "enums.h" +#include "colormac.h" +#include "light.h" +#include "api_arrayelt.h" + +#include "swrast/swrast.h" +#include "array_cache/acache.h" +#include "tnl/tnl.h" +#include "tnl/t_pipeline.h" +#include "swrast_setup/swrast_setup.h" + + +/* ============================================================= + * Alpha blending + */ + +static void r200AlphaFunc( GLcontext *ctx, GLenum func, GLchan ref ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + int pp_misc = rmesa->hw.ctx.cmd[CTX_PP_MISC]; + + R200_STATECHANGE( rmesa, ctx ); + + pp_misc &= ~(R200_ALPHA_TEST_OP_MASK | R200_REF_ALPHA_MASK); + pp_misc |= (ref & R200_REF_ALPHA_MASK); + + switch ( func ) { + case GL_NEVER: + pp_misc |= R200_ALPHA_TEST_FAIL; + break; + case GL_LESS: + pp_misc |= R200_ALPHA_TEST_LESS; + break; + case GL_EQUAL: + pp_misc |= R200_ALPHA_TEST_EQUAL; + break; + case GL_LEQUAL: + pp_misc |= R200_ALPHA_TEST_LEQUAL; + break; + case GL_GREATER: + pp_misc |= R200_ALPHA_TEST_GREATER; + break; + case GL_NOTEQUAL: + pp_misc |= R200_ALPHA_TEST_NEQUAL; + break; + case GL_GEQUAL: + pp_misc |= R200_ALPHA_TEST_GEQUAL; + break; + case GL_ALWAYS: + pp_misc |= R200_ALPHA_TEST_PASS; + break; + } + + rmesa->hw.ctx.cmd[CTX_PP_MISC] = pp_misc; +} + +static void r200BlendEquation( GLcontext *ctx, GLenum mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint b = rmesa->hw.ctx.cmd[CTX_RB3D_BLENDCNTL] & ~R200_COMB_FCN_MASK; + + switch ( mode ) { + case GL_FUNC_ADD_EXT: + case GL_LOGIC_OP: + b |= R200_COMB_FCN_ADD_CLAMP; + break; + + case GL_FUNC_SUBTRACT_EXT: + b |= R200_COMB_FCN_SUB_CLAMP; + break; + + case GL_FUNC_REVERSE_SUBTRACT_EXT: + b |= R200_COMB_FCN_RSUB_CLAMP; + break; + + case GL_MIN_EXT: + b |= R200_COMB_FCN_MIN; + break; + + case GL_MAX_EXT: + b |= R200_COMB_FCN_MAX; + break; + + default: + break; + } + + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_RB3D_BLENDCNTL] = b; + if ( ctx->Color.ColorLogicOpEnabled ) { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_ROP_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] &= ~R200_ROP_ENABLE; + } +} + +static void r200BlendFunc( GLcontext *ctx, GLenum sfactor, GLenum dfactor ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint b = rmesa->hw.ctx.cmd[CTX_RB3D_BLENDCNTL] & + ~(R200_SRC_BLEND_MASK | R200_DST_BLEND_MASK); + + switch ( ctx->Color.BlendSrcRGB ) { + case GL_ZERO: + b |= R200_SRC_BLEND_GL_ZERO; + break; + case GL_ONE: + b |= R200_SRC_BLEND_GL_ONE; + break; + case GL_DST_COLOR: + b |= R200_SRC_BLEND_GL_DST_COLOR; + break; + case GL_ONE_MINUS_DST_COLOR: + b |= R200_SRC_BLEND_GL_ONE_MINUS_DST_COLOR; + break; + case GL_SRC_ALPHA: + b |= R200_SRC_BLEND_GL_SRC_ALPHA; + break; + case GL_ONE_MINUS_SRC_ALPHA: + b |= R200_SRC_BLEND_GL_ONE_MINUS_SRC_ALPHA; + break; + case GL_DST_ALPHA: + b |= R200_SRC_BLEND_GL_DST_ALPHA; + break; + case GL_ONE_MINUS_DST_ALPHA: + b |= R200_SRC_BLEND_GL_ONE_MINUS_DST_ALPHA; + break; + case GL_SRC_ALPHA_SATURATE: + b |= R200_SRC_BLEND_GL_SRC_ALPHA_SATURATE; + break; + case GL_CONSTANT_COLOR: + b |= R200_SRC_BLEND_GL_CONST_COLOR; + break; + case GL_ONE_MINUS_CONSTANT_COLOR: + b |= R200_SRC_BLEND_GL_ONE_MINUS_CONST_COLOR; + break; + case GL_CONSTANT_ALPHA: + b |= R200_SRC_BLEND_GL_CONST_ALPHA; + break; + case GL_ONE_MINUS_CONSTANT_ALPHA: + b |= R200_SRC_BLEND_GL_ONE_MINUS_CONST_ALPHA; + break; + default: + break; + } + + switch ( ctx->Color.BlendDstRGB ) { + case GL_ZERO: + b |= R200_DST_BLEND_GL_ZERO; + break; + case GL_ONE: + b |= R200_DST_BLEND_GL_ONE; + break; + case GL_SRC_COLOR: + b |= R200_DST_BLEND_GL_SRC_COLOR; + break; + case GL_ONE_MINUS_SRC_COLOR: + b |= R200_DST_BLEND_GL_ONE_MINUS_SRC_COLOR; + break; + case GL_SRC_ALPHA: + b |= R200_DST_BLEND_GL_SRC_ALPHA; + break; + case GL_ONE_MINUS_SRC_ALPHA: + b |= R200_DST_BLEND_GL_ONE_MINUS_SRC_ALPHA; + break; + case GL_DST_ALPHA: + b |= R200_DST_BLEND_GL_DST_ALPHA; + break; + case GL_ONE_MINUS_DST_ALPHA: + b |= R200_DST_BLEND_GL_ONE_MINUS_DST_ALPHA; + break; + case GL_CONSTANT_COLOR: + b |= R200_DST_BLEND_GL_CONST_COLOR; + break; + case GL_ONE_MINUS_CONSTANT_COLOR: + b |= R200_DST_BLEND_GL_ONE_MINUS_CONST_COLOR; + break; + case GL_CONSTANT_ALPHA: + b |= R200_DST_BLEND_GL_CONST_ALPHA; + break; + case GL_ONE_MINUS_CONSTANT_ALPHA: + b |= R200_DST_BLEND_GL_ONE_MINUS_CONST_ALPHA; + break; + default: + break; + } + + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_RB3D_BLENDCNTL] = b; +} + +static void r200BlendFuncSeparate( GLcontext *ctx, + GLenum sfactorRGB, GLenum dfactorRGB, + GLenum sfactorA, GLenum dfactorA ) +{ + r200BlendFunc( ctx, sfactorRGB, dfactorRGB ); +} + + +/* ============================================================= + * Depth testing + */ + +static void r200DepthFunc( GLcontext *ctx, GLenum func ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] &= ~R200_Z_TEST_MASK; + + switch ( ctx->Depth.Func ) { + case GL_NEVER: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_TEST_NEVER; + break; + case GL_LESS: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_TEST_LESS; + break; + case GL_EQUAL: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_TEST_EQUAL; + break; + case GL_LEQUAL: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_TEST_LEQUAL; + break; + case GL_GREATER: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_TEST_GREATER; + break; + case GL_NOTEQUAL: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_TEST_NEQUAL; + break; + case GL_GEQUAL: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_TEST_GEQUAL; + break; + case GL_ALWAYS: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_TEST_ALWAYS; + break; + } +} + + +static void r200DepthMask( GLcontext *ctx, GLboolean flag ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + R200_STATECHANGE( rmesa, ctx ); + + if ( ctx->Depth.Mask ) { + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_WRITE_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] &= ~R200_Z_WRITE_ENABLE; + } +} + + +/* ============================================================= + * Fog + */ + + +static void r200Fogfv( GLcontext *ctx, GLenum pname, const GLfloat *param ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + union { int i; float f; } c, d; + GLchan col[4]; + GLuint i; + + c.i = rmesa->hw.fog.cmd[FOG_C]; + d.i = rmesa->hw.fog.cmd[FOG_D]; + + switch (pname) { + case GL_FOG_MODE: + if (!ctx->Fog.Enabled) + return; + R200_STATECHANGE(rmesa, tcl); + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] &= ~R200_TCL_FOG_MASK; + switch (ctx->Fog.Mode) { + case GL_LINEAR: + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] |= R200_TCL_FOG_LINEAR; + if (ctx->Fog.Start == ctx->Fog.End) { + c.f = 1.0F; + d.f = 1.0F; + } + else { + c.f = ctx->Fog.End/(ctx->Fog.End-ctx->Fog.Start); + d.f = -1.0/(ctx->Fog.End-ctx->Fog.Start); + } + break; + case GL_EXP: + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] |= R200_TCL_FOG_EXP; + c.f = 0.0; + d.f = -ctx->Fog.Density; + break; + case GL_EXP2: + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] |= R200_TCL_FOG_EXP2; + c.f = 0.0; + d.f = -(ctx->Fog.Density * ctx->Fog.Density); + break; + default: + return; + } + break; + case GL_FOG_DENSITY: + switch (ctx->Fog.Mode) { + case GL_EXP: + c.f = 0.0; + d.f = -ctx->Fog.Density; + break; + case GL_EXP2: + c.f = 0.0; + d.f = -(ctx->Fog.Density * ctx->Fog.Density); + break; + default: + break; + } + break; + case GL_FOG_START: + case GL_FOG_END: + if (ctx->Fog.Mode == GL_LINEAR) { + if (ctx->Fog.Start == ctx->Fog.End) { + c.f = 1.0F; + d.f = 1.0F; + } else { + c.f = ctx->Fog.End/(ctx->Fog.End-ctx->Fog.Start); + d.f = -1.0/(ctx->Fog.End-ctx->Fog.Start); + } + } + break; + case GL_FOG_COLOR: + R200_STATECHANGE( rmesa, ctx ); + UNCLAMPED_FLOAT_TO_RGB_CHAN( col, ctx->Fog.Color ); + i = r200PackColor( 4, col[0], col[1], col[2], 0 ); + rmesa->hw.ctx.cmd[CTX_PP_FOG_COLOR] &= ~R200_FOG_COLOR_MASK; + rmesa->hw.ctx.cmd[CTX_PP_FOG_COLOR] |= i; + break; + case GL_FOG_COORDINATE_SOURCE_EXT: + /* What to do? + */ + break; + default: + return; + } + + if (c.i != rmesa->hw.fog.cmd[FOG_C] || d.i != rmesa->hw.fog.cmd[FOG_D]) { + R200_STATECHANGE( rmesa, fog ); + rmesa->hw.fog.cmd[FOG_C] = c.i; + rmesa->hw.fog.cmd[FOG_D] = d.i; + } +} + + +/* ============================================================= + * Scissoring + */ + + +static GLboolean intersect_rect( XF86DRIClipRectPtr out, + XF86DRIClipRectPtr a, + XF86DRIClipRectPtr b ) +{ + *out = *a; + if ( b->x1 > out->x1 ) out->x1 = b->x1; + if ( b->y1 > out->y1 ) out->y1 = b->y1; + if ( b->x2 < out->x2 ) out->x2 = b->x2; + if ( b->y2 < out->y2 ) out->y2 = b->y2; + if ( out->x1 >= out->x2 ) return GL_FALSE; + if ( out->y1 >= out->y2 ) return GL_FALSE; + return GL_TRUE; +} + + +void r200RecalcScissorRects( r200ContextPtr rmesa ) +{ + XF86DRIClipRectPtr out; + int i; + + /* Grow cliprect store? + */ + if (rmesa->state.scissor.numAllocedClipRects < rmesa->numClipRects) { + while (rmesa->state.scissor.numAllocedClipRects < rmesa->numClipRects) { + rmesa->state.scissor.numAllocedClipRects += 1; /* zero case */ + rmesa->state.scissor.numAllocedClipRects *= 2; + } + + if (rmesa->state.scissor.pClipRects) + FREE(rmesa->state.scissor.pClipRects); + + rmesa->state.scissor.pClipRects = + MALLOC( rmesa->state.scissor.numAllocedClipRects * + sizeof(XF86DRIClipRectRec) ); + + if (!rmesa->state.scissor.numAllocedClipRects) { + rmesa->state.scissor.numAllocedClipRects = 0; + return; + } + } + + out = rmesa->state.scissor.pClipRects; + rmesa->state.scissor.numClipRects = 0; + + for ( i = 0 ; i < rmesa->numClipRects ; i++ ) { + if ( intersect_rect( out, + &rmesa->pClipRects[i], + &rmesa->state.scissor.rect ) ) { + rmesa->state.scissor.numClipRects++; + out++; + } + } +} + + +static void r200UpdateScissor( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if ( rmesa->dri.drawable ) { + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; + + int x = ctx->Scissor.X; + int y = dPriv->h - ctx->Scissor.Y - ctx->Scissor.Height; + int w = ctx->Scissor.X + ctx->Scissor.Width - 1; + int h = dPriv->h - ctx->Scissor.Y - 1; + + rmesa->state.scissor.rect.x1 = x + dPriv->x; + rmesa->state.scissor.rect.y1 = y + dPriv->y; + rmesa->state.scissor.rect.x2 = w + dPriv->x + 1; + rmesa->state.scissor.rect.y2 = h + dPriv->y + 1; + + r200RecalcScissorRects( rmesa ); + } +} + + +static void r200Scissor( GLcontext *ctx, + GLint x, GLint y, GLsizei w, GLsizei h ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if ( ctx->Scissor.Enabled ) { + R200_FIREVERTICES( rmesa ); /* don't pipeline cliprect changes */ + r200UpdateScissor( ctx ); + } + +} + + +/* ============================================================= + * Culling + */ + +static void r200CullFace( GLcontext *ctx, GLenum unused ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint s = rmesa->hw.set.cmd[SET_SE_CNTL]; + GLuint t = rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL]; + + s |= R200_FFACE_SOLID | R200_BFACE_SOLID; + t &= ~(R200_CULL_FRONT | R200_CULL_BACK); + + if ( ctx->Polygon.CullFlag ) { + switch ( ctx->Polygon.CullFaceMode ) { + case GL_FRONT: + s &= ~R200_FFACE_SOLID; + t |= R200_CULL_FRONT; + break; + case GL_BACK: + s &= ~R200_BFACE_SOLID; + t |= R200_CULL_BACK; + break; + case GL_FRONT_AND_BACK: + s &= ~(R200_FFACE_SOLID | R200_BFACE_SOLID); + t |= (R200_CULL_FRONT | R200_CULL_BACK); + break; + } + } + + if ( rmesa->hw.set.cmd[SET_SE_CNTL] != s ) { + R200_STATECHANGE(rmesa, set ); + rmesa->hw.set.cmd[SET_SE_CNTL] = s; + } + + if ( rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] != t ) { + R200_STATECHANGE(rmesa, tcl ); + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] = t; + } +} + +static void r200FrontFace( GLcontext *ctx, GLenum mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + R200_STATECHANGE( rmesa, set ); + rmesa->hw.set.cmd[SET_SE_CNTL] &= ~R200_FFACE_CULL_DIR_MASK; + + R200_STATECHANGE( rmesa, tcl ); + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] &= ~R200_CULL_FRONT_IS_CCW; + + switch ( mode ) { + case GL_CW: + rmesa->hw.set.cmd[SET_SE_CNTL] |= R200_FFACE_CULL_CW; + break; + case GL_CCW: + rmesa->hw.set.cmd[SET_SE_CNTL] |= R200_FFACE_CULL_CCW; + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] |= R200_CULL_FRONT_IS_CCW; + break; + } +} + +/* ============================================================= + * Point state + */ +static void r200PointSize( GLcontext *ctx, GLfloat size ) +{ + fprintf(stderr, "%s: %f\n", __FUNCTION__, size ); +} + +/* ============================================================= + * Line state + */ +static void r200LineWidth( GLcontext *ctx, GLfloat widthf ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + R200_STATECHANGE( rmesa, lin ); + R200_STATECHANGE( rmesa, set ); + + /* Line width is stored in U6.4 format. + */ + rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] &= ~0xffff; + rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] |= (GLuint)(ctx->Line._Width * 16.0); + + if ( widthf > 1.0 ) { + rmesa->hw.set.cmd[SET_SE_CNTL] |= R200_WIDELINE_ENABLE; + } else { + rmesa->hw.set.cmd[SET_SE_CNTL] &= ~R200_WIDELINE_ENABLE; + } +} + +static void r200LineStipple( GLcontext *ctx, GLint factor, GLushort pattern ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + R200_STATECHANGE( rmesa, lin ); + rmesa->hw.lin.cmd[LIN_RE_LINE_PATTERN] = + ((((GLuint)factor & 0xff) << 16) | ((GLuint)pattern)); +} + + +/* ============================================================= + * Masks + */ +static void r200ColorMask( GLcontext *ctx, + GLboolean r, GLboolean g, + GLboolean b, GLboolean a ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint mask = r200PackColor( rmesa->r200Screen->cpp, + ctx->Color.ColorMask[RCOMP], + ctx->Color.ColorMask[GCOMP], + ctx->Color.ColorMask[BCOMP], + ctx->Color.ColorMask[ACOMP] ); + + GLuint flag = rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] & ~R200_PLANE_MASK_ENABLE; + + if (!(r && g && b && a)) + flag |= R200_PLANE_MASK_ENABLE; + + if ( rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] != flag ) { + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] = flag; + } + + if ( rmesa->hw.msk.cmd[MSK_RB3D_PLANEMASK] != mask ) { + R200_STATECHANGE( rmesa, msk ); + rmesa->hw.msk.cmd[MSK_RB3D_PLANEMASK] = mask; + } +} + + +/* ============================================================= + * Polygon state + */ + +static void r200PolygonOffset( GLcontext *ctx, + GLfloat factor, GLfloat units ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLfloat constant = units * rmesa->state.depth.scale; + +/* factor *= 2; */ +/* constant *= 2; */ + +/* fprintf(stderr, "%s f:%f u:%f\n", __FUNCTION__, factor, constant); */ + + R200_STATECHANGE( rmesa, zbs ); + rmesa->hw.zbs.cmd[ZBS_SE_ZBIAS_FACTOR] = *(GLuint *)&factor; + rmesa->hw.zbs.cmd[ZBS_SE_ZBIAS_CONSTANT] = *(GLuint *)&constant; +} + +static void r200PolygonStipple( GLcontext *ctx, const GLubyte *mask ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint i; + drmRadeonStipple stipple; + + /* Must flip pattern upside down. + */ + for ( i = 0 ; i < 32 ; i++ ) { + rmesa->state.stipple.mask[31 - i] = ((GLuint *) mask)[i]; + } + + /* TODO: push this into cmd mechanism + */ + R200_FIREVERTICES( rmesa ); + LOCK_HARDWARE( rmesa ); + + /* FIXME: Use window x,y offsets into stipple RAM. + */ + stipple.mask = rmesa->state.stipple.mask; + drmCommandWrite( rmesa->dri.fd, DRM_RADEON_STIPPLE, + &stipple, sizeof(drmRadeonStipple) ); + UNLOCK_HARDWARE( rmesa ); +} + +static void r200PolygonMode( GLcontext *ctx, GLenum face, GLenum mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLboolean flag = (ctx->_TriangleCaps & DD_TRI_UNFILLED) != 0; + + /* Can't generally do unfilled via tcl, but some good special + * cases work. + */ + TCL_FALLBACK( ctx, R200_TCL_FALLBACK_UNFILLED, flag); + if (rmesa->TclFallback) { + r200ChooseRenderState( ctx ); + r200ChooseVertexState( ctx ); + } +} + + +/* ============================================================= + * Rendering attributes + * + * We really don't want to recalculate all this every time we bind a + * texture. These things shouldn't change all that often, so it makes + * sense to break them out of the core texture state update routines. + */ + +/* Examine lighting and texture state to determine if separate specular + * should be enabled. + */ +static void r200UpdateSpecular( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + CARD32 p = rmesa->hw.ctx.cmd[CTX_PP_CNTL]; + + R200_STATECHANGE( rmesa, tcl ); + R200_STATECHANGE( rmesa, vtx ); + + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] &= ~(3<<R200_VTX_COLOR_0_SHIFT); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] &= ~(3<<R200_VTX_COLOR_1_SHIFT); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] &= ~R200_OUTPUT_COLOR_0; + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] &= ~R200_OUTPUT_COLOR_1; + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] &= ~R200_LIGHTING_ENABLE; + p &= ~R200_SPECULAR_ENABLE; + + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] |= R200_DIFFUSE_SPECULAR_COMBINE; + + + if (ctx->Light.Enabled && + ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR) { + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] |= + ((R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT) | + (R200_VTX_FP_RGBA << R200_VTX_COLOR_1_SHIFT)); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] |= R200_OUTPUT_COLOR_0; + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] |= R200_OUTPUT_COLOR_1; + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] |= R200_LIGHTING_ENABLE; + p |= R200_SPECULAR_ENABLE; + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] &= + ~R200_DIFFUSE_SPECULAR_COMBINE; + } + else if (ctx->Light.Enabled) { + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] |= + ((R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT)); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] |= R200_OUTPUT_COLOR_0; + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] |= R200_LIGHTING_ENABLE; + } else if (ctx->Fog.ColorSumEnabled ) { + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] |= + ((R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT) | + (R200_VTX_FP_RGBA << R200_VTX_COLOR_1_SHIFT)); + p |= R200_SPECULAR_ENABLE; + } else { + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] |= + ((R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT)); + } + + if (ctx->Fog.Enabled) { + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] |= + ((R200_VTX_FP_RGBA << R200_VTX_COLOR_1_SHIFT)); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] |= R200_OUTPUT_COLOR_1; + } + + if ( rmesa->hw.ctx.cmd[CTX_PP_CNTL] != p ) { + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_PP_CNTL] = p; + } + + /* Update vertex/render formats + */ + if (rmesa->TclFallback) { + r200ChooseRenderState( ctx ); + r200ChooseVertexState( ctx ); + } +} + + +/* ============================================================= + * Materials + */ + + +/* Update on colormaterial, material emmissive/ambient, + * lightmodel.globalambient + */ +static void update_global_ambient( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + float *fcmd = (float *)R200_DB_STATE( glt ); + + /* Need to do more if both emmissive & ambient are PREMULT: + */ + if ((rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_1] & + ((3 << R200_FRONT_EMISSIVE_SOURCE_SHIFT) | + (3 << R200_FRONT_AMBIENT_SOURCE_SHIFT))) == 0) + { + COPY_3V( &fcmd[GLT_RED], + ctx->Light.Material[0].Emission); + ACC_SCALE_3V( &fcmd[GLT_RED], + ctx->Light.Model.Ambient, + ctx->Light.Material[0].Ambient); + } + else + { + COPY_3V( &fcmd[GLT_RED], ctx->Light.Model.Ambient ); + } + + R200_DB_STATECHANGE(rmesa, &rmesa->hw.glt); +} + +/* Update on change to + * - light[p].colors + * - light[p].enabled + * - material, + * - colormaterial enabled + * - colormaterial bitmask + */ +static void update_light_colors( GLcontext *ctx, GLuint p ) +{ + struct gl_light *l = &ctx->Light.Light[p]; + +/* fprintf(stderr, "%s\n", __FUNCTION__); */ + + if (l->Enabled) { + r200ContextPtr rmesa = R200_CONTEXT(ctx); + float *fcmd = (float *)R200_DB_STATE( lit[p] ); + GLuint bitmask = ctx->Light.ColorMaterialBitmask; + struct gl_material *mat = &ctx->Light.Material[0]; + + COPY_4V( &fcmd[LIT_AMBIENT_RED], l->Ambient ); + COPY_4V( &fcmd[LIT_DIFFUSE_RED], l->Diffuse ); + COPY_4V( &fcmd[LIT_SPECULAR_RED], l->Specular ); + + if (!ctx->Light.ColorMaterialEnabled) + bitmask = 0; + + if ((bitmask & FRONT_AMBIENT_BIT) == 0) + SELF_SCALE_3V( &fcmd[LIT_AMBIENT_RED], mat->Ambient ); + + if ((bitmask & FRONT_DIFFUSE_BIT) == 0) + SELF_SCALE_3V( &fcmd[LIT_DIFFUSE_RED], mat->Diffuse ); + + if ((bitmask & FRONT_SPECULAR_BIT) == 0) + SELF_SCALE_3V( &fcmd[LIT_SPECULAR_RED], mat->Specular ); + + R200_DB_STATECHANGE( rmesa, &rmesa->hw.lit[p] ); + } +} + +/* Also fallback for asym colormaterial mode in twoside lighting... + */ +static void check_twoside_fallback( GLcontext *ctx ) +{ + GLboolean fallback = GL_FALSE; + + if (ctx->Light.Enabled && ctx->Light.Model.TwoSide) { + if (memcmp( &ctx->Light.Material[0], + &ctx->Light.Material[1], + sizeof(struct gl_material)) != 0) + fallback = GL_TRUE; + else if (ctx->Light.ColorMaterialEnabled && + (ctx->Light.ColorMaterialBitmask & BACK_MATERIAL_BITS) != + ((ctx->Light.ColorMaterialBitmask & FRONT_MATERIAL_BITS)<<1)) + fallback = GL_TRUE; + } + + TCL_FALLBACK( ctx, R200_TCL_FALLBACK_LIGHT_TWOSIDE, fallback ); +} + +static void r200ColorMaterial( GLcontext *ctx, GLenum face, GLenum mode ) +{ + if (ctx->Light.ColorMaterialEnabled) { + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint light_model_ctl1 = rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_1]; + GLuint mask = ctx->Light.ColorMaterialBitmask; + + /* Default to PREMULT: + */ + light_model_ctl1 &= ~((0xf << R200_FRONT_EMISSIVE_SOURCE_SHIFT) | + (0xf << R200_FRONT_AMBIENT_SOURCE_SHIFT) | + (0xf << R200_FRONT_DIFFUSE_SOURCE_SHIFT) | + (0xf << R200_FRONT_SPECULAR_SOURCE_SHIFT)); + + if (mask & FRONT_EMISSION_BIT) { + light_model_ctl1 |= (R200_LM1_SOURCE_VERTEX_COLOR_0 << + R200_FRONT_EMISSIVE_SOURCE_SHIFT); + } + + if (mask & FRONT_AMBIENT_BIT) { + light_model_ctl1 |= (R200_LM1_SOURCE_VERTEX_COLOR_0 << + R200_FRONT_AMBIENT_SOURCE_SHIFT); + } + + if (mask & FRONT_DIFFUSE_BIT) { + light_model_ctl1 |= (R200_LM1_SOURCE_VERTEX_COLOR_0 << + R200_FRONT_DIFFUSE_SOURCE_SHIFT); + } + + if (mask & FRONT_SPECULAR_BIT) { + light_model_ctl1 |= (R200_LM1_SOURCE_VERTEX_COLOR_0 << + R200_FRONT_SPECULAR_SOURCE_SHIFT); + } + + if (light_model_ctl1 != rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_1]) { + GLuint p; + + R200_STATECHANGE( rmesa, tcl ); + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_1] = light_model_ctl1; + + for (p = 0 ; p < MAX_LIGHTS; p++) + update_light_colors( ctx, p ); + } + } + + check_twoside_fallback( ctx ); +} + +void r200UpdateMaterial( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLfloat *fcmd = (GLfloat *)R200_DB_STATE( mtl[0] ); + GLuint p; + GLuint mask = ~0; + + if (ctx->Light.ColorMaterialEnabled) + mask &= ~ctx->Light.ColorMaterialBitmask; + + if (R200_DEBUG & DEBUG_STATE) + fprintf(stderr, "%s\n", __FUNCTION__); + + + if (mask & FRONT_EMISSION_BIT) { + fcmd[MTL_EMMISSIVE_RED] = ctx->Light.Material[0].Emission[0]; + fcmd[MTL_EMMISSIVE_GREEN] = ctx->Light.Material[0].Emission[1]; + fcmd[MTL_EMMISSIVE_BLUE] = ctx->Light.Material[0].Emission[2]; + fcmd[MTL_EMMISSIVE_ALPHA] = ctx->Light.Material[0].Emission[3]; + } + if (mask & FRONT_AMBIENT_BIT) { + fcmd[MTL_AMBIENT_RED] = ctx->Light.Material[0].Ambient[0]; + fcmd[MTL_AMBIENT_GREEN] = ctx->Light.Material[0].Ambient[1]; + fcmd[MTL_AMBIENT_BLUE] = ctx->Light.Material[0].Ambient[2]; + fcmd[MTL_AMBIENT_ALPHA] = ctx->Light.Material[0].Ambient[3]; + } + if (mask & FRONT_DIFFUSE_BIT) { + fcmd[MTL_DIFFUSE_RED] = ctx->Light.Material[0].Diffuse[0]; + fcmd[MTL_DIFFUSE_GREEN] = ctx->Light.Material[0].Diffuse[1]; + fcmd[MTL_DIFFUSE_BLUE] = ctx->Light.Material[0].Diffuse[2]; + fcmd[MTL_DIFFUSE_ALPHA] = ctx->Light.Material[0].Diffuse[3]; + } + if (mask & FRONT_SPECULAR_BIT) { + fcmd[MTL_SPECULAR_RED] = ctx->Light.Material[0].Specular[0]; + fcmd[MTL_SPECULAR_GREEN] = ctx->Light.Material[0].Specular[1]; + fcmd[MTL_SPECULAR_BLUE] = ctx->Light.Material[0].Specular[2]; + fcmd[MTL_SPECULAR_ALPHA] = ctx->Light.Material[0].Specular[3]; + } + if (mask & FRONT_SHININESS_BIT) { + fcmd[MTL_SHININESS] = ctx->Light.Material[0].Shininess; + } + + if (R200_DB_STATECHANGE( rmesa, &rmesa->hw.mtl[0] )) { + for (p = 0 ; p < MAX_LIGHTS; p++) + update_light_colors( ctx, p ); + + check_twoside_fallback( ctx ); + update_global_ambient( ctx ); + } + else if (R200_DEBUG & (DEBUG_PRIMS|DEBUG_STATE)) + fprintf(stderr, "%s: Elided noop material call\n", __FUNCTION__); +} + +/* _NEW_LIGHT + * _NEW_MODELVIEW + * _MESA_NEW_NEED_EYE_COORDS + * + * Uses derived state from mesa: + * _VP_inf_norm + * _h_inf_norm + * _Position + * _NormDirection + * _ModelViewInvScale + * _NeedEyeCoords + * _EyeZDir + * + * which are calculated in light.c and are correct for the current + * lighting space (model or eye), hence dependencies on _NEW_MODELVIEW + * and _MESA_NEW_NEED_EYE_COORDS. + */ +static void update_light( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + /* Have to check these, or have an automatic shortcircuit mechanism + * to remove noop statechanges. (Or just do a better job on the + * front end). + */ + { + GLuint tmp = rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0]; + + if (ctx->_NeedEyeCoords) + tmp &= ~R200_LIGHT_IN_MODELSPACE; + else + tmp |= R200_LIGHT_IN_MODELSPACE; + + if (tmp != rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0]) + { + R200_STATECHANGE( rmesa, tcl ); + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] = tmp; + } + } + + { + GLfloat *fcmd = (GLfloat *)R200_DB_STATE( eye ); + fcmd[EYE_X] = ctx->_EyeZDir[0]; + fcmd[EYE_Y] = ctx->_EyeZDir[1]; + fcmd[EYE_Z] = - ctx->_EyeZDir[2]; + fcmd[EYE_RESCALE_FACTOR] = ctx->_ModelViewInvScale; + R200_DB_STATECHANGE( rmesa, &rmesa->hw.eye ); + } + + +/* R200_STATECHANGE( rmesa, glt ); */ + + if (ctx->Light.Enabled) { + GLint p; + for (p = 0 ; p < MAX_LIGHTS; p++) { + if (ctx->Light.Light[p].Enabled) { + struct gl_light *l = &ctx->Light.Light[p]; + GLfloat *fcmd = (GLfloat *)R200_DB_STATE( lit[p] ); + + if (l->EyePosition[3] == 0.0) { + COPY_3FV( &fcmd[LIT_POSITION_X], l->_VP_inf_norm ); + COPY_3FV( &fcmd[LIT_DIRECTION_X], l->_h_inf_norm ); + fcmd[LIT_POSITION_W] = 0; + fcmd[LIT_DIRECTION_W] = 0; + } else { + COPY_4V( &fcmd[LIT_POSITION_X], l->_Position ); + fcmd[LIT_DIRECTION_X] = -l->_NormDirection[0]; + fcmd[LIT_DIRECTION_Y] = -l->_NormDirection[1]; + fcmd[LIT_DIRECTION_Z] = -l->_NormDirection[2]; + fcmd[LIT_DIRECTION_W] = 0; + } + + R200_DB_STATECHANGE( rmesa, &rmesa->hw.lit[p] ); + } + } + } +} + +static void r200Lightfv( GLcontext *ctx, GLenum light, + GLenum pname, const GLfloat *params ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLint p = light - GL_LIGHT0; + struct gl_light *l = &ctx->Light.Light[p]; + GLfloat *fcmd = (GLfloat *)rmesa->hw.lit[p].cmd; + + + switch (pname) { + case GL_AMBIENT: + case GL_DIFFUSE: + case GL_SPECULAR: + update_light_colors( ctx, p ); + break; + + case GL_SPOT_DIRECTION: + /* picked up in update_light */ + break; + + case GL_POSITION: { + /* positions picked up in update_light, but can do flag here */ + GLuint flag = (p&1)? R200_LIGHT_1_IS_LOCAL : R200_LIGHT_0_IS_LOCAL; + GLuint idx = TCL_PER_LIGHT_CTL_0 + p/2; + + R200_STATECHANGE(rmesa, tcl); + if (l->EyePosition[3] != 0.0F) + rmesa->hw.tcl.cmd[idx] |= flag; + else + rmesa->hw.tcl.cmd[idx] &= ~flag; + break; + } + + case GL_SPOT_EXPONENT: + R200_STATECHANGE(rmesa, lit[p]); + fcmd[LIT_SPOT_EXPONENT] = params[0]; + break; + + case GL_SPOT_CUTOFF: { + GLuint flag = (p&1) ? R200_LIGHT_1_IS_SPOT : R200_LIGHT_0_IS_SPOT; + GLuint idx = TCL_PER_LIGHT_CTL_0 + p/2; + + R200_STATECHANGE(rmesa, lit[p]); + fcmd[LIT_SPOT_CUTOFF] = l->_CosCutoff; + + R200_STATECHANGE(rmesa, tcl); + if (l->SpotCutoff != 180.0F) + rmesa->hw.tcl.cmd[idx] |= flag; + else + rmesa->hw.tcl.cmd[idx] &= ~flag; + + break; + } + + case GL_CONSTANT_ATTENUATION: + R200_STATECHANGE(rmesa, lit[p]); + fcmd[LIT_ATTEN_CONST] = params[0]; + break; + case GL_LINEAR_ATTENUATION: + R200_STATECHANGE(rmesa, lit[p]); + fcmd[LIT_ATTEN_LINEAR] = params[0]; + break; + case GL_QUADRATIC_ATTENUATION: + R200_STATECHANGE(rmesa, lit[p]); + fcmd[LIT_ATTEN_QUADRATIC] = params[0]; + break; + default: + return; + } + +} + + + + +static void r200LightModelfv( GLcontext *ctx, GLenum pname, + const GLfloat *param ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + switch (pname) { + case GL_LIGHT_MODEL_AMBIENT: + update_global_ambient( ctx ); + break; + + case GL_LIGHT_MODEL_LOCAL_VIEWER: + R200_STATECHANGE( rmesa, tcl ); + if (ctx->Light.Model.LocalViewer) + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] |= R200_LOCAL_VIEWER; + else + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] &= ~R200_LOCAL_VIEWER; + break; + + case GL_LIGHT_MODEL_TWO_SIDE: + R200_STATECHANGE( rmesa, tcl ); + if (ctx->Light.Model.TwoSide) + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] |= R200_LIGHT_TWOSIDE; + else + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] &= ~R200_LIGHT_TWOSIDE; + + check_twoside_fallback( ctx ); + + if (rmesa->TclFallback) { + r200ChooseRenderState( ctx ); + r200ChooseVertexState( ctx ); + } + break; + + case GL_LIGHT_MODEL_COLOR_CONTROL: + r200UpdateSpecular(ctx); + break; + + default: + break; + } +} + +static void r200ShadeModel( GLcontext *ctx, GLenum mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint s = rmesa->hw.set.cmd[SET_SE_CNTL]; + + s &= ~(R200_DIFFUSE_SHADE_MASK | + R200_ALPHA_SHADE_MASK | + R200_SPECULAR_SHADE_MASK | + R200_FOG_SHADE_MASK); + + switch ( mode ) { + case GL_FLAT: + s |= (R200_DIFFUSE_SHADE_FLAT | + R200_ALPHA_SHADE_FLAT | + R200_SPECULAR_SHADE_FLAT | + R200_FOG_SHADE_FLAT); + break; + case GL_SMOOTH: + s |= (R200_DIFFUSE_SHADE_GOURAUD | + R200_ALPHA_SHADE_GOURAUD | + R200_SPECULAR_SHADE_GOURAUD | + R200_FOG_SHADE_GOURAUD); + break; + default: + return; + } + + if ( rmesa->hw.set.cmd[SET_SE_CNTL] != s ) { + R200_STATECHANGE( rmesa, set ); + rmesa->hw.set.cmd[SET_SE_CNTL] = s; + } +} + + +/* ============================================================= + * User clip planes + */ + +static void r200ClipPlane( GLcontext *ctx, GLenum plane, const GLfloat *eq ) +{ + GLint p = (GLint) plane - (GLint) GL_CLIP_PLANE0; + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLint *ip = (GLint *)ctx->Transform._ClipUserPlane[p]; + + R200_STATECHANGE( rmesa, ucp[p] ); + rmesa->hw.ucp[p].cmd[UCP_X] = ip[0]; + rmesa->hw.ucp[p].cmd[UCP_Y] = ip[1]; + rmesa->hw.ucp[p].cmd[UCP_Z] = ip[2]; + rmesa->hw.ucp[p].cmd[UCP_W] = ip[3]; +} + +static void r200UpdateClipPlanes( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint p; + + for (p = 0; p < ctx->Const.MaxClipPlanes; p++) { + if (ctx->Transform.ClipEnabled[p]) { + GLint *ip = (GLint *)ctx->Transform._ClipUserPlane[p]; + + R200_STATECHANGE( rmesa, ucp[p] ); + rmesa->hw.ucp[p].cmd[UCP_X] = ip[0]; + rmesa->hw.ucp[p].cmd[UCP_Y] = ip[1]; + rmesa->hw.ucp[p].cmd[UCP_Z] = ip[2]; + rmesa->hw.ucp[p].cmd[UCP_W] = ip[3]; + } + } +} + + +/* ============================================================= + * Stencil + */ + +static void r200StencilFunc( GLcontext *ctx, GLenum func, + GLint ref, GLuint mask ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint refmask = ((ctx->Stencil.Ref << R200_STENCIL_REF_SHIFT) | + (ctx->Stencil.ValueMask << R200_STENCIL_MASK_SHIFT)); + + R200_STATECHANGE( rmesa, ctx ); + R200_STATECHANGE( rmesa, msk ); + + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] &= ~R200_STENCIL_TEST_MASK; + rmesa->hw.msk.cmd[MSK_RB3D_STENCILREFMASK] &= ~(R200_STENCIL_REF_MASK| + R200_STENCIL_VALUE_MASK); + + switch ( ctx->Stencil.Function ) { + case GL_NEVER: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_TEST_NEVER; + break; + case GL_LESS: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_TEST_LESS; + break; + case GL_EQUAL: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_TEST_EQUAL; + break; + case GL_LEQUAL: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_TEST_LEQUAL; + break; + case GL_GREATER: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_TEST_GREATER; + break; + case GL_NOTEQUAL: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_TEST_NEQUAL; + break; + case GL_GEQUAL: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_TEST_GEQUAL; + break; + case GL_ALWAYS: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_TEST_ALWAYS; + break; + } + + rmesa->hw.msk.cmd[MSK_RB3D_STENCILREFMASK] |= refmask; +} + +static void r200StencilMask( GLcontext *ctx, GLuint mask ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + R200_STATECHANGE( rmesa, msk ); + rmesa->hw.msk.cmd[MSK_RB3D_STENCILREFMASK] &= ~R200_STENCIL_WRITE_MASK; + rmesa->hw.msk.cmd[MSK_RB3D_STENCILREFMASK] |= + (ctx->Stencil.WriteMask << R200_STENCIL_WRITEMASK_SHIFT); +} + +static void r200StencilOp( GLcontext *ctx, GLenum fail, + GLenum zfail, GLenum zpass ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] &= ~(R200_STENCIL_FAIL_MASK | + R200_STENCIL_ZFAIL_MASK | + R200_STENCIL_ZPASS_MASK); + + switch ( ctx->Stencil.FailFunc ) { + case GL_KEEP: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_FAIL_KEEP; + break; + case GL_ZERO: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_FAIL_ZERO; + break; + case GL_REPLACE: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_FAIL_REPLACE; + break; + case GL_INCR: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_FAIL_INC; + break; + case GL_DECR: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_FAIL_DEC; + break; + case GL_INVERT: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_FAIL_INVERT; + break; + } + + switch ( ctx->Stencil.ZFailFunc ) { + case GL_KEEP: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZFAIL_KEEP; + break; + case GL_ZERO: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZFAIL_ZERO; + break; + case GL_REPLACE: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZFAIL_REPLACE; + break; + case GL_INCR: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZFAIL_INC; + break; + case GL_DECR: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZFAIL_DEC; + break; + case GL_INVERT: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZFAIL_INVERT; + break; + } + + switch ( ctx->Stencil.ZPassFunc ) { + case GL_KEEP: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZPASS_KEEP; + break; + case GL_ZERO: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZPASS_ZERO; + break; + case GL_REPLACE: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZPASS_REPLACE; + break; + case GL_INCR: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZPASS_INC; + break; + case GL_DECR: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZPASS_DEC; + break; + case GL_INVERT: + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_STENCIL_ZPASS_INVERT; + break; + } +} + +static void r200ClearStencil( GLcontext *ctx, GLint s ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + rmesa->state.stencil.clear = + ((GLuint) ctx->Stencil.Clear | + (0xff << R200_STENCIL_MASK_SHIFT) | + (ctx->Stencil.WriteMask << R200_STENCIL_WRITEMASK_SHIFT)); +} + + +/* ============================================================= + * Window position and viewport transformation + */ + +/* + * To correctly position primitives: + */ +#define SUBPIXEL_X 0.125 + +void r200UpdateWindow( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; + GLfloat xoffset = (GLfloat)dPriv->x; + GLfloat yoffset = (GLfloat)dPriv->y + dPriv->h; + const GLfloat *v = ctx->Viewport._WindowMap.m; + + GLfloat sx = v[MAT_SX]; + GLfloat tx = v[MAT_TX] + xoffset + SUBPIXEL_X; + GLfloat sy = - v[MAT_SY]; + GLfloat ty = (- v[MAT_TY]) + yoffset; + GLfloat sz = v[MAT_SZ] * rmesa->state.depth.scale; + GLfloat tz = v[MAT_TZ] * rmesa->state.depth.scale; + + R200_FIREVERTICES( rmesa ); + R200_STATECHANGE( rmesa, vpt ); + + rmesa->hw.vpt.cmd[VPT_SE_VPORT_XSCALE] = *(GLuint *)&sx; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_XOFFSET] = *(GLuint *)&tx; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_YSCALE] = *(GLuint *)&sy; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_YOFFSET] = *(GLuint *)&ty; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_ZSCALE] = *(GLuint *)&sz; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_ZOFFSET] = *(GLuint *)&tz; +} + + + +static void r200Viewport( GLcontext *ctx, GLint x, GLint y, + GLsizei width, GLsizei height ) +{ + /* Don't pipeline viewport changes, conflict with window offset + * setting below. Could apply deltas to rescue pipelined viewport + * values, or keep the originals hanging around. + */ + R200_FIREVERTICES( R200_CONTEXT(ctx) ); + r200UpdateWindow( ctx ); +} + +static void r200DepthRange( GLcontext *ctx, GLclampd nearval, + GLclampd farval ) +{ + r200UpdateWindow( ctx ); +} + +void r200UpdateViewportOffset( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; + GLfloat xoffset = (GLfloat)dPriv->x; + GLfloat yoffset = (GLfloat)dPriv->y + dPriv->h; + const GLfloat *v = ctx->Viewport._WindowMap.m; + + GLfloat tx = v[MAT_TX] + xoffset; + GLfloat ty = (- v[MAT_TY]) + yoffset; + + if ( rmesa->hw.vpt.cmd[VPT_SE_VPORT_XOFFSET] != tx || + rmesa->hw.vpt.cmd[VPT_SE_VPORT_YOFFSET] != ty ) + { + /* Note: this should also modify whatever data the context reset + * code uses... + */ + rmesa->hw.vpt.cmd[VPT_SE_VPORT_XOFFSET] = *(GLuint *)&tx; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_YOFFSET] = *(GLuint *)&ty; + + /* update polygon stipple x/y screen offset */ + { + GLuint stx, sty; + GLuint m = rmesa->hw.msc.cmd[MSC_RE_MISC]; + + m &= ~(R200_STIPPLE_X_OFFSET_MASK | + R200_STIPPLE_Y_OFFSET_MASK); + + /* add magic offsets, then invert */ + stx = 31 - ((rmesa->dri.drawable->x - 1) & R200_STIPPLE_COORD_MASK); + sty = 31 - ((rmesa->dri.drawable->y + rmesa->dri.drawable->h - 1) + & R200_STIPPLE_COORD_MASK); + + m |= ((stx << R200_STIPPLE_X_OFFSET_SHIFT) | + (sty << R200_STIPPLE_Y_OFFSET_SHIFT)); + + if ( rmesa->hw.msc.cmd[MSC_RE_MISC] != m ) { + R200_STATECHANGE( rmesa, msc ); + rmesa->hw.msc.cmd[MSC_RE_MISC] = m; + } + } + } + + r200UpdateScissor( ctx ); +} + + + +/* ============================================================= + * Miscellaneous + */ + +static void r200ClearColor( GLcontext *ctx, const GLchan c[4] ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + rmesa->state.color.clear = r200PackColor( rmesa->r200Screen->cpp, + c[0], c[1], c[2], c[3] ); +} + + +static void r200RenderMode( GLcontext *ctx, GLenum mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + FALLBACK( rmesa, R200_FALLBACK_RENDER_MODE, (mode != GL_RENDER) ); +} + + +static GLuint r200_rop_tab[] = { + R200_ROP_CLEAR, + R200_ROP_AND, + R200_ROP_AND_REVERSE, + R200_ROP_COPY, + R200_ROP_AND_INVERTED, + R200_ROP_NOOP, + R200_ROP_XOR, + R200_ROP_OR, + R200_ROP_NOR, + R200_ROP_EQUIV, + R200_ROP_INVERT, + R200_ROP_OR_REVERSE, + R200_ROP_COPY_INVERTED, + R200_ROP_OR_INVERTED, + R200_ROP_NAND, + R200_ROP_SET, +}; + +static void r200LogicOpCode( GLcontext *ctx, GLenum opcode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint rop = (GLuint)opcode - GL_CLEAR; + + ASSERT( rop < 16 ); + + R200_STATECHANGE( rmesa, msk ); + rmesa->hw.msk.cmd[MSK_RB3D_ROPCNTL] = r200_rop_tab[rop]; +} + + +void r200SetCliprects( r200ContextPtr rmesa, GLenum mode ) +{ + __DRIdrawablePrivate *dPriv = rmesa->dri.drawable; + + switch ( mode ) { + case GL_FRONT_LEFT: + rmesa->numClipRects = dPriv->numClipRects; + rmesa->pClipRects = (XF86DRIClipRectPtr)dPriv->pClipRects; + break; + case GL_BACK_LEFT: + /* Can't ignore 2d windows if we are page flipping. + */ + if ( dPriv->numBackClipRects == 0 || rmesa->doPageFlip ) { + rmesa->numClipRects = dPriv->numClipRects; + rmesa->pClipRects = (XF86DRIClipRectPtr)dPriv->pClipRects; + } + else { + rmesa->numClipRects = dPriv->numBackClipRects; + rmesa->pClipRects = (XF86DRIClipRectPtr)dPriv->pBackClipRects; + } + break; + default: + fprintf(stderr, "bad mode in r200SetCliprects\n"); + return; + } + + if (rmesa->state.scissor.enabled) + r200RecalcScissorRects( rmesa ); +} + + +static void r200SetDrawBuffer( GLcontext *ctx, GLenum mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_DRI) + fprintf(stderr, "%s %s\n", __FUNCTION__, + _mesa_lookup_enum_by_nr( mode )); + + R200_FIREVERTICES(rmesa); /* don't pipeline cliprect changes */ + + switch ( mode ) { + case GL_FRONT_LEFT: + FALLBACK( rmesa, R200_FALLBACK_DRAW_BUFFER, GL_FALSE ); + if ( rmesa->doPageFlip && rmesa->sarea->pfCurrentPage == 1 ) { + rmesa->state.color.drawOffset = rmesa->r200Screen->backOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->backPitch; + } else { + rmesa->state.color.drawOffset = rmesa->r200Screen->frontOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->frontPitch; + } + r200SetCliprects( rmesa, GL_FRONT_LEFT ); + break; + case GL_BACK_LEFT: + FALLBACK( rmesa, R200_FALLBACK_DRAW_BUFFER, GL_FALSE ); + if ( rmesa->doPageFlip && rmesa->sarea->pfCurrentPage == 1 ) { + rmesa->state.color.drawOffset = rmesa->r200Screen->frontOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->frontPitch; + } else { + rmesa->state.color.drawOffset = rmesa->r200Screen->backOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->backPitch; + } + r200SetCliprects( rmesa, GL_BACK_LEFT ); + break; + default: + FALLBACK( rmesa, R200_FALLBACK_DRAW_BUFFER, GL_TRUE ); + return; + } + + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_RB3D_COLOROFFSET] = (rmesa->state.color.drawOffset & + R200_COLOROFFSET_MASK); + rmesa->hw.ctx.cmd[CTX_RB3D_COLORPITCH] = rmesa->state.color.drawPitch; +} + + +/* ============================================================= + * State enable/disable + */ + +static void r200Enable( GLcontext *ctx, GLenum cap, GLboolean state ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint p, flag; + + if ( R200_DEBUG & DEBUG_STATE ) + fprintf( stderr, __FUNCTION__"( %s = %s )\n", + _mesa_lookup_enum_by_nr( cap ), + state ? "GL_TRUE" : "GL_FALSE" ); + + switch ( cap ) { + /* Fast track this one... + */ + case GL_TEXTURE_1D: + case GL_TEXTURE_2D: + case GL_TEXTURE_3D: + break; + + case GL_ALPHA_TEST: + R200_STATECHANGE( rmesa, ctx ); + if (state) { + rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= R200_ALPHA_TEST_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_PP_CNTL] &= ~R200_ALPHA_TEST_ENABLE; + } + break; + + case GL_BLEND: + R200_STATECHANGE( rmesa, ctx ); + if (state) { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_ALPHA_BLEND_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] &= ~R200_ALPHA_BLEND_ENABLE; + } + if ( ctx->Color.ColorLogicOpEnabled ) { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_ROP_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] &= ~R200_ROP_ENABLE; + } + break; + + case GL_CLIP_PLANE0: + case GL_CLIP_PLANE1: + case GL_CLIP_PLANE2: + case GL_CLIP_PLANE3: + case GL_CLIP_PLANE4: + case GL_CLIP_PLANE5: + p = cap-GL_CLIP_PLANE0; + R200_STATECHANGE( rmesa, tcl ); + if (state) { + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] |= (R200_UCP_ENABLE_0<<p); + r200ClipPlane( ctx, cap, NULL ); + } + else { + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] &= ~(R200_UCP_ENABLE_0<<p); + } + break; + + case GL_COLOR_MATERIAL: + r200ColorMaterial( ctx, 0, 0 ); + if (!state) + r200UpdateMaterial( ctx ); + break; + + case GL_CULL_FACE: + r200CullFace( ctx, 0 ); + break; + + case GL_DEPTH_TEST: + R200_STATECHANGE(rmesa, ctx ); + if ( state ) { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_Z_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] &= ~R200_Z_ENABLE; + } + break; + + case GL_DITHER: + R200_STATECHANGE(rmesa, ctx ); + if ( state ) { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_DITHER_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] &= ~R200_DITHER_ENABLE; + } + break; + + case GL_FOG: + R200_STATECHANGE(rmesa, ctx ); + if ( state ) { + rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= R200_FOG_ENABLE; + r200Fogfv( ctx, GL_FOG_MODE, 0 ); + } else { + rmesa->hw.ctx.cmd[CTX_PP_CNTL] &= ~R200_FOG_ENABLE; + R200_STATECHANGE(rmesa, tcl); + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] &= ~R200_TCL_FOG_MASK; + } + r200UpdateSpecular( ctx ); /* for PK_SPEC */ + break; + + case GL_LIGHT0: + case GL_LIGHT1: + case GL_LIGHT2: + case GL_LIGHT3: + case GL_LIGHT4: + case GL_LIGHT5: + case GL_LIGHT6: + case GL_LIGHT7: + R200_STATECHANGE(rmesa, tcl); + p = cap - GL_LIGHT0; + if (p&1) + flag = (R200_LIGHT_1_ENABLE | + R200_LIGHT_1_ENABLE_AMBIENT | + R200_LIGHT_1_ENABLE_SPECULAR); + else + flag = (R200_LIGHT_0_ENABLE | + R200_LIGHT_0_ENABLE_AMBIENT | + R200_LIGHT_0_ENABLE_SPECULAR); + + if (state) + rmesa->hw.tcl.cmd[p/2 + TCL_PER_LIGHT_CTL_0] |= flag; + else + rmesa->hw.tcl.cmd[p/2 + TCL_PER_LIGHT_CTL_0] &= ~flag; + + /* + */ + update_light_colors( ctx, p ); + break; + + case GL_LIGHTING: + r200UpdateSpecular(ctx); + check_twoside_fallback( ctx ); + break; + + case GL_LINE_SMOOTH: + R200_STATECHANGE( rmesa, ctx ); + if ( state ) { + rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= R200_ANTI_ALIAS_LINE; + } else { + rmesa->hw.ctx.cmd[CTX_PP_CNTL] &= ~R200_ANTI_ALIAS_LINE; + } + break; + + case GL_LINE_STIPPLE: + R200_STATECHANGE( rmesa, set ); + if ( state ) { + rmesa->hw.set.cmd[SET_RE_CNTL] |= R200_PATTERN_ENABLE; + } else { + rmesa->hw.set.cmd[SET_RE_CNTL] &= ~R200_PATTERN_ENABLE; + } + break; + + case GL_COLOR_LOGIC_OP: + R200_STATECHANGE( rmesa, ctx ); + if ( state ) { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_ROP_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] &= ~R200_ROP_ENABLE; + } + break; + + case GL_NORMALIZE: + R200_STATECHANGE( rmesa, tcl ); + if ( state ) { + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] |= R200_NORMALIZE_NORMALS; + } else { + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] &= ~R200_NORMALIZE_NORMALS; + } + break; + + /* Pointsize registers on r200 don't seem to do anything. Maybe + * have to pass pointsizes as vertex parameters? In any case, + * setting pointmin == pointsizemax == 1.0, and doing nothing + * for aa is enough to satisfy conform. + */ + case GL_POINT_SMOOTH: + break; + + /* These don't really do anything, as we don't use the 3vtx + * primitives yet. + */ +#if 0 + case GL_POLYGON_OFFSET_POINT: + R200_STATECHANGE( rmesa, set ); + if ( state ) { + rmesa->hw.set.cmd[SET_SE_CNTL] |= R200_ZBIAS_ENABLE_POINT; + } else { + rmesa->hw.set.cmd[SET_SE_CNTL] &= ~R200_ZBIAS_ENABLE_POINT; + } + break; + + case GL_POLYGON_OFFSET_LINE: + R200_STATECHANGE( rmesa, set ); + if ( state ) { + rmesa->hw.set.cmd[SET_SE_CNTL] |= R200_ZBIAS_ENABLE_LINE; + } else { + rmesa->hw.set.cmd[SET_SE_CNTL] &= ~R200_ZBIAS_ENABLE_LINE; + } + break; +#endif + + case GL_POLYGON_OFFSET_FILL: + R200_STATECHANGE( rmesa, set ); + if ( state ) { + rmesa->hw.set.cmd[SET_SE_CNTL] |= R200_ZBIAS_ENABLE_TRI; + } else { + rmesa->hw.set.cmd[SET_SE_CNTL] &= ~R200_ZBIAS_ENABLE_TRI; + } + break; + + case GL_POLYGON_SMOOTH: + R200_STATECHANGE( rmesa, ctx ); + if ( state ) { + rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= R200_ANTI_ALIAS_POLY; + } else { + rmesa->hw.ctx.cmd[CTX_PP_CNTL] &= ~R200_ANTI_ALIAS_POLY; + } + break; + + case GL_POLYGON_STIPPLE: + R200_STATECHANGE(rmesa, set ); + if ( state ) { + rmesa->hw.set.cmd[SET_RE_CNTL] |= R200_STIPPLE_ENABLE; + } else { + rmesa->hw.set.cmd[SET_RE_CNTL] &= ~R200_STIPPLE_ENABLE; + } + break; + + case GL_RESCALE_NORMAL_EXT: { + GLboolean tmp = ctx->_NeedEyeCoords ? state : !state; + R200_STATECHANGE( rmesa, tcl ); + if ( tmp ) { + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] |= R200_RESCALE_NORMALS; + } else { + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] &= ~R200_RESCALE_NORMALS; + } + break; + } + + case GL_SCISSOR_TEST: + R200_FIREVERTICES( rmesa ); + rmesa->state.scissor.enabled = state; + r200UpdateScissor( ctx ); + break; + + case GL_STENCIL_TEST: + if ( rmesa->state.stencil.hwBuffer ) { + R200_STATECHANGE( rmesa, ctx ); + if ( state ) { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_STENCIL_ENABLE; + } else { + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] &= ~R200_STENCIL_ENABLE; + } + } else { + FALLBACK( rmesa, R200_FALLBACK_STENCIL, state ); + } + break; + + case GL_TEXTURE_GEN_Q: + case GL_TEXTURE_GEN_R: + case GL_TEXTURE_GEN_S: + case GL_TEXTURE_GEN_T: + /* Picked up in r200UpdateTextureState. + */ + rmesa->recheck_texgen[ctx->Texture.CurrentUnit] = GL_TRUE; + break; + + case GL_COLOR_SUM_EXT: + r200UpdateSpecular ( ctx ); + break; + + default: + return; + } +} + + +void r200LightingSpaceChange( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLboolean tmp; + + if (R200_DEBUG & DEBUG_STATE) + fprintf(stderr, "%s %d BEFORE %x\n", __FUNCTION__, ctx->_NeedEyeCoords, + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0]); + + if (ctx->_NeedEyeCoords) + tmp = ctx->Transform.RescaleNormals; + else + tmp = !ctx->Transform.RescaleNormals; + + R200_STATECHANGE( rmesa, tcl ); + if ( tmp ) { + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] |= R200_RESCALE_NORMALS; + } else { + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] &= ~R200_RESCALE_NORMALS; + } + + if (R200_DEBUG & DEBUG_STATE) + fprintf(stderr, "%s %d AFTER %x\n", __FUNCTION__, ctx->_NeedEyeCoords, + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0]); +} + +/* ============================================================= + * Deferred state management - matrices, textures, other? + */ + + + + +static void upload_matrix( r200ContextPtr rmesa, GLfloat *src, int idx ) +{ + float *dest = ((float *)R200_DB_STATE( mat[idx] ))+MAT_ELT_0; + int i; + + + for (i = 0 ; i < 4 ; i++) { + *dest++ = src[i]; + *dest++ = src[i+4]; + *dest++ = src[i+8]; + *dest++ = src[i+12]; + } + + R200_DB_STATECHANGE( rmesa, &rmesa->hw.mat[idx] ); +} + +static void upload_matrix_t( r200ContextPtr rmesa, GLfloat *src, int idx ) +{ + float *dest = ((float *)R200_DB_STATE( mat[idx] ))+MAT_ELT_0; + memcpy(dest, src, 16*sizeof(float)); + R200_DB_STATECHANGE( rmesa, &rmesa->hw.mat[idx] ); +} + + +static void update_texturematrix( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + GLuint tpc = rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_0]; + GLuint compsel = rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL]; + int unit; + + if (R200_DEBUG & DEBUG_STATE) + fprintf(stderr, "%s before COMPSEL: %x\n", __FUNCTION__, + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL]); + + rmesa->TexMatEnabled = 0; + rmesa->TexMatCompSel = 0; + + for (unit = 0 ; unit < 2; unit++) { + if (!ctx->Texture.Unit[unit]._ReallyEnabled) + continue; + + if (ctx->TextureMatrix[unit].type != MATRIX_IDENTITY) { + rmesa->TexMatEnabled |= (R200_TEXGEN_TEXMAT_0_ENABLE| + R200_TEXMAT_0_ENABLE) << unit; + + rmesa->TexMatCompSel |= R200_OUTPUT_TEX_0 << unit; + + if (rmesa->TexGenEnabled & (R200_TEXMAT_0_ENABLE << unit)) { + /* Need to preconcatenate any active texgen + * obj/eyeplane matrices: + */ + _math_matrix_mul_matrix( &rmesa->tmpmat, + &rmesa->TexGenMatrix[unit], + &ctx->TextureMatrix[unit] ); + upload_matrix( rmesa, rmesa->tmpmat.m, R200_MTX_TEX0+unit ); + } + else { + upload_matrix( rmesa, ctx->TextureMatrix[unit].m, + R200_MTX_TEX0+unit ); + } + } + else if (rmesa->TexGenEnabled & (R200_TEXMAT_0_ENABLE << unit)) { + upload_matrix( rmesa, rmesa->TexGenMatrix[unit].m, + R200_MTX_TEX0+unit ); + } + } + + tpc = (rmesa->TexMatEnabled | rmesa->TexGenEnabled); + if (tpc != rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_0] || + rmesa->TexGenInputs != rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_1]) { + R200_STATECHANGE(rmesa, tcg); + rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_0] = tpc; + rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_1] = rmesa->TexGenInputs; + } + + compsel &= ~R200_OUTPUT_TEX_MASK; + compsel |= rmesa->TexMatCompSel | rmesa->TexGenCompSel; + if (compsel != rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL]) { + R200_STATECHANGE(rmesa, vtx); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] = compsel; + } +} + + + +void r200ValidateState( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint new_state = rmesa->NewGLState; + + if (new_state & _NEW_TEXTURE) { + r200UpdateTextureState( ctx ); + new_state |= rmesa->NewGLState; /* may add TEXTURE_MATRIX */ + } + + /* Need an event driven matrix update? + */ + if (new_state & (_NEW_MODELVIEW|_NEW_PROJECTION)) + upload_matrix( rmesa, ctx->_ModelProjectMatrix.m, R200_MTX_MVP ); + + /* Need these for lighting (shouldn't upload otherwise) + */ + if (new_state & (_NEW_MODELVIEW)) { + upload_matrix( rmesa, ctx->ModelView.m, R200_MTX_MV ); + upload_matrix_t( rmesa, ctx->ModelView.inv, R200_MTX_IMV ); + } + + /* Does this need to be triggered on eg. modelview for + * texgen-derived objplane/eyeplane matrices? + */ + if (new_state & (_NEW_TEXTURE|_NEW_TEXTURE_MATRIX)) { + update_texturematrix( ctx ); + } + + if (new_state & (_NEW_LIGHT|_NEW_MODELVIEW|_MESA_NEW_NEED_EYE_COORDS)) { + update_light( ctx ); + } + + /* emit all active clip planes if projection matrix changes. + */ + if (new_state & (_NEW_PROJECTION)) { + if (ctx->Transform._AnyClip) + r200UpdateClipPlanes( ctx ); + } + + + rmesa->NewGLState = 0; +} + + +static void r200InvalidateState( GLcontext *ctx, GLuint new_state ) +{ + _swrast_InvalidateState( ctx, new_state ); + _swsetup_InvalidateState( ctx, new_state ); + _ac_InvalidateState( ctx, new_state ); + _tnl_InvalidateState( ctx, new_state ); + _ae_invalidate_state( ctx, new_state ); + R200_CONTEXT(ctx)->NewGLState |= new_state; + r200VtxfmtInvalidate( ctx ); +} + +static void r200WrapRunPipeline( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + TNLcontext *tnl = TNL_CONTEXT(ctx); + + if (0) + fprintf(stderr, "%s, newstate: %x\n", __FUNCTION__, rmesa->NewGLState); + + /* Validate state: + */ + if (rmesa->NewGLState) + r200ValidateState( ctx ); + + if (tnl->vb.Material) { + TCL_FALLBACK( ctx, R200_TCL_FALLBACK_MATERIAL, GL_TRUE ); + } + + /* Run the pipeline. + */ + _tnl_run_pipeline( ctx ); + + if (tnl->vb.Material) { + TCL_FALLBACK( ctx, R200_TCL_FALLBACK_MATERIAL, GL_FALSE ); + r200UpdateMaterial( ctx ); /* not needed any more? */ + } +} + + +/* Initialize the driver's state functions. + */ +void r200InitStateFuncs( GLcontext *ctx ) +{ + ctx->Driver.UpdateState = r200InvalidateState; + ctx->Driver.LightingSpaceChange = r200LightingSpaceChange; + + ctx->Driver.SetDrawBuffer = r200SetDrawBuffer; + + ctx->Driver.AlphaFunc = r200AlphaFunc; + ctx->Driver.BlendEquation = r200BlendEquation; + ctx->Driver.BlendFunc = r200BlendFunc; + ctx->Driver.BlendFuncSeparate = r200BlendFuncSeparate; + ctx->Driver.ClearColor = r200ClearColor; + ctx->Driver.ClearDepth = NULL; + ctx->Driver.ClearIndex = NULL; + ctx->Driver.ClearStencil = r200ClearStencil; + ctx->Driver.ClipPlane = r200ClipPlane; + ctx->Driver.ColorMask = r200ColorMask; + ctx->Driver.CullFace = r200CullFace; + ctx->Driver.DepthFunc = r200DepthFunc; + ctx->Driver.DepthMask = r200DepthMask; + ctx->Driver.DepthRange = r200DepthRange; + ctx->Driver.Enable = r200Enable; + ctx->Driver.Fogfv = r200Fogfv; + ctx->Driver.FrontFace = r200FrontFace; + ctx->Driver.Hint = NULL; + ctx->Driver.IndexMask = NULL; + ctx->Driver.LightModelfv = r200LightModelfv; + ctx->Driver.Lightfv = r200Lightfv; + ctx->Driver.LineStipple = r200LineStipple; + ctx->Driver.LineWidth = r200LineWidth; + ctx->Driver.LogicOpcode = r200LogicOpCode; + ctx->Driver.PolygonMode = r200PolygonMode; + ctx->Driver.PolygonOffset = r200PolygonOffset; + ctx->Driver.PolygonStipple = r200PolygonStipple; + ctx->Driver.PointSize = r200PointSize; + ctx->Driver.RenderMode = r200RenderMode; + ctx->Driver.Scissor = r200Scissor; + ctx->Driver.ShadeModel = r200ShadeModel; + ctx->Driver.StencilFunc = r200StencilFunc; + ctx->Driver.StencilMask = r200StencilMask; + ctx->Driver.StencilOp = r200StencilOp; + ctx->Driver.Viewport = r200Viewport; + + /* Pixel path fallbacks + */ + ctx->Driver.Accum = _swrast_Accum; + ctx->Driver.Bitmap = _swrast_Bitmap; + ctx->Driver.CopyPixels = _swrast_CopyPixels; + ctx->Driver.DrawPixels = _swrast_DrawPixels; + ctx->Driver.ReadPixels = _swrast_ReadPixels; + + /* Swrast hooks for imaging extensions: + */ + ctx->Driver.CopyColorTable = _swrast_CopyColorTable; + ctx->Driver.CopyColorSubTable = _swrast_CopyColorSubTable; + ctx->Driver.CopyConvolutionFilter1D = _swrast_CopyConvolutionFilter1D; + ctx->Driver.CopyConvolutionFilter2D = _swrast_CopyConvolutionFilter2D; + + TNL_CONTEXT(ctx)->Driver.NotifyMaterialChange = r200UpdateMaterial; + TNL_CONTEXT(ctx)->Driver.RunPipeline = r200WrapRunPipeline; +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_state.h b/xc/lib/GL/mesa/src/drv/r200/r200_state.h new file mode 100644 index 000000000..b5055716b --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_state.h @@ -0,0 +1,68 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_STATE_H__ +#define __R200_STATE_H__ + +#ifdef GLX_DIRECT_RENDERING + +#include "r200_context.h" + +extern void r200InitState( r200ContextPtr rmesa ); +extern void r200InitStateFuncs( GLcontext *ctx ); + +extern void r200UpdateMaterial( GLcontext *ctx ); + +extern void r200SetCliprects( r200ContextPtr rmesa, GLenum mode ); +extern void r200RecalcScissorRects( r200ContextPtr rmesa ); +extern void r200UpdateViewportOffset( GLcontext *ctx ); +extern void r200UpdateWindow( GLcontext *ctx ); + +extern void r200ValidateState( GLcontext *ctx ); + +extern void r200PrintDirty( r200ContextPtr rmesa, + const char *msg ); + + +extern void r200Fallback( GLcontext *ctx, GLuint bit, GLboolean mode ); +#define FALLBACK( rmesa, bit, mode ) do { \ + if ( 0 ) fprintf( stderr, "FALLBACK in "__FUNCTION__": #%d=%d\n", \ + bit, mode ); \ + r200Fallback( rmesa->glCtx, bit, mode ); \ +} while (0) + +extern void r200LightingSpaceChange( GLcontext *ctx ); + +#endif +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_state_init.c b/xc/lib/GL/mesa/src/drv/r200/r200_state_init.c new file mode 100644 index 000000000..417b30ef9 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_state_init.c @@ -0,0 +1,691 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_ioctl.h" +#include "r200_state.h" +#include "r200_tcl.h" +#include "r200_tex.h" +#include "r200_swtcl.h" +#include "r200_vtxfmt.h" + +#include "mem.h" +#include "mmath.h" +#include "enums.h" +#include "colormac.h" +#include "light.h" +#include "api_arrayelt.h" + +#include "swrast/swrast.h" +#include "array_cache/acache.h" +#include "tnl/tnl.h" +#include "tnl/t_pipeline.h" +#include "swrast_setup/swrast_setup.h" + +/* ============================================================= + * State initialization + */ + +void r200PrintDirty( r200ContextPtr rmesa, const char *msg ) +{ + struct r200_state_atom *l; + + fprintf(stderr, msg); + fprintf(stderr, ": "); + + foreach(l, &(rmesa->hw.dirty)) { + fprintf(stderr, "%s, ", l->name); + } + + fprintf(stderr, "\n"); +} + +static int cmdpkt( int id ) +{ + drmRadeonCmdHeader h; + h.i = 0; + h.packet.cmd_type = RADEON_CMD_PACKET; + h.packet.packet_id = id; + return h.i; +} + +static int cmdvec( int offset, int stride, int count ) +{ + drmRadeonCmdHeader h; + h.i = 0; + h.vectors.cmd_type = RADEON_CMD_VECTORS; + h.vectors.offset = offset; + h.vectors.stride = stride; + h.vectors.count = count; + return h.i; +} + +static int cmdscl( int offset, int stride, int count ) +{ + drmRadeonCmdHeader h; + h.i = 0; + h.scalars.cmd_type = RADEON_CMD_SCALARS; + h.scalars.offset = offset; + h.scalars.stride = stride; + h.scalars.count = count; + return h.i; +} + +static int cmdscl2( int offset, int stride, int count ) +{ + drmRadeonCmdHeader h; + h.i = 0; + h.scalars.cmd_type = RADEON_CMD_SCALARS2; + h.scalars.offset = offset - 0x100; + h.scalars.stride = stride; + h.scalars.count = count; + return h.i; +} + +#define CHECK( NM, FLAG ) \ +static GLboolean check_##NM( GLcontext *ctx, int idx ) \ +{ \ + (void) idx; \ + return FLAG; \ +} + +#define TCL_CHECK( NM, FLAG ) \ +static GLboolean check_##NM( GLcontext *ctx, int idx ) \ +{ \ + r200ContextPtr rmesa = R200_CONTEXT(ctx); \ + (void) idx; \ + return !rmesa->TclFallback && (FLAG); \ +} + + +static GLboolean check_firsttime( GLcontext *ctx, int idx ) +{ + static int firsttime = 1; + int i = firsttime; + firsttime = 0; + return i; +} + +CHECK( always, GL_TRUE ) +CHECK( tex_any, ctx->Texture._ReallyEnabled ) +CHECK( tex, ctx->Texture.Unit[idx]._ReallyEnabled ) +CHECK( fog, ctx->Fog.Enabled ) +TCL_CHECK( tcl, GL_TRUE ) +TCL_CHECK( tcl_tex_any, ctx->Texture._ReallyEnabled ) +TCL_CHECK( tcl_tex, ctx->Texture.Unit[idx]._ReallyEnabled ) +TCL_CHECK( tcl_lighting, ctx->Light.Enabled ) +TCL_CHECK( tcl_eyespace_or_lighting, ctx->_NeedEyeCoords || ctx->Light.Enabled ) +TCL_CHECK( tcl_light, ctx->Light.Enabled && ctx->Light.Light[idx].Enabled ) +TCL_CHECK( tcl_ucp, ctx->Transform.ClipEnabled[idx] ) +/* TCL_CHECK( tcl_eyespace_or_fog, ctx->_NeedEyeCoords || ctx->Fog.Enabled ) */ + + +static GLboolean check_tcl_eyespace_or_fog( GLcontext *ctx, int idx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + int res; + (void) idx; + res = !rmesa->TclFallback && (ctx->_NeedEyeCoords || ctx->Fog.Enabled); + fprintf(stderr, "%s: %d\n", __FUNCTION__, res); + return res; +} + + +/* Initialize the context's hardware state. + */ +void r200InitState( r200ContextPtr rmesa ) +{ + GLcontext *ctx = rmesa->glCtx; + GLuint color_fmt, depth_fmt, i; + + switch ( rmesa->r200Screen->cpp ) { + case 2: + color_fmt = R200_COLOR_FORMAT_RGB565; + break; + case 4: + color_fmt = R200_COLOR_FORMAT_ARGB8888; + break; + default: + fprintf( stderr, "Error: Unsupported pixel depth... exiting\n" ); + exit( -1 ); + } + + rmesa->state.color.clear = 0x00000000; + + switch ( ctx->Visual.depthBits ) { + case 16: + rmesa->state.depth.scale = 1.0 / (GLfloat)0xffff; + depth_fmt = R200_DEPTH_FORMAT_16BIT_INT_Z; + rmesa->state.stencil.clear = 0x00000000; + break; + case 24: + rmesa->state.depth.scale = 1.0 / (GLfloat)0xffffff; + depth_fmt = R200_DEPTH_FORMAT_24BIT_INT_Z; + rmesa->state.stencil.clear = 0xff000000; + break; + default: + fprintf( stderr, "Error: Unsupported depth %d... exiting\n", + ctx->Visual.depthBits ); + exit( -1 ); + } + + /* Only have hw stencil when depth buffer is 24 bits deep */ + rmesa->state.stencil.hwBuffer = ( ctx->Visual.stencilBits > 0 && + ctx->Visual.depthBits == 24 ); + + rmesa->Fallback = 0; + + if ( ctx->Visual.doubleBufferMode && rmesa->sarea->pfCurrentPage == 0 ) { + rmesa->state.color.drawOffset = rmesa->r200Screen->backOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->backPitch; + } else { + rmesa->state.color.drawOffset = rmesa->r200Screen->frontOffset; + rmesa->state.color.drawPitch = rmesa->r200Screen->frontPitch; + } + + rmesa->state.pixel.readOffset = rmesa->state.color.drawOffset; + rmesa->state.pixel.readPitch = rmesa->state.color.drawPitch; + + /* Initialize lists: + */ + make_empty_list(&(rmesa->hw.dirty)); rmesa->hw.dirty.name = "DIRTY"; + make_empty_list(&(rmesa->hw.clean)); rmesa->hw.clean.name = "CLEAN"; + + +#define ALLOC_STATE( ATOM, CHK, SZ, NM, IDX ) \ + do { \ + rmesa->hw.ATOM.cmd_size = SZ; \ + rmesa->hw.ATOM.cmd = (int *)CALLOC(SZ * sizeof(int)); \ + rmesa->hw.ATOM.lastcmd = (int *)CALLOC(SZ * sizeof(int)); \ + rmesa->hw.ATOM.name = NM; \ + rmesa->hw.ATOM.idx = IDX; \ + rmesa->hw.ATOM.check = check_##CHK; \ + insert_at_head(&(rmesa->hw.dirty), &(rmesa->hw.ATOM)); \ + } while (0) + + + /* Allocate state buffers: + */ + ALLOC_STATE( ctx, always, CTX_STATE_SIZE, "CTX/context", 0 ); + ALLOC_STATE( set, always, SET_STATE_SIZE, "SET/setup", 0 ); + ALLOC_STATE( lin, always, LIN_STATE_SIZE, "LIN/line", 0 ); + ALLOC_STATE( msk, always, MSK_STATE_SIZE, "MSK/mask", 0 ); + ALLOC_STATE( vpt, always, VPT_STATE_SIZE, "VPT/viewport", 0 ); + ALLOC_STATE( vtx, always, VTX_STATE_SIZE, "VTX/vertex", 0 ); + ALLOC_STATE( vap, always, VAP_STATE_SIZE, "VAP/vap", 0 ); + ALLOC_STATE( vte, always, VTE_STATE_SIZE, "VTE/vte", 0 ); + ALLOC_STATE( msc, always, MSC_STATE_SIZE, "MSC/misc", 0 ); + ALLOC_STATE( cst, firsttime, CST_STATE_SIZE, "CST/constant", 0 ); + ALLOC_STATE( zbs, always, ZBS_STATE_SIZE, "ZBS/zbias", 0 ); + ALLOC_STATE( tam, tex_any, TAM_STATE_SIZE, "TAM/tam", 0 ); + ALLOC_STATE( tf, tex_any, TF_STATE_SIZE, "TF/tfactor", 0 ); + ALLOC_STATE( tex[0], tex_any, TEX_STATE_SIZE, "TEX/tex-0", 0 ); + ALLOC_STATE( tex[1], tex_any, TEX_STATE_SIZE, "TEX/tex-1", 1 ); + + ALLOC_STATE( tcl, tcl, TCL_STATE_SIZE, "TCL/tcl", 0 ); + ALLOC_STATE( msl, tcl, MSL_STATE_SIZE, "MSL/matrix-select", 0 ); + ALLOC_STATE( tcg, tcl, TCG_STATE_SIZE, "TCG/texcoordgen", 0 ); + ALLOC_STATE( mtl[0], tcl_lighting, MTL_STATE_SIZE, "MTL0/material0", 0 ); + ALLOC_STATE( grd, tcl, GRD_STATE_SIZE, "GRD/guard-band", 0 ); + ALLOC_STATE( fog, fog, FOG_STATE_SIZE, "FOG/fog", 0 ); + ALLOC_STATE( glt, tcl_lighting, GLT_STATE_SIZE, "GLT/light-global", 0 ); + ALLOC_STATE( eye, tcl_lighting, EYE_STATE_SIZE, "EYE/eye-vector", 0 ); + ALLOC_STATE( mat[R200_MTX_MV], tcl, MAT_STATE_SIZE, "MAT/modelview", 0 ); + ALLOC_STATE( mat[R200_MTX_IMV], tcl, MAT_STATE_SIZE, "MAT/it-modelview", 0 ); + ALLOC_STATE( mat[R200_MTX_MVP], tcl, MAT_STATE_SIZE, "MAT/modelproject", 0 ); + ALLOC_STATE( mat[R200_MTX_TEX0], tcl_tex, MAT_STATE_SIZE, "MAT/texmat0", 0 ); + ALLOC_STATE( mat[R200_MTX_TEX1], tcl_tex, MAT_STATE_SIZE, "MAT/texmat1", 1 ); + ALLOC_STATE( ucp[0], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-0", 0 ); + ALLOC_STATE( ucp[1], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-1", 1 ); + ALLOC_STATE( ucp[2], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-2", 2 ); + ALLOC_STATE( ucp[3], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-3", 3 ); + ALLOC_STATE( ucp[4], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-4", 4 ); + ALLOC_STATE( ucp[5], tcl_ucp, UCP_STATE_SIZE, "UCP/userclip-5", 5 ); + ALLOC_STATE( lit[0], tcl_light, LIT_STATE_SIZE, "LIT/light-0", 0 ); + ALLOC_STATE( lit[1], tcl_light, LIT_STATE_SIZE, "LIT/light-1", 1 ); + ALLOC_STATE( lit[2], tcl_light, LIT_STATE_SIZE, "LIT/light-2", 2 ); + ALLOC_STATE( lit[3], tcl_light, LIT_STATE_SIZE, "LIT/light-3", 3 ); + ALLOC_STATE( lit[4], tcl_light, LIT_STATE_SIZE, "LIT/light-4", 4 ); + ALLOC_STATE( lit[5], tcl_light, LIT_STATE_SIZE, "LIT/light-5", 5 ); + ALLOC_STATE( lit[6], tcl_light, LIT_STATE_SIZE, "LIT/light-6", 6 ); + ALLOC_STATE( lit[7], tcl_light, LIT_STATE_SIZE, "LIT/light-7", 7 ); + ALLOC_STATE( pix[0], always, PIX_STATE_SIZE, "PIX/pixstage-0", 0 ); + ALLOC_STATE( pix[1], tex, PIX_STATE_SIZE, "PIX/pixstage-1", 1 ); + + + /* Fill in the packet headers: + */ + rmesa->hw.ctx.cmd[CTX_CMD_0] = cmdpkt(RADEON_EMIT_PP_MISC); + rmesa->hw.ctx.cmd[CTX_CMD_1] = cmdpkt(RADEON_EMIT_PP_CNTL); + rmesa->hw.ctx.cmd[CTX_CMD_2] = cmdpkt(RADEON_EMIT_RB3D_COLORPITCH); + rmesa->hw.lin.cmd[LIN_CMD_0] = cmdpkt(RADEON_EMIT_RE_LINE_PATTERN); + rmesa->hw.lin.cmd[LIN_CMD_1] = cmdpkt(RADEON_EMIT_SE_LINE_WIDTH); + rmesa->hw.msk.cmd[MSK_CMD_0] = cmdpkt(RADEON_EMIT_RB3D_STENCILREFMASK); + rmesa->hw.vpt.cmd[VPT_CMD_0] = cmdpkt(RADEON_EMIT_SE_VPORT_XSCALE); + rmesa->hw.set.cmd[SET_CMD_0] = cmdpkt(RADEON_EMIT_SE_CNTL); + rmesa->hw.msc.cmd[MSC_CMD_0] = cmdpkt(RADEON_EMIT_RE_MISC); + rmesa->hw.cst.cmd[CST_CMD_0] = cmdpkt(R200_EMIT_PP_CNTL_X); + rmesa->hw.cst.cmd[CST_CMD_1] = cmdpkt(R200_EMIT_RB3D_DEPTHXY_OFFSET); + rmesa->hw.cst.cmd[CST_CMD_2] = cmdpkt(R200_EMIT_RE_AUX_SCISSOR_CNTL); + rmesa->hw.cst.cmd[CST_CMD_3] = cmdpkt(R200_EMIT_RE_SCISSOR_TL_0); + rmesa->hw.cst.cmd[CST_CMD_4] = cmdpkt(R200_EMIT_SE_VAP_CNTL_STATUS); + rmesa->hw.cst.cmd[CST_CMD_5] = cmdpkt(R200_EMIT_RE_POINTSIZE); + rmesa->hw.cst.cmd[CST_CMD_6] = cmdpkt(R200_EMIT_TCL_INPUT_VTX_VECTOR_ADDR_0); + rmesa->hw.tam.cmd[TAM_CMD_0] = cmdpkt(R200_EMIT_PP_TAM_DEBUG3); + rmesa->hw.tf.cmd[TF_CMD_0] = cmdpkt(R200_EMIT_TFACTOR_0); + rmesa->hw.tex[0].cmd[TEX_CMD_0] = cmdpkt(R200_EMIT_PP_TXFILTER_0); + rmesa->hw.tex[0].cmd[TEX_CMD_1] = cmdpkt(R200_EMIT_PP_TXOFFSET_0); + rmesa->hw.tex[1].cmd[TEX_CMD_0] = cmdpkt(R200_EMIT_PP_TXFILTER_1); + rmesa->hw.tex[1].cmd[TEX_CMD_1] = cmdpkt(R200_EMIT_PP_TXOFFSET_1); + rmesa->hw.pix[0].cmd[PIX_CMD_0] = cmdpkt(R200_EMIT_PP_TXCBLEND_0); + rmesa->hw.pix[1].cmd[PIX_CMD_0] = cmdpkt(R200_EMIT_PP_TXCBLEND_1); + rmesa->hw.zbs.cmd[ZBS_CMD_0] = cmdpkt(RADEON_EMIT_SE_ZBIAS_FACTOR); + rmesa->hw.tcl.cmd[TCL_CMD_0] = cmdpkt(R200_EMIT_TCL_LIGHT_MODEL_CTL_0); + rmesa->hw.tcl.cmd[TCL_CMD_1] = cmdpkt(R200_EMIT_TCL_UCP_VERT_BLEND_CTL); + rmesa->hw.tcg.cmd[TCG_CMD_0] = cmdpkt(R200_EMIT_TEX_PROC_CTL_2); + rmesa->hw.msl.cmd[MSL_CMD_0] = cmdpkt(R200_EMIT_MATRIX_SELECT_0); + rmesa->hw.vap.cmd[VAP_CMD_0] = cmdpkt(R200_EMIT_VAP_CTL); + rmesa->hw.vtx.cmd[VTX_CMD_0] = cmdpkt(R200_EMIT_VTX_FMT_0); + rmesa->hw.vtx.cmd[VTX_CMD_1] = cmdpkt(R200_EMIT_OUTPUT_VTX_COMP_SEL); + rmesa->hw.vtx.cmd[VTX_CMD_2] = cmdpkt(R200_EMIT_SE_VTX_STATE_CNTL); + rmesa->hw.vte.cmd[VTE_CMD_0] = cmdpkt(R200_EMIT_VTE_CNTL); + rmesa->hw.mtl[0].cmd[MTL_CMD_0] = + cmdvec( R200_VS_MAT_0_EMISS, 1, 16 ); + rmesa->hw.mtl[0].cmd[MTL_CMD_1] = + cmdscl2( R200_SS_MAT_0_SHININESS, 1, 1 ); + rmesa->hw.grd.cmd[GRD_CMD_0] = + cmdscl( R200_SS_VERT_GUARD_CLIP_ADJ_ADDR, 1, 4 ); + rmesa->hw.fog.cmd[FOG_CMD_0] = + cmdvec( R200_VS_FOG_PARAM_ADDR, 1, 4 ); + rmesa->hw.glt.cmd[GLT_CMD_0] = + cmdvec( R200_VS_GLOBAL_AMBIENT_ADDR, 1, 4 ); + rmesa->hw.eye.cmd[EYE_CMD_0] = + cmdvec( R200_VS_EYE_VECTOR_ADDR, 1, 4 ); + + rmesa->hw.mat[R200_MTX_MV].cmd[MAT_CMD_0] = + cmdvec( R200_VS_MATRIX_0_MV, 1, 16); + rmesa->hw.mat[R200_MTX_IMV].cmd[MAT_CMD_0] = + cmdvec( R200_VS_MATRIX_1_INV_MV, 1, 16); + rmesa->hw.mat[R200_MTX_MVP].cmd[MAT_CMD_0] = + cmdvec( R200_VS_MATRIX_2_MVP, 1, 16); + rmesa->hw.mat[R200_MTX_TEX0].cmd[MAT_CMD_0] = + cmdvec( R200_VS_MATRIX_3_TEX0, 1, 16); + rmesa->hw.mat[R200_MTX_TEX1].cmd[MAT_CMD_0] = + cmdvec( R200_VS_MATRIX_4_TEX1, 1, 16); + + for (i = 0 ; i < 8; i++) { + rmesa->hw.lit[i].cmd[LIT_CMD_0] = + cmdvec( R200_VS_LIGHT_AMBIENT_ADDR + i, 8, 24 ); + rmesa->hw.lit[i].cmd[LIT_CMD_1] = + cmdscl( R200_SS_LIGHT_DCD_ADDR + i, 8, 7 ); + } + + for (i = 0 ; i < 6; i++) { + rmesa->hw.ucp[i].cmd[UCP_CMD_0] = + cmdvec( R200_VS_UCP_ADDR + i, 1, 4 ); + } + + /* Initial Harware state: + */ + rmesa->hw.ctx.cmd[CTX_PP_MISC] = (R200_ALPHA_TEST_PASS | + R200_RIGHT_HAND_CUBE_OGL); + + rmesa->hw.ctx.cmd[CTX_PP_FOG_COLOR] = (R200_FOG_VERTEX | + R200_FOG_USE_SPEC_ALPHA); + + rmesa->hw.ctx.cmd[CTX_RE_SOLID_COLOR] = 0x00000000; + + rmesa->hw.ctx.cmd[CTX_RB3D_BLENDCNTL] = (R200_COMB_FCN_ADD_CLAMP | + R200_SRC_BLEND_GL_ONE | + R200_DST_BLEND_GL_ZERO ); + + rmesa->hw.ctx.cmd[CTX_RB3D_DEPTHOFFSET] = + rmesa->r200Screen->depthOffset; + + rmesa->hw.ctx.cmd[CTX_RB3D_DEPTHPITCH] = + ((rmesa->r200Screen->depthPitch & + R200_DEPTHPITCH_MASK) | + R200_DEPTH_ENDIAN_NO_SWAP); + + rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] = (depth_fmt | + R200_Z_TEST_LESS | + R200_STENCIL_TEST_ALWAYS | + R200_STENCIL_FAIL_KEEP | + R200_STENCIL_ZPASS_KEEP | + R200_STENCIL_ZFAIL_KEEP | + R200_Z_WRITE_ENABLE); + + rmesa->hw.ctx.cmd[CTX_PP_CNTL] = (R200_ANTI_ALIAS_NONE + | R200_TEX_BLEND_0_ENABLE); + + rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] = color_fmt; + + rmesa->hw.ctx.cmd[CTX_RB3D_COLOROFFSET] = (rmesa->state.color.drawOffset & + R200_COLOROFFSET_MASK); + + rmesa->hw.ctx.cmd[CTX_RB3D_COLORPITCH] = ((rmesa->state.color.drawPitch & + R200_COLORPITCH_MASK) | + R200_COLOR_ENDIAN_NO_SWAP); + + rmesa->hw.set.cmd[SET_SE_CNTL] = (R200_FFACE_CULL_CCW | + R200_BFACE_SOLID | + R200_FFACE_SOLID | + R200_FLAT_SHADE_VTX_LAST | + R200_DIFFUSE_SHADE_GOURAUD | + R200_ALPHA_SHADE_GOURAUD | + R200_SPECULAR_SHADE_GOURAUD | + R200_FOG_SHADE_GOURAUD | + R200_VTX_PIX_CENTER_OGL | + R200_ROUND_MODE_TRUNC | + R200_ROUND_PREC_8TH_PIX); + + rmesa->hw.set.cmd[SET_RE_CNTL] = (R200_PERSPECTIVE_ENABLE | + R200_SCISSOR_ENABLE); + + rmesa->hw.lin.cmd[LIN_RE_LINE_PATTERN] = + ((0x0000 & R200_LINE_PATTERN_MASK) | + (0 << R200_LINE_REPEAT_COUNT_SHIFT) | + (0 << R200_LINE_PATTERN_START_SHIFT) | + R200_LINE_PATTERN_LITTLE_BIT_ORDER); + + rmesa->hw.lin.cmd[LIN_RE_LINE_STATE] = + ((0 << R200_LINE_CURRENT_PTR_SHIFT) | + (1 << R200_LINE_CURRENT_COUNT_SHIFT)); + + rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] = (1 << 4); + + rmesa->hw.msk.cmd[MSK_RB3D_STENCILREFMASK] = + ((0x00 << R200_STENCIL_REF_SHIFT) | + (0xff << R200_STENCIL_MASK_SHIFT) | + (0xff << R200_STENCIL_WRITEMASK_SHIFT)); + + rmesa->hw.msk.cmd[MSK_RB3D_ROPCNTL] = R200_ROP_COPY; + rmesa->hw.msk.cmd[MSK_RB3D_PLANEMASK] = 0xffffffff; + + rmesa->hw.tam.cmd[TAM_DEBUG3] = 0; + + rmesa->hw.msc.cmd[MSC_RE_MISC] = + ((0 << R200_STIPPLE_X_OFFSET_SHIFT) | + (0 << R200_STIPPLE_Y_OFFSET_SHIFT) | + R200_STIPPLE_BIG_BIT_ORDER); + + + rmesa->hw.cst.cmd[CST_PP_CNTL_X] = 0; + rmesa->hw.cst.cmd[CST_RB3D_DEPTHXY_OFFSET] = 0; + rmesa->hw.cst.cmd[CST_RE_AUX_SCISSOR_CNTL] = 0x0; + rmesa->hw.cst.cmd[CST_RE_SCISSOR_TL_0] = 0; + rmesa->hw.cst.cmd[CST_RE_SCISSOR_BR_0] = 0; + rmesa->hw.cst.cmd[CST_SE_VAP_CNTL_STATUS] = 0; + rmesa->hw.cst.cmd[CST_RE_POINTSIZE] = 0x100010; + rmesa->hw.cst.cmd[CST_SE_TCL_INPUT_VTX_0] = + (0x0 << R200_VERTEX_POSITION_ADDR__SHIFT); + rmesa->hw.cst.cmd[CST_SE_TCL_INPUT_VTX_1] = + (0x02 << R200_VTX_COLOR_0_ADDR__SHIFT) | + (0x03 << R200_VTX_COLOR_1_ADDR__SHIFT); + rmesa->hw.cst.cmd[CST_SE_TCL_INPUT_VTX_2] = + (0x06 << R200_VTX_TEX_0_ADDR__SHIFT) | + (0x07 << R200_VTX_TEX_1_ADDR__SHIFT) | + (0x08 << R200_VTX_TEX_2_ADDR__SHIFT) | + (0x09 << R200_VTX_TEX_3_ADDR__SHIFT); + rmesa->hw.cst.cmd[CST_SE_TCL_INPUT_VTX_3] = + (0x0A << R200_VTX_TEX_4_ADDR__SHIFT) | + (0x0B << R200_VTX_TEX_5_ADDR__SHIFT); + + + rmesa->hw.vpt.cmd[VPT_SE_VPORT_XSCALE] = 0x00000000; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_XOFFSET] = 0x00000000; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_YSCALE] = 0x00000000; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_YOFFSET] = 0x00000000; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_ZSCALE] = 0x00000000; + rmesa->hw.vpt.cmd[VPT_SE_VPORT_ZOFFSET] = 0x00000000; + + rmesa->hw.tex[0].cmd[TEX_PP_TXFILTER] = R200_BORDER_MODE_OGL; + rmesa->hw.tex[0].cmd[TEX_PP_TXFORMAT] = + (R200_TXFORMAT_ST_ROUTE_STQ0 | + (2 << R200_TXFORMAT_WIDTH_SHIFT) | + (2 << R200_TXFORMAT_HEIGHT_SHIFT)); + rmesa->hw.tex[0].cmd[TEX_PP_TXOFFSET] = 0; + rmesa->hw.tex[0].cmd[TEX_PP_BORDER_COLOR] = 0; + rmesa->hw.tex[0].cmd[TEX_PP_TXFORMAT_X] = + (/* R200_TEXCOORD_PROJ | */ + 0x100000); /* Small default bias */ + + rmesa->hw.pix[0].cmd[PIX_PP_TXCBLEND] = + (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_DIFFUSE_COLOR | + R200_TXC_OP_MADD); + + rmesa->hw.pix[0].cmd[PIX_PP_TXCBLEND2] = + ((0 << R200_TXC_TFACTOR_SEL_SHIFT) | + R200_TXC_SCALE_1X | + R200_TXC_CLAMP_0_1 | + R200_TXC_OUTPUT_REG_R0); + + rmesa->hw.pix[0].cmd[PIX_PP_TXABLEND] = + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_DIFFUSE_ALPHA | + R200_TXA_OP_MADD); + + rmesa->hw.pix[0].cmd[PIX_PP_TXABLEND2] = + ((0 << R200_TXA_TFACTOR_SEL_SHIFT) | + R200_TXA_SCALE_1X | + R200_TXA_CLAMP_0_1 | + R200_TXA_OUTPUT_REG_R0); + + rmesa->hw.tex[1].cmd[TEX_PP_TXFILTER] = R200_BORDER_MODE_OGL; + rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] = + (R200_TXFORMAT_ST_ROUTE_STQ1 | + (2 << R200_TXFORMAT_WIDTH_SHIFT) | + (2 << R200_TXFORMAT_HEIGHT_SHIFT)); + rmesa->hw.tex[1].cmd[TEX_PP_TXOFFSET] = 0; + rmesa->hw.tex[1].cmd[TEX_PP_BORDER_COLOR] = 0; + rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT_X] = + (/* R200_TEXCOORD_PROJ | */ + 0x100000); /* Small default bias */ + + rmesa->hw.pix[1].cmd[PIX_PP_TXCBLEND] = + (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_DIFFUSE_COLOR | + R200_TXC_OP_MADD); + + rmesa->hw.pix[1].cmd[PIX_PP_TXCBLEND2] = + ((0 << R200_TXC_TFACTOR_SEL_SHIFT) | + R200_TXC_SCALE_1X | + R200_TXC_CLAMP_0_1 | + R200_TXC_OUTPUT_REG_R0); + + rmesa->hw.pix[1].cmd[PIX_PP_TXABLEND] = + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_DIFFUSE_ALPHA | + R200_TXA_OP_MADD); + + rmesa->hw.pix[1].cmd[PIX_PP_TXABLEND2] = + ((0 << R200_TXA_TFACTOR_SEL_SHIFT) | + R200_TXA_SCALE_1X | + R200_TXA_CLAMP_0_1 | + R200_TXA_OUTPUT_REG_R0); + + rmesa->hw.tf.cmd[TF_TFACTOR_0] = 0; + rmesa->hw.tf.cmd[TF_TFACTOR_1] = 0; + rmesa->hw.tf.cmd[TF_TFACTOR_2] = 0; + rmesa->hw.tf.cmd[TF_TFACTOR_3] = 0; + rmesa->hw.tf.cmd[TF_TFACTOR_4] = 0; + rmesa->hw.tf.cmd[TF_TFACTOR_5] = 0; + + rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL] = + (R200_VAP_TCL_ENABLE | + (0x9 << R200_VAP_VF_MAX_VTX_NUM__SHIFT)); + + rmesa->hw.vte.cmd[VTE_SE_VTE_CNTL] = + (R200_VPORT_X_SCALE_ENA | + R200_VPORT_Y_SCALE_ENA | + R200_VPORT_Z_SCALE_ENA | + R200_VPORT_X_OFFSET_ENA | + R200_VPORT_Y_OFFSET_ENA | + R200_VPORT_Z_OFFSET_ENA | + R200_VTX_W0_FMT); + + + rmesa->hw.vtx.cmd[VTX_VTXFMT_0] = 0; + rmesa->hw.vtx.cmd[VTX_VTXFMT_1] = 0; + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] = + ((R200_VTX_Z0 | R200_VTX_W0 | + (R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT))); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_1] = 0; + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] = (R200_OUTPUT_XYZW); + rmesa->hw.vtx.cmd[VTX_STATE_CNTL] = R200_VSC_UPDATE_USER_COLOR_0_ENABLE; + + + /* Matrix selection */ + rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_0] = + (R200_MTX_MV << R200_MODELVIEW_0_SHIFT); + + rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_1] = + (R200_MTX_IMV << R200_IT_MODELVIEW_0_SHIFT); + + rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_2] = + (R200_MTX_MVP << R200_MODELPROJECT_0_SHIFT); + + rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_3] = + ((R200_MTX_TEX0 << R200_TEXMAT_0_SHIFT) | + (R200_MTX_TEX1 << R200_TEXMAT_1_SHIFT) | + (R200_MTX_TEX2 << R200_TEXMAT_2_SHIFT) | + (R200_MTX_TEX3 << R200_TEXMAT_3_SHIFT)); + + rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_4] = + ((R200_MTX_TEX4 << R200_TEXMAT_4_SHIFT) | + (R200_MTX_TEX5 << R200_TEXMAT_5_SHIFT)); + + + /* General TCL state */ + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] = + (R200_SPECULAR_LIGHTS | + R200_DIFFUSE_SPECULAR_COMBINE | + R200_LOCAL_LIGHT_VEC_GL); + + rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_1] = + ((R200_LM1_SOURCE_LIGHT_PREMULT << R200_FRONT_EMISSIVE_SOURCE_SHIFT) | + (R200_LM1_SOURCE_LIGHT_PREMULT << R200_FRONT_AMBIENT_SOURCE_SHIFT) | + (R200_LM1_SOURCE_LIGHT_PREMULT << R200_FRONT_DIFFUSE_SOURCE_SHIFT) | + (R200_LM1_SOURCE_LIGHT_PREMULT << R200_FRONT_SPECULAR_SOURCE_SHIFT) | + (R200_LM1_SOURCE_LIGHT_PREMULT << R200_BACK_EMISSIVE_SOURCE_SHIFT) | + (R200_LM1_SOURCE_LIGHT_PREMULT << R200_BACK_AMBIENT_SOURCE_SHIFT) | + (R200_LM1_SOURCE_LIGHT_PREMULT << R200_BACK_DIFFUSE_SOURCE_SHIFT) | + (R200_LM1_SOURCE_LIGHT_PREMULT << R200_BACK_SPECULAR_SOURCE_SHIFT)); + + rmesa->hw.tcl.cmd[TCL_PER_LIGHT_CTL_0] = 0; /* filled in via callbacks */ + rmesa->hw.tcl.cmd[TCL_PER_LIGHT_CTL_1] = 0; + rmesa->hw.tcl.cmd[TCL_PER_LIGHT_CTL_2] = 0; + rmesa->hw.tcl.cmd[TCL_PER_LIGHT_CTL_3] = 0; + + rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] = + (R200_UCP_IN_CLIP_SPACE | + R200_CULL_FRONT_IS_CCW); + + /* Texgen/Texmat state */ + rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_2] = 0x0; /* masks??? */ + rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_3] = + ((0 << R200_TEXGEN_0_INPUT_TEX_SHIFT) | + (1 << R200_TEXGEN_1_INPUT_TEX_SHIFT) | + (2 << R200_TEXGEN_2_INPUT_TEX_SHIFT) | + (3 << R200_TEXGEN_3_INPUT_TEX_SHIFT) | + (4 << R200_TEXGEN_4_INPUT_TEX_SHIFT) | + (5 << R200_TEXGEN_5_INPUT_TEX_SHIFT)); + rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_0] = 0; + rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_1] = + ((0 << R200_TEXGEN_0_INPUT_SHIFT) | + (1 << R200_TEXGEN_1_INPUT_SHIFT) | + (2 << R200_TEXGEN_2_INPUT_SHIFT) | + (3 << R200_TEXGEN_3_INPUT_SHIFT) | + (4 << R200_TEXGEN_4_INPUT_SHIFT) | + (5 << R200_TEXGEN_5_INPUT_SHIFT)); + rmesa->hw.tcg.cmd[TCG_TEX_CYL_WRAP_CTL] = 0; + + rmesa->TexGenInputs = rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_1]; + + + for (i = 0 ; i < 8; i++) { + struct gl_light *l = &ctx->Light.Light[i]; + GLenum p = GL_LIGHT0 + i; + *(float *)&(rmesa->hw.lit[i].cmd[LIT_RANGE_CUTOFF]) = FLT_MAX; + + ctx->Driver.Lightfv( ctx, p, GL_AMBIENT, l->Ambient ); + ctx->Driver.Lightfv( ctx, p, GL_DIFFUSE, l->Diffuse ); + ctx->Driver.Lightfv( ctx, p, GL_SPECULAR, l->Specular ); + ctx->Driver.Lightfv( ctx, p, GL_POSITION, 0 ); + ctx->Driver.Lightfv( ctx, p, GL_SPOT_DIRECTION, 0 ); + ctx->Driver.Lightfv( ctx, p, GL_SPOT_EXPONENT, &l->SpotExponent ); + ctx->Driver.Lightfv( ctx, p, GL_SPOT_CUTOFF, &l->SpotCutoff ); + ctx->Driver.Lightfv( ctx, p, GL_CONSTANT_ATTENUATION, + &l->ConstantAttenuation ); + ctx->Driver.Lightfv( ctx, p, GL_LINEAR_ATTENUATION, + &l->LinearAttenuation ); + ctx->Driver.Lightfv( ctx, p, GL_QUADRATIC_ATTENUATION, + &l->QuadraticAttenuation ); + } + + ctx->Driver.LightModelfv( ctx, GL_LIGHT_MODEL_AMBIENT, + ctx->Light.Model.Ambient ); + + TNL_CONTEXT(ctx)->Driver.NotifyMaterialChange( ctx ); + + for (i = 0 ; i < 6; i++) { + ctx->Driver.ClipPlane( ctx, GL_CLIP_PLANE0 + i, NULL ); + } + + ctx->Driver.Fogfv( ctx, GL_FOG_MODE, 0 ); + ctx->Driver.Fogfv( ctx, GL_FOG_DENSITY, &ctx->Fog.Density ); + ctx->Driver.Fogfv( ctx, GL_FOG_START, &ctx->Fog.Start ); + ctx->Driver.Fogfv( ctx, GL_FOG_END, &ctx->Fog.End ); + ctx->Driver.Fogfv( ctx, GL_FOG_COLOR, ctx->Fog.Color ); + ctx->Driver.Fogfv( ctx, GL_FOG_COORDINATE_SOURCE_EXT, 0 ); + + rmesa->hw.grd.cmd[GRD_VERT_GUARD_CLIP_ADJ] = IEEE_ONE; + rmesa->hw.grd.cmd[GRD_VERT_GUARD_DISCARD_ADJ] = IEEE_ONE; + rmesa->hw.grd.cmd[GRD_HORZ_GUARD_CLIP_ADJ] = IEEE_ONE; + rmesa->hw.grd.cmd[GRD_HORZ_GUARD_DISCARD_ADJ] = IEEE_ONE; + + rmesa->hw.eye.cmd[EYE_X] = 0; + rmesa->hw.eye.cmd[EYE_Y] = 0; + rmesa->hw.eye.cmd[EYE_Z] = IEEE_ONE; + rmesa->hw.eye.cmd[EYE_RESCALE_FACTOR] = IEEE_ONE; + + r200LightingSpaceChange( ctx ); + + rmesa->lost_context = 1; +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_swtcl.c b/xc/lib/GL/mesa/src/drv/r200/r200_swtcl.c new file mode 100644 index 000000000..ec26785bd --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_swtcl.c @@ -0,0 +1,1140 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "glheader.h" +#include "mtypes.h" +#include "colormac.h" +#include "enums.h" +#include "mem.h" +#include "mmath.h" +#include "macros.h" + +#include "swrast_setup/swrast_setup.h" +#include "math/m_translate.h" +#include "tnl/tnl.h" +#include "tnl/t_context.h" +#include "tnl/t_pipeline.h" + +#include "r200_context.h" +#include "r200_ioctl.h" +#include "r200_state.h" +#include "r200_swtcl.h" +#include "r200_tcl.h" + +/*********************************************************************** + * Build render functions from dd templates * + ***********************************************************************/ + + +#define R200_XYZW_BIT 0x01 +#define R200_RGBA_BIT 0x02 +#define R200_SPEC_BIT 0x04 +#define R200_TEX0_BIT 0x08 +#define R200_TEX1_BIT 0x10 +#define R200_PTEX_BIT 0x20 +#define R200_MAX_SETUP 0x40 + +static void flush_last_swtcl_prim( r200ContextPtr rmesa ); + +static struct { + void (*emit)( GLcontext *, GLuint, GLuint, void *, GLuint ); + interp_func interp; + copy_pv_func copy_pv; + GLboolean (*check_tex_sizes)( GLcontext *ctx ); + GLuint vertex_size; + GLuint vertex_stride_shift; + GLuint vertex_format; +} setup_tab[R200_MAX_SETUP]; + + +static int se_vtx_fmt_0[] = { + 0, + + (R200_VTX_XY | + R200_VTX_Z0 | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT)), + + (R200_VTX_XY | + R200_VTX_Z0 | + R200_VTX_W0 | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT) | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_1_SHIFT)), + + (R200_VTX_XY | + R200_VTX_Z0 | + R200_VTX_W0 | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT) | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_1_SHIFT)), + + (R200_VTX_XY | + R200_VTX_Z0 | + R200_VTX_W0 | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT) | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_1_SHIFT)), + + (R200_VTX_XY | + R200_VTX_Z0 | + R200_VTX_W0 | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT) | + (R200_VTX_PK_RGBA << R200_VTX_COLOR_1_SHIFT)) +}; + +static int se_vtx_fmt_1[] = { + 0, + 0, + 0, + ((2 << R200_VTX_TEX0_COMP_CNT_SHIFT)), + ((2 << R200_VTX_TEX0_COMP_CNT_SHIFT) | + (2 << R200_VTX_TEX1_COMP_CNT_SHIFT)), + ((3 << R200_VTX_TEX0_COMP_CNT_SHIFT) | + (3 << R200_VTX_TEX1_COMP_CNT_SHIFT)), +}; + +#define TINY_VERTEX_FORMAT 1 +#define NOTEX_VERTEX_FORMAT 2 +#define TEX0_VERTEX_FORMAT 3 +#define TEX1_VERTEX_FORMAT 4 +#define PROJ_TEX1_VERTEX_FORMAT 5 +#define TEX2_VERTEX_FORMAT 0 +#define TEX3_VERTEX_FORMAT 0 +#define PROJ_TEX3_VERTEX_FORMAT 0 + +#define DO_XYZW (IND & R200_XYZW_BIT) +#define DO_RGBA (IND & R200_RGBA_BIT) +#define DO_SPEC (IND & R200_SPEC_BIT) +#define DO_FOG (IND & R200_SPEC_BIT) +#define DO_TEX0 (IND & R200_TEX0_BIT) +#define DO_TEX1 (IND & R200_TEX1_BIT) +#define DO_TEX2 0 +#define DO_TEX3 0 +#define DO_PTEX (IND & R200_PTEX_BIT) + +#define VERTEX r200Vertex +#define VERTEX_COLOR r200_color_t +#define GET_VIEWPORT_MAT() 0 +#define GET_TEXSOURCE(n) n +#define GET_VERTEX_FORMAT() R200_CONTEXT(ctx)->swtcl.vertex_format +#define GET_VERTEX_STORE() R200_CONTEXT(ctx)->swtcl.verts +#define GET_VERTEX_STRIDE_SHIFT() R200_CONTEXT(ctx)->swtcl.vertex_stride_shift +#define GET_UBYTE_COLOR_STORE() &R200_CONTEXT(ctx)->UbyteColor +#define GET_UBYTE_SPEC_COLOR_STORE() &R200_CONTEXT(ctx)->UbyteSecondaryColor + +#define HAVE_HW_VIEWPORT 1 +#define HAVE_HW_DIVIDE (IND & ~(R200_XYZW_BIT|R200_RGBA_BIT)) +#define HAVE_TINY_VERTICES 1 +#define HAVE_RGBA_COLOR 1 +#define HAVE_NOTEX_VERTICES 1 +#define HAVE_TEX0_VERTICES 1 +#define HAVE_TEX1_VERTICES 1 +#define HAVE_TEX2_VERTICES 0 +#define HAVE_TEX3_VERTICES 0 +#define HAVE_PTEX_VERTICES 1 + +#define CHECK_HW_DIVIDE (!(ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE| \ + DD_TRI_UNFILLED))) + +#define IMPORT_QUALIFIER +#define IMPORT_FLOAT_COLORS r200_import_float_colors +#define IMPORT_FLOAT_SPEC_COLORS r200_import_float_spec_colors + +#define INTERP_VERTEX setup_tab[R200_CONTEXT(ctx)->swtcl.SetupIndex].interp +#define COPY_PV_VERTEX setup_tab[R200_CONTEXT(ctx)->swtcl.SetupIndex].copy_pv + + +/*********************************************************************** + * Generate pv-copying and translation functions * + ***********************************************************************/ + +#define TAG(x) r200_##x +#define IND ~0 +#include "tnl_dd/t_dd_vb.c" +#undef IND + + +/*********************************************************************** + * Generate vertex emit and interp functions * + ***********************************************************************/ + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT) +#define TAG(x) x##_wg +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_TEX0_BIT) +#define TAG(x) x##_wgt0 +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_TEX0_BIT|R200_PTEX_BIT) +#define TAG(x) x##_wgpt0 +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_TEX0_BIT|R200_TEX1_BIT) +#define TAG(x) x##_wgt0t1 +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_TEX0_BIT|R200_TEX1_BIT|\ + R200_PTEX_BIT) +#define TAG(x) x##_wgpt0t1 +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_SPEC_BIT) +#define TAG(x) x##_wgfs +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_SPEC_BIT|\ + R200_TEX0_BIT) +#define TAG(x) x##_wgfst0 +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_SPEC_BIT|\ + R200_TEX0_BIT|R200_PTEX_BIT) +#define TAG(x) x##_wgfspt0 +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_SPEC_BIT|\ + R200_TEX0_BIT|R200_TEX1_BIT) +#define TAG(x) x##_wgfst0t1 +#include "tnl_dd/t_dd_vbtmp.h" + +#define IND (R200_XYZW_BIT|R200_RGBA_BIT|R200_SPEC_BIT|\ + R200_TEX0_BIT|R200_TEX1_BIT|R200_PTEX_BIT) +#define TAG(x) x##_wgfspt0t1 +#include "tnl_dd/t_dd_vbtmp.h" + + +/*********************************************************************** + * Initialization + ***********************************************************************/ + +static void init_setup_tab( void ) +{ + init_wg(); + init_wgt0(); + init_wgpt0(); + init_wgt0t1(); + init_wgpt0t1(); + init_wgfs(); + init_wgfst0(); + init_wgfspt0(); + init_wgfst0t1(); + init_wgfspt0t1(); +} + + + +void r200PrintSetupFlags(char *msg, GLuint flags ) +{ + fprintf(stderr, "%s(%x): %s%s%s%s%s%s\n", + msg, + (int)flags, + (flags & R200_XYZW_BIT) ? " xyzw," : "", + (flags & R200_RGBA_BIT) ? " rgba," : "", + (flags & R200_SPEC_BIT) ? " spec/fog," : "", + (flags & R200_TEX0_BIT) ? " tex-0," : "", + (flags & R200_TEX1_BIT) ? " tex-1," : "", + (flags & R200_PTEX_BIT) ? " proj-tex," : ""); +} + + +static void r200RenderStart( GLcontext *ctx ) +{ + TNLcontext *tnl = TNL_CONTEXT(ctx); + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + if (!setup_tab[rmesa->swtcl.SetupIndex].check_tex_sizes(ctx)) { + GLuint ind = rmesa->swtcl.SetupIndex |= (R200_PTEX_BIT|R200_RGBA_BIT); + + /* R200 handles projective textures nicely; just have to change + * up to the new vertex format. + */ + if (setup_tab[ind].vertex_format != rmesa->swtcl.vertex_format) { + int i; + R200_NEWPRIM(rmesa); + i = rmesa->swtcl.vertex_format = setup_tab[ind].vertex_format; + rmesa->swtcl.vertex_size = setup_tab[ind].vertex_size; + rmesa->swtcl.vertex_stride_shift = setup_tab[ind].vertex_stride_shift; + +/* fprintf(stderr, "%s: set vertex_format %d\n", __FUNCTION__, */ +/* rmesa->swtcl.vertex_format); */ +/* r200PrintSetupFlags( "setup flags", rmesa->swtcl.SetupIndex); */ + + R200_STATECHANGE( rmesa, vtx ); + rmesa->hw.vtx.cmd[VTX_VTXFMT_0] = se_vtx_fmt_0[i]; + rmesa->hw.vtx.cmd[VTX_VTXFMT_1] = se_vtx_fmt_1[i]; + } + + if (!(ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE|DD_TRI_UNFILLED))) { + tnl->Driver.Render.Interp = setup_tab[rmesa->swtcl.SetupIndex].interp; + tnl->Driver.Render.CopyPV = setup_tab[rmesa->swtcl.SetupIndex].copy_pv; + } + } + + if (rmesa->dma.flush != 0 && + rmesa->dma.flush != flush_last_swtcl_prim) + rmesa->dma.flush( rmesa ); +} + + +void r200BuildVertices( GLcontext *ctx, GLuint start, GLuint count, + GLuint newinputs ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + GLubyte *v = ((GLubyte *)rmesa->swtcl.verts + + (start << rmesa->swtcl.vertex_stride_shift)); + GLuint stride = 1 << rmesa->swtcl.vertex_stride_shift; + + newinputs |= rmesa->swtcl.SetupNewInputs; + rmesa->swtcl.SetupNewInputs = 0; + +/* fprintf(stderr, "%s: vertex_format %d\n", __FUNCTION__, */ +/* rmesa->swtcl.vertex_format); */ + + if (!newinputs) + return; + + setup_tab[rmesa->swtcl.SetupIndex].emit( ctx, start, count, v, stride ); +} + +void r200ChooseVertexState( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + TNLcontext *tnl = TNL_CONTEXT(ctx); + GLuint ind = (R200_XYZW_BIT | R200_RGBA_BIT); + + if (!rmesa->TclFallback || rmesa->Fallback) + return; + + if (ctx->Fog.Enabled || (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR)) + ind |= R200_SPEC_BIT; + + if (ctx->Texture._ReallyEnabled & 0x0f0) + ind |= R200_TEX0_BIT|R200_TEX1_BIT; + else if (ctx->Texture._ReallyEnabled & 0x00f) + ind |= R200_TEX0_BIT; + + rmesa->swtcl.SetupIndex = ind; + + if (ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE|DD_TRI_UNFILLED)) { + tnl->Driver.Render.Interp = r200_interp_extras; + tnl->Driver.Render.CopyPV = r200_copy_pv_extras; + } + else { + tnl->Driver.Render.Interp = setup_tab[ind].interp; + tnl->Driver.Render.CopyPV = setup_tab[ind].copy_pv; + } + + if (setup_tab[ind].vertex_format != rmesa->swtcl.vertex_format) { + int i; + R200_NEWPRIM(rmesa); + i = rmesa->swtcl.vertex_format = setup_tab[ind].vertex_format; + rmesa->swtcl.vertex_size = setup_tab[ind].vertex_size; + rmesa->swtcl.vertex_stride_shift = setup_tab[ind].vertex_stride_shift; + +/* fprintf(stderr, "%s: set vertex_format %d\n", __FUNCTION__, */ +/* rmesa->swtcl.vertex_format); */ +/* r200PrintSetupFlags( "setup flags", rmesa->swtcl.SetupIndex); */ + + R200_STATECHANGE( rmesa, vtx ); + rmesa->hw.vtx.cmd[VTX_VTXFMT_0] = se_vtx_fmt_0[i]; + rmesa->hw.vtx.cmd[VTX_VTXFMT_1] = se_vtx_fmt_1[i]; + } + + { + GLuint vte = rmesa->hw.vte.cmd[VTE_SE_VTE_CNTL]; + GLuint vap = rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL]; + GLuint needproj; + + /* HW perspective divide is a win, but tiny vertex formats are a + * bigger one. + */ + if (setup_tab[ind].vertex_format == TINY_VERTEX_FORMAT || + (ctx->_TriangleCaps & (DD_TRI_LIGHT_TWOSIDE|DD_TRI_UNFILLED))) { + needproj = GL_TRUE; + vte |= R200_VTX_XY_FMT | R200_VTX_Z_FMT; + vte &= ~R200_VTX_W0_FMT; + vap |= R200_VAP_FORCE_W_TO_ONE; + } + else { + needproj = GL_FALSE; + vte &= ~(R200_VTX_XY_FMT | R200_VTX_Z_FMT); + vte |= R200_VTX_W0_FMT; + vap &= ~R200_VAP_FORCE_W_TO_ONE; + } + + _tnl_need_projected_coords( ctx, needproj ); + if (vte != rmesa->hw.vte.cmd[VTE_SE_VTE_CNTL]) { + R200_STATECHANGE( rmesa, vte ); + rmesa->hw.vte.cmd[VTE_SE_VTE_CNTL] = vte; + } + if (vap != rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL]) { + R200_STATECHANGE( rmesa, vap ); + rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL] = vap; + } + } +} + + +/* Flush vertices in the current dma region. + */ +static void flush_last_swtcl_prim( r200ContextPtr rmesa ) +{ + if (R200_DEBUG & DEBUG_IOCTL) + fprintf(stderr, "%s\n", __FUNCTION__); + + if (rmesa->dma.current.buf) { + struct r200_dma_region *current = &rmesa->dma.current; + GLuint current_offset = (rmesa->dri.agp_buffer_offset + + current->buf->buf->idx * RADEON_BUFFER_SIZE + + current->start); + + assert (!(rmesa->swtcl.hw_primitive & R200_VF_PRIM_WALK_IND)); + + assert (current->start + + rmesa->swtcl.numverts * rmesa->swtcl.vertex_size * 4 == + current->ptr); + + if (rmesa->dma.current.start != rmesa->dma.current.ptr) { + r200EmitVertexAOS( rmesa, + rmesa->swtcl.vertex_size, + current_offset); + + r200EmitVbufPrim( rmesa, + rmesa->swtcl.hw_primitive, + rmesa->swtcl.numverts); + } + + rmesa->swtcl.numverts = 0; + current->start = current->ptr; + + rmesa->dma.flush = 0; + } +} + + +/* Alloc space in the current dma region. + */ +static __inline void *r200AllocDmaLowVerts( r200ContextPtr rmesa, + int nverts, int vsize ) +{ + GLuint bytes = vsize * nverts; + + if ( rmesa->dma.current.ptr + bytes > rmesa->dma.current.end ) + r200RefillCurrentDmaRegion( rmesa ); + + if (!rmesa->dma.flush) { + rmesa->dma.flush = flush_last_swtcl_prim; + } + + assert( vsize == rmesa->swtcl.vertex_size * 4 ); + assert( rmesa->dma.flush == flush_last_swtcl_prim ); + assert (rmesa->dma.current.start + + rmesa->swtcl.numverts * rmesa->swtcl.vertex_size * 4 == + rmesa->dma.current.ptr); + + + { + GLubyte *head = rmesa->dma.current.address + rmesa->dma.current.ptr; + rmesa->dma.current.ptr += bytes; + rmesa->swtcl.numverts += nverts; + return head; + } + +} + + + + +void r200_emit_contiguous_verts( GLcontext *ctx, GLuint start, GLuint count ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint vertex_size = rmesa->swtcl.vertex_size * 4; + CARD32 *dest = r200AllocDmaLowVerts( rmesa, count-start, vertex_size ); + setup_tab[rmesa->swtcl.SetupIndex].emit( ctx, start, count, dest, + vertex_size ); +} + + + +void r200_emit_indexed_verts( GLcontext *ctx, GLuint start, GLuint count ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + r200AllocDmaRegionVerts( rmesa, + &rmesa->swtcl.indexed_verts, + count - start, + rmesa->swtcl.vertex_size * 4, + 64); + + setup_tab[rmesa->swtcl.SetupIndex].emit( + ctx, start, count, + rmesa->swtcl.indexed_verts.address + rmesa->swtcl.indexed_verts.start, + rmesa->swtcl.vertex_size * 4 ); +} + + +/* + * Render unclipped vertex buffers by emitting vertices directly to + * dma buffers. Use strip/fan hardware primitives where possible. + * Try to simulate missing primitives with indexed vertices. + */ +#define HAVE_POINTS 1 +#define HAVE_LINES 1 +#define HAVE_LINE_STRIPS 1 +#define HAVE_TRIANGLES 1 +#define HAVE_TRI_STRIPS 1 +#define HAVE_TRI_STRIP_1 0 +#define HAVE_TRI_FANS 1 +#define HAVE_QUADS 1 +#define HAVE_QUAD_STRIPS 1 +#define HAVE_POLYGONS 1 +#define HAVE_ELTS 1 + +static const GLuint hw_prim[GL_POLYGON+1] = { + R200_VF_PRIM_POINTS, + R200_VF_PRIM_LINES, + 0, + R200_VF_PRIM_LINE_STRIP, + R200_VF_PRIM_TRIANGLES, + R200_VF_PRIM_TRIANGLE_STRIP, + R200_VF_PRIM_TRIANGLE_FAN, + R200_VF_PRIM_QUADS, + R200_VF_PRIM_QUAD_STRIP, + R200_VF_PRIM_POLYGON +}; + +static __inline void r200DmaPrimitive( r200ContextPtr rmesa, GLenum prim ) +{ + R200_NEWPRIM( rmesa ); + rmesa->swtcl.hw_primitive = hw_prim[prim]; + assert(rmesa->dma.current.ptr == rmesa->dma.current.start); +} + +static __inline void r200EltPrimitive( r200ContextPtr rmesa, GLenum prim ) +{ + R200_NEWPRIM( rmesa ); + rmesa->swtcl.hw_primitive = hw_prim[prim] | R200_VF_PRIM_WALK_IND; +} + + +static void VERT_FALLBACK( GLcontext *ctx, + GLuint start, + GLuint count, + GLuint flags ) +{ + TNLcontext *tnl = TNL_CONTEXT(ctx); + tnl->Driver.Render.PrimitiveNotify( ctx, flags & PRIM_MODE_MASK ); + tnl->Driver.Render.BuildVertices( ctx, start, count, ~0 ); + tnl->Driver.Render.PrimTabVerts[flags&PRIM_MODE_MASK]( ctx, start, count, flags ); + R200_CONTEXT(ctx)->swtcl.SetupNewInputs = VERT_CLIP; +} + +static void ELT_FALLBACK( GLcontext *ctx, + GLuint start, + GLuint count, + GLuint flags ) +{ + TNLcontext *tnl = TNL_CONTEXT(ctx); + tnl->Driver.Render.PrimitiveNotify( ctx, flags & PRIM_MODE_MASK ); + tnl->Driver.Render.BuildVertices( ctx, start, count, ~0 ); + tnl->Driver.Render.PrimTabElts[flags&PRIM_MODE_MASK]( ctx, start, count, flags ); + R200_CONTEXT(ctx)->swtcl.SetupNewInputs = VERT_CLIP; +} + + +#define LOCAL_VARS r200ContextPtr rmesa = R200_CONTEXT(ctx) +#define ELTS_VARS GLushort *dest +#define INIT( prim ) r200DmaPrimitive( rmesa, prim ) +#define ELT_INIT(prim) r200EltPrimitive( rmesa, prim ) +#define NEW_PRIMITIVE() R200_NEWPRIM( rmesa ) +#define NEW_BUFFER() r200RefillCurrentDmaRegion( rmesa ) +#define GET_CURRENT_VB_MAX_VERTS() \ + (((int)rmesa->dma.current.end - (int)rmesa->dma.current.ptr) / (rmesa->swtcl.vertex_size*4)) +#define GET_SUBSEQUENT_VB_MAX_VERTS() \ + ((RADEON_BUFFER_SIZE) / (rmesa->swtcl.vertex_size*4)) + +#define GET_CURRENT_VB_MAX_ELTS() \ + ((R200_CMD_BUF_SZ - (rmesa->store.cmd_used + 16)) / 2) +#define GET_SUBSEQUENT_VB_MAX_ELTS() \ + ((R200_CMD_BUF_SZ - 1024) / 2) + + + +/* How do you extend an existing primitive? + */ +#define ALLOC_ELTS(nr) \ +do { \ + if (rmesa->dma.flush == r200FlushElts && \ + rmesa->store.cmd_used + nr*2 < R200_CMD_BUF_SZ) { \ + \ + dest = (GLushort *)(rmesa->store.cmd_buf + \ + rmesa->store.cmd_used); \ + rmesa->store.cmd_used += nr*2; \ + } \ + else { \ + if (rmesa->dma.flush) { \ + rmesa->dma.flush( rmesa ); \ + } \ + \ + r200EmitVertexAOS( rmesa, \ + rmesa->swtcl.vertex_size, \ + (rmesa->dri.agp_buffer_offset + \ + rmesa->swtcl.indexed_verts.buf->buf->idx * \ + RADEON_BUFFER_SIZE + \ + rmesa->swtcl.indexed_verts.start)); \ + \ + dest = r200AllocEltsOpenEnded( rmesa, \ + rmesa->swtcl.hw_primitive, \ + nr ); \ + } \ +} while (0) + +#define ALLOC_ELTS_NEW_PRIMITIVE(nr) ALLOC_ELTS( nr ) + +#define EMIT_ELT(offset, x) (dest)[offset] = (GLushort) (x) +#if defined(__i386__) +#define EMIT_TWO_ELTS(offset, x, y) *(GLuint *)(dest+offset) = ((y)<<16)|(x); +#endif +#define INCR_ELTS( nr ) dest += nr +#define RELEASE_ELT_VERTS() \ + r200ReleaseDmaRegion( rmesa, &rmesa->swtcl.indexed_verts, __FUNCTION__ ) +#define EMIT_VERTS( ctx, j, nr ) \ + r200_emit_contiguous_verts(ctx, j, (j)+(nr)) +#define EMIT_INDEXED_VERTS( ctx, start, count ) \ + r200_emit_indexed_verts( ctx, start, count ) + + +#define TAG(x) r200_dma_##x +#include "tnl_dd/t_dd_dmatmp.h" + + +/**********************************************************************/ +/* Render pipeline stage */ +/**********************************************************************/ + + +static GLboolean r200_run_render( GLcontext *ctx, + struct gl_pipeline_stage *stage ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *VB = &tnl->vb; + GLuint i, length, flags = 0; + render_func *tab = TAG(render_tab_verts); + + if (rmesa->swtcl.indexed_verts.buf && (!VB->Elts || stage->changed_inputs)) + RELEASE_ELT_VERTS(); + + + + if ((R200_DEBUG & DEBUG_VERTS) || /* No debug */ + VB->ClipOrMask || /* No clipping */ + rmesa->swtcl.RenderIndex != 0 || /* No per-vertex manipulations */ + ctx->Line.StippleFlag) /* No stipple -- fix me? */ + return GL_TRUE; + + if (VB->Elts) { + tab = TAG(render_tab_elts); + if (!rmesa->swtcl.indexed_verts.buf) + if (!TAG(emit_elt_verts)(ctx, 0, VB->Count)) + return GL_TRUE; /* too many vertices */ + } + + tnl->Driver.Render.Start( ctx ); + + for (i = 0 ; !(flags & PRIM_LAST) ; i += length) + { + flags = VB->Primitive[i]; + length = VB->PrimitiveLength[i]; + + if (R200_DEBUG & DEBUG_PRIMS) + fprintf(stderr, "r200_render.c: prim %s %d..%d\n", + _mesa_lookup_enum_by_nr(flags & PRIM_MODE_MASK), + i, i+length); + + if (length) + tab[flags & PRIM_MODE_MASK]( ctx, i, i + length, flags ); + } + + tnl->Driver.Render.Finish( ctx ); + + return GL_FALSE; /* finished the pipe */ +} + + + +static void r200_check_render( GLcontext *ctx, + struct gl_pipeline_stage *stage ) +{ + GLuint inputs = VERT_OBJ|VERT_CLIP|VERT_RGBA; + + if (ctx->RenderMode == GL_RENDER) { + if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) + inputs |= VERT_SPEC_RGB; + + if (ctx->Texture.Unit[0]._ReallyEnabled) + inputs |= VERT_TEX(0); + + if (ctx->Texture.Unit[1]._ReallyEnabled) + inputs |= VERT_TEX(1); + + if (ctx->Fog.Enabled) + inputs |= VERT_FOG_COORD; + } + + stage->inputs = inputs; +} + + +static void dtr( struct gl_pipeline_stage *stage ) +{ + (void)stage; +} + + +const struct gl_pipeline_stage _r200_render_stage = +{ + "r200 render", + (_DD_NEW_SEPARATE_SPECULAR | + _NEW_TEXTURE| + _NEW_FOG| + _NEW_RENDERMODE), /* re-check (new inputs) */ + 0, /* re-run (always runs) */ + GL_TRUE, /* active */ + 0, 0, /* inputs (set in check_render), outputs */ + 0, 0, /* changed_inputs, private */ + dtr, /* destructor */ + r200_check_render, /* check - initially set to alloc data */ + r200_run_render /* run */ +}; + + + +/**************************************************************************/ + + +static const GLuint reduced_hw_prim[GL_POLYGON+1] = { + R200_VF_PRIM_POINTS, + R200_VF_PRIM_LINES, + R200_VF_PRIM_LINES, + R200_VF_PRIM_LINES, + R200_VF_PRIM_TRIANGLES, + R200_VF_PRIM_TRIANGLES, + R200_VF_PRIM_TRIANGLES, + R200_VF_PRIM_TRIANGLES, + R200_VF_PRIM_TRIANGLES, + R200_VF_PRIM_TRIANGLES +}; + +static void r200RasterPrimitive( GLcontext *ctx, GLuint hwprim ); +static void r200RenderPrimitive( GLcontext *ctx, GLenum prim ); +static void r200ResetLineStipple( GLcontext *ctx ); + +#undef HAVE_QUADS +#define HAVE_QUADS 0 + +#undef HAVE_QUAD_STRIPS +#define HAVE_QUAD_STRIPS 0 + +/*********************************************************************** + * Emit primitives as inline vertices * + ***********************************************************************/ + +#define CTX_ARG r200ContextPtr rmesa +#define CTX_ARG2 rmesa +#define GET_VERTEX_DWORDS() rmesa->swtcl.vertex_size +#define ALLOC_VERTS( n, size ) r200AllocDmaLowVerts( rmesa, n, size * 4 ) +#define LOCAL_VARS \ + r200ContextPtr rmesa = R200_CONTEXT(ctx); \ + const GLuint shift = rmesa->swtcl.vertex_stride_shift; \ + const char *r200verts = (char *)rmesa->swtcl.verts; +#define VERT(x) (r200Vertex *)(r200verts + (x << shift)) +#define VERTEX r200Vertex +#define DO_DEBUG_VERTS (1 && (R200_DEBUG & DEBUG_VERTS)) +#define PRINT_VERTEX(v) r200_print_vertex(rmesa->glCtx, v) +#undef TAG +#define TAG(x) r200_##x +#include "tnl_dd/t_dd_triemit.h" + + +/*********************************************************************** + * Macros for t_dd_tritmp.h to draw basic primitives * + ***********************************************************************/ + +#define QUAD( a, b, c, d ) r200_quad( rmesa, a, b, c, d ) +#define TRI( a, b, c ) r200_triangle( rmesa, a, b, c ) +#define LINE( a, b ) r200_line( rmesa, a, b ) +#define POINT( a ) r200_point( rmesa, a ) + +/*********************************************************************** + * Build render functions from dd templates * + ***********************************************************************/ + +#define R200_TWOSIDE_BIT 0x01 +#define R200_UNFILLED_BIT 0x02 +#define R200_MAX_TRIFUNC 0x04 + + +static struct { + points_func points; + line_func line; + triangle_func triangle; + quad_func quad; +} rast_tab[R200_MAX_TRIFUNC]; + + +#define DO_FALLBACK 0 +#define DO_UNFILLED (IND & R200_UNFILLED_BIT) +#define DO_TWOSIDE (IND & R200_TWOSIDE_BIT) +#define DO_FLAT 0 +#define DO_OFFSET 0 +#define DO_TRI 1 +#define DO_QUAD 1 +#define DO_LINE 1 +#define DO_POINTS 1 +#define DO_FULL_QUAD 1 + +#define HAVE_RGBA 1 +#define HAVE_SPEC 1 +#define HAVE_INDEX 0 +#define HAVE_BACK_COLORS 0 +#define HAVE_HW_FLATSHADE 1 +#define TAB rast_tab + +#define DEPTH_SCALE 1.0 +#define UNFILLED_TRI unfilled_tri +#define UNFILLED_QUAD unfilled_quad +#define VERT_X(_v) _v->v.x +#define VERT_Y(_v) _v->v.y +#define VERT_Z(_v) _v->v.z +#define AREA_IS_CCW( a ) (a < 0) +#define GET_VERTEX(e) (rmesa->swtcl.verts + (e<<rmesa->swtcl.vertex_stride_shift)) + +#define VERT_SET_RGBA( v, c ) v->ui[coloroffset] = *(GLuint *)c +#define VERT_COPY_RGBA( v0, v1 ) v0->ui[coloroffset] = v1->ui[coloroffset] +#define VERT_SAVE_RGBA( idx ) color[idx] = v[idx]->ui[coloroffset] +#define VERT_RESTORE_RGBA( idx ) v[idx]->ui[coloroffset] = color[idx] + +#define VERT_SET_SPEC( v, c ) if (havespec) COPY_3V(v->ub4[5], c ) +#define VERT_COPY_SPEC( v0, v1 ) if (havespec) COPY_3V(v0->ub4[5], v1->ub4[5]) +#define VERT_SAVE_SPEC( idx ) if (havespec) spec[idx] = v[idx]->ui[5] +#define VERT_RESTORE_SPEC( idx ) if (havespec) v[idx]->ui[5] = spec[idx] + +#undef LOCAL_VARS +#undef TAG +#undef INIT + +#define LOCAL_VARS(n) \ + r200ContextPtr rmesa = R200_CONTEXT(ctx); \ + GLuint color[n], spec[n]; \ + GLuint coloroffset = (rmesa->swtcl.vertex_size == 4 ? 3 : 4); \ + GLboolean havespec = (rmesa->swtcl.vertex_size > 4); \ + (void) color; (void) spec; (void) coloroffset; (void) havespec; + +/*********************************************************************** + * Helpers for rendering unfilled primitives * + ***********************************************************************/ + +#define RASTERIZE(x) r200RasterPrimitive( ctx, reduced_hw_prim[x] ) +#define RENDER_PRIMITIVE rmesa->swtcl.render_primitive +#define TAG(x) x +#include "tnl_dd/t_dd_unfilled.h" +#undef IND + + +/*********************************************************************** + * Generate GL render functions * + ***********************************************************************/ + + +#define IND (0) +#define TAG(x) x +#include "tnl_dd/t_dd_tritmp.h" + +#define IND (R200_TWOSIDE_BIT) +#define TAG(x) x##_twoside +#include "tnl_dd/t_dd_tritmp.h" + +#define IND (R200_UNFILLED_BIT) +#define TAG(x) x##_unfilled +#include "tnl_dd/t_dd_tritmp.h" + +#define IND (R200_TWOSIDE_BIT|R200_UNFILLED_BIT) +#define TAG(x) x##_twoside_unfilled +#include "tnl_dd/t_dd_tritmp.h" + + +static void init_rast_tab( void ) +{ + init(); + init_twoside(); + init_unfilled(); + init_twoside_unfilled(); +} + +/**********************************************************************/ +/* Render unclipped begin/end objects */ +/**********************************************************************/ + +#define VERT(x) (r200Vertex *)(r200verts + (x << shift)) +#define RENDER_POINTS( start, count ) \ + for ( ; start < count ; start++) \ + r200_point( rmesa, VERT(start) ) +#define RENDER_LINE( v0, v1 ) \ + r200_line( rmesa, VERT(v0), VERT(v1) ) +#define RENDER_TRI( v0, v1, v2 ) \ + r200_triangle( rmesa, VERT(v0), VERT(v1), VERT(v2) ) +#define RENDER_QUAD( v0, v1, v2, v3 ) \ + r200_quad( rmesa, VERT(v0), VERT(v1), VERT(v2), VERT(v3) ) +#define INIT(x) do { \ + r200RenderPrimitive( ctx, x ); \ +} while (0) +#undef LOCAL_VARS +#define LOCAL_VARS \ + r200ContextPtr rmesa = R200_CONTEXT(ctx); \ + const GLuint shift = rmesa->swtcl.vertex_stride_shift; \ + const char *r200verts = (char *)rmesa->swtcl.verts; \ + const GLuint * const elt = TNL_CONTEXT(ctx)->vb.Elts; \ + const GLboolean stipple = ctx->Line.StippleFlag; \ + (void) elt; (void) stipple; +#define RESET_STIPPLE if ( stipple ) r200ResetLineStipple( ctx ); +#define RESET_OCCLUSION +#define PRESERVE_VB_DEFS +#define ELT(x) (x) +#define TAG(x) r200_##x##_verts +#include "tnl/t_vb_rendertmp.h" +#undef ELT +#undef TAG +#define TAG(x) r200_##x##_elts +#define ELT(x) elt[x] +#include "tnl/t_vb_rendertmp.h" + + + +/**********************************************************************/ +/* Choose render functions */ +/**********************************************************************/ + +void r200ChooseRenderState( GLcontext *ctx ) +{ + TNLcontext *tnl = TNL_CONTEXT(ctx); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint index = 0; + GLuint flags = ctx->_TriangleCaps; + + if (!rmesa->TclFallback || rmesa->Fallback) + return; + + if (flags & DD_TRI_LIGHT_TWOSIDE) index |= R200_TWOSIDE_BIT; + if (flags & DD_TRI_UNFILLED) index |= R200_UNFILLED_BIT; + + if (index != rmesa->swtcl.RenderIndex) { + tnl->Driver.Render.Points = rast_tab[index].points; + tnl->Driver.Render.Line = rast_tab[index].line; + tnl->Driver.Render.ClippedLine = rast_tab[index].line; + tnl->Driver.Render.Triangle = rast_tab[index].triangle; + tnl->Driver.Render.Quad = rast_tab[index].quad; + + if (index == 0) { + tnl->Driver.Render.PrimTabVerts = r200_render_tab_verts; + tnl->Driver.Render.PrimTabElts = r200_render_tab_elts; + tnl->Driver.Render.ClippedPolygon = r200_fast_clipped_poly; + } else { + tnl->Driver.Render.PrimTabVerts = _tnl_render_tab_verts; + tnl->Driver.Render.PrimTabElts = _tnl_render_tab_elts; + tnl->Driver.Render.ClippedPolygon = _tnl_RenderClippedPolygon; + } + + rmesa->swtcl.RenderIndex = index; + } +} + + +/**********************************************************************/ +/* High level hooks for t_vb_render.c */ +/**********************************************************************/ + + +static void r200RasterPrimitive( GLcontext *ctx, GLuint hwprim ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (rmesa->swtcl.hw_primitive != hwprim) { + R200_NEWPRIM( rmesa ); + rmesa->swtcl.hw_primitive = hwprim; + } +} + +static void r200RenderPrimitive( GLcontext *ctx, GLenum prim ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + rmesa->swtcl.render_primitive = prim; + if (prim < GL_TRIANGLES || !(ctx->_TriangleCaps & DD_TRI_UNFILLED)) + r200RasterPrimitive( ctx, reduced_hw_prim[prim] ); +} + +static void r200RenderFinish( GLcontext *ctx ) +{ +} + +static void r200ResetLineStipple( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + R200_STATECHANGE( rmesa, lin ); +} + + +/**********************************************************************/ +/* Transition to/from hardware rasterization. */ +/**********************************************************************/ + +static char *fallbackStrings[] = { + "Texture mode", + "glDrawBuffer(GL_FRONT_AND_BACK)", + "glEnable(GL_STENCIL) without hw stencil buffer", + "glRenderMode(selection or feedback)", + "glBlendEquation", + "glBlendFunc(mode != ADD)" + "R200_NO_RAST" +}; + + +static char *getFallbackString(GLuint bit) +{ + int i = 0; + while (bit > 1) { + i++; + bit >>= 1; + } + return fallbackStrings[i]; +} + + +void r200Fallback( GLcontext *ctx, GLuint bit, GLboolean mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + TNLcontext *tnl = TNL_CONTEXT(ctx); + GLuint oldfallback = rmesa->Fallback; + + if (mode) { + rmesa->Fallback |= bit; + if (oldfallback == 0) { + R200_FIREVERTICES( rmesa ); + TCL_FALLBACK( ctx, R200_TCL_FALLBACK_RASTER, GL_TRUE ); + _swsetup_Wakeup( ctx ); + _tnl_need_projected_coords( ctx, GL_TRUE ); + rmesa->swtcl.RenderIndex = ~0; + if (R200_DEBUG & DEBUG_FALLBACKS) { + fprintf(stderr, "R200 begin rasterization fallback: 0x%x %s\n", + bit, getFallbackString(bit)); + } + } + } + else { + rmesa->Fallback &= ~bit; + if (oldfallback == bit) { + _swrast_flush( ctx ); + tnl->Driver.Render.Start = r200RenderStart; + tnl->Driver.Render.PrimitiveNotify = r200RenderPrimitive; + tnl->Driver.Render.Finish = r200RenderFinish; + tnl->Driver.Render.BuildVertices = r200BuildVertices; + tnl->Driver.Render.ResetLineStipple = r200ResetLineStipple; + TCL_FALLBACK( ctx, R200_TCL_FALLBACK_RASTER, GL_FALSE ); + if (rmesa->TclFallback) { + /* These are already done if rmesa->TclFallback goes to + * zero above. But not if it doesn't (R200_NO_TCL for + * example?) + */ + r200ChooseVertexState( ctx ); + r200ChooseRenderState( ctx ); + } + if (R200_DEBUG & DEBUG_FALLBACKS) { + fprintf(stderr, "R200 end rasterization fallback: 0x%x %s\n", + bit, getFallbackString(bit)); + } + } + } +} + + +/**********************************************************************/ +/* Initialization. */ +/**********************************************************************/ + +void r200InitSwtcl( GLcontext *ctx ) +{ + TNLcontext *tnl = TNL_CONTEXT(ctx); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint size = TNL_CONTEXT(ctx)->vb.Size; + static int firsttime = 1; + + if (firsttime) { + init_rast_tab(); + init_setup_tab(); + firsttime = 0; + } + + tnl->Driver.Render.Start = r200RenderStart; + tnl->Driver.Render.Finish = r200RenderFinish; + tnl->Driver.Render.PrimitiveNotify = r200RenderPrimitive; + tnl->Driver.Render.ResetLineStipple = r200ResetLineStipple; + tnl->Driver.Render.BuildVertices = r200BuildVertices; + + rmesa->swtcl.verts = (char *)ALIGN_MALLOC( size * 16 * 4, 32 ); + rmesa->swtcl.RenderIndex = ~0; + rmesa->swtcl.render_primitive = GL_TRIANGLES; + rmesa->swtcl.hw_primitive = 0; +} + + +void r200DestroySwtcl( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (rmesa->swtcl.indexed_verts.buf) + r200ReleaseDmaRegion( rmesa, &rmesa->swtcl.indexed_verts, __FUNCTION__ ); + + if (rmesa->swtcl.verts) { + ALIGN_FREE(rmesa->swtcl.verts); + rmesa->swtcl.verts = 0; + } + + if (rmesa->UbyteSecondaryColor.Ptr) { + ALIGN_FREE(rmesa->UbyteSecondaryColor.Ptr); + rmesa->UbyteSecondaryColor.Ptr = 0; + } + + if (rmesa->UbyteColor.Ptr) { + ALIGN_FREE(rmesa->UbyteColor.Ptr); + rmesa->UbyteColor.Ptr = 0; + } +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_swtcl.h b/xc/lib/GL/mesa/src/drv/r200/r200_swtcl.h new file mode 100644 index 000000000..3bf4ab5fd --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_swtcl.h @@ -0,0 +1,75 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_SWTCL_H__ +#define __R200_SWTCL_H__ + +#include "mtypes.h" +#include "swrast/swrast.h" +#include "r200_context.h" + +extern void r200InitSwtcl( GLcontext *ctx ); +extern void r200DestroySwtcl( GLcontext *ctx ); + +extern void r200ChooseRenderState( GLcontext *ctx ); +extern void r200ChooseVertexState( GLcontext *ctx ); + +extern void r200CheckTexSizes( GLcontext *ctx ); + +extern void r200BuildVertices( GLcontext *ctx, GLuint start, GLuint count, + GLuint newinputs ); + +extern void r200PrintSetupFlags(char *msg, GLuint flags ); + + +extern void r200_emit_contiguous_verts( GLcontext *ctx, + GLuint start, + GLuint count ); + +extern void r200_emit_indexed_verts( GLcontext *ctx, + GLuint start, + GLuint count ); + +extern void r200_translate_vertex( GLcontext *ctx, + const r200Vertex *src, + SWvertex *dst ); + +extern void r200_print_vertex( GLcontext *ctx, const r200Vertex *v ); + +extern void r200_import_float_colors( GLcontext *ctx ); +extern void r200_import_float_spec_colors( GLcontext *ctx ); + + + +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_tcl.c b/xc/lib/GL/mesa/src/drv/r200/r200_tcl.c new file mode 100644 index 000000000..42b796b15 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_tcl.c @@ -0,0 +1,546 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_state.h" +#include "r200_ioctl.h" +#include "r200_tex.h" +#include "r200_tcl.h" +#include "r200_swtcl.h" +#include "r200_maos.h" + +#include "mmath.h" +#include "mtypes.h" +#include "enums.h" +#include "colormac.h" +#include "light.h" + +#include "array_cache/acache.h" +#include "tnl/tnl.h" +#include "tnl/t_pipeline.h" + + + +#define HAVE_POINTS 1 +#define HAVE_LINES 1 +#define HAVE_LINE_LOOP 0 +#define HAVE_LINE_STRIPS 1 +#define HAVE_TRIANGLES 1 +#define HAVE_TRI_STRIPS 1 +#define HAVE_TRI_STRIP_1 0 +#define HAVE_TRI_FANS 1 +#define HAVE_QUADS 0 /* hw quad verts in wrong order??? */ +#define HAVE_QUAD_STRIPS 1 +#define HAVE_POLYGONS 1 +#define HAVE_ELTS 1 + + +#define HW_POINTS R200_VF_PRIM_POINTS +#define HW_LINES R200_VF_PRIM_LINES +#define HW_LINE_LOOP 0 +#define HW_LINE_STRIP R200_VF_PRIM_LINE_STRIP +#define HW_TRIANGLES R200_VF_PRIM_TRIANGLES +#define HW_TRIANGLE_STRIP_0 R200_VF_PRIM_TRIANGLE_STRIP +#define HW_TRIANGLE_STRIP_1 0 +#define HW_TRIANGLE_FAN R200_VF_PRIM_TRIANGLE_FAN +#define HW_QUADS R200_VF_PRIM_QUADS +#define HW_QUAD_STRIP R200_VF_PRIM_QUAD_STRIP +#define HW_POLYGON R200_VF_PRIM_POLYGON + + +static GLboolean discrete_prim[0x10] = { + 0, /* 0 none */ + 1, /* 1 points */ + 1, /* 2 lines */ + 0, /* 3 line_strip */ + 1, /* 4 tri_list */ + 0, /* 5 tri_fan */ + 0, /* 6 tri_strip */ + 0, /* 7 tri_w_flags */ + 1, /* 8 rect list (unused) */ + 1, /* 9 3vert point */ + 1, /* a 3vert line */ + 0, /* b point sprite */ + 0, /* c line loop */ + 1, /* d quads */ + 0, /* e quad strip */ + 0, /* f polygon */ +}; + + +#define LOCAL_VARS r200ContextPtr rmesa = R200_CONTEXT(ctx) +#define ELTS_VARS GLushort *dest + +#define ELT_INIT(prim, hw_prim) \ + r200TclPrimitive( ctx, prim, hw_prim | R200_VF_PRIM_WALK_IND ) + +#define GET_ELTS() rmesa->tcl.Elts + + +#define NEW_PRIMITIVE() R200_NEWPRIM( rmesa ) +#define NEW_BUFFER() r200RefillCurrentDmaRegion( rmesa ) + +/* Don't really know how many elts will fit in what's left of cmdbuf, + * as there is state to emit, etc: + */ + +#if 0 +#define GET_CURRENT_VB_MAX_ELTS() \ + ((R200_CMD_BUF_SZ - (rmesa->store.cmd_used + 16)) / 2) +#define GET_SUBSEQUENT_VB_MAX_ELTS() ((R200_CMD_BUF_SZ - 16) / 2) +#else +/* Testing on isosurf shows a maximum around here. Don't know if it's + * the card or driver or kernel module that is causing the behaviour. + */ +#define GET_CURRENT_VB_MAX_ELTS() 300 +#define GET_SUBSEQUENT_VB_MAX_ELTS() 300 +#endif + +#define RESET_STIPPLE() do { \ + R200_STATECHANGE( rmesa, lin ); \ + r200EmitState( rmesa ); \ +} while (0) + +#define AUTO_STIPPLE( mode ) do { \ + R200_STATECHANGE( rmesa, lin ); \ + if (mode) \ + rmesa->hw.lin.cmd[LIN_RE_LINE_PATTERN] |= \ + R200_LINE_PATTERN_AUTO_RESET; \ + else \ + rmesa->hw.lin.cmd[LIN_RE_LINE_PATTERN] &= \ + ~R200_LINE_PATTERN_AUTO_RESET; \ + r200EmitState( rmesa ); \ +} while (0) + + +/* How do you extend an existing primitive? + */ +#define ALLOC_ELTS(nr) \ +do { \ + if (rmesa->dma.flush == r200FlushElts && \ + rmesa->store.cmd_used + nr*2 < R200_CMD_BUF_SZ) { \ + \ + dest = (GLushort *)(rmesa->store.cmd_buf + \ + rmesa->store.cmd_used); \ + rmesa->store.cmd_used += nr*2; \ + } \ + else { \ + if (rmesa->dma.flush) \ + rmesa->dma.flush( rmesa ); \ + \ + r200EmitAOS( rmesa, \ + rmesa->tcl.aos_components, \ + rmesa->tcl.nr_aos_components, \ + 0 ); \ + \ + dest = r200AllocEltsOpenEnded( rmesa, \ + rmesa->tcl.hw_primitive, \ + nr ); \ + } \ +} while (0) + + + +/* TODO: Try to extend existing primitive if both are identical, + * discrete and there are no intervening state changes. (Somewhat + * duplicates changes to DrawArrays code) + */ +static void EMIT_PRIM( GLcontext *ctx, + GLenum prim, + GLuint hwprim, + GLuint start, + GLuint count) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + r200TclPrimitive( ctx, prim, hwprim ); + + r200EmitAOS( rmesa, + rmesa->tcl.aos_components, + rmesa->tcl.nr_aos_components, + start ); + + /* Why couldn't this packet have taken an offset param? + */ + r200EmitVbufPrim( rmesa, + rmesa->tcl.hw_primitive, + count - start ); +} + + + +/* Try & join small primitives + */ +#if 0 +#define PREFER_DISCRETE_ELT_PRIM( NR, PRIM ) 0 +#else +#define PREFER_DISCRETE_ELT_PRIM( NR, PRIM ) \ + ((NR) < 20 || \ + ((NR) < 40 && \ + rmesa->tcl.hw_primitive == (PRIM| \ + R200_VF_TCL_OUTPUT_VTX_ENABLE| \ + R200_VF_PRIM_WALK_IND))) +#endif + +#define EMIT_ELT(offset, x) (dest)[offset] = (GLushort) (x) +#if defined(__i386__) +#define EMIT_TWO_ELTS(offset, x, y) *(GLuint *)(dest+offset) = ((y)<<16)|(x); +#endif +#define INCR_ELTS( nr ) dest += nr +#define RELEASE_ELT_VERTS() \ + r200ReleaseArrays( ctx, ~0 ) + + + +#define TAG(x) tcl_##x +#include "tnl_dd/t_dd_dmatmp2.h" + +/**********************************************************************/ +/* External entrypoints */ +/**********************************************************************/ + +void r200EmitPrimitive( GLcontext *ctx, + GLuint first, + GLuint last, + GLuint flags ) +{ + tcl_render_tab_verts[flags&PRIM_MODE_MASK]( ctx, first, last, flags ); +} + +void r200EmitEltPrimitive( GLcontext *ctx, + GLuint first, + GLuint last, + GLuint flags ) +{ + tcl_render_tab_elts[flags&PRIM_MODE_MASK]( ctx, first, last, flags ); +} + +void r200TclPrimitive( GLcontext *ctx, + GLenum prim, + int hw_prim ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint newprim = hw_prim | R200_VF_TCL_OUTPUT_VTX_ENABLE; + + if (newprim != rmesa->tcl.hw_primitive || + !discrete_prim[hw_prim&0xf]) { + R200_NEWPRIM( rmesa ); + rmesa->tcl.hw_primitive = newprim; + } +} + + +/**********************************************************************/ +/* Render pipeline stage */ +/**********************************************************************/ + + +/* TCL render. + */ +static GLboolean r200_run_tcl_render( GLcontext *ctx, + struct gl_pipeline_stage *stage ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + TNLcontext *tnl = TNL_CONTEXT(ctx); + struct vertex_buffer *VB = &tnl->vb; + GLuint i,flags = 0,length; + + /* TODO: separate this from the swtnl pipeline + */ + if (rmesa->TclFallback) + return GL_TRUE; /* fallback to software t&l */ + + if (R200_DEBUG & DEBUG_PRIMS) + fprintf(stderr, "%s\n", __FUNCTION__); + + if (VB->Count == 0) + return GL_FALSE; + + r200ReleaseArrays( ctx, stage->changed_inputs ); + r200EmitArrays( ctx, stage->inputs ); + + rmesa->tcl.Elts = VB->Elts; + + for (i = VB->FirstPrimitive ; !(flags & PRIM_LAST) ; i += length) + { + flags = VB->Primitive[i]; + length = VB->PrimitiveLength[i]; + + if (R200_DEBUG & DEBUG_PRIMS) + fprintf(stderr, "%s: prim %s %d..%d\n", + __FUNCTION__, + _mesa_lookup_enum_by_nr(flags & PRIM_MODE_MASK), + i, i+length); + + if (!length) + continue; + + if (rmesa->tcl.Elts) + r200EmitEltPrimitive( ctx, i, i+length, flags ); + else + r200EmitPrimitive( ctx, i, i+length, flags ); + } + + return GL_FALSE; /* finished the pipe */ +} + + + +static void r200_check_tcl_render( GLcontext *ctx, + struct gl_pipeline_stage *stage ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint inputs = VERT_OBJ; + + /* Validate state: + */ + if (rmesa->NewGLState) + r200ValidateState( ctx ); + + if (0) + fprintf(stderr, "%s: RE %d TGE %d NN %d\n", + __FUNCTION__, + ctx->Texture.Unit[0]._ReallyEnabled, + ctx->Texture.Unit[0].TexGenEnabled, + rmesa->TexGenNeedNormals[0]); + + if (ctx->RenderMode == GL_RENDER) { + /* Make all this event-driven: + */ + if (ctx->Light.Enabled) { + inputs |= VERT_NORM; + + if (ctx->Light.ColorMaterialEnabled) { + inputs |= VERT_RGBA; + } + } + else { + inputs |= VERT_RGBA; + + if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) { + inputs |= VERT_SPEC_RGB; + } + } + + if (ctx->Texture.Unit[0]._ReallyEnabled) { + if (ctx->Texture.Unit[0].TexGenEnabled) { + if (rmesa->TexGenNeedNormals[0]) { + inputs |= VERT_NORM; + } + } else { + inputs |= VERT_TEX(0); + } + } + + if (ctx->Texture.Unit[1]._ReallyEnabled) { + if (ctx->Texture.Unit[1].TexGenEnabled) { + if (rmesa->TexGenNeedNormals[1]) { + inputs |= VERT_NORM; + } + } else { + inputs |= VERT_TEX(1); + } + } + + stage->inputs = inputs; + stage->active = 1; + } + else + stage->active = 0; +} + +static void r200_init_tcl_render( GLcontext *ctx, + struct gl_pipeline_stage *stage ) +{ + stage->check = r200_check_tcl_render; + stage->check( ctx, stage ); +} + +static void dtr( struct gl_pipeline_stage *stage ) +{ + (void)stage; +} + + +/* Initial state for tcl stage. + */ +const struct gl_pipeline_stage _r200_tcl_stage = +{ + "r200 render", + (_DD_NEW_SEPARATE_SPECULAR | + _NEW_LIGHT| + _NEW_TEXTURE| + _NEW_FOG| + _NEW_RENDERMODE), /* re-check (new inputs) */ + 0, /* re-run (always runs) */ + GL_TRUE, /* active */ + 0, 0, /* inputs (set in check_render), outputs */ + 0, 0, /* changed_inputs, private */ + dtr, /* destructor */ + r200_init_tcl_render, /* check - initially set to alloc data */ + r200_run_tcl_render /* run */ +}; + + + +/**********************************************************************/ +/* Validate state at pipeline start */ +/**********************************************************************/ + + +/*----------------------------------------------------------------------- + * Manage TCL fallbacks + */ + + +static void transition_to_swtnl( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + TNLcontext *tnl = TNL_CONTEXT(ctx); + + R200_NEWPRIM( rmesa ); + rmesa->swtcl.vertex_format = 0; + + r200ChooseVertexState( ctx ); + r200ChooseRenderState( ctx ); + + _mesa_validate_all_lighting_tables( ctx ); + + tnl->Driver.NotifyMaterialChange = + _mesa_validate_all_lighting_tables; + + r200ReleaseArrays( ctx, ~0 ); + + /* Still using the D3D based hardware-rasterizer from the radeon; + * need to put the card into D3D mode to make it work: + */ + R200_STATECHANGE( rmesa, vap ); + rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL] &= ~R200_VAP_TCL_ENABLE; + rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL] |= R200_VAP_D3D_TEX_DEFAULT; + + R200_STATECHANGE( rmesa, vte ); + rmesa->hw.vte.cmd[VTE_SE_VTE_CNTL] &= ~R200_VTX_W0_FMT; + + R200_STATECHANGE( rmesa, set ); + rmesa->hw.set.cmd[SET_RE_CNTL] |= (R200_VTX_STQ0_D3D | + R200_VTX_STQ1_D3D); +} + + +static void transition_to_hwtnl( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + TNLcontext *tnl = TNL_CONTEXT(ctx); + + _tnl_need_projected_coords( ctx, GL_FALSE ); + + r200UpdateMaterial( ctx ); + + tnl->Driver.NotifyMaterialChange = r200UpdateMaterial; + + if ( rmesa->dma.flush ) + rmesa->dma.flush( rmesa ); + + rmesa->dma.flush = 0; + rmesa->swtcl.vertex_format = 0; + + if (rmesa->swtcl.indexed_verts.buf) + r200ReleaseDmaRegion( rmesa, &rmesa->swtcl.indexed_verts, + __FUNCTION__ ); + + R200_STATECHANGE( rmesa, vap ); + rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL] |= R200_VAP_TCL_ENABLE; + rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL] &= ~(R200_VAP_FORCE_W_TO_ONE | + R200_VAP_D3D_TEX_DEFAULT); + + R200_STATECHANGE( rmesa, vte ); + rmesa->hw.vte.cmd[VTE_SE_VTE_CNTL] &= ~(R200_VTX_XY_FMT|R200_VTX_Z_FMT); + rmesa->hw.vte.cmd[VTE_SE_VTE_CNTL] |= R200_VTX_W0_FMT; + + R200_STATECHANGE( rmesa, set ); + rmesa->hw.set.cmd[SET_RE_CNTL] &= ~(R200_VTX_STQ0_D3D | + R200_VTX_STQ1_D3D); + + + if (R200_DEBUG & DEBUG_FALLBACKS) + fprintf(stderr, "R200 end tcl fallback\n"); +} + +static char *fallbackStrings[] = { + "Rasterization fallback", + "Unfilled triangles", + "Twosided lighting, differing materials", + "Materials in VB (maybe between begin/end)", + "Texgen unit 0", + "Texgen unit 1", + "Texgen unit 2", + "User disable" +}; + + +static char *getFallbackString(GLuint bit) +{ + int i = 0; + while (bit > 1) { + i++; + bit >>= 1; + } + return fallbackStrings[i]; +} + + + +void r200TclFallback( GLcontext *ctx, GLuint bit, GLboolean mode ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint oldfallback = rmesa->TclFallback; + + if (mode) { + rmesa->TclFallback |= bit; + if (oldfallback == 0) { + if (R200_DEBUG & DEBUG_FALLBACKS) + fprintf(stderr, "R200 begin tcl fallback %s\n", + getFallbackString( bit )); + transition_to_swtnl( ctx ); + } + } + else { + rmesa->TclFallback &= ~bit; + if (oldfallback == bit) { + if (R200_DEBUG & DEBUG_FALLBACKS) + fprintf(stderr, "R200 end tcl fallback %s\n", + getFallbackString( bit )); + transition_to_hwtnl( ctx ); + } + } +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_tcl.h b/xc/lib/GL/mesa/src/drv/r200/r200_tcl.h new file mode 100644 index 000000000..2920f7eb0 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_tcl.h @@ -0,0 +1,65 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_TCL_H__ +#define __R200_TCL_H__ + +#ifdef GLX_DIRECT_RENDERING + +#include "r200_context.h" + +extern void r200TclPrimitive( GLcontext *ctx, GLenum prim, int hw_prim ); +extern void r200EmitEltPrimitive( GLcontext *ctx, GLuint first, GLuint last, + GLuint flags ); +extern void r200EmitPrimitive( GLcontext *ctx, GLuint first, GLuint last, + GLuint flags ); + +extern void r200TclFallback( GLcontext *ctx, GLuint bit, GLboolean mode ); + +#define R200_TCL_FALLBACK_RASTER 0x1 /* rasterization */ +#define R200_TCL_FALLBACK_UNFILLED 0x2 /* unfilled tris */ +#define R200_TCL_FALLBACK_LIGHT_TWOSIDE 0x4 /* twoside tris */ +#define R200_TCL_FALLBACK_MATERIAL 0x8 /* material in vb */ +#define R200_TCL_FALLBACK_TEXGEN_0 0x10 /* texgen, unit 0 */ +#define R200_TCL_FALLBACK_TEXGEN_1 0x20 /* texgen, unit 1 */ +#define R200_TCL_FALLBACK_TEXGEN_2 0x40 /* texgen, unit 2 */ +#define R200_TCL_FALLBACK_TCL_DISABLE 0x80 /* user disable */ + +#define R200_MAX_TCL_VERTSIZE (4*4) /* using maos now... */ + +#define TCL_FALLBACK( ctx, bit, mode ) r200TclFallback( ctx, bit, mode ) + + +#endif +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_tex.c b/xc/lib/GL/mesa/src/drv/r200/r200_tex.c new file mode 100644 index 000000000..ba7254b3a --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_tex.c @@ -0,0 +1,661 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_state.h" +#include "r200_ioctl.h" +#include "r200_swtcl.h" +#include "r200_tex.h" + +#include "colormac.h" +#include "context.h" +#include "enums.h" +#include "image.h" +#include "mem.h" +#include "mmath.h" +#include "simple_list.h" +#include "texformat.h" +#include "texstore.h" +#include "texutil.h" + + +/* ============================================================= + * Utility functions: + */ + +static void r200SetTexWrap( r200TexObjPtr t, GLenum swrap, GLenum twrap ) +{ + t->pp_txfilter &= ~(R200_CLAMP_S_MASK | R200_CLAMP_T_MASK); + + switch ( swrap ) { + case GL_REPEAT: + t->pp_txfilter |= R200_CLAMP_S_WRAP; + break; + case GL_CLAMP: + t->pp_txfilter |= R200_CLAMP_S_CLAMP_LAST; + break; + case GL_CLAMP_TO_EDGE: + t->pp_txfilter |= R200_CLAMP_S_CLAMP_LAST; + break; + } + + switch ( twrap ) { + case GL_REPEAT: + t->pp_txfilter |= R200_CLAMP_T_WRAP; + break; + case GL_CLAMP: + t->pp_txfilter |= R200_CLAMP_T_CLAMP_LAST; + break; + case GL_CLAMP_TO_EDGE: + t->pp_txfilter |= R200_CLAMP_T_CLAMP_LAST; + break; + } +} + +static void r200SetTexMaxAnisotropy( r200TexObjPtr t, GLfloat max ) +{ + t->pp_txfilter &= ~R200_MAX_ANISO_MASK; + + if ( max == 1.0 ) { + t->pp_txfilter |= R200_MAX_ANISO_1_TO_1; + } else if ( max <= 2.0 ) { + t->pp_txfilter |= R200_MAX_ANISO_2_TO_1; + } else if ( max <= 4.0 ) { + t->pp_txfilter |= R200_MAX_ANISO_4_TO_1; + } else if ( max <= 8.0 ) { + t->pp_txfilter |= R200_MAX_ANISO_8_TO_1; + } else { + t->pp_txfilter |= R200_MAX_ANISO_16_TO_1; + } +} + +static void r200SetTexFilter( r200TexObjPtr t, GLenum minf, GLenum magf ) +{ + GLuint anisotropy = (t->pp_txfilter & R200_MAX_ANISO_MASK); + + t->pp_txfilter &= ~(R200_MIN_FILTER_MASK | R200_MAG_FILTER_MASK); + + if ( anisotropy == R200_MAX_ANISO_1_TO_1 ) { + switch ( minf ) { + case GL_NEAREST: + t->pp_txfilter |= R200_MIN_FILTER_NEAREST; + break; + case GL_LINEAR: + t->pp_txfilter |= R200_MIN_FILTER_LINEAR; + break; + case GL_NEAREST_MIPMAP_NEAREST: + t->pp_txfilter |= R200_MIN_FILTER_NEAREST_MIP_NEAREST; + break; + case GL_NEAREST_MIPMAP_LINEAR: + t->pp_txfilter |= R200_MIN_FILTER_LINEAR_MIP_NEAREST; + break; + case GL_LINEAR_MIPMAP_NEAREST: + t->pp_txfilter |= R200_MIN_FILTER_NEAREST_MIP_LINEAR; + break; + case GL_LINEAR_MIPMAP_LINEAR: + t->pp_txfilter |= R200_MIN_FILTER_LINEAR_MIP_LINEAR; + break; + } + } else { + switch ( minf ) { + case GL_NEAREST: + t->pp_txfilter |= R200_MIN_FILTER_ANISO_NEAREST; + break; + case GL_LINEAR: + t->pp_txfilter |= R200_MIN_FILTER_ANISO_LINEAR; + break; + case GL_NEAREST_MIPMAP_NEAREST: + case GL_LINEAR_MIPMAP_NEAREST: + t->pp_txfilter |= R200_MIN_FILTER_ANISO_NEAREST_MIP_NEAREST; + break; + case GL_NEAREST_MIPMAP_LINEAR: + case GL_LINEAR_MIPMAP_LINEAR: + t->pp_txfilter |= R200_MIN_FILTER_ANISO_NEAREST_MIP_LINEAR; + break; + } + } + + switch ( magf ) { + case GL_NEAREST: + t->pp_txfilter |= R200_MAG_FILTER_NEAREST; + break; + case GL_LINEAR: + t->pp_txfilter |= R200_MAG_FILTER_LINEAR; + break; + } +} + +static void r200SetTexBorderColor( r200TexObjPtr t, GLubyte c[4] ) +{ + t->pp_border_color = r200PackColor( 4, c[0], c[1], c[2], c[3] ); +} + + +static r200TexObjPtr r200AllocTexObj( struct gl_texture_object *texObj ) +{ + r200TexObjPtr t; + + t = CALLOC_STRUCT( r200_tex_obj ); + if (!t) + return NULL; + + if ( R200_DEBUG & DEBUG_TEXTURE ) { + fprintf( stderr, __FUNCTION__"( %p, %p )\n", texObj, t ); + } + + t->tObj = texObj; + make_empty_list( t ); + + /* Initialize non-image-dependent parts of the state: + */ + r200SetTexWrap( t, texObj->WrapS, texObj->WrapT ); + r200SetTexMaxAnisotropy( t, texObj->MaxAnisotropy ); + r200SetTexFilter( t, texObj->MinFilter, texObj->MagFilter ); + r200SetTexBorderColor( t, texObj->BorderColor ); + return t; +} + + +static const struct gl_texture_format * +r200ChooseTextureFormat( GLcontext *ctx, GLint internalFormat, + GLenum format, GLenum type ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + const GLboolean do32bpt = ( rmesa->r200Screen->cpp == 4 ); + + switch ( internalFormat ) { + case 4: + case GL_RGBA: + if ( format == GL_BGRA ) { + if ( type == GL_UNSIGNED_INT_8_8_8_8_REV ) { + return &_mesa_texformat_argb8888; + } + else if ( type == GL_UNSIGNED_SHORT_4_4_4_4_REV ) { + return &_mesa_texformat_argb4444; + } + else if ( type == GL_UNSIGNED_SHORT_1_5_5_5_REV ) { + return &_mesa_texformat_argb1555; + } + } + return do32bpt ? &_mesa_texformat_rgba8888 : &_mesa_texformat_argb4444; + + case 3: + case GL_RGB: + if ( format == GL_RGB && type == GL_UNSIGNED_SHORT_5_6_5 ) { + return &_mesa_texformat_rgb565; + } + return do32bpt ? &_mesa_texformat_rgba8888 : &_mesa_texformat_rgb565; + + case GL_RGBA8: + case GL_RGB10_A2: + case GL_RGBA12: + case GL_RGBA16: + return do32bpt ? &_mesa_texformat_rgba8888 : &_mesa_texformat_argb4444; + + case GL_RGBA4: + case GL_RGBA2: + return &_mesa_texformat_argb4444; + + case GL_RGB5_A1: + return &_mesa_texformat_argb1555; + + case GL_RGB8: + case GL_RGB10: + case GL_RGB12: + case GL_RGB16: + return do32bpt ? &_mesa_texformat_rgba8888 : &_mesa_texformat_rgb565; + + case GL_RGB5: + case GL_RGB4: + case GL_R3_G3_B2: + return &_mesa_texformat_rgb565; + + case GL_ALPHA: + case GL_ALPHA4: + case GL_ALPHA8: + case GL_ALPHA12: + case GL_ALPHA16: + return &_mesa_texformat_al88; + + case 1: + case GL_LUMINANCE: + case GL_LUMINANCE4: + case GL_LUMINANCE8: + case GL_LUMINANCE12: + case GL_LUMINANCE16: + return &_mesa_texformat_al88; + + case 2: + case GL_LUMINANCE_ALPHA: + case GL_LUMINANCE4_ALPHA4: + case GL_LUMINANCE6_ALPHA2: + case GL_LUMINANCE8_ALPHA8: + case GL_LUMINANCE12_ALPHA4: + case GL_LUMINANCE12_ALPHA12: + case GL_LUMINANCE16_ALPHA16: + return &_mesa_texformat_al88; + + case GL_INTENSITY: + case GL_INTENSITY4: + case GL_INTENSITY8: + case GL_INTENSITY12: + case GL_INTENSITY16: + /* At the moment, glean & conform both fail using the i8 internal + * format. + */ + return &_mesa_texformat_al88; +/* return &_mesa_texformat_i8; */ + + default: + _mesa_problem(ctx, "unexpected texture format in r200ChoosTexFormat"); + return NULL; + } + + return NULL; /* never get here */ +} + + +static void r200TexImage1D( GLcontext *ctx, GLenum target, GLint level, + GLint internalFormat, + GLint width, GLint border, + GLenum format, GLenum type, const GLvoid *pixels, + const struct gl_pixelstore_attrib *packing, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + r200TexObjPtr t = (r200TexObjPtr) texObj->DriverData; + + if ( t ) { + r200SwapOutTexObj( rmesa, t ); + } + else { + t = r200AllocTexObj( texObj ); + if (!t) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage1D"); + return; + } + texObj->DriverData = t; + } + + /* Note, this will call r200ChooseTextureFormat */ + _mesa_store_teximage1d(ctx, target, level, internalFormat, + width, border, format, type, pixels, + &ctx->Unpack, texObj, texImage); + + t->dirty_images |= (1 << level); +} + + +static void r200TexSubImage1D( GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, + GLsizei width, + GLenum format, GLenum type, + const GLvoid *pixels, + const struct gl_pixelstore_attrib *packing, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + r200TexObjPtr t = (r200TexObjPtr)texObj->DriverData; + + assert( t ); /* this _should_ be true */ + if ( t ) { + r200SwapOutTexObj( rmesa, t ); + t->dirty_images |= (1 << level); + } + else { + t = r200AllocTexObj(texObj); + if (!t) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage1D"); + return; + } + texObj->DriverData = t; + } + + _mesa_store_texsubimage1d(ctx, target, level, xoffset, width, + format, type, pixels, packing, texObj, + texImage); + + t->dirty_images |= (1 << level); +} + + +static void r200TexImage2D( GLcontext *ctx, GLenum target, GLint level, + GLint internalFormat, + GLint width, GLint height, GLint border, + GLenum format, GLenum type, const GLvoid *pixels, + const struct gl_pixelstore_attrib *packing, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + r200TexObjPtr t = (r200TexObjPtr)texObj->DriverData; + +/* fprintf(stderr, "%s\n", __FUNCTION__); */ + + if ( t ) { + r200SwapOutTexObj( rmesa, t ); + } + else { + t = r200AllocTexObj( texObj ); + if (!t) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexImage2D"); + return; + } + texObj->DriverData = t; + } + + /* Note, this will call r200ChooseTextureFormat */ + _mesa_store_teximage2d(ctx, target, level, internalFormat, + width, height, border, format, type, pixels, + &ctx->Unpack, texObj, texImage); + + t->dirty_images |= (1 << level); +} + + +static void r200TexSubImage2D( GLcontext *ctx, GLenum target, GLint level, + GLint xoffset, GLint yoffset, + GLsizei width, GLsizei height, + GLenum format, GLenum type, + const GLvoid *pixels, + const struct gl_pixelstore_attrib *packing, + struct gl_texture_object *texObj, + struct gl_texture_image *texImage ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + r200TexObjPtr t = (r200TexObjPtr) texObj->DriverData; + +/* fprintf(stderr, "%s\n", __FUNCTION__); */ + + assert( t ); /* this _should_ be true */ + if ( t ) { + r200SwapOutTexObj( rmesa, t ); + } + else { + t = r200AllocTexObj(texObj); + if (!t) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glTexSubImage2D"); + return; + } + texObj->DriverData = t; + } + + _mesa_store_texsubimage2d(ctx, target, level, xoffset, yoffset, width, + height, format, type, pixels, packing, texObj, + texImage); + + t->dirty_images |= (1 << level); +} + + +static void r200TexEnv( GLcontext *ctx, GLenum target, + GLenum pname, const GLfloat *param ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint unit = ctx->Texture.CurrentUnit; + struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; + + if ( R200_DEBUG & DEBUG_STATE ) { + fprintf( stderr, "%s( %s )\n", + __FUNCTION__, _mesa_lookup_enum_by_nr( pname ) ); + } + + switch ( pname ) { + case GL_TEXTURE_ENV_COLOR: { + GLubyte c[4]; + GLuint envColor; + UNCLAMPED_FLOAT_TO_RGBA_CHAN( c, texUnit->EnvColor ); + envColor = r200PackColor( 4, c[0], c[1], c[2], c[3] ); + if ( rmesa->hw.tf.cmd[TF_TFACTOR_0 + unit] != envColor ) { + R200_STATECHANGE( rmesa, tf ); + rmesa->hw.tf.cmd[TF_TFACTOR_0 + unit] = envColor; + } + break; + } + + case GL_TEXTURE_LOD_BIAS_EXT: { + GLfloat bias; + GLuint b; + const int fixed_one = 0x8000000; + + /* The R200's LOD bias is a signed 2's complement value with a + * range of -16.0 <= bias < 16.0. + * + * NOTE: Add a small bias to the bias for conform mipsel.c test. + */ + bias = *param + .01; + bias = CLAMP( bias, -16.0, 16.0 ); + b = (int)(bias * fixed_one) & R200_LOD_BIAS_MASK; + + if ( (rmesa->hw.tex[unit].cmd[TEX_PP_TXFORMAT_X] & R200_LOD_BIAS_MASK) != b ) { + R200_STATECHANGE( rmesa, tex[unit] ); + rmesa->hw.tex[unit].cmd[TEX_PP_TXFORMAT_X] &= ~R200_LOD_BIAS_MASK; + rmesa->hw.tex[unit].cmd[TEX_PP_TXFORMAT_X] |= b; + } + break; + } + + default: + return; + } +} + +static void r200TexParameter( GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj, + GLenum pname, const GLfloat *params ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + r200TexObjPtr t = (r200TexObjPtr) texObj->DriverData; + + if ( R200_DEBUG & (DEBUG_STATE|DEBUG_TEXTURE) ) { + fprintf( stderr, __FUNCTION__"( %s )\n", + _mesa_lookup_enum_by_nr( pname ) ); + } + + if ( ( target != GL_TEXTURE_2D ) && + ( target != GL_TEXTURE_1D ) ) + return; + + switch ( pname ) { + case GL_TEXTURE_MIN_FILTER: + case GL_TEXTURE_MAG_FILTER: + case GL_TEXTURE_MAX_ANISOTROPY_EXT: + r200SetTexMaxAnisotropy( t, texObj->MaxAnisotropy ); + r200SetTexFilter( t, texObj->MinFilter, texObj->MagFilter ); + break; + + case GL_TEXTURE_WRAP_S: + case GL_TEXTURE_WRAP_T: + r200SetTexWrap( t, texObj->WrapS, texObj->WrapT ); + break; + + case GL_TEXTURE_BORDER_COLOR: + r200SetTexBorderColor( t, texObj->BorderColor ); + break; + + case GL_TEXTURE_BASE_LEVEL: + case GL_TEXTURE_MAX_LEVEL: + case GL_TEXTURE_MIN_LOD: + case GL_TEXTURE_MAX_LOD: + /* This isn't the most efficient solution but there doesn't appear to + * be a nice alternative for R200. Since there's no LOD clamping, + * we just have to rely on loading the right subset of mipmap levels + * to simulate a clamped LOD. + */ + r200SwapOutTexObj( rmesa, t ); + break; + + default: + return; + } + + /* Mark this texobj as dirty (one bit per tex unit) + */ + t->dirty_state = TEX_ALL; +} + + + +static void r200BindTexture( GLcontext *ctx, GLenum target, + struct gl_texture_object *texObj ) +{ + r200TexObjPtr t = (r200TexObjPtr) texObj->DriverData; + GLuint unit = ctx->Texture.CurrentUnit; + + if ( R200_DEBUG & (DEBUG_STATE|DEBUG_TEXTURE) ) { + fprintf( stderr, __FUNCTION__"( %p ) unit=%d\n", texObj, unit ); + } + + if ( target == GL_TEXTURE_2D || target == GL_TEXTURE_1D ) { + if ( !t ) { + t = r200AllocTexObj( texObj ); + texObj->DriverData = t; + } + } +} + +static void r200DeleteTexture( GLcontext *ctx, + struct gl_texture_object *texObj ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + r200TexObjPtr t = (r200TexObjPtr) texObj->DriverData; + + if ( R200_DEBUG & (DEBUG_STATE|DEBUG_TEXTURE) ) { + fprintf( stderr, __FUNCTION__"( %p )\n", texObj ); + } + + if ( t ) { + if ( rmesa ) { + R200_FIREVERTICES( rmesa ); + } + r200DestroyTexObj( rmesa, t ); + texObj->DriverData = NULL; + } +} + +static GLboolean r200IsTextureResident( GLcontext *ctx, + struct gl_texture_object *texObj ) +{ + r200TexObjPtr t = (r200TexObjPtr) texObj->DriverData; + + return ( t && t->memBlock ); +} + + +static void r200InitTextureObjects( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + struct gl_texture_object *texObj; + GLuint tmp = ctx->Texture.CurrentUnit; + + ctx->Texture.CurrentUnit = 0; + + texObj = ctx->Texture.Unit[0].Current1D; + r200BindTexture( ctx, GL_TEXTURE_1D, texObj ); + move_to_tail( &rmesa->texture.swapped, + (r200TexObjPtr)texObj->DriverData ); + + texObj = ctx->Texture.Unit[0].Current2D; + r200BindTexture( ctx, GL_TEXTURE_2D, texObj ); + move_to_tail( &rmesa->texture.swapped, + (r200TexObjPtr)texObj->DriverData ); + + ctx->Texture.CurrentUnit = 1; + + texObj = ctx->Texture.Unit[1].Current1D; + r200BindTexture( ctx, GL_TEXTURE_1D, texObj ); + move_to_tail( &rmesa->texture.swapped, + (r200TexObjPtr)texObj->DriverData ); + + texObj = ctx->Texture.Unit[1].Current2D; + r200BindTexture( ctx, GL_TEXTURE_2D, texObj ); + move_to_tail( &rmesa->texture.swapped, + (r200TexObjPtr)texObj->DriverData ); + + ctx->Texture.CurrentUnit = tmp; +} + +/* Need: + * - Same GEN_MODE for all active bits + * - Same EyePlane/ObjPlane for all active bits when using Eye/Obj + * - STRQ presumably all supported (matrix means incoming R values + * can end up in STQ, this has implications for vertex support, + * presumably ok if maos is used, though?) + * + * Basically impossible to do this on the fly - just collect some + * basic info & do the checks from ValidateState(). + */ +static void r200TexGen( GLcontext *ctx, + GLenum coord, + GLenum pname, + const GLfloat *params ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint unit = ctx->Texture.CurrentUnit; + rmesa->recheck_texgen[unit] = GL_TRUE; +} + + +void r200InitTextureFuncs( GLcontext *ctx ) +{ + ctx->Driver.ChooseTextureFormat = r200ChooseTextureFormat; + ctx->Driver.TexImage1D = r200TexImage1D; + ctx->Driver.TexImage2D = r200TexImage2D; + ctx->Driver.TexImage3D = _mesa_store_teximage3d; + ctx->Driver.TexSubImage1D = r200TexSubImage1D; + ctx->Driver.TexSubImage2D = r200TexSubImage2D; + ctx->Driver.TexSubImage3D = _mesa_store_texsubimage3d; + ctx->Driver.CopyTexImage1D = _swrast_copy_teximage1d; + ctx->Driver.CopyTexImage2D = _swrast_copy_teximage2d; + ctx->Driver.CopyTexSubImage1D = _swrast_copy_texsubimage1d; + ctx->Driver.CopyTexSubImage2D = _swrast_copy_texsubimage2d; + ctx->Driver.CopyTexSubImage3D = _swrast_copy_texsubimage3d; + ctx->Driver.TestProxyTexImage = _mesa_test_proxy_teximage; + + ctx->Driver.BindTexture = r200BindTexture; + ctx->Driver.CreateTexture = NULL; /* FIXME: Is this used??? */ + ctx->Driver.DeleteTexture = r200DeleteTexture; + ctx->Driver.IsTextureResident = r200IsTextureResident; + ctx->Driver.PrioritizeTexture = NULL; + ctx->Driver.ActiveTexture = NULL; + ctx->Driver.UpdateTexturePalette = NULL; + + ctx->Driver.TexEnv = r200TexEnv; + ctx->Driver.TexParameter = r200TexParameter; + ctx->Driver.TexGen = r200TexGen; + + r200InitTextureObjects( ctx ); +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_tex.h b/xc/lib/GL/mesa/src/drv/r200/r200_tex.h new file mode 100644 index 000000000..1b3adb8ab --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_tex.h @@ -0,0 +1,55 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_TEX_H__ +#define __R200_TEX_H__ + +#ifdef GLX_DIRECT_RENDERING + +extern void r200UpdateTextureState( GLcontext *ctx ); + +extern int r200UploadTexImages( r200ContextPtr rmesa, r200TexObjPtr t ); + +extern void r200AgeTextures( r200ContextPtr rmesa, int heap ); +extern void r200DestroyTexObj( r200ContextPtr rmesa, r200TexObjPtr t ); +extern void r200SwapOutTexObj( r200ContextPtr rmesa, r200TexObjPtr t ); + +extern void r200PrintLocalLRU( r200ContextPtr rmesa, int heap ); +extern void r200PrintGlobalLRU( r200ContextPtr rmesa, int heap ); +extern void r200UpdateTexLRU( r200ContextPtr rmesa, r200TexObjPtr t ); + +extern void r200InitTextureFuncs( GLcontext *ctx ); + +#endif +#endif /* __R200_TEX_H__ */ diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_texmem.c b/xc/lib/GL/mesa/src/drv/r200/r200_texmem.c new file mode 100644 index 000000000..7ad4287f6 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_texmem.c @@ -0,0 +1,573 @@ +/* $XFree86$ */ +/************************************************************************** + +Copyright (C) Tungsten Graphics 2002. All Rights Reserved. +The Weather Channel, Inc. funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 +license. This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation on the rights to use, copy, modify, merge, publish, +distribute, sub license, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND +NON-INFRINGEMENT. IN NO EVENT SHALL ATI, VA LINUX SYSTEMS AND/OR THEIR +SUPPLIERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER +IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR +IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE +SOFTWARE. + +**************************************************************************/ + +/* + * Authors: + * Kevin E. Martin <martin@valinux.com> + * Gareth Hughes <gareth@valinux.com> + * + */ + +#include "r200_context.h" +#include "r200_state.h" +#include "r200_ioctl.h" +#include "r200_swtcl.h" +#include "r200_tex.h" + +#include "context.h" +#include "colormac.h" +#include "mmath.h" +#include "macros.h" +#include "simple_list.h" +#include "enums.h" +#include "mem.h" + + + +/* Destroy hardware state associated with texture `t'. + */ +void r200DestroyTexObj( r200ContextPtr rmesa, r200TexObjPtr t ) +{ + if ( !t ) + return; + + if ( R200_DEBUG & DEBUG_TEXTURE ) { + fprintf( stderr, __FUNCTION__"( %p, %p )\n", t, t->tObj ); + } + + if ( t->memBlock ) { + mmFreeMem( t->memBlock ); + t->memBlock = NULL; + } + + if ( t->tObj ) + t->tObj->DriverData = NULL; + + if ( rmesa ) { + if ( t == rmesa->state.texture.unit[0].texobj ) { + rmesa->state.texture.unit[0].texobj = NULL; + remove_from_list( &rmesa->hw.tex[0] ); + make_empty_list( &rmesa->hw.tex[0] ); + } + + if ( t == rmesa->state.texture.unit[1].texobj ) { + rmesa->state.texture.unit[1].texobj = NULL; + remove_from_list( &rmesa->hw.tex[1] ); + make_empty_list( &rmesa->hw.tex[1] ); + } + } + + remove_from_list( t ); + FREE( t ); +} + + +/* Keep track of swapped out texture objects. + */ +void r200SwapOutTexObj( r200ContextPtr rmesa, r200TexObjPtr t ) +{ + if ( R200_DEBUG & DEBUG_TEXTURE ) { + fprintf( stderr, __FUNCTION__"( %p, %p )\n", t, t->tObj ); + } + + if ( t->memBlock ) { + mmFreeMem( t->memBlock ); + t->memBlock = NULL; + } + + t->dirty_images = ~0; + move_to_tail( &rmesa->texture.swapped, t ); +} + +/* Print out debugging information about texture LRU. + */ +void r200PrintLocalLRU( r200ContextPtr rmesa, int heap ) +{ + r200TexObjPtr t; + int sz = 1 << (rmesa->r200Screen->logTexGranularity[heap]); + + fprintf( stderr, "\nLocal LRU, heap %d:\n", heap ); + + foreach ( t, &rmesa->texture.objects[heap] ) { + if (!t->tObj) { + fprintf( stderr, "Placeholder %d at 0x%x sz 0x%x\n", + t->memBlock->ofs / sz, + t->memBlock->ofs, + t->memBlock->size ); + } else { + fprintf( stderr, "Texture at 0x%x sz 0x%x\n", + t->memBlock->ofs, + t->memBlock->size ); + } + } + + fprintf( stderr, "\n" ); +} + +void r200PrintGlobalLRU( r200ContextPtr rmesa, int heap ) +{ + radeon_tex_region_t *list = rmesa->sarea->texList[heap]; + int i, j; + + fprintf( stderr, "\nGlobal LRU, heap %d list %p:\n", heap, list ); + + for ( i = 0, j = RADEON_NR_TEX_REGIONS ; i < RADEON_NR_TEX_REGIONS ; i++ ) { + fprintf( stderr, "list[%d] age %d next %d prev %d\n", + j, list[j].age, list[j].next, list[j].prev ); + j = list[j].next; + if ( j == RADEON_NR_TEX_REGIONS ) break; + } + + if ( j != RADEON_NR_TEX_REGIONS ) { + fprintf( stderr, "Loop detected in global LRU\n" ); + for ( i = 0 ; i < RADEON_NR_TEX_REGIONS ; i++ ) { + fprintf( stderr, "list[%d] age %d next %d prev %d\n", + i, list[i].age, list[i].next, list[i].prev ); + } + } + + fprintf( stderr, "\n" ); +} + +/* Reset the global texture LRU. + */ +static void r200ResetGlobalLRU( r200ContextPtr rmesa, int heap ) +{ + radeon_tex_region_t *list = rmesa->sarea->texList[heap]; + int sz = 1 << rmesa->r200Screen->logTexGranularity[heap]; + int i; + + /* + * (Re)initialize the global circular LRU list. The last element in + * the array (RADEON_NR_TEX_REGIONS) is the sentinal. Keeping it at + * the end of the array allows it to be addressed rationally when + * looking up objects at a particular location in texture memory. + */ + for ( i = 0 ; (i+1) * sz <= rmesa->r200Screen->texSize[heap] ; i++ ) { + list[i].prev = i-1; + list[i].next = i+1; + list[i].age = 0; + } + + i--; + list[0].prev = RADEON_NR_TEX_REGIONS; + list[i].prev = i-1; + list[i].next = RADEON_NR_TEX_REGIONS; + list[RADEON_NR_TEX_REGIONS].prev = i; + list[RADEON_NR_TEX_REGIONS].next = 0; + rmesa->sarea->texAge[heap] = 0; +} + +/* Update the local and glock texture LRUs. + */ +void r200UpdateTexLRU(r200ContextPtr rmesa, r200TexObjPtr t ) +{ + int heap = t->heap; + radeon_tex_region_t *list = rmesa->sarea->texList[heap]; + int sz = rmesa->r200Screen->logTexGranularity[heap]; + int start = t->memBlock->ofs >> sz; + int end = (t->memBlock->ofs + t->memBlock->size-1) >> sz; + int i; + + rmesa->texture.age[heap] = ++rmesa->sarea->texAge[heap]; + + if ( !t->memBlock ) { + fprintf( stderr, "no memblock\n\n" ); + return; + } + + /* Update our local LRU */ + move_to_head( &rmesa->texture.objects[heap], t ); + + /* Update the global LRU */ + for ( i = start ; i <= end ; i++ ) { + list[i].in_use = 1; + list[i].age = rmesa->texture.age[heap]; + + /* remove_from_list(i) */ + list[(CARD32)list[i].next].prev = list[i].prev; + list[(CARD32)list[i].prev].next = list[i].next; + + /* insert_at_head(list, i) */ + list[i].prev = RADEON_NR_TEX_REGIONS; + list[i].next = list[RADEON_NR_TEX_REGIONS].next; + list[(CARD32)list[RADEON_NR_TEX_REGIONS].next].prev = i; + list[RADEON_NR_TEX_REGIONS].next = i; + } + + if ( 0 ) { + r200PrintGlobalLRU( rmesa, t->heap ); + r200PrintLocalLRU( rmesa, t->heap ); + } +} + +/* Update our notion of what textures have been changed since we last + * held the lock. This pertains to both our local textures and the + * textures belonging to other clients. Keep track of other client's + * textures by pushing a placeholder texture onto the LRU list -- these + * are denoted by (tObj == NULL). + */ +static void r200TexturesGone( r200ContextPtr rmesa, int heap, + int offset, int size, int in_use ) +{ + r200TexObjPtr t, tmp; + + foreach_s ( t, tmp, &rmesa->texture.objects[heap] ) { + if ( t->memBlock->ofs >= offset + size || + t->memBlock->ofs + t->memBlock->size <= offset ) + continue; + + /* It overlaps - kick it out. Need to hold onto the currently + * bound objects, however. + */ + r200SwapOutTexObj( rmesa, t ); + } + + if ( in_use ) { + t = (r200TexObjPtr) CALLOC( sizeof(*t) ); + if ( !t ) return; + + t->memBlock = mmAllocMem( rmesa->texture.heap[heap], size, 0, offset ); + if ( !t->memBlock ) { + fprintf( stderr, "Couldn't alloc placeholder sz %x ofs %x\n", + (int)size, (int)offset ); + mmDumpMemInfo( rmesa->texture.heap[heap] ); + return; + } + insert_at_head( &rmesa->texture.objects[heap], t ); + } +} + +/* Update our client's shared texture state. If another client has + * modified a region in which we have textures, then we need to figure + * out which of our textures has been removed, and update our global + * LRU. + */ +void r200AgeTextures( r200ContextPtr rmesa, int heap ) +{ + RADEONSAREAPrivPtr sarea = rmesa->sarea; + + if ( sarea->texAge[heap] != rmesa->texture.age[heap] ) { + int sz = 1 << rmesa->r200Screen->logTexGranularity[heap]; + int nr = 0; + int idx; + + for ( idx = sarea->texList[heap][RADEON_NR_TEX_REGIONS].prev ; + idx != RADEON_NR_TEX_REGIONS && nr < RADEON_NR_TEX_REGIONS ; + idx = sarea->texList[heap][idx].prev, nr++ ) + { + /* If switching texturing schemes, then the SAREA might not + * have been properly cleared, so we need to reset the + * global texture LRU. + */ + if ( idx * sz > rmesa->r200Screen->texSize[heap] ) { + nr = RADEON_NR_TEX_REGIONS; + break; + } + + if ( sarea->texList[heap][idx].age > rmesa->texture.age[heap] ) { + r200TexturesGone( rmesa, heap, idx * sz, sz, + sarea->texList[heap][idx].in_use ); + } + } + + if ( nr == RADEON_NR_TEX_REGIONS ) { + r200TexturesGone( rmesa, heap, 0, + rmesa->r200Screen->texSize[heap], 0 ); + r200ResetGlobalLRU( rmesa, heap ); + } + + rmesa->texture.age[heap] = sarea->texAge[heap]; + } +} + + +/* ============================================================= + * Texture image conversions + */ + +/* Upload the texture image associated with texture `t' at level `level' + * at the address relative to `start'. + */ +static void r200UploadSubImage( r200ContextPtr rmesa, + r200TexObjPtr t, + GLint hwlevel, + GLint x, GLint y, GLint width, GLint height ) +{ + struct gl_texture_image *texImage; + const struct gl_texture_format *texFormat; + GLint texelsPerDword = 0; + GLuint format, pitch, offset; + GLint imageWidth, imageHeight; + GLint ret; + drmRadeonTexture tex; + drmRadeonTexImage tmp; + int level = hwlevel + t->firstLevel; + + if ( R200_DEBUG & DEBUG_TEXTURE ) { + fprintf( stderr, "%s level %d %dx%d\n", __FUNCTION__, + level, width, height); + } + + /* Ensure we have a valid texture to upload */ + if ( ( hwlevel < 0 ) || ( hwlevel >= RADEON_MAX_TEXTURE_LEVELS ) ) { + _mesa_problem(NULL, "bad texture level in r200UploadSubimage"); + return; + } + + texImage = t->tObj->Image[level]; + if ( !texImage ) { + if ( R200_DEBUG & DEBUG_TEXTURE ) + fprintf( stderr, __FUNCTION__ ": texImage %d is NULL!\n", level ); + return; + } + if ( !texImage->Data ) { + if ( R200_DEBUG & DEBUG_TEXTURE ) + fprintf( stderr, __FUNCTION__ ": image data is NULL!\n" ); + return; + } + + texFormat = texImage->TexFormat; + + switch ( texFormat->TexelBytes ) { + case 1: + texelsPerDword = 4; + break; + case 2: + texelsPerDword = 2; + break; + case 4: + texelsPerDword = 1; + break; + } + + format = t->pp_txformat & R200_TXFORMAT_FORMAT_MASK; + + imageWidth = texImage->Width; + imageHeight = texImage->Height; + + offset = t->bufAddr; + pitch = (t->image[0].width * texFormat->TexelBytes) / 64; + +#if 0 + /* Bump the performace counter */ + rmesa->c_textureBytes += (dwords << 2); +#endif + + if ( R200_DEBUG & (DEBUG_TEXTURE|DEBUG_IOCTL) ) + { + GLint imageX = 0; + GLint imageY = 0; + GLint blitX = t->image[hwlevel].x; + GLint blitY = t->image[hwlevel].y; + GLint blitWidth = t->image[hwlevel].width; + GLint blitHeight = t->image[hwlevel].height; + fprintf( stderr, " upload image: %d,%d at %d,%d\n", + imageWidth, imageHeight, imageX, imageY ); + fprintf( stderr, " upload blit: %d,%d at %d,%d\n", + blitWidth, blitHeight, blitX, blitY ); + fprintf( stderr, " blit ofs: 0x%07x pitch: 0x%x " + "level: %d/%d format: %x\n", + (GLuint)offset, (GLuint)pitch, hwlevel, level, format ); + } + + t->image[hwlevel].data = texImage->Data; + + tex.offset = offset; + tex.pitch = pitch; + tex.format = format; + tex.width = imageWidth; + tex.height = imageHeight; + tex.image = &tmp; + + memcpy( &tmp, &t->image[hwlevel], sizeof(drmRadeonTexImage) ); + + do { + ret = drmCommandWriteRead( rmesa->dri.fd, DRM_RADEON_TEXTURE, + &tex, sizeof(drmRadeonTexture) ); + if (ret) { + if (R200_DEBUG & DEBUG_IOCTL) + fprintf(stderr, "DRM_RADEON_TEXTURE: again!\n"); + usleep(1); + } + } while ( ret && errno == EAGAIN ); + + if ( ret ) { + UNLOCK_HARDWARE( rmesa ); + fprintf( stderr, "DRM_R200_TEXTURE: return = %d\n", ret ); + fprintf( stderr, " offset=0x%08x pitch=0x%x format=%d\n", + offset, pitch, format ); + fprintf( stderr, " image width=%d height=%d\n", + imageWidth, imageHeight ); + fprintf( stderr, " blit width=%d height=%d data=%p\n", + t->image[hwlevel].width, t->image[hwlevel].height, + t->image[hwlevel].data ); + exit( 1 ); + } +} + +/* Upload the texture images associated with texture `t'. This might + * require removing our own and/or other client's texture objects to + * make room for these images. + */ +int r200UploadTexImages( r200ContextPtr rmesa, r200TexObjPtr t ) +{ + const int numLevels = t->lastLevel - t->firstLevel + 1; + int heap; + r200TexObjPtr t0 = rmesa->state.texture.unit[0].texobj; + r200TexObjPtr t1 = rmesa->state.texture.unit[1].texobj; + + if ( R200_DEBUG & (DEBUG_TEXTURE|DEBUG_IOCTL) ) { + fprintf( stderr, __FUNCTION__"( %p, %p ) sz=%d lvls=%d-%d\n", + rmesa->glCtx, t->tObj, t->totalSize, + t->firstLevel, t->lastLevel ); + } + + if ( !t || t->totalSize == 0 ) + return 0; + + if (R200_DEBUG & DEBUG_SYNC) { + fprintf(stderr, "\nSyncing\n\n"); + R200_FIREVERTICES( rmesa ); + r200Finish( rmesa->glCtx ); + } + + LOCK_HARDWARE( rmesa ); + + /* Choose the heap appropriately */ + heap = t->heap = RADEON_CARD_HEAP; +#if 0 + if ( !rmesa->r200Screen->IsPCI && + t->totalSize > rmesa->r200Screen->texSize[heap] ) { + heap = t->heap = RADEON_AGP_HEAP; + } +#endif + + /* Do we need to eject LRU texture objects? */ + if ( !t->memBlock ) { + /* Allocate a memory block on a 4k boundary (1<<12 == 4096) */ + t->memBlock = mmAllocMem( rmesa->texture.heap[heap], + t->totalSize, 12, 0 ); + +#if 0 + /* Try AGP before kicking anything out of local mem */ + if ( !t->memBlock && heap == R200_CARD_HEAP ) { + t->memBlock = mmAllocMem( rmesa->texture.heap[R200_AGP_HEAP], + t->totalSize, 12, 0 ); + + if ( t->memBlock ) + heap = t->heap = R200_AGP_HEAP; + } +#endif + + /* Kick out textures until the requested texture fits */ + while ( !t->memBlock ) { + if ( rmesa->texture.objects[heap].prev == t0 || + rmesa->texture.objects[heap].prev == t1 ) { + fprintf( stderr, + "r200UploadTexImages: ran into bound texture\n" ); + UNLOCK_HARDWARE( rmesa ); + return -1; + } + if ( rmesa->texture.objects[heap].prev == + &rmesa->texture.objects[heap] ) { + if ( rmesa->r200Screen->IsPCI ) { + fprintf( stderr, "r200UploadTexImages: upload texture " + "failure on local texture heaps, sz=%d\n", + t->totalSize ); + UNLOCK_HARDWARE( rmesa ); + return -1; +#if 0 + } else if ( heap == R200_CARD_HEAP ) { + heap = t->heap = R200_AGP_HEAP; + continue; +#endif + } else { + fprintf( stderr, "r200UploadTexImages: upload texture " + "failure on both local and AGP texture heaps, " + "sz=%d\n", + t->totalSize ); + UNLOCK_HARDWARE( rmesa ); + return -1; + } + } + + r200SwapOutTexObj( rmesa, rmesa->texture.objects[heap].prev ); + + t->memBlock = mmAllocMem( rmesa->texture.heap[heap], + t->totalSize, 12, 0 ); + } + + /* Set the base offset of the texture image */ + t->bufAddr = rmesa->r200Screen->texOffset[heap] + t->memBlock->ofs; + t->pp_txoffset = t->bufAddr; + +#if 0 + /* Fix AGP texture offsets */ + if ( heap == RADEON_AGP_HEAP ) { + t->pp_txoffset -= rmesa->r200Screen->texOffset[heap]; + t->pp_txoffset += rmesa->dri.agp_texture_offset; + t->bufAddr = t->pp_txoffset; + } +#endif + + /* Mark this texobj as dirty on all units: + */ + t->dirty_state = TEX_ALL; + } + + /* Let the world know we've used this memory recently */ + r200UpdateTexLRU( rmesa, t ); + + /* Upload any images that are new */ + if (t->dirty_images) { + int hwlevel; + for ( hwlevel = 0 ; hwlevel < numLevels ; hwlevel++ ) { + if ( t->dirty_images & (1 << (hwlevel+t->firstLevel)) ) { + r200UploadSubImage( rmesa, t, hwlevel, + 0, 0, + t->image[hwlevel].width, + t->image[hwlevel].height ); + } + } + t->dirty_images = 0; + } + + + UNLOCK_HARDWARE( rmesa ); + + if (R200_DEBUG & DEBUG_SYNC) { + fprintf(stderr, "\nSyncing\n\n"); + r200Finish( rmesa->glCtx ); + } + + return 0; +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_texstate.c b/xc/lib/GL/mesa/src/drv/r200/r200_texstate.c new file mode 100644 index 000000000..4d663db5c --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_texstate.c @@ -0,0 +1,1388 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_state.h" +#include "r200_ioctl.h" +#include "r200_swtcl.h" +#include "r200_tex.h" +#include "r200_tcl.h" + +#include "colormac.h" +#include "context.h" +#include "enums.h" +#include "macros.h" +#include "mem.h" +#include "mmath.h" +#include "simple_list.h" +#include "texformat.h" + + +static void r200SetTexImages( r200ContextPtr rmesa, + struct gl_texture_object *tObj ) +{ + r200TexObjPtr t = (r200TexObjPtr)tObj->DriverData; + const struct gl_texture_image *baseImage = tObj->Image[tObj->BaseLevel]; + GLint totalSize; + GLint texelsPerDword = 0, blitWidth = 0, blitPitch = 0; + GLint x, y, width, height; + GLint i; + GLint firstLevel=0, lastLevel=0, numLevels; + GLint log2Width, log2Height; + GLuint txformat = 0; + + /* Set the hardware texture format + */ + switch (baseImage->TexFormat->MesaFormat) { + case MESA_FORMAT_I8: + txformat = R200_TXFORMAT_I8; + break; + case MESA_FORMAT_AL88: + txformat = R200_TXFORMAT_AI88; + break; + case MESA_FORMAT_RGBA8888: + txformat = R200_TXFORMAT_RGBA8888; + break; + case MESA_FORMAT_ARGB8888: + txformat = R200_TXFORMAT_ARGB8888; + break; + case MESA_FORMAT_RGB565: + txformat = R200_TXFORMAT_RGB565; + break; + case MESA_FORMAT_ARGB1555: + txformat = R200_TXFORMAT_ARGB1555; + break; + case MESA_FORMAT_ARGB4444: + txformat = R200_TXFORMAT_ARGB4444; + break; + default: + _mesa_problem(NULL, "unexpected texture format in r200TexImage2D"); + return; + } + + t->pp_txformat &= ~(R200_TXFORMAT_FORMAT_MASK | + R200_TXFORMAT_ALPHA_IN_MAP); + t->pp_txformat |= txformat; + + if ( txformat == R200_TXFORMAT_RGBA8888 || + txformat == R200_TXFORMAT_ARGB4444 || + txformat == R200_TXFORMAT_ARGB1555 || + txformat == R200_TXFORMAT_AI88 ) { + t->pp_txformat |= R200_TXFORMAT_ALPHA_IN_MAP; + } + + /* The R200 has a 64-byte minimum pitch for all blits. We + * calculate the equivalent number of texels to simplify the + * calculation of the texture image area. + */ + switch ( baseImage->TexFormat->TexelBytes ) { + case 1: + texelsPerDword = 4; + blitPitch = 64; + break; + case 2: + texelsPerDword = 2; + blitPitch = 32; + break; + case 4: + texelsPerDword = 1; + blitPitch = 16; + break; + } + + /* Select the larger of the two widths for our global texture image + * coordinate space. As the R200 has very strict offset rules, we + * can't upload mipmaps directly and have to reference their location + * from the aligned start of the whole image. + */ + blitWidth = MAX2( baseImage->Width, blitPitch ); + + /* Calculate mipmap offsets and dimensions. + */ + totalSize = 0; + x = 0; + y = 0; + + /* Compute which mipmap levels we really want to send to the hardware. + * This depends on the base image size, GL_TEXTURE_MIN_LOD, + * GL_TEXTURE_MAX_LOD, GL_TEXTURE_BASE_LEVEL, and GL_TEXTURE_MAX_LEVEL. + * Yes, this looks overly complicated, but it's all needed. + */ +/* fprintf(stderr, */ +/* "\nBaseLevel %d MinLod %f MaxLod %f MaxLevel %d\n", */ +/* tObj->BaseLevel, tObj->MinLod, tObj->MaxLod, */ +/* tObj->MaxLevel); */ + + + firstLevel = tObj->BaseLevel + (GLint)(tObj->MinLod + 0.5); + firstLevel = MAX2(firstLevel, tObj->BaseLevel); + lastLevel = tObj->BaseLevel + (GLint)(tObj->MaxLod + 0.5); + lastLevel = MAX2(lastLevel, tObj->BaseLevel); + lastLevel = MIN2(lastLevel, tObj->BaseLevel + baseImage->MaxLog2); + lastLevel = MIN2(lastLevel, tObj->MaxLevel); + lastLevel = MAX2(firstLevel, lastLevel); /* need at least one level */ + + /* save these values */ + t->firstLevel = firstLevel; + t->lastLevel = lastLevel; + + + numLevels = lastLevel - firstLevel + 1; + + log2Width = tObj->Image[firstLevel]->WidthLog2; + log2Height = tObj->Image[firstLevel]->HeightLog2; + + for ( i = 0 ; i < numLevels ; i++ ) { + const struct gl_texture_image *texImage; + GLuint size; + + texImage = tObj->Image[i + firstLevel]; + if ( !texImage ) + break; + + width = texImage->Width; + height = texImage->Height; + + /* Texture images have a minimum pitch of 32 bytes (half of the + * 64-byte minimum pitch for blits). For images that have a + * width smaller than this, we must pad each texture image + * scanline out to this amount. + */ + if ( width < blitPitch / 2 ) { + width = blitPitch / 2; + } + + size = width * height * baseImage->TexFormat->TexelBytes; + totalSize += size; + assert( (totalSize & 31) == 0 ); + + while ( width < blitWidth && height > 1 ) { + width *= 2; + height /= 2; + } + + assert(i < RADEON_MAX_TEXTURE_LEVELS); + t->image[i].x = x; + t->image[i].y = y; + + t->image[i].width = width; + t->image[i].height = height; + + /* While blits must have a pitch of at least 64 bytes, mipmaps + * must be aligned on a 32-byte boundary (just like each texture + * image scanline). + */ + if ( width >= blitWidth ) { + y += height; + } else { + x += width; + if ( x >= blitWidth ) { + x = 0; + y++; + } + } + + if ( 0 ) + fprintf( stderr, "level=%d p=%d %dx%d -> %dx%d at (%d,%d)\n", + i, blitWidth, baseImage->Width, baseImage->Height, + t->image[i].width, t->image[i].height, + t->image[i].x, t->image[i].y ); + } + + /* Align the total size of texture memory block. + */ + t->totalSize = (totalSize + RADEON_OFFSET_MASK) & ~RADEON_OFFSET_MASK; + + /* Hardware state: + */ + t->pp_txfilter &= ~R200_MAX_MIP_LEVEL_MASK; + t->pp_txfilter |= (numLevels - 1) << R200_MAX_MIP_LEVEL_SHIFT; + + t->pp_txformat &= ~(R200_TXFORMAT_WIDTH_MASK | + R200_TXFORMAT_HEIGHT_MASK); + t->pp_txformat |= ((log2Width << R200_TXFORMAT_WIDTH_SHIFT) | + (log2Height << R200_TXFORMAT_HEIGHT_SHIFT)); + + t->dirty_state = TEX_ALL; + + r200UploadTexImages( rmesa, t ); +} + + + +/* ================================================================ + * Texture combine functions + */ + +#define R200_DISABLE 0 +#define R200_REPLACE 1 +#define R200_MODULATE 2 +#define R200_DECAL 3 +#define R200_BLEND 4 +#define R200_ADD 5 +#define R200_MAX_COMBFUNC 6 + +static GLuint r200_color_combine[][R200_MAX_COMBFUNC] = +{ + /* Unit 0: + */ + { + /* Disable combiner stage + */ + (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_DIFFUSE_COLOR | + R200_TXC_OP_MADD), + + /* GL_REPLACE = 0x00802800 + */ + (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_MADD), + + /* GL_MODULATE = 0x00800142 + */ + (R200_TXC_ARG_A_DIFFUSE_COLOR | /* current starts in DIFFUSE */ + R200_TXC_ARG_B_R0_COLOR | + R200_TXC_ARG_C_ZERO | + R200_TXC_OP_MADD), + + /* GL_DECAL = 0x008c2d42 + */ + (R200_TXC_ARG_A_DIFFUSE_COLOR | + R200_TXC_ARG_B_R0_COLOR | + R200_TXC_ARG_C_R0_ALPHA | + R200_TXC_OP_LERP), + + /* GL_BLEND = 0x008c2902 + */ + (R200_TXC_ARG_A_DIFFUSE_COLOR | + R200_TXC_ARG_B_TFACTOR_COLOR | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_LERP), + + /* GL_ADD = 0x00812802 + */ + (R200_TXC_ARG_A_DIFFUSE_COLOR | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_COMP_ARG_B | + R200_TXC_OP_MADD), + }, + + /* Unit 1: + */ + { + /* Disable combiner stage + */ + (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_MADD), + + /* GL_REPLACE = 0x00803000 + */ + (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R1_COLOR | + R200_TXC_OP_MADD), + + /* GL_MODULATE = 0x00800182 + */ + (R200_TXC_ARG_A_R0_COLOR | /* current in R0 thereafter */ + R200_TXC_ARG_B_R1_COLOR | + R200_TXC_ARG_C_ZERO | + R200_TXC_OP_MADD), + + /* GL_DECAL = 0x008c3582 + */ + (R200_TXC_ARG_A_R0_COLOR | + R200_TXC_ARG_B_R1_COLOR | + R200_TXC_ARG_C_R1_ALPHA | + R200_TXC_OP_LERP), + + /* GL_BLEND = 0x008c3102 + */ + (R200_TXC_ARG_A_R0_COLOR | + R200_TXC_ARG_B_TFACTOR_COLOR | + R200_TXC_ARG_C_R1_COLOR | + R200_TXC_OP_LERP), + + /* GL_ADD = 0x00813002 + */ + (R200_TXC_ARG_A_R0_COLOR | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R1_COLOR | + R200_TXC_COMP_ARG_B | + R200_TXC_OP_MADD), + }, + + /* Unit 2: + */ + { + /* Disable combiner stage + */ + (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R0_COLOR | + R200_TXC_OP_MADD), + + /* GL_REPLACE = 0x00803800 + */ + (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R2_COLOR | + R200_TXC_OP_MADD), + + /* GL_MODULATE = 0x008001c2 + */ + (R200_TXC_ARG_A_R0_COLOR | + R200_TXC_ARG_B_R2_COLOR | + R200_TXC_ARG_C_ZERO | + R200_TXC_OP_MADD), + + /* GL_DECAL = 0x008c3dc2 + */ + (R200_TXC_ARG_A_R0_COLOR | + R200_TXC_ARG_B_R2_COLOR | + R200_TXC_ARG_C_R2_ALPHA | + R200_TXC_OP_LERP), + + /* GL_BLEND = 0x008c3902 + */ + (R200_TXC_ARG_A_R0_COLOR | + R200_TXC_ARG_B_TFACTOR_COLOR | + R200_TXC_ARG_C_R2_COLOR | + R200_TXC_OP_LERP), + + /* GL_ADD = 0x00813802 + */ + (R200_TXC_ARG_A_R0_COLOR | + R200_TXC_ARG_B_ZERO | + R200_TXC_ARG_C_R2_COLOR | + R200_TXC_COMP_ARG_B | + R200_TXC_OP_MADD), + } +}; + +static GLuint r200_alpha_combine[][R200_MAX_COMBFUNC] = +{ + /* Unit 0: + */ + { + /* Disable combiner stage + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_DIFFUSE_ALPHA | + R200_TXA_OP_MADD), + + + /* GL_REPLACE = 0x00800500 + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD), + + /* GL_MODULATE = 0x00800051 + */ + (R200_TXA_ARG_A_DIFFUSE_ALPHA | + R200_TXA_ARG_B_R0_ALPHA | + R200_TXA_ARG_C_ZERO | + R200_TXA_OP_MADD), + + /* GL_DECAL = 0x00800100 + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_DIFFUSE_ALPHA | + R200_TXA_OP_MADD), + + /* GL_BLEND = 0x00800051 + */ + (R200_TXA_ARG_A_DIFFUSE_ALPHA | + R200_TXA_ARG_B_TFACTOR_ALPHA | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_LERP), + + /* GL_ADD = 0x00800051 + */ + (R200_TXA_ARG_A_DIFFUSE_ALPHA | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXC_COMP_ARG_B | + R200_TXA_OP_MADD), + }, + + /* Unit 1: + */ + { + /* Disable combiner stage + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD), + + /* GL_REPLACE = 0x00800600 + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R1_ALPHA | + R200_TXA_OP_MADD), + + /* GL_MODULATE = 0x00800061 + */ + (R200_TXA_ARG_A_R0_ALPHA | + R200_TXA_ARG_B_R1_ALPHA | + R200_TXA_ARG_C_ZERO | + R200_TXA_OP_MADD), + + /* GL_DECAL = 0x00800100 + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD), + + /* GL_BLEND = 0x00800061 + */ + (R200_TXA_ARG_A_R0_ALPHA | + R200_TXA_ARG_B_TFACTOR_ALPHA | + R200_TXA_ARG_C_R1_ALPHA | + R200_TXA_OP_LERP), + + /* GL_ADD = 0x00800061 + */ + (R200_TXA_ARG_A_R0_ALPHA | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R1_ALPHA | + R200_TXC_COMP_ARG_B | + R200_TXA_OP_MADD), + }, + + /* Unit 2: + */ + { + /* Disable combiner stage + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD), + + /* GL_REPLACE = 0x00800700 + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R2_ALPHA | + R200_TXA_OP_MADD), + + /* GL_MODULATE = 0x00800071 + */ + (R200_TXA_ARG_A_R0_ALPHA | + R200_TXA_ARG_B_R2_ALPHA | + R200_TXA_ARG_C_ZERO | + R200_TXA_OP_MADD), + + /* GL_DECAL = 0x00800100 + */ + (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R0_ALPHA | + R200_TXA_OP_MADD), + + /* GL_BLEND = 0x00800071 + */ + (R200_TXA_ARG_A_R0_ALPHA | + R200_TXA_ARG_B_TFACTOR_ALPHA | + R200_TXA_ARG_C_R2_ALPHA | + R200_TXA_OP_LERP), + + /* GL_ADD = 0x00800021 + */ + (R200_TXA_ARG_A_R0_ALPHA | + R200_TXA_ARG_B_ZERO | + R200_TXA_ARG_C_R2_ALPHA | + R200_TXC_COMP_ARG_B | + R200_TXA_OP_MADD), + } +}; + + +/* GL_ARB_texture_env_combine support + */ + +/* The color tables have combine functions for GL_SRC_COLOR, + * GL_ONE_MINUS_SRC_COLOR, GL_SRC_ALPHA and GL_ONE_MINUS_SRC_ALPHA. + */ +static GLuint r200_register_color[][R200_MAX_TEXTURE_UNITS] = +{ + { + R200_TXC_ARG_A_R0_COLOR, + R200_TXC_ARG_A_R1_COLOR, + R200_TXC_ARG_A_R2_COLOR + }, + { + R200_TXC_ARG_A_R0_COLOR | R200_TXC_COMP_ARG_A, + R200_TXC_ARG_A_R1_COLOR | R200_TXC_COMP_ARG_A, + R200_TXC_ARG_A_R2_COLOR | R200_TXC_COMP_ARG_A + }, + { + R200_TXC_ARG_A_R0_ALPHA, + R200_TXC_ARG_A_R1_ALPHA, + R200_TXC_ARG_A_R2_ALPHA + }, + { + R200_TXC_ARG_A_R0_ALPHA | R200_TXC_COMP_ARG_A, + R200_TXC_ARG_A_R1_ALPHA | R200_TXC_COMP_ARG_A, + R200_TXC_ARG_A_R2_ALPHA | R200_TXC_COMP_ARG_A + }, +}; + +static GLuint r200_tfactor_color[] = +{ + R200_TXC_ARG_A_TFACTOR_COLOR, + R200_TXC_ARG_A_TFACTOR_COLOR | R200_TXC_COMP_ARG_A, + R200_TXC_ARG_A_TFACTOR_ALPHA, + R200_TXC_ARG_A_TFACTOR_ALPHA | R200_TXC_COMP_ARG_A +}; + +static GLuint r200_primary_color[] = +{ + R200_TXC_ARG_A_DIFFUSE_COLOR, + R200_TXC_ARG_A_DIFFUSE_COLOR | R200_TXC_COMP_ARG_A, + R200_TXC_ARG_A_DIFFUSE_ALPHA, + R200_TXC_ARG_A_DIFFUSE_ALPHA | R200_TXC_COMP_ARG_A +}; + + +/* The alpha tables only have GL_SRC_ALPHA and GL_ONE_MINUS_SRC_ALPHA. + */ +static GLuint r200_register_alpha[][R200_MAX_TEXTURE_UNITS] = +{ + { + R200_TXA_ARG_A_R0_ALPHA, + R200_TXA_ARG_A_R1_ALPHA, + R200_TXA_ARG_A_R2_ALPHA + }, + { + R200_TXA_ARG_A_R0_ALPHA | R200_TXC_COMP_ARG_A, + R200_TXA_ARG_A_R1_ALPHA | R200_TXC_COMP_ARG_A, + R200_TXA_ARG_A_R2_ALPHA | R200_TXC_COMP_ARG_A + }, +}; + +static GLuint r200_tfactor_alpha[] = +{ + R200_TXA_ARG_A_TFACTOR_ALPHA, + R200_TXA_ARG_A_TFACTOR_ALPHA | R200_TXC_COMP_ARG_A +}; + +static GLuint r200_primary_alpha[] = +{ + R200_TXA_ARG_A_DIFFUSE_ALPHA, + R200_TXA_ARG_A_DIFFUSE_ALPHA | R200_TXC_COMP_ARG_A +}; + + + +/* Extract the arg from slot A, shift it into the correct argument slot + * and set the corresponding complement bit. + */ +#define R200_COLOR_ARG( n, arg ) \ +do { \ + color_combine |= \ + ((color_arg[n] & R200_TXC_ARG_A_MASK) \ + << R200_TXC_ARG_##arg##_SHIFT); \ + color_combine |= \ + ((color_arg[n] >> R200_TXC_COMP_ARG_A_SHIFT) \ + << R200_TXC_COMP_ARG_##arg##_SHIFT); \ +} while (0) + +#define R200_ALPHA_ARG( n, arg ) \ +do { \ + alpha_combine |= \ + ((alpha_arg[n] & R200_TXA_ARG_A_MASK) \ + << R200_TXA_ARG_##arg##_SHIFT); \ + alpha_combine |= \ + ((alpha_arg[n] >> R200_TXA_COMP_ARG_A_SHIFT) \ + << R200_TXA_COMP_ARG_##arg##_SHIFT); \ +} while (0) + + +/* ================================================================ + * Texture unit state management + */ + +static void r200UpdateTextureEnv( GLcontext *ctx, int unit ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; + GLuint color_combine, alpha_combine; + GLuint color_scale = rmesa->hw.pix[unit].cmd[PIX_PP_TXCBLEND2]; + GLuint alpha_scale = rmesa->hw.pix[unit].cmd[PIX_PP_TXABLEND2]; + + if ( R200_DEBUG & DEBUG_TEXTURE ) { + fprintf( stderr, __FUNCTION__"( %p, %d )\n", ctx, unit ); + } + + /* Set the texture environment state. Isn't this nice and clean? + * The R200 will automagically set the texture alpha to 0xff when + * the texture format does not include an alpha component. This + * reduces the amount of special-casing we have to do, alpha-only + * textures being a notable exception. + */ + if ( !texUnit->_ReallyEnabled ) { + /* Don't cache these results. + */ + rmesa->state.texture.unit[unit].format = 0; + rmesa->state.texture.unit[unit].envMode = 0; + color_combine = r200_color_combine[unit][R200_DISABLE]; + alpha_combine = r200_alpha_combine[unit][R200_DISABLE]; + } + else { + const struct gl_texture_object *tObj = texUnit->_Current; + const GLenum format = tObj->Image[tObj->BaseLevel]->Format; + GLuint color_arg[3], alpha_arg[3]; + GLuint i, numColorArgs = 0, numAlphaArgs = 0; + + switch ( texUnit->EnvMode ) { + case GL_REPLACE: + switch ( format ) { + case GL_RGBA: + case GL_LUMINANCE_ALPHA: + case GL_INTENSITY: + color_combine = r200_color_combine[unit][R200_REPLACE]; + alpha_combine = r200_alpha_combine[unit][R200_REPLACE]; + break; + case GL_ALPHA: + color_combine = r200_color_combine[unit][R200_DISABLE]; + alpha_combine = r200_alpha_combine[unit][R200_REPLACE]; + break; + case GL_LUMINANCE: + case GL_RGB: + color_combine = r200_color_combine[unit][R200_REPLACE]; + alpha_combine = r200_alpha_combine[unit][R200_DISABLE]; + break; + case GL_COLOR_INDEX: + default: + return; + } + break; + + case GL_MODULATE: + switch ( format ) { + case GL_RGBA: + case GL_LUMINANCE_ALPHA: + case GL_INTENSITY: + color_combine = r200_color_combine[unit][R200_MODULATE]; + alpha_combine = r200_alpha_combine[unit][R200_MODULATE]; + break; + case GL_ALPHA: + color_combine = r200_color_combine[unit][R200_DISABLE]; + alpha_combine = r200_alpha_combine[unit][R200_MODULATE]; + break; + case GL_RGB: + case GL_LUMINANCE: + color_combine = r200_color_combine[unit][R200_MODULATE]; + alpha_combine = r200_alpha_combine[unit][R200_DISABLE]; + break; + case GL_COLOR_INDEX: + default: + return; + } + break; + + case GL_DECAL: + switch ( format ) { + case GL_RGBA: + case GL_RGB: + color_combine = r200_color_combine[unit][R200_DECAL]; + alpha_combine = r200_alpha_combine[unit][R200_DISABLE]; + break; + case GL_ALPHA: + case GL_LUMINANCE: + case GL_LUMINANCE_ALPHA: + case GL_INTENSITY: + color_combine = r200_color_combine[unit][R200_DISABLE]; + alpha_combine = r200_alpha_combine[unit][R200_DISABLE]; + break; + case GL_COLOR_INDEX: + default: + return; + } + break; + + case GL_BLEND: + switch ( format ) { + case GL_RGBA: + case GL_RGB: + case GL_LUMINANCE: + case GL_LUMINANCE_ALPHA: + color_combine = r200_color_combine[unit][R200_BLEND]; + alpha_combine = r200_alpha_combine[unit][R200_MODULATE]; + break; + case GL_ALPHA: + color_combine = r200_color_combine[unit][R200_DISABLE]; + alpha_combine = r200_alpha_combine[unit][R200_MODULATE]; + break; + case GL_INTENSITY: + color_combine = r200_color_combine[unit][R200_BLEND]; + alpha_combine = r200_alpha_combine[unit][R200_BLEND]; + break; + case GL_COLOR_INDEX: + default: + return; + } + break; + + case GL_ADD: + switch ( format ) { + case GL_RGBA: + case GL_RGB: + case GL_LUMINANCE: + case GL_LUMINANCE_ALPHA: + color_combine = r200_color_combine[unit][R200_ADD]; + alpha_combine = r200_alpha_combine[unit][R200_MODULATE]; + break; + case GL_ALPHA: + color_combine = r200_color_combine[unit][R200_DISABLE]; + alpha_combine = r200_alpha_combine[unit][R200_MODULATE]; + break; + case GL_INTENSITY: + color_combine = r200_color_combine[unit][R200_ADD]; + alpha_combine = r200_alpha_combine[unit][R200_ADD]; + break; + case GL_COLOR_INDEX: + default: + return; + } + break; + + case GL_COMBINE: + /* Don't cache these results. + */ + rmesa->state.texture.unit[unit].format = 0; + rmesa->state.texture.unit[unit].envMode = 0; + + /* Step 0: + * Calculate how many arguments we need to process. + */ + switch ( texUnit->CombineModeRGB ) { + case GL_REPLACE: + numColorArgs = 1; + break; + case GL_MODULATE: + case GL_ADD: + case GL_ADD_SIGNED: + case GL_SUBTRACT: + case GL_DOT3_RGB: + case GL_DOT3_RGBA: + numColorArgs = 2; + break; + case GL_INTERPOLATE: + numColorArgs = 3; + break; + default: + return; + } + + switch ( texUnit->CombineModeA ) { + case GL_REPLACE: + numAlphaArgs = 1; + break; + case GL_SUBTRACT: + case GL_MODULATE: + case GL_ADD: + case GL_ADD_SIGNED: + numAlphaArgs = 2; + break; + case GL_INTERPOLATE: + numAlphaArgs = 3; + break; + default: + return; + } + + /* Step 1: + * Extract the color and alpha combine function arguments. + */ + for ( i = 0 ; i < numColorArgs ; i++ ) { + const GLuint op = texUnit->CombineOperandRGB[i] - GL_SRC_COLOR; + assert(op >= 0); + assert(op <= 3); + switch ( texUnit->CombineSourceRGB[i] ) { + case GL_TEXTURE: + color_arg[i] = r200_register_color[op][unit]; + break; + case GL_CONSTANT: + color_arg[i] = r200_tfactor_color[op]; + break; + case GL_PRIMARY_COLOR: + color_arg[i] = r200_primary_color[op]; + break; + case GL_PREVIOUS: + if (unit == 0) + color_arg[i] = r200_primary_color[op]; + else + color_arg[i] = r200_register_color[op][0]; + break; + default: + return; + } + } + + for ( i = 0 ; i < numAlphaArgs ; i++ ) { + const GLuint op = texUnit->CombineOperandA[i] - GL_SRC_ALPHA; + assert(op >= 0); + assert(op <= 1); + switch ( texUnit->CombineSourceA[i] ) { + case GL_TEXTURE: + alpha_arg[i] = r200_register_alpha[op][unit]; + break; + case GL_CONSTANT: + alpha_arg[i] = r200_tfactor_alpha[op]; + break; + case GL_PRIMARY_COLOR: + alpha_arg[i] = r200_primary_alpha[op]; + break; + case GL_PREVIOUS: + if (unit == 0) + alpha_arg[i] = r200_primary_alpha[op]; + else + alpha_arg[i] = r200_register_alpha[op][0]; + break; + default: + return; + } + } + + /* Step 2: + * Build up the color and alpha combine functions. + */ + switch ( texUnit->CombineModeRGB ) { + case GL_REPLACE: + color_combine = (R200_TXC_ARG_A_ZERO | + R200_TXC_ARG_B_ZERO | + R200_TXC_OP_MADD); + R200_COLOR_ARG( 0, C ); + break; + case GL_MODULATE: + color_combine = (R200_TXC_ARG_C_ZERO | + R200_TXC_OP_MADD); + R200_COLOR_ARG( 0, A ); + R200_COLOR_ARG( 1, B ); + break; + case GL_ADD: + color_combine = (R200_TXC_ARG_B_ZERO | + R200_TXC_COMP_ARG_B | + R200_TXC_OP_MADD); + R200_COLOR_ARG( 0, A ); + R200_COLOR_ARG( 1, C ); + break; + case GL_SUBTRACT: + color_combine = (R200_TXC_ARG_B_ZERO | + R200_TXC_COMP_ARG_B | + R200_TXC_NEG_ARG_C | + R200_TXC_OP_MADD); + R200_COLOR_ARG( 0, A ); + R200_COLOR_ARG( 1, C ); + break; + case GL_ADD_SIGNED: + color_combine = (R200_TXC_ARG_B_ZERO | + R200_TXC_COMP_ARG_B | + R200_TXC_BIAS_ARG_C | /* new */ + R200_TXC_OP_MADD); /* was ADDSIGNED */ + R200_COLOR_ARG( 0, A ); + R200_COLOR_ARG( 1, C ); + break; + case GL_INTERPOLATE: + color_combine = (R200_TXC_OP_LERP); + R200_COLOR_ARG( 0, B ); + R200_COLOR_ARG( 1, A ); + R200_COLOR_ARG( 2, C ); + break; + case GL_DOT3_RGB: + case GL_DOT3_RGBA: + color_combine = (R200_TXC_ARG_C_ZERO | + R200_TXC_OP_DOT3); + R200_COLOR_ARG( 0, A ); + R200_COLOR_ARG( 1, B ); + break; + default: + return; + } + + switch ( texUnit->CombineModeA ) { + case GL_REPLACE: + alpha_combine = (R200_TXA_ARG_A_ZERO | + R200_TXA_ARG_B_ZERO | + R200_TXA_OP_MADD); + R200_ALPHA_ARG( 0, C ); + break; + case GL_MODULATE: + alpha_combine = (R200_TXA_ARG_C_ZERO | + R200_TXA_OP_MADD); + R200_ALPHA_ARG( 0, A ); + R200_ALPHA_ARG( 1, B ); + break; + case GL_ADD: + alpha_combine = (R200_TXA_ARG_B_ZERO | + R200_TXC_COMP_ARG_B | + R200_TXA_OP_MADD); + R200_ALPHA_ARG( 0, A ); + R200_ALPHA_ARG( 1, C ); + break; + case GL_SUBTRACT: + alpha_combine = (R200_TXA_ARG_B_ZERO | + R200_TXC_COMP_ARG_B | + R200_TXC_NEG_ARG_C | + R200_TXA_OP_MADD); + R200_ALPHA_ARG( 0, A ); + R200_ALPHA_ARG( 1, C ); + break; + case GL_ADD_SIGNED: + alpha_combine = (R200_TXA_ARG_B_ZERO | + R200_TXC_COMP_ARG_B | + R200_TXC_BIAS_ARG_C | /* new */ + R200_TXA_OP_MADD); /* was ADDSIGNED */ + R200_ALPHA_ARG( 0, A ); + R200_ALPHA_ARG( 1, C ); + break; + case GL_INTERPOLATE: + alpha_combine = (R200_TXA_OP_LERP); + R200_ALPHA_ARG( 0, B ); + R200_ALPHA_ARG( 1, A ); + R200_ALPHA_ARG( 2, C ); + break; + default: + return; + } + + if ( texUnit->CombineModeRGB == GL_DOT3_RGB ) { + alpha_combine |= R200_TXA_DOT_ALPHA; + } + + /* Step 3: + * Apply the scale factor. The EXT extension has a somewhat + * unnecessary restriction that the scale must be 4x. The ARB + * extension will likely drop this and we can just apply the + * scale factors regardless. + */ + color_scale &= ~R200_TXC_SCALE_MASK; + alpha_scale &= ~R200_TXA_SCALE_MASK; + if ( texUnit->CombineModeRGB != GL_DOT3_RGB && + texUnit->CombineModeRGB != GL_DOT3_RGBA ) { + color_scale |= texUnit->CombineScaleShiftRGB<< R200_TXC_SCALE_SHIFT; + alpha_scale |= texUnit->CombineScaleShiftA << R200_TXA_SCALE_SHIFT; + } + else + { + color_scale |= R200_TXC_SCALE_4X; + alpha_scale |= R200_TXA_SCALE_4X; + } + + /* All done! + */ + break; + + default: + return; + } + } + + if ( rmesa->hw.pix[unit].cmd[PIX_PP_TXCBLEND] != color_combine || + rmesa->hw.pix[unit].cmd[PIX_PP_TXABLEND] != alpha_combine || + rmesa->hw.pix[unit].cmd[PIX_PP_TXCBLEND2] != color_scale || + rmesa->hw.pix[unit].cmd[PIX_PP_TXABLEND2] != alpha_scale) { + R200_STATECHANGE( rmesa, pix[unit] ); + rmesa->hw.pix[unit].cmd[PIX_PP_TXCBLEND] = color_combine; + rmesa->hw.pix[unit].cmd[PIX_PP_TXABLEND] = alpha_combine; + rmesa->hw.pix[unit].cmd[PIX_PP_TXCBLEND2] = color_scale; + rmesa->hw.pix[unit].cmd[PIX_PP_TXABLEND2] = alpha_scale; + } +} + +#define TEXOBJ_TXFILTER_MASK (R200_MAX_MIP_LEVEL_MASK | \ + R200_MIN_FILTER_MASK | \ + R200_MAG_FILTER_MASK | \ + R200_MAX_ANISO_MASK | \ + R200_CLAMP_S_MASK | \ + R200_CLAMP_T_MASK) + +#define TEXOBJ_TXFORMAT_MASK (R200_TXFORMAT_WIDTH_MASK | \ + R200_TXFORMAT_HEIGHT_MASK | \ + R200_TXFORMAT_FORMAT_MASK | \ + R200_TXFORMAT_ALPHA_IN_MAP) + + + +static void import_tex_obj_state( r200ContextPtr rmesa, + int unit, + r200TexObjPtr texobj ) +{ + GLuint *cmd = R200_DB_STATE( tex[unit] ); + + cmd[TEX_PP_TXFILTER] &= ~TEXOBJ_TXFILTER_MASK; + cmd[TEX_PP_TXFILTER] |= texobj->pp_txfilter & TEXOBJ_TXFILTER_MASK; + cmd[TEX_PP_TXFORMAT] &= ~TEXOBJ_TXFORMAT_MASK; + cmd[TEX_PP_TXFORMAT] |= texobj->pp_txformat & TEXOBJ_TXFORMAT_MASK; + cmd[TEX_PP_TXSIZE] = texobj->pp_txsize; /* NPOT only! */ + cmd[TEX_PP_TXPITCH] = texobj->pp_txpitch; /* NPOT only! */ + cmd[TEX_PP_TXOFFSET] = texobj->pp_txoffset; + cmd[TEX_PP_BORDER_COLOR] = texobj->pp_border_color; + texobj->dirty_state &= ~(1<<unit); + + R200_DB_STATECHANGE( rmesa, &rmesa->hw.tex[unit] ); +} + + + + +static void set_texgen_matrix( r200ContextPtr rmesa, + GLuint unit, + GLfloat *s_plane, + GLfloat *t_plane ) +{ + static const GLfloat scale_identity[4] = { 1,1,1,1 }; + + if (!TEST_EQ_4V( s_plane, scale_identity) || + !(TEST_EQ_4V( t_plane, scale_identity))) { + rmesa->TexGenEnabled |= R200_TEXMAT_0_ENABLE<<unit; + rmesa->TexGenMatrix[unit].m[0] = s_plane[0]; + rmesa->TexGenMatrix[unit].m[4] = s_plane[1]; + rmesa->TexGenMatrix[unit].m[8] = s_plane[2]; + rmesa->TexGenMatrix[unit].m[12] = s_plane[3]; + + rmesa->TexGenMatrix[unit].m[1] = t_plane[0]; + rmesa->TexGenMatrix[unit].m[5] = t_plane[1]; + rmesa->TexGenMatrix[unit].m[9] = t_plane[2]; + rmesa->TexGenMatrix[unit].m[13] = t_plane[3]; + rmesa->NewGLState |= _NEW_TEXTURE_MATRIX; + } +} + +/* Ignoring the Q texcoord for now. + * + * Returns GL_FALSE if fallback required. + */ +static GLboolean r200_validate_texgen( GLcontext *ctx, GLuint unit ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; + GLuint inputshift = R200_TEXGEN_0_INPUT_SHIFT + unit*4; + GLuint tmp = rmesa->TexGenEnabled; + + rmesa->TexGenCompSel &= ~(R200_OUTPUT_TEX_0 << unit); + rmesa->TexGenEnabled &= ~(R200_TEXGEN_TEXMAT_0_ENABLE<<unit); + rmesa->TexGenEnabled &= ~(R200_TEXMAT_0_ENABLE<<unit); + rmesa->TexGenInputs &= ~(R200_TEXGEN_INPUT_MASK<<inputshift); + rmesa->TexGenNeedNormals[unit] = 0; + + if (0) + fprintf(stderr, "%s unit %d\n", __FUNCTION__, unit); + + if ((texUnit->TexGenEnabled & (S_BIT|T_BIT)) == 0) { + /* Disabled, no fallback: + */ + rmesa->TexGenInputs |= + (R200_TEXGEN_INPUT_TEXCOORD_0+unit) << inputshift; + return GL_TRUE; + } + else if (texUnit->TexGenEnabled & Q_BIT) { + /* Very easy to do this, in fact would remove a fallback case + * elsewhere, but I haven't done it yet... Fallback: + */ + fprintf(stderr, "fallback Q_BIT\n"); + return GL_FALSE; + } + else if ((texUnit->TexGenEnabled & (S_BIT|T_BIT)) != (S_BIT|T_BIT) || + texUnit->GenModeS != texUnit->GenModeT) { + /* Mixed modes, fallback: + */ +/* fprintf(stderr, "fallback mixed texgen\n"); */ + return GL_FALSE; + } + else + rmesa->TexGenEnabled |= R200_TEXGEN_TEXMAT_0_ENABLE << unit; + + switch (texUnit->GenModeS) { + case GL_OBJECT_LINEAR: + rmesa->TexGenInputs |= R200_TEXGEN_INPUT_OBJ << inputshift; + set_texgen_matrix( rmesa, unit, + texUnit->ObjectPlaneS, + texUnit->ObjectPlaneT); + break; + + case GL_EYE_LINEAR: + rmesa->TexGenInputs |= R200_TEXGEN_INPUT_EYE << inputshift; + set_texgen_matrix( rmesa, unit, + texUnit->EyePlaneS, + texUnit->EyePlaneT); + break; + + case GL_REFLECTION_MAP_NV: + rmesa->TexGenNeedNormals[unit] = GL_TRUE; + rmesa->TexGenInputs |= R200_TEXGEN_INPUT_EYE_REFLECT<<inputshift; + break; + + case GL_NORMAL_MAP_NV: + rmesa->TexGenNeedNormals[unit] = GL_TRUE; + rmesa->TexGenInputs |= R200_TEXGEN_INPUT_EYE_NORMAL<<inputshift; + break; + + case GL_SPHERE_MAP: + rmesa->TexGenNeedNormals[unit] = GL_TRUE; + rmesa->TexGenInputs |= R200_TEXGEN_INPUT_SPHERE<<inputshift; + break; + + default: + /* Unsupported mode, fallback: + */ + /* fprintf(stderr, "fallback unsupported texgen\n"); */ + return GL_FALSE; + } + + rmesa->TexGenCompSel |= R200_OUTPUT_TEX_0 << unit; + + if (tmp != rmesa->TexGenEnabled) { + rmesa->NewGLState |= _NEW_TEXTURE_MATRIX; + } + + if (0) + fprintf(stderr, "%s unit %d neednormals %d\n", __FUNCTION__, unit, + rmesa->TexGenNeedNormals[unit]); + + return GL_TRUE; +} + + + + +static GLboolean r200UpdateTextureUnit( GLcontext *ctx, int unit ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit]; + + if ( texUnit->_ReallyEnabled & (TEXTURE0_1D|TEXTURE0_2D) ) { + struct gl_texture_object *tObj = texUnit->_Current; + r200TexObjPtr t = (r200TexObjPtr) tObj->DriverData; + GLenum format; + + /* Fallback if there's a texture border */ + if ( tObj->Image[tObj->BaseLevel]->Border > 0 ) + return GL_FALSE; + + /* Upload teximages (not pipelined) + */ + if ( t->dirty_images ) { + R200_FIREVERTICES( rmesa ); + r200SetTexImages( rmesa, tObj ); + /* Fallback if we can't upload: + */ + if ( !t->memBlock ) + return GL_FALSE; + } + + /* Update state if this is a different texture object to last + * time. + */ + if ( rmesa->state.texture.unit[unit].texobj != t ) { + rmesa->state.texture.unit[unit].texobj = t; + t->dirty_state |= 1<<unit; + r200UpdateTexLRU( rmesa, t ); /* XXX: should be locked! */ + } + + + /* Newly enabled? + */ + if ( 1|| !(rmesa->hw.ctx.cmd[CTX_PP_CNTL] & (R200_TEX_0_ENABLE<<unit))) { + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= (R200_TEX_0_ENABLE | + R200_TEX_BLEND_0_ENABLE) << unit; + + R200_STATECHANGE( rmesa, vtx ); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_1] |= 4 << (unit * 3); + + rmesa->recheck_texgen[unit] = GL_TRUE; + } + + if (t->dirty_state & (1<<unit)) { + import_tex_obj_state( rmesa, unit, t ); + } + + if (rmesa->recheck_texgen[unit]) { + GLboolean fallback = !r200_validate_texgen( ctx, unit ); + TCL_FALLBACK( ctx, (R200_TCL_FALLBACK_TEXGEN_0<<unit), fallback); + rmesa->recheck_texgen[unit] = 0; + rmesa->NewGLState |= _NEW_TEXTURE_MATRIX; + } + + format = tObj->Image[tObj->BaseLevel]->Format; + if ( rmesa->state.texture.unit[unit].format != format || + rmesa->state.texture.unit[unit].envMode != texUnit->EnvMode ) { + rmesa->state.texture.unit[unit].format = format; + rmesa->state.texture.unit[unit].envMode = texUnit->EnvMode; + r200UpdateTextureEnv( ctx, unit ); + } + + } + else if ( texUnit->_ReallyEnabled ) { + /* 3d textures, etc: + */ + return GL_FALSE; + } + else if (rmesa->hw.ctx.cmd[CTX_PP_CNTL] & (R200_TEX_0_ENABLE<<unit)) { + /* Texture unit disabled */ + rmesa->state.texture.unit[unit].texobj = 0; + R200_STATECHANGE( rmesa, ctx ); + rmesa->hw.ctx.cmd[CTX_PP_CNTL] &= ~((R200_TEX_0_ENABLE | + R200_TEX_BLEND_0_ENABLE) << unit); + rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= R200_TEX_BLEND_0_ENABLE; + + R200_STATECHANGE( rmesa, tcl ); + rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_1] &= ~(7 << (unit * 3)); + + if (rmesa->TclFallback & (R200_TCL_FALLBACK_TEXGEN_0<<unit)) { + TCL_FALLBACK( ctx, (R200_TCL_FALLBACK_TEXGEN_0<<unit), GL_FALSE); + } + + /* Actually want to keep all units less than max active texture + * enabled, right? Fix this for >2 texunits. + */ + if (unit == 0) + r200UpdateTextureEnv( ctx, unit ); + + + { + GLuint inputshift = R200_TEXGEN_0_INPUT_SHIFT + unit*4; + GLuint tmp = rmesa->TexGenEnabled; + + rmesa->TexGenEnabled &= ~(R200_TEXGEN_TEXMAT_0_ENABLE<<unit); + rmesa->TexGenEnabled &= ~(R200_TEXMAT_0_ENABLE<<unit); + rmesa->TexGenEnabled &= ~(R200_TEXGEN_INPUT_MASK<<inputshift); + rmesa->TexGenNeedNormals[unit] = 0; + rmesa->TexGenCompSel &= ~(R200_OUTPUT_TEX_0 << unit); + rmesa->TexGenInputs &= ~(R200_TEXGEN_INPUT_MASK<<inputshift); + + if (tmp != rmesa->TexGenEnabled) { + rmesa->recheck_texgen[unit] = GL_TRUE; + rmesa->NewGLState |= _NEW_TEXTURE_MATRIX; + } + } + } + + return GL_TRUE; +} + +void r200UpdateTextureState( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLboolean ok; + GLuint dbg; + + ok = (r200UpdateTextureUnit( ctx, 0 ) && + r200UpdateTextureUnit( ctx, 1 )); + + FALLBACK( rmesa, R200_FALLBACK_TEXTURE, !ok ); + + if (rmesa->TclFallback) + r200ChooseVertexState( ctx ); + + /* + * T0 hang workaround ------------- + */ +#if 1 + if ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_ENABLE_MASK) == R200_TEX_0_ENABLE && + (rmesa->hw.tex[0].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK) > R200_MIN_FILTER_LINEAR) { + + R200_STATECHANGE(rmesa, ctx); + R200_STATECHANGE(rmesa, tex[1]); + rmesa->hw.ctx.cmd[CTX_PP_CNTL] |= R200_TEX_1_ENABLE; + rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] &= ~TEXOBJ_TXFORMAT_MASK; + rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] |= 0x08000000; + } + else { + if ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_1_ENABLE) && + (rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] & 0x08000000)) { + R200_STATECHANGE(rmesa, tex[1]); + rmesa->hw.tex[1].cmd[TEX_PP_TXFORMAT] &= ~0x08000000; + } + } +#endif + +#if 1 + /* + * Texture cache LRU hang workaround ------------- + */ + dbg = 0x0; + if (((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_0_ENABLE) && + ((((rmesa->hw.tex[0].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & + 0x04) == 0))) + { + dbg |= 0x02; + } + + if (((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_1_ENABLE) && + ((((rmesa->hw.tex[1].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & + 0x04) == 0))) + { + dbg |= 0x04; + } + + if (dbg != rmesa->hw.tam.cmd[TAM_DEBUG3]) { + R200_STATECHANGE( rmesa, tam ); + rmesa->hw.tam.cmd[TAM_DEBUG3] = dbg; + if (0) printf("TEXCACHE LRU HANG WORKAROUND %x\n", dbg); + } +#endif +} + +/* + also tests for higher texunits: + + ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_2_ENABLE) && + ((((rmesa->hw.tex[2].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 0x04) == 0)) || + ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_4_ENABLE) && + ((((rmesa->hw.tex[4].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 0x04) == 0))) + + ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_3_ENABLE) && + ((((rmesa->hw.tex[3].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 0x04) == 0)) || + ((rmesa->hw.ctx.cmd[CTX_PP_CNTL] & R200_TEX_5_ENABLE) && + ((((rmesa->hw.tex[5].cmd[TEX_PP_TXFILTER] & R200_MIN_FILTER_MASK)) & 0x04) == 0))) + +*/ diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt.c b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt.c new file mode 100644 index 000000000..cf786a252 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt.c @@ -0,0 +1,1136 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "r200_context.h" +#include "r200_state.h" +#include "r200_ioctl.h" +#include "r200_tex.h" +#include "r200_tcl.h" +#include "r200_vtxfmt.h" + +#include "api_noop.h" +#include "api_arrayelt.h" +#include "context.h" +#include "mem.h" +#include "mmath.h" +#include "mtypes.h" +#include "enums.h" +#include "glapi.h" +#include "colormac.h" +#include "light.h" +#include "state.h" +#include "vtxfmt.h" + +#include "tnl/tnl.h" +#include "tnl/t_context.h" +#include "tnl/t_array_api.h" + +struct r200_vb vb; + +static void r200FlushVertices( GLcontext *, GLuint ); + +static void count_func( const char *name, struct dynfn *l ) +{ + int i = 0; + struct dynfn *f; + foreach (f, l) i++; + if (i) fprintf(stderr, "%s: %d\n", name, i ); +} + +static void count_funcs( r200ContextPtr rmesa ) +{ + count_func( "Vertex2f", &rmesa->vb.dfn_cache.Vertex2f ); + count_func( "Vertex2fv", &rmesa->vb.dfn_cache.Vertex2fv ); + count_func( "Vertex3f", &rmesa->vb.dfn_cache.Vertex3f ); + count_func( "Vertex3fv", &rmesa->vb.dfn_cache.Vertex3fv ); + count_func( "Color4ub", &rmesa->vb.dfn_cache.Color4ub ); + count_func( "Color4ubv", &rmesa->vb.dfn_cache.Color4ubv ); + count_func( "Color3ub", &rmesa->vb.dfn_cache.Color3ub ); + count_func( "Color3ubv", &rmesa->vb.dfn_cache.Color3ubv ); + count_func( "Color4f", &rmesa->vb.dfn_cache.Color4f ); + count_func( "Color4fv", &rmesa->vb.dfn_cache.Color4fv ); + count_func( "Color3f", &rmesa->vb.dfn_cache.Color3f ); + count_func( "Color3fv", &rmesa->vb.dfn_cache.Color3fv ); + count_func( "SecondaryColor3f", &rmesa->vb.dfn_cache.SecondaryColor3fEXT ); + count_func( "SecondaryColor3fv", &rmesa->vb.dfn_cache.SecondaryColor3fvEXT ); + count_func( "SecondaryColor3ub", &rmesa->vb.dfn_cache.SecondaryColor3ubEXT ); + count_func( "SecondaryColor3ubv", &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT ); + count_func( "Normal3f", &rmesa->vb.dfn_cache.Normal3f ); + count_func( "Normal3fv", &rmesa->vb.dfn_cache.Normal3fv ); + count_func( "TexCoord2f", &rmesa->vb.dfn_cache.TexCoord2f ); + count_func( "TexCoord2fv", &rmesa->vb.dfn_cache.TexCoord2fv ); + count_func( "TexCoord1f", &rmesa->vb.dfn_cache.TexCoord1f ); + count_func( "TexCoord1fv", &rmesa->vb.dfn_cache.TexCoord1fv ); + count_func( "MultiTexCoord2fARB", &rmesa->vb.dfn_cache.MultiTexCoord2fARB ); + count_func( "MultiTexCoord2fvARB", &rmesa->vb.dfn_cache.MultiTexCoord2fvARB ); + count_func( "MultiTexCoord1fARB", &rmesa->vb.dfn_cache.MultiTexCoord1fARB ); + count_func( "MultiTexCoord1fvARB", &rmesa->vb.dfn_cache.MultiTexCoord1fvARB ); +} + + +void r200_copy_to_current( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s\n", __FUNCTION__); + + assert(ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT); + assert(vb.context == ctx); + + if (rmesa->vb.vtxfmt_0 & R200_VTX_N0) { + ctx->Current.Normal[0] = vb.normalptr[0]; + ctx->Current.Normal[1] = vb.normalptr[1]; + ctx->Current.Normal[2] = vb.normalptr[2]; + } + + switch( VTX_COLOR(rmesa->vb.vtxfmt_0, 0) ) { + case R200_VTX_PK_RGBA: + ctx->Current.Color[0] = UBYTE_TO_FLOAT( vb.ubytecolorptr[0] ); + ctx->Current.Color[1] = UBYTE_TO_FLOAT( vb.ubytecolorptr[1] ); + ctx->Current.Color[2] = UBYTE_TO_FLOAT( vb.ubytecolorptr[2] ); + ctx->Current.Color[3] = UBYTE_TO_FLOAT( vb.ubytecolorptr[3] ); + break; + + case R200_VTX_FP_RGB: + ctx->Current.Color[0] = vb.floatcolorptr[0]; + ctx->Current.Color[1] = vb.floatcolorptr[1]; + ctx->Current.Color[2] = vb.floatcolorptr[2]; + break; + + case R200_VTX_FP_RGBA: + ctx->Current.Color[0] = vb.floatcolorptr[0]; + ctx->Current.Color[1] = vb.floatcolorptr[1]; + ctx->Current.Color[2] = vb.floatcolorptr[2]; + ctx->Current.Color[3] = vb.floatcolorptr[3]; + break; + + default: + break; + } + + if (VTX_COLOR(rmesa->vb.vtxfmt_0, 1) == R200_VTX_PK_RGBA) { + ctx->Current.SecondaryColor[0] = UBYTE_TO_FLOAT( vb.ubytespecptr[0] ); + ctx->Current.SecondaryColor[1] = UBYTE_TO_FLOAT( vb.ubytespecptr[1] ); + ctx->Current.SecondaryColor[2] = UBYTE_TO_FLOAT( vb.ubytespecptr[2] ); + } + + if (rmesa->vb.vtxfmt_1 & (7 << R200_VTX_TEX0_COMP_CNT_SHIFT)) { + ctx->Current.Texcoord[0][0] = vb.texcoordptr[0][0]; + ctx->Current.Texcoord[0][1] = vb.texcoordptr[0][1]; + ctx->Current.Texcoord[0][2] = 0.0F; + ctx->Current.Texcoord[0][3] = 1.0F; + } + + if (rmesa->vb.vtxfmt_1 & (7 << R200_VTX_TEX1_COMP_CNT_SHIFT)) { + ctx->Current.Texcoord[1][0] = vb.texcoordptr[1][0]; + ctx->Current.Texcoord[1][1] = vb.texcoordptr[1][1]; + ctx->Current.Texcoord[1][2] = 0.0F; + ctx->Current.Texcoord[1][3] = 1.0F; + } + + ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT; +} + +static GLboolean discreet_gl_prim[GL_POLYGON+1] = { + 1, /* 0 points */ + 1, /* 1 lines */ + 0, /* 2 line_strip */ + 0, /* 3 line_loop */ + 1, /* 4 tris */ + 0, /* 5 tri_fan */ + 0, /* 6 tri_strip */ + 1, /* 7 quads */ + 0, /* 8 quadstrip */ + 0, /* 9 poly */ +}; + +static void flush_prims( r200ContextPtr rmesa ) +{ + int i,j; + struct r200_dma_region tmp = rmesa->dma.current; + + tmp.buf->refcount++; + tmp.aos_size = vb.vertex_size; + tmp.aos_stride = vb.vertex_size; + tmp.aos_start = GET_START(&tmp); + + rmesa->dma.current.ptr = rmesa->dma.current.start += + (vb.initial_counter - vb.counter) * vb.vertex_size * 4; + + rmesa->tcl.vertex_format = rmesa->vb.vtxfmt_0; + rmesa->tcl.aos_components[0] = &tmp; + rmesa->tcl.nr_aos_components = 1; + rmesa->dma.flush = 0; + + /* Optimize the primitive list: + */ + if (rmesa->vb.nrprims > 1) { + for (j = 0, i = 1 ; i < rmesa->vb.nrprims; i++) { + int pj = rmesa->vb.primlist[j].prim & 0xf; + int pi = rmesa->vb.primlist[i].prim & 0xf; + + if (pj == pi && discreet_gl_prim[pj] && + rmesa->vb.primlist[i].start == rmesa->vb.primlist[j].end) { + rmesa->vb.primlist[j].end = rmesa->vb.primlist[i].end; + } + else { + j++; + if (j != i) rmesa->vb.primlist[j] = rmesa->vb.primlist[i]; + } + } + rmesa->vb.nrprims = j+1; + } + + if (rmesa->vb.vtxfmt_0 != rmesa->hw.vtx.cmd[VTX_VTXFMT_0] || + rmesa->vb.vtxfmt_1 != rmesa->hw.vtx.cmd[VTX_VTXFMT_1]) { + R200_STATECHANGE( rmesa, vtx ); + rmesa->hw.vtx.cmd[VTX_VTXFMT_0] = rmesa->vb.vtxfmt_0; + rmesa->hw.vtx.cmd[VTX_VTXFMT_1] = rmesa->vb.vtxfmt_1; + } + + + for (i = 0 ; i < rmesa->vb.nrprims; i++) { + if (R200_DEBUG & DEBUG_PRIMS) + fprintf(stderr, "vtxfmt prim %d: %s %d..%d\n", i, + _mesa_lookup_enum_by_nr( rmesa->vb.primlist[i].prim & + PRIM_MODE_MASK ), + rmesa->vb.primlist[i].start, + rmesa->vb.primlist[i].end); + + if (rmesa->vb.primlist[i].start < rmesa->vb.primlist[i].end) + r200EmitPrimitive( vb.context, + rmesa->vb.primlist[i].start, + rmesa->vb.primlist[i].end, + rmesa->vb.primlist[i].prim ); + } + + rmesa->vb.nrprims = 0; + r200ReleaseDmaRegion( rmesa, &tmp, __FUNCTION__ ); +} + + +static void start_prim( r200ContextPtr rmesa, GLuint mode ) +{ + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s %d\n", __FUNCTION__, vb.initial_counter - vb.counter); + + rmesa->vb.primlist[rmesa->vb.nrprims].start = vb.initial_counter - vb.counter; + rmesa->vb.primlist[rmesa->vb.nrprims].prim = mode; +} + +static void note_last_prim( r200ContextPtr rmesa, GLuint flags ) +{ + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s %d\n", __FUNCTION__, vb.initial_counter - vb.counter); + + if (rmesa->vb.prim[0] != GL_POLYGON+1) { + rmesa->vb.primlist[rmesa->vb.nrprims].prim |= flags; + rmesa->vb.primlist[rmesa->vb.nrprims].end = vb.initial_counter - vb.counter; + + if (++(rmesa->vb.nrprims) == R200_MAX_PRIMS) + flush_prims( rmesa ); + } +} + + +static void copy_vertex( r200ContextPtr rmesa, GLuint n, GLfloat *dst ) +{ + GLuint i; + GLfloat *src = (GLfloat *)(rmesa->dma.current.address + + rmesa->dma.current.ptr + + (rmesa->vb.primlist[rmesa->vb.nrprims].start + n) * + vb.vertex_size * 4); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "copy_vertex %d\n", rmesa->vb.primlist[rmesa->vb.nrprims].start + n); + + for (i = 0 ; i < vb.vertex_size; i++) { + dst[i] = src[i]; + } +} + +/* NOTE: This actually reads the copied vertices back from uncached + * memory. Could also use the counter/notify mechanism to populate + * tmp on the fly as vertices are generated. + */ +static GLuint copy_dma_verts( r200ContextPtr rmesa, GLfloat (*tmp)[15] ) +{ + GLuint ovf, i; + GLuint nr = (vb.initial_counter - vb.counter) - rmesa->vb.primlist[rmesa->vb.nrprims].start; + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s %d verts\n", __FUNCTION__, nr); + + switch( rmesa->vb.prim[0] ) + { + case GL_POINTS: + return 0; + case GL_LINES: + ovf = nr&1; + for (i = 0 ; i < ovf ; i++) + copy_vertex( rmesa, nr-ovf+i, tmp[i] ); + return i; + case GL_TRIANGLES: + ovf = nr%3; + for (i = 0 ; i < ovf ; i++) + copy_vertex( rmesa, nr-ovf+i, tmp[i] ); + return i; + case GL_QUADS: + ovf = nr&3; + for (i = 0 ; i < ovf ; i++) + copy_vertex( rmesa, nr-ovf+i, tmp[i] ); + return i; + case GL_LINE_STRIP: + if (nr == 0) + return 0; + copy_vertex( rmesa, nr-1, tmp[0] ); + return 1; + case GL_LINE_LOOP: + case GL_TRIANGLE_FAN: + case GL_POLYGON: + if (nr == 0) + return 0; + else if (nr == 1) { + copy_vertex( rmesa, 0, tmp[0] ); + return 1; + } else { + copy_vertex( rmesa, 0, tmp[0] ); + copy_vertex( rmesa, nr-1, tmp[1] ); + return 2; + } + case GL_TRIANGLE_STRIP: + ovf = MIN2( nr-1, 2 ); + for (i = 0 ; i < ovf ; i++) + copy_vertex( rmesa, nr-ovf+i, tmp[i] ); + return i; + case GL_QUAD_STRIP: + ovf = MIN2( nr-1, 2 ); + if (nr > 2) ovf += nr&1; + for (i = 0 ; i < ovf ; i++) + copy_vertex( rmesa, nr-ovf+i, tmp[i] ); + return i; + default: + assert(0); + return 0; + } +} + +static void VFMT_FALLBACK_OUTSIDE_BEGIN_END( const char *caller ) +{ + GLcontext *ctx = vb.context; + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS)) + fprintf(stderr, "%s from %s\n", __FUNCTION__, caller); + + if (ctx->Driver.NeedFlush) + r200FlushVertices( ctx, ctx->Driver.NeedFlush ); + + if (ctx->NewState) + _mesa_update_state( ctx ); /* clear state so fell_back sticks */ + + _tnl_wakeup_exec( ctx ); + + assert( rmesa->dma.flush == 0 ); + rmesa->vb.fell_back = GL_TRUE; + rmesa->vb.installed = GL_FALSE; +/* vb.context = 0; */ +} + + +static void VFMT_FALLBACK( const char *caller ) +{ + GLcontext *ctx = vb.context; + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLfloat tmp[3][15]; + GLuint i, prim; + GLuint ind0 = rmesa->vb.vtxfmt_0; + GLuint ind1 = rmesa->vb.vtxfmt_1; + GLuint nrverts; + GLfloat alpha = 1.0; + + if (R200_DEBUG & (DEBUG_FALLBACKS|DEBUG_VFMT)) + fprintf(stderr, "%s from %s\n", __FUNCTION__, caller); + + if (rmesa->vb.prim[0] == GL_POLYGON+1) { + VFMT_FALLBACK_OUTSIDE_BEGIN_END( __FUNCTION__ ); + return; + } + + /* Copy vertices out of dma: + */ + nrverts = copy_dma_verts( rmesa, tmp ); + + /* Finish the prim at this point: + */ + note_last_prim( rmesa, 0 ); + flush_prims( rmesa ); + + /* Update ctx->Driver.CurrentExecPrimitive and swap in swtnl. + */ + prim = rmesa->vb.prim[0]; + ctx->Driver.CurrentExecPrimitive = GL_POLYGON+1; + _tnl_wakeup_exec( ctx ); + + assert(rmesa->dma.flush == 0); + rmesa->vb.fell_back = GL_TRUE; + rmesa->vb.installed = GL_FALSE; + vb.context = 0; + glBegin( prim ); + + if (rmesa->vb.installed_color_3f_sz == 4) + alpha = ctx->Current.Color[3]; + + /* Replay saved vertices + */ + for (i = 0 ; i < nrverts; i++) { + GLuint offset = 3; + if (ind0 & R200_VTX_N0) { + glNormal3fv( &tmp[i][offset] ); + offset += 3; + } + + if (VTX_COLOR(ind0, 0) == R200_VTX_PK_RGBA) { + glColor4ubv( (GLubyte *)&tmp[i][offset] ); + offset++; + } + else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGBA) { + glColor4fv( &tmp[i][offset] ); + offset+=4; + } + else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGB) { + glColor3fv( &tmp[i][offset] ); + offset+=3; + } + + if (VTX_COLOR(ind0, 1) == R200_VTX_PK_RGBA) { + _glapi_Dispatch->SecondaryColor3ubvEXT( (GLubyte *)&tmp[i][offset] ); + offset++; + } + + if (ind1 & (7 << R200_VTX_TEX0_COMP_CNT_SHIFT)) { + glTexCoord2fv( &tmp[i][offset] ); + offset += 2; + } + + if (ind1 & (7 << R200_VTX_TEX1_COMP_CNT_SHIFT)) { + glMultiTexCoord2fvARB( GL_TEXTURE1_ARB, &tmp[i][offset] ); + offset += 2; + } + + glVertex3fv( &tmp[i][0] ); + } + + /* Replay current vertex + */ + if (ind0 & R200_VTX_N0) + glNormal3fv( vb.normalptr ); + + if (VTX_COLOR(ind0, 0) == R200_VTX_PK_RGBA) + glColor4ubv( vb.ubytecolorptr ); + else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGBA) + glColor4fv( vb.floatcolorptr ); + else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGB) { + if (rmesa->vb.installed_color_3f_sz == 4 && alpha != 1.0) + glColor4f( vb.floatcolorptr[0], + vb.floatcolorptr[1], + vb.floatcolorptr[2], + alpha ); + else + glColor3fv( vb.floatcolorptr ); + } + + if (VTX_COLOR(ind0, 1) == R200_VTX_PK_RGBA) + _glapi_Dispatch->SecondaryColor3ubvEXT( vb.ubytespecptr ); + + if (ind1 & (7 << R200_VTX_TEX0_COMP_CNT_SHIFT)) + glTexCoord2fv( vb.texcoordptr[0] ); + + if (ind1 & (7 << R200_VTX_TEX1_COMP_CNT_SHIFT)) + glMultiTexCoord2fvARB( GL_TEXTURE1_ARB, vb.texcoordptr[1] ); +} + + + +static void wrap_buffer( void ) +{ + GLcontext *ctx = vb.context; + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLfloat tmp[3][15]; + GLuint i, nrverts; + + if (R200_DEBUG & (DEBUG_VFMT|DEBUG_PRIMS)) + fprintf(stderr, "%s %d\n", __FUNCTION__, vb.initial_counter - vb.counter); + + /* Don't deal with parity. + */ + if ((((vb.initial_counter - vb.counter) - + rmesa->vb.primlist[rmesa->vb.nrprims].start) & 1)) { + vb.counter++; + vb.initial_counter++; + return; + } + + /* Copy vertices out of dma: + */ + nrverts = copy_dma_verts( rmesa, tmp ); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%d vertices to copy\n", nrverts); + + + /* Finish the prim at this point: + */ + note_last_prim( rmesa, 0 ); + flush_prims( rmesa ); + + /* Get new buffer + */ + r200RefillCurrentDmaRegion( rmesa ); + + /* Reset counter, dmaptr + */ + vb.dmaptr = (int *)(rmesa->dma.current.ptr + rmesa->dma.current.address); + vb.counter = (rmesa->dma.current.end - rmesa->dma.current.ptr) / + (vb.vertex_size * 4); + vb.counter--; + vb.initial_counter = vb.counter; + vb.notify = wrap_buffer; + + rmesa->dma.flush = flush_prims; + start_prim( rmesa, rmesa->vb.prim[0] ); + + + /* Reemit saved vertices + */ + for (i = 0 ; i < nrverts; i++) { + if (R200_DEBUG & DEBUG_VERTS) { + int j; + fprintf(stderr, "re-emit vertex %d to %p\n", i, vb.dmaptr); + if (R200_DEBUG & DEBUG_VERBOSE) + for (j = 0 ; j < vb.vertex_size; j++) + fprintf(stderr, "\t%08x/%f\n", *(int*)&tmp[i][j], tmp[i][j]); + } + + memcpy( vb.dmaptr, tmp[i], vb.vertex_size * 4 ); + vb.dmaptr += vb.vertex_size; + vb.counter--; + } +} + + + +static GLboolean check_vtx_fmt( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + GLuint ind0 = R200_VTX_Z0; + GLuint ind1 = 0; + + if (rmesa->TclFallback || rmesa->vb.fell_back || ctx->CompileFlag) + return GL_FALSE; + + if (ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) + ctx->Driver.FlushVertices( ctx, FLUSH_UPDATE_CURRENT ); + + /* Make all this event-driven: + */ + if (ctx->Light.Enabled) { + ind0 |= R200_VTX_N0; + + /* TODO: make this data driven: If we receive only ubytes, send + * color as ubytes. Also check if converting (with free + * checking for overflow) is cheaper than sending floats + * directly. + */ + if (ctx->Light.ColorMaterialEnabled) { + if (1 || ctx->Color.AlphaEnabled) + ind0 |= R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT; + else + ind0 |= R200_VTX_FP_RGB << R200_VTX_COLOR_0_SHIFT; + } + } + else { + /* TODO: make this data driven? + */ + ind0 |= R200_VTX_PK_RGBA << R200_VTX_COLOR_0_SHIFT; + + if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) { + ind0 |= R200_VTX_PK_RGBA << R200_VTX_COLOR_1_SHIFT; + } + } + + if (ctx->Texture.Unit[0]._ReallyEnabled) { + if (ctx->Texture.Unit[0].TexGenEnabled) { + if (rmesa->TexGenNeedNormals[0]) { + ind0 |= R200_VTX_N0; + } + } else { + if (ctx->Current.Texcoord[0][2] != 0.0F || + ctx->Current.Texcoord[0][3] != 1.0) { + if (R200_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS)) + fprintf(stderr, "%s: rq0\n", __FUNCTION__); + return GL_FALSE; + } + ind1 |= 2 << R200_VTX_TEX0_COMP_CNT_SHIFT; + } + } + + if (ctx->Texture.Unit[1]._ReallyEnabled) { + if (ctx->Texture.Unit[1].TexGenEnabled) { + if (rmesa->TexGenNeedNormals[1]) { + ind0 |= R200_VTX_N0; + } + } else { + if (ctx->Current.Texcoord[1][2] != 0.0F || + ctx->Current.Texcoord[1][3] != 1.0) { + if (R200_DEBUG & (DEBUG_VFMT|DEBUG_FALLBACKS)) + fprintf(stderr, "%s: rq1\n", __FUNCTION__); + return GL_FALSE; + } + ind1 |= 2 << R200_VTX_TEX1_COMP_CNT_SHIFT; + } + } + + if (R200_DEBUG & (DEBUG_VFMT|DEBUG_STATE)) + fprintf(stderr, "%s: format: 0x%x, 0x%x\n", __FUNCTION__, ind0, ind1 ); + + R200_NEWPRIM(rmesa); + rmesa->vb.vtxfmt_0 = ind0; + rmesa->vb.vtxfmt_1 = ind1; + rmesa->vb.prim = &ctx->Driver.CurrentExecPrimitive; + + vb.vertex_size = 3; + vb.normalptr = ctx->Current.Normal; + vb.ubytecolorptr = 0; + vb.floatcolorptr = ctx->Current.Color; + vb.texcoordptr[0] = ctx->Current.Texcoord[0]; + vb.texcoordptr[1] = ctx->Current.Texcoord[1]; + + /* Run through and initialize the vertex components in the order + * the hardware understands: + */ + if (ind0 & R200_VTX_N0) { + vb.normalptr = &vb.vertex[vb.vertex_size].f; + vb.vertex_size += 3; + vb.normalptr[0] = ctx->Current.Normal[0]; + vb.normalptr[1] = ctx->Current.Normal[1]; + vb.normalptr[2] = ctx->Current.Normal[2]; + } + + if (VTX_COLOR(ind0, 0) == R200_VTX_PK_RGBA) { + vb.ubytecolorptr = &vb.vertex[vb.vertex_size].ub4[0]; + vb.vertex_size += 1; + UNCLAMPED_FLOAT_TO_RGBA_CHAN( vb.ubytecolorptr, ctx->Current.Color ); + } + else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGBA) { + vb.floatcolorptr = &vb.vertex[vb.vertex_size].f; + vb.vertex_size += 4; + vb.floatcolorptr[0] = ctx->Current.Color[0]; + vb.floatcolorptr[1] = ctx->Current.Color[1]; + vb.floatcolorptr[2] = ctx->Current.Color[2]; + vb.floatcolorptr[3] = ctx->Current.Color[3]; + } + else if (VTX_COLOR(ind0, 0) == R200_VTX_FP_RGB) { + vb.floatcolorptr = &vb.vertex[vb.vertex_size].f; + vb.vertex_size += 3; + vb.floatcolorptr[0] = ctx->Current.Color[0]; + vb.floatcolorptr[1] = ctx->Current.Color[1]; + vb.floatcolorptr[2] = ctx->Current.Color[2]; + } + + if (VTX_COLOR(ind0, 1) == R200_VTX_PK_RGBA) { + vb.ubytespecptr = &vb.vertex[vb.vertex_size].ub4[0]; + vb.vertex_size += 1; + UNCLAMPED_FLOAT_TO_RGB_CHAN( vb.ubytespecptr, + ctx->Current.SecondaryColor ); + } + + + if (ind1 & (7 << R200_VTX_TEX0_COMP_CNT_SHIFT)) { + vb.texcoordptr[0] = &vb.vertex[vb.vertex_size].f; + vb.vertex_size += 2; + vb.texcoordptr[0][0] = ctx->Current.Texcoord[0][0]; + vb.texcoordptr[0][1] = ctx->Current.Texcoord[0][1]; + } + + if (ind1 & (7 << R200_VTX_TEX1_COMP_CNT_SHIFT)) { + vb.texcoordptr[1] = &vb.vertex[vb.vertex_size].f; + vb.vertex_size += 2; + vb.texcoordptr[1][0] = ctx->Current.Texcoord[1][0]; + vb.texcoordptr[1][1] = ctx->Current.Texcoord[1][1]; + } + + if (rmesa->vb.installed_vertex_format != rmesa->vb.vtxfmt_0) { + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "reinstall on vertex_format change\n"); + _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt ); + rmesa->vb.installed_vertex_format = rmesa->vb.vtxfmt_0; + } + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s -- success\n", __FUNCTION__); + + return GL_TRUE; +} + + +void r200VtxfmtInvalidate( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + rmesa->vb.recheck = GL_TRUE; + rmesa->vb.fell_back = GL_FALSE; +} + + +static void r200NewList( GLcontext *ctx, GLuint list, GLenum mode ) +{ + VFMT_FALLBACK_OUTSIDE_BEGIN_END( __FUNCTION__ ); +} + + +static void r200VtxfmtValidate( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s\n", __FUNCTION__); + + if (ctx->Driver.NeedFlush) + ctx->Driver.FlushVertices( ctx, ctx->Driver.NeedFlush ); + + rmesa->vb.recheck = GL_FALSE; + + if (check_vtx_fmt( ctx )) { + if (!rmesa->vb.installed) { + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "reinstall (new install)\n"); + + _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt ); + ctx->Driver.FlushVertices = r200FlushVertices; + ctx->Driver.NewList = r200NewList; + rmesa->vb.installed = GL_TRUE; + vb.context = ctx; + } + else if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s: already installed", __FUNCTION__); + } + else { + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s: failed\n", __FUNCTION__); + + if (rmesa->vb.installed) { + if (rmesa->dma.flush) + rmesa->dma.flush( rmesa ); + _tnl_wakeup_exec( ctx ); + rmesa->vb.installed = GL_FALSE; + vb.context = 0; + } + } +} + + + +/* Materials: + */ +static void r200_Materialfv( GLenum face, GLenum pname, + const GLfloat *params ) +{ + GLcontext *ctx = vb.context; + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s\n", __FUNCTION__); + + if (rmesa->vb.prim[0] != GL_POLYGON+1) { + VFMT_FALLBACK( __FUNCTION__ ); + glMaterialfv( face, pname, params ); + return; + } + _mesa_noop_Materialfv( face, pname, params ); + r200UpdateMaterial( vb.context ); +} + + +/* Begin/End + */ +static void r200_Begin( GLenum mode ) +{ + GLcontext *ctx = vb.context; + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s( %s )\n", __FUNCTION__, + _mesa_lookup_enum_by_nr( mode )); + + if (mode > GL_POLYGON) { + _mesa_error( ctx, GL_INVALID_ENUM, "glBegin" ); + return; + } + + if (rmesa->vb.prim[0] != GL_POLYGON+1) { + _mesa_error( ctx, GL_INVALID_OPERATION, "glBegin" ); + return; + } + + if (ctx->NewState) + _mesa_update_state( ctx ); + + if (rmesa->NewGLState) + r200ValidateState( ctx ); + + if (rmesa->vb.recheck) + r200VtxfmtValidate( ctx ); + + if (!rmesa->vb.installed) { + glBegin( mode ); + return; + } + + + if (rmesa->dma.flush && vb.counter < 12) { + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s: flush almost-empty buffers\n", __FUNCTION__); + flush_prims( rmesa ); + } + + /* Need to arrange to save vertices here? Or always copy from dma (yuk)? + */ + if (!rmesa->dma.flush) { + if (rmesa->dma.current.ptr + 12*vb.vertex_size*4 > + rmesa->dma.current.end) { + R200_NEWPRIM( rmesa ); + r200RefillCurrentDmaRegion( rmesa ); + } + + vb.dmaptr = (int *)(rmesa->dma.current.address + rmesa->dma.current.ptr); + vb.counter = (rmesa->dma.current.end - rmesa->dma.current.ptr) / + (vb.vertex_size * 4); + vb.counter--; + vb.initial_counter = vb.counter; + vb.notify = wrap_buffer; + rmesa->dma.flush = flush_prims; + vb.context->Driver.NeedFlush |= FLUSH_STORED_VERTICES; + } + + + rmesa->vb.prim[0] = mode; + start_prim( rmesa, mode | PRIM_BEGIN ); +} + + + +static void r200_End( void ) +{ + GLcontext *ctx = vb.context; + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s\n", __FUNCTION__); + + if (rmesa->vb.prim[0] == GL_POLYGON+1) { + _mesa_error( ctx, GL_INVALID_OPERATION, "glEnd" ); + return; + } + + note_last_prim( rmesa, PRIM_END ); + rmesa->vb.prim[0] = GL_POLYGON+1; +} + + +/* Fallback on difficult entrypoints: + */ +#define PRE_LOOPBACK( FUNC ) \ +do { \ + if (R200_DEBUG & DEBUG_VFMT) \ + fprintf(stderr, "%s\n", __FUNCTION__); \ + VFMT_FALLBACK( __FUNCTION__ ); \ +} while (0) +#define TAG(x) r200_fallback_##x +#include "vtxfmt_tmp.h" + + + +static GLboolean r200NotifyBegin( GLcontext *ctx, GLenum p ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s\n", __FUNCTION__); + + assert(!rmesa->vb.installed); + + if (ctx->NewState) + _mesa_update_state( ctx ); + + if (rmesa->NewGLState) + r200ValidateState( ctx ); + + if (ctx->Driver.NeedFlush) + ctx->Driver.FlushVertices( ctx, ctx->Driver.NeedFlush ); + + if (rmesa->vb.recheck) + r200VtxfmtValidate( ctx ); + + if (!rmesa->vb.installed) { + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s -- failed\n", __FUNCTION__); + return GL_FALSE; + } + + r200_Begin( p ); + return GL_TRUE; +} + +static void r200FlushVertices( GLcontext *ctx, GLuint flags ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "%s\n", __FUNCTION__); + + assert(rmesa->vb.installed); + assert(vb.context == ctx); + + if (flags & FLUSH_UPDATE_CURRENT) { + r200_copy_to_current( ctx ); + if (R200_DEBUG & DEBUG_VFMT) + fprintf(stderr, "reinstall on update_current\n"); + _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt ); + ctx->Driver.NeedFlush &= ~FLUSH_UPDATE_CURRENT; + } + + if (flags & FLUSH_STORED_VERTICES) { + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + assert (rmesa->dma.flush == 0 || + rmesa->dma.flush == flush_prims); + if (rmesa->dma.flush == flush_prims) + flush_prims( R200_CONTEXT( ctx ) ); + ctx->Driver.NeedFlush &= ~FLUSH_STORED_VERTICES; + } +} + + + +/* At this point, don't expect very many versions of each function to + * be generated, so not concerned about freeing them? + */ + + +void r200VtxfmtInit( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + GLvertexformat *vfmt = &(rmesa->vb.vtxfmt); + + MEMSET( vfmt, 0, sizeof(GLvertexformat) ); + + /* Hook in chooser functions for codegen, etc: + */ + r200VtxfmtInitChoosers( vfmt ); + + /* Handled fully in supported states, but no codegen: + */ + vfmt->Materialfv = r200_Materialfv; + vfmt->ArrayElement = _ae_loopback_array_elt; /* generic helper */ + vfmt->Rectf = _mesa_noop_Rectf; /* generic helper */ + vfmt->Begin = r200_Begin; + vfmt->End = r200_End; + + /* Fallback for performance reasons: (Fix with cva/elt path here and + * dmatmp2.h style primitive-merging) + * + * These should call NotifyBegin(), as should _tnl_EvalMesh, to allow + * a driver-hook. + */ + vfmt->DrawArrays = r200_fallback_DrawArrays; + vfmt->DrawElements = r200_fallback_DrawElements; + vfmt->DrawRangeElements = r200_fallback_DrawRangeElements; + + + /* Not active in supported states; just keep ctx->Current uptodate: + */ + vfmt->FogCoordfvEXT = _mesa_noop_FogCoordfvEXT; + vfmt->FogCoordfEXT = _mesa_noop_FogCoordfEXT; + vfmt->EdgeFlag = _mesa_noop_EdgeFlag; + vfmt->EdgeFlagv = _mesa_noop_EdgeFlagv; + vfmt->Indexi = _mesa_noop_Indexi; + vfmt->Indexiv = _mesa_noop_Indexiv; + + + /* Active but unsupported -- fallback if we receive these: + */ + vfmt->CallList = r200_fallback_CallList; + vfmt->EvalCoord1f = r200_fallback_EvalCoord1f; + vfmt->EvalCoord1fv = r200_fallback_EvalCoord1fv; + vfmt->EvalCoord2f = r200_fallback_EvalCoord2f; + vfmt->EvalCoord2fv = r200_fallback_EvalCoord2fv; + vfmt->EvalMesh1 = r200_fallback_EvalMesh1; + vfmt->EvalMesh2 = r200_fallback_EvalMesh2; + vfmt->EvalPoint1 = r200_fallback_EvalPoint1; + vfmt->EvalPoint2 = r200_fallback_EvalPoint2; + vfmt->TexCoord3f = r200_fallback_TexCoord3f; + vfmt->TexCoord3fv = r200_fallback_TexCoord3fv; + vfmt->TexCoord4f = r200_fallback_TexCoord4f; + vfmt->TexCoord4fv = r200_fallback_TexCoord4fv; + vfmt->MultiTexCoord3fARB = r200_fallback_MultiTexCoord3fARB; + vfmt->MultiTexCoord3fvARB = r200_fallback_MultiTexCoord3fvARB; + vfmt->MultiTexCoord4fARB = r200_fallback_MultiTexCoord4fARB; + vfmt->MultiTexCoord4fvARB = r200_fallback_MultiTexCoord4fvARB; + vfmt->Vertex4f = r200_fallback_Vertex4f; + vfmt->Vertex4fv = r200_fallback_Vertex4fv; + + (void)r200_fallback_vtxfmt; + + TNL_CONTEXT(ctx)->Driver.NotifyBegin = r200NotifyBegin; + + vb.context = ctx; + rmesa->vb.enabled = 1; + rmesa->vb.prim = &ctx->Driver.CurrentExecPrimitive; + rmesa->vb.primflags = 0; + + make_empty_list( &rmesa->vb.dfn_cache.Vertex2f ); + make_empty_list( &rmesa->vb.dfn_cache.Vertex2fv ); + make_empty_list( &rmesa->vb.dfn_cache.Vertex3f ); + make_empty_list( &rmesa->vb.dfn_cache.Vertex3fv ); + make_empty_list( &rmesa->vb.dfn_cache.Color4ub ); + make_empty_list( &rmesa->vb.dfn_cache.Color4ubv ); + make_empty_list( &rmesa->vb.dfn_cache.Color3ub ); + make_empty_list( &rmesa->vb.dfn_cache.Color3ubv ); + make_empty_list( &rmesa->vb.dfn_cache.Color4f ); + make_empty_list( &rmesa->vb.dfn_cache.Color4fv ); + make_empty_list( &rmesa->vb.dfn_cache.Color3f ); + make_empty_list( &rmesa->vb.dfn_cache.Color3fv ); + make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3fEXT ); + make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3fvEXT ); + make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3ubEXT ); + make_empty_list( &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT ); + make_empty_list( &rmesa->vb.dfn_cache.Normal3f ); + make_empty_list( &rmesa->vb.dfn_cache.Normal3fv ); + make_empty_list( &rmesa->vb.dfn_cache.TexCoord2f ); + make_empty_list( &rmesa->vb.dfn_cache.TexCoord2fv ); + make_empty_list( &rmesa->vb.dfn_cache.TexCoord1f ); + make_empty_list( &rmesa->vb.dfn_cache.TexCoord1fv ); + make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord2fARB ); + make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord2fvARB ); + make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord1fARB ); + make_empty_list( &rmesa->vb.dfn_cache.MultiTexCoord1fvARB ); + + r200InitCodegen( &rmesa->vb.codegen ); +} + +static void free_funcs( struct dynfn *l ) +{ + struct dynfn *f, *tmp; + foreach_s (f, tmp, l) { + remove_from_list( f ); + ALIGN_FREE( f->code ); + FREE( f ); + } +} + +void r200VtxfmtUnbindContext( GLcontext *ctx ) +{ + if (R200_CONTEXT(ctx)->vb.installed) { + assert(vb.context == ctx); + VFMT_FALLBACK_OUTSIDE_BEGIN_END( __FUNCTION__ ); + } + + TNL_CONTEXT(ctx)->Driver.NotifyBegin = 0; +} + + +void r200VtxfmtMakeCurrent( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + +#if defined(THREADS) + static GLboolean ThreadSafe = GL_FALSE; /* In thread-safe mode? */ + if (!ThreadSafe) { + static unsigned long knownID; + static GLboolean firstCall = GL_TRUE; + if (firstCall) { + knownID = _glthread_GetID(); + firstCall = GL_FALSE; + } + else if (knownID != _glthread_GetID()) { + ThreadSafe = GL_TRUE; + + if (R200_DEBUG & (DEBUG_DRI|DEBUG_VFMT)) + fprintf(stderr, "**** Multithread situation!\n"); + } + } + if (ThreadSafe) + return; +#endif + + if (rmesa->vb.enabled) { + TNL_CONTEXT(ctx)->Driver.NotifyBegin = r200NotifyBegin; + } +} + + +void r200VtxfmtDestroy( GLcontext *ctx ) +{ + r200ContextPtr rmesa = R200_CONTEXT( ctx ); + + count_funcs( rmesa ); + free_funcs( &rmesa->vb.dfn_cache.Vertex2f ); + free_funcs( &rmesa->vb.dfn_cache.Vertex2fv ); + free_funcs( &rmesa->vb.dfn_cache.Vertex3f ); + free_funcs( &rmesa->vb.dfn_cache.Vertex3fv ); + free_funcs( &rmesa->vb.dfn_cache.Color4ub ); + free_funcs( &rmesa->vb.dfn_cache.Color4ubv ); + free_funcs( &rmesa->vb.dfn_cache.Color3ub ); + free_funcs( &rmesa->vb.dfn_cache.Color3ubv ); + free_funcs( &rmesa->vb.dfn_cache.Color4f ); + free_funcs( &rmesa->vb.dfn_cache.Color4fv ); + free_funcs( &rmesa->vb.dfn_cache.Color3f ); + free_funcs( &rmesa->vb.dfn_cache.Color3fv ); + free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3ubEXT ); + free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3ubvEXT ); + free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3fEXT ); + free_funcs( &rmesa->vb.dfn_cache.SecondaryColor3fvEXT ); + free_funcs( &rmesa->vb.dfn_cache.Normal3f ); + free_funcs( &rmesa->vb.dfn_cache.Normal3fv ); + free_funcs( &rmesa->vb.dfn_cache.TexCoord2f ); + free_funcs( &rmesa->vb.dfn_cache.TexCoord2fv ); + free_funcs( &rmesa->vb.dfn_cache.TexCoord1f ); + free_funcs( &rmesa->vb.dfn_cache.TexCoord1fv ); + free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord2fARB ); + free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord2fvARB ); + free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord1fARB ); + free_funcs( &rmesa->vb.dfn_cache.MultiTexCoord1fvARB ); +} + diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt.h b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt.h new file mode 100644 index 000000000..7a04d971f --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt.h @@ -0,0 +1,128 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#ifndef __R200_VTXFMT_H__ +#define __R200_VTXFMT_H__ + +#ifdef GLX_DIRECT_RENDERING + +#include "r200_context.h" + + + +extern struct r200_vb vb; + + +extern void r200VtxfmtUpdate( GLcontext *ctx ); +extern void r200VtxfmtInit( GLcontext *ctx ); +extern void r200VtxfmtInvalidate( GLcontext *ctx ); +extern void r200VtxfmtDestroy( GLcontext *ctx ); +extern void r200VtxfmtInitChoosers( GLvertexformat *vfmt ); + +extern void r200VtxfmtMakeCurrent( GLcontext *ctx ); +extern void r200VtxfmtUnbindContext( GLcontext *ctx ); + +extern void r200_copy_to_current( GLcontext *ctx ); + +#define DFN( FUNC, CACHE) \ +do { \ + char *start = (char *)&FUNC; \ + char *end = (char *)&FUNC##_end; \ + insert_at_head( &CACHE, dfn ); \ + dfn->key[0] = key[0]; \ + dfn->key[1] = key[1]; \ + dfn->code = ALIGN_MALLOC( end - start, 16 ); \ + memcpy (dfn->code, start, end - start); \ +} \ +while ( 0 ) + +#define FIXUP( CODE, OFFSET, CHECKVAL, NEWVAL ) \ +do { \ + int *icode = (int *)(CODE+OFFSET); \ + assert (*icode == CHECKVAL); \ + *icode = (int)NEWVAL; \ +} while (0) + + +/* Useful for figuring out the offsets: + */ +#define FIXUP2( CODE, OFFSET, CHECKVAL, NEWVAL ) \ +do { \ + while (*(int *)(CODE+OFFSET) != CHECKVAL) OFFSET++; \ + fprintf(stderr, "%s/%d CVAL %x OFFSET %d VAL %x\n", __FUNCTION__, \ + __LINE__, CHECKVAL, OFFSET, (int)(NEWVAL)); \ + *(int *)(CODE+OFFSET) = (int)(NEWVAL); \ + OFFSET += 4; \ +} while (0) + +/* + */ +void r200InitCodegen( struct dfn_generators *gen ); +void r200InitX86Codegen( struct dfn_generators *gen ); +void r200InitSSECodegen( struct dfn_generators *gen ); + + + +/* Defined in r200_vtxfmt_x86.c + */ +struct dynfn *r200_makeX86Vertex2f( GLcontext *, const int * ); +struct dynfn *r200_makeX86Vertex2fv( GLcontext *, const int * ); +struct dynfn *r200_makeX86Vertex3f( GLcontext *, const int * ); +struct dynfn *r200_makeX86Vertex3fv( GLcontext *, const int * ); +struct dynfn *r200_makeX86Color4ub( GLcontext *, const int * ); +struct dynfn *r200_makeX86Color4ubv( GLcontext *, const int * ); +struct dynfn *r200_makeX86Color3ub( GLcontext *, const int * ); +struct dynfn *r200_makeX86Color3ubv( GLcontext *, const int * ); +struct dynfn *r200_makeX86Color4f( GLcontext *, const int * ); +struct dynfn *r200_makeX86Color4fv( GLcontext *, const int * ); +struct dynfn *r200_makeX86Color3f( GLcontext *, const int * ); +struct dynfn *r200_makeX86Color3fv( GLcontext *, const int * ); +struct dynfn *r200_makeX86SecondaryColor3ubEXT( GLcontext *, const int * ); +struct dynfn *r200_makeX86SecondaryColor3ubvEXT( GLcontext *, const int * ); +struct dynfn *r200_makeX86SecondaryColor3fEXT( GLcontext *, const int * ); +struct dynfn *r200_makeX86SecondaryColor3fvEXT( GLcontext *, const int * ); +struct dynfn *r200_makeX86Normal3f( GLcontext *, const int * ); +struct dynfn *r200_makeX86Normal3fv( GLcontext *, const int * ); +struct dynfn *r200_makeX86TexCoord2f( GLcontext *, const int * ); +struct dynfn *r200_makeX86TexCoord2fv( GLcontext *, const int * ); +struct dynfn *r200_makeX86TexCoord1f( GLcontext *, const int * ); +struct dynfn *r200_makeX86TexCoord1fv( GLcontext *, const int * ); +struct dynfn *r200_makeX86MultiTexCoord2fARB( GLcontext *, const int * ); +struct dynfn *r200_makeX86MultiTexCoord2fvARB( GLcontext *, const int * ); +struct dynfn *r200_makeX86MultiTexCoord1fARB( GLcontext *, const int * ); +struct dynfn *r200_makeX86MultiTexCoord1fvARB( GLcontext *, const int * ); + + +#endif +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_c.c b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_c.c new file mode 100644 index 000000000..bc6a49546 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_c.c @@ -0,0 +1,723 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include "mtypes.h" +#include "colormac.h" +#include "simple_list.h" +#include "api_noop.h" +#include "vtxfmt.h" + +#include "r200_vtxfmt.h" + +/* Fallback versions of all the entrypoints for situations where + * codegen isn't available. This is still a lot faster than the + * vb/pipeline implementation in Mesa. + */ +static void r200_Vertex3f( GLfloat x, GLfloat y, GLfloat z ) +{ + int i; + + *vb.dmaptr++ = *(int *)&x; + *vb.dmaptr++ = *(int *)&y; + *vb.dmaptr++ = *(int *)&z; + + for (i = 3; i < vb.vertex_size; i++) + *vb.dmaptr++ = vb.vertex[i].i; + + if (--vb.counter == 0) + vb.notify(); +} + + +static void r200_Vertex3fv( const GLfloat *v ) +{ + int i; + + *vb.dmaptr++ = *(int *)&v[0]; + *vb.dmaptr++ = *(int *)&v[1]; + *vb.dmaptr++ = *(int *)&v[2]; + + for (i = 3; i < vb.vertex_size; i++) + *vb.dmaptr++ = vb.vertex[i].i; + + if (--vb.counter == 0) + vb.notify(); +} + + +static void r200_Vertex2f( GLfloat x, GLfloat y ) +{ + int i; + + *vb.dmaptr++ = *(int *)&x; + *vb.dmaptr++ = *(int *)&y; + *vb.dmaptr++ = 0; + + for (i = 3; i < vb.vertex_size; i++) + *vb.dmaptr++ = vb.vertex[i].i; + + if (--vb.counter == 0) + vb.notify(); +} + + +static void r200_Vertex2fv( const GLfloat *v ) +{ + int i; + + *vb.dmaptr++ = *(int *)&v[0]; + *vb.dmaptr++ = *(int *)&v[1]; + *vb.dmaptr++ = 0; + + for (i = 3; i < vb.vertex_size; i++) + *vb.dmaptr++ = vb.vertex[i].i; + + if (--vb.counter == 0) + vb.notify(); +} + + + +/* Color for ubyte (packed) color formats: + */ +static void r200_Color3ub_ub( GLubyte r, GLubyte g, GLubyte b ) +{ + GLubyte *dest = vb.ubytecolorptr; + dest[0] = r; + dest[1] = g; + dest[2] = b; + dest[3] = 0xff; +} + +static void r200_Color3ubv_ub( const GLubyte *v ) +{ + GLubyte *dest = vb.ubytecolorptr; + dest[0] = v[0]; + dest[1] = v[1]; + dest[2] = v[2]; + dest[3] = 0xff; +} + +static void r200_Color4ub_ub( GLubyte r, GLubyte g, GLubyte b, GLubyte a ) +{ + GLubyte *dest = vb.ubytecolorptr; + dest[0] = r; + dest[1] = g; + dest[2] = b; + dest[3] = a; +} + +static void r200_Color4ubv_ub( const GLubyte *v ) +{ + GLubyte *dest = vb.ubytecolorptr; + *(int *)dest = *(int *)v; +} + + +static void r200_Color3f_ub( GLfloat r, GLfloat g, GLfloat b ) +{ + GLubyte *dest = vb.ubytecolorptr; + UNCLAMPED_FLOAT_TO_UBYTE( dest[0], r ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[1], g ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[2], b ); + dest[3] = 255; +} + +static void r200_Color3fv_ub( const GLfloat *v ) +{ + GLubyte *dest = vb.ubytecolorptr; + UNCLAMPED_FLOAT_TO_UBYTE( dest[0], v[0] ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[1], v[1] ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[2], v[2] ); + dest[3] = 255; +} + +static void r200_Color4f_ub( GLfloat r, GLfloat g, GLfloat b, GLfloat a ) +{ + GLubyte *dest = vb.ubytecolorptr; + UNCLAMPED_FLOAT_TO_UBYTE( dest[0], r ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[1], g ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[2], b ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[3], a ); +} + +static void r200_Color4fv_ub( const GLfloat *v ) +{ + GLubyte *dest = vb.ubytecolorptr; + UNCLAMPED_FLOAT_TO_UBYTE( dest[0], v[0] ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[1], v[1] ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[2], v[2] ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[3], v[3] ); +} + + +/* Color for float color+alpha formats: + */ +static void r200_Color3ub_4f( GLubyte r, GLubyte g, GLubyte b ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = UBYTE_TO_FLOAT(r); + dest[1] = UBYTE_TO_FLOAT(g); + dest[2] = UBYTE_TO_FLOAT(b); + dest[3] = 1.0; +} + +static void r200_Color3ubv_4f( const GLubyte *v ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = UBYTE_TO_FLOAT(v[0]); + dest[1] = UBYTE_TO_FLOAT(v[1]); + dest[2] = UBYTE_TO_FLOAT(v[2]); + dest[3] = 1.0; +} + +static void r200_Color4ub_4f( GLubyte r, GLubyte g, GLubyte b, GLubyte a ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = UBYTE_TO_FLOAT(r); + dest[1] = UBYTE_TO_FLOAT(g); + dest[2] = UBYTE_TO_FLOAT(b); + dest[3] = UBYTE_TO_FLOAT(a); +} + +static void r200_Color4ubv_4f( const GLubyte *v ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = UBYTE_TO_FLOAT(v[0]); + dest[1] = UBYTE_TO_FLOAT(v[1]); + dest[2] = UBYTE_TO_FLOAT(v[2]); + dest[3] = UBYTE_TO_FLOAT(v[3]); +} + + +static void r200_Color3f_4f( GLfloat r, GLfloat g, GLfloat b ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = r; + dest[1] = g; + dest[2] = b; + dest[3] = 1.0; +} + +static void r200_Color3fv_4f( const GLfloat *v ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = v[0]; + dest[1] = v[1]; + dest[2] = v[2]; + dest[3] = 1.0; +} + +static void r200_Color4f_4f( GLfloat r, GLfloat g, GLfloat b, GLfloat a ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = r; + dest[1] = g; + dest[2] = b; + dest[3] = a; +} + +static void r200_Color4fv_4f( const GLfloat *v ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = v[0]; + dest[1] = v[1]; + dest[2] = v[2]; + dest[3] = v[3]; +} + + +/* Color for float color formats: + */ +static void r200_Color3ub_3f( GLubyte r, GLubyte g, GLubyte b ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = UBYTE_TO_FLOAT(r); + dest[1] = UBYTE_TO_FLOAT(g); + dest[2] = UBYTE_TO_FLOAT(b); +} + +static void r200_Color3ubv_3f( const GLubyte *v ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = UBYTE_TO_FLOAT(v[0]); + dest[1] = UBYTE_TO_FLOAT(v[1]); + dest[2] = UBYTE_TO_FLOAT(v[2]); +} + +static void r200_Color4ub_3f( GLubyte r, GLubyte g, GLubyte b, GLubyte a ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = UBYTE_TO_FLOAT(r); + dest[1] = UBYTE_TO_FLOAT(g); + dest[2] = UBYTE_TO_FLOAT(b); + vb.context->Current.Color[3] = UBYTE_TO_FLOAT(a); +} + +static void r200_Color4ubv_3f( const GLubyte *v ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = UBYTE_TO_FLOAT(v[0]); + dest[1] = UBYTE_TO_FLOAT(v[1]); + dest[2] = UBYTE_TO_FLOAT(v[2]); + vb.context->Current.Color[3] = UBYTE_TO_FLOAT(v[3]); +} + + +static void r200_Color3f_3f( GLfloat r, GLfloat g, GLfloat b ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = r; + dest[1] = g; + dest[2] = b; +} + +static void r200_Color3fv_3f( const GLfloat *v ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = v[0]; + dest[1] = v[1]; + dest[2] = v[2]; +} + +static void r200_Color4f_3f( GLfloat r, GLfloat g, GLfloat b, GLfloat a ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = r; + dest[1] = g; + dest[2] = b; + vb.context->Current.Color[3] = a; +} + +static void r200_Color4fv_3f( const GLfloat *v ) +{ + GLfloat *dest = vb.floatcolorptr; + dest[0] = v[0]; + dest[1] = v[1]; + dest[2] = v[2]; + vb.context->Current.Color[3] = v[3]; +} + + +/* Secondary Color: + */ +static void r200_SecondaryColor3ubEXT( GLubyte r, GLubyte g, GLubyte b ) +{ + GLubyte *dest = vb.ubytespecptr; + dest[0] = r; + dest[1] = g; + dest[2] = b; + dest[3] = 0xff; +} + +static void r200_SecondaryColor3ubvEXT( const GLubyte *v ) +{ + GLubyte *dest = vb.ubytespecptr; + dest[0] = v[0]; + dest[1] = v[1]; + dest[2] = v[2]; + dest[3] = 0xff; +} + +static void r200_SecondaryColor3fEXT( GLfloat r, GLfloat g, GLfloat b ) +{ + GLubyte *dest = vb.ubytespecptr; + UNCLAMPED_FLOAT_TO_UBYTE( dest[0], r ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[1], g ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[2], b ); + dest[3] = 255; +} + +static void r200_SecondaryColor3fvEXT( const GLfloat *v ) +{ + GLubyte *dest = vb.ubytespecptr; + UNCLAMPED_FLOAT_TO_UBYTE( dest[0], v[0] ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[1], v[1] ); + UNCLAMPED_FLOAT_TO_UBYTE( dest[2], v[2] ); + dest[3] = 255; +} + + + +/* Normal + */ +static void r200_Normal3f( GLfloat n0, GLfloat n1, GLfloat n2 ) +{ + GLfloat *dest = vb.normalptr; + dest[0] = n0; + dest[1] = n1; + dest[2] = n2; +} + +static void r200_Normal3fv( const GLfloat *v ) +{ + GLfloat *dest = vb.normalptr; + dest[0] = v[0]; + dest[1] = v[1]; + dest[2] = v[2]; +} + + +/* TexCoord + */ +static void r200_TexCoord1f( GLfloat s ) +{ + GLfloat *dest = vb.texcoordptr[0]; + dest[0] = s; + dest[1] = 0; +} + +static void r200_TexCoord1fv( const GLfloat *v ) +{ + GLfloat *dest = vb.texcoordptr[0]; + dest[0] = v[0]; + dest[1] = 0; +} + +static void r200_TexCoord2f( GLfloat s, GLfloat t ) +{ + GLfloat *dest = vb.texcoordptr[0]; + dest[0] = s; + dest[1] = t; +} + +static void r200_TexCoord2fv( const GLfloat *v ) +{ + GLfloat *dest = vb.texcoordptr[0]; + dest[0] = v[0]; + dest[1] = v[1]; +} + + +/* MultiTexcoord + */ +static void r200_MultiTexCoord1fARB( GLenum target, GLfloat s ) +{ + GLfloat *dest = vb.texcoordptr[(target - GL_TEXTURE0_ARB)&1]; + dest[0] = s; + dest[1] = 0; +} + +static void r200_MultiTexCoord1fvARB( GLenum target, const GLfloat *v ) +{ + GLfloat *dest = vb.texcoordptr[(target - GL_TEXTURE0_ARB)&1]; + dest[0] = v[0]; + dest[1] = 0; +} + +static void r200_MultiTexCoord2fARB( GLenum target, GLfloat s, GLfloat t ) +{ + GLfloat *dest = vb.texcoordptr[(target - GL_TEXTURE0_ARB)&1]; + dest[0] = s; + dest[1] = t; +} + +static void r200_MultiTexCoord2fvARB( GLenum target, const GLfloat *v ) +{ + GLfloat *dest = vb.texcoordptr[(target - GL_TEXTURE0_ARB)&1]; + dest[0] = v[0]; + dest[1] = v[1]; +} + +static struct dynfn *lookup( struct dynfn *l, const int *key ) +{ + struct dynfn *f; + + foreach( f, l ) { + if (f->key[0] == key[0] && f->key[1] == key[1]) + return f; + } + + return 0; +} + +/* Can't use the loopback template for this: + */ + +#define CHOOSE(FN, FNTYPE, MASK0, MASK1, ARGS1, ARGS2 ) \ +static void choose_##FN ARGS1 \ +{ \ + r200ContextPtr rmesa = R200_CONTEXT(vb.context); \ + int key[2]; \ + struct dynfn *dfn; \ + \ + key[0] = rmesa->vb.vtxfmt_0 & MASK0; \ + key[1] = rmesa->vb.vtxfmt_1 & MASK1; \ + \ + dfn = lookup( &rmesa->vb.dfn_cache.FN, key ); \ + if (dfn == 0) \ + dfn = rmesa->vb.codegen.FN( vb.context, key ); \ + else if (R200_DEBUG & DEBUG_CODEGEN) \ + fprintf(stderr, "%s -- cached codegen\n", __FUNCTION__ ); \ + \ + if (dfn) \ + vb.context->Exec->FN = (FNTYPE)(dfn->code); \ + else { \ + if (R200_DEBUG & DEBUG_CODEGEN) \ + fprintf(stderr, "%s -- generic version\n", __FUNCTION__ ); \ + vb.context->Exec->FN = r200_##FN; \ + } \ + \ + vb.context->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT; \ + vb.context->Exec->FN ARGS2; \ +} + + + +/* For the _3f case, only allow one color function to be hooked in at + * a time. Eventually, use a similar mechanism to allow selecting the + * color component of the vertex format based on client behaviour. + * + * Note: Perform these actions even if there is a codegen or cached + * codegen version of the chosen function. + */ +#define CHOOSE_COLOR(FN, FNTYPE, NR, MASK0, MASK1, ARGS1, ARGS2 ) \ +static void choose_##FN ARGS1 \ +{ \ + GLcontext *ctx = vb.context; \ + r200ContextPtr rmesa = R200_CONTEXT(vb.context); \ + int key[2]; \ + struct dynfn *dfn; \ + \ + key[0] = rmesa->vb.vtxfmt_0 & MASK0; \ + key[1] = rmesa->vb.vtxfmt_1 & MASK1; \ + \ + if (VTX_COLOR(rmesa->vb.vtxfmt_0,0) == R200_VTX_PK_RGBA) { \ + ctx->Exec->FN = r200_##FN##_ub; \ + } \ + else if (VTX_COLOR(rmesa->vb.vtxfmt_0,0) == R200_VTX_FP_RGB) { \ + \ + if (rmesa->vb.installed_color_3f_sz != NR) { \ + rmesa->vb.installed_color_3f_sz = NR; \ + if (NR == 3) ctx->Current.Color[3] = 1.0; \ + if (ctx->Driver.NeedFlush & FLUSH_UPDATE_CURRENT) { \ + r200_copy_to_current( ctx ); \ + _mesa_install_exec_vtxfmt( ctx, &rmesa->vb.vtxfmt ); \ + ctx->Exec->FN ARGS2; \ + return; \ + } \ + } \ + \ + ctx->Exec->FN = r200_##FN##_3f; \ + } \ + else { \ + ctx->Exec->FN = r200_##FN##_4f; \ + } \ + \ + \ + dfn = lookup( &rmesa->vb.dfn_cache.FN, key ); \ + if (!dfn) dfn = rmesa->vb.codegen.FN( ctx, key ); \ + \ + if (dfn) { \ + if (R200_DEBUG & DEBUG_CODEGEN) \ + fprintf(stderr, "%s -- codegen version\n", __FUNCTION__ ); \ + ctx->Exec->FN = (FNTYPE)dfn->code; \ + } \ + else if (R200_DEBUG & DEBUG_CODEGEN) \ + fprintf(stderr, "%s -- 'c' version\n", __FUNCTION__ ); \ + \ + ctx->Driver.NeedFlush |= FLUSH_UPDATE_CURRENT; \ + ctx->Exec->FN ARGS2; \ +} + + + + + +/* VTXFMT_0 + */ +#define MASK_XYZW (R200_VTX_W0|R200_VTX_Z0) +#define MASK_NORM (MASK_XYZW|R200_VTX_N0) +#define MASK_COLOR (MASK_NORM |(R200_VTX_COLOR_MASK<<R200_VTX_COLOR_0_SHIFT)) +#define MASK_SPEC (MASK_COLOR|(R200_VTX_COLOR_MASK<<R200_VTX_COLOR_1_SHIFT)) + +/* VTXFMT_1 + */ +#define MASK_ST0 (0x7 << R200_VTX_TEX0_COMP_CNT_SHIFT) + + + +typedef void (*p4f)( GLfloat, GLfloat, GLfloat, GLfloat ); +typedef void (*p3f)( GLfloat, GLfloat, GLfloat ); +typedef void (*p2f)( GLfloat, GLfloat ); +typedef void (*p1f)( GLfloat ); +typedef void (*pe2f)( GLenum, GLfloat, GLfloat ); +typedef void (*pe1f)( GLenum, GLfloat ); +typedef void (*p4ub)( GLubyte, GLubyte, GLubyte, GLubyte ); +typedef void (*p3ub)( GLubyte, GLubyte, GLubyte ); +typedef void (*pfv)( const GLfloat * ); +typedef void (*pefv)( GLenum, const GLfloat * ); +typedef void (*pubv)( const GLubyte * ); + + +CHOOSE(Normal3f, p3f, MASK_NORM, 0, + (GLfloat a,GLfloat b,GLfloat c), (a,b,c)) +CHOOSE(Normal3fv, pfv, MASK_NORM, 0, + (const GLfloat *v), (v)) + +CHOOSE_COLOR(Color4ub, p4ub, 4, MASK_COLOR, 0, + (GLubyte a,GLubyte b, GLubyte c, GLubyte d), (a,b,c,d)) +CHOOSE_COLOR(Color4ubv, pubv, 4, MASK_COLOR, 0, + (const GLubyte *v), (v)) +CHOOSE_COLOR(Color3ub, p3ub, 3, MASK_COLOR, 0, + (GLubyte a,GLubyte b, GLubyte c), (a,b,c)) +CHOOSE_COLOR(Color3ubv, pubv, 3, MASK_COLOR, 0, + (const GLubyte *v), (v)) + +CHOOSE_COLOR(Color4f, p4f, 4, MASK_COLOR, 0, + (GLfloat a,GLfloat b, GLfloat c, GLfloat d), (a,b,c,d)) +CHOOSE_COLOR(Color4fv, pfv, 4, MASK_COLOR, 0, + (const GLfloat *v), (v)) +CHOOSE_COLOR(Color3f, p3f, 3, MASK_COLOR, 0, + (GLfloat a,GLfloat b, GLfloat c), (a,b,c)) +CHOOSE_COLOR(Color3fv, pfv, 3, MASK_COLOR, 0, + (const GLfloat *v), (v)) + + +CHOOSE(SecondaryColor3ubEXT, p3ub, MASK_SPEC, 0, + (GLubyte a,GLubyte b, GLubyte c), (a,b,c)) +CHOOSE(SecondaryColor3ubvEXT, pubv, MASK_SPEC, 0, + (const GLubyte *v), (v)) +CHOOSE(SecondaryColor3fEXT, p3f, MASK_SPEC, 0, + (GLfloat a,GLfloat b, GLfloat c), (a,b,c)) +CHOOSE(SecondaryColor3fvEXT, pfv, MASK_SPEC, 0, + (const GLfloat *v), (v)) + +CHOOSE(TexCoord2f, p2f, ~0, MASK_ST0, + (GLfloat a,GLfloat b), (a,b)) +CHOOSE(TexCoord2fv, pfv, ~0, MASK_ST0, + (const GLfloat *v), (v)) +CHOOSE(TexCoord1f, p1f, ~0, MASK_ST0, + (GLfloat a), (a)) +CHOOSE(TexCoord1fv, pfv, ~0, MASK_ST0, + (const GLfloat *v), (v)) + +CHOOSE(MultiTexCoord2fARB, pe2f, ~0, ~0, + (GLenum u,GLfloat a,GLfloat b), (u,a,b)) +CHOOSE(MultiTexCoord2fvARB, pefv, ~0, ~0, + (GLenum u,const GLfloat *v), (u,v)) +CHOOSE(MultiTexCoord1fARB, pe1f, ~0, ~0, + (GLenum u,GLfloat a), (u,a)) +CHOOSE(MultiTexCoord1fvARB, pefv, ~0, ~0, + (GLenum u,const GLfloat *v), (u,v)) + +CHOOSE(Vertex3f, p3f, ~0, ~0, + (GLfloat a,GLfloat b,GLfloat c), (a,b,c)) +CHOOSE(Vertex3fv, pfv, ~0, ~0, + (const GLfloat *v), (v)) +CHOOSE(Vertex2f, p2f, ~0, ~0, + (GLfloat a,GLfloat b), (a,b)) +CHOOSE(Vertex2fv, pfv, ~0, ~0, + (const GLfloat *v), (v)) + + + + + +void r200VtxfmtInitChoosers( GLvertexformat *vfmt ) +{ + vfmt->Color3f = choose_Color3f; + vfmt->Color3fv = choose_Color3fv; + vfmt->Color3ub = choose_Color3ub; + vfmt->Color3ubv = choose_Color3ubv; + vfmt->Color4f = choose_Color4f; + vfmt->Color4fv = choose_Color4fv; + vfmt->Color4ub = choose_Color4ub; + vfmt->Color4ubv = choose_Color4ubv; + vfmt->SecondaryColor3fEXT = choose_SecondaryColor3fEXT; + vfmt->SecondaryColor3fvEXT = choose_SecondaryColor3fvEXT; + vfmt->SecondaryColor3ubEXT = choose_SecondaryColor3ubEXT; + vfmt->SecondaryColor3ubvEXT = choose_SecondaryColor3ubvEXT; + vfmt->MultiTexCoord1fARB = choose_MultiTexCoord1fARB; + vfmt->MultiTexCoord1fvARB = choose_MultiTexCoord1fvARB; + vfmt->MultiTexCoord2fARB = choose_MultiTexCoord2fARB; + vfmt->MultiTexCoord2fvARB = choose_MultiTexCoord2fvARB; + vfmt->Normal3f = choose_Normal3f; + vfmt->Normal3fv = choose_Normal3fv; + vfmt->TexCoord1f = choose_TexCoord1f; + vfmt->TexCoord1fv = choose_TexCoord1fv; + vfmt->TexCoord2f = choose_TexCoord2f; + vfmt->TexCoord2fv = choose_TexCoord2fv; + vfmt->Vertex2f = choose_Vertex2f; + vfmt->Vertex2fv = choose_Vertex2fv; + vfmt->Vertex3f = choose_Vertex3f; + vfmt->Vertex3fv = choose_Vertex3fv; +} + + +static struct dynfn *codegen_noop( GLcontext *ctx, const int *key ) +{ + (void) ctx; (void) key; + return 0; +} + +void r200InitCodegen( struct dfn_generators *gen ) +{ + gen->Vertex3f = codegen_noop; + gen->Vertex3fv = codegen_noop; + gen->Color4ub = codegen_noop; + gen->Color4ubv = codegen_noop; + gen->Normal3f = codegen_noop; + gen->Normal3fv = codegen_noop; + gen->TexCoord2f = codegen_noop; + gen->TexCoord2fv = codegen_noop; + gen->MultiTexCoord2fARB = codegen_noop; + gen->MultiTexCoord2fvARB = codegen_noop; + gen->Vertex2f = codegen_noop; + gen->Vertex2fv = codegen_noop; + gen->Color3ub = codegen_noop; + gen->Color3ubv = codegen_noop; + gen->Color4f = codegen_noop; + gen->Color4fv = codegen_noop; + gen->Color3f = codegen_noop; + gen->Color3fv = codegen_noop; + gen->SecondaryColor3fEXT = codegen_noop; + gen->SecondaryColor3fvEXT = codegen_noop; + gen->SecondaryColor3ubEXT = codegen_noop; + gen->SecondaryColor3ubvEXT = codegen_noop; + gen->TexCoord1f = codegen_noop; + gen->TexCoord1fv = codegen_noop; + gen->MultiTexCoord1fARB = codegen_noop; + gen->MultiTexCoord1fvARB = codegen_noop; + + if (!getenv("R200_NO_CODEGEN")) { +#if defined(USE_X86_ASM) + r200InitX86Codegen( gen ); +#endif + +#if defined(USE_SSE_ASM) + r200InitSSECodegen( gen ); +#endif + } +} diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_sse.c b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_sse.c new file mode 100644 index 000000000..b375574fd --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_sse.c @@ -0,0 +1,96 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include <stdio.h> +#include <assert.h> +#include "mem.h" +#include "simple_list.h" +#include "r200_vtxfmt.h" + +#if defined(USE_SSE_ASM) + +/* Build specialized versions of the immediate calls on the fly for + * the current state. ???P4 SSE2 versions??? + */ + + +static struct dynfn *makeSSENormal3fv( GLcontext *ctx, const int *key ) +{ + /* Requires P4 (sse2?) + */ + static unsigned char temp[] = { + 0x8b, 0x44, 0x24, 0x04, /* mov 0x4(%esp,1),%eax */ + 0xba, 0x78, 0x56, 0x34, 0x12, /* mov $0x12345678,%edx */ + 0xf3, 0x0f, 0x7e, 0x00, /* movq (%eax),%xmm0 */ + 0x66, 0x0f, 0x6e, 0x48, 0x08, /* movd 0x8(%eax),%xmm1 */ + 0x66, 0x0f, 0xd6, 0x42, 0x0c, /* movq %xmm0,0xc(%edx) */ + 0x66, 0x0f, 0x7e, 0x4a, 0x14, /* movd %xmm1,0x14(%edx) */ + 0xc3, /* ret */ + }; + + + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + insert_at_head( &rmesa->vb.dfn_cache.Normal3fv, dfn ); + dfn->key[0] = key[0]; + dfn->key[1] = key[1]; + + dfn->code = ALIGN_MALLOC( sizeof(temp), 16 ); + memcpy (dfn->code, temp, sizeof(temp)); + FIXUP(dfn->code, 5, 0x0, (int)vb.normalptr); + return dfn; +} + +void r200InitSSECodegen( struct dfn_generators *gen ) +{ + /* Need to: + * - check kernel sse support + * - check p4/sse2 + */ + (void) makeSSENormal3fv; +} + + +#else + +void r200InitX86Codegen( struct dfn_generators *gen ) +{ + (void) gen; +} + +#endif + + + + diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_x86.c b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_x86.c new file mode 100644 index 000000000..28978e35b --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_vtxfmt_x86.c @@ -0,0 +1,463 @@ +/* $XFree86$ */ +/* +Copyright (C) The Weather Channel, Inc. 2002. All Rights Reserved. + +The Weather Channel (TM) funded Tungsten Graphics to develop the +initial release of the Radeon 8500 driver under the XFree86 license. +This notice must be preserved. + +Permission is hereby granted, free of charge, to any person obtaining +a copy of this software and associated documentation files (the +"Software"), to deal in the Software without restriction, including +without limitation the rights to use, copy, modify, merge, publish, +distribute, sublicense, and/or sell copies of the Software, and to +permit persons to whom the Software is furnished to do so, subject to +the following conditions: + +The above copyright notice and this permission notice (including the +next paragraph) shall be included in all copies or substantial +portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, +EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF +MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. +IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE +LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION +OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION +WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. +*/ + +/* + * Authors: + * Keith Whitwell <keith@tungstengraphics.com> + */ + +#include <stdio.h> +#include <assert.h> +#include "mem.h" +#include "mmath.h" +#include "simple_list.h" +#include "r200_vtxfmt.h" + +#if defined(USE_X86_ASM) + +#define EXTERN( FUNC ) \ +extern const char *FUNC; \ +extern const char *FUNC##_end + +EXTERN ( _x86_Normal3fv ); +EXTERN ( _x86_Normal3f ); +EXTERN ( _x86_Vertex3fv_6 ); +EXTERN ( _x86_Vertex3fv_8 ); +EXTERN ( _x86_Vertex3fv ); +EXTERN ( _x86_Vertex3f_4 ); +EXTERN ( _x86_Vertex3f_6 ); +EXTERN ( _x86_Vertex3f ); +EXTERN ( _x86_Color4ubv_ub ); +EXTERN ( _x86_Color4ubv_4f ); +EXTERN ( _x86_Color4ub_ub ); +EXTERN ( _x86_Color3fv_3f ); +EXTERN ( _x86_Color3f_3f ); +EXTERN ( _x86_TexCoord2fv ); +EXTERN ( _x86_TexCoord2f ); +EXTERN ( _x86_MultiTexCoord2fvARB ); +EXTERN ( _x86_MultiTexCoord2fvARB_2 ); +EXTERN ( _x86_MultiTexCoord2fARB ); +EXTERN ( _x86_MultiTexCoord2fARB_2 ); + + +/* Build specialized versions of the immediate calls on the fly for + * the current state. Generic x86 versions. + */ + +struct dynfn *r200_makeX86Vertex3f( GLcontext *ctx, const int *key ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x 0x%08x %d\n", __FUNCTION__, + key[0], key[1], vb.vertex_size ); + + switch (vb.vertex_size) { + case 4: { + + DFN ( _x86_Vertex3f_4, rmesa->vb.dfn_cache.Vertex3f ); + FIXUP(dfn->code, 2, 0x0, (int)&vb.dmaptr); + FIXUP(dfn->code, 25, 0x0, (int)&vb.vertex[3]); + FIXUP(dfn->code, 36, 0x0, (int)&vb.counter); + FIXUP(dfn->code, 46, 0x0, (int)&vb.dmaptr); + FIXUP(dfn->code, 51, 0x0, (int)&vb.counter); + FIXUP(dfn->code, 60, 0x0, (int)&vb.notify); + break; + } + case 6: { + + DFN ( _x86_Vertex3f_6, rmesa->vb.dfn_cache.Vertex3f ); + FIXUP(dfn->code, 3, 0x0, (int)&vb.dmaptr); + FIXUP(dfn->code, 28, 0x0, (int)&vb.vertex[3]); + FIXUP(dfn->code, 34, 0x0, (int)&vb.vertex[4]); + FIXUP(dfn->code, 40, 0x0, (int)&vb.vertex[5]); + FIXUP(dfn->code, 57, 0x0, (int)&vb.counter); + FIXUP(dfn->code, 63, 0x0, (int)&vb.dmaptr); + FIXUP(dfn->code, 70, 0x0, (int)&vb.counter); + FIXUP(dfn->code, 79, 0x0, (int)&vb.notify); + break; + } + default: { + + DFN ( _x86_Vertex3f, rmesa->vb.dfn_cache.Vertex3f ); + FIXUP(dfn->code, 3, 0x0, (int)&vb.vertex[3]); + FIXUP(dfn->code, 9, 0x0, (int)&vb.dmaptr); + FIXUP(dfn->code, 37, 0x0, vb.vertex_size-3); + FIXUP(dfn->code, 44, 0x0, (int)&vb.counter); + FIXUP(dfn->code, 50, 0x0, (int)&vb.dmaptr); + FIXUP(dfn->code, 56, 0x0, (int)&vb.counter); + FIXUP(dfn->code, 67, 0x0, (int)&vb.notify); + break; + } + } + + return dfn; +} + + + +struct dynfn *r200_makeX86Vertex3fv( GLcontext *ctx, const int *key ) +{ + r200ContextPtr rmesa = R200_CONTEXT(ctx); + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x 0x%08x %d\n", __FUNCTION__, + key[0], key[1], vb.vertex_size ); + + switch (vb.vertex_size) { + case 6: { + + DFN ( _x86_Vertex3fv_6, rmesa->vb.dfn_cache.Vertex3fv ); + FIXUP(dfn->code, 1, 0x00000000, (int)&vb.dmaptr); + FIXUP(dfn->code, 27, 0x0000001c, (int)&vb.vertex[3]); + FIXUP(dfn->code, 33, 0x00000020, (int)&vb.vertex[4]); + FIXUP(dfn->code, 45, 0x00000024, (int)&vb.vertex[5]); + FIXUP(dfn->code, 56, 0x00000000, (int)&vb.dmaptr); + FIXUP(dfn->code, 61, 0x00000004, (int)&vb.counter); + FIXUP(dfn->code, 67, 0x00000004, (int)&vb.counter); + FIXUP(dfn->code, 76, 0x00000008, (int)&vb.notify); + break; + } + + + case 8: { + + DFN ( _x86_Vertex3fv_8, rmesa->vb.dfn_cache.Vertex3fv ); + FIXUP(dfn->code, 1, 0x00000000, (int)&vb.dmaptr); + FIXUP(dfn->code, 27, 0x0000001c, (int)&vb.vertex[3]); + FIXUP(dfn->code, 33, 0x00000020, (int)&vb.vertex[4]); + FIXUP(dfn->code, 45, 0x0000001c, (int)&vb.vertex[5]); + FIXUP(dfn->code, 51, 0x00000020, (int)&vb.vertex[6]); + FIXUP(dfn->code, 63, 0x00000024, (int)&vb.vertex[7]); + FIXUP(dfn->code, 74, 0x00000000, (int)&vb.dmaptr); + FIXUP(dfn->code, 79, 0x00000004, (int)&vb.counter); + FIXUP(dfn->code, 85, 0x00000004, (int)&vb.counter); + FIXUP(dfn->code, 94, 0x00000008, (int)&vb.notify); + break; + } + + + + default: { + + DFN ( _x86_Vertex3fv, rmesa->vb.dfn_cache.Vertex3fv ); + FIXUP(dfn->code, 8, 0x01010101, (int)&vb.dmaptr); + FIXUP(dfn->code, 32, 0x00000006, vb.vertex_size-3); + FIXUP(dfn->code, 37, 0x00000058, (int)&vb.vertex[3]); + FIXUP(dfn->code, 45, 0x01010101, (int)&vb.dmaptr); + FIXUP(dfn->code, 50, 0x02020202, (int)&vb.counter); + FIXUP(dfn->code, 58, 0x02020202, (int)&vb.counter); + FIXUP(dfn->code, 67, 0x0, (int)&vb.notify); + break; + } + } + + return dfn; +} + +struct dynfn *r200_makeX86Normal3fv( GLcontext *ctx, const int *key ) +{ + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + int i = 0; + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] ); + + DFN ( _x86_Normal3fv, rmesa->vb.dfn_cache.Normal3fv ); + + FIXUP2(dfn->code, i, 0x0, (int)vb.normalptr); + FIXUP2(dfn->code, i, 0x4, 4+(int)vb.normalptr); + FIXUP2(dfn->code, i, 0x8, 8+(int)vb.normalptr); + fprintf(stderr, "%s done\n", __FUNCTION__); + return dfn; +} + +struct dynfn *r200_makeX86Normal3f( GLcontext *ctx, const int *key ) +{ + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] ); + + DFN ( _x86_Normal3f, rmesa->vb.dfn_cache.Normal3f ); + FIXUP(dfn->code, 1, 0x12345678, (int)vb.normalptr); + return dfn; +} + +struct dynfn *r200_makeX86Color4ubv( GLcontext *ctx, const int *key ) +{ + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] ); + + if (VTX_COLOR(key[0],0) == R200_VTX_PK_RGBA) { + DFN ( _x86_Color4ubv_ub, rmesa->vb.dfn_cache.Color4ubv); + FIXUP(dfn->code, 5, 0x12345678, (int)vb.ubytecolorptr); + return dfn; + } + else { + + DFN ( _x86_Color4ubv_4f, rmesa->vb.dfn_cache.Color4ubv); + FIXUP(dfn->code, 2, 0x00000000, (int)_mesa_ubyte_to_float_color_tab); + FIXUP(dfn->code, 27, 0xdeadbeaf, (int)vb.floatcolorptr); + FIXUP(dfn->code, 33, 0xdeadbeaf, (int)vb.floatcolorptr+4); + FIXUP(dfn->code, 55, 0xdeadbeaf, (int)vb.floatcolorptr+8); + FIXUP(dfn->code, 61, 0xdeadbeaf, (int)vb.floatcolorptr+12); + return dfn; + } +} + +struct dynfn *r200_makeX86Color4ub( GLcontext *ctx, const int *key ) +{ + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] ); + + if (VTX_COLOR(key[0],0) == R200_VTX_PK_RGBA) { + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + DFN ( _x86_Color4ub_ub, rmesa->vb.dfn_cache.Color4ub ); + FIXUP(dfn->code, 18, 0x0, (int)vb.ubytecolorptr); + FIXUP(dfn->code, 24, 0x0, (int)vb.ubytecolorptr+1); + FIXUP(dfn->code, 30, 0x0, (int)vb.ubytecolorptr+2); + FIXUP(dfn->code, 36, 0x0, (int)vb.ubytecolorptr+3); + return dfn; + } + else + return 0; +} + + +struct dynfn *r200_makeX86Color3fv( GLcontext *ctx, const int *key ) +{ + if (VTX_COLOR(key[0],0) != R200_VTX_FP_RGB) + return 0; + else + { + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] ); + + DFN ( _x86_Color3fv_3f, rmesa->vb.dfn_cache.Color3fv ); + FIXUP(dfn->code, 5, 0x0, (int)vb.floatcolorptr); + return dfn; + } +} + +struct dynfn *r200_makeX86Color3f( GLcontext *ctx, const int *key ) +{ + if (VTX_COLOR(key[0],0) != R200_VTX_FP_RGB) + return 0; + else + { + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x\n", __FUNCTION__, key[0] ); + + DFN ( _x86_Color3f_3f, rmesa->vb.dfn_cache.Color3f ); + FIXUP(dfn->code, 1, 0x12345678, (int)vb.floatcolorptr); + return dfn; + } +} + + + +struct dynfn *r200_makeX86TexCoord2fv( GLcontext *ctx, const int *key ) +{ + + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x 0x%08x\n", __FUNCTION__, key[0], key[1] ); + + DFN ( _x86_TexCoord2fv, rmesa->vb.dfn_cache.TexCoord2fv ); + FIXUP(dfn->code, 5, 0x12345678, (int)vb.texcoordptr[0]); + return dfn; +} + +struct dynfn *r200_makeX86TexCoord2f( GLcontext *ctx, const int *key ) +{ + + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x 0x%08x\n", __FUNCTION__, key[0], key[1] ); + + DFN ( _x86_TexCoord2f, rmesa->vb.dfn_cache.TexCoord2f ); + FIXUP(dfn->code, 1, 0x12345678, (int)vb.texcoordptr[0]); + return dfn; +} + +struct dynfn *r200_makeX86MultiTexCoord2fvARB( GLcontext *ctx, const int *key ) +{ +#if 0 + static char temp[] = { + 0x8b, 0x44, 0x24, 0x04, /* mov 0x4(%esp,1),%eax */ + 0x8b, 0x4c, 0x24, 0x08, /* mov 0x8(%esp,1),%ecx */ + 0x2d, 0xc0, 0x84, 0x00, 0x00, /* sub $0x84c0,%eax */ + 0x83, 0xe0, 0x01, /* and $0x1,%eax */ + 0x8b, 0x11, /* mov (%ecx),%edx */ + 0xc1, 0xe0, 0x03, /* shl $0x3,%eax */ + 0x8b, 0x49, 0x04, /* mov 0x4(%ecx),%ecx */ + 0x89, 0x90, 0, 0, 0, 0,/* mov %edx,DEST(%eax) */ + 0x89, 0x88, 0, 0, 0, 0,/* mov %ecx,DEST+8(%eax) */ + 0xc3, /* ret */ + }; + static char temp2[] = { + 0x8b, 0x44, 0x24, 0x04, /* mov 0x4(%esp,1),%eax */ + 0x8b, 0x4c, 0x24, 0x08, /* mov 0x8(%esp,1),%ecx */ + 0x2d, 0xc0, 0x84, 0x00, 0x00, /* sub $0x84c0,%eax */ + 0x83, 0xe0, 0x01, /* and $0x1,%eax */ + 0x8b, 0x14, 0x85, 0, 0, 0, 0, /* mov DEST(,%eax,4),%edx */ + 0x8b, 0x01, /* mov (%ecx),%eax */ + 0x89, 0x02, /* mov %eax,(%edx) */ + 0x8b, 0x41, 0x04, /* mov 0x4(%ecx),%eax */ + 0x89, 0x42, 0x04, /* mov %eax,0x4(%edx) */ + 0xc3, /* ret */ + }; +#endif + + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x 0x%08x\n", __FUNCTION__, key[0], key[1] ); + + if (vb.texcoordptr[1] == vb.texcoordptr[0]+4) { + DFN ( _x86_MultiTexCoord2fvARB, rmesa->vb.dfn_cache.MultiTexCoord2fvARB ); + FIXUP(dfn->code, 26, 0xdeadbeef, (int)vb.texcoordptr[0]); + FIXUP(dfn->code, 32, 0xdeadbeef, (int)vb.texcoordptr[0]+4); + } else { + DFN ( _x86_MultiTexCoord2fvARB_2, rmesa->vb.dfn_cache.MultiTexCoord2fvARB ); + FIXUP(dfn->code, 19, 0x0, (int)vb.texcoordptr); + } + return dfn; +} + +struct dynfn *r200_makeX86MultiTexCoord2fARB( GLcontext *ctx, + const int *key ) +{ +#if 0 + static char temp[] = { + 0x8b, 0x44, 0x24, 0x04, /* mov 0x4(%esp,1),%eax */ + 0x8b, 0x54, 0x24, 0x08, /* mov 0x8(%esp,1),%edx */ + 0x2d, 0xc0, 0x84, 0x00, 0x00, /* sub $0x84c0,%eax */ + 0x8b, 0x4c, 0x24, 0x0c, /* mov 0xc(%esp,1),%ecx */ + 0x83, 0xe0, 0x01, /* and $0x1,%eax */ + 0xc1, 0xe0, 0x03, /* shl $0x3,%eax */ + 0x89, 0x90, 0, 0, 0, 0, /* mov %edx,DEST(%eax) */ + 0x89, 0x88, 0, 0, 0, 0, /* mov %ecx,DEST+8(%eax) */ + 0xc3, /* ret */ + }; + + static char temp2[] = { + 0x8b, 0x44, 0x24, 0x04, /* mov 0x4(%esp,1),%eax */ + 0x8b, 0x54, 0x24, 0x08, /* mov 0x8(%esp,1),%edx */ + 0x2d, 0xc0, 0x84, 0x00, 0x00, /* sub $0x84c0,%eax */ + 0x8b, 0x4c, 0x24, 0x0c, /* mov 0xc(%esp,1),%ecx */ + 0x83, 0xe0, 0x01, /* and $0x1,%eax */ + 0x8b, 0x04, 0x85, 0, 0, 0, 0, /* mov DEST(,%eax,4),%eax */ + 0x89, 0x10, /* mov %edx,(%eax) */ + 0x89, 0x48, 0x04, /* mov %ecx,0x4(%eax) */ + 0xc3, /* ret */ + }; +#endif + struct dynfn *dfn = MALLOC_STRUCT( dynfn ); + r200ContextPtr rmesa = R200_CONTEXT(ctx); + + if (R200_DEBUG & DEBUG_CODEGEN) + fprintf(stderr, "%s 0x%08x 0x%08x\n", __FUNCTION__, key[0], key[1] ); + + if (vb.texcoordptr[1] == vb.texcoordptr[0]+4) { + DFN ( _x86_MultiTexCoord2fARB, rmesa->vb.dfn_cache.MultiTexCoord2fARB ); + FIXUP(dfn->code, 25, 0xdeadbeef, (int)vb.texcoordptr[0]); + FIXUP(dfn->code, 31, 0xdeadbeef, (int)vb.texcoordptr[0]+4); + } + else { + /* Note: this might get generated multiple times, even though the + * actual emitted code is the same. + */ + DFN ( _x86_MultiTexCoord2fARB_2, rmesa->vb.dfn_cache.MultiTexCoord2fARB ); + FIXUP(dfn->code, 23, 0x0, (int)vb.texcoordptr); + } + return dfn; +} + + +void r200InitX86Codegen( struct dfn_generators *gen ) +{ + gen->Vertex3f = r200_makeX86Vertex3f; + gen->Vertex3fv = r200_makeX86Vertex3fv; + gen->Color4ub = r200_makeX86Color4ub; /* PKCOLOR only */ + gen->Color4ubv = r200_makeX86Color4ubv; /* PKCOLOR only */ + gen->Normal3f = r200_makeX86Normal3f; + gen->Normal3fv = r200_makeX86Normal3fv; + gen->TexCoord2f = r200_makeX86TexCoord2f; + gen->TexCoord2fv = r200_makeX86TexCoord2fv; + gen->MultiTexCoord2fARB = r200_makeX86MultiTexCoord2fARB; + gen->MultiTexCoord2fvARB = r200_makeX86MultiTexCoord2fvARB; + gen->Color3f = r200_makeX86Color3f; + gen->Color3fv = r200_makeX86Color3fv; + + /* Not done: + */ +/* gen->Vertex2f = r200_makeX86Vertex2f; */ +/* gen->Vertex2fv = r200_makeX86Vertex2fv; */ +/* gen->Color3ub = r200_makeX86Color3ub; */ +/* gen->Color3ubv = r200_makeX86Color3ubv; */ +/* gen->Color4f = r200_makeX86Color4f; */ +/* gen->Color4fv = r200_makeX86Color4fv; */ +/* gen->TexCoord1f = r200_makeX86TexCoord1f; */ +/* gen->TexCoord1fv = r200_makeX86TexCoord1fv; */ +/* gen->MultiTexCoord1fARB = r200_makeX86MultiTexCoord1fARB; */ +/* gen->MultiTexCoord1fvARB = r200_makeX86MultiTexCoord1fvARB; */ +} + + +#else + +void r200InitX86Codegen( struct dfn_generators *gen ) +{ + (void) gen; +} + +#endif diff --git a/xc/lib/GL/mesa/src/drv/r200/r200_vtxtmp_x86.S b/xc/lib/GL/mesa/src/drv/r200/r200_vtxtmp_x86.S new file mode 100644 index 000000000..cfb0ecbd8 --- /dev/null +++ b/xc/lib/GL/mesa/src/drv/r200/r200_vtxtmp_x86.S @@ -0,0 +1,408 @@ +/* $XFree86$ */ +/************************************************************************** + +Copyright 2002 Tungsten Graphics Inc., Cedar Park, Texas. + +All Rights Reserved. + +Permission is hereby granted, free of charge, to any person obtaining a +copy of this software and associated documentation files (the "Software"), +to deal in the Software without restriction, including without limitation +on the rights to use, copy, modify, merge, publish, distribute, sub +license, and/or sell copies of the Software, and to permit persons to whom +the Software is furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice (including the next +paragraph) shall be included in all copies or substantial portions of the +Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL +ATI, TUNGSTEN GRAPHICS AND/OR THEIR SUPPLIERS BE LIABLE FOR ANY CLAIM, +DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR +OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE +USE OR OTHER DEALINGS IN THE SOFTWARE. + +**************************************************************************/ + +#define GLOBL( x ) \ +.globl x; \ +x##: + +.data +.align 4 +GLOBL( _x86_Normal3fv) + movl 4(%esp), %eax /* load 'v' off stack */ + movl (%eax), %ecx /* load v[0] */ + movl 4(%eax), %edx /* load v[1] */ + movl 8(%eax), %eax /* load v[2] */ + movl %ecx, 0 /* store v[0] to current vertex */ + movl %edx, 4 /* store v[1] to current vertex */ + movl %eax, 8 /* store v[2] to current vertex */ + ret +GLOBL ( _x86_Normal3fv_end ) + +/* + vertex 3f vertex size 4 +*/ + +GLOBL ( _x86_Vertex3f_4 ) + movl (0), %ecx + movl 4(%esp), %eax + movl 8(%esp), %edx + movl %eax, (%ecx) + movl %edx, 4(%ecx) + movl 12(%esp), %eax + movl (0), %edx + movl %eax, 8(%ecx) + movl %edx, 12(%ecx) + movl (0), %eax + addl $16, %ecx + dec %eax + movl %ecx, (0) + movl %eax, (0) + je .1 + ret +.1: jmp *0 + +GLOBL ( _x86_Vertex3f_4_end ) + +/* + vertex 3f vertex size 6 +*/ +GLOBL ( _x86_Vertex3f_6 ) + push %edi + movl (0), %edi + movl 8(%esp), %eax + movl 12(%esp), %edx + movl 16(%esp), %ecx + movl %eax, (%edi) + movl %edx, 4(%edi) + movl %ecx, 8(%edi) + movl (0), %eax + movl (0), %edx + movl (0), %ecx + movl %eax, 12(%edi) + movl %edx, 16(%edi) + movl %ecx, 20(%edi) + addl $24, %edi + movl (0), %eax + movl %edi, (0) + dec %eax + pop %edi + movl %eax, (0) + je .2 + ret +.2: jmp *0 +GLOBL ( _x86_Vertex3f_6_end ) +/* + vertex 3f generic size +*/ +GLOBL ( _x86_Vertex3f ) + push %edi + push %esi + movl $0, %esi + movl (0), %edi + movl 12(%esp), %eax + movl 16(%esp), %edx + movl 20(%esp), %ecx + movl %eax, (%edi) + movl %edx, 4(%edi) + movl %ecx, 8(%edi) + addl $12, %edi + movl $0, %ecx + repz movsl %ds:(%esi), %es:(%edi) + movl (0), %eax + movl %edi, (0) + dec %eax + movl %eax, (0) + pop %esi + pop %edi + je .3 + ret +.3: jmp *0 + +GLOBL ( _x86_Vertex3f_end ) + +/* + Vertex 3fv vertex size 6 +*/ +GLOBL ( _x86_Vertex3fv_6 ) + movl (0), %eax + movl 4(%esp), %ecx + movl (%ecx), %edx + movl %edx, (%eax) + movl 4(%ecx), %edx + movl 8(%ecx), %ecx + movl %edx, 4(%eax) + movl %ecx, 8(%eax) + movl (28), %edx + movl (32), %ecx + movl %edx, 12(%eax) + movl %ecx, 16(%eax) + movl (36), %edx + movl %edx, 20(%eax) + addl $24, %eax + movl %eax, 0 + movl 4, %eax + dec %eax + movl %eax, 4 + je .4 + ret +.4: jmp *8 + +GLOBL ( _x86_Vertex3fv_6_end ) + +/* + Vertex 3fv vertex size 8 +*/ +GLOBL ( _x86_Vertex3fv_8 ) + movl (0), %eax + movl 4(%esp), %ecx + movl (%ecx), %edx + movl %edx ,(%eax) + movl 4(%ecx) ,%edx + movl 8(%ecx) ,%ecx + movl %edx, 4(%eax) + movl %ecx, 8(%eax) + movl (28), %edx + movl (32), %ecx + movl %edx, 12(%eax) + movl %ecx, 16(%eax) + movl (28), %edx + movl (32), %ecx + movl %edx, 20(%eax) + movl %ecx, 24(%eax) + movl (36), %edx + movl %edx, 28(%eax) + addl $32, %eax + movl %eax, (0) + movl 4, %eax + dec %eax + movl %eax, (4) + je .5 + ret +.5: jmp *8 + +GLOBL ( _x86_Vertex3fv_8_end ) + +/* + Vertex 3fv generic vertex size +*/ +GLOBL ( _x86_Vertex3fv ) + movl 4(%esp), %edx + push %edi + push %esi + movl (0x1010101), %edi + movl (%edx), %eax + movl 4(%edx), %ecx + movl 8(%edx), %esi + movl %eax, (%edi) + movl %ecx, 4(%edi) + movl %esi, 8(%edi) + addl $12, %edi + movl $6, %ecx + movl $0x58, %esi + repz movsl %ds:(%esi), %es:(%edi) + movl %edi, (0x1010101) + movl (0x2020202), %eax + pop %esi + pop %edi + dec %eax + movl %eax, (0x2020202) + je .6 + ret +.6: jmp *0 +GLOBL ( _x86_Vertex3fv_end ) + +/* + Normal 3f +*/ +GLOBL ( _x86_Normal3f ) + movl $0x12345678, %edx + movl 4(%esp), %eax + movl %eax, (%edx) + movl 8(%esp), %eax + movl %eax, 4(%edx) + movl 12(%esp), %eax + movl %eax, 8(%edx) + ret +GLOBL ( _x86_Normal3f_end ) + +/* + Color 4ubv_ub +*/ +GLOBL ( _x86_Color4ubv_ub ) + movl 4(%esp), %eax + movl $0x12345678, %edx + movl (%eax), %eax + movl %eax, (%edx) + ret +GLOBL ( _x86_Color4ubv_ub_end ) + +/* + Color 4ubv 4f +*/ +GLOBL ( _x86_Color4ubv_4f ) + push %ebx + movl $0, %edx + xor %eax, %eax + xor %ecx, %ecx + movl 8(%esp), %ebx + movl (%ebx), %ebx + mov %bl, %al + mov %bh, %cl + movl (%edx,%eax,4),%eax + movl (%edx,%ecx,4),%ecx + movl %eax, (0xdeadbeaf) + movl %ecx, (0xdeadbeaf) + xor %eax, %eax + xor %ecx, %ecx + shr $16, %ebx + mov %bl, %al + mov %bh, %cl + movl (%edx,%eax,4), %eax + movl (%edx,%ecx,4), %ecx + movl %eax, (0xdeadbeaf) + movl %ecx, (0xdeadbeaf) + pop %ebx + ret +GLOBL ( _x86_Color4ubv_4f_end ) + +/* + + Color4ub_ub +*/ +GLOBL( _x86_Color4ub_ub ) + push %ebx + movl 8(%esp), %eax + movl 12(%esp), %edx + movl 16(%esp), %ecx + movl 20(%esp), %ebx + mov %al, (0) + mov %dl, (0) + mov %cl, (0) + mov %bl, (0) + pop %ebx + ret +GLOBL( _x86_Color4ub_ub_end ) + +/* + Color3fv_3f +*/ +GLOBL( _x86_Color3fv_3f ) + movl 4(%esp), %eax + movl $0, %edx + movl (%eax), %ecx + movl %ecx, (%edx) + movl 4(%eax), %ecx + movl %ecx, 4(%edx) + movl 8(%eax), %ecx + movl %ecx, 8(%edx) + ret +GLOBL( _x86_Color3fv_3f_end ) + +/* + Color3f_3f +*/ +GLOBL( _x86_Color3f_3f ) + movl $0x12345678, %edx + movl 4(%esp), %eax + movl %eax, (%edx) + movl 8(%esp,1), %eax + movl %eax, 4(%edx) + movl 12(%esp), %eax + movl %eax, 8(%edx) + ret +GLOBL( _x86_Color3f_3f_end ) + +/* + TexCoord2fv +*/ + +GLOBL( _x86_TexCoord2fv ) + movl 4(%esp), %eax + movl $0x12345678, %edx + movl (%eax), %ecx + movl 4(%eax), %eax + movl %ecx, (%edx) + movl %eax, 4(%edx) + ret + +GLOBL( _x86_TexCoord2fv_end ) +/* + TexCoord2f +*/ +GLOBL( _x86_TexCoord2f ) + movl $0x12345678, %edx + movl 4(%esp), %eax + movl 8(%esp), %ecx + movl %eax, (%edx) + movl %ecx, 4(%edx) + ret +GLOBL( _x86_TexCoord2f_end ) + +/* + MultiTexCoord2fvARB st0/st1 +*/ +GLOBL( _x86_MultiTexCoord2fvARB ) + + movl 4(%esp), %eax + movl 8(%esp), %ecx + sub $0x84c0, %eax + and $1, %eax + movl (%ecx), %edx + shl $3, %eax + movl 4(%ecx), %ecx + movl %edx, 0xdeadbeef(%eax) + movl %ecx, 0xdeadbeef(%eax) + ret +GLOBL( _x86_MultiTexCoord2fvARB_end ) +/* + MultiTexCoord2fvARB +*/ + +GLOBL( _x86_MultiTexCoord2fvARB_2 ) + movl 4(%esp,1), %eax + movl 8(%esp,1), %ecx + sub $0x84c0, %eax + and $0x1, %eax + movl 0(,%eax,4), %edx + movl (%ecx), %eax + movl %eax, (%edx) + movl 4(%ecx), %eax + movl %eax, 4(%edx) + ret + +GLOBL( _x86_MultiTexCoord2fvARB_2_end ) + +/* + MultiTexCoord2fARB st0/st1 +*/ +GLOBL( _x86_MultiTexCoord2fARB ) + movl 4(%esp), %eax + movl 8(%esp), %edx + sub $0x84c0, %eax + movl 12(%esp), %ecx + and $1, %eax + shl $3, %eax + movl %edx, 0xdeadbeef(%eax) + movl %ecx, 0xdeadbeef(%eax) + ret +GLOBL( _x86_MultiTexCoord2fARB_end ) + +/* + MultiTexCoord2fARB +*/ +GLOBL( _x86_MultiTexCoord2fARB_2 ) + movl 4(%esp), %eax + movl 8(%esp), %edx + sub $0x84c0, %eax + movl 12(%esp,1), %ecx + and $1,%eax + movl 0(,%eax,4), %eax + movl %edx, (%eax) + movl %ecx, 4(%eax) + ret +GLOBL( _x86_MultiTexCoord2fARB_2_end ) diff --git a/xc/lib/GL/mesa/src/drv/radeon/radeon_ioctl.c b/xc/lib/GL/mesa/src/drv/radeon/radeon_ioctl.c index cdd6c4f15..1d777ce00 100644 --- a/xc/lib/GL/mesa/src/drv/radeon/radeon_ioctl.c +++ b/xc/lib/GL/mesa/src/drv/radeon/radeon_ioctl.c @@ -730,6 +730,17 @@ void radeonPageFlip( const __DRIdrawablePrivate *dPriv ) LOCK_HARDWARE( rmesa ); + /* Need to do this for the perf box placement: + */ + if (rmesa->dri.drawable->numClipRects) + { + XF86DRIClipRectPtr box = rmesa->dri.drawable->pClipRects; + XF86DRIClipRectPtr b = rmesa->sarea->boxes; + b[0] = box[0]; + rmesa->sarea->nbox = 1; + } + + /* Throttle the frame rate -- only allow one pending swap buffers * request at a time. */ @@ -913,7 +924,8 @@ static void radeonClear( GLcontext *ctx, GLbitfield mask, GLboolean all, depth_boxes[n].f[RADEON_CLEAR_Y1] = (float)b[n].y1; depth_boxes[n].f[RADEON_CLEAR_X2] = (float)b[n].x2; depth_boxes[n].f[RADEON_CLEAR_Y2] = (float)b[n].y2; - depth_boxes[n].f[RADEON_CLEAR_DEPTH] = ctx->Depth.Clear; + depth_boxes[n].f[RADEON_CLEAR_DEPTH] = + (float)rmesa->state.depth.clear; } ret = drmCommandWrite( rmesa->dri.fd, DRM_RADEON_CLEAR, diff --git a/xc/lib/GL/mesa/src/drv/radeon/radeon_tcl.c b/xc/lib/GL/mesa/src/drv/radeon/radeon_tcl.c index e55a55a0d..e66c150b0 100644 --- a/xc/lib/GL/mesa/src/drv/radeon/radeon_tcl.c +++ b/xc/lib/GL/mesa/src/drv/radeon/radeon_tcl.c @@ -209,9 +209,9 @@ static void EMIT_PRIM( GLcontext *ctx, /* Try & join small primitives */ #if 0 -#define PREFER_DISCREET_ELT_PRIM( NR, PRIM ) 0 +#define PREFER_DISCRETE_ELT_PRIM( NR, PRIM ) 0 #else -#define PREFER_DISCREET_ELT_PRIM( NR, PRIM ) \ +#define PREFER_DISCRETE_ELT_PRIM( NR, PRIM ) \ ((NR) < 20 || \ ((NR) < 40 && \ rmesa->tcl.hw_primitive == (PRIM| \ diff --git a/xc/lib/GL/mesa/src/drv/tdfx/tdfx_tris.c b/xc/lib/GL/mesa/src/drv/tdfx/tdfx_tris.c index e5c51ad86..51239e97f 100644 --- a/xc/lib/GL/mesa/src/drv/tdfx/tdfx_tris.c +++ b/xc/lib/GL/mesa/src/drv/tdfx/tdfx_tris.c @@ -1167,24 +1167,6 @@ static void tdfxRenderFinish( GLcontext *ctx ) } - -/* - * These functions are used when we're software rendering, and - * lock/unlock the hardware (for span reading/writing). - */ -static void tdfxSwSetupStart( GLcontext *ctx ) -{ - tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx); - LOCK_HARDWARE(fxMesa); -} - -static void tdfxSwSetupFinish( GLcontext *ctx ) -{ - tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx); - UNLOCK_HARDWARE(fxMesa); -} - - /**********************************************************************/ /* Manage total rasterization fallbacks */ /**********************************************************************/ @@ -1257,7 +1239,6 @@ void tdfxFallback( GLcontext *ctx, GLuint bit, GLboolean mode ) void tdfxDDInitTriFuncs( GLcontext *ctx ) { TNLcontext *tnl = TNL_CONTEXT(ctx); - SScontext *swsetup = SWSETUP_CONTEXT(ctx); tdfxContextPtr fxMesa = TDFX_CONTEXT(ctx); static int firsttime = 1; @@ -1276,9 +1257,5 @@ void tdfxDDInitTriFuncs( GLcontext *ctx ) tnl->Driver.Render.BuildVertices = tdfxBuildVertices; tnl->Driver.Render.Multipass = NULL; - - swsetup->Driver.Start = tdfxSwSetupStart; - swsetup->Driver.Finish = tdfxSwSetupFinish; - (void) tdfx_print_vertex; } diff --git a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon.h b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon.h index 7b20e4a2c..68fa761bd 100644 --- a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon.h +++ b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon.h @@ -387,6 +387,7 @@ typedef struct { unsigned long agpOffset; unsigned char *AGP; /* Map */ int agpMode; + int agpFastWrite; CARD32 pciCommand; diff --git a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_common.h b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_common.h index ca046dd52..3d6198ca2 100644 --- a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_common.h +++ b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_common.h @@ -83,7 +83,8 @@ typedef struct { enum { DRM_RADEON_INIT_CP = 0x01, - DRM_RADEON_CLEANUP_CP = 0x02 + DRM_RADEON_CLEANUP_CP = 0x02, + DRM_RADEON_INIT_R200_CP = 0x03 } func; unsigned long sarea_priv_offset; int is_pci; @@ -296,7 +297,47 @@ typedef struct { #define RADEON_EMIT_SE_ZBIAS_FACTOR 18 /* zbias/2 */ #define RADEON_EMIT_SE_TCL_OUTPUT_VTX_FMT 19 /* tcl/11 */ #define RADEON_EMIT_SE_TCL_MATERIAL_EMMISSIVE_RED 20 /* material/17 */ -#define RADEON_MAX_STATE_PACKETS 21 +#define R200_EMIT_PP_TXCBLEND_0 21 /* tex0/4 */ +#define R200_EMIT_PP_TXCBLEND_1 22 /* tex1/4 */ +#define R200_EMIT_PP_TXCBLEND_2 23 /* tex2/4 */ +#define R200_EMIT_PP_TXCBLEND_3 24 /* tex3/4 */ +#define R200_EMIT_PP_TXCBLEND_4 25 /* tex4/4 */ +#define R200_EMIT_PP_TXCBLEND_5 26 /* tex5/4 */ +#define R200_EMIT_PP_TXCBLEND_6 27 /* /4 */ +#define R200_EMIT_PP_TXCBLEND_7 28 /* /4 */ +#define R200_EMIT_TCL_LIGHT_MODEL_CTL_0 29 /* tcl/6 */ +#define R200_EMIT_TFACTOR_0 30 /* tf/6 */ +#define R200_EMIT_VTX_FMT_0 31 /* vtx/4 */ +#define R200_EMIT_VAP_CTL 32 /* vap/1 */ +#define R200_EMIT_MATRIX_SELECT_0 33 /* msl/5 */ +#define R200_EMIT_TEX_PROC_CTL_2 34 /* tcg/5 */ +#define R200_EMIT_TCL_UCP_VERT_BLEND_CTL 35 /* tcl/1 */ +#define R200_EMIT_PP_TXFILTER_0 36 /* tex0/6 */ +#define R200_EMIT_PP_TXFILTER_1 37 /* tex1/6 */ +#define R200_EMIT_PP_TXFILTER_2 38 /* tex2/6 */ +#define R200_EMIT_PP_TXFILTER_3 39 /* tex3/6 */ +#define R200_EMIT_PP_TXFILTER_4 40 /* tex4/6 */ +#define R200_EMIT_PP_TXFILTER_5 41 /* tex5/6 */ +#define R200_EMIT_PP_TXOFFSET_0 42 /* tex0/1 */ +#define R200_EMIT_PP_TXOFFSET_1 43 /* tex1/1 */ +#define R200_EMIT_PP_TXOFFSET_2 44 /* tex2/1 */ +#define R200_EMIT_PP_TXOFFSET_3 45 /* tex3/1 */ +#define R200_EMIT_PP_TXOFFSET_4 46 /* tex4/1 */ +#define R200_EMIT_PP_TXOFFSET_5 47 /* tex5/1 */ +#define R200_EMIT_VTE_CNTL 48 /* vte/1 */ +#define R200_EMIT_OUTPUT_VTX_COMP_SEL 49 /* vtx/1 */ +#define R200_EMIT_PP_TAM_DEBUG3 50 /* tam/1 */ +#define R200_EMIT_PP_CNTL_X 51 /* cst/1 */ +#define R200_EMIT_RB3D_DEPTHXY_OFFSET 52 /* cst/1 */ +#define R200_EMIT_RE_AUX_SCISSOR_CNTL 53 /* cst/1 */ +#define R200_EMIT_RE_SCISSOR_TL_0 54 /* cst/2 */ +#define R200_EMIT_RE_SCISSOR_TL_1 55 /* cst/2 */ +#define R200_EMIT_RE_SCISSOR_TL_2 56 /* cst/2 */ +#define R200_EMIT_SE_VAP_CNTL_STATUS 57 /* cst/1 */ +#define R200_EMIT_SE_VTX_STATE_CNTL 58 /* cst/1 */ +#define R200_EMIT_RE_POINTSIZE 59 /* cst/1 */ +#define R200_EMIT_TCL_INPUT_VTX_VECTOR_ADDR_0 60 /* cst/4 */ +#define RADEON_MAX_STATE_PACKETS 61 /* Commands understood by cmd_buffer ioctl. More can be added but @@ -308,23 +349,24 @@ typedef struct { #define RADEON_CMD_DMA_DISCARD 4 /* discard current dma buf */ #define RADEON_CMD_PACKET3 5 /* emit hw packet */ #define RADEON_CMD_PACKET3_CLIP 6 /* emit hw packet wrapped in cliprects */ +#define RADEON_CMD_SCALARS2 7 /* R200 stopgap */ typedef union { int i; struct { - char cmd_type, pad0, pad1, pad2; + unsigned char cmd_type, pad0, pad1, pad2; } header; struct { - char cmd_type, packet_id, pad0, pad1; + unsigned char cmd_type, packet_id, pad0, pad1; } packet; struct { - char cmd_type, offset, stride, count; + unsigned char cmd_type, offset, stride, count; } scalars; struct { - char cmd_type, offset, stride, count; + unsigned char cmd_type, offset, stride, count; } vectors; struct { - char cmd_type, buf_idx, pad0, pad1; + unsigned char cmd_type, buf_idx, pad0, pad1; } dma; } drmRadeonCmdHeader; diff --git a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.c b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.c index aa2e6bf3d..b2f3b813b 100644 --- a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.c +++ b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.c @@ -341,202 +341,13 @@ static void RADEONDRISwapContext(ScreenPtr pScreen, DRISyncType syncType, } } -/* The Radeon has depth tiling on all the time, so we have to convert - * the x,y coordinates into the memory bus address (mba) in the same - * manner as the engine. In each case, the linear block address (ba) - * is calculated, and then wired with x and y to produce the final - * memory address. - */ -static CARD32 radeon_mba_z16(RADEONInfoPtr info, - int x, int y) -{ - CARD32 pitch = info->frontPitch; - CARD32 ba, address = 0; /* a[0] = 0 */ - - ba = (y / 16) * (pitch / 32) + (x / 32); - - address |= (x & 0x7) << 1; /* a[1..3] = x[0..2] */ - address |= (y & 0x7) << 4; /* a[4..6] = y[0..2] */ - address |= (x & 0x8) << 4; /* a[7] = x[3] */ - address |= (ba & 0x3) << 8; /* a[8..9] = ba[0..1] */ - address |= (y & 0x8) << 7; /* a[10] = y[3] */ - address |= ((x & 0x10) ^ (y & 0x10)) << 7; /* a[11] = x[4] ^ y[4] */ - address |= (ba & ~0x3u) << 10; /* a[12..] = ba[2..] */ - - return address; -} - -static CARD32 radeon_mba_z32(RADEONInfoPtr info, - int x, int y) -{ - CARD32 pitch = info->frontPitch; - CARD32 ba, address = 0; /* a[0..1] = 0 */ - - ba = (y / 16) * (pitch / 16) + (x / 16); - - address |= (x & 0x7) << 2; /* a[2..4] = x[0..2] */ - address |= (y & 0x3) << 5; /* a[5..6] = y[0..1] */ - address |= - (((x & 0x10) >> 2) ^ (y & 0x4)) << 5; /* a[7] = x[4] ^ y[2] */ - address |= (ba & 0x3) << 8; /* a[8..9] = ba[0..1] */ - - address |= (y & 0x8) << 7; /* a[10] = y[3] */ - address |= - (((x & 0x8) << 1) ^ (y & 0x10)) << 7; /* a[11] = x[3] ^ y[4] */ - address |= (ba & ~0x3u) << 10; /* a[12..] = ba[2..] */ - - return address; -} - -/* 16-bit depth buffer functions */ -#define WRITE_DEPTH16(_x, _y, d) \ - *(CARD16 *)(pointer)(buf + radeon_mba_z16(info, (_x), (_y))) = (d) - -#define READ_DEPTH16(d, _x, _y) \ - (d) = *(CARD16 *)(pointer)(buf + radeon_mba_z16(info, (_x), (_y))) - -/* 24 bit depth, 8 bit stencil depthbuffer functions */ -#define WRITE_DEPTH32(_x, _y, d) \ -do { \ - CARD32 tmp = \ - *(CARD32 *)(pointer)(buf + radeon_mba_z32(info, (_x), (_y))); \ - tmp &= 0xff000000; \ - tmp |= ((d) & 0x00ffffff); \ - *(CARD32 *)(pointer)(buf + radeon_mba_z32(info, (_x), (_y))) = tmp; \ -} while (0) - -#define READ_DEPTH32(d, _x, _y) \ - d = *(CARD32 *)(pointer)(buf + radeon_mba_z32(info, (_x), (_y))) & 0x00ffffff - -/* Screen to screen copy of data in the depth buffer */ -static void RADEONScreenToScreenCopyDepth(ScrnInfoPtr pScrn, - int xa, int ya, - int xb, int yb, - int w, int h) -{ - RADEONInfoPtr info = RADEONPTR(pScrn); - int xstart, xend, xdir; - int ystart, yend, ydir; - int x, y, d; - unsigned char *buf = info->FB + info->depthOffset; - - if (xa < xb) xdir = -1, xstart = w-1, xend = 0; - else xdir = 1, xstart = 0, xend = w-1; - - if (ya < yb) ydir = -1, ystart = h-1, yend = 0; - else ydir = 1, ystart = 0, yend = h-1; - - switch (pScrn->bitsPerPixel) { - case 16: - for (x = xstart; x != xend; x += xdir) { - for (y = ystart; y != yend; y += ydir) { - READ_DEPTH16(d, xa+x, ya+y); - WRITE_DEPTH16(xb+x, yb+y, d); - } - } - break; - case 32: - for (x = xstart; x != xend; x += xdir) { - for (y = ystart; y != yend; y += ydir) { - READ_DEPTH32(d, xa+x, ya+y); - WRITE_DEPTH32(xb+x, yb+y, d); - } - } - break; - default: break; - } -} /* Initialize the state of the back and depth buffers. */ static void RADEONDRIInitBuffers(WindowPtr pWin, RegionPtr prgn, CARD32 indx) { - /* FIXME: This routine needs to have acceleration turned on */ - ScreenPtr pScreen = pWin->drawable.pScreen; - ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum]; - RADEONInfoPtr info = RADEONPTR(pScrn); - RADEONSAREAPrivPtr pSAREAPriv; - BoxPtr pbox; - int nbox; - unsigned int color, depth, stencil; - unsigned int color_mask, depth_mask, flags; - drmRadeonClearType clear; - drmRadeonClearRect depth_boxes[RADEON_NR_SAREA_CLIPRECTS]; - - /* FIXME: This should be based on the __GLXvisualConfig info */ - color = 0; - switch (pScrn->bitsPerPixel) { - case 16: - depth = 0x0000ffff; - color_mask = 0x0000ffff; - depth_mask = 0xffffffff; - stencil = 0x00000000; - flags = RADEON_BACK | RADEON_DEPTH; - break; - case 32: - depth = 0x00ffffff; - color_mask = 0xffffffff; - depth_mask = 0x00ffffff; - stencil = 0xff000000; - flags = RADEON_BACK | RADEON_DEPTH /*| RADEON_STENCIL*/; - break; - default: - return; - } - - /* FIXME: Copy XAAPaintWindow() and use REGION_TRANSLATE() */ - /* FIXME: Only initialize the back and depth buffers for contexts - that request them */ - - FLUSH_RING(); - - pSAREAPriv = (RADEONSAREAPrivPtr)DRIGetSAREAPrivate(pScreen); - - clear.flags = flags; - clear.clear_color = color; - clear.clear_depth = depth; - clear.color_mask = color_mask; - clear.depth_mask = depth_mask; - clear.depth_boxes = depth_boxes; - - pbox = REGION_RECTS(prgn); - nbox = REGION_NUM_RECTS(prgn); - - for (; nbox; nbox--, pbox++) { - int ret; - - /* DRM_RADEON_CLEAR uses the clip rects to draw instead of the - rect passed to it; however, it uses the rect for the depth - clears */ - pSAREAPriv->boxes[0].x1 = pbox->x1; - pSAREAPriv->boxes[0].y1 = pbox->y1; - pSAREAPriv->boxes[0].x2 = pbox->x2; - pSAREAPriv->boxes[0].y2 = pbox->y2; - pSAREAPriv->nbox = 1; - - depth_boxes[0].f[RADEON_CLEAR_X1] = (float)pbox->x1; - depth_boxes[0].f[RADEON_CLEAR_Y1] = (float)pbox->y1; - depth_boxes[0].f[RADEON_CLEAR_X2] = (float)pbox->x2; - depth_boxes[0].f[RADEON_CLEAR_Y2] = (float)pbox->y2; - depth_boxes[0].f[RADEON_CLEAR_DEPTH] = (float)depth; - - ret = drmCommandWrite(info->drmFD, DRM_RADEON_CLEAR, - &clear, sizeof(drmRadeonClearType)); - if (ret) { - xf86DrvMsg(pScrn->scrnIndex, X_ERROR, - "[dri] DRIInitBuffers timed out, resetting engine...\n"); - RADEONEngineReset(pScrn); - RADEONEngineRestore(pScrn); - RADEONCP_RESET(pScrn, info); - RADEONCP_START(pScrn, info); - return; - } - } - - /* Mark the X server as the last context owner */ - pSAREAPriv->ctxOwner = DRIGetContext(pScreen); - - RADEONSelectBuffer(pScrn, RADEON_FRONT); - info->accel->NeedToSync = TRUE; + /* NOOP. There's no need for the 2d driver to be clearing buffers + * for the 3d client. It knows how to do that on its own. + */ } /* Copy the back and depth buffers when the X server moves a window. @@ -669,13 +480,6 @@ static void RADEONDRIMoveBuffers(WindowPtr pParent, DDXPointRec ptOldOrg, xa, ya, destx, desty, w, h); - RADEONSelectBuffer(pScrn, RADEON_DEPTH); - - if (info->depthMoves) - RADEONScreenToScreenCopyDepth(pScrn, - xa, ya, - destx, desty, - w, h); } RADEONSelectBuffer(pScrn, RADEON_FRONT); @@ -722,6 +526,11 @@ static Bool RADEONDRIAgpInit(RADEONInfoPtr info, ScreenPtr pScreen) case 1: default: mode |= RADEON_AGP_1X_MODE; } + if (info->agpFastWrite) + { + mode |= RADEON_AGP_FW_MODE; + } + xf86DrvMsg(pScreen->myNum, X_INFO, "[agp] Mode 0x%08lx [AGP 0x%04x/0x%04x; Card 0x%04x/0x%04x]\n", mode, vendor, device, @@ -993,7 +802,11 @@ static int RADEONDRIKernelInit(RADEONInfoPtr info, ScreenPtr pScreen) memset(&drmInfo, 0, sizeof(drmRadeonInit)); - drmInfo.func = DRM_RADEON_INIT_CP; + if (info->IsR200) + drmInfo.func = DRM_RADEON_INIT_R200_CP; + else + drmInfo.func = DRM_RADEON_INIT_CP; + drmInfo.sarea_priv_offset = sizeof(XF86DRISAREARec); drmInfo.is_pci = info->IsPCI; drmInfo.cp_mode = info->CPMode; @@ -1086,162 +899,6 @@ static void RADEONDRICPInit(ScrnInfoPtr pScrn) RADEONSelectBuffer(pScrn, RADEON_FRONT); } -/* Initialize the DRI specific hardware state stored in the SAREA. - Currently, this involves setting up the 3D hardware state. */ -static void RADEONDRISAREAInit(ScreenPtr pScreen, - RADEONSAREAPrivPtr pSAREAPriv) -{ - ScrnInfoPtr pScrn = xf86Screens[pScreen->myNum]; - RADEONInfoPtr info = RADEONPTR(pScrn); - radeon_context_regs_t *ctx; - radeon_texture_regs_t *tex; - CARD32 color_fmt, depth_fmt; - int i; - - switch (info->CurrentLayout.pixel_code) { - case 16: - color_fmt = RADEON_COLOR_FORMAT_RGB565; - depth_fmt = RADEON_DEPTH_FORMAT_16BIT_INT_Z; - break; - case 32: - color_fmt = RADEON_COLOR_FORMAT_ARGB8888; - depth_fmt = RADEON_DEPTH_FORMAT_24BIT_INT_Z; - break; - default: - xf86DrvMsg(pScreen->myNum, X_ERROR, - "[dri] RADEONDRISAREAInit failed: Unsupported depth (%d bpp). Disabling DRI.\n", - info->CurrentLayout.pixel_code); - return; - } - - /* Initialize the context state */ - ctx = &pSAREAPriv->ContextState; - - ctx->pp_misc = (RADEON_ALPHA_TEST_PASS | - RADEON_CHROMA_FUNC_FAIL | - RADEON_CHROMA_KEY_NEAREST | - RADEON_SHADOW_FUNC_EQUAL | - RADEON_SHADOW_PASS_1 | - RADEON_RIGHT_HAND_CUBE_OGL); - - ctx->pp_fog_color = ((0x00000000 & RADEON_FOG_COLOR_MASK) | - RADEON_FOG_VERTEX | - RADEON_FOG_USE_DEPTH); - - ctx->re_solid_color = 0x00000000; - - ctx->rb3d_blendcntl = (RADEON_SRC_BLEND_GL_ONE | - RADEON_DST_BLEND_GL_ZERO ); - - ctx->rb3d_depthoffset = info->depthOffset; - - ctx->rb3d_depthpitch = ((info->depthPitch & RADEON_DEPTHPITCH_MASK) | - RADEON_DEPTH_ENDIAN_NO_SWAP); - - ctx->rb3d_zstencilcntl = (depth_fmt | - RADEON_Z_TEST_LESS | - RADEON_STENCIL_TEST_ALWAYS | - RADEON_STENCIL_FAIL_KEEP | - RADEON_STENCIL_ZPASS_KEEP | - RADEON_STENCIL_ZFAIL_KEEP | - RADEON_Z_WRITE_ENABLE); - - ctx->pp_cntl = (RADEON_SCISSOR_ENABLE | - RADEON_ANTI_ALIAS_NONE); - - ctx->rb3d_cntl = (RADEON_PLANE_MASK_ENABLE | - color_fmt | (1<<15)); - - ctx->rb3d_coloroffset = (info->backOffset & RADEON_COLOROFFSET_MASK); - - ctx->re_width_height = ((0x7ff << RADEON_RE_WIDTH_SHIFT) | - (0x7ff << RADEON_RE_HEIGHT_SHIFT)); - - ctx->rb3d_colorpitch = ((info->backPitch & RADEON_COLORPITCH_MASK) | - RADEON_COLOR_ENDIAN_NO_SWAP); - - ctx->se_cntl = (RADEON_FFACE_CULL_CW | - RADEON_BFACE_SOLID | - RADEON_FFACE_SOLID | - RADEON_FLAT_SHADE_VTX_LAST | - RADEON_DIFFUSE_SHADE_GOURAUD | - RADEON_ALPHA_SHADE_GOURAUD | - RADEON_SPECULAR_SHADE_GOURAUD | - RADEON_FOG_SHADE_GOURAUD | - RADEON_VPORT_XY_XFORM_ENABLE | - RADEON_VTX_PIX_CENTER_OGL | - RADEON_ROUND_MODE_TRUNC | - RADEON_ROUND_PREC_8TH_PIX); - - ctx->se_coord_fmt = (RADEON_VTX_XY_PRE_MULT_1_OVER_W0 | - RADEON_VTX_Z_PRE_MULT_1_OVER_W0 | - RADEON_TEX1_W_ROUTING_USE_Q1); - - ctx->re_line_pattern = ((0x0000 & RADEON_LINE_PATTERN_MASK) | - (0 << RADEON_LINE_REPEAT_COUNT_SHIFT) | - (0 << RADEON_LINE_PATTERN_START_SHIFT) | - RADEON_LINE_PATTERN_LITTLE_BIT_ORDER); - - ctx->re_line_state = ((0 << RADEON_LINE_CURRENT_PTR_SHIFT) | - (0 << RADEON_LINE_CURRENT_COUNT_SHIFT)); - - ctx->se_line_width = 0x0000000; - - ctx->pp_lum_matrix = 0x00000000; - - ctx->pp_rot_matrix_0 = 0x00000000; - ctx->pp_rot_matrix_1 = 0x00000000; - - ctx->rb3d_stencilrefmask = - (CARD32)((0x000 << RADEON_STENCIL_REF_SHIFT) | - (0x0ff << RADEON_STENCIL_MASK_SHIFT) | - (0x0ff << RADEON_STENCIL_WRITEMASK_SHIFT)); - - ctx->rb3d_ropcntl = 0x00000000; - ctx->rb3d_planemask = 0xffffffff; - - ctx->se_vport_xscale = 0x00000000; - ctx->se_vport_xoffset = 0x00000000; - ctx->se_vport_yscale = 0x00000000; - ctx->se_vport_yoffset = 0x00000000; - ctx->se_vport_zscale = 0x00000000; - ctx->se_vport_zoffset = 0x00000000; - -#if X_BYTE_ORDER == X_LITTLE_ENDIAN - ctx->se_cntl_status = RADEON_VC_NO_SWAP; -#else - ctx->se_cntl_status = RADEON_VC_32BIT_SWAP; -#endif - - if (info->IsM6) - ctx->se_cntl_status |= RADEON_TCL_BYPASS; - - ctx->re_top_left = ((0 << RADEON_RE_LEFT_SHIFT) | - (0 << RADEON_RE_TOP_SHIFT) ); - - ctx->re_misc = ((0 << RADEON_STIPPLE_X_OFFSET_SHIFT) | - (0 << RADEON_STIPPLE_Y_OFFSET_SHIFT) | - RADEON_STIPPLE_LITTLE_BIT_ORDER); - - /* Initialize the texture state */ - for (i = 0; i < RADEON_MAX_TEXTURE_UNITS; i++) { - tex = &pSAREAPriv->TexState[i]; - - tex->pp_txfilter = 0x00000000; - tex->pp_txformat = 0x00000000; - tex->pp_txoffset = 0x00000000; - tex->pp_txcblend = 0x00000000; - tex->pp_txablend = 0x00000000; - tex->pp_tfactor = 0x00000000; - tex->pp_border_color = 0x00000000; - } - - /* Mark the context as dirty */ - pSAREAPriv->dirty = RADEON_UPLOAD_ALL; - - /* Mark the X server as the last context owner */ - pSAREAPriv->ctxOwner = DRIGetContext(pScreen); -} /* Initialize the screen-specific data structures for the DRI and the Radeon. This is the main entry point to the device-specific @@ -1300,7 +957,12 @@ Bool RADEONDRIScreenInit(ScreenPtr pScreen) info->pDRIInfo = pDRIInfo; pDRIInfo->drmDriverName = RADEON_DRIVER_NAME; - pDRIInfo->clientDriverName = RADEON_DRIVER_NAME; + + if (info->IsR200) + pDRIInfo->clientDriverName = R200_DRIVER_NAME; + else + pDRIInfo->clientDriverName = RADEON_DRIVER_NAME; + pDRIInfo->busIdString = xalloc(64); sprintf(pDRIInfo->busIdString, "PCI:%d:%d:%d", @@ -1399,26 +1061,37 @@ Bool RADEONDRIScreenInit(ScreenPtr pScreen) /* Check the radeon DRM version */ version = drmGetVersion(info->drmFD); if (version) { - if (version->version_major != 1 || + int req_minor, req_patch; + + if (info->IsR200) { + req_minor = 5; + req_patch = 0; + } + else { #if X_BYTE_ORDER == X_LITTLE_ENDIAN - version->version_minor < 1) { + req_minor = 1; + req_patch = 0; #else - version->version_minor < 2 || - (version->version_minor == 2 && version->version_patchlevel == 0)) { + req_minor = 2; + req_patch = 1; #endif + } + + if (version->version_major != 1 || + version->version_minor < req_minor || + (version->version_minor == req_minor && + version->version_patchlevel < req_patch)) { /* incompatible drm version */ xf86DrvMsg(pScreen->myNum, X_ERROR, "[dri] RADEONDRIScreenInit failed because of a version mismatch.\n" "[dri] radeon.o kernel module version is %d.%d.%d " -#if X_BYTE_ORDER == X_LITTLE_ENDIAN - "but version 1.1.x is needed.\n" -#else - "but version 1.2.x (x > 0) is needed.\n" -#endif + "but version 1.%d.%d or newer is needed.\n" "[dri] Disabling DRI.\n", version->version_major, version->version_minor, - version->version_patchlevel); + version->version_patchlevel, + req_minor, + req_patch); drmFreeVersion(version); RADEONDRICloseScreen(pScreen); return FALSE; @@ -1525,8 +1198,6 @@ Bool RADEONDRIFinishScreenInit(ScreenPtr pScreen) pSAREAPriv = (RADEONSAREAPrivPtr)DRIGetSAREAPrivate(pScreen); memset(pSAREAPriv, 0, sizeof(*pSAREAPriv)); - RADEONDRISAREAInit(pScreen, pSAREAPriv); - pRADEONDRI = (RADEONDRIPtr)info->pDRIInfo->devPrivate; pRADEONDRI->deviceID = info->Chipset; @@ -1801,22 +1472,16 @@ RADEONDRITransitionTo2d(ScreenPtr pScreen) RADEONInfoPtr info = RADEONPTR(pScrn); RADEONSAREAPrivPtr pSAREAPriv = DRIGetSAREAPrivate(pScreen); - /* Go back to the front buffer if things were left in a flipped state. - */ - if (pSAREAPriv->pfCurrentPage != 0) { - /* Won't work as we're not holding the lock at this point: - */ -/* drmRadeonFlipBuffers( info->drmFD ); */ - } - /* Shut down shadowing if we've made it back to the front page: */ if (pSAREAPriv->pfCurrentPage == 0) { pSAREAPriv->pfAllowPageFlip = 0; } -/* else */ -/* xf86DrvMsg(pScreen->myNum, X_WARNING, */ -/* "[dri] RADEONDRITransitionTo2d failed to unflip buffers.\n"); */ + else { + xf86DrvMsg(pScreen->myNum, X_WARNING, + "[dri] RADEONDRITransitionTo2d: " + "kernel failed to unflip buffers.\n"); + } info->have3DWindows = 0; } diff --git a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.h b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.h index 35c34693b..ef46cfd37 100644 --- a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.h +++ b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_dri.h @@ -44,6 +44,7 @@ #define RADEON_DEFAULT_CP_PIO_MODE RADEON_CSQ_PRIPIO_INDPIO #define RADEON_DEFAULT_CP_BM_MODE RADEON_CSQ_PRIBM_INDBM #define RADEON_DEFAULT_AGP_MODE 1 +#define RADEON_DEFAULT_AGP_FAST_WRITE 0 #define RADEON_DEFAULT_AGP_SIZE 8 /* MB (must be a power of 2 and > 4MB) */ #define RADEON_DEFAULT_RING_SIZE 1 /* MB (must be page aligned) */ #define RADEON_DEFAULT_BUFFER_SIZE 2 /* MB (must be page aligned) */ diff --git a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_driver.c b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_driver.c index ac3323018..580eaf0e4 100644 --- a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_driver.c +++ b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_driver.c @@ -126,6 +126,7 @@ typedef enum { OPTION_NO_SECURITY, OPTION_USEC_TIMEOUT, OPTION_AGP_MODE, + OPTION_AGP_FW, OPTION_AGP_SIZE, OPTION_RING_SIZE, OPTION_BUFFER_SIZE, @@ -147,6 +148,7 @@ const OptionInfoRec RADEONOptions[] = { { OPTION_CP_PIO, "CPPIOMode", OPTV_BOOLEAN, {0}, FALSE }, { OPTION_USEC_TIMEOUT, "CPusecTimeout", OPTV_INTEGER, {0}, FALSE }, { OPTION_AGP_MODE, "AGPMode", OPTV_INTEGER, {0}, FALSE }, + { OPTION_AGP_FW, "AGPFastWrite", OPTV_BOOLEAN, {0}, FALSE }, { OPTION_AGP_SIZE, "AGPSize", OPTV_INTEGER, {0}, FALSE }, { OPTION_RING_SIZE, "RingSize", OPTV_INTEGER, {0}, FALSE }, { OPTION_BUFFER_SIZE, "BufferSize", OPTV_INTEGER, {0}, FALSE }, @@ -1899,6 +1901,7 @@ static Bool RADEONPreInitDRI(ScrnInfoPtr pScrn) info->ringSize = RADEON_DEFAULT_RING_SIZE; info->bufSize = RADEON_DEFAULT_BUFFER_SIZE; info->agpTexSize = RADEON_DEFAULT_AGP_TEX_SIZE; + info->agpFastWrite = RADEON_DEFAULT_AGP_FAST_WRITE; info->CPusecTimeout = RADEON_DEFAULT_CP_TIMEOUT; @@ -1914,6 +1917,14 @@ static Bool RADEONPreInitDRI(ScrnInfoPtr pScrn) "Using AGP %dx mode\n", info->agpMode); } + if ((info->agpFastWrite = xf86ReturnOptValBool(info->Options, + OPTION_AGP_FW, FALSE))) { + xf86DrvMsg(pScrn->scrnIndex, X_CONFIG, "Enabling AGP Fast Write\n"); + } else { + xf86DrvMsg(pScrn->scrnIndex, X_INFO, + "AGP Fast Write disabled by default\n"); + } + if (xf86GetOptValInteger(info->Options, OPTION_AGP_SIZE, (int *)&(info->agpSize))) { switch (info->agpSize) { @@ -2529,10 +2540,6 @@ Bool RADEONScreenInit(int scrnIndex, ScreenPtr pScreen, int argc, char **argv) (pScrn->displayWidth * pScrn->virtualY * info->CurrentLayout.pixel_bytes * 3 + 1023) / 1024); info->directRenderingEnabled = FALSE; - } else if (info->IsR200) { - info->directRenderingEnabled = FALSE; - xf86DrvMsg(scrnIndex, X_WARNING, - "Direct rendering not yet supported on Radeon 8500\n"); } else { if(info->IsSecondary) info->directRenderingEnabled = FALSE; diff --git a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_reg.h b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_reg.h index 1d56204e9..9f78421b1 100644 --- a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_reg.h +++ b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_reg.h @@ -166,7 +166,8 @@ # define RADEON_AGP_1X_MODE 0x01 # define RADEON_AGP_2X_MODE 0x02 # define RADEON_AGP_4X_MODE 0x04 -# define RADEON_AGP_MODE_MASK 0x07 +# define RADEON_AGP_FW_MODE 0x10 +# define RADEON_AGP_MODE_MASK 0x17 #define RADEON_AMCGPIO_A_REG 0x01a0 #define RADEON_AMCGPIO_EN_REG 0x01a8 #define RADEON_AMCGPIO_MASK 0x0194 diff --git a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_version.h b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_version.h index e06aead72..6969bd9d1 100644 --- a/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_version.h +++ b/xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_version.h @@ -1,6 +1,6 @@ -/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_version.h,v 1.1 2000/11/02 16:55:46 tsi Exp $ */ +/* $XFree86: xc/programs/Xserver/hw/xfree86/drivers/ati/radeon_version.h,v 1.3 2002/01/16 16:22:28 tsi Exp $ */ /* - * Copyright 2000 by Marc Aurele La France (TSI @ UQV), tsi@ualberta.ca + * Copyright 2000 through 2002 by Marc Aurele La France (TSI @ UQV), tsi@xfree86.org * * Permission to use, copy, modify, distribute, and sell this software and its * documentation for any purpose is hereby granted without fee, provided that @@ -26,6 +26,7 @@ #define RADEON_NAME "RADEON" #define RADEON_DRIVER_NAME "radeon" +#define R200_DRIVER_NAME "r200" #define RADEON_VERSION_NAME "4.0.1" diff --git a/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/Makefile.bsd b/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/Makefile.bsd index 9c87d963d..f26fd53d9 100644 --- a/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/Makefile.bsd +++ b/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/Makefile.bsd @@ -1,6 +1,6 @@ # $FreeBSD$ # i810, i830 & sis are not complete -SUBDIR = tdfx mga r128 radeon gamma # i810 sis i830 +SUBDIR = radeon # r128 i810 sis i830 tdfx mga gamma .include <bsd.subdir.mk> diff --git a/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/drm_drv.h b/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/drm_drv.h index fb0454ddf..81ca644a4 100644 --- a/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/drm_drv.h +++ b/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/drm_drv.h @@ -1207,7 +1207,7 @@ int DRM(unlock)( DRM_IOCTL_ARGS ) DRM(dma_schedule)( dev, 1 ); #endif - /* FIXME: Do we ever really need to check this??? + /* FIXME: Do we ever really need to check this? */ if ( 1 /* !dev->context_flag */ ) { if ( DRM(lock_free)( dev, &dev->lock.hw_lock->lock, diff --git a/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/radeon_drv.c b/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/radeon_drv.c index fe69fbb82..d02ab9594 100644 --- a/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/radeon_drv.c +++ b/xc/programs/Xserver/hw/xfree86/os-support/bsd/drm/kernel/radeon_drv.c @@ -52,6 +52,7 @@ drm_chipinfo_t DRM(devicelist)[] = { {0x1002, 0x5157, 1, "ATI Radeon QW 7500 (AGP)"}, {0x1002, 0x5159, 1, "ATI Radeon QY VE (AGP)"}, {0x1002, 0x515A, 1, "ATI Radeon QZ VE (AGP)"}, + {0x1002, 0x514C, 1, "ATI Radeon QL 8500 (AGP)"}, {0, 0, 0, NULL} }; diff --git a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon.h b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon.h index ad36c8685..885e4297f 100644 --- a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon.h +++ b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon.h @@ -25,6 +25,7 @@ * * Authors: * Gareth Hughes <gareth@valinux.com> + * Keith Whitwell <keith@tungstengraphics.com> */ #ifndef __RADEON_H__ @@ -43,14 +44,14 @@ #define __HAVE_SG 1 #define __HAVE_PCI_DMA 1 -#define DRIVER_AUTHOR "Gareth Hughes, VA Linux Systems Inc." +#define DRIVER_AUTHOR "Gareth Hughes, Keith Whitwell, others." #define DRIVER_NAME "radeon" #define DRIVER_DESC "ATI Radeon" -#define DRIVER_DATE "20020714" +#define DRIVER_DATE "20020611" #define DRIVER_MAJOR 1 -#define DRIVER_MINOR 4 +#define DRIVER_MINOR 5 #define DRIVER_PATCHLEVEL 0 /* Interface history: @@ -63,6 +64,10 @@ * - Add support for new radeon packets (keith) * - Add getparam ioctl (keith) * - Add flip-buffers ioctl, deprecate fullscreen foo (keith). + * 1.4 - Add scratch registers to get_param ioctl. + * 1.5 - Add r200 packets to cmdbuf ioctl + * - Add r200 function to init ioctl + * - Add 'scalar2' instruction to cmdbuf */ #define DRIVER_IOCTLS \ [DRM_IOCTL_NR(DRM_IOCTL_DMA)] = { radeon_cp_buffers, 1, 0 }, \ @@ -104,15 +109,6 @@ */ #define __HAVE_DMA 1 -#if 0 -/* GH: Remove this for now... */ -#define __HAVE_DMA_QUIESCENT 1 -#define DRIVER_DMA_QUIESCENT() do { \ - drm_radeon_private_t *dev_priv = dev->dev_private; \ - return radeon_do_cp_idle( dev_priv ); \ -} while (0) -#endif - /* Buffer customization: */ #define DRIVER_BUF_PRIV_T drm_radeon_buf_priv_t diff --git a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_cp.c b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_cp.c index 8250c09b5..01069e498 100644 --- a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_cp.c +++ b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_cp.c @@ -44,6 +44,266 @@ /* CP microcode (from ATI) */ +static u32 R200_cp_microcode[][2] = { + { 0x21007000, 0000000000 }, + { 0x20007000, 0000000000 }, + { 0x000000ab, 0x00000004 }, + { 0x000000af, 0x00000004 }, + { 0x66544a49, 0000000000 }, + { 0x49494174, 0000000000 }, + { 0x54517d83, 0000000000 }, + { 0x498d8b64, 0000000000 }, + { 0x49494949, 0000000000 }, + { 0x49da493c, 0000000000 }, + { 0x49989898, 0000000000 }, + { 0xd34949d5, 0000000000 }, + { 0x9dc90e11, 0000000000 }, + { 0xce9b9b9b, 0000000000 }, + { 0x000f0000, 0x00000016 }, + { 0x352e232c, 0000000000 }, + { 0x00000013, 0x00000004 }, + { 0x000f0000, 0x00000016 }, + { 0x352e272c, 0000000000 }, + { 0x000f0001, 0x00000016 }, + { 0x3239362f, 0000000000 }, + { 0x000077ef, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x00000020, 0x0000001a }, + { 0x00004000, 0x0000001e }, + { 0x00061000, 0x00000002 }, + { 0x00000020, 0x0000001a }, + { 0x00004000, 0x0000001e }, + { 0x00061000, 0x00000002 }, + { 0x00000020, 0x0000001a }, + { 0x00004000, 0x0000001e }, + { 0x00000016, 0x00000004 }, + { 0x0003802a, 0x00000002 }, + { 0x040067e0, 0x00000002 }, + { 0x00000016, 0x00000004 }, + { 0x000077e0, 0x00000002 }, + { 0x00065000, 0x00000002 }, + { 0x000037e1, 0x00000002 }, + { 0x040067e1, 0x00000006 }, + { 0x000077e0, 0x00000002 }, + { 0x000077e1, 0x00000002 }, + { 0x000077e1, 0x00000006 }, + { 0xffffffff, 0000000000 }, + { 0x10000000, 0000000000 }, + { 0x0003802a, 0x00000002 }, + { 0x040067e0, 0x00000006 }, + { 0x00007675, 0x00000002 }, + { 0x00007676, 0x00000002 }, + { 0x00007677, 0x00000002 }, + { 0x00007678, 0x00000006 }, + { 0x0003802b, 0x00000002 }, + { 0x04002676, 0x00000002 }, + { 0x00007677, 0x00000002 }, + { 0x00007678, 0x00000006 }, + { 0x0000002e, 0x00000018 }, + { 0x0000002e, 0x00000018 }, + { 0000000000, 0x00000006 }, + { 0x0000002f, 0x00000018 }, + { 0x0000002f, 0x00000018 }, + { 0000000000, 0x00000006 }, + { 0x01605000, 0x00000002 }, + { 0x00065000, 0x00000002 }, + { 0x00098000, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x64c0603d, 0x00000004 }, + { 0x00080000, 0x00000016 }, + { 0000000000, 0000000000 }, + { 0x0400251d, 0x00000002 }, + { 0x00007580, 0x00000002 }, + { 0x00067581, 0x00000002 }, + { 0x04002580, 0x00000002 }, + { 0x00067581, 0x00000002 }, + { 0x00000046, 0x00000004 }, + { 0x00005000, 0000000000 }, + { 0x00061000, 0x00000002 }, + { 0x0000750e, 0x00000002 }, + { 0x00019000, 0x00000002 }, + { 0x00011055, 0x00000014 }, + { 0x00000055, 0x00000012 }, + { 0x0400250f, 0x00000002 }, + { 0x0000504a, 0x00000004 }, + { 0x00007565, 0x00000002 }, + { 0x00007566, 0x00000002 }, + { 0x00000051, 0x00000004 }, + { 0x01e655b4, 0x00000002 }, + { 0x4401b0dc, 0x00000002 }, + { 0x01c110dc, 0x00000002 }, + { 0x2666705d, 0x00000018 }, + { 0x040c2565, 0x00000002 }, + { 0x0000005d, 0x00000018 }, + { 0x04002564, 0x00000002 }, + { 0x00007566, 0x00000002 }, + { 0x00000054, 0x00000004 }, + { 0x00401060, 0x00000008 }, + { 0x00101000, 0x00000002 }, + { 0x000d80ff, 0x00000002 }, + { 0x00800063, 0x00000008 }, + { 0x000f9000, 0x00000002 }, + { 0x000e00ff, 0x00000002 }, + { 0000000000, 0x00000006 }, + { 0x00000080, 0x00000018 }, + { 0x00000054, 0x00000004 }, + { 0x00007576, 0x00000002 }, + { 0x00065000, 0x00000002 }, + { 0x00009000, 0x00000002 }, + { 0x00041000, 0x00000002 }, + { 0x0c00350e, 0x00000002 }, + { 0x00049000, 0x00000002 }, + { 0x00051000, 0x00000002 }, + { 0x01e785f8, 0x00000002 }, + { 0x00200000, 0x00000002 }, + { 0x00600073, 0x0000000c }, + { 0x00007563, 0x00000002 }, + { 0x006075f0, 0x00000021 }, + { 0x20007068, 0x00000004 }, + { 0x00005068, 0x00000004 }, + { 0x00007576, 0x00000002 }, + { 0x00007577, 0x00000002 }, + { 0x0000750e, 0x00000002 }, + { 0x0000750f, 0x00000002 }, + { 0x00a05000, 0x00000002 }, + { 0x00600076, 0x0000000c }, + { 0x006075f0, 0x00000021 }, + { 0x000075f8, 0x00000002 }, + { 0x00000076, 0x00000004 }, + { 0x000a750e, 0x00000002 }, + { 0x0020750f, 0x00000002 }, + { 0x00600079, 0x00000004 }, + { 0x00007570, 0x00000002 }, + { 0x00007571, 0x00000002 }, + { 0x00007572, 0x00000006 }, + { 0x00005000, 0x00000002 }, + { 0x00a05000, 0x00000002 }, + { 0x00007568, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x00000084, 0x0000000c }, + { 0x00058000, 0x00000002 }, + { 0x0c607562, 0x00000002 }, + { 0x00000086, 0x00000004 }, + { 0x00600085, 0x00000004 }, + { 0x400070dd, 0000000000 }, + { 0x000380dd, 0x00000002 }, + { 0x00000093, 0x0000001c }, + { 0x00065095, 0x00000018 }, + { 0x040025bb, 0x00000002 }, + { 0x00061096, 0x00000018 }, + { 0x040075bc, 0000000000 }, + { 0x000075bb, 0x00000002 }, + { 0x000075bc, 0000000000 }, + { 0x00090000, 0x00000006 }, + { 0x00090000, 0x00000002 }, + { 0x000d8002, 0x00000006 }, + { 0x00005000, 0x00000002 }, + { 0x00007821, 0x00000002 }, + { 0x00007800, 0000000000 }, + { 0x00007821, 0x00000002 }, + { 0x00007800, 0000000000 }, + { 0x01665000, 0x00000002 }, + { 0x000a0000, 0x00000002 }, + { 0x000671cc, 0x00000002 }, + { 0x0286f1cd, 0x00000002 }, + { 0x000000a3, 0x00000010 }, + { 0x21007000, 0000000000 }, + { 0x000000aa, 0x0000001c }, + { 0x00065000, 0x00000002 }, + { 0x000a0000, 0x00000002 }, + { 0x00061000, 0x00000002 }, + { 0x000b0000, 0x00000002 }, + { 0x38067000, 0x00000002 }, + { 0x000a00a6, 0x00000004 }, + { 0x20007000, 0000000000 }, + { 0x01200000, 0x00000002 }, + { 0x20077000, 0x00000002 }, + { 0x01200000, 0x00000002 }, + { 0x20007000, 0000000000 }, + { 0x00061000, 0x00000002 }, + { 0x0120751b, 0x00000002 }, + { 0x8040750a, 0x00000002 }, + { 0x8040750b, 0x00000002 }, + { 0x00110000, 0x00000002 }, + { 0x000380dd, 0x00000002 }, + { 0x000000bd, 0x0000001c }, + { 0x00061096, 0x00000018 }, + { 0x844075bd, 0x00000002 }, + { 0x00061095, 0x00000018 }, + { 0x840075bb, 0x00000002 }, + { 0x00061096, 0x00000018 }, + { 0x844075bc, 0x00000002 }, + { 0x000000c0, 0x00000004 }, + { 0x804075bd, 0x00000002 }, + { 0x800075bb, 0x00000002 }, + { 0x804075bc, 0x00000002 }, + { 0x00108000, 0x00000002 }, + { 0x01400000, 0x00000002 }, + { 0x006000c4, 0x0000000c }, + { 0x20c07000, 0x00000020 }, + { 0x000000c6, 0x00000012 }, + { 0x00800000, 0x00000006 }, + { 0x0080751d, 0x00000006 }, + { 0x000025bb, 0x00000002 }, + { 0x000040c0, 0x00000004 }, + { 0x0000775c, 0x00000002 }, + { 0x00a05000, 0x00000002 }, + { 0x00661000, 0x00000002 }, + { 0x0460275d, 0x00000020 }, + { 0x00004000, 0000000000 }, + { 0x00007999, 0x00000002 }, + { 0x00a05000, 0x00000002 }, + { 0x00661000, 0x00000002 }, + { 0x0460299b, 0x00000020 }, + { 0x00004000, 0000000000 }, + { 0x01e00830, 0x00000002 }, + { 0x21007000, 0000000000 }, + { 0x00005000, 0x00000002 }, + { 0x00038042, 0x00000002 }, + { 0x040025e0, 0x00000002 }, + { 0x000075e1, 0000000000 }, + { 0x00000001, 0000000000 }, + { 0x000380d9, 0x00000002 }, + { 0x04007394, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, + { 0000000000, 0000000000 }, +}; + + static u32 radeon_cp_microcode[][2] = { { 0x21007000, 0000000000 }, { 0x20007000, 0000000000 }, @@ -345,6 +605,8 @@ static int radeon_do_pixcache_flush( drm_radeon_private_t *dev_priv ) u32 tmp; int i; + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + tmp = RADEON_READ( RADEON_RB2D_DSTCACHE_CTLSTAT ); tmp |= RADEON_RB2D_DC_FLUSH_ALL; RADEON_WRITE( RADEON_RB2D_DSTCACHE_CTLSTAT, tmp ); @@ -369,6 +631,8 @@ static int radeon_do_wait_for_fifo( drm_radeon_private_t *dev_priv, { int i; + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) { int slots = ( RADEON_READ( RADEON_RBBM_STATUS ) & RADEON_RBBM_FIFOCNT_MASK ); @@ -387,6 +651,8 @@ static int radeon_do_wait_for_idle( drm_radeon_private_t *dev_priv ) { int i, ret; + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + ret = radeon_do_wait_for_fifo( dev_priv, 64 ); if ( ret ) return ret; @@ -420,11 +686,26 @@ static void radeon_cp_load_microcode( drm_radeon_private_t *dev_priv ) radeon_do_wait_for_idle( dev_priv ); RADEON_WRITE( RADEON_CP_ME_RAM_ADDR, 0 ); - for ( i = 0 ; i < 256 ; i++ ) { - RADEON_WRITE( RADEON_CP_ME_RAM_DATAH, - radeon_cp_microcode[i][1] ); - RADEON_WRITE( RADEON_CP_ME_RAM_DATAL, - radeon_cp_microcode[i][0] ); + + if (dev_priv->is_r200) + { + DRM_INFO("Loading R200 Microcode\n"); + for ( i = 0 ; i < 256 ; i++ ) + { + RADEON_WRITE( RADEON_CP_ME_RAM_DATAH, + R200_cp_microcode[i][1] ); + RADEON_WRITE( RADEON_CP_ME_RAM_DATAL, + R200_cp_microcode[i][0] ); + } + } + else + { + for ( i = 0 ; i < 256 ; i++ ) { + RADEON_WRITE( RADEON_CP_ME_RAM_DATAH, + radeon_cp_microcode[i][1] ); + RADEON_WRITE( RADEON_CP_ME_RAM_DATAL, + radeon_cp_microcode[i][0] ); + } } } @@ -736,12 +1017,10 @@ static int radeon_do_init_cp( drm_device_t *dev, drm_radeon_init_t *init ) return DRM_ERR(EINVAL); } + dev_priv->is_r200 = (init->func == RADEON_INIT_R200_CP); + dev_priv->do_boxes = 1; dev_priv->cp_mode = init->cp_mode; - /* Simple idle check. - */ - atomic_set( &dev_priv->idle_count, 0 ); - /* We don't support anything other than bus-mastering ring mode, * but the ring can be in either AGP or PCI space for the ring * read pointer. @@ -1028,6 +1307,7 @@ int radeon_cp_init( DRM_IOCTL_ARGS ) switch ( init.func ) { case RADEON_INIT_CP: + case RADEON_INIT_R200_CP: return radeon_do_init_cp( dev, &init ); case RADEON_CLEANUP_CP: return radeon_do_cleanup_cp( dev ); @@ -1169,6 +1449,14 @@ int radeon_fullscreen( DRM_IOCTL_ARGS ) * completed rendering. * * KW: It's also a good way to find free buffers quickly. + * + * KW: Ideally this loop wouldn't exist, and freelist_get wouldn't + * sleep. However, bugs in older versions of radeon_accel.c mean that + * we essentially have to do this, else old clients will break. + * + * However, it does leave open a potential deadlock where all the + * buffers are held by other clients, which can't release them because + * they can't get the lock. */ drm_buf_t *radeon_freelist_get( drm_device_t *dev ) @@ -1193,17 +1481,56 @@ drm_buf_t *radeon_freelist_get( drm_device_t *dev ) buf_priv = buf->dev_private; if ( buf->pid == 0 || (buf->pending && buf_priv->age <= done_age) ) { + dev_priv->stats.requested_bufs++; buf->pending = 0; return buf; } start = 0; } - DRM_UDELAY( 1 ); + + if (t) { + DRM_UDELAY( 1 ); + dev_priv->stats.freelist_loops++; + } } DRM_ERROR( "returning NULL!\n" ); return NULL; } +#if 0 +drm_buf_t *radeon_freelist_get( drm_device_t *dev ) +{ + drm_device_dma_t *dma = dev->dma; + drm_radeon_private_t *dev_priv = dev->dev_private; + drm_radeon_buf_priv_t *buf_priv; + drm_buf_t *buf; + int i, t; + int start; + u32 done_age = DRM_READ32(&dev_priv->scratch[1]); + + if ( ++dev_priv->last_buf >= dma->buf_count ) + dev_priv->last_buf = 0; + + start = dev_priv->last_buf; + dev_priv->stats.freelist_loops++; + + for ( t = 0 ; t < 2 ; t++ ) { + for ( i = start ; i < dma->buf_count ; i++ ) { + buf = dma->buflist[i]; + buf_priv = buf->dev_private; + if ( buf->pid == 0 || (buf->pending && + buf_priv->age <= done_age) ) { + dev_priv->stats.requested_bufs++; + buf->pending = 0; + return buf; + } + } + start = 0; + } + + return NULL; +} +#endif void radeon_freelist_reset( drm_device_t *dev ) { @@ -1228,11 +1555,23 @@ int radeon_wait_ring( drm_radeon_private_t *dev_priv, int n ) { drm_radeon_ring_buffer_t *ring = &dev_priv->ring; int i; + u32 last_head = GET_RING_HEAD(ring); for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) { - radeon_update_ring_snapshot( ring ); + u32 head = GET_RING_HEAD(ring); + + ring->space = (head - ring->tail) * sizeof(u32); + if ( ring->space <= 0 ) + ring->space += ring->size; if ( ring->space > n ) return 0; + + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + + if (head != last_head) + i = 0; + last_head = head; + DRM_UDELAY( 1 ); } @@ -1251,7 +1590,7 @@ static int radeon_cp_get_buffers( drm_device_t *dev, drm_dma_t *d ) for ( i = d->granted_count ; i < d->request_count ; i++ ) { buf = radeon_freelist_get( dev ); - if ( !buf ) return DRM_ERR(EAGAIN); + if ( !buf ) return DRM_ERR(EBUSY); /* NOTE: broken client */ buf->pid = DRM_CURRENTPID; diff --git a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_drm.h b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_drm.h index 3802e46c5..6469bfb80 100644 --- a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_drm.h +++ b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_drm.h @@ -89,7 +89,47 @@ #define RADEON_EMIT_SE_ZBIAS_FACTOR 18 /* zbias/2 */ #define RADEON_EMIT_SE_TCL_OUTPUT_VTX_FMT 19 /* tcl/11 */ #define RADEON_EMIT_SE_TCL_MATERIAL_EMMISSIVE_RED 20 /* material/17 */ -#define RADEON_MAX_STATE_PACKETS 21 +#define R200_EMIT_PP_TXCBLEND_0 21 /* tex0/4 */ +#define R200_EMIT_PP_TXCBLEND_1 22 /* tex1/4 */ +#define R200_EMIT_PP_TXCBLEND_2 23 /* tex2/4 */ +#define R200_EMIT_PP_TXCBLEND_3 24 /* tex3/4 */ +#define R200_EMIT_PP_TXCBLEND_4 25 /* tex4/4 */ +#define R200_EMIT_PP_TXCBLEND_5 26 /* tex5/4 */ +#define R200_EMIT_PP_TXCBLEND_6 27 /* /4 */ +#define R200_EMIT_PP_TXCBLEND_7 28 /* /4 */ +#define R200_EMIT_TCL_LIGHT_MODEL_CTL_0 29 /* tcl/7 */ +#define R200_EMIT_TFACTOR_0 30 /* tf/7 */ +#define R200_EMIT_VTX_FMT_0 31 /* vtx/5 */ +#define R200_EMIT_VAP_CTL 32 /* vap/1 */ +#define R200_EMIT_MATRIX_SELECT_0 33 /* msl/5 */ +#define R200_EMIT_TEX_PROC_CTL_2 34 /* tcg/5 */ +#define R200_EMIT_TCL_UCP_VERT_BLEND_CTL 35 /* tcl/1 */ +#define R200_EMIT_PP_TXFILTER_0 36 /* tex0/6 */ +#define R200_EMIT_PP_TXFILTER_1 37 /* tex1/6 */ +#define R200_EMIT_PP_TXFILTER_2 38 /* tex2/6 */ +#define R200_EMIT_PP_TXFILTER_3 39 /* tex3/6 */ +#define R200_EMIT_PP_TXFILTER_4 40 /* tex4/6 */ +#define R200_EMIT_PP_TXFILTER_5 41 /* tex5/6 */ +#define R200_EMIT_PP_TXOFFSET_0 42 /* tex0/1 */ +#define R200_EMIT_PP_TXOFFSET_1 43 /* tex1/1 */ +#define R200_EMIT_PP_TXOFFSET_2 44 /* tex2/1 */ +#define R200_EMIT_PP_TXOFFSET_3 45 /* tex3/1 */ +#define R200_EMIT_PP_TXOFFSET_4 46 /* tex4/1 */ +#define R200_EMIT_PP_TXOFFSET_5 47 /* tex5/1 */ +#define R200_EMIT_VTE_CNTL 48 /* vte/1 */ +#define R200_EMIT_OUTPUT_VTX_COMP_SEL 49 /* vtx/1 */ +#define R200_EMIT_PP_TAM_DEBUG3 50 /* tam/1 */ +#define R200_EMIT_PP_CNTL_X 51 /* cst/1 */ +#define R200_EMIT_RB3D_DEPTHXY_OFFSET 52 /* cst/1 */ +#define R200_EMIT_RE_AUX_SCISSOR_CNTL 53 /* cst/1 */ +#define R200_EMIT_RE_SCISSOR_TL_0 54 /* cst/2 */ +#define R200_EMIT_RE_SCISSOR_TL_1 55 /* cst/2 */ +#define R200_EMIT_RE_SCISSOR_TL_2 56 /* cst/2 */ +#define R200_EMIT_SE_VAP_CNTL_STATUS 57 /* cst/1 */ +#define R200_EMIT_SE_VTX_STATE_CNTL 58 /* cst/1 */ +#define R200_EMIT_RE_POINTSIZE 59 /* cst/1 */ +#define R200_EMIT_TCL_INPUT_VTX_VECTOR_ADDR_0 60 /* cst/4 */ +#define RADEON_MAX_STATE_PACKETS 61 /* Commands understood by cmd_buffer ioctl. More can be added but @@ -101,24 +141,25 @@ #define RADEON_CMD_DMA_DISCARD 4 /* discard current dma buf */ #define RADEON_CMD_PACKET3 5 /* emit hw packet */ #define RADEON_CMD_PACKET3_CLIP 6 /* emit hw packet wrapped in cliprects */ +#define RADEON_CMD_SCALARS2 7 /* r200 stopgap */ typedef union { int i; struct { - char cmd_type, pad0, pad1, pad2; + unsigned char cmd_type, pad0, pad1, pad2; } header; struct { - char cmd_type, packet_id, pad0, pad1; + unsigned char cmd_type, packet_id, pad0, pad1; } packet; struct { - char cmd_type, offset, stride, count; + unsigned char cmd_type, offset, stride, count; } scalars; struct { - char cmd_type, offset, stride, count; + unsigned char cmd_type, offset, stride, count; } vectors; struct { - char cmd_type, buf_idx, pad0, pad1; + unsigned char cmd_type, buf_idx, pad0, pad1; } dma; } drm_radeon_cmd_header_t; @@ -327,7 +368,8 @@ typedef struct { typedef struct drm_radeon_init { enum { RADEON_INIT_CP = 0x01, - RADEON_CLEANUP_CP = 0x02 + RADEON_CLEANUP_CP = 0x02, + RADEON_INIT_R200_CP = 0x03, } func; unsigned long sarea_priv_offset; int is_pci; diff --git a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_drv.h b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_drv.h index 15c0d4dda..7c341b39e 100644 --- a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_drv.h +++ b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_drv.h @@ -79,12 +79,25 @@ typedef struct drm_radeon_private { int writeback_works; int usec_timeout; + + int is_r200; + int is_pci; unsigned long phys_pci_gart; dma_addr_t bus_pci_gart; - atomic_t idle_count; - + struct { + u32 boxes; + int freelist_timeouts; + int freelist_loops; + int requested_bufs; + int last_frame_reads; + int last_clear_reads; + int clears; + int texture_uploads; + } stats; + + int do_boxes; int page_flipping; int current_page; u32 crtc_offset; @@ -134,14 +147,6 @@ extern drm_buf_t *radeon_freelist_get( drm_device_t *dev ); extern int radeon_wait_ring( drm_radeon_private_t *dev_priv, int n ); -static __inline__ void -radeon_update_ring_snapshot( drm_radeon_ring_buffer_t *ring ) -{ - ring->space = (GET_RING_HEAD(ring) - ring->tail) * sizeof(u32); - if ( ring->space <= 0 ) - ring->space += ring->size; -} - extern int radeon_do_cp_idle( drm_radeon_private_t *dev_priv ); extern int radeon_do_cleanup_cp( drm_device_t *dev ); extern int radeon_do_cleanup_pageflip( drm_device_t *dev ); @@ -159,6 +164,14 @@ extern int radeon_cp_cmdbuf( DRM_IOCTL_ARGS ); extern int radeon_cp_getparam( DRM_IOCTL_ARGS ); extern int radeon_cp_flip( DRM_IOCTL_ARGS ); +/* Flags for stats.boxes + */ +#define RADEON_BOX_DMA_IDLE 0x1 +#define RADEON_BOX_RING_FULL 0x2 +#define RADEON_BOX_FLIP 0x4 +#define RADEON_BOX_WAIT_IDLE 0x8 +#define RADEON_BOX_TEXTURE_LOAD 0x10 + /* Register definitions, register access macros and drmAddMap constants @@ -282,6 +295,7 @@ extern int radeon_cp_flip( DRM_IOCTL_ARGS ); # define RADEON_STENCIL_ENABLE (1 << 7) # define RADEON_Z_ENABLE (1 << 8) #define RADEON_RB3D_DEPTHOFFSET 0x1c24 +#define RADEON_RB3D_DEPTHPITCH 0x1c28 #define RADEON_RB3D_PLANEMASK 0x1d84 #define RADEON_RB3D_STENCILREFMASK 0x1d7c #define RADEON_RB3D_ZCACHE_MODE 0x3250 @@ -513,6 +527,62 @@ extern int radeon_cp_flip( DRM_IOCTL_ARGS ); #define RADEON_TXFORMAT_ARGB8888 6 #define RADEON_TXFORMAT_RGBA8888 7 +#define R200_PP_TXCBLEND_0 0x2f00 +#define R200_PP_TXCBLEND_1 0x2f10 +#define R200_PP_TXCBLEND_2 0x2f20 +#define R200_PP_TXCBLEND_3 0x2f30 +#define R200_PP_TXCBLEND_4 0x2f40 +#define R200_PP_TXCBLEND_5 0x2f50 +#define R200_PP_TXCBLEND_6 0x2f60 +#define R200_PP_TXCBLEND_7 0x2f70 +#define R200_SE_TCL_LIGHT_MODEL_CTL_0 0x2268 +#define R200_PP_TFACTOR_0 0x2ee0 +#define R200_SE_VTX_FMT_0 0x2088 +#define R200_SE_VAP_CNTL 0x2080 +#define R200_SE_TCL_MATRIX_SEL_0 0x2230 +#define R200_SE_TCL_TEX_PROC_CTL_2 0x22a8 +#define R200_SE_TCL_UCP_VERT_BLEND_CTL 0x22c0 +#define R200_PP_TXFILTER_5 0x2ca0 +#define R200_PP_TXFILTER_4 0x2c80 +#define R200_PP_TXFILTER_3 0x2c60 +#define R200_PP_TXFILTER_2 0x2c40 +#define R200_PP_TXFILTER_1 0x2c20 +#define R200_PP_TXFILTER_0 0x2c00 +#define R200_PP_TXOFFSET_5 0x2d78 +#define R200_PP_TXOFFSET_4 0x2d60 +#define R200_PP_TXOFFSET_3 0x2d48 +#define R200_PP_TXOFFSET_2 0x2d30 +#define R200_PP_TXOFFSET_1 0x2d18 +#define R200_PP_TXOFFSET_0 0x2d00 +#define R200_RE_AUX_SCISSOR_CNTL 0x26f0 +#define R200_SE_VTE_CNTL 0x20b0 +#define R200_SE_TCL_OUTPUT_VTX_COMP_SEL 0x2250 +#define R200_PP_TAM_DEBUG3 0x2d9c +#define R200_PP_CNTL_X 0x2cc4 +#define R200_SE_VAP_CNTL_STATUS 0x2140 +#define R200_RE_SCISSOR_TL_0 0x1cd8 +#define R200_RE_SCISSOR_TL_1 0x1ce0 +#define R200_RE_SCISSOR_TL_2 0x1ce8 +#define R200_RB3D_DEPTHXY_OFFSET 0x1d60 +#define R200_RE_AUX_SCISSOR_CNTL 0x26f0 +#define R200_SE_VTX_STATE_CNTL 0x2180 +#define R200_RE_POINTSIZE 0x2648 +#define R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0 0x2254 + + +#define SE_VAP_CNTL__TCL_ENA_MASK 0x00000001 +#define SE_VAP_CNTL__FORCE_W_TO_ONE_MASK 0x00010000 +#define SE_VAP_CNTL__VF_MAX_VTX_NUM__SHIFT 0x00000012 +#define SE_VTE_CNTL__VTX_XY_FMT_MASK 0x00000100 +#define SE_VTE_CNTL__VTX_Z_FMT_MASK 0x00000200 +#define SE_VTX_FMT_0__VTX_Z0_PRESENT_MASK 0x00000001 +#define SE_VTX_FMT_0__VTX_W0_PRESENT_MASK 0x00000002 +#define SE_VTX_FMT_0__VTX_COLOR_0_FMT__SHIFT 0x0000000b +#define R200_3D_DRAW_IMMD_2 0xC0003500 +#define R200_SE_VTX_FMT_1 0x208c +#define R200_RE_CNTL 0x1c50 + + /* Constants */ #define RADEON_MAX_USEC_TIMEOUT 100000 /* 100 ms */ @@ -620,30 +690,16 @@ do { \ } \ } while (0) + +/* Perfbox functionality only. + */ #define RING_SPACE_TEST_WITH_RETURN( dev_priv ) \ do { \ - drm_radeon_ring_buffer_t *ring = &dev_priv->ring; int i; \ - if ( ring->space < ring->high_mark ) { \ - for ( i = 0 ; i < dev_priv->usec_timeout ; i++ ) { \ - radeon_update_ring_snapshot( ring ); \ - if ( ring->space >= ring->high_mark ) \ - goto __ring_space_done; \ - DRM_UDELAY( 1 ); \ - } \ - DRM_ERROR( "ring space check from memory failed, reading register...\n" ); \ - /* If ring space check fails from RAM, try reading the \ - register directly */ \ - ring->space = 4 * ( RADEON_READ( RADEON_CP_RB_RPTR ) - ring->tail ); \ - if ( ring->space <= 0 ) \ - ring->space += ring->size; \ - if ( ring->space >= ring->high_mark ) \ - goto __ring_space_done; \ - \ - DRM_ERROR( "ring space check failed!\n" ); \ - return DRM_ERR(EBUSY); \ + if (!(dev_priv->stats.boxes & RADEON_BOX_DMA_IDLE)) { \ + u32 head = GET_RING_HEAD(&dev_priv->ring); \ + if (head == dev_priv->ring.tail) \ + dev_priv->stats.boxes |= RADEON_BOX_DMA_IDLE; \ } \ - __ring_space_done: \ - ; \ } while (0) #define VB_AGE_TEST_WITH_RETURN( dev_priv ) \ @@ -710,16 +766,15 @@ do { \ } \ if (((dev_priv->ring.tail + _nr) & mask) != write) { \ DRM_ERROR( \ - "ADVANCE_RING(): mismatch: nr: %x write: %x\n", \ + "ADVANCE_RING(): mismatch: nr: %x write: %x line: %d\n", \ ((dev_priv->ring.tail + _nr) & mask), \ - write); \ + write, __LINE__); \ } else \ dev_priv->ring.tail = write; \ } while (0) #define COMMIT_RING() do { \ - radeon_flush_write_combine(); \ - RADEON_WRITE( RADEON_CP_RB_WPTR, dev_priv->ring.tail ); \ + RADEON_WRITE( RADEON_CP_RB_WPTR, dev_priv->ring.tail ); \ } while (0) #define OUT_RING( x ) do { \ @@ -760,6 +815,4 @@ do { \ } while (0) -#define RADEON_PERFORMANCE_BOXES 0 - #endif /* __RADEON_DRV_H__ */ diff --git a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_state.c b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_state.c index 1cc6bde80..7f84e739a 100644 --- a/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_state.c +++ b/xc/programs/Xserver/hw/xfree86/os-support/shared/drm/kernel/radeon_state.c @@ -239,18 +239,50 @@ static struct { { RADEON_SE_ZBIAS_FACTOR,2,"RADEON_SE_ZBIAS_FACTOR" }, { RADEON_SE_TCL_OUTPUT_VTX_FMT,11,"RADEON_SE_TCL_OUTPUT_VTX_FMT" }, { RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED,17,"RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED" }, + { R200_PP_TXCBLEND_0, 4, "R200_PP_TXCBLEND_0" }, + { R200_PP_TXCBLEND_1, 4, "R200_PP_TXCBLEND_1" }, + { R200_PP_TXCBLEND_2, 4, "R200_PP_TXCBLEND_2" }, + { R200_PP_TXCBLEND_3, 4, "R200_PP_TXCBLEND_3" }, + { R200_PP_TXCBLEND_4, 4, "R200_PP_TXCBLEND_4" }, + { R200_PP_TXCBLEND_5, 4, "R200_PP_TXCBLEND_5" }, + { R200_PP_TXCBLEND_6, 4, "R200_PP_TXCBLEND_6" }, + { R200_PP_TXCBLEND_7, 4, "R200_PP_TXCBLEND_7" }, + { R200_SE_TCL_LIGHT_MODEL_CTL_0, 6, "R200_SE_TCL_LIGHT_MODEL_CTL_0" }, + { R200_PP_TFACTOR_0, 6, "R200_PP_TFACTOR_0" }, + { R200_SE_VTX_FMT_0, 4, "R200_SE_VTX_FMT_0" }, + { R200_SE_VAP_CNTL, 1, "R200_SE_VAP_CNTL" }, + { R200_SE_TCL_MATRIX_SEL_0, 5, "R200_SE_TCL_MATRIX_SEL_0" }, + { R200_SE_TCL_TEX_PROC_CTL_2, 5, "R200_SE_TCL_TEX_PROC_CTL_2" }, + { R200_SE_TCL_UCP_VERT_BLEND_CTL, 1, "R200_SE_TCL_UCP_VERT_BLEND_CTL" }, + { R200_PP_TXFILTER_0, 6, "R200_PP_TXFILTER_0" }, + { R200_PP_TXFILTER_1, 6, "R200_PP_TXFILTER_1" }, + { R200_PP_TXFILTER_2, 6, "R200_PP_TXFILTER_2" }, + { R200_PP_TXFILTER_3, 6, "R200_PP_TXFILTER_3" }, + { R200_PP_TXFILTER_4, 6, "R200_PP_TXFILTER_4" }, + { R200_PP_TXFILTER_5, 6, "R200_PP_TXFILTER_5" }, + { R200_PP_TXOFFSET_0, 1, "R200_PP_TXOFFSET_0" }, + { R200_PP_TXOFFSET_1, 1, "R200_PP_TXOFFSET_1" }, + { R200_PP_TXOFFSET_2, 1, "R200_PP_TXOFFSET_2" }, + { R200_PP_TXOFFSET_3, 1, "R200_PP_TXOFFSET_3" }, + { R200_PP_TXOFFSET_4, 1, "R200_PP_TXOFFSET_4" }, + { R200_PP_TXOFFSET_5, 1, "R200_PP_TXOFFSET_5" }, + { R200_SE_VTE_CNTL, 1, "R200_SE_VTE_CNTL" }, + { R200_SE_TCL_OUTPUT_VTX_COMP_SEL, 1, "R200_SE_TCL_OUTPUT_VTX_COMP_SEL" }, + { R200_PP_TAM_DEBUG3, 1, "R200_PP_TAM_DEBUG3" }, + { R200_PP_CNTL_X, 1, "R200_PP_CNTL_X" }, + { R200_RB3D_DEPTHXY_OFFSET, 1, "R200_RB3D_DEPTHXY_OFFSET" }, + { R200_RE_AUX_SCISSOR_CNTL, 1, "R200_RE_AUX_SCISSOR_CNTL" }, + { R200_RE_SCISSOR_TL_0, 2, "R200_RE_SCISSOR_TL_0" }, + { R200_RE_SCISSOR_TL_1, 2, "R200_RE_SCISSOR_TL_1" }, + { R200_RE_SCISSOR_TL_2, 2, "R200_RE_SCISSOR_TL_2" }, + { R200_SE_VAP_CNTL_STATUS, 1, "R200_SE_VAP_CNTL_STATUS" }, + { R200_SE_VTX_STATE_CNTL, 1, "R200_SE_VTX_STATE_CNTL" }, + { R200_RE_POINTSIZE, 1, "R200_RE_POINTSIZE" }, + { R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0, 4, "R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0" }, }; - - - - - - - -#if RADEON_PERFORMANCE_BOXES /* ================================================================ * Performance monitoring functions */ @@ -259,10 +291,12 @@ static void radeon_clear_box( drm_radeon_private_t *dev_priv, int x, int y, int w, int h, int r, int g, int b ) { - u32 pitch, offset; u32 color; RING_LOCALS; + x += dev_priv->sarea_priv->boxes[0].x1; + y += dev_priv->sarea_priv->boxes[0].y1; + switch ( dev_priv->color_fmt ) { case RADEON_COLOR_FORMAT_RGB565: color = (((r & 0xf8) << 8) | @@ -275,8 +309,11 @@ static void radeon_clear_box( drm_radeon_private_t *dev_priv, break; } - offset = dev_priv->back_offset; - pitch = dev_priv->back_pitch >> 3; + BEGIN_RING( 4 ); + RADEON_WAIT_UNTIL_3D_IDLE(); + OUT_RING( CP_PACKET0( RADEON_DP_WRITE_MASK, 0 ) ); + OUT_RING( 0xffffffff ); + ADVANCE_RING(); BEGIN_RING( 6 ); @@ -288,7 +325,12 @@ static void radeon_clear_box( drm_radeon_private_t *dev_priv, RADEON_ROP3_P | RADEON_GMC_CLR_CMP_CNTL_DIS ); - OUT_RING( (pitch << 22) | (offset >> 5) ); + if ( dev_priv->page_flipping && dev_priv->current_page == 1 ) { + OUT_RING( dev_priv->front_pitch_offset ); + } else { + OUT_RING( dev_priv->back_pitch_offset ); + } + OUT_RING( color ); OUT_RING( (x << 16) | y ); @@ -299,16 +341,57 @@ static void radeon_clear_box( drm_radeon_private_t *dev_priv, static void radeon_cp_performance_boxes( drm_radeon_private_t *dev_priv ) { - if ( atomic_read( &dev_priv->idle_count ) == 0 ) { - radeon_clear_box( dev_priv, 64, 4, 8, 8, 0, 255, 0 ); - } else { - atomic_set( &dev_priv->idle_count, 0 ); + /* Collapse various things into a wait flag -- trying to + * guess if userspase slept -- better just to have them tell us. + */ + if (dev_priv->stats.last_frame_reads > 1 || + dev_priv->stats.last_clear_reads > dev_priv->stats.clears) { + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; } -} -#endif + if (dev_priv->stats.freelist_loops) { + dev_priv->stats.boxes |= RADEON_BOX_WAIT_IDLE; + } + /* Purple box for page flipping + */ + if ( dev_priv->stats.boxes & RADEON_BOX_FLIP ) + radeon_clear_box( dev_priv, 4, 4, 8, 8, 255, 0, 255 ); + + /* Red box if we have to wait for idle at any point + */ + if ( dev_priv->stats.boxes & RADEON_BOX_WAIT_IDLE ) + radeon_clear_box( dev_priv, 16, 4, 8, 8, 255, 0, 0 ); + + /* Blue box: lost context? + */ + /* Yellow box for texture swaps + */ + if ( dev_priv->stats.boxes & RADEON_BOX_TEXTURE_LOAD ) + radeon_clear_box( dev_priv, 40, 4, 8, 8, 255, 255, 0 ); + + /* Green box if hardware never idles (as far as we can tell) + */ + if ( !(dev_priv->stats.boxes & RADEON_BOX_DMA_IDLE) ) + radeon_clear_box( dev_priv, 64, 4, 8, 8, 0, 255, 0 ); + + + /* Draw bars indicating number of buffers allocated + * (not a great measure, easily confused) + */ + if (dev_priv->stats.requested_bufs) { + if (dev_priv->stats.requested_bufs > 100) + dev_priv->stats.requested_bufs = 100; + + radeon_clear_box( dev_priv, 4, 16, + dev_priv->stats.requested_bufs, 4, + 196, 128, 128 ); + } + + memset( &dev_priv->stats, 0, sizeof(dev_priv->stats) ); + +} /* ================================================================ * CP command dispatch functions */ @@ -328,6 +411,8 @@ static void radeon_cp_dispatch_clear( drm_device_t *dev, RING_LOCALS; DRM_DEBUG( "flags = 0x%x\n", flags ); + dev_priv->stats.clears++; + if ( dev_priv->page_flipping && dev_priv->current_page == 1 ) { unsigned int tmp = flags; @@ -336,120 +421,251 @@ static void radeon_cp_dispatch_clear( drm_device_t *dev, if ( tmp & RADEON_BACK ) flags |= RADEON_FRONT; } + if ( flags & (RADEON_FRONT | RADEON_BACK) ) { + + BEGIN_RING( 4 ); + + /* Ensure the 3D stream is idle before doing a + * 2D fill to clear the front or back buffer. + */ + RADEON_WAIT_UNTIL_3D_IDLE(); + + OUT_RING( CP_PACKET0( RADEON_DP_WRITE_MASK, 0 ) ); + OUT_RING( clear->color_mask ); + + ADVANCE_RING(); + + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->ctx_owner = 0; + + for ( i = 0 ; i < nbox ; i++ ) { + int x = pbox[i].x1; + int y = pbox[i].y1; + int w = pbox[i].x2 - x; + int h = pbox[i].y2 - y; + + DRM_DEBUG( "dispatch clear %d,%d-%d,%d flags 0x%x\n", + x, y, w, h, flags ); + + if ( flags & RADEON_FRONT ) { + BEGIN_RING( 6 ); + + OUT_RING( CP_PACKET3( RADEON_CNTL_PAINT_MULTI, 4 ) ); + OUT_RING( RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_SOLID_COLOR | + (dev_priv->color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_P | + RADEON_GMC_CLR_CMP_CNTL_DIS ); + + OUT_RING( dev_priv->front_pitch_offset ); + OUT_RING( clear->clear_color ); + + OUT_RING( (x << 16) | y ); + OUT_RING( (w << 16) | h ); + + ADVANCE_RING(); + } + + if ( flags & RADEON_BACK ) { + BEGIN_RING( 6 ); + + OUT_RING( CP_PACKET3( RADEON_CNTL_PAINT_MULTI, 4 ) ); + OUT_RING( RADEON_GMC_DST_PITCH_OFFSET_CNTL | + RADEON_GMC_BRUSH_SOLID_COLOR | + (dev_priv->color_fmt << 8) | + RADEON_GMC_SRC_DATATYPE_COLOR | + RADEON_ROP3_P | + RADEON_GMC_CLR_CMP_CNTL_DIS ); + + OUT_RING( dev_priv->back_pitch_offset ); + OUT_RING( clear->clear_color ); + + OUT_RING( (x << 16) | y ); + OUT_RING( (w << 16) | h ); + + ADVANCE_RING(); + } + } + } + /* We have to clear the depth and/or stencil buffers by * rendering a quad into just those buffers. Thus, we have to * make sure the 3D engine is configured correctly. */ - if ( flags & (RADEON_DEPTH | RADEON_STENCIL) ) { - rb3d_cntl = depth_clear->rb3d_cntl; + if ( dev_priv->is_r200 && + (flags & (RADEON_DEPTH | RADEON_STENCIL)) ) { - if ( flags & RADEON_DEPTH ) { - rb3d_cntl |= RADEON_Z_ENABLE; - } else { - rb3d_cntl &= ~RADEON_Z_ENABLE; - } + int tempPP_CNTL; + int tempRE_CNTL; + int tempRB3D_CNTL; + int tempRB3D_ZSTENCILCNTL; + int tempRB3D_STENCILREFMASK; + int tempRB3D_PLANEMASK; + int tempSE_CNTL; + int tempSE_VTE_CNTL; + int tempSE_VTX_FMT_0; + int tempSE_VTX_FMT_1; + int tempSE_VAP_CNTL; + int tempRE_AUX_SCISSOR_CNTL; - if ( flags & RADEON_STENCIL ) { - rb3d_cntl |= RADEON_STENCIL_ENABLE; - rb3d_stencilrefmask = clear->depth_mask; /* misnamed field */ - } else { - rb3d_cntl &= ~RADEON_STENCIL_ENABLE; - rb3d_stencilrefmask = 0x00000000; - } - } + tempPP_CNTL = 0; + tempRE_CNTL = 0; - for ( i = 0 ; i < nbox ; i++ ) { - int x = pbox[i].x1; - int y = pbox[i].y1; - int w = pbox[i].x2 - x; - int h = pbox[i].y2 - y; + tempRB3D_CNTL = depth_clear->rb3d_cntl; + tempRB3D_CNTL &= ~(1<<15); /* unset radeon magic flag */ - DRM_DEBUG( "dispatch clear %d,%d-%d,%d flags 0x%x\n", - x, y, w, h, flags ); + tempRB3D_ZSTENCILCNTL = depth_clear->rb3d_zstencilcntl; + tempRB3D_STENCILREFMASK = 0x0; - if ( flags & (RADEON_FRONT | RADEON_BACK) ) { - BEGIN_RING( 4 ); + tempSE_CNTL = depth_clear->se_cntl; - /* Ensure the 3D stream is idle before doing a - * 2D fill to clear the front or back buffer. - */ - RADEON_WAIT_UNTIL_3D_IDLE(); - OUT_RING( CP_PACKET0( RADEON_DP_WRITE_MASK, 0 ) ); - OUT_RING( clear->color_mask ); - ADVANCE_RING(); + /* Disable TCL */ - /* Make sure we restore the 3D state next time. - */ - dev_priv->sarea_priv->ctx_owner = 0; - } + tempSE_VAP_CNTL = (/* SE_VAP_CNTL__FORCE_W_TO_ONE_MASK | */ + (0x9 << SE_VAP_CNTL__VF_MAX_VTX_NUM__SHIFT)); - if ( flags & RADEON_FRONT ) { - BEGIN_RING( 6 ); + tempRB3D_PLANEMASK = 0x0; - OUT_RING( CP_PACKET3( RADEON_CNTL_PAINT_MULTI, 4 ) ); - OUT_RING( RADEON_GMC_DST_PITCH_OFFSET_CNTL | - RADEON_GMC_BRUSH_SOLID_COLOR | - (dev_priv->color_fmt << 8) | - RADEON_GMC_SRC_DATATYPE_COLOR | - RADEON_ROP3_P | - RADEON_GMC_CLR_CMP_CNTL_DIS ); + tempRE_AUX_SCISSOR_CNTL = 0x0; - OUT_RING( dev_priv->front_pitch_offset ); - OUT_RING( clear->clear_color ); + tempSE_VTE_CNTL = + SE_VTE_CNTL__VTX_XY_FMT_MASK | + SE_VTE_CNTL__VTX_Z_FMT_MASK; - OUT_RING( (x << 16) | y ); - OUT_RING( (w << 16) | h ); + /* Vertex format (X, Y, Z, W)*/ + tempSE_VTX_FMT_0 = + SE_VTX_FMT_0__VTX_Z0_PRESENT_MASK | + SE_VTX_FMT_0__VTX_W0_PRESENT_MASK; + tempSE_VTX_FMT_1 = 0x0; - ADVANCE_RING(); + + /* + * Depth buffer specific enables + */ + if (flags & RADEON_DEPTH) { + /* Enable depth buffer */ + tempRB3D_CNTL |= RADEON_Z_ENABLE; + } else { + /* Disable depth buffer */ + tempRB3D_CNTL &= ~RADEON_Z_ENABLE; } - if ( flags & RADEON_BACK ) { - BEGIN_RING( 6 ); + /* + * Stencil buffer specific enables + */ + if ( flags & RADEON_STENCIL ) { + tempRB3D_CNTL |= RADEON_STENCIL_ENABLE; + tempRB3D_STENCILREFMASK = clear->depth_mask; + } else { + tempRB3D_CNTL &= ~RADEON_STENCIL_ENABLE; + tempRB3D_STENCILREFMASK = 0x00000000; + } - OUT_RING( CP_PACKET3( RADEON_CNTL_PAINT_MULTI, 4 ) ); - OUT_RING( RADEON_GMC_DST_PITCH_OFFSET_CNTL | - RADEON_GMC_BRUSH_SOLID_COLOR | - (dev_priv->color_fmt << 8) | - RADEON_GMC_SRC_DATATYPE_COLOR | - RADEON_ROP3_P | - RADEON_GMC_CLR_CMP_CNTL_DIS ); + BEGIN_RING( 26 ); + RADEON_WAIT_UNTIL_2D_IDLE(); + + OUT_RING_REG( RADEON_PP_CNTL, tempPP_CNTL ); + OUT_RING_REG( R200_RE_CNTL, tempRE_CNTL ); + OUT_RING_REG( RADEON_RB3D_CNTL, tempRB3D_CNTL ); + OUT_RING_REG( RADEON_RB3D_ZSTENCILCNTL, + tempRB3D_ZSTENCILCNTL ); + OUT_RING_REG( RADEON_RB3D_STENCILREFMASK, + tempRB3D_STENCILREFMASK ); + OUT_RING_REG( RADEON_RB3D_PLANEMASK, tempRB3D_PLANEMASK ); + OUT_RING_REG( RADEON_SE_CNTL, tempSE_CNTL ); + OUT_RING_REG( R200_SE_VTE_CNTL, tempSE_VTE_CNTL ); + OUT_RING_REG( R200_SE_VTX_FMT_0, tempSE_VTX_FMT_0 ); + OUT_RING_REG( R200_SE_VTX_FMT_1, tempSE_VTX_FMT_1 ); + OUT_RING_REG( R200_SE_VAP_CNTL, tempSE_VAP_CNTL ); + OUT_RING_REG( R200_RE_AUX_SCISSOR_CNTL, + tempRE_AUX_SCISSOR_CNTL ); + ADVANCE_RING(); - OUT_RING( dev_priv->back_pitch_offset ); - OUT_RING( clear->clear_color ); + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->ctx_owner = 0; - OUT_RING( (x << 16) | y ); - OUT_RING( (w << 16) | h ); + for ( i = 0 ; i < nbox ; i++ ) { + + /* Funny that this should be required -- + * sets top-left? + */ + radeon_emit_clip_rect( dev_priv, + &sarea_priv->boxes[i] ); + BEGIN_RING( 14 ); + OUT_RING( CP_PACKET3( R200_3D_DRAW_IMMD_2, 12 ) ); + OUT_RING( (RADEON_PRIM_TYPE_RECT_LIST | + RADEON_PRIM_WALK_RING | + (3 << RADEON_NUM_VERTICES_SHIFT)) ); + OUT_RING( depth_boxes[i].ui[CLEAR_X1] ); + OUT_RING( depth_boxes[i].ui[CLEAR_Y1] ); + OUT_RING( depth_boxes[i].ui[CLEAR_DEPTH] ); + OUT_RING( 0x3f800000 ); + OUT_RING( depth_boxes[i].ui[CLEAR_X1] ); + OUT_RING( depth_boxes[i].ui[CLEAR_Y2] ); + OUT_RING( depth_boxes[i].ui[CLEAR_DEPTH] ); + OUT_RING( 0x3f800000 ); + OUT_RING( depth_boxes[i].ui[CLEAR_X2] ); + OUT_RING( depth_boxes[i].ui[CLEAR_Y2] ); + OUT_RING( depth_boxes[i].ui[CLEAR_DEPTH] ); + OUT_RING( 0x3f800000 ); ADVANCE_RING(); } + } + else if ( (flags & (RADEON_DEPTH | RADEON_STENCIL)) ) { - if ( flags & (RADEON_DEPTH | RADEON_STENCIL) ) { + rb3d_cntl = depth_clear->rb3d_cntl; - radeon_emit_clip_rect( dev_priv, - &sarea_priv->boxes[i] ); + if ( flags & RADEON_DEPTH ) { + rb3d_cntl |= RADEON_Z_ENABLE; + } else { + rb3d_cntl &= ~RADEON_Z_ENABLE; + } - BEGIN_RING( 28 ); + if ( flags & RADEON_STENCIL ) { + rb3d_cntl |= RADEON_STENCIL_ENABLE; + rb3d_stencilrefmask = clear->depth_mask; /* misnamed field */ + } else { + rb3d_cntl &= ~RADEON_STENCIL_ENABLE; + rb3d_stencilrefmask = 0x00000000; + } - RADEON_WAIT_UNTIL_2D_IDLE(); + BEGIN_RING( 13 ); + RADEON_WAIT_UNTIL_2D_IDLE(); - OUT_RING( CP_PACKET0( RADEON_PP_CNTL, 1 ) ); - OUT_RING( 0x00000000 ); - OUT_RING( rb3d_cntl ); + OUT_RING( CP_PACKET0( RADEON_PP_CNTL, 1 ) ); + OUT_RING( 0x00000000 ); + OUT_RING( rb3d_cntl ); + + OUT_RING_REG( RADEON_RB3D_ZSTENCILCNTL, + depth_clear->rb3d_zstencilcntl ); + OUT_RING_REG( RADEON_RB3D_STENCILREFMASK, + rb3d_stencilrefmask ); + OUT_RING_REG( RADEON_RB3D_PLANEMASK, + 0x00000000 ); + OUT_RING_REG( RADEON_SE_CNTL, + depth_clear->se_cntl ); + ADVANCE_RING(); - OUT_RING_REG( RADEON_RB3D_ZSTENCILCNTL, - depth_clear->rb3d_zstencilcntl ); - OUT_RING_REG( RADEON_RB3D_STENCILREFMASK, - rb3d_stencilrefmask ); - OUT_RING_REG( RADEON_RB3D_PLANEMASK, - 0x00000000 ); - OUT_RING_REG( RADEON_SE_CNTL, - depth_clear->se_cntl ); + /* Make sure we restore the 3D state next time. + */ + dev_priv->sarea_priv->ctx_owner = 0; - /* Radeon 7500 doesn't like vertices without - * color. + for ( i = 0 ; i < nbox ; i++ ) { + + /* Funny that this should be required -- + * sets top-left? */ + radeon_emit_clip_rect( dev_priv, + &sarea_priv->boxes[i] ); + + BEGIN_RING( 15 ); + OUT_RING( CP_PACKET3( RADEON_3D_DRAW_IMMD, 13 ) ); OUT_RING( RADEON_VTX_Z_PRESENT | RADEON_VTX_PKCOLOR_PRESENT); @@ -459,6 +675,7 @@ static void radeon_cp_dispatch_clear( drm_device_t *dev, RADEON_VTX_FMT_RADEON_MODE | (3 << RADEON_NUM_VERTICES_SHIFT)) ); + OUT_RING( depth_boxes[i].ui[CLEAR_X1] ); OUT_RING( depth_boxes[i].ui[CLEAR_Y1] ); OUT_RING( depth_boxes[i].ui[CLEAR_DEPTH] ); @@ -475,10 +692,6 @@ static void radeon_cp_dispatch_clear( drm_device_t *dev, OUT_RING( 0x0 ); ADVANCE_RING(); - - /* Make sure we restore the 3D state next time. - */ - dev_priv->sarea_priv->ctx_owner = 0; } } @@ -506,11 +719,12 @@ static void radeon_cp_dispatch_swap( drm_device_t *dev ) RING_LOCALS; DRM_DEBUG( "\n" ); -#if RADEON_PERFORMANCE_BOXES + /* Do some trivial performance monitoring... */ - radeon_cp_performance_boxes( dev_priv ); -#endif + if (dev_priv->do_boxes) + radeon_cp_performance_boxes( dev_priv ); + /* Wait for the 3D stream to idle before dispatching the bitblt. * This will prevent data corruption between the two streams. @@ -579,20 +793,21 @@ static void radeon_cp_dispatch_flip( drm_device_t *dev ) { drm_radeon_private_t *dev_priv = dev->dev_private; RING_LOCALS; - DRM_DEBUG( "page=%d\n", dev_priv->current_page ); + DRM_DEBUG( "%s: page=%d pfCurrentPage=%d\n", + __FUNCTION__, + dev_priv->current_page, + dev_priv->sarea_priv->pfCurrentPage); -#if RADEON_PERFORMANCE_BOXES /* Do some trivial performance monitoring... */ - radeon_cp_performance_boxes( dev_priv ); -#endif + if (dev_priv->do_boxes) { + dev_priv->stats.boxes |= RADEON_BOX_FLIP; + radeon_cp_performance_boxes( dev_priv ); + } BEGIN_RING( 4 ); RADEON_WAIT_UNTIL_3D_IDLE(); -/* - RADEON_WAIT_UNTIL_PAGE_FLIPPED(); -*/ OUT_RING( CP_PACKET0( RADEON_CRTC_OFFSET, 0 ) ); if ( dev_priv->current_page == 0 ) { @@ -847,6 +1062,8 @@ static int radeon_cp_dispatch_texture( drm_device_t *dev, int ret = 0, i; RING_LOCALS; + dev_priv->stats.boxes |= RADEON_BOX_TEXTURE_LOAD; + /* FIXME: Be smarter about this... */ buf = radeon_freelist_get( dev ); @@ -1611,6 +1828,30 @@ static __inline__ int radeon_emit_scalars( return 0; } +/* God this is ugly + */ +static __inline__ int radeon_emit_scalars2( + drm_radeon_private_t *dev_priv, + drm_radeon_cmd_header_t header, + drm_radeon_cmd_buffer_t *cmdbuf ) +{ + int sz = header.scalars.count; + int *data = (int *)cmdbuf->buf; + int start = ((unsigned int)header.scalars.offset) + 0x100; + int stride = header.scalars.stride; + RING_LOCALS; + + BEGIN_RING( 3+sz ); + OUT_RING( CP_PACKET0( RADEON_SE_TCL_SCALAR_INDX_REG, 0 ) ); + OUT_RING( start | (stride << RADEON_SCAL_INDX_DWORD_STRIDE_SHIFT)); + OUT_RING( CP_PACKET0_TABLE( RADEON_SE_TCL_SCALAR_DATA_REG, sz-1 ) ); + OUT_RING_USER_TABLE( data, sz ); + ADVANCE_RING(); + cmdbuf->buf += sz * sizeof(int); + cmdbuf->bufsz -= sz * sizeof(int); + return 0; +} + static __inline__ int radeon_emit_vectors( drm_radeon_private_t *dev_priv, drm_radeon_cmd_header_t header, @@ -1775,6 +2016,7 @@ int radeon_cp_cmdbuf( DRM_IOCTL_ARGS ) switch (header.header.cmd_type) { case RADEON_CMD_PACKET: + DRM_DEBUG("RADEON_CMD_PACKET\n"); if (radeon_emit_packets( dev_priv, header, &cmdbuf )) { DRM_ERROR("radeon_emit_packets failed\n"); return DRM_ERR(EINVAL); @@ -1782,6 +2024,7 @@ int radeon_cp_cmdbuf( DRM_IOCTL_ARGS ) break; case RADEON_CMD_SCALARS: + DRM_DEBUG("RADEON_CMD_SCALARS\n"); if (radeon_emit_scalars( dev_priv, header, &cmdbuf )) { DRM_ERROR("radeon_emit_scalars failed\n"); return DRM_ERR(EINVAL); @@ -1789,6 +2032,7 @@ int radeon_cp_cmdbuf( DRM_IOCTL_ARGS ) break; case RADEON_CMD_VECTORS: + DRM_DEBUG("RADEON_CMD_VECTORS\n"); if (radeon_emit_vectors( dev_priv, header, &cmdbuf )) { DRM_ERROR("radeon_emit_vectors failed\n"); return DRM_ERR(EINVAL); @@ -1796,6 +2040,7 @@ int radeon_cp_cmdbuf( DRM_IOCTL_ARGS ) break; case RADEON_CMD_DMA_DISCARD: + DRM_DEBUG("RADEON_CMD_DMA_DISCARD\n"); idx = header.dma.buf_idx; if ( idx < 0 || idx >= dma->buf_count ) { DRM_ERROR( "buffer index %d (of %d max)\n", @@ -1813,6 +2058,7 @@ int radeon_cp_cmdbuf( DRM_IOCTL_ARGS ) break; case RADEON_CMD_PACKET3: + DRM_DEBUG("RADEON_CMD_PACKET3\n"); if (radeon_emit_packet3( dev, &cmdbuf )) { DRM_ERROR("radeon_emit_packet3 failed\n"); return DRM_ERR(EINVAL); @@ -1820,12 +2066,20 @@ int radeon_cp_cmdbuf( DRM_IOCTL_ARGS ) break; case RADEON_CMD_PACKET3_CLIP: + DRM_DEBUG("RADEON_CMD_PACKET3_CLIP\n"); if (radeon_emit_packet3_cliprect( dev, &cmdbuf, orig_nbox )) { DRM_ERROR("radeon_emit_packet3_clip failed\n"); return DRM_ERR(EINVAL); } break; + case RADEON_CMD_SCALARS2: + DRM_DEBUG("RADEON_CMD_SCALARS2\n"); + if (radeon_emit_scalars2( dev_priv, header, &cmdbuf )) { + DRM_ERROR("radeon_emit_scalars2 failed\n"); + return DRM_ERR(EINVAL); + } + break; default: DRM_ERROR("bad cmd_type %d at %p\n", header.header.cmd_type, @@ -1835,6 +2089,7 @@ int radeon_cp_cmdbuf( DRM_IOCTL_ARGS ) } + DRM_DEBUG("DONE\n"); COMMIT_RING(); return 0; } @@ -1863,12 +2118,14 @@ int radeon_cp_getparam( DRM_IOCTL_ARGS ) value = dev_priv->agp_buffers_offset; break; case RADEON_PARAM_LAST_FRAME: + dev_priv->stats.last_frame_reads++; value = GET_SCRATCH( 0 ); break; case RADEON_PARAM_LAST_DISPATCH: value = GET_SCRATCH( 1 ); break; case RADEON_PARAM_LAST_CLEAR: + dev_priv->stats.last_clear_reads++; value = GET_SCRATCH( 2 ); break; default: |