Merge branch 'master' into gammagamma

Conflicts:
	pixman/pixman-transformed.c

18 files changed, 838 insertions, 156 deletions

diff --git a/configure.ac b/configure.ac
index 030d063d..ebb55575 100644
--- a/configure.ac
+++ b/configure.ac
@@ -63,11 +63,18 @@ AM_INIT_AUTOMAKE([dist-bzip2])
 
 AM_CONFIG_HEADER(config.h)
 
+AC_CANONICAL_HOST
+
 AC_PROG_CC
 AC_PROG_LIBTOOL
 AC_CHECK_FUNCS([getisax])
 AC_C_BIGENDIAN
 AC_C_INLINE
+
+# Checks for Sun Studio compilers
+AC_CHECK_DECL([__SUNPRO_C], [SUNCC="yes"], [SUNCC="no"])
+AC_CHECK_DECL([__amd64], [AMD64_ABI="yes"], [AMD64_ABI="no"])
+
 # 
 # We ignore pixman_major in the version here because the major version should
 # always be encoded in the actual library name. Ie., the soname is:
@@ -141,12 +148,23 @@ AC_MSG_RESULT($have_sunstudio8)
 dnl ===========================================================================
 dnl Check for MMX
 
-MMX_CFLAGS="-mmmx -Winline"
+if test "x$MMX_CFLAGS" = "x" ; then
+   if test "x$SUNCC" = "xyes"; then
+      # Sun Studio doesn't have an -xarch=mmx flag, so we have to use sse
+      # but if we're building 64-bit, mmx & sse support is on by default and
+      # -xarch=sse throws an error instead
+      if test "$AMD64_ABI" = "no" ; then
+         MMX_CFLAGS="-xarch=sse"
+      fi
+   else
+      MMX_CFLAGS="-mmmx -Winline"
+   fi
+fi
 
 have_mmx_intrinsics=no
 AC_MSG_CHECKING(whether to use MMX intrinsics)
 xserver_save_CFLAGS=$CFLAGS
-CFLAGS="$CFLAGS $MMX_CFLAGS"
+CFLAGS="$MMX_CFLAGS $CFLAGS"
 AC_COMPILE_IFELSE([
 #if defined(__GNUC__) && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 4))
 error "Need GCC >= 3.4 for MMX intrinsics"
@@ -183,12 +201,21 @@ AM_CONDITIONAL(USE_MMX, test $have_mmx_intrinsics = yes)
 dnl ===========================================================================
 dnl Check for SSE2
 
-SSE2_CFLAGS="-mmmx -msse2 -Winline"
+if test "x$SSE2_CFLAGS" = "x" ; then
+   if test "x$SUNCC" = "xyes"; then
+      # SSE2 is enabled by default in the Sun Studio 64-bit environment
+      if test "$AMD64_ABI" = "no" ; then
+         SSE2_CFLAGS="-xarch=sse2"
+      fi
+   else
+      SSE2_CFLAGS="-mmmx -msse2 -Winline"
+   fi
+fi
 
 have_sse2_intrinsics=no
 AC_MSG_CHECKING(whether to use SSE2 intrinsics)
 xserver_save_CFLAGS=$CFLAGS
-CFLAGS="$CFLAGS -msse2 $SSE2_CFLAGS"
+CFLAGS="$SSE2_CFLAGS $CFLAGS"
 
 AC_COMPILE_IFELSE([
 #if defined(__GNUC__) && (__GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 2))
@@ -226,10 +253,31 @@ fi
 
 AM_CONDITIONAL(USE_SSE2, test $have_sse2_intrinsics = yes)
 
-dnl ========================================================
+dnl ===========================================================================
+dnl Other special flags needed when building code using MMX or SSE instructions
+case $host_os in
+   solaris*)
+      # When building 32-bit binaries, apply a mapfile to ensure that the
+      # binaries aren't flagged as only able to run on MMX+SSE capable CPUs
+      # since they check at runtime before using those instructions
+      if test "$AMD64_ABI" = "no" ; then
+         HWCAP_LDFLAGS='-Wl,-M,$(srcdir)/solaris-hwcap.mapfile'
+      fi
+      if test "x$MMX_LDFLAGS" = "x" ; then
+         MMX_LDFLAGS="$HWCAP_LDFLAGS"
+      fi
+      if test "x$SSE2_LDFLAGS" = "x" ; then
+	 SSE2_LDFLAGS="$HWCAP_LDFLAGS"
+      fi
+      ;;
+esac
+
 AC_SUBST(MMX_CFLAGS)
+AC_SUBST(MMX_LDFLAGS)
 AC_SUBST(SSE2_CFLAGS)
+AC_SUBST(SSE2_LDFLAGS)
 
+dnl ===========================================================================
 dnl Check for VMX/Altivec
 if test -n "`$CC -v 2>&1 | grep version | grep Apple`"; then
     VMX_CFLAGS="-faltivec"
diff --git a/pixman/Makefile.am b/pixman/Makefile.am
index c4612ea5..b8589be6 100644
--- a/pixman/Makefile.am
+++ b/pixman/Makefile.am
@@ -54,6 +54,7 @@ libpixman_mmx_la_SOURCES = \
 	pixman-mmx.h
 libpixman_mmx_la_CFLAGS = $(DEP_CFLAGS) $(MMX_CFLAGS)
 libpixman_mmx_la_LIBADD = $(DEP_LIBS)
+libpixman_1_la_LDFLAGS += $(MMX_LDFLAGS)
 libpixman_1_la_LIBADD += libpixman-mmx.la
 endif
 
@@ -77,6 +78,7 @@ libpixman_sse2_la_SOURCES = \
 	pixman-sse2.h
 libpixman_sse2_la_CFLAGS = $(DEP_CFLAGS) $(SSE2_CFLAGS)
 libpixman_sse2_la_LIBADD = $(DEP_LIBS)
+libpixman_1_la_LDFLAGS += $(SSE2_LDFLAGS)
 libpixman_1_la_LIBADD += libpixman-sse2.la
 endif
 
diff --git a/pixman/pixman-access.c b/pixman/pixman-access.c
index bca45810..2b428769 100644
--- a/pixman/pixman-access.c
+++ b/pixman/pixman-access.c
@@ -121,6 +121,36 @@ fbFetch_x8b8g8r8 (bits_image_t *pict, int x, int y, int width, uint32_t *buffer)
 }
 
 static FASTCALL void
+fbFetch_b8g8r8a8 (bits_image_t *pict, int x, int y, int width, uint32_t *buffer)
+{
+    const uint32_t *bits = pict->bits + y*pict->rowstride;
+    const uint32_t *pixel = (uint32_t *)bits + x;
+    const uint32_t *end = pixel + width;
+    while (pixel < end) {
+	uint32_t p = READ(pict, pixel++);
+	*buffer++ = ((p & 0xff000000) >> 24) |
+	    ((p & 0x00ff0000) >> 8) |
+	    ((p & 0x0000ff00) << 8) |
+	    ((p & 0x000000ff) << 24);
+    }
+}
+
+static FASTCALL void
+fbFetch_b8g8r8x8 (bits_image_t *pict, int x, int y, int width, uint32_t *buffer)
+{
+    const uint32_t *bits = pict->bits + y*pict->rowstride;
+    const uint32_t *pixel = (uint32_t *)bits + x;
+    const uint32_t *end = pixel + width;
+    while (pixel < end) {
+	uint32_t p = READ(pict, pixel++);
+	*buffer++ = 0xff000000 |
+	    ((p & 0xff000000) >> 24) |
+	    ((p & 0x00ff0000) >> 8) |
+	    ((p & 0x0000ff00) << 8);
+    }
+}
+
+static FASTCALL void
 fbFetch_a2b10g10r10 (bits_image_t *pict, int x, int y, int width, uint64_t *buffer)
 {
     const uint32_t *bits = pict->bits + y*pict->rowstride;
@@ -694,6 +724,8 @@ fetchProc32 ACCESS(pixman_fetchProcForPicture32) (bits_image_t * pict)
     case PIXMAN_x8r8g8b8: return fbFetch_x8r8g8b8;
     case PIXMAN_a8b8g8r8: return fbFetch_a8b8g8r8;
     case PIXMAN_x8b8g8r8: return fbFetch_x8b8g8r8;
+    case PIXMAN_b8g8r8a8: return fbFetch_b8g8r8a8;
+    case PIXMAN_b8g8r8x8: return fbFetch_b8g8r8x8;
     /* These two require wide compositing */
     case PIXMAN_a2b10g10r10: return NULL;
     case PIXMAN_x2b10g10r10: return NULL;
@@ -845,6 +877,30 @@ fbFetchPixel_x8b8g8r8 (bits_image_t *pict, int offset, int line)
 }
 
 static FASTCALL uint32_t
+fbFetchPixel_b8g8r8a8 (bits_image_t *pict, int offset, int line)
+{
+    uint32_t *bits = pict->bits + line*pict->rowstride;
+    uint32_t  pixel = READ(pict, (uint32_t *)bits + offset);
+
+    return ((pixel & 0xff000000) >> 24 |
+	    (pixel & 0x00ff0000) >> 8 |
+	    (pixel & 0x0000ff00) << 8 |
+	    (pixel & 0x000000ff) << 24);
+}
+
+static FASTCALL uint32_t
+fbFetchPixel_b8g8r8x8 (bits_image_t *pict, int offset, int line)
+{
+    uint32_t *bits = pict->bits + line*pict->rowstride;
+    uint32_t  pixel = READ(pict, (uint32_t *)bits + offset);
+
+    return ((0xff000000) |
+	    (pixel & 0xff000000) >> 24 |
+	    (pixel & 0x00ff0000) >> 8 |
+	    (pixel & 0x0000ff00) << 8);
+}
+
+static FASTCALL uint32_t
 fbFetchPixel_r8g8b8 (bits_image_t *pict, int offset, int line)
 {
     uint32_t *bits = pict->bits + line*pict->rowstride;
@@ -1284,6 +1340,8 @@ fetchPixelProc32 ACCESS(pixman_fetchPixelProcForPicture32) (bits_image_t * pict)
     case PIXMAN_x8r8g8b8: return fbFetchPixel_x8r8g8b8;
     case PIXMAN_a8b8g8r8: return fbFetchPixel_a8b8g8r8;
     case PIXMAN_x8b8g8r8: return fbFetchPixel_x8b8g8r8;
+    case PIXMAN_b8g8r8a8: return fbFetchPixel_b8g8r8a8;
+    case PIXMAN_b8g8r8x8: return fbFetchPixel_b8g8r8x8;
     /* These two require wide compositing */
     case PIXMAN_a2b10g10r10: return fbFetchPixel32_generic_lossy;
     case PIXMAN_x2b10g10r10: return fbFetchPixel32_generic_lossy;
@@ -1429,6 +1487,33 @@ fbStore_x8b8g8r8 (pixman_image_t *image,
 }
 
 static FASTCALL void
+fbStore_b8g8r8a8 (pixman_image_t *image,
+		  uint32_t *bits, const uint32_t *values, int x, int width, const pixman_indexed_t * indexed)
+{
+    int i;
+    uint32_t *pixel = (uint32_t *)bits + x;
+    for (i = 0; i < width; ++i)
+	WRITE(image, pixel++,
+	    ((values[i] >> 24) & 0x000000ff) |
+	    ((values[i] >>  8) & 0x0000ff00) |
+	    ((values[i] <<  8) & 0x00ff0000) |
+	    ((values[i] << 24) & 0xff000000));
+}
+
+static FASTCALL void
+fbStore_b8g8r8x8 (pixman_image_t *image,
+		  uint32_t *bits, const uint32_t *values, int x, int width, const pixman_indexed_t * indexed)
+{
+    int i;
+    uint32_t *pixel = (uint32_t *)bits + x;
+    for (i = 0; i < width; ++i)
+	WRITE(image, pixel++,
+	    ((values[i] >>  8) & 0x0000ff00) |
+	    ((values[i] <<  8) & 0x00ff0000) |
+	    ((values[i] << 24) & 0xff000000));
+}
+
+static FASTCALL void
 fbStore_r8g8b8 (pixman_image_t *image,
 		uint32_t *bits, const uint32_t *values, int x, int width,
 		const pixman_indexed_t * indexed)
@@ -1817,6 +1902,8 @@ storeProc32 ACCESS(pixman_storeProcForPicture32) (bits_image_t * pict)
     case PIXMAN_x8r8g8b8: return fbStore_x8r8g8b8;
     case PIXMAN_a8b8g8r8: return fbStore_a8b8g8r8;
     case PIXMAN_x8b8g8r8: return fbStore_x8b8g8r8;
+    case PIXMAN_b8g8r8a8: return fbStore_b8g8r8a8;
+    case PIXMAN_b8g8r8x8: return fbStore_b8g8r8x8;
 
         /* 24bpp formats */
     case PIXMAN_r8g8b8: return fbStore_r8g8b8;
diff --git a/pixman/pixman-image.c b/pixman/pixman-image.c
index e80c4793..bd52f253 100644
--- a/pixman/pixman-image.c
+++ b/pixman/pixman-image.c
@@ -631,6 +631,8 @@ color_to_pixel (pixman_color_t *color,
 	  format == PIXMAN_x8r8g8b8	||
 	  format == PIXMAN_a8b8g8r8	||
 	  format == PIXMAN_x8b8g8r8	||
+	  format == PIXMAN_b8g8r8a8	||
+	  format == PIXMAN_b8g8r8x8	||
 	  format == PIXMAN_r5g6b5	||
 	  format == PIXMAN_b5g6r5	||
 	  format == PIXMAN_a8))
@@ -645,6 +647,13 @@ color_to_pixel (pixman_color_t *color,
 	    ((c & 0x0000ff00) >>  0) |
 	    ((c & 0x000000ff) << 16);
     }
+    if (PIXMAN_FORMAT_TYPE (format) == PIXMAN_TYPE_BGRA)
+    {
+	c = ((c & 0xff000000) >> 24) |
+	    ((c & 0x00ff0000) >>  8) |
+	    ((c & 0x0000ff00) <<  8) |
+	    ((c & 0x000000ff) << 24);
+    }
 
     if (format == PIXMAN_a8)
 	c = c >> 24;
@@ -703,7 +712,11 @@ pixman_image_fill_rectangles (pixman_op_t		    op,
 		pixman_box32_t *boxes;
 
 		pixman_region32_init_rect (&fill_region, rects[i].x, rects[i].y, rects[i].width, rects[i].height);
-		pixman_region32_intersect (&fill_region, &fill_region, &dest->common.clip_region);
+		if (!pixman_region32_intersect (&fill_region,
+						&fill_region,
+						&dest->common.clip_region))
+		    return FALSE;
+
 
 		boxes = pixman_region32_rectangles (&fill_region, &n_boxes);
 		for (j = 0; j < n_boxes; ++j)
@@ -761,6 +774,8 @@ pixman_image_can_get_solid (pixman_image_t *image)
     case PIXMAN_x8r8g8b8:
     case PIXMAN_a8b8g8r8:
     case PIXMAN_x8b8g8r8:
+    case PIXMAN_b8g8r8a8:
+    case PIXMAN_b8g8r8x8:
     case PIXMAN_r8g8b8:
     case PIXMAN_b8g8r8:
     case PIXMAN_r5g6b5:
diff --git a/pixman/pixman-mmx.c b/pixman/pixman-mmx.c
index caeeafc0..7d0a8023 100644
--- a/pixman/pixman-mmx.c
+++ b/pixman/pixman-mmx.c
@@ -71,15 +71,19 @@
  * possible.
  */
 
-/* --------------- MMX primitivess ------------------------------------ */
+/* --------------- MMX primitives ------------------------------------- */
 
 #ifdef __GNUC__
-typedef unsigned long long ullong;
-typedef ullong mmxdatafield;
-#endif
-#ifdef _MSC_VER
-typedef unsigned __int64 ullong;
+typedef uint64_t mmxdatafield;
+#else
 typedef __m64 mmxdatafield;
+/* If __m64 is defined as a struct or union, define M64_MEMBER to be the
+   name of the member used to access the data */
+# ifdef _MSC_VER
+#  define M64_MEMBER m64_u64
+# elif defined(__SUNPRO_C)
+#  define M64_MEMBER l_
+# endif
 #endif
 
 typedef struct
@@ -101,42 +105,31 @@ typedef struct
     mmxdatafield mmx_000000000000ffff;
 } MMXData;
 
+#if defined(_MSC_VER)
+# define MMXDATA_INIT(field, val) { val##UI64 }
+#elif defined(M64_MEMBER)	/* __m64 is a struct, not an integral type */
+# define MMXDATA_INIT(field, val) field =   { val##ULL }
+#else				/* __m64 is an integral type */
+# define MMXDATA_INIT(field, val) field =   val##ULL
+#endif
+
 static const MMXData c =
 {
-#ifdef __GNUC__
-    .mmx_4x00ff =			0x00ff00ff00ff00ffULL,
-    .mmx_4x0080 =			0x0080008000800080ULL,
-    .mmx_565_rgb =			0x000001f0003f001fULL,
-    .mmx_565_unpack_multiplier =	0x0000008404100840ULL,
-    .mmx_565_r =			0x000000f800000000ULL,
-    .mmx_565_g =			0x0000000000fc0000ULL,
-    .mmx_565_b =			0x00000000000000f8ULL,
-    .mmx_mask_0 =			0xffffffffffff0000ULL,
-    .mmx_mask_1 =			0xffffffff0000ffffULL,
-    .mmx_mask_2 =			0xffff0000ffffffffULL,
-    .mmx_mask_3 =			0x0000ffffffffffffULL,
-    .mmx_full_alpha =			0x00ff000000000000ULL,
-    .mmx_ffff0000ffff0000 =		0xffff0000ffff0000ULL,
-    .mmx_0000ffff00000000 =		0x0000ffff00000000ULL,
-    .mmx_000000000000ffff =		0x000000000000ffffULL,
-#endif
-#ifdef _MSC_VER
-    { 0x00ff00ff00ff00ffUI64 },
-    { 0x0080008000800080UI64 },
-    { 0x000001f0003f001fUI64 },
-    { 0x0000008404100840UI64 },
-    { 0x000000f800000000UI64 },
-    { 0x0000000000fc0000UI64 },
-    { 0x00000000000000f8UI64 },
-    { 0xffffffffffff0000UI64 },
-    { 0xffffffff0000ffffUI64 },
-    { 0xffff0000ffffffffUI64 },
-    { 0x0000ffffffffffffUI64 },
-    { 0x00ff000000000000UI64 },
-    { 0xffff0000ffff0000UI64 },
-    { 0x0000ffff00000000UI64 },
-    { 0x000000000000ffffUI64 },
-#endif
+    MMXDATA_INIT(.mmx_4x00ff,			0x00ff00ff00ff00ff),
+    MMXDATA_INIT(.mmx_4x0080,			0x0080008000800080),
+    MMXDATA_INIT(.mmx_565_rgb,			0x000001f0003f001f),
+    MMXDATA_INIT(.mmx_565_unpack_multiplier,	0x0000008404100840),
+    MMXDATA_INIT(.mmx_565_r,			0x000000f800000000),
+    MMXDATA_INIT(.mmx_565_g,			0x0000000000fc0000),
+    MMXDATA_INIT(.mmx_565_b,			0x00000000000000f8),
+    MMXDATA_INIT(.mmx_mask_0,			0xffffffffffff0000),
+    MMXDATA_INIT(.mmx_mask_1,			0xffffffff0000ffff),
+    MMXDATA_INIT(.mmx_mask_2,			0xffff0000ffffffff),
+    MMXDATA_INIT(.mmx_mask_3,			0x0000ffffffffffff),
+    MMXDATA_INIT(.mmx_full_alpha,		0x00ff000000000000),
+    MMXDATA_INIT(.mmx_ffff0000ffff0000,		0xffff0000ffff0000),
+    MMXDATA_INIT(.mmx_0000ffff00000000,		0x0000ffff00000000),
+    MMXDATA_INIT(.mmx_000000000000ffff,		0x000000000000ffff),
 };
 
 #ifdef __GNUC__
@@ -145,43 +138,35 @@ static const MMXData c =
 #    else
 #        define MC(x) ((__m64)c.mmx_##x)
 #    endif
-#endif
-
-#ifdef _MSC_VER
+#else
 #    define MC(x) c.mmx_##x
 #endif
 
 static force_inline __m64
-M64 (ullong x)
+M64 (uint64_t x)
 {
 #ifdef __ICC
     return _mm_cvtsi64_m64 (x);
-#elif defined (__GNUC__)
-    return (__m64)x;
-#endif
-
-#ifdef _MSC_VER
+#elif defined M64_MEMBER	/* __m64 is a struct, not an integral type */
     __m64 res;
 
-    res.m64_u64 = x;
+    res.M64_MEMBER = x;
     return res;
+#else				/* __m64 is an integral type */
+    return (__m64)x;
 #endif
 }
 
-static force_inline ullong
-ULLONG (__m64 x)
+static force_inline uint64_t
+UINT64 (__m64 x)
 {
 #ifdef __ICC
     return _mm_cvtm64_si64 (x);
-#elif defined (__GNUC__)
-    return (ullong)x;
-#endif
-
-#ifdef _MSC_VER
-    ullong res;
-
-    res = x.m64_u64;
+#elif defined M64_MEMBER	/* __m64 is a struct, not an integral type */
+    uint64_t res = x.M64_MEMBER;
     return res;
+#else				/* __m64 is an integral type */
+    return (uint64_t)x;
 #endif
 }
 
@@ -1059,10 +1044,10 @@ fbCompositeSolid_nx0565mmx (pixman_op_t op,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    ullong d = *dst;
+	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (M64(d), 0);
 	    vdest = pack565(over(vsrc, vsrca, vdest), vdest, 0);
-	    *dst = ULLONG(vdest);
+	    *dst = UINT64(vdest);
 
 	    w--;
 	    dst++;
@@ -1089,10 +1074,10 @@ fbCompositeSolid_nx0565mmx (pixman_op_t op,
 
 	while (w)
 	{
-	    ullong d = *dst;
+	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (M64(d), 0);
 	    vdest = pack565(over(vsrc, vsrca, vdest), vdest, 0);
-	    *dst = ULLONG(vdest);
+	    *dst = UINT64(vdest);
 
 	    w--;
 	    dst++;
@@ -1517,12 +1502,12 @@ fbCompositeSrc_8888x0565mmx (pixman_op_t op,
 	while (w && (unsigned long)dst & 7)
 	{
 	    __m64 vsrc = load8888 (*src);
-	    ullong d = *dst;
+	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (M64(d), 0);
 
 	    vdest = pack565(over(vsrc, expand_alpha(vsrc), vdest), vdest, 0);
 
-	    *dst = ULLONG(vdest);
+	    *dst = UINT64(vdest);
 
 	    w--;
 	    dst++;
@@ -1560,12 +1545,12 @@ fbCompositeSrc_8888x0565mmx (pixman_op_t op,
 	while (w)
 	{
 	    __m64 vsrc = load8888 (*src);
-	    ullong d = *dst;
+	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (M64(d), 0);
 
 	    vdest = pack565(over(vsrc, expand_alpha(vsrc), vdest), vdest, 0);
 
-	    *dst = ULLONG(vdest);
+	    *dst = UINT64(vdest);
 
 	    w--;
 	    dst++;
@@ -1596,7 +1581,7 @@ fbCompositeSolidMask_nx8x8888mmx (pixman_op_t op,
     int	dstStride, maskStride;
     uint16_t	w;
     __m64	vsrc, vsrca;
-    ullong	srcsrc;
+    uint64_t	srcsrc;
 
     CHECKPOINT();
 
@@ -1606,7 +1591,7 @@ fbCompositeSolidMask_nx8x8888mmx (pixman_op_t op,
     if (srca == 0)
 	return;
 
-    srcsrc = (ullong)src << 32 | src;
+    srcsrc = (uint64_t)src << 32 | src;
 
     fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
     fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
@@ -1626,7 +1611,7 @@ fbCompositeSolidMask_nx8x8888mmx (pixman_op_t op,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    ullong m = *mask;
+	    uint64_t m = *mask;
 
 	    if (m)
 	    {
@@ -1643,13 +1628,13 @@ fbCompositeSolidMask_nx8x8888mmx (pixman_op_t op,
 
 	while (w >= 2)
 	{
-	    ullong m0, m1;
+	    uint64_t m0, m1;
 	    m0 = *mask;
 	    m1 = *(mask + 1);
 
 	    if (srca == 0xff && (m0 & m1) == 0xff)
 	    {
-		*(ullong *)dst = srcsrc;
+		*(uint64_t *)dst = srcsrc;
 	    }
 	    else if (m0 | m1)
 	    {
@@ -1673,7 +1658,7 @@ fbCompositeSolidMask_nx8x8888mmx (pixman_op_t op,
 
 	while (w)
 	{
-	    ullong m = *mask;
+	    uint64_t m = *mask;
 
 	    if (m)
 	    {
@@ -1701,7 +1686,7 @@ pixman_fill_mmx (uint32_t *bits,
 		 int height,
 		 uint32_t xor)
 {
-    ullong	fill;
+    uint64_t	fill;
     __m64	vfill;
     uint32_t	byte_width;
     uint8_t	*byte_line;
@@ -1744,7 +1729,7 @@ pixman_fill_mmx (uint32_t *bits,
 	stride *= 4;
     }
 
-    fill = ((ullong)xor << 32) | xor;
+    fill = ((uint64_t)xor << 32) | xor;
     vfill = M64(fill);
 
 #ifdef __GNUC__
@@ -1867,7 +1852,7 @@ fbCompositeSolidMaskSrc_nx8x8888mmx (pixman_op_t op,
     int	dstStride, maskStride;
     uint16_t	w;
     __m64	vsrc, vsrca;
-    ullong	srcsrc;
+    uint64_t	srcsrc;
 
     CHECKPOINT();
 
@@ -1881,7 +1866,7 @@ fbCompositeSolidMaskSrc_nx8x8888mmx (pixman_op_t op,
 	return;
     }
 
-    srcsrc = (ullong)src << 32 | src;
+    srcsrc = (uint64_t)src << 32 | src;
 
     fbComposeGetStart (pDst, xDst, yDst, uint32_t, dstStride, dstLine, 1);
     fbComposeGetStart (pMask, xMask, yMask, uint8_t, maskStride, maskLine, 1);
@@ -1901,7 +1886,7 @@ fbCompositeSolidMaskSrc_nx8x8888mmx (pixman_op_t op,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    ullong m = *mask;
+	    uint64_t m = *mask;
 
 	    if (m)
 	    {
@@ -1922,13 +1907,13 @@ fbCompositeSolidMaskSrc_nx8x8888mmx (pixman_op_t op,
 
 	while (w >= 2)
 	{
-	    ullong m0, m1;
+	    uint64_t m0, m1;
 	    m0 = *mask;
 	    m1 = *(mask + 1);
 
 	    if (srca == 0xff && (m0 & m1) == 0xff)
 	    {
-		*(ullong *)dst = srcsrc;
+		*(uint64_t *)dst = srcsrc;
 	    }
 	    else if (m0 | m1)
 	    {
@@ -1944,7 +1929,7 @@ fbCompositeSolidMaskSrc_nx8x8888mmx (pixman_op_t op,
 	    }
 	    else
 	    {
-		*(ullong *)dst = 0;
+		*(uint64_t *)dst = 0;
 	    }
 
 	    mask += 2;
@@ -1956,7 +1941,7 @@ fbCompositeSolidMaskSrc_nx8x8888mmx (pixman_op_t op,
 
 	while (w)
 	{
-	    ullong m = *mask;
+	    uint64_t m = *mask;
 
 	    if (m)
 	    {
@@ -1998,7 +1983,7 @@ fbCompositeSolidMask_nx8x0565mmx (pixman_op_t op,
     int	dstStride, maskStride;
     uint16_t	w;
     __m64	vsrc, vsrca, tmp;
-    ullong srcsrcsrcsrc, src16;
+    uint64_t srcsrcsrcsrc, src16;
 
     CHECKPOINT();
 
@@ -2015,10 +2000,10 @@ fbCompositeSolidMask_nx8x0565mmx (pixman_op_t op,
     vsrca = expand_alpha (vsrc);
 
     tmp = pack565(vsrc, _mm_setzero_si64(), 0);
-    src16 = ULLONG(tmp);
+    src16 = UINT64(tmp);
 
-    srcsrcsrcsrc = (ullong)src16 << 48 | (ullong)src16 << 32 |
-	(ullong)src16 << 16 | (ullong)src16;
+    srcsrcsrcsrc = (uint64_t)src16 << 48 | (uint64_t)src16 << 32 |
+	(uint64_t)src16 << 16 | (uint64_t)src16;
 
     while (height--)
     {
@@ -2032,15 +2017,15 @@ fbCompositeSolidMask_nx8x0565mmx (pixman_op_t op,
 
 	while (w && (unsigned long)dst & 7)
 	{
-	    ullong m = *mask;
+	    uint64_t m = *mask;
 
 	    if (m)
 	    {
-		ullong d = *dst;
+		uint64_t d = *dst;
 		__m64 vd = M64(d);
 		__m64 vdest = in_over(vsrc, vsrca, expand_alpha_rev (M64 (m)), expand565(vd, 0));
 		vd = pack565(vdest, _mm_setzero_si64(), 0);
-		*dst = ULLONG(vd);
+		*dst = UINT64(vd);
 	    }
 
 	    w--;
@@ -2052,7 +2037,7 @@ fbCompositeSolidMask_nx8x0565mmx (pixman_op_t op,
 
 	while (w >= 4)
 	{
-	    ullong m0, m1, m2, m3;
+	    uint64_t m0, m1, m2, m3;
 	    m0 = *mask;
 	    m1 = *(mask + 1);
 	    m2 = *(mask + 2);
@@ -2060,7 +2045,7 @@ fbCompositeSolidMask_nx8x0565mmx (pixman_op_t op,
 
 	    if (srca == 0xff && (m0 & m1 & m2 & m3) == 0xff)
 	    {
-		*(ullong *)dst = srcsrcsrcsrc;
+		*(uint64_t *)dst = srcsrcsrcsrc;
 	    }
 	    else if (m0 | m1 | m2 | m3)
 	    {
@@ -2090,15 +2075,15 @@ fbCompositeSolidMask_nx8x0565mmx (pixman_op_t op,
 
 	while (w)
 	{
-	    ullong m = *mask;
+	    uint64_t m = *mask;
 
 	    if (m)
 	    {
-		ullong d = *dst;
+		uint64_t d = *dst;
 		__m64 vd = M64(d);
 		__m64 vdest = in_over(vsrc, vsrca, expand_alpha_rev (M64(m)), expand565(vd, 0));
 		vd = pack565(vdest, _mm_setzero_si64(), 0);
-		*dst = ULLONG(vd);
+		*dst = UINT64(vd);
 	    }
 
 	    w--;
@@ -2152,12 +2137,12 @@ fbCompositeSrc_8888RevNPx0565mmx (pixman_op_t op,
 	while (w && (unsigned long)dst & 7)
 	{
 	    __m64 vsrc = load8888 (*src);
-	    ullong d = *dst;
+	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (M64(d), 0);
 
 	    vdest = pack565(over_rev_non_pre(vsrc, vdest), vdest, 0);
 
-	    *dst = ULLONG(vdest);
+	    *dst = UINT64(vdest);
 
 	    w--;
 	    dst++;
@@ -2213,12 +2198,12 @@ fbCompositeSrc_8888RevNPx0565mmx (pixman_op_t op,
 	while (w)
 	{
 	    __m64 vsrc = load8888 (*src);
-	    ullong d = *dst;
+	    uint64_t d = *dst;
 	    __m64 vdest = expand565 (M64(d), 0);
 
 	    vdest = pack565(over_rev_non_pre(vsrc, vdest), vdest, 0);
 
-	    *dst = ULLONG(vdest);
+	    *dst = UINT64(vdest);
 
 	    w--;
 	    dst++;
@@ -2282,7 +2267,7 @@ fbCompositeSrc_8888RevNPx8888mmx (pixman_op_t op,
 
 	while (w >= 2)
 	{
-	    ullong s0, s1;
+	    uint64_t s0, s1;
 	    unsigned char a0, a1;
 	    __m64 d0, d1;
 
@@ -2376,10 +2361,10 @@ fbCompositeSolidMask_nx8888x0565Cmmx (pixman_op_t op,
 
 	    if (m)
 	    {
-		ullong d = *q;
+		uint64_t d = *q;
 		__m64 vdest = expand565 (M64(d), 0);
 		vdest = pack565 (in_over (vsrc, vsrca, load8888 (m), vdest), vdest, 0);
-		*q = ULLONG(vdest);
+		*q = UINT64(vdest);
 	    }
 
 	    twidth--;
@@ -2419,10 +2404,10 @@ fbCompositeSolidMask_nx8888x0565Cmmx (pixman_op_t op,
 	    m = *(uint32_t *)p;
 	    if (m)
 	    {
-		ullong d = *q;
+		uint64_t d = *q;
 		__m64 vdest = expand565(M64(d), 0);
 		vdest = pack565 (in_over(vsrc, vsrca, load8888(m), vdest), vdest, 0);
-		*q = ULLONG(vdest);
+		*q = UINT64(vdest);
 	    }
 
 	    twidth--;
@@ -2780,7 +2765,7 @@ fbCompositeSrcAdd_8888x8888mmx (pixman_op_t 	op,
 	while (w >= 2)
 	{
 	    dst64 = _mm_adds_pu8(*(__m64*)src, *(__m64*)dst);
-	    *(ullong*)dst = ULLONG(dst64);
+	    *(uint64_t*)dst = UINT64(dst64);
 	    dst += 2;
 	    src += 2;
 	    w -= 2;
@@ -2864,7 +2849,7 @@ pixman_blt_mmx (uint32_t *src_bits,
 
 	while (w >= 64)
 	{
-#ifdef __GNUC__
+#if defined (__GNUC__) || (defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590))
 	    __asm__ (
 		"movq	  (%1),	  %%mm0\n"
 		"movq	 8(%1),	  %%mm1\n"
@@ -2993,7 +2978,7 @@ fbCompositeOver_x888x8x8888mmx (pixman_op_t      op,
 
 	while (w--)
 	{
-	    ullong m = *mask;
+	    uint64_t m = *mask;
 
 	    if (m)
 	    {
diff --git a/pixman/pixman-pict.c b/pixman/pixman-pict.c
index 13a70454..33eb8b08 100644
--- a/pixman/pixman-pict.c
+++ b/pixman/pixman-pict.c
@@ -1969,6 +1969,8 @@ pixman_image_composite (pixman_op_t      op,
             pSrc->common.transform->matrix[2][1] == 0 &&
             pSrc->common.transform->matrix[2][2] == pixman_fixed_1) {
             func = fbCompositeSrcScaleNearest;
+            /* Like the general path, fbCompositeSrcScaleNearest handles all the repeat types itself */
+            srcRepeat = FALSE;
         }
     } else if ((pSrc->type == BITS || pixman_image_can_get_solid (pSrc)) && (!pMask || pMask->type == BITS)
         && !srcTransform && !maskTransform
diff --git a/pixman/pixman-private.h b/pixman/pixman-private.h
index 1380fa3b..debd7234 100644
--- a/pixman/pixman-private.h
+++ b/pixman/pixman-private.h
@@ -70,7 +70,7 @@
 /* 'inline' is available only in C++ in MSVC */
 #   define inline __inline
 #   define force_inline __forceinline
-#elif defined __GNUC__
+#elif defined __GNUC__ || (defined(__SUNPRO_C) && (__SUNPRO_C >= 0x590))
 #   define inline __inline__
 #   define force_inline __inline__ __attribute__ ((__always_inline__))
 #else
diff --git a/pixman/pixman-region.c b/pixman/pixman-region.c
index 01a28be7..dec2c9de 100644
--- a/pixman/pixman-region.c
+++ b/pixman/pixman-region.c
@@ -572,7 +572,7 @@ pixman_region_appendNonO (
 {									\
     int newRects;							\
     if ((newRects = rEnd - r)) {					\
-	RECTALLOC(newReg, newRects);					\
+	RECTALLOC_BAIL(newReg, newRects, bail);					\
 	memmove((char *)PIXREGION_TOP(newReg),(char *)r, 			\
               newRects * sizeof(box_type_t));				\
 	newReg->data->numRects += newRects;				\
@@ -752,7 +752,8 @@ pixman_op(
 		bot = MIN(r1->y2, r2y1);
 		if (top != bot)	{
 		    curBand = newReg->data->numRects;
-		    pixman_region_appendNonO(newReg, r1, r1BandEnd, top, bot);
+		    if (!pixman_region_appendNonO(newReg, r1, r1BandEnd, top, bot))
+			goto bail;
 		    Coalesce(newReg, prevBand, curBand);
 		}
 	    }
@@ -763,7 +764,8 @@ pixman_op(
 		bot = MIN(r2->y2, r1y1);
 		if (top != bot) {
 		    curBand = newReg->data->numRects;
-		    pixman_region_appendNonO(newReg, r2, r2BandEnd, top, bot);
+		    if (!pixman_region_appendNonO(newReg, r2, r2BandEnd, top, bot))
+			goto bail;
 		    Coalesce(newReg, prevBand, curBand);
 		}
 	    }
@@ -779,8 +781,12 @@ pixman_op(
 	ybot = MIN(r1->y2, r2->y2);
 	if (ybot > ytop) {
 	    curBand = newReg->data->numRects;
-	    (* overlapFunc)(newReg, r1, r1BandEnd, r2, r2BandEnd, ytop, ybot,
-			    pOverlap);
+	    if (!(* overlapFunc)(newReg,
+				 r1, r1BandEnd,
+				 r2, r2BandEnd,
+				 ytop, ybot,
+				 pOverlap))
+		goto bail;
 	    Coalesce(newReg, prevBand, curBand);
 	}
 
@@ -805,7 +811,10 @@ pixman_op(
 	/* Do first nonOverlap1Func call, which may be able to coalesce */
 	FindBand(r1, r1BandEnd, r1End, r1y1);
 	curBand = newReg->data->numRects;
-	pixman_region_appendNonO(newReg, r1, r1BandEnd, MAX(r1y1, ybot), r1->y2);
+	if (!pixman_region_appendNonO(newReg,
+				      r1, r1BandEnd,
+				      MAX(r1y1, ybot), r1->y2))
+	    goto bail;
 	Coalesce(newReg, prevBand, curBand);
 	/* Just append the rest of the boxes  */
 	AppendRegions(newReg, r1BandEnd, r1End);
@@ -814,7 +823,10 @@ pixman_op(
 	/* Do first nonOverlap2Func call, which may be able to coalesce */
 	FindBand(r2, r2BandEnd, r2End, r2y1);
 	curBand = newReg->data->numRects;
-	pixman_region_appendNonO(newReg, r2, r2BandEnd, MAX(r2y1, ybot), r2->y2);
+	if (!pixman_region_appendNonO(newReg,
+				      r2, r2BandEnd,
+				      MAX(r2y1, ybot), r2->y2))
+	    goto bail;
 	Coalesce(newReg, prevBand, curBand);
 	/* Append rest of boxes */
 	AppendRegions(newReg, r2BandEnd, r2End);
@@ -840,6 +852,11 @@ pixman_op(
     }
 
     return TRUE;
+
+bail:
+    if (oldData)
+	free(oldData);
+    return pixman_break (newReg);
 }
 
 /*-
diff --git a/pixman/pixman-utils.c b/pixman/pixman-utils.c
index 6e04df62..8734dc53 100644
--- a/pixman/pixman-utils.c
+++ b/pixman/pixman-utils.c
@@ -461,6 +461,8 @@ pixman_format_supported_destination (pixman_format_code_t format)
     case PIXMAN_x8r8g8b8:
     case PIXMAN_a8b8g8r8:
     case PIXMAN_x8b8g8r8:
+    case PIXMAN_b8g8r8a8:
+    case PIXMAN_b8g8r8x8:
     case PIXMAN_r8g8b8:
     case PIXMAN_b8g8r8:
     case PIXMAN_r5g6b5:
@@ -531,6 +533,8 @@ pixman_format_supported_source (pixman_format_code_t format)
     case PIXMAN_x8r8g8b8:
     case PIXMAN_a8b8g8r8:
     case PIXMAN_x8b8g8r8:
+    case PIXMAN_b8g8r8a8:
+    case PIXMAN_b8g8r8x8:
     case PIXMAN_r8g8b8:
     case PIXMAN_b8g8r8:
     case PIXMAN_r5g6b5:
diff --git a/pixman/pixman.h b/pixman/pixman.h
index 42c397ce..96839f87 100644
--- a/pixman/pixman.h
+++ b/pixman/pixman.h
@@ -648,10 +648,12 @@ struct pixman_indexed
 #define PIXMAN_TYPE_GRAY	5
 #define PIXMAN_TYPE_YUY2	6
 #define PIXMAN_TYPE_YV12	7
+#define PIXMAN_TYPE_BGRA	8
 
 #define PIXMAN_FORMAT_COLOR(f)				\
 	(PIXMAN_FORMAT_TYPE(f) == PIXMAN_TYPE_ARGB ||	\
-	 PIXMAN_FORMAT_TYPE(f) == PIXMAN_TYPE_ABGR)
+	 PIXMAN_FORMAT_TYPE(f) == PIXMAN_TYPE_ABGR ||	\
+	 PIXMAN_FORMAT_TYPE(f) == PIXMAN_TYPE_BGRA)
 
 /* 32bpp formats */
 typedef enum {
@@ -659,6 +661,8 @@ typedef enum {
     PIXMAN_x8r8g8b8 =	PIXMAN_FORMAT(32,PIXMAN_TYPE_ARGB,0,8,8,8),
     PIXMAN_a8b8g8r8 =	PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,8,8,8,8),
     PIXMAN_x8b8g8r8 =	PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,0,8,8,8),
+    PIXMAN_b8g8r8a8 =	PIXMAN_FORMAT(32,PIXMAN_TYPE_BGRA,8,8,8,8),
+    PIXMAN_b8g8r8x8 =	PIXMAN_FORMAT(32,PIXMAN_TYPE_BGRA,0,8,8,8),
     PIXMAN_x2b10g10r10 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,0,10,10,10),
     PIXMAN_a2b10g10r10 = PIXMAN_FORMAT(32,PIXMAN_TYPE_ABGR,2,10,10,10),
 
diff --git a/pixman/solaris-hwcap.mapfile b/pixman/solaris-hwcap.mapfile
new file mode 100644
index 00000000..7f439a95
--- /dev/null
+++ b/pixman/solaris-hwcap.mapfile
@@ -0,0 +1,36 @@
+###############################################################################
+#
+# Copyright 2009 Sun Microsystems, Inc.  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 the rights to use, copy, modify, merge, publish,
+# distribute, and/or sell copies of the Software, and to permit persons
+# to whom the Software is furnished to do so, provided that the above
+# copyright notice(s) and this permission notice appear in all copies of
+# the Software and that both the above copyright notice(s) and this
+# permission notice appear in supporting documentation.
+#
+# 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
+# OF THIRD PARTY RIGHTS. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
+# HOLDERS INCLUDED IN THIS NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL
+# INDIRECT OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING
+# FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT,
+# NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION
+# WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+#
+# Except as contained in this notice, the name of a copyright holder
+# shall not be used in advertising or otherwise to promote the sale, use
+# or other dealings in this Software without prior written authorization
+# of the copyright holder.
+#
+###############################################################################
+#
+# Override the linker's detection of CMOV/MMX/SSE instructions so this
+# library isn't flagged as only usable on CPU's with those ISA's, since it
+# checks at runtime for availability before calling them
+
+hwcap_1 = V0x0 FPU OVERRIDE;
diff --git a/test/Makefile.am b/test/Makefile.am
index 33e2200d..cccb6f80 100644
--- a/test/Makefile.am
+++ b/test/Makefile.am
@@ -1,19 +1,35 @@
-if HAVE_GTK
+TEST_LDADD = $(top_builddir)/pixman/libpixman-1.la
+INCLUDES = -I$(top_srcdir)/pixman -I$(top_builddir)/pixman 
+
 TESTPROGRAMS =			\
+	region-test		\
+	scaling-test		\
+	fetch-test
+
+fetch_test_LDADD = $(TEST_LDADD)
+region_test_LDADD = $(TEST_LDADD)
+scaling_test_LDADD = $(TEST_LDADD)
+
+# GTK using test programs
+
+if HAVE_GTK
+
+GTK_LDADD = $(TEST_LDADD) $(GTK_LIBS)
+
+TESTPROGRAMS +=			\
 	clip-test		\
 	composite-test		\
 	gradient-test		\
-	region-test		\
-	fetch-test
+	trap-test
 
 noinst_PROGRAMS = $(TESTPROGRAMS)
 
-INCLUDES = -I$(top_srcdir)/pixman -I$(top_builddir)/pixman $(GTK_CFLAGS)
+INCLUDES += $(GTK_CFLAGS)
 
-composite_test_LDADD =	$(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
-gradient_test_LDADD = $(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
-fetch_test_LDADD = $(top_builddir)/pixman/libpixman-1.la
-region_test_LDADD = $(top_builddir)/pixman/libpixman-1.la
-clip_test_LDADD = $(top_builddir)/pixman/libpixman-1.la $(GTK_LIBS)
+composite_test_LDADD = $(GTK_LDADD)
+gradient_test_LDADD = $(GTK_LDADD)
+clip_test_LDADD = $(GTK_LDADD)
+trap_test_LDADD = $(GTK_LDADD)
 
 endif
+
diff --git a/test/clip-test.c b/test/clip-test.c
index 457e97a2..4995083c 100644
--- a/test/clip-test.c
+++ b/test/clip-test.c
@@ -20,14 +20,27 @@ pixbuf_from_argb32 (uint32_t *bits,
 	for (w = 0; w < width; ++w)
 	{
 	    uint32_t argb = bits[h * stride + w];
-	    guint32 abgr;
-	    
-	    abgr = (argb & 0xff000000) |
-		(argb & 0xff) << 16 |
-		(argb & 0x00ff00) |
-		(argb & 0xff0000) >> 16;
-	    
-	    p_bits[h * (p_stride / 4) + w] = abgr;
+	    guint r, g, b, a;
+	    char *pb = p_bits;
+
+	    pb += h * p_stride + w * 4;
+
+	    r = (argb & 0x00ff0000) >> 16;
+	    g = (argb & 0x0000ff00) >> 8;
+	    b = (argb & 0x000000ff) >> 0;
+	    a = (argb & 0xff000000) >> 24;
+
+	    if (a)
+	    {
+		r = (r * 255) / a;
+		g = (g * 255) / a;
+		b = (b * 255) / a;
+	    }
+
+	    pb[0] = r;
+	    pb[1] = g;
+	    pb[2] = b;
+	    pb[3] = a;
 	}
     }
     
diff --git a/test/composite-test.c b/test/composite-test.c
index d6596f49..388c8e64 100644
--- a/test/composite-test.c
+++ b/test/composite-test.c
@@ -1,10 +1,9 @@
 #include <stdlib.h>
 #include <stdio.h>
 #include "pixman.h"
-
 #include <gtk/gtk.h>
 
-static GdkPixbuf *
+GdkPixbuf *
 pixbuf_from_argb32 (uint32_t *bits,
 		    int width,
 		    int height,
@@ -15,20 +14,36 @@ pixbuf_from_argb32 (uint32_t *bits,
     int p_stride = gdk_pixbuf_get_rowstride (pixbuf);
     guint32 *p_bits = (guint32 *)gdk_pixbuf_get_pixels (pixbuf);
     int w, h;
-
+    
     for (h = 0; h < height; ++h)
     {
 	for (w = 0; w < width; ++w)
 	{
 	    uint32_t argb = bits[h * stride + w];
-	    guint32 rgba;
-
-	    rgba = (argb << 8) | (argb >> 24);
-
-	    p_bits[h * (p_stride / 4) + w] = rgba;
+	    guint r, g, b, a;
+	    char *pb = p_bits;
+
+	    pb += h * p_stride + w * 4;
+
+	    r = (argb & 0x00ff0000) >> 16;
+	    g = (argb & 0x0000ff00) >> 8;
+	    b = (argb & 0x000000ff) >> 0;
+	    a = (argb & 0xff000000) >> 24;
+
+	    if (a)
+	    {
+		r = (r * 255) / a;
+		g = (g * 255) / a;
+		b = (b * 255) / a;
+	    }
+
+	    pb[0] = r;
+	    pb[1] = g;
+	    pb[2] = b;
+	    pb[3] = a;
 	}
     }
-
+    
     return pixbuf;
 }
 
diff --git a/test/gradient-test.c b/test/gradient-test.c
index 8a99ff0e..806256bb 100644
--- a/test/gradient-test.c
+++ b/test/gradient-test.c
@@ -20,14 +20,27 @@ pixbuf_from_argb32 (uint32_t *bits,
 	for (w = 0; w < width; ++w)
 	{
 	    uint32_t argb = bits[h * stride + w];
-	    guint32 abgr;
-	    
-	    abgr = (argb & 0xff000000) |
-		(argb & 0xff) << 16 |
-		(argb & 0x00ff00) |
-		(argb & 0xff0000) >> 16;
-	    
-	    p_bits[h * (p_stride / 4) + w] = abgr;
+	    guint r, g, b, a;
+	    char *pb = p_bits;
+
+	    pb += h * p_stride + w * 4;
+
+	    r = (argb & 0x00ff0000) >> 16;
+	    g = (argb & 0x0000ff00) >> 8;
+	    b = (argb & 0x000000ff) >> 0;
+	    a = (argb & 0xff000000) >> 24;
+
+	    if (a)
+	    {
+		r = (r * 255) / a;
+		g = (g * 255) / a;
+		b = (b * 255) / a;
+	    }
+
+	    pb[0] = r;
+	    pb[1] = g;
+	    pb[2] = b;
+	    pb[3] = a;
 	}
     }
     
diff --git a/test/scaling-test-bisect.rb b/test/scaling-test-bisect.rb
new file mode 100644
index 00000000..2e4aff30
--- /dev/null
+++ b/test/scaling-test-bisect.rb
@@ -0,0 +1,38 @@
+#!/usr/bin/env ruby
+
+if not ARGV[0] or not ARGV[1] then
+    printf("Please provide two 'scaling-test' static binaries in the command line.\n\n")
+    printf("The first should be linked with the correct reference pixman library.\n")
+    printf("The second binrary should be linked with the pixman library to be tested.\n")
+    exit(0)
+end
+
+$MAX = 10000
+$MIN = 1
+$AVG = 0
+
+if `#{ARGV[0]} #{$MAX} 2>/dev/null` == `#{ARGV[1]} #{$MAX} 2>/dev/null` then
+    printf("test ok\n")
+    exit(0)
+end
+
+printf("test failed, bisecting...\n")
+
+while $MAX != $MIN + 1 do
+    $AVG = (($MIN + $MAX) / 2).to_i
+    res1 = `#{ARGV[0]} #{$AVG} 2>/dev/null`
+    res2 = `#{ARGV[1]} #{$AVG} 2>/dev/null`
+    if res1 != res2 then
+        $MAX = $AVG
+    else
+        $MIN = $AVG
+    end
+end
+
+printf("-- ref --\n")
+printf("%s\n", `#{ARGV[0]} -#{$MAX}`)
+printf("-- new --\n")
+printf("%s\n", `#{ARGV[1]} -#{$MAX}`)
+
+printf("\nFailed test number is %d, you can reproduce the problematic conditions\n", $MAX)
+printf("by running 'scaling-test -%d'\n", $MAX)
diff --git a/test/scaling-test.c b/test/scaling-test.c
new file mode 100644
index 00000000..ff24b1d4
--- /dev/null
+++ b/test/scaling-test.c
@@ -0,0 +1,261 @@
+#include <assert.h>
+#include <stdlib.h>
+#include <stdio.h>
+#include "pixman.h"
+
+/* A primitive pseudorandom number generator, taken from POSIX.1-2001 example */
+
+static uint32_t lcg_seed;
+
+uint32_t lcg_rand(void)
+{
+    lcg_seed = lcg_seed * 1103515245 + 12345;
+    return ((uint32_t)(lcg_seed / 65536) % 32768);
+}
+
+void lcg_srand(uint32_t seed)
+{
+    lcg_seed = seed;
+}
+
+uint32_t lcg_rand_n(int max)
+{
+    return lcg_rand() % max;
+}
+
+/*----------------------------------------------------------------------------*\
+ *  CRC-32 version 2.0.0 by Craig Bruce, 2006-04-29.
+ *
+ *  This program generates the CRC-32 values for the files named in the
+ *  command-line arguments.  These are the same CRC-32 values used by GZIP,
+ *  PKZIP, and ZMODEM.  The Crc32_ComputeBuf() can also be detached and
+ *  used independently.
+ *
+ *  THIS PROGRAM IS PUBLIC-DOMAIN SOFTWARE.
+ *
+ *  Based on the byte-oriented implementation "File Verification Using CRC"
+ *  by Mark R. Nelson in Dr. Dobb's Journal, May 1992, pp. 64-67.
+ *
+ *  v1.0.0: original release.
+ *  v1.0.1: fixed printf formats.
+ *  v1.0.2: fixed something else.
+ *  v1.0.3: replaced CRC constant table by generator function.
+ *  v1.0.4: reformatted code, made ANSI C.  1994-12-05.
+ *  v2.0.0: rewrote to use memory buffer & static table, 2006-04-29.
+\*----------------------------------------------------------------------------*/
+
+/*----------------------------------------------------------------------------*\
+ *  NAME:
+ *     Crc32_ComputeBuf() - computes the CRC-32 value of a memory buffer
+ *  DESCRIPTION:
+ *     Computes or accumulates the CRC-32 value for a memory buffer.
+ *     The 'inCrc32' gives a previously accumulated CRC-32 value to allow
+ *     a CRC to be generated for multiple sequential buffer-fuls of data.
+ *     The 'inCrc32' for the first buffer must be zero.
+ *  ARGUMENTS:
+ *     inCrc32 - accumulated CRC-32 value, must be 0 on first call
+ *     buf     - buffer to compute CRC-32 value for
+ *     bufLen  - number of bytes in buffer
+ *  RETURNS:
+ *     crc32 - computed CRC-32 value
+ *  ERRORS:
+ *     (no errors are possible)
+\*----------------------------------------------------------------------------*/
+
+static uint32_t Crc32_ComputeBuf( uint32_t inCrc32, const void *buf,
+                                       size_t bufLen )
+{
+    static const uint32_t crcTable[256] = {
+   0x00000000,0x77073096,0xEE0E612C,0x990951BA,0x076DC419,0x706AF48F,0xE963A535,
+   0x9E6495A3,0x0EDB8832,0x79DCB8A4,0xE0D5E91E,0x97D2D988,0x09B64C2B,0x7EB17CBD,
+   0xE7B82D07,0x90BF1D91,0x1DB71064,0x6AB020F2,0xF3B97148,0x84BE41DE,0x1ADAD47D,
+   0x6DDDE4EB,0xF4D4B551,0x83D385C7,0x136C9856,0x646BA8C0,0xFD62F97A,0x8A65C9EC,
+   0x14015C4F,0x63066CD9,0xFA0F3D63,0x8D080DF5,0x3B6E20C8,0x4C69105E,0xD56041E4,
+   0xA2677172,0x3C03E4D1,0x4B04D447,0xD20D85FD,0xA50AB56B,0x35B5A8FA,0x42B2986C,
+   0xDBBBC9D6,0xACBCF940,0x32D86CE3,0x45DF5C75,0xDCD60DCF,0xABD13D59,0x26D930AC,
+   0x51DE003A,0xC8D75180,0xBFD06116,0x21B4F4B5,0x56B3C423,0xCFBA9599,0xB8BDA50F,
+   0x2802B89E,0x5F058808,0xC60CD9B2,0xB10BE924,0x2F6F7C87,0x58684C11,0xC1611DAB,
+   0xB6662D3D,0x76DC4190,0x01DB7106,0x98D220BC,0xEFD5102A,0x71B18589,0x06B6B51F,
+   0x9FBFE4A5,0xE8B8D433,0x7807C9A2,0x0F00F934,0x9609A88E,0xE10E9818,0x7F6A0DBB,
+   0x086D3D2D,0x91646C97,0xE6635C01,0x6B6B51F4,0x1C6C6162,0x856530D8,0xF262004E,
+   0x6C0695ED,0x1B01A57B,0x8208F4C1,0xF50FC457,0x65B0D9C6,0x12B7E950,0x8BBEB8EA,
+   0xFCB9887C,0x62DD1DDF,0x15DA2D49,0x8CD37CF3,0xFBD44C65,0x4DB26158,0x3AB551CE,
+   0xA3BC0074,0xD4BB30E2,0x4ADFA541,0x3DD895D7,0xA4D1C46D,0xD3D6F4FB,0x4369E96A,
+   0x346ED9FC,0xAD678846,0xDA60B8D0,0x44042D73,0x33031DE5,0xAA0A4C5F,0xDD0D7CC9,
+   0x5005713C,0x270241AA,0xBE0B1010,0xC90C2086,0x5768B525,0x206F85B3,0xB966D409,
+   0xCE61E49F,0x5EDEF90E,0x29D9C998,0xB0D09822,0xC7D7A8B4,0x59B33D17,0x2EB40D81,
+   0xB7BD5C3B,0xC0BA6CAD,0xEDB88320,0x9ABFB3B6,0x03B6E20C,0x74B1D29A,0xEAD54739,
+   0x9DD277AF,0x04DB2615,0x73DC1683,0xE3630B12,0x94643B84,0x0D6D6A3E,0x7A6A5AA8,
+   0xE40ECF0B,0x9309FF9D,0x0A00AE27,0x7D079EB1,0xF00F9344,0x8708A3D2,0x1E01F268,
+   0x6906C2FE,0xF762575D,0x806567CB,0x196C3671,0x6E6B06E7,0xFED41B76,0x89D32BE0,
+   0x10DA7A5A,0x67DD4ACC,0xF9B9DF6F,0x8EBEEFF9,0x17B7BE43,0x60B08ED5,0xD6D6A3E8,
+   0xA1D1937E,0x38D8C2C4,0x4FDFF252,0xD1BB67F1,0xA6BC5767,0x3FB506DD,0x48B2364B,
+   0xD80D2BDA,0xAF0A1B4C,0x36034AF6,0x41047A60,0xDF60EFC3,0xA867DF55,0x316E8EEF,
+   0x4669BE79,0xCB61B38C,0xBC66831A,0x256FD2A0,0x5268E236,0xCC0C7795,0xBB0B4703,
+   0x220216B9,0x5505262F,0xC5BA3BBE,0xB2BD0B28,0x2BB45A92,0x5CB36A04,0xC2D7FFA7,
+   0xB5D0CF31,0x2CD99E8B,0x5BDEAE1D,0x9B64C2B0,0xEC63F226,0x756AA39C,0x026D930A,
+   0x9C0906A9,0xEB0E363F,0x72076785,0x05005713,0x95BF4A82,0xE2B87A14,0x7BB12BAE,
+   0x0CB61B38,0x92D28E9B,0xE5D5BE0D,0x7CDCEFB7,0x0BDBDF21,0x86D3D2D4,0xF1D4E242,
+   0x68DDB3F8,0x1FDA836E,0x81BE16CD,0xF6B9265B,0x6FB077E1,0x18B74777,0x88085AE6,
+   0xFF0F6A70,0x66063BCA,0x11010B5C,0x8F659EFF,0xF862AE69,0x616BFFD3,0x166CCF45,
+   0xA00AE278,0xD70DD2EE,0x4E048354,0x3903B3C2,0xA7672661,0xD06016F7,0x4969474D,
+   0x3E6E77DB,0xAED16A4A,0xD9D65ADC,0x40DF0B66,0x37D83BF0,0xA9BCAE53,0xDEBB9EC5,
+   0x47B2CF7F,0x30B5FFE9,0xBDBDF21C,0xCABAC28A,0x53B39330,0x24B4A3A6,0xBAD03605,
+   0xCDD70693,0x54DE5729,0x23D967BF,0xB3667A2E,0xC4614AB8,0x5D681B02,0x2A6F2B94,
+   0xB40BBE37,0xC30C8EA1,0x5A05DF1B,0x2D02EF8D };
+    uint32_t crc32;
+    unsigned char *byteBuf;
+    size_t i;
+
+    /** accumulate crc32 for buffer **/
+    crc32 = inCrc32 ^ 0xFFFFFFFF;
+    byteBuf = (unsigned char*) buf;
+    for (i=0; i < bufLen; i++) {
+        crc32 = (crc32 >> 8) ^ crcTable[ (crc32 ^ byteBuf[i]) & 0xFF ];
+    }
+    return( crc32 ^ 0xFFFFFFFF );
+}
+
+
+#define MAX_SRC_WIDTH  100
+#define MAX_SRC_HEIGHT 100
+#define MAX_DST_WIDTH  100
+#define MAX_DST_HEIGHT 100
+#define MAX_STRIDE     16
+
+/*
+ * Composite operation with pseudorandom images
+ */
+uint32_t test_composite(uint32_t initcrc, int testnum, int verbose)
+{
+    int i;
+    pixman_image_t *src_img;
+    pixman_image_t *dst_img;
+    pixman_transform_t transform;
+    pixman_region16_t clip;
+    int src_width, src_height;
+    int dst_width, dst_height;
+    int src_stride, dst_stride;
+    int src_x, src_y;
+    int dst_x, dst_y;
+    int src_bpp;
+    int dst_bpp;
+    int w, h;
+    int scale_x = 32768, scale_y = 32768;
+    int op;
+    int repeat = 0;
+    int src_fmt, dst_fmt;
+    uint32_t *srcbuf;
+    uint32_t *dstbuf;
+    uint32_t crc32;
+
+    lcg_srand(testnum);
+
+    src_bpp = /*(lcg_rand_n(2) == 0) ? 2 :*/ 4;
+    dst_bpp = /*(lcg_rand_n(2) == 0) ? 2 :*/ 4;
+    op = (lcg_rand_n(2) == 0) ? PIXMAN_OP_SRC : PIXMAN_OP_OVER;
+
+    src_width = lcg_rand_n(MAX_SRC_WIDTH) + 1;
+    src_height = lcg_rand_n(MAX_SRC_HEIGHT) + 1;
+    dst_width = lcg_rand_n(MAX_DST_WIDTH) + 1;
+    dst_height = lcg_rand_n(MAX_DST_HEIGHT) + 1;
+    src_stride = src_width * src_bpp + lcg_rand_n(MAX_STRIDE) * src_bpp;
+    dst_stride = dst_width * dst_bpp + lcg_rand_n(MAX_STRIDE) * dst_bpp;
+
+    src_x = -src_width + lcg_rand_n(src_width * 3);
+    src_y = -src_height + lcg_rand_n(src_height * 3);
+    dst_x = -dst_width + lcg_rand_n(dst_width * 2);
+    dst_y = -dst_height + lcg_rand_n(dst_height * 3);
+
+    srcbuf = (uint32_t *)malloc(src_stride * src_height);
+    dstbuf = (uint32_t *)malloc(dst_stride * dst_height);
+    for (i = 0; i < src_stride * src_height; i++)
+        *((uint8_t *)srcbuf + i) = lcg_rand_n(256);
+    for (i = 0; i < dst_stride * dst_height; i++)
+        *((uint8_t *)dstbuf + i) = lcg_rand_n(256);
+
+    src_fmt = src_bpp == 4 ? (lcg_rand_n(2) == 0 ?
+        PIXMAN_a8r8g8b8 : PIXMAN_x8r8g8b8) : PIXMAN_r5g6b5;
+
+    dst_fmt = dst_bpp == 4 ? (lcg_rand_n(2) == 0 ?
+        PIXMAN_a8r8g8b8 : PIXMAN_x8r8g8b8) : PIXMAN_r5g6b5;
+
+    src_img = pixman_image_create_bits(
+        src_fmt, src_width, src_height, srcbuf, src_stride);
+
+    dst_img = pixman_image_create_bits(
+        dst_fmt, dst_width, dst_height, dstbuf, dst_stride);
+
+    if (lcg_rand_n(2) == 0) {
+	scale_x = 32768 + lcg_rand_n(65536);
+	scale_y = 32768 + lcg_rand_n(65536);
+        pixman_transform_init_scale(&transform, scale_x, scale_y);
+        pixman_image_set_transform(src_img, &transform);
+    }
+
+    switch (lcg_rand_n(4)) {
+        case 0: repeat = PIXMAN_REPEAT_NONE; break;
+        case 1: repeat = PIXMAN_REPEAT_NORMAL; break;
+        case 2: repeat = PIXMAN_REPEAT_PAD; break;
+        case 3: repeat = PIXMAN_REPEAT_REFLECT; break;
+    }
+    pixman_image_set_repeat(src_img, repeat);
+
+    w = lcg_rand_n(MAX_DST_WIDTH * 2);
+    h = lcg_rand_n(MAX_DST_HEIGHT * 2);
+
+    if (verbose) {
+        printf("op=%d, scale_x=%d, scale_y=%d, repeat=%d\n",
+            op, scale_x, scale_y, repeat);
+        printf("src_width=%d, src_height=%d, dst_width=%d, dst_height=%d\n",
+            src_width, src_height, dst_width, dst_height);
+        printf("src_x=%d, src_y=%d, dst_x=%d, dst_y=%d\n",
+            src_x, src_y, dst_x, dst_y);
+        printf("w=%d, h=%d\n", w, h);
+    }
+
+    pixman_image_composite (op, src_img, NULL, dst_img,
+                            src_x, src_y, 0, 0, dst_x, dst_y, w, h);
+
+    if (verbose) {
+        int j;
+        for (i = 0; i < dst_height; i++) {
+            for (j = 0; j < dst_stride; j++) {
+                printf("%02X ", *((uint8_t *)dstbuf + i * dst_stride + j));
+            }
+            printf("\n");
+        }
+    }
+
+    pixman_image_unref (src_img);
+    pixman_image_unref (dst_img);
+
+    crc32 = Crc32_ComputeBuf(initcrc, dstbuf, dst_stride * dst_height);
+    free(srcbuf);
+    free(dstbuf);
+    return crc32;
+}
+
+int main(int argc, char *argv[])
+{
+    int i, n = 0;
+    uint32_t crc = 0;
+
+    if (argc >= 2)
+        n = atoi(argv[1]);
+
+    if (n == 0) n = 10000;
+
+    if (n < 0) {
+        crc = test_composite(0, -n, 1);
+        printf("crc32=%08X\n", crc);
+    }
+    else {
+        for (i = 1; i <= n; i++)
+        {
+            crc = test_composite(crc, i, 0);
+        }
+        printf("crc32=%08X\n", crc);
+    }
+    return 0;
+}
diff --git a/test/trap-test.c b/test/trap-test.c
new file mode 100644
index 00000000..7750e5a2
--- /dev/null
+++ b/test/trap-test.c
@@ -0,0 +1,126 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <gtk/gtk.h>
+#include "pixman.h"
+
+GdkPixbuf *
+pixbuf_from_argb32 (uint32_t *bits,
+		    int width,
+		    int height,
+		    int stride)
+{
+    GdkPixbuf *pixbuf = gdk_pixbuf_new (GDK_COLORSPACE_RGB, TRUE,
+					8, width, height);
+    int p_stride = gdk_pixbuf_get_rowstride (pixbuf);
+    guint32 *p_bits = (guint32 *)gdk_pixbuf_get_pixels (pixbuf);
+    int w, h;
+    
+    for (h = 0; h < height; ++h)
+    {
+	for (w = 0; w < width; ++w)
+	{
+	    uint32_t argb = bits[h * stride + w];
+	    guint r, g, b, a;
+	    char *pb = p_bits;
+
+	    pb += h * p_stride + w * 4;
+
+	    r = (argb & 0x00ff0000) >> 16;
+	    g = (argb & 0x0000ff00) >> 8;
+	    b = (argb & 0x000000ff) >> 0;
+	    a = (argb & 0xff000000) >> 24;
+
+	    if (a)
+	    {
+		r = (r * 255) / a;
+		g = (g * 255) / a;
+		b = (b * 255) / a;
+	    }
+
+	    pb[0] = r;
+	    pb[1] = g;
+	    pb[2] = b;
+	    pb[3] = a;
+	}
+    }
+    
+    return pixbuf;
+}
+
+static gboolean
+on_expose (GtkWidget *widget, GdkEventExpose *expose, gpointer data)
+{
+    GdkPixbuf *pixbuf = data;
+    
+    gdk_draw_pixbuf (widget->window, NULL,
+		     pixbuf, 0, 0, 0, 0,
+		     gdk_pixbuf_get_width (pixbuf),
+		     gdk_pixbuf_get_height (pixbuf),
+		     GDK_RGB_DITHER_NONE,
+		     0, 0);
+    
+    return TRUE;
+}
+
+static void
+show_window (uint32_t *bits, int w, int h, int stride)
+{
+    GdkPixbuf *pixbuf;
+    
+    GtkWidget *window = gtk_window_new (GTK_WINDOW_TOPLEVEL);
+    
+    pixbuf = pixbuf_from_argb32 (bits, w, h, stride);
+    
+    g_signal_connect (window, "expose_event", G_CALLBACK (on_expose), pixbuf);
+    g_signal_connect (window, "delete_event", G_CALLBACK (gtk_main_quit), NULL);
+    
+    gtk_widget_show (window);
+    
+    gtk_main ();
+}
+
+int
+main (int argc, char **argv)
+{
+#define WIDTH 200
+#define HEIGHT 200
+
+    pixman_image_t *src_img;
+    pixman_image_t *mask_img;
+    pixman_image_t *dest_img;
+    pixman_trap_t trap;
+    pixman_color_t white = { 0x0000, 0xffff, 0x0000, 0xffff };
+    uint32_t *bits = malloc (WIDTH * HEIGHT * 4);
+    uint32_t *mbits = malloc (WIDTH * HEIGHT);
+
+    memset (mbits, 0, WIDTH * HEIGHT);
+    memset (bits, 0xff, WIDTH * HEIGHT * 4);
+    
+    trap.top.l = pixman_int_to_fixed (50) + 0x8000;
+    trap.top.r = pixman_int_to_fixed (150) + 0x8000;
+    trap.top.y = pixman_int_to_fixed (30);
+
+    trap.bot.l = pixman_int_to_fixed (50) + 0x8000;
+    trap.bot.r = pixman_int_to_fixed (150) + 0x8000;
+    trap.bot.y = pixman_int_to_fixed (150);
+
+    mask_img = pixman_image_create_bits (PIXMAN_a8, WIDTH, HEIGHT, mbits, WIDTH);
+    src_img = pixman_image_create_solid_fill (&white);
+    dest_img = pixman_image_create_bits (PIXMAN_a8r8g8b8, WIDTH, HEIGHT, bits, WIDTH * 4);
+    
+    pixman_add_traps (mask_img, 0, 0, 1, &trap);
+
+    pixman_image_composite (PIXMAN_OP_OVER,
+			    src_img, mask_img, dest_img,
+			    0, 0, 0, 0, 0, 0, WIDTH, HEIGHT);
+    
+    gtk_init (&argc, &argv);
+    
+    show_window (bits, WIDTH, HEIGHT, WIDTH);
+    
+    pixman_image_unref (src_img);
+    pixman_image_unref (dest_img);
+    free (bits);
+    
+    return 0;
+}