tgsi: Add support for global / private / input MEMORY

Extend the MEMORY file support to differentiate between global, private
and shared memory, as well as "input" memory.

"MEMORY[x], INPUT" is intended to access OpenCL kernel parameters, a
special memory type is added for this, since the actual storage of these
(e.g. UBO-s) may differ per implementation. The uploading of kernel
parameters is handled by launch_grid, "MEMORY[x], INPUT" allows drivers
to use an access mechanism for parameter reads which matches with the
upload method.

Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu> (v1)
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com> (v2)

7 files changed, 50 insertions, 25 deletions

diff --git a/src/gallium/auxiliary/tgsi/tgsi_build.c b/src/gallium/auxiliary/tgsi/tgsi_build.c
index f767b08913..7e30bb646e 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_build.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_build.c
@@ -111,7 +111,7 @@ tgsi_default_declaration( void )
    declaration.Local = 0;
    declaration.Array = 0;
    declaration.Atomic = 0;
-   declaration.Shared = 0;
+   declaration.MemType = TGSI_MEMORY_TYPE_GLOBAL;
    declaration.Padding = 0;
 
    return declaration;
@@ -128,7 +128,7 @@ tgsi_build_declaration(
    unsigned local,
    unsigned array,
    unsigned atomic,
-   unsigned shared,
+   unsigned mem_type,
    struct tgsi_header *header )
 {
    struct tgsi_declaration declaration;
@@ -146,7 +146,7 @@ tgsi_build_declaration(
    declaration.Local = local;
    declaration.Array = array;
    declaration.Atomic = atomic;
-   declaration.Shared = shared;
+   declaration.MemType = mem_type;
    header_bodysize_grow( header );
 
    return declaration;
@@ -406,7 +406,7 @@ tgsi_build_full_declaration(
       full_decl->Declaration.Local,
       full_decl->Declaration.Array,
       full_decl->Declaration.Atomic,
-      full_decl->Declaration.Shared,
+      full_decl->Declaration.MemType,
       header );
 
    if (maxsize <= size)
diff --git a/src/gallium/auxiliary/tgsi/tgsi_dump.c b/src/gallium/auxiliary/tgsi/tgsi_dump.c
index c8b91bba53..6d39ef2108 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_dump.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_dump.c
@@ -365,8 +365,13 @@ iter_declaration(
    }
 
    if (decl->Declaration.File == TGSI_FILE_MEMORY) {
-      if (decl->Declaration.Shared)
-         TXT(", SHARED");
+      switch (decl->Declaration.MemType) {
+      /* Note: ,GLOBAL is optional / the default */
+      case TGSI_MEMORY_TYPE_GLOBAL:  TXT(", GLOBAL");  break;
+      case TGSI_MEMORY_TYPE_SHARED:  TXT(", SHARED");  break;
+      case TGSI_MEMORY_TYPE_PRIVATE: TXT(", PRIVATE"); break;
+      case TGSI_MEMORY_TYPE_INPUT:   TXT(", INPUT");   break;
+      }
    }
 
    if (decl->Declaration.File == TGSI_FILE_SAMPLER_VIEW) {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c b/src/gallium/auxiliary/tgsi/tgsi_text.c
index 77598d2cb7..028633c941 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_text.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
@@ -1390,8 +1390,18 @@ static boolean parse_declaration( struct translate_ctx *ctx )
             ctx->cur = cur;
          }
       } else if (file == TGSI_FILE_MEMORY) {
-         if (str_match_nocase_whole(&cur, "SHARED")) {
-            decl.Declaration.Shared = 1;
+         if (str_match_nocase_whole(&cur, "GLOBAL")) {
+            /* Note this is a no-op global is the default */
+            decl.Declaration.MemType = TGSI_MEMORY_TYPE_GLOBAL;
+            ctx->cur = cur;
+         } else if (str_match_nocase_whole(&cur, "SHARED")) {
+            decl.Declaration.MemType = TGSI_MEMORY_TYPE_SHARED;
+            ctx->cur = cur;
+         } else if (str_match_nocase_whole(&cur, "PRIVATE")) {
+            decl.Declaration.MemType = TGSI_MEMORY_TYPE_PRIVATE;
+            ctx->cur = cur;
+         } else if (str_match_nocase_whole(&cur, "INPUT")) {
+            decl.Declaration.MemType = TGSI_MEMORY_TYPE_INPUT;
             ctx->cur = cur;
          }
       } else {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.c b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
index 0dd5ea76f3..297e257b32 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.c
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.c
@@ -191,7 +191,7 @@ struct ureg_program
 
    struct ureg_tokens domain[2];
 
-   bool use_shared_memory;
+   bool use_memory[TGSI_MEMORY_TYPE_COUNT];
 };
 
 static union tgsi_any_token error_tokens[32];
@@ -730,13 +730,14 @@ struct ureg_src ureg_DECL_buffer(struct ureg_program *ureg, unsigned nr,
    return reg;
 }
 
-/* Allocate a shared memory area.
+/* Allocate a memory area.
  */
-struct ureg_src ureg_DECL_shared_memory(struct ureg_program *ureg)
+struct ureg_src ureg_DECL_memory(struct ureg_program *ureg,
+                                 unsigned memory_type)
 {
-   struct ureg_src reg = ureg_src_register(TGSI_FILE_MEMORY, 0);
+   struct ureg_src reg = ureg_src_register(TGSI_FILE_MEMORY, memory_type);
 
-   ureg->use_shared_memory = true;
+   ureg->use_memory[memory_type] = true;
    return reg;
 }
 
@@ -1673,7 +1674,7 @@ emit_decl_buffer(struct ureg_program *ureg,
 }
 
 static void
-emit_decl_shared_memory(struct ureg_program *ureg)
+emit_decl_memory(struct ureg_program *ureg, unsigned memory_type)
 {
    union tgsi_any_token *out = get_tokens(ureg, DOMAIN_DECL, 2);
 
@@ -1682,11 +1683,11 @@ emit_decl_shared_memory(struct ureg_program *ureg)
    out[0].decl.NrTokens = 2;
    out[0].decl.File = TGSI_FILE_MEMORY;
    out[0].decl.UsageMask = TGSI_WRITEMASK_XYZW;
-   out[0].decl.Shared = true;
+   out[0].decl.MemType = memory_type;
 
    out[1].value = 0;
-   out[1].decl_range.First = 0;
-   out[1].decl_range.Last = 0;
+   out[1].decl_range.First = memory_type;
+   out[1].decl_range.Last = memory_type;
 }
 
 static void
@@ -1861,8 +1862,10 @@ static void emit_decls( struct ureg_program *ureg )
       emit_decl_buffer(ureg, ureg->buffer[i].index, ureg->buffer[i].atomic);
    }
 
-   if (ureg->use_shared_memory)
-      emit_decl_shared_memory(ureg);
+   for (i = 0; i < TGSI_MEMORY_TYPE_COUNT; i++) {
+      if (ureg->use_memory[i])
+         emit_decl_memory(ureg, i);
+   }
 
    if (ureg->const_decls.nr_constant_ranges) {
       for (i = 0; i < ureg->const_decls.nr_constant_ranges; i++) {
diff --git a/src/gallium/auxiliary/tgsi/tgsi_ureg.h b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
index 74324678a9..b4258fdbaa 100644
--- a/src/gallium/auxiliary/tgsi/tgsi_ureg.h
+++ b/src/gallium/auxiliary/tgsi/tgsi_ureg.h
@@ -340,7 +340,7 @@ struct ureg_src
 ureg_DECL_buffer(struct ureg_program *ureg, unsigned nr, bool atomic);
 
 struct ureg_src
-ureg_DECL_shared_memory(struct ureg_program *ureg);
+ureg_DECL_memory(struct ureg_program *ureg, unsigned memory_type);
 
 static inline struct ureg_src
 ureg_imm4f( struct ureg_program *ureg,
diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp
index 39cd986c23..5d744aff33 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_from_tgsi.cpp
@@ -864,7 +864,7 @@ public:
    std::vector<Resource> resources;
 
    struct MemoryFile {
-      bool shared;
+      uint8_t mem_type; // TGSI_MEMORY_TYPE_*
    };
    std::vector<MemoryFile> memoryFiles;
 
@@ -1222,7 +1222,7 @@ bool Source::scanDeclaration(const struct tgsi_full_declaration *decl)
       break;
    case TGSI_FILE_MEMORY:
       for (i = first; i <= last; ++i)
-         memoryFiles[i].shared = decl->Declaration.Shared;
+         memoryFiles[i].mem_type = decl->Declaration.MemType;
       break;
    case TGSI_FILE_NULL:
    case TGSI_FILE_TEMPORARY:
@@ -1527,7 +1527,8 @@ Converter::makeSym(uint tgsiFile, int fileIdx, int idx, int c, uint32_t address)
 
    sym->reg.fileIndex = fileIdx;
 
-   if (tgsiFile == TGSI_FILE_MEMORY && code->memoryFiles[fileIdx].shared)
+   if (tgsiFile == TGSI_FILE_MEMORY &&
+       code->memoryFiles[fileIdx].mem_type == TGSI_MEMORY_TYPE_SHARED)
       sym->setFile(FILE_MEMORY_SHARED);
 
    if (idx >= 0) {
diff --git a/src/gallium/include/pipe/p_shader_tokens.h b/src/gallium/include/pipe/p_shader_tokens.h
index 5c460276d7..e1979dd1eb 100644
--- a/src/gallium/include/pipe/p_shader_tokens.h
+++ b/src/gallium/include/pipe/p_shader_tokens.h
@@ -117,6 +117,12 @@ enum tgsi_file_type {
 #define TGSI_CYLINDRICAL_WRAP_Z (1 << 2)
 #define TGSI_CYLINDRICAL_WRAP_W (1 << 3)
 
+#define TGSI_MEMORY_TYPE_GLOBAL        0 /* OpenCL global              */
+#define TGSI_MEMORY_TYPE_SHARED        1 /* OpenCL local / GLSL shared */
+#define TGSI_MEMORY_TYPE_PRIVATE       2 /* OpenCL private             */
+#define TGSI_MEMORY_TYPE_INPUT         3 /* OpenCL kernel input params */
+#define TGSI_MEMORY_TYPE_COUNT         4
+
 struct tgsi_declaration
 {
    unsigned Type        : 4;  /**< TGSI_TOKEN_TYPE_DECLARATION */
@@ -130,8 +136,8 @@ struct tgsi_declaration
    unsigned Local       : 1;  /**< optimize as subroutine local variable? */
    unsigned Array       : 1;  /**< extra array info? */
    unsigned Atomic      : 1;  /**< atomic only? for TGSI_FILE_BUFFER */
-   unsigned Shared      : 1;  /**< shared storage for TGSI_FILE_MEMORY */
-   unsigned Padding     : 4;
+   unsigned MemType     : 2;  /**< TGSI_MEMORY_TYPE_x for TGSI_FILE_MEMORY */
+   unsigned Padding     : 3;
 };
 
 struct tgsi_declaration_range