GPU: Vertex Format: ADd function for safe GLSL attrib name

This remove code duplication and use base63 encoding of the hash.
Use mumur hash to have more randomness.

6 files changed, 108 insertions, 50 deletions

diff --git a/source/blender/draw/intern/draw_cache_extract_mesh.c b/source/blender/draw/intern/draw_cache_extract_mesh.c
index f9d6c9ed582..ea1813464c3 100644
--- a/source/blender/draw/intern/draw_cache_extract_mesh.c
+++ b/source/blender/draw/intern/draw_cache_extract_mesh.c
@@ -1570,23 +1570,17 @@ static void *extract_uv_tan_init(const MeshRenderData *mr, void *buf)
   bool orco_allocated = false;
   const bool use_orco_tan = mr->cache->cd_used.tan_orco != 0;
 
-  /* XXX FIXME XXX */
-  /* We use a hash to identify each data layer based on its name.
-   * Gawain then search for this name in the current shader and bind if it exists.
-   * NOTE : This is prone to hash collision.
-   * One solution to hash collision would be to format the cd layer name
-   * to a safe glsl var name, but without name clash.
-   * NOTE 2 : Replicate changes to code_generate_vertex_new() in gpu_codegen.c */
   for (int i = 0; i < MAX_MTFACE; i++) {
     if (uv_layers & (1 << i)) {
-      char attr_name[32];
+      char attr_name[32], attr_safe_name[GPU_MAX_SAFE_ATTRIB_NAME];
       const char *layer_name = CustomData_get_layer_name(cd_ldata, CD_MLOOPUV, i);
-      uint hash = BLI_ghashutil_strhash_p(layer_name);
+
+      GPU_vertformat_safe_attrib_name(layer_name, attr_safe_name, GPU_MAX_SAFE_ATTRIB_NAME);
       /* UV layer name. */
-      BLI_snprintf(attr_name, sizeof(attr_name), "u%u", hash);
+      BLI_snprintf(attr_name, sizeof(attr_name), "u%s", attr_safe_name);
       GPU_vertformat_attr_add(&format, attr_name, GPU_COMP_F32, 2, GPU_FETCH_FLOAT);
       /* Auto layer name. */
-      BLI_snprintf(attr_name, sizeof(attr_name), "a%u", hash);
+      BLI_snprintf(attr_name, sizeof(attr_name), "a%s", attr_safe_name);
       GPU_vertformat_alias_add(&format, attr_name);
       /* Active render layer name. */
       if (i == CustomData_get_render_layer(cd_ldata, CD_MLOOPUV)) {
@@ -1610,11 +1604,11 @@ static void *extract_uv_tan_init(const MeshRenderData *mr, void *buf)
 
   for (int i = 0; i < MAX_MTFACE; i++) {
     if (tan_layers & (1 << i)) {
-      char attr_name[32];
+      char attr_name[32], attr_safe_name[GPU_MAX_SAFE_ATTRIB_NAME];
       const char *layer_name = CustomData_get_layer_name(cd_ldata, CD_MLOOPUV, i);
-      uint hash = BLI_ghashutil_strhash_p(layer_name);
+      GPU_vertformat_safe_attrib_name(layer_name, attr_safe_name, GPU_MAX_SAFE_ATTRIB_NAME);
       /* Tangent layer name. */
-      BLI_snprintf(attr_name, sizeof(attr_name), "t%u", hash);
+      BLI_snprintf(attr_name, sizeof(attr_name), "t%s", attr_safe_name);
       GPU_vertformat_attr_add(&format, attr_name, GPU_COMP_F32, 4, GPU_FETCH_FLOAT);
       /* Active render layer name. */
       if (i == CustomData_get_render_layer(cd_ldata, CD_MLOOPUV)) {
@@ -1687,10 +1681,10 @@ static void *extract_uv_tan_init(const MeshRenderData *mr, void *buf)
   }
 
   if (use_orco_tan) {
-    char attr_name[32];
+    char attr_name[32], attr_safe_name[GPU_MAX_SAFE_ATTRIB_NAME];
     const char *layer_name = CustomData_get_layer_name(cd_ldata, CD_TANGENT, 0);
-    uint hash = BLI_ghashutil_strhash_p(layer_name);
-    BLI_snprintf(attr_name, sizeof(*attr_name), "t%u", hash);
+    GPU_vertformat_safe_attrib_name(layer_name, attr_safe_name, GPU_MAX_SAFE_ATTRIB_NAME);
+    BLI_snprintf(attr_name, sizeof(*attr_name), "t%s", attr_safe_name);
     GPU_vertformat_attr_add(&format, attr_name, GPU_COMP_F32, 4, GPU_FETCH_FLOAT);
     GPU_vertformat_alias_add(&format, "t");
     GPU_vertformat_alias_add(&format, "at");
@@ -1779,20 +1773,13 @@ static void *extract_vcol_init(const MeshRenderData *mr, void *buf)
   CustomData *cd_ldata = &mr->me->ldata;
   uint32_t vcol_layers = mr->cache->cd_used.vcol;
 
-  /* XXX FIXME XXX */
-  /* We use a hash to identify each data layer based on its name.
-   * Gawain then search for this name in the current shader and bind if it exists.
-   * NOTE : This is prone to hash collision.
-   * One solution to hash collision would be to format the cd layer name
-   * to a safe glsl var name, but without name clash.
-   * NOTE 2 : Replicate changes to code_generate_vertex_new() in gpu_codegen.c */
   for (int i = 0; i < 8; i++) {
     if (vcol_layers & (1 << i)) {
-      char attr_name[32];
+      char attr_name[32], attr_safe_name[GPU_MAX_SAFE_ATTRIB_NAME];
       const char *layer_name = CustomData_get_layer_name(cd_ldata, CD_MLOOPCOL, i);
-      uint hash = BLI_ghashutil_strhash_p(layer_name);
+      GPU_vertformat_safe_attrib_name(layer_name, attr_safe_name, GPU_MAX_SAFE_ATTRIB_NAME);
 
-      BLI_snprintf(attr_name, sizeof(attr_name), "c%u", hash);
+      BLI_snprintf(attr_name, sizeof(attr_name), "c%s", attr_safe_name);
       GPU_vertformat_attr_add(&format, attr_name, GPU_COMP_U8, 4, GPU_FETCH_INT_TO_FLOAT_UNIT);
 
       if (i == CustomData_get_render_layer(cd_ldata, CD_MLOOPCOL)) {
@@ -1804,7 +1791,7 @@ static void *extract_vcol_init(const MeshRenderData *mr, void *buf)
       /* Gather number of auto layers. */
       /* We only do vcols that are not overridden by uvs */
       if (CustomData_get_named_layer_index(cd_ldata, CD_MLOOPUV, layer_name) == -1) {
-        BLI_snprintf(attr_name, sizeof(attr_name), "a%u", hash);
+        BLI_snprintf(attr_name, sizeof(attr_name), "a%s", attr_safe_name);
         GPU_vertformat_alias_add(&format, attr_name);
       }
     }
diff --git a/source/blender/draw/intern/draw_cache_impl_mesh.c b/source/blender/draw/intern/draw_cache_impl_mesh.c
index f498771b596..b23dac838e6 100644
--- a/source/blender/draw/intern/draw_cache_impl_mesh.c
+++ b/source/blender/draw/intern/draw_cache_impl_mesh.c
@@ -244,10 +244,12 @@ static void mesh_cd_extract_auto_layers_names_and_srgb(Mesh *me,
   for (int i = 0; i < uv_len; i++) {
     if ((cd_used.uv & (1 << i)) != 0) {
       const char *name = CustomData_get_layer_name(cd_ldata, CD_MLOOPUV, i);
-      uint hash = BLI_ghashutil_strhash_p(name);
+      char safe_name[GPU_MAX_SAFE_ATTRIB_NAME];
+      GPU_vertformat_safe_attrib_name(name, safe_name, GPU_MAX_SAFE_ATTRIB_NAME);
+      auto_ofs += BLI_snprintf_rlen(
+          auto_names + auto_ofs, auto_names_len - auto_ofs, "ba%s", safe_name);
       /* +1 to include '\0' terminator. */
-      auto_ofs += 1 + BLI_snprintf_rlen(
-                          auto_names + auto_ofs, auto_names_len - auto_ofs, "ba%u", hash);
+      auto_ofs += 1;
     }
   }
 
@@ -257,10 +259,12 @@ static void mesh_cd_extract_auto_layers_names_and_srgb(Mesh *me,
       const char *name = CustomData_get_layer_name(cd_ldata, CD_MLOOPCOL, i);
       /* We only do vcols that are not overridden by a uv layer with same name. */
       if (CustomData_get_named_layer_index(cd_ldata, CD_MLOOPUV, name) == -1) {
-        uint hash = BLI_ghashutil_strhash_p(name);
+        char safe_name[GPU_MAX_SAFE_ATTRIB_NAME];
+        GPU_vertformat_safe_attrib_name(name, safe_name, GPU_MAX_SAFE_ATTRIB_NAME);
+        auto_ofs += BLI_snprintf_rlen(
+            auto_names + auto_ofs, auto_names_len - auto_ofs, "ba%s", safe_name);
         /* +1 to include '\0' terminator. */
-        auto_ofs += 1 + BLI_snprintf_rlen(
-                            auto_names + auto_ofs, auto_names_len - auto_ofs, "ba%u", hash);
+        auto_ofs += 1;
         auto_is_srgb[auto_is_srgb_ofs] = true;
         auto_is_srgb_ofs++;
       }
diff --git a/source/blender/gpu/GPU_shader.h b/source/blender/gpu/GPU_shader.h
index 124f1f1ff8a..f4a94c7759a 100644
--- a/source/blender/gpu/GPU_shader.h
+++ b/source/blender/gpu/GPU_shader.h
@@ -396,7 +396,9 @@ void GPU_shader_free_builtin_shaders(void);
 
 /* Vertex attributes for shaders */
 
-#define GPU_MAX_ATTR 32
+/* Hardware limit is 16. Position attribute is always needed so we reduce to 15.
+ * This makes sure the GPUVertexFormat name buffer does not overflow. */
+#define GPU_MAX_ATTR 15
 
 typedef struct GPUVertAttrLayers {
   struct {
diff --git a/source/blender/gpu/GPU_vertex_format.h b/source/blender/gpu/GPU_vertex_format.h
index fc468436499..97c1828b593 100644
--- a/source/blender/gpu/GPU_vertex_format.h
+++ b/source/blender/gpu/GPU_vertex_format.h
@@ -32,8 +32,10 @@
 
 #define GPU_VERT_ATTR_MAX_LEN 16
 #define GPU_VERT_ATTR_MAX_NAMES 6
-#define GPU_VERT_ATTR_NAME_AVERAGE_LEN 11
-#define GPU_VERT_ATTR_NAMES_BUF_LEN ((GPU_VERT_ATTR_NAME_AVERAGE_LEN + 1) * GPU_VERT_ATTR_MAX_LEN)
+#define GPU_VERT_ATTR_NAMES_BUF_LEN 256
+#define GPU_VERT_FORMAT_MAX_NAMES 63 /* More than enough, actual max is ~30. */
+/* Computed as GPU_VERT_ATTR_NAMES_BUF_LEN / 30 (actual max format name). */
+#define GPU_MAX_SAFE_ATTRIB_NAME 8
 
 typedef enum {
   GPU_COMP_I8,
@@ -80,8 +82,8 @@ BLI_STATIC_ASSERT(GPU_VERT_ATTR_NAMES_BUF_LEN <= 256,
 typedef struct GPUVertFormat {
   /** 0 to 16 (GPU_VERT_ATTR_MAX_LEN). */
   uint attr_len : 5;
-  /** Total count of active vertex attribute. */
-  uint name_len : 5;
+  /** Total count of active vertex attribute names. (max GPU_VERT_FORMAT_MAX_NAMES) */
+  uint name_len : 6;
   /** Stride in bytes, 1 to 1024. */
   uint stride : 11;
   /** Has the format been packed. */
@@ -117,6 +119,8 @@ BLI_INLINE const char *GPU_vertformat_attr_name_get(const GPUVertFormat *format,
   return format->names + attr->names[n_idx];
 }
 
+void GPU_vertformat_safe_attrib_name(const char *attrib_name, char *r_safe_name, uint max_len);
+
 /* format conversion */
 
 typedef struct GPUPackedNormal {
diff --git a/source/blender/gpu/intern/gpu_codegen.c b/source/blender/gpu/intern/gpu_codegen.c
index 3e635b3198a..0e15fdd000b 100644
--- a/source/blender/gpu/intern/gpu_codegen.c
+++ b/source/blender/gpu/intern/gpu_codegen.c
@@ -46,6 +46,7 @@
 #include "GPU_shader.h"
 #include "GPU_texture.h"
 #include "GPU_uniformbuffer.h"
+#include "GPU_vertex_format.h"
 
 #include "BLI_sys_types.h" /* for intptr_t support */
 
@@ -1011,19 +1012,24 @@ static char *code_generate_vertex(ListBase *nodes, const char *vert_code, bool u
               ds, "#define att%d %s\n", input->attr_id, attr_prefix_get(input->attr_type));
         }
         else {
-          uint hash = BLI_ghashutil_strhash_p(input->attr_name);
+          char attr_safe_name[GPU_MAX_SAFE_ATTRIB_NAME];
+          GPU_vertformat_safe_attrib_name(
+              input->attr_name, attr_safe_name, GPU_MAX_SAFE_ATTRIB_NAME);
           BLI_dynstr_appendf(ds,
-                             "DEFINE_ATTR(%s, %s%u);\n",
+                             "DEFINE_ATTR(%s, %s%s);\n",
                              GPU_DATATYPE_STR[input->type],
                              attr_prefix_get(input->attr_type),
-                             hash);
-          BLI_dynstr_appendf(
-              ds, "#define att%d %s%u\n", input->attr_id, attr_prefix_get(input->attr_type), hash);
+                             attr_safe_name);
+          BLI_dynstr_appendf(ds,
+                             "#define att%d %s%s\n",
+                             input->attr_id,
+                             attr_prefix_get(input->attr_type),
+                             attr_safe_name);
           /* Auto attribute can be vertex color byte buffer.
            * We need to know and convert them to linear space in VS. */
           if (input->attr_type == CD_AUTO_FROM_NAME) {
-            BLI_dynstr_appendf(ds, "uniform bool ba%u;\n", hash);
-            BLI_dynstr_appendf(ds, "#define att%d_is_srgb ba%u\n", input->attr_id, hash);
+            BLI_dynstr_appendf(ds, "uniform bool ba%s;\n", attr_safe_name);
+            BLI_dynstr_appendf(ds, "#define att%d_is_srgb ba%s\n", input->attr_id, attr_safe_name);
           }
         }
         BLI_dynstr_appendf(ds,
diff --git a/source/blender/gpu/intern/gpu_vertex_format.c b/source/blender/gpu/intern/gpu_vertex_format.c
index f672d350afa..2cfb17b1568 100644
--- a/source/blender/gpu/intern/gpu_vertex_format.c
+++ b/source/blender/gpu/intern/gpu_vertex_format.c
@@ -31,6 +31,8 @@
 #include <string.h>
 
 #include "BLI_utildefines.h"
+#include "BLI_string.h"
+#include "BLI_ghash.h"
 
 #define PACK_DEBUG 0
 
@@ -149,9 +151,9 @@ uint GPU_vertformat_attr_add(GPUVertFormat *format,
                              GPUVertFetchMode fetch_mode)
 {
 #if TRUST_NO_ONE
-  assert(format->name_len < GPU_VERT_ATTR_MAX_LEN); /* there's room for more */
-  assert(format->attr_len < GPU_VERT_ATTR_MAX_LEN); /* there's room for more */
-  assert(!format->packed);                          /* packed means frozen/locked */
+  assert(format->name_len < GPU_VERT_FORMAT_MAX_NAMES); /* there's room for more */
+  assert(format->attr_len < GPU_VERT_ATTR_MAX_LEN);     /* there's room for more */
+  assert(!format->packed);                              /* packed means frozen/locked */
   assert((comp_len >= 1 && comp_len <= 4) || comp_len == 8 || comp_len == 12 || comp_len == 16);
 
   switch (comp_type) {
@@ -197,7 +199,7 @@ void GPU_vertformat_alias_add(GPUVertFormat *format, const char *alias)
 {
   GPUVertAttr *attr = &format->attrs[format->attr_len - 1];
 #if TRUST_NO_ONE
-  assert(format->name_len < GPU_VERT_ATTR_MAX_LEN); /* there's room for more */
+  assert(format->name_len < GPU_VERT_FORMAT_MAX_NAMES); /* there's room for more */
   assert(attr->name_len < GPU_VERT_ATTR_MAX_NAMES);
 #endif
   format->name_len++; /* multiname support */
@@ -218,6 +220,59 @@ int GPU_vertformat_attr_id_get(const GPUVertFormat *format, const char *name)
   return -1;
 }
 
+/* Encode 4 original bytes into 6 safe bytes. */
+static void safe_bytes(char out[6], const char data[4])
+{
+  char safe_chars[63] = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_";
+
+  uint32_t in = *(uint32_t *)data;
+  for (int i = 0; i < 6; i++) {
+    /* Encoding in base63 */
+    out[i] = safe_chars[in % 63u];
+    in /= 63u;
+  }
+}
+
+#if 0 /* For when we can use 11chars names. */
+/* Encode 8 original bytes into 11 safe bytes. */
+static void safe_bytes(char out[11], const char data[8])
+{
+  char safe_chars[63] = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789_";
+
+  uint64_t in = *(uint64_t *)data;
+  for (int i = 0; i < 11; i++) {
+    /* Encoding in base63 */
+    out[i] = safe_chars[in % 63lu];
+    in /= 63lu;
+  }
+}
+#endif
+
+/* Warning: Always add a prefix to the result of this function as
+ * the generated string can start with a number and not be a valid attribute name. */
+void GPU_vertformat_safe_attrib_name(const char *attrib_name,
+                                     char *r_safe_name,
+                                     uint UNUSED(max_len))
+{
+  char data[4] = {0};
+  /* We use a hash to identify each data layer based on its name.
+   * NOTE: This is still prone to hash collision but the risks are very low.*/
+  *(uint *)data = BLI_ghashutil_strhash_p_murmur(attrib_name);
+  /* Convert to safe bytes characters. */
+  safe_bytes(r_safe_name, data);
+  /* End the string */
+  r_safe_name[6] = '\0';
+
+  /* TOOD(fclem) When UV and Tangent buffers will be separated, name buffer will have plenty more
+   * space. In this case, we can think of having ~13 chars for each name which would be enough to
+   * encode some of the input string along with the hash to reduce colision possibility. */
+
+  BLI_assert(GPU_MAX_SAFE_ATTRIB_NAME >= 7);
+#if 0 /* For debugging */
+  printf("%s > %x > %s\n", attrib_name, *(uint32_t *)data, r_safe_name);
+#endif
+}
+
 /* Make attribute layout non-interleaved.
  * Warning! This does not change data layout!
  * Use direct buffer access to fill the data.