i965: Track non-compressible sampling of renderbuffers

[mesa.git] / src / mesa / drivers / dri / i965 / brw_state_cache.c

diff --git a/src/mesa/drivers/dri/i965/brw_state_cache.c b/src/mesa/drivers/dri/i965/brw_state_cache.c
index 9e00c837407e328a8907eafbf6c1ff871a0248b0..0e98e654c1c7c67e41afc2de4a59999c5f701d93 100644 (file)
--- a/src/mesa/drivers/dri/i965/brw_state_cache.c
+++ b/src/mesa/drivers/dri/i965/brw_state_cache.c
@@ -51,6 +51,7 @@
 #include "brw_wm.h"
 #include "brw_gs.h"
 #include "brw_cs.h"
+#include "brw_program.h"
 
 #define FILE_DEBUG_FLAG DEBUG_STATE
 
@@ -137,7 +138,7 @@ bool
 brw_search_cache(struct brw_cache *cache,
                  enum brw_cache_id cache_id,
                  const void *key, GLuint key_size,
-                 uint32_t *inout_offset, void *out_aux)
+                 uint32_t *inout_offset, void *inout_aux)
 {
    struct brw_context *brw = cache->brw;
    struct brw_cache_item *item;
@@ -155,11 +156,12 @@ brw_search_cache(struct brw_cache *cache,
    if (item == NULL)
       return false;
 
-   *(void **)out_aux = ((char *)item->key + item->key_size);
+   void *aux = ((char *) item->key) + item->key_size;
 
-   if (item->offset != *inout_offset) {
+   if (item->offset != *inout_offset || aux != *((void **) inout_aux)) {
       brw->ctx.NewDriverState |= (1 << cache_id);
       *inout_offset = item->offset;
+      *((void **) inout_aux) = aux;
    }
 
    return true;
@@ -197,6 +199,7 @@ brw_cache_new_bo(struct brw_cache *cache, uint32_t new_size)
     * that depend on it (state base address on gen5+, or unit state before).
     */
    brw->ctx.NewDriverState |= BRW_NEW_PROGRAM_CACHE;
+   brw->batch.state_base_address_emitted = false;
 }
 
 /**
@@ -208,7 +211,7 @@ brw_lookup_prog(const struct brw_cache *cache,
                 const void *data, unsigned data_size)
 {
    const struct brw_context *brw = cache->brw;
-   int i;
+   unsigned i;
    const struct brw_cache_item *item;
 
    for (i = 0; i < cache->size; i++) {
@@ -349,15 +352,6 @@ brw_init_caches(struct brw_context *brw)
                                  4096, 64);
    if (brw->has_llc)
       drm_intel_gem_bo_map_unsynchronized(cache->bo);
-
-   cache->aux_compare[BRW_CACHE_VS_PROG] = brw_vs_prog_data_compare;
-   cache->aux_compare[BRW_CACHE_GS_PROG] = brw_gs_prog_data_compare;
-   cache->aux_compare[BRW_CACHE_FS_PROG] = brw_wm_prog_data_compare;
-   cache->aux_compare[BRW_CACHE_CS_PROG] = brw_cs_prog_data_compare;
-   cache->aux_free[BRW_CACHE_VS_PROG] = brw_stage_prog_data_free;
-   cache->aux_free[BRW_CACHE_GS_PROG] = brw_stage_prog_data_free;
-   cache->aux_free[BRW_CACHE_FS_PROG] = brw_stage_prog_data_free;
-   cache->aux_free[BRW_CACHE_CS_PROG] = brw_stage_prog_data_free;
 }
 
 static void
@@ -371,9 +365,12 @@ brw_clear_cache(struct brw_context *brw, struct brw_cache *cache)
    for (i = 0; i < cache->size; i++) {
       for (c = cache->items[i]; c; c = next) {
         next = c->next;
-         if (cache->aux_free[c->cache_id]) {
+         if (c->cache_id == BRW_CACHE_VS_PROG ||
+             c->cache_id == BRW_CACHE_GS_PROG ||
+             c->cache_id == BRW_CACHE_FS_PROG ||
+             c->cache_id == BRW_CACHE_CS_PROG) {
             const void *item_aux = c->key + c->key_size;
-            cache->aux_free[c->cache_id](item_aux);
+            brw_stage_prog_data_free(item_aux);
          }
         free((void *)c->key);
         free(c);
@@ -391,8 +388,27 @@ brw_clear_cache(struct brw_context *brw, struct brw_cache *cache)
    /* We need to make sure that the programs get regenerated, since
     * any offsets leftover in brw_context will no longer be valid.
     */
-   brw->NewGLState |= ~0;
-   brw->ctx.NewDriverState |= ~0ull;
+   brw->NewGLState = ~0;
+   brw->ctx.NewDriverState = ~0ull;
+   brw->state.pipelines[BRW_RENDER_PIPELINE].mesa = ~0;
+   brw->state.pipelines[BRW_RENDER_PIPELINE].brw = ~0ull;
+   brw->state.pipelines[BRW_COMPUTE_PIPELINE].mesa = ~0;
+   brw->state.pipelines[BRW_COMPUTE_PIPELINE].brw = ~0ull;
+
+   /* Also, NULL out any stale program pointers. */
+   brw->vs.prog_data = NULL;
+   brw->vs.base.prog_data = NULL;
+   brw->tcs.prog_data = NULL;
+   brw->tcs.base.prog_data = NULL;
+   brw->tes.prog_data = NULL;
+   brw->tes.base.prog_data = NULL;
+   brw->gs.prog_data = NULL;
+   brw->gs.base.prog_data = NULL;
+   brw->wm.prog_data = NULL;
+   brw->wm.base.prog_data = NULL;
+   brw->cs.prog_data = NULL;
+   brw->cs.base.prog_data = NULL;
+
    intel_batchbuffer_flush(brw);
 }