i965: Move BRW_NEW_*_PROG_DATA flags to .brw (not .cache).

I put the BRW_NEW_*_PROG_DATA flags at the beginning so that
brw_state_cache.c can still continue using 1 << brw_cache_id.

I also added a comment explaining the difference between
BRW_NEW_*_PROG_DATA and BRW_NEW_*_PROGRAM, as it took me a long time
to remember it.

Non-mechanical changes:
- brw_state_cache.c and brw_ff_gs.c now signal .brw, not .cache.
- brw_state_upload.c - INTEL_DEBUG=state changes.
- brw_context.h - bit definition merging.

v2: Correct the explanation of BRW_NEW_*_PROG_DATA to mention
    state-based recompiles, and nix the "proper subset" claim,
    as it's false. (Caught by Kristian Høgsberg).

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>

34 files changed, 137 insertions, 115 deletions

diff --git a/src/mesa/drivers/dri/i965/brw_binding_tables.c b/src/mesa/drivers/dri/i965/brw_binding_tables.c
index 2e843d5..7ffd7b2 100644
--- a/src/mesa/drivers/dri/i965/brw_binding_tables.c
+++ b/src/mesa/drivers/dri/i965/brw_binding_tables.c
@@ -111,8 +111,8 @@ const struct brw_tracked_state brw_vs_binding_table = {
       .mesa = 0,
       .brw = BRW_NEW_BATCH |
              BRW_NEW_VS_CONSTBUF |
+             BRW_NEW_VS_PROG_DATA |
              BRW_NEW_SURFACES,
-      .cache = BRW_NEW_VS_PROG_DATA
    },
    .emit = brw_vs_upload_binding_table,
 };
@@ -131,8 +131,8 @@ const struct brw_tracked_state brw_wm_binding_table = {
    .dirty = {
       .mesa = 0,
       .brw = BRW_NEW_BATCH |
+             BRW_NEW_FS_PROG_DATA |
              BRW_NEW_SURFACES,
-      .cache = BRW_NEW_FS_PROG_DATA
    },
    .emit = brw_upload_wm_binding_table,
 };
@@ -155,8 +155,8 @@ const struct brw_tracked_state brw_gs_binding_table = {
       .mesa = 0,
       .brw = BRW_NEW_BATCH |
              BRW_NEW_GS_CONSTBUF |
+             BRW_NEW_GS_PROG_DATA |
              BRW_NEW_SURFACES,
-      .cache = BRW_NEW_GS_PROG_DATA
    },
    .emit = brw_gs_upload_binding_table,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_clip_state.c b/src/mesa/drivers/dri/i965/brw_clip_state.c
index 0e1aa58..09a2523 100644
--- a/src/mesa/drivers/dri/i965/brw_clip_state.c
+++ b/src/mesa/drivers/dri/i965/brw_clip_state.c
@@ -167,10 +167,10 @@ const struct brw_tracked_state brw_clip_unit = {
                _NEW_TRANSFORM |
                _NEW_VIEWPORT,
       .brw   = BRW_NEW_BATCH |
+               BRW_NEW_CLIP_PROG_DATA |
                BRW_NEW_CURBE_OFFSETS |
                BRW_NEW_PROGRAM_CACHE |
                BRW_NEW_URB_FENCE,
-      .cache = BRW_NEW_CLIP_PROG_DATA
    },
    .emit = brw_upload_clip_unit,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_context.h b/src/mesa/drivers/dri/i965/brw_context.h
index a671ce1..e678bde 100644
--- a/src/mesa/drivers/dri/i965/brw_context.h
+++ b/src/mesa/drivers/dri/i965/brw_context.h
@@ -149,8 +149,21 @@ struct brw_vec4_prog_key;
 struct brw_wm_prog_key;
 struct brw_wm_prog_data;
 
+enum brw_cache_id {
+   BRW_CACHE_FS_PROG,
+   BRW_CACHE_BLORP_BLIT_PROG,
+   BRW_CACHE_SF_PROG,
+   BRW_CACHE_VS_PROG,
+   BRW_CACHE_FF_GS_PROG,
+   BRW_CACHE_GS_PROG,
+   BRW_CACHE_CLIP_PROG,
+
+   BRW_MAX_CACHE
+};
+
 enum brw_state_id {
-   BRW_STATE_URB_FENCE,
+   /* brw_cache_ids must come first - see brw_state_cache.c */
+   BRW_STATE_URB_FENCE = BRW_MAX_CACHE,
    BRW_STATE_FRAGMENT_PROGRAM,
    BRW_STATE_GEOMETRY_PROGRAM,
    BRW_STATE_VERTEX_PROGRAM,
@@ -191,6 +204,38 @@ enum brw_state_id {
    BRW_NUM_STATE_BITS
 };
 
+/**
+ * BRW_NEW_*_PROG_DATA and BRW_NEW_*_PROGRAM are similar, but distinct.
+ *
+ * BRW_NEW_*_PROGRAM relates to the gl_shader_program/gl_program structures.
+ * When the currently bound shader program differs from the previous draw
+ * call, these will be flagged.  They cover brw->{stage}_program and
+ * ctx->{Stage}Program->_Current.
+ *
+ * BRW_NEW_*_PROG_DATA is flagged when the effective shaders change, from a
+ * driver perspective.  Even if the same shader is bound at the API level,
+ * we may need to switch between multiple versions of that shader to handle
+ * changes in non-orthagonal state.
+ *
+ * Additionally, multiple shader programs may have identical vertex shaders
+ * (for example), or compile down to the same code in the backend.  We combine
+ * those into a single program cache entry.
+ *
+ * BRW_NEW_*_PROG_DATA occurs when switching program cache entries, which
+ * covers the brw_*_prog_data structures, and brw->*.prog_offset.
+ */
+#define BRW_NEW_FS_PROG_DATA            (1ull << BRW_CACHE_FS_PROG)
+/* XXX: The BRW_NEW_BLORP_BLIT_PROG_DATA dirty bit is unused (as BLORP doesn't
+ * use the normal state upload paths), but the cache is still used.  To avoid
+ * polluting the brw_state_cache code with special cases, we retain the dirty
+ * bit for now.  It should eventually be removed.
+ */
+#define BRW_NEW_BLORP_BLIT_PROG_DATA    (1ull << BRW_CACHE_BLORP_BLIT_PROG)
+#define BRW_NEW_SF_PROG_DATA            (1ull << BRW_CACHE_SF_PROG)
+#define BRW_NEW_VS_PROG_DATA            (1ull << BRW_CACHE_VS_PROG)
+#define BRW_NEW_FF_GS_PROG_DATA         (1ull << BRW_CACHE_FF_GS_PROG)
+#define BRW_NEW_GS_PROG_DATA            (1ull << BRW_CACHE_GS_PROG)
+#define BRW_NEW_CLIP_PROG_DATA          (1ull << BRW_CACHE_CLIP_PROG)
 #define BRW_NEW_URB_FENCE               (1ull << BRW_STATE_URB_FENCE)
 #define BRW_NEW_FRAGMENT_PROGRAM        (1ull << BRW_STATE_FRAGMENT_PROGRAM)
 #define BRW_NEW_GEOMETRY_PROGRAM        (1ull << BRW_STATE_GEOMETRY_PROGRAM)
@@ -692,18 +737,6 @@ struct brw_gs_prog_data
  */
 #define SHADER_TIME_STRIDE 64
 
-enum brw_cache_id {
-   BRW_CACHE_FS_PROG,
-   BRW_CACHE_BLORP_BLIT_PROG,
-   BRW_CACHE_SF_PROG,
-   BRW_CACHE_VS_PROG,
-   BRW_CACHE_FF_GS_PROG,
-   BRW_CACHE_GS_PROG,
-   BRW_CACHE_CLIP_PROG,
-
-   BRW_MAX_CACHE
-};
-
 struct brw_cache_item {
    /**
     * Effectively part of the key, cache_id identifies what kind of state
@@ -775,16 +808,6 @@ enum shader_time_shader_type {
    ST_FS16_RESET,
 };
 
-/* Flags for brw->state.cache.
- */
-#define BRW_NEW_FS_PROG_DATA                (1 << BRW_CACHE_FS_PROG)
-#define BRW_NEW_BLORP_BLIT_PROG_DATA        (1 << BRW_CACHE_BLORP_BLIT_PROG)
-#define BRW_NEW_SF_PROG_DATA                (1 << BRW_CACHE_SF_PROG)
-#define BRW_NEW_VS_PROG_DATA                (1 << BRW_CACHE_VS_PROG)
-#define BRW_NEW_FF_GS_PROG_DATA             (1 << BRW_CACHE_FF_GS_PROG)
-#define BRW_NEW_GS_PROG_DATA                (1 << BRW_CACHE_GS_PROG)
-#define BRW_NEW_CLIP_PROG_DATA              (1 << BRW_CACHE_CLIP_PROG)
-
 struct brw_vertex_buffer {
    /** Buffer object containing the uploaded vertex data */
    drm_intel_bo *bo;
diff --git a/src/mesa/drivers/dri/i965/brw_curbe.c b/src/mesa/drivers/dri/i965/brw_curbe.c
index 15105a5..c3d3b9d 100644
--- a/src/mesa/drivers/dri/i965/brw_curbe.c
+++ b/src/mesa/drivers/dri/i965/brw_curbe.c
@@ -142,9 +142,9 @@ static void calculate_curbe_offsets( struct brw_context *brw )
 const struct brw_tracked_state brw_curbe_offsets = {
    .dirty = {
       .mesa = _NEW_TRANSFORM,
-      .brw  = BRW_NEW_CONTEXT,
-      .cache = BRW_NEW_VS_PROG_DATA |
-               BRW_NEW_FS_PROG_DATA
+      .brw  = BRW_NEW_CONTEXT |
+              BRW_NEW_FS_PROG_DATA |
+              BRW_NEW_VS_PROG_DATA,
    },
    .emit = calculate_curbe_offsets
 };
@@ -311,10 +311,10 @@ const struct brw_tracked_state brw_constant_buffer = {
       .mesa = _NEW_PROGRAM_CONSTANTS,
       .brw  = BRW_NEW_BATCH |
               BRW_NEW_CURBE_OFFSETS |
+              BRW_NEW_FS_PROG_DATA |
               BRW_NEW_PSP | /* Implicit - hardware requires this, not used above */
-              BRW_NEW_URB_FENCE,
-      .cache = BRW_NEW_VS_PROG_DATA |
-               BRW_NEW_FS_PROG_DATA
+              BRW_NEW_URB_FENCE |
+              BRW_NEW_VS_PROG_DATA,
    },
    .emit = brw_upload_constant_buffer,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_draw_upload.c b/src/mesa/drivers/dri/i965/brw_draw_upload.c
index ca70fd3..4ff5815 100644
--- a/src/mesa/drivers/dri/i965/brw_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_draw_upload.c
@@ -870,8 +870,8 @@ const struct brw_tracked_state brw_vertices = {
    .dirty = {
       .mesa = _NEW_POLYGON,
       .brw = BRW_NEW_BATCH |
-             BRW_NEW_VERTICES,
-      .cache = BRW_NEW_VS_PROG_DATA,
+             BRW_NEW_VERTICES |
+             BRW_NEW_VS_PROG_DATA,
    },
    .emit = brw_emit_vertices,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_ff_gs.c b/src/mesa/drivers/dri/i965/brw_ff_gs.c
index f2e9663..653c4b6 100644
--- a/src/mesa/drivers/dri/i965/brw_ff_gs.c
+++ b/src/mesa/drivers/dri/i965/brw_ff_gs.c
@@ -230,7 +230,7 @@ brw_upload_ff_gs_prog(struct brw_context *brw)
    populate_key(brw, &key);
 
    if (brw->ff_gs.prog_active != key.need_gs_prog) {
-      brw->state.dirty.cache |= BRW_NEW_FF_GS_PROG_DATA;
+      brw->state.dirty.brw |= BRW_NEW_FF_GS_PROG_DATA;
       brw->ff_gs.prog_active = key.need_gs_prog;
    }
 
@@ -252,8 +252,8 @@ const struct brw_tracked_state brw_ff_gs_prog = {
    .dirty = {
       .mesa  = _NEW_LIGHT,
       .brw   = BRW_NEW_PRIMITIVE |
-               BRW_NEW_TRANSFORM_FEEDBACK,
-      .cache = BRW_NEW_VS_PROG_DATA
+               BRW_NEW_TRANSFORM_FEEDBACK |
+               BRW_NEW_VS_PROG_DATA,
    },
    .emit = brw_upload_ff_gs_prog
 };
diff --git a/src/mesa/drivers/dri/i965/brw_gs_state.c b/src/mesa/drivers/dri/i965/brw_gs_state.c
index cc646de..cb46ace 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_state.c
@@ -93,9 +93,9 @@ const struct brw_tracked_state brw_gs_unit = {
       .mesa  = 0,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_CURBE_OFFSETS |
+               BRW_NEW_FF_GS_PROG_DATA |
                BRW_NEW_PROGRAM_CACHE |
                BRW_NEW_URB_FENCE,
-      .cache = BRW_NEW_FF_GS_PROG_DATA
    },
    .emit = brw_upload_gs_unit,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_gs_surface_state.c b/src/mesa/drivers/dri/i965/brw_gs_surface_state.c
index 267880b..00fcc31 100644
--- a/src/mesa/drivers/dri/i965/brw_gs_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_gs_surface_state.c
@@ -58,8 +58,8 @@ const struct brw_tracked_state brw_gs_pull_constants = {
    .dirty = {
       .mesa = _NEW_PROGRAM_CONSTANTS,
       .brw = BRW_NEW_BATCH |
-             BRW_NEW_GEOMETRY_PROGRAM,
-      .cache = BRW_NEW_GS_PROG_DATA,
+             BRW_NEW_GEOMETRY_PROGRAM |
+             BRW_NEW_GS_PROG_DATA,
    },
    .emit = brw_upload_gs_pull_constants,
 };
@@ -85,8 +85,8 @@ const struct brw_tracked_state brw_gs_ubo_surfaces = {
    .dirty = {
       .mesa = _NEW_PROGRAM,
       .brw = BRW_NEW_BATCH |
+             BRW_NEW_GS_PROG_DATA |
              BRW_NEW_UNIFORM_BUFFER,
-      .cache = BRW_NEW_GS_PROG_DATA,
    },
    .emit = brw_upload_gs_ubo_surfaces,
 };
@@ -110,8 +110,8 @@ const struct brw_tracked_state brw_gs_abo_surfaces = {
    .dirty = {
       .mesa = _NEW_PROGRAM,
       .brw = BRW_NEW_ATOMIC_BUFFER |
-             BRW_NEW_BATCH,
-      .cache = BRW_NEW_GS_PROG_DATA,
+             BRW_NEW_BATCH |
+             BRW_NEW_GS_PROG_DATA,
    },
    .emit = brw_upload_gs_abo_surfaces,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_misc_state.c b/src/mesa/drivers/dri/i965/brw_misc_state.c
index 16b4db6..d4b6e7c 100644
--- a/src/mesa/drivers/dri/i965/brw_misc_state.c
+++ b/src/mesa/drivers/dri/i965/brw_misc_state.c
@@ -117,10 +117,10 @@ const struct brw_tracked_state brw_psp_urb_cbs = {
    .dirty = {
       .mesa = 0,
       .brw = BRW_NEW_BATCH |
+             BRW_NEW_FF_GS_PROG_DATA |
              BRW_NEW_GEN4_UNIT_STATE |
              BRW_NEW_STATE_BASE_ADDRESS |
              BRW_NEW_URB_FENCE,
-      .cache = BRW_NEW_FF_GS_PROG_DATA,
    },
    .emit = upload_psp_urb_cbs,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_sf_state.c b/src/mesa/drivers/dri/i965/brw_sf_state.c
index c653e3b..3306fb5 100644
--- a/src/mesa/drivers/dri/i965/brw_sf_state.c
+++ b/src/mesa/drivers/dri/i965/brw_sf_state.c
@@ -306,9 +306,9 @@ const struct brw_tracked_state brw_sf_unit = {
                _NEW_SCISSOR,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_PROGRAM_CACHE |
+               BRW_NEW_SF_PROG_DATA |
                BRW_NEW_SF_VP |
                BRW_NEW_URB_FENCE,
-      .cache = BRW_NEW_SF_PROG_DATA,
    },
    .emit = upload_sf_unit,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_state_cache.c b/src/mesa/drivers/dri/i965/brw_state_cache.c
index 29b1597..94086a9 100644
--- a/src/mesa/drivers/dri/i965/brw_state_cache.c
+++ b/src/mesa/drivers/dri/i965/brw_state_cache.c
@@ -157,7 +157,7 @@ brw_search_cache(struct brw_cache *cache,
    *(void **)out_aux = ((char *)item->key + item->key_size);
 
    if (item->offset != *inout_offset) {
-      brw->state.dirty.cache |= (1 << cache_id);
+      brw->state.dirty.brw |= (1 << cache_id);
       *inout_offset = item->offset;
    }
 
@@ -339,7 +339,7 @@ brw_upload_cache(struct brw_cache *cache,
 
    *out_offset = item->offset;
    *(void **)out_aux = (void *)((char *)item->key + item->key_size);
-   cache->brw->state.dirty.cache |= 1 << cache_id;
+   cache->brw->state.dirty.brw |= 1 << cache_id;
 }
 
 void
diff --git a/src/mesa/drivers/dri/i965/brw_state_upload.c b/src/mesa/drivers/dri/i965/brw_state_upload.c
index 5ad0668..875f434 100644
--- a/src/mesa/drivers/dri/i965/brw_state_upload.c
+++ b/src/mesa/drivers/dri/i965/brw_state_upload.c
@@ -486,6 +486,13 @@ static struct dirty_bit_map mesa_bits[] = {
 };
 
 static struct dirty_bit_map brw_bits[] = {
+   DEFINE_BIT(BRW_NEW_FS_PROG_DATA),
+   DEFINE_BIT(BRW_NEW_BLORP_BLIT_PROG_DATA),
+   DEFINE_BIT(BRW_NEW_SF_PROG_DATA),
+   DEFINE_BIT(BRW_NEW_VS_PROG_DATA),
+   DEFINE_BIT(BRW_NEW_FF_GS_PROG_DATA),
+   DEFINE_BIT(BRW_NEW_GS_PROG_DATA),
+   DEFINE_BIT(BRW_NEW_CLIP_PROG_DATA),
    DEFINE_BIT(BRW_NEW_URB_FENCE),
    DEFINE_BIT(BRW_NEW_FRAGMENT_PROGRAM),
    DEFINE_BIT(BRW_NEW_GEOMETRY_PROGRAM),
@@ -528,13 +535,6 @@ static struct dirty_bit_map brw_bits[] = {
 };
 
 static struct dirty_bit_map cache_bits[] = {
-   DEFINE_BIT(BRW_NEW_FS_PROG_DATA),
-   DEFINE_BIT(BRW_NEW_BLORP_BLIT_PROG_DATA),
-   DEFINE_BIT(BRW_NEW_SF_PROG_DATA),
-   DEFINE_BIT(BRW_NEW_VS_PROG_DATA),
-   DEFINE_BIT(BRW_NEW_FF_GS_PROG_DATA),
-   DEFINE_BIT(BRW_NEW_GS_PROG_DATA),
-   DEFINE_BIT(BRW_NEW_CLIP_PROG_DATA),
    {0, 0, 0}
 };
 
@@ -648,7 +648,6 @@ void brw_upload_state(struct brw_context *brw)
 
    if (unlikely(INTEL_DEBUG & DEBUG_STATE)) {
       STATIC_ASSERT(ARRAY_SIZE(brw_bits) == BRW_NUM_STATE_BITS + 1);
-      STATIC_ASSERT(ARRAY_SIZE(cache_bits) == BRW_MAX_CACHE + 1);
 
       brw_update_dirty_count(mesa_bits, state->mesa);
       brw_update_dirty_count(brw_bits, state->brw);
diff --git a/src/mesa/drivers/dri/i965/brw_urb.c b/src/mesa/drivers/dri/i965/brw_urb.c
index 09a9065..e76db6a 100644
--- a/src/mesa/drivers/dri/i965/brw_urb.c
+++ b/src/mesa/drivers/dri/i965/brw_urb.c
@@ -212,9 +212,9 @@ done:
 const struct brw_tracked_state brw_recalculate_urb_fence = {
    .dirty = {
       .mesa = 0,
-      .brw = BRW_NEW_CURBE_OFFSETS,
-      .cache = BRW_NEW_SF_PROG_DATA |
-               BRW_NEW_VS_PROG_DATA,
+      .brw = BRW_NEW_CURBE_OFFSETS |
+             BRW_NEW_SF_PROG_DATA |
+             BRW_NEW_VS_PROG_DATA,
    },
    .emit = recalculate_urb_fence
 };
diff --git a/src/mesa/drivers/dri/i965/brw_vs_state.c b/src/mesa/drivers/dri/i965/brw_vs_state.c
index 09ecd3a..998a225 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_state.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_state.c
@@ -192,8 +192,8 @@ const struct brw_tracked_state brw_vs_unit = {
                BRW_NEW_PROGRAM_CACHE |
                BRW_NEW_SAMPLER_STATE_TABLE |
                BRW_NEW_URB_FENCE |
-               BRW_NEW_VERTEX_PROGRAM,
-      .cache = BRW_NEW_VS_PROG_DATA,
+               BRW_NEW_VERTEX_PROGRAM |
+               BRW_NEW_VS_PROG_DATA,
    },
    .emit = brw_upload_vs_unit,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_vs_surface_state.c b/src/mesa/drivers/dri/i965/brw_vs_surface_state.c
index 263f391..288aeb7 100644
--- a/src/mesa/drivers/dri/i965/brw_vs_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_vs_surface_state.c
@@ -129,8 +129,8 @@ const struct brw_tracked_state brw_vs_pull_constants = {
    .dirty = {
       .mesa = _NEW_PROGRAM_CONSTANTS,
       .brw = BRW_NEW_BATCH |
-             BRW_NEW_VERTEX_PROGRAM,
-      .cache = BRW_NEW_VS_PROG_DATA,
+             BRW_NEW_VERTEX_PROGRAM |
+             BRW_NEW_VS_PROG_DATA,
    },
    .emit = brw_upload_vs_pull_constants,
 };
@@ -155,8 +155,8 @@ const struct brw_tracked_state brw_vs_ubo_surfaces = {
    .dirty = {
       .mesa = _NEW_PROGRAM,
       .brw = BRW_NEW_BATCH |
-             BRW_NEW_UNIFORM_BUFFER,
-      .cache = BRW_NEW_VS_PROG_DATA,
+             BRW_NEW_UNIFORM_BUFFER |
+             BRW_NEW_VS_PROG_DATA,
    },
    .emit = brw_upload_vs_ubo_surfaces,
 };
@@ -180,8 +180,8 @@ const struct brw_tracked_state brw_vs_abo_surfaces = {
    .dirty = {
       .mesa = _NEW_PROGRAM,
       .brw = BRW_NEW_ATOMIC_BUFFER |
-             BRW_NEW_BATCH,
-      .cache = BRW_NEW_VS_PROG_DATA,
+             BRW_NEW_BATCH |
+             BRW_NEW_VS_PROG_DATA,
    },
    .emit = brw_upload_vs_abo_surfaces,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_wm_state.c b/src/mesa/drivers/dri/i965/brw_wm_state.c
index 51f48b4..12cbc72 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_state.c
@@ -255,10 +255,10 @@ const struct brw_tracked_state brw_wm_unit = {
       .brw = BRW_NEW_BATCH |
              BRW_NEW_CURBE_OFFSETS |
              BRW_NEW_FRAGMENT_PROGRAM |
+             BRW_NEW_FS_PROG_DATA |
              BRW_NEW_PROGRAM_CACHE |
              BRW_NEW_SAMPLER_STATE_TABLE |
              BRW_NEW_STATS_WM,
-      .cache = BRW_NEW_FS_PROG_DATA,
    },
    .emit = brw_upload_wm_unit,
 };
diff --git a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
index f74c320..8169ef8 100644
--- a/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
+++ b/src/mesa/drivers/dri/i965/brw_wm_surface_state.c
@@ -495,8 +495,8 @@ const struct brw_tracked_state brw_wm_pull_constants = {
    .dirty = {
       .mesa = _NEW_PROGRAM_CONSTANTS,
       .brw = BRW_NEW_BATCH |
-             BRW_NEW_FRAGMENT_PROGRAM,
-      .cache = BRW_NEW_FS_PROG_DATA,
+             BRW_NEW_FRAGMENT_PROGRAM |
+             BRW_NEW_FS_PROG_DATA,
    },
    .emit = brw_upload_wm_pull_constants,
 };
@@ -892,8 +892,8 @@ const struct brw_tracked_state brw_wm_ubo_surfaces = {
    .dirty = {
       .mesa = _NEW_PROGRAM,
       .brw = BRW_NEW_BATCH |
+             BRW_NEW_FS_PROG_DATA |
              BRW_NEW_UNIFORM_BUFFER,
-      .cache = BRW_NEW_FS_PROG_DATA,
    },
    .emit = brw_upload_wm_ubo_surfaces,
 };
@@ -943,8 +943,8 @@ const struct brw_tracked_state brw_wm_abo_surfaces = {
    .dirty = {
       .mesa = _NEW_PROGRAM,
       .brw = BRW_NEW_ATOMIC_BUFFER |
-             BRW_NEW_BATCH,
-      .cache = BRW_NEW_FS_PROG_DATA,
+             BRW_NEW_BATCH |
+             BRW_NEW_FS_PROG_DATA,
    },
    .emit = brw_upload_wm_abo_surfaces,
 };
diff --git a/src/mesa/drivers/dri/i965/gen6_clip_state.c b/src/mesa/drivers/dri/i965/gen6_clip_state.c
index defceba..e8c16ca 100644
--- a/src/mesa/drivers/dri/i965/gen6_clip_state.c
+++ b/src/mesa/drivers/dri/i965/gen6_clip_state.c
@@ -187,9 +187,9 @@ const struct brw_tracked_state gen6_clip_state = {
                _NEW_LIGHT |
                _NEW_TRANSFORM,
       .brw   = BRW_NEW_CONTEXT |
+               BRW_NEW_FS_PROG_DATA |
                BRW_NEW_META_IN_PROGRESS |
                BRW_NEW_RASTERIZER_DISCARD,
-      .cache = BRW_NEW_FS_PROG_DATA
    },
    .emit = upload_clip_state,
 };
@@ -201,9 +201,9 @@ const struct brw_tracked_state gen7_clip_state = {
                _NEW_POLYGON |
                _NEW_TRANSFORM,
       .brw   = BRW_NEW_CONTEXT |
+               BRW_NEW_FS_PROG_DATA |
                BRW_NEW_META_IN_PROGRESS |
                BRW_NEW_RASTERIZER_DISCARD,
-      .cache = BRW_NEW_FS_PROG_DATA
    },
    .emit = upload_clip_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen6_gs_state.c b/src/mesa/drivers/dri/i965/gen6_gs_state.c
index 4ac591c..35f5be7 100644
--- a/src/mesa/drivers/dri/i965/gen6_gs_state.c
+++ b/src/mesa/drivers/dri/i965/gen6_gs_state.c
@@ -57,8 +57,8 @@ const struct brw_tracked_state gen6_gs_push_constants = {
                _NEW_TRANSFORM,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_GEOMETRY_PROGRAM |
+               BRW_NEW_GS_PROG_DATA |
                BRW_NEW_PUSH_CONSTANT_ALLOCATION,
-      .cache = BRW_NEW_GS_PROG_DATA,
    },
    .emit = gen6_upload_gs_push_constants,
 };
@@ -200,10 +200,10 @@ const struct brw_tracked_state gen6_gs_state = {
                _NEW_TRANSFORM,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_CONTEXT |
+               BRW_NEW_FF_GS_PROG_DATA |
                BRW_NEW_GEOMETRY_PROGRAM |
+               BRW_NEW_GS_PROG_DATA |
                BRW_NEW_PUSH_CONSTANT_ALLOCATION,
-      .cache = BRW_NEW_FF_GS_PROG_DATA |
-               BRW_NEW_GS_PROG_DATA,
    },
    .emit = upload_gs_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen6_sf_state.c b/src/mesa/drivers/dri/i965/gen6_sf_state.c
index 591890d..ccb6b30 100644
--- a/src/mesa/drivers/dri/i965/gen6_sf_state.c
+++ b/src/mesa/drivers/dri/i965/gen6_sf_state.c
@@ -444,10 +444,10 @@ const struct brw_tracked_state gen6_sf_state = {
                _NEW_SCISSOR,
       .brw   = BRW_NEW_CONTEXT |
                BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA |
                BRW_NEW_GEOMETRY_PROGRAM |
                BRW_NEW_PRIMITIVE |
                BRW_NEW_VUE_MAP_GEOM_OUT,
-      .cache = BRW_NEW_FS_PROG_DATA
    },
    .emit = upload_sf_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen6_urb.c b/src/mesa/drivers/dri/i965/gen6_urb.c
index 7fb24e4..107a4f2 100644
--- a/src/mesa/drivers/dri/i965/gen6_urb.c
+++ b/src/mesa/drivers/dri/i965/gen6_urb.c
@@ -128,10 +128,10 @@ const struct brw_tracked_state gen6_urb = {
    .dirty = {
       .mesa = 0,
       .brw = BRW_NEW_CONTEXT |
-             BRW_NEW_GEOMETRY_PROGRAM,
-      .cache = BRW_NEW_GS_PROG_DATA |
-               BRW_NEW_VS_PROG_DATA |
-               BRW_NEW_FF_GS_PROG_DATA,
+             BRW_NEW_FF_GS_PROG_DATA |
+             BRW_NEW_GEOMETRY_PROGRAM |
+             BRW_NEW_GS_PROG_DATA |
+             BRW_NEW_VS_PROG_DATA,
    },
    .emit = gen6_upload_urb,
 };
diff --git a/src/mesa/drivers/dri/i965/gen6_vs_state.c b/src/mesa/drivers/dri/i965/gen6_vs_state.c
index 100f567..fc0fd1d 100644
--- a/src/mesa/drivers/dri/i965/gen6_vs_state.c
+++ b/src/mesa/drivers/dri/i965/gen6_vs_state.c
@@ -151,8 +151,8 @@ const struct brw_tracked_state gen6_vs_push_constants = {
                _NEW_TRANSFORM,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_PUSH_CONSTANT_ALLOCATION |
-               BRW_NEW_VERTEX_PROGRAM,
-      .cache = BRW_NEW_VS_PROG_DATA,
+               BRW_NEW_VERTEX_PROGRAM |
+               BRW_NEW_VS_PROG_DATA,
    },
    .emit = gen6_upload_vs_push_constants,
 };
@@ -265,8 +265,8 @@ const struct brw_tracked_state gen6_vs_state = {
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_CONTEXT |
                BRW_NEW_PUSH_CONSTANT_ALLOCATION |
-               BRW_NEW_VERTEX_PROGRAM,
-      .cache = BRW_NEW_VS_PROG_DATA
+               BRW_NEW_VERTEX_PROGRAM |
+               BRW_NEW_VS_PROG_DATA,
    },
    .emit = upload_vs_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen6_wm_state.c b/src/mesa/drivers/dri/i965/gen6_wm_state.c
index d83d4c8..4ac91af 100644
--- a/src/mesa/drivers/dri/i965/gen6_wm_state.c
+++ b/src/mesa/drivers/dri/i965/gen6_wm_state.c
@@ -59,8 +59,8 @@ const struct brw_tracked_state gen6_wm_push_constants = {
       .mesa  = _NEW_PROGRAM_CONSTANTS,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA |
                BRW_NEW_PUSH_CONSTANT_ALLOCATION,
-      .cache = BRW_NEW_FS_PROG_DATA,
    },
    .emit = gen6_upload_wm_push_constants,
 };
@@ -305,8 +305,8 @@ const struct brw_tracked_state gen6_wm_state = {
                _NEW_PROGRAM_CONSTANTS,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA |
                BRW_NEW_PUSH_CONSTANT_ALLOCATION,
-      .cache = BRW_NEW_FS_PROG_DATA
    },
    .emit = upload_wm_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen7_gs_state.c b/src/mesa/drivers/dri/i965/gen7_gs_state.c
index 820c496..0f49139 100644
--- a/src/mesa/drivers/dri/i965/gen7_gs_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_gs_state.c
@@ -156,8 +156,8 @@ const struct brw_tracked_state gen7_gs_state = {
       .mesa  = _NEW_TRANSFORM,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_CONTEXT |
-               BRW_NEW_GEOMETRY_PROGRAM,
-      .cache = BRW_NEW_GS_PROG_DATA
+               BRW_NEW_GEOMETRY_PROGRAM |
+               BRW_NEW_GS_PROG_DATA,
    },
    .emit = upload_gs_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen7_sf_state.c b/src/mesa/drivers/dri/i965/gen7_sf_state.c
index ae6bf13..c260bba 100644
--- a/src/mesa/drivers/dri/i965/gen7_sf_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_sf_state.c
@@ -93,10 +93,10 @@ const struct brw_tracked_state gen7_sbe_state = {
                _NEW_PROGRAM,
       .brw   = BRW_NEW_CONTEXT |
                BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA |
                BRW_NEW_GEOMETRY_PROGRAM |
                BRW_NEW_PRIMITIVE |
                BRW_NEW_VUE_MAP_GEOM_OUT,
-      .cache = BRW_NEW_FS_PROG_DATA
    },
    .emit = upload_sbe_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen7_urb.c b/src/mesa/drivers/dri/i965/gen7_urb.c
index bcf20ff..d90e48c 100644
--- a/src/mesa/drivers/dri/i965/gen7_urb.c
+++ b/src/mesa/drivers/dri/i965/gen7_urb.c
@@ -315,9 +315,9 @@ const struct brw_tracked_state gen7_urb = {
    .dirty = {
       .mesa = 0,
       .brw = BRW_NEW_CONTEXT |
-             BRW_NEW_GEOMETRY_PROGRAM,
-      .cache = BRW_NEW_VS_PROG_DATA |
-               BRW_NEW_GS_PROG_DATA,
+             BRW_NEW_GEOMETRY_PROGRAM |
+             BRW_NEW_GS_PROG_DATA |
+             BRW_NEW_VS_PROG_DATA,
    },
    .emit = gen7_upload_urb,
 };
diff --git a/src/mesa/drivers/dri/i965/gen7_vs_state.c b/src/mesa/drivers/dri/i965/gen7_vs_state.c
index 1d3ac88..3b0126d 100644
--- a/src/mesa/drivers/dri/i965/gen7_vs_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_vs_state.c
@@ -114,8 +114,8 @@ const struct brw_tracked_state gen7_vs_state = {
       .mesa  = _NEW_TRANSFORM,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_CONTEXT |
-               BRW_NEW_VERTEX_PROGRAM,
-      .cache = BRW_NEW_VS_PROG_DATA
+               BRW_NEW_VERTEX_PROGRAM |
+               BRW_NEW_VS_PROG_DATA,
    },
    .emit = upload_vs_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen7_wm_state.c b/src/mesa/drivers/dri/i965/gen7_wm_state.c
index 83a5152..575eaf5 100644
--- a/src/mesa/drivers/dri/i965/gen7_wm_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_wm_state.c
@@ -132,8 +132,8 @@ const struct brw_tracked_state gen7_wm_state = {
                _NEW_MULTISAMPLE |
                _NEW_POLYGON,
       .brw   = BRW_NEW_BATCH |
-               BRW_NEW_FRAGMENT_PROGRAM,
-      .cache = BRW_NEW_FS_PROG_DATA,
+               BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA,
    },
    .emit = upload_wm_state,
 };
@@ -276,8 +276,8 @@ const struct brw_tracked_state gen7_ps_state = {
                _NEW_COLOR |
                _NEW_MULTISAMPLE,
       .brw   = BRW_NEW_BATCH |
-               BRW_NEW_FRAGMENT_PROGRAM,
-      .cache = BRW_NEW_FS_PROG_DATA
+               BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA,
    },
    .emit = upload_ps_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen8_depth_state.c b/src/mesa/drivers/dri/i965/gen8_depth_state.c
index 9a862c5..fc0e3fd 100644
--- a/src/mesa/drivers/dri/i965/gen8_depth_state.c
+++ b/src/mesa/drivers/dri/i965/gen8_depth_state.c
@@ -372,8 +372,8 @@ const struct brw_tracked_state gen8_pma_fix = {
               _NEW_DEPTH |
               _NEW_MULTISAMPLE |
               _NEW_STENCIL,
-      .brw = BRW_NEW_FRAGMENT_PROGRAM,
-      .cache = BRW_NEW_FS_PROG_DATA,
+      .brw = BRW_NEW_FRAGMENT_PROGRAM |
+             BRW_NEW_FS_PROG_DATA,
    },
    .emit = gen8_emit_pma_stall_workaround
 };
diff --git a/src/mesa/drivers/dri/i965/gen8_draw_upload.c b/src/mesa/drivers/dri/i965/gen8_draw_upload.c
index a189c19..9e8e24b 100644
--- a/src/mesa/drivers/dri/i965/gen8_draw_upload.c
+++ b/src/mesa/drivers/dri/i965/gen8_draw_upload.c
@@ -234,8 +234,8 @@ const struct brw_tracked_state gen8_vertices = {
    .dirty = {
       .mesa = _NEW_POLYGON,
       .brw = BRW_NEW_BATCH |
-             BRW_NEW_VERTICES,
-      .cache = BRW_NEW_VS_PROG_DATA,
+             BRW_NEW_VERTICES |
+             BRW_NEW_VS_PROG_DATA,
    },
    .emit = gen8_emit_vertices,
 };
diff --git a/src/mesa/drivers/dri/i965/gen8_gs_state.c b/src/mesa/drivers/dri/i965/gen8_gs_state.c
index 2901538..526a826 100644
--- a/src/mesa/drivers/dri/i965/gen8_gs_state.c
+++ b/src/mesa/drivers/dri/i965/gen8_gs_state.c
@@ -122,8 +122,8 @@ const struct brw_tracked_state gen8_gs_state = {
       .mesa  = _NEW_TRANSFORM,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_CONTEXT |
-               BRW_NEW_GEOMETRY_PROGRAM,
-      .cache = BRW_NEW_GS_PROG_DATA
+               BRW_NEW_GEOMETRY_PROGRAM |
+               BRW_NEW_GS_PROG_DATA,
    },
    .emit = gen8_upload_gs_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen8_ps_state.c b/src/mesa/drivers/dri/i965/gen8_ps_state.c
index 24e19d4..4267bf2 100644
--- a/src/mesa/drivers/dri/i965/gen8_ps_state.c
+++ b/src/mesa/drivers/dri/i965/gen8_ps_state.c
@@ -89,8 +89,8 @@ const struct brw_tracked_state gen8_ps_extra = {
       .mesa  = _NEW_MULTISAMPLE,
       .brw   = BRW_NEW_CONTEXT |
                BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA |
                BRW_NEW_NUM_SAMPLES,
-      .cache = BRW_NEW_FS_PROG_DATA,
    },
    .emit = upload_ps_extra,
 };
@@ -128,8 +128,8 @@ const struct brw_tracked_state gen8_wm_state = {
    .dirty = {
       .mesa  = _NEW_LINE |
                _NEW_POLYGON,
-      .brw   = BRW_NEW_CONTEXT,
-      .cache = BRW_NEW_FS_PROG_DATA,
+      .brw   = BRW_NEW_CONTEXT |
+               BRW_NEW_FS_PROG_DATA,
    },
    .emit = upload_wm_state,
 };
@@ -252,8 +252,8 @@ const struct brw_tracked_state gen8_ps_state = {
    .dirty = {
       .mesa  = _NEW_MULTISAMPLE,
       .brw   = BRW_NEW_BATCH |
-               BRW_NEW_FRAGMENT_PROGRAM,
-      .cache = BRW_NEW_FS_PROG_DATA
+               BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA,
    },
    .emit = upload_ps_state,
 };
diff --git a/src/mesa/drivers/dri/i965/gen8_sf_state.c b/src/mesa/drivers/dri/i965/gen8_sf_state.c
index edeab1e..91d5801 100644
--- a/src/mesa/drivers/dri/i965/gen8_sf_state.c
+++ b/src/mesa/drivers/dri/i965/gen8_sf_state.c
@@ -133,8 +133,8 @@ const struct brw_tracked_state gen8_sbe_state = {
                _NEW_PROGRAM,
       .brw   = BRW_NEW_CONTEXT |
                BRW_NEW_FRAGMENT_PROGRAM |
+               BRW_NEW_FS_PROG_DATA |
                BRW_NEW_VUE_MAP_GEOM_OUT,
-      .cache = BRW_NEW_FS_PROG_DATA,
    },
    .emit = upload_sbe,
 };
diff --git a/src/mesa/drivers/dri/i965/gen8_vs_state.c b/src/mesa/drivers/dri/i965/gen8_vs_state.c
index cc13fc6..00f2731 100644
--- a/src/mesa/drivers/dri/i965/gen8_vs_state.c
+++ b/src/mesa/drivers/dri/i965/gen8_vs_state.c
@@ -84,8 +84,8 @@ const struct brw_tracked_state gen8_vs_state = {
       .mesa  = _NEW_TRANSFORM,
       .brw   = BRW_NEW_BATCH |
                BRW_NEW_CONTEXT |
-               BRW_NEW_VERTEX_PROGRAM,
-      .cache = BRW_NEW_VS_PROG_DATA
+               BRW_NEW_VERTEX_PROGRAM |
+               BRW_NEW_VS_PROG_DATA,
    },
    .emit = upload_vs_state,
 };