i965: Move binding table update packets to binding table setup time.

This keeps us from needing to reemit all the other stage state just
because a surface changed.

Improves unoptimized glamor x11perf -f8text by 1.10201% +/- 0.489869%
(n=296). [v1]

v2:
 - Drop binding table packets from Gen8 unit state as well.
 - Pass _3DSTATE_BINDING_TABLE_POINTERS_XS to brw_upload_binding_table,
   cutting even more code.
v3: Don't forget to drop them from 3DSTATE_GS (botched refactor in v2).

Signed-off-by: Eric Anholt <[email protected]> [v1]
Reviewed-by: Kenneth Graunke <[email protected]> [v1]
Signed-off-by: Kenneth Graunke <[email protected]> [v2, v3]
Reviewed-by: Eric Anholt <[email protected]> [v3]

7 files changed, 17 insertions, 39 deletions

diff --git a/src/mesa/drivers/dri/i965/brw_binding_tables.c b/src/mesa/drivers/dri/i965/brw_binding_tables.c
index 0de5d1a0737..1cd4bba6753 100644
--- a/src/mesa/drivers/dri/i965/brw_binding_tables.c
+++ b/src/mesa/drivers/dri/i965/brw_binding_tables.c
@@ -52,6 +52,7 @@
  */
 static void
 brw_upload_binding_table(struct brw_context *brw,
+                         uint32_t packet_name,
                          GLbitfield brw_new_binding_table,
                          struct brw_stage_state *stage_state)
 {
@@ -82,6 +83,13 @@ brw_upload_binding_table(struct brw_context *brw,
    }
 
    brw->state.dirty.brw |= brw_new_binding_table;
+
+   if (brw->gen >= 7) {
+      BEGIN_BATCH(2);
+      OUT_BATCH(packet_name << 16 | (2 - 2));
+      OUT_BATCH(stage_state->bind_bo_offset);
+      ADVANCE_BATCH();
+   }
 }
 
 /**
@@ -93,7 +101,9 @@ brw_upload_binding_table(struct brw_context *brw,
 static void
 brw_vs_upload_binding_table(struct brw_context *brw)
 {
-   brw_upload_binding_table(brw, BRW_NEW_VS_BINDING_TABLE, &brw->vs.base);
+   brw_upload_binding_table(brw,
+                            _3DSTATE_BINDING_TABLE_POINTERS_VS,
+                            BRW_NEW_VS_BINDING_TABLE, &brw->vs.base);
 }
 
 const struct brw_tracked_state brw_vs_binding_table = {
@@ -112,7 +122,9 @@ const struct brw_tracked_state brw_vs_binding_table = {
 static void
 brw_upload_wm_binding_table(struct brw_context *brw)
 {
-   brw_upload_binding_table(brw, BRW_NEW_PS_BINDING_TABLE, &brw->wm.base);
+   brw_upload_binding_table(brw,
+                            _3DSTATE_BINDING_TABLE_POINTERS_PS,
+                            BRW_NEW_PS_BINDING_TABLE, &brw->wm.base);
 }
 
 const struct brw_tracked_state brw_wm_binding_table = {
@@ -132,7 +144,9 @@ brw_gs_upload_binding_table(struct brw_context *brw)
    if (brw->geometry_program == NULL)
       return;
 
-   brw_upload_binding_table(brw, BRW_NEW_GS_BINDING_TABLE, &brw->gs.base);
+   brw_upload_binding_table(brw,
+                            _3DSTATE_BINDING_TABLE_POINTERS_GS,
+                            BRW_NEW_GS_BINDING_TABLE, &brw->gs.base);
 }
 
 const struct brw_tracked_state brw_gs_binding_table = {
diff --git a/src/mesa/drivers/dri/i965/gen7_gs_state.c b/src/mesa/drivers/dri/i965/gen7_gs_state.c
index b179d1954b1..d18ae155185 100644
--- a/src/mesa/drivers/dri/i965/gen7_gs_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_gs_state.c
@@ -66,12 +66,6 @@ upload_gs_state(struct brw_context *brw)
    /* CACHE_NEW_GS_PROG */
    const struct brw_vec4_prog_data *prog_data = &brw->gs.prog_data->base;
 
-   /* BRW_NEW_GS_BINDING_TABLE */
-   BEGIN_BATCH(2);
-   OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_GS << 16 | (2 - 2));
-   OUT_BATCH(stage_state->bind_bo_offset);
-   ADVANCE_BATCH();
-
    /* CACHE_NEW_SAMPLER */
    BEGIN_BATCH(2);
    OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_GS << 16 | (2 - 2));
diff --git a/src/mesa/drivers/dri/i965/gen7_vs_state.c b/src/mesa/drivers/dri/i965/gen7_vs_state.c
index c4f1d264bff..0d9859bec59 100644
--- a/src/mesa/drivers/dri/i965/gen7_vs_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_vs_state.c
@@ -75,12 +75,6 @@ upload_vs_state(struct brw_context *brw)
    if (!brw->is_haswell)
       gen7_emit_vs_workaround_flush(brw);
 
-   /* BRW_NEW_VS_BINDING_TABLE */
-   BEGIN_BATCH(2);
-   OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_VS << 16 | (2 - 2));
-   OUT_BATCH(stage_state->bind_bo_offset);
-   ADVANCE_BATCH();
-
    /* CACHE_NEW_SAMPLER */
    BEGIN_BATCH(2);
    OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_VS << 16 | (2 - 2));
diff --git a/src/mesa/drivers/dri/i965/gen7_wm_state.c b/src/mesa/drivers/dri/i965/gen7_wm_state.c
index 38067e60c8b..ca3e2753935 100644
--- a/src/mesa/drivers/dri/i965/gen7_wm_state.c
+++ b/src/mesa/drivers/dri/i965/gen7_wm_state.c
@@ -143,12 +143,6 @@ upload_ps_state(struct brw_context *brw)
    const int max_threads_shift = brw->is_haswell ?
       HSW_PS_MAX_THREADS_SHIFT : IVB_PS_MAX_THREADS_SHIFT;
 
-   /* BRW_NEW_PS_BINDING_TABLE */
-   BEGIN_BATCH(2);
-   OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_PS << 16 | (2 - 2));
-   OUT_BATCH(brw->wm.base.bind_bo_offset);
-   ADVANCE_BATCH();
-
    /* CACHE_NEW_SAMPLER */
    BEGIN_BATCH(2);
    OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_PS << 16 | (2 - 2));
diff --git a/src/mesa/drivers/dri/i965/gen8_gs_state.c b/src/mesa/drivers/dri/i965/gen8_gs_state.c
index c018d7084fb..97fbf84f7ff 100644
--- a/src/mesa/drivers/dri/i965/gen8_gs_state.c
+++ b/src/mesa/drivers/dri/i965/gen8_gs_state.c
@@ -36,12 +36,6 @@ gen8_upload_gs_state(struct brw_context *brw)
    /* CACHE_NEW_GS_PROG */
    const struct brw_vec4_prog_data *prog_data = &brw->gs.prog_data->base;
 
-   /* BRW_NEW_GS_BINDING_TABLE */
-   BEGIN_BATCH(2);
-   OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_GS << 16 | (2 - 2));
-   OUT_BATCH(stage_state->bind_bo_offset);
-   ADVANCE_BATCH();
-
    /* CACHE_NEW_SAMPLER */
    BEGIN_BATCH(2);
    OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_GS << 16 | (2 - 2));
diff --git a/src/mesa/drivers/dri/i965/gen8_ps_state.c b/src/mesa/drivers/dri/i965/gen8_ps_state.c
index 561fc96e7d3..7d8f9544ee6 100644
--- a/src/mesa/drivers/dri/i965/gen8_ps_state.c
+++ b/src/mesa/drivers/dri/i965/gen8_ps_state.c
@@ -136,12 +136,6 @@ upload_ps_state(struct brw_context *brw)
    struct gl_context *ctx = &brw->ctx;
    uint32_t dw3 = 0, dw6 = 0, dw7 = 0;
 
-   /* BRW_NEW_PS_BINDING_TABLE */
-   BEGIN_BATCH(2);
-   OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_PS << 16 | (2 - 2));
-   OUT_BATCH(brw->wm.base.bind_bo_offset);
-   ADVANCE_BATCH();
-
    /* CACHE_NEW_SAMPLER */
    BEGIN_BATCH(2);
    OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_PS << 16 | (2 - 2));
diff --git a/src/mesa/drivers/dri/i965/gen8_vs_state.c b/src/mesa/drivers/dri/i965/gen8_vs_state.c
index 02a0176a702..373cfe4b6f8 100644
--- a/src/mesa/drivers/dri/i965/gen8_vs_state.c
+++ b/src/mesa/drivers/dri/i965/gen8_vs_state.c
@@ -62,12 +62,6 @@ upload_vs_state(struct brw_context *brw)
    /* CACHE_NEW_VS_PROG */
    const struct brw_vec4_prog_data *prog_data = &brw->vs.prog_data->base;
 
-   /* BRW_NEW_VS_BINDING_TABLE */
-   BEGIN_BATCH(2);
-   OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_VS << 16 | (2 - 2));
-   OUT_BATCH(stage_state->bind_bo_offset);
-   ADVANCE_BATCH();
-
    /* CACHE_NEW_SAMPLER */
    BEGIN_BATCH(2);
    OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_VS << 16 | (2 - 2));