[Mesa-dev] [PATCH 2/2] i965/fs: dump some extra lowerings/optimizations on optimize

[Alejandro Piñeiro]

Specifically, assign_constant_locations, lower_constant_loads and
split_virtual_grfs. For split and assign we assume that the lowering
will result on changes, as it uses to be the case, to avoid adding
progress checks there.
---

I added them because on my wip code I was getting some strange
assembly coming from one of those optimizations, and wanted to
know which one.

 src/intel/compiler/brw_fs.cpp | 29 +++++++++++++++++------------
 src/intel/compiler/brw_fs.h   |  6 +++---
 2 files changed, 20 insertions(+), 15 deletions(-)

diff --git a/src/intel/compiler/brw_fs.cpp b/src/intel/compiler/brw_fs.cpp
index 382a0fe..2a63449 100644
--- a/src/intel/compiler/brw_fs.cpp
+++ b/src/intel/compiler/brw_fs.cpp
@@ -1658,7 +1658,7 @@ fs_visitor::assign_gs_urb_setup()
  * contiguous sets of GRFs.  If we split, we'll end up with reduced
  * live intervals and better dead code elimination and coalescing.
  */
-void
+bool
 fs_visitor::split_virtual_grfs()
 {
    /* Compact the register file so we eliminate dead vgrfs.  This
@@ -1775,6 +1775,8 @@ fs_visitor::split_virtual_grfs()
       }
    }
    invalidate_live_intervals();
+
+   return true;
 }
 
 /**
@@ -1914,12 +1916,12 @@ set_push_pull_constant_loc(unsigned uniform, int *chunk_start,
  * So, this will push some of them out to the pull constant buffer and
  * update the program to load them.
  */
-void
+bool
 fs_visitor::assign_constant_locations()
 {
    /* Only the first compile gets to decide on locations. */
    if (dispatch_width != min_dispatch_width)
-      return;
+      return false;
 
    bool is_live[uniforms];
    memset(is_live, 0, sizeof(is_live));
@@ -2087,16 +2089,19 @@ fs_visitor::assign_constant_locations()
    if (stage == MESA_SHADER_COMPUTE)
       brw_cs_prog_data(stage_prog_data)->thread_local_id_index =
          new_thread_local_id_index;
+
+   return true;
 }
 
 /**
  * Replace UNIFORM register file access with either UNIFORM_PULL_CONSTANT_LOAD
  * or VARYING_PULL_CONSTANT_LOAD instructions which load values into VGRFs.
  */
-void
+bool
 fs_visitor::lower_constant_loads()
 {
    const unsigned index = stage_prog_data->binding_table.pull_constants_start;
+   bool progress = false;
 
    foreach_block_and_inst_safe (block, fs_inst, inst, cfg) {
       /* Set up the annotation tracking for new generated instructions. */
@@ -2137,6 +2142,7 @@ fs_visitor::lower_constant_loads()
                                inst->src[i].offset % 4;
 
          brw_mark_surface_used(prog_data, index);
+         progress = true;
       }
 
       if (inst->opcode == SHADER_OPCODE_MOV_INDIRECT &&
@@ -2158,9 +2164,12 @@ fs_visitor::lower_constant_loads()
          inst->remove(block);
 
          brw_mark_surface_used(prog_data, index);
+         progress = true;
       }
    }
    invalidate_live_intervals();
+
+   return progress;
 }
 
 bool
@@ -5650,14 +5659,6 @@ fs_visitor::optimize()
     */
    bld = fs_builder(this, 64);
 
-   assign_constant_locations();
-   lower_constant_loads();
-
-   validate();
-
-   split_virtual_grfs();
-   validate();
-
 #define OPT(pass, args...) ({                                           \
       pass_num++;                                                       \
       bool this_progress = pass(args);                                  \
@@ -5680,6 +5681,10 @@ fs_visitor::optimize()
    int iteration = 0;
    int pass_num = 0;
 
+   OPT(assign_constant_locations);
+   OPT(lower_constant_loads);
+   OPT(split_virtual_grfs);
+
    OPT(opt_drop_redundant_mov_to_flags);
 
    do {
diff --git a/src/intel/compiler/brw_fs.h b/src/intel/compiler/brw_fs.h
index 6c8c027..dc6cb6e 100644
--- a/src/intel/compiler/brw_fs.h
+++ b/src/intel/compiler/brw_fs.h
@@ -122,10 +122,10 @@ public:
                                    int first_payload_node);
    int choose_spill_reg(struct ra_graph *g);
    void spill_reg(int spill_reg);
-   void split_virtual_grfs();
+   bool split_virtual_grfs();
    bool compact_virtual_grfs();
-   void assign_constant_locations();
-   void lower_constant_loads();
+   bool assign_constant_locations();
+   bool lower_constant_loads();
    void invalidate_live_intervals();
    void calculate_live_intervals();
    void calculate_register_pressure();
-- 
2.9.3