Mesa (master): i965/fs: Use per-channel interference for register_coalesce_2().

[Eric Anholt]

Module: Mesa
Branch: master
Commit: b6af650a095034eaa2de93bf6cf2985d7fdfce89
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=b6af650a095034eaa2de93bf6cf2985d7fdfce89

Author: Eric Anholt <eric at anholt.net>
Date:   Tue Apr 30 14:30:19 2013 -0700

i965/fs: Use per-channel interference for register_coalesce_2().

This will let us coalesce into texture-from-GRF arguments, which would
otherwise be prevented due to the live interval for the whole vgrf
extending across all the MOVs setting up the channels of the message

v2 (Kenneth Graunke): Rebase for renames.

Reviewed-by: Kenneth Graunke <kenneth at whitecape.org>

---

 src/mesa/drivers/dri/i965/brw_fs.cpp               |   33 ++++++--------------
 .../drivers/dri/i965/brw_fs_live_variables.cpp     |   13 ++++++++
 src/mesa/drivers/dri/i965/brw_fs_live_variables.h  |    3 ++
 3 files changed, 26 insertions(+), 23 deletions(-)

diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp b/src/mesa/drivers/dri/i965/brw_fs.cpp
index 74cc9ce..da31b3e 100644
--- a/src/mesa/drivers/dri/i965/brw_fs.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
@@ -2063,11 +2063,16 @@ fs_visitor::register_coalesce_2()
 	  inst->src[0].smear != -1 ||
 	  inst->dst.file != GRF ||
 	  inst->dst.type != inst->src[0].type ||
-	  virtual_grf_sizes[inst->src[0].reg] != 1 ||
-	  virtual_grf_interferes(inst->dst.reg, inst->src[0].reg)) {
+	  virtual_grf_sizes[inst->src[0].reg] != 1) {
 	 continue;
       }
 
+      int var_from = live_intervals->var_from_reg(&inst->src[0]);
+      int var_to = live_intervals->var_from_reg(&inst->dst);
+
+      if (live_intervals->vars_interfere(var_from, var_to))
+         continue;
+
       int reg_from = inst->src[0].reg;
       assert(inst->src[0].reg_offset == 0);
       int reg_to = inst->dst.reg;
@@ -2091,31 +2096,13 @@ fs_visitor::register_coalesce_2()
       }
 
       inst->remove();
-
-      /* We don't need to recalculate live intervals inside the loop despite
-       * invalidating them; we only use them for the interferes test, and we
-       * must have had a situation where the intervals were:
-       *
-       *  from  to
-       *  ^
-       *  |
-       *  v
-       *        ^
-       *        |
-       *        v
-       *
-       * Some register R that might get coalesced with one of these two could
-       * only be referencing "to", otherwise "from"'s range would have been
-       * longer.  R's range could also only start at the end of "to" or later,
-       * otherwise it will conflict with "to" when we try to coalesce "to"
-       * into Rw anyway.
-       */
-      invalidate_live_intervals();
-
       progress = true;
       continue;
    }
 
+   if (progress)
+      invalidate_live_intervals();
+
    return progress;
 }
 
diff --git a/src/mesa/drivers/dri/i965/brw_fs_live_variables.cpp b/src/mesa/drivers/dri/i965/brw_fs_live_variables.cpp
index 41176c7..50aa7a6 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_live_variables.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_live_variables.cpp
@@ -244,6 +244,12 @@ fs_live_variables::compute_start_end()
    }
 }
 
+int
+fs_live_variables::var_from_reg(fs_reg *reg)
+{
+   return var_from_vgrf[reg->reg] + reg->reg_offset;
+}
+
 fs_live_variables::fs_live_variables(fs_visitor *v, cfg_t *cfg)
    : v(v), cfg(cfg)
 {
@@ -335,6 +341,13 @@ fs_visitor::calculate_live_intervals()
 }
 
 bool
+fs_live_variables::vars_interfere(int a, int b)
+{
+   return !(end[b] <= start[a] ||
+            end[a] <= start[b]);
+}
+
+bool
 fs_visitor::virtual_grf_interferes(int a, int b)
 {
    return !(virtual_grf_end[a] <= virtual_grf_start[b] ||
diff --git a/src/mesa/drivers/dri/i965/brw_fs_live_variables.h b/src/mesa/drivers/dri/i965/brw_fs_live_variables.h
index fa14eec..82575d8 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_live_variables.h
+++ b/src/mesa/drivers/dri/i965/brw_fs_live_variables.h
@@ -66,6 +66,9 @@ public:
    void compute_live_variables();
    void compute_start_end();
 
+   bool vars_interfere(int a, int b);
+   int var_from_reg(fs_reg *reg);
+
    fs_visitor *v;
    cfg_t *cfg;
    void *mem_ctx;