summaryrefslogtreecommitdiffstats
path: root/src/glsl/opt_algebraic.cpp
diff options
context:
space:
mode:
authorKenneth Graunke <kenneth@whitecape.org>2014-02-28 23:15:49 -0800
committerKenneth Graunke <kenneth@whitecape.org>2014-03-02 13:35:03 -0800
commit3f37dd913f3461fb197e878739a269e639852667 (patch)
tree20dd1bc68dbf37d3f45666cbff893d6e1e292156 /src/glsl/opt_algebraic.cpp
parentecb71cfa66f4257661579a0afa5f9c56c7dbfce2 (diff)
downloadexternal_mesa3d-3f37dd913f3461fb197e878739a269e639852667.zip
external_mesa3d-3f37dd913f3461fb197e878739a269e639852667.tar.gz
external_mesa3d-3f37dd913f3461fb197e878739a269e639852667.tar.bz2
glsl: Fix broken LRP algebraic optimization.
opt_algebraic was translating lrp(x, 0, a) into add(x, -mul(x, a)). Unfortunately, this references "x" twice, which is invalid in the IR, leading to assertion failures in the validator. Normally, cloning IR solves this. However, "x" could actually be an arbitrary expression tree, so copying it could result in huge piles of wasted computation. This is why we avoid reusing subexpressions. Instead, transform it into mul(x, add(1.0, -a)), which is equivalent but doesn't need two references to "x". Fixes a regression since d5fa8a95621169, which isn't in any stable branches. Fixes 18 shaders in shader-db (bastion and yofrankie). Signed-off-by: Kenneth Graunke <kenneth@whitecape.org> Reviewed-by: Ian Romanick <ian.d.romanick@intel.com> Reviewed-by: Matt Turner <mattst88@gmail.com>
Diffstat (limited to 'src/glsl/opt_algebraic.cpp')
-rw-r--r--src/glsl/opt_algebraic.cpp4
1 files changed, 3 insertions, 1 deletions
diff --git a/src/glsl/opt_algebraic.cpp b/src/glsl/opt_algebraic.cpp
index 778638c..5c49a78 100644
--- a/src/glsl/opt_algebraic.cpp
+++ b/src/glsl/opt_algebraic.cpp
@@ -571,7 +571,9 @@ ir_algebraic_visitor::handle_expression(ir_expression *ir)
} else if (is_vec_zero(op_const[0])) {
return mul(ir->operands[1], ir->operands[2]);
} else if (is_vec_zero(op_const[1])) {
- return add(ir->operands[0], neg(mul(ir->operands[0], ir->operands[2])));
+ unsigned op2_components = ir->operands[2]->type->vector_elements;
+ ir_constant *one = new(mem_ctx) ir_constant(1.0f, op2_components);
+ return mul(ir->operands[0], add(one, neg(ir->operands[2])));
}
break;