Two fixes to the vpermilvar optimization.

The instcomine logic to handle vpermilvar's pd and 256 variants was incorrect.
The _256 variants have indexes into the individual 128 bit lanes and in all
cases it also has to mask out unused bits.

llvm-svn: 207577

1 files changed, 4 insertions, 4 deletions

diff --git a/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll b/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
index a3e978141ad..9ae024181bc 100644
--- a/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
+++ b/llvm/test/Transforms/InstCombine/vec_demanded_elts.ll
@@ -318,7 +318,7 @@ define <4 x float> @test_vpermilvar_ps(<4 x float> %v) {
 declare <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float>, <8 x i32>)
 define <8 x float> @test_vpermilvar_ps_256(<8 x float> %v) {
 ; CHECK-LABEL: @test_vpermilvar_ps_256(
-; CHECK: shufflevector <8 x float> %v, <8 x float> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
+; CHECK: shufflevector <8 x float> %v, <8 x float> undef, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 7, i32 6, i32 5, i32 4>
   %a = tail call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> %v, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>)
   ret <8 x float> %a
 }
@@ -327,15 +327,15 @@ declare <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double>, <2 x i32>)
 define <2 x double> @test_vpermilvar_pd(<2 x double> %v) {
 ; CHECK-LABEL: @test_vpermilvar_pd(
 ; CHECK: shufflevector <2 x double> %v, <2 x double> undef, <2 x i32> <i32 1, i32 0>
-  %a = tail call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> %v, <2 x i32> <i32 1, i32 0>)
+  %a = tail call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> %v, <2 x i32> <i32 2, i32 0>)
   ret <2 x double> %a
 }
 
 declare <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double>, <4 x i32>)
 define <4 x double> @test_vpermilvar_pd_256(<4 x double> %v) {
 ; CHECK-LABEL: @test_vpermilvar_pd_256(
-; CHECK: shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
-  %a = tail call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> %v, <4 x i32> <i32 3, i32 2, i32 1, i32 0>)
+; CHECK: shufflevector <4 x double> %v, <4 x double> undef, <4 x i32> <i32 1, i32 0, i32 3, i32 2>
+  %a = tail call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> %v, <4 x i32> <i32 3, i32 1, i32 2, i32 0>)
   ret <4 x double> %a
 }