[llvm] r261095 - [X86][SSE] Update pshufb mask test to use a real input instead of undef

Simon Pilgrim via llvm-commits llvm-commits at lists.llvm.org
Wed Feb 17 06:56:58 PST 2016


Author: rksimon
Date: Wed Feb 17 08:56:58 2016
New Revision: 261095

URL: http://llvm.org/viewvc/llvm-project?rev=261095&view=rev
Log:
[X86][SSE] Update pshufb mask test to use a real input instead of undef

We are getting better at combining constant pshufb masks - this test would've failed once we decode bitcasted masks as well.

Modified:
    llvm/trunk/test/CodeGen/X86/pshufb-mask-comments.ll

Modified: llvm/trunk/test/CodeGen/X86/pshufb-mask-comments.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/pshufb-mask-comments.ll?rev=261095&r1=261094&r2=261095&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/pshufb-mask-comments.ll (original)
+++ llvm/trunk/test/CodeGen/X86/pshufb-mask-comments.ll Wed Feb 17 08:56:58 2016
@@ -36,18 +36,18 @@ define <16 x i8> @test3(<16 x i8> %V) {
 
 ; Test that we won't crash when the constant was reused for another instruction.
 
-define <16 x i8> @test4(<2 x i64>* %V) {
+define <16 x i8> @test4(<16 x i8> %V, <2 x i64>* %P) {
 ; CHECK-LABEL: test4:
 ; CHECK:       # BB#0:
-; CHECK-NEXT:    movdqa {{.*#+}} xmm0 = [1084818905618843912,506097522914230528]
-; CHECK-NEXT:    movdqa %xmm0, (%rdi)
-; CHECK-NEXT:    pshufb %xmm0, %xmm0
+; CHECK-NEXT:    movdqa {{.*#+}} xmm1 = [1084818905618843912,506097522914230528]
+; CHECK-NEXT:    movdqa %xmm1, (%rdi)
+; CHECK-NEXT:    pshufb %xmm1, %xmm0
 ; CHECK-NEXT:    retq
   %1 = insertelement <2 x i64> undef, i64 1084818905618843912, i32 0
   %2 = insertelement <2 x i64>    %1, i64  506097522914230528, i32 1
-  store <2 x i64> %2, <2 x i64>* %V, align 16
+  store <2 x i64> %2, <2 x i64>* %P, align 16
   %3 = bitcast <2 x i64> %2 to <16 x i8>
-  %4 = tail call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> undef, <16 x i8> %3)
+  %4 = tail call <16 x i8> @llvm.x86.ssse3.pshuf.b.128(<16 x i8> %V, <16 x i8> %3)
   ret <16 x i8> %4
 }
 




More information about the llvm-commits mailing list