CodeGen peephole: fold redundant phys reg copies

[oota-llvm.git] / test / CodeGen / X86 / 2011-10-21-widen-cmp.ll

diff --git a/test/CodeGen/X86/2011-10-21-widen-cmp.ll b/test/CodeGen/X86/2011-10-21-widen-cmp.ll
index aa1a60026f43909e50fe4038da29eb80811a15bf..cb4648c382f79eda383d742d0cc714df570617a5 100644 (file)
--- a/test/CodeGen/X86/2011-10-21-widen-cmp.ll
+++ b/test/CodeGen/X86/2011-10-21-widen-cmp.ll
@@ -1,15 +1,23 @@
-; RUN: llc < %s -march=x86-64 -mcpu=corei7 | FileCheck %s
-
-target triple = "x86_64-unknown-linux-gnu"
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 | FileCheck %s
 
 ; Check that a <4 x float> compare is generated and that we are
 ; not stuck in an endless loop.
 
-; CHECK: cmp_2_floats
-; CHECK: cmpordps
-; CHECK: ret
-
 define void @cmp_2_floats() {
+; CHECK-LABEL: cmp_2_floats:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cmpordps %xmm0, %xmm0
+; CHECK-NEXT:    pmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
+; CHECK-NEXT:    psllq $32, %xmm0
+; CHECK-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
+; CHECK-NEXT:    psrad $31, %xmm0
+; CHECK-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5],xmm0[6,7]
+; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; CHECK-NEXT:    pslld $31, %xmm0
+; CHECK-NEXT:    blendvps %xmm0, %xmm0
+; CHECK-NEXT:    movlps %xmm0, (%rax)
+; CHECK-NEXT:    retq
 entry:
   %0 = fcmp oeq <2 x float> undef, undef
   %1 = select <2 x i1> %0, <2 x float> undef, <2 x float> undef
@@ -17,14 +25,31 @@ entry:
   ret void
 }
 
-; CHECK: cmp_2_doubles
-; CHECK: cmpordpd
-; CHECK: blendvpd
-; CHECK: ret
 define void @cmp_2_doubles() {
+; CHECK-LABEL: cmp_2_doubles:
+; CHECK:       # BB#0: # %entry
+; CHECK-NEXT:    cmpordpd %xmm0, %xmm0
+; CHECK-NEXT:    blendvpd %xmm0, %xmm0
+; CHECK-NEXT:    movapd %xmm0, (%rax)
+; CHECK-NEXT:    retq
 entry:
   %0 = fcmp oeq <2 x double> undef, undef
   %1 = select <2 x i1> %0, <2 x double> undef, <2 x double> undef
   store <2 x double> %1, <2 x double>* undef
   ret void
 }
+
+define void @mp_11193(<8 x float> * nocapture %aFOO, <8 x float>* nocapture %RET) nounwind {
+; CHECK-LABEL: mp_11193:
+; CHECK:       # BB#0: # %allocas
+; CHECK-NEXT:    movl $-1082130432, (%rsi) # imm = 0xFFFFFFFFBF800000
+; CHECK-NEXT:    retq
+allocas:
+  %bincmp = fcmp olt <8 x float> <float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 1.000000e+00, float 9.000000e+00, float 1.000000e+00, float 9.000000e+00, float 1.000000e+00> , <float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00>
+  %t = extractelement <8 x i1> %bincmp, i32 0
+  %ft = sitofp i1 %t to float
+  %pp = bitcast <8 x float>* %RET to float*
+  store float %ft, float* %pp
+  ret void
+}
+