[llvm] r298774 - [X86][SSE] Add extra computeNumSignBits test case for D31311.

[Simon Pilgrim via llvm-commits]

Author: rksimon
Date: Sat Mar 25 10:43:36 2017
New Revision: 298774

URL: http://llvm.org/viewvc/llvm-project?rev=298774&view=rev
Log:
[X86][SSE] Add extra computeNumSignBits test case for D31311.

Modified:
    llvm/trunk/test/CodeGen/X86/known-signbits-vector.ll

Modified: llvm/trunk/test/CodeGen/X86/known-signbits-vector.ll
URL: http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/X86/known-signbits-vector.ll?rev=298774&r1=298773&r2=298774&view=diff
==============================================================================
--- llvm/trunk/test/CodeGen/X86/known-signbits-vector.ll (original)
+++ llvm/trunk/test/CodeGen/X86/known-signbits-vector.ll Sat Mar 25 10:43:36 2017
@@ -96,3 +96,50 @@ define float @signbits_ashr_extract_sito
   %3 = sitofp i64 %2 to float
   ret float %3
 }
+
+define float @signbits_ashr_insert_ashr_extract_sitofp(i64 %a0, i64 %a1) nounwind {
+; X32-LABEL: signbits_ashr_insert_ashr_extract_sitofp:
+; X32:       # BB#0:
+; X32-NEXT:    pushl %ebp
+; X32-NEXT:    movl %esp, %ebp
+; X32-NEXT:    andl $-8, %esp
+; X32-NEXT:    subl $16, %esp
+; X32-NEXT:    movl 8(%ebp), %eax
+; X32-NEXT:    movl 12(%ebp), %ecx
+; X32-NEXT:    shrdl $30, %ecx, %eax
+; X32-NEXT:    sarl $30, %ecx
+; X32-NEXT:    vmovd %eax, %xmm0
+; X32-NEXT:    vpinsrd $1, %ecx, %xmm0, %xmm0
+; X32-NEXT:    vpinsrd $2, 16(%ebp), %xmm0, %xmm0
+; X32-NEXT:    vpinsrd $3, 20(%ebp), %xmm0, %xmm0
+; X32-NEXT:    vpsrad $3, %xmm0, %xmm1
+; X32-NEXT:    vpsrlq $3, %xmm0, %xmm0
+; X32-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
+; X32-NEXT:    vmovq %xmm0, {{[0-9]+}}(%esp)
+; X32-NEXT:    fildll {{[0-9]+}}(%esp)
+; X32-NEXT:    fstps {{[0-9]+}}(%esp)
+; X32-NEXT:    flds {{[0-9]+}}(%esp)
+; X32-NEXT:    movl %ebp, %esp
+; X32-NEXT:    popl %ebp
+; X32-NEXT:    retl
+;
+; X64-LABEL: signbits_ashr_insert_ashr_extract_sitofp:
+; X64:       # BB#0:
+; X64-NEXT:    sarq $30, %rdi
+; X64-NEXT:    vmovq %rsi, %xmm0
+; X64-NEXT:    vmovq %rdi, %xmm1
+; X64-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; X64-NEXT:    vpsrad $3, %xmm0, %xmm1
+; X64-NEXT:    vpsrlq $3, %xmm0, %xmm0
+; X64-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
+; X64-NEXT:    vmovq %xmm0, %rax
+; X64-NEXT:    vcvtsi2ssq %rax, %xmm2, %xmm0
+; X64-NEXT:    retq
+  %1 = ashr i64 %a0, 30
+  %2 = insertelement <2 x i64> undef, i64 %1, i32 0
+  %3 = insertelement <2 x i64> %2, i64 %a1, i32 1
+  %4 = ashr <2 x i64> %3, <i64 3, i64 3>
+  %5 = extractelement <2 x i64> %4, i32 0
+  %6 = sitofp i64 %5 to float
+  ret float %6
+}