[X86][SSE] Add bool vector extraction test cases from PR15215

llvm-svn: 314813

1 files changed, 134 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/X86/bool-vector.ll b/llvm/test/CodeGen/X86/bool-vector.ll
new file mode 100644
index 00000000000..0ee5c066be8
--- /dev/null
+++ b/llvm/test/CodeGen/X86/bool-vector.ll
@@ -0,0 +1,134 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=X32 --check-prefix=X32-SSE2
+; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X32 --check-prefix=X32-AVX2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=X64 --check-prefix=X64-SSE2
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX2
+
+define i32 @PR15215_bad(<4 x i32> %input) {
+; X32-SSE2-LABEL: PR15215_bad:
+; X32-SSE2:       # BB#0: # %entry
+; X32-SSE2-NEXT:    pslld $31, %xmm0
+; X32-SSE2-NEXT:    psrad $31, %xmm0
+; X32-SSE2-NEXT:    movmskps %xmm0, %eax
+; X32-SSE2-NEXT:    retl
+;
+; X32-AVX2-LABEL: PR15215_bad:
+; X32-AVX2:       # BB#0: # %entry
+; X32-AVX2-NEXT:    vpslld $31, %xmm0, %xmm0
+; X32-AVX2-NEXT:    vpsrad $31, %xmm0, %xmm0
+; X32-AVX2-NEXT:    vmovmskps %xmm0, %eax
+; X32-AVX2-NEXT:    retl
+;
+; X64-SSE2-LABEL: PR15215_bad:
+; X64-SSE2:       # BB#0: # %entry
+; X64-SSE2-NEXT:    pslld $31, %xmm0
+; X64-SSE2-NEXT:    psrad $31, %xmm0
+; X64-SSE2-NEXT:    movmskps %xmm0, %eax
+; X64-SSE2-NEXT:    retq
+;
+; X64-AVX2-LABEL: PR15215_bad:
+; X64-AVX2:       # BB#0: # %entry
+; X64-AVX2-NEXT:    vpslld $31, %xmm0, %xmm0
+; X64-AVX2-NEXT:    vpsrad $31, %xmm0, %xmm0
+; X64-AVX2-NEXT:    vmovmskps %xmm0, %eax
+; X64-AVX2-NEXT:    retq
+entry:
+  %0 = trunc <4 x i32> %input to <4 x i1>
+  %1 = bitcast <4 x i1> %0 to i4
+  %2 = zext i4 %1 to i32
+  ret i32 %2
+}
+
+define i32 @PR15215_good(<4 x i32> %input) {
+; X32-SSE2-LABEL: PR15215_good:
+; X32-SSE2:       # BB#0: # %entry
+; X32-SSE2-NEXT:    pushl %esi
+; X32-SSE2-NEXT:  .Lcfi0:
+; X32-SSE2-NEXT:    .cfi_def_cfa_offset 8
+; X32-SSE2-NEXT:  .Lcfi1:
+; X32-SSE2-NEXT:    .cfi_offset %esi, -8
+; X32-SSE2-NEXT:    movd %xmm0, %eax
+; X32-SSE2-NEXT:    andl $1, %eax
+; X32-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; X32-SSE2-NEXT:    movd %xmm1, %ecx
+; X32-SSE2-NEXT:    andl $1, %ecx
+; X32-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
+; X32-SSE2-NEXT:    movd %xmm1, %edx
+; X32-SSE2-NEXT:    andl $1, %edx
+; X32-SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
+; X32-SSE2-NEXT:    movd %xmm0, %esi
+; X32-SSE2-NEXT:    andl $1, %esi
+; X32-SSE2-NEXT:    leal (%eax,%ecx,2), %eax
+; X32-SSE2-NEXT:    leal (%eax,%edx,4), %eax
+; X32-SSE2-NEXT:    leal (%eax,%esi,8), %eax
+; X32-SSE2-NEXT:    popl %esi
+; X32-SSE2-NEXT:    retl
+;
+; X32-AVX2-LABEL: PR15215_good:
+; X32-AVX2:       # BB#0: # %entry
+; X32-AVX2-NEXT:    pushl %esi
+; X32-AVX2-NEXT:  .Lcfi0:
+; X32-AVX2-NEXT:    .cfi_def_cfa_offset 8
+; X32-AVX2-NEXT:  .Lcfi1:
+; X32-AVX2-NEXT:    .cfi_offset %esi, -8
+; X32-AVX2-NEXT:    vmovd %xmm0, %eax
+; X32-AVX2-NEXT:    andl $1, %eax
+; X32-AVX2-NEXT:    vpextrd $1, %xmm0, %ecx
+; X32-AVX2-NEXT:    andl $1, %ecx
+; X32-AVX2-NEXT:    vpextrd $2, %xmm0, %edx
+; X32-AVX2-NEXT:    andl $1, %edx
+; X32-AVX2-NEXT:    vpextrd $3, %xmm0, %esi
+; X32-AVX2-NEXT:    andl $1, %esi
+; X32-AVX2-NEXT:    leal (%eax,%ecx,2), %eax
+; X32-AVX2-NEXT:    leal (%eax,%edx,4), %eax
+; X32-AVX2-NEXT:    leal (%eax,%esi,8), %eax
+; X32-AVX2-NEXT:    popl %esi
+; X32-AVX2-NEXT:    retl
+;
+; X64-SSE2-LABEL: PR15215_good:
+; X64-SSE2:       # BB#0: # %entry
+; X64-SSE2-NEXT:    movd %xmm0, %eax
+; X64-SSE2-NEXT:    andl $1, %eax
+; X64-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; X64-SSE2-NEXT:    movd %xmm1, %ecx
+; X64-SSE2-NEXT:    andl $1, %ecx
+; X64-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
+; X64-SSE2-NEXT:    movd %xmm1, %edx
+; X64-SSE2-NEXT:    andl $1, %edx
+; X64-SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
+; X64-SSE2-NEXT:    movd %xmm0, %esi
+; X64-SSE2-NEXT:    andl $1, %esi
+; X64-SSE2-NEXT:    leal (%rax,%rcx,2), %eax
+; X64-SSE2-NEXT:    leal (%rax,%rdx,4), %eax
+; X64-SSE2-NEXT:    leal (%rax,%rsi,8), %eax
+; X64-SSE2-NEXT:    retq
+;
+; X64-AVX2-LABEL: PR15215_good:
+; X64-AVX2:       # BB#0: # %entry
+; X64-AVX2-NEXT:    vmovd %xmm0, %eax
+; X64-AVX2-NEXT:    andl $1, %eax
+; X64-AVX2-NEXT:    vpextrd $1, %xmm0, %ecx
+; X64-AVX2-NEXT:    andl $1, %ecx
+; X64-AVX2-NEXT:    vpextrd $2, %xmm0, %edx
+; X64-AVX2-NEXT:    andl $1, %edx
+; X64-AVX2-NEXT:    vpextrd $3, %xmm0, %esi
+; X64-AVX2-NEXT:    andl $1, %esi
+; X64-AVX2-NEXT:    leal (%rax,%rcx,2), %eax
+; X64-AVX2-NEXT:    leal (%rax,%rdx,4), %eax
+; X64-AVX2-NEXT:    leal (%rax,%rsi,8), %eax
+; X64-AVX2-NEXT:    retq
+entry:
+  %0 = trunc <4 x i32> %input to <4 x i1>
+  %1 = extractelement <4 x i1> %0, i32 0
+  %e1 = select i1 %1, i32 1, i32 0
+  %2 = extractelement <4 x i1> %0, i32 1
+  %e2 = select i1 %2, i32 2, i32 0
+  %3 = extractelement <4 x i1> %0, i32 2
+  %e3 = select i1 %3, i32 4, i32 0
+  %4 = extractelement <4 x i1> %0, i32 3
+  %e4 = select i1 %4, i32 8, i32 0
+  %5 = or i32 %e1, %e2
+  %6 = or i32 %5, %e3
+  %7 = or i32 %6, %e4
+  ret i32 %7
+}