From eeb4a0c7efaf65610ac8cded1280dfd139a75c9c Mon Sep 17 00:00:00 2001
From: Chandler Carruth <chandlerc@gmail.com>
Date: Wed, 1 Oct 2014 20:49:54 +0000
Subject: [PATCH] [x86] Add a 32-bit run to the sext test, and remove a sad
 vec_sext.ll test file.

This old test had a bunch of functions that were never even checked. =/
The only thing it really did was to make sure that we did something
reasonable in 32-bit mode with SSE4.1. Adding another run line to the
main vector-sext.ll test seems a better way to do that.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218810 91177308-0d34-0410-b5e6-96231b3b80d8
---
 test/CodeGen/X86/vec_sext.ll    |  80 --------------
 test/CodeGen/X86/vector-sext.ll | 181 ++++++++++++++++++++++++++++++++
 2 files changed, 181 insertions(+), 80 deletions(-)
 delete mode 100644 test/CodeGen/X86/vec_sext.ll

diff --git a/test/CodeGen/X86/vec_sext.ll b/test/CodeGen/X86/vec_sext.ll
deleted file mode 100644
index 0fd66a30939..00000000000
--- a/test/CodeGen/X86/vec_sext.ll
+++ /dev/null
@@ -1,80 +0,0 @@
-; RUN: llc < %s -march=x86-64 -mattr=+avx | FileCheck %s
-; RUN: llc < %s -march=x86 -mattr=+avx | FileCheck %s
-
-define<4 x i32> @func_16_32() {
-  %F = load <4 x i16>* undef
-  %G = sext <4 x i16> %F to <4 x i32>
-  %H = load <4 x i16>* undef
-  %Y = sext <4 x i16> %H to <4 x i32>
-  %T = add <4 x i32> %Y, %G
-  store <4 x i32>%T , <4 x i32>* undef
-  ret <4 x i32> %T
-}
-
-define<4 x i64> @func_16_64() {
-  %F = load <4 x i16>* undef
-  %G = sext <4 x i16> %F to <4 x i64>
-  %H = load <4 x i16>* undef
-  %Y = sext <4 x i16> %H to <4 x i64>
-  %T = xor <4 x i64> %Y, %G
-  store <4 x i64>%T , <4 x i64>* undef
-  ret <4 x i64> %T
-}
-
-define<4 x i64> @func_32_64() {
-  %F = load <4 x i32>* undef
-  %G = sext <4 x i32> %F to <4 x i64>
-  %H = load <4 x i32>* undef
-  %Y = sext <4 x i32> %H to <4 x i64>
-  %T = or <4 x i64> %Y, %G
-  ret <4 x i64> %T
-}
-
-define<4 x i16> @func_8_16() {
-  %F = load <4 x i8>* undef
-  %G = sext <4 x i8> %F to <4 x i16>
-  %H = load <4 x i8>* undef
-  %Y = sext <4 x i8> %H to <4 x i16>
-  %T = add <4 x i16> %Y, %G
-  ret <4 x i16> %T
-}
-
-define<4 x i32> @func_8_32() {
-  %F = load <4 x i8>* undef
-  %G = sext <4 x i8> %F to <4 x i32>
-  %H = load <4 x i8>* undef
-  %Y = sext <4 x i8> %H to <4 x i32>
-  %T = sub <4 x i32> %Y, %G
-  ret <4 x i32> %T
-}
-
-define<4 x i64> @func_8_64() {
-  %F = load <4 x i8>* undef
-  %G = sext <4 x i8> %F to <4 x i64>
-  %H = load <4 x i8>* undef
-  %Y = sext <4 x i8> %H to <4 x i64>
-  %T = add <4 x i64> %Y, %G
-  ret <4 x i64> %T
-}
-
-define<4 x i32> @const_16_32() {
-  %G = sext <4 x i16> <i16 0, i16 3, i16 8, i16 7> to <4 x i32>
-  ret <4 x i32> %G
-}
-
-define<4 x i64> @const_16_64() {
-  %G = sext <4 x i16> <i16 0, i16 3, i16 8, i16 7> to <4 x i64>
-  ret <4 x i64> %G
-}
-
-define <4 x i32> @sextload(<4 x i16>* %ptr) {
-; From PR20767 - make sure that we correctly use SSE4.1 to do sign extension
-; loads for both 32-bit and 64-bit x86 targets.
-; CHECK-LABEL: sextload:
-; CHECK:         vpmovsxwd {{.*}}, %xmm0
-; CHECK-NEXT:    ret
-entry:
-  %l = load<4 x i16>* %ptr
-  %m = sext<4 x i16> %l to <4 x i32>
-  ret <4 x i32> %m
-}
diff --git a/test/CodeGen/X86/vector-sext.ll b/test/CodeGen/X86/vector-sext.ll
index ead1b953dc5..eae5b297655 100644
--- a/test/CodeGen/X86/vector-sext.ll
+++ b/test/CodeGen/X86/vector-sext.ll
@@ -3,6 +3,9 @@
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+sse4.1 | FileCheck %s --check-prefix=SSE --check-prefix=SSE41
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+avx | FileCheck %s --check-prefix=AVX --check-prefix=AVX1
 ; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=x86-64 -mattr=+avx2 | FileCheck %s --check-prefix=AVX --check-prefix=AVX2
+;
+; Just one 32-bit run to make sure we do reasonable things there.
+; RUN: llc < %s -mtriple=i686-apple-darwin -mcpu=i686 -mattr=+sse4.1 | FileCheck %s --check-prefix=X32-SSE41
 
 define <8 x i32> @sext_8i16_to_8i32(<8 x i16> %A) nounwind uwtable readnone ssp {
 ; SSE2-LABEL: sext_8i16_to_8i32:
@@ -52,6 +55,17 @@ define <8 x i32> @sext_8i16_to_8i32(<8 x i16> %A) nounwind uwtable readnone ssp
 ; AVX2:       ## BB#0:
 ; AVX2-NEXT:    vpmovsxwd %xmm0, %ymm0
 ; AVX2-NEXT:    retq
+;
+; X32-SSE41-LABEL: sext_8i16_to_8i32:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movdqa %xmm0, %xmm1
+; X32-SSE41-NEXT:    pmovzxwd %xmm1, %xmm0
+; X32-SSE41-NEXT:    punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
+; X32-SSE41-NEXT:    pslld $16, %xmm1
+; X32-SSE41-NEXT:    psrad $16, %xmm1
+; X32-SSE41-NEXT:    pslld $16, %xmm0
+; X32-SSE41-NEXT:    psrad $16, %xmm0
+; X32-SSE41-NEXT:    retl
 
   %B = sext <8 x i16> %A to <8 x i32>
   ret <8 x i32>%B
@@ -137,6 +151,25 @@ define <4 x i64> @sext_4i32_to_4i64(<4 x i32> %A) nounwind uwtable readnone ssp
 ; AVX2:       ## BB#0:
 ; AVX2-NEXT:    vpmovsxdq %xmm0, %ymm0
 ; AVX2-NEXT:    retq
+;
+; X32-SSE41-LABEL: sext_4i32_to_4i64:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    pmovzxdq %xmm0, %xmm2
+; X32-SSE41-NEXT:    movd %xmm2, %eax
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pextrd $2, %xmm2, %ecx
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm2
+; X32-SSE41-NEXT:    sarl $31, %ecx
+; X32-SSE41-NEXT:    pinsrd $3, %ecx, %xmm2
+; X32-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,0,3,0]
+; X32-SSE41-NEXT:    movd %xmm1, %eax
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pextrd $2, %xmm1, %ecx
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm1
+; X32-SSE41-NEXT:    sarl $31, %ecx
+; X32-SSE41-NEXT:    pinsrd $3, %ecx, %xmm1
+; X32-SSE41-NEXT:    movdqa %xmm2, %xmm0
+; X32-SSE41-NEXT:    retl
 
   %B = sext <4 x i32> %A to <4 x i64>
   ret <4 x i64>%B
@@ -166,6 +199,12 @@ define <4 x i32> @load_sext_test1(<4 x i16> *%ptr) {
 ; AVX:       ## BB#0:
 ; AVX-NEXT:    vpmovsxwd (%rdi), %xmm0
 ; AVX-NEXT:    retq
+;
+; X32-SSE41-LABEL: load_sext_test1:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    pmovsxwd (%eax), %xmm0
+; X32-SSE41-NEXT:    retl
 
  %X = load <4 x i16>* %ptr
  %Y = sext <4 x i16> %X to <4 x i32>
@@ -205,6 +244,12 @@ define <4 x i32> @load_sext_test2(<4 x i8> *%ptr) {
 ; AVX:       ## BB#0:
 ; AVX-NEXT:    vpmovsxbd (%rdi), %xmm0
 ; AVX-NEXT:    retq
+;
+; X32-SSE41-LABEL: load_sext_test2:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    pmovsxbd (%eax), %xmm0
+; X32-SSE41-NEXT:    retl
  %X = load <4 x i8>* %ptr
  %Y = sext <4 x i8> %X to <4 x i32>
  ret <4 x i32>%Y
@@ -238,6 +283,12 @@ define <2 x i64> @load_sext_test3(<2 x i8> *%ptr) {
 ; AVX:       ## BB#0:
 ; AVX-NEXT:    vpmovsxbq (%rdi), %xmm0
 ; AVX-NEXT:    retq
+;
+; X32-SSE41-LABEL: load_sext_test3:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    pmovsxbq (%eax), %xmm0
+; X32-SSE41-NEXT:    retl
  %X = load <2 x i8>* %ptr
  %Y = sext <2 x i8> %X to <2 x i64>
  ret <2 x i64>%Y
@@ -271,6 +322,12 @@ define <2 x i64> @load_sext_test4(<2 x i16> *%ptr) {
 ; AVX:       ## BB#0:
 ; AVX-NEXT:    vpmovsxwq (%rdi), %xmm0
 ; AVX-NEXT:    retq
+;
+; X32-SSE41-LABEL: load_sext_test4:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    pmovsxwq (%eax), %xmm0
+; X32-SSE41-NEXT:    retl
  %X = load <2 x i16>* %ptr
  %Y = sext <2 x i16> %X to <2 x i64>
  ret <2 x i64>%Y
@@ -304,6 +361,12 @@ define <2 x i64> @load_sext_test5(<2 x i32> *%ptr) {
 ; AVX:       ## BB#0:
 ; AVX-NEXT:    vpmovsxdq (%rdi), %xmm0
 ; AVX-NEXT:    retq
+;
+; X32-SSE41-LABEL: load_sext_test5:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    pmovsxdq (%eax), %xmm0
+; X32-SSE41-NEXT:    retl
  %X = load <2 x i32>* %ptr
  %Y = sext <2 x i32> %X to <2 x i64>
  ret <2 x i64>%Y
@@ -333,6 +396,12 @@ define <8 x i16> @load_sext_test6(<8 x i8> *%ptr) {
 ; AVX:       ## BB#0:
 ; AVX-NEXT:    vpmovsxbw (%rdi), %xmm0
 ; AVX-NEXT:    retq
+;
+; X32-SSE41-LABEL: load_sext_test6:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    pmovsxbw (%eax), %xmm0
+; X32-SSE41-NEXT:    retl
  %X = load <8 x i8>* %ptr
  %Y = sext <8 x i8> %X to <8 x i16>
  ret <8 x i16>%Y
@@ -428,6 +497,27 @@ define <4 x i64> @sext_4i1_to_4i64(<4 x i1> %mask) {
 ; AVX2-NEXT:    vpsrad $31, %xmm0, %xmm0
 ; AVX2-NEXT:    vpmovsxdq %xmm0, %ymm0
 ; AVX2-NEXT:    retq
+;
+; X32-SSE41-LABEL: sext_4i1_to_4i64:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    pslld $31, %xmm0
+; X32-SSE41-NEXT:    psrad $31, %xmm0
+; X32-SSE41-NEXT:    pmovzxdq %xmm0, %xmm2
+; X32-SSE41-NEXT:    movd %xmm2, %eax
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pextrd $2, %xmm2, %ecx
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm2
+; X32-SSE41-NEXT:    sarl $31, %ecx
+; X32-SSE41-NEXT:    pinsrd $3, %ecx, %xmm2
+; X32-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,0,3,0]
+; X32-SSE41-NEXT:    movd %xmm1, %eax
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pextrd $2, %xmm1, %ecx
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm1
+; X32-SSE41-NEXT:    sarl $31, %ecx
+; X32-SSE41-NEXT:    pinsrd $3, %ecx, %xmm1
+; X32-SSE41-NEXT:    movdqa %xmm2, %xmm0
+; X32-SSE41-NEXT:    retl
   %extmask = sext <4 x i1> %mask to <4 x i64>
   ret <4 x i64> %extmask
 }
@@ -482,6 +572,18 @@ define <16 x i16> @sext_16i8_to_16i16(<16 x i8> *%ptr) {
 ; AVX2-NEXT:    vmovdqa (%rdi), %xmm0
 ; AVX2-NEXT:    vpmovsxbw %xmm0, %ymm0
 ; AVX2-NEXT:    retq
+;
+; X32-SSE41-LABEL: sext_16i8_to_16i16:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    movdqa (%eax), %xmm1
+; X32-SSE41-NEXT:    pmovzxbw %xmm1, %xmm0
+; X32-SSE41-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
+; X32-SSE41-NEXT:    psllw $8, %xmm1
+; X32-SSE41-NEXT:    psraw $8, %xmm1
+; X32-SSE41-NEXT:    psllw $8, %xmm0
+; X32-SSE41-NEXT:    psraw $8, %xmm0
+; X32-SSE41-NEXT:    retl
  %X = load <16 x i8>* %ptr
  %Y = sext <16 x i8> %X to <16 x i16>
  ret <16 x i16> %Y
@@ -577,6 +679,27 @@ define <4 x i64> @sext_4i8_to_4i64(<4 x i8> %mask) {
 ; AVX2-NEXT:    vpsrad $24, %xmm0, %xmm0
 ; AVX2-NEXT:    vpmovsxdq %xmm0, %ymm0
 ; AVX2-NEXT:    retq
+;
+; X32-SSE41-LABEL: sext_4i8_to_4i64:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    pslld $24, %xmm0
+; X32-SSE41-NEXT:    psrad $24, %xmm0
+; X32-SSE41-NEXT:    pmovzxdq %xmm0, %xmm2
+; X32-SSE41-NEXT:    movd %xmm2, %eax
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pextrd $2, %xmm2, %ecx
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm2
+; X32-SSE41-NEXT:    sarl $31, %ecx
+; X32-SSE41-NEXT:    pinsrd $3, %ecx, %xmm2
+; X32-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,0,3,0]
+; X32-SSE41-NEXT:    movd %xmm1, %eax
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pextrd $2, %xmm1, %ecx
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm1
+; X32-SSE41-NEXT:    sarl $31, %ecx
+; X32-SSE41-NEXT:    pinsrd $3, %ecx, %xmm1
+; X32-SSE41-NEXT:    movdqa %xmm2, %xmm0
+; X32-SSE41-NEXT:    retl
   %extmask = sext <4 x i8> %mask to <4 x i64>
   ret <4 x i64> %extmask
 }
@@ -672,6 +795,35 @@ define <4 x i64> @load_sext_4i8_to_4i64(<4 x i8> *%ptr) {
 ; AVX2:       ## BB#0:
 ; AVX2-NEXT:    vpmovsxbq (%rdi), %ymm0
 ; AVX2-NEXT:    retq
+;
+; X32-SSE41-LABEL: load_sext_4i8_to_4i64:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    movd (%eax), %xmm0
+; X32-SSE41-NEXT:    pmovzxbd %xmm0, %xmm1
+; X32-SSE41-NEXT:    pmovzxbq %xmm0, %xmm2
+; X32-SSE41-NEXT:    movd %xmm2, %eax
+; X32-SSE41-NEXT:    movsbl %al, %eax
+; X32-SSE41-NEXT:    movd %eax, %xmm0
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm0
+; X32-SSE41-NEXT:    pextrd $2, %xmm2, %eax
+; X32-SSE41-NEXT:    movsbl %al, %eax
+; X32-SSE41-NEXT:    pinsrd $2, %eax, %xmm0
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pinsrd $3, %eax, %xmm0
+; X32-SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,0,3,0]
+; X32-SSE41-NEXT:    movd %xmm2, %eax
+; X32-SSE41-NEXT:    movsbl %al, %eax
+; X32-SSE41-NEXT:    movd %eax, %xmm1
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm1
+; X32-SSE41-NEXT:    pextrd $2, %xmm2, %eax
+; X32-SSE41-NEXT:    movsbl %al, %eax
+; X32-SSE41-NEXT:    pinsrd $2, %eax, %xmm1
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pinsrd $3, %eax, %xmm1
+; X32-SSE41-NEXT:    retl
  %X = load <4 x i8>* %ptr
  %Y = sext <4 x i8> %X to <4 x i64>
  ret <4 x i64>%Y
@@ -761,6 +913,35 @@ define <4 x i64> @load_sext_4i16_to_4i64(<4 x i16> *%ptr) {
 ; AVX2:       ## BB#0:
 ; AVX2-NEXT:    vpmovsxwq (%rdi), %ymm0
 ; AVX2-NEXT:    retq
+;
+; X32-SSE41-LABEL: load_sext_4i16_to_4i64:
+; X32-SSE41:       ## BB#0:
+; X32-SSE41-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X32-SSE41-NEXT:    movsd (%eax), %xmm0
+; X32-SSE41-NEXT:    pmovzxwd %xmm0, %xmm1
+; X32-SSE41-NEXT:    pmovzxwq %xmm0, %xmm2
+; X32-SSE41-NEXT:    movd %xmm2, %eax
+; X32-SSE41-NEXT:    cwtl
+; X32-SSE41-NEXT:    movd %eax, %xmm0
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm0
+; X32-SSE41-NEXT:    pextrd $2, %xmm2, %eax
+; X32-SSE41-NEXT:    cwtl
+; X32-SSE41-NEXT:    pinsrd $2, %eax, %xmm0
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pinsrd $3, %eax, %xmm0
+; X32-SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[2,0,3,0]
+; X32-SSE41-NEXT:    movd %xmm2, %eax
+; X32-SSE41-NEXT:    cwtl
+; X32-SSE41-NEXT:    movd %eax, %xmm1
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pinsrd $1, %eax, %xmm1
+; X32-SSE41-NEXT:    pextrd $2, %xmm2, %eax
+; X32-SSE41-NEXT:    cwtl
+; X32-SSE41-NEXT:    pinsrd $2, %eax, %xmm1
+; X32-SSE41-NEXT:    sarl $31, %eax
+; X32-SSE41-NEXT:    pinsrd $3, %eax, %xmm1
+; X32-SSE41-NEXT:    retl
  %X = load <4 x i16>* %ptr
  %Y = sext <4 x i16> %X to <4 x i64>
  ret <4 x i64>%Y