From 592fba5b4db731c3638e1521b074a610eed0e2bb Mon Sep 17 00:00:00 2001
From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Mon, 15 Jul 2019 15:43:04 +0000
Subject: [PATCH] [X86][SSE] Add PACKSS with zero shuffle masks.

This is an example of expansion due to D61129 - it should combine back to a PACKSS with a zero operand.

llvm-svn: 366077
---
 test/CodeGen/X86/packss.ll | 61 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 61 insertions(+)

diff --git a/test/CodeGen/X86/packss.ll b/test/CodeGen/X86/packss.ll
index 2e22501a986..5ac486281e7 100644
--- a/test/CodeGen/X86/packss.ll
+++ b/test/CodeGen/X86/packss.ll
@@ -264,3 +264,64 @@ define <8 x i16> @trunc_ashr_v4i64_demandedelts(<4 x i64> %a0) {
   %5 = trunc <8 x i32> %4 to <8 x i16>
   ret <8 x i16> %5
 }
+
+define <16 x i8> @packsswb_icmp_128_zero(<8 x i16> %a0) {
+; SSE-LABEL: packsswb_128_zero:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pxor %xmm1, %xmm1
+; SSE-NEXT:    pcmpeqw %xmm0, %xmm1
+; SSE-NEXT:    packsswb %xmm0, %xmm1
+; SSE-NEXT:    movq {{.*#+}} xmm0 = xmm1[0],zero
+; SSE-NEXT:    ret{{[l|q]}}
+;
+; AVX-LABEL: packsswb_128_zero:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpeqw %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],zero,zero,zero,zero,zero,zero,zero,zero
+; AVX-NEXT:    ret{{[l|q]}}
+  %1 = icmp eq <8 x i16> %a0, zeroinitializer
+  %2 = sext <8 x i1> %1 to <8 x i8>
+  %3 = shufflevector <8 x i8> %2, <8 x i8> zeroinitializer, <16 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
+  ret <16 x i8> %3
+}
+
+define <32 x i8> @packsswb_icmp_zero_256(<16 x i16> %a0) {
+; SSE-LABEL: packsswb_zero_256:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pxor %xmm2, %xmm2
+; SSE-NEXT:    pcmpeqw %xmm2, %xmm1
+; SSE-NEXT:    pcmpeqw %xmm2, %xmm0
+; SSE-NEXT:    packsswb %xmm1, %xmm0
+; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,1],xmm0[2,3]
+; SSE-NEXT:    pslldq {{.*#+}} xmm0 = zero,zero,zero,zero,zero,zero,zero,zero,xmm0[0,1,2,3,4,5,6,7]
+; SSE-NEXT:    movaps %xmm2, %xmm1
+; SSE-NEXT:    ret{{[l|q]}}
+;
+; AVX1-LABEL: packsswb_zero_256:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm1
+; AVX1-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
+; AVX1-NEXT:    ret{{[l|q]}}
+;
+; AVX2-LABEL: packsswb_zero_256:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vpcmpeqw %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm2
+; AVX2-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,0,2,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
+; AVX2-NEXT:    ret{{[l|q]}}
+  %1 = icmp eq <16 x i16> %a0, zeroinitializer
+  %2 = sext <16 x i1> %1 to <16 x i8>
+  %3 = shufflevector <16 x i8> zeroinitializer, <16 x i8> %2, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
+  ret <32 x i8> %3
+}