[X86] Add test showing failure to remove sign splats from PACKSS intrinsics

[RKSimon]

PACKSS intrinsic calls are only expanded to X86ISD::PACKSS nodes during legalisation, after which time we fail to handle cases where ASHR sign splats (now lowered to X86ISD::VSRAI) are unnecessary/

Add additional example of FREEZE(PACKSS()) as that's an issue as well.

[llvmbot]

@llvm/pr-subscribers-backend-x86

Author: Simon Pilgrim (RKSimon)

Changes

PACKSS intrinsic calls are only expanded to X86ISD::PACKSS nodes during legalisation, after which time we fail to handle cases where ASHR sign splats (now lowered to X86ISD::VSRAI) are unnecessary/

Add additional example of FREEZE(PACKSS()) as that's an issue as well.

---

Full diff: https://github.com/llvm/llvm-project/pull/161518.diff

1 Files Affected:

• (added) llvm/test/CodeGen/X86/combine-pack.ll (+55)

diff --git a/llvm/test/CodeGen/X86/combine-pack.ll b/llvm/test/CodeGen/X86/combine-pack.ll
new file mode 100644
index 0000000000000..9e740b04073e0
--- /dev/null
+++ b/llvm/test/CodeGen/X86/combine-pack.ll
@@ -0,0 +1,55 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=SSE
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=AVX
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=AVX
+
+declare <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32>, <4 x i32>)
+
+; TODO: Failure to remove unnecessary signsplat
+define <8 x i16> @combine_packss_v4i32_signsplat(<4 x i32> %a0, <4 x i32> %a1) {
+; SSE-LABEL: combine_packss_v4i32_signsplat:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pcmpgtd %xmm1, %xmm0
+; SSE-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    psraw $15, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: combine_packss_v4i32_signsplat:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpcmpgtd %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpackssdw %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpsraw $15, %xmm0, %xmm0
+; AVX-NEXT:    retq
+  %cmp = icmp sgt <4 x i32> %a0, %a1
+  %ext = sext <4 x i1> %cmp to <4 x i32>
+  %pack = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %ext, <4 x i32> splat (i32 -1))
+  %signsplat = ashr <8 x i16> %pack, splat (i16 15)
+  ret <8 x i16> %signsplat
+}
+
+; TODO: Failure to remove unnecessary signsplat through freeze
+define <8 x i16> @combine_packss_v4i32_freeze_signsplat(<4 x i32> %a0, <4 x i32> %a1) {
+; SSE-LABEL: combine_packss_v4i32_freeze_signsplat:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pcmpgtd %xmm1, %xmm0
+; SSE-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    psraw $15, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: combine_packss_v4i32_freeze_signsplat:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpcmpgtd %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpackssdw %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpsraw $15, %xmm0, %xmm0
+; AVX-NEXT:    retq
+  %cmp = icmp sgt <4 x i32> %a0, %a1
+  %ext = sext <4 x i1> %cmp to <4 x i32>
+  %pack = call <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32> %ext, <4 x i32> splat (i32 -1))
+  %freeze = freeze <8 x i16> %pack
+  %signsplat = ashr <8 x i16> %freeze, splat (i16 15)
+  ret <8 x i16> %signsplat
+}