[X86] Show failure to fold freeze(gfni()) -> gfni(freeze(),freeze()) for all gfni instructions

Author: RKSimon

llvm/test/CodeGen/X86/combine-gfni.ll

@@ -0,0 +1,104 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+gfni | FileCheck %s --check-prefixes=SSE
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+gfni,+avx | FileCheck %s --check-prefixes=AVX
+; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx512vl,+gfni,+avx512bw | FileCheck %s --check-prefixes=AVX512
+
+define <16 x i8> @gf2p8affineqb_freeze(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %a2) {
+; SSE-LABEL: gf2p8affineqb_freeze:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pxor %xmm3, %xmm3
+; SSE-NEXT:    pcmpgtb %xmm2, %xmm3
+; SSE-NEXT:    gf2p8affineqb $11, %xmm1, %xmm1
+; SSE-NEXT:    pand %xmm3, %xmm1
+; SSE-NEXT:    pandn %xmm0, %xmm3
+; SSE-NEXT:    por %xmm1, %xmm3
+; SSE-NEXT:    movdqa %xmm3, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: gf2p8affineqb_freeze:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vgf2p8affineqb $11, %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
+;
+; AVX512-LABEL: gf2p8affineqb_freeze:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpmovb2m %xmm2, %k1
+; AVX512-NEXT:    vgf2p8affineqb $11, %xmm1, %xmm1, %xmm1
+; AVX512-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
+; AVX512-NEXT:    retq
+  %i = icmp slt <16 x i8> %a2, zeroinitializer
+  %g = call <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8> %a1, <16 x i8> %a1, i8 11)
+  %f = freeze <16 x i8> %g
+  %r = select <16 x i1> %i, <16 x i8> %f, <16 x i8> %a0
+  ret <16 x i8> %r
+}
+
+define <16 x i8> @gf2p8affineinvqb_freeze(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %a2) {
+; SSE-LABEL: gf2p8affineinvqb_freeze:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pxor %xmm3, %xmm3
+; SSE-NEXT:    pcmpgtb %xmm2, %xmm3
+; SSE-NEXT:    gf2p8affineinvqb $11, %xmm1, %xmm1
+; SSE-NEXT:    pand %xmm3, %xmm1
+; SSE-NEXT:    pandn %xmm0, %xmm3
+; SSE-NEXT:    por %xmm1, %xmm3
+; SSE-NEXT:    movdqa %xmm3, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: gf2p8affineinvqb_freeze:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vgf2p8affineinvqb $11, %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
+;
+; AVX512-LABEL: gf2p8affineinvqb_freeze:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpmovb2m %xmm2, %k1
+; AVX512-NEXT:    vgf2p8affineinvqb $11, %xmm1, %xmm1, %xmm1
+; AVX512-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
+; AVX512-NEXT:    retq
+  %i = icmp slt <16 x i8> %a2, zeroinitializer
+  %g = call <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8> %a1, <16 x i8> %a1, i8 11)
+  %f = freeze <16 x i8> %g
+  %r = select <16 x i1> %i, <16 x i8> %f, <16 x i8> %a0
+  ret <16 x i8> %r
+}
+
+define <16 x i8> @gf2p8mulb_freeze(<16 x i8> %a0, <16 x i8> %a1, <16 x i8> %a2) {
+; SSE-LABEL: gf2p8mulb_freeze:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pxor %xmm3, %xmm3
+; SSE-NEXT:    pcmpgtb %xmm2, %xmm3
+; SSE-NEXT:    gf2p8mulb %xmm1, %xmm1
+; SSE-NEXT:    pand %xmm3, %xmm1
+; SSE-NEXT:    pandn %xmm0, %xmm3
+; SSE-NEXT:    por %xmm1, %xmm3
+; SSE-NEXT:    movdqa %xmm3, %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: gf2p8mulb_freeze:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vgf2p8mulb %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
+;
+; AVX512-LABEL: gf2p8mulb_freeze:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpmovb2m %xmm2, %k1
+; AVX512-NEXT:    vgf2p8mulb %xmm1, %xmm1, %xmm1
+; AVX512-NEXT:    vmovdqu8 %xmm1, %xmm0 {%k1}
+; AVX512-NEXT:    retq
+  %i = icmp slt <16 x i8> %a2, zeroinitializer
+  %g = call <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8> %a1, <16 x i8> %a1)
+  %f = freeze <16 x i8> %g
+  %r = select <16 x i1> %i, <16 x i8> %f, <16 x i8> %a0
+  ret <16 x i8> %r
+}
+
+declare <16 x i8> @llvm.x86.vgf2p8affineqb.128(<16 x i8>, <16 x i8>, i8)
+declare <32 x i8> @llvm.x86.vgf2p8affineqb.256(<32 x i8>, <32 x i8>, i8)
+declare <16 x i8> @llvm.x86.vgf2p8affineinvqb.128(<16 x i8>, <16 x i8>, i8)
+declare <32 x i8> @llvm.x86.vgf2p8affineinvqb.256(<32 x i8>, <32 x i8>, i8)
+declare <16 x i8> @llvm.x86.vgf2p8mulb.128(<16 x i8>, <16 x i8>)
+declare <32 x i8> @llvm.x86.vgf2p8mulb.256(<32 x i8>, <32 x i8>)