[X86] Add test coverage showing failure to concatenate matching rotate nodes

Author: RKSimon

llvm/test/CodeGen/X86/vector-shuffle-combining-avx512bwvl.ll

@@ -98,6 +98,36 @@ define <16 x i8> @combine_shuffle_vrotli_v4i32(<4 x i32> %a0) {
 }
 declare <4 x i32> @llvm.fshl.v4i32(<4 x i32>, <4 x i32>, <4 x i32>)
 
+
+define <8 x i32> @concat_vrotli_v4i32(<4 x i32> %a0, <4 x i32> %a1) {
+; CHECK-LABEL: concat_vrotli_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vprold $3, %xmm0, %xmm0
+; CHECK-NEXT:    vprold $3, %xmm1, %xmm1
+; CHECK-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; CHECK-NEXT:    ret{{[l|q]}}
+  %r0 = tail call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a0, <4 x i32> %a0, <4 x i32> splat (i32 3))
+  %r1 = tail call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a1, <4 x i32> %a1, <4 x i32> splat (i32 3))
+  %shuffle = shufflevector <4 x i32> %r0, <4 x i32> %r1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+  ret <8 x i32> %shuffle
+}
+
+define <8 x i32> @concat_vrotlv_v4i32(<4 x i32> %a0, <4 x i32> %a1, <8 x i32> %a2) {
+; CHECK-LABEL: concat_vrotlv_v4i32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vextracti128 $1, %ymm2, %xmm3
+; CHECK-NEXT:    vprolvd %xmm2, %xmm0, %xmm0
+; CHECK-NEXT:    vprolvd %xmm3, %xmm1, %xmm1
+; CHECK-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm0
+; CHECK-NEXT:    ret{{[l|q]}}
+  %lo = shufflevector <8 x i32> %a2, <8 x i32> poison, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
+  %hi = shufflevector <8 x i32> %a2, <8 x i32> poison, <4 x i32> <i32 4, i32 5, i32 6, i32 7>
+  %r0 = tail call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a0, <4 x i32> %a0, <4 x i32> %lo)
+  %r1 = tail call <4 x i32> @llvm.fshl.v4i32(<4 x i32> %a1, <4 x i32> %a1, <4 x i32> %hi)
+  %shuffle = shufflevector <4 x i32> %r0, <4 x i32> %r1, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
+  ret <8 x i32> %shuffle
+}
+
 define void @PR46178(ptr %0) {
 ; X86-LABEL: PR46178:
 ; X86:       # %bb.0:
@@ -174,11 +204,11 @@ define i64 @PR55050() {
 ; X86-NEXT:    xorl %edx, %edx
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    testb %dl, %dl
-; X86-NEXT:    jne .LBB10_2
+; X86-NEXT:    jne .LBB12_2
 ; X86-NEXT:  # %bb.1: # %if
 ; X86-NEXT:    xorl %eax, %eax
 ; X86-NEXT:    xorl %edx, %edx
-; X86-NEXT:  .LBB10_2: # %exit
+; X86-NEXT:  .LBB12_2: # %exit
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: PR55050: