[AVX512] Fix TERNLOG when folding broadcast

Patch to fix ternlog instructions with a folded
broadcast. The broadcast decorator, e.g. {1toX}, was missing.

Differential Revision: https://reviews.llvm.org/D38649

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@315122 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86InstrAVX512.td b/lib/Target/X86/X86InstrAVX512.td
index 0c52fa889e8112f48d7b9cc930942e1a7a141a99..fe2fe5378717848fa45edb3c1d13d1a2884ff4d6 100644 (file)
--- a/lib/Target/X86/X86InstrAVX512.td
+++ b/lib/Target/X86/X86InstrAVX512.td
@@ -9399,26 +9399,26 @@ multiclass avx512_ternlog<bits<8> opc, string OpcodeStr, SDNode OpNode,
                    (OpNode (X86VBroadcast (_.ScalarLdFrag addr:$src3)),
                     _.RC:$src2, _.RC:$src1, (i8 imm:$src4)),
                    _.RC:$src1)),
-            (!cast<Instruction>(NAME#_.ZSuffix#rmik) _.RC:$src1, _.KRCWM:$mask,
+            (!cast<Instruction>(NAME#_.ZSuffix#rmbik) _.RC:$src1, _.KRCWM:$mask,
              _.RC:$src2, addr:$src3, (VPTERNLOG321_imm8 imm:$src4))>;
   def : Pat<(_.VT (vselect _.KRCWM:$mask,
                    (OpNode _.RC:$src2, _.RC:$src1,
                     (X86VBroadcast (_.ScalarLdFrag addr:$src3)),
                     (i8 imm:$src4)), _.RC:$src1)),
-            (!cast<Instruction>(NAME#_.ZSuffix#rmik) _.RC:$src1, _.KRCWM:$mask,
+            (!cast<Instruction>(NAME#_.ZSuffix#rmbik) _.RC:$src1, _.KRCWM:$mask,
              _.RC:$src2, addr:$src3, (VPTERNLOG213_imm8 imm:$src4))>;
   def : Pat<(_.VT (vselect _.KRCWM:$mask,
                    (OpNode _.RC:$src2,
                     (X86VBroadcast (_.ScalarLdFrag addr:$src3)),
                     _.RC:$src1, (i8 imm:$src4)),
                    _.RC:$src1)),
-            (!cast<Instruction>(NAME#_.ZSuffix#rmik) _.RC:$src1, _.KRCWM:$mask,
+            (!cast<Instruction>(NAME#_.ZSuffix#rmbik) _.RC:$src1, _.KRCWM:$mask,
              _.RC:$src2, addr:$src3, (VPTERNLOG231_imm8 imm:$src4))>;
   def : Pat<(_.VT (vselect _.KRCWM:$mask,
                    (OpNode (X86VBroadcast (_.ScalarLdFrag addr:$src3)),
                     _.RC:$src1, _.RC:$src2, (i8 imm:$src4)),
                    _.RC:$src1)),
-            (!cast<Instruction>(NAME#_.ZSuffix#rmik) _.RC:$src1, _.KRCWM:$mask,
+            (!cast<Instruction>(NAME#_.ZSuffix#rmbik) _.RC:$src1, _.KRCWM:$mask,
              _.RC:$src2, addr:$src3, (VPTERNLOG312_imm8 imm:$src4))>;
 }
 

diff --git a/test/CodeGen/X86/avx512-vpternlog-commute.ll b/test/CodeGen/X86/avx512-vpternlog-commute.ll
index c917e0b17f1ce1d19fb2e13c70b053d7cf3fda75..5e1b28c1983412d980393162b1806b6b3a0f3e30 100644 (file)
--- a/test/CodeGen/X86/avx512-vpternlog-commute.ll
+++ b/test/CodeGen/X86/avx512-vpternlog-commute.ll
@@ -1008,7 +1008,7 @@ define <16 x i32> @vpternlog_v16i32_012_broadcast0_mask1(i32* %x0ptr, <16 x i32>
 ; CHECK-LABEL: vpternlog_v16i32_012_broadcast0_mask1:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovd %esi, %k1
-; CHECK-NEXT:    vpternlogd $92, (%rdi), %zmm1, %zmm0 {%k1}
+; CHECK-NEXT:    vpternlogd $92, (%rdi){1to16}, %zmm1, %zmm0 {%k1}
 ; CHECK-NEXT:    retq
   %x0scalar = load i32, i32* %x0ptr
   %vecinit.i = insertelement <16 x i32> undef, i32 %x0scalar, i32 0
@@ -1023,7 +1023,7 @@ define <16 x i32> @vpternlog_v16i32_012_broadcast0_mask2(i32* %x0ptr, <16 x i32>
 ; CHECK-LABEL: vpternlog_v16i32_012_broadcast0_mask2:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovd %esi, %k1
-; CHECK-NEXT:    vpternlogd $58, (%rdi), %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vpternlogd $58, (%rdi){1to16}, %zmm0, %zmm1 {%k1}
 ; CHECK-NEXT:    vmovdqa64 %zmm1, %zmm0
 ; CHECK-NEXT:    retq
   %x0scalar = load i32, i32* %x0ptr
@@ -1039,7 +1039,7 @@ define <16 x i32> @vpternlog_v16i32_012_broadcast1_mask2(<16 x i32> %x0, i32* %x
 ; CHECK-LABEL: vpternlog_v16i32_012_broadcast1_mask2:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovd %esi, %k1
-; CHECK-NEXT:    vpternlogd $46, (%rdi), %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vpternlogd $46, (%rdi){1to16}, %zmm0, %zmm1 {%k1}
 ; CHECK-NEXT:    vmovdqa64 %zmm1, %zmm0
 ; CHECK-NEXT:    retq
   %x1scalar = load i32, i32* %x1ptr
@@ -1055,7 +1055,7 @@ define <16 x i32> @vpternlog_v16i32_012_broadcast2_mask1(<16 x i32> %x0, <16 x i
 ; CHECK-LABEL: vpternlog_v16i32_012_broadcast2_mask1:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    kmovd %esi, %k1
-; CHECK-NEXT:    vpternlogd $78, (%rdi), %zmm0, %zmm1 {%k1}
+; CHECK-NEXT:    vpternlogd $78, (%rdi){1to16}, %zmm0, %zmm1 {%k1}
 ; CHECK-NEXT:    vmovdqa64 %zmm1, %zmm0
 ; CHECK-NEXT:    retq
   %x2scalar = load i32, i32* %x2ptr