[SLP] Fix PR35047: Fix default cost model for cast op in X86.

[ABataev]

Summary

The cost calculation for default case on X86 target does not always
follow correct wayt because of missing 4-th argument in
BaseT::getCastInstrCost() call. Added this missing parameter.

[RKSimon]

LGTM

lib/Target/X86/X86TargetTransformInfo.cpp

int X86TTIImpl::getCastInstrCost(unsigned Opcode, Type *Dst, Type *Src,
 
   if (ST->hasSSE2()) {
     if (const auto *Entry = ConvertCostTableLookup(SSE2ConversionTbl, ISD,
                                                    DstTy.getSimpleVT(),
                                                    SrcTy.getSimpleVT()))
       return Entry->Cost;
   }
 
-  return BaseT::getCastInstrCost(Opcode, Dst, Src);
+  return BaseT::getCastInstrCost(Opcode, Dst, Src, I);
 }
 
 int X86TTIImpl::getCmpSelInstrCost(unsigned Opcode, Type *ValTy, Type *CondTy,
                                    const Instruction *I) {
   // Legalize the type.
   std::pair<int, MVT> LT = TLI->getTypeLegalizationCost(DL, ValTy);
 
   MVT MTy = LT.second;

test/Transforms/SLPVectorizer/X86/cast.ll

 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
-; RUN: opt < %s -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -basicaa -slp-vectorizer -dce -S | FileCheck %s --check-prefix=CHECK --check-prefix=SSE
-; RUN: opt < %s -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7-avx -basicaa -slp-vectorizer -dce -S | FileCheck %s --check-prefix=CHECK --check-prefix=AVX
+; RUN: opt < %s -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7 -basicaa -slp-vectorizer -dce -S | FileCheck %s
+; RUN: opt < %s -mtriple=x86_64-apple-macosx10.8.0 -mcpu=corei7-avx -basicaa -slp-vectorizer -dce -S | FileCheck %s
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64-S128"
 
 ; int test_sext_4i8_to_4i32(int * restrict A, char * restrict B) {
 ;     A[0] = B[0];
 ;     A[1] = B[1];
 ;     A[2] = B[2];
 ;     A[3] = B[3];
 ; }
 
 define i32 @test_sext_4i8_to_4i32(i32* noalias nocapture %A, i8* noalias nocapture %B) {
 ; CHECK-LABEL: @test_sext_4i8_to_4i32(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = bitcast i8* %B to <4 x i8>*
+; CHECK-NEXT:    [[TMP0:%.*]] = bitcast i8* [[B:%.*]] to <4 x i8>*
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x i8>, <4 x i8>* [[TMP0]], align 1
 ; CHECK-NEXT:    [[TMP2:%.*]] = sext <4 x i8> [[TMP1]] to <4 x i32>
-; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* %A to <4 x i32>*
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[A:%.*]] to <4 x i32>*
 ; CHECK-NEXT:    store <4 x i32> [[TMP2]], <4 x i32>* [[TMP3]], align 4
 ; CHECK-NEXT:    ret i32 undef
 ;
 entry:
   %0 = load i8, i8* %B, align 1
   %conv = sext i8 %0 to i32
   store i32 %conv, i32* %A, align 4
   %arrayidx2 = getelementptr inbounds i8, i8* %B, i64 1
entry:
   %arrayidx10 = getelementptr inbounds i32, i32* %A, i64 3
   store i32 %conv9, i32* %arrayidx10, align 4
   ret i32 undef
 }
 
 define i32 @test_zext_4i16_to_4i32(i32* noalias nocapture %A, i16* noalias nocapture %B) {
 ; CHECK-LABEL: @test_zext_4i16_to_4i32(
 ; CHECK-NEXT:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = bitcast i16* %B to <4 x i16>*
+; CHECK-NEXT:    [[TMP0:%.*]] = bitcast i16* [[B:%.*]] to <4 x i16>*
 ; CHECK-NEXT:    [[TMP1:%.*]] = load <4 x i16>, <4 x i16>* [[TMP0]], align 1
 ; CHECK-NEXT:    [[TMP2:%.*]] = zext <4 x i16> [[TMP1]] to <4 x i32>
-; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* %A to <4 x i32>*
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i32* [[A:%.*]] to <4 x i32>*
 ; CHECK-NEXT:    store <4 x i32> [[TMP2]], <4 x i32>* [[TMP3]], align 4
 ; CHECK-NEXT:    ret i32 undef
 ;
 entry:
   %0 = load i16, i16* %B, align 1
   %conv = zext i16 %0 to i32
   store i32 %conv, i32* %A, align 4
   %arrayidx2 = getelementptr inbounds i16, i16* %B, i64 1
entry:
   %3 = load i16, i16* %arrayidx8, align 1
   %conv9 = zext i16 %3 to i32
   %arrayidx10 = getelementptr inbounds i32, i32* %A, i64 3
   store i32 %conv9, i32* %arrayidx10, align 4
   ret i32 undef
 }
 
 define i64 @test_sext_4i16_to_4i64(i64* noalias nocapture %A, i16* noalias nocapture %B) {
-; SSE-LABEL: @test_sext_4i16_to_4i64(
-; SSE-NEXT:  entry:
-; SSE-NEXT:    [[TMP0:%.*]] = bitcast i16* %B to <2 x i16>*
-; SSE-NEXT:    [[TMP1:%.*]] = load <2 x i16>, <2 x i16>* [[TMP0]], align 1
-; SSE-NEXT:    [[TMP2:%.*]] = sext <2 x i16> [[TMP1]] to <2 x i64>
-; SSE-NEXT:    [[TMP3:%.*]] = bitcast i64* %A to <2 x i64>*
-; SSE-NEXT:    store <2 x i64> [[TMP2]], <2 x i64>* [[TMP3]], align 4
-; SSE-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* %B, i64 2
-; SSE-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds i64, i64* %A, i64 2
-; SSE-NEXT:    [[TMP4:%.*]] = bitcast i16* [[ARRAYIDX5]] to <2 x i16>*
-; SSE-NEXT:    [[TMP5:%.*]] = load <2 x i16>, <2 x i16>* [[TMP4]], align 1
-; SSE-NEXT:    [[TMP6:%.*]] = sext <2 x i16> [[TMP5]] to <2 x i64>
-; SSE-NEXT:    [[TMP7:%.*]] = bitcast i64* [[ARRAYIDX7]] to <2 x i64>*
-; SSE-NEXT:    store <2 x i64> [[TMP6]], <2 x i64>* [[TMP7]], align 4
-; SSE-NEXT:    ret i64 undef
-;
-; AVX-LABEL: @test_sext_4i16_to_4i64(
-; AVX-NEXT:  entry:
-; AVX-NEXT:    [[TMP0:%.*]] = bitcast i16* %B to <4 x i16>*
-; AVX-NEXT:    [[TMP1:%.*]] = load <4 x i16>, <4 x i16>* [[TMP0]], align 1
-; AVX-NEXT:    [[TMP2:%.*]] = sext <4 x i16> [[TMP1]] to <4 x i64>
-; AVX-NEXT:    [[TMP3:%.*]] = bitcast i64* %A to <4 x i64>*
-; AVX-NEXT:    store <4 x i64> [[TMP2]], <4 x i64>* [[TMP3]], align 4
-; AVX-NEXT:    ret i64 undef
+; CHECK-LABEL: @test_sext_4i16_to_4i64(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = bitcast i16* [[B:%.*]] to <2 x i16>*
+; CHECK-NEXT:    [[TMP1:%.*]] = load <2 x i16>, <2 x i16>* [[TMP0]], align 1
+; CHECK-NEXT:    [[TMP2:%.*]] = sext <2 x i16> [[TMP1]] to <2 x i64>
+; CHECK-NEXT:    [[TMP3:%.*]] = bitcast i64* [[A:%.*]] to <2 x i64>*
+; CHECK-NEXT:    store <2 x i64> [[TMP2]], <2 x i64>* [[TMP3]], align 4
+; CHECK-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, i16* [[B]], i64 2
+; CHECK-NEXT:    [[ARRAYIDX7:%.*]] = getelementptr inbounds i64, i64* [[A]], i64 2
+; CHECK-NEXT:    [[TMP4:%.*]] = bitcast i16* [[ARRAYIDX5]] to <2 x i16>*
+; CHECK-NEXT:    [[TMP5:%.*]] = load <2 x i16>, <2 x i16>* [[TMP4]], align 1
+; CHECK-NEXT:    [[TMP6:%.*]] = sext <2 x i16> [[TMP5]] to <2 x i64>
+; CHECK-NEXT:    [[TMP7:%.*]] = bitcast i64* [[ARRAYIDX7]] to <2 x i64>*
+; CHECK-NEXT:    store <2 x i64> [[TMP6]], <2 x i64>* [[TMP7]], align 4
+; CHECK-NEXT:    ret i64 undef
 ;
 entry:
   %0 = load i16, i16* %B, align 1
   %conv = sext i16 %0 to i64
   store i64 %conv, i64* %A, align 4
   %arrayidx2 = getelementptr inbounds i16, i16* %B, i64 1
   %1 = load i16, i16* %arrayidx2, align 1
   %conv3 = sext i16 %1 to i64

test/Transforms/SLPVectorizer/X86/load-merge.ll

 ; CHECK-NEXT:    [[TMP0:%.*]] = load i8, i8* [[DATA:%.*]], align 1
 ; CHECK-NEXT:    [[CONV:%.*]] = zext i8 [[TMP0]] to i32
 ; CHECK-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, i8* [[DATA]], i64 1
 ; CHECK-NEXT:    [[TMP1:%.*]] = load i8, i8* [[ARRAYIDX1]], align 1
 ; CHECK-NEXT:    [[CONV2:%.*]] = zext i8 [[TMP1]] to i32
 ; CHECK-NEXT:    [[SHL3:%.*]] = shl nuw nsw i32 [[CONV2]], 8
 ; CHECK-NEXT:    [[OR:%.*]] = or i32 [[SHL3]], [[CONV]]
 ; CHECK-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[DATA]], i64 2
+; CHECK-NEXT:    [[TMP2:%.*]] = load i8, i8* [[ARRAYIDX4]], align 1
+; CHECK-NEXT:    [[CONV5:%.*]] = zext i8 [[TMP2]] to i32
+; CHECK-NEXT:    [[SHL6:%.*]] = shl nuw nsw i32 [[CONV5]], 16
+; CHECK-NEXT:    [[OR7:%.*]] = or i32 [[OR]], [[SHL6]]
 ; CHECK-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, i8* [[DATA]], i64 3
-; CHECK-NEXT:    [[TMP2:%.*]] = bitcast i8* [[ARRAYIDX4]] to <2 x i8>*
-; CHECK-NEXT:    [[TMP3:%.*]] = load <2 x i8>, <2 x i8>* [[TMP2]], align 1
-; CHECK-NEXT:    [[TMP4:%.*]] = zext <2 x i8> [[TMP3]] to <2 x i32>
-; CHECK-NEXT:    [[TMP5:%.*]] = shl nuw <2 x i32> [[TMP4]], <i32 16, i32 24>
-; CHECK-NEXT:    [[TMP6:%.*]] = extractelement <2 x i32> [[TMP5]], i32 0
-; CHECK-NEXT:    [[OR7:%.*]] = or i32 [[OR]], [[TMP6]]
-; CHECK-NEXT:    [[TMP7:%.*]] = extractelement <2 x i32> [[TMP5]], i32 1
-; CHECK-NEXT:    [[OR11:%.*]] = or i32 [[OR7]], [[TMP7]]
+; CHECK-NEXT:    [[TMP3:%.*]] = load i8, i8* [[ARRAYIDX8]], align 1
+; CHECK-NEXT:    [[CONV9:%.*]] = zext i8 [[TMP3]] to i32
+; CHECK-NEXT:    [[SHL10:%.*]] = shl nuw i32 [[CONV9]], 24
+; CHECK-NEXT:    [[OR11:%.*]] = or i32 [[OR7]], [[SHL10]]
 ; CHECK-NEXT:    ret i32 [[OR11]]
 ;
 entry:
   %0 = load i8, i8* %data, align 1
   %conv = zext i8 %0 to i32
   %arrayidx1 = getelementptr inbounds i8, i8* %data, i64 1
   %1 = load i8, i8* %arrayidx1, align 1
   %conv2 = zext i8 %1 to i32