Skip to content

Commit c03a0f1

Browse files
farzonlfarzonl
committed
instead of detecting the conflicts lets pick the smallest value for the alloca then keep the cast but change the input type.
1 parent de3024d commit c03a0f1

File tree

2 files changed

+116
-58
lines changed

2 files changed

+116
-58
lines changed

‎llvm/lib/Target/DirectX/DXILLegalizePass.cpp

Lines changed: 29 additions & 17 deletions
Original file line numberDiff line numberDiff line change
@@ -145,11 +145,23 @@ static void fixI8UseChain(Instruction &I,
145145
}
146146

147147
if (auto *Cast = dyn_cast<CastInst>(&I)) {
148-
149-
if (Cast->getSrcTy()->isIntegerTy(8)) {
150-
ToRemove.push_back(Cast);
151-
Cast->replaceAllUsesWith(ReplacedValues[Cast->getOperand(0)]);
148+
if (!Cast->getSrcTy()->isIntegerTy(8))
149+
return;
150+
151+
ToRemove.push_back(Cast);
152+
auto* Replacement =ReplacedValues[Cast->getOperand(0)];
153+
if (Cast->getType() == Replacement->getType()) {
154+
Cast->replaceAllUsesWith(Replacement);
155+
return;
152156
}
157+
Value* AdjustedCast = nullptr;
158+
if (Cast->getOpcode() == Instruction::ZExt)
159+
AdjustedCast = Builder.CreateZExtOrTrunc(Replacement, Cast->getType());
160+
if (Cast->getOpcode() == Instruction::SExt)
161+
AdjustedCast = Builder.CreateSExtOrTrunc(Replacement, Cast->getType());
162+
163+
if(AdjustedCast)
164+
Cast->replaceAllUsesWith(AdjustedCast);
153165
}
154166
}
155167

@@ -160,8 +172,9 @@ static void upcastI8AllocasAndUses(Instruction &I,
160172
if (!AI || !AI->getAllocatedType()->isIntegerTy(8))
161173
return;
162174

163-
std::optional<Type *> TargetType;
164-
bool Conflict = false;
175+
Type *SmallestType = nullptr;
176+
177+
// Gather all cast targets
165178
for (User *U : AI->users()) {
166179
auto *Load = dyn_cast<LoadInst>(U);
167180
if (!Load)
@@ -170,21 +183,20 @@ static void upcastI8AllocasAndUses(Instruction &I,
170183
auto *Cast = dyn_cast<CastInst>(LU);
171184
if (!Cast)
172185
continue;
173-
Type *T = Cast->getType();
174-
if (!TargetType)
175-
TargetType = T;
176-
177-
if (TargetType.value() != T) {
178-
Conflict = true;
179-
break;
180-
}
186+
Type *Ty = Cast->getType();
187+
if (!SmallestType ||
188+
Ty->getPrimitiveSizeInBits() < SmallestType->getPrimitiveSizeInBits())
189+
SmallestType = Ty;
181190
}
182191
}
183-
if (!TargetType || Conflict)
184-
return;
185192

193+
if (!SmallestType)
194+
return; // no valid casts found
195+
196+
// Replace alloca
186197
IRBuilder<> Builder(AI);
187-
AllocaInst *NewAlloca = Builder.CreateAlloca(TargetType.value());
198+
auto *NewAlloca =
199+
Builder.CreateAlloca(SmallestType);
188200
ReplacedValues[AI] = NewAlloca;
189201
ToRemove.push_back(AI);
190202
}

‎llvm/test/CodeGen/DirectX/legalize-i8-alloca.ll

Lines changed: 87 additions & 41 deletions
Original file line numberDiff line numberDiff line change
@@ -1,53 +1,99 @@
1+
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
12
; RUN: opt -S -passes='dxil-legalize' -mtriple=dxil-pc-shadermodel6.3-library %s | FileCheck %s
23

34
define void @const_i8_store() {
4-
%accum.i.flat = alloca [1 x i32], align 4
5-
%i = alloca i8, align 4
6-
store i8 1, ptr %i
7-
%i8.load = load i8, ptr %i
8-
%z = zext i8 %i8.load to i32
9-
%gep = getelementptr i32, ptr %accum.i.flat, i32 0
10-
store i32 %z, ptr %gep, align 4
11-
ret void
5+
; CHECK-LABEL: define void @const_i8_store() {
6+
; CHECK-NEXT: [[ACCUM_I_FLAT:%.*]] = alloca [1 x i32], align 4
7+
; CHECK-NEXT: [[TMP1:%.*]] = alloca i32, align 4
8+
; CHECK-NEXT: store i32 1, ptr [[TMP1]], align 4
9+
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
10+
; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[ACCUM_I_FLAT]], i32 0
11+
; CHECK-NEXT: store i32 [[TMP2]], ptr [[GEP]], align 4
12+
; CHECK-NEXT: ret void
13+
;
14+
%accum.i.flat = alloca [1 x i32], align 4
15+
%i = alloca i8, align 4
16+
store i8 1, ptr %i
17+
%i8.load = load i8, ptr %i
18+
%z = zext i8 %i8.load to i32
19+
%gep = getelementptr i32, ptr %accum.i.flat, i32 0
20+
store i32 %z, ptr %gep, align 4
21+
ret void
1222
}
1323

1424
define void @const_add_i8_store() {
15-
%accum.i.flat = alloca [1 x i32], align 4
16-
%i = alloca i8, align 4
17-
%add_i8 = add nsw i8 3, 1
18-
store i8 %add_i8, ptr %i
19-
%i8.load = load i8, ptr %i
20-
%z = zext i8 %i8.load to i32
21-
%gep = getelementptr i32, ptr %accum.i.flat, i32 0
22-
store i32 %z, ptr %gep, align 4
23-
ret void
25+
; CHECK-LABEL: define void @const_add_i8_store() {
26+
; CHECK-NEXT: [[ACCUM_I_FLAT:%.*]] = alloca [1 x i32], align 4
27+
; CHECK-NEXT: [[TMP1:%.*]] = alloca i32, align 4
28+
; CHECK-NEXT: store i32 4, ptr [[TMP1]], align 4
29+
; CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
30+
; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[ACCUM_I_FLAT]], i32 0
31+
; CHECK-NEXT: store i32 [[TMP2]], ptr [[GEP]], align 4
32+
; CHECK-NEXT: ret void
33+
;
34+
%accum.i.flat = alloca [1 x i32], align 4
35+
%i = alloca i8, align 4
36+
%add_i8 = add nsw i8 3, 1
37+
store i8 %add_i8, ptr %i
38+
%i8.load = load i8, ptr %i
39+
%z = zext i8 %i8.load to i32
40+
%gep = getelementptr i32, ptr %accum.i.flat, i32 0
41+
store i32 %z, ptr %gep, align 4
42+
ret void
2443
}
2544

2645
define void @var_i8_store(i1 %cmp.i8) {
27-
%accum.i.flat = alloca [1 x i32], align 4
28-
%i = alloca i8, align 4
29-
%select.i8 = select i1 %cmp.i8, i8 1, i8 2
30-
store i8 %select.i8, ptr %i
31-
%i8.load = load i8, ptr %i
32-
%z = zext i8 %i8.load to i32
33-
%gep = getelementptr i32, ptr %accum.i.flat, i32 0
34-
store i32 %z, ptr %gep, align 4
35-
ret void
46+
; CHECK-LABEL: define void @var_i8_store(
47+
; CHECK-SAME: i1 [[CMP_I8:%.*]]) {
48+
; CHECK-NEXT: [[ACCUM_I_FLAT:%.*]] = alloca [1 x i32], align 4
49+
; CHECK-NEXT: [[TMP1:%.*]] = alloca i32, align 4
50+
; CHECK-NEXT: [[TMP2:%.*]] = select i1 [[CMP_I8]], i32 1, i32 2
51+
; CHECK-NEXT: store i32 [[TMP2]], ptr [[TMP1]], align 4
52+
; CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4
53+
; CHECK-NEXT: [[GEP:%.*]] = getelementptr i32, ptr [[ACCUM_I_FLAT]], i32 0
54+
; CHECK-NEXT: store i32 [[TMP3]], ptr [[GEP]], align 4
55+
; CHECK-NEXT: ret void
56+
;
57+
%accum.i.flat = alloca [1 x i32], align 4
58+
%i = alloca i8, align 4
59+
%select.i8 = select i1 %cmp.i8, i8 1, i8 2
60+
store i8 %select.i8, ptr %i
61+
%i8.load = load i8, ptr %i
62+
%z = zext i8 %i8.load to i32
63+
%gep = getelementptr i32, ptr %accum.i.flat, i32 0
64+
store i32 %z, ptr %gep, align 4
65+
ret void
3666
}
3767

3868
define void @conflicting_cast(i1 %cmp.i8) {
39-
%accum.i.flat = alloca [2 x i32], align 4
40-
%i = alloca i8, align 4
41-
%select.i8 = select i1 %cmp.i8, i8 1, i8 2
42-
store i8 %select.i8, ptr %i
43-
%i8.load = load i8, ptr %i
44-
%z = zext i8 %i8.load to i16
45-
%gep1 = getelementptr i16, ptr %accum.i.flat, i32 0
46-
store i16 %z, ptr %gep1, align 2
47-
%gep2 = getelementptr i16, ptr %accum.i.flat, i32 1
48-
store i16 %z, ptr %gep2, align 2
49-
%z2 = zext i8 %i8.load to i32
50-
%gep3 = getelementptr i32, ptr %accum.i.flat, i32 1
51-
store i32 %z2, ptr %gep3, align 4
52-
ret void
53-
}
69+
; CHECK-LABEL: define void @conflicting_cast(
70+
; CHECK-SAME: i1 [[CMP_I8:%.*]]) {
71+
; CHECK-NEXT: [[ACCUM_I_FLAT:%.*]] = alloca [2 x i32], align 4
72+
; CHECK-NEXT: [[TMP1:%.*]] = alloca i16, align 2
73+
; CHECK-NEXT: [[TMP2:%.*]] = select i1 [[CMP_I8]], i32 1, i32 2
74+
; CHECK-NEXT: store i32 [[TMP2]], ptr [[TMP1]], align 4
75+
; CHECK-NEXT: [[TMP3:%.*]] = load i16, ptr [[TMP1]], align 2
76+
; CHECK-NEXT: [[GEP1:%.*]] = getelementptr i16, ptr [[ACCUM_I_FLAT]], i32 0
77+
; CHECK-NEXT: store i16 [[TMP3]], ptr [[GEP1]], align 2
78+
; CHECK-NEXT: [[GEP2:%.*]] = getelementptr i16, ptr [[ACCUM_I_FLAT]], i32 1
79+
; CHECK-NEXT: store i16 [[TMP3]], ptr [[GEP2]], align 2
80+
; CHECK-NEXT: [[TMP4:%.*]] = zext i16 [[TMP3]] to i32
81+
; CHECK-NEXT: [[GEP3:%.*]] = getelementptr i32, ptr [[ACCUM_I_FLAT]], i32 1
82+
; CHECK-NEXT: store i32 [[TMP4]], ptr [[GEP3]], align 4
83+
; CHECK-NEXT: ret void
84+
;
85+
%accum.i.flat = alloca [2 x i32], align 4
86+
%i = alloca i8, align 4
87+
%select.i8 = select i1 %cmp.i8, i8 1, i8 2
88+
store i8 %select.i8, ptr %i
89+
%i8.load = load i8, ptr %i
90+
%z = zext i8 %i8.load to i16
91+
%gep1 = getelementptr i16, ptr %accum.i.flat, i32 0
92+
store i16 %z, ptr %gep1, align 2
93+
%gep2 = getelementptr i16, ptr %accum.i.flat, i32 1
94+
store i16 %z, ptr %gep2, align 2
95+
%z2 = zext i8 %i8.load to i32
96+
%gep3 = getelementptr i32, ptr %accum.i.flat, i32 1
97+
store i32 %z2, ptr %gep3, align 4
98+
ret void
99+
}

0 commit comments

Comments
 (0)