mirror of
https://github.com/llvm/llvm-project.git
synced 2025-04-25 21:36:05 +00:00
7eca38ce76
Add nuw attribute to inbounds GEPs where the expression used to form the GEP is an addition of unsigned indices. Relands #105496, which was reverted because it exposed a miscompilation arising from #98608. This is now fixed by #106512.
127 lines
5.3 KiB
C++
127 lines
5.3 KiB
C++
// RUN: %clang_cc1 -std=c++11 -triple x86_64-apple-darwin %s -emit-llvm -o - | FileCheck -check-prefixes=X64,CHECK %s
|
|
// RUN: %clang_cc1 -std=c++11 -triple amdgcn %s -emit-llvm -o - | FileCheck -check-prefixes=AMDGCN,CHECK %s
|
|
|
|
template<typename T>
|
|
struct S {
|
|
static int n;
|
|
};
|
|
template<typename T> int S<T>::n = 5;
|
|
|
|
int f() {
|
|
// Make sure that the reference here is enough to trigger the instantiation of
|
|
// the static data member.
|
|
// CHECK: @_ZN1SIiE1nE = linkonce_odr{{.*}} global i32 5
|
|
int a[S<int>::n];
|
|
return sizeof a;
|
|
}
|
|
|
|
void test0(void *array, int n) {
|
|
// CHECK-LABEL: define{{.*}} void @_Z5test0Pvi(
|
|
// AMDGCN: [[ARRAY0:%.*]] = alloca ptr, align 8, addrspace(5)
|
|
// AMDGCN-NEXT: [[N0:%.*]] = alloca i32, align 4, addrspace(5)
|
|
// AMDGCN-NEXT: [[REF0:%.*]] = alloca ptr, align 8, addrspace(5)
|
|
// AMDGCN-NEXT: [[S0:%.*]] = alloca i16, align 2, addrspace(5)
|
|
// AMDGCN-NEXT: [[ARRAY:%.*]] = addrspacecast ptr addrspace(5) [[ARRAY0]] to ptr
|
|
// AMDGCN-NEXT: [[N:%.*]] = addrspacecast ptr addrspace(5) [[N0]] to ptr
|
|
// AMDGCN-NEXT: [[REF:%.*]] = addrspacecast ptr addrspace(5) [[REF0]] to ptr
|
|
// AMDGCN-NEXT: [[S:%.*]] = addrspacecast ptr addrspace(5) [[S0]] to ptr
|
|
// X64: [[ARRAY:%.*]] = alloca ptr, align 8
|
|
// X64-NEXT: [[N:%.*]] = alloca i32, align 4
|
|
// X64-NEXT: [[REF:%.*]] = alloca ptr, align 8
|
|
// X64-NEXT: [[S:%.*]] = alloca i16, align 2
|
|
// CHECK-NEXT: store ptr
|
|
// CHECK-NEXT: store i32
|
|
|
|
// Capture the bounds.
|
|
// CHECK-NEXT: [[T0:%.*]] = load i32, ptr [[N]], align 4
|
|
// CHECK-NEXT: [[DIM0:%.*]] = zext i32 [[T0]] to i64
|
|
// CHECK-NEXT: [[T0:%.*]] = load i32, ptr [[N]], align 4
|
|
// CHECK-NEXT: [[T1:%.*]] = add nsw i32 [[T0]], 1
|
|
// CHECK-NEXT: [[DIM1:%.*]] = zext i32 [[T1]] to i64
|
|
typedef short array_t[n][n+1];
|
|
|
|
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[ARRAY]], align 8
|
|
// CHECK-NEXT: store ptr [[T0]], ptr [[REF]], align 8
|
|
array_t &ref = *(array_t*) array;
|
|
|
|
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[REF]]
|
|
// CHECK-NEXT: [[T1:%.*]] = mul nsw i64 1, [[DIM1]]
|
|
// CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i16, ptr [[T0]], i64 [[T1]]
|
|
// CHECK-NEXT: [[T3:%.*]] = getelementptr inbounds i16, ptr [[T2]], i64 2
|
|
// CHECK-NEXT: store i16 3, ptr [[T3]]
|
|
ref[1][2] = 3;
|
|
|
|
// CHECK-NEXT: [[T0:%.*]] = load ptr, ptr [[REF]]
|
|
// CHECK-NEXT: [[T1:%.*]] = mul nsw i64 4, [[DIM1]]
|
|
// CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i16, ptr [[T0]], i64 [[T1]]
|
|
// CHECK-NEXT: [[T3:%.*]] = getelementptr inbounds i16, ptr [[T2]], i64 5
|
|
// CHECK-NEXT: [[T4:%.*]] = load i16, ptr [[T3]]
|
|
// CHECK-NEXT: store i16 [[T4]], ptr [[S]], align 2
|
|
short s = ref[4][5];
|
|
|
|
// CHECK-NEXT: ret void
|
|
}
|
|
|
|
|
|
void test2(int b) {
|
|
// CHECK-LABEL: define{{.*}} void {{.*}}test2{{.*}}(i32 noundef %b)
|
|
int varr[b];
|
|
// AMDGCN: %__end1 = alloca ptr, align 8, addrspace(5)
|
|
// AMDGCN: [[END:%.*]] = addrspacecast ptr addrspace(5) %__end1 to ptr
|
|
// get the address of %b by checking the first store that stores it
|
|
//CHECK: store i32 %b, ptr [[PTR_B:%.*]]
|
|
|
|
// get the size of the VLA by getting the first load of the PTR_B
|
|
//CHECK: [[VLA_NUM_ELEMENTS_PREZEXT:%.*]] = load i32, ptr [[PTR_B]]
|
|
//CHECK-NEXT: [[VLA_NUM_ELEMENTS_PRE:%.*]] = zext i32 [[VLA_NUM_ELEMENTS_PREZEXT]]
|
|
|
|
b = 15;
|
|
//CHECK: store i32 15, ptr [[PTR_B]]
|
|
|
|
// Now get the sizeof, and then divide by the element size
|
|
|
|
|
|
//CHECK: [[VLA_SIZEOF:%.*]] = mul nuw i64 4, [[VLA_NUM_ELEMENTS_PRE]]
|
|
//CHECK-NEXT: [[VLA_NUM_ELEMENTS_POST:%.*]] = udiv i64 [[VLA_SIZEOF]], 4
|
|
//CHECK-NEXT: [[VLA_END_PTR:%.*]] = getelementptr inbounds nuw i32, ptr {{%.*}}, i64 [[VLA_NUM_ELEMENTS_POST]]
|
|
//X64-NEXT: store ptr [[VLA_END_PTR]], ptr %__end1
|
|
//AMDGCN-NEXT: store ptr [[VLA_END_PTR]], ptr [[END]]
|
|
for (int d : varr) 0;
|
|
}
|
|
|
|
void test3(int b, int c) {
|
|
// CHECK-LABEL: define{{.*}} void {{.*}}test3{{.*}}(i32 noundef %b, i32 noundef %c)
|
|
int varr[b][c];
|
|
// AMDGCN: %__end1 = alloca ptr, align 8, addrspace(5)
|
|
// AMDGCN: [[END:%.*]] = addrspacecast ptr addrspace(5) %__end1 to ptr
|
|
// get the address of %b by checking the first store that stores it
|
|
//CHECK: store i32 %b, ptr [[PTR_B:%.*]]
|
|
//CHECK-NEXT: store i32 %c, ptr [[PTR_C:%.*]]
|
|
|
|
// get the size of the VLA by getting the first load of the PTR_B
|
|
//CHECK: [[VLA_DIM1_PREZEXT:%.*]] = load i32, ptr [[PTR_B]]
|
|
//CHECK-NEXT: [[VLA_DIM1_PRE:%.*]] = zext i32 [[VLA_DIM1_PREZEXT]]
|
|
//CHECK: [[VLA_DIM2_PREZEXT:%.*]] = load i32, ptr [[PTR_C]]
|
|
//CHECK-NEXT: [[VLA_DIM2_PRE:%.*]] = zext i32 [[VLA_DIM2_PREZEXT]]
|
|
|
|
b = 15;
|
|
c = 15;
|
|
//CHECK: store i32 15, ptr [[PTR_B]]
|
|
//CHECK: store i32 15, ptr [[PTR_C]]
|
|
// Now get the sizeof, and then divide by the element size
|
|
|
|
// multiply the two dimensions, then by the element type and then divide by the sizeof dim2
|
|
//CHECK: [[VLA_DIM1_X_DIM2:%.*]] = mul nuw i64 [[VLA_DIM1_PRE]], [[VLA_DIM2_PRE]]
|
|
//CHECK-NEXT: [[VLA_SIZEOF:%.*]] = mul nuw i64 4, [[VLA_DIM1_X_DIM2]]
|
|
//CHECK-NEXT: [[VLA_SIZEOF_DIM2:%.*]] = mul nuw i64 4, [[VLA_DIM2_PRE]]
|
|
//CHECK-NEXT: [[VLA_NUM_ELEMENTS:%.*]] = udiv i64 [[VLA_SIZEOF]], [[VLA_SIZEOF_DIM2]]
|
|
//CHECK-NEXT: [[VLA_END_INDEX:%.*]] = mul nsw i64 [[VLA_NUM_ELEMENTS]], [[VLA_DIM2_PRE]]
|
|
//CHECK-NEXT: [[VLA_END_PTR:%.*]] = getelementptr inbounds nuw i32, ptr {{%.*}}, i64 [[VLA_END_INDEX]]
|
|
//X64-NEXT: store ptr [[VLA_END_PTR]], ptr %__end
|
|
//AMDGCN-NEXT: store ptr [[VLA_END_PTR]], ptr [[END]]
|
|
|
|
for (auto &d : varr) 0;
|
|
}
|
|
|
|
|