3dd5a298bf
Summary: Right now we annotate C++'s `operator new` with `noalias` attribute, which very much is healthy for optimizations. However as per [[ http://eel.is/c++draft/basic.stc.dynamic.allocation | `[basic.stc.dynamic.allocation]` ]], there are more promises on global `operator new`, namely: * non-`std::nothrow_t` `operator new` *never* returns `nullptr` * If `std::align_val_t align` parameter is taken, the pointer will also be `align`-aligned * ~~global `operator new`-returned pointer is `__STDCPP_DEFAULT_NEW_ALIGNMENT__`-aligned ~~ It's more caveated than that. Supplying this information may not cause immediate landslide effects on any specific benchmarks, but it for sure will be healthy for optimizer in the sense that the IR will better reflect the guarantees provided in the source code. The caveat is `-fno-assume-sane-operator-new`, which currently prevents emitting `noalias` attribute, and is automatically passed by Sanitizers ([[ https://bugs.llvm.org/show_bug.cgi?id=16386 | PR16386 ]]) - should it also cover these attributes? The problem is that the flag is back-end-specific, as seen in `test/Modules/explicit-build-flags.cpp`. But while it is okay to add `noalias` metadata in backend, we really should be adding at least the alignment metadata to the AST, since that allows us to perform sema checks on it. Reviewers: erichkeane, rjmccall, jdoerfert, eugenis, rsmith Reviewed By: rsmith Subscribers: xbolva00, jrtc27, atanasyan, nlopes, cfe-commits Tags: #llvm, #clang Differential Revision: https://reviews.llvm.org/D73380
148 lines
5.8 KiB
C++
148 lines
5.8 KiB
C++
// RUN: %clang_cc1 -std=c++11 -triple i386-unknown-unknown %s -emit-llvm -o - | FileCheck %s
|
|
// RUN: %clang_cc1 -std=c++11 -triple i386-unknown-unknown %s -emit-llvm -fsanitize=signed-integer-overflow -o - | FileCheck --check-prefix=SIO %s
|
|
|
|
// CHECK: @[[ABC4:.*]] = {{.*}} constant [4 x i8] c"abc\00"
|
|
// CHECK: @[[ABC15:.*]] = {{.*}} constant [15 x i8] c"abc\00\00\00\00
|
|
|
|
// CHECK-LABEL: define void @_Z2fni
|
|
void fn(int n) {
|
|
// CHECK: icmp ult i{{32|64}} %{{[^ ]+}}, 3
|
|
// CHECK: store i32 1
|
|
// CHECK: store i32 2
|
|
// CHECK: store i32 3
|
|
// CHECK: sub {{.*}}, 12
|
|
// CHECK: call void @llvm.memset
|
|
new int[n] { 1, 2, 3 };
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z11const_exactv
|
|
void const_exact() {
|
|
// CHECK-NOT: icmp ult i{{32|64}} %{{[^ ]+}}, 3
|
|
// CHECK-NOT: icmp eq i32*
|
|
new int[3] { 1, 2, 3 };
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z16const_sufficientv
|
|
void const_sufficient() {
|
|
// CHECK-NOT: icmp ult i{{32|64}} %{{[^ ]+}}, 3
|
|
new int[4] { 1, 2, 3 };
|
|
// CHECK: ret void
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z22check_array_value_initv
|
|
void check_array_value_init() {
|
|
struct S;
|
|
new (int S::*[3][4][5]) ();
|
|
|
|
// CHECK: call noalias nonnull i8* @_Zna{{.}}(i{{32 240|64 480}})
|
|
// CHECK: getelementptr inbounds i{{32|64}}, i{{32|64}}* {{.*}}, i{{32|64}} 60
|
|
|
|
// CHECK: phi
|
|
// CHECK: store i{{32|64}} -1,
|
|
// CHECK: getelementptr inbounds i{{32|64}}, i{{32|64}}* {{.*}}, i{{32|64}} 1
|
|
// CHECK: icmp eq
|
|
// CHECK: br i1
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z15string_nonconsti
|
|
void string_nonconst(int n) {
|
|
// CHECK: icmp slt i{{32|64}} %{{[^ ]+}}, 4
|
|
// FIXME: Conditionally throw an exception rather than passing -1 to alloc function
|
|
// CHECK: select
|
|
// CHECK: %[[PTR:.*]] = call noalias nonnull i8* @_Zna{{.}}(i{{32|64}}
|
|
// CHECK: call void @llvm.memcpy{{.*}}(i8* align {{[0-9]+}} %[[PTR]], i8* align {{[0-9]+}} getelementptr inbounds ([4 x i8], [4 x i8]* @[[ABC4]], i32 0, i32 0), i32 4,
|
|
// CHECK: %[[REST:.*]] = getelementptr inbounds i8, i8* %[[PTR]], i32 4
|
|
// CHECK: %[[RESTSIZE:.*]] = sub {{.*}}, 4
|
|
// CHECK: call void @llvm.memset{{.*}}(i8* align {{[0-9]+}} %[[REST]], i8 0, i{{32|64}} %[[RESTSIZE]],
|
|
new char[n] { "abc" };
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z12string_exactv
|
|
void string_exact() {
|
|
// CHECK-NOT: icmp
|
|
// CHECK: %[[PTR:.*]] = call noalias nonnull i8* @_Zna{{.}}(i{{32|64}} 4)
|
|
// CHECK: call void @llvm.memcpy{{.*}}(i8* align {{[0-9]+}} %[[PTR]], i8* align {{[0-9]+}} getelementptr inbounds ([4 x i8], [4 x i8]* @[[ABC4]], i32 0, i32 0), i32 4,
|
|
// CHECK-NOT: memset
|
|
new char[4] { "abc" };
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z17string_sufficientv
|
|
void string_sufficient() {
|
|
// CHECK-NOT: icmp
|
|
// CHECK: %[[PTR:.*]] = call noalias nonnull i8* @_Zna{{.}}(i{{32|64}} 15)
|
|
// FIXME: For very large arrays, it would be preferable to emit a small copy and a memset.
|
|
// CHECK: call void @llvm.memcpy{{.*}}(i8* align {{[0-9]+}} %[[PTR]], i8* align {{[0-9]+}} getelementptr inbounds ([15 x i8], [15 x i8]* @[[ABC15]], i32 0, i32 0), i32 15,
|
|
// CHECK-NOT: memset
|
|
new char[15] { "abc" };
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z10aggr_exactv
|
|
void aggr_exact() {
|
|
// CHECK-NOT: icmp
|
|
// CHECK: %[[MEM:.*]] = call noalias nonnull i8* @_Zna{{.}}(i{{32|64}} 16)
|
|
// CHECK: %[[PTR0:.*]] = bitcast i8* %[[MEM]] to %[[AGGR:.*]]*
|
|
// CHECK: %[[FIELD:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR0]], i32 0, i32 0{{$}}
|
|
// CHECK: store i32 1, i32* %[[FIELD]]
|
|
// CHECK: %[[FIELD:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR0]], i32 0, i32 1{{$}}
|
|
// CHECK: store i32 2, i32* %[[FIELD]]
|
|
// CHECK: %[[PTR1:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR0]], i32 1{{$}}
|
|
// CHECK: %[[FIELD:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR1]], i32 0, i32 0{{$}}
|
|
// CHECK: store i32 3, i32* %[[FIELD]]
|
|
// CHECK: %[[FIELD:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR1]], i32 0, i32 1{{$}}
|
|
// CHECK: store i32 0, i32* %[[FIELD]]
|
|
// CHECK-NOT: store
|
|
// CHECK-NOT: memset
|
|
struct Aggr { int a, b; };
|
|
new Aggr[2] { 1, 2, 3 };
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z15aggr_sufficienti
|
|
void aggr_sufficient(int n) {
|
|
// CHECK: icmp ult i32 %{{.*}}, 2
|
|
// CHECK: %[[MEM:.*]] = call noalias nonnull i8* @_Zna{{.}}(
|
|
// CHECK: %[[PTR0:.*]] = bitcast i8* %[[MEM]] to %[[AGGR:.*]]*
|
|
// CHECK: %[[FIELD:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR0]], i32 0, i32 0{{$}}
|
|
// CHECK: store i32 1, i32* %[[FIELD]]
|
|
// CHECK: %[[FIELD:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR0]], i32 0, i32 1{{$}}
|
|
// CHECK: store i32 2, i32* %[[FIELD]]
|
|
// CHECK: %[[PTR1:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR0]], i32 1{{$}}
|
|
// CHECK: %[[FIELD:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR1]], i32 0, i32 0{{$}}
|
|
// CHECK: store i32 3, i32* %[[FIELD]]
|
|
// CHECK: %[[FIELD:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR1]], i32 0, i32 1{{$}}
|
|
// CHECK: store i32 0, i32* %[[FIELD]]
|
|
// CHECK: %[[PTR2:.*]] = getelementptr inbounds %[[AGGR]], %[[AGGR]]* %[[PTR1]], i32 1{{$}}
|
|
// CHECK: %[[REMAIN:.*]] = sub i32 {{.*}}, 16
|
|
// CHECK: %[[MEM:.*]] = bitcast %[[AGGR]]* %[[PTR2]] to i8*
|
|
// CHECK: call void @llvm.memset{{.*}}(i8* align {{[0-9]+}} %[[MEM]], i8 0, i32 %[[REMAIN]],
|
|
struct Aggr { int a, b; };
|
|
new Aggr[n] { 1, 2, 3 };
|
|
}
|
|
|
|
// SIO-LABEL: define void @_Z14constexpr_testv
|
|
void constexpr_test() {
|
|
// SIO: call noalias nonnull i8* @_Zna{{.}}(i32 4)
|
|
new int[0+1]{0};
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z13unknown_boundv
|
|
void unknown_bound() {
|
|
struct Aggr { int x, y, z; };
|
|
new Aggr[]{1, 2, 3, 4};
|
|
// CHECK: call {{.*}}_Znaj(i32 24)
|
|
// CHECK: store i32 1
|
|
// CHECK: store i32 2
|
|
// CHECK: store i32 3
|
|
// CHECK: store i32 4
|
|
// CHECK: store i32 0
|
|
// CHECK: store i32 0
|
|
// CHECK-NOT: store
|
|
// CHECK: }
|
|
}
|
|
|
|
// CHECK-LABEL: define void @_Z20unknown_bound_stringv
|
|
void unknown_bound_string() {
|
|
new char[]{"hello"};
|
|
// CHECK: call {{.*}}_Znaj(i32 6)
|
|
// CHECK: memcpy{{.*}} i32 6,
|
|
}
|