maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/CodeGenCXX/pod-member-memcpys.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

283 lines
13 KiB
C++
Raw Normal View History

Don't pass -O0 to clang_cc1, it is the default. llvm-svn: 189910
2013-09-04 12:12:25 +08:00
// RUN: %clang_cc1 -triple x86_64-apple-darwin10 -emit-llvm -std=c++03 -fexceptions -fcxx-exceptions -o - %s | FileCheck %s
// RUN: %clang_cc1 -triple i386-apple-darwin10 -emit-llvm -std=c++03 -o - %s | FileCheck --check-prefix=CHECK-2 %s
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
struct POD {
int w, x, y, z;
};
struct PODLike {
int w, x, y, z;
PODLike();
~PODLike();
};
struct NonPOD {
NonPOD();
NonPOD(const NonPOD&);
NonPOD& operator=(const NonPOD&);
};
struct Basic {
int a, b, c, d;
NonPOD np;
int w, x, y, z;
};
struct PODMember {
int a, b, c, d;
POD p;
NonPOD np;
int w, x, y, z;
};
struct PODLikeMember {
int a, b, c, d;
PODLike pl;
NonPOD np;
int w, x, y, z;
};
struct ArrayMember {
int a, b, c, d;
int e[12];
NonPOD np;
int f[12];
int w, x, y, z;
};
Fix an off-by-one mistake in IRGen's copy-construction special cases in the presence of zero-length arrays. Patch by Joran Bigalet! llvm-svn: 358115
2019-04-11 02:07:18 +08:00
struct ZeroLengthArrayMember {
NonPOD np;
int a;
int b[0];
int c;
};
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
struct VolatileMember {
int a, b, c, d;
volatile int v;
NonPOD np;
int w, x, y, z;
};
struct BitfieldMember {
int a, b, c, d;
NonPOD np;
int w : 6;
int x : 6;
int y : 6;
int z : 6;
};
Use the correct alignment for POD-member memcpys where the first field is a bitfield. CGBitField::StorageAlignment holds the alignment in chars, but emitMemcpy had been treating it as if it were held in bits, leading to underaligned memcpys. Related to PR15348. Thanks very much to Chandler for the diagnosis. llvm-svn: 176163
2013-02-27 12:14:49 +08:00
struct BitfieldMember2 {
unsigned a : 1;
unsigned b, c, d;
NonPOD np;
};
CodeGen: FieldMemcpyizer didn't handle copies starting inside bitfields It's possible to construct cases where the first field we are trying to copy is in the middle of an IR field. In some complicated cases, we would fail to use an appropriate offset inside the object. Earlier builds of clang seemed to miscompile the code by copying an insufficient number of bytes. Up until now, we would assert: the copying offset was insufficiently aligned. This fixes PR21232. llvm-svn: 219524
2014-10-11 02:57:10 +08:00
struct BitfieldMember3 {
virtual void f();
int : 8;
int x : 1;
int y;
};
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
struct InnerClassMember {
struct {
int a, b, c, d;
} a;
int b, c, d, e;
NonPOD np;
int w, x, y, z;
};
struct ReferenceMember {
ReferenceMember(int &a, int &b, int &c, int &d)
: a(a), b(b), c(c), d(d) {}
int &a;
int &b;
NonPOD np;
int &c;
int &d;
};
Use ASTContext::getDeclAlign(<Field Decl>) to get the alignment of the first field to be memcpy'd, rather instead of ASTContext::getTypeAlign(<Field Type>). For packed structs the alignment of a field may be less than the alignment of the field's type. <rdar://problem/13338585> llvm-svn: 176512
2013-03-06 04:27:24 +08:00
struct __attribute__((packed)) PackedMembers {
char c;
NonPOD np;
int w, x, y, z;
};
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// COPY-ASSIGNMENT OPERATORS:
// Assignment operators are output in the order they're encountered.
#define CALL_AO(T) void callAO##T(T& a, const T& b) { a = b; }
CALL_AO(Basic)
CALL_AO(PODMember)
CALL_AO(PODLikeMember)
CALL_AO(ArrayMember)
Fix an off-by-one mistake in IRGen's copy-construction special cases in the presence of zero-length arrays. Patch by Joran Bigalet! llvm-svn: 358115
2019-04-11 02:07:18 +08:00
CALL_AO(ZeroLengthArrayMember)
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
CALL_AO(VolatileMember)
CALL_AO(BitfieldMember)
CALL_AO(InnerClassMember)
Use ASTContext::getDeclAlign(<Field Decl>) to get the alignment of the first field to be memcpy'd, rather instead of ASTContext::getTypeAlign(<Field Type>). For packed structs the alignment of a field may be less than the alignment of the field's type. <rdar://problem/13338585> llvm-svn: 176512
2013-03-06 04:27:24 +08:00
CALL_AO(PackedMembers)
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// Basic copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.Basic* @_ZN5BasicaSERKS_(%struct.Basic* %this, %struct.Basic* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Fix the test on -Asserts build. llvm-svn: 189888
2013-09-04 07:52:53 +08:00
// CHECK: ret %struct.Basic*
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// PODMember copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.PODMember* @_ZN9PODMemberaSERKS_(%struct.PODMember* %this, %struct.PODMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 32, i1 {{.*}})
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Fix the test on -Asserts build. llvm-svn: 189888
2013-09-04 07:52:53 +08:00
// CHECK: ret %struct.PODMember*
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// PODLikeMember copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.PODLikeMember* @_ZN13PODLikeMemberaSERKS_(%struct.PODLikeMember* %this, %struct.PODLikeMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 32, i1 {{.*}})
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Fix the test on -Asserts build. llvm-svn: 189888
2013-09-04 07:52:53 +08:00
// CHECK: ret %struct.PODLikeMember*
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// ArrayMember copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.ArrayMember* @_ZN11ArrayMemberaSERKS_(%struct.ArrayMember* %this, %struct.ArrayMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 64, i1 {{.*}})
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 64, i1 {{.*}})
Fix the test on -Asserts build. llvm-svn: 189888
2013-09-04 07:52:53 +08:00
// CHECK: ret %struct.ArrayMember*
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
Fix an off-by-one mistake in IRGen's copy-construction special cases in the presence of zero-length arrays. Patch by Joran Bigalet! llvm-svn: 358115
2019-04-11 02:07:18 +08:00
// ZeroLengthArrayMember copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.ZeroLengthArrayMember* @_ZN21ZeroLengthArrayMemberaSERKS_(%struct.ZeroLengthArrayMember* %this, %struct.ZeroLengthArrayMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Fix an off-by-one mistake in IRGen's copy-construction special cases in the presence of zero-length arrays. Patch by Joran Bigalet! llvm-svn: 358115
2019-04-11 02:07:18 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 8, i1 {{.*}})
// CHECK: ret %struct.ZeroLengthArrayMember*
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// VolatileMember copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.VolatileMember* @_ZN14VolatileMemberaSERKS_(%struct.VolatileMember* %this, %struct.VolatileMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: load volatile i32, i32* {{.*}}, align 4
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// CHECK: store volatile i32 {{.*}}, align 4
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Fix the test on -Asserts build. llvm-svn: 189888
2013-09-04 07:52:53 +08:00
// CHECK: ret %struct.VolatileMember*
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// BitfieldMember copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.BitfieldMember* @_ZN14BitfieldMemberaSERKS_(%struct.BitfieldMember* %this, %struct.BitfieldMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 1 {{.*}} align 1 {{.*}}i64 3, i1 {{.*}})
Fix the test on -Asserts build. llvm-svn: 189888
2013-09-04 07:52:53 +08:00
// CHECK: ret %struct.BitfieldMember*
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// InnerClass copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.InnerClassMember* @_ZN16InnerClassMemberaSERKS_(%struct.InnerClassMember* %this, %struct.InnerClassMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 32, i1 {{.*}})
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Fix the test on -Asserts build. llvm-svn: 189888
2013-09-04 07:52:53 +08:00
// CHECK: ret %struct.InnerClassMember*
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
Use ASTContext::getDeclAlign(<Field Decl>) to get the alignment of the first field to be memcpy'd, rather instead of ASTContext::getTypeAlign(<Field Type>). For packed structs the alignment of a field may be less than the alignment of the field's type. <rdar://problem/13338585> llvm-svn: 176512
2013-03-06 04:27:24 +08:00
// PackedMembers copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.PackedMembers* @_ZN13PackedMembersaSERKS_(%struct.PackedMembers* %this, %struct.PackedMembers* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
// CHECK: call nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %struct.NonPOD* @_ZN6NonPODaSERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 1 {{.*}} align 1 {{.*}}i64 16, i1 {{.*}})
Fix the test on -Asserts build. llvm-svn: 189888
2013-09-04 07:52:53 +08:00
// CHECK: ret %struct.PackedMembers*
Use ASTContext::getDeclAlign(<Field Decl>) to get the alignment of the first field to be memcpy'd, rather instead of ASTContext::getTypeAlign(<Field Type>). For packed structs the alignment of a field may be less than the alignment of the field's type. <rdar://problem/13338585> llvm-svn: 176512
2013-03-06 04:27:24 +08:00
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// COPY-CONSTRUCTORS:
// Clang outputs copy-constructors in the reverse of the order that
// copy-constructor calls are encountered. Add functions that call the copy
// constructors of the classes above in reverse order here.
#define CALL_CC(T) T callCC##T(const T& b) { return b; }
Use ASTContext::getDeclAlign(<Field Decl>) to get the alignment of the first field to be memcpy'd, rather instead of ASTContext::getTypeAlign(<Field Type>). For packed structs the alignment of a field may be less than the alignment of the field's type. <rdar://problem/13338585> llvm-svn: 176512
2013-03-06 04:27:24 +08:00
CALL_CC(PackedMembers)
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// PackedMembers copy-assignment:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN13PackedMembersC2ERKS_(%struct.PackedMembers* %this, %struct.PackedMembers* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// CHECK: call void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 1 {{.*}} align 1 {{.*}}i64 16, i1 {{.*}})
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// CHECK: ret void
Use the correct alignment for POD-member memcpys where the first field is a bitfield. CGBitField::StorageAlignment holds the alignment in chars, but emitMemcpy had been treating it as if it were held in bits, leading to underaligned memcpys. Related to PR15348. Thanks very much to Chandler for the diagnosis. llvm-svn: 176163
2013-02-27 12:14:49 +08:00
CALL_CC(BitfieldMember2)
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// BitfieldMember2 copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-2-LABEL: define linkonce_odr void @_ZN15BitfieldMember2C2ERKS_(%struct.BitfieldMember2* %this, %struct.BitfieldMember2* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK-2: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 false)
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// CHECK-2: call void @_ZN6NonPODC1ERKS_
// CHECK-2: ret void
CodeGen: FieldMemcpyizer didn't handle copies starting inside bitfields It's possible to construct cases where the first field we are trying to copy is in the middle of an IR field. In some complicated cases, we would fail to use an appropriate offset inside the object. Earlier builds of clang seemed to miscompile the code by copying an insufficient number of bytes. Up until now, we would assert: the copying offset was insufficiently aligned. This fixes PR21232. llvm-svn: 219524
2014-10-11 02:57:10 +08:00
CALL_CC(BitfieldMember3)
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// BitfieldMember3 copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN15BitfieldMember3C2ERKS_(%struct.BitfieldMember3* %this, %struct.BitfieldMember3* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 8 {{.*}} align 8 {{.*}}i64 8, i1 false)
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// CHECK: ret void
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
CALL_CC(ReferenceMember)
// ReferenceMember copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN15ReferenceMemberC2ERKS_(%struct.ReferenceMember* %this, %struct.ReferenceMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 8 {{.*}} align 8 {{.*}}i64 16, i1 {{.*}})
Don't run optimizations in a clang test. llvm-svn: 189880
2013-09-04 07:14:03 +08:00
// CHECK: call void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 8 {{.*}} align 8 {{.*}}i64 16, i1 {{.*}})
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// CHECK: ret void
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
CALL_CC(InnerClassMember)
// InnerClass copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN16InnerClassMemberC2ERKS_(%struct.InnerClassMember* %this, %struct.InnerClassMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 32, i1 {{.*}})
Don't run optimizations in a clang test. llvm-svn: 189880
2013-09-04 07:14:03 +08:00
// CHECK: call void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// CHECK: ret void
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
CALL_CC(BitfieldMember)
// BitfieldMember copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN14BitfieldMemberC2ERKS_(%struct.BitfieldMember* %this, %struct.BitfieldMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Don't run optimizations in a clang test. llvm-svn: 189880
2013-09-04 07:14:03 +08:00
// CHECK: call void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 1 {{.*}} align 1 {{.*}}i64 3, i1 {{.*}})
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// CHECK: ret void
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
CALL_CC(VolatileMember)
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// VolatileMember copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN14VolatileMemberC2ERKS_(%struct.VolatileMember* %this, %struct.VolatileMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: load volatile i32, i32* {{.*}}, align 4
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// CHECK: store volatile i32 {{.*}}, align 4
Don't run optimizations in a clang test. llvm-svn: 189880
2013-09-04 07:14:03 +08:00
// CHECK: call void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// CHECK: ret void
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
CALL_CC(ArrayMember)
// ArrayMember copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN11ArrayMemberC2ERKS_(%struct.ArrayMember* %this, %struct.ArrayMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 64, i1 {{.*}})
Don't run optimizations in a clang test. llvm-svn: 189880
2013-09-04 07:14:03 +08:00
// CHECK: call void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 64, i1 {{.*}})
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// CHECK: ret void
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
CALL_CC(PODLikeMember)
// PODLikeMember copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN13PODLikeMemberC2ERKS_(%struct.PODLikeMember* %this, %struct.PODLikeMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 32, i1 {{.*}})
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// CHECK: invoke void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
// CHECK: ret void
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
// CHECK: landingpad
// CHECK: invoke void @_ZN7PODLikeD1Ev
Re-apply r174919 - smarter copy/move assignment/construction, with fixes for bitfield related issues. The original commit broke Takumi's builder. The bug was caused by bitfield sizes being determined by their underlying type, rather than the field info. A similar issue with bitfield alignments showed up on closer testing. Both have been fixed in this patch. llvm-svn: 175389
2013-02-17 15:22:09 +08:00
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
CALL_CC(PODMember)
// PODMember copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN9PODMemberC2ERKS_(%struct.PODMember* %this, %struct.PODMember* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 32, i1 {{.*}})
Don't run optimizations in a clang test. llvm-svn: 189880
2013-09-04 07:14:03 +08:00
// CHECK: call void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
CodeGen: FieldMemcpyizer didn't handle copies starting inside bitfields It's possible to construct cases where the first field we are trying to copy is in the middle of an IR field. In some complicated cases, we would fail to use an appropriate offset inside the object. Earlier builds of clang seemed to miscompile the code by copying an insufficient number of bytes. Up until now, we would assert: the copying offset was insufficiently aligned. This fixes PR21232. llvm-svn: 219524
2014-10-11 02:57:10 +08:00
// CHECK: ret void
Emit DeferredDeclsToEmit in a DFS order. Currently we emit DeferredDeclsToEmit in reverse order. This patch changes that. The advantages of the change are that * The output order is a bit closer to the source order. The change to test/CodeGenCXX/pod-member-memcpys.cpp is a good example. * If we decide to deffer more, it will not cause as large changes in the estcases as it would without this patch. llvm-svn: 226751
2015-01-22 08:24:57 +08:00
CALL_CC(Basic)
// Basic copy-constructor:
[CGCall] Annotate references with "align" attribute. If we're going to assume references are dereferenceable, we should also assume they're aligned: otherwise, we can't actually dereference them. See also D80072. Differential Revision: https://reviews.llvm.org/D80166
2020-05-19 02:29:11 +08:00
// CHECK-LABEL: define linkonce_odr void @_ZN5BasicC2ERKS_(%struct.Basic* %this, %struct.Basic* nonnull align {{[0-9]+}} dereferenceable({{[0-9]+}}) %0)
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Don't run optimizations in a clang test. llvm-svn: 189880
2013-09-04 07:14:03 +08:00
// CHECK: call void @_ZN6NonPODC1ERKS_
Change memcpy/memove/memset to have dest and source alignment attributes (Step 1). Summary: Upstream LLVM is changing the the prototypes of the @llvm.memcpy/memmove/memset intrinsics. This change updates the Clang tests for this change. The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change removes the alignment argument in favour of placing the alignment attribute on the source and destination pointers of the memory intrinsic call. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) At this time the source and destination alignments must be the same (Step 1). Step 2 of the change, to be landed shortly, will relax that contraint and allow the source and destination to have different alignments. llvm-svn: 322964
2018-01-20 01:12:54 +08:00
// CHECK: call void @llvm.memcpy.p0i8.p0i8.i64({{.*}} align 4 {{.*}} align 4 {{.*}}i64 16, i1 {{.*}})
Use ASTContext::getDeclAlign(<Field Decl>) to get the alignment of the first field to be memcpy'd, rather instead of ASTContext::getTypeAlign(<Field Type>). For packed structs the alignment of a field may be less than the alignment of the field's type. <rdar://problem/13338585> llvm-svn: 176512
2013-03-06 04:27:24 +08:00
// CHECK: ret void