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/microsoft-abi-member-pointe...

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

940 lines
36 KiB
C++
Raw Normal View History

[FileCheck] Add -allow-deprecated-dag-overlap to failing clang tests See https://reviews.llvm.org/D47106 for details. Reviewed By: probinson Differential Revision: https://reviews.llvm.org/D47172 llvm-svn: 336844
2018-07-12 04:26:20 +08:00
// RUN: %clang_cc1 -std=c++11 -Wno-uninitialized -fno-rtti -emit-llvm %s -o - -triple=i386-pc-win32 -fms-extensions | FileCheck -allow-deprecated-dag-overlap %s
MS ABI: Don't ICE for pointers to pointers to members of incomplete classes CodeGen would try to come up with an LLVM IR type for a pointer to member type on the way to forming an LLVM IR type for a pointer to pointer to member type. However, if the pointer to member representation has not been locked in yet, we would not be able to come up with a pointer to member IR type. In these cases, make the pointer to member type an incomplete type. This will make the pointer to pointer to member type a pointer to an incomplete type. If the class eventually obtains an inheritance model, we will make the pointer to member type represent the actual inheritance model. Differential Revision: http://reviews.llvm.org/D5373 llvm-svn: 218084
2014-09-19 06:05:54 +08:00
// RUN: %clang_cc1 -std=c++11 -Wno-uninitialized -fno-rtti -emit-llvm %s -o - -triple=x86_64-pc-win32 -fms-extensions | FileCheck %s -check-prefix=X64
// RUN: %clang_cc1 -std=c++11 -Wno-uninitialized -fno-rtti -emit-llvm %s -o - -triple=i386-pc-win32 -DINCOMPLETE_VIRTUAL -fms-extensions -verify
// RUN: %clang_cc1 -std=c++11 -Wno-uninitialized -fno-rtti -emit-llvm %s -o - -triple=i386-pc-win32 -DINCOMPLETE_VIRTUAL -DMEMFUN -fms-extensions -verify
Fix null MSInheritanceAttr deref in CXXRecordDecl::getMSInheritanceModel() Ensure latest MPT decl has a MSInheritanceAttr when instantiating templates, to avoid null MSInheritanceAttr deref in CXXRecordDecl::getMSInheritanceModel(). See PR#37399 for repo / details. Patch by Andrew Rogers! Differential Revision: https://reviews.llvm.org/D46664 llvm-svn: 333680
2018-06-01 02:42:29 +08:00
namespace pr37399 {
template <typename T>
struct Functor {
void (T::*PtrToMemberFunction)();
};
// CHECK-DAG: %"struct.pr37399::Functor" = type { i8* }
template <typename SomeType>
class SimpleDerivedFunctor;
template <typename SomeType>
class SimpleDerivedFunctor : public Functor<SimpleDerivedFunctor<SomeType>> {};
// CHECK-DAG: %"class.pr37399::SimpleDerivedFunctor" = type { %"struct.pr37399::Functor" }
SimpleDerivedFunctor<void> SimpleFunctor;
// CHECK-DAG: @"?SimpleFunctor@pr37399@@3V?$SimpleDerivedFunctor@X@1@A" = dso_local global %"class.pr37399::SimpleDerivedFunctor" zeroinitializer, align 4
short Global = 0;
template <typename SomeType>
class DerivedFunctor;
template <typename SomeType>
class DerivedFunctor
: public Functor<DerivedFunctor<void>> {
public:
void Foo() {
Global = 42;
}
};
class MultipleBase {
public:
MultipleBase() : Value() {}
short Value;
};
// CHECK-DAG: %"class.pr37399::MultipleBase" = type { i16 }
template <typename SomeType>
class MultiplyDerivedFunctor;
template <typename SomeType>
class MultiplyDerivedFunctor
: public Functor<MultiplyDerivedFunctor<void>>,
public MultipleBase {
public:
void Foo() {
MultipleBase::Value = 42*2;
}
};
class VirtualBase {
public:
VirtualBase() : Value() {}
short Value;
};
// CHECK-DAG: %"class.pr37399::VirtualBase" = type { i16 }
template <typename SomeType>
class VirtBaseFunctor
: public Functor<SomeType>,
public virtual VirtualBase{};
template <typename SomeType>
class VirtuallyDerivedFunctor;
template <typename SomeType>
class VirtuallyDerivedFunctor
: public VirtBaseFunctor<VirtuallyDerivedFunctor<void>>,
public virtual VirtualBase {
public:
void Foo() {
VirtualBase::Value = 42*3;
}
};
} // namespace pr37399
pr37399::DerivedFunctor<int> BFunctor;
// CHECK-DAG: @"?BFunctor@@3V?$DerivedFunctor@H@pr37399@@A" = dso_local global %"[[BFUNCTOR:class.pr37399::DerivedFunctor(\.[0-9]+)?]]" zeroinitializer, align 8
// CHECK-DAG: %"[[BFUNCTOR]]" = type { %"[[BFUNCTORBASE:struct.pr37399::Functor(\.[0-9]+)?]]" }
// CHECK-DAG: %"[[BFUNCTORBASE]]" = type { { i8*, i32, i32, i32 } }
pr37399::DerivedFunctor<void> AFunctor;
// CHECK-DAG: @"?AFunctor@@3V?$DerivedFunctor@X@pr37399@@A" = dso_local global %"[[AFUNCTOR:class.pr37399::DerivedFunctor(\.[0-9]+)?]]" zeroinitializer, align 8
// CHECK-DAG: %"[[AFUNCTOR]]" = type { %"[[AFUNCTORBASE:struct.pr37399::Functor(\.[0-9]+)?]]" }
// CHECK-DAG: %"[[AFUNCTORBASE]]" = type { { i8*, i32, i32, i32 } }
pr37399::MultiplyDerivedFunctor<int> DFunctor;
// CHECK-DAG: @"?DFunctor@@3V?$MultiplyDerivedFunctor@H@pr37399@@A" = dso_local global %"[[DFUNCTOR:class.pr37399::MultiplyDerivedFunctor(\.[0-9]+)?]]" zeroinitializer, align 8
// CHECK-DAG: %"[[DFUNCTOR]]" = type { %"[[DFUNCTORBASE:struct.pr37399::Functor(\.[0-9]+)?]]", %"class.pr37399::MultipleBase", [6 x i8] }
// CHECK-DAG: %"[[DFUNCTORBASE]]" = type { { i8*, i32, i32, i32 } }
pr37399::MultiplyDerivedFunctor<void> CFunctor;
// CHECK-DAG: @"?CFunctor@@3V?$MultiplyDerivedFunctor@X@pr37399@@A" = dso_local global %"[[CFUNCTOR:class.pr37399::MultiplyDerivedFunctor(\.[0-9]+)?]]" zeroinitializer, align 8
// CHECK-DAG: %"[[CFUNCTOR]]" = type { %"[[CFUNCTORBASE:struct.pr37399::Functor(\.[0-9]+)?]]", %"class.pr37399::MultipleBase", [6 x i8] }
// CHECK-DAG: %"[[CFUNCTORBASE]]" = type { { i8*, i32, i32, i32 } }
pr37399::VirtuallyDerivedFunctor<int> FFunctor;
// CHECK-DAG: @"?FFunctor@@3V?$VirtuallyDerivedFunctor@H@pr37399@@A" = dso_local global %"[[FFUNCTOR:class.pr37399::VirtuallyDerivedFunctor(\.[0-9]+)?]]" zeroinitializer, align 8
// CHECK-DAG: %"[[FFUNCTOR]]" = type { %"class.pr37399::VirtBaseFunctor.base", %"class.pr37399::VirtualBase" }
pr37399::VirtuallyDerivedFunctor<void> EFunctor;
// CHECK-DAG: @"?EFunctor@@3V?$VirtuallyDerivedFunctor@X@pr37399@@A" = dso_local global %"[[EFUNCTOR:class.pr37399::VirtuallyDerivedFunctor(\.[0-9]+)?]]" zeroinitializer, align 8
// CHECK-DAG: %"[[EFUNCTOR]]" = type { %"class.pr37399::VirtBaseFunctor.base", %"class.pr37399::VirtualBase" }
// CHECK-DAG: %"class.pr37399::VirtBaseFunctor.base" = type <{ %"[[VFUNCTORBASE:struct.pr37399::Functor(\.[0-9]+)?]]", i32*, [4 x i8] }>
// CHECK-DAG: %"[[VFUNCTORBASE]]" = type { { i8*, i32, i32, i32 } }
namespace pr37399 {
void SingleInheritanceFnPtrCall() {
BFunctor.PtrToMemberFunction = &DerivedFunctor<void>::Foo;
(AFunctor.*(BFunctor.PtrToMemberFunction))();
}
void MultipleInheritanceFnPtrCall() {
DFunctor.PtrToMemberFunction = &MultiplyDerivedFunctor<void>::Foo;
Global = CFunctor.MultipleBase::Value;
(CFunctor.*(DFunctor.PtrToMemberFunction))();
Global = CFunctor.MultipleBase::Value;
}
void VirtualInheritanceFnPtrCall() {
FFunctor.PtrToMemberFunction = &VirtuallyDerivedFunctor<void>::Foo;
Global = EFunctor.VirtualBase::Value;
(EFunctor.*(FFunctor.PtrToMemberFunction))();
Global = EFunctor.VirtualBase::Value;
}
} // namespace pr37399
[MS] Fix constexpr data member pointer conversions Constexpr data member conversions work by starting with the class that originally introduced the field, and converting from there to the type that the user desires. Before this change, Clang was using the inheritance model from the final destination class type instead of the model from the class that originally introduced the field. To fix this, find the relevant FieldDecl and take its parent class instead of using the member pointer type the user provided. Indirect field decls require some special handling to find the parent class. Fixes PR43803
2019-10-29 08:05:34 +08:00
namespace pr43803 {
// This case is interesting because it exercises conversion between member
// pointer types when emitting constants.
struct B;
struct C { int B::*option; };
extern const C table[3];
struct A {
int x, y;
// Test the indirect case.
struct {
int z;
};
};
struct B : A {};
const C table[] = {
{&B::x},
{&B::y},
{&B::z},
};
// CHECK: @"?table@pr43803@@3QBUC@1@B" = dso_local constant [3 x %"struct.pr43803::C"]
// CHECK-SAME: [%"struct.pr43803::C" { { i32, i32, i32 } zeroinitializer, [4 x i8] undef },
// CHECK-SAME: %"struct.pr43803::C" { { i32, i32, i32 } { i32 4, i32 0, i32 0 }, [4 x i8] undef },
// CHECK-SAME: %"struct.pr43803::C" { { i32, i32, i32 } { i32 8, i32 0, i32 0 }, [4 x i8] undef }]
}
[MS ABI] Allow a member pointers' converted type to change Member pointers in the MS ABI are tricky for a variety of reasons. The size of a member pointer is indeterminate until the program reaches a point where the representation is required to be known. However, *pointers* to member pointers may exist without knowing the pointee type's representation. In these cases, we synthesize an opaque LLVM type for the pointee type. However, we can be in a situation where the underlying member pointer's representation became known mid-way through the program. To account for this, we attempted to manicure CodeGen's type-cache so that we can replace the opaque member pointer type with the real deal while leaving the pointer types unperturbed. This, unfortunately, is a problematic approach to take as we will violate CodeGen's invariants. These violations are mostly harmless but let's do the right thing instead: invalidate the type-cache if a member pointer's LLVM representation changes. This fixes PR26313. llvm-svn: 258839
2016-01-27 03:30:26 +08:00
struct PR26313_Y;
typedef void (PR26313_Y::*PR26313_FUNC)();
struct PR26313_X {
PR26313_FUNC *ptr;
PR26313_X();
};
PR26313_X::PR26313_X() {}
void PR26313_f(PR26313_FUNC *p) { delete p; }
struct PR26313_Z;
int PR26313_Z::**a = nullptr;
int PR26313_Z::*b = *a;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-DAG: @"?a@@3PAPQPR26313_Z@@HA" = dso_local global %0* null, align 4
// CHECK-DAG: @"?b@@3PQPR26313_Z@@HQ1@" = dso_local global { i32, i32, i32 } { i32 0, i32 0, i32 -1 }, align 4
[MS ABI] Allow a member pointers' converted type to change Member pointers in the MS ABI are tricky for a variety of reasons. The size of a member pointer is indeterminate until the program reaches a point where the representation is required to be known. However, *pointers* to member pointers may exist without knowing the pointee type's representation. In these cases, we synthesize an opaque LLVM type for the pointee type. However, we can be in a situation where the underlying member pointer's representation became known mid-way through the program. To account for this, we attempted to manicure CodeGen's type-cache so that we can replace the opaque member pointer type with the real deal while leaving the pointer types unperturbed. This, unfortunately, is a problematic approach to take as we will violate CodeGen's invariants. These violations are mostly harmless but let's do the right thing instead: invalidate the type-cache if a member pointer's LLVM representation changes. This fixes PR26313. llvm-svn: 258839
2016-01-27 03:30:26 +08:00
MS ABI: Don't ICE for pointers to pointers to members of incomplete classes CodeGen would try to come up with an LLVM IR type for a pointer to member type on the way to forming an LLVM IR type for a pointer to pointer to member type. However, if the pointer to member representation has not been locked in yet, we would not be able to come up with a pointer to member IR type. In these cases, make the pointer to member type an incomplete type. This will make the pointer to pointer to member type a pointer to an incomplete type. If the class eventually obtains an inheritance model, we will make the pointer to member type represent the actual inheritance model. Differential Revision: http://reviews.llvm.org/D5373 llvm-svn: 218084
2014-09-19 06:05:54 +08:00
namespace PR20947 {
struct A;
int A::**a = nullptr;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-DAG: @"?a@PR20947@@3PAPQA@1@HA" = dso_local global %{{.*}}* null, align 4
MS ABI: Don't ICE for pointers to pointers to members of incomplete classes CodeGen would try to come up with an LLVM IR type for a pointer to member type on the way to forming an LLVM IR type for a pointer to pointer to member type. However, if the pointer to member representation has not been locked in yet, we would not be able to come up with a pointer to member IR type. In these cases, make the pointer to member type an incomplete type. This will make the pointer to pointer to member type a pointer to an incomplete type. If the class eventually obtains an inheritance model, we will make the pointer to member type represent the actual inheritance model. Differential Revision: http://reviews.llvm.org/D5373 llvm-svn: 218084
2014-09-19 06:05:54 +08:00
struct B;
int B::*&b = b;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-DAG: @"?b@PR20947@@3AAPQB@1@HA" = dso_local global %{{.*}}* null, align 4
MS ABI: Don't ICE for pointers to pointers to members of incomplete classes CodeGen would try to come up with an LLVM IR type for a pointer to member type on the way to forming an LLVM IR type for a pointer to pointer to member type. However, if the pointer to member representation has not been locked in yet, we would not be able to come up with a pointer to member IR type. In these cases, make the pointer to member type an incomplete type. This will make the pointer to pointer to member type a pointer to an incomplete type. If the class eventually obtains an inheritance model, we will make the pointer to member type represent the actual inheritance model. Differential Revision: http://reviews.llvm.org/D5373 llvm-svn: 218084
2014-09-19 06:05:54 +08:00
}
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
MS ABI: Add another test for PR20017 llvm-svn: 223733
2014-12-09 09:36:45 +08:00
namespace PR20017 {
template <typename T>
struct A {
int T::*m_fn1() { return nullptr; }
};
struct B;
auto a = &A<B>::m_fn1;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-DAG: @"?a@PR20017@@3P8?$A@UB@PR20017@@@1@AEPQB@1@HXZQ21@" = dso_local global i8* bitcast ({ i32, i32, i32 } ({{.*}}*)* @"?m_fn1@?$A@UB@PR20017@@@PR20017@@QAEPQB@2@HXZ" to i8*), align 4
MS ABI: Add another test for PR20017 llvm-svn: 223733
2014-12-09 09:36:45 +08:00
}
Sema: Do not assert when dereferencing member pointer using virtual inheritance with an incomplete class type The MS ABI requires that we determine the vbptr offset if have a virtual inheritance model. Instead, raise an error pointing to the diagnostic when this happens. This fixes PR18583. Differential Revision: http://llvm-reviews.chandlerc.com/D2842 llvm-svn: 201824
2014-02-21 07:22:07 +08:00
#ifndef INCOMPLETE_VIRTUAL
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
struct B1 {
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
void foo();
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
int b;
};
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
struct B2 {
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
int b2;
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
void foo();
};
struct Single : B1 {
void foo();
};
struct Multiple : B1, B2 {
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
int m;
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
void foo();
};
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
struct Virtual : virtual B1 {
int v;
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
void foo();
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
};
[ms-cxxabi] Implement member data pointers for non-dynamic classes Summary: For non-dynamic classes (no virtual bases), member data pointers are simple offsets from the base of the record. Dynamic classes use an aggregate for member data pointers and are therefore currently unsupported. Unlike Itanium, the ms ABI uses 0 to represent null for polymorphic classes. Non-polymorphic classes use -1 like Itanium, since 0 is a valid field offset. Reviewers: rjmccall CC: timurrrr, cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D558 llvm-svn: 177753
2013-03-23 03:02:54 +08:00
struct POD {
int a;
int b;
};
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
struct Polymorphic {
virtual void myVirtual();
int a;
int b;
};
// This class uses the virtual inheritance model, yet its vbptr offset is not 0.
// We still use zero for the null field offset, despite it being a valid field
// offset.
struct NonZeroVBPtr : POD, Virtual {
int n;
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
void foo();
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
};
struct Unspecified;
[ms-cxxabi] Fix assert in unspecified inheritance memptr emission If a class is using the unspecified inheritance model for member pointers and later we find the class is defined to use single inheritance, zero out the vbptr offset field of the member pointer when it is formed. llvm-svn: 192664
2013-10-15 09:18:02 +08:00
struct UnspecSingle;
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
// Check that we can lower the LLVM types and get the null initializers right.
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
int Single ::*s_d_memptr;
int Polymorphic::*p_d_memptr;
int Multiple ::*m_d_memptr;
int Virtual ::*v_d_memptr;
int NonZeroVBPtr::*n_d_memptr;
int Unspecified::*u_d_memptr;
[ms-cxxabi] Fix assert in unspecified inheritance memptr emission If a class is using the unspecified inheritance model for member pointers and later we find the class is defined to use single inheritance, zero out the vbptr offset field of the member pointer when it is formed. llvm-svn: 192664
2013-10-15 09:18:02 +08:00
int UnspecSingle::*us_d_memptr;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?s_d_memptr@@3PQSingle@@HQ1@" = dso_local global i32 -1, align 4
// CHECK: @"?p_d_memptr@@3PQPolymorphic@@HQ1@" = dso_local global i32 0, align 4
// CHECK: @"?m_d_memptr@@3PQMultiple@@HQ1@" = dso_local global i32 -1, align 4
// CHECK: @"?v_d_memptr@@3PQVirtual@@HQ1@" = dso_local global { i32, i32 }
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: { i32 0, i32 -1 }, align 4
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?n_d_memptr@@3PQNonZeroVBPtr@@HQ1@" = dso_local global { i32, i32 }
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: { i32 0, i32 -1 }, align 4
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?u_d_memptr@@3PQUnspecified@@HQ1@" = dso_local global { i32, i32, i32 }
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: { i32 0, i32 0, i32 -1 }, align 4
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?us_d_memptr@@3PQUnspecSingle@@HQ1@" = dso_local global { i32, i32, i32 }
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: { i32 0, i32 0, i32 -1 }, align 4
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
void (Single ::*s_f_memptr)();
void (Multiple::*m_f_memptr)();
void (Virtual ::*v_f_memptr)();
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?s_f_memptr@@3P8Single@@AEXXZQ1@" = dso_local global i8* null, align 4
// CHECK: @"?m_f_memptr@@3P8Multiple@@AEXXZQ1@" = dso_local global { i8*, i32 } zeroinitializer, align 4
// CHECK: @"?v_f_memptr@@3P8Virtual@@AEXXZQ1@" = dso_local global { i8*, i32, i32 } zeroinitializer, align 4
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
// We can define Unspecified after locking in the inheritance model.
[ms-cxxabi] Fix vbptr offsets in memptrs when the vbptr is in an nvbase Also addresses a review comment from John from on r180985 by removing the "== -1" check, since it's now reusing the correct code which has the comment. llvm-svn: 183318
2013-06-05 23:58:29 +08:00
struct Unspecified : Multiple, Virtual {
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
void foo();
int u;
};
[ms-cxxabi] Fix assert in unspecified inheritance memptr emission If a class is using the unspecified inheritance model for member pointers and later we find the class is defined to use single inheritance, zero out the vbptr offset field of the member pointer when it is formed. llvm-svn: 192664
2013-10-15 09:18:02 +08:00
struct UnspecSingle {
void foo();
};
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// Test memptr emission in a constant expression.
namespace Const {
void (Single ::*s_f_mp)() = &Single::foo;
void (Multiple ::*m_f_mp)() = &B2::foo;
void (Virtual ::*v_f_mp)() = &Virtual::foo;
void (Unspecified::*u_f_mp)() = &Unspecified::foo;
[ms-cxxabi] Fix assert in unspecified inheritance memptr emission If a class is using the unspecified inheritance model for member pointers and later we find the class is defined to use single inheritance, zero out the vbptr offset field of the member pointer when it is formed. llvm-svn: 192664
2013-10-15 09:18:02 +08:00
void (UnspecSingle::*us_f_mp)() = &UnspecSingle::foo;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?s_f_mp@Const@@3P8Single@@AEXXZQ2@" =
// CHECK: global i8* bitcast ({{.*}} @"?foo@Single@@QAEXXZ" to i8*), align 4
// CHECK: @"?m_f_mp@Const@@3P8Multiple@@AEXXZQ2@" =
// CHECK: global { i8*, i32 } { i8* bitcast ({{.*}} @"?foo@B2@@QAEXXZ" to i8*), i32 4 }, align 4
// CHECK: @"?v_f_mp@Const@@3P8Virtual@@AEXXZQ2@" =
// CHECK: global { i8*, i32, i32 } { i8* bitcast ({{.*}} @"?foo@Virtual@@QAEXXZ" to i8*), i32 0, i32 0 }, align 4
// CHECK: @"?u_f_mp@Const@@3P8Unspecified@@AEXXZQ2@" =
// CHECK: global { i8*, i32, i32, i32 } { i8* bitcast ({{.*}} @"?foo@Unspecified@@QAEXXZ" to i8*), i32 0, i32 0, i32 0 }, align 4
// CHECK: @"?us_f_mp@Const@@3P8UnspecSingle@@AEXXZQ2@" =
// CHECK: global { i8*, i32, i32, i32 } { i8* bitcast ({{.*}} @"?foo@UnspecSingle@@QAEXXZ" to i8*), i32 0, i32 0, i32 0 }, align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
}
namespace CastParam {
// This exercises ConstExprEmitter instead of ValueDecl::evaluateValue. The
// extra reinterpret_cast for the parameter type requires more careful folding.
// FIXME: Or does it? If reinterpret_casts are no-ops, we should be able to
// strip them in evaluateValue() and just proceed as normal with an APValue.
struct A {
int a;
void foo(A *p);
};
struct B { int b; };
struct C : B, A { int c; };
void (A::*ptr1)(void *) = (void (A::*)(void *)) &A::foo;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?ptr1@CastParam@@3P8A@1@AEXPAX@ZQ21@" =
// CHECK: global i8* bitcast (void ({{.*}})* @"?foo@A@CastParam@@QAEXPAU12@@Z" to i8*), align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// Try a reinterpret_cast followed by a memptr conversion.
void (C::*ptr2)(void *) = (void (C::*)(void *)) (void (A::*)(void *)) &A::foo;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?ptr2@CastParam@@3P8C@1@AEXPAX@ZQ21@" =
// CHECK: global { i8*, i32 } { i8* bitcast (void ({{.*}})* @"?foo@A@CastParam@@QAEXPAU12@@Z" to i8*), i32 4 }, align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
void (C::*ptr3)(void *) = (void (C::*)(void *)) (void (A::*)(void *)) (void (A::*)(A *)) 0;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?ptr3@CastParam@@3P8C@1@AEXPAX@ZQ21@" =
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: global { i8*, i32 } zeroinitializer, align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
struct D : C {
virtual void isPolymorphic();
int d;
};
// Try a cast that changes the inheritance model. Null for D is 0, but null for
// C is -1. We need the cast to long in order to hit the non-APValue path.
int C::*ptr4 = (int C::*) (int D::*) (long D::*) 0;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?ptr4@CastParam@@3PQC@1@HQ21@" = dso_local global i32 -1, align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// MSVC rejects this but we accept it.
int C::*ptr5 = (int C::*) (long D::*) 0;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: @"?ptr5@CastParam@@3PQC@1@HQ21@" = dso_local global i32 -1, align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
}
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
struct UnspecWithVBPtr;
int UnspecWithVBPtr::*forceUnspecWithVBPtr;
struct UnspecWithVBPtr : B1, virtual B2 {
int u;
void foo();
};
// Test emitting non-virtual member pointers in a non-constexpr setting.
void EmitNonVirtualMemberPointers() {
void (Single ::*s_f_memptr)() = &Single::foo;
void (Multiple ::*m_f_memptr)() = &Multiple::foo;
void (Virtual ::*v_f_memptr)() = &Virtual::foo;
void (Unspecified::*u_f_memptr)() = &Unspecified::foo;
void (UnspecWithVBPtr::*u2_f_memptr)() = &UnspecWithVBPtr::foo;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local void @"?EmitNonVirtualMemberPointers@@YAXXZ"() {{.*}} {
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
// CHECK: alloca i8*, align 4
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: alloca { i8*, i32 }, align 4
// CHECK: alloca { i8*, i32, i32 }, align 4
// CHECK: alloca { i8*, i32, i32, i32 }, align 4
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: store i8* bitcast (void (%{{.*}}*)* @"?foo@Single@@QAEXXZ" to i8*), i8** %{{.*}}, align 4
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
// CHECK: store { i8*, i32 }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: { i8* bitcast (void (%{{.*}}*)* @"?foo@Multiple@@QAEXXZ" to i8*), i32 0 },
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: { i8*, i32 }* %{{.*}}, align 4
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
// CHECK: store { i8*, i32, i32 }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: { i8* bitcast (void (%{{.*}}*)* @"?foo@Virtual@@QAEXXZ" to i8*), i32 0, i32 0 },
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: { i8*, i32, i32 }* %{{.*}}, align 4
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
// CHECK: store { i8*, i32, i32, i32 }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: { i8* bitcast (void (%{{.*}}*)* @"?foo@Unspecified@@QAEXXZ" to i8*), i32 0, i32 0, i32 0 },
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: { i8*, i32, i32, i32 }* %{{.*}}, align 4
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
// CHECK: store { i8*, i32, i32, i32 }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: { i8* bitcast (void (%{{.*}}*)* @"?foo@UnspecWithVBPtr@@QAEXXZ" to i8*),
Forgot to update to a test in r240043 llvm-svn: 240049
2015-06-19 04:34:05 +08:00
// CHECK: i32 0, i32 0, i32 0 },
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: { i8*, i32, i32, i32 }* %{{.*}}, align 4
[ms-cxxabi] Emit non-virtual member function pointers Without any conversion, this is pretty straightforward. Most of the fields can be zeros. The order is: - field offset or pointer - nonvirtual adjustment (for MI functions) - vbptr offset (for unspecified) - virtual adjustment offset (for virtual inheritance) Differential Revision: http://llvm-reviews.chandlerc.com/D699 llvm-svn: 180985
2013-05-03 09:15:11 +08:00
// CHECK: ret void
// CHECK: }
}
[ms-cxxabi] Implement member data pointers for non-dynamic classes Summary: For non-dynamic classes (no virtual bases), member data pointers are simple offsets from the base of the record. Dynamic classes use an aggregate for member data pointers and are therefore currently unsupported. Unlike Itanium, the ms ABI uses 0 to represent null for polymorphic classes. Non-polymorphic classes use -1 like Itanium, since 0 is a valid field offset. Reviewers: rjmccall CC: timurrrr, cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D558 llvm-svn: 177753
2013-03-23 03:02:54 +08:00
void podMemPtrs() {
int POD::*memptr;
memptr = &POD::a;
memptr = &POD::b;
if (memptr)
memptr = 0;
// Check that member pointers use the right offsets and that null is -1.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local void @"?podMemPtrs@@YAXXZ"() {{.*}} {
[ms-cxxabi] Implement member data pointers for non-dynamic classes Summary: For non-dynamic classes (no virtual bases), member data pointers are simple offsets from the base of the record. Dynamic classes use an aggregate for member data pointers and are therefore currently unsupported. Unlike Itanium, the ms ABI uses 0 to represent null for polymorphic classes. Non-polymorphic classes use -1 like Itanium, since 0 is a valid field offset. Reviewers: rjmccall CC: timurrrr, cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D558 llvm-svn: 177753
2013-03-23 03:02:54 +08:00
// CHECK: %[[memptr:.*]] = alloca i32, align 4
// CHECK-NEXT: store i32 0, i32* %[[memptr]], align 4
// CHECK-NEXT: store i32 4, i32* %[[memptr]], align 4
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: %[[memptr_val:.*]] = load i32, i32* %[[memptr]], align 4
[ms-cxxabi] Implement member data pointers for non-dynamic classes Summary: For non-dynamic classes (no virtual bases), member data pointers are simple offsets from the base of the record. Dynamic classes use an aggregate for member data pointers and are therefore currently unsupported. Unlike Itanium, the ms ABI uses 0 to represent null for polymorphic classes. Non-polymorphic classes use -1 like Itanium, since 0 is a valid field offset. Reviewers: rjmccall CC: timurrrr, cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D558 llvm-svn: 177753
2013-03-23 03:02:54 +08:00
// CHECK-NEXT: %{{.*}} = icmp ne i32 %[[memptr_val]], -1
// CHECK-NEXT: br i1 %{{.*}}, label %{{.*}}, label %{{.*}}
// CHECK: store i32 -1, i32* %[[memptr]], align 4
// CHECK: ret void
// CHECK: }
}
void polymorphicMemPtrs() {
int Polymorphic::*memptr;
memptr = &Polymorphic::a;
memptr = &Polymorphic::b;
if (memptr)
memptr = 0;
// Member pointers for polymorphic classes include the vtable slot in their
// offset and use 0 to represent null.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local void @"?polymorphicMemPtrs@@YAXXZ"() {{.*}} {
[ms-cxxabi] Implement member data pointers for non-dynamic classes Summary: For non-dynamic classes (no virtual bases), member data pointers are simple offsets from the base of the record. Dynamic classes use an aggregate for member data pointers and are therefore currently unsupported. Unlike Itanium, the ms ABI uses 0 to represent null for polymorphic classes. Non-polymorphic classes use -1 like Itanium, since 0 is a valid field offset. Reviewers: rjmccall CC: timurrrr, cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D558 llvm-svn: 177753
2013-03-23 03:02:54 +08:00
// CHECK: %[[memptr:.*]] = alloca i32, align 4
// CHECK-NEXT: store i32 4, i32* %[[memptr]], align 4
// CHECK-NEXT: store i32 8, i32* %[[memptr]], align 4
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: %[[memptr_val:.*]] = load i32, i32* %[[memptr]], align 4
[ms-cxxabi] Implement member data pointers for non-dynamic classes Summary: For non-dynamic classes (no virtual bases), member data pointers are simple offsets from the base of the record. Dynamic classes use an aggregate for member data pointers and are therefore currently unsupported. Unlike Itanium, the ms ABI uses 0 to represent null for polymorphic classes. Non-polymorphic classes use -1 like Itanium, since 0 is a valid field offset. Reviewers: rjmccall CC: timurrrr, cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D558 llvm-svn: 177753
2013-03-23 03:02:54 +08:00
// CHECK-NEXT: %{{.*}} = icmp ne i32 %[[memptr_val]], 0
// CHECK-NEXT: br i1 %{{.*}}, label %{{.*}}, label %{{.*}}
// CHECK: store i32 0, i32* %[[memptr]], align 4
// CHECK: ret void
// CHECK: }
}
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
bool nullTestDataUnspecified(int Unspecified::*mp) {
return mp;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local zeroext i1 @"?nullTestDataUnspecified@@YA_NPQUnspecified@@H@Z"{{.*}} {
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: %{{.*}} = load { i32, i32, i32 }, { i32, i32, i32 }* %{{.*}}, align 4
// CHECK: store { i32, i32, i32 } {{.*}} align 4
// CHECK: %[[mp:.*]] = load { i32, i32, i32 }, { i32, i32, i32 }* %{{.*}}, align 4
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[mp0:.*]] = extractvalue { i32, i32, i32 } %[[mp]], 0
// CHECK: %[[cmp0:.*]] = icmp ne i32 %[[mp0]], 0
// CHECK: %[[mp1:.*]] = extractvalue { i32, i32, i32 } %[[mp]], 1
// CHECK: %[[cmp1:.*]] = icmp ne i32 %[[mp1]], 0
MS ABI: Fix logic bug in member pointer null test code This code is trying to test if the pointer is *not* null. Therefore we should use 'or' instead of 'and' to combine the results of 'icmp ne'. This logic is consistent with the general member pointer comparison code in EmitMemberPointerComparison. llvm-svn: 207815
2014-05-02 08:05:16 +08:00
// CHECK: %[[and0:.*]] = or i1 %[[cmp0]], %[[cmp1]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[mp2:.*]] = extractvalue { i32, i32, i32 } %[[mp]], 2
// CHECK: %[[cmp2:.*]] = icmp ne i32 %[[mp2]], -1
MS ABI: Fix logic bug in member pointer null test code This code is trying to test if the pointer is *not* null. Therefore we should use 'or' instead of 'and' to combine the results of 'icmp ne'. This logic is consistent with the general member pointer comparison code in EmitMemberPointerComparison. llvm-svn: 207815
2014-05-02 08:05:16 +08:00
// CHECK: %[[and1:.*]] = or i1 %[[and0]], %[[cmp2]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: ret i1 %[[and1]]
// CHECK: }
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
// Pass this large type indirectly.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// X64-LABEL: define dso_local zeroext i1 @"?nullTestDataUnspecified@@
IR: print value numbers for unnamed function arguments For consistency with normal instructions and clarity when reading IR, it's best to print the %0, %1, ... names of function arguments in definitions. Also modifies the parser to accept IR in that form for obvious reasons. llvm-svn: 367755
2019-08-03 22:28:34 +08:00
// X64: ({ i32, i32, i32 }* %0)
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
}
bool nullTestFunctionUnspecified(void (Unspecified::*mp)()) {
return mp;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local zeroext i1 @"?nullTestFunctionUnspecified@@YA_NP8Unspecified@@AEXXZ@Z"{{.*}} {
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: %{{.*}} = load { i8*, i32, i32, i32 }, { i8*, i32, i32, i32 }* %{{.*}}, align 4
// CHECK: store { i8*, i32, i32, i32 } {{.*}} align 4
// CHECK: %[[mp:.*]] = load { i8*, i32, i32, i32 }, { i8*, i32, i32, i32 }* %{{.*}}, align 4
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[mp0:.*]] = extractvalue { i8*, i32, i32, i32 } %[[mp]], 0
// CHECK: %[[cmp0:.*]] = icmp ne i8* %[[mp0]], null
// CHECK: ret i1 %[[cmp0]]
// CHECK: }
}
int loadDataMemberPointerVirtual(Virtual *o, int Virtual::*memptr) {
return o->*memptr;
// Test that we can unpack this aggregate member pointer and load the member
// data pointer.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local i32 @"?loadDataMemberPointerVirtual@@YAHPAUVirtual@@PQ1@H@Z"{{.*}} {
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[o:.*]] = load %{{.*}}*, %{{.*}}** %{{.*}}, align 4
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: %[[memptr:.*]] = load { i32, i32 }, { i32, i32 }* %{{.*}}, align 4
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[memptr0:.*]] = extractvalue { i32, i32 } %[[memptr:.*]], 0
// CHECK: %[[memptr1:.*]] = extractvalue { i32, i32 } %[[memptr:.*]], 1
// CHECK: %[[v6:.*]] = bitcast %{{.*}}* %[[o]] to i8*
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[vbptr:.*]] = getelementptr inbounds i8, i8* %[[v6]], i32 0
MS ABI: Emit more canonical vbptr stores and loads This eliminates some i8* GEPs and makes the IR that clang emits a bit more canonical. More work is needed for vftables, but that isn't a clear win so I plan to send it for review. llvm-svn: 220398
2014-10-23 01:26:00 +08:00
// CHECK: %[[vbptr_a:.*]] = bitcast i8* %[[vbptr]] to i32**
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[vbtable:.*]] = load i32*, i32** %[[vbptr_a:.*]]
MS ABI: Emit more canonical vbptr stores and loads This eliminates some i8* GEPs and makes the IR that clang emits a bit more canonical. More work is needed for vftables, but that isn't a clear win so I plan to send it for review. llvm-svn: 220398
2014-10-23 01:26:00 +08:00
// CHECK: %[[memptr1_shr:.*]] = ashr exact i32 %[[memptr1]], 2
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[v7:.*]] = getelementptr inbounds i32, i32* %[[vbtable]], i32 %[[memptr1_shr]]
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[vbase_offs:.*]] = load i32, i32* %[[v7]]
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[v10:.*]] = getelementptr inbounds i8, i8* %[[vbptr]], i32 %[[vbase_offs]]
// CHECK: %[[offset:.*]] = getelementptr inbounds i8, i8* %[[v10]], i32 %[[memptr0]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[v11:.*]] = bitcast i8* %[[offset]] to i32*
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[v12:.*]] = load i32, i32* %[[v11]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: ret i32 %[[v12]]
// CHECK: }
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
// A two-field data memptr on x64 gets coerced to i64 and is passed in a
// register or memory.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// X64-LABEL: define dso_local i32 @"?loadDataMemberPointerVirtual@@YAHPEAUVirtual@@PEQ1@H@Z"
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
// X64: (%struct.Virtual* %o, i64 %memptr.coerce)
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
}
int loadDataMemberPointerUnspecified(Unspecified *o, int Unspecified::*memptr) {
return o->*memptr;
// Test that we can unpack this aggregate member pointer and load the member
// data pointer.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local i32 @"?loadDataMemberPointerUnspecified@@YAHPAUUnspecified@@PQ1@H@Z"{{.*}} {
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[o:.*]] = load %{{.*}}*, %{{.*}}** %{{.*}}, align 4
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: %[[memptr:.*]] = load { i32, i32, i32 }, { i32, i32, i32 }* %{{.*}}, align 4
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[memptr0:.*]] = extractvalue { i32, i32, i32 } %[[memptr:.*]], 0
// CHECK: %[[memptr1:.*]] = extractvalue { i32, i32, i32 } %[[memptr:.*]], 1
// CHECK: %[[memptr2:.*]] = extractvalue { i32, i32, i32 } %[[memptr:.*]], 2
// CHECK: %[[base:.*]] = bitcast %{{.*}}* %[[o]] to i8*
// CHECK: %[[is_vbase:.*]] = icmp ne i32 %[[memptr2]], 0
// CHECK: br i1 %[[is_vbase]], label %[[vadjust:.*]], label %[[skip:.*]]
//
Widen the checks in the ms abi memptr test to work under NDEBUG llvm-svn: 179311
2013-04-12 03:01:17 +08:00
// CHECK: [[vadjust]]
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[vbptr:.*]] = getelementptr inbounds i8, i8* %[[base]], i32 %[[memptr1]]
MS ABI: Emit more canonical vbptr stores and loads This eliminates some i8* GEPs and makes the IR that clang emits a bit more canonical. More work is needed for vftables, but that isn't a clear win so I plan to send it for review. llvm-svn: 220398
2014-10-23 01:26:00 +08:00
// CHECK: %[[vbptr_a:.*]] = bitcast i8* %[[vbptr]] to i32**
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[vbtable:.*]] = load i32*, i32** %[[vbptr_a:.*]]
MS ABI: Emit more canonical vbptr stores and loads This eliminates some i8* GEPs and makes the IR that clang emits a bit more canonical. More work is needed for vftables, but that isn't a clear win so I plan to send it for review. llvm-svn: 220398
2014-10-23 01:26:00 +08:00
// CHECK: %[[memptr2_shr:.*]] = ashr exact i32 %[[memptr2]], 2
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[v7:.*]] = getelementptr inbounds i32, i32* %[[vbtable]], i32 %[[memptr2_shr]]
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[vbase_offs:.*]] = load i32, i32* %[[v7]]
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[base_adj:.*]] = getelementptr inbounds i8, i8* %[[vbptr]], i32 %[[vbase_offs]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
//
Widen the checks in the ms abi memptr test to work under NDEBUG llvm-svn: 179311
2013-04-12 03:01:17 +08:00
// CHECK: [[skip]]
// CHECK: %[[new_base:.*]] = phi i8* [ %[[base]], %{{.*}} ], [ %[[base_adj]], %[[vadjust]] ]
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[offset:.*]] = getelementptr inbounds i8, i8* %[[new_base]], i32 %[[memptr0]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[v11:.*]] = bitcast i8* %[[offset]] to i32*
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[v12:.*]] = load i32, i32* %[[v11]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: ret i32 %[[v12]]
// CHECK: }
}
void callMemberPointerSingle(Single *o, void (Single::*memptr)()) {
(o->*memptr)();
// Just look for an indirect thiscall.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local void @"?callMemberPointerSingle@@{{.*}} {{.*}} {
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: call x86_thiscallcc void %{{.*}}(%{{.*}} %{{.*}})
// CHECK: ret void
// CHECK: }
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// X64-LABEL: define dso_local void @"?callMemberPointerSingle@@
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
// X64: (%struct.Single* %o, i8* %memptr)
// X64: bitcast i8* %{{[^ ]*}} to void (%struct.Single*)*
// X64: ret void
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
}
void callMemberPointerMultiple(Multiple *o, void (Multiple::*memptr)()) {
(o->*memptr)();
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local void @"?callMemberPointerMultiple@@{{.*}} {
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[memptr0:.*]] = extractvalue { i8*, i32 } %{{.*}}, 0
// CHECK: %[[memptr1:.*]] = extractvalue { i8*, i32 } %{{.*}}, 1
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[this_adjusted:.*]] = getelementptr inbounds i8, i8* %{{.*}}, i32 %[[memptr1]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[this:.*]] = bitcast i8* %[[this_adjusted]] to {{.*}}
// CHECK: %[[fptr:.*]] = bitcast i8* %[[memptr0]] to {{.*}}
// CHECK: call x86_thiscallcc void %[[fptr]](%{{.*}} %[[this]])
// CHECK: ret void
// CHECK: }
}
void callMemberPointerVirtualBase(Virtual *o, void (Virtual::*memptr)()) {
(o->*memptr)();
// This shares a lot with virtual data member pointers.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local void @"?callMemberPointerVirtualBase@@{{.*}} {
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[memptr0:.*]] = extractvalue { i8*, i32, i32 } %{{.*}}, 0
// CHECK: %[[memptr1:.*]] = extractvalue { i8*, i32, i32 } %{{.*}}, 1
// CHECK: %[[memptr2:.*]] = extractvalue { i8*, i32, i32 } %{{.*}}, 2
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[vbptr:.*]] = getelementptr inbounds i8, i8* %{{.*}}, i32 0
MS ABI: Emit more canonical vbptr stores and loads This eliminates some i8* GEPs and makes the IR that clang emits a bit more canonical. More work is needed for vftables, but that isn't a clear win so I plan to send it for review. llvm-svn: 220398
2014-10-23 01:26:00 +08:00
// CHECK: %[[vbptr_a:.*]] = bitcast i8* %[[vbptr]] to i32**
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[vbtable:.*]] = load i32*, i32** %[[vbptr_a:.*]]
MS ABI: Emit more canonical vbptr stores and loads This eliminates some i8* GEPs and makes the IR that clang emits a bit more canonical. More work is needed for vftables, but that isn't a clear win so I plan to send it for review. llvm-svn: 220398
2014-10-23 01:26:00 +08:00
// CHECK: %[[memptr2_shr:.*]] = ashr exact i32 %[[memptr2]], 2
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[v7:.*]] = getelementptr inbounds i32, i32* %[[vbtable]], i32 %[[memptr2_shr]]
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[vbase_offs:.*]] = load i32, i32* %[[v7]]
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: %[[v10:.*]] = getelementptr inbounds i8, i8* %[[vbptr]], i32 %[[vbase_offs]]
// CHECK: %[[this_adjusted:.*]] = getelementptr inbounds i8, i8* %[[v10]], i32 %[[memptr1]]
[ms-cxxabi] Implement member pointer emission and dereferencing Summary: Handles all inheritance models for both data and function member pointers. Also implements isZeroInitializable() and refactors some of the null member pointer code. MSVC supports converting member pointers through virtual bases, which clang does not (yet?) support. Implementing that extension is covered by http://llvm.org/15713 Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D613 llvm-svn: 179305
2013-04-12 02:13:19 +08:00
// CHECK: %[[fptr:.*]] = bitcast i8* %[[memptr0]] to void ({{.*}})
// CHECK: %[[this:.*]] = bitcast i8* %[[this_adjusted]] to {{.*}}
// CHECK: call x86_thiscallcc void %[[fptr]](%{{.*}} %[[this]])
// CHECK: ret void
// CHECK: }
}
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
bool compareSingleFunctionMemptr(void (Single::*l)(), void (Single::*r)()) {
return l == r;
// Should only be one comparison here.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local zeroext i1 @"?compareSingleFunctionMemptr@@YA_NP8Single@@AEXXZ0@Z"{{.*}} {
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
// CHECK-NOT: icmp
// CHECK: %[[r:.*]] = icmp eq
// CHECK-NOT: icmp
// CHECK: ret i1 %[[r]]
// CHECK: }
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// X64-LABEL: define dso_local zeroext i1 @"?compareSingleFunctionMemptr@@
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
// X64: (i8* %{{[^,]*}}, i8* %{{[^)]*}})
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
}
bool compareNeqSingleFunctionMemptr(void (Single::*l)(), void (Single::*r)()) {
return l != r;
// Should only be one comparison here.
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local zeroext i1 @"?compareNeqSingleFunctionMemptr@@YA_NP8Single@@AEXXZ0@Z"{{.*}} {
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
// CHECK-NOT: icmp
// CHECK: %[[r:.*]] = icmp ne
// CHECK-NOT: icmp
// CHECK: ret i1 %[[r]]
// CHECK: }
}
bool unspecFuncMemptrEq(void (Unspecified::*l)(), void (Unspecified::*r)()) {
return l == r;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local zeroext i1 @"?unspecFuncMemptrEq@@YA_NP8Unspecified@@AEXXZ0@Z"{{.*}} {
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
// CHECK: %[[lhs0:.*]] = extractvalue { i8*, i32, i32, i32 } %[[l:.*]], 0
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[r:.*]], 0
// CHECK: %[[cmp0:.*]] = icmp eq i8* %[[lhs0]], %{{.*}}
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[l]], 1
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[r]], 1
// CHECK: %[[cmp1:.*]] = icmp eq i32
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[l]], 2
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[r]], 2
// CHECK: %[[cmp2:.*]] = icmp eq i32
// CHECK: %[[res12:.*]] = and i1 %[[cmp1]], %[[cmp2]]
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[l]], 3
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[r]], 3
// CHECK: %[[cmp3:.*]] = icmp eq i32
// CHECK: %[[res123:.*]] = and i1 %[[res12]], %[[cmp3]]
// CHECK: %[[iszero:.*]] = icmp eq i8* %[[lhs0]], null
// CHECK: %[[bits_or_null:.*]] = or i1 %[[res123]], %[[iszero]]
// CHECK: %{{.*}} = and i1 %[[bits_or_null]], %[[cmp0]]
// CHECK: ret i1 %{{.*}}
// CHECK: }
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// X64-LABEL: define dso_local zeroext i1 @"?unspecFuncMemptrEq@@
IR: print value numbers for unnamed function arguments For consistency with normal instructions and clarity when reading IR, it's best to print the %0, %1, ... names of function arguments in definitions. Also modifies the parser to accept IR in that form for obvious reasons. llvm-svn: 367755
2019-08-03 22:28:34 +08:00
// X64: ({ i8*, i32, i32, i32 }* %0, { i8*, i32, i32, i32 }* %1)
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
}
bool unspecFuncMemptrNeq(void (Unspecified::*l)(), void (Unspecified::*r)()) {
return l != r;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local zeroext i1 @"?unspecFuncMemptrNeq@@YA_NP8Unspecified@@AEXXZ0@Z"{{.*}} {
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
// CHECK: %[[lhs0:.*]] = extractvalue { i8*, i32, i32, i32 } %[[l:.*]], 0
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[r:.*]], 0
// CHECK: %[[cmp0:.*]] = icmp ne i8* %[[lhs0]], %{{.*}}
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[l]], 1
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[r]], 1
// CHECK: %[[cmp1:.*]] = icmp ne i32
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[l]], 2
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[r]], 2
// CHECK: %[[cmp2:.*]] = icmp ne i32
// CHECK: %[[res12:.*]] = or i1 %[[cmp1]], %[[cmp2]]
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[l]], 3
// CHECK: %{{.*}} = extractvalue { i8*, i32, i32, i32 } %[[r]], 3
// CHECK: %[[cmp3:.*]] = icmp ne i32
// CHECK: %[[res123:.*]] = or i1 %[[res12]], %[[cmp3]]
// CHECK: %[[iszero:.*]] = icmp ne i8* %[[lhs0]], null
// CHECK: %[[bits_or_null:.*]] = and i1 %[[res123]], %[[iszero]]
// CHECK: %{{.*}} = or i1 %[[bits_or_null]], %[[cmp0]]
// CHECK: ret i1 %{{.*}}
// CHECK: }
}
bool unspecDataMemptrEq(int Unspecified::*l, int Unspecified::*r) {
return l == r;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local zeroext i1 @"?unspecDataMemptrEq@@YA_NPQUnspecified@@H0@Z"{{.*}} {
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
// CHECK: extractvalue { i32, i32, i32 } %{{.*}}, 0
// CHECK: extractvalue { i32, i32, i32 } %{{.*}}, 0
// CHECK: icmp eq i32
// CHECK: extractvalue { i32, i32, i32 } %{{.*}}, 1
// CHECK: extractvalue { i32, i32, i32 } %{{.*}}, 1
// CHECK: icmp eq i32
// CHECK: extractvalue { i32, i32, i32 } %{{.*}}, 2
// CHECK: extractvalue { i32, i32, i32 } %{{.*}}, 2
// CHECK: icmp eq i32
// CHECK: and i1
// CHECK: and i1
// CHECK: ret i1
// CHECK: }
Win64: Pass member pointers larger than 8 bytes by reference The Win64 ABI docs on MSDN say that arguments bigger than 8 bytes are passed by reference. Prior to this change, we were only applying this logic to RecordType arguments. This affects both the Itanium and Microsoft C++ ABIs. Reviewers: majnemer Differential Revision: http://reviews.llvm.org/D3587 llvm-svn: 207817
2014-05-02 08:51:20 +08:00
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// X64-LABEL: define dso_local zeroext i1 @"?unspecDataMemptrEq@@
IR: print value numbers for unnamed function arguments For consistency with normal instructions and clarity when reading IR, it's best to print the %0, %1, ... names of function arguments in definitions. Also modifies the parser to accept IR in that form for obvious reasons. llvm-svn: 367755
2019-08-03 22:28:34 +08:00
// X64: ({ i32, i32, i32 }* %0, { i32, i32, i32 }* %1)
[ms-cxxabi] Implement member pointer comparisons Summary: Like Itanium, comparisons are basically bitwise comparisons of the two values, with an exception for null member function pointers. If two function pointers are null, only the function pointer field matters for comparison purposes. The rest of the bits can be arbitrary. We take advantage of this in isZeroInitializable(), and it may matter once we start emitting conversions. Reviewers: rjmccall CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D695 llvm-svn: 180800
2013-05-01 04:15:14 +08:00
}
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
void (Multiple::*convertB2FuncToMultiple(void (B2::*mp)()))() {
return mp;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local i64 @"?convertB2FuncToMultiple@@YAP8Multiple@@AEXXZP8B2@@AEXXZ@Z"{{.*}} {
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK: store
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[mp:.*]] = load i8*, i8** %{{.*}}, align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK: icmp ne i8* %[[mp]], null
// CHECK: br i1 %{{.*}} label %{{.*}}, label %{{.*}}
//
// memptr.convert: ; preds = %entry
// CHECK: insertvalue { i8*, i32 } undef, i8* %[[mp]], 0
// CHECK: insertvalue { i8*, i32 } %{{.*}}, i32 4, 1
// CHECK: br label
//
// memptr.converted: ; preds = %memptr.convert, %entry
// CHECK: phi { i8*, i32 } [ zeroinitializer, %{{.*}} ], [ {{.*}} ]
// CHECK: }
}
void (B2::*convertMultipleFuncToB2(void (Multiple::*mp)()))() {
// FIXME: cl emits warning C4407 on this code because of the representation
// change. We might want to do the same.
return static_cast<void (B2::*)()>(mp);
// FIXME: We should return i8* instead of i32 here. The ptrtoint cast prevents
// LLVM from optimizing away the branch. This is likely a bug in
// lib/CodeGen/TargetInfo.cpp with how we classify memptr types for returns.
//
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local i32 @"?convertMultipleFuncToB2@@YAP8B2@@AEXXZP8Multiple@@AEXXZ@Z"{{.*}} {
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK: store
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: %[[src:.*]] = load { i8*, i32 }, { i8*, i32 }* %{{.*}}, align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK: extractvalue { i8*, i32 } %[[src]], 0
// CHECK: icmp ne i8* %{{.*}}, null
// CHECK: br i1 %{{.*}}, label %{{.*}}, label %{{.*}}
//
// memptr.convert: ; preds = %entry
// CHECK: %[[fp:.*]] = extractvalue { i8*, i32 } %[[src]], 0
// CHECK: br label
//
// memptr.converted: ; preds = %memptr.convert, %entry
// CHECK: phi i8* [ null, %{{.*}} ], [ %[[fp]], %{{.*}} ]
// CHECK: }
}
namespace Test1 {
struct A { int a; };
struct B { int b; };
struct C : virtual A { int c; };
struct D : B, C { int d; };
void (D::*convertCToD(void (C::*mp)()))() {
return mp;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local void @"?convertCToD@Test1@@YAP8D@1@AEXXZP8C@1@AEXXZ@Z"{{.*}} {
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK: store
[MS ABI] Fix the preferred alignment of member pointers Member pointers in the MS ABI have different alignment depending on whether they were created on the stack or live in a record. llvm-svn: 235681
2015-04-24 09:25:05 +08:00
// CHECK: load { i8*, i32, i32 }, { i8*, i32, i32 }* %{{.*}}, align 4
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK: extractvalue { i8*, i32, i32 } %{{.*}}, 0
// CHECK: icmp ne i8* %{{.*}}, null
// CHECK: br i1 %{{.*}}, label %{{.*}}, label %{{.*}}
//
// memptr.convert: ; preds = %entry
// CHECK: extractvalue { i8*, i32, i32 } %{{.*}}, 0
[MS ABI] Rework member pointer conversion Member pointers in the MS ABI are made complicated due to the following: - Virtual methods in the most derived class (MDC) might live in a vftable in a virtual base. - There are four different representations of member pointer: single inheritance, multiple inheritance, virtual inheritance and the "most general" representation. - Bases might have a *more* general representation than classes which derived from them, a most surprising result. We believed that we could treat all member pointers as-if they were a degenerate case of the multiple inheritance model. This fell apart once we realized that implementing standard member pointers using this ABI requires referencing members with a non-zero vbindex. On a bright note, all but the virtual inheritance model operate rather similarly. The virtual inheritance member pointer representation awkwardly requires a virtual base adjustment in order to refer to entities in the MDC. However, the first virtual base might be quite far from the start of the virtual base. This means that we must add a negative non-virtual displacement. However, things get even more complicated. The most general representation interprets vbindex zero differently from the virtual inheritance model: it doesn't reference the vbtable at all. It turns out that this complexity can increase for quite some time: consider a derived to base conversion from the most general model to the multiple inheritance model... To manage this complexity we introduce a concept of "normalized" member pointer which allows us to treat all three models as the most general model. Then we try to figure out how to map this generalized member pointer onto the destination member pointer model. I've done my best to furnish the code with comments explaining why each adjustment is performed. This fixes PR23878. llvm-svn: 240384
2015-06-23 15:31:11 +08:00
// CHECK: %[[nvoff:.*]] = extractvalue { i8*, i32, i32 } %{{.*}}, 1
// CHECK: %[[vbidx:.*]] = extractvalue { i8*, i32, i32 } %{{.*}}, 2
// CHECK: %[[is_nvbase:.*]] = icmp eq i32 %[[vbidx]], 0
// CHECK: %[[nv_disp:.*]] = add nsw i32 %[[nvoff]], 4
// CHECK: %[[nv_adj:.*]] = select i1 %[[is_nvbase]], i32 %[[nv_disp]], i32 0
// CHECK: %[[dst_adj:.*]] = select i1 %[[is_nvbase]], i32 4, i32 0
// CHECK: %[[adj:.*]] = sub nsw i32 %[[nv_adj]], %[[dst_adj]]
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK: insertvalue { i8*, i32, i32 } undef, i8* {{.*}}, 0
// CHECK: insertvalue { i8*, i32, i32 } {{.*}}, i32 %[[adj]], 1
// CHECK: insertvalue { i8*, i32, i32 } {{.*}}, i32 {{.*}}, 2
// CHECK: br label
//
// memptr.converted: ; preds = %memptr.convert, %entry
// CHECK: phi { i8*, i32, i32 } [ { i8* null, i32 0, i32 -1 }, {{.*}} ], [ {{.*}} ]
// CHECK: }
}
}
namespace Test2 {
// Test that we dynamically convert between different null reps.
struct A { int a; };
struct B : A { int b; };
struct C : A {
int c;
virtual void hasVfPtr();
};
int A::*reinterpret(int B::*mp) {
return reinterpret_cast<int A::*>(mp);
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local i32 @"?reinterpret@Test2@@YAPQA@1@HPQB@1@H@Z"{{.*}} {
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK-NOT: select
// CHECK: ret i32
// CHECK: }
}
int A::*reinterpret(int C::*mp) {
return reinterpret_cast<int A::*>(mp);
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: define dso_local i32 @"?reinterpret@Test2@@YAPQA@1@HPQC@1@H@Z"{{.*}} {
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[mp:.*]] = load i32, i32*
[ms-cxxabi] Implement member pointer conversions Summary: This only supports converting along non-virtual inheritance paths by changing the field offset or the non-virtual base adjustment. This implements three kinds of conversions: - codegen for Value conversions - Constant emission for APValue - Constant folding for CastExprs In almost all constant initialization settings EmitMemberPointer(APValue) is called, except when the expression contains a reinterpret cast. reinterpret casts end up being a big corner case because the null value changes between different kinds of member pointers. Reviewers: rsmith CC: cfe-commits Differential Revision: http://llvm-reviews.chandlerc.com/D741 llvm-svn: 181543
2013-05-10 05:01:17 +08:00
// CHECK: %[[cmp:.*]] = icmp ne i32 %[[mp]], 0
// CHECK: select i1 %[[cmp]], i32 %[[mp]], i32 -1
// CHECK: }
}
}
[ms-cxxabi] bitcast to i8* to deref a data member pointer This was causing us to miscompile llvm::SymbolTableListTraits::getListOwner(), which uses data member pointers. llvm-svn: 196545
2013-12-06 06:44:07 +08:00
namespace Test3 {
// Make sure we cast 'this' to i8* before using GEP.
struct A {
int a;
int b;
};
int *load_data(A *a, int A::*mp) {
return &(a->*mp);
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define dso_local i32* @"?load_data@Test3@@YAPAHPAUA@1@PQ21@H@Z"{{.*}} {
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: %[[a:.*]] = load %"struct.Test3::A"*, %"struct.Test3::A"** %{{.*}}, align 4
// CHECK: %[[mp:.*]] = load i32, i32* %{{.*}}, align 4
[ms-cxxabi] bitcast to i8* to deref a data member pointer This was causing us to miscompile llvm::SymbolTableListTraits::getListOwner(), which uses data member pointers. llvm-svn: 196545
2013-12-06 06:44:07 +08:00
// CHECK: %[[a_i8:.*]] = bitcast %"struct.Test3::A"* %[[a]] to i8*
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: getelementptr inbounds i8, i8* %[[a_i8]], i32 %[[mp]]
[ms-cxxabi] bitcast to i8* to deref a data member pointer This was causing us to miscompile llvm::SymbolTableListTraits::getListOwner(), which uses data member pointers. llvm-svn: 196545
2013-12-06 06:44:07 +08:00
// CHECK: }
}
}
MS ABI: Include the vfptr offset in memptrs to virtual methods Virtual methods expect 'this' to point to the vfptr containing the virtual method, and this extends to virtual member pointer thunks. The relevant vfptr is always at offset zero on entry to the thunk, and no this adjustment is needed. Previously we would not include the vfptr adjustment in the member pointer, and we'd look at the vfptr offset when loading from the vftable in the thunk. Fixes PR18917. llvm-svn: 201835
2014-02-21 10:27:32 +08:00
namespace Test4 {
struct A { virtual void f(); };
struct B { virtual void g(); };
struct C : A, B { virtual void g(); };
void (C::*getmp())() {
return &C::g;
}
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define dso_local i64 @"?getmp@Test4@@YAP8C@1@AEXXZXZ"()
// CHECK: store { i8*, i32 } { i8* bitcast (void (%"struct.Test4::C"*, ...)* @"??_9C@Test4@@$BA@AE" to i8*), i32 4 }, { i8*, i32 }* %{{.*}}
MS ABI: Include the vfptr offset in memptrs to virtual methods Virtual methods expect 'this' to point to the vfptr containing the virtual method, and this extends to virtual member pointer thunks. The relevant vfptr is always at offset zero on entry to the thunk, and no this adjustment is needed. Previously we would not include the vfptr adjustment in the member pointer, and we'd look at the vfptr offset when loading from the vftable in the thunk. Fixes PR18917. llvm-svn: 201835
2014-02-21 10:27:32 +08:00
//
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define linkonce_odr x86_thiscallcc void @"??_9C@Test4@@$BA@AE"(%"struct.Test4::C"* %this, ...) {{.*}} comdat
MS ABI: Include the vfptr offset in memptrs to virtual methods Virtual methods expect 'this' to point to the vfptr containing the virtual method, and this extends to virtual member pointer thunks. The relevant vfptr is always at offset zero on entry to the thunk, and no this adjustment is needed. Previously we would not include the vfptr adjustment in the member pointer, and we'd look at the vfptr offset when loading from the vftable in the thunk. Fixes PR18917. llvm-svn: 201835
2014-02-21 10:27:32 +08:00
// CHECK-NOT: getelementptr
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK: load void (%"struct.Test4::C"*, ...)**, void (%"struct.Test4::C"*, ...)*** %{{.*}}
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK: getelementptr inbounds void (%"struct.Test4::C"*, ...)*, void (%"struct.Test4::C"*, ...)** %{{.*}}, i64 0
MS ABI: Include the vfptr offset in memptrs to virtual methods Virtual methods expect 'this' to point to the vfptr containing the virtual method, and this extends to virtual member pointer thunks. The relevant vfptr is always at offset zero on entry to the thunk, and no this adjustment is needed. Previously we would not include the vfptr adjustment in the member pointer, and we'd look at the vfptr offset when loading from the vftable in the thunk. Fixes PR18917. llvm-svn: 201835
2014-02-21 10:27:32 +08:00
// CHECK-NOT: getelementptr
[opaque pointer types] Explicit non-pointer type for call expressions (migration for recent LLVM change to textual IR for calls) llvm-svn: 235147
2015-04-17 07:25:00 +08:00
// CHECK: musttail call x86_thiscallcc void (%"struct.Test4::C"*, ...) %
MS ABI: Include the vfptr offset in memptrs to virtual methods Virtual methods expect 'this' to point to the vfptr containing the virtual method, and this extends to virtual member pointer thunks. The relevant vfptr is always at offset zero on entry to the thunk, and no this adjustment is needed. Previously we would not include the vfptr adjustment in the member pointer, and we'd look at the vfptr offset when loading from the vftable in the thunk. Fixes PR18917. llvm-svn: 201835
2014-02-21 10:27:32 +08:00
}
s/pr2007/20007/ in a test llvm-svn: 210834
2014-06-13 05:46:38 +08:00
namespace pr20007 {
MS ABI: Fix forming pointers to members of a base class Previously we would calculate the inheritance model of a class when requiring a pointer to member type of that class to be complete. The inheritance model is used to figure out how many fields are used by the member pointer. However, once we require a pointer to member of a derived class type to be complete, we can form pointers to members of bases without calculating the inheritance model for those bases. This was causing crashes on this simple test case: struct A { void f(); void f(int); }; struct B : public A {}; void g() { void (B::*a)() = &B::f; } Now we calculate the inheritance models of all base classes when completing a member pointer type. Fixes PR2007. llvm-svn: 210813
2014-06-13 03:49:17 +08:00
struct A {
void f();
void f(int);
};
struct B : public A {};
void test() { void (B::*a)() = &B::f; }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define dso_local void @"?test@pr20007@@YAXXZ"
// CHECK: store i8* bitcast (void (%"struct.pr20007::A"*)* @"?f@A@pr20007@@QAEXXZ" to i8*)
MS ABI: Fix forming pointers to members of a base class Previously we would calculate the inheritance model of a class when requiring a pointer to member type of that class to be complete. The inheritance model is used to figure out how many fields are used by the member pointer. However, once we require a pointer to member of a derived class type to be complete, we can form pointers to members of bases without calculating the inheritance model for those bases. This was causing crashes on this simple test case: struct A { void f(); void f(int); }; struct B : public A {}; void g() { void (B::*a)() = &B::f; } Now we calculate the inheritance models of all base classes when completing a member pointer type. Fixes PR2007. llvm-svn: 210813
2014-06-13 03:49:17 +08:00
}
s/pr2007/20007/ in a test llvm-svn: 210834
2014-06-13 05:46:38 +08:00
namespace pr20007_kw {
MS ABI: Fix forming pointers to members of a base class Previously we would calculate the inheritance model of a class when requiring a pointer to member type of that class to be complete. The inheritance model is used to figure out how many fields are used by the member pointer. However, once we require a pointer to member of a derived class type to be complete, we can form pointers to members of bases without calculating the inheritance model for those bases. This was causing crashes on this simple test case: struct A { void f(); void f(int); }; struct B : public A {}; void g() { void (B::*a)() = &B::f; } Now we calculate the inheritance models of all base classes when completing a member pointer type. Fixes PR2007. llvm-svn: 210813
2014-06-13 03:49:17 +08:00
struct A {
void f();
void f(int);
};
struct __single_inheritance B;
struct B : public A {};
void test() { void (B::*a)() = &B::f; }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define dso_local void @"?test@pr20007_kw@@YAXXZ"
// CHECK: store i8* bitcast (void (%"struct.pr20007_kw::A"*)* @"?f@A@pr20007_kw@@QAEXXZ" to i8*)
MS ABI: Fix forming pointers to members of a base class Previously we would calculate the inheritance model of a class when requiring a pointer to member type of that class to be complete. The inheritance model is used to figure out how many fields are used by the member pointer. However, once we require a pointer to member of a derived class type to be complete, we can form pointers to members of bases without calculating the inheritance model for those bases. This was causing crashes on this simple test case: struct A { void f(); void f(int); }; struct B : public A {}; void g() { void (B::*a)() = &B::f; } Now we calculate the inheritance models of all base classes when completing a member pointer type. Fixes PR2007. llvm-svn: 210813
2014-06-13 03:49:17 +08:00
}
MS ABI: Don't force bases to have an inheritance model Previously, assigning an inheritance model to a derived class would trigger further assiginments to the various bases of the class. This was done to fix a bug where we couldn't handle an implicit base-to-derived conversion for pointers-to-members when the conversion was ambiguous at an earlier point. However, this is not how the MS scheme works. Instead, assign inheritance models to *just* the class which owns to declaration we ended up referencing. N.B. This result is surprising in many ways. It means that it is possible for a base to have a "larger" inheritance model than it's derived classes. It also means that bases in the conversion path do not get assigned a model. struct A { void f(); void f(int); }; struct B : A {}; struct C : B {}; void f() { void (C::*x)() = &A::f; } We can only begin to assign an inheritance model *after* we've seen the address-of but *before* we've done the implicit conversion the more derived pointer-to-member type. After that point, both 'A' and 'C' will have an inheritance model but 'B' will not. Surprising, right? llvm-svn: 215174
2014-08-08 08:10:39 +08:00
namespace pr20007_pragma {
struct A {
void f();
void f(int);
};
struct B : public A {};
void test() { (void)(void (B::*)()) &B::f; }
#pragma pointers_to_members(full_generality, virtual_inheritance)
static_assert(sizeof(int B::*) == 4, "");
static_assert(sizeof(int A::*) == 4, "");
#pragma pointers_to_members(best_case)
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define dso_local void @"?test@pr20007_pragma@@YAXXZ"
MS ABI: Don't force bases to have an inheritance model Previously, assigning an inheritance model to a derived class would trigger further assiginments to the various bases of the class. This was done to fix a bug where we couldn't handle an implicit base-to-derived conversion for pointers-to-members when the conversion was ambiguous at an earlier point. However, this is not how the MS scheme works. Instead, assign inheritance models to *just* the class which owns to declaration we ended up referencing. N.B. This result is surprising in many ways. It means that it is possible for a base to have a "larger" inheritance model than it's derived classes. It also means that bases in the conversion path do not get assigned a model. struct A { void f(); void f(int); }; struct B : A {}; struct C : B {}; void f() { void (C::*x)() = &A::f; } We can only begin to assign an inheritance model *after* we've seen the address-of but *before* we've done the implicit conversion the more derived pointer-to-member type. After that point, both 'A' and 'C' will have an inheritance model but 'B' will not. Surprising, right? llvm-svn: 215174
2014-08-08 08:10:39 +08:00
}
namespace pr20007_pragma2 {
struct A {
};
struct B : public A {
void f();
};
void test() { (void)&B::f; }
#pragma pointers_to_members(full_generality, virtual_inheritance)
static_assert(sizeof(int B::*) == 4, "");
static_assert(sizeof(int A::*) == 12, "");
#pragma pointers_to_members(best_case)
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define dso_local void @"?test@pr20007_pragma2@@YAXXZ"
MS ABI: Don't force bases to have an inheritance model Previously, assigning an inheritance model to a derived class would trigger further assiginments to the various bases of the class. This was done to fix a bug where we couldn't handle an implicit base-to-derived conversion for pointers-to-members when the conversion was ambiguous at an earlier point. However, this is not how the MS scheme works. Instead, assign inheritance models to *just* the class which owns to declaration we ended up referencing. N.B. This result is surprising in many ways. It means that it is possible for a base to have a "larger" inheritance model than it's derived classes. It also means that bases in the conversion path do not get assigned a model. struct A { void f(); void f(int); }; struct B : A {}; struct C : B {}; void f() { void (C::*x)() = &A::f; } We can only begin to assign an inheritance model *after* we've seen the address-of but *before* we've done the implicit conversion the more derived pointer-to-member type. After that point, both 'A' and 'C' will have an inheritance model but 'B' will not. Surprising, right? llvm-svn: 215174
2014-08-08 08:10:39 +08:00
}
[MS ABI] Lock-in the derived memptr rep. for base-to-derived conversions We would get this right in the case where an explicit cast was formed but not when we were performing an implicit conversion. This fixes PR23828. llvm-svn: 239625
2015-06-13 01:55:44 +08:00
namespace pr23823 {
struct Base { void Method(); };
struct Child : Base {};
void use(void (Child::*const &)());
void f() { use(&Child::Method); }
#pragma pointers_to_members(full_generality, virtual_inheritance)
static_assert(sizeof(int Base::*) == 4, "");
static_assert(sizeof(int Child::*) == 4, "");
#pragma pointers_to_members(best_case)
}
MS ABI: Fix inheritance model calculation in CRTP CRTP-like patterns involve a class which inherits from another class using itself as a template parameter. However, the base class itself may try to create a pointer-to-member which involves the derived class. This is problematic because we may not have finished parsing the most derived classes' base specifiers yet. It turns out that MSVC simply uses the unspecified inheritance model instead of doing anything fancy. This fixes PR19987. llvm-svn: 210886
2014-06-13 14:43:46 +08:00
namespace pr19987 {
template <typename T>
struct S {
int T::*x;
};
struct U : S<U> {};
static_assert(sizeof(S<U>::x) == 12, "");
}
Sema: Do not assert when dereferencing member pointer using virtual inheritance with an incomplete class type The MS ABI requires that we determine the vbptr offset if have a virtual inheritance model. Instead, raise an error pointing to the diagnostic when this happens. This fixes PR18583. Differential Revision: http://llvm-reviews.chandlerc.com/D2842 llvm-svn: 201824
2014-02-21 07:22:07 +08:00
#else
struct __virtual_inheritance A;
#ifdef MEMFUN
int foo(A *a, int (A::*mp)()) {
return (a->*mp)(); // expected-error{{requires a complete class type}}
}
#else
int foo(A *a, int A::*mp) {
return a->*mp; // expected-error{{requires a complete class type}}
}
#endif
#endif
[MS ABI] Rework member pointer conversion Member pointers in the MS ABI are made complicated due to the following: - Virtual methods in the most derived class (MDC) might live in a vftable in a virtual base. - There are four different representations of member pointer: single inheritance, multiple inheritance, virtual inheritance and the "most general" representation. - Bases might have a *more* general representation than classes which derived from them, a most surprising result. We believed that we could treat all member pointers as-if they were a degenerate case of the multiple inheritance model. This fell apart once we realized that implementing standard member pointers using this ABI requires referencing members with a non-zero vbindex. On a bright note, all but the virtual inheritance model operate rather similarly. The virtual inheritance member pointer representation awkwardly requires a virtual base adjustment in order to refer to entities in the MDC. However, the first virtual base might be quite far from the start of the virtual base. This means that we must add a negative non-virtual displacement. However, things get even more complicated. The most general representation interprets vbindex zero differently from the virtual inheritance model: it doesn't reference the vbtable at all. It turns out that this complexity can increase for quite some time: consider a derived to base conversion from the most general model to the multiple inheritance model... To manage this complexity we introduce a concept of "normalized" member pointer which allows us to treat all three models as the most general model. Then we try to figure out how to map this generalized member pointer onto the destination member pointer model. I've done my best to furnish the code with comments explaining why each adjustment is performed. This fixes PR23878. llvm-svn: 240384
2015-06-23 15:31:11 +08:00
namespace pr23878 {
struct A { virtual void g(); };
struct B { virtual void f(); };
struct C : virtual B { void f(); };
struct D : A, C {};
typedef void (D::*DMemPtrTy)();
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define dso_local void @"?get_memptr@pr23878@@YAP8D@1@AEXXZXZ"
// CHECK: @"??_9C@pr23878@@$BA@AE" to i8*), i32 0, i32 4
[MS ABI] Rework member pointer conversion Member pointers in the MS ABI are made complicated due to the following: - Virtual methods in the most derived class (MDC) might live in a vftable in a virtual base. - There are four different representations of member pointer: single inheritance, multiple inheritance, virtual inheritance and the "most general" representation. - Bases might have a *more* general representation than classes which derived from them, a most surprising result. We believed that we could treat all member pointers as-if they were a degenerate case of the multiple inheritance model. This fell apart once we realized that implementing standard member pointers using this ABI requires referencing members with a non-zero vbindex. On a bright note, all but the virtual inheritance model operate rather similarly. The virtual inheritance member pointer representation awkwardly requires a virtual base adjustment in order to refer to entities in the MDC. However, the first virtual base might be quite far from the start of the virtual base. This means that we must add a negative non-virtual displacement. However, things get even more complicated. The most general representation interprets vbindex zero differently from the virtual inheritance model: it doesn't reference the vbtable at all. It turns out that this complexity can increase for quite some time: consider a derived to base conversion from the most general model to the multiple inheritance model... To manage this complexity we introduce a concept of "normalized" member pointer which allows us to treat all three models as the most general model. Then we try to figure out how to map this generalized member pointer onto the destination member pointer model. I've done my best to furnish the code with comments explaining why each adjustment is performed. This fixes PR23878. llvm-svn: 240384
2015-06-23 15:31:11 +08:00
DMemPtrTy get_memptr() { return &D::f; }
}
[MSVC] Crash fix: assigning of overloaded member function pointer caused assertion Original class was not marked with inheritance attribute and it causes a crash on codegen. Differential Revision: http://reviews.llvm.org/D11828 llvm-svn: 244428
2015-08-10 12:07:49 +08:00
class C {};
typedef void (C::*f)();
class CA : public C {
public:
void OnHelp(void);
int OnHelp(int);
};
// CHECK-LABEL: foo_fun
void foo_fun() {
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK: store i8* bitcast (void (%class.CA*)* @"?OnHelp@CA@@QAEXXZ" to i8*), i8**
[MSVC] Crash fix: assigning of overloaded member function pointer caused assertion Original class was not marked with inheritance attribute and it causes a crash on codegen. Differential Revision: http://reviews.llvm.org/D11828 llvm-svn: 244428
2015-08-10 12:07:49 +08:00
f func = (f)&CA::OnHelp;
}
[MS ABI] Don't crash on references to pointers to members in args We know that a reference can always be dereferenced. However, we don't always know the number of bytes if the reference's pointee type is incomplete. This case was correctly handled but we didn't consider the case where the type is complete but we cannot calculate its size for ABI specific reasons. In this specific case, a member pointer's size is available only under certain conditions. This fixes PR24703. llvm-svn: 247188
2015-09-10 04:57:59 +08:00
namespace PR24703 {
struct S;
void f(int S::*&p) {}
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: define dso_local void @"?f@PR24703@@YAXAAPQS@1@H@Z"(
[MS ABI] Don't crash on references to pointers to members in args We know that a reference can always be dereferenced. However, we don't always know the number of bytes if the reference's pointee type is incomplete. This case was correctly handled but we didn't consider the case where the type is complete but we cannot calculate its size for ABI specific reasons. In this specific case, a member pointer's size is available only under certain conditions. This fixes PR24703. llvm-svn: 247188
2015-09-10 04:57:59 +08:00
}
[MSVC] Crash fix: assigning of overloaded member function pointer caused assertion Original class was not marked with inheritance attribute and it causes a crash on codegen. Differential Revision: http://reviews.llvm.org/D11828 llvm-svn: 244428
2015-08-10 12:07:49 +08:00
[MS ABI] Select a pointer to member representation more often Given a reference to a pointer to member whose class's inheritance model is unspecified, make sure we come up with an inheritance model in plausible places. One place we were missing involved LValue to RValue conversion, another involved unary type traits. llvm-svn: 247248
2015-09-10 15:20:05 +08:00
namespace ReferenceToMPTWithIncompleteClass {
struct S;
struct J;
struct K;
extern K *k;
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: @"?f@ReferenceToMPTWithIncompleteClass@@YAIAAPQS@1@H@Z"(
[MS ABI] Select a pointer to member representation more often Given a reference to a pointer to member whose class's inheritance model is unspecified, make sure we come up with an inheritance model in plausible places. One place we were missing involved LValue to RValue conversion, another involved unary type traits. llvm-svn: 247248
2015-09-10 15:20:05 +08:00
// CHECK: ret i32 12
unsigned f(int S::*&p) { return sizeof p; }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: @"?g@ReferenceToMPTWithIncompleteClass@@YA_NAAPQJ@1@H0@Z"(
[MS ABI] Select a pointer to member representation more often Given a reference to a pointer to member whose class's inheritance model is unspecified, make sure we come up with an inheritance model in plausible places. One place we were missing involved LValue to RValue conversion, another involved unary type traits. llvm-svn: 247248
2015-09-10 15:20:05 +08:00
bool g(int J::*&p, int J::*&q) { return p == q; }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: @"?h@ReferenceToMPTWithIncompleteClass@@YAHAAPQK@1@H@Z"(
[MS ABI] Select a pointer to member representation more often Given a reference to a pointer to member whose class's inheritance model is unspecified, make sure we come up with an inheritance model in plausible places. One place we were missing involved LValue to RValue conversion, another involved unary type traits. llvm-svn: 247248
2015-09-10 15:20:05 +08:00
int h(int K::*&p) { return k->*p; }
}
[MS ABI] Select an inheritance model in template arguments We used to only select an inheritance model if the pointer to member was nullptr. Instead, select a model regardless of the member pointer's value. N.B. This bug was exposed by making member pointers report true for isIncompleteType but has been latent since the member pointer scheme's inception. llvm-svn: 247464
2015-09-12 04:18:09 +08:00
namespace PMFInTemplateArgument {
template <class C, int (C::*M)(int)>
void JSMethod();
class A {
int printd(int);
void printd();
};
void A::printd() { JSMethod<A, &A::printd>(); }
[MS] Don't escape MS C++ names with \01 It is not needed after LLVM r327734. Now it will be easier to copy-paste IR symbol names from Clang. llvm-svn: 327738
2018-03-17 04:36:49 +08:00
// CHECK-LABEL: @"??$JSMethod@VA@PMFInTemplateArgument@@$1?printd@12@AAEHH@Z@PMFInTemplateArgument@@YAXXZ"(
[MS ABI] Select an inheritance model in template arguments We used to only select an inheritance model if the pointer to member was nullptr. Instead, select a model regardless of the member pointer's value. N.B. This bug was exposed by making member pointers report true for isIncompleteType but has been latent since the member pointer scheme's inception. llvm-svn: 247464
2015-09-12 04:18:09 +08:00
}