Without this patch, clang will not wrap in an ElaboratedType node types written
without a keyword and nested name qualifier, which goes against the intent that
we should produce an AST which retains enough details to recover how things are
written.
The lack of this sugar is incompatible with the intent of the type printer
default policy, which is to print types as written, but to fall back and print
them fully qualified when they are desugared.
An ElaboratedTypeLoc without keyword / NNS uses no storage by itself, but still
requires pointer alignment due to pre-existing bug in the TypeLoc buffer
handling.
---
Troubleshooting list to deal with any breakage seen with this patch:
1) The most likely effect one would see by this patch is a change in how
a type is printed. The type printer will, by design and default,
print types as written. There are customization options there, but
not that many, and they mainly apply to how to print a type that we
somehow failed to track how it was written. This patch fixes a
problem where we failed to distinguish between a type
that was written without any elaborated-type qualifiers,
such as a 'struct'/'class' tags and name spacifiers such as 'std::',
and one that has been stripped of any 'metadata' that identifies such,
the so called canonical types.
Example:
```
namespace foo {
struct A {};
A a;
};
```
If one were to print the type of `foo::a`, prior to this patch, this
would result in `foo::A`. This is how the type printer would have,
by default, printed the canonical type of A as well.
As soon as you add any name qualifiers to A, the type printer would
suddenly start accurately printing the type as written. This patch
will make it print it accurately even when written without
qualifiers, so we will just print `A` for the initial example, as
the user did not really write that `foo::` namespace qualifier.
2) This patch could expose a bug in some AST matcher. Matching types
is harder to get right when there is sugar involved. For example,
if you want to match a type against being a pointer to some type A,
then you have to account for getting a type that is sugar for a
pointer to A, or being a pointer to sugar to A, or both! Usually
you would get the second part wrong, and this would work for a
very simple test where you don't use any name qualifiers, but
you would discover is broken when you do. The usual fix is to
either use the matcher which strips sugar, which is annoying
to use as for example if you match an N level pointer, you have
to put N+1 such matchers in there, beginning to end and between
all those levels. But in a lot of cases, if the property you want
to match is present in the canonical type, it's easier and faster
to just match on that... This goes with what is said in 1), if
you want to match against the name of a type, and you want
the name string to be something stable, perhaps matching on
the name of the canonical type is the better choice.
3) This patch could expose a bug in how you get the source range of some
TypeLoc. For some reason, a lot of code is using getLocalSourceRange(),
which only looks at the given TypeLoc node. This patch introduces a new,
and more common TypeLoc node which contains no source locations on itself.
This is not an inovation here, and some other, more rare TypeLoc nodes could
also have this property, but if you use getLocalSourceRange on them, it's not
going to return any valid locations, because it doesn't have any. The right fix
here is to always use getSourceRange() or getBeginLoc/getEndLoc which will dive
into the inner TypeLoc to get the source range if it doesn't find it on the
top level one. You can use getLocalSourceRange if you are really into
micro-optimizations and you have some outside knowledge that the TypeLocs you are
dealing with will always include some source location.
4) Exposed a bug somewhere in the use of the normal clang type class API, where you
have some type, you want to see if that type is some particular kind, you try a
`dyn_cast` such as `dyn_cast<TypedefType>` and that fails because now you have an
ElaboratedType which has a TypeDefType inside of it, which is what you wanted to match.
Again, like 2), this would usually have been tested poorly with some simple tests with
no qualifications, and would have been broken had there been any other kind of type sugar,
be it an ElaboratedType or a TemplateSpecializationType or a SubstTemplateParmType.
The usual fix here is to use `getAs` instead of `dyn_cast`, which will look deeper
into the type. Or use `getAsAdjusted` when dealing with TypeLocs.
For some reason the API is inconsistent there and on TypeLocs getAs behaves like a dyn_cast.
5) It could be a bug in this patch perhaps.
Let me know if you need any help!
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D112374
This reverts commit 7c51f02eff because it
stills breaks the LLDB tests. This was re-landed without addressing the
issue or even agreement on how to address the issue. More details and
discussion in https://reviews.llvm.org/D112374.
Without this patch, clang will not wrap in an ElaboratedType node types written
without a keyword and nested name qualifier, which goes against the intent that
we should produce an AST which retains enough details to recover how things are
written.
The lack of this sugar is incompatible with the intent of the type printer
default policy, which is to print types as written, but to fall back and print
them fully qualified when they are desugared.
An ElaboratedTypeLoc without keyword / NNS uses no storage by itself, but still
requires pointer alignment due to pre-existing bug in the TypeLoc buffer
handling.
---
Troubleshooting list to deal with any breakage seen with this patch:
1) The most likely effect one would see by this patch is a change in how
a type is printed. The type printer will, by design and default,
print types as written. There are customization options there, but
not that many, and they mainly apply to how to print a type that we
somehow failed to track how it was written. This patch fixes a
problem where we failed to distinguish between a type
that was written without any elaborated-type qualifiers,
such as a 'struct'/'class' tags and name spacifiers such as 'std::',
and one that has been stripped of any 'metadata' that identifies such,
the so called canonical types.
Example:
```
namespace foo {
struct A {};
A a;
};
```
If one were to print the type of `foo::a`, prior to this patch, this
would result in `foo::A`. This is how the type printer would have,
by default, printed the canonical type of A as well.
As soon as you add any name qualifiers to A, the type printer would
suddenly start accurately printing the type as written. This patch
will make it print it accurately even when written without
qualifiers, so we will just print `A` for the initial example, as
the user did not really write that `foo::` namespace qualifier.
2) This patch could expose a bug in some AST matcher. Matching types
is harder to get right when there is sugar involved. For example,
if you want to match a type against being a pointer to some type A,
then you have to account for getting a type that is sugar for a
pointer to A, or being a pointer to sugar to A, or both! Usually
you would get the second part wrong, and this would work for a
very simple test where you don't use any name qualifiers, but
you would discover is broken when you do. The usual fix is to
either use the matcher which strips sugar, which is annoying
to use as for example if you match an N level pointer, you have
to put N+1 such matchers in there, beginning to end and between
all those levels. But in a lot of cases, if the property you want
to match is present in the canonical type, it's easier and faster
to just match on that... This goes with what is said in 1), if
you want to match against the name of a type, and you want
the name string to be something stable, perhaps matching on
the name of the canonical type is the better choice.
3) This patch could exposed a bug in how you get the source range of some
TypeLoc. For some reason, a lot of code is using getLocalSourceRange(),
which only looks at the given TypeLoc node. This patch introduces a new,
and more common TypeLoc node which contains no source locations on itself.
This is not an inovation here, and some other, more rare TypeLoc nodes could
also have this property, but if you use getLocalSourceRange on them, it's not
going to return any valid locations, because it doesn't have any. The right fix
here is to always use getSourceRange() or getBeginLoc/getEndLoc which will dive
into the inner TypeLoc to get the source range if it doesn't find it on the
top level one. You can use getLocalSourceRange if you are really into
micro-optimizations and you have some outside knowledge that the TypeLocs you are
dealing with will always include some source location.
4) Exposed a bug somewhere in the use of the normal clang type class API, where you
have some type, you want to see if that type is some particular kind, you try a
`dyn_cast` such as `dyn_cast<TypedefType>` and that fails because now you have an
ElaboratedType which has a TypeDefType inside of it, which is what you wanted to match.
Again, like 2), this would usually have been tested poorly with some simple tests with
no qualifications, and would have been broken had there been any other kind of type sugar,
be it an ElaboratedType or a TemplateSpecializationType or a SubstTemplateParmType.
The usual fix here is to use `getAs` instead of `dyn_cast`, which will look deeper
into the type. Or use `getAsAdjusted` when dealing with TypeLocs.
For some reason the API is inconsistent there and on TypeLocs getAs behaves like a dyn_cast.
5) It could be a bug in this patch perhaps.
Let me know if you need any help!
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D112374
This reverts commit bdc6974f92 because it
breaks all the LLDB tests that import the std module.
import-std-module/array.TestArrayFromStdModule.py
import-std-module/deque-basic.TestDequeFromStdModule.py
import-std-module/deque-dbg-info-content.TestDbgInfoContentDequeFromStdModule.py
import-std-module/forward_list.TestForwardListFromStdModule.py
import-std-module/forward_list-dbg-info-content.TestDbgInfoContentForwardListFromStdModule.py
import-std-module/list.TestListFromStdModule.py
import-std-module/list-dbg-info-content.TestDbgInfoContentListFromStdModule.py
import-std-module/queue.TestQueueFromStdModule.py
import-std-module/stack.TestStackFromStdModule.py
import-std-module/vector.TestVectorFromStdModule.py
import-std-module/vector-bool.TestVectorBoolFromStdModule.py
import-std-module/vector-dbg-info-content.TestDbgInfoContentVectorFromStdModule.py
import-std-module/vector-of-vectors.TestVectorOfVectorsFromStdModule.py
https://green.lab.llvm.org/green/view/LLDB/job/lldb-cmake/45301/
Without this patch, clang will not wrap in an ElaboratedType node types written
without a keyword and nested name qualifier, which goes against the intent that
we should produce an AST which retains enough details to recover how things are
written.
The lack of this sugar is incompatible with the intent of the type printer
default policy, which is to print types as written, but to fall back and print
them fully qualified when they are desugared.
An ElaboratedTypeLoc without keyword / NNS uses no storage by itself, but still
requires pointer alignment due to pre-existing bug in the TypeLoc buffer
handling.
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D112374
clang currently calls MarkVTableUsed() for classes that get their virtual
methods called or that participate in a dynamic_cast. This is unnecessary,
since CodeGen only emits vtables when it generates constructor, destructor, and
vtt code. (*)
Note that Sema::MarkVTableUsed() doesn't cause the emission of a vtable.
Its main user-visible effect is that it instantiates virtual member functions
of template classes, to make sure that if codegen decides to write a vtable
all the entries in the vtable are defined.
While this shouldn't change the behavior of codegen (other than being faster),
it does make clang more permissive: virtual methods of templates (in particular
destructors) end up being instantiated less often. In particular, classes that
have members that are smart pointers to incomplete types will now get their
implicit virtual destructor instantiated less frequently. For example, this
used to not compile but does now compile:
template <typename T> struct OwnPtr {
~OwnPtr() { static_assert((sizeof(T) > 0), "TypeMustBeComplete"); }
};
class ScriptLoader;
struct Base { virtual ~Base(); };
struct Sub : public Base {
virtual void someFun() const {}
OwnPtr<ScriptLoader> m_loader;
};
void f(Sub *s) { s->someFun(); }
The more permissive behavior matches both gcc (where this is not often
observable, since in practice most things with virtual methods have a key
function, and Sema::DefineUsedVTables() skips vtables for classes with key
functions) and cl (which is my motivation for this change) – this fixes
PR20337. See this issue and the review thread for some discussions about
optimizations.
This is similar to r213109 in spirit. r225761 was a prerequisite for this
change.
Various tests relied on "a->f()" marking a's vtable as used (in the sema
sense), switch these to just construct a on the stack. This forces
instantiation of the implicit constructor, which will mark the vtable as used.
(*) The exception is -fapple-kext mode: In this mode, qualified calls to
virtual functions (`a->Base::f()`) still go through the vtable, and since the
vtable pointer off this doesn't point to Base's vtable, this needs to reference
Base's vtable directly. To keep this working, keep referencing the vtable for
virtual calls in apple kext mode.
llvm-svn: 227073
There are two things in a C++ program that need to read the vtable pointer:
Constructors and destructors. (A few other operations -- virtual calls,
dynamic cast, rtti -- read the vtable pointer off a this pointer, but for
this they don't need the vtable symbol.) Implicit constructors and destructors
and explicit constructors already marked the vtable as used, but explicit
destructors didn't.
Note that the only thing sema's "mark a class's vtable used" does is to mark all
final overriders of the class as referenced, it does _not_ cause emission of
the vtable itself. This is done on demand by codegen, independent of sema,
since sema might emit functions that are not referenced. (The exception are
vtables that are forced via key functions -- these are forced onto codegen
by sema.)
This bug went unnoticed for years because it doesn't have observable effects
(yet -- I want to change this in PR20337, which is why I noticed this).
r213109 made it so that _calls_ to constructors don't mark the vtable used.
Currently, _calls_ to destructors still mark the vtable used. If that
wasn't the case, this program would tickle the problem:
test.h:
template <typename T>
struct B {
int* p;
virtual ~B() { delete p; }
virtual void f() {}
};
struct __attribute__((visibility("default"))) C {
C();
B<int> m;
};
test2.cc:
#include "test.h"
int main() {
C* c = new C;
delete c;
}
test3.cc:
#include "test.h"
C::C() {}
# This bin/clang++ binary doesn't MarkVTableUsed() for virtual dtor calls:
$ bin/clang++ -shared test3.cc -std=c++11 -O2 -fvisibility=hidden \
-fvisibility-inlines-hidden -o libtest3.dylib
$ bin/clang++ test2.cc -std=c++11 -O2 -fvisibility=hidden \
-fvisibility-inlines-hidden libtest3.dylib
Undefined symbols for architecture x86_64:
"B<int>::f()", referenced from:
vtable for B<int> in test2-af8f4f.o
ld: symbol(s) not found for architecture x86_64
What's happening here is that there's a copy of B's vtable hidden in
libtest3.dylib, because C's constructor caused an implicit instantiation of that
(and implicit constructors generate vtables).
test2.cc calls C's destructDr, which destroys the B<int> member,
which wants to overwrite the vtable back to B (think of B as the base of a class
hierarchy, and of hierarchical destruction -- maybe we shouldn't do the vtable
writing in destructors of final classes), but there's nothing in test2.cc that
marks B's vtable used. So codegen writes out the vtable, but since it wasn't
marked used, sema didn't mark all the virtual functions (in particular f())
as used.
Note that this change makes us reject programs we didn't reject before (see
the included Sema test case), but both gcc and cl also reject this code, and
clang used to reject it before r213109.
llvm-svn: 225761
Templates don't have key functions (cf computeKeyFunction() in
RecordLayoutBuilder.cpp), so don't have something that looks like one.
Also, instead of a vcall to force generation of the vtable, just construct
the object. This is how the repro on PR5557 (what the test is for) worked too.
llvm-svn: 225741
This fixes compilation errors about incomplete types used with WebKit's
RefPtr template. Simply calling an out of line constructor should not
instantiate all inline and defaulted virtual methods.
Tested by building and testing several big piles of code on Linux.
Reviewers: rsmith
Differential Revision: http://reviews.llvm.org/D4429
llvm-svn: 213109
This makes the C++ ABI depend entirely on the target: MS ABI for -win32 triples,
Itanium otherwise. It's no longer possible to do weird combinations.
To be able to run a test with a specific ABI without constraining it to a
specific triple, new substitutions are added to lit: %itanium_abi_triple and
%ms_abi_triple can be used to get the current target triple adjusted to the
desired ABI. For example, if the test suite is running with the i686-pc-win32
target, %itanium_abi_triple will expand to i686-pc-mingw32.
Differential Revision: http://llvm-reviews.chandlerc.com/D2545
llvm-svn: 199250
In preparation for making the Win32 triple imply MS ABI mode,
make all tests pass in this mode, or make them use the Itanium
mode explicitly.
Differential Revision: http://llvm-reviews.chandlerc.com/D2401
llvm-svn: 199130
"used" (e.g., we will refer to the vtable in the generated code) and
when they are defined (i.e., because we've seen the key function
definition). Previously, we were effectively tracking "potential
definitions" rather than uses, so we were a bit too eager about emitting
vtables for classes without key functions.
The new scheme:
- For every use of a vtable, Sema calls MarkVTableUsed() to indicate
the use. For example, this occurs when calling a virtual member
function of the class, defining a constructor of that class type,
dynamic_cast'ing from that type to a derived class, casting
to/through a virtual base class, etc.
- For every definition of a vtable, Sema calls MarkVTableUsed() to
indicate the definition. This happens at the end of the translation
unit for classes whose key function has been defined (so we can
delay computation of the key function; see PR6564), and will also
occur with explicit template instantiation definitions.
- For every vtable defined/used, we mark all of the virtual member
functions of that vtable as defined/used, unless we know that the key
function is in another translation unit. This instantiates virtual
member functions when needed.
- At the end of the translation unit, Sema tells CodeGen (via the
ASTConsumer) which vtables must be defined (CodeGen will define
them) and which may be used (for which CodeGen will define the
vtables lazily).
From a language perspective, both the old and the new schemes are
permissible: we're allowed to instantiate virtual member functions
whenever we want per the standard. However, all other C++ compilers
were more lazy than we were, and our eagerness was both a performance
issue (we instantiated too much) and a portability problem (we broke
Boost test cases, which now pass).
Notes:
(1) There's a ton of churn in the tests, because the order in which
vtables get emitted to IR has changed. I've tried to isolate some of
the larger tests from these issues.
(2) Some diagnostics related to
implicitly-instantiated/implicitly-defined virtual member functions
have moved to the point of first use/definition. It's better this
way.
(3) I could use a review of the places where we MarkVTableUsed, to
see if I missed any place where the language effectively requires a
vtable.
Fixes PR7114 and PR6564.
llvm-svn: 103718
referenced unless we see one of them defined (or the key function
defined, if it as one) or if we need the vtable for something. Fixes
PR7114.
llvm-svn: 103497
deterministic and work properly with templates. Once a class that
needs a vtable has been defined, we now do one if two things:
- If the class has no key function, we place the class on a list of
classes whose virtual functions will need to be "marked" at the
end of the translation unit. The delay until the end of the
translation unit is needed because we might see template
specializations of these virtual functions.
- If the class has a key function, we do nothing; when the key
function is defined, the class will be placed on the
aforementioned list.
At the end of the translation unit, we "mark" all of the virtual
functions of the classes on the list as used, possibly causing
template instantiation and other classes to be added to the
list. This gets LLVM's lib/Support/CommandLine.cpp compiling again.
llvm-svn: 92821
- This is designed to make it obvious that %clang_cc1 is a "test variable"
which is substituted. It is '%clang_cc1' instead of '%clang -cc1' because it
can be useful to redefine what gets run as 'clang -cc1' (for example, to set
a default target).
llvm-svn: 91446
Matheus Izvekov