David Majnemer [Thu, 25 Sep 2014 19:43:56 +0000 (19:43 +0000)]
MS ABI: Avoid hashing back reference keys in mangler
This patch replaces the back reference StringMap from the MS mangler
with a SmallVector of strings. My previous patches reduced the number of
hashes involved in back reference lookups, this one removes them
completely. The back reference map contains at most 10 entries, which
are likely to be of varying sizes and different initial subsequences,
and which can easily became huge (due to templates and namespaces).
The solution presented is the simplest possible one. Nevertheless, it's
enough to reduce compilation times for a particular test case from 11.1s
to 9s, versus 8.58s for the Itanium ABI. Possible further improvements
include using a sorted vector (carefully to not introduce an extra
comparison), storing the string contents in a common arena, and/or keep
the string storage in the context for reuse.
Richard Smith [Wed, 24 Sep 2014 23:55:00 +0000 (23:55 +0000)]
Fix handling of preincrement on bit-fields. This gives a bit-field in C++, but
we were failing to find that bit-field when performing integer promotions. This
brings us closer to following the standard, and closer to GCC.
In C, this change is technically a regression: we get bit-field promotions
completely wrong in C, promoting cases that are categorically not bit-field
designators. This change makes us do so slightly more consistently, though.
The attribute documentation now conforms to Aaron Ballman's renaming of the
thread safety attributes, as well as the new paper that is due to be published
in the conference on Source Code Analysis and Manipulation (SCAM 2014) later
this week. In addition, recent changes to the analysis, such as checking
of references and negative capabilities, are now documented.
Downgrade error about stdcall decls with no prototype to a warning
Fixes PR21027. The MIDL compiler produces code that does this.
If we wanted to improve the warning, I think we could do this:
void __stdcall f(); // Don't warn without -Wstrict-prototypes.
void g() {
f(); // Might warn, the user probably meant for f to take no args.
f(1, 2, 3); // Warn, we have no idea what args f takes.
f(1); // Error, this is insane, one of these calls is broken.
}
David Blaikie [Wed, 24 Sep 2014 17:01:27 +0000 (17:01 +0000)]
Include debug info for types referenced only via explicit cast expressions.
Most of the debug info emission is powered essentially from function
definitions - if we emit the definition of a function, we emit the types
of its parameters, the members of those types, and so on and so forth.
For types that aren't referenced even indirectly due to this - because
they only appear in temporary expressions, not in any named variable, we
need to explicitly emit/add them as is done here. This is not the only
case of such code, and we might want to consider handling "void
func(void*); ... func(new T());" (currently debug info for T is not
emitted) at some point, though GCC doesn't. There's a much broader
solution to these issues, but it's a lot of work for possibly marginal
gain (but might help us improve the default -fno-standalone-debug
behavior to be even more aggressive in some places). See the original
review thread for more details.
Patch by jyoti allur (jyoti.yalamanchili@gmail.com)!
David Majnemer [Wed, 24 Sep 2014 11:04:09 +0000 (11:04 +0000)]
Sema: Inherit the flexible array property from struct fields
A record which contains a flexible array member is itself a flexible
array member. A struct which contains such a record should also
consider itself to be a flexible array member.
Follow-up to r218292: Add more REVERTIBLE_TYPE_TRAITs.
r218292 reverted r197496 because it broke things. In addition to breaking
things, r197496 also made all traits starting with __is_ revertible.
Reinstantiate that part of r197496 because code out there (e.g. libc++) depends
on this behavior. Fixes PR21045.
David Majnemer [Tue, 23 Sep 2014 22:58:15 +0000 (22:58 +0000)]
MS ABI: Pure virtual functions don't contribute to vtordisps
Usually, overriding a virtual function defined in a virtual base
required emission of a vtordisp slot in the record. However no vtordisp
is needed if the overriding function is pure; it should be impossible to
observe the pure virtual method.
Richard Trieu [Tue, 23 Sep 2014 22:52:42 +0000 (22:52 +0000)]
Improve -Wuninitialized to take into account field ordering with initializer
lists. Since the fields are inititalized one at a time, using a field with
lower index to initialize a higher indexed field should not be warned on.
Richard Smith [Tue, 23 Sep 2014 21:05:52 +0000 (21:05 +0000)]
PR18793: If we try to EnterTokenStream when our current lexer is a caching
lexer, add the token buffer underneath the caching lexer where possible and
push the tokens directly into the caching lexer otherwise. We previously
put the lexer into a corrupted state where we could not guarantee to provide
the tokens in the right order and would sometimes assert.
Richard Smith [Tue, 23 Sep 2014 20:31:39 +0000 (20:31 +0000)]
Don't perform ADL when looking up operator=; there is no non-member form of
that function, and apart from being slow, this is unnecessary: ADL can trigger
instantiations that are not permitted here. The standard isn't *completely*
clear here, but this seems like the intent, and in any case this approach is
permitted by [temp.inst]p7.
Don't use comdats for initializers on platforms that don't support it
In particular, pre-.init_array ELF uses the .ctors section mechanism.
MinGW COFF also uses .ctors, now that I think about it. Therefore,
restrict this optimization to the two platforms that are currently known
to work: ELF with .init_array and COFF with .CRT$XCU.
We need to walk the class hierarchy twice: once in depth-first base
specifier order for mangling and again in depth-first layout order for
vftable layout.
Vftable layout seems to depend on the full path from the most derived
class to the base containing the vfptr.
Return true from UnifySection when emitting a diagnostic
Test Plan: I noticed this through code inspection. The callers use the return value to remove the SectionAttr if a diagnostic is emitted, but I don't think the failure to do so is observable right now.
Fix evatuated value of __builtin_object_size according to its
'type' argument when it cannot be determined which objects ptr
points to at compile time. rdar://18334276
Daniel Sanders [Mon, 22 Sep 2014 13:27:06 +0000 (13:27 +0000)]
[mips] Correct alignment of vectors passed in varargs for the O32 ABI.
Summary:
Vectors are normally 16-byte aligned, however the O32 ABI enforces a
maximum alignment of 8-bytes since the base of the stack is 8-byte aligned.
Previously, this was enforced on the caller side, but not on the callee side.
This fixes the output of OpenCL's printf when given vectors.
[OPENMP] Codegen for 'omp critical' directive.
This patch adds codegen for constructs:
#pragma omp critical [name]
<body>
It generates global variable ".gomp_critical_user_[name].var" of type int32[8]. Then it generates library call "kmpc_critical(loc, gtid, .gomp_critical_user_[name].var)", code for <body> statement and final call "kmpc_end_critical(loc, gtid, .gomp_critical_user_[name].var)".
Differential Revision: http://reviews.llvm.org/D5202
ms-inline-asm: Scope inline asm labels to functions
Summary:
This fixes PR20023. In order to implement this scoping rule, we piggy
back on the existing LabelDecl machinery, by creating LabelDecl's that
will carry the "internal" name of the inline assembly label, which we
will rewrite the asm label to.
- Remove the push and pop instructions that were saving and restoring %ebx
before and after cpuid in 32-bit pic mode. We were doing this to ensure we
don't lose the GOT address in pic register %ebx, but this isn't necessary
because the GOT address is kept in a virtual register.
- In 64-bit mode, preserve base register %rbx around cpuid.
Follow-up to r214408: Warn on other callee-cleanup functions without prototype too.
According to lore, we used to verifier-fail on:
void __thiscall f();
int main() { f(1); }
So that's fixed now. System headers use prototype-less __stdcall functions,
so make that a warning that's DefaultError -- then it fires on regular code
but is suppressed in system headers.
Since it's used in system headers, we have codegen tests for this; massage
them slightly so that they still compile.
This patch makes sure that the dllexport attribute is transferred to the alias when such alias is created. It only affects the Itanium ABI because for the MSVC ABI a workaround is in place to not generate aliases of dllexport ctors/dtors.
A new CodeGenModule function is provided, CodeGenModule::setAliasAttributes, to factor the code for transferring attributes to aliases.
Samuel Benzaquen [Fri, 19 Sep 2014 16:10:03 +0000 (16:10 +0000)]
Make DynTypedNode have the dynamic type of the object, instead of its static type.
Summary:
Make DynTypedNode have the dynamic type of the object, instead of its static type.
Some optimizations that are in the works require that the nodes have the right type.
Robert Khasanov [Fri, 19 Sep 2014 10:29:22 +0000 (10:29 +0000)]
[x86] Add _addcarry_u{32|64} and _subborrow_u{32|64}.
They are added to adxintrin.h but outside __ADX__ block.
These intrinics generates adc and sbb correspondingly that were available before ADX
Daniel Jasper [Fri, 19 Sep 2014 08:28:43 +0000 (08:28 +0000)]
clang-format: Prevent column layout if elements aren't uniform enough.
This patch only considers the difference between the length of the
shortest and longest element, but we might want to look at other
features (token count, etc.) in future.
Rafael Espindola [Fri, 19 Sep 2014 01:54:22 +0000 (01:54 +0000)]
Don't use the third field of llvm.global_ctors for MachO.
The field is defined as:
If the third field is present, non-null, and points to a global variable or function, the initializer function will only run if the associated data from the current module is not discarded.
Thread Safety Analysis: add new warning flag, -Wthread-safety-reference, which
warns when a guarded variable is passed by reference as a function argument.
This is released as a separate warning flag, because it could potentially
break existing code that uses thread safety analysis.
David Majnemer [Thu, 18 Sep 2014 22:05:54 +0000 (22:05 +0000)]
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.
Patch to check at compile time for overflow when
__builtin___memcpy_chk and similar builtins are
being used. Patch by Jacques Fortier (with added
clang tests). rdar://11076881
Hans Wennborg [Thu, 18 Sep 2014 16:01:32 +0000 (16:01 +0000)]
Revert r217995 and follow-ups:
r218053: Use exceptions() instead of getNumExceptions()/getExceptionType() to avoid
r218011: Work around MSVC parser bug by putting redundant braces around the body of
r217997: Skip parens when detecting whether we're instantiating a function declaration.
r217995: Instantiate exception specifications when instantiating function types (other
The Windows build was broken for 16 hours and no one had any good ideas of how to
fix it. Reverting for now to make the builders green. See the cfe-commits thread [1] for
more info.
This was the build error (from [2]):
C:\bb-win7\ninja-clang-i686-msc17-R\llvm-project\clang\lib\Sema\SemaTemplateInstantiate.cpp(1590) : error C2668: '`anonymous-namespace'::TemplateInstantiator::TransformFunctionProtoType' : ambiguous call to overloaded function
C:\bb-win7\ninja-clang-i686-msc17-R\llvm-project\clang\lib\Sema\SemaTemplateInstantiate.cpp(1313): could be 'clang::QualType `anonymous-namespace'::TemplateInstantiator::TransformFunctionProtoType<clang::Sema::SubstFunctionDeclType::<lambda_756edcbe7bd5c7584849a6e3a1491735>>(clang::TypeLocBuilder &,clang::FunctionProtoTypeLoc,clang::CXXRecordDecl *,unsigned int,Fn)'
with
[
Fn=clang::Sema::SubstFunctionDeclType::<lambda_756edcbe7bd5c7584849a6e3a1491735>
]
c:\bb-win7\ninja-clang-i686-msc17-r\llvm-project\clang\lib\sema\TreeTransform.h(4532): or 'clang::QualType clang::TreeTransform<Derived>::TransformFunctionProtoType<clang::Sema::SubstFunctionDeclType::<lambda_756edcbe7bd5c7584849a6e3a1491735>>(clang::TypeLocBuilder &,clang::FunctionProtoTypeLoc,clang::CXXRecordDecl *,unsigned int,Fn)'
with
[
Derived=`anonymous-namespace'::TemplateInstantiator,
Fn=clang::Sema::SubstFunctionDeclType::<lambda_756edcbe7bd5c7584849a6e3a1491735>
]
while trying to match the argument list '(clang::TypeLocBuilder, clang::FunctionProtoTypeLoc, clang::CXXRecordDecl *, unsigned int, clang::Sema::SubstFunctionDeclType::<lambda_756edcbe7bd5c7584849a6e3a1491735>)'
Use exceptions() instead of getNumExceptions()/getExceptionType() to avoid
accesses to incorrect exception types when getExceptionSpecType() != EST_Dynamic
This fixes a crash in test/CXX/except/except.spec/template.cpp that happens in
certain build configurations.
Change -Wbind-to-temporary-copy from an ExtWarn to an Extension.
The reasoning is that this construct is accepted by all compilers and valid in
C++11, so it doesn't seem like a useful warning to have enabled by default.
Building with -pedantic, -Wbind-to-temporary-copy, or -Wc++98-compat still
shows the warning.
The motivation is that I built re2, and this was the only warning that was
emitted during the build. Both changing re2 to fix the warning and detecting
clang and suppressing the warning in re2's build seem inferior than just giving
the compiler a good default for this warning.
Also move the cxx98compat version of this warning to CXX98CompatPedantic, and
update tests accordingly.
David Majnemer [Thu, 18 Sep 2014 00:42:05 +0000 (00:42 +0000)]
Sema: Diagnose undefined structs used as Microsoft anonymous structs
Previously, we would not mark structs containing anonymous structs as
invalid. Later, horrific things would occur when trying to determine
the size of the parent record.
Instead, require the struct to be a complete type when used as an
anonymous struct. Mark both the anonymous field for the struct and the
parent context as invalid (this is similar to what we do when a struct
contains a field with an incomplete type.)
Richard Smith [Wed, 17 Sep 2014 23:57:05 +0000 (23:57 +0000)]
Instantiate exception specifications when instantiating function types (other
than the type of a function declaration). We previously didn't instantiate
these at all! This also covers the pathological case where the only mention of
a parameter pack is within the exception specification; this gives us a second
way (other than alias templates) to reach the horrible state where a type
contains an unexpanded pack, but its canonical type does not.
Extend ARM ACLE support (Section 6.5.1) for AArch32. Define __ARM_FP if
hardware floating point support is available as per the value defined by the
ACLE.
We would end up marking the vtable of the derived class as used for no
reason. Because the call itself is qualified, it is never virtual, and
the vtable of the derived class isn't helpful. We would end up rejecting
code that MSVC accepts for no benefit.
Translate -lfoo to -lfoo.lib while making sure that -lfoo.lib stays as
-lfoo.lib. Also, these arguments were being passed twice: once
explicitly via AddAllArgs, and again implicitly as linker inputs. Now
they are passed once.
Rafael Espindola [Tue, 16 Sep 2014 15:18:21 +0000 (15:18 +0000)]
Add support for putting constructors and destructos in explicit comdats.
There are situations when clang knows that the C1 and C2 constructors
or the D1 and D2 destructors are identical. We already optimize some
of these cases, but cannot optimize it when the GlobalValue is
weak_odr.
The problem with weak_odr is that an old TU seeing the same code will
have a C1 and a C2 comdat with the corresponding symbols. We cannot
suddenly start putting the C2 symbol in the C1 comdat as we cannot
guarantee that the linker will not pick a .o with only C1 in it.
The solution implemented by GCC is to expand the ABI to have a comdat
whose name uses a C5/D5 suffix and always has both symbols. That is
what this patch implements.
Richard Smith [Tue, 16 Sep 2014 01:24:02 +0000 (01:24 +0000)]
Reject a slightly-sneaky way to perform a read of mutable state from within a
constexpr function. Part of this fix is a tentative fix for an as-yet-unfiled
core issue (we're missing a prohibition against reading mutable members from
unions via a trivial constructor/assignment, since that doesn't perform an
lvalue-to-rvalue conversion on the members).
Objective-C. Prevents a crash generating AST for a
a property assignment due to numerous user errors.
Cannot come up with a reasonable test case due to
array of user errors before the crash point.
rdar://17813651.
Major rewrite of linking strategy for sanitizer runtimes on Linux.
Change 1: we used to add static sanitizer runtimes at the
very beginning of the linker invocation, even before crtbegin.o, which
is gross and not correct in general. Fix this: now addSanitizerRuntimes()
adds all sanitizer-related link flags to the end of the linker invocation
being constructed. It means, that we should call this function in the
correct place, namely, before AddLinkerInputs() to make sure sanitizer
versions of library functions will be preferred.
Change 2: Put system libraries sanitizer libraries depend on at the
end of the linker invocation, where all the rest system libraries are
located. Respect --nodefaultlibs and --nostdlib flags. This is another way
to fix PR15823. Original fix landed in r215940 put "-lpthread" and friends
immediately after static ASan runtime, before the user linker inputs.
This caused significant slowdown in dynamic linker for large binaries
linked against thousands of shared objects. Instead, to mark system
libraries as DT_NEEDED we prepend them with "--no-as-needed" flag,
discarding the "-Wl,--as-needed" flag that could be provided by the user.
Otherwise, this change is a code cleanup. Instead of having a special method
for each sanitizer, we introduce a function collectSanitizerRuntimes() that
analyzes -fsanitize= flags and returns the set of static and shared
libraries that needs to be linked.
projects / clang / log