Easwaran Raman [Sat, 12 Dec 2015 00:31:02 +0000 (00:31 +0000)]
Attach maximum function count to Module when using PGO mode
This sets the maximum entry count among all functions in the program to the
module using module flags. This allows the optimizer to use this information.
Richard Smith [Fri, 11 Dec 2015 22:41:00 +0000 (22:41 +0000)]
[modules] If the semantic and lexical DC of a decl are the same, write out the
second one as 0 instead of writing the same bits to the module file twice.
This typically reduces PCM file size by about 1%.
Richard Smith [Fri, 11 Dec 2015 22:39:52 +0000 (22:39 +0000)]
Allow non-defining declarations of class template partial specializations to
have a nested name specifier. Strictly speaking, forward declarations of class
template partial specializations are not permitted at all, but that seems like
an obvious wording defect, and if we allow them without a nested name specifier
we should also allow them with a nested name specifier.
Nick Lewycky [Fri, 11 Dec 2015 21:28:55 +0000 (21:28 +0000)]
Error on redeclaring with a conflicting asm label and on redeclaring with an asm label after the first ODR-use. Detects problems like the one in PR22830 where gcc and clang both compiled the file but with different behaviour.
[PGO] Stop using invalid char in instr variable names.
(This is part-2 of the patch -- fixing test cases)
Before the patch, -fprofile-instr-generate compile will fail
if no integrated-as is specified when the file contains
any static functions (the -S output is also invalid).
This patch fixed the issue. With the change, the index format
version will be bumped up by 1. Backward compatibility is
preserved with this change.
This improves the coverage for the multilib directories used for ARM. Also add
tests covering the internal triple (thumbv7-*). The Juno board can be run in
this configuration.
Ben Langmuir [Thu, 10 Dec 2015 23:41:39 +0000 (23:41 +0000)]
[VFS] Fix status() of opened redirected file
Make RedirectedFileSystem::openFilForRead(path)->status() the same as
RedirectedFileSystem::status(path). Previously we would just get the
status of the underlying real file, which would not have the IsVFSMapped
bit set.
This fixes rebuilding a module that has an include that is relative to
the includer where we will lookup the real path of that file before we
lookup the VFS location.
John McCall [Thu, 10 Dec 2015 23:31:01 +0000 (23:31 +0000)]
In Objective-C, ignore attempts to redefine the ARC/GC qualifier macros.
This works around existing system headers which unconditionally
redefine these macros.
This is reasonably safe to do because we used to warn about it anyway
(outside of system headers). Continue to warn if the redefinition
would have changed the expansion. Still permit redefinition if the
macro is explicitly #undef'ed first.
Douglas Gregor [Thu, 10 Dec 2015 23:02:09 +0000 (23:02 +0000)]
Objective-C properties: merge attributes when redeclaring 'readonly' as 'readwrite' in an extension.
r251874 stopped back-patching the AST when an Objective-C 'readonly'
property is redeclared in a class extension as 'readwrite'. However,
it did not properly handle merging of Objective-C property attributes
(e.g., getter name, ownership, atomicity) to the redeclaration,
leading to bad metadata. Merge (and check!) those property attributes
so we get the right metadata and reasonable ASTs. Fixes
rdar://problem/23823989.
Do not generate DW_TAG_imported_module for anonymous namespaces (even nested) for all the platforms except PS4.
For PS4, generate explicit import for anonymous namespaces and mark it by DW_AT_artificial attribute.
These attributes were previously unexposed. Expose them through the libclang
interfaces. Add tests that cover both the MSVC spelling and the GNU spelling.
Alexey Bataev [Thu, 10 Dec 2015 08:20:58 +0000 (08:20 +0000)]
[OPENMP] Fixed processing of predetermined data-sharing attributes
Predetermined data-shared attributes for local variables are now considered as implicit. Also, patch prohibits changin of DSA for static memebers of classes.
Alexey Bataev [Thu, 10 Dec 2015 05:47:10 +0000 (05:47 +0000)]
[OPENMP] Make -fopenmp to turn on OpenMP support by default, clang part
Patch turns on OpenMP support in clang by default after fixing OpenMP buildbots.
Differential Revision: http://reviews.llvm.org/D13803
Faisal Vali [Thu, 10 Dec 2015 05:36:39 +0000 (05:36 +0000)]
Fix PR24694 (CWG1591): Deducing array bound and element type from initializer list
https://llvm.org/bugs/show_bug.cgi?id=24694
http://wg21.link/cwg1591
Teach DeduceFromInitializerList in SemaTemplateDeduction.cpp to deduce against array (constant and dependent sized) parameters (really, reference to arrays since they don't decay to pointers), by checking if the template parameter is either one of those kinds of arrays, and if so, deducing each initializer list element against the element type, and then deducing the array bound if needed.
In brief, this patch enables the following code:
template<class T, int N> int *f(T (&&)[N]);
int *ip = f({1, 2, 3});
Alexey Bataev [Thu, 10 Dec 2015 04:38:18 +0000 (04:38 +0000)]
[MSVC] Fix for http://llvm.org/PR25636: indexed accessor property not supported correctly.
All problems described in http://llvm.org/PR25636 are implemented except for return value of the 'put' property. This patch fixes this problem with the indexed properties
Differential Revision: http://reviews.llvm.org/D15174
Richard Smith [Thu, 10 Dec 2015 01:11:47 +0000 (01:11 +0000)]
PR25416: Improve performance of processing inline assembly consisting of many
implicitly-concatenated string literals. When looking for the start of a token
in the inline assembly, start from the end of the previous token, not the start
of the entire string.
Douglas Gregor [Wed, 9 Dec 2015 22:57:32 +0000 (22:57 +0000)]
Objective-C properties: loosen 'atomic' checking for readonly properties.
r251874 reworked the way we handle properties declared within
Objective-C class extensions, which had the effective of tightening up
property checking in a number of places. In this particular class of
cases, we end up complaining about "atomic" mismatches between an
implicitly-atomic, readonly property and a nonatomic, readwrite
property, which doesn't make sense because "atomic" is essentially
irrelevant to readonly properties.
Therefore, suppress this diagnostic when the readonly property is
implicitly atomic. Fixes rdar://problem/23803109.
Douglas Gregor [Tue, 8 Dec 2015 22:45:17 +0000 (22:45 +0000)]
Objective-C properties: fix bogus use of "isa<>" on a QualType.
The code used "isa" to check the type and then "getAs" to look through
sugar; we need to look through the sugar when checking, too, otherwise
any kind of sugar (nullability qualifiers in the example; or a
typedef) will thwart this semantic check. Fixes rdar://problem/23804250.
[Sema] Add warning when comparing nonnull and null
Currently, we emit warnings in some cases where nonnull function
parameters are compared against null. This patch extends this support
to warn when comparing the result of `returns_nonnull` functions
against null.
More specifically, we will now warn cases like:
int *foo() __attribute__((returns_nonnull));
int main() {
if (foo() == NULL) {} // warning: will always evaluate to false
}
Simon Pilgrim [Tue, 8 Dec 2015 21:16:45 +0000 (21:16 +0000)]
[X86][AVX2] Stripped backend codegen tests
As discussed on the ml, backend tests need to be put in llvm/test/CodeGen/X86 as fast-isel tests using IR that is as close to what is generated here as possible.
Alexey Bataev [Tue, 8 Dec 2015 12:06:20 +0000 (12:06 +0000)]
[OPENMP 4.5] Parsing/sema for 'num_tasks' clause.
OpenMP 4.5 adds directives 'taskloop' and 'taskloop simd'. These directives support clause 'num_tasks'. Patch adds parsing/semantic analysis for this clause.
Richard Smith [Tue, 8 Dec 2015 03:21:47 +0000 (03:21 +0000)]
Explicitly permit undefined behavior in constant initializers for global
variables in C, in the cases where we can constant-fold it to a value
regardless (such as floating-point division by zero and signed integer
overflow). Strictly enforcing this rule breaks too much code.
Reid Kleckner [Tue, 8 Dec 2015 01:08:09 +0000 (01:08 +0000)]
[diagnostics] Avoid crashes while printing macro backtraces
When attempting to map a source into a given level of macro expansion,
this code was ignoring the possibility that the start and end of the
range might take wildly different paths through the tree of macro
expansions. It was assuming that the begin spelling location would
always precede the end spelling location, which is false. A macro can
easily transpose its arguments.
This also fixes a related issue where there are extra macro arguments
between the begin location and the end location. In this situation, we
now highlight the entire macro invocation.
Devin Coughlin [Mon, 7 Dec 2015 23:01:53 +0000 (23:01 +0000)]
[analyzer] Fix crash when lambda captures a variable-length array.
When a C++ lambda captures a variable-length array, it creates a capture
field to store the size of the array. The initialization expression for this
capture is null, which led the analyzer to crash when initializing the field.
To avoid this, use the size expression from the VLA type to determine the
initialization value.
Teresa Johnson [Mon, 7 Dec 2015 20:40:36 +0000 (20:40 +0000)]
Remove target specifier from new tests
Hopefully fix the remaining bot failure from r254927. Remove
target specification since it shouldn't be needed, and this causes
an error when trying to check the pass execution structure in
test/CodeGen/thinlto_backend.c on non-x86 arches.
Justin Bogner [Mon, 7 Dec 2015 20:04:57 +0000 (20:04 +0000)]
AST: defer to TypeLoc::copy in TypeLoc::initializeFullCopy
If we're initializing a TypeLoc from one that's been allocated with
different alignment, memcpy will get the padding wrong. The `copy`
method already checks and handles this case, so we should just defer
to it.
This also drops the `const` off of the `initializeFullCopy`
declarations, since it isn't even remotely true (and the compiler
notices when we try to call copy() instead of tricking it with
memcpy).
Teresa Johnson [Mon, 7 Dec 2015 19:21:34 +0000 (19:21 +0000)]
[ThinLTO] Option to invoke ThinLTO backend passes and importing
Summary:
Adds new option -fthinlto-index=<file> to invoke the LTO pipeline
along with function importing via clang using the supplied function
summary index file. This supports invoking the parallel ThinLTO
backend processes in a distributed build environment via clang.
Additionally, this causes the module linker to be invoked on the bitcode
file being compiled to perform any necessary promotion and renaming of
locals that are exported via the function summary index file.
Add a couple tests that confirm we get expected errors when we try to
use the new option on a file that isn't bitcode, or specify an invalid
index file. The tests also confirm that we trigger the expected function
import pass.
Alexey Bataev [Mon, 7 Dec 2015 12:52:51 +0000 (12:52 +0000)]
[OPENMP 4.5] parsing/sema support for 'grainsize' clause.
OpenMP 4.5 adds 'taksloop' and 'taskloop simd' directives, which have 'grainsize' clause. Patch adds parsing/sema analysis of this clause.
Alexey Bataev [Mon, 7 Dec 2015 10:51:44 +0000 (10:51 +0000)]
[OPENMP 4.5] parsing/sema support for 'nogroup' clause.
OpenMP 4.5 adds 'taskloop' and 'taskloop simd' directives. These directives have new 'nogroup' clause. Patch adds basic parsing/sema support for this clause.
Faisal Vali [Mon, 7 Dec 2015 02:37:44 +0000 (02:37 +0000)]
Fix PR20334: invalid assertion while diagnosing list initialization failure
https://llvm.org/bugs/show_bug.cgi?id=20334
Unfortunately, clang currently checks for a certain brokenness of implementations of std::initializer_list in CodeGen (void
AggExprEmitter::VisitCXXStdInitializerListExpr), not in SemaInit. Until that is fixed, make sure we don't let broken attempts that are aggregates leak through into sema, which allows maintenance of expected invariants, and avoids triggering an assertion.
Serge Pavlov [Sun, 6 Dec 2015 14:32:39 +0000 (14:32 +0000)]
[PGO] Instrument only base constructors and destructors.
Constructors and destructors may be represented by several functions
in IR. Only base structors correspond to source code, others are
small pieces of code and eventually call the base variant. In this
case instrumentation of non-base structors has little sense, this
fix remove it. Now profile data of a declaration corresponds to
exactly one function in IR, it agrees with the current logic of the
profile data loading.
Simon Pilgrim [Sat, 5 Dec 2015 11:12:23 +0000 (11:12 +0000)]
[X86][3DNow!] Stripped backend codegen tests
As discussed on the ml, backend tests need to be put in llvm/test/CodeGen/X86 as fast-isel tests using IR that is as close to what is generated here as possible.
Simon Pilgrim [Sat, 5 Dec 2015 10:37:35 +0000 (10:37 +0000)]
[X86][F16C] Stripped backend codegen tests
As discussed on the ml, backend tests need to be put in llvm/test/CodeGen/X86 as fast-isel tests using IR that is as close to what is generated here as possible.
Craig Topper [Sat, 5 Dec 2015 07:41:44 +0000 (07:41 +0000)]
[AST] Use std::copy and std::fill to simplify some memcpy and memset calls. Also const-correct some methods being used since the std::copy catches the mismatch where memcpy didn't.
Reid Kleckner [Sat, 5 Dec 2015 01:52:14 +0000 (01:52 +0000)]
Revert "[x86] Exclusion of incorrect include headers paths for MCU target"
This reverts commit r254195.
From the description, I suspect that the wrong patch was committed here,
and this is causing assertion failures in EmitDeferred() when the global
value ends up being a bitcast of a global.
Devin Coughlin [Sat, 5 Dec 2015 00:22:36 +0000 (00:22 +0000)]
[analyzer] Fix MemRegion crash casting non-struct to derived struct (PR25426).
This commit prevents MemRegion::getAsOffset() from crashing when the analyzed
program casts a symbolic region of a non-record type to some derived type and
then attempts to access a field of the base type.
Nico Weber [Fri, 4 Dec 2015 19:14:14 +0000 (19:14 +0000)]
Fix debug info for Objective-C properties from class extensions after r251874
After r251874, properties from class extensions no longer show up in
ObjCInterfaceDecl::properties(). Make debug info emission explicitly
look for properties in class extensions before looking at direct properties.
Also add a test that checks for this. There are three interesting cases:
1. A property is only declared in a class extension, and the @implementation
is in a different file. This used to generated a DIObjcProperty before
r251874 and does again with this fix.
2. A property is declared as readonly in the class itself and redeclared as
readwrite in a class extension. clang before r251874 put the DIObjcProperty
on the first declaration. clang after r251874 didn't emit any DIObjcProperty,
and clang with this fix puts it on the readwrite redeclaration (which is
what lookup finds). This seems like a progression.
3. Like 2, but with an @implementation in the same file. In this case,
the property debug info gets generated a second time through the ivar
from the definition. In this case, lookup and declaration code need
to agree on the line number so that the DIObjcProperty isn't emitted
twice. In this case, clang before r251874 emitted one DIObjcProperty
on the first declaration, clang with r251874 emitted one on the second
declaration, and clang with this patch still does the latter.
projects / clang / log