Matt Arsenault [Fri, 1 Jul 2016 17:38:14 +0000 (17:38 +0000)]
Emit more intrinsics for builtin functions
This is important for building libclc. Since r273039 tests are failing
due to now emitting calls to these functions instead of emitting the
DAG node. The libm function names are implemented for OpenCL, and should
call the locally defined versions, so -fno-builtin is used. The IR
Some functions use the __builtins and expect the intrinsics to be
emitted. Without this we end up with nobuiltin calls to intrinsics
or to unsupported library calls.
Eric Fiselier [Fri, 1 Jul 2016 01:24:09 +0000 (01:24 +0000)]
[Feature] Add a builtin for indexing into parameter packs. Patch by Louis Dionne.
This patch adds a __nth_element builtin that allows fetching the n-th type of a
parameter pack with very little compile-time overhead. The patch was inspired by
r252036 and r252115 by David Majnemer, which add a similar __make_integer_seq
builtin for efficiently creating a std::integer_sequence.
Reviewed as D15421. http://reviews.llvm.org/D15421
Samuel Antao [Thu, 30 Jun 2016 21:22:08 +0000 (21:22 +0000)]
[OpenMP] Use fopenmp prefix for all options introduced by the offloading implementation.
Summary: This patch changes the options used by offloading to start with -fopenmp instead of -fomp. This makes the option naming more consistent and materializes a suggestion by Richard Smith in http://reviews.llvm.org/D9888.
Justin Lebar [Thu, 30 Jun 2016 20:29:29 +0000 (20:29 +0000)]
Fix ASTMatchersNodeTest to work on Windows.
It was failing because it had an explicit check for whether we're on
Windows.
There are a few other similar explicit checks in this file which I
didn't remove because they serve as reasonable documentation that the
test doesn't work with a Windows triple.
Richard Smith [Thu, 30 Jun 2016 20:24:30 +0000 (20:24 +0000)]
Fix typo-correction crash if a typo occurs within the operand of a
function-style cast to a non-dependent type which is then used in an invalid
way. We'd lose the "type dependent" bit here, and downstream Sema processing
would then discard the expression if it was used in a context where its type
rendered it invalid.
Summary:
This lets LLVM perform IPO over these functions. In particular, it
allows LLVM to emit ld.global.nc for loads to __restrict pointers in
kernels that are never written to.
Justin Lebar [Thu, 30 Jun 2016 18:12:25 +0000 (18:12 +0000)]
Don't instantiate a full host toolchain in ASTMatchersTest.
Summary:
This test was stat()'ing large swaths of /usr/lib hundreds of times, as
every invocation of matchesConditionally*() created a new Linux
toolchain.
In addition to being slow, perf indicated this was causing substantial
contention in the kernel.
Something is...interesting in the kernel, as without this patch I
sometimes see ~11m spent in the kernel, and sometimes ~5m. This
corresponds to bimodal ninja check-clang times of ~30s and ~20s.
It's not clear to me exactly what causes the bimodality. In any case,
this change makes this test run in 2.5s, down from 17s, and it seems to
cause us to get the 20s ninja check-clang time unconditionally.
Reid Kleckner [Thu, 30 Jun 2016 17:41:31 +0000 (17:41 +0000)]
[codeview] Emit qualified display names if -gline-tables-only is on
When -gmlt is on, we don't emit namespace or class scope information,
and the CodeView emission code in LLVM can't compute the fully qualified
name. If we know LLVM won't be able to get the name right, go ahead and
emit the qualified name in the frontend.
We could change our -gmlt emission strategy to include those scopes when
emitting codeview, but that would increase memory usage and slow down
LTO and add more complexity to debug info emission.
The same problem exists when you debug a -gmlt binary with GDB, so we
should consider removing '&& EmitCodeView' from the condition here at
some point in the future after evaluating the impact on object file
size.
Dmitry Polukhin [Thu, 30 Jun 2016 09:40:38 +0000 (09:40 +0000)]
[GCC] PR23529 Mangler part of attrbute abi_tag support
Original patch by Stefan Bühler http://reviews.llvm.org/D12834
Difference between original and this one:
- fixed all failing tests
- fixed mangling for global variable outside namespace
- emit ABI tags for guards and local names
- clang-format + other stylistic changes
- significantly reworked patch according to Richard's suggestions
Sema part, committed before http://reviews.llvm.org/D17567
Nikolay Haustov [Thu, 30 Jun 2016 09:06:33 +0000 (09:06 +0000)]
AMDGPU: Set amdgpu_kernel calling convention for OpenCL kernels.
Summary:
Summary:
Change Clang calling convention SpirKernel to OpenCLKernel.
Set calling convention OpenCLKernel for amdgcn as well.
Add virtual method .getOpenCLKernelCallingConv() to TargetCodeGenInfo
and use it to set target calling convention for AMDGPU and SPIR.
Update tests.
[ClangConfig] Unset some variables after we are done using them to configure ClangConfig.cmake.in files.
This ensures that the values do not bleed over in between computations. It may
make sense in the future to just refactor this code into functions to provide
"true scoping".
[ClangConfig] Instead of hard coding the ClangTargets location to /ClangTargets.cmake, follow LLVM's example and use a pre-computed cmake variable @CLANG_CONFIG_EXPORTS_FILE@.
This just makes ClangConfig more consistent with LLVMConfig.
Richard Smith [Wed, 29 Jun 2016 21:17:59 +0000 (21:17 +0000)]
P0305R1: Parsing support for init-statements in 'if' and 'switch' statements.
No semantic analysis yet.
This is a pain to disambiguate correctly, because the parsing rules for the
declaration form of a condition and of an init-statement are quite different --
for a token sequence that looks like a declaration, we frequently need to
disambiguate all the way to the ')' or ';'.
We could do better here in some cases by stopping disambiguation once we've
decided whether we've got an expression or not (rather than keeping going until
we know whether it's an init-statement declaration or a condition declaration),
by unifying our parsing code for the two types of declaration and moving the
syntactic checks into Sema; if this has a measurable impact on parsing
performance, I'll look into that.
[ClangConfig] Copy ClangConfig.cmake to ${CMAKE_BINARY_DIR}/${CLANG_INSTALL_PACKAGE_DIR} instead of to ${CLANG_BINARY_DIR}/${CLANG_INSTALL_PACKAGE_DIR}.
This is an obvious bug since ClangConfig.cmake looks for ClangTargets.cmake in
${CMAKE_CURRENT_LIST_DIR}. But ClangTargets.cmake is in
${CMAKE_BINARY_DIR}/${CLANG_INSTALL_PACKAGE_DIR}, so it will always fail with an
in tree build.
In the case where clang is built out of tree, this is still correct since
CMAKE_BINARY_DIR and CLANG_BINARY_DIR will be the same.
Reid Kleckner [Wed, 29 Jun 2016 18:29:21 +0000 (18:29 +0000)]
Re-land "[MS] Don't expect vftables to be provided for extern template instantiations"
Reverts r273305 and re-instates r273296.
We needed to fix a bug in Sema::MarkVTableUsed to ensure that operator
delete lookup occurs when the vtable is referenced. We already had a
special case to look up operator delete when dllimport was used, but I
think should really mark virtual destructors referenced any time the
vtable is used.
Adam Nemet [Wed, 29 Jun 2016 04:55:31 +0000 (04:55 +0000)]
[Diag] Add getter shouldAlwaysPrint. NFC
For the new hotness attribute, the API will take the pass rather than
the pass name so we can no longer play the trick of AlwaysPrint being a
special pass name. This adds a getter to help the transition.
Richard Smith [Wed, 29 Jun 2016 01:10:27 +0000 (01:10 +0000)]
Mark inheriting constructors as deleted if the corresponding defaulted default
constructor would be; this is effectively required by P0136R1. This has the
effect of exposing the validity of the base class initialization steps to
SFINAE checks.
Sean Silva [Wed, 29 Jun 2016 00:29:23 +0000 (00:29 +0000)]
Revert "[PS4] Tighten up a test (noticed in passing)"
This reverts commit r269709.
r262285 changed this deliberately so that the test would not be
sensitive to which binaries are in the same directory as clang.
See the commit message of that commit for more background.
Manman Ren [Tue, 28 Jun 2016 23:01:49 +0000 (23:01 +0000)]
ObjC Class Property: diagnostics when accessing a class property using instance.
When a class property is accessed with an object instance, before this commit,
we try to apply a typo correction of the same property:
property 'c' not found on object of type 'A *'; did you mean 'c'?
With this commit, we correctly emit a diagnostics:
property 'c' is a class property; did you mean to access it with class 'A'?
Manman Ren [Tue, 28 Jun 2016 20:55:30 +0000 (20:55 +0000)]
AvailabilityAttr: we accept "macos" as the platform name.
We continue accepting "macosx" but canonicalize it to "macos", When emitting
diagnostics, we use "macOS" instead of "OS X".
The PlatformName in TargetInfo is changed from "macosx" to "macos" so we can
directly compare the Platform in AvailabilityAttr with the PlatformName
in TargetInfo.
Replace inheriting constructors implementation with new approach, voted into
C++ last year as a DR against C++11.
Instead of synthesizing a set of derived class constructors for each inherited
base class constructor, we make the constructors of the base class visible to
constructor lookup in the derived class, using the normal rules for
using-declarations.
For constructors, UsingShadowDecl now has a ConstructorUsingShadowDecl derived
class that tracks the requisite additional information. We create shadow
constructors (not found by name lookup) in the derived class to model the
actual initialization, and have a new expression node,
CXXInheritedCtorInitExpr, to model the initialization of a base class from such
a constructor. (This initialization is special because it performs real perfect
forwarding of arguments.)
In cases where argument forwarding is not possible (for inalloca calls,
variadic calls, and calls with callee parameter cleanup), the shadow inheriting
constructor is not emitted and instead we directly emit the initialization code
into the caller of the inherited constructor.
Note that this new model is not perfectly compatible with the old model in some
corner cases. In particular:
* if B inherits a private constructor from A, and C uses that constructor to
construct a B, then we previously required that A befriends B and B
befriends C, but the new rules require A to befriend C directly, and
* if a derived class has its own constructors (and so its implicit default
constructor is suppressed), it may still inherit a default constructor from
a base class
Chris Bieneman [Tue, 28 Jun 2016 18:32:22 +0000 (18:32 +0000)]
[CMake] Connect check-compiler-rt to check-all
When using the LLVM_BUILD_EXTERNAL_COMPILER_RT option with
LLVM_ENABLE_TESTS we should also bind check-compiler-rt to check-all so
that the compiler-rt tests run too.
David Majnemer [Tue, 28 Jun 2016 03:13:16 +0000 (03:13 +0000)]
[clang-cl] Define _MSVC_LANG
Recently, Microsoft added support for a flag, /std, which controls which
version of the language rules MSVC should use.
MSVC hasn't updated __cplusplus though.
Instead, they added a new macro, _MSVC_LANG, which is defined in a
similar fashion to __cplusplus. This is used to indicate which mode the
compiler is in.
Jordan Rose [Tue, 28 Jun 2016 01:02:31 +0000 (01:02 +0000)]
Avoid accessing an invalid PresumedLoc.
DiagnosticNoteRenderer asserts trying to emit its "while building
module Foo imported from bar.h:5" note when the presumed location
of the import is invalid. This assertion was added in r267914,
where most uses of 'getFilename' were updated to test 'isValid'
instead. This one must have been missed.
I can't come up with a test because this location is always valid
in C-based code, but external clients that manually import modules
(*cough*Swift*cough*) sometimes provide invalid SourceLocations.
Richard Smith [Mon, 27 Jun 2016 19:43:46 +0000 (19:43 +0000)]
Add simple, stupid, pattern-based fuzzer / reducer for modules bugs. I've
already used this to find and reduce quite a few bugs, and it works pretty well
if you can find the right patterns.
[ExprConstant] Fix PR28314 - crash while evluating objectsize.
This fixes a crash in code like:
```
struct A {
struct B b;
char c[1];
}
int foo(struct A* a) { return __builtin_object_size(a->c, 0); }
```
We wouldn't check whether the structs we were examining were invalid,
and getting the layout of an invalid struct is (unsurprisingly) A Bad
Thing. With this patch, we'll always return conservatively if we see an
invalid struct, since I'm assuming the presence of an invalid struct
means that our compilation failed (so having a conservative result isn't
such a big deal).
Carlo Bertolli [Mon, 27 Jun 2016 14:55:37 +0000 (14:55 +0000)]
Resubmission of http://reviews.llvm.org/D21564 after fixes.
[OpenMP] Initial implementation of parse and sema for composite pragma 'distribute parallel for'
This patch is an initial implementation for #distribute parallel for.
The main differences that affect other pragmas are:
The implementation of 'distribute parallel for' requires blocking of the associated loop, where blocks are "distributed" to different teams and iterations within each block are scheduled to parallel threads within each team. To implement blocking, sema creates two additional worksharing directive fields that are used to pass the team assigned block lower and upper bounds through the outlined function resulting from 'parallel'. In this way, scheduling for 'for' to threads can use those bounds.
As a consequence of blocking, the stride of 'distribute' is not 1 but it is equal to the blocking size. This is returned by the runtime and sema prepares a DistIncrExpr variable to hold that value.
As a consequence of blocking, the global upper bound (EnsureUpperBound) expression of the 'for' is not the original loop upper bound (e.g. in for(i = 0 ; i < N; i++) this is 'N') but it is the team-assigned block upper bound. Sema creates a new expression holding the calculation of the actual upper bound for 'for' as UB = min(UB, PrevUB), where UB is the loop upper bound, and PrevUB is the team-assigned block upper bound.
Richard Smith [Sat, 25 Jun 2016 00:15:56 +0000 (00:15 +0000)]
Implement C++17 P0386R2, inline variables. (The 'inline' specifier gives a
variable weak discardable linkage and partially-ordered initialization, and is
implied for constexpr static data members.)
Chris Bieneman [Fri, 24 Jun 2016 20:21:12 +0000 (20:21 +0000)]
[CMake] Remove CLANG_APPEND_VC_REV option
I added this option in r257827 to try and add compatibility with autoconf. At the time I misunderstood the problem.
Our CMake automatically generates the SVN revision information and generates a build action to update it so builds don't need to be re-configured on SCM update (which is a better solution than we had in autoconf).
The problem I was actually seeing was isolated cases where SVN revision information isn't available because the repository structures have been removed. This happens in some automated testing systems.
This patch allows SVN_REVISION to be overridden if the build configuration could not find the SCM repository structures, and removes the code from my original patch because it is unnecessary.
Carlo Bertolli [Fri, 24 Jun 2016 18:53:35 +0000 (18:53 +0000)]
[OpenMP] Initial implementation of parse and sema for composite pragma 'distribute parallel for'
http://reviews.llvm.org/D21564
This patch is an initial implementation for #distribute parallel for.
The main differences that affect other pragmas are:
The implementation of 'distribute parallel for' requires blocking of the associated loop, where blocks are "distributed" to different teams and iterations within each block are scheduled to parallel threads within each team. To implement blocking, sema creates two additional worksharing directive fields that are used to pass the team assigned block lower and upper bounds through the outlined function resulting from 'parallel'. In this way, scheduling for 'for' to threads can use those bounds.
As a consequence of blocking, the stride of 'distribute' is not 1 but it is equal to the blocking size. This is returned by the runtime and sema prepares a DistIncrExpr variable to hold that value.
As a consequence of blocking, the global upper bound (EnsureUpperBound) expression of the 'for' is not the original loop upper bound (e.g. in for(i = 0 ; i < N; i++) this is 'N') but it is the team-assigned block upper bound. Sema creates a new expression holding the calculation of the actual upper bound for 'for' as UB = min(UB, PrevUB), where UB is the loop upper bound, and PrevUB is the team-assigned block upper bound.
projects / clang / log