Commit Graph

2704 Commits

Author SHA1 Message Date
Yaxun Liu 6d96f16347 CodeGen: Cast alloca to expected address space
Alloca always returns a pointer in alloca address space, which may
be different from the type defined by the language. For example,
in C++ the auto variables are in the default address space. Therefore
cast alloca to the expected address space when necessary.

Differential Revision: https://reviews.llvm.org/D32248

llvm-svn: 303370
2017-05-18 18:51:09 +00:00
James Y Knight eb96e44aea [SPARC] Support 'f' and 'e' inline asm constraints.
Patch by Patrick Boettcher.

Differential Revision: https://reviews.llvm.org/D29117

llvm-svn: 302913
2017-05-12 16:01:23 +00:00
Simon Pilgrim 3511348dbb [X86][LWP] Add clang support for LWP instructions.
This patch adds support for the the LightWeight Profiling (LWP) instructions which are available on all AMD Bulldozer class CPUs (bdver1 to bdver4).

Differential Revision: https://reviews.llvm.org/D32770

llvm-svn: 302418
2017-05-08 12:09:45 +00:00
Weiming Zhao be380c711c [ARM] Limit the diagnose when an ISR calls a regular function
Summary:
When the function is compiled with soft-float or on CPU with no FPU, we
don't need to diagnose for a call from an ISR to a regular function.

Reviewers: jroelofs, eli.friedman

Reviewed By: jroelofs

Subscribers: aemerson, rengolin, javed.absar, cfe-commits

Differential Revision: https://reviews.llvm.org/D32918

llvm-svn: 302274
2017-05-05 19:25:29 +00:00
Alex Lorenz ce4518f21d DiagnosticsEngine should clear DelayedDiagID before reporting the
delayed diagnostic

This fix avoids an infinite recursion that was uncovered in one of our internal
tests by r301992. The testcase is the most reduced version of that
auto-generated test.

This is an improved version of the reverted commit r302037. The previous fix
actually managed to expose another subtle bug whereby `fatal_too_many_errors`
error was reported twice, with the second report setting the
`FatalErrorOccurred` flag. That prevented the notes that followed the diagnostic
the caused `fatal_too_many_errors` to be emitted. This commit ensures that notes
that follow `fatal_too_many_errors` but that belong to the diagnostic that
caused `fatal_too_many_errors` won't be emitted by setting the
`FatalErrorOccurred` when emitting `fatal_too_many_errors`.

rdar://31962618

llvm-svn: 302151
2017-05-04 13:56:51 +00:00
Alex Lorenz 5191e80b87 Revert r302037
The commit caused the following two buildbot failures:

    Clang :: Misc/error-limit-multiple-notes.cpp
    Clang :: Misc/error-limit.c

llvm-svn: 302046
2017-05-03 16:22:16 +00:00
Alex Lorenz d7dfec818b DiagnosticsEngine should clear DelayedDiagID before reporting the
delayed diagnostic

This avoids an infinite loop that was uncovered in one of our internal tests
by r301992. The testcase is the most reduced version of that auto-generated
test.

rdar://31962618

llvm-svn: 302037
2017-05-03 15:41:16 +00:00
Richard Smith e37391c4fe [modules] Round-trip -Werror flag through explicit module build.
The intent for an explicit module build is that the diagnostics produced within
the module are those that were configured when the module was built, not those
that are enabled within a user of the module. This includes diagnostics that
don't actually show up until the module is used (for instance, diagnostics
produced during template instantiation and weird cases like -Wpadded).

We serialized and restored the diagnostic state for individual warning groups,
but previously did not track the state for flags like -Werror and -Weverything,
which are implemented as separate bits rather than as part of the diagnostics
mapping information.

llvm-svn: 301992
2017-05-03 00:28:49 +00:00
Duncan P. N. Exon Smith 16b1ac950b Darwin: Define __STDC_NO_THREADS__ on Darwin targets
Darwin doesn't support C11 threads.h.  Define `__STDC_NO_THREADS__` so
that users can check for it.

rdar://problem/18461003

llvm-svn: 301508
2017-04-27 01:47:22 +00:00
Carlo Bertolli ffafe10fac [OpenMP] Prepare sema to support combined constructs with omp distribute and omp for
https://reviews.llvm.org/D32237

This patch prepares sema with additional fields to support all those composite and combined constructs of OpenMP that include pragma 'distribute' and 'for', such as 'distribute parallel for'. It also extends the regression tests for 'distribute parallel for' and adds a new one.

llvm-svn: 300802
2017-04-20 00:39:39 +00:00
Alex Lorenz 9e7bf161ea Add #pragma clang attribute
This is a recommit of r300539 that was reverted in r300543 due to test failures.
The original commit message is displayed below:

The new '#pragma clang attribute' directive can be used to apply attributes to
multiple declarations. An attribute must satisfy the following conditions to
be supported by the pragma:
- It must have a subject list that's defined in the TableGen file.
- It must be documented.
- It must not be late parsed.
- It must have a GNU/C++11 spelling.

Differential Revision: https://reviews.llvm.org/D30009

llvm-svn: 300556
2017-04-18 14:33:39 +00:00
Oliver Stannard 233310f3e6 [ARM,AArch64] Define __ELF__ for arm-none-eabihf and AArch64
This macro is defined for arm-none-eabi as of r266625, but it should also be
defined for eabihf and aarch64.

llvm-svn: 300549
2017-04-18 13:12:36 +00:00
Alex Lorenz 3bfe962afa Revert r300539 - Add #pragma clang attribute
Some tests fail on the Windows buildbots. I will have to investigate more.
This commit reverts r300539, r300540 and r300542.

llvm-svn: 300543
2017-04-18 10:46:41 +00:00
Alex Lorenz 0a849f47d2 Add #pragma clang attribute
The new '#pragma clang attribute' directive can be used to apply attributes to
multiple declarations. An attribute must satisfy the following conditions to
be supported by the pragma:
- It must have a subject list that's defined in the TableGen file.
- It must be documented.
- It must not be late parsed.
- It must have a GNU/C++11 spelling.

Differential Revision: https://reviews.llvm.org/D30009

llvm-svn: 300539
2017-04-18 09:41:47 +00:00
Eric Christopher d26d8839d8 When we turn on vsx it should also turn on altivec explicitly, same
with disabling it as well as disabling all vsx specific features when
turning off altivec.

Fixes PR32663.

llvm-svn: 300395
2017-04-15 06:15:00 +00:00
Konstantin Zhuravlyov e668b1cd1e [AMDGPU][GFX9] Set +fp32-denormals for >=gfx900 unless -cl-denorms-are-zero is set
Differential Revision: https://reviews.llvm.org/D31482

llvm-svn: 300306
2017-04-14 05:33:57 +00:00
Duncan P. N. Exon Smith 900f817591 Serialization: Simulate -Werror settings in implicit modules
r293123 started serializing diagnostic pragma state for modules.  This
makes the serialization work properly for implicit modules.

An implicit module build (using Clang's internal build system) uses the
same PCM file location for different `-Werror` levels.

E.g., if a TU has `-Werror=format` and tries to load a PCM built without
`-Werror=format`, a new PCM will be built in its place (and the new PCM
should have the same signature, since r297655).  In the other direction,
if a TU does not have `-Werror=format` and tries to load a PCM built
with `-Werror=format`, it should "just work".

The idea is to evolve the PCM toward the strictest -Werror flags that
anyone tries.

r293123 started serializing the diagnostic pragma state for each PCM.
Since this encodes the -Werror settings at module-build time, it breaks
the implicit build model.

This commit filters the diagnostic state in order to simulate the
current compilation's diagnostic settings.  Firstly, it ignores the
module's serialized first diagnostic state, replacing it with the state
from this compilation's command-line.  Secondly, if a pragma warning was
upgraded to error/fatal when generating the PCM (e.g., due to `-Werror`
on the command-line), it checks whether it should still be upgraded in
its current context.

llvm-svn: 300025
2017-04-12 03:58:58 +00:00
David Blaikie e6b7c28d17 Modular Codegen: Add/use a bit in serialized function definitions to track whether they are the subject of modular codegen
Some decls are created not where they are written, but in other module
files/users (implicit special members and function template implicit
specializations). To correctly identify them, use a bit next to the definition
to track the modular codegen property.

Discussed whether the module file bit could be omitted in favor of
reconstituting from the modular codegen decls list - best guess today is that
the efficiency improvement of not having to deserialize the whole list whenever
any function is queried by a module user is worth it for the small size
increase of this redundant (list + bit-on-def) representation.

Reviewers: rsmith

Differential Revision: https://reviews.llvm.org/D29901

llvm-svn: 299982
2017-04-11 20:46:34 +00:00
Yaxun Liu b34ec829be [OpenCL] Map default address space to alloca address space
For OpenCL, the private address space qualifier is 0 in AST. Before this change, 0 address space qualifier
is always mapped to target address space 0. As now target private address space is specified by
alloca address space in data layout, address space qualifier 0 needs to be mapped to alloca addr space specified by the data layout.

This change has no impact on targets whose alloca addr space is 0.

With contributions from Matt Arsenault, Tony Tye and Wen-Heng (Jack) Chung

Differential Revision: https://reviews.llvm.org/D31404

llvm-svn: 299965
2017-04-11 17:24:23 +00:00
Alex Lorenz f127821140 [Parser][ObjC++] Improve diagnostics and recovery when C++ keywords are used
as identifiers in Objective-C++

This commit improves the 'expected identifier' errors that are presented when a
C++ keyword is used as an identifier in Objective-C++ by mentioning that this is
a C++ keyword in the diagnostic message. It also improves the error recovery:
the parser will now treat the C++ keywords as identifiers to prevent unrelated
parsing errors.

rdar://20626062

Differential Revision: https://reviews.llvm.org/D26503

llvm-svn: 299950
2017-04-11 15:01:53 +00:00
Yaxun Liu b122ed9181 [AMDGPU] Temporarily change constant address space from 4 to 2 for the new address space mapping
Change constant address space from 4 to 2 for the new address space mapping in Clang.

Differential Revision: https://reviews.llvm.org/D31771

llvm-svn: 299691
2017-04-06 19:18:36 +00:00
Chih-Hung Hsieh a0b99e45e9 [Basic] getColumnNumber returns location of CR+LF on Windows
When fixing a Clang-Tidy bug in D31406,
reuse of FileID enabled the missing highlightRange function.
Assertion in highlightRange failed because the end-of-range column
number was 2 + the last column of a line on Windows.
This fix is required to enable D31406.

Differential Revision: https://reviews.llvm.org/D31713

llvm-svn: 299681
2017-04-06 18:36:50 +00:00
Dean Michael Berris 835832d37a [XRay] Add -fxray-{always,never}-instrument= flags to clang
Summary:
The -fxray-always-instrument= and -fxray-never-instrument= flags take
filenames that are used to imbue the XRay instrumentation attributes
using a whitelist mechanism (similar to the sanitizer special cases
list). We use the same syntax and semantics as the sanitizer blacklists
files in the implementation.

As implemented, we respect the attributes that are already defined in
the source file (i.e. those that have the
[[clang::xray_{always,never}_instrument]] attributes) before applying
the always/never instrument lists.

Reviewers: rsmith, chandlerc

Subscribers: jfb, mgorny, cfe-commits

Differential Revision: https://reviews.llvm.org/D30388

llvm-svn: 299041
2017-03-30 00:29:36 +00:00
Eric Christopher fc6ffede65 Default enable the rtm feature only on skylake and later for now because Intel disabled the feature on some haswell and broadwell processors:
http://www.intel.com/content/dam/www/public/us/en/documents/specification-updates/core-m-processor-family-spec-update.pdf

the -mrtm option will still work normally.

llvm-svn: 298956
2017-03-28 23:03:19 +00:00
Erik Verbruggen dfffaf579f FileManager: mark virtual file entries as valid entries
The getVirtualFile method would create entries for e.g. libclang's
CXUnsavedFile but not mark them as valid. The effect is that a lookup
through getFile where the file name is not exactly matching the virtual
file (e.g. through mixing slashes and backslashes on Windows) would
result in a normal file "lookup", and re-using the file entry found
by using the UniqueID, and overwrite the file entry fields. Because the
lookup involves opening the file, and moving it into the file entry, the
file is now open. The SourceManager keys its buffers on the UniqueID
(which is still the same), so it will find an already loaded buffer.
Because only the loading a buffer from disk will close the file, the
FileEntry will hold on to an open file for as long as the FileManager
is around. As the FileManager will only get destroyed at a reparse,
you can't safe to the "leaked" and locked file on Windows.

llvm-svn: 298905
2017-03-28 09:18:05 +00:00
Eric Christopher 1e6fedbb83 _CALL_LINUX is only defined on 64-bit ppc linux platforms, not 32-bit.
Adjust and add a test for the 32-bit side.

llvm-svn: 298778
2017-03-25 19:26:04 +00:00
Yaxun Liu 354097b05d [AMDGPU] Make AMDGPUTargetInfo::AS private
llvm-svn: 298773
2017-03-25 11:34:41 +00:00
Eric Christopher 19d8c1fed2 Update the comment on not yet generated preprocessor defines to remove __LONGDOUBLE128.
llvm-svn: 298771
2017-03-25 06:38:57 +00:00
Eric Christopher ee21410bdc Add the __LONGDOUBLE128 define for ppc targets that have 128 bit long doubles.
llvm-svn: 298770
2017-03-25 06:37:23 +00:00
Eric Christopher 0d36116394 Define __HAVE_BSWAP__ on ppc to match gcc since we support both builtins as well.
llvm-svn: 298769
2017-03-25 05:40:13 +00:00
Yaxun Liu 3464f92e23 [AMDGPU] Switch address space mapping by triple environment amdgiz
For target environment amdgiz and amdgizcl (giz means Generic Is Zero), AMDGPU will use new address space mapping where generic address space is 0 and private address space is 5. The data layout is also changed correspondingly.

Differential Revision: https://reviews.llvm.org/D31210

llvm-svn: 298767
2017-03-25 03:46:25 +00:00
Eric Christopher 2fd2eda0ac Add the _CALL_LINUX preprocessor define for ppc linux platforms.
This typically is only for a new enough linker (bfd >= 2.16.2 or gold), but
our ppc suppport post-dates this and it should work on all linux platforms. It
is guaranteed to work on all elfv2 platforms.

llvm-svn: 298765
2017-03-25 03:33:59 +00:00
Eric Christopher 153dad4ca9 __BIGGEST_ALIGNMENT__ has always been 16 on all power platforms rather
than the default of 8 in clang, fix and update tests accordingly.

llvm-svn: 298761
2017-03-25 02:55:21 +00:00
Eric Christopher b70819e6a2 Add preprocessor defines for a bare powerpc64le triple/cpu.
The le triple didn't exist until power8, so use that as a default (this
also matches what gcc does).

llvm-svn: 298759
2017-03-25 02:29:18 +00:00
Eric Christopher 3646e6289d Move setting of LangOpts based on target flags out of CompilerInstance
and into TargetInfo::adjust so that it gets called in more places
throughout the compiler (AST serialization in particular).

Should fix PPC modules after removing of faltivec.

llvm-svn: 298487
2017-03-22 06:36:09 +00:00
Eric Christopher 758aad76d8 Remove the -faltivec alias option and replace it with -maltivec everywhere.
The alias was only ever used on darwin and had some issues there,
and isn't used in practice much. Also fixes a problem with -mno-altivec
not turning off -maltivec.

Also add a diagnostic for faltivec/fno-altivec that directs users to use
maltivec options and include the altivec.h file explicitly.

llvm-svn: 298449
2017-03-21 22:06:18 +00:00
Benjamin Kramer f6021ecddc Apply clang-tidy's performance-unnecessary-value-param to parts of clang.
No functionality change intended.

llvm-svn: 298443
2017-03-21 21:35:04 +00:00
Eric Christopher 74fa24ff27 Turn on HTM on power8 and later (including powerpc64le) since it's
available by default on those cpus and configurations.

llvm-svn: 298307
2017-03-20 21:12:53 +00:00
Duncan P. N. Exon Smith 030d7d6daa Reapply "Modules: Cache PCMs in memory and avoid a use-after-free"
This reverts commit r298185, effectively reapplying r298165, after fixing the
new unit tests (PR32338).  The memory buffer generator doesn't null-terminate
the MemoryBuffer it creates; this version of the commit informs getMemBuffer
about that to avoid the assert.

Original commit message follows:

----

Clang's internal build system for implicit modules uses lock files to
ensure that after a process writes a PCM it will read the same one back
in (without contention from other -cc1 commands).  Since PCMs are read
from disk repeatedly while invalidating, building, and importing, the
lock is not released quickly.  Furthermore, the LockFileManager is not
robust in every environment.  Other -cc1 commands can stall until
timeout (after about eight minutes).

This commit changes the lock file from being necessary for correctness
to a (possibly dubious) performance hack.  The remaining benefit is to
reduce duplicate work in competing -cc1 commands which depend on the
same module.  Follow-up commits will change the internal build system to
continue after a timeout, and reduce the timeout.  Perhaps we should
reconsider blocking at all.

This also fixes a use-after-free, when one part of a compilation
validates a PCM and starts using it, and another tries to swap out the
PCM for something new.

The PCMCache is a new type called MemoryBufferCache, which saves memory
buffers based on their filename.  Its ownership is shared by the
CompilerInstance and ModuleManager.

- The ModuleManager stores PCMs there that it loads from disk, never
touching the disk if the cache is hot.

- When modules fail to validate, they're removed from the cache.

- When a CompilerInstance is spawned to build a new module, each
already-loaded PCM is assumed to be valid, and is frozen to avoid
the use-after-free.

- Any newly-built module is written directly to the cache to avoid the
round-trip to the filesystem, making lock files unnecessary for
correctness.

Original patch by Manman Ren; most testcases by Adrian Prantl!

llvm-svn: 298278
2017-03-20 17:58:26 +00:00
Renato Golin f1966cf646 Revert "Modules: Cache PCMs in memory and avoid a use-after-free"
This reverts commit r298165, as it broke the ARM builds.

llvm-svn: 298185
2017-03-18 12:31:32 +00:00
Duncan P. N. Exon Smith 079c40e886 Modules: Cache PCMs in memory and avoid a use-after-free
Clang's internal build system for implicit modules uses lock files to
ensure that after a process writes a PCM it will read the same one back
in (without contention from other -cc1 commands).  Since PCMs are read
from disk repeatedly while invalidating, building, and importing, the
lock is not released quickly.  Furthermore, the LockFileManager is not
robust in every environment.  Other -cc1 commands can stall until
timeout (after about eight minutes).

This commit changes the lock file from being necessary for correctness
to a (possibly dubious) performance hack.  The remaining benefit is to
reduce duplicate work in competing -cc1 commands which depend on the
same module.  Follow-up commits will change the internal build system to
continue after a timeout, and reduce the timeout.  Perhaps we should
reconsider blocking at all.

This also fixes a use-after-free, when one part of a compilation
validates a PCM and starts using it, and another tries to swap out the
PCM for something new.

The PCMCache is a new type called MemoryBufferCache, which saves memory
buffers based on their filename.  Its ownership is shared by the
CompilerInstance and ModuleManager.

  - The ModuleManager stores PCMs there that it loads from disk, never
    touching the disk if the cache is hot.

  - When modules fail to validate, they're removed from the cache.

  - When a CompilerInstance is spawned to build a new module, each
    already-loaded PCM is assumed to be valid, and is frozen to avoid
    the use-after-free.

  - Any newly-built module is written directly to the cache to avoid the
    round-trip to the filesystem, making lock files unnecessary for
    correctness.

Original patch by Manman Ren; most testcases by Adrian Prantl!

llvm-svn: 298165
2017-03-17 22:55:13 +00:00
Krzysztof Parzyszek ff7f6675e7 [Hexagon] Recognize hexagonv62 as a valid target CPU
llvm-svn: 297778
2017-03-14 20:29:23 +00:00
Juergen Ributzka f978743907 Reapply [VFS] Ignore broken symlinks in the directory iterator.
Modified the tests to accept any iteration order, to run only on Unix, and added
additional error reporting to investigate SystemZ bot issue.

The VFS directory iterator and recursive directory iterator behave differently
from the LLVM counterparts. Once the VFS iterators hit a broken symlink they
immediately abort. The LLVM counterparts don't stat entries unless they have to
descend into the next directory, which allows to recover from this issue by
clearing the error code and skipping to the next entry.

This change adds similar behavior to the VFS iterators. There should be no
change in current behavior in the current CLANG source base, because all
clients have loop exit conditions that also check the error code.

This fixes rdar://problem/30934619.

Differential Revision: https://reviews.llvm.org/D30768

llvm-svn: 297693
2017-03-14 00:14:40 +00:00
Matt Arsenault bf5e3e4391 AMDGPU: Make 0 the private nullptr value
We can't actually pretend that 0 is valid for address space 0.
r295877 added a workaround to stop allocating user objects
there, so we can use 0 as the invalid pointer.

Some of the tests seemed to be using private as the non-0 null
test address space, so add copies using local to make sure
this is still stressed.

llvm-svn: 297659
2017-03-13 19:47:53 +00:00
Duncan P. N. Exon Smith 60fa28882e Modules: Use hash of PCM content for SIGNATURE
Change ASTFileSignature from a random 32-bit number to the hash of the
PCM content.

  - Move definition ASTFileSignature to Basic/Module.h so Module and
    ASTSourceDescriptor can use it.

  - Change the signature from uint64_t to std::array<uint32_t,5>.

  - Stop using (saving/reading) the size and modification time of PCM
    files when there is a valid SIGNATURE.

  - Add UNHASHED_CONTROL_BLOCK, and use it to store the SIGNATURE record
    and other records that shouldn't affect the hash.  Because implicit
    modules reuses the same file for multiple levels of -Werror, this
    includes DIAGNOSTIC_OPTIONS and DIAG_PRAGMA_MAPPINGS.

This helps to solve a PCH + implicit Modules dependency issue: PCH files
are handled by the external build system, whereas implicit modules are
handled by internal compiler build system.  This prevents invalidating a
PCH when the compiler overwrites a PCM file with the same content
(modulo the diagnostic differences).

Design and original patch by Manman Ren!

llvm-svn: 297655
2017-03-13 18:45:08 +00:00
Juergen Ributzka 31d01baa9f Revert "Reapply [VFS] Ignore broken symlinks in the directory iterator."
Still broken on Windows and SystemZ bot ... sorry for the noise.

llvm-svn: 297533
2017-03-11 00:14:50 +00:00
Juergen Ributzka 46541f1b0b Reapply [VFS] Ignore broken symlinks in the directory iterator.
Modified the tests to accept any iteration order.

The VFS directory iterator and recursive directory iterator behave differently
from the LLVM counterparts. Once the VFS iterators hit a broken symlink they
immediately abort. The LLVM counterparts allow to recover from this issue by
clearing the error code and skipping to the next entry.

This change adds the same functionality to the VFS iterators. There should be
no change in current behavior in the current CLANG source base, because all
clients have loop exit conditions that also check the error code.

This fixes rdar://problem/30934619.

Differential Revision: https://reviews.llvm.org/D30768

llvm-svn: 297528
2017-03-10 22:49:04 +00:00
Juergen Ributzka bc1c5b1d04 Revert r297510 "[VFS] Ignore broken symlinks in the directory iterator."
The tests are failing on one of the bots.

llvm-svn: 297517
2017-03-10 21:46:51 +00:00
Juergen Ributzka 4a25956838 [VFS] Remove the Path variable from RealFSDirIter. NFC.
This variable is set, but never used.

llvm-svn: 297511
2017-03-10 21:23:29 +00:00
Juergen Ributzka fd958fca0b [VFS] Ignore broken symlinks in the directory iterator.
The VFS directory iterator and recursive directory iterator behave differently
from the LLVM counterparts. Once the VFS iterators hit a broken symlink they
immediately abort. The LLVM counterparts allow to recover from this issue by
clearing the error code and skipping to the next entry.

This change adds the same functionality to the VFS iterators. There should be
no change in current behavior in the current CLANG source base, because all
clients have loop exit conditions that also check the error code.

This fixes rdar://problem/30934619.

Differential Revision: https://reviews.llvm.org/D30768

llvm-svn: 297510
2017-03-10 21:23:27 +00:00