Sanjay Patel [Wed, 29 Mar 2017 21:58:52 +0000 (21:58 +0000)]
[CodeGen] clean up and add tests for scalar and-of-setcc; NFC
https://bugs.llvm.org/show_bug.cgi?id=32401
llvm-svn: 299034
Adam Nemet [Wed, 29 Mar 2017 21:54:24 +0000 (21:54 +0000)]
Use FPContractModeKind universally
FPContractModeKind is the codegen option flag which is already ternary (off,
on, fast). This makes it universally the type for the contractable info
across the front-end:
* In FPOptions (i.e. in the Sema + in the expression nodes).
* In LangOpts::DefaultFPContractMode which is the option that initializes
FPOptions in the Sema.
Another way to look at this change is that before fp-contractable on/off were
the only states handled to the front-end:
* For "on", FMA folding was performed by the front-end
* For "fast", we simply forwarded the flag to TargetOptions to handle it in
LLVM
Now off/on/fast are all exposed because for fast we will generate
fast-math-flags during CodeGen.
This is toward moving fp-contraction=fast from an LLVM TargetOption to a
FastMathFlag in order to fix PR25721.
---
This is a recommit of r299027 with an adjustment to the test
CodeGenCUDA/fp-contract.cu. The test assumed that even
though -ffp-contract=on is passed FE-based folding of FMA won't happen.
This is obviously wrong since the user is asking for this explicitly with the
option. CUDA is different that -ffp-contract=fast is on by default.
The test used to "work" because contract=fast and contract=on were maintained
separately and we didn't fold in the FE because contract=fast was on due to
the target-default. This patch consolidates the contract=on/fast/off state
into a ternary state hence the change in behavior.
---
Differential Revision: https://reviews.llvm.org/D31167
llvm-svn: 299033
Francis Ricci [Wed, 29 Mar 2017 21:49:47 +0000 (21:49 +0000)]
Move current thread data out of lsan_common on linux
Summary:
Now that we have a platform-specific non-common lsan file, use
it to store non-common lsan data.
Reviewers: kubamracek
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31472
llvm-svn: 299032
Francis Ricci [Wed, 29 Mar 2017 21:49:13 +0000 (21:49 +0000)]
Remove failing check from platform specific darwin lsan initializer
Summary:
We currently don't have any platform specific darwin
lsan modules, don't force failure if they don't exist.
Reviewers: kubamracek
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31473
llvm-svn: 299031
Davide Italiano [Wed, 29 Mar 2017 21:34:58 +0000 (21:34 +0000)]
[X86IselLowering] Remove extraneous semicolon. NFCI.
Unbreaks the build with GCC -Werror.
llvm-svn: 299030
Adam Nemet [Wed, 29 Mar 2017 21:24:19 +0000 (21:24 +0000)]
Revert "Use FPContractModeKind universally"
This reverts commit r299027.
It's causing a test failure in clang's CodeGenCUDE/fp-contract.cu
llvm-svn: 299029
Pavel Labath [Wed, 29 Mar 2017 21:01:14 +0000 (21:01 +0000)]
Centralize libc++ test skipping logic
Summary:
This aims to replace the different decorators we've had on each libc++
test with a single solution. Each libc++ will be assigned to the
"libc++" category and a single central piece of code will decide whether
we are actually able to run libc++ test in the given configuration by
enabling or disabling the category (while giving the user the
opportunity to override this).
I started this effort because I wanted to get libc++ tests running on
android, and none of the existing decorators worked for this use case:
- skipIfGcc - incorrect, we can build libc++ executables on android
with gcc (in fact, after this, we can now do it on linux as well)
- lldbutil.skip_if_library_missing - this checks whether libc++.so is
loaded in the proces, which fails in case of a statically linked
libc++ (this makes copying executables to the remote target easier to
manage).
To make this work I needed to split out the pseudo_barrier code from the
force-included file, as libc++'s atomic does not play well with gcc on
linux, and this made every test fail, even though we need the code only
in the threading tests.
So far, I am only annotating one of the tests with this category. If
this does not break anything, I'll proceed to update the rest.
Reviewers: jingham, zturner, EricWF
Subscribers: srhines, lldb-commits
Differential Revision: https://reviews.llvm.org/D30984
llvm-svn: 299028
Adam Nemet [Wed, 29 Mar 2017 20:39:49 +0000 (20:39 +0000)]
Use FPContractModeKind universally
FPContractModeKind is the codegen option flag which is already ternary (off,
on, fast). This makes it universally the type for the contractable info
across the front-end:
* In FPOptions (i.e. in the Sema + in the expression nodes).
* In LangOpts::DefaultFPContractMode which is the option that initializes
FPOptions in the Sema.
Another way to look at this change is that before fp-contractable on/off were
the only states handled to the front-end:
* For "on", FMA folding was performed by the front-end
* For "fast", we simply forwarded the flag to TargetOptions to handle it in
LLVM
Now off/on/fast are all exposed because for fast we will generate
fast-math-flags during CodeGen.
This is toward moving fp-contraction=fast from an LLVM TargetOption to a
FastMathFlag in order to fix PR25721.
Differential Revision: https://reviews.llvm.org/D31167
llvm-svn: 299027
Huihui Zhang [Wed, 29 Mar 2017 20:23:56 +0000 (20:23 +0000)]
revert test commit r299024
llvm-svn: 299026
Rui Ueyama [Wed, 29 Mar 2017 20:15:29 +0000 (20:15 +0000)]
Remove parameter names from template instantiations to make them compact. NFC.
llvm-svn: 299025
Huihui Zhang [Wed, 29 Mar 2017 20:10:45 +0000 (20:10 +0000)]
test commit, add blank line
llvm-svn: 299024
Kamil Rytarowski [Wed, 29 Mar 2017 19:52:24 +0000 (19:52 +0000)]
Add NetBSD path for Debugging Information in Separate Files
Summary:
NetBSD stores debug information files in the `/usr/libdata/debug` path.
This change fixes debugging distribution executables, e.g. `look`(1):
```
$ lldb /usr/bin/look
(lldb) target create "/usr/bin/look"
Current executable set to '/usr/bin/look' (x86_64).
(lldb) b main
Breakpoint 1: where = look`main + 22 at look.c:107, address = 0x0000000000000da6
(lldb) r
Process 23473 launched: '/usr/bin/look' (x86_64)
Process 23473 stopped
* thread #1, stop reason = breakpoint 1.1
frame #0: 0x0000000186600da6 look`main(argc=1, argv=0x00007f7fffc7c488) at look.c:107
104
105 string = NULL;
106 file = _PATH_WORDS;
-> 107 termchar = '\0';
108 while ((ch = getopt(argc, argv, "dft:")) != -1)
109 switch(ch) {
110 case 'd':
(lldb)
```
There is no `/usr/lib/debug` path on NeBSD, so remove it from search.
Sponsored by <The NetBSD Foundation>
Reviewers: jingham, emaste, kettenis, labath, joerg
Reviewed By: labath
Subscribers: aprantl, #lldb
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D31461
llvm-svn: 299023
Davide Italiano [Wed, 29 Mar 2017 19:39:46 +0000 (19:39 +0000)]
[DAGCombiner] Remove else after return. NFCI.
llvm-svn: 299022
Rui Ueyama [Wed, 29 Mar 2017 19:35:44 +0000 (19:35 +0000)]
Revert r298815: Do not use assert to report broken input files.
This reverts commit because this really shouldn't happen unless
there's a bug in LLD.
llvm-svn: 299021
Sean Callanan [Wed, 29 Mar 2017 19:32:59 +0000 (19:32 +0000)]
Move the definition of SBListener::GetSP() to SBListener.cpp.
This is the requirement for all functions in the public API,
to eliminate weak symbol definitions.
llvm-svn: 299020
Adrian McCarthy [Wed, 29 Mar 2017 19:27:08 +0000 (19:27 +0000)]
Re-land: "Make NativeExeSymbol a concrete subclass of NativeRawSymbol [PDB]"
This should work on all platforms now that r299006 has landed. Tested locally
on Windows and Linux.
This moves exe symbol-specific method implementations out of NativeRawSymbol
into a concrete subclass. Also adds implementations for hasCTypes and
hasPrivateSymbols and a simple test to ensure the native reader can access the
summary information for the executable from the PDB.
Original Differential Revision: https://reviews.llvm.org/D31059
llvm-svn: 299019
Rafael Espindola [Wed, 29 Mar 2017 19:26:26 +0000 (19:26 +0000)]
Add ifunc support to ModuleSymbolTable.
Do that by creating a global_values, which is similar to
global_objects, but also iterates over aliases and ifuncs.
llvm-svn: 299018
Matthew Simpson [Wed, 29 Mar 2017 18:23:08 +0000 (18:23 +0000)]
[InstCombine] Correct the check for vector GEPs
Some of the GEP combines (e.g., descaling) can't handle vector GEPs. We have an
existing check that attempts to bail out if given a vector GEP. However, the
check only tests the GEP's pointer operand. A GEP results in a vector of
pointers if at least one of its operands is vector-typed (e.g., its pointer
operand could be a scalar, but its index could be a vector). We should just
check the type of the GEP itself. This should fix PR32414.
Reference: https://bugs.llvm.org/show_bug.cgi?id=32414
Differential Revision: https://reviews.llvm.org/D31470
llvm-svn: 299017
Filipe Cabecinhas [Wed, 29 Mar 2017 18:17:22 +0000 (18:17 +0000)]
Add allocator_frees_and_returns_null_on_realloc_zero=false flag for compatibility with allocators which allow a realloc(p, 0) and don't free the pointer.
Summary:
I know of two implementations that do this (ASan is not protecting against accessing the returned memory for now, just like malloc(0)):
SIE libc on the PS4
dlmalloc has a flag for this
This allows us to properly support this behaviour.
Reviewers: vsk, kcc
Subscribers: llvm-commits, kubamracek
Differential Revision: https://reviews.llvm.org/D31295
llvm-svn: 299016
Brian Kelley [Wed, 29 Mar 2017 18:16:38 +0000 (18:16 +0000)]
[Objective-C] Miscellaneous -fobjc-weak Fixes
Summary: After examining the remaining uses of LangOptions.ObjCAutoRefCount, found a some additional places to also check for ObjCWeak not covered by previous test cases. Added a test file to verify all the code paths that were changed.
Reviewers: rsmith, doug.gregor, rjmccall
Reviewed By: rjmccall
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D31007
llvm-svn: 299015
Brian Kelley [Wed, 29 Mar 2017 18:09:02 +0000 (18:09 +0000)]
[Objective-C] Fix "weak-unavailable" warning with -fobjc-weak
Summary: clang should produce the same errors Objective-C classes that cannot be assigned to weak pointers under both -fobjc-arc and -fobjc-weak. Check for ObjCWeak along with ObjCAutoRefCount when analyzing pointer conversions. Add an -fobjc-weak pass to the existing arc-unavailable-for-weakref test cases to verify the behavior is the same.
Reviewers: rsmith, doug.gregor, rjmccall
Reviewed By: rjmccall
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D31006
llvm-svn: 299014
Sanjay Patel [Wed, 29 Mar 2017 18:08:01 +0000 (18:08 +0000)]
[DAGCombiner] unify type checks and add asserts; NFCI
We had a mix of type checks and usage that wasn't very clear.
llvm-svn: 299013
Graydon Hoare [Wed, 29 Mar 2017 17:58:41 +0000 (17:58 +0000)]
Unbreak windows bot.
llvm-svn: 299012
Brian Kelley [Wed, 29 Mar 2017 17:55:11 +0000 (17:55 +0000)]
[Objective-C] Fix "repeated use of weak" warning with -fobjc-weak
Summary: -Warc-repeated-use-of-weak should produce the same warnings with -fobjc-weak as it does with -objc-arc. Also check for ObjCWeak along with ObjCAutoRefCount when recording the use of an evaluated weak variable. Add a -fobjc-weak run to the existing arc-repeated-weak test case and adapt it slightly to work in both modes.
Reviewers: rsmith, doug.gregor, jordan_rose, rjmccall
Reviewed By: rjmccall
Subscribers: arphaman, rjmccall, cfe-commits
Differential Revision: https://reviews.llvm.org/D31005
llvm-svn: 299011
Brian Kelley [Wed, 29 Mar 2017 17:40:35 +0000 (17:40 +0000)]
[Objective-C] Fix __weak type traits with -fobjc-weak
Summary: Similar to ARC, in ObjCWeak Objective-C object pointers qualified with a weak lifetime are not POD or trivial types. Update the type trait code to reflect this. Copy and adapt the arc-type-traits.mm test case to verify correctness.
Reviewers: rsmith, doug.gregor, rjmccall
Reviewed By: rjmccall
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D31004
llvm-svn: 299010
Graydon Hoare [Wed, 29 Mar 2017 17:33:09 +0000 (17:33 +0000)]
[PCH] Attach instance's dependency collectors to PCH external AST sources.
Summary:
When a PCH is included via -include-pch, clang should treat the
current TU as dependent on the sourcefile that the PCH was generated from.
This is currently _partly_ accomplished by InitializePreprocessor calling
AddImplicitIncludePCH to synthesize an implicit #include of the sourcefile,
into the preprocessor's Predefines buffer.
For FrontendActions such as PreprocessOnlyAction (which is, curiously, what the
driver winds up running one of in response to a plain clang -M) this is
sufficient: the preprocessor cranks over its Predefines and emits a dependency
reference to the initial sourcefile.
For other FrontendActions (for example -emit-obj or -fsyntax-only) the
Predefines buffer is reset to the suggested predefines buffer from the PCH, so
the dependency edge is lost. The result is that clang emits a .d file in those
cases that lacks a reference to the .h file responsible for the input (and in
Swift's case, our .swiftdeps file winds up not including a reference to the
source file for a PCH bridging header.)
This patch fixes the problem by taking a different tack: ignoring the
Predefines buffer (which seems a bit like a hack anyways) and directly
attaching the CompilerInstance's DependencyCollectors (and legacy
DependencyFileGenerator) to the ASTReader for the external AST.
This approach is similar to the one chosen in earlier consultation with Bruno
and Ben, and I think it's the least-bad solution, given several options.
Reviewers: bruno, benlangmuir, doug.gregor
Reviewed By: bruno, doug.gregor
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D31378
llvm-svn: 299009
Brian Kelley [Wed, 29 Mar 2017 17:31:42 +0000 (17:31 +0000)]
[Objective-C] C++ Classes with __weak Members non-POD Types when using -fobjc-weak
Summary: When adding an Objective-C retainable type member to a C++ class, also check the LangOpts.ObjCWeak flag and the lifetime qualifier so __weak qualified Objective-C pointer members cause the class to be a non-POD type with non-trivial special members, so the compiler always emits the necessary runtime calls for copying, moving, and destroying the weak member. Otherwise, Objective-C++ classes with weak Objective-C pointer members compiled with -fobjc-weak exhibit undefined behavior if the C++ class is classified as a POD type.
Reviewers: rsmith, benlangmuir, doug.gregor, rjmccall
Reviewed By: rjmccall
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D31003
llvm-svn: 299008
Brian Kelley [Wed, 29 Mar 2017 17:18:05 +0000 (17:18 +0000)]
Test Commit
Remove trailing whitespace.
llvm-svn: 299007
Adrian McCarthy [Wed, 29 Mar 2017 17:11:27 +0000 (17:11 +0000)]
llvm-pdbdump: If we don't change the color, don't reset the color.
The -output-color option was successful at suppressing color changes, but
was still allowing color resets.
Differential Revision: https://reviews.llvm.org/D31468
llvm-svn: 299006
Weiming Zhao [Wed, 29 Mar 2017 16:59:09 +0000 (16:59 +0000)]
[Builtins] Mark mulsc3_test as UNSUPPORTED for armhf; NFC
The same test fails on clang-cmake-armv7-a15-full biuld bot
but passes the clang-cmake-thumbv7-a15-full.
For now, we mark it as UNSUPPORTED for armhf target.
Bug 32457 tracks it.
llvm-svn: 299005
Simon Pilgrim [Wed, 29 Mar 2017 15:41:58 +0000 (15:41 +0000)]
[X86] Tidied up comment - we don't custom lower add/sub i64 on i686 anymore. NFCI.
llvm-svn: 299004
Sanjay Patel [Wed, 29 Mar 2017 15:37:33 +0000 (15:37 +0000)]
[DAGCombiner] reduce code duplication by rearranging checks; NFCI
llvm-svn: 299002
Daniel Sanders [Wed, 29 Mar 2017 15:37:18 +0000 (15:37 +0000)]
[tablegen][globalisel] Convert the SelectionDAG importer to a tree walking approach. NFC
Summary:
But don't actually inspect the tree any deeper than we already do. This
change is NFC but the next one will enable full traversal of the
source/destination patterns.
Depends on D30535
Reviewers: t.p.northover, qcolombet, aditya_nandakumar, rovka, ab
Subscribers: igorb, dberris, llvm-commits, kristof.beyls
Differential Revision: https://reviews.llvm.org/D30536
llvm-svn: 299001
Simon Pilgrim [Wed, 29 Mar 2017 15:27:24 +0000 (15:27 +0000)]
Spelling mistakes in comments. NFCI.
llvm-svn: 299000
Sven van Haastregt [Wed, 29 Mar 2017 15:25:06 +0000 (15:25 +0000)]
[MachineVerifier] Drop a spurious const
As of r298987 the argument is a value that we std::move, so it
shouldn't be const anymore.
llvm-svn: 298999
George Rimar [Wed, 29 Mar 2017 15:23:28 +0000 (15:23 +0000)]
[ELF] - Prepare GnuHashTableSection<ELFT> for detemplation.
This patch changes GnuHashTableSection implementation
to avoid depend on uintX_t and other ELFT stuff, reducing amount of
changes for following patch(es).
Differential revision: https://reviews.llvm.org/D31184
llvm-svn: 298998
Weiming Zhao [Wed, 29 Mar 2017 14:55:51 +0000 (14:55 +0000)]
[Compiler-RT][Builtins] Remove XFAIL for mulsc3; NFC
It was XFAILed in r298974. However, the problem was not exposed on
the buildbot because hardfp flag was not passed during the test.
We can fix the CMAKE to pass the same flag as building the lib to the
RUN line to see if the problem is still there. For now, we remove the
XFAIL.
llvm-svn: 298997
Filipe Cabecinhas [Wed, 29 Mar 2017 14:42:27 +0000 (14:42 +0000)]
Cleanup in preparation for D30703. NFCI
Make the enumerators follow the coding convention and start with OW_...
llvm-svn: 298996
Rafael Espindola [Wed, 29 Mar 2017 14:27:38 +0000 (14:27 +0000)]
Fix GettingStarted.rst statement.
Patch by Wei-Ren Chen!
llvm-svn: 298995
Rafael Espindola [Wed, 29 Mar 2017 14:20:38 +0000 (14:20 +0000)]
Remove unused argument.
llvm-svn: 298994
Simon Pilgrim [Wed, 29 Mar 2017 13:00:00 +0000 (13:00 +0000)]
[X86][AVX2] Prevent unary interleaving patterns from calling lowerVectorShuffleAsSplitOrBlend (PR32453)
llvm-svn: 298993
Egor Churaev [Wed, 29 Mar 2017 12:09:39 +0000 (12:09 +0000)]
Reapplied r298976 [OpenCL] Added parsing for OpenCL vector types.
llvm-svn: 298992
Simon Pilgrim [Wed, 29 Mar 2017 12:09:25 +0000 (12:09 +0000)]
[AMDGPU] Tidy up computeKnownBitsForTargetNode/ComputeNumSignBitsForTargetNode arguments. NFCI.
Based on comment in D31249.
llvm-svn: 298991
Simon Pilgrim [Wed, 29 Mar 2017 10:47:18 +0000 (10:47 +0000)]
[X86][MMX] Added generic sitofp test to compare against existing cvtdq2ps test.
llvm-svn: 298989
Simon Pilgrim [Wed, 29 Mar 2017 10:44:51 +0000 (10:44 +0000)]
[X86] Removed old comment. NFCI.
No longer makes sense as the previous opcode mnemonic it was referring to is long gone.
llvm-svn: 298988
Sven van Haastregt [Wed, 29 Mar 2017 09:08:25 +0000 (09:08 +0000)]
[MachineVerifier] Avoid reference to nullptr
Instantiation of the MachineVerifierPass through
PassInfo::getNormalCtor would yield a segfault since the default
constructor of the MachineVerifierPass takes a reference to nullptr.
Patch by Simone Pellegrini.
Differential Revision: https://reviews.llvm.org/D31387
llvm-svn: 298987
Eric Christopher [Wed, 29 Mar 2017 07:40:44 +0000 (07:40 +0000)]
Move the x86 cpu feature rtm from Haswell to Skylake matching clang commit r298956.
llvm-svn: 298986
Craig Topper [Wed, 29 Mar 2017 07:31:56 +0000 (07:31 +0000)]
[AVX-512] Remove explicit KMOVWrk from isel patterns. COPY_TO_REGCLASS to GR32 is enough.
llvm-svn: 298985
Craig Topper [Wed, 29 Mar 2017 06:55:28 +0000 (06:55 +0000)]
[AVX-512] Remove explicit KMOVWrk/KMOVWKr instructions from patterns where we can just use COPY_TO_REGCLASS instead.
This will result in a KMOVW or KMOVD being emitted during register allocation. And in at least some cases this might allow the register coalescer to remove the copy all together.
llvm-svn: 298984
Dean Michael Berris [Wed, 29 Mar 2017 06:10:12 +0000 (06:10 +0000)]
[XRay] Update FDR log reader to be aware of buffer sizes per thread.
Summary:
It is problematic for this reader that it expects to read data from
several threads, but the header or message format does not define
framing. Since the buffers are reused, we can't rely on skipping
zeroed out data as a synchronization method either.
There is an argument that this is not version compatible with the format
the reader expected previously. I argue that since the writer wrote garbage
past the end of buffer record, there is no currently working reader to
compromise.
The corresponding writer change is posted to D31384.
Reviewers: dberris, pelikan
Reviewed By: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31385
llvm-svn: 298983
Dean Michael Berris [Wed, 29 Mar 2017 05:56:37 +0000 (05:56 +0000)]
[XRay] [compiler-rt] Write buffer length to FDR log before writing buffer.
Summary:
Currently the FDR log writer, upon flushing, dumps a sequence of buffers from
its freelist to disk. A reader can read the first buffer up to an EOB record,
but then it is unclear how far ahead to scan to find the next threads traces.
There are a few ways to handle this problem.
1. The reader has externalized knowledge of the buffer size.
2. The size of buffers is in the file header or otherwise encoded in the log.
3. Only write out the portion of the buffer with records. When released, the
buffers are marked with a size.
4. The reader looks for memory that matches a pattern and synchronizes on it.
2 and 3 seem the most flexible and 2 does not rule 3 out.
This is an implementation of 2.
In addition, the function handler for fdr more aggressively checks for
finalization and makes an attempt to release its buffer.
Reviewers: pelikan, dberris
Reviewed By: dberris
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31384
llvm-svn: 298982
Egor Churaev [Wed, 29 Mar 2017 05:40:45 +0000 (05:40 +0000)]
Reverted r298976 [OpenCL] Added parsing for OpenCL vector types.
llvm-svn: 298978
Dean Michael Berris [Wed, 29 Mar 2017 05:19:24 +0000 (05:19 +0000)]
[XRay][compiler-rt] Add an end-to-end test for FDR Logging
Summary:
This change exercises the end-to-end functionality defined in the FDR
logging implementation. We also prepare for being able to run traces
generated by the FDR logging implementation from being analysed with the
llvm-xray command that comes with the LLVM distribution.
This also unblocks D31385, D31384, and D31345.
Reviewers: kpw, pelikan
Subscribers: llvm-commits, mgorny
Differential Revision: https://reviews.llvm.org/D31452
llvm-svn: 298977
Egor Churaev [Wed, 29 Mar 2017 05:08:18 +0000 (05:08 +0000)]
[OpenCL] Added parsing for OpenCL vector types.
Reviewers: cfe-commits, Anastasia
Reviewed By: Anastasia
Subscribers: yaxunl, bader
Differential Revision: https://reviews.llvm.org/D31183
llvm-svn: 298976
Dean Michael Berris [Wed, 29 Mar 2017 04:55:45 +0000 (04:55 +0000)]
[XRay][tools] Handle "no subcommand" case for llvm-xray
Summary:
Currently the llvm-xray commandline tool fails to handle the case for
when no subcommand is provided in a graceful manner. This fixes that to
print the help message explaining the subcommands and the available
options.
Reviewers: pcc, pelikan
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31409
llvm-svn: 298975
Weiming Zhao [Wed, 29 Mar 2017 03:36:46 +0000 (03:36 +0000)]
[Builtin] Unxfail tests for armhf
Summary:
Originally, a few tests fail for armhf target due to:
1) COMPILER_RT_ARMHF_TARGET was not set when building the lib
2) COMPILER_RT_ABI should not be defined as `__attribute__((pcs("aapcs")))` for armhf when building for both lib and tests
This address https://bugs.llvm.org//show_bug.cgi?id=32261
mulsc3_test.c is a newly exposed issue, which will be addressed separately.
Reviewers: rengolin, compnerd
Reviewed By: compnerd
Subscribers: aemerson, llvm-commits, mgorny
Differential Revision: https://reviews.llvm.org/D31448
llvm-svn: 298974
Rui Ueyama [Wed, 29 Mar 2017 01:36:37 +0000 (01:36 +0000)]
Attempt to fix buildbot.
llvm-svn: 298971
Kamil Rytarowski [Wed, 29 Mar 2017 01:10:21 +0000 (01:10 +0000)]
Remove dead include <sys/user.h> from the NetBSD code.
llvm-svn: 298970
Rui Ueyama [Wed, 29 Mar 2017 00:49:50 +0000 (00:49 +0000)]
Use uint64_t instead of uintX_t and size_t.
uint64_t is simpler and less error-prone than target or host-dependent types.
llvm-svn: 298969
Rui Ueyama [Wed, 29 Mar 2017 00:49:29 +0000 (00:49 +0000)]
Change the order of parameters. NFC.
If a function takes a size and an alignment, we usually pass them
in that order instead of the reverse order.
llvm-svn: 298968
Juergen Ributzka [Wed, 29 Mar 2017 00:24:36 +0000 (00:24 +0000)]
Revert "Added `applyAtomicChanges` function."
This broke GreenDragon:
http://lab.llvm.org:8080/green/job/clang-stage2-cmake-modulesRDA_build/4776/
Reverting this commit and all follow-up commits.
llvm-svn: 298967
Juergen Ributzka [Wed, 29 Mar 2017 00:24:34 +0000 (00:24 +0000)]
Revert "Try to unbreak buildbots after r298913."
llvm-svn: 298966
Juergen Ributzka [Wed, 29 Mar 2017 00:24:32 +0000 (00:24 +0000)]
Revert "clangToolingRefactor: Add libdeps upon clangFormat in r298913."
llvm-svn: 298965
Rui Ueyama [Wed, 29 Mar 2017 00:09:40 +0000 (00:09 +0000)]
Remove a comment that is taken from gdb man page.
This is not an original comment but taken from gdb man page, so
a pointer to the original document should suffice.
llvm-svn: 298964
Adam Nemet [Tue, 28 Mar 2017 23:46:14 +0000 (23:46 +0000)]
[SDAG] Remove -enable-fmf-dag
This is no longer needed as spotted by Sanjay in
https://reviews.llvm.org/D31165.
llvm-svn: 298963
Adam Nemet [Tue, 28 Mar 2017 23:46:12 +0000 (23:46 +0000)]
[SDAG] Handle VectorReduction in SDNodeFlags::intersectWith
Spotted by Sanjay in https://reviews.llvm.org/D31165
llvm-svn: 298962
Adam Nemet [Tue, 28 Mar 2017 23:46:08 +0000 (23:46 +0000)]
[SDAG] Add AllowContract to SNodeFlags
Properly propagate the FMF from the LLVM IR to this flag.
This is toward moving fp-contraction=fast from an LLVM TargetOption to a
FastMathFlag in order to fix PR25721.
Differential Revision: https://reviews.llvm.org/D31165
llvm-svn: 298961
Peter Collingbourne [Tue, 28 Mar 2017 23:35:34 +0000 (23:35 +0000)]
More accurate header inclusions. NFC.
llvm-svn: 298960
Craig Topper [Tue, 28 Mar 2017 23:29:35 +0000 (23:29 +0000)]
[AVX-512] Add test case that was supposed to go with r298957.
llvm-svn: 298959
Jim Ingham [Tue, 28 Mar 2017 23:25:34 +0000 (23:25 +0000)]
Print the error if dsymForUUID sometimes produces bad plists.
Not much we can do about it but at least we can print the bad
plist and the error.
llvm-svn: 298958
Craig Topper [Tue, 28 Mar 2017 23:20:37 +0000 (23:20 +0000)]
[AVX-512] Punt on fast-isel of truncates to i1 when AVX512 is enabled.
We should be masking the value and emitting a register copy like we do in non-fast isel. Instead we were just updating the value map and emitting nothing.
After r298928 we started seeing cases where we would create a copy from GR8 to GR32 because the source register in a VK1 to GR32 copy was replaced by the GR8 going into a truncate.
This fixes PR32451.
llvm-svn: 298957
Eric Christopher [Tue, 28 Mar 2017 23:03:19 +0000 (23:03 +0000)]
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
Guozhi Wei [Tue, 28 Mar 2017 22:55:01 +0000 (22:55 +0000)]
[PPC] In PPCBoolRetToInt change the bool value to i64 if the target is ppc64
In PPCBoolRetToInt bool value is changed to i32 type. On ppc64 it may introduce an extra zero extension for the return value. This patch changes the integer type to i64 to avoid the zero extension on ppc64.
This patch fixed PR32442.
Differential Revision: https://reviews.llvm.org/D31407
llvm-svn: 298955
Sanjay Patel [Tue, 28 Mar 2017 22:45:53 +0000 (22:45 +0000)]
[DAGCombiner] reduce code duplication with local variables; NFCI
llvm-svn: 298954
Kamil Rytarowski [Tue, 28 Mar 2017 22:43:17 +0000 (22:43 +0000)]
Add support for tracing hello-world application on NetBSD
Summary:
This patch is a stripped down from features a NetBSD process
code (patch is kept under 2k LOC). This code has assumption that
there is only one thread within a debugged process. The only
debugger trap supported is software breakpoint (TRAP_BRKPT).
The generic platform code requires to add dummy function for
watchpoints etc. These functions are currently empty.
This code is not the final platform support as is and it's treated as
a base to extend, refactor and address issues afterwards.
Supported features:
- handle software breakpoints,
- correctly attach to a tracee,
- support NetBSD specific ptrace(2),
- monitor process termination,
- monitor SIGTRAP events,
- monitor SIGSTOP events,
- monitor other signals events,
- resume the whole process,
- get memory region info perms,
- read memory from tracee,
- write memory to tracee,
- read ELF AUXV,
- x86_64 GPR read and write code
For the generic framework include:
- halt,
- detach,
- signal,
- kill,
- allocatememory,
- deallocatememory,
- update threads,
- getarchitecture,
- getfileloadaddress,
- and others.
This code has preliminary AddThread code.
Out of interest in this patch:
- exec() traps,
- hardware debug register traps,
- single step trap,
- thread creation/termination trap,
- process fork(2), vfork(2) and vfork(2) done traps,
- syscall entry and exit trap,
- threads,
- FPR registers,
- retrieving tracee's thread name,
- non x86_64 support.
This code can be used to start a hello world application and trace it.
This code can be used by other BSD systems as a starting point to get similar
capabilities.
Sponsored by <The NetBSD Foundation>
Reviewers: emaste, joerg, kettenis, labath
Subscribers: mgorny, #lldb
Tags: #lldb
Differential Revision: https://reviews.llvm.org/D31374
llvm-svn: 298953
Eric Christopher [Tue, 28 Mar 2017 22:37:43 +0000 (22:37 +0000)]
Add a similar test for tailcall optimization as in r270287 for aarch64.
llvm-svn: 298952
Peter Collingbourne [Tue, 28 Mar 2017 22:31:35 +0000 (22:31 +0000)]
LTO: Replace InputFile::Symbol::getFlags() with predicate accessors. NFC.
This makes the predicates independent of the flag representation
and makes the code a little easier to read.
llvm-svn: 298951
Sanjay Patel [Tue, 28 Mar 2017 22:25:25 +0000 (22:25 +0000)]
[DAG] fix formatting; NFC
llvm-svn: 298950
Sanjay Patel [Tue, 28 Mar 2017 22:22:50 +0000 (22:22 +0000)]
[DAGCombiner] remove redundant conditions and duplicated code; NFCI
llvm-svn: 298949
Stanislav Mekhanoshin [Tue, 28 Mar 2017 22:13:51 +0000 (22:13 +0000)]
[AMDGPU] Boost unroll threshold for loops reading local memory
This is less important than increase threshold for private memory,
but still brings performance improvements in a wide range of tests.
Unrolling more for local memory serves three purposes: it allows
to combine ds operations if offset becomes static, saves registers
used for offsets in case of static offsets, and allows better lds
latency hiding.
Differential Revision: https://reviews.llvm.org/D31412
llvm-svn: 298948
Francis Ricci [Tue, 28 Mar 2017 21:56:45 +0000 (21:56 +0000)]
Postpone lsan tls allocation until required
Summary:
This prevents InternalAlloc from being called before the sanitizers
are fully initialized.
Reviewers: kubamracek, kcc
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31306
llvm-svn: 298947
Francis Ricci [Tue, 28 Mar 2017 21:56:44 +0000 (21:56 +0000)]
Enable i386 builds for darwin lsan
Summary:
Now that __thread is no longer used for lsan on darwin, i386 builds
can be enabled.
Reviewers: kcc, kubamracek
Subscribers: danalbert, srhines, mgorny, llvm-commits
Differential Revision: https://reviews.llvm.org/D29995
llvm-svn: 298946
Stanislav Mekhanoshin [Tue, 28 Mar 2017 21:48:54 +0000 (21:48 +0000)]
[AMDGPU] Fix recorded region boundaries in max-occupancy scheduler
This is incorrect to record region boundaries before scheduling,
it may change after scheduling. As a result second pass may see less
instructions to schedule than it should.
Differential Revision: https://reviews.llvm.org/D31434
llvm-svn: 298945
Sanjay Patel [Tue, 28 Mar 2017 21:40:41 +0000 (21:40 +0000)]
[DAGCombiner] rename variables in foldAndOfSetCCs for easier reading; NFCI
llvm-svn: 298944
Simon Pilgrim [Tue, 28 Mar 2017 21:32:11 +0000 (21:32 +0000)]
[X86][MMX] Match MMX fp_to_sint conversions from XMM registers
We currently perform the various fp_to_sint XMM conversion and then transfer to the MMX register (on 32-bit via the stack).
This patch improves support for MOVDQ2Q XMM to MMX transfers and adds the XMM->MMX fp_to_sint direct conversion patterns. The SSE2 specifications are the same as for XMM->XMM and XMM->MMX rounding/exceptions/etc.
Differential Revision: https://reviews.llvm.org/D30868
llvm-svn: 298943
Bob Haarman [Tue, 28 Mar 2017 21:20:06 +0000 (21:20 +0000)]
COFF: include archive name in LTO object name
Summary: In the ELF linker, we create the buffer identifier for bitcode files by appending the object name to the archive name. This change makes the COFF linker do the same. Without the change, ThinLTO builds can fail with an error message about multiple ThinLTO modules per object file, caused by object files contained in different archives having the same name.
Reviewers: pcc, ruiu
Reviewed By: pcc
Subscribers: mehdi_amini
Differential Revision: https://reviews.llvm.org/D31402
llvm-svn: 298942
Matt Arsenault [Tue, 28 Mar 2017 20:33:12 +0000 (20:33 +0000)]
Fix crashing on TargetCustom PseudoSourceValues
Default to something more reasonable if printCustom isn't implemented.
llvm-svn: 298941
Sanjay Patel [Tue, 28 Mar 2017 20:28:16 +0000 (20:28 +0000)]
[DAGCombiner] clean up foldAndOfSetCCs; NFCI
1. Fix bogus comment.
2. Early exit to reduce indent.
3. Change node pointer param to what it really is: an SDLoc.
llvm-svn: 298940
Adam Nemet [Tue, 28 Mar 2017 20:11:52 +0000 (20:11 +0000)]
[IR] Add AllowContract to FastMathFlags
-ffp-contract=fast does not currently work with LTO because it's passed as a
TargetOption to the backend rather than in the IR. This adds it to
FastMathFlags.
This is toward fixing PR25721
Differential Revision: https://reviews.llvm.org/D31164
llvm-svn: 298939
Sanjay Patel [Tue, 28 Mar 2017 19:58:46 +0000 (19:58 +0000)]
[DAGCombiner] add helper function for and-of-setcc folds; NFC
This is just a cut and paste followed by clang-format. Clean up to follow.
llvm-svn: 298938
Shoaib Meenai [Tue, 28 Mar 2017 19:33:31 +0000 (19:33 +0000)]
[libc++] Add a key function for bad_function_call
Summary:
bad_function_call is currently an empty class, so any object files using
that class will end up with their own copy of its typeinfo, typeinfo
name and vtable, leading to unnecessary duplication that has to be
resolved by the dynamic linker. Instead, give bad_function_call a key
function and put a definition for that key function in libc++ itself, to
centralize the typeinfo and vtable.
This is consistent with the behavior for other exception classes. The
key functions are defined in libc++ rather than libc++abi since the
class is defined in the libc++ versioning namespace, so ABI
compatibility with libstdc++ is not a concern.
Guard this change behind an ABI macro, since it isn't backwards
compatible (i.e., clients built against the new libc++ headers wouldn't
be able to run against an older libc++ library).
Reviewers: mclow.lists, EricWF
Subscribers: mgorny, cfe-commits
Differential Revision: https://reviews.llvm.org/D27387
llvm-svn: 298937
Mehdi Amini [Tue, 28 Mar 2017 18:55:44 +0000 (18:55 +0000)]
Add support for -fno-builtin to LTO and ThinLTO to libLTO
Reviewers: tejohnson, pcc
Subscribers: Prazek, dexonsmith, llvm-commits
Differential Revision: https://reviews.llvm.org/D30791
llvm-svn: 298936
Stanislav Mekhanoshin [Tue, 28 Mar 2017 18:23:24 +0000 (18:23 +0000)]
[AMDGPU] Split -amdgpu-early-inline-all option
Previously it was covered by the internalization. It turns out we cannot
run internalizer in FE, it break separate compilation tests. Thus early
inliner gets its own option.
Differential Revision: https://reviews.llvm.org/D31429
llvm-svn: 298935
Rui Ueyama [Tue, 28 Mar 2017 18:11:52 +0000 (18:11 +0000)]
Do not set entsize for .gnu.hash.
.gnu.hash happen to contain only 32-bit integers for 32-bit arch,
but the section contents are not uniform array members, so setting
entsize doesn't make much sense. This behavior seems to have been
blindly copied from GNU linkers.
llvm-svn: 298934
Sanjay Patel [Tue, 28 Mar 2017 17:23:49 +0000 (17:23 +0000)]
[x86] use VPMOVMSK to replace memcmp libcalls for 32-byte equality
Follow-up to:
https://reviews.llvm.org/rL298775
llvm-svn: 298933
Weiming Zhao [Tue, 28 Mar 2017 17:15:11 +0000 (17:15 +0000)]
Revert "Dont emit Mapping symbols for sections that contain only data."
It breaks some lld tests.
This reverts commit
3a50eea6d9732ab40e9a7aebe6be777b53a8b35c.
llvm-svn: 298932
Nirav Dave [Tue, 28 Mar 2017 17:09:49 +0000 (17:09 +0000)]
[SDAG] Deal with deleted node in PromoteIntShiftOp
Deal with case that initial node is deleted during dag-combine leading
to an assertional failure in promoteIntShiftOp.
Fixes PR32420.
Reviewers: spatel, RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D31403
llvm-svn: 298931
Zvi Rackover [Tue, 28 Mar 2017 16:45:23 +0000 (16:45 +0000)]
Add reproducer test for pr32449. NFC.
llvm-svn: 298930
Simon Pilgrim [Tue, 28 Mar 2017 16:40:38 +0000 (16:40 +0000)]
[X86][AVX2] Add support for combining v16i16 shuffles to VPBLENDW
llvm-svn: 298929
Craig Topper [Tue, 28 Mar 2017 16:35:29 +0000 (16:35 +0000)]
[AVX-512] Fix accidental uses of AH/BH/CH/DH after copies to/from mask registers
We've had several bugs(PR32256, PR32241) recently that resulted from usages of AH/BH/CH/DH either before or after a copy to/from a mask register.
This ultimately occurs because we create COPY_TO_REGCLASS with VK1 and GR8. Then in CopyToFromAsymmetricReg in X86InstrInfo we find a 32-bit super register for the GR8 to emit the KMOV with. But as these tests are demonstrating, its possible for the GR8 register to be a high register and we end up doing an accidental extra or insert from bits 15:8.
I think the best way forward is to stop making copies directly between mask registers and GR8/GR16. Instead I think we should restrict to only copies between mask registers and GR32/GR64 and use EXTRACT_SUBREG/INSERT_SUBREG to handle the conversion from GR32 to GR16/8 or vice versa.
Unfortunately, this complicates fastisel a bit more now to create the subreg extracts where we used to create GR8 copies. We can probably make a helper function to bring down the repitition.
This does result in KMOVD being used for copies when BWI is available because we don't know the original mask register size. This caused a lot of deltas on tests because we have to split the checks for KMOVD vs KMOVW based on BWI.
Differential Revision: https://reviews.llvm.org/D30968
llvm-svn: 298928