Vitaly Buka [Tue, 30 Aug 2016 00:57:40 +0000 (00:57 +0000)]
[asan] Disable test on darwin bot
According logs asan detects the bug but string with file name is not found.
I will investigate and fix the test.
llvm-svn: 280038
Eric Fiselier [Tue, 30 Aug 2016 00:54:37 +0000 (00:54 +0000)]
Fix PR28831 - Bad logic around CMAKE_OSX_SYSROOT and CMAKE_OSX_ARCHITECTURES.
See https://llvm.org/bugs/show_bug.cgi?id=28831 for more information about
the changes.
llvm-svn: 280037
Piotr Padlewski [Tue, 30 Aug 2016 00:46:26 +0000 (00:46 +0000)]
NFC: add early exit in ModuleSummaryAnalysis
Summary:
Changed this code because it was not very readable.
The one question that I got after changing it is, should we
count calls to intrinsics? We don't add them to caller summary,
so maybe we shouldn't also count them?
Reviewers: tejohnson, eraman, mehdi_amini
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23949
llvm-svn: 280036
Richard Smith [Tue, 30 Aug 2016 00:44:54 +0000 (00:44 +0000)]
C++ Modules TS: driver support for building modules.
This works as follows: we add --precompile to the existing gamut of options for
specifying how far to go when compiling an input (-E, -c, -S, etc.). This flag
specifies that an input is taken to the precompilation step and no further, and
this can be specified when building a .pcm from a module interface or when
building a .pch from a header file.
The .cppm extension (and some related extensions) are implicitly recognized as
C++ module interface files. If --precompile is /not/ specified, the file is
compiled (via a .pcm) to a .o file containing the code for the module (and then
potentially also assembled and linked, if -S, -c, etc. are not specified). We
do not yet suppress the emission of object code for other users of the module
interface, so for now this will only work if everything in the .cppm file has
vague linkage.
As with the existing support for module-map modules, prebuilt modules can be
provided as compiler inputs either via the -fmodule-file= command-line argument
or via files named ModuleName.pcm in one of the directories specified via
-fprebuilt-module-path=.
This also exposes the -fmodules-ts cc1 flag in the driver. This is still
experimental, and in particular, the concrete syntax is subject to change as
the Modules TS evolves in the C++ committee. Unlike -fmodules, this flag does
not enable support for implicitly loading module maps nor building modules via
the module cache, but those features can be turned on separately and used in
conjunction with the Modules TS support.
llvm-svn: 280035
Hal Finkel [Tue, 30 Aug 2016 00:22:22 +0000 (00:22 +0000)]
[PowerPC] Add triple to test/CodeGen/PowerPC/atomic-2.ll for ppc64le
Otherwise, running the test on Darwin systems will not work.
llvm-svn: 280034
Duncan P. N. Exon Smith [Tue, 30 Aug 2016 00:18:43 +0000 (00:18 +0000)]
Rename unittests/ADT/ilistTestTemp.cpp => IListTest.cpp
And rename the tests inside from ilistTest to IListTest. This makes the
file sort properly in the CMakeLists.txt (previously, sorting would
throw it down to the end of the list) and is consistent with the tests
I've added more recently.
Why use IListNodeBaseTest.cpp (and a test name of IListNodeBaseTest)?
- ilist_node_base_test is the obvious thing, since this is testing
ilist_node_base. However, gtest disallows underscores in test names.
- ilist_node_baseTest fails for the same reason.
- ilistNodeBaseTest is weird, because it isn't in our usual
TitleCaseTest form that we use for tests, and it also doesn't have the
name of the tested class in it.
- IlistNodeBaseTest matches TitleCaseTest, but "Ilist" is hard to read,
and really "ilist" is an abbreviation for "IntrusiveList" so the
lowercase "list" is strange.
- That left IListNodeBaseTest.
Note: I made this move in two stages, with a temporary filename of
ilistTestTemp in between in r279524. This was in the hopes of avoiding
problems on Git and SVN clients on case-insensitive filesystems,
particularly on buildbots with incremental checkouts.
llvm-svn: 280033
Duncan P. N. Exon Smith [Tue, 30 Aug 2016 00:13:12 +0000 (00:13 +0000)]
ADT: Give ilist<T>::reverse_iterator a handle to the current node
Reverse iterators to doubly-linked lists can be simpler (and cheaper)
than std::reverse_iterator. Make it so.
In particular, change ilist<T>::reverse_iterator so that it is *never*
invalidated unless the node it references is deleted. This matches the
guarantees of ilist<T>::iterator.
(Note: MachineBasicBlock::iterator is *not* an ilist iterator, but a
MachineInstrBundleIterator<MachineInstr>. This commit does not change
MachineBasicBlock::reverse_iterator, but it does update
MachineBasicBlock::reverse_instr_iterator. See note at end of commit
message for details on bundle iterators.)
Given the list (with the Sentinel showing twice for simplicity):
[Sentinel] <-> A <-> B <-> [Sentinel]
the following is now true:
1. begin() represents A.
2. begin() holds the pointer for A.
3. end() represents [Sentinel].
4. end() holds the poitner for [Sentinel].
5. rbegin() represents B.
6. rbegin() holds the pointer for B.
7. rend() represents [Sentinel].
8. rend() holds the pointer for [Sentinel].
The changes are #6 and #8. Here are some properties from the old
scheme (which used std::reverse_iterator):
- rbegin() held the pointer for [Sentinel] and rend() held the pointer
for A;
- operator*() cost two dereferences instead of one;
- converting from a valid iterator to its valid reverse_iterator
involved a confusing increment; and
- "RI++->erase()" left RI invalid. The unintuitive replacement was
"RI->erase(), RE = end()".
With vector-like data structures these properties are hard to avoid
(since past-the-beginning is not a valid pointer), and don't impose a
real cost (since there's still only one dereference, and all iterators
are invalidated on erase). But with lists, this was a poor design.
Specifically, the following code (which obviously works with normal
iterators) now works with ilist::reverse_iterator as well:
for (auto RI = L.rbegin(), RE = L.rend(); RI != RE;)
fooThatMightRemoveArgFromList(*RI++);
Converting between iterator and reverse_iterator for the same node uses
the getReverse() function.
reverse_iterator iterator::getReverse();
iterator reverse_iterator::getReverse();
Why doesn't iterator <=> reverse_iterator conversion use constructors?
In order to catch and update old code, reverse_iterator does not even
have an explicit conversion from iterator. It wouldn't be safe because
there would be no reasonable way to catch all the bugs from the changed
semantic (see the changes at call sites that are part of this patch).
Old code used this API:
std::reverse_iterator::reverse_iterator(iterator);
iterator std::reverse_iterator::base();
Here's how to update from old code to new (that incorporates the
semantic change), assuming I is an ilist<>::iterator and RI is an
ilist<>::reverse_iterator:
[Old] ==> [New]
reverse_iterator(I) (--I).getReverse()
reverse_iterator(I) ++I.getReverse()
--reverse_iterator(I) I.getReverse()
reverse_iterator(++I) I.getReverse()
RI.base() (--RI).getReverse()
RI.base() ++RI.getReverse()
--RI.base() RI.getReverse()
(++RI).base() RI.getReverse()
delete &*RI, RE = end() delete &*RI++
RI->erase(), RE = end() RI++->erase()
=======================================
Note: bundle iterators are out of scope
=======================================
MachineBasicBlock::iterator, also known as
MachineInstrBundleIterator<MachineInstr>, is a wrapper to represent
MachineInstr bundles. The idea is that each operator++ takes you to the
beginning of the next bundle. Implementing a sane reverse iterator for
this is harder than ilist. Here are the options:
- Use std::reverse_iterator<MBB::i>. Store a handle to the beginning of
the next bundle. A call to operator*() runs a loop (usually
operator--() will be called 1 time, for unbundled instructions).
Increment/decrement just works. This is the status quo.
- Store a handle to the final node in the bundle. A call to operator*()
still runs a loop, but it iterates one time fewer (usually
operator--() will be called 0 times, for unbundled instructions).
Increment/decrement just works.
- Make the ilist_sentinel<MachineInstr> *always* store that it's the
sentinel (instead of just in asserts mode). Then the bundle iterator
can sniff the sentinel bit in operator++().
I initially tried implementing the end() option as part of this commit,
but updating iterator/reverse_iterator conversion call sites was
error-prone. I have a WIP series of patches that implements the final
option.
llvm-svn: 280032
Evgeniy Stepanov [Mon, 29 Aug 2016 23:42:34 +0000 (23:42 +0000)]
[cfi] Export __cfi_check when linking with -fsanitize-cfi-cross-dso.
Multi-DSO CFI model requires every DSO to export a __cfi_check function.
llvm-svn: 280031
Jan Vesely [Mon, 29 Aug 2016 23:21:46 +0000 (23:21 +0000)]
AMDGPU/R600: Cleanup DAGCombine
Move SDLoc initialization to comon place.
fall back to AMDGPU version in one place
Differential Revision: https://reviews.llvm.org/D23900
llvm-svn: 280030
Evgeniy Stepanov [Mon, 29 Aug 2016 23:15:46 +0000 (23:15 +0000)]
Fix typo in test.
llvm-svn: 280028
Lang Hames [Mon, 29 Aug 2016 23:10:20 +0000 (23:10 +0000)]
[ORC] Fix unit-test breakage from r280016.
Void functions returning error now boolean convert to 'false' if they succeed.
Unit tests updated to reflect this.
llvm-svn: 280027
Vitaly Buka [Mon, 29 Aug 2016 22:59:02 +0000 (22:59 +0000)]
[asan] Attempt to fix test on darwin bot
llvm-svn: 280026
Michael Kuperstein [Mon, 29 Aug 2016 22:49:05 +0000 (22:49 +0000)]
Fix typo in comment. NFC.
llvm-svn: 280025
Teresa Johnson [Mon, 29 Aug 2016 22:46:56 +0000 (22:46 +0000)]
[ThinLTO] Indirect call promotion fixes for promoted local functions
Summary:
Fix a couple issues limiting the application of indirect call promotion
in ThinLTO mode:
- Invoke indirect call promotion before globalopt, since it may
eliminate imported functions which appear unreferenced.
- Invoke indirect call promotion with InLTO=true so that the PGOFuncName
metadata is used to get the name for locals which would have been
renamed during promotion.
Reviewers: davidxl, mehdi_amini
Subscribers: Prazek, llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D24004
llvm-svn: 280024
Chris Bieneman [Mon, 29 Aug 2016 22:26:00 +0000 (22:26 +0000)]
[CMake] Trying to fix the bots I broke
See: http://lab.llvm.org:8011/builders/libcxx-libcxxabi-x86_64-linux-ubuntu-cxx1z/builds/981/steps/build.libcxxabi/logs/stdio
llvm-svn: 280023
Hal Finkel [Mon, 29 Aug 2016 22:25:36 +0000 (22:25 +0000)]
[PowerPC] Fix i8/i16 atomics for little-Endian targets without partword atomics
For little-Endian PowerPC, we generally target only P8 and later by default.
However, generic (older) 64-bit configurations are still an option, and in that
case, partword atomics are not available (e.g. stbcx.). To lower i8/i16 atomics
without true i8/i16 atomic operations, we emulate using i32 atomics in
combination with a bunch of shifting and masking, etc. The amount by which to
shift in little-Endian mode is different from the amount in big-Endian mode (it
is inverted -- meaning we can leave off the xor when computing the amount).
Fixes PR22923.
llvm-svn: 280022
Chris Bieneman [Mon, 29 Aug 2016 22:12:21 +0000 (22:12 +0000)]
[CMake] Use -std=c++11 if supported
Summary: This patch adds a check for if -std=c++11 is a supported flag, and adds it to CMAKE_CXX_FLAGS if it is supported.
Reviewers: EricWF
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D24007
llvm-svn: 280021
Chad Rosier [Mon, 29 Aug 2016 22:09:51 +0000 (22:09 +0000)]
[SLP] Return a boolean value for these static helpers. NFC.
Differential Revision: https://reviews.llvm.org/D24008
llvm-svn: 280020
Jan Vesely [Mon, 29 Aug 2016 22:05:06 +0000 (22:05 +0000)]
AMDGPU/R600: Remove MergeVectorStores from legalization
This is handled by DAGCombiner in a more generic way
Differential Revision: https://reviews.llvm.org/D23970
llvm-svn: 280019
Rui Ueyama [Mon, 29 Aug 2016 22:01:21 +0000 (22:01 +0000)]
Make lld actually compatible with gold in terms of filler handling.
GNU gold handles output section fillers as 32-bit values.
This patch makes LLD compatible with that behavior.
Differential revision: https://reviews.llvm.org/D23181
llvm-svn: 280018
Lang Hames [Mon, 29 Aug 2016 21:57:52 +0000 (21:57 +0000)]
[ORC][RPC] Fix typo in RPC comments: call primitives on void functions return
future<Error>, not future<bool>.
llvm-svn: 280017
Lang Hames [Mon, 29 Aug 2016 21:56:30 +0000 (21:56 +0000)]
[ORC][RPC] Make the future type of an Orc RPC call Error/Expected rather than
Optional.
For void functions the return type of a nonblocking call changes from
Expected<future<Optional<bool>>> to Expected<future<Error>>, and for functions
returning T the return type changes from Expected<future<Optional<T>>> to
Expected<future<Expected<T>>>.
Inner results need to be checked (since the RPC connection may have dropped
out before a result came back) and Error/Expected provide stronger checking
requirements. It also allows us drop the crufty 'optionalToError' function and
just collapse Errors in the single-threaded call primitives.
llvm-svn: 280016
Saleem Abdulrasool [Mon, 29 Aug 2016 21:33:37 +0000 (21:33 +0000)]
libc++: perform configuration checks with -nodefaultlibs
We're compiling libc++ with -nodefaultlibs, so we should also pass this
option during the configuration checks to ensure those checks are
consistent with the actual build.
The primary motivation here is to ease cross-compilation against a
non-standard set of C++ libraries. Previously, the configuration checks
would attempt to link against the standard C++ libraries, which would
cause link failures when cross-compiling, even though the actual library
link would go through correctly (because of the use of -nodefaultlibs
and explicitly specifying any needed libraries). This is more correct
even ignoring the motivation, however.
Patch by Shoaib Meenai!
llvm-svn: 280015
Saleem Abdulrasool [Mon, 29 Aug 2016 21:33:01 +0000 (21:33 +0000)]
COFF: disambiguate make_unique (NFC)
This disambiguates `llvm::make_unqiue` and `std::make_unique` for the Windows
buildbots.
llvm-svn: 280014
Chris Bieneman [Mon, 29 Aug 2016 21:26:32 +0000 (21:26 +0000)]
[CMake] Make LLVMConfig.cmake variable names match in-tree names
With the runtimes build we're trying to use LLVMConfig.cmake as a way of providing LLVM_* variables that are needed to behave as if the project is building in tree. To make this work we need to rename two variables by dropping the "S" from the end. This makes the variables match the in-tree names.
This renames:
LLVM_INCLUDE_DIRS -> LLVM_INCLUDE_DIR
LLVM_LIBRARY_DIRS -> LLVM_LIBRARY_DIR
The versions ending in S are not used in-tree anywhere. This also cleans up LLVM_LIBRARY_DIR being set to the same value with and without the "S".
llvm-svn: 280013
Saleem Abdulrasool [Mon, 29 Aug 2016 21:20:46 +0000 (21:20 +0000)]
COFF: add beginnings of debug directory creation
The IMAGE_FILE_HEADER structure contains a (RVA, size) to an array of
COFF_DEBUG_DIRECTORY records. Each one of these records contains an RVA to a OMF
Debug Directory. These OMF debug directories are derived into newer types such
as PDB70, PDB20, etc. This constructs a PDB70 structure which will allow us to
associate a GUID with a build to actually tie debug information.
llvm-svn: 280012
Tim Northover [Mon, 29 Aug 2016 21:00:00 +0000 (21:00 +0000)]
GlobalISel: use multi-dimensional arrays for legalize actions.
Instead of putting all possible requests into a single table, we can perform
the extremely dense lookup based on opcode and type-index in constant time
using multi-dimensional array-like things.
This roughly halves the time spent doing legalization, which was dominated by
queries against the Actions table.
llvm-svn: 280011
Adrian Prantl [Mon, 29 Aug 2016 20:46:59 +0000 (20:46 +0000)]
Fix a bug preventing the cause for a module file-not-found from being displayed
llvm-svn: 280010
Adrian Prantl [Mon, 29 Aug 2016 20:46:56 +0000 (20:46 +0000)]
Fix a bug preventing the cause of a module-out-of-date error from being printed
llvm-svn: 280009
Easwaran Raman [Mon, 29 Aug 2016 20:45:51 +0000 (20:45 +0000)]
Fix a thinko in r278189.
llvm-svn: 280008
Eric Fiselier [Mon, 29 Aug 2016 20:43:38 +0000 (20:43 +0000)]
Fix or suppress GCC warnings during build.
Summary:
Currently a number of GCC warnings are emitted when building libc++. This patch fixes or ignores all of them. The primary changes are:
* Work around strict aliasing issues in `typeinfo::hash_code()` by using __attribute__((may_alias)). However I think a non-aliasing `hash_code()` implementation is possible. Further investigation needed.
* Add `_LIBCPP_UNREACHABLE()` to switch in `strstream.cpp` to avoid -Wpotentially-uninitialized.
* Fix -Wunused-value warning in `__all` by adding a void cast.
* Ignore -Wattributes for now. There are a number of real attribute issues when using GCC but enabling the warning is too noisy.
* Ignore -Wliteral-suffix since it warns about the use of reserved identifiers. Note Only GCC 7.0 supports disabling this warning.
* Ignore -Wc++14-compat since it warns about the sized new/delete overloads.
Reviewers: EricWF
Differential Revision: https://reviews.llvm.org/D24003
llvm-svn: 280007
Saleem Abdulrasool [Mon, 29 Aug 2016 20:42:07 +0000 (20:42 +0000)]
AMDGPU: fix mismatch tags, NFC
llvm-svn: 280006
Saleem Abdulrasool [Mon, 29 Aug 2016 20:42:03 +0000 (20:42 +0000)]
ExecutionEngine: fix a bug in the movt/movw relocator
According to the arm arm specifications, 4 bytes are needed for a shift instead
of 8, this was causing the movt instruction to write to a different register
sometimes.
Patch by Walter Erquinigo!
llvm-svn: 280005
Chris Bieneman [Mon, 29 Aug 2016 20:18:52 +0000 (20:18 +0000)]
[CMake] Builtins build needs LLVM_*_OUTPUT_INTDIR variables
This allows the builtins archives to build into the correct subdirectory under the binary dir. Addresses the issue discussed in D24001.
llvm-svn: 280002
Matthew Simpson [Mon, 29 Aug 2016 20:14:04 +0000 (20:14 +0000)]
[LV] Move insertelement sequence after scalar definitions
After r279649 when getting a vector value from VectorLoopValueMap, we create an
insertelement sequence on-demand if the value has been scalarized instead of
vectorized. We previously inserted this insertelement sequence before the
value's first vector user. However, this insert location is problematic if that
user is the phi node of a first-order recurrence. With this patch, we move the
insertelement sequence after the last scalar instruction we created when
scalarizing the value. Thus, the value's vector definition in the new loop will
immediately follow its scalar definitions. This should fix PR30183.
Reference: https://llvm.org/bugs/show_bug.cgi?id=30183
llvm-svn: 280001
Zachary Turner [Mon, 29 Aug 2016 19:58:14 +0000 (19:58 +0000)]
Convert GetNameColonValue to return StringRefs.
StringExtractor::GetNameColonValue() looks for a substring of the
form "<name>:<value>" and returns <name> and <value> to the caller.
This results in two unnecessary string copies, since the name and
value are not translated in any way and simply returned as-is.
By converting this to return StringRefs we can get rid of hundreds
of string copies.
llvm-svn: 280000
Eric Fiselier [Mon, 29 Aug 2016 19:50:49 +0000 (19:50 +0000)]
Turn On -DLIBCXX_ENABLE_BENCHMARKS by default.
This patch enables the `cxx-benchmarks` target by default. Note that the target
still has to be manually invoked since it isn't included in the default 'make'
rule.
This patch also gets the benchmarks building w/ GCC. The build previously
required the '-stdlib=libc++' flag but upstream patches to Google Benchmark
now allow the library to build w/ libc++ and GCC.
These changes should make the benchmarks easier to build and test.
llvm-svn: 279999
Krzysztof Parzyszek [Mon, 29 Aug 2016 19:50:15 +0000 (19:50 +0000)]
Propagate TBAA info in SelectionDAG::getIndexedLoad
Patch by Pranav Bhandarkar.
llvm-svn: 279998
Zachary Turner [Mon, 29 Aug 2016 19:45:59 +0000 (19:45 +0000)]
Copy StringExtractor to StdStringExtractor.
I have some improvements to make to StringExtractor that require
using LLVM. debugserver can't take a dependency on LLVM but uses
this file, so I'm forking it off into StdStringExtractor and
StringExtractor, so that StringExtractor can take advantage of
some performance improvements and readability improvements that
LLVM can provide.
llvm-svn: 279997
Douglas Katzman [Mon, 29 Aug 2016 19:42:57 +0000 (19:42 +0000)]
[Myriad]: add missing 'mcpu' values
Should have been done with r276646.
llvm-svn: 279996
Tom Stellard [Mon, 29 Aug 2016 19:42:52 +0000 (19:42 +0000)]
AMDGPU/SI: Implement a custom MachineSchedStrategy
Summary:
GCNSchedStrategy re-uses most of GenericScheduler, it's just uses
a different method to compute the excess and critical register
pressure limits.
It's not enabled by default, to enable it you need to pass -misched=gcn
to llc.
Shader DB stats:
32464 shaders in 17874 tests
Totals:
SGPRS: 1542846 -> 1643125 (6.50 %)
VGPRS: 1005595 -> 904653 (-10.04 %)
Spilled SGPRs: 29929 -> 27745 (-7.30 %)
Spilled VGPRs: 334 -> 352 (5.39 %)
Scratch VGPRs: 1612 -> 1624 (0.74 %) dwords per thread
Code Size:
36688188 ->
37034900 (0.95 %) bytes
LDS: 1913 -> 1913 (0.00 %) blocks
Max Waves: 254101 -> 265125 (4.34 %)
Wait states: 0 -> 0 (0.00 %)
Totals from affected shaders:
SGPRS: 1338220 -> 1438499 (7.49 %)
VGPRS: 886221 -> 785279 (-11.39 %)
Spilled SGPRs: 29869 -> 27685 (-7.31 %)
Spilled VGPRs: 334 -> 352 (5.39 %)
Scratch VGPRs: 1612 -> 1624 (0.74 %) dwords per thread
Code Size:
34315716 ->
34662428 (1.01 %) bytes
LDS: 1551 -> 1551 (0.00 %) blocks
Max Waves: 188127 -> 199151 (5.86 %)
Wait states: 0 -> 0 (0.00 %)
Reviewers: arsenm, mareko, nhaehnle, MatzeB, atrick
Subscribers: arsenm, kzhuravl, llvm-commits
Differential Revision: https://reviews.llvm.org/D23688
llvm-svn: 279995
Zachary Turner [Mon, 29 Aug 2016 19:30:26 +0000 (19:30 +0000)]
Remove std::atomic from lldb::Address.
std::atomic<uint64_t> requires 64-bit alignment in order to
guarantee atomicity. Normally the compiler is pretty good about
aligning types, but an exception to this is when the type is
passed by value as a function parameter. In this case, if your
stack is 4-byte aligned, most modern compilers (including clang
as of LLVM 4.0) fail to align the type, rendering the atomicity
ineffective.
A deeper investigation of the class's implementation suggests
that the use of atomic was in vain anyway, because if the class
were to be shared amongst multiple threads, there were already
other data races present, and that the proper way to ensure
thread-safe access to this data would be to use a mutex from a
higher level.
Since the std::atomic was not serving its intended purpose anyway,
and since the presence of it generates compiler errors on some
platforms that cannot be workaround, we remove std::atomic from
Address here. Although unlikely, if data races do resurface
the proper fix should involve a mutex from a higher level, or an
attempt to limit the Address's access to a single thread.
llvm-svn: 279994
Vitaly Buka [Mon, 29 Aug 2016 19:28:34 +0000 (19:28 +0000)]
[asan] Enable new stack poisoning with store instruction by default
Reviewers: eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23968
llvm-svn: 279993
Tim Northover [Mon, 29 Aug 2016 19:27:20 +0000 (19:27 +0000)]
GlobalISel: switch to SmallVector for pending legalizations.
std::queue was doing far to many heap allocations to be healthy.
llvm-svn: 279992
Tom Stellard [Mon, 29 Aug 2016 19:15:22 +0000 (19:15 +0000)]
AMDGPU/SI: Improve SILoadStoreOptimizer and run it before the scheduler
Summary:
The SILoadStoreOptimizer can now look ahead more then one instruction when
looking for instructions to merge, which greatly improves the number of
loads/stores that we are able to merge.
Moving the pass before scheduling avoids increasing register pressure after
the scheduler, so that the scheduler's register pressure estimates will be
more accurate. It also gives more consistent results, since it is no longer
affected by minor scheduling changes.
Reviewers: arsenm
Subscribers: arsenm, kzhuravl, llvm-commits
Differential Revision: https://reviews.llvm.org/D23814
llvm-svn: 279991
Tim Northover [Mon, 29 Aug 2016 19:12:20 +0000 (19:12 +0000)]
ASan: remove variable only used in assertions build
llvm-svn: 279990
Eric Fiselier [Mon, 29 Aug 2016 19:12:01 +0000 (19:12 +0000)]
Update Google Benchmark library.
llvm-svn: 279989
Tim Northover [Mon, 29 Aug 2016 19:07:16 +0000 (19:07 +0000)]
GlobalISel: legalize frem to a libcall on AArch64.
llvm-svn: 279988
Tim Northover [Mon, 29 Aug 2016 19:07:08 +0000 (19:07 +0000)]
GlobalISel: rework CallLowering so that it can be used for libcalls too.
There should be no functional change here, I'm just making the implementation
of "frem" (to libcall) legalization easier for a followup.
llvm-svn: 279987
Matt Arsenault [Mon, 29 Aug 2016 19:01:48 +0000 (19:01 +0000)]
AMDGPU/R600: Fix fixups used for constant arrays
Fixes bug 29289
llvm-svn: 279986
Kyle Butt [Mon, 29 Aug 2016 18:27:12 +0000 (18:27 +0000)]
IfConversion: Fix branch predication bug.
This bug shows up with diamonds that share unpredicable, unanalyzable branches.
There's an included test case from Hexagon. What was happening was that we were
attempting to predicate the branch instruction despite the fact that it was
checked to be the same. Now for unanalyzable branches we skip over the branch
instructions when predicating the block.
Differential Revision: https://reviews.llvm.org/D23939
llvm-svn: 279985
Vitaly Buka [Mon, 29 Aug 2016 18:17:21 +0000 (18:17 +0000)]
Use store operation to poison allocas for lifetime analysis.
Summary:
Calling __asan_poison_stack_memory and __asan_unpoison_stack_memory for small
variables is too expensive.
Code is disabled by default and can be enabled by -asan-experimental-poisoning.
PR27453
Reviewers: eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23947
llvm-svn: 279984
Kostya Serebryany [Mon, 29 Aug 2016 17:45:43 +0000 (17:45 +0000)]
[scudo] use 32 bits of ASLR entropy instead of just 24 when shuffling allocated chunks
llvm-svn: 279983
Vitaly Buka [Mon, 29 Aug 2016 17:41:29 +0000 (17:41 +0000)]
[asan] Separate calculation of ShadowBytes from calculating ASanStackFrameLayout
Summary: No functional changes, just refactoring to make D23947 simpler.
Reviewers: eugenis
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23954
llvm-svn: 279982
Vitaly Buka [Mon, 29 Aug 2016 17:16:59 +0000 (17:16 +0000)]
[asan] Remove runtime flag detect_stack_use_after_scope
Summary:
We are going to use store instructions to poison some allocas.
Runtime flag will require branching in instrumented code on every lifetime
intrinsic. We'd like to avoid that.
Reviewers: eugenis
Subscribers: llvm-commits, kubabrecka
Differential Revision: https://reviews.llvm.org/D23967
llvm-svn: 279981
David Majnemer [Mon, 29 Aug 2016 17:14:08 +0000 (17:14 +0000)]
[SimplifyCFG] Hoisting invalidates metadata
We forgot to remove optimization metadata when performing hosting during
FoldTwoEntryPHINode.
This fixes PR29163.
llvm-svn: 279980
Reid Kleckner [Mon, 29 Aug 2016 16:35:43 +0000 (16:35 +0000)]
Make vec_fabs.ll pass with MSVC 2013
We should revert this change once we drop support for MSVC 2013.
llvm-svn: 279979
Reid Kleckner [Mon, 29 Aug 2016 16:24:57 +0000 (16:24 +0000)]
Try to fix clang-offload-bunder.c test once more
llvm-svn: 279978
Teresa Johnson [Mon, 29 Aug 2016 16:22:23 +0000 (16:22 +0000)]
[gold] Fix test accidentally regressed for newer gold
With r279911 I accidentally regressed the gold/X86/start-lib-common.ll
test for newer golds (v1.12+) that honor the --start-lib/--end-lib.
Remove the alignment which should not be there to make this work with
both old and new gold linkers.
Additionally, now that we have a subdirectory for v1.12+ gold tests,
copy this test there and check specifically for the v1.12+ behavior.
llvm-svn: 279977
Evandro Menezes [Mon, 29 Aug 2016 16:04:37 +0000 (16:04 +0000)]
[AArch64] Adjust the scheduling model for Exynos M1.
Further refine the model for loads.
llvm-svn: 279976
Anna Thomas [Mon, 29 Aug 2016 15:41:59 +0000 (15:41 +0000)]
[StatepointsForGC] Rematerialize in the presence of PHIs
Summary:
While walking the use chain for identifying rematerializable values in RS4GC,
add the case where the current value and base value are the same PHI nodes.
This will aid rematerialization of geps and casts instead of relocating.
Reviewers: sanjoy, reames, igor
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D23920
llvm-svn: 279975
Teresa Johnson [Mon, 29 Aug 2016 15:33:01 +0000 (15:33 +0000)]
[LTO] Remove extraneous output
Remove some debugging output to stderr that snuck in with r279576.
llvm-svn: 279974
Sanjay Patel [Mon, 29 Aug 2016 15:27:17 +0000 (15:27 +0000)]
[Constant] remove fdiv and frem from canTrap()
Assuming the default FP env, we should not treat fdiv and frem any differently in terms of
trapping behavior than any other FP op. Ie, FP ops do not trap with the default FP env.
This matches how we treat the fdiv/frem in IR with isSafeToSpeculativelyExecute() and in
the backend after:
https://reviews.llvm.org/rL279970
llvm-svn: 279973
Sanjay Patel [Mon, 29 Aug 2016 14:57:53 +0000 (14:57 +0000)]
[SimplifyCFG] rename test file, regenerate checks, and add test
The fdiv test shows a problem similar to:
https://reviews.llvm.org/rL279970
llvm-svn: 279972
Gor Nishanov [Mon, 29 Aug 2016 14:34:12 +0000 (14:34 +0000)]
[Coroutines] Part 9: Add cleanup subfunction.
Summary:
[Coroutines] Part 9: Add cleanup subfunction.
This patch completes coroutine heap allocation elision. Now, the heap elision example from docs\Coroutines.rst compiles and produces expected result (see test/Transform/Coroutines/ex3.ll)
Intrinsic Changes:
* coro.free gets a token parameter tying it to coro.id to allow reliably discovering all coro.frees associated with a particular coroutine.
* coro.id gets an extra parameter that points back to a coroutine function. This allows to check whether a coro.id describes the enclosing function or it belongs to a different function that was later inlined.
CoroSplit now creates three subfunctions:
# f$resume - resume logic
# f$destroy - cleanup logic, followed by a deallocation code
# f$cleanup - just the cleanup code
CoroElide pass during devirtualization replaces coro.destroy with either f$destroy or f$cleanup depending whether heap elision is performed or not.
Other fixes, improvements:
* Fixed buglet in Shape::buildFrame that was not creating coro.save properly if coroutine has more than one suspend point.
* Switched to using variable width suspend index field (no longer limited to 32 bit index field can be as little as i1 or as large as i<whatever-size_t-is>)
Reviewers: majnemer
Subscribers: llvm-commits, mehdi_amini
Differential Revision: https://reviews.llvm.org/D23844
llvm-svn: 279971
Sanjay Patel [Mon, 29 Aug 2016 13:32:41 +0000 (13:32 +0000)]
[TargetLowering] remove fdiv and frem from canOpTrap() (PR29114)
Assuming the default FP env, we should not treat fdiv and frem any differently in terms of
trapping behavior than any other FP op. Ie, FP ops do not trap with the default FP env.
This matches how we treat these ops in IR with isSafeToSpeculativelyExecute(). There's a
similar bug in Constant::canTrap().
This bug manifests in PR29114:
https://llvm.org/bugs/show_bug.cgi?id=29114
...as a sequence of scalar divisions instead of a vector division on x86 for a <3 x float>
type.
Differential Revision: https://reviews.llvm.org/D23974
llvm-svn: 279970
Krzysztof Parzyszek [Mon, 29 Aug 2016 13:15:35 +0000 (13:15 +0000)]
Do not use MRI::getMaxLaneMaskForVReg as a mask covering whole register
MRI::getMaxLaneMaskForVReg does not always cover the whole register.
For example, on X86 the upper 16 bits of EAX cannot be accessed via
any subregister. Consequently, there is no lane mask that only covers
that part of EAX. The getMaxLaneMaskForVReg will return the union of
the lane masks for all subregisters, and in case of EAX, that union
will not cover the upper 16 bits.
This fixes https://llvm.org/bugs/show_bug.cgi?id=29132
llvm-svn: 279969
Tom Stellard [Mon, 29 Aug 2016 13:06:10 +0000 (13:06 +0000)]
AMDGPU/SI: Improve register allocation hints for sopk instructions
Summary:
For shrinking SOPK instructions, we were creating a hint to tell the
register allocator to use the register allocated for src0 for the dst
operand as well. However, this seems to not work sometimes depending
on the order virtual registers are assigned physical registers.
To fix this, I've added a second allocation hint which does the reverse,
asks that the register allocated for dst is used for src0.
Reviewers: arsenm
Subscribers: arsenm, llvm-commits, kzhuravl
Differential Revision: https://reviews.llvm.org/D23862
llvm-svn: 279968
Rafael Espindola [Mon, 29 Aug 2016 12:47:22 +0000 (12:47 +0000)]
Use the correct ctor/dtor section for dynamic-no-pic.
llvm-svn: 279967
Benjamin Kramer [Mon, 29 Aug 2016 12:41:32 +0000 (12:41 +0000)]
Mark test as XFAIL instead of disabling it everywhere.
There is no lit feature 'X86' so this test is just disabled completely.
Make it XFAIL until a solution is found.
llvm-svn: 279966
Rafael Espindola [Mon, 29 Aug 2016 12:33:42 +0000 (12:33 +0000)]
Move code only used by codegen out of MC. NFC.
MC itself never needs to know about these sections.
llvm-svn: 279965
Haojian Wu [Mon, 29 Aug 2016 12:26:33 +0000 (12:26 +0000)]
Fix -Wunused-but-set-variable warning.
Summary: A follow-up fix on r279958.
Reviewers: bkramer
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D23989
llvm-svn: 279964
Tom Stellard [Mon, 29 Aug 2016 12:05:32 +0000 (12:05 +0000)]
AMDGPU/SI: Query AA, if available, in areMemAccessesTriviallyDisjoint()
Summary:
The SILoadStoreOptimizer will need to use AliasAnalysis here in order to
move it before scheduling.
Reviewers: arsenm
Subscribers: arsenm, llvm-commits, kzhuravl
Differential Revision: https://reviews.llvm.org/D23813
llvm-svn: 279963
Igor Kudrin [Mon, 29 Aug 2016 11:48:50 +0000 (11:48 +0000)]
[Coverage] Prevent creating a redundant counter if a nested body ends with a macro.
If there were several nested statements arranged in a way that all of them
end up with the same macro, then the expansion of this macro was assigned
with all the corresponding counters of these statements.
As a result, the wrong counter value was shown for the macro in llvm-cov.
This patch fixes the issue by preventing adding a counter for an expanded
source range if it already has an assigned counter, which is expected
to come from the most specific statement.
Differential Revision: https://reviews.llvm.org/D23160
llvm-svn: 279962
Igor Breger [Mon, 29 Aug 2016 09:12:31 +0000 (09:12 +0000)]
Fixed a bug in type legalizer for masked gather.
The problem occurs when the Node doesn't updated in place , UpdateNodeOperation() return the node that already exist.
In this case assert fail in PromoteIntegerOperand() , N have 2 results ( val + chain).
Differential Revision: http://reviews.llvm.org/D23756
llvm-svn: 279961
Igor Breger [Mon, 29 Aug 2016 08:52:52 +0000 (08:52 +0000)]
[AVX512] In some cases KORTEST instruction may be used instead of ZEXT + TEST sequence.
Differential Revision: http://reviews.llvm.org/D23490
llvm-svn: 279960
Haojian Wu [Mon, 29 Aug 2016 08:48:15 +0000 (08:48 +0000)]
[InstructionSelect] NumBlocks isn't defined in DEBUG build.
Summary: A follow-up fixing on http://llvm.org/viewvc/llvm-project?view=revision&revision=279905.
Reviewers: bkramer
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D23985
llvm-svn: 279959
Craig Topper [Mon, 29 Aug 2016 04:49:31 +0000 (04:49 +0000)]
[X86] Don't lower FABS/FNEG masking directly to a ConstantPool load. Just create a ConstantFPSDNode and let that be lowered.
This allows broadcast loads to used when available.
llvm-svn: 279958
Craig Topper [Mon, 29 Aug 2016 04:49:27 +0000 (04:49 +0000)]
[AVX-512] Always use v8i64 when converting 512-bit FAND/FOR/FXOR/FANDN to integer operations when DQI isn't supported. This is consistent with the recent changes to promote logical operations to i64 vectors.
llvm-svn: 279957
Craig Topper [Mon, 29 Aug 2016 04:49:24 +0000 (04:49 +0000)]
[AVX-512] Add 512-bit fabs tests with and without AVX512DQ.
llvm-svn: 279956
Eric Fiselier [Mon, 29 Aug 2016 01:43:41 +0000 (01:43 +0000)]
Fix pair::operator=(TupleLike&&).
This assignment operator was previously broken since the SFINAE always resulted
in substitution failure. This caused assignments to turn into
copy construction + assignment.
This patch was originally committed as r279953 but was reverted due to warnings
in the test-suite. This new patch corrects those warnings.
llvm-svn: 279955
Eric Fiselier [Mon, 29 Aug 2016 01:39:54 +0000 (01:39 +0000)]
Revert r279953 - Fix pair::operator=(TupleLike&&)
The test emits warnings causing the test-suite to fail. Since I want this
patch merged into 3.9 I'll recommit it with a clean test.
llvm-svn: 279954
Eric Fiselier [Mon, 29 Aug 2016 01:09:47 +0000 (01:09 +0000)]
Fix pair::operator=(TupleLike&&).
This assignment operator was previously broken since the SFINAE always resulted
in substitution failure. This caused assignments to turn into
copy construction + assignment.
llvm-svn: 279953
Lang Hames [Mon, 29 Aug 2016 00:54:29 +0000 (00:54 +0000)]
[Orc] Simplify LogicalDylib and move it back inside CompileOnDemandLayer. Also
switch to using one indirect stub manager per logical dylib rather than one per
input module.
LogicalDylib is a helper class used by the CompileOnDemandLayer to manage
symbol resolution between modules during lazy compilation. In particular, it
ensures that internal symbols resolve correctly even in the case where multiple
input modules contain the same internal symbol name (which must to be promoted
to external hidden linkage so that functions in any given module can be split
out by lazy compilation). LogicalDylib's resolution scheme (before this commit)
required one stub-manager per input module. This made recompilation of functions
(by adding a module containing a new definition) difficult, as the stub manager
for any given symbol was bound to the module that supplied the original
definition. By using one stubs manager for the whole logical dylib symbols can
be more easily replaced, although support for doing this is not included in this
patch (it will be implemented in a follow up).
llvm-svn: 279952
Craig Topper [Sun, 28 Aug 2016 22:20:51 +0000 (22:20 +0000)]
[AVX-512] Add support for selecting 512-bit VPABSB/VPABSW when BWI is available.
llvm-svn: 279951
Craig Topper [Sun, 28 Aug 2016 22:20:48 +0000 (22:20 +0000)]
[AVX-512] Add patterns for selecting 128/256-bit EVEX VPABS instructions.
llvm-svn: 279950
Craig Topper [Sun, 28 Aug 2016 22:20:45 +0000 (22:20 +0000)]
[AVX-512] Add testcases showing that we don't emit 512-bit vpabsb/vpabsw. Will be fixed in a future commit.
llvm-svn: 279949
Eric Fiselier [Sun, 28 Aug 2016 22:14:37 +0000 (22:14 +0000)]
Implement C++17 std::sample.
This patch implements the std::sample function added to C++17 from LFTS. It
also removes the std::experimental::sample implementation which now forwards
to std::sample.
llvm-svn: 279948
Eric Fiselier [Sun, 28 Aug 2016 21:55:00 +0000 (21:55 +0000)]
Mark LWG 2716 as complete - shuffle and sample disallows lvalue URNGs.
Libc++'s implementation of shuffle and sample already support lvalue and rvalue
RNG's. This patch adds tests for both categories and marks the issue as complete.
This patch also contains drive-by change for std::experimental::sample which
improves the diagnostics produced when the correct iterator categories are
not supplied.
llvm-svn: 279947
Saleem Abdulrasool [Sun, 28 Aug 2016 21:33:30 +0000 (21:33 +0000)]
AST: improve layout of SimpleTypoCorrector
Add the "explicit" specifier to the single-argument constructor of
SimpleTypoCorrector. Reorder the fields to remove excessive padding (8 bytes).
Patch by Alexander Shaposhnikov!
llvm-svn: 279946
Eric Fiselier [Sun, 28 Aug 2016 21:26:01 +0000 (21:26 +0000)]
Implement LWG 2711. Constrain path members.
llvm-svn: 279945
Sylvestre Ledru [Sun, 28 Aug 2016 20:33:42 +0000 (20:33 +0000)]
Fix some typos in the doc
llvm-svn: 279944
Sylvestre Ledru [Sun, 28 Aug 2016 20:29:18 +0000 (20:29 +0000)]
Fix some typos in the doc
llvm-svn: 279943
Sylvestre Ledru [Sun, 28 Aug 2016 20:22:34 +0000 (20:22 +0000)]
Fix a typo in the doc: overriden -> overridden
llvm-svn: 279942
Saleem Abdulrasool [Sun, 28 Aug 2016 20:10:33 +0000 (20:10 +0000)]
EHABI: fail on WMMX vops without WMMX support
When the unwinder is built without WMMX support, if we encounter a WMMX register
virtual operation, early rather than attempting to continue as we would not have
saved the register set anyways. This should never come down this path, but,
just in case, help it abort more explicitly.
llvm-svn: 279941
Eric Fiselier [Sun, 28 Aug 2016 18:33:08 +0000 (18:33 +0000)]
[Docs] Update libc++ target names after r279675.
llvm-svn: 279940
Sanjay Patel [Sun, 28 Aug 2016 18:31:32 +0000 (18:31 +0000)]
[x86] add tests for <3 x N> vector types (PR29114)
llvm-svn: 279939
Eric Fiselier [Sun, 28 Aug 2016 18:28:51 +0000 (18:28 +0000)]
[CMake] Be more consistent about naming targets and components in libc++abi
This patch (and commit summary) mirror r279675 by Chris B which was applied
to libc++ but not libc++abi.
Summary:
The point of this patch is to have a consistent convention for naming build,
check and install targets so that the targets can be constructed from the
project name.
This change renames a bunch of CMake components and targets from libcxxabi to
cxxabi. For each renamed target I've added a convenience target that matches
the old target name and depends on the new target. This will preserve function
of the old targets so that the change doesn't break the world. We can evaluate
if it is worth removing the extra targets later.
llvm-svn: 279938
Sanjay Patel [Sun, 28 Aug 2016 18:18:00 +0000 (18:18 +0000)]
[InstCombine] use m_APInt to allow icmp (and X, Y), C folds for splat constant vectors
llvm-svn: 279937
Eric Fiselier [Sun, 28 Aug 2016 18:16:18 +0000 (18:16 +0000)]
[CMake] Apply r279151 cleanup to libc++abi.
This patch applies changes similar to those in r279515 to libc++abi.
Summary of changes in this patch:
* Renamed variable LLVM_CONFIG -> LLVM_CONFIG_PATH
* Renamed variable LIBCXXABI_BUILT_STANDALONE -> LIBCXXABI_STANDALONE_BUILD
* Add an include of AddLLVM in the tests subdirectory for add_lit_testsuite.
llvm-svn: 279936