Shoaib Meenai [Mon, 31 Oct 2016 15:09:10 +0000 (15:09 +0000)]
[libc++] Add configuration define for off_t functions
Create this define in __config and use it elsewhere, instead of checking
the operating system/library defines in other files. The aim is to
reduce the usage of _WIN32 outside __config. No functional change.
Differential Revision: https://reviews.llvm.org/D25741
llvm-svn: 285582
Malcolm Parsons [Mon, 31 Oct 2016 14:48:49 +0000 (14:48 +0000)]
Fix link to check
llvm-svn: 285581
Rafael Espindola [Mon, 31 Oct 2016 14:44:41 +0000 (14:44 +0000)]
Simplify. NFC.
No need to subtract then add Sec->getVA().
llvm-svn: 285580
Malcolm Parsons [Mon, 31 Oct 2016 14:43:37 +0000 (14:43 +0000)]
[clang-tidy] Enhance modernize-use-auto to casts
Summary:
Extend modernize-use-auto to cases when a variable is assigned with a cast.
e.g.
Type *Ptr1 = dynamic_cast<Type*>(Ptr2);
http://llvm.org/PR25499
Reviewers: angelgarcia, aaron.ballman, klimek, Prazek, alexfh
Subscribers: Prazek, Eugene.Zelenko, cfe-commits
Tags: #clang-tools-extra
Differential Revision: https://reviews.llvm.org/D25316
llvm-svn: 285579
Ulrich Weigand [Mon, 31 Oct 2016 14:38:05 +0000 (14:38 +0000)]
[SystemZ] Add -march=archX aliases
For compatibility with other compilers on the platform, allow specifying
levels of the z/Architecture instead of model names with -march. In
particular, the following aliases are now supported:
-march=arch8 equals -march=z10
-march=arch9 equals -march=z196
-march=arch10 equals -march=zEC12
-march=arch11 equals -march=z13
This parallels the equivalent (and prerequisite) LLVM change in r285577.
llvm-svn: 285578
Ulrich Weigand [Mon, 31 Oct 2016 14:33:29 +0000 (14:33 +0000)]
[SystemZ] Rework processor feature definitions and add -mcpu=archX support
This patch implements two changes:
- Move processor feature definition into a new file SystemZFeatures.td,
and provide explicit lists of supported and unsupported features for
each level of the z/Architecture. This allows specifying unsupported
features in the scheduler definition files for each processor.
- Add optional aliases for the -mcpu processor names according to the
level of the z/Architecture, for compatibility with other compilers
on the platform. The supported aliases are:
-mcpu=arch8 equals -mcpu=z10
-mcpu=arch9 equals -mcpu=z196
-mcpu=arch10 equals -mcpu=zEC12
-mcpu=arch11 equals -mcpu=z13
llvm-svn: 285577
Ulrich Weigand [Mon, 31 Oct 2016 14:28:43 +0000 (14:28 +0000)]
[SystemZ] Guard LEFR/LFER with FeatureVector
The LEFR/LFER pseudos are aliases for vector instructions and should
therefore be guared by FeatureVector. If they aren't, the TableGen
scheduler definition checking might complain that there is no data
for those pseudos for pre-z13 machines.
No functional change intended.
llvm-svn: 285576
Ulrich Weigand [Mon, 31 Oct 2016 14:25:05 +0000 (14:25 +0000)]
[SystemZ] Correctly diagnose missing features in AsmParser
Currently, when using an instruction that is not supported on the
currently selected architecture, the LLVM assembler is likely to
diagnose an "invalid operand" instead of a "missing feature".
This is because many operands require a custom parser in order to
be processed correctly, and if an instruction is not available
according to the current feature set, the generated parser code
will also not detect the associated custom operand parsers.
Fixed by temporarily enabling all features while parsing operands.
The missing features will then be correctly detected when actually
parsing the instruction itself.
llvm-svn: 285575
Ulrich Weigand [Mon, 31 Oct 2016 14:21:36 +0000 (14:21 +0000)]
[SystemZ] Fix encoding of MVCK and .insn ss
LLVM currently treats the first operand of MVCK as if it were a
regular base+index+displacement address. However, it is in fact
a base+displacement combined with a length register field.
While the two might look syntactically similar, there are two
semantic differences:
- %r0 is a valid length register, even though it cannot be used
as an index register.
- In an expression with just a single register like 0(%rX), the
register is treated as base with normal addresses, while it is
treated as the length register (with an empty base) for MVCK.
Fixed by adding a new operand parser class BDRAddr and reworking
the assembler parser to distinguish between address + length
register operands and regular addresses.
llvm-svn: 285574
Michael Zuckerman [Mon, 31 Oct 2016 14:16:57 +0000 (14:16 +0000)]
second attempt at r285565.
llvm-svn: 285573
Roger Ferrer Ibanez [Mon, 31 Oct 2016 14:14:13 +0000 (14:14 +0000)]
Fix archetypes.hpp under libcpp-no-extensions and std level < 14
Under -fno-exceptions TEST_THROW becomes abort / __builtin_abort which returns
void. This causes a type mismatch in the conditional operator when testing the
library in C++98,03,11 modes.
Use a comma operator to workaround this problem.
Differential Revision: https://reviews.llvm.org/D26147
llvm-svn: 285572
Eric Fiselier [Mon, 31 Oct 2016 14:14:04 +0000 (14:14 +0000)]
Spell libcxxabi-no-threads correctly
llvm-svn: 285571
Daniel Jasper [Mon, 31 Oct 2016 13:23:00 +0000 (13:23 +0000)]
Skip over AnnotatedLines with >50 levels of nesting; don't format them.
Reasoning:
- ExpressionParser uses a lot of stack for these, bad in some environments.
- Our formatting algorithm is N^3 and gets really slow.
- The resulting formatting is unlikely to be any good.
- This is probably generated code we're formatting by accident.
We treat these as unparseable, and signal incomplete formatting. 50 is
an arbitrary number, I've only seen real problems from ~150 levels.
Patch by Sam McCall. Thank you.
llvm-svn: 285570
Daniel Jasper [Mon, 31 Oct 2016 13:18:25 +0000 (13:18 +0000)]
clang-format: [JS] Fix missing space after 'yield'.
Before:
class X {
delete(val) {
return null;
}
* gen() {
yield[1, 2];
}
* gen() {
yield{a: 1};
}
};
After:
class X {
delete(val) {
return null;
}
* gen() {
yield [1, 2];
}
* gen() {
yield {a: 1};
}
};
llvm-svn: 285569
Dorit Nuzman [Mon, 31 Oct 2016 13:17:31 +0000 (13:17 +0000)]
Second attempt at r285517.
llvm-svn: 285568
Rafael Espindola [Mon, 31 Oct 2016 13:14:53 +0000 (13:14 +0000)]
Correctly merge visibility of linkerscript created symbols.
llvm-svn: 285567
Jonas Paulsson [Mon, 31 Oct 2016 13:05:48 +0000 (13:05 +0000)]
[SystemZ] Model 2 VBU units (not 1) in SystemZScheduleZ13.td.
NFC.
Review: Ulrich Weigand.
llvm-svn: 285566
Michael Zuckerman [Mon, 31 Oct 2016 12:49:36 +0000 (12:49 +0000)]
revert r285563 fail in test CodeGen/avx512-inline-asm-kregisters-basics.c
llvm-svn: 285565
Alexey Bataev [Mon, 31 Oct 2016 12:10:53 +0000 (12:10 +0000)]
Improved cost model for FDIV and FSQRT, by Andrew Tischenko
There is a bug describing poor cost model for floating point operations:
Bug 29083 - [X86][SSE] Improve costs for floating point operations. This
patch is the second one in series of patches dealing with cost model.
Differential Revision: https://reviews.llvm.org/D25722
llvm-svn: 285564
Michael Zuckerman [Mon, 31 Oct 2016 12:05:41 +0000 (12:05 +0000)]
[x86][inline-asm] Introducing (AVX512) k0-k7 registers for inline-asm usage
Commit on behalf of mharoush
After LGTM and check all:
This patch enables usage of k registers in inline assembly syntax.
Adding triple
Reviewer: 1. rnk
2. delena
Differential Revision: https://reviews.llvm.org/D25011
llvm-svn: 285563
Pavel Labath [Mon, 31 Oct 2016 11:53:13 +0000 (11:53 +0000)]
Use the right dwarf attributes for parsing location list entries (fixup for r285441)
Note that the parsing code here is still incorrect wrt. the new draft of the
dwarf 5 spec (seconds arguments to DW_LLE_startx_length should be uleb128, not
u32). Once we have compilers actually emitting dwarf conformant with the new
spec, we'll need to revisit this and figure out the proper behavior there.
This should unbreak the linux bot.
llvm-svn: 285562
Rafael Espindola [Mon, 31 Oct 2016 11:47:37 +0000 (11:47 +0000)]
Fix this test when we have clang-offload-bundler.exe.
llvm-svn: 285561
Manuel Klimek [Mon, 31 Oct 2016 11:40:14 +0000 (11:40 +0000)]
Add triple to test so it does not fail on windows.
llvm-svn: 285560
Manuel Klimek [Mon, 31 Oct 2016 11:18:39 +0000 (11:18 +0000)]
Delete .s file that did not test anything, and check in test that works.
In D26098, Davide Italiano submitted a .s file instead of the .ll file
that was the last stage of the review.
llvm-svn: 285559
Eugene Leviant [Mon, 31 Oct 2016 11:03:01 +0000 (11:03 +0000)]
Add getOffset to ELFSectionRef
Differential revision: https://reviews.llvm.org/D26074
llvm-svn: 285558
Alexey Bader [Mon, 31 Oct 2016 10:26:31 +0000 (10:26 +0000)]
[OpenCL] Setting constant address space for array initializers
Summary: Setting constant address space for global constants used for memcpy-initialization of arrays.
Patch by Alexey Sotkin.
Reviewers: bader, yaxunl, Anastasia
Subscribers: cfe-commits, AlexeySotkin
Differential Revision: https://reviews.llvm.org/D25305
llvm-svn: 285557
Michael Zuckerman [Mon, 31 Oct 2016 10:12:36 +0000 (10:12 +0000)]
Revert reviosion 285555
llvm-svn: 285556
Michael Zuckerman [Mon, 31 Oct 2016 09:37:59 +0000 (09:37 +0000)]
[x86][inline-asm] Introducing (AVX512) k0-k7 registers for inline-asm usage
Commit on behalf of mharoush
After LGTM and check all:
This patch enables usage of k registers in inline assembly syntax.
Reviewer: 1. rnk
2. delena
Differential Revision: https://reviews.llvm.org/D25011
llvm-svn: 285555
Roger Ferrer Ibanez [Mon, 31 Oct 2016 08:47:53 +0000 (08:47 +0000)]
Change from "XFAIL: libcpp-no-exceptions" to "UNSUPPORTED: libcpp-no-exceptions" tests that only check exceptions and nothing else
This is a follow up of D24562.
These tests do not check anything but exceptions, so it makes sense to mark
them as UNSUPPORTED under a library built without exceptions.
Differential Revision: https://reviews.llvm.org/D26075
llvm-svn: 285550
Eric Liu [Mon, 31 Oct 2016 08:28:29 +0000 (08:28 +0000)]
[change-namespace] fix namespace specifiers of template arguments.
llvm-svn: 285549
Elad Cohen [Mon, 31 Oct 2016 08:21:54 +0000 (08:21 +0000)]
[Modules] Add a command line option for loading the clang builtins modulemap.
-fbuiltin-module-map loads the clang builtins modulemap file. (This is
equivalent to -fmodule-map-file=<resource dir>/include/module.modulemap)
Differential Revision: https://reviews.llvm.org/D25767
llvm-svn: 285548
Craig Topper [Mon, 31 Oct 2016 06:24:00 +0000 (06:24 +0000)]
[AVX-512] Add a regular expression to a test that was missed in r285540.
llvm-svn: 285547
Craig Topper [Mon, 31 Oct 2016 05:55:57 +0000 (05:55 +0000)]
[AVX-512] Add missing patterns for selecting masked vector extracts that started from shuffles.
llvm-svn: 285546
Craig Topper [Mon, 31 Oct 2016 05:49:11 +0000 (05:49 +0000)]
[AVX-512] Use selectd instead of selectps for _mm256_mask_extracti32x4_epi32.
llvm-svn: 285545
David Majnemer [Mon, 31 Oct 2016 05:37:48 +0000 (05:37 +0000)]
Add support for __builtin_alloca_with_align
__builtin_alloca always uses __BIGGEST_ALIGNMENT__ for the alignment of
the allocation. __builtin_alloca_with_align allows the programmer to
specify the alignment of the allocation.
This fixes PR30658.
llvm-svn: 285544
Serge Pavlov [Mon, 31 Oct 2016 05:11:12 +0000 (05:11 +0000)]
Make output of ast-print closer to C++ code
Put semicolon after non-defining method declaration and a class
specialization body.
llvm-svn: 285543
Chris Bieneman [Mon, 31 Oct 2016 04:48:19 +0000 (04:48 +0000)]
[Test Suite] Pull generateSource into lldbtest
Summary:
Convert tests using LLDB headers to use generateSource to put the right include paths in place regardless of whether or not you're building a framework.
This also abstracted generateSource out of TestPublicAPIHeaders.py into lldbtest.py.
Reviewers: tfiala, zturner
Subscribers: lldb-commits
Differential Revision: https://reviews.llvm.org/D25887
llvm-svn: 285542
Chris Bieneman [Mon, 31 Oct 2016 04:48:10 +0000 (04:48 +0000)]
[Test Suite] Properly respect --framework option
Summary:
dotest.py has a framework option that is not respected. This patch makes the framework path properly configurable via the --framework option.
This patch also adds a function to the lldbtest.Base class named "hasDarwinFramework" which allows us to not rely on the host platform to determine if a framework is present. If running on Darwin, and not building a framework, this will follow the *nix code paths which are appropriate for Darwin.
Reviewers: tfiala
Subscribers: lldb-commits, mgorny
Differential Revision: https://reviews.llvm.org/D25886
llvm-svn: 285541
Craig Topper [Mon, 31 Oct 2016 04:30:56 +0000 (04:30 +0000)]
[AVX-512] Remove masked vector extract builtins and replace with native shufflevectors and selects.
Unfortunately, the backend currently doesn't fold masks into the instructions correctly when they come from these shufflevectors. I'll work on that in a future commit.
llvm-svn: 285540
Craig Topper [Mon, 31 Oct 2016 04:30:51 +0000 (04:30 +0000)]
[AVX-512] Remove many of the masked 128/256-bit shift builtins and replace them with unmasked builtins and selects.
llvm-svn: 285539
Eric Fiselier [Mon, 31 Oct 2016 03:42:50 +0000 (03:42 +0000)]
Add 'inline' but not 'always_inline' to std::strings destructor.
Adding both 'inline' and 'always_inline' to the destructor has been contentious.
However most of the performance benefits can be gained by only adding 'inline',
and there is no reason to hold up that change while discussing the other.
llvm-svn: 285538
Eric Fiselier [Mon, 31 Oct 2016 03:40:29 +0000 (03:40 +0000)]
Remove additional function template definitions from the dylib
llvm-svn: 285537
Sanjoy Das [Mon, 31 Oct 2016 03:32:45 +0000 (03:32 +0000)]
Make a test case more rigorous; NFC
llvm-svn: 285536
Sanjoy Das [Mon, 31 Oct 2016 03:32:43 +0000 (03:32 +0000)]
[SCEV] Try to order n-ary expressions in CompareValueComplexity
llvm-svn: 285535
Sanjoy Das [Mon, 31 Oct 2016 03:32:39 +0000 (03:32 +0000)]
[SCEV] Reduce boilerplate in unit tests
llvm-svn: 285534
Artem Dergachev [Mon, 31 Oct 2016 03:08:48 +0000 (03:08 +0000)]
[analyzer] NumberObjectConversion: support more types, misc updates.
Support CFNumberRef and OSNumber objects, which may also be accidentally
converted to plain integers or booleans.
Enable explicit boolean casts by default in non-pedantic mode.
Improve handling for warnings inside macros.
Improve error messages.
Differential Revision: https://reviews.llvm.org/D25731
llvm-svn: 285533
Eric Fiselier [Mon, 31 Oct 2016 02:46:25 +0000 (02:46 +0000)]
Optimize filesystem::path by providing weaker exception guarantees.
path uses string::append to construct, append, and concatenate paths. Unfortunatly
string::append has a strong exception safety guaranteed and if it can't prove
that the iterator operations don't throw then it will allocate a temporary
string copy to append to. However this extra allocation and copy is very
undesirable for path which doesn't have the same exception guarantees.
To work around this this patch adds string::__append_forward_unsafe which exposes
the std::string::append interface for forward iterators without enforcing
that the iterator is noexcept.
llvm-svn: 285532
Eric Fiselier [Mon, 31 Oct 2016 02:07:23 +0000 (02:07 +0000)]
Fix _LIBCPP_EXTERN_TEMPLATE_INLINE_VISIBILITY to always have default visibility.
This prevent the symbols from being both externally available and hidden, which
causes them to be linked incorrectly. This is only a problem when the address
of the function is explicitly taken since it will always be inlined otherwise.
This patch fixes the issues that caused r285456 to be reverted, and can
now be reapplied.
llvm-svn: 285531
Eric Fiselier [Sun, 30 Oct 2016 23:53:50 +0000 (23:53 +0000)]
Improve performance of constructing filesystem::path from strings.
This patch fixes a performance bug when constructing or appending to a path
from a string or c-string. Previously we called 'push_back' to append every
single character. This caused multiple re-allocation and copies when at most
one reallocation is necessary. The new behavior is to simply call
`string::append` so it can correctly handle reallocation.
For large strings this change is a ~4x improvement. This also makes our path
faster to construct than libstdc++'s.
llvm-svn: 285530
Sanjoy Das [Sun, 30 Oct 2016 23:52:56 +0000 (23:52 +0000)]
[SCEV] In CompareValueComplexity, order global values by their name
llvm-svn: 285529
Sanjoy Das [Sun, 30 Oct 2016 23:52:53 +0000 (23:52 +0000)]
[SCEV] Use auto for consistency with an upcoming change; NFC
llvm-svn: 285528
Sanjoy Das [Sun, 30 Oct 2016 23:52:50 +0000 (23:52 +0000)]
Clean up test a little bit; NFC
llvm-svn: 285527
Eric Fiselier [Sun, 30 Oct 2016 23:30:38 +0000 (23:30 +0000)]
Rewrite std::filesystem::path iterators and parser
This patch entirely rewrites the parsing logic for paths. Unlike the previous
implementation this one stores information about the current state; For example
if we are in a trailing separator or a root separator. This avoids the need for
extra lookahead (and extra work) when incrementing or decrementing an iterator.
Roughly this gives us a 15% speedup over the previous implementation.
Unfortunately this implementation is still a lot slower than libstdc++'s.
Because libstdc++ pre-parses and splits the path upon construction their
iterators are trivial to increment/decrement. This makes libc++ lazy parsing
100x slower than libstdc++. However the pre-parsing libstdc++ causes a ton
of extra and unneeded allocations when constructing the string. For example
`path("/foo/bar/")` would require at least 5 allocations with libstdc++
whereas libc++ uses only one. The non-allocating behavior is much preferable
when you consider filesystem usages like 'exists("/foo/bar/")'.
Even then libc++'s path seems to be twice as slow to simply construct compared
to libstdc++. More investigation is needed about this.
llvm-svn: 285526
Mehdi Amini [Sun, 30 Oct 2016 23:26:13 +0000 (23:26 +0000)]
Fix clang installed path to handle case where clang is invoked through a symlink
This code path is used when generating the path to libLTO.dylib, which
is passed to the linker as `-lto_library'.
Without this, if clang is invoked through a symlink, libLTO is
searched in a path relative to where the symlink is instead of
where clang is actually installed.
Fix PR30811.
Patch by: Jack Howarth
Differential Revision: https://reviews.llvm.org/D26116
llvm-svn: 285525
Eric Fiselier [Sun, 30 Oct 2016 22:53:00 +0000 (22:53 +0000)]
Add start of filesystem benchmarks
llvm-svn: 285524
Eric Fiselier [Sun, 30 Oct 2016 20:05:52 +0000 (20:05 +0000)]
Mark thread exit test as unsupported w/o threads
llvm-svn: 285523
Sanjay Patel [Sun, 30 Oct 2016 18:19:35 +0000 (18:19 +0000)]
[DAG] x | x --> x
llvm-svn: 285522
Sanjay Patel [Sun, 30 Oct 2016 18:13:30 +0000 (18:13 +0000)]
[DAG] x & x --> x
llvm-svn: 285521
Sanjay Patel [Sun, 30 Oct 2016 18:04:19 +0000 (18:04 +0000)]
[x86] add tests for basic logic op folds
llvm-svn: 285520
Michael Zuckerman [Sun, 30 Oct 2016 14:54:05 +0000 (14:54 +0000)]
Fixing "type" issue for (epi32)
and replaceing hardcoded inf with clang builtin inf "__builtin_inff()" for float ({max|min}_{pd|ps})
llvm-svn: 285519
Dorit Nuzman [Sun, 30 Oct 2016 14:34:57 +0000 (14:34 +0000)]
Revert r285517 due to build failures.
llvm-svn: 285518
Dorit Nuzman [Sun, 30 Oct 2016 12:23:26 +0000 (12:23 +0000)]
[LoopVectorize] Make interleaved-accesses analysis less conservative about
possible pointer-wrap-around concerns, in some cases.
Before this patch, collectConstStridedAccesses (part of interleaved-accesses
analysis) called getPtrStride with [Assume=false, ShouldCheckWrap=true] when
examining all candidate pointers. This is too conservative. Instead, this
patch makes collectConstStridedAccesses use an optimistic approach, calling
getPtrStride with [Assume=true, ShouldCheckWrap=false], and then, once the
candidate interleave groups have been formed, revisits the pointer-wrapping
analysis but only where it matters: namely, in groups that have gaps, and where
the gaps are not at the very end of the group (in which case the loop is
peeled). This second time getPtrStride is called with [Assume=false,
ShouldCheckWrap=true], but this could further be improved to using Assume=true,
once we also add the logic to track that we are not going to meet the scev
runtime checks threshold.
Differential Revision: https://reviews.llvm.org/D25276
llvm-svn: 285517
Craig Topper [Sun, 30 Oct 2016 07:11:34 +0000 (07:11 +0000)]
[AVX-512] Remove masked 128/256-bit builtins for vpmaddwd and vpmaddubsw. Replace with unmasked builtins and select.
llvm-svn: 285516
Craig Topper [Sun, 30 Oct 2016 06:56:16 +0000 (06:56 +0000)]
[X86] Use intrinsics table for PMADDUBSW and PMADDWD so that we can use the legacy intrinsics to select EVEX encoded instructions when available.
This removes a couple tablegen classes that become unused after this change. Another class gained an additional parameter to allow PMADDUBSW to specify a different result type from its input type.
llvm-svn: 285515
Hongbin Zheng [Sun, 30 Oct 2016 06:07:59 +0000 (06:07 +0000)]
[Polly] Remove the unused POLLY_LINK_LIBS for linking polly into
tools
Differential Revision: https://reviews.llvm.org/D25861
llvm-svn: 285514
Teresa Johnson [Sun, 30 Oct 2016 05:40:44 +0000 (05:40 +0000)]
[ThinLTO] Use per-summary flag to prevent exporting locals used in inline asm
Summary:
Instead of using the workaround of suppressing the entire index for
modules that call inline asm that may reference locals, use the
NoRename flag on the summary for any locals in the llvm.used set, and
add a reference edge from any functions containing inline asm.
This avoids issues from having no summaries despite the module defining
global values, which was preventing more aggressive index-based
optimization. It will be followed by a subsequent patch to make a
similar fix for local references in module level asm (to fix PR30610).
Reviewers: mehdi_amini
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D26121
llvm-svn: 285513
Teresa Johnson [Sun, 30 Oct 2016 05:15:23 +0000 (05:15 +0000)]
[ThinLTO] Correctly resolve linkonce when importing aliasee
Summary:
When we have an aliasee that is linkonce, while we can't convert
the non-prevailing copies to available_externally, we still need to
convert the prevailing copy to weak. If a reference to the aliasee
is exported, not converting a copy to weak will result in undefined
references when the linkonce is removed in its original module.
Add a new test and update existing tests.
Reviewers: mehdi_amini
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D26076
llvm-svn: 285512
NAKAMURA Takumi [Sun, 30 Oct 2016 02:58:48 +0000 (02:58 +0000)]
clang/test/Driver/openmp-offload.c: Relax expressions if "ld.exe" exists, like mingw.
llvm-svn: 285511
Craig Topper [Sun, 30 Oct 2016 00:02:55 +0000 (00:02 +0000)]
[X86] Don't use loadv2i64 on SSE version of PMULHRSW. Use memopv2i64 instead.
This bug was introduced in r285501.
llvm-svn: 285510
NAKAMURA Takumi [Sat, 29 Oct 2016 23:14:18 +0000 (23:14 +0000)]
NativeFormatting.cpp: Fix build for mingw. Where would writePadding() be?
llvm-svn: 285509
Teresa Johnson [Sat, 29 Oct 2016 21:52:23 +0000 (21:52 +0000)]
[ThinLTO] Rename doPromoteLocalToGlobal to shouldPromoteLocalToGlobal (NFC)
Rename as suggested in code review for D26063.
llvm-svn: 285508
Teresa Johnson [Sat, 29 Oct 2016 21:31:48 +0000 (21:31 +0000)]
[ThinLTO] Use NoPromote flag in summary during promotion
Summary:
Replace the check of whether a GV has a section with the flag check
in the summary. This is in preparation for using the NoPromote flag
to convey other situations when we can't promote (e.g. locals used in
inline asm).
Reviewers: mehdi_amini
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D26063
llvm-svn: 285507
Peter Collingbourne [Sat, 29 Oct 2016 20:57:12 +0000 (20:57 +0000)]
IR: Remove a no longer needed assert.
This assert was checking for a miscompile in a version of GCC that
we no longer support.
llvm-svn: 285506
Craig Topper [Sat, 29 Oct 2016 19:02:14 +0000 (19:02 +0000)]
[AVX-512] Remove 128/256-bit masked pmulhrsw/pmulhuw/pmulhw builtins and use unmasked builtins and select instead.
llvm-svn: 285505
Craig Topper [Sat, 29 Oct 2016 19:02:10 +0000 (19:02 +0000)]
[AVX-512] Remove masked 128/256-bit sqrt builtins and replace them with unmasked builtins and a select.
llvm-svn: 285504
Craig Topper [Sat, 29 Oct 2016 19:02:07 +0000 (19:02 +0000)]
[AVX-512] Remove masked 128/256-bit pmuludq/pmuldq builtins and replace them with unmasked builtins and a select.
llvm-svn: 285503
Craig Topper [Sat, 29 Oct 2016 19:02:03 +0000 (19:02 +0000)]
[AVX-512] Remove masked 128/256-bit floating point max/min builtins. Use unmasked builtins with select instead.
llvm-svn: 285502
Craig Topper [Sat, 29 Oct 2016 18:41:45 +0000 (18:41 +0000)]
[X86] Use intrinsics table for VPMULHRSW intrincis so that the legacy intrinsics can select EVEX encoded instructions when available.
This requires a minor rename of the instructions due to the use of different tablegen classes and how the names are concatenated.
llvm-svn: 285501
Richard Smith [Sat, 29 Oct 2016 17:28:48 +0000 (17:28 +0000)]
Factor finding of libc++ include path out of building -cc1 arguments.
llvm-svn: 285500
Sanjay Patel [Sat, 29 Oct 2016 16:21:19 +0000 (16:21 +0000)]
[ValueTracking] recognize more variants of smin/smax
Try harder to detect obfuscated min/max patterns: the initial pattern was added with D9352 / rL236202.
There was a bug fix for PR27137 at rL264996, but I think we can do better by folding the corresponding
smax pattern and commuted variants.
The codegen tests demonstrate the effect of ValueTracking on the backend via SelectionDAGBuilder. We
can't expose these differences minimally in IR because we don't have smin/smax intrinsics for IR.
Differential Revision: https://reviews.llvm.org/D26091
llvm-svn: 285499
Sanjay Patel [Sat, 29 Oct 2016 16:02:57 +0000 (16:02 +0000)]
[x86] add tests for smin/smax matchSelPattern (D26091)
llvm-svn: 285498
Piotr Padlewski [Sat, 29 Oct 2016 15:28:30 +0000 (15:28 +0000)]
[Devirtualization] Decorate vfunction load with invariant.load
Summary:
This patch was introduced one year ago, but because my google account
was disabled, I didn't get email with failing buildbot and I missed
revert of this commit. There was small but in test regex.
I am back.
Reviewers: rsmith, rengolin
Subscribers: nlewycky, rjmccall, cfe-commits
Differential Revision: https://reviews.llvm.org/D26117
llvm-svn: 285497
Piotr Padlewski [Sat, 29 Oct 2016 15:28:25 +0000 (15:28 +0000)]
NFC small format
llvm-svn: 285496
Sanjay Patel [Sat, 29 Oct 2016 15:22:04 +0000 (15:22 +0000)]
[InstCombine] re-use bitcasted compare operands in selects (PR28001)
These mixed bitcast patterns show up with SSE/AVX intrinsics because we bitcast function parameters to <2 x i64>.
The bitcasts obfuscate the expected min/max forms as shown in PR28001:
https://llvm.org/bugs/show_bug.cgi?id=28001#c6
Differential Revision: https://reviews.llvm.org/D25943
llvm-svn: 285495
Simon Pilgrim [Sat, 29 Oct 2016 11:29:39 +0000 (11:29 +0000)]
[DAGCombiner] (REAPPLIED) Add vector demanded elements support to computeKnownBits
Currently computeKnownBits returns the common known zero/one bits for all elements of vector data, when we may only be interested in one/some of the elements.
This patch adds a DemandedElts argument that allows us to specify the elements we actually care about. The original computeKnownBits implementation calls with a DemandedElts demanding all elements to match current behaviour. Scalar types set this to 1.
The approach was found to be easier than trying to add a per-element known bits solution, for a similar usefulness given the combines where computeKnownBits is typically used.
I've only added support for a few opcodes so far (the ones that have proven straightforward to test), all others will default to demanding all elements but can be updated in due course.
DemandedElts support could similarly be added to computeKnownBitsForTargetNode in a future commit.
This looked like this had caused compile time regressions on some buildbots (and was reverted in rL285381), but appears to have just been a harmless bystander!
Differential Revision: https://reviews.llvm.org/D25691
llvm-svn: 285494
Michael Zuckerman [Sat, 29 Oct 2016 10:29:20 +0000 (10:29 +0000)]
[X86][AVX512][Clang][Intrinsics][reduce] Adding missing reduce (max|min) intrinsics to Clang .
After LGTM and Check-all
Vector-reduction arithmetic accepts vectors as inputs and produces
scalars as outputs.This class of vector operation forms the basis
of many scientific computations. In vector-reduction arithmetic,
the evaluation off is independent of the order of the input elements of V.
Reviewer: 1. craig.topper
2. igorb
Differential Revision: https://reviews.llvm.org/D25988
llvm-svn: 285493
Elena Demikhovsky [Sat, 29 Oct 2016 08:44:46 +0000 (08:44 +0000)]
Fixed FMA + FNEG combine.
Masked form of FMA should be omitted in this optimization.
Differential Revision: https://reviews.llvm.org/D25984
llvm-svn: 285492
Tobias Grosser [Sat, 29 Oct 2016 06:19:34 +0000 (06:19 +0000)]
[ScopDetect] Use SCEVRewriteVisitor to simplify SCEVRemoveSMax rewriter
ScalarEvolution got at some pointer a SCEVRewriteVisitor. Use it to simplify
our SCEVRemoveSMax visitor.
llvm-svn: 285491
Matt Arsenault [Sat, 29 Oct 2016 04:05:06 +0000 (04:05 +0000)]
AMDGPU: Use 1/2pi inline imm on VI
I'm guessing at how it is supposed to be printed
llvm-svn: 285490
Rui Ueyama [Sat, 29 Oct 2016 01:28:06 +0000 (01:28 +0000)]
Attempt to fix a buildbot really this time.
llvm-svn: 285489
Matthias Braun [Sat, 29 Oct 2016 01:03:41 +0000 (01:03 +0000)]
AArch64DeadRegisterDefinitionsPass: Cleanup; NFC
- Fix doxygen file comment
- reduce indentation in loop
- Factor out some common subexpressions
- Move independent helper function out of class
- Fix Changed flag (this is not strictly NFC but a bugfix, but the flag
seems ignored anyway)
llvm-svn: 285488
Rui Ueyama [Sat, 29 Oct 2016 00:56:44 +0000 (00:56 +0000)]
Define calculateDbgStreamSize for consistency.
llvm-svn: 285487
Tim Shen [Sat, 29 Oct 2016 00:51:41 +0000 (00:51 +0000)]
[APFloat] Remove the redundent function body of uninitialized ctor, which should be done in r285468
llvm-svn: 285486
Mehdi Amini [Sat, 29 Oct 2016 00:50:02 +0000 (00:50 +0000)]
Revert "Added 'inline' attribute to basic_string's destructor"
This reverts commit r285456, which broke LTO bootstrap on Darwin.
llvm-svn: 285485
Todd Fiala [Sat, 29 Oct 2016 00:29:15 +0000 (00:29 +0000)]
Limit LLDB_EXPORT_ALL_SYMBOLS to lldb symbols
LLDB_EXPORT_ALL_SYMBOLS used to instruct the build to export all
the symbols in liblldb on CMake builds. This change limits the
CMake define to only add in the lldb_private namespace to the
symbols that normally get exported, such that we export all the
symbols in the public lldb namespace and the lldb_private namespace.
This is a fix for:
https://llvm.org/bugs/show_bug.cgi?id=30822
Reviewers: labath, beanz
Subscribers: lldb-commits
Differential Revision: https://reviews.llvm.org/D26093
llvm-svn: 285484
Zachary Turner [Sat, 29 Oct 2016 00:27:22 +0000 (00:27 +0000)]
Resubmit "Add support for advanced number formatting."
This resubmits r284436 and r284437, which were reverted in
r284462 as they were breaking the AArch64 buildbot.
The breakage on AArch64 turned out to be a miscompile which is
still not fixed, but is actively tracked at llvm.org/pr30748.
This resubmission re-writes the code in a way so as to make the
miscompile not happen.
llvm-svn: 285483
Anna Zaks [Sat, 29 Oct 2016 00:27:07 +0000 (00:27 +0000)]
[asan] Set the darwin deployment target for the dead-strip test
ASan dead-strip support relies on a linker option that only exists
in 10.11 and later, so the LLVM instrumentation checks for the deployment
target. This test does not pass when clang is built to choose lower
deployment target by default but runs on newer host.
(Note, the REQUIRES: osx-ld64-live_support clause only checks the host
and not the target OS.)
Differential Revision: https://reviews.llvm.org/D26107
llvm-svn: 285482
Rui Ueyama [Fri, 28 Oct 2016 23:57:37 +0000 (23:57 +0000)]
Do not print out Flags field twice.
llvm-svn: 285481
Davide Italiano [Fri, 28 Oct 2016 23:55:32 +0000 (23:55 +0000)]
[DAGCombiner] Fix a crash visiting `AND` nodes.
Instead of asserting that the shift count is != 0 we just bail out
as it's not profitable trying to optimize a node which will be
removed anyway.
Differential Revision: https://reviews.llvm.org/D26098
llvm-svn: 285480
Tom Stellard [Fri, 28 Oct 2016 23:53:48 +0000 (23:53 +0000)]
AMDGPU/SI: Don't use non-0 waitcnt values when waiting on Flat instructions
Summary:
Flat instruction can return out of order, so we need always need to wait
for all the outstanding flat operations.
Reviewers: tony-tye, arsenm
Subscribers: kzhuravl, wdng, nhaehnle, llvm-commits, yaxunl
Differential Revision: https://reviews.llvm.org/D25998
llvm-svn: 285479