Simon Atanasyan [Wed, 19 Nov 2014 05:50:59 +0000 (05:50 +0000)]
[Mips] Remove redundant const_cast
No functional changes.
llvm-svn: 222322
Simon Atanasyan [Wed, 19 Nov 2014 05:50:53 +0000 (05:50 +0000)]
[Mips] Make the function return type constant pointer
No functional changes.
llvm-svn: 222321
Simon Atanasyan [Wed, 19 Nov 2014 05:50:47 +0000 (05:50 +0000)]
[Mips] Move setup of LLD_R_MIPS_STO_PLT relocation to the getPLTEntry() routine
No functional changes.
llvm-svn: 222320
David Blaikie [Wed, 19 Nov 2014 05:49:42 +0000 (05:49 +0000)]
Remove StringMap::GetOrCreateValue in favor of StringMap::insert
Having two ways to do this doesn't seem terribly helpful and
consistently using the insert version (which we already has) seems like
it'll make the code easier to understand to anyone working with standard
data structures. (I also updated many references to the Entry's
key and value to use first() and second instead of getKey{Data,Length,}
and get/setValue - for similar consistency)
Also removes the GetOrCreateValue functions so there's less surface area
to StringMap to fix/improve/change/accommodate move semantics, etc.
llvm-svn: 222319
Eric Fiselier [Wed, 19 Nov 2014 05:49:03 +0000 (05:49 +0000)]
Overhaul and separate nullptr_t tests to pass with C++03.
The standard requires that nullptr_t can be reinterpret_cast to an integral type
at least the size of nullptr_t. There is no way to emulate this conversion in
the C++03 nullptr_t implementation. The test for this conversion has been moved
to a new test and marked XFAIL with c++03.
This recommits what was originally r222296.
llvm-svn: 222318
David Blaikie [Wed, 19 Nov 2014 05:48:40 +0000 (05:48 +0000)]
clang-format a recent commit I made
llvm-svn: 222317
Eric Fiselier [Wed, 19 Nov 2014 05:41:29 +0000 (05:41 +0000)]
Revert r222296 to fix bad commit message
llvm-svn: 222316
Nico Weber [Wed, 19 Nov 2014 05:21:39 +0000 (05:21 +0000)]
Wrap to 80 columns, no behavior change.
llvm-svn: 222315
Eric Fiselier [Wed, 19 Nov 2014 05:00:57 +0000 (05:00 +0000)]
Remove accidentally commited line in commit. Sorry
llvm-svn: 222314
Shankar Easwaran [Wed, 19 Nov 2014 03:51:50 +0000 (03:51 +0000)]
[ELF] Handle ctors/dtors sections
Move functionality from MIPS target to DefaultLayout.
No change in functionality.
llvm-svn: 222313
Shankar Easwaran [Wed, 19 Nov 2014 03:51:48 +0000 (03:51 +0000)]
[ELF] Fix creation of segments.
Linker was creating a separate output segment in some cases if input sections
had huge alignments. This patch fixes the issue.
llvm-svn: 222312
Shankar Easwaran [Wed, 19 Nov 2014 03:51:45 +0000 (03:51 +0000)]
[ELF] Create input and output section names
No change in functionality.
llvm-svn: 222311
Shankar Easwaran [Wed, 19 Nov 2014 03:51:43 +0000 (03:51 +0000)]
[ELF] Rename MergedSection to OutputSection.
No change in functionality.
llvm-svn: 222310
Shankar Easwaran [Wed, 19 Nov 2014 03:51:41 +0000 (03:51 +0000)]
[ELF] Sectionkey should also contain path.
No change in functionality.
llvm-svn: 222309
Peter Collingbourne [Wed, 19 Nov 2014 03:34:20 +0000 (03:34 +0000)]
Teach llvm-build to avoid touching LibraryDependencies.inc unless the contents
change. This saves us from rebuilding llvm-config each time we reconfigure.
llvm-svn: 222308
Peter Collingbourne [Wed, 19 Nov 2014 03:34:17 +0000 (03:34 +0000)]
Expose LLVM version string via macro in llvm-config.h, and modify Go bindings
to make use of it.
llvm-svn: 222307
David Blaikie [Wed, 19 Nov 2014 03:06:06 +0000 (03:06 +0000)]
Standardize on StringMap::insert, removing uses of StringMap::GetOrCreateValue.
llvm-svn: 222306
David Blaikie [Wed, 19 Nov 2014 03:05:07 +0000 (03:05 +0000)]
Remove uses of StringMap::GetOrCreateValue in favor of stl-compatible API usage
llvm-svn: 222305
David Blaikie [Wed, 19 Nov 2014 03:04:29 +0000 (03:04 +0000)]
Update for LLVM API change
llvm-svn: 222304
David Blaikie [Wed, 19 Nov 2014 02:56:13 +0000 (02:56 +0000)]
Update for LLVM API change
llvm-svn: 222303
David Blaikie [Wed, 19 Nov 2014 02:56:00 +0000 (02:56 +0000)]
Make StringSet::insert return pair<iterator, bool> like other self-associative containers
StringSet is still a bit dodgy in that it exposes the raw iterator of
the StringMap parent, which exposes the weird detail that StringSet
actually has a 'value'... but anyway, this is useful for a handful of
clients that want to reference the newly inserted/persistent string data
in the StringSet/Map/Entry/thing.
llvm-svn: 222302
Jason Molenda [Wed, 19 Nov 2014 02:29:52 +0000 (02:29 +0000)]
Add additional checks to the SavedLocationForRegister method
where it is retrieving the Return Address register contents
on a target where that's a thing. If we fail to get a valid
RA, we force a switch to the fallback unwind plan. This patch
adds a sanity check for that fallback unwind plan -- it must
get a valid CFA for this frame in addition to being able to
retrieve the caller's PC -- and it correctly marks the unwind
rules as failing if the fallback unwind plan fails.
<rdar://problem/
19010211>
llvm-svn: 222301
Nick Kledzik [Wed, 19 Nov 2014 02:21:53 +0000 (02:21 +0000)]
[mach-o] propagate dylib version numbers
Mach-o does not use a simple SO_NEEDED to track dependent dylibs. Instead,
the linker copies four things from each dylib to each client: the runtime path
(aka "install name"), the build time, current version (dylib build number), and
compatibility version The build time is no longer used (it cause every rebuild
of a dylib to be different). The compatibility version is usually just 1.0
and never changes, or the dylib becomes incompatible.
This patch copies that information into the NormalizedMachO format and
propagates it to clients.
llvm-svn: 222300
Rui Ueyama [Wed, 19 Nov 2014 02:07:10 +0000 (02:07 +0000)]
llvm-readobj: fix off-by-one error in COFFDumper
It printed out base relocation table header as table entry.
This patch also makes llvm-readobj to not skip ABSOLUTE entries
becuase it was confusing.
llvm-svn: 222299
Eric Fiselier [Wed, 19 Nov 2014 01:45:12 +0000 (01:45 +0000)]
Cleanup quick_exit tests and get them passing in C++03.
Wrap the original test in _LIBCPP_HAS_QUICK_EXIT so it only runs when we have
quick_exit and add two new tests that check that when _LIBCPP_HAS_QUICK_EXIT
is not defined then no definition of std::at_quick_exit or std::quick_exit are
available.
llvm-svn: 222298
Kuba Brecka [Wed, 19 Nov 2014 01:31:59 +0000 (01:31 +0000)]
Make the ASan OS X DYLD_INSERT_LIBRARIES detection path-independent
Reviewed at http://reviews.llvm.org/D6238
ASan on Darwin during launch reads DYLD_INSERT_LIBRARIES env. variable and if it's not set or if the ASan dylib is not present in there, it relaunches the process. The check whether the dylib is present in the variable is now trying to find a full path in there. This fails in the scenarios where we want to copy the dylib to the executable's directory or somewhere else and set the DYLD_INSERT_LIBRARIES manually, see http://reviews.llvm.org/D6018.
Let's change the search in DYLD_INSERT_LIBRARIES to only look for the filename of the dylib and not the full path.
llvm-svn: 222297
Eric Fiselier [Wed, 19 Nov 2014 01:31:56 +0000 (01:31 +0000)]
diff --git a/test/language.support/support.types/nullptr_t.pass.cpp b/test/language.support/support.types/nullptr_t.pass.cpp
index 6c15fef..4d7c8b0 100644
--- a/test/language.support/support.types/nullptr_t.pass.cpp
+++ b/test/language.support/support.types/nullptr_t.pass.cpp
@@ -18,42 +18,62 @@ struct A
A(std::nullptr_t) {}
};
+template <class T>
+void test_conversions()
+{
+ {
+ T p = 0;
+ assert(p == nullptr);
+ }
+ {
+ T p = nullptr;
+ assert(p == nullptr);
+ assert(nullptr == p);
+ assert(!(p != nullptr));
+ assert(!(nullptr != p));
+ }
+}
+
+template <class T>
+void test_comparisons()
+{
+ T p = nullptr;
+ assert(p == nullptr);
+ assert(p <= nullptr);
+ assert(p >= nullptr);
+ assert(!(p != nullptr));
+ assert(!(p < nullptr));
+ assert(!(p > nullptr));
+ assert(nullptr == p);
+ assert(nullptr <= p);
+ assert(nullptr >= p);
+ assert(!(nullptr != p));
+ assert(!(nullptr < p));
+ assert(!(nullptr > p));
+}
+
+
int main()
{
static_assert(sizeof(std::nullptr_t) == sizeof(void*),
"sizeof(std::nullptr_t) == sizeof(void*)");
- A* p = 0;
- assert(p == nullptr);
- void (A::*pmf)() = 0;
-#ifdef __clang__
- // GCC 4.2 can't handle this
- assert(pmf == nullptr);
-#endif
- int A::*pmd = 0;
- assert(pmd == nullptr);
- A a1(nullptr);
- A a2(0);
- bool b = nullptr;
- assert(!b);
- assert(nullptr == nullptr);
- assert(nullptr <= nullptr);
- assert(nullptr >= nullptr);
- assert(!(nullptr != nullptr));
- assert(!(nullptr < nullptr));
- assert(!(nullptr > nullptr));
- A* a = nullptr;
- assert(a == nullptr);
- assert(a <= nullptr);
- assert(a >= nullptr);
- assert(!(a != nullptr));
- assert(!(a < nullptr));
- assert(!(a > nullptr));
- assert(nullptr == a);
- assert(nullptr <= a);
- assert(nullptr >= a);
- assert(!(nullptr != a));
- assert(!(nullptr < a));
- assert(!(nullptr > a));
- std::ptrdiff_t i = reinterpret_cast<std::ptrdiff_t>(nullptr);
- assert(i == 0);
+
+ {
+ test_conversions<std::nullptr_t>();
+ test_conversions<void*>();
+ test_conversions<A*>();
+ test_conversions<void(*)()>();
+ test_conversions<void(A::*)()>();
+ test_conversions<int A::*>();
+ }
+ {
+ test_comparisons<std::nullptr_t>();
+ test_comparisons<void*>();
+ test_comparisons<A*>();
+ test_comparisons<void(*)()>();
+ }
+ {
+ bool b = nullptr;
+ assert(!b);
+ }
}
diff --git a/test/language.support/support.types/nullptr_t_integral_cast.fail.cpp b/test/language.support/support.types/nullptr_t_integral_cast.fail.cpp
new file mode 100644
index 0000000..92bd879
--- /dev/null
+++ b/test/language.support/support.types/nullptr_t_integral_cast.fail.cpp
@@ -0,0 +1,17 @@
+//===----------------------------------------------------------------------===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is dual licensed under the MIT and the University of Illinois Open
+// Source Licenses. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// typedef decltype(nullptr) nullptr_t;
+
+#include <cstddef>
+
+int main()
+{
+ std::ptrdiff_t i = static_cast<std::ptrdiff_t>(nullptr);
+}
diff --git a/test/language.support/support.types/nullptr_t_integral_cast.pass.cpp b/test/language.support/support.types/nullptr_t_integral_cast.pass.cpp
new file mode 100644
index 0000000..34c7a93
--- /dev/null
+++ b/test/language.support/support.types/nullptr_t_integral_cast.pass.cpp
@@ -0,0 +1,24 @@
+//===----------------------------------------------------------------------===//
+//
+// The LLVM Compiler Infrastructure
+//
+// This file is dual licensed under the MIT and the University of Illinois Open
+// Source Licenses. See LICENSE.TXT for details.
+//
+//===----------------------------------------------------------------------===//
+
+// NOTE: nullptr_t emulation cannot handle a reinterpret_cast to an
+// integral type
+// XFAIL: c++98, c++03
+
+// typedef decltype(nullptr) nullptr_t;
+
+
+#include <cstddef>
+#include <cassert>
+
+int main()
+{
+ std::ptrdiff_t i = reinterpret_cast<std::ptrdiff_t>(nullptr);
+ assert(i == 0);
+}
llvm-svn: 222296
Jim Ingham [Wed, 19 Nov 2014 01:28:13 +0000 (01:28 +0000)]
Add "-k" and "-K" options to the driver, that allow you to register
some commands that will get run if the target crashes.
Also fix the bug where the local .lldbinit file was not getting
sourced before not after the target was created from the file options on the
driver command line.
<rdar://problem/
19019843>
llvm-svn: 222295
Rafael Espindola [Wed, 19 Nov 2014 01:02:22 +0000 (01:02 +0000)]
Use a memcpy so that type based alias analysis sees the change.
The other option would be to do something like
if (that.isSingleWord())
VAL = that.VAL;
else
pVal = that.pVal
This bug was causing 86TTI::getIntImmCost to be miscompiled in a LTO
bootstrap in stage2, causing the build of stage3 to fail.
LLVM is getting quiet good at exploiting this. Not sure if there is anything
a sanitizer could do to help
llvm-svn: 222294
Rui Ueyama [Wed, 19 Nov 2014 00:48:06 +0000 (00:48 +0000)]
Fix MSVC warning.
This patch fixes the following MSVC warning.
warning C4334: '<<' : result of 32-bit shift implicitly
converted to 64 bits (was 64-bit shift intended?)
llvm-svn: 222293
Weiming Zhao [Wed, 19 Nov 2014 00:29:14 +0000 (00:29 +0000)]
[Aarch64] Customer lowering of CTPOP to SIMD should check for NEON availability
llvm-svn: 222292
Kostya Serebryany [Wed, 19 Nov 2014 00:24:11 +0000 (00:24 +0000)]
[asan] initial support for experimental basic-block tracing; also add tests for various levels of -fsanitize-coverage
llvm-svn: 222291
Kostya Serebryany [Wed, 19 Nov 2014 00:22:58 +0000 (00:22 +0000)]
[asan] add experimental basic-block tracing to asan-coverage; also fix -fsanitize-coverage=3 which was broken by r221718
llvm-svn: 222290
Rui Ueyama [Wed, 19 Nov 2014 00:18:07 +0000 (00:18 +0000)]
llvm-readobj: teach it how to dump COFF base relocation table
llvm-svn: 222289
Kostya Serebryany [Wed, 19 Nov 2014 00:17:31 +0000 (00:17 +0000)]
Introduce llvm::SplitAllCriticalEdges
Summary:
move the code from BreakCriticalEdges::runOnFunction()
into a separate utility function llvm::SplitAllCriticalEdges()
so that it can be used independently.
No functionality change intended.
Test Plan: check-llvm
Reviewers: nlewycky
Reviewed By: nlewycky
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D6313
llvm-svn: 222288
Manman Ren [Wed, 19 Nov 2014 00:13:26 +0000 (00:13 +0000)]
Revert r222039 because of bot failure.
http://lab.llvm.org:8080/green/job/clang-Rlto_master/298/
Hopefully, bot will be green. If not, we will re-submit the commit.
llvm-svn: 222287
Matt Arsenault [Wed, 19 Nov 2014 00:01:31 +0000 (00:01 +0000)]
R600/SI: Implement areMemAccessesTriviallyDisjoint
This partially makes up for not having address spaces
used for alias analysis in some simple cases.
This is not yet enabled by default so shouldn't change anything yet.
llvm-svn: 222286
Matt Arsenault [Tue, 18 Nov 2014 23:57:33 +0000 (23:57 +0000)]
R600/SI: Set hasSideEffects = 0 on load and store instructions.
Assuming unmodeled side effects interferes with some scheduling
opportunities.
Don't put it in the base class of DS instructions since there
are a few weird effecting, non load/store instructions there.
llvm-svn: 222285
Daniel Jasper [Tue, 18 Nov 2014 23:55:27 +0000 (23:55 +0000)]
clang-format: Add option to disable alignment after opening brackets
Before:
SomeFunction(parameter,
parameter);
After:
SomeFunction(parameter,
parameter);
Patch by Harry Terkelsen, thank you!
llvm-svn: 222284
Daniel Jasper [Tue, 18 Nov 2014 23:48:01 +0000 (23:48 +0000)]
clang-format: Fix space between generic type parameter and square
bracket
Before:
public Foo<X, Y> [] foos;
After:
public Foo<X, Y>[] foos;
Patch by Harry Terkelsen. Thank you!
llvm-svn: 222283
Eric Fiselier [Tue, 18 Nov 2014 23:46:18 +0000 (23:46 +0000)]
Modify tests to check that va_copy is only defined in C++11 and beyond.
llvm-svn: 222282
Simon Pilgrim [Tue, 18 Nov 2014 23:38:19 +0000 (23:38 +0000)]
[X86][AVX] 256-bit vector stack unaligned load/stores identification
Under many circumstances the stack is not 32-byte aligned, resulting in the use of the vmovups/vmovupd/vmovdqu instructions when inserting ymm reloads/spills.
This minor patch adds these instructions to the isFrameLoadOpcode/isFrameStoreOpcode helpers so that they can be correctly identified and not be treated as folded reloads/spills.
This has also been noticed by http://llvm.org/bugs/show_bug.cgi?id=18846 where it was causing redundant spills - I've added a reduced test case at test/CodeGen/X86/pr18846.ll
Differential Revision: http://reviews.llvm.org/D6252
llvm-svn: 222281
Enrico Granata [Tue, 18 Nov 2014 23:36:25 +0000 (23:36 +0000)]
Shuffle APIs around a little bit, so that if you pass custom summary options, we don't end up caching the summary hence obtained. You may want to obtain an uncapped summary, but this should not be reflected in the summary we cache. The drawback is that we don't cache as aggressively as we could, but at least you get to have different summaries with different options without having to reset formatters or the SBValue at each step
llvm-svn: 222280
Justin Bogner [Tue, 18 Nov 2014 23:03:14 +0000 (23:03 +0000)]
profile: Robustify instrprof tests
Change these tests not to rely on the exact metadata numbers the
profile data gets.
llvm-svn: 222279
Eric Fiselier [Tue, 18 Nov 2014 23:01:57 +0000 (23:01 +0000)]
Flush out test cases for tuples constructor SFINAE
llvm-svn: 222278
Enrico Granata [Tue, 18 Nov 2014 22:54:45 +0000 (22:54 +0000)]
Add the ability for the NSString and libc++ std::string formatters to retrieve uncapped data
llvm-svn: 222277
Enrico Granata [Tue, 18 Nov 2014 22:47:33 +0000 (22:47 +0000)]
Allow dsymutil to exists in a path with spaces in it
llvm-svn: 222276
Ismail Pazarbasi [Tue, 18 Nov 2014 22:45:59 +0000 (22:45 +0000)]
Revert "git-svn test commit"
This reverts commit
aa8d370ee798f75bc05a1ae9a240bc3e5b6870ac.
llvm-svn: 222275
Colin LeMahieu [Tue, 18 Nov 2014 22:45:47 +0000 (22:45 +0000)]
[Hexagon] Adding A2_and instruction.
llvm-svn: 222274
Ismail Pazarbasi [Tue, 18 Nov 2014 22:44:51 +0000 (22:44 +0000)]
git-svn test commit
llvm-svn: 222273
Chad Rosier [Tue, 18 Nov 2014 22:41:49 +0000 (22:41 +0000)]
[FastISel][AArch64] Also allow folding of sign-/zero-extend and arithmetic
shift-right for booleans (i1).
Arithmetic shift-right immediate with sign-/zero-extensions also works for
boolean values. Update the assert and the test cases to reflect that fact.
llvm-svn: 222272
Eric Christopher [Tue, 18 Nov 2014 22:40:27 +0000 (22:40 +0000)]
More override warning cleanup.
llvm-svn: 222271
Chad Rosier [Tue, 18 Nov 2014 22:38:42 +0000 (22:38 +0000)]
[FastISel][AArch64] Also allow folding of sign-/zero-extend and logical
shift-right for booleans (i1).
Logical shift-right immediate with sign-/zero-extensions also works for boolean
values. Update the assert and the test cases to reflect that fact.
llvm-svn: 222270
Eric Christopher [Tue, 18 Nov 2014 22:36:15 +0000 (22:36 +0000)]
Fix 80-column violations, trailing whitespace, comment text.
llvm-svn: 222269
Eric Christopher [Tue, 18 Nov 2014 22:36:11 +0000 (22:36 +0000)]
This test also requires an aarch64 target.
llvm-svn: 222268
Justin Bogner [Tue, 18 Nov 2014 22:17:30 +0000 (22:17 +0000)]
Fix an inaccurate comment
llvm-svn: 222267
Nick Kledzik [Tue, 18 Nov 2014 22:08:50 +0000 (22:08 +0000)]
[mach-o] for 32-bit arm, in -r mode support B (not BL) to undefined symbol
When fixing up BL instructions, the linker has to compare the thumbness of the
target to decide if the instruction needs to be converted to BLX. But with B
instruction there is no BX, so the linker asserts if the target is not the
same thumbness. This assert was firing in -r mode when the target was undefined
which it interpreted as being non-thumb.
Test case change is to add a B (in both thumb and arm code) to an undefined
symbol and round trip through -r mode.
llvm-svn: 222266
David Majnemer [Tue, 18 Nov 2014 22:06:45 +0000 (22:06 +0000)]
InstCombine: Fix another infinite loop caused by visitFPTrunc
We would attempt to replace an frem's operand with the same operand.
This would cause InstCombine to think real work was done, causing
InstCombine to enter an infinite loop.
This fixes the second part of PR21576.
llvm-svn: 222265
Fariborz Jahanian [Tue, 18 Nov 2014 21:57:54 +0000 (21:57 +0000)]
[Sema] Patch to issue warning on comparing parameters with
nonnull attribute when comparison is always true/false.
Original patch by Steven Wu. I have added extra code to prevent issuing of
warning when the nonnull parameter is modified prior to the comparison.
This addition prevents false positives in the most obvious cases.
There may still be false positive warnings in some cases (as one of my tests
indicates), but benefit far outweighs such cases. rdar://
18712242
llvm-svn: 222264
Colin LeMahieu [Tue, 18 Nov 2014 21:51:51 +0000 (21:51 +0000)]
[Hexagon] Adding A2_sub instruction
Renaming test files.
llvm-svn: 222263
Ismail Pazarbasi [Tue, 18 Nov 2014 21:46:06 +0000 (21:46 +0000)]
Find SWIG with CMake
SWIG is searched under certain paths within python script. CMake can
detect SWIG with find_package(SWIG). This is used iff user checks
LLDB_ENABLE_PYTHON_SCRIPTS_SWIG_API_GENERATION. If
buildSwigWrapperClasses.py does not receive swigExecutable argument,
then the script will use its current search implementation.
llvm-svn: 222262
David Majnemer [Tue, 18 Nov 2014 21:30:02 +0000 (21:30 +0000)]
Revert "Revert r222040 because of bot failure."
This reverts commit r222203, reverting r222040 didn't end up turning the
bot green.
llvm-svn: 222261
Eric Fiselier [Tue, 18 Nov 2014 21:28:49 +0000 (21:28 +0000)]
Add support for LLVM_USE_SANITIZER=Thread to test suite
llvm-svn: 222260
Eric Fiselier [Tue, 18 Nov 2014 21:26:45 +0000 (21:26 +0000)]
Add support for LLVM_USE_SANITIZER=Thread
llvm-svn: 222259
Eric Fiselier [Tue, 18 Nov 2014 21:23:38 +0000 (21:23 +0000)]
Add support for LLVM_USE_SANITIZER=Thread
Summary: This will help in testing libc++ and libc++abi with tsan.
Reviewers: samsonov
Reviewed By: samsonov
Subscribers: samsonov, llvm-commits
Differential Revision: http://reviews.llvm.org/D6283
llvm-svn: 222258
Juergen Ributzka [Tue, 18 Nov 2014 21:20:17 +0000 (21:20 +0000)]
[FastISel][AArch64] Follow-up fix for "Fix shift-immediate emission for "zero" shifts."
Shifts also perform sign-/zero-extends to larger types, which requires us to emit
an integer extend instead of a simple COPY.
Related to PR21594.
llvm-svn: 222257
Matt Arsenault [Tue, 18 Nov 2014 21:06:58 +0000 (21:06 +0000)]
R600/SI: Move SIFixSGPRCopies to inst selector passes
This should expose more of the actually used VALU
instructions to the machine optimization passes.
This also should help getting i1 handling into a better state.
For not entirly understood reasons, this fixes the split-scalar-i64-add.ll
test where a 64-bit add would only partially be moved to the VALU
resulting in use of undefined VCC.
llvm-svn: 222256
Juergen Ributzka [Tue, 18 Nov 2014 21:02:40 +0000 (21:02 +0000)]
[AArch64] Don't optimize all compare instructions.
"optimizeCompareInstr" converts compares (cmp/cmn) into plain sub/add
instructions when the flags are not used anymore. This conversion is valid for
most instructions, but not all. Some instructions that don't set the flags
(e.g. sub with immediate) can set the SP, whereas the flag setting version uses
the same encoding for the "zero" register.
Update the code to also check for the return register before performing the
optimization to make sure that a cmp doesn't suddenly turn into a sub that sets
the stack pointer.
I don't have a test case for this, because it isn't easy to trigger.
llvm-svn: 222255
Owen Anderson [Tue, 18 Nov 2014 20:50:19 +0000 (20:50 +0000)]
Fix an incorrect chain operand when expanding INSERT_VECTOR operations through the stack.
Patch by Daniil Troshkov!
llvm-svn: 222254
Tom Stellard [Tue, 18 Nov 2014 20:39:39 +0000 (20:39 +0000)]
R600/SI: Make sure resource descriptors are always stored in SGPRs
llvm-svn: 222253
Eric Fiselier [Tue, 18 Nov 2014 20:37:53 +0000 (20:37 +0000)]
[libcxxabi] Refactor CMakeLists.txt's handling of compile and link flags to suppress warnings.
Summary:
This patch mirrors the recent change to libc++ found here http://reviews.llvm.org/D6277.
This fixes PR20395 (http://llvm.org/bugs/show_bug.cgi?id=20395).
Reviewers: jroelofs, mclow.lists, danalbert
Reviewed By: danalbert
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D6286
llvm-svn: 222252
Marshall Clow [Tue, 18 Nov 2014 20:37:47 +0000 (20:37 +0000)]
Marked LWG 2399 as complete. I committed a test for this earlier today.
llvm-svn: 222251
Chad Rosier [Tue, 18 Nov 2014 20:34:01 +0000 (20:34 +0000)]
[Reassociate] Use test cases that can actually be optimized to verify optional
flags are cleared. The reassociation pass was just reordering the leaf nodes
in the previous test cases.
llvm-svn: 222250
Colin LeMahieu [Tue, 18 Nov 2014 20:28:11 +0000 (20:28 +0000)]
[Hexagon] Converting from ADD_rr to A2_add which has encoding bits.
Adding test to show correct instruction selection and encoding.
llvm-svn: 222249
Chad Rosier [Tue, 18 Nov 2014 20:21:54 +0000 (20:21 +0000)]
[Reassociate] Rename local variable to not use same name as a member
variable. NFC.
llvm-svn: 222248
Juergen Ributzka [Tue, 18 Nov 2014 19:58:59 +0000 (19:58 +0000)]
[FastISel][AArch64] Fix shift-immediate emission for "zero" shifts.
This change emits a COPY for a shift-immediate with a "zero" shift value.
This fixes PR21594 where we emitted a shift instruction with an incorrect
immediate operand.
llvm-svn: 222247
Greg Clayton [Tue, 18 Nov 2014 19:45:23 +0000 (19:45 +0000)]
Fixed the stop hook test after recent editline changes.
llvm-svn: 222246
Ed Maste [Tue, 18 Nov 2014 19:30:13 +0000 (19:30 +0000)]
Add decorator for test that fails on FreeBSD after editline rework
llvm.org/21599
llvm-svn: 222245
Jozef Kolek [Tue, 18 Nov 2014 19:20:34 +0000 (19:20 +0000)]
Test commit to verify that commit access works.
llvm-svn: 222244
Jim Ingham [Tue, 18 Nov 2014 19:12:13 +0000 (19:12 +0000)]
Patch from dawn@burble.org to make the --silent-run do what it says, not the opposite of what it says.
llvm-svn: 222243
Marshall Clow [Tue, 18 Nov 2014 18:14:53 +0000 (18:14 +0000)]
Add a test for LWG issue #2399. We already implement this, but now we have a test as well.
llvm-svn: 222242
Philip Reames [Tue, 18 Nov 2014 17:46:32 +0000 (17:46 +0000)]
Tweak EarlyCSE to recognize series of dead stores
EarlyCSE is giving up on the current instruction immediately when it recognizes that the current instruction makes a previous store trivially dead. There's no reason to do this. Once the previous store has been deleted, it's perfectly legal to remember the value of the current store (for value forwarding) and the fact the store occurred (it could be dead too!).
Reviewed by: Hal
Differential Revision: http://reviews.llvm.org/D6301
llvm-svn: 222241
Marshall Clow [Tue, 18 Nov 2014 17:35:16 +0000 (17:35 +0000)]
Update status of LWG issues 2340, 2396 and 2401. In all three cases, these are things that we already do.
llvm-svn: 222240
Manman Ren [Tue, 18 Nov 2014 16:45:34 +0000 (16:45 +0000)]
Remove triple in testing case to recover an arm bot.
llvm-svn: 222239
Marshall Clow [Tue, 18 Nov 2014 16:15:00 +0000 (16:15 +0000)]
Since Eric poisoned the comma operator on all our test iterators, we no longer need 'comma_iterator'. Remove it from the test suite.
llvm-svn: 222238
NAKAMURA Takumi [Tue, 18 Nov 2014 15:05:01 +0000 (15:05 +0000)]
OMPAtomicDirective::Create(): Fix \param. [-Wdocumentation]
llvm-svn: 222237
Evgeniy Stepanov [Tue, 18 Nov 2014 14:28:57 +0000 (14:28 +0000)]
[asan] Re-enable zero_page_pc test on Android.
It's not failing anymore. Not sure what has changed (the bot was down for a
long time), but lets keep it this way.
llvm-svn: 222236
Jeroen Ketema [Tue, 18 Nov 2014 14:19:27 +0000 (14:19 +0000)]
Don't include <stddef.h>
Including a standard or system header isn't allowed in OpenCL.
The type "size_t" needs to be explicitely defined now.
v2: Use __SIZE_TYPE__ instead of unsigned int.
v3: Define ptrdiff_t and NULL.
Patch-by: Jean-Sébastien Pédron
Reviewed-by: Jeroen Ketema
Reviewed-by: Jan Vesely
llvm-svn: 222235
Evgeniy Stepanov [Tue, 18 Nov 2014 14:04:12 +0000 (14:04 +0000)]
[asan] Fix CMake CMP0046 warning in standalone build.
Make FileCheck an "imported dependency" in standalone build.
llvm-svn: 222234
NAKAMURA Takumi [Tue, 18 Nov 2014 12:23:19 +0000 (12:23 +0000)]
CallGraphTest.cpp: Remove invalid tests. ++S might step over F if S == F.
MSVC Runtime detects "Assertion failed: vector iterator not incrementable"
llvm-svn: 222233
Evgeniy Stepanov [Tue, 18 Nov 2014 10:33:15 +0000 (10:33 +0000)]
[msan] Remove MSanDR and supporting code.
MSanDR is a dynamic instrumentation tool that can instrument the code
(prebuilt libraries and such) that could not be instrumented at compile time.
This code is unused (to the best of our knowledge) and unmaintained, and
starting to bit-rot.
llvm-svn: 222232
Alexey Bataev [Tue, 18 Nov 2014 10:14:22 +0000 (10:14 +0000)]
[OPENMP] Additional processing of 'omp atomic read' directive.
According to OpenMP standard, Section 2.12.6, atomic Construct, '#pragma omp atomic read' is allowed to be used only for expression statements of form 'v = x;', where x and v (as applicable) are both l-value expressions with scalar type. Patch adds checks for it.
llvm-svn: 222231
David Majnemer [Tue, 18 Nov 2014 09:31:41 +0000 (09:31 +0000)]
InstCombine: Fold away tautological masked compares
It is impossible for (x & INT_MAX) == 0 && x == INT_MAX to ever be true.
While this sort of reasoning should normally live in InstSimplify,
the machinery that derives this result is not trivial to split out.
llvm-svn: 222230
David Majnemer [Tue, 18 Nov 2014 09:31:36 +0000 (09:31 +0000)]
InstCombine: Clean up foldLogOpOfMaskedICmps
No functional change intended.
llvm-svn: 222229
Dmitry Vyukov [Tue, 18 Nov 2014 06:44:43 +0000 (06:44 +0000)]
tsan: add description of AcquireGlobal function
llvm-svn: 222228
Justin Hibbits [Tue, 18 Nov 2014 06:17:20 +0000 (06:17 +0000)]
Add PIC-level support to Clang.
Summary:
This distinguishes between -fpic and -fPIC now, with the additions in LLVM for
PIC level support.
Test Plan: No regressions
Reviewers: echristo, rafael
Reviewed By: rafael
Subscribers: rnk, emaste, llvm-commits
Differential Revision: http://reviews.llvm.org/D5400
llvm-svn: 222227
Jason Molenda [Tue, 18 Nov 2014 05:57:42 +0000 (05:57 +0000)]
Fix up the code in the FuncUnwinders class that
retrieves the personality routine addr and the
LSDA addr. Don't bother checking with the
"non-call site" unwind plan - this kind of
information is only going to come from the
call site unwind plan.
llvm-svn: 222226
Jason Molenda [Tue, 18 Nov 2014 05:43:11 +0000 (05:43 +0000)]
Add documentation for the SBTarget::ReadInstructions and
SBTarget::GetInstructions APIs so it's a little clearer to
understand which should be used.
<rdar://problem/
18787018>
llvm-svn: 222225
Argyrios Kyrtzidis [Tue, 18 Nov 2014 05:24:18 +0000 (05:24 +0000)]
[ASTReader] Add a convenience function to retrieve all the input files of a module file.
llvm-svn: 222224
Justin Bogner [Tue, 18 Nov 2014 05:22:39 +0000 (05:22 +0000)]
docs: Modernize some examples in WritingAnLLVMPass
llvm-svn: 222223