platform/upstream/llvm.git
9 years agoMake convert_to_integral.pass.cpp more platform generic.
Eric Fiselier [Wed, 11 Feb 2015 01:18:05 +0000 (01:18 +0000)]
Make convert_to_integral.pass.cpp more platform generic.

Don't depend on the underlying types of enums and wchar_t.

llvm-svn: 228781

9 years agoFix libcxxabi's library and object root for tests.
Eric Fiselier [Wed, 11 Feb 2015 01:07:48 +0000 (01:07 +0000)]
Fix libcxxabi's library and object root for tests.

llvm-svn: 228779

9 years agoRemove default definition for libcxx_obj_dir because it doesn't make sense
Eric Fiselier [Wed, 11 Feb 2015 01:03:44 +0000 (01:03 +0000)]
Remove default definition for libcxx_obj_dir because it doesn't make sense

llvm-svn: 228778

9 years agoAs part of the cleanup when a process dies, tell watchpoints to forget their previous...
Enrico Granata [Wed, 11 Feb 2015 00:37:54 +0000 (00:37 +0000)]
As part of the cleanup when a process dies, tell watchpoints to forget their previously recorded values

Because types are not reliably protected against the death of their owners, having ValueObjects lurking around like that past the useful lifetime of their owner processes is a potential source of crashes
That is - in itself - worth fixing at some point, but for this case, watchpoints holding on to old values don't offer enough value to make the larger fix worth

Fixes rdar://19788756

llvm-svn: 228777

9 years agoR600/SI: Store immediate offsets > 12-bits in soffset
Tom Stellard [Wed, 11 Feb 2015 00:34:35 +0000 (00:34 +0000)]
R600/SI: Store immediate offsets > 12-bits in soffset

This will save us from having to extend these offsets to 64-bits
and storing them in a pair of vgprs.

llvm-svn: 228776

9 years agoR600/SI: Add soffset operand to mubuf addr64 instruction
Tom Stellard [Wed, 11 Feb 2015 00:34:32 +0000 (00:34 +0000)]
R600/SI: Add soffset operand to mubuf addr64 instruction

We were previously hard-coding soffset to 0.

llvm-svn: 228775

9 years agoFix warning due to unused private member variable.
Zachary Turner [Wed, 11 Feb 2015 00:33:00 +0000 (00:33 +0000)]
Fix warning due to unused private member variable.

llvm-svn: 228774

9 years agoFix some warnings due to -Wcovered-switch-default.
Zachary Turner [Wed, 11 Feb 2015 00:13:39 +0000 (00:13 +0000)]
Fix some warnings due to -Wcovered-switch-default.

llvm-svn: 228773

9 years ago[UBSan] Add report deduplication for -fsanitize=function.
Alexey Samsonov [Wed, 11 Feb 2015 00:05:31 +0000 (00:05 +0000)]
[UBSan] Add report deduplication for -fsanitize=function.

Summary:
Make sure we don't print the error report from -fsanitize=function
twice for the same source location, as we do in another UBSan handlers.

Test Plan: check-ubsan test suite

Reviewers: rsmith, pcc

Reviewed By: pcc

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D7524

llvm-svn: 228772

9 years agoFixed a crasher that could happen if:
Greg Clayton [Wed, 11 Feb 2015 00:04:33 +0000 (00:04 +0000)]
Fixed a crasher that could happen if:
- you have a type that contains a typedef to a VectorType or an ExtVectorType
- that type is returned from an ARM function that LLDB steps over so we try to figure out the return type
- we try to determine if the type is a homogeneous aggregate type and we crash

We get not using getAs() when we should have been and using llvm::cast caused an assertion crash when the typedef type didn't return a valid VectorType or ExtVectorType.

<rdar://problem/19646550>

llvm-svn: 228771

9 years agoAdded CoreMedia.cpp to CMakeLists.txt for egranata.
Chaoren Lin [Wed, 11 Feb 2015 00:00:39 +0000 (00:00 +0000)]
Added CoreMedia.cpp to CMakeLists.txt for egranata.

llvm-svn: 228770

9 years agoEmit landing pads for SEH even if nounwind is present
Reid Kleckner [Wed, 11 Feb 2015 00:00:21 +0000 (00:00 +0000)]
Emit landing pads for SEH even if nounwind is present

Disabling exceptions applies nounwind to lots of functions. SEH catches
asynch exceptions, so emit the landing pad anyway.

llvm-svn: 228769

9 years agoConvert std::make_unique<> to llvm::make_unique<>.
Zachary Turner [Tue, 10 Feb 2015 23:46:48 +0000 (23:46 +0000)]
Convert std::make_unique<> to llvm::make_unique<>.

llvm-svn: 228768

9 years ago[mips] Add __clear_cache() definition for non-Android systems
Petar Jovanovic [Tue, 10 Feb 2015 23:36:19 +0000 (23:36 +0000)]
[mips] Add __clear_cache() definition for non-Android systems

Make sure clear_cache() builtin has an appropriate definition for Linux.
Call syscall(NR_cacheflush, ...).

Differential Revision: http://reviews.llvm.org/D7205

llvm-svn: 228767

9 years agoAdd the missing testcase for r228764.
Adrian Prantl [Tue, 10 Feb 2015 23:32:56 +0000 (23:32 +0000)]
Add the missing testcase for r228764.

llvm-svn: 228766

9 years agoFix makeLibCall argument (signed) in SoftenFloatRes_XINT_TO_FP function
Petar Jovanovic [Tue, 10 Feb 2015 23:30:14 +0000 (23:30 +0000)]
Fix makeLibCall argument (signed) in SoftenFloatRes_XINT_TO_FP function

The isSigned argument of makeLibCall function was hard-coded to false
(unsigned). This caused zero extension on MIPS64 soft float.
As the result SingleSource/Benchmarks/Stanford/FloatMM test and
SingleSource/UnitTests/2005-07-17-INT-To-FP test failed.
The solution was to use the proper argument.

Patch by Strahinja Petrovic.

Differential Revision: http://reviews.llvm.org/D7292

llvm-svn: 228765

9 years agoDebug Info: Support variables that are described by more than one MMI
Adrian Prantl [Tue, 10 Feb 2015 23:18:28 +0000 (23:18 +0000)]
Debug Info: Support variables that are described by more than one MMI
table entry. This happens when SROA splits up an alloca and the resulting
allocas cannot be lowered to SSA values because their address is passed
to a function.

Fixes PR22502.

llvm-svn: 228764

9 years agoFix indentation.
Adrian Prantl [Tue, 10 Feb 2015 23:18:15 +0000 (23:18 +0000)]
Fix indentation.

llvm-svn: 228763

9 years agoFix a couple typos in the previous commit
Enrico Granata [Tue, 10 Feb 2015 23:17:07 +0000 (23:17 +0000)]
Fix a couple typos in the previous commit

llvm-svn: 228762

9 years agoEarlyCSE: Add check lines for test added in r228760
David Majnemer [Tue, 10 Feb 2015 23:11:02 +0000 (23:11 +0000)]
EarlyCSE: Add check lines for test added in r228760

llvm-svn: 228761

9 years agoEarlyCSE: It isn't safe to CSE across synchronization boundaries
David Majnemer [Tue, 10 Feb 2015 23:09:43 +0000 (23:09 +0000)]
EarlyCSE: It isn't safe to CSE across synchronization boundaries

This fixes PR22514.

llvm-svn: 228760

9 years agoAdd an LLDB summary for CMTime. Fixes rdar://15370376
Enrico Granata [Tue, 10 Feb 2015 23:02:25 +0000 (23:02 +0000)]
Add an LLDB summary for CMTime. Fixes rdar://15370376

llvm-svn: 228759

9 years agoAdd missing function and header include.
Zachary Turner [Tue, 10 Feb 2015 22:56:21 +0000 (22:56 +0000)]
Add missing function and header include.

llvm-svn: 228758

9 years agoExtract attach core logic from SBTarget::Attach* methods into unified SBTarget::Attac...
Oleksiy Vyalov [Tue, 10 Feb 2015 22:49:57 +0000 (22:49 +0000)]
Extract attach core logic from SBTarget::Attach* methods into unified SBTarget::AttachToProcess and make it work with platform for remote attach purposes.

http://reviews.llvm.org/D7471

llvm-svn: 228757

9 years agoOops. Don't call Windows functions on non-windows.
Zachary Turner [Tue, 10 Feb 2015 22:47:14 +0000 (22:47 +0000)]
Oops.  Don't call Windows functions on non-windows.

llvm-svn: 228756

9 years agoRewrite llvm-pdbdump in terms of LLVMDebugInfoPDB.
Zachary Turner [Tue, 10 Feb 2015 22:43:25 +0000 (22:43 +0000)]
Rewrite llvm-pdbdump in terms of LLVMDebugInfoPDB.

This makes llvm-pdbdump available on all platforms, although it
will currently fail to create a dumper if there is no PDB reader
implementation for the current platform.

It implements dumping of compilands and children, which is less
information than was previously available, but it has to be
rewritten from scratch using the new set of interfaces, so the
rest of the functionality will be added back in subsequent commits.

llvm-svn: 228755

9 years agoX86: @llvm.frameaddress should defer to SelectionDAG for Win CFI
David Majnemer [Tue, 10 Feb 2015 22:00:34 +0000 (22:00 +0000)]
X86: @llvm.frameaddress should defer to SelectionDAG for Win CFI

llvm-svn: 228754

9 years agoGNU: Rename parseZOption because it actually parses only -z max-page-size.
Rui Ueyama [Tue, 10 Feb 2015 21:40:51 +0000 (21:40 +0000)]
GNU: Rename parseZOption because it actually parses only -z max-page-size.

llvm-svn: 228753

9 years agoFix build due to mismatched function signatures.
Zachary Turner [Tue, 10 Feb 2015 21:40:29 +0000 (21:40 +0000)]
Fix build due to mismatched function signatures.

llvm-svn: 228752

9 years ago[Object] Reformat the code with clang-format
Simon Atanasyan [Tue, 10 Feb 2015 21:38:25 +0000 (21:38 +0000)]
[Object] Reformat the code with clang-format

No functional changes.

llvm-svn: 228751

9 years agoUnittest: Do s/_context/_ctx/g.
Rui Ueyama [Tue, 10 Feb 2015 21:28:52 +0000 (21:28 +0000)]
Unittest: Do s/_context/_ctx/g.

llvm-svn: 228750

9 years agoGNU: Add --no-export-dynamic command line option.
Rui Ueyama [Tue, 10 Feb 2015 21:27:31 +0000 (21:27 +0000)]
GNU: Add --no-export-dynamic command line option.

llvm-svn: 228749

9 years agoX86: Make @llvm.frameaddress work correctly with Windows unwind codes
David Majnemer [Tue, 10 Feb 2015 21:22:05 +0000 (21:22 +0000)]
X86: Make @llvm.frameaddress work correctly with Windows unwind codes

Simply loading or storing the frame pointer is not sufficient for
Windows targets.  Instead, create a synthetic frame object that we will
lower later.  References to this synthetic object will be replaced with
the correct reference to the frame address.

llvm-svn: 228748

9 years agoProvide DIA implementation of DebugInfoPDB.
Zachary Turner [Tue, 10 Feb 2015 21:17:52 +0000 (21:17 +0000)]
Provide DIA implementation of DebugInfoPDB.

This implements DebugInfoPDB when the DIA SDK is present on the system.
Specifically, this means that the following conditions are met:
  1) You are building on Windows.
  2) You are building with MSVC.
  3) Visual Studio did not corrupt the installation of DIA due to a
     known issue with side-by-side installations of VS2012 and VS2013.
If all of these conditions are true, you will be able to pass a value
of PDB_Reader::DIA to PDB::createPdbReader().

There are no tests for this yet, as any test will be in the form of a
lit test which tests the llvm-pdbdump.exe, which still needs to be
rewritten in terms of this library.

llvm-svn: 228747

9 years agoReformat (and remove some tabs) to make debugging this code a
Eric Christopher [Tue, 10 Feb 2015 21:15:06 +0000 (21:15 +0000)]
Reformat (and remove some tabs) to make debugging this code a
little easier to step through.

llvm-svn: 228746

9 years agoNow use the __debugbreak intrinsic instead of calling RaiseException; it requires...
Aaron Ballman [Tue, 10 Feb 2015 21:13:04 +0000 (21:13 +0000)]
Now use the __debugbreak intrinsic instead of calling RaiseException; it requires no forward declares and still calls VEH.

llvm-svn: 228745

9 years agoFix 'process launch -i' for remote processes
Vince Harron [Tue, 10 Feb 2015 21:09:04 +0000 (21:09 +0000)]
Fix 'process launch -i' for remote processes

We want to forward stdin when stdio is not disabled and when we're not
redirecting stdin from a file.

renamed m_stdio_disable to m_stdin_forward and inverted value because
that's what we want to remember.

There was previously a bug that if you redirected stdin from a file,
stdout and stderr would also be redirected to /dev/null

Adds support for remote target to TestProcessIO.py

Fixes ProcessIOTestCase.test_stdin_redirection_with_dwarf for remote
Linux targets

llvm-svn: 228744

9 years agoDriver: move non-positional parameters out of switch. NFC.
Rui Ueyama [Tue, 10 Feb 2015 20:57:43 +0000 (20:57 +0000)]
Driver: move non-positional parameters out of switch. NFC.

llvm-svn: 228743

9 years agoFix overly prescriptive test that broken on Mac after r228725.
Daniel Jasper [Tue, 10 Feb 2015 20:49:05 +0000 (20:49 +0000)]
Fix overly prescriptive test that broken on Mac after r228725.

llvm-svn: 228742

9 years agofix docs typo
Nico Weber [Tue, 10 Feb 2015 20:43:54 +0000 (20:43 +0000)]
fix docs typo

llvm-svn: 228741

9 years agoFix ASan's Noinst unit tests
Kuba Brecka [Tue, 10 Feb 2015 20:37:57 +0000 (20:37 +0000)]
Fix ASan's Noinst unit tests

We currently skip all "Noinst" unit tests on OS X, which was probably caused when we removed the "allow_reexec" flag. The MaybeReexec function fails to re-execute when the runtime is linked statically, because there is no dylib to use. This patch adds an explicit DisableReexec function that is used from asan_noinst_test.cc and the runtime then doesn't try to re-execute.

Reviewed at http://reviews.llvm.org/D7493

llvm-svn: 228740

9 years ago[PowerPC] Remove the --no-tls-optimize workaround from the clang driver
Bill Schmidt [Tue, 10 Feb 2015 20:36:08 +0000 (20:36 +0000)]
[PowerPC] Remove the --no-tls-optimize workaround from the clang driver

llvm-svn: 228739

9 years agoChanging the status code generated by LLVM_BUILTIN_TRAP on Windows to be something...
Aaron Ballman [Tue, 10 Feb 2015 20:13:52 +0000 (20:13 +0000)]
Changing the status code generated by LLVM_BUILTIN_TRAP on Windows to be something categorized as a valid error code. Fixes crashing uses (such as not --crash) with existing sys::Wait behavior.

llvm-svn: 228738

9 years ago[analyzer] Pass actual state to alloc/dealloc handling functions.
Anton Yartsev [Tue, 10 Feb 2015 20:13:08 +0000 (20:13 +0000)]
[analyzer] Pass actual state to alloc/dealloc handling functions.

The state obtained from CheckerContext::getState() may be outdated by the time the alloc/dealloc handling function is called (e.g. the state was modified but the transition was not performed). State argument was added to all alloc/dealloc handling functions in order to get the latest state and to allow sequential calls to those functions.

llvm-svn: 228737

9 years agoFixed memory corruption problem.
Andrey Churbanov [Tue, 10 Feb 2015 20:10:21 +0000 (20:10 +0000)]
Fixed memory corruption problem.

llvm-svn: 228736

9 years agoCheck for backtraces in tests which are verifying pretty stack traces from a crashing...
Pete Cooper [Tue, 10 Feb 2015 19:53:38 +0000 (19:53 +0000)]
Check for backtraces in tests which are verifying pretty stack traces from a crashing clang.

PrettyStackTrace now requires the ENABLE_BACKTRACES option.  We need to check for that here or these tests fail llvm-lit.

Reviewed by chandlerc

llvm-svn: 228735

9 years agoUpdated the kmp_omp_struct_info_t structure used by debuggers.
Andrey Churbanov [Tue, 10 Feb 2015 19:53:13 +0000 (19:53 +0000)]
Updated the kmp_omp_struct_info_t structure used by debuggers.

llvm-svn: 228734

9 years agoAdding support for llvm.eh.begincatch and llvm.eh.endcatch intrinsics and beginning...
Andrew Kaylor [Tue, 10 Feb 2015 19:52:43 +0000 (19:52 +0000)]
Adding support for llvm.eh.begincatch and llvm.eh.endcatch intrinsics and beginning the documentation of native Windows exception handling.

Differential Revision: http://reviews.llvm.org/D7398

llvm-svn: 228733

9 years ago[UBSan] Reduce the number of getCallerLocation() calls.
Alexey Samsonov [Tue, 10 Feb 2015 19:50:20 +0000 (19:50 +0000)]
[UBSan] Reduce the number of getCallerLocation() calls.

getCallerLocation() is expensive as it issues a call to symbolizer.
(In fact, this function has a memory leak at the moment, but this
will be fixed in the nearest future). We should only call it if
we're actually going to print an error report, in particular,
once for every reported source location.

__ubsan_handle_type_mismatch: call getCallerLocation() only if
provided source location is invalid, and only if the report is not
deduplicated.

__ubsan_handle_float_cast_overflow: call getSourceLocation with
correct CallerPC (the one in user code, not in UBSan handler). Source
location for this check is not currently emitted by frontend.

llvm-svn: 228732

9 years agoDeadArgElim: arguments affect all returned sub-values by default.
Tim Northover [Tue, 10 Feb 2015 19:49:18 +0000 (19:49 +0000)]
DeadArgElim: arguments affect all returned sub-values by default.

Unless we meet an insertvalue on a path from some value to a return, that value
will be live if *any* of the return's components are live, so all of those
components must be added to the MaybeLiveUses.

Previously we were deleting arguments if sub-value 0 turned out to be dead.

llvm-svn: 228731

9 years agoOpenMP 4.0 standard function omp_is_initial_device() implemented.
Andrey Churbanov [Tue, 10 Feb 2015 19:47:09 +0000 (19:47 +0000)]
OpenMP 4.0 standard function omp_is_initial_device() implemented.

llvm-svn: 228730

9 years agoDon't crash when evaluating a DWARF expression has a DW_OP_bra with nothing on the...
Greg Clayton [Tue, 10 Feb 2015 19:43:15 +0000 (19:43 +0000)]
Don't crash when evaluating a DWARF expression has a DW_OP_bra with nothing on the value stack.

<rdar://problem/18919125>

llvm-svn: 228729

9 years agoFix up r228725, missed change in PPCSubtarget definition
Bill Schmidt [Tue, 10 Feb 2015 19:31:55 +0000 (19:31 +0000)]
Fix up r228725, missed change in PPCSubtarget definition

llvm-svn: 228728

9 years agoAdded comment on format of local labels in asm code
Andrey Churbanov [Tue, 10 Feb 2015 19:31:17 +0000 (19:31 +0000)]
Added comment on format of local labels in asm code

llvm-svn: 228727

9 years agoIR: Add MDNode::replaceWithPermanent()
Duncan P. N. Exon Smith [Tue, 10 Feb 2015 19:13:46 +0000 (19:13 +0000)]
IR: Add MDNode::replaceWithPermanent()

Add new API for converting temporaries that may self-reference.
Self-referencing nodes are not allowed to be uniqued, so sending them
into `replaceWithUniqued()` is dangerous (and this commit adds
assertions that prevent it).

`replaceWithPermanent()` has similar semantics to `get()` followed by
calls to `replaceOperandWith()`.  In particular, if there's a
self-reference, it returns a distinct node; otherwise, it returns a
uniqued one.  Like `replaceWithUniqued()` and `replaceWithDistinct()`
(well, it calls out to them) it mutates the temporary node in place if
possible, only calling `replaceAllUsesWith()` on a uniquing collision.

llvm-svn: 228726

9 years ago[PowerPC] Fix reverted patch r227976 to avoid register assignment issues
Bill Schmidt [Tue, 10 Feb 2015 19:09:05 +0000 (19:09 +0000)]
[PowerPC] Fix reverted patch r227976 to avoid register assignment issues

See full discussion in http://reviews.llvm.org/D7491.

We now hide the add-immediate and call instructions together in a
separate pseudo-op, which is tagged to define GPR3 and clobber the
call-killed registers.  The PPCTLSDynamicCall pass prior to RA now
expands this op into the two separate addi and call ops, with explicit
definitions of GPR3 on both instructions, and explicit clobbers on the
call instruction.  The pass is now marked as requiring and preserving
the LiveIntervals and SlotIndexes analyses, and fixes these up after
the replacement sequences are introduced.

Self-hosting has been verified on LE P8 and BE P7 with various
optimization levels, etc.  It has also been verified with the
--no-tls-optimize flag workaround removed.

llvm-svn: 228725

9 years agoX86: Emit Win64 SaveXMM opcodes at the right offset in the right order
David Majnemer [Tue, 10 Feb 2015 19:01:47 +0000 (19:01 +0000)]
X86: Emit Win64 SaveXMM opcodes at the right offset in the right order

Walk the instructions marked FrameSetup and consider any stores of XMM
registers to the stack as needing a SaveXMM opcode.

This fixes PR22521.

Differential Revision: http://reviews.llvm.org/D7527

llvm-svn: 228724

9 years agoUse override rather than virtual.
Rui Ueyama [Tue, 10 Feb 2015 18:59:37 +0000 (18:59 +0000)]
Use override rather than virtual.

llvm-svn: 228723

9 years agoGNU: Use StringRef::empty instead of a boolean flag.
Rui Ueyama [Tue, 10 Feb 2015 18:55:39 +0000 (18:55 +0000)]
GNU: Use StringRef::empty instead of a boolean flag.

This local variable name did not follow the style guide,
and it is not actually needed.

llvm-svn: 228722

9 years agoAdded CFI directives to asm code in order to have correct backtraces in gdb.
Andrey Churbanov [Tue, 10 Feb 2015 18:51:52 +0000 (18:51 +0000)]
Added CFI directives to asm code in order to have correct backtraces in gdb.

llvm-svn: 228721

9 years agoMake ABI header not found a warning, not an error.
Dan Albert [Tue, 10 Feb 2015 18:46:57 +0000 (18:46 +0000)]
Make ABI header not found a warning, not an error.

Since we've added a new header to libc++abi (__cxxabi_config.h), we
now have a case where we might not always find all the ABI headers:
building libc++ against the system's libc++abi on Darwin.

Since this isn't actually a fatal error, degrade it to a warning.

llvm-svn: 228720

9 years ago[PowerPC] Support the (old) cntlz instruction alias
Hal Finkel [Tue, 10 Feb 2015 18:45:02 +0000 (18:45 +0000)]
[PowerPC] Support the (old) cntlz instruction alias

Some old assembly code uses the cntlz alias for cntlzw, binutils supports this,
and we should too. Fixes PR22519.

llvm-svn: 228719

9 years agoThe usage of tt_state flag is replaced by an array of two task_team pointers.
Andrey Churbanov [Tue, 10 Feb 2015 18:37:43 +0000 (18:37 +0000)]
The usage of tt_state flag is replaced by an array of two task_team pointers.

llvm-svn: 228718

9 years agoDriver: use hasArg instead of getLastArg if return value is not used.
Rui Ueyama [Tue, 10 Feb 2015 18:34:46 +0000 (18:34 +0000)]
Driver: use hasArg instead of getLastArg if return value is not used.

llvm-svn: 228717

9 years agoFailsafe directory for shared objects and support executables.
Chaoren Lin [Tue, 10 Feb 2015 18:30:34 +0000 (18:30 +0000)]
Failsafe directory for shared objects and support executables.

llvm-svn: 228716

9 years agoAdding x86 to supported architectures on x86_64.
Chaoren Lin [Tue, 10 Feb 2015 18:30:31 +0000 (18:30 +0000)]
Adding x86 to supported architectures on x86_64.

llvm-svn: 228715

9 years agoFix build on Windows which was broken by a recent commit.
Hafiz Abid Qadeer [Tue, 10 Feb 2015 18:12:44 +0000 (18:12 +0000)]
Fix build on Windows which was broken by a recent commit.
This file MICmnStreamStdinLinux.cpp is wrongly added in the windows build.
It has no use there. CMakeList.txt for lldb-mi needs to be re-factored to
include the right file for each build. This is quick fix to un-break the build.

llvm-svn: 228714

9 years agoAdd a test case for new unrolling heuristics.
Michael Zolotukhin [Tue, 10 Feb 2015 17:54:54 +0000 (17:54 +0000)]
Add a test case for new unrolling heuristics.

THe heuristics were added in r228265 and r228434.

llvm-svn: 228713

9 years agoAdd pragma system header to some experimental headers and add newlines to files.
Eric Fiselier [Tue, 10 Feb 2015 17:32:49 +0000 (17:32 +0000)]
Add pragma system header to some experimental headers and add newlines to files.

llvm-svn: 228712

9 years agoFix more issues exposed by -pedantic-errors in c++03 mode
Eric Fiselier [Tue, 10 Feb 2015 17:20:18 +0000 (17:20 +0000)]
Fix more issues exposed by -pedantic-errors in c++03 mode

llvm-svn: 228711

9 years agoFix the LLDB build under Debian Kfreebsd
Sylvestre Ledru [Tue, 10 Feb 2015 17:16:13 +0000 (17:16 +0000)]
Fix the LLDB build under Debian Kfreebsd

Summary:
I don't know if there is a better way for the change in source/Host/freebsd/ThisThread.cpp

Reviewers: emaste

Subscribers: hansw, emaste, lldb-commits

Differential Revision: http://reviews.llvm.org/D7441

llvm-svn: 228710

9 years agoFix multiple problems of lldb-mi blocking on input monitoring and needing a return.
Hafiz Abid Qadeer [Tue, 10 Feb 2015 17:10:39 +0000 (17:10 +0000)]
Fix multiple problems of lldb-mi blocking on input monitoring and needing a return.

Summary:
One of the problem is reported here.
http://llvm.org/bugs/show_bug.cgi?id=22411

A fix was committed for this problem that works only for OSX. This revision
extends that fix to other system. The select system call has some limitation with
multi-threaded application which have been addresses here.

LLDB-mi exits if quit command is given but needs an extra retur if -gdb-exit is
given. That issue has also been addressed.

Reviewers: ki.stfu, emaste

Reviewed By: ki.stfu

Subscribers: lldb-commits

Differential Revision: http://reviews.llvm.org/D7529

llvm-svn: 228709

9 years ago[Hexagon] Adding vector load with post-increment instructions. Adding decoder functi...
Colin LeMahieu [Tue, 10 Feb 2015 16:59:36 +0000 (16:59 +0000)]
[Hexagon] Adding vector load with post-increment instructions.  Adding decoder function for 64bit control register class.

llvm-svn: 228708

9 years agoFixed a problem where we might omit some breakpoints when using DWARF in .o files.
Greg Clayton [Tue, 10 Feb 2015 16:53:40 +0000 (16:53 +0000)]
Fixed a problem where we might omit some breakpoints when using DWARF in .o files.

The problem occurred when we had incorrect address ranges in the debug map that included the padding between functions causing the end address of a line table entry to fall into an inlinked (next function) address range.

<rdar://problem/19721144>

llvm-svn: 228707

9 years agoRemove use of zero length arrays in tests. Get tests passing with -pedantic-errors
Eric Fiselier [Tue, 10 Feb 2015 16:51:29 +0000 (16:51 +0000)]
Remove use of zero length arrays in tests. Get tests passing with -pedantic-errors

llvm-svn: 228706

9 years ago[libcxx] Fix PR 22468 - std::function<void()> does not accept non-void-returning...
Eric Fiselier [Tue, 10 Feb 2015 16:48:45 +0000 (16:48 +0000)]
[libcxx] Fix PR 22468 - std::function<void()> does not accept non-void-returning functions

Summary:
The bug can be found here: http://llvm.org/bugs/show_bug.cgi?id=22468

`__invoke_void_return_wrapper` is needed to properly handle calling a function that returns a value but where the std::function return type is void. Without this '-Wsystem-headers' will cause `function::operator()(...)` to not compile.

Reviewers: eugenis, K-ballo, mclow.lists

Reviewed By: mclow.lists

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D7444

llvm-svn: 228705

9 years ago[libcxx] Properly convert the count arguments to the *_n algorithms before use.
Eric Fiselier [Tue, 10 Feb 2015 16:46:42 +0000 (16:46 +0000)]
[libcxx] Properly convert the count arguments to the *_n algorithms before use.

Summary:
The requirement on the `Size` type passed to *_n algorithms is that it is convertible to an integral type. This means we can't use a variable of type `Size` directly. Instead we need to convert it to an integral type first.  The problem is finding out what integral type to convert it to.  `__convert_to_integral` figures out what integral type to convert it to and performs the conversion, It also promotes the resulting integral type so that it is at least as big as an integer. `__convert_to_integral` also has a special case for converting enums. This should only work on non-scoped enumerations because it does not apply an explicit conversion from the enum to its underlying type.

Reviewers: chandlerc, mclow.lists

Reviewed By: mclow.lists

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D7449

llvm-svn: 228704

9 years ago[mips][microMIPS] Implement movep instruction
Zoran Jovanovic [Tue, 10 Feb 2015 16:36:20 +0000 (16:36 +0000)]
[mips][microMIPS] Implement movep instruction
Differential Revision: http://reviews.llvm.org/D7465

llvm-svn: 228703

9 years ago[ASan/Win] NFC, just minor simplification of the code
Timur Iskhodzhanov [Tue, 10 Feb 2015 16:29:05 +0000 (16:29 +0000)]
[ASan/Win] NFC, just minor simplification of the code

llvm-svn: 228702

9 years ago[ASan/Win] Add the executable's directory to the symbol search path
Timur Iskhodzhanov [Tue, 10 Feb 2015 16:17:01 +0000 (16:17 +0000)]
[ASan/Win] Add the executable's directory to the symbol search path

This should fix symbolization in those cases when the .exe file is moved together with the .pdb

llvm-svn: 228701

9 years agoTwo comment typo fixes in lib/CodeGen/SelectionDAG/DAGCombiner.cpp.
Jonas Paulsson [Tue, 10 Feb 2015 15:34:29 +0000 (15:34 +0000)]
Two comment typo fixes in lib/CodeGen/SelectionDAG/DAGCombiner.cpp.

llvm-svn: 228700

9 years agoExplicitly initialize a flag in a default constructor.
Paul Robinson [Tue, 10 Feb 2015 15:30:02 +0000 (15:30 +0000)]
Explicitly initialize a flag in a default constructor.
Works around a Visual C++ issue.

Patch by Douglas Yung!

llvm-svn: 228699

9 years agoFix use of C++11 extensions in C++03 code.
Eric Fiselier [Tue, 10 Feb 2015 15:17:46 +0000 (15:17 +0000)]
Fix use of C++11 extensions in C++03 code.

llvm-svn: 228698

9 years ago[ARM] Add tests for armv6s[-]m being an alias to armv6[-]m
Bradley Smith [Tue, 10 Feb 2015 15:16:49 +0000 (15:16 +0000)]
[ARM] Add tests for armv6s[-]m being an alias to armv6[-]m

llvm-svn: 228697

9 years ago[ARM] Add armv6s[-]m as an alias to armv6[-]m
Bradley Smith [Tue, 10 Feb 2015 15:15:08 +0000 (15:15 +0000)]
[ARM] Add armv6s[-]m as an alias to armv6[-]m

llvm-svn: 228696

9 years agoFixed module-map-checker to recurse umbrella dirs when collecting headers.
John Thompson [Tue, 10 Feb 2015 14:58:07 +0000 (14:58 +0000)]
Fixed module-map-checker to recurse umbrella dirs when collecting headers.

llvm-svn: 228695

9 years agoAdd translationUnitDecl matcher.
Samuel Benzaquen [Tue, 10 Feb 2015 14:46:45 +0000 (14:46 +0000)]
Add translationUnitDecl matcher.

Summary: Add translationUnitDecl matcher.

Reviewers: alexfh

Subscribers: klimek, cfe-commits

Differential Revision: http://reviews.llvm.org/D7512

llvm-svn: 228694

9 years agoRenamed module.map to module.modulemap (modularize).
John Thompson [Tue, 10 Feb 2015 14:45:30 +0000 (14:45 +0000)]
Renamed module.map to module.modulemap (modularize).

llvm-svn: 228693

9 years agoRenamed module.map to module.modulemap.
John Thompson [Tue, 10 Feb 2015 14:29:16 +0000 (14:29 +0000)]
Renamed module.map to module.modulemap.

llvm-svn: 228692

9 years agoRe-committing r228628 with a fix for 64-bit builds.
Aaron Ballman [Tue, 10 Feb 2015 14:28:11 +0000 (14:28 +0000)]
Re-committing r228628 with a fix for 64-bit builds.

On Windows, we now use RaiseException to generate the kind of trap we require (one which calls our vectored exception handler), and fall back to using a volatile write to simulate a trap elsewhere.

llvm-svn: 228691

9 years agoFix typo in cmake example docs
Renato Golin [Tue, 10 Feb 2015 14:15:58 +0000 (14:15 +0000)]
Fix typo in cmake example docs

Patch by Vinicius Tinti.

llvm-svn: 228690

9 years agoThe -I option needs a space, apparently, from the actual implementation.
John Thompson [Tue, 10 Feb 2015 13:51:17 +0000 (13:51 +0000)]
The -I option needs a space, apparently, from the actual implementation.

llvm-svn: 228689

9 years ago[X86][AVX2] Missing AVX2 memory folding instructions
Simon Pilgrim [Tue, 10 Feb 2015 13:22:57 +0000 (13:22 +0000)]
[X86][AVX2] Missing AVX2 memory folding instructions

Added most of the missing vector folding patterns for AVX2 (as well as fixing the vpermpd and verpmq patterns)

Differential Revision: http://reviews.llvm.org/D7492

llvm-svn: 228688

9 years ago[mips][microMIPS] Add disassembler tests for 16-bit instructions BREAK16 and SDBBP16
Jozef Kolek [Tue, 10 Feb 2015 13:20:51 +0000 (13:20 +0000)]
[mips][microMIPS] Add disassembler tests for 16-bit instructions BREAK16 and SDBBP16

Differential Revision: http://reviews.llvm.org/D7443

llvm-svn: 228687

9 years agoBugfix for missed dependency from store to load in buildSchedGraph().
Jonas Paulsson [Tue, 10 Feb 2015 13:03:32 +0000 (13:03 +0000)]
Bugfix for missed dependency from store to load in buildSchedGraph().

Background: When handling underlying objects for a store, the vector
of previous mem uses, mapped to the same Value, is afterwards cleared
(regardless of ThisMayAlias). This means that during handling of the
next store using the same Value, adjustChainDeps() must be called,
otherwise a dependency might be missed.

For example, three spill/reload (NonAliasing) memory accesses using
the same Value 'a', with different offsets:

    SU(2): store  @a
    SU(1): store  @a, Offset:1
    SU(0): load   @a

In this case we have:

* SU(1) does not need a dep against SU(0). Therefore,SU(0) ends up in
  RejectMemNodes and is removed from the mem-uses list (AliasMemUses
  or NonAliasMemUses), as this list is cleared.

* SU(2) needs a dep against SU(0). Therefore, SU(2) must check
  RejectMemNodes by calling adjustChainDeps().

Previously, for store SUs, adjustChainDeps() was only called if
MayAlias was true, missing the S(2) to S(0) dependency in the case
above. The fix is to always call adjustChainDeps(), regardless of
MayAlias, since this applies both for AliasMemUses and
NonAliasMemUses.

No testcase found for any in-tree target.

llvm-svn: 228686

9 years ago[X86][XOP] Added XOP memory folding patterns + tests
Simon Pilgrim [Tue, 10 Feb 2015 12:57:17 +0000 (12:57 +0000)]
[X86][XOP] Added XOP memory folding patterns + tests

This patch adds the complete AMD Bulldozer XOP instruction set to the memory folding pattern tables for stack folding, etc.

Note: Many of the XOP instructions have multiple table entries as it can fold loads from different sources.

Differential Revision: http://reviews.llvm.org/D7484

llvm-svn: 228685

9 years ago[mips][microMIPS] Fix disassembling of 16-bit microMIPS instructions LWM16 and SWM16
Jozef Kolek [Tue, 10 Feb 2015 12:41:13 +0000 (12:41 +0000)]
[mips][microMIPS] Fix disassembling of 16-bit microMIPS instructions LWM16 and SWM16

Differential Revision: http://reviews.llvm.org/D7436

llvm-svn: 228683

9 years ago[X86][FastIsel] Avoid introducing legacy SSE instructions if the target has AVX.
Andrea Di Biagio [Tue, 10 Feb 2015 12:04:41 +0000 (12:04 +0000)]
[X86][FastIsel] Avoid introducing legacy SSE instructions if the target has AVX.

This patch teaches X86FastISel how to select AVX instructions for scalar
float/double convert operations.

Before this patch, X86FastISel always selected legacy SSE instructions
for FPExt (from float to double) and FPTrunc (from double to float).

For example:
\code
  define double @foo(float %f) {
    %conv = fpext float %f to double
    ret double %conv
  }
\end code

Before (with -mattr=+avx -fast-isel) X86FastIsel selected a CVTSS2SDrr which is
legacy SSE:
  cvtss2sd %xmm0, %xmm0

With this patch, X86FastIsel selects a VCVTSS2SDrr instead:
  vcvtss2sd %xmm0, %xmm0, %xmm0

Added test fast-isel-fptrunc-fpext.ll to check both the register-register and
the register-memory float/double conversion variants.

Differential Revision: http://reviews.llvm.org/D7438

llvm-svn: 228682

9 years agoAdd thread-id field in *stopped notification (MI)
Ilia K [Tue, 10 Feb 2015 12:02:03 +0000 (12:02 +0000)]
Add thread-id field in *stopped notification (MI)

Summary:
Add thread-id field in *stopped notification (MI) + tests

All tests pass on OS X

Reviewers: zturner, clayborg, abidh

Reviewed By: clayborg

Subscribers: lldb-commits, zturner, clayborg, abidh

Differential Revision: http://reviews.llvm.org/D7501

llvm-svn: 228681

9 years ago[ELF][ARM] Add veneer generation to branch instructions
Denis Protivensky [Tue, 10 Feb 2015 09:31:42 +0000 (09:31 +0000)]
[ELF][ARM] Add veneer generation to branch instructions

When calling ARM code from Thumb and vice versa,
a veneer that switches instruction set should be generated.
Added veneer generation for ARM_JUMP24 ARM_THM_JUMP24 instructions.

Differential Revision: http://reviews.llvm.org/D7502

llvm-svn: 228680