platform/upstream/llvm.git
11 years agoAdd a new function to ConstantExpr - getAsInstruction. This returns its Instruction...
James Molloy [Sat, 17 Nov 2012 17:56:30 +0000 (17:56 +0000)]
Add a new function to ConstantExpr - getAsInstruction. This returns its Instruction* corollary, which may be useful if a user
wishes to transform a ConstantExpr so that one of its operands is no longer constant.

llvm-svn: 168262

11 years agoEnable inlining of 4 byte atomic ops on ppc32, 8 byte atomic ops on ppc64.
Benjamin Kramer [Sat, 17 Nov 2012 17:30:55 +0000 (17:30 +0000)]
Enable inlining of 4 byte atomic ops on ppc32, 8 byte atomic ops on ppc64.

Also fixes a bit/byte mismatch when checking if a target supports atomic ops of a certain size.

llvm-svn: 168260

11 years agoPlug a memory leak in the GCOV profiling emitter, which never released the edge table...
Benjamin Kramer [Sat, 17 Nov 2012 13:49:37 +0000 (13:49 +0000)]
Plug a memory leak in the GCOV profiling emitter, which never released the edge table memory.

llvm-svn: 168259

11 years agoRemove copy ctor that provides no value over the default.
Benjamin Kramer [Sat, 17 Nov 2012 09:14:31 +0000 (09:14 +0000)]
Remove copy ctor that provides no value over the default.

It's also simpler to just copy the words than mangling bits like this ctor did.

llvm-svn: 168258

11 years agoFurther reduce "-fsyntax-only -Wuninitialized" time on sqlite3.c by another 2.5%...
Ted Kremenek [Sat, 17 Nov 2012 07:18:30 +0000 (07:18 +0000)]
Further reduce "-fsyntax-only -Wuninitialized" time on sqlite3.c by another 2.5% using intelligent pruning of blocks during the final reporting pass.

llvm-svn: 168257

11 years agoreally fix permissions
Pawel Wodnicki [Sat, 17 Nov 2012 06:38:44 +0000 (06:38 +0000)]
really fix permissions

llvm-svn: 168256

11 years agofix permissions
Pawel Wodnicki [Sat, 17 Nov 2012 06:35:19 +0000 (06:35 +0000)]
fix permissions

llvm-svn: 168255

11 years agoadding whose code is it anywa tools
Pawel Wodnicki [Sat, 17 Nov 2012 06:24:37 +0000 (06:24 +0000)]
adding whose code is it anywa tools

llvm-svn: 168254

11 years agoRemoving utf-8 smart quote and trailing whitespace
Joe Abbey [Sat, 17 Nov 2012 05:13:16 +0000 (05:13 +0000)]
Removing utf-8 smart quote and trailing whitespace

llvm-svn: 168253

11 years agoutils: remove the isl directory after cloning cloog
Sebastian Pop [Sat, 17 Nov 2012 04:57:41 +0000 (04:57 +0000)]
utils: remove the isl directory after cloning cloog

This fixes my previous commit, thanks to "Dmitry N. Mikushin"
<maemarcus@gmail.com>

llvm-svn: 168252

11 years agoSuppressing the 'direct base ‘{anonymous}::S1’ inaccessible in ‘{anonymous}::D9’
Joe Abbey [Sat, 17 Nov 2012 04:54:22 +0000 (04:54 +0000)]
Suppressing the 'direct base ‘{anonymous}::S1’ inaccessible in ‘{anonymous}::D9’
due to ambiguity' warning.

llvm-svn: 168251

11 years agoutils: remove existing cloog and isl dirs before git cloning
Sebastian Pop [Sat, 17 Nov 2012 03:43:48 +0000 (03:43 +0000)]
utils: remove existing cloog and isl dirs before git cloning

Patch adapted from "Dmitry N. Mikushin" <maemarcus@gmail.com>.

llvm-svn: 168250

11 years agoSilence the buildbots for this test while I figure out the triple
Andrew Trick [Sat, 17 Nov 2012 03:39:26 +0000 (03:39 +0000)]
Silence the buildbots for this test while I figure out the triple

llvm-svn: 168249

11 years agoBroaden isSchedulingBoundary to check aliases of SP.
Andrew Trick [Sat, 17 Nov 2012 03:35:11 +0000 (03:35 +0000)]
Broaden isSchedulingBoundary to check aliases of SP.

On PPC the stack pointer is X1, but ADJCALLSTACK writes R1.

Fixes PR14315: Register regmask dependency problem with misched.

llvm-svn: 168248

11 years agoPhi speculation improvement for BasicAA
Hal Finkel [Sat, 17 Nov 2012 02:33:15 +0000 (02:33 +0000)]
Phi speculation improvement for BasicAA

This is a partial solution to PR14351. It removes some of the special
significance of the first incoming phi value in the phi aliasing checking logic
in BasicAA. In the context of a loop, the old logic assumes that the first
incoming value is the interesting one (meaning that it is the one that comes
from outside the loop), but this is often not the case.  With this change, we
now test first the incoming value that comes from a block other than the parent
of the phi being tested.

llvm-svn: 168245

11 years agoPer agreement with Doug, take ownership of Clang's libSema.
Richard Smith [Sat, 17 Nov 2012 02:26:54 +0000 (02:26 +0000)]
Per agreement with Doug, take ownership of Clang's libSema.

llvm-svn: 168244

11 years agoSwitch -Wuninitialized to use a reverse-post order traversal as
Ted Kremenek [Sat, 17 Nov 2012 02:00:00 +0000 (02:00 +0000)]
Switch -Wuninitialized to use a reverse-post order traversal as
an initial baseline for enqueued blocks, but use a simple DFS stack
for propagating changes quickly up back edges.

This provides a 3.5% reduction in -fsyntax-only time on sqlite3.c.

llvm-svn: 168241

11 years agoMark FP_EXTEND form v2f32 to v2f64 as "expand" for ARM NEON. Patch by Pete Couperus.
Eli Friedman [Sat, 17 Nov 2012 01:52:46 +0000 (01:52 +0000)]
Mark FP_EXTEND form v2f32 to v2f64 as "expand" for ARM NEON.  Patch by Pete Couperus.

llvm-svn: 168240

11 years agoAdd missing features for misc x86 CPUs to CPU feature translation. Patch by Jung...
Eli Friedman [Sat, 17 Nov 2012 01:43:10 +0000 (01:43 +0000)]
Add missing features for misc x86 CPUs to CPU feature translation.  Patch by Jung-uk Kim.

llvm-svn: 168239

11 years agoHan Ming's commit in r168228 had a bunch of 4-space tabs
Jason Molenda [Sat, 17 Nov 2012 01:41:04 +0000 (01:41 +0000)]
Han Ming's commit in r168228 had a bunch of 4-space tabs
in the source files.  Expand to spaces.  No content changes,
just whitespace.

llvm-svn: 168238

11 years agoClean up X86 target feature translation code slightly. No intended functional change...
Eli Friedman [Sat, 17 Nov 2012 01:16:19 +0000 (01:16 +0000)]
Clean up X86 target feature translation code slightly.  No intended functional change.  Patch by Jung-uk Kim.

llvm-svn: 168237

11 years ago[fast-isel] Add the -verify-machineinstrs to these test cases. The remaining
Chad Rosier [Sat, 17 Nov 2012 00:42:06 +0000 (00:42 +0000)]
[fast-isel] Add the -verify-machineinstrs to these test cases.  The remaining
test cases require fixes to fast-isel before the verifier can be enabled.
Part of rdar://12594152

llvm-svn: 168233

11 years agoFollow up on <rdar://12720514>. Removed commented out code.
Han Ming Ong [Sat, 17 Nov 2012 00:33:14 +0000 (00:33 +0000)]
Follow up on <rdar://12720514>. Removed commented out code.

llvm-svn: 168232

11 years agoLoopVectorizer: Add initial support for pointer induction variables (for example...
Nadav Rotem [Sat, 17 Nov 2012 00:27:03 +0000 (00:27 +0000)]
LoopVectorizer: Add initial support for pointer induction variables (for example: *dst++ = *src++).
At the moment we still require to have an integer induction variable (for example: i++).

llvm-svn: 168231

11 years agoInitial implementation of MipsTargetLowering::isLegalAddressingMode.
Akira Hatanaka [Sat, 17 Nov 2012 00:25:41 +0000 (00:25 +0000)]
Initial implementation of MipsTargetLowering::isLegalAddressingMode.

llvm-svn: 168230

11 years agoRename methods like PairSRegs() to createSRegpairNode() to meet our coding
Weiming Zhao [Sat, 17 Nov 2012 00:23:35 +0000 (00:23 +0000)]
Rename methods like PairSRegs() to createSRegpairNode() to meet our coding
style requirement.

llvm-svn: 168229

11 years ago<rdar://problem/12720514> Sub-TLF: Provide service to profile the inferior
Han Ming Ong [Sat, 17 Nov 2012 00:21:04 +0000 (00:21 +0000)]
<rdar://problem/12720514> Sub-TLF: Provide service to profile the inferior

This allows client to query profiling states on the inferior.

llvm-svn: 168228

11 years agoTeach SROA rewriteVectorizedStoreInst to handle cases when the loaded value is narrow...
Evan Cheng [Sat, 17 Nov 2012 00:05:06 +0000 (00:05 +0000)]
Teach SROA rewriteVectorizedStoreInst to handle cases when the loaded value is narrower than the stored value. rdar://12713675

llvm-svn: 168227

11 years agoMarking remote mcjit tests as XFAIL for cygwin (hopefully only temporarily).
Andrew Kaylor [Sat, 17 Nov 2012 00:02:50 +0000 (00:02 +0000)]
Marking remote mcjit tests as XFAIL for cygwin (hopefully only temporarily).

llvm-svn: 168226

11 years agoClaim ownership of MCJIT
Andrew Kaylor [Fri, 16 Nov 2012 23:56:48 +0000 (23:56 +0000)]
Claim ownership of MCJIT

llvm-svn: 168225

11 years agoAdded symbols page for Mac OS X that explains how DebugSymbols.framework can be custo...
Greg Clayton [Fri, 16 Nov 2012 23:50:32 +0000 (23:50 +0000)]
Added symbols page for Mac OS X that explains how DebugSymbols.framework can be customized.

llvm-svn: 168224

11 years agoAdd some basic support for the fission addr forms to DebugInfo.
Eric Christopher [Fri, 16 Nov 2012 23:44:11 +0000 (23:44 +0000)]
Add some basic support for the fission addr forms to DebugInfo.

llvm-svn: 168223

11 years agoTypo.
Chad Rosier [Fri, 16 Nov 2012 23:41:41 +0000 (23:41 +0000)]
Typo.

llvm-svn: 168222

11 years agoMarking remote mcjit tests as XFAIL for mingw32 (hopefully only temporarily).
Andrew Kaylor [Fri, 16 Nov 2012 23:38:16 +0000 (23:38 +0000)]
Marking remote mcjit tests as XFAIL for mingw32 (hopefully only temporarily).

llvm-svn: 168221

11 years agoFix expected error message in test.
Filipe Cabecinhas [Fri, 16 Nov 2012 23:18:41 +0000 (23:18 +0000)]
Fix expected error message in test.

llvm-svn: 168220

11 years agoAdd constant definitions for fission dwarf attributes, forms, etc.
Eric Christopher [Fri, 16 Nov 2012 23:04:31 +0000 (23:04 +0000)]
Add constant definitions for fission dwarf attributes, forms, etc.

llvm-svn: 168218

11 years agoRevert r167799. It's not really correct, and it doesn't fix the problem that it was...
Bill Wendling [Fri, 16 Nov 2012 23:03:00 +0000 (23:03 +0000)]
Revert r167799. It's not really correct, and it doesn't fix the problem that it was intended to fix.

llvm-svn: 168217

11 years agoClaim the driver.
Chad Rosier [Fri, 16 Nov 2012 22:49:27 +0000 (22:49 +0000)]
Claim the driver.

llvm-svn: 168215

11 years agoTestcase overriding-ftemplate-comments.cpp: use [[@LINE]] to make it less fragile.
Dmitri Gribenko [Fri, 16 Nov 2012 22:34:07 +0000 (22:34 +0000)]
Testcase overriding-ftemplate-comments.cpp: use [[@LINE]] to make it less fragile.

llvm-svn: 168213

11 years ago[driver] Add the missing TY_PP_ObjCXX_Alias case to the isCXX function.
Chad Rosier [Fri, 16 Nov 2012 22:31:39 +0000 (22:31 +0000)]
[driver] Add the missing TY_PP_ObjCXX_Alias case to the isCXX function.

This was causing different behavior when using -x objective-c++-cpp-output as
compared to -x objc++-cpp-output.  Specifically, the driver was not adding the
-fcxx-exceptions flag in the latter case.
rdar://12680312

llvm-svn: 168212

11 years agoRemove default public copy ctors.
Benjamin Kramer [Fri, 16 Nov 2012 22:22:20 +0000 (22:22 +0000)]
Remove default public copy ctors.

They are just useless and prevent SmallVector from picking an optimized codepath
for memcpyable elements.

llvm-svn: 168211

11 years agoMarking remote mcjit tests as XFAIL for ARM (hopefully only temporarily).
Andrew Kaylor [Fri, 16 Nov 2012 22:21:04 +0000 (22:21 +0000)]
Marking remote mcjit tests as XFAIL for ARM (hopefully only temporarily).

llvm-svn: 168210

11 years agoistreambuf_iterator increment should call sbumpc instead of snextc. Patch
Howard Hinnant [Fri, 16 Nov 2012 22:17:23 +0000 (22:17 +0000)]
istreambuf_iterator increment should call sbumpc instead of snextc.  Patch
by Kimball Thurston.  This fixes http://llvm.org/bugs/show_bug.cgi?id=14358.

llvm-svn: 168209

11 years agoRemove trailing spaces.
Jakub Staszak [Fri, 16 Nov 2012 22:07:00 +0000 (22:07 +0000)]
Remove trailing spaces.

llvm-svn: 168208

11 years agoRemove hard coded registers in ARM ldrexd and strexd instructions
Weiming Zhao [Fri, 16 Nov 2012 21:55:34 +0000 (21:55 +0000)]
Remove hard coded registers in ARM ldrexd and strexd instructions

This patch replaces the hard coded GPR pair [R0, R1] of
Intrinsic:arm_ldrexd and [R2, R3] of Intrinsic:arm_strexd with
even/odd GPRPair reg class.
Similar to the lowering of atomic_64 operation.

llvm-svn: 168207

11 years agoStmtDumper: remove incomplete support for limiting the maximum dump depth.
Dmitri Gribenko [Fri, 16 Nov 2012 21:43:31 +0000 (21:43 +0000)]
StmtDumper: remove incomplete support for limiting the maximum dump depth.
There are better ways of limiting the amount of information if there is a need
for that.

Patch by Philip Craig.

llvm-svn: 168206

11 years ago<rdar://problem/12238339>
Greg Clayton [Fri, 16 Nov 2012 21:36:10 +0000 (21:36 +0000)]
<rdar://problem/12238339>

Make sure architectures are obeyed for skinny mach files.

llvm-svn: 168205

11 years agoFix lldb to compile with top of tree LLVM/clang.
Greg Clayton [Fri, 16 Nov 2012 21:35:22 +0000 (21:35 +0000)]
Fix lldb to compile with top of tree LLVM/clang.

llvm-svn: 168204

11 years agoUse array_pod_sort instead of std::sort.
Andrew Trick [Fri, 16 Nov 2012 21:33:38 +0000 (21:33 +0000)]
Use array_pod_sort instead of std::sort.

llvm-svn: 168203

11 years agotypo
Andrew Trick [Fri, 16 Nov 2012 21:33:35 +0000 (21:33 +0000)]
typo

llvm-svn: 168202

11 years agoAdd myself as owner of XCore Backend.
Richard Osborne [Fri, 16 Nov 2012 21:26:36 +0000 (21:26 +0000)]
Add myself as owner of XCore Backend.

llvm-svn: 168201

11 years agoMake sure FABS on v2f32 and v4f32 is legal on ARM NEON
Anton Korobeynikov [Fri, 16 Nov 2012 21:15:20 +0000 (21:15 +0000)]
Make sure FABS on v2f32 and v4f32 is legal on ARM NEON
This fixes PR14359

llvm-svn: 168200

11 years agoFix handling of aliases to functions.
Richard Osborne [Fri, 16 Nov 2012 21:12:38 +0000 (21:12 +0000)]
Fix handling of aliases to functions.

An alias to a function should use pc relative addressing.

llvm-svn: 168199

11 years ago[NVPTX] Order global variables in def-use order before emiting them in the final...
Justin Holewinski [Fri, 16 Nov 2012 21:03:51 +0000 (21:03 +0000)]
[NVPTX] Order global variables in def-use order before emiting them in the final assembly

llvm-svn: 168198

11 years agoPreserve address space of forward-referenced global variables in the LL parser
Justin Holewinski [Fri, 16 Nov 2012 21:03:47 +0000 (21:03 +0000)]
Preserve address space of forward-referenced global variables in the LL parser

Before, the parser would assert on the following code:

@a2 = global i8 addrspace(1)* @a
@a = addrspace(1) global i8 0

because the type of @a was "i8*" instead of "i8 addrspace(1)*" when parsing
the initializer for @a2.

llvm-svn: 168197

11 years agoMake this easier to understand, as suggested by Chandler.
Duncan Sands [Fri, 16 Nov 2012 20:53:08 +0000 (20:53 +0000)]
Make this easier to understand, as suggested by Chandler.

llvm-svn: 168196

11 years agoAdded program header emission
Hemant Kulkarni [Fri, 16 Nov 2012 20:51:32 +0000 (20:51 +0000)]
Added program header emission

llvm-svn: 168195

11 years agoRemove conditions from 'else if' that were guaranteed by preceding 'if'.
Craig Topper [Fri, 16 Nov 2012 20:01:39 +0000 (20:01 +0000)]
Remove conditions from 'else if' that were guaranteed by preceding 'if'.

llvm-svn: 168191

11 years ago<rdar://problem/12712928>
Greg Clayton [Fri, 16 Nov 2012 20:00:00 +0000 (20:00 +0000)]
<rdar://problem/12712928>

Removed an assert that was for debug only.

llvm-svn: 168190

11 years agoReleaseNotes.html updates reflecting PPC64 Elf Linux efforts and progress.
Will Schmidt [Fri, 16 Nov 2012 19:46:35 +0000 (19:46 +0000)]
ReleaseNotes.html updates reflecting PPC64 Elf Linux efforts and progress.

llvm-svn: 168189

11 years agoInstructionSimplify should be able to simplify A+B==B+A to 'true'
Duncan Sands [Fri, 16 Nov 2012 19:41:26 +0000 (19:41 +0000)]
InstructionSimplify should be able to simplify A+B==B+A to 'true'
but wasn't due to the same logic bug that caused PR14361.

llvm-svn: 168186

11 years agoUsing const cast to alleviate a warning.
Joe Abbey [Fri, 16 Nov 2012 19:38:42 +0000 (19:38 +0000)]
Using const cast to alleviate a warning.

A PR is being filed to address some code issues here.

llvm-svn: 168185

11 years agoUpdate RecordingMemoryManager as per change to RuntimeDyld base class
Daniel Malea [Fri, 16 Nov 2012 19:17:44 +0000 (19:17 +0000)]
Update RecordingMemoryManager as per change to RuntimeDyld base class
made in upstream llvm (r168114)

llvm-svn: 168184

11 years agoFactor out the final FADD that's common to multiple code paths in the visitLog* funct...
Craig Topper [Fri, 16 Nov 2012 19:08:44 +0000 (19:08 +0000)]
Factor out the final FADD that's common to multiple code paths in the visitLog* functions.

llvm-svn: 168183

11 years agomove irrelevant attribution.
Chris Lattner [Fri, 16 Nov 2012 18:58:23 +0000 (18:58 +0000)]
move irrelevant attribution.

llvm-svn: 168182

11 years agoFix PR14361: wrong simplification of A+B==B+A. You may think that the old logic
Duncan Sands [Fri, 16 Nov 2012 18:55:49 +0000 (18:55 +0000)]
Fix PR14361: wrong simplification of A+B==B+A.  You may think that the old logic
replaced by this patch is equivalent to the new logic, but you'd be wrong, and
that's exactly where the bug was.  There's a similar bug in instsimplify which
manifests itself as instsimplify failing to simplify this, rather than doing it
wrong, see next commit.

llvm-svn: 168181

11 years agoAdding new tests to test lli's pseudo-remote feature (-remote-mcjit).
Andrew Kaylor [Fri, 16 Nov 2012 18:51:59 +0000 (18:51 +0000)]
Adding new tests to test lli's pseudo-remote feature (-remote-mcjit).

llvm-svn: 168180

11 years agoFinally add myself to the credits.
Craig Topper [Fri, 16 Nov 2012 18:44:36 +0000 (18:44 +0000)]
Finally add myself to the credits.

llvm-svn: 168178

11 years agoFix partial-match-bind-behavior with forEachDescendant() matchers.
Daniel Jasper [Fri, 16 Nov 2012 18:39:22 +0000 (18:39 +0000)]
Fix partial-match-bind-behavior with forEachDescendant() matchers.

The problem is that a partial match of an (explicit or implicit) allOf matcher
binds results, i.e.

recordDecl(decl().bind("x"), hasName("A"))

can very well bind a record that is not named "A". With this fix, the common
cases of stumbling over this bug are fixed by the BoundNodesMap overwriting the
results of a partial match. An error can still be created with a weird
combination of anyOf and allOf (see inactive test). We need to decide whether
this is worth fixing, as the fix will have performance impact.

Review: http://llvm-reviews.chandlerc.com/D124
llvm-svn: 168177

11 years agoSimplifyCFG: Don't assume non-null ScalarTargetTransformInfo.
Hans Wennborg [Fri, 16 Nov 2012 18:22:08 +0000 (18:22 +0000)]
SimplifyCFG: Don't assume non-null ScalarTargetTransformInfo.

Patch by Pekka Jääskeläinen!

llvm-svn: 168176

11 years ago[analyzer] SATestBuild.py: fix broken string concatenation
Jordan Rose [Fri, 16 Nov 2012 17:41:21 +0000 (17:41 +0000)]
[analyzer] SATestBuild.py: fix broken string concatenation

This has been broken for a while, but the branch was never being taken.
(We were trying to do 'str + floatVal'; now we do 'str % intVal' and use
the '%d' format.)

llvm-svn: 168174

11 years agoWork around a layering violation from Target to CodeGen.
Benjamin Kramer [Fri, 16 Nov 2012 17:32:33 +0000 (17:32 +0000)]
Work around a layering violation from Target to CodeGen.

Technically this is still a layering violation but it's header-only which makes
it less harmful. No functionality change.

llvm-svn: 168173

11 years agoMake libASTMatchers link its dependencies. libRewriteFrontend depends on libRewriteCore.
Benjamin Kramer [Fri, 16 Nov 2012 17:30:58 +0000 (17:30 +0000)]
Make libASTMatchers link its dependencies. libRewriteFrontend depends on libRewriteCore.

LLVM_USED_LIBS does nothing here.

llvm-svn: 168172

11 years agollvm/test/CodeGen/X86/hipe-cc*.ll: Add explicit -mcpu, or they don't expect to pass...
NAKAMURA Takumi [Fri, 16 Nov 2012 16:07:37 +0000 (16:07 +0000)]
llvm/test/CodeGen/X86/hipe-cc*.ll: Add explicit -mcpu, or they don't expect to pass on Atom.

llvm-svn: 168171

11 years agoubsan: fix broken test case by using brand new [[@LINE]] FileCheck feature
Alexey Samsonov [Fri, 16 Nov 2012 13:38:03 +0000 (13:38 +0000)]
ubsan: fix broken test case by using brand new [[@LINE]] FileCheck feature

llvm-svn: 168170

11 years agoUBSan: enable CMake build on Darwin. Fix two failing tests and disable the test which...
Alexey Samsonov [Fri, 16 Nov 2012 12:57:07 +0000 (12:57 +0000)]
UBSan: enable CMake build on Darwin. Fix two failing tests and disable the test which produces linker errors on Darwin.

llvm-svn: 168169

11 years agoUBSan: enable proper linking with UBsan runtime on Darwin. Turn on building ubsan...
Alexey Samsonov [Fri, 16 Nov 2012 12:53:14 +0000 (12:53 +0000)]
UBSan: enable proper linking with UBsan runtime on Darwin. Turn on building ubsan on OS X in 'make' build system. Patch by Jean-Daniel Dupas.

llvm-svn: 168168

11 years agoUBSan: fix support on Darwin in 'make' build system. Patch by Jean-Daniel Dupas.
Alexey Samsonov [Fri, 16 Nov 2012 12:51:23 +0000 (12:51 +0000)]
UBSan: fix support on Darwin in 'make' build system. Patch by Jean-Daniel Dupas.

llvm-svn: 168167

11 years agoAdd the Erlang/HiPE calling convention, patch by Yiannis Tsiouris.
Duncan Sands [Fri, 16 Nov 2012 12:36:39 +0000 (12:36 +0000)]
Add the Erlang/HiPE calling convention, patch by Yiannis Tsiouris.

llvm-svn: 168166

11 years agoUpdate credits file regarding SROA work... Also include my other primary
Chandler Carruth [Fri, 16 Nov 2012 11:45:08 +0000 (11:45 +0000)]
Update credits file regarding SROA work... Also include my other primary
email address for completeness.

llvm-svn: 168165

11 years agoAdd myself as the code owner for a bunch of stuff.
Chandler Carruth [Fri, 16 Nov 2012 11:43:01 +0000 (11:43 +0000)]
Add myself as the code owner for a bunch of stuff.

Notably, I'm reviewing this stuff already, and this makes it clear I'm
on the hook for it.

I'm relying on Eric Christopher and Michael Spencer to help with Support
when there are platform specific issues, but Chris wants a single point
of contact, and I'm happy with that.

llvm-svn: 168164

11 years agoasan: sync with gcc r193553
Dmitry Vyukov [Fri, 16 Nov 2012 11:26:05 +0000 (11:26 +0000)]
asan: sync with gcc r193553

llvm-svn: 168151

11 years ago[TSan] use explicit ctor for BlockingCall
Alexey Samsonov [Fri, 16 Nov 2012 11:21:07 +0000 (11:21 +0000)]
[TSan] use explicit ctor for BlockingCall

llvm-svn: 168150

11 years agoAdd MCJIT test case for running global constructors.
Amara Emerson [Fri, 16 Nov 2012 11:17:00 +0000 (11:17 +0000)]
Add MCJIT test case for running global constructors.

llvm-svn: 168149

11 years agoAdd R_ARM_TARGET1 relocation to MCJIT and ensure that the runtime dyld reads
Amara Emerson [Fri, 16 Nov 2012 11:11:59 +0000 (11:11 +0000)]
Add R_ARM_TARGET1 relocation to MCJIT and ensure that the runtime dyld reads
all symbols during object loading, not just global ones.

This fixes JIT execution of code using llvm.global_ctors with internal
linkage constructors.

llvm-svn: 168148

11 years agoConstant::IsThreadDependent(): Use dyn_cast<Constant> instead of cast
Hans Wennborg [Fri, 16 Nov 2012 10:33:25 +0000 (10:33 +0000)]
Constant::IsThreadDependent(): Use dyn_cast<Constant> instead of cast

It turns out that the operands of a Constant are not always themselves
Constant. For example, one of the operands of BlockAddress is
BasicBlock, which is not a Constant.

This should fix the dragonegg-x86_64-linux-gcc-4.6-test build which
broke in r168037.

llvm-svn: 168147

11 years ago[TSan] use llvm-symbolizer to run tsan tests
Alexey Samsonov [Fri, 16 Nov 2012 10:16:14 +0000 (10:16 +0000)]
[TSan] use llvm-symbolizer to run tsan tests

llvm-svn: 168146

11 years agoStore this Decl* as a Decl* instead of a uintptr_t. No functionality change.
Nick Lewycky [Fri, 16 Nov 2012 08:40:59 +0000 (08:40 +0000)]
Store this Decl* as a Decl* instead of a uintptr_t. No functionality change.

llvm-svn: 168145

11 years ago[DebugInfo] Generate address ranges for compile units even if .debug_aranges is prese...
Alexey Samsonov [Fri, 16 Nov 2012 08:36:25 +0000 (08:36 +0000)]
[DebugInfo] Generate address ranges for compile units even if .debug_aranges is present: it is often the case that .debug_aranges section contains ranges only for a small subset of compile units. Test cases will be added in separate commits.

llvm-svn: 168144

11 years agoFactor some common code to reduce compile size.
Craig Topper [Fri, 16 Nov 2012 07:48:23 +0000 (07:48 +0000)]
Factor some common code to reduce compile size.

llvm-svn: 168143

11 years agoLoopVectorize: Division reductions generate incorrect code. Remove the part of the...
Nadav Rotem [Fri, 16 Nov 2012 06:51:17 +0000 (06:51 +0000)]
LoopVectorize: Division reductions generate incorrect code. Remove the part of the code that deals with divs.
Thanks to Paul Redmond for catching this while reviewing the code.

llvm-svn: 168142

11 years agoUse roundps/pd for llvm.ceil, llvm.trunc, llvm.rint, and llvm.nearbyint of vector...
Craig Topper [Fri, 16 Nov 2012 06:37:56 +0000 (06:37 +0000)]
Use roundps/pd for llvm.ceil, llvm.trunc, llvm.rint, and llvm.nearbyint of vector types.

llvm-svn: 168141

11 years agoChange RegisterContextLLDB's unwind logging to report which stack frame
Jason Molenda [Fri, 16 Nov 2012 06:15:40 +0000 (06:15 +0000)]
Change RegisterContextLLDB's unwind logging to report which stack frame
finally was able to restore a register, instead of just reporting the
frames that couldn't supply the reg.

llvm-svn: 168139

11 years agoFix typo in a comment.
Craig Topper [Fri, 16 Nov 2012 06:10:48 +0000 (06:10 +0000)]
Fix typo in a comment.

llvm-svn: 168138

11 years agoclang/unittests: Fixup corresponding to Doug's r168136.
NAKAMURA Takumi [Fri, 16 Nov 2012 04:40:11 +0000 (04:40 +0000)]
clang/unittests: Fixup corresponding to Doug's r168136.

llvm-svn: 168137

11 years agoSince CreateTargetInfo is taking ownership of the target options, pass
Douglas Gregor [Fri, 16 Nov 2012 04:24:59 +0000 (04:24 +0000)]
Since CreateTargetInfo is taking ownership of the target options, pass
it as a pointer.

llvm-svn: 168136

11 years agoRemove unused diagnostics
Douglas Gregor [Fri, 16 Nov 2012 04:10:20 +0000 (04:10 +0000)]
Remove unused diagnostics

llvm-svn: 168135

11 years agoRevert r168132, "clang/test/lit.cfg: Suppress the feature 'dev-fd-fs' for now." I...
NAKAMURA Takumi [Fri, 16 Nov 2012 03:50:55 +0000 (03:50 +0000)]
Revert r168132, "clang/test/lit.cfg: Suppress the feature 'dev-fd-fs' for now." I missed Daniel's r168095 by my accident.

llvm-svn: 168134

11 years ago[libclang] When caching code-completion results, pass the CachedCompletionAllocator
Argyrios Kyrtzidis [Fri, 16 Nov 2012 03:34:57 +0000 (03:34 +0000)]
[libclang] When caching code-completion results, pass the CachedCompletionAllocator
to the CodeCompletionTUInfo that is going to be used to get the results.

Previously we would use ASTUnit's CodeCompletionTUInfo which has its own allocator
that will go away when we reparse. That could result in a use-after-free bug when
getting the parent context name from a CodeCompletionString.

Addresses rdar://12568377.

llvm-svn: 168133

11 years agoclang/test/lit.cfg: Suppress the feature 'dev-fd-fs' for now.
NAKAMURA Takumi [Fri, 16 Nov 2012 03:28:51 +0000 (03:28 +0000)]
clang/test/lit.cfg: Suppress the feature 'dev-fd-fs' for now.

None of my buildhost (centos6 x86-64 and cygwin) is passing. Investigating.

llvm-svn: 168132