platform/upstream/llvm.git
9 years agoRestores r228382, which was reverted in r228406.
Sameer Sahasrabuddhe [Wed, 25 Feb 2015 05:48:23 +0000 (05:48 +0000)]
Restores r228382, which was reverted in r228406.

The original commit failed to handle "shift assign" (<<=), which
broke the test mentioned in r228406. This is now fixed and the
test added to the lit tests under SemaOpenCL.

*** Original commit message from r228382 ***

OpenCL: handle shift operator with vector operands

Introduce a number of checks:
1. If LHS is a scalar, then RHS cannot be a vector.
2. Operands must be of integer type.
3. If both are vectors, then the number of elements must match.

Relax the requirement for "usual arithmetic conversions":
When LHS is a vector, a scalar RHS can simply be expanded into a
vector; OpenCL does not require that its rank be lower than the LHS.
For example, the following code is not an error even if the implicit
type of the constant literal is "int".

  char2 foo(char2 v) { return v << 1; }

Consolidate existing tests under CodeGenOpenCL, and add more tests
under SemaOpenCL.

llvm-svn: 230464

9 years agoFix CommandLineCompletionTestCase.test_watchpoint_xxx tests after r230408
Ilia K [Wed, 25 Feb 2015 05:47:10 +0000 (05:47 +0000)]
Fix CommandLineCompletionTestCase.test_watchpoint_xxx tests after r230408

llvm-svn: 230463

9 years ago[IC] Turn non-null MD on pointer loads to range MD on integer loads.
Charles Davis [Wed, 25 Feb 2015 05:10:25 +0000 (05:10 +0000)]
[IC] Turn non-null MD on pointer loads to range MD on integer loads.

Summary:
This change fixes the FIXME that you recently added when you committed
(a modified version of) my patch.  When `InstCombine` combines a load and
store of an pointer to those of an equivalently-sized integer, it currently
drops any `!nonnull` metadata that might be present.  This change replaces
`!nonnull` metadata with `!range !{ 1, -1 }` metadata instead.

Reviewers: chandlerc

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D7621

llvm-svn: 230462

9 years agoRevert 230419, 230425, 230432.
Nico Weber [Wed, 25 Feb 2015 04:39:21 +0000 (04:39 +0000)]
Revert 230419, 230425, 230432.

They don't build on Windows.
http://lab.llvm.org:8011/builders/sanitizer-windows/ went red for example.

llvm-svn: 230461

9 years agoProduce less broken basic block sequences for __finally blocks.
Nico Weber [Wed, 25 Feb 2015 04:05:18 +0000 (04:05 +0000)]
Produce less broken basic block sequences for __finally blocks.

The way cleanups (such as PerformSEHFinally) get emitted is that codegen
generates some initialization code, then calls the cleanup's Emit() with the
insertion point set to a good place, then the cleanup is supposed to emit its
stuff, and then codegen might tack in a jump or similar to where the insertion
point is after the cleanup.

The PerformSEHFinally cleanup tries to just stash away the block it's supposed
to codegen into, and then does codegen later, into that stashed block.  However,
after codegen'ing the __finally block, it used to set the insertion point to
the finally's continuation block (where the __finally cleanup goes when its body
is completed after regular, non-exceptional control flow).  That's not correct,
as that block can (and generally does) already ends in a jump.  Instead,
remember the insertion point that was current before the __finally got emitted,
and restore that.

Fixes two of the crashes in PR22553.

llvm-svn: 230460

9 years agoAdd comments for two CleanupKinds.
Nico Weber [Wed, 25 Feb 2015 03:58:36 +0000 (03:58 +0000)]
Add comments for two CleanupKinds.

llvm-svn: 230459

9 years agodocs: Document optimizations in control flow integrity design doc.
Peter Collingbourne [Wed, 25 Feb 2015 03:35:03 +0000 (03:35 +0000)]
docs: Document optimizations in control flow integrity design doc.

llvm-svn: 230458

9 years agoAdd some missing #includes and forward declarations found by modules build.
Richard Smith [Wed, 25 Feb 2015 03:12:03 +0000 (03:12 +0000)]
Add some missing #includes and forward declarations found by modules build.

llvm-svn: 230457

9 years agoRevert "Adapt clang-tools-extra to clang module format changes."
Adrian Prantl [Wed, 25 Feb 2015 02:46:37 +0000 (02:46 +0000)]
Revert "Adapt clang-tools-extra to clang module format changes."

This reverts commit 230424.

llvm-svn: 230456

9 years agoRevert "Add a missing target requirement."
Adrian Prantl [Wed, 25 Feb 2015 02:46:29 +0000 (02:46 +0000)]
Revert "Add a missing target requirement."

This reverts commit 230430.

llvm-svn: 230455

9 years agoRevert "Wrap clang module files in a Mach-O, ELF, or COFF container."
Adrian Prantl [Wed, 25 Feb 2015 02:44:04 +0000 (02:44 +0000)]
Revert "Wrap clang module files in a Mach-O, ELF, or COFF container."

llvm-svn: 230454

9 years agoRevert "Add a missing target requirement."
Adrian Prantl [Wed, 25 Feb 2015 02:43:38 +0000 (02:43 +0000)]
Revert "Add a missing target requirement."

llvm-svn: 230453

9 years agobuild: check if atomic routines are implicitly provided
Saleem Abdulrasool [Wed, 25 Feb 2015 02:38:03 +0000 (02:38 +0000)]
build: check if atomic routines are implicitly provided

It is possible for the atomic routines to be provided by the compiler without
requiring any additional libraries.  Check if that is the case before checking
for a library.

Patch by Matt Glazar!

llvm-svn: 230452

9 years agoWhen FileSpec::Resolve is given a bare file like "ls",
Jason Molenda [Wed, 25 Feb 2015 02:35:25 +0000 (02:35 +0000)]
When FileSpec::Resolve is given a bare file like "ls",
and llvm::sys::fs::make_absolute prepends the current
working directory to that path, leave the original
bare file name unchanged if $cwd/ls doesn't exist.

http://reviews.llvm.org/D7477
<rdar://problem/18775190>

llvm-svn: 230451

9 years agoFix bots failing on an explicit triple
Reid Kleckner [Wed, 25 Feb 2015 02:32:00 +0000 (02:32 +0000)]
Fix bots failing on an explicit triple

Why do we need a registered target for this?

llvm-svn: 230450

9 years agoDon't crash on missing '{' after __except or __finally, PR22687.
Nico Weber [Wed, 25 Feb 2015 02:22:06 +0000 (02:22 +0000)]
Don't crash on missing '{' after __except or __finally, PR22687.

Also add some general test/Parser coverage for SEH blocks.

llvm-svn: 230449

9 years agoReland r230314 "Fix codegen for virtual methods that are (re-) exported from multiple...
Reid Kleckner [Wed, 25 Feb 2015 02:19:56 +0000 (02:19 +0000)]
Reland r230314 "Fix codegen for virtual methods that are (re-) exported from multiple modules."

This reverts the revert from commit r230406.

The changes in r230445 and r230446 make the test pass on Windows now.

llvm-svn: 230448

9 years agoMS ABI: Add triple to test relying on key functions
Reid Kleckner [Wed, 25 Feb 2015 02:16:15 +0000 (02:16 +0000)]
MS ABI: Add triple to test relying on key functions

llvm-svn: 230447

9 years agoMS ABI: Try to respect external AST source record layouts
Reid Kleckner [Wed, 25 Feb 2015 02:16:09 +0000 (02:16 +0000)]
MS ABI: Try to respect external AST source record layouts

Covered by existing tests in test/CodeGen/override-layout.c and
test/CodeGenCXX/override-layout.cpp. Seriously, they found real bugs in
my code. :)

llvm-svn: 230446

9 years agoAdd missing MD canonicalization to MS vtable builder
Reid Kleckner [Wed, 25 Feb 2015 02:16:02 +0000 (02:16 +0000)]
Add missing MD canonicalization to MS vtable builder

llvm-svn: 230445

9 years ago[asan] one more attempt to fix windows build
Kostya Serebryany [Wed, 25 Feb 2015 02:00:26 +0000 (02:00 +0000)]
[asan] one more attempt to fix windows build

llvm-svn: 230432

9 years agoAdd a missing target requirement.
Adrian Prantl [Wed, 25 Feb 2015 01:52:10 +0000 (01:52 +0000)]
Add a missing target requirement.

llvm-svn: 230430

9 years agoAdd a missing target requirement.
Adrian Prantl [Wed, 25 Feb 2015 01:52:05 +0000 (01:52 +0000)]
Add a missing target requirement.

llvm-svn: 230429

9 years agoRemove slow and apparently pointless updating of all identifiers at the start
Richard Smith [Wed, 25 Feb 2015 01:45:32 +0000 (01:45 +0000)]
Remove slow and apparently pointless updating of all identifiers at the start
of writing out an AST file.

llvm-svn: 230428

9 years ago[modules] Add include/llvm/IR/DebugInfoFlags.def to the textual headers list.
Richard Smith [Wed, 25 Feb 2015 01:44:09 +0000 (01:44 +0000)]
[modules] Add include/llvm/IR/DebugInfoFlags.def to the textual headers list.

llvm-svn: 230427

9 years agoFold ParseSEHTryBlockCommon() into its only caller. No behavior change.
Nico Weber [Wed, 25 Feb 2015 01:43:27 +0000 (01:43 +0000)]
Fold ParseSEHTryBlockCommon() into its only caller. No behavior change.

ParseCXXTryBlockCommon() makes sense because it has two callers due to
function try blocks.  There are no SEH function try blocks.

llvm-svn: 230426

9 years ago[asan] attempting to fix the windows build
Kostya Serebryany [Wed, 25 Feb 2015 01:41:42 +0000 (01:41 +0000)]
[asan] attempting to fix the windows build

llvm-svn: 230425

9 years agoAdapt clang-tools-extra to clang module format changes.
Adrian Prantl [Wed, 25 Feb 2015 01:32:04 +0000 (01:32 +0000)]
Adapt clang-tools-extra to clang module format changes.
- add clangCodeGen.a to the tools that need it
- tweak pp-trace command line handling to not conflict with clang's.

llvm-svn: 230424

9 years agoWrap clang module files in a Mach-O, ELF, or COFF container.
Adrian Prantl [Wed, 25 Feb 2015 01:31:45 +0000 (01:31 +0000)]
Wrap clang module files in a Mach-O, ELF, or COFF container.
This is a necessary prerequisite for debugging with modules.
The .pcm files become containers that hold the serialized AST which allows
us to store debug information in the module file that can be shared by all
object files that were built importing the module.

This reapplies r230044 with a fixed configure+make build and updated
dependencies and testcase requirements. Over the last iteration this
version adds
- missing target requirements for testcases that specify an x86 triple,
- a missing clangCodeGen.a dependency to libClang.a in the make build.

rdar://problem/19104245

llvm-svn: 230423

9 years agoSuppress 'warning C4062: enumerator X in switch of enum Y is not handled' from system...
Michael J. Spencer [Wed, 25 Feb 2015 01:30:13 +0000 (01:30 +0000)]
Suppress 'warning C4062: enumerator X in switch of enum Y is not handled' from system header.

llvm-svn: 230422

9 years ago[PowerPC] Add triples to QPX tests
Hal Finkel [Wed, 25 Feb 2015 01:26:59 +0000 (01:26 +0000)]
[PowerPC] Add triples to QPX tests

Some of these tests fail on Darwin systems because of a lack of a triple;
fix that.

llvm-svn: 230421

9 years ago[GC] Document the recently added PlaceSafepoints and RewriteGCForStatepoints passes
Philip Reames [Wed, 25 Feb 2015 01:23:59 +0000 (01:23 +0000)]
[GC] Document the recently added PlaceSafepoints and RewriteGCForStatepoints passes

llvm-svn: 230420

9 years ago[asan] add __asan_default_suppressions() hook
Kostya Serebryany [Wed, 25 Feb 2015 01:15:37 +0000 (01:15 +0000)]
[asan] add __asan_default_suppressions() hook

llvm-svn: 230419

9 years agoFix logging in GDBRemoteCommunicationServerPlatform.
Oleksiy Vyalov [Wed, 25 Feb 2015 01:11:38 +0000 (01:11 +0000)]
Fix logging in GDBRemoteCommunicationServerPlatform.

llvm-svn: 230418

9 years agoMake RedeclarableResult's ownership-transferring constructor be a move
Richard Smith [Wed, 25 Feb 2015 01:11:29 +0000 (01:11 +0000)]
Make RedeclarableResult's ownership-transferring constructor be a move
constructor.

llvm-svn: 230417

9 years agollvm-dis: Stop crashing when dropping debug info
Duncan P. N. Exon Smith [Wed, 25 Feb 2015 01:10:03 +0000 (01:10 +0000)]
llvm-dis: Stop crashing when dropping debug info

Since r199356, we've printed a warning when dropping debug info.
r225562 started crashing on that, since it registered a diagnostic
handler that only expected errors.  This fixes the handler to expect
other severities.  As a side effect, it now prints "error: " at the
start of error messages, similar to `llvm-as`.

There was a testcase for r199356, but it only really checked the
assembler.  Move `test/Bitcode/drop-debug-info.ll` to `test/Assembler`,
and introduce `test/Bitcode/drop-debug-info.3.5.ll` (and companion
`.bc`) to test the bitcode reader.

Note: tools/gold/gold-plugin.cpp has an equivalent bug, but I'm not sure
what the best fix is there.  I'll file a PR.

llvm-svn: 230416

9 years ago[opaque pointer type] Bitcode support for explicit type parameter on GEP.
David Blaikie [Wed, 25 Feb 2015 01:08:52 +0000 (01:08 +0000)]
[opaque pointer type] Bitcode support for explicit type parameter on GEP.

Like r230414, add bitcode support including backwards compatibility, for
an explicit type parameter to GEP.

At the suggestion of Duncan I tried coalescing the two older bitcodes into a
single new bitcode, though I did hit a wrinkle: I couldn't figure out how to
create an explicit abbreviation for a record with a variable number of
arguments (the indicies to the gep). This means the discriminator between
inbounds and non-inbounds gep is a full variable-length field I believe? Is my
understanding correct? Is there a way to create such an abbreviation? Should I
just use two bitcodes as before?

Reviewers: dexonsmith

Differential Revision: http://reviews.llvm.org/D7736

llvm-svn: 230415

9 years ago[opaque pointer type] bitcode support for explicit type parameter to the load instruction
David Blaikie [Wed, 25 Feb 2015 01:07:20 +0000 (01:07 +0000)]
[opaque pointer type] bitcode support for explicit type parameter to the load instruction

Summary:
I've taken my best guess at this, but I've cargo culted in places & so
explanations/corrections would be great.

This seems to pass all the tests (check-all, covering clang and llvm) so I
believe that pretty well exercises both the backwards compatibility and common
(same version) compatibility given the number of checked in bitcode files we
already have. Is that a reasonable approach to testing here? Would some more
explicit tests be desired?

1) is this the right way to do back-compat in this case (looking at the number
  of entries in the bitcode record to disambiguate between the old schema and
  the new?)

2) I don't quite understand the logarithm logic to choose the encoding type of
  the type parameter in the abbreviation description, but I found another
  instruction doing the same thing & it seems to work. Is that the right
  approach?

Reviewers: dexonsmith

Differential Revision: http://reviews.llvm.org/D7655

llvm-svn: 230414

9 years ago[PowerPC] Add support for the QPX vector instruction set
Hal Finkel [Wed, 25 Feb 2015 01:06:45 +0000 (01:06 +0000)]
[PowerPC] Add support for the QPX vector instruction set

This adds support for the QPX vector instruction set, which is used by the
enhanced A2 cores on the IBM BG/Q supercomputers. QPX vectors are 256 bytes
wide, holding 4 double-precision floating-point values. Boolean values, modeled
here as <4 x i1> are actually also represented as floating-point values
(essentially  { -1, 1 } for { false, true }). QPX shares many features with
Altivec and VSX, but is distinct from both of them. One major difference is
that, instead of adding completely-separate vector registers, QPX vector
registers are extensions of the scalar floating-point registers (lane 0 is the
corresponding scalar floating-point value). The operations supported on QPX
vectors mirrors that supported on the scalar floating-point values (with some
additional ones for permutations and logical/comparison operations).

I've been maintaining this support out-of-tree, as part of the bgclang project,
for several years. This is not the entire bgclang patch set, but is most of the
subset that can be cleanly integrated into LLVM proper at this time. Adding
this to the LLVM backend is part of my efforts to rebase bgclang to the current
LLVM trunk, but is independently useful (especially for codes that use LLVM as
a JIT in library form).

The assembler/disassembler test coverage is complete. The CodeGen test coverage
is not, but I've included some tests, and more will be added as follow-up work.

llvm-svn: 230413

9 years ago[modules] Fix a bug that would result in a build with P paths through a module
Richard Smith [Wed, 25 Feb 2015 00:56:02 +0000 (00:56 +0000)]
[modules] Fix a bug that would result in a build with P paths through a module
graph with M modules to take O(P) time, not just O(M) time, when using explicit
module builds.

llvm-svn: 230412

9 years agoSupport SHF_MERGE sections in COMDATs.
Rafael Espindola [Wed, 25 Feb 2015 00:52:15 +0000 (00:52 +0000)]
Support SHF_MERGE sections in COMDATs.

This patch unifies the comdat and non-comdat code paths. By doing this
it add missing features to the comdat side and removes the fixed
section assumptions from the non-comdat side.

In ELF there is no one true section for "4 byte mergeable" constants.
We are better off computing the required properties of the section
and asking the context for it.

llvm-svn: 230411

9 years agoBitcodeWriter: Refactor common computation of bits required for a type index.
David Blaikie [Wed, 25 Feb 2015 00:51:52 +0000 (00:51 +0000)]
BitcodeWriter: Refactor common computation of bits required for a type index.

Suggested by Duncan. Happy to bikeshed the name, cache the result, etc.

llvm-svn: 230410

9 years ago[asan] add suppressions for odr violations
Kostya Serebryany [Wed, 25 Feb 2015 00:49:12 +0000 (00:49 +0000)]
[asan] add suppressions for odr violations

llvm-svn: 230409

9 years agoIf you try to auto-complete "target symbols<TAB>" you get "target symbolsadd" instead...
Greg Clayton [Wed, 25 Feb 2015 00:32:43 +0000 (00:32 +0000)]
If you try to auto-complete "target symbols<TAB>" you get "target symbolsadd" instead of "target symbols ".

Fix this by returning the fact that the "symbols" word is complete if there is nothing else to complete after the "symbols" word.

<rdar://problem/19164599>

llvm-svn: 230408

9 years agoFix consistently wrong sphinx markup
Philip Reames [Wed, 25 Feb 2015 00:22:07 +0000 (00:22 +0000)]
Fix consistently wrong sphinx markup

I'd been using '' where I should have been using ``.

llvm-svn: 230407

9 years agoRevert r230314, "Fix codegen for virtual methods that are (re-) exported from multipl...
NAKAMURA Takumi [Wed, 25 Feb 2015 00:18:22 +0000 (00:18 +0000)]
Revert r230314, "Fix codegen for virtual methods that are (re-) exported from multiple modules."

It crashes for targeting (i686|x86_64)-win32.

  clang: clang/lib/AST/VTableBuilder.cpp:142: {anonymous}::FinalOverriders::OverriderInfo {anonymous}::FinalOverriders::getOverrider(const clang::CXXMethodDecl*, clang::CharUnits) const: Assertion `OverridersMap.count(std::make_pair(MD, BaseOffset)) && "Did not find overrider!"' failed.

llvm-svn: 230406

9 years agoUpdate the GC docs to explicitly mention both gcroot and gc.statepoint
Philip Reames [Wed, 25 Feb 2015 00:18:04 +0000 (00:18 +0000)]
Update the GC docs to explicitly mention both gcroot and gc.statepoint

Also, fix confusing bit of the gcroot documentation that bit me personally.

llvm-svn: 230405

9 years agoMake this test even more OS and register allocation neutral.
Eric Christopher [Wed, 25 Feb 2015 00:12:11 +0000 (00:12 +0000)]
Make this test even more OS and register allocation neutral.

llvm-svn: 230404

9 years ago[GC] Sync documentation with code naming
Philip Reames [Tue, 24 Feb 2015 23:57:26 +0000 (23:57 +0000)]
[GC] Sync documentation with code naming

Fixing an issue pointed out by Sean Silva.  Thanks!

llvm-svn: 230403

9 years agoMore GC documentation cleanup
Philip Reames [Tue, 24 Feb 2015 23:51:37 +0000 (23:51 +0000)]
More GC documentation cleanup

llvm-svn: 230402

9 years agoFixed Xcode project after recent changes:
Greg Clayton [Tue, 24 Feb 2015 23:50:21 +0000 (23:50 +0000)]
Fixed Xcode project after recent changes:

"After recent changes, some code has become redundant. This revision tries to remove
the un-used code and tidy up the rest.

Following 4 files have been removed. I have updated CMake files and checked that it builds
fine on Linux and Windows. Can somebody update the xcode related file accordingly?

tools/lldb-mi/MICmnStreamStdinLinux.cpp
tools/lldb-mi/MICmnStreamStdinLinux.h
tools/lldb-mi/MICmnStreamStdinWindows.cpp
tools/lldb-mi/MICmnStreamStdinWindows.h"

llvm-svn: 230401

9 years agoMake this test not dependent upon the triple. All that was needed
Eric Christopher [Tue, 24 Feb 2015 23:43:26 +0000 (23:43 +0000)]
Make this test not dependent upon the triple. All that was needed
was some flexibility in the check line for the comment basic block.

llvm-svn: 230400

9 years ago[CMake] Clean up lldb-mi CMakeLists.txt.
Zachary Turner [Tue, 24 Feb 2015 23:36:12 +0000 (23:36 +0000)]
[CMake] Clean up lldb-mi CMakeLists.txt.

It was maintaining two huge file lists that are identical except
for 1 file.  Make this work the proper way.

llvm-svn: 230398

9 years agoMore GC doc cleanup
Philip Reames [Tue, 24 Feb 2015 23:34:24 +0000 (23:34 +0000)]
More GC doc cleanup

Mostly minor wording changes for readability.  Nothing major to see here.

llvm-svn: 230397

9 years ago[CMake] Set policy CMP0051 to OLD globally.
Zachary Turner [Tue, 24 Feb 2015 23:32:47 +0000 (23:32 +0000)]
[CMake] Set policy CMP0051 to OLD globally.

When you use generator expressions in a library sources list,
and then later access the SOURCES property, the OLD behavior
(CMake 3.0 and earlier) would not include these expressions in
the SOURCES property.  The NEW behavior (starting in CMake 3.1)
is that they do include the generator expressions in the SOURCES
property.

Differential Revision: http://reviews.llvm.org/D7870
Reviewed By: Chris Bieneman

llvm-svn: 230396

9 years agoclang-format: Change location of stashed token
Jacques Pienaar [Tue, 24 Feb 2015 23:23:24 +0000 (23:23 +0000)]
clang-format: Change location of stashed token

Commit of patch in http://reviews.llvm.org/D7871

llvm-svn: 230395

9 years agoLowerBitSets: Introduce global layout builder.
Peter Collingbourne [Tue, 24 Feb 2015 23:17:02 +0000 (23:17 +0000)]
LowerBitSets: Introduce global layout builder.

The builder is based on a layout algorithm that tries to keep members of
small bit sets together. The new layout compresses Chromium's bit sets to
around 15% of their original size.

Differential Revision: http://reviews.llvm.org/D7796

llvm-svn: 230394

9 years agoImprove the getting started instructions in the GC docs
Philip Reames [Tue, 24 Feb 2015 23:12:27 +0000 (23:12 +0000)]
Improve the getting started instructions in the GC docs

This is still gcroot vs gc.statepoint agnostic.  I'm just trying to clarify the general documentation at this point.

llvm-svn: 230393

9 years agoPrologEpilogInserter: Clean up math in calculateFrameObjectOffsets
David Majnemer [Tue, 24 Feb 2015 23:08:13 +0000 (23:08 +0000)]
PrologEpilogInserter: Clean up math in calculateFrameObjectOffsets

There is no need to open-code the alignment calculation, we have a
handy RoundUpToAlignment function which "Does The Right Thing (TM)".

llvm-svn: 230392

9 years agoremove function names from comments; NFC
Sanjay Patel [Tue, 24 Feb 2015 22:43:06 +0000 (22:43 +0000)]
remove function names from comments; NFC

llvm-svn: 230391

9 years agoFix qLaunchGDBServer packet parsing in GDBRemoteCommunicationServerPlatform.
Oleksiy Vyalov [Tue, 24 Feb 2015 22:23:39 +0000 (22:23 +0000)]
Fix qLaunchGDBServer packet parsing in GDBRemoteCommunicationServerPlatform.

llvm-svn: 230390

9 years agoResubmit "[CMake] Change lldbAPI to be a CMake OBJECT library."
Zachary Turner [Tue, 24 Feb 2015 22:17:57 +0000 (22:17 +0000)]
Resubmit "[CMake] Change lldbAPI to be a CMake OBJECT library."

This resubmits r230380.  The primary cause of the failure was
actually just a warning, which we can disable at the CMake level
in a followup patch on the LLVM side.  The other thing which was
actually an error on the bot should be able to be fixed with
a clean.

llvm-svn: 230389

9 years agoReapplied D7816 & rL230177 & rL230278 - with an additional fix toensure that the...
Simon Pilgrim [Tue, 24 Feb 2015 22:08:56 +0000 (22:08 +0000)]
Reapplied D7816 & rL230177 & rL230278 - with an additional fix toensure that the smallest build vector input scalar type is always used. Additional (crash) test cases already committed.

llvm-svn: 230388

9 years agoInstrProf: Simplify a couple of tests after r230383
Justin Bogner [Tue, 24 Feb 2015 21:49:28 +0000 (21:49 +0000)]
InstrProf: Simplify a couple of tests after r230383

Most of the checks in these two tests were actually testing the
behaviour of the instrprof LLVM pass. Now that we're testing that
specifically in LLVM's test suite, it's better if we only test the
frontend's behaviour here.

llvm-svn: 230387

9 years agoAdded test case for PR22678 (check CONCAT_VECTORS DAG combiner pass doesn't introduce...
Simon Pilgrim [Tue, 24 Feb 2015 21:46:23 +0000 (21:46 +0000)]
Added test case for PR22678 (check CONCAT_VECTORS DAG combiner pass doesn't introduce illegal types)

llvm-svn: 230386

9 years agoCUDA: Add option to allow host device functions to call host functions
Jacques Pienaar [Tue, 24 Feb 2015 21:45:33 +0000 (21:45 +0000)]
CUDA: Add option to allow host device functions to call host functions

Commiting code from review http://reviews.llvm.org/D7841

llvm-svn: 230385

9 years agoPR22673 again: diagnose use of the used decl, not the found decl. This is also
Richard Smith [Tue, 24 Feb 2015 21:44:43 +0000 (21:44 +0000)]
PR22673 again: diagnose use of the used decl, not the found decl. This is also
wrong (DiagnoseUseOfDecl should take both), but it's more consistent with what
we do in other places.

llvm-svn: 230384

9 years agoInstrProf: Test for appropriate linkage of the profiling structures
Justin Bogner [Tue, 24 Feb 2015 21:42:42 +0000 (21:42 +0000)]
InstrProf: Test for appropriate linkage of the profiling structures

This test checks that the symbols instrprof creates have appropriate
linkage. The tests already exist in clang in a slightly different form
from before we sunk profile generation into an LLVM pass, but that's
an awkward place for them now. I'll remove/simplify the clang versions
shortly.

llvm-svn: 230383

9 years agoRevert "[CMake] Change lldbAPI to be a CMake OBJECT library."
Zachary Turner [Tue, 24 Feb 2015 21:28:01 +0000 (21:28 +0000)]
Revert "[CMake] Change lldbAPI to be a CMake OBJECT library."

This reverts commit r230380.  It causes CMake to fail on certain
machines with an error about needing to use string(STRIP_GENEX).

llvm-svn: 230382

9 years agoPR22673: Don't forget to check a constructor for deletedness when we use it to
Richard Smith [Tue, 24 Feb 2015 21:16:19 +0000 (21:16 +0000)]
PR22673: Don't forget to check a constructor for deletedness when we use it to
implicitly construct a temporary in a reference binding.

llvm-svn: 230381

9 years ago[CMake] Change lldbAPI to be a CMake OBJECT library.
Zachary Turner [Tue, 24 Feb 2015 20:58:39 +0000 (20:58 +0000)]
[CMake] Change lldbAPI to be a CMake OBJECT library.

An OBJECT library is a special type of CMake library that produces
no archive, has no link interface, and no link inputs.  It is like
a regular archive, just without the physical output.  To link
against an OBJECT library, you reference it in the *source* file
list of a library using the special syntax $<TARGET_OBJECTS:lldbAPI>.
This will cause every object file to be passed to the linker
independently, as opposed to a single archive being passed to the
linker.

This is *extremely* important on Windows.  lldbAPI exports all of the
SB classes using __declspec(dllexport).  Unfortunately for technical
reasons it is not possible (well, extremely difficult) to get the
linker to propagate a __declspec(dllexport) attribute from a symbol
in an object file in an archive to a DLL that links against that
archive.  The solution to this is for the DLL to link the object files
directly.  So lldbAPI must be an OBJECT library.

This fixes an issue that has been present since the duplicated
lldbAPI file lists were removed, which would cause linker failures.

As a side effect, this also makes LLDB_DISABLE_PYTHON=1 work again
on Windows, which was previously totally broken.

llvm-svn: 230380

9 years ago[CMake] Don't automatically copy python27.dll.
Zachary Turner [Tue, 24 Feb 2015 20:58:24 +0000 (20:58 +0000)]
[CMake] Don't automatically copy python27.dll.

This is generating problems when you have built both debug and
release python.  For now I just want to get CMake to work, I
will work on a more robust fix later.  In the meantime you may
need to copy python27(_d).dll manually to ninja\bin after
building.

llvm-svn: 230379

9 years agoFixing eol-style
Andrew Kaylor [Tue, 24 Feb 2015 20:49:35 +0000 (20:49 +0000)]
Fixing eol-style

llvm-svn: 230378

9 years agoAdd test that we reject invocations of deleted constructors. One of these
Richard Smith [Tue, 24 Feb 2015 20:41:36 +0000 (20:41 +0000)]
Add test that we reject invocations of deleted constructors. One of these
constructs is not rejected by should be; that's PR22673.

llvm-svn: 230370

9 years agoReleaseNotes: add notes about VS2014 Update 4 requirement
Hans Wennborg [Tue, 24 Feb 2015 20:36:40 +0000 (20:36 +0000)]
ReleaseNotes: add notes about VS2014 Update 4 requirement

llvm-svn: 230363

9 years agoMS extensions: Properly diagnose address of MS property decl
Reid Kleckner [Tue, 24 Feb 2015 20:29:40 +0000 (20:29 +0000)]
MS extensions: Properly diagnose address of MS property decl

Summary: Fixes PR22671.

Reviewers: rsmith

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D7863

llvm-svn: 230362

9 years agoFix typo: qual -> equal
Arch D. Robison [Tue, 24 Feb 2015 20:11:49 +0000 (20:11 +0000)]
Fix typo: qual -> equal

llvm-svn: 230361

9 years agoAdd .fpu directives to ARM unwind save & restore functions.
Jonathan Roelofs [Tue, 24 Feb 2015 20:09:29 +0000 (20:09 +0000)]
Add .fpu directives to ARM unwind save & restore functions.

TODO: The iwmmx register save & restore functions still need the same treatment.
      I didn't do that in this patch because the integrated assembler has a bug
      where it refuses to build them on -march=armv6-m, even with
         .march armv5t
         .arm
      in front of those functions.

This should fix PR22384.

http://reviews.llvm.org/D7258

llvm-svn: 230360

9 years agoStarting to cleanup the garbage collection documentation
Philip Reames [Tue, 24 Feb 2015 19:44:46 +0000 (19:44 +0000)]
Starting to cleanup the garbage collection documentation

In this change:
- Put the getting started section first
- Create a dedicated section to document the built in collector strategies
- Move discuss of ShadowStack into new section
- Add placeholders for erlang, ocaml, and statepoint-example collectors

There will be many more changes following.  I plan on full integrating the documentation for gc.statepoint and gc.root.  I want to make it much clearer on how to get started and what users should expect in terms of effort.

llvm-svn: 230359

9 years agoRevert:
Eric Christopher [Tue, 24 Feb 2015 19:11:00 +0000 (19:11 +0000)]
Revert:

Author: Simon Pilgrim <llvm-dev@redking.me.uk>
Date:   Mon Feb 23 23:04:28 2015 +0000

    Fix based on post-commit comment on D7816 & rL230177 - BUILD_VECTOR operand truncation was using the the BV's output scalar type instead of the input type.

and

Author: Simon Pilgrim <llvm-dev@redking.me.uk>
Date:   Sun Feb 22 18:17:28 2015 +0000

    [DagCombiner] Generalized BuildVector Vector Concatenation

    The CONCAT_VECTORS combiner pass can transform the concat of two BUILD_VECTOR nodes into a single BUILD_VECTOR node.

    This patch generalises this to support any number of BUILD_VECTOR nodes, and also permits UNDEF nodes to be included as well.

    This was noticed as AVX vec128 -> vec256 canonicalization sometimes creates a CONCAT_VECTOR with a real vec128 lower and an vec128 UNDEF upper.

    Differential Revision: http://reviews.llvm.org/D7816

as the root cause of PR22678 which is causing an assertion inside the DAG combiner.

I'll follow up to the main thread as well.

llvm-svn: 230358

9 years agoRename UpdateRegAllocHint to match style guidelines.
Eric Christopher [Tue, 24 Feb 2015 19:10:57 +0000 (19:10 +0000)]
Rename UpdateRegAllocHint to match style guidelines.

llvm-svn: 230357

9 years ago[asan] fix odr-violation.cc test to always use slow unwinding (hopefully fixes the...
Kostya Serebryany [Tue, 24 Feb 2015 18:52:38 +0000 (18:52 +0000)]
[asan] fix odr-violation.cc test to always use slow unwinding (hopefully fixes the ARM bot)

llvm-svn: 230356

9 years agoAArch64: Relax assert about large shift sizes.
Matthias Braun [Tue, 24 Feb 2015 18:52:04 +0000 (18:52 +0000)]
AArch64: Relax assert about large shift sizes.

The reason why these large shift sizes happen is because OpaqueConstants
currently inhibit alot of DAG combining, but that has to be addressed in
another commit (like the proposal in D6946).

Differential Revision: http://reviews.llvm.org/D6940

llvm-svn: 230355

9 years agoDAGCombiner: Move variable definitions closer to use; NFC
Matthias Braun [Tue, 24 Feb 2015 18:52:01 +0000 (18:52 +0000)]
DAGCombiner: Move variable definitions closer to use; NFC

llvm-svn: 230354

9 years agoDAGCombiner: Move variable declaration closer to definiion; NFC
Matthias Braun [Tue, 24 Feb 2015 18:51:59 +0000 (18:51 +0000)]
DAGCombiner: Move variable declaration closer to definiion; NFC

llvm-svn: 230353

9 years agoR600/SI: Remove isel mubuf legalization
Tom Stellard [Tue, 24 Feb 2015 17:59:19 +0000 (17:59 +0000)]
R600/SI: Remove isel mubuf legalization

We legalize mubuf instructions post-instruction selection, so this
code is no longer needed.

llvm-svn: 230352

9 years agoAdd comment in MiSignalTestCase.test_lldbmi_stopped_when_stopatentry_local back and...
Ilia K [Tue, 24 Feb 2015 17:54:34 +0000 (17:54 +0000)]
Add comment in MiSignalTestCase.test_lldbmi_stopped_when_stopatentry_local back and revert expecting string.

llvm-svn: 230351

9 years agoARM: Simplify PCS handling.
Tim Northover [Tue, 24 Feb 2015 17:22:40 +0000 (17:22 +0000)]
ARM: Simplify PCS handling.

The backend should now be able to handle all AAPCS rules based on argument
type, which means Clang no longer has to duplicate the register-counting logic
and the CodeGen can be significantly simplified.

llvm-svn: 230349

9 years agoARM: treat [N x i32] and [N x i64] as AAPCS composite types
Tim Northover [Tue, 24 Feb 2015 17:22:34 +0000 (17:22 +0000)]
ARM: treat [N x i32] and [N x i64] as AAPCS composite types

The logic is almost there already, with our special homogeneous aggregate
handling. Tweaking it like this allows front-ends to emit AAPCS compliant code
without ever having to count registers or add discarded padding arguments.

Only arrays of i32 and i64 are needed to model AAPCS rules, but I decided to
apply the logic to all integer arrays for more consistency.

llvm-svn: 230348

9 years agoAdjust an lldb-mi test to fix failure on Linux.
Hafiz Abid Qadeer [Tue, 24 Feb 2015 17:20:43 +0000 (17:20 +0000)]
Adjust an lldb-mi test to fix failure on Linux.

llvm-svn: 230347

9 years agoRemove redundant code from lldb-mi.
Hafiz Abid Qadeer [Tue, 24 Feb 2015 17:08:20 +0000 (17:08 +0000)]
Remove redundant code from lldb-mi.

Summary:
After recent changes, some code has become redundant. This revision tries to remove
the un-used code and tidy up the rest.

Following 4 files have been removed. I have updated CMake files and checked that it builds
fine on Linux and Windows. Can somebody update the xcode related file accordingly?

tools/lldb-mi/MICmnStreamStdinLinux.cpp
tools/lldb-mi/MICmnStreamStdinLinux.h
tools/lldb-mi/MICmnStreamStdinWindows.cpp
tools/lldb-mi/MICmnStreamStdinWindows.h

Reviewers: clayborg, ki.stfu

Reviewed By: clayborg, ki.stfu

Subscribers: lldb-commits

Differential Revision: http://reviews.llvm.org/D7834

llvm-svn: 230345

9 years ago[ASan/Win] Add support for sanitizer allocator hooks, __asan_default_options and...
Timur Iskhodzhanov [Tue, 24 Feb 2015 17:07:22 +0000 (17:07 +0000)]
[ASan/Win] Add support for sanitizer allocator hooks, __asan_default_options and __asan_on_error

llvm-svn: 230344

9 years agoRevert "Raising minimum required CMake version to 2.8.12.2."
Tobias Grosser [Tue, 24 Feb 2015 16:39:46 +0000 (16:39 +0000)]
Revert "Raising minimum required CMake version to 2.8.12.2."

This reverts commit r230062.

Debian stable (wheezy) ships still with cmake 2.8.9.

The commit broke my LLVM/Polly buildbot, to my knowledge our only Linux+cmake
buildbot.

llvm-svn: 230343

9 years agosimplify control flow; NFC
Sanjay Patel [Tue, 24 Feb 2015 16:26:02 +0000 (16:26 +0000)]
simplify control flow; NFC

llvm-svn: 230342

9 years agoRevert r230280: "Bugfix: SCEVExpander incorrectly marks increment operations as no...
Hans Wennborg [Tue, 24 Feb 2015 16:19:29 +0000 (16:19 +0000)]
Revert r230280: "Bugfix: SCEVExpander incorrectly marks increment operations as no-wrap"

This caused PR22674, failing this assert:

Instructions.h:2281: llvm::Value* llvm::PHINode::getOperand(unsigned int) const: Assertion `i_nocapture < OperandTraits<PHINode>::operands(this) && "getOperand() out of range!"' failed.

llvm-svn: 230341

9 years agoAllow non-affine control flow -- Code Generation
Johannes Doerfert [Tue, 24 Feb 2015 16:16:32 +0000 (16:16 +0000)]
Allow non-affine control flow -- Code Generation

  This is the code generation for region statements that are created
  when non-affine control flow was present in the input. A new
  generator, similar to the block or vector generator, for regions is
  used to traverse and copy the region statement and to adjust the
  control flow inside the new region in the end.

llvm-svn: 230340

9 years ago[x32] Mark RBX as reserved when EBX is the base pointer.
Michael Kuperstein [Tue, 24 Feb 2015 16:13:16 +0000 (16:13 +0000)]
[x32] Mark RBX as reserved when EBX is the base pointer.

This should have gone into r230334.

llvm-svn: 230339

9 years agofix typo in comment; NFC
Sanjay Patel [Tue, 24 Feb 2015 16:11:05 +0000 (16:11 +0000)]
fix typo in comment; NFC

llvm-svn: 230338

9 years agoOmit a default argument
Johannes Doerfert [Tue, 24 Feb 2015 16:02:16 +0000 (16:02 +0000)]
Omit a default argument

llvm-svn: 230337

9 years ago[FIX] Silence warning
Johannes Doerfert [Tue, 24 Feb 2015 16:00:41 +0000 (16:00 +0000)]
[FIX] Silence warning

llvm-svn: 230336