platform/upstream/llvm.git
10 years agoX86: stifle GCC warning
Saleem Abdulrasool [Thu, 12 Jun 2014 17:56:18 +0000 (17:56 +0000)]
X86: stifle GCC warning

lib/Target/X86/X86TargetTransformInfo.cpp: In member function ‘virtual unsigned int {anonymous}::X86TTI::getIntImmCost(unsigned int, unsigned int, const llvm::APInt&, llvm::Type*) const’:
lib/Target/X86/X86TargetTransformInfo.cpp:920:60: warning: enumeral and non-enumeral type in conditional expression [enabled by default]

This seems like an unhelpful warning, but there doesnt seem to be a controlling
flag, so add an explicit cast to silence the warning.

llvm-svn: 210806

10 years agoTrying to fix the windows build.
Rafael Espindola [Thu, 12 Jun 2014 17:49:35 +0000 (17:49 +0000)]
Trying to fix the windows build.

llvm-svn: 210805

10 years agoBasic: fix warnings from GCC
Saleem Abdulrasool [Thu, 12 Jun 2014 17:43:37 +0000 (17:43 +0000)]
Basic: fix warnings from GCC

tools/clang/lib/Basic/DiagnosticIDs.cpp: In function ‘clang::DiagnosticIDs::Level toLevel(clang::diag::Severity)’:
tools/clang/lib/Basic/DiagnosticIDs.cpp:382:1: warning: control reaches end of non-void function [-Wreturn-type]

tools/clang/lib/Format/Format.cpp: In member function ‘virtual std::string clang::format::ParseErrorCategory::message(int) const’:
tools/clang/lib/Format/Format.cpp:282:1: warning: control reaches end of non-void function [-Wreturn-type]

Add a default cases that asserts that we handle the severity, parse error.

llvm-svn: 210804

10 years agoRemove system_error.h.
Rafael Espindola [Thu, 12 Jun 2014 17:38:55 +0000 (17:38 +0000)]
Remove system_error.h.

This is a minimal change to remove the header. I will remove the occurrences
of "using std::error_code" in a followup patch.

llvm-svn: 210803

10 years agoInclude system_error directly.
Rafael Espindola [Thu, 12 Jun 2014 17:19:42 +0000 (17:19 +0000)]
Include system_error directly.

llvm-svn: 210802

10 years agoinclude system_error directly.
Rafael Espindola [Thu, 12 Jun 2014 17:15:58 +0000 (17:15 +0000)]
include system_error directly.

llvm-svn: 210801

10 years agoRun llvm/utils/sort_includes.py in a few files.
Rafael Espindola [Thu, 12 Jun 2014 17:12:28 +0000 (17:12 +0000)]
Run llvm/utils/sort_includes.py in a few files.

This will reduce the noise in a followup patch.

llvm-svn: 210800

10 years agoUses #include "..." instead of #include <...> for llvm headers.
Rafael Espindola [Thu, 12 Jun 2014 17:08:11 +0000 (17:08 +0000)]
Uses #include "..." instead of #include <...> for llvm headers.

llvm-svn: 210799

10 years agoMore tests for // rdar://17259812
Fariborz Jahanian [Thu, 12 Jun 2014 16:55:43 +0000 (16:55 +0000)]
More tests for // rdar://17259812

llvm-svn: 210798

10 years agoinclude system_error directly.
Rafael Espindola [Thu, 12 Jun 2014 16:53:02 +0000 (16:53 +0000)]
include system_error directly.

llvm-svn: 210797

10 years ago[Mips] Make exe-dynamic.test test case independent from external input files.
Simon Atanasyan [Thu, 12 Jun 2014 16:46:47 +0000 (16:46 +0000)]
[Mips] Make exe-dynamic.test test case independent from external input files.

llvm-svn: 210796

10 years agoObjective-C. Use isObjCRetainableType in my last
Fariborz Jahanian [Thu, 12 Jun 2014 16:12:30 +0000 (16:12 +0000)]
Objective-C. Use isObjCRetainableType in my last
patch. NFC.

llvm-svn: 210795

10 years agoadding re-include guards into lib/Support/reg*.h
Artyom Skrobov [Thu, 12 Jun 2014 16:07:56 +0000 (16:07 +0000)]
adding re-include guards into lib/Support/reg*.h

llvm-svn: 210794

10 years agoRemove pimpl class from PassRegistry.
Zachary Turner [Thu, 12 Jun 2014 16:06:51 +0000 (16:06 +0000)]
Remove pimpl class from PassRegistry.

Since removeRegistrationListener is no longer called during static
destruction, we can get rid of the pimpl in PassRegistry.

This should clean up the code somewhat, increase clarity, and also
allows us to put the Lock as a member of the class, instead of as a
ManagedStatic.

As part of this change, the PassInfo class is moved from
PassSupport.h to its own file, to eliminate the otherwise circular
header dependency between PassRegistry.h and PassSupport.h

Reviewed by: rnk, dblaikie

Differential Revision: http://reviews.llvm.org/D4107

llvm-svn: 210793

10 years agoRevert "SelectionDAG: Enable (and (setcc x), (setcc y)) -> (setcc (and x, y)) for...
Tom Stellard [Thu, 12 Jun 2014 16:04:47 +0000 (16:04 +0000)]
Revert "SelectionDAG: Enable (and (setcc x), (setcc y)) -> (setcc (and x, y)) for vectors"

This reverts commit r210540, adds a testcase for the regression it
caused, and marks the R600 test it was supposed to fix as XFAIL.

llvm-svn: 210792

10 years agoExplicitly initialize all fields to avoid -Wmissing-field-initializers warning.
Eli Bendersky [Thu, 12 Jun 2014 15:47:57 +0000 (15:47 +0000)]
Explicitly initialize all fields to avoid -Wmissing-field-initializers warning.

llvm-svn: 210791

10 years agoDisable the load/store optimization pass for Thumb-1.
James Molloy [Thu, 12 Jun 2014 15:18:33 +0000 (15:18 +0000)]
Disable the load/store optimization pass for Thumb-1.

Moritz's changes have improved codegen a lot, but further testing showed significant correctness problems. Disable by default until these have been worked out.

Patch by Moritz Roth!

llvm-svn: 210789

10 years ago[mips][mips64r6] bc1[tf] are not available on MIPS32r6/MIPS64r6
Daniel Sanders [Thu, 12 Jun 2014 15:00:17 +0000 (15:00 +0000)]
[mips][mips64r6] bc1[tf] are not available on MIPS32r6/MIPS64r6

Summary:
Also tightened up the acceptable condition operand for these instructions
on MIPS-I to MIPS-III. Support for $fcc[1-7] was added in MIPS-IV. Prior
to that only $fcc0 is acceptable.

We currently don't optimize (BEQZ (NOT $a), $target) and similar. It's
probably best to do this in InstCombine.

Depends on D4111

Reviewers: jkolek, zoran.jovanovic, vmedic

Reviewed By: vmedic

Differential Revision: http://reviews.llvm.org/D4112

llvm-svn: 210787

10 years ago[mips][mips64r6] bc2[ft] are not available on MIPS32r6/MIPS64r6
Daniel Sanders [Thu, 12 Jun 2014 14:54:13 +0000 (14:54 +0000)]
[mips][mips64r6] bc2[ft] are not available on MIPS32r6/MIPS64r6

Summary:
These instructions are not implemented for any MIPS ISA so we only need
testcases.

Depends on D4110

Reviewers: jkolek, zoran.jovanovic, vmedic

Reviewed By: vmedic

Differential Revision: http://reviews.llvm.org/D4111

llvm-svn: 210786

10 years agoDon't import error_code into the lld namespace.
Rafael Espindola [Thu, 12 Jun 2014 14:53:47 +0000 (14:53 +0000)]
Don't import error_code into the lld namespace.

llvm-svn: 210785

10 years ago[mips][mips64r6] [sl][duw]xc1 are not available on MIPS32r6/MIPS64r6
Daniel Sanders [Thu, 12 Jun 2014 14:19:28 +0000 (14:19 +0000)]
[mips][mips64r6] [sl][duw]xc1 are not available on MIPS32r6/MIPS64r6

Summary:
Folded mips64-fp-indexed-ls.ll into fp-indexed-ls.ll. To do so, the zext's in
mips64-fp-indexed-ls.ll were changed to implicit sign extensions (performed
by getelementptr). This does not affect the purpose of the test.

Depends on D4004

Reviewers: zoran.jovanovic, jkolek, vmedic

Reviewed By: vmedic

Differential Revision: http://reviews.llvm.org/D4110

llvm-svn: 210784

10 years agoReplace llvm::error_code with std::error_code.
Rafael Espindola [Thu, 12 Jun 2014 14:11:22 +0000 (14:11 +0000)]
Replace llvm::error_code with std::error_code.

llvm-svn: 210783

10 years agoThis removes TODO added in http://reviews.llvm.org/D3658
Dinesh Dwivedi [Thu, 12 Jun 2014 14:06:00 +0000 (14:06 +0000)]
This removes TODO added in reviews.llvm.org/D3658

The patch transforms

ABS(NABS(X)) -> ABS(X)
NABS(ABS(X)) -> NABS(X)

Differential Revision: http://reviews.llvm.org/D4040

llvm-svn: 210782

10 years agoreplace llvm::error_code with std::error_code.
Rafael Espindola [Thu, 12 Jun 2014 14:04:54 +0000 (14:04 +0000)]
replace llvm::error_code with std::error_code.

llvm-svn: 210781

10 years agoReplace llvm::error_code with std::error_code.
Rafael Espindola [Thu, 12 Jun 2014 14:02:15 +0000 (14:02 +0000)]
Replace llvm::error_code with std::error_code.

llvm-svn: 210780

10 years ago[mips][mips64r6] prefx is not available on MIPS32r6/MIPS64r6
Daniel Sanders [Thu, 12 Jun 2014 13:51:27 +0000 (13:51 +0000)]
[mips][mips64r6] prefx is not available on MIPS32r6/MIPS64r6

Summary: We haven't implemented this instruction so we only add a test case.

Reviewers: vmedic, zoran.jovanovic, jkolek

Reviewed By: jkolek

Differential Revision: http://reviews.llvm.org/D4004

llvm-svn: 210779

10 years ago[mips][mips64r6] 80 col corrections that should have been in r210777.
Daniel Sanders [Thu, 12 Jun 2014 13:42:04 +0000 (13:42 +0000)]
[mips][mips64r6] 80 col corrections that should have been in r210777.

llvm-svn: 210778

10 years ago[mips][mips64r6] c.cond.fmt, mov[fntz], and mov[fntz].[ds] are not available on MIPS3...
Daniel Sanders [Thu, 12 Jun 2014 13:39:06 +0000 (13:39 +0000)]
[mips][mips64r6] c.cond.fmt, mov[fntz], and mov[fntz].[ds] are not available on MIPS32r6/MIPS64r6

Summary:
c.cond.fmt has been replaced by cmp.cond.fmt. Where c.cond.fmt wrote to
dedicated condition registers, cmp.cond.fmt writes 1 or 0 to normal FGR's
(like the GPR comparisons).

mov[fntz] have been replaced by seleqz and selnez. These instructions
conditionally zero a register based on a bool in a GPR. The results can
then be or'd together to act as a select without, for example, requiring a third
register read port.

mov[fntz].[ds] have been replaced with sel.[ds]

MIPS64r6 currently generates unnecessary sign-extensions for most selects.
This is because the result of a SETCC is currently an i32. Bits 32-63 are
undefined in i32 and the behaviour of seleqz/selnez would otherwise depend
on undefined bits. Later, we will fix this by making the result of SETCC an
i64 on MIPS64 targets.

Depends on D3958

Reviewers: jkolek, vmedic, zoran.jovanovic

Reviewed By: vmedic, zoran.jovanovic

Differential Revision: http://reviews.llvm.org/D4003

llvm-svn: 210777

10 years agoReplace llvm::error_code with std::error_code.
Rafael Espindola [Thu, 12 Jun 2014 13:32:11 +0000 (13:32 +0000)]
Replace llvm::error_code with std::error_code.

llvm-svn: 210776

10 years ago[mips][mips64r6] jalx is not available on MIPS32r6/MIPS64r6
Daniel Sanders [Thu, 12 Jun 2014 12:58:20 +0000 (12:58 +0000)]
[mips][mips64r6] jalx is not available on MIPS32r6/MIPS64r6

Summary: Depends on D3957

Reviewers: jkolek, zoran.jovanovic, vmedic

Reviewed By: vmedic

Differential Revision: http://reviews.llvm.org/D3958

llvm-svn: 210775

10 years agoAllow lowercase messages in this test again.
Rafael Espindola [Thu, 12 Jun 2014 12:40:04 +0000 (12:40 +0000)]
Allow lowercase messages in this test again.

I misunderstood where the message was being converted.

llvm-svn: 210774

10 years ago[mips][mips64r6] Add R_MIPS_PC19_S2
Zoran Jovanovic [Thu, 12 Jun 2014 12:40:00 +0000 (12:40 +0000)]
[mips][mips64r6] Add R_MIPS_PC19_S2
Differential Revision: http://reviews.llvm.org/D3866

llvm-svn: 210773

10 years agoDon't import make_error_code into the llvm namespace.
Rafael Espindola [Thu, 12 Jun 2014 11:58:49 +0000 (11:58 +0000)]
Don't import make_error_code into the llvm namespace.

llvm-svn: 210772

10 years ago[mips] Use MTHC1 when it is available (MIPS32r2 and later) for both FP32 and FP64
Daniel Sanders [Thu, 12 Jun 2014 11:55:58 +0000 (11:55 +0000)]
[mips] Use MTHC1 when it is available (MIPS32r2 and later) for both FP32 and FP64

Summary:
To make this work for both AFGR64 and FGR64 register sets, I've had to make the
instruction definition consistent with the white lie (that it reads the lower
32-bits of the register) when they are generated by expandBuildPairF64().

Corrected the definition of hasMips32r2() and hasMips64r2() to include
MIPS32r6 and MIPS64r6.

Depends on D3956

Reviewers: jkolek, zoran.jovanovic, vmedic

Reviewed By: vmedic

Differential Revision: http://reviews.llvm.org/D3957

llvm-svn: 210771

10 years ago[mips][mips64r6] Add bgec and bgeuc instructions
Zoran Jovanovic [Thu, 12 Jun 2014 11:47:44 +0000 (11:47 +0000)]
[mips][mips64r6] Add bgec and bgeuc instructions
Differential Revision: http://reviews.llvm.org/D4017

llvm-svn: 210770

10 years ago[X86] Teach how to dump the name of target node RDTSCP_DAG.
Andrea Di Biagio [Thu, 12 Jun 2014 11:37:24 +0000 (11:37 +0000)]
[X86] Teach how to dump the name of target node RDTSCP_DAG.

When I originally added node RDTSCP_DAG (r207127) I forgot to add
a string name for it in method 'getTargetNodeName'.

No functional change intended.

llvm-svn: 210769

10 years agoFix msvc unittest build.
Rafael Espindola [Thu, 12 Jun 2014 11:35:17 +0000 (11:35 +0000)]
Fix msvc unittest build.

Looks like msvc has an asymmetrical operator ==.

llvm-svn: 210768

10 years agoA follow-up to r210260: updated a comment. No functional changes.
Alexander Kornienko [Thu, 12 Jun 2014 11:25:45 +0000 (11:25 +0000)]
A follow-up to r210260: updated a comment. No functional changes.

llvm-svn: 210767

10 years agoCMake: don't install the internal config.h header
Alp Toker [Thu, 12 Jun 2014 11:25:18 +0000 (11:25 +0000)]
CMake: don't install the internal config.h header

Background:
  http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-June/073707.html

llvm-svn: 210766

10 years agoTrack clang changes from r210764
Alp Toker [Thu, 12 Jun 2014 11:14:32 +0000 (11:14 +0000)]
Track clang changes from r210764

llvm-svn: 210765

10 years agoDiagnosticsEngine: update severity setters to new terminology
Alp Toker [Thu, 12 Jun 2014 11:13:52 +0000 (11:13 +0000)]
DiagnosticsEngine: update severity setters to new terminology

llvm-svn: 210764

10 years ago[mips][mips64r6] madd.[ds], msub.[ds], nmadd.[ds], and nmsub.[ds] are not available...
Daniel Sanders [Thu, 12 Jun 2014 11:04:18 +0000 (11:04 +0000)]
[mips][mips64r6] madd.[ds], msub.[ds], nmadd.[ds], and nmsub.[ds] are not available on MIPS32r6/MIPS64r6

Summary:
This patch updates both the assembler and the code generator.

MIPS32r6/MIPS64r6 replaces them with maddf.[ds] and msubf.[ds] which are fused
multiply-add/sub operations. We don't emit these yet, this patch only prevents the removed instructions from being emitted.

Depends on D3955

Reviewers: jkolek, zoran.jovanovic, vmedic

Reviewed By: vmedic

Differential Revision: http://reviews.llvm.org/D3956

llvm-svn: 210763

10 years ago[mips][mips64r6] madd/maddu/msub/msubu are not available on MIPS32r6/MIPS64r6
Daniel Sanders [Thu, 12 Jun 2014 10:54:16 +0000 (10:54 +0000)]
[mips][mips64r6] madd/maddu/msub/msubu are not available on MIPS32r6/MIPS64r6

Summary:
This patch disables madd/maddu/msub/msubu in both the assembler and code
generator.

Depends on D3896

Reviewers: jkolek, zoran.jovanovic, vmedic

Reviewed By: vmedic

Differential Revision: http://reviews.llvm.org/D3955

llvm-svn: 210762

10 years ago[X86] Teach how to combine AVX and AVX2 horizontal binop on packed 256-bit vectors.
Andrea Di Biagio [Thu, 12 Jun 2014 10:53:48 +0000 (10:53 +0000)]
[X86] Teach how to combine AVX and AVX2 horizontal binop on packed 256-bit vectors.

This patch adds target combine rules to match:
 - [AVX] Horizontal add/sub of packed single/double precision floating point
   values from 256-bit vectors;
 - [AVX2] Horizontal add/sub of packed integer values from 256-bit vectors.

llvm-svn: 210761

10 years ago[mips][mips64r6] Replace m[tf]hi, m[tf]lo, mult, multu, dmult, dmultu, div, ddiv...
Daniel Sanders [Thu, 12 Jun 2014 10:44:10 +0000 (10:44 +0000)]
[mips][mips64r6] Replace m[tf]hi, m[tf]lo, mult, multu, dmult, dmultu, div, ddiv, divu, ddivu for MIPS32r6/MIPS64.

Summary:
The accumulator-based (HI/LO) multiplies and divides from earlier ISA's have
been removed and replaced with GPR-based equivalents. For example:
  div $1, $2
  mflo $3
is now:
  div $3, $1, $2

This patch disables the accumulator-based multiplies and divides for
MIPS32r6/MIPS64r6 and uses the GPR-based equivalents instead.

Renamed expandPseudoDiv to insertDivByZeroTrap to better describe the
behaviour of the function.

MipsDelaySlotFiller now invalidates the liveness information when moving
instructions to the delay slot. Without this, divrem.ll will abort since
%GP ends up used before it is defined.

Reviewers: vmedic, zoran.jovanovic, jkolek

Reviewed By: jkolek

Differential Revision: http://reviews.llvm.org/D3896

llvm-svn: 210760

10 years agoTrack clang changes from r210758
Alp Toker [Thu, 12 Jun 2014 10:16:11 +0000 (10:16 +0000)]
Track clang changes from r210758

llvm-svn: 210759

10 years agoComplete the switch from mappings to declarative diagnostic severities
Alp Toker [Thu, 12 Jun 2014 10:15:20 +0000 (10:15 +0000)]
Complete the switch from mappings to declarative diagnostic severities

This begins to address cognitive dissonance caused by treating the Note
diagnostic level as a severity in the diagnostic engine.

No change in functionality.

llvm-svn: 210758

10 years ago[mips] Move CHECK lines to the same line as the instruction it's testing
Matheus Almeida [Thu, 12 Jun 2014 09:50:17 +0000 (09:50 +0000)]
[mips] Move CHECK lines to the same line as the instruction it's testing
for consistency with the other tests.

No functional changes.

llvm-svn: 210757

10 years agoR600/SI: Use a register set to -1 for data0 on ds_inc*/ds_dec*
Matt Arsenault [Thu, 12 Jun 2014 08:21:54 +0000 (08:21 +0000)]
R600/SI: Use a register set to -1 for data0 on ds_inc*/ds_dec*

There is not such thing as a 0-data ds instruction, and the data
operand needs to be a vgpr set to something meaningful.

llvm-svn: 210756

10 years agoDo not use namespace polly inside a header.
Andreas Simbuerger [Thu, 12 Jun 2014 07:26:25 +0000 (07:26 +0000)]
Do not use namespace polly inside a header.

In general this fixes ambiguity that can arise from using
a different namespace that declares the same symbols as
we do.

One example inside llvm would be:
  createIndVarSimplifyPass(..);

Which can be found in:
  llvm/Transforms/Scalar.h
and
  polly/LinkAllPasses.h

llvm-svn: 210755

10 years agoFix typo
Andreas Simbuerger [Thu, 12 Jun 2014 07:26:02 +0000 (07:26 +0000)]
Fix typo

llvm-svn: 210754

10 years agoCheck for an empty error log.
Andreas Simbuerger [Thu, 12 Jun 2014 07:25:08 +0000 (07:25 +0000)]
Check for an empty error log.

Fixes #19976.

The error log does not contain an error, in case we reject a candidate
without generating a diagnostic message by using invalid<>(...). This is
the case for the top-level region of a function.

The patch comes without a test-case because adding a useful one requires
additional code just for triggering it. Before the patch it would only trigger,
if we try to print the CFG with Scop error annotations.

llvm-svn: 210753

10 years agoMove getDebugLocation to ScopDetectionDiagnostic
Andreas Simbuerger [Thu, 12 Jun 2014 07:23:04 +0000 (07:23 +0000)]
Move getDebugLocation to ScopDetectionDiagnostic

llvm-svn: 210752

10 years agoUse StringRef to simplify code. No functional change.
Craig Topper [Thu, 12 Jun 2014 05:32:35 +0000 (05:32 +0000)]
Use StringRef to simplify code. No functional change.

llvm-svn: 210751

10 years agoUse ArrayRef in some function parameters instead of a pointer and count. No functiona...
Craig Topper [Thu, 12 Jun 2014 05:32:27 +0000 (05:32 +0000)]
Use ArrayRef in some function parameters instead of a pointer and count. No functional change.

llvm-svn: 210750

10 years agoAvoid in-class initializer from r210747
Alp Toker [Thu, 12 Jun 2014 04:27:37 +0000 (04:27 +0000)]
Avoid in-class initializer from r210747

Turns out MSVC doesn't like this. Sorry for the noise!

llvm-svn: 210749

10 years agoAvoid anonymous namespace in header from r210747
Alp Toker [Thu, 12 Jun 2014 04:21:14 +0000 (04:21 +0000)]
Avoid anonymous namespace in header from r210747

llvm-svn: 210748

10 years agoAvoid redundant allocations in the linker optimisation hint
Alp Toker [Thu, 12 Jun 2014 04:02:46 +0000 (04:02 +0000)]
Avoid redundant allocations in the linker optimisation hint

llvm-svn: 210747

10 years ago[FastISel][x86] Add testcase for r210719.
Juergen Ributzka [Thu, 12 Jun 2014 03:54:05 +0000 (03:54 +0000)]
[FastISel][x86] Add testcase for r210719.

llvm-svn: 210746

10 years agoAdd a std:: prefix in cases where ADL would have failed on windows.
Rafael Espindola [Thu, 12 Jun 2014 03:53:36 +0000 (03:53 +0000)]
Add a std:: prefix in cases where ADL would have failed on windows.

This is in preparation for removing make_error_code from the llvm namespace.

llvm-svn: 210745

10 years agoFix typo.
Rafael Espindola [Thu, 12 Jun 2014 03:31:26 +0000 (03:31 +0000)]
Fix typo.

Thanks to Alp Toker for noticing.

llvm-svn: 210744

10 years ago[x86] Improve frameaddress test from r210709.
Juergen Ributzka [Thu, 12 Jun 2014 03:29:29 +0000 (03:29 +0000)]
[x86] Improve frameaddress test from r210709.

llvm-svn: 210743

10 years ago[FastISel] Add support for the stackmap intrinsic.
Juergen Ributzka [Thu, 12 Jun 2014 03:29:26 +0000 (03:29 +0000)]
[FastISel] Add support for the stackmap intrinsic.

This implements target-independent FastISel lowering for the stackmap intrinsic.

llvm-svn: 210742

10 years agoDon't use make_error_code from the llvm namespace.
Rafael Espindola [Thu, 12 Jun 2014 03:13:49 +0000 (03:13 +0000)]
Don't use make_error_code from the llvm namespace.

llvm-svn: 210741

10 years agoPrefix generic_category with std::.
Rafael Espindola [Thu, 12 Jun 2014 02:52:22 +0000 (02:52 +0000)]
Prefix generic_category with std::.

Sorry I missed these before.

llvm-svn: 210740

10 years agoGive clang-format its own error category.
Rafael Espindola [Thu, 12 Jun 2014 02:50:04 +0000 (02:50 +0000)]
Give clang-format its own error category.

The posix errno values are probably to the best thing to use for
describing parse errors.

This should also fix the mingw build.

llvm-svn: 210739

10 years agoDriver: use more range-based for loops
Saleem Abdulrasool [Thu, 12 Jun 2014 02:08:04 +0000 (02:08 +0000)]
Driver: use more range-based for loops

Mechanical change converting some of the simpler for loops into range-based for
loops.  NFC.

llvm-svn: 210738

10 years agoDon't put generic_category in the llvm namespace.
Rafael Espindola [Thu, 12 Jun 2014 02:00:39 +0000 (02:00 +0000)]
Don't put generic_category in the llvm namespace.

llvm-svn: 210737

10 years agoUse generic_category from the std namespace.
Rafael Espindola [Thu, 12 Jun 2014 01:59:06 +0000 (01:59 +0000)]
Use generic_category from the std namespace.

llvm-svn: 210736

10 years agoUse generic_category from the std namespace.
Rafael Espindola [Thu, 12 Jun 2014 01:57:33 +0000 (01:57 +0000)]
Use generic_category from the std namespace.

llvm-svn: 210735

10 years agoFix verifier for GlobalAliases to avoid recursing into global initializers.
Bob Wilson [Thu, 12 Jun 2014 01:46:54 +0000 (01:46 +0000)]
Fix verifier for GlobalAliases to avoid recursing into global initializers.

The verifier follows GlobalAlias operands so that it can detect cycles of
alias definitions. It was doing this in a way that caused it to also recurse
through initializers for the GlobalValue aliasees, and it would fail when
an initializer refers to a global that is a declaration and not a definition.
This patch causes it to stop recursing when it hits a global definition.
<rdar://problem/17277451>

llvm-svn: 210734

10 years agoDon't import error_category into the llvm namespace.
Rafael Espindola [Thu, 12 Jun 2014 01:45:43 +0000 (01:45 +0000)]
Don't import error_category into the llvm namespace.

llvm-svn: 210733

10 years agoUse error_category from the std namespace.
Rafael Espindola [Thu, 12 Jun 2014 01:44:19 +0000 (01:44 +0000)]
Use error_category from the std namespace.

llvm-svn: 210732

10 years agoDon't import error_condition into the llvm namespace.
Rafael Espindola [Thu, 12 Jun 2014 01:29:42 +0000 (01:29 +0000)]
Don't import error_condition into the llvm namespace.

llvm-svn: 210731

10 years agoUse error_condition from the std namespace.
Rafael Espindola [Thu, 12 Jun 2014 01:28:12 +0000 (01:28 +0000)]
Use error_condition from the std namespace.

llvm-svn: 210730

10 years agoUsed mapWindowsError. I missed these in the initial transition.
Rafael Espindola [Thu, 12 Jun 2014 01:25:33 +0000 (01:25 +0000)]
Used mapWindowsError. I missed these in the initial transition.

llvm-svn: 210729

10 years agoRemove some low hanging fruit from system_error.h
Rafael Espindola [Thu, 12 Jun 2014 01:10:47 +0000 (01:10 +0000)]
Remove some low hanging fruit from system_error.h

llvm-svn: 210728

10 years agoErrno should use generic_category.
Rafael Espindola [Thu, 12 Jun 2014 00:57:40 +0000 (00:57 +0000)]
Errno should use generic_category.

Sorry, no testcase, just noticed while trying to remove llvm's system_error.h

llvm-svn: 210727

10 years ago[modules] Add first-draft module maps for Clang. These don't cover everything
Richard Smith [Thu, 12 Jun 2014 00:32:32 +0000 (00:32 +0000)]
[modules] Add first-draft module maps for Clang. These don't cover everything
yet, but they're enough for a clean bootstrap (with a few local patches that
are yet to be committed).

llvm-svn: 210726

10 years agoTry to fix the mingw build.
Rafael Espindola [Thu, 12 Jun 2014 00:24:39 +0000 (00:24 +0000)]
Try to fix the mingw build.

* MingW needs mapWindowsError.
* MingW is missing some entries in std::errc, but we don't use them.

llvm-svn: 210725

10 years agoDo not register and de-register PassRegistrationListeners during
Zachary Turner [Thu, 12 Jun 2014 00:16:36 +0000 (00:16 +0000)]
Do not register and de-register PassRegistrationListeners during
construction and destruction.

PassRegistrationListener is intended for use as a generic listener.
In some cases, PassRegistrationListener-derived classes were being
created, and automatically registered and de-registered in static
constructors and destructors.  Since ManagedStatics are destroyed
prior to program shutdown, this leads to errors where an attempt is
made to access a ManagedStatic that has already been destroyed.

Reviewed by: rnk, dblaikie

Differential Revision: http://reviews.llvm.org/D4106

llvm-svn: 210724

10 years agoPR19966: don't crash/assert when __underlying_type is applied to an incomplete
Richard Smith [Thu, 12 Jun 2014 00:01:45 +0000 (00:01 +0000)]
PR19966: don't crash/assert when __underlying_type is applied to an incomplete
enumeration type. I've also filed a LWG issue pointing out that this should be
ill-formed.

llvm-svn: 210723

10 years agoAccomodate for message differences on windows.
Rafael Espindola [Wed, 11 Jun 2014 23:53:06 +0000 (23:53 +0000)]
Accomodate for message differences on windows.

llvm-svn: 210722

10 years agoTeach LoopUnrollPass to respect loop unrolling hints in metadata.
Eli Bendersky [Wed, 11 Jun 2014 23:15:35 +0000 (23:15 +0000)]
Teach LoopUnrollPass to respect loop unrolling hints in metadata.

See http://reviews.llvm.org/D4090 for more details.

The Clang change that produces this metadata was committed in r210667

Patch by Mark Heffernan.

llvm-svn: 210721

10 years ago[FastISel][X86] Add support for the sqrt intrinsic.
Juergen Ributzka [Wed, 11 Jun 2014 23:11:02 +0000 (23:11 +0000)]
[FastISel][X86] Add support for the sqrt intrinsic.

llvm-svn: 210720

10 years ago[FastIsel][X86] Add support for lowering the first 8 floating-point arguments.
Juergen Ributzka [Wed, 11 Jun 2014 23:10:58 +0000 (23:10 +0000)]
[FastIsel][X86] Add support for lowering the first 8 floating-point arguments.

llvm-svn: 210719

10 years agoMulti-line expressions in Xcode now have a space between the line number and the...
Greg Clayton [Wed, 11 Jun 2014 23:10:41 +0000 (23:10 +0000)]
Multi-line expressions in Xcode now have a space between the line number and the expression text.

<rdar://problem/17238093>

llvm-svn: 210718

10 years agoDon't acquire the mutex during the destructor of PassRegistry.
Zachary Turner [Wed, 11 Jun 2014 23:03:31 +0000 (23:03 +0000)]
Don't acquire the mutex during the destructor of PassRegistry.

This destructor is run as part of static program termination, and
so all ManagedStatics (including this lock) will have been
destroyed by llvm_shutdown.  Furthermore, if there is actually
a race condition during static program termination, then we are
just hiding a bug somewhere else, because other threads should
not be running at this point.

llvm-svn: 210717

10 years agoImplement get_magic with generic tools and inline it.
Rafael Espindola [Wed, 11 Jun 2014 22:53:00 +0000 (22:53 +0000)]
Implement get_magic with generic tools and inline it.

llvm-svn: 210716

10 years agoDon't inherit dllimport to inline move assignment operators
Hans Wennborg [Wed, 11 Jun 2014 22:44:39 +0000 (22:44 +0000)]
Don't inherit dllimport to inline move assignment operators

Current MSVC versions don't have move assignment operators, so we
can't rely on them being available in the dll. If we have the
definition, we can just use that directly. This breaks pointer
equality, but should work fine otherwise.

When there is an MSVC version that supports move assignment,
we can key this off the -fmsc-ver option.

http://reviews.llvm.org/D4105

llvm-svn: 210715

10 years agoFix test build on windows.
Rafael Espindola [Wed, 11 Jun 2014 22:23:45 +0000 (22:23 +0000)]
Fix test build on windows.

llvm-svn: 210714

10 years agoDon't slice SemaDiagnosticBuilder
Reid Kleckner [Wed, 11 Jun 2014 21:57:15 +0000 (21:57 +0000)]
Don't slice SemaDiagnosticBuilder

I wasn't able to figure out how to emit this diagnostic from a SFINAE
context, so I don't have a test.

llvm-svn: 210713

10 years agoRemove unused has_magic.
Rafael Espindola [Wed, 11 Jun 2014 21:53:22 +0000 (21:53 +0000)]
Remove unused has_magic.

This will allow inlining get_magic, which should in turn fix one of the mingw
build problems after the switch to std::error_code.

llvm-svn: 210712

10 years agoMove Simple.h and Alias.h to include/Core.
Rui Ueyama [Wed, 11 Jun 2014 21:47:51 +0000 (21:47 +0000)]
Move Simple.h and Alias.h to include/Core.

Because the files in Core actually depend on these files.

Differential Revision: http://reviews.llvm.org/D4000

llvm-svn: 210710

10 years ago[FastISel][X86] Add support for the frameaddress intrinsic.
Juergen Ributzka [Wed, 11 Jun 2014 21:44:44 +0000 (21:44 +0000)]
[FastISel][X86] Add support for the frameaddress intrinsic.

llvm-svn: 210709

10 years agoFix the VS2012 build.
Hans Wennborg [Wed, 11 Jun 2014 21:24:13 +0000 (21:24 +0000)]
Fix the VS2012 build.

It didn't handle the in-class member initializer and compound literal.

llvm-svn: 210707

10 years agoObjective-C. Accept '__attribute__((__ns_returns_retained__))'
Fariborz Jahanian [Wed, 11 Jun 2014 21:22:53 +0000 (21:22 +0000)]
Objective-C. Accept '__attribute__((__ns_returns_retained__))'
for function/methods returning block in MRR mode as well.
// rdar://17259812

llvm-svn: 210706

10 years ago[AArch64] Basic Sched Model for Cortex-A57.
Chad Rosier [Wed, 11 Jun 2014 21:06:56 +0000 (21:06 +0000)]
[AArch64] Basic Sched Model for Cortex-A57.
Patch by Dave Estes<cestes@codeaurora.org>
Differential Revision: http://reviews.llvm.org/D4008

llvm-svn: 210705

10 years agoQuick fix for the windows build.
Rafael Espindola [Wed, 11 Jun 2014 21:04:06 +0000 (21:04 +0000)]
Quick fix for the windows build.

We have to look for both std::make_error_code as well as our overloads.

llvm-svn: 210704

10 years agoR600: Set correct InstrItinClass for instructions using *Helper classes
Tom Stellard [Wed, 11 Jun 2014 20:51:42 +0000 (20:51 +0000)]
R600: Set correct InstrItinClass for instructions using *Helper classes

We weren't doing this before, so all instruction using the *Helper
classes were considered for any ALU slot.

This fixes a hang in the builtin-char-clz-1.0.generated.cl piglit test.

llvm-svn: 210703