platform/upstream/llvm.git
9 years ago[Sanitizer] Be consistent about separating ==%PID== and logged data.
Alexey Samsonov [Tue, 31 Mar 2015 18:16:42 +0000 (18:16 +0000)]
[Sanitizer] Be consistent about separating ==%PID== and logged data.

See https://code.google.com/p/address-sanitizer/issues/detail?id=385.

llvm-svn: 233720

9 years agoELF: Replace a macro with an inlined function.
Rui Ueyama [Tue, 31 Mar 2015 18:15:05 +0000 (18:15 +0000)]
ELF: Replace a macro with an inlined function.

FINDV4BITMASK macro is defined as a macro so that the macro body is inlined.
We should use inlined functions instead of macros.

llvm-svn: 233719

9 years agoELF: Do not use multiple inheritance.
Rui Ueyama [Tue, 31 Mar 2015 18:10:24 +0000 (18:10 +0000)]
ELF: Do not use multiple inheritance.

Multiple inheritance is casually used here. Rewriting to not
using multiple inheritance reduces the complexity of the code
and also makes it shorter.

llvm-svn: 233718

9 years agoTestPrintStackTraces - made XFAIL more precise
Vince Harron [Tue, 31 Mar 2015 17:45:54 +0000 (17:45 +0000)]
TestPrintStackTraces - made XFAIL more precise

Works with x86_64 inferior, fails w/i386 inferior - updated test to
reflect

llvm-svn: 233717

9 years agoEnabled a bunch of tests on Linux
Vince Harron [Tue, 31 Mar 2015 17:44:37 +0000 (17:44 +0000)]
Enabled a bunch of tests on Linux

Removed expectedFailureLinux from failures that I was unable to
reproduce, updated and improved some other comments near XFAIL tests

Differential Revision: http://reviews.llvm.org/D8676

llvm-svn: 233716

9 years agoAdd sm_37 target to Clang for NVPTX
Eli Bendersky [Tue, 31 Mar 2015 17:03:16 +0000 (17:03 +0000)]
Add sm_37 target to Clang for NVPTX

Support for this target was added in LLVM r233575 and r233583

llvm-svn: 233715

9 years agoFix DynamicLoaderMacOSXDYLD to deal with new shared cache changes.
Greg Clayton [Tue, 31 Mar 2015 17:02:36 +0000 (17:02 +0000)]
Fix DynamicLoaderMacOSXDYLD to deal with new shared cache changes.

<rdar://problem/20357466>

llvm-svn: 233714

9 years agoImplement fmax using __builtin_fmax
Tom Stellard [Tue, 31 Mar 2015 16:59:23 +0000 (16:59 +0000)]
Implement fmax using __builtin_fmax

This ensures correct handling of NaNi.

This has been tested with piglit, OpenCV, and the ocl conformance tests.

llvm-svn: 233713

9 years agoImplement fmin using __builtin_fmin
Tom Stellard [Tue, 31 Mar 2015 16:59:21 +0000 (16:59 +0000)]
Implement fmin using __builtin_fmin

This ensures correct handling of NaN.

This has been tested with piglit, OpenCV, and the ocl conformance tests.

llvm-svn: 233712

9 years ago[libcxx] Optimize vectors uninitialized construction of trivial types from an iterato...
Eric Fiselier [Tue, 31 Mar 2015 16:54:19 +0000 (16:54 +0000)]
[libcxx] Optimize vectors uninitialized construction of trivial types from an iterator range.

Summary:
In certain cases vector can use memcpy to construct a range of elements at the back of the vector. We currently don't do this resulting in terrible code gen in non-optimized mode and a
very large slowdown compared to libstdc++.

This patch adds a `__construct_forward_range(Allocator, Iter, Iter, _Ptr&)` and `__construct_forward_range(Allocator, Tp*, Tp*, Tp*&)` functions to `allocator_traits` which act similarly to the existing `__construct_forward(...)` functions.

This patch also changes vectors `__construct_at_end(Iter, Iter)` to be `__construct_at_end(Iter, Iter, SizeType)` where SizeType is the size of the range. `__construct_at_end(Iter, Iter, SizeType)` now calls `allocator_traits<Tp>::__construct_forward_range(...)`.

This patch is based off the design of `__swap_out_circular_buffer(...)` which uses `allocator_traits<Tp>::__construct_forward(...)`.

On my machine this code performs 4x better than the current implementation when tested against `std::vector<int>`.

Reviewers: howard.hinnant, titus, kcc, mclow.lists

Reviewed By: mclow.lists

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D8109

llvm-svn: 233711

9 years agoUse DEBUG_TYPE and DEBUG instead of DEBUG_WITH_TYPE.
Rui Ueyama [Tue, 31 Mar 2015 16:48:57 +0000 (16:48 +0000)]
Use DEBUG_TYPE and DEBUG instead of DEBUG_WITH_TYPE.

llvm-svn: 233710

9 years agoAArch64: fix v8.1 sqrdmlah tests on Darwin platforms
Tim Northover [Tue, 31 Mar 2015 16:41:38 +0000 (16:41 +0000)]
AArch64: fix v8.1 sqrdmlah tests on Darwin platforms

llvm-svn: 233709

9 years agoRework r233647, "llvm/examples: Suppress building a few JIT examples."
NAKAMURA Takumi [Tue, 31 Mar 2015 16:41:25 +0000 (16:41 +0000)]
Rework r233647, "llvm/examples: Suppress building a few JIT examples."

It didn't work with "install".

llvm-svn: 233708

9 years ago[ASan/Win] UnmapOrDie should not fail on zero address / size
Timur Iskhodzhanov [Tue, 31 Mar 2015 16:39:20 +0000 (16:39 +0000)]
[ASan/Win] UnmapOrDie should not fail on zero address / size

llvm-svn: 233707

9 years agoSema: Accept pointers to any address space for builtin functions
Tom Stellard [Tue, 31 Mar 2015 16:39:02 +0000 (16:39 +0000)]
Sema: Accept pointers to any address space for builtin functions

As long as they don't have an address space explicitly defined.

This allows builtins with pointer arguments to be used with OpenCL.

llvm-svn: 233706

9 years agoFix FreeBSD test runs after r233311
Ed Maste [Tue, 31 Mar 2015 16:37:10 +0000 (16:37 +0000)]
Fix FreeBSD test runs after r233311

On FreeBSD LLDB's triple ends up as e.g. "x86_64-unknown-freebsd10.1"
but getPlatform() consumers expect just the name with no version
number.

llvm-svn: 233705

9 years ago[X86, AVX] try to lowerVectorShuffleAsElementInsertion() for all 256-bit vector sub...
Sanjay Patel [Tue, 31 Mar 2015 16:32:11 +0000 (16:32 +0000)]
[X86, AVX] try to lowerVectorShuffleAsElementInsertion() for all 256-bit vector sub-types

I suggested this change in D7898 (http://llvm.org/viewvc/llvm-project?view=revision&revision=231354)

It improves the v4i64 case although not optimally. This AVX codegen:

  vmovq {{.*#+}} xmm0 = mem[0],zero
  vxorpd %ymm1, %ymm1, %ymm1
  vblendpd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3]

Becomes:

  vmovsd {{.*#+}} xmm0 = mem[0],zero

Unfortunately, this doesn't completely solve PR22685. There are still at least 2 problems under here:

    We're not handling v32i8 / v16i16.
    We're not getting the FP / int domains right for instruction selection.

But since this patch alone appears to do no harm, reduces code duplication, and helps v4i64,
I'm submitting this patch ahead of fixing the above.

Differential Revision: http://reviews.llvm.org/D8341

llvm-svn: 233704

9 years agoChanged a function to static.
Hafiz Abid Qadeer [Tue, 31 Mar 2015 16:30:29 +0000 (16:30 +0000)]
Changed a function to static.
A temp object was being created to call StripOffFileName. This function
is not dependent on class members so I am making it static.
No regression on testsuite on Linux.

llvm-svn: 233703

9 years ago[clang-tidy] Clarify message for the google-explicit-constructor check
Alexander Kornienko [Tue, 31 Mar 2015 16:24:44 +0000 (16:24 +0000)]
[clang-tidy] Clarify message for the google-explicit-constructor check

Use "constructors that are callable with a single argument" instead of
"single-argument constructors" when referring to constructors using default
arguments or parameter packs.

llvm-svn: 233702

9 years agotypos; NFC
Sanjay Patel [Tue, 31 Mar 2015 16:17:51 +0000 (16:17 +0000)]
typos; NFC

llvm-svn: 233701

9 years ago[SystemZ] Add Analysis to required_libraries (fall-out from r233688)
Ulrich Weigand [Tue, 31 Mar 2015 15:16:13 +0000 (15:16 +0000)]
[SystemZ] Add Analysis to required_libraries (fall-out from r233688)

Should fix build failures with cmake builds

llvm-svn: 233700

9 years ago[docs] add cross-reference
Scott Douglass [Tue, 31 Mar 2015 15:07:53 +0000 (15:07 +0000)]
[docs] add cross-reference

llvm-svn: 233699

9 years agoclang-format: [JS] Support getters, setters and methods in object literals.
Daniel Jasper [Tue, 31 Mar 2015 14:34:15 +0000 (14:34 +0000)]
clang-format: [JS] Support getters, setters and methods in object literals.

llvm-svn: 233698

9 years agoForce braces on the else branch if they are being added to the if branch.
Samuel Benzaquen [Tue, 31 Mar 2015 13:53:03 +0000 (13:53 +0000)]
Force braces on the else branch if they are being added to the if branch.

Summary:
Force braces on the else branch if they are being added to the if branch.
This ensures consistency in the transformed code.

Reviewers: alexfh

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D8708

llvm-svn: 233697

9 years agoExpand MUX instructions early on Hexagon
Krzysztof Parzyszek [Tue, 31 Mar 2015 13:35:12 +0000 (13:35 +0000)]
Expand MUX instructions early on Hexagon

This time with all files included.

llvm-svn: 233696

9 years agoRevert 233694. Weak SVN-fu.
Krzysztof Parzyszek [Tue, 31 Mar 2015 13:32:32 +0000 (13:32 +0000)]
Revert 233694.  Weak SVN-fu.

llvm-svn: 233695

9 years agoExpand MUX instructions early on Hexagon
Krzysztof Parzyszek [Tue, 31 Mar 2015 13:29:17 +0000 (13:29 +0000)]
Expand MUX instructions early on Hexagon

llvm-svn: 233694

9 years ago[AArch64] Add v8.1a "Rounding Double Multiply Add/Subtract" extension
Vladimir Sukharev [Tue, 31 Mar 2015 13:15:48 +0000 (13:15 +0000)]
[AArch64] Add v8.1a "Rounding Double Multiply Add/Subtract" extension

Reviewers: t.p.northover, jmolloy

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8502

llvm-svn: 233693

9 years ago[Mips] Remove redundant #include
Simon Atanasyan [Tue, 31 Mar 2015 13:04:08 +0000 (13:04 +0000)]
[Mips] Remove redundant #include

No functional changes.

llvm-svn: 233692

9 years ago[Mips] Do not use MIPS assembler and disassembler in the tests
Simon Atanasyan [Tue, 31 Mar 2015 13:04:01 +0000 (13:04 +0000)]
[Mips] Do not use MIPS assembler and disassembler in the tests

No functional changes.

llvm-svn: 233691

9 years ago[SystemZ] Support RISBGN instruction on zEC12
Ulrich Weigand [Tue, 31 Mar 2015 12:58:17 +0000 (12:58 +0000)]
[SystemZ] Support RISBGN instruction on zEC12

So far, we do not yet support any instruction specific to zEC12.
Most of the facilities added with zEC12 are indeed not very useful
to compiler code generation, but there is one exception: the
miscellaneous-extensions facility provides the RISBGN instruction,
which is a variant of RISBG that does not set the condition code.

Add support for this facility, MC support for RISBGN, and CodeGen
support for prefering RISBGN over RISBG on zEC12, unless we can
actually make use of the condition code set by RISBG.

llvm-svn: 233690

9 years ago[SystemZ] Use POPCNT instruction on z196
Ulrich Weigand [Tue, 31 Mar 2015 12:56:33 +0000 (12:56 +0000)]
[SystemZ] Use POPCNT instruction on z196

We already exploit a number of instructions specific to z196,
but not yet POPCNT.  Add support for the population-count
facility, MC support for the POPCNT instruction, CodeGen
support for using POPCNT, and implement the getPopcntSupport
TargetTransformInfo hook.

llvm-svn: 233689

9 years ago[SystemZ] Provide basic TargetTransformInfo implementation
Ulrich Weigand [Tue, 31 Mar 2015 12:52:27 +0000 (12:52 +0000)]
[SystemZ] Provide basic TargetTransformInfo implementation

This hooks up the TargetTransformInfo machinery for SystemZ,
and provides an implementation of getIntImmCost.

In addition, the patch adds the isLegalICmpImmediate and
isLegalAddImmediate TargetLowering overrides, and updates
a couple of test cases where we now generate slightly
better code.

llvm-svn: 233688

9 years ago[Sanitizer RT] Put the Symbolizer module name string ownership in order
Timur Iskhodzhanov [Tue, 31 Mar 2015 12:50:05 +0000 (12:50 +0000)]
[Sanitizer RT] Put the Symbolizer module name string ownership in order

Reviewed at http://reviews.llvm.org/D8666

llvm-svn: 233687

9 years agoFix the operand encoding in the test instruction.
Rafael Espindola [Tue, 31 Mar 2015 12:31:55 +0000 (12:31 +0000)]
Fix the operand encoding in the test instruction.

Fixes pr22995.

llvm-svn: 233686

9 years ago[LLDB][MIPS] Read/Write register for MIPS64
Mohit K. Bhakkad [Tue, 31 Mar 2015 12:01:27 +0000 (12:01 +0000)]
[LLDB][MIPS] Read/Write register for MIPS64
Patch by Sagar Thakur

Reviewers: clayborg, tberghammer.

Subscribers: jaydeep, bhushan, mohit.bhakkad, llvm-commits.

Differential Revision: http://reviews.llvm.org/D8695

llvm-svn: 233685

9 years ago[PowerPC] Remove TargetMachine CPU auto-detection
Ulrich Weigand [Tue, 31 Mar 2015 12:01:06 +0000 (12:01 +0000)]
[PowerPC] Remove TargetMachine CPU auto-detection

As was done for X86 in r206094.

llvm-svn: 233684

9 years agoFix type detection for 'char' variables
Tamas Berghammer [Tue, 31 Mar 2015 10:21:50 +0000 (10:21 +0000)]
Fix type detection for 'char' variables

A char can have signed and unsigned encoding but previously lldb always
assumed it is signed. This CL adds a logic to detect the encoding of
'char' types based on the default encoding on the target architecture.
It fixes variable printing and expression evaluation on architectures
where 'char' is signed by default.

Differential revision: http://reviews.llvm.org/D8636

llvm-svn: 233682

9 years ago[SDAG] Move TRUNCATE splitting logic into a helper, and use
James Molloy [Tue, 31 Mar 2015 10:20:58 +0000 (10:20 +0000)]
[SDAG] Move TRUNCATE splitting logic into a helper, and use
 it more liberally.

SplitVecOp_TRUNCATE has logic for recursively splitting oversize vectors
that need more than one round of splitting to become legal. There are many
other ISD nodes that could benefit from this logic, so factor it out and
use it for FP_TO_UINT,FP_TO_SINT,SINT_TO_FP,UINT_TO_FP and FTRUNC.

llvm-svn: 233681

9 years agoFix Windows build after r233679
Tamas Berghammer [Tue, 31 Mar 2015 10:11:04 +0000 (10:11 +0000)]
Fix Windows build after r233679

llvm-svn: 233680

9 years agoMove several plugin to its own namespace
Tamas Berghammer [Tue, 31 Mar 2015 09:52:22 +0000 (09:52 +0000)]
Move several plugin to its own namespace

Affected paths:
* Plugins/Platform/Android/*
* Plugins/Platform/Linux/*
* Plugins/Platform/gdb-server/*
* Plugins/Process/Linux/*
* Plugins/Process/gdb-remote/*

Differential revision: http://reviews.llvm.org/D8654

llvm-svn: 233679

9 years agoCreate new test for expression evaluation where char type matters
Tamas Berghammer [Tue, 31 Mar 2015 09:48:53 +0000 (09:48 +0000)]
Create new test for expression evaluation where char type matters

char is a special type in C++ which can be signed/unsigned and have to
be distinguished both from "signed char" and from "unsigned char". This
test check for this behaviour during the expression evaluation with
different compiler settings.

Differential revision: http://reviews.llvm.org/D8657

llvm-svn: 233678

9 years agoMake new test slightly more generic to allow running it under
Daniel Jasper [Tue, 31 Mar 2015 09:08:16 +0000 (09:08 +0000)]
Make new test slightly more generic to allow running it under
non-standard test runners.

llvm-svn: 233677

9 years agoAlso use dl_iterate_phdr() on CloudABI.
Ed Schouten [Tue, 31 Mar 2015 07:59:53 +0000 (07:59 +0000)]
Also use dl_iterate_phdr() on CloudABI.

The unwinder now works on CloudABI. All exception-related libc++ tests
now pass.

This change was actually part of D8169, which also adds support for
FreeBSD. That, however, still requires some more polishing.

llvm-svn: 233676

9 years agoDo not scale tile loops
Tobias Grosser [Tue, 31 Mar 2015 07:52:36 +0000 (07:52 +0000)]
Do not scale tile loops

We now generate tile loops as:

 for (int c1 = 0; c1 <= 47; c1 += 1)
   for (int c2 = 0; c2 <= 47; c2 += 1)
     for (int c3 = 0; c3 <= 31; c3 += 1)
       for (int c4 = 0; c4 <= 31; c4 += 4)
         #pragma simd
         for (int c5 = c4; c5 <= c4 + 3; c5 += 1)
           Stmt_for_body3(32 * c1 + c3, 32 * c2 + c5);

instead of

 for (int c1 = 0; c1 <= 1535; c1 += 32)
   for (int c2 = 0; c2 <= 1535; c2 += 32)
     for (int c3 = 0; c3 <= 31; c3 += 1)
       for (int c4 = 0; c4 <= 31; c4 += 4)
         #pragma simd
         for (int c5 = c4; c5 <= c4 + 3; c5 += 1)
           Stmt_for_body3(c1 + c3, c2 + c5);

Run-time performance-wise this makes little difference, but this gives a large
reduction in compile time (10-30% on 17 LNT benchmarks). Apparently the isl
AST generator is not yet very efficient in generating the latter.

llvm-svn: 233675

9 years ago[X86] Stop changing result of getHostCPUName based on whether the processor supports...
Craig Topper [Tue, 31 Mar 2015 06:18:31 +0000 (06:18 +0000)]
[X86] Stop changing result of getHostCPUName based on whether the processor supports AVX. getHostCPUFeatures should be used instead to determine whether to support AVX.

llvm-svn: 233674

9 years agoMake llc use getHostCPUFeatures when 'native' is specified for cpu.
Craig Topper [Tue, 31 Mar 2015 05:52:57 +0000 (05:52 +0000)]
Make llc use getHostCPUFeatures when 'native' is specified for cpu.

This is necessary for x86 where not all Sandybridge, Ivybrige, Haswell, and Broadwell CPUs support AVX. Currently we modify the CPU name back to Nehalem for this case, but that turns off additional features for these CPUs.

llvm-svn: 233673

9 years ago[X86] Use getHostCPUFeatures when 'native' is specified for cpu.
Craig Topper [Tue, 31 Mar 2015 05:45:00 +0000 (05:45 +0000)]
[X86] Use getHostCPUFeatures when 'native' is specified for cpu.

This is necessary because not aall Sandybridge, Ivybrige, Haswell, and Broadwell CPUs support AVX. Currently we modify the CPU name back to Nehalem for this case, but that turns off additional features for these CPUs.

llvm-svn: 233672

9 years ago[X86] Be more robust against unknown Intel family 6 models. Use feature flags to...
Craig Topper [Tue, 31 Mar 2015 05:42:45 +0000 (05:42 +0000)]
[X86] Be more robust against unknown Intel family 6 models. Use feature flags to guess what it might be.

llvm-svn: 233671

9 years agoFix TestSendSignal.py for remote.
Chaoren Lin [Tue, 31 Mar 2015 04:26:34 +0000 (04:26 +0000)]
Fix TestSendSignal.py for remote.

Summary: There is an extra eStateConnected before eStateRunning with remote.

Reviewers: clayborg, ovyalov

Reviewed By: clayborg, ovyalov

Subscribers: lldb-commits

Differential Revision: http://reviews.llvm.org/D8714

llvm-svn: 233670

9 years ago[libcxx] Add code coverage configuration to CMake and LIT.
Eric Fiselier [Tue, 31 Mar 2015 04:15:45 +0000 (04:15 +0000)]
[libcxx] Add code coverage configuration to CMake and LIT.

Summary:
This patch adds configuration to CMake and LIT for running the libc++ test-suite to generate code coverage.

To use code coverage use following instructions.

* Find the clang resource dir using `$CXX -print-search-dirs`. Let <library-dir> be the first library search directory.
* `cmake <regular-options> -DLIBCXX_GENERATE_COVERAGE=ON -DLIBCXX_COVERAGE_LIBRARY=<library-dir>/lib/<platform>/libclang_rt.profile.a <source>`
* `make cxx`
* `make check-libcxx`
* `make generate-libcxx-coverage`

The reason I want this patch upstreamed is so I can setup a bot that generates code coverage and posts in online for every revision.

Reviewers: mclow.lists, jroelofs, danalbert

Reviewed By: danalbert

Differential Revision: http://reviews.llvm.org/D8716

llvm-svn: 233669

9 years ago[X86] Generate MOVNT for all vector types.
Ahmed Bougacha [Tue, 31 Mar 2015 03:16:51 +0000 (03:16 +0000)]
[X86] Generate MOVNT for all vector types.

We used to miss non-Q YMM integer vectors, and, non-Q/D XMM integer
vectors.
While there, change the v4i32 patterns to prefer MOVNTDQ.

llvm-svn: 233668

9 years agotools: Unify how verifyModule() is called
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 03:07:23 +0000 (03:07 +0000)]
tools: Unify how verifyModule() is called

Unify the error messages for the various tools when `verifyModule()`
fails on an input module.  The "brave new way" is:

    lltool: path/to/input.ll: error: input module is broken!

llvm-svn: 233667

9 years ago[bpf] mark mov instructions as ReMaterializable
Alexei Starovoitov [Tue, 31 Mar 2015 02:49:58 +0000 (02:49 +0000)]
[bpf] mark mov instructions as ReMaterializable

loading immediate into register is cheap, so take advantage of remat.

llvm-svn: 233666

9 years agoVerifier: Don't return early from verifyTypeRefs()
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 02:37:13 +0000 (02:37 +0000)]
Verifier: Don't return early from verifyTypeRefs()

We'll no longer crash in the `verifyTypeRefs()` (used to be called
`verifyDebugInfo()`), so there's no reason to return early here.  Remove
the `EverBroken` member since this was the only use!

llvm-svn: 233665

9 years agoVerifier: Explicitly verify type references
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 02:27:32 +0000 (02:27 +0000)]
Verifier: Explicitly verify type references

`verifyDebugInfo()` was doing two things:

  - Asserting on unresolved type references.
  - Calling `Verify()` functions for various types of debug info.

The `Verify()` functions have been gutted, so rename the function to
`verifyTypeRefs()` and explicitly check those.  Instead of assertions,
we get nice error messages now.

llvm-svn: 233664

9 years agoVerifier: Move over DISubprogram::Verify()
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 02:09:55 +0000 (02:09 +0000)]
Verifier: Move over DISubprogram::Verify()

Move over the remaining (somewhat complicated) check from
`DISubprogram::Verify()`.  I suspect this check could be optimized --
e.g., it would be nice not to do another full traversal here -- but it's
not exactly obvious how.  For now, just bring it over as is.

Once we have a better model for the "canonical" subprogram of a
`Function`, we should enforce that all `!dbg` attachments lead to the
canonical one.

llvm-svn: 233663

9 years agoDebugInfo: Rewrite llvm::getDISubprogram(), NFC
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 02:06:28 +0000 (02:06 +0000)]
DebugInfo: Rewrite llvm::getDISubprogram(), NFC

Simplify implementation of `llvm::getDISubprogram()`.  I might go
through and see how difficult it is to update the users, since this
function doesn't really seem necessary anymore.

llvm-svn: 233662

9 years agoDebugInfo: Hide isScopeRef() and isTypeRef() behind NDEBUG
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 01:47:55 +0000 (01:47 +0000)]
DebugInfo: Hide isScopeRef() and isTypeRef() behind NDEBUG

The copies of these in `lib/IR/DebugInfo.cpp` are apparently [1] only
used in assertions now, so hide them behind `#ifndef NDEBUG`.

[1]: http://lab.llvm.org:8011/builders/lld-x86_64-darwin13/builds/9238

llvm-svn: 233661

9 years agoDebugInfo: Factor out RETURN_FROM_RAW, etc.
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 01:47:37 +0000 (01:47 +0000)]
DebugInfo: Factor out RETURN_FROM_RAW, etc.

Remove the helper macros `RETURN_FROM_RAW()`,
`RETURN_DESCRIPTOR_FROM_RAW()`, and `RETURN_REF_FROM_RAW()`, since they
don't do anything too special anymore.  This loses an `assert(get())`,
but I'm hoping any crashes were shaken out when r232844 landed a few
weeks ago.

llvm-svn: 233660

9 years agoAllow enabling CCache through an env variable. This helps enable/disable the option...
Eric Fiselier [Tue, 31 Mar 2015 01:41:32 +0000 (01:41 +0000)]
Allow enabling CCache through an env variable. This helps enable/disable the option on buildbots.

llvm-svn: 233659

9 years agoVerifier: Check reference flags in debug info
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 01:28:58 +0000 (01:28 +0000)]
Verifier: Check reference flags in debug info

Move over checks of `&` and `&&` flags.

llvm-svn: 233658

9 years agoVerifier: Move more debug info checks away from Verify()
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 01:28:22 +0000 (01:28 +0000)]
Verifier: Move more debug info checks away from Verify()

Most of these checks were already in the `Verifier` so this is more of a
cleanup.  Now almost everything is over there.

Now that require a `name:` for `MDGlobalVariable`, add a check in
`LLParser` for it.

llvm-svn: 233657

9 years agoDebugInfo: Move debug info flags to the new hierarchy
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 01:19:51 +0000 (01:19 +0000)]
DebugInfo: Move debug info flags to the new hierarchy

Move definition of the debug info flags to the new hierarchy, but leave
them duplicated in `DIDescriptor` for now to reduce code churn.  My
current plan is to remove `DIDescriptor` entirely, so the duplication
should go away naturally.

llvm-svn: 233656

9 years agoFix call to llvm::Target::createMCInstPrinter.
Chaoren Lin [Tue, 31 Mar 2015 00:59:13 +0000 (00:59 +0000)]
Fix call to llvm::Target::createMCInstPrinter.

llvm-svn: 233655

9 years agoVerifier: Move checks over from DIDescriptor::Verify()
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 00:47:15 +0000 (00:47 +0000)]
Verifier: Move checks over from DIDescriptor::Verify()

Move over some more checks from `DIDescriptor::Verify()`, and change
`LLParser` to require non-null `file:` fields in compile units.

I've ignored the comment in test/Assembler/metadata-null-operands.ll
since I disagree with it.  At the time that test was written (r229960),
the debug info verifier wasn't on by default, so my comment there is in
the context of not expecting the verifier to be useful.  It is now, and
besides that, since r233394 we can check when parsing textual IR whether
an operand is null that shouldn't be.

llvm-svn: 233654

9 years ago[AArch64] Fix poor codegen for add immediate.
Quentin Colombet [Tue, 31 Mar 2015 00:31:13 +0000 (00:31 +0000)]
[AArch64] Fix poor codegen for add immediate.
We used to match the register variant before the immediate when the register
argument could be implicitly zero-extended.

llvm-svn: 233653

9 years agoIncreate backlog of lldb-platform's listener socket
Vince Harron [Tue, 31 Mar 2015 00:27:10 +0000 (00:27 +0000)]
Increate backlog of lldb-platform's listener socket

lldb-platform's listener socket only had a backlog of one connection.
That means that if more than one client connected simultaneously, the
connection would be refused. The test suite can be run remotely with
dozens of threads connecting simultaneously. Raised this limit to 100
to effectively eliminate lost connections.

Test Plan:
run tests against a remote target

Differential Revision: http://reviews.llvm.org/D8696

llvm-svn: 233652

9 years agoFix socket leak in lldb-server platform
Vince Harron [Tue, 31 Mar 2015 00:24:51 +0000 (00:24 +0000)]
Fix socket leak in lldb-server platform

lldb-server server process was holding on to a connection to every
client that connected. This continued until the process hit it's
file limit and exited.

lldb-server p --listen *:5432 --server

run remote tests against that server more than 1000 times

Differential Revision: http://reviews.llvm.org/D8697

llvm-svn: 233651

9 years agoDebugInfo: Remove LexicalBlockFile scope/context distinction
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 00:10:37 +0000 (00:10 +0000)]
DebugInfo: Remove LexicalBlockFile scope/context distinction

Two things here:

 1. I read `getScope()` and `getContext()` backwards in r233640.  There
    was no need for `getScopeOfScope()`.  Obviously not enough test
    coverage here (as I said in that commit, I'm going to come back to
    that), but anyway I'm reverting to the behaviour before r233640.
 2. The callers that use `DILexicalBlockFile::getContext()` don't seem
    to care about the difference.  Just have it redirect to `getScope()`
    so I can't get confused again.

llvm-svn: 233650

9 years agoUpdate for llvm commit r233648.
Eric Christopher [Tue, 31 Mar 2015 00:10:23 +0000 (00:10 +0000)]
Update for llvm commit r233648.

llvm-svn: 233649

9 years agoReplace the MCSubtargetInfo parameter with a Triple when creating
Eric Christopher [Tue, 31 Mar 2015 00:10:04 +0000 (00:10 +0000)]
Replace the MCSubtargetInfo parameter with a Triple when creating
an MCInstPrinter. Update all callers and use where we wanted a Triple
previously.

llvm-svn: 233648

9 years agollvm/examples: Suppress building a few JIT examples.
NAKAMURA Takumi [Tue, 31 Mar 2015 00:03:43 +0000 (00:03 +0000)]
llvm/examples: Suppress building a few JIT examples.

  examples/ExceptionDemo/CMakeFiles/ExceptionDemo.dir/ExceptionDemo.cpp.o:(.data.rel.ro._ZTIN4llvm18MCJITMemoryManagerE[_ZTIN4llvm18MCJITMemoryManagerE]+0x10): undefined reference to `typeinfo for llvm::RuntimeDyld::MemoryManager'

llvm-svn: 233647

9 years agollvm/examples: Add RuntimeDyld as libdeps.
NAKAMURA Takumi [Tue, 31 Mar 2015 00:03:36 +0000 (00:03 +0000)]
llvm/examples: Add RuntimeDyld as libdeps.

llvm-svn: 233646

9 years agoFix .gitignore to ignore symlinks.
Chaoren Lin [Tue, 31 Mar 2015 00:03:00 +0000 (00:03 +0000)]
Fix .gitignore to ignore symlinks.

llvm-svn: 233645

9 years agoLexicalScopes: Cleanup remaining uses of DebugLoc
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 23:58:59 +0000 (23:58 +0000)]
LexicalScopes: Cleanup remaining uses of DebugLoc

llvm-svn: 233644

9 years agoDebugLoc: Remove getFromDILexicalBlock()
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 23:47:26 +0000 (23:47 +0000)]
DebugLoc: Remove getFromDILexicalBlock()

The only user of `DebugLoc::getFromDILexicalBlock()` was creating a new
`MDLocation` as convenient API for passing an `MDScope`.  Stop doing
that, and remove the API.  If in the future we actually *want* to create
new DebugLocs, calling `MDLexicalBlock::get()` makes more sense.

llvm-svn: 233643

9 years ago[Objective-C patch]. Amend TransformObjCMessageExpr to handle call
Fariborz Jahanian [Mon, 30 Mar 2015 23:30:24 +0000 (23:30 +0000)]
[Objective-C patch]. Amend TransformObjCMessageExpr to handle call
to 'super' of instance/class methods and not assert.
rdar://20350364

llvm-svn: 233642

9 years agoMake the new tests better; make sure that we're testing the case where no reallocatio...
Marshall Clow [Mon, 30 Mar 2015 23:26:16 +0000 (23:26 +0000)]
Make the new tests better; make sure that we're testing the case where no reallocation has to happen

llvm-svn: 233641

9 years agoLexicalScopes: Use debug info hierarchy pervasively
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 23:21:21 +0000 (23:21 +0000)]
LexicalScopes: Use debug info hierarchy pervasively

Pervasively use the types provided by the debug info hierarchy rather
than `MDNode` in `LexicalScopes`.

I noticed (again, I guess, based on comments in the implementation?)
that `DILexicalBlockFile::getScope()` returns something different from
`DILexicalBlockFile::getContext()`.  I created a local helper for
getting the same logic from `MDLexicalBlockFile` called
`getScopeOfScope()`.  I still don't really understand it, but I've added
some FIXMEs and I'll come back to it (I suspect the way we encode these
objects isn't really ideal).

Note that my previous commit r233610 accidentally changed behaviour in
`findLexicalScope()` -- it transitioned from a call to
`DILexicalBlockFile::getScope()` to `MDLexicalBlockFile::getScope()`
(sounds right, doesn't it?) -- so I've fixed that as a drive-by.  No
tests failed with my error, so it looks like we're missing some coverage
here... when I come back to understand the logic, I'll see if I can add
some.

Other than the fix to `findLexicalScope()`, no functionality change.

llvm-svn: 233640

9 years agoSilence an unused variable warning.
David Majnemer [Mon, 30 Mar 2015 23:14:45 +0000 (23:14 +0000)]
Silence an unused variable warning.

No functional change intended.

llvm-svn: 233639

9 years agoMove lib/Fuzzer docs from a README.txt to a proper .rst file.
Kostya Serebryany [Mon, 30 Mar 2015 23:05:30 +0000 (23:05 +0000)]
Move lib/Fuzzer docs from a README.txt to a proper .rst file.

Summary:
Move lib/Fuzzer docs from a README.txt to a proper .rst file.
This change does not add any content, just formatting.

Test Plan: n/a

Reviewers: samsonov

Reviewed By: samsonov

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8710

llvm-svn: 233638

9 years ago[fuzzer] when a single unit takes over 1 second to run and it is the slowest one...
Kostya Serebryany [Mon, 30 Mar 2015 23:04:35 +0000 (23:04 +0000)]
[fuzzer] when a single unit takes over 1 second to run and it is the slowest one so far, print it.

llvm-svn: 233637

9 years ago[WinEH] Run cleanup handlers when an exception is thrown
David Majnemer [Mon, 30 Mar 2015 22:58:10 +0000 (22:58 +0000)]
[WinEH] Run cleanup handlers when an exception is thrown

Generate tables in the .xdata section representing what actions to take
when an exception is thrown.  This currently fills in state for
cleanups, catch handlers are still unfinished.

llvm-svn: 233636

9 years agoTransfer implicit operands when expanding the RET_ReallyLR pseudo instruction.
Juergen Ributzka [Mon, 30 Mar 2015 22:45:56 +0000 (22:45 +0000)]
Transfer implicit operands when expanding the RET_ReallyLR pseudo instruction.

When we expand the RET_ReallyLR pseudo instruction we also need to transfer the
implicit operands.

The return register is an implicit operand and without it the liveness
calculation generates an incorrect live-out set for the patchpoint.

This fixes rdar://problem/19068476.

llvm-svn: 233635

9 years ago[fuzzer] print various stats in a unified way
Kostya Serebryany [Mon, 30 Mar 2015 22:44:03 +0000 (22:44 +0000)]
[fuzzer] print various stats in a unified way

llvm-svn: 233624

9 years ago[bpf] add support for bswap instructions
Alexei Starovoitov [Mon, 30 Mar 2015 22:40:40 +0000 (22:40 +0000)]
[bpf] add support for bswap instructions

BPF has cpu_to_be and cpu_to_le instructions.
For now assume little endian and generate cpu_to_be for ISD::BSWAP.

llvm-svn: 233620

9 years ago[bpf] fix build
Alexei Starovoitov [Mon, 30 Mar 2015 22:40:35 +0000 (22:40 +0000)]
[bpf] fix build

fix build broken due to r233599.
Would still need to switch to MDLocation long term.

llvm-svn: 233619

9 years ago[Mips] Move the `Elf_RegInfo` structure declaration to the separate file
Simon Atanasyan [Mon, 30 Mar 2015 22:36:07 +0000 (22:36 +0000)]
[Mips] Move the `Elf_RegInfo` structure declaration to the separate file

No functional changes.

llvm-svn: 233618

9 years ago[ELF] Change type of `OutputSection::setType()` argument to int64_t
Simon Atanasyan [Mon, 30 Mar 2015 22:36:01 +0000 (22:36 +0000)]
[ELF] Change type of `OutputSection::setType()` argument to int64_t

Type of `OutputSection::_type` field is int64_t. This change makes
the field's and the argument's types consistent and allows to assign
full range of values to the `OutputSection::_type` field.

llvm-svn: 233617

9 years ago[Mips] clang-format the code
Simon Atanasyan [Mon, 30 Mar 2015 22:35:56 +0000 (22:35 +0000)]
[Mips] clang-format the code

No functional changes.

llvm-svn: 233616

9 years agoRename const char *Triple argument to TT to avoid shadowing llvm::Triple.
Eric Christopher [Mon, 30 Mar 2015 22:31:14 +0000 (22:31 +0000)]
Rename const char *Triple argument to TT to avoid shadowing llvm::Triple.

llvm-svn: 233615

9 years agoRemove unused MCSubtargetInfo argument from the X86 MCInstPrinter ctors.
Eric Christopher [Mon, 30 Mar 2015 22:16:37 +0000 (22:16 +0000)]
Remove unused MCSubtargetInfo argument from the X86 MCInstPrinter ctors.

llvm-svn: 233614

9 years agoDFSan-based fuzzer (proof of concept).
Kostya Serebryany [Mon, 30 Mar 2015 22:09:51 +0000 (22:09 +0000)]
DFSan-based fuzzer (proof of concept).

Summary:
This adds a simple DFSan-based (i.e. taint-guided) fuzzer mutator,
see the comments for details.

Test Plan: a test added

Reviewers: samsonov, pcc

Reviewed By: samsonov, pcc

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D8669

llvm-svn: 233613

9 years agoRemove unused MCSubtargetInfo argument from the Sparc MCInstPrinter ctors.
Eric Christopher [Mon, 30 Mar 2015 22:09:43 +0000 (22:09 +0000)]
Remove unused MCSubtargetInfo argument from the Sparc MCInstPrinter ctors.

llvm-svn: 233612

9 years agoRemove unused MCSubtargetInfo argument from the NVPTX MCInstPrinter ctors.
Eric Christopher [Mon, 30 Mar 2015 22:03:16 +0000 (22:03 +0000)]
Remove unused MCSubtargetInfo argument from the NVPTX MCInstPrinter ctors.

llvm-svn: 233611

9 years agoLexicalScopes: Use MDLocation directly instead of DebugLoc
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 21:54:46 +0000 (21:54 +0000)]
LexicalScopes: Use MDLocation directly instead of DebugLoc

There's no benefit to using `DebugLoc` here.  Moreover, this will let a
follow-up commit work with `MDScope` directly instead of `DebugLoc`.

llvm-svn: 233610

9 years agoRemove unused MCSubtargetInfo argument from the ARM MCInstPrinter ctors.
Eric Christopher [Mon, 30 Mar 2015 21:52:28 +0000 (21:52 +0000)]
Remove unused MCSubtargetInfo argument from the ARM MCInstPrinter ctors.

llvm-svn: 233609

9 years agoRemove unused MCSubtargetInfo argument from the AArch64 MCInstPrinter ctors.
Eric Christopher [Mon, 30 Mar 2015 21:52:26 +0000 (21:52 +0000)]
Remove unused MCSubtargetInfo argument from the AArch64 MCInstPrinter ctors.

llvm-svn: 233608

9 years agoRemove unused Target argument from MCInstPrinter ctor functions.
Eric Christopher [Mon, 30 Mar 2015 21:52:21 +0000 (21:52 +0000)]
Remove unused Target argument from MCInstPrinter ctor functions.

llvm-svn: 233607