platform/upstream/llvm.git
6 years ago[clang-tidy] implement check for goto
Jonas Toth [Wed, 17 Jan 2018 10:27:41 +0000 (10:27 +0000)]
[clang-tidy] implement check for goto

The usage of `goto` is discourage in C++ since forever. This check implements
a warning for every `goto`. Even though there are (rare) valid use cases for
`goto`, better high level constructs should be used.

`goto` is used sometimes in C programs to free resources at the end of
functions in the case of errors. This pattern is better implemented with
RAII in C++.

Reviewers: aaron.ballman, alexfh, hokein

Reviewed By: aaron.ballman

Subscribers: lebedev.ri, jbcoe, Eugene.Zelenko, klimek, nemanjai, mgorny, xazax.hun, kbarton, cfe-commits

Differential Revision: https://reviews.llvm.org/D41815

llvm-svn: 322626

6 years ago[ELF] - Stop mixing order of -defsym/-script commands.
George Rimar [Wed, 17 Jan 2018 10:24:49 +0000 (10:24 +0000)]
[ELF] - Stop mixing order of -defsym/-script commands.

Previously we always handled -defsym after other commands in command line.
That made impossible to overload values set by -defsym from linker script:

 test.script:
  foo = 0x22;
-defsym=foo=0x11 -script t.script
would always set foo to 0x11.

That is inconstent with common logic which allows to override command line
options. it is inconsistent with bfd behavior and seems breaks assumption that
-defsym is the same as linker script assignment, as -defsyms always handled out of
command line order.

Patch fixes the handling order.

Differential revision: https://reviews.llvm.org/D42054

llvm-svn: 322625

6 years ago[clang-tidy] fix minor formatting issue
Jonas Toth [Wed, 17 Jan 2018 10:23:50 +0000 (10:23 +0000)]
[clang-tidy] fix minor formatting issue

llvm-svn: 322624

6 years agoAllow usage of X86-prefixes as separate instrs.
Andrew V. Tischenko [Wed, 17 Jan 2018 10:12:06 +0000 (10:12 +0000)]
Allow usage of X86-prefixes as separate instrs.
Differential Revision: https://reviews.llvm.org/D42102

llvm-svn: 322623

6 years ago[OMPT] Add Workaround for Intel Compiler Bug
Joachim Protze [Wed, 17 Jan 2018 10:06:06 +0000 (10:06 +0000)]
[OMPT] Add Workaround for Intel Compiler Bug

Add Workaround for Intel Compiler Bug with Case#: 03138964

A critical region within a nested task causes a segfault in icc 14-18:

int main()
{
  #pragma omp parallel num_threads(2)
  #pragma omp master
    #pragma omp task
      #pragma omp task
        #pragma omp critical
          printf("test\n");
}
When the critical region is in a separate function, the segault does not occur.
So we add noinline to make sure that the function call stays there.

Differential Revision: https://reviews.llvm.org/D41182

llvm-svn: 322622

6 years ago[OMPT] Rename ompt_mutex_impl_t to kmp_mutex_impl
Joachim Protze [Wed, 17 Jan 2018 10:06:01 +0000 (10:06 +0000)]
[OMPT] Rename ompt_mutex_impl_t to kmp_mutex_impl

The defintion is not part of the spec and thus should not have the prefix
"ompt_" but rather a prefix that indicates that this is implementation
specific.

Differential Revision: https://reviews.llvm.org/D41166

llvm-svn: 322621

6 years ago[OMPT] Return appropiate values for ompt runtime entry points for non-OpenMP threads
Joachim Protze [Wed, 17 Jan 2018 10:05:55 +0000 (10:05 +0000)]
[OMPT] Return appropiate values for ompt runtime entry points for non-OpenMP threads

When the current thread is not an (initialized) OpenMP thread, the runtime
entry points return values that correspond to "not available" or similar

Differential Revision: https://reviews.llvm.org/D41167

llvm-svn: 322620

6 years ago[MC] Fix -stack-size-section on ARM
Sean Eveson [Wed, 17 Jan 2018 09:01:29 +0000 (09:01 +0000)]
[MC] Fix -stack-size-section on ARM

Change symbol values in the stack_size section from being 8 bytes, to being a target dependent size.

Differential Revision: https://reviews.llvm.org/D42108

llvm-svn: 322619

6 years ago[CodeGen] Fix a crash on mangling multiversioned functions
George Burgess IV [Wed, 17 Jan 2018 04:46:04 +0000 (04:46 +0000)]
[CodeGen] Fix a crash on mangling multiversioned functions

`multiVersionSortPriority` expects features to have no prefix. We
currently carry them around in the format "+${feature}".

llvm-svn: 322618

6 years agolibcxx: Stop using private MSVC macros in the exception implementation.
Peter Collingbourne [Wed, 17 Jan 2018 04:37:04 +0000 (04:37 +0000)]
libcxx: Stop using private MSVC macros in the exception implementation.

Inline the provided "fallback" definitions (which seem to always be
taken) that expand to __cdecl into users. The fallback definitions
for the *CRTIMP* macros were wrong in the case where the CRT is being
linked statically, so define our own macro as a replacement.

Differential Revision: https://reviews.llvm.org/D42158

llvm-svn: 322617

6 years ago[X86] In LowerBUILD_VECTOR, rename ExtVT to EltVT so it makes sense.
Craig Topper [Wed, 17 Jan 2018 03:58:21 +0000 (03:58 +0000)]
[X86] In LowerBUILD_VECTOR, rename ExtVT to EltVT so it makes sense.

llvm-svn: 322616

6 years ago[X86] Remove duplicate lines from scheduler models. NFC
Craig Topper [Wed, 17 Jan 2018 03:50:21 +0000 (03:50 +0000)]
[X86] Remove duplicate lines from scheduler models. NFC

llvm-svn: 322615

6 years ago[pdbutil] Replace 0 byte PDB input with correct version to fix failing unit test
Aaron Smith [Wed, 17 Jan 2018 03:48:07 +0000 (03:48 +0000)]
[pdbutil] Replace 0 byte PDB input with correct version to fix failing unit test

llvm-svn: 322614

6 years ago[Support] Return an enum instead of an unsigned; NFC.
George Burgess IV [Wed, 17 Jan 2018 03:12:06 +0000 (03:12 +0000)]
[Support] Return an enum instead of an unsigned; NFC.

We seem to be (logically) returning ArchExtKinds here in all cases, so
the return type should reflect that.

The static_cast is necessary because `A.ID` is actually an `unsigned`,
presumably since we use `decltype(A)` to represent extended attributes
for both ARM and AArch64, which use distinct `ArchExtKinds`.

We can't trivially make the same change for ARM, because one of the
values it returns is the bitwise-or of two `ARM::ArchExtKind`s.

llvm-svn: 322613

6 years ago[analyzer] support a mode to only show relevant lines in HTML diagnostics
George Karpenkov [Wed, 17 Jan 2018 02:59:11 +0000 (02:59 +0000)]
[analyzer] support a mode to only show relevant lines in HTML diagnostics

HTML diagnostics can be an overwhelming blob of pages of code.
This patch adds a checkbox which filters this list down to only the
lines *relevant* to the counterexample by e.g. skipping branches which
analyzer has assumed to be infeasible at a time.

The resulting amount of output is much smaller, and often fits on one
screen, and also provides a much more readable diagnostics.

Differential Revision: https://reviews.llvm.org/D41378

llvm-svn: 322612

6 years agoMake DiagnosticsEngine() take DiagOpts as DiagnosticsEngine.
Nico Weber [Wed, 17 Jan 2018 02:55:27 +0000 (02:55 +0000)]
Make DiagnosticsEngine() take DiagOpts as DiagnosticsEngine.

No behavior change, but makes it a bit clearer that DiagnosticsEngine adds a
ref to DiagOpts.

llvm-svn: 322611

6 years agoFix build error - 'default label in switch which covers all enumeration values'
Aaron Smith [Wed, 17 Jan 2018 01:49:01 +0000 (01:49 +0000)]
Fix build error - 'default label in switch which covers all enumeration values'

llvm-svn: 322610

6 years ago[WebAssembly] Define __heap_base global
Sam Clegg [Wed, 17 Jan 2018 01:34:31 +0000 (01:34 +0000)]
[WebAssembly] Define __heap_base global

This is an immutable exported global representing
the start of the heap area.  It is a page aligned.

Differential Revision: https://reviews.llvm.org/D42030

llvm-svn: 322609

6 years agoFix pretty printing the unspecified param of a variadic function
Aaron Smith [Wed, 17 Jan 2018 01:22:03 +0000 (01:22 +0000)]
Fix pretty printing the unspecified param of a variadic function

Summary:
 - Fix a bug in PrettyBuiltinDumper that returns "void" as the name for
  an unspecified builtin type. Since the unspecified param of a variadic
  function is considered a builtin of unspecified type in PDBs, we set
  "..." for its name.

  - Provide a method to determine if a PDBSymbolFunc is variadic in
  PrettyFunctionDumper since PDBSymbolFunc::getArgument() doesn't return the
  last unspecified-type param.

  - Add a pretty-func-dumper.test to test pretty dumping of variadic
  functions.

Reviewers: zturner, llvm-commits

Reviewed By: zturner

Differential Revision: https://reviews.llvm.org/D41801

llvm-svn: 322608

6 years agocfi: Add a blacklist entry for MSVC's std::get_temporary_buffer function.
Peter Collingbourne [Wed, 17 Jan 2018 01:15:33 +0000 (01:15 +0000)]
cfi: Add a blacklist entry for MSVC's std::get_temporary_buffer function.

Differential Revision: https://reviews.llvm.org/D42150

llvm-svn: 322607

6 years ago[WebAssembly] Create synthetic __dso_handle symbol
Sam Clegg [Wed, 17 Jan 2018 01:10:28 +0000 (01:10 +0000)]
[WebAssembly] Create synthetic __dso_handle symbol

This is used by __cxa_ataxit to determine the currently
executing DLL.  Once we fully support DLLs this will need
to be set to some address within the DLL.

The ELF linker added support for this symbol here:
https://reviews.llvm.org/D33856

Differential Revision: https://reviews.llvm.org/D42024

llvm-svn: 322606

6 years ago[LLD][COFF] Report error when file will exceed Windows maximum image size (4GB)
Rui Ueyama [Wed, 17 Jan 2018 01:08:02 +0000 (01:08 +0000)]
[LLD][COFF] Report error when file will exceed Windows maximum image size (4GB)

Patch by Colden Cullen.

Currently, when a large PE (>4 GiB) is to be produced, a crash occurs
because:

1. Calling setOffset with a number greater than UINT32_MAX causes the
   PointerToRawData to overflow

2. When adding the symbol table to the end of the file, the last section's
   offset was used to calculate file size. Because this had overflowed,
   this number was too low, and the file created would not be large enough.
   This lead to the actual crash I saw, which was a buffer overrun.

This change:

1. Adds comment to setOffset, clarifying that overflow can occur, but it's
   somewhat safe because the error will be handled elsewhere

2. Adds file size check after all output data has been created This matches
   the MS link.exe error, which looks prints as: "LINK : fatal error
   LNK1248: image size (10000EFC9) exceeds maximum allowable size
   (FFFFFFFF)"

3. Changes calculate of the symbol table offset to just use the existing
   FileSize. This should match the previous calculations, but doesn't rely
   on the use of a u32 that can overflow.

4. Removes trivial usage of a magic number that bugged me while I was
   debugging the issue

I'm not sure how to add a test for this outside of adding 4GB of object
files to the repo. If there's an easier way, let me know and I'll be
happy to add a test.

Differential Revision: https://reviews.llvm.org/D42010

llvm-svn: 322605

6 years ago[libFuzzer] Support using libc++
Petr Hosek [Wed, 17 Jan 2018 00:42:48 +0000 (00:42 +0000)]
[libFuzzer] Support using libc++

This is needed in case the users of libFuzzer use libc++ in their
code, which the fuzz target (libFuzzer) will be linked against.
When libc++ source is available, we build a private version of it
and link it against libFuzzer which allows using the same static
library against codebases which use both libc++ and libstdc++.

Differential Revision: https://reviews.llvm.org/D37631

llvm-svn: 322604

6 years agoRemove the hardcoded macos deployment targets altogether
Jason Molenda [Wed, 17 Jan 2018 00:22:27 +0000 (00:22 +0000)]
Remove the hardcoded macos deployment targets altogether
from the xcode project files.  We'll build for the current
OS only, by default.

llvm-svn: 322603

6 years agoadd ID as a special acronym to objc property declaration check for property names...
Yan Zhang [Wed, 17 Jan 2018 00:19:35 +0000 (00:19 +0000)]
add ID as a special acronym to objc property declaration check for property names like bundleID.allow using acronyms as suffix.

Reviewers: benhamilton, hokein

Reviewed By: benhamilton

Subscribers: klimek, cfe-commits

Differential Revision: https://reviews.llvm.org/D42143

llvm-svn: 322602

6 years ago[hwasan] Rename sized load/store callbacks to be consistent with ASan.
Evgeniy Stepanov [Tue, 16 Jan 2018 23:15:08 +0000 (23:15 +0000)]
[hwasan] Rename sized load/store callbacks to be consistent with ASan.

Summary: __hwasan_load is now __hwasan_loadN.

Reviewers: kcc

Subscribers: hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D42138

llvm-svn: 322601

6 years ago[X86][BTVER2] Fix scheduling of VCMPSD/VCMPSS instructions
Simon Pilgrim [Tue, 16 Jan 2018 22:15:41 +0000 (22:15 +0000)]
[X86][BTVER2] Fix scheduling of VCMPSD/VCMPSS instructions

For some reason they don't have a trailing i like the packed equivalents.

llvm-svn: 322600

6 years ago[CallSiteSplitting] Pass list of (BB, Conditions) pairs to splitCallSite.
Florian Hahn [Tue, 16 Jan 2018 22:13:15 +0000 (22:13 +0000)]
[CallSiteSplitting] Pass list of (BB, Conditions) pairs to splitCallSite.

This removes some duplication from splitCallSite and makes it easier to
add additional code dealing with each predecessor. It also allows us to
split for more than 2 predecessors, although that is not enabled for
now.

Reviewers: junbuml, mcrosier, davidxl, davide

Reviewed By: junbuml

Differential Revision: https://reviews.llvm.org/D41858

llvm-svn: 322599

6 years ago[X86][BTVER2] Use instrs instead of instregex for low match counts (PR35955)
Simon Pilgrim [Tue, 16 Jan 2018 22:08:43 +0000 (22:08 +0000)]
[X86][BTVER2] Use instrs instead of instregex for low match counts (PR35955)

llvm-svn: 322598

6 years ago[X86][BTVER2] Use instrs instead of instregex for single use matches (PR35955)
Simon Pilgrim [Tue, 16 Jan 2018 21:44:48 +0000 (21:44 +0000)]
[X86][BTVER2] Use instrs instead of instregex for single use matches (PR35955)

llvm-svn: 322597

6 years ago[PPC64] Added vector registers.
Martin Storsjo [Tue, 16 Jan 2018 20:54:10 +0000 (20:54 +0000)]
[PPC64] Added vector registers.

The Registers_ppc64 class needed a couple of changes, both to accommodate the
new registers as well as to handle the overlaps of VS register set
without wasting space.

The save/restore code of V and VS registers was added.
As VS registers depend on the VMX extension, they are processed only if
VMX support is detected (_ARCH_PWR8 for now).

Patch by Leandro Lupori!

Differential Revision: https://reviews.llvm.org/D41906

llvm-svn: 322596

6 years agoSpecify inline for isWhitespace in CommandLine.cpp
Rui Ueyama [Tue, 16 Jan 2018 20:52:32 +0000 (20:52 +0000)]
Specify inline for isWhitespace in CommandLine.cpp

Patch by Takuto Ikuta.

In chromium's component build, there are many directive sections and
commandline parsing takes much time.
This patch is for speed up of lld in RelWithDebInfo build by forcing
inline heavily called isWhitespace function.

10 times link perf stats of blink_core.dll changed like below.

master:
TotalSeconds: 9.8764878
TotalSeconds: 10.1455242
TotalSeconds: 10.075279
TotalSeconds: 10.3397347
TotalSeconds: 9.8361665
TotalSeconds: 9.9544441
TotalSeconds: 9.8960686
TotalSeconds: 9.8877865
TotalSeconds: 10.0551879
TotalSeconds: 10.0492254
Avg: 10.01159047

with this patch:
TotalSeconds: 8.8696762
TotalSeconds: 9.1021585
TotalSeconds: 9.0233893
TotalSeconds: 9.1886175
TotalSeconds: 9.156954
TotalSeconds: 9.0978564
TotalSeconds: 9.1316824
TotalSeconds: 8.8354606
TotalSeconds: 9.2549431
TotalSeconds: 9.4473085
Avg: 9.11080465

llvm-svn: 322595

6 years ago[ExecutionEngine] Rename JITSymbol::isStrongDefinition to isStrong.
Lang Hames [Tue, 16 Jan 2018 20:39:51 +0000 (20:39 +0000)]
[ExecutionEngine] Rename JITSymbol::isStrongDefinition to isStrong.

For symmetry with isWeak, isCommon.

llvm-svn: 322594

6 years agoAdd context to why test was disabled on Windows
Richard Trieu [Tue, 16 Jan 2018 19:53:06 +0000 (19:53 +0000)]
Add context to why test was disabled on Windows

test/Modules/odr_hash-Friend.cpp triggers an assertion in MicrosoftMangle.cpp
This has been reported in PR35939

llvm-svn: 322593

6 years agoMove target MV resolver to COMDAT
Erich Keane [Tue, 16 Jan 2018 19:49:52 +0000 (19:49 +0000)]
Move target MV resolver to COMDAT

As reported here: https://bugs.llvm.org/show_bug.cgi?id=35921
The resolver functions should be in their own
COMDAT regions. This patch sets that up.

Differential Revision: https://reviews.llvm.org/D42110

llvm-svn: 322592

6 years ago[PPC] Add a new register XER aliased to CARRY
Guozhi Wei [Tue, 16 Jan 2018 19:28:50 +0000 (19:28 +0000)]
[PPC] Add a new register XER aliased to CARRY

When "xer" is specified as clobbered register in inline assembler, clang can accept it, but llvm simply ignore it when lowered to machine instructions. It may cause problems later in scheduler.

This patch adds a new register XER aliased to CARRY, and adds it to register class CARRYRC. Now PPCTargetLowering::getRegForInlineAsmConstraint can return correct register number for inline asm constraint "{xer}", and scheduler behave correctly.

Differential Revision: https://reviews.llvm.org/D41967

llvm-svn: 322591

6 years agoInline foot gun into only valid use.
Rafael Espindola [Tue, 16 Jan 2018 19:28:28 +0000 (19:28 +0000)]
Inline foot gun into only valid use.

Symbol had both Visibility and getVisibility() and they had different
meanings. That is just too easy to get wrong.

getVisibility() would compute the visibility of a particular symbol
(foo in bar.o), and Visibility stores the computed value we will put
in the output.

There is only one case when we want what getVisibility() provides, so
inline it.

llvm-svn: 322590

6 years ago[DOCS] Updated current status of OpenMP support, NFC.
Alexey Bataev [Tue, 16 Jan 2018 19:22:49 +0000 (19:22 +0000)]
[DOCS] Updated current status of OpenMP support, NFC.

llvm-svn: 322589

6 years ago[hwasan] Build runtime library with -fPIC, not -fPIE.
Evgeniy Stepanov [Tue, 16 Jan 2018 19:21:45 +0000 (19:21 +0000)]
[hwasan] Build runtime library with -fPIC, not -fPIE.

Summary: -fPIE can not be used when building a shared library.

Reviewers: alekseyshl, peter.smith

Subscribers: kubamracek, llvm-commits, mgorny

Differential Revision: https://reviews.llvm.org/D42121

llvm-svn: 322588

6 years ago[OPENMP] Add support for `depend` clauses on `target teams distribute
Alexey Bataev [Tue, 16 Jan 2018 19:18:24 +0000 (19:18 +0000)]
[OPENMP] Add support for `depend` clauses on `target teams distribute
parallel for simd` directives.

Added codegen for `depend` clauses on `#pragma omp target teams
distribute parallel for simd` directives.

llvm-svn: 322587

6 years agoFix another case we used the wrong visibility.
Rafael Espindola [Tue, 16 Jan 2018 19:02:46 +0000 (19:02 +0000)]
Fix another case we used the wrong visibility.

In here too we want the computed output visibility.

llvm-svn: 322586

6 years ago[OPENMP] Add support for `depend` on `target teams distribute parallel
Alexey Bataev [Tue, 16 Jan 2018 19:02:33 +0000 (19:02 +0000)]
[OPENMP] Add support for `depend` on `target teams distribute parallel
for` directives.

Added codegen for `depend` clauses on `#pragma omp target teams
distribute parallel for` directives.

llvm-svn: 322585

6 years ago[CodeGen] Skip some instructions that shouldn't affect shrink-wrapping
Francis Visoiu Mistrih [Tue, 16 Jan 2018 18:55:26 +0000 (18:55 +0000)]
[CodeGen] Skip some instructions that shouldn't affect shrink-wrapping

r320606 checked for MI.isMetaInstruction which skips all DBG_VALUEs.

This also skips IMPLICIT_DEFs and other instructions that may def / read
a register.

Differential Revision: https://reviews.llvm.org/D42119

llvm-svn: 322584

6 years agoAdd an extra test. NFC.
Rafael Espindola [Tue, 16 Jan 2018 18:53:09 +0000 (18:53 +0000)]
Add an extra test. NFC.

Without this all test would pass if the visibility checks were removed
from SymbolTable::addShared and SymbolTable::addUndefined.

llvm-svn: 322583

6 years ago[GlobalISel][TableGen] Add support for SDNodeXForm
Volkan Keles [Tue, 16 Jan 2018 18:44:05 +0000 (18:44 +0000)]
[GlobalISel][TableGen] Add support for SDNodeXForm

Summary:
This patch adds CustomRenderer which renders the matched
operands to the specified instruction.

Targets can enable the matching of SDNodeXForm by adding
a definition that inherits from GICustomOperandRenderer and
GISDNodeXFormEquiv as follows.

def gi_imm8 : GICustomOperandRenderer<"renderImm8”>,
                       GISDNodeXFormEquiv<imm8_xform>;

Custom renderer functions should be of the form:
void render(MachineInstrBuilder &MIB, const MachineInstr &I);

Reviewers: dsanders, ab, rovka

Reviewed By: dsanders

Subscribers: kristof.beyls, javed.absar, llvm-commits, mgrang, qcolombet

Differential Revision: https://reviews.llvm.org/D42012

llvm-svn: 322582

6 years ago[IslTools] dumpPw: Dump same structure pieces together.
Michael Kruse [Tue, 16 Jan 2018 18:39:42 +0000 (18:39 +0000)]
[IslTools] dumpPw: Dump same structure pieces together.

Print same or similar structure elements together. Previously, the
value could take more importance that the space structure if visited
first in the space nest tree.

Before:
{
  Left[0] -> Right[i]: i >= 0;
  Left[1] -> AnotherRight[i];
  Left[2] -> Right[-1]
}

After:
{
  Left[0] -> Right[i]: i >= 0;
  Left[2] -> Right[-1];
  Left[1] -> AnotherRight[i]
}

llvm-svn: 322581

6 years agoFix another case we were using the wrong visibility.
Rafael Espindola [Tue, 16 Jan 2018 18:21:23 +0000 (18:21 +0000)]
Fix another case we were using the wrong visibility.

llvm-svn: 322580

6 years ago[SLP] Fix for PR32164: Improve vectorization of reverse order of extract operations.
Alexey Bataev [Tue, 16 Jan 2018 18:17:01 +0000 (18:17 +0000)]
[SLP] Fix for PR32164: Improve vectorization of reverse order of extract operations.

Summary: Sometimes vectorization of insertelement instructions with extractelement operands may produce an extra shuffle operation, if these operands are in the reverse order. Patch tries to improve this situation by the reordering of the operands to remove this extra shuffle operation.

Reviewers: mkuper, hfinkel, RKSimon, spatel

Subscribers: mzolotukhin, llvm-commits

Differential Revision: https://reviews.llvm.org/D33954

llvm-svn: 322579

6 years ago[OPENMP] Add support for `depend` clauses on `target parallel for simd`
Alexey Bataev [Tue, 16 Jan 2018 17:55:15 +0000 (17:55 +0000)]
[OPENMP] Add support for `depend` clauses on `target parallel for simd`
directives.

Added codegen for `depend` clauses on `#pragma omp target parallel for
simd` directives.

llvm-svn: 322578

6 years ago[OPENMP] Add support for `depend` clauses on `target parallel for`
Alexey Bataev [Tue, 16 Jan 2018 17:41:04 +0000 (17:41 +0000)]
[OPENMP] Add support for `depend` clauses on `target parallel for`
directives.

Added codegen for `depend` clause on `#pragma omp target parallel for`
directives.

llvm-svn: 322577

6 years agoUse the combined visibility when computing dso_local.
Rafael Espindola [Tue, 16 Jan 2018 17:34:26 +0000 (17:34 +0000)]
Use the combined visibility when computing dso_local.

We track both the combined visibility that will be used for the output
symbol and the original input visibility of the selected symbol.

Almost everything should use the computed visibility.

I will make the names less confusing an a followup patch.

llvm-svn: 322576

6 years ago[OPENMP] Add support for `depend` clauses on `target teams distribute
Alexey Bataev [Tue, 16 Jan 2018 17:22:50 +0000 (17:22 +0000)]
[OPENMP] Add support for `depend` clauses on `target teams distribute
simd` directives.

Added codegen for `depend` clauses on `#pragma omp target teams
distribute simd` directives.

llvm-svn: 322575

6 years ago[X86][MMX] Accept UNDEF upper bits for MOVD GR32->MMX
Simon Pilgrim [Tue, 16 Jan 2018 17:01:31 +0000 (17:01 +0000)]
[X86][MMX] Accept UNDEF upper bits for MOVD GR32->MMX

llvm-svn: 322574

6 years agoAdd va_start()/va_copy()/va_end to Builtins.def
Nico Weber [Tue, 16 Jan 2018 16:55:41 +0000 (16:55 +0000)]
Add va_start()/va_copy()/va_end to Builtins.def

That way, clang suggests including stdarg.h when these are used in C files.
https://reviews.llvm.org/D42085

llvm-svn: 322573

6 years agoSet dso_local in lld.
Rafael Espindola [Tue, 16 Jan 2018 16:49:05 +0000 (16:49 +0000)]
Set dso_local in lld.

We were already doing this in gold, but not in lld.

llvm-svn: 322572

6 years ago[OPENMP] Add support for `depend` clause on `target teams distribute`.
Alexey Bataev [Tue, 16 Jan 2018 16:46:46 +0000 (16:46 +0000)]
[OPENMP] Add support for `depend` clause on `target teams distribute`.

Added codegen for `depend` clauses on `#pragma omp target teams
distribute` directives.

llvm-svn: 322571

6 years ago[OPENMP] Add support for `depend` clauses on `target parallel` directive.
Alexey Bataev [Tue, 16 Jan 2018 16:27:49 +0000 (16:27 +0000)]
[OPENMP] Add support for `depend` clauses on `target parallel` directive.

Added codegen for `depend` clauses on `#pragma omp target parallel`
directives.

llvm-svn: 322570

6 years ago[OPENMP] Add support for `depend` clauses on `target teams`.
Alexey Bataev [Tue, 16 Jan 2018 15:57:07 +0000 (15:57 +0000)]
[OPENMP] Add support for `depend` clauses on `target teams`.

Added codegen for `depend` clause on `#pragma omp target teams`
directives.

llvm-svn: 322569

6 years ago[LiveDebugValues] update kill-after-spill test with target triple
Petar Jovanovic [Tue, 16 Jan 2018 15:57:03 +0000 (15:57 +0000)]
[LiveDebugValues] update kill-after-spill test with target triple

Set target triple to "x86_64-unknown-linux-gnu".

llvm-svn: 322568

6 years agoMore constexpr algorithms from P0202. search/search_n
Marshall Clow [Tue, 16 Jan 2018 15:48:27 +0000 (15:48 +0000)]
More constexpr algorithms from P0202. search/search_n

llvm-svn: 322566

6 years ago[SystemZ] Support vector registers with inline asm
Ulrich Weigand [Tue, 16 Jan 2018 15:39:23 +0000 (15:39 +0000)]
[SystemZ] Support vector registers with inline asm

Allow using vector register names and the "v" constraint
in inline asm to ensure compatibility with GCC.

llvm-svn: 322562

6 years ago[OPENMP] Add support for `depend` clauses on `target simd`.
Alexey Bataev [Tue, 16 Jan 2018 15:05:16 +0000 (15:05 +0000)]
[OPENMP] Add support for `depend` clauses on `target simd`.

Added codegen for `depend` clauses on `#pragma omp target simd`
directives.

llvm-svn: 322559

6 years agoChange an internal table of constants for the poisson distribution from
Marshall Clow [Tue, 16 Jan 2018 14:54:36 +0000 (14:54 +0000)]
Change an internal table of constants for the poisson distribution from
type 'result_type' to 'double'. The only thing that we ever do with
these numbers is to promote them to 'double' and use them in a division.
For small result_types, the values were getting truncated, skewing the
results. Thanks to James Nagurne for the suggestion.

llvm-svn: 322556

6 years ago[LiveDebugValues] recognize spilled reg killed in instruction after spill
Petar Jovanovic [Tue, 16 Jan 2018 14:46:05 +0000 (14:46 +0000)]
[LiveDebugValues] recognize spilled reg killed in instruction after spill

Current condition for spill instruction recognition in LiveDebugValues does
not recognize case when register is spilled and killed in next instruction.

Patch by Nikola Prica.

Differential Revision: https://reviews.llvm.org/D41226

llvm-svn: 322554

6 years ago[X86][MMX] Improve MMX constant generation
Simon Pilgrim [Tue, 16 Jan 2018 14:21:28 +0000 (14:21 +0000)]
[X86][MMX] Improve MMX constant generation

Extend the MMX zero code to take any constant with zero'd upper 32-bits

llvm-svn: 322553

6 years ago[NFC] fix trivial typos in documents
Hiroshi Inoue [Tue, 16 Jan 2018 13:19:48 +0000 (13:19 +0000)]
[NFC] fix trivial typos in documents

"the the" -> "the"

llvm-svn: 322552

6 years ago[NFC] fix trivial typo in document
Hiroshi Inoue [Tue, 16 Jan 2018 13:19:31 +0000 (13:19 +0000)]
[NFC] fix trivial typo in document

"the the" -> "the"

llvm-svn: 322551

6 years agoSquash -Wcovered-switch-default wairning
Sam McCall [Tue, 16 Jan 2018 12:54:28 +0000 (12:54 +0000)]
Squash -Wcovered-switch-default wairning

llvm-svn: 322549

6 years agoEnsure code complete with !LoadExternal sees all local decls.
Sam McCall [Tue, 16 Jan 2018 12:33:46 +0000 (12:33 +0000)]
Ensure code complete with !LoadExternal sees all local decls.

Summary:
noload_lookups() was too lazy: in addition to avoiding external decls, it
avoided populating the lazy lookup structure for internal decls.
This is the right behavior for the existing callsite in ASTDumper, but I think
it's not a very useful default, so we populate it by default.

While here:
 - remove an unused test file accidentally added in r322371.
 - remove lookups_begin()/lookups_end() in favor of lookups().begin(), which is
   more common and more efficient.

Reviewers: ilya-biryukov

Subscribers: cfe-commits, rsmith

Differential Revision: https://reviews.llvm.org/D42077

llvm-svn: 322548

6 years ago[clangd] Avoid combinatorial explosion in CodeCompleteTests.
Sam McCall [Tue, 16 Jan 2018 12:21:24 +0000 (12:21 +0000)]
[clangd] Avoid combinatorial explosion in CodeCompleteTests.

Summary:
This test dominates our unit test runtime, and the change speeds it up by 10x.
We lose coverage of some combinations of flags, but I'm not sure that's finding
many bugs.

3300 -> 300ms on my machine (3800 -> 800ms for the whole of CompletionTest).

Reviewers: ilya-biryukov

Subscribers: klimek, cfe-commits

Differential Revision: https://reviews.llvm.org/D42063

llvm-svn: 322547

6 years agoAdd missing CINDEX_LINKAGE
Ivan Donchevskii [Tue, 16 Jan 2018 12:11:59 +0000 (12:11 +0000)]
Add missing CINDEX_LINKAGE

Follow up for [libclang] Add PrintingPolicy for pretty printing declarations

Differential Revision: https://reviews.llvm.org/D39903

llvm-svn: 322546

6 years ago[X86][I86,I186,I286,I386,I486,PPRO, MMX]: Adding full coverage of MC encoding for...
Gadi Haber [Tue, 16 Jan 2018 11:33:45 +0000 (11:33 +0000)]
[X86][I86,I186,I286,I386,I486,PPRO, MMX]: Adding full coverage of MC encoding for the I86, I186, I286, I386, I486, PPRO and MMX isa sets.<NFC>

NFC.
 Adding MC regressions tests to cover the I86, I186, I286, I386, I486, PPRO and MMX isa sets.
 This patch is part of a larger task to cover MC encoding of all X86 ISA Sets.
 Started in revision: https://reviews.llvm.org/D39952

Reviewers: zvi, RKSimon, AndreiGrischenko, craig.topper
Differential Revision: https://reviews.llvm.org/D40879

Change-Id: I231a35861611bfd3d23c74cc59507373f021a629
llvm-svn: 322544

6 years ago[DebugInfo] Unify dumping of address ranges
Jonas Devlieghere [Tue, 16 Jan 2018 11:17:57 +0000 (11:17 +0000)]
[DebugInfo] Unify dumping of address ranges

Summary:
This patch unifies the printing of address ranges as [0x0, 0x1).

rdar://34822059

Reviewers: aprantl, dblaikie

Subscribers: mehdi_amini, llvm-commits

Differential Revision: https://reviews.llvm.org/D42056

llvm-svn: 322543

6 years ago[CodeGen] Remove special case of printing subRegIdx from MachineInstr::print
Francis Visoiu Mistrih [Tue, 16 Jan 2018 10:53:14 +0000 (10:53 +0000)]
[CodeGen] Remove special case of printing subRegIdx from MachineInstr::print

Support in MachineOperand has been added in r320209. No need to special
case this anymore.

llvm-svn: 322542

6 years ago[CodeGen][NFC] Correct case for printSubRegIdx
Francis Visoiu Mistrih [Tue, 16 Jan 2018 10:53:11 +0000 (10:53 +0000)]
[CodeGen][NFC] Correct case for printSubRegIdx

llvm-svn: 322541

6 years ago[libclang] Add PrintingPolicy for pretty printing declarations
Jonathan Coe [Tue, 16 Jan 2018 10:19:56 +0000 (10:19 +0000)]
[libclang] Add PrintingPolicy for pretty printing declarations

Summary:
Introduce clang_getCursorPrettyPrinted() for pretty printing
declarations. Expose also PrintingPolicy, so the user gets more
fine-grained control of the entities being printed.

The already existing clang_getCursorDisplayName() is pretty limited -
for example, it does not handle return types, parameter names or default
arguments for function declarations. Addressing these issues in
clang_getCursorDisplayName() would mean to duplicate existing code
(e.g. clang::DeclPrinter), so rather expose new API to access the
existing functionality.

Reviewed By: jbcoe

Subscribers: cfe-commits

Tags: #clang

Patch by nik (Nikolai Kosjar)

Differential Revision: https://reviews.llvm.org/D39903

llvm-svn: 322540

6 years agoAdd a value_type to ArrayRef.
Clement Courbet [Tue, 16 Jan 2018 09:11:20 +0000 (09:11 +0000)]
Add a value_type to ArrayRef.

Summary: Not sure this needs a review or not. Erring on the safe side.

Reviewers: dblaikie

Differential Revision: https://reviews.llvm.org/D41666

llvm-svn: 322538

6 years ago[X86][XSAVE]: Adding full coverage of MC encoding for the XSAVE isa sets.<NFC>
Gadi Haber [Tue, 16 Jan 2018 08:50:29 +0000 (08:50 +0000)]
[X86][XSAVE]: Adding full coverage of MC encoding for the XSAVE isa sets.<NFC>

NFC.
 Adding MC regressions tests to cover the XSAVE ISA sets.
 This patch is part of a larger task to cover MC encoding of all X86 ISA Sets started in revision: https://reviews.llvm.org/D39952

Reviewers: zvi, RKSimon, AndreiGrischenko, craig.topper
Differential Revision: https://reviews.llvm.org/D41282

Change-Id: I325bf8f421f78c80179a04fc39033366759cbe45
llvm-svn: 322537

6 years ago[FileCheck] - Fix possible buffer out of bounds access when parsing --check-prefix.
George Rimar [Tue, 16 Jan 2018 08:09:24 +0000 (08:09 +0000)]
[FileCheck] - Fix possible buffer out of bounds access when parsing --check-prefix.

FileCheck tool crashes when trying to parse --check-prefix argument if there is no any
data after it.

For example test like following would crash if there are no symbols and no EOL mark after `boom`:

# REQUIRES: x86
# RUN: <skipped few lines>
# RUN: llvm-readobj -t %t | FileCheck %s --check-prefix=boom

Patch fixes the issue.

Differential revision: https://reviews.llvm.org/D42057

llvm-svn: 322536

6 years ago[BPF] Mark pseudo insn patterns as isCodeGenOnly
Yonghong Song [Tue, 16 Jan 2018 07:27:20 +0000 (07:27 +0000)]
[BPF] Mark pseudo insn patterns as isCodeGenOnly

These pseudos are not supposed to be visible to user.

This patch reduced the auto-generated instruction matcher. For example,
the following words are removed from keyword list of LLVM BPF assembler.

-  MCK__35_, // '#'
-  MCK__COLON_, // ':'
-  MCK__63_, // '?'
-  MCK_ADJCALLSTACKDOWN, // 'ADJCALLSTACKDOWN'
-  MCK_ADJCALLSTACKUP, // 'ADJCALLSTACKUP'
-  MCK_PSEUDO, // 'PSEUDO'
-  MCK_Select, // 'Select'

Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Jiong Wang <jiong.wang@netronome.com>
llvm-svn: 322535

6 years ago[BPF] Teach DAG2DAG AND elimination about load intrinsics
Yonghong Song [Tue, 16 Jan 2018 07:27:19 +0000 (07:27 +0000)]
[BPF] Teach DAG2DAG AND elimination about load intrinsics

As commented on the existing code:

  // The Reg operand should be a virtual register, which is defined
  // outside the current basic block. DAG combiner has done a pretty
  // good job in removing truncating inside a single basic block.

However, when the Reg operand comes from bpf_load_[byte | half | word]
intrinsics, the generic optimizer doesn't understand their results are
zero extended, so these single basic block elimination opportunities were
missed.

Acked-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Jiong Wang <jiong.wang@netronome.com>
llvm-svn: 322534

6 years ago[SROA] fix assetion failure
Hiroshi Inoue [Tue, 16 Jan 2018 06:23:05 +0000 (06:23 +0000)]
[SROA] fix assetion failure

This patch fixes the assertion failure in SROA reported in PR35657.
PR35657 reports the assertion failure due to r319522 (splitting for non-whole-alloca slices), but this problem can happen even without r319522.

The problem exists in a check for reusing an existing alloca when rewriting partitions. As the original comment said, we can reuse the existing alloca if the new alloca has the same type and offset with the existing one. But the code checks only type of the alloca and then check the offset using an assert.
In a corner case with out-of-bounds access (e.g. @PR35657 function added in unit test), it is possible that the two allocas have the same type but different offsets.

This patch makes the check of the offset in the if condition, and re-enables the splitting for non-whole-alloca slices.

Differential Revision: https://reviews.llvm.org/D41981

llvm-svn: 322533

6 years ago[X86] Revisit the fix I made years ago to make 'xchgl %eax, %eax' not encode using...
Craig Topper [Tue, 16 Jan 2018 06:07:16 +0000 (06:07 +0000)]
[X86] Revisit the fix I made years ago to make 'xchgl %eax, %eax' not encode using the 0x90 encoding in 64-bit mode.

Prior to this we had a separate instruction and register class that excluded eax to prevent matching the instruction that would encode with 0x90.

This patch changes this to just use an InstAlias to force xchgl %eax, %eax to use XCHG32rr instruction in 64-bit mode. This gets rid of the separate instruction and register class.

llvm-svn: 322532

6 years ago[X86] Make 'xchgq %rax, %rax' an alias for the 0x90 nop encoding to match gas.
Craig Topper [Tue, 16 Jan 2018 06:07:14 +0000 (06:07 +0000)]
[X86] Make 'xchgq %rax, %rax' an alias for the 0x90 nop encoding to match gas.

Previously we encoded it as 0x48 0x90.

llvm-svn: 322531

6 years ago[Sema] Fix a crash on invalid features in multiversioning
George Burgess IV [Tue, 16 Jan 2018 03:01:50 +0000 (03:01 +0000)]
[Sema] Fix a crash on invalid features in multiversioning

We were trying to emit a diag::err_bad_multiversion_option diagnostic,
which expects an int as its first argument, with a string argument. As
it happens, the string `Feature` that was causing this was shadowing an
int `Feature` from the surrounding scope. :)

llvm-svn: 322530

6 years agoMore constexpr algorithms from P0202: lower_bound, upper_bound, equal_range, binary_s...
Marshall Clow [Tue, 16 Jan 2018 02:34:41 +0000 (02:34 +0000)]
More constexpr algorithms from P0202: lower_bound, upper_bound, equal_range, binary_search

llvm-svn: 322529

6 years agoActually CALL the constexpr tests.
Marshall Clow [Tue, 16 Jan 2018 02:11:13 +0000 (02:11 +0000)]
Actually CALL the constexpr tests.

llvm-svn: 322528

6 years agoMore constexpr (re P0202) - equal and mismatch
Marshall Clow [Tue, 16 Jan 2018 02:04:10 +0000 (02:04 +0000)]
More constexpr (re P0202) - equal and mismatch

llvm-svn: 322527

6 years agoAvoid Wparentheses warning.
Simon Pilgrim [Mon, 15 Jan 2018 22:40:06 +0000 (22:40 +0000)]
Avoid Wparentheses warning.

llvm-svn: 322526

6 years ago[X86][MMX] Add support for MMX zero vector creation
Simon Pilgrim [Mon, 15 Jan 2018 22:32:40 +0000 (22:32 +0000)]
[X86][MMX] Add support for MMX zero vector creation

As mentioned on PR35869, (and came up recently on D41517) we don't create a MMX zero register via the PXOR but instead perform a spill to stack from a XMM zero register.

This patch adds support for direct MMX zero vector creation and should make it easier to add better constant vector creation in the future as well.

Differential Revision: https://reviews.llvm.org/D41908

llvm-svn: 322525

6 years ago[X86][SSE] Add custom execution domain fixing for BLENDPD/BLENDPS/PBLENDD/PBLENDW...
Simon Pilgrim [Mon, 15 Jan 2018 22:18:45 +0000 (22:18 +0000)]
[X86][SSE] Add custom execution domain fixing for BLENDPD/BLENDPS/PBLENDD/PBLENDW (PR34873)

Add support for custom execution domain fixing and implement support for BLENDPD/BLENDPS/PBLENDD/PBLENDW.

Differential Revision: https://reviews.llvm.org/D42042

llvm-svn: 322524

6 years ago[x86] add tests to show missed constant shrinking (PR35907); NFC
Sanjay Patel [Mon, 15 Jan 2018 21:57:41 +0000 (21:57 +0000)]
[x86] add tests to show missed constant shrinking (PR35907); NFC

llvm-svn: 322523

6 years ago[x86] regenerate test checks; NFC
Sanjay Patel [Mon, 15 Jan 2018 21:32:39 +0000 (21:32 +0000)]
[x86] regenerate test checks; NFC

llvm-svn: 322522

6 years ago[x86] regenerate test checks; NFC
Sanjay Patel [Mon, 15 Jan 2018 21:28:52 +0000 (21:28 +0000)]
[x86] regenerate test checks; NFC

llvm-svn: 322521

6 years ago[docs] Only LLVM IR bitstreams begin with 'BC'
Brian Gesiak [Mon, 15 Jan 2018 21:23:32 +0000 (21:23 +0000)]
[docs] Only LLVM IR bitstreams begin with 'BC'

Summary:
The LLVM Bitcode File Format documentation states that all bitstreams
begin with the magic number 'BC', and that generic bitstream analyzer
tools may check for this number in order to determine whether the
stream is a bitstream.

However, in practice:

* Only LLVM IR bitcode begins with 'BC'. Other bitstreams -- Clang
  AST files and precompiled headers, Clang serialized diagnostics,
  Swift modules -- do not start with 'BC'. A tool that actually checked
  for 'BC' would only be able to recognize LLVM IR.
* The `llvm-bcanalyzer`, arguably the most used generic bitstream
  analyzer tool, does not check for a magic number 'BC' (except to
  determine whether the file is LLVM IR).

Update the bitcode format documentation to make it clear that not all
bitstreams begin with 'BC', and that tools should not rely on that
particular magic number value.

Test Plan:
Build the `docs-llvm-html` target and confirm the changes render in
a Safari web browser.

Reviewers: harlanhaskins, eugenis, mehdi_amini, pcc, angerman

Reviewed By: angerman

Subscribers: angerman, llvm-commits

Differential Revision: https://reviews.llvm.org/D42002

llvm-svn: 322520

6 years ago[x86] regenerate test checks; NFC
Sanjay Patel [Mon, 15 Jan 2018 21:22:46 +0000 (21:22 +0000)]
[x86] regenerate test checks; NFC

llvm-svn: 322519

6 years agoRevert 319303: Add _Float128 as alias to __float128 to enable compilations on Fedora2...
Erich Keane [Mon, 15 Jan 2018 21:16:25 +0000 (21:16 +0000)]
Revert 319303: Add _Float128 as alias to __float128 to enable compilations on Fedora27/glibc2

Differential Revision: https://reviews.llvm.org/D40673

llvm-svn: 322518

6 years ago[Driver] Suggest valid integrated tools
Brian Gesiak [Mon, 15 Jan 2018 21:05:40 +0000 (21:05 +0000)]
[Driver] Suggest valid integrated tools

Summary:
There are only two valid integrated Clang driver tools: `-cc1` and
`-cc1as`. If a user asks for an unknown tool, such as `-cc1asphalt`,
an error message is displayed to indicate that there is no such tool,
but the message doesn't indicate what the valid options are.

Include the valid options in the error message.

Test Plan: `check-clang`

Reviewers: sepavloff, bkramer, phosek

Reviewed By: bkramer

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D42004

llvm-svn: 322517

6 years ago[OPENMP] Update status of OpenMP support, NFC.
Alexey Bataev [Mon, 15 Jan 2018 21:01:29 +0000 (21:01 +0000)]
[OPENMP] Update status of OpenMP support, NFC.

llvm-svn: 322516

6 years ago[OPENMP] Initial codegen for `target teams distribute parallel for
Alexey Bataev [Mon, 15 Jan 2018 20:59:40 +0000 (20:59 +0000)]
[OPENMP] Initial codegen for `target teams distribute parallel for
simd`.

Added host codegen + codegen for devices with default codegen for
`#pragma omp target teams distribute parallel for simd` directive.

llvm-svn: 322515