llvm-svn: 212819

[CMake] add_llvm_library: Add "RUNTIME DESTINATION bin" to install(). It affects add_library(SHARED) for Win32.DLL.

llvm-svn: 212818

[PATCH] [x32] Introduce -mx32 flag enforcing x32 mode for other triplets. By Pavel Chupin (@pavel.v.chupin).

-mx32 flag setup target environment to GNUX32 and can be used for 
other 32/64-bit triplets (i386-unknown-linux, x86_64-unknown-linux) to
turn on x32 mode. Compatible with GCC -mx32 flag.

Differential Revision: http://reviews.llvm.org/D4470

llvm-svn: 212817

Including the scratch buffer size in the initial reservation eliminates the
subsequent malloc+move operation and offers a healthier constant growth with
less memory wastage.

When doing this, take care to avoid invalidating the source buffer.

llvm-svn: 212816

This fixes '___asan_init_v4 already defined' errors when linking some of Chromium DLLs.
Looks like one of the DLL is using a .lib produced while linking another DLL and it exploded after r212699.
I'm trying to come up with a small testcase...

llvm-svn: 212815

Reviewers: djasper, sbenza, bkramer

Reviewed By: bkramer

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D4463

llvm-svn: 212814

llvm-svn: 212813

ACLE 2.0 allows __fp16 to be used as a function argument or return
type. This enables this for AArch64.

llvm-svn: 212812

llvm-svn: 212811

Add FileCheck -implicit-check-not option to allow stricter tests without adding too many CHECK-NOTs manually.

Summary:
Add FileCheck -implicit-check-not option which allows specifying a
pattern that should only occur in the input when explicitly matched by a
positive check. This feature allows checking tool diagnostics in a way
clang -verify does it for compiler diagnostics.

The option has been tested on a number of clang-tidy checks, I'll post a link to
the clang-tidy patch to this thread.

Once there's an agreement on the general direction, I can add tests and
documentation.

Reviewers: djasper, bkramer

Reviewed By: bkramer

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D4462

llvm-svn: 212810

llvm-svn: 212809

Also add a few comments.

<rdar://problem/17581756>

llvm-svn: 212808

Reviewed at http://reviews.llvm.org/D4471

llvm-svn: 212807

llvm-svn: 212806

Returns a warning when using an unknown optimization flag.
This patch includes -finline-limit as one of those ignored flags.
More options will be moved in this group

Patch by Arthur Marble <arthur@info9.net> in the context of
Debian Google Summer of code 2014.

Reviewers: rnk, Sylvestre
llvm-svn: 212805

llvm-svn: 212804

llvm-svn: 212803

This commit fixes bug http://llvm.org/bugs/show_bug.cgi?id=20103.

Thanks to Qwertyuiop for the report and the proposed fix.

llvm-svn: 212802

llvm-svn: 212801

They where just left over from copy-pasting.

Reported-by: Johannes Doerfert <jdoerfert@codeaurora.org>
llvm-svn: 212800

Our versions are not exactly as fast as libc's, and
MSan uses them heavily (even compared to other sanitizers).

This will break if libc version of mem* are instrumented,
but they never are, and if they are, we should be able
to fix it on libc side.

llvm-svn: 212799

llvm-svn: 212798

The goal is to find code like the example below, which is likely a typo
where someone meant to write "if (*b)".
bool *b = SomeFunction();
 if (b) {
   // b never dereferenced
 }

This checker naturally has a relatively high false positive rate so it
applies some heuristics to avoid cases where the pointer is checked for
nullptr before being written.

Differential Revision: http://reviews.llvm.org/D4458

llvm-svn: 212797

Thanks to Johannes Doerfert for narrowing down the bug.

Reported-by: Chris Jenneisch <chrisj@codeaurora.org>
llvm-svn: 212796

No functional change.  As I was trying to understand this function, I found
that variables were reused with confusing names and the broadcast case was a
bit too implicit.  Hopefully, this is an improvement.

llvm-svn: 212795

It was computing the VL/n case as:
  MemObjSize = VectorByteSize / ElemByteSize / Divider * ElemByteSize

ElemByteSize not only falls out but VectorByteSize/Divider now actually
matches the definition of VL/n.

Also some formatting fixes.

llvm-svn: 212794

Revert "Reapply "DebugInfo: Ensure that all debug location scope chains from instructions within a function, lead to the function itself.""

This reverts commit r212776.

Nope, still seems to be failing on the sanitizer bots... but hey, not
the msan self-host anymore, it's failing in asan now. I'll start looking
there next.

llvm-svn: 212793

llvm-svn: 212792

llvm-svn: 212791

llvm-svn: 212790

llvm-svn: 212789

into their container; we won't find them there. These things are already being
merged when they're added to their primary template's folding set, so this
merging is redundant (and causes us to reject-valid because we think we've
found an odr violation).

llvm-svn: 212788

llvm-svn: 212787

Addressing review comments from r212784.

llvm-svn: 212786

This patch fixes a number of issues with embedded Python on
Windows.  In particular:

1) The script that builds the python modules was normalizing the
   case of python filenames during copies.  The module name is
   the filename, and is case-sensitive, so this was breaking code.

2) Changes the build to not attempt to link against python27.lib
   (e.g. the release library) when linking against msvcrt debug
   library.  Doing a debug build of LLDB with embedded python
   support now requires you to provide your own self-compiled
   debug version of python.

3) Don't import termios when initializing the interpreter.  This
   is part of a larger effort to remove the dependency on termios
   since it is not available on Windows.  This particular instance
   was unnecessary and unused.

Reviewed by: Todd Fiala

Differential Revision: http://reviews.llvm.org/D4441

llvm-svn: 212785

The relevant portion of C++ standard says [namespace.memdef]p3:

  If the name in a friend declaration is neither qualified nor a
  template-id and the declaration is a function or an
  elaborated-type-specifier, the lookup to determine whether the entity
  has been previously declared shall not consider any scopes outside the
  innermost enclosing namespace.

MSVC does not implement that rule for types.  If there is a type in an
enclosing namespace, they consider an unqualified tag declaration with
the same name to be a redeclaration of the type from another namespace.

Implementing compatibility is a simple matter of disabling our
implementation of this rule for types, which was added in r177473.

Reviewers: rsmith

Differential Revision: http://reviews.llvm.org/D4443

llvm-svn: 212784

Fixes include:
- Don't say that "<arch>-apple-ios" is compatible with "<arch>-apple-macosx"
- Fixed DynamicLoaderMacOSXDYLD so specify an architecture that was converted solely from a cputype and subtype, just specify the file + UUID.
- Fixed PlatformiOSSimulator::GetSupportedArchitectureAtIndex() so it returns the correct archs
- Fixed SymbolFileDWARFDebugMap to load .o files correctly by just specifying the architecture without the vendor and OS now that "<arch>-apple-ios" is not compatible with "<arch>-apple-macosx" so we can load .o files correctly for DWARF with debug map
- Fixed the coded in TargetList::CreateTarget() so it does the right thing with an underspecified triple where just the arch is specified.

llvm-svn: 212783

Partially fix PR20058: reduce compile time for loop unrolling with very high count by reducing calls to SE->forgetLoop

llvm-svn: 212782

This patch flips the default value for -gcolumn-info to be on by
default. I discussed the rationale and provided compile/size data
in:

http://lists.cs.uiuc.edu/pipermail/llvmdev/2014-June/074290.html

This also updates the documentation and some tests that relied on
the lack of column information. Some tests had column information
in the expected output, but it was wrong (the tsan tests). Others
were using the driver to execute.

llvm-svn: 212781

RuntimeDyldChecker.

This allows us to remove one of the six remaining object files in the LLVM
source tree.

llvm-svn: 212780