This function adds an extra path argument to lto_module_create_from_memory.
The path argument will be passed to makeBuffer to make sure the MemoryBuffer
has a name and the created module has a module identifier.

This is mainly for emitting warning messages from the linker. When we emit
warning message on a module, we can use the module identifier.

rdar://15985737

llvm-svn: 201114

A const ObjectFile needs to be able to provide its name. For an IRObjectFile,
that means being able to call the mangler. Since each IRObjectFile can have
a different mangling, it is natural for them to contain a Mangler which is
therefore also const.

llvm-svn: 201113

Replaced cast and vreinterepret operations with
code to reinterpret bitwise the types float16_t and
int16_t.

llvm-svn: 201112

Patch by Brad King!

llvm-svn: 201111

llvm-svn: 201110

llvm-svn: 201109

llvm-svn: 201108

Truncation is just accessing a subregister for any multiple of
the register size, so it's free.

llvm-svn: 201107

llvm-svn: 201106

Introduce a notion of a 'current representation method' for
pointers-to-members.

When starting out, this is set to 'best case' (representation method is
chosen by examining the class, selecting the smallest representation
that would work given the class definition or lack thereof).

This pragma allows the translation unit to dictate exactly what
representation to use, similar to how the inheritance model keywords
operate.

N.B.  PCH support is forthcoming.

Differential Revision: http://llvm-reviews.chandlerc.com/D2723

llvm-svn: 201105

The crux of the issue is that LCSSA doesn't preserve stateful alias
analyses. Before r200067, LICM didn't cause LCSSA to run in the LTO pass
manager, where LICM runs essentially without any of the other loop
passes. As a consequence the globalmodref-aa pass run before that loop
pass manager was able to survive the loop pass manager and be used by
DSE to eliminate stores in the function called from the loop body in
Adobe-C++/loop_unroll (and similar patterns in other benchmarks).

When LICM was taught to preserve LCSSA it had to require it as well.
This caused it to be run in the loop pass manager and because it did not
preserve AA, the stateful AA was lost. Most of LLVM's AA isn't stateful
and so this didn't manifest in most cases. Also, in most cases LCSSA was
already running, and so there was no interesting change.

The real kicker is that LCSSA by its definition (injecting PHI nodes
only) trivially preserves AA! All we need to do is mark it, and then
everything goes back to working as intended. It probably was blocking
some other weird cases of stateful AA but the only one I have is
a 1000-line IR test case from loop_unroll, so I don't really have a good
test case here.

Hopefully this fixes the regressions on performance that have been seen
since that revision.

llvm-svn: 201104

Made sure we pass along the file action paths for stdin/stdout/stderr to the XPC service.
[Reviewed by Greg Clayton]

llvm-svn: 201103

Create a new diagnostic, -Wignored-pragmas and use it to handle any
case where a pragma would have a side effect but is ignored.

llvm-svn: 201102

Fix for PR17606 - result_of (and INVOKE) works incorrectly for member function pointers with ref qualifiers. Also a drive-by fix for common_type in C++03 mode. Thanks to Michel Morin for the bug report and the proposed fix.

llvm-svn: 201101

llvm-svn: 201100

No functional change, this code was just superfluous.

llvm-svn: 201099

This fixes the oversight from r159077.

llvm-svn: 201098

DS instructions that access local memory can only uses addresses that
are less than or equal to the value of M0.  When M0 is uninitialized,
then we experience undefined behavior.

This patch also changes the behavior to emit S_WQM_B64 on pixel shaders
no matter what kind of DS instruction is used.

llvm-svn: 201097

This doesn't change any functionality, since we only have two shader
types (compute and pixel) that use local memory.  We're just changing
the logic to match the documentation.

llvm-svn: 201096

Thanks to Chandler for the catch.

llvm-svn: 201095

Now that both ARM backends use the same implementation for vshll operations,
the code can be shared. This is also a necessary LLVM/Clang interface update.

llvm-svn: 201094

Similarly to the vshrn instructions, these are simple zext/sext + trunc
operations. Using normal LLVM IR should allow for better code, and more sharing
with the AArch64 backend.

llvm-svn: 201093

Although the interface to el_push should be a constant parameter (as it is on
Darwin), certain Linux distributions currently ship a header which does not
provide proper const correctness.  This causes compilation failures on Linux.
Strip the constness on the parameter, which whilst incorrect, is mostly
harmless.  The parameter will not be changed by the interface and so it is
acceptable to do this.  When distributions have updated to a more correct
declaration, it would be nice to revert this change.

Addresses PR18784.

llvm-svn: 201092

llvm-svn: 201091

For A- and R-class processors, r12 is not normally callee-saved, but is for
interrupt handlers. See AAPCS, 5.3.1.1, "Use of IP by the linker".

llvm-svn: 201089

llvm-svn: 201088

llvm-svn: 201087

Now the backend supports the natural LLVM IR, we can shamelessly steal the
AArch64 front-end code to implement the vshrn intrinsic on 32-bit ARM.

llvm-svn: 201086

vshrn is just the combination of a right shift and a truncate (and the limits
on the immediate value actually mean the signedness of the shift doesn't
matter). Using that representation allows us to get rid of an ARM-specific
intrinsic, share more code with AArch64 and hopefully get better code out of
the mid-end optimisers.

llvm-svn: 201085

This change adds a copy of <ucontext.h> for Android found in google-breakpad
that is missing from the official NDK.
ASan SEGV handler is still disabled by default and can be enabled with
ASAN_OPTIONS=handle_segv.

llvm-svn: 201084

llvm-svn: 201083

llvm-svn: 201082

llvm-svn: 201081

 the addition of Mips64 tests.

No functional changes.

llvm-svn: 201080

llvm-svn: 201079

This way it's possible to share the instruction's description for LSA and
DLSA (to be added).

No functional changes.

llvm-svn: 201078

llvm-svn: 201077

The CMake install(DIRECTORY) command documents that it sets permissions
on directories it is asked to install.  Since the <prefix>/include
directory may not be exclusive to the LLVM/Clang installation, we should
not ask CMake to manage permissions of that directory for us.  Instead,
give only our own include/clang and include/clang-c subdirectories to
the install(DIRECTORY) command.

Fixes PR4500. Patch by Brad King.

llvm-svn: 201076

The CMake install(DIRECTORY) command documents that it sets permissions
on directories it is asked to install.  Since the <prefix>/include
directory may not be exclusive to the LLVM installation, we should not
ask CMake to manage permissions of that directory for us.  Instead, give
only our own include/llvm and include/llvm-c subdirectories to the
install(DIRECTORY) command.

Fixes PR4500. Patch by Brad King.

llvm-svn: 201075

llvm-svn: 201074