This reverts commit r208912.

It was committed accidentally without review.

llvm-svn: 208914

We have to iterate over all the calls that were inlined to find out if
any were musttail.

Sink another variable down to where its used.

llvm-svn: 208913

The allocas going out of scope are immediately killed by the return
instruction.

Reviewers: chandlerc

Differential Revision: http://reviews.llvm.org/D3630

llvm-svn: 208912

llvm-svn: 208911

The interesting case is what happens when you inline a musttail call
through a musttail call site.  In this case, we can't break perfect
forwarding or allow any stack growth.

Instead of merging control flow from the inlined return instruction
after a musttail call into the body of the caller, leave the inlined
return instruction in the caller so that the musttail call stays in the
tail position.

More work is required in http://reviews.llvm.org/D3630 to handle the
case where the inlined function has dynamic allocas or byval arguments.

Reviewers: chandlerc

Differential Revision: http://reviews.llvm.org/D3491

llvm-svn: 208910

llvm-svn: 208909

The symlink needs to point to a relative path, so we don't break
building in a chroot.

Tested-by: Laurent Carlier <lordheavym@gmail.org>
llvm-svn: 208908

llvm-svn: 208907

types to hold relocation's offset and addend.

llvm-svn: 208906

llvm-svn: 208905

Revert the current implementation and C API. New implementation and C APIs are
in the works.

llvm-svn: 208904

how CMake drives a windows link and how our custom command does.

llvm-svn: 208903

llvm-svn: 208902

When using the VFS, we want the virtual header location when searching
for a framework module, since that will be the one in the correct
directory structure for the module.

I'll add a regression test once I finish reducing the larger one I have.

llvm-svn: 208901

llvm-svn: 208900

llvm-svn: 208899

llvm-svn: 208898

llvm-svn: 208897

for sanitizers to pass the C++ compilation and exe linking flags through
from the host CMake configuration. We pass the target flags afterward,
allowing them to trump flags as needed. This is particularly important
when the flags direct Clang, even the just-built-Clang, toward the
standard library, linker, and other tools to use.

llvm-svn: 208896

Added target specific combine rules to fold blend intrinsics according
to the following rules:
 1) fold(blend A, A, Mask) -> A;
 2) fold(blend A, B, <allZeros>) -> A;
 3) fold(blend A, B, <allOnes>) -> B.

Added two new tests to verify that the new folding rules work for all
the optimized blend intrinsics.

llvm-svn: 208895

Differential Revision: http://reviews.llvm.org/D3712

llvm-svn: 208894

llvm-svn: 208893

Differential Revision: http://reviews.llvm.org/D3711

llvm-svn: 208892

Differential Revision: http://reviews.llvm.org/D3710

llvm-svn: 208891

Differential Revision: http://reviews.llvm.org/D3709

llvm-svn: 208890

llvm-svn: 208889

We now use SReg_* for integer types and VReg_* for floating-point types.
This should help simplify the SIFixSGPRCopies pass and no longer causes
ISel to insert a COPY after termiator instuctions that output a value.

This change is covered by exisitng tests.

llvm-svn: 208888

This prevents a future commit from regressing the load-i1.ll test.

llvm-svn: 208887

llvm-svn: 208886

llvm-svn: 208885

llvm-svn: 208884

Summary:
Make checks filtering more intuitive and easy to use. Remove
-disable-checks and change the format of -checks= to a comma-separated list of
globs with optional '-' prefix to denote exclusion. The -checks= option is now
cumulative, so it modifies defaults, not overrides them. Each glob adds or
removes to the current set of checks, so the filter can be refined or overriden
by adding globs.

Example:
  The default value for -checks= is
  '*,-clang-analyzer-alpha*,-llvm-include-order,-llvm-namespace-comment,-google-*',
  which allows all checks except for the ones named clang-analyzer-alpha* and
  others specified with the leading '-'. To allow all google-* checks one can
  write:
    clang-tidy -checks=google-* ...
  If one needs only google-* checks, we first need to remove everything (-*):
    clang-tidy -checks=-*,google-*
  etc.

I'm not sure if we need to change something here, so I didn't touch the docs
yet.

Reviewers: klimek, alexfh

Reviewed By: alexfh

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D3770

llvm-svn: 208883

Patch by Adam Strzelecki, thank you!

llvm-svn: 208882

llvm-svn: 208881

Previously, TableGen assumed that every aliased operand consumed precisely 1
MachineInstr slot (this was reasonable because until a couple of days ago,
nothing more complicated was eligible for printing).

This allows a couple more ARM64 aliases to print so we can remove the special
code.

On the X86 side, I've gone for explicit AT&T size specifiers as the default, so
turned off a few of the aliases that would have just started printing.

llvm-svn: 208880

llvm-svn: 208879

The mode is enabled with -DTSAN_NO_HISTORY=1 flag.
Intended mostly for research purposes (how fast can it go w/o history).

llvm-svn: 208878

Summary: Depends on D3728

Reviewers: jkolek, zoran.jovanovic, vmedic

Reviewed By: vmedic

Differential Revision: http://reviews.llvm.org/D3729

llvm-svn: 208877

computeKnownBits, consolidate them into one assert at the end of
computeKnownBits itself.

llvm-svn: 208876

llvm-svn: 208875