of loops.

Previously, two consecutive calls to function "func" would result in the
following sequence of instructions:

1. load $16, %got(func)($gp) // load address of lazy-binding stub.
2. move $25, $16
3. jalr $25                  // jump to lazy-binding stub.
4. nop
5. move $25, $16
6. jalr $25                  // jump to lazy-binding stub again.

With this patch, the second call directly jumps to func's address, bypassing
the lazy-binding resolution routine:

1. load $25, %got(func)($gp) // load address of lazy-binding stub.
2. jalr $25                  // jump to lazy-binding stub.
3. nop
4. load $25, %got(func)($gp) // load resolved address of func.
5. jalr $25                  // directly jump to func.

llvm-svn: 191591

Differential Revision: http://llvm-reviews.chandlerc.com/D1766

llvm-svn: 191590

llvm-svn: 191589

closest ISA.  armv6 wouldn't have the handful of
T32 instructions that the Cortex-M0 ISA specifies.
<rdar://problem/15099306> 

llvm-svn: 191588

It uses the T16 and a few T32 instructions from the ARMv7-A ISA.
<rdar://problem/15099306> 

llvm-svn: 191587

Struct-path aware TBAA generates tags to specify the access path,
while scalar TBAA only generates tags to scalar types.

llvm-svn: 191586

llvm-svn: 191585

if property name is a valid identifier in the underlying
language. // rdar://15044184

llvm-svn: 191584

llvm-svn: 191583

llvm-svn: 191582

Stop using some locally defined mach-o constants.

llvm-svn: 191581

llvm-svn: 191580

llvm-svn: 191579

resolved by lazy-binding.

llvm-svn: 191578

tomorrow is complete.

There is a missing warning due to a serious issue with template
instantiation in Clang (and potentially in the core language).

llvm-svn: 191577

llvm-svn: 191576

llvm-svn: 191575

llvm-svn: 191574

llvm-svn: 191573

uses.

This fixes one of the two remaining failures to implement [[deprecated]]
as specified for C++14.

llvm-svn: 191572

llvm-svn: 191571

[stackprotector] Refactor the StackProtector pass from a single .cpp file into StackProtector.h and StackProtector.cpp.

No functionality change.  Future patches will add analysis which will be used
in other passes (PEI, StackSlot).  The end goal is to support ssp-strong stack
layout rules.

WIP.

Differential Revision: http://llvm-reviews.chandlerc.com/D1521

llvm-svn: 191570

It is mentioned in the LLVM coding standard that _begin() and _end() suffixes
should be used.

llvm-svn: 191569

Fix OS Version reporting bug detected by TestPlatform for some Linux 3.x kernels that do not report the update version
- should resolve the current failure on the Linux clang buildbot

llvm-svn: 191568

This will eventually need to be refactored to better handle COPY relocations,
as other relocations can also generate them. I'm not yet sure the exact
circumstances in which they are needed yet.

llvm-svn: 191567

llvm-svn: 191566

llvm-svn: 191565

llvm-svn: 191564

range that includes a tab character will cause out-of-bounds access to the
fixit string.

llvm-svn: 191563

llvm-svn: 191562

llvm-svn: 191561

llvm-svn: 191560

C++ operators are represented by CXXOperatorName now, which is already
documented.

llvm-svn: 191559

llvm-svn: 191558

llvm-objdump: Dump COFF import table if -private-headers option is given.
llvm-svn: 191557

the other "Get*TypeName" functions.

llvm-svn: 191556

llvm-svn: 191555

llvm-svn: 191554

llvm-svn: 191553

Currently foldSelectICmpAndOr asserts if the "or" involves a vector
containing several of the same power of two. We can easily avoid this by
only performing the fold on integer types, like foldSelectICmpAnd does.

Fixes <rdar://problem/15012516>

llvm-svn: 191552