llvm-svn: 296203

llvm-svn: 296202

llvm-svn: 296200

llvm-svn: 296199

llvm-svn: 296197

llvm-svn: 296195

llvm-svn: 296194

llvm-svn: 296192

llvm-svn: 296191

llvm-svn: 296189

------------------------------------------------------------------------
r288115 | abataev | 2016-11-29 09:21:14 +0100 (Tue, 29 Nov 2016) | 8 lines

[SLPVectorizer] Improved support of partial tree vectorization.

Currently SLP vectorizer tries to vectorize a binary operation and dies
immediately after unsuccessful the first unsuccessfull attempt. Patch
tries to improve the situation, trying to vectorize all binary
operations of all children nodes in the binop tree.

Differential Revision: https://reviews.llvm.org/D25517
------------------------------------------------------------------------

llvm-svn: 296185

------------------------------------------------------------------------
r295990 | jvesely | 2017-02-23 08:12:21 -0800 (Thu, 23 Feb 2017) | 5 lines

AMDGPU/SI: Fix trunc i16 pattern

Hit on ASICs that support 16bit instructions.

Differential Revision: https://reviews.llvm.org/D30281
------------------------------------------------------------------------

llvm-svn: 296158

------------------------------------------------------------------------
r296003 | mcrosier | 2017-02-23 10:49:03 -0800 (Thu, 23 Feb 2017) | 32 lines

[Reassociate] Add negated value of negative constant to the Duplicates list.

In OptimizeAdd, we scan the operand list to see if there are any common factors
between operands that can be factored out to reduce the number of multiplies
(e.g., 'A*A+A*B*C+D' -> 'A*(A+B*C)+D'). For each operand of the operand list, we
only consider unique factors (which is tracked by the Duplicate set). Now if we
find a factor that is a negative constant, we add the negated value as a factor
as well, because we can percolate the negate out. However, we mistakenly don't
add this negated constant to the Duplicates set.

Consider the expression A*2*-2 + B. Obviously, nothing to factor.

For the added value A*2*-2 we over count 2 as a factor without this change,
which causes the assert reported in PR30256.  The problem is that this code is
assuming that all the multiply operands of the add are already reassociated.
This change avoids the issue by making OptimizeAdd tolerate multiplies which
haven't been completely optimized; this sort of works, but we're doing wasted
work: we'll end up revisiting the add later anyway.

Another possible approach would be to enforce RPO iteration order more strongly.
If we have RedoInsts, we process them immediately in RPO order, rather than
waiting until we've finished processing the whole function. Intuitively, it
seems like the natural approach: reassociation works on expression trees, so
the optimization only works in one direction. That said, I'm not sure how
practical that is given the current Reassociate; the "optimal" form for an
expression depends on its use list (see all the uses of "user_back()"), so
Reassociate is really an iterative optimization of sorts, so any changes here
would probably get messy.

PR30256

Differential Revision: https://reviews.llvm.org/D30228
------------------------------------------------------------------------

llvm-svn: 296156

------------------------------------------------------------------------
r293604 | sammccall | 2017-01-30 21:23:20 -0800 (Mon, 30 Jan 2017) | 12 lines

In VirtualCallChecker, handle indirect calls

Summary:
In VirtualCallChecker, handle indirect calls.

getDirectCallee() can be nullptr, and dyn_cast(nullptr) is UB

Reviewers: bkramer

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D29303
------------------------------------------------------------------------

llvm-svn: 296154

------------------------------------------------------------------------
r296063 | hans | 2017-02-23 17:16:34 -0800 (Thu, 23 Feb 2017) | 4 lines

Revert r291477 "[Frontend] Correct values of ATOMIC_*_LOCK_FREE to match builtin"

It caused PR31864. There is a patch in progress to fix that, but let's
revert in the meantime.
------------------------------------------------------------------------

llvm-svn: 296152

------------------------------------------------------------------------
r295474 | hahnfeld | 2017-02-17 10:32:58 -0800 (Fri, 17 Feb 2017) | 6 lines

[OpenMP] Fix cancellation point in task with no cancel

With tasks, the cancel may happen in another task. This has a different
region info which means that we can't find it here.

Differential Revision: https://reviews.llvm.org/D30091
------------------------------------------------------------------------

llvm-svn: 296151

llvm-svn: 296150

------------------------------------------------------------------------
r296030 | hans | 2017-02-23 14:29:00 -0800 (Thu, 23 Feb 2017) | 7 lines

Revert r282872 "CVP. Turn marking adds as no wrap on by default"

While not CVP's fault, this caused miscompiles (PR31181). Reverting
until those are resolved.

(This also reverts the follow-ups r288154 and r288161 which removed the
flag.)
------------------------------------------------------------------------

llvm-svn: 296148

------------------------------------------------------------------------
r295474 | hahnfeld | 2017-02-17 10:32:58 -0800 (Fri, 17 Feb 2017) | 6 lines

[OpenMP] Fix cancellation point in task with no cancel

With tasks, the cancel may happen in another task. This has a different
region info which means that we can't find it here.

Differential Revision: https://reviews.llvm.org/D30091
------------------------------------------------------------------------

llvm-svn: 296139

------------------------------------------------------------------------
r295762 | eugenis | 2017-02-21 12:17:34 -0800 (Tue, 21 Feb 2017) | 3 lines

Fix PR31896.

Address of an alias of a global with offset is incorrectly lowered as an address of the global (i.e. ignoring offset).
------------------------------------------------------------------------

llvm-svn: 296002

------------------------------------------------------------------------
r295843 | jvesely | 2017-02-22 07:01:42 -0800 (Wed, 22 Feb 2017) | 3 lines

[OpenCL] r600 needs OpenCL kernel calling convention

Differential Revision: https://reviews.llvm.org/D30236
------------------------------------------------------------------------

llvm-svn: 296001

------------------------------------------------------------------------
r295473 | hahnfeld | 2017-02-17 10:32:51 -0800 (Fri, 17 Feb 2017) | 13 lines

[OpenMP] Remove barriers at cancel and cancellation point

This resolves a deadlock with the cancel directive when there is no explicit
cancellation point. In that case, the implicit barrier acts as cancellation
point. After removing the barrier after cancel, the now unmatched barrier for
the explicit cancellation point has to go as well.

This has probably worked before rL255992: With the calls for the explicit
barrier, it was sure that all threads passed a barrier before exiting.

Reported by Simon Convent and Joachim Protze!

Differential Revision: https://reviews.llvm.org/D30088
------------------------------------------------------------------------

llvm-svn: 296000

Support lowering AEABI TLS access (__aeabi_read_tp) with long calls.
This requires adjusting the call sequence to use an indirect call to get
full addressability.

Resolves PR31769!

By Saleem Abdulrasool!

llvm-svn: 295910

------------------------------------------------------------------------
r294954 | chapuni | 2017-02-13 06:59:53 -0800 (Mon, 13 Feb 2017) | 5 lines

Fix r291495 -- Normalize LLVM_CMAKE_PATH in clang standalone build.

CMake handles paths with slashes. It caused cmake/install failure on msbuild.exe.

Note, Other llvm-config-oriented variables have been normalized since they are stored in the cache attributed with PATH.
------------------------------------------------------------------------

------------------------------------------------------------------------
r295592 | chapuni | 2017-02-18 19:17:31 -0800 (Sat, 18 Feb 2017) | 1 line

clang/CMakeLists.txt: Rework r294954 -- use file(TO_CMAKE_PATH).
------------------------------------------------------------------------

llvm-svn: 295760

------------------------------------------------------------------------
r295610 | brad | 2017-02-19 11:33:26 -0800 (Sun, 19 Feb 2017) | 4 lines

Link static PIE programs against rcrt0.o on OpenBSD

Patch by Stefan Kempf.

------------------------------------------------------------------------

llvm-svn: 295752

------------------------------------------------------------------------
r293273 | chandlerc | 2017-01-27 02:27:32 -0800 (Fri, 27 Jan 2017) | 19 lines

[LICM] When we are recomputing the alias sets for a subloop, we cannot
skip sub-subloops.

The logic to skip subloops dated from when this code was shared with the
cached case. Once it was factored out to only run in the case of
recomputed subloops it became a dangerous bug. If a subsubloop contained
an interfering instruction it would be silently skipped from the alias
sets for LICM.

With the old pass manager this was extremely hard to trigger as it would
require failing to visit these subloops with the LICM pass but then
visiting the outer loop somehow. I've not yet contrived any test case
that actually manages to trigger this.

But with the new pass manager we don't do the cross-loop caching hack
that the old PM does and so we recompute alias set information from
first principles. While this seems much cleaner and simpler it exposed
this bug and would subtly miscompile code due to failing to correctly
model the aliasing constraints of deeply nested loops.
------------------------------------------------------------------------

llvm-svn: 295748

------------------------------------------------------------------------
r295486 | adrian | 2017-02-17 11:42:32 -0800 (Fri, 17 Feb 2017) | 6 lines

Debug Info: Sort frame index expressions before emitting them.
This fixes PR31381, which caused an assertion and/or invalid debug info.

This affects debug variables that have multiple fragments in the MMI
side (i.e.: in the stack frame) table.
rdar://problem/30571676
------------------------------------------------------------------------

------------------------------------------------------------------------
r295490 | adrian | 2017-02-17 12:02:26 -0800 (Fri, 17 Feb 2017) | 1 line

Fix windows bots by locking down the target triple on this testcase.
------------------------------------------------------------------------

llvm-svn: 295747

------------------------------------------------------------------------
r295512 | matze | 2017-02-17 15:15:03 -0800 (Fri, 17 Feb 2017) | 8 lines

AArch64LoadStoreOptimizer: Correctly clear kill flags

When promoting the Load of a Store-Load pair to a COPY all kill flags
between the store and the load need to be cleared.

rdar://30402435

Differential Revision: https://reviews.llvm.org/D30110
------------------------------------------------------------------------

llvm-svn: 295744

llvm-svn: 295743

llvm-svn: 295742

------------------------------------------------------------------------
r293124 | mkuper | 2017-01-25 17:04:11 -0800 (Wed, 25 Jan 2017) | 9 lines

[LoopUnroll] Properly update loopinfo for runtime unrolling by 2

Even when we don't create a remainder loop (that is, when we unroll by 2), we
may duplicate nested loops into the remainder. This is complicated by the fact
the remainder may itself be either inserted into an outer loop, or at the top
level. In the latter case, we may need to create new top-level loops.

Differential Revision: https://reviews.llvm.org/D29156

------------------------------------------------------------------------

llvm-svn: 295380

------------------------------------------------------------------------
r295230 | arnolds | 2017-02-15 12:43:43 -0800 (Wed, 15 Feb 2017) | 11 lines

AddressSanitizer: don't track swifterror memory addresses

They are register promoted by ISel and so it makes no sense to treat them as
memory.

Inserting calls to the thread sanitizer would also generate invalid IR.

You would hit:

"swifterror value can only be loaded and stored from, or as a swifterror
argument!"
------------------------------------------------------------------------

llvm-svn: 295377

------------------------------------------------------------------------
r295215 | arnolds | 2017-02-15 10:57:06 -0800 (Wed, 15 Feb 2017) | 11 lines

ThreadSanitizer: don't track swifterror memory addresses

They are register promoted by ISel and so it makes no sense to treat them as
memory.

Inserting calls to the thread sanitizer would also generate invalid IR.

You would hit:

"swifterror value can only be loaded and stored from, or as a swifterror
argument!"
------------------------------------------------------------------------

llvm-svn: 295376

------------------------------------------------------------------------
r295224 | rsmith | 2017-02-15 11:57:10 -0800 (Wed, 15 Feb 2017) | 2 lines

PR24440: Do not silently discard a fold-expression appearing as the operand of a cast-expression.

------------------------------------------------------------------------

llvm-svn: 295375

------------------------------------------------------------------------
r295213 | mkuper | 2017-02-15 10:37:26 -0800 (Wed, 15 Feb 2017) | 10 lines

[DAG] Don't try to create an INSERT_SUBVECTOR with an illegal source

We currently can't legalize those, but we should really not be creating
them in the first place, since legalization would probably look similar to the
way we legalize CONCAT_VECTORS - basically replace the INSERT with a BUILD.

This fixes PR311956.

Differential Revision: https://reviews.llvm.org/D29961

------------------------------------------------------------------------

llvm-svn: 295374

LDC, the LLVM-based D compiler, is already ready for LLVM 4.0.

llvm-svn: 295367

------------------------------------------------------------------------
r292271 | aaron | 2017-01-17 21:48:31 +0000 (Tue, 17 Jan 2017) | 2 lines

Silence some Sphinx diagnostics in an attempt to get the documentation builder back to green (http://lab.llvm.org:8011/builders/llvm-sphinx-docs/builds/1895).
------------------------------------------------------------------------

llvm-svn: 295344

------------------------------------------------------------------------
r295313 | stulova | 2017-02-16 04:49:29 -0800 (Thu, 16 Feb 2017) | 9 lines

[OpenCL][Doc] Added OpenCL vendor extension description to user manual doc

Added description of a new feature that allows to specify
vendor extension in flexible way using compiler pragma instead
of modifying source code directly (committed in clang@r289979).

Review: D29829


------------------------------------------------------------------------

llvm-svn: 295340

Review: D29830
llvm-svn: 295315

------------------------------------------------------------------------
r294129 | gberry | 2017-02-05 10:28:14 -0800 (Sun, 05 Feb 2017) | 16 lines

[SelectionDAG] In InstrEmitter, handle EXTRACT_SUBREG of a physical register.

Summary:
Without this change, the getVR() call would hit an assert since it was
being passed a physical register.

Update the AArch64/ldst-opt.ll test with a case that triggers this
behavior by adding a run with strict-align, which causes an unaligned
STR XZR instruction to be split into byte stores, creating an
EXTRACT_SUBREG of XZR that triggers the original problem.

Reviewers: bogner, qcolombet, MatzeB, atrick

Subscribers: aemerson, mcrosier, llvm-commits

Differential Revision: https://reviews.llvm.org/D29495
------------------------------------------------------------------------

llvm-svn: 295250