Looks like for 'and' and 'or' we end up performing at least some of the transformations this is bocking in a round about way anyway.

For 'and sext(cmp1), sext(cmp2) we end up later turning it into 'select cmp1, sext(cmp2), 0'. Then we optimize that back to sext (and cmp1, cmp2). This is the same result we would have gotten if shouldOptimizeCast hadn't blocked it. We do something analogous for 'or'.

With this patch we allow that transformation to happen directly in foldCastedBitwiseLogic. And we now support the same thing for 'xor'. This is definitely opening up many other cases, but since we already went around it for some cases hopefully it's ok.

Differential Revision: https://reviews.llvm.org/D36213

llvm-svn: 311508

llvm-svn: 311504

Summary:
ScopDetection used to check if a loop withing a region was infinite and emitted a diagnostic in such cases. After r310940 there's no point checking against that situation, as infinite loops don't appear in regions anymore.

The test failure was observed on these two polly buildbots:
http://lab.llvm.org:8011/builders/polly-arm-linux/builds/8368
http://lab.llvm.org:8011/builders/polly-amd64-linux/builds/10310

This patch XFAILs `ReportLoopHasNoExit.ll` and turns infinite loop detection into an assert.

Reviewers: grosser, sanjoy, bollu

Reviewed By: grosser

Subscribers: efriedma, aemerson, kristof.beyls, dberlin, llvm-commits

Tags: #polly

Differential Revision: https://reviews.llvm.org/D36776

llvm-svn: 311503

This reverts commit r311492.

llvm-svn: 311499

Summary:
There is no need to emit alias metadata for scalars, as basicaa will easily
distinguish them from arrays. This reduces the size of the metadata we generate.
This is especially useful after we moved to -polly-position=before-vectorizer,
where a lot more scalar dependences are introduced, which increased the size of
the alias analysis metadata and made us commonly reach the limits after which
we do not emit alias metadata that have been introduced to prevent quadratic
growth of this alias metadata.

This improves 2mm performance from 1.5 seconds to 0.17 seconds.

Reviewers: Meinersbur, bollu, singam-sanjay

Reviewed By: Meinersbur

Subscribers: pollydev, llvm-commits

Tags: #polly

Differential Revision: https://reviews.llvm.org/D37028

llvm-svn: 311498

This reverts commit r311468 because it broke some CFI bots.

llvm-svn: 311497

llvm-svn: 311496

The struct tag is going away in soon-to-be-released glibc 2.26 and the
stack_t typedef seems to have been there forever.

Patch by Bernhard Rosenkraenzer!

llvm-svn: 311495

[Polly] Satisfy Clang-format for r311489 changes, but it's weird that Clang-format didn't complain about headers order in previous versions (NFC).

llvm-svn: 311494

This patch adds printing for DW_AT_type DIEs like it's currently already
the case for DW_AT_specification DIEs.

llvm-svn: 311492

We can't reuse the llvm.assume instruction's bitcast because it may not
dominate every user of the vtable pointer.

Differential Revision: https://reviews.llvm.org/D36994

llvm-svn: 311491

Summary:
Use the initialexec TLS type and eliminate calls to the TLS
wrapper.  Fixes the sanitizer-x86_64-linux-fuzzer bot failure.

Reviewers: vitalybuka, kcc

Reviewed By: kcc

Subscribers: hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D37026

llvm-svn: 311490

llvm-svn: 311489

As before, not every platform supports LTO. Make sure the platform
we're targeting is one that supports it (regardless of the *host*
platform).

llvm-svn: 311488

With tests fixed for Windows style paths now that they are going
through path canonicalization.

llvm-svn: 311487

llvm-svn: 311484

JSON serializer fails to escape newlines and backslashes. Let's fix that.

Differential Revision: https://reviews.llvm.org/D34322

llvm-svn: 311483

as it breaks builtin standalone build on some bots.

llvm-svn: 311482

This is the same assertion as in https://reviews.llvm.org/D25206 that is
triggered when RecordLayoutBuilder tries to compute the size of a field
(for capture "typo_boo" in the test case) whose type hasn't been
deduced.

The fix is to add CorrectDelayedTyposInExpr call to the cases when we
aren't disambiguating between an Obj-C message send and a lambda
expression.

rdar://problem/31760839

Reviewers: rsmith, ahatanak

Reviewed By: arphaman

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D36853

llvm-svn: 311480

of class fails to map class static variable.

If the global variable is captured and it has several redeclarations,
sometimes it may lead to a compiler crash. Patch fixes this by working
only with canonical declarations.

llvm-svn: 311479

llvm-svn: 311477

Summary:
Rename stop-after to stop-diff-after. When building LLVM with
-DLLVM_BUILD_LLVM_DYLIB=ON, stop-after collides with the stop-after
already present in LLVM.

Reviewers: johannes, arphaman

Subscribers: klimek, aheejin, cfe-commits

Differential Revision: https://reviews.llvm.org/D36989

llvm-svn: 311476

llvm-svn: 311474

Currently, in case of GEMM and the pattern matching based optimizations, we
use only the SLP Vectorizer out of two LLVM vectorizers. Since the Loop
Vectorizer can get in the way of optimal code generation, we disable the Loop
Vectorizer for the innermost loop using mark nodes and emitting the
corresponding metadata.

Reviewed-by: Tobias Grosser <tobias@grosser.es>

Differential Revision: https://reviews.llvm.org/D36928

llvm-svn: 311473

llvm-svn: 311472

This was originally a `#define`. It is much easier to play around with
this as an environment variable when we run on large programs.

Differential Revision: https://reviews.llvm.org/D37012

llvm-svn: 311471

llvm-svn: 311470

The total number of loops is useful as a baseline comparing how many
loops have been optimized in different configurations.

llvm-svn: 311469

Patch by Rafael Espíndola.

This is PR34053.

The implementation is a bit of a hack, given the precise location where
IsPreemtible is set, it cannot be used from
SymbolTable::handleAnonymousVersion.

I could add another method to SymbolTable if you think that would be
better.

Differential Revision: https://reviews.llvm.org/D36499

llvm-svn: 311468

Summary:
This patch teaches ADCE to preserve both DominatorTrees and PostDominatorTrees.

This is reapplies the original patch r311057 that was reverted in r311381.
The previous version wasn't using the batch update api for updating dominators,
which in vary rare cases caused assertion failures.

This also fixes PR34258.

Reviewers: dberlin, chandlerc, sanjoy, davide, grosser, brzycki

Reviewed By: davide

Subscribers: grandinj, zhendongsu, llvm-commits, david2050

Differential Revision: https://reviews.llvm.org/D35869

llvm-svn: 311467

I was contacted by Jesper Antonsson from Ericsson who ran into problems
with r311181 in their test suites with for an out-of-tree target.
Because of the latter I don't have a reproducer, but we definitely don't
want to modify the data structure on which we are iterating inside the
loop.

llvm-svn: 311466

I don't see anything Darwin-specific here, so I made the target generic x86-64.

llvm-svn: 311465

I've replaced the two OS-specific runs with a generic run because
there's no functional difference in the resulting output that
we're checking. Also, the script still doesn't work with a Win
target.

llvm-svn: 311463

Update IR generated to retain section information for external declarations. 
This is related to https://reviews.llvm.org/D36487

Patch By: eandrews
Differential Revision: https://reviews.llvm.org/D36712

llvm-svn: 311459

Summary:
This patch is an alternative to https://reviews.llvm.org/D36614, by resolving a
non-idempotency issue by breaking non-trailing comments:

Consider formatting the following code with column limit at `V`:
```
                    V
const /* comment comment */ A = B;
```
The comment is not a trailing comment, breaking before it doesn't bring it under
the column limit. The formatter breaks after it, resulting in:

```
                    V
const /* comment comment */
    A = B;
```
For a next reformat, the formatter considers the comment as a trailing comment,
so it is free to break it further, resulting in:

```
                    V
const /* comment
         comment */
    A = B;
```
This patch improves the situation by directly producing the third case.

Reviewers: djasper

Reviewed By: djasper

Subscribers: cfe-commits, klimek

Differential Revision: https://reviews.llvm.org/D37007

llvm-svn: 311457

Summary:
This patch fixes a regression after https://reviews.llvm.org/rL305665,
which updates the structure of the `lines` variable.

Reviewers: djasper

Reviewed By: djasper

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D37011

llvm-svn: 311456

Rename the function getSupportedNanEncoding() to getIEEE754Standard(), since
this function will be used for non-nan related features.

Patch by Aleksandar Beserminji.

Differential Revision: https://reviews.llvm.org/D36824

llvm-svn: 311454

Add Armv8.3-A to the architecture to the TargetParser unittests.

Differential Revision: https://reviews.llvm.org/D36748

llvm-svn: 311450

llvm-svn: 311449

Armv8.3-A adds instructions that convert a double-precision floating
point number to a signed 32-bit integer with round towards zero,
designed for improving Javascript performance.

Differential Revision: https://reviews.llvm.org/D36785

llvm-svn: 311448