llvm-svn: 321109

Clang-format.

llvm-svn: 321107

Summary:
We use Diagnostic as a key to find the corresponding FixIt when we do
the "apply-fix", but the "severity" field could be omitted, in some cases,
the codeAction request sent from LSP clients (e.g. VScode) doesn't include the
`severity` field, which makes clangd fail to find the FixIt.

Test the following code in VScode, before the fix, no FixIt shown.

```
void main() {}
^~~~
```

Reviewers: sammccall

Reviewed By: sammccall

Subscribers: klimek, ilya-biryukov, cfe-commits

Differential Revision: https://reviews.llvm.org/D41280

llvm-svn: 321106

Store data relocations with their respective segment.
This allows relocations to be applied as each segment
is written (and therefore in parallel).

Differential Revision: https://reviews.llvm.org/D41410

llvm-svn: 321105

Another followup to my refactoring in r321036: Turns out we can end up
with an x86 darwin target that is not macos (simulator triples can look
like i386-apple-ios) so we need the x86/32bit check in all cases.

llvm-svn: 321104

Also remove unneeded braces and old comment.

llvm-svn: 321103

in '-target'

rdar://35742458

Differential Revision: https://reviews.llvm.org/D41076

llvm-svn: 321102

Verify that -lookup takes a 64-bit address.

llvm-svn: 321101

[AMDGPU] Turn off MergeConsecutiveStores() before Instruction Selection for AMDGPU. Commit dbbb6c5fc3642987430866dffdf710df4f616ac7 turned on MergeConsecutiveStores() before Instruction Selection for all targets. Enough AMDGPU compiles go into an infinite loop ( MergeConsecutiveStores() merges two stores; LegalizeStoreOps() un-merges; MergeConsecutiveStores() re-merges, etc. ) to warrant turning it off until the issues can be addressed.

Differential Revision: https://reviews.llvm.org/D41377

llvm-svn: 321100

OS instead of inferring it from SDK / environment

The OS version is specified in -target should be used instead of the one in an
environment variable / SDK name.

rdar://35813850

Differential Revision: https://reviews.llvm.org/D40998

llvm-svn: 321099

do CSE for to => do CSE to

llvm-svn: 321098

Patch by Nicholas Wilson

llvm-svn: 321097

Patch by Nicholas Wilson

llvm-svn: 321096

When building with cmake on green gragon or on ci.swift.org, this test fails.

rdar://problem/36134350

llvm-svn: 321095

llvm-svn: 321094

llvm-svn: 321093

Reviewers: sammccall

Reviewed By: sammccall

Subscribers: klimek, ilya-biryukov, cfe-commits

Differential Revision: https://reviews.llvm.org/D41289

llvm-svn: 321092

This doesn't add legalizer support, just prevents crashing so that we
can gracefully fall back to SDAG.

Fixes PR35690.

llvm-svn: 321091

The Clang option -foptimization-record-file= controls which file an
optimization record is output to. Optimization records are output if you
use the Clang option -fsave-optimization-record. If you specify the
first option without the second, you get a warning that the command line
argument was unused. Passing -foptimization-record-file= should imply
-fsave-optimization-record.

This fixes PR33670

Patch by: Dmitry Venikov <venikov@phystech.edu>

Differential revision: https://reviews.llvm.org/D39834

llvm-svn: 321090

Summary:
Extend overlapping store elision to handle overwrites of stores by
larger stores.

Nontemporal tests have been modified to add memory dependencies to
prevent store elision.

Reviewers: craig.topper, rnk, t.p.northover

Subscribers: javed.absar, hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D40969

llvm-svn: 321089

This change add support for init functions in the linking
section, but only in -r/--relocatable mode.

Differential Revision: https://reviews.llvm.org/D41375

llvm-svn: 321088

Going to add unit tests in the next patch. (Haha!) But seriously there's
some work to do first - need to extract the markers-in-source-code
parser from CodeComplete test and make it more flexible, to allow
annotated ranges etc.

llvm-svn: 321087

llvm-svn: 321086

We try to prevent shuffle combining to value types that would stop the folding of masked operations, but by just returning early, we were failing to try different shuffle types.

The TODOs are all still relevant here to improve codegen but we're lacking test examples.

llvm-svn: 321085

Work towards the unification of MIR and debug output by refactoring the
interfaces.

Before this patch we printed "<call frame instruction>" in the debug
output.

llvm-svn: 321084

Summary: Use symbol index to populate completion results for qualfified IDs e.g. "nx::A^".

Reviewers: ilya-biryukov, sammccall

Reviewed By: ilya-biryukov, sammccall

Subscribers: rwols, klimek, mgorny, cfe-commits, sammccall

Differential Revision: https://reviews.llvm.org/D41281

llvm-svn: 321083

Original commit is at: https://reviews.llvm.org/D41295.

llvm-svn: 321082

llvm-svn: 321081

Summary: This is to be consistent with latest Movidius MDK releases.
Also, don't inherit any gcc paths for shave triple.

Reviewers: jyknight

Subscribers: emaste, fedor.sergeev

Differential Revision: https://reviews.llvm.org/D41295

llvm-svn: 321080

The refactoring in r281640 made -dot-cfg-only ignore the "-only" part.

llvm-svn: 321079

Negative values never disabled the pruning - they simply set high values for the pruning interval.

The behaviour now is that negative values set the maximum pruning interval (which appears to have been the intention from the start) see https://reviews.llvm.org/D41231.

I have adjusted the comments to reflect this, removed any inaccurate statements, and corrected any typos I spotted in the English.

Differential Revision: https://reviews.llvm.org/D41279

llvm-svn: 321078

borked by: rL284966 (see: https://reviews.llvm.org/D25730).

Previously, Interval was unsigned (see: CachePruning.h), replacing the type with std::chrono::seconds (which is signed) causes a regression in behaviour because the c-api intends negative values to translate to large positive intervals to *effectively* disable the pruning (see comments on: setCachePruningInterval()).

Differential Revision: https://reviews.llvm.org/D41231

llvm-svn: 321077

llvm-svn: 321076

This is a follow-up fix of r320814.  A test case is also added.

llvm-svn: 321075

As mentioned in D38318 and D40865, modern Intel processors prefer to combine multiple shuffles to a variable shuffle mask (PSHUFB/VPERMPS etc.) instead of having multiple stage 'fixed' shuffles which put more pressure on Port 5 (at the expense of extra shuffle mask loads).

This patch provides a FeatureFastVariableShuffle target flag for Haswell+ CPUs that prefers combining 2 or more fixed shuffles to a single variable shuffle (default is 3 shuffles).

The long term aim is to drive more of this from schedule data (probably via the MC) but we're not close to being ready for that yet.

Differential Revision: https://reviews.llvm.org/D41323

llvm-svn: 321074

Summary:
- Moved these functions to SourceCode.h
- added unit tests
- fix off by one in positionToOffset: Offset - 1 in final calculation was wrong
- fixed formatOnType which had an equal and opposite off-by-one
- positionToOffset and offsetToPosition both consistently clamp to beginning/end
  of file when input is out of range
- gave variables more descriptive names
- removed windows line ending fixmes where there is nothing to fix
- elaborated on UTF-8 fixmes

This will conflict with Eric's D41281, but in a pretty easy-to-resolve way.

Reviewers: ioeric

Subscribers: klimek, mgorny, ilya-biryukov, cfe-commits

Differential Revision: https://reviews.llvm.org/D41351

llvm-svn: 321073

Also adds a simple test case.

llvm-svn: 321072

Summary:
The motivation here is LLDB, where we need to fixup relocations in
mmapped files before their contents can be read correctly.  The
MemoryBuffer class does exactly what we need, *except* that it maps the
file in read-only mode.

WritableMemoryBuffer reuses the existing machinery for opening and
mmapping a file. The only difference is in the argument to the
mapped_file_region constructor -- we create a private copy-on-write
mapping, so that we can make changes to the mapped data, but the changes
aren't carried over to the underlying file.

This patch is based on an initial version by Zachary Turner.

Reviewers: mehdi_amini, rnk, rafael, dblaikie, zturner

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D40291

llvm-svn: 321071

Extension to D39729 which performed this for vXi16, with the same bit flipping to handle SMAX/SMIN/UMAX cases, vXi8 UMIN horizontal reductions can be performed.

This makes use of the fact that by performing a pair-wise i8 SHUFFLE/UMIN before PHMINPOSUW, we both get the UMIN of each pair but also zero-extend the upper bits ready for v8i16.

Differential Revision: https://reviews.llvm.org/D41294

llvm-svn: 321070

llvm-svn: 321069