platform/upstream/llvm.git
6 years agoAMDGPU: Turn D16 for MIMG instructions into a regular operand
Nicolai Haehnle [Thu, 21 Jun 2018 13:36:01 +0000 (13:36 +0000)]
AMDGPU: Turn D16 for MIMG instructions into a regular operand

Summary:
This allows us to reduce the number of different machine instruction
opcodes, which reduces the table sizes and helps flatten the TableGen
multiclass hierarchies.

We can do this because for each hardware MIMG opcode, we have a full set
of IMAGE_xxx_Vn_Vm machine instructions for all required sizes of vdata
and vaddr registers. Instead of having separate D16 machine instructions,
a packed D16 instructions loading e.g. 4 components can simply use the
same V2 opcode variant that non-D16 instructions use.

We still require a TSFlag for D16 buffer instructions, because the
D16-ness of buffer instructions is part of the opcode. Renaming the flag
should help avoid future confusion.

The one non-obvious code change is that for gather4 instructions, the
disassembler can no longer automatically decide whether to use a V2 or
a V4 variant. The existing logic which choose the correct variant for
other MIMG instruction is extended to cover gather4 as well.

As a bonus, some of the assembler error messages are now more helpful
(e.g., complaining about a wrong data size instead of a non-existing
instruction).

While we're at it, delete a whole bunch of dead legacy TableGen code.

Change-Id: I89b02c2841c06f95e662541433e597f5d4553978

Reviewers: arsenm, rampitec, kzhuravl, artem.tamazov, dp, rtaylor

Subscribers: wdng, yaxunl, dstuttard, tpr, t-tye, llvm-commits

Differential Revision: https://reviews.llvm.org/D47434

llvm-svn: 335222

6 years agoTableGen: Allow foreach in multiclass to depend on template args
Nicolai Haehnle [Thu, 21 Jun 2018 13:35:44 +0000 (13:35 +0000)]
TableGen: Allow foreach in multiclass to depend on template args

Summary:
This also allows inner foreach loops to have a list that depends on
the iteration variable of an outer foreach loop. The test cases show
some very simple examples of how this can be used.

This was perhaps the last remaining major non-orthogonality in the
TableGen frontend.

Change-Id: I79b92d41a5c0e7c03cc8af4000c5e1bda5ef464d

Reviewers: tra, simon_tatham, craig.topper, MartinO, arsenm

Subscribers: wdng, llvm-commits

Differential Revision: https://reviews.llvm.org/D47431

llvm-svn: 335221

6 years agoFix line endings in recently updated test file
Ivan Donchevskii [Thu, 21 Jun 2018 12:39:24 +0000 (12:39 +0000)]
Fix line endings in recently updated test file

llvm-svn: 335220

6 years ago[llvm-mca] Updates comment in code, and remove some stale comments. NFC
Andrea Di Biagio [Thu, 21 Jun 2018 12:14:49 +0000 (12:14 +0000)]
[llvm-mca] Updates comment in code, and remove some stale comments. NFC

Also, rename fields `TotalMappings` and `NumUsedMappings` in struct
RegisterMappingTracker into `NumPhysRegs` and `NumUsedPhysRegs`.

llvm-svn: 335219

6 years ago[clangd] Expose 'shouldCollectSymbol' helper from SymbolCollector.
Eric Liu [Thu, 21 Jun 2018 12:12:26 +0000 (12:12 +0000)]
[clangd] Expose 'shouldCollectSymbol' helper from SymbolCollector.

Summary: This allows tools to examine symbols that would be collected in a symbol index. For example, a tool that measures index-based completion quality would be interested in references to symbols that are collected in the index.

Reviewers: sammccall

Reviewed By: sammccall

Subscribers: ilya-biryukov, MaskRay, jkorous, cfe-commits

Differential Revision: https://reviews.llvm.org/D48418

llvm-svn: 335218

6 years ago[DA] Enable -da-delinearize by default
David Green [Thu, 21 Jun 2018 11:53:16 +0000 (11:53 +0000)]
[DA] Enable -da-delinearize by default

This enables da-delinearize in Dependence Analysis for delinearizing array
accesses into multiple dimensions. This can help to increase the power of
Dependence analysis on multi-dimensional arrays and prevent having to fall
back to the slower and less accurate MIV tests. It adds static checks on the
bounds of the arrays to ensure that one dimension doesn't overflow into
another, and brings our code in line with our tests.

Differential Revision: https://reviews.llvm.org/D45872

llvm-svn: 335217

6 years ago[X86][AVX] Reduce v4f64/v4i64 shuffle costs (PR37882)
Simon Pilgrim [Thu, 21 Jun 2018 11:37:13 +0000 (11:37 +0000)]
[X86][AVX] Reduce v4f64/v4i64 shuffle costs (PR37882)

These were being over cautious for costs for one/two op general shuffles - VSHUFPD doesn't have to replicate the same shuffle in both lanes like VSHUFPS does.

llvm-svn: 335216

6 years ago[SLPVectorizer][X86] Add horizontal add/sub tests
Simon Pilgrim [Thu, 21 Jun 2018 11:16:10 +0000 (11:16 +0000)]
[SLPVectorizer][X86] Add horizontal add/sub tests

Shows PR37882 perf regression

llvm-svn: 335215

6 years ago[DebugInfo] Make sure all DBG_VALUEs' reguse operands have IsDebug property
Mikael Holmen [Thu, 21 Jun 2018 10:03:34 +0000 (10:03 +0000)]
[DebugInfo] Make sure all DBG_VALUEs' reguse operands have IsDebug property

Summary:
In some cases, these operands lacked the IsDebug property, which is meant to signal that
they should not affect codegen. This patch adds a check for this property in the
MachineVerifier and adds it where it was missing.

This includes refactorings to use MachineInstrBuilder construction functions instead of
manually setting up the intrinsic everywhere.

Patch by: JesperAntonsson

Reviewers: aprantl, rnk, echristo, javed.absar

Reviewed By: aprantl

Subscribers: qcolombet, sdardis, nemanjai, JDevlieghere, atanasyan, llvm-commits

Differential Revision: https://reviews.llvm.org/D48319

llvm-svn: 335214

6 years agoCODE_OWNERS: Take ownership of the MIPS backend
Simon Atanasyan [Thu, 21 Jun 2018 09:59:44 +0000 (09:59 +0000)]
CODE_OWNERS: Take ownership of the MIPS backend

As agreed with Simon Dardis, I'm taking over as a code owner
for the MIPS backend.

llvm-svn: 335213

6 years ago[clangd] Fix proximity signal output format. NFC
Eric Liu [Thu, 21 Jun 2018 09:51:28 +0000 (09:51 +0000)]
[clangd] Fix proximity signal output format. NFC

llvm-svn: 335212

6 years ago[Sema] Fix overloaded static functions for templates
Ivan Donchevskii [Thu, 21 Jun 2018 08:34:50 +0000 (08:34 +0000)]
[Sema] Fix overloaded static functions for templates

Apply almost the same fix as https://reviews.llvm.org/D36390 but for templates.

Differential Revision: https://reviews.llvm.org/D43453

llvm-svn: 335211

6 years ago[DAGCombine] Fix alignment for offset loads/stores
David Green [Thu, 21 Jun 2018 08:30:07 +0000 (08:30 +0000)]
[DAGCombine] Fix alignment for offset loads/stores

The alignment parameter to getExtLoad is treated as a base alignment,
not the alignment of the load (base + offset). When we infer a better
alignment for a Ptr we need to ensure that it applies to the base to
prevent the alignment on the load from being wrong.

This fixes a bug where the alignment could then be used to incorrectly
prove noalias between a load and a store, leading to a miscompile.

Differential Revision: https://reviews.llvm.org/D48029

llvm-svn: 335210

6 years agoFix double-float constant truncation warnings. NFCI.
Simon Pilgrim [Thu, 21 Jun 2018 07:54:47 +0000 (07:54 +0000)]
Fix double-float constant truncation warnings. NFCI.

llvm-svn: 335209

6 years agoRemove FIXME comment about WIP.
Eric Christopher [Thu, 21 Jun 2018 07:15:19 +0000 (07:15 +0000)]
Remove FIXME comment about WIP.
This is the only line other than the function signature remaining
of the original patch.

llvm-svn: 335208

6 years agoAdd some explanatory text to the associated symbol support.
Eric Christopher [Thu, 21 Jun 2018 07:15:14 +0000 (07:15 +0000)]
Add some explanatory text to the associated symbol support.

llvm-svn: 335207

6 years agoRecommit r333268: [IPSCCP] Use PredicateInfo to propagate facts from cmp instructions.
Florian Hahn [Thu, 21 Jun 2018 07:15:08 +0000 (07:15 +0000)]
Recommit r333268: [IPSCCP] Use PredicateInfo to propagate facts from cmp instructions.

r335150 should resolve the issues with the clang-with-thin-lto-ubuntu
and clang-with-lto-ubuntu builders.

Original message:
This patch updates IPSCCP to use PredicateInfo to propagate
facts to true branches predicated by EQ and to false branches
predicated by NE.

As a follow up, we should be able to extend it to also propagate additional
facts about nonnull.

Reviewers: davide, mssimpso, dberlin, efriedma

Reviewed By: davide, dberlin

llvm-svn: 335206

6 years ago[DebugInfo] Keep DBG_VALUE undef in LiveDebugVariables
Mikael Holmen [Thu, 21 Jun 2018 07:02:46 +0000 (07:02 +0000)]
[DebugInfo] Keep DBG_VALUE undef in LiveDebugVariables

Summary:
Fixes PR36579.

For cases where we had e.g.

 DBG_VALUE 42
 [...]
 DBG_VALUE undef

LiveDebugVariables would discard all undef DBG_VALUEs and then it would
look like the variable had the value 42 throughout the rest of the
function, which is incorrect.

With this patch we don't remove all undef DBG_VALUEs in LiveDebugVariables
so they will be kept after register allocation just like other DBG_VALUEs
which will yield more correct debug information.

Reviewers: aprantl

Reviewed By: aprantl

Subscribers: bjope, Ka-Ka, JDevlieghere, llvm-commits

Differential Revision: https://reviews.llvm.org/D48277

llvm-svn: 335205

6 years ago[X86] Go through some tests that still reference old intrinsics that have been autoup...
Craig Topper [Thu, 21 Jun 2018 06:17:16 +0000 (06:17 +0000)]
[X86] Go through some tests that still reference old intrinsics that have been autoupgraded and replace them with the upgraded IR.

This is mostly the stack folding tests and is by no means a thorough audit of tests.

llvm-svn: 335204

6 years ago[PM/LoopUnswitch] Add partial non-trivial unswitching for invariant
Chandler Carruth [Thu, 21 Jun 2018 06:14:03 +0000 (06:14 +0000)]
[PM/LoopUnswitch] Add partial non-trivial unswitching for invariant
conditions feeding a chain of `and`s or `or`s for a branch.

Much like with full non-trivial unswitching, we rely on the pass manager
to handle iterating until all of the profitable unswitches have been
done. This is to allow other more profitable unswitches to fire on any
of the cloned, simpler versions of the loop if viable.

Threading the partial unswiching through the non-trivial unswitching
logic motivated some minor refactorings. If those are too disruptive to
make it reasonable to review this patch, I can separate them out, but
it'll be somewhat timeconsuming so I wanted to send it for initial
review as-is. Feel free to tell me whether it warrants pulling apart.

I've tried to re-use (and factor out) logic form the partial trivial
unswitching, but not as much could be shared as I had haped. Still, this
wasn't as bad as I naively expected.

Some basic testing is added, but I probably need more. Suggestions for
things you'd like to see tested more than welcome. One thing I'd like to
do is add some testing that when we schedule this with loop-instsimplify
it effectively cleans up the cruft created.

Last but not least, this uncovered a bug that has been in loop cloning
the entire time for non-trivial unswitching. Specifically, we didn't
correctly add the outer-most cloned loop to the list of cloned loops.
This meant that LCSSA wouldn't be updated for it hypothetically, and
more significantly that we would never visit it in the loop pass
manager. I noticed this while checking loop-instsimplify by hand. I'll
try to separate this bugfix out into its own patch with a more focused
test. But it is just one line, so shouldn't significantly confuse the
review here.

After this patch, the only missing "feature" in this unswitch I'm aware
of us non-trivial unswitching of switches. I'll try implementing *full*
non-trivial unswitching of switches (which is at least a sound thing to
implement), but *partial* non-trivial unswitching of switches is
something I don't see any sound and principled way to implement. I also
have no interesting test cases for the latter, so I'm not really
worried. The rest of the things that need to be ported are bug-fixes and
more narrow / targeted support for specific issues.

Differential Revision: https://reviews.llvm.org/D47522

llvm-svn: 335203

6 years ago[RISC-V] Fix a test case to not include label names as those aren't
Chandler Carruth [Thu, 21 Jun 2018 05:42:05 +0000 (05:42 +0000)]
[RISC-V] Fix a test case to not include label names as those aren't
stable in non-asserts builds. This fixes a test failure in release
config.

llvm-svn: 335202

6 years agoProvenanceAnalysis: Store WeakTrackingVH instead of Value* in UnderlyingValue Cache.
Michael Zolotukhin [Thu, 21 Jun 2018 05:14:00 +0000 (05:14 +0000)]
ProvenanceAnalysis: Store WeakTrackingVH instead of Value* in UnderlyingValue Cache.

Summary:
Since the value stored in the cache might be deleted or replaced with
something else, we need to use tracking ValueHandlers instead of plain
Value pointers. It was discovered in one of internal builds, and
unfortunately there is no small reproducer for the issue.

The cache was introduced in rL327328.

Reviewers: ahatanak, pete

Subscribers: hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D48407

llvm-svn: 335201

6 years ago[X86] Remove masking from the 512-bit floating point max/min builtins. Use select...
Craig Topper [Thu, 21 Jun 2018 05:01:01 +0000 (05:01 +0000)]
[X86] Remove masking from the 512-bit floating point max/min builtins. Use select in IR instead.

llvm-svn: 335200

6 years ago[X86] Remove masking from 512-bit floating max/min intrinsics. Use select instruction...
Craig Topper [Thu, 21 Jun 2018 05:00:56 +0000 (05:00 +0000)]
[X86] Remove masking from 512-bit floating max/min intrinsics. Use select instruction instead.

llvm-svn: 335199

6 years agoRevert "[SCEV] Improve zext(A /u B) and zext(A % B)"
Tim Shen [Thu, 21 Jun 2018 02:15:32 +0000 (02:15 +0000)]
Revert "[SCEV] Improve zext(A /u B) and zext(A % B)"

This reverts commit r335197, as some bots are not happy.

llvm-svn: 335198

6 years ago[SCEV] Improve zext(A /u B) and zext(A % B)
Tim Shen [Thu, 21 Jun 2018 01:49:07 +0000 (01:49 +0000)]
[SCEV] Improve zext(A /u B) and zext(A % B)

Summary:
Try to match udiv and urem patterns, and sink zext down to the leaves.

I'm not entirely sure why some unrelated tests change, but the added <nsw>s seem right.

Reviewers: sanjoy

Subscribers: jlebar, hiraditya, bixia, llvm-commits

Differential Revision: https://reviews.llvm.org/D48338

llvm-svn: 335197

6 years agoRevert "Add python tool to dump and construct header maps"
Bruno Cardoso Lopes [Thu, 21 Jun 2018 01:23:58 +0000 (01:23 +0000)]
Revert "Add python tool to dump and construct header maps"

This reverts commit fcfa2dd517ec1a6045a81e8247e346d630a22618.

Broke bots:

http://lab.llvm.org:8011/builders/clang-x64-ninja-win7/builds/11315
http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/10411/steps/test-check-all/logs/stdio

llvm-svn: 335196

6 years agoRevert "Warning for framework headers using double quote includes"
Bruno Cardoso Lopes [Thu, 21 Jun 2018 01:23:51 +0000 (01:23 +0000)]
Revert "Warning for framework headers using double quote includes"

This reverts commit 9b5ff2db7e31c4bb11a7d468260b068b41c7c285.

Broke bots:

http://lab.llvm.org:8011/builders/clang-x64-ninja-win7/builds/11315
http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/10411/steps/test-check-all/logs/stdio

llvm-svn: 335195

6 years agoRevert "Fix hmaptool cmake file to work on Windows"
Bruno Cardoso Lopes [Thu, 21 Jun 2018 01:23:42 +0000 (01:23 +0000)]
Revert "Fix hmaptool cmake file to work on Windows"

This reverts commit 63711c3cd337a0d22617579a904af07481139611, due to
breaking bots:

http://lab.llvm.org:8011/builders/clang-x64-ninja-win7/builds/11315
http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/10411/steps/test-check-all/logs/stdio

llvm-svn: 335194

6 years agoASan docs: no_sanitize("address") works on globals.
Evgeniy Stepanov [Thu, 21 Jun 2018 00:16:32 +0000 (00:16 +0000)]
ASan docs: no_sanitize("address") works on globals.

Summary: Mention that no_sanitize attribute can be used with globals.

Reviewers: alekseyshl

Subscribers: cfe-commits

Differential Revision: https://reviews.llvm.org/D48390

llvm-svn: 335193

6 years ago[WebAssembly] Error on mismatched function signature in final output
Sam Clegg [Thu, 21 Jun 2018 00:12:25 +0000 (00:12 +0000)]
[WebAssembly] Error on mismatched function signature in final output

During symbol resolution, emit warnings for function signature
mismatches.  During GC, if any mismatched symbol is marked as live
then generate an error.

This means that we only error out if the mismatch is written to the
final output.  i.e. if we would generate an invalid wasm file.

Differential Revision: https://reviews.llvm.org/D48394

llvm-svn: 335192

6 years agoWhen a dependent alignas is applied to a non-dependent typedef,
Richard Smith [Wed, 20 Jun 2018 23:36:55 +0000 (23:36 +0000)]
When a dependent alignas is applied to a non-dependent typedef,
prioritize the error for the bad subject over the error for the
dependent / non-dependent mismatch.

llvm-svn: 335191

6 years agoFix hmaptool cmake file to work on Windows
Bruno Cardoso Lopes [Wed, 20 Jun 2018 23:08:43 +0000 (23:08 +0000)]
Fix hmaptool cmake file to work on Windows

Unbreak a few windows buildbots:
http://lab.llvm.org:8011/builders/clang-x64-ninja-win7/builds/11315
http://lab.llvm.org:8011/builders/llvm-clang-x86_64-expensive-checks-win/builds/10411/steps/test-check-all/logs/stdio

llvm-svn: 335190

6 years agoUse cast instead of dyn_cast_or_null.
Akira Hatanaka [Wed, 20 Jun 2018 22:56:59 +0000 (22:56 +0000)]
Use cast instead of dyn_cast_or_null.

This addresses John's post-commit review feedback.

https://reviews.llvm.org/rC335021#inline-2038

llvm-svn: 335189

6 years ago[DWARF] Improved error reporting for range lists.
Wolfgang Pieb [Wed, 20 Jun 2018 22:56:37 +0000 (22:56 +0000)]
[DWARF] Improved error reporting for range lists.
Errors found processing the DW_AT_ranges attribute are propagated by lower level
routines and reported by their callers.

Reviewer: JDevlieghere

Differential Revision: https://reviews.llvm.org/D48344

llvm-svn: 335188

6 years ago[WebAssembly] Minor cleanup to test inputs. NFC.
Sam Clegg [Wed, 20 Jun 2018 22:52:33 +0000 (22:52 +0000)]
[WebAssembly] Minor cleanup to test inputs. NFC.

Update load-undefined.test such that it doesn't rely on
ret32 and ret64 having default visibility.

Split out from: https://reviews.llvm.org/D48394

Differential Revision: https://reviews.llvm.org/D48403

llvm-svn: 335187

6 years ago[WebAssembly] Update function signature mismatch error message. NFC.
Sam Clegg [Wed, 20 Jun 2018 22:45:03 +0000 (22:45 +0000)]
[WebAssembly] Update function signature mismatch error message. NFC.

We don't start our error messages with capital letters.

Split out from https://reviews.llvm.org/D48394

Differential Revision: https://reviews.llvm.org/D48400

llvm-svn: 335186

6 years ago[mips] Add microMIPS specific addressing patterns.
Simon Dardis [Wed, 20 Jun 2018 22:40:12 +0000 (22:40 +0000)]
[mips] Add microMIPS specific addressing patterns.

These are identical but use microMIPS instructions instead of MIPS instructions.

Also, flatten the 'let AdditionalPredicates = [InMicroMips]' by using the
ISA_MICROMIPS adjective. Add tests for constant materialization.

Reviewers: atanasyan, abeserminji, smaksimovic

Differential Revision: https://reviews.llvm.org/D48275

llvm-svn: 335185

6 years agoWarning for framework headers using double quote includes
Bruno Cardoso Lopes [Wed, 20 Jun 2018 22:11:59 +0000 (22:11 +0000)]
Warning for framework headers using double quote includes

Introduce -Wquoted-include-in-framework-header, which should fire a warning
whenever a quote include appears in a framework header and suggest a fix-it.
For instance, for header A.h added in the tests, this is how the warning looks
like:

./A.framework/Headers/A.h:2:10: warning: double-quoted include "A0.h" in framework header, expected angle-bracketed instead [-Wquoted-include-in-framework-header]
#include "A0.h"
         ^~~~~~
         <A/A0.h>
./A.framework/Headers/A.h:3:10: warning: double-quoted include "B.h" in framework header, expected angle-bracketed instead [-Wquoted-include-in-framework-header]
#include "B.h"
         ^~~~~
         <B.h>

This helps users to prevent frameworks from using local headers when in fact
they should be targetting system level ones.

The warning is off by default.

Differential Revision: https://reviews.llvm.org/D47157

rdar://problem/37077034

llvm-svn: 335184

6 years agoGeneralize MergeBlockIntoPredecessor. Replace uses of MergeBasicBlockIntoOnlyPred.
Alina Sbirlea [Wed, 20 Jun 2018 22:01:04 +0000 (22:01 +0000)]
Generalize MergeBlockIntoPredecessor. Replace uses of MergeBasicBlockIntoOnlyPred.

Summary:
Two utils methods have essentially the same functionality. This is an attempt to merge them into one.
1. lib/Transforms/Utils/Local.cpp : MergeBasicBlockIntoOnlyPred
2. lib/Transforms/Utils/BasicBlockUtils.cpp : MergeBlockIntoPredecessor

Prior to the patch:
1. MergeBasicBlockIntoOnlyPred
Updates either DomTree or DeferredDominance
Moves all instructions from Pred to BB, deletes Pred
Asserts BB has single predecessor
If address was taken, replace the block address with constant 1 (?)

2. MergeBlockIntoPredecessor
Updates DomTree, LoopInfo and MemoryDependenceResults
Moves all instruction from BB to Pred, deletes BB
Returns if doesn't have a single predecessor
Returns if BB's address was taken

After the patch:
Method 2. MergeBlockIntoPredecessor is attempting to become the new default:
Updates DomTree or DeferredDominance, and LoopInfo and MemoryDependenceResults
Moves all instruction from BB to Pred, deletes BB
Returns if doesn't have a single predecessor
Returns if BB's address was taken

Uses of MergeBasicBlockIntoOnlyPred that need to be replaced:

1. lib/Transforms/Scalar/LoopSimplifyCFG.cpp
Updated in this patch. No challenges.

2. lib/CodeGen/CodeGenPrepare.cpp
Updated in this patch.
  i. eliminateFallThrough is straightforward, but I added using a temporary array to avoid the iterator invalidation.
  ii. eliminateMostlyEmptyBlock(s) methods also now use a temporary array for blocks
Some interesting aspects:
  - Since Pred is not deleted (BB is), the entry block does not need updating.
  - The entry block was being updated with the deleted block in eliminateMostlyEmptyBlock. Added assert to make obvious that BB=SinglePred.
  - isMergingEmptyBlockProfitable assumes BB is the one to be deleted.
  - eliminateMostlyEmptyBlock(BB) does not delete BB on one path, it deletes its unique predecessor instead.
  - adding some test owner as subscribers for the interesting tests modified:
    test/CodeGen/X86/avx-cmp.ll
    test/CodeGen/AMDGPU/nested-loop-conditions.ll
    test/CodeGen/AMDGPU/si-annotate-cf.ll
    test/CodeGen/X86/hoist-spill.ll
    test/CodeGen/X86/2006-11-17-IllegalMove.ll

3. lib/Transforms/Scalar/JumpThreading.cpp
Not covered in this patch. It is the only use case using the DeferredDominance.
I would defer to Brian Rzycki to make this replacement.

Reviewers: chandlerc, spatel, davide, brzycki, bkramer, javed.absar

Subscribers: qcolombet, sanjoy, nemanjai, nhaehnle, jlebar, tpr, kbarton, RKSimon, wmi, arsenm, llvm-commits

Differential Revision: https://reviews.llvm.org/D48202

llvm-svn: 335183

6 years agoRelated to PR37768: improve diagnostics for class name shadowing.
Richard Smith [Wed, 20 Jun 2018 21:58:20 +0000 (21:58 +0000)]
Related to PR37768: improve diagnostics for class name shadowing.

Diagnose the name of the class being shadowed by using declarations, and
improve the diagnostics for the case where the name of the class is
shadowed by a non-static data member in a class with constructors.  In
the latter case, we now always give the "member with the same name as
its class" diagnostic regardless of the relative order of the member and
the constructor, rather than giving an inscrutible diagnostic if the
constructor appears second.

llvm-svn: 335182

6 years agoFix WasmEHFuncInfo.h to include what it uses
Bruno Cardoso Lopes [Wed, 20 Jun 2018 21:43:49 +0000 (21:43 +0000)]
Fix WasmEHFuncInfo.h to include what it uses

This fixes clang+llvm build with Modules and local submodule visibility.

llvm-svn: 335181

6 years agoImprove SBThread's stepping API using SBError parameter.
Alexander Polyakov [Wed, 20 Jun 2018 21:43:16 +0000 (21:43 +0000)]
Improve SBThread's stepping API using SBError parameter.

Summary: The new methods will allow to get error messages from stepping API.

Reviewers: aprantl, clayborg, labath, jingham

Reviewed By: aprantl, clayborg, jingham

Subscribers: apolyakov, labath, jingham, clayborg, lemo, lldb-commits

Differential Revision: https://reviews.llvm.org/D47991

llvm-svn: 335180

6 years ago[MemorySSA] Add convenience APIs in updater to avoid needing MSSA.
Alina Sbirlea [Wed, 20 Jun 2018 21:30:29 +0000 (21:30 +0000)]
[MemorySSA] Add convenience APIs in updater to avoid needing MSSA.

Summary:
Ideally passes should not need to pass MSSA around and do all updates through the updater.
Add convenience APIs to help with that.

Reviewers: george.burgess.iv

Subscribers: sanjoy, jlebar, llvm-commits, Prazek

Differential Revision: https://reviews.llvm.org/D48334

llvm-svn: 335179

6 years agoRemove myself from the release testers list. (NFC)
Simon Dardis [Wed, 20 Jun 2018 21:25:50 +0000 (21:25 +0000)]
Remove myself from the release testers list. (NFC)

llvm-svn: 335178

6 years agoAdd python tool to dump and construct header maps
Bruno Cardoso Lopes [Wed, 20 Jun 2018 21:16:37 +0000 (21:16 +0000)]
Add python tool to dump and construct header maps

Header maps are binary files used by Xcode, which are used to map
header names or paths to other locations. Clang has support for
those since its inception, but there's not a lot of header map
testing around.

Since it's a binary format, testing becomes pretty much brittle
and its hard to even know what's inside if you don't have the
appropriate tools.

Add a python based tool that allows creating and dumping header
maps based on a json description of those. While here, rewrite
tests to use the tool and remove the binary files from the tree.

This tool was initially written by Daniel Dunbar.

Differential Revision: https://reviews.llvm.org/D46485

rdar://problem/39994722

llvm-svn: 335177

6 years ago[Dominators] Simplify child lists and make them deterministic
Benjamin Kramer [Wed, 20 Jun 2018 21:12:59 +0000 (21:12 +0000)]
[Dominators] Simplify child lists and make them deterministic

This fixes an extremely subtle non-determinism that can only be
triggered by an unfortunate alignment of passes. In my case:

- JumpThreading does large dominator tree updates
- CorrelatedValuePropagation preserves domtree now
- LICM codegen depends on the order of children on domtree nodes

The last part is non-deterministic if the update was stored in a set.
But it turns out that the set is completely unnecessary, updates are
deduplicated at an earlier stage so we can just use a vector, which is
both more efficient and doesn't destroy the input ordering.

I didn't manage to get the 240 MB IR file reduced enough, triggering
this bug requires a lot of jump threading, so landing this without a
test case.

Differential Revision: https://reviews.llvm.org/D48392

llvm-svn: 335176

6 years ago[MS] Make sure __GetExceptionInfo works on types with no linkage
Reid Kleckner [Wed, 20 Jun 2018 21:12:20 +0000 (21:12 +0000)]
[MS] Make sure __GetExceptionInfo works on types with no linkage

Fixes PR36327

llvm-svn: 335175

6 years ago[MemorySSA] Verify Phi incoming blocks are block predecessors.
Alina Sbirlea [Wed, 20 Jun 2018 21:06:13 +0000 (21:06 +0000)]
[MemorySSA] Verify Phi incoming blocks are block predecessors.

Summary: Make the MemorySSA verify also check that all Phi incoming blocks are block predecessors.

Reviewers: george.burgess.iv

Subscribers: sanjoy, jlebar, Prazek, llvm-commits

Differential Revision: https://reviews.llvm.org/D48333

llvm-svn: 335174

6 years ago[X86] Use setcc ISD opcode for AVX512 integer comparisons all the way to isel
Craig Topper [Wed, 20 Jun 2018 21:05:02 +0000 (21:05 +0000)]
[X86] Use setcc ISD opcode for AVX512 integer comparisons all the way to isel

I don't believe there is any real reason to have separate X86 specific opcodes for vector compares. Setcc has the same behavior just uses a different encoding for the condition code.

I had to change the CondCodeAction for SETLT and SETLE to prevent some transforms from changing SETGT lowering.

Differential Revision: https://reviews.llvm.org/D43608

llvm-svn: 335173

6 years ago[CMake] Convert paths to the right form in standalone builds on Windows
Martin Storsjo [Wed, 20 Jun 2018 21:03:34 +0000 (21:03 +0000)]
[CMake] Convert paths to the right form in standalone builds on Windows

The paths output from llvm-config --cmakedir and from clang
--print-libgcc-file-name can contain backslashes, while CMake
can't handle the paths in this form.

This matches what compiler-rt already does (since SVN r203789
and r293195).

Differential Revision: https://reviews.llvm.org/D48356

llvm-svn: 335172

6 years ago[CMake] Convert paths to the right form in standalone builds on Windows
Martin Storsjo [Wed, 20 Jun 2018 20:59:18 +0000 (20:59 +0000)]
[CMake] Convert paths to the right form in standalone builds on Windows

The paths output from llvm-config --cmakedir and from clang
--print-libgcc-file-name can contain backslashes, while CMake
can't handle the paths in this form.

This matches what compiler-rt already does (since SVN r203789
and r293195).

Differential Revision: https://reviews.llvm.org/D48355

llvm-svn: 335171

6 years ago[SLPVectorizer] Provide InstructionsState down the BoUpSLP vectorization call tree
Simon Pilgrim [Wed, 20 Jun 2018 20:54:52 +0000 (20:54 +0000)]
[SLPVectorizer] Provide InstructionsState down the BoUpSLP vectorization call tree

As described in D48359, this patch pushes InstructionsState down the BoUpSLP call hierarchy instead of the corresponding raw OpValue. This makes it easier to track the alternate opcode etc. and avoids us having to call getAltOpcode which makes it difficult to support more than one alternate opcode.

Differential Revision: https://reviews.llvm.org/D48382

llvm-svn: 335170

6 years ago[CMake] Convert paths to the right form in standalone builds on Windows
Martin Storsjo [Wed, 20 Jun 2018 20:53:19 +0000 (20:53 +0000)]
[CMake] Convert paths to the right form in standalone builds on Windows

The paths output from llvm-config --cmakedir and from clang
--print-libgcc-file-name can contain backslashes, while CMake
can't handle the paths in this form.

This matches what compiler-rt already does (since SVN r203789
and r293195).

Differential Revision: https://reviews.llvm.org/D48353

llvm-svn: 335169

6 years ago[CUDA] Removed unused __nvvm_* builtins with non-generic pointers.
Artem Belevich [Wed, 20 Jun 2018 20:34:04 +0000 (20:34 +0000)]
[CUDA] Removed unused __nvvm_* builtins with non-generic pointers.

They were hot even hooked into CGBuiltin's machinery. Even if they were,
CUDA does not support AS-specific pointers, so there would be no legal way
no way to call these builtins.

This came up in D47154.

Differential Revision: https://reviews.llvm.org/D47845

llvm-svn: 335168

6 years agoAllow binop C1, (select cc, CF, CT) -> select folding
Stanislav Mekhanoshin [Wed, 20 Jun 2018 20:24:20 +0000 (20:24 +0000)]
Allow binop C1, (select cc, CF, CT) -> select folding

Previously this folding was done only if select is a first operand.
However, for non-commutative operations constant may go before
select.

Differential Revision: https://reviews.llvm.org/D48223

llvm-svn: 335167

6 years ago[TSan] fix build and couple of unit tests on FreeBSD
David Carlier [Wed, 20 Jun 2018 20:17:44 +0000 (20:17 +0000)]
[TSan] fix build and couple of unit tests on FreeBSD

Similarly to Msan adding -pie provokes linkage issue, was spotted with pie_test.cc
Set to XFAIL for couple of unit tests.

Reviewers: vitalybuka, krytarowski, dim

Reviewed By: vitalybuka

Differential Revision: https://reviews.llvm.org/D48317

llvm-svn: 335166

6 years ago[InstCombine] fix typo in test comment; NFC
Sanjay Patel [Wed, 20 Jun 2018 20:16:45 +0000 (20:16 +0000)]
[InstCombine] fix typo in test comment; NFC

llvm-svn: 335165

6 years ago[Lsan] intercept thr_exit on FreeBSD
David Carlier [Wed, 20 Jun 2018 20:13:25 +0000 (20:13 +0000)]
[Lsan] intercept thr_exit on FreeBSD

Intercepts thr_exit call on FreeBSD.
Disable pthread key workflow.
The pthread key create approach does not function under FreeBSD as the libpthread is not initialised enough at this stage.

Reviewers: vitalybuka, krytarowski, dim

Reviewed By: vitalybuka

Differential Revision: https://reviews.llvm.org/D48268

llvm-svn: 335164

6 years agoRemove some instances of manual UUID pretty-printing
Pavel Labath [Wed, 20 Jun 2018 20:13:04 +0000 (20:13 +0000)]
Remove some instances of manual UUID pretty-printing

Identical functionality is already offered by the UUID::getAsString
method.

llvm-svn: 335163

6 years ago[mips] Correct predicates for loads, bit manipulation instructions and some pseudos
Simon Dardis [Wed, 20 Jun 2018 19:59:58 +0000 (19:59 +0000)]
[mips] Correct predicates for loads, bit manipulation instructions and some pseudos

Additionally, correct the definition of the rdhwr instruction.

Reviewers: atanasyan, abeserminji, smaksimovic

Differential Revision: https://reviews.llvm.org/D48216

llvm-svn: 335162

6 years agoAMDGPU: Fix scalar_to_vector for v4i16/v4f16
Matt Arsenault [Wed, 20 Jun 2018 19:45:48 +0000 (19:45 +0000)]
AMDGPU: Fix scalar_to_vector for v4i16/v4f16

llvm-svn: 335161

6 years agoAMDGPU: Fix missing C++ mode comment
Matt Arsenault [Wed, 20 Jun 2018 19:45:40 +0000 (19:45 +0000)]
AMDGPU: Fix missing C++ mode comment

llvm-svn: 335160

6 years agoFixed test in prior build where FileCheck tried to match against
Leonard Chan [Wed, 20 Jun 2018 19:34:05 +0000 (19:34 +0000)]
Fixed test in prior build where FileCheck tried to match against
`common` when declaring a global variable when we primarily care about
the value assigned in the test.

llvm-svn: 335159

6 years ago[Hexagon] Replace .ll test for expanding post-ra pesudos with .mir
Krzysztof Parzyszek [Wed, 20 Jun 2018 19:22:27 +0000 (19:22 +0000)]
[Hexagon] Replace .ll test for expanding post-ra pesudos with .mir

llvm-svn: 335158

6 years ago[IR] add/use isIntDivRem convenience function
Sanjay Patel [Wed, 20 Jun 2018 19:02:17 +0000 (19:02 +0000)]
[IR] add/use isIntDivRem convenience function

There are more existing potential users of this,
but I've limited this patch to the first couple
that I found to minimize typo risk.

llvm-svn: 335157

6 years ago[PM/LoopUnswitch] Support partial trivial unswitching.
Chandler Carruth [Wed, 20 Jun 2018 18:57:07 +0000 (18:57 +0000)]
[PM/LoopUnswitch] Support partial trivial unswitching.

The idea of partial unswitching is to take a *part* of a branch's
condition that is loop invariant and just unswitching that part. This
primarily makes sense with i1 conditions of branches as opposed to
switches. When dealing with i1 conditions, we can easily extract loop
invariant inputs to a a branch and unswitch them to test them entirely
outside the loop.

As part of this, we now create much more significant cruft in the loop
body, so this relies on adding cleanup passes to the loop pipeline and
revisiting unswitched loops to do that cleanup before continuing to
process them.

This already appears to be more powerful at unswitching than the old
loop unswitch pass, and so I'd appreciate pretty careful review in case
I'm just missing some correctness checks. The `LIV-loop-condition` test
case is not unswitched by the old unswitch pass, but is with this pass.

Thanks to Sanjoy and Fedor for the review!

Differential Revision: https://reviews.llvm.org/D46706

llvm-svn: 335156

6 years agoFixed test that failed when checking what variable a value was stored
Leonard Chan [Wed, 20 Jun 2018 18:48:05 +0000 (18:48 +0000)]
Fixed test that failed when checking what variable a value was stored
in for fixed point types.

llvm-svn: 335155

6 years ago[RISCV] Accept fmv.s.x and fmv.x.s as mnemonic aliases for fmv.w.x and fmv.x.w
Alex Bradbury [Wed, 20 Jun 2018 18:42:25 +0000 (18:42 +0000)]
[RISCV] Accept fmv.s.x and fmv.x.s as mnemonic aliases for fmv.w.x and fmv.x.w

These instructions were renamed in version 2.2 of the user-level ISA spec, but
the old name should also be accepted by standard tools.

llvm-svn: 335154

6 years ago[MachineOutliner] Add debug info test for the outliner
Jessica Paquette [Wed, 20 Jun 2018 18:41:11 +0000 (18:41 +0000)]
[MachineOutliner] Add debug info test for the outliner

The outliner emits debug info. Add a test that outlines a function
and uses llvm-dwarfdump to check the emitted DWARF for correctness.

llvm-svn: 335153

6 years ago[Local] Generalize insertReplacementDbgValues, NFC
Vedant Kumar [Wed, 20 Jun 2018 18:40:14 +0000 (18:40 +0000)]
[Local] Generalize insertReplacementDbgValues, NFC

This utility should operate on Values, not Instructions. While I'm here,
I've also made it possible to skip emitting replacement dbg.values for
certain debug users (by having RewriteExpr return nullptr).

llvm-svn: 335152

6 years ago[InstCombine] add vector select of binops tests (PR37806)
Sanjay Patel [Wed, 20 Jun 2018 17:48:43 +0000 (17:48 +0000)]
[InstCombine] add vector select of binops tests (PR37806)

These represent the most basic requested transform - a matching
operand and 2 constant operands.

llvm-svn: 335151

6 years ago[PredicateInfo] Order instructions in different BBs by DFSNumIn.
Florian Hahn [Wed, 20 Jun 2018 17:42:01 +0000 (17:42 +0000)]
[PredicateInfo] Order instructions in different BBs by DFSNumIn.

Using OrderedInstructions::dominates as comparator for instructions in
BBs without dominance relation can cause a non-deterministic order
between such instructions. That in turn can cause us to materialize
copies in a non-deterministic order. While this does not effect
correctness, it causes some minor non-determinism in the final generated
code, because values have slightly different labels.

Without this patch, running -print-predicateinfo on a reasonably large
module produces slightly different output on each run.

This patch uses the dominator trees DFSInNum to order instruction from
different BBs, which should enforce a deterministic ordering and
guarantee that dominated instructions come after the instructions that
dominate them.

Reviewers: dberlin, efriedma, davide

Reviewed By: efriedma

Differential Revision: https://reviews.llvm.org/D48230

llvm-svn: 335150

6 years agoMake test sources compatible with android+libcxx+modules
Pavel Labath [Wed, 20 Jun 2018 17:32:48 +0000 (17:32 +0000)]
Make test sources compatible with android+libcxx+modules

In a modules build, android is very picky about which symbols are
visible after including libc++ headers (e.g. <cstdio> defines only
std::printf and not ::printf).

This consolidates the tests where this was an issue to always include
the <c???> version of the headers and prefixes the symbols with std:: as
necessary.

Apart from that, there is no functional change in the tests.

llvm-svn: 335149

6 years ago[Fixed Point Arithmetic] Fixed Point Precision Bits and Fixed Point Literals
Leonard Chan [Wed, 20 Jun 2018 17:19:40 +0000 (17:19 +0000)]
[Fixed Point Arithmetic] Fixed Point Precision Bits and Fixed Point Literals

This diff includes the logic for setting the precision bits for each primary fixed point type in the target info and logic for initializing a fixed point literal.

Fixed point literals are declared using the suffixes

```
hr: short _Fract
uhr: unsigned short _Fract
r: _Fract
ur: unsigned _Fract
lr: long _Fract
ulr: unsigned long _Fract
hk: short _Accum
uhk: unsigned short _Accum
k: _Accum
uk: unsigned _Accum
```
Errors are also thrown for illegal literal values

```
unsigned short _Accum u_short_accum = 256.0uhk;   // expected-error{{the integral part of this literal is too large for this unsigned _Accum type}}
```

Differential Revision: https://reviews.llvm.org/D46915

llvm-svn: 335148

6 years ago[Sanitizers] Remove OOM/BadRequest allocator error handling policies.
Alex Shlyapnikov [Wed, 20 Jun 2018 17:10:33 +0000 (17:10 +0000)]
[Sanitizers] Remove OOM/BadRequest allocator error handling policies.

Summary:
Remove the generic error nadling policies and handle each allocator error
explicitly. Although more verbose, it allows for more comprehensive, precise
and actionable allocator related failure reports.

This finishes up the series of changes of the particular sanitizer
allocators, improves the internal allocator error reporting and removes
now unused policies.

Reviewers: vitalybuka, cryptoad

Subscribers: kubamracek, delcypher, #sanitizers, llvm-commits

Differential Revision: https://reviews.llvm.org/D48328

llvm-svn: 335147

6 years ago[DWARF] Don't keep a ref to possibly stack allocated data.
Paul Robinson [Wed, 20 Jun 2018 17:08:46 +0000 (17:08 +0000)]
[DWARF] Don't keep a ref to possibly stack allocated data.

llvm-svn: 335146

6 years agoIRMover: Account for matching types present across modules
Vlad Tsyrklevich [Wed, 20 Jun 2018 16:50:56 +0000 (16:50 +0000)]
IRMover: Account for matching types present across modules

Summary:
Due to uniqueing of DICompositeTypes, it's possible for a type from one
module to be loaded into another earlier module without being renamed.
Then when the defining module is being IRMoved, the type can be used as
a Mapping destination before being loaded, such that when it's requested
using TypeMapTy::get() it will fail with an assertion that the type is a
source type when it's actually a type in both the source and
destination modules. Correctly handle that case by allowing a non-opaque
non-literal struct type be present in both modules.

Fix for PR37684.

Reviewers: pcc, tejohnson

Reviewed By: pcc, tejohnson

Subscribers: tobiasvk, mehdi_amini, steven_wu, llvm-commits, kcc

Differential Revision: https://reviews.llvm.org/D47898

llvm-svn: 335145

6 years ago[Local] Add a utility to insert replacement dbg.values, NFC
Vedant Kumar [Wed, 20 Jun 2018 16:50:25 +0000 (16:50 +0000)]
[Local] Add a utility to insert replacement dbg.values, NFC

The purpose of this utility is to make it easier for optimizations to
insert replacement dbg.values for instructions they are deleting. This
is useful in situations where salvageDebugInfo is inapplicable, say,
because the new dbg.value cannot refer to an operand of the dying value.

The utility is called insertReplacementDbgValues.

It assumes that the instruction 'From' is going to be deleted, and
inserts replacement dbg.values for each debug user of 'From'. The
newly-inserted dbg.values refer to 'To' instead of 'From'. Each
replacement dbg.value has the same location and variable as the debug
user it replaces, has a DIExpression determined by the result of
'RewriteExpr' applied to an old debug user of 'From', and is placed
before 'InsertBefore'.

This should simplify future patches, like D48331.

llvm-svn: 335144

6 years agoRemove a redundant initialization. NFC
Paul Robinson [Wed, 20 Jun 2018 16:12:03 +0000 (16:12 +0000)]
Remove a redundant initialization. NFC

llvm-svn: 335143

6 years ago[SLPVectorizer] Move isOneOf after InstructionsState type. NFCI.
Simon Pilgrim [Wed, 20 Jun 2018 16:11:00 +0000 (16:11 +0000)]
[SLPVectorizer] Move isOneOf after InstructionsState type. NFCI.

A future patch will have isOneOf use InstructionsState.

llvm-svn: 335142

6 years ago[DAG] Don't map a TableId to itself in the ReplacedValues map
Bjorn Pettersson [Wed, 20 Jun 2018 16:06:09 +0000 (16:06 +0000)]
[DAG] Don't map a TableId to itself in the ReplacedValues map

Summary:
Found some regressions (infinite loop in DAGTypeLegalizer::RemapId)
after r334880. This patch makes sure that we do map a TableId to
itself.

Reviewers: niravd

Reviewed By: niravd

Subscribers: llvm-commits

Differential Revision: https://reviews.llvm.org/D48364

llvm-svn: 335141

6 years ago[ELF][MIPS] Fill a primary-GOT as much as possible
Simon Atanasyan [Wed, 20 Jun 2018 15:58:48 +0000 (15:58 +0000)]
[ELF][MIPS] Fill a primary-GOT as much as possible

While building a Global Offset Table try to fill the primary GOT as much
as possible because the primary GOT can be accessed in the most
effective way. If it is not possible, try to fill the last GOT in the
multi-GOT list, and finally create a new GOT if both attempts failed.

llvm-svn: 335140

6 years agoSimplify. No behavior change.
Nico Weber [Wed, 20 Jun 2018 15:57:38 +0000 (15:57 +0000)]
Simplify. No behavior change.

llvm-svn: 335139

6 years agominor: fixed typo in debug print
Andrey Churbanov [Wed, 20 Jun 2018 15:54:11 +0000 (15:54 +0000)]
minor: fixed typo in debug print

llvm-svn: 335138

6 years ago[DAG] Fix and-mask folding when narrowing loads.
Nirav Dave [Wed, 20 Jun 2018 15:36:29 +0000 (15:36 +0000)]
[DAG] Fix and-mask folding when narrowing loads.

Summary:
Check that and masks are strictly smaller than implicit mask from
narrowed load.

Fixes PR37820.

Reviewers: samparker, RKSimon, nemanjai

Subscribers: hiraditya, llvm-commits

Differential Revision: https://reviews.llvm.org/D48335

llvm-svn: 335137

6 years ago[LIT] Enable testing of LLVM gold plugin on Mac OS X
Eugene Leviant [Wed, 20 Jun 2018 15:32:47 +0000 (15:32 +0000)]
[LIT] Enable testing of LLVM gold plugin on Mac OS X

Differential revision: https://reviews.llvm.org/D48350

llvm-svn: 335136

6 years ago[WebAssembly] Update know failures for the wasm waterfall
Sam Clegg [Wed, 20 Jun 2018 15:17:12 +0000 (15:17 +0000)]
[WebAssembly] Update know failures for the wasm waterfall

Summary:
The waterfall no longer builds .s files and no longers uses
the wasm-o when it builds object files.

Subscribers: dschuff, jgravelle-google, aheejin, sunfish, llvm-commits

Differential Revision: https://reviews.llvm.org/D48371

llvm-svn: 335135

6 years ago[SLPVectorizer] Use InstructionsState to record AltOpcode
Simon Pilgrim [Wed, 20 Jun 2018 15:13:40 +0000 (15:13 +0000)]
[SLPVectorizer] Use InstructionsState to record AltOpcode

This is part of a move towards generalizing the alternate opcode mechanism and not just supporting (F)Add/(F)Sub counterparts.

The patch embeds the AltOpcode in the InstructionsState instead of calling getAltOpcode so often.

I'm hoping to eventually remove all uses of getAltOpcode and handle alternate opcode selection entirely within getSameOpcode, that will require us to use InstructionsState throughout the BoUpSLP call hierarchy (similar to some of the changes in D28907), which I will begin in future patches.

Differential Revision: https://reviews.llvm.org/D48359

llvm-svn: 335134

6 years agoFix-up for r335123: Set default apple_platform to "osx".
Kuba Mracek [Wed, 20 Jun 2018 14:59:16 +0000 (14:59 +0000)]
Fix-up for r335123: Set default apple_platform to "osx".

llvm-svn: 335133

6 years agoMake sure TestNumThreads works with libc++
Pavel Labath [Wed, 20 Jun 2018 14:54:34 +0000 (14:54 +0000)]
Make sure TestNumThreads works with libc++

The problem was that with libc++ the std::unique_lock declaration was
completely inlined, so there was no line table entry in the main.cpp
file to set a breakpoint on. Therefore, the breakpoint got moved to the
next line, but that meant the test would deadlock as the thread would
stop with the lock already held.

I fix that issue by adding a dummy statement before the std::unique_lock
line to anchor the breakpoint.

I think this should fix the issue because of which this test was
disabled on darwin, but someone should verify that before enabling it.

llvm-svn: 335132

6 years ago[llvm-mca] use APint::operator[] to obtain the bit value. NFC
Andrea Di Biagio [Wed, 20 Jun 2018 14:30:17 +0000 (14:30 +0000)]
[llvm-mca] use APint::operator[] to obtain the bit value. NFC

llvm-svn: 335131

6 years ago[SLPVectorizer] Relax "alternate" opcode vectorisation to work with any SK_Select...
Simon Pilgrim [Wed, 20 Jun 2018 14:26:28 +0000 (14:26 +0000)]
[SLPVectorizer] Relax "alternate" opcode vectorisation to work with any SK_Select shuffle pattern

D47985 saw the old SK_Alternate 'alternating' shuffle mask replaced with the SK_Select mask which accepts either input operand for each lane, equivalent to a vector select with a constant condition operand.

This patch updates SLPVectorizer to make full use of this SK_Select shuffle pattern by removing the 'isOdd()' limitation.

The AArch64 regression will be fixed by D48172.

Differential Revision: https://reviews.llvm.org/D48174

llvm-svn: 335130

6 years ago[InstSimplify] Fix missed optimization in simplifyUnsignedRangeCheck()
Sanjay Patel [Wed, 20 Jun 2018 14:22:49 +0000 (14:22 +0000)]
[InstSimplify] Fix missed optimization in simplifyUnsignedRangeCheck()

For both operands are unsigned, the following optimizations are valid, and missing:

   1. X > Y && X != 0 --> X > Y
   2. X > Y || X != 0 --> X != 0
   3. X <= Y || X != 0 --> true
   4. X <= Y || X == 0 --> X <= Y
   5. X > Y && X == 0 --> false

unsigned foo(unsigned x, unsigned y) { return x > y && x != 0; }
should fold to x > y, but I found we haven't done it right now.
besides, unsigned foo(unsigned x, unsigned y) { return x < y && y != 0; }
Has been folded to x < y, so there may be a bug.

Patch by: Li Jia He!

Differential Revision: https://reviews.llvm.org/D47922

llvm-svn: 335129

6 years ago[InstSimplify] Add tests for missed optimizations in simplifyUnsignedRangeCheck ...
Sanjay Patel [Wed, 20 Jun 2018 14:03:13 +0000 (14:03 +0000)]
[InstSimplify] Add tests for missed optimizations in simplifyUnsignedRangeCheck (NFC)

These are the baseline tests for the functional change in D47922.

Patch by Li Jia He!

Differential Revision: https://reviews.llvm.org/D48000

llvm-svn: 335128

6 years ago[RISCV] Add InstAlias definitions for fgt.{s|d}, fge.{s|d}
Alex Bradbury [Wed, 20 Jun 2018 14:03:02 +0000 (14:03 +0000)]
[RISCV] Add InstAlias definitions for fgt.{s|d}, fge.{s|d}

These are produced by GCC and supported by GAS, but not currently contained in
the pseudoinstruction listing in the RISC-V ISA manual.

llvm-svn: 335127

6 years ago[X86] Adding a test for PR37879
Mikhail Dvoretckii [Wed, 20 Jun 2018 14:01:57 +0000 (14:01 +0000)]
[X86] Adding a test for PR37879

llvm-svn: 335126

6 years ago[ADT] Allow llvm::hash_code as DenseMap key.
Sam McCall [Wed, 20 Jun 2018 13:56:25 +0000 (13:56 +0000)]
[ADT] Allow llvm::hash_code as DenseMap key.

Summary:
This is useful when hash collisions are unlikely and acceptable, e.g. in clangd
completion ranking.

Reviewers: ioeric

Subscribers: ilya-biryukov, llvm-commits

Differential Revision: https://reviews.llvm.org/D48361

llvm-svn: 335125

6 years ago[Hexagon] Remove 'T' from HasVNN predicates, NFC
Krzysztof Parzyszek [Wed, 20 Jun 2018 13:56:09 +0000 (13:56 +0000)]
[Hexagon] Remove 'T' from HasVNN predicates, NFC

Patch by Sumanth Gundapaneni.

llvm-svn: 335124

6 years ago[sanitizer] Unify and generalize Apple platforms in CMake and lit test configs
Kuba Mracek [Wed, 20 Jun 2018 13:33:42 +0000 (13:33 +0000)]
[sanitizer] Unify and generalize Apple platforms in CMake and lit test configs

There's more platforms than just "ios" and "iossim" that we should support, and adding more lit config variables for each platform isn't great. Let's generalize and have a single value that determines what the platform under test is.

Differential Revision: https://reviews.llvm.org/D48309

llvm-svn: 335123