platform/upstream/llvm.git
8 years ago[SCEV] Try to reuse existing value during SCEV expansion
Wei Mi [Thu, 4 Feb 2016 01:27:38 +0000 (01:27 +0000)]
[SCEV] Try to reuse existing value during SCEV expansion

Current SCEV expansion will expand SCEV as a sequence of operations
and doesn't utilize the value already existed. This will introduce
redundent computation which may not be cleaned up throughly by
following optimizations.

This patch introduces an ExprValueMap which is a map from SCEV to the
set of equal values with the same SCEV. When a SCEV is expanded, the
set of values is checked and reused whenever possible before generating
a sequence of operations.

The original commit triggered regressions in Polly tests. The regressions
exposed two problems which have been fixed in current version.

1. Polly will generate a new function based on the old one. To generate an
instruction for the new function, it builds SCEV for the old instruction,
applies some tranformation on the SCEV generated, then expands the transformed
SCEV and insert the expanded value into new function. Because SCEV expansion
may reuse value cached in ExprValueMap, the value in old function may be
inserted into new function, which is wrong.
   In SCEVExpander::expand, there is a logic to check the cached value to
be used should dominate the insertion point. However, for the above
case, the check always passes. That is because the insertion point is
in a new function, which is unreachable from the old function. However
for unreachable node, DominatorTreeBase::dominates thinks it will be
dominated by any other node.
   The fix is to simply add a check that the cached value to be used in
expansion should be in the same function as the insertion point instruction.

2. When the SCEV is of scConstant type, expanding it directly is cheaper than
reusing a normal value cached. Although in the cached value set in ExprValueMap,
there is a Constant type value, but it is not easy to find it out -- the cached
Value set is not sorted according to the potential cost. Existing reuse logic
in SCEVExpander::expand simply chooses the first legal element from the cached
value set.
   The fix is that when the SCEV is of scConstant type, don't try the reuse
logic. simply expand it.

Differential Revision: http://reviews.llvm.org/D12090

llvm-svn: 259736

8 years agoFix undefined behavior when compiling in C++14 mode (with sized deletion
Richard Smith [Thu, 4 Feb 2016 01:21:16 +0000 (01:21 +0000)]
Fix undefined behavior when compiling in C++14 mode (with sized deletion
enabled): ensure that we do not invoke the sized deallocator for MemoryBuffer
subclasses that have tail-allocated data.

llvm-svn: 259735

8 years agoFix predefine for __NSConstantString struct type
Ben Langmuir [Thu, 4 Feb 2016 00:55:24 +0000 (00:55 +0000)]
Fix predefine for __NSConstantString struct type

Per review feedback the name was wrong and it can be used outside
Objective-C.

Unfortunately, making the internal struct visible broke some ASTMatchers
tests that assumed that the first record decl would be from user code,
rather than a builtin type.  I'm worried that this will also affect
users' code.  So this patch adds a typedef to wrap the internal struct
and only makes the typedef visible to namelookup.  This is sufficient to
allow the ASTReader to merge the decls we need without making the struct
itself visible.

rdar://problem/24425801

llvm-svn: 259734

8 years ago[codeview] Don't attempt a cross-section label diff
Reid Kleckner [Thu, 4 Feb 2016 00:21:42 +0000 (00:21 +0000)]
[codeview] Don't attempt a cross-section label diff

This only comes up when we're trying to find the next .cv_loc label.

Fixes PR26467

llvm-svn: 259733

8 years ago[libFuzzer] hot fix a test
Kostya Serebryany [Thu, 4 Feb 2016 00:12:28 +0000 (00:12 +0000)]
[libFuzzer] hot fix a test

llvm-svn: 259732

8 years ago[libFuzzer] don't write the test unit when a leak is detected (since we don't know...
Kostya Serebryany [Thu, 4 Feb 2016 00:02:17 +0000 (00:02 +0000)]
[libFuzzer] don't write the test unit when a leak is detected (since we don't know which unit causes the leak)

llvm-svn: 259731

8 years ago[SimplifyCFG] Fix for "endless" loop after dead code removal (Alternative to
Gerolf Hoflehner [Wed, 3 Feb 2016 23:54:25 +0000 (23:54 +0000)]
[SimplifyCFG] Fix for "endless" loop after dead code removal (Alternative to
D16251)

Summary:
This is a simpler fix to the problem than the dominator approach in
http://reviews.llvm.org/D16251. It adds only values into the gather() while loop
that have been seen before.

The actual endless loop is in the constant compare gather() routine in
Utils/SimplifyCFG.cpp. The same value ret.0.off0.i is pushed back into the
queue:
%.ret.0.off0.i = or i1 %.ret.0.off0.i, %cmp10.i

Here is what happens at the IR level:

for.cond.i:                                       ; preds = %if.end6.i,
%if.end.i54
%ix.0.i = phi i32 [ 0, %if.end.i54 ], [ %inc.i55, %if.end6.i ]
%ret.0.off0.i = phi i1 [false, %if.end.i54], [%.ret.0.off0.i, %if.end6.i] <<<
%cmp2.i = icmp ult i32 %ix.0.i, %11
br i1 %cmp2.i, label %for.body.i, label %LBJ_TmpSimpleNeedExt.exit

if.end6.i:                                        ; preds = %for.body.i
%cmp10.i = icmp ugt i32 %conv.i, %add9.i
%.ret.0.off0.i = or i1 %ret.0.off0.i, %cmp10.i <<<

When if.end.i54 gets eliminated which removes the definition of ret.0.off0.i.
The result is the expression %.ret.0.off0.i = or i1 %.ret.0.off0.i, %cmp10.i
(Note the first ‘or’ operand is now %.ret.0.off0.i, and *NOT* %ret.0.off0.i).
And
now there is use of .ret.0.off0.i before a definition which triggers the
“endless” loop in gather():

while(!DFT.empty()) {

    V = DFT.pop_back_val();   // V is .ret.0.off0.i

    if (Instruction *I = dyn_cast<Instruction>(V)) {
      // If it is a || (or && depending on isEQ), process the operands.
      if (I->getOpcode() == (isEQ ? Instruction::Or : Instruction::And)) {
        DFT.push_back(I->getOperand(1));  // This is now .ret.0.off0.i also
        DFT.push_back(I->getOperand(0));

        continue; // “endless loop” for .ret.0.off0.i
      }

Reviewers: reames, ahatanak

Subscribers: llvm-commits

Differential Revision: http://reviews.llvm.org/D16839

llvm-svn: 259730

8 years agoAdd support for -sdk_version cmdline option.
Pete Cooper [Wed, 3 Feb 2016 23:39:05 +0000 (23:39 +0000)]
Add support for -sdk_version cmdline option.

This option is emitted in the min_version load commands.

Note, there's currently a difference in behaviour compared to ld64 in
that we emit a warning if we generate a min_version load command and
didn't give an sdk_version.  We need to decide what the correct behaviour
is here as its possible we want to emit an error and force clients to
provide the option.

llvm-svn: 259729

8 years agoBump DiagnosticSemaKinds count; we're close to hitting it.
Manman Ren [Wed, 3 Feb 2016 23:35:29 +0000 (23:35 +0000)]
Bump DiagnosticSemaKinds count; we're close to hitting it.

 $ grep '= DIAG_START_SEMA' include/clang/Basic/DiagnosticIDs.h
       DIAG_START_ANALYSIS      = DIAG_START_SEMA            + 3000
 $ grep 'def ' include/clang/Basic/DiagnosticSemaKinds.td | wc -l
       2994

llvm-svn: 259728

8 years ago[InstrProfiling] Fix a comment (NFC)
Vedant Kumar [Wed, 3 Feb 2016 23:22:43 +0000 (23:22 +0000)]
[InstrProfiling] Fix a comment (NFC)

llvm-svn: 259727

8 years agoUnify the target opcode enum in TargetOpcodes.h and the FixedInstrs array in
David L Kreitzer [Wed, 3 Feb 2016 23:17:32 +0000 (23:17 +0000)]
Unify the target opcode enum in TargetOpcodes.h and the FixedInstrs array in
CodeGenTarget.cpp to avoid the ordering dependence. NFCI.

Differential Revision: http://reviews.llvm.org/D16826

llvm-svn: 259726

8 years agoMinor code cleanups. NFC.
Junmo Park [Wed, 3 Feb 2016 23:16:39 +0000 (23:16 +0000)]
Minor code cleanups. NFC.

llvm-svn: 259725

8 years agoFix missing module qualification of subprocess.PIPE.
Zachary Turner [Wed, 3 Feb 2016 22:53:18 +0000 (22:53 +0000)]
Fix missing module qualification of subprocess.PIPE.

llvm-svn: 259724

8 years agoPrint the OffsetStart field's relocation
David Majnemer [Wed, 3 Feb 2016 22:45:21 +0000 (22:45 +0000)]
Print the OffsetStart field's relocation

llvm-svn: 259723

8 years agorangify; NFCI
Sanjay Patel [Wed, 3 Feb 2016 22:44:14 +0000 (22:44 +0000)]
rangify; NFCI

llvm-svn: 259722

8 years agoReapply r259624, it is likely not the commit causing the bot failures.
Quentin Colombet [Wed, 3 Feb 2016 22:41:00 +0000 (22:41 +0000)]
Reapply r259624, it is likely not the commit causing the bot failures.
Original message:
Make CF constant string decl visible to name lookup to fix module errors

The return type of the __builtin___*StringMakeConstantString functions
is a pointer to a struct, so we need that struct to be visible to name
lookup so that we will correctly merge multiple declarations of that
type if they come from different modules.

Incidentally, to make this visible to name lookup we need to rename the
type to __NSConstantString, since the real NSConstantString is an
Objective-C interface type.  This shouldn't affect anyone outside the
compiler since users of the constant string builtins cast the result
immediately to CFStringRef.

Since this struct type is otherwise implicitly created by the AST
context and cannot access namelookup, we make this a predefined type
and initialize it in Sema.

Note: this issue of builtins that refer to types not visible to name
lookup technically also affects other builtins (e.g. objc_msgSendSuper),
but in all other cases the builtin is a library builtin and the issue
goes away if you include the library that defines the types it uses,
unlike for these constant string builtins.

rdar://problem/24425801

llvm-svn: 259721

8 years agoclean up; NFC
Sanjay Patel [Wed, 3 Feb 2016 22:37:37 +0000 (22:37 +0000)]
clean up; NFC

llvm-svn: 259720

8 years ago[llvm-readobj] Add support for dumping S_DEFRANGE symbols
David Majnemer [Wed, 3 Feb 2016 22:36:46 +0000 (22:36 +0000)]
[llvm-readobj] Add support for dumping S_DEFRANGE symbols

llvm-svn: 259719

8 years agoAdd generation of LC_VERSION_MIN load commands.
Pete Cooper [Wed, 3 Feb 2016 22:28:29 +0000 (22:28 +0000)]
Add generation of LC_VERSION_MIN load commands.

If the command line contains something like -macosx_version_min and we
don't explicitly disable generation with -no_version_load_command then
we generate the LC_VERSION_MIN command in the output file.

There's a couple of FIXME's in here.  These will be handled soon with
more tests but I didn't want to grow this patch any more than it already was.

rdar://problem/24472630

llvm-svn: 259718

8 years ago[cfi] Safe handling of unaddressable vtable pointers (compiler-rt).
Evgeniy Stepanov [Wed, 3 Feb 2016 22:19:04 +0000 (22:19 +0000)]
[cfi] Safe handling of unaddressable vtable pointers (compiler-rt).

Avoid crashing when printing diagnostics for vtable-related CFI
errors. In diagnostic mode, the frontend does an additional check of
the vtable pointer against the set of all known vtable addresses and
lets the runtime handler know if it is safe to inspect the vtable.

http://reviews.llvm.org/D16824

llvm-svn: 259717

8 years ago[cfi] Safe handling of unaddressable vtable pointers (clang).
Evgeniy Stepanov [Wed, 3 Feb 2016 22:18:55 +0000 (22:18 +0000)]
[cfi] Safe handling of unaddressable vtable pointers (clang).

Avoid crashing when printing diagnostics for vtable-related CFI
errors. In diagnostic mode, the frontend does an additional check of
the vtable pointer against the set of all known vtable addresses and
lets the runtime handler know if it is safe to inspect the vtable.

http://reviews.llvm.org/D16823

llvm-svn: 259716

8 years agoRevert r259624 - Make CF constant string decl visible to name lookup to fix module...
Quentin Colombet [Wed, 3 Feb 2016 22:14:53 +0000 (22:14 +0000)]
Revert r259624 - Make CF constant string decl visible to name lookup to fix module errors.

This breaks some internal bots in stage2: clang seg fault.
Looking with Ben to see what is going on.

llvm-svn: 259715

8 years agoPass socket scheme as part of debug server listen URL.
Oleksiy Vyalov [Wed, 3 Feb 2016 22:02:43 +0000 (22:02 +0000)]
Pass socket scheme as part of debug server listen URL.

http://reviews.llvm.org/D16861

llvm-svn: 259714

8 years agoReplace static const int with enum to fix obnoxious linker errors about a missing...
Reid Kleckner [Wed, 3 Feb 2016 21:45:39 +0000 (21:45 +0000)]
Replace static const int with enum to fix obnoxious linker errors about a missing definition

llvm-svn: 259712

8 years ago[unittests] Move TargetRegistry test from Support to MC
Reid Kleckner [Wed, 3 Feb 2016 21:41:24 +0000 (21:41 +0000)]
[unittests] Move TargetRegistry test from Support to MC

This removes the dependency from SupportTests to all of the LLVM
backends, and makes it link faster.

llvm-svn: 259705

8 years agoSilence -Wsign-conversion issue in ProgramTest.cpp
Reid Kleckner [Wed, 3 Feb 2016 21:41:12 +0000 (21:41 +0000)]
Silence -Wsign-conversion issue in ProgramTest.cpp

Unfortunately, ProgramInfo::ProcessId is signed on Unix and unsigned on
Windows, breaking the standard fix of using '0U' in the gtest
expectation.

llvm-svn: 259704

8 years agoFix pointers to go on the right hand side. NFC.
Ana Pazos [Wed, 3 Feb 2016 21:34:39 +0000 (21:34 +0000)]
Fix pointers to go on the right hand side. NFC.

Summary:
Fixed pointers to go on the right hand side following coding guidelines. NFC.

Patch by Mandeep Singh Grang.

Reviewers: majnemer, arsenm, sanjoy

Differential Revision: http://reviews.llvm.org/D16866

llvm-svn: 259703

8 years ago[LoopStrengthReduce] Don't rewrite PHIs with incoming values from CatchSwitches
David Majnemer [Wed, 3 Feb 2016 21:30:34 +0000 (21:30 +0000)]
[LoopStrengthReduce] Don't rewrite PHIs with incoming values from CatchSwitches

Bail out if we have a PHI on an EHPad that gets a value from a
CatchSwitchInst.  Because the CatchSwitchInst cannot be split, there is
no good place to stick any instructions.

This fixes PR26373.

llvm-svn: 259702

8 years ago[ScalarEvolutionExpander] Simplify findInsertPointAfter
David Majnemer [Wed, 3 Feb 2016 21:30:31 +0000 (21:30 +0000)]
[ScalarEvolutionExpander] Simplify findInsertPointAfter

No functional change is intended.  The loop could only execute, at most,
once.

llvm-svn: 259701

8 years ago[codeview] Remove EmitLabelDiff in favor emitAbsoluteSymbolDiff
Reid Kleckner [Wed, 3 Feb 2016 21:24:42 +0000 (21:24 +0000)]
[codeview] Remove EmitLabelDiff in favor emitAbsoluteSymbolDiff

llvm-svn: 259700

8 years ago[codeview] Use the MCStreamer interface directly instead of AsmPrinter
Reid Kleckner [Wed, 3 Feb 2016 21:15:48 +0000 (21:15 +0000)]
[codeview] Use the MCStreamer interface directly instead of AsmPrinter

This is mostly about having shorter lines and standardizing on one
interface, but it also avoids some needless indirection.

No functional change.

llvm-svn: 259697

8 years ago[DWARFDebug] Fix another case of overlapping ranges
Keno Fischer [Wed, 3 Feb 2016 21:13:33 +0000 (21:13 +0000)]
[DWARFDebug] Fix another case of overlapping ranges

Summary:
In r257979, I added code to ensure that we wouldn't merge DebugLocEntries if
the pieces they describe overlap. Unfortunately, I failed to cover the case,
where there may have multiple active Expressions in the entry, in which case we
need to make sure that no two values overlap before we can perform the merge.

This fixed PR26148.

Reviewers: aprantl
Differential Revision: http://reviews.llvm.org/D16742

llvm-svn: 259696

8 years agoAddress NDEBUG-related linkage issues for Value::assertModuleIsMaterialized()
Todd Fiala [Wed, 3 Feb 2016 21:13:23 +0000 (21:13 +0000)]
Address NDEBUG-related linkage issues for Value::assertModuleIsMaterialized()

The IR/Value class had a linkage issue present when LLVM was built
as a library, and the LLVM library build time had different settings
for NDEBUG than the client of the LLVM library.  Clients could get
into a state where the LLVM lib expected
Value::assertModuleIsMaterialized() to be inline-defined in the header
but clients expected that method to be defined in the LLVM library.

See this llvm-commits thread for more details:
http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20160201/329667.html

llvm-svn: 259695

8 years agotest: make test case more robust against removal of unrelated instructions
Tobias Grosser [Wed, 3 Feb 2016 21:10:11 +0000 (21:10 +0000)]
test: make test case more robust against removal of unrelated instructions

llvm-svn: 259693

8 years agoFix addend computation for IRELATIVE relocations.
Rafael Espindola [Wed, 3 Feb 2016 21:02:48 +0000 (21:02 +0000)]
Fix addend computation for IRELATIVE relocations.

llvm-svn: 259692

8 years ago[SelectionDAG] Fix CombineToPreIndexedLoadStore O(n^2) behavior
Tim Shen [Wed, 3 Feb 2016 20:58:55 +0000 (20:58 +0000)]
[SelectionDAG] Fix CombineToPreIndexedLoadStore O(n^2) behavior

This patch consists of two parts: a performance fix in DAGCombiner.cpp
and a correctness fix in SelectionDAG.cpp.

The test case tests the bug that's uncovered by the performance fix, and
fixed by the correctness fix.

The performance fix keeps the containers required by the
hasPredecessorHelper (which is a lazy DFS) and reuse them. Since
hasPredecessorHelper is called in a loop, the overall efficiency reduced
from O(n^2) to O(n), where n is the number of SDNodes.

The correctness fix keeps iterating the neighbor list even if it's time
to early return. It will return after finishing adding all neighbors to
Worklist, so that no neighbors are discarded due to the original early
return.

llvm-svn: 259691

8 years ago[CUDA] added declarations for device-side system calls
Artem Belevich [Wed, 3 Feb 2016 20:53:58 +0000 (20:53 +0000)]
[CUDA] added declarations for device-side system calls

...and std:: wrappers for free/malloc.

llvm-svn: 259690

8 years agoFix sign conversion warnings in LLDB Python unittests
Reid Kleckner [Wed, 3 Feb 2016 20:48:09 +0000 (20:48 +0000)]
Fix sign conversion warnings in LLDB Python unittests

llvm-svn: 259689

8 years agoEnsure that we substitute into the declaration of a template parameter pack
Richard Smith [Wed, 3 Feb 2016 20:40:30 +0000 (20:40 +0000)]
Ensure that we substitute into the declaration of a template parameter pack
(that is not a pack expansion) during template argument deduction, even if we
deduced that the pack would be empty.

llvm-svn: 259688

8 years agoRefactor conversion of deduced template arguments to reduce repetition.
Richard Smith [Wed, 3 Feb 2016 20:15:01 +0000 (20:15 +0000)]
Refactor conversion of deduced template arguments to reduce repetition.

llvm-svn: 259687

8 years ago[NetBSD] Remove dead code.
Davide Italiano [Wed, 3 Feb 2016 20:13:50 +0000 (20:13 +0000)]
[NetBSD] Remove dead code.

PR: http://reviews.llvm.org/D16818
llvm-svn: 259686

8 years agoRemove a stray ;.
Jim Ingham [Wed, 3 Feb 2016 19:49:03 +0000 (19:49 +0000)]
Remove a stray ;.

llvm-svn: 259685

8 years agoThe SetStopInfo from a Mach Exception was setting the stop
Jim Ingham [Wed, 3 Feb 2016 19:45:31 +0000 (19:45 +0000)]
The SetStopInfo from a Mach Exception was setting the stop
reason to None when we stop due to a trace, then noticed that
we were on a breakpoint that was not valid for the current thread.
That should actually have set it back to trace.

This was pr26441 (<rdar://problem/24470203>)

llvm-svn: 259684

8 years agoMinor performance tweaks to llvm-tblgen (and a few that might be a good idea)
Reid Kleckner [Wed, 3 Feb 2016 19:34:28 +0000 (19:34 +0000)]
Minor performance tweaks to llvm-tblgen (and a few that might be a good idea)

Summary:
This patch adds a reserve call to an expensive function
(`llvm::LoadIntrinsics`), and may fix a few other low hanging
performance fruit (I've put them in comments for now, so we can
discuss).

**Motivation:**

As I'm sure other developers do, when I build LLVM, I build the entire
project with the same config (`Debug`, `MinSizeRel`, `Release`, or
`RelWithDebInfo`). However, the `Debug` config also builds llvm-tblgen
in `Debug` mode. Later build steps that run llvm-tblgen then can
actually be the slowest steps in the entire build. Nobody likes slow
builds.

Reviewers: rnk, dblaikie

Differential Revision: http://reviews.llvm.org/D16832

Patch by Alexander G. Riccio

llvm-svn: 259683

8 years agore.results.form: Format out-of-range subexpression references as null
Duncan P. N. Exon Smith [Wed, 3 Feb 2016 19:30:20 +0000 (19:30 +0000)]
re.results.form: Format out-of-range subexpression references as null

Rather than crashing in match_results::format() when a reference to a
marked subexpression is out of range, format the subexpression as empty
(i.e., replace it with an empty string).  Note that
match_results::operator[]() has a range-check and returns a null match
in this case, so this just re-uses that logic.

llvm-svn: 259682

8 years ago[Sema debugger support] Require non-void types to be complete in unknown-anytype...
Douglas Gregor [Wed, 3 Feb 2016 19:13:08 +0000 (19:13 +0000)]
[Sema debugger support] Require non-void types to be complete in unknown-anytype casts.

When performing a cast from an __unknown_anytype function call to a
non-void type, we need to make sure that type is complete. Fixes
rdar://problem/23959960.

llvm-svn: 259681

8 years agoMove some android platform functions to lldbplatformutil.
Zachary Turner [Wed, 3 Feb 2016 19:12:30 +0000 (19:12 +0000)]
Move some android platform functions to lldbplatformutil.

My eventual goal is to move all of the test decorators to their
own module such as `decorators.py`.  But some of the decorators
use existing functions in `lldbtest.py` and conceptually the
functions are probably more appropriately placed in lldbplatformutil.
Moreover, lldbtest.py is a huge file with a ton of random utility
functions scattered around, so this patch also workds toward the
goal of reducing the footprint of this one module to a more
reasonable size.

So this patch moves some of them over to lldbplatformutil with the
eventual goal of moving decorators over to their own module.

Reviewed By: Tamas Berghammer, Pavel Labath
Differential Revision: http://reviews.llvm.org/D16830

llvm-svn: 259680

8 years agoELF: Remove dead code.
Rui Ueyama [Wed, 3 Feb 2016 19:07:56 +0000 (19:07 +0000)]
ELF: Remove dead code.

R_X86_64_DTPOFF64 relocs are for the dynamic linker.
Static linkers create them but don't consume them.

llvm-svn: 259679

8 years agoAdded Anastasia Stulova as a code owner for OpenCL
Anastasia Stulova [Wed, 3 Feb 2016 18:51:19 +0000 (18:51 +0000)]
Added Anastasia Stulova as a code owner for OpenCL

Reviewers: Chris Lattner

Subscribers: cfe-dev
llvm-svn: 259678

8 years agoFix regression from r259622: the operand of an increment that is the operand of
Richard Smith [Wed, 3 Feb 2016 18:48:43 +0000 (18:48 +0000)]
Fix regression from r259622: the operand of an increment that is the operand of
a cast expression is not the operand of a cast expression itself, so if it's
parenthesized we need to form a ParenExpr not a ParenListExpr.

llvm-svn: 259677

8 years agoARM: support TLS for WoA
Saleem Abdulrasool [Wed, 3 Feb 2016 18:21:59 +0000 (18:21 +0000)]
ARM: support TLS for WoA

Add support for TLS access for Windows on ARM.  This generates a similar access
to MSVC for ARM.

The changes to the tablegen data is needed to support loading an external symbol
global that is not for a call.  The adjustments to the DAG to DAG transforms are
needed to preserve the 32-bit move.

llvm-svn: 259676

8 years agoRevert r259662, which caused regressions on polly tests.
Wei Mi [Wed, 3 Feb 2016 18:05:57 +0000 (18:05 +0000)]
Revert r259662, which caused regressions on polly tests.

llvm-svn: 259675

8 years ago[InstCombine] Revert r238452: Fold IntToPtr and PtrToInt into preceding loads.
Quentin Colombet [Wed, 3 Feb 2016 18:04:13 +0000 (18:04 +0000)]
[InstCombine] Revert r238452: Fold IntToPtr and PtrToInt into preceding loads.

According to git bisect, this is the root cause of a miscompile for Regex in
libLLVMSupport. I am still working on reducing a test case.
The actual bug may be elsewhere and this commit just exposed it.

Anyway, at the moment, to reproduce, follow these steps:
1. Build clang and libLTO in release mode.
2. Create a new build directory <stage2> and cd into it.
3. Use clang and libLTO from #1 to build llvm-extract in Release mode + asserts
   using -O2 -flto
4. Run llvm-extract  -ralias '.*bar' -S test/Other/extract-alias.ll

Result:
program doesn't contain global named '.*bar'!

Expected result:
@a0a0bar = alias void ()* @bar
@a0bar = alias void ()* @bar

declare void @bar()

Note: In step #3, if you don't use lto or asserts, the miscompile disappears.
llvm-svn: 259674

8 years ago[ScheduleDAGInstrs::buildSchedGraph()] Handling of memory dependecies rewritten.
Jonas Paulsson [Wed, 3 Feb 2016 17:52:29 +0000 (17:52 +0000)]
[ScheduleDAGInstrs::buildSchedGraph()] Handling of memory dependecies rewritten.

Recommited, after some fixing with test cases.

Updated test cases:
test/CodeGen/AArch64/arm64-misched-memdep-bug.ll
test/CodeGen/AArch64/tailcall_misched_graph.ll

Temporarily disabled test cases:
test/CodeGen/AMDGPU/split-vector-memoperand-offsets.ll
test/CodeGen/PowerPC/ppc64-fastcc.ll (partially updated)
test/CodeGen/PowerPC/vsx-fma-m.ll
test/CodeGen/PowerPC/vsx-fma-sp.ll

http://reviews.llvm.org/D8705
Reviewers: Hal Finkel, Andy Trick.

llvm-svn: 259673

8 years agoFix comments /NFC
Xinliang David Li [Wed, 3 Feb 2016 17:51:16 +0000 (17:51 +0000)]
Fix comments /NFC

llvm-svn: 259672

8 years agoAdd builtins for bitreverse intrinsic
Matt Arsenault [Wed, 3 Feb 2016 17:49:38 +0000 (17:49 +0000)]
Add builtins for bitreverse intrinsic

Follow the naming convention that bswap uses since it's a
similar sort of operation.

llvm-svn: 259671

8 years agoclang-format: Fix formatting of ternary expressions with comments.
Daniel Jasper [Wed, 3 Feb 2016 17:27:10 +0000 (17:27 +0000)]
clang-format: Fix formatting of ternary expressions with comments.

Before:
  int i = aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ?
      /*bbbbbbbbbbbbbbbbbbbb=*/bbbbbbbbbbbbbbbbbbbbbbbbb :
       ccccccccccccccccccccccccccc;

After:
  int i = aaaaaaaaaaaaaaaaaaaaaaaaaaaaaa ?
      /*bbbbbbbbbbbbbbbbbbbb=*/bbbbbbbbbbbbbbbbbbbbbbbbb :
              ccccccccccccccccccccccccccc;

llvm-svn: 259670

8 years agoDo not expect /dev/null (or NUL) non-mmap'able.
Rui Ueyama [Wed, 3 Feb 2016 17:25:11 +0000 (17:25 +0000)]
Do not expect /dev/null (or NUL) non-mmap'able.

On some Windows environment, this test did not fail, because opening NUL
with FileOutputBuffer didn't fail. Thanks to George Rimar for reporting.

llvm-svn: 259669

8 years ago[clang-tidy] Fix a crash issue on misc-virtual-near-miss check.
Haojian Wu [Wed, 3 Feb 2016 17:21:44 +0000 (17:21 +0000)]
[clang-tidy] Fix a crash issue on misc-virtual-near-miss check.

Summary:
The crash is caused by triggering a Assertion failed in DeclCXX.h when the
check detects non-defined class return type in a class method declaration.

Reviewers: congliu, alexfh

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D16854

llvm-svn: 259668

8 years agoAdd back the ABITest makefiles
Chris Bieneman [Wed, 3 Feb 2016 17:16:01 +0000 (17:16 +0000)]
Add back the ABITest makefiles

These files are standalone and not integrated with CMake, so we probably want them.

llvm-svn: 259667

8 years ago[Unittest] Clean up formatting, NFC
Joseph Tremoulet [Wed, 3 Feb 2016 17:11:24 +0000 (17:11 +0000)]
[Unittest] Clean up formatting, NFC

Summary:
Use an early return to reduce indentation.
Remove unused local.

Reviewers: dblaikie, lhames

Subscribers: lhames, llvm-commits

Differential Revision: http://reviews.llvm.org/D16513

llvm-svn: 259663

8 years ago[SCEV] Try to reuse existing value during SCEV expansion
Wei Mi [Wed, 3 Feb 2016 17:05:12 +0000 (17:05 +0000)]
[SCEV] Try to reuse existing value during SCEV expansion

Current SCEV expansion will expand SCEV as a sequence of operations
and doesn't utilize the value already existed. This will introduce
redundent computation which may not be cleaned up throughly by
following optimizations.

This patch introduces an ExprValueMap which is a map from SCEV to the
set of equal values with the same SCEV. When a SCEV is expanded, the
set of values is checked and reused whenever possible before generating
a sequence of operations.

Differential Revision: http://reviews.llvm.org/D12090

llvm-svn: 259662

8 years agoSimplify. NFC.
Rafael Espindola [Wed, 3 Feb 2016 16:53:39 +0000 (16:53 +0000)]
Simplify. NFC.

llvm-svn: 259660

8 years agoProperly build shared libraries if LLVM_LINK_LLVM_DYLIB is enabled.
Tobias Grosser [Wed, 3 Feb 2016 16:29:04 +0000 (16:29 +0000)]
Properly build shared libraries if LLVM_LINK_LLVM_DYLIB is enabled.

Contributed-by: Jack Howarth <howarthjw@gmail.com>
llvm-svn: 259659

8 years agowww: Simplify 'build & install' descriptions
Tobias Grosser [Wed, 3 Feb 2016 16:18:24 +0000 (16:18 +0000)]
www: Simplify 'build & install' descriptions

We remove information for older versions of Polly and also shorten the overall
text. This should make it a lot easier for people to get to the important code
wight away.

llvm-svn: 259658

8 years ago[ARM] Move GNUEABI divmod to __aeabi_divmod*
Renato Golin [Wed, 3 Feb 2016 16:10:54 +0000 (16:10 +0000)]
[ARM] Move GNUEABI divmod to __aeabi_divmod*

The GNU toolchain emits __aeabi_divmod for soft-divide on ARM cores
which happens to be a lot faster than __divsi3/__modsi3 when the core
has hardware divide instructions. Do the same here.

Fixes PR26450.

llvm-svn: 259657

8 years ago[MachineCopyPropagation] Fix comment. NFC
Jun Bum Lim [Wed, 3 Feb 2016 15:56:27 +0000 (15:56 +0000)]
[MachineCopyPropagation] Fix comment. NFC

Reviewers: MatzeB, qcolombet, jmolloy, mcrosier

Subscribers: llvm-commits, mcrosier

Differential Revision: http://reviews.llvm.org/D16806

llvm-svn: 259656

8 years ago[mips] Remove redundant inclusions of MipsAnalyzeImmediate.h
Daniel Sanders [Wed, 3 Feb 2016 15:54:12 +0000 (15:54 +0000)]
[mips] Remove redundant inclusions of MipsAnalyzeImmediate.h

llvm-svn: 259655

8 years ago[OpenMP] Parsing + sema for target parallel for directive.
Arpith Chacko Jacob [Wed, 3 Feb 2016 15:46:42 +0000 (15:46 +0000)]
[OpenMP] Parsing + sema for target parallel for directive.

Summary:
This patch adds parsing + sema for the target parallel for directive along with testcases.

Reviewers: ABataev

Differential Revision: http://reviews.llvm.org/D16759

llvm-svn: 259654

8 years agotScopInfo: Shorten comment slightly
Tobias Grosser [Wed, 3 Feb 2016 15:42:38 +0000 (15:42 +0000)]
tScopInfo: Shorten comment slightly

llvm-svn: 259653

8 years agoMinor cleanup to remove casts and improve some const correctness. NFC.
Aaron Ballman [Wed, 3 Feb 2016 15:20:51 +0000 (15:20 +0000)]
Minor cleanup to remove casts and improve some const correctness. NFC.

Patch by Alexander Riccio.

llvm-svn: 259652

8 years ago[OpenCL] Adding reserved operator logical xor for OpenCL
Anastasia Stulova [Wed, 3 Feb 2016 15:17:14 +0000 (15:17 +0000)]
[OpenCL] Adding reserved operator logical xor for OpenCL

This patch adds the reserved operator ^^ when compiling for OpenCL (spec v1.1 s6.3.g),
which results in a more meaningful error message.

Patch by Neil Hickey!

Review: http://reviews.llvm.org/D13280

M    test/SemaOpenCL/unsupported.cl
M    include/clang/Basic/TokenKinds.def
M    include/clang/Basic/DiagnosticParseKinds.td
M    lib/Basic/OperatorPrecedence.cpp
M    lib/Lex/Lexer.cpp
M    lib/Parse/ParseExpr.cpp

llvm-svn: 259651

8 years agotsan: disable flaky mmap_stress test
Dmitry Vyukov [Wed, 3 Feb 2016 15:10:00 +0000 (15:10 +0000)]
tsan: disable flaky mmap_stress test

llvm-svn: 259650

8 years ago[DemandedBits] Revert r249687 due to PR26071
James Molloy [Wed, 3 Feb 2016 15:05:06 +0000 (15:05 +0000)]
[DemandedBits] Revert r249687 due to PR26071

This regresses a test in LoopVectorize, so I'll need to go away and think about how to solve this in a way that isn't broken.

From the writeup in PR26071:

What's happening is that ComputeKnownZeroes is telling us that all bits except the LSB are zero. We're then deciding that only the LSB needs to be demanded from the icmp's inputs.

This is where we're wrong - we're assuming that after simplification the bits that were known zero will continue to be known zero. But they're not - during trivialization the upper bits get changed (because an XOR isn't shrunk), so the icmp fails.

The fault is in demandedbits - its contract does clearly state that a non-demanded bit may either be zero or one.

llvm-svn: 259649

8 years agoProvide match function to look over an entire TU again.
Daniel Jasper [Wed, 3 Feb 2016 14:29:55 +0000 (14:29 +0000)]
Provide match function to look over an entire TU again.

llvm-svn: 259648

8 years agoForgot to remove file in previous commit.
Yury Gribov [Wed, 3 Feb 2016 13:36:31 +0000 (13:36 +0000)]
Forgot to remove file in previous commit.

llvm-svn: 259647

8 years ago[analyzer] AnalysisConsumer: print fully-qualified function name while displaying...
Yury Gribov [Wed, 3 Feb 2016 13:35:33 +0000 (13:35 +0000)]
[analyzer] AnalysisConsumer: print fully-qualified function name while displaying progress

-analyzer-display progress option prints only function names which may be ambiguous. This patch forces AnalysisConsumer to print fully-qualified function names.
Patch by Alex Sidorin!

Differential Revision: http://reviews.llvm.org/D16804

llvm-svn: 259646

8 years agoFix for PR 26381
Nemanja Ivanovic [Wed, 3 Feb 2016 12:53:38 +0000 (12:53 +0000)]
Fix for PR 26381

Simple fix - Constant values were not being sign extended in FastIsel.

llvm-svn: 259645

8 years agoAdd ability to override JIT expr compiler options.
Aidan Dodds [Wed, 3 Feb 2016 12:33:05 +0000 (12:33 +0000)]
Add ability to override JIT expr compiler options.

Runtimes should be able to pass custom compilation options to the JIT for their stack frame. This patch adds a custom expression options member class to LanguageOptions, and modifies the clang expression evaluator to check the current runtime for those options. If those options are available on the runtime, they are passed to the clang compiler.

Committed for Luke Drummond.
Differential Revision: http://reviews.llvm.org/D15527

llvm-svn: 259644

8 years ago[clang-tidy] bug fix: Don't warn on partial template specialization in `misc-definiti...
Haojian Wu [Wed, 3 Feb 2016 12:10:27 +0000 (12:10 +0000)]
[clang-tidy] bug fix: Don't warn on partial template specialization in `misc-definitions-in-headers` check.

Reviewers: alexfh

Subscribers: cfe-commits

Differential Revision: http://reviews.llvm.org/D16578

llvm-svn: 259643

8 years agoRemove skipUnlessListedRemote test decorator
Pavel Labath [Wed, 3 Feb 2016 11:51:25 +0000 (11:51 +0000)]
Remove skipUnlessListedRemote test decorator

This decorator was used in only one test, and it's behaviour was quite complicated. It skipped
if:
- test was remote
- platform was *not* android

I am not aware of anyone running tests with this configuration (and even then, I am not aware of
a reason why the test should not pass), but if TestLoadUnload starts breaking for you after this
commit, please disable the test with
@expectedFailureAll(remote=True, oslist=[YOUR_PLATFORM])

llvm-svn: 259642

8 years ago[mips] Add SHF_MIPS_GPREL flag to the MIPS .sbss and .sdata sections
Simon Atanasyan [Wed, 3 Feb 2016 11:50:22 +0000 (11:50 +0000)]
[mips] Add SHF_MIPS_GPREL flag to the MIPS .sbss and .sdata sections

MIPS ABI states that .sbss and .sdata sections must have SHF_MIPS_GPREL
flag. See Figure 4–7 on page 69 in the following document:
ftp://www.linux-mips.org/pub/linux/mips/doc/ABI/mipsabi.pdf.

Differential Revision: http://reviews.llvm.org/D15740

llvm-svn: 259641

8 years agoclang-tidy: [misc-unused-parameters] Ignore template instantiations.
Daniel Jasper [Wed, 3 Feb 2016 11:33:18 +0000 (11:33 +0000)]
clang-tidy: [misc-unused-parameters] Ignore template instantiations.

No functional changes intended as we should already do the
corresponding fixes when visiting the primary template. There are
existing tests that verify that we do change unused parameters of
templated functions.

llvm-svn: 259640

8 years ago-inline-asm][X86] Add ability to use AVX512 in MS inline asm
Marina Yatsina [Wed, 3 Feb 2016 11:32:08 +0000 (11:32 +0000)]
-inline-asm][X86] Add ability to use AVX512 in MS inline asm

Defined the new AVX512 registers in clang inline asm.
Fixed a bug in the MC subtarget info creation during the parsing of MS asm statement - now it receives the actual CPU and target features information.

Differential Revision: http://reviews.llvm.org/D16757

llvm-svn: 259639

8 years agoFix an off-by-one in SocketTest::DecodeHostAndPort
Pavel Labath [Wed, 3 Feb 2016 11:12:23 +0000 (11:12 +0000)]
Fix an off-by-one in SocketTest::DecodeHostAndPort

65535 is still a valid port. This should fix the android failures we were getting when we chose
to connect over 65535 to the remote lldb-server.

llvm-svn: 259638

8 years agoReduce initial Sema memory consumption by 400KB. By Elisavet Sakellari.
Axel Naumann [Wed, 3 Feb 2016 10:45:22 +0000 (10:45 +0000)]
Reduce initial Sema memory consumption by 400KB. By Elisavet Sakellari.

llvm-svn: 259637

8 years ago[TableGen] Add 'register alternative name matching' support
Dylan McKay [Wed, 3 Feb 2016 10:30:16 +0000 (10:30 +0000)]
[TableGen] Add 'register alternative name matching' support

Summary:
This adds a new attribute which targets can set in TableGen which causes a function to be generated which matches register alternative names. This is very similar to `ShouldEmitMatchRegisterName`, except it works on alt names.

This patch is currently used by the out of tree part of the AVR backend. It reduces code duplication greatly, and has the effect that you do not need to hardcode altname to register mappings in C++.

It will not work on targets which have registers which share the same aliases.

Reviewers: stoklund, arsenm, dsanders, hfinkel, vkalintiris

Subscribers: hfinkel, dylanmckay, llvm-commits

Differential Revision: http://reviews.llvm.org/D16312

llvm-svn: 259636

8 years ago[X86][AVX] Add support for 64-bit VZEXT_LOAD of 256/512-bit vectors to EltsFromConsec...
Simon Pilgrim [Wed, 3 Feb 2016 09:41:59 +0000 (09:41 +0000)]
[X86][AVX] Add support for 64-bit VZEXT_LOAD of 256/512-bit vectors to EltsFromConsecutiveLoads

Follow up to D16217 and D16729

This change uncovered an odd pattern where VZEXT_LOAD v4i64 was being lowered to a load of the lower v2i64 (so the 2nd i64 destination element wasn't being zeroed), I can't find any use/reason for this and have removed the pattern and replaced it so only the 1st i64 element is loaded and the upper bits all zeroed. This matches the description for X86ISD::VZEXT_LOAD

Differential Revision: http://reviews.llvm.org/D16768

llvm-svn: 259635

8 years ago[RenderScript] Use LLVM DWARF language enum
Ewan Crawford [Wed, 3 Feb 2016 09:17:03 +0000 (09:17 +0000)]
[RenderScript] Use LLVM DWARF language enum

A DWARF language vender extension for RenderScript was added to LLVM in r259348(http://reviews.llvm.org/D16409)
We should use this generated enum instead of the hardcoded value.

RenderScript is also based on C99 with some extensions, so we want to use ClangASTContext when RS is detected.

Reviewers:  clayborg
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D16766

llvm-svn: 259634

8 years agoScopInfo: Extend description of the access relation.
Tobias Grosser [Wed, 3 Feb 2016 06:37:38 +0000 (06:37 +0000)]
ScopInfo: Extend description of the access relation.

llvm-svn: 259633

8 years agoAdd a compatibility test
Xinliang David Li [Wed, 3 Feb 2016 06:27:38 +0000 (06:27 +0000)]
Add a compatibility test

llvm-svn: 259632

8 years agoFix a typo in comment
Xinliang David Li [Wed, 3 Feb 2016 06:24:11 +0000 (06:24 +0000)]
Fix a typo in comment

llvm-svn: 259631

8 years agoFix uninitiazed variable use problem
Xinliang David Li [Wed, 3 Feb 2016 06:23:16 +0000 (06:23 +0000)]
Fix uninitiazed variable use problem

llvm-svn: 259630

8 years agoRevert "Support loads with differently sized types from a single array"
Tobias Grosser [Wed, 3 Feb 2016 05:53:27 +0000 (05:53 +0000)]
Revert "Support loads with differently sized types from a single array"

This reverts commit (@259587). It needs some further discussions.

llvm-svn: 259629

8 years agoclang-format: [JS/TypeScript] Support "enum" as an optional property name, too.
Daniel Jasper [Wed, 3 Feb 2016 05:33:44 +0000 (05:33 +0000)]
clang-format: [JS/TypeScript] Support "enum" as an optional property name, too.

Before:
  enum?: string
  [];

After:
  enum?: string[];

llvm-svn: 259628

8 years agoSync up with master file
Xinliang David Li [Wed, 3 Feb 2016 04:09:02 +0000 (04:09 +0000)]
Sync up with master file

llvm-svn: 259627

8 years ago[PGO] Profile summary reader/writer support
Xinliang David Li [Wed, 3 Feb 2016 04:08:18 +0000 (04:08 +0000)]
[PGO] Profile summary reader/writer support

With this patch, the profile summary data will be available in indexed
profile data file so that profiler reader/compiler optimizer can start
to make use of.

Differential Revision: http://reviews.llvm.org/D16258

llvm-svn: 259626

8 years agoLowerBitSets: Don't bother to do any work if the llvm.bitset.test intrinsic is unused.
Peter Collingbourne [Wed, 3 Feb 2016 03:48:46 +0000 (03:48 +0000)]
LowerBitSets: Don't bother to do any work if the llvm.bitset.test intrinsic is unused.

llvm-svn: 259625

8 years agoMake CF constant string decl visible to name lookup to fix module errors
Ben Langmuir [Wed, 3 Feb 2016 03:26:19 +0000 (03:26 +0000)]
Make CF constant string decl visible to name lookup to fix module errors

The return type of the __builtin___*StringMakeConstantString functions
is a pointer to a struct, so we need that struct to be visible to name
lookup so that we will correctly merge multiple declarations of that
type if they come from different modules.

Incidentally, to make this visible to name lookup we need to rename the
type to __NSConstantString, since the real NSConstantString is an
Objective-C interface type.  This shouldn't affect anyone outside the
compiler since users of the constant string builtins cast the result
immediately to CFStringRef.

Since this struct type is otherwise implicitly created by the AST
context and cannot access namelookup, we make this a predefined type
and initialize it in Sema.

Note: this issue of builtins that refer to types not visible to name
lookup technically also affects other builtins (e.g. objc_msgSendSuper),
but in all other cases the builtin is a library builtin and the issue
goes away if you include the library that defines the types it uses,
unlike for these constant string builtins.

rdar://problem/24425801

llvm-svn: 259624

8 years agoAdd #include "llvm/Support/raw_ostream.h" to fix Windows build.
Peter Collingbourne [Wed, 3 Feb 2016 03:16:37 +0000 (03:16 +0000)]
Add #include "llvm/Support/raw_ostream.h" to fix Windows build.

llvm-svn: 259623