Quentin Colombet [Tue, 31 Mar 2015 00:31:13 +0000 (00:31 +0000)]
[AArch64] Fix poor codegen for add immediate.
We used to match the register variant before the immediate when the register
argument could be implicitly zero-extended.
llvm-svn: 233653
Vince Harron [Tue, 31 Mar 2015 00:27:10 +0000 (00:27 +0000)]
Increate backlog of lldb-platform's listener socket
lldb-platform's listener socket only had a backlog of one connection.
That means that if more than one client connected simultaneously, the
connection would be refused. The test suite can be run remotely with
dozens of threads connecting simultaneously. Raised this limit to 100
to effectively eliminate lost connections.
Test Plan:
run tests against a remote target
Differential Revision: http://reviews.llvm.org/D8696
llvm-svn: 233652
Vince Harron [Tue, 31 Mar 2015 00:24:51 +0000 (00:24 +0000)]
Fix socket leak in lldb-server platform
lldb-server server process was holding on to a connection to every
client that connected. This continued until the process hit it's
file limit and exited.
lldb-server p --listen *:5432 --server
run remote tests against that server more than 1000 times
Differential Revision: http://reviews.llvm.org/D8697
llvm-svn: 233651
Duncan P. N. Exon Smith [Tue, 31 Mar 2015 00:10:37 +0000 (00:10 +0000)]
DebugInfo: Remove LexicalBlockFile scope/context distinction
Two things here:
1. I read `getScope()` and `getContext()` backwards in r233640. There
was no need for `getScopeOfScope()`. Obviously not enough test
coverage here (as I said in that commit, I'm going to come back to
that), but anyway I'm reverting to the behaviour before r233640.
2. The callers that use `DILexicalBlockFile::getContext()` don't seem
to care about the difference. Just have it redirect to `getScope()`
so I can't get confused again.
llvm-svn: 233650
Eric Christopher [Tue, 31 Mar 2015 00:10:23 +0000 (00:10 +0000)]
Update for llvm commit r233648.
llvm-svn: 233649
Eric Christopher [Tue, 31 Mar 2015 00:10:04 +0000 (00:10 +0000)]
Replace the MCSubtargetInfo parameter with a Triple when creating
an MCInstPrinter. Update all callers and use where we wanted a Triple
previously.
llvm-svn: 233648
NAKAMURA Takumi [Tue, 31 Mar 2015 00:03:43 +0000 (00:03 +0000)]
llvm/examples: Suppress building a few JIT examples.
examples/ExceptionDemo/CMakeFiles/ExceptionDemo.dir/ExceptionDemo.cpp.o:(.data.rel.ro._ZTIN4llvm18MCJITMemoryManagerE[_ZTIN4llvm18MCJITMemoryManagerE]+0x10): undefined reference to `typeinfo for llvm::RuntimeDyld::MemoryManager'
llvm-svn: 233647
NAKAMURA Takumi [Tue, 31 Mar 2015 00:03:36 +0000 (00:03 +0000)]
llvm/examples: Add RuntimeDyld as libdeps.
llvm-svn: 233646
Chaoren Lin [Tue, 31 Mar 2015 00:03:00 +0000 (00:03 +0000)]
Fix .gitignore to ignore symlinks.
llvm-svn: 233645
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 23:58:59 +0000 (23:58 +0000)]
LexicalScopes: Cleanup remaining uses of DebugLoc
llvm-svn: 233644
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 23:47:26 +0000 (23:47 +0000)]
DebugLoc: Remove getFromDILexicalBlock()
The only user of `DebugLoc::getFromDILexicalBlock()` was creating a new
`MDLocation` as convenient API for passing an `MDScope`. Stop doing
that, and remove the API. If in the future we actually *want* to create
new DebugLocs, calling `MDLexicalBlock::get()` makes more sense.
llvm-svn: 233643
Fariborz Jahanian [Mon, 30 Mar 2015 23:30:24 +0000 (23:30 +0000)]
[Objective-C patch]. Amend TransformObjCMessageExpr to handle call
to 'super' of instance/class methods and not assert.
rdar://
20350364
llvm-svn: 233642
Marshall Clow [Mon, 30 Mar 2015 23:26:16 +0000 (23:26 +0000)]
Make the new tests better; make sure that we're testing the case where no reallocation has to happen
llvm-svn: 233641
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 23:21:21 +0000 (23:21 +0000)]
LexicalScopes: Use debug info hierarchy pervasively
Pervasively use the types provided by the debug info hierarchy rather
than `MDNode` in `LexicalScopes`.
I noticed (again, I guess, based on comments in the implementation?)
that `DILexicalBlockFile::getScope()` returns something different from
`DILexicalBlockFile::getContext()`. I created a local helper for
getting the same logic from `MDLexicalBlockFile` called
`getScopeOfScope()`. I still don't really understand it, but I've added
some FIXMEs and I'll come back to it (I suspect the way we encode these
objects isn't really ideal).
Note that my previous commit r233610 accidentally changed behaviour in
`findLexicalScope()` -- it transitioned from a call to
`DILexicalBlockFile::getScope()` to `MDLexicalBlockFile::getScope()`
(sounds right, doesn't it?) -- so I've fixed that as a drive-by. No
tests failed with my error, so it looks like we're missing some coverage
here... when I come back to understand the logic, I'll see if I can add
some.
Other than the fix to `findLexicalScope()`, no functionality change.
llvm-svn: 233640
David Majnemer [Mon, 30 Mar 2015 23:14:45 +0000 (23:14 +0000)]
Silence an unused variable warning.
No functional change intended.
llvm-svn: 233639
Kostya Serebryany [Mon, 30 Mar 2015 23:05:30 +0000 (23:05 +0000)]
Move lib/Fuzzer docs from a README.txt to a proper .rst file.
Summary:
Move lib/Fuzzer docs from a README.txt to a proper .rst file.
This change does not add any content, just formatting.
Test Plan: n/a
Reviewers: samsonov
Reviewed By: samsonov
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D8710
llvm-svn: 233638
Kostya Serebryany [Mon, 30 Mar 2015 23:04:35 +0000 (23:04 +0000)]
[fuzzer] when a single unit takes over 1 second to run and it is the slowest one so far, print it.
llvm-svn: 233637
David Majnemer [Mon, 30 Mar 2015 22:58:10 +0000 (22:58 +0000)]
[WinEH] Run cleanup handlers when an exception is thrown
Generate tables in the .xdata section representing what actions to take
when an exception is thrown. This currently fills in state for
cleanups, catch handlers are still unfinished.
llvm-svn: 233636
Juergen Ributzka [Mon, 30 Mar 2015 22:45:56 +0000 (22:45 +0000)]
Transfer implicit operands when expanding the RET_ReallyLR pseudo instruction.
When we expand the RET_ReallyLR pseudo instruction we also need to transfer the
implicit operands.
The return register is an implicit operand and without it the liveness
calculation generates an incorrect live-out set for the patchpoint.
This fixes rdar://problem/
19068476.
llvm-svn: 233635
Kostya Serebryany [Mon, 30 Mar 2015 22:44:03 +0000 (22:44 +0000)]
[fuzzer] print various stats in a unified way
llvm-svn: 233624
Alexei Starovoitov [Mon, 30 Mar 2015 22:40:40 +0000 (22:40 +0000)]
[bpf] add support for bswap instructions
BPF has cpu_to_be and cpu_to_le instructions.
For now assume little endian and generate cpu_to_be for ISD::BSWAP.
llvm-svn: 233620
Alexei Starovoitov [Mon, 30 Mar 2015 22:40:35 +0000 (22:40 +0000)]
[bpf] fix build
fix build broken due to r233599.
Would still need to switch to MDLocation long term.
llvm-svn: 233619
Simon Atanasyan [Mon, 30 Mar 2015 22:36:07 +0000 (22:36 +0000)]
[Mips] Move the `Elf_RegInfo` structure declaration to the separate file
No functional changes.
llvm-svn: 233618
Simon Atanasyan [Mon, 30 Mar 2015 22:36:01 +0000 (22:36 +0000)]
[ELF] Change type of `OutputSection::setType()` argument to int64_t
Type of `OutputSection::_type` field is int64_t. This change makes
the field's and the argument's types consistent and allows to assign
full range of values to the `OutputSection::_type` field.
llvm-svn: 233617
Simon Atanasyan [Mon, 30 Mar 2015 22:35:56 +0000 (22:35 +0000)]
[Mips] clang-format the code
No functional changes.
llvm-svn: 233616
Eric Christopher [Mon, 30 Mar 2015 22:31:14 +0000 (22:31 +0000)]
Rename const char *Triple argument to TT to avoid shadowing llvm::Triple.
llvm-svn: 233615
Eric Christopher [Mon, 30 Mar 2015 22:16:37 +0000 (22:16 +0000)]
Remove unused MCSubtargetInfo argument from the X86 MCInstPrinter ctors.
llvm-svn: 233614
Kostya Serebryany [Mon, 30 Mar 2015 22:09:51 +0000 (22:09 +0000)]
DFSan-based fuzzer (proof of concept).
Summary:
This adds a simple DFSan-based (i.e. taint-guided) fuzzer mutator,
see the comments for details.
Test Plan: a test added
Reviewers: samsonov, pcc
Reviewed By: samsonov, pcc
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D8669
llvm-svn: 233613
Eric Christopher [Mon, 30 Mar 2015 22:09:43 +0000 (22:09 +0000)]
Remove unused MCSubtargetInfo argument from the Sparc MCInstPrinter ctors.
llvm-svn: 233612
Eric Christopher [Mon, 30 Mar 2015 22:03:16 +0000 (22:03 +0000)]
Remove unused MCSubtargetInfo argument from the NVPTX MCInstPrinter ctors.
llvm-svn: 233611
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 21:54:46 +0000 (21:54 +0000)]
LexicalScopes: Use MDLocation directly instead of DebugLoc
There's no benefit to using `DebugLoc` here. Moreover, this will let a
follow-up commit work with `MDScope` directly instead of `DebugLoc`.
llvm-svn: 233610
Eric Christopher [Mon, 30 Mar 2015 21:52:28 +0000 (21:52 +0000)]
Remove unused MCSubtargetInfo argument from the ARM MCInstPrinter ctors.
llvm-svn: 233609
Eric Christopher [Mon, 30 Mar 2015 21:52:26 +0000 (21:52 +0000)]
Remove unused MCSubtargetInfo argument from the AArch64 MCInstPrinter ctors.
llvm-svn: 233608
Eric Christopher [Mon, 30 Mar 2015 21:52:21 +0000 (21:52 +0000)]
Remove unused Target argument from MCInstPrinter ctor functions.
llvm-svn: 233607
Quentin Colombet [Mon, 30 Mar 2015 21:50:44 +0000 (21:50 +0000)]
[RegisterCoalescer] Fix a potential misuse of direct operand index in the
terminal rule.
Spot by code inspection.
llvm-svn: 233606
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 21:47:17 +0000 (21:47 +0000)]
DebugInfo: Use the new DebugLoc API from r233573
This should fix the build [1] after r233599 removed the old API.
[1]: http://lab.llvm.org:8011/builders/perf-x86_64-penryn-O3-polly-parallel-fast/builds/5265
llvm-svn: 233605
David Blaikie [Mon, 30 Mar 2015 21:41:43 +0000 (21:41 +0000)]
[opaque pointer type] Change GetElementPtrInst::getIndexedType to take the pointee type
This pushes the use of PointerType::getElementType up into several
callers - I'll essentially just have to keep pushing that up the stack
until I can eliminate every call to it...
llvm-svn: 233604
Rafael Espindola [Mon, 30 Mar 2015 21:36:43 +0000 (21:36 +0000)]
Fix PR23045.
Keep a note in the materializer that we are stripping debug info so that
user doing a lazy read of the module don't hit outdated formats.
Thanks to Duncan for suggesting the fix.
llvm-svn: 233603
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 21:35:14 +0000 (21:35 +0000)]
Transforms: Update unit tests to use verifyModule()
Since I'm slowly gutting `DISubprogram` and `DICompileUnit`, update the
`CloneFunc` unit tests to call `verifyModule()` (where the checks are
moving to).
llvm-svn: 233602
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 21:32:28 +0000 (21:32 +0000)]
DwarfDebug: Avoid creating new DebugLocs in the backend
Don't use `DebugLoc::getFnDebugLoc()`, which creates new `MDLocation`s,
in the backend. We just want to grab the subprogram here anyway.
llvm-svn: 233601
Paul Robinson [Mon, 30 Mar 2015 21:27:28 +0000 (21:27 +0000)]
Explain how to abandon a review.
llvm-svn: 233600
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 21:19:50 +0000 (21:19 +0000)]
DebugInfo: Remove dead code from old DebugLoc API
Remove old API for `DebugLoc` now that all the callers have been
updated. If this broke your out-of-tree build, here's a quick map from
the old API to the new one:
DebugLoc DebugLoc::getFromMDLocation(MDNode *)
=> DebugLoc::DebugLoc(MDLocation *)
=> explicit DebugLoc::DebugLoc(MDNode *) // works with broken code
MDNode *DebugLoc::getAsMDNode(LLVMContext &)
=> MDLocation *DebugLoc::get()
=> DebugLoc::operator MDLocation *()
=> MDNode *DebugLoc::getAsMDNode() // works with broken code
bool DebugLoc::isUnknown()
=> DebugLoc::operator MDLocation *()
i.e.: if (MDLocation *DL = ...)
=> DebugLoc::operator bool() // works with broken code
i.e.: if (DebugLoc DL = ...)
void DebugLoc::getScopeAndInlinedAt(MDNode *&, MDNode *&)
=> use: MDNode *DebugLoc::getScope()
and: MDLocation *DebugLoc::getInlinedAt()
MDNode *DebugLoc::getScopeNode(LLVMContext &)
=> MDNode *DebugLoc::getInlinedAtScope()
void DebugLoc::dump(LLVMContext &)
=> void DebugLoc::dump()
void DebugLoc::getFnDebugLoc(LLVMContext &)
=> void DebugLoc::getFnDebugLoc()
MDNode *DebugLoc::getScope(LLVMContext &)
=> MDNode *DebugLoc::getScope()
MDNode *DebugLoc::getInlinedAt(LLVMContext &)
=> MDLocation *DebugLoc::getInlinedAt()
I've noted above the only functions that won't crash on broken code (due
to downcasting to `MDLocation`). If your code could be dealing with
broken IR (i.e., you haven't run the verifier yet, or you've used a
temporary node that will eventually (but not yet) get RAUW'ed to an
`MDLocation`), you need to restrict yourself to those.
llvm-svn: 233599
Rafael Espindola [Mon, 30 Mar 2015 21:18:36 +0000 (21:18 +0000)]
Use range loops and add missing braces.
llvm-svn: 233598
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 21:05:29 +0000 (21:05 +0000)]
Transforms: Fix a use of the old DebugLoc in unittests
Missed this one before.
llvm-svn: 233597
David Blaikie [Mon, 30 Mar 2015 20:42:56 +0000 (20:42 +0000)]
[opaque pointer type] More IRBuilder::createGEP (non-inbounds) migrations: CodeGenPrepare and SimplifyLibCalls
llvm-svn: 233596
Peter Collingbourne [Mon, 30 Mar 2015 20:41:21 +0000 (20:41 +0000)]
MC: For variable symbols, maintain MCSymbol::Section as a cache.
This fixes the visibility of symbols in certain edge cases involving aliases
with multiple levels of indirection.
Fixes PR19582.
Differential Revision: http://reviews.llvm.org/D8586
llvm-svn: 233595
Derek Schuff [Mon, 30 Mar 2015 20:31:33 +0000 (20:31 +0000)]
Add driver support for Native Client SDK
Add Tool and ToolChain support for clang to target the NaCl OS using the NaCl
SDK for x86-32, x86-64 and ARM.
Includes nacltools::Assemble and Link which are derived from gnutools. They
are similar to Linux but different enought that they warrant their own class.
Also includes a NaCl_TC in ToolChains derived from Generic_ELF with library
and include paths suitable for an SDK and independent of the system tools.
Differential Revision: http://reviews.llvm.org/D8590
llvm-svn: 233594
Jordan Rose [Mon, 30 Mar 2015 20:18:04 +0000 (20:18 +0000)]
[analyzer] Add test for previous commit.
Again, this is being applied in a separate commit so that the previous commit
can be reverted while leaving the test in place.
rdar://problem/
20335433
llvm-svn: 233593
Jordan Rose [Mon, 30 Mar 2015 20:18:00 +0000 (20:18 +0000)]
[analyzer] Disable all retain count diagnostics on values that come from ivars.
This is imitating a pre-r228174 state where ivars are not considered tracked by
default, but with the addition that even ivars /with/ retain count information
(e.g. "[_ivar retain]; [ivar _release];") are not being tracked as well. This is
to ensure that we don't regress on values accessed through both properties and
ivars, which is what r228174 was trying to fix.
The issue occurs in code like this:
[_contentView retain];
[_contentView removeFromSuperview];
[self addSubview:_contentView]; // invalidates 'self'
[_contentView release];
In this case, the call to -addSubview: may change the value of self->_contentView,
and so the analyzer can't be sure that we didn't leak the original _contentView.
This is a correct conservative view of the world, but not a useful one. Until we
have a heuristic that allows us to not consider this a leak, not emitting a
diagnostic is our best bet.
This commit disables all of the ivar-related retain count tests, but does not
remove them to ensure that we don't crash trying to evaluate either valid or
erroneous code. The next commit will add a new test for the example above so
that this commit (and the previous one) can be reverted wholesale when a better
solution is implemented.
Rest of rdar://problem/
20335433
llvm-svn: 233592
Jordan Rose [Mon, 30 Mar 2015 20:17:47 +0000 (20:17 +0000)]
[analyzer] Don't special-case ivars backing +0 properties.
Give up this checking in order to continue tracking that these values came from
direct ivar access, which will be important in the next commit.
Part of rdar://problem/
20335433
llvm-svn: 233591
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 20:04:06 +0000 (20:04 +0000)]
llvm-dis: Use the new `DebugLoc` API, NFC
Update tools/llvm-dis to use the new `DebugLoc` API.
llvm-svn: 233590
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 20:01:41 +0000 (20:01 +0000)]
DebugInfo: Use new LLVM API for DebugLoc
Use the new API for `DebugLoc` added in r233573 before the old one
disappears.
llvm-svn: 233589
Richard Smith [Mon, 30 Mar 2015 19:52:53 +0000 (19:52 +0000)]
Updating code owners file per discussion with Doug at the 2014-10 dev meeting.
llvm-svn: 233588
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 19:49:49 +0000 (19:49 +0000)]
Transforms: Use the new DebugLoc API, NFC
Update lib/Analysis and lib/Transforms to use the new `DebugLoc` API.
llvm-svn: 233587
Kit Barton [Mon, 30 Mar 2015 19:40:59 +0000 (19:40 +0000)]
[PPC] Move argument range checks for HTM and crypto builtins to Sema
The argument range checks for the HTM and Crypto builtins were implemented in
CGBuiltin.cpp, not in Sema. This change moves them to the appropriate location
in SemaChecking.cpp. It requires the creation of a new method in the Sema class
to do checks for PPC-specific builtins.
http://reviews.llvm.org/D8672
llvm-svn: 233586
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 19:40:05 +0000 (19:40 +0000)]
IR: Use the new DebugLoc API, NFC
Update lib/IR and lib/Bitcode to use the new `DebugLoc` API. Added an
explicit conversion to `bool` (avoiding a conversion to `MDLocation`),
since a couple of these use cases need to handle broken code.
llvm-svn: 233585
Paul Robinson [Mon, 30 Mar 2015 19:37:44 +0000 (19:37 +0000)]
Verify 'optnone' can run DAG combiner when appropriate.
Adds a test to verify the behavior that r233153 restored: 'optnone'
does not spuriously disable the DAG combiner, and in fact there are
cases where the DAG combiner must run (even at -O0 or 'optnone') in
order for codegen to succeed.
Differential Revision: http://reviews.llvm.org/D8614
llvm-svn: 233584
Justin Holewinski [Mon, 30 Mar 2015 19:30:55 +0000 (19:30 +0000)]
[NVPTX] Associate a minimum PTX version for each SM architecture
When a new SM architecture is introduced, it is only supported by the
current PTX version and later. Make sure we are using at least the
minimum PTX version for the target architecture.
This also removes support for PTX ISA < 3.2.
llvm-svn: 233583
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 19:14:47 +0000 (19:14 +0000)]
CodeGen: Use the new DebugLoc API, NFC
Update lib/CodeGen (and lib/Target) to use the new `DebugLoc` API.
llvm-svn: 233582
Kit Barton [Mon, 30 Mar 2015 18:46:15 +0000 (18:46 +0000)]
[PPC] Modify HTM and crypto intrinsics to inherit from GCCBuiltin
There was a change to the way some of the HTM and crypto builtins are being
handled in Clang. Previously, some of the builtins were dealt with in the
CodeGenFunction::EmitPPCBuiltinExpr method (in order to do range checking on
constant arguments). These check will been moved to Sema
http://reviews.llvm.org/D8672), which means those builtins will not be handled
in the EmitPPCBuiltinExpr method anymore. As a result, the definition of the
intrinsics in IntrinsicsPowerPC.td needs to be modified to inherit from the
GCCBuiltin definition.
http://reviews.llvm.org/D8673
llvm-svn: 233581
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 18:45:11 +0000 (18:45 +0000)]
DebugInfo: Reflow printDebugLoc() to use early returns, NFC
llvm-svn: 233580
Lang Hames [Mon, 30 Mar 2015 18:37:01 +0000 (18:37 +0000)]
[Orc] Separate callback manager selection from callback manager construction for
OrcLazyJIT in lli.
Separating these concerns enables more natural error handling.
llvm-svn: 233579
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 18:29:18 +0000 (18:29 +0000)]
Bitcode: Reflow code to use early continues, NFC
llvm-svn: 233578
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 18:23:28 +0000 (18:23 +0000)]
SelectionDAG: Reflow code to use early returns, NFC
llvm-svn: 233577
Justin Holewinski [Mon, 30 Mar 2015 18:12:50 +0000 (18:12 +0000)]
[NVPTX] Add options for PTX 4.1/4.2 and SM 3.2/3.7/5.2/5.3
llvm-svn: 233575
Ulrich Weigand [Mon, 30 Mar 2015 18:08:50 +0000 (18:08 +0000)]
[SystemZ] Fix fallout from r233543 on no-assert builds
Test cases must not check for symbolic variable names that are not
present in IR generated by no-assert builds.
Fixed by testing a more complete subset of the va_arg dataflow,
without relying on variable names.
llvm-svn: 233574
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 18:07:40 +0000 (18:07 +0000)]
DebugInfo: Write new DebugLoc API
Rewrite `DebugLoc` with a cleaner API that reflects its current status
as a wrapper around an `MDLocation` pointer.
- Add accessors/constructors to/from `MDLocation`.
- Simplify construction from `MDNode`.
- Remove unnecessary `LLVMContext` from APIs.
- Drop some API that isn't useful any more.
- Rewrite documentation.
Actually, I've left the old API behind temporarily at the bottom of the
class so that I can update callers in separate commits. I'll remove it
once the callers are updated.
llvm-svn: 233573
Tobias Grosser [Mon, 30 Mar 2015 17:56:50 +0000 (17:56 +0000)]
Drop some CLooG leftovers
llvm-svn: 233572
Tobias Grosser [Mon, 30 Mar 2015 17:55:40 +0000 (17:55 +0000)]
Delete some leftovers from scoplib
llvm-svn: 233571
Tobias Grosser [Mon, 30 Mar 2015 17:54:01 +0000 (17:54 +0000)]
Drop libpluto support
We do not have buildbots or anything that tests this functionality, hence it
most likely bitrots. People interested to use this functionality can always
recover it from svn history.
llvm-svn: 233570
Adrian McCarthy [Mon, 30 Mar 2015 17:46:36 +0000 (17:46 +0000)]
Work around lack of %zd printf format specifier in MSVC libs.
llvm-svn: 233569
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 17:41:24 +0000 (17:41 +0000)]
DebugInfo: Implement MDLocation::getInlinedAtScope()
Write `MDLocation::getInlinedAtScope()` and use it to re-implement
`DebugLoc::getScopeNode()` (and simplify `DISubprogram::Verify()`).
This follows the inlined-at linked list and returns the scope of the
deepest/last location.
llvm-svn: 233568
Tobias Grosser [Mon, 30 Mar 2015 17:28:57 +0000 (17:28 +0000)]
Update isl to
285e92aea
This is mostly a set of schedule tree enhancements which are not yet directly
useful to Polly.
llvm-svn: 233567
Tobias Grosser [Mon, 30 Mar 2015 17:22:28 +0000 (17:22 +0000)]
Delinearization of expressions that contain array size parameters
This allows us to delinerize code such as:
A[][n]
for (i
for (j
A[i][n-j-1] = ...
which would previously have been delinearize to an access A[i+1][-j-1].
To recover the correct access we apply the piecewise expression:
{ A[i][j] -> A[i-1][i+N]: i < 0; A[i][j] -> A[i][i]: i >= 0}
This approach generalizes to higher dimensions.
llvm-svn: 233566
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 17:21:38 +0000 (17:21 +0000)]
Verifier: Add operand checks for remaining debug info
llvm-svn: 233565
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 17:06:38 +0000 (17:06 +0000)]
DebugInfo: Simplify logic in DISubprogram::Verify(), NFC
Simplify the logic in `DISubprogram::Verify()` by using the new debug
info hierarchy directly instead of the `DebugLoc` wrapper.
llvm-svn: 233563
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 17:04:06 +0000 (17:04 +0000)]
Verifier: Loosen r233559 check for 'function:' field in MDSubprogram
Stop worrying about what the `function:` field is in `MDSubprogram`,
since it could be a bitcast [1]. Just check its type and leave it at
that.
[1]: http://lab.llvm.org:8080/green/job/clang-stage2-configure-Rlto_build/3540/
llvm-svn: 233562
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 16:37:48 +0000 (16:37 +0000)]
Verifier: Add operand checks for MDLexicalBlock
Add operand checks for `MDLexicalBlock` and `MDLexicalBlockFile`. Like
`MDLocalVariable` and `MDLocation`, these nodes always require a scope.
There was no test bitrot to fix here (just updated the serialization
tests in test/Assembler/mdlexicalblock.ll).
llvm-svn: 233561
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 16:21:28 +0000 (16:21 +0000)]
DebugInfo: Rename some testcases
Momentarily (but never in tree), the `scope:` field was called
`parent:`. Apparently a few testcases were left behind with "parent" in
the name, so rename them.
llvm-svn: 233560
Duncan P. N. Exon Smith [Mon, 30 Mar 2015 16:19:15 +0000 (16:19 +0000)]
Verifier: Check operands of MDSubprogram nodes
Check operands of `MDSubprogram`s in the verifier, and update the
accessors and factory functions to use more specific types.
There were a lot of broken testcases, which I fixed in r233466. If you
have out-of-tree tests for debug info, you probably need similar changes
to the ones I made there.
llvm-svn: 233559
Yaron Keren [Mon, 30 Mar 2015 16:10:39 +0000 (16:10 +0000)]
Trying to fix the Hexagon and debian-fast bots arm32_neon_vcnt_upgrade.ll test.
llvm-svn: 233558
Marshall Clow [Mon, 30 Mar 2015 16:07:11 +0000 (16:07 +0000)]
While testing Erik's code coverage scripts, I found a hole in the test suite - vector::assign where a reallocation was not required had no tests. Add some
llvm-svn: 233557
Tobias Grosser [Mon, 30 Mar 2015 16:05:48 +0000 (16:05 +0000)]
Add option to only run scop detection
llvm-svn: 233556
Yaron Keren [Mon, 30 Mar 2015 15:42:36 +0000 (15:42 +0000)]
Remove more superfluous .str() and replace std::string concatenation with Twine.
Following r233392, http://llvm.org/viewvc/llvm-project?rev=233392&view=rev.
llvm-svn: 233555
Sanjay Patel [Mon, 30 Mar 2015 15:31:32 +0000 (15:31 +0000)]
more space; NFC
llvm-svn: 233554
Simon Pilgrim [Mon, 30 Mar 2015 15:25:51 +0000 (15:25 +0000)]
[X86] Ensure integer domain on scalar i64 load/store stack folding tests. NFC
llvm-svn: 233553
Eric Fiselier [Mon, 30 Mar 2015 15:22:20 +0000 (15:22 +0000)]
[libcxx] Fix PR22771 - Support access control SFINAE in the library version of is_convertible.
Summary:
Currently the conversion check does not take place in a context where access control SFINAE is applied. This patch changes the context of the test expression so that SFINAE occurs if access control does not permit the conversion.
Related bug: https://llvm.org/bugs/show_bug.cgi?id=22771
Reviewers: mclow.lists, rsmith, dim
Reviewed By: dim
Subscribers: dim, rodrigc, emaste, cfe-commits
Differential Revision: http://reviews.llvm.org/D8461
llvm-svn: 233552
Rafael Espindola [Mon, 30 Mar 2015 15:16:30 +0000 (15:16 +0000)]
Fix Sphinx warning " Title underline too short.".
llvm-svn: 233551
Simon Atanasyan [Mon, 30 Mar 2015 15:07:16 +0000 (15:07 +0000)]
[ELF] Use override keyword instead of virtual
No functional changes.
llvm-svn: 233550
Simon Atanasyan [Mon, 30 Mar 2015 15:07:11 +0000 (15:07 +0000)]
[Mips] Do not use llvm::Optional for GP0 value and TLS section address
Use of llvm::Optional is redundant here. Initializing by default value 0
is enough.
No functional changes.
llvm-svn: 233549
Simon Atanasyan [Mon, 30 Mar 2015 15:07:05 +0000 (15:07 +0000)]
[Mips] Rename class method merge => mergeHeaderFlags
No functional changes.
llvm-svn: 233548
Robert Flack [Mon, 30 Mar 2015 14:12:17 +0000 (14:12 +0000)]
Replace sys.platform skips in tests with @skip decorators which check against remote platform.
Adds @skipIfPlatform and @skipUnlessPlatform decorators which will skip if /
unless the target platform is in the provided platform list.
Test Plan:
ninja check-lldb shows no regressions.
When running cross platform, tests which cannot run on the target platform are
skipped.
Differential Revision: http://reviews.llvm.org/D8665
llvm-svn: 233547
Tamas Berghammer [Mon, 30 Mar 2015 14:05:37 +0000 (14:05 +0000)]
Fix check in TestNoreturnUnwind
llvm-svn: 233546
Rafael Espindola [Mon, 30 Mar 2015 13:59:06 +0000 (13:59 +0000)]
Save a std::string.
The group names are always symbol names, so we can use a StringRef.
llvm-svn: 233545
Ulrich Weigand [Mon, 30 Mar 2015 13:50:21 +0000 (13:50 +0000)]
[SystemZ] Fix definition of IntMaxType / Int64Type
Like on other 64-bit platforms, Int64Type should be SignedLong
on SystemZ, not SignedLongLong as per default. This could cause
ABI incompatibilities in certain cases (e.g. name mangling).
llvm-svn: 233544
Ulrich Weigand [Mon, 30 Mar 2015 13:49:01 +0000 (13:49 +0000)]
[SystemZ] Fix some ABI corner cases
Running the GCC's inter-compiler ABI compatibility test suite uncovered
a couple of errors in clang's SystemZ ABI implementation. These all
affect only rare corner cases:
- Short vector types
GCC synthetic vector types defined with __attribute__ ((vector_size ...))
are always passed and returned by reference. (This is not documented in
the official ABI document, but is the de-facto ABI implemented by GCC.)
clang would do that only for vector sizes >= 16 bytes, but not for shorter
vector types.
- Float-like aggregates and empty bitfields
clang would consider any aggregate containing an empty bitfield as
first element to be a float-like aggregate. That's obviously wrong.
According to the ABI doc, the presence of an empty bitfield makes
an aggregate to be *not* float-like. However, due to a bug in GCC,
empty bitfields are ignored in C++; this patch changes clang to be
compatible with this "feature" of GCC.
- Float-like aggregates and va_arg
The va_arg implementation would mis-detect some aggregates as float-like
that aren't actually passed as such. This applies to aggregates that
have only a single element of type float or double, but using an aligned
attribute that increases the total struct size to more than 8 bytes.
This error occurred because the va_arg implement used to have an copy
of the float-like aggregate detection logic (i.e. it would call the
isFPArgumentType routine, but not perform the size check).
To simplify the logic, this patch removes the duplicated logic and
instead simply checks the (possibly coerced) LLVM argument type as
already determined by classifyArgumentType.
llvm-svn: 233543
Daniel Sanders [Mon, 30 Mar 2015 13:47:23 +0000 (13:47 +0000)]
[mips] Add support for 'ZC' inline assembly memory constraint.
Summary: Also add tests for 'R' and 'm'.
Reviewers: atanasyan
Reviewed By: atanasyan
Subscribers: cfe-commits
Differential Revision: http://reviews.llvm.org/D8449
llvm-svn: 233542
Ulrich Weigand [Mon, 30 Mar 2015 13:46:59 +0000 (13:46 +0000)]
[SystemZ] Fix LLVM crash on unoptimized code
Compiling the following function with -O0 would crash, since LLVM would
hit an assertion in getTestUnderMaskCond:
int test(unsigned long x)
{
return x >= 0 && x <= 15;
}
Fixed by detecting the case in the caller of getTestUnderMaskCond.
llvm-svn: 233541
Ulrich Weigand [Mon, 30 Mar 2015 13:46:25 +0000 (13:46 +0000)]
[SystemZ] Remove TargetMachine CPU auto-detection
As was done for X86 in r206094.
llvm-svn: 233540
Rafael Espindola [Mon, 30 Mar 2015 13:39:16 +0000 (13:39 +0000)]
Special case the creation of relocation sections.
These sections are never looked up and we know when have to create them. Use
that to save adding them to the regular map and avoid a symbol->string->symbol
conversion for the group symbol.
This also makes the implementation independent of the details of how unique
sections are implemented.
llvm-svn: 233539