Michael Gottesman [Fri, 20 Feb 2015 00:02:49 +0000 (00:02 +0000)]
[objc-arc-contract] We can not move retains over instructions which can not conservatively be proven to not decrement the retain's RCIdentity.
I also cleaned up the code to make it more understandable for mere mortals.
<rdar://problem/
19853758>
llvm-svn: 229937
Michael Gottesman [Fri, 20 Feb 2015 00:02:45 +0000 (00:02 +0000)]
[objc-arc] Add the predicate CanDecrementRefCount.
This is different from CanAlterRefCount since CanDecrementRefCount is
attempting to prove specifically whether or not an instruction can
decrement instead of the more general question of whether it can
decrement or increment.
llvm-svn: 229936
Eric Fiselier [Thu, 19 Feb 2015 23:57:46 +0000 (23:57 +0000)]
Fix incorrect locale mapping in config.py on OSX
llvm-svn: 229935
Jordan Rose [Thu, 19 Feb 2015 23:57:04 +0000 (23:57 +0000)]
[analyzer] RetainCountChecker: don't try to track ivars known to be nil.
We expect in general that any nil value has no retain count information
associated with it; violating this results in unexpected state unification
/later/ when we decide to throw the information away. Unexpectedly caching
out can lead to an assertion failure or crash.
rdar://problem/
19862648
llvm-svn: 229934
Duncan P. N. Exon Smith [Thu, 19 Feb 2015 23:56:07 +0000 (23:56 +0000)]
IR: Fix MDType fields from unsigned to uint64_t
When trying to match the current schema with the new debug info
hierarchy, I downgraded `SizeInBits`, `AlignInBits` and `OffsetInBits`
to 32-bits (oops!). Caught this while testing my upgrade script to move
the hierarchy into place. Bump it back up to 64-bits and update tests.
llvm-svn: 229933
Ahmed Bougacha [Thu, 19 Feb 2015 23:52:41 +0000 (23:52 +0000)]
[ARM] Re-re-apply VLD1/VST1 base-update combine.
This re-applies r223862, r224198, r224203, and r224754, which were
reverted in r228129 because they exposed Clang misalignment problems
when self-hosting.
The combine caused the crashes because we turned ISD::LOAD/STORE nodes
to ARMISD::VLD1/VST1_UPD nodes. When selecting addressing modes, we
were very lax for the former, and only emitted the alignment operand
(as in "[r1:128]") when it was larger than the standard alignment of
the memory type.
However, for ARMISD nodes, we just used the MMO alignment, no matter
what. In our case, we turned ISD nodes to ARMISD nodes, and this
caused the alignment operands to start being emitted.
And that's how we exposed alignment problems that were ignored before
(but I believe would have been caught with SCTRL.A==1?).
To fix this, we can just mirror the hack done for ISD nodes: only
take into account the MMO alignment when the access is overaligned.
Original commit message:
We used to only combine intrinsics, and turn them into VLD1_UPD/VST1_UPD
when the base pointer is incremented after the load/store.
We can do the same thing for generic load/stores.
Note that we can only combine the first load/store+adds pair in
a sequence (as might be generated for a v16f32 load for instance),
because other combines turn the base pointer addition chain (each
computing the address of the next load, from the address of the last
load) into independent additions (common base pointer + this load's
offset).
rdar://
19717869, rdar://
14062261.
llvm-svn: 229932
Eric Christopher [Thu, 19 Feb 2015 23:52:35 +0000 (23:52 +0000)]
Only use the initialized MCInstrInfo if it's been initialized already
during SetupMachineFunction. This is also the single use of MII
and it'll be changing to TargetInstrInfo (which is MachineFunction
based) in the next commit here.
llvm-svn: 229931
Duncan P. N. Exon Smith [Thu, 19 Feb 2015 23:48:17 +0000 (23:48 +0000)]
DebugInfo: Match Name and DisplayName in testcase
There's no way for `DIBuilder` to create a subprogram or global variable
where `getName()` and `getDisplayName()` give different answers. This
testcase managed to achieve the feat though. This was probably just
left behind in some sort of upgrade along the way.
llvm-svn: 229930
Ahmed Bougacha [Thu, 19 Feb 2015 23:30:37 +0000 (23:30 +0000)]
[ARM] Minor cleanup to CombineBaseUpdate. NFC.
In preparation for a future patch:
- rename isLoad to isLoadOp: the former is confusing, and can be taken
to refer to the fact that the node is an ISD::LOAD. (it isn't, yet.)
- change formatting here and there.
- add some comments.
- const-ify bools.
llvm-svn: 229929
Eric Christopher [Thu, 19 Feb 2015 23:29:42 +0000 (23:29 +0000)]
Migrate away a use of the subtarget (and TargetMachine) from
AsmPrinterDwarf since the information is on the MCRegisterInfo
via the MCContext and MMI that we already have on the AsmPrinter.
llvm-svn: 229928
Eric Fiselier [Thu, 19 Feb 2015 23:26:54 +0000 (23:26 +0000)]
Move to using -fdiagnostics-color=always on both GCC and Clang
llvm-svn: 229927
Duncan P. N. Exon Smith [Thu, 19 Feb 2015 23:25:21 +0000 (23:25 +0000)]
IR: Add missing null operand to MDSubroutineType
Add missing `nullptr` from `MDSubroutineType`'s operands for
`MDCompositeTypeBase::getIdentifier()` (and add tests for all the other
unused fields). This highlights just how crazy it is that
`MDSubroutineType` inherits from `MDCompositeTypeBase`.
llvm-svn: 229926
Ahmed Bougacha [Thu, 19 Feb 2015 23:13:10 +0000 (23:13 +0000)]
[CodeGen] Use ArrayRef instead of std::vector&. NFC.
The former lets us use SmallVectors. Do so in ARM and AArch64.
llvm-svn: 229925
Alexey Samsonov [Thu, 19 Feb 2015 22:56:49 +0000 (22:56 +0000)]
[Sanitizer] Drop LibIgnore dependency on SuppressionContext. NFC.
Let each LibIgnore user (for now it's only TSan) manually go
through SuppressionContext and pass ignored library templates to
LibIgnore.
llvm-svn: 229924
Alexey Samsonov [Thu, 19 Feb 2015 22:56:47 +0000 (22:56 +0000)]
[Sanitizer] Move TemplateMatch() to sanitizer_common.cc. NFC.
llvm-svn: 229923
Hans Wennborg [Thu, 19 Feb 2015 22:39:24 +0000 (22:39 +0000)]
Don't dllexport inline methods when targeting MinGW.
MinGW neither imports nor exports such methods. The import bit was
committed earlier, in r221154, and this takes care of the export part.
This also partially fixes PR22591.
llvm-svn: 229922
Dimitry Andric [Thu, 19 Feb 2015 22:32:33 +0000 (22:32 +0000)]
Add support for analyzing FreeBSD kernel printf extensions.
This adds a new __freebsd_kprintf__ format string type, which enables
checking when used in __attribute__((format(...))) attributes. It can
check the FreeBSD kernel specific %b, %D, %r and %y specifiers, using
existing diagnostic messages. Also adds test cases for all these
specifiers.
Differential Revision: http://reviews.llvm.org/D7154
llvm-svn: 229921
Tobias Grosser [Thu, 19 Feb 2015 22:16:12 +0000 (22:16 +0000)]
Fix formatting
llvm-svn: 229920
Fariborz Jahanian [Thu, 19 Feb 2015 21:52:41 +0000 (21:52 +0000)]
[Objective-C Sema]. Don't warn about use of
property accessors in @selector not implemented
because they will be auto-synthesized. rdar://
16607480
llvm-svn: 229919
David Majnemer [Thu, 19 Feb 2015 21:41:48 +0000 (21:41 +0000)]
Itanium ABI: Pack expansions change the arity of expressions to unknown
llvm-svn: 229918
Eric Christopher [Thu, 19 Feb 2015 21:29:51 +0000 (21:29 +0000)]
MCTargetOptions reside on the TargetMachine that we always have via
TargetOptions.
llvm-svn: 229917
Eric Christopher [Thu, 19 Feb 2015 21:24:23 +0000 (21:24 +0000)]
Remove a call to TargetMachine::getSubtarget from the inline
asm support in the asm printer. If we can get a subtarget from
the machine function then we should do so, otherwise we can
go ahead and create a default one since we're at the module
level.
llvm-svn: 229916
Reid Kleckner [Thu, 19 Feb 2015 21:13:45 +0000 (21:13 +0000)]
Avoid using a COMDAT for sized delete on MachO
llvm-svn: 229915
Colin LeMahieu [Thu, 19 Feb 2015 21:10:50 +0000 (21:10 +0000)]
[Hexagon] Moving remaining methods off of HexagonMCInst in to HexagonMCInstrInfo and eliminating HexagonMCInst class.
llvm-svn: 229914
Reid Kleckner [Thu, 19 Feb 2015 21:01:34 +0000 (21:01 +0000)]
Put the implicit weak sized deallocation funciton in C++14 in a comdat
Fixes PR22635.
llvm-svn: 229913
Greg Fitzgerald [Thu, 19 Feb 2015 20:42:23 +0000 (20:42 +0000)]
Fix heap-buffer-overflow bugs identified by the Address Sanitizer
Differential Revision: http://reviews.llvm.org/D7733
llvm-svn: 229912
Benjamin Kramer [Thu, 19 Feb 2015 20:24:04 +0000 (20:24 +0000)]
MC: Allow multiple comma-separated expressions on the .uleb128 directive.
For compatiblity with GNU as. Binutils documents this as
'.uleb128 expressions'. Subtle, isn't it?
llvm-svn: 229911
Ben Langmuir [Thu, 19 Feb 2015 20:23:22 +0000 (20:23 +0000)]
Revert "Mangle the IsSystem bit into the .pcm file name"
While I investigate some possible problems with this patch.
This reverts commit r228966
llvm-svn: 229910
Argyrios Kyrtzidis [Thu, 19 Feb 2015 20:12:20 +0000 (20:12 +0000)]
[PCH/Modules] Check that the specific module cache path the PCH was built with, is the same as
the one in the current compiler invocation. If they differ reject the PCH.
This protects against the badness occurring from getting modules loaded from different module caches (see crashes).
rdar://
19889860
llvm-svn: 229909
Benjamin Kramer [Thu, 19 Feb 2015 20:04:02 +0000 (20:04 +0000)]
SSAUpdater: Use range-based for. NFC.
llvm-svn: 229908
Eric Christopher [Thu, 19 Feb 2015 19:52:25 +0000 (19:52 +0000)]
Remove unused argument from emitInlineAsmStart.
llvm-svn: 229907
Michael Gottesman [Thu, 19 Feb 2015 19:51:36 +0000 (19:51 +0000)]
[objc-arc] Convert the bodies of ARCInstKind predicates into covered switches.
This is much better than the previous manner of just using
short-curcuiting booleans from:
1. A "naive" efficiency perspective: we do not have to rely on the
compiler to change the short circuiting boolean operations into a
switch.
2. An understanding perspective by making the implicit behavior of
negative predicates explicit.
3. A maintainability perspective through the covered switch flag making
it easy to know where to update code when adding new ARCInstKinds.
llvm-svn: 229906
Michael Gottesman [Thu, 19 Feb 2015 19:51:32 +0000 (19:51 +0000)]
[objc-arc] Change the InstructionClass to be an enum class called ARCInstKind.
I also renamed ObjCARCUtil.cpp -> ARCInstKind.cpp. That file only contained
items related to ARCInstKind anyways.
llvm-svn: 229905
Chris Bieneman [Thu, 19 Feb 2015 19:50:52 +0000 (19:50 +0000)]
Checking if TARGET_OS_IPHONE is defined isn't good enough for 10.7 and earlier.
Older versions of the TargetConditionals header always defined TARGET_OS_IPHONE to something (0 or 1), so we need to test not only for the existence but also if it is 1.
This resolves PR22631.
llvm-svn: 229904
Colin LeMahieu [Thu, 19 Feb 2015 19:49:27 +0000 (19:49 +0000)]
[Hexagon] Moving more functions off of HexagonMCInst and in to HexagonMCInstrInfo.
llvm-svn: 229903
David Majnemer [Thu, 19 Feb 2015 19:35:18 +0000 (19:35 +0000)]
CodeGen: Weak reference temporaries belong in a COMDAT
llvm-svn: 229902
Ilia K [Thu, 19 Feb 2015 19:26:52 +0000 (19:26 +0000)]
Add -stack-select-frame command (MI)
Summary:
Add -stack-select-frame command + test.
All tests pass on OS X.
Reviewers: abidh, emaste, clayborg
Reviewed By: clayborg
Subscribers: lldb-commits, emaste, clayborg, abidh
Differential Revision: http://reviews.llvm.org/D7765
llvm-svn: 229901
David Majnemer [Thu, 19 Feb 2015 19:25:17 +0000 (19:25 +0000)]
CodeGen: static constexpr data members should have a linkonce_odr init
Classes can be defined in multiple translation units. This means that
the static constexpr data members should have identical initializers in
all translation units. Implement this by giving the reference temporary
linkonce_odr linkage.
llvm-svn: 229900
Adam Nemet [Thu, 19 Feb 2015 19:15:21 +0000 (19:15 +0000)]
[LoopAccesses] Change LAA:getInfo to return a constant reference
As expected, this required a few more const-correctness fixes.
Based on Hal's feedback on D7684.
llvm-svn: 229899
Adam Nemet [Thu, 19 Feb 2015 19:15:19 +0000 (19:15 +0000)]
[LoopAccesses] Add -analyze support
The LoopInfo in combination with depth_first is used to enumerate the
loops.
Right now -analyze is not yet complete. It only prints the result of
the analysis, the report and the run-time checks. Printing the unsafe
depedences will require a bit more reshuffling which I'd like to do in a
follow-on to this patchset. Unsafe dependences are currently checked
via -debug-only=loop-accesses in the new test.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229898
Adam Nemet [Thu, 19 Feb 2015 19:15:15 +0000 (19:15 +0000)]
[LoopAccesses] Split out LoopAccessReport from VectorizerReport
The only difference between these two is that VectorizerReport adds a
vectorizer-specific prefix to its messages. When LAA is used in the
vectorizer context the prefix is added when we promote the
LoopAccessReport into a VectorizerReport via one of the constructors.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229897
Adam Nemet [Thu, 19 Feb 2015 19:15:13 +0000 (19:15 +0000)]
[LoopAccesses] Add missing const to APIs in VectorizationReport
When I split out LoopAccessReport from this, I need to create some temps
so constness becomes necessary.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229896
Adam Nemet [Thu, 19 Feb 2015 19:15:10 +0000 (19:15 +0000)]
[LoopAccesses] Add canAnalyzeLoop
This allows the analysis to be attempted with any loop. This feature
will be used with -analysis. (LV only requests the analysis on loops
that have already satisfied these tests.)
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229895
Adam Nemet [Thu, 19 Feb 2015 19:15:07 +0000 (19:15 +0000)]
[LoopAccesses] Change debug messages from LV to LAA
Also add pass name as an argument to VectorizationReport::emitAnalysis.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229894
Adam Nemet [Thu, 19 Feb 2015 19:15:04 +0000 (19:15 +0000)]
[LoopAccesses] Create the analysis pass
This is a function pass that runs the analysis on demand. The analysis
can be initiated by querying the loop access info via LAA::getInfo. It
either returns the cached info or runs the analysis.
Symbolic stride information continues to reside outside of this analysis
pass. We may move it inside later but it's not a priority for me right
now. The idea is that Loop Distribution won't support run-time stride
checking at least initially.
This means that when querying the analysis, symbolic stride information
can be provided optionally. Whether stride information is used can
invalidate the cache entry and rerun the analysis. Note that if the
loop does not have any symbolic stride, the entry should be preserved
across Loop Distribution and LV.
Since currently the only user of the pass is LV, I just check that the
symbolic stride information didn't change when using a cached result.
On the LV side, LoopVectorizationLegality requests the info object
corresponding to the loop from the analysis pass. A large chunk of the
diff is due to LAI becoming a pointer from a reference.
A test will be added as part of the -analyze patch.
Also tested that with AVX, we generate identical assembly output for the
testsuite (including the external testsuite) before and after.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229893
Adam Nemet [Thu, 19 Feb 2015 19:15:00 +0000 (19:15 +0000)]
[LoopAccesses] Cache the result of canVectorizeMemory
LAA will be an on-demand analysis pass, so we need to cache the result
of the analysis. canVectorizeMemory is renamed to analyzeLoop which
computes the result. canVectorizeMemory becomes the query function for
the cached result.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229892
Adam Nemet [Thu, 19 Feb 2015 19:14:56 +0000 (19:14 +0000)]
[LoopAccesses] Stash the report from the analysis rather than emitting it
The transformation passes will query this and then emit them as part of
their own report. The currently only user LV is modified to do just
that.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229891
Adam Nemet [Thu, 19 Feb 2015 19:14:52 +0000 (19:14 +0000)]
[LoopAccesses] Make VectorizerParams global + fix for cyclic dep
As LAA is becoming a pass, we can no longer pass the params to its
constructor. This changes the command line flags to have external
storage. These can now be accessed both from LV and LAA.
VectorizerParams is moved out of LoopAccessInfo in order to shorten the
code to access it.
This commits also has the fix (D7731) to the break dependence cycle
between the analysis and vector libraries.
This is part of the patchset that converts LoopAccessAnalysis into an
actual analysis pass.
llvm-svn: 229890
Adam Nemet [Thu, 19 Feb 2015 19:14:34 +0000 (19:14 +0000)]
Revert "Reformat."
This reverts commit r229651.
I'd like to ultimately revert r229650 but this reformat stands in the
way. I'll reformat the affected files once the the loop-access pass is
fully committed.
llvm-svn: 229889
David Blaikie [Thu, 19 Feb 2015 19:06:04 +0000 (19:06 +0000)]
[orcjit] Include CMake support for the fully_lazy example and fix the build
Not sure if/how to make the CMake build use C++14 for the examples, so
let's stick to C++11 for now.
llvm-svn: 229888
Filipe Cabecinhas [Thu, 19 Feb 2015 19:00:33 +0000 (19:00 +0000)]
Make the _mm256_insert_epi64 definition more consistent
Use long long for the epi64 argument, like the other intrinsics.
NFC since this is only defined in 64-bit mode, not in 32-bit.
Fix suggested by H. J. Lu!
llvm-svn: 229886
Colin LeMahieu [Thu, 19 Feb 2015 19:00:00 +0000 (19:00 +0000)]
[Hexagon] Creating HexagonMCInstrInfo namespace as landing zone for static functions detached from HexagonMCInst.
llvm-svn: 229885
Eric Christopher [Thu, 19 Feb 2015 18:46:25 +0000 (18:46 +0000)]
Fix grammar in documentation.
Patch by Ralph Campbell!
llvm-svn: 229884
Eric Christopher [Thu, 19 Feb 2015 18:46:23 +0000 (18:46 +0000)]
Update and remove a few calls to TargetMachine::getSubtargetImpl
out of the asm printer.
llvm-svn: 229883
Kostya Serebryany [Thu, 19 Feb 2015 18:45:37 +0000 (18:45 +0000)]
[fuzzer] split main() into FuzzerDriver() that takes a callback as a parameter and a tiny main() in a separate file
llvm-svn: 229882
Ben Langmuir [Thu, 19 Feb 2015 18:22:35 +0000 (18:22 +0000)]
Assume the original file is created before release in LockFileManager
This is true in clang, and let's us remove the problematic code that
waits around for the original file and then times out if it doesn't get
created in short order. This caused any 'dead' lock file or legitimate
time out to cause a cascade of timeouts in any processes waiting on the
same lock (even if they only just showed up).
llvm-svn: 229881
Kostya Serebryany [Thu, 19 Feb 2015 18:21:12 +0000 (18:21 +0000)]
[fuzzer] properly annotate fallthrough, add one more entry to FAQ
llvm-svn: 229880
Johannes Doerfert [Thu, 19 Feb 2015 18:11:50 +0000 (18:11 +0000)]
[REFACTOR] Simplify the SCoP detection interface a bit
llvm-svn: 229879
Pavel Labath [Thu, 19 Feb 2015 18:10:55 +0000 (18:10 +0000)]
Factor out common URL handling code in PlatformRemoteGDBServer
Summary:
this also gets rid of a compiler warning in release builds by using a dynamically allocated
buffer. Therefore, a size assertion is not necessary (and probably should have been an error in
the first place).
Reviewers: tberghammer
Subscribers: lldb-commits
Differential Revision: http://reviews.llvm.org/D7751
llvm-svn: 229878
Johannes Doerfert [Thu, 19 Feb 2015 18:09:39 +0000 (18:09 +0000)]
[NFC] Generalize getIslCompatibleName interface.
llvm-svn: 229877
Oleksiy Vyalov [Thu, 19 Feb 2015 17:58:04 +0000 (17:58 +0000)]
Prevent LLGS from crashing when exiting - make NativeProcessLinux to wait until ThreadStateCoordinator is fully stopped before entering ~NativeProcessLinux.
http://reviews.llvm.org/D7692
llvm-svn: 229875
Colin LeMahieu [Thu, 19 Feb 2015 17:38:39 +0000 (17:38 +0000)]
[Hexagon] Removing static variable holding MCInstrInfo.
llvm-svn: 229872
Benjamin Kramer [Thu, 19 Feb 2015 17:19:43 +0000 (17:19 +0000)]
LSR: Move set instead of copying. NFC.
llvm-svn: 229871
Sanjay Patel [Thu, 19 Feb 2015 16:59:11 +0000 (16:59 +0000)]
add X86 load folding tests for unary math ops
X86 load folding is fragile; eg, the tests here
don't work without AVX even though they should. This
is because we have a mix of tablegen patterns that have
been added over time, and we have a load folding table
used by the peephole optimizer that has to be kept in
sync with the ever-changing ISA and tablegen defs.
llvm-svn: 229870
John Thompson [Thu, 19 Feb 2015 16:47:27 +0000 (16:47 +0000)]
Added module map coverage support, extracted from module-map-checker.
llvm-svn: 229869
Ilia K [Thu, 19 Feb 2015 16:33:16 +0000 (16:33 +0000)]
Fix usage of m_bThreadInvalid in -thread-info (MI)
llvm-svn: 229868
Tobias Grosser [Thu, 19 Feb 2015 16:24:57 +0000 (16:24 +0000)]
Drop left over code from FinalRead statement removal.
We removed FinalRead statements in r152319.
llvm-svn: 229867
Marshall Clow [Thu, 19 Feb 2015 16:17:46 +0000 (16:17 +0000)]
Make basic_streambuf::xsputn write characters in chunks whenever possible, instead of one at a time. References PR#10193
llvm-svn: 229866
Daniel Jasper [Thu, 19 Feb 2015 16:14:18 +0000 (16:14 +0000)]
clang-format: [js] Support ES6 module exports.
Patch by Martin Probst, thank you!
llvm-svn: 229865
Rafael Espindola [Thu, 19 Feb 2015 16:08:20 +0000 (16:08 +0000)]
Avoid conversion to float when creating ConstantDataArray/ConstantDataVector.
Patch by Raoux, Thomas F!
llvm-svn: 229864
Daniel Jasper [Thu, 19 Feb 2015 16:07:32 +0000 (16:07 +0000)]
clang-format: [js] Support ES6 module imports.
Patch by Martin Probst.
llvm-svn: 229863
Daniel Jasper [Thu, 19 Feb 2015 16:03:16 +0000 (16:03 +0000)]
clang-format: [js] Do not fall through for JS structural elements.
Patch by Martin Probst. Thank you.
llvm-svn: 229862
Benjamin Kramer [Thu, 19 Feb 2015 15:26:17 +0000 (15:26 +0000)]
Demote vectors to arrays. No functionality change.
llvm-svn: 229861
Timur Iskhodzhanov [Thu, 19 Feb 2015 15:25:26 +0000 (15:25 +0000)]
[ASan/Win] Thread sanitizer common interface through asan_win_dll_thunk.cc
llvm-svn: 229860
Chandler Carruth [Thu, 19 Feb 2015 15:21:57 +0000 (15:21 +0000)]
[x86] Delete still more piles of complex code now that we have a good
systematic lowering of v8i16.
This required a slight strategy shift to prefer unpack lowerings in more
places. While this isn't a cut-and-dry win in every case, it is in the
overwhelming majority. There are only a few places where the old
lowering would probably be a touch faster, and then only by a small
margin.
In some cases, this is yet another significant improvement.
llvm-svn: 229859
Timur Iskhodzhanov [Thu, 19 Feb 2015 15:15:33 +0000 (15:15 +0000)]
[Sanitizers] Move the common sanitizer interface from sanitizer_internal_defs.h to a new sanitizer_interface_internal.h file
Reviewed at http://reviews.llvm.org/D7758
llvm-svn: 229858
Ilia K [Thu, 19 Feb 2015 15:14:17 +0000 (15:14 +0000)]
Add -stack-info-frame command (MI)
Summary:
Add -stack-info-frame command + test.
All tests pass on OS X.
Reviewers: emaste, clayborg, abidh
Reviewed By: abidh
Subscribers: lldb-commits, clayborg, emaste, abidh
Differential Revision: http://reviews.llvm.org/D7750
llvm-svn: 229857
Chandler Carruth [Thu, 19 Feb 2015 15:06:13 +0000 (15:06 +0000)]
[x86] Teach the unpack lowering how to lower with an initial unpack in
addition to lowering to trees rooted in an unpack.
This saves shuffles and or registers in many various ways, lets us
handle another class of v4i32 shuffles pre SSE4.1 without domain
crosses, etc.
llvm-svn: 229856
John Thompson [Thu, 19 Feb 2015 14:31:48 +0000 (14:31 +0000)]
Pruned some unneeded code and comments.
llvm-svn: 229855
Tobias Grosser [Thu, 19 Feb 2015 14:28:36 +0000 (14:28 +0000)]
Add missing comments to member variables
Reported-by: Johannes Doerfert
llvm-svn: 229854
Chandler Carruth [Thu, 19 Feb 2015 14:08:24 +0000 (14:08 +0000)]
[x86] Dramatically improve v8i16 shuffle lowering by not using its
terribly complex partial blend logic.
This code path was one of the more complex and bug prone when it first
went in and it hasn't faired much better. Ultimately, with the simpler
basis for unpack lowering and support bit-math blending, this is
completely obsolete. In the worst case without this we generate
different but equivalent instructions. However, in many cases we
generate much better code. This is especially true when blends or pshufb
is available.
This does expose one (minor) weakness of the unpack lowering that I'll
try to address.
In case you were wondering, this is actually a big part of what I've
been trying to pull off in the recent string of commits.
llvm-svn: 229853
Nathan Sidwell [Thu, 19 Feb 2015 14:03:22 +0000 (14:03 +0000)]
FIX PR 18432, default args, friends & late-parsed members.
Sema::MergeCXXFunctionDecl: propagate hasUnparsedDefaultArg to new decl.
Parser::HandleMemberFunctionDeclDelays: check hasUnparsedDefaultArg
flag.
Parser::ParseLexedMethodDeclaration: handle inherited unparsed default
arg case.
llvm-svn: 229852
Chandler Carruth [Thu, 19 Feb 2015 13:56:49 +0000 (13:56 +0000)]
[x86] Remove the final fallback in the v8i16 lowering that isn't really
needed, and significantly improve the SSSE3 path.
This makes the new strategy much more clear. If we can blend, we just go
with that. If we can't blend, we try to permute into an unpack so
that we handle cases where the unpack doing the blend also simplifies
the shuffle. If that fails and we've got SSSE3, we now call into
factored-out pshufb lowering code so that we leverage the fact that
pshufb can set up a blend for us while shuffling. This generates great
code, especially because we *know* we don't have a fast blend at this
point. Finally, we fall back on decomposing into permutes and blends
because we do at least have a bit-math-based blend if we need to use
that.
This pretty significantly improves some of the v8i16 code paths. We
never need to form pshufb for the single-input shuffles because we have
effective target-specific combines to form it there, but we were missing
its effectiveness in the blends.
llvm-svn: 229851
Anton Yartsev [Thu, 19 Feb 2015 13:36:20 +0000 (13:36 +0000)]
[analyzer] Different handling of alloca().
+ separate bug report for "Free alloca()" error to be able to customize checkers responsible for this error.
+ Muted "Free alloca()" error for NewDelete checker that is not responsible for c-allocated memory, turned on for unix.MismatchedDeallocator checker.
+ RefState for alloca() - to be able to detect usage of zero-allocated memory by upcoming ZeroAllocDereference checker.
+ AF_Alloca family to handle alloca() consistently - keep proper family in RefState, handle 'alloca' by getCheckIfTracked() facility, etc.
+ extra tests.
llvm-svn: 229850
Chandler Carruth [Thu, 19 Feb 2015 13:15:12 +0000 (13:15 +0000)]
[x86] Simplify the pre-SSSE3 v16i8 lowering significantly by decomposing
them into permutes and a blend with the generic decomposition logic.
This works really well in almost every case and lets the code only
manage the expansion of a single input into two v8i16 vectors to perform
the actual shuffle. The blend-based merging is often much nicer than the
pack based merging that this replaces. The only place where it isn't we
end up blending between two packs when we could do a single pack. To
handle that case, just teach the v2i64 lowering to handle these blends
by digging out the operands.
With this we're down to only really random permutations that cause an
explosion of instructions.
llvm-svn: 229849
Jean-Daniel Dupas [Thu, 19 Feb 2015 12:38:54 +0000 (12:38 +0000)]
[Mach-O] Rename enum typename for consistency. NFC
Typename shouldn't mix camel case and underscore.
Thanks to Rui for the remark.
llvm-svn: 229848
Chandler Carruth [Thu, 19 Feb 2015 12:10:37 +0000 (12:10 +0000)]
[x86] Remove the insanely over-aggressive unpack lowering strategy for
v16i8 shuffles, and replace it with new facilities.
This uses precise patterns to match exact unpacks, and the new
generalized unpack lowering only when we detect a case where we will
have to shuffle both inputs anyways and they terminate in exactly
a blend.
This fixes all of the blend horrors that I uncovered by always lowering
blends through the vector shuffle lowering. It also removes *sooooo*
much of the crazy instruction sequences required for v16i8 lowering
previously. Much cleaner now.
The only "meh" aspect is that we sometimes use pshufb+pshufb+unpck when
it would be marginally nicer to use pshufb+pshufb+por. However, the
difference there is *tiny*. In many cases its a win because we re-use
the pshufb mask. In others, we get to avoid the pshufb entirely. I've
left a FIXME, but I'm dubious we can really do better than this. I'm
actually pretty happy with this lowering now.
For SSE2 this exposes some horrors that were really already there. Those
will have to fixed by changing a different path through the v16i8
lowering.
llvm-svn: 229846
Jozef Kolek [Thu, 19 Feb 2015 11:51:32 +0000 (11:51 +0000)]
[mips][microMIPS] Make usage of AND16, OR16 and XOR16 by code generator
Differential Revision: http://reviews.llvm.org/D7611
llvm-svn: 229845
Chandler Carruth [Thu, 19 Feb 2015 11:43:37 +0000 (11:43 +0000)]
[x86] The SELECT x86 DAG combine also does legalization. It used to rely
on things not being marked as either custom or legal, but we now do
custom lowering of more VSELECT nodes. To cope with this, manually
replicate the legality tests here. These have to stay in sync with the
set of tests used in the custom lowering of VSELECT.
Ideally, we wouldn't do any of this combine-based-legalization when we
have an actual custom legalization step for VSELECT, but I'm not going
to be able to rewrite all of that today.
I don't have a test case for this currently, but it was found when
compiling a number of the test-suite benchmarks. I'll try to reduce
a test case and add it.
This should at least fix the test-suite fallout on build bots.
llvm-svn: 229844
Pavel Labath [Thu, 19 Feb 2015 11:41:12 +0000 (11:41 +0000)]
Silence unused variable warnings in release builds
llvm-svn: 229843
Igor Laevsky [Thu, 19 Feb 2015 11:39:04 +0000 (11:39 +0000)]
Add few simple tests to check statepoint placement for invoke instructions.
Differential Revision: http://reviews.llvm.org/D7535
llvm-svn: 229842
Michael Kuperstein [Thu, 19 Feb 2015 11:38:11 +0000 (11:38 +0000)]
Reverting r229831 due to multiple ARM/PPC/MIPS build-bot failures.
llvm-svn: 229841
Igor Laevsky [Thu, 19 Feb 2015 11:28:47 +0000 (11:28 +0000)]
Implement invoke statepoint verification.
Differential Revision: http://reviews.llvm.org/D7366
llvm-svn: 229840
Yaron Keren [Thu, 19 Feb 2015 11:21:11 +0000 (11:21 +0000)]
Spelling correction.
llvm-svn: 229839
Igor Laevsky [Thu, 19 Feb 2015 11:02:11 +0000 (11:02 +0000)]
Add invoke related functionality into StatepointSite classes.
Differential Revision: http://reviews.llvm.org/D7364
llvm-svn: 229838
Elena Demikhovsky [Thu, 19 Feb 2015 10:48:04 +0000 (10:48 +0000)]
AVX-512: Full implementation for VRNDSCALESS/SD instructions and intrinsics.
llvm-svn: 229837
Chandler Carruth [Thu, 19 Feb 2015 10:46:52 +0000 (10:46 +0000)]
[x86] Add support for bit-wise blending and use it in the v8 and v16
lowering paths. I'm going to be leveraging this to simplify a lot of the
overly complex lowering of v8 and v16 shuffles in pre-SSSE3 modes.
Sadly, this isn't profitable on v4i32 and v2i64. There, the float and
double blending instructions for pre-SSE4.1 are actually pretty good,
and we can't beat them with bit math. And once SSE4.1 comes around we
have direct blending support and this ceases to be relevant.
Also, some of the test cases look odd because the domain fixer
canonicalizes these to floating point domain. That's OK, it'll use the
integer domain when it matters and some day I may be able to update
enough of LLVM to canonicalize the other way.
This restores almost all of the regressions from teaching x86's vselect
lowering to always use vector shuffle lowering for blends. The remaining
problems are because the v16 lowering path is still doing crazy things.
I'll be re-arranging that strategy in more detail in subsequent commits
to finish recovering the performance here.
llvm-svn: 229836
Chandler Carruth [Thu, 19 Feb 2015 10:36:19 +0000 (10:36 +0000)]
[x86,sdag] Two interrelated changes to the x86 and sdag code.
First, don't combine bit masking into vector shuffles (even ones the
target can handle) once operation legalization has taken place. Custom
legalization of vector shuffles may exist for these patterns (making the
predicate return true) but that custom legalization may in some cases
produce the exact bit math this matches. We only really want to handle
this prior to operation legalization.
However, the x86 backend, in a fit of awesome, relied on this. What it
would do is mark VSELECTs as expand, which would turn them into
arithmetic, which this would then match back into vector shuffles, which
we would then lower properly. Amazing.
Instead, the second change is to teach the x86 backend to directly form
vector shuffles from VSELECT nodes with constant conditions, and to mark
all of the vector types we support lowering blends as shuffles as custom
VSELECT lowering. We still mark the forms which actually support
variable blends as *legal* so that the custom lowering is bypassed, and
the legal lowering can even be used by the vector shuffle legalization
(yes, i know, this is confusing. but that's how the patterns are
written).
This makes the VSELECT lowering much more sensible, and in fact should
fix a bunch of bugs with it. However, as you'll see in the test cases,
right now what it does is point out the *hilarious* deficiency of the
new vector shuffle lowering when it comes to blends. Fortunately, my
very next patch fixes that. I can't submit it yet, because that patch,
somewhat obviously, forms the exact and/or pattern that the DAG combine
is matching here! Without this patch, teaching the vector shuffle
lowering to produce the right code infloops in the DAG combiner. With
this patch alone, we produce terrible code but at least lower through
the right paths. With both patches, all the regressions here should be
fixed, and a bunch of the improvements (like using 2 shufps with no
memory loads instead of 2 andps with memory loads and an orps) will
stay. Win!
There is one other change worth noting here. We had hilariously wrong
vectorization cost estimates for vselect because we fell through to the
code path that assumed all "expand" vector operations are scalarized.
However, the "expand" lowering of VSELECT is vector bit math, most
definitely not scalarized. So now we go back to the correct if horribly
naive cost of "1" for "not scalarized". If anyone wants to add actual
modeling of shuffle costs, that would be cool, but this seems an
improvement on its own. Note the removal of 16 and 32 "costs" for doing
a blend. Even in SSE2 we can blend in fewer than 16 instructions. ;] Of
course, we don't right now because of OMG bad code, but I'm going to fix
that. Next patch. I promise.
llvm-svn: 229835
Daniel Jasper [Thu, 19 Feb 2015 09:56:13 +0000 (09:56 +0000)]
Make -fmodules-decluse and -fmodules-strict-decluse compatible options.
They don't actually influence the result of the module compilation.
llvm-svn: 229834
Mohit K. Bhakkad [Thu, 19 Feb 2015 09:14:43 +0000 (09:14 +0000)]
[LSan][MIPS64] Enable LSan testing for mips64/mips64el
Patch by Sagar Thakur
Reviewers: petarj, earthdok, kcc.
Subscribers: samsonov, dsanders, mohit.bhakkad, Anand.Takale, llvm-commits.
Differential Revision: http://reviews.llvm.org/D7124
llvm-svn: 229833