Walter Lee [Thu, 16 Nov 2017 17:03:25 +0000 (17:03 +0000)]
[asan] Ensure that the minimum redzone is at least SHADOW_GRANULARITY
This is required by the Asan run-time.
Differential Revision: https://reviews.llvm.org/D39472
llvm-svn: 318422
Walter Lee [Thu, 16 Nov 2017 17:03:00 +0000 (17:03 +0000)]
[asan] Fix small X86_64 ShadowOffset for non-default shadow scale
The requirement is that shadow memory must be aligned to page
boundaries (4k in this case). Use a closed form equation that always
satisfies this requirement.
Differential Revision: https://reviews.llvm.org/D39471
llvm-svn: 318421
Sanjay Patel [Thu, 16 Nov 2017 17:01:09 +0000 (17:01 +0000)]
[InstCombine] regenerate test checks; NFC
llvm-svn: 318420
Rafael Espindola [Thu, 16 Nov 2017 17:00:48 +0000 (17:00 +0000)]
Revert "[Support][CachePruning] Fix regression in pruning interval"
This reverts commit r318397.
It broke tools/gold/X86/cache.ll.
llvm-svn: 318419
Sanjay Patel [Thu, 16 Nov 2017 16:59:49 +0000 (16:59 +0000)]
[InstCombine] regenerate test checks; NFC
Also, remove some unnecessary bits. I don't think we need fcmp in any test here either?
llvm-svn: 318418
Sanjay Patel [Thu, 16 Nov 2017 16:38:42 +0000 (16:38 +0000)]
[InstCombine] regenerate test checks; NFC
llvm-svn: 318417
Sanjay Patel [Thu, 16 Nov 2017 16:36:48 +0000 (16:36 +0000)]
[InstCombine] regenerate test checks; NFC
llvm-svn: 318416
Philip Pfaffe [Thu, 16 Nov 2017 16:35:19 +0000 (16:35 +0000)]
[SI] Fix a potential use-after-free
Summary:
There is a potential use-after-free bug in Scop::buildSchedule(Region *,
LoopStackTy &, LoopInfo &). Before, we took a reference to LoopStack.back()
which is a use after free, since back is popped off further below. This didn't
crash before by pure chance, since LoopStack is actually a vector, and the
memory isn't freed upon pop. I turned this into an iterator-based algorithm.
Reviewers: grosser, bollu, Meinersbur
Reviewed By: Meinersbur
Subscribers: llvm-commits, pollydev
Differential Revision: https://reviews.llvm.org/D39979
llvm-svn: 318415
Yaxun Liu [Thu, 16 Nov 2017 16:33:04 +0000 (16:33 +0000)]
Update tests for llvm.invariant.group.barrier becoming mangled
Differential Revision: https://reviews.llvm.org/D40062
llvm-svn: 318414
Yaxun Liu [Thu, 16 Nov 2017 16:32:16 +0000 (16:32 +0000)]
Let llvm.invariant.group.barrier accepts pointer to any address space
llvm.invariant.group.barrier may accept pointers to arbitrary address space.
This patch let it accept pointers to i8 in any address space and returns
pointer to i8 in the same address space.
Differential Revision: https://reviews.llvm.org/D39973
llvm-svn: 318413
Ilya Biryukov [Thu, 16 Nov 2017 16:25:18 +0000 (16:25 +0000)]
[clangd] Use in-memory preambles in clangd.
Reviewers: klimek, bkramer, sammccall
Reviewed By: sammccall
Subscribers: cfe-commits
Differential Revision: https://reviews.llvm.org/D39843
llvm-svn: 318412
Ilya Biryukov [Thu, 16 Nov 2017 16:25:01 +0000 (16:25 +0000)]
Allow to store precompiled preambles in memory.
Summary:
These preambles are built by ASTUnit and clangd. Previously, preambles
were always stored on disk.
In-memory preambles are routed back to the compiler as virtual files in
a custom VFS.
Interface of ASTUnit does not allow to use in-memory preambles, as
ASTUnit::CodeComplete receives FileManager as a parameter, so we can't
change VFS used by the compiler inside the CodeComplete method.
A follow-up commit will update clangd in clang-tools-extra to use
in-memory preambles.
Reviewers: klimek, sammccall, bkramer
Reviewed By: klimek
Subscribers: ioeric, cfe-commits
Differential Revision: https://reviews.llvm.org/D39842
llvm-svn: 318411
Kostya Kortchinsky [Thu, 16 Nov 2017 16:19:44 +0000 (16:19 +0000)]
[sanitizer] Use runtime checks instead of API level for Android logging
Summary:
Recent Bionic have a slew of `async_safe_*` logging functions that are
basically the liblog ones but included within the libc. They have the advantage
of not allocating memory. `async_safe_write_log` does no formatting and is
likely the best candidate for logging.
Use a weak definition to try and use it. Also, avoid API level checks (as
the toolchain is compiled at a rather low API level) for `__android_log_write`
in favor of a weak definition as well.
Keep the fallback to `syslog` if nothing else was found.
I tried to overhaul the code block to only have a single #if SANITIZER_ANDROID
but I am not particularly attached to the form. LMKWYT.
Reviewers: eugenis
Reviewed By: eugenis
Subscribers: srhines, kubamracek, llvm-commits
Differential Revision: https://reviews.llvm.org/D40100
llvm-svn: 318410
Igor Laevsky [Thu, 16 Nov 2017 16:17:32 +0000 (16:17 +0000)]
[FuzzMutate] Fix windows build after rL318407.
Add correct library dependence.
llvm-svn: 318409
Sanjay Patel [Thu, 16 Nov 2017 15:25:59 +0000 (15:25 +0000)]
[InstSimplify] add tests for fcmp ord/uno; NFC
llvm-svn: 318408
Igor Laevsky [Thu, 16 Nov 2017 15:23:08 +0000 (15:23 +0000)]
[FuzzMutate] NFC. Move parseModule and writeModule from llvm-isel-fuzzer into FuzzMutate.
This is to be able to reuse them in the llvm-opt-fuzzer.
llvm-svn: 318407
John Brawn [Thu, 16 Nov 2017 15:15:00 +0000 (15:15 +0000)]
Remove stray comma in sink-addrmode test
The extra comma meant it wasn't correctly checking that we weren't getting an
extra getelementptr.
llvm-svn: 318406
Sanjay Patel [Thu, 16 Nov 2017 14:40:51 +0000 (14:40 +0000)]
[InstCombine] include 'sub' in the list of narrow-able binops
// trunc (binop X, C) --> binop (trunc X, C')
// trunc (binop (ext X), Y) --> binop X, (trunc Y)
I'm grouping sub with the other binops because that makes the code simpler
and the transforms are valid:
https://rise4fun.com/Alive/UeF
...so even though we don't expect a sub with constant Op1 or any of the
other opcodes with constant Op0 due to canonicalization rules, we might as
well handle those situations if non-canonical code somehow reaches this
point (it should just make instcombine more efficient in reaching its
end goal).
This should solve the problem that later manifests in the vectorizers in
PR35295:
https://bugs.llvm.org/show_bug.cgi?id=35295
llvm-svn: 318404
Aaron Smith [Thu, 16 Nov 2017 14:33:09 +0000 (14:33 +0000)]
[DebugInfo/PDB] Adding getUndecoratedNameEx and IPDB interfaces for IDiaEnumTables and IDiaTable.
Initial changes to support debugging PE/COFF files with LLDB on Windows through DIA SDK.
There is another set of changes required on the LLDB side before this does anything.
Differential Revision: https://reviews.llvm.org/D39517
llvm-svn: 318403
Simon Pilgrim [Thu, 16 Nov 2017 14:13:48 +0000 (14:13 +0000)]
[X86] Add scheduling tests for SHLD/SHRD
llvm-svn: 318402
Javed Absar [Thu, 16 Nov 2017 13:49:27 +0000 (13:49 +0000)]
[SCEV] simplify loop. NFC.
Change loop to range-based
llvm-svn: 318401
Aaron Smith [Thu, 16 Nov 2017 13:42:28 +0000 (13:42 +0000)]
Test commit. Add a missing dash to the standard llvm file header; NFC.
llvm-svn: 318400
Pavel Labath [Thu, 16 Nov 2017 13:38:57 +0000 (13:38 +0000)]
Implement core dump debugging for PPC64le
Summary: Implement core dump debugging for PPC64le.
Reviewers: labath
Reviewed By: labath
Subscribers: JDevlieghere, krytarowski, clayborg, labath, lbianc, nemanjai, gut, anajuliapc, mgorny, kbarton, lldb-commits
Differential Revision: https://reviews.llvm.org/D39681
Patch by Alexandre Yukio Yamashita <alexandre.yamashita@eldorado.org.br>
llvm-svn: 318399
Diana Picus [Thu, 16 Nov 2017 13:32:47 +0000 (13:32 +0000)]
[ARM GlobalISel] Add tests for BIC. NFC
Add instruction selector tests for BICrr and BICri, which are handled by
TableGen.
llvm-svn: 318398
Ben Dunbobbin [Thu, 16 Nov 2017 13:15:56 +0000 (13:15 +0000)]
[Support][CachePruning] Fix regression in pruning interval
Fixed broken comparison.
borked by: rL284966 (see: https://reviews.llvm.org/D25730).
Differential Revision: https://reviews.llvm.org/D40119
llvm-svn: 318397
Peter Smith [Thu, 16 Nov 2017 13:00:17 +0000 (13:00 +0000)]
[builtins][ARM] re-enable mulsc3_test.c on Arm
The mulsc3_test.c was marked as unsupported due to PR32457, the underlying
cause of this PR was fixed in PR28164 so we can remove the unsupported as
it is no longer needed.
Differential Revision: https://reviews.llvm.org/D40076
llvm-svn: 318396
Walter Lee [Thu, 16 Nov 2017 12:57:19 +0000 (12:57 +0000)]
[asan] Fix size/alignment issues with non-default shadow scale
Fix a couple places where the minimum alignment/size should be a
function of the shadow granularity:
- alignment of AllGlobals
- the minimum left redzone size on the stack
Added a test to verify that the metadata_array is properly aligned
for shadow scale of 5, to be enabled when we add build support
for testing shadow scale of 5.
Differential Revision: https://reviews.llvm.org/D39470
llvm-svn: 318395
George Rimar [Thu, 16 Nov 2017 12:33:36 +0000 (12:33 +0000)]
[ELF] - Don't emit broken relocations for SHF_MERGE sections when --emit-relocs is used.
Previously our relocations we rewrote were broken for that case.
We emited incorrect addend and broken relocation info field
because did not produce section symbol for mergeable synthetic sections.
Differential revision: https://reviews.llvm.org/D40070
llvm-svn: 318394
Diana Picus [Thu, 16 Nov 2017 12:29:28 +0000 (12:29 +0000)]
[ARM GlobalISel] Add tests for REVSH patterns. NFC
Add instruction selector tests for some of the REVSH patterns handled by
TableGen.
llvm-svn: 318393
Yaxun Liu [Thu, 16 Nov 2017 12:22:19 +0000 (12:22 +0000)]
Fix pointer EVT in SelectionDAGBuilder::visitAlloca
SelectionDAGBuilder::visitAlloca assumes alloca address space is 0, which is
incorrect for triple amdgcn---amdgiz and causes isel failure.
This patch fixes that.
Differential Revision: https://reviews.llvm.org/D40095
llvm-svn: 318392
Martin Storsjo [Thu, 16 Nov 2017 12:06:42 +0000 (12:06 +0000)]
[COFF] Don't write long section names for sections that will be mapped at runtime
Sections that will be mapped at runtime will only have the short
section name available, since the string table it points into isn't
mapped. Therefore prefer truncating those names over writing a
long name that is unavailable at runtime.
This allows libunwind to find the .eh_frame section at runtime even
if the module was built with debug info enabled.
Differential Revision: https://reviews.llvm.org/D40025
llvm-svn: 318391
Sam Parker [Thu, 16 Nov 2017 11:28:26 +0000 (11:28 +0000)]
[DAGCombine] Enable more srl -> load combines
Change the calculation for the desired ValueType for non-sign
extending loads, as in those cases we don't care about the
higher bits. This creates a smaller ExtVT and allows for such
combinations as:
(srl (zextload i16, [addr]), 8) -> (zextload i8, [addr + 1])
Differential Revision: https://reviews.llvm.org/D40034
llvm-svn: 318390
Simon Dardis [Thu, 16 Nov 2017 10:50:44 +0000 (10:50 +0000)]
[mips][mt] Add missing test cases from r318207
llvm-svn: 318389
Andrey Churbanov [Thu, 16 Nov 2017 10:45:07 +0000 (10:45 +0000)]
Exclude untied tasks from checking of task scheduling constraint (TSC).
This can improve performance of tests with untied tasks.
Differential Revision: https://reviews.llvm.org/D39613
llvm-svn: 318388
Benjamin Kramer [Thu, 16 Nov 2017 10:18:07 +0000 (10:18 +0000)]
Assert correct removal of SUnit in LatencyPriorityQueue
The LatencyPriorityQueue doesn't currently check whether the SU being removed really exists in the Queue.
This method fails quietly when SU is not found and removes the last element from the Queue, leading to unexpected behavior.
Unfortunately, this only occurs on our custom target, with the custom scheduler. In our case, when remove() is invoked, it removes the wrong SU at the end of the Queue, which is only discovered later when VerifyScheduledDAG() is invoked and finds that some nodes were not scheduled at all.
As this is only reproducible with a lot of proprietary code, I'm hopeful this assert is straightforward enough to not necessitate a test.
Patch by Ondrej Glasnak!
Differential Revision: https://reviews.llvm.org/D40084
llvm-svn: 318387
Mohammed Agabaria [Thu, 16 Nov 2017 09:38:32 +0000 (09:38 +0000)]
[TTI][X86] update costs of interleaved load\store of i64\double
This patch contains more accurate cost of interelaved load\store of stride 2 for the types int64\double on AVX2.
Reviewers: delena, RKSimon, craig.topper, dorit
Reviewed By: dorit
Differential Revision: https://reviews.llvm.org/D40008
llvm-svn: 318385
Martin Storsjo [Thu, 16 Nov 2017 07:22:44 +0000 (07:22 +0000)]
[COFF] Improve the autoexport check for symbols from import libraries with -opt:noref
If -opt:noref is specified, they can end up with isLive() == 1
when the autoexport check is run.
To reduce the risk of potential issues, only consider exporting
DefinedRegular and DefinedCommon, nothing else.
Differential Revision: https://reviews.llvm.org/D40014
llvm-svn: 318384
Martin Storsjo [Thu, 16 Nov 2017 07:16:36 +0000 (07:16 +0000)]
[docs] Mention that dwarf unwinding should be supported on arm64/windows
This didn't require any further changes to libunwind as long as win64
in general is handled correctly.
Differential Revision: https://reviews.llvm.org/D39632
llvm-svn: 318383
Mikael Holmen [Thu, 16 Nov 2017 07:01:23 +0000 (07:01 +0000)]
[MachineRegisterInfo] Avoid having dbg.values affect code generation
Summary:
Use use_nodbg_empty() rather than use_empty() in
MachineRegisterInfo::EmitLiveInCopies() when determining if a livein
register has any uses or not. Otherwise a single dbg.value can make us
generate different code, meaning -g would affect code generation.
Found when compiling code for my out-of-tree target. Unfortunately I
haven't been able to reproduce the problem on X86 or any of the other
in-tree targets that I tried, so no test case.
Reviewers: MatzeB
Reviewed By: MatzeB
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D39044
llvm-svn: 318382
Max Kazantsev [Thu, 16 Nov 2017 06:06:27 +0000 (06:06 +0000)]
[IRCE] Fix SCEVExpander's usage in IRCE
When expanding exit conditions for pre- and postloops, we may end up expanding a
recurrency from the loop to in its loop's preheader. This produces incorrect IR.
This patch ensures that IRCE uses SCEVExpander correctly and only expands code which
is safe to expand in this particular location.
Differentian Revision: https://reviews.llvm.org/D39234
llvm-svn: 318381
Craig Topper [Thu, 16 Nov 2017 06:02:05 +0000 (06:02 +0000)]
[X86] Update TTI to report that v1iX/v1fX types aren't legal for masked gather/scatter/load/store.
The type legalizer will try to scalarize these operations if it sees them, but there is no handling for scalarizing them. This leads to a fatal error. With this change they will now be scalarized by the mem intrinsic scalarizing pass before SelectionDAG.
llvm-svn: 318380
Craig Topper [Thu, 16 Nov 2017 06:02:03 +0000 (06:02 +0000)]
[SelectionDAG] Use report_fatal_error instead of llvm_unreachable in some code that can be reached if targets don't configure things correctly.
For example, this is currently reachable by X86 if you use a masked store intrinsic with a v1iX type.
Using a fatal error seems like a better user experience if someone were to encounter this on a release build. There are several other similar places that have been converted from unreachable to fatal error previously.
llvm-svn: 318379
Marshall Clow [Thu, 16 Nov 2017 05:48:32 +0000 (05:48 +0000)]
More of P0600 - '[[nodiscard]] in the Library' mark empty() as nodiscard in filesystem::path
llvm-svn: 318378
Max Kazantsev [Thu, 16 Nov 2017 05:10:56 +0000 (05:10 +0000)]
[SCEV][NFC] Introduce isSafeToExpandAt function to SCEVExpander
This function checks that:
1) It is safe to expand a SCEV;
2) It is OK to materialize it at the specified location.
For example, attempt to expand a loop's AddRec to the same loop's preheader should fail.
Differential Revision: https://reviews.llvm.org/D39236
llvm-svn: 318377
Kostya Serebryany [Thu, 16 Nov 2017 04:52:05 +0000 (04:52 +0000)]
[libFuzzer] disable test/fuzzer/merge-sigusr.test on Mac while investigating the bot failure
llvm-svn: 318376
Marshall Clow [Thu, 16 Nov 2017 04:48:34 +0000 (04:48 +0000)]
More of P0600 - '[[nodiscard]] in the Library' mark empty() as nodiscard in match_results. <regex>
llvm-svn: 318375
Eric Christopher [Thu, 16 Nov 2017 03:25:02 +0000 (03:25 +0000)]
Fix thinko in last commit.
llvm-svn: 318374
Eric Christopher [Thu, 16 Nov 2017 03:18:15 +0000 (03:18 +0000)]
Add NDEBUG checks around LLVM_DUMP_METHOD functions for Wunused-function warnings.
llvm-svn: 318373
Eric Christopher [Thu, 16 Nov 2017 03:18:13 +0000 (03:18 +0000)]
Need to work around the gcc Wunused-function bug as far back as gcc 6.1, update accordingly.
llvm-svn: 318372
Eric Christopher [Thu, 16 Nov 2017 03:18:09 +0000 (03:18 +0000)]
Add NDEBUG checks around LLVM_DUMP_METHOD functions for Wunused-function warnings.
llvm-svn: 318371
Yaxun Liu [Thu, 16 Nov 2017 02:54:49 +0000 (02:54 +0000)]
Fix APInt bit size in processDbgDeclares
processDbgDeclares assumes pointer size is the same for different addr spaces.
It uses pointer size for addr space 0 for all pointers, which causes assertion
in stripAndAccumulateInBoundsConstantOffsets for amdgcn---amdgiz since
pointer in addr space 5 has different size than in addr space 0.
This patch fixes that.
Differential Revision: https://reviews.llvm.org/D40085
llvm-svn: 318370
Evgeniy Stepanov [Thu, 16 Nov 2017 02:52:19 +0000 (02:52 +0000)]
[asan] Fallback to non-ifunc dynamic shadow on android<22.
Summary: Android < 22 does not support ifunc.
Reviewers: pcc
Subscribers: srhines, kubamracek, hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D40116
llvm-svn: 318369
Craig Topper [Thu, 16 Nov 2017 02:07:45 +0000 (02:07 +0000)]
[X86] Custom type legalize v2f32 masked gathers instead of trying to cleanup after type legalization.
llvm-svn: 318368
Alex Lorenz [Thu, 16 Nov 2017 01:31:27 +0000 (01:31 +0000)]
[DeclPrinter] Extract function PrintConstructorInitializers, NFC
Patch by Nikolai Kosjar!
Differential Revision: https://reviews.llvm.org/D40066
llvm-svn: 318367
Yan Zhang [Thu, 16 Nov 2017 01:28:29 +0000 (01:28 +0000)]
add check to avoid throwing objc exception according to Google Objective-C guide
Summary:
This is a small check to avoid throwing objc exceptions.
In specific it will detect the usage of @throw statement and throw warning.
Reviewers: hokein, benhamilton
Reviewed By: hokein, benhamilton
Subscribers: cfe-commits, mgorny
Differential Revision: https://reviews.llvm.org/D40058
llvm-svn: 318366
Alex Lorenz [Thu, 16 Nov 2017 01:28:25 +0000 (01:28 +0000)]
[DeclPrinter] Honor TerseOutput for constructors
Patch by Nikolai Kosjar!
Differential Revision: https://reviews.llvm.org/D39957
llvm-svn: 318365
Bob Haarman [Thu, 16 Nov 2017 01:22:01 +0000 (01:22 +0000)]
[coff] correctly emit safeseh entries for handlers defined in dlls
Summary:
We previously assumed that all SafeSEH handlers are
DefinedRegular symbols. This is not the case for handlers defined in
DLLs. As a result, we were failing to emit entries in the SafeSEH
table for those handlers. This change fixes that.
Fixes PR35324.
Reviewers: rnk, ruiu
Reviewed By: rnk
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D40102
llvm-svn: 318364
Bob Haarman [Thu, 16 Nov 2017 01:16:52 +0000 (01:16 +0000)]
LTO: clarify why we need to gracefully handle sys::fs::rename failures
Reviewers: pcc, rafael
Reviewed By: pcc
Subscribers: mehdi_amini, llvm-commits, hiraditya
Differential Revision: https://reviews.llvm.org/D40094
llvm-svn: 318362
Rafael Espindola [Thu, 16 Nov 2017 01:06:36 +0000 (01:06 +0000)]
Convert a use of createUniqueFile to TempFile::create.
llvm-svn: 318361
Yonghong Song [Thu, 16 Nov 2017 00:52:30 +0000 (00:52 +0000)]
bpf: enable llvm-objdump to print out symbolized jmp target
Add hook in BPF backend so that llvm-objdump can print out
the jmp target with label names, e.g.,
...
if r1 != 2 goto 6 <LBB0_2>
...
goto 7 <LBB0_4>
...
LBB0_2:
...
LBB0_4:
...
Signed-off-by: Yonghong Song <yhs@fb.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
llvm-svn: 318358
Jason Molenda [Thu, 16 Nov 2017 00:50:29 +0000 (00:50 +0000)]
Fix alignment of arm64 fpu register context structure
so it has the same padding as the kernel's definition
which is written in terms of uint128_t. Original patch
by Ryan Mansfield.
<rdar://problem/
35468499>
llvm-svn: 318357
Daniel Sanders [Thu, 16 Nov 2017 00:46:35 +0000 (00:46 +0000)]
[globalisel][tablegen] Generate rule coverage and use it to identify untested rules
Summary:
This patch adds a LLVM_ENABLE_GISEL_COV which, like LLVM_ENABLE_DAGISEL_COV,
causes TableGen to instrument the generated table to collect rule coverage
information. However, LLVM_ENABLE_GISEL_COV goes a bit further than
LLVM_ENABLE_DAGISEL_COV. The information is written to files
(${CMAKE_BINARY_DIR}/gisel-coverage-* by default). These files can then be
concatenated into ${LLVM_GISEL_COV_PREFIX}-all after which TableGen will
read this information and use it to emit warnings about untested rules.
This technique could also be used by SelectionDAG and can be further
extended to detect hot rules and give them priority over colder rules.
Usage:
* Enable LLVM_ENABLE_GISEL_COV in CMake
* Build the compiler and run some tests
* cat gisel-coverage-[0-9]* > gisel-coverage-all
* Delete lib/Target/*/*GenGlobalISel.inc*
* Build the compiler
Known issues:
* ${LLVM_GISEL_COV_PREFIX}-all must be generated as a manual
step due to a lack of a portable 'cat' command. It should be the
concatenation of all ${LLVM_GISEL_COV_PREFIX}-[0-9]* files.
* There's no mechanism to discard coverage information when the ruleset
changes
Depends on D39742
Reviewers: ab, qcolombet, t.p.northover, aditya_nandakumar, rovka
Reviewed By: rovka
Subscribers: vsk, arsenm, nhaehnle, mgorny, kristof.beyls, javed.absar, igorb, llvm-commits
Differential Revision: https://reviews.llvm.org/D39747
llvm-svn: 318356
Reid Kleckner [Thu, 16 Nov 2017 00:32:19 +0000 (00:32 +0000)]
Try to fix WebAssembly build after r318352
llvm-svn: 318355
Craig Topper [Thu, 16 Nov 2017 00:19:59 +0000 (00:19 +0000)]
[GVNHoist] Fix a signed/unsigned comparison warning that occurs in 32-bit builds with gcc.
std::distance returns ptrdiff_t which is signed. 64-bit builds don't notice because type promotion widens the unsigned first.
llvm-svn: 318354
Rong Xu [Thu, 16 Nov 2017 00:14:05 +0000 (00:14 +0000)]
[CodeGen] Fix the branch probability assertion in r318202
Due to integer precision, we might have numerator greater than denominator in
the branch probability scaling. Add a check to prevent this from happening.
llvm-svn: 318353
Daniel Sanders [Wed, 15 Nov 2017 23:55:44 +0000 (23:55 +0000)]
Add backend name to Target to enable runtime info to be fed back into TableGen
Summary:
Make it possible to feed runtime information back to tablegen to enable
profile-guided tablegen-eration, detection of untested tablegen definitions, etc.
Being a cross-compiler by nature, LLVM will potentially collect data for multiple
architectures (e.g. when running 'ninja check'). We therefore need a way for
TableGen to figure out what data applies to the backend it is generating at the
time. This patch achieves that by including the name of the 'def X : Target ...'
for the backend in the TargetRegistry.
Reviewers: qcolombet
Reviewed By: qcolombet
Subscribers: jholewinski, arsenm, jyknight, aditya_nandakumar, sdardis, nemanjai, ab, nhaehnle, t.p.northover, javed.absar, qcolombet, llvm-commits, fedor.sergeev
Differential Revision: https://reviews.llvm.org/D39742
llvm-svn: 318352
Evandro Menezes [Wed, 15 Nov 2017 23:49:58 +0000 (23:49 +0000)]
[AArch64] Adjust the cost model for Exynos M1 and M2
Fix the modeling of FP stores.
llvm-svn: 318351
Aditya Nandakumar [Wed, 15 Nov 2017 23:45:04 +0000 (23:45 +0000)]
[GISel][NFC]: Move getOpcodeDef from the LegalizationArtifactCombiner into GlobalISel/Utils for use elsewhere
llvm-svn: 318350
Alex Shlyapnikov [Wed, 15 Nov 2017 23:42:48 +0000 (23:42 +0000)]
[LSan] Export __lsan_init
Summary:
lsan_preinit.cc is meant to be linked into executable and calls
lsan_init from .preinit_array section. But if liblsan is a shared library,
then this doesn't work, because the symbol is not exported. This patch fixes
that. The counterparts like asan_init or __tsan_init already do have
SANITIZER_INTERFACE_ATTRIBUTE.
Committing on behalf of jakubjelinek.
Reviewers: alekseyshl, kcc
Subscribers: llvm-commits
Differential revision: https://reviews.llvm.org/D39889
llvm-svn: 318349
Davide Italiano [Wed, 15 Nov 2017 23:39:41 +0000 (23:39 +0000)]
[POSIX] Replace assert with llvm_unreachable(). NFCI.
llvm-svn: 318348
NAKAMURA Takumi [Wed, 15 Nov 2017 23:04:44 +0000 (23:04 +0000)]
clang/module.modulemap: clang/Basic/X86Target.def may be textual header.
llvm-svn: 318347
Erich Keane [Wed, 15 Nov 2017 22:36:24 +0000 (22:36 +0000)]
Add X86Target.def that was forgotten in r30734
llvm-svn: 318345
Erich Keane [Wed, 15 Nov 2017 22:25:39 +0000 (22:25 +0000)]
Split x86 "Processor" info into its own def file. [NFC]
A first step toward removing the repetition of
features/CPU info in the x86 target info, this
patch pulls all the processor information out into
its own .def file.
Differential Revision: https://reviews.llvm.org/D40093
llvm-svn: 318343
Sanjay Patel [Wed, 15 Nov 2017 22:19:55 +0000 (22:19 +0000)]
[InstCombine] add sub narrowing tests; NFC
This might be the root cause of PR35295:
https://bugs.llvm.org/show_bug.cgi?id=35295
llvm-svn: 318342
Eugene Zelenko [Wed, 15 Nov 2017 22:00:04 +0000 (22:00 +0000)]
[AST, Sema] Fix some Clang-tidy modernize and Include What You Use warnings; other minor fixes (NFC).
llvm-svn: 318341
Matt Arsenault [Wed, 15 Nov 2017 21:51:43 +0000 (21:51 +0000)]
AMDGPU: Replace i64 add/sub lowering
Use VOP3 add/addc like usual.
This has some tradeoffs. Inline immediates fold
a little better, but other constants are worse off.
SIShrinkInstructions could be made smarter to handle
these cases.
This allows us to avoid selecting scalar adds where we
need to track the carry in scc and replace its users.
This makes it easier to use the carryless VALU adds.
llvm-svn: 318340
Sam McCall [Wed, 15 Nov 2017 21:50:53 +0000 (21:50 +0000)]
[clangd] Loosen more brittle tests
llvm-svn: 318339
Dan Gohman [Wed, 15 Nov 2017 21:38:33 +0000 (21:38 +0000)]
[WebAssembly] Update cfg-stackify.ll to remove the workaround added in r318288.
Remove -switch-peel-threshold=100 and update the expected results in test10
in cfg-stackify.ll.
llvm-svn: 318338
Sam McCall [Wed, 15 Nov 2017 21:33:56 +0000 (21:33 +0000)]
[clangd] Revert broken r318329 and disable tests on PS4
(Clearly this is not a permanent solution)
llvm-svn: 318337
Francis Ricci [Wed, 15 Nov 2017 21:19:20 +0000 (21:19 +0000)]
Fix some typos in a comment.
llvm-svn: 318336
Evandro Menezes [Wed, 15 Nov 2017 21:06:22 +0000 (21:06 +0000)]
[AArch64] Refactor the loads and stores optimizer
Move remaining inline matching of instructions of some optimizations into
separate functions, like in the other optimizations. Otherwise, NFC.
Differential revision: https://reviews.llvm.org/D40090
llvm-svn: 318335
Rafael Espindola [Wed, 15 Nov 2017 20:55:53 +0000 (20:55 +0000)]
Simplify file handling in dsymutil.
This moves the file handling out of DwarfLinker.cpp.
This fixes what is at least an oddity if not a bug. DwarfLinker.cpp
was using ToolOutputFile, which uses RemoveFileOnSignal. The issue is
that dsymutil.cpp uses that too. It is now clear from the interface
that only dsymutil.cpp is responsible for creating and deleting files.
llvm-svn: 318334
Craig Topper [Wed, 15 Nov 2017 20:42:49 +0000 (20:42 +0000)]
[X86] Add some explanatory comments to the ProcessorFeatures enum in Host.cpp.
llvm-svn: 318331
Craig Topper [Wed, 15 Nov 2017 20:42:47 +0000 (20:42 +0000)]
[X86] Add a return to the end of a switch to prevent an accidental fallthrough in the future.
llvm-svn: 318330
Sam McCall [Wed, 15 Nov 2017 20:10:14 +0000 (20:10 +0000)]
[clangd] Fix flag name from r318327
llvm-svn: 318329
Marshall Clow [Wed, 15 Nov 2017 20:02:27 +0000 (20:02 +0000)]
More of P0600 - '[[nodiscard]] in the Library' mark empty() as nodiscard in string, string_view, and the free function std::empty(). Removed tabs from <string_view>, which is why the diff is so big.
llvm-svn: 318328
Sam McCall [Wed, 15 Nov 2017 19:38:09 +0000 (19:38 +0000)]
[clangd] Try to unbreak tests on PS4 by targeting PC explicitly
llvm-svn: 318327
Casey Carter [Wed, 15 Nov 2017 19:14:45 +0000 (19:14 +0000)]
[test] Alignment must be > __STDCPP_DEFAULT_NEW_ALIGNMENT__ to call aligned new
Differential Revision: D39221
llvm-svn: 318325
Jake Ehrlich [Wed, 15 Nov 2017 19:13:31 +0000 (19:13 +0000)]
[llvm-objcopy] Change -O binary to respect section removal and behave like GNU objcopy
The original -O binary implementation just copied segment data from the
object and dumped it into a file. This doesn't take into account any
operations performed on objects such as section removal. GNU objcopy has
some specific behavior that we'd also like to respect. For instance
using -O binary and -j <some_section> will dump <some_section> to a
file. This change implements GNU objcopy style -O binary to as close of
an approximation as I can determine.
Differential Revision: https://reviews.llvm.org/D39713
llvm-svn: 318324
Sanjay Patel [Wed, 15 Nov 2017 19:12:01 +0000 (19:12 +0000)]
[InstCombine] trunc (binop X, C) --> binop (trunc X, C')
Note that one-use and shouldChangeType() are checked ahead of the switch.
Without the narrowing folds, we can produce inferior vector code as shown in PR35299:
https://bugs.llvm.org/show_bug.cgi?id=35299
llvm-svn: 318323
Rafael Espindola [Wed, 15 Nov 2017 19:09:22 +0000 (19:09 +0000)]
Use TempFile in lto caching.
This requires a small change to TempFile: allowing a discard after a
failed keep.
With this the cache now handles signals and reuses a fd instead of
reopening the file.
llvm-svn: 318322
Sean Fertile [Wed, 15 Nov 2017 18:58:27 +0000 (18:58 +0000)]
[PowerPC] Implement mayBeEmittedAsTailCall for PPC
Implements TargetLowering callback 'mayBeEmittedAsTailCall' that enables
CodeGenPrepare to duplicate returns when they might enable a tail-call.
Differential Revision: https://reviews.llvm.org/D39777
llvm-svn: 318321
Reid Kleckner [Wed, 15 Nov 2017 18:51:12 +0000 (18:51 +0000)]
[InstCombine] Salvage debug info during initial DCE
InstCombine salvages debug info for every instruction it erases from its
worklist, but it wasn't doing it during its initial DCE when populating
its worklist. This fixes that.
This should help improve availability of 'this' in optimized debug info
when casts are necessary.
llvm-svn: 318320
Sanjay Patel [Wed, 15 Nov 2017 18:09:43 +0000 (18:09 +0000)]
[InstCombine] add tests for missing trunc folds; NFC
As noted in PR35299:
https://bugs.llvm.org/show_bug.cgi?id=35299
...this is likely the root cause for a mis-vectorization transform.
llvm-svn: 318319
Vedant Kumar [Wed, 15 Nov 2017 18:05:19 +0000 (18:05 +0000)]
[docs] Mention opt -metarenamer in the bugpoint docs
Thanks to arsenm and davide for the suggestion!
llvm-svn: 318318
Ilya Biryukov [Wed, 15 Nov 2017 18:04:56 +0000 (18:04 +0000)]
[clangd] clang-format the source code. NFC.
llvm-svn: 318317
Sam McCall [Wed, 15 Nov 2017 17:53:46 +0000 (17:53 +0000)]
[clangd] Fix time units in clangd performance trace
llvm-svn: 318316
Hans Wennborg [Wed, 15 Nov 2017 17:47:58 +0000 (17:47 +0000)]
Try to fix test/SemaCXX/deleted-operator.cpp after r318309
The number of 'built-in candidate' notes now varies since __float128 may
or may not be a candidate depending on the target.
llvm-svn: 318314
Marshall Clow [Wed, 15 Nov 2017 17:47:09 +0000 (17:47 +0000)]
Clean up the tests for free data(), size() and empty()
llvm-svn: 318313
Evandro Menezes [Wed, 15 Nov 2017 17:39:37 +0000 (17:39 +0000)]
[AArch64] Adjust the cost model for Exynos M1 and M2
Fix the modeling of loads and stores using the pre or post indexed
addressing modes.
llvm-svn: 318312