Tatyana Krasnukha [Thu, 4 Oct 2018 11:39:55 +0000 (11:39 +0000)]
Re-commit r343500 "Fix build with GCC < 5.0 (PR39131)"
Occasionally didn't commit actual fix the first time.
llvm-svn: 343769
Martin Bohme [Thu, 4 Oct 2018 11:36:39 +0000 (11:36 +0000)]
[clang-tidy] Sequence statements with multiple parents correctly (PR39149)
Summary:
Before this fix, the bugprone-use-after-move check could incorrectly
conclude that a use and move in a function template were not sequenced.
For details, see
https://bugs.llvm.org/show_bug.cgi?id=39149
Reviewers: alexfh, hokein, aaron.ballman, JonasToth
Reviewed By: aaron.ballman
Subscribers: xazax.hun, cfe-commits
Tags: #clang-tools-extra
Differential Revision: https://reviews.llvm.org/D52782
llvm-svn: 343768
Haojian Wu [Thu, 4 Oct 2018 11:03:55 +0000 (11:03 +0000)]
[Index] Respect "IndexFunctionLocals" option for type loc.
Summary:
Previously, clang index ignored local symbols defined in the function body even
IndexFunctionLocals is true.
Reviewers: sammccall
Reviewed By: sammccall
Subscribers: ilya-biryukov, ioeric, arphaman, kadircet, cfe-commits
Differential Revision: https://reviews.llvm.org/D52877
llvm-svn: 343767
Andrea Di Biagio [Thu, 4 Oct 2018 10:36:49 +0000 (10:36 +0000)]
[llvm-mca] Check for inconsistencies when constructing instruction descriptors.
This should help with catching inconsistent definitions of instructions with
zero opcodes, which also declare to consume scheduler/pipeline resources.
llvm-svn: 343766
Simon Pilgrim [Thu, 4 Oct 2018 10:25:52 +0000 (10:25 +0000)]
Fix MSVC "not all control paths return a value" warning. NFCI.
llvm-svn: 343765
Sam McCall [Thu, 4 Oct 2018 10:10:35 +0000 (10:10 +0000)]
[clangd] Revert accidental flag change
llvm-svn: 343764
Haojian Wu [Thu, 4 Oct 2018 09:56:08 +0000 (09:56 +0000)]
[clangd] Use canonical declarations in ReferenceFinder.
Summary:
handleDeclOccurrencce reports a canonical declartion, so stick to use
canonical declarations to determine whether a declaration is in the
target set.
Also fix a previous ref test which misses a matched label (it fails without this
patch).
Reviewers: sammccall
Subscribers: ilya-biryukov, ioeric, MaskRay, jkorous, arphaman, kadircet, cfe-commits
Differential Revision: https://reviews.llvm.org/D52871
llvm-svn: 343763
George Rimar [Thu, 4 Oct 2018 09:31:15 +0000 (09:31 +0000)]
[ELF] - Simplify. NFCI.
Assign the `Link` to parent directly.
llvm-svn: 343762
Petr Pavlu [Thu, 4 Oct 2018 09:25:44 +0000 (09:25 +0000)]
[constexpr] Fix ICE when memcpy() is given a pointer to an incomplete array
Fix code for constant evaluation of __builtin_memcpy() and
__builtin_memmove() that would attempt to divide by zero when given two
pointers to an incomplete array.
Differential Revision: https://reviews.llvm.org/D51855
llvm-svn: 343761
Sam McCall [Thu, 4 Oct 2018 09:16:12 +0000 (09:16 +0000)]
[clangd] Support refs() in dex. Largely cloned from MemIndex.
Reviewers: hokein
Subscribers: ilya-biryukov, ioeric, MaskRay, jkorous, arphaman, kadircet, cfe-commits
Differential Revision: https://reviews.llvm.org/D52726
llvm-svn: 343760
Sam McCall [Thu, 4 Oct 2018 08:30:03 +0000 (08:30 +0000)]
[clangd] clangd-indexer: Drop support for MR-via-YAML
Summary:
It's slow, and the open-source reduce implementation doesn't scale properly.
While here, tidy up some dead headers and comments.
Reviewers: kadircet
Subscribers: ilya-biryukov, ioeric, MaskRay, jkorous, arphaman, cfe-commits
Differential Revision: https://reviews.llvm.org/D52517
llvm-svn: 343759
Sjoerd Meijer [Thu, 4 Oct 2018 07:38:53 +0000 (07:38 +0000)]
[AArch64][ARM] Context sensitive meaning of crypto
For AArch64, crypto means:
- sm4 + sha3 + sha2 + aes for Armv8.4-A and up, and
- sha2 + aes for Armv8.3-A and earlier.
For AArch32:
Crypto means sha2 + aes, because the Armv8.2-A crypto instructions
were added to AArch64 only.
Differential Revision: https://reviews.llvm.org/D50179
llvm-svn: 343758
Alex Bradbury [Thu, 4 Oct 2018 07:35:52 +0000 (07:35 +0000)]
[RISCV][NFC] Remove dead CHECK lines from vararg.ll test
The RISCV32 check prefix is no longer used so these lines are dead.
llvm-svn: 343757
Alex Bradbury [Thu, 4 Oct 2018 07:28:49 +0000 (07:28 +0000)]
[RISCV] Bugfix for floats passed on the stack with the ILP32 ABI on RV32F
f32 values passed on the stack would previously cause an assertion in
unpackFromMemLoc.. This would only trigger in the presence of the F extension
making f32 a legal type. Otherwise the f32 would be legalized.
This patch fixes that by keeping LocVT=f32 when a float is passed on the
stack. It also adds test coverage for this case, and tests that also
demonstrate lw/sw/flw/fsw will be selected when most profitable. i.e. there is
no unnecessary i32<->f32 conversion in registers.
llvm-svn: 343756
Clement Courbet [Thu, 4 Oct 2018 07:07:16 +0000 (07:07 +0000)]
[llvm-exegesis][NFC] Test sched class names only in !NDEBUG mode.
Sched classes have no names in NDEBUG.
llvm-svn: 343755
Craig Topper [Thu, 4 Oct 2018 06:13:27 +0000 (06:13 +0000)]
[X86] Merge matchANDXORWithAllOnesAsANDNP into combineANDXORWithAllOnesIntoANDNP. NFCI
It's the only caller and the logic pretty easy to combine.
llvm-svn: 343754
Petr Hosek [Thu, 4 Oct 2018 05:38:53 +0000 (05:38 +0000)]
[CMake] Use just basename when copying C++ ABI headers
This avoids duplicate directories when the filename includes path.
Fixes PR39145
Differential Revision: https://reviews.llvm.org/D52762
llvm-svn: 343753
Alex Bradbury [Thu, 4 Oct 2018 05:27:50 +0000 (05:27 +0000)]
[RISCV][NFC] Fix naming of RISCVISelLowering::{LowerRETURNADDR,LowerFRAMEADDR}
Rename to lowerRETURNADDR, lowerFRAMEADDR in order to be consistent with the
LLVM coding style and the other functions in this file.
llvm-svn: 343752
Fangrui Song [Thu, 4 Oct 2018 03:32:47 +0000 (03:32 +0000)]
[llvm-exegesis] Unbreak analysis-uops-variant.test introduced in D52825
A `defined(NDEBUG) && !defined(LLVM_ENABLE_DUMP)` build does not call
writeEscaped and there will be no `SBWriteZeroLatency` in the output.
llvm-svn: 343751
Craig Topper [Thu, 4 Oct 2018 02:40:35 +0000 (02:40 +0000)]
[LegalizeIntegerTypes] Fix typo in comment. NFC
llvm-svn: 343750
Matt Morehouse [Thu, 4 Oct 2018 00:35:24 +0000 (00:35 +0000)]
Use std::make_pair rather than brace initialization.
r343732 broke the Windows bot. Seems like the compiler on that bot
doesn't like brace initialization.
llvm-svn: 343749
Shoaib Meenai [Thu, 4 Oct 2018 00:07:59 +0000 (00:07 +0000)]
[ELF] Fix typo. NFC
llvm-svn: 343748
George Karpenkov [Thu, 4 Oct 2018 00:07:45 +0000 (00:07 +0000)]
[analyzer] [quickfix] Temporarily disabling a failing test.
llvm-svn: 343747
Derek Schuff [Wed, 3 Oct 2018 23:56:52 +0000 (23:56 +0000)]
[WebAssembly] Add WebAssembly to LLVM_ALL_TARGETS
Summary:
After fixing memory leaks in rL343362 and rL343733 the sanitizer builds are
clean and we should be good to build by default again.
Differential Revision: https://reviews.llvm.org/D52850
llvm-svn: 343746
Shoaib Meenai [Wed, 3 Oct 2018 23:53:11 +0000 (23:53 +0000)]
[ELF] Fix crash on invalid undefined local symbols
r320770 made LLD handle invalid DSOs where local symbols were found in
the global part of the symbol table. Unfortunately, it didn't handle the
case where those local symbols were also undefined, and r326242 exposed
an assertion failure in that case. Just warn on that case instead of
crashing, by moving the local binding check before the undefined symbol
addition.
The input file for the test is crafted by hand, since I don't know of
any tool that would produce such a broken DSO. I also don't understand
what it even means for a symbol to be undefined but have STB_LOCAL
binding - I don't think that combination makes any sense - but we have
found broken DSOs of this nature that we were linking against. I've
included detailed instructions on how to produce the DSO in the test.
Differential Revision: https://reviews.llvm.org/D52815
llvm-svn: 343745
Jordan Rupprecht [Wed, 3 Oct 2018 23:40:04 +0000 (23:40 +0000)]
[llvm-nm] Update clang tests based on changed llvm-nm output
llvm-svn: 343744
Jordan Rupprecht [Wed, 3 Oct 2018 23:39:59 +0000 (23:39 +0000)]
[llvm-nm] Update lld tests based on changed llvm-nm output
llvm-svn: 343743
Jordan Rupprecht [Wed, 3 Oct 2018 23:39:49 +0000 (23:39 +0000)]
[llvm-nm] Print an explicit "no symbols" message when an object file has no symbols
Summary:
GNU nm (and other nm implementations, such as "go tool nm") prints an explicit "no symbols" message when an object file has no symbols. Currently llvm-nm just doesn't print anything. Adding an explicit "no symbols" message will allow llvm-nm to be used in place of nm: some scripts and build processes use `nm <file> | grep "no symbols"` as a test to see if a file has no symbols. It will also be more familiar to anyone used to nm.
That said, the format implemented here is slightly different, in that it doesn't print the tool name in the message (which IMHO is not useful to include).
Demo:
```
$ for nm in nm bin/llvm-nm ; do echo "nm implementation: $nm"; $nm /tmp/foo{1,2}.o; echo; done
nm implementation: nm
/tmp/foo1.o:
nm: /tmp/foo1.o: no symbols
/tmp/foo2.o:
0000000000000000 T foo2
nm implementation: bin/llvm-nm
/tmp/foo1.o:
no symbols
/tmp/foo2.o:
0000000000000000 T foo2
```
Reviewers: MaskRay
Reviewed By: MaskRay
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D52810
llvm-svn: 343742
Alex Bradbury [Wed, 3 Oct 2018 23:30:16 +0000 (23:30 +0000)]
[RISCV] Handle redundant SplitF64+BuildPairF64 pairs in a DAGCombine
r343712 performed this optimisation during instruction selection. As Eli
Friedman pointed out in post-commit review, implementing this as a DAGCombine
might allow opportunities for further optimisations.
llvm-svn: 343741
Nick Desaulniers [Wed, 3 Oct 2018 23:09:29 +0000 (23:09 +0000)]
[SEMA] split ExtWarn dupl-decl-spec's into Extension and ExtWarn
Summary:
For types deduced from typedef's and typeof's, don't warn for duplicate
declaration specifiers in C90 unless -pedantic.
Create a third diagnostic type for duplicate declaration specifiers.
Previously, we had an ExtWarn and a Warning. This change adds a third,
Extension, which only warns when -pedantic is set, staying silent
otherwise.
Fixes PR32985.
Reviewers: rsmith
Reviewed By: rsmith
Subscribers: srhines, cfe-commits
Differential Revision: https://reviews.llvm.org/D52849
llvm-svn: 343740
Thomas Lively [Wed, 3 Oct 2018 23:02:23 +0000 (23:02 +0000)]
[WebAssembly] Bitselect intrinsic and instruction
Summary: Depends on D52755.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, sunfish, llvm-commits
Differential Revision: https://reviews.llvm.org/D52805
llvm-svn: 343739
Fangrui Song [Wed, 3 Oct 2018 22:56:26 +0000 (22:56 +0000)]
Revert "[ASTMatchers] Fix DynamicASTMatchersTests again"
This reverts commit
8a6631a983ec9c1d22cc77c5f55a524a651740f0.
The last fix seems good in Debug mode.
llvm-svn: 343738
Alex Bradbury [Wed, 3 Oct 2018 22:53:25 +0000 (22:53 +0000)]
[RISCV][NFC] Refactor LocVT<->ValVT converstion in RISCVISelLowering
There was some duplicated logic for using the LocInfo of a CCValAssign in
order to convert from the ValVT to LocVT or vice versa. Resolve this by
factoring out convertLocVTFromValVT from unpackFromRegLoc. Also rename
packIntoRegLoc to the more appropriate convertValVTToLocVT and call these
helper functions consistently.
llvm-svn: 343737
Artem Dergachev [Wed, 3 Oct 2018 22:48:00 +0000 (22:48 +0000)]
[analyzer] [tests] [quickfix] Make more test more resilient for non-defaut -std.
It is important to specify the version of the standard because tests should
test the same thing regardless of the current default version of the standard.
llvm-svn: 343736
George Karpenkov [Wed, 3 Oct 2018 22:31:09 +0000 (22:31 +0000)]
[analyzer] Do not crash if the assumption added in TrustNonNullChecker is enough to make the state unfeasible
rdar://
43541814
Differential Revision: https://reviews.llvm.org/D52848
llvm-svn: 343735
Derek Schuff [Wed, 3 Oct 2018 22:25:32 +0000 (22:25 +0000)]
[WebAssembly] Refactor use of signatures
Update use of WebAssemblySignature to go along with D52580
Differential Revision: https://reviews.llvm.org/D52622
llvm-svn: 343734
Derek Schuff [Wed, 3 Oct 2018 22:22:48 +0000 (22:22 +0000)]
[WebAssembly] Refactor WasmSignature and use it for MCSymbolWasm
MCContext does not destroy MCSymbols on shutdown. So, rather than putting
SmallVectors (which may heap-allocate) inside MCSymbolWasm, use unowned pointer
to a WasmSignature instead. The signatures are now owned by the AsmPrinter.
Also uses WasmSignature instead of param and result vectors in TargetStreamer,
and leaves some TODOs for further simplification.
Differential Revision: https://reviews.llvm.org/D52580
llvm-svn: 343733
Rui Ueyama [Wed, 3 Oct 2018 22:20:26 +0000 (22:20 +0000)]
Minor refacotring of Relocations.cpp. NFC.
This patch splits ThunkCreator::mergeThunks into two smaller functions.
Also adds blank lines to various places so that the code doesn't look
too dense.
llvm-svn: 343732
Daniel Sanders [Wed, 3 Oct 2018 22:05:31 +0000 (22:05 +0000)]
[machineverifier] Detect PHI's that are preceeded by non-PHI's
If present, PHI nodes must appear before non-PHI nodes in a basic block. The
register allocator relies on this and will fail to eliminate PHI's that do not
meet this requirement.
llvm-svn: 343731
Simon Atanasyan [Wed, 3 Oct 2018 22:02:23 +0000 (22:02 +0000)]
[mips] Remove -allow-deprecated-dag-overlap flag from tests. NFC
Fix DAG check statements in MIPS codegen tests to remove
-allow-deprecated-dag-overlap flag.
llvm-svn: 343730
Sean Fertile [Wed, 3 Oct 2018 21:58:44 +0000 (21:58 +0000)]
[PPC64] Test documenting toc-restore after interposable recursive call. [NFC]
A test verifying that toc restores are properly inserted following recursive
calls, as well as briefly describing why they are needed.
Differential Revision: https://reviews.llvm.org/D52564
llvm-svn: 343729
Sean Fertile [Wed, 3 Oct 2018 21:58:42 +0000 (21:58 +0000)]
[PPC64] Add test documenting toc-restore when linking with -Bsymbolic. [NFC]
Differential Revision: https://reviews.llvm.org/D52523
llvm-svn: 343728
Sanjay Patel [Wed, 3 Oct 2018 21:44:59 +0000 (21:44 +0000)]
[InstCombine] allow SimplifyDemandedVectorElts to work with FP binops
We're a long way from D50992 and D51553, but this is where we have to start.
We weren't back-propagating undefs into binop constant values for anything but
add/sub/mul/and/or/xor.
This is likely because we have to be careful about not introducing UB/poison
with div/rem/shift. But I suspect we already are getting the poison part wrong
for add/sub/mul (although it may not be possible to expose the bug currently
because we use SimplifyDemandedVectorElts from a limited set of opcodes).
See the discussion/implementation from D48987 and D49047.
This patch just enables functionality for FP ops because those do not have
UB/poison potential.
llvm-svn: 343727
Jan Kratochvil [Wed, 3 Oct 2018 21:42:54 +0000 (21:42 +0000)]
Fix buildbot regression by rL339929: NameError: global name 'test_directory' is not defined
With buildbot slave under test - I get after rL339929:
http://lab.llvm.org:8014/builders/lldb-x86_64-fedora-28-cmake/builds/243/steps/test1/logs/stdio
File "/home/buildbot/lldbroot/lldb-x86_64-fedora-28-cmake/scripts/../llvm/tools/lldb/test/dotest.py", line 7, in <module>
lldbsuite.test.run_suite()
File "/quad/home/buildbot/lldbroot/lldb-x86_64-fedora-28-cmake/llvm/tools/lldb/packages/Python/lldbsuite/test/dotest.py", line 1177, in run_suite
configuration.results_formatter_object)
File "/quad/home/buildbot/lldbroot/lldb-x86_64-fedora-28-cmake/llvm/tools/lldb/packages/Python/lldbsuite/test/dosep.py", line 1692, in main
dst = core.replace(test_directory, "")[1:]
NameError: global name 'test_directory' is not defined
Patch by Vedant Kumar.
Differential Revision: https://reviews.llvm.org/D51874
llvm-svn: 343726
Louis Dionne [Wed, 3 Oct 2018 21:36:16 +0000 (21:36 +0000)]
[libc++][NFC] Add error messages to a couple of static_asserts in span
Summary:
Add error messages to a couple of static_asserts in span to match the
style used in the rest of the file. Also fix an extra paren typo in a
assert error message.
Committed on behalf of Jason Lovett.
Reviewers: ldionne
Subscribers: libcxx-commits
Differential Revision: https://reviews.llvm.org/D52841
llvm-svn: 343725
Heejin Ahn [Wed, 3 Oct 2018 21:30:15 +0000 (21:30 +0000)]
Make meanings of variables clearer in action table generation (NFC)
Summary:
Reviewers: kristina, zhmu, dschuff, rnk
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D52680
llvm-svn: 343724
Craig Topper [Wed, 3 Oct 2018 21:10:29 +0000 (21:10 +0000)]
[X86] Stop promoting vector ISD::SELECT to vXi64.
The additional patterns needed for this aren't overwhelming and introducing extra bitcasts during lowering limits our ability to do computeNumSignBits. Not that I have a good example of that for select. I'm just becoming increasingly grumpy about promotion of AND/OR/XOR. SELECT was just a lot easier to fix.
llvm-svn: 343723
Fangrui Song [Wed, 3 Oct 2018 21:00:44 +0000 (21:00 +0000)]
[ASTMatchers] Fix DynamicASTMatchersTests again
llvm-svn: 343722
Fangrui Song [Wed, 3 Oct 2018 20:53:53 +0000 (20:53 +0000)]
[test] Fix -Wunused-variable in rC343665
llvm-svn: 343721
Stephen Kelly [Wed, 3 Oct 2018 20:53:02 +0000 (20:53 +0000)]
Remove stray character from docs
llvm-svn: 343720
Stephen Kelly [Wed, 3 Oct 2018 20:52:57 +0000 (20:52 +0000)]
Update documentation for correctness
llvm-svn: 343719
Shafik Yaghmour [Wed, 3 Oct 2018 20:52:56 +0000 (20:52 +0000)]
Adding skipIf to std::variant libc++ data-formatter test since get is not available before macOS 10.14
Patch by Shafik Yaghmour
llvm-svn: 343718
Stephen Kelly [Wed, 3 Oct 2018 20:52:51 +0000 (20:52 +0000)]
Add matchers missing from dynamic AST registry
llvm-svn: 343717
Stephen Kelly [Wed, 3 Oct 2018 20:52:45 +0000 (20:52 +0000)]
Sort list of matchers
llvm-svn: 343716
George Karpenkov [Wed, 3 Oct 2018 20:46:50 +0000 (20:46 +0000)]
[analyzer] [tests] [quickfix] Make the test more resilient for a non-defaut std configuration
llvm-svn: 343715
Sanjay Patel [Wed, 3 Oct 2018 20:35:25 +0000 (20:35 +0000)]
[InstCombine] add tests for binop undef-into-constant propagation; NFC
llvm-svn: 343714
Craig Topper [Wed, 3 Oct 2018 20:28:43 +0000 (20:28 +0000)]
[X86] Add CMOV_VK2/VK4 pseudos and remove lowering code that turned v2i1/v4i1 SELECT into v8i1.
llvm-svn: 343713
Alex Bradbury [Wed, 3 Oct 2018 20:12:10 +0000 (20:12 +0000)]
[RISCV][NFCI] Handle redundant splitf64+buildpairf64 pairs during instruction selection
Although we can't write a tablegen pattern to remove redundant
splitf64+buildf64 pairs due to the multiple return values, we can handle it
with some C++ selection code. This is simpler than removing them after
instruction selection through RISCVDAGToDAGISel::PostprocessISelDAG, as was
done previously.
llvm-svn: 343712
Patrick Lyster [Wed, 3 Oct 2018 20:07:58 +0000 (20:07 +0000)]
[OPENMP] Add reverse_offload clause to requires directive
llvm-svn: 343711
Craig Topper [Wed, 3 Oct 2018 19:48:26 +0000 (19:48 +0000)]
[X86] Add CMOV pseudos for VR128X and VR256X register classes. Use them when AVX512VL is enabled.
This allows the phi nodes to be generated with the correct register class when expanded.
llvm-svn: 343710
Craig Topper [Wed, 3 Oct 2018 19:48:23 +0000 (19:48 +0000)]
[X86] Don't break CMOV pseudo instructions down by type. Just by register class.
The register class is all that's important for the pseudo instructions. We can use patterns to handle the different types.
llvm-svn: 343709
Simon Pilgrim [Wed, 3 Oct 2018 19:02:38 +0000 (19:02 +0000)]
[X86] PUSH/POP 'mem-mem' instructions are not RMW - these are 2 different addresses
This patch adds a 'WriteCopy' [WriteLoad, WriteStore] schedule sequence instead to better model the behaviour
Found by @andreadb during llvm-mca testing on btver2 which was crashing on "zero uop" WriteRMW only instructions
llvm-svn: 343708
Matthew Voss [Wed, 3 Oct 2018 18:45:04 +0000 (18:45 +0000)]
Add template type and value parameter metadata nodes to template variable specializations
Summary: Add an optional attribute referring to a tuple of type and value template parameter nodes to the DIGlobalVariable node. This allows us to record the parameters of template variable specializations.
Reviewers: dblaikie, aprantl, probinson, JDevlieghere, clayborg, jingham
Reviewed By: JDevlieghere
Subscribers: cfe-commits
Tags: #debug-info
Differential Revision: https://reviews.llvm.org/D52058
llvm-svn: 343707
Matthew Voss [Wed, 3 Oct 2018 18:44:53 +0000 (18:44 +0000)]
Emit template type and value parameter DIEs for template variables.
Summary:
Ensure the TemplateParam attribute of the DIGlobalVariable node is translated into the proper DIEs.
Resolves https://bugs.llvm.org/show_bug.cgi?id=22119
Reviewers: dblaikie, probinson, aprantl, JDevlieghere, clayborg, whitequark, deadalnix
Reviewed By: dblaikie
Subscribers: llvm-commits
Tags: #debug-info
Differential Revision: https://reviews.llvm.org/D52057
llvm-svn: 343706
Simon Pilgrim [Wed, 3 Oct 2018 18:38:28 +0000 (18:38 +0000)]
[X86] Move Atomic binops to use WriteALURMW schedule class
These were being tagged as <WriteALULd, WriteRMW> instead of properly using the RMW sequence
llvm-svn: 343705
Martin Storsjo [Wed, 3 Oct 2018 18:31:53 +0000 (18:31 +0000)]
[COFF] Cope with weak aliases produced by GNU tools
When GNU tools create a weak alias, they produce a strong symbol
named .weak.<weaksymbol>.<relatedstrongsymbol>.
GNU ld allows many such weak alternatives for the same weak symbol, and
the linker picks the first one encountered.
This can't be reproduced by assembling from .s files, since llvm-mc
produces symbols named .weak.<weaksymbol>.default in these cases.
Differential Revision: https://reviews.llvm.org/D52601
llvm-svn: 343704
Julie Hockett [Wed, 3 Oct 2018 18:25:27 +0000 (18:25 +0000)]
[clang-doc] Avoid parsing undefined base classes
Don't try to parse base classes for declarations that are not
definitions (segfaults, as there is no DefinitionData to access).
Differential Revision: https://reviews.llvm.org/D52313
llvm-svn: 343703
Martin Storsjo [Wed, 3 Oct 2018 18:24:05 +0000 (18:24 +0000)]
[test] Use --sysroot instead of -B in print-multi-directory.c
This avoids finding a similar matching GCC installation outside
of the test directory tree in the surrounding environment, which
would make the test fail. (This happened on Ubuntu 16.04.)
Differential Revision: https://reviews.llvm.org/D52533
llvm-svn: 343702
Simon Pilgrim [Wed, 3 Oct 2018 18:18:50 +0000 (18:18 +0000)]
[X86][Btver2] Fix MMX PSHUFB schedule
Match AMD Fam16h SOG + llvm-exegesis tests
llvm-svn: 343701
Simon Pilgrim [Wed, 3 Oct 2018 18:05:01 +0000 (18:05 +0000)]
[X86] Move Atomic CMPXCHG to WriteCMPXCHGRMW schedule class
llvm-svn: 343700
Mandeep Singh Grang [Wed, 3 Oct 2018 17:24:21 +0000 (17:24 +0000)]
[COFF, ARM64] Add _ReadWriteBarrier intrinsic
Reviewers: rnk, mstorsjo, compnerd, TomTan, haripul, javed.absar
Reviewed By: rnk
Subscribers: kristof.beyls, chrib, jfb, cfe-commits
Differential Revision: https://reviews.llvm.org/D52809
llvm-svn: 343699
Nico Weber [Wed, 3 Oct 2018 17:01:39 +0000 (17:01 +0000)]
lld-link: Several tweaks to default entry point selection.
Three related changes:
1. link.exe uses the presence of main and wmain to decide if it should call
mainCRTStartup or wmainCRTStartup, even if /nodefaultlib is passed. For
compatibility, remove FindMain logic.
2. Default to the non-wide entrypoint if main is not found. This has two effects:
2a. In normal links, lld-link now prints
lld-link: error: undefined symbol: _main
>>> referenced by f:\dd\vctools\crt\vcstartup\src\startup\exe_common.inl:78
>>> libcmt.lib(exe_main.obj):("int __cdecl invoke_main(void)" (?invoke_main@@YAHXZ))
>>> referenced by f:\dd\vctools\crt\vcstartup\src\startup\exe_common.inl:283
>>> libcmt.lib(exe_main.obj):("int __cdecl __scrt_common_main_seh(void)" (?__scrt_common_main_seh@@YAHXZ))
instead of
lld-link: error: entry point must be defined
This is arguably a better error message, since it now mentions that _main is
missing. (This matches link.exe's diagnostic in this case.)
2b. With /nodefautlib, we now default to mainCRTStartup if no main() is
present, again matching link.exe. This makes r337407 obsolete.
This means if you have a cc file containing both mainCRTStartup and
wmainCRTStartup and you pass /nodefaultlib /subsystem:console, lld-link will
now call mainCRTStartup, matching link.exe
3. Print a warning if both main and wmain are present, similar to link.exe's
LNK4067.
Differential Revision: https://reviews.llvm.org/D52832
llvm-svn: 343698
Simon Pilgrim [Wed, 3 Oct 2018 16:45:26 +0000 (16:45 +0000)]
[X86] Add SkylakeClient uops counter - same as the other Intel models.
llvm-svn: 343697
Daniel Sanders [Wed, 3 Oct 2018 16:29:24 +0000 (16:29 +0000)]
Correct implementation of -verify-machineinstrs such that it's still overridable for EXPENSIVE_CHECKS
-verify-machineinstrs was implemented as a simple bool. As a result, the
'VerifyMachineCode == cl::BOU_UNSET' used by EXPENSIVE_CHECKS to make it on by
default but possible to disable didn't work as intended. Changed
-verify-machineinstrs to a boolOrDefault to correct this.
llvm-svn: 343696
Adrian Prantl [Wed, 3 Oct 2018 16:24:14 +0000 (16:24 +0000)]
Skip test with older versions of clang
llvm-svn: 343695
Sanjay Patel [Wed, 3 Oct 2018 15:46:03 +0000 (15:46 +0000)]
[InstCombine] clean up foldVectorBinop(); NFC
1. Fix include ordering.
2. Improve variable name (width is bitwidth not number-of-elements).
3. Add local Opcode variable to reduce code duplication.
llvm-svn: 343694
Daniel Sanders [Wed, 3 Oct 2018 15:43:39 +0000 (15:43 +0000)]
[globalisel][combines] Don't sink G_TRUNC down to use if that use is a G_PHI
This fixes a problem where the register allocator fails to eliminate a PHI
because there's a non-PHI in the middle of the PHI instructions at the start
of a BB.
This G_TRUNC can be better placed but this at least fixes the correctness issue
quickly. I'll follow up with a patch to the verifier to catch this kind of bug
in future.
llvm-svn: 343693
Sanjay Patel [Wed, 3 Oct 2018 15:20:58 +0000 (15:20 +0000)]
[InstCombine] name change: foldShuffledBinop -> foldVectorBinop; NFC
This function will deal with more than shuffles with D50992, and I
have another potential per-element fold that could live here.
llvm-svn: 343692
Andrea Di Biagio [Wed, 3 Oct 2018 15:02:44 +0000 (15:02 +0000)]
[llvm-mca] Add support for move elimination in class RegisterFile.
This patch teaches class RegisterFile how to analyze register writes from
instructions that are move elimination candidates.
In particular, it teaches it how to check if a move can be effectively eliminated
by the underlying PRF, and (if necessary) how to perform move elimination.
The long term goal is to allow processor models to describe instructions that
are valid move elimination candidates.
The idea is to let register file definitions in tablegen declare if/when moves
can be eliminated.
This patch is a non functional change.
The logic that performs move elimination is currently disabled. A future patch
will add support for move elimination in the processor models, and enable this
new code path.
llvm-svn: 343691
Simon Pilgrim [Wed, 3 Oct 2018 14:51:09 +0000 (14:51 +0000)]
[llvm-exegesis] Avoid yaml parser from calling sscanf for obvious non-matches (PR39102)
deserializeMCOperand - ensure that we at least match the first character of the sscanf pattern before calling
This reduces llvm-exegesis uops analysis of the instructions supported from btver2 from 5m13s to 2m1s on debug builds.
llvm-svn: 343690
Nirav Dave [Wed, 3 Oct 2018 14:13:30 +0000 (14:13 +0000)]
[X86] Correctly use SSE registers if no-x87 is selected.
Fix use of SSE1 registers for f32 ops in no-x87 mode.
Notably, allow use of SSE instructions for f32 operations in 64-bit
mode (but not 32-bit which is disallowed by callign convention).
Also avoid translating memset/memcopy/memmove into SSE registers
without X87 for 32-bit mode.
This fixes PR38738.
Reviewers: nickdesaulniers, craig.topper
Subscribers: hiraditya, llvm-commits
Differential Revision: https://reviews.llvm.org/D52555
llvm-svn: 343689
Alex Bradbury [Wed, 3 Oct 2018 13:13:13 +0000 (13:13 +0000)]
[RISCV][NFC] Refactor RISCVDAGToDAGISel::Select
Introduce and use a switch on the opcode.
llvm-svn: 343688
James Henderson [Wed, 3 Oct 2018 13:00:20 +0000 (13:00 +0000)]
[ThinLTO]Expose cache entry expiration time option in llvm-lto and fix a test
Two cases in a ThinLTO test were passing for the wrong reasons, since
rL340374. The tests were supposed to be testing that files were being
pruned due to the cache size, but they were in fact being pruned because
they were older than the default expiration period of 1 week.
This change fixes the tests by explicitly setting the expiration time to
the maximum value. This required the option to be exposed in llvm-lto.
By assigning all files in the cache a similar time, it is possible to see
that the newest files are still being kept, and that we aren't passing
for the wrong reason again. In the event that the entry expiration were
to expire for them, then the test would start failing, because these
files would be removed too.
Reviewed by: rnk, inglorion
Differential Revision: https://reviews.llvm.org/D51992
llvm-svn: 343687
Jonas Paulsson [Wed, 3 Oct 2018 12:51:19 +0000 (12:51 +0000)]
[RA CopyHints] Fix compile-time regression
This patch makes sure that a register is only hinted once to RA. In extreme
cases the same register can otherwise be hinted numerous times and cause a
compile time slowdown.
Review: Simon Pilgrim
https://reviews.llvm.org/D52826
llvm-svn: 343686
Clement Courbet [Wed, 3 Oct 2018 12:48:50 +0000 (12:48 +0000)]
[llvm-exegesis][NFC] Revert rL343682 "Fix unused variable warning".
That was not the proper fix: the variable is used in debug mode.
llvm-svn: 343685
Clement Courbet [Wed, 3 Oct 2018 12:35:35 +0000 (12:35 +0000)]
[llvm-exegesis] Fix rL343680 in release mode.
llvm-svn: 343684
Pavel Labath [Wed, 3 Oct 2018 12:29:33 +0000 (12:29 +0000)]
Pull FixupBreakpointPCAsNeeded into base class
Summary:
This function existed (with identical code) in both NativeProcessLinux
and NativeProcessNetBSD, and it is likely that it would be useful to any
future implementation of NativeProcessProtocol.
Therefore I move it to the base class.
Reviewers: krytarowski
Subscribers: lldb-commits
Differential Revision: https://reviews.llvm.org/D52719
llvm-svn: 343683
Clement Courbet [Wed, 3 Oct 2018 12:27:43 +0000 (12:27 +0000)]
[llvm-exegesis][NFC] Fix unused variable warning.
llvm-svn: 343682
Aaron Puchert [Wed, 3 Oct 2018 11:58:19 +0000 (11:58 +0000)]
Thread safety analysis: Unwrap __builtin_expect in getTrylockCallExpr
Summary:
When people are really sure they'll get the lock they sometimes use
__builtin_expect. It's also used by some assertion implementations.
Asserting that try-lock succeeded is basically the same as asserting
that the lock is not held by anyone else (and acquiring it).
Reviewers: aaron.ballman, delesley
Reviewed By: aaron.ballman
Subscribers: kristina, cfe-commits
Differential Revision: https://reviews.llvm.org/D52398
llvm-svn: 343681
Clement Courbet [Wed, 3 Oct 2018 11:50:25 +0000 (11:50 +0000)]
[llvm-exegesis] Resolve variant classes in analysis.
Summary: See PR38884.
Reviewers: gchatelet
Subscribers: tschuett, RKSimon, llvm-commits
Differential Revision: https://reviews.llvm.org/D52825
llvm-svn: 343680
Alex Bradbury [Wed, 3 Oct 2018 11:35:22 +0000 (11:35 +0000)]
[RISCV] Gate float<->int and double<->int conversion patterns on IsRV32
The patterns as defined are correct only when XLen==32.
This is another preparatory patch for a set of patches that flesh out RV64
codegen.
llvm-svn: 343679
Alex Bradbury [Wed, 3 Oct 2018 11:14:26 +0000 (11:14 +0000)]
[RISCV] Remove XLenVT==i32 assumptions from RISCVInstrInfo td
1. brcond operates on an condition.
2. atomic_fence and the pseudo AMO instructions should all take xlen immediates
This allows the same definitions and patterns to work for RV64 (XLenVT==i64).
llvm-svn: 343678
Alex Bradbury [Wed, 3 Oct 2018 11:04:59 +0000 (11:04 +0000)]
[RISCV] Gate simm32 materialisation pattern and SW pattern on IsRV32
These patterns are not correct for RV64.
llvm-svn: 343677
Florian Hahn [Wed, 3 Oct 2018 11:01:23 +0000 (11:01 +0000)]
[LoopInterchange] Remove unused variable PreserveLCSSA (NFC).
llvm-svn: 343676
Alex Bradbury [Wed, 3 Oct 2018 10:59:42 +0000 (10:59 +0000)]
[RISCV] Remove RV64 test lines from umulo-128-legalisation-lowering.ll
The generated code is incorrect anyway, and this test adds noise to the
upcoming set of patches that flesh out RV64 support.
llvm-svn: 343675
Jonas Toth [Wed, 3 Oct 2018 10:59:19 +0000 (10:59 +0000)]
[CodeGen] NFC fix pedantic warning from extra semicolon
llvm-svn: 343674
Jonas Toth [Wed, 3 Oct 2018 10:37:19 +0000 (10:37 +0000)]
[clang-tidy] NFC reorder registering in CppCoreGuidelines module
llvm-svn: 343673
Tim Renouf [Wed, 3 Oct 2018 10:29:43 +0000 (10:29 +0000)]
[AMDGPU] Fix for negative offsets in buffer/tbuffer intrinsics
Summary:
The new buffer/tbuffer intrinsics handle an out-of-range immediate
offset by moving/adding offset&-4096 to a vgpr, leaving an in-range
immediate offset, with a chance of the move/add being CSEd for similar
loads/stores.
However it turns out that a negative offset in a vgpr is illegal, even
if adding the immediate offset makes it legal again.
Therefore, this commit disables the offset&-4096 thing if the offset is
negative.
Differential Revision: https://reviews.llvm.org/D52683
Change-Id: Ie02f0a74f240a138dc2a29d17cfbd9e350e4ed13
llvm-svn: 343672
Simon Pilgrim [Wed, 3 Oct 2018 10:28:43 +0000 (10:28 +0000)]
[X86][Btver2] Most RMW instructions don't require an additional uop
Remove uop on WriteRMW and move it into the few instructions that need it.
Match AMD Fam16h SOG + llvm-exegesis tests
llvm-svn: 343671
Simon Pilgrim [Wed, 3 Oct 2018 10:01:13 +0000 (10:01 +0000)]
[X86] ALU/ADC RMW instructions should use the WriteRMW sequence class
I was expecting this to be a nfc but Silvermont seems to be setup a little differently:
// A folded store needs a cycle on MEC_RSV for the store data, but it does not need an extra port cycle to recompute the address.
def : WriteRes<WriteRMW, [SLM_MEC_RSV]>;
So moving from WriteStore to WriteRMW reduces predicted port pressure, confirmed by @craig.topper that this is correct.
Differential Revision: https://reviews.llvm.org/D52740
llvm-svn: 343670