Wang, Pengfei [Tue, 10 Nov 2020 12:01:45 +0000 (20:01 +0800)]
[CodeGen][X86] Remove unused check-prefixes from some avx512 tests. NFCI.
Kazushi (Jam) Marukawa [Tue, 10 Nov 2020 04:42:24 +0000 (13:42 +0900)]
[VE] Support inline assembly with vector regsiters
Support inline assembly with vector registers. Add a regression test also.
Reviewed By: simoll
Differential Revision: https://reviews.llvm.org/D91146
Sander de Smalen [Tue, 10 Nov 2020 08:32:23 +0000 (08:32 +0000)]
[LoopVectorizer] NFC: Propagate ElementCount to more interfaces.
Interfaces changed to take `ElementCount` as parameters:
* LoopVectorizationPlanner::buildVPlans
* LoopVectorizationPlanner::buildVPlansWithVPRecipes
* LoopVectorizationCostModel::selectVectorizationFactor
This patch is NFC for fixed-width vectors.
Reviewed By: dmgreen, ctetreau
Differential Revision: https://reviews.llvm.org/D90879
Michał Górny [Mon, 9 Nov 2020 11:05:54 +0000 (12:05 +0100)]
[lldb] [test] Rename '.categories' to 'categories'
Make category-specifying files visible. There is really no good reason
to keep them hidden, and having them visible increases the chances
that someone will actually spot them.
Differential Revision: https://reviews.llvm.org/D91065
Michał Górny [Mon, 9 Nov 2020 09:47:22 +0000 (10:47 +0100)]
[lldb] [test] Skip ObjC-based tests via 'objc' category
Replace the plethora of ObjC-implied 'skipUnlessDarwin' decorators
with marking tests as 'objc' category (whenever missing), and skip all
ObjC tests on non-Darwin platforms. I have used '.categories' file
wherever it was present already or all (>1) tests were relying on ObjC,
and explicit add_test_categories() where there was only one test.
Differential Revision: https://reviews.llvm.org/D91056
Joe Ellis [Tue, 10 Nov 2020 10:55:20 +0000 (10:55 +0000)]
[SVE][AArch64] Improve specificity of vectorization legality TypeSize test
The test was using -O2, where -loop-vectorize will suffice.
Reviewed By: fpetrogalli
Differential Revision: https://reviews.llvm.org/D90685
Simon Pilgrim [Tue, 10 Nov 2020 10:48:48 +0000 (10:48 +0000)]
[X86] Remove unused check-prefixes from some vector tests
Simon Pilgrim [Tue, 10 Nov 2020 10:17:46 +0000 (10:17 +0000)]
Add a llvm_unreachable to fix MSVC unhandled enum warning. NFCI.
Stefan Gränitz [Tue, 10 Nov 2020 10:36:59 +0000 (11:36 +0100)]
[lldb][ObjectFile] Relocate sections for in-memory objects (e.g. received via JITLoaderGDB)
Part 2 of a fix for JITed code debugging. This has been a regression from 5.0 to 6.0 and it's still reproducible on current master: https://bugs.llvm.org/show_bug.cgi?id=36209 Part 1 was D61611 a while ago.
The in-memory object files we obtain from JITLoaderGDB are not yet relocated. It looks like this used to happen on the LLDB side and my guess is that it broke with D38142. (However, it's hard to tell because the whole thing was broken already due to the bug in part 1.) The patch moved relocation resolution to a later point in time and didn't apply it to in-memory objects. I am not aware of any reason why we wouldn't resolve relocations per-se, so I made it unconditional here. On Debian, it fixes the bug for me and all tests in `check-lldb` are still fine.
Reviewed By: labath
Differential Revision: https://reviews.llvm.org/D90769
Mirko Brkusanin [Fri, 6 Nov 2020 14:36:17 +0000 (15:36 +0100)]
[GlobalISel] Add combine for (x | mask) -> x when (x | mask) == x
If we have a mask, and a value x, where (x | mask) == x, we can drop the OR
and just use x.
Differential Revision: https://reviews.llvm.org/D90952
Mirko Brkusanin [Fri, 6 Nov 2020 13:15:10 +0000 (14:15 +0100)]
[GlobalISel] Expand combine for (x & mask) -> x when (x & mask) == x
We can use KnownBitsAnalysis to cover cases when mask is not trivial. It can
also help with cases when mask is not constant but can still be folded into
one. Since 'and' is comutative we should treat both operands as possible
replacements.
Differential Revision: https://reviews.llvm.org/D90674
Mirko Brkusanin [Fri, 30 Oct 2020 14:35:42 +0000 (15:35 +0100)]
[AMDGPU][GlobalISel] Combine shift + logic + shift with constant operands
This sequence of instructions can be simplified if they are single use and
some operands are constants. Additional combines may be applied afterwards.
Differential Revision: https://reviews.llvm.org/D90223
Mirko Brkusanin [Thu, 29 Oct 2020 17:12:21 +0000 (18:12 +0100)]
[AMDGPU][GlobalISel] Fold a chain of two shift instructions with constant operands
Sequence of same shift instructions with constant operands can be combined into
a single shift instruction.
Differential Revision: https://reviews.llvm.org/D90217
James Henderson [Fri, 6 Nov 2020 13:55:10 +0000 (13:55 +0000)]
[lld][ELF][test] Add additional --symbol-ordering-file testing
This covers a few cases that aren't otherwise tested:
1) Non-ascii symbol names are ordered.
2) Comments, whitespace and blank lines are trimmed.
3) Missing order files result in an error.
Reviewed by: MaskRay, grimar
Differential Revision: https://reviews.llvm.org/D90933
James Henderson [Tue, 3 Nov 2020 14:41:09 +0000 (14:41 +0000)]
[lld][ELF] Add additional time trace categories
I noticed when running a large link with the --time-trace option that
there were several areas which were missing any specific time trace
categories (aside from the generic link/ExecuteLinker categories). This
patch adds new categories to fill most of the "gaps", or to provide more
detail than was previously provided.
Reviewed by: MaskRay, grimar, russell.gallop
Differential Revision: https://reviews.llvm.org/D90686
Alex Zinenko [Mon, 9 Nov 2020 16:29:21 +0000 (17:29 +0100)]
[mlir] ODS-backed python binding generator for custom op classes
Introduce an ODS/Tablegen backend producing Op wrappers for Python bindings
based on the ODS operation definition. Usage:
mlir-tblgen -gen-python-op-bindings -Iinclude <path/to/Ops.td> \
-bind-dialect=<dialect-name>
Reviewed By: mehdi_amini
Differential Revision: https://reviews.llvm.org/D90960
Kazushi (Jam) Marukawa [Tue, 10 Nov 2020 04:42:24 +0000 (13:42 +0900)]
[VE] Support inline assembly
Support inline assembly with scalar registers. Add a regression test also.
Reviewed By: simoll
Differential Revision: https://reviews.llvm.org/D91119
Alex Zinenko [Fri, 6 Nov 2020 14:09:54 +0000 (15:09 +0100)]
[mlir] Support slicing for operands in results in Python bindings
Slicing, that is element access with `[being:end:step]` structure, is
a common Python idiom for sequence-like containers. It is also necessary
to support custom accessor for operations with variadic operands and
results (an operation an return a slice of its operands that correspond
to the given variadic group).
Add generic utility to support slicing in Python bindings and use it
for operation operands and results.
Depends On D90923
Reviewed By: stellaraccident, mehdi_amini
Differential Revision: https://reviews.llvm.org/D90936
Haojian Wu [Tue, 10 Nov 2020 08:13:32 +0000 (09:13 +0100)]
[clangd] Bump index version number.
https://reviews.llvm.org/D89670 changed the Ref structure, we need to
bump the version to invalidate all stored stale data, otherwise we will
get ` Error while reading shard: malformed or truncated refs` when
building the background index.
Differential Revision: https://reviews.llvm.org/D91131
Kirill Bobyrev [Tue, 10 Nov 2020 09:17:31 +0000 (10:17 +0100)]
[clangd] Update remote index documentation
* Even though remote index is still somewhat experimental, it can now be
used withing clangd itself: this should be the primary way of trying
it out
* Remove `protobuf-compiler` from list of needed Debian packages as it
`protobuf-compiler-grpc` already depends on it
* Bump recommended gRPC version to 1.32.3
Jay Foad [Mon, 9 Nov 2020 10:46:38 +0000 (10:46 +0000)]
[AMDGPU] Remove an unused return value. NFC.
Differential Revision: https://reviews.llvm.org/D91063
Kirill Bobyrev [Tue, 10 Nov 2020 09:08:42 +0000 (10:08 +0100)]
[clangd] Enhance Clangd rename testing coverage
We plan to eliminate error-prone and obsolete Clang-Rename API from Clangd. To
do that, we will introduce Decl canonicalization rules that will make renaming
code simpler and easier to maintain (D71880).
To ensure smooth transition to the new implementation, many Clang-Rename tests
will be adopted in Clangd to prevent any possible regressions. This patch is
the first in the chain of test migration patches. It improves existing tests
and adopts tests from Clang-Rename's alias, class and enum testing files.
Reviewed By: hokein
Differential Revision: https://reviews.llvm.org/D91102
Wang, Pengfei [Tue, 10 Nov 2020 08:06:13 +0000 (16:06 +0800)]
[CodeGen][X86] Remove unused check-prefixes. NFCI.
Artur Bialas [Tue, 10 Nov 2020 08:49:12 +0000 (09:49 +0100)]
[mlir][spirv] Add VectorInsertDynamicOp and vector.insertelement lowering
VectorInsertDynamicOp in SPIRV dialect
conversion from vector.insertelement to spirv VectorInsertDynamicOp
Differential Revision: https://reviews.llvm.org/D90927
Martin Storsjö [Fri, 6 Nov 2020 07:34:17 +0000 (09:34 +0200)]
[libcxx] [test] Simplify the fs helper header for posix cases. NFC.
Differential Revision: https://reviews.llvm.org/D91073
Martin Storsjö [Mon, 9 Nov 2020 11:59:52 +0000 (13:59 +0200)]
[libcxx] [test] Call create_directory_symlink where relevant
Differential Revision: https://reviews.llvm.org/D91072
Martin Storsjö [Mon, 9 Nov 2020 11:59:13 +0000 (13:59 +0200)]
[libcxx] [test] Make a separate create_directory_symlink helper
This more closely mirrors the public API, instead of using an
obscure bool parameter.
Differential Revision: https://reviews.llvm.org/D91071
Martin Probst [Tue, 10 Nov 2020 08:16:49 +0000 (09:16 +0100)]
clang-format: [JS] support new assignment operators.
Before:
a && = b;
After:
a &&= b;
These operators are new additions in ES2021.
Differential Revision: https://reviews.llvm.org/D91132
Max Kazantsev [Tue, 10 Nov 2020 08:00:36 +0000 (15:00 +0700)]
[NFC] Add flag to disable IV widening in indvar instance
This allows us to have control over IV widening in the pipeline.
Haojian Wu [Tue, 10 Nov 2020 07:49:57 +0000 (08:49 +0100)]
[clangd] Add basic conflict detection for the rename.
With this patch, we reject the rename if the new name would conflict with
any other decls in the decl context of the renamed decl.
Differential Revision: https://reviews.llvm.org/D89790
David Blaikie [Tue, 10 Nov 2020 07:27:47 +0000 (23:27 -0800)]
Roll otherwise-unused variable into assert
Esme-Yi [Tue, 10 Nov 2020 06:52:39 +0000 (06:52 +0000)]
[PowerPC] Add an ISEL pattern for Mul with Imm.
Summary: This patch try to do the following transformation if the multiplier doen't fit int16:
(mul X, c1 << c2) -> (rldicr (mulli X, c1) c2)
Reviewed By: jsji, steven.zhang
Differential Revision: https://reviews.llvm.org/D87384
Max Kazantsev [Tue, 10 Nov 2020 06:46:32 +0000 (13:46 +0700)]
[NFC] Different way of getting step
Richard Smith [Mon, 21 Sep 2020 06:18:04 +0000 (23:18 -0700)]
[c++20] For P0732R2 / P1907R1: Basic code generation and name
mangling support for non-type template parameters of class type and
template parameter objects.
The Itanium side of this follows the approach I proposed in
https://github.com/itanium-cxx-abi/cxx-abi/issues/47 on 2020-09-06.
The MSVC side of this was determined empirically by observing MSVC's
output.
Differential Revision: https://reviews.llvm.org/D89998
River Riddle [Tue, 10 Nov 2020 05:50:47 +0000 (21:50 -0800)]
[mlir][Asm] Add support for using an alias for trailing operation locations
Locations often get very long and clutter up operations when printed inline with them. This revision adds support for using aliases with trailing operation locations, and makes printing with aliases the default behavior. Aliases in the trailing location take the form `loc(<alias>)`, such as `loc(#loc0)`. As with all aliases, using `mlir-print-local-scope` can be used to disable them and get the inline behavior.
Differential Revision: https://reviews.llvm.org/D90652
River Riddle [Tue, 10 Nov 2020 05:50:31 +0000 (21:50 -0800)]
[mlir][AsmPrinter] Refactor printing to only print aliases for attributes/types that will exist in the output.
This revision refactors the way that attributes/types are considered when generating aliases. Instead of considering all of the attributes/types of every operation, we perform a "fake" print step that prints the operations using a dummy printer to collect the attributes and types that would actually be printed during the real process. This removes a lot of attributes/types from consideration that generally won't end up in the final output, e.g. affine map attributes in an `affine.apply`/`affine.for`.
This resolves a long standing TODO w.r.t aliases, and helps to have a much cleaner textual output format. As a datapoint to the latter, as part of this change several tests were identified as testing for the presence of attributes aliases that weren't actually referenced by the custom form of any operation.
To ensure that this wouldn't cause a large degradation in compile time due to the second full print, I benchmarked this change on a very large module with a lot of operations(The file is ~673M/~4.7 million lines long). This file before this change take ~6.9 seconds to print in the custom form, and ~7 seconds after this change. In the custom assembly case, this added an average of a little over ~100 miliseconds to the compile time. This increase was due to the way that argument attributes on functions are structured and how they get printed; i.e. with a better representation the negative impact here can be greatly decreased. When printing in the generic form, this revision had no observable impact on the compile time. This benchmarking leads me to believe that the impact of this change on compile time w.r.t printing is closely related to `print` methods that perform a lot of additional/complex processing outside of the OpAsmPrinter.
Differential Revision: https://reviews.llvm.org/D90512
Max Kazantsev [Tue, 10 Nov 2020 05:37:12 +0000 (12:37 +0700)]
[SCEV] Drop cached ranges of AddRecs after flag update
Our range computation methods benefit from no-wrap flags. But if the ranges
were first computed before the flags were set, the cached range will be too
pessimistic.
We need to drop cached ranges whenever we sharpen AddRec's no wrap flags.
Differential Revision: https://reviews.llvm.org/D89847
Reviewed By: fhahn
Haowei Wu [Tue, 10 Nov 2020 04:06:20 +0000 (20:06 -0800)]
[scan-build] Supprot relative 'file' in cdb.
Excluded folders in scan build is turned to absolute path before
comapre to 'file' in cdb. 'file' in cdb might be a path relative
to 'directory', so we need to turn it to absolute path before
comparison.
Patch by Yu Shan
Differential Revision: https://reviews.llvm.org/D90362
Carl Ritson [Tue, 10 Nov 2020 03:16:26 +0000 (12:16 +0900)]
[AMDGPU] Fix lowering of S_MOV_{B32,B64}_term
If the source of S_MOV_{B32,B64}_term is an immediate then it
cannot be lowered to a COPY.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D90451
Qiu Chaofan [Tue, 10 Nov 2020 02:52:13 +0000 (10:52 +0800)]
[PowerPC] [Clang] Port SSE4.1-compatible insert intrinsics
This patch adds three intrinsics compatible to x86's SSE 4.1 on PowerPC
target, with tests:
- _mm_insert_epi8
- _mm_insert_epi32
- _mm_insert_epi64
The intrinsics implementation is contributed by Paul Clarke.
Reviewed By: jsji
Differential Revision: https://reviews.llvm.org/D89242
Arthur Eubanks [Mon, 9 Nov 2020 19:40:49 +0000 (11:40 -0800)]
[NewPM] Port -separate-const-offset-from-gep
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D91095
Michael Kruse [Tue, 10 Nov 2020 00:47:41 +0000 (18:47 -0600)]
[OMPIRBuilder] Start 'Create' methods with lower case. NFC.
For consistency with the IRBuilder, OpenMPIRBuilder has method names starting with 'Create'. However, the LLVM coding style has methods names starting with lower case letters, as all other OpenMPIRBuilder already methods do. The clang-tidy configuration used by Phabricator also warns about the naming violation, adding noise to the reviews.
This patch renames all `OpenMPIRBuilder::CreateXYZ` methods to `OpenMPIRBuilder::createXYZ`, and updates all in-tree callers.
I tested check-llvm, check-clang, check-mlir and check-flang to ensure that I did not miss a caller.
Reviewed By: mehdi_amini, fghanim
Differential Revision: https://reviews.llvm.org/D91109
Kazu Hirata [Tue, 10 Nov 2020 01:29:40 +0000 (17:29 -0800)]
[BranchProbabilityInfo] Use SmallVector (NFC)
This patch simplifies BranchProbabilityInfo by changing the type of
Probs.
Without this patch:
DenseMap<Edge, BranchProbability> Probs
maps an ordered pair of a BasicBlock* and a successor index to an edge
probability.
With this patch:
DenseMap<const BasicBlock *, SmallVector<BranchProbability, 2>> Probs
maps a BasicBlock* to a vector of edge probabilities.
BranchProbabilityInfo has a property that for a given basic block, we
either have edge probabilities for all successors or do not have any
edge probability at all. This property combined with the current map
type leads to a somewhat complicated algorithm in eraseBlock to erase
map entries one by one while increasing the successor index.
The new map type allows us to remove the all edge probabilities for a
given basic block in a more intuitive manner, namely:
Probs.erase(BB);
Differential Revision: https://reviews.llvm.org/D91017
Xun Li [Tue, 10 Nov 2020 01:24:49 +0000 (17:24 -0800)]
[Coroutine] Move all used local allocas to the .resume function
Prior to D89768, any alloca that's used after suspension points will be put on to the coroutine frame, and hence they will always be reloaded in the resume function.
However D89768 introduced a more precise way to determine whether an alloca should live on the frame. Allocas that are only used within one suspension region (hence does not need to live across suspension points) will not be put on the frame. They will remain local to the resume function.
When creating the new entry for the .resume function, the existing logic only moved all the allocas from the old entry to the new entry. This covers every alloca from the old entry. However allocas that's defined afer coro.begin are put into a separate basic block during CoroSplit (the PostSpill basic block). We need to make sure these allocas are moved to the new entry as well if they are used.
This patch walks through all allocas, and check if they are still used but are not reachable from the new entry, if so, we move them to the new entry.
Differential Revision: https://reviews.llvm.org/D90977
Jonas Devlieghere [Tue, 10 Nov 2020 00:36:47 +0000 (16:36 -0800)]
[lldb] Don't use ::exit but instead return from the driver loop (NFC)
This fixes a reproducer test failure that was caused by the undefined
order in which global destructors run. More concretely, the static
instance of the RealFileSystem had been destroyed before we finalized
the reproducer, which uses it to copy files into the reproducer. By
exiting normally, we call SBDebugger::Terminate and finalize the
reproducer before any static dtors are run.
Jonas Devlieghere [Tue, 10 Nov 2020 00:36:03 +0000 (16:36 -0800)]
[lldb] Destory the debugger in the Driver dtor (NFC)
Sam Clegg [Sun, 8 Nov 2020 22:28:13 +0000 (14:28 -0800)]
[libc++] Remove emscripten handling from exception_fallback.ipp
Emscripten doesn't use this file (at least not anymore), it uses
exception_libcxxabi.ipp since _LIBCPPABI_VERSION is defined.
Differential Revision: https://reviews.llvm.org/D91041
Josh Stone [Mon, 9 Nov 2020 23:36:14 +0000 (15:36 -0800)]
Enable opt-bisect for the new pass manager
This instruments a should-run-optional-pass callback using the existing
OptBisect class to decide if new passes should be skipped. Passes that
force isRequired never reach this at all, so they are not included in
"BISECT:" output nor its pass count.
The test case is resurrected from r267022, an early version of D19172
that had new pass manager support (later reverted and redone without).
Reviewed By: aeubanks
Differential Revision: https://reviews.llvm.org/D87951
Arthur Eubanks [Mon, 9 Nov 2020 18:35:23 +0000 (10:35 -0800)]
[update_test_checks] Allow opt to have .exe file extension
Reviewed By: rnk
Differential Revision: https://reviews.llvm.org/D91091
Jonas Devlieghere [Mon, 9 Nov 2020 23:25:59 +0000 (15:25 -0800)]
[lldb] Make GetSelectedOrDummyTarget return the target by reference (NFC)
Return references from GetDummyTarget and GetSelectedOrDummyTarget. This
matches how the APIs are already used in practice.
LLVM GN Syncbot [Mon, 9 Nov 2020 23:03:42 +0000 (23:03 +0000)]
[gn build] Port
dbfa69c5024
Lei Zhang [Mon, 9 Nov 2020 22:57:20 +0000 (17:57 -0500)]
[mlir][spirv] Expose more query APIs directly on TargetEnv
This allows us to omit one level of indirection when querying
the information from the underlying attribute.
Reviewed By: hanchung, ThomasRaoux
Differential Revision: https://reviews.llvm.org/D91080
Jan Svoboda [Mon, 9 Nov 2020 20:51:42 +0000 (15:51 -0500)]
Port some floating point options to new option marshalling infrastructure
This ports a number of OpenCL and fast-math flags for floating point
over to the new marshalling infrastructure.
As part of this, `Opt{In,Out}FFlag` were enhanced to allow other flags to
imply them, via `DefaultAnyOf<>`. For example:
```
defm signed_zeros : OptOutFFlag<"signed-zeros", ...,
"LangOpts->NoSignedZero",
DefaultAnyOf<[cl_no_signed_zeros, menable_unsafe_fp_math]>>;
```
defines `-fsigned-zeros` (`false`) and `-fno-signed-zeros` (`true`)
linked to the keypath `LangOpts->NoSignedZero`, defaulting to `false`,
but set to `true` implicitly if one of `-cl-no-signed-zeros` or
`-menable-unsafe-fp-math` is on.
Note that the initial patch was written Daniel Grumberg.
Differential Revision: https://reviews.llvm.org/D82756
Sam McCall [Mon, 9 Nov 2020 22:58:51 +0000 (23:58 +0100)]
Revert "[Syntax] Add minimal TableGen for syntax nodes. NFC"
This reverts commit
55120f74ca12faea0e90fe552c85c14485f1fd91.
Segfaults during build:
http://lab.llvm.org:8011/#/builders/36/builds/1310
Keith Smiley [Mon, 9 Nov 2020 22:47:17 +0000 (14:47 -0800)]
[llvm-cov][NFC] Fix typo in format comment
Sam McCall [Sat, 31 Oct 2020 20:09:11 +0000 (21:09 +0100)]
[Syntax] Add minimal TableGen for syntax nodes. NFC
So far, only used to generate Kind and implement classof().
My plan is to have this general-purpose Nodes.inc in the style of AST
DeclNodes.inc etc, and additionally a special-purpose backend generating
the actual class definitions. But baby steps...
Differential Revision: https://reviews.llvm.org/D90540
Kirill Bobyrev [Mon, 9 Nov 2020 22:03:39 +0000 (23:03 +0100)]
Revert "[llvm] CMake: Force MSVC to read code as UTF-8"
This reverts commit
4d81c8adb6ed9840257f6cb6b93f60856d422a15.
Apparently, we have code that indirectly uses #pragma
execution_character_set which depends on utf-8 not being set:
http://lab.llvm.org:8011/#/builders/127/builds/1161/steps/4/logs/stdio
Rahul Joshi [Mon, 9 Nov 2020 19:43:45 +0000 (11:43 -0800)]
[MLIR] Add setPublic(), setPrivate(), and setNested() to Symbol interface
- Add shorter helper functions to set visibility for Symbols.
Differential Revision: https://reviews.llvm.org/D91096
Louis Dionne [Mon, 9 Nov 2020 21:10:13 +0000 (16:10 -0500)]
[runtimes] Avoid overwriting the rpath unconditionally
When building the runtimes, it's very important not to add rpaths unless
the user explicitly asks for them (the standard way being CMAKE_INSTALL_RPATH),
or to change the install name dir unless the user requests it (via
CMAKE_INSTALL_NAME_DIR).
llvm_setup_rpath() would override the install_name_dir of the runtimes
even if CMAKE_INSTALL_NAME_DIR was specified to something, which is wrong
and in fact even "dangerous" for the runtimes.
This issue was discovered when trying to build libc++ and libc++abi as
system libraries for Apple, where we set the install name dir to /usr/lib
explicitly. llvm_setup_rpath() would cause libc++ to have the wrong install
name dir, and for basically everything on the system to fail to load.
This was discovered just now because we previously used something closer
to a standalone build, where llvm_setup_rpath() wouldn't exist, and hence
not be used.
This is a revert of the following commits:
libunwind:
3a667b9bd8b741f5ac1d8d47857140a3d70737fb
libc++abi:
4877063e195dfcc128451bbf3dd7b03d04d2562f
libc++:
88434fe05fdb112a33052c4d8a91c9e989cb032d
Those added llvm_setup_rpath() for consistency, so it seems reasonable
to revert.
Differential Revision: https://reviews.llvm.org/D91099
Kirill Bobyrev [Mon, 9 Nov 2020 21:49:49 +0000 (22:49 +0100)]
[clang] Simplify buildSyntaxTree API
Follow-up on https://reviews.llvm.org/D88553#inline-837013
Reviewed By: sammccall
Differential Revision: https://reviews.llvm.org/D90672
Kirill Bobyrev [Mon, 9 Nov 2020 21:47:16 +0000 (22:47 +0100)]
[llvm] CMake: Force MSVC to read code as UTF-8
Symptoms: https://github.com/clangd/clangd/issues/571
Reviewed By: kadircet
Differential Revision: https://reviews.llvm.org/D90116
Hansang Bae [Wed, 21 Oct 2020 13:50:31 +0000 (08:50 -0500)]
[OpenMP] Remove obsolete Fortran module file
Modern Fortran compilers support Fortran 90, so we do not need to use
the source code for Fortran compilers that do not support Fortran 90.
Differential Revision: https://reviews.llvm.org/D90077
Kadir Cetinkaya [Mon, 9 Nov 2020 20:54:38 +0000 (21:54 +0100)]
[clangd] Fix shared-lib builds
This breaks a cyclic dependency. clangDeamon doesn't need to depend on
clangdRemoteIndex yet.
Michał Górny [Mon, 9 Nov 2020 14:47:37 +0000 (15:47 +0100)]
[lldb] [Process/NetBSD] Correct DS/ES/FS/GS register sizes
Fix DS/ES/FS/GS register sizes in getter/setter for NetBSD. Apparently
only CS and SS registers are specified as 64/32-bit in LLDB, while
the others are specified as 16-bit.
Differential Revision: https://reviews.llvm.org/D91090
Michael Kruse [Mon, 9 Nov 2020 20:13:17 +0000 (14:13 -0600)]
[OpenMPIRBuilder] Implement CreateCanonicalLoop.
CreateCanonicalLoop generates a standardized control flow structure for OpenMP canonical for loops. The structure can be consumed by loop-associated directives such as worksharing-loop, distribute, simd etc. as well as loop transformations such as tile and unroll.
This is a first design without considering all complexities yet. The control-flow emits more basic block than strictly necessary, but these will be optimized by CFGSimplify anyway, provide a nice separation of concerns and might later be useful with more complex scenarios. I successfully implemented a basic tile construct using this API, which is not part of this patch.
The fundamental building block is the CreateCanonicalLoop that only takes the loop trip count and operates on the logical iteration spaces only. An overloaded CreateCanonicalLoop for using LB, UB, Increment is provided as well, but at least for C++, Clang will need to implement a loop counter to logical induction variable mapping anyway, since iterator overload resolution cannot be done in LLVMFrontend.
As there currently is no user for CreateCanonicalLoop, it is only called from unittests. Similarly, CanonicalLoopInfo::eraseFromParent() is used in my file implementation and might be generally useful for implementing loop-associated constructs, but is not used in this patch itself.
The following non-exhaustive list describes not yet covered items:
* collapse clause (including non-rectangular and non-perfectly nested); idea is to provide a OpenMPIRBuilder::collapseLoopNest method consuming multiple nested loops and returning a new CanonicalLoopInfo that can be used for loop-associated directives.
* simarly: ordered clause for DOACROSS loops
* branch weights
* Cancellation point (?)
* AllocaIP
* break statement (if needed at all)
* Exceptions (if not completely handled in the front-end)
* Using it in Clang; this requires implementing at least one loop-associated construct.
* ...
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D90830
Alexander Belyaev [Mon, 9 Nov 2020 15:47:02 +0000 (16:47 +0100)]
[mlir][std] Add ExpandOps pass.
The pass combines patterns of ExpandAtomic, ExpandMemRefReshape,
StdExpandDivs passes. The pass is meant to legalize STD for conversion to LLVM.
Differential Revision: https://reviews.llvm.org/D91082
Arthur Eubanks [Mon, 9 Nov 2020 20:54:13 +0000 (12:54 -0800)]
[NewPM] Add unique-internal-linkage-names to PassRegistry.def
Pass was already ported, just not properly hooked up.
Arthur Eubanks [Mon, 9 Nov 2020 20:47:57 +0000 (12:47 -0800)]
[test] Completely pin PR37334-break-crit-edges-require-dt.ll to legacy PM
This tests legacy PM specific code (https://bugs.llvm.org/show_bug.cgi?id=37334).
This is failing under the new PM because -loop-sink is a function pass,
not a loop pass, and doesn't run -loop-simplify.
Duncan P. N. Exon Smith [Thu, 5 Nov 2020 22:12:17 +0000 (17:12 -0500)]
Tooling: Remove dead code for ToolingInvocation::mapVirtualFile
Follows through on
c4cb3b10dc8c50e46c9fb1b7ae95e3c3c94975d3's FIXME
dating back to 2015. Anyone using this should migrate to
InMemoryFileSystem and/or ClangTool::mapVirtualFile.
Differential Revision: https://reviews.llvm.org/D90885
Arthur Eubanks [Mon, 9 Nov 2020 20:39:51 +0000 (12:39 -0800)]
[test][NewPM] Pin -flattencfg test to legacy PM
It is not used in the optimization pipeline and can be ported later.
Nikita Popov [Mon, 9 Nov 2020 19:42:57 +0000 (20:42 +0100)]
[BasicAA] Add test for decomposition limit (NFC)
Test behavior before/at/after the GEP decomposition limit.
Arthur Eubanks [Mon, 9 Nov 2020 20:19:03 +0000 (12:19 -0800)]
[test][NewPM] Fix LoopLoadElim tests under NPM
Snehasish Kumar [Sat, 7 Nov 2020 00:24:25 +0000 (16:24 -0800)]
[llvm] Check the debug info line table for basic block sections.
Extend the existing basic block sections debug info test to check for the correctness of the generated line table.
Differential Revision: https://reviews.llvm.org/D90989
Kirill Bobyrev [Mon, 9 Nov 2020 20:18:35 +0000 (21:18 +0100)]
[clangd] NFC: Fix a typo in Tracer name
Fangrui Song [Mon, 9 Nov 2020 20:10:44 +0000 (12:10 -0800)]
[FastISel][test] %llc_dwarf -mtriple -> llc -mtriple
if config.target_triple contains 'windows-msvc', %llc_dwarf has a -mtriple:
llc.exe: for the --mtriple option: may only occur zero or one times!
Rahul Joshi [Mon, 9 Nov 2020 19:57:52 +0000 (11:57 -0800)]
[MLIR] Fix GCC build failure
António Afonso [Mon, 9 Nov 2020 19:45:35 +0000 (11:45 -0800)]
Revert "Check if debug line sequences are starting after the first code segment"
This reverts commit
265a38fbc547adc0e2b71a98dac4d0a60b0dd63c.
Stephen Kelly [Mon, 9 Nov 2020 18:59:30 +0000 (18:59 +0000)]
Fix use of directly-nested traverse() matchers
Stephen Kelly [Mon, 9 Nov 2020 18:59:00 +0000 (18:59 +0000)]
Fix trailing whitespace
My editor keeps on changing this and I keep having to revert it.
Stephen Kelly [Fri, 6 Nov 2020 16:05:16 +0000 (16:05 +0000)]
Change algorithms to return iterators
Make it possible to inspect the matched node, possibly to ignore it.
Differential Revision: https://reviews.llvm.org/D90983
Rahul Joshi [Mon, 9 Nov 2020 16:23:55 +0000 (08:23 -0800)]
[MLIR] Change FuncOp assembly syntax to print visibility inline instead of in attrib dict.
- Change syntax for FuncOp to be `func <visibility>? @name` instead of printing the
visibility in the attribute dictionary.
- Since printFunctionLikeOp() and parseFunctionLikeOp() are also used by other
operations, make the "inline visibility" an opt-in feature.
- Updated unit test to use and check the new syntax.
Differential Revision: https://reviews.llvm.org/D90859
Eric Astor [Mon, 9 Nov 2020 18:22:37 +0000 (13:22 -0500)]
[ms] [llvm-ml] Support MASM's relational operators (EQ, LT, etc.)
Support the named relational operators (EQ, LT, etc.).
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D89733
Rahul Joshi [Mon, 9 Nov 2020 17:01:39 +0000 (09:01 -0800)]
[MLIR] Support `global_memref` and `get_global_memref` in standard -> LLVM conversion.
- Convert `global_memref` to LLVM::GlobalOp.
- Convert `get_global_memref` to a memref descriptor with a pointer to the first element
of the global stashed in it.
- Extend unit test and a mlir-cpu-runner test to validate the generated LLVM IR.
Differential Revision: https://reviews.llvm.org/D90803
David Zarzycki [Wed, 28 Oct 2020 11:18:09 +0000 (07:18 -0400)]
[SelectionDAG] Enable CTPOP optimization fine tuning
Add a TLI hook to allow SelectionDAG to fine tune the conversion of CTPOP to a chain of "x & (x - 1)" when CTPOP isn't legal.
A subsequent patch will attempt to fine tune the X86 code gen.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D89952
Aleksandr Platonov [Mon, 9 Nov 2020 18:39:01 +0000 (21:39 +0300)]
[clangd][remote] Check an index file correctly
There is not reason to check `std::make_unique<...>(..)` return value,
but `clangd::clang::loadIndex()` returns `nullptr` if an index file could not be loaded (e.g. incorrect version).
Reviewed By: kadircet
Differential Revision: https://reviews.llvm.org/D91049
Fangrui Song [Mon, 9 Nov 2020 18:26:02 +0000 (10:26 -0800)]
[FastISel][test] Add ELF triple after D90877
MachO has different symbol naming and thus on a MachO platform this test has a
different behavior without a triple.
Francesco Petrogalli [Mon, 9 Nov 2020 18:18:28 +0000 (18:18 +0000)]
[llvm][AArch64] Allow TB(N)Z to drop signext for sign bit tests.
For example if the sign extension is only used in for TBZ, and the value is used elsewhere with a zero extension, this can eliminate a sign extension.
Reviewed By: samparker
Differential Revision: https://reviews.llvm.org/D90606
Peter Steinfeld [Mon, 9 Nov 2020 15:42:12 +0000 (07:42 -0800)]
[flang] Avoid calling the linker when "-c" option is used
The title says it all.
Eric Astor [Mon, 9 Nov 2020 16:49:41 +0000 (11:49 -0500)]
[ms] [llvm-ml] Support REPEAT/FOR/WHILE macro-like directives
Support MASM's REPEAT, FOR, FORC, and WHILE macro-like directives.
Also adds support for macro argument substitution inside quoted strings, and additional testing for macro directives.
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D89732
Fangrui Song [Mon, 9 Nov 2020 18:19:12 +0000 (10:19 -0800)]
[libunwind] Delete unused codeOffsetAtStackDecrement/registersInOtherRegisters/sameValueUsed
ld64 uses them to create compact unwind from DWARF call frame information.
When the code was ported to libunwind, the variables were not deleted.
Reviewed By: #libunwind, compnerd
Differential Revision: https://reviews.llvm.org/D91039
David Green [Mon, 9 Nov 2020 18:18:43 +0000 (18:18 +0000)]
[ARM] Remove MI variable aliasing. NFC
This was accidentally using the same name for two different variables in
the same line. Whilst it seems to work for some compilers, others have
trouble and it is probably not a fantastic idea.
Craig Topper [Mon, 9 Nov 2020 18:05:51 +0000 (10:05 -0800)]
[RISCV] Make ctlz/cttz cheap to speculatively execute so CodeGenPrepare won't insert a zero check.
Add additional isel patterns for ctzw/clzw instructions.
Differential Revision: https://reviews.llvm.org/D91040
Craig Topper [Mon, 9 Nov 2020 18:01:55 +0000 (10:01 -0800)]
[RISCV] Add isel patterns for using PACK for zext.h and zext.w.
Differential Revision: https://reviews.llvm.org/D91024
Craig Topper [Mon, 9 Nov 2020 17:57:38 +0000 (09:57 -0800)]
[RISCV] Make SIGN_EXTEND_INREG from i8/i16 legal when Zbb extension is enabled.
This produces better code for sign extend to i64 on RV32 target.
Differential Revision: https://reviews.llvm.org/D91023
Fangrui Song [Mon, 9 Nov 2020 18:08:44 +0000 (10:08 -0800)]
[ELF] Special case static_assert for _WIN32
I don't have a Windows machine. Hope someone can test why its InputSection is
still larger.
Craig Topper [Mon, 9 Nov 2020 17:45:22 +0000 (09:45 -0800)]
[RISCV] Add isel patterns to match sbset/sbclr/sbinv/sbext even if the shift amount isn't masked.
This uses the shiftop PatFrags to handle the masked shift amount
and unmasked shift amount cases. That also checks XLen as part
of the masked amount check so we don't need separate RV32 and RV64
patterns.
Differential Revision: https://reviews.llvm.org/D91016
Fangrui Song [Mon, 9 Nov 2020 17:55:09 +0000 (09:55 -0800)]
[ELF] Make InputSection smaller
On LP64/Windows platforms, this decreases sizeof(InputSection) from 208 (larger
on Windows) to 184.
For a large executable (7.6GiB, inputSections.size()=5105122,
make<InputSection> called 4835760 times), this decreases cgroup
memory.max_usage_in_bytes by 0.6%
Reviewed By: grimar
Differential Revision: https://reviews.llvm.org/D91018
Simon Pilgrim [Mon, 9 Nov 2020 17:50:03 +0000 (17:50 +0000)]
[X86] vector-narrow-binop.ll - remove unused check-prefixes
Simon Pilgrim [Mon, 9 Nov 2020 17:46:49 +0000 (17:46 +0000)]
[HardwareLoops] Remove unused check-prefixes
Just use default CHECK
Paul Robinson [Thu, 5 Nov 2020 21:09:10 +0000 (13:09 -0800)]
[FastISel] Reduce spills around mem-intrinsic calls
FastISel generates instructions to materialize "local values" at the
top of a block, in the hope that these values could be reused within
the block. To reduce spills and restores, FastISel treats calls as
sub-block boundaries, flushing the "local value map" at each call.
This patch treats the mem* intrinsics as if they were calls, because
at O0 generally they are calls. Eliminating these spills/restores is
actually better for debugging (especially a "continue at this line"
command), code size, stack frame size, and maybe even performance.
Differential Revision: https://reviews.llvm.org/D90877