Anastasia Stulova [Fri, 15 Jan 2021 17:19:16 +0000 (17:19 +0000)]
[OpenCL][Docs] Fixed cross-section reference in OpenCLSupport
Tags: #clang
Utkarsh Saxena [Thu, 14 Jan 2021 17:01:25 +0000 (18:01 +0100)]
[clangd] Update CC Ranking model with better sampling.
A better sampling strategy was used to generate the dataset for this
model.
New signals introduced in this model:
- NumNameInContext: Number of words in the context that matches the name
of the candidate.
- FractionNameInContext: Fraction of the words in context matching the
name of the candidate.
We remove the signal `IsForbidden` from the model and down rank
forbidden signals aggresively.
Differential Revision: https://reviews.llvm.org/D94697
Andrzej Warzynski [Fri, 15 Jan 2021 16:45:01 +0000 (16:45 +0000)]
[flang][driver] Copy test file into a temp dir when testing (nfc)
The following driver invocation will generate an output file
in the same directory as the input file:
```
flang-new -fc1 -test-io test-input.f90
```
This is the desired behaviour. However, when testing we need to make
sure that we don't pollute the source directory. To this end, copy the
input file into a temporary directory before testing.
This is similar to https://reviews.llvm.org/D94243.
Craig Topper [Fri, 15 Jan 2021 17:05:34 +0000 (09:05 -0800)]
[TargetLowering] Don't speculatively call ComputeNumSignBits. NFC
These methods are recursive so a little costly.
We only look at the result in one place in this function and it's
conditional. We also only need the second call if the first had
enough returned enough sign bits.
Peter Steinfeld [Fri, 15 Jan 2021 15:04:20 +0000 (07:04 -0800)]
[flang] Disallow INTENT attribute on procedure dummy arguments
C843 states that "An entity with the INTENT attribute shall be a dummy
data object or a dummy procedure pointer." This change enforces that
and fixes some tests that erroneously violated this rule.
Differential Revision: https://reviews.llvm.org/D94781
Simon Pilgrim [Fri, 15 Jan 2021 16:25:05 +0000 (16:25 +0000)]
[X86][SSE] Attempt to fold shuffle(binop(),binop()) -> binop(shuffle(),shuffle())
If this will help us fold shuffles together, then push the shuffle through the merged binops.
Ideally this would be performed in DAGCombiner::visitVECTOR_SHUFFLE but getting an efficient+legal merged shuffle can be tricky - on SSE we can be confident that for 32/64-bit elements vectors shuffles should easily fold.
Sam Tebbs [Fri, 15 Jan 2021 14:43:10 +0000 (14:43 +0000)]
[ARM][Block placement] Check the predecessor exists before processing it
Not all machine loops will have a predecessor. so the pass needs to
check it before continuing.
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D94780
Jan Svoboda [Fri, 15 Jan 2021 14:44:09 +0000 (15:44 +0100)]
[clang][cli] NFC: Promote ParseLangArgs and ParseCodeGenArgs to members
This patch promotes `ParseLangArgs` and `ParseCodeGenArgs` to members of `CompilerInvocation`. That will be useful in the following patch D94682, where we need to access protected members of `LangOptions` and `CodeGenOptions`. Both of those classes already have `friend CompilerInvocation`.
This is cleaner than keeping those functions freestanding and having to specify the exact signature of both in extra `friend` declarations.
Reviewed By: Bigcheese
Differential Revision: https://reviews.llvm.org/D94681
Jan Svoboda [Fri, 15 Jan 2021 14:43:18 +0000 (15:43 +0100)]
[clang][cli] NFC: Parse some LangOpts after the defaults are set
This patch ensures we only parse the necessary options before calling `setLangDefaults` (explained in D94678).
Because neither `LangOpts.CFProtectionBranch` nor `LangOpts.SYCLIsDevice` are used in `setLangDefaults`, this is a NFC.
Reviewed By: Bigcheese
Differential Revision: https://reviews.llvm.org/D94680
Jan Svoboda [Fri, 15 Jan 2021 14:42:29 +0000 (15:42 +0100)]
[clang][cli] NFC: Add PIE parsing for precompiled input and IR
This patch effectively reverts a small part of D83979.
When we stop parsing `LangOpts` unconditionally in `parseSimpleArgs` (above the diff) and move them back to `ParseLangArgs` (called in `else` branch) in D94682, `LangOpts.PIE` would never get parsed in this `if` branch. This patch ensures this doesn't happen.
Right now, this causes `LangOpts.PIE` to be parsed twice, but that will be immediately corrected in D94682.
Reviewed By: Bigcheese
Differential Revision: https://reviews.llvm.org/D94679
Michael Munday [Fri, 15 Jan 2021 15:32:31 +0000 (15:32 +0000)]
[RISCV][NFC] Fix order of parameters in cmov ge/le tests
The first parameter should be selected if the condition is true and
the last parameter if the condition is false. Prior to this change
it was the other way round which was confusing.
Differential Revision: https://reviews.llvm.org/D94729
Guillaume Chatelet [Fri, 15 Jan 2021 15:32:02 +0000 (15:32 +0000)]
[libc] CopyAlignedBlocks can now specify alignment on top of block size
This has been requested in D92236
Differential Revision: https://reviews.llvm.org/D94770
Valentin Clement [Fri, 15 Jan 2021 15:29:37 +0000 (10:29 -0500)]
[mlir] Add better support for f80 and f128
Add builtin f80 and f128 following @schweitz proposition
https://llvm.discourse.group/t/rfc-adding-better-support-for-higher-precision-floating-point/2526/5
Reviewed By: ftynse, rriddle
Differential Revision: https://reviews.llvm.org/D94737
Simon Pilgrim [Fri, 15 Jan 2021 15:08:17 +0000 (15:08 +0000)]
[DAG] visitVECTOR_SHUFFLE - MergeInnerShuffle - improve shuffle(shuffle(x,y),shuffle(x,y)) merging
MergeInnerShuffle currently attempts to merge shuffle(shuffle(x,y),z) patterns into a single shuffle, using 1 or 2 of the x,y,z ops.
However if we already match 2 ops we might be able to handle the third op if its also a shuffle that references one of the previous ops, allowing us to handle some cases like:
shuffle(shuffle(x,y),shuffle(x,y))
shuffle(shuffle(shuffle(x,z),y),z)
shuffle(shuffle(x,shuffle(x,y)),z)
etc.
This isn't an exhaustive match and is dependent on the order the candidate ops are encountered - if one of the matched ops was a shuffle that was peek-able we don't go back and try to split that, I haven't found much need for that amount of analysis yet.
This is a preliminary patch that will allow us to later improve x86 HADD/HSUB matching - but needs to be reviewed separately as its in generic code and affects existing Thumb2 tests.
Differential Revision: https://reviews.llvm.org/D94671
Jamie Schmeiser [Fri, 15 Jan 2021 14:51:34 +0000 (09:51 -0500)]
Set option default for enabling memory ssa for new pass manager loop sink pass to true.
Summary:
Set the default for the option enabling memory ssa use in the loop sink
pass to true for the new pass manager.
Author: Jamie Schmeiser <schmeise@ca.ibm.com>
Reviewed By: asbirlea (Alina Sbirlea)
Differential Revision: https://reviews.llvm.org/D92486
Simon Pilgrim [Fri, 15 Jan 2021 14:42:42 +0000 (14:42 +0000)]
[X86] Add umin knownbits/demandedbits ult test for D94532
Jan Svoboda [Fri, 15 Jan 2021 13:47:55 +0000 (14:47 +0100)]
[clang][cli] Parse & generate options necessary for LangOptions defaults manually
It turns out we need to handle `LangOptions` separately from the rest of the options. `LangOptions` used to be conditionally parsed only when `!(DashX.getFormat() == InputKind::Precompiled || DashX.getLanguage() == Language::LLVM_IR)` and we need to restore this order (for more info, see D94682).
D94682 moves the parsing of marshalled `LangOpts` from `parseSimpleArgs` back to `ParseLangArgs`.
We need to parse marshalled `LangOpts` **after** `ParseLangArgs` calls `setLangDefaults`. This will enable future patches, where values of some `LangOpts` depend on the defaults.
However, two language options (`-finclude-default-header` and `-fdeclare-opencl-builtins`) need to be parsed **before** `ParseLangArgs` calls `setLangDefaults`, because they are necessary for setting up OpenCL defaults correctly.
This patch implements this by removing their marshalling info and manually parsing (and generating) them exactly where necessary.
Reviewed By: Bigcheese
Differential Revision: https://reviews.llvm.org/D94678
Anastasia Stulova [Fri, 15 Jan 2021 14:25:32 +0000 (14:25 +0000)]
[OpenCL][Docs] Fixed malformed table in OpenCLSupport
Tags: #clang
Stephan Herhut [Fri, 15 Jan 2021 14:03:08 +0000 (15:03 +0100)]
[SVE] Fix unused variable.
Introduced by [SVE] Restrict the usage of REINTERPRET_CAST.
Differential Revision: https://reviews.llvm.org/D94773
Lei Zhang [Fri, 15 Jan 2021 13:57:07 +0000 (08:57 -0500)]
[mlir][linalg] Support generating builders for named op attributes
This commit adds support to generate an additional builder for
each named op that has attributes. This gives better experience
when creating the named ops.
Along the way adds support for i64.
Reviewed By: hanchung
Differential Revision: https://reviews.llvm.org/D94733
Sam Tebbs [Fri, 15 Jan 2021 11:42:44 +0000 (11:42 +0000)]
[ARM] Don't run the block placement pass at O0
The block placement pass shouldn't run unless optimisations are enabled.
Differential Revision: https://reviews.llvm.org/D94691
Andy Wingo [Fri, 27 Nov 2020 08:19:46 +0000 (09:19 +0100)]
[WebAssembly] MC layer writes table symbols to object files
Now that the linker handles table symbols, we can allow the frontend to
produce them.
Depends on D91870.
Differential Revision: https://reviews.llvm.org/D92215
Simon Pilgrim [Fri, 15 Jan 2021 13:41:40 +0000 (13:41 +0000)]
[X86][AVX] combineHorizOpWithShuffle - support target shuffles in HOP(SHUFFLE(X,Y),SHUFFLE(X,Y)) -> SHUFFLE(HOP(X,Y))
Be more aggressive on (AVX2+) folds of lane shuffles of 256-bit horizontal ops by working on target/faux shuffles as well.
Raphael Isemann [Fri, 15 Jan 2021 13:43:26 +0000 (14:43 +0100)]
[lldb][docs] Translate ASCII art to restructured text formatting
This translates most of the old ASCII art in our documentation to the
equivalent in restructured text (which the new version of the LLDB docs
is using).
Adam Czachorowski [Thu, 14 Jan 2021 17:21:23 +0000 (18:21 +0100)]
[clangd] Set correct CWD when using compile_flags.txt
This fixes a bug where clangd would attempt to set CWD to the
compile_flags.txt file itself.
Differential Revision: https://reviews.llvm.org/D94699
Adam Czachorowski [Thu, 14 Jan 2021 21:41:10 +0000 (22:41 +0100)]
[clangd] Make ExpandAutoType not available on template params.
We cannot expand auto when used inside a template param (C++17 feature),
so do not offer it there.
Differential Revision: https://reviews.llvm.org/D94719
Raphael Isemann [Fri, 15 Jan 2021 13:07:45 +0000 (14:07 +0100)]
Revert "[lldb][docs] Use sphinx instead of epydoc to generate LLDB's Python reference"
This reverts commit
bab121a1b66e85390cad019ec921febcba35519d. It seems the
docs buildbot doesn't have the required Python headers.
Stefan Gränitz [Fri, 15 Jan 2021 12:32:02 +0000 (13:32 +0100)]
[Orc] Fix OrcV2Examples after D94690
Differential Revision: https://reviews.llvm.org/D94690
Raphael Isemann [Fri, 15 Jan 2021 12:24:24 +0000 (13:24 +0100)]
[lldb][docs] Use sphinx instead of epydoc to generate LLDB's Python reference
Currently LLDB uses epydoc to generate the Python API reference for the website.
epydoc however is unmaintained since more than a decade and no longer works with
Python 3. Also whatever setup we had once for generating the documentation on
the website server no longer seems to work, so the current website documentation
has been stale since more than a year.
This patch replaces epydoc with sphinx and its automodapi plugin that can
generate Python API references. LLVM already uses sphinx for the rest of the
documentation, so this way we are more consistent with the rest of LLVM. The
only new dependency is the automodapi plugin for sphinx.
This patch effectively does the following things:
* Remove the epydoc code.
* Make a new dummy Python API page in our website that just calls the Sphinx
command for generated the API documentation.
* Add a mock _lldb module that is only used when generating the Python API.
This way we don't have to build all of LLDB to generate the API reference.
Some notes:
* The long list of skips is necessary due to boilerplate functions that SWIG
is generating. Sadly automodapi is not really scriptable from what I can see,
so we have to blacklist this stuff manually.
* The .gitignore change because automodapi wants a subfolder of our
documentation directory to place generated documentation files there. The path
is also what is used on the website, so we can't really workaround this
(without copying the whole `docs` dir somewhere else when we build).
* We have to use environment variables to pass our build path to our sphinx
configuration. Sphinx doesn't support passing variables onto that script.
Reviewed By: JDevlieghere
Differential Revision: https://reviews.llvm.org/D94489
Hsiangkai Wang [Fri, 15 Jan 2021 03:23:18 +0000 (11:23 +0800)]
[NFC][RISCV] Remove useless code in RISCVRegisterInfo.td.
Differential Revision: https://reviews.llvm.org/D94750
Stefan Gränitz [Fri, 15 Jan 2021 11:26:04 +0000 (12:26 +0100)]
[Orc] Allow LLJITBuilder's CreateObjectLinkingLayer to return errors
It can be useful for an ObjectLinkingLayerCreator to allow callee errors to get propagated to the builder. Specifically, this is the case when the ObjectLayer uses the EHFrameRegistrationPlugin, because it requires a TPCEHFrameRegistrar and instantiation for it may fail (e.g. if the required registration symbols are missing in the target process).
Reviewed By: lhames
Differential Revision: https://reviews.llvm.org/D94690
Stefan Gränitz [Fri, 15 Jan 2021 11:25:57 +0000 (12:25 +0100)]
[Orc][NFC] Turn LLJIT member ObjTransformLayer into unique_ptr
All other layers in LLJIT are stored as unique_ptr's already. At this point, it is not strictly necessary for ObjTransformLayer, but it makes a follow-up change more straightforward.
Reviewed By: lhames
Differential Revision: https://reviews.llvm.org/D94689
Paul Walker [Wed, 13 Jan 2021 11:45:54 +0000 (11:45 +0000)]
[SVE] Restrict the usage of REINTERPRET_CAST.
In order to limit the number of combinations of REINTERPRET_CAST,
whilst at the same time prevent overlap with BITCAST, this patch
establishes the following rules:
1. The operand and result element types must be the same.
2. The operand and/or result type must be an unpacked type.
Differential Revision: https://reviews.llvm.org/D94593
Sam Elliott [Fri, 15 Jan 2021 11:20:28 +0000 (11:20 +0000)]
[RISCV] Optimize Branch Comparisons
I noticed in D94450 that there were quite a few places where we generate
the sequence:
```
xN <- comparison ...
xN <- xor xN, 1
bnez xN, symbol
```
Given we know the XOR will be used by BRCOND, which only looks at the lowest
bit, I think we can remove the XOR and just invert the branch condition in
these cases?
The case mostly seems to come up in floating point tests, where there is often
more logic to combine the results of multiple SETCCs, rather than a single
(BRCOND (SETCC ...) ...).
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D94535
Muhammad Omair Javaid [Fri, 15 Jan 2021 11:12:20 +0000 (16:12 +0500)]
DynamicRegisterInfo calculate offsets in separate function
This patch pull offset calculation logic out of DynamicRegisterInfo::Finalize
into a separate function. We are going to call this function whenever we
update SVE register sizes.
Reviewed By: labath
Differential Revision: https://reviews.llvm.org/D94008
Muhammad Omair Javaid [Fri, 15 Jan 2021 10:33:31 +0000 (15:33 +0500)]
[LLDB] Add per-thread register infos shared pointer in gdb-remote
In gdb-remote process we have register infos defind as a refernce object of
GDBRemoteDynamicRegisterInfo class. In past register infos have remained
constant througout the life time of a process.
This has changed after AArch64 SVE support where register infos will have
per-thread configuration. SVE registers will have per-thread size and can
be updated while running. This patch aims to build up for that support by
changing GDBRemoteDynamicRegisterInfo reference to a shared pointer deinfed
per-thread.
Reviewed By: labath
Differential Revision: https://reviews.llvm.org/D82857
Ilya Golovenko [Fri, 15 Jan 2021 07:47:28 +0000 (10:47 +0300)]
[clangd] exclude symbols from document outline which do not originate from the main file
Differential Revision: https://reviews.llvm.org/D94753
Georgii Rymar [Thu, 14 Jan 2021 09:35:01 +0000 (12:35 +0300)]
[llvm-readobj][test] - Remove excessive YAML fields from tests.
This removes excessive YAML keys from `SHT_GNU_verdef` sections.
Those keys are set by default.
Differential revision: https://reviews.llvm.org/D94660
Georgii Rymar [Wed, 13 Jan 2021 14:16:55 +0000 (17:16 +0300)]
[yaml2obj/obj2yaml] - Refine handling of SHT_GNU_verdef sections.
This patch:
1) Makes `Version`, `Flags`, `VersionNdx` and `Hash` fields to be `Optional<>`.
2) Disallows dumping version definitions that have `vd_version != 1`.
`vd_version` identifies the version of the structure itself.
(https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/symversion.html,
https://docs.oracle.com/cd/E19683-01/816-7777/chapter6-80869/index.html)
3) Stops dumping default values for `Version`, `Flags`, `VersionNdx` and `Hash` fields.
4) Refines testing.
Differential revision: https://reviews.llvm.org/D94659
Oliver Stannard [Wed, 6 Jan 2021 12:12:58 +0000 (12:12 +0000)]
[ARM][GISel] Treat calls as variadic even if only fixed arguments provided
For the ARM hard-float calling convention, calls to variadic functions
need to be treated diffrently, even if only the fixed arguments are
provided.
This fixes GCC-C-execute-pr68390 in the test-suite, which is failing on
the ARM GlobaISel bot.
Georgii Rymar [Thu, 14 Jan 2021 11:45:05 +0000 (14:45 +0300)]
[llvm-nm] - Simplify the code in dumpSymbolNamesFromObject. NFC.
It is possible to simplify the logic that extracts symbol names.
D94667 made the `NMSymbol::Name` to be `std::string`,
what allowed this simplification.
Differential revision: https://reviews.llvm.org/D94669
Guillaume Chatelet [Fri, 15 Jan 2021 09:26:45 +0000 (09:26 +0000)]
[libc] Allow customization of memcpy via flags.
- Adds LLVM_LIBC_IS_DEFINED macro to libc/src/__support/common.h
- Adds a few knobs to memcpy to help with experimentations:
- LLVM_LIBC_MEMCPY_X86_USE_ONLY_REPMOVSB replaces the implementation with a single call to rep;movsb
- LLVM_LIBC_MEMCPY_X86_USE_REPMOVSB_FROM_SIZE customizes where the usage of rep;movsb
Differential Revision: https://reviews.llvm.org/D94692
Georgii Rymar [Thu, 14 Jan 2021 10:45:44 +0000 (13:45 +0300)]
[llvm-nm] - Move MachO specific logic out from the dumpSymbolNamesFromObject(). NFC.
`dumpSymbolNamesFromObject` is the method that dumps symbol names.
It has 563 lines, mostly because of huge piece of MachO specific code.
In this patch I move it to separate helper method.
The new size of `dumpSymbolNamesFromObject` is 93 lines. With it it becomes
much easier to maintain it.
I had to change the type of 2 name fields to `std::string`, because MachO logic
uses temporarily buffer strings (e.g `ExportsNameBuffer`, `BindsNameBuffer` etc):
```
std::string ExportsNameBuffer;
raw_string_ostream EOS(ExportsNameBuffer);
```
these buffers were moved to `dumpSymbolsFromDLInfoMachO` by this patch and
invalidated after return. Technically, before this patch we had a situation
when local pointers (symbol names) were assigned to members of global static `SymbolList`,
what is dirty by itself.
Differential revision: https://reviews.llvm.org/D94667
Alok Kumar Sharma [Fri, 15 Jan 2021 08:10:33 +0000 (13:40 +0530)]
[Debuginfo][DW_OP_implicit_pointer] (1/7) Support for DW_OP_LLVM_implicit_pointer
New dwarf operator DW_OP_LLVM_implicit_pointer is introduced (present only in LLVM IR)
This operator is required as it is different than DWARF operator
DW_OP_implicit_pointer in representation and specification (number
and types of operands) and later can not be used as multiple level.
Reviewed By: aprantl
Differential Revision: https://reviews.llvm.org/D84113
Igor Kudrin [Thu, 14 Jan 2021 08:03:44 +0000 (15:03 +0700)]
[libcxx testing] Fix UB in tests for std::lock_guard
If mutex::try_lock() is called in a thread that already owns the mutex,
the behavior is undefined. The patch fixes the issue by creating another
thread, where the call is allowed.
Differential Revision: https://reviews.llvm.org/D94656
Amara Emerson [Fri, 15 Jan 2021 08:57:51 +0000 (00:57 -0800)]
[AArch64][GlobalISel] Fix fallbacks introduced for G_SITOFP in
8f283cafddfa8d6d01a94b48cdc5d25817569e91
If we have an integer->fp convert that has differing sizes, e.g. s32 to s64,
then don't try to convert it to AArch64::G_SITOF since it won't select.
Georgii Rymar [Fri, 15 Jan 2021 08:57:32 +0000 (11:57 +0300)]
[llvm-readobj] - Fix the compilation with GCC < 7.0.
This addressed post commit comments for D93900.
GCC had an issue and requires placing a specialization of
`printUnwindInfo` to a namespace to compile:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=56480
Qiu Chaofan [Fri, 15 Jan 2021 08:40:41 +0000 (16:40 +0800)]
[Clang] Mutate long-double math builtins into f128 under IEEE-quad
Under -mabi=ieeelongdouble on PowerPC, IEEE-quad floating point semantic
is used for long double. This patch mutates call to related builtins
into f128 version on PowerPC. And in theory, this should be applied to
other targets when their backend supports IEEE 128-bit style libcalls.
GCC already has these mutations except nansl, which is not available on
PowerPC along with other variants (nans, nansf).
Reviewed By: RKSimon, nemanjai
Differential Revision: https://reviews.llvm.org/D92080
Nikita Popov [Fri, 15 Jan 2021 07:59:17 +0000 (08:59 +0100)]
Revert "Reapply "ADT: Fix reference invalidation in SmallVector::push_back and single-element insert""
This reverts commit
260a856c2abcef49c7cb3bdcd999701db3e2af38.
This reverts commit
3043e5a5c33c4c871f4a1dfd621a8839f9a1f0b3.
This reverts commit
49142991a685bd427d7e877c29c77371dfb7634c.
This change had a larger than anticipated compile-time impact,
possibly because the small value optimization is not working as
intended. See D93779.
Andy Wingo [Thu, 14 Jan 2021 09:15:56 +0000 (10:15 +0100)]
[WebAssembly] Add support for table linking to wasm-ld
This patch adds support to wasm-ld for linking multiple table references
together, in a manner similar to wasm globals. The indirect function
table is synthesized as needed.
To manage the transitional period in which the compiler doesn't yet
produce TABLE_NUMBER relocations and doesn't residualize table symbols,
the linker will detect object files which have table imports or
definitions, but no table symbols. In that case it will synthesize
symbols for the defined and imported tables.
As a change, relocatable objects are now written with table symbols,
which can cause symbol renumbering in some of the tests. If no object
file requires an indirect function table, none will be written to the
file. Note that for legacy ObjFile inputs, this test is conservative: as
we don't have relocs for each use of the indirecy function table, we
just assume that any incoming indirect function table should be
propagated to the output.
Differential Revision: https://reviews.llvm.org/D91870
KAWASHIMA Takahiro [Mon, 4 Jan 2021 02:25:42 +0000 (11:25 +0900)]
[AArch64] Add Fujitsu A64FX scheduling model
Basic support of A64FX was added in D75594 but its scheduling model
was missing. This commit adds the scheduling model. Also, this commit
amends/adds some subtarget parameters of A64FX.
The A64FX Microarchitecture Manual, which is source information of
this commit, is on GitHub.
https://github.com/fujitsu/A64FX/
Differential Revision: https://reviews.llvm.org/D93791
Jan Svoboda [Wed, 13 Jan 2021 11:50:19 +0000 (12:50 +0100)]
[clang][cli] Specify KeyPath prefixes via TableGen classes
It turns out we need to handle `LangOptions` separately from the rest of the options. `LangOptions` used to be conditionally parsed only when `!(DashX.getFormat() == InputKind::Precompiled || DashX.getLanguage() == Language::LLVM_IR)` and we need to restore this order (for more info, see D94682).
We could do this similarly to how `DiagnosticOptions` are handled: via a counterpart to the `IsDiag` mix-in (e.g. `IsLang`). These mix-ins would prefix the option key path with the appropriate `CompilerInvocation::XxxOpts` member. However, this solution would be problematic, as we'd now have two kinds of options (`Lang` and `Diag`) with seemingly incomplete key paths in the same file. To understand what `CompilerInvocation` member an option affects, one would need to read the whole option definition and notice the `IsDiag` or `IsLang` class.
Instead, this patch introduces more robust way to handle different kinds of options separately: via the `KeyPathAndMacroPrefix` class. We have one specialization of that class per `CompilerInvocation` member (e.g. `LangOpts`, `DiagnosticOpts`, etc.). Now, instead of specifying a key path with `"LangOpts->UndefPrefixes"`, we use `LangOpts<"UndefPrefixes">`. This keeps the readability intact (you don't have to look for the `IsLang` mix-in, the key path is complete on its own) and allows us to specify a custom macro prefix within `LangOpts`.
Reviewed By: Bigcheese
Differential Revision: https://reviews.llvm.org/D94676
Jan Svoboda [Thu, 14 Jan 2021 07:34:17 +0000 (08:34 +0100)]
[clang][cli] NFC: Decrease the scope of ParseCodeGenArgs parameters
Instead of passing the whole `TargetOptions` and `FrontendOptions` to `ParseCodeGenArgs` give it only the necessary members.
This makes tracking the dependencies between various parsers and option groups easier.
Reviewed By: Bigcheese
Differential Revision: https://reviews.llvm.org/D94675
Jan Svoboda [Thu, 14 Jan 2021 07:26:12 +0000 (08:26 +0100)]
[clang][cli] NFC: Decrease the scope of ParseLangArgs parameters
Instead of passing the whole `TargetOptions` and `PreprocessorOptions` to `ParseLangArgs` give it only the necessary members.
This makes tracking the dependencies between various parsers and option groups easier.
Reviewed By: Bigcheese
Differential Revision: https://reviews.llvm.org/D94674
Aart Bik [Thu, 14 Jan 2021 20:04:49 +0000 (12:04 -0800)]
[mlir][sparse] retry sparse-only for cyclic iteration graphs
This is a very minor improvement during iteration graph construction.
If the first attempt considering the dimension order of all tensors fails,
a second attempt is made using the constraints of sparse tensors only.
Dense tensors prefer dimension order (locality) but provide random access
if needed, enabling the compilation of more sparse kernels.
Reviewed By: penpornk
Differential Revision: https://reviews.llvm.org/D94709
Yashaswini [Fri, 15 Jan 2021 05:42:49 +0000 (11:12 +0530)]
Add Semantic check for Flang OpenMP 4.5 - 2.7.1 Do Loop restrictions on single directive and firstprivate clause.
Semantic checks added to check the worksharing 'single' region closely nested inside a worksharing 'do' region. And also to check whether the 'do' iteration variable is a variable in 'Firstprivate' clause.
Files:
check-directive-structure.h
check-omp-structure.h
check-omp-structure.cpp
Testcases:
omp-do01-positivecase.f90
omp-do01.f90
omp-do05-positivecase.f90
omp-do05.f90
Reviewed by: Kiran Chandramohan @kiranchandramohan , Valentin Clement @clementval
Differential Revision: https://reviews.llvm.org/D93205
Kazu Hirata [Fri, 15 Jan 2021 04:30:34 +0000 (20:30 -0800)]
[llvm] Remove redundant return and continue statements (NFC)
Identified with readability-redundant-control-flow.
Kazu Hirata [Fri, 15 Jan 2021 04:30:32 +0000 (20:30 -0800)]
[llvm] Use llvm::drop_begin (NFC)
Kazu Hirata [Fri, 15 Jan 2021 04:30:31 +0000 (20:30 -0800)]
[CodeGen, Transforms] Use llvm::sort (NFC)
Cheng Wang [Wed, 25 Nov 2020 13:12:48 +0000 (21:12 +0800)]
[libc] Add memmove implementation.
Use `memcpy` rather than copying bytes one by one, for there might be large
size structs to move.
Reviewed By: gchatelet, sivachandra
Differential Revision: https://reviews.llvm.org/D93195
Amara Emerson [Thu, 14 Jan 2021 18:56:04 +0000 (10:56 -0800)]
[AArch64][GlobalISel] Add selection support for fpr bank source variants of G_SITOFP and G_UITOFP.
In order to import patterns for these, we need to define new ops that can map to
the AArch64ISD::[SU]ITOF nodes. We then transform fpr->fpr variants of the generic
opcodes to these custom opcodes in preisel-lowering. We have to do it here and
not the PostLegalizer combiner because this has to run after regbankselect.
Differential Revision: https://reviews.llvm.org/D94702
Yitzhak Mandelbaum [Tue, 22 Dec 2020 15:02:56 +0000 (15:02 +0000)]
[libTooling] Change `addInclude` to use expansion locs.
This patch changes the default range used to anchor the include insertion to use
an expansion loc. This ensures that the location is valid, when the user relies
on the default range.
Driveby: extend a FIXME for a problem that was emphasized by this change; fix some spellings.
Differential Revision: https://reviews.llvm.org/D93703
Jon Chesterfield [Fri, 15 Jan 2021 02:16:32 +0000 (02:16 +0000)]
[libomptarget][nvptx] Reduce calls to cuda header
[libomptarget][nvptx] Reduce calls to cuda header
Remove use of clock_t in favour of a builtin. Drop a preprocessor branch.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D94731
Jian Cai [Fri, 15 Jan 2021 01:33:20 +0000 (17:33 -0800)]
Reland "[AsmParser] make .ascii support spaces as separators"
This relands commit
e0963ae274be5b071d1e1b00f5e4e019483c09e9, which was
reverted on commit
82c4153e66fa284729da86a8d6c302d4b8cec86c due to a
test failure, which turned out to be a false positive.
Amara Emerson [Thu, 14 Jan 2021 18:53:56 +0000 (10:53 -0800)]
[AArch64][GlobalISel] Assign FPR banks to loads which are used by integer->float conversions.
G_[US]ITOFP users of loads on AArch64 can operate on both gpr and fpr banks for scalars.
Because of this, if their source is a load, then that load can be assigned to an fpr
bank and therefore avoid having to do a cross bank copy via a gpr->fpr conversion.
Differential Revision: https://reviews.llvm.org/D94701
Tim Keith [Fri, 15 Jan 2021 00:31:48 +0000 (16:31 -0800)]
[flang] Fix use-associated procedure in generic
When a use-associated procedure was included in a generic, we weren't
correctly recording that fact. The ultimate symbol was added rather than
the local symbol.
Also, improve the message emitted for the specific procedure by
mentioning the module it came from.
This fixes one of the problems in https://bugs.llvm.org/show_bug.cgi?id=48648.
Differential Revision: https://reviews.llvm.org/D94696
Jon Chesterfield [Fri, 15 Jan 2021 00:19:48 +0000 (00:19 +0000)]
[libomptarget][nvptx][nfc] Move target_impl functions out of header
[libomptarget][nvptx][nfc] Move target_impl functions out of header
This removes most of the differences between the two target_impl.h.
Also change name mangling from C to C++ for __kmpc_impl_*_lock.
Reviewed By: tianshilei1992
Differential Revision: https://reviews.llvm.org/D94728
MaheshRavishankar [Thu, 14 Jan 2021 23:41:12 +0000 (15:41 -0800)]
[mlir][Linalg] NFC: Verify tiling on linalg.generic operation on tensors.
With the recent changes to linalg on tensor semantics, the tiling
operations works out-of-the-box for generic operations. Add a test to
verify that and some minor refactoring.
Differential Revision: https://reviews.llvm.org/D93077
MaheshRavishankar [Thu, 14 Jan 2021 23:41:04 +0000 (15:41 -0800)]
[mlir][Linalg] Add canonicalization of linalg op -> dim op.
Add canonicalization to replace use of the result of a linalg
operation on tensors in a dim operation, to use one of the operands of
the linalg operations instead. This allows the linalg op itself to be
deleted when all its non-dim uses are removed (say through tiling, etc.)
Differential Revision: https://reviews.llvm.org/D93076
Shilei Tian [Fri, 15 Jan 2021 00:14:19 +0000 (19:14 -0500)]
[OpenMP] Remove omptarget-nvptx from deps as it is no longer a valid target
`omptarget-nvptx` is still a dependence for `check-libomptarget-nvtpx`
although it has been removed by D94573.
Reviewed By: JonChesterfield
Differential Revision: https://reviews.llvm.org/D94725
Sanjay Patel [Thu, 14 Jan 2021 22:26:48 +0000 (17:26 -0500)]
[SLP] remove unnecessary state in matching reductions
This is NFC-intended. I'm still trying to figure out
how the loop where this is used works. It does not
seem like we require this data at all, but it's
hard to confirm given the complicated predicates.
MaheshRavishankar [Thu, 14 Jan 2021 22:58:04 +0000 (14:58 -0800)]
[mlir][Linalg] Add canonicalization to remove no-op linalg operations.
linalg.generic/indexed_generic operations on tensors whose body is
just yielding the (non-induction variable) arguments of the operation
can be canonicalized by replacing uses of the result with the
corresponding arguments.
Differential Revision: https://reviews.llvm.org/D94581
Sam McCall [Thu, 14 Jan 2021 22:54:51 +0000 (23:54 +0100)]
[clangd] Reduce logspam for CDB scanning
Andrew Young [Thu, 14 Jan 2021 10:07:29 +0000 (02:07 -0800)]
[mlir] Remove over specified memory effects
The standard and gpu dialect both have `alloc` operations which use the
memory effect `MemAlloc`. In both cases, it is specified on both the
operation itself and on the result. This results in two memory effects
being created for these operations. When `MemAlloc` is defined on an
operation, it represents some background effect which the compiler
cannot reason about, and inhibits the ability of the compiler to
remove dead `std.alloc` operations. This change removes the uneeded
`MemAlloc` effect from these operations and leaves the effect on the
result, which allows dead allocs to be erased.
There is the same problem, but to a lesser extent, with MemFree, MemRead
and MemWrite. Over-specifying these traits is not currently inhibiting
any optimization.
Differential Revision: https://reviews.llvm.org/D94662
Sam Elliott [Thu, 14 Jan 2021 22:32:17 +0000 (22:32 +0000)]
[RISCV][NFC] Regenerate Calling Convention Tests
This regenerates these tests using utils/update_llc_test_checks.py so
that future changes in this area don't have the noise of lots of `@plt`
lines being added.
I also removed the `nounwind`s from the stack-realignment.ll test to
increase coverage on the generated call frame information.
Teresa Johnson [Thu, 14 Jan 2021 22:07:11 +0000 (14:07 -0800)]
[LTO] Test format fix (NFC)
As requested in D91583, use ';;' instead of ';' to preceed comments in
lld test. I did this in the equivalent gold test as well.
Alexandre Ganea [Thu, 14 Jan 2021 22:03:13 +0000 (17:03 -0500)]
Re-land [Support] On Windows, take the affinity mask into account
The number of hardware threads available to a ThreadPool can be limited if setting an affinity mask.
For example:
> start /B /AFFINITY 0xF lld-link.exe ...
Would let LLD only use 4 hyper-threads.
Previously, there was an outstanding issue on Windows Server 2019 on dual-CPU machines, which was preventing from using both CPU sockets. In normal conditions, when no affinity mask was set, ProcessorGroup::AllThreads was different from ProcessorGroup::UsableThreads. The previous code in llvm/lib/Support/Windows/Threading.inc L201 was improperly assuming those two values to be equal, and consequently was limiting the execution to only one CPU socket.
Differential Revision: https://reviews.llvm.org/D92419
Craig Topper [Thu, 14 Jan 2021 21:41:35 +0000 (13:41 -0800)]
[RISCV] Optimize select_cc after fp compare expansion
Some FP compares expand to a sequence ending with (xor X, 1) to invert the result. If
the consumer is a select_cc we can likely get rid of this xor by fixing
up the select_cc condition.
This patch combines (select_cc (xor X, 1), 0, setne, trueV, falseV) -
(select_cc X, 0, seteq, trueV, falseV) if we can prove X is 0/1.
Reviewed By: lenary
Differential Revision: https://reviews.llvm.org/D94546
Douglas Yung [Thu, 14 Jan 2021 21:26:21 +0000 (13:26 -0800)]
Add -fexceptions to test as it uses them and fails on platforms where it is not on by default (like the PS4).
Roland McGrath [Thu, 14 Jan 2021 01:22:45 +0000 (17:22 -0800)]
[libc] Use #undef isascii in specific header
Standard C allows all standard headers to declare macros for all
their functions. So after possibly including any standard header
like <ctype.h>, it's perfectly normal for any and all of the
functions it declares to be defined as macros. Standard C requires
explicit `#undef` before using that identifier in a way that is not
compatible with function-like macro definitions.
The C standard's rules for this are extended to POSIX as well for
the interfaces it defines, and it's the expected norm for
nonstandard extensions declared by standard C library headers too.
So far the only place this has come up for llvm-libc's code is with
the isascii function in Fuchsia's libc. But other cases can arise
for any standard (or common extension) function names that source
code in llvm-libc is using in nonstandard ways, i.e. as C++
identifiers.
The only correct and robust way to handle the possible inclusion of
standard C library headers when building llvm-libc source code is to
use `#undef` explicitly for each identifier before using it. The
easy and obvious place to do that is in the per-function header.
This requires that all code, such as test code, that might include
any standard C library headers, e.g. via utils/UnitTest/Test.h, make
sure to include those *first* before the per-function header.
This change does that for isascii and its test. But it should be
done uniformly for all the code and documented as a consistent
convention so new implementation files are sure to get this right.
Reviewed By: sivachandra
Differential Revision: https://reviews.llvm.org/D94642
Nico Weber [Thu, 14 Jan 2021 21:19:25 +0000 (16:19 -0500)]
[gn build] (manually) port
387d3c24792f
Jinsong Ji [Thu, 14 Jan 2021 20:35:56 +0000 (20:35 +0000)]
[PowerPC] Only use some extend mne if assembler is modern enough
Legacy AIX assembly might not support all extended mnes,
add one feature bit to control the generation in MC,
and avoid generating them by default on AIX.
Reviewed By: sfertile
Differential Revision: https://reviews.llvm.org/D94458
Sean Silva [Sat, 9 Jan 2021 00:25:47 +0000 (16:25 -0800)]
[mlir][docs] Bring bufferization docs up to date.
This spilts out BufferDeallocationInternals.md, since buffer
deallocation is not part of bufferization per se.
Differential Revision: https://reviews.llvm.org/D94351
Adam Czachorowski [Thu, 14 Jan 2021 19:54:16 +0000 (20:54 +0100)]
[clang] Do not crash when CXXRecordDecl has a non-CXXRecordDecl base.
This can happen on some invalid code, like the included test case.
Differential Revision: https://reviews.llvm.org/D94704
Tres Popp [Thu, 14 Jan 2021 20:04:37 +0000 (21:04 +0100)]
[mlir] Remove TosaToLinalg dependency on all Passes
TosaToLinalg was depending on its header file indirectly through
Passes.h rather than directly. This removes that indirection.
Differential Revision: https://reviews.llvm.org/D94706
River Riddle [Thu, 14 Jan 2021 19:57:17 +0000 (11:57 -0800)]
[mlir][PatternRewriter] Add a new hook to selectively replace uses of an operation
This revision adds a new `replaceOpWithIf` hook that replaces uses of an operation that satisfy a given functor. If all uses are replaced, the operation gets erased in a similar manner to `replaceOp`. DialectConversion support will be added in a followup as this requires adjusting how replacements are tracked there.
Differential Revision: https://reviews.llvm.org/D94632
Craig Topper [Thu, 14 Jan 2021 19:44:02 +0000 (11:44 -0800)]
[RISCV] Merge Utils library into MCTargetDesc
MCTargetDesc includes headers from Utils and Utils includes headers
from MCTargetDesc. So from a library layering perspective it makes sense
for them to be in the same library. I guess the other option might be to
move the tablegen includes from RISCVMCTargetDesc.h to RISCVBaseInfo.h
so that RISCVBaseInfo.h didn't need to include RISCVMCTargetDesc.h.
Everything else that depends on Utils also depends on MCTargetDesc so
having one library seemed simpler.
Differential Revision: https://reviews.llvm.org/D93168
Fangrui Song [Thu, 14 Jan 2021 19:46:22 +0000 (11:46 -0800)]
[Driver] -gsplit-dwarf: Produce .dwo regardless of -gN for IR input
This generalizes D94647 to IR input, as suggested by @tejohnson.
Ideally the driver should just forward split dwarf options, but doing this currently will cause `clang -gsplit-dwarf -c a.c` to create a .dwo with just `.strtab`.
Reviewed By: dblaikie, tejohnson
Differential Revision: https://reviews.llvm.org/D94655
River Riddle [Thu, 14 Jan 2021 19:35:15 +0000 (11:35 -0800)]
[mlir][OpFormatGen] Format enum attribute cases as keywords when possible
In the overwhelmingly common case, enum attribute case strings represent valid identifiers in MLIR syntax. This revision updates the format generator to format as a keyword in these cases, removing the need to wrap values in a string. The parser still retains the ability to parse the string form, but the printer will use the keyword form when applicable.
Differential Revision: https://reviews.llvm.org/D94575
River Riddle [Thu, 14 Jan 2021 19:34:55 +0000 (11:34 -0800)]
[mlir][ODS] Add new RangedTypesMatchWith operation predicate
This is a variant of TypesMatchWith that provides support for variadic arguments. This is necessary because ranges generally can't use the default operator== comparators for checking equality.
Differential Revision: https://reviews.llvm.org/D94574
Nikita Popov [Tue, 10 Nov 2020 19:43:46 +0000 (20:43 +0100)]
[BasicAA] Handle recursive queries more efficiently
An alias query currently works out roughly like this:
* Look up location pair in cache.
* Perform BasicAA logic (including cache lookup and insertion...)
* Perform a recursive query using BestAAResults.
* Look up location pair in cache (and thus do not recurse into BasicAA)
* Query all the other AA providers.
* Query all the other AA providers.
This is a lot of unnecessary work, all ultimately caused by the
BestAAResults query at the end of aliasCheck(). The reason we perform
it, is that aliasCheck() is getting called recursively, and we of
course want those recursive queries to also make use of other AA
providers, not just BasicAA. We can solve this by making the recursive
queries directly use BestAAResults (which will check both BasicAA
and other providers), rather than recursing into aliasCheck().
There are some tradeoffs:
* We can no longer pass through the precomputed underlying object
to aliasCheck(). This is not a major concern, because nowadays
getUnderlyingObject() is quite cheap.
* Results from other AA providers are no longer cached inside
BasicAA. The way this worked was already a bit iffy, in that a
result could be cached, but if it was MayAlias, we'd still end
up re-querying other providers anyway. If we want to cache
non-BasicAA results, we should do that in a more principled manner.
In any case, despite those tradeoffs, this works out to be a decent
compile-time improvment. I think it also simplifies the mental model
of how BasicAA works. It took me quite a while to fully understand
how these things interact.
Differential Revision: https://reviews.llvm.org/D90094
Mehdi Amini [Thu, 14 Jan 2021 19:28:49 +0000 (19:28 +0000)]
Add newline to terminate debug message (NFC)
Rob Suderman [Tue, 5 Jan 2021 22:26:48 +0000 (14:26 -0800)]
[MLIR][TOSA] First lowerings from Tosa to Linalg
Initial commit to add support for lowering from TOSA to Linalg. The focus is on
the essential infrastructure for these lowerings and integration with existing
passes.
Includes lowerings for a subset of operations including:
abs, add, sub, pow, and, or, xor, left shift, right shift, tanh
Lit tests are used to validate correctness.
Differential Revision: https://reviews.llvm.org/D94247
Erich Keane [Thu, 14 Jan 2021 19:00:49 +0000 (11:00 -0800)]
[NFC] Update test to not check for 'opaque' in the file name.
The intent presumably is to avoid generating 'opaque' in the IR, but the
header contains the filename. Thus, having the workspace in a directory
with opaque in it causes this test to fail.
This just adds a 'CHECK' line on target-triple, which is the last line
of the IR-header.
Valentin Clement [Thu, 14 Jan 2021 19:19:35 +0000 (14:19 -0500)]
[openacc] Rename generated file from ACC.cpp.inc to ACC.inc to match D92955
This patch rename the tablegen generated file ACC.cpp.inc to ACC.inc in order
to match what was done in D92955. This file is included in header file as well as .cpp
file so it make more sense.
Reviewed By: sameeranjoshi
Differential Revision: https://reviews.llvm.org/D93485
Mitch Phillips [Thu, 14 Jan 2021 01:28:10 +0000 (17:28 -0800)]
[GWP-ASan] Minor refactor of optional components.
In preparation for the inbuilt options parser, this is a minor refactor
of optional components including:
- Putting certain optional elements in the right header files,
according to their function and their dependencies.
- Cleaning up some old and mostly-dead code.
- Moving some functions into anonymous namespaces to prevent symbol
export.
Reviewed By: cryptoad, eugenis
Differential Revision: https://reviews.llvm.org/D94117
Shilei Tian [Thu, 14 Jan 2021 18:55:05 +0000 (13:55 -0500)]
[OpenMP] Dropped unnecessary define when compiling deviceRTLs for NVPTX
The comment said CUDA 9 header files use the `nv_weak` attribute which
`clang` is not yet prepared to handle. It's three years ago and now things have
changed. Based on my test, removing the definition doesn't have any problem on
my machine with CUDA 11.1 installed.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D94700
Hiroshi Yamauchi [Mon, 11 Jan 2021 19:02:37 +0000 (11:02 -0800)]
[X86] Add the FSRM feature (Fast Short Rep Mov) to Zen3.
Note -x86-use-fsrm-for-memcpy is still disabled by default and there's no
default behavior change.
Differential Revision: https://reviews.llvm.org/D94436
Zequan Wu [Thu, 14 Jan 2021 03:14:25 +0000 (19:14 -0800)]
[clang][MSVC] Fix missing MSInheritanceAttr in template specialization.
Fix PR48687.
Differential Revision: https://reviews.llvm.org/D94646
Shilei Tian [Thu, 14 Jan 2021 18:34:18 +0000 (13:34 -0500)]
[OpenMP] Drop the static library libomptarget-nvptx
For NVPTX target, OpenMP provides a static library `libomptarget-nvptx`
built by NVCC, and another bitcode `libomptarget-nvptx-sm_{$sm}.bc` generated by
Clang. When compiling an OpenMP program, the `.bc` file will be fed to `clang`
in the second run on the program that compiles the target part. Then the generated
PTX file will be fed to `ptxas` to generate the object file, and finally the driver
invokes `nvlink` to generate the binary, where the static library will be appened
to `nvlink`.
One question is, why do we need two libraries? The only difference is, the static
library contains `omp_data.cu` and the bitcode library doesn't. It's unclear why
they were implemented in this way, but per D94565, there is no issue if we also
include the file into the bitcode library. Therefore, we can safely drop the
static library.
This patch is about the change in OpenMP. The driver will be updated as well if
this patch is accepted.
Reviewed By: jdoerfert, JonChesterfield
Differential Revision: https://reviews.llvm.org/D94573