Max Kazantsev [Thu, 26 Nov 2020 10:30:44 +0000 (17:30 +0700)]
Return "[SCEV] Use isBasicBlockEntryGuardedByCond in isLoopBackedgeGuardedByCond", 2nd try
Reverted because the compile time impact is still too high.
isKnownViaNonRecursiveReasoning is used twice, we can do it just once.
Differential Revision: https://reviews.llvm.org/D92152
Max Kazantsev [Thu, 26 Nov 2020 10:27:42 +0000 (17:27 +0700)]
Revert "[SCEV] Use isBasicBlockEntryGuardedByCond in isLoopBackedgeGuardedByCond"
This reverts commit
3d4c0460ec6040fc071e56dc113afd181294591e.
Compile time impact is still high. Need to understand why.
Differential Revision: https://reviews.llvm.org/D92153
Georgii Rymar [Wed, 25 Nov 2020 10:58:40 +0000 (13:58 +0300)]
[llvm-readelf/obj] - Report a warning when the value of the DT_PLTREL dynamic tag is invalid.
We report an error for unknown `DT_PLTREL` values.
This switches the error to warning.
Differential revision: https://reviews.llvm.org/D92087
Max Kazantsev [Thu, 26 Nov 2020 09:59:39 +0000 (16:59 +0700)]
[SCEV] Use isBasicBlockEntryGuardedByCond in isLoopBackedgeGuardedByCond
Previously we tried to using isKnownPredicateAt, but it makes an
extra query to isKnownPredicate, which has negative impact on compile
time. Let's try to use more lightweight isBasicBlockEntryGuardedByCond.
Differential Revision: https://reviews.llvm.org/D92152
Craig Topper [Thu, 26 Nov 2020 10:02:38 +0000 (02:02 -0800)]
[RISCV] Add isel patterns for sbsetw/sbclrw/sbinvw with sext_inreg as the root.
This handles cases were the input isn't known to be sign extended.
Craig Topper [Thu, 26 Nov 2020 09:54:07 +0000 (01:54 -0800)]
[RISCV] Add test cases for missed opportunities to use sbsetw/sbclrw/sbinvw when the result isn't known to be sign extended.
If the input isn't sign extended, but the output of the or/xor/and
is used by a sign_inreg we can still use sbsetw/sbclrw/sbinvw.
Marek Kurdej [Thu, 26 Nov 2020 09:40:52 +0000 (10:40 +0100)]
[llvm-profgen] [docs] Fix invalid header. Add to ToC. NFC.
Max Kazantsev [Thu, 26 Nov 2020 09:42:43 +0000 (16:42 +0700)]
Revert "[SCEV] Use isKnownPredicateAt in isLoopBackedgeGuardedByCond"
This reverts commit
14f2ad0e3cc54d5eb254b545a469e8ffdb62b119.
Reverting to investigate compile time drop.
Differential Revision: https://reviews.llvm.org/D92152
Mark de Wever [Thu, 26 Nov 2020 09:37:04 +0000 (10:37 +0100)]
[NFC][libc++] Mark LWG3296 as complete.
I recalled Marshall had already made this change. The change is
committed in
e3f89a989a23b9bfcb9f9d01172cebb63db627e1.
Marek Kurdej [Thu, 26 Nov 2020 09:20:09 +0000 (10:20 +0100)]
[libc++] Fix gcc warning -Wsign-compare.
Marek Kurdej [Thu, 26 Nov 2020 09:07:16 +0000 (10:07 +0100)]
[libc++] [P0966] [C++20] Fix bug PR45368 by correctly implementing P0966: string::reserve should not shrink.
This patch fixes the implementation as well as the tests that didn't actually test the wanted behaviour.
You'll find all the details in the bug report.
It adds as well deprecation warning for reserve() (without argument) and adds a test.
http://wg21.link/P0966R1
https://bugs.llvm.org/show_bug.cgi?id=45368
https://reviews.llvm.org/D54992
Reviewed By: ldionne, #libc
Differential Revision: https://reviews.llvm.org/D91778
Jay Foad [Wed, 25 Nov 2020 11:51:23 +0000 (11:51 +0000)]
[AMDGPU] Introduce and use isGFX10Plus. NFC.
It's more future-proof to use isGFX10Plus from the start, on the
assumption that future architectures will be based on current
architectures.
Also make use of the existing isGFX9Plus in a few places.
Differential Revision: https://reviews.llvm.org/D92092
Marek Kurdej [Thu, 26 Nov 2020 08:03:50 +0000 (09:03 +0100)]
[libc++] [docs] Migrate C++ status pages to RestructuredText (RST).
Currently, papers and issues are in separate .csv files (that is easier to update), but I can put them inline.Transforming current html tables into rst are done by the script (attached to the patch FYI but I'll remove it before committing).
I'll of course update RST files before committing to match any modifications that may happen in master branch.
This patch moves the status pages in www/ to RST format in docs/.
It also does some other minor changes: fix copyright year and broken comment end, adds substitutions for coherence (and add colors, but that can be removed easily).
It adds as well redirects from old to new status pages.
Reviewed By: ldionne, #libc
Differential Revision: https://reviews.llvm.org/D92076
Raphael Isemann [Thu, 26 Nov 2020 08:40:02 +0000 (09:40 +0100)]
[lldb] Fix TestThreadStepOut.py after "Flush local value map on every instruction"
After
cf1c774d6ace59c5adc9ab71b31e762c1be695b1, Clang seems to generate code
that is more similar to icc/Clang, so we can use the same line numbers for
all compilers in this test.
Nathan Ridge [Thu, 26 Nov 2020 02:31:09 +0000 (21:31 -0500)]
[clangd] Do not treat line as inactive if skipped range ends at character position 0
Fixes https://github.com/clangd/clangd/issues/602
Differential Revision: https://reviews.llvm.org/D92148
Zhengyang Liu [Thu, 26 Nov 2020 07:10:36 +0000 (00:10 -0700)]
Fix use-of-uninitialized-value in rG75f50e15bf8f
Differential Revision: https://reviews.llvm.org/D71126
Marek Kurdej [Thu, 26 Nov 2020 08:31:20 +0000 (09:31 +0100)]
[libc++] [www] Fix HTML. NFC.
Needed for a future automatic update to RST.
Martin Storsjö [Thu, 26 Nov 2020 08:09:32 +0000 (10:09 +0200)]
[OpenMP][OMPT] Fix building with OMPT disabled after
6d3b81664a4b79
Marek Kurdej [Thu, 26 Nov 2020 07:58:44 +0000 (08:58 +0100)]
[libc++] [libc++abi] Mark a few tests as unsupported/xfail on gcc-7/8/9.
This should make the builder http://lab.llvm.org:8011/#/builders/101/ happy.
It uses gcc-9 and not Tip-Of-Trunk as its name indicates BTW.
GCC-10 passes all these tests.
Fix gcc warnings: -Wsign-compare, -Wparentheses, -Wpragmas.
Reviewed By: ldionne, #libc, #libc_abi
Differential Revision: https://reviews.llvm.org/D92099
Kazu Hirata [Thu, 26 Nov 2020 07:52:52 +0000 (23:52 -0800)]
[Support] Use llvm::is_contained (NFC)
Georgii Rymar [Thu, 26 Nov 2020 07:22:29 +0000 (10:22 +0300)]
[llvm-readobj] - Fix a warning.
This addresses post review comment for D92018.
The warning was:
```
error: loop variable 'Note' is always a copy because the range of type 'iterator_range<llvm::object::ELFFile<llvm::object::ELFType<llvm::support::big, true> >::Elf_Note_Iterator>' (aka 'iterator_range<Elf_Note_Iterator_Impl<ELFType<(llvm::support::endianness)0U, true> > >') does not return a reference [-Werror,-Wrange-loop-analysis]
for (const typename ELFT::Note &Note : Obj.notes(S, Err))
```
Max Kazantsev [Thu, 26 Nov 2020 06:20:02 +0000 (13:20 +0700)]
[SCEV] Use isKnownPredicateAt in isLoopBackedgeGuardedByCond
A piece of code in `isLoopBackedgeGuardedByCond` basically duplicates
the dominators traversal from `isBlockEntryGuardedByCond` called from
`isKnownPredicateAt`, but it's less powerful because it does not give context
to `isImpliedCond`. This patch reuses the `isKnownPredicateAt `function there,
reducing the amount of code duplication and making it more powerful.
Differential Revision: https://reviews.llvm.org/D92152
Reviewed By: skatkov
Craig Topper [Thu, 26 Nov 2020 05:31:41 +0000 (21:31 -0800)]
[RISCV] Add isel pattern to match (i64 (sra (shl X, 32), C)) to SRAIW if C > 32.
Max Kazantsev [Thu, 26 Nov 2020 05:34:55 +0000 (12:34 +0700)]
[IndVars] Use isLoopBackedgeGuardedByCond for last iteration check
Use more context to prove contextual facts about the last iteration. It is
only executed when the backedge is taken, so we can use `isLoopBackedgeGuardedByCond`
to make this check.
Differential Revision: https://reviews.llvm.org/D91535
Reviewed By: skatkov
Craig Topper [Thu, 26 Nov 2020 04:35:23 +0000 (20:35 -0800)]
[RISCV] Remove unused PatFrag argument from the tablegen class used for c.beqz/c.bnez. NFC
Craig Topper [Thu, 26 Nov 2020 03:12:46 +0000 (19:12 -0800)]
[LegalizerTypes] Add support for scalarizing the operand of an FP_EXTEND when the result type is legal.
Max Kazantsev [Thu, 26 Nov 2020 03:49:44 +0000 (10:49 +0700)]
[LoopLoadElim] Make sure all loops are in simplify form. PR48150
LoopLoadElim may end up expanding an AddRec from a loop
which is not the current loop. This loop may not be in simplify
form. We figure it out after the no-return point, so cannot bail
in this case.
AddRec requires simplify form to expand. The only way to ensure
this does not crash is to simplify all loops beforehand.
The issue only exists in new PM. Old PM requests LoopSimplify
required pass and it simplifies all loops before the opt begins.
Differential Revision: https://reviews.llvm.org/D91525
Reviewed By: asbirlea, aeubanks
Stella Laurenzo [Thu, 26 Nov 2020 01:45:58 +0000 (17:45 -0800)]
[mlir][Python] Add additional capsule interop types.
Differential Revision: https://reviews.llvm.org/D92144
Sam Clegg [Mon, 23 Nov 2020 23:41:07 +0000 (15:41 -0800)]
[lld][WebAssembly] Ensure stub symbols always get address 0
Without this extra flag we can't distingish between stub functions and
functions that happen to have address 0 (relative to __table_base).
Adding this flag bit the base symbol class actually avoids growing the
SymbolUnion struct which would not be true if we added it to the
FunctionSymbol subclass (due to bitbacking).
The previous approach of setting it's table index to zero worked for
normal static relocations but not for `-fPIC` code.
See https://github.com/emscripten-core/emscripten/issues/12819
Differential Revision: https://reviews.llvm.org/D92038
Nico Weber [Thu, 26 Nov 2020 02:11:50 +0000 (21:11 -0500)]
[gn build] sync script: try to make a loop clearer
We want to find 'sources = [', but only if it's not the prefix
of 'sources = []' -- we're looking for a non-empty list.
No behavior change.
Nico Weber [Thu, 26 Nov 2020 02:05:11 +0000 (21:05 -0500)]
[gn build] sync script: only compute tokloc when adding files
No behavior change, but maybe a bit clearer.
Nico Weber [Thu, 26 Nov 2020 01:56:03 +0000 (20:56 -0500)]
[gn build] sync script: handle multiple source lists if only deleting
With this, changes like
b534beabeed3ba can be merged automatically.
Nathan Ridge [Tue, 24 Nov 2020 01:59:36 +0000 (20:59 -0500)]
[clangd] Collect main file refs by default
This is needed for call hierarchy to be able to find callers of
main-file-only functions.
Differential Revision: https://reviews.llvm.org/D92000
Aart Bik [Wed, 25 Nov 2020 20:29:05 +0000 (12:29 -0800)]
[mlir][sparse] add ability to select pointer/index storage type
This change gives sparse compiler clients more control over selecting
individual types for the pointers and indices in the sparse storage schemes.
Narrower width obviously results in smaller memory footprints, but the
range should always suffice for the maximum number of entries or index value.
Reviewed By: penpornk
Differential Revision: https://reviews.llvm.org/D92126
Nico Weber [Thu, 26 Nov 2020 01:19:46 +0000 (20:19 -0500)]
[gn build] (manually) port
b534beabeed3ba
Richard Smith [Thu, 26 Nov 2020 01:00:23 +0000 (17:00 -0800)]
Part of C++ DR 39: a class member lookup is not ambiguous if it finds the
same type in multiple base classes.
Not even if the type is introduced by distinct declarations (for
example, two typedef declarations, or a typedef and a class definition).
Zhengyang Liu [Tue, 24 Nov 2020 21:55:24 +0000 (14:55 -0700)]
Adding PoisonValue for representing poison value explicitly in IR
Define ConstantData::PoisonValue.
Add support for poison value to LLLexer/LLParser/BitcodeReader/BitcodeWriter.
Add support for poison value to llvm-c interface.
Add support for poison value to OCaml binding.
Add m_Poison in PatternMatch.
Differential Revision: https://reviews.llvm.org/D71126
Richard Smith [Wed, 25 Nov 2020 23:22:51 +0000 (15:22 -0800)]
Refactor and simplify class scope name lookup.
This is partly in preparation for an upcoming change that can change the
order in which DeclContext lookup results are presented.
In passing, fix some obvious errors where name lookup's notion of a
"static member function" missed static member function templates, and
where its notion of "same set of declarations" was confused by the same
declarations appearing in a different order.
Amy Huang [Tue, 24 Nov 2020 19:26:14 +0000 (11:26 -0800)]
[CodeView] Avoid emitting empty debug globals subsection.
In https://reviews.llvm.org/D89072 I added static const data members
to the debug subsection for globals. It skipped emitting an S_CONSTANT if it
didn't have a value, which meant the subsection could be empty.
This patch fixes the empty subsection issue.
Differential Revision: https://reviews.llvm.org/D92049
Reid Kleckner [Thu, 26 Nov 2020 00:08:01 +0000 (16:08 -0800)]
Revert builtins fp16 support: tests do not pass on Mac
Revert "[compiler-rt] [builtins] Support conversion between fp16 and fp128" & dependency
Revert "[compiler-rt] [builtins] Use _Float16 on extendhfsf2, truncdfhf2 __truncsfhf2 if available"
This reverts commit
7a948298813c913841a36ed0b460db0856fe7082.
This reverts commit
1fb91fcf9cfe849c7e9996597c491306e34e7abc.
Sean Silva [Thu, 26 Nov 2020 00:10:34 +0000 (16:10 -0800)]
[NFC] Fix pattern name.
It still had the old name from before ElementwiseMappable was added.
Craig Topper [Wed, 25 Nov 2020 23:07:34 +0000 (15:07 -0800)]
[RISCV] Add an implementation of isFMAFasterThanFMulAndFAdd
Start with an assumption that FMA is faster than Fmul+FAdd. If thats not true
on some particular implementation we can add a tuning parameter in the future.
I've update the fmuladd test cases and added new test cases for fast math flag
based contraction.
Differential Revision: https://reviews.llvm.org/D91987
Craig Topper [Wed, 25 Nov 2020 22:54:26 +0000 (14:54 -0800)]
[SelectionDAGBuilder] Add SPF_NABS support to visitSelect
We currently don't match this which limits the effectiveness of D91120 until
InstCombine starts canonicalizing to llvm.abs. This should be easy to remove
if/when we remove the SPF_ABS handling.
Differential Revision: https://reviews.llvm.org/D92118
Tom Stellard [Wed, 25 Nov 2020 22:43:18 +0000 (14:43 -0800)]
github actions: Use llvmbot token for main branch sync
The default github actions token cannot push to restricted branches, so we need to use a token from the llvmbot user.
David Blaikie [Wed, 25 Nov 2020 21:34:15 +0000 (13:34 -0800)]
DebugInfo: remove unnecessary mtriple from test/DebugInfo/X86/abstract_origin.ll
The test already specifies a triple in the IR itself.
Based on post-commit feedback from Luís Marques.
Louis Dionne [Wed, 25 Nov 2020 21:14:28 +0000 (16:14 -0500)]
[libc++] NFC: Reindent non-lockfree-atomics feature
Eric Astor [Wed, 25 Nov 2020 20:37:00 +0000 (15:37 -0500)]
[ms] [llvm-ml] Implement the expression expansion operator
In text-item contexts, %expr expands to a string containing the results of evaluating `expr`.
Reviewed By: thakis
Differential Revision: https://reviews.llvm.org/D89736
Marius Brehler [Thu, 19 Nov 2020 22:21:00 +0000 (22:21 +0000)]
[mlir] Add build configuration for Quant unittest
Craig Topper [Wed, 25 Nov 2020 20:40:25 +0000 (12:40 -0800)]
[RISCV] Add test cases to check that we use (smax X, (neg X)) for abs with Zbb extension.
Craig Topper [Wed, 25 Nov 2020 20:10:40 +0000 (12:10 -0800)]
[RISCV] Make SMIN/SMAX/UMIN/UMAX legal with Zbb extension.
This is the logically correct thing to do. But it generates worse
code for i32 umin/umax on the rv64 due to type legalize requesting
zext even though the arguments are sext. Maybe we can teach type
legalizer to use sext for umin/umax for RISCV.
It's also producing possibly worse code on i64 on RV32 since we
still end up with selects that become branches. But this seems
like something we could improve in type legalization or DAG combine.
Hopefully this makes D92095 work for RISCV with Zbb.
Louis Dionne [Thu, 12 Nov 2020 20:14:33 +0000 (15:14 -0500)]
[libc++] Factor out common logic for calling aligned allocation
There were a couple of places where we needed to call the underlying
platform's aligned allocation/deallocation function. Instead of having
the same logic all over the place, extract the logic into a pair of
helper functions __libcpp_aligned_alloc and __libcpp_aligned_free.
The code in libcxxabi/src/fallback_malloc.cpp looks like it could be
simplified after this change -- I purposefully did not simplify it
further to keep this change as straightforward as possible, since it
is touching very important parts of the library.
Also, the changes in libcxx/src/new.cpp and libcxxabi/src/stdlib_new_delete.cpp
are basically the same -- I just kept both source files in sync.
The underlying reason for this refactoring is to make it easier to support
platforms that provide aligned allocation through C11's aligned_alloc
function instead of posix_memalign. After this change, we'll only have
to add support for that in a single place.
Differential Revision: https://reviews.llvm.org/D91379
Sam McCall [Wed, 25 Nov 2020 20:31:18 +0000 (21:31 +0100)]
[clangd] Track deprecation of 'member' semantic token type in LSP.
Frank Laub [Wed, 25 Nov 2020 20:22:01 +0000 (20:22 +0000)]
[MLIR][Affine] Add custom builders for AffineVectorLoadOp/AffineVectorStoreOp
Adding missing custom builders for AffineVectorLoadOp & AffineVectorStoreOp. In practice, it is difficult to correctly construct these ops without these builders (because the AffineMap is not included at construction time).
Differential Revision: https://reviews.llvm.org/D86380
Reid Kleckner [Tue, 24 Nov 2020 19:50:33 +0000 (11:50 -0800)]
[MS] Add more 128bit cmpxchg intrinsics for AArch64
The MSVC STL for requires this on ARM64.
Requested in https://llvm.org/pr47099
Depends on D92061
Differential Revision: https://reviews.llvm.org/D92062
Reid Kleckner [Tue, 24 Nov 2020 22:48:05 +0000 (14:48 -0800)]
[MS] Fix double evaluation of MSVC builtin arguments
This code got quite twisted because we consider some MSVC builtins to be
target agnostic, and some to be target specific. Target specific
intrinsics have a pattern of doing up-front argument evaluation, while
general intrinsics do not evaluate their arguments up front. As we tried
to share codepaths between the target-specific and target-agnostic
handling, we ended up doing double evaluation.
Instead, have each target handle MSVC intrinsics consistently before up
front argument evaluation. This requires passing less data around and is
more consistent with target independent intrinsic handling.
See D50979 for past examples of this bug. I noticed this while looking
into adding some more intrinsics.
Differential Revision: https://reviews.llvm.org/D92061
Simon Pilgrim [Wed, 25 Nov 2020 19:00:33 +0000 (19:00 +0000)]
[Hexagon] Add support for ISD::SMAX/SMIN/UMAX/UMIN instead of custom dag patterns
This should handle the basic integer min/max handling - the HVX ops are still TODO.
This is some necessary cleanup work for min/max ops to eventually help us move the add/sub sat patterns into DAGCombine - D91876.
Differential Revision: https://reviews.llvm.org/D92112
Craig Topper [Wed, 25 Nov 2020 18:42:08 +0000 (10:42 -0800)]
[RISCV] Add test cases showing that we don't recognize the select form of NABS in SelectionDAGBuilder so we end up with branches. NFC
There's a FIXME that it should produce (sub 0, (abs)).
Paul Robinson [Wed, 25 Nov 2020 18:39:09 +0000 (13:39 -0500)]
Remove static function unused after cf1c774.
Caused some -Werror bot failures.
Simon Pilgrim [Wed, 25 Nov 2020 18:39:56 +0000 (18:39 +0000)]
[CostModel][X86] Refresh ISD::ABS costs
Update costs now that D92095 and D92102 have tweaked the SSE2 implementation
The SSE42 BLENDVPD cost can actually be used on SSE41 as we don't attempt to generate PCMPGT anymore
Add scalar i16/i32/i64 costs as we can do this cheaply with CMOV
LLVM GN Syncbot [Wed, 25 Nov 2020 18:35:53 +0000 (18:35 +0000)]
[gn build] Port
73fdd998701
Nathan James [Wed, 25 Nov 2020 18:35:34 +0000 (18:35 +0000)]
[clangd] Implement clang-tidy options from config
Added some new ClangTidyOptionsProvider like classes designed for clangd work flow.
These providers are designed to source the options on the worker thread but in a thread safe manner.
This is done through making the options getter take a pointer to the filesystem used by the worker thread which natuarally is from a ThreadsafeFS.
Internal caching in the providers is also guarded.
The providers don't inherit from `ClangTidyOptionsProvider` instead they share a base class which is able to create a provider for the `ClangTidyContext` using a specific FileSystem.
This approach means one provider can be used for multiple contexts even though `ClangTidyContext` owns its provider.
Depends on D90531
Reviewed By: sammccall
Differential Revision: https://reviews.llvm.org/D91029
Simon Pilgrim [Wed, 25 Nov 2020 18:05:41 +0000 (18:05 +0000)]
[DAG] Legalize abs(x) -> umin(x,sub(0,x)) iff umin/sub are legal
If umin() is legal, this is likely to result in smaller codegen expansion for abs(x) than the xor(add,ashr) method.
Followup to D92095
Alive2: https://alive2.llvm.org/ce/z/8nuX6s https://alive2.llvm.org/ce/z/q2hB9w
Paul Robinson [Wed, 18 Nov 2020 21:27:14 +0000 (16:27 -0500)]
[FastISel] Flush local value map on ever instruction
Local values are constants or addresses that can't be folded into
the instruction that uses them. FastISel materializes these in a
"local value" area that always dominates the current insertion
point, to try to avoid materializing these values more than once
(per block).
https://reviews.llvm.org/D43093 added code to sink these local
value instructions to their first use, which has two beneficial
effects. One, it is likely to avoid some unnecessary spills and
reloads; two, it allows us to attach the debug location of the
user to the local value instruction. The latter effect can
improve the debugging experience for debuggers with a "set next
statement" feature, such as the Visual Studio debugger and PS4
debugger, because instructions to set up constants for a given
statement will be associated with the appropriate source line.
There are also some constants (primarily addresses) that could be
produced by no-op casts or GEP instructions; the main difference
from "local value" instructions is that these are values from
separate IR instructions, and therefore could have multiple users
across multiple basic blocks. D43093 avoided sinking these, even
though they were emitted to the same "local value" area as the
other instructions. The patch comment for D43093 states:
Local values may also be used by no-op casts, which adds the
register to the RegFixups table. Without reversing the RegFixups
map direction, we don't have enough information to sink these
instructions.
This patch undoes most of D43093, and instead flushes the local
value map after(*) every IR instruction, using that instruction's
debug location. This avoids sometimes incorrect locations used
previously, and emits instructions in a more natural order.
This does mean materialized values are not re-used across IR
instruction boundaries; however, only about 5% of those values
were reused in an experimental self-build of clang.
(*) Actually, just prior to the next instruction. It seems like
it would be cleaner the other way, but I was having trouble
getting that to work.
Differential Revision: https://reviews.llvm.org/D91734
Craig Topper [Wed, 25 Nov 2020 17:43:16 +0000 (09:43 -0800)]
[RISCV] Custom type legalize i32 fshl/fshr on RV64 with Zbt.
This adds custom opcodes for FSLW/FSRW so we can type legalize
fshl/fshr without needing to match a sign_extend_inreg.
I've used the operand order from fshl/fshr to make the isel
pattern similar to the non-W form. It was also hard to decide
another order since the register instruction has the shift amount
as the second operand, but the immediate instruction has it as
the third operand.
Differential Revision: https://reviews.llvm.org/D91479
Johannes Doerfert [Wed, 25 Nov 2020 17:49:30 +0000 (11:49 -0600)]
[OpenMP][Docs] Add more content, call coordinates, FAQ entries, links
Johannes Doerfert [Wed, 25 Nov 2020 15:59:50 +0000 (09:59 -0600)]
[Flang][Docs] Update call information and add two more calls
Call information have been updated and the OpenMP in LLVM as well as
Classic Flang call have been added.
AndreyChurbanov [Wed, 25 Nov 2020 17:40:23 +0000 (20:40 +0300)]
[OpenMP] libomp: fix non-X86, non-AARCH64 builds
Commit https://reviews.llvm.org/rG7b5254223acbf2ef9cd278070c5a84ab278d7e5f
broke the build for some architectures, because macro KMP_PREFIX_UNDERSCORE
was defined only for x86, x86_64 and aarch64. This patch defines it for other
architectures (as a no-op).
Differential Revision: https://reviews.llvm.org/D92027
Joachim Protze [Wed, 18 Nov 2020 11:49:19 +0000 (12:49 +0100)]
[OpenMP][OMPT] Introduce a guard to handle OMPT return address
This is an alternative approach to address inconsistencies pointed out in: D90078
This patch makes sure that the return address is reset, when leaving the scope.
In some cases, I had to move the macro out of an if-statement to have it in the
right scope, in some cases I added an additional block to restrict the scope.
This patch does not handle inconsistencies, which might occur if the return
address is still set when we call into the application.
Test case (repeated_calls.c) provided by @hbae
Differential Revision: https://reviews.llvm.org/D91692
Isabel Thärigen [Tue, 27 Oct 2020 13:05:28 +0000 (14:05 +0100)]
[OpenMP][OMPT] Implement verbose tool loading
OpenMP 5.1 introduces the new env variable
OMP_TOOL_VERBOSE_INIT=(disabled|stdout|stderr|<filename>) to enable verbose
loading and initialization of OMPT tools.
This env variable helps to understand the cause when loading of a tool fails
(e.g., undefined symbols or dependency not in LD_LIBRARY_PATH)
Output of OMP_TOOL_VERBOSE_INIT is added for OMP_DISPLAY_ENV
Tests for this patch are integrated into the different existing tool loading
tests, making these tests more verbose. An Archer specific verbose test is
integrated into an existing Archer test.
Patch prepared by: Isabel Thärigen
Differential Revision: https://reviews.llvm.org/D91464
Nico Weber [Wed, 25 Nov 2020 17:08:10 +0000 (12:08 -0500)]
[gn build] Download prebuilt mac-arm64 binary now that it exists
Sadly requires an ugly workaround for an ugly bug, but still nicer than
building locally.
Endre Fülöp [Fri, 16 Oct 2020 07:37:42 +0000 (09:37 +0200)]
[clang][test] Fix prefix operator++ signature in iterators
Prefix operator++ should return the iterator incremented by reference.
Differential Revision: https://reviews.llvm.org/D89528
Fangrui Song [Wed, 25 Nov 2020 17:00:55 +0000 (09:00 -0800)]
[ELF] Rename adjustRelaxExpr to adjustTlsExpr and delete the unused `data` parameter. NFC
Reviewed By: psmith
Differential Revision: https://reviews.llvm.org/D91995
Joe Ellis [Tue, 24 Nov 2020 10:08:04 +0000 (10:08 +0000)]
[SVE] Fix TypeSize warning in RuntimePointerChecking::insert
The TypeSize warning would occur because RuntimePointerChecking::insert
was not scalable vector aware. The fix is to use
ScalarEvolution::getSizeOfExpr to grab the size of types.
Differential Revision: https://reviews.llvm.org/D90171
Fangrui Song [Wed, 25 Nov 2020 16:43:26 +0000 (08:43 -0800)]
[ELF] Add TargetInfo::adjustGotPcExpr for `R_GOT_PC` relaxations. NFC
With this change, `TargetInfo::adjustRelaxExpr` is only related to TLS
relaxations and a subsequent clean-up can delete the `data` parameter.
Differential Revision: https://reviews.llvm.org/D92079
Roman Lebedev [Wed, 25 Nov 2020 16:38:51 +0000 (19:38 +0300)]
[AMDGPU] Actually fully update opt-pipeline.ll test to account for -loop-idiom vs -indvars switch
Andy Wingo [Wed, 25 Nov 2020 16:31:05 +0000 (08:31 -0800)]
[MC][WebAssembly] Only emit indirect function table import if needed
The indirect function table, synthesized by the linker, is needed if and
only if there are TABLE_INDEX relocs.
Differential Revision: https://reviews.llvm.org/D91637
Roman Lebedev [Wed, 25 Nov 2020 16:17:25 +0000 (19:17 +0300)]
[PassManager] Run Induction Variable Simplification pass *after* Recognize loop idioms pass, not before
Currently, `-indvars` runs first, and then immediately after `-loop-idiom` does.
I'm not really sure if `-loop-idiom` requires `-indvars` to run beforehand,
but i'm *very* sure that `-indvars` requires `-loop-idiom` to run afterwards,
as it can be seen in the phase-ordering test.
LoopIdiom runs on two types of loops: countable ones, and uncountable ones.
For uncountable ones, IndVars obviously didn't make any change to them,
since they are uncountable, so for them the order should be irrelevant.
For countable ones, well, they should have been countable before IndVars
for IndVars to make any change to them, and since SCEV is used on them,
it shouldn't matter if IndVars have already canonicalized them.
So i don't really see why we'd want the current ordering.
Should this cause issues, it will give us a reproducer test case
that shows flaws in this logic, and we then could adjust accordingly.
While this is quite likely beneficial in-the-wild already,
it's a required part for the full motivational pattern
behind `left-shift-until-bittest` loop idiom (D91038).
Reviewed By: dmgreen
Differential Revision: https://reviews.llvm.org/D91800
Andy Wingo [Wed, 25 Nov 2020 15:54:31 +0000 (07:54 -0800)]
[WebAssembly] Factor out WasmTableType in binary format
This commit factors out a WasmTableType definition from WasmTable, as is
the case for WasmGlobal and other data types. Also add support for
extracting the SymbolName for a table from the linking section's symbol
table.
Differential Revision: https://reviews.llvm.org/D91849
Sebastian Neubauer [Fri, 23 Oct 2020 09:20:20 +0000 (11:20 +0200)]
[AMDGPU] Emit stack frame size in metadata
Add .shader_functions to pal metadata, which contains the stack frame
size for all non-entry-point functions.
Differential Revision: https://reviews.llvm.org/D90036
Valentin Clement [Wed, 25 Nov 2020 15:28:02 +0000 (10:28 -0500)]
[flang][openacc] Semantic check for cache directive
Add semantic check for the cache directive. According to section 2.10 from the specification:
A var in a cache directive must be a single array element or a simple subarray.
Reviewed By: kiranchandramohan
Differential Revision: https://reviews.llvm.org/D90184
Simon Pilgrim [Wed, 25 Nov 2020 15:03:03 +0000 (15:03 +0000)]
[DAG] Legalize abs(x) -> smax(x,sub(0,x)) iff smax/sub are legal
If smax() is legal, this is likely to result in smaller codegen expansion for abs(x) than the xor(add,ashr) method.
This is also what PowerPC has been doing for its abs implementation, so it lets us get rid of a load of custom lowering code there (and which was never updated when they added smax lowering).
Alive2: https://alive2.llvm.org/ce/z/xRk3cD
Differential Revision: https://reviews.llvm.org/D92095
Simon Pilgrim [Wed, 25 Nov 2020 11:50:16 +0000 (11:50 +0000)]
[PowerPC] Regenerate vec_select.ll tests and add <1 x i128> test case
Guillaume Chatelet [Wed, 25 Nov 2020 14:23:31 +0000 (14:23 +0000)]
Fix case mismatch between definition and declaration
Arthur O'Dwyer [Tue, 24 Nov 2020 14:59:26 +0000 (09:59 -0500)]
[libc++] ADL-proof <variant> by adding _VSTD:: qualification on calls.
Differential Revision: https://reviews.llvm.org/D92036
Adam Czachorowski [Tue, 24 Nov 2020 19:47:37 +0000 (20:47 +0100)]
[clangd] PopulateSwitch: disable on dependent enums.
If the enum is a dependent type, we would crash somewhere in
getIntWidth(). -Wswitch diagnostic doesn't work on dependent enums
either.
Differential Revision: https://reviews.llvm.org/D92051
Cullen Rhodes [Wed, 18 Nov 2020 18:13:08 +0000 (18:13 +0000)]
[LAA] NFC: Rename [get]MaxSafeRegisterWidth -> [get]MaxSafeVectorWidthInBits
MaxSafeRegisterWidth is a misnomer since it actually returns the maximum
safe vector width. Register suggests it relates directly to a physical
register where it could be a vector spanning one or more physical
registers.
Reviewed By: sdesmalen
Differential Revision: https://reviews.llvm.org/D91727
Mark de Wever [Wed, 25 Nov 2020 12:46:08 +0000 (13:46 +0100)]
Revert "[libc++] P1645 constexpr for <numeric>"
This reverts commit
eb9b063539c34d0d4dd14e8516eeb77bb8b9e4bd.
The commit fails to build on build bots using LLVM 8.
Aaron Ballman [Wed, 25 Nov 2020 12:33:38 +0000 (07:33 -0500)]
Fix a typo in the documentation to unbreak the sphinx builder.
Georgii Rymar [Tue, 24 Nov 2020 11:54:48 +0000 (14:54 +0300)]
[llvm-readelf/obj] - Stop using `reportError` when dumping notes.
This starts using `reportUniqueWarnings` instead of `reportError`
in the code that is responsible for dumping notes.
Differential revision: https://reviews.llvm.org/D92021
Mark de Wever [Tue, 24 Nov 2020 13:55:55 +0000 (14:55 +0100)]
[libc++] P1645 constexpr for <numeric>
Implements P1645: constexpr for <numeric> algorithms
Reviewed By: ldionne, #libc
Differential Revision: https://reviews.llvm.org/D90569
Sam McCall [Wed, 25 Nov 2020 11:49:18 +0000 (12:49 +0100)]
[clangd] Use TimePoint<> instead of system_clock::time_point, it does matter after all.
Mark Murray [Mon, 16 Nov 2020 13:11:35 +0000 (13:11 +0000)]
[ARM][AArch64] Adding Neoverse N2 CPU support
Add support for the Neoverse N2 CPU to the ARM and AArch64 backends.
Differential Revision: https://reviews.llvm.org/D91695
Simon Pilgrim [Tue, 24 Nov 2020 17:36:58 +0000 (17:36 +0000)]
SemaExpr.cpp - use castAs<> instead of getAs<> as we dereference the pointer directly. NFCI.
castAs<> will assert the correct cast type instead of just returning null, which we then try to dereference immediately.
Simon Pilgrim [Tue, 24 Nov 2020 16:56:24 +0000 (16:56 +0000)]
TargetInfo.cpp - use castAs<> instead of getAs<> as we dereference the pointer directly. NFCI.
castAs<> will assert the correct cast type instead of just returning null, which we then try to dereference immediately.
Simon Pilgrim [Tue, 24 Nov 2020 16:55:26 +0000 (16:55 +0000)]
CGCall.cpp - use castAs<> instead of getAs<> as we dereference the pointer directly. NFCI.
castAs<> will assert the correct cast type instead of just returning null, which we then try to dereference immediately in the setUsedBits call.
Simon Pilgrim [Tue, 24 Nov 2020 16:51:55 +0000 (16:51 +0000)]
SemaExpr.cpp - use castAs<> instead of getAs<> as we dereference the pointer directly. NFCI.
castAs<> will assert the correct cast type instead of just returning null, which we then try to dereference immediately.
Simon Pilgrim [Tue, 24 Nov 2020 16:20:03 +0000 (16:20 +0000)]
DetectDeadLanes.cpp - remove unused headers. NFCI.
Kerry McLaughlin [Wed, 25 Nov 2020 10:54:31 +0000 (10:54 +0000)]
[SVE][CodeGen] Add a DAG combine to extend mscatter indices
This patch adds a target-specific DAG combine for mscatter to promote indices
with element types i8 or i16 before legalisation, plus various tests with illegal types.
Reviewed By: sdesmalen
Differential Revision: https://reviews.llvm.org/D90945
LLVM GN Syncbot [Wed, 25 Nov 2020 11:13:15 +0000 (11:13 +0000)]
[gn build] Port
d95db1693cb
Sam McCall [Wed, 23 Sep 2020 17:58:32 +0000 (19:58 +0200)]
[clangd] Extract common file-caching logic from ConfigProvider.
The plan is to use this to use this for .clang-format, .clang-tidy, and
compile_commands.json. (Currently the former two are reparsed every
time, and the latter is cached forever and changes are never seen).
Differential Revision: https://reviews.llvm.org/D88172