platform/upstream/llvm.git
2 years ago[mlir][Linalg] Properly propagate transform result in ScalarizeOp
Nicolas Vasilache [Tue, 27 Dec 2022 14:14:58 +0000 (06:14 -0800)]
[mlir][Linalg] Properly propagate transform result in ScalarizeOp

2 years ago[RS4GC] Rematerialize derived pointers before uses.
Denis Antrushin [Tue, 29 Nov 2022 11:47:35 +0000 (18:47 +0700)]
[RS4GC] Rematerialize derived pointers before uses.

Introduce an option to rematerialize derived pointers immediately
before their uses instead of after every statepoint. This can be
beneficial when pointer is live across many statepoints but has
few uses.
Initial implementation is simple and rematerializes derived pointer
before every use, even if there are several uses in the same block
or rematerialization instructions can be hoisted etc.
Transformation is considered profitable if we would insert less
instructions than we would insert after every live statepoint.

Depends on D138910D138911

Reviewed By: anna, skatkov

Differential Revision: https://reviews.llvm.org/D138912

2 years ago[SLP]Fix PR59693: Do not crash trying to set insert point for buildvector
Alexey Bataev [Tue, 27 Dec 2022 13:12:16 +0000 (05:12 -0800)]
[SLP]Fix PR59693: Do not crash trying to set insert point for buildvector
of extractvalues.

No need to get the last instruction only for vectorized extractvalues,
for gathered(buildvector sequence) still need to get the insertion
  point.

2 years ago[mlir] NFC - Expose scf::canonicalizeMinMaxOp
Nicolas Vasilache [Tue, 27 Dec 2022 13:47:02 +0000 (05:47 -0800)]
[mlir] NFC - Expose scf::canonicalizeMinMaxOp

2 years agoReapply [MergedLoadStoreMotion] Convert tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 11:20:05 +0000 (12:20 +0100)]
Reapply [MergedLoadStoreMotion] Convert tests to opaque pointers (NFC)

Reapply after reapplying dependent revision.

2 years ago[LoadStoreVectorizer] Convert tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 12:09:25 +0000 (13:09 +0100)]
[LoadStoreVectorizer] Convert tests to opaque pointers (NFC)

2 years ago[LoadStoreVectorize] Regenerate test checks (NFC)
Nikita Popov [Tue, 27 Dec 2022 11:57:52 +0000 (12:57 +0100)]
[LoadStoreVectorize] Regenerate test checks (NFC)

2 years ago[LoadStoreVectorizer] Convert some tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 11:53:30 +0000 (12:53 +0100)]
[LoadStoreVectorizer] Convert some tests to opaque pointers (NFC)

2 years ago[LoopBoundSplit] Convert tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 11:52:22 +0000 (12:52 +0100)]
[LoopBoundSplit] Convert tests to opaque pointers (NFC)

2 years agoReapply [MergeLoadStoreMotion] Don't require GEP for sinking
Nikita Popov [Tue, 27 Dec 2022 11:17:58 +0000 (12:17 +0100)]
Reapply [MergeLoadStoreMotion] Don't require GEP for sinking

Reapply with a fix for a failing debuginfo assignment tracking test.

-----

Allow sinking stores where both operands are the same, don't require
them to have an identical GEP in each block.

This came up when migrating tests to opaque pointers, where
zero-index GEPs are omitted.

2 years ago[GVNHoist] Make test more robust (NFC)
Nikita Popov [Tue, 27 Dec 2022 11:44:17 +0000 (12:44 +0100)]
[GVNHoist] Make test more robust (NFC)

Make sure these stores cannot be sunk, which might defeat the
intent of the test.

2 years agoRevert "[MergeLoadStoreMotion] Don't require GEP for sinking"
Nikita Popov [Tue, 27 Dec 2022 11:37:49 +0000 (12:37 +0100)]
Revert "[MergeLoadStoreMotion] Don't require GEP for sinking"

I missed a test failure in the DebugInfo directory.

This reverts commit 2c15b9d9e1a898cfd849db81b36d278eac3ef24e.
This reverts commit fb435e1cb5842e1437436e9e7378dfc4106fdad8.

2 years ago[MergedLoadStoreMotion] Convert tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 11:20:05 +0000 (12:20 +0100)]
[MergedLoadStoreMotion] Convert tests to opaque pointers (NFC)

2 years ago[MergeLoadStoreMotion] Don't require GEP for sinking
Nikita Popov [Tue, 27 Dec 2022 11:17:58 +0000 (12:17 +0100)]
[MergeLoadStoreMotion] Don't require GEP for sinking

Allow sinking stores where both operands are the same, don't require
them to have an identical GEP in each block.

This came up when migrating tests to opaque pointers, where
zero-index GEPs are omitted.

2 years ago[MergedLoadStoreMotion] Add tests for store without GEPs (NFC)
Nikita Popov [Tue, 27 Dec 2022 11:03:59 +0000 (12:03 +0100)]
[MergedLoadStoreMotion] Add tests for store without GEPs (NFC)

MergedLoadStoreMotion currently only handles the case where each
store has it's own GEP. It fails to handle the case where the
store argument is exactly the same.

2 years ago[Tests] Rename InstMerge -> MergedLoadStoreMotion (NFC)
Nikita Popov [Tue, 27 Dec 2022 10:52:33 +0000 (11:52 +0100)]
[Tests] Rename InstMerge -> MergedLoadStoreMotion (NFC)

These are tests for the MergeLoadStoreMotion pass, so name them
accordingly.

2 years ago[reland][libc][NFC] Add -fno-lax-vector-conversions compilation flag
Guillaume Chatelet [Tue, 27 Dec 2022 08:25:32 +0000 (08:25 +0000)]
[reland][libc][NFC] Add -fno-lax-vector-conversions compilation flag

Now that a3d2c344773cc4fc95136fd67245880b34d8e335 has been submitted.

2 years ago[libc][NFC] Fix lax vector conversion for aarch64
Guillaume Chatelet [Tue, 27 Dec 2022 10:16:23 +0000 (10:16 +0000)]
[libc][NFC] Fix lax vector conversion for aarch64

2 years ago[InterleavedAccess] Convert tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 09:57:34 +0000 (10:57 +0100)]
[InterleavedAccess] Convert tests to opaque pointers (NFC)

2 years ago[LCSSA] Convert tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 09:56:49 +0000 (10:56 +0100)]
[LCSSA] Convert tests to opaque pointers (NFC)

2 years ago[Internalize] Convert tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 09:54:35 +0000 (10:54 +0100)]
[Internalize] Convert tests to opaque pointers (NFC)

2 years ago[InferFunctionAttrs] Convert tests to opaque pointers (NFC)
Nikita Popov [Tue, 27 Dec 2022 09:53:42 +0000 (10:53 +0100)]
[InferFunctionAttrs] Convert tests to opaque pointers (NFC)

2 years ago[clang] Use a StringRef instead of a raw char pointer to store builtin and call infor...
serge-sans-paille [Mon, 26 Dec 2022 08:19:09 +0000 (09:19 +0100)]
[clang] Use a StringRef instead of a raw char pointer to store builtin and call information

This avoids recomputing string length that is already known at compile time.

It has a slight impact on preprocessing / compile time, see

https://llvm-compile-time-tracker.com/compare.php?from=3f36d2d579d8b0e8824d9dd99bfa79f456858f88&to=e49640c507ddc6615b5e503144301c8e41f8f434&stat=instructions:u

This a recommit of e953ae5bbc313fd0cc980ce021d487e5b5199ea4 and the subsequent fixes caa713559bd38f337d7d35de35686775e8fb5175 and 06b90e2e9c991e211fecc97948e533320a825470.

The above patchset caused some version of GCC to take eons to compile clang/lib/Basic/Targets/AArch64.cpp, as spotted in aa171833ab0017d9732e82b8682c9848ab25ff9e.
The fix is to make BuiltinInfo tables a compilation unit static variable, instead of a private static variable.

Differential Revision: https://reviews.llvm.org/D139881

2 years agoRevert "[libc][NFC] Add -fno-lax-vector-conversions compilation flag"
Guillaume Chatelet [Tue, 27 Dec 2022 08:30:19 +0000 (08:30 +0000)]
Revert "[libc][NFC] Add -fno-lax-vector-conversions compilation flag"

This breaks aarch64 build.

This reverts commit 32f4c3f1033e32b7d2b7ba310c46c8c0d4013eb0.

2 years ago[libc][NFC] Add -fno-lax-vector-conversions compilation flag
Guillaume Chatelet [Tue, 27 Dec 2022 08:25:32 +0000 (08:25 +0000)]
[libc][NFC] Add -fno-lax-vector-conversions compilation flag

2 years ago[NFC][Clang][RISCV] Rename data member 'DefaultPolicy' to 'PolicyAttrs'
eopXD [Thu, 22 Dec 2022 09:20:09 +0000 (01:20 -0800)]
[NFC][Clang][RISCV] Rename data member 'DefaultPolicy' to 'PolicyAttrs'

The naming here is strange since the value may still be updated.

Reviewed By: kito-cheng, khchen

Differential Revision: https://reviews.llvm.org/D140389

2 years ago[RFC][OpenMP] Update to Python3 for lit test
Shilei Tian [Tue, 27 Dec 2022 02:39:33 +0000 (21:39 -0500)]
[RFC][OpenMP] Update to Python3 for lit test

I think it's reasonable to upgrade to Python 3 for LIT test requirement because `lit` itself (`llvm/utils/lit/lit.py`) already switched to Python 3. In addition, LLVM already requires Python 3.6 to be the minimum version (https://llvm.org/docs/GettingStarted.html#software).

Reviewed By: jdoerfert, jhuber6

Differential Revision: https://reviews.llvm.org/D139855

2 years ago[Test] Add tests with logical AND/OR
Max Kazantsev [Tue, 27 Dec 2022 01:58:12 +0000 (08:58 +0700)]
[Test] Add tests with logical AND/OR

2 years ago[mlir] Include type_traits to support latest libc++
Fangrui Song [Tue, 27 Dec 2022 01:37:03 +0000 (17:37 -0800)]
[mlir] Include type_traits to support latest libc++

for std::make_unsigned_t

2 years ago[libc++] Granularize <type_traits> includes in <iterator>
Nikolas Klauser [Tue, 20 Dec 2022 20:13:12 +0000 (21:13 +0100)]
[libc++] Granularize <type_traits> includes in <iterator>

Reviewed By: Mordante, #libc

Spies: libcxx-commits

Differential Revision: https://reviews.llvm.org/D140621

2 years ago[mlir] Typos in affine dialect
Kai Sasaki [Tue, 27 Dec 2022 00:30:00 +0000 (09:30 +0900)]
[mlir] Typos in affine dialect

2 years ago[VPlan] Remove redundant blocks by merging them into predecessors.
Florian Hahn [Mon, 26 Dec 2022 22:46:24 +0000 (22:46 +0000)]
[VPlan] Remove redundant blocks by merging them into predecessors.

Add and run VPlan transform to fold blocks with a single predecessor
into the predecessor. This remove redundant blocks and addresses a TODO
to replace special handling for the vector latch VPBB.

Reviewed By: Ayal

Differential Revision: https://reviews.llvm.org/D139927

2 years ago[DAGCombine] `combineShuffleToZeroExtendVectorInReg()`: widen shuffle elements before...
Roman Lebedev [Mon, 26 Dec 2022 20:45:37 +0000 (23:45 +0300)]
[DAGCombine] `combineShuffleToZeroExtendVectorInReg()`: widen shuffle elements before trying to match

We might have sunk a bitcast into shuffle, and now it might be operating
on more fine-grained elements than what we'd match, so we must not be
dependent on whatever the granularity the shuffle happened to be in,
but transform it into the one canonical for us - with widest elements.

2 years ago[NFC][X86] Add some tests that can be matched as ZERO_EXTEND_VECTOR_INREG
Roman Lebedev [Mon, 26 Dec 2022 21:41:59 +0000 (00:41 +0300)]
[NFC][X86] Add some tests that can be matched as ZERO_EXTEND_VECTOR_INREG

2 years ago[DAG] `combineShuffleToZeroExtendVectorInReg()`: try to match w/ commuted operands
Roman Lebedev [Mon, 26 Dec 2022 19:49:01 +0000 (22:49 +0300)]
[DAG] `combineShuffleToZeroExtendVectorInReg()`: try to match w/ commuted operands

We don't have any reason to expect that the operand we will match
is on any particular hand of the shuffle, so we should try both.

2 years ago[AArch64] Custom lower `ISD::ZERO_EXTEND_VECTOR_INREG`
Roman Lebedev [Mon, 26 Dec 2022 18:46:20 +0000 (21:46 +0300)]
[AArch64] Custom lower `ISD::ZERO_EXTEND_VECTOR_INREG`

The baseline legalization for `ISD::ZERO_EXTEND_VECTOR_INREG`
(`VectorLegalizer::ExpandZERO_EXTEND_VECTOR_INREG`),
blends-in the zeros, but as mentioned e.g.
in b4bd0a404fe26071dab0854dfd9767974909c7c4,
there is no such thing for AArch64.

So some of the shuffles that would be nicely lowered
by `LowerVECTOR_SHUFFLE()`, e.g. into `ZIP1`,
would now be unrecognizable after round-tripping
through `ISD::ZERO_EXTEND_VECTOR_INREG` recognition & legalization.

The most obvious solution is to just custom-lower
`ISD::ZERO_EXTEND_VECTOR_INREG` as the `ZIP1`-with-zeros,
like it would have been originally in that test case.

2 years ago[DAGCombiner] Add a most basic `combineShuffleToZeroExtendVectorInReg()`
Roman Lebedev [Sun, 25 Dec 2022 19:16:30 +0000 (22:16 +0300)]
[DAGCombiner] Add a most basic `combineShuffleToZeroExtendVectorInReg()`

Sometimes we end up with a shuffles in DAG that would be
better represented as a `ISD::ZERO_EXTEND_VECTOR_INREG`,
and a failure to do so causes suboptimal codegen in a number of cases,
especially when we will then cast vector to scalar.

I acknowledge, the test changes here are rather underwhelming,
but as with all of codegen, it's always a yak shawing,
and this is the most stripped down version of the patch
that shows *some* effect without having insurmountable amount
of fallout to deal with. The next change resolves this regression.

The transformation will be extended in follow-ups.

2 years ago[NFC][AArch64] Add a few vector shuffle tests that should be `zip1`
Roman Lebedev [Mon, 26 Dec 2022 18:06:20 +0000 (21:06 +0300)]
[NFC][AArch64] Add a few vector shuffle tests that should be `zip1`

At least, they are equivalent to the `@vzipNoBlend`, which is lowered into zip1.

2 years agoCommit changes to the Code of Conduct that make it more clear regarding behavior...
tlattner [Mon, 26 Dec 2022 19:29:44 +0000 (11:29 -0800)]
Commit changes to the Code of Conduct that make it more clear regarding behavior outside of LLVM spaces that impact the safety of our community members. Discussion may be found here: https://discourse.llvm.org/t/code-of-conduct-changes-related-to-llvm-project-policy-changes/64197

2 years ago[InstCombine] do not add "nuw" to 1<<X if the "1" has undefined elements
Sanjay Patel [Mon, 26 Dec 2022 18:12:44 +0000 (13:12 -0500)]
[InstCombine] do not add "nuw" to 1<<X if the "1" has undefined elements

This was noted as a potential miscompile in the post-commit feedback
for the patch that added this fold:
d4493dd1ed58ac3f1eab0

2 years ago[InstCombine] replace undef in vector tests with poison; NFC
Sanjay Patel [Mon, 26 Dec 2022 17:34:47 +0000 (12:34 -0500)]
[InstCombine] replace undef in vector tests with poison; NFC

I left a few of the existing undef tests in place for extra
coverage and because one of those was noted as a miscompile in
the post-commit feedback for d4493dd1ed58ac3f1eab0, but we are
transitioning to poison, so it is more valuable to test the
expected IR going forward.

2 years ago[lldb] [utils] Fix linking lit-cpuid to LLVM dylib
Michał Górny [Mon, 26 Dec 2022 13:30:35 +0000 (14:30 +0100)]
[lldb] [utils] Fix linking lit-cpuid to LLVM dylib

Use `LINK_COMPONENTS` instead of manual `target_link_libraries` to link
lit-cpuid to LLVM components.  This ensures that dylib is used along
with `LLVM_LINK_LLVM_DYLIB` rather than linking to component libraries
that may not be installed.

This fixes build failure on Gentoo after a dep on TargetParser component
was added in f09cf34d00625e57dea5317a3ac0412c07292148.

Differential Revision: https://reviews.llvm.org/D140671

2 years agoRevert "[IPSCCP] Enable specialization of functions."
Florian Hahn [Mon, 26 Dec 2022 16:02:59 +0000 (16:02 +0000)]
Revert "[IPSCCP] Enable specialization of functions."

This reverts commit 2656572d485127cc30b8fe9752024d2a0f1c50db.

It looks like CINT2017rate/502.gcc_r gets mis-compiled with LTO + PGO on
AArch64 with function specialization.

2 years ago[TwoAddressInstruction] Constrain RegClass when processing a statepoint
Danila Malyutin [Fri, 23 Dec 2022 16:47:31 +0000 (19:47 +0300)]
[TwoAddressInstruction] Constrain RegClass when processing a statepoint

This transformation could've triggered a verifier assert if RegA and RegB
were of different reg classes. Fix this by constraining as the comment
for replaceRegWith suggests.

Differential Revision: https://reviews.llvm.org/D140672

2 years ago[Xtensa 10/10] Add relaxations and fixups. Add rest part of Xtensa Core Instructions
Andrei Safronov [Mon, 26 Dec 2022 11:00:00 +0000 (12:00 +0100)]
[Xtensa 10/10] Add relaxations and fixups. Add rest part of Xtensa Core Instructions

Add branch/jump/call/l32r instructions and fixups support. Add R_XTENSA_32/R_XTENSA_SLOT0_OP
relocations in object files generation. Modify tests to support new instructions.
Add tests for relocations and fixups.

Differential Revision: https://reviews.llvm.org/D64836

2 years ago[Xtensa 9/10] Add basic support of Xtensa disassembler
Andrei Safronov [Mon, 26 Dec 2022 10:58:36 +0000 (11:58 +0100)]
[Xtensa 9/10] Add basic support of Xtensa disassembler

Differential Revision: https://reviews.llvm.org/D64835

2 years ago[Xtensa 8/10] Add support of the Xtensa shift/load/store/move and processor control...
Andrei Safronov [Mon, 26 Dec 2022 10:56:06 +0000 (11:56 +0100)]
[Xtensa 8/10] Add support of the Xtensa shift/load/store/move and processor control instructions

Add new subset of Core Instructions (not full yet). Add appropriate operands description,
modify asm parser, printer and code emitter. Modify tests to support new instructions.

Differential Revision: https://reviews.llvm.org/D64834

2 years ago[Xtensa 7/10] Add Xtensa instruction printer
Andrei Safronov [Mon, 26 Dec 2022 10:55:06 +0000 (11:55 +0100)]
[Xtensa 7/10] Add Xtensa instruction printer

Add printer for current instructions and operands subsets.
Also add basic tests of the Xtensa instructions.

Differential Revision: https://reviews.llvm.org/D64833

2 years ago[Xtensa 6/10] Add Xtensa basic assembler parser
Andrei Safronov [Mon, 26 Dec 2022 10:53:44 +0000 (11:53 +0100)]
[Xtensa 6/10] Add Xtensa basic assembler parser

Currently parse just described in *.td files Xtensa instructions and operands subsets.

Differential Revision: https://reviews.llvm.org/D64832

2 years ago[Xtensa 5/10] Add Xtensa MCTargetDescr initial functionality
Andrei Safronov [Mon, 26 Dec 2022 10:52:38 +0000 (11:52 +0100)]
[Xtensa 5/10] Add Xtensa MCTargetDescr initial functionality

Differential Revision: https://reviews.llvm.org/D64831

2 years ago[Xtensa 4/10] Add basic *td files with Xtensa architecture description
Andrei Safronov [Mon, 26 Dec 2022 10:49:11 +0000 (11:49 +0100)]
[Xtensa 4/10] Add basic *td files with Xtensa architecture description

Add initial Xtensa.td file with target machine description. Add XtensaInstrInfo.td,
currently describe just susbet of Core Instructions like ALU, Processor control,
memory barrier and some move instructions. Add descriptions of the instructions
formats(XtensaInstrInfo.td) and some immediate instruction operands(XtensaOperands.td).
Add General Registers and Special Registers classes.

Differential Revision: https://reviews.llvm.org/D64830

2 years ago[Xtensa 3/10] Add initial version of the Xtensa backend
Andrei Safronov [Mon, 26 Dec 2022 10:45:59 +0000 (11:45 +0100)]
[Xtensa 3/10] Add initial version of the Xtensa backend

Add Xtensa MCTargetDesc stub. Add XtensaTargetMachine and XtensaTargetInfo.
Modify llib/Target/LLVMBuild.txt. Now Xtensa target could be builded as EXPERIMENTAL.

Differential Revision: https://reviews.llvm.org/D64829

2 years ago[Xtensa 2/10] Add Xtensa ELF definitions
Andrei Safronov [Mon, 26 Dec 2022 10:39:46 +0000 (11:39 +0100)]
[Xtensa 2/10] Add Xtensa ELF definitions

Add file with Xtensa ELF relocations. Add Xtensa support to ELF.h,
ELFObject.h and ELFYAML.cpp. Add simple test of Xtensa ELF representation in YAML.

Differential Revision: https://reviews.llvm.org/D64827

2 years ago[Xtensa 1/10] Recognize Xtensa in triple parsing code
Andrei Safronov [Mon, 26 Dec 2022 10:37:28 +0000 (11:37 +0100)]
[Xtensa 1/10] Recognize Xtensa in triple parsing code

I'm sharing initial set of patches that adds LLVM backend for Xtensa architecture.
It is based on this LLVM fork https://github.com/espressif/llvm-xtensa.
I prepared patches by similar way like it was already successfully done for RISCV, i.e. incrementally add an initial MC layer for Xtensa by small chunks which could be reviewable.

Differential Revision: https://reviews.llvm.org/D64826

2 years ago[LoongArch][test] Regenerate checks for the ghc-cc.ll test case
WANG Xuerui [Mon, 26 Dec 2022 12:15:55 +0000 (20:15 +0800)]
[LoongArch][test] Regenerate checks for the ghc-cc.ll test case

Seems the codegen was stale (the extra `ret` after the tail call should
not be there, and indeed it is not emitted by the current code), plus
the whitespaces are different from the update_llc_test_checks.py style.
Simply regenerate it to fix the test failure.

Differential Revision: https://reviews.llvm.org/D140670

2 years ago[LoongArch] Add GHC Calling Convention
Lin Runze [Mon, 26 Dec 2022 10:37:09 +0000 (18:37 +0800)]
[LoongArch] Add GHC Calling Convention

This is modeled after [[ https://reviews.llvm.org/D89788 | the RISCV GHC calling convention]]
and matches [[ https://gitlab.haskell.org/ghc/ghc/-/merge_requests/9292 | the corresponding GHC change ]].

Reviewed By: xen0n, wangleiat

Differential Revision: https://reviews.llvm.org/D137495

2 years ago[clang] [OpenMP] Test amdgcn_openmp_device_math_c.c test on 32-bit platforms
Michał Górny [Mon, 26 Dec 2022 09:21:44 +0000 (10:21 +0100)]
[clang] [OpenMP] Test amdgcn_openmp_device_math_c.c test on 32-bit platforms

Explicitly pass triple to the test compiler to prevent failure when
the host triple is 32-bit.  This is the same solution
as f74e3d2f81d2aae47d6032fc1d23114460d48a37, thanks to Joseph Huber
for it.

2 years ago[Test] Update inverse test for turn-to-invariant to what they meant to be
Max Kazantsev [Mon, 26 Dec 2022 08:58:18 +0000 (15:58 +0700)]
[Test] Update inverse test for turn-to-invariant to what they meant to be

They were supposed to test inverted branches with OR condition, not AND.
Fixed this now.

2 years ago[IndVars][NFC] Factor out condition creation in optimizeLoopExitWithUnknownExitCount
Max Kazantsev [Mon, 26 Dec 2022 08:00:27 +0000 (15:00 +0700)]
[IndVars][NFC] Factor out condition creation in optimizeLoopExitWithUnknownExitCount

This is a preparation step to support optimization of conditions that are not immediately ICmp.

2 years ago[Test] Add test showing potential conflict b/w AND elimination and IV widening
Max Kazantsev [Mon, 26 Dec 2022 07:37:49 +0000 (14:37 +0700)]
[Test] Add test showing potential conflict b/w AND elimination and IV widening

2 years agoRevert "[clang] Use a StringRef instead of a raw char pointer to store builtin and...
Vitaly Buka [Mon, 26 Dec 2022 06:54:27 +0000 (22:54 -0800)]
Revert "[clang] Use a StringRef instead of a raw char pointer to store builtin and call information"
Revert "Fix lldb option handling since e953ae5bbc313fd0cc980ce021d487e5b5199ea4 (part 2)"
Revert "Fix lldb option handling since e953ae5bbc313fd0cc980ce021d487e5b5199ea4"

GCC build hangs on this bot https://lab.llvm.org/buildbot/#/builders/37/builds/19104
compiling CMakeFiles/obj.clangBasic.dir/Targets/AArch64.cpp.d

The bot uses GNU 11.3.0, but I can reproduce locally with gcc (Debian 12.2.0-3) 12.2.0.

This reverts commit caa713559bd38f337d7d35de35686775e8fb5175.
This reverts commit 06b90e2e9c991e211fecc97948e533320a825470.
This reverts commit e953ae5bbc313fd0cc980ce021d487e5b5199ea4.

2 years ago[NFC][Clang][RISCV] Reduce for-loop with SmallVector utility
eopXD [Sun, 25 Dec 2022 15:38:01 +0000 (07:38 -0800)]
[NFC][Clang][RISCV] Reduce for-loop with SmallVector utility

As topic, this commit reduces the for-loops with utilities of SmallVector.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D140661

2 years ago[Clang] Add `nvptx-arch` tool to query installed NVIDIA GPUs
Joseph Huber [Tue, 20 Dec 2022 22:02:07 +0000 (16:02 -0600)]
[Clang] Add `nvptx-arch` tool to query installed NVIDIA GPUs

We already have a tool called `amdgpu-arch` which returns the GPUs on
the system. This is used to determine the default architecture when
doing offloading. This patch introduces a similar tool `nvptx-arch`.
Right now we use the detected GPU at compile time. This is unhelpful
when building on a login node and moving execution to a compute node for
example. This will allow us to better choose a default architecture when
targeting NVPTX. Also we can probably use this with CMake's `native`
setting for CUDA now.

CUDA since 11.6 provides `__nvcc_device_query` which has a similar
function but it is probably better to define this locally if we want to
depend on it in clang.

Reviewed By: tianshilei1992

Differential Revision: https://reviews.llvm.org/D140433

2 years ago[RISCV] Implement assembler support for XTHeadVdot
Jojo R [Wed, 9 Nov 2022 08:17:20 +0000 (16:17 +0800)]
[RISCV] Implement assembler support for XTHeadVdot

This patch implements the T-Head vendor extensions (XTHeadVdot),
which is documented here, it's based on standard vector extension v1.0:
  https://github.com/T-head-Semi/thead-extension-spec

2 years ago[NFC] Elminate some needless nested-map complexity.
Owen Anderson [Sat, 24 Dec 2022 05:03:23 +0000 (22:03 -0700)]
[NFC] Elminate some needless nested-map complexity.

Reviewed By: nikic

Differential Revision: https://reviews.llvm.org/D140648

2 years agoRemove workaround for libstdc++ 4.8.
Owen Anderson [Sun, 25 Dec 2022 04:46:05 +0000 (21:46 -0700)]
Remove workaround for libstdc++ 4.8.

Reviewed By: nikic

Differential Revision: https://reviews.llvm.org/D140656

2 years ago[DebugInfo] make DW_LANG_C11 respect -gstrict-dwarf
Chen Zheng [Thu, 22 Dec 2022 10:57:11 +0000 (05:57 -0500)]
[DebugInfo] make DW_LANG_C11 respect -gstrict-dwarf

Reviewed By: dblaikie

Differential Revision: https://reviews.llvm.org/D140544

2 years ago[compiler-rt] Fix build errors when using gcc on LoongArch
Youling Tang [Mon, 26 Dec 2022 01:33:54 +0000 (09:33 +0800)]
[compiler-rt] Fix build errors when using gcc on LoongArch

- GCC does not recognize $fcsr0, uses $r0 instead.
- GCC does not implement __builtin_thread_pointer, which can be
  obtained directly through $tp.

Reviewed By: SixWeining

Differential Revision: https://reviews.llvm.org/D140545

2 years ago[NFC][DAGCombiner] `canCombineShuffleToAnyExtendVectorInreg()`: take matcher as callback
Roman Lebedev [Mon, 26 Dec 2022 00:56:58 +0000 (03:56 +0300)]
[NFC][DAGCombiner] `canCombineShuffleToAnyExtendVectorInreg()`: take matcher as callback

2 years ago[NFC][DAG] `canCombineShuffleToAnyExtendVectorInreg()`: check for legal op before...
Roman Lebedev [Sun, 25 Dec 2022 22:43:49 +0000 (01:43 +0300)]
[NFC][DAG] `canCombineShuffleToAnyExtendVectorInreg()`: check for legal op before matching

Likewise as with legal types check, might as well not match if won't use.

2 years ago[NFC][Analysis] Implement `getShuffleMaskWithWidestElts()` wrapper (+tests)
Roman Lebedev [Sun, 25 Dec 2022 21:41:12 +0000 (00:41 +0300)]
[NFC][Analysis] Implement `getShuffleMaskWithWidestElts()` wrapper (+tests)

It will be needed in an upcoming patch to implement some shuffle combining.

2 years ago[NFC][DAGCombiner] Extract `canCombineShuffleToAnyVectorExtendInReg()` helper
Roman Lebedev [Sun, 25 Dec 2022 21:03:45 +0000 (00:03 +0300)]
[NFC][DAGCombiner] Extract `canCombineShuffleToAnyVectorExtendInReg()` helper

Adding zero-ext support isn't as straight-forward, and it's easier
to to so in a new function, but this helper is useful there.
This does not change any existing behaviour.

2 years ago[NFC][DAG] `combineShuffleToVectorExtend()`: check that the type is legal first
Roman Lebedev [Sun, 25 Dec 2022 18:20:43 +0000 (21:20 +0300)]
[NFC][DAG] `combineShuffleToVectorExtend()`: check that the type is legal first

There is no point in doing any of the potentially-costly matching
if we will inevitably give up anyway.

2 years ago[SPIRV] support __spirv_Load/Store builtin functions
Ilia Diachkov [Sun, 25 Dec 2022 21:16:14 +0000 (00:16 +0300)]
[SPIRV] support __spirv_Load/Store builtin functions

The patch adds support for the builtin functions __spirv_Load and
__spirv_Store. One test is added to demonstrate the improvement.

Differential Revision: https://reviews.llvm.org/D140490

2 years ago[VPlan] Use VPBB in sinkScalarOperands directly. (NFC)
Florian Hahn [Sun, 25 Dec 2022 21:34:58 +0000 (21:34 +0000)]
[VPlan] Use VPBB in sinkScalarOperands directly. (NFC)

Suggested by @Ayal in D139790.

2 years ago[DebugInfo] Fix: Variables that have no non-empty values being emitted when they...
Stephen Tozer [Tue, 20 Dec 2022 10:01:56 +0000 (10:01 +0000)]
[DebugInfo] Fix: Variables that have no non-empty values being emitted when they have a DBG_VALUE_LIST

This patch fixes a simple bug where `DbgValueHistoryMap::hasNonEmptyLocation` was incorrectly handling DBG_VALUE_LIST instructions, treating empty values as non-empty, causing empty variables to be emitted into DWARF.

Reviewed By: Orlando

Differential Revision: https://reviews.llvm.org/D133925

2 years ago[gn build] Port f0756e086010
LLVM GN Syncbot [Sun, 25 Dec 2022 21:05:12 +0000 (21:05 +0000)]
[gn build] Port f0756e086010

2 years agoRevert "[clang-format] Add an option to format integer literal separators"
Vitaly Buka [Sun, 25 Dec 2022 20:49:35 +0000 (12:49 -0800)]
Revert "[clang-format] Add an option to format integer literal separators"
Revert "[clang-format] Disable FixRanges in IntegerLiteralSeparatorTest"

Breaks buildbots, details in https://reviews.llvm.org/D140543

This reverts commit 879bd9146a2c9ea395abd7c1ebd0f76f414a4967.
This reverts commit 46c94e5067b5f396c24bb950505c79bc819bd4b8.

2 years agoRevert "[RISCV] Enable the LocalStackSlotAllocation pass support."
Craig Topper [Sun, 25 Dec 2022 20:40:53 +0000 (12:40 -0800)]
Revert "[RISCV] Enable the LocalStackSlotAllocation pass support."

This reverts commit 180397cdded67a8fdf56f92a0b70d32f0dac8af6.

This seems to cause llvm-testsuite failures.

2 years ago[libc++] Implement constexpr {isfinite, isinf, isnan, isnormal}
Nikolas Klauser [Sun, 18 Dec 2022 23:02:44 +0000 (00:02 +0100)]
[libc++] Implement constexpr {isfinite, isinf, isnan, isnormal}

This starts implementing P0533

Reviewed By: Mordante, #libc

Spies: libcxx-commits

Differential Revision: https://reviews.llvm.org/D140277

2 years ago[OpenMP] Fix test on 32-bit platforms
Joseph Huber [Sun, 25 Dec 2022 15:47:04 +0000 (09:47 -0600)]
[OpenMP] Fix test on 32-bit platforms

Summary:
This test didn't specify the triple so it defaulted to the user's, if
this was 32-bit then it failed due to a diagnostic message.

2 years ago[NFC][ADT] Rename StringMapEntry *Create() into StringMapEntry *create.
Alexey Lapshin [Sun, 25 Dec 2022 14:25:18 +0000 (15:25 +0100)]
[NFC][ADT] Rename StringMapEntry *Create() into StringMapEntry *create.

2 years ago[NFC][SupportTests] Adjust `UnsignedDivideUsingMagic()` for readability
Roman Lebedev [Sun, 25 Dec 2022 13:43:20 +0000 (16:43 +0300)]
[NFC][SupportTests] Adjust `UnsignedDivideUsingMagic()` for readability

2 years ago[IPSCCP] Enable specialization of functions.
Alexandros Lamprineas [Sun, 25 Dec 2022 08:05:21 +0000 (10:05 +0200)]
[IPSCCP] Enable specialization of functions.

This patch enables Function Specialization by default at all
optimization levels except Os, Oz.

Compilation Time Overhead:
--------------------------
Measured the Instruction Count increase (Geomean) for CTMark from
the llvm-testsuite as in https://llvm-compile-time-tracker.com.
 * {-O3, Non-LTO}: +0.136% Instruction Count
 * {-O3, LTO}: +0.346% Instruction Count

Performance Uplift:
-------------------
Measured +9.121% score increase for 505.mcf_r from SPEC Int 2017
(Tested on Neoverse N1 with -O3 + LTO)

Correctness Testing:
--------------------
 * Passes bootstrap Clang with ASAN + LTO + FuncSpec aggressive options:
   { MaxClonesThreshold=10,
     SmallFunctionThreshold=10,
     AvgLoopIterationCount=30,
     SpecializeOnAddresses=true,
     EnableSpecializationForLiteralConstant=true,
     FuncSpecializationMaxIters=10 }
 * Builds Chromium and passes its unittests with the above options + ThinLTO.

For more info please refer to
https://discourse.llvm.org/t/rfc-should-we-enable-function-specialization/61518

Differential Revision: https://reviews.llvm.org/D140210

2 years ago[Doc] Replace PYTHON_EXECUTABLE with Python3_EXECUTABLE
eopXD [Sat, 24 Dec 2022 13:12:35 +0000 (05:12 -0800)]
[Doc] Replace PYTHON_EXECUTABLE with Python3_EXECUTABLE

As topic, the variable to specify the python executable now should be this.
This is probably something that was left out in D78762.

Reviewed By: JDevlieghere

Differential Revision: https://reviews.llvm.org/D140652

2 years ago[gn build] Port 46c94e5067b5
LLVM GN Syncbot [Sun, 25 Dec 2022 02:55:37 +0000 (02:55 +0000)]
[gn build] Port 46c94e5067b5

2 years ago[gn build] Port 066b492b747a
LLVM GN Syncbot [Sun, 25 Dec 2022 02:55:36 +0000 (02:55 +0000)]
[gn build] Port 066b492b747a

2 years ago[clang-format] Disable FixRanges in IntegerLiteralSeparatorTest
Owen Pan [Sun, 25 Dec 2022 02:47:23 +0000 (18:47 -0800)]
[clang-format] Disable FixRanges in IntegerLiteralSeparatorTest

The FixRanges unit test from 46c94e5067b5 breaks the build bots.
Disable it for now.

2 years agoAdds support for GOT relocations to i386/ELF backend
Kshitij Jain [Mon, 19 Dec 2022 04:07:42 +0000 (04:07 +0000)]
Adds support for GOT relocations to i386/ELF backend

This CR adds support for GOT relocations to the JITLink i386/ELF backend.

Reviewed By: lhames

Differential Revision: https://reviews.llvm.org/D140279

2 years agoUpdates and adds tests for i386/ELF JITLink backend
Kshitij Jain [Sun, 20 Nov 2022 04:09:29 +0000 (04:09 +0000)]
Updates and adds tests for i386/ELF JITLink backend

This CR modifies the existing 32 bit pcrel relocation test to
include the case when the relocation target might be present at
a smaller address than the address of the location that needs to be
patched.

Additionally, it adds a test for 16 bit absolute relocation.

Reviewed By: sunho

Differential Revision: https://reviews.llvm.org/D138372

2 years ago[NFC] Add exhaustive test coverage for `{Un}signedDivisionByConstantInfo`
Roman Lebedev [Sat, 24 Dec 2022 21:48:05 +0000 (00:48 +0300)]
[NFC] Add exhaustive test coverage for `{Un}signedDivisionByConstantInfo`

Use this wrapper if you want to try brute-forcing wider bit widths:
https://godbolt.org/z/3xGzTM881

I've brute-forced i16 for both signed and unsigned, and we're all good.
As mentioned in https://reviews.llvm.org/D140636

2 years ago[clang-format] Add an option to format integer literal separators
Owen Pan [Thu, 22 Dec 2022 09:21:17 +0000 (01:21 -0800)]
[clang-format] Add an option to format integer literal separators

Closes #58949.

Differential Revision: https://reviews.llvm.org/D140543

2 years ago[ConstraintElim] Add missing CHECK lines.
Florian Hahn [Sat, 24 Dec 2022 21:52:28 +0000 (21:52 +0000)]
[ConstraintElim] Add missing CHECK lines.

2 years ago[ConstraintElim] Add extra test with chained GEPs without inbounds.
Florian Hahn [Sat, 24 Dec 2022 21:51:26 +0000 (21:51 +0000)]
[ConstraintElim] Add extra test with chained GEPs without inbounds.

2 years ago[ConstraintElim] Convert tests to use opaque pointers (NFC).
Florian Hahn [Sat, 24 Dec 2022 21:46:46 +0000 (21:46 +0000)]
[ConstraintElim] Convert tests to use opaque pointers (NFC).

2 years ago[LV] Use SCEV to check if the trip count <= VF * UF.
Florian Hahn [Sat, 24 Dec 2022 18:33:36 +0000 (18:33 +0000)]
[LV] Use SCEV to check if the trip count <= VF * UF.

Just comparing constant trip counts causes LV to miss cases where the
vector loop body only executes once.

The motivation for this is to remove the need for unrolling to remove
vector loop back-edges, if the body only executes once in more cases.

Reviewed By: Ayal

Differential Revision: https://reviews.llvm.org/D133017

2 years ago[NFC][Codegen][X86] Autogenerate check lines in shift-i256.ll
Roman Lebedev [Sat, 24 Dec 2022 16:18:50 +0000 (19:18 +0300)]
[NFC][Codegen][X86] Autogenerate check lines in shift-i256.ll

2 years ago[NFC][Codegen][AVR] Make shift.ll autogenerate-able
Roman Lebedev [Sat, 24 Dec 2022 16:12:27 +0000 (19:12 +0300)]
[NFC][Codegen][AVR] Make shift.ll autogenerate-able

2 years ago[NFC][Codegen] Add tests with oversized shifts by non-byte-multiple
Roman Lebedev [Sat, 24 Dec 2022 15:39:57 +0000 (18:39 +0300)]
[NFC][Codegen] Add tests with oversized shifts by non-byte-multiple

2 years ago[NFC][Codegen] Rename tests for oversized shifts by byte multiple
Roman Lebedev [Sat, 24 Dec 2022 15:39:48 +0000 (18:39 +0300)]
[NFC][Codegen] Rename tests for oversized shifts by byte multiple