Kazushi (Jam) Marukawa [Fri, 5 Jun 2020 13:43:29 +0000 (15:43 +0200)]
[VE] Support branch instructions in MC layer
Summary:
Add regression tests of asmparser, mccodeemitter, and disassembler for
branch instructions. In order to support them, we enhance asmparser
by adding splitting mnemonic mechanism, e.g. "bgt.l.t" into "b", "gt",
and ".l.t", and parsing mechanism for AS style memory addressing.
We also implment encoding and decoding mechanism for branch instructions.
Differential Revision: https://reviews.llvm.org/D81215
Simon Pilgrim [Fri, 5 Jun 2020 11:47:50 +0000 (12:47 +0100)]
SyncDependenceAnalysis.h - remove orphan method declarations. NFCI.
These have been there since the header was added by D51491 but afaict without an implementation, all functionality is actually in DivergenceAnalysis
Simon Pilgrim [Fri, 5 Jun 2020 11:17:32 +0000 (12:17 +0100)]
TargetLibraryInfo.h - reduce Triple.h include to forward declaration. NFC.
Move implicit include dependencies down to source files.
Simon Pilgrim [Fri, 5 Jun 2020 10:31:35 +0000 (11:31 +0100)]
ScopedNoAliasAA.h - remove unnecessary InstrTypes.h include. NFC.
Ties Stuij [Fri, 5 Jun 2020 13:26:51 +0000 (14:26 +0100)]
[clang][BFloat] Add create/set/get/dup intrinsics
Summary:
This patch is part of a series that adds support for the Bfloat16 extension of
the Armv8.6-a architecture, as detailed here:
https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-architecture-developments-armv8-6-a
The bfloat type and its properties are specified in the Arm Architecture
Reference Manual:
https://developer.arm.com/docs/ddi0487/latest/arm-architecture-reference-manual-armv8-for-armv8-a-architecture-profile
The following people contributed to this patch:
- Luke Cheeseman
- Momchil Velikov
- Luke Geeson
- Ties Stuij
- Mikhail Maltsev
Reviewers: t.p.northover, sdesmalen, fpetrogalli, LukeGeeson, stuij, labrinea
Reviewed By: labrinea
Subscribers: miyuki, dmgreen, labrinea, kristof.beyls, ilya-biryukov, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79710
Guillaume Chatelet [Thu, 4 Jun 2020 21:56:01 +0000 (21:56 +0000)]
[Alignment][NFC] Migrate CallingConv tablegen code
Summary:
We first migrate the generated code, more patches to come.
This patch is part of a series to introduce an Alignment type.
See this thread for context: http://lists.llvm.org/pipermail/llvm-dev/2019-July/133851.html
See this patch for the introduction of the type: https://reviews.llvm.org/D64790
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D81196
madhur13490 [Thu, 7 May 2020 17:27:32 +0000 (17:27 +0000)]
Fix a typo in AMDGPU docs
Reviewers: t-tye, arsenm
Reviewed By: arsenm
Subscribers: kzhuravl, jvesely, wdng, nhaehnle, yaxunl, dstuttard, tpr, kerbowa, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D81247
Louis Dionne [Thu, 4 Jun 2020 18:54:38 +0000 (14:54 -0400)]
[libc++] Link against libatomic when it is found
Before this patch, we tried detecting whether small atomics were available
without linking against libatomic. However, that's not really what we want
to know -- instead, we want to know what's required in order to support
atomics fully, which is to link against libatomic when it's provided.
That is both much simpler, and it doesn't suffer the problem that we would
not link against libatomic when small atomics didn't require it, which
lead to non-lockfree atomics never working.
Furthermore, because we understand that some platforms might not want to
(or be able to) ship non-lockfree atomics, we add that notion to the test
suite, independently of a potential extern library.
After this patch, we therefore:
(1) Link against libatomic when it is provided
(2) Independently detect whether non-lockfree atomics are supported in
the test suite, regardless of whether that means we're linking against
an external library or not (which is an implementation detail).
Differential Revision: https://reviews.llvm.org/D81190
Ties Stuij [Fri, 5 Jun 2020 12:06:01 +0000 (13:06 +0100)]
[clang][BFloat] add NEON emitter for bfloat
Summary:
This patch adds the bfloat16_t struct typedefs (e.g. bfloat16x8x2_t) to
arm_neon.h
This patch is part of a series implementing the Bfloat16 extension of the
Armv8.6-a architecture, as detailed here:
https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-architecture-developments-armv8-6-a
The bfloat type, and its properties are specified in the Arm Architecture
Reference Manual:
https://developer.arm.com/docs/ddi0487/latest/arm-architecture-reference-manual-armv8-for-armv8-a-architecture-profile
The following people contributed to this patch:
- Luke Cheeseman
- Simon Tatham
- Ties Stuij
Reviewers: t.p.northover, fpetrogalli, sdesmalen, az, LukeGeeson
Reviewed By: fpetrogalli
Subscribers: SjoerdMeijer, LukeGeeson, pbarrio, mgorny, kristof.beyls, ilya-biryukov, MaskRay, jkorous, arphaman, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79708
HazemAbdelhafez [Fri, 5 Jun 2020 12:38:40 +0000 (08:38 -0400)]
[MLIR][SPIRV] Support flat, location, and noperspective decorations
Add support for flat, location, and noperspective decorations in the
serializer and deserializer to be able to process basic shader files
for graphics applications.
Differential Revision: https://reviews.llvm.org/D80837
Denis Antrushin [Fri, 5 Jun 2020 12:52:14 +0000 (15:52 +0300)]
Nicolas Vasilache [Fri, 5 Jun 2020 12:22:15 +0000 (08:22 -0400)]
[mlir][Vector] Move temporary alloc to top of the function alloca when lowering vector_transfers
Recently introduced allocation hoisting is quite conservative on the cases when it triggers.
This revision makes it such that the allocations for vector transfer lowerings are hoisted
to the top of the function.
This should be revisited in the context of parallelism and is a temporary workaround.
Differential Revision: https://reviews.llvm.org/D81253
Dineshkumar Bhaskaran [Fri, 5 Jun 2020 12:41:37 +0000 (12:41 +0000)]
[MsgPack] Added a convenience operator
Summary: Added "not equal to" operator for DocNode comparison
Reviewers: arsenm, scott.linder, saiislam
Reviewed By: saiislam
Subscribers: wdng, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D81250
Joachim Protze [Fri, 5 Jun 2020 12:29:22 +0000 (14:29 +0200)]
Remove mention of counter from Archer readme
The feature was removed before upstreaming Archer, so the documentation is wrong
Ties Stuij [Fri, 5 Jun 2020 09:35:59 +0000 (10:35 +0100)]
[ARM] Add poly64_t on AArch32.
Summary:
The poly64 types are guarded with ifdefs for AArch64 only. This is wrong. This
was also incorrectly documented in the ACLE spec, but this has been rectified in
the latest release. See paragraph 13.1.2 "Vector data types":
https://developer.arm.com/docs/101028/latest
This patch was written by Alexandros Lamprineas.
Reviewers: ostannard, sdesmalen, fpetrogalli, labrinea, t.p.northover, LukeGeeson
Reviewed By: ostannard
Subscribers: pbarrio, LukeGeeson, kristof.beyls, danielkiss, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79711
Denis Antrushin [Thu, 4 Jun 2020 16:11:13 +0000 (19:11 +0300)]
[TargetLowering][NFC] More efficient emitPatchpoint().
Current implementation of emitPatchpoint() is very inefficient:
for every FrameIndex operand if creates new MachineInstr with
that operand expanded and all other copied as is.
Since PATCHPOINT/STATEPOINT instructions may have *a lot* of
FrameIndex operands, we end up creating and erasing many
machine instructions. But we can do it in single pass, with only
one new machine instruction generated.
Reviewed By: reames
Differential Revision: https://reviews.llvm.org/D81181
Tom Weaver [Fri, 5 Jun 2020 11:53:56 +0000 (12:53 +0100)]
[Dexter] Add DexLimitSteps command and ConditionalController
* Adds DexLimitSteps Command.
* Add ConditionalController, a new DebuggerController type.
* 5 regression tests
* documentation
* recommit, fixed accidental adding of unnecessary file
Reviewers: jmorse
Differential Revision: https://reviews.llvm.org/D79786
Stefan Pintilie [Fri, 5 Jun 2020 11:37:21 +0000 (06:37 -0500)]
[PowerPC] Remove extra nop after notoc call
Calls that are marked as @notoc do not require the extra nop after the call
for the TOC restore.
Differential Revision: https://reviews.llvm.org/D81081
Nicolas Vasilache [Fri, 5 Jun 2020 11:38:55 +0000 (07:38 -0400)]
[mlir][Linalg] Add missing CMake dependency on SCFTransforms
Kerry McLaughlin [Fri, 5 Jun 2020 10:11:50 +0000 (11:11 +0100)]
[CodeGen][SVE] Legalisation of extends with scalable types
Summary:
This patch adds legalisation of extensions where the operand
of the extend is a legal scalable type but the result is not.
EXTRACT_SUBVECTOR is used to split the result, before
being replaced by target-specific [S|U]UNPK[HI|LO] operations.
For example:
```
zext <vscale x 16 x i8> %a to <vscale x 16 x i16>
```
should emit:
```
uunpklo z2.h, z0.b
uunpkhi z1.h, z0.b
```
Reviewers: sdesmalen, efriedma, david-arm
Reviewed By: efriedma
Subscribers: tschuett, hiraditya, rkruppe, psnobl, huihuiz, cfe-commits, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D79587
Kadir Cetinkaya [Fri, 5 Jun 2020 11:02:35 +0000 (13:02 +0200)]
Revert "[InstCombine] Simplify compare of Phi with constant inputs against a constant"
This reverts commit
16b7eb6dd1247dbe322061d33636a054d6c954dc.
Breaks build bots, see
http://lab.llvm.org:8011/builders/clang-x86_64-debian-fast/builds/29888
for an example.
Kadir Cetinkaya [Fri, 5 Jun 2020 11:01:08 +0000 (13:01 +0200)]
[clang][test] Put output into temp directory
To unbreak builds that happen on a read-only directory
Stefan Pintilie [Fri, 5 Jun 2020 10:47:20 +0000 (05:47 -0500)]
[PowerPC][NFC] Add more PC Relative tests
Modify the pcrel.ll test file to add more testing for PC Relative.
Nicolas Vasilache [Fri, 5 Jun 2020 10:35:46 +0000 (06:35 -0400)]
[mlir][Linalg] Add a hoistRedundantVectorTransfers helper function
This revision adds a helper function to hoist vector.transfer_read /
vector.transfer_write pairs out of immediately enclosing scf::ForOp
iteratively, if the following conditions are true:
1. The 2 ops access the same memref with the same indices.
2. All operands are invariant under the enclosing scf::ForOp.
3. No uses of the memref either dominate the transfer_read or are
dominated by the transfer_write (i.e. no aliasing between the write and
the read across the loop)
To improve hoisting opportunities, call the `moveLoopInvariantCode` helper
function on the candidate loop above which to hoist. Hoisting the transfers
results in scf::ForOp yielding the value that originally transited through
memory.
This revision additionally exposes `moveLoopInvariantCode` as a helper in
LoopUtils.h and updates SliceAnalysis to support return scf::For values and
allow hoisting across multiple scf::ForOps.
Differential Revision: https://reviews.llvm.org/D81199
Xing GUO [Fri, 5 Jun 2020 10:48:27 +0000 (18:48 +0800)]
[ObjectYAML][test] Reuse the chack tag "DWARF-HEADER". NFC.
Kadir Cetinkaya [Fri, 5 Jun 2020 10:25:09 +0000 (12:25 +0200)]
[lldb] Handle new BFloat16 type
Max Kazantsev [Fri, 5 Jun 2020 09:43:45 +0000 (16:43 +0700)]
[InstCombine] Simplify compare of Phi with constant inputs against a constant
We can simplify
```
icmp <pred> phi(C1, C2, ...), C
```
with
```
phi(icmp(C1, C), icmp(C2, C), ...)
```
provided that all comparison of constants are constants themselves.
Differential Revision: https://reviews.llvm.org/D81151
Reviewed By: lebedev.ri
Kadir Cetinkaya [Fri, 5 Jun 2020 10:01:03 +0000 (12:01 +0200)]
[mlir] Make PassManager.h self-contained
Forward declaring llvm::errs is not enough, as it is used as a default
parameter with a type that references the base class. So the class
hierarchy must be visible.
Jan Kratochvil [Fri, 5 Jun 2020 09:59:26 +0000 (11:59 +0200)]
[lldb] Fix UBSan regression in GetSLEB128
It regressed recently by my: https://reviews.llvm.org/D81119
Kazushi (Jam) Marukawa [Fri, 5 Jun 2020 09:55:47 +0000 (11:55 +0200)]
[VE] Support fixed-point operation instructions in MC layer
Summary:
Add regression tests of asmparser, mccodeemitter, and disassembler for
fixed-point operation instructions. In order to support them, we add
MImm parser to asmparser. Also add a new MPD instruction which is one
of multiply instructions.
Differential Revision: https://reviews.llvm.org/D81207
Daniel Grumberg [Fri, 5 Jun 2020 09:46:38 +0000 (10:46 +0100)]
NFC: Formatting fix in TableGEn/OptParserEmitter.cpp
Simon Pilgrim [Fri, 5 Jun 2020 09:45:42 +0000 (10:45 +0100)]
MemorySSAUpdater.h - reduce unnecessary includes to forward declarations. NFC.
Remove unnecessary MemoryAccess forward declaration as its already included from MemorySSA.h
Move implicit include dependencies down to source files.
Simon Pilgrim [Fri, 5 Jun 2020 09:41:31 +0000 (10:41 +0100)]
Fix MSVC "not all control paths return a value" warning. NFC.
Add llvm_unreachable after switch statement for TestLanguage enum
Ties Stuij [Thu, 4 Jun 2020 23:20:02 +0000 (00:20 +0100)]
[ARM] Add __bf16 as new Bfloat16 C Type
Summary:
This patch upstreams support for a new storage only bfloat16 C type.
This type is used to implement primitive support for bfloat16 data, in
line with the Bfloat16 extension of the Armv8.6-a architecture, as
detailed here:
https://community.arm.com/developer/ip-products/processors/b/processors-ip-blog/posts/arm-architecture-developments-armv8-6-a
The bfloat type, and its properties are specified in the Arm Architecture
Reference Manual:
https://developer.arm.com/docs/ddi0487/latest/arm-architecture-reference-manual-armv8-for-armv8-a-architecture-profile
In detail this patch:
- introduces an opaque, storage-only C-type __bf16, which introduces a new bfloat IR type.
This is part of a patch series, starting with command-line and Bfloat16
assembly support. The subsequent patches will upstream intrinsics
support for BFloat16, followed by Matrix Multiplication and the
remaining Virtualization features of the armv8.6-a architecture.
The following people contributed to this patch:
- Luke Cheeseman
- Momchil Velikov
- Alexandros Lamprineas
- Luke Geeson
- Simon Tatham
- Ties Stuij
Reviewers: SjoerdMeijer, rjmccall, rsmith, liutianle, RKSimon, craig.topper, jfb, LukeGeeson, fpetrogalli
Reviewed By: SjoerdMeijer
Subscribers: labrinea, majnemer, asmith, dexonsmith, kristof.beyls, arphaman, danielkiss, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D76077
Alexander Belyaev [Fri, 5 Jun 2020 09:16:53 +0000 (11:16 +0200)]
[Mlir] Implement printer, parser, verifier and builder for shape.reduce.
Differential Revision: https://reviews.llvm.org/D81186
Roman Lebedev [Fri, 5 Jun 2020 09:18:15 +0000 (12:18 +0300)]
[NFC][SCEV] Add test with 'or' with no common bits set
Sam Parker [Fri, 5 Jun 2020 09:09:56 +0000 (10:09 +0100)]
[CostModel] Unify getMemoryOpCost
Use getMemoryOpCost from the generic implementation of getUserCost
and have getInstructionThroughput return the result of that for loads
and stores.
This also means that the X86 implementation of getUserCost can be
removed with the functionality folded into its getMemoryOpCost.
Differential Revision: https://reviews.llvm.org/D80984
Tres Popp [Fri, 5 Jun 2020 08:12:00 +0000 (10:12 +0200)]
[mlir] NFC formatting cleanup.
Tres Popp [Wed, 20 May 2020 13:56:40 +0000 (15:56 +0200)]
[mlir] Canonicalization of shape.assuming
Summary:
This will inline the region to a shape.assuming in the case that the
input witness is found to be statically true.
Differential Revision: https://reviews.llvm.org/D80302
Tres Popp [Wed, 20 May 2020 13:54:57 +0000 (15:54 +0200)]
[mlir] Folding and canonicalization of shape.cstr_eq
In the case of all inputs being constant and equal, cstr_eq will be
replaced with a true_witness.
Differential Revision: https://reviews.llvm.org/D80303
Tres Popp [Wed, 20 May 2020 13:52:55 +0000 (15:52 +0200)]
[mlir] Canonicalization and folding of shape.cstr_broadcastable
This allows replacing of this op with a true witness in the case of both
inputs being const_shapes and being found to be broadcastable.
Differential Revision: https://reviews.llvm.org/D80304
Tres Popp [Wed, 20 May 2020 14:12:57 +0000 (16:12 +0200)]
[mlir] Add folding for shape.any
If any input to shape.any is a const_shape, shape.any can be replaced
with that input.
Differential Revision: https://reviews.llvm.org/D80305
Tres Popp [Thu, 4 Jun 2020 10:43:13 +0000 (12:43 +0200)]
[mlir] Folding of shape.assuming_all
This allows assuming_all to be replaced when all inputs are known to be
statically passing witnesses.
Differential Revision: https://reviews.llvm.org/D80306
Tres Popp [Wed, 20 May 2020 13:56:12 +0000 (15:56 +0200)]
[mlir] Add a shape op that returns a constant witness
This will later be used during canonicalization and folding steps to replace
statically known passing constraints.
Differential Revision: https://reviews.llvm.org/D80307
Alexander Belyaev [Thu, 4 Jun 2020 20:40:48 +0000 (22:40 +0200)]
[Mlir] Add assembly format for `shape.mul`.
Differential Revision: https://reviews.llvm.org/D81194
Roman Lebedev [Fri, 5 Jun 2020 08:23:16 +0000 (11:23 +0300)]
[NFC][SCEV] Some tests for shifts by bitwidth-2/bitwidth-1 w/ no-wrap flags
Hans Wennborg [Fri, 5 Jun 2020 08:32:40 +0000 (10:32 +0200)]
Fix gcc5 build of ASTMatchersTest.h
After a180d54 the build was failing with:
In file included from /work/llvm.monorepo/clang/unittests/ASTMatchers/ASTMatchersNodeTest.cpp:9:0:
/work/llvm.monorepo/clang/unittests/ASTMatchers/ASTMatchersTest.h:
In function ‘llvm::ArrayRef<clang::TestLanguage> clang::ast_matchers::langCxx11OrLater()’:
/work/llvm.monorepo/clang/unittests/ASTMatchers/ASTMatchersTest.h:64:10:
error: could not convert ‘(const clang::TestLanguage*)(& Result)’ from
‘const clang::TestLanguage*’ to ‘llvm::ArrayRef<clang::TestLanguage>’
return Result;
^
Haojian Wu [Wed, 3 Jun 2020 08:34:05 +0000 (10:34 +0200)]
[clangd] Populate the parse options to CodeCompletion/SignatureHelp.
Summary: A followup of D79938.
Reviewers: sammccall
Subscribers: ilya-biryukov, MaskRay, jkorous, arphaman, kadircet, usaxena95, cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81066
Kazushi (Jam) Marukawa [Fri, 5 Jun 2020 08:03:10 +0000 (10:03 +0200)]
[VE] Add AND/OR/XOR regression tests
Summary:
Add AND/OR/XOR regression tests and separate bit-op tests from exisiting
tests.
Differential Revision: https://reviews.llvm.org/D81206
Vitaly Buka [Fri, 5 Jun 2020 06:52:38 +0000 (23:52 -0700)]
[StackSafety,NFC] Switch tests to aarch64
Uday Bondhugula [Mon, 25 May 2020 12:10:44 +0000 (17:40 +0530)]
[MLIR] Update linalg.conv lowering to use affine load in the absence of padding
Update linalg to affine lowering for convop to use affine load for input
whenever there is no padding. It had always been using std.loads because
max in index functions (needed for non-zero padding if not materializing
zeros) couldn't be represented in the non-zero padding cases.
In the future, the non-zero padding case could also be made to use
affine - either by materializing or using affine.execute_region. The
latter approach will not impact the scf/std output obtained after
lowering out affine.
Differential Revision: https://reviews.llvm.org/D81191
Jonas Devlieghere [Fri, 5 Jun 2020 06:34:05 +0000 (23:34 -0700)]
Revert "Set the captures on a CXXRecordDecl representing a lambda closure type"
This reverts commit
c13dd74e311d2ac70dd3ea663d800307d1aa5b6b.
Jonas Devlieghere [Fri, 5 Jun 2020 06:25:01 +0000 (23:25 -0700)]
Revert "PR46209: properly determine whether a copy assignment operator is"
This reverts commit
c57f8a3a20540fcf9fbf98c0a73f381ec32fce2a.
Jan Kratochvil [Fri, 5 Jun 2020 06:28:06 +0000 (08:28 +0200)]
[nfc] [lldb] clang-format #include files order
Fangrui Song [Fri, 5 Jun 2020 06:25:19 +0000 (23:25 -0700)]
[Driver] Migrate some -f/-fno options to use OptInFFlag and OptOutFFlag
Also assign OptInFFlag and OptOutFFlag to f_Group.
Max Kazantsev [Fri, 5 Jun 2020 05:00:44 +0000 (12:00 +0700)]
Revert "[InstCombine][NFC] Factor out constant check"
This reverts commit
9bdb91889020b3e61cba26adb1b9c64a24c09f95.
This refactoring proved to not be useful.
Xing GUO [Fri, 5 Jun 2020 04:15:34 +0000 (12:15 +0800)]
[DWARFYAML][debug_aranges] Replace InitialLength with Format and Length.
This patch addresses the comment in [D80972](https://reviews.llvm.org/D80972#inline-744217).
Before this patch, the initial length field of .debug_aranges section should be declared as:
```
## 32-bit DWARF
debug_aranges:
- Length:
TotalLength: 0x20
Version: 2
...
## 64-bit DWARF
debug_aranges:
- Length:
TotalLength: 0xffffffff
TotalLength64: 0x20
Version: 2
...
```
After this patch:
```
## 32-bit DWARF
debug_aranges:
- [[Format: DWARF32]] ## Optional
Length: 0x20
Version: 2
...
## 64-bit DWARF
debug_aranges:
- Format: DWARF64
Length: 0x20
Version: 2
```
Current implementation of generating DWARF64 .debug_aranges section is buggy. A follow-up patch will improve it and add test cases for DWARF64.
Reviewed By: jhenderson
Differential Revision: https://reviews.llvm.org/D81063
Vitaly Buka [Fri, 5 Jun 2020 02:44:42 +0000 (19:44 -0700)]
[StackSafety,NFC] Ignore callee declarations
It's going to fail FunctionInfo lookup anyway.
Petr Hosek [Fri, 5 Jun 2020 03:18:35 +0000 (20:18 -0700)]
[AddressSanitizer] Don't use weak linkage for __{start,stop}_asan_globals
It should not be necessary to use weak linkage for these. Doing so
implies interposablity and thus PIC generates indirections and
dynamic relocations, which are unnecessary and suboptimal. Aside
from this, ASan instrumentation never introduces GOT indirection
relocations where there were none before--only new absolute relocs
in RELRO sections for metadata, which are less problematic for
special linkage situations that take pains to avoid GOT generation.
Patch By: mcgrathr
Differential Revision: https://reviews.llvm.org/D80605
Fangrui Song [Fri, 5 Jun 2020 02:28:54 +0000 (19:28 -0700)]
[Driver] Migrate some -f/-fno options to use OptInFFlag and OptOutFFlag
Richard Smith [Fri, 5 Jun 2020 02:16:05 +0000 (19:16 -0700)]
PR46209: properly determine whether a copy assignment operator is
trivial.
We previously took a shortcut by assuming that if a subobject had a
trivial copy assignment operator (with a few side-conditions), we would
always invoke it, and could avoid going through overload resolution.
That turns out to not be correct in the presenve of ref-qualifiers (and
also won't be the case for copy-assignments with requires-clauses
either). Use the same logic for lazy declaration of copy-assignments
that we use for all other special member functions.
Richard Smith [Fri, 5 Jun 2020 02:15:22 +0000 (19:15 -0700)]
Set the captures on a CXXRecordDecl representing a lambda closure type
before marking it complete.
No functionality change intended.
Philip Reames [Fri, 5 Jun 2020 01:15:04 +0000 (18:15 -0700)]
[Statepoint] Migrate a few tests to gc-live bundle format and fix assert
The assert was missed in
0e7c7705, migrating the test revealed the problem.
Vedant Kumar [Thu, 28 May 2020 22:57:11 +0000 (15:57 -0700)]
[LiveDebugValues] Cache LexicalScopes::getMachineBasicBlocks, NFCI
Summary:
Cache the results from getMachineBasicBlocks in LexicalScopes to speed
up UserValueScopes::dominates queries. This replaces the caching done
in UserValueScopes. Compared to the old caching method, this reduces
memory traffic when a VarLoc is copied (e.g. when a VarLocMap grows),
and enables caching across basic blocks.
When compiling sqlite 3.5.7 (CTMark version), this patch reduces the
number of calls to getMachineBasicBlocks from 10,207 to 1,093. I also
measured a small compile-time reduction (~ 0.1% of total wall time, on
average, on my machine).
As a drive-by, I made the DebugLoc in UserValueScopes a const reference
to cut down on MetadataTracking traffic.
Reviewers: jmorse, Orlando, aprantl, nikic
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D80957
River Riddle [Wed, 3 Jun 2020 01:20:38 +0000 (18:20 -0700)]
[mlir] Refactor BoolAttr to be a special case of IntegerAttr
This simplifies a lot of handling of BoolAttr/IntegerAttr. For example, a lot of places currently have to handle both IntegerAttr and BoolAttr. In other places, a decision is made to pick one which can lead to surprising results for users. For example, DenseElementsAttr currently uses BoolAttr for i1 even if the user initialized it with an Array of i1 IntegerAttrs.
Differential Revision: https://reviews.llvm.org/D81047
Mircea Trofin [Wed, 3 Jun 2020 00:48:02 +0000 (17:48 -0700)]
[docs] Referenced llvm workflow in HowToAddABuilder
Reviewers: gkistanova, dblaikie
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D81046
Dan Gohman [Thu, 4 Jun 2020 22:55:08 +0000 (15:55 -0700)]
[WebAssembly] Fix a testcase to be independent of the sysroot default
As a followup to D62922, add a sysroot command-line option to this test
to ensure that the output is independent of any default sysroot options,
and adjust the reactor test to be more consistent with the command test.
Nicolas Vasilache [Thu, 4 Jun 2020 22:55:26 +0000 (18:55 -0400)]
[mlir][Linalg] Add a hoistViewAllocOps helper function
This revision adds a helper function to hoist alloc/dealloc pairs and
alloca op out of immediately enclosing scf::ForOp if both conditions are true:
1. all operands are defined outside the loop.
2. all uses are ViewLikeOp or DeallocOp.
This is now considered Linalg-specific and will be generalized on a per-need basis.
Differential Revision: https://reviews.llvm.org/D81152
Jan Korous [Thu, 4 Jun 2020 22:55:44 +0000 (15:55 -0700)]
[Analyzer][NoUncountedMembersChecker] Fix crash for C structs
Fixes https://bugs.llvm.org/show_bug.cgi?id=46177
Fixes second bug reported in https://bugs.llvm.org/show_bug.cgi?id=46142
Philip Reames [Thu, 4 Jun 2020 22:22:07 +0000 (15:22 -0700)]
[Statepoint] Switch RS4GC to using gc-live bundle form
Now that we have an operand based form for the GC arguments to a statepoint intrinsic, update RS4GC to use it and update tests to reflect. This is pretty straight forward. I nearly landed without review, but figured a second set of eyes didn't hurt.
Differential Revision: https://reviews.llvm.org/D81121
Petr Hosek [Thu, 4 Jun 2020 22:47:05 +0000 (15:47 -0700)]
[Fuchsia] Rely on linker switch rather than dead code ref for profile runtime
Follow the model used on Linux, where the clang driver passes the
linker a -u switch to force the profile runtime to be linked in,
rather than having every TU emit a dead function with a reference.
Differential Revision: https://reviews.llvm.org/D79835
Petr Hosek [Thu, 4 Jun 2020 22:44:10 +0000 (15:44 -0700)]
Revert "[Fuchsia] Rely on linker switch rather than dead code ref for profile runtime"
This reverts commit
d51054217403b47f452619e11318bd214749a845 since
it broke several bots.
Julian Lettner [Thu, 4 Jun 2020 21:48:42 +0000 (14:48 -0700)]
Make linter happy
Vedant Kumar [Thu, 4 Jun 2020 21:32:46 +0000 (14:32 -0700)]
[docs] HowToUpdateDebugInfo: Minor cleanups
- Change the reference to salvageDebugInfoOrUndef to salvageDebugInfo
(in accordance with https://reviews.llvm.org/D78369).
- Reorganize a few sections in preparation for an upcoming change that
attempts to specify rules for updating debug locations.
- Fix some intra-document links.
- Some spelling / wording fixes.
Yuanfang Chen [Tue, 2 Jun 2020 00:09:40 +0000 (17:09 -0700)]
[Docs] Add the entry for `Advanced builds` in UserGuide.rst
Also add a link to it from ThinLTO.rst.
Alexey Bataev [Thu, 4 Jun 2020 21:33:13 +0000 (17:33 -0400)]
Fix undefined behaviour when trying to deref nullptr.
Craig Topper [Thu, 4 Jun 2020 19:21:04 +0000 (12:21 -0700)]
[Reassociate] Teach ConvertShiftToMul to preserve nsw flag if the shift amount is not bitwidth - 1.
Multiply and shl have different signed overflow behavior in
some cases. But it looks like we should be ok as long as the
shift amount is less than bitwidth - 1.
Alive2: http://volta.cs.utah.edu:8080/z/MM4WZP
Differential Revision: https://reviews.llvm.org/D81189
Matt Arsenault [Thu, 4 Jun 2020 18:37:39 +0000 (14:37 -0400)]
AMDGPU: Fix overriding global FP atomic feature predicates
Global TableGen let override blocks are pretty dangerous and override
any local special cases. In this case, the broader HasFlatGlobalInsts
was overriding the more specific predicate for
FeatureAtomicFaddInsts. Make sure HasFlatGlobalInsts is implied by
FeatureAtomicFaddInsts, and make sure the right predicate is used.
One issue with independently setting the subtarget features on
incompatible targets is all of the encoding families do not define all
opcodes. This will hit an assert on gfx10 for example, since we set
the encoding independently based on the generation and not based on a
feature.
Matt Arsenault [Thu, 28 May 2020 00:12:11 +0000 (20:12 -0400)]
AMDGPU: Select strict_fmul
Matt Arsenault [Sat, 23 May 2020 17:56:02 +0000 (13:56 -0400)]
AMDGPU: Select strict_fma
Like with strict_fadd, the legalization is scalarizing the v4f16 when
it should split.
Matt Arsenault [Sat, 23 May 2020 14:41:40 +0000 (10:41 -0400)]
AMDGPU: Select strict_fadd
Diego Caballero [Thu, 4 Jun 2020 21:08:49 +0000 (14:08 -0700)]
[mlir] Add support for bf16 to StandardToLLVM conversion
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D81127
Matt Arsenault [Mon, 1 Jun 2020 18:04:02 +0000 (14:04 -0400)]
AMDGPU: Add test for fdiv nofpexcept preservation
This logically belongs with
89d48ccabe6a950369b2bd922b1d8e987b856ac7,
but this order was needed to avoid regressions before adding
mayRaiseFPExceptions to relevant instructions.
Matt Arsenault [Wed, 27 May 2020 01:56:53 +0000 (21:56 -0400)]
AMDGPU: Set mayRaiseFPException
This may be missing a few overrides to set it off still in some
special cases. Since the flags set during selection should now be
reliably preserved, this should not change codegen for non-strictfp
functions.
Sanjay Patel [Thu, 4 Jun 2020 21:25:22 +0000 (17:25 -0400)]
[InstCombine] avoid crashing on select-shuffle detection
As mentioned in the post-commit comments of D81013 -
the mask check API has to assume the shuffle is
not length-changing, but we have not ruled that out
in this code. Use the ShuffleVectorInst call instead.
Petr Hosek [Thu, 4 Jun 2020 21:24:33 +0000 (14:24 -0700)]
[Fuchsia] Rely on linker switch rather than dead code ref for profile runtime
Follow the model used on Linux, where the clang driver passes the
linker a -u switch to force the profile runtime to be linked in,
rather than having every TU emit a dead function with a reference.
Patch By: mcgrathr
Differential Revision: https://reviews.llvm.org/D79835
Shilei Tian [Thu, 4 Jun 2020 20:58:37 +0000 (16:58 -0400)]
[OpenMP] Improve D2D memcpy to use more efficient driver API
Summary:
In current implementation, D2D memcpy is first to copy data back to host and then
copy from host to device. This is very efficient if the device supports D2D
memcpy, like CUDA.
In this patch, D2D memcpy will first try to use native supported driver API. If
it fails, fall back to original way. It is worth noting that D2D memcpy in this
scenerio contains two ideas:
- Same devices: this is the D2D memcpy in the CUDA context.
- Different devices: this is the PeerToPeer memcpy in the CUDA context.
My implementation merges this two parts. It chooses the best API according to
the source device and destination device.
Reviewers: jdoerfert, AndreyChurbanov, grokos
Reviewed By: jdoerfert
Subscribers: yaxunl, guansong, sstefan1, openmp-commits
Tags: #openmp
Differential Revision: https://reviews.llvm.org/D80649
Yaxun (Sam) Liu [Fri, 24 Apr 2020 20:41:24 +0000 (16:41 -0400)]
[CUDA][HIP] Fix implicit HD function resolution
recommit
e03394c6a6ff with fix
When implicit HD function calls a function in device compilation,
if one candidate is an implicit HD function, current resolution rule is:
D wins over HD and H
HD and H are equal
this caused regression when there is an otherwise worse D candidate
This patch changes that to
D, HD and H are all equal
The rationale is that we already know for host compilation there is already
a valid candidate in HD and H candidates that will not cause error. Allowing
HD and H gives us a fall back candidate that will not cause error. If D wins,
that means D has to be a better match otherwise, therefore D should also
be a valid candidate that will not cause error. In this way, we can guarantee
no regression.
Differential Revision: https://reviews.llvm.org/D80450
Matt Arsenault [Thu, 4 Jun 2020 19:31:28 +0000 (15:31 -0400)]
AMDGPU: Fix using unencodable instructions in tests
There are a number of MIR tests using instructions on subtargets where
they don't really exist. These are some of the easy cases that don't
require splitting up test functions.
Matt Arsenault [Thu, 4 Jun 2020 19:21:20 +0000 (15:21 -0400)]
AMDGPU/GlobalISel: Fix making LDS FP atomics legal on SI/CI
Matt Arsenault [Thu, 4 Jun 2020 19:14:23 +0000 (15:14 -0400)]
AMDGPU/GlobalISel: Fix trying to use wave32 for gfx9 test
Alexey Bataev [Mon, 4 May 2020 20:19:31 +0000 (16:19 -0400)]
[OPENMP50]Codegen for inscan reductions in worksharing directives.
Summary:
Implemented codegen for reduction clauses with inscan modifiers in
worksharing constructs.
Emits the code for the directive with inscan reductions.
The code is the following:
```
size num_iters = <num_iters>;
<type> buffer[num_iters];
for (i: 0..<num_iters>) {
<input phase>;
buffer[i] = red;
}
for (int k = 0; k != ceil(log2(num_iters)); ++k)
for (size cnt = last_iter; cnt >= pow(2, k); --k)
buffer[i] op= buffer[i-pow(2,k)];
for (0..<num_iters>) {
red = InclusiveScan ? buffer[i] : buffer[i-1];
<scan phase>;
}
```
Reviewers: jdoerfert
Subscribers: yaxunl, guansong, arphaman, cfe-commits, caomhin
Tags: #clang
Differential Revision: https://reviews.llvm.org/D79948
Thomas Lively [Thu, 4 Jun 2020 20:25:10 +0000 (13:25 -0700)]
[WebAssembly] Lower llvm.debugtrap properly
Summary:
Unlike normal traps, debug traps are allowed to return and can have
additional instructions in the same basic block. Without explicit
backend support for debug traps, they are lowered in ISel as normal
traps. Since normal traps are lowered in the WebAssembly backend to
the UNREACHABLE instruction, which is a terminator, using debug traps
could lead to invalid MBBs when there are additional instructions
after the trap. This patch fixes the issue by lowering debug traps to
a new version of the UNREACHABLE instruction, DEBUG_UNREACHABLE, that
is not a terminator.
An alternative approach would have been to make UNREACHABLE not a
terminator, but that breaks a large number of tests. In particular, it
would require removing the traps inserted after noreturn calls to
@llvm.wasm.throw because otherwise the terminator throw would be
followed by a non-terminator UNREACHABLE and we would be back to
having invalid MBBs. Overall the approach in this patch seems simpler.
Reviewers: aheejin, dschuff
Subscribers: sbc100, jgravelle-google, hiraditya, sunfish, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D81055
Pete Steinfeld [Tue, 26 May 2020 22:12:15 +0000 (15:12 -0700)]
[flang] Fixed crash on forward referenced `len` parameter
Summary:
Using a forward reference to define a `len` parameter causes a crash.
The underlying cause was that a previously declared type had an
erroneous expression for its `LEN` param value. When this expression
was referenced to evaluate a subsequent expression, bad things happened.
I fixed this by putting in code to detect this case.
Reviewers: tskeith, klausler, DavidTruby
Subscribers: llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D80593
Huihui Zhang [Thu, 4 Jun 2020 19:43:51 +0000 (12:43 -0700)]
[NFC] Move test vscale-factor-out-constant.ll to AArch64 sub-directory.
Vscale scalable vector is specific to AArch64 target.
Bring back 'uglygep' check.
Eric Schweitz [Tue, 2 Jun 2020 20:58:12 +0000 (13:58 -0700)]
[flang] Add the conversions for types.
Part of lowering is to convert the front-end types to their FIR dialect representations. These conversions are done by here in the ConvertType module.
proactively update the code to conform better with LLVM coding conventions
Differential Revision: https://reviews.llvm.org/D81034
aartbik [Wed, 3 Jun 2020 23:54:26 +0000 (16:54 -0700)]
[mlir] [VectorOps] Add missing comments to CreateMaskOp lowering
Summary: Add missing comment to CreateMask. Fixed typo in ConstantMask comment.
Reviewers: nicolasvasilache, rriddle, reidtatge, ftynse
Reviewed By: ftynse
Subscribers: mehdi_amini, rriddle, jpienaar, shauheen, antiagainst, nicolasvasilache, arpith-jacob, mgester, lucyrfox, liufengdb, stephenneuendorffer, Joonsoo, grosul1, frgossen, Kayjukh, jurahul
Tags: #mlir
Differential Revision: https://reviews.llvm.org/D81125
Florian Hahn [Thu, 4 Jun 2020 19:15:21 +0000 (20:15 +0100)]
[SemaOverload] Use iterator_range to iterate over VectorTypes (NFC).
We can simplify the code a bit by using iterator_range instead of
plain iterators. Matrix type support here (added in
6f6e91d19337)
already uses an iterator_range.
Reviewers: rjmccall, arphaman, jfb, Bigcheese
Reviewed By: rjmccall
Differential Revision: https://reviews.llvm.org/D81138
Dmitri Gribenko [Thu, 4 Jun 2020 18:20:33 +0000 (20:20 +0200)]
AST Matchers test: use arrays instead of vectors
Subscribers: cfe-commits
Tags: #clang
Differential Revision: https://reviews.llvm.org/D81180
Valentin Clement [Thu, 4 Jun 2020 18:47:13 +0000 (14:47 -0400)]
[flang] avoid GCC < 8 compiler failure after D80794
Summary:
Patch D80794 remove the custom flags for release build for flang.
This leads to build failure with GCC < 8. This patch add upperbound check
in order to avoid the -Werror=array-bounds to trigger a build failure.
```
/home/4vn/versioning/llvm-project/flang/lib/Decimal/big-radix-floating-point.h:183:29: error: array subscript is above array bounds [-Werror=array-bounds]
digit_[j] = digit_[j + remove];
~~~~~~^
/home/4vn/versioning/llvm-project/flang/lib/Decimal/big-radix-floating-point.h:183:29: error: array subscript is above array bounds [-Werror=array-bounds]
digit_[j] = digit_[j + remove];
~~~~~~^
/home/4vn/versioning/llvm-project/flang/lib/Decimal/big-radix-floating-point.h:183:29: error: array subscript is above array bounds [-Werror=array-bounds]
digit_[j] = digit_[j + remove];
~~~~~~^
/home/4vn/versioning/llvm-project/flang/lib/Decimal/big-radix-floating-point.h:183:29: error: array subscript is above array bounds [-Werror=array-bounds]
digit_[j] = digit_[j + remove];
~~~~~~^
/home/4vn/versioning/llvm-project/flang/lib/Decimal/big-radix-floating-point.h:183:29: error: array subscript is above array bounds [-Werror=array-bounds]
digit_[j] = digit_[j + remove];
```
```
/home/4vn/versioning/llvm-project/flang/include/flang/Evaluate/integer.h:809:28: error: array subscript is above array bounds [-Werror=array-bounds]
xy += product[to];
~~~~~~~^
/home/4vn/versioning/llvm-project/flang/include/flang/Evaluate/integer.h:810:22: error: array subscript is above array bounds [-Werror=array-bounds]
product[to] = xy & partMask;
~~~~~~~^
/home/4vn/versioning/llvm-project/flang/include/flang/Evaluate/integer.h:809:28: error: array subscript is above array bounds [-Werror=array-bounds]
xy += product[to];
~~~~~~~^
```
Reviewers: DavidTruby, sscalpone, jdoerfert
Reviewed By: DavidTruby
Subscribers: llvm-commits
Tags: #llvm, #flang
Differential Revision: https://reviews.llvm.org/D81179