Zhaoshi Zheng [Thu, 17 Sep 2020 22:14:14 +0000 (15:14 -0700)]
[NFC] Test Commit
Derek Schuff [Sat, 8 Aug 2020 04:23:11 +0000 (21:23 -0700)]
Support dwarf fission for wasm object files
Initial support for dwarf fission sections (-gsplit-dwarf) on wasm.
The most interesting change is support for writing 2 files (.o and .dwo) in the
wasm object writer. My approach moves object-writing logic into its own function
and calls it twice, swapping out the endian::Writer (W) in between calls.
It also splits the import-preparation step into its own function (and skips it when writing a dwo).
Differential Revision: https://reviews.llvm.org/D85685
Florian Hahn [Thu, 17 Sep 2020 21:09:53 +0000 (22:09 +0100)]
[MemorySSA] Be more conservative when traversing MemoryPhis.
I think we need to be even more conservative when traversing memory
phis, to make sure we catch any loop carried dependences.
This approach updates fillInCurrentPair to use unknown sizes for
locations when we walk over a phi, unless the location is guaranteed to
be loop-invariant for any possible loop. Using an unknown size for
locations should ensure we catch all memory accesses to locations after
the given memory location, which includes loop-carried dependences.
Reviewed By: asbirlea
Differential Revision: https://reviews.llvm.org/D87778
Arthur Eubanks [Thu, 17 Sep 2020 20:57:28 +0000 (13:57 -0700)]
[NewPM] Fix pr45927.ll under NPM
Alexander Shaposhnikov [Fri, 11 Sep 2020 05:05:20 +0000 (22:05 -0700)]
[llvm-install-name-tool] Update the command-line guide
Nikita Popov [Sat, 5 Sep 2020 09:38:39 +0000 (11:38 +0200)]
[InstCombine] Canonicalize SPF_ABS to abs intrinc
Enable canonicalization of SPF_ABS and SPF_NABS to the abs intrinsic.
To be conservative, the one-use check on the comparison is retained,
this may be relaxed if all goes well.
It's pretty likely that this will uncover places that missing
handling for the abs() intrinsic. Please report any seen performance
regressions.
Differential Revision: https://reviews.llvm.org/D87188
Whitney Tsang [Thu, 17 Sep 2020 17:53:26 +0000 (17:53 +0000)]
[LoopUnrollAndJam] Allow unroll and jam loops forced by user.
Summary: Allow unroll and jam loops forced by user.
LoopUnrollAndJamPass is still disabled by default in the NPM pipeline,
and can be controlled by -enable-npm-unroll-and-jam.
Reviewed By: Meinersbur, dmgreen
Differential Revision: https://reviews.llvm.org/D87786
Nikita Popov [Thu, 17 Sep 2020 19:22:37 +0000 (21:22 +0200)]
[GVN] Use that assume(!X) implies X==false (PR47496)
We already use that assume(X) implies X==true, do the same for
assume(!X) implying X==false. This fixes PR47496.
Nikita Popov [Thu, 17 Sep 2020 18:39:29 +0000 (20:39 +0200)]
[GVN] Add additional assume tests (NFC)
The other assume tests seem to be dealing with equalities in
particular. Test implication for the condition itself, especially
the negated case from PR47496.
Florian Hahn [Thu, 17 Sep 2020 15:45:02 +0000 (16:45 +0100)]
[SCEV] Add test cases for max BTC with loop guard info.
This adds test cases for PR40961 and PR47247. They illustrate cases in
which the max backedge-taken count can be improved by information from
the loop guards.
Victor Huang [Thu, 17 Sep 2020 19:13:29 +0000 (14:13 -0500)]
Disable hoisting MI to hotter basic blocks when using pgo
This is a follow up patch for https://reviews.llvm.org/D63676 to
enable the feature when using pgo.
Differential Revision: https://reviews.llvm.org/D85240
Vitaly Buka [Thu, 17 Sep 2020 19:15:00 +0000 (12:15 -0700)]
[Lsan] Use fp registers to search for pointers
X86 can use xmm registers for pointers operations. e.g. for std::swap.
I don't know yet if it's possible on other platforms.
NT_X86_XSTATE includes all registers from NT_FPREGSET so
the latter used only if the former is not available. I am not sure how
reasonable to expect that but LLD has such fallback in
NativeRegisterContextLinux_x86_64::ReadFPR.
Reviewed By: morehouse
Differential Revision: https://reviews.llvm.org/D87754
Jon Roelofs [Thu, 17 Sep 2020 19:13:22 +0000 (12:13 -0700)]
AArch64::ArchKind's underlying type is uint64_t
LLVM GN Syncbot [Thu, 17 Sep 2020 19:09:34 +0000 (19:09 +0000)]
[gn build] Port
7e4c6fb8546
Andrew Litteken [Thu, 17 Sep 2020 17:28:09 +0000 (12:28 -0500)]
[IRSim] Adding IR Instruction Mapper
This introduces the IRInstructionMapper, and the associated wrapper for
instructions, IRInstructionData, that maps IR level Instructions to
unsigned integers.
Mapping is done mainly by using the "isSameOperationAs" comparison
between two instructions. If they return true, the opcode, result type,
and operand types of the instruction are used to hash the instruction
with an unsigned integer. The mapper accepts instruction ranges, and
adds each resulting integer to a list, and each wrapped instruction to
a separate list.
At present, branches, phi nodes are not mapping and exception handling
is illegal. Debug instructions are not considered.
The different mapping schemes are tested in
unittests/Analysis/IRSimilarityIdentifierTest.cpp
Recommit of:
b04c1a9d3127730c05e8a22a0e931a12a39528df
Differential Revision: https://reviews.llvm.org/D86968
Cameron McInally [Thu, 17 Sep 2020 18:54:46 +0000 (13:54 -0500)]
[SVE][WIP] Implement lowering for fixed length VSELECT to Scalable
Map fixed length VSELECT to its Scalable equivalent.
Differential Revision: https://reviews.llvm.org/D85364
Reid Kleckner [Thu, 4 Jun 2020 01:08:55 +0000 (18:08 -0700)]
[PDB] Split TypeServerSource and extend type index map lifetime
Extending the lifetime of these type index mappings does increase memory
usage (+2% in my case), but it decouples type merging from symbol
merging. This is a pre-requisite for two changes that I have in mind:
- parallel type merging: speeds up slow type merging
- defered symbol merging: avoid heap allocating (relocating) all symbols
This eliminates CVIndexMap and moves its data into TpiSource. The maps
are also split into a SmallVector and ArrayRef component, so that the
ipiMap can alias the tpiMap for /Z7 object files, and so that both maps
can simply alias the PDB type server maps for /Zi files.
Splitting TypeServerSource establishes that all input types to be merged
can be identified with two 32-bit indices:
- The index of the TpiSource object
- The type index of the record
This is useful, because this information can be stored in a single
64-bit atomic word to enable concurrent hashtable insertion.
One last change is that now all object files with debugChunks get a
TpiSource, even if they have no type info. This avoids some null checks
and special cases.
Differential Revision: https://reviews.llvm.org/D87736
Amara Emerson [Thu, 17 Sep 2020 18:17:18 +0000 (11:17 -0700)]
[AArch64][GlobalISel] Widen G_EXTRACT_VECTOR_ELT element types if < 8b.
In order to not unnecessarily promote the source vector to greater than our
native vector size of 128b, I've added some cascading rules to widen based on
the number of elements.
Amara Emerson [Thu, 17 Sep 2020 18:16:02 +0000 (11:16 -0700)]
[AArch64][GlobalISel] Make <8 x s16> and <16 x s8> legal for shifts.
Sanjay Patel [Thu, 17 Sep 2020 18:22:05 +0000 (14:22 -0400)]
[VectorCombine] limit load+insert transform to one-use
As discussed in:
https://llvm.org/PR47558
...there are several potential fixes/follow-ups visible
in the test case, but this is the quickest and safest
fix of the perf regression.
Craig Topper [Thu, 17 Sep 2020 17:33:34 +0000 (10:33 -0700)]
[X86] Don't match x87 register inline asm constraints unless the VT is floating point or its a clobber
The register class picked will be the RFP80 register class which has a f80 VT. The code in SelectionDAGBuilder that generates copies around inline assembly doesn't know how to handle an integer and floating point type of different bit widths.
The test case is derived from this https://godbolt.org/z/sEa659 which gcc accepts but clang crashes on. This patch just gives a more graceful error. I'm not sure if the single element struct case is special in gcc. Adding another field to the struct makes gcc reject it. If we want to support this correctly I think we need a change in the frontend to give us the true element type. Right now the frontend just realizes the constraint can take a memory argument so creates an integer type of the same size and bitcasts.
Differential Revision: https://reviews.llvm.org/D87485
Navdeep Kumar [Thu, 17 Sep 2020 18:07:21 +0000 (23:37 +0530)]
[MLIR][Affine] Add parametric tile size support for affine.for tiling
Add support to tile affine.for ops with parametric sizes (i.e., SSA
values). Currently supports hyper-rectangular loop nests with constant
lower bounds only. Move methods
- moveLoopBody(*)
- getTileableBands(*)
- checkTilingLegality(*)
- tilePerfectlyNested(*)
- constructTiledIndexSetHyperRect(*)
to allow reuse with constant tile size API. Add a test pass -test-affine
-parametric-tile to test parametric tiling.
Differential Revision: https://reviews.llvm.org/D87353
Abhishek Varma [Thu, 17 Sep 2020 18:00:47 +0000 (23:30 +0530)]
[MLIR] Support for return values in Affine.For yield
Add support for return values in affine.for yield along the same lines
as scf.for and affine.parallel.
Signed-off-by: Abhishek Varma <abhishek.varma@polymagelabs.com>
Differential Revision: https://reviews.llvm.org/D87437
Yaxun (Sam) Liu [Thu, 17 Sep 2020 17:53:38 +0000 (13:53 -0400)]
Revert "[NFC] Refactor DiagnosticBuilder and PartialDiagnostic"
This reverts commit
ee5519d323571c4a9a7d92cb817023c9b95334cd.
Yaxun (Sam) Liu [Thu, 17 Sep 2020 17:53:25 +0000 (13:53 -0400)]
Revert "[CUDA][HIP] Defer overloading resolution diagnostics for host device functions"
This reverts commit
7f1f89ec8d9944559042bb6d3b1132eabe3409de.
This reverts commit
40df06cdafc010002fc9cfe1dda73d689b7d27a6.
Sanjay Patel [Thu, 17 Sep 2020 17:49:48 +0000 (13:49 -0400)]
[VectorCombine] rearrange bailouts for load insert for efficiency; NFC
Sanjay Patel [Thu, 17 Sep 2020 17:21:58 +0000 (13:21 -0400)]
[VectorCombine] add test for multi-use load (PR47558); NFC
Jinsong Ji [Thu, 17 Sep 2020 17:43:41 +0000 (17:43 +0000)]
[PowerPC][AIX] Don't hardcode python invoke command line
We shouldn't assume python exists, we should let lit
to decide whether it is python or python3 and expand the path.
Adrian Prantl [Thu, 17 Sep 2020 17:46:03 +0000 (10:46 -0700)]
Add missing include
jerryyin [Thu, 17 Sep 2020 15:47:33 +0000 (08:47 -0700)]
[AMDGPU] Fix ROCm unit test memref initialization
Raul Tambre [Fri, 4 Sep 2020 16:10:09 +0000 (19:10 +0300)]
[Sema] Introduce BuiltinAttr, per-declaration builtin-ness
Instead of relying on whether a certain identifier is a builtin, introduce BuiltinAttr to specify a declaration as having builtin semantics.
This fixes incompatible redeclarations of builtins, as reverting the identifier as being builtin due to one incompatible redeclaration would have broken rest of the builtin calls.
Mostly-compatible redeclarations of builtins also no longer have builtin semantics. They don't call the builtin nor inherit their attributes.
A long-standing FIXME regarding builtins inside a namespace enclosed in extern "C" not being recognized is also addressed.
Due to the more correct handling attributes for builtin functions are added in more places, resulting in more useful warnings.
Tests are updated to reflect that.
Intrinsics without an inline definition in intrin.h had `inline` and `static` removed as they had no effect and caused them to no longer be recognized as builtins otherwise.
A pthread_create() related test is XFAIL-ed, as it relied on it being recognized as a builtin based on its name.
The builtin declaration syntax is too restrictive and doesn't allow custom structs, function pointers, etc.
It seems to be the only case and fixing this would require reworking the current builtin syntax, so this seems acceptable.
Fixes PR45410.
Reviewed By: rsmith, yutsumi
Differential Revision: https://reviews.llvm.org/D77491
Matt Morehouse [Thu, 17 Sep 2020 16:23:35 +0000 (09:23 -0700)]
[DFSan] Add bcmp wrapper.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D87801
Eduardo Caldas [Thu, 17 Sep 2020 09:32:46 +0000 (09:32 +0000)]
[SyntaxTree][Synthesis] Fix allocation in `createTree` for more general use
Prior to this change `createTree` could not create arbitrary syntax
trees. Now it dispatches to the constructor of the concrete syntax tree
according to the `NodeKind` passed as argument. This allows reuse inside
the Synthesis API. # Please enter the commit message for your changes.
Lines starting
Differential Revision: https://reviews.llvm.org/D87820
Bogdan Graur [Thu, 17 Sep 2020 16:04:21 +0000 (18:04 +0200)]
[amdgpu] Compilation fix for Release
Reviewed By: bkramer
Differential Revision: https://reviews.llvm.org/D87838
Sanjay Patel [Thu, 17 Sep 2020 13:02:26 +0000 (09:02 -0400)]
[InstSimplify] add tests for FP constant miscompile; NFC (PR43907)
David Green [Thu, 17 Sep 2020 15:58:35 +0000 (16:58 +0100)]
[ARM] Expand distributing increments to also handle existing pre/post inc instructions.
This extends the distributing postinc code in load/store optimizer to
also handle the case where there is an existing pre/post inc instruction,
where subsequent instructions can be modified to use the adjusted
offset from the increment. This can save us having to keep the old
register live past the increment instruction.
Differential Revision: https://reviews.llvm.org/D83377
Amara Emerson [Wed, 16 Sep 2020 19:14:40 +0000 (12:14 -0700)]
[AArch64][GlobalISel] Fix bug in fewVectorElts action while legalizing oversize G_FPTRUNC vectors.
For <8 x s32> = fptrunc <8 x s64> the fewerElementsVector action tries to break
down the source vector into the final source vectors of <2 x s64> using unmerge.
This fixes a crash due to using the wrong number of elements for the breakdown
type.
Also add some legalizer tests for explicitly G_FPTRUNC which we didn't have.
Differential Revision: https://reviews.llvm.org/D87814
Hanhan Wang [Thu, 17 Sep 2020 15:54:16 +0000 (08:54 -0700)]
[mlir][Vector] Add a folder for vector.broadcast
Fold the operation if the source is a scalar constant or splat constant.
Update transform-patterns-matmul-to-vector.mlir because the broadcast ops are folded in the conversion.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D87703
Yaxun (Sam) Liu [Thu, 17 Sep 2020 15:51:09 +0000 (11:51 -0400)]
Fix build failure in clangd
Simon Pilgrim [Thu, 17 Sep 2020 15:00:02 +0000 (16:00 +0100)]
ModuloSchedule.cpp - remove unnecessary includes. NFCI.
Already included in ModuloSchedule.h
Matt Morehouse [Thu, 17 Sep 2020 15:43:26 +0000 (08:43 -0700)]
Revert "[DFSan] Add bcmp wrapper."
This reverts commit
559f9198125392bfa8e7d462aa8e87fcf5030185 due to bot
failure.
Max Kazantsev [Thu, 17 Sep 2020 15:36:41 +0000 (22:36 +0700)]
[Test] Add tests showing that IndVars cannot prove (X + 1 > X)
Valentin Clement [Thu, 17 Sep 2020 15:34:28 +0000 (11:34 -0400)]
[flang][openacc] Lower clauses on loop construct to OpenACC dialect
Lower OpenACCLoopConstruct and most of the clauses to the OpenACC acc.loop operation in MLIR.
This patch refelcts what can be upstream from PR flang-compiler/f18-llvm-project#419
Reviewed By: SouraVX
Differential Revision: https://reviews.llvm.org/D87389
Valentin Clement [Thu, 17 Sep 2020 15:33:31 +0000 (11:33 -0400)]
[mlir][openacc] Change operand type from index to AnyInteger in parallel op
This patch change the type of operands async, wait, numGangs, numWorkers and vectorLength from index
to AnyInteger to fit with acc.loop and the OpenACC specification.
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D87712
David Green [Thu, 17 Sep 2020 15:33:03 +0000 (16:33 +0100)]
[ARM] Add more MVE postinc distribution tests. NFC
Yaxun (Sam) Liu [Wed, 16 Sep 2020 19:42:08 +0000 (15:42 -0400)]
[CUDA][HIP] Defer overloading resolution diagnostics for host device functions
In CUDA/HIP a function may become implicit host device function by
pragma or constexpr. A host device function is checked in both
host and device compilation. However it may be emitted only
on host or device side, therefore the diagnostics should be
deferred until it is known to be emitted.
Currently clang is only able to defer certain diagnostics. This causes
false alarms and limits the usefulness of host device functions.
This patch lets clang defer all overloading resolution diagnostics for host device functions.
An option -fgpu-defer-diag is added to control this behavior. By default
it is off.
It is NFC for other languages.
Differential Revision: https://reviews.llvm.org/D84364
Sanne Wouda [Sat, 12 Sep 2020 00:17:42 +0000 (01:17 +0100)]
[AArch64] Match pairwise add/fadd pattern
D75689 turns the faddp pattern into a shuffle with vector add.
Match this new pattern in target-specific DAG combine, rather than ISel,
because legalization (for v2f32) turns it into a bit of a mess.
- extended to cover f16, f32, f64 and i64
Sanne Wouda [Fri, 4 Sep 2020 15:58:02 +0000 (16:58 +0100)]
Precommit test updates
Matt Morehouse [Thu, 17 Sep 2020 15:22:54 +0000 (08:22 -0700)]
[DFSan] Add bcmp wrapper.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D87801
Alexey Bataev [Wed, 16 Sep 2020 16:19:06 +0000 (12:19 -0400)]
[OpenMP 5.0] Fix user-defined mapper privatization in tasks
This patch fixes the problem that user-defined mapper array is not correctly privatized inside a task. This problem causes openmp/libomptarget/test/offloading/target_depend_nowait.cpp fails.
Differential Revision: https://reviews.llvm.org/D84470
Xun Li [Thu, 17 Sep 2020 15:12:46 +0000 (08:12 -0700)]
[Coroutine] Fix a bug where Coroutine incorrectly spills phi and invoke defs before CoroBegin
When a spill definition is before CoroBegin, we cannot spill it to the frame immediately after the definition. We have to spill it after the frame is ready.
The current implementation handles it properly for any other kinds of instructions except for PhINode and InvokeInst, which could also be defined before CoroBegin.
This patch fixes it by moving the CoroBegin dominance check earlier, so that it covers all cases.
Added a test.
Differential Revision: https://reviews.llvm.org/D87810
Louis Dionne [Thu, 30 Jul 2020 14:00:53 +0000 (10:00 -0400)]
[libc++] Remove some workarounds for missing variadic templates
We don't support GCC in C++03 mode, and Clang provides variadic templates
even in C++03 mode. So there's effectively no supported compiler that
doesn't support variadic templates.
This effectively gets rid of all uses of _LIBCPP_HAS_NO_VARIADICS, but
some workarounds for the lack of variadics remain.
Michael Liao [Wed, 9 Sep 2020 20:48:03 +0000 (16:48 -0400)]
[amdgpu] Lower SGPR-to-VGPR copy in the final phase of ISel.
- Need to lower COPY from SGPR to VGPR to a real instruction as the
standard COPY is used where the source and destination are from the
same register bank so that we potentially coalesc them together and
save one COPY. Considering that, backend optimizations, such as CSE,
won't handle them. However, the copy from SGPR to VGPR always needs
materializing to a native instruction, it should be lowered into a
real one before other backend optimizations.
Differential Revision: https://reviews.llvm.org/D87556
David Green [Thu, 17 Sep 2020 15:00:51 +0000 (16:00 +0100)]
[ARM] Sink splats to MVE intrinsics
The predicated MVE intrinsics are generated as, for example,
llvm.arm.mve.add.predicated(x, splat(y). p). We need to sink the splat
value back into the loop, like we do for other instructions, so we can
re-select qr variants.
Differential Revision: https://reviews.llvm.org/D87693
Kamil Rytarowski [Thu, 17 Sep 2020 14:57:30 +0000 (16:57 +0200)]
[compiler-rt] [scudo] Fix typo in function attribute
Fixes the build after landing https://reviews.llvm.org/D87562
Stephan Herhut [Wed, 16 Sep 2020 08:01:54 +0000 (10:01 +0200)]
[mlir][Standard] Canonicalize chains of tensor_cast operations
Adds a pattern that replaces a chain of two tensor_cast operations by a single tensor_cast operation if doing so will not remove constraints on the shapes.
Kamil Rytarowski [Thu, 17 Sep 2020 14:46:32 +0000 (16:46 +0200)]
[compiler-rt] [hwasan] Replace INLINE with inline
Fixes the build after landing D87562.
Kamil Rytarowski [Thu, 17 Sep 2020 14:34:59 +0000 (16:34 +0200)]
[compiler-rt] [netbsd] Include <sys/dkbad.h>
Fixes build on NetBSD/sparc64.
alex-t [Wed, 16 Sep 2020 16:54:29 +0000 (19:54 +0300)]
[AMDGPU] should expand ROTL i16 to shifts.
Instruction combining pass turns library rotl implementation to llvm.fshl.i16.
In the selection dag the intrinsic is turned to ISD::ROTL node that cannot be selected.
Need to expand it to shifts again.
Reviewed By: rampitec, arsenm
Differential Revision: https://reviews.llvm.org/D87618
Kamil Rytarowski [Thu, 17 Sep 2020 14:27:48 +0000 (16:27 +0200)]
[compiler-rt] [tsan] [netbsd] Catch unsupported LONG_JMP_SP_ENV_SLOT
Error out during build for unsupported CPU.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D87602
Kamil Rytarowski [Thu, 17 Sep 2020 14:04:50 +0000 (16:04 +0200)]
[compiler-rt] Replace INLINE with inline
This fixes the clash with BSD headers.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D87562
Simon Pilgrim [Thu, 17 Sep 2020 14:05:45 +0000 (15:05 +0100)]
LiveDebugVariables.cpp - remove unnecessary Compiler.h include. NFCI.
Already included in LiveDebugVariables.h
Simon Pilgrim [Thu, 17 Sep 2020 14:03:53 +0000 (15:03 +0100)]
DwarfExpression.cpp - remove unnecessary includes. NFCI.
Already included in DwarfExpression.h
Simon Pilgrim [Thu, 17 Sep 2020 14:00:11 +0000 (15:00 +0100)]
ValueList.cpp - remove unnecessary includes. NFCI.
Already included in ValueList.h
Kamil Rytarowski [Thu, 17 Sep 2020 14:02:59 +0000 (16:02 +0200)]
[compiler-rt] Avoid pulling libatomic to sanitizer tests
Avoid fallbacking to software emulated compiler atomics, that are usually
provided by libatomic, which is not always present.
This fixes the test on NetBSD, which does not provide libatomic in base.
Reviewed By: vitalybuka
Differential Revision: https://reviews.llvm.org/D87568
Simon Pilgrim [Thu, 17 Sep 2020 13:45:46 +0000 (14:45 +0100)]
SafeStackLayout.cpp - remove unnecessary StackLifetime.h include. NFCI.
Already included in SafeStackLayout.h
jerryyin [Wed, 16 Sep 2020 15:57:37 +0000 (08:57 -0700)]
[AMDGPU] Bump to ROCm 3.7 dependency hip_hcc->amdhip64
Differential Revision: https://reviews.llvm.org/D87773
Simon Pilgrim [Thu, 17 Sep 2020 13:27:15 +0000 (14:27 +0100)]
InstCombiner.h - remove unnecessary KnownBits.h include. NFCI.
Move the include down to cpp files with an implicit dependency.
Yvan Roux [Thu, 17 Sep 2020 13:13:55 +0000 (15:13 +0200)]
[ARM][MachineOutliner] Add missing testcase for calls.
Florian Hahn [Wed, 16 Sep 2020 17:44:40 +0000 (18:44 +0100)]
[MemorySSA] Add another loop clobber test case.
Kerry McLaughlin [Thu, 17 Sep 2020 10:52:14 +0000 (11:52 +0100)]
[SVE][CodeGen] Lower floating point -> integer conversions
This patch adds new ISD nodes, FCVTZS_MERGE_PASSTHRU &
FCVTZU_MERGE_PASSTHRU, which are used to lower scalable vector
FP_TO_SINT/FP_TO_UINT operations and the following intrinsics:
- llvm.aarch64.sve.fcvtzu
- llvm.aarch64.sve.fcvtzs
Reviewed By: efriedma, paulwalker-arm
Differential Revision: https://reviews.llvm.org/D87232
Georgii Rymar [Thu, 17 Sep 2020 12:36:06 +0000 (15:36 +0300)]
[obj2yaml] - Don't emit EM_NONE.
When ELF header's `e_machine == 0`, we emit:
```
Machine: EM_NONE
```
We can avoid doing this, because yaml2obj sets the
`e_machine` field to `EM_NONE` by default.
Differential revision: https://reviews.llvm.org/D87829
Georgii Rymar [Tue, 15 Sep 2020 13:17:08 +0000 (16:17 +0300)]
[llvm-readelf/obj][test] - Document what we print in various places for unnamed section symbols.
We have an issue with `ELFDumper<ELFT>::getSymbolSectionName`:
1) It is used deeply for both LLVM/GNU styles and might return LLVM-style only
values to describe symbols: "Undefined", "Processor Specific", "Absolute", etc.
2) `getSymbolSectionName` is used by `getFullSymbolName` and these special values
might appear in instead of symbol names in many places.
This occurs for unnamed section symbols.
It was not noticed because for most cases I've found it is unexpected to have an
unnamed section symbol. This patch documents the existent behavior, adds tests and FIXMEs.
Differential revision: https://reviews.llvm.org/D87763
Sanjay Patel [Thu, 17 Sep 2020 12:39:23 +0000 (08:39 -0400)]
[SLP] sort candidates to increase chance of optimal compare reduction
This is one (small) part of improving PR41312:
https://llvm.org/PR41312
As shown there and in the smaller tests here, if we have some member of the
reduction values that does not match the others, we want to push it to the
end (bring the matching members forward and together).
In the regression tests, we have 5 candidates for the 4 slots of the reduction.
If the one "wrong" compare is grouped with the others, it prevents forming the
ideal v4i1 compare reduction.
Differential Revision: https://reviews.llvm.org/D87772
Jessica Clarke [Thu, 17 Sep 2020 12:44:01 +0000 (13:44 +0100)]
[clang][docs] Fix documentation of -O
D79916 changed the behaviour from -O2 to -O1 but the documentation was
not updated to reflect this.
Simon Pilgrim [Thu, 17 Sep 2020 12:28:14 +0000 (13:28 +0100)]
Remove unnecessary forward declarations. NFCI.
All of these forward declarations are fully defined in headers that are directly included.
Mikael Holmen [Thu, 17 Sep 2020 12:20:34 +0000 (14:20 +0200)]
[ConstraintSystem] Remove local variable that is set but not read [NFC]
gcc 7.4 warns about it.
mydeveloperday [Thu, 17 Sep 2020 12:22:26 +0000 (13:22 +0100)]
[clang-format][regression][PR47461] ifdef causes catch to be seen as a function
https://bugs.llvm.org/show_bug.cgi?id=47461
The following change {D80940} caused a regression in code which ifdef's around the try and catch block cause incorrect brace placement around the catch
```
try
{
}
catch (...) {
// This is not a small function
bar = 1;
}
}
```
The brace after the catch will be placed on a newline
Reviewed By: curdeius
Differential Revision: https://reviews.llvm.org/D87291
Simon Pilgrim [Thu, 17 Sep 2020 12:08:42 +0000 (13:08 +0100)]
MetadataLoader.cpp - remove unnecessary StringRef include. NFCI.
Already included in MetadataLoader.h
Simon Pilgrim [Thu, 17 Sep 2020 11:52:23 +0000 (12:52 +0100)]
SymbolizableObjectFile.h - remove unnecessary includes. NFCI.
Use forward declarations where possible, move includes down to SymbolizableObjectFile.cpp and avoid duplicate includes.
Sam Parker [Thu, 17 Sep 2020 12:07:46 +0000 (13:07 +0100)]
[NFC][ARM] Tail fold test changes
Run update script on one test and add another.
David Spickett [Thu, 17 Sep 2020 12:07:44 +0000 (13:07 +0100)]
Revert "[lldb] Don't send invalid region addresses to lldb server"
This reverts commit
c687af0c30b4dbdc9f614d5e061c888238e0f9c5
due to a test failure on Windows.
David Green [Thu, 17 Sep 2020 11:39:21 +0000 (12:39 +0100)]
[ARM] Additional tests for qr intrinsics in loops. NFC
Simon Pilgrim [Thu, 17 Sep 2020 11:18:27 +0000 (12:18 +0100)]
DwarfStringPool.cpp - remove unnecessary StringRef include. NFCI.
Already included in DwarfStringPool.h
Simon Pilgrim [Thu, 17 Sep 2020 11:12:00 +0000 (12:12 +0100)]
DwarfFile.h - remove unnecessary includes. NFCI.
Use forward declarations where possible, move includes down to DwarfFile.cpp and avoid duplicate includes.
David Green [Thu, 17 Sep 2020 11:10:23 +0000 (12:10 +0100)]
[ARM] Extra fp16 bitcast tests. NFC
Alex Zinenko [Thu, 17 Sep 2020 10:59:57 +0000 (12:59 +0200)]
[mlir] turn clang-format back on in C API test
C API test uses FileCheck comments inside C code and needs to
temporarily switch off clang-format to prevent it from messing with
FileCheck directives. A recently landed commit forgot to turn it back on
after a block of FileCheck comments. Fix that.
Nico Weber [Thu, 17 Sep 2020 10:33:24 +0000 (06:33 -0400)]
[gn build] (manually) port
c9af34027bc
Vincent Zhao [Wed, 16 Sep 2020 15:04:09 +0000 (16:04 +0100)]
[MLIR] Turns swapId into a FlatAffineConstraints member func
`swapId` used to be a static function in `AffineStructures.cpp`. This diff makes it accessible from the external world by turning it into a member function of `FlatAffineConstraints`. This will be very helpful for other projects that need to manipulate the content of `FlatAffineConstraints`.
Differential Revision: https://reviews.llvm.org/D87766
Simon Pilgrim [Wed, 16 Sep 2020 18:02:20 +0000 (19:02 +0100)]
[AsmPrinter] DwarfDebug - use DebugLoc const references where possible. NFC.
Avoid unnecessary copies.
Simon Pilgrim [Wed, 16 Sep 2020 18:01:42 +0000 (19:01 +0100)]
[AMDGPU] Remove orphan SITargetLowering::LowerINT_TO_FP declaration. NFCI.
Method implementation no longer exists.
Simon Pilgrim [Wed, 16 Sep 2020 17:52:28 +0000 (18:52 +0100)]
[AsmPrinter] Remove orphan DwarfUnit::shareAcrossDWOCUs declaration. NFCI.
Method implementation no longer exists.
Jakub Lichman [Thu, 17 Sep 2020 09:26:30 +0000 (09:26 +0000)]
[mlir][Linalg] Convolution tiling added to ConvOp vectorization pass
ConvOp vectorization supports now only convolutions of static shapes with dimensions
of size either 3(vectorized) or 1(not) as underlying vectors have to be of static
shape as well. In this commit we add support for convolutions of any size as well as
dynamic shapes by leveraging existing matmul infrastructure for tiling of both input
and kernel to sizes accepted by the previous version of ConvOp vectorization.
In the future this pass can be extended to take "tiling mask" as a user input which
will enable vectorization of user specified dimensions.
Differential Revision: https://reviews.llvm.org/D87676
Cullen Rhodes [Fri, 11 Sep 2020 15:18:44 +0000 (15:18 +0000)]
[clang][aarch64] ACLE: Support implicit casts between GNU and SVE vectors
This patch adds support for implicit casting between GNU vectors and SVE
vectors when `__ARM_FEATURE_SVE_BITS==N`, as defined by the Arm C
Language Extensions (ACLE, version 00bet5, section 3.7.3.3) for SVE [1].
This behavior makes it possible to use GNU vectors with ACLE functions
that operate on VLAT. For example:
typedef int8_t vec __attribute__((vector_size(32)));
vec f(vec x) { return svasrd_x(svptrue_b8(), x, 1); }
Tests are also added for implicit casting between GNU and fixed-length
SVE vectors created by the 'arm_sve_vector_bits' attribute. This
behavior makes it possible to use VLST with existing interfaces that
operate on GNUT. For example:
typedef int8_t vec1 __attribute__((vector_size(32)));
void f(vec1);
#if __ARM_FEATURE_SVE_BITS==256 && __ARM_FEATURE_SVE_VECTOR_OPERATORS
typedef svint8_t vec2 __attribute__((arm_sve_vector_bits(256)));
void g(vec2 x) { f(x); } // OK
#endif
The `__ARM_FEATURE_SVE_VECTOR_OPERATORS` feature macro indicates
interoperability with the GNU vector extension. This is the first patch
providing support for this feature, which once complete will be enabled
by the `-msve-vector-bits` flag, as the `__ARM_FEATURE_SVE_BITS` feature
currently is.
[1] https://developer.arm.com/documentation/100987/latest
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D87607
David Spickett [Tue, 15 Sep 2020 13:49:48 +0000 (14:49 +0100)]
[lldb] Don't send invalid region addresses to lldb server
Previously when <addr> in "memory region <addr>" didn't
parse correctly, we'd print an error then also ask lldb-server
for a region containing LLDB_INVALID_ADDRESS.
(lldb) memory region not_an_address
error: invalid address argument "not_an_address"...
error: Server returned invalid range
Only send the command to lldb-server if the address
parsed correctly.
(lldb) memory region not_an_address
error: invalid address argument "not_an_address"...
Reviewed By: labath
Differential Revision: https://reviews.llvm.org/D87694
Rainer Orth [Thu, 17 Sep 2020 09:17:11 +0000 (11:17 +0200)]
[X86] Fix stack alignment on 32-bit Solaris/x86
On Solaris/x86, several hundred 32-bit tests `FAIL`, all in the same way:
env ASAN_OPTIONS=halt_on_error=false ./halt_on_error_suppress_equal_pcs.cpp.tmp
Segmentation Fault (core dumped)
They segfault during startup:
Thread 2 received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1 (LWP 1)]
0x080f21f0 in __sanitizer::internal_mmap(void*, unsigned long, int, int, int, unsigned long long) () at /vol/llvm/src/llvm-project/dist/compiler-rt/lib/sanitizer_common/sanitizer_solaris.cpp:65
65 int prot, int flags, int fd, OFF_T offset) {
1: x/i $pc
=> 0x80f21f0 <_ZN11__sanitizer13internal_mmapEPvmiiiy+16>: movaps 0x30(%esp),%xmm0
(gdb) p/x $esp
$3 = 0xfeffd488
The problem is that `movaps` expects 16-byte alignment, while 32-bit Solaris/x86
only guarantees 4-byte alignment following the i386 psABI.
This patch updates `X86Subtarget::initSubtargetFeatures` accordingly,
handles Solaris/x86 in the corresponding testcase, and allows for some
variation in address alignment in
`compiler-rt/test/ubsan/TestCases/TypeCheck/vptr.cpp`.
Tested on `amd64-pc-solaris2.11` and `x86_64-pc-linux-gnu`.
Differential Revision: https://reviews.llvm.org/D87615
Douglas Yung [Thu, 17 Sep 2020 08:28:32 +0000 (01:28 -0700)]
Revert "Re-land: Add new hidden option -print-changed which only reports changes to IR"
The test added in this commit is failing on Windows bots:
http://lab.llvm.org:8011/builders/llvm-clang-win-x-armv7l/builds/1269
This reverts commit
f9e6d1edc0dad9afb26e773aa125ed62c58f7080 and follow-up commit
6859d95ea2d0f3fe0de2923a3f642170e66a1a14.
Roman Lebedev [Thu, 17 Sep 2020 08:08:26 +0000 (11:08 +0300)]
[NFC] EliminateDuplicatePHINodes(): small-size optimization: if there are <= 32 PHI's, O(n^2) algo is faster (geomean -0.08%)
This is functionally equivalent to the old implementation.
As per https://llvm-compile-time-tracker.com/compare.php?from=
5f4e9bf6416e45eba483a4e5e263749989fdb3b3&to=
4739e6e4eb54d3736e6457249c0919b30f6c855a&stat=instructions
this is a clear geomean compile-time regression-free win with overall geomean of `-0.08%`
32 PHI's appears to be the sweet spot; both the 16 and 64 performed worse:
https://llvm-compile-time-tracker.com/compare.php?from=
5f4e9bf6416e45eba483a4e5e263749989fdb3b3&to=
c4efe1fbbfdf0305ac26cd19eacb0c7774cdf60e&stat=instructions
https://llvm-compile-time-tracker.com/compare.php?from=
5f4e9bf6416e45eba483a4e5e263749989fdb3b3&to=
e4989d1c67010d3339d1a40ff5286a31f10cfe82&stat=instructions
If we have more PHI's than that, we fall-back to the original DenseSet-based implementation,
so the not-so-fast cases will still be handled.
However compile-time isn't the main motivation here.
I can name at least 3 limitations of this CSE:
1. Assumes that all PHI nodes have incoming basic blocks in the same order (can be fixed while keeping the DenseMap)
2. Does not special-handle `undef` incoming values (i don't see how we can do this with hashing)
3. Does not special-handle backedge incoming values (maybe can be fixed by hashing backedge as some magical value)
Reviewed By: efriedma
Differential Revision: https://reviews.llvm.org/D87408
Jay Foad [Wed, 16 Sep 2020 10:13:45 +0000 (11:13 +0100)]
[SplitKit] Only copy live lanes
When splitting a live interval with subranges, only insert copies for
the lanes that are live at the point of the split. This avoids some
unnecessary copies and fixes a problem where copying dead lanes was
generating MIR that failed verification. The test case for this is
test/CodeGen/AMDGPU/splitkit-copy-live-lanes.mir.
Without this fix, some earlier live range splitting would create %430:
%430 [256r,848r:0)[848r,2584r:1) 0@256r 1@848r L0000000000000003 [848r,2584r:0) 0@848r L0000000000000030 [256r,2584r:0) 0@256r weight:1.480938e-03
...
256B undef %430.sub2:vreg_128 = V_LSHRREV_B32_e32 16, %20.sub1:vreg_128, implicit $exec
...
848B %430.sub0:vreg_128 = V_AND_B32_e32 %92:sreg_32, %20.sub1:vreg_128, implicit $exec
...
2584B %431:vreg_128 = COPY %430:vreg_128
Then RAGreedy::tryLocalSplit would split %430 into %432 and %433 just
before 848B giving:
%432 [256r,844r:0) 0@256r L0000000000000030 [256r,844r:0) 0@256r weight:3.066802e-03
%433 [844r,848r:0)[848r,2584r:1) 0@844r 1@848r L0000000000000030 [844r,2584r:0) 0@844r L0000000000000003 [844r,844d:0)[848r,2584r:1) 0@844r 1@848r weight:2.831776e-03
...
256B undef %432.sub2:vreg_128 = V_LSHRREV_B32_e32 16, %20.sub1:vreg_128, implicit $exec
...
844B undef %433.sub0:vreg_128 = COPY %432.sub0:vreg_128 {
internal %433.sub2:vreg_128 = COPY %432.sub2:vreg_128
848B }
%433.sub0:vreg_128 = V_AND_B32_e32 %92:sreg_32, %20.sub1:vreg_128, implicit $exec
...
2584B %431:vreg_128 = COPY %433:vreg_128
Note that the copy from %432 to %433 at 844B is a curious
bundle-without-a-BUNDLE-instruction that SplitKit creates deliberately,
and it includes a copy of .sub0 which is not live at this point, and
that causes it to fail verification:
*** Bad machine code: No live subrange at use ***
- function: zextload_global_v64i16_to_v64i64
- basic block: %bb.0 (0x7faed48) [0B;2848B)
- instruction: 844B undef %433.sub0:vreg_128 = COPY %432.sub0:vreg_128
- operand 1: %432.sub0:vreg_128
- interval: %432 [256r,844r:0) 0@256r L0000000000000030 [256r,844r:0) 0@256r weight:3.066802e-03
- at: 844B
Using real bundles with a BUNDLE instruction might also fix this
problem, but the current fix is less invasive and also avoids some
unnecessary copies.
https://bugs.llvm.org/show_bug.cgi?id=47492
Differential Revision: https://reviews.llvm.org/D87757
Jay Foad [Wed, 16 Sep 2020 19:28:02 +0000 (20:28 +0100)]
[AMDGPU] Generate test checks for splitkit-copy-bundle.mir
This is a pre-commit for D87757 "[SplitKit] Only copy live lanes".