wanglei [Fri, 18 Nov 2022 08:33:49 +0000 (16:33 +0800)]
[LoongArch] Implement assembler branches pseudo instructions
These instructions always output the canonical mnemonic. The GNU tools
emit the canonical mnemonic for the branch pseudo instructions as well
(e.g. "bgt" will be recognised by the assembler but never printed by
objdump).
Reviewed By: xen0n
Differential Revision: https://reviews.llvm.org/D138100
Tobias Gysi [Fri, 18 Nov 2022 08:44:26 +0000 (09:44 +0100)]
[mlir][llvm] Import additional debug info from LLVMIR.
Add a DebugImporter to convert LLVMIR debug metadata into
MLIR debug attributes. It is the counterpart to the
DebugTranslation class and supports the same attributes.
The revision only supports the translation of instruction,
function, and module debug information. The import of
intrinsics is left to a later revision.
Depends on D138206
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D138209
Tobias Gysi [Fri, 18 Nov 2022 08:35:15 +0000 (09:35 +0100)]
[mlir][llvm] Make some debug info attribute parameters optional.
The revision makes specific debug information attribute parameters
optional since some of them can be omitted in LLVMIR. The additional
flexibility enables a later revision that will support importing
debug information from LLVMIR. A special case is the types parameter
of the SubroutineTypeAttr. For void functions, its first entry is
null in LLVMIR. This revision splits the type parameter in an optional
resultType parameter and an argumentTypes array to support this corner
case.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D138206
Martin Storsjö [Fri, 18 Nov 2022 08:27:46 +0000 (10:27 +0200)]
[include-cleaner] Fix building with LLVM_LINK_LLVM_DYLIB=ON on mingw
Make sure to use clang_target_link_libraries, which skips linking
against libraries and links against libclang-cpp instead, if
LLVM_LINK_LLVM_DYLIB is enabled.
This fixes errors like these:
ld.lld: error: duplicate symbol: clang::PPCallbacks::~PPCallbacks()
>>> defined at libclangLex.a(PPCallbacks.cpp.obj)
>>> defined at libclang-cpp.dll
Mikhail Goncharov [Fri, 18 Nov 2022 08:25:45 +0000 (09:25 +0100)]
Revert "[opt][clang] Enable using -module-summary/-flto=thin with -S/-emit-llvm"
This reverts commit
34ab474348e2623cc67abddb7bb662385297ac65.
as it has introduced circular dependency lib - analysis
Adrian Kuegel [Fri, 18 Nov 2022 07:46:54 +0000 (08:46 +0100)]
[llvm][Bazel] Remove dependency to avoid cycyle.
Differential Revision: https://reviews.llvm.org/D138266
Jean Perier [Fri, 18 Nov 2022 07:38:11 +0000 (08:38 +0100)]
[flang] Add hlfir.designate definition
Add hlfir.designate operation definition. This is the HLFIR building
blocks to represent Fortran designator. An hlfir.designator is a
Fortran "part-ref" and an optional susbstring or part ref.
See the operation added description for more detail, and
https://github.com/llvm/llvm-project/blob/main/flang/docs/HighLevelFIR.md
for the rational of this operation.
Differential Revision: https://reviews.llvm.org/D138121
Chuanqi Xu [Thu, 17 Nov 2022 07:01:32 +0000 (15:01 +0800)]
[Coroutines] Handle the writes to promise alloca prior to
llvm.coro.begin
Previously we've taken care of the writes to allocas prior to llvm.coro.begin.
However, since the promise alloca is special so that we never handled it
before. For the long time, since the programmers can't access the
promise_type due to the c++ language specification, we still failed to
recognize the problem until a recent report:
https://github.com/llvm/llvm-project/issues/57861
And we've tested many codes that the problem gone away after we handle
the writes to the promise alloca prior to @llvm.coro.begin()
prope until a recent report:
https://github.com/llvm/llvm-project/issues/57861
And we've tested many codes that the problem gone away after we handle
the writes to the promise alloca prior to @llvm.coro.begin() properly.
Closes https://github.com/llvm/llvm-project/issues/57861
Guray Ozen [Fri, 18 Nov 2022 07:38:03 +0000 (08:38 +0100)]
[mlir] [transform] Error for duplicated processor mapping
In a nested loop nest, it is not feasible to map different loops to the same processing unit; for an example, check the code below. This modification includes a check in this circumstance.
```
scf.foreach_thread (%i, %j) in (%c32, %c32) {...}
{ mapping = [#gpu.thread<x>, #gpu.thread<x>] }
```
Note: It also deletes a test because it is not possible to reproduce this error.
Depends on D138020
Reviewed By: ftynse
Differential Revision: https://reviews.llvm.org/D138032
Vitaly Buka [Fri, 18 Nov 2022 07:30:54 +0000 (23:30 -0800)]
[NFC][OpenMP] Remove unused label
Matt Jacobson [Thu, 17 Nov 2022 04:42:15 +0000 (23:42 -0500)]
[Driver] move FreeBSD header search path management to the driver
This matches OpenBSD, and it supports Swift's use of clang for its C interop
functionality. Recent changes to Swift use AddClangSystemIncludeArgs() to
inspect the cc1 args; this doesn't work for platforms where cc1 adds standard
include paths implicitly. See:
<https://github.com/apple/swift/commit/
cf3354222d9f480de74db390f53a6dcc749fde14>
Also clean up InitHeaderSearch, making it clearer which targets manage header
search paths in the driver.
Differential Revision: https://reviews.llvm.org/D138183
Vitaly Buka [Fri, 18 Nov 2022 07:17:15 +0000 (23:17 -0800)]
[NFC][OpenMP] Fix const cast warning
Vitaly Buka [Fri, 18 Nov 2022 07:16:31 +0000 (23:16 -0800)]
[test][OpenMP] Expect more tsan reports
Vitaly Buka [Fri, 18 Nov 2022 07:05:17 +0000 (23:05 -0800)]
[NFC][OpenMP] Remove extra ";"
Vitaly Buka [Fri, 18 Nov 2022 07:03:01 +0000 (23:03 -0800)]
[NFC][OpenMP] Fix warning about non-virtual dtor
Vitaly Buka [Fri, 18 Nov 2022 06:34:30 +0000 (22:34 -0800)]
[NFC][asan] Rename variables in test
luxufan [Fri, 18 Nov 2022 06:56:36 +0000 (14:56 +0800)]
[RegisterScavenger][RISCV] Don't search for FrameSetup instrs if we were searching from Non-FrameSetup instrs
Otherwise, the spill position may point to position where before
FrameSetup instructions. In which case, the spill instruction may store
to caller's frame since the stack pointer has not been adjustted.
Fixes https://github.com/llvm/llvm-project/issues/58286
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D135693
Chen Zheng [Thu, 17 Nov 2022 05:18:04 +0000 (05:18 +0000)]
[PowerPC] mark dead def for ctr be clobber.
TLS pseudo ADDIStlsgdHA will have such def. This dead def should
also prevent PPC from generating CTR loops.
Martin Storsjö [Tue, 8 Nov 2022 09:45:41 +0000 (11:45 +0200)]
[openmp] [test] Fix warnings about printf format mismatches on Windows
This fixes warnings like this:
```
openmp/runtime/test/omp_testsuite.h:107:60: warning: format specifies type 'unsigned int' but the argument has type 'DWORD' (aka 'unsigned long') [-Wformat]
fprintf(stderr, "CreateThread() failed: Error #%u.\n", GetLastError());
~~ ^~~~~~~~~~~~~~
%lu
```
Differential Revision: https://reviews.llvm.org/D137747
Martin Storsjö [Tue, 8 Nov 2022 13:31:40 +0000 (15:31 +0200)]
[openmp] Only try to add the -mrtm flag on i386/x86_64
Clang does accept the flag for any architecture, but prints a
warning:
clang-16: warning: argument unused during compilation: '-mrtm' [-Wunused-command-line-argument]
This flag is documented as an x86 specific flag, so don't try to
add it for other architectures, to silence this warning.
Differential Revision: https://reviews.llvm.org/D137744
Han-Kuan Chen [Tue, 15 Nov 2022 11:34:44 +0000 (03:34 -0800)]
[RISCV] Make lowerVECTOR_SHUFFLEAsVSlidedown follow source until not EXTRACT_SUBVECTOR.
Current lowerVECTOR_SHUFFLEAsVSlidedown only seeks whether input are
EXTRACT_SUBVECTOR and their source are same. The commit will make the
function seek input and their source until they are not
EXTRACT_SUBVECTOR.
Differential Revision: https://reviews.llvm.org/D138025
Han-Kuan Chen [Tue, 15 Nov 2022 12:02:41 +0000 (04:02 -0800)]
[RISCV] Pre-commit test.
Differential Revision: https://reviews.llvm.org/D138024
Vitaly Buka [Fri, 18 Nov 2022 06:29:19 +0000 (22:29 -0800)]
[asan] Simplify the test
Mehdi Amini [Thu, 17 Nov 2022 09:09:30 +0000 (09:09 +0000)]
Apply clang-tidy fixes for llvm-else-after-return in SparseTensorCodegen.cpp (NFC)
Mehdi Amini [Thu, 17 Nov 2022 08:06:15 +0000 (08:06 +0000)]
Apply clang-tidy fixes for bugprone-argument-comment in LinalgTransformOps.cpp (NFC)
Mehdi Amini [Wed, 16 Nov 2022 21:29:28 +0000 (21:29 +0000)]
MLIR: Add --fix-error to clang-tidy script
Alexander Shaposhnikov [Fri, 18 Nov 2022 04:26:08 +0000 (04:26 +0000)]
[opt][clang] Enable using -module-summary/-flto=thin with -S/-emit-llvm
Enable using -module-summary with -S
(similarly to what currently can be achieved with opt <input> -o - | llvm-dis).
This is a recommit of
ef9e62469.
Test plan: ninja check-all
Differential revision: https://reviews.llvm.org/D137768
Fazlay Rabbi [Fri, 18 Nov 2022 00:20:14 +0000 (16:20 -0800)]
[OpenMP] Initial parsing/sema for 'strict' modifier with 'grainsize' clause
This patch gives basic parsing and semantic analysis support for 'strict'
modifier with 'grainsize' clause of 'taskloop' construct introduced in
OpenMP 5.1 (section 2.12.2)
Differential Revision: https://reviews.llvm.org/D138217
Lei Zhang [Fri, 18 Nov 2022 04:46:18 +0000 (23:46 -0500)]
[mlir][spirv] Support attribute in MapMemRefStorageClassPass
MemRef memory space actually can be an attribute. Update the
map function signature to accept an attribute. The default
mappings can still only covers numeric ones, but this allows
downstream callers to extend with custom memory spaces.
Reviewed By: kuhar
Differential Revision: https://reviews.llvm.org/D138257
Matt Arsenault [Sun, 31 May 2020 17:23:20 +0000 (13:23 -0400)]
AMDGPU/GlobalISel: Make strict fadd, fmul and fma legal
Timm Bäder [Thu, 17 Nov 2022 14:49:30 +0000 (15:49 +0100)]
[clang][Parse][NFC] Remove unused CommaLocs parameters
Timm Bäder [Thu, 17 Nov 2022 11:12:11 +0000 (12:12 +0100)]
[clang][Parser][NFC] Simplify ParseParenExprOrCondition
Differential Revision: https://reviews.llvm.org/D138194
Chuanqi Xu [Fri, 18 Nov 2022 03:49:33 +0000 (11:49 +0800)]
[NFC] Remove unused codes after
4a7be42d922af
There are still some unused codes remained after
4a7be42d922af. And this
patch tries to remove them.
Chuanqi Xu [Fri, 18 Nov 2022 02:45:08 +0000 (10:45 +0800)]
[C++20] [Modules] Don't emit macro definitions with named module
It is meaningless to emit macro definitions for named modules. With some
small experiments, the size of the module for the named modules reduced
2%~4% after this patch.
Matt Arsenault [Fri, 18 Nov 2022 02:05:30 +0000 (18:05 -0800)]
AMDGPU/GlobalISel: Fix not selecting modifiers for f16 fma on gfx9
VOP3OpSel wasn't trying to match any modifiers. Just try to match the
basic case, like the DAG does.
YingChi Long [Tue, 1 Nov 2022 16:08:40 +0000 (00:08 +0800)]
[VP] Add support for vp.inttoptr & vp.ptrtoint
Add vp.inttoptr & vp.ptrtoint support by lowering them into
vp.zext / vp.truncate with in SelectionDAGBuilder.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D137169
Chuanqi Xu [Fri, 18 Nov 2022 02:13:40 +0000 (10:13 +0800)]
[C++20] [Modules] Remove unmaintained Header Module
Currently there is a -emit-header-module mode, which can combine several
headers together as a module interface. However, this breaks our
assumption (for standard c++ modules) about module interface. The module
interface should come from a module interface unit. And if it is a
header, it should be a header unit. And currently we have no ideas to
combine several headers together.
So I think this mode is an experimental one and it is not maintained and
it is not used. So it will be better to remove them.
Reviewed By: Bigcheese, dblaikie, bruno
Differential Revision: https://reviews.llvm.org/D137609
Qiu Chaofan [Fri, 18 Nov 2022 02:28:24 +0000 (10:28 +0800)]
[Clang] Fix behavior of -ffp-model option when overriden
-ffp-model=strict -ffp-model=fast will still enable strict exception
handling behavior, therefore clang still emits constrained FP operations
in IR.
-ffp-model=fast -ffp-model=strict emits two warnings: one for strict
overriding fast, the other for strict overriding strict, which is
confusing.
Reviewed By: zahiraam
Differential Revision: https://reviews.llvm.org/D137618
Volodymyr Sapsai [Fri, 8 Jul 2022 20:10:31 +0000 (13:10 -0700)]
[ODRHash] Detect duplicate `ObjCProtocolDecl` ODR mismatches during parsing.
When during parsing we encountered a duplicate `ObjCProtocolDecl`, we
were always emitting an error. With this change we accept
* when a previous `ObjCProtocolDecl` is in a hidden [sub]module;
* parsed `ObjCProtocolDecl` is the same as the previous one.
And in case of mismatches we provide more detailed error messages.
rdar://
93069080
Differential Revision: https://reviews.llvm.org/D130327
Alexander Shaposhnikov [Fri, 18 Nov 2022 01:47:56 +0000 (01:47 +0000)]
[IR] Split out IR printing passes into IRPrinter
This diff splits out (from LLVMCore) IR printing passes into IRPrinter.
This structure is similar to what we already have for IRReader and
enables us to avoid circular dependencies between LLVMCore and Analysis
(this is a preparation for https://reviews.llvm.org/D137768).
The legacy interface is left unchanged, once the legacy pass manager
is removed (in the future) we will be able to clean it up further.
The bazel build configuration has been updated as well.
Test plan:
1/ Tested the following cmake configurations: static/dynamic linking * lld/gold * clang/gcc
2/ bazel build --config=generic_clang @llvm-project//...
Differential revision: https://reviews.llvm.org/D138081
Alexey Bataev [Fri, 18 Nov 2022 00:46:51 +0000 (16:46 -0800)]
[SLP]Fix PR59053: trying to erase instruction with users.
Need to count the reduced values, vectorized in the tree but not in the top node. Such scalars still must be extracted out of the vector node instead of the original scalar.
Fangrui Song [Fri, 18 Nov 2022 01:23:12 +0000 (01:23 +0000)]
[Sink] Process basic blocks with a single successor
This condition seems unnecessary.
Reviewed By: arsenm
Differential Revision: https://reviews.llvm.org/D93511
Volodymyr Sapsai [Wed, 6 Jul 2022 21:58:26 +0000 (14:58 -0700)]
[ODRHash] Hash `ObjCPropertyDecl` and diagnose discovered mismatches.
Differential Revision: https://reviews.llvm.org/D130326
Diego Caballero [Fri, 18 Nov 2022 01:18:11 +0000 (01:18 +0000)]
Revert "[mlir][Vector] Re-define masking semantics in vector.transfer ops"
This reverts commit
6c59c5cd08c879c9d1cfa653711613244a7c39bf.
Konstantin Varlamov [Fri, 18 Nov 2022 01:13:53 +0000 (17:13 -0800)]
[libc++][NFC] Add a missing include in pre-C++14 language modes.
Diego Caballero [Wed, 16 Nov 2022 00:43:50 +0000 (00:43 +0000)]
[mlir][Vector] Re-define masking semantics in vector.transfer ops
Masking hasn't been widely used in vector transfer ops and the semantics
of the mask operand were a bit loose. This patch states that the mask
operand in a vector transfer op is applied to the read/write part of the
operation and, therefore, its shape should match the shape of the
elements read/written from/into the memref/tensor regardless of any
permutation/broadcasting also applied by the transfer operation.
Reviewers: nicolasvasilache
Differential Revision: https://reviews.llvm.org/D138079
Fangrui Song [Fri, 18 Nov 2022 00:58:04 +0000 (00:58 +0000)]
[CMake] Remove a duplicate clangLex entry
Fangrui Song [Fri, 18 Nov 2022 00:56:04 +0000 (00:56 +0000)]
[include-cleaner] Fix -DBUILD_SHARED_LIBS=on build
and apply an include-cleaner fix for the tool itself:)
addPPCallbacks requires the vtable of PPChainedCallbacks and needs a clangLex
dependency.
Alexander Shaposhnikov [Fri, 18 Nov 2022 00:53:56 +0000 (00:53 +0000)]
[include-cleaner] Fix the build of include-cleaner with LLVM_BUILD_LLVM_DYLIB=ON
Konstantin Varlamov [Fri, 18 Nov 2022 00:51:53 +0000 (16:51 -0800)]
[libc++] Fix a missing include in pre-C++17 language modes.
Brad Smith [Fri, 18 Nov 2022 00:46:23 +0000 (19:46 -0500)]
[Linux] Revert
1e56821bac02a5d3c6249bbf3ef43b8b569d2551
The glibc issue mentioned in #47994 has been fixed upstream.
David Blaikie [Fri, 18 Nov 2022 00:12:42 +0000 (00:12 +0000)]
Update lambda mangling test to C++17
add some side effects so some (I guess guaranteed?) constant folding
doesn't happen, keeping the test testing the things it was testing..
The test passes with 14, 17, and 20 - so let's just leave the version
off so it might be able to be updated/used in C++20 when the default
changes to C++20 in the future.
Jennifer Yu [Wed, 16 Nov 2022 21:29:14 +0000 (13:29 -0800)]
[OPENMP5.1] Initial support for severity clause
Differential Revision:https://reviews.llvm.org/D138227
Yabin Cui [Fri, 18 Nov 2022 00:03:46 +0000 (00:03 +0000)]
[OpenMP] kmp_affinity.h: add RISCV64 support
In D135552 the #else is added, which causes build error when
building openmp on RISCV64. This patch fixed the error:
"Unknown or unsupported architecture"
Reviewed By: pirama
Differential Revision: https://reviews.llvm.org/D138241
Krzysztof Parzyszek [Wed, 16 Nov 2022 16:57:01 +0000 (08:57 -0800)]
[Hexagon] Add instruction definitions for Hexagon v71, v71t, and v73
This includes instruction formats, definitions, encodings, scheduling
classes, and builtins/intrinsics.
New and improved version of
76536989ba, so much so that even clang
builds with it.
Michael Jones [Wed, 19 Oct 2022 20:36:59 +0000 (13:36 -0700)]
[libc] add scanf entrypoints
This patch adds scanf, sscanf, and fscanf entrypoints. It also adds unit
tests for sscanf and a basic test to fscanf. The scanf function is
basically impossible to test in an automated fashion due to it recieving
user input.
Reviewed By: sivachandra, lntue
Differential Revision: https://reviews.llvm.org/D138076
Doru Bercea [Wed, 16 Nov 2022 23:23:48 +0000 (17:23 -0600)]
Fix declare target implementation to support enter.
Benjamin Kramer [Thu, 17 Nov 2022 23:26:08 +0000 (00:26 +0100)]
[Affine] Replace a magic -1 with ShapedType::kDynamicSize
Adrian Prantl [Thu, 17 Nov 2022 23:24:35 +0000 (15:24 -0800)]
Comment function
Matt Arsenault [Thu, 17 Nov 2022 23:07:23 +0000 (15:07 -0800)]
AMDGPU: Convert a test to opaque pointers
Fangrui Song [Thu, 17 Nov 2022 23:14:32 +0000 (23:14 +0000)]
Revert "[Hexagon] Add instruction definitions for Hexagon v71, v71t, and v73"
This reverts commit
766536989baf9431ff535670d01c4c19d7351fad.
The commit caused:
clang/include/clang/Basic/BuiltinsHexagonDep.def:1896:69: error: use of undeclared identifier 'HVXV73'
TARGET_BUILTIN(__builtin_HEXAGON_V6_vadd_sf_bf, "V32iV16iV16i", "", HVXV73)
when building `clang`.
v1nh1shungry [Thu, 17 Nov 2022 21:30:35 +0000 (21:30 +0000)]
[clang-tidy] Fix misc-unused-using-decls for user-defined literals
Current version complains unused using-declaration even if the target
user-defined literal is used.
Reviewed By: ymandel
Differential Revision: https://reviews.llvm.org/D138204
Fangrui Song [Thu, 17 Nov 2022 22:45:29 +0000 (22:45 +0000)]
Revert D135411 "Add generic KCFI operand bundle lowering"
This reverts commit
eb2a57ebc7aaad551af30462097a9e06c96db925.
llvm/include/llvm/Transforms/Instrumentation/KCFI.h including
llvm/CodeGen is a layering violation. We should use an approach where
Instrumementation/ doesn't need to include CodeGen/.
Sorry for not spotting this in the review.
Craig Topper [Thu, 17 Nov 2022 22:03:05 +0000 (14:03 -0800)]
[X86] Use correctly sized floating point literals in *zero_ps/pd.
This avoids depending on int->float or double->float conversion.
Improving codegen with #pragma STDC FENV_ACCESS ON.
Really we should improve constant folding somewhere, but this was
a cheap and easy improvement.
Fixes PR59052.
Murali Vijayaraghavan [Sat, 5 Nov 2022 03:50:40 +0000 (03:50 +0000)]
Simplifying the SplitReduction logic that uses the control to get the
dimension where the extra parallel dimension is inserted
Currently, the innerParallel and non innerParallel strategies use two
different ways to fix for where the extra loop is inserted and where the
extra dimension for the intermediate result is inserted - innerParallel
adds the extra (parallel) loop right after the pre-existing reduction
loop, whereas non innerParallel adds the reduction loop in the successor
to the index supplied by control, and the parallel loop in the index
supplied by the control. The semantics of the index supplied by the
control is supposed to only control where the extra tensor dimension is
inserted in the intermediate tensor. Conflating this index with where
the reduction (and parallel) loops are inserted leads to more complex
(and confusing) logic overall. This differential removes conflating the
two uses of the index, and keeps the reduction and parallel loops in the
same vicinity and uses the supplied index to only determine the position
of the extra tensor dimension. It also simplifies the code by merging
the two strategies in a lot more places.
Differential Revision: https://reviews.llvm.org/D137478
Krzysztof Parzyszek [Wed, 16 Nov 2022 16:57:01 +0000 (08:57 -0800)]
[Hexagon] Add instruction definitions for Hexagon v71, v71t, and v73
This includes instruction formats, definitions, encodings, scheduling
classes, and builtins/intrinsics.
Krzysztof Parzyszek [Thu, 17 Nov 2022 21:56:50 +0000 (13:56 -0800)]
[Hexagon] Add ELF flags for Hexagon v71, v71t, and v73
Sami Tolvanen [Thu, 18 Aug 2022 23:53:09 +0000 (23:53 +0000)]
Add generic KCFI operand bundle lowering
The KCFI sanitizer emits "kcfi" operand bundles to indirect
call instructions, which the LLVM back-end lowers into an
architecture-specific type check with a known machine instruction
sequence. Currently, KCFI operand bundle lowering is supported only
on 64-bit X86 and AArch64 architectures.
As a lightweight forward-edge CFI implementation that doesn't
require LTO is also useful for non-Linux low-level targets on
other machine architectures, add a generic KCFI operand bundle
lowering pass that's only used when back-end lowering support is not
available and allows -fsanitize=kcfi to be enabled in Clang on all
architectures.
Reviewed By: nickdesaulniers, MaskRay
Differential Revision: https://reviews.llvm.org/D135411
Ben Langmuir [Thu, 17 Nov 2022 00:07:34 +0000 (16:07 -0800)]
[clang] Use InMemoryModuleCache for readASTFileControlBlock NFC
When a pcm has already been loaded from disk, reuse it from the
InMemoryModuleCache in readASTFileControlBlock. This avoids potentially
reading it again.
As noted in the FIXME, ideally we would also add the module to the cache
if it will be used again later, but that could modify its build state
and we do not have enough context currenlty to know if it's correct.
Differential Revision: https://reviews.llvm.org/D138160
usama hameed [Mon, 7 Nov 2022 01:36:57 +0000 (17:36 -0800)]
[UBSan] Implementation NonUniqueARMRTTIBit __type_name has higher bits set on arm64
Differential Revision: https://reviews.llvm.org/D137518
Roman Lebedev [Thu, 17 Nov 2022 21:40:50 +0000 (00:40 +0300)]
[NFC][SROA] Add more tests with non-speculatable `load`s of `select`s
bixia1 [Thu, 17 Nov 2022 20:25:41 +0000 (12:25 -0800)]
[mlir][sparse] Make three integration tests run with the codegen path.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D138233
bixia1 [Thu, 17 Nov 2022 18:19:57 +0000 (10:19 -0800)]
[mlir][sparse] Fix a problem in the new operator rewriter.
The getSparseTensorReaderNextX functions should return void.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D138226
Krzysztof Parzyszek [Wed, 16 Nov 2022 23:30:12 +0000 (15:30 -0800)]
[Hexagon] Improve inserting/extracting to/from scalar predicates
Fixes https://github.com/llvm/llvm-project/issues/59042.
Matt Arsenault [Fri, 11 Nov 2022 20:29:48 +0000 (12:29 -0800)]
LangRef: Fix repeated word
Krzysztof Parzyszek [Thu, 17 Nov 2022 19:54:51 +0000 (11:54 -0800)]
[Hexagon] Fix lowering loads/stores of scalar vNi1
Don't treat them as i1, all predicate bits need to be loaded or stored.
Krzysztof Parzyszek [Thu, 17 Nov 2022 18:55:00 +0000 (10:55 -0800)]
[Hexagon] Use BUILD_PAIR instead of HexagonISD::COMBINE in lowering
Martin Storsjö [Tue, 8 Nov 2022 09:37:15 +0000 (11:37 +0200)]
[openmp] [test] Unbreak tests that rely on hidden_helper_task on Windows
The hidden helper task is only enabled on Linux (kmp_runtime.cpp
initializes __kmp_enable_hidden_helper to TRUE for linux but to
FALSE for any other OS), and the __kmp_stg_parse_use_hidden_helper
function always makes it disabled on non-Linux OSes too.
Add a lit test feature for hidden helper tasks (only made available
on Linux) and mark two tests as requiring this feature.
Disable hidden helper tasks in the test that doesn't really involve
them, for consistent behaviour across platforms.
Differential Revision: https://reviews.llvm.org/D137749
Guillaume Chatelet [Thu, 17 Nov 2022 20:20:57 +0000 (20:20 +0000)]
Revert D119547 "[libc][bazel] Add tests to the bazel build"
The introducion of `constraint_setting` generated issues with downstream
integration.
This reverts commit
3c13d83ad59b5932328c0a99b0a0008e1da8b1d8.
Vaibhav Yenamandra [Thu, 17 Nov 2022 19:47:02 +0000 (14:47 -0500)]
Refactor StaticAnalyzer to use `clang::SarifDocumentWriter`
Refactor StaticAnalyzer to use clang::SarifDocumentWriter for
serializing sarif diagnostics.
Uses clang::SarifDocumentWriter to generate SARIF output in the
StaticAnalyzer.
Various bugfixes are also made to clang::SarifDocumentWriter.
Summary of changes:
clang/lib/Basic/Sarif.cpp:
* Fix bug in adjustColumnPos introduced from prev move, it now uses
FullSourceLoc::getDecomposedExpansionLoc which provides the correct
location (in the presence of macros) instead of
FullSourceLoc::getDecomposedLoc.
* Fix createTextRegion so that it handles caret ranges correctly,
this should bring it to parity with the previous implementation.
clang/test/Analysis/diagnostics/Inputs/expected-sarif:
* Update the schema URL to the offical website
* Add the emitted defaultConfiguration sections to all rules
* Annotate results with the "level" property
clang/lib/StaticAnalyzer/Core/SarifDiagnostics.cpp:
* Update SarifDiagnostics class to hold a clang::SarifDocumentWriter
that it uses to convert diagnostics to SARIF.
Artem Belevich [Thu, 17 Nov 2022 19:35:34 +0000 (11:35 -0800)]
[CUDA] make use of deprecated texture API conditional on CUDA version.
Peiming Liu [Thu, 17 Nov 2022 17:49:23 +0000 (17:49 +0000)]
[mlir][sparse] fix crash on sparse_tensor.foreach operation on tensors with complex<T> elements.
Reviewed By: aartbik, bixia
Differential Revision: https://reviews.llvm.org/D138223
Evgenii Stepanov [Thu, 3 Nov 2022 19:42:12 +0000 (12:42 -0700)]
[scudo] Extend basic test to cover zero size allocs.
Differential Revision: https://reviews.llvm.org/D137354
Jakub Kuderski [Thu, 17 Nov 2022 19:10:16 +0000 (14:10 -0500)]
[bazel][mlir] Add missing dependency for D138178
Jakub Kuderski [Thu, 17 Nov 2022 19:01:50 +0000 (14:01 -0500)]
[mlir][arith] Add `index_cast` and `index_castui` support to WIE
Reviewed By: Mogball
Differential Revision: https://reviews.llvm.org/D138225
Sami Tolvanen [Wed, 16 Nov 2022 23:08:00 +0000 (23:08 +0000)]
[X86][KCFI] Add support for memory operand unfolding
When the Linux kernel is compiled without -mretpoline, KCFI fails
ungracefully because it doesn't handle indirect calls with a memory
target operand. Since the KCFI check will need to load the target
address into a register for validating the type hash anyway, simply
unfold memory operands in indirect calls that need a KCFI check.
Fixes #59017
Sanjay Patel [Thu, 17 Nov 2022 18:35:23 +0000 (13:35 -0500)]
[Passes] reduce duplicated CHECK lines in tests; NFC
I missed this file when updating similarly with
f44e8464027c.
Adrian Prantl [Thu, 17 Nov 2022 18:49:44 +0000 (10:49 -0800)]
Attempt ro fix Windows build errors.
bixia1 [Thu, 17 Nov 2022 16:16:19 +0000 (08:16 -0800)]
[mlir][sparse] Extend the operator new rewriter to handle isSymmetric flag.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D138214
Jakub Kuderski [Thu, 17 Nov 2022 18:45:01 +0000 (13:45 -0500)]
[mlir][spirv][math] Fix crash on unsupported types in math-to-spirv
Fail to match conversion patterns when source op has unsupported types.
Fixes: https://github.com/llvm/llvm-project/issues/58749
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D138178
Tomas Matheson [Mon, 24 Oct 2022 13:25:37 +0000 (14:25 +0100)]
[ARM][AArch64] Use StringRef in TargetParser structs
The invalid case is now represented by an empty StringRef rather than
a nullptr.
Previously ARCH_FEATURE was build from SUB_ARCH by prepending "+".
This is now reverse, so that the "+arch-feature" is now visible in
the .def, which is a bit clearer. This meant converting one StringSwitch
into a loop.
Removed getters which are now mostly unnecessary.
Removed some old FIXMEs.
Differential Revision: https://reviews.llvm.org/D138026
Mengxuan Cai [Thu, 17 Nov 2022 18:33:37 +0000 (13:33 -0500)]
[LoopInterchange] Refactor and rewrite validDepInterchange()
The current code of validDepInterchange() enumerates cases that are
legal for interchange. This could be simplified by checking
lexicographically order of the swapped direction matrix.
Reviewed By: congzhe, Meinersbur, bmahjour
Differential Revision: https://reviews.llvm.org/D137461
Jakub Kuderski [Thu, 17 Nov 2022 18:36:02 +0000 (13:36 -0500)]
[mlir][arith] Add Max*I and Min*I support to WIE
This includes LIT IR tests and runtime tests.
Reviewed By: Mogball
Differential Revision: https://reviews.llvm.org/D138184
Roman Lebedev [Thu, 17 Nov 2022 18:17:56 +0000 (21:17 +0300)]
[Pipelines] Introduce SROA after (final, run-time) loop unrolling
Now that we are done with loop unrolling, be it either by LoopVectorizer,
or LoopUnroll passes, some variable-offset GEP's into alloca's could have
become constant-offset, thus enabling SROA and alloca promotion,
yet we don't capitalize on that, which is surprizing.
While it would be good to not introduce one more SROA invocation,
but instead move the one from `PassBuilder::buildFunctionSimplificationPipeline()`,
the existing test coverage says that is a bad idea,
though it would be fine compile-time wise: https://llvm-compile-time-tracker.com/compare.php?from=
b150d34c47efbd8fa09604bce805c0920360f8d7&to=
5a9a5c855158b482552be8c7af3e73d67fa44805&stat=instructions
So instead, i add yet another SROA run.
I have checked, and it needs to be at least after said final loop unrolling.
This is still fine compile-time wise: https://llvm-compile-time-tracker.com/compare.php?from=
70324cd88328c0924e605fa81b696572560aa5c9&to=
fb489bbef687ad821c3173a931709f9cad9aee8a&stat=instructions
I've encountered this in a real code, `SROA-after-final-loop-unrolling.ll` has been reduced from https://godbolt.org/z/fsdMhETh3
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D136806
Slava Zakharin [Tue, 15 Nov 2022 19:09:59 +0000 (11:09 -0800)]
[flang] Propagate fastmath flags during intrinsics simplification.
In general, the meaning of fastmath flags on a call during inlining
is that the call's operation flags must be ignored. For user functions
that means that the fastmath flags used for the function definition
override any call site's fastmath flags. For intrinsic functions
we can use the call site's fastmath flags, but we have to make sure
that the call sites with different flags produce/use different
simplified versions of the same intrinsic function.
Differential Revision: https://reviews.llvm.org/D138048
Sanjay Patel [Thu, 17 Nov 2022 17:54:58 +0000 (12:54 -0500)]
[Passes] reduce duplicated CHECK lines in tests; NFC
bixia1 [Thu, 17 Nov 2022 15:32:58 +0000 (07:32 -0800)]
[mlir][sparse] Run an integration test with codegen.
Reviewed By: Peiming
Differential Revision: https://reviews.llvm.org/D138213
Roman Lebedev [Thu, 17 Nov 2022 17:23:51 +0000 (20:23 +0300)]
[Analysis] `isSafeToLoadUnconditionally()`: `lifetime` intrinsics can be ignored
In practice this means that we can speculate more loads in SROA.
This e.g. comes up in https://godbolt.org/z/G8716s6sj,
although we are missing second half of the puzzle to optimize that.
Roman Lebedev [Thu, 17 Nov 2022 17:02:12 +0000 (20:02 +0300)]
[NFC][SROA] Add test for select speculation failures
Stanislav Mekhanoshin [Thu, 21 Apr 2022 23:23:11 +0000 (16:23 -0700)]
[AMDGPU] Allow finer grain control of an unaligned access speed
A target can return if a misaligned access is 'fast' as defined
by the target or not. In reality there can be different levels
of 'fast' and 'slow'. This patch changes the boolean 'Fast'
argument of the allowsMisalignedMemoryAccesses family of functions
to an unsigned representing its speed.
A target can still define it as it wants and the direct translation
of the current code uses 0 and 1 for current false and true. This
makes the change an NFC.
Subsequent patch will start using an actual value of speed in
the load/store vectorizer to compare if a vectorized access going
to be not just fast, but not slower than before.
Differential Revision: https://reviews.llvm.org/D124217