Chia-hung Duan [Thu, 19 Jan 2023 17:52:12 +0000 (17:52 +0000)]
[scudo] Calling getStats requires holding lock
We didn't acquire the mutex while accessing those lock protected data,
this CL fixes it and now we don't need to disable the allocator while
reading its states.
Differential Revision: https://reviews.llvm.org/D142149
Peter Klausler [Sat, 11 Feb 2023 19:10:58 +0000 (11:10 -0800)]
[flang] Conform with standard (mostly) for character length mismatches on arguments
Fortran 2018 defines some flavors of dummy arguments to require exact
matching of character lengths between dummy and actual arguments;
these situations tend to be those in which the interface must be
explicit and a descriptor is involved: assumed shape, assumed rank,
allocatable, and pointer.
Fortran allows an actual argument in other cases to have a longer
length than the dummy argument; as a common extension, we support a
shorter actual argument as well by means of blank padding, but should
emit a warning.
Differential Revision: https://reviews.llvm.org/D143821
bixia1 [Mon, 13 Feb 2023 20:43:03 +0000 (12:43 -0800)]
[mlir][sparse] Extend readCOOIndices to support overhead types beyond index_type.
This is to prepare for implementing the C API for reading a COO tensor to the
given buffers for indices and values.
Reviewed By: wrengr
Differential Revision: https://reviews.llvm.org/D143861
Peter Klausler [Tue, 7 Feb 2023 01:08:06 +0000 (17:08 -0800)]
[flang] Diagnose REPEAT with negative NCOPIES=
Emit an error when the NCOPIES= argument to the intrinsic function
REPEAT has a negative value. (The current implementation crashes, which
isn't informative.)
Differential Revision: https://reviews.llvm.org/D143820
Peiming Liu [Mon, 13 Feb 2023 17:46:06 +0000 (17:46 +0000)]
[mlir][sparse] fix a memory leakage when converting from a tensor slice
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D143929
Joseph Huber [Mon, 13 Feb 2023 22:38:59 +0000 (16:38 -0600)]
[libc] Write stub files to a new directory to avoid conflicts
Summary:
This hack with stub files is used to make the final object archive
have human-understandable names. We currently output these into the
current binary directory, which sometimes interferes with the actual
source file. Put these in their own directory to be certain they don't
conflict.
Peter Klausler [Mon, 13 Feb 2023 22:22:02 +0000 (14:22 -0800)]
[flang][build] Fix build issue reported on recent commit
Some compiler (not specified) reported to issue an error on a
"default:" clause in a switch statement whose cases cover all of
the values of an "enum class". Since other compilers/versions
are known to complain in the other direction, change the switch
statement to a cascade of ifs.
Alexander Yermolovich [Mon, 13 Feb 2023 22:08:40 +0000 (14:08 -0800)]
Revert "[LLDB] Enable 64 bit debug/type offset"
This reverts commit
f36fe009c0fc1d655bfc6168730bedfa1b36e622.
Nawrin Sultana [Wed, 25 Jan 2023 22:01:02 +0000 (16:01 -0600)]
[OpenMP] Add check for target allocator regardless of the availability of libmemkind
Current runtime implementation only checks for target allocator when libmemkind is
not available. This patch adds checks for target allocator regardless of the
presence of libmemkind library.
Differential Revision: https://reviews.llvm.org/D142582
Jakub Kuderski [Mon, 13 Feb 2023 21:51:37 +0000 (16:51 -0500)]
[mlir][tosa] Enable `apply_scale` unrolling
Make `tosa.apply_scale` implement `VectorUnrollOpInterface` so that we
can unroll it.
Reviewed By: rsuderman
Differential Revision: https://reviews.llvm.org/D143944
Peter Jung [Mon, 13 Feb 2023 22:00:22 +0000 (14:00 -0800)]
[Release] Produce bolt tarball
Source tarball's are used from some distribution to build the project
Reviewed By: Amir
Differential Revision: https://reviews.llvm.org/D143809
Benjamin Kramer [Mon, 13 Feb 2023 21:45:21 +0000 (22:45 +0100)]
[lldb] Use portable format string PRIx64
Fanbo Meng [Mon, 13 Feb 2023 15:31:49 +0000 (10:31 -0500)]
[runtimes] Set LLVM_ENABLE_PER_TARGET_RUNTIME_DIR_default to ON for OS390
Reviewed By: abhina.sreeskantharajan, zibi
Differential Revision: https://reviews.llvm.org/D143916
Kévin Petit [Mon, 13 Feb 2023 21:24:13 +0000 (21:24 +0000)]
libclc: remove sqrt/rsqrt from clspv SOURCES
https://reviews.llvm.org/D134040
Patch by: Aaron Greig <aaron.greig@codeplay.com>
Florian Hahn [Mon, 13 Feb 2023 21:24:57 +0000 (21:24 +0000)]
[VPlan] Use properlyDominates predicate for ordering FOR users.
The current implementation may return true for A < B and B < A, which
may cause issues if the sort implementation assures this property of the
comperator. This should fix a crash with MSVC.
Markus Böck [Mon, 13 Feb 2023 16:32:08 +0000 (17:32 +0100)]
[mlir] Make the vast majority of integration and runner tests work on Windows
This patch contains the changes required to make the vast majority of integration and runner tests run on Windows.
Historically speaking, the JIT support for Windows has been lacking behind, but recent versions of ORC JIT have now caught up and works for basically all examples in repo.
Sadly due to these tests previously not working on Windows, basically all of them are making unix-like assumptions about things like filenames, paths, shell syntax etc.
This patch fixes all these issues in one big swoop and enables Windows support for the vast majority of integration tests.
More specifically, following changes had to be done:
* The various JIT runners used paths to the runtime libraries that assumed a Unix toolchain layout and filenames. I abstracted the specific path and filename of these runtime libraries away by making the paths to the runtime libraries be passed from cmake into lit. This now also allows a much more convenient syntax: `--shared-libs=%mlir_c_runner_utils` instead of `--shared-libs=%mlir_lib_dir/lib/libmlir_c_runner_utils%shlibext`
* Some tests using python set environment variables using the `ENV=VALUE cmd` format. This works on Unix, but on Windows it has to prefixed using `env ENV=VALUE cmd`
* Some tests used C functions that are simply not available or exported on Windows (`fabsf`, `aligned_alloc`). These tests have either been adjusted or explicitly marked as `UNSUPPORTED`
Some tests remain disabled on Windows as before:
* In SparseTensor some tests have non-trivial logic for finding the runtime libraries which seems to be required for the use of emulators. I do not have the time to port these so I simply kept them disabled
* Some tests requiring special hardware which I simply cannot test remain disabled on Windows. These include usage of AVX512 or AMX
The tests for `mlir-vulkan-runner` and `mlir-spirv-runner` all work now as well and so do the vast majority of `mlir-cpu-runner`.
Differential Revision: https://reviews.llvm.org/D143925
Markus Böck [Mon, 13 Feb 2023 11:39:02 +0000 (12:39 +0100)]
[mlir][SPIRVToLLVM] Add pass option to emit opaque-pointers
Part of https://discourse.llvm.org/t/rfc-switching-the-llvm-dialect-and-dialect-lowerings-to-opaque-pointers/68179
This patch adds the pass option and required changes to the patterns to support the emission of LLVM IR opaque pointers. Given how close SPIRV semantics are to LLVM IR semantics this boils down to just a few changes:
* Making sure that GEP and alloca are built with the explicit base pointer type
* creating opaque pointers instead of typed pointers if requested
* omitting pointer to pointer bitcasts
Differential Revision: https://reviews.llvm.org/D143900
Markus Böck [Sun, 12 Feb 2023 22:55:47 +0000 (23:55 +0100)]
[mlir][LLVM] Verify correct pointer casts with `llvm.bitcast`
`llvm.bitcast` has so far not had a verifier and this allowed various bugs to sneak into the codebase (including within tests!) which could only be caught once translated to actual LLVM IR. This patch fixes those problematic cases by now verifying bitcasts on pointers are done correctly.
Specifically, it verifies that if pointers are involved, that both result and source types are pointers, that this also applies to vector of pointers and that pointer casts are of the same address space.
The only thing left unverified is the general case of "source type size does not match result type size". I think this case is less trivial and more prone to false positives, so I did not yet implement it.
Differential Revision: https://reviews.llvm.org/D143868
Markus Böck [Sun, 12 Feb 2023 22:52:16 +0000 (23:52 +0100)]
[mlir][GPU] add required address space cast when lowering to LLVM
The runtime functions `memset` and `memcpy` are lowered are declared with pointers to the default address space (0) while their ops however are compatible with memrefs taking any address space.
Such cases do not cause any issues with MLIRs LLVM Dialect due to `bitcast`s verifier being too lenient at the moment, but actual LLVM IR does not allow casting between address spaces using `bitcast`: https://godbolt.org/z/3a1z97rc9
This patch fixes the issue by inserting an address space cast before the bitcast, to first cast the pointer into the correct address space before doing the bitcast.
Differential Revision: https://reviews.llvm.org/D143866
Kai Sasaki [Mon, 13 Feb 2023 20:54:42 +0000 (12:54 -0800)]
[mlir][tosa] Fix segmentation fault in case of folding unranked tensor
Trying to fold the unranked tensor for "tosa.equal" crashes due to null reference.
We need to check the dynamic cast result beforehand. This is reported in
https://github.com/llvm/llvm-project/issues/60192.
Reviewed By: rsuderman
Differential Revision: https://reviews.llvm.org/D143034
Alexander Yermolovich [Mon, 13 Feb 2023 20:49:32 +0000 (12:49 -0800)]
[LLDB] Enable 64 bit debug/type offset
This came out of from https://discourse.llvm.org/t/dwarf-dwp-4gb-limit/63902
With big binaries we can have .dwp files where .debug_info.dwo section can grow
beyond 4GB. We would like to support this in LLVM and in LLDB.
The plan is to enable manual parsing of cu/tu index in DWARF library
(https://reviews.llvm.org/D137882), and then
switch internal index data structure to 64 bit.
For the second part is to enable 64bit offset support in LLDB with
this patch.
Depends on D139955
Reviewed By: labath
Differential Revision: https://reviews.llvm.org/D138618
Tue Ly [Mon, 13 Feb 2023 20:52:51 +0000 (15:52 -0500)]
[libc][bazel] Update math function unit tests' dependency computation.
Peter Klausler [Tue, 7 Feb 2023 00:40:21 +0000 (16:40 -0800)]
[flang] Warn on mismatched DATA substring sizes rather than crashing
When a DATA statement initializes a substring with a character constant
of the wrong length, do the right thing with blank padding or truncation,
and emit a warning. Current code is crashing due to an unhandled error
reported from the low-level data image initialization framework.
Differential Revision: https://reviews.llvm.org/D143819
Sam Clegg [Fri, 10 Feb 2023 22:33:52 +0000 (14:33 -0800)]
[lld][WebAssembly] Limit size of shared 64-bit memories of 2^^34
This is current limit in v8. See
https://github.com/WebAssembly/memory64/issues/33 how we might change
this in the future.
Differential Revision: https://reviews.llvm.org/D143783
Joseph Huber [Mon, 6 Feb 2023 17:01:20 +0000 (11:01 -0600)]
[libc] Add a loader utility for AMDHSA architectures for testing
This is the first attempt to get some testing support for GPUs in LLVM's
libc. We want to be able to compile for and call generic code while on
the device. This is difficult as most GPU applications also require the
support of large runtimes that may contain their own bugs (e.g. CUDA /
HIP / OpenMP / OpenCL / SYCL). The proposed solution is to provide a
"loader" utility that allows us to execute a "main" function on the GPU.
This patch implements a simple loader utility targeting the AMDHSA
runtime called `amdhsa_loader` that takes a GPU program as its first
argument. It will then attempt to load a predetermined `_start` kernel
inside that image and launch execution. The `_start` symbol is provided
by a `start` utility function that will be linked alongside the
application. Thus, this should allow us to run arbitrary code on the
user's GPU with the following steps for testing.
```
clang++ Start.cpp --target=amdgcn-amd-amdhsa -mcpu=<arch> -ffreestanding -nogpulib -nostdinc -nostdlib -c
clang++ Main.cpp --target=amdgcn-amd-amdhsa -mcpu=<arch> -nogpulib -nostdinc -nostdlib -c
clang++ Start.o Main.o --target=amdgcn-amd-amdhsa -o image
amdhsa_loader image <args, ...>
```
We determine the `-mcpu` value using the `amdgpu-arch` utility provided
either by `clang` or `rocm`. If `amdgpu-arch` isn't found or returns an
error we shouldn't run the tests as the machine does not have a valid
HSA compatible GPU. Alternatively we could make this utility in-source
to avoid the external dependency.
This patch provides a single test for this untility that simply checks
to see if we can compile an application containing a simple `main`
function and execute it.
The proposed solution in the future is to create an alternate
implementation of the LibcTest.cpp source that can be compiled and
launched using this utility. This approach should allow us to use the
same test sources as the other applications.
This is primarily a prototype, suggestions for how to better integrate
this with the existing LibC infastructure would be greatly appreciated.
The loader code should also be cleaned up somewhat. An implementation
for NVPTX will need to be written as well.
Reviewed By: sivachandra, JonChesterfield
Differential Revision: https://reviews.llvm.org/D139839
John Brawn [Mon, 13 Feb 2023 19:33:54 +0000 (19:33 +0000)]
[extract_symbols.py] Be more permissive when examining nm output
The regex which checks for undefined symbols in the nm output was
written assuming GNU nm, but other versions of nm (e.g. llvm-nm) have
slightly different output. Make the regex more permissive to account
for this.
Alexander Shaposhnikov [Mon, 13 Feb 2023 18:29:16 +0000 (18:29 +0000)]
[Clang][LLVM] Enable __arithmetic_fence and fprotect-parens on AArch64
Enable __arithmetic_fence and fprotect-parens on AArch64.
Test plan: ninja check-clang check-clang-tools check-llvm
Differential revision: https://reviews.llvm.org/D143781
Krzysztof Parzyszek [Mon, 13 Feb 2023 19:22:11 +0000 (11:22 -0800)]
Revert "[Hexagon] Add release note for LLVM 16"
This reverts commit
140bc411b245260680d90fd981635317b261ef0e.
The release notes update has been backported to releases/16.x, and
can now be reverted in main.
Jakub Kuderski [Mon, 13 Feb 2023 19:18:12 +0000 (14:18 -0500)]
[mlir][arith] Allow to specify `constFoldBinaryOp` result type
This enables us to use the common fold helpers on elementwise ops that
produce different result type than operand types, e.g., `arith.cmpi` or
`arith.addui_extended`.
Use the updated helper to teach `arith.cmpi` to fold constant vectors.
Reviewed By: Mogball
Differential Revision: https://reviews.llvm.org/D143779
Jakub Kuderski [Mon, 13 Feb 2023 18:38:11 +0000 (13:38 -0500)]
[mlir][arith] Fold selection over constant vector conditions
Also add missing tests for the scalar and splat cases.
Reviewed By: antiagainst, Mogball
Differential Revision: https://reviews.llvm.org/D143801
Alex Langford [Fri, 10 Feb 2023 23:51:49 +0000 (15:51 -0800)]
[lldb] Fix image lookup crash
lldb may crash when performing `image lookup --verbose --address $ADDR`.
The ExecutionContext that gets passed into DWARFExpression::Evaluate may
be valid but unpopulated. However, in one specific case, we were
assuming that it has a valid Target and using it without checking first.
We reach this codepath when we attempt to get information about an
address that doesn't map to a CompileUnit in the module containing the
requested address. lldb then checks to see if it maps to a global
variable, so lldb has to evaluate the location of each global variable
in the module. If a location expression contains DW_OP_deref_size that
uses a FileAddress, we hit this code path. The simplest test case is to
take a module that has a global variable with DW_OP_deref_size in its
location expression, attempt to read an address that doesn't map to a
CompileUnit (e.g. 0x0) and ensure we don't crash.
Differential Revision: https://reviews.llvm.org/D143792
Vlad Serebrennikov [Mon, 13 Feb 2023 07:43:09 +0000 (10:43 +0300)]
[clang] Add test for CWG2396
Also mark [[ https://wg21.link/cwg1291 | CWG1291 ]] as "na".
[[ https://wg21.link/p1787 | P1787 ]]: CWG1291 and CWG2396 are resolved by explicitly specifying how to look up names in a conversion-type-id.
Wording: see changes to [basic.lookup.unqual]/5 and [basic.lookup.qual]/2.
"Calling a conversion function" example in [[ https://wg21.link/p1787 | P1787 ]] is also relevant.
Differential Revision: https://reviews.llvm.org/D142316
Craig Topper [Mon, 13 Feb 2023 18:39:55 +0000 (10:39 -0800)]
[RISCV] Make vsetvli intrinsics default to MA.
The vsetvli insertion pass can replace it with MU if needed by
a using instruction. The vsetvli insertion pass will not convert
MU to MA so we need to start at MA.
Reviewed By: eopXD
Differential Revision: https://reviews.llvm.org/D143790
Vadim Paretsky (Intel Americas Inc) [Mon, 13 Feb 2023 18:29:10 +0000 (10:29 -0800)]
[OpenMP] generate the Windows import library that imports by name rather
than ordinal
This check-in changes the OpenMP build script to generate the Windows
import library that imports by name rather than ordinal to reduce
ordinals order dependency and promote runtime flavors compatibility
going forward. The existing ordinals ordering is preserved to maintain
backward compatibility.
Differential Revision: https://reviews.llvm.org/D143431
Jake Hall [Mon, 13 Feb 2023 14:10:20 +0000 (14:10 +0000)]
[mlir] Add Float8E5M2FNUZ and Float8E4M3FNUZ types to MLIR
Float8E5M2FNUZ and Float8E4M3FNUZ have been added to APFloat in D141863.
This change adds these types as MLIR builtin types alongside Float8E5M2
and Float8E4M3FN (added in D133823 and D138075).
Reviewed By: krzysz00
Differential Revision: https://reviews.llvm.org/D143744
Peter Klausler [Mon, 6 Feb 2023 01:25:28 +0000 (17:25 -0800)]
[flang] Emit portability warning for extension
f18 accepts statement labels in fixed form source even if they follow
a semicolon -- i.e., they're not in the fixed form's label field.
Emit a warning for such usage.
Differential Revision: https://reviews.llvm.org/D143817
Han Zhu [Mon, 30 Jan 2023 23:45:47 +0000 (15:45 -0800)]
[mlir] Avoid building some shared libraries when PIC is off
When `LLVM_ENABLE_PIC = OFF`, shared libraries cannot be built against code
that's compiled without -fPIC. Example error message:
``
ld.lld: error: relocation R_X86_64_32 cannot be used against local symbol;
recompile with -fPIC
>>> defined in lib/libLLVMSupport.a(StringMap.cpp.o)
>>> referenced by StringMap.cpp
>>> StringMap.cpp.o:(llvm::StringMapImpl::StringMapImpl(unsigned
>>> int, unsigned int)) in archive lib/libLLVMSupport.a
``
Similar to [how libclang handles
this](https://github.com/llvm/llvm-project/blob/main/clang/tools/clang-shlib/CMakeLists.txt#L2-L4),
skip building these shared libraries when `LLVM_ENABLE_PIC = OFF`.
Differential Revision: https://reviews.llvm.org/D142941
Michael Buch [Mon, 13 Feb 2023 18:04:44 +0000 (18:04 +0000)]
Revert "Reland "[lldb][DWARFASTParserClang] Attach linkage name to ctors/dtors if missing""
This reverts commit
19128792e2aa320c1a149f7f93638cbd7f3c83c6.
As pointed out in https://reviews.llvm.org/D143652 this implementation
doesn't quite work for subobject constructors/destructors because DWARF
can map multiple definitions of a ctor/dtor to the same specification DIE.
With the current implementation we would pick the first definition we
find and use that linkage name which means we can sometimes pick the
wrong dtor/ctor and fail to execute a valid expression.
Differential Revision: https://reviews.llvm.org/D143652
Michael Buch [Mon, 13 Feb 2023 18:03:44 +0000 (18:03 +0000)]
Revert "[lldb][Test][NFC] TestExternalCtorDtorLookup: add more instantiations to test linkage name is correct"
This reverts commit
3c160d3051dc5d637bd35cfa673871338589ad8e.
This patch depends on the reverted patch in https://reviews.llvm.org/D143652
Differential Revision: https://reviews.llvm.org/D143652
Philip Reames [Mon, 13 Feb 2023 17:42:25 +0000 (09:42 -0800)]
[docs] Speculative link syntax fix
Kazu Hirata [Mon, 13 Feb 2023 17:40:27 +0000 (09:40 -0800)]
[ADT] Add llvm::rotl and llvm::rotr to bit.h
Differential Revision: https://reviews.llvm.org/D143882
Mark de Wever [Mon, 13 Feb 2023 17:32:04 +0000 (18:32 +0100)]
[libc++][doc] Updates format status page.
This adds the papers voted in during the Issaquah plenary.
Corentin Jabot [Mon, 13 Feb 2023 17:24:20 +0000 (09:24 -0800)]
[clang] fix DR status in cxx_status.html
Alex Brachet [Mon, 13 Feb 2023 17:24:40 +0000 (17:24 +0000)]
[Driver] Add --vfsoverlay flag
This flag implies `-ivfsoverlay`, and additionally passes the same
argument to the linker if it supports it. At present the only linker
which does is lld-link, so this functionality has only been added to
the MSVC toolchain. Additionally this option has been made a
CoreOption so that clang-cl can use it without `-Xclang`
Differential Revision: https://reviews.llvm.org/D141808
Peter Klausler [Mon, 6 Feb 2023 23:25:53 +0000 (15:25 -0800)]
[flang] Allow for user-defined intrinsic operators (regression)
The recent code that uses an iterative rather than recursive walk
over the parse tree when processing expressions needs to allow for
the rare case in which an intrinsic operator (e.g., +) is extended
by a generic interface in the program.
Differential Revision: https://reviews.llvm.org/D143818
Hongtao Yu [Fri, 10 Feb 2023 01:48:53 +0000 (17:48 -0800)]
[CSSPGO][Preinliner] Set default value of sample-profile-inline-limit-max to 50000.
The previous threshold 3000 is too small to enable any inlining for giant functions which come in with bigger size than that. In real world, I've seen a big hot function with 34000 dissasembly size. Motivated by that I'm changing the value to 50000.
With the new value the allowance size growth should still be reasonable, as it is also bounded by another threshold, i.e, --sample-profile-inline-growth-limit , which defaults to 12. The new value should mostly only affect giant functions.
I've seen for serveral internal services, the new threshold boosts performance, and it has neutral impact for other services without hot giant functions. So far I haven't seen any performance regression with that.
Reviewed By: wenlei
Differential Revision: https://reviews.llvm.org/D143696
Roy Jacobson [Mon, 13 Feb 2023 17:14:22 +0000 (19:14 +0200)]
[Clang] Export CanPassInRegisters as a type trait
While working on D140664, I thought it would be nice to be able to write tests
for parameter passing ABI. Currently we test this by dumping the AST and
matching the results which makes it hard to write new tests.
Adding this builtin will allow writing better ABI tests which
can help improve our coverage in this area.
While less useful, maybe some users would also find it useful for asserting
against pessimisations for their classes.
Reviewed By: erichkeane
Differential Revision: https://reviews.llvm.org/D141775
Philip Reames [Mon, 13 Feb 2023 17:07:28 +0000 (09:07 -0800)]
[RISCV][docs] Describe status of zicsr and zifencei
This has been a repeated point of recent confusion. There are a couple attempts in flight to change the way we handle the extensions, but for the moment, let's just document what's going on.
Differential Revision: https://reviews.llvm.org/D14392
Younan Zhang [Mon, 13 Feb 2023 11:17:26 +0000 (19:17 +0800)]
[Tooling/Inclusion] Add symbol mappings for `std::experimental::filesystem`
Clangd maintains a symbol map from standard library, in order to prevent
unexpected header/symbol leaks from internal files. (e.g. files under
`bits/` for libstdc++) This symbol map was generated by a python script
that parses pages of offline cppreference archive. The script didn't
handle the case for `std::experimental::`, where most symbols are from
TS. It works well as symbols are directly laid out in the corresponding
header under `experimental` directory for most of time.
However, libstdc++'s implementation split symbols of TS FS into a few
header files located in `experimental/bits`. This would make the code
completion provide internal headers when we simply select the symbols.
There are slightly differences between TS FS and C++17 FS. Some
functions like `system_complete` was replaced by `absolute` and
relative-related operations were introduced later by another proposal.
Even so, all mainstream implementation are based on N4100, the final
filesystem TS draft that was published in 2014 and from which symbols
we've added are exported.
This fixes https://github.com/clangd/clangd/issues/1481
Reviewed By: kadircet
Differential Revision: https://reviews.llvm.org/D142836
Michael Buch [Fri, 10 Feb 2023 15:38:10 +0000 (15:38 +0000)]
Reland "[lldb][DWARFASTParserClang] Correctly resolve imported namespaces during expression evaluation"
This relands a patch previously reverted
in `
181d6e24ca3c09bfd6ec7c3b20affde3e5ea9b40`.
This wasn't quite working on Linux because we
weren't populating the manual DWARF index with
`DW_TAG_imported_declaration`. The relanded patch
does this.
**Summary**
This patch makes the expression evaluator understand
namespace aliases.
This will become important once `std::ranges` become
more widespread since `std::views` is defined as:
```
namespace std {
namespace ranges::views {}
namespace views = ranges::views;
}
```
**Testing**
* Added API test
Differential Revision: https://reviews.llvm.org/D143398
Philipp Tomsich [Mon, 13 Feb 2023 16:56:20 +0000 (17:56 +0100)]
[RISCV] Fix -Wbraced-scalar-init warning. NFC
Fix the following warning:
/lib/Target/RISCV/RISCVISelLowering.cpp:315:24: error: braces around scalar initializer [-Werror,-Wbraced-scalar-init]
setOperationAction({ISD::CTLZ}, XLenVT, Legal);
^~~~~~~~~~~
Lang Hames [Mon, 13 Feb 2023 16:28:07 +0000 (08:28 -0800)]
[ORC] StaticLibraryDefinitionGenerator -- support in-memory universal binaries.
Add new StaticLibraryDefinitionGenerator::Create methods to support construction
from in-memory universal binaries.
Louis Dionne [Fri, 3 Feb 2023 22:27:39 +0000 (17:27 -0500)]
[libc++] Reorganize exception_guard
We try to keep as little code as possible within #ifdef.
Differential Revision: https://reviews.llvm.org/D143299
Christudasan Devadasan [Fri, 10 Feb 2023 07:54:07 +0000 (13:24 +0530)]
[AMDGPU] Allow architected SGPRs for workgroup IDs
Some subtargets use architected SGPRs for workgroup
IDs instead of the regular SGPRs. This patch enables
the support for the same and is guarded under the
subtarget feature FeatureArchitectedSGPRs.
Reviewed By: foad
Differential Revision: https://reviews.llvm.org/D143707
Christudasan Devadasan [Fri, 10 Feb 2023 06:22:00 +0000 (11:52 +0530)]
[AMDGPU] Add switch to enable architected SGPRs.
Reviewed By: rampitec, foad
Differential Revision: https://reviews.llvm.org/D143706
John Brawn [Mon, 13 Feb 2023 15:59:20 +0000 (15:59 +0000)]
[extract_symbols.py] Adjust usage of nm again
The previous change to extract_symbols.py means that building on macOS
with LLVM_EXPORT_SYMBOLS_FOR_PLUGINS fails as we try to export some
local symbols, as the regex used to match external symbols wasn't good
enough. Solve this by using the -g option to nm, so we only get
external symbols and don't have to check for local symbols at all.
Mirko Müller [Mon, 13 Feb 2023 16:14:23 +0000 (16:14 +0000)]
[AArch64] Add NZCV Def for TLSDESC_CALLSEQ
The glibc and older musl handlers of tlsdesc_dynamic use a cmp instruction
which will clobber NZCV.
See glibc's _dl_tlsdesc_dynamic:
https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/aarch64/dl-tlsdesc.S;hb=refs/heads/release/2.37/master
See v1.1.21 Musl's __tlsdesc_dynamic:
https://git.musl-libc.org/cgit/musl/tree/src/ldso/aarch64/tlsdesc.s?h=v1.1.21
Differential Revision: https://reviews.llvm.org/D143157
Dinar Temirbulatov [Mon, 13 Feb 2023 16:07:06 +0000 (16:07 +0000)]
[DAGCombine] Allow DAGCombine to remove dead masked stores.
Remove a dead masked store if another one has the same base pointer and mask or
the following store has all true constant mask and size if equal or bigger to
the first store.
Differential Revision: https://reviews.llvm.org/D143069
Jakub Kuderski [Mon, 13 Feb 2023 16:01:51 +0000 (11:01 -0500)]
[mlir][spirv] Drop outdated TODO comment
We do not plan to handle wide vector types with SPIR-V arrays anymore.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D143777
Philipp Tomsich [Mon, 13 Feb 2023 15:29:31 +0000 (16:29 +0100)]
[RISCV] Add vendor-defined XTheadBb (basic bit-manipulation) extension
The vendor-defined XTHeadBb (predating the standard Zbb extension)
extension adds some bit-manipulation extensions with somewhat similar
semantics as some of the Zbb instructions.
It is supported by the C9xx cores (e.g., found in the wild in the
Allwinner D1) by Alibaba T-Head.
The current (as of this commit) public documentation for XTHeadBb is
available from:
https://github.com/T-head-Semi/thead-extension-spec/releases/download/2.2.2/xthead-2023-01-30-2.2.2.pdf
Support for these instructions has already landed in GNU Binutils:
https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=
8254c3d2c94ae5458095ea6c25446ba89134b9da
Depends on D143036
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D143439
Tobias Hieta [Mon, 13 Feb 2023 15:58:13 +0000 (16:58 +0100)]
[NFC] Fix using-declspec.cpp test with non-C++17 compilers
Matthias Springer [Mon, 13 Feb 2023 15:35:47 +0000 (16:35 +0100)]
[mlir][bufferization] Better propagation of bufferizesToMemoryWrite through regions
`bufferizesToMemoryWrite(OpResult)` looks for OpOperands that bufferize to memory writes inside the region of the defining op (if it has one). Currently, if the reverse use-def chain stops at any value inside of the region, the OpResult is considered to bufferize to a memory write.
It is always safe to have false positives among `bufferizesToMemoryWrite`, so the previous implementation is also correct. However, it can lead to additional buffer copies.
Differential Revision: https://reviews.llvm.org/D142223
Peter Klausler [Wed, 8 Feb 2023 18:25:44 +0000 (10:25 -0800)]
[flang] Fix new assertion failure
Don't compute the address of element [0] of a vector when the vector
is empty, even if the address is not used.
Differential Revision: https://reviews.llvm.org/D143824
Hans Wennborg [Mon, 13 Feb 2023 15:17:02 +0000 (16:17 +0100)]
Revert "[libc++] Remove _LIBCPP_CONSTEVAL"
It causes mysterious memory leaks when comparing std::string, see GitHub
Issue #60709 and the code review.
> All supported compilers support `consteval`, so there is no more need for the macro.
>
> Reviewed By: ldionne, Mordante, #libc
>
> Spies: libcxx-commits
>
> Differential Revision: https://reviews.llvm.org/D143489
This reverts commit
aaef3b82f4f0dd3924a3491b381d5015ff472b86.
Philipp Tomsich [Mon, 13 Feb 2023 15:10:11 +0000 (16:10 +0100)]
[RISCV] Add vendor-defined XTHeadBs (single-bit) extension
The vendor-defined XTHeadBs (predating the standard Zbs extension)
extension adds a bit-test instruction (th.tst) with similar semantics
as bexti from Zbs. It is supported by the C9xx cores (e.g., found in
the wild in the Allwinner D1) by Alibaba T-Head.
The current (as of this commit) public documentation for XTHeadBs is
available from:
https://github.com/T-head-Semi/thead-extension-spec/releases/download/2.2.2/xthead-2023-01-30-2.2.2.pdf
Support for these instructions has already landed in GNU Binutils:
https://sourceware.org/git/?p=binutils-gdb.git;a=commit;h=
8254c3d2c94ae5458095ea6c25446ba89134b9da
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D143036
Guillaume Chatelet [Mon, 13 Feb 2023 15:22:31 +0000 (15:22 +0000)]
[libc][NFC] Fix typo in header guards
WuXinlong [Mon, 13 Feb 2023 11:27:20 +0000 (19:27 +0800)]
[RISCV] edit document of Zcb to match the Implementation
This patch reorder the list of Experimental Extensions in doc,
and change the spec verision in doc to match the actual Implementation
if this patch is so small that it is not worth to submit separately, Apologies for bothering
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D143908
Stefan Pintilie [Mon, 13 Feb 2023 14:59:23 +0000 (09:59 -0500)]
[PowerPC] Fix float materialization patterns.
Two of the float materialization patterns use the VSSRC regsiter class. This
register class is not available before Power 8. The patterns will stay the same
for Power 8 and up but must use the class F4RC for Power 7 and earlier.
This patch fixes those patterns.
Reviewed By: nemanjai, amyk, #powerpc
Differential Revision: https://reviews.llvm.org/D142120
Phoebe Wang [Mon, 13 Feb 2023 14:33:20 +0000 (22:33 +0800)]
[X86][FP16] Combine two steps conversions into direct conversion
When both v8i64 and v4f16 are not legal in a v8i64->v8f16 conversion, legalizer will breaks it into v8i64->v4i64->v4f32->v8f32->v8f16.
Given we support v4i64->v8f16, we can combine them with a shuffle instruction.
Reviewed By: LuoYuanke
Differential Revision: https://reviews.llvm.org/D143872
Anton Sidorenko [Mon, 13 Feb 2023 14:10:39 +0000 (17:10 +0300)]
Recommit [YAML IO] Check that mapping doesn't contain duplicating keys
The revert reason is fixed in D143727 (test changes).
According to YAML specification keys must be unique for a mapping node:
"The content of a mapping node is an unordered set of key/value node pairs, with
the restriction that each of the keys is unique".
Differential Revision: https://reviews.llvm.org/D140474
Valentin Clement [Mon, 13 Feb 2023 14:42:25 +0000 (15:42 +0100)]
[flang] Handle correctly optional intrinsic scalar to unlimited polymorphic optional
When an optional intrinsic scalar is passed to a function expecting an
unlimited polymorphic dummy argument, the presence test must be done
before the emboxing otherwise it will result in a program crash.
Depends on D143888
Reviewed By: jeanPerier, PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D143889
Tobias Hieta [Thu, 9 Feb 2023 09:53:39 +0000 (10:53 +0100)]
[clang] Handle __declspec() attributes in using
This patch fixes so that declspec attributes are forwarded
to the alias declaration.
Before this patch this would assert:
class Test { int a; };
using AlignedTest = __declspec(align(16)) const Test;
static_assert(alignof(AlignedTest) == 16, "error");
But afterwards it behaves the same as MSVC does and doesn't
assert.
Fixes: llvm/llvm-project#60513
Reviewed By: aaron.ballman
Differential Revision: https://reviews.llvm.org/D143632
Valentin Clement [Mon, 13 Feb 2023 14:41:19 +0000 (15:41 +0100)]
[flang] Fix output type when emboxing intrinsic type to unlimited polymorphic box
When emboxing an intrinsic type to a polymorphic descriptor, directly set its
type to `fir.class<none>`.
`fir.class<i32>` is not a real type used anywhere in lowering so make it right directly
avoid unnecessary convert op to `fir.class<none>`. Also `fir.class<i32>` would not be
recognized as unlimited polymorphic.
Reviewed By: jeanPerier, PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D143888
Xiang Li [Sun, 12 Feb 2023 17:19:17 +0000 (12:19 -0500)]
[mlir] support complex type in DenseElementsAttr::get.
Fixes #60662 https://github.com/llvm/llvm-project/issues/60662
Allow ComplexType when create DenseElementsAttr.
Also allow build ConstantOp for integer complex.
Differential Revision: https://reviews.llvm.org/D143848
David Green [Mon, 13 Feb 2023 14:35:10 +0000 (14:35 +0000)]
[AArch64] Reassociate sub(x, add(m1, m2)) to sub(sub(x, m1), m2)
The mid end will reassociate sub(sub(x, m1), m2) to sub(x, add(m1, m2)). This
reassociates it back to allow the creation of more mls instructions.
Differential Revision: https://reviews.llvm.org/D143143
Florian Hahn [Mon, 13 Feb 2023 13:51:14 +0000 (13:51 +0000)]
[VPlan] Fix iterator invalidation in adjustFixedOrderRecurrences.
adjustFixedOrderRecurrences may insert instructions after immediately
after the PHI nodes in the block. This invalidates the phis() iterator.
To avoid crashing/accessing invalid recipes, first collect all
first-order recurrence phi recipes.
This should fix a crash reported by @dmgreen after D142589 landed.
Louis Dionne [Fri, 10 Feb 2023 23:10:35 +0000 (15:10 -0800)]
[libc++] Fix bug in allocate_shared_for_overwrite
Instead of destroying the object with allocator::destroy, we must
call its destructor directly. As a fly-by also mark LWG3008 as
fixed since it is handled by our implementation.
This was pointed out by Tim Song in https://reviews.llvm.org/D140913.
Differential Revision: https://reviews.llvm.org/D143791
Leandro Lupori [Thu, 2 Feb 2023 20:28:32 +0000 (20:28 +0000)]
[flang] Avoid infinite recursion in common block check
Don't call CheckCommonBlockDerivedType() recursively if the
derived type symbol is the same symbol that is already being
processed. This can happen when a component is a pointer of the
same type as its parent component, for instance.
Fixes #60230
Reviewed By: klausler
Differential Revision: https://reviews.llvm.org/D143211
Louis Dionne [Tue, 7 Feb 2023 17:20:35 +0000 (09:20 -0800)]
[libc++] Implement P2136R3 std::invoke_r
Differential Revision: https://reviews.llvm.org/D143610
Anton Sidorenko [Fri, 10 Feb 2023 12:37:13 +0000 (15:37 +0300)]
[Test][lldb] Fix YAML mapping keys duplication. NFC.
YAML specification does not allow keys duplication an a mapping. However, YAML
parser in LLVM does not have any check on that and uses only the last key entry.
In this change duplicated keys are merged to satisfy the spec.
Differential Revision: https://reviews.llvm.org/D143727
Simon Pilgrim [Mon, 13 Feb 2023 13:28:42 +0000 (13:28 +0000)]
Revert rG0b0a38a7a229b70d7261771ba0e702843bd34e97 : "[X86] combineX86ShufflesRecursively - don't widen shuffle subvector inputs"
Reports of miscompiles, that I'm still trying to triage - reverting for now
Quentin Colombet [Fri, 10 Feb 2023 16:21:38 +0000 (17:21 +0100)]
[mlir][MemRef][Transform] Don't apply multibuffer on "useless" allocs
`alloc`s that have users outside of loops are guaranteed to fail in
`multibuffer`.
Instead of exposing ourselves to that failure in the transform dialect,
filter out the `alloc`s that fall in this category.
To be able to do this filtering we have to change the `multibuffer`
transform op from `TransformEachOpTrait` to a plain `TransformOp`. This is
because `TransformEachOpTrait` expects that every successful `applyToOne`
returns a non-empty result.
Couple of notes:
- I changed the assembly syntax to make sure we only get `alloc` ops as
input. (And added a test case to make sure we reject invalid inputs.)
- `multibuffer` can still fail pretty easily when you know its limitations.
See the updated `op failed to multibuffer` test case for instance.
Longer term, instead of leaking/coupling the actual implementation (in
this case the checks normally done in `memref::multiBuffer`) with the
transform dialect (the added check in `::apply`), we may want to refactor
how we structure the underlying implementation. E.g., we could imagine a
`canApply` method for all the implementations that we want to hook up in
the transform dialect.
This has some implications on how not to duplicate work between
`canApply` and the actual implementation but I thought I throw that here
to have us think about it :).
Differential Revision: https://reviews.llvm.org/D143747
Nico Weber [Mon, 13 Feb 2023 13:03:00 +0000 (08:03 -0500)]
[gn] port
c4f7cc867299 (CoroTests)
Joseph Huber [Thu, 2 Feb 2023 16:58:16 +0000 (10:58 -0600)]
[Libomptarget] Remove dependency on the DeviceRTL from the GPU plugins
The GPU plugins have a dependency on the device libraries. Sometimes we
cannot build the device libraries because the user does not have a valid
`clang` to use or it was explicitly disabled. Currently this leads to a
transitive failure because we cannot meet this dependency. This patch
simply removes that dependency.
Fixes https://github.com/llvm/llvm-project/issues/60457
Reviewed By: tianshilei1992
Differential Revision: https://reviews.llvm.org/D143196
Michael Buch [Mon, 13 Feb 2023 12:50:31 +0000 (12:50 +0000)]
[lldb][Test][NFC] TestExternalCtorDtorLookup: add more instantiations to test linkage name is correct
Tomas Matheson [Tue, 7 Feb 2023 16:20:46 +0000 (16:20 +0000)]
[AArch64] Fix LSE2/LSE128/RCPC3 precedence
D142712 added tests for when both lse2 and lse128 are available, but
in practice there is no way to enable LSE128 without LSE2 from clang:
LSE128 is a v9 only feature and LSE2 has been mandatory since v8.4,
and +/-lse2 can not be specified on the clang command line.
Therefore it makes more sense that lse2+lse128 should emit lse128
instructions, otherwise they will not be emitted at all.
It also makes sense to remove the lse128-only backend tests if that set
of attributes is never set by the frontend.
Differential Revision: https://reviews.llvm.org/D143506
Nikolas Klauser [Mon, 26 Dec 2022 15:24:01 +0000 (16:24 +0100)]
[libc++] Granularize <type_traits> includes in <algorithm>
Reviewed By: Mordante, #libc
Spies: libcxx-commits, wenlei
Differential Revision: https://reviews.llvm.org/D140673
Janek van Oirschot [Fri, 10 Feb 2023 17:03:30 +0000 (17:03 +0000)]
Reapply "[AMDGPU] Modify adjustInliningThreshold to also consider the cost of passing function arguments through the stack"
Reapplies
142c28ffa1323e9a8d53200a22c80d5d778e0d0f as part of D140242 which got reverted due to amdgpu openmp test failures.
This diff fixes said failures by eliding most of `adjustInliningThresholdUsingCallee` for indirect calls as the callee function is unavailable for indirect calls.
Reviewed By: arsenm, #amdgpu
Differential Revision: https://reviews.llvm.org/D143498
Quentin Colombet [Fri, 10 Feb 2023 12:43:33 +0000 (13:43 +0100)]
[mlir][MemRef][TransformOps] Fix error reporting for multibuffer
Multibuffer will fail to apply on allocs that are used outside of loops.
This was properly caught in the current implementation but the way we report
it was broken.
Notes cannot be emitted on their own, they need to be attached to another
main diagnostic.
Long story short, change the severity of the report from Note to Error.
Differential Revision: https://reviews.llvm.org/D143729
Alexey Lapshin [Thu, 2 Feb 2023 16:17:52 +0000 (17:17 +0100)]
[DWARFLinker] Refactor cloneAddressAttribute().
As a preparation for implementing DWARFv5 address ranges generation,
this patch refactors cloneAddressAttribute() method. It has special
handling for addresses which can be relocated in some unrelated value,
for applying relocations twice, for indexed addresses. Instead of
all these special handlings this patch uses general handling:
Read attribute value from InputDIE and apply PCOffset.
Another thing is that current handling of DW_FORM_addrx misses the
fact that relocations might be applied twice in some cases. This
patch fixes this problem also.
Differential Revision: https://reviews.llvm.org/D143269
Guillaume Chatelet [Mon, 13 Feb 2023 11:14:13 +0000 (11:14 +0000)]
[libc] Fix LIBC_LOOP_NOUNROLL being a noop
David Stuttard [Thu, 19 Jan 2023 13:55:49 +0000 (13:55 +0000)]
[Coroutines] Modify CoroFrame materializable into a callback
This change makes it possible to optionally provide a different callback to
determine if an instruction is materializable.
By default the behaviour is unchanged.
Differential Revision: https://reviews.llvm.org/D142621
David Stuttard [Mon, 23 Jan 2023 12:52:44 +0000 (12:52 +0000)]
[Coroutines] Improve rematerialization stage
As originally implemented, the rematerialization of valid instructions across
the suspend point would iterate 4 times, meaning that up to 4 instructions could
be rematerialized.
This implementation changes that approach to instead build a graph of
rematerializable instructions, then move all of them. This is faster than the
original approach and is not limited to an arbitrary limit.
Differential Revision: https://reviews.llvm.org/D142620
David Stuttard [Thu, 12 Jan 2023 11:08:12 +0000 (11:08 +0000)]
[Coroutines] Presubmit test for more coro remats
Added more tests that check for >4 instructions.
Also added a retcon-remat test that checks rematerialization into a suspend
block predecessor (such as when remat for a retcon suspend happens).
Differential Revision: https://reviews.llvm.org/D142619
Samuel Parker [Mon, 13 Feb 2023 11:00:54 +0000 (11:00 +0000)]
[ARM] O3-pipeline fix
Tom Eccles [Wed, 7 Dec 2022 17:52:31 +0000 (17:52 +0000)]
[flang] RFC: -fstack-arrays
An RFC describing how -fstack-arrays will be implemented.
Differential Revision: https://reviews.llvm.org/D139617
Tom Eccles [Fri, 27 Jan 2023 18:01:44 +0000 (18:01 +0000)]
[flang][hlfir] lower hlfir.sum into fir runtime call
This duplicates some argument lowering and return value processing from
flang/lib/Lower/ConvertCall.cpp. The existing code in CovertCall lowers
directly into the fir.call (without the hlfir transformational intrinsic
operation), and it is too tied to the lowering code to move into
flang/lib/Optimizer to allow for more complete re-use here.
Differential Revision: https://reviews.llvm.org/D143512
Tom Eccles [Mon, 23 Jan 2023 09:52:31 +0000 (09:52 +0000)]
[flang] lower sum intrinsic to hlfir.sum operation
Differential Revision: https://reviews.llvm.org/D142898
Tom Eccles [Tue, 17 Jan 2023 17:37:15 +0000 (17:37 +0000)]
[flang] add hlfir.sum operation
Add an HLFIR operation for the SUM transformational intrinsic, according
to the design set out in flang/doc/HighLevelFIR.md.
I decided to make hlfir.sum very lenient about the form of its
arguments. This allows the sum intrinsic to be lowered to only this HLFIR
operation, without needing several operations to convert and box
arguments. Having only one operation generated for the intrinsic
invocation should make optimisation passes on HLFIR simpler.
Differential Revision: https://reviews.llvm.org/D142897
Markus Böck [Sun, 12 Feb 2023 20:34:21 +0000 (21:34 +0100)]
[mlir][SCFToOpenMP] Add pass option to emit LLVM opaque pointers
Part of https://discourse.llvm.org/t/rfc-switching-the-llvm-dialect-and-dialect-lowerings-to-opaque-pointers/68179
There were luckily only very few changes that had to be made. To allow users to also specify the pass option from C++ code I have also migrated the pass to use autogenerated constructors to autogenerate a pass option struct.
Differential Revision: https://reviews.llvm.org/D143855