David Sherwood [Wed, 19 Oct 2022 14:27:12 +0000 (14:27 +0000)]
[AArch64][SVE2] Add the SVE2.1 contiguous load to multiple consecutive vector
This patch adds the assembly/disassembly for the following instructions:
ld1* : Contiguous load of bytes to multiple consecutive vectors -
(scalar + scalar) and (scalar + immediate)
ldnt1* : Contiguous load non-temporal of bytes to multiple consecutive
vectors - (scalar + scalar) and (scalar + immediate)
The reference can be found here:
https://developer.arm.com/documentation/ddi0602/2022-09
Differential Revision: https://reviews.llvm.org/D136680
Stefan Gränitz [Tue, 1 Nov 2022 09:13:43 +0000 (10:13 +0100)]
[JITLink] Revert template typename change after
2a3b257a935a (NFC)
As pointed out in the review https://reviews.llvm.org/D135523#C3696894NL515
Tiezhu Yang [Tue, 1 Nov 2022 04:56:55 +0000 (12:56 +0800)]
[LLDB][LoongArch] Add LoongArch ArchSpec and subtype detection
Define LoongArch architecture subtypes, add the LoongArch ArchSpec bits,
and inspect the ELF header to detect the right subtype based on ELF class.
Here is a simple test:
```
[loongson@linux ~]$ cat hello.c
int main()
{
printf("Hello, World!\n");
return 0;
}
[loongson@linux ~]$ clang hello.c -g -o hello
```
Without this patch:
```
[loongson@linux ~]$ llvm-project/llvm/build/bin/lldb hello
(lldb) target create "hello"
error: '/home/loongson/hello' doesn't contain any 'host' platform architectures: unknown
```
With this patch:
```
[loongson@linux ~]$ llvm-project/llvm/build/bin/lldb hello
(lldb) target create "hello"
Current executable set to '/home/loongson/hello' (loongarch64).
(lldb) run
Process 735167 launched: '/home/loongson/hello' (loongarch64)
Hello, World!
Process 735167 exited with status = 0 (0x00000000)
(lldb) quit
[loongson@linux ~]$ llvm-project/llvm/build/bin/llvm-lit llvm-project/lldb/test/Shell/ObjectFile/ELF/loongarch-arch.yaml
llvm-lit: /home/loongson/llvm-project/llvm/utils/lit/lit/llvm/config.py:456: note: using clang: /home/loongson/llvm-project/llvm/build/bin/clang
-- Testing: 1 tests, 1 workers --
PASS: lldb-shell :: ObjectFile/ELF/loongarch-arch.yaml (1 of 1)
Testing Time: 0.09s
Passed: 1
```
Reviewed By: SixWeining, xen0n, DavidSpickett
Differential Revision: https://reviews.llvm.org/D137057
Nikita Popov [Thu, 27 Oct 2022 10:45:08 +0000 (12:45 +0200)]
[IR] Take operand bundles into account for call argument readonly/writeonly
We currently only take operand bundle effects into account when
querying the function-level memory attributes. However, I believe
that we also need to do the same for parameter attributes. For
example, a call with deopt bundle to a function with readnone
parameter attribute cannot treat that parameter as readnone,
because the deopt bundle may read it.
Differential Revision: https://reviews.llvm.org/D136834
Matt Arsenault [Tue, 1 Nov 2022 07:40:46 +0000 (00:40 -0700)]
AMDGPU: Add testcase from issue 58639
This was fixed in
bf789b1957efd2482e1dbd164d91a6612a450fe3
Yeting Kuo [Mon, 24 Oct 2022 15:40:48 +0000 (23:40 +0800)]
[VP][RISCV] Add vp.rint and RISC-V support.
FRINT uses dynamic rounding mode instead of static rounding mode. The patch
rename VFCVT_X_F_VL to VFCVT_RM_X_F_VL for static rounding mode uses and added
new ISDNode VFCVT_X_F_VL directly selected to PseudoVFCVT_X_F_V.
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D136662
Matt Arsenault [Sat, 29 Oct 2022 18:32:04 +0000 (11:32 -0700)]
AMDGPU/GlobalISel: Add missing run line for gfx7 packed operations
We were only testing targets with legal 16-bit operations.
Matt Arsenault [Tue, 20 Sep 2022 22:17:25 +0000 (18:17 -0400)]
AtomicExpand: Use InstSimplifyFolder
Automatically cleanup operations if we know the atomic has higher
alignment.
Michał Górny [Fri, 28 Oct 2022 20:13:35 +0000 (22:13 +0200)]
[lit] Add pyproject.toml and fix build w/ modern setuptools backend
Add a `pyproject.toml` file that provides build system information
for PEP 517-compliant builders. While all the commonly used builders
provide fallback to running `setup.py` for backwards compatibility, this
ensures the best forward compatibility. It also provides a reliable way
of specifying the minimum required setuptools version. Effectively, it
will make it possible to remove `setup.py` in favor of purely
declarative configuration in the future, or even switch to a different
build system.
Update `setup.py` to explicitly add the current directory to `sys.path`
for importing `lit`. This is necessary, as the modern setuptools
backend does not guarantee that the current directory is present there.
Differential Revision: https://reviews.llvm.org/D136976
Arnab Dutta [Mon, 31 Oct 2022 11:49:51 +0000 (17:19 +0530)]
[MLIR] Fix bug in the method constructing semi affine expression from flattened form
Set proper offset to the second element of the index pair, so that
we do not have same index values for more than one local expression.
Reviewed By: springerm
Differential Revision: https://reviews.llvm.org/D137062
Michael Buch [Tue, 1 Nov 2022 05:56:15 +0000 (22:56 -0700)]
[lldb][Test] Fix TestFrameFormatNameWithArgs.test on Windows/Linux
* Windows doesn't support setting these breakpoints by basename
* On Linux std::function arguments aren't formatted as such
Alexander Shaposhnikov [Tue, 1 Nov 2022 05:55:28 +0000 (05:55 +0000)]
[IR][Instruction] Annotate more methods with LLVM_READONLY
This diff annotates more methods of the class Instruction with
LLVM_READONLY (similarly to isAssociative/isCommutative).
Test plan: ninja check-all
Differential revision: https://reviews.llvm.org/D137053
skc7 [Wed, 26 Oct 2022 04:33:13 +0000 (04:33 +0000)]
[SLP] Extend reordering data of tree entry to support PHI nodes
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D136757
Michał Górny [Tue, 1 Nov 2022 04:49:24 +0000 (05:49 +0100)]
Revert "[compiler-rt] Switch from llvm-config to find_package(LLVM)"
This reverts commit
37acf9bdd4686397a6d37000e8d3fe35f456a860. It broke
a buildbot.
Sameer Sahasrabuddhe [Tue, 1 Nov 2022 04:26:58 +0000 (09:56 +0530)]
[AAPointerInfo] refactor how offsets and Access objects are tracked
AAPointerInfo now maintains a list of all Access objects that it owns, along
with the following maps:
- OffsetBins: OffsetAndSize -> { Access }
- InstTupleMap: RemoteI x LocalI -> Access
A RemoteI is any instruction that accesses memory. RemoteI is different from
LocalI if and only if LocalI is a call; then RemoteI is some instruction in the
callgraph starting from LocalI.
Motivation: When AAPointerInfo recomputes the offset for an instruction, it sets
the value to Unknown if the new offset is not the same as the old offset. The
instruction must now be moved from its current bin to the bin corresponding to
the new offset. This happens for example, when:
- A PHINode has operands that result in different offsets.
- The same remote inst is reachable from the same local inst via different paths
in the callgraph:
```
A (local inst)
|
B
/ \
C1 C2
\ /
D (remote inst)
```
This fixes a bug where a store is incorrectly eliminated in a lit test.
Reviewed By: jdoerfert
Differential Revision: https://reviews.llvm.org/D136526
Matt Arsenault [Sat, 29 Oct 2022 01:50:19 +0000 (18:50 -0700)]
llvm-reduce: Write bitcode temporary files for bitcode inputs
Most tools accept .ll or .bc inputs interchangably, but some don't.
Default to writing temporary files that match the input. This
will also aid reducing deserialization bugs.
Matt Arsenault [Sat, 29 Oct 2022 01:24:21 +0000 (18:24 -0700)]
llvm-reduce: Drop guessing output format based on file extension
Matt Arsenault [Wed, 26 Oct 2022 20:05:46 +0000 (13:05 -0700)]
llvm-reduce: Support emitting bitcode for final result
Previously, this unconditionally emitted text IR. I ran
into a bug that manifested in broken disassembly, so the
desired output was the bitcode format. If the input format
was binary bitcode, the requested output file ends in .bc,
or an explicit -output-bitcode option was used, emit bitcode.
Michał Górny [Sat, 29 Oct 2022 17:03:37 +0000 (19:03 +0200)]
[compiler-rt] Switch from llvm-config to find_package(LLVM)
Replace the use of the deprecated `llvm-config` tool with LLVM's CMake
files for detecting LLVM in standalone builds.
Differential Revision: https://reviews.llvm.org/D137024
Matt Arsenault [Sat, 29 Oct 2022 00:17:00 +0000 (17:17 -0700)]
Fix missing C++ mode comment
Matt Arsenault [Mon, 24 Oct 2022 16:08:54 +0000 (09:08 -0700)]
llvm-reduce: Report file opening errors
This was also trying to write the bitcode to the failed file
on failure, which asserts. Also, consistently use
ToolOutputFile, instead of one path manually removing
the temp file.
Matt Arsenault [Thu, 29 Sep 2022 20:21:49 +0000 (16:21 -0400)]
AtomicExpand: Don't create unused instructions for some atomicrmw
This wasn't used by every atomicrmw expansion.
Matt Arsenault [Sat, 29 Oct 2022 19:34:03 +0000 (12:34 -0700)]
Hexagon: Register null TargetStreamer
Unlike just about every target, I don't
see an obvious null dereference this fixes.
Matt Arsenault [Sat, 29 Oct 2022 19:34:42 +0000 (12:34 -0700)]
Hexagon: Add missing C++ mode comment
Matt Arsenault [Sat, 29 Oct 2022 19:38:55 +0000 (12:38 -0700)]
NVPTX: Register null target streamer
Matt Arsenault [Fri, 28 Oct 2022 20:42:37 +0000 (13:42 -0700)]
X86: Implement null target streamer
There should no need for null checks in the AsmPrinter
Matt Arsenault [Fri, 28 Oct 2022 17:32:15 +0000 (10:32 -0700)]
AArch64: Register null MCTargetStreamer
Fixes crashes with clang -emit-codegen-only
Florian Mayer [Tue, 1 Nov 2022 00:39:05 +0000 (17:39 -0700)]
Revert "[InstCombine] fold mul with decremented "shl -1" factor"
This reverts commit
bc886e9b587b9d009f49b12eaaa9ebc1c71905a1.
Broke MSAN bootstrap buildbots with Assertion `RangeAfterCopy % ExtraScale == 0 && "Extra instruction requires immediate to be aligned"' failed.
Florian Mayer [Mon, 31 Oct 2022 23:34:17 +0000 (16:34 -0700)]
[NFC] use has_value instead of hasValue
LLVM GN Syncbot [Mon, 31 Oct 2022 23:37:46 +0000 (23:37 +0000)]
[gn build] Port
627465cf9fa7
Nikolas Klauser [Wed, 19 Oct 2022 09:06:12 +0000 (11:06 +0200)]
[libc++][PMR] Deprecate the implementation in std::experimental
This commit deprecates <experimental/memory_resource> since we now ship the non-experimental
version of it. Per the libc++ policy [1], we are deprecating the experimental feature in
upcoming LLVM 16 and will remove it entirely in LLVM 18.
[1]: https://libcxx.llvm.org/DesignDocs/ExperimentalFeatures.html#id4
Reviewed By: ldionne, #libc
Spies: EricWF, libcxx-commits
Differential Revision: https://reviews.llvm.org/D136245
Nikolas Klauser [Tue, 18 Oct 2022 19:40:49 +0000 (21:40 +0200)]
[libc++][PMR] Move the pmr::memory_resource destructor into the dylib
This avoids emitting the VTable of `pmr::memory_resource` in every TU.
Reviewed By: ldionne
Spies: EricWF, nemanjai, libcxx-commits
Differential Revision: https://reviews.llvm.org/D136196
Nikolas Klauser [Tue, 18 Oct 2022 19:33:32 +0000 (21:33 +0200)]
[libc++][PMR] Mark completed papers and issues as such
Reviewed By: ldionne, #libc
Spies: krytarowski, Quuxplusone, tschuett, libcxx-commits
Differential Revision: https://reviews.llvm.org/D136194
Nikolas Klauser [Tue, 18 Oct 2022 18:40:13 +0000 (20:40 +0200)]
[libc++][PMR] Add attributes
This allows the compiler to do more optimizations.
Reviewed By: ldionne, #libc
Spies: libcxx-commits, krytarowski
Differential Revision: https://reviews.llvm.org/D136191
Nikolas Klauser [Wed, 12 Oct 2022 23:03:58 +0000 (01:03 +0200)]
[libc++] Move preferred_name declarations into the forward declaring headers and add pmr preferred names
We currently define the preferred names in multiple places. `basic_string` and `basic_string_view` also have a lot of aliases, which makes the declarations quite long. So let's only add the preferred names in forward-declaring headers to make the implementation more readable and have all the preferred names in one place.
Reviewed By: ldionne
Spies: EricWF, krytarowski, libcxx-commits
Differential Revision: https://reviews.llvm.org/D135824
Ilia Diachkov [Sat, 29 Oct 2022 00:07:02 +0000 (03:07 +0300)]
[SPIRV] support the enqueue_kernel builtin function
The patch adds support of the enqueue_kernel builtin function.
It is implemented in the same way as in the SPIRV translator.
2 LIT tests are added to show the improvement.
Differential Revision: https://reviews.llvm.org/D137018
Anlun Xu [Mon, 31 Oct 2022 22:26:58 +0000 (15:26 -0700)]
[Bazel] Add llvm:Support to Bazel deps of SPIRVCommonConversion
Differential Revision: https://reviews.llvm.org/D137119
Usman Nadeem [Mon, 31 Oct 2022 22:08:11 +0000 (15:08 -0700)]
[JumpThreading] Put a limit on the PHI nodes when duplicating a BB.
Do not duplicate a BB if it has a lot of PHI nodes.
If a threadable chain is too long then the number of duplicated PHI nodes
can add up, leading to a substantial increase in compile time when rewriting
the SSA.
Fixes https://github.com/llvm/llvm-project/issues/58203
Differential Revision: https://reviews.llvm.org/D136716
The threshold of 76 in this patch is reasonably high and reduces the compile
time of cldwat2m_macro.f90 in SPEC2017/cam4 from 80+min to <2min.
Change-Id: I153c89a8e0d89b206a5193dc1b908c67e320717e
Craig Topper [Mon, 31 Oct 2022 19:37:51 +0000 (12:37 -0700)]
[RISCV] Fix crash a vector add has a 4x sext and zext operand.
We can narrow one of the extends and keep the other original by
using a vwaddu.wv or vwadd.wv.
We were previously forgetting to keep the original operand and
instead took the source of its extend. This resulted in a type
mismatch that later failed with an impossible physical register copy.
To fix this I've refactored some code to maintain information about
whether the source needs to be extended at all for longer so we could
use it in materialize.
Differential Revision: https://reviews.llvm.org/D137106
Evgeny Shulgin [Sat, 22 Oct 2022 20:17:56 +0000 (20:17 +0000)]
[libc++] Add test for checking progress on P0533R9
This test checks progress on P0533R9 (https://wg21.link/p0533r9).
Whenever a desired function becomes constexpr, the programmer won't forget
to switch `ASSERT_NOT_CONSTEXPR_CXX23` to `ASSERT_CONSTEXPR_CXX23` and
eventually to change the paper's implementation status. The test also works
as a reference list of unimplemented functions.
Reviewed By: philnik, #libc
Differential Revision: https://reviews.llvm.org/D136538
LLVM GN Syncbot [Mon, 31 Oct 2022 21:45:27 +0000 (21:45 +0000)]
[gn build] Port
84ea6b6f78df
LLVM GN Syncbot [Mon, 31 Oct 2022 21:45:26 +0000 (21:45 +0000)]
[gn build] Port
0d01300aacf6
Jonas Devlieghere [Mon, 31 Oct 2022 17:16:32 +0000 (10:16 -0700)]
[lldb] Emit diagnostic events in the diagnostic dump
Connect the diagnostic events with the diagnostic infrastructure.
- Emit existing diagnostic events (warnings and errors) to the
diagnostic log.
- Introduce a new diagnostic event (info) that's used exclusively for
diagnostic logging and does not get broadcast.
Differential revision: https://reviews.llvm.org/D136648
Jonas Devlieghere [Mon, 31 Oct 2022 17:11:12 +0000 (10:11 -0700)]
[lldb] Add a "diagnostics dump" command
Add a "diagnostics dump" command to, as the name implies, dump the
diagnostics to disk. The goal of this command is to let the user
generate the diagnostics in case of an issue that doesn't cause the
debugger to crash.
This command is also critical for testing, where we don't want to cause
a crash to emit the diagnostics.
Differential revision: https://reviews.llvm.org/D135622
Jonas Devlieghere [Fri, 28 Oct 2022 21:26:11 +0000 (14:26 -0700)]
[lldb] Add diagnostics
Around this time last year, I said on the mailing list [1] that I wanted
to to transform the reproducers into something that resembles a
sysdiagnose on Apple platforms: a collection of files containing a
variety of information to help diagnose bugs or troubleshoot issues.
This patch adds that framework. Based on lessons learned from the
reproducers, I've intentionally tried to keep it small and simple.
Different parts of LLDB can register callbacks (this is necessary for
layering purposes) that will get called when the diagnostics should be
generated.
[1] https://lists.llvm.org/pipermail/lldb-dev/2021-September/017045.html
Differential revision: https://reviews.llvm.org/D134991
Sanjay Patel [Mon, 31 Oct 2022 21:32:34 +0000 (17:32 -0400)]
[InstCombine] adjust branch on logical-and fold
The transform was just added with:
115d2f69a515cd756fa51
...but as noted in post-commit feedback, it was
confusingly coded. Now, we create the final
expected canonicalized form directly and put
an extra use check on the match, so we should
not ever end up with more instructions.
Sanjay Patel [Mon, 31 Oct 2022 21:29:03 +0000 (17:29 -0400)]
[InstCombine] add test for branch on logical-and; NFC
More coverage for the change
115d2f69a515cd756fa51 as
suggested in post-commit feedback.
Given that the transform is canonicalized with another
'not' op, we should adjust the use checks and create
that form directly.
Krzysztof Parzyszek [Mon, 31 Oct 2022 21:06:39 +0000 (14:06 -0700)]
[Hexagon] Add LLVM codegen testcases for V6_[add|sub]carryo intrinsics
Tue Ly [Mon, 31 Oct 2022 20:04:41 +0000 (16:04 -0400)]
[libc][math] Add place-holder implementation for pow function.
Add place-holder implementation for pow function to unblock libc demo
examples.
Reviewed By: michaelrj
Differential Revision: https://reviews.llvm.org/D137109
Tue Ly [Mon, 31 Oct 2022 19:42:00 +0000 (15:42 -0400)]
[libc][math] Add place-holder implementation for asin to unblock demo examples.
Add a place-holder implementation for asin to unblock libc demo
examples.
Reviewed By: michaelrj
Differential Revision: https://reviews.llvm.org/D137105
Peter Klausler [Mon, 31 Oct 2022 21:17:02 +0000 (14:17 -0700)]
[flang] Rework line of code to dodge clang 16 warning
Recode a non-short-circuiting conjunction of two Boolean function calls
into separate statements to avoid a warning from clang 16.
Jakub Kuderski [Mon, 31 Oct 2022 21:00:45 +0000 (17:00 -0400)]
[mlir][arith][spirv] Hard fail in `-convert-arith-to-spirv`
Turn legalization failures into hard failures to make sure that we do
not miss conversion pattern application failures.
Add a message on type conversion failure.
Move unsupported cases into a separate test file.
Reviewed By: antiagainst
Differential Revision: https://reviews.llvm.org/D137102
Dan Gohman [Tue, 4 Oct 2022 19:50:10 +0000 (12:50 -0700)]
[wasm-ld] Allow importing/exporting the output module's memory with arbitrary names
This adds an `--export-memory` option to wasm-ld which allows passing
a name to give to the exported memory, and extends `--import-memory` to
allow passing a <module>,<name> pair specifying where the memory should
be imported from.
This is based on https://reviews.llvm.org/D131376, with the main
difference being that it only supports exporting memory by one name.
Differential Revision: https://reviews.llvm.org/D135898
chenglin.bi [Mon, 31 Oct 2022 20:45:59 +0000 (04:45 +0800)]
[InstCombine] Precommit test for D136582; NFC
Krzysztof Parzyszek [Mon, 31 Oct 2022 19:28:25 +0000 (12:28 -0700)]
[Hexagon] Add builtins and intrinsics for V6_v[add|sub]carryo
Philip Reames [Mon, 31 Oct 2022 20:34:28 +0000 (13:34 -0700)]
[InstCombine] Allow simplify demanded transformations on scalable vectors
Differential Revision: https://reviews.llvm.org/D136475
Paul Robinson [Mon, 31 Oct 2022 20:14:56 +0000 (13:14 -0700)]
[lit] Fix typo in REQUIRES clause
Patrick Walton [Mon, 24 Oct 2022 22:31:35 +0000 (15:31 -0700)]
[AliasAnalysis] Introduce getModRefInfoMask() as a generalization of pointsToConstantMemory().
The pointsToConstantMemory() method returns true only if the memory pointed to
by the memory location is globally invariant. However, the LLVM memory model
also has the semantic notion of *locally-invariant*: memory that is known to be
invariant for the life of the SSA value representing that pointer. The most
common example of this is a pointer argument that is marked readonly noalias,
which the Rust compiler frequently emits.
It'd be desirable for LLVM to treat locally-invariant memory the same way as
globally-invariant memory when it's safe to do so. This patch implements that,
by introducing the concept of a *ModRefInfo mask*. A ModRefInfo mask is a bound
on the Mod/Ref behavior of an instruction that writes to a memory location,
based on the knowledge that the memory is globally-constant memory (in which
case the mask is NoModRef) or locally-constant memory (in which case the mask
is Ref). ModRefInfo values for an instruction can be combined with the
ModRefInfo mask by simply using the & operator. Where appropriate, this patch
has modified uses of pointsToConstantMemory() to instead examine the mask.
The most notable optimization change I noticed with this patch is that now
redundant loads from readonly noalias pointers can be eliminated across calls,
even when the pointer is captured. Internally, before this patch,
AliasAnalysis was assigning Ref to reads from constant memory; now AA can
assign NoModRef, which is a tighter bound.
Differential Revision: https://reviews.llvm.org/D136659
Paul Robinson [Mon, 31 Oct 2022 19:58:53 +0000 (12:58 -0700)]
[lit] Remove undefined keyword from a Driver test
`clang-driver` hasn't been a valid keyword for a long time.
The test fails when the keyword is removed, so added XFAIL: *
to reflect the true status of the test (XFAIL not UNSUPPORTED)
and filed issue #58711 to get it fixed.
Sanjay Patel [Mon, 31 Oct 2022 19:49:03 +0000 (15:49 -0400)]
[InstCombine] canonicalize branch with logical-and-not condition
https://alive2.llvm.org/ce/z/EfHlWN
In the motivating case from issue #58313,
this allows forming a duplicate 'not' op
which then gets CSE'd and simplifyCFG'd
and combined into the expected 'xor'.
Sanjay Patel [Mon, 31 Oct 2022 18:00:05 +0000 (14:00 -0400)]
[InstCombine] add tests for branch on logical and/or; NFC
Philip Reames [Mon, 31 Oct 2022 18:54:37 +0000 (11:54 -0700)]
[test] Add coverage for sign bits and known non-zero for scalable vectors
Peter Klausler [Mon, 17 Oct 2022 19:51:00 +0000 (12:51 -0700)]
[flang] Fix pointer definition semantic checking via refactoring
The infrastructure in semantics that is used to check that the
left-hand sides of normal assignment statements are really definable
variables was not being used to check whether the LHSs of pointer assignments
are modifiable, and so most cases of unmodifiable pointers are left
undiagnosed. Rework the semantics checking for pointer assignments,
NULLIFY statements, pointer dummy arguments, &c. so that cases of
unmodifiable pointers are properly caught. This has been done
by extracting all the various definability checking code that has
been implemented for different contexts in Fortran into one new
facility.
The new consolidated definability checking code returns messages
meant to be attached as "because: " explanations to context-dependent
errors like "left-hand side of assignment is not definable".
These new error message texts and their attached explanations
affect many existing tests, which have been updated. The testing
infrastructure was extended by another patch to properly compare
warnings and explanatory messages, which had been ignored until
recently.
Differential Revision: https://reviews.llvm.org/D136979
Jeff Niu [Thu, 27 Oct 2022 20:43:41 +0000 (13:43 -0700)]
[mlir][scf] Add scf-to-cf lowering for `scf.index_switch`
This patch adds lowering from `scf.index_switch` to `cf.switch.
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D136883
Philip Reames [Mon, 31 Oct 2022 18:36:14 +0000 (11:36 -0700)]
[test] Refresh an autogened test to pickup naming change
Matheus Izvekov [Mon, 31 Oct 2022 18:34:23 +0000 (19:34 +0100)]
Revert "[clang] ASTImporter: Fix importing of va_list types and declarations"
This reverts commit
5f820c0f55cd9d4f4520cc92eae0a1c4afe05a54.
Apparently it breaks aarch64 buildbots.
https://lab.llvm.org/buildbot#builders/188/builds/21591
Peter Klausler [Thu, 13 Oct 2022 22:22:55 +0000 (15:22 -0700)]
[flang] Be more precise about CHARACTER known length discrepancies
Many intrinsic functions in Fortran require that two or more of their
arguments have types that agree in the values of all of their type
parameters, while others only require the same type category and kind
type parameters but not lengths, including the important case of
CHARACTER. The intrinsic procedure tables need to be adjusted in
some cases so that discrepancies in character lengths that are known
at compilation time can be diagnosed as errors where they should be,
as in for example MOVE_ALLOC().
Differential Revision: https://reviews.llvm.org/D137032
Philip Reames [Mon, 31 Oct 2022 18:15:48 +0000 (11:15 -0700)]
Address post commit style comment from 087bb0f
Paul Robinson [Mon, 31 Oct 2022 17:51:29 +0000 (10:51 -0700)]
[lit] Remove undefined keyword from an MCJIT test
Changed 'XFAIL: mcjit-ia32' to 'XFAIL: i686 && windows' because the
'mcjit-ia32 feature isn't defined anywhere, or used anywhere else.
I found this as part of other work to clean up lit keywords, so
I figured I would go ahead and fix it. Verified by hacking my
lit.site.cfg.py so host_triple = target_triple = "i686-pc-windows-msvc"
and the test correctly reported XFAIL, while "i686-pc-linux" passes.
Paul Robinson [Mon, 31 Oct 2022 17:44:49 +0000 (10:44 -0700)]
Revert "[lit] Define keyword used by MCJIT test"
This reverts commit
6d859266803e2a9060c4e8770f92cc2c7bd05a3b.
Causing an XPASS on 32-bit Gentoo, as reported by mgorny.
Fangrui Song [Mon, 31 Oct 2022 17:45:00 +0000 (10:45 -0700)]
[utils] Remove ancient scripts using llvm-config --src-root
The last effective changes for the 4 scripts were more than 10 years ago.
They do grep and lines of code counting which have mature modern replacements
(rg, tokei, etc).
Reviewed By: mgorny
Differential Revision: https://reviews.llvm.org/D136969
Craig Topper [Mon, 31 Oct 2022 17:32:22 +0000 (10:32 -0700)]
[X86][Driver] Remove stale FIXME. NFC
We use getHostCPUFeatures in x86::getX86TargetFeatures so I think
this FIXME is taken care of already.
Reviewed By: RKSimon
Differential Revision: https://reviews.llvm.org/D137054
Peter Klausler [Wed, 12 Oct 2022 19:59:48 +0000 (12:59 -0700)]
[flang] Always diagnose incompatible function results when associating function pointers
There are some exceptional cases where the compiler permits association of
procedure pointers or dummy arguments with slightly incompatible procedure
targets, but they should not override any incompatibilty of function
result types.
(Includes a second fix to resolve the original motivating test failure, in
which a COMPLEX intrinsic function was getting its result kind divided by
two due to an implicit C++ conversion of the kind to a "*kind" parse tree
node, and those legacy "COMPLEX*size" type designators' values are twice
the type kind value.)
Differential Revision: https://reviews.llvm.org/D136964
Dhruva Chakrabarti [Fri, 17 Jun 2022 23:05:39 +0000 (16:05 -0700)]
[OpenMP] [OMPT] [2/8] Implemented a connector for communication of OMPT callbacks between libraries.
This is part of a set of patches implementing OMPT target callback support and has been split out of the originally submitted https://reviews.llvm.org/D113728. The overall design can be found in https://rice.app.box.com/s/pf3gix2hs4d4o1aatwir1set05xmjljc
The purpose of this patch is to provide a way to register tool-provided callbacks into libomp when libomptarget is loaded.
Introduced a cmake variable LIBOMPTARGET_OMPT_SUPPORT that can be used to control OMPT target support. It follows host OMPT support, controlled by LIBOMP_HAVE_OMPT_SUPPORT.
Added a connector that can be used to communicate between OMPT implementations in libomp and libomptarget or libomptarget and a plugin.
Added a global constructor in libomptarget that uses the connector to force registration of tool-provided callbacks in libomp. A pair of init and fini functions are provided to libomp as part of the connect process which will be used to register the tool-provided callbacks in libomptarget.
Depends on D123429
Patch from John Mellor-Crummey <johnmc@rice.edu>
(With contributions from Dhruva Chakrabarti <Dhruva.Chakrabarti@amd.com>)
Reviewed By: dreachem
Differential Revision: https://reviews.llvm.org/D123572
Craig Topper [Mon, 31 Oct 2022 17:04:36 +0000 (10:04 -0700)]
[X86] Enable EVEX GFNI instructions without avx512bw.
We only really need avx512bw for masking 256 or 512 bit GFNI
instructions due to the need for v32i1 or v64i1.
I wanted to enable 128-bit intrinsics with avx512vl, but the
__builtin_ia32_selectb_128 used in the header file requires avx512bw.
The codegen test for the same is also not using a masked instruction
because vselect with v16i1 mask and v16i8 is not legal so is expanded
before isel. To fix these issues we need a mask specific builtin and a
mask specific ISD opcode.
Fixes PR58687.
Reviewed By: pengfei
Differential Revision: https://reviews.llvm.org/D137036
Katherine Rasmussen [Fri, 28 Oct 2022 00:55:18 +0000 (17:55 -0700)]
[flang] Add atomic_and to list of intrinsics
Add the atomic subroutine, atomic_and, to the list of
intrinsic subroutines, add its last dummy argument to a check
for coindexed-object, and update test.
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D136907
Matheus Izvekov [Sun, 23 Oct 2022 14:57:12 +0000 (16:57 +0200)]
[clang] Instantiate alias templates with sugar
This makes use of the changes introduced in D134604, in order to
instantiate alias templates witn a final sugared substitution.
This comes at no additional relevant cost.
Since we don't track / unique them in specializations, we wouldn't be
able to resugar them later anyway.
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D136565
Matheus Izvekov [Sun, 23 Oct 2022 11:41:57 +0000 (13:41 +0200)]
[clang] Instantiate NTTPs and template default arguments with sugar
This makes use of the changes introduced in D134604, in order to
instantiate non-type template parameters and default template arguments
with a final sugared substitution.
This comes at no additional relevant cost.
Since we don't track / unique them in specializations, we wouldn't be
able to resugar them later anyway.
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D136564
Matheus Izvekov [Sat, 29 Oct 2022 18:36:31 +0000 (20:36 +0200)]
[clang] ASTImporter: Fix importing of va_list types and declarations
This fixes a problem where __va_list_tag was not correctly imported,
possibly leading to multiple definitions with different types.
This adds __va_list_tag to it's proper scope, so that the ASTImporter
can find it.
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D136886
Matheus Izvekov [Fri, 28 Oct 2022 20:22:39 +0000 (22:22 +0200)]
[clang] Fix handling of unexpanded packs in template argument expressions.
Closes #58679.
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D136977
Matheus Izvekov [Fri, 28 Oct 2022 16:23:02 +0000 (18:23 +0200)]
[clang] Improve error recovery for pack expansion of expressions
Closes #58673.
Signed-off-by: Matheus Izvekov <mizvekov@gmail.com>
Differential Revision: https://reviews.llvm.org/D136962
Alex Brachet [Mon, 31 Oct 2022 16:55:53 +0000 (16:55 +0000)]
[libc] Implement getopt
Differential Revision: https://reviews.llvm.org/D133487
Alexey Bataev [Mon, 31 Oct 2022 15:54:49 +0000 (08:54 -0700)]
[SLP]Fix a crash in the analysis of the compatible cmp operands.
We can skip the analysis of the operands opcodes, can compare directly
them in some cases.
Sameer Sahasrabuddhe [Mon, 31 Oct 2022 16:45:56 +0000 (22:15 +0530)]
[AAPointerInfo] check for Unknown offsets in callee
When translating offset info from the callee at a call site, first check if the
offset is Unknown. Any offset in the caller should be added only if the callee
offset is valid.
Differential Revision: https://reviews.llvm.org/D137011
LLVM GN Syncbot [Mon, 31 Oct 2022 15:53:48 +0000 (15:53 +0000)]
[gn build] Port
aee2a35ac4ab
Mengxuan Cai [Mon, 31 Oct 2022 15:45:44 +0000 (11:45 -0400)]
[LoopFuse] Ensure loops are in loop simplified form under new PM
Loop Fusion (Function Pass) requires loops in simplified form. With
legacy-pm, loop-simplify pass is added as a dependency for loop-fusion.
But the new pass manager does not always ensure this format. This patch
tries to invoke simplifyLoop() on loops that are not in simplified form
only for new PM.
Reviewed By: aeubanks
Differential Revision: https://reviews.llvm.org/D136781
Freddy Ye [Mon, 31 Oct 2022 15:01:31 +0000 (23:01 +0800)]
[X86] Add AVX-NE-CONVERT instructions.
For more details about these instructions, please refer to the latest ISE document: https://www.intel.com/content/www/us/en/develop/download/intel-architecture-instruction-set-extensions-programming-reference.html
Reviewed By: pengfei
Differential Revision: https://reviews.llvm.org/D135930
Lang Hames [Mon, 31 Oct 2022 05:14:30 +0000 (22:14 -0700)]
[JITLink][AArch46] Update aarch64 pointer jump stub sequence, add utilities.
Updates the aarch64 pointer jump stub sequence to:
ADRP x16, <ptr>@page21
LDR x16, [x16, <ptr>@pageoff12]
BR x16
from:
LDR x16, <ptr>@ldrimm19
BR x16
The old sequence can only reference pointers within +/-1Mb, which may not be
enough for large object files. The new sequence can reach pointers within
+/-4Gb. (A future pre-fixup-pass could apply range-based optimizations to
turn this into an ldr-br-nop sequence.)
Also adds createPointerJumpStubBlock and createAnonymousPointerJumpStub
utilities along the same lines as their x86-64 counterparts.
The PLTTableManager is updated to use the new utility functions.
Valentin Clement [Mon, 31 Oct 2022 15:33:45 +0000 (16:33 +0100)]
[flang] Fix code generation for polymorphic deallocation
This patch fix issues found during call to deallocate.
fir.class can be handled the same way as fir.box in fir.store operation
code generation. In MutableBox::createNewFirBox, the fir.class is
also already a boxed entity.
Reviewed By: PeteSteinfeld
Differential Revision: https://reviews.llvm.org/D137079
Serge Pavlov [Mon, 31 Oct 2022 12:09:32 +0000 (19:09 +0700)]
[Support] Do not run test on AIX
A part of the unit test CommandLineTest/BadResponseFile, added in
the commit
fd3d7a9f8cbb (Handle errors in expansion of response files)
fails on AIX. The test checks if an error is issued when `file` in
`@file` is a path to directory. This change excludes that check on
AIX platform.
Differential Revision: https://reviews.llvm.org/D136090
bixia1 [Fri, 28 Oct 2022 17:26:54 +0000 (10:26 -0700)]
[mlir][sparse] Implement the rewrite for sparse_tensor.push_back a value n times.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D136654
Jan Sjodin [Fri, 28 Oct 2022 17:54:59 +0000 (13:54 -0400)]
[OpenMP] [OMPIRBuilder] Create a new datatype to hold the unique target region info
Re-apply of:
3d0e9edd8e53fb72e85084f4170513159212839a
Reverted in:
0cb65b0a585c8b3d4a8a2aefe994a8fc907934f8
A function parameter was using the wrong type 'llvm::TargetRegion' instead of
'const llvm:: TargetRegion&', which caused the error in the address sanitizer.
The correct type is now used.
This patch puts the individual target region information attributes into a
struct so that the nested mappings are not needed and passing the information
around is simplified.
Reviewed By: jdoerfert, mikerice
Differential Revision: https://reviews.llvm.org/D136601
Lechen Yu [Mon, 31 Oct 2022 14:36:58 +0000 (15:36 +0100)]
[libomptarget] Fix a race condition in checkDeviceAndCtors
When multiple threads invoke checkDeviceAndCtors, both of them may read true
from the shared variable Device.HasPendingGlobals, and then invoke initLibrary
redundantly. Therefore only protecting the access to Device.HasPendingGlobals
is not sufficient to guarantee that initLibrary is invoked just once.
To fix this race condition, we move the invocation of initLibrary into the
critical section, and remove the same lock inside initLibrary.
Differential Revision: https://reviews.llvm.org/D136952
Jean Perier [Mon, 31 Oct 2022 14:32:29 +0000 (15:32 +0100)]
[flang][NFC] move constant lowering into its own unit
This patch moves intrinsic evaluate::Constant<T> lowering into its own
unit outside of ScalarExpr and genarr lowering so that it can
be used by the new lowering without any changes.
DerivedType lowering cannot be shared at that stage because it is too
correlated with the current lowering (requires structure constructor
and designator lowering).
The code had to be refactored quite a bit so that it could be carved
out, but the only "functional" change is that the length of character
arrays lowered by genarr is now `index` instead of `i64` (see test change).
One non-functional benefit of the change is that `toEvExpr` is not
needed anymore and some compile time copies of big constant arrays
that it was causing are removed (see old calls in previous genarr code),
although I am not sure any compile time speed-ups are visible here.
Differential Revision: https://reviews.llvm.org/D136955
Michael Buch [Mon, 31 Oct 2022 14:11:18 +0000 (14:11 +0000)]
[lldb][Test] Fix TestFrameFormatNameWithArgs.test on Linux
Be less strict about the `std::` namespace string. Depending
on platform it may contain an inline namespace, e.g., `__1`.
rkayaith [Fri, 28 Oct 2022 21:17:54 +0000 (17:17 -0400)]
[mlir][doc] Update op optional-group asm format doc
Update docs to describe the behaviour when a default-valued attribute is
used as an optional-group's anchor after D134993.
Reviewed By: Mogball
Differential Revision: https://reviews.llvm.org/D136987
Haojian Wu [Tue, 18 Oct 2022 20:25:46 +0000 (22:25 +0200)]
[include-cleaner] Add a data-structure to capture IWYU pragmas.
PragmaIncludes captures the pragma-based header-mapping information, it is used
in the "Location => Header" step to determine the final spelling header for a
symbol (rather than the header directive).
The structure is by design to be used inside the include-cleaner library
and clangd.
Differential Revision: https://reviews.llvm.org/D136071
Michał Górny [Mon, 31 Oct 2022 14:20:48 +0000 (15:20 +0100)]
Revert "[llvm] [cmake] Set EXCLUDE_FROM_ALL on gtest and TestingSupport"
This reverts commit
64959325eb470489ca5757b9bd6eef97b402a0da. It broke
the flang-aarch64-out-of-tree buildbot.
David Spickett [Mon, 31 Oct 2022 11:35:54 +0000 (11:35 +0000)]
[LLDB][AArch64] Add SME2 to disassembler test
+all includes it since
5d67b051e29c2bde42a5004634296b88542c096a.
Brendon Cahoon [Sun, 9 Oct 2022 22:34:57 +0000 (17:34 -0500)]
[BasicBlockUtils] Add a new way for CreateControlFlowHub()
The existing way of creating the predicate in the guard blocks uses
a boolean value per outgoing block. This increases the number of live
booleans as the number of outgoing blocks increases. The new way added
in this change is to store one integer to represent the outgoing block
we want to branch to, then at each guard block, an integer equality
check is performed to decide which a specific outgoing block is taken.
Using an integer reduces the number of live values and decreases
register pressure especially in cases where there are a large number
of outgoing blocks. The integer based approach is used when the
number of outgoing blocks crosses a threshold, which is currently set
to 32.
Patch by Ruiling Song.
Differential review: https://reviews.llvm.org/D127831