'nohost' is only valid on routine, and states that the compiler
shouldn't compile this routine for the host. It has no arguments, so no
checking is required besides putting it in the AST.
If mask is a scalar, it always converts to !fir.box<!fir.array<1xi1>>.
The wrong value may be picked up when passing to the function
on the big endian platform. This patch is to do the conversion
based on the original type of the mask and convert the value to
i1 after the load.
These 4 clauses are mutually exclusive, AND require at least one of
them. Additionally, gang has some additional restrictions in that only
the 'dim' specifier is permitted. This patch implements all of this, and
ends up refactoring the handling of each of these clauses for
readabililty.
emitSCEVChecks checks if SCEVCheckCond matches zero, and returns
nullptr. However, it sets SCEVCheckCond as used before it does this,
which prevents it from being removed during cleanup, resulting in
unreachable blocks being emitted. Fix this.
OpPassManager contains a field of type std::unique_ptr which
is not guaranteed to be trivially relocatable so we cannot use
llvm::array_pod_sort.
Reviewers: River707, joker-eph
Reviewed By: joker-eph
Pull Request: https://github.com/llvm/llvm-project/pull/129968
Spec can be found here https://github.com/intel/llvm/pull/15225
TODO for future patches:
- During spec review need to decide whether only FunctionCall or Atomic
instructions can be decorated and if not - move the code around adding
handling for other instructions;
- Handle optional string metadata;
- Handle LLVM atomic instructions;
- Handle SPIR-V friendly atomic calls returning via sret argument.
Signed-off-by: Sidorov, Dmitry <dmitry.sidorov@intel.com>
This patch replaces all instances of ubuntu-latest with ubuntu-24.04
(outside of the entries in libc++) based on the guidelines in the LLVM
CI best practices doc (https://llvm.org/docs/CIBestPractices.html).
This patch replaces all instances of ubuntu-latest with ubuntu-24.04
based on the guidelines in the LLVM CI best practices doc
(https://llvm.org/docs/CIBestPractices.html).
This PR moves vector.insert canonicalizers for DenseElementsAttr (splat
and non splat case) to folders. Folders are local, and it's always
better to implement a folder than a canonicalizer.
This PR is mostly NFC-ish, because the functionality mostly remains
same, but is now run as part of a folder, which is why some tests are
changed, because GreedyPatternRewriter tries to fold by default.
Updates some instances of plain `return failure();` in VectorToSCF.cpp
with `return notifyMatchFailure();` and a description (usually copied
from the nearby comment).
There's many more "plain" `return failure();` left, but ATM I only
have the cycles for the ones updated here.
This change implements variable linkage types in ClangIR except for
common linkage which requires Comdat support.
---------
Co-authored-by: Morris Hafner <mhafner@nvidia.com>
Co-authored-by: Henrich Lauko <xlauko@mail.muni.cz>
Need to use consistent storages for unique elements, when going to
iterate over them to avoid non-determinism in reused elements analysis.
Fixes#130082
Clang's prior documentation for the `CPATH` environment variable stated that
paths it specifies are added as system header search paths. The actual behavior
is that such paths are treated as though they were passed via `-I` options at
the end of the driver command line and are thus added as non-system (user)
header search paths.
The documentation additionally claimed that empty path entries in the `CPATH`
environment variable are ignored. This was also incorrect; Clang treats empty
entries as nominating the compiler's current working directory; as though `.`
was specified.
Clang's behavior is consistent with gcc as documented at
https://gcc.gnu.org/onlinedocs/gcc/Environment-Variables.html#index-CPATH.
This change aligns Clang's documentation with the behavior actually observed.
Additional editorial changes are included to clarify that the related
`C_INCLUDE_PATH`, `CPLUS_INCLUDE_PATH`, `OBJC_INCLUDE_PATH`, and
`OBJCPLUS_INCLUDE_PATH` environment variables specify additional paths that
are treated as system header search paths (in contrast to `CPATH`).
Fixes issue #49742.
Instead of manually adding a note pointing to the relevant template
parameter to every relevant error, which is very easy to miss, this
patch adds a new instantiation context note, so that this can work using
RAII magic.
This fixes a bunch of places where these notes were missing, and is more
future-proof.
Some diagnostics are reworked to make better use of this note:
- Errors about missing template arguments now refer to the parameter
which is missing an argument.
- Template Template parameter mismatches now refer to template
parameters as parameters instead of arguments.
It's likely this will add the note to some diagnostics where the
parameter is not super relevant, but this can be reworked with time and
the decrease in maintenance burden makes up for it.
This bypasses the templight dumper for the new context entry, as the
tests are very hard to update.
This depends on #125453, which is needed to avoid losing the context
note for errors occuring during template argument deduction.
vitalybuka identified a fix here that fixes the issue, and lets us make
fewer copies! This applies his patch plus reapplys the original.
This reverts commit c4c29b95a6e6809017e71e85f33faecfe85d88b2.
Drive-by: Enables test coverage for `ranges::stable_sort` with proxy
iterators, and changes "constexpr in" to "constexpr since" in comments
in `<algorithm>`.
When defining functions with two or more format attributes, if the
format strings don't have the same format family, there is a false
positive warning that the incorrect kind of format string is being
passed at forwarded format string call sites.
This happens because we check that the format string family of each
format attribute is compatible before we check that we're using the
associated format parameter. The fix is to move the check down one
scope, after we've established that we are checking the right parameter.
Tests are updated to include a true negative and a true positive of this
situation.
From what I understand, we only create VPReductionRecipes for in-loop
reductions, and we don't currently support in-loop AnyOf reductions.
We only create VPReductionRecipes in the !PhiR->isInLoop() section of
adjustRecipesForReductions, and this comment from the initial patch
seems to confirm this
https://reviews.llvm.org/D108136#anchor-inline-1038338, so I think we
can remove this check in the condition logic.
I checked compiling SPEC 2017 with -prefer-inloop-predicates and the
added assertion doesn't trigger.
We previously allowed duplicates of auto/seq/independent on a 'loop'
construct. This is disallowed by the restriction (which says exactly one
of...), so this patch ensures they are disallowed.
Need to use consistent storages for unique elements, when going to
iterate over them to avoid non-determinism in reused elements analysis.
Fixes#130082
pointer allocation is done through `AllocateValidatedPointerPayload`.
This function was not updated to use the registered allocators in the
descriptor to perform the allocation. This patch makes use of the
allocator.
The footer word is not set and not checked for allocator other than the
default one. The support will likely come in a follow up patch but this
will necessitate more functions to be registered to be able to set and
get the footer value when the allocation in on the device.
When generating aliases from `OpAsm{Dialect,Type,Attr}Interface`, the
result would be sanitized and if the alias provided by the interface has
a trailing digit, AsmPrinter would attach an underscore to it to
presumably prevent confliction.
#### Motivation
There are two reasons to motivate the change from the old behavior to
the proposed behavior
1. If the type/attribute can generate unique alias from its content,
then the extra trailing underscore added by AsmPrinter will be strange
```mlir
func.func @add(%ct: !ct_L0_) -> !ct_L0_
%ct_0 = bgv.add %ct, %ct : (!ct_L0_, !ct_L0_) -> !ct_L0_
%ct_1 = bgv.add %ct_0, %ct_0 : (!ct_L0_, !ct_L0_) -> !ct_L0_
%ct_2 = bgv.add %ct_1, %ct_1 : (!ct_L0_, !ct_L0_) -> !ct_L0_
return %ct_2 : !ct_L0_
}
```
Which aesthetically would be better if we have `(!ct_L0, !ct_L0) ->
!ct_L0`
2. The Value name behavior is that, for the first instance, use no
suffix `_N`, which can be similarly applied to alias name. See the IR
above where the first one is called `%ct` and others are called `%ct_N`.
See `uniqueValueName` for detail.
#### Conflict detection
```mlir
!test.type<a = 3> // suggest !name0
!test.type<a = 4> // suggest !name0
!test.another<b = 3> // suggest !name0_
!test.another<b = 4> // suggest !name0_
```
The conflict detection is based on `nameCounts` in `initializeAliases`,
where
In the original way, the first two will get sanitized to `!name0_` and
`initializeAlias` can assign unique id `0, 1, 2, 3` to them.
In the current way, the `initializeAlias` uses `usedAliases` to track
which name has been used, and use such information to generate a suffix
id that will make the printed alias name unique.
The result for the above example is `!name0, !name0_1, !name0_,
!name0_2` now.
