| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
arguments by ArrayRef instead
Silence pass by value warnings
Fixes #89210
|
|
Lambdas used in the initializer of a local class were not mangling the
name of the member.
Fixes #88906
|
|
This patch extends verification of the `omp.parallel` operation to check
it is correctly defined when taking a loop wrapper role.
In OpenMP, a PARALLEL construct can be either a (potenially combined)
block construct or a loop construct, when appearing as part of a
composite construct. This is currently the case for the DISTRIBUTE
PARALLEL DO/FOR and DISTRIBUTE PARALLEL DO/FOR SIMD exclusively.
When used to represent the PARALLEL leaf of a composite construct, it
must follow the rules of a wrapper loop operation in MLIR, and this is
what this patch ensures. No additional restrictions are introduced for
PARALLEL block constructs.
|
|
This patch replaces some `saveInsertionPoint`, `restoreInsertionPoint`
call pairs for an `InsertionGuard` instance where it makes sense within
Flang OpenMP lowering to make further modifications less error-prone.
|
|
This commit fixes a bug in the DICompositeType element ignore mode. It
seems that vectors require the presence of elements, as they otherwise
do not pass the verifier.
|
|
fixes #88348
proof:
https://alive2.llvm.org/ce/z/fJnM7t
test will be added later
---------
Signed-off-by: ZelinMa557 <3388706467@qq.com>
|
|
difference (#79842)"
This reverts commit 950bb097e11d6ee26533c00519c62df994322228
|
|
unrolling for k dimension (#88591)
Fixes correctness issue with current smmla unrolling patterns whereby
unrolling K dimension would only include the result from the last tile
along K. Updates patterns to feed previous smmla output of the previous
tile into the next one along K.
|
|
Adds a pass that folds aliasing ops into XeGPU ops.
|
|
__hipRegisterFatBinary only accepts one pointer argument. It is expected
to get the fat binary size from the header.
This patch adds a file size field to the header of the compressed
bundle.
|
|
It caused the \headerfile directive to be considered as part of
the brief description in some of our tooling.
|
|
|
|
Interestingly, this case crashed from the very beginning of the project,
at least starting by clang-3.
As a "fix" I just do the same thing as we do for concrete integers. It
might not be the best we could do, but arguably, it's still better than
crashing.
Fixes #89185
|
|
Needed for a future patch.
|
|
|
|
This change cleans up call sites. Next step is to mark the member
functions deprecated.
See https://mlir.llvm.org/deprecation and
https://discourse.llvm.org/t/preferred-casting-style-going-forward.
|
|
|
|
|
|
Defers some of the setup work until the remark is actually emitted. Move
big lambda into a helper function, and only use value capture of a
pointer. Prepare to have different messages.
|
|
The new collapse test cases define `MAX_THREADS` to be 256 and use all
available threads/logical processors on the system. This triples the
testing time on an AIX machine that has 128 logical processors. This
patch changes to use half of available logical processors to avoid over
subscribing because there are other libomp tests running at the same
time, including 2 other such collapse tests.
|
|
Related to commit b2323f43e3cdb52b4e15a7d4f434cd5c64740dd4
|
|
accept UNDEF in a splat vector of -1
Make sure we use getNOT instead of reusing the allones (with undefs) vector
|
|
This was copied from vec_add_of_not_with_undef instead of vec_add_of_not_decrement - replacing the second sub with an add
|
|
This PR adds the TPEI (Test Pending External Interruption) instruction,
along with the facility that contains it. This is a millicoded system
instruction that is not used for code generation, so it will be used
exclusively by the Assembler and Disassembler.
Accordingly, this commit also adds tests for both.
|
|
Use bool instead of function_ref.
|
|
This commit explicitly specifies the matching mode (C library function,
any non-method function, or C++ method) for the `CallDescription`s
constructed in various checkers where this transition was easy and
straightforward.
This change won't cause major functional changes, but isn't NFC because
it ensures that e.g. call descriptions for a non-method function won't
accidentally match a method that has the same name.
Separate commits will perform (or have already performed) this change in
other checkers. My goal is to ensure that the call description mode is
always explicitly specified and eliminate (or strongly restrict) the
vague "may be either a method or a simple function" mode that's the
current default.
|
|
`shouldExpandReduction` (#88874)
In the existing version, SystemZTTIImpl::shouldExpandReduction will
create a `cast` error when handling vector reduction intrinsics that
do not have the vector to reduce as their first operand, such as
`llvm.vector.reduce.fadd` and `llvm.vector.reduce.fmul`.
This commit fixes that problem by moving the cast into the case
statement that handles the specific intrinsic, where the vector
operand position is well-known.
|
|
The types we get out of expressions will not have an associated symbol
file, so the current method of looking up the type will fail. Instead, I
plumb the query through the TypeSystem class. This correctly finds the
type in both cases (importing it into the expression AST if needed). I
haven't measured, but it should also be more efficient than doing a type
lookup (at least, after the type has already been found once).
|
|
The implementation follows the ELF implementation.
|
|
This commit introduces a flag to allow skipping the potentially
recursive import of DICompositeType elements. This patch is essentially a
bandaid for the still broken recursive debug type import.
Some of our downstream inputs are produced by excessive usage of
template meta programming, and thus contain tens of thousands of types
that all participate in such recursions. Unfortunately, the series of
patches that introduces type support is not easily revertible due to
being around for a while now and Modular depending on it.
We can consider to revert this change once the type importer has show to
be very performant, but for now we are talking second vs hours to import
specific files.
|
|
OpenMP 5.2: Section 5.5.5: A procedure pointer must not appear in a
reduction clause.
Fixes #87915
|
|
This commit extends the data layout to support scalable vectors. For
scalable vectors, the `TypeSize`'s scalable field is set accordingly,
and the alignment information remains the same as for normal vectors.
This behavior is in sync with what LLVM's data layout queries are
producing.
Before this change, scalable vectors incorrectly returned the same size
as "normal" vectors.
|
|
Need to check that at least single bit is cleared for unsigned nodes
before reducing their size. Otherwise they might be treated as signed in
signed nodes.
|
|
Allow to fold or/and-and to BSL instuction for scalable vectors.
|
|
This reverts commit 74e07ab523122d6a8347b25770062ab331b6bb84.
It might be that Mask.getBitWidth() == Mask.countl_zero() (32 in my
case) and zero bitwidth2 causes the crash.
|
|
This patch fixes
https://github.com/llvm/llvm-project/pull/85592#issuecomment-2065865020.
I found this while fixing flag propagation in my
["vectorizer"](https://github.com/dtcxzyw/llvm-codegen-benchmark/blob/main/vectorize.cpp).
|
|
|
|
Solves SWDEV-449592
|
|
Restrict the types which are valid for EmitC operations. Use what is
currently supported by the emitter as restriction. Define a utility
functions for valid types, such that they can be used to restrict the
operations in table gen as well as being available for reuse in dialect
conversions.
|
|
Introduce new subclasses of VPWidenMemoryRecipe for VP
(vector-predicated) loads and stores to address multiple TODOs from
https://github.com/llvm/llvm-project/pull/76172
Note that the introduction of the new recipes also improves code-gen for
VP gather/scatters by removing the redundant header mask. With the new
approach, it is not sufficient to look at users of the widened canonical
IV to find all uses of the header mask.
In some cases, a widened IV is used instead of separately widening the
canonical IV. To handle that, first collect all VPValues representing header
masks (by looking at users of both the canonical IV and widened inductions
that are canonical) and then checking all users (recursively) of those header
masks.
Depends on https://github.com/llvm/llvm-project/pull/87411.
PR: https://github.com/llvm/llvm-project/pull/87816
|
|
These were added in https://reviews.llvm.org/D14208, which look like
they attempt to detect abs from xor+add+ashr. They do not appear to be
detecting the correct value for the src input though, which I think is
intended to be the sub(zext, zext) part of the pattern. We have pattens
from abs now, so the old invalid patterns can be removed.
Fixes #88784
|
|
Previously we use 'unsigned' as the type of ID in 'CreateDeserialized'.
And the type of `DeclID` in serialization is 'uint32_t', so there is
minor inconsistency.
Also more importantly, if we want to extend the type of DeclID from
uint32_t to uint64_t, we may be in trouble due to we forgot updating the
a lot of 'CreateDeserialized'.
So this patch tries to use semantical type 'DeclID' for
'*Decl::CreateDeserialized' to make sure it is tightly consistent.
|
|
For some reason, when I merged #89235, two lines were mis-formatted.
This patch corrects this; while I'm here, I'm also correcting other
existing formatting errors.
|
|
OpenCL environment (#89199)
If there is no information about OpenCL version we are forced to
generate OpenCL 1.0 by default for the OpenCL environment to avoid
puzzling run-times with Unknown/0.0 version output. For a reference,
LLVM-SPIRV Translator avoids potential issues with run-times in a
similar manner.
|
|
This class no longer serves any purpose; see also the discussion here:
https://reviews.llvm.org/D155204#inline-1503204
A lot of existing tests in TransferTest.cpp check for the existence of
`RecordValue`s. Some of these checks are now simply redundant and have
been
removed. In other cases, tests were checking for the existence of a
`RecordValue` as a way of testing whether a record has been initialized.
I have
typically changed these test to instead check whether a field of the
record has
a value.
|
|
Before this change -gdwarf-5 on AIX will cause backend crash, because
some DWARF5 sections are not defined in XCOFF.
Explicitly statement -gdwarf-5 as unsupported in frontend on AIX.
|
|
`PropagateResultObject()`. (#89235)
|
|
(#89243)
Similarly to #87965, delete check lines which do not have corresponding
FileCheck run lines in tiny-model-pic.ll (while having them tested in
tiny-model-static.ll).
|
|
when unset -fopenmp tries to find the library itself and in some
configurations this test fails
fix for #88545 8f07a67f9731dfcd490f8aaefac34d95f207b39c
|
|
Commit 5f87957fefb21d454f2f (pull-requst #80515) corrected some codegen
problems related to _BitInt types being used as shift exponents. But it
did not fix it properly for the special case when the shift count
operand is a signed _BitInt.
The basic problem is the same as the one solved for unsigned _BitInt. As
we use an unsigned comparison to see if the shift exponent is
out-of-bounds, then we need to find an unsigned maximum allowed shift
amount to use in the check. Normally the shift amount is limited by
bitwidth of the LHS of the shift. However, when the RHS type is small in
relation to the LHS then we need to use a value that fits inside the
bitwidth of the RHS instead.
The earlier fix simply used the unsigned maximum when deterining the max
shift amount based on the RHS type. It did however not take into
consideration that the RHS type could have a signed representation. In
such situations we need to use the signed maximum instead. Otherwise we
do not recognize a negative shift exponent as UB.
|
