Although nice to have to prove the freeze can be moved, this can fail
immediately after freeze(op(...)) -> op(freeze(),freeze(),...) creation
if any of the new freeze nodes now prevents value tracking from seeing
through to the source values (e.g. shift amounts/element indices are in
bounds etc.).
This will allow us to remove the isGuaranteedNotToBeUndefOrPoison checks
inside canCreateUndefOrPoison that were discussed on #146361
Firstly, this commit requires that all types are signless in the strict
mode of the validation pass. This is because signless types on
operations are required by the TOSA specification. The "strict" mode in
the validation pass is the final check for TOSA conformance to the
specification, which can often be used for conversion to other formats.
In addition, a conversion pass `--tosa-convert-integer-type-to-signless`
is provided to allow a user to convert all integer types to signless.
The intention is that this pass can be run before the validation pass.
Following use of this pass, input/output information should be carried
independently by the user.
When compiling with `-march=armv9-a+nosve` we found that Clang still
defines the `__ARM_FEATURE_SVE2` macro, which is explicitly set in
`setArchFeatures` when compiling for armv9-a.
After some experimenting, I found out that the list of features passed
into `AArch64TargetInfo::handleTargetFeatures` has already been expanded
and takes into account `+no[feature]` and has already expanded features
like `armv9-a`.
From that I conclude that `setArchFeatures` is no longer required.
Verify that the alignments specified by clang TargetInfo match the
alignments specified by LLVM data layout, which will hopefully prevent
accidental mismatches in the future.
This currently contains opt-outs for a number of of existing mismatches.
I'm also skipping the verification if options like `-malign-double` are
used, or a language that mandates sizes/alignments that differ from C.
The verification happens in CodeGen, as we can't have an IR dependency
in Basic.
In OpenMP Version 5.1, the tile and unroll directives were added. When
using these directives, it is possible to nest them within other OpenMP
Loop Constructs. This patch enables the semantics to allow for this
behaviour on these specific directives. Any nested loops will be stored
within the initial Loop Construct until reaching the DoConstruct itself.
Relevant tests have been added, and previous behaviour has been retained
with no changes.
See also, #110008
For the checks whether certain intrinsics are used, work with intrinsic
IDs instead of intrinsic names.
This also exposes that some of the checks were incorrect, because the
intrinsics were overloaded. There is no efficient way to determine
whether these are used. This change explicitly documents which
intrinsics are not checked for this reason.
As the data layout a few lines further up specifies, the int, long and
pointer alignment should be 16 instead of the default of 32.
The long long alignment is also incorrect, but that would require a
change to the data layout as well.
Comparison with GCC, which consistently uses 2 byte alignment:
https://gcc.godbolt.org/z/K3x6a7dEf At least based on some spot checks,
the changes to bit field layout also make use match GCC now.
This was found by https://github.com/llvm/llvm-project/pull/144720.
They are left over from our previous attempt at DWARF64. The new attempt
is not using them, and they also don't have equivalents in the llvm
DWARFDataExtractor class.
We added LNLP sched model recently, PFM counter bounding names needs to
match cpu string.
llvm-exegesis wont produce results without correct naming.
Co-authored-by: mattarde <mattarde@intel.com>
reading, and one bug in the new RegisterContextUnifiedCore class.
The PR I landed a few days ago to allow Mach-O corefiles to augment
their registers with additional per-thread registers in metadata exposed
a few bugs in the x86_64 corefile reader when running under different CI
environments. It also showed a bug in my RegisterContextUnifiedCore
class where I wasn't properly handling lookups of unknown registers
(e.g. the LLDB_GENERIC_RA when debugging an intel target).
The Mach-O x86_64 corefile support would say that it had fpu & exc
registers available in every corefile, regardless of whether they were
actually present. It would only read the bytes for the first register
flavor in the LC_THREAD, the GPRs, but it read them incorrectly, so
sometimes you got more register context than you'd expect. The LC_THREAD
register context specifies a flavor and the number of uint32_t words;
the ObjectFileMachO method would read that number of uint64_t's,
exceeding the GPR register space, but it was followed by FPU and then
EXC register space so it didn't crash. If you had a corefile with GPR
and EXC register bytes, it would be written into the GPR and then FPU
register areas, with zeroes filling out the rest of the context.
The code following `llvm_unreachable` is optimized out in Release builds. In this case, `Embedder::create` do not seem to return `nullptr` causing `CreateInvalidMode` test to break. Hence removing `llvm_unreachable`.
This patch fixes:
llvm/unittests/Analysis/FunctionPropertiesAnalysisTest.cpp:132:12:
error: moving a local object in a return statement prevents copy
elision [-Werror,-Wpessimizing-move]
This change simplifies the API by removing the error handling complexity.
- Changed `Embedder::create()` to return `std::unique_ptr<Embedder>` directly instead of `Expected<std::unique_ptr<Embedder>>`
- Updated documentation and tests to reflect the new API
- Added death test for invalid IR2Vec kind in debug mode
- In release mode, simply returns nullptr for invalid kinds instead of creating an error
(Tracking issue - #141817)
For non-SHF_ALLOC sections, sh_addr is set to 0.
Skip sections without SHF_ALLOC flag, so `minVA` will not be set to 0
with non-SHF_ALLOC sections, and the size of non-SHF_ALLOC sections will
not contribute to `maxVA`.
Due to the changes made in #143479, size of `FeatureMap` reflects the number of features used. Hence replacing `NumberOfFeatures` with `FeatureMap.size()` in the `DevelopmentModeInlineAdvisor`.
This patch makes determining alignment and width of BitInt to be target
ABI specific and makes it consistent with [Procedure Call Standard for
the LoongArch™ Architecture] for LoongArch target
(https://github.com/loongson/la-abi-specs/blob/release/lapcs.adoc).
The TestOpenACCInterfaces test loads dialects including HLFIR and DLTI
(for data layout). The appropriate dependencies were missing leading to
link failures:
TestOpenACCInterfaces.cpp:(.text._ZNK12_GLOBAL__N_124TestFIROpenACCInterfaces20getDependentDialectsERN4mlir15DialectRegistryE+0x66):
undefined reference to
`mlir::detail::TypeIDResolver<hlfir::hlfirDialect, void>::id'
TestOpenACCInterfaces.cpp:(.text._ZNK12_GLOBAL__N_124TestFIROpenACCInterfaces20getDependentDialectsERN4mlir15DialectRegistryE+0x141):
undefined reference to `mlir::detail::TypeIDResolver<mlir::DLTIDialect,
void>::id'
This PR is part of https://discourse.llvm.org/t/rfc-profile-information-propagation-unittesting/73595
In a slight departure from the RFC, instead of a brand-new `MD_prof_unknown` kind, this adds a first operand to `MD_prof` metadata. This makes it easy to replace with valid metadata (only one `MD_prof`), otherwise sites inserting valid `MD_prof` would also have to check to remove the `unknown` one.
The patch just introduces the notion and fixes the verifier accordingly. Existing APIs working (esp. reading) `MD_prof` will be updated subsequently.
Fixes#141840
This PR implements support for the `memcpy` intrinsic in the DXIL
CBuffer Access pass with the following restrictions:
- The CBuffer Access must be the `src` operand of `memcpy` and must be
direct (i.e., not a GEP)
- The type of the CBuffer Access must be of an Array Type
These restrictions greatly simplify the implementation of `memcpy` yet
still covers the known uses in DML shaders.
Furthermore, to prevent errors like #141840 from occurring silently
again, this PR adds error reporting for unsupported users of globals in
the DXIL CBuffer Access pass.
fir.class is treated similarly as fir.box - but it has one key
distinction which is that it doesn't hold an element type. Thus the
categorization logic was mishandling this case for this reason (and also
the fact that it assumed that a base object is always a fir.ref).
This PR improves this handling and adds appropriate test exercising both
a class and a class field to ensure categorization works.
Instead of looking up the narrower reduction type via getRecurrenceType
we can generate the needed extend directly at constructiond re-use the
truncated value from the loop.
PR: https://github.com/llvm/llvm-project/pull/141860
This PR ensures we correctly restore the cursor column after resizing
the statusline. To ensure we have space for the statusline, we have to
emit a newline to move up everything on screen. The newline causes the
cursor to move to the start of the next line, which needs to be undone.
Normally, we would use escape codes to save & restore the cursor
position, but that doesn't work here, as the cursor position may have
(purposely) changed. Instead, we move the cursor up one line using an
escape code, but we weren't restoring the column.
Interestingly, Editline was able to recover from this issue through the
LineInfo struct which contains the buffer and the cursor location, which
allows us to compute the column. This PR addresses the bug by having
Editline "refresh" the cursor position.
Fixes#134064
