This test tracks bug of MachineVerifier to check live range segment for
EHa. Async exception can happen at any place within seh scope, not only
the call instruction. Need to teach MachineVerifier to know that.

False negative/positive rate has decreased to the degree that these
extra confirmations no longer provide any value, but only create
friction in the happy case.

When we have false analysis, people usually need to apply the fixes and
run the builds to discover the failure. At that point they can add
relevant IWYU pragmas to guide analysis, and will be more likely to
create bug reports due to extra annoyance :)

... and only look at equivalence of substituted expressions, not results
of constraint satisfaction.
This is required by the standard when matching redeclarations.

Fixes #74314.

There is already some existing machinery for that in
`TemplateInstantiator` and `Sema` exposed separate functions for
substituting expressions with intention to do that:
- `Sema::SubstExpr` should not evaluate constraints.
- `Sema::SubstConstraintExpr` should.

However, both functions used to be equivalent. Introduce a new function
that does not evaluate constraint and use it when matching declarations.

Also change implementation of `SubstConstraintExpr` to call `SubstExpr`
directly so it's obvious they behave in the same way and add a FIXME to
call out that we might need to revamp this approach in the future.

Don't rebuild the demanded elts mask every time.

If we successfully cast only the first base node as GlobalAddressSDNode / ConstantPoolSDNode / FrameIndexSDNode then we can early out as we know that base won't cast as a later type.

Noticed while investigating profiles for potential compile time improvements.

Don't wait to test that all base ptr matches have succeeded

Rename UndefZero -> UndefSel (undefined elements from Sel operand).

llvm-project/llvm/lib/CodeGen/TargetLoweringBase.cpp:570:12: error: unused variable 'ModeN' [-Werror,-Wunused-variable]
  570 |   unsigned ModeN, ModelN;
      |            ^~~~~
llvm-project/llvm/lib/CodeGen/TargetLoweringBase.cpp:570:19: error: unused variable 'ModelN' [-Werror,-Wunused-variable]
  570 |   unsigned ModeN, ModelN;
      |                   ^~~~~~
2 errors generated.

This patch implement the GlobalISel counterpart to
4d7df43f.

Bytecode versions have named constants that should be used instead of
magic numbers.

The change in c1eab576 fixed the
behavior of `getDiscardableAttrDictionary` for ops that are not using
properties to only return discardable attributes. Bytecode writer was
relying on the wrong behavior and would assume all attributes are
discardable, without appropriate testing. Fix that and add a test.

This reverts commit 42ec9761.

Reason: Broke the sanitizer buildbots. See more information on the
github comment thread at
https://github.com/llvm/llvm-project/commit/42ec976184acd40436acd7104ad715c60ca3e7ed

This reverts commit 53edf12e.

Reason: Broke the sanitizer buildbots with a memory leak. More
information available on
https://github.com/llvm/llvm-project/commit/53edf12e526704cc251b6a6917319c7cb7a653a0

The dataflow analysis framework within MLIR allows to customize the
transfer function when a `call-like` operation is encuntered.

The check to see if the analysis was executed in intraprocedural mode
was executed after the check to see if the callee had the
CallableOpInterface, and thus intraprocedural analyses would behave as
interpocedural ones when performing indirect calls.

This commit fixes the issue by performing the check for
intraprocedurality first.

Dense forward analyses were already behaving correctly.
https://github.com/llvm/llvm-project/blob/main/mlir/lib/Analysis/DataFlow/DenseAnalysis.cpp#L63



Co-authored-by: massimo <mo.fioravanti@gmail.com>

Vectors are always bit-packed and don't respect the elements' alignment
requirements. This is different from arrays. This means offsets of
vector GEPs need to be computed differently than offsets of array GEPs.

This PR fixes many places that rely on an incorrect pattern
that always relies on `DL.getTypeAllocSize(GTI.getIndexedType())`.
We replace these by usages of  `GTI.getSequentialElementStride(DL)`, 
which is a new helper function added in this PR.

This changes behavior for GEPs into vectors with element types for which
the (bit) size and alloc size is different. This includes two cases:

* Types with a bit size that is not a multiple of a byte, e.g. i1.
GEPs into such vectors are questionable to begin with, as some elements
  are not even addressable.
* Overaligned types, e.g. i16 with 32-bit alignment.

Existing tests are unaffected, but a miscompilation of a new test is fixed.

---------

Co-authored-by: Nikita Popov <github@npopov.com>

* Replaced `Any` with static types where appropriate
* Removed undocumented `__str__` and `__repr__` -- these are always
defined via `object`

Updates `generic` as `linalg.generic` (for consistency and to avoid
ambiguity) and a few other fixes.

I'm reasonably confident iff wasn't an abbreviation for if and only if
here and it was starting to bug me quite a bit.

Co-authored-by: Bogdan Graur <bgraur@google.com>

According to
https://www.nongnu.org/avr-libc/user-manual/inline_asm.html, 'G' only
represents floating point constant '0.0'. And avr-gcc also rejects other
non-zero FP values.

This pull request is a follow-up of patch:
https://github.com/llvm/llvm-project/pull/68225 and it explicitly
specifies OpenMP device libraries for Fortran OpenMP tests.

This adds legalization, notably libcall lowering for fpowi. It is a
little different to other methods as the function takes both a float and
integer register. Otherwise all vectors get scalarized and fp16 is
promoted to fp32.

Remove `nosync` as discussed in
https://github.com/llvm/llvm-project/pull/73613

[LV] Change loops' interleave count computation

A set of microbenchmarks in llvm-test-suite (https://github.com/llvm/llvm-test-suite/pull/56), when tested on a AArch64 platform, demonstrates that loop interleaving is beneficial when the vector loop runs at least twice or when the epilogue loop trip count (TC) is minimal. Therefore, we choose interleaving count (IC) between TC/VF & TC/2*VF (VF = vectorization factor), such that remainder TC for the epilogue loop is minimum while the IC is maximum in case the remainder TC is same for both.

The initial tests for this change were submitted in PRs:
https://github.com/llvm/llvm-project/pull/70272 and https://github.com/llvm/llvm-project/pull/74689.

This patch cleans up the duplicate code for folding commutative binops
over `select/phi/minmax`.

Related commits:
+ select support:
https://github.com/llvm/llvm-project/commit/88cc35b27e6c7966ab2463fa06d3dd970e88df64
+ phi support:
https://github.com/llvm/llvm-project/commit/8674a023bcacb677ce48b8831e2ae35b5aa2d8ef
+ minmax support:
https://github.com/llvm/llvm-project/commit/624973806c5644ccfa84805319b5852edb68d48d

We can use RISCVISD::VMERGE_VL with an undef passthru operand.

I had to rewrite the FMA patterns to handle both undef and non-undef
cases so we can get the tail policy.

Reuse the patterns governing the previous input file being formatted if
the current input file is from the same directory.

Make it so that PDL in pattern rewrites can be optionally disabled.

PDL is still enabled by default and not optional bazel. So this should
be a NOP for most folks, while enabling other to disable.

This only works with tests disabled. With tests enabled this still
compiles but tests fail as there is no lit config to disable tests that
depend on PDL rewrites yet.

Opcode values changed, trivial fix.

In addition to `lhs()` and `rhs()` to return left and right operands,
add `res()` to return the result value.

As suggested in https://github.com/llvm/llvm-project/pull/76210, this
patch re-organize the mc tests for apx promoted instrs, instr tests
within same cpuid would be listed in one test.
Also add explicit prefix {evex} tests and 8 displacement memory test,
promoted instrs need set No_CD8 to avoid AVX512 compress encoding.

Four variants: promoted legacy, ND (new data destination), NF (no flags
update) and NF_ND (NF + ND).

The syntax of NF instructions is aligned with GNU binutils.
https://sourceware.org/pipermail/binutils/2023-September/129545.html