I originally thought we needed to add module file inputs for modular
deps at the same time as outputs because they depend on the
lookupModuleOutput callback, but this is not the case: they only depend
on the callback results for other modules, which have already been
computed by this point. So move them earlier so that they're set in the
CompilerInvocation at the same time as other inputs. This makes the
code easier to understand.

This change is effectively NFC, though it technically changes the module
exact value of the context hash.

Differential Revision: https://reviews.llvm.org/D142392

The object is now correct by construction.

This is the 15th commit of a patch-set that aims to change the default policy
for RVV intrinsics from TAMU to TAMA.

Please refer to the cover letter in the 1st commit (D141573) for an
overview.

Depends on D141793.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D141796

After this commit, the non-policy variants of `vid` and `viota` are no
longer available for an oveloaded version since the default policy is
now TAMA and the masked-off operand is removed.

Be noted that ALL RVV intrinsics now operate under the general
assumption that a policy behavior is "agnostic" unless specified.
Therefore this patch also changes the semantic of policy intrinsics
with the suffix of `_ta` and `tu`. These intrinsics don't have their
mask policy specified and was assumed to be undisturbed. It is now
changed to agnostic.

This is the 14th commit of a patch-set that aims to change the default policy
for RVV intrinsics from TAMU to TAMA.

Please refer to the cover letter in the 1st commit (D141573) for an
overview.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D141793

These files were oversights in D141198. The test cases are now under its
exact mnemonics. For example, test cases of `vle8`, `vle16`, `vle32`,
and `vle64` were under `vle.c`. Now they are exist under `vle8.c`,
`vle16.c`, `vle32.c`, and `vle64.c`, respectively.

This is the 13th commit of a patch-set that aims to change the default policy
for RVV intrinsics from TAMU to TAMA.

Please refer to the cover letter in the 1st commit (D141573) for an
overview.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D141792

The current way creates a fallacy that checking for
`PolicyAttrs == TAIL_AGNOSTIC` is implicitly equivalant to
`TAIL_AGNOSTIC_MASK_UNDISTURBED`. This works under the assumption that
an unmasked intrinsic has a policy of TAMU. The expression here is
mis-leading and will not be correct when the default policy is not
TAMU.

As this patch-set targets to change the default policy from TAMU to
TAMA, this commit is necessary before changing the default.

This is the 12th commit of a patch-set that aims to change the default policy
for RVV intrinsics from TAMU to TAMA.

Please refer to the cover letter in the 1st commit (D141573) for an
overview.

Reviewed By: craig.topper

Differential Revision: https://reviews.llvm.org/D141789

The build bots are failing.
This reverts commit c84d74f5.

The paper
- P2286R8 Formatting ranges
is fully implemented modulo its feature test macro. This macro has been
revised by
- LWG3750 Too many papers bump __cpp_lib_format
The new macro depends on
- P2585R0 Improving default container formatting
This paper revises parts of P2286R8 and adds new formatter
specializations. The specialization for debug strings has some wording
issues, which is addresses in this paper
- P2733R0 Fix handling of empty specifiers in std::format

Reviewed By: #libc, ldionne

Differential Revision: https://reviews.llvm.org/D142153

The attribute can be removed now as preceding patches have removed its
users.

This is the 11th commit of a patch-set that aims to change the default policy
for RVV intrinsics from TAMU to TAMA.

Please refer to the cover letter in the 1st commit (D141573) for an
overview.

Reviewed By: craig.topper, kito-cheng

Differential Revision: https://reviews.llvm.org/D141768

This patch fixes a subtle bug in how we create lvalues to reference-typed
fields. In the rare case that the field is umodeled because of the depth limit
on field modeling, the lvalue created can be malformed. This patch prevents that
and adds some related assertions to other code dealing with lvalues for
references.

Differential Revision: https://reviews.llvm.org/D142468

This implements the same behavior as D141997 but makes sure that the same detection mechanism is used between CMake and source code.

Differential Revision: https://reviews.llvm.org/D142108

[mlir][Linalg] NFC - Expose packing transpose implementation as a standalone functional-style API call

This patch adds the logic necessary to target the sparse-tensor dialect
integration tests for SVE. As the LLVM backend for AArch64 does not
currently support product reductions, the corresponding tests are
disabled for SVE.

Not all tests have been updated yet. The remaining tests will be
refactored in a separate patch shortly.

Differential Revision: https://reviews.llvm.org/D121304



Co-authored-by: Andrzej Warzynski <andrzej.warzynski@arm.com>

Caller of the two utilities is always companied with a predicate to
check for `!IsMask`, so we don't need to check for the mask policy
here.

This also removes dependency to `Policy::PolicyType::Omit`, which will
be removed.

This is the 10th commit of a patch-set that aims to change the default policy
for RVV intrinsics from TAMU to TAMA.

Please refer to the cover letter in the 1st commit (D141573) for an
overview.

Reviewed By: kito-cheng

Differential Revision: https://reviews.llvm.org/D141767

`MaskedPrototype` is initialized and used nowhere, this is a bug. The
existing codebase works correctly upon this bug because the default
policy for unmasked intrinsics is set to TAMU.

This is something to be fixed because when the default policy is
changed to TAMA, clang will generate incorrect result.

This is the 9th commit of a patch-set that aims to change the default policy
for RVV intrinsics from TAMU to TAMA.

Please refer to the cover letter in the 1st commit (D141573) for an
overview.

Reviewed By: kito-cheng

Differential Revision: https://reviews.llvm.org/D141764

Extra test coverage suggested for D140514.

This fixes the rendering of the items at https://llvm.org/docs/GettingStarted.html#stand-alone-builds

Differential Revision: https://reviews.llvm.org/D142457

See <https://github.com/llvm/llvm-project/issues/57261> for full
details. Essentially, a previous version of the psABI indicated (by my
reading) that integer scalars passed on the stack were anyext. A [later
commit](https://github.com/riscv-non-isa/riscv-elf-psabi-doc/commit/cec39a064ee0e5b0129973fffab7e3ad1710498f)
changed this to indicate that they are in fact signext/zeroext just as
if they were passed in registers.

This patch adds the change in the release notes but doesn't add a flag
to retain the old behaviour. The hope is that it's sufficiently hard to
trigger an issue due to this that it isn't worthwhile doing so.

Differential Revision: https://reviews.llvm.org/D140401

Generalize and share code for operand bundle handling. Drop the anonymous namespace (all other helper functions are local static). Rename the existing funclet test for cleanup-pads.

Reviewed By: compnerd

Differential Revision: https://reviews.llvm.org/D137945

When optimizing retain-release-sequences we insert (and delete) ObjC runtime calls. These calls need a funclet operand bundle that refers to the enclosing funclet pad whenever they are inserted in a WinEH funclet. WinEH funclets can contain multiple basic blocks. In order to find the enclosing funclet pad, we have to calculate the funclet coloring first.

Reviewed By: ahatanak

Differential Revision: https://reviews.llvm.org/D137944

The op doesn't need to be a LinalgOp, implementing TilingInterface and
DestinationStyleOpInterace is sufficient.

Reviewed By: nicolasvasilache, ftynse

Differential Revision: https://reviews.llvm.org/D142460

This patch fix issues reported for Ubuntu and possibly other platforms:
https://github.com/llvm/llvm-project/issues/45290

The latest comment on this issue points out that using dlsym rather than
the weak symbol approach to call TSan annotation functions fixes the issue
for Ubuntu.

Differential Revision: https://reviews.llvm.org/D142378

Also removes a dead argument from some tests.

Test if 2 values have different or same signbits:
(X u>> BitWidth - 1) == zext (Y s> -1) --> (X ^ Y) < 0
(X u>> BitWidth - 1) != zext (Y s> -1) --> (X ^ Y) > -1

https://alive2.llvm.org/ce/z/qMwMhj

As noted in #60242, these patterns regressed between the
14.0 and 15.0 releases - probably due to a change in
canonicalization of related patterns.

The related patterns for testing if 2 values are both
pos/neg appear to be handled already.

These were added with e76c95fb with the
right test names, but the predicates weren't updated
to match.

When emboxing to a polymorphic entity without a type source box,
the type desc address must be retrived from the input type and
not from the box type.

Reviewed By: jeanPerier

Differential Revision: https://reviews.llvm.org/D142435

Only updates the assert to check where the box is a BaseBoxType
instead of a BoxType since ClassType are also valid in that case.

Reviewed By: jeanPerier

Differential Revision: https://reviews.llvm.org/D142442

This addresses issues found by:
https://lab.llvm.org/buildbot/#/builders/92/builds/39084

The original logic resulted in inserting an integer vector into
a floating point one and vice versa. Patch also adds the missing
assert that would have caught the issue.

Differential Revision: https://reviews.llvm.org/D142303

The LLVMTestingAnnotations library that is now used by unittests
is not installed as part of LLVM.  In order to make it possible to build
unittests when performing the standalone build of clang, build
the library from LLVM sources locally.  This mirrors the existing logic
for LLVMTestingSupport.

Differential Revision: https://reviews.llvm.org/D142449

Declare callbacks extern weak (if no existing declaration exists), and
only call if the function address is non-null.

This allows to attach semantic metadata to binaries where no user of
that metadata exists, avoiding to have to link empty stub callbacks.

Once the binary is linked (statically or dynamically) against a tool
runtime that implements the callbacks, the respective callbacks will be
called. This vastly simplifies gradual deployment of tools using the
metadata, esp. avoiding having to recompile large codebases with
different compiler flags (which negatively impacts compiler caches).

Reviewed By: dvyukov, vitalybuka

Differential Revision: https://reviews.llvm.org/D142408

Differential Revision: https://reviews.llvm.org/D136138

 Add the following intrinsic:
  ADD single & multi
  SUB single & multi
 NOTE: These intrinsics are still in development and are subject to future changes.

Reviewed By: david-arm

Differential Revision: https://reviews.llvm.org/D142114

This helps to unblock our internal integrate that now relies on C++20
configuration. I will follow up with a proposal to re-enable the deleted
constructors soon, but it is a bit involved to avoid increasing the
sizes of struct.