Add a pass that propagates sharding information throughout the graph.
After this pass, each of the operations' operands and results is
annotated with a mesh.shard operation.

The pass is driven by a newly added ShardingInterface, and an implementation
for element-wise and matmul ops in the TOSA dialect is provided.

Loosen up the matching so that a missing leaf debug frame in the profile
does not prevent matching an allocation context if we can match further
up the inlined call context. This relies on the pre-inliner, which was
already the default when performing normal PGO feedback along with the
MemProf feedback, but to ensure matching is not affected by the presence
of PGO, enable the pre-inliner for MemProf feedback as well.

This builds on top of the work started in c3a302d3 to convert
LocateSymbolFile to a SymbolLocator plugin. This commit moves
LocateExecutableSymbolFile.

In g++, empty unions are not ignored like empty structs when flattening
structs to examine whether the structs can be passed via FARs in C++.
This patch aligns clang++ with g++.

Fix https://github.com/llvm/llvm-project/issues/70890.

I'm assuming that because sys::path::append defaults to native path
this fails. Since dsymutil is a Darwin tool, we can just hardcode the
UNIX path separator.

This builds on top of the work started in c3a302d3 to convert
LocateSymbolFile to a SymbolLocator plugin. This commit moves
FindSymbolFileInBundle.

This reverts commit 9d9400d7.
This reverts commit bda763ae.

The buildbot is broken and tests are failing.

/llvm-project/mlir/lib/Dialect/Mesh/IR/MeshOps.cpp:73:1: error: non-void function does not return a value in all control paths [-Werror,-Wreturn-type]
}
^
1 error generated.

Since smaia and ssaia are ratified now, we could remove their
experimental- prefix.

/llvm-project/mlir/lib/Target/LLVM/ROCDL/Target.cpp:181:40: error: unused variable 'targetMachine' [-Werror,-Wunused-variable]
  std::optional<llvm::TargetMachine *> targetMachine =
                                       ^
1 error generated.

Add a pass that propagates sharding information throughout the graph.
After this pass, each of the operations' operands and results is
annotated with a `mesh.shard` operation, and the operations themselves
are added with sharding option attributes.

The pass is driven by  a newly added `ShardingInterface`, and an implementation
for element-wise and matmul ops in the TOSA dialect is provided.

The ErrnoSetterMatcher target was renamed in a previous patch, but not
all uses were caught. This patch fixes those that remain.

Enable passing in MlirAsmState optionally (allow for passing in null) to
allow using the more efficient print calling API. The existing print
behavior results in a new AsmState is implicitly created by walking the
parent op and renumbering values. This makes the cost more explicit and
avoidable (by reusing an AsmState).

The current Darwin arm64e ABI on AArch64 systems using ARMv8.3 & newer
cores, adds authentication bits to the vtable pointer address. The
vtable address must be in addressable memory, so running it through
Process::FixDataAddress will be a no-op on other targets.

This was originally a downstream change that I hadn't upstreamed yet,
and it was surfaced by Greg's changes in
https://github.com/llvm/llvm-project/pull/67599
so I needed to update the local patch, and was reminded that I should
upstream this.

This prepares actual "direct IR codegen" for loose compressed and 2:4.
Also bit of cleanup of stale TODOs

Edited lld/ELF/Options.td to cdsort as well

CDSort function reordering outperforms the existing default heuristic (
hfsort/C^3) in terms of the performance of generated binaries while
being (almost) as fast. Thus, the suggestion is to change the default.
The speedup is up to 1.5% perf for large front-end binaries, and can be
moderate/neutral for "small" benchmarks.

High-level **perf impact** on two selected binaries:
clang-10 binary (built with LTO+AutoFDO/CSSPGO): wins on top of C^3 in
[0.3%..0.8%]
rocksDB-8 binary (built with LTO+CSSPGO): wins on top of C^3 in
[0.8%..1.5%]

More detailed measurements on the clang binary is at
[here](https://reviews.llvm.org/D152834#4445042)

The motivating use case is to have type metadata on vtables if IR instrumentation is on (without the requirement of`-fwhole-program-vtables` or `-flto`).

A related rfc is in
https://discourse.llvm.org/t/rfc-dynamic-type-profiling-and-optimizations-in-llvm/74600

Not used in all cases. Shown by python test ops.

    This commit contains the initial scaffolding to convert the
    functionality currently implemented in LocateSymbolFile to a plugin
    architecture. The plugin approach allows us to easily add new ways to
    find symbols and fixes some issues with the current implementation.

    For instance, currently we (ab)use the host OS to include support for
    querying the DebugSymbols framework on macOS. The plugin approach
    retains all the benefits (including the ability to compile this out on
    other platforms) while maintaining a higher level of separation with the
    platform independent code.

    To limit the scope of this patch, I've only converted a single function:
    LocateExecutableObjectFile. Future commits will convert the remaining
    LocateSymbolFile functions and eventually remove LocateSymbolFile. To
    make reviewing easier, that will done as follow-ups.

Previously only unwrapped values were considered for default values in
builders, expand to Attributes given the change to populate defaults. To avoid
overlap between wrapped and unwrapped, the starting index is incremented
depending on if the smallest starting index for default valued args can be
materialized in unwrapped form or not.

Given the above limitation this is pretty small change.

`optional` is unfortunately meant to also represent conditionally set which
means we cannot allow nullptr for all Attributes marked optional at the moment.

Reviewed By: Mogball

Differential Revision: https://reviews.llvm.org/D140705

Opaque ptr cleanup effort (NFC)

This patch adjusts the inline function decl check for LLVM libc to
ignore implicit functions. For the moment the plan is to ignore these
and mark the class with a macro so that it can be given the appropriate
properties without explicitly defining all its ctors/dtors.

- The latest TOSA spec for CustomOp is having attributes
  - operator_name, domain_name, and implementation_attrs
- Updated the dialect to align with the spec change

Signed-off-by: Jerry Ge <jerry.ge@arm.com>

This commit removes the no longer required `llvm.ptr<i8>` bitcasts from
AMX's legalize-for-llvm-export pass. Typed pointers have been
deprecated for a while now and it's planned to soon remove them from
the LLVM dialect.

Related PSA: https://discourse.llvm.org/t/psa-removal-of-typed-pointers-from-the-llvm-dialect/74502

Fix a bug introduced in 98c90a13 (ISel: introduce vector ISD::LRINT,
ISD::LLRINT; custom RISCV lowering), where ISD::LRINT and ISD::LLRINT
used WidenVecRes_Unary to widen the vector result. This leads to
incorrect CodeGen for RISC-V fixed-vectors of length 3, and a crash in
SelectionDAG when we try to lower llvm.lrint.vxi32.vxf64 on i686. Fix
the bug by implementing a correct WidenVecRes_XRINT.

Fixes #71187.

[ELF] Improve undefined symbol message w/ DW_TAG_variable of the enclosing symbol but w/o line number information (#70854)

The undefined symbol message suggests the source line when line number
information is available (see https://reviews.llvm.org/D31481).
When the undefined symbol is from a global variable, we won't get the
line information.
```
extern int undef;
namespace ns {
int *var[] = {
  &undef
};
// DW_TAG_variable(DW_AT_decl_file/DW_AT_decl_line) is available while
// line number information is unavailable.
}

ld.lld: error: undefined symbol: undef
>>> referenced by undef-debug2.cc
>>>               undef-debug2.o:(ns::var)
```

This patch utilizes `getEnclosingSymbol` to locate `var` and find
DW_TAG_variable for `var`:
```
ld.lld: error: undefined symbol: undef
>>> referenced by undef-debug2.cc:3 (/tmp/c/undef-debug2.cc:3)
>>>               undef-debug2.o:(ns::var)
```

It seems that currently this section is only created by the mold linker
if 2 conditions are met: 1. The PLT function was called directly. 2. The
indirect access to PLT function was found (e.g. through ADRP
relocation). Although mold created symbol for every plt entry I've
removed them in yaml file to check that .plt.got was truly disassembled
by bolt.

Some specific implementation of the offload may want more customization, and
even avoid using LLVM in-tree to dispatch the ISA translation to a custom
solution. This refactoring makes it possible for such implementation to work
without even configuring the target backend in LLVM.

Reviewers: fabianmcg

Reviewed By: fabianmcg

Pull Request: https://github.com/llvm/llvm-project/pull/71165

This commit removes all LLVM dialect typed pointers from the target
tests. Typed pointers have been deprecated for a while now and it's
planned to soon remove them from the LLVM dialect.

Related PSA:
https://discourse.llvm.org/t/psa-removal-of-typed-pointers-from-the-llvm-dialect/74502

This commit removes all LLVM dialect typed pointers from the integration
tests. Typed pointers have been deprecated for a while now and it's
planned to soon remove them from the LLVM dialect.

Related PSA:
https://discourse.llvm.org/t/psa-removal-of-typed-pointers-from-the-llvm-dialect/74502

Revert "Port Swift's merge function pass to llvm: merging functions that differ in constants (#68235)"

This reverts commit 19b5495b.

PR landed without approval, with severe quality issues.