The NPM inliner does not remove arbitrary unused internal functions, and
that is not the point of this test.

Some used analyses are optional but were relied upon in tests, the
legacy PM runs them in a different order than the NPM.

Revert "Revert "Expand existing loopsink testing to also test loopsinking using new pass manager and fix LICM bug.""

This reverts commit 562addba.

Reverted change too quickly, the failing test cases passed on the next build.
So reverting revert (to include the changes).

Make use of the more flexible constraint handling added in
a8a79c90 to decompose add nuw/sub nuw.

Summary:
This patch adds support for passing in the original delcaration name in the source file to the libomptarget runtime. This will allow the runtime to provide more intelligent debugging messages. This patch takes the original expression parsed from the OpenMP map / update clause and provides a textual representation if it was explicitly mapped, otherwise it takes the name of the variable declaration as a fallback. The information in passed to the runtime in a global array of strings that matches the existing ident_t source location strings using ";name;filename;column;row;;"

Reviewers: jdoerfert

Differential Revision: https://reviews.llvm.org/D89802

This is the same fix as 23aeadb8,
just for CloneScopedAliasMetadata rather than PropagateCallSiteMetadata.

In this case the previous outcome was incorrectly dropped metadata,
as it was not part of the computed metadata map.

The real change in the test is that the first load now retains
metadata, the rest of the changes are due to changes in metadata
numbering.

The metadata from the %c load gets dropped.

Revert "Expand existing loopsink testing to also test loopsinking using new pass manager and fix LICM bug."

This reverts commit d4ba28bd.

This reverts commit 16eb853f.

The test is failing on some Linux build bots. See the review for
an example.

In some cases, when deserializing a `CXXMethodDecl` of a `CXXSpecializationTemplateDecl`,
the call to `FunctionDecl::setPure()` happens before the `DefinitionData` member has been
populated (which appears to happen lower down in a `mergeRedeclarable` call), causing a
crash (https://reviews.llvm.org/P8228).

This diff fixes this by deferring the `FunctionDecl::setPure()` till after the `DefinitionData` has
been filled in.

Reviewed By: lxfind

Differential Revision: https://reviews.llvm.org/D86853

The VMap also contains a mapping from Argument => Instruction,
where the instruction is part of the original function, not the
inlined one. The code was assuming that all the instructions in
the VMap were inlined.

This was a pre-existing problem for the loop access metadata, but
was extended to the more common noalias metadata by
27f647d1, thus causing miscompiles.

There is a similar assumption inside CloneAliasScopeMetadata(), so
that one likely needs to be fixed as well.

The test shows noalias metadata being incorrectly applied to the
instruction producing the argument for the call.

Summary:
Expand existing loopsink testing to also test loopsinking using new pass
manager.  Enable memoryssa for loopsink with new pass manager.  This
combination exposed a bug that was previously fixed for loopsink
without memoryssa.  When sinking an instruction into a loop, the source
block may not be part of the loop but still needs to be checked for
pointer invalidation.  This is the fix for bugzilla #39695 (PR 54659)
expanded to also work with memoryssa.

Respond to review comments.  Enable Memory SSA in legacy Loop Sink pass
under EnableMSSALoopDependency option control.  Update tests accordingly.

Respond to review comments.  Add options controlling whether memoryssa is
used for loop sink, defaulting to off.  Expand testing based on these
options.

Respond to review comments.  Properly indicated preserved analyses.

Author: Jamie Schmeiser <schmeise@ca.ibm.com>
Reviewed By: asbirlea (Alina Sbirlea)
Differential Revision: https://reviews.llvm.org/D90249

As reported in https://reviews.llvm.org/D91383#2401825, this
assert breaks external -aa-eval tests. We'll have to fix this
case before re-enabling it.

If the containing allocator build uses -DGWP_ASAN_DEFAULT_ENABLED=false
then the option will default to false.  For e.g. Scudo, this is simpler
and more efficient than using -DSCUDO_DEFAULT_OPTIONS=... to set gwp-asan
options that have to be parsed from the string at startup.

Reviewed By: hctim

Differential Revision: https://reviews.llvm.org/D91463

This commit does the renaming mentioned in the title in order to bring
'spv' dialect closer to the MLIR naming conventions.

Reviewed By: antiagainst

Differential Revision: https://reviews.llvm.org/D91715

Adds support for setting the `Rule` field. In the process, refactors the code that accesses that field and adds a constructor that doesn't require a rule argument.

This feature is needed by checks that must set the rule *after* the check class
is constructed. For example, any check that maintains state to be accessed from
the rule needs this support. Since the object's fields are not initialized when
the superclass constructor is called, they can't be (safely) captured by a rule
passed to the existing constructor.  This patch allows constructing the check
superclass fully before setting the rule.

As a driveby fix, removed the "optional" from the rule, since rules are just a
set of cases, so empty rules are evident.

Differential Revision: https://reviews.llvm.org/D91544

This reverts commit c6ef6e16.

Basically, publicly linked libraries have a different semantic than components,
which link libraries privately.

Differential Revision: https://reviews.llvm.org/D91461

This reverts commit 0016ab6f.

Fix: Consume error from Expected<T>.

This is specifically required by the upcoming ExpandVectorPredication
pass (D78203) to recognize llvm.vp.* intrinsics that may ignore their
predicates.

This makes lld/MachO look more like lld/COFF and lld/ELF, as discussed
in D91640.

In some cases, the wrong amount of registers was reserved.

Also enable more v3f16 tests.

Differential Revision: https://reviews.llvm.org/D90847

This resolves LWG2503.

This reverts commit 43243208.

Consistently use SUnit *SU (or drop the argname entirely if not used like the other HazardRecognizer methods).

Silences cppcheck warnings.

We're just performing a null pointer check, we don't need the actual variable.

This patch creates a SystemZ folder in clang/test/CodeGen to contain systemz-related lit tests.

Reviewed By: muiez

Differential Revision: https://reviews.llvm.org/D91628

Adding features in OpenMP 5.1 specification, as documented in feature change history, to the 5.1 table. I alphabetized the rows of the table according to the category. For deprecating master construct, I just used 'other' as the category.

Reviewed By: jdoerfert

Differential Revision: https://reviews.llvm.org/D90802

Fixes test failures when building just `check-lld` in a clean build dir.

after 9aa78982 https://reviews.llvm.org/D90930

Differential Revision: https://reviews.llvm.org/D91160

Xref layer changes for textdocument/implementation (https://microsoft.github.io/language-server-protocol/specification#textDocument_implementation)

This currently shows all functions (implementations) that overrides a virtual function.

Differential Revision: https://reviews.llvm.org/D91702

Support more instructions in SpeculativeExecution pass:
- ExtractValue
- InsertValue
- Trunc
- Freeze

Differential Revision: https://reviews.llvm.org/D91688

Patch by: @laytonio (Layton Kifer)

Differential Revision: https://reviews.llvm.org/D91671

The compiler should treat array subscript with base pointer as a first
pointer in complex data, it is used only for member expression with base
pointer.

Differential Revision: https://reviews.llvm.org/D91660

We have workarounds for two different cases where vccz can get out of
sync with the value in vcc. This fixes them in two ways:

1. Fix the case where the def of vcc was in a previous basic block, by
pessimistically assuming that vccz might be incorrect at a basic block
boundary.

2. Fix the handling of pre-existing waitcnt instructions by calling
generateWaitcntInstBefore before examining ScoreBrackets to determine
whether there's an outstanding smem read operation.

Differential Revision: https://reviews.llvm.org/D91636

[Reassociate] Don't convert add-like-or's into add's if they appear to be part of load-combining idiom

As Wei Mi is reporting in post-commit review
  https://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20201116/853479.html
teaching -reassociate about add-like-or's (70472f34) results in breaking apart
load widening patterns, and reassociating them.

For now, simply exclude any such `or` that appears to be a root of
load widening idiom from the or->add transformation.

Note that the heuristic is greedy, it doesn't ensure that loads
can *actually* be widened into a single load.