For AVX2+ targets this requires us to also recognise v4f64 concat(broadcast(x),broadcast(y)) -> movddup(concat(x,y))

[llvm-profgen] CSProfileGenerator::generateLineNumBasedProfile - use cast<> instead of dyn_cast<> to avoid dereference of nullptr

The pointer is always dereferenced immediately below, so assert the cast is correct instead of returning nullptr

This is a simple test fix that canonicalises the SVE mulh tests.

[NFCI][SCEV] `computeExitLimitFromCondFromBinOp()`: rely on `getSequentialMinMaxExpr()` constant relaxation

`getSequentialMinMaxExpr()` has been taught to perform this relaxation,
so rely on that now. Not sure this can be tested.

Currently, `computeExitLimitFromCondFromBinOp()` does that directly.

Clean up return value on affineDataCopyGenerate utility. Return the
actual success/failure status instead of the "number of bytes" which
isn't being used in the codebase in any way. The success/failure status
wasn't being sent out earlier.

Differential Revision: https://reviews.llvm.org/D117209

This patch implements two builtins specified in D111529.
The last __builtin_reduce_add will be seperated into another one.

Differential Revision: https://reviews.llvm.org/D116736

fix typo in comment: libcstd++ -> libstdc++

Reviewed By: wallace

Differential Revision: https://reviews.llvm.org/D117288

By default, copies are inserted right before the tensor OpOperand use. With this change, `bufferize` implementation can change the insertion point. This is needed for some ops where it would be illegal to insert a copy right before the use.

Differential Revision: https://reviews.llvm.org/D117291

Fold `memref.copy %x, %x`.

Differential Revision: https://reviews.llvm.org/D117224

The case with an inner while loop wasn't tested before. Same for outer loop with a ForeachMacro.

Test case showing a foldable icmp ('icmp ult i8* [[SCEVGEP1]], [[SCEVGEP1]]').
This can be simplified in a follow-up change.

This reverts commit 07f9fb8b.

Differential Revision: https://reviews.llvm.org/D117220

In the absence of maps, we can lower memref.copy to a memcpy.

Differential Revision: https://reviews.llvm.org/D116099

This patch creates functions which might be used to dump types.
This functionality was already implemented by  DWARFTypePrinter.
Now it could be reused. It will help D96035, which uses DWARFTypePrinter.

Differential Revision: https://reviews.llvm.org/D117134

If the source/dest is a cast that does not change shape/element type, the cast can be skipped.

Differential Revision: https://reviews.llvm.org/D117215

Since we don't merge/expand non-sequential umin exprs into umin_seq exprs,
we may have umin_seq(umin(umin_seq())) chain, and the innermost umin_seq
can have duplicate operands still.

Make the definitions of hpmcounter3-hpmcounter31,
hpmcounter3h-hpmcounter31h, mhpmcounter3-mhpmcounter31,
mhpmcounter3h-mhpmcounter31h, pmpaddr0-pmpaddr63, mhpmevent3-31, and
pmpcfg0-15 substantially less repetitive using a foreach loop.

Differential Revision: https://reviews.llvm.org/D117227

LLVM dialect supports terminators with repeated successor blocks that take
different operands. This cannot be directly expressed in LLVM IR though since
it uses the number of the predecessor block to differentiate values in its PHI
nodes. Therefore, the translation to LLVM IR inserts dummy blocks to forward
arguments in case of repeated succesors with arguments. The insertion works
correctly. However, when connecting PHI nodes to their source values, the
assertion of the insertion having worked correctly was incorrect: it would only
trigger if repeated blocks were adjacent in the successor list (not guaranteed
by anything) and would not check if the successors have operands (no need for
dummy blocks in absence of operands since no PHIs are being created). Change
the assertion to only trigger in case of duplicate successors with operands,
and don't expect them to be adjacent.

Reviewed By: wsmoses

Differential Revision: https://reviews.llvm.org/D117214

Depends on D117038.

Reviewed By: lebedev.ri

Differential Revision: https://reviews.llvm.org/D117039

The invalid undef value already triggers a verifier failure, but
then the upwards scan from the cleanuppad ends up asserting. Make
sure this is handled gacefully instead.

Based on a finding by ClangTidy readability-container-size-empty check.

Take advantage of D117117 to simplify {{\[}}[[ to [[[.

TestIOHandlerPythonREPLSigint.py is failing on Arm/Linux buildbot. I am
marking it as skip for now.

Previously, a strange trailing comment was produced:
```
namespace out { namespace {
}} // namespace out::
```
(mind the "out::").

Reviewed By: MyDeveloperDay, owenpan

Differential Revision: https://reviews.llvm.org/D117289

This doesn't require callers to put the pointer operand and the indices
in a container like a vector when calling the function. This is not
really an issue with the existing callers. But when using it from
IRBuilder the inputs are available as separate pointer value and indices
ArrayRef.

Reviewed By: lebedev.ri

Differential Revision: https://reviews.llvm.org/D117038

When masked scatter intrinsic does a uniform store to a destination
address from a source vector, and in this case, the mask is all one value.
This patch replaces the masked scatter with an extracted element of the
last lane of the source vector and stores it in the destination vector.
This patch also folds when the value in the masked scatter is a splat.
In this case, the mask cannot be all zero, and it folds to a scalar store
of the value in the destination pointer.

Differential Revision: https://reviews.llvm.org/D115724

The reinterpret load code will convert undef values into zero.
Check the uniform value case before it to produce a better result
for all-undef initializers.

However, the uniform value handling will return the uniform value
even if the access is out of bounds, while the reinterpret load
code will return undef. Add an explicit check to retain the
previous result in this case.

If we're loading from an all-undef value, we sometimes still
return zero rather than undef.

Respect the user choice, e.g. -DCMAKE_CXX_ARCHIVE_FINISH=: (to skip the
(usually) no-op step).

[clang-check] Adjust argument adjusters for clang-check to strip options blocking the static analyzer

Output generation options (like `-save-temps`) will make the analyzer not executed even `--analyze` option is provided in the driver arguments.
Besides, the original approach of adding `--analyze` option will not work when (more than one) `-fsyntax-only` options are provided in the driver arguments.

This patch fixes these two problems by using the syntax-only adjuster to remove output generation options and manually filter out redundant `-fsyntax-only` options.

In the new implementation, the adjusters added by `ClangTool` will not be removed but used as dependencies for clang-check adjusters for analyzer options.

Reviewed By: sammccall

Differential Revision: https://reviews.llvm.org/D116329

With the introduction of this flag, it is no longer necessary to enable noundef analysis with 4 separate flags.
(-Xclang -enable-noundef-analysis -mllvm -msan-eager-checks=1).
This change only covers the introduction into the compiler.

This is a follow up to: https://reviews.llvm.org/D116855

Reviewed By: vitalybuka

Differential Revision: https://reviews.llvm.org/D116633