As nullptr is a legitimate value, change the BaseTypeMetadataCache hash lookup/insertion to use find and
insert rather than the subscript operator. 

Also adjust getBaseTypeInfoHelper to do no insertion, but let getBaseTypeInfo do that.

Add an additional email address that has been used.

This patch replaces the IR based truncateToMinimalBitwidths with a VPlan
version. This has 3 benefits:
1) the VPlan-based version is simpler; we don't need to implement
   special codegen for each supported instruction type like the IR based
   one.
2) Removes a dependency on the cost-model after VPlan execution and
3) Removes a use of getVPValue that uses underlying values after VPlan
   execution (See removed FIXME).

Depends on D149081.

Depends on D149079.

Reviewed By: Ayal

Differential Revision: https://reviews.llvm.org/D149903

This dependency appears unused and it bloats the size of the Python
bindings for non-GPU users.

After inserting a select for the final value, update the VPlan def-use
chains. At the moment, the incorrect live-out doesn't cause a
mis-compile, as computing the final reduction value is not yet modeled
in VPlan.

In memory fold table, we have

```
 {X86::KMOVDkk, X86::KMOVDkm, 0},
 {X86::KMOVDkk_EVEX, X86::KMOVDkm_EVEX, 0}
```

where `KMOVDkm_EVEX` can use EGPR as base and index registers, while
`KMOVDkm` can't. Hence, though `KMOVkk` does not have any GPR operands,
we prefer to use `KMOVDkk_EVEX` to help register allocation.

It will be compressed to `KMOVDkk` in EVEX2VEX pass if memory folding
does not happen.

Avoids spurious diffs from change in how metadata is checked in
a future commit.

This reverts commit 030047c4.

Breaks Qt and is inconsistent with GCC.

See the following issue for details:

Fixes #74069

The existing lowering of tosa.conv2d emits a separate linalg.generic
operator to add the bias after computing the computation.

This change eliminates that additional step by using the generated
linalg.conv_2d_* operator by using the bias value as the input to the
linalg.conv_2d operation.

Rather than:

    %init = tensor.empty()
    %conv = linalg.conv_2d ins(%A, %B) %outs(%init)

    %init = tensor.empty()
    %bias = linalg.generic ins(%conv, %bias) outs(%init2) {
      // perform add operation
    }

The lowering now produces:

    %init = tensor.empty()
    %bias_expanded = linalg.broadcast ins(%bias) outs(%init)
    %conv = linalg.conv_2d ins(%A, %B) %outs(%bias)

This is the same strategy as
https://github.com/llvm/llvm-project/pull/73049 applied to convolutions.

llvm-project/llvm/lib/Target/AArch64/AArch64FrameLowering.cpp:497:21:
error: unused variable 'MFI' [-Werror,-Wunused-variable]
  MachineFrameInfo &MFI = MF.getFrameInfo();
                    ^
1 error generated.

The minimum GCC version was bumped up from 4.8 to 5.1 and then even newer
awhile ago so garbage collect the pre 4.9 workaround.

https://reviews.llvm.org/D66188

Add support for probing for dynamic allocas (variable-size objects and
outgoing stack arguments).

Co-authored-by: Oliver Stannard <oliver.stannard@linaro.org>

This patch simultaneously adds tests for `CTPOP`.

This relands 07cec73d with fix tests.

This reverts commit 07cec73d.

This patch simultaneously adds tests for `CTPOP`.

Identified with clangd.

Identified with clangd.

Identified with clangd.

 By default, `isShuffleMaskLegal` always returns true, which can result
 in the expansion of `BUILD_VECTOR` into a `VECTOR_SHUFFLE` node in
 certain situations. Subsequently, the `VECTOR_SHUFFLE` node is expanded
 again into a `BUILD_VECTOR`, leading to an infinite loop.
 To address this, we always return false, allowing the expansion of
 `BUILD_VECTOR` through the stack.

-Fix case where an op is scheduled in stage 0 and used with a distance
of 1
-Fix case where we don't peel the epilogue and a value not part of the
last stage is used outside the loop.

This is the failure:
https://lab.llvm.org/buildbot/#/builders/236/builds/7728/steps/10/logs/stdio

This started with eef8e1d2, but because
there were a couple of patches that came after that I had to revert all
3 of them because of merge conflicts.

Add a style option to specify the order that property attributes should
appear in ObjC property declarations (property attributes are things like
`nonatomic, strong, nullable`).

Closes #71323.

Differential Revision: https://reviews.llvm.org/D150083

This check was trying to find cases of invalid API usage:
incorrect/missing handle side effects and/or incorrect rewriter usage.

This check is not implemented correctly and can report false positives
in case of pointer reuse (different op created at same location). It is
unclear if such a check can be implemented given that we have both
tracking listener-based handle updates and handle consumption.

Fixes #72931.

Without folding the result of the initial tensor.dim, the
ReifyResultShapes implementation would be incorrect because it would
return a dynamic shape for a static result shape.

This moves the offload entry logic into classes and provides convenient
accessors. No functional change intended but we can now print all
offload entries (and later look them up), tested via
`OMPTARGET_DUMP_OFFLOAD_ENTRIES=<device_no>`.

When building debug version of LLVM with
`LLVM_USE_RELATIVE_PATHS_IN_FILES=On` would cause source paths to be
incorrect, and be prefixed by the build directory. This lead to source
locations like the following: `../build/llvm/...`. Such paths do not
exist, and existing debuggers can't adjust their search location because
of the incorrect prefix. Ultimately, this happened because the relative
path creation goes in the wrong direction, from source-dir to build-dir
instead of from build-dir to source-dir.

This patch swaps the directionality of the relative paths so that they
get a proper prefix from the build directory. Given a build dir at
`/build` and a project directory at `/llvm-project`, we get source
locations like: `../llvm-project/llvm/lib/Transforms/...`, which a
debugger can resolve once pointed to the correct project directory.

This is not completely NFC since we now check all 4 requirements and the
test is checking the good and the bad case for combining flags.