Replace the pattern with
x * sqrt(z/y)

---------

Co-authored-by: Matt Arsenault <arsenm2@gmail.com>

Makes sure that "scalability" flags in the
`CastAwayExtractStridedSliceLeadingOneDim` pattern are correctly
updated.

Add an additional test with a struct with multiple fields.

... and set an explicit target triple.

Should fix buildbot issues like:

https://lab.llvm.org/buildbot/#/builders/245/builds/20379/steps/5/logs/FAIL__Clang__warn-unused-filescoped_cpp

Fixes #10518
Fixes #67914
Fixes #78388
Also addresses the second example in #49103

This patch is based on suggestion from @cor3ntin in
https://github.com/llvm/llvm-project/issues/67914#issuecomment-1896011898

Disable libstdc++ assertions in the runtime library just like in
https://reviews.llvm.org/D143168.

[Clang-tidy] bugprone-too-small-loop-variable - false-negative when const variable is used as loop bound (#81183)

Changed LibASTMatcher to give an appropriate warning
when a const loop bound is initialized with a function declaration.

Fixes: #79580

On Windows the libc++ test suite sees the MSVC STL headers and may
conclude these are libc++ headers when inspecting the name. Modules
guard against forgetting to export new headers. Finding MSVC STL's
headers gives false positives. Since the CI tests non-Windows platforms
too, the validation will be disabled on Windows.

Fixes: https://github.com/llvm/llvm-project/issues/79010



---------

Co-authored-by: Louis Dionne <ldionne.2@gmail.com>

Using the `-dir` suffix for directories makes it easier to understand.

Fixes: https://github.com/llvm/llvm-project/issues/78310

Use the "main" transform-interpreter pass instead of the test pass.
This, along with the previously introduced debug extension, now allow
tutorials to no longer depend on test passes and extensions.

This was spotted by @philnik.

These work-arounds were slated for removal in LLVM-18, but missed the
deadline.

Some changes in libc++ affect the dylib. These changes are not present
on systems that use the system dylib. Currently that are the Apple
backdeployment targets. Figuring out which MacOS versions to target is
not trivial for non-Apple engineers. These shorthands make it easier to
select the proper feature make a test UNSUPPORTED or XFAIL.

During the design discussion with Louis we considered whether or not to
add preprocessor definitions to allow partial disabling of a test. This
would be useful when an existing feature is changed by modifying the
dylib. In the end we decided not to add this feature to avoid additional
complexity in the tests. Instead the test will be disabled for that
target.

```
/usr/local/google/home/lntue/experiment/llvm/llvm-project/libc/test/src/__support/FPUtil/fpbits_test.cpp:268:2: warning: extra ';' outside of a function is incompatible with C++98 [-Wc++98-compat-extra-semi]
};
 ^
1 warning generated.
```

```
In file included from /usr/local/google/home/lntue/experiment/llvm/llvm-project/libc/test/src/sys/socket/linux/bind_test.cpp:17:
/usr/local/google/home/lntue/experiment/llvm/llvm-project/libc/test/UnitTest/Test.h:17:9: warning: 'libc_make_test_file_path' macro redefined [-Wmacro-redefined]
#define libc_make_test_file_path(file_name) (file_name)
        ^
/usr/local/google/home/lntue/experiment/llvm/llvm-project/libc/test/UnitTest/LibcTest.h:20:9: note: previous definition is here
#define libc_make_test_file_path(file_name)                                    \
        ^
1 warning generated.
```

This is what test/Sema/const-eval.c does as well and without specifying
it, some windows builders are broken:

https://lab.llvm.org/buildbot/#/builders/265/builds/2453

Adds a new output target "srec" to write SREC files from ELF inputs.

https://en.wikipedia.org/wiki/SREC_(file_format)

We should only warn if the default version is the one that is unused.

Fixes: https://github.com/llvm/llvm-project/issues/80227

Fixes: https://github.com/llvm/llvm-project/issues/81043

Remove a stale FIXME comment and improve the test.

[Flang] Fix NULLIFY statement that returns too early for multiple procedure pointer objects. (#81164)

The current code that handles NULLIFY statement for procedure pointer
returns after the 1st object.
This PR is to remove the `return` so it can nullify multiple procedure
pointer objects.

Add tests showing missed simplifications due to phase ordering.

Instead of returning false, just ignore the operation and return true;
This gives us the desired diagnostic behavior in the added test case.

Covers CWG issues
[1150](https://cplusplus.github.io/CWG/issues/1150.html),
[1487](https://cplusplus.github.io/CWG/issues/1487.html),
[1567](https://cplusplus.github.io/CWG/issues/1567.html),
[1738](https://cplusplus.github.io/CWG/issues/1738.html),
[2273](https://cplusplus.github.io/CWG/issues/2273.html),
[2277](https://cplusplus.github.io/CWG/issues/2277.html),
[2356](https://cplusplus.github.io/CWG/issues/2356.html),
[2504](https://cplusplus.github.io/CWG/issues/2504.html).

On top of the wording in proposed resolutions,
[P0136R1](https://wg21.link/p0136r1) "Rewording inheriting constructors
(core issue 1941 et al)" is a very relevant paper.

Note that status for 1738 `sup P0136R1` is not officially recognized by
CWG, but saying `yes` or `no` seems even more confusing to me. Official
resolution is to reject certain code, but Clang is the only
implementation that still rejects it to this day:
https://godbolt.org/z/b1W8jc1o5. GCC rejected it until 9, now it's
accepted: https://godbolt.org/z/of6oh4sdT

The fold for icmp (gep (p, i1), gep (p, i2)) to icmp (i1, i2) is
currently limited to one of the GEPs either having one use or a constant
offset. I believe this is to avoid duplicating complex arithmetic both
in the GEP and the offset comparison.

This patch instead does the same thing that the indexed compare fold
does, which is to rewrite the GEP into i8 form if necessary, so that the
offset arithmetic is not repeated after the transform.

I ran into this problem in a case where there are multiple conditions on
the same pointer, which prevents them from getting folded.

yaml2obj creates invalid object files even when the input was created by
obj2yaml using a valid object file. On the other hand, yaml2obj is used
to intentionally create invalid object files for testing purposes.

This update balances using specified input values when provided and
computing file offsets and sizes if necessary.

Return *this from operator++. NFC, this just allows using ++Iter in
an expression in future patches.

Almost all uses of `*TreePatternNode` expect it to be non-null. There
was the occasional check that it wasn't, which I have removed. Making
them references makes it clear that they exist.

This was attempted in 2018 (1b465767)
for `TreePatternNode::getChild()` but that was reverted.

These are added to the ArmSVE dialect for consistency with LLVM, which
registers SME2 intrinsics that don't require ZA under SVE.

This mechanism is introduced by #68324.

This refactor makes the prototype and attributes clear.

Reviewers: asb, kito-cheng, philnik777, topperc, preames

Reviewed By: topperc

Pull Request: https://github.com/llvm/llvm-project/pull/80280

RISCV target will use this parameter, so we need a way to specify
it.

Reviewers: AaronBallman, philnik777

Reviewed By: AaronBallman

Pull Request: https://github.com/llvm/llvm-project/pull/80279