Fix clang-tidy: warning: do not use 'else' after 'return'

clang-format everything for CI

clang-format (ignoring indentation change ARMAsmPrinter.cpp in match rest of file)

Rebase and update tests

Ping. @ldionne? It doesn't seem like there is a lot of interest in this fix, since it's been up for over 2 months now in more or less the same form. I'm happy to keep making the suggested changes if it has a chance of landing, otherwise please let me know and I will just close it.

LGTM, I will leave it to the RISC-V people to approve the predecessor

If we are just keeping -riscv-no-aliases to avoid changing many tests it might be better to just remove it now. Otherwise it will likely stay forever. Maybe it is used externally though, I don't know.

Thanks everyone for the comments.

It was disabled here: https://github.com/llvm/llvm-test-suite/commit/87d67af9d8565d068b6706c081b7ae07addcb882

Thanks @Quuxplusone. If @ldionne can't be tempted into taking a look is there anyone else appropriate?

Ping @Quuxplusone

From the history it seems like @lhames might be best suited to comment on the test change

Ping

Rebase

Rewrite the patch after the changes in D101311 went in.

Thanks for the feedback!

Remove dead code paths and fix assert string

• Add tests for v8m.baseline
• Add two new pseudo instructions tCMP_SWAP_8 and tCMP_SWAP_16 to limit register for UXT instruction
• Add some asserts in ExpandCMP_SWAP

Add tests for v8m.baseline

Yes the additional argument changes the ABI and should be an opt-in for the user. This can be done by adding a new define in include/__config in the block #if defined(_LIBCPP_ABI_UNSTABLE) || _LIBCPP_ABI_VERSION >= 2.

Place breaking ABI change behind a macro

Minimize the CHECK lines in the test

Minimize the CHECK lines in the test

Ping

Closing in favour of D100944

Minor change to trigger rebuild

• At more triples to the test (Thumb1/Thumb2)
• Remove tests for floating point atomics. They tend to be bitcast to ints by the frontend and the support for them is patchy across architecture versions.
• Rebase on top of pre-commit test in D101912

Rebase to try to get CI to pass

In D101606#2728320, @dmgreen wrote:

Sounds good to me.

Whilst we are here, are any of the other uses of bitcast in arm_mve.td potentially a problem? I took a quick look and because they both converting the inputs and the outputs, I believe they will be OK. (Two wrongs make a right, if you will).

Use --check-prefixes=CHECK,CHECK-BE etc to combine common blocks.
Sorry for the churn.

remove old check lines that were not automatically removed

Use update_cc_test_checks

I will merge this today

Friendly ping

Fix bad merge

Fix bad merge

The same problem exists in the ARM backend, which I fixed internally by implementing new pseudos for the atomicrmw instructions, but hadn't got round to upstreaming it yet. Here it is for comparison: D101164.

clang-format

• [NFCI] Factor out ELF section unique ID calculation into function
• [CodeGen][ELF] Create unique sections for different flags

Hi @bd1976llvm, I was planning to update the description, add reviewers and close the other review once the CI was passing. I'll also update the patch taking your comments into consideration.

Ping. @Quuxplusone were there any outstanding issues you wanted me to address?

Update abilist for MacOS C++20

Ping

Update formatting to match the rest of the file