Pushed fixes to the release/10.x branch, should we revert?

Hi Tom,

We got some user requests for (clangd) cherrypicks in the next point release, which nudged us in digging up a list of other bugfixes (mostly crashers) to backport.

And this is where I messed up: we reviewed the list and then pushed them to the 10.x branch, forgetting about “Patches applied to the release branch may only be applied … with approval from the release manager”.
(The excuse: I’ve mostly been active with major releases, and we had an understanding with Hans that any clangd-only changes were fine up to RC2 or so. Point releases are different of course and this is your call, I’d just forgotten the actual policy)

Anyway, I’m happy to revert any/all of these changes if you don’t think they belong on the branch (or pending a decision). They’re all bugfixes and suitable in principle for a point release, but up to you whether they belong in this one.
The commits are this range. Of these all affect clangd only, except a trivial clang crash fix by Richard Smith and a small tweak to a clang-tidy check. (There are a couple of changes to clang/lib/Syntax that are only used in clangd).

Let me know what you’d prefer (or just go ahead and revert if that’s easiest).
Sorry about the churn.


Thank you! I filed https://bugs.llvm.org/show_bug.cgi?id=46270 and attempted to summarize. Please let me know what else is needed!

Haojian and Kadir (CCed) are familiar with the patches and happy to have them merged.
The clang-proper patches are the ones where we don’t have owner-level expertise: I’ll ask Richard to sign off on cbc9b92df4582617314b08d1ecef41d355733874 (which is the one he didn’t author himself).

Thanks, Sam