[clang] [Clang] Allow devirtualization involving array subscripts with constant indices when the pointee type is known (PR #207540)

[Aaron Ballman via cfe-commits]

AaronBallman wrote:

> > > Just a heads up that we're seeing crashes after this. It's very possible that that's due to UB on our side of course, but given that, is this something that UBSan could check for and flag?
> > 
> > 
> > I would expect UBSan to flag this sort of thing but would not be surprised if it doesn't yet. Given that this landed just before the branch, if UBSan doesn't catch it, should we pull the change back until UBsan has coverage to help people fix their code (so basically, pull the change out of 23.x and keep it likely for 24.x)? (Not suggesting we go that route or not, just raising the question so we're thinking in terms of user experience for the release.)
> 
> We should see the repro first.

Yup, agreed. :-)

https://github.com/llvm/llvm-project/pull/207540