madhur13490 wrote:

Thanks — the TupleType refactor, the IRTypeMapper tests, and the predicate 
i1/i8 comments+test all look good.

Couple of things I'd like to nail down though. Switching the unimplemented 
default from `Direct` to `Ignore` is a nice touch, but it doesn't actually get 
us past the debug cross-check — `Ignore` still doesn't match the classic path's 
Indirect for a register-exhausted SVE tuple, so a debug clang still aborts 
under `-fexperimental-abi-lowering` (I rebuilt with your latest and it still 
reproduces). That's fine as a known gap, I just think it should be called out 
in the PR description, and it'd be good to know #218799 is landing right behind 
this so that window stays small.

Same deal with `__mfp8` and the fixed-length `svmfloat8` — those still hit 
`reportFatalInternalError` even though the classic path handles them. Happy to 
treat it as not-implemented-yet, just want it acknowledged.

LGTM, otherwise. Thanks!


https://github.com/llvm/llvm-project/pull/221375
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to