koachan wrote:
> Have you checked whether these code models are actually implemented and how
> complete they are? For example, if one model is very far from complete,
> perhaps we should be honest and say it is unsupported.
All three I listed is implemented in some capacity; the one that is the most
well-exercised, so to say, is the `medmid`/`medium` one, given that it is the
default one used by LLVM. For the rest, there's a couple tests at
`llvm/test/CodeGen/SPARC/{blockaddr,constpool,globals,func-addr}.ll` but while
the tests seem to be okay, I haven't really tried using them to do some
bigger-scale compiles...
https://github.com/llvm/llvm-project/pull/79485
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
