lenary wrote: > > I think we shouldn't do this for RVA22, so as not to break existing users > > of that profile, who may have a toolchain that doesn't support B. > > This change makes sense to me for the RV*23 profiles, especially since your > > change to RVM23 has been accepted. > > For RVA22, I had a look at the commit history for extension support in > binutils. From what I can see (@kito-cheng please correct me if I'm wrong): > > * zic64b support was only committed in > 25f05199bb7e35820c23e802424484accb7936b1 in July 2024 > * B support was committed in c144f638337944101131d9fe6de4ab908f6d4c2d in May > 2024 > > So given we emit zic64b anyway (as it has always been in the RVA22 spec), I > think we probably do have freedom to add in B here without breaking binutils > that would have previously successfully assembled our rva22u64 output.
Alright, then I don't have any worries about RVA22. https://github.com/llvm/llvm-project/pull/113942 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits