frasercrmck wrote: > > Oh excuse me, there's currently no target-specific behaviour. I missed the > > HasFallback in > > HasFallback only applies to the two-argument form.
Oh yes, thank you. I've fixed that now. The behaviour is unconditionally zero-is-undef. > > I don't have a strong opinion about the names. > > The whole "fallback behavior" thing was written back in the early days of > clang, before we even had ubsan. I doubt we would handle that the same way > today... and I don't think we need to carry it forward to new variants. If > the user cares, they can explicitly request `__builtin_elementwise_ctz(x, > 32)` or whatever. The builtins are now renamed to cttz/ctlz. @arsenm I haven't at this point updated the documentation. I note that it always says "It is undefined behaviour if the first argument is 0 and no second argument is provided.". My literal reading of this is already that it's always undefined behaviour for all targets (as you're requesting), but perhaps with the baggage of the old builtins in mind it's still not clear? What do you think? https://github.com/llvm/llvm-project/pull/131995 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
