darkbuck wrote: For the codegen part, when we generate `fshl`/`fshr`, why we need to generate `urem` to canonicalize that shift amount for `unsigned` count. Based on [LangRef](https://llvm.org/docs/LangRef.html#llvm-fshl-intrinsic), ``that shift amount is treated as unsigned amount modulo the element size of the arguments''. That `urem` is redundant and generate unnecessary code.
https://github.com/llvm/llvm-project/pull/160259 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
