On Feb 11, 2021, Alexandre Oliva <ol...@adacore.com> wrote: > How does this look?
> for gcc/ChangeLog > PR tree-optimization/94092 > * builtins.c (try_store_by_multiple_pieces): New. > (expand_builtin_memset_args): Use it. If target_char_cast > fails, proceed as for non-constant val. Pass len's ctz to... > * expr.c (clear_storage_hints): ... this. Try store by > multiple pieces after setmem. > (clear_storage): Adjust. > * expr.h (clear_storage_hints): Likewise. > (try_store_by_multiple_pieces): Declare. > * tree-loop-distribution.c: Include builtins.h. > (generate_memset_builtin): Propagate dst_base alignmen to mem. > * tree-ssanames.c (get_nonzero_bits): Zero out low bits of > integral types, when a MULT_EXPR INTEGER_CST operand ensures > the result will be a multiple of a power of two. I forgot to mention this passed regstrap on x86_64-linux-gnu, as well as some cross testing of riscv32-elf. I've also regstrapped it on x86_64-linux-gnu along with a patch for testing purposes, that tried try_store_by_multiple_pieces before setmem in all 3 locations where they are called, which gives me some confidence that the implementation is reasonably robust. Is this ok to install? (if not right now, perhaps in stage1) -- Alexandre Oliva, happy hacker https://FSFLA.org/blogs/lxo/ Free Software Activist GNU Toolchain Engineer Vim, Vi, Voltei pro Emacs -- GNUlius Caesar
