https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117093
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |NEW
Last reconfirmed| |2024-10-12
Ever confirmed|0 |1
--- Comment #2 from Richard Biener <rguenth at gcc dot gnu.org> ---
Indeed. This probably best fits the bswap pass which could be enhanced to
discover vector permutes.
There are some match.pd patterns doing that for very simple cases:
/* Simplify vector inserts of other vector extracts to a permute. */
(simplify
(bit_insert @0 (BIT_FIELD_REF@2 @1 @rsize @rpos) @ipos)
but here it doesn't work out greedily (also bit_insert into vec_perm
folding is missing).
I think attacking this from bswap is better (but the pass only runs late).
