https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119974
Richard Sandiford <rsandifo at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Last reconfirmed| |2025-04-29
CC| |rsandifo at gcc dot gnu.org
Status|UNCONFIRMED |WAITING
Ever confirmed|0 |1
--- Comment #1 from Richard Sandiford <rsandifo at gcc dot gnu.org> ---
Removing the PTEST wouldn't be correct in isolation, since RDFFRS+B.LAST would
test the last predicate bit (as though the PTEST were predicated on a ptrue_b)
rather than the last active element in ptest_d.
If we did more dataflow analysis, we could prove that each byte of FFR is
either 0 or 0xff and so convert a ptrue_d test into a ptrue_b test. But
really, I think it'd be better for the programmer to do this themselves.
