https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120652
Li Pan <pan2.li at intel dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |pan2.li at intel dot com
--- Comment #1 from Li Pan <pan2.li at intel dot com> ---
`-march=rv64gcv_zvl1024b` will make this case even failure eariler, at first
time hit expand_select_vl, the maybe_legitimize_operands will copy the const
rtx of input from index 2 to 1 if equals.
`-march=rv64gcv_zvl256` will hit this at the 6th time, while
`-march=rv64gcv_zvl1024b` will hit this at the 1st time. It sounds like the
copy_rtx makes the rtx_to_poly_int64 from the copied rtx get incorrect nunits.
