https://gcc.gnu.org/bugzilla/show_bug.cgi?id=93908
Jakub Jelinek <jakub at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|NEW |ASSIGNED
Assignee|unassigned at gcc dot gnu.org |jakub at gcc dot gnu.org
--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I'd say the bug is in the way how we try to lower the zero_extract store:
Trying 19, 13, 21 -> 22:
Failed to match this instruction:
(set (zero_extract:DI (mem/j:HI (plus:DI (reg/v/f:DI 60 [ s ])
(const_int 10 [0xa])) [0 +0 S2 A16])
(const_int 2 [0x2])
(const_int 7 [0x7]))
(const_int -1 [0xffffffffffffffff]))
Successfully matched this instruction:
(set (reg:HI 67)
(const_int -128 [0xffffffffffffff80]))
Successfully matched this instruction:
(set (mem/j:HI (plus:DI (reg/v/f:DI 60 [ s ])
(const_int 10 [0xa])) [0 +0 S2 A16])
(ior:HI (mem/j:HI (plus:DI (reg/v/f:DI 60 [ s ])
(const_int 10 [0xa])) [0 +0 S2 A16])
(reg:HI 67)))
I'd say storing a -1 into a zero_extract is fine, at least IMHO we shouldn't
rely on it being zero extended.
