[Bug target/82989] [7/8 regression ] Inexplicable use of NEON for 64-bit math

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82989

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
In both adddi3_neon and lshrdi3_neon the GPR alternatives use exactly the same
number of ?s as the NEON ones with avoid_neon_for_64bits, just the
neon_for_64bits alternatives don't have any.  So I don't see how it is a
preference of not using NEON, when TARGET_PREFER_NEON_64BITS is true, neon is
clearly preferred, otherwise there is no preferencing at all.  And indeed, for
DImode shifts by 32 the GPR regs should be prefered.  Anyway, leaving this to
ARM maintainers, don't know about the arch enough to do something.