[Bug target/70321] [6/7 Regression] STV generates less optimized code

[ienkovich at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=70321

--- Comment #14 from Ilya Enkovich <ienkovich at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #13)
> So shall we still consider moving 32-bit stv pass before the combiner
> instead of after it?  The argument about andn is no longer relevant, we have
> a one_cmpldi3_doubleword pattern now.

Previously moving the pass caused regressions. If no regressions appear now
then we should do it. It might enable some additional optimizations.