On Tue, Dec 17, 2019 at 05:34:32PM +0100, Lukas Czerner wrote:
> > If so, I would prefer
> > fsparam_flag_no("barrier", Opt_barrier), // barrier | nobarrier
> > fsparam_u32("barrier", Opt_barrier), // barrier=<number>
> > as the solution, with fs_parse() having been taught to allow argument-bearing
> > and argument-less options with the same name, picking the right one. That
> > way Opt_nobarrier gets removed as well...
> >
> > I'll push a branch with that stuff later today; will post when it's out...
>
> That would be great, thanks.
It took longer than I hoped, sorry ;-/ The current patchset is in
#untested.fs_parse; the really interesting part is up to
"turn fs_param_is_... into functions".
One surprising source of PITA around your patchset is ext4_show_options().
It pretty much forces you into keeping "no..." forms separate, even though
normally you could just say
fsparam_flag_no("quota", Opt_quota),
and get rid of Opt_noquota, etc.
If you keep that dependency, it'll need to be documented - right in
fs/ext4/super.c, to make sure we don't get "optimizing" followups breaking
the hell out of things.
Said that, I really doubt that token2str() is a good idea. It might make
more sense to start with separating _ext4_show_options() from that
machinery.
Another thing is that all fsparam_bool() users are modifying user-visible
ABI; use fparam_flag_no() + fsparam_u32() with the same name and same
opt - that'll give you the existing behaviour.
