Re: fixing coda in -current

[Maxime Villard <max%m00nbsd.net@localhost>]

Le 20/11/2018 à 16:18, Greg Troxel a écrit :
> I used to use it, and may again.  So I'd like to see it stay, partly
> because I think it's good to keep NetBS relevant in the fileystem
> research world.  I am expecting to see new upstream activity.

The problem is that CODA is not relevant in this world either... Developed
since 1987, hasn't gone very far, seems to have been dropped by everybody
except us and Linux. There appears to be some activity in the Linux source
from time to time, but nothing really significant at a first glance. Our
code, on the other side, hasn't moved at all.

The research world of today is much more around ZFS and HammerFS, neither
of which we support correctly. Keeping dead wood of the previous century for
the sake of "research" is really bad IMHO.

Filesystems (just like network drivers) consume APIs, and now we find
ourselves making blind changes without even testing, because there are just
so many things to test. So here we are, someone did try to mount a CODA
filesystem, and figured out that there are obvious KASSERTs that are
firing, because the vnode locking was revamped but not tested there...

We have the CODA code in the source tree, but we don't maintain it, and no
one else does either. I think keeping dead wood like that doesn't serve any
particular purpose, except making life harder for people that want to
revamp stuff.

But I'll leave it there.

> But, I think it makes sense to remove it from GENERIC, and perhaps have
> whatever don't-autoload treatment, so that people only get it if they
> explicitly ask for it.  That way it should not bother others.

I agree that we should disable it from GENERIC. It is as experimental as
several other FSes that are disabled, so it does make sense.