christos@zoulas.com (Christos Zoulas) writes:

> On Apr 22,  2:08pm, enami@sm.sony.co.jp (enami tsugutomo) wrote:
> -- Subject: Re: statfs changes
> 
> Previously the code did not check in *some* filesystems for length
> limitations, and truncated partially the statfs info or did not
> set it correctly (i.e. it bailed out early if a copy{,in}str()
> failed. Now it enforces the limits on all the filesystems.
> Unfortunately the limit is 90 characters now. Having incorrect
> statfs info can have security implications.  We could increase the
> limits, or it is trivial to remove the checking...

Hmm.

- Possibiliy we need to extend struct statfs so that it can have full
  path.

- Current error handling when set_statfs_info() failed is poor.  E.g.,
  ffs_mount returns an error but doesn't undo the actual mount work.

- fs_fsmnt (in case of ffs) isn't updated.

enami.