Already we have fixed above issue. But it is not in a first merge

patches bunch. FAI, I include a patch that has should been patched

after p19. Try it. (unionfs6 for stable, unionfs for current).
First, we should merge current unionfs patch into current tree.

Above issue patch will be merged after that work. Well, whatsoever,

we should do first merge work done :)
--

   Daichi GOTO, http://people.freebsd.org/~daichi

[...]
I'm compiling this patch on my test system right now.
Another issue is userland support for unionfs, e.g. in fstat,

as described on this page:
http://c0mplx.org/src/fstat-unionfs-patch/
Do you plan to analyse/investigate this topic ?
There's another topic if one uses unionfs in jail() setups: How to

backup the files, and only those files that a different from

the base ?
If I traverse a mounted unionfs, how do I know where data is coming

from, the lower mount or the higher mount ?
If I can't tell the difference, I'll backup quite a lot of stuff

multiple times.
I've experimented a little and found no easy way to tell lower

from upper unless I open the file (which sounds expensive).

Have a look at http://c0mplx.org/src/isunionfs.c -- does this

sound like a way to go ?
--

pi@c0mplx.org           +49 171 3101372                        13 years to go !

_______________________________________________

freebsd-current@freebsd.org mailing list

http://lists.freebsd.org/mailman/listinfo/freebsd-current

To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"

It looks like interesting. But your patch has a issues.
  +		/* fprintf(stderr,"found upper vnode\n"); */

  +		res = ufs_filestat(&upper, fsp);
and
  +		/* fprintf(stderr,"found lower vnode\n"); */

  +		res = ufs_filestat(&lower, fsp);
depend on UFS. It must treat both UFS and no UFS fs.
And I am not a maintainer of fstat(1). Please contact to

maintainer. Perhaps Ed Maste(emaste) is maintainer I suppose
The isunionfs.c looks like interesting, too :) But your

program is not complete. 'below' option gives it non-correct

work. Addition it does not consider unionfs and nullfs

combibation or something like that. To check upper/lower

completely, you need the same way of your fstat(1) patch.
But your idea looks interesting :)  Keep your concern of

unionfs. To get keep concern is very good for us!
--

   Daichi GOTO, http://people.freebsd.org/~daichi

_______________________________________________

freebsd-current@freebsd.org mailing list

http://lists.freebsd.org/mailman/listinfo/freebsd-current

To unsubscribe, send any mail to "freebsd-current-unsubscribe@freebsd.org"

If you have a reasonable number of files you can maybe get away with the =
"brute force" approach - gather stat's for all files in the base FS,=20

gather stat's of all files "above" and calculate an intersection with=20

different inode numbers.

Ok, I just looked at your .c file and this is exactly what you do, sorry =
for the noise :)