vfsx17.patch available - continuing vfs work (expert developers only)

YONETANI Tomokazu qhwt+dfly at les.ath.cx
Fri Nov 5 06:50:17 PST 2004


On Fri, Nov 05, 2004 at 02:56:48PM +0100, Simon 'corecode' Schubert wrote:
> On 05.11.2004, at 11:44, YONETANI Tomokazu wrote:
> >>    The next patch is ready.  I haven't dealt with the nfs server, 
> >>unionfs,
> >>    or nullfs yet, I've been working around the edges cleaning things 
> >>up.
> >>
> >>	fetch http://leaf.dragonflybsd.org/~dillon/vfsx17.patch
> >
> >I'm running make -j100 buildworld with /usr/obj NFS-lookback mounted
> >from /home/nfs to see if it survives (the previous one didn't finish).
> >I'll run other benchmarks on that mount point if it's finished.

The box rebooted about 30 minutes after starting buildworld and
unfortunately I missed whatever was displayed on the console.

> >One thing I noticed is that, with the following line in /etc/exports
> >/home/nfs 127.0.0.1
> >
> >(and is mounted to /usr/obj) one would usually get `permision denied'
> >(unless you specified one of -map* options) if you for example try to 
> >create
> >a file on the NFS mounted filesystem as root; with vfsx17.patch, it 
> >creates
> >a file with uid = 4294967296 instead.  Is this normal?
> 
> If I understand correctly, this is about the same I always (ot not) 
> expected to happen: default root mount to nobody.

Correction: it was 4294967294, not 4294967296. Here's what I did and
what happened. The number looks like -2 interpreted as 32-bit unsigned
integer.

# cd /usr/obj
# touch a; su -m nobody -c 'touch b'
# ls -la
drwxrwxrwt   3 root        wheel       512 Nov  5 23:35 .
drwxr-xr-x  12 root        wheel       512 Nov  2 14:33 ..
-rw-r--r--   1 4294967294  wheel         0 Nov  5 23:35 a
-rw-r--r--   1 nobody      wheel         0 Nov  5 23:35 b

But it happens with or without vfsx17.patch anyway, so the patch is not
responsible.





More information about the Kernel mailing list