On Tue December 13 2011, Vasanth Ragavendran wrote:
> Hi
> 
> If I introduce ${HOSTCC} into Makefile in aufs-util (and aufs2-util) for
> > compileing c2sh, c2tmac and ver (probably), then will the problem solved?
> > How do you think?
> >
> > I am considering a new ioctl to get the aufs version dynamicall. It if
> > goes well, ver.c won't be necessary probably. But it may bring another
> > confusion.
> >
> >
> > It might work, however the plink.c which contains a O_CLOEXEC defined in
> fcntl.h creates problem. i guess the location of this file containing
> O_CLOEXEC differs from architecture to architecture (i'm not too sure of
> this though)
>

That might be related to why umount.aufs:auplink:plink.c is generating
the "Inappropriate IOCTL for device message" and refusing to take
down an auFS stack created using the (default) plink option.
 
> plink.c --> /usr/include/fcntl.h --> /usr/include/bits/fcntl.h
> 
> 
> u had mentioned it would be located this way but for my scenario it was
> this way
> 
> /usr/include/linux/fcntl.h --> /usr/include/asm/fcntl.h -->
> /usr/include/asm-generic/fcntl.h
> 
> so i literally changed the header file as <linunx/fcntl.h> in the plink.c
> file. It was able to cross-compile only after that.
> 
> Further, back to my problem with the aufs working.
> 
> Ok, as long as you don't use /etc/mtab, mount.aufs is unrelated to the
> path of branches.
> The pathname of mount.aufs is totally upto your mount(8).
> The mount(8) in util-linux packages expects it is placed under /sbin.
> 
> Okay currently i'm having this way. in the initramfs i have a sbin folder
> with the mount.aufs, umount.aufs, auibusy, auplink (all cross compiled).
> here i do a aufs mount to /aufs folder and mount(8) (pls note this mount(8)
> is a link to the busybox bin file) returns these and pls note there is no
> /etc/mtab file and these values are the same when i do a "cat /proc/mount".
> 

Which may be one source of the confusion.
With the introduction of namespaces in the kernel, the proper place
to look is:
cat /proc/self/mount

Since the (visible) mounts vary by namespace and a process might
be in any namespace - lookup must now be done reference to /proc/self/*

And in the August, 2011 kernel (3.0.4) used in my au/chroot example
is horribly broken.

Mike
> rootfs on / type rootfs (rw)
> /proc on /proc type proc (rw,relatime)
> tmpfs on /tmp type tmpfs (rw,relatime)
> /dev/mtdblock2 on /tmp/jffs2 type jffs2 (rw,relatime)
> /dev/loop0 on /tmp/sqfsh type squashfs (ro,relatime)
> aufs on /aufs type aufs (rw,relatime,si=c50d032a)
> 
> After this i do a move to tmp location inside /aufs folder. the move is
> successful and again mount(8) or cat /proc/mount returns these
> 
> rootfs on / type rootfs (rw)
> /proc on /proc type proc (rw,relatime)
> tmpfs on /aufs/tmp type tmpfs (rw,relatime)
> /dev/mtdblock2 on /aufs/tmp/jffs2 type jffs2 (rw,relatime)
> /dev/loop0 on /aufs/tmp/sqfsh type squashfs (ro,relatime)
> aufs on /aufs type aufs (rw,relatime,si=c50d032a)
> 
> As u can see the locations have been moved under /aufs. these changes are
> highlighted above. and the /aufs/tmp contains everything which was present
> before in /tmp. then i do a umount /proc.
> And then i do a switch root to /aufs and here comes the problem again. the
> mount(8) or cat /proc/mount in the aufs filesystem returns these
> 
> rootfs / rootfs rw 0 0
> tmpfs /tmp tmpfs rw,relatime 0 0
> /dev/mtdblock2 /tmp/jffs2 jffs2 rw,relatime 0 0
> /dev/loop0 /tmp/sqfsh squashfs ro,relatime 0 0
> aufs / aufs rw,relatime,si=c50d032a 0 0
> /proc /proc proc rw,relatime 0 0
> devpts /dev/pts devpts rw,relatime,mode=622 0 0
> none /tmp ramfs rw,relatime 0 0
> 
> it still seems to have the old mount values and further the mount -o move
> which was performed during the initramfs doesn't seem to be reflected which
> are highlighted again above. and /tmp in the aufs filesystem doesn't
> contain anything. now this is working correct? i'm doubtful.
> 
> Are you sure that your "mount -o move" succeeded?
> When aufs shows its entry for /proc/mounts, the path of branch is
> calculated (generated dynamically).
> And you set the aufs module parmeter 'brs' to 0, right?
> 
> I don't set the brs as I don't have the CONFIG_AUFS_SYSAUFS paramater.
> 
> Thanks and Regards
> R.Vasanth Ragavendran.
> 



------------------------------------------------------------------------------
Systems Optimization Self Assessment
Improve efficiency and utilization of IT resources. Drive out cost and 
improve service delivery. Take 5 minutes to use this Systems Optimization 
Self Assessment. http://www.accelacomm.com/jaw/sdnl/114/51450054/

Reply via email to