Re: sysfs root link count broken in 2.6.22-git5

[Oliver Pinter]

On 7/18/07, Jean Delvare <khali@xxxxxxxxxxxx> wrote:
> Hi Greg,
>
> On Tue, 17 Jul 2007 20:38:28 -0700, Greg KH wrote:
> > On Tue, Jul 17, 2007 at 11:05:30PM +0200, Jean Delvare wrote:
> > > This breaks libsensors. libsensors uses libsysfs, and libsysfs is not
> > > very smart in that it will initialize successfully even if sysfs is not
> > > mounted.
> >
> > libsysfs isn't smart at all, and isn't even supported anymore.  I'd
> > really suggest droping it entirely, it isn't worth it.
>
> Agreed, except that I do not have the time for this right now. I want
> to get lm-sensors-3.0.0 ready for a release candidate first. What
> really matters for this is to get the API ready. Implementation details
> will come later.
>
> > > So I added tests after the initialization, to make sure that
> > > sysfs is really there. These tests are:
> > > * The mount point exists.
> > > * The mount point is really mounted.
> >
> > Do you know of a 2.6 based distro that does not mount sysfs at /sys?  We
> > took that check out a long time ago in udev and no one has complained :)
>
> I don't know of any 2.6-based distro not mounting sysfs at /sys, but I
> know of 2.4-based distros not mounting sysfs at all ;) libsensors
> supports both 2.4 and 2.6 kernels, so being able to tell whether sysfs
> is mounted or not, matters.

debian has mounted the sysfs

/etc/rcS.d/S02mountkernfs.sh:

#
	# Mount sysfs on /sys
	#
	# Only mount sysfs if it is supported (kernel >= 2.6)
	if grep -E -qs "sysfs\$" /proc/filesystems
	then
		domount sysfs "" /sys sysfs -onodev,noexec,nosuid
	fi

	# Mount /var/run and /var/lock as tmpfs if enabled
	if [ yes = "$RAMRUN" ] ; then
		RUN_OPT=
		[ "${RUN_SIZE:=$TMPFS_SIZE}" ] && RUN_OPT=",size=$RUN_SIZE"
		domount tmpfs "" /var/run varrun -omode=0755,nosuid$RUN_OPT
		touch /var/run/.ramfs
	fi
	if [ yes = "$RAMLOCK" ] ; then
		LOCK_OPT=
		[ "${LOCK_SIZE:=$TMPFS_SIZE}" ] && LOCK_OPT=",size=$LOCK_SIZE"
		domount tmpfs "" /var/lock varlock -omode=1777,nodev,noexec,nosuid$LOCK_OPT
		touch /var/lock/.ramfs
	fi


> > > The code looks like:
> > >
> > >        if (sysfs_get_mnt_path(sensors_sysfs_mount, NAME_MAX)
> > >          || stat(sensors_sysfs_mount, &statbuf) < 0
> > >          || statbuf.st_nlink <= 2)      /* Empty directory */
> > >                 return 0;           /* Failure */
> > >
> > > This works OK with 2.6.22.1, but the last test fails with the current
> > > git kernel even when sysfs is mounted.
> >
> > Yeah, but is checking the number of hard links in the directory a safe
> > way to always verify that it isn't empty?
>
> I think so, yes. To the best of my knowledge, it has worked on all
> Unix-like systems for decades. There are other ways, but this is by far
> the less expensive.
>
> >                                            Isn't there some glibc
> > function that can detect the mount point of a filesystem or directory?
> > Something in glibc parses /proc/mounts for something, I can't remember
> > what it is right now though, sorry.
>
> Maybe getmntent(3)? Sure I could use this, but how expensive compared
> to a single stat(2).
>
> > Again, I recommend dropping libsysfs, it's gone from some distros
> > already :)
>
> Really? I'm curious how such distributions support libsensors and the
> other tools which still rely on libsysfs. If they have already
> converted libsensors for me, that would be good news :)
>
> > And yes, the bug should be fixed, I agree.  Thanks for letting us know.
>
> Tejun already fixed it, that was quick :)
>
> --
> Jean Delvare
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/