[lug] Mount process in uninterruptible sleep

[Riggs, Rob]

Your machine needs to be rebooted. Linux has had similar mount problems in
the kernel all the way back to 1.x (and likely since inception). There are a
number of things that can cause this. My guess would be that the vfat
filesystem choked on the data in your disk image. (Are you sure it's vfat
rather than msdos?)

If this hang can be repeated on the most recent kernel in the series you are
using (2.2 or 2.4), then it would be a good idea to post a message to the
linux-kernel list describing the problem, how to reproduce it, and offer
them access to the data file.

The safest thing to do at this point is to go into single user mode (telinit
1), kill any remaining processes (except init, your shell, the errant mount
command, etc.), then unmount, or remount read-only, all of the filesystems
that you can. You can then try a clean shutdown or reboot, but my bet is
that it will hang and you'll need to press the big red button.

Or you could kill the shell, leave the mount process in its current state
and continue on as if life were normal. But by the time you do get around to
rebooting your machine (months from now, no doubt), you will have forgotten
about the stray mount process and have to hit the big red button with your
disks mounted read/write. fsck, fsck, fsck.

-----Original Message-----
From: Chris Riddoch [mailto:socket at peakpeak.com]
Sent: Tuesday, July 17, 2001 11:15 PM
To: lug at lug.boulder.co.us
Subject: [lug] Mount process in uninterruptible sleep


Hi, everyone.

This should be simple, for somebody...

I used 'dd' to copy a vfat-formatted floppy to a file, disk.img.  Then
I made a 'mnt' directory in the current directory.  And then:

socket at magrathea:~/old-disks$ sudo mount -t vfat -o ro,loop disk.img mnt/

It hung.  It was about 2:00 AM last night when this happened, and
decided there wouldn't be any harm in leaving it until later to fix.
Too tired to think.

And now, I've just discovered that terminal under a pile of others...

socket at magrathea:~$ ps 10105
  PID TTY      STAT   TIME COMMAND
10105 pts/2    D      0:00 mount -t vfat -o ro,loop disk.img mnt/
socket at magrathea:~$

This mount process is unkillable - even by root, even by -9.  I'm
tempted to just try to kill its parent terminal, but I'm wary of
potentially leaving a random mount process lying about.  That 'D' in
the stat column, according to the manpage, means it's blocked in
uninterruptible sleep (usually waiting for IO) Well, it's been waiting
an awfully long time on a highly responsive hard drive, so I suspect
it's not IO.

What's going on, anyway?

--
Chris Riddoch         |  epistemological  | Send spam to: uce at ftc.gov
socket at peakpeak.com   |  humility         | 
_______________________________________________
Web Page:  http://lug.boulder.co.us
Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug