Bug 5244 - Psplash fails to unmount /mnt/.psplash
Summary: Psplash fails to unmount /mnt/.psplash
Status: RESOLVED FIXED
Alias: None
Product: System Startup
Classification: Runtime
Component: system-startup (show other bugs)
Version: 1.5
Hardware: Other arm
: Medium normal
Target Milestone: 1.5.1
Assignee: Saul Wold
QA Contact:
URL:
Whiteboard:
Depends on:
Blocks:
 
Reported: 2013-09-20 09:56 UTC by Diego
Modified: 2013-11-18 22:16 UTC (History)
3 users (show)

See Also:
OS type for building Yocto: ---
Type of Regression: ---
Verified:
Documentation change: No (bug/feature does not impact docs)

Attachments
Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description Diego 2013-09-20 09:56:37 UTC 
I'm on a core-image-base image on a i.MX6Q Boundary Devices Nitrogen6x.

One of the last things I get at boot, just before login, is:

umount: /mnt/.psplash: target is busy.
        (In some cases useful info about processes that use            
         the device is found by lsof(8) or fuser(1))

and the same at shutdown:
Disabling non-boot CPUs ...
CPU1: shutdown
CPU2: shutdown
CPU3: shutdown
Power down.
umount: /mnt/.psplash: target is busy.
        (In some cases useful info about processes that use
         the device is found by lsof(8) or fuser(1))

If I try to "umount /mnt/.psplash" after login I can unmount it correctly, so it's probably just trying to umount it too early.
Comment 1 Saul Wold 2013-09-24 21:58:46 UTC 
Is this reproducible after the first boot? I was able to reproduce the first 
failure during boot once, but not on subsequent boot ups.

Can you add an "fuser /mnt/.psplash to your /etc/init.d/rc I think there is a 
race condition and the psplash is not finishing before
the umount occurs.
Comment 2 Diego 2013-09-25 08:03:59 UTC 
(In reply to comment #1)
> Is this reproducible after the first boot? I was able to reproduce the first 
> failure during boot once, but not on subsequent boot ups.
> 
> Can you add an "fuser /mnt/.psplash to your /etc/init.d/rc I think there is
> a 
> race condition and the psplash is not finishing before
> the umount occurs.

Yeah, it's just a matter of istants, because I've added hacky line to rc:
        fuser /mnt/.psplash
        cat /proc/$(fuser /mnt/.psplash | cut -d" " -f1)/comm
and sometimes I get:
73 
psplash
while sometimes I get:
73 
cat: /proc/73/comm: No such file or directory
and then I get no error from umount.

So yeah, umount should wait for psplash to finish, not just hope it finished.
Comment 3 Saul Wold 2013-10-15 22:30:24 UTC 
Since you can get consistent failure, can you try to following.

Since we know psplash is exiting, it's safe to umount regardless of the state 
of the process, so I wonder if the -l (lazy option) would help and if not that, 
the -f (force), again just edit the /etc/init.d/rc file and add that option to 
the umount and reboot. 

Please report the results here.
Comment 4 Diego 2013-10-28 09:52:38 UTC 
(In reply to comment #3)
> Since you can get consistent failure, can you try to following.
> 
> Since we know psplash is exiting, it's safe to umount regardless of the
> state 
> of the process, so I wonder if the -l (lazy option) would help and if not
> that, 
> the -f (force), again just edit the /etc/init.d/rc file and add that option
> to 
> the umount and reboot. 
> 
> Please report the results here.

Adding the "-l" option to umount seems to fix the problem. Tried 10 reboots and I did not encounter the error anymore.