usb stick not mounted after hibernate

[Ian Jackson]

Binary package hint: hal

Using a hard disk install made from the i386 livecd 20070411:

I inserted a USB stick containing a few jpgs. This worked fine and produced an appropriate nautilus window. I left this window open and did some other unrelated tests. After I'd finished the other tests I asked the machine to hibernate (with the dialogue from the top-right corner power icon) - still with the nautilus window open and the usb stick inserted.

When the machine had hibernated, I restarted it (with the usb stick still inserted). On restart I notice that the nautilus window for the usb stick is not present; a terminal window shows me that it isn't mounted. However it does show up in /proc/scsi/scsi and I can say "cfdisk /dev/sdb" successfully.

Removing and reinserting the stick makes it work.

[Ian Jackson (ijackson)]

Ian Jackson writes ("[Bug 105563] usb stick not mounted after hibernate"):
> When the machine had hibernated, I restarted it (with the usb stick
> still inserted). On restart I notice that the nautilus window for the
> usb stick is not present; a terminal window shows me that it isn't
> mounted. However it does show up in /proc/scsi/scsi and I can say
> "cfdisk /dev/sdb" successfully.

On another similar install on a different machine, when I resumed from
hibernation I got an "unsafe device removal" warning bubble.

Ian.

[hendrixski (hendrixski)]

CONFIRMED! I have the same problem on my Toshiba Portégé, running ubuntu feisty. This is REALLY annoying on a laptop! (well, tablet PC really, but it's a laptop). In order to put data on, or pull data off of a USB I have to reboot! This is unacceptable.

Is this connected to other things not waking up after hibernate? Like tuner cards, which makes mythtv users pull their hair out!

 - hendrixski

[hendrixski (hendrixski)]

attached are the outputs of lspci -vvnn > lspci-vvnn.log to show my system specs
uname -a is 2.6.20-16-lowlatency #2 SMP PREEMPT Thu Jun 7 20:23:03 UTC 2007 i686 GNU/Linux

[hendrixski (hendrixski)]

oops, should have put this in as well
After I rebooted the laptop I ran dmesg -c to clean the log, then inserted the USB stick (as can be seen in the log) and it worked, I removed it properly, then I hibernated and tried again but this time NOTHING. the output of dmesg>dmesg.log is also attached.

I hope that helps :-) Please pleaseplease please get this taken care of.

[Christian Schlauer (cs-usenet)]

I can confirm Ian Jackson's bug and his comment (comment #1 above). As I understand it, the problem is in the kernel. I found this:

,----[ <http://lwn.net/Articles/241465/> ]
| The highly experimental "USB persist" feature attempts to maintain
| the state of USB devices when they lose power. The driving
| motivation between this patch is to be able to suspend a system
| containing filesystems on USB storage and still have those
| filesystems mounted and working at resume time.
`----

So this will be ("highly experimental") in kernel 2.6.23.

[Peter Cordes (peter-cordes)]

On Hardy (2.6.24), there are still problems. CONFIG_USB_PERSIST=y, in /boot/config-2.6.24-16-generic.

 I don't have time right now to test a lot of stuff, for one thing because my Toshiba A70 is really slow to boot. (kernel seems to stick for a while before before detecting hard drives...)

 I hibernated with a Patriot brand USB stick mounted (by gnome/nautilus), and when I came back there seemed to be two mount points. /media/Patriot on /dev/sda1, and /media/Patriot_ on /dev/sdb1. I didn't notice until I'd unmounted one of them, but I think the kernel detected the USB stick again, instead of realizing that it was the same one present when it hibernated. I did remove and reinsert the USB stick, after a while, so that's in the kernel log, too.

 My kern.log shows this. Note the sda insertion on May 7, 23:50, then the hibernate. After the resume on May 8, 14:15, the USB stick is detected as sdb. There are some FAT directory read errors, which makes it look like the kernel really did get something crossed up. This might have been the original mount which became stale after resume from hibernation? I don't know if there ever was a /dev/sda1 block dev after the resume.

 Eventually just umount and pumount, and clicking on unmount in nautilus, got rid of /media/Patriot*, and the mount entries, so I didn't have to hack mtab or anything.

[dreamerns (dreamerns-ns-linux)]

Does this problem still persist?

[Jonathan Thomas (echidnaman)]

We are closing this bug report because it lacks the information we need to investigate the problem, as requested in the previous comments. Please reopen it if you can give us the missing information, and don't hesitate to submit bug reports in the future. To reopen the bug report you can click on the current status, under the "Status" column, and change the status back to "New". Thanks again!