Opened 11 years ago

Last modified 8 years ago

#9062 assigned enhancement

Panic triggered by mass storage device unplug

Reported by: przemub Owned by: nobody
Priority: normal Milestone: R1
Component: System/Kernel Version: R1/Development
Keywords: panic kernel fat unplug Cc:
Blocked By: Blocking:
Platform: All

Description

hrev44697, x86, gcc4h.

When I unplugged my Sony Ericsson X8 phone (FAT32-formatted) without unmount, I quickly fetched up in Kernel Debugging Land.

Interestingly, when I did similar with my digital camera, it did not occur.

I was able to continue with entering co twice.

Attachments (1)

syslog (156.8 KB ) - added by przemub 11 years ago.

Download all attachments as: .zip

Change History (5)

by przemub, 11 years ago

Attachment: syslog added

comment:1 by luroh, 11 years ago

Component: File Systems/FATDrivers/USB
Owner: changed from nobody to mmlr
Status: newassigned

comment:2 by mmlr, 11 years ago

Component: Drivers/USBSystem/Kernel
Owner: changed from mmlr to nobody
Type: bugenhancement

This is not a bug. The KDL indicates that you just lost data. I maintain that it is a user error to unplug a device with mounted volumes. For later production releases the panic should be removed and an appropriate error message should be shown to the user instead. This is neither USB nor filesystem related though, and it concerns other types of problems as well (i.e. an actual device failure).

in reply to:  2 comment:3 by przemub, 11 years ago

Replying to mmlr:

For later production releases the panic should be removed and an appropriate error message should be shown to the user instead.

This is exactly what I'd love to. Not everyone must know there is no need to restart computer in this case.

Although, panic may discourage users from brutal treatment of their drives for life :].

Last edited 11 years ago by przemub (previous) (diff)

comment:4 by pulkomandy, 8 years ago

That KDL is now gone, but still we can get into problems as there is no warning message at all, and the drive is not unmounted.

Ideally, the user should be requested to re-plug the volume, then the writing should resume.

Note: See TracTickets for help on using tickets.