Bug 15421

Message in kernel 2.6.31-14-generic

[  278.769885] sd 6:0:0:0: [sdc] Unhandled sense code
[  278.769890] sd 6:0:0:0: [sdc] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[  278.769894] sd 6:0:0:0: [sdc] Sense Key : Medium Error [current] 
[  278.769898] sd 6:0:0:0: [sdc] Add. Sense: Unrecovered read error
[  278.769902] end_request: I/O error, dev sdc, sector 6312799
[  278.769907] __ratelimit: 8 callbacks suppressed
[  278.769910] Buffer I/O error on device sdc1, logical block 789092
[  278.769914] Buffer I/O error on device sdc1, logical block 789093
[  278.769935] Buffer I/O error on device sdc1, logical block 789094
[  278.769938] Buffer I/O error on device sdc1, logical block 789095
[  278.769941] Buffer I/O error on device sdc1, logical block 789096
[  278.769943] Buffer I/O error on device sdc1, logical block 789097
[  278.769946] Buffer I/O error on device sdc1, logical block 789098
[  278.769949] Buffer I/O error on device sdc1, logical block 789099
[  278.769952] Buffer I/O error on device sdc1, logical block 789100
[  278.769954] Buffer I/O error on device sdc1, logical block 789101

Created attachment 25295 [details]
/2.6.31-14-generic_dmesg.txt

Please do not file Ubuntu kernel issues here, file them in the ubuntu bug database.

If you can duplicate this on a kernel.org kernel, like 2.6.34-rc2 right now,
then we can work on it.

Can you?

Not work for my kernel 2.6.34-rc2, testing with 2.6.33.1?

Not work for my kernel 2.6.34-rc2, testing with 2.6.33.1?

See also my bugreport here containing links to the problem being discussed on linux-usb a while ago: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/515023

I can also confirm this bug with a very recent (git-checkout from today) kernel from Linus' tree (post .34-rc2)

New release - new test:
mainline:  	2.6.34-rc3  	2010-03-30

Created attachment 25777 [details]
dmesg_2.6.34-rc2.26032010.txt

Created attachment 25778 [details]
config-2.6.34-rc2.26032010

*** Bug 15081 has been marked as a duplicate of this bug. ***

Created attachment 25785 [details]
boot_dmesg-2.6.34-rc3

Created attachment 25786 [details]
current_dmesg-2.6.34-rc3

Created attachment 25787 [details]
config-2.6.34-rc3-31032010

*** Bug 15236 has been marked as a duplicate of this bug. ***

Neither of the two dmesg logs in comments #8 and #12 contains anything about an I/O error.  Each one does contain a failed read of sector 63, but in both cases the read was retried successfully.  Is this because the dmesg logs don't cover the time when you tried to mount the drive?

Please collect a usbmon trace showing what happens when the I/O error occurs.  Instructions are in Documentation/usb/usbmon.txt.  The output from usbmon is a lot more compact and more informative than the usb-storage debugging messages.

Also, note that this most likely is not a bug in the kernel.  Instead it is probably caused by a bug in the Jmicron converter (those things are notoriously buggy) combined with a user program issuing commands the converter can't handle.

Alan, we already discussed this some time back on linux-usb (topic was "strange problem regarding non-mounting usb hard drive). The date was the last days of January. You found out that something was sending two Ata passhrough commands with the second one being problematic. You also mentioned then it was unlikely to be a kernel bug.
I then took the issue tu Ubuntu's launchpad where it bounced back to the kernel. They released a fix for a specific imx51-kernel (don't have the patch at hand but I'll look it up) so maybe you could take a look at this bugreport here and decide whether this could be apply to mainline as well: https://bugs.launchpad.net/ubuntu/+source/linux-fsl-imx51/+bug/499881

Created attachment 25818 [details]
Patch for imx51 from Ubuntu

I extracted the patch from this original diff where the copyright information can also be found: https://launchpad.net/ubuntu/+archive/primary/+files/linux-fsl-imx51_2.6.31-607.12.diff.gz

Created attachment 25819 [details]
original usbmon trace

You might want to read through this thread:

    http://marc.info/?t=126927150300011&r=1&w=2

It's essentially the same problem as you are having.  It turns out the offending command is sent by the hdparm program, as invoked automatically by udev through

    /lib/udev/rules.d/85-hdparm.rules

If you have this file, and you rename it or move it somewhere else so that udev doesn't see it, you may find the problem has gone away.

In the end, something like the Ubuntu patch may be needed.  I don't like the way they wrote it, however.  Still, if there's no other choice then we'll add something similar.

Thanks, Alan, renaming that file is a workaround - at least with the mainline kernel I compiled for myself. I remember trying this some time ago when I was on another kernel (it was either Ubuntu's -32 version or one the Ubuntu -33 mainline build) and back then it didn't have any effect.
Ho-hum, I'll go file a bug against udev in Ubuntu now, thanks Alan!

Sorry, I mistook that file for a different one from udev someone suggested renaming and which of course didn't change anything - the issue is kernel-independent...