Kernel Bug Tracker – Bug 2727
CIFS Kernel error message
Last modified: 2005-03-14 07:05:46 UTC
Distribution: Fedora Core 2
Hardware Environment: Dell 1750 w Perc/DI
Software Environment: duplicity (http://www.nongnu.org/duplicity/)
I'm using duplicity to backup a cifs mount locally (remote server: Dell NAS 745
using Windows 2003 Appliance). After 4GB data transferred (over 100GB) i noticed
six occurences of this missage in dmesg:
Badness in sleep_on_timeout at kernel/sched.c:2528
[<42ac9f97>] SendReceive+0x1c1/0x336 [cifs]
[<42abc155>] CIFSFindFirst+0x1f0/0x278 [cifs]
[<42ac5d4b>] cifs_readdir+0x27d/0xa60 [cifs]
Steps to reproduce:
-mount a cifs share over windows 2003 appliance
-do duplicity-like file access (non linear reads, it reads an amount of data,
compress/sign it, writes it locally, then read more data, etc...)
sleep_on_timeout is a deprecated (and broken) API that shouldn't be used in new
code; however this seems to have been added recently to CIFS.....
The warning in question is that sleep_on_timeout() is called without the BKL
helt, which is invalid usage of that API.
I had not noticed this defect - will presumably simply need to change to
wait_event_interruptible with timeout
I looked around and did not see a replacement for this function in
include/linux/wait.h (uninterruptible wait on event with timeout). Is the
best available way to remove this function to replace with schedule_timeout
inside a loop checking if the event is satisfied?
Fixed long ago. Probably in the 2.6.9 merge.