New: kernel hangs when trying to read atip wiith cdrecord
You did write:
Quoted 7 lines
Quoted 7 lines
Executing 'read buffer' command on Bus 1 Target 0, Lun 0 timeout 40s
CDB: 3C 00 00 00 00 00 00 FC 00 00So the kernel doesn't timeout/return for some reason. You can find the full log here:
http://muenchen-surf.de/lohmaier/cdrecord_atip_log
Quoted 8 lines
Quoted 3 lines
If that information is not enough to handle the process, please tell me what else you need. Thank you.
We seem to have an awful lot of these "CD burner doesn't work but it did in 2.4" reports.
Does anyone have the vaguest inklink of how we broke it?
There are various reasons....
One important problem of course is the fact that various distributors (e.g. RedHat, Suse, Debian, Ubuntu) are ungrading cdrecord slower than the Linux kernel changes it's interfaces. Related to is problem: the Linux Kernel folks do not inform me before breaking Linux kernel interfaces, so I am informed too late in order to add a workaround to cdrecord in time.
The most recent interface changes have been:
- Instead of fixing the bug that allowed anyone to send arbitrary SCSI
commands to CD/DVD-recorders (caused by not requiring write permission on the node), the behavior of SCSI generic was changed (in 2.6.8.1).
Old behavior: open device as root and send commands as user. New behavior: open device as user and send commands as root.
- Some time between January 2006 and 2004, a new rlimit "RLIMIT_NOFILE"
was added and implemented in a way that is not compatible to previous kernel behavior.
Old behavior: mlockall(MCL_CURRENT|MCL_FUTURE) done as root was honored
for the same process if later continued as user.
New behavior: mlockall(MCL_CURRENT|MCL_FUTURE) done as root is no longer
honored for the same process if later continued as user.
Another problem is caused by the fact that various distributors (e.g. RedHat, Suse, Debian, Ubuntu) are applying patches to cdrecord that break cdrecord.
The problem mentioned in this thread seems to be caused by the fact that Linux sometimes ignores timeouts. I have no idea how to help in this (timeout) case.
Jörg
Signature · 4 lines
EMail:[email protected] (home) Jörg Schilling D-13353 Berlin [email protected] (uni) [email protected] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
Quoted 3 lines · On Sun, 25 Jun 2006 12:53:55 +0200 Joerg Schilling…
The problem mentioned in this thread seems to be caused by the fact that Linux sometimes ignores timeouts. I have no idea how to help in this (timeout) case.
OK, thanks. So is that likely to be an IDE bug, or a SCSI bug or an IDE-CD bug?
Quoted 9 lines · Andrew Morton
Quoted 3 lines · On Sun, 25 Jun 2006 12:53:55 +0200 Joerg Schilling…
The problem mentioned in this thread seems to be caused by the fact that Linux sometimes ignores timeouts. I have no idea how to help in this (timeout) case.
OK, thanks. So is that likely to be an IDE bug, or a SCSI bug or an IDE-CD bug?
I am not sure if the log from the OP includes all information.
I've seen already messages like this:
cdrecord: Input/output error. Cannot set SG_SET_TIMEOUT.
This should be something that I would never to expect to happen.
If the OP does not see a similar message, it seems that the call is accepted but later ignored. I have no idea why this happens.
Jörg
Signature · 4 lines
EMail:[email protected] (home) Jörg Schilling D-13353 Berlin [email protected] (uni) [email protected] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily
Quoted 10 lines · Joerg Schilling
I am not sure if the log from the OP includes all information.
I've seen already messages like this:
cdrecord: Input/output error. Cannot set
SG_SET_TIMEOUT.This should be something that I would never to expect to happen.
If the OP does not see a similar message, it seems that the call is accepted but later ignored. I have no idea why this happens.
Let me add some notes:
Some drives do not support (or correctly support) the 'read buffer' command. As I don't have all drives, I cannot tell how the drive from the OP behaves on Solaris. All drives I've seen so far return an error on Solaris - not a timeout, so it is not clear, whether the hang on Linux would result in a timeout on Solaris.
It may be that the /dev/hd* interface ignores timeouts, but it may also be that the bus situation did cause a kernel hang. As my experience shows that all drives behave always the same, it should be simple to trace the problem for the owner of a drive.
Jörg
Signature · 4 lines
EMail:[email protected] (home) Jörg Schilling D-13353 Berlin [email protected] (uni) [email protected] (work) Blog: http://schily.blogspot.com/ URL: http://cdrecord.berlios.de/old/private/ ftp://ftp.berlios.de/pub/schily