xmon & SCSI ATA devices
Hello,
xmon and SCSI SATA device driver were installed on my system. When I invoked xmon explicitly for kernel debugging, there're probably pending SCSI requests issued. So those SCSI requests complained timeout when I quited from xmon. I want to find a way to suspend SCSI device before invoking xmon and resume that before quiting from xmon. Anybody knew there is a way to do this?
Thanks,
Gavin
Quoted 9 lines · gshan
Hello,
xmon and SCSI SATA device driver were installed on my system. When I invoked xmon explicitly for kernel debugging, there're probably pending SCSI requests issued.
So those SCSI requests complained timeout when I quited from xmon. I want to find a way to suspend SCSI device before invoking xmon and resume that before
quiting from xmon. Anybody knew there is a way to do this?
Well, it's non trivial. xmon is very low level and doesn't muck around with drivers etc...
We could add hacks to avoid those timeouts or even do what you suggest with suspending devices, but that would make entering xmon a -lot- more fragile. The idea is that xmon relies on very little kernel services and can be entered even when things are utterly wrong.
To be honest, I'm tempted to leave that as it is. Most of the time, getting into xmon is a one way trip....
Cheers,
Ben.
Quoted 6 lines
Thanks,
Gavin
_______________________________________________
Linuxppc-dev mailing list
[email protected]
https://lists.ozlabs.org/listinfo/linuxppc-dev
Benjamin Herrenschmidt wrote: On Thu, 2009-12-24 at 02:34 -0500, gshan wrote: Hello, xmon and SCSI SATA device driver were installed on my system. When I invoked xmon explicitly for kernel debugging, there're probably pending SCSI requests issued. So those SCSI requests complained timeout when I quited from xmon. I want to find a way to suspend SCSI device before invoking xmon and resume that before quiting from xmon. Anybody knew there is a way to do this?
Quoted 4 lines · gshan
Actually, the situation I described above has caused lots of disk issue. The DPT (disk partition table) might be lost because of this. I think it's broken the rule: xmon is assisting to resolve kernel issue, not cause more.
Well, mostly xmon kicks in when you already oopsed or crashed so .. :-)
As I said, getting into the SCSI stack or similar would make the whole thing extremely fragile. But you may want to hack something for yourself, feel free to do so.
Cheers,
Ben.
Benjamin Herrenschmidt wrote: On Tue, 2009-12-29 at 23:09 -0500, gshan wrote: Actually, the situation I described above has caused lots of disk issue. The DPT (disk partition table) might be lost because of this. I think it's broken the rule: xmon is assisting to resolve kernel issue, not cause more.