/...a few hundreds reboots later.../
No, it is not this patch. It is something else. And whatever it is, it
is already present in 2.6.25-rc3.
To reproduce it, I need to plug an SBP-2 device in and out, then shut
the machine down (e.g. shutdown -h now, whereas shutdown -r now does not
seem to trigger the bug).
Since it happens after all filesystems were unmounted or r/o-mounted, I
can't capture the log output easily (perhaps with a netconsole or so)
but it also can't do any damage at that point anymore.
It does not happen with ohci1394 + sbp2, so it is obviously located in
the firewire stack.
I am now gradually removing debug options from the kernel to see which
debug facility is making the fuzz...
--
Stefan Richter
-=====-==--- --== --=-=
http://arcgraph.de/sr/
--