> Alot of the mount options just don't get processed on remount. I
> complained about this back when noatime got into the 2.0.x source.
Someone posted a one-liner fix to invalidate inodes upon
umount/remount/whatever. Any chance of this making 2.0.34 Alan?
> > Also, we _always_ seem to get oopses when after a few days uptime I stop
> > the networking, rmmod 3c59x then insmod and start the networking. Latest
> > single oops available upon request.
>
> Are you sure you stopped _all_ programs that were using the network? I've
God no. There were several ftpd's on the go, at least. The actual oops was
an in.ftpd process. I'm sorry, if ifconfig'ing down a device and/or
rmmod'ing the device driver leaves dangling pointers associated with
network packets and/or connections, it should be considered a kernel bug.
I don't mind if the packets are dropped, rerouted, connections forcibly
closed, whatever, but an oops shouldn't occur.
Besides when networking is going south, it's useful to ifconfig the device
down and up without the hassle of monotonous checking for connections,
open sockets etc. Indeed it is fun seing a connection survive the
ditching and reinstating of an interface.
Cheers
Chris
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu