On Wed, Mar 26, 2003 at 11:34:37AM -0800, Greg KH wrote:
> > Anyway I'll test out a 2.5 kernel when I'm back in the office
> > tomorrow, I can devote some time to tracking down the problem if you
> > can give me some pointers on where to start. I'd like to be able to
> > feel confident that this will work reliably under 2.4, otherwise I
> > guess I need to look for alternate solutions.
>
> The problem is in the race on close() in the usb-serial.c code. In 2.5
> that logic has been rewritten to (hopefully) get rid of the race. That
> is what will need to be backported, once people test that this fixes the
> issue.
OK. 2.5.66 compiled and booted.
I've jumpered the hardware back to how it was originally when I
experienced the problem.
I've been working happily for about 10 mins with:
while /bin/true; do
for i in *; do
cat $i >/dev/ttyUSB0
done
done
No Oopsen or errors in dmesg as yet. (Before I was getting many errors
about 0 size writes).
I can keep working under 2.5.66 for now to see if I experience any
problems, but it would appear that the race is gone in 2.5.66
(CONFIG_PREEMPT=y)
If you'd like me to try any patches against 2.4 just let me know.
Thanks again,
--(o- Chris Sykes -- GPG Key: http://www.sigsegv.plus.com/key.txt //\ "Don't worry. Everything is getting nicely out of control ..." V_/_ Douglas Adams - The Salmon of Doubt
This archive was generated by hypermail 2b29 : Mon Mar 31 2003 - 22:00:27 EST