Patrik Staehli wrote:
The freezing lockups only happen when the interrupt load of the system is relatively high and the mouse is being moved during that time.
For all of its relative stability, the "usb-ohci" driver did have some issues ... many of which are fixed only in the 2.5 code, though a few less-invasive fixes made it into 2.4.
However, I seem to recall the 2.4 code now has some Geode-specific tweaks, which aren't yet in the 2.5 code ... so while I'll suggest that you try 2.5 on your system, that'll be with the hope that you will do so and port those 2.4 patches forward! The 2.5 code can run pretty easily on 2.4, though I've not done it in quite a while. Just portability tweaks. If that 2.5 code does the job for you ... you'd not be the only person to benefit from an option to use such a backport.
Further investigations (making procfs outputs from usb-ohci) have uncovered that the 'Skip' flag for the frozen usb mouse is set even though there are still transfer descriptors pending.Descriptors, multiple? My initial guess is that you're getting
bit by one of the unlink bugs. High interrupt load means high
latencies ... and some of the bugs fixed would absolutely goof
up "usb-ohci" as latencies climbed, specifically on unlink paths.
(But why were you having multiple descriptors with a mouse?)
- Dave
-------------------------------------------------------
This SF.NET email is sponsored by: Thawte.com
Understand how to protect your customers personal information by implementing
SSL on your Apache Web Server. Click here to get our FREE Thawte Apache Guide: http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0029en
_______________________________________________
[EMAIL PROTECTED]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
