On Wednesday 19 March 2003 15:36, David Brownell wrote: > Duncan Sands wrote: > > From time to time when booting 2.4.21-pre5, the > > hotplug script for my usb modem seems to be run > > three times. If I then try to restart the hotplug > > subsystem, I systematically get an oops: > > > > ehci-hcd 00:10.3: remove state 3 > > usb.c: USB disconnect on device 00:10.3-0 address 1 > > usb.c: USB bus 1 deregistered > > usb.c: USB disconnect on device 00:0a.0-0 address 1 > > usb.c: USB bus 2 deregistered > > usb.c: USB disconnect on device 00:10.2-0 address 1 > > usb.c: USB bus 3 deregistered > > usb.c: USB disconnect on device 00:10.1-0 address 1 > > usb.c: USB disconnect on device 00:10.1-2 address 2 > > usb.c: USB bus 4 deregistered > > usb.c: USB disconnect on device 00:10.0-0 address 1 > > usb.c: USB bus 5 deregistered > > So five root hubs got removed through the "rmmod" sequence. > One of them had a device connected to it at that time. > > That one, 00:10.1, would seem to be a UHCI controller, > given that 00:10.3 is a VT8235 (VIA), and that we heard > > from the UHCI controller: > > kmem_cache_destroy: Can't free all objects c158ea08 > > uhci: not all urb_priv's were freed > > Not clear why that didn't shut down completely; maybe > Johannes knows. I could believe it was related to the > way your device enumerated three times ... could be that > something didn't get cleaned up properly. > > The "oops" below is because the old slab cache didn't get > destroyed, so that a new one (with the same name) couldn't > get created. (Me, I'd rather see kmem_cache_create just > fail rather than BUG out...) > > > Nothing here seems to implicate the EHCI code.
Hi Dave, thanks for your comments. I didn't realize that kmem_cache_create would BUG, I assumed the USB code was failing to deal with the error condition. By the way, I wanted to thank you for your clear explanation of the possibly problematic interaction of DMA and CPU writes due to cache flushing. I'm afraid your email got eaten by a massive system failure here - but fortunately I archived it in a few neurons first. Ciao, Duncan. ------------------------------------------------------- This SF.net email is sponsored by: Does your code think in ink? You could win a Tablet PC. Get a free Tablet PC hat just for playing. What are you waiting for? http://ads.sourceforge.net/cgi-bin/redirect.pl?micr5043en _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
