> Yes, you've found a bug. Try the attached patch if it fixes your problem
> and doesn't introduce any other side effect - it shouldn't, but I didn't
> have enough time to test it.
This patch solves part of the problem -- I can now hot-plug the mouse while
the system is running and things keep working.
However, the USB subsystem doesn't recover from a system suspend, and there
appear to be some bugs in the usb-uchi module.
When the system resumes after a suspend, the kernel starts reporting:
usb_control/bulk_msg: timeout
usb.c: USB device not responding, giving up (error = -110)
Attempting to remove the mousedev module at this point fails because gpm
still has /dev/usb/mouse open. Killing gpm frees up mousedev for
unloading.
The timeout message are coming from the usb-uhci module -- I still see them
even if I unload mousedev, hid, and input. When I unload usb-uhci, I get
the following errors:
kmem_shrink: Invalid cache addr c7fefb00
kmem_destroy: Can't free all objects c7fefb00
usb-uhci.c: uhci_desc_kmem remained
kmem_shrink: Invalid cache addr c7fefb60
kmem_destroy: Can't free all objects c7fefb60
usb-uhci.c: urb_priv_kmem remained
When I reload the module, I get:
kmem_create: Dup name - uhci_desc
kmem_create: Dup name - urb_priv
I've tried to work around this with some apm script that perform the
following actions on a suspend:
* stop gpm
* unload USB modules
And then on a resume:
* reload USB modules
* start gpm
This seems to work -- with the exception of the errors noted above -- but
seems particularly fragile. Originally I had this all in a single script
that was run when the system resumed, and this caused an "unable to handle
kernel paging request" error along with the following call trace:
Call Trace: [usb_internal_control_msg+94/112] [usb_control_msg+116/152]
[usb_get_descriptor+152/184] [usb_new_device+59/424] [RCSid+23488/24628]
[usb_hub_port_connect_change+634/664] [usb_hub_events+149/500]
[clas_info+1395/4332] [NR_TYPES+457/4040] [usb_hub_thread+57/92]
[get_options+0/112] [kernel_thread+35/48]
-- Lars
--
Lars Kellogg-Stedman <[EMAIL PROTECTED]> --> http://www.larsshack.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
