I am running a patched kernel 2.4.21 with acpi and swsusp ver 1.0.3
with usb compiled in as a module.
When kernel loads for the first time everything works fine, all usb
modules can be loaded and unloaded properly. After suspending and
restarting, the computer comes up fine and everything works. The
problem is that at this point, when I try to unload the usbcore module
it gets to the point of calling usb_hub_cleanup in drivers/usb/hub.c.
At this points it tried to kill khubd with killproc, which works fine
(the process is stoped), with a return value of 0. The problem is that
at this point the function locks on the call
wait_for_completion(&khubd_exited); which never returns, and rmmod gets
locked. I tried changing the DECLARE_COMPLETION call so that it will be
redone each time the module starts but it didn't solve the problem. Any
ideas on how to further persue this or whether there is a known
solution?
Thanx
--
Micha Feigin
[email protected]
On Wed, Aug 06, 2003 at 10:42:06PM +0300, Micha Feigin wrote:
> I am running a patched kernel 2.4.21 with acpi and swsusp ver 1.0.3
> with usb compiled in as a module.
>
> When kernel loads for the first time everything works fine, all usb
> modules can be loaded and unloaded properly. After suspending and
> restarting, the computer comes up fine and everything works. The
> problem is that at this point, when I try to unload the usbcore module
> it gets to the point of calling usb_hub_cleanup in drivers/usb/hub.c.
> At this points it tried to kill khubd with killproc, which works fine
> (the process is stoped), with a return value of 0. The problem is that
> at this point the function locks on the call
> wait_for_completion(&khubd_exited); which never returns, and rmmod gets
> locked. I tried changing the DECLARE_COMPLETION call so that it will be
> redone each time the module starts but it didn't solve the problem. Any
> ideas on how to further persue this or whether there is a known
> solution?
Unload the usb modules before suspending.
Good luck,
greg k-h
On Thu, 2003-08-07 at 00:23, Greg KH wrote:
> On Wed, Aug 06, 2003 at 10:42:06PM +0300, Micha Feigin wrote:
> > I am running a patched kernel 2.4.21 with acpi and swsusp ver 1.0.3
> > with usb compiled in as a module.
> >
> > When kernel loads for the first time everything works fine, all usb
> > modules can be loaded and unloaded properly. After suspending and
> > restarting, the computer comes up fine and everything works. The
> > problem is that at this point, when I try to unload the usbcore module
> > it gets to the point of calling usb_hub_cleanup in drivers/usb/hub.c.
> > At this points it tried to kill khubd with killproc, which works fine
> > (the process is stoped), with a return value of 0. The problem is that
> > at this point the function locks on the call
> > wait_for_completion(&khubd_exited); which never returns, and rmmod gets
> > locked. I tried changing the DECLARE_COMPLETION call so that it will be
> > redone each time the module starts but it didn't solve the problem. Any
> > ideas on how to further persue this or whether there is a known
> > solution?
>
> Unload the usb modules before suspending.
>
> Good luck,
>
> greg k-h
Guess I wasn't clear enough. I unload all the modules (except for the fw
ones actually since for some reason they refuse to do so).
This includes the usb modules, before I suspend.
There some more info on the subject though. Apparently this happens only
when calling the hibernation script (from the swsusp page on
sourceforge) from acpid in response to the sleep button. When called
from the command line I don't see this behavior.
--
Micha Feigin
[email protected]