3com OfficeConnect 11g doesn't work after waking up (prism54)

[Kevin Freeman]

On Sun, 2004-11-14 at 21:32 +0000, Simon Michelson wrote:
> I've just installed FC3 on my IBM ThinkPad T21.  On the whole the
> system is working well.  However, there is one problem which I'd really
> like to get fixed: when the system wakes up, the wireless card doesn't 
> work.
> 
> The card is a 3com OfficeConnect version 1.0.  I've downloaded the
> firmware and put it in /lib/firmware/isl3890.  The card is configured 
> as eth1 and works.  There are link and activity lights on the card.  
> While the machine is running I can do 
> # ifdown eth1 && sleep 5 && ifup eth1 
> and get the expected behaviour (including the link light on the card 
> disappearing).  Messages like the following get logged but don't seem
> to be a problem (remember "ibm" is just the hostname):
>   Nov 14 21:14:51 ibm hald[2297]: Timed out waiting for hotplug event 307. Rebasing to 308
> 
> I put the laptop to sleep by typing
> # echo 3 > /proc/acpi/sleep
> 
> It goes to sleep fine, and I wake it up by pressing the function button.
> The following is logged:
> 
> Nov 14 21:41:31 ibm kernel: eth1: got suspend request (state 3)
> Nov 14 21:41:33 ibm kernel: eth1: got resume request
> 
> As the machine wakes up, the link light comes on first.  Then the activity
> light comes on and stays on solidly.  When I try and ping something on
> the network, nothing happens.  Bringing the interface up and down doesn't
> fix the problem (though sometimes the activity light will disappear after
> a while).  While you're bringing the interface up and down, messages
> like this are logged:
> Nov 14 21:47:43 ibm kernel: eth1: timeout waiting for mgmt response
> Nov 14 21:47:48 ibm kernel: NETDEV WATCHDOG: eth1: transmit timed out
> Nov 14 21:47:50 ibm kernel: eth1: timeout waiting for mgmt response
> Nov 14 21:47:51 ibm kernel: NETDEV WATCHDOG: eth1: transmit timed out
> Nov 14 21:47:51 ibm kernel: eth1: timeout waiting for mgmt response
> Nov 14 21:47:52 ibm kernel: eth1: timeout waiting for mgmt response
> Nov 14 21:47:52 ibm kernel: NETDEV WATCHDOG: eth1: transmit timed out
> Nov 14 21:47:53 ibm kernel: eth1: timeout waiting for mgmt response
> Nov 14 21:47:53 ibm kernel: eth1: mgmt tx queue is still full
> 
> The only way I've found to fix it so far is to rmmod prism54 && modprobe
> prism54.  Note that just taking out and putting
> in the card doesn't work without doing the "rmmod prism54".  
> 
> I'd be most grateful for any help fixing this problem.  Also, do
> let me know if there's any further information I should provide to
> this list on the matter.
> 
>   Simon.
> 

I do not use suspend/resume but occasionally run into the same bug.  And
I "solve" it in the same manner that you posted.  The bugzilla comment
below indicates a workaround - I have not encountered the bug since the
workaround has been posted.

http://prism54.org/cgi-bin/bugzilla/show_bug.cgi?id=56#c14

Kevin Freeman