Jonathan, thanks for the response. I won't have physical access to the laptop again until the 
weekend, but I attach an edited syslog, which may give the developers some clues.

An additional piece of information which I see in there is that some 6 hours after booting the new 
kernel, the ath5k driver starts to report "wlan0: moving STA 30:46:9a:3c:12:70 to state 2", state 1, 
state 0; and at that instant wpa_supplicant reports that the wireless is disconnected: 
CTRL-EVENT-DISCONNECTED bssid=30:46:9a:3c:12:70 reason=4

The log shows that connection is re-established after several attempts, but the same sequence then 
repeats 4 hours later, and every few hours after that, until after about half a day the connection 
stays up. Eventually though, after 1 day 18 hours from boot, the disconnect sequence re-occurs and 
then, immediately, the "gain calibration timeout" messages start at 2 per second, and the wireless 
is borked permanently, including across a reboot.

The laptop is running fine, with no errors reported, on the old Kernel, (3.0.0), after a power 
cycle, and the iwconfig looks like this:

wlan0     IEEE 802.11bg  ESSID:"Annaghvarn"
           Mode:Managed  Frequency:2.427 GHz  Access Point: 30:46:9A:3C:12:70
           Bit Rate=54 Mb/s   Tx-Power=20 dBm
           Retry  long limit:7   RTS thr:off   Fragment thr:off
           Power Management:off
           Link Quality=43/70  Signal level=-67 dBm
           Rx invalid nwid:0  Rx invalid crypt:0  Rx invalid frag:0
           Tx excessive retries:0  Invalid misc:1   Missed beacon:0



On 01/04/12 21:52, Jonathan Nieder wrote:
> Steve Graham wrote:
>
>> [Subject: Bug#665881: not Debian-specific]
>
> Please keep in mind that these messages appear as emails in a crowded
> inbox, so the subject line can be a good place to put valuable
> context.
>
>> I have just compiled a custom 3.3.0 kernel for an Acer netbook,
>> previously running 3.0.0, and see the same problem as reported by
>> Hans. In my case, after 36 hours of uptime with the new kernel, the
>> ath5k began to report constant "gain calibration timeout" errors and
>> connectivity was lost.
>>
>> Rebooting with the previous kernel without powering off did not
>> restore wireless connectivity. The old ath5k was reporting "gain
>> calibration timeout" too. Powering off and on, and booting the old
>> 3.0 kernel seems to have restored correct functionality.
>
> Thanks!  This agrees with Hans's observations, too.
>
> Please send a summary of your observations to
> linux-wireless@vger.kernel.org, cc-ing Hans, ath5k-devel@lists.ath5k.org,
> and either me or this bug log so we can track it.  Be sure to mention:
>
>   - precise steps to reproduce, expected result, actual result, and how
>     the difference indicates a bug (should be simple in this case)
>   - how reproducible and how quickly reproducible it is (e.g., within
>     5 hours 50% of the time?)
>   - which kernel versions you have tested and what happened with each
>   - full "dmesg" output from a working and a non-working boot, as
>     attachments
>   - a pointer to http://bugs.debian.org/665881 for the backstory
>   - any workarounds or other weird observations
>
> Hopefully someone upstream may have ideas for commands or patches to
> try in order to narrow down the problem.
>
> Thanks and hope that helps,
> Jonathan
>