Bug 22842 - iwl3945 suddenly stops working
iwl3945 suddenly stops working
Status: CLOSED INSUFFICIENT_DATA
Product: Drivers
Classification: Unclassified
Component: network-wireless
All Linux
: P1 normal
Assigned To: Stanislaw Gruszka
:
Depends on:
Blocks: 16444
  Show dependency treegraph
 
Reported: 2010-11-14 11:14 UTC by Felipe Contreras
Modified: 2012-07-23 12:02 UTC (History)
7 users (show)

See Also:
Kernel Version: 2.6.36
Tree: Mainline
Regression: Yes


Attachments
Debug log (640.31 KB, application/x-xz)
2010-11-19 23:37 UTC, Felipe Contreras
Details

Description Felipe Contreras 2010-11-14 11:14:16 UTC
Just like that, there's no warning or anything.

I can recover by trying:
 * bring down/up the interface
 * rfkill
 * reload the module
 * suspend/resume

None of these alone makes it work, I don't know what exactly does, I just try all of them until it works again.

This seems to have started on 2.6.36, but I'm not sure.
Comment 1 Andrew Morton 2010-11-16 22:36:42 UTC
Recategorised as network-wireless
Comment 2 Florian Mickler 2010-11-17 08:27:35 UTC
Does turning on debugging in the driver give some clues?
Comment 3 Felipe Contreras 2010-11-17 09:58:39 UTC
(In reply to comment #2)
> Does turning on debugging in the driver give some clues?

What kind of debugging?

 CONFIG_IWLWIFI_DEBUG
 echo 0x43fff > /sys/class/net/wlan0/device/debug_level

?

I will try that, however, a bit more of clarification: I can reconnect to the access point just fine, even get an IP from DHCP, and the other APs are listed correctly. However, once connected I cannot even ping the gateway.
Comment 4 John W. Linville 2010-11-17 13:43:21 UTC
Perhaps we could see some dmesg output from after the failure?
Comment 5 wey-yi.w.guy 2010-11-17 15:37:04 UTC
echo 0x43fff > /sys/class/net/wlan0/device/debug_level will help a lot

Thanks
Wey
Comment 6 Felipe Contreras 2010-11-19 23:37:00 UTC
Created attachment 37682 [details]
Debug log

Here is a debug log.

At this point issuing a ping command returns nothing but iwl3945_mac_tx seems to enter and leave.

This time I didn't try to make any changes (bring down the iface, rkfill) and it seems after some time it recovered.
Comment 7 Felipe Contreras 2010-11-25 21:25:30 UTC
So? What else do you need?

This issue is very annoying.

I'm going to try 2.6.35 again and see if I can narrow down the issue.
Comment 8 Felipe Contreras 2010-11-25 21:26:25 UTC
BTW. Why can't I change the status from NEEDINFO? The info has been provided.
Comment 9 Florian Mickler 2010-11-25 22:07:14 UTC
The configuration of this bugzilla seems to be not ideal. I changed it for you.
Comment 10 Felipe Contreras 2010-11-25 22:25:11 UTC
(In reply to comment #9)
> The configuration of this bugzilla seems to be not ideal. I changed it for you.

Thanks, I've filed a bug for that:
https://bugzilla.kernel.org/show_bug.cgi?id=23782

Other projects use needinfo as a flag.
Comment 11 John W. Linville 2011-01-25 20:02:26 UTC
Wey-Yi, is the infor Felipe provided sufficient for you?
Comment 12 wey-yi.w.guy 2011-01-25 20:32:01 UTC
got the log, will have someone look into it.
Felipe, I will ask for more information if needed.

Thanks
Wey
Comment 13 Florian Mickler 2011-03-29 21:30:52 UTC
Is this still a problem in 2.6.38.y?
Comment 14 Stanislaw Gruszka 2011-03-31 13:39:45 UTC
If problem is not solved, does disable_hw_scan=1 helps? If not, bisection will be needed to solve it ...

Attachment should change MIME type to application/x-xz (I have no permissions to change this).
Comment 15 Felipe Contreras 2011-03-31 21:59:57 UTC
(In reply to comment #14)
> If problem is not solved, does disable_hw_scan=1 helps? If not, bisection will
> be needed to solve it ...

Unfortunately I have a new laptop and I haven't managed to use the old one at all. I'll try to use it a bit more to see if I can reproduce.

> Attachment should change MIME type to application/x-xz (I have no permissions
> to change this).

Done.
Comment 16 Stanislaw Gruszka 2012-01-18 14:32:00 UTC
Ok, I'm closing the bug with insufficient data resolution.

Note You need to log in before you can comment on or make changes to this bug.