[pkg-wpa-devel] Bug#768130: Bug#768130: wpa_supplicant[1689]: nl80211: send_and_recv->nl_recvmsgs failed: -33" when trying to start apache

[Stefan Lippers-Hollmann]

Hi

On 2015-02-23, Willem van den Akker wrote:
> I have a laptop with an Intel Corporation Centrino Wireless-N 1030 wifi
> adapter.
> (Network controller: Intel Corporation Centrino Wireless-N 1030 [Rainbow
> Peak] (rev 34))
> 
> This adapter worked fine until there was an upgrade to wpa_suppliant
> 2.x.
> After that many 'nl80211: send_and_recv->nl_recvmsgs failed: -33'
> messages are found
> in the log and the wifi connection is down for about 15 seconds. This
> happens every 2 minutes.
> 
> Feb 23 17:40:14 notebook wpa_supplicant[1117]: nl80211:
> send_and_recv->nl_recvmsgs failed: -33
> 
> This makes the jessie for this kind of laptops almost unusable. 
> 
> I have tried the suggestion 'modprobe iwlwifi 11n-disable=1' 2, 4 or 8.
> But then can cannot 
> connect at all to my AP.
> 
> So I think there is a major problem for the Jessie iwlwifi users.
> 
> I hope for a fast solution ;)

As mentioned before, I do not believe that this is a bug in 
wpasupplicant, but rather a kernel issue - as I've never 
seen this (or similar) bugs myself with non-iwlwifi drivers
(as I unfortunately don't have access to modern intel wlan 
cards) and it's the kernel's job to provide a driver 
agnostic API between mac80211 based kernel drivers and 
userspace, namely wpa_supplicant.

Therefore we -well you or the original submitter, as I don't
have access to iwlwifi based devices- really need to establish
if this really is a regression in the wpasupplicant package or
in the kernel, which probably was updated several times within
a similar time frame of the wpa_supplicant 2.2/ 2.3 uploads.
The absence of bugreports for non-iwlwifi devices suggests 
otherwise.

In order to debug this, you can try kernel 3.19 from experimental, 
check if you have the most current firmware (ucode) for your wlan
card (firmware-iwlwifi might not carry the newer firmware blobs
due to the freeze, dmesg might tell) and do comparative tests with
the different wpasupplicant versions from

	http://snapshot.debian.org/package/wpa/

My hunch remains that this is probably a kernel regression, but
this can only be confirmed with those tests and by someone who
can reproduce the issue/ owns the affected hardware - and if this
really is a bug in wpa_supplicant, ideally by a git bisection 
between the last known-good and the first known-broken version.

Regards
	Stefan Lippers-Hollmann
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 819 bytes
Desc: Digitale Signatur von OpenPGP
URL: <http://lists.alioth.debian.org/pipermail/pkg-wpa-devel/attachments/20150223/ec64e507/attachment.sig>