[ath9k-devel] nohwcrypt requirement for virtual interfaces

[ath9k-devel] nohwcrypt requirement for virtual interfaces

[Goo Mail]

Hi all,

Can anyone tell me if the nohwcrypt=1 module parameter is still required
when using ath9k with multiple virtual interfaces in AP mode?

I recently ran into some weird
issues<https://forum.openwrt.org/viewtopic.php?pid=105006>where a
client on one of my VIFs was not seeing broadcast ethernet traffic
(specifically, ARP requests) sent by the base station through a bridged
interface, even though plenty of other wireless traffic on that VIF was
passed normally. I noticed that the VIFs worked fine and even passed the
broadcast traffic properly when I turned the encryption off, but it broke as
soon as I configured the VIFs to use WPA2-PSK.

I was about to file a bug report, but I just now ran across a year-old
linux-wireless
thread <http://lwn.net/Articles/321690/> mentioning that nohwcrypt=1 was
required when using ath9k with virtual interfaces. I just tried flipping
that parameter on and things now seem to be working much better as far as
broadcast traffic. However, the linux-wireless thread hinted that the
requirement for nohwcrypt might be changed in the future depending on key
cache handling and such.

Is nohwcrypt still required, and if so, is there any more specific set of
situations where it is definitely needed?

Thanks,
Scott
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20100323/ebac5e8a/attachment.htm 

[ath9k-devel] nohwcrypt requirement for virtual interfaces

[yingqiang Ma]

I encountered this issue. And it turns to be cause by the group key.

Ath9k only can install one group key into the hardware, so the group key of
the first VIF will be overwritten by the following VIF.

look at the code here:
ath9k/main.c::ath_key_config:
    if (!(key->flags & IEEE80211_KEY_FLAG_PAIRWISE)) {
        /* For now, use the default keys for broadcast keys. This may
         * need to change with virtual interfaces. */
        idx = key->keyidx;

The idx eventually will be always 1 for each VIF.

We workaround it by set an identical group key for all VIFs. Hope someone
can fix it soon.

Regards,
Daniel

2010/3/24 Goo Mail <goomail59@gmail.com>

> Hi all,
>
> Can anyone tell me if the nohwcrypt=1 module parameter is still required
> when using ath9k with multiple virtual interfaces in AP mode?
>
> I recently ran into some weird issues<https://forum.openwrt.org/viewtopic.php?pid=105006>where a client on one of my VIFs was not seeing broadcast ethernet traffic
> (specifically, ARP requests) sent by the base station through a bridged
> interface, even though plenty of other wireless traffic on that VIF was
> passed normally. I noticed that the VIFs worked fine and even passed the
> broadcast traffic properly when I turned the encryption off, but it broke as
> soon as I configured the VIFs to use WPA2-PSK.
>
> I was about to file a bug report, but I just now ran across a year-old linux-wireless
> thread <http://lwn.net/Articles/321690/> mentioning that nohwcrypt=1 was
> required when using ath9k with virtual interfaces. I just tried flipping
> that parameter on and things now seem to be working much better as far as
> broadcast traffic. However, the linux-wireless thread hinted that the
> requirement for nohwcrypt might be changed in the future depending on key
> cache handling and such.
>
> Is nohwcrypt still required, and if so, is there any more specific set of
> situations where it is definitely needed?
>
> Thanks,
> Scott
>
>
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20100324/b3e93bfb/attachment-0001.htm