wpa_supplicant 2.4/2.5 issue with some Intel chipsets

[Peer, Ilan]

> -----Original Message-----
> From: Evangelos Foutras [mailto:evangelos at foutrelis.com]
> Sent: Sunday, December 27, 2015 13:29
> To: Peer, Ilan; Jouni Malinen
> Cc: hostap at lists.infradead.org
> Subject: Re: wpa_supplicant 2.4/2.5 issue with some Intel chipsets
> 
> On 27/12/15 12:42, Peer, Ilan wrote:
> >> On 24/12/15 20:49, Jouni Malinen wrote:
> >>> On Fri, Dec 18, 2015 at 04:07:58PM +0200, Evangelos Foutras wrote:
> >>>> While I am not personally experiencing any issues with
> >>>> wpa_supplicant
> >>>> 2.5 (not using an Intel wifi card though), we have had several
> >>>> people reporting connection issues downstream in Arch Linux.
> >>>>
> >>>> It appears to be related to the creation and/or usage of P2P
> >>>> management interfaces on some Intel chipsets. (Pardon any wrong
> >>>> terminology, I am not familiar with the wireless stack.)
> >>>>
> >>>> A common error message appears to be:
> >>>>
> >>>>   Could not read interface p2p-dev-wlp2s0 flags: No such device
> >>>>
> >
> > This print is probably harmless. I've just sent a patch that fixes this issue. See
> http://patchwork.ozlabs.org/patch/561134/.
> >
> > Regards,
> >
> > Ilan.
> 
> With that out of the way, is it possible that, when called with the -W flag,
> wpa_supplicant does wpa_supplicant_ctrl_iface_wait() for the same control
> interface twice? (Once for the regular interface and one for the P2P one,
> which AFAICS inherits it when no P2P conf file is given.)
> 
> The above would explain the behavior reported by our users but I am not
> sure if that is the case.

There is no specific description of the issues seen by the users but this is indeed possible. I've just sent 2 additional patches to handle this, can you give them a try?

http://patchwork.ozlabs.org/patch/561147/
http://patchwork.ozlabs.org/patch/561146/

Thanks in advance,

Ilan.