Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Christian Lamparter <chunkeey@gmail.com>
To:     Frank =?ISO-8859-1?Q?Sch=E4fer?= <fschaefer.oss@googlemail.com>
Cc:     chunkeey@googlemail.com, j@w1.fi, linux-wireless@vger.kernel.org
Subject: Re: [Bug] carl9170 + wpa_supplicant: P2P mode doesn't work with separate interfaces
Date:   Sat, 18 Apr 2020 18:30:13 +0200
Message-ID: <8972271.5nXPVzACVl@debian64>
In-Reply-To: <3a9d86b6-744f-e670-8792-9167257edef8@googlemail.com>
References: <3a9d86b6-744f-e670-8792-9167257edef8@googlemail.com>
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain; charset="iso-8859-1"
Sender: linux-wireless-owner@vger.kernel.org
Precedence: bulk

Hi,

On Saturday, 18 April 2020 13:20:27 CEST Frank Sch=E4fer wrote:
> I'm currently testing Miraclecast=20
> (https://github.com/albfan/miraclecast), an open source=20
> Wifi-Display/Miracast implementation.
> In one of my setups, I'm using a carl9170 device as sink, which fails=20
> with the following debugging output:
>=20
It's been a very long time. But I do remember meddling with P2P. Part of the
reason is that the carl9170 driver needed these virtual interfaces initiali=
zed
in a specific order.

So for P2P GO+CLIENT, you would have to initialize the P2P-GO interface fir=
st
and the P2P Client interface second. Which is backwards of what the wpa_sup=
plicant
does.

Same is true for STA+AP (Repeater). The AP has to start first before the STA
can be brought up (Also the STA must not interfere with channel operations,=
 as
the fast channel change feature doesn't work right on the AR9170 Hardware).

This is documented in the driver:
<https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/dr=
ivers/net/wireless/ath/carl9170/main.c#n625>

and was changed as part of this commit:
<https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/=
drivers/net/wireless/ath/carl9170/main.c?id=3D7f878b0d96e6b3bd27d736fb2fb7e=
3cc94b16b26>

that identifies the wpa_supplicant commit
["nl80211: Automatically use concurrent P2P if possible"] as the reason.

So since this broke, something must have changed since 2012 I guess.
But I need more details, before I can do something.

Regards,
Christian