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;
Content-Type: text/plain;
        charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 13.4 \(3608.120.23.2.4\))
Subject: Re: [PATCH] mac80211: fix regression in sta connection monitor
From:   Georgi Valkov <gvalkov@abv.bg>
In-Reply-To: <099b1679-87c4-1ac5-978b-a1110db93f54@nbd.name>
Date:   Thu, 1 Oct 2020 16:59:30 +0300
Cc:     linux-wireless@vger.kernel.org, johannes@sipsolutions.net,
        openwrt-devel@lists.openwrt.org
Content-Transfer-Encoding: quoted-printable
Message-Id: <9F798A1A-4CBA-47E9-9F2A-E407EDEBCD90@abv.bg>
References: <20200921154545.91971-1-nbd@nbd.name>
 <7A099B27-5278-4D2A-91A1-6B795CB0E221@abv.bg>
 <099b1679-87c4-1ac5-978b-a1110db93f54@nbd.name>
To:     Felix Fietkau <nbd@nbd.name>
Precedence: bulk

Hi Felix!

I can confirm that the original regression is now fully resolved in v3 =
of your patch.
Further testing showed that the second issue: group key handshake failed =
(RSN) after 4 tries is unrelated. Apparently OpenWRT does not support =
group rekeying when WDS is used, so it always fails and should be =
disabled: option wpa_group_rekey =E2=80=980=E2=80=99.

Please excuse me for taking your time with the unrelated issue. There =
was no way for me to know that, since the disconnect remained, and it =
may take up to 24 hours to detect.

Thank you for your help!
Georgi

> On 2020-09-22, at 11:56 AM, Felix Fietkau <nbd@nbd.name> wrote:
>=20
> On 2020-09-22 09:28, Georgi Valkov wrote:
>> Hi Felix!
>>=20
>> In the original code before the regression, as well as in the fix I
>> proposed, probe_send_count was always reset if ack is true, while in
>> your fix it will not be reset when non-null frames are received. The
>> connection failed again after 10 hours and the STA remained offline =
for
>> two minutes instead of two seconds. I should also note that I spent
>> multiple days testing the fix I proposed and I had no disconnects.
>> Please revise your fix again!
> Thanks for the info. Handling non-null frames in that function is
> incomplete, because it does not cover the 802.3 tx status path.
> I'll make v2 patch that checks the timestamp of the last ACK in
> ieee80211_sta_work instead.
>=20
> - Felix
>=20