Received-SPF: pass (google.com: best guess record for domain of linux-wireless-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Toke =?utf-8?Q?H=C3=B8iland-J=C3=B8rgensen?= <toke@redhat.com>
To:     Johannes Berg <johannes@sipsolutions.net>,
        Jens Axboe <axboe@kernel.dk>,
        Emmanuel Grumbach <emmanuel.grumbach@intel.com>,
        Luca Coelho <luciano.coelho@intel.com>
Cc:     "linux-wireless\@vger.kernel.org" <linux-wireless@vger.kernel.org>,
        Networking <netdev@vger.kernel.org>
Subject: Re: iwlwifi warnings in 5.5-rc1
In-Reply-To: <3420d73e667b01ec64bf0cc9da6232b41e862860.camel@sipsolutions.net>
References: <ceb74ea2-6a1b-4cef-8749-db21a2ee4311@kernel.dk> <9727368004ceef03f72d259b0779c2cf401432e1.camel@sipsolutions.net> <878snjgs5l.fsf@toke.dk> <3420d73e667b01ec64bf0cc9da6232b41e862860.camel@sipsolutions.net>
Date:   Wed, 11 Dec 2019 11:23:02 +0100
Message-ID: <875zingnzt.fsf@toke.dk>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Sender: linux-wireless-owner@vger.kernel.org
Precedence: bulk

Johannes Berg <johannes@sipsolutions.net> writes:

> On Wed, 2019-12-11 at 09:53 +0100, Toke H=C3=B8iland-J=C3=B8rgensen wrote=
:
>> Johannes Berg <johannes@sipsolutions.net> writes:
>>=20
>> > On Tue, 2019-12-10 at 13:46 -0700, Jens Axboe wrote:
>> > > Hi,
>> > >=20
>> > > Since the GRO issue got fixed, iwlwifi has worked fine for me.
>> > > However, on every boot, I get some warnings:
>> > >=20
>> > > ------------[ cut here ]------------
>> > > STA b4:75:0e:99:1f:e0 AC 2 txq pending airtime underflow: 4294967088=
, 208
>> >=20
>> > Yeah, we've seen a few reports of this.
>>=20
>> FWIW I've tried reproducing but I don't get the error with the 8265 /
>> 8275 chip in my laptop. I've thought about sending a patch for mac80211
>> to just clear the tx_time_est field after calling
>> ieee80211_sta_update_pending_airtime() - that should prevent any errors
>> from double-reporting of skbs (which is what I'm guessing is going on
>> here).
>
> It does feel like it, but I'm not sure how that'd be possible?
>
> OK, I talked with Emmanuel and I think it's the GSO path - it'll end up
> with skb_clone() and then report both of them back.

Right, figured it was something like that; just couldn't find the place
in the driver where it did that from my cursory browsing.

> Regardless, I think I'll probably have to disable AQL and make it more
> opt-in for the driver - I found a bunch of other issues ...

Issues like what? Making it opt-in was going to be my backup plan; I was
kinda hoping we could work out any issues so it would be a "no harm"
kind of thing that could be left as always-on. Maybe that was a bit too
optimistic; but it's also a pain having to keep track of which drivers
have which features/fixes...

-Toke