Received-SPF: pass (google.com: domain of linux-wireless-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <abcf4b9aed8adad05841234dad103ced15f9bfb2.camel@sipsolutions.net>
Subject: Re: traceability of wifi packet drops
From:   Johannes Berg <johannes@sipsolutions.net>
To:     Jakub Kicinski <kuba@kernel.org>,
        Eric Dumazet <edumazet@google.com>
Cc:     netdev@vger.kernel.org, linux-wireless@vger.kernel.org
Date:   Tue, 28 Mar 2023 09:37:43 +0200
In-Reply-To: <20230327180950.79e064da@kernel.org>
References: <00659771ed54353f92027702c5bbb84702da62ce.camel@sipsolutions.net>
         <20230327180950.79e064da@kernel.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.46.4 (3.46.4-1.fc37) 
MIME-Version: 1.0
Precedence: bulk

On Mon, 2023-03-27 at 18:09 -0700, Jakub Kicinski wrote:
> > Anyone have any great ideas?
>=20
> We need something that'd scale to more subsystems, so I don't think
> having all the definitions in enum skb_drop_reason directly is an
> option.

Fair enough.

> My knee jerk idea would be to either use the top 8 bits of the
> skb reason enum to denote the space. And then we'd say 0 is core
> 1 is wifi (enum ieee80211_rx_result) etc. Within the WiFi space=20
> you can use whatever encoding you like.

Right. That's not _that_ far from what I proposed above, except you pull
the core out=20

> On a quick look nothing is indexed by the reason directly, so no
> problems with using the high bits.

I think you missed he drop_reasons[] array in skbuff.c, but I guess we
could just not add these to the DEFINE_DROP_REASON() macro (and couldn't
really add them anyway).

The only user seems to be drop_monitor, which anyway checks the array
bounds (in the trace hit function.)

Or we change the design of this to actually have each subsystem provide
an array/a callback for their namespace, if the strings are important?
Some registration/unregistration might be needed for modules, but that
could be done.

> Option #2 is to add one main drop reason called SKB_DROP_REASON_MAC80211
> and have a separate tracepoint which exposes the detailed wifi
> reason and any necessary context. mac80211 would then have its own
> wrapper around kfree_skb_reason() which triggers the tracepoint.

Yeah, I considered doing that with just the line number, but who knows
what people might want to use this for in the end, so that's not a great
idea either I guess :-)

I would prefer the version with the drop reasons, since then also you
only have to worry about one tracepoint.

johannes