[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <af2dae8e1628b43afc898262b64c10128f7d1a7d.camel@sipsolutions.net>
Date: Fri, 19 Jul 2024 09:31:50 -0700
From: Johannes Berg <johannes@...solutions.net>
To: Jiri Pirko <jiri@...nulli.us>
Cc: netdev@...r.kernel.org, Eric Dumazet <edumazet@...gle.com>,
syzbot+2120b9a8f96b3fa90bad@...kaller.appspotmail.com
Subject: Re: [RFC PATCH 2/2] net: bonding: don't call ethtool methods under
RCU
On Fri, 2024-07-19 at 11:50 +0200, Jiri Pirko wrote:
> Thu, Jul 18, 2024 at 09:20:17PM CEST, johannes@...solutions.net wrote:
> > From: Johannes Berg <johannes.berg@...el.com>
> >
> > Currently, bond_miimon_inspect() is called under RCU, but it
> > calls ethtool ops. Since my earlier commit facd15dfd691
> > ("net: core: synchronize link-watch when carrier is queried")
> > this is no longer permitted in the general ethtool case, but
> > it was already not permitted for many drivers such as USB in
> > which it can sleep to do MDIO register accesses etc.
> >
> > Therefore, it's better to simply not do this. Change bonding
> > to acquire the RTNL for the MII monitor work directly to call
> > the bond_miimon_inspect() function and thus ethtool ops.
>
> Is there a good reason why to directly query device here using whatever?
See Jay's email for that, I think?
> I mean, why netif_oper_up() would not return the correct bool here?
> Introduction of periodic rtnl locking for no good reason is not probably
> something we should do :/
>
Yeah, fair.
johannes
Powered by blists - more mailing lists