[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20211213081556.1a575a28@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
Date: Mon, 13 Dec 2021 08:15:56 -0800
From: Jakub Kicinski <kuba@...nel.org>
To: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
Cc: netdev@...r.kernel.org, "David S. Miller" <davem@...emloft.net>,
Eric Dumazet <edumazet@...gle.com>,
Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH net-next] net: dev: Always serialize on Qdisc::busylock
in __dev_xmit_skb() on PREEMPT_RT.
On Mon, 13 Dec 2021 11:45:59 +0100 Sebastian Andrzej Siewior wrote:
> On 2021-12-10 20:32:56 [-0800], Jakub Kicinski wrote:
> > On Fri, 10 Dec 2021 16:41:44 +0100 Sebastian Andrzej Siewior wrote:
> > > - contended = qdisc_is_running(q);
> > > + if (!IS_ENABLED(CONFIG_PREEMPT_RT))
> > > + contended = qdisc_is_running(q);
> > > + else
> > > + contended = true;
> >
> > Why not:
> >
> > contended = qdisc_is_running(q) || IS_ENABLED(CONFIG_PREEMPT_RT);
>
> I could do that. But I would swap the two arguments so that the
> IS_ENABLED macro comes first and if true qdisc_is_running() is optimized
> away.
FWIW I disagree. My version was more readable. The sprinkling of the
PREEMPT_RT tosh is getting a little annoying. Trying to regress the
clarity of the code should be top of mind here.
Powered by blists - more mailing lists