Message-ID: <254513dc01188ca3ff161fcb456bc96a610821a1.camel@gmx.de>
Subject: Re: [rcf/patch] netpoll: Make it RT friendly
From:   Mike Galbraith <efault@gmx.de>
To:     Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc:     Thomas Gleixner <tglx@linutronix.de>,
        LKML <linux-kernel@vger.kernel.org>,
        RT <linux-rt-users@vger.kernel.org>
Date:   Wed, 15 Dec 2021 20:54:40 +0100
In-Reply-To: <YboSe4NfbCeoqF1j@linutronix.de>
References: <20211118163405.adczki7ibk22bw7e@linutronix.de>
         <773fd8246a1ec4ef79142d9e31b8ba4163a0d496.camel@gmx.de>
         <YboSe4NfbCeoqF1j@linutronix.de>
Content-Type: text/plain; charset="UTF-8"
User-Agent: Evolution 3.42.2 
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Wed, 2021-12-15 at 17:06 +0100, Sebastian Andrzej Siewior wrote:
>
> What is the context for netpoll_send_skb()? Why do we need to disable BH
> + RCU on RT?

Because rcu_dereference_bh() says it better be, but..

> If interrupts are never disabled, doesn't this break the assumption made
> in netpoll_tx_running()?
>
> queue_process() is also busted.

..that makes it kinda moot, and wasn't unexpected. It works, or at
least netconsole does, and lockdep was fine with it, but I figured it
unlikely to fly.  In fact, I had already taken the silence as a "Surely
you jest" or such.

	-Mike