Re: [PATCH 2.5] fixes for airo.c

From: Javier Achirica <achirica@telefonica.net>
To: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Daniel Ritz <daniel.ritz@gmx.ch>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	linux-net <linux-net@vger.kernel.org>
Subject: Re: [PATCH 2.5] fixes for airo.c
Date: Wed, 23 Jul 2003 20:20:20 +0200 (MEST)	[thread overview]
Message-ID: <Pine.SOL.4.30.0307232018120.15976-100000@tudela.mad.ttd.net> (raw)
In-Reply-To: <1058983403.5516.101.camel@dhcp22.swansea.linux.org.uk>

On 23 Jul 2003, Alan Cox wrote:

> On Mer, 2003-07-23 at 18:56, Daniel Ritz wrote:
> > > You cannot use down() in xmit, as it may be called in interrupt context. I
> > > know it slows things down, but that's the only way I figured out of
> > > handling a transmission while the card is processing a long command.
> >
> > hu? no. you can do a down() as xmit is never called from interrupt context. and
> > the dev->hard_start_xmit() calls are serialized with the dev->xmit_lock. the
> > serialization is broken by the schedule_work() thing.
>
> If you are about to start a long command why not mark the device busy
> for transmit before starting ?

I thought about that some time ago. The problem I have in some cases is
that there are commands that, based on the status of the radio, may be
very fast or very long, I didn't think that marking the devide busy "just
in case" before every command was very efficient.

Javier Achirica