Return-path: Received: from ey-out-2122.google.com ([74.125.78.26]:20689 "EHLO ey-out-2122.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755513AbZAXU4u (ORCPT ); Sat, 24 Jan 2009 15:56:50 -0500 Received: by ey-out-2122.google.com with SMTP id 22so1111749eye.37 for ; Sat, 24 Jan 2009 12:56:48 -0800 (PST) Message-ID: <497B808F.6060009@gmail.com> (sfid-20090124_215655_334774_C68D1B6E) Date: Sat, 24 Jan 2009 21:56:47 +0100 From: Artur Skawina MIME-Version: 1.0 To: Artur Skawina CC: Christian Lamparter , linux-wireless@vger.kernel.org, Larry Finger Subject: Re: [RFC][PATCH v2] p54usb: rx refill revamp References: <200901232245.15216.chunkeey@web.de> <200901240215.44226.chunkeey@web.de> <497A967F.2010900@gmail.com> <200901241206.18690.chunkeey@web.de> <497B71E9.6080703@gmail.com> In-Reply-To: <497B71E9.6080703@gmail.com> Content-Type: text/plain; charset=UTF-8 Sender: linux-wireless-owner@vger.kernel.org List-ID: Artur Skawina wrote: > Christian Lamparter wrote: >> On Saturday 24 January 2009 05:18:07 Artur Skawina wrote: >>> [My version schedules the work for every urb, even the poisoned ones] >> well, there's now a hard limit... no change of a endless loop now. > > The whole point of the poisoning was to prevent resubmission when > canceling the urbs -- if you work around that manually, you could just > as well kill them, instead of poisoning. > I don't understand why want to add extra code to the rx irq just to > avoid scheduling a work when downing the i/f, and keep a nasty failure > case. The difference in down() performance is not going to be measurable, > and even if it was, it wouldn't matter. Oh, and we could always do something like if (likely(atomic_read(&urb->reject)==0)) queue_work(priv->common.hw->workqueue, &priv->rx_refill_work); which should catch most cases then urbs are either killed or poisoned. artur