From mboxrd@z Thu Jan 1 00:00:00 1970 From: Joe Perches Subject: Re: [PATCH 02/12] net: axienet: Handle 0 packet receive gracefully Date: Tue, 05 May 2015 11:53:49 -0700 Message-ID: <1430852029.6575.13.camel@perches.com> References: <7fb84f65a61bbe0fdb4b61a871cf4d4f7910955d.1430817941.git.michal.simek@xilinx.com> <55b3c89b3549c61d62b8440636516fd572870842.1430817941.git.michal.simek@xilinx.com> <1430834264.7191.9.camel@perches.com> <554910D1.1030706@xilinx.com> Mime-Version: 1.0 Content-Type: text/plain; charset="ISO-8859-1" Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, Peter Crosthwaite , =?ISO-8859-1?Q?S=F6ren?= Brinkmann , monstr@monstr.eu, John Linn , Anirudha Sarangi , linux-kernel@vger.kernel.org, linux-arm-kernel@lists.infradead.org To: Michal Simek Return-path: In-Reply-To: <554910D1.1030706@xilinx.com> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Tue, 2015-05-05 at 20:49 +0200, Michal Simek wrote: > On 05/05/2015 03:57 PM, Joe Perches wrote: > > On Tue, 2015-05-05 at 11:25 +0200, Michal Simek wrote: > >> From: Peter Crosthwaite > >> > >> The AXI-DMA rx-delay interrupt can sometimes be triggered > >> when there are 0 outstanding packets received. This is due > >> to the fact that the receive function will greedily consume > >> as many packets as possible on interrupt. So if two packets > >> (with a very particular timing) arrive in succession they > >> will each cause the rx-delay interrupt, but the first interrupt > >> will consume both packets. > >> This means the second interrupt is a 0 packet receive. > >> > >> This is mostly OK, except that the tail pointer register is > >> updated unconditionally on receive. Currently the tail pointer > >> is always set to the current bd-ring descriptor under > >> the assumption that the hardware has moved onto the next > >> descriptor. What this means for length 0 recv is the current > >> descriptor that the hardware is potentially yet to use will > >> be marked as the tail. This causes the hardware to think > >> its run out of descriptors deadlocking the whole rx path. > >> > >> Fixed by updating the tail pointer to the most recent > >> successfully consumed descriptor. > > > > I think some of this would be good to have as comments > > in the code instead of just in the changelog. > Is it really needed? If yes, no problem to add it but git blame can > point you to that. That's up to you. I think that useful but concealed information is always hard to follow or find.