From mboxrd@z Thu Jan 1 00:00:00 1970 From: Felix Fietkau Subject: Re: [PATCH v4 4/9] bgmac: simplify rx DMA error handling Date: Mon, 13 Apr 2015 17:05:38 +0200 Message-ID: <552BDB42.1080104@openwrt.org> References: <1428933162-26763-1-git-send-email-nbd@openwrt.org> <1428933162-26763-4-git-send-email-nbd@openwrt.org> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Network Development , Hauke Mehrtens To: =?UTF-8?B?UmFmYcWCIE1pxYJlY2tp?= Return-path: Received: from arrakis.dune.hu ([78.24.191.176]:46397 "EHLO arrakis.dune.hu" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932228AbbDMPFx (ORCPT ); Mon, 13 Apr 2015 11:05:53 -0400 In-Reply-To: Sender: netdev-owner@vger.kernel.org List-ID: On 2015-04-13 16:53, Rafa=C5=82 Mi=C5=82ecki wrote: > On 13 April 2015 at 15:52, Felix Fietkau wrote: >> Unmap the DMA buffer before checking it. If it is poisoned, map it a= gain >> and pass it back to the hardware. >=20 > Where exactly do you map it again? I can't see it. It's not handled i= n > bgmac_dma_rx_setup_desc. Also keep in mind that mapping may fail - > this is why old code doesn't unmap anything until a new buffer is > really ready. How do you want to deal with mapping error after > unmapping a buffer? You're right, seems that I accidentally dropped a chunk of the patch. - Felix