From mboxrd@z Thu Jan 1 00:00:00 1970 From: Gabriel Krisman Bertazi Subject: Re: [PATCH v4] bnx2x: Alloc 4k fragment for each rx ring buffer element Date: Thu, 28 May 2015 10:26:43 -0300 Message-ID: <87zj4osui4.fsf@sycorax.br.ibm.com> References: <553AD097.4020209@gmx.de> <87383jhxj9.fsf@dhcp-9-18-235-171.br.ibm.com> <20150430.160537.232956261880122754.davem@davemloft.net> <874mnsxhnm.fsf_-_@dhcp-9-18-235-171.br.ibm.com> <87wq02t6ce.fsf_-_@dhcp-9-18-235-171.br.ibm.com> <55658781.8060104@redhat.com> Mime-Version: 1.0 Content-Type: text/plain Cc: Michal Schmidt , David Miller , "LinoSanfilippo\@gmx.de" , Ariel Elior , netdev , "cascardo\@cascardo.eti.br" , "brking\@linux.vnet.ibm.com" To: Yuval Mintz Return-path: Received: from e24smtp01.br.ibm.com ([32.104.18.85]:56894 "EHLO e24smtp01.br.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751556AbbE1N0u (ORCPT ); Thu, 28 May 2015 09:26:50 -0400 Received: from /spool/local by e24smtp01.br.ibm.com with IBM ESMTP SMTP Gateway: Authorized Use Only! Violators will be prosecuted for from ; Thu, 28 May 2015 10:26:48 -0300 Received: from d24relay02.br.ibm.com (d24relay02.br.ibm.com [9.13.184.26]) by d24dlp01.br.ibm.com (Postfix) with ESMTP id 1C8B33520062 for ; Thu, 28 May 2015 09:25:42 -0400 (EDT) Received: from d24av01.br.ibm.com (d24av01.br.ibm.com [9.8.31.91]) by d24relay02.br.ibm.com (8.14.9/8.14.9/NCO v10.0) with ESMTP id t4SDPVQj64946242 for ; Thu, 28 May 2015 10:25:31 -0300 Received: from d24av01.br.ibm.com (localhost [127.0.0.1]) by d24av01.br.ibm.com (8.14.4/8.14.4/NCO v10.0 AVout) with ESMTP id t4SDQiPY021526 for ; Thu, 28 May 2015 10:26:45 -0300 In-Reply-To: (Yuval Mintz's message of "Thu, 28 May 2015 07:24:22 +0000") Sender: netdev-owner@vger.kernel.org List-ID: Yuval Mintz writes: > Actually, this upsets me greatly. We didn't see it on a system with 4KB > pages, but this means you've actually tried to 'sell' us a fastpath fix that > was never tested on machines for which it was meant as an improvement. The iteration that inserted this bug was such a quick fix that I didn`t bother rerunning it. It just modified the type variable. Take my word it was a one time honest mistake. The last iteration, as well as all of the previous ones were tested on Power servers with bnx2x adapters. The tests included stressing the device and several iterations of hotplugging the adapter, helping us to verify correctness of the rx queue operation as well as verifying that DMA mapping/unmapping were correct. -- Gabriel Krisman Bertazi