Return-Path: Received: from mail-oi0-f41.google.com ([209.85.218.41]:34950 "EHLO mail-oi0-f41.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751671AbcFVNsz (ORCPT ); Wed, 22 Jun 2016 09:48:55 -0400 MIME-Version: 1.0 In-Reply-To: <576A92BA.3020703@grimberg.me> References: <20160620155751.10809.22262.stgit@manet.1015granger.net> <20160620161200.10809.45762.stgit@manet.1015granger.net> <576A92BA.3020703@grimberg.me> From: Or Gerlitz Date: Wed, 22 Jun 2016 16:47:58 +0300 Message-ID: Subject: Re: [PATCH v3 25/25] IB/mlx4: Workaround for mlx4_alloc_priv_pages() array allocator To: Sagi Grimberg Cc: Chuck Lever , Yishai Hadas , "linux-rdma@vger.kernel.org" , "linux-nfs@vger.kernel.org" , Leon Romanovsky Content-Type: text/plain; charset=UTF-8 Sender: linux-nfs-owner@vger.kernel.org List-ID: On Wed, Jun 22, 2016 at 4:29 PM, Sagi Grimberg wrote: > >>> Ensure the MR's PBL array never occupies the last 8 bytes of a page. >>> This eliminates random "Local Protection Error" flushes when SLUB >>> debugging is enabled. >>> >>> Fixes: 1b2cd0fc673c ('IB/mlx4: Support the new memory registration API') >> >> >> Can't the driver advertize smaller quantity for what's occupies later >> those last eight bytes (255 or 511 of >> attr XX instead of 256 or 512)? > Not sure I understand your question. Are you suggesting that the driver > would expose that it's capable of 256 pages per MR? I am asking if we can advertize something else what would cause not to fall into the last eight bytes on a page, e.g if we fall there since we advertize for some attr N, lets advertize N-1 just asking