From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756053Ab0A2Wym (ORCPT ); Fri, 29 Jan 2010 17:54:42 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755943Ab0A2Wyl (ORCPT ); Fri, 29 Jan 2010 17:54:41 -0500 Received: from sj-iport-6.cisco.com ([171.71.176.117]:63287 "EHLO sj-iport-6.cisco.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752339Ab0A2Wyk (ORCPT ); Fri, 29 Jan 2010 17:54:40 -0500 Authentication-Results: sj-iport-6.cisco.com; dkim=neutral (message not signed) header.i=none X-IronPort-Anti-Spam-Filtered: true X-IronPort-Anti-Spam-Result: ApoEAIr2YkurR7H+/2dsb2JhbADDGpdOhEIE X-IronPort-AV: E=Sophos;i="4.49,371,1262563200"; d="scan'208";a="475123963" From: Roland Dreier To: Alex Chiang Cc: linux-rdma@vger.kernel.org, justin.chen@hp.com, linux-kernel@vger.kernel.org Subject: Re: [PATCH 0/7] Increase maximum Infiniband HCAs per-system References: <20100129214039.17745.38679.stgit@bob.kio> X-Message-Flag: Warning: May contain useful information Date: Fri, 29 Jan 2010 14:54:36 -0800 In-Reply-To: <20100129214039.17745.38679.stgit@bob.kio> (Alex Chiang's message of "Fri, 29 Jan 2010 14:44:57 -0700") Message-ID: User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org > Justin Chen discovered that Linux "only" supports 32 IB cards in > a single system when testing a larger system with 40 cards and > discovered that OFED only reported 32 HCAs. > > This patchset doubles the number of HCAs allowed per system in a > backwards-compatible manner. Looks like a really great start, some nice cleanups as well the added functionality. I've been meaning to use pahole for a while... Have you considered drivers/infiniband/core/user_mad.c and ucm.c? I think user_mad.c is somewhat more important, as that is what allows an adapter to be used for running the SM. So I think we're still left with some potential issues around lots of adapters in one system. (I think use of ucm by real apps is minimal to nonexistent, but someday we should deal with that too) - R.