From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: net-next panic in ovs call to arch_fast_hash2 since e5a2c899 Date: Thu, 13 Nov 2014 21:45:49 -0500 (EST) Message-ID: <20141113.214549.1520205472319716774.davem@davemloft.net> References: <12086.1415931332@famine> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: netdev@vger.kernel.org, discuss@openvswitch.org, pshelar@nicira.com, ogerlitz@mellanox.com To: jay.vosburgh@canonical.com Return-path: Received: from shards.monkeyblade.net ([149.20.54.216]:34936 "EHLO shards.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S933518AbaKNCpy (ORCPT ); Thu, 13 Nov 2014 21:45:54 -0500 In-Reply-To: <12086.1415931332@famine> Sender: netdev-owner@vger.kernel.org List-ID: From: Jay Vosburgh Date: Thu, 13 Nov 2014 18:15:32 -0800 > The "have feature" function, __intel_crc4_2_hash2, does not > clobber %r8, and so the call does not panic on a system with > X86_FEATURE_XMM4_2, although I'm not sure if that's a deliberate > compiler action or just happenstance because __intel_crc4_2_hash2 uses > fewer registers than __jhash2. Perhaps alternative calls can only be used with assembler routines that use specific calling conventions, and they therefore generally don't work with C functions?