From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751773AbdAWVBL (ORCPT ); Mon, 23 Jan 2017 16:01:11 -0500 Received: from mga11.intel.com ([192.55.52.93]:43911 "EHLO mga11.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751746AbdAWVBJ (ORCPT ); Mon, 23 Jan 2017 16:01:09 -0500 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.33,275,1477983600"; d="scan'208";a="34422920" Date: Mon, 23 Jan 2017 12:57:25 -0800 From: Yu-cheng Yu To: Dave Hansen Cc: fenghua.yu@intel.com, dvlasenk@redhat.com, peterz@infradead.org, oleg@redhat.com, mingo@kernel.org, linux-kernel@vger.kernel.org, brgerst@gmail.com, luto@kernel.org, bp@alien8.de, jpoimboe@redhat.com, haokexin@gmail.com, hpa@zytor.com, quentin.casasnovas@oracle.com, tglx@linutronix.de, torvalds@linux-foundation.org, riel@redhat.com, linux-tip-commits@vger.kernel.org Subject: Re: [tip:x86/urgent] x86/fpu: Set the xcomp_bv when we fake up a XSAVES area Message-ID: <20170123205725.GA3920@test-lenovo> References: <1485075023-30161-1-git-send-email-haokexin@gmail.com> <20170123165529.GA4996@test-lenovo> <2be814b7-9fd6-7955-b4e3-6ecb4ef76052@linux.intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <2be814b7-9fd6-7955-b4e3-6ecb4ef76052@linux.intel.com> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Jan 23, 2017 at 09:23:06AM -0800, Dave Hansen wrote: > On 01/23/2017 08:55 AM, Yu-cheng Yu wrote: > >> The best fix here would be not to paper over the issue in the copy > >> function but find where it got clobbered, or where some initialization > >> code failed to set it. > > > > Someone else reported different issues from the same bug and a different > > patch was just tested OK this morning. I think that adding xfeatures bits > > to xcomp_bv should have been done in fpstate_init(). > > Right. So where did it get cleared out? It is not set until a task triggers XSAVES. We did not set it in fpstate_init() because there is no valid data at the time. The problem happens when Linux copies data to the XSAVES area, like we see here; the kernel is not expected to change XSAVES format (xcomp_bv) but xcomp_bv is still blank (except bit 63). Because XSAVES format is not changed after boot time and xcomp_bv does not affect INIT optimization, why don't we fix the problem in fpstate_init()? Yu-cheng