From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754806AbdCTPNF (ORCPT ); Mon, 20 Mar 2017 11:13:05 -0400 Received: from mx1.redhat.com ([209.132.183.28]:54828 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752912AbdCTPLA (ORCPT ); Mon, 20 Mar 2017 11:11:00 -0400 DMARC-Filter: OpenDMARC Filter v1.3.2 mx1.redhat.com 1E41464D97 Authentication-Results: ext-mx09.extmail.prod.ext.phx2.redhat.com; dmarc=none (p=none dis=none) header.from=redhat.com Authentication-Results: ext-mx09.extmail.prod.ext.phx2.redhat.com; spf=pass smtp.mailfrom=mst@redhat.com DKIM-Filter: OpenDKIM Filter v2.11.0 mx1.redhat.com 1E41464D97 Date: Mon, 20 Mar 2017 16:34:22 +0200 From: "Michael S. Tsirkin" To: Alex Williamson Cc: Cao jin , linux-kernel@vger.kernel.org, kvm@vger.kernel.org, qemu-devel@nongnu.org, izumi.taku@jp.fujitsu.com Subject: Re: [PATCH] vfio pci: kernel support of error recovery only for non fatal error Message-ID: <20170320163316-mutt-send-email-mst@kernel.org> References: <1488180523-18137-1-git-send-email-caoj.fnst@cn.fujitsu.com> <20170313160619.28622002@t450s.home> <58CFD01F.1060508@cn.fujitsu.com> <20170320083056.3f2a5603@t450s.home> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20170320083056.3f2a5603@t450s.home> X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.38]); Mon, 20 Mar 2017 14:34:29 +0000 (UTC) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, Mar 20, 2017 at 08:30:56AM -0600, Alex Williamson wrote: > > > What about the case where the user has not registered for receiving > > > non-fatal errors, now we send an error signal on both error_detected > > > and slot_reset. Is that useful/desirable? > > > > > > > Not desirable, but seems not harmful, guest user will stop anyway. How > > to avoid this case gracefully seems not easy. > > "Not harmful" is presuming the behavior of the user. QEMU might not be > the only consumer of these events. Is it possible to receive a > slot_reset without first receiving an error_detected? If not then we > can easily track our action for one to decide on the behavior for the > other. Thanks, > > Alex I would just pass maximum info to userspace and let it decide. -- MST