From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:50083) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1f115X-00047c-5g for qemu-devel@nongnu.org; Tue, 27 Mar 2018 22:45:24 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1f115U-00047w-2k for qemu-devel@nongnu.org; Tue, 27 Mar 2018 22:45:23 -0400 Received: from mx3-rdu2.redhat.com ([66.187.233.73]:53644 helo=mx1.redhat.com) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1f115T-00046s-ST for qemu-devel@nongnu.org; Tue, 27 Mar 2018 22:45:19 -0400 References: <20180327113452.19613-1-dgilbert@redhat.com> <0cb312f6-6c36-25ac-3c9a-0d9626abac56@redhat.com> <20180327142610.GC2837@work-vm> From: Jason Wang Message-ID: <962659da-2196-4dbf-9dd7-1a6f19105a86@redhat.com> Date: Wed, 28 Mar 2018 10:45:08 +0800 MIME-Version: 1.0 In-Reply-To: <20180327142610.GC2837@work-vm> Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH 0/3] e1000 migration changes for 2.12 List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: "Dr. David Alan Gilbert" Cc: qemu-devel@nongnu.org, eswierk@skyportsystems.com, quintela@redhat.com, peterx@redhat.com On 2018=E5=B9=B403=E6=9C=8827=E6=97=A5 22:26, Dr. David Alan Gilbert wrot= e: > * Jason Wang (jasowang@redhat.com) wrote: >> >> On 2018=E5=B9=B403=E6=9C=8827=E6=97=A5 19:34, Dr. David Alan Gilbert (= git) wrote: >>> From: "Dr. David Alan Gilbert" >>> >>> Hi Ed, Jason, >>> This set of patches change the e1000 migration code to make >>> it easier to keep with compatibility with older versions in backwards >>> migration; but I do need some advice whether I need to do more as we= ll. >>> >>> I think the first and second patch are fairly uncontrovercial and I >>> would like them for 2.12, since it'll make any future changes easier. >>> The third one changes the default behaviour, so again I'd prefer it b= ut >>> lets see what you think. >> The patches looks good to me. So for the changes of default behavior, = did >> you mean we can make the migration to older versions work? > Right. > >>> My question however, without knowing the internals of the e1000, is >>> whether when ommitting the subsection, should the code in 2.12 be >>> changing the data it sends back in the main section of data? >> I'm not sure I get the meaning here. But it looks to me turning it off= for >> old machine types makes sense, otherwise, management need to set it >> explicitly. > OK, let me expand the question a bit. > If I understand correctly the d6244b description, there's two pieces of > state (TSO and non-TSO) where there used to be only one. > Now, with the new format we migrate both pieces of state, but lets thin= k > about what happens if we have to migrate only one piece. > > a) 2.11->2.11 migrated the only piece it knew about; so I guess the onl= y > problem was really the UDP corruption mentioned in the commit. > > b) 2.11->2.12 - it receives the wrongly merged piece of state and puts = it > in - well which of the two states does it load it into? What's the > effect of that? So I think what we can do in this case is keep the (buggy) behavior=20 here. E.g put the state into both props and tso_props. > > c) 2.12(+my mod)->2.11 ok, so 2.12 will have filled in both sets of sta= te > internally; but now it's only going to send one of them over to 2.11 - > which one gets sent to 2.11? Is it the one that 2.11 is expecting? Then we can keep the behavior of 2.11 when migrate_tso_props (probably=20 need a better name) is false we will use props for all contexts. > > d) 2.12(+my mod)->2.12(+my mod) with an old machine type, again we're o= nly > going to send one set of data (same as c) - but what's 2.12 going to > make of the one piece of state received? So if we do like above, guest will see buggy device after migration.=20 (But do we really care this case?). Thanks > > (b) is an existing question. > > Dave > >> Thanks >> >>> Dave >>> >>> >>> Dr. David Alan Gilbert (3): >>> e1000: Convert v3 fields to subsection >>> e1000: wire new subsection to property >>> e1000: Old machine types, turn new subsection off >>> >>> hw/net/e1000.c | 46 ++++++++++++++++++++++++++++++++++-------= ----- >>> include/hw/compat.h | 4 ++++ >>> 2 files changed, 38 insertions(+), 12 deletions(-) >>> > -- > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK