From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hongyang Yang Subject: Re: [PATCH 07/10 V7] libxl: use the API to setup/teardown network buffering [and 1 more messages] Date: Tue, 13 May 2014 09:41:34 +0800 Message-ID: <5371784E.80003@cn.fujitsu.com> References: <1397540297-32184-1-git-send-email-yanghy@cn.fujitsu.com> <1392023972-24675-1-git-send-email-laijs@cn.fujitsu.com> <1392023972-24675-8-git-send-email-laijs@cn.fujitsu.com> <21268.49462.367220.516477@mariner.uk.xensource.com> <1397540297-32184-6-git-send-email-yanghy@cn.fujitsu.com> <21335.58413.243916.380152@mariner.uk.xensource.com> <21347.49908.432388.413355@mariner.uk.xensource.com> <5369C7D5.5040803@cn.fujitsu.com> <21360.51750.941378.425720@mariner.uk.xensource.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; Format="flowed" Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <21360.51750.941378.425720@mariner.uk.xensource.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xen.org Errors-To: xen-devel-bounces@lists.xen.org To: Ian Jackson , rshriram@cs.ubc.ca Cc: Lai Jiangshan , FNST-Wen Congyang , Stefano Stabellini , Andrew Cooper , Jiang Yunhong , Dong Eddie , "xen-devel@lists.xen.org" , Roger Pau Monne , Ian Campbell List-Id: xen-devel@lists.xenproject.org On 05/12/2014 09:18 PM, Ian Jackson wrote: > Shriram Rajagopalan writes ("Re: [Xen-devel] [PATCH 07/10 V7] libxl: use the API to setup/teardown network buffering [and 1 more messages]"): >> On Wed, May 7, 2014 at 12:42 AM, Hongyang Yang wrote: > ... >> This patchset is based on the current remus implementation >> (without netbuffer) witch is integrated into suspend/resume >> code, if as you suggested we pick option 1, the whole remus >> structure needs refactoring. we're working on it, may take >> quiet a while. > ... >> Just to make sure I am on the same page with you folks, >> Ian, when you talked about the two options (1) & (2), did you mean the >> entire remus implementation inside libxl or just "this" setup/teardown code >> base? > > I think it is OK for some of the Remus code to fit into the rest of > libxl code via method (1) and some via method (2). > > But any particular subfunction should do strictly one or the other. I > think, overall, (1) is better. It would be nice to do (1) everywhere. > But a series can be acceptable even if it contains some (2). > Thanks for the reply, that's more clear and solved my doubts too. Thanks, Yang. > Thanks, > Ian. > > For reference, the (1) and (2) we are referring to are these: > > 1. Make the remus part of this be a fully self-contained standard > asynchronous callback-based suboperation, like libxl__xswait, > libxl__bootloader, et al. > > In this case you should rigorously follow the existing patterns, > defining a clear interface between the two parts, providing a > callback function set by the caller, etc. > > 2. Integrate the remus part into the suspend/resume code in an > ad hoc fashion, with extremely clear comments everywhere about the > expected interface, and no extraneous moving parts. > > I'm sure you know that but I wanted to save future readers, and anyone > not following this in detail, the effort of chasing it down. > . >