From: Yoshiaki Tamura Subject: Re: [RFC][PATCH 12/13] Kemari: use signal to save qemu state for Kemari Date: Sat, 07 Mar 2009 01:20:27 +0900 Message-ID: <49B14D4B.7090300@lab.ntt.co.jp> References: <49B0B8DC.5000606@lab.ntt.co.jp> <49B0C6FF.8090903@lab.ntt.co.jp> <49B10B8D.6090505@eu.citrix.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <49B10B8D.6090505@eu.citrix.com> List-Unsubscribe: , List-Post: List-Help: List-Subscribe: , Sender: xen-devel-bounces@lists.xensource.com Errors-To: xen-devel-bounces@lists.xensource.com To: Stefano Stabellini Cc: "ospk-vm@lab.ntt.co.jp" , Ian Pratt , xen-devel , Ian Jackson , Keir Fraser List-Id: xen-devel@lists.xenproject.org Stefano Stabellini wrote: > Yoshiaki Tamura wrote: > >> This patch implements signal handler for saving QEMU status quickly. It also >> includes a feature borrowed from KVM-17, which saves QEMU status quickly >> without repositioning file offset by fseek. > > We got rid of the signal interface not so long ago because it doesn't > work for stubdoms. > > The new way of asking qemu to save the VM state is writing a command on > xenstore (look at xenstore.c:xenstore_process_dm_command_event). > Would that work for you? > > > xenstore is attractive because it can send more information between components than signals. However, we used the signal handler to avoid the latency of xenstore. Although we haven't measured the latency, we need a simple and quick interface to save the QEMU state because Kemari transfers the saved state on every synchronization which happens quite often. Have you measured, or any thoughts on the latency of xenstore compared to signals?