> -----Original Message----- > From: virtualization-bounces@xxxxxxxxxxxxxx > [mailto:virtualization-bounces@xxxxxxxxxxxxxx] On Behalf Of Andi Kleen > Sent: Wednesday, September 28, 2005 12:59 PM > To: virtualization@xxxxxxxxxxxxxx; Martin J. Bligh > Cc: cgriffin@xxxxxxxxxx; Chris Wright; steven@xxxxxxxxxxxxx; > Walker, Bruce J (HP-Labs) > Subject: Re: Phone meeting about kernel virtualization hooks > > On Wednesday 28 September 2005 20:41, Martin J. Bligh wrote: > > > > > Would be splendid if we can keep that up-to-date with every > major release, > > plus -rc releases if possible. I know it's a lot of work ... > > It's one of the problems imho that the xen tree is always > based on a ancient > tree (last release). It would be much easier if it tracked > mainline more > closely. Agreed but I'd suggest that this is a symptom of the real problem: Updating the Xen -sparse tree is still a very manual process. For someone who's done it a few times (e.g. Keir or Ian), it "only" takes a couple hours. As more of the preparatory patches get into linux, the effort will go down, but it is still a manual process. I am intrigued by Rik's proposal of creating a daily mirror that restructures the directories; if this could also automagically create patches** and attempt to apply them to several Linux trees (e.g. last dot release, current -rcX, current -mm), the process would very quickly uncover potential Xenlinux problem spots and encourage fixes to them. ** Per my previous message, a three phase patch would be best: First a patch for existing Linux files, then a patch for new files in existing Linux directories, then a patch for new files in new directories.