[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] VM checkpoint image format
On Sun, 22 Dec 2013, Andrew Cooper wrote: On 21/12/2013 16:31, Prateek Sharma wrote:Hello all, Is the VM-save image format specified anywhere? (the file written to by xc_domain_save). I am looking for what fields are saved in what order. Or is the xc_domain_save function itself the specification? Can I change the order of fields saved (thus changing the on-disk image format) and expect no breakages? Ofcourse, I will need to change the xc_domain_restore function as well. Thanks! --PrateekThere is no documentation that I am aware of, other than the code itself, and the code is a rats nest. The format itself is quite lacking in terms of extensibility, and makes some assumptions which are not actually accurate. In my copious free time (of which I have none), or when XenServer decided that it actually wants migration which works in combination with ballooning (or any of the other more interesting memory modes), I have a plan to make a "migration protocol v2" which shall be formally specified in a document, and have ample room for forward extensibility in a backward compatible manner. As for changing the order of things. So long as your xc_domain_save() and xc_domain_restore() have similar expectations of what order to put/find chunks of the stream, it should work fine. It would help if you described what changes you wish to make. We might be able to point you at the correct area. ~Andrew I'm attempting to implement VM lazy-restore. To achieve that, it would be nice if page-table pages, vcpu state, GDT pages, etc are located _before_ the rest of the pages. Is there some implicit dependency in the ordering of pages and the VCPU context? Is there a way to access (and hence save) all the page-table pages of a guest? Thanks! --Prateek _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxx http://lists.xen.org/xen-devel
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |