[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 0/3] Migration on QEMU upstream



On Wed, 2011-11-23 at 17:40 +0000, Anthony PERARD wrote:
> Hi all,
> 
> This patch series introduce migration with QEMU upstream.
> 
> A patch series for QEMU will be sent later meant to "fix" the default behavior
> of QEMU during save/restore.
> 
> There is two new QMP commands implemented:
>   - getfd: This give a file descriptor to QEMU through the unix socket.

This sounds more like a setfd command.

>   - migration: This ask QEMU to save its states in a fd previously sent.

Why can the fd not be included in this message directly? How is a given
{s,g}etfd call associated with the migration command? What happens if
some other command also wants to receive an fd in the future?

I presume that a bunch of this will become  more obvious when the qemu
side is posted, there's a qemu protocol spec in that right?

> In order to call "getfd", an alternative qmp_send have been implemented in
> libxl.

You could also have added optional msg_control{len} parameters to the
existing command. I don't know if that is better though.

Ian.

> 
> Regards,
> 
> 
> Anthony PERARD (3):
>   libxl_qmp: Cut qmp_send function.
>   libxl_qmp: Command migrate.
>   libxl: Introduce migrate with the new QEMU.
> 
>  tools/libxl/libxl_dm.c       |    6 +-
>  tools/libxl/libxl_dom.c      |   28 +++++++-
>  tools/libxl/libxl_internal.h |    2 +
>  tools/libxl/libxl_qmp.c      |  161 +++++++++++++++++++++++++++++++++++++----
>  4 files changed, 176 insertions(+), 21 deletions(-)
> 



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.