On Fri, 2015-05-08 at 22:14 +0100, Andrew Cooper wrote:
> There are some records which should only be sent once in the stream, and not
> repeated for each checkpoint. {start,end}_of_stream() become per-checkpoint,
> and a new start_of_stream() is introduced.
>
> There is no resulting change record order, but the X86_PV_INFO and
> X86_PV_P2M_FRAMES records are positively identified as once per stream, rather
> than once per checkpoint.
In the case of the latter judging from the comments this is currently an
implementation detail and in principal we want to be able to resend the
p2m frames as well?
[...]
> diff --git a/tools/libxc/xc_sr_save_x86_pv.c b/tools/libxc/xc_sr_save_x86_pv.c
> index a668221..f63f40b 100644
> --- a/tools/libxc/xc_sr_save_x86_pv.c
> +++ b/tools/libxc/xc_sr_save_x86_pv.c
> @@ -816,6 +816,12 @@ static int x86_pv_start_of_stream(struct xc_sr_context
> *ctx)
> if ( rc )
> return rc;
>
> + /*
> + * Ideally should be able to change during migration. Currently
> + * corruption will occur if the contents or location of the P2M changes
> + * during the live migration loop. If one is very lucky, the breakage
> + * will not be subtle.
> + */
> rc = write_x86_pv_p2m_frames(ctx);
> if ( rc )
> return rc;
This is the comment I'm basing the above on.
Ian.
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel
