Re: [PATCH] migration: stop tracking ram writes when cancelling background migration

[Peter Xu]

On Fri, May 26, 2023 at 01:59:08PM +0200, Fiona Ebner wrote:
> Currently, it is only done when the iteration finishes successfully.
> Not cleaning up the userfaultfd write protection can lead to
> symptoms/issues such as the process hanging in memmove or GDB not
> being able to attach.
> 
> Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>

Reviewed-by: Peter Xu <peterx@redhat.com>

> For the success case, the stuff in between the old call and new call
> site should not depend on tracking to already be stopped, right?

Yes I think so.  There's only device states to be flushed and no guest
memory should be touched.

Even if we'll touch them, since we've finished migrating all of them so
they're already unprotected anyway (actually, even during ram save it's
read-only here from bg thread), so not any problem I can see.

Thanks,

-- 
Peter Xu