Re: [Qemu-devel] [PATCH v3 11/29] vhost+postcopy: Transmit 'listen' to client

[Peter Xu]

On Tue, Mar 06, 2018 at 11:20:56AM +0000, Dr. David Alan Gilbert wrote:
> * Peter Xu (address@hidden) wrote:
> > On Mon, Mar 05, 2018 at 05:42:42PM +0000, Dr. David Alan Gilbert wrote:
> > > * Peter Xu (address@hidden) wrote:
> > > > On Fri, Feb 16, 2018 at 01:16:07PM +0000, Dr. David Alan Gilbert (git) 
> > > > wrote:
> > > > 
> > > > [...]
> > > > 
> > > > >  typedef struct VuVirtqElement {
> > > > > diff --git a/docs/interop/vhost-user.txt b/docs/interop/vhost-user.txt
> > > > > index 621543e654..bdec9ec0e8 100644
> > > > > --- a/docs/interop/vhost-user.txt
> > > > > +++ b/docs/interop/vhost-user.txt
> > > > > @@ -682,6 +682,12 @@ Master message types
> > > > >        the slave must open a userfaultfd for later use.
> > > > >        Note that at this stage the migration is still in precopy mode.
> > > > >  
> > > > > + * VHOST_USER_POSTCOPY_LISTEN
> > > > > +      Id: 27
> > > > > +      Master payload: N/A
> > > > > +
> > > > > +      Master advises slave that a transition to postcopy mode has 
> > > > > happened.
> > > > 
> > > > Could we add something to explain why this listen needs to be
> > > > broadcasted to clients?  Since I failed to find it out quickly
> > > > myself. :(
> > > 
> > > I've changed this to:
> > > 
> > >  * VHOST_USER_POSTCOPY_LISTEN
> > >       Id: 29
> > >       Master payload: N/A
> > > 
> > >       Master advises slave that a transition to postcopy mode has 
> > > happened.
> > >       The slave must ensure that shared memory is registered with 
> > > userfaultfd
> > >       to cause faulting of non-present pages.
> > 
> > But shouldn't this be assured by the SET_MEM_TABLE call?
> 
> Yes, it is the set_mem_table that does the register - but it only
> registers it with userfaultfd if it's received a 'listen' notification,
> otherwise it assumes it's a normal precopy migration.

I think I got the picture now.  Please add this after the enhanced
document:

Reviewed-by: Peter Xu <address@hidden>

Thanks,

-- 
Peter Xu