Re: [Qemu-devel] [PATCH 2/3] migration: Remove use of old MigrationParams

[Dr. David Alan Gilbert]

* Juan Quintela (address@hidden) wrote:
> "Dr. David Alan Gilbert" <address@hidden> wrote:
> > * Juan Quintela (address@hidden) wrote:
> >> "Dr. David Alan Gilbert" <address@hidden> wrote:
> >> > * Juan Quintela (address@hidden) wrote:
> >> >> Markus Armbruster <address@hidden> wrote:
> >> >> > Juan Quintela <address@hidden> writes:
> >> >> >
> >> >> >> Eric Blake <address@hidden> wrote:
> >> >> 
> >> >> 
> >> >> >>> Or is the proposal that we are also going to simplify the QMP 
> >> >> >>> 'migrate'
> >> >> >>> command to get rid of crufty parameters?
> >> >> >>
> >> >> >> I didn't read it that way, but I would not oppose O:-)
> >> >> >>
> >> >> >> Later, Juan.
> >> >> >
> >> >> > I'm not too familiar with this stuff, so please correct my
> >> >> > misunderstandings.
> >> >> >
> >> >> > "Normal" migration configuration is global state, i.e. it applies to 
> >> >> > all
> >> >> > future migrations.
> >> >> >
> >> >> > Except the "migrate" command's flags apply to just the migration 
> >> >> > kicked
> >> >> > off by that command.
> >> >> >
> >> >> > QMP command "migrate" has two flags "blk" (HMP: -b) and "inc" (HMP: 
> >> >> > -i).
> >> >> > !blk && inc makes no sense and is silently treated like !blk && !inc.
> >> >> >
> >> >> > There's a third flag "detach" (HMP: -d), but it does nothing in QMP.
> >> >> 
> >> >> As qmp command is asynchronous, you can think that -d is *always* on in
> >> >> QMP O:-)
> >> >> 
> >> >> > You'd like to deprecate these flags in favour of "normal" 
> >> >> > configuration.
> >> >> > However, we need to maintain QMP backward compatibility at least for a
> >> >> > while.  HMP backward compatibility is nice to have, but not required.
> >> >> >
> >> >> > First step is to design the new interface you want.  Second step is to
> >> >> > figure out backward compatibility.
> >> >> >
> >> >> > The new interface adds a block migration tri-state (off,
> >> >> > non-incremental, incremental) to global state, default off.  Whether
> >> >> > it's done as two bools or an enum of three values doesn't matter here.
> >> >> 
> >> >> Tristates will complicate it.  I still think that:
> >> >> 
> >> >> - capability: block_migration
> >> >> - parameter: block_shared
> >> >> 
> >> >> Makes more sense, no?
> >> >
> >> > I don't understand what making block_shared a parameter gives you as
> >> > opposed to simply having two capabilities.
> >> >
> >> > (And how did we get 'shared'? We started off with block & incremental)
> >> 
> >> The variables on MigrationParams:
> >> 
> >> struct MigrationParams {
> >>     bool blk;
> >>     bool shared;
> >> };
> >> 
> >> 
> >> I can move to incremental.  I am not sure which one is clearer.
> >> 
> >> The advantage of having shared as a parameter is that we forget about
> >> all this dependency bussiness.  Is the same than compression_threads
> >> paramter, you setup to whichever value that you want.  But you don't get
> >> compression_threads until you set the compress capability.
> >> 
> >> So, in this case we will have:
> >> 
> >> block capability: Are we using block migration or not
> >> block-incremental parameter: If we are using block migration, are we
> >>       using incremental copying of the block layer?
> >
> > If it's still a boolean why does having it as a parameter remove the
> > dependency?
> 
> Forget -b/-i.
> 
> migration_set_parameter compression_threads 8
> 
> migrate <foo>
> 
> We don't use compression_threads at all
> 
> migrate_set_capability compress
> 
> migrate <foo>
> 
> Now, we use compression threads.
> 
> So, compression_threads parameter is a parameter that is only used when
> compress capability is enabled.
> 
> Same for block migration.  Block_incremental parameter is used only when
> block migration capability is setup.  No dependency check needed at all.
> 
> Or I am losing something obvious here?

Ah, you've made up a new rule - I don't think it's a bad rule
but is it true? Do we always enable a capability before we use a
parameter? I don't think so - I think the tls parameters don't have
a capability.
My previous rule was just that if it was a bool it was a capability
and you can have whatever dependencies you like there - or none.

Dave

> Later, Juan.
--
Dr. David Alan Gilbert / address@hidden / Manchester, UK