On Tue, 7 Jun 2011 18:59:12 +0530
Amit Shah <amit.s...@redhat.com> wrote:
> On (Mon) 06 Jun 2011 [11:38:03], Luiz Capitulino wrote:
> > On Mon, 6 Jun 2011 17:10:32 +0530
> > Amit Shah <amit.s...@redhat.com> wrote:
> >
> > > On (Fri) 03 Jun 2011 [16:03:57], Luiz Capitulino wrote:
> > >
> > > > +static int tray_open(const char *device, int remove, int force)
> > > > +{
> > > > + BlockDriverState *bs;
> > > > +
> > > > + bs = bdrv_removable_find(device);
> > > > + if (!bs) {
> > > > + return -1;
> > > > + }
> > > > +
> > > > + if (bdrv_eject(bs, 1, force) < 0) {
> > > > + /* FIXME: will report undefined error in QMP */
> > > > + return -1;
> > > > + }
> > > > +
> > > > + if (remove) {
> > > > + bdrv_close(bs);
> > > > + }
> > > > +
> > > > + return 0;
> > > > +}
> > >
> > > What's the reason to tie the 'remove' with tray open?
> >
> > In my first try I had a command called 'blockdev-media-remove', but then
> > I had the impression that I was going too far as the only reason a client
> > would ever want to open the tray is to remove the media.
>
> Not necessary -- CD/DVD writers eject and reload trays after erasing
> media -- at least they used to.
But that's not done by the VM's user/client.
I think I'll add it anyway, as it's what people seem to expect from an
API like this.
> > > Won't it be
> > > simpler to have it separated out, perhaps a 'change' event instead of
> > > 'insert' that can accept NULL which means just remove medium?
> >
> > You meant 'command' instead of 'event', right?
> >
> > I don't think a change command makes sense, because it's just a shortcut
> > to open/remove/insert/close.
>
> Yes, command, sorry.
>
> And by 'change' I don't mean the current monitor change command --
> that's a badly-named one.
>
> By change I mean just that -- replace the media. And that should
> succeed only if tray is open. And tray remains open after the change.
The same thing can be done with blockdev-media-remove and
blockdev-media-insert, so I don't think this adds value.
