Re: [Qemu-devel] [RFC 10/10] QMP/HMP: change: Use QMP tray commands

[Amit Shah]

On (Mon) 06 Jun 2011 [11:45:06], Luiz Capitulino wrote:
> On Mon, 6 Jun 2011 17:18:31 +0530
> Amit Shah <address@hidden> wrote:
> 
> > On (Fri) 03 Jun 2011 [16:04:02], Luiz Capitulino wrote:
> > > This commit rewrites change in terms of blockdev-tray-open,
> > > blockdev-media-insert and blockdev-tray-close.
> > > 
> > > There should be no visible changes in HMP or QMP, except that
> > > the use of this command causes the BLOCK_TRAY_OPEN *and*
> > > BLOCK_TRAY_CLOSE events to emitted.
> > > 
> > > Signed-off-by: Luiz Capitulino <address@hidden>
> > > ---
> > >  blockdev.c |   43 +++++++++----------------------------------
> > >  1 files changed, 9 insertions(+), 34 deletions(-)
> > > 
> > > diff --git a/blockdev.c b/blockdev.c
> > > index 36c56fd..e9edcb3 100644
> > > --- a/blockdev.c
> > > +++ b/blockdev.c
> > > @@ -648,23 +648,6 @@ out:
> > >      return ret;
> > >  }
> > >  
> > > -static int eject_device(Monitor *mon, BlockDriverState *bs, int force)
> > > -{
> > > -    if (!force) {
> > > -        if (!bdrv_is_removable(bs)) {
> > > -            qerror_report(QERR_DEVICE_NOT_REMOVABLE,
> > > -                           bdrv_get_device_name(bs));
> > > -            return -1;
> > > -        }
> > > -        if (bdrv_is_locked(bs)) {
> > > -            qerror_report(QERR_DEVICE_LOCKED, bdrv_get_device_name(bs));
> > > -            return -1;
> > > -        }
> > > -    }
> > > -    bdrv_close(bs);
> > > -    return 0;
> > > -}
> > > -
> > >  static BlockDriverState *bdrv_removable_find(const char *name)
> > >  {
> > >      BlockDriverState *bs;
> > > @@ -819,30 +802,22 @@ int do_change_block(Monitor *mon, const char 
> > > *device,
> > >                      const char *filename, const char *fmt)
> > >  {
> > >      BlockDriverState *bs;
> > > -    BlockDriver *drv = NULL;
> > > -    int bdrv_flags;
> > >  
> > > -    bs = bdrv_find(device);
> > > -    if (!bs) {
> > > -        qerror_report(QERR_DEVICE_NOT_FOUND, device);
> > > +    if (tray_open(device, 1, 0) < 0) { /* XXX: should we force? */
> > >          return -1;
> > >      }
> > > -    if (fmt) {
> > > -        drv = bdrv_find_whitelisted_format(fmt);
> > > -        if (!drv) {
> > > -            qerror_report(QERR_INVALID_BLOCK_FORMAT, fmt);
> > > -            return -1;
> > > -        }
> > > -    }
> > > -    if (eject_device(mon, bs, 0) < 0) {
> > > +
> > > +    if (media_insert(device, filename, fmt) < 0) {
> > >          return -1;
> > >      }
> > > -    bdrv_flags = bdrv_is_read_only(bs) ? 0 : BDRV_O_RDWR;
> > > -    bdrv_flags |= bdrv_is_snapshot(bs) ? BDRV_O_SNAPSHOT : 0;
> > > -    if (bdrv_open(bs, filename, bdrv_flags, drv) < 0) {
> > > -        qerror_report(QERR_OPEN_FILE_FAILED, filename);
> > > +
> > > +    if (tray_close(device) < 0) {
> > > +        /* XXX: open the tray and remove the media? */
> > >          return -1;
> > >      }
> > 
> > So if the new medium points to a non-existent file, close will fail,
> 
> media_insert() will and tray_close() won't be called. That's what you
> meant, right?

Yes.

> > and leave the tray in the open state.  A next change command will fail
> > because the tray is already open.  Not a good thing.
> 
> tray_open() shouldn't fail if the tray is already open.

Ah, that's right.

                Amit