Re: [Qemu-devel] [PATCH] block: Add bdrv_forbid_ext_snapshots.

[Kevin Wolf]

Am 26.09.2013 um 04:01 hat Jeff Cody geschrieben:
> On Wed, Sep 25, 2013 at 04:23:22PM +0200, Benoît Canet wrote:
> > Drivers having a bs->file where set to recurse the call to their child.
> > Protocol and drivers designed to be on the bottom of the stack where set to 
> > allow
> > snapshots.
> > Future protocols like quorum where creating snapshots does not make sense
> > without block filters will be set to forbid snapshots.
> > 
> > Signed-off-by: Benoit Canet <address@hidden>

> > diff --git a/block.c b/block.c
> > index 4a98250..ff296df 100644
> > --- a/block.c
> > +++ b/block.c
> > @@ -4651,3 +4651,30 @@ int bdrv_amend_options(BlockDriverState *bs, 
> > QEMUOptionParameter *options)
> >      }
> >      return bs->drv->bdrv_amend_options(bs, options);
> >  }
> > +
> > +bool bdrv_is_ext_snapshot_forbidden(BlockDriverState *bs)
> > +{
> 
> I think either:
> A) Name this function bdrv_forbid_ext_snapshots(), or
> B) Name the BlockDriver function ptr to .bdrv_is_ext_snapshot_forbidden
> 
> The idea being that this function and the BlockDriver function ptr
> should have the same name (e.g. bdrv_has_zero_init, and
> bs->drv->bdrv_has_zero_init, etc..)

Yes, I agree, some consistent naming is desirable. I don't think
bdrv_forbid_ext_snapshots() is a good name, because it implies that
calling this function is what forbids the snapshot (i.e. an action
similar to adding a migration blocker), whereas in fact it just checks
whether snapshots are forbidden.

How about bdrv_ext_snapshot_allowed(), which avoid double negations when
we check for "not forbidden"? Or perhaps even bdrv_check_ext_snapshot(),
which would be a more generic name that could be extended to the
three-way distinction we intended to have in the end:

- External snapshots are forbidden
- May snapshot, but below this BDS (ask bs->file; this is for filters)
- Do the snapshot here

Kevin