Am 22.02.2014 um 01:01 hat Eric Blake geschrieben:
> On 02/21/2014 03:11 PM, Kevin Wolf wrote:
> > Instead of making the backing file contents visible again after a discard
> > request, set the zero flag if possible (i.e. on version >= 3).
> >
> > Signed-off-by: Kevin Wolf <kw...@redhat.com>
> > Reviewed-by: Eric Blake <ebl...@redhat.com>
> > ---
> > block/qcow2-cluster.c | 22 ++++++++++++++++++++--
> > tests/qemu-iotests/046 | 18 ++++++++++++++----
> > 2 files changed, 34 insertions(+), 6 deletions(-)
> >
> > diff --git a/block/qcow2-cluster.c b/block/qcow2-cluster.c
> > index c57f39d..36c1bed 100644
> > --- a/block/qcow2-cluster.c
> > +++ b/block/qcow2-cluster.c
> > @@ -1367,13 +1367,31 @@ static int discard_single_l2(BlockDriverState *bs,
> > uint64_t offset,
> > uint64_t old_offset;
> >
> > old_offset = be64_to_cpu(l2_table[l2_index + i]);
> > - if ((old_offset & L2E_OFFSET_MASK) == 0) {
> > +
> > + /*
> > + * Make sure that a discarded area reads back as zeroes for v3
> > images
> > + * (we cannot do it for v2 without actually writing a zero-filled
> > + * buffer). We can skip the operation if the cluster is already
> > marked
> > + * as zero, or if it's unallocated and we don't have a backing
> > file.
> > + *
>
> Possible idea for a followup:
>
> Is it possible with v2 images to write a single sector/cluster with all
> 0s, and to make all other writes of all-0 data bump the reference count
> of the magic known-zero cluster, so that we are at least conserving disk
> space by heavily reusing the known cluster?No, that doesn't work because when you have multiple virtual clusters pointing to the same physical cluster in a single snapshot, you can't guarantee any more that QCOW_OFLAG_COPIED is set correctly. Ironically, I guess you could still do it in v3 when you set the dirty flag of lazy refcounting, but we don't have that in v2 either... > Then again, we may be reaching the point where no one cares about v2 > images as much. Right, that too. Kevin
pgph8UYeq_xd9.pgp
Description: PGP signature
