Re: [Qemu-devel] [PATCH] block: change default of .has_zero_init to 0

[Kevin Wolf <kwolf@redhat.com>]

Am 28.06.2013 um 10:16 hat Peter Lieven geschrieben:
> On 28.06.2013 10:06, Kevin Wolf wrote:
> >Am 27.06.2013 um 15:52 hat Peter Lieven geschrieben:
> >>.has_zero_init defaults to 1 for all formats and protocols.
> >>
> >>this is a dangerous default since this means that all
> >>new added drivers need to manually overwrite it to 0 if
> >>they do not ensure that a device is zero initialized
> >>after bdrv_create().
> >>
> >>if a driver needs to explicitly set this value to
> >>1 its easier to verify the correctness in the review process.
> >>
> >>during review of the existing drivers it turned out
> >>that ssh and gluster had a wrong default of 1.
> >>both protocols support host_devices as backend
> >>which are not by default zero initialized. this
> >>wrong assumption will lead to possible corruption
> >>if qemu-img convert is used to write to such a backend.
> >>
> >>a similar problem with the wrong default existed for
> >>iscsi mose likely because the driver developer did
> >>oversee the default value of 1.
> >>
> >>Signed-off-by: Peter Lieven <pl@kamp.de>
> >>---
> >>  block.c               |    8 +++++++-
> >>  block/qcow.c          |    1 +
> >>  block/qcow2.c         |    1 +
> >>  block/qed.c           |    1 +
> >>  block/raw-posix.c     |   10 +---------
> >>  block/raw-win32.c     |    7 +------
> >>  block/rbd.c           |    1 +
> >>  block/sheepdog.c      |    1 +
> >>  block/vdi.c           |    1 +
> >>  block/vmdk.c          |    1 +
> >>  include/block/block.h |    1 +
> >>  11 files changed, 17 insertions(+), 16 deletions(-)
> >You forgot cow, which is also a simple case that can be handled in this
> >patch.
> ups.
> >
> >vpc is still easy, but a bit more complicated than a constant return 1,
> >because it depends on the subformat whether a new image will inherit the
> >has_zero_init property from the underlying storage or whether it always
> >produces zeros (for VHD_FIXED type images, it's basically raw + footer).
> >I'll send a separate patch for this.
> shall I leave this to the new 0 default until your patch is ready?

Yes, please leave vpc alone. I guess my patch will be queued before
yours anyway. ;-)

> >A similar situation exists for vmdk, I think, just that the difference
> >can be per extent there. I guess we need to return 0 if one of the
> >extents is flat and has an underlying storage returning 0. I'll leave
> >this part to Fam. For now, please remove the bdrv_has_zero_init_1 for
> >vmdk from this patch as it's unsafe.
> ok I will mention that the value is changed for vmdk (and vpc) in the
> commit message.

Good idea for vmdk. For vpc, don't mention it, because see above.

Kevin