2014-02-13 8:19 GMT+08:00 Eric Blake <ebl...@redhat.com>:
> On 02/11/2014 11:33 PM, Chunyan Liu wrote:
> > cow.c: replace QEMUOptionParameter with QemuOpts
> >
> > Signed-off-by: Dong Xu Wang <wdon...@linux.vnet.ibm.com>
> > Signed-off-by: Chunyan Liu <cy...@suse.com>
> > ---
> > block/cow.c | 46 ++++++++++++++++++++++------------------------
> > 1 files changed, 22 insertions(+), 24 deletions(-)
> >
>
> > +static QemuOptsList cow_create_opts = {
> > + .name = "cow-create-opts",
> > + .head = QTAILQ_HEAD_INITIALIZER(cow_create_opts.head),
> > + .desc = {
> > + {
> > + .name = BLOCK_OPT_SIZE,
> > + .type = QEMU_OPT_SIZE,
> > + .help = "Virtual disk size"
>
> Oh, these QemuOpts really ARE constant strings in the general case, and
> passing these strings to free() could be a disaster. I hope we're okay
> in what we do; in fact, maybe it's worth a patch to QemuOptsList that
> adds a bool record of whether the list is dynamically allocated (false
> by default for all static initialization, and true when malloc'ing a
> list) and then asserting that attempts to free (parts of) a list are
> only done on a dynamically allocated list.
>
> Well, create_opts specified here won't be freed by free() in existing code
I thi
nk. Those using it would malloc and memcpy from it, and that would be
freed. The
same way as existing create_options handling.
> > @@ -414,14 +412,14 @@ static BlockDriver bdrv_cow = {
> > .bdrv_probe = cow_probe,
> > .bdrv_open = cow_open,
> > .bdrv_close = cow_close,
> > - .bdrv_create = cow_create,
> > + .bdrv_create2 = cow_create,
>
> Might as well keep alignment looking nice.
>
> --
> Eric Blake eblake redhat com +1-919-301-3266
> Libvirt virtualization library http://libvirt.org
>
>
