Re: [PATCH v5 8/8] libvduse: Add support for reconnecting

[Yongji Xie <xieyongji@bytedance.com>]

On Thu, May 19, 2022 at 5:44 PM Stefan Hajnoczi <stefanha@redhat.com> wrote:
>
> On Thu, May 19, 2022 at 04:25:13PM +0800, Yongji Xie wrote:
> > On Wed, May 18, 2022 at 10:03 PM Stefan Hajnoczi <stefanha@redhat.com> wrote:
> > >
> > > On Wed, May 04, 2022 at 03:40:51PM +0800, Xie Yongji wrote:
> > > > @@ -291,6 +294,15 @@ static int vduse_blk_exp_create(BlockExport *exp, BlockExportOptions *opts,
> > > >          return -ENOMEM;
> > > >      }
> > > >
> > > > +    vblk_exp->recon_file = g_strdup_printf("%s/vduse-blk-%s",
> > > > +                                           g_get_tmp_dir(), exp->id);
> > >
> > > g_get_tmp_dir() returns the $TMPDIR environment variable. This means
> > > exp->id must be unique across the host. Please document this.
> > >
> >
> > Now we also use exp->id as the name of vduse device which should also
> > be unique across the host. So I'm not sure if it's better to add a new
> > unique id for vduse since the exp->id is now used by all block
> > exports.
>
> Good point, I forgot that the VDUSE device name must be unique! It's a
> little more flexible to have a separate vduse-name= option which is used
> for the VDUSE device name and the reconnection filename, but also a
> little more typing. I'm not sure if it's necessary to separate the two
> concepts.
>

OK, let's reuse the exp->id and document this first. We can add this
new option if we need it in the future.

Thanks,
Yongji