Hi Leon,
> -----Original Message-----
> From: linux-rdma-owner@xxxxxxxxxxxxxxx <linux-rdma-
> owner@xxxxxxxxxxxxxxx> On Behalf Of Leon Romanovsky
> Sent: Wednesday, May 1, 2019 12:31 PM
> To: Jason Gunthorpe <jgg@xxxxxxxxxxxx>
> Cc: Doug Ledford <dledford@xxxxxxxxxx>; RDMA mailing list <linux-
> rdma@xxxxxxxxxxxxxxx>
> Subject: Re: [PATCH rdma-next] RDMA/device: Don't fire uevent before
> device is fully initialized
>
> On Wed, May 01, 2019 at 01:28:55PM +0000, Jason Gunthorpe wrote:
> > On Wed, May 01, 2019 at 08:46:19AM +0300, Leon Romanovsky wrote:
> > > From: Leon Romanovsky <leonro@xxxxxxxxxxxx>
> > >
> > > When the refcount is 0 the device is invisible to netlink. However
> > > in the patch below the refcount = 1 was moved to after the device_add().
> > > This creates a race where userspace can issue a netlink query after
> > > the
> > > device_add() event and not see the device as visible.
> > >
> > > Ensure that no uevent is fired before device is fully registered.
> > >
> > > Fixes: d79af7242bb2 ("RDMA/device: Expose ib_device_try_get(()")
> > > Signed-off-by: Leon Romanovsky <leonro@xxxxxxxxxxxx>
> > > drivers/infiniband/core/device.c | 12 ++++++++++++
> > > 1 file changed, 12 insertions(+)
> > >
> > > diff --git a/drivers/infiniband/core/device.c
> > > b/drivers/infiniband/core/device.c
> > > index 8ae4906a60e7..4cdc8588df7f 100644
> > > +++ b/drivers/infiniband/core/device.c
> > > @@ -808,6 +808,7 @@ static int add_one_compat_dev(struct ib_device
> *device,
> > > cdev->dev.release = compatdev_release;
> > > dev_set_name(&cdev->dev, "%s", dev_name(&device->dev));
> > >
> > > + dev_set_uevent_suppress(&device->dev, true);
> > > ret = device_add(&cdev->dev);
> > > if (ret)
> > > goto add_err;
> >
> > compat devices definitely should not be doing this..
>
> Why? They have the same problematic pattern:
> device_add -> ..... -> enable_device.
>
compat devices are added in enable_device_and_get() after refcount is set.
ib_register_device()
enable_device_and_get()
refcount=2
[...]
add_compat_devs()
