RE: [PATCH] vmbus: use put_device() if device_register fail

[Michael Kelley (EOSG)]

> -----Original Message-----
> From: linux-kernel-owner@xxxxxxxxxxxxxxx <linux-kernel-owner@xxxxxxxxxxxxxxx> On Behalf
> Of KY Srinivasan
> Sent: Sunday, March 18, 2018 8:02 PM
> To: Arvind Yadav <arvind.yadav.cs@xxxxxxxxx>; Stephen Hemminger
> <sthemmin@xxxxxxxxxxxxx>; Haiyang Zhang <haiyangz@xxxxxxxxxxxxx>
> Cc: devel@xxxxxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
> Subject: RE: [PATCH] vmbus: use put_device() if device_register fail
> 
> > -----Original Message-----
> > From: devel <driverdev-devel-bounces@xxxxxxxxxxxxxxxxxxxxxx> On Behalf
> > Of Arvind Yadav
> > Sent: Saturday, March 17, 2018 11:48 PM
> > To: Stephen Hemminger <sthemmin@xxxxxxxxxxxxx>; Haiyang Zhang
> > <haiyangz@xxxxxxxxxxxxx>
> > Cc: devel@xxxxxxxxxxxxxxxxxxxxxx; linux-kernel@xxxxxxxxxxxxxxx
> > Subject: [PATCH] vmbus: use put_device() if device_register fail
> >
> > if device_register() returned an error. Always use put_device()
> > to give up the reference initialized.
> >
> > Signed-off-by: Arvind Yadav <arvind.yadav.cs@xxxxxxxxx>
> > ---
> >  drivers/hv/vmbus_drv.c | 1 +
> >  1 file changed, 1 insertion(+)
> >
> > diff --git a/drivers/hv/vmbus_drv.c b/drivers/hv/vmbus_drv.c
> > index bc65c4d..25da2f3 100644
> > --- a/drivers/hv/vmbus_drv.c
> > +++ b/drivers/hv/vmbus_drv.c
> > @@ -1358,6 +1358,7 @@ int vmbus_device_register(struct hv_device
> > *child_device_obj)
> >  	ret = device_register(&child_device_obj->device);
> >  	if (ret) {
> >  		pr_err("Unable to register child device\n");
> > +		put_device(&child_device_obj->device);
> 
> If the registration failed; we would not have acquired the reference on the device and so
> we should not be dropping the reference in the failure path.

Looking at the code for device_register() and its constituent parts
device_initialize() and device_add(), Arvind is correct.  device_initialize()
creates the object with a reference count of 1.  device_add() does not
decrement that reference count or free the object, even if it fails.  The
comments for device_register() call this out as well.

Michael  

> 
> K. Y
> >  		return ret;
> >  	}
> >
> > --
> > 2.7.4