On Tue, May 24, 2016 at 09:44:43AM -0700, Bart Van Assche wrote:
> On 05/24/2016 09:34 AM, Greg KH wrote:
> > On Tue, May 24, 2016 at 09:25:05AM -0700, Bart Van Assche wrote:
> > > > +static inline long h_send_crq(struct ibmvscsis_adapter *adapter,
> > > > + u64 word1, u64 word2)
> > > > +{
> > > > + long rc;
> > > > + struct vio_dev *vdev = adapter->dma_dev;
> > > > +
> > > > + pr_debug("ibmvscsis: ibmvscsis_send_crq(0x%x, 0x%016llx,
> > > > 0x%016llx)\n",
> > > > + vdev->unit_address, word1, word2);
> > > > +
> > >
> > > As Joe Perches already asked, please define pr_fmt() instead of including
> > > the kernel module name in every pr_debug() statement.
> >
> > Even better, as this is a driver, it should be using dev_*() calls
> > instead of pr_*() calls to properly identify the device and driver that
> > is making the message. No driver should be using pr_*() except in
> > _very_ limited usages.
>
> Hi Greg,
>
> The reason I recommended pr_debug() is because ibmvscsis is a LIO target
> driver and I don't think a struct device is associated with a LIO target
> driver. See e.g. target_register_template() in
> drivers/target/target_core_configfs.c.
That seems messed up, there should be a struct device somewhere to
use...
