Re: [PATCH v1 01/12] gna: add driver module

[Greg Kroah-Hartman]

On Mon, Mar 01, 2021 at 11:18:59AM +0100, Maciej Kwapulinski wrote:
> 
> Andy Shevchenko <andy.shevchenko@xxxxxxxxx> writes:
> 
> > On Tue, Feb 16, 2021 at 6:11 PM Maciej Kwapulinski
> > <maciej.kwapulinski@xxxxxxxxxxxxxxx> wrote:
> >>
> ....
> >> +static int __init gna_drv_init(void)
> >> +{
> >> +       int ret;
> >> +
> >> +       mutex_init(&gna_drv_priv.lock);
> >> +
> >> +       gna_class = class_create(THIS_MODULE, "gna");
> >> +       if (IS_ERR(gna_class)) {
> >> +               pr_err("class device create failed\n");
> >> +               return PTR_ERR(gna_class);
> >> +       }
> >> +       gna_class->devnode = gna_devnode;
> >> +
> >> +       ret = pci_register_driver(&gna_driver);
> >
> > Is it possible to decouple a PCI glue driver from the class as many
> > other existing examples are doing?
> >
> 
> I see many pci drivers (including staging) that do have it glued though.
> 
> Examples are:
> 1. "static int __init kp2000_pcie_init(void)" (commit on May 20 09:34:11
> 2019)
> 2. "static int __init hl_init(void)" (commit on Mon Feb 18 09:46:43 2019)
> 
> Please give me more details.

Never use a staging driver for any type of example, _EXECPT_ for a bad
one.  There's a reason the code is in staging and not in the "real" part
of the kernel.

thanks,

greg k-h