Re: [patch] 2.4.2-pre3: parport_pc init_module bug

[Andrew Morton <andrewm@uow.edu.au>]

Jeff Garzik wrote:
> 
> Bad patch.  It should be
> 
>         if (r >= 0) {
>                 registered_parport = 1;
>                 if (r > 0)
>                         count += r;
>         }
> 
> If pci_register_driver returns < 0, the driver is not registered with
> the system.

eh?

pci_register_driver(struct pci_driver *drv)
{
        struct pci_dev *dev;
        int count = 0;

        list_add_tail(&drv->node, &pci_drivers);
        pci_for_each_dev(dev) {
                if (!pci_dev_driver(dev))
                        count += pci_announce_device(drv, dev);
        }
        return count;
}

Maybe you're thinking of pci_module_init?

Now, if there were some actual COMMENTS (scary, I know) in the pci
code which described the API, stuff like this wouldn't happen.

-