Re: [PATCH] video: hpfb: Unregister DIO driver on init failure

[Geert Uytterhoeven]

Hi Haoxiang,

CC hp300

On Tue, 23 Jun 2026 at 06:41, Haoxiang Li <haoxiang_li2024@xxxxxxx> wrote:
> hpfb_init() registers the DIO driver via dio_register_driver().
> If a later error occurs, the function returns directly without
> unregistering the DIO driver. Unregister the DIO driver before
> returning from these error paths.
>
> Signed-off-by: Haoxiang Li <haoxiang_li2024@xxxxxxx>

Thanks for your patch, which is now commit d6c3e2402523ce01 ("fbdev:
hpfb: Unregister DIO driver on init failure") in fbdev/for-next

> --- a/drivers/video/fbdev/hpfb.c
> +++ b/drivers/video/fbdev/hpfb.c
> @@ -407,10 +407,13 @@ static int __init hpfb_init(void)

As per the comment out-of-context above, this driver supports devices
on two types of buses:

        /* Topcats can be on the internal IO bus or real DIO devices.
         * The internal variant sits at 0x560000; it has primary
         * and secondary ID registers just like the DIO version.
         * So we merge the two detection routines.

>         err = copy_from_kernel_nofault(&i, (unsigned char *)INTFBVADDR + DIO_IDOFF, 1);
>
>         if (!err && (i == DIO_ID_FBUFFER) && topcat_sid_ok(sid = DIO_SECID(INTFBVADDR))) {
> -               if (!request_mem_region(INTFBPADDR, DIO_DEVSIZE, "Internal Topcat"))
> +               if (!request_mem_region(INTFBPADDR, DIO_DEVSIZE, "Internal Topcat")) {
> +                       dio_unregister_driver(&hpfb_driver);
>                         return -EBUSY;
> +               }
>                 printk(KERN_INFO "Internal Topcat found (secondary id %02x)\n", sid);
>                 if (hpfb_init_one(INTFBPADDR, INTFBVADDR)) {
> +                       dio_unregister_driver(&hpfb_driver);
>                         return -ENOMEM;
>                 }
>         }

Hence if the detection or initialization on the internal bus fails,
other devices on the DIO bus must not be force-unbound.

This is also the reason why any error returned by
copy_from_kernel_nofault() is not considered fatal.

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@xxxxxxxxxxxxxx

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds