Re: [RFC PATCH] amd64-agp: do not bind to pci driver if probing fails

[Lukas Wunner]

On Sat, Jun 21, 2025 at 04:55:52AM +0300, Ahmed Salem wrote:
> --- a/drivers/char/agp/amd64-agp.c
> +++ b/drivers/char/agp/amd64-agp.c
> @@ -768,10 +768,15 @@ int __init agp_amd64_init(void)
>  
>  		/* Look for any AGP bridge */
>  		agp_amd64_pci_driver.id_table = agp_amd64_pci_promisc_table;
> -		err = driver_attach(&agp_amd64_pci_driver.driver);
> -		if (err == 0 && agp_bridges_found == 0) {
> +		if ((int *)agp_amd64_pci_driver.probe != 0) {
>  			pci_unregister_driver(&agp_amd64_pci_driver);
>  			err = -ENODEV;
> +		} else {
> +			err = driver_attach(&agp_amd64_pci_driver.driver);
> +			if (err == 0 && agp_bridges_found == 0) {
> +				pci_unregister_driver(&agp_amd64_pci_driver);
> +				err = -ENODEV;
> +			}

Is the "probe" member in agp_amd64_pci_driver overwritten with a
zero pointer anywhere?  I don't see that it is, so it seems the
else-branch is never entered.

I had already prepared a fix for this, but waited for 0-day to
crunch through it.  I've just submitted it, so that's what I had
in mind:

https://lore.kernel.org/r/f8ff40f35a9a5836d1371f60e85c09c5735e3c5e.1750497201.git.lukas@xxxxxxxxx/

Thanks,

Lukas