Re: [-mm patch] aic7xxx: check irq validity (was Re: 2.6.18-mm2)

[Frederik Deweerdt <deweerdt@free.fr>]

On Sat, Sep 30, 2006 at 07:58:18PM -0400, Jeff Garzik wrote:
> Frederik Deweerdt wrote:
> >On Sat, Sep 30, 2006 at 12:43:24AM +0100, Alan Cox wrote:
> >>Ar Gwe, 2006-09-29 am 23:50 +0000, ysgrifennodd Frederik Deweerdt:
> >>>Does this patch makes sense in that case? If yes, I'll put up a patch
> >>>for the remaining cases in the drivers/scsi/aic7xxx/ directory.
> >>>Also, aic7xxx's coding style would put parenthesis around the returned
> >>>value, should I follow it?
> >>Yes - but perhaps with a warning message so users know why ?
> >>
> >>As to coding style - kernel style is unbracketed so I wouldnt worry
> >>about either.
> >>
> >Thanks for the advices. The following patch checks whenever the irq is valid before issuing a
> >request_irq() for AIC7XXX and AIC79XX. An error message is displayed to
> >let the user know what went wrong.
> >Regards,
> >Frederik
> >Signed-off-by: Frederik Deweerdt <frederik.deweerdt@gmail.com>
> 
> Actually, rather than adding this check to every driver, I would rather do something like the attached patch:  create a 
> pci_request_irq(), and pass a struct pci_device to it.  Then the driver author doesn't have to worry about such details.
> 
That's better, indeed. 
[...]
> +#ifndef ARCH_VALIDATE_PCI_IRQ
> +int pci_valid_irq(struct pci_dev *pdev)
> +{
> +	if (pdev->irq == 0)
> +		return -EINVAL;
                        ^^^^^^
Woulnd't this rather be ENODEV? Admitedly, from pci_valid_irq() (or
is_irq_valid()) point of view, it _has_ been passed an invalid value. But
from userspace's point of view, it's like the device was not present.

Regards,
Frederik