Re: [patch 0/2] acpi: driverregistration again can report ENODEV

[Frank Seidel <fseidel@suse.de>]

[-- Attachment #1: Type: text/plain, Size: 1063 bytes --]

On Monday 23 October 2006 17:34, Bjorn Helgaas wrote:
> I think the current tree contains a merge error in asus_acpi_init().
> It says:
>
> 	if (!asus_hotk_found) {
> 		...
> 		return result;
>
> But "result" is always zero at this point (it came from
> acpi_bus_register_driver(), which returns either zero or a negative
> error value).
>
> My original change:
>  
> http://kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=co
>mmitdiff;h=578b333bfe8eb1360207a08a53c321822a8f40f3 contained:
>
> 	if (!asus_hotk_found) {
> 		...
> 		return -ENODEV;
>
> which is what we want.  I think the -ENODEV got lost when resolving
> conflicts with the "propagate correct return value" patches that
> happened at about the same time.
After some more testings and digging through the code.. i
agree with you that this is way to go (for now).
So, i sadfully now have to vote against my own patch ;-)

Len, please don't commit it. As Bjorn wrote, fixing the asus_acpi module
with return -ENODEV is the better solution.

Thanks,
Frank

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]