Re: [PATCH] HID: roccat: fixing actual/startup profile sysfs attribute in koneplus

[Stefan Achatz <erazor_de@users.sourceforge.net>]

Am Dienstag, den 12.04.2011, 13:27 +0200 schrieb Jiri Kosina:
On Sun, 10 Apr 2011, Stefan Achatz wrote:
> 
> > startup_profile and actual_profile didn't work as expected. Also
> > as the actual profile is persistent, the distinction between the
> > two was ambiguous. And the event is now propagated through chardev.
> > The userland tool has been updated to support this change.
> 
> Fortunately, this is 'testing' ABI. But still, I'd prefer to follow good 
> practices here, and minimize incompatible ABI changes.
> 
> Is the userspace tool able to handle the situations:
> 
> - too old tool running to kernel containing this patch
> - new tool running on older kernel
> 
> ?

Lets split this in the three possible actions, two of them broken in the
actual module code:

1 reading actual_profile: shows stale data.
2 reading startup_profile: ok.
3 writing startup_profile: does not work because of a wrong writelength:

> > -	return koneplus_send(usb_dev, KONEPLUS_USB_COMMAND_STARTUP_PROFILE,
> > -			&buf, sizeof(struct koneplus_profile_buttons));
> > +	return koneplus_send(usb_dev, KONEPLUS_USB_COMMAND_ACTUAL_PROFILE,
> > +			&buf, sizeof(struct koneplus_actual_profile));

My userspace tools are not in any distros package repository, I keep new
tools backward compatible and I do a rapid release approach.
With these points and the actual modules flaws it didn't make much sense
for me to keep startup_profile for compatibility, but I wouldn't
struggle against it. If you feel better with the compatibility approach,
I can have an updated patch ready within minutes on request.
Stefan