Re: gpio-ucb1400

[Guenter Roeck <linux@roeck-us.net>]

On Fri, Mar 29, 2013 at 08:46:39PM +0100, Jean Delvare wrote:
> Hi all,
> 
> In September 2009, a driver for the GPIO function of the UCB1400 chip
> was added to the kernel tree. The probe function of this driver requires
> ucbdata to be set. The only place where this happens is in function
> ucb1400_gpio_set_data(). This function was never call, and still isn't.
> So this is dead code for 3.5 years as far as the upstream kernel is
> concerned.
> 
> To make things worse, this driver can't be built as a module, for no
> good reason that I can see.
> 
> Marek, can you explain what was the point of submitting this driver that
> nobody can use?
> 
> I would like either this driver to be fixed so that it can be used (and
> that would IMHO start with dropping the ugly ucb1400_gpio_set_data hook
> and global variable ucbdata), or this driver to be dropped from the
> kernel tree. If the driver is kept, it should be adjusted so that it can
> be built as a module.
> 
> If I overlooked something, please let me know.
> 
Interestingly, the author made an attempt to fix that with [1]. It looks like
the rest of that series was merged, but this patch wasn't, though I don't find
any information about the reason.

Guenter

[1] http://lists.infradead.org/pipermail/linux-arm-kernel/2010-October/028656.html