Re: [PATCH] ARM: OMAP: mux: add config for 16xx SPI pins

[Tony Lindgren <tony@atomide.com>]

* Mark Howell <mhowell@northlink.com> [060830 03:59]:
> Mark Howell wrote:
> >This patch adds pin mux info for the SPI master/slave interface on 
> >OMAP16xx. Data from OMAP 1611/1612 TRM and errata. Works for me on my
> > 1611/H2 with current git kernel.
> >
> 
> I have since spent some lab time with a scope and figured out that I
> misunderstood the pull-up config bits in the mux definition structure
> (they are inverted vs. what is actually written to the pull-up control 
> registers). The patch I posted before works, but it can be better.

OK
 
> I'll post a patch to set better default values for 16xx SPI pin mux
> after I get a little more lab time. I don't think this is a burning
> issue for anyone 'cept me, since eons had passed without anyone spec'ing
> these pins for 16xx SPI anyway :-)
> 
> But this brings up a question... there are some other pins where I'm
> suspicious of the default pull-up config in mux.c, such as for the
> u-wire interface. Won't there be a bit of a power penalty for turning on
> pull-up or pull-down when it isn't necessary? Is there some established 
> opinion or practice in this group regarding pin mux pull-up config to 
> which I can refer?

But the pull_ena bit is 0 so they are disabled..

Somebody from TI please correct if I'm wrong, but if a device works
without pull-up or pull-down they should not be used. And if device pins
are not used, the pins should not be floating to save power.

In general, the pin muxing should be done in the bootloader for production
boards as it's board specific and static. During driver development and
debugging, pin muxing can be done in the kernel to make development
easier.

Regards,

Tony