On Fri, Mar 07, 2014 at 10:48:41AM +0800, Mark Brown wrote:
> On Thu, Mar 06, 2014 at 05:05:39PM +0000, Charles Keepax wrote:
>
> > The following patch added support for spi controllers with a dedicated
> > chip select pin:
> >
> > commit 3146beec21b64f4551fcf0ac148381d54dc41b1b
> > spi: s3c64xx: Added provision for dedicated cs pin
> >
> > It updated the device tree binding to require a "cs-gpio" property to be
> > specified on the spi controller node if chip selects will be given as
> > GPIOs per slave, rather than the controller having a dedicated internal
> > chip select pin.
>
> No, it doesn't - it's saying that if the device has a "cs-gpio" property
> then to use that as the chip select. It's not a boolean, it's a GPIO
> specifier. Looking at the code it looks like the intention is to search
> all children for a cs-gpio during the controller probe, it's possible
> that this isn't working correctly.
That is basically part of my question is the current setup doing
what it is intended to? The Samsung binding has controller-data
blocks on each of the slaves that specify the gpio for that
slave.
@@ -1326,7 +1340,11 @@ static int s3c64xx_spi_probe(struct platform_device *pdev)
sdd->cntrlr_info = sci;
sdd->pdev = pdev;
sdd->sfr_start = mem_res->start;
+ sdd->cs_gpio = true;
if (pdev->dev.of_node) {
+ if (!of_find_property(pdev->dev.of_node, "cs-gpio", NULL))
+ sdd->cs_gpio = false;
This part of the original patch adds the check, I guess the
mistake could be an assumption that of_find_property will
recursively check all the children, which it won't. This will
just check the controller node itself for a cs-gpio property but
as these are specified within sub-nodes they won't be found.
Hence currently you need to add one at the controller level.
Thanks,
Charles
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
