On 10.08.2017 13:35, Greg Ungerer wrote:
On 10/08/17 19:35, Vladimir Zapolskiy wrote:
Hi Oleksij,
...
Errors in DTB (or platform data) may confuse a driver and lead to runtime
misbehaviour. You describe an error in a board DTB, which is definitely
better to handle in the SPI driver, but I don't think it is strictly
mandatory to do it, because DTB errors are supposed to be fixed in DTB.
May be one day a formal check of DTBs against Documentation/devicetree
descriptions will be added and such DTB errors could be captured on DTB
compilation stage.
I completely agree with Vladmir here. Since "cs-gpios" defines the
number of chip selects, as per the code you point out, it is the range
limit. So if a DTB defines it wrongly then you can expect some things
not to work right. The spi code quite rightly relies on the DTB
definitions to be correct for proper operation.
And in this case:
cs-gpios = <&gpio1 0 0>, <&gpio1 1 0>, <&gpio1 2 0>, <&gpio1 3 0>, <0>;
we should produce a 3 bit value b100 which will be shifted left and
"or"-ed with other ctrl bits.
So the register settings will be wrong and the device will not work.
You can't really expect any other behavior from an incorrect DTB.
Ok :)
--
To unsubscribe from this list: send the line "unsubscribe linux-spi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
