Hi Ricardo,
On 06/05/2020 10:21, Ricardo Cañuelo wrote:
Hi Tomi, thanks for reviewing the patch.
On mar 28-04-2020 12:49:28, Tomi Valkeinen wrote:
I don't quite understand this. We cannot have negative numbers in dts files? Or we can, but dt_binding_check doesn't handle them correctly? Or that int32 is not supported in yaml bindings?
AFAICT, you can have negative numbers in dts files (see [1] and [2]) and
This is also my understanding after some googling. And there's even of_property_read_s32() in the kernel.
the DT schema certainly supports signed integers, but dt_binding_check seems to interpret all cells as unsigned 32bit integers because that's what they are, really. In kernel code this is not a problem because you
Well, this is in the nitpick category, and maybe not even relevant, but I don't think that's correct. They're just bits. Some pieces of SW happen to use u32 containers to store the bits. But what the bits mean is not related to the container.
can cast the value back to a signed int before you run your own sanity checks on them.
Doesn't all this just point to a bug or missing feature in dt_binding_check? That's not a reason to change the ABI.
Tomi