On Mon, Feb 03, 2025 at 06:39:09PM +0100, Krzysztof Kozlowski wrote:
> On 03/02/2025 17:46, Uwe Kleine-König wrote:
> > +
> > + chosen {
> > + stdout-path = "serial0:115200n8";
> > + };
> > +
> > + memory@0 {
> > + /* 1 GiB */
> > + device_type = "memory";
> > + reg = <0x0 0x40000000>;
> > + };
> > +
> > + soc {
> > + fpga: bus@ff200000 {
> > + compatible = "simple-bus";
> > + reg = <0xff200000 0x00200000>;
>
>
> Are you sure simple-bus allows this? IOW, no warnings from simple-bus
> schema? To my knowledge, simple-bus allows only objects, except the few
> allowed properties (and reg is not allowed because it is bus, not a device).
I double checked to have a recent dtschema[1] and removing the reg line
doesn't change the output of
make [...] intel/socfpga/socfpga_cyclone5_de10nano.dtb CHECK_DTBS=1
But I agree that in the presence of
ranges = <0x00000000 0xff200000 0x00200000>;
the reg property isn't useful.
I'll wait a bit for more comments and then will send a v4 without that
reg line. (Unless Dinh is happy with my patch otherwise and applies it
fixing up for this issue themselves. *hinthint* :-)
> Nice coincidence with other patchset from Intel also claiming FPGA with
> resources is the bus, although there it could be root port.
There results no action item for me from this statement, right?
Best regards
Uwe
[1] pip3 install dtschema --upgrade --user --break-system-packages
Attachment:
signature.asc
Description: PGP signature
