Sander Vanheule <san...@svanheule.net> writes: > I don't know how others feel about this, but I find something like > "ih-type = <IH_TYPE_FILESYSTEM>", > to be clearer than > "ih-type = <7>". > Would it be an option to create a dt-bindings header? For > example,'include/dt-indings/mtd/partition/uimage.h' (to mirror the DT > documentation paths) could include definitions for IH_TYPE_*. > mtdsplit_uimage.c could even include this header to prevent duplicate > definitions.
This is a good idea. I will do that for the next version. I see that mainline has an example of a shared header in include/dt-bindings/input/linux-event-codes.h which is a symlink to include/uapi/linux/input-event-codes.h Something similar could be done here. >> kernel: mtdsplit_uimage: replace "netgear,uimage" parser >> kernel: mtdsplit_uimage: replace "edimax,uimage" parser > > This patch removes "edimax,uimage", while also introducing the DT > properties to do this. For the other properties you split this up. If > there won't be any formal documentation, maybe a separate patch could > serve this purpose? Yes, the splitting should have been more consistent. Will fix that for the next version. And I do consider writing up a formal DT binding as the first patch, following the guidelines in https://www.kernel.org/doc/html/latest/devicetree/bindings/submitting-patches.html Not because I think this is material for upstream. But I agree that non-obvious stuff like these properties should be documented. If nothing else, then for my own good. Bjørn _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org https://lists.openwrt.org/mailman/listinfo/openwrt-devel
