On 09/05/18 14:13, Rob Herring wrote: > On Wed, Sep 5, 2018 at 3:36 PM Frank Rowand <frowand.list@xxxxxxxxx> wrote: >> >> +cc: devicetree-compiler@xxxxxxxxxxxxxxx >> >> For context on the device-tree compiler list, this patch is for >> the Linux kernel. >> >> On 09/04/18 01:05, Tero Kristo wrote: >>> Add simple support for building DT overlays. This just compiles any >>> given .dtso file under the base arch/arm/boot/dts directory into >>> a .dtbo. >>> >>> Signed-off-by: Tero Kristo <t-kristo@xxxxxx> >>> --- >>> arch/arm/Makefile | 2 +- >>> 1 file changed, 1 insertion(+), 1 deletion(-) >>> >>> diff --git a/arch/arm/Makefile b/arch/arm/Makefile >>> index d1516f8..ab555a0 100644 >>> --- a/arch/arm/Makefile >>> +++ b/arch/arm/Makefile >>> @@ -339,7 +339,7 @@ $(BOOT_TARGETS): vmlinux >>> $(INSTALL_TARGETS): >>> $(Q)$(MAKE) $(build)=$(boot) MACHINE=$(MACHINE) $@ >>> >>> -%.dtb: | scripts >>> +%.dtb %.dtbo: | scripts > > This will need to change with my rework of the DT build rules[1]. > >>> $(Q)$(MAKE) $(build)=$(boot)/dts MACHINE=$(MACHINE) $(boot)/dts/$@ >>> >>> PHONY += dtbs dtbs_install >>> >> >> The idea of adding knowledge of .dtso and .dtbo files to the dtc >> compiler, but that idea and the related patches have fallen by >> the wayside. >> >> Current overlay sources and blobs are properly handled by dtc when >> they are .dts and .dtb files. >> >> If the dtc compiler is updated with knowledge of .dtso and .dtbo >> then these changes would be needed. At the moment the distinction >> of a .dtso and .dtbo is a naming convention that provides a clue >> about the intended use of the file. But dtc does not know these >> file name extensions, requiring a specific flag to tell dtc to >> treat them as .dts or .dtb files. I think the negative outweighs >> the positive for this patch. > > I don't follow what you are proposing. We should decide on a convention though. > > We shouldn't need .dtso because that is identified with the /plugin/ > tag. Distinguishing overlay dtbs with .dtbo seems like a good idea > though. I noted technical reasons that using .dtbo has both good and bad aspects. I too at times just fall back to a "sounds good" or "sounds bad" type response (and will continue to do so in the future), but in this case please provide a somewhat stronger reasoning for using .dtbo since I stated that I feel the in technical balance the negatives outweigh the positives. If dtc adds knowledge of .dtbo my opinion of the balance will immediately reverse. > > Rob > > [1] https://lkml.org/lkml/2018/8/21/770 >
