Hi Krzysztof Kozlowski and Marijn Suijten On 02/08/22 13:16, Krzysztof Kozlowski wrote: > On 01/08/2022 20:53, Marijn Suijten wrote: >> On 2022-08-01 16:55:11, Joel Selvaraj wrote: >> Any summary what changed since v1? > > +1 (although I already reviewed it) Will add them in v3. >> I haven't re-read what was discussed in v1, but doing it this way causes >> git to _not_ record this as a rename but instead state that everything >> has been removed from sdm845-xiaomi-beryllium-tianma.dts, and a new file >> sdm845-xiaomi-beryllium-common.dtsi was introduced with inconveniently >> almost identical contents (see the unnecessary size of the patch that >> follows). > > The patch should be formatted a bit different. I agree that if combined > with first patch and proper settings (-M10% -C10%, optionally also > experiment with -B although here it looks not needed). > > I reviewed the diff side-by-sie and there were differences (labels) > tricky to spot. If you generate the patch correctly, not much of review > is needed... >> Instead, I'd keep the original patch with a rename from >> sdm845-xiaomi-beryllium.dts to sdm845-xiaomi-beryllium-common.dtsi, and >> _also_ update the existing: >> >> dtb-$(CONFIG_ARCH_QCOM) += sdm845-xiaomi-beryllium.dtb >> >> in Makefile to match this rename so that it keeps compiling, even if >> that means we treat a .dtsi as a .dts which may (likely) not be treated >> correctly by existing build rules. >> >> If it doesn't - and this approach is probably frowned upon anyway - it >> is perhaps easiest to generalize sdm845-xiaomi-beryllium.dtb (as >> suggested above) _and_ introduce sdm845-xiaomi-beryllium-tianma.dtb >> _and_ update Makefile in a _single_ patch, such that everyting keeps >> compiling and stays consistent wrt how git treats renames. Later >> patches update the compatible and add the ebbg variant. >> >> - Marijn > > Best regards, > Krzysztof I will work on v3 based on both of your comments. Regards, Joel Selvaraj
