On 22/07/2022 19:23, Doug Anderson wrote: > Hi, > > On Fri, Jul 22, 2022 at 10:14 AM Krzysztof Kozlowski > <krzysztof.kozlowski@xxxxxxxxxx> wrote: >> >> On 21/07/2022 20:29, Doug Anderson wrote: >>> Hi, >>> >>> On Thu, Jul 21, 2022 at 9:52 AM Krzysztof Kozlowski >>> <krzysztof.kozlowski@xxxxxxxxxx> wrote: >>>> >>>> On 21/07/2022 18:43, Doug Anderson wrote: >>>>> Hi, >>>>> >>>>> On Thu, Jul 21, 2022 at 9:33 AM Krzysztof Kozlowski >>>>> <krzysztof.kozlowski@xxxxxxxxxx> wrote: >>>>>> >>>>>> On 21/07/2022 15:37, Doug Anderson wrote: >>>>>>> >>>>>>> Not worth sending a new version for, but normally I expect the >>>>>>> bindings to be patch #1 and the dts change to be patch #2. In any >>>>>>> case: >>>>>>> >>>>>>> Reviewed-by: Douglas Anderson <dianders@xxxxxxxxxxxx> >>>>>> >>>>>> I would say worth v4, because otherwise patches is not bisectable. >>>>> >>>>> You're saying because `dtbs_check` will fail between the two? >>>> >>>> Yes >>> >>> OK. Then I assume you agree that reversing the order of the patches >>> won't help, only combining the two patches into one. >>> >>> >>>>> How does >>>>> flipping the order help? If `dtbs_check` needs to be bisectable then >>>>> these two need to be one patch, but I was always under the impression >>>>> that we wanted bindings patches separate from dts patches. >>>> >>>> I don't think anyone said that bindings patches must be separate from >>>> DTS. The only restriction is DTS cannot go with drivers. >>> >>> I have always heard that best practice is to have bindings in a patch >>> by themselves. >> >> Yes, bindings must be separate patch, no one here objects this. You said >> they cannot go together via one maintainer tree or I misunderstood? >> >>> If I've misunderstood and/or folks have changed their >>> minds, that's fine, but historically I've been told to keep them >>> separate. >> >> Nothing changed. Bindings must be separate. They will be applied by >> maintainer and, if correctly ordered, this is bisectable. > > OK, I think this is the disconnect here. > > No matter what order Jimmy's patches land in, it won't be bisectable > from the standpoint of "make dtbs_check". This is what I've been > trying to say. > > * If the bindings land first then the device tree won't have sku6 and > will fail "make dtbs_check" > > * If the dts lands first then the bindings won't have sku6 and will > fail "make dtbs_check". > > Am I missing something? Ah, you're right... The patch changes the bindings of a board instead of bringing a new variant. Yeah, this cannot be bisectable if kept separate, thus order does no matter. > > So when you said "I don't think anyone said that bindings patches must > be separate from DTS" and that you cared about "make dtbs_check" being > bisectable that you were saying you wanted these squashed into one > patch. I guess that's not the case. > Best regards, Krzysztof
