On Fri, Jan 20, 2017 at 10:59 AM, Bruce Ashfield <bruce.ashfi...@windriver.com> wrote: > On 2017-01-20 09:19 AM, George McCollister wrote: >> >> On Thu, Jan 19, 2017 at 5:24 PM, Bruce Ashfield >> <bruce.ashfi...@windriver.com> wrote: >>> >>> On 2017-01-19 6:20 PM, George McCollister wrote: >>>> >>>> >>>> What's the correct way to use a custom bsp .scc file located in my >>>> custom bsp layer? For purposes of this email let's say it's name is >>>> myplatform-preempt-rt.scc. If I specifically add it to the >>>> yocto-linux-rt_4.4 SRC_URI it appears to get processed twice by scc >>>> according to the meta-series output file. It then tries to apply >>>> patches to standard/preempt-rt/base that don't apply. When building on >>>> a intel-core2-32 platform, intel-core2-32-preempt-rt.scc only gets >>>> processed once and no incorrect patches are applied. >>> >>> >>> >>> Something could have broken, I never rule out bugs! >>> >>> Morty had some different semantics than the newer branches (which >>> are simplified compared to the auto-resume logic in old releases). >>> >>> That being said, the capability to have a BSP description in the >>> SRC_URI is there, and did work during that release cycle. To have >>> it work, you do need the right format in the BSP .scc file, in >>> particular something with the KMACHINE and KTYPE descriptions will >>> be seen as an entry point, and not trigger the re-application of >>> the inherited patches. >>> >>> Is there any way that I can see the contents of that .scc file ? >>> Otherwise, I can spin up a morty build and mock something up >>> myself. >> >> >> Bruce, >> >> I stripped down my bsp later and pushed it github. You can find the >> BSP .scc (and the files it includes) here: >> >> https://github.com/gmccollister/meta-orion-minimal-bsp/tree/master/recipes-kernel/linux/linux-yocto/orion-io >> >> The linux-yocto-rt_4.4.bbappend that adds orion-io-preempt-rt.scc to >> the SRC_URI is here: >> >> https://github.com/gmccollister/meta-orion-minimal-bsp/blob/master/recipes-kernel/linux/linux-yocto-rt_4.4.bbappend >> > > Thanks, this helped .. and did show me the behaviour that you were > mentioning. > > I had to scratch my head for minute, since I tested this scenario > myself when doing the major re-work in morty for how the kernel > meta data s processed. > > In morty I was able to delete about 40% of all the code and simplified > the processing. There used to be logic that would process the patch > series, resume processing, wrap .cfg/.scc files, detect BSP definitions > and do special processing .. etc, etc. It was a rats nest of corner > cases. > > In morty, if something is on the SRC_URI .. it is applied to the > tree. No exceptions. Hence why when your BSP definition is on the > SRC_URI .. it tries to reapply all the patches to the tree. > > The way that out of tree/recipe space meta data is handled is now > the same as the way the core linux-yocto 'kernel-meta' is handled. > You specify either a git repo or a directory on the SRC_URI and > tag it as having meta-data. Those directories are searched for > a BSP that matches the machine that is being built. Once found, > it is processed like any other BSP .. no special cases. > > What looks to have been missed was documenting *how* this should > be done with a directory in a layer, versus a separate git > repository. If you'd like to open a bug against me, I'll make sure > that it is documented in the upcoming release (and hopefully on > morty as well).
Done: https://bugzilla.yoctoproject.org/show_bug.cgi?id=10946 > > The following change to your bbappend triggered the right processing > in my case: > > SRC_URI_append_orion-io = " \ > - file://orion-io-preempt-rt.scc \ > + file://orion-io;type=kmeta;destsuffix=orion-io \ > " Perfect, that works. Thanks for clearing this up! > > All that says, is that the directory orion-io (where you had the > .scc/.cfg files) is kernel meta data, and that it can be found > in the directory called 'orion-io' after the fetcher pulls it > into WORKDIR. > > From there, it is found, and the configuration, etc, applied to > the build. > > Bruce > > > >> -George >> >>> >>> Bruce >>> >>> >>>> >>>> If I don't include it in the yocto-linux-rt_4.4 SRC_URI it's never >>>> applied at all. I'm wondering if this a new problem with the last big >>>> push to morty since I didn't notice this before. >>>> >>>> I also verified that I can make myplatform-preempt-rt.scc work by >>>> doing the following (only shows as processed once in meta-series and >>>> no incorrect patches applied): >>>> bitbake -c cleanall linux-yocto-rt >>>> bitbake -c unpack linux-yocto-rt >>>> Manually copy myplatform-preempt-rt.scc and related .scc, .cfg files >>>> to the following path: >>>> >>>> >>>> work/myplatform-oe-linux-gnueabi/linux-yocto-rt/4.4.36+gitAUTOINC+b846fc6436_61d7bf47fe-r0/kernel-meta/bsp/myplatform/ >>>> bitbake -c compile linux-yocto-rt >>>> >>>> Regards, >>>> George McCollister >>>> >>> > -- _______________________________________________ yocto mailing list yocto@yoctoproject.org https://lists.yoctoproject.org/listinfo/yocto
