On 01/20/2017 07:30 AM, Burton, Ross wrote:
On 20 January 2017 at 13:29, Maxin B. John <maxin.j...@intel.com <mailto:maxin.j...@intel.com>> wrote: > The issue here is that in a multilib build the ${base_libdir} is resolving to /lib64 which completely breaks some tools that fully expect /lib/modules to be a fixed location regardless if the target is 32/64 or something else. > I would expect the kernel to be ok with this change if we provide the kernel command line parameter: "firmware_class.path=${base_libdir}" or similar.
You are correct that this could work, but that is not a parameter I would expect to set in the default case. It would only be something I used to execute a system rollback where I need firmware to get the device running again.
I believe that using nonarch_base_libdir is the proper fix here, we need to understand why this didn't just work. Ross
When I had tried the search and replace I still ended up with the following problem, which is why I elected to just revert to what worked previously. WARNING: linux-yocto-4.8.12+gitAUTOINC+3edb4de355_9bcb4ea3fa-r0 do_package: QA Issue: linux-yocto: Files/directories were installed but not shipped in any package: /lib64 /lib/modules/4.8.17-yocto-standard/modules.builtin /lib64/firmware /lib64/firmware/cpia2 /lib64/firmware/cpia2/stv0672_vp4.bin Please set FILES such that these items are packaged. Alternatively if they are unneeded, avoid installing them or delete them within do_install. linux-yocto: 5 installed and not shipped files. [installed-vs-shipped] I am sure this problem can be corrected and debugged. Cheers, Jason.
-- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core