02:03 <hggg123> Hi.. Hope this is the right forum.. I am trying to bringup SMP on a new soc with 2 arm cores. Now when trying to start 2nd core, with multple fixes along the way, I got to the point where secondary_startup (head.S) tries to enable the MMU. Now we set up the TTBR1 with the pagetables already setup for core0 and then enable MMU. At this point, it takes a data abort, basically because the identity mapping

02:03 <hggg123> entries do not exist. But this seems to be a classic problem. So how does it work on other socs?

02:53 <hggg123> so __turn_mmu_on does seem to be in .idmap.text section.. So guess the real question is why the mapping does not exist in my case.

04:40 <hggg123> found the problem..virt_to_idmap() is not defined properly for XIP_KERNEL which was something I was trying to get working with SMP at the same time. Thanks for your help.. Cheers!

14:54 <arnd> krzk, robher: can you confirm that the addition of lots of "bootph-all" properties in https://git.kernel.org/pub/scm/linux/kernel/git/ti/linux.git/commit/?h=ti-k3-dt-for-v6.7&id=4669288219a7f77c6ff992d10ce6a20660863979 etc is what we want for the dts files in the kernel? I hadn't seen those before, but I can see it following the schema in https://github.com/devicetree-org/dt-schema/blob/main/dtschema/schemas/bootph.yaml

15:18 <robher> arnd: Want is an overstatement, but allowed, yes. One difference with the prior u-boot only tags, is that parent nodes are supposed to be implicitly included. Before, the node would obviously remain, but all the properties could be dropped.

15:19 <robher> I suspect without enforcing that, folks have just done a search and replace...