09:12 <Grimler> krzk: do you still have an arndale-octa board? If yes, could you share a dmesg from it? I am trying to mainline another exynos5420 device (galaxy tab s), and am having some trouble bringing up the secondary CPUs. CCI has to be disabled for this device (as for arndale-octa), and I am able to bring up the first 4 CPUs after a small patch to arch/arm/mach-exynos/{firmware,platsmp}.c

09:12 <Grimler> krzk: the last 4 CPUs (the small ones) are not brought up though. Since arndale-octa also has disabled CCI I am curious if all CPUs are brought up at boot on that device

09:19 <krzk> Grimler: I have, but not with me at the moment. Also, I stopped booting it under my CI, so the last boot was two years ago: https://krzk.eu/#/builders/32

09:20 <krzk> Grimler: Octa has the same issue - you will have only 4 cpus. It was a pain to work on it, so I gave up. Not worth to waste time on 5420.

09:32 <Grimler> krzk: thanks! Identical error in those logs to what I am getting. Exynos542X is at least rather well supported currently, so there's plenty of material to dig into for someone (i.e. me) who wants to get into kernel hacking :)

11:25 <krzk> Grimler: you can check out branches on my github: var/exynos-secondary-cpu-hotplug and wip/fix-arndale-octa-cpus

11:26 <krzk> and vendor kernel it's your friend for decoding these crazy SMC calls...

12:21 <Grimler> krzk: thanks! I'll probably send a patchset to add base support with only 4 CPUs first, as I am sure it will take me many months to understand all the assembly code involved here

12:22 <Grimler> krzk: do you remember if CCI worked (no more "imprecise external aborts" https://github.com/torvalds/linux/commit/25217fef3551) for arndale after your fixes in the fix-arndale-octa-cpus branch?

13:03 <krzk> Grimler: no, it did not work