13:13 <flacks> robmur01: can you be more specific about how rkclk_init() is integrated into downstream u-boot? ayufan's and rockchip's u-boot trees only use rkclk_init() once per chip clock driver.

13:13 <flacks> there's no other usage of rkclk_init() throughout both trees

13:15 <flacks> but anyway, after some investigation, I got boot time to be a *lot* quicker using the miniloader - `capacity-dmips-mhz` has to be removed from `rk3399.dtsi` in linux - that prevents the soft lockup at boot and prevents the various hangs

13:16 <flacks> removed for all 6 processors*

13:20 <flacks> from ~160s before linux hands off control to the OS down to ~13s :)

13:21 <flacks> anarsoul: FWIW :)

15:40 <maz> flacks: or not. the real solution is to fix your firmware.

15:41 <maz> flacks: adding bugs to the DT isn't a viable, long term solution.

17:22 <anarsoul> flacks: it doesn't sound like a fix to me

19:14 <sphalerite> hanetzer: do you happen

19:14 <sphalerite> oops

19:15 <sphalerite> did you get suspend working at all with coreboot?

19:16 <hanetzer> sphalerite: never tried it

19:17 <sphalerite> istr it working with an older kernel and the stock firmware, but here coreboot runs and then… nothing

21:41 <flacks> maz: u-boot rk3399.dtsi doesn't have that setting. linux DTs are older

21:59 <anarsoul> flacks: linux DTs cannot be older since u-boot DTs are downstream

22:10 <flacks> well that's comparing u-boot upstream vs latest linux stable. rk3399.dtsi between them is quite diverged

22:11 <flacks> u-boot HEAD*

22:12 <anarsoul> flacks: because u-boot has a lot of u-boot specific stuff in it and it hasn't been split properly into rk3399-u-boot.dtsi

22:16 <flacks> oh that makes sense

22:18 <flacks> so are you implying that having `capacity-dmips-mhz` is correct and shouldn't be removed? u-boot rk3399.dtsi doesn't have it.

22:20 <flacks> anarsoul : ^

22:20 <anarsoul> yes

22:31 <flacks> I see