12:24 <forkbomb> has anyone tried getting kexec working on exynos?

12:25 <forkbomb> i have it sort of working - i get a serial console from the second kernel - but it doesn't finish booting

12:25 <forkbomb> i'm curious if anyone's had any more success

12:25 <mszyprow> forkbomb: I've never tried...

12:25 <forkbomb> ok

12:26 <forkbomb> btw, i think the display issue on i9300 might be because trats2 and i9300 have different LCD reset gpios

12:27 <mszyprow> forkbomb: it would be strange, because I've used i9300 schematic for trats2 development...

12:28 <forkbomb> could it be an earlier revision?

12:30 <forkbomb> because in the vendor kernel, the only references to gpy4(5) being the lcd reset line are in 'gpio-rev01-midas.h' and 'gpio-rev02-midas.h'

12:31 <mszyprow> forkbomb: I've just checked and it looks that the mainline dts has bug

12:32 <mszyprow> forkbomb: it should use gpf2-1 for mlcd_rst

12:32 <mszyprow> forkbomb: but I would need to check with the hardware

12:32 <forkbomb> ah, ok

12:33 <mszyprow> forkbomb: and you are right that some old i9300/trats2 prototype used gpy4-5

12:34 <mszyprow> maybe it worked because our bootloader did everything that was needed....

12:34 <forkbomb> yeah, maybe

12:34 <forkbomb> i also noticed that the delay values specified in dts don't seem to be used by the s6e8aa0 panel driver

12:34 <forkbomb> they're just hardcoded :P

12:36 <forkbomb> mszyprow: is there a reason cpuidle is only enabled on trats2 for exynos4412?

12:37 <mszyprow> well, old, early dt-times and conversion from board files to dt had its cost ;) feel free to send a patch

12:37 <forkbomb> i will eventually

12:37 <forkbomb> the board files are definitely quite painful though :)

12:37 <mszyprow> afair additional cpuidle modes (coupled) requires support from bootloader

12:38 <forkbomb> ah, i see

12:38 <mszyprow> and it was not working on odroid x2/u3

13:39 <forkbomb> mszyprow: occasionally i get "[ 10.386450] s3c-i2c 138d0000.i2c: timeout waiting for bus idle" when i try and kexec. any idea how i should properly cleanup i2c before kexecing?

13:44 <forkbomb> but with 'clk_ignore_unused', i have it booting properly most of the time

14:01 <mszyprow> forkbomb: no idea, stopping i2c transfers should be enough

14:04 <forkbomb> ok, guess I just have to figure out how to do that :)