11:18 <krzk> Viciouss: you cut the paste log... the cause is not there

11:19 <krzk> Unless it is the start with "INFO"?

11:42 <Viciouss> krzk: this is all that comes up in dmesg

12:04 <krzk> ah, ok :)

12:06 <krzk> if someone has time, could investigate possible deadlock in pl330 around UART, visible on Exynos5422 from time to time during reboot:

12:06 <krzk> https://krzk.eu/#/builders/21/builds/3872/steps/39/logs/serial0

12:06 <krzk> later there is also null-pointer in pl330_tasklet but maybe that's related to above?

13:03 <chewitt_> @krzk speaking of Exynos 5422 .. mesa/panfrost is running on Odroid XU3/4 now :)

13:05 <chewitt> FYI, one of the pm-domains changes has been flagged for negative performance impact

13:05 <chewitt> https://github.com/chewitt/linux/commit/71a533f97eb67809b88a75a880172829826f2ae1

13:06 <chewitt> I've asked the guy who's been poking things (Pavel) to grab some stats and report it

13:27 <krzk> chewitt: cool!

17:02 <krzk> chewitt: that exynos power domain is unlikely to cause performance degradation - I don't see there any code which could do it... more likely is that for old driver some domains for example do not work at all, therefore maybe PM runtime is disabled?

17:03 <krzk> chewitt: different topic - it's quite surprising how many patches you like to carry over instead of upstreaming them. Like with everything - not having things upstream is a costly choice, bringing troubles in debugging and reporting bugs (for any reporting buh answer will be: test it on mainline).

17:05 <chewitt> @krzk I loathe patches

17:05 <chewitt> the Amlogic patchset that I carry is mostly WIP device-tree hacking .. nothing too serious

17:05 <chewitt> the Exynos patchset is somewhat derived from the Hardkernel images

17:06 <chewitt> I'm not convinced half of the patches are required

17:06 <chewitt> my personal problem is that i'm not a coding developer, so figuring out what's essential vs. random garbage is hard

17:09 <chewitt> most of the Exynos patches have subject lines and zero description, so if you can't read the code it's not simple

17:10 <chewitt> my personal mission will be to have a minimal patchset that supports Kodi with hardware decode working

17:18 <krzk> I understand