00:39 <lumag_> robclark, yep, CTL_START are used only for cmd mode, so this is correct

00:40 <lumag_> robclark, another point to check: which resources (INTF, LM, PP, etc) are allocated in each of the cases?

00:41 <robclark> thx to addition to atomic state dump, I can tell they match in both cases ;-)

00:42 <abhinav___> robclark: there is one more block missing in the snapshot. i do not think it will vary but just-in case, can youn include this one?

00:42 <abhinav___> https://www.irccloud.com/pastebin/8UsyHAFF/

00:42 <abhinav___> I will submit a change for this

00:43 <robclark> ok

00:50 <lumag_> robclark, good :-)

00:51 <lumag_> abhinav___, feel free to include my r-b when you will submit this change :-D

00:53 <robclark> abhinav___: lm state matches in both cases

00:55 <robclark> abhinav___: btw, for the regs that are double-buffered (ie. latch of flush+vblank), do the unbuffered values show up anywhere in the I/O space (ie. if it was a flush issue, would it show up in register dumps)?

01:20 <sboyd> is there a flush harder button?

01:21 <abhinav___> robclark: so in other words you are asking if a particular register value is not flushed will it show stale value or the new one when you read? It will show the new value in the I/O state but HW will be using the old value till the flush

01:22 <abhinav___> sboyd: :)

01:25 <robclark> abhinav___: I meant is there another register that shows the flushed/latched value..

01:34 <abhinav___> robclark: no, there isnt such one

01:36 <robclark> hmm, ok.. that would have been useful to debug flush issues (but at least I don't *think* that is what this is)

01:39 <abhinav___> yeah i also dont think so ..... i think the solid fill on the planes will give some answer

01:39 <abhinav___> we will try to fix up that path in the bug

01:39 <abhinav___> will leave a note to kalyan tonight

01:44 <robclark> sg

18:30 <aka_[m]> Do anyone have idea what and if something should be set inside bw_table of venus/core.c for avg_10bit if downstream vidc.dtsi doesnt contain ubwc-10bit profile? i only has ubwc on 8953/8937

18:31 <aka_[m]> s/i/it/