<mwalle>
kholk: i just tried your patches but I get
<mwalle>
[ 3.094831] mediatek-drm mediatek-drm.9.auto: probe with driver mediatek-drm failed with error -22
<mwalle>
[ 3.121116] mediatek-drm mediatek-drm.11.auto: probe with driver mediatek-drm failed with error -22
<mwalle>
I just use vdosys1 the first one isn't set up
<kholk>
mwalle: yes I tried only with the second vdosys, main at vdo0, ext at vdo1
<mwalle>
mh, let me try with a dummy dsi
<kholk>
you have to declare both because otherwise
<kholk>
the hardcoded stuff kicks in
<kholk>
logic is - if there's an OF graph, use that - otherwise, use the hardcoded paths
<kholk>
that was done to keep compatibility with older devicetrees
<mwalle>
kholk: hm, unfortunately. on the first there is only hdmi, which isn't supported yet
<kholk>
there's no way around that if not to add a "mt8195b" compatible
<kholk>
ouch.
<mwalle>
kholk: can't we disable the vdosys0 completely?
<kholk>
I don't think we can, never tried though.
<kholk>
it's all because of the hardcoded paths in mediatek-drm anyway
<kholk>
if you go to mtk_drm_drv.c and remove the hardcoded paths, it shouldn't complain anymore
<kholk>
but that's a hack
<mwalle>
also, not sure if we want to keep that _main and _ext thing, which I saw in your example dts
<mwalle>
because, that sounds like vdosys0 is the main and vdosys1 is the second path, which it is not - or shouldn't be
<kholk>
0 is always main, so logically I wrote main...
<kholk>
...same for ext, it's always 1
<kholk>
(0 meaning endpoint@0)
<mwalle>
mh, then.. what is main and what is ext?
<kholk>
main is supposed to be the primary display
<kholk>
ext is supposed to be external display (I stole the names from mediatek-drm itself)
<kholk>
or any other secondary display anyway
<kholk>
mediatek-drm is naming them as "main", "ext" and "third"
<kholk>
(and I of course disagree with the naming)
<kholk>
btw, food time, brb :-)
<mwalle>
enjoy!
<mwalle>
kholk: well, yeah, but we know that the driver is hardcoding everything and make assumptions which aren't true for all setups
<mwalle>
so moving towards the dt graph, we should get rid of these assumptions
<mwalle>
kholk: oh and i completely missed that all in the ext chain are endpoint@1.
<kholk>
yeah, I didn't really want to do that initially, but I had no other choice
<kholk>
reordering the pipelines (between main,ext,third) is not trivial at all and requires some restructuring of the mediatek-drm driver
<kholk>
that's all because it supports multiple vdosys, so it's getting multiple instances...
<kholk>
they eventually converge into one - luckily - but the probe stage is like that
<kholk>
the only way of doing that would be to actually perform the reordering after everything probed and before the component children are actually initialized
<kholk>
but then again, that's not trivial at all
<kholk>
my urgency was to actually stop getting more of that hardcoded me... erm... :-)
<kholk>
so that the debt doesn't get higher
<kholk>
and we stop getting more before it's too late....... :-)
<kholk>
(and now it's already *almost* too late)
schroes_ is now known as schroes__
schroes__ is now known as schroes
paricbat has joined #linux-mediatek
paricbat has quit [Remote host closed the connection]
paricbat has joined #linux-mediatek
paricbat has quit [Changing host]
paricbat has joined #linux-mediatek
eballetbo has quit [Quit: Connection closed for inactivity]
paricbat has quit [Quit: paricbat]
paricbat has joined #linux-mediatek
paricbat has quit [Changing host]
paricbat has joined #linux-mediatek
paricbat has quit [Quit: paricbat]
paricbat has joined #linux-mediatek
paricbat has quit [Changing host]
paricbat has joined #linux-mediatek
paricbat has quit [Remote host closed the connection]
paricbat has joined #linux-mediatek
schroes has quit [Read error: Connection reset by peer]
schroes has joined #linux-mediatek
ungeskriptet has quit [Quit: Ping timeout (120 seconds)]
paricbat has quit [Remote host closed the connection]