<mmind00>
inode: so you got the hdmi working on your board? ... judging by the patch you sent :-)
<inode>
i did
<inode>
your suggestions and past patches helped me a lot, thanks
<inode>
for now the trick is that i have to have a monitor connected at boot, i suspect this might change if/when i find out what the offset for the HDMI HPD IO enable pin is in one of those GRF_SOC_CONx registers
<inode>
i also have to explictly select a suitable resolution and refresh rate via the cmdline for the HDMI-A-1 connector
<inode>
something about receiving EDID is a bit hit or miss without it
<inode>
on some occassions there will be no issues with allowing the device to use the modeline info received
<inode>
on most occassions it will seemingly select the best modeline advertised, and the monitors and tv's i've tried will display the tty0 console for a few seconds, but then the screen firmware will complain about the timing being wrong
<inode>
this more or less co-incides with "nobody cared about irq 35"
<inode>
but i get that error being displayed without any adverse effects even when the video output is good and i've selected my resolution of choice manually
<inode>
it seems to come from dw_hdmi_cec_thread
warpme_ has quit [Quit: warpme_]
<inode>
my current problem is that after 30 minutes after boot, the device becomes unresponsive over ethernet, UART2 and tty0
<inode>
this also happens if i use the android kernel shipped with the device, but do not use the initrd that was intended for it
<inode>
is this maybe due to power management (suspend/hibernate?) support that is lacking in mainline?
<inode>
i'd guess that in the android case there might be some userland process that is feeding some sort of watchdog to prevent suspension
<mmind00>
inode: actually that irq 35 is the _main_ dw-hdmi interrupt, so your devicetree-node should reference that
<mmind00>
that is also the irq where the main hotplug event will happen
<inode>
it does reference it
<mmind00>
inode: and more importantly it's the only irq in the binding + driver