austriancoder changed the topic of #etnaviv to: #etnaviv - the home of the reverse-engineered Vivante GPU driver - Logs https://freenode.irclog.whitequark.org/etnaviv
_whitelogger has joined #etnaviv
_whitelogger has joined #etnaviv
pcercuei has quit [Quit: dodo]
_whitelogger has joined #etnaviv
lynxeye has joined #etnaviv
<cengiz_io> hello again marex, card0 is etnaviv, card1 is display-subsystem-card
_whitelogger has joined #etnaviv
pcercuei has joined #etnaviv
<cengiz_io> what could actually cause "modetest -c" to show zero connectors?
<cengiz_io> kmstest says "All OK"
<cengiz_io> where do we define connectors for drm?
<cengiz_io> There are no useful docs for this, sorry for keeping on asking desperately
<daniels> is modetest actually looking at the right card?
<daniels> I would suggest using https://github.com/ascent12/drm_info instead of modetest
<daniels> modetest is ... not very good
<cengiz_io> daniels I'm trying both cards card0 and card1
<cengiz_io> will try to compile drm_info now. thanks. the worst part is I've tried every example devicetree definition I could ever find over the last two weeks
<cengiz_io> none of them helped
<cengiz_io> can panel-simple work with drm?
<pcercuei> yes
<pcercuei> I believe you use a custom (non-upstream) device tree?
<cengiz_io> pcercuei yes I do
<pcercuei> make sure your panel and display are correctly connected together with ports
<cengiz_io> pcercuei that's the part I'm struggling most. totally cargo cult programming that..
<cengiz_io> daniels can you take a quick look at this drm_info.json ? anything irregular? other than no connectors... https://gist.github.com/cengizIO/3bb36a8e4e3f0ae0629f5d70bd590268
<pcercuei> can you share your devicetree?
berton has joined #etnaviv
<daniels> yeah, no connectors and no encoders makes me think that you're either missing some modules (bridge, power regulator, panel, etc) or missing the device tree
<pcercuei> it could be that the panel driver does not probe. Did you check dmesg?
<cengiz_io> I've checked dmesg, panel-simple is nowhere to be seen but imx-drm seems to initialize perfectly
<pcercuei> rgrep "tianma,tm050rdzg03" drivers/gpu/drm/panel/
<pcercuei> returns nothing
<cengiz_io> that's in a patch on my tree
<pcercuei> ok
<pcercuei> ok, I see the problem
<pcercuei> lines 37 to 52 should be inside a ports { } node
<cengiz_io> let me take a look
<cengiz_io> I've wrapped them now. how did you figure that out?
<pcercuei> experience :)
<cengiz_io> because I've copied this from some example and they didn't add that... :/
* cengiz_io will bang his head against a concrete wall if this works
<cengiz_io> well pcercuei I got bad news and good news
<cengiz_io> good news: now dtc is complaining about port definitions. it didn't parse them before. thanks for that
<cengiz_io> bad news: still no simple panel getting probed or any connectors listed
<pcercuei> well it's not normal that it complains
<cengiz_io> pcercuei I should be able to see something related to ipu1_ds0 in dmesg right? there are only capture devices listed in dmesg
<cengiz_io> just warnings
<pcercuei> you can't have warnings and expect stuff to work ;)
<cengiz_io> Warning (reg_format): /display/ports/port@0:reg: property has invalid length (4 bytes) (#address-cells == 2, #size-cells == 1)
<pcercuei> you need to have the #address-cells and #size-cells properly set, like this: https://github.com/torvalds/linux/blob/master/arch/mips/boot/dts/ingenic/jz4725b.dtsi#L318-L329
<cengiz_io> where do those numbers come from?
<pcercuei> devicetree documentation
<pcercuei> this is for another SoC and another driver btw, which expects the IPU to be at port 8
<pcercuei> in your case it's port 0
<cengiz_io> oh ok
lynxeye has quit [Quit: Leaving.]
adjtm has joined #etnaviv
berton has quit [Quit: Leaving]
adjtm has quit [Quit: Leaving]
<cengiz_io> is there any other requirement to probe panel-simple other than using a matching compatible string in devicetree and enabling CONFIG_DRM_PANEL_SIMPLE?
flto has quit [Quit: Leaving]