[Bug 103347] [dp] intel_dp.c:1554 WARN_ON(len <=8)

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=103347

--- Comment #17 from shashank.sharma at intel.com <shashank.sharma at intel.com> ---
(In reply to F.Hartung from comment #16)
> So to be honest I do not see where that should lead.
> 
> The forum is not only talking about HDMI 2.0 and 10bit color capabilities,
> mostly they / we talked about the buggy audio pass-through or no audio
> output at all. Most of the HDMI firmware updates fixed those issues like
> this 
> 
> http://nucblog.net/2017/02/apollo-lake-and-kaby-lake-nuc-hd-audio-
> bitstreaming-firmware-fix/
> 
Yes we have fair idea about both of these issues, and that's why I mentioned
these are advance use cases, whereas your problem is basic. Based on the logs
shared, there is no LSPCON device attached !!
> 
> The uploaded dmesg output literally states: 
> 
> [    1.488203] i915 0000:00:02.0: vgaarb: changed VGA decodes:
> olddecodes=io+mem,decodes=io+mem:owns=io+mem
> 
> [    1.488607] [drm] Finished loading DMC firmware i915/bxt_dmc_ver1_07.bin
> (v1.7)
> 
> [    1.489424] [drm] forcing DP-2 connector on
> 
> [    1.511305] [drm] Initialized i915 1.6.0 20171023 for 0000:00:02.0 on
> minor 0
> 
> So yes IMHO the i915 driver is loaded or at least it's firmware and stuff,
> also the DP-2 connector is mentioned several times.
> 
These are literally the only places I915 gets mentioned about in the uploaded
logs, whereas if I915 driver is loaded, we will see a lot lot more logs about
I915 and LSPCON.

> This is the Asrock support page https://asrock.com/mb/Intel/J3455-ITX/
> 
> Feel free to show me the wrong port I'm using. Last time I checked the
> connectors it had a DVI-D, D-SUB and a single HDMI. Afaik there is no "real"
> DP connector but the internal DP 1.4 port is converted by the lspcon to HDMI
> 2.0 because the apollo lake plattform itself provides no HDMI 2.0 just 1.4.

Ok, in the image shown, from the top view it was not possible for me to detect
if there are 2 HMDI ports stacked, or there is only one port. 

Now if that's the only HDMI port, then there is a very high possibility that
I915 driver is not even loaded, so LSPCON is not enabled on that port, so it's
enumerated as HDMI-1 and so on. 

My theory says:
- I915 driver is not loaded.
- LSPCON is not detected, and configured/enabled.
- There is hardcoded EDID for connector DP-2:
[   14.353325] [drm] Got external EDID base block and 1 extension from
"edid/edid.bin" for connector "DP-2"
- So modeset is bypassing the actual DP detect sequence, and while doing the
Link training for modeset, its getting this problem as there is no real DPCD
for DP.    

Next steps:
- For us to debug anything, dmesg logs of bootup with drm.debug=0x0e is a must.
please provide these logs, and based on that, we can analyze the next steps. 

- Shashank

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-gfx-bugs/attachments/20180417/d7f2d4b7/attachment-0001.html>