[Bug 200645] 4.18-rc regression bisected to e03fd3f30: amdgpu polaris11/rx460 only activates on one output/monitor of two

[bugzilla-daemon at bugzilla.kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=200645

--- Comment #7 from Duncan (1i5t5.duncan at cox.net) ---
So out of curiosity I found my old DVI-D/DisplayPort adapter and tried
switching outputs.

First thing I noticed is that the DVI/HDMI cable I'm using is only single-link
DVI, as is the DVI/DisplayPort adapter, so if I put the 4k monitor on it, it's
automatically limited to 1080p.

Which means that until I go buy a new cable, the only way I have to connect the
4k monitor and /get/ 4k is via the HDMI/HDMI cable and the card's HDMI
connector.

That leaves the DVI or DisplayPort card connector for the 1920x1080 TV.  I can
connect it directly to the card DVI-D using the HDMI/DVI-D cable, as I had it
when I first filed this bug, or add the DVI-D/DisplayPort adaptor as well, and
connect it to the card's DisplayPort, as I have it now.

Either way, connecting the 1920x1080 via DVI-D or DisplayPort, the bug still
manifests -- without the cap to 8bpc, trying to boot 4.18-rc1+ blanks that
monitor as soon as the drmfb code takes over from the early-boot/VGA, while the
HDMI-connected 4k monitor continues to work normally (at 1080p until X starts
due to the video= kernel commandline option, at native 3840x2160 in X).

And the problem could also be partially due to the single-link cable I'm still
using for that monitor, enough for 8bpc 1920x1080, but perhaps not enough for
10bpc, even if both ends support it (I'm not actually sure whether the 1080p
limited TV does or not).

-- 
You are receiving this mail because:
You are watching the assignee of the bug.