Hi,
Any clue/hint/fix regarding X11 flickering cursor ? HWCursor ?
rock-5b-debian-bullseye-xfce4-arm64-20221122-1158-gpt.img.xz / 5.10.110-34-rockchip-gca15bbe36e6c
Alex
Hi,
Any clue/hint/fix regarding X11 flickering cursor ? HWCursor ?
rock-5b-debian-bullseye-xfce4-arm64-20221122-1158-gpt.img.xz / 5.10.110-34-rockchip-gca15bbe36e6c
Alex
https://www.armbian.com/rock-5b/
Use armbian jammy + panfork-mesa instead. It works just fine.
Same here on jammy
Cursor is blinking for me aswell when running glmark for example, but only when 3d operations are going on. With 2d tuxes it does not flash. Using todayâs panfork.
I can confirm, latest armbian panfork, it is still doing this.
Registering the hardware cursor in dtb can solve this problem
Based on https://github.com/rockchip-linux/kernel/commit/473a8fb13d5845b42151d045cc48284607ab0f38
I did https://github.com/ChisBread/linux-orangepi/commit/63eb2dd31f886dcdb3f8d4d595c41f2a12afce2f#
for orange pi 5
I believe it also works on rock 5b
definitely worth trying ! thanks for your feedback
Iâve actually gone and tried this after digging through the driver and dts tree to see why no hardware cursor was available.
I suspect it works fine if youâre only running one or maybe two displays, but it fails with 3.
Hi bread, not familiar enough with linux to be able to patch it myself. Do you think these updates will make it to regular jammy updates eventually?
Waiting for this PR to be merged, but maybe it will be closed https://github.com/radxa/kernel/pull/50
I have X11 working and i donât have the flickering cursor. I always thought this issue would be on Wayland only.
In what circumstance this happens?
We can have two scenarios:
Update:
I currently use ssh session to do most of the things, and not always at the desktop. But now paying attention to the cursor, i can see the cursor flickering a bit while moving it, but almost imperceptible, is that the flickering you are experiencing?
My scenario is 1)
Can you disclose what displays you are using and which port is connected to?
2 HP 27q displays, 2560x1440, attached to both HDMI ports.
A no-name 1920x1080 display, connected to the USB-C port with a dongle.
Mind sharing a picture of the connection?
Here is my cursor: https://drive.google.com/file/d/1iQ9H_gYYxe1uMjp7EFPmjZ9cXJMldUkp/view?usp=sharing
works perfectly, may requires some adjustments like Sarah said, but clearly fixes the issue for now in a basic dual screen setup.
https://drive.google.com/drive/folders/155CXeKE0Yc71yHKKLDGxIQ8JDoXaInBY?usp=sharing
Pictures are my monitor setup and the Rock5b with itâs connections. Video is (poorly shot) how the screen responds when the cursor properties are enabled.
Connections:
Left display: USB-C dongle. 1920x1080.
Center display: right HDMI port. 2560x1440 with rotation.
Right display: left HDMI port. 2560x1440 without rotation.
All displays are 60hz, though I donât suspect that makes too much difference here.
The log is the kernel & system messages, if thatâs helpful.
Essentially: USB-C and left HDMI fail to output video - but the cursor is perfectly smooth. The center display, which was misconfigured because I couldnât see to login and signed into a second account on the machine, displayed my desktop otherwise correctly, until I moved the cursor to those other displays.
Ha ha, I got dizzy following your cursor.
You are on Wayland. I donât have any experience with Wayland. vop2 kept crashing i think. The entire screen flickered.
I can only comment on your boot log, if you allow me, that does not look good. I think you need at least 27W. I donât push my Rock 5B that much. I can now use the 65W PD on my setup.
Nope, thatâs Xorg. Wayland had its own issues - maybe recent changes have fixed this, but GNOME would shit a brick starting a Wayland session.
Edit: maybe that user was Wayland. I couldnât see which account I signed in with. (boot.log says it was.)
Iâll double-check again once I completely disable Wayland on here.
As far as the PSU goes: thatâs a 65W USB-C brick from a laptop. sensors
reports that itâs configured for 58W, which makes sense as some of it is drained off the top by the adapter in the USB-C port.
Last up on here: Yep, it was Wayland. gdm is starting up with Wayland enabled, leaving me with mangled displays. Dropping into a console and using âstartxâ, it came up fine. Oh well, itâs not like I ever got Wayland working properly anyway - even when I got close, there were things that absolutely didnât work.
Ok. You should check for the latest xserver fix, but i think you could try some other kernel and see if does not crash that much, eventually, i will try dual screen, but i canât help much.
Yep, see my last edit. I read my log and found that it was, in fact, Wayland. I went to a console and explicitly ran âstartxâ (since the text-mode console worked just fine) and it all came up and is working.