Display Artifacts

[m-weigand]

This is a meta-issue that tries to collect all relevant information regarding the various screen-corruption issues encountered on the Pinenote.

Report relevant information

• kernel version and source
• u-boot version
• desktop environment (i.e., Gnome or sway)
• before/after a suspend-cycle
• used waveform

[jazn]

• Kernel: Linux pineNote 5.17.0-rc6-next-20220304-g18d666ad5da0-dirty #1 SMP Fri Sep 30 21:19:30 CEST 2022 aarch64 GNU/Linux
• u-boot: orginal uboot patched to start linux automaticly
• DE: wayland/sway
• suspend cycle: After reboot works fine. Then after suspend, unsuspected cycle it works fine for a longer time (did not monitored that yet) until it starts to add random lines. Lines go away after forced refresh. Lines go away after suspend, unsuspected cycle. But after first time it reoccur in few minutes after clearing. After reboot it works fine for a longer time. Probably until next suspend, unsuspected cycle.
• waveform: orginal_firmware]$ sha256sum waveform.bin 987c3ce11b0ce17db6da2bce492bcc147c4473c252b5fff43a5048062b7b8939 waveform.bin

Sample image:

[Godzil]

How does the screen refresh are triggered? Does the app control the refresh or does the driver do refresh the screen at fixed interval or when it detect some changes in the framebuffer?

When we look on the 4th white line, on top of the word "Mezczyzna", we can see in the lines there are inverted letter in the "random" lines and it really look to me like when there is a synchronisation issues in the video out on a normal display.

With the driver can you force refresh the screen with a full flash?

How does the lines appears? Are they appearing during a refresh or do they just slowly appear on the screen with no clear explanations?

On the first case, something is wrong with the framebuffer, or random data are sent to the screen or there are sync issues (check the vsync/hsync signals) If it is the latter, the screen may not be properly DC balanced and the VCOM voltage could be wrong.

I had weird behaviours with eink screen in the past where I could make the ink migrate long after the last refresh on a completely unplugged screen because some charge were left on the display.

The DC imbalance could be an issue in Rockchip driver, have you had a look at the eink driver from brand that used the same rockchip hardware? I'm pretty sure Bookeen use the exact same hardware with their Notea, but they never released the GPL sources, and I'm pretty sure other people must have use the same SoC.

[jazn]

Kernel driver controls refresh.

I don't remember exactly but as I recall lines appeared on drawing elements, not slowly. This suggests synchronization issues, I agree.