Summary: | vt video corruption when switching from xserver to tty | ||||||
---|---|---|---|---|---|---|---|
Product: | Base System | Reporter: | Damien <nixuser1980> | ||||
Component: | kern | Assignee: | freebsd-bugs (Nobody) <bugs> | ||||
Status: | New --- | ||||||
Severity: | Affects Only Me | CC: | emaste, manu, nixuser1980, ota | ||||
Priority: | --- | Keywords: | vt | ||||
Version: | CURRENT | ||||||
Hardware: | Any | ||||||
OS: | Any | ||||||
Attachments: |
|
Description
Damien
2019-02-10 07:44:12 UTC
Created attachment 201884 [details]
Corruption to vt under i915kms after switching from xserver
Additional Information: I can confirm that this only occurs when switching from X server to tty without shutting down X first. If X server is shut down, when the switch to the tty occurs, any corruption is now gone. Also, if the corruption persists due to X server being shut down forceably i.e. through kill etc., restarting X and then shutting it down gracefully prevents the corruption from being displayed when it X server terminates. It should be noted that no corruption is ever seen when the tty is switched to the active X server. I've not observed this on my X220 w/ i915 graphics. What's the version of your X server and drm kmod packages? I recently started using a very old laptop and also saw video corruption, too. I haven't care much about it but seeing this made me curious how this incident will go and if I also get benefit, too... I can reproduce this on my new laptop, Lenovo X1 Carbon 7th Generation, with: $ pciconf -lv | grep -B2 -A2 Graphics vgapci0@pci0:0:2:0: class=0x030000 rev=0x02 hdr=0x00 vendor=0x8086 device=0x3ea0 subvendor=0x17aa subdevice=0x2292 vendor = 'Intel Corporation' device = 'UHD Graphics 620 (Whiskey Lake)' class = display subclass = VGA There is a few stuff worth to try: - Use 13.0-CURRENT with the latest drm-current-kmod (based on Linux 5.2) - Set hw.i915kms.enable_fbc=0 in loader.conf - Set hw.i915kms.enable_psr=0 in loader.conf I don't have the issue on all my intel systems so that might take a while to debug ... |