Bug 250072 - graphics/drm-fbsd12.0-kmod: panic: !list_empty(&fence->cb_list) failed
Summary: graphics/drm-fbsd12.0-kmod: panic: !list_empty(&fence->cb_list) failed
Status: New
Alias: None
Product: Ports & Packages
Classification: Unclassified
Component: Individual Port(s) (show other bugs)
Version: Latest
Hardware: Any Any
: --- Affects Only Me
Assignee: freebsd-x11 (Nobody)
Keywords: panic
Depends on:
Reported: 2020-10-03 13:30 UTC by Pascal Abresch
Modified: 2020-10-05 15:55 UTC (History)
1 user (show)

See Also:

core file text (407.33 KB, text/plain)
2020-10-03 13:30 UTC, Pascal Abresch
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Pascal Abresch 2020-10-03 13:30:15 UTC
Created attachment 218490 [details]
core file text

Occured while running hikari on X11 with love2d and love2d on X11 directly
Comment 1 Mark Linimon freebsd_committer freebsd_triage 2020-10-04 03:43:31 UTC
From original Summary:

BUG ON !list_empty(&fence->cb_list) failed at /wrkdirs/usr/ports/graphics/drm-fbsd12.0-kmod/work/kms-drm-99da0ba/linuxkpi/gplv2/include/linux/dma-fence.h:91
Comment 2 Niclas Zeising freebsd_committer 2020-10-05 08:01:54 UTC
Hikari is a Wayland compositor.  How can you run that on X11?
Comment 3 Niclas Zeising freebsd_committer 2020-10-05 08:02:41 UTC
Duplicate of 250070 ?
Comment 4 Jan Beich freebsd_committer 2020-10-05 14:07:24 UTC
(In reply to Niclas Zeising from comment #2)
> Hikari is a Wayland compositor.  How can you run that on X11?

Via wlroots X11 backend. It works similar to Xephyr/Xnest but hardware-accelerated: running Hikari on X11 would create a window where one can manage Wayland and X11 clients according to Hikari rules. Such nesting isn't limited by depth but at some point performance may degrade e.g., on Skylake under Xorg "cage cage cage cage vkcube-wayland" runs with 60FPS but adding another "cage" drops to 45FPS.

See also https://old.reddit.com/r/linux_gaming/comments/iudig1/gamescope_continues_advancing_as_waylandvulkan/

(In reply to Niclas Zeising from comment #3)
> Duplicate of 250070 ?

Likely different panic/backtrace on the same radeonkms version. Confirming on 5.4-lts would be nice but until drm-fbsd12.2-kmod or similar appears it'd require upgrading to -CURRENT.
Comment 5 Pascal Abresch 2020-10-05 15:55:47 UTC
I opened this second ticket becase the error message is more explicit, i have quite a number of page faults but this one had a specific assert beeing triggered, so i do not consider it a dupe. :)