And here's the drm.debug=0xff output leading to the oops:
[ 109.575582] [drm:drm_stub_open] [ 109.575609] [drm:drm_open_helper] pid = 2430, minor = 0 [ 109.575630] [drm:ipp_subdrv_open] done priv[0xed9b7e10] [ 109.575647] [drm:drm_setup] [ 109.575699] [drm:drm_ioctl] pid=2430, dev=0xe200, auth=1, DRM_IOCTL_VERSION [ 109.575719] [drm:drm_ioctl] pid=2430, dev=0xe200, auth=1, DRM_IOCTL_VERSION [ 109.575764] [drm:drm_release] open_count = 1 [ 109.575786] [drm:drm_release] pid = 2430, device = 0xe200, open_count = 1 [ 109.575810] [drm:drm_lastclose] [ 109.575856] [drm:drm_atomic_state_alloc] Allocate atomic state edbdb100 [ 109.575876] [drm:drm_atomic_get_crtc_state] Added [CRTC:21] ed10d600 state to edbdb100 [ 109.575893] [drm:drm_atomic_get_plane_state] Added [PLANE:17] edbdb900 state to edbdb100 [ 109.575906] [drm:drm_atomic_set_crtc_for_plane] Link plane state edbdb900 to [NOCRTC] [ 109.575918] [drm:drm_atomic_set_fb_for_plane] Set [NOFB] for plane state edbdb900 [ 109.575933] [drm:drm_atomic_add_affected_connectors] Adding all current connectors for [CRTC:21] to edbdb100 [ 109.575947] [drm:drm_atomic_check_only] checking edbdb100 [ 109.575965] [drm:drm_atomic_commit] commiting edbdb100 [ 109.575996] Unable to handle kernel NULL pointer dereference at virtual address 000000a4
I've tried to do this in userspace by running modetest through gdb, but I can't seem to break when the segfault occurs (modetest is already dead when gdb attempts to break).
I hope this makes it easier to isolate the issue.
With best wishes, Tobias