On Sat, Mar 09, 2024 at 02:42:25PM +1300, Avon Robertson wrote: > I apologise for making this thread so disjointed. I need to > find a method to include long mails such as this to a previous > short post. > > The serial output from a boot of the kernel that includes > Jonathon's reverting diff is below.
This diff should avoid the deref you saw, but I fear it will just move the problem to another function. Index: sys/dev/pci/drm/amd/pm/powerplay/hwmgr/smu7_clockpowergating.c =================================================================== RCS file: /cvs/src/sys/dev/pci/drm/amd/pm/powerplay/hwmgr/smu7_clockpowergating.c,v diff -u -p -r1.1 smu7_clockpowergating.c --- sys/dev/pci/drm/amd/pm/powerplay/hwmgr/smu7_clockpowergating.c 7 Jul 2021 02:38:30 -0000 1.1 +++ sys/dev/pci/drm/amd/pm/powerplay/hwmgr/smu7_clockpowergating.c 12 Mar 2024 04:33:25 -0000 @@ -115,7 +115,8 @@ void smu7_powergate_uvd(struct pp_hwmgr { struct smu7_hwmgr *data = (struct smu7_hwmgr *)(hwmgr->backend); - data->uvd_power_gated = bgate; + if (data) + data->uvd_power_gated = bgate; if (bgate) { amdgpu_device_ip_set_powergating_state(hwmgr->adev,