https://bugzilla.kernel.org/show_bug.cgi?id=215436
--- Comment #26 from spasswolf@web.de --- For the extra call hpd_state = 0: [ 30.684201] wlp4s0: deauthenticating from 54:67:51:3d:a2:e0 by local choice (Reason: 3=DEAUTH_LEAVING) [ 30.794571] amdgpu 0000:08:00.0: amdgpu: Calling optimize_bandwidth from dc_commit_state_no_check [ 30.794609] amdgpu 0000:08:00.0: amdgpu: Calling update_clocks from dcn20_optimize_bandwidth [ 30.794613] amdgpu 0000:08:00.0: amdgpu: adev->in_s0ix = 0, hpd_state = 1 [ 35.026604] PM: suspend entry (s2idle) [ 35.256153] Filesystems sync: 0.229 seconds [ 35.257044] Freezing user space processes ... (elapsed 0.002 seconds) done. [ 35.259636] OOM killer disabled. [ 35.259636] Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done. [ 35.260642] printk: Suspending console(s) (use no_console_suspend to debug) [ 35.264212] amdgpu 0000:08:00.0: amdgpu: amdgpu_pmops_suspend: adev->in_s0ix = 1 [ 35.264369] amdgpu 0000:08:00.0: amdgpu: calling update_clocks from dcn21_optimize_pwr_state [ 35.264379] amdgpu 0000:08:00.0: amdgpu: adev->in_s0ix = 1, hpd_state = 0 [ 35.264381] amdgpu 0000:08:00.0: amdgpu: calling rn_vbios_smu_set_dcn_low_power
So one could go back to the old version of rn_update_clocks with if (display_count == 0 && !hpd_state) {