[REGRESSION] drm/amd/display: backlight brightness set to 0 at amdgpu initialization

[Mario Limonciello]

On 7/19/25 10:25 AM, Mario Limonciello wrote:
> 
> 
> On 7/18/25 9:12 PM, Lauri Tirkkonen wrote:
>> On Fri, Jul 18 2025 20:14:08 -0500, Mario Limonciello wrote:
>>> OK, I think we need to do two things to figure out what's going on.
>>>
>>> 1) Let's shift over to 6.16-rc6.  Once we've got a handle on the 
>>> situation
>>> there we can iron out if there are other patches missing or this is also
>>> broken for you in 6.16.  If it's not working as expected there either we
>>> need it fixed there first anyway.
>>
>> Same behavior on 6.16-rc6: brightness is set to 0 (max 399000),
>> minimally visible.
>>
>>> 2) The starting brightness I don't expect to be "0".  We need to see 
>>> what
>>> values were read out from the firmware. There is a debugging message 
>>> we can
>>> catch if you boot with drm.debug=0x106.  Keep in mind you probably 
>>> need to
>>> increase log_buf_len if your ring buffer is set too small too.
>>>
>>> https://github.com/torvalds/linux/ 
>>> commit/4b61b8a390511a1864f26cc42bab72881e93468d
>>>
>>> PS: I would rather you add logs into a gist, pastebin or a bug 
>>> somewhere if
>>> you can.
>>
>> [    3.210757] amdgpu 0000:03:00.0: 
>> [drm:amdgpu_dm_connector_late_register [amdgpu]] Backlight caps: min: 
>> 1000, max: 400000, ac 100, dc 32
>>
>> full dmesg: https://termbin.com/o2q3
>>
> 
> Thanks for sharing.  Can you get me an updated output with 
> drm.debug=0x106 set and with this applied?  I want to see what values 
> were set.
> 
> diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/ 
> drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
> index 5df7f8c34231..cdc43cfb39dd 100644
> --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
> +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
> @@ -10489,8 +10489,12 @@ static void amdgpu_dm_atomic_commit_tail(struct 
> drm_atomic_state *state)
>          /* restore the backlight level */
>          for (i = 0; i < dm->num_of_edps; i++) {
>                  if (dm->backlight_dev[i] &&
> -                   (dm->actual_brightness[i] != dm->brightness[i]))
> +                   (dm->actual_brightness[i] != dm->brightness[i])) {
> +                       drm_WARN(adev_to_drm(adev), true,
> +                                       "Backlight level %d does not 
> match actual brightness %d for edp %d\n",
> +                                       dm->brightness[i], dm- 
>  >actual_brightness[i], i);
>                          amdgpu_dm_backlight_set_level(dm, i, dm- 
>  >brightness[i]);
> +                   }
>          }
> 
>          /*
> 
> Also, does turning off custom brightness curves 
> (amdgpu.dcdebugmask=0x40000) help?
> 

In advance of getting that updated log; I have a theory what's going on. 
  I think the first value programmed to brightness happens as part of 
that very first modeset.

If that's what's going on, then I think doing an explicit programming 
cycle at the backlight registration will help.  Here's a potential patch.

diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c 
b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index b19e7964060d..4b99efbaf481 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -4996,6 +4996,8 @@ amdgpu_dm_register_backlight_device(struct 
amdgpu_dm_connector *aconnector)
                 dm->backlight_dev[aconnector->bl_idx] = NULL;
         } else
                 drm_dbg_driver(drm, "DM: Registered Backlight device: 
%s\n", bl_name);
+
+ 
amdgpu_dm_backlight_update_status(dm->backlight_dev[aconnector->bl_idx]);
  }

  static int initialize_plane(struct amdgpu_display_manager *dm,