Re: AMDGPU - Regression: Black screen due to commit 16dc8bc27c2aa3c93905d3e885e27f1e3535f09a

[Mario Limonciello <superm1@kernel.org>]

On 6/21/25 11:07 AM, Walt Holman wrote:
> On Sat, Jun 21, 2025 at 11:03 AM Walt Holman <waltholman09@gmail.com> wrote:
>>
>> On Sat, Jun 21, 2025 at 10:52 AM Mario Limonciello <superm1@kernel.org> wrote:
>>>
>>>
>>>
>>> On 6/21/25 10:18 AM, Walt Holman wrote:
>>>> Hello,
>>>>
>>>> With the latest drm fixes this week on 6.16-rc2, I am experiencing a
>>>> black screen instead of the sddm greeter and the GPU appears to be
>>>> locked up. I can ssh into the laptop and reboot it, but that's about
>>>> it. I have bisected the commit to commit id:
>>>> 16dc8bc27c2aa3c93905d3e885e27f1e3535f09a and upon reverting the
>>>> commit, the system works as normal. The hardware is an Asus Rog
>>>> Zephyrus G16 with AMD Ryzen AI 9 HX 370 w/ Radeon 890M video. I'm able
>>>> to test patches etc.. if need be.
>>>
>>> Hi there,
>>>
>>> By chance do you have an OLED panel?  If so can you please try the patch
>>> attached to this bug?
>>>
>>> https://gitlab.freedesktop.org/drm/amd/-/issues/4338
>>>
>>> Thanks,
>>>
>>>>
>>>> 16dc8bc27c2aa3c93905d3e885e27f1e3535f09a is the first bad commit
>>>> commit 16dc8bc27c2aa3c93905d3e885e27f1e3535f09a (HEAD)
>>>> Author: Mario Limonciello <mario.limonciello@amd.com>
>>>> Date:   Thu May 29 09:46:32 2025 -0500
>>>>
>>>>       drm/amd/display: Export full brightness range to userspace
>>>>
>>>>       [WHY]
>>>>       Userspace currently is offered a range from 0-0xFF but the PWM is
>>>>       programmed from 0-0xFFFF.  This can be limiting to some software
>>>>       that wants to apply greater granularity.
>>>>
>>>>       [HOW]
>>>>       Convert internally to firmware values only when mapping custom
>>>>       brightness curves because these are in 0-0xFF range. Advertise full
>>>>       PWM range to userspace.
>>>>
>>>>       Cc: Mario Limonciello <mario.limonciello@amd.com>
>>>>       Cc: Alex Deucher <alexander.deucher@amd.com>
>>>>       Reviewed-by: Roman Li <roman.li@amd.com>
>>>>       Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
>>>>       Signed-off-by: Alex Hung <alex.hung@amd.com>
>>>>       Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
>>>>       Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
>>>>       (cherry picked from commit 8dbd72cb790058ce52279af38a43c2b302fdd3e5)
>>>>       Cc: stable@vger.kernel.org
>>>>
>>>>    drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 41
>>>> +++++++++++++++++++++++++++--------------
>>>>    1 file changed, 27 insertions(+), 14 deletions(-)
>>>>
>>>
>>
>> Yes, I do have an OLED panel and that patch does make it visible
>> again. It is still very dark, but visible.
>>
>> -Walt
> 
> Upon rebooting again, the display was much brighter and is fine. Just
> the first boot after that was dark.
> 
> -Walt

Thanks, this makes sense.  I suspect that because 0 means "off" for your 
panel that the patch you bisected to exposed running at 0 (or near 
enough to 0) that it caused this behavior.

That patch you tested will be upstreamed in a future promotion, and I 
think we can take it in a -fixes PR later in the 6.16 fixes cycle.

But there is one more thing I would like to confirm - on your system can 
you run with drm.debug=0x106 on the kernel command line and share me the 
output from this debug print?

https://github.com/torvalds/linux/blob/v6.16-rc2/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c#L4933