[REGRESSION][BISECTED] intel iGPU with HDMI PLL stopped working at 1080p@120Hz 1efd5384

[REGRESSION][BISECTED] intel iGPU with HDMI PLL stopped working at 1080p@120Hz 1efd5384

[Vas Novikov]

[-- Attachment #1.1.1: Type: text/plain, Size: 2427 bytes --]

Hi dear LKML and community, this is my first post here, so I'd 
appreciate any guidance or redirection if it's due.

Starting from commit 
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1efd5384277eb71fce20922579061cd3acdb07cf,
HDMI handling for certain refresh rates on my intel iGPU is broken.

The error is still present in 6.16rc1.

Specifically, this is the command that disambiguates the newer broken 
kernels:

         xrandr --output HDMI-1 --auto --scale 1x1 --mode 1920x1080 
--rate 120 --pos 0x0 --output eDP-1 --off

The important parts are 1920x1080 and 120Hz. When run on commits prior 
to the bisected above, it behaves as expected, delivering 1920x1080 @ 
120Hz. When run on kernel builds with the above commit included (that 
commit or later), the monitor goes completely blank. After about 30 
seconds, it shuts down entirely (which I assume means that from the 
monitor's perspective, HDMI got "disconnected").

On this link you can see my original report in the ArchLinux community, 
where Christian Heusel (@gromit) kindly guided me through the bisection 
process and built the bisection images for me to try: 
https://gitlab.archlinux.org/archlinux/packaging/packages/linux/-/issues/145#note_277482 
  This link also contains the bisection history.

Additional info:

* The monitor and the notebook are connected via an HDMI cable, the 
monitor itself is a 4k@120Hz monitor.

* According to `lsmod | grep -E "(i915|Xe)"`, I'm using the i915 kernel 
driver for the GPU.

* The GPU is an iGPU from intel, specifically `Intel Core Ultra 7 155H`.

* One symptom that disambiguates the working and non-working kernels, 
besides whether they actually have the bug, is that the broken kernels 
cause xrandr to additionally report the 144.05 refresh rate as possible 
for the monitor, whereas the non-broken kernels consistently cause 
xrandr to only list refresh rate 120 and below as possible. I'm only 
ever testing the refresh rate of 120, but the presence of the 144.05 
rate is correlated.

If any other information or anything is needed, please write.


Thank you,
Vas

----

#regzbot introduced: 1efd5384277eb71fce20922579061cd3acdb07cf
#regzbot title: intel iGPU with HDMI PLL stopped working at 1080p@120Hz 
1efd5384
#regzbot link: 
https://gitlab.archlinux.org/archlinux/packaging/packages/linux/-/issues/145


[-- Attachment #1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 669 bytes --]

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 236 bytes --]

Re: [REGRESSION][BISECTED] intel iGPU with HDMI PLL stopped working at 1080p@120Hz 1efd5384

[Jani Nikula]

On Tue, 17 Jun 2025, Vas Novikov <vasya.novikov@gmail.com> wrote:
> If any other information or anything is needed, please write.

Does [1] help?

If not, please file a bug as described at [2], in particular attach
dmesg with debugs enabled all the way from boot to reproducing the
problem.

Thanks,
Jani.


[1] https://lore.kernel.org/r/20250613061246.1118579-1-ankit.k.nautiyal@intel.com
[2] https://drm.pages.freedesktop.org/intel-docs/how-to-file-i915-bugs.html


-- 
Jani Nikula, Intel

Re: [REGRESSION][BISECTED] intel iGPU with HDMI PLL stopped working at 1080p@120Hz 1efd5384

[Vas Novikov]

Hi Jani and everyone,

On 17/06/2025 12.33, Jani Nikula wrote:
> Does [1] help?

The patch works. (Applied on top of 6.16.0-rc2-1.1-mainline, built by 
Christian @gromit who helped again.)

The patch (or the new kernel) also have a side effect of xrandr allowing 
a completely new refresh rate, ~144Hz. This new refresh also seems to 
work (I cannot easily disambiguate 144 versus 120, but I can tell it's 
not 60Hz). So as far as my hardware is concerned, this patch leaves the 
whole system working in all scenarios that I've tested.

Thanks!


Kind regards,
Vas

Re: [REGRESSION][BISECTED] intel iGPU with HDMI PLL stopped working at 1080p@120Hz 1efd5384

[Jani Nikula]

On Tue, 17 Jun 2025, Vas Novikov <vasya.novikov@gmail.com> wrote:
> Hi Jani and everyone,
>
> On 17/06/2025 12.33, Jani Nikula wrote:
>> Does [1] help?
>
> The patch works. (Applied on top of 6.16.0-rc2-1.1-mainline, built by 
> Christian @gromit who helped again.)
>
> The patch (or the new kernel) also have a side effect of xrandr allowing 
> a completely new refresh rate, ~144Hz. This new refresh also seems to 
> work (I cannot easily disambiguate 144 versus 120, but I can tell it's 
> not 60Hz). So as far as my hardware is concerned, this patch leaves the 
> whole system working in all scenarios that I've tested.

Thanks a lot for testing! Ankit will send a v2 of it, and I think we'll
have it in mainline and backported to stable in a few weeks.

There's no need to file that bug report, this will suffice.


BR,
Jani.


-- 
Jani Nikula, Intel

Re: [REGRESSION][BISECTED] intel iGPU with HDMI PLL stopped working at 1080p@120Hz 1efd5384

[Nautiyal, Ankit K]

On 6/17/2025 7:20 PM, Jani Nikula wrote:
> On Tue, 17 Jun 2025, Vas Novikov <vasya.novikov@gmail.com> wrote:
>> Hi Jani and everyone,
>>
>> On 17/06/2025 12.33, Jani Nikula wrote:
>>> Does [1] help?
>> The patch works. (Applied on top of 6.16.0-rc2-1.1-mainline, built by
>> Christian @gromit who helped again.)
>>
>> The patch (or the new kernel) also have a side effect of xrandr allowing
>> a completely new refresh rate, ~144Hz. This new refresh also seems to
>> work (I cannot easily disambiguate 144 versus 120, but I can tell it's
>> not 60Hz). So as far as my hardware is concerned, this patch leaves the
>> whole system working in all scenarios that I've tested.
> Thanks a lot for testing! Ankit will send a v2 of it, and I think we'll
> have it in mainline and backported to stable in a few weeks.
>
> There's no need to file that bug report, this will suffice.


Thanks Vas for reporting and testing and Christian for the help with 
bisection/builds.

Thanks Jani for first pointing out the sparse warning, and then for 
quickly pointing out the fix.

I have sent the v2 of the fix [1].

[1] 
https://lore.kernel.org/all/20250618130951.1596587-2-ankit.k.nautiyal@intel.com/

Regards,

Ankit


>
>
> BR,
> Jani.
>
>