Issue or bug concerning DRRS on old i7-4980HQ

Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

Hello all.

Addresses as per MANTAINERS, and Ville as per Git history...

I've been trying to assist a new Linux user with an old 2015 Alienware 
17 R2, i7-4980HQ (Iris Pro 5200) together with NVIDIA GTX980M.

The user experienced flickering (in a brightness sense) and had traced 
this back to Intel DRRS. I supplied a simple test patch as

--- linux-6.8.0/drivers/gpu/drm/i915/display/intel_dp.c.orig	
+++ linux-6.8.0/drivers/gpu/drm/i915/display/intel_dp.c
@@ -2727,6 +2727,8 @@
  {
  	struct drm_i915_private *i915 = to_i915(connector->base.dev);

+	return false;
+
  	if (pipe_config->vrr.enable)
  		return false;

which does indeed work. Set out to then solve it right(er) and ran into 
the debugfs drrs ctl file, for him

/sys/kernel/debug/dri/1/crtc-0/i915_drrs_ctl

We haven't however been able to do anything useful with that. If we echo 
0 into it it seems that DRRS does temporarily go to "DRRS active: no" in 
the corresponding i915_drrs_status file, but only seconds later it again 
shows as "DRRS active: yes", and his problem returns/remains.

It's making the system largely unusable with Linux -- or are we/am I 
missing something as to configuration?

The user is as said a new Linux user and will need to stick with generic 
distribution kernels, Ubuntu here, but the debugfs thing seems to then 
not work and a kernel parameter appears to not exist. When searching I 
found this issue also at for example

https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1928138

and on the list at

https://lists.freedesktop.org/archives/intel-gfx/2017-October/145050.html

Is there something we/he could do to get DRRS disabled with a generic 
distribution kernel? Should the i915_drrs_ctl thing have worked?

And if not, is there a possibility for a kernel parameter that does 
something akin to that above working test patch (but then right)?

Kind regards,
Rene Herman

✗ LGCI.VerificationFailed: failure for Issue or bug concerning DRRS on old i7-4980HQ

[Patchwork]

== Series Details ==

Series: Issue or bug concerning DRRS on old i7-4980HQ
URL   : https://patchwork.freedesktop.org/series/152362/
State : failure

== Summary ==

Address 'rene.herman@gmail.com' is not on the allowlist, which prevents CI from being triggered for this patch.
If you want Intel GFX CI to accept this address, please contact the script maintainers at i915-ci-infra@lists.freedesktop.org.
Exception occurred during validation, bailing out!

Re: Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

Here's that Ubuntu launchpad bug on freedesktop.org, but including a 
possibly interesting fix from 5 months ago in the currently last comment:

https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441

When Windows 10 goes out of support I expect a fair few of these older 
systems (supposedly without TPM) will get Linufied.

Regards,
Rene

On 31-07-2025 04:49, René Herman wrote:

> Hello all.
> 
> Addresses as per MANTAINERS, and Ville as per Git history...
> 
> I've been trying to assist a new Linux user with an old 2015 Alienware 
> 17 R2, i7-4980HQ (Iris Pro 5200) together with NVIDIA GTX980M.
> 
> The user experienced flickering (in a brightness sense) and had traced 
> this back to Intel DRRS. I supplied a simple test patch as
> 
> --- linux-6.8.0/drivers/gpu/drm/i915/display/intel_dp.c.orig
> +++ linux-6.8.0/drivers/gpu/drm/i915/display/intel_dp.c
> @@ -2727,6 +2727,8 @@
>   {
>       struct drm_i915_private *i915 = to_i915(connector->base.dev);
> 
> +    return false;
> +
>       if (pipe_config->vrr.enable)
>           return false;
> 
> which does indeed work. Set out to then solve it right(er) and ran into 
> the debugfs drrs ctl file, for him
> 
> /sys/kernel/debug/dri/1/crtc-0/i915_drrs_ctl
> 
> We haven't however been able to do anything useful with that. If we echo 
> 0 into it it seems that DRRS does temporarily go to "DRRS active: no" in 
> the corresponding i915_drrs_status file, but only seconds later it again 
> shows as "DRRS active: yes", and his problem returns/remains.
> 
> It's making the system largely unusable with Linux -- or are we/am I 
> missing something as to configuration?
> 
> The user is as said a new Linux user and will need to stick with generic 
> distribution kernels, Ubuntu here, but the debugfs thing seems to then 
> not work and a kernel parameter appears to not exist. When searching I 
> found this issue also at for example
> 
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1928138
> 
> and on the list at
> 
> https://lists.freedesktop.org/archives/intel-gfx/2017-October/145050.html
> 
> Is there something we/he could do to get DRRS disabled with a generic 
> distribution kernel? Should the i915_drrs_ctl thing have worked?
> 
> And if not, is there a possibility for a kernel parameter that does 
> something akin to that above working test patch (but then right)?
> 
> Kind regards,
> Rene Herman

Re: Issue or bug concerning DRRS on old i7-4980HQ

[Jani Nikula]

On Thu, 31 Jul 2025, René Herman <rene.herman@gmail.com> wrote:
> Here's that Ubuntu launchpad bug on freedesktop.org, but including a 
> possibly interesting fix from 5 months ago in the currently last comment:
>
> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441

Might be useful to attach dmesg and VBT to the bug as described at [1].

BR,
Jani.


[1] https://drm.pages.freedesktop.org/intel-docs/how-to-file-i915-bugs.html

-- 
Jani Nikula, Intel

Re: Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

On 01-08-2025 11:19, Jani Nikula wrote:

> On Thu, 31 Jul 2025, René Herman <rene.herman@gmail.com> wrote:
>> Here's that Ubuntu launchpad bug on freedesktop.org, but including a
>> possibly interesting fix from 5 months ago in the currently last comment:
>>
>> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441
> 
> Might be useful to attach dmesg and VBT to the bug as described at [1].

Just in case: please note the freedesktop.org bug is not our bug but 
just a 4 year old same one. I also unfortunately don't have the hardware 
on hand, and it seems the person I am assisting with this can't due to 
vacation get to it right now.

Although I'm not a graphics/display person, seems to me it's more or 
less fixed by Arun Murthy (added to the To: list) in the last, still 
recent-ish comment on that bug?

https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441#note_2803858

I.e., could while working with the person with the hardware not figure 
out what the point of that i915_drrs_ctl would be if the system 
immediately enabled it again anyway, so Arun's approach seems to make 
sense to me.

This is probably going to be a more common issue soon when these older 
no-TPM Haswell systems get installed with Linux at the end of Windows 10 
support.

Regards,
Rene

Re: Issue or bug concerning DRRS on old i7-4980HQ

[Jani Nikula]

On Sat, 02 Aug 2025, René Herman <rene.herman@gmail.com> wrote:
> On 01-08-2025 11:19, Jani Nikula wrote:
>
>> On Thu, 31 Jul 2025, René Herman <rene.herman@gmail.com> wrote:
>>> Here's that Ubuntu launchpad bug on freedesktop.org, but including a
>>> possibly interesting fix from 5 months ago in the currently last comment:
>>>
>>> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441
>> 
>> Might be useful to attach dmesg and VBT to the bug as described at [1].
>
> Just in case: please note the freedesktop.org bug is not our bug but 
> just a 4 year old same one. I also unfortunately don't have the hardware 
> on hand, and it seems the person I am assisting with this can't due to 
> vacation get to it right now.
>
> Although I'm not a graphics/display person, seems to me it's more or 
> less fixed by Arun Murthy (added to the To: list) in the last, still 
> recent-ish comment on that bug?
>
> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441#note_2803858
>
> I.e., could while working with the person with the hardware not figure 
> out what the point of that i915_drrs_ctl would be if the system 
> immediately enabled it again anyway, so Arun's approach seems to make 
> sense to me.

Hacking debugfs is not a fix, it's a hack.

> This is probably going to be a more common issue soon when these older 
> no-TPM Haswell systems get installed with Linux at the end of Windows 10 
> support.

On the upside, we'll get more people with the hardware to test.


BR,
Jani.


-- 
Jani Nikula, Intel

Re: Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

On 04-08-2025 10:21, Jani Nikula wrote:

> On Sat, 02 Aug 2025, René Herman <rene.herman@gmail.com> wrote:

>> I.e., could while working with the person with the hardware not figure
>> out what the point of that i915_drrs_ctl would be if the system
>> immediately enabled it again anyway, so Arun's approach seems to make
>> sense to me.
> 
> Hacking debugfs is not a fix, it's a hack.

So I take it that's a no then on looking at/fixing the DRRS bug? Please 
don't get me wrong; as said to Arun as well, I have no experience with 
either this code or really anything concerning this technology even: I 
am assuming I'm missing something when I couldn't figure out why the 
DRRS disable CTL existed when the system activated it again immediately 
anyway.

I am/was trying that help that user, and maybe you guys as to having 
this issue on the map; that already reported bug is 4 year's old, so it 
didn't seem to otherwise have a lot of traction.

Kind regards,
Rene

Re: Issue or bug concerning DRRS on old i7-4980HQ

[Jani Nikula]

On Mon, 04 Aug 2025, René Herman <rene.herman@gmail.com> wrote:
> On 04-08-2025 10:21, Jani Nikula wrote:
>
>> On Sat, 02 Aug 2025, René Herman <rene.herman@gmail.com> wrote:
>
>>> I.e., could while working with the person with the hardware not figure
>>> out what the point of that i915_drrs_ctl would be if the system
>>> immediately enabled it again anyway, so Arun's approach seems to make
>>> sense to me.
>> 
>> Hacking debugfs is not a fix, it's a hack.
>
> So I take it that's a no then on looking at/fixing the DRRS bug?

All I'm saying is that any fix must not depend on the user having to use
a debugfs or a module parameter. It needs to work out of the box.

BR,
Jani.


> Please 
> don't get me wrong; as said to Arun as well, I have no experience with 
> either this code or really anything concerning this technology even: I 
> am assuming I'm missing something when I couldn't figure out why the 
> DRRS disable CTL existed when the system activated it again immediately 
> anyway.
>
> I am/was trying that help that user, and maybe you guys as to having 
> this issue on the map; that already reported bug is 4 year's old, so it 
> didn't seem to otherwise have a lot of traction.
>
> Kind regards,
> Rene

-- 
Jani Nikula, Intel

Re: Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

On 04-08-2025 17:42, Jani Nikula wrote:

> On Mon, 04 Aug 2025, René Herman <rene.herman@gmail.com> wrote:

> All I'm saying is that any fix must not depend on the user having to use
> a debugfs or a module parameter. It needs to work out of the box.

Makes sense to me. Now, while repeating me being clueless about this 
code and even technology outright, I've wondered if the thing to do 
might even just be refusing to enable/use DRRS on this old Haswell-era 
hardware

Probably not a theoretically/technically sound/interesting  plan -- but 
it sure would solve things for both me/us and the Ubuntu/freedesktop.org 
bug-reporter, and frankly, I don't think anyone on a system like that in 
2025 would know or care in the first place. They just very much mind the 
flickering...

Maybe stupid; maybe regresses a ton of Haswell-era systems on which it 
does work fine. No idea...

Regards,
Rene

Re: Issue or bug concerning DRRS on old i7-4980HQ

[Jani Nikula]

On Mon, 04 Aug 2025, René Herman <rene.herman@gmail.com> wrote:
> On 04-08-2025 17:42, Jani Nikula wrote:
>
>> On Mon, 04 Aug 2025, René Herman <rene.herman@gmail.com> wrote:
>
>> All I'm saying is that any fix must not depend on the user having to use
>> a debugfs or a module parameter. It needs to work out of the box.
>
> Makes sense to me. Now, while repeating me being clueless about this 
> code and even technology outright, I've wondered if the thing to do 
> might even just be refusing to enable/use DRRS on this old Haswell-era 
> hardware
>
> Probably not a theoretically/technically sound/interesting  plan -- but 
> it sure would solve things for both me/us and the Ubuntu/freedesktop.org 
> bug-reporter, and frankly, I don't think anyone on a system like that in 
> 2025 would know or care in the first place. They just very much mind the 
> flickering...
>
> Maybe stupid; maybe regresses a ton of Haswell-era systems on which it 
> does work fine. No idea...

There's been a bunch of updates on DRRS in the past year, and I have yet
to see a dmesg with debugs enabled reproducing the issue on a recent
kernel. We need data before doing anything else. The discussion in this
thread is not actionable. For all I know the issue could've been fixed
already.

BR,
Jani.

-- 
Jani Nikula, Intel

Re: Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

On 04-08-2025 18:03, Jani Nikula wrote:

> There's been a bunch of updates on DRRS in the past year, and I have
> yet to see a dmesg with debugs enabled reproducing the issue on a
> recent kernel. We need data before doing anything else. The
> discussion in this thread is not actionable. For all I know the
> issue could've been fixed already.
I'll try and talk the (new to Linux) user with the hardware through 
installing a current mainline kernel, but expect this may at least need 
to wait for him to be back from that mentioned vacation. Will follow up 
if there's anything to report.

Regards,
Rene.

Re: Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

On 04-08-2025 18:11, René Herman wrote:

> I'll try and talk the (new to Linux) user with the hardware through 
> installing a current mainline kernel, but expect this may at least
> need to wait for him to be back from that mentioned vacation. Will
> follow up if there's anything to report.
We managed to install the 6.16 mainline kernel and generate the 
requested dmesg and VBT logs, with the requested "drm.debug=0xe 
log_buf_len=4M ignore_loglevel" kernel parameters. They are attached at

https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441#note_3064300

Hope this helps. Many thanks for looking.

Regards,
Rene

Re: Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

On 21-08-2025 21:24, René Herman wrote:

> We managed to install the 6.16 mainline kernel and generate the 
> requested dmesg and VBT logs, with the requested "drm.debug=0xe 
> log_buf_len=4M ignore_loglevel" kernel parameters. They are attached at
> 
> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441#note_3064300
> 
> Hope this helps. Many thanks for looking.

Hi all.

Was this useful?

Kind regards,
Rene.

Re: Issue or bug concerning DRRS on old i7-4980HQ

[Murthy, Arun R]

On 02-08-2025 16:06, René Herman wrote:
> On 01-08-2025 11:19, Jani Nikula wrote:
>
>> On Thu, 31 Jul 2025, René Herman <rene.herman@gmail.com> wrote:
>>> Here's that Ubuntu launchpad bug on freedesktop.org, but including a
>>> possibly interesting fix from 5 months ago in the currently last 
>>> comment:
>>>
>>> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441
>>
>> Might be useful to attach dmesg and VBT to the bug as described at [1].
>
> Just in case: please note the freedesktop.org bug is not our bug but 
> just a 4 year old same one. I also unfortunately don't have the 
> hardware on hand, and it seems the person I am assisting with this 
> can't due to vacation get to it right now.
>
> Although I'm not a graphics/display person, seems to me it's more or 
> less fixed by Arun Murthy (added to the To: list) in the last, still 
> recent-ish comment on that bug?

This was a debug patch provided to root case the issue and not a proper fix!

Thanks and Regards,
Arun R Murthy
--------------------

>
> https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/3441#note_2803858
>
> I.e., could while working with the person with the hardware not figure 
> out what the point of that i915_drrs_ctl would be if the system 
> immediately enabled it again anyway, so Arun's approach seems to make 
> sense to me.
>
> This is probably going to be a more common issue soon when these older 
> no-TPM Haswell systems get installed with Linux at the end of Windows 
> 10 support.
>
> Regards,
> Rene
>

Re: Issue or bug concerning DRRS on old i7-4980HQ

[René Herman]

On 04-08-2025 10:48, Murthy, Arun R wrote:

> On 02-08-2025 16:06, René Herman wrote:

>> Although I'm not a graphics/display person, seems to me it's more
>> or less fixed by Arun Murthy (added to the To: list) in the last,
>> still recent-ish comment on that bug?
> 
> This was a debug patch provided to root case the issue and not a
> proper fix!
OK, but it's a real question I have, not meant flippantly or anythng: 
what is then the idea of the DRRS CTL when it's enable/activated again 
immediately anyway? I'm not at all familiar with this code or even 
technology; I assume I'm missing something.

Regards,
Rene

Re: Issue or bug concerning DRRS on old i7-4980HQ

[Jani Nikula]

On Mon, 04 Aug 2025, René Herman <rene.herman@gmail.com> wrote:
> On 04-08-2025 10:48, Murthy, Arun R wrote:
>
>> On 02-08-2025 16:06, René Herman wrote:
>
>>> Although I'm not a graphics/display person, seems to me it's more
>>> or less fixed by Arun Murthy (added to the To: list) in the last,
>>> still recent-ish comment on that bug?
>> 
>> This was a debug patch provided to root case the issue and not a
>> proper fix!
> OK, but it's a real question I have, not meant flippantly or anythng: 
> what is then the idea of the DRRS CTL when it's enable/activated again 
> immediately anyway? I'm not at all familiar with this code or even 
> technology; I assume I'm missing something.

The debugfs is mainly for testing, where the scenarios that re-enable
DRRS are under control.

BR,
Jani.


-- 
Jani Nikula, Intel