Bug #17238

Display flickering since Tails 4.0 (Intel DPST?)

Added by kdr4 2019-11-15 09:39:48 . Updated 2019-11-22 10:51:31 .

Status:
New
Priority:
Normal
Assignee:
Category:
Hardware support
Target version:
Start date:
Due date:
% Done:

0%

Feature Branch:
Type of work:
Research
Blueprint:

Starter:
Affected tool:
Deliverable for:

Description

I think Intel’s DPST (Display Power Saving Technology) has become activated on my device since using Tais 4.0. This results in a noticeable dimming and brightening of the display according to what’s on the screen, sometimes quite quickly so that it drags attention from what your doing. I think sometimes it may result in the entire screen going black very briefly (like 0.25 secs), I notice this sometimes. This behaviour was not present in 3.16 or before.

It is distracting and there are plenty of people that have complained about this when using Windows (see here, for example: https://dancharblog.wordpress.com/2019/04/14/disable-intel-dpst-adaptive-contrast-brightness-on-new-uwd-dch-drivers/ ).

Disabling in Windows requires a Registry change but is effective and survives reboots. It would be great to either disable it by default in Tails 4.0 or give users the option.


Subtasks


History

#1 Updated by intrigeri 2019-11-16 10:07:58

  • Category set to Hardware support

#2 Updated by intrigeri 2019-11-16 10:16:11

I’ve found no way to control this on Linux :/
Maybe you can disable this in the firmware/BIOS settings?

#3 Updated by kdr4 2019-11-22 08:40:01

Unfortunately not, and I’ve never come across any reports that this is something that is ever exposed in the firmware or BIOS. According to a Reddit thread I found (https://www.reddit.com/r/Surface/comments/c9ug2b/fyi_you_can_disable_dpst_intels_horrible/) there are several ways to do this on Windows: via the Intel Graphics Command Center available on the Microsoft Store; via third-party app/script DPST-Control; or directly via the Registry. Maybe these will help identify a way to address this on Linux.

This really constitutes a considerable regression since 3.16 and earlier, where DPST was not active on my device. Is there any way Tails could be rolled back to whatever driver it was using before?

#4 Updated by intrigeri 2019-11-22 10:51:31

Hi,

> Unfortunately not, and I’ve never come across any reports that this is something that is ever exposed in the firmware or BIOS.

:/

> According to a Reddit thread I found (https://www.reddit.com/r/Surface/comments/c9ug2b/fyi_you_can_disable_dpst_intels_horrible/) there are several ways to do this on Windows: via the Intel Graphics Command Center available on the Microsoft Store; via third-party app/script DPST-Control; or directly via the Registry. Maybe these will help identify a way to address this on Linux.

Yes, it would be nice if someone else than me tried (since I already tried and failed).

If the problem indeed comes from DPST, and if there’s indeed no way for a user to turn it off (we could document such a workaround), I doubt we at Tails can do much about it. Best would be to try and reproduce on a recent Linux distro (Debian unstable, latest Fedora or Ubuntu non-LTS) and report this directly to that distro or upstream: then you can provide whatever debugging info they’ll need, and test workarounds or candidate fixes they propose.

> Is there any way Tails could be rolled back to whatever driver it was using before?

From an engineering pov: I’m afraid this would require lots of work, both for the initial implementation and for regular maintenance.

From a user-centric PoV: chances are that such a downgrade would also bring back problems that 4.0 fixed for other Intel graphics hardware. At the moment we don’t know how many of our users are badly affected by this potentially-DPST-related regression so it’s hard to tell which way would serve the majority of our users best.