Intel Linux Graphics Driver Patches For Fast Soft-RC6 Yield Big Energy Use Improvement

[phoronix]

Phoronix: Intel Linux Graphics Driver Patches For Fast Soft-RC6 Yield Big Energy Use Improvement

Longtime open-source Intel Linux graphics driver developer Chris Wilson has sent out a set of 19 patches for what he calls fast soft-RC6 support and is a "substantial" improvement over the current driver code for Intel graphics power-savings...

http://www.phoronix.com/scan.php?page=news_item&px=Intel-Patches-Fast-Soft-RC6

[M@yeulC]

What is RC6? What GPU families are expected to benefit from this?

[liskin]

Don't expect anything exciting. It's just that the security fix for CVE-2019-0154 made the RC6 power saving state practically unavailable, meaning the GPU is always on and keeps the CPU from entering deeper package C-states, wasting 1-2 watts. Chris' fix aims to bring the old behaviour back. Upstream bug report with more details here: https://bugs.freedesktop.org/show_bug.cgi?id=112315

[brent]

So THAT is why battery life has been crap for me for the last few days. Sigh.

This regression is really bad, the workaround/fix should be included in 5.3.x and earlier stable kernels as well!

[liskin]

So THAT is why battery life has been crap for me for the last few days. Sigh.

I think it is okay-ish to temporarily revert the patch as it only fixes a Denial-of-Service:

The second issue, assigned CVE-2019-0154 and is rated Moderate, allows an unprivileged, local attacker to create an invalid state while the GPU is in low-power mode. This allows the system to become inaccessible and requires a system restart to resume normal operations.

[willmore]

Casting shade at xorg? Why, because they architected their driver poorly to start with?

[brent]

I think it is okay-ish to temporarily revert the patch as it only fixes a Denial-of-Service:

Sure, but I don't want to hack around with patched kernels. This shouldn't be necessary.