X.Org 21.1.0 sees a release with Variable refresh rate support in the modesetting driver

I was under the impression that nobody would work on new X.org features anymore and it would even hardly be maintained?

Quoting: EikeI was under the impression that nobody would work on new X.org features anymore and it would even hardly be maintained?

I think that this is maintainers realizing some bugs can't just be left to rot and decided to patch up before they leave.

Quoting: TrainDocI think that this is maintainers realizing some bugs can't just be left to rot and decided to patch up before they leave.

Variable refresh rate support sounds like a major thing to me, though?

Quoting: Eike

Quoting: TrainDocI think that this is maintainers realizing some bugs can't just be left to rot and decided to patch up before they leave.

Variable refresh rate support sounds like a major thing to me, though?

But the question is, what did it improve?
VRR has been working on X for long time.. the only issue has been that it has been working only with 1 monitor active.
Does that change that? what the F is modesetting driver, what does it improve for the end user?

Quoting: Xpander

Quoting: Eike

Quoting: TrainDocI think that this is maintainers realizing some bugs can't just be left to rot and decided to patch up before they leave.

Variable refresh rate support sounds like a major thing to me, though?

But the question is, what did it improve?
VRR has been working on X for long time.. the only issue has been that it has been working only with 1 monitor active.

Oh man!
I've been actually using it on X already!
Yes, sorry, it's obviously not the big change I thought.

QuoteVariable refresh rate support in the modesetting driver.

It is probably improved by this:

• 1*) https://cgit.freedesktop.org/xorg/xserver/commit/?id=017ce263376aa64a495c4d71a140a24b1dff7054

and this:

• 2*) https://cgit.freedesktop.org/xorg/xserver/commit/?id=68f01c0f02ece4f6efe9ce18be81244246a1e114

Quoting: XpanderVRR has been working on X for long time.. the only issue has been that it has been working only with 1 monitor active.
Does that change that?

Yes, tldr:

• 1*) In vrr+nonvrr setups, in the past the vrr capability depended on the order you plug the monitors(!) (or the order in which the system probed them, so a matter of fluck), now it is handled correctly.


• 2*) Also, when refresh rates differ, it is not only possible to vsync all the outputs to a minimum common multiple (so no vrr) , but even vsync just the primary one to his own frequency and let others basically tear (the latter is the new feature).

Quotewhat the F is modesetting driver

From man modesetting:
"modesetting is an Xorg driver for KMS devices."

It is a generic driver for xorg that works across multiple gpus, given that the underlying kernel supports them via KMS/DRM.
Optionally, it can be 2D accelerated via Glamour if the dri driver implements opengl.
Take intel cards, if you use the xf86-video-modesetting driver, than xorg will re-use already available kernel calls to show things, if you use xf86-video-intel, than different calls will be performed.
It is also a thing that nvidia could just support modesetting driver and fully integrates into the kernel and still have his proprietary 3D blob living in the dri part of his driver (nvidia-utils under Arch).

-> This is what i've understood. <-


Last edited by kokoko3k on 28 October 2021 at 11:12 am UTC

Look forward to this in mainline arch repo. Its a feature I've been waiting for a long time and useful for me since I don't use Wayland yet due to some bugs and missing features.

Thanks for explanation.

The modesetting driver part i could search myself also, but my point actually was that there are so many little things. like if you search for what modesetting driver is, then you get anther set of questions. What is KMS what is DRM..then you search for this and even more new questions arise :)


Last edited by Xpander on 28 October 2021 at 11:18 am UTC

Quoting: XpanderThanks for explanation.

The modesetting driver part i could search myself also, but my point actually was that there are so many little things. like if you search for what modesetting driver is, then you get anther set of questions. What is KMS what is DRM..then you search for this and even more new questions arise :)

Lol, you're right, you need to get one big picture out of so many little bits.
Check this out:
https://www.bitwiz.org.uk/s/how-dri-and-drm-work.html

Quoting: Guest

Quoting: XpanderThanks for explanation.

The modesetting driver part i could search myself also, but my point actually was that there are so many little things. like if you search for what modesetting driver is, then you get anther set of questions. What is KMS what is DRM..then you search for this and even more new questions arise :)

Yes, that's how 'learning' works - step by step

Yeah, except their point is that it's a situation where you don't get to learn step by step. Learning about one step tells you nothing meaningful because it's presented in a way that leaves it meaningless unless you learn about this other step, which in turn tells you nothing unless you learn about still other steps.