Microsoft Reworks The "DXGKRNL" Driver It Wants To Get Into The Linux Kernel

[Doomsdayrs]

It should be rejected for attempting to compromise FOSS standards.

Microsoft can keep it as their own proprietary kernel module. I do not see how Linux users, big or small, will ever benefit from this.

[Quackdoc]

The problem is that Windows drivers need to be signed and it gets harder and harder to install non-signed drivers. And signing drivers cost money. That's why you get signed virtio-drivers for Windows from RedHat only if you have a subscription.

So having something official would be great.

And due to the advancements in containerization my personal need for hardware-accellerated Linux VMs on Linux is nearing 0.

this really isn't an issue RHEL want's accelerated windows VMs as much as anyone else, fedora provides digitally signed drivers for free, many distros even do it too. Vulkan signed drivers are not an issue, sure it costs money, but it isn't a lot in the grand scheme of things, D3D10umd is more of an issue than anything because it's a pain in the arse to set it up.

[Quackdoc]

Until there is, these changes should not be mainlined. Simply put, does not contribute anything any of us need or want, and does nothing but encouraging migrating OFF linux.

this is completely false, before DXGKRNL was only compatible with microsoft's closed source d3d12 implementation. Hyper-v is still a competent hypervisor, this will be really good for linux guests running on hyper-v. the goal of linux isn't to take over the world. for servers running hyper-v this will be immensely useful.

and the reason why it would "encourage" users to migrate away is a bad point, no one will migrate away from linux except people who needed linux for their work. if you don't want people to migrate away the answer is making a good linux desktop which doesn't turn users away from it, which at this point seems to be a truly futile task.

[Charlie68]

To me, the things Microsoft is doing are of zero interest.

[Termy]

Until there is, these changes should not be mainlined. Simply put, does not contribute anything any of us need or want, and does nothing but encouraging migrating OFF linux.

This - all the way.

[Volta]

Typical Linux copium and whining with the "Oh boo hoo we are not capable enough to make something as good as DirectX by ourselves so Microsoft is bad and they must give away the API they spent billions to create and maintain for free because I say so"

Typical moron. Why we need helping m$ in the first place?

[Volta]

Hyper-v is still a competent hypervisor, this will be really good for linux guests running on hyper-v. the goal of linux isn't to take over the world. for servers running hyper-v this will be immensely useful.

And this is completely false.

[CochainComplex]

Typical Linux copium and whining with the "Oh boo hoo we are not capable enough to make something as good as DirectX by ourselves so Microsoft is bad and they must give away the API they spent billions to create and maintain for free because I say so"

Vulkan is as good as DirectX but has almost no lobbyism in the Gameengine Sector. Besides it was the FOSS community reverse engineering the whole DirectX API (plus windows itself) to make it run almost on par with windows performancewise without any documentation or help provided by MS.

[skeevy420]

if you don't want people to migrate away the answer is making a good linux desktop which doesn't turn users away from it, which at this point seems to be a truly futile task.

That won't happen until the kernel has a more stable ABI. IMHO, that should be the LTS releases. Let the stable, interim, releases break things with new drivers and changes, but have it all get back in line by the end of the year and the next LTS release.

It would also help if Red Hat/IBM would just drop GNOME and Ubuntu would drop Snaps. I say that because the two projects are major contributions from both those respective companies that have major, major push-back from the Linux community at large. Last time I saw the numbers, GNOME barely had a 25% market share and it's the desktop shipped with nearly every mainstream distribution. KDE has the same with market share while being shipped as the default from fewer distributions. After KDE, it's environments like Mate, Cinnamon, and XFCE -- or, What are Linux desktops that look and feel like older Windows desktops? for $500.


GNOME thinks they're all slick by shipping a minimized desktop and letting the user configure it with plugins. But first the user might have to install a package or two, will have to visit the right web site, will have to install a browser extension, and finally install their plugins just to then have to be vigilant by keeping up with versioning dependencies and ensuring that plugins don't conflict. That requires the user to RTFM and then some. The "minimized desktop and configure it with plugins" concept sounds nice when discussing it. It truly does. It just isn't very nice in actual use.

Snaps are similar -- Ubuntu has a large market share yet many of their users (and forks) opt out of Snaps for Flats. Why? Flats are more universal, seem to have more support from random developers, and the both have the same downside -- you gotta wait a bit when you run the first program so the runtime loads. Pro-tip -- put a Snap/Flat like Yakuake or Guake in your desktop auto-start list so you hide the runtime start with your logon tasks and you get a drop down terminal. It really doesn't help Snaps that Ubuntu likes to drop projects. When will they drop Snaps? Why get into the habit of using Snaps if Ubuntu will drop it? Especially if a competitor exists and is used by a company like Valve for Steam.

GNOME and Snaps both double efforts and their end-goals of unifying and simplifying the Linux desktop ecosystem have the opposite effects as intended -- they complicate things, fracture communities, and double up efforts for end-users and developers alike.

GNOME and Snaps represent major Linux companies completely ignoring users, current market trends, and how people who use computers want to use their computers.

IMHO, If Red Hat wanted to help they'd fork Qt6 and turn it into GTK5 while turning GNOME into a KDE feature like -- Settings > Appearance > Desktop Shell. If those were done then damn near every desktop environment could re-write themselves as a KDE Desktop Shell which would greatly unify the Linux Desktop.

If Ubuntu wants to help: quit trying to be special with NIH. Y'all have the Linux niche reputation as Google Quitters. Frigging do something and stick with it....just not Snaps or any other problem that has been solved in some form or fashion.

If sure the reverse of all of that can be said about KDE and Flats. That's just how it works. I'm sure one or more of y'all will let me know

Well, that covers my desktop ranting for the next week or so

For the love of God don't any of y'all reply to this in full. There was too much text the first time

[Lbibass]

It should be rejected for attempting to compromise FOSS standards.

Microsoft can keep it as their own proprietary kernel module. I do not see how Linux users, big or small, will ever benefit from this.

People who want to use linux, but maybe don't have the know-how, or simply don't want to install linux on their own hardware outside of a VM. If someone, lets say a student, is taking two different classes, one where they have to have access to machine learning tools that are only available on linux, but also needs to have Solidworks for a design class, this would be perfect.

So I can see its use cases. In addition, having this in mainline would help encourage participation from the rest of the FOSS community. I see nothing but good coming out of this.