I had delayed downloading this update for a couple of weeks, but now I had to do it and I find out this.
It seems optimus-manager is broken in Wayland, so I tried installing plasma-x11-session
And yet the problem remains. It seems the latest optimus-manager is borked in every possible way, so not even working with xorg!
I have tested optimus-manager --print-mode
The resutls are pretty much the same:
ERROR: the latest GPU setup attempt failed at Xorg pre-start hook. Log at /var/log/optimus-manager/switch/switch-20260114T204339.log Cannot execute command because of previous errors.
The log says this:
[17] INFO: # Xorg pre-start hook [17] INFO: Previous state was: {'type': 'pending_pre_xorg_start', 'requested_mode': 'nvidia', 'current_mode': None} [18] INFO: Requested mode is: nvidia [42] INFO: Available modules: ['nouveau', 'bbswitch', 'acpi_call', 'nvidia', 'nvidia_drm', 'nvidia_modeset', 'nvidia_uvm'] [42] INFO: Unloading modules ['nouveau'] (if loaded) [45] INFO: Loading module bbswitch [51] INFO: Setting GPU power to ON via bbswitch [80] INFO: Loading module nvidia [1311] ERROR: Xorg pre-start setup error Traceback (most recent call last): File "/usr/lib/python3.13/site-packages/optimus_manager/kernel.py", line 245, in _load_module subprocess.check_call( ~~~~~~~~~~~~~~~~~~~~~^ f"modprobe {module} {' '.join(options)}", ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ shell=True, text=True, stderr=subprocess.PIPE, stdout=subprocess.DEVNULL) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.13/subprocess.py", line 419, in check_call raise CalledProcessError(retcode, cmd) subprocess.CalledProcessError: Command 'modprobe nvidia NVreg_UsePageAttributeTable=1' returned non-zero exit status 1. The above exception was the direct cause of the following exception: Traceback (most recent call last): File "/usr/lib/python3.13/site-packages/optimus_manager/hooks/pre_xorg_start.py", line 51, in main setup_kernel_state(config, prev_state, requested_mode) ~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.13/site-packages/optimus_manager/kernel.py", line 22, in setup_kernel_state _nvidia_up(config, hybrid=(requested_mode == "hybrid")) ~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.13/site-packages/optimus_manager/kernel.py", line 95, in _nvidia_up _load_nvidia_modules(config, available_modules) ~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.13/site-packages/optimus_manager/kernel.py", line 164, in _load_nvidia_modules _load_module(available_modules, "nvidia", options=nvidia_options) ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/lib/python3.13/site-packages/optimus_manager/kernel.py", line 249, in _load_module raise KernelSetupError(f"Error running modprobe for {module}: {e.stderr}") from e optimus_manager.kernel.KernelSetupError: Error running modprobe for nvidia: None [1318] INFO: Removing /etc/X11/xorg.conf.d/10-optimus-manager.conf (if present) [1318] INFO: Writing state {'type': 'pre_xorg_start_failed', 'switch_id': '20260114T204339', 'requested_mode': 'nvidia'}
Tried a few things, asked AI for help and ultimately decided to try envycontrol. But to do that I will have to completely uninstall optimus-manager. I wonder what is wrong with it, why can’t the devs keep things working correctly?
WIsh me luck, I will update this post. If envycontrol fixes this then at least things will be functional, but I guess I will still miss some of the features that optimus-manager provided.
optimus-manager is not in the repos and is thus not supported by Manjaro. It is an AUR package:
pamac search optimus
nvidia-exec-git v0.2.9-1 AUR
Run programs in nvidia optimus setups with power management for Xorg
and Wayland without log out
nvidia-exec 0.3.2-1 AUR
Run programs in nvidia optimus setups with power management for Xorg
and Wayland without log out
dl-switch 1-1 AUR
Easy way to switch between displaylink and optimus
ttf-optimusprinceps 1.0-2 AUR
Optimus Princeps font by Manfred Klein
plymouth-theme-optimus-git r28.32a0d92-1 AUR
The plymouth theme collection by adi1090x
plasma6-applets-optimus-gpu-switcher-git r138.a298529-1 AUR
KDE Plasma widget to change the GPU mode using the EnvyControl command
line tool.
plasma6-applets-optimus-gpu-switcher 1.0.2-1 extra
KDE Plasma widget to change the GPU mode using the EnvyControl command
line tool.
optimus-manager-qt-kde 1.6.9-1 AUR
A Qt interface for Optimus Manager that allows to configure and switch
GPUs on Optimus laptops using the tray menu KDE
optimus-manager-qt-git 1.6.6.r0.g98d630c-1 AUR
A Qt interface for Optimus Manager that allows to configure and switch
GPUs on Optimus laptops using the tray menu
optimus-manager-qt 1.6.9-3 AUR
A Qt interface for Optimus Manager that allows to configure and switch
GPUs on Optimus laptops using the tray menu
optimus-manager-git 4:0-1 AUR
Allows using Nvidia Optimus laptop graphics
The only optimus package in the repos, plasma6-applets-optimus-gpu-switcher, has envycontrol as a required dependency:
pamac info plasma6-applets-optimus-gpu-switcher
Name : plasma6-applets-optimus-gpu-switcher
Version : 1.0.2-1
Description : KDE Plasma widget to change the GPU mode using the
EnvyControl command line tool.
URL : https://github.com/enielrodriguez/optimus-gpu-switcher
Licenses : GPL-3.0-or-later
Repository : extra
Installed Size : 75.6 kB
Groups : --
Depends On : envycontrol plasma-workspace
Optional Dependencies : libnotify: desktop notifications [Installed]
zenity: display GTK+ dialogs [Installed]
Provides : --
Replaces : plasma5-applets-optimus-gpu-switcher
Conflicts With : plasma5-applets-optimus-gpu-switcher
Packager : Mark Wagie <mark@manjaro.org>
Build Date : Tue 02 Apr 2024 01:56:02
Validated By : MD5 Sum SHA-256 Sum Signature
Hopefully envycontrol will turn out to be a suitable alternative for you.
I did what I said I would do, uninstalled optimus thing and installed envycontrol. Rebooted, now shows in systray, but it’s still in intel mode. I use the envycontrol option in systray to switch to nvidia, ot asks for password and then says I must reboot. So I reboot to… black screen. Now what? I need to use this PC is the only one I have!
I’ve updated the topic title to more accurately reflect the issue that you have described in your posts.
I can’t help with the graphics issue as I have no experience with Nvidia. However, to assist those who are able to offer help, some system info may be useful.
Well, to revert things I had to CTRL+ALT+F2 , login to the terminal there and issue this command:
sudo envycontrol -s hybrid (Thanks to AI I was able to get out of there)
However I’m still puzzled, why the full nvidia mode boots to black screen? Anyways, with that I’m back to the uncertainty. Even though the system works, it doesn’t give me full Nvidia, only hybrid or intel, unlike optimus-manager, that one was giving me full Nvidia mode, I wonder what’s wrong this time? It feels like a downgrade anyways.
Results of inxi -zv8 (I’m only including what I consider necesary)
Man, this hits me hard, I have been using this laptop with Manjaro for years and I will need to keep using it for many years, I just hope things will keep working.
Thanks for that, but could you please do it properly?
[MiniHowTo] Present code or command output
Note: Grave accent (or, backtick) characters are used.
For inlinecode or to show a single command:
please follow this simple example: code here.
For multi-linecommand output:
place text beween two rows of three (```) backtick characters to form a scrollable text box, or …
use the </> button in the Compose window to achieve the same result.
The empty row between these rows of backtick characters is where to paste any multi-line command output (such as inxi).
An example of a pre-formatted text enclosure:
What is a Quark?
A. In Physics, an elementary particle and fundamental constituent of matter.
B. A character from the Star Trek: "Deep Space Nine" television series.
C. A soft, creamy, usually unsalted cheese traditional to central Europe.
That driver is no longer supported. That’s why you get a black screen. You should switch to the 390 or 470 series.
Well, come next year you will have to switch to a different desktop environment then, because Plasma 6.8 will be dropping all X11 support. That, or stick with the Intel GPU.
Uhm… quite strange, for some reason the text I paste looks like that even with the pre-formatted text enclosure. What could I be doing wrong? I select the text and press the button and that’s the result. If I press the button without selecting text, the whole post becomes enclosed by the dark rectangle. I had to switch from the "rich text” to the “standard markdown editor” to see those backtick characters, the rich text doesn’t seem to work the way I expected.
Results of inxi -zv8 (I’m only including what I consider necesary)
That only works if you are in the standard markdown mode, in rich text mode if you do that you get the results I was getting.
Well, in order to be user friendly and user intuitive, consistency is key. It should work in a similar way to quoting, but it doesn’t.
I didn’t choose that version, I chose to install the DKMS version (nvidia-dkms) and that thing seems to have automatically compiled the 550-580 version. Also, it doesn’t make sense to me that you say lower number versions are more up to date, are you kidding?
That sounds like you telling me I’m screwed just for existing. Anyways, I have managed to get it working on Wayland, but the question still remains, why full Nvidia mode is gone.
I didn’t say they are more up-to-date. What I said was that the 550 series isn’t supported anymore — it’s not even in the repos anymore — and I’m not sure whether the 570 series still supports such an old GPU as what’s in your machine.
You could try the 570 driver series, though.
Ha, aren’t we all?
Look, I didn’t make those decisions, and neither did Manjaro. It was KDE themselves who have announced that Plasma 6.8 won’t be supporting X11 anymore.
Wayland is the future, and it’s already the default display server protocol in Plasma as of the release of Plasma 6. GNOME has already dropped X11 right now. Xfce still supports it, because Xfce is very conservative, and its Wayland support is still experimental.
It’s because the old Nvidia drivers do not support Wayland. Only the latest Nvidia drivers do, but those in turn do not support your GPU.
In other words, yes, you’re screwed. By Nvidia, not by us. But if it’s any consolation, you don’t really need the proprietary Nvidia drivers, because nouveau — which is a Free & Open Source driver in the kernel itself — will do for most use cases, and your Intel driver will be even better, because unlike nouveau, it didn’t have to be reverse-engineered.
Granted, I am not a gamer — I don’t know whether you are, so please don’t put any words in my mouth — but my now almost 7-year-old desktop system here has an onboard Intel GPU, and it has served me well so far. I can watch movies in UHD (1920 x 1080), and there’s no shortage of snappiness on account of the desktop effects in Plasma.
So unless you are a gamer, you’ll do perfectly fine just using the Intel GPU in your machine.
Yes I’m a gamer and I play on Linux in 10 year old Nvidia hardware.
nouveau devs should outsmart Nvidia devs and do better reverse engineering or something then?
if we want to get to have the year of Linux in the desktop I mean.
It’s hard to reverse-engineer something of which the specifications are being kept secret.
Intel and AMD both offer adequate performance with Free & Open Source drivers. Nvidia is the only player in the GPU field that refuses to open up their specs.
But we don’t want that. The whole thing is a red herring, because GNU/Linux has never been about market shares or competition.
That said, GNU/Linux currently has a desktop market share of well over 6% already, which isn’t all too bad for an operating system that — with the exception of a small handful of computer brands — doesn’t come preinstalled and isn’t being advertised.
The fact that GNU/Linux runs about 85% of the internet — with the remainder primarily taken up by other UNIX-family operating systems like BSD and a couple of still proprietary UNIX versions like AIX, HP/UX and Solaris — and that it runs on all of the world’s top-500 supercomputers, as well as that it is the dominant guest operating system on Microsoft’s own Azure cloud, should say more about Microsoft’s marketing strategy of pushing Windows down everybody’s throat than about the qualities of GNU/Linux.
And the irony is that people who buy one of those preinstalled-with-Windows computers think that they are getting Windows for free, while the OEM license is actually included in the price of the machine.
Given everything I said in the previous paragraph, I don’t see GNU/Linux as a failure at all. In fact, if anything, the fact that Microsoft Windows can only be successful in one particular market segment — and only so because it comes preinstalled on most new desktop and laptop computers — says more than enough about what a poor excuse of an operating system Windows really is.
And if you then start looking at how Windows works under the hood, and how it was put together, that’s when the real fun begins.
Linus Torvalds succeeded because he achieved that, he built things that were even better than his competitors, exceeded the performance of Windows in server hardware (hence why Windows Server failed and Linux is king on servers)
Then you fail to understand reality itself. Market is a reflection of reality itself, if you are good, you steamroll there and everywhere else, if you aren’t then you struggle to survive or just fail. If we keep this mindset we will effectively become obsolete and completely useless. Just wait until all the good engineers / devs leave or get so old that they have to retire.
With a mindset like yours, Linux will die in a few decades.
It’s not only gaming what I do here!
What about other video intensive activities? Recording or editing video with KdenLive, etc.
What if I need to do some renderings? I have Blender installed!
You will force people like me back to Windows!
Apples and oranges. Linus wasn’t interested in competition.
Linus was using Minix on his HP Vectra 386 computer, and Minix is a microkernel-based UNIX-style operating system of which the source code was open but read-only, and at the time, it only existed in a 16-bit version for the Intel 8086/8088 processor family. Linus wanted to explore the power of his Intel 80386 processor, but the Minix license forbade him from modifying the source code, and that’s why he started writing his own kernel, but without making it a microkernel.
It was just a hobby project at first, and Linus has said many times that had GNU been completely production-ready at the time — meaning with its own GNU HURD microkernel, which still isn’t production-ready today — or if 386BSD had been available already, then he would never have even bothered, because all he wanted was a working 32-bit UNIX system.
Even though Linus had already ported a couple of GNU tools to his new kernel — because those tools were readily available and freely licensed — he then posted about his endeavor in the comp.os.minix Usenet discussion group, asking for anyone interested to pitch in. The GNU people were also monitoring said group, and they immediately contacted Linus and started porting just about the entire GNU userland to the Linux kernel.
That’s how the ball got rolling, and when Linus then relicensed his kernel under the GNU GPLv2, the Free Software Foundation — the organization behind GNU — considered its objective attained: they could now offer a fully functional and fully freely licensed UNIX-family operating system to whoever wanted it.
No, I’m afraid the failure to understand would be at your own hands, then.
I’m guessing that you haven’t read a word of what I said with regard to supercomputers, or the fact that 85% of the internet runs on GNU/Linux, or the fact that GNU/Linux currently has a desktop market share of 6% — and it continues growing! — in spite of the lack of advertising or being pushed down people’s throats like those Windows OEM licenses.
That’s what the Windows fanboys have been telling us GNU/Linux advocates for decades already, yes. And now they’re eating crow.
I hope they have some salt and pepper handy.
In order to use blender successfully, you need a much more powerful machine than the one you have. They recommend at least 32 GiB of RAM.
As for kdenlive, I have used kdenlive and rendered a video on a machine with only 4 GiB of RAM, two processor cores, and a built-in Radeon GPU with only 128 MiB of video memory. So you don’t need Nvidia for that.
If people like you return to Windows, then it’s not because anybody’s forcing them, but because they are Windows users at heart, who only seek out GNU/Linux because of Microsoft’s cessation of support for older Windows versions and older computers, or because of some other Microsoft-related idiosyncrasy.
They are not coming to GNU/Linux out of an interest in the operating system itself, and they are full of very flawed preconceived notions about what GNU/Linux is, why it was created, and what its objectives are.
Either way, returning to the topic, you have been presented with potential solutions, and now it’s up to you to at least give them a try. I will sum them up for you again…:
Install the 390 or 470 series drivers. They are still supported, while the driver you’re using right now — i.e. the 550 series — is not.
You could try the 570 driver, but there’s no guarantee it’ll work. As I understand it, it does not support old GPUs like the one in your machine anymore. But you could give it a try.
This version is the rolling version - which currently is at 590 - which dropped support for three (3) GPU generations.
The script nvidia-driver-assistant is likely to suggest the 575xx driver for your GPU but if you prefer dkms to rebuild on kernel update - that is also an option
For your system I will suggest nvidia-570xx-dkms[-open] or nvidia-575xx-dkms[-open] (optionally the open driver) which should work for your GPU.
For switching GPU, I think envycontrol is a better choice, but you may also have succes with the nvidia-prime package which enables you to run select applications on the Nvidia GPU.
You also have the choice of using mhwd to install the hybrid driver
I’m on almost the same hardware (on xfce/x11), 4th gen Intel/Nvidia 860M hybrid with envycontrol. The nvidia 860M works fine with the 575xx driver.
To get envycontrol working best use Manjaro Settings Manager to check for installed hybrid setups first, most likely video-hybrid-intel-nvidia-470xx-prime , remove it and install video-hybrid-intel-nvidia-575xx-prime .
Whatever anyone comes and does, people should be free to do it. Linux was about freedom after all. Wheter we like it or not, we exist in a world where the only way to make things work is the market, and GNU / Linux distros market strategy has usually been a Pay what you want business model that is similar to freemium models where paid users sustain things for everyone else. But paying is not mandatory, as it’s voluntary via donations. That’s one of the things that makes the whole thing so awesome! Anyone is free to use the software!
Hasn’t this been the situation with most hardware always? Isn’t most hardware compatibility on Linux reverse engineered? You misunderstand a basic principle here. Have we forgotten the time when Torvalds himself flipped the finger at Nvidia? Lots of things have happened since then, and at least they are giving us proprietary drivers for free, they could be giving us nothing nada! You are taking that for granted! (In fact I think I may have been making the same mistake of taking it for granted some years ago).
Open hardware could be a long term solution, but it needs to grow more, that implies things such as investment, market promotion, making it competitive, etc. Heck! I would not be using an Nvidia video card if I a better Open Source option existed, but it doesn’t exist! Also, I could be using AMD but since it’s not as popular as Nvidia, it was not an available option in the local market for me to buy!
Anyways, there will always be people who want to use their hardware (whatever it is) and not use Microsoft Windows. There’s nothing wrong with reverse engineering, thanks to it most hardware can actually work on Linux. Linux should remain a viable option for anyone, not something exclusively for a very small niche of people, that was a good beginning, but staying there implies the dangers of becoming obsolete and fading away, because that’s how reality works.
All in all, I have been worried about people like me being able to keep using their hardware for a long time (specially Nvidia due to how often I have had problems with it) and I have been thinking about this.
Nouveau is clearly not the most used solution, and the reason is how it fails to work even with some older GPUs. You want the Nouveau project to thrive don’t you? At least I would like to see it thrive, but for that things must change.
The solution is definitely NOT to force people to use their hardware crippled by incomplete software. If there’s any real solution, It definitely requires people in the community changing that mindset and finding out better solutions. I’m glad you bring this to the topic because it gives me a chance to present some ideas I have been thinking about.
A possible solution does imply having a better reverse engineering. The reason it’s hard to reverse engineer a graphics card is not that specifications are kept secret (absurd argument, you reverse engineer anything precisely because there are no open source specifications), it’s that the hardware complexity has grown exponentially to the point that a single engineer can’t do everything by himself! An engineer doesn’t even have to outsmart another engineer, he has to outsmart teams of several well paid engineers in large companies!
And yet, what Nvidia is constructing with all the effort they are putting dedicated to AI developement is precisely what could help us! AI! We, the community, could benefit from AI to reverse engineer any hardware! Not only GPUs! Reverse engineering has been of great importance to make Linux work with most hardware, but while it’s a hard process, AI is the perfect tool to automate it and make it progress faster! AI won’t replace engineers, it’s just a tool, it lacks a will of it’s own.
So, Instead of having the Nouveau project stagnate for years trying to solve making some GPUs work, that time could be reduced to only a few months! However that does imply having people in the community organize to make it work.
Open source AI models are already a thing, so, the Nouveau engineers could create a model of their own, that helps them with the reverse engineering process, that would offload some of their work, reducing their stress and leverage their engineering capabilities so they can at least have a chance of outsmarting Nvidia engineers. It would be an adittional effort that could actually be a gamechanger! Why? Because once the Nouveau engineers are able to crack the code and make better drivers, that would also enable them to make another AI that helps them build an open hardware alternative to both AMD and Nvidia! Alas, I can’t do this myself, I don’t have the resources, but I do have these ideas that have potential to help the community, and I hope that anyone reading this can understand it.
One of the reasons I think this could work, is how AI has been helping me to do many things. For example I got out of the black screen problem quickly because I asked AI.
Without AI I would have been wandering the web, wondering if there was a way to revert the problem with envycontrol not knowing what command put on the terminal. Maybe I would have tried reinstalling the distro or maybe even rethink my options and go back to distro hopping out of rage and frustration. AI helped me calm down and showed me that the solution was a single command away, instead of sitting there puzzled and in despair.
AI being helpful however, didn’t imply replacing my interactions with people in this forum.
I’m thankful for your replies guys, AI can’t do everything, but it could be the key to keep Linux working in the long term future.
Back to the main topic:
I have installed the DKMS version because it’s required by Lutris. Since I had followed the instructions here:
When I installed this instance of Manjaro on this laptop, I remember I had followed the instructions there
And I had used the provided command for Arch based distros that was given there to install everything back then:
Thanks for pointing that out. I chose nvidia-dkms because at that moment it seemed to be the best option, I would have imagined that the devs would have done the best effort to make the rolling release option the most compatible one and to keep all the old hardware working for as long as possible without bothering the users to switch versions, but I can imagine that the decision to drop 3 GPU generations came from Nvidia themselves.
Thanks for pointing that out, knowing that someone else with similarly old hardware is still able to make it work is a relief. It was specially useful helping me decide to go through the rabbit hole of hell of fixing the next problems I had.
I tried that command and found out how screwed I was. It returned a bunch of errors. I was on a loop of dependency deadlocks, because I have lots of stuff installed in this laptop. So, to get out of this, I also had some help of AI.
The conversation would be lost, but I still have the text, as I copied it to save it.
If you want me to share it and how it helped me, I would like to, but this post has already become too long.
They are. But what they generally seem to forget is that with great freedom comes great responsibility.
Not that kind of freedom. It’s about freedom in the sense of licensing.
No. The Linux kernel — which is where hardware support lives — has always been sponsored by some of the biggest names in the industry, from Intel and AMD over Adaptec, QLogic, Oracle (and Sun Microsystems before it was acquired by Oracle), Hewlett-Packard, SGI (Silicon Graphics), IBM, you name it.
Of course not. But from what you write farther down, it would appear that you would be the one misinterpreting things.
Nvidia was already supplying Linux drivers long before Linus flipped them the bird, even going back way over 20 years. But they have always been proprietary drivers.
And before nouveau, there was nv, which Nvidia sponsored, albeit that it was barely usable at all — no acceleration whatsoever — and then Nvidia pulled out of nv. That’s why a couple of people began developing nouveau as an alternative.
What happened at one point was that Linus chose to change the kernel interface in such a way that anything interfacing with the kernel directly had to do so by way of something written as Free & Open Source, even if only for debugging reasons.
Nvidia refused to comply, which is why Linus flipped them the bird.
I agree. I’d much rather be running a RISC-V workstation than an x86-64 machine. But alas, capitalism has reserved machines like that for the wealthy, and even then still, not every GNU/Linux distribution supports it.
It wasn’t meant to be elitist or exclusive — even though my own comments are often misinterpreted along that same line — but someone new to GNU/Linux should at least have the willingness to learn and to get to know the system, and to keep it maintained.
If you want to drive a car, then you need to get a driver’s license, and then once you own your car, there are traffic regulations you should adhere to. Knowing how to ride a bicycle isn’t quite the same thing as driving a car, and especially not when it’s a Formula One car — which, compared to the offerings from Microsoft and Apple, GNU/Linux more or less is.
Ideally we should be allowed to use our hardware for as long as it’s still functional. But both Apple and Microsoft have found ways to force us into buying newer hardware and contributing to the growing pile of e-waste.
Actually, no. I’d rather go with something that’s fully supported, i.e. Intel or AMD.
Yes. It takes a lot of effort from many qualified people to do things right, and even then, in the USA there is such a thing a software patents — the EU does not recognize patents on software — which could make life very difficult for the engineers.
This is one of the reasons why the GNU-proper kernel — the HURD — is still not production-ready, by the way. It’s a microkernel design based upon GNU Mach, but the problem is that GNU Mach could not use the methods used in the original Mach kernel because of patents. So they had to work around those, with as a result that GNU Mach became slow as molasses.
