Solved Should I remove NVIDIA drivers on an Intel GPU (HP computer)?

Solved issue

909mjolnir

Active Member
Joined
Mar 20, 2025
Messages
112
Reaction score
93
Credits
1,059
UPDATE: Solved overall. Not really a problem either way. Context was better explained by others in a different yet similar post. Sorry about the fuss.

Code:
I looked in octopi (pacman) on my Manjaro XFCE/MATE system (HP laptop computer) and I have NVIDIA drivers installed for some reason.
Is this odd?  My computer has Intel hardware and absolutely no NVIDIA hardware.  Could I delete the drivers and have it still run?

Thanks in advance. 

I took a look at HardInfo2;
I have an Intel GeminiLake [Mesa Intel UHD Graphics 600]; the in use driver is i915. 
How do I remove the NVIDIA stuff?
 
Last edited:


UPDATE: Okay, so I bit the bullet and deleted the NVIDIA system files with sudo pacman -Rndd on the individual names after I verified that the Intel video drivers were still installed. Then I rebooted and the system still works. So I guess it's not so bad.

I'm getting some wierd GPE type mismatch errors in dmesg and on screen during boot, but from what I've read that's because NVIDIA delivered partially broken driver(s) that don't match the ACPI specifications. That's why I ventured into this. The errors on my non-quiet boot screen were neon aqua colored (bluish green, teal). I also get those on shutdown all of a sudden. I hadn't had these warnings until a recent system update.

Maybe I won't worry about it. What do you think?
 
Normally the kernel only loads the drivers it needs, but this isn't always 100% true,
It doesn't hurt anything having them on there, but it doesn't really help either.

Normally nvidia drivers don't get installed by default.
 
If you have the drive space just leave them.
 
Thanks for the replies. I'm thinking about what you guys said.

I ended up deleting them before reading this forum.
Everything still works so maybe I didn't really need them.
I don't usually delete drivers at all.

I also deleted some kind of NVIDIA binary which is responsible for downloading more NVIDIA stuff.
It said so in it's description within Octopi. So maybe this is a good thing that I did this.

My whole system is Intel graphics / HP so it's just wierd to get other stuff specific to the GPU.
I can still watch videos and everything so I don't notice any loss.

Anyways thanks.
 
I forgot to mention, I'm still getting some kind of GPE type mismatch ACPI warnings during bootup. But my system starts up so fast I can't read the messages. But they aren't the normal ones, and they are neon aqua (bluish/greenish) and not the normal [OK] stuff loaded, types of messages.
 
that's odd - wonder how the drivers got there if you dont have the hardware. Manjaro is an Arch derived distro and I know Arch will install the nvidia drivers if that setting is enabled in archinstall... still, odd.
 
I forgot to mention, I'm still getting some kind of GPE type mismatch ACPI warnings during bootup. But my system starts up so fast I can't read the messages. But they aren't the normal ones, and they are neon aqua (bluish/greenish) and not the normal [OK] stuff loaded, types of messages.
A lot of logs can be viewed as @CaffeineAddict has outlined in post #8.

Whilst there are numerous messages that pass by very swiftly on the screen during boot up, unfortunately, not all of them are recoverable. The reason is that they are fed by the kernel into a "ring buffer" which is a buffer of a limited size determined during the compilation of the kernel. When the messages fill the buffer, the older ones have to be moved out to make room for the next lot of messages, hence the "ring" aspect where there's a cyclical process of messages coming in replacing those removed to make room for them in the limited space. Unfortunately, those that are removed from the buffer are not normally recoverable in a log file. I guess one could film them. The following command does show the kernel messages currently in the buffer:
Code:
dmesg
It doesn't however, show the removed ones. Nevertheless there are logs from other software like systemd's journalctl, rsyslog, syslog-ng, ulogd2 and others which report kernel and system processes beyond the early boot up. If there is no issue with booting, then there is little for the user to be concerned about, other than perhaps interest. It's a different matter if the kernel panics and booting fails.

On the matter of the GPE and ACPI warnings, again, if there is no discernible effect on the functioning of the machine, then these messages can be ignored. Unfortunately, many manufacturers of UEFI/BIOS firmware and software do not implement all the ACPI specifications, usually as an economy measure, whereas the kernel expects the standards to be observed and where it finds they are not, it notes it on screen. Usually though, the kernel is well enough equipped to deal with the issue and functionality is not much affected if at all.

If there are in fact ACPI or GPE problems there are ways of dealing with them, but often it's unnecessary.
 
Thanks very much for the recent helpful replies.

I don't want post up all my journal and log info here because there's some private personal information in there that I noticed.
However,...

I took a peek at the /var/log/boot.log and the journalctl -b and I must say that both techniques are helpful.
I already have been looking at dmesg > log.txt ; featherpad log.txt, but your techniques are more helpful.

The downside is that I don't really know how to respond to some of the stuff. for example, my wifi NIC seems to be spamming with interrupts, but those didn't show up on most of the logging. The main errors I get during boot seem to be ACPI stuff.
The journalctl -b reveals other stuff, including some dbus and xfce errors I have never seen before, and even a message about a tainted kernel(!).

Should I be worried about a tainted kernel? And what do I do about it if it is "tainted"?
 
Should I be worried about a tainted kernel? And what do I do about it if it is "tainted"?

Usually not, most of the time when you see this... it's your Linux distro vendor that "tainted" it.
All that really means, is it has some source code that isn't part of the main kernel.org kernel. The other thing
that sometimes causes this, is when you mix signed packages with un-signed packages. This happens frequently
with nvidia's drivers for some distros. Not all packages do this, but the ones that contain kernel modules (like nvidia)
sometimes do.
 
Okay thanks. That makes sense. I've been trying to read up on this topic too. It's a strange way to describe a kernel status, it's so dramatic LOL. "TAINTED" wooooo (scary music). LOL

But anyhow, I think I was able to reduce some (but not all of the messages) by reducing the number of kernel parameters I was using, and by removing some of the proprietary drivers (I already had free drivers installed).

As it turns out, some of the messages are just about devices that I blacklisted (and I don't mind that) in the kernel parameters.

But the very first "blue" messages I can't read except for the "GPE type mismatch" or something like that. The internet says not to worry about that though, I guess.

There's some other stuff printed there about PNP but I can't read it, and it doesn't show up in the /var/log/boot.log, most everything looks OK. There are a few odd errors here and there, but since my system seems okay, I'm guessing I should ignore them.

My only question is, why am I all of sudden getting blue items during boot when I never had that before?
Is it from system upgrade? Is it from the newer Linux kernel? Does it mean I got hacked?
 
Okay thanks. That makes sense. I've been trying to read up on this topic too. It's a strange way to describe a kernel status, it's so dramatic LOL. "TAINTED" wooooo (scary music). LOL

But anyhow, I think I was able to reduce some (but not all of the messages) by reducing the number of kernel parameters I was using, and by removing some of the proprietary drivers (I already had free drivers installed).

As it turns out, some of the messages are just about devices that I blacklisted (and I don't mind that) in the kernel parameters.

But the very first "blue" messages I can't read except for the "GPE type mismatch" or something like that. The internet says not to worry about that though, I guess.

There's some other stuff printed there about PNP but I can't read it, and it doesn't show up in the /var/log/boot.log, most everything looks OK. There are a few odd errors here and there, but since my system seems okay, I'm guessing I should ignore them.

My only question is, why am I all of sudden getting blue items during boot when I never had that before?
Is it from system upgrade? Is it from the newer Linux kernel? Does it mean I got hacked?
The blue text on boot up is largely informational messages or lower-priority kernel messages. Some kernels have it and some don't since it depends on how the kernel and bootloader were configured. The white text is largely about initializing hardware, loading drivers, and setting up system facilities that are needed. A blue screen on the other hand, can be a kernel panic, but that's quite obvious since the machine stops functioning.

If a system boots up normally and functions as expected, then there's usually no problem.

The likelihood of a home user's linux computer being hacked after it has been installed with a default installation of a linux distro that hasn't been altered, ranges from small to negligible to extremely remote. Hackers generally like larger targets from which there can be a return to them of some sort. One can always check uncertainties by engaging with more experienced users, forums, distro documentation, kernel docs etc. Questioning helps.

Some hacking of linux is possible of course. There are rootkit attacks, hardware and software vulnerabilities that can be taken advantage of. Mostly these vulnerabilites are identified early in the FOSS biosphere and are either resolved, or plugged or neutralised in some way. Some have existed for many years but remained so obscure that there is no evidence they were taken advantage of, e.g.: https://www.securin.io/articles/a-1...xposes-linux-to-privilege-escalation-attacks/, and here: https://www.bleepingcomputer.com/ne...-decade-old-needrestart-flaw-that-gives-root/

It's quite common for such vulnerabilities to be discovered by security specialists and investigators or firms in the linux world who then construct proof of concept code to show how the exploit can succeed in attacking a system without there existing any evidence of use of that exploit in the wider community. Then the developers usually code away the holes and life goes on.

It's the case that some users are uncertain in their usage of a linux installation's functioning and see things that are unfamiliar, odd looking or seemingly coming out of nowhere and so are unable to attribute the occurrences to any process which consequently raises their suspicions. Experience gathering knowledge over time helps with discriminating between the troubling and benign.

In relation to some major vulnerabilities in linux, one can check whether or not the kernel they are using has had mitigations applied to it against some of the relatively recent identified threats. One can check the "Vulnerablities:" section of the output of the lscpu command thus:
Code:
$ lscpu
<snip>
Vulnerabilities:     
  Gather data sampling:   Not affected
  Itlb multihit:          Not affected
  L1tf:                   Not affected
  Mds:                    Not affected
  Meltdown:               Not affected
  Mmio stale data:        Not affected
  Reg file data sampling: Mitigation; Clear Register File
  Retbleed:               Not affected
  Spec rstack overflow:   Not affected
  Spec store bypass:      Mitigation; Speculative Store Bypass disabled vi
                          a prctl
  Spectre v1:             Mitigation; usercopy/swapgs barriers and __user
                          pointer sanitization
  Spectre v2:             Mitigation; Enhanced / Automatic IBRS; IBPB cond
                          itional; RSB filling; PBRSB-eIBRS SW sequence; B
                          HI BHI_DIS_S
  Srbds:                  Not affected
  Tsx async abort:        Not affected
This kernel is well covered for the identified vulnerabilities. The actual exploits can be found in docs online for the interested reader.
 
Last edited:
The likelihood of a home user's linux computer being hacked after it has been installed with a default installation of a linux distro that hasn't been altered, ranges from small to negligible to extremely remote. Hackers generally like larger targets from which there can be a return to them of some sort. One can always check uncertainties by engaging with more experienced users, forums, distro documentation, kernel docs etc. Questioning helps.

Unless there's a broad-spectrum attack (of which I know of nothing current - and I pay attention to this stuff), you're so unlikely to be an individual target as to exclude it from the realm of possibilities. It's just not going to happen, as @osprey says.

If you're the target of a spear-phishing attack, chances are you won't know until it's much too late.

Fortunately, you're not that special. (Nor am I.)

I could be financially compromised but I have real-world defenses against those and just your regular bank/credit card companies are likely safe enough until you go giving away information. Those types of attacks don't care what OS you're using.
 
The likelihood of a home user's linux computer being hacked after it has been installed with a default installation of a linux distro that hasn't been altered, ranges from small to negligible to extremely remote. Hackers generally like larger targets from which there can be a return to them of some sort. One can always check uncertainties by engaging with more experienced users, forums, distro documentation, kernel docs etc. Questioning helps.
Your ISP (or agencies to whom ISP must comply by the law) is the primary attacker on a fresh installed system.
Ofc. as long they have the interest in you.

Ordinary hackers are unlikely to find you as soon as you install system.
 
Kernel is tainted by anything out of default tree: usually sound drivers, network card, video. If you don't want kernel to be tainted just install GNU libre kernel where all blobs are removed and only true open source drivers are allowed, but you are risking that your devices may not work.

Also it is easy to change boot messages color: mine default is green and errors are red.

I removed everything from (custom) kernel that is not related to my laptop hardware and options that I don't use. Same goes with all the software. Got rid of stuff not used.
System is superstable. If you do keep it, nothing will happen just sits there so it is up to you what do you do.


Patching kernel and hardening kernel against attack are two different things.
To see how well your kernel will defend itself try this tool:
Default kernels will fail miserably this test but not to worry. For average user default is good enough.

Finally nobody is going to be interested in small fish, unless you are working on secret secret :)
Home users just are victims of pishing or such (un)willingly giving up some information.
So it is nearly impossible to get hacked after fresh OS install.
 
Last edited:
It's a strange way to describe a kernel status, it's so dramatic LOL. "TAINTED" wooooo (scary music). LOL
One way how to avoid tainted kernel warning is to sign all your drivers, the kernel and boot loader with MOK (Machine Owner Key) and use it to setup custom secure boot without MS BS keys.
 
hey guys, thanks for all the replies and explanations.
I guess I'll not worry any more about this stuff.

It's just the first time in my life that I ever saw those blue messages.
Everything else flows by as [OK] in white and green on black and reads typically as "success" messages. Also, there's no boot delay. My whole system boots up faster than I type my password.

Maybe it's different because I upgraded my kernel with MSM and also the system was updated via pacman. I just don't know. I won't worry about the "taint" though.

thanks again.
 
If anybody is interested, I've been finding out that my wifi/bluetooth chip has a controversial (error prone) reputation within Manjaro Linux. And a lot of the blue error messages seem to be kernel messages that I didn't get before, some are just unimportant warnings but others were related to my wifi chip and/or it's power management were spamming.

As it turns out, there was a driver available from AUR which required some linux-headers (which hadn't been installed for some odd reason). So long story short, I installed the linux-headers and the matching kernel (which was a downgrade to the LTS kernel). And I'm trying out the LTS kernel (v612) instead of what I had been using (v614).

The goal is to at least prevent some unhandled interrupts from spamming (from Realek wifi).
Also, it seems like I was getting some kind of firewall activity(!!)

Maybe everything is okay enough.
 
The AUR driver didn't hurt anything but it didn't help anything.
However, I was able to reduce some of the spurious interrupts coming from the wifi chip by setting log levels at the kernel parameter in GRUB.

This reduced some non-error messages too.
This is what I added to my GRUB cmdline:

Code:
loglevel=3 systemd.show_status=auto rd.udev.log_level=3

I learned about it from the Arch Wiki, in a section on how to "silence" the boot process.

UPDATE: I think everything is working mostly okay now. I vaguely remember when I first removed the loglevel=3 thingy and then after a few linux system upgrades (and whatever else) is when I first started getting the turquoise messages flooding the screen. I still like to see the boot and shutdown process so that's why I don't use the "quiet splash" messages.
 
Last edited:


Follow Linux.org

Members online


Top