Do you folks update video drivers when not having a problem?

[BarkingGhostar]

This is the first time I noticed a chipset manufacture-issued (?) driver release. While I do have some problems with the 1060 card in Windows 7 on a 4K Acer monitor, I have not had any problems in Linux. This showed up in the Updates Manager yesterday but I haven't installed it.

 

[whm1974]

I used to back when we had to manually install closed source drivers. But not anymore.

 

[you2]

I don't. If it aint broke don't break it.

 

[whm1974]

I don't. If it aint broke don't break it.

Unless you are playing the latest games that have issues with older close source drivers, I wouldn't even bother unless you get performance improvements.

 

[Red Squirrel]

I typically don't, I find video drivers can make or break the system, if it works I leave it alone.

 

[mv2devnull]

NVidia provides drivers as binary blob "sources". https://www.nvidia.com/object/unix.html
They update to add features and to fix bugs.
You need to "install" the driver for each kernel version (manually or with DKMS).

Maintainers of ELRepo yum repository (for rpm-based "Enterprise Linux" distros) "compile" those blobs as RPM-packages.
That version has relaxed relation to kernel; same kernel module works with multiple kernel versions.
At the moment ELRepo has version 418.43 and legacy versions 390.116, 340.107.
The package update is (almost) trivial and therefore I let it happen (most of the time).

I don't know how Ubuntu repos function.

I do know that Geforce 1060 is supported by the 418. https://www.nvidia.com/Download/driverResults.aspx/142958/en-us
Therefore, getting an offer for 384 that is older than most recent legacy series (390) is somewhat odd.

 

[ultimatebob]

I think that there is a security hole in the older NVidia drivers. You might want to install that one.

 

[Red Squirrel]

What would be the attack surface on a video driver? Does it listen on any ports? That's stupid if it does.

 

[you2]

If someone can run an application on the host machine they can use security flaws to gain root access. This includes getting the user of the machine to run an infected blob. With root access they can pretty much do anything like enable remote access.

What would be the attack surface on a video driver? Does it listen on any ports? That's stupid if it does.

 

[Red Squirrel]

Oh so they already need to have remote access then? Does not seem THAT bad, then again with how insecure browsers are these days it would be no big deal for a malicious website to run said blob to then enable something worse.

 

[you2]

Or an email attachment.

Oh so they already need to have remote access then? Does not seem THAT bad, then again with how insecure browsers are these days it would be no big deal for a malicious website to run said blob to then enable something worse.

 

[Red Squirrel]

Or an email attachment.

Don't you have to actually execute those manually though? Or is there tricks to make it open automatically as soon as you open the email?

 

[you2]

I never execute them - and my mail client (thunderbird) require manual execution. However one thing that occur to me last night (and I don't know the answer to this question) is if you could have the browser render an image that would trigger the bugs. I haven't looked into the explicit problem and naively I would expect it to be more complex than simple rendering but still it begs the question if an application had the right set of primitives to induce the vulnerability and you simply need to require the drawing information to take advantage of it then the issue would be pretty sever. This sort of thing was the case with previous zip vunerbilities as well as most gif viewers - where the data triggered the bug in the common software for accessing these files to gain access to the machine.

Don't you have to actually execute those manually though? Or is there tricks to make it open automatically as soon as you open the email?

 

[Amol S.]

This is the first time I noticed a chipset manufacture-issued (?) driver release. While I do have some problems with the 1060 card in Windows 7 on a 4K Acer monitor, I have not had any problems in Linux. This showed up in the Updates Manager yesterday but I haven't installed it.
View attachment 4298

Don't do it. There have been complaints in thd past that NVIDIA drivers for Ubuntu on Virtual Machines have caused issues with the Ubuntu OS on the VM.

 

[mv2devnull]

Don't do it. There have been complaints in thd past that NVIDIA drivers for Ubuntu on Virtual Machines have caused issues with the Ubuntu OS on the VM.

Where does OP state that the Ubuntu is in VM?

 

[Amol S.]

Where does OP state that the Ubuntu is in VM?

I am trying to be a step ahead. The OP also did not state that the Ubuntu is not on VM. We don't know, thus we must assume all possibilities.

 

[corkyg]

Short answer . . . no!