PaX

joined 2 years ago
[–] PaX@hexbear.net 32 points 6 months ago (2 children)
[–] PaX@hexbear.net 1 points 6 months ago* (last edited 6 months ago)

You can check to see what drivers were compiled as modules or into your kernel by reading the kernel configuration at /proc/config.gz or /boot/*config*

There might also be out-of-tree (not included with the kernel) drivers installed as packages on your system but this is very rare outside of like... having an NVIDIA card and running the closed-source vendor driver

[–] PaX@hexbear.net 7 points 6 months ago* (last edited 6 months ago)

The vast majority of drivers are included with the Linux kernel now (in tree) so the difference usually comes down to kernel version (newer kernels have more drivers, of course) or kernel configuration set at compile-time (this can be anything from including or not including drivers, to turning driver features on and off, or more fundamental changes beyond drivers)

You can get kernel version info from uname -a and a lot of the time, probably most of the time (this is also down to configuration), you can get kernel configuration info from /proc/config.gz (use gzip -d to decompress) or something like /boot/config

Then you can run diff on configurations of 2 different distro kernels you're interested in to see how the 2 distribution's kernels were set up differently

This could also be caused by different setups of userspace tools or UI that interact with these drivers in different, sometimes worse ways but this is usually much less likely in my experience (most Linux distros do things like this the same way these days tbh)

Oh, also, there are a lot of drivers that require vendor-supplied firmware or binary blobs to function and most of the time distros don't bake these into the kernel (although it is possible) and different distros might have more or less of these blobs available or installed by default or they might be packaged differently. The kernel should print an error message if it can't find blobs it needs though

I guess there's kinda a lot to consider lol. Sorry if all of this is obvious

What hardware are you talking about specifically?

[–] PaX@hexbear.net 3 points 6 months ago* (last edited 6 months ago)

Ohh that's true, I didn't think about that. It would be difficult to route anything through it unless you were connected directly to it with nothing in-between because no other router would forward packets destined for somewhere else to my machine (except maybe in the extremely unlikely case of source routing?). It seems obvious now lol, thank you!

I'll write some firewall rules just in case

 

can someone else use my machine as a router to forward traffic to anywhere else on the internet?

I'm not entirely sure what the security implications of that would even be if true but probably nothing good

I don't have any other routes in my routing table other than my "default" route and this machine is reachable via a globally routable IPv4 address. Also I think there are probably other machines on the same subnet (cloud VPS)

[–] PaX@hexbear.net 2 points 7 months ago* (last edited 7 months ago) (1 children)

I see. Our motherboards have different chipsets (I have an X570 in mine). It probably has nothing to do with my issue...

Hoping those kernel parameters fix it. I wish I could help further. PCs are just a bottomless, mostly undocumented rabbithole :(

[–] PaX@hexbear.net 0 points 7 months ago (3 children)

What motherboard do you have? Also what happens exactly when the lock-ups happen? Have you ever been playing audio when the lock-ups happen and does it loop or stop or keep playing?

I recently had to "fix" (workaround) a similar issue in the OpenBSD kernel with a specific hardware peripheral on my PC (running a 2nd-gen Ryzen), the High Definition Audio controller. For whatever reason (and only when I was running OpenBSD) interrupts from the HDA controller (to let the CPU know to refill audio buffers) would just randomly stop making it to the CPU and audio would loop for a few seconds and then shut off. I spent a long time trying to figure out what causes it and reading Linux driver code but I couldn't find a cause or why only OpenBSD would trigger it. I ended up having to write kind of a hacky polling mode into the HDA driver. My only guess is some of these AMD-chipset-having motherboards have faulty interrupt controllers.

Maybe there is a similar issue with your system and timer interrupts aren't making it to your CPU or something. But I'm not really an expert on PC architecture and idek if it even works like that on PCs lol

Sorry for so many questions but do you also have any kernel logs available from when this happens?

[–] PaX@hexbear.net 3 points 7 months ago

I simply do both

It's not working agony-deep

[–] PaX@hexbear.net 4 points 7 months ago

There are some purpose-built ARM Linux laptops available but as an owner of an unused Pinebook Pro.... can't recommend agony-deep

Walking the path of a PC hater is not easy

[–] PaX@hexbear.net 2 points 7 months ago* (last edited 7 months ago)

Rip out the fan and connect the processor heatsink to a heatpipe

Then carry around a cup of water to dip the heatpipe into

This is not a bit, I am a real hardware designer

[–] PaX@hexbear.net 3 points 8 months ago* (last edited 8 months ago)

Xorg? Wayland? You have bespoke protocols just for windowed graphics? I'm happy with my /dev/draw and /dev/wsys/* grillman

Unix is a zombie OS that should probably die

7
deleted by creator (hexbear.net)
submitted 1 year ago* (last edited 1 year ago) by PaX@hexbear.net to c/piracy@lemmy.dbzer0.com
view more: next ›