Random Segmentation Faults on AMD CPUs

[SniverDaBest]

Hello!
I was messing around with cavOS for fun, and found that a lot of programs will randomly seg-fault when trying to run them.
I haven't tested this outside of Qemu.

Running ls a bunch results in this:

In the serial console, it shows GP-faults every time it seg-faults in the terminal.

======================================= REGDUMP =======================================
+ RAX=0000002108421084 RBX=00001000000a4920 RCX=0000700000049bd0 RDX=0000000000000016 +
+ RSI=0000700000000000 RDI=000070000004a698 RBP=00001000000a58a0 RSP=ffff80000cc3cfd8 +
+ R8 =00001000000a58a0 R9 =0000000000041066 R10=0000000000000000 R11=0000000000000000 +
+ R12=000070000041e4e0 R13=0000600000000010 R14=0000000000000000 R15=0000000000000200 +
+ RIP=ffffffff8008d4b4 RFL=0000000000010086 _CS=0000000000000028 _SS=0000000000000030 +
+ ERR=0000000000000048 INT=000000000000000d _DS=0000000000000030 RES=0000000000000000 +
=======================================================================================
[regdump] regs{ffff80000cc3cf18} timerTicks{278542} task{ffff800006002600} taskId{162}
[isr::task] [u] Killing task{162} because of General Protection Fault!

And sometimes, it just quits the program before showing a seg-fault.
Or sometimes it runs the program, and then seg-faults in the middle of it. I got this to happen with ls and fastfetch.

[malwarepad]

I am 99% sure your host has an AMD CPU and you are running cavOS on KVM accelerated qemu. I've seen this happen a lot on other people's machines unfortunately. I'm guessing that there is some small AMD-specific issue somewhere in the syscall/irq/isr code that I cannot debug and fix since I only own Intel machines.

[SniverDaBest]

When I get the chance to, I will look into this.

[SniverDaBest]

Disabling KVM does indeed fix things.
I'm looking through the code to see if I can find anything that could hint at the root cause of the faulting.

[9xbt]

Disabling KVM does indeed fix things. I'm looking through the code to see if I can find anything that could hint at the root cause of the faulting.

Put that RIP through addr2line and see where it's happening.

[SniverDaBest]

I personally can't find anything wrong, but I do know where it's occurring.
Using objdump, I found that it was GPF'ing at src/kernel/cpu/isr.asm, line 35. (or the iretq in asm_finalize)
I don't know where the issue is... I never got syscall working when I was doing my x86 OS.
I'll try running with KVM on a couple other AMD systems, since it could just be an issue with my CPU, even if it's very unlikely.

[SniverDaBest]

It's not just my model of CPU... Tested on an older Zen4 CPU in my laptop and it gets the same issue.