You are not logged in.
I built a desktop computer with a Ryzen 5 3600, Radeon 5600xt, and 550W PSU about a year ago and I've had complete and seemingly random system freezes ever since I got it. The freezes can be so severe to where SysReq keys don't worry, other times it plays the audio on loop and lets me use the SysReq. Additionally, I get segmentation faults in programs like electron, chromium, spotify multiple times each day (im talking like at least 1 of them for about every 3 hours of uptime, at least). Sometimes these correlate to a full system freeze, but not always. I've found that the system is most prone to freezing when It is more or less idle (load average < 1) and the freezing only occurs when the X server is running. I've left the system on TTY for many hours and encountered no problems. This has all happened while using basically the most latest arch linux kernel at the time.
Things I have tried with no success;
- Use a "typical current idle" and "disable c6 states" in UEFI as recommended elsewhere for Ryzen processors. I also tried the kernel parameters to disable the C-states. Ultimately neither stopped the freezes nor segfaults.
- Many iterations of `memtest` led to no errors
- I've left the desktop on tty2 while the X server ran on tty1 in the background, this eventually led to a crashed X server.
- The computer never freezes under heavy load, like when playing a game.
- I almost never use windows, but I've left it idle on windows and seen the computer do a complete unprompted restart. Could do some more testing here.
- Another windows afk test while leaving spotify, chrome, and discord opened led to a blue screen within about 10 minutes with error codes `driver overran stack_buffer`, and another test gave `kernel_auto_boost_invalid_lock_release`
I try to take a look at my Xorg and journalctl logs whenever this happens, but usually they are not very useful. Each log almost always has a segfault somewhere, but the segfault isn't always what actually seems to correlate to the freeze. I've linked a few interesting journalctl and Xorg logs from times when the computer froze/crashed. I can't make sense of most of them, but maybe someone else can.
Xorg crash logs;
- Xorg crash while on different tty
- Random crash
Journalctl logs;
- bug around 12:20:17
- stack trace at 09:04:42
- BUG: soft lockup around 14:42:00
I also have coredumps for most of the segfaults but they show the same info as the stack trace in journalctl, as far as I know.
At this point, I'm considering trying to return the CPU under warranty. When I first installed it, the computer would not boot properly, but when I reseated it that fixed the problem, so I've been skeptical for a while now of its reliability. With the constant segmentation faults and computer freezes, I can't think of anything else except for it being a hardware problem (still strange no crashes without X server running, maybe GPU the problem instead?). Do any of you think it's worth for me to try and RMA and return the CPU to fix this issue? The computer is basically unusable with how frequent the freezing has gotten lately.
Last edited by lowfye (2021-05-21 19:22:15)
Offline