Random System Crashes F40 KDE

Our Neighbors Asahi Linux Asahi General
, , , ,
42

Continuing this slightly off-topic memory discussion here - agreeing with the OP that the system crashes with reboot (kernel oops) regression is solved.

My system still locked up once today, again as previously mentioned with an open zoom session and many other tabs in firefox. It showed 100% zram and 20% swapfile usage before hanging/locking. I now changed the swap priorities to be equal and this seems to be much more stable. Maybe this should be done for all 8GB limited users @marcan ?

NAME               TYPE      SIZE   USED PRIO
/var/swap/swapfile file        8G 967.3M  100
/dev/zram0         partition 7.4G 805.6M  100
43

I was able to reproduce the problem (twice) just now and instead of locking the whole system, now the Firefox zoom tab crashes. That’s great because I mind much less! And it’s a reminder to stop working and actually pay attention to the zoom call :slight_smile:

44

Are you getting OOM kills for the zoom tab, or some other kernel message when it happens?

Setting the swapfile priorities to equal doesn’t make any sense, because zram is supposed to be used first before spilling to disk (otherwise it sort of defeats the whole purpose of having zram). If zram specifically is causing hangs we need to figure out what’s going on there, not just fudge the priorities to make it less likely.

45

Im a new Fedora user who installed the KDE spin 2 days ago. Yesterday It started happening for me as well. I was always using Firefox, Elisa and Stremio while it happened. I thought it was Stremio, but it happened again 10 minutes ago with stremio off, just Firefox and Elisa.

NVIDIA-SMI Info:
550.90.07 Driver Version: 550.90.07 CUDA Version: 12.4

Fastfetch Info:
OS: Fedora Linux 40 (KDE Plasma) x86_64
Host: 82JY (Legion 5 17ACH6H)
Kernel: Linux 6.9.4-200.fc40.x86_64
Uptime: 8 mins
Packages: 2568 (rpm), 28 (flatpak)
Shell: bash 5.2.26
Display (CMN175C): 1920x1080 @ 144Hz [Built-in]
DE: KDE Plasma 6.0.5
WM: KWin (Wayland)
WM Theme: WillowDarkShader
Theme: Fusion (BreezeDark) [QT], Breeze [GTK3/4]
Icons: breeze-dark [QT], breeze-dark [GTK3/4]
Font: Noto Sans (11pt) [QT], Noto Sans (11pt) [GTK3/4]
Cursor: breeze (36px)
Terminal: konsole 24.5.0
CPU: AMD Ryzen 5 5600H (12) @ 4.28 GHz
GPU: NVIDIA GeForce RTX 3060 Mobile / Max-Q [Discrete]
Memory: 4.86 GiB / 15.46 GiB (31%)
Swap: 0 B / 8.00 GiB (0%)
Disk (/): 220.31 GiB / 929.93 GiB (24%) - btrfs
Battery: 59% [AC Connected]
Locale: en_US.UTF-8

46

Well it doesn’t happen often anymore now with both zram and swapfile at the same priority. I think these journalctl messages relate to the firefox tab crashes (which previously would hang the whole system, when I had the default priorities; journalctl remains empty then until the manual reboot):

Jun 14 15:50:11 pieter-macmini firefox[25983]: Crash Annotation GraphicsCriticalError: |[0][GFX1-]: RenderCompositorSWGL failed mapping default framebuffer, no dt (t=1724.74) |[1][GFX1]: DMABuf surface import failed! (t=5897.56) [GFX1]>
Jun 14 15:50:11 pieter-macmini firefox[25983]: Crash Annotation GraphicsCriticalError: |[0][GFX1-]: RenderCompositorSWGL failed mapping default framebuffer, no dt (t=1724.74) |[1][GFX1]: DMABuf surface import failed! (t=5897.56) |[2][G>
Jun 14 15:50:11 pieter-macmini firefox[25983]: Crash Annotation GraphicsCriticalError: |[0][GFX1-]: RenderCompositorSWGL failed mapping default framebuffer, no dt (t=1724.74) |[1][GFX1]: DMABuf surface import failed! (t=5897.56) |[2][G>
Jun 14 15:50:13 pieter-macmini firefox[25983]: [Parent 25983, IPC I/O Parent] WARNING: Message needs unreceived descriptors channel:ffff18136100 message-type:11403270 header()->num_handles:1 num_fds:0 fds_i:0: file /builddir/build/BUIL>
Jun 14 15:50:13 pieter-macmini firefox[38384]: Exiting due to channel error.
47

Hi Michael, it seems you’re not on apple silicon - shown by your x86_64 architecture. The discussion here is about Asahi Linux which is for apple silicon only.

48

This post is quite old and there doesn’t seem to be a solution pointed out.

Did the problem get solved? Did it solve itself in an update?

49

It is hopefully fixed by: PSA: Transitioning from zram+swap to zswap