https://bugzilla.suse.com/show_bug.cgi?id=1199355 https://bugzilla.suse.com/show_bug.cgi?id=1199355#c31 --- Comment #31 from Anthony Iliopoulos --- (In reply to Stefan Seyfried from comment #30)

Regarding stack overflow or such:

This is an 8GB raspberry pi4 and it is only crashing when running a huge VM (7.5GB) as OBS build job, building a package (kodi) which actually uses all this memory. It's always this one build job crashing the machine, relatively reliably.

This is also what the stacktrace corroborates and not related to any stack overflow I think. At the point of crash the qemu process is trying to do buffered io over uring and the page cache is trying to allocate a page which is failing due to the low memory conditions. At that point the page allocator is going into direct reclaim and trying to shrink the memory by swapping out some dirty pages (which is why it is doing submit_bio). This all looks normal. The patch I've mentioned is already applied to linux-block/for-next tree [1], but I don't know if this indeed fixes the issue as I'm not that familiar with the block layer. Better follow Hannes' suggestions there :-) [1] https://git.kernel.org/pub/scm/linux/kernel/git/axboe/linux-block.git/commit/?h=for-next&id=5d05426e2d5fd7df8afc866b78c36b37b00188b7 -- You are receiving this mail because: You are the assignee for the bug.