ZFS generating "kernel BUG at mm/usercopy.c:102!" #15918
Labels
Component: Memory Management
kernel memory management
Type: Defect
Incorrect behavior (e.g. crash, hang)
Comments
|
Will check and update, thanks! |
|
Same here: I'm running: |
|
Interesting facts:
|
|
similar issue here: zfs version: kernel: |
|
I can't reproduce it (yet), but I've been reading code. On the kernel side, the two errors come from this part of if (is_vmalloc_addr(ptr) && !pagefault_disabled()) {
struct vmap_area *area = find_vmap_area(addr);
if (!area)
usercopy_abort("vmalloc", "no area", to_user, 0, n);
if (n > area->va_end - addr) {
offset = addr - area->va_start;
usercopy_abort("vmalloc", NULL, to_user, offset, n);
}
return;
}That is, the pointer is partially or entirely outside a valid virtual allocation. So it's effectively a use-after-free, or perhaps, a failed allocation. Given proximity to OOM makes me wonder if another thread freed something in response to a reclaim request, and by the time we get here its gone. If nothing else, we could use similar checking logic as above ourselves, and if we hit it, log something and return |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
System information
Ubuntu 22.04 running 6.2.0-1018-aws #18~22.04.1-Ubuntu.
zfs-2.1.5-1ubuntu6~22.04.2
zfs-kmod-2.1.9-2ubuntu1.1
on AWS u-12tb1.112xlarge x86_64 instance.
Describe the problem you're observing
We are getting these message in dmesg:
The text was updated successfully, but these errors were encountered: