Linus Torvalds wrote: > On Thu, Jan 11, 2018 at 6:11 AM, Tetsuo Handa > <penguin-kernel@xxxxxxxxxxxxxxxxxxx> wrote: > > > > I retested with some debug printk() patch. > > Could you perhaps enable KASAN too? Unfortunately, KASAN is not available for x86_32 kernels. Thus, I'm stuck. > So presumably "page->mem_cgroup" was just a random pointer. Which > probably means that "page" itself is not actually a page pointer, sinc > e I assume there was no memory hotplug going on here? Nothing special. No memory hotplug etc. > Most (all?) of your other oopses seem to have somewhat similar > patterns: shrink_inactive_list() -> rmap_walk_file() -> oops due to > garbage. Yes. In most cases, problems are detected by that sequence.