Re: [mm 4.15-rc7] Random oopses under memory pressure.

[Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>]

Tetsuo Handa wrote:
> Michal Hocko wrote:
> > All of those seem to be file pages. So maybe try to use a different FS.
> 
> Maybe that's the next thing I should try.

xfs versus ext4 => Both triggers the bug

Linus Torvalds wrote:
> On Thu, Jan 11, 2018 at 5:31 PM, Tetsuo Handa
> <penguin-kernel@i-love.sakura.ne.jp> wrote:
> >
> > Thus, I suspect that somewhere is confusing HighMem pages and !HighMem pages.
> 
> Hmm. I can't even imagine how you'd do that.
> 
> Sure, if you take page_address() to get a kmap'ed linear address, and
> then feed that linear address back to virt_to_page(), you'd certainly
> get a crazy page. But that would be insane.. I don't see how you'd do
> that.
> 
> Hmm. Do you have CONFIG_DEBUG_VIRTUAL enabled? That should catch at
> least the above case, it should enable a debugging version of
> __virt_to_phys() and use it.
> 

CONFIG_DEBUG_VIRTUAL=y versus CONFIG_DEBUG_VIRTUAL=n => No difference

No mem= parameter versus mem=768M => Cannot trigger if mem=768M (i.e. not using HighMem)

CONFIG_SLUB=y versus CONFIG_SLAB=y => Both triggers the bug

CONFIG_DEBUG_PAGEALLOC=y versus CONFIG_DEBUG_PAGEALLOC=n => Cannot trigger if CONFIG_DEBUG_PAGEALLOC=n

I don't know whether there is a bug in CONFIG_DEBUG_PAGEALLOC=y code.
Config is at http://I-love.SAKURA.ne.jp/tmp/config-4.15-rc7-min .

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>