Re: [PATCH] stackdepot: Use a raw spinlock in stack depot

[Dmitry Vyukov]

On Tue, May 4, 2021 at 5:34 PM Andi Kleen <ak@xxxxxxxxxxxxxxx> wrote:
> > So why is this a false positive that we just need to silence?
> > I see LOCKDEP is saying we are doing something wrong, and your
> > description just describes how we are doing something wrong :)
> > If this is a special false positive case, it would be good to have a
> > comment on DEFINE_RAW_SPINLOCK explaining why we are using it.
> >
> > I wonder why we never saw this on syzbot. Is it an RT kernel or some
> > other special config?
>
> This happened in a special configuration that triggered ACPI errors at
> boot time.
>
> It's probably not something that is normally executed, as well as syzbot is
>
> probably not exercising bootup anyways.
>
> > A similar issue was discussed recently for RT kernel:
> > https://groups.google.com/g/kasan-dev/c/MyHh8ov-ciU/m/nahiuqFLAQAJ
> > And I think it may be fixable in the same way -- make stackdepot not
> > allocate in contexts where it's not OK to allocate.
>
>
> Yes that's a good idea. I've seen also other errors about the allocator
> triggered
>
> by stack depot being in the wrong context. Probably doing that would be
> the right
>
> fix. But I actually tried to switch depot to GFP_ATOMIC allocations
> (from GFP_NOWAIT),
>
> but it didn't help, so I'm not fully sure what needs to be changed.

We may not allocate at all, see may_prealloc idea here:
https://groups.google.com/g/kasan-dev/c/MyHh8ov-ciU/m/k1LXBmonAQAJ