Re: [PATCH] x86/apic: Fix circular locking dependency between console and hrtimer locks

[Leon Romanovsky]

On Mon, Apr 27, 2020 at 05:35:18PM +0200, Thomas Gleixner wrote:
> Leon Romanovsky <leon@xxxxxxxxxx> writes:
> > But if we are talking about other weirdnesses, I have another splat in my
> > QEMU machine, which is different, but from the same code area.
>
> It's a completely different code area, really. This triggers in
> deactivate_slab().
>
> > [    1.383968] ACPI: Added _OSI(Module Device)
> > [    1.385684] ACPI: Added _OSI(Processor Device)
> > [    1.389345] ACPI: Added _OSI(3.0 _SCP Extensions)
> > [    1.389345] ACPI: Added _OSI(Processor Aggregator Device)
> > [    1.393454] ACPI: Added _OSI(Linux-Dell-Video)
> > [    1.394920] ACPI: Added _OSI(Linux-Lenovo-NV-HDMI-Audio)
> > [    1.396481] ACPI: Added _OSI(Linux-HPI-Hybrid-Graphics)
> > [    1.793418] ACPI: 1 ACPI AML tables successfully acquired and loaded
> > [    1.845452]
> > [    1.846768] =============================
> > [    1.849293] [ BUG: Invalid wait context ]
>
> Do you have CONFIG_PROVE_RAW_LOCK_NESTING enabled? If yes, please
> disable it. The Kconfig option explains that this will trigger
> splats.

Yes, I have it enabled.

Thanks for the tip.

>
> If not, then this is mysterious but has absolutely nothing to do with
> the other thing. It's also completely unrelated to ACPI ...
>
> Thanks,
>
>         tglx