[PATCH v2] panic, kexec: Make __crash_kexec() NMI safe

[Valentin Schneider]

On 23/06/22 11:31, Sebastian Andrzej Siewior wrote:
> On 2022-06-20 12:15:20 [+0100], Valentin Schneider wrote:
>> Attempting to get a crash dump out of a debug PREEMPT_RT kernel via an NMI
>> panic() doesn't work. The cause of that lies in the PREEMPT_RT definition
>> of mutex_trylock():
>> 
> …
>
>> Fixes: 6ce47fd961fa ("rtmutex: Warn if trylock is called from hard/softirq context")
>> Signed-off-by: Valentin Schneider <vschneid at redhat.com>
>
> Reviewed-by: Sebastian Andrzej Siewior <bigeasy at linutronix.de>
>
>> ---
>> v1 -> v2
>> ++++++++
>> 
>> o Changed from Peterson-like synchronization to simpler atomic_cmpxchg
>>   (Petr)
>> o Slightly reworded changelog
>> o Added Fixes: tag. Technically should be up to since kexec can happen
>>   in an NMI, but that isn't such a clear target
>
> RT-wise it would be needed for each release.

So git tells me the Fixes: commit dates from v4.2; from [1] and [2] I get
that both current longterm stable and stable-rt trees go as far back as
v4.9, so I'm guessing if that gets picked up for the stable trees then it
should make its way into the stable -rt trees?

[1]: https://wiki.linuxfoundation.org/realtime/preempt_rt_versions
[2]: https://www.kernel.org/category/releases.html

> There is also a mutex_unlock() in case an image is missing. This can go
> via the scheduler if there is a waiter which does not look good with the
> NMI in the picture.
>
> Sebastian