[PATCH] Fix kexec abort due to IPI from panic().

[Seiji Aguchi]

Hi Eric,

Thank you for your quick response!

>Disabling interrupts is fine,

Thanks!

>However that call to kmsg_dump(KMSG_DUMP_KEXEC) is a bug as it introduces locks 
>into a path that should not be taking locks.

I'd like to find a way that kexec coexists with kmsg_dump(KMSG_DUMP_KEXEC) because 
kmsg_dump is a useful troubleshooting feature as well.

So, I will improve kmsg_dump(KMSG_DUMP_KEXEC) if there are some bugs.
Could you please let me know your concern?
It is helpful for me if you have an example scenario kexec fails.

>Nothing in the crash_kexec path should even have the option of blocking.

Do you mean I need to change kmsg_dump(KMSG_DUMP_KEXEC) to lockless?

Seiji