RE: [RFC][PATCH -next] make pstore/kmsg_dump run after stoppingother cpus in panic path

[Seiji Aguchi]

Hi,

>> afacit this assumes that (reason == KMSG_DUMP_PANIC) if in_nmi().  Is
>> that always the case, and will it always be the case in the future?
>

Currently, when kernel is in nmi context and kmsg_dump() are called, its reason is always "KMSG_DUMP_PANIC".

>
>Perhaps a 'BUG_ON(in_nmi() && reason != KMSG_DUMP_PANIC)'?

I don't think BUG_ON() is needed.

If someone would like to log messages in the case of "in_nmi() && reason != KMSG_DUMP_PANIC",
he/she will add a new trigger as follows.

We can discuss implementation of kmsg_dump() at that time.
But I can't see any use case now.

	371 static notrace __kprobes void 
	372 unknown_nmi_error(unsigned char reason, struct pt_regs *regs) 
	373 {
	374         if (notify_die(DIE_NMIUNKNOWN, "nmi", regs, reason, 2, SIGINT) == 
	375                         NOTIFY_STOP) 
	376                 return; 
	377 #ifdef CONFIG_MCA 
	378         /* 
	379          * Might actually be able to figure out what the guilty party 
	380          * is: 
	381          */ 
	382         if (MCA_bus) { 
	383                 mca_handle_nmi(); 
	384                 return; 
	385         } 
	386 #endif 
	387         pr_emerg("Uhhuh. NMI received for unknown reason %02x on CPU %d.\n", 
	388                  reason, smp_processor_id()); 
	389  
	390         pr_emerg("Do you have a strange power saving mode enabled?\n"); 
	391         if (unknown_nmi_panic || panic_on_unrecovered_nmi) 
	392                 panic("NMI: Not continuing"); 
	393  
	394         pr_emerg("Dazed and confused, but trying to continue\n");
		    *kmsg_dump(NMI);*
	395 }

>>
>> I felt that the spin_trylock() approach was less horrid than this.  I
>> assume that the new approach will cause lockdep to go berzerk?
>
>Heh.  Good point.  That is probably a good test case.  Though finding a
>working GHES implementation in the firmware isn't easy these days, making
>it hard to test. :-/

Thank you for supporting my patch.
But I can resend another patch below in accordance with Chen's request.
We can avoid potential deplock issue with this patch.
And this is simpler than trylock approach.

> <snip>
> If(!panic)
> 	spin_lock_irqsave();
>
>       .
>       .
> If(!panic)
> 	spin_unlock_restore();
> <snip>

Seiji
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/