Re: [PATCH] PPC64 mmu_context_init needs to run earlier

[Benjamin Herrenschmidt]

> That said, maybe the problem is that we shouldn't even get far enough into 
> the fork() logic to ever get into a new MMU context if driver_init ends up 
> being called before we're ready. 

Agreed. I chatted with Andrew about this, and I think we need
call_usermodehelper to be in a "plugged" state during boot, where it
queues up events but doesn't exec's userland. It remains to be decided
at what point during boot (during initcalls ? after initcalls) we can
"unplug" it tho...

I think it's definitely bogus to try to run userland in the middle of
arch_initcall's.... 

We need this plug/unplug logic (as I wrote separately to linux-pm) for
suspend as well, since we can't affort calling userlands once we have
started suspending devices (and frozen userland).

Ben.


-
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/