Re: possible spinlock optimizations

Ingo Molnar (mingo@chiara.csoma.elte.hu)
Tue, 28 Sep 1999 22:41:38 +0200 (CEST)


On Tue, 28 Sep 1999, Andrea Arcangeli wrote:

> If a spinlock got acquired two times by the same cpu then you'll sti() and
> the NMI won't trigger the oops with Manfred's idea applyed.

well, if you read Chuck's suggestion:

[...]
> besides, making the IRQ-masked spinlocks interruptible might mean that
> we're more likely to interrupt a deadlock via SysRq, right?

(unless i'm grossly misunderstanding Chuck) he is (correctly) suggesting
that the sti-change would make the kernel more debuggable, because 'hard
lockup' spinlock deadlocks would become at least Sysrq-debuggable. [Chuck
please correct me if i'm misinterpreting you]

what i said was that in all these cases the new NMI watchdog already
'debugs' the lockup, so the (quite valid) point is not relevant anymore.
ie.: the sti-change does not make the kernel more debuggable.

-- mingo

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