Re: Spinlocks
From: Pete Zaitcev (OTID3) (oted3354ac022639a1_at_comcast.net)
Date: 10/19/05
- Next message: Pete Zaitcev (OTID3): "Re: Spinlocks"
- Previous message: Pete Zaitcev (OTID3): "Re: Spinlocks"
- In reply to: Kasper Dupont: "Re: Spinlocks"
- Next in thread: Kasper Dupont: "Re: Spinlocks"
- Reply: Kasper Dupont: "Re: Spinlocks"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Tue, 18 Oct 2005 15:35:06 -0700
On Tue, 18 Oct 2005 13:08:53 +0200, Kasper Dupont wrote:
> "Peter T. Breuer" wrote:
>>
>> So long as he disables interrupts on ALL CPUs at once, there should be
>> no problem, since the resource (the interrupt generator) is now guarded.
>
> Disabling interrupts across all CPUs is a very expensive
> operation. And I'm not even sure it would give you the
> protection you need.
It would. That's how cli() worked in Alan's SMP. But yes, it's hideously
expensive, because a cross-call is involved.
-- Pete
- Next message: Pete Zaitcev (OTID3): "Re: Spinlocks"
- Previous message: Pete Zaitcev (OTID3): "Re: Spinlocks"
- In reply to: Kasper Dupont: "Re: Spinlocks"
- Next in thread: Kasper Dupont: "Re: Spinlocks"
- Reply: Kasper Dupont: "Re: Spinlocks"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]