Re: [PATCH] Fix x86_64 _spin_lock_irqsave()

[Martin Bligh <mbligh@mbligh.org>]

Andrew Morton wrote:
> On Thu, 24 Aug 2006 13:10:09 +1000
> Nick Piggin <nickpiggin@yahoo.com.au> wrote:
> 
> 
>>Edward Falk wrote:
>>
>>>Add spin_lock_string_flags and _raw_spin_lock_flags() to 
>>>asm-x86_64/spinlock.h so that _spin_lock_irqsave() has the same 
>>>semantics on x86_64 as it does on i386 and does *not* have interrupts 
>>>disabled while it is waiting for the lock.
>>>
>>>This fix is courtesy of Michael Davidson
>>
>>So, what's the bug? You shouldn't rely on these semantics anyway
>>because you should never expect to wait for a spinlock for so long
>>(and it may be the case that irqs can't be enabled anyway).
>>
>>BTW. you should be cc'ing Andi Kleen (x86+/-64 maintainer) on
>>this type of stuff.
>>
>>No comments on the merits of adding this feature. I suppose parity
>>with i386 is a good thing, though.
>>
> 
> 
> We put this into x86 ages ago and Andi ducked the x86_64 patch at the time.
> 
> I don't recall any reports about the x86 patch (Zwane?) improving or
> worsening anything.  I guess there are some theoretical interrupt latency
> benefits.

Spinlocks are indeed meant to be held for a short time, but irq
disabling is meant to be shorter.

I think the real question is: what is the justification for disabling
interrupts when spinning for a lock? We should never disable interrupts
unless we have to.

M.