Re: [PATCH V2 0/6][RFC] futex: FUTEX_LOCK with optional adaptive spinning

[Avi Kivity <avi@redhat.com>]

On 04/06/2010 12:54 AM, Darren Hart wrote:
> Avi Kivity wrote:
>> On 04/05/2010 11:23 PM, Darren Hart wrote:
>>> In-Reply-To:
>>>
>>> NOT FOR INCLUSION
>>>
>>> The following patch series implements a new experimental kernel side 
>>> futex mutex
>>> via new FUTEX_LOCK and FUTEX_LOCK_ADAPTIVE futex op codes. The 
>>> adaptive spin
>>> follows the kernel mutex model of allowing one spinner until the 
>>> lock is
>>> released or the owner is descheduled. The patch currently allows the 
>>> user to
>>> specify if they want no spinning, a single adaptive spinner, or 
>>> multiple
>>> spinners (aggressive adaptive spinning, or aas... which I have 
>>> mistyped as "ass"
>>> enough times to realize a better term is indeed required :-).
>>
>> An interesting (but perhaps difficult to achieve) optimization would 
>> be to spin in userspace.
>
> I couldn't think of a lightweight way to determine when the owner has 
> been scheduled out in userspace. Kernel assistance is required. You 
> could do this on the schedule() side of things, but I figured I'd get 
> some strong pushback if I tried to add a hook into descheduling that 
> flipped a bit in the futex value stating the owner was about to 
> deschedule(). Still, that might be something to explore.

In the futex value it's hopeless (since a thread can hold many locks), 
but I don't think it's unreasonable to set a bit in the thread local 
storage area.  The futex format would then need to be extended to 
contain a pointer to this bit.


>
>>
>> How many cores (or hardware threads) does this machine have? 
>
> Sorry, I meant to include that. I tested on an 8 CPU (no hardware 
> threads) 2.6 GHz Opteron 2218 (2 QuadCore CPUs) system.
>
> > At 10%
>> duty cycle you have 25 waiters behind the lock on average.  I don't 
>> think this is realistic, and it means that spinning is invoked only 
>> rarely.
>
> Perhaps some instrumentation is in order, it seems to get invoked 
> enough to achieve some 20% increase in lock/unlock iterations. Perhaps 
> another metric would be of more value - such as average wait time?

Why measure an unrealistic workload?

>
>> I'd be interested in seeing runs where the average number of waiters 
>> is 0.2, 0.5, 1, and 2, corresponding to moderate-to-bad contention.
>> 25 average waiters on compute bound code means the application needs 
>> to be rewritten, no amount of mutex tweaking will help it.
>
> Perhaps something NR_CPUS threads would be of more interest? 

That seems artificial.

> At 10% that's about .8 and at 25% the 2 of your upper limit. I could 
> add a few more duty-cycle points and make 25% the max. I'll kick that 
> off and post the results... probably tomorrow, 10M iterations takes a 
> while, but makes the results relatively stable.

Thanks.  But why not vary the number of threads as well?

>
>> Does the wakeup code select the spinning waiter, or just a random 
>> waiter?
>
> The wakeup code selects the highest priority task in fifo order to 
> wake-up - however, under contention it is most likely going to go back 
> to sleep as another waiter will steal the lock out from under it. This 
> locking strategy is unashamedly about as "unfair" as it gets.

Best to avoid the wakeup if we notice the lock was stolen.


-- 
Do not meddle in the internals of kernels, for they are subtle and quick to panic.