svn commit: r193011 - in head: . share/man/man9 sys/cddl/compat/opensolaris/sys sys/conf sys/kern sys/sys

[Attilio Rao]

2009/5/29 Robert Watson <rwatson at freebsd.org>:
> On Fri, 29 May 2009, Attilio Rao wrote:
>
>>  Reverse the logic for ADAPTIVE_SX option and enable it by default.
>>  Introduce for this operation the reverse NO_ADAPTIVE_SX option.
>>  The flag SX_ADAPTIVESPIN to be passed to sx_init_flags(9) gets suppressed
>>  and the new flag, offering the reversed logic, SX_NOADAPTIVE is added.
>>
>>  Additively implements adaptive spininning for sx held in shared mode.
>>  The spinning limit can be handled through sysctls in order to be tuned
>>  while the code doesn't reach the release, after which time they should
>>  be dropped probabilly.
>
> The princple here has always been that sleepable locks are likely to be used
> in sleepable ways, and therefore that adaptive locking (spinning) was
> unproductive.  In light of these performance results, have you considered
> making the same change to lockmgr (assuming it's not there already)?  Also,
> is adaptive locking for sx locks disabled in Xen by default in the same was
> as mutexes?

Yes, lockmgrs are the next ones to be handled in regard of this.

I'm actually prodding key people abuout experimenting adaptive
spinning and reporting to me. Let's see if they can get some
interesting benchmark in edge cases (Xen too).

Thanks,
Attilio


-- 
Peace can only be achieved by understanding - A. Einstein