Re: RFC: Ideal Adaptive Spinning Conditions

[john cooper]

Peter W. Morreale wrote:

> Right.  This was *critical* for the adaptive rtmutex.   Note in the RT
> patch, everybody spins as long as the current owner is on CPU.   
> 
> FWIW, IIRC, Solaris has a heuristic approach where incoming tasks spin
> for a period of time before going to sleep.  (Cray UINCOS did the same) 

IIRC Solaris mutexes are declared either simple spin or adaptive,
an acquisition attempt of the latter only checking the hold
status of the mutex and if held the owner's run status before
making the spin vs. block decision.

I don't believe mixing of the two mutex types within a given path
was permissible as a previously acquired simple spin mutex could
remain held when a subsequent adaptive mutex decided to block.
Although presumably an elevated IPL could have been sufficient to
flag that scenario.

-john
  
-- 
john.cooper@xxxxxxxxxxxxxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/