Andres Freund <andres@anarazel.de> writes:
> On 2017-05-11 16:27:48 -0400, Peter Eisentraut wrote:
>> (So without contention fast-path locking beats the extra dance that
>> open_share_lock() does.)
> That's kind of surprising, I really wouldn't have thought it'd be faster
> without. I guess it's the overhead of sigsetjmp(). Cool.
My results (posted nearby) lead me to suspect that the improvement
Peter sees from 9.1 to 9.2 has little to do with fastpath locking
and a lot to do with some improvement or other in subtransaction
lock management.
regards, tom lane
