Re: Performance degradation in commit ac1d794

On Mon, Mar 21, 2016 at 6:09 PM, Andres Freund <andres(at)anarazel(dot)de> wrote:
> On 2016-03-21 11:52:43 +1300, Thomas Munro wrote:
>> * I would be interested in writing a kqueue implementation of this for
>> *BSD (and MacOSX?) at some point if someone doesn't beat me to it.
>
> I hoped that somebody would do that - that'd afaics be the only major
> API missing.

Here's a first swing at it, though I need to find some spare time to
test and figure out if some details like error conditions and EOF are
handled correctly. It could in theory minimise the number of syscalls
it makes by buffering changes (additions and hopefully one day
removals) and then use a single kevent syscall to apply all
modifications to the set and begin waiting, but for now it mirrors the
epoll code. One user-visible difference compared to epoll/poll/select
is that it delivers readable and writable events separately if both
conditions are true for a single fd. It builds and appears to work
correctly on FreeBSD 10.2 and MacOSX 10.10.2. Sharing this early
version in case any BSD users have any feedback. I hope to do some
testing this weekend.

--
Thomas Munro
http://www.enterprisedb.com