Nice work! Re: EXPERIMENTAL: tevent_kqueue support

[Ira Cooper]

On Wed, Mar 6, 2013 at 10:18 AM, Stefan (metze) Metzmacher
<metze at samba.org>wrote:

> Am 06.03.2013 14:16, schrieb Ira Cooper:
> > On Wed, Mar 6, 2013 at 5:34 AM, Stefan (metze) Metzmacher
> > <metze at samba.org>wrote:
> >
> >> Am 06.03.2013 11:21, schrieb Andrew Bartlett:
> >>> On Fri, 2013-03-01 at 16:36 -0800, Jeremy Allison wrote:
> >>>> On Fri, Mar 01, 2013 at 08:29:19PM +0100, Stefan (metze) Metzmacher
> >> wrote:
> >>>>
> >>>>> Yes, more clear than commit 83ce1f9d9504156ea1b8e2a4f5655a48b5bcd1f1
> is
> >>>>> not possible :-)
> >>>>
> >>>> Ok, reviewed and pushed to autobuild.
> >>>>
> >>>> Here is the bug I created to back-port all the
> >>>> tevent library changes to v4-0-test so we
> >>>> can these released. Patchset is attached to
> >>>> the bug.
> >>>>
> >>>> https://bugzilla.samba.org/show_bug.cgi?id=9695
> >>>
> >>> I've only looked at this from the sidelines, but I wanted to say a big
> >>> congratulations to all of you involved in this effort.  As I understand
> >>> it, this isn't just a minor update to tevent as the version number
> >>> suggests, it really is a major fix of some of the long-standing issues
> >>> while also adding an important operating mode for FreeBSD.
> >>
> >> Thanks! There are some FreeBSD fixes, but the kqueue support is still
> >> not there
> >> and I'm not sure it will make it in future. The problem is that
> >> the kernel doesn't tell the caller the full details about the
> >> file descriptor state, read and write are always reported independently.
> >>
> >>
> > Metze,
> >
> > I'm not sure I understand what you mean here.  Can we not support cases
> > where read/write are reported independently?
> >
> > That may be a problem for Solaris/illumos "port" support.
>
> Solaris/illumos has /dev/poll, correct?
> I haven't looked into details, but that should provide the same features
> as poll(),
> which means the kernel reports POLLIN and POLLOUT together.


In re-reading the doc, for port_associate, I think you are right.  It'll do
that.  But it will only give 1 event per fd.

http://docs.oracle.com/cd/E23824_01/html/821-1465/port-associate-3c.html#scrolltoc

Thanks,

-Ira