Re: Overhauling our interrupt handling

From: Kyotaro HORIGUCHI <horiguchi(dot)kyotaro(at)lab(dot)ntt(dot)co(dot)jp>
To: andres(at)2ndquadrant(dot)com
Cc: pgsql-hackers(at)postgresql(dot)org, hlinnakangas(at)vmware(dot)com, tgl(at)sss(dot)pgh(dot)pa(dot)us, robertmhaas(at)gmail(dot)com
Subject: Re: Overhauling our interrupt handling
Date: 2015-01-16 01:41:02
Message-ID: 20150116.104102.131609138.horiguchi.kyotaro@lab.ntt.co.jp
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hello,

> > I think I should finilize my commitfest item for this issue, with
> > .. "Rejected"?
>
> Fine with me.

done.

> > > 0001: Replace walsender's latch with the general shared latch.
> > >
> > > New patch that removes ImmediateInteruptOK behaviour from walsender. I
> > > think that's a rather good idea, because walsender currently seems to
> > > assume WaitLatchOrSocket is reentrant - which I don't think is really
> > > guaranteed.
> > > Hasn't been reviewed yet, but I think it's not far from being
> > > committable.
> >
> > Deesn't this patchset containing per-socket basis non-blocking
> > control for win32? It should make the code (above the win32
> > socket layer itself) more simpler.
>
> I don't think so - we still rely on it unfortunately.

Does "it" mean win32_noblock? Or the nonblocking bare win32
socket? The win32-per-sock-blkng-cntl patch in the below message
should cover both of them.

http://www.postgresql.org/message-id/54060AE5.5020502@vmware.com

If you are saying it should be a patch separate from this, I'll
do so.

regareds,

--
Kyotaro Horiguchi
NTT Open Source Software Center

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Dilip kumar 2015-01-16 03:19:13 Re: TODO : Allow parallel cores to be used by vacuumdb [ WIP ]
Previous Message Peter Eisentraut 2015-01-16 01:30:15 Re: pg_rewind in contrib