Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 3 Aug 2012 00:17:46 +1000 (EST)
From:      Bruce Evans <brde@optusnet.com.au>
To:        davidxu@freebsd.org
Cc:        src-committers@freebsd.org, Giovanni Trematerra <gianni@freebsd.org>, svn-src-all@freebsd.org, Konstantin Belousov <kib@freebsd.org>, bde@freebsd.org, Bruce Evans <brde@optusnet.com.au>, svn-src-head@freebsd.org
Subject:   Re: svn commit: r238936 - in head/sys: fs/fifofs kern sys
Message-ID:  <20120803001044.X2863@besplex.bde.org>
In-Reply-To: <501A72AE.30309@gmail.com>
References:  <201207310548.q6V5mZHf091624@svn.freebsd.org> <CACfq090si0g7QG9UR%2Bmb%2BO5M4KH0=WPfWnF5NzZjYzKBp8hQ5Q@mail.gmail.com> <50179581.9070805@gmail.com> <CACfq091K=1DUOee4LorPPSoEV8KnT570GGXDA01mHDrV12kKiQ@mail.gmail.com> <20120802051805.P3345@besplex.bde.org> <5019ED93.8060802@gmail.com> <20120802173441.X1291@besplex.bde.org> <501A72AE.30309@gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 2 Aug 2012, David Xu wrote:

> On 2012/8/2 16:12, Bruce Evans wrote:
>> ...
>> I made similar patches for sockets (to set POLLHUP on hangup (now in
>> -current) and to not set POLLIN on hangup unless there is still data
>> to be read).  I started killing POLLINIGNEOF for sockets. -current
>> added it for nameless pipes instead :-(.  With the new fifo
>> implementation, POLLINIGNEOF is even more of a mistake for sockets,
>> but more needed for pipes since named pipes are fifos.
>> ...
> I think you can kill POLLINIGNEOF at all,  I have grepped, and there is no 
> external user,
> only pipe and socket code use it internally. The POLLINIGNEOF is confusing 
> because
> it has same prefix with POLLIN, POLLOUT and other POLL flags.

Did you grep all of google for it :-).  All of ports should be enough.

Uses of it in the kernel are certainly gone, but it was intentionally
put in the user API to previde a workaround for the policy that was
hard-coded in the kernel.  The policy changed slightly, and you could
set POLLINIGNEOF to either go back to the old policy or get the new
policy for more cases.  Hopefully this was never actually used except
for testing and termporary workarounds.  But it was expanded to work
on nameless pipes as well as fifos and sockets.

Bruce



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20120803001044.X2863>