Re: [Exim] Possible Bug in exim 3.953

Startseite
Nachricht löschen
Nachricht beantworten
Autor: Jochen Erwied
Datum:  
To: Philip Hazel
CC: Exim Mailinglist
Betreff: Re: [Exim] Possible Bug in exim 3.953
Monday, February 25, 2002, 10:18:42 AM, you wrote:

> Well, Exim 3.95x doesn't intend to set a debug flag on any sockets. I
> don't know enough about sockets to try anything like that. I don't
> remember making any changes to the way Exim handles sockets from the
> 3.3x releases. I don't think I should hold up the 4.00 release for
> this... we'll have to track it down afterwards.


Well, after digging through the source, I think you do set debugging
for the socket. This is the relevant part (and fix):

--- daemon.c.orig       Mon Feb 25 13:32:14 2002
+++ daemon.c    Mon Feb 25 13:32:32 2002
@@ -853,5 +853,5 @@
     /* Set TCP_NODELAY; Exim does its own buffering. */


-    setsockopt(listen_sockets[sk], SOL_SOCKET, TCP_NODELAY, (uschar *)(&on),
+    setsockopt(listen_sockets[sk], IPPROTO_TCP, TCP_NODELAY, (uschar *)(&on),
       sizeof(on));


TCP_NODELAY is identical to SO_DEBUG. So when used in combination with
SOL_SOCKET, debugging on kernel-level is turned on, at least for
Linux.

The patch is for the just released 4.00, by the way.

--
mbs GmbH      | http://www.mbs-software.de
Jochen Erwied | mailto:Jochen.Erwied@mbs-software.de
Roemerstr. 15 | phone: +49-2151-7294-0
47809 Krefeld | fax:   +49-2151-7294-50