[milters] Archive

Lists Index Date Thread Search

Article: 1861
From: Anthony Howe
Date: 2008-05-13 10:42:49 -0400
Subject: Re: milter-null: X-Null-Tag differs between 32bit- and

Removal...........: milters-request@milter.info?subject=remove
More information..: http://www.milter.info/#Support
--------------------------------------------------------

Andreas Piper uttered...:
> I am using 32bit-servers for sending my mails. Some of my receiving servers 
> are 64bit-architecture, and on these the X-Null-Tag-value differs from the 
> value calculated by the 32bit-systems (using identical input-parameters), 
> leading to false positives.
> 
> Some digging in the sources showed, that the reason for this behaviour is the 
> differing size of time_t ("4" for 32bit, "8"for 64bit), wich is
used as 
> length-parameter in md5_append (milter-null.c lines 402 and 500) in the 
> creation of the X-Null-Tag-value.

OK. I've done some study into this issue and come up with a machine 
independent solution that should avoid issues WRT different sizes for 
time_t for the next release.

> *but* with this change I now observe in the log some error messages of type
> 
> May 11 23:35:54 vhrz151 sm-mta[25345]: m4BLZpIg025345: Milter (milter-null): 
> write(D) returned -1, expected 123: Broken pipe
> May 11 23:35:54 vhrz151 sm-mta[25345]: m4BLZpIg025345: Milter (milter-null): 
> to error state
> 
> which did not occur before.
> 
> Do you have an idea how to solve this problem?

The above errors seriously doubt are due to you changing the length 
constant. I suspect some other issue is at play.

-- 
Anthony C Howe        Twitter: SirWumpus                    SnertSoft
+33 6 11 89 73 78       Skype: SirWumpus        BarricadeMX & Milters
http://www.snert.com/     ICQ: 7116561
     http://www.snertsoft.com/

Lists Index Date Thread Search