system/3099: sendmail with MILTER option generates premature EOM

>Number: 3099
>Category: system
>Synopsis: sendmail with -DMILTER generates premature EOM
>Confidential: no
LAMP, Inc.
  Mr Dale Carstensen

  LAMP, Inc.                   office (505)662-2524
  1504 S Sage                     FAX (505)662-3588

	System      : OpenBSD 3.2
	Architecture: OpenBSD.i386
	Machine     : i386

I want to use the milter feature of sendmail to sort out SPAM, perhaps also virii or worms. I built sendmail with -DMILTER and the other Makefile mods to sendmail required. The resulting sendmail rejected every message sent to it as a daemon with the message

 collect: premature EOM: Resource temporarily unavailable

So I debugged the problem. I still don't know why, but when collect gets called the file structure it uses to read from the network has f_timeout set to -1 instead of something reasonable, but only if the sendmail has been compiled with -DMILTER. So I forced it to be an hour instead, which seems to be the intended default timeout. That eliminates the symptom. The authors of sendmail and/or libmilter may want to fix this some other way, but a context diff of my fix to sendmail/collect.c is included.
>How-To-Repeat:
        

Since I don't want a sendmail daemon running on port 25 that rejects all messages, I used -C/etc/mail/sendmail.cf-miltertest which had these alterations from my version of the proto.cf version and -bd:

231,234c231,235

< O DaemonPortOptions=Family=inet, address=0.0.0.0, Name=MTA
< O DaemonPortOptions=Family=inet6, address=::, Name=MTA6, M=O
< O DaemonPortOptions=Family=inet, address=0.0.0.0, Port=587, Name=MSA, M=E
< O DaemonPortOptions=Family=inet6, address=::, Port=587, Name=MSA6, M=O, M=E

---

> #O DaemonPortOptions=Family=inet, address=0.0.0.0, Name=MTA

> O DaemonPortOptions=Family=inet, address=0.0.0.0, Name=MTA, Port=2525

> #O DaemonPortOptions=Family=inet6, address=::, Name=MTA6, M=O

> #O DaemonPortOptions=Family=inet, address=0.0.0.0, Port=587, Name=MSA, M=E

> #O DaemonPortOptions=Family=inet6, address=::, Port=587, Name=MSA6, M=O, M=E


That is, in /usr/src/gnu/usr.sbin/sendmail/sendmail, with ./sendmail
having been made with the MILTER options:

#./sendmail -C/etc/mail/sendmail.cf-miltertest -bd

Then I used telnet to send a message:

telnet localhost 2525
ehlo myhost.dom
mail from: myemail@myhost.dom
rcpt to: myemail@myhost.dom
data

at which point I would get

354 Enter mail, end with "." on a line by itself
Connection closed by foreign host.

before I could enter any text.

After the fix, I could enter headers, a blank line, body lines and the
"." to finish my message, and "quit" end the session.  I tried setting
fp->f_timeout = 3600 with gdb, but it didn't appear to change the
symptoms.  Eventually I saw that the timeout is in milliseconds, so
using 3600000 did work.  I just couldn't tell the difference between
no timeout and 3.6 seconds when I was switching windows between gdb
and the telnet.

I would kill the sendmail process to clean up.  Also note that this
corrupts the contents of /var/run/sendmail.pid, so that should be
cleaned up, too, at least on openBSD.

>Fix:

	
*** src/gnu/usr.sbin/sendmail/sendmail/collect.c.orig   Tue Feb 18 10:13:00 2003
--- src/gnu/usr.sbin/sendmail/sendmail/collect.c        Tue Feb 18 11:25:13 2003
***************
*** 308,313 ****
--- 308,314 ----
        df = NULL;
        ignrdot = smtpmode ? false : IgnrDot;
Do you need help?
In addition to our free technical library, Pantek provides professional IT services to companies who utilize Open Source software:
Learn more about Pantek's Technical Support offerings
Place an order for 24/7/365 Technical Support here
Learn more about Pantek, who we are, and what we do
Chat with a Live Operator
Related openbsd.org Links
advocacy
announce
bugs
ipv6
misc
ports
ports-changes
security-announce
smp
source-changes
tech
Request more information about Pantek
        dbto = smtpmode ? TimeOuts.to_datablock : 0;
+       fp->f_timeout = dbto * 1000;    /* keep -DMILTER from killing collect */
        c = SM_IO_EOF;
        inputerr = false;
        headeronly = hdrp != NULL;

>Release-Note: