Bug #958

Sendmail bug back?

Added by dave1 almost 7 years ago. Updated over 6 years ago.

Status:ClosedStart date:
Priority:UrgentDue date:
Assignee:-% Done:

0%

Category:-
Target version:-

Description

I have sendmail randomly coredumping on signal 11 on my freshly upgraded
dragonfly 1.12 box. It looks like the bug Matt fixed, so I spent some
time making sure that patch to main.c actually found it's way into this
binary. (That doesn't mean I actually succeeded at this but it's quite
likely I did.)

# gdb sendmail sendmail.core
GNU gdb 6.7.1
...
Core was generated by `sendmail'.
Program terminated with signal 11, Segmentation fault.
#0 newenvelope (e=0x80fc740, parent=0x80fc740, rpool=0x483810c0)
at /usr/src/usr.sbin/sendmail/../../contrib/sendmail-8.14/sendmail/envelope
.c:97
97 /usr/src/usr.sbin/sendmail/../../contrib/sendmail-8.14/sendmail/envelop
e.c: No such file or directory.
in /usr/src/usr.sbin/sendmail/../../contrib/sendmail-8.14/sendmail/enve
lope.c
(gdb) bt
#0 newenvelope (e=0x80fc740, parent=0x80fc740, rpool=0x483810c0)
at /usr/src/usr.sbin/sendmail/../../contrib/sendmail-8.14/sendmail/envelope
.c:97
#1 0x0809b565 in smtp (nullserver=0x0, d_flags=0x80fe18c, e=0x80fc740)
at /usr/src/usr.sbin/sendmail/../../contrib/sendmail-8.14/sendmail/srvrsmtp
.c:3789
#2 0x080aff32 in main (argc=5, argv=<value optimized out>, envp=0xbfbff6f8)
at main.c:2641

I've placed the corefile and the binary on leaf in my "crash" directory,
for perusal of better eyes than mine. Thanks in advance.
--
Dave Hayes - Consultant - Altadena CA, USA -
>>> The opinions expressed above are entirely my own <<<

"Never give people anything they ask for until at least a
day has passed," said Nasrudin. "For experience shows that
they only appreciate something when they have had the
opportunity of doubting whether they will get it or not."

newenvelope.p2 (1.3 KB) dragonfly-bugs, 03/04/2008 03:38 PM

History

#1 Updated by dragonfly-bugs almost 7 years ago

It would have been nice to contact sendmail.org to tell them about it...
Anyway, a patch is attached.

#2 Updated by dillon almost 7 years ago

:I have sendmail randomly coredumping on signal 11 on my freshly upgraded
:dragonfly 1.12 box. It looks like the bug Matt fixed, so I spent some
:time making sure that patch to main.c actually found it's way into this
:binary. (That doesn't mean I actually succeeded at this but it's quite
:likely I did.)
:
:#1 0x0809b565 in smtp (nullserver=0x0, d_flags=0x80fe18c, e=0x80fc740)
: at /usr/src/usr.sbin/sendmail/../../contrib/sendmail-8.14/sendmail/srvrsmtp
:.c:3789
:#2 0x080aff32 in main (argc=5, argv=<value optimized out>, envp=0xbfbff6f8)
:at main.c:2641
:
:I've placed the corefile and the binary on leaf in my "crash" directory,
:for perusal of better eyes than mine. Thanks in advance.
:--
:Dave Hayes - Consultant - Altadena CA, USA -
:>>> The opinions expressed above are entirely my own <<<

Hell. It's the same bug, but in a different location.

Fetch this patch file and apply it in /usr/src/usr.sbin/sendmail,
then recompile sendmail. Tell me if that fixes the problem.

fetch http://apollo.backplane.com/DFlyMisc/sendmail02.patch

-Matt

#3 Updated by dillon almost 7 years ago

:
:It would have been nice to contact sendmail.org to tell them about it...
:Anyway, a patch is attached.

Ach. two more places. I only caught the second one.

Don't worry, contacting sendmail was next on my list. Can I assume
you will take care of it, Claus ?

I should have checked the other use cases when I reported the
original bug.

-Matt

#4 Updated by dragonfly-bugs almost 7 years ago

On Tue, Mar 04, 2008, Matthew Dillon wrote:

Yes, it is in our CVS tree for a while already (currently in alpha
testing as you might have guessed from the diff).

Someone else reported one of those and I found the third. That
should be all...

#5 Updated by dave1 almost 7 years ago

Well actually I just did a "cvs up -Pd" in /usr/src/usr.sbin/sendmail
to pick up your commits.

With the previous sendmail binary, coredumps happened within 15 minutes
of starting the server. I've been running with the current binary for
approximately one hour with no coredumps.

I also had a user who had problems sending mail under the old binary.
This user has run the same test case and found no issues or problems.
--
Dave Hayes - Consultant - Altadena CA, USA -
>>> The opinions expressed above are entirely my own <<<

Anarchism is founded on the observation that since few
are wise enough to rule themselves, even fewer are wise
enough to rule others.
- Edward Abbey

#6 Updated by dillon almost 7 years ago

:With the previous sendmail binary, coredumps happened within 15 minutes
:of starting the server. I've been running with the current binary for
:approximately one hour with no coredumps.
:
:I also had a user who had problems sending mail under the old binary.
:This user has run the same test case and found no issues or problems.
:--
:Dave Hayes - Consultant - Altadena CA, USA -
:>>> The opinions expressed above are entirely my own <<<

Yah, I decided it could go in immediately when Claus posted the
more comprehensive patch.

I am going to roll 1.12.1 on wednesday. There was another MFC that
went in a few days ago that we also want to get in there, as well.

-Matt
Matthew Dillon
<>

Also available in: Atom PDF