kernel: pptpcm[32528]: segfault at d6a1ec30 ip 0804d6b1 sp bfa3e0e0 error 5 in pptp[8048000+f000]

[mizu]

Ubuntu Jaunty,

VPN tunnel through network-manager-pptp

uname -a
Linux hostname 2.6.28-11-generic #40-Ubuntu SMP Fri Apr 3 17:39:51 UTC 2009 i686 GNU/Linux

 pppd[957]: Plugin /usr/lib/pppd/2.4.5//nm-pptp-pppd-plugin.so loaded.
 pppd[957]: pppd 2.4.5 started by root, uid 0
 pppd[957]: Using interface ppp0
 pppd[957]: Connect: ppp0 <--> /dev/pts/4
 kernel: [16196.412502] pptpcm[32528]: segfault at d6a1ec30 ip 0804d6b1 sp bfa3e0e0 error 5 in pptp[8048000+f000]
 pppd[957]: CHAP authentication succeeded
 kernel: [16197.613515] PPP MPPE Compression module registered
 pppd[957]: MPPE 128-bit stateless compression enabled
 pppd[957]: local IP address 10.1.1.252
 pppd[957]: remote IP address 10.199.6.1

[apanloco (apanloco)]

My syslog is filled with pptpcm segfaults, on Lucid. And the reliability of my pptp connection is just not there.

[Paul Crawford (psc-sat)]

Seeing the same sort of problem with 10.04 i386 on my AMD-based system:

uname -a
Linux paul-ubuntu 2.6.32-25-generic #44-Ubuntu SMP Fri Sep 17 20:26:08 UTC 2010 i686 GNU/Linux

/var/log/messages
Sep 29 20:57:12 paul-ubuntu pppd[4121]: Modem hangup
Sep 29 20:57:12 paul-ubuntu pppd[4121]: Connect time 4.9 minutes.
Sep 29 20:57:12 paul-ubuntu pppd[4121]: Sent 10792614 bytes, received 1046078 bytes.
Sep 29 20:57:12 paul-ubuntu kernel: [ 3061.825783] pptpcm[4134]: segfault at c0cf4a64 ip 0804d3f0 sp bfcd2c40 error 5 in pptp[8048000+e000]
Sep 29 20:57:12 paul-ubuntu pppd[4121]: Connection terminated.
Sep 29 20:57:14 paul-ubuntu pppd[4121]: Exit.

I don't know the reason for the VPN 'Modem hangup' but I am pretty sure it should not cause a segfault! Have now enabled apport and will see if it happens again soon.

[Paul Crawford (psc-sat)]

Got an apport trace & core dump and logged as bug #656560 so these two are probably duplicates.

Also similar are Bug #314121 and maybe https://bugzilla.redhat.com/show_bug.cgi?id=504131 (which it seems is not being investigated as that version of Fedora was end-of-life, despite the bug being old and unfixed!)

[Jakob Moellers (jakob-moellers)]

Discovered a strange behaviour:

At home, my the VPN-connection can be established but in my second flat the problem described above occurs (same as #3).

Other strange thing: in my second flat, my Windows Vista computer can connect to the VPN without any problems.

Using Ubuntu 10.04, 2.6.32-25-generic.

/var/log/messages:

Oct 27 20:09:05 jakob-unilaptop pppd[1972]: Plugin /usr/lib/pppd/2.4.5//nm-pptp-pppd-plugin.so loaded.
Oct 27 20:09:05 jakob-unilaptop pppd[1972]: pppd 2.4.5 started by root, uid 0
Oct 27 20:09:05 jakob-unilaptop pppd[1972]: Using interface ppp0
Oct 27 20:09:05 jakob-unilaptop pppd[1972]: Connect: ppp0 <--> /dev/pts/1
Oct 27 20:09:07 jakob-unilaptop kernel: [ 190.018549] pptpcm[1983]: segfault at c0c30864 ip 0804d3f0 sp bf87e050 error 5 in pptp[8048000+e000]
Oct 27 20:09:07 jakob-unilaptop pppd[1972]: Modem hangup
Oct 27 20:09:07 jakob-unilaptop pppd[1972]: Connection terminated.
Oct 27 20:09:07 jakob-unilaptop pppd[1972]: Exit.

[Paul Crawford (psc-sat)]

I built a version of pptp with debug enabled, and caught the segmentation fault again. This time I get the following in the syslog file:

{{{
Nov 16 23:07:33 paul-ubuntu pptp[5592]: nm-pptp-service-5464 log[decaps_gre:pptp_gre.c:414]: buffering packet 1323 (expecting 1322, lost or reordered)
Nov 16 23:08:09 paul-ubuntu pptp[5598]: nm-pptp-service-5464 log[logecho:pptp_ctrl.c:677]: Echo Request received.
Nov 16 23:08:09 paul-ubuntu pptp[5598]: nm-pptp-service-5464 log[ctrlp_rep:pptp_ctrl.c:251]: Sent control packet type is 6 'Echo-Reply'
Nov 16 23:08:09 paul-ubuntu pptp[5598]: nm-pptp-service-5464 log[pptp_read_some:pptp_ctrl.c:551]: read error: Connection reset by peer
Nov 16 23:08:09 paul-ubuntu pptp[5598]: nm-pptp-service-5464 log[callmgr_main:pptp_callmgr.c:258]: Closing connection (shutdown)
Nov 16 23:08:09 paul-ubuntu pptp[5598]: nm-pptp-service-5464 log[pptp_send_ctrl_packet:pptp_ctrl.c:622]: write error: Broken pipe
Nov 16 23:08:09 paul-ubuntu pptp[5598]: nm-pptp-service-5464 log[call_callback:pptp_callmgr.c:79]: Closing connection (call state)
Nov 16 23:08:09 paul-ubuntu kernel: [ 6282.854809] pptpcm[5598]: segfault at c0bbb964 ip 0804d3f0 sp bf8dcb40 error 5 in pptp[8048000+e000]
}}}

Then using gdb on the core bump extracted from apport's crash report I get the following:

{{{
(gdb) where
#0 0x0804d3f0 in pptp_fd_set (conn=0x96f6858, read_set=0xbf8dcc3c, write_set=0xbf8dcbbc, max_fd=0xbf8dce34) at pptp_ctrl.c:464
#1 0x08052e7f in callmgr_main (argc=3, argv=0xbf8dcf1c, envp=0xbf8de2dc) at pptp_callmgr.c:266
#2 0x0804a8e0 in launch_callmgr (inetaddr=..., phonenr=0x0, argc=5, argv=0xbf8de2c4, envp=0xbf8de2dc) at pptp.c:505
#3 0x0804a722 in open_callmgr (inetaddr=..., phonenr=0x0, argc=5, argv=0xbf8de2c4, envp=0xbf8de2dc, pty_fd=0, gre_fd=4) at pptp.c:474
#4 0x0804a225 in main (argc=5, argv=0xbf8de2c4, envp=0xbf8de2dc) at pptp.c:375
}}}

The call that causes the fault appears to be the one at pptp_callmgr.c line 266 where it is doing the 'shutdown' steps.

[Jakob Möllers (jakob-moellers-googlemail)]

Solved the problem for me:

I changed my internet provider. It turned out that my internet provider blocked the VPN connection.

Now everything works perfectly.

So you guys could ask your provider, too...

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Eugene (eugene-tarasov)]

It's not in network-manager. This bug in pptp. Other distros are in fire too...