[pptp-server] PPTPCTRL Kills my Linux Server!

[Jim Morris]

Help!

I've been using PoPToP to provide VPN services to Windows 98 clients on
two different networks, using two different Linux servers.

The system I really *NEED* to run PPTPD on has locked up and/or rebooted
three times in the past month, with PPTPCTRL as the apparent cause,
based on the register dump I end up with in my syslog.  The reboot
usually happens right when a client connects, or a couple of minutes
after a client connects to PPTPD, and the log messages in
/var/log/messages look like below - this is from todays lockup, which
took an hour to recover from, due to having to manually run e2fsck on a
couple of 18GB drives:

12:33:04 pptpd[9122]: GRE: Bad checksum from pppd
12:33:28 last message repeated 2 times
12:33:31 pppd[9123]: Protocol-Reject for unsupported protocol 0x4040
12:34:32 kernel: Unable to handle kernel NULL pointer dereference at
    virtual address 00000070
12:34:32 kernel: current->tss.cr3 = 0ab1d000, %cr3 = 0ab1d000
12:34:32 kernel: *pde = 00000000
12:34:32 kernel: Oops: 0002
12:34:32 kernel: CPU:    1
12:34:32 kernel: EIP:    0010:[<d088d21e>]
12:34:32 kernel: EFLAGS: 00010296
12:34:32 kernel: eax: 00000800   ebx: 0000008b   ecx: cde4ac00  edx:
00000000
12:34:32 kernel: esi: cde4ae60   edi: cde4ac48   ebp: cde4ac48  esp:
ca06beb8
12:34:32 kernel: ds: 0018   es: 0018   ss: 0018
12:34:32 kernel: Process pptpctrl (pid: 9122, process nr: 78,
stackpage=ca06b000)
12:34:32 kernel: Stack: ca547000 00000000 ffff8b3f 00000000 cde4ac48
00000000
    cd6e68e0 cde4af2e
12:34:32 kernel:        0000002c d088cef4 cde4ac00 cde4ac00 ca547000
ca4a7000
    ca06bf8c d088cb0b
12:34:32 kernel:        cde4ac00 ca547000 ca4a7000 c01b316a ca547000
00000000
    00000000 c01a98c2
12:34:32 kernel: Call Trace: [<d088cef4>] [<d088cb0b>]
[pty_unthrottle+38/72]
    [check_unthrottle+42/48] [read_chan+1510/1792] [tty_read+176/208]
    [sys_read+194/232]
12:34:32 kernel:        [system_call+52/56] [startup_32+43/164]
12:34:32 kernel: Code: f0 ff 4a 70 0f 94 c0 84 c0 74 09 52 e8 91 64 8c
ef 83 c4 04

Both systems run Redhat 6.0, with the Linux 2.2.14 kernel.  One system
is SMP, whereas the other is a uniprocessor machine.  PPPD 2.3.7 (stock
for Redhat 6.0) is in use.  I've tried PPTPD v1.0.0 and v1.1.1 - problem
has occured with both.  I've also had the problem with Linux kernel
2.2.11 and 2.2.14.  The system having the problem is the SMP system - in
one case, the Kernel "oops" message occured on CPU 0 - on the other two
cases, on CPU 1, as above.

PPTPD seems to be the only thing that has caused a system lockup or
reboot on this particular server since I built it back last July....

Any help would be greatly appreciated!
 --
/------------------------------------------------\
| Jim Morris  | Business:  jmorris at rtc-group.com |
|             | Personal:  Jim at Morris.net        |
|------------------------------------------------|
|       AOL Instant Messenger:  JFM2001          |
\------------------------------------------------/