Every time any client connects to the PPTP daemon poptop, the kernel panics with a oops on an interrupt handler Reproducible: Always Steps to Reproduce: 1. Activate pptpd 2. Connect with any client 3. Watch the system lock up Actual Results: The system prints out a oops message and freezes Expected Results: Activate a VPN connection using PPTP The firewall is running nothing, except a cron daemon. The only "unusual" thing is that I'm using grsecurity at high level. I'm using the latest kernel source package from gentoo and the latest available pptpd daemon.
Also, connecting FROM that machine TO a poptop server with the pptpclient package causes a kernel panic... this makes me wonder wether it could be an hardware problem instead...
Recompiled the kernel without grsec... everything is shitty as usual :|
which kernel ?
2.4.20-r1
I've got the same problem - and I don't know whether it's the kernel or one of the packages (ppp or pptpd). After it didn't work on the gentoo-box I've re-installed gentoo on the same machine, but with the same kernel-config. And again, it doesn't work. On another machine everything worked fine from the beginning on... P.S.: I'm using kernel 2.4.20-r5
I have a related problem. I have kernel panics when _disconnect_ing from PPTP when a firewall with NAT using iptables is enabled. Other people report that the same problem does not affect the Vanilla kernel here (haven't tried this myself): http://forums.gentoo.org/viewtopic.php?t=44363
Yes you are perfectly right: the problem is on DISCONNECTION, in fact. I may add that using a precompiled kernel from other distros works nicely with the same conf...
Could we please have a register dump and can you generate the necessary information from that dump using 'ksymoops' please? If this bug is no longer reproducable, I'll have to close it as NEEDINFO...
Created attachment 16948 [details] kernel panic, 2.4.20-gentoo-r6 with pptop NAT module and pptpd running written manually from screen 2003-09-02, may contain typos. (added as attachment to preserve layout)
This has happened to me on two separate machines on several occasions (nearly identical kernel configurations, but not quite, manually configured each time). It has happened on both 2.4.20-gentoo-r5 and -r6 . It seems only to happen when the iptables NAT modules for pptp are loaded, without them it has been working fine. I have seen several such bug-reports, unfortunately I can't find the most relevant right now. Here is one regarding gentoo though: https://bugzilla.netfilter.org/cgi-bin/bugzilla/show_bug.cgi?id=54 I now have a test machine which I can crash as often as I want, if anyone wants info (but only until the middle of next week). Maybe the pptp modules patched into the Gentoo-sources could be updated? Sorry about the attachment, can someone delete it? I only seem to be able to edit it. Oops: 0000 CPU: 0 EIP: 0010:[<d08a30f3>] Tainted: P EFLAGS: 00010213 eax: 00000000 ebx: fffffff4 ecx: cec93ecc edx: 00000006 esi: 00000000 edi: cedbba68 ebp: cf1b896c esp: cec05d44 ds: 0018 es: 0018 ss: 0018 Process pptpctrl (pid:23813, stackpage=cec05000) Stack: cedbbb60 cedbba00 cedbba98 00000014 d08a3378 cedbba00 00000014 00000000 cf1b896c 00000001 fb01a8c0 00000001 00000000 cedbba00 cec05e24 d0889040 00000003 d0885169 cf1b8958 00000028 cedbba00 00000003 cec05da4 00000001 Call Trace: [<d08a3378>] [<d0889040>] [<d0885169>] [<c0383570>] [<c0375a24>] [<c0383570>] [<c0383570>] [<c0375da6>] [<c0383570>] [<d0888b5c>] [<c0382233>] [<c0383570>] [<c03934e7>] [<c0369200>] [<c03940a4>] [<c0368d1b>] [<c0394e6e>] [<c038a962>] [<c0202b7c>] [<c03a8a04>] [<c0364ae9>] [<c036518b>] [<c0239a35>] [<c0237946>] [<c0237a28>] [<c01f56df>] Code: 8b 43 18 85 c0 75 16 89 1c 24 e8 9e e2 ff ff 8b 36 39 fe 75 <0>Kernel panic: Aiee, killing interrupt handler! In interrupt hanlder - not syncing
minor addition: It works fine if the following modules (in addition to slhc, ppp_generic, ppp_mppe, bsd_comp) are loaded: ip_tables, iptable_filter, ip_conntrack, ipt_conntrack, iptable_nat, ip_conntrack_proto_gre, ip_nat_proto_gre I tested connecting and disconnection between each insmod. It crashes upon connecting after ip_conntrack_pptp has been loaded. The output this time was: Oops: 0000 CPU: 0 EIP: 0010:[<d08a00f3>] Tainted: P EFLAGS: 00010217 eax: 00000000 ebx: fffffff4 ecx: cef8aecc edx: 00000002 esi: 00000000 edi: ce647e88 ebp: cf09388c esp: ce7cfcd8 ds: 0018 es: 0018 ss: 0018 Process pptpctrl (pid: 27365, stackpage=ce7cf000) Stack: ce647480 cf0938ac ce647e20 000000a8 d08a04fe ce647e20 00000000 00000001 cf456c00 ce647e20 ce647eb8 000000a8 d08a0360 cf09388c cf0938a0 00000094 ce647e20 00000003 fb01a8c0 00000001 00000094 ce647e20 ce7cfdd8 d0882040 Call Trace: [<d08a04fe>] [<d09a0360>] [<d0882040>] [<d087e169>] [<c0383570>] [<c0375a24>] [<c0383570>] [<c0383570>] [<c0375da6>] [<c0383570>] [<d0881b5c>] [<c0382233>] [<c0382570>] [<c039912d>] [<c039318e>] [<c03934e7>] [<c0369200>] [<c03940a4>] [<c0368d1b>] [<c0388115>] [<c02bb4fe>] [<c03a9652>] [<c0364b65>] [<c02141d6>] [<c0364dfe>] [<c0238747>] [<c01f56df>] Code: 8b 43 18 85 c0 75 16 89 1c 24 e8 9e c2 ff ff 8b 36 39 fe 75 <0>Kernel panic: Aiee, killing interrupt handler! In interrupt handler - not syncing
Can you stick that OOPS into a text file, emerge 'ksymoops' and run: ksymoops < file_with_oops > file_with_output ... and paste file_with_output to bugzilla please?
ping Arne. This bug is getting old. Plasmaroo asked you to do the following: Can you stick that OOPS into a text file, emerge 'ksymoops' and run: ksymoops < file_with_oops > file_with_output ... and paste file_with_output to bugzilla please? Unfortunately the OOPS text isn't useful without this information, so if you are running a newer kernel and/or the problem has gone away, we'll need to close this bug. Thanks.
No response from reporter, closing NEEDINFO