I set up a PPTP server on my linode (a freshly installed Ubuntu 10.04 LTS image) and I consistently get a kernel panic when trying to VPN in from a client. This seems to happen from some clients but not from others (all using the same credentials). For those that it does happen, it happens consistently. Any ideas on how to resolve this?

The kernel panic shows this:

BUG: unable to handle kernel NULL pointer dereference at   (null)
IP: [<c0193144>] page_address+0x14/0xe0
*pdpt = 000000001eaf3007 *pde = 0000000000000000 
Oops: 0000 [#1] SMP 
last sysfs file: /sys/kernel/uevent_seqnum
Modules linked in:

Pid: 0, comm: swapper Not tainted 2.6.38.3-linode32 #1  
EIP: 0061:[<c0193144>] EFLAGS: 00010286 CPU: 0
EIP is at page_address+0x14/0xe0
EAX: 00000000 EBX: 00000000 ECX: 00000251 EDX: 00000250
ESI: 00000250 EDI: de6b7980 EBP: 20411000 ESP: df40feb4
 DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0069
Process swapper (pid: 0, ti=df40e000 task=c079df20 task.ti=c0788000)
Stack:
 deb58340 00000250 de6b7980 20411000 c049e742 00000000 000000d0 dee8a000
 00d02280 deb58be4 00000010 deb58000 deb59010 000000c0 deb58340 00000001
 deb58340 de6af8d8 c049fcff 00000000 00000001 df40ff9c dfbba380 df40ff78
Call Trace:
 [<c049e742>] ? xennet_alloc_rx_buffers+0x1c2/0x300
 [<c049fcff>] ? xennet_poll+0x4df/0xc20
 [<c04fc43a>] ? net_rx_action+0x9a/0x130
 [<c01380ec>] ? __do_softirq+0x7c/0x130
 [<c0138070>] ? __do_softirq+0x0/0x130
 <IRQ> 
 [<c0137fe5>] ? irq_exit+0x65/0x70
 [<c043a03d>] ? xen_evtchn_do_upcall+0x1d/0x30
 [<c0109487>] ? xen_do_upcall+0x7/0xc
 [<c01013a7>] ? hypercall_page+0x3a7/0x1010
 [<c0105b8f>] ? xen_safe_halt+0xf/0x20
 [<c010f66f>] ? default_idle+0x2f/0x60
 [<c0107ed2>] ? cpu_idle+0x42/0x70
 [<c07ca8ac>] ? start_kernel+0x2da/0x2df
 [<c07ca410>] ? unknown_bootoption+0x0/0x190
 [<c07cdaa5>] ? xen_start_kernel+0x530/0x538
Code: 89 c2 b8 2c 1e 85 c0 e9 3f ff ff ff 0f 0b eb fe 8d b4 26 00 00 00 00 83 ec 10 89 1c 24 89 c3 89 74 24 04 89 7c 24 08 89 6c 24 0c <8b> 00 c1 e8 1e 69 c0 80 03 00 00 05 40 05 7c c0 2b 80 4c 03 00 
EIP: [<c0193144>] page_address+0x14/0xe0 SS:ESP 0069:df40feb4
CR2: 0000000000000000
---[ end trace 19ddaabd0d19ad12 ]---
Kernel panic - not syncing: Fatal exception in interrupt
Pid: 0, comm: swapper Tainted: G      D     2.6.38.3-linode32 #1
Call Trace:
 [<c063cfbf>] ? panic+0x57/0x13e
 [<c010bec6>] ? oops_end+0x96/0xa0
 [<c011e362>] ? no_context+0xc2/0x190
 [<c011e58f>] ? bad_area_nosemaphore+0xf/0x20
 [<c011e943>] ? do_page_fault+0x223/0x3e0
 [<c0184325>] ? __alloc_pages_nodemask+0xf5/0x670
 [<c011e720>] ? do_page_fault+0x0/0x3e0
 [<c063fea6>] ? error_code+0x5a/0x60
 [<c011e720>] ? do_page_fault+0x0/0x3e0
 [<c0193144>] ? page_address+0x14/0xe0
 [<c049e742>] ? xennet_alloc_rx_buffers+0x1c2/0x300
 [<c049fcff>] ? xennet_poll+0x4df/0xc20
 [<c04fc43a>] ? net_rx_action+0x9a/0x130
 [<c01380ec>] ? __do_softirq+0x7c/0x130
 [<c0138070>] ? __do_softirq+0x0/0x130
 <IRQ>  [<c0137fe5>] ? irq_exit+0x65/0x70
 [<c043a03d>] ? xen_evtchn_do_upcall+0x1d/0x30
 [<c0109487>] ? xen_do_upcall+0x7/0xc
 [<c01013a7>] ? hypercall_page+0x3a7/0x1010
 [<c0105b8f>] ? xen_safe_halt+0xf/0x20
 [<c010f66f>] ? default_idle+0x2f/0x60
 [<c0107ed2>] ? cpu_idle+0x42/0x70
 [<c07ca8ac>] ? start_kernel+0x2da/0x2df
 [<c07ca410>] ? unknown_bootoption+0x0/0x190
 [<c07cdaa5>] ? xen_start_kernel+0x530/0x538

Try running one of the older Linode kernels instead of 2.6.38.3-linode32. If that fixes it, raise a ticket.

_________________
/ Peter

Downgrading to an older kernel makes the panic go away. Raised a ticket with Linode who say it is being investigated.

Does this happen with Latest 2.6 Paravirt - which is now 2.6.39 based?

http://www.linode.com/kernels/
http://www.linode.com/kernels/rss.xml

-Chris

2.6.39 does not resolve the issue. It still happens.

I have exactly the same problem: with anything above 2.6.18.8-linode22, my Linode in London panics (maybe it worked 5 or 6 times top with the latest paravirt).

On a side note yesterday I have spent hours trying to figure out why with kernel 2.6.18.8 the GRE negotiation was failing while tcpdump was spitting out ICMP related errors. No matter how many times I have rebooted my router and my Linode negotiations would simply fail.

Today (27th June, 11PM GMT+1) everything worked flawlessly with the same configuration, the same router and the same clients, it was really frustrating.

How is the ticket going?

The ticket is going nowhere. You might want to file one of your own.

Apparently it's a regression:

https://bugzilla.redhat.com/show_bug.cgi?id=248716

you may want to include the above link to the bug report you have already opened with XEN-devel.