Crash at tcp_input+1401 in VBoxDD.dylib of 2.2.2r46594

[Chris Bergmann]

VM spontaneously closed during network activity. Guest is configured as a firewall with two NICs (NAT and local.) No meaningful VM logs found. Related MacOSX crash reporter log attached. Excerpt as folows

Thread 14 Crashed:
0   VBoxDD.dylib                  	0x18b73ac9 tcp_input + 1401
1   VBoxDD.dylib                  	0x18b63a6f slirp_select_poll + 863
2   VBoxDD.dylib                  	0x18b45164 drvNATAsyncIoThread(PDMDRVINS*, PDMTHREAD*) + 276
3   VBoxVMM.dylib                 	0x01db71d5 pdmR3ThreadMain(RTTHREADINT*, void*) + 213
4   VBoxRT.dylib                  	0x003103e0 rtThreadMain + 64
5   VBoxRT.dylib                  	0x0033c514 rtThreadNativeMain(void*) + 132
6   libSystem.B.dylib             	0x95ecc095 _pthread_start + 321
7   libSystem.B.dylib             	0x95ecbf52 thread_start + 34

[Chris Bergmann]

MacOSX CrashReporter log

[Chris Bergmann]

Issue appears reproducible. After restarting the guest, incoming network traffic (a SVN HTTPS connection which is to be forwarded to another VM) causes immediate termination of the VM. It's worth noting that I have added a port-forward for the NAT interface as described in the user manual.

[Chris Bergmann]

Workaround: switch to bridged networking.

[Robin Green]

I am also seeing this crash. However, port-forwarding did work previously for me, with this version of VirtualBox - but previously I was only forwarding UDP, which presumably would not cause a crash in a function called tcp_input - so I guess that explains it.

[Robin Green]

Also, I now have two guest NICs, and previously, when it wasn't crashing, I only had one NIC. Don't know if that is relevant.

[vasily Levchenko]

was fixed in 2.2.4. Fill free to reopen if problem still persist