ohNet crash with Parallels installed

[simoncn]

I've had a report of an ohNet crash on Mac OS X with Parallels installed. I'm attaching the Mac OS X crash dump. This shows that OsNetworkSocketMulticastAddMembership returned an error code, and Os::NetworkSocketMulticastAddMembership threw an exception to report this. The exception wasn't caught by ohNet code, which caused the C++ library to call abort().

My guess is that ohNet was trying to set up an SSDP multicast listener for one of the additional network adapters that Parallels creates. The error from OsNetworkSocketMulticastAddMembership is probably correct, but it shouldn't cause the process to abort.

[simoncn]

(24-09-2012 10:30 AM)simoncn Wrote:  I've had a report of an ohNet crash on Mac OS X with Parallels installed. I'm attaching the Mac OS X crash dump. This shows that OsNetworkSocketMulticastAddMembership returned an error code, and Os::NetworkSocketMulticastAddMembership threw an exception to report this. The exception wasn't caught by ohNet code, which caused the C++ library to call abort().

My guess is that ohNet was trying to set up an SSDP multicast listener for one of the additional network adapters that Parallels creates. The error from OsNetworkSocketMulticastAddMembership is probably correct, but it shouldn't cause the process to abort.

I've had a further update from the user that this crash isn't caused by an adapter created by Parallels, but an adapter created by a program that tries to connect to a VPN server.

[simonc]

It sounds like the best thing to do in this case would be to simply catch then ignore the exception. I'll do this during device construction only. Exceptions that happen on adapter changes (which happens post-hibernation) will still need to be thrown. This latter class of exception is already handled so shouldn't cause you any problems.

I'll get this committed in the next couple of days.

[simoncn]

(26-09-2012 12:14 PM)simonc Wrote:  It sounds like the best thing to do in this case would be to simply catch then ignore the exception. I'll do this during device construction only. Exceptions that happen on adapter changes (which happens post-hibernation) will still need to be thrown. This latter class of exception is already handled so shouldn't cause you any problems.

I'll get this committed in the next couple of days.

Thanks very much! I'm putting together a new release of MinimServer, and it would be good if I can include this fix.

[simonc]

(26-09-2012 12:14 PM)simonc Wrote:  I'll get this committed in the next couple of days.

I've made some local changes that should fix this. They should make it onto github this evening.