I've run gdb several times on the native program (gdb --args marionnet.native); the signal (it's a SIGTERM) always arrived during a select call, when I tested (but not always from a death_monitor function, as in the attached log). This is not very significant anyway, since select can be used as a high-resolution sleep.
sysdeps/unix/syscall-template.S is part of the libc sources. It would be nice to start UML under a debugger, to see the point where it sends a signal to its parent... ** Attachment added: "gdb session; of course I trigger the bug by starting up a machine with a 64bit kernel" https://bugs.launchpad.net/marionnet/+bug/734785/+attachment/1907714/+files/session -- You received this bug notification because you are a member of marionnet-dev, which is subscribed to marionnet. https://bugs.launchpad.net/bugs/734785 Title: VM kernel crash may crash marionnet itself Status in Marionnet - A virtual network laboratory: New Bug description: When playing with new VM kernels (compiled with ARCH=um), we made a 64 bits kernel... we tryied to run it with a VM in 32 bits... silly idea ! Of course the VM crashed but surprisingly marionnet also crashed in less than a second. The debug option only showed "terminated", has if marionnet have been killed by some regular signal. (Marionnet version 0.90.6). _______________________________________________ Mailing list: https://launchpad.net/~marionnet-dev Post to : [email protected] Unsubscribe : https://launchpad.net/~marionnet-dev More help : https://help.launchpad.net/ListHelp
