https://bugzilla.novell.com/show_bug.cgi?id=850920 https://bugzilla.novell.com/show_bug.cgi?id=850920#c2 Petr Tesařík <ptesarik@suse.com> changed: What |Removed |Added ---------------------------------------------------------------------------- AssignedTo|bnc-team-gnome@forge.provo. |jengelh@inai.de |novell.com | --- Comment #2 from Petr Tesařík <ptesarik@suse.com> 2013-11-19 10:22:45 UTC --- I tried to debug this issue, and it seems that Ekiga runs an endless registration loop under certain circumstances, since each subscriber() initializes a PThread object, which in turn creates a pair of filedescriptors with pipe(). At some point, the process runs out of file descriptors... The loop is apparently created "indirectly" by looping the value of compat_mode in Opal::Account::handle_registration_event() over: SIPRegister::e_FullyCompliant SIPRegister::e_CannotRegisterMultipleContacts SIPRegister::e_CannotRegisterPrivateContacts SIPRegister::e_HasApplicationLayerGateway and back to SIPRegister::e_FullyCompliant The strace shows that each pipe() + clone() is preceded by a timed out poll() syscall with a timeout of 0. I guess that the zero timeout is used, because the seconds member of the corresponding message is set to 0. I'm unable to follow the code path to determine what went wrong, and why the algorithm goes wile only after some time, but I suspect this bug is not specific to openSUSE. Should I report it there, or can you do it for me? -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.