Michael Mann
changed
bug 1419
Comment # 5
on bug 1419
from Michael Mann
I could not reproduce this on the current trunk (rev 46640), running WinXP. I
thought about adding an "extra layer" of protection in
signal_pipe_check_running() (in dumpcap.c) that would involve checking the
(Wireshark) process ID to see if its still running, but I haven't been able to
duplicate a Wireshark crash that keeps dumpcap around in order to test the
"extra layer" works.
Originally I wanted to replace the current "signaling" (since that's all the
named pipe is used for in Windows) with just checking the process ID, but that
doesn't allow for a "graceful" exit of dumpcap on a capture restart.
I didn't run Wireshark until it ran out of memory, but I did add an intentional
NULL dereference to a dissector that was executed during a live capture.
Wireshark crashed, but it took dumpcap with it.
Tempted to mark this as RESOLVED.
You are receiving this mail because:
- You are the assignee for the bug.
- You are watching all bug changes.