abrt 1.1.1 detected a crash. architecture: x86_64 Attached file: backtrace cmdline: /usr/sbin/modem-manager comment: Typically, the first two or three connections drop with a crash in modemmanager after a few seconds/minutes. After the 2nd or 3rd quick-drop, the connection establishes and seems to stay pretty stable. component: ModemManager executable: /usr/sbin/modem-manager global_uuid: 5d6da46489da8ca23cf51a8e3f006e6c867e1e2a kernel: 2.6.33.5-112.fc13.x86_64 package: ModemManager-0.3-13.git20100507.fc13 rating: 3 reason: Process /usr/sbin/modem-manager was killed by signal 11 (SIGSEGV) release: Fedora release 13 (Goddard) How to reproduce ----- 1. Connect to Verizon via Novatel U760 usb device 2. modemmanager quickly segfaults
Created attachment 424143 [details] File: backtrace
Could you do the following for me (all as root)? 1) make sure the U760 is unplugged 2) service NetworkManager stop 3) killall -TERM modem-manager 4) gdb /usr/sbin/modem-manager 5) type "run --debug" at the (gdb) prompt 6) plug your modem in 7) wait for it to show up in the NM menu 8) connect it 9) when MM crashes, gdb will drop you back to the "(gdb)" prompt, type "t a a bt" here to get a backtrace (stands for "thread apply all backtrace") 10) you may have to hit return a few times when gdb asks to get it to print all the backtrace 11) attach everything (the MM debug output before the crash, and the backtrace) to this bug At the (gdb) prompt you can type 'quit' to exit. Thanks!
FWIW, I pulled down the ModemManager SRPM from Koji and updated it w/ the 6/22/2010 code from git (ModemManager-0.4-1.git20100622.fc13.x86_64 now) and mm is no longer bombing out. Now, the connection does drop periodically but that seems to be from PPP (and it may just be the 3G connection dropping sporadically) but with that connection definition set to auto-connect, NM winds up restarting after MM cycles through and re-discovers the ttyUSBx devices, etc. If needbe, I can downgrade back to the stock F13 ModemManager and capture the backtraces that you need.
Ok, thanks for the update. I'll close this, it could have been the one of a few fixes to MM upstream that fixed your issue.