NM broken after suspend

[Dennis Heinson]

Binary package hint: network-manager

Looking at previous bug reports, this issue should have been fixed. However, upon restart from suspend (standby) mode on my Thinkpad T41 with Atheros Wi-Fi, NM does not detect any devices.

Ifconfig/iwconfig show them, however.
I use a WPA2-Network.
The wired network is displayed as grey, but I do not have a cable connected either.

Please send me logging instructions or point me thereto, and I will collect the necessary data.

[Dennis Heinson (dheinson)]

It's in Hardy, daily build from Feb-08, and with all update packages installed.

[MsTiFtS (mstifts)]

Same misbehavior here, when I suspend to RAM, NM won't be able to connect to my wireless networks afterwards. If I do it all manually, it works, but doesn't show my wireless card (ipw3945) any more.
Killing NM and the dispatcher and reloading them solves a part of the problem, but sometimes it won't be able to connect to the network then. (LED always blinking around but not getting solid on) Unloading and reloading the kernel module and the userspace daemon while before relaunching NM solves that.

Today at work, after a suspend to disk, NM started to catch and rethrow a SIGSEGV in an infinite loop, eating up all my CPU.
Excerpt of the output of strace NetworkManager --no-daemon:
writev(2, [{"NetworkManager: nm_dbus_escape_o"..., 102}, {"\n", 1}], 2
NetworkManager: nm_dbus_escape_object_path: assertion `g_utf8_validate (utf8_string, -1, NULL)' failed
) = 103
send(3, "<11>Feb 11 10:00:15 NetworkManag"..., 122, MSG_NOSIGNAL) = 122
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
time(NULL) = 1202720415
stat64("/etc/localtime", {st_mode=S_IFREG|0644, st_size=2295, ...}) = 0
writev(2, [{"NetworkManager: <WARN> nm_signa"..., 88}], 1
NetworkManager: <WARN> nm_signal_handler(): Caught signal 11. Generating backtrace...
) = 88
send(3, "<12>Feb 11 10:00:15 NetworkManag"..., 108, MSG_NOSIGNAL) = 108
stat64("/usr/lib/network-manager/nm-crash-logger", {st_mode=S_IFREG|0755, st_size=4276, ...}) = 0
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0xb7b646f8) = 8824
waitpid(8824, NetworkManager: ******************* START **********************************
NetworkManager: (no debugging symbols found)
ptrace: Operation not permitted.
/home/micha/8812: No such file or directory.
NetworkManager: Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1".
NetworkManager: ******************* END **********************************
[{WIFEXITED(s) && WEXITSTATUS(s) == 0}], 0) = 8824
--- SIGCHLD (Child exited) @ 0 (0) ---
rt_sigaction(SIGSEGV, NULL, NULL, 8) = 0
tgkill(8812, 8812, SIGSEGV) = 0
sigreturn() = ? (mask now [])
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
sigreturn() = ? (mask now [])
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
sigreturn() = ? (mask now [])
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
sigreturn() = ? (mask now [])
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
sigreturn() = ? (mask now [])
--- SIGSEGV (Segmentation fault) @ 0 (0) ---
etc. etc. etc.

Syslog doesn't show anything. The SIGSEGVing behavior was reproducible until I rebooted.
In Gutsy that happened on every NM restart and just went away with a reboot, in Hardy this was the first time.
I don't yet know whether this occurred because of the suspend to disk or because of the other networks at work, I'll try to reproduce that again later. When it happens again, I'll also have a look whether restarting DBUS fixes it.