Right having closed bug 289982, I'm moving attention to this. Something in the sequence of events as NM tries to bring up both eth0 and wlan0 at boot/login is messing things up.
Attached is a syslog that hopefully documents the problem. I recompiled the kernel to enable debug printks in drivers/net/ps3_gelic_*.c so there's plenty of detail about what the gelic_wireless driver is doing.
Essentially this log shows
1. Boot - ending with eth0 being brought up and plenty of wlan scanning
2. Me logging in (search for pam_sm_authenticate)
3. NM trying to connect to my wifi network (search for "starting connection 'Auto munckfish'")
4. Then what seems to be repeated failures to connect/associate delimited by ...
Right having closed bug 289982, I'm moving attention to this. Something in the sequence of events as NM tries to bring up both eth0 and wlan0 at boot/login is messing things up.
Attached is a syslog that hopefully documents the problem. I recompiled the kernel to enable debug printks in drivers/ net/ps3_ gelic_* .c so there's plenty of detail about what the gelic_wireless driver is doing.
Essentially this log shows
1. Boot - ending with eth0 being brought up and plenty of wlan scanning authenticate)
2. Me logging in (search for pam_sm_
3. NM trying to connect to my wifi network (search for "starting connection 'Auto munckfish'")
4. Then what seems to be repeated failures to connect/associate delimited by ...
Feb 23 13:34:48 ps3 NetworkManager: <info> (wlan0): supplicant connection state: associating -> disconnected
Feb 23 13:34:48 ps3 NetworkManager: <info> (wlan0): supplicant connection state: disconnected -> scanning
and sometimes
Feb 23 13:34:55 ps3 NetworkManager: <info> wlan0: link timed out.
5. Then at some point NM applet pops up the pasword dialog (search for "asking for new secrets"), and we run through the same loop all over again.