Description of problem: NM dies when wrong WEP key entered in the Passphrase dialog Version-Release number of selected component (if applicable): CVS HEAD 1.436 How reproducible: Every time Steps to Reproduce: * Login as user with no NM gconf or keyring data stored, NM service running, nm-applet running, AP broadcasting its essid * Click on progress bar for AP * Note the gconf now contains the essid and timestamp * type wrong HEX WEP key in Passphrase dialog and enter the default keyring password when prompted. OK * NM service dies but PID file still exists. nm-applet still running. * gconf now also contains the key_type * restart the NM service * enter correct HEX WEP key. * connection made and WEP key updated in keyring. Actual results: NM dies. Expected results: After a failure to connect, NM should display an error message telling the user that the WEP key is wrong. Additional info:
Created attachment 117252 [details] patch for NetworkManagerDevice.c The patch fixes the bug reported above. Under the conditions described, the allowed_ap is NULL at the point when the auth_meth for the allowed_ap is first updated. This update can be avoided using an if statement.
Ok, updated and committed to CVS. Thanks for the ping.