NM falsely thinks wlan is off if eth is turned off

Bug #391154 reported by Vany
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Network Manager Applet
Invalid
Medium
network-manager (Ubuntu)
Invalid
Undecided
Unassigned

Bug Description

when eth1(ipw2200) is turned of by button nm-applet thinks, that wlan0 (pcmcia atheros) is turned off to.
But when i connect with wpa_supplicant manually - card works.

(second card is a workarround for bug in wpa_supplicant and ipw2200)

ProblemType: Bug
Architecture: i386
DistroRelease: Ubuntu 9.04
ExecutablePath: /usr/bin/nm-applet
IfupdownConfig:
 auto lo
 iface lo inet loopback
Package: network-manager-gnome 0.7.1~rc4.1-0ubuntu2
ProcEnviron:
 SHELL=/bin/bash
 PATH=(custom, user)
 LANG=C
SourcePackage: network-manager-applet
Uname: Linux 2.6.28-13-generic i686

Revision history for this message
Vany (ivan-serezhkin) wrote :
Revision history for this message
Tony Espy (awe) wrote :

If you look at the attached syslog, you'll find that the following relevant log messages:

Jun 23 18:31:48 vfs NetworkManager: <info> Wireless now disabled by radio killswitch
Jun 23 18:31:48 vfs NetworkManager: <info> (eth1): device state change: 3 -> 2
Jun 23 18:31:48 vfs NetworkManager: <info> (eth1): deactivating device (reason: 0).
Jun 23 18:31:48 vfs NetworkManager: <info> (eth1): taking down device.
Jun 23 18:31:48 vfs NetworkManager: <info> (wlan0): device state change: 3 -> 2
Jun 23 18:31:48 vfs NetworkManager: <info> (wlan0): deactivating device (reason: 0).
Jun 23 18:31:48 vfs NetworkManager: <info> (wlan0): taking down device.

Changed in network-manager-applet (Ubuntu):
status: New → Incomplete
Changed in network-manager-applet:
status: Unknown → New
Revision history for this message
Tony Espy (awe) wrote :

Looking at the source code for version network-manager-0.7.1~rc4.1.cf199a964, found in jaunty:

The first message is logged by nm-manager.c:hal_manager_rfkill_changed_cb().

This function is registered as a callback function for the signal "rfkill_changed". The signal in turn is fired by the function nm-hal-manager.c:killswitch_getpower_done(), which is registered when the dbus_g_proxy_begin_call() is made to send the GetPower DBUS request to Hal for a single device's killswitch state. The problem is, NetworkManager only keeps a single per HalManager instance of the rfkill state. See nm-hal-manager.c:NMHalManagerPrivate. So as soon as one active kill switch is found, "rfkill_changed" is fired, which results in hal_manager_rfkill_changed_cb disabling wireless.

Game over. I verified that this code still exists as is in 0.7.1 final.

Changed in network-manager-applet (Ubuntu):
status: Incomplete → Confirmed
Changed in network-manager-applet:
importance: Unknown → Medium
status: New → Confirmed
Changed in network-manager-applet:
status: Confirmed → Invalid
Revision history for this message
Thomas Hood (jdthood) wrote :

@Tony: Has this bug been fixed yet?

affects: network-manager-applet (Ubuntu) → network-manager (Ubuntu)
Thomas Hood (jdthood)
summary: - nm-applet can't detect wireless cards power separately
+ NM falsely thinks wlan is off if eth is turned off
Revision history for this message
Bryan Quigley (bryanquigley) wrote :

No response to question from 2012, closing.

Changed in network-manager (Ubuntu):
status: Confirmed → Invalid
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.