NM does not memorize network activation state on suspend/resume or restart

Bug #387345 reported by pittipatti
48
This bug affects 9 people
Affects Status Importance Assigned to Milestone
Linux
Fix Released
Medium
NetworkManager
Fix Released
Medium
network-manager (Ubuntu)
Fix Released
Wishlist
Unassigned

Bug Description

Binary package hint: network-manager

After resuming from suspend (or restarting) I would expect the system to be in the same networking state as as before the suspend.
In particular when networking was disabled before suspending, I would expect it to still be disabled after resuming.

But, networking allways gets enabled after resuming or restarting the system.

CVE References

pittipatti (pittipatti)
summary: - NM does not memorize network state on suspend/resume or restart
+ NM does not memorize network activation state on suspend/resume or
+ restart
Revision history for this message
Alexander Sack (asac) wrote :

i think restarting is ok to forget the state. howver, suspend/resume sounds reasonable.

Changed in network-manager (Ubuntu):
importance: Undecided → Wishlist
status: New → Triaged
Changed in network-manager:
status: Unknown → Confirmed
Revision history for this message
Arnaud Faucher (arnaud-faucher) wrote :

In my case (Acer laptop on Karmic), the lost state in suspend/resume was the result of a bug caused by a recent rewrite of rfkill in the kernel (still present in 2.6.31-rc4).

The patch located at http://www.kernel.org/pub/linux/kernel/people/linville/wireless-2.6/0018-acer-wmi-fix-rfkill-conversion.patch solved this. Therefore, it may not be a problem related to network-manager, but to the kernel itself.

Revision history for this message
Arnaud Faucher (arnaud-faucher) wrote :

Bugzilla kernel bug #13659 for HP machines is closely related to the one being observed on my Acer.

Changed in linux:
importance: Undecided → Unknown
status: New → Unknown
Changed in linux:
status: Unknown → Fix Released
Revision history for this message
Arnaud Faucher (arnaud-faucher) wrote :

This bug is fixed for me with the latest version of the linux kernel (package version 2.6.31-5.24).

Changed in network-manager (Ubuntu):
status: Triaged → Fix Released
Changed in network-manager:
status: Confirmed → In Progress
Changed in network-manager:
status: In Progress → Fix Released
Revision history for this message
fugounashi (fugounashi+launchpad) wrote :

this bug (disabled wifi enabled on resume) is still present in network-manager 0.8-0ubuntu3 / lucid 10.04

a work-around is to use rfkill instead, which is probably what network-manager should be doing anyway...

papukaija (papukaija)
tags: added: lucid
Revision history for this message
papukaija (papukaija) wrote :

In response to comment 5: I think that Arnaud did a mistake when he marked this bug as fixed in Ubuntu's NM. The fix will probably go to NM 0.8.1

Changed in network-manager (Ubuntu):
status: Fix Released → Confirmed
tags: added: not-a-wishlist
Changed in network-manager:
importance: Unknown → Medium
Revision history for this message
Launchpad Janitor (janitor) wrote :

This bug was fixed in the package network-manager - 0.8.2-0ubuntu1

---------------
network-manager (0.8.2-0ubuntu1) natty; urgency=low

  * new upstream release 0.8.2
    - update timestamp in active system connections every 5 mins
    - ignore authorization for sleep/wake requests (but restrict to root)
    - add timestamp to default auto wired connections (LP: #447067)
    - don't crash if for some reason HW addresses aren't available (bgo #630534)
    - ensure Enabled state is preserved regardless of rfkill (LP: #387345)
    - add dhcp4-change and dhcp6-change events (bgo #563654)
    - validate Enable/Disable WiFi and WWAN requests
    - handle VPN service installation/removal dynamically (LP: #458595)
    - let plugins forbid VPN connections from getting the default route
      (LP: #598357)
    - work around dbus-glib property access bug (CVE-2010-1172)
    - ensure dhcp_manager exists before trying to unref it (bgo #626610)
  * drop dhcp3 support patch, it was applied upstream.
    - remove debian/patches/99-support-dhcp3.patch
    - update debian/patches/series
  * drop cherry-picked patches to /etc/hosts handling, their changes are
    included in the source tarball:
    - remove debian/patches/lp659872_dont_overwite_hosts_aliases_ee9ce60.patch
    - remove debian/patches/lp632896_cleanup_etc_hosts.patch
    - update debian/patches/series
  * Properly document workflow for working on NM packaging.
    - add debian/README.source
  * Slight change to GET_SOURCE script. Looks like we were looking for the
    "orig" keyword in the wrong variable.
    - update debian/rules
  * Fix typo in README.source
    - update debian/README.source
  * network-manager Breaks: dhcp3-client (<< 4.1.1-P1-11) to provide seemless
    upgrades to isc-dhcp 4 and avoid unnecessary compatibility issues
  * switch Build-Depends from dhcp3-client to isc-dhcp-client >= 4.1.1 to make
    sure we don't build with an old version of DHCP.
  * libnm-glib2 also Breaks: dhcp3-client (<< 4.1.1-P1-11)
    - debian/control
  * libnm-util1 Breaks: dhcp3-client (<< 4.1.1-P1-11) as well
    - debian/control
  * Patch various files to support building with kernel headers 2.6.36 and
    newer (otherwise it conflicts with libc headers for interface structures)
    - debian/patches/do-not-include-netif.patch
    - debian/patches/series
 -- Mathieu Trudel-Lapierre <email address hidden> Tue, 09 Nov 2010 13:01:48 -0500

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

Duplicates of this bug

Other bug subscribers

Remote bug watches

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