[PATCH 0/2] Mesh rejoin workaround
Jason Mobarak
jam at cozybit.com
Thu Mar 6 20:31:00 EST 2014
Hello,
This is an RFC on our mesh rejoin workaround. We have a test which requires
that a mesh node be able to rejoin the network after going "out of range", then
coming back into range.
We see the following behavior:
(1) Mesh nodes 1, 2, and 3 join a mesh
(2) The firmware sends QoS NULL frames to other STAs, presumably tracking
whether nodes are still alive
(3) Mesh node 1 goes out of range
(4) QoS NULL frame is not ACKed, so wcn36xx gets a notification that the
firmware has removed the STA
(5) Mesh node 1 comes back into range
(6) Mesh node 1 is no longer able to send data to mesh
If after #4, instead of allowing the STA to be removed, we immediately re-add
the STA, the node is able communicate on the mesh once it comes within range
again.
Obviously, this is a non-ideal fix... Do you all have any ideas on how to
better workaround the issue of the firmware removing the STA?
Our test suite can be seen accessed here:
- https://github.com/cozybit/wcn36xx-tests
Specifcally I'm running this test:
- https://github.com/cozybit/wcn36xx-tests/blob/master/scripts/run_multi_rejoin
Chun-Yeow Yeoh (1):
wcn36xx: fix the mesh STA for rejoin
Jason Mobarak (1):
Revert "wcn36xx: Pad TIM PVM if needed"
Chun-Yeow Yeoh (1):
wcn36xx: fix the mesh STA for rejoin
Jason Mobarak (1):
Revert "wcn36xx: Pad TIM PVM if needed"
hal.h | 3 ---
main.c | 45 +++++++++++++++++++++++++++++++---
smd.c | 84 +++++++++++++++++++++++++++++++++++++--------------------------
smd.h | 9 ++++++-
wcn36xx.h | 6 +++++
5 files changed, 105 insertions(+), 42 deletions(-)
--
1.9.0
More information about the wcn36xx
mailing list