Re: NM roaming



Thank you very much for the detailed explanation.

Will look into wpa_supplicant and mac80211 then.

BR.


2012/12/10 Dan Williams <dcbw redhat com>
On Mon, 2012-12-10 at 14:25 -0200, Franco Miceli wrote:
> Hi,
>
>
> I am working on a project that aims to assess how wireless roaming
> works.
>
>
> I am currently working with nm 0.8 and am lost trying to find out
> where in the code the roaming happens.
>
>
> Could anyone point me at some direction as to where to start?

As others have said, it can happen at different levels.  NM actually
never makes a decision to roam, that's all left to wpa_supplicant and
the kernel.  There are two basic cases:

1) "thick" firmware that contains much of the 802.11 stack will almost
always control roaming in the firmware; the firmware monitors current
channel conditions and scan results and determines when to jump to a new
access point, does so, then notifies wpa_supplicant of the change.  An
example driver here is 'orinoco'.

2) "thin" firmware leaves most of the 802.11 functionality to the
kernel, where it's implemented in the 'mac80211' kernel module.  Here,
the kernels' mac80211 stack just passes signal and channel information
up to wpa_supplicant, which may decided to roam based on that
information.  Example drivers are ath5k, ath9k, iwlwifi, and more.

BTW, NM 0.8 is pretty old, 0.9 has been out for a long time.  But the
roaming procedures are not different between the two, it's still all
controlled by the kernel and wpa_supplicant.

Dan




--
Ing. Franco Miceli
CITS - Plan Ceibal - Investigación & Desarrollo
Av. Italia 6201 - Montevideo, Uruguay
CP: 11500
Tel: (598 2) 601 5773 int.: 2227


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]