Lizhong Li wrote: > Hi, > > I'd like to bring up this topic since the WiFi team have a strong > interest to NWAM project and testing, and the NWAM QE and developers > also have some questions to query. > > As I knew, perhaps the main focus is how the nwam daemon works with WiFi > drivers smoothly and how to test it, the following link may be helpful. > > http://opensolaris.org/os/project/nwam/p1spec/daemon/ > On this page, you may want to take extra care of this section "7.4 nwamd > as Event Dispatcher", > and this section ? "7.5 nwamd as Policy Engine". > > Now nwamd can deal with such events: > > RTM_NEWADDR, New address for IP interface, Routing socket > RTM_DELADDR, Address removed, Routing socket > DL_NOTE_LINK_UP, Wired link is up, Wireless link is connected to AP, > DLPI notification > DL_NOTE_LINK_DOWN, Wired link is down, Wireless link is disconnected > from AP, DLPI notification > EC_DEV_ADD, NIC hotplug inserted, sysevent > EC_DEV_REMOVE, NIC hotplug removed, sysevent > WLAN scan done, dladm_wlan_scan(), nwamd > WLAN connection state, dladm_wlan_getlinkattr(), nwamd > more information needed, no recognized WLAN available,nwamd > more information needed, WEP or WPA key needed,nwamd > > For testing, we are developing a test "fake" NIC driver which is able to > pose as a wired or wireless NIC, this is supposed to simulate link > up/down, device plug/unplug, and such events, without require any > physical infrastructure support. Cecilia has a question about it, > whether the "fake" driver could simulate the real behavior like real > wifi drivers since the different wifi drivers have the different > policies to deal with some events, e.g., they'll wait various time to > retry when they lose the connection to AP, then the 'fake' driver how to > handle it ?
About this part, my question was: - Under different situations, different WiFi drivers behaviors depend on themselves. I was curious about how one fake driver simulates different situations/environments and different cards/drivers. When we develop a new driver, we used to test it with gldv3/WiFi test suite under different situations many times to ensure it works. So, it seems necessary for pseudo driver to have an algorithm to simulate such difference. -Cecilia > And could this "fake driver" simulate all the events which the nic > deriver have ? >
