quaker,

> 
> wladm_cmd_scan(), wladm_get_bssid(), wladm_get_essid()
> are not really necessary because they seem to be implementable
> by extending the existing interfaces.
> 
> for wladm_cmd_scan(), you can just change wladm_scan() to
> check if func is NULL and only do the scan if it is.
> 
> | EXPLAIN
> | wladm_cmd_scan() just sends WL_SCAN command to driver,
> | and return immediately, don't wait to get the scan results;
> | It is different from the wladm_scan().
> | To be clear, I rename wladm_cmd_scan to dladm_wlan_scan_only().
>

why can't you just change dladm_wlan_scan() to behave as you
described? e.g. if func is NULL, just send WL_SCAN and return.

 
> for get_bssid,essid, you should just use wladm_get_link_attr()
> and extend the wladm_link_attr_t with the fields la_bssid and
> la_essid. these represent the local bssid/essid. corresponding
> WLADM_LINK_ATTR_ESSID/BSSID flags need to be added as well.
> 
> | EXPLAIN
> | The result of wladm_get_link_attr() is only available when link is
> | CONNECTED. If not connected, it will return nothing.
> | Wpa service need to get ESSID, BSSID when doing a connection/key
> | negotiation, so need to keep these two APIs.
>

you are describing the current implementation. I don't see why
you can't change dladm_wlan_get_linkattr() to retrieve bssid and
essid on a non-connected link. as I mentioned above, you should do
this by adding two fields la_bssid and la_essid to the structure
dladm_wlan_linkattr_t.

 
> also, I don't like the exposure of the wl_* data structures to
> clients of libwladm. for each one of these structures, you
> should create a similar one called wladm_xxx; and in your new
> wladm routines, you translate these types to your wl_xxx types.
> the reason for doing this is to keep libwladm clients insulated
> from ioctl interface changes.
> 
> | ACCEPT
> | use void * replacing spsecific data structure.
> 

as discussed offline, this is not right because the client still
has to include net/wpa.h.

eric
 
 
This message posted from opensolaris.org
_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to