PATCH: send_mgmt_frame in driver.c

[Gunter Burchardt]

> Agreed. However, I want to make the new 0.2 stable branch before these
> changes. After this, we can continue doing changes in the development
> branch (CVS trunk) and I'm more open to experimental code being added.
> Though, I would rather not add a new directory to make the build system
> more complex at this point.

Ok. No patch before new development branch!

> Yes, indeed. hostapd_ioctl() and hostap_ioctl_*() functions should
> disappear from the exported interface and only used within
> (hostap_)driver.c. Luckily, these have quite limited use outside
> driver.c, so this does not require that much changes.

There are only 2 places hostapd_ioctl() is used outside from driver.c .
In sta_info.c it is used to get the inactivity time. I will make a new
function pointer like get_inactivity_time() (or shorter) for this
issue. I will send you this short patch soon.

The second point is in ieee802_11.c . I think this issue is more
complex according to make a good interface for handling of new
stations. There must be a function like add_sta() in driver.c for the
hostap_ioctl() . But its a bit confusing for me when to call it. It
is possible to merge this new function add_sta() in
hostapd_new_assoc_sta() ? It is possible that a sta can associate twice
with an ap? What should add_sta() do in if it is called with a already
sta? In which case a assoc isn't a new station (is there a reassoc case
defined in 802.11?) ?

regards
gunter