Wired

Jouni Malinen jkmaline at cc.hut.fi
Thu Jun 17 00:59:51 EDT 2004


On Wed, Jun 09, 2004 at 08:05:02AM +0200, Gunter Burchardt wrote:

> My first mail seems to be to huge for this mailing list.

Which is a good sign that separating the patch into couple of shorter
files might have been helpful.. ;-)

> Next patch is ready. It compiles/starts up/authenticates a client via
> 1x/deauthenticates it and shutdown. Im not able to test WPA. I had to
> change some calls in wpa.c for the new void *priv pointer. driver.c is
> nearly clean from hapd. hapd is only needed by debug-messages. We
> should add some defines for debugging in driver specific stuff without
> hapd in it!

Thanks! I merged most of it into CVS tree. Some cleanup was needed for
indentation and trailing white space etc., but other than that, I did
not too much changes. However, I dropped this part for now:

>  - function pointer for driver specific sta dump
>    (hostapd_dump_sta_state), moved this function to driver.c

driver.c should be low-level code and processing struct sta_info in it
does not sound a good direction. Actually, I ended up removing
struct sta_info completely from driver.c.

struct sta_info may need to be handled somewhat separately since
different driver interfaces might either share the same structure or
use their own (or have some kind of sub-structure within that
structure etc.).

Some of these changes removed some hapd uses from driver.c, but yes, the
remaining debug cases would benefit from some cleanup.

> Ok this patch has 1017 lines. I hope this is ok.

It was a bit long and partly because of this, it took me some time to
go through everything. In other words, smaller patch files would be
prefered.

> Next step is to move hapd->sock to hapd->driver.data->sock. Therefor i
> move ieee802_1x_send_data() to a new file send.c . I will also add a
> function to send 802.11 management frames to send.c . driver_info get 
> 2 new function pointer for these functions.

OK. Please replace struct sta_info with u8 *addr (i.e., only sta->addr
is needed) when moving ieee802_1x_send_data(). In addition, that could
actually be renamed to ieee802_1x_send_eapol() (and something shorter,
like send_eapol for function pointer).

Btw, please use another thread and more descriptive subject line for
next patch. I have a bit too many messages in NEW state waiting and I
might miss something if the subject line does not describe the contents.
Including PATCH on the line could also be helpful for the same reason.

-- 
Jouni Malinen                                            PGP id EFC895FA



More information about the HostAP mailing list