Rate mismatch with Devicescape IEEE 802.11 stack and reported encryption methodes in beacon

[Olivier H]

Sorry my first message wasn't send cause of the size limit.

2006/8/23, Jouni Malinen <jkmaline at cc.hut.fi>:

> On Tue, Aug 22, 2006 at 09:18:13PM +0200, Olivier H wrote:
>
> > I'm currently testing the rt2x00 driver (http://rt2x00.serialmonkey.com)
> > wich use the devicescape stack in master mode with hostapd.
> > I had a few issues regarding the rates (always slow in G mode)
> >
> > The problem is that in driver_devicescape.c/i802_set_rate_sets() the
> mode is
> > set by HOSTAPD_MODE_IEEE80211*
> > And then back in the stack in
> > ieee80211/ieee80211_ioctl.c/ieee80211_ioctl_set_rate_sets() the mode is
> > compared to MODE_IEEE80211*
>
> Hmm.. that should not be the case. hostapd is supposed to use HOSTAPD_*
> internally and convert it into MODE_* whenever interacting with the
> driver code.
>
> > And both are different:
>
> Yes, by design.
>
> > So when i run hostapd in G mode, the stack will assume B mode.
> > So the easiest way to solve this is to set the hostapd values same as in
> the
> > stack:
> >
> > typedef enum {
> >    HOSTAPD_MODE_IEEE80211B = 1,
> >    HOSTAPD_MODE_IEEE80211G = 3,
> >    HOSTAPD_MODE_IEEE80211A = 0,
> >    NUM_HOSTAPD_MODES = 3
> > } hostapd_hw_mode;
> >
> > What do you think of that? Can it be include in next release?
>
> No, that is not acceptable. The correct fix would be to change
> i802_set_rate_sets to convert mode parameter to use proper enum when
> talking to the driver.


I tough so, but i just picked the easiest and fastest way to make sure this
was really  the way to solve the problem.
modes weren't translated, so i made a patch to do it. se attached file

> Then an another issue i get with my Intel(r) PROSet/Wireless Software
> > v10.1. I
> > can't sucessfully associate to the AP when WPA is enabled.
> > It was working with the v9.
>
> Did I understand this correctly: You were able to use Intel client
> software version 9 with the AP that is running hostapd, but when
> changing to Intel software version 10.1 (and without modifying AP) the
> connection does not work?


Exactly. the v10.1 fail on the wpa handshaking (don't remember exactly at
which stage, first or second maybe)
The v9 correctly handle the situation and work great in WPA


> The problem is that the encryption for the network is reported as
> > "WPA-entreprise" (sorry in french, i don't know if the english name is
> > different)
> > But In fact it's "WPA-Personel", no username or whatever needed, just
> the
> > passphrase or PSK.
> > So the new version only want to be in the mode that the network told
> him.
> >
> > Is this a knowed issue? or can you tell me where this information is
> send so
> > i can see what i can do.
>
> I haven't heard this before. Could you please send your hostapd
> configuration file? If you can easily capture wireless frames, it would
> also be interesting to see what exactly is in the Beacon and Probe
> Response frames sent by the AP.
>
>
Apparently, the encryption method of the AP is in the beacon, but when i
want to configure the intel client it also start authenticating to so  the
AP to see what info are needed (auth and assoc_req).
I can't capture with my windows laptop, I'll try to do it directly on the
machine running hostapd.

Enclose, my configuration file you asked and the patch.

Olivier
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.shmoo.com/pipermail/hostap/attachments/20060823/2557c296/attachment.htm 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: patch_hostapd_rate_sets.mode.diff
Type: application/octet-stream
Size: 837 bytes
Desc: not available
Url : http://lists.shmoo.com/pipermail/hostap/attachments/20060823/2557c296/attachment.obj 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hostapd.conf
Type: application/octet-stream
Size: 527 bytes
Desc: not available
Url : http://lists.shmoo.com/pipermail/hostap/attachments/20060823/2557c296/attachment-0001.obj