Date: Tue, 8 Dec 2009 11:42:46 +0100 From: Andre Albsmeier <Andre.Albsmeier@siemens.com> To: Rui Paulo <rpaulo@FreeBSD.org> Cc: Andre Albsmeier <Andre.Albsmeier@siemens.com>, freebsd-mobile@FreeBSD.org Subject: Re: iwi: Possibly wrong interpretation of beacon->number in if_iwi.c? Message-ID: <20091208104246.GA18375@curry.mchp.siemens.de> In-Reply-To: <9A2736AF-321C-4F39-9AE5-BD436F7C2808@FreeBSD.org> References: <20091207163607.GA15625@curry.mchp.siemens.de> <9A2736AF-321C-4F39-9AE5-BD436F7C2808@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 08-Dec-2009 at 01:05:21 +0000, Rui Paulo wrote:
> On 7 Dec 2009, at 16:36, Andre Albsmeier wrote:
> > What do people think?
>
>
> This could be a new firmware interface. You might want to check out
> how the latest Linux driver does this.
They do it the same way (comparing the whole 32bit ints).
I don't know about the firmware interface (no docs). Originally,
FreeBSD-6 and FreeBSD-7 use the same fw (V3.0). 8+ uses V3.1.
Our code in FreeBSD-6 is the same as in 7. I didn't have problems with
6 in the very same environment, however, I remember I had to raise
bmissthreshold a bit which means that I actually had missed beacons
(the whole code in question is only triggered on a beacon miss). I don't
have any problems at home where I am only 2m away from the AP (possibly
no missed beacons there).
I have also seen other people report similarly insane numbers. I have
also found reports on the net where the values were reasonable.
As I said, I use fw V3.1 and not the original V3.0. I did so in
FreeBSD-6 as well.
To summarise things, here are my observations of how beacon->number
behaves with different software:
fw3.0 fw3.1
FreeBSD-6 OK OK
FreeBSD-7 insane insane
"insane" means that in 99% of all cases beacon->number is something
like 0xabcdXYXY (abcd can be 0000).
Maybe someone who got the docs can jump in here.
-Andre
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20091208104246.GA18375>