Linksys Firmware problem.

[Pavel Roskin]

On Mon, 1 Sep 2003 wim at djo.tudelft.nl wrote:

> > On Mon, 1 Sep 2003 wim at djo.tudelft.nl wrote:
> >
> > > It may be that 1.4.2 has a bug, but that bug can't be 'fatal for the Host
> > > AP mode'.  I've got a BenQ AWL500 access point, build around an AWL100
> > > PCMCIA card (prism2.5). The AP works nicely, although it's running on
> > > STA 1.4.2.
> >
> > How do you know that there is no firmware loaded into the RAM?
>
> I can't see that. But it's hard to believe that the manufacturer puts buggy
> firmware into their cards, just to load better firmware when they actually
> want to use it. If it were just a license problem, they wouldn't circumvent
> that by silently use another version, although it's less visible that way.
>
> > How do you know that the card operates in the HostAP mode?
>
> Can the access point be an access point by not being in access point mode?
> Enlighten me.

Of course.  If tertiary firmware is loaded, the card works as a firmware
AP.  In this case, many AP functions are managed by the firmware, not by
the host software.

OK, that's how I understand the situation.  We may have 3 cases:

1) The existing secondary firmware 1.4.2 is used in the HostAP mode.
2) Some other secondary firmware is loaded and used in the HostAP mode.
3) Tertiary firmware is loaded and used in the firmware AP mode.

Since we know that firmware 1.4.2 is buggy (it's acknowledged in the
release noted for 1.4.9), it's quite unlikely that it's used.  The choice
between variants 2 and 3 is likely determined by the processor speed,
desired features and the licensing cost of the tertiary firmware.

The argument that the manufacturer would have flashed the better firmware
doesn't hold for several reasons:

1) It's easier to update card firmware as part of the AP firmware (i.e.
firmware for the whole product) if flashing is not involved.

2) Hardware manufacturers are very conservative.  If firmware 1.4.2 is
known to initialize and accept RAM download, they would go with it rather
than upgrade.

3) Cards with 1.4.2 firmware may have been produced already by another
company and could be bought cheaply.  Hardware manufacturers are reluctant
to change anything in the hardware made by others.

4) Licensing cost of flashable tertiary firmware can be higher that the
cost of the corresponding RAM firmware.

5) OEMs prefer reliability over speed.  Failure means equipment return or
support call.  Slow startup is often unnoticed or tolerated by the users.

Don't get me wrong.  The manufacturers are conservative about things that
don't need to be changed, but they care about correct functionality and
competitive features.  That's why I think that using buggy firmware for
the actual wireless operation is very unlikely.

> So, they sell El Cheapo No Brand cards with their name on in and they
> construct a workaround on their CPU board that uses that card. Hmmm.
> Maybe I'll once get so far to check that.. (There are several flash
> updates for the AWL500 and these file are byte-swapped but appear
> otherwise not compressed, so maybe part of it may be recognised somehow
> as station firmware.)

I've downloaded the image and there is word "Tertiary" in it.  It seems
that the tertiary firmware is present in the image as an unencrypted
binary.

-- 
Regards,
Pavel Roskin