PRISM2_NON_VOLATILE_DOWNLOAD in kernel 2.6.14

[Pavel Roskin]

Quoting Jar <jar at pcuf.fi>:

>
> > I see.  Could you please send the output of the following command?
> >
> > hostap_rid wlan0 get ffff
>
> [root at gw hostap-utils-0.4.7]# ./hostap_rid wlan0 get ffff
> 70 31 30 30 30 33 63 30 2e 68 65 78 00 00 53 31 30 31 30 35 30 36 2e 48 45 58
> 00 00

OK, that's p10003c0.hex S1010506.HEX

p10003c0.hex doesn't include the modem-firmware records, and all s1* firmwares I
have use variant 1, not 0.  It would be interesting to know if the original
secondary firmware was s10008c0.hex or something else, but this information is
lost.

> > Please run it before attempts to load any firmware.  It should give the
> names of
> > the original firmwares.
>
> The card has already v1.5.6 in it but it makes no differ, it is always beeing
> like
> that. I have updated from 0.8.0-->1.3.6-->1.4.9-->1.5.6 and always prism_srec
> falsely complains about imcompatibility.

Why "falsely"?  You make it sound like prism2_srec is doing something definitely
wrong, but I think it acts according to the specification.

Let's see http://home.eunet.cz/jt/wifi/RM025-prism-dpm-ver3d0.pdf

8.2.2.4 says variant 0 should not be used in production, so you have a
preliminary version of the hardware, and its compatibility with the production
firmware is not guaranteed.

The same section says that the compatibility ranges are defined for certain
variants.  I assume that it means that in case of mismatching variants the
compatibility ranges should not be compared, and firmware download should not
be performed.

> > In any case, hostap is erring on the safe side, which is the right thing to
> do.
>
> Yes, but the message is false alarm. I can flash with '-i' parameter and
> everything
> is OK after flashing.

Again, you are assuming that prism2_srec is wrong because the firmware works on
your hardware.  However, there is no definite way to know that all features are
working correctly.

An opposite situation would be easier - if prism2_srec allows loading the
firmware that obviously breaks some functionality, then we could say that
prism2_srec is wrong.

Working with preliminary hardware assumes that you take some risks, including
risk of incompatibility with production firmware.  prism2_srec refuses to take
that risk without you realizing it.

> I can also flash with tohose same files with windows
> tool and
> it doesn't warn anything. Sometimes it seems to be too strict with some
> firmware
> files.

Just because a closed source tool does it, it doesn't mean it's the right thing
to do.  I, for one, cannot examine the sources of that software to find the
reasons why it works like this.

Maybe it has a database of the cards with variant 0 hardware that are known to
be compatible with production firmware.  Perhaps prism2_srec could do it as
well.

As for the impact of this issue, here the only report I could find:
http://lists.shmoo.com/pipermail/hostap/2002-November/000329.html

I don't think many people are using those cards with Host AP.

> I agree with you, prism_srec is safe. I have also flashed couple of cards and
> they
> are always worked 100% after that.

OK, good to know.  I hope my patch with be applied to the kernel.

--
Regards,
Pavel Roskin