[Prism54-devel] Crosscompiling for prism54?

Gerhard Weiß g.weiss@hermstedt.de
Mon, 10 Nov 2003 10:11:34 +0100


Am 08.11.2003 um 08:38 schrieb Herbert Valerio Riedel:

> On Fri, Nov 07, 2003 at 02:05:41PM +0100, Gerhard Weiß wrote:
>> this might be a newbie question, but I started on Linux only a few
>> weeks ago...
>
> great! hope you'll become as addicted as I became years ago... =)
>
>> I've got an i386 Linux for compiling a kernel for an PPC405 system
> sounds sweet :-)
at least it is quite small - and a 405 together with VIA686 is a quite 
complete
system.
>
>> (which we build from scratch). This works quite good so far. Until 
>> now,
>> we compiled all driver statically into the kernel.
>
>> Now I've got a couple of Prism54 mini-pci cards here, which I like to
>> incorporate, so first step was to upgrade to 2.4.23-pre9 (done). But
>> after compiling in the prism stuff (statically again) it doesn't seem
>> to find the card, while I can see the card pretty well (lspci as well
>> as from bootloader).
> what's the exact kernel-log output for the prism54 driver while
> initializing?
Exact none - thats my problem
>
>> So, is static linking of the prism driver supported at all? If, so,
>> there does the matching start from? Currently I cannot find the point
>> where the first device matching should happen (since here I should
>> start debugging, right?)
>
> read up about the hotplug support for pci-like devices...
I thought that already and started with that last friday - knowing now 
that this
is the right way, I'll go on.

>
>> If only driver modules should work - would anybody be so kind to
>> explain or point me into the right direction, how to set up that 
>> module
>> stuff when cross-compiling the kernel?
>
> well, static linkage does not work yet, especially for cards that a
> plugged in permanently; (reason is: the firmware can't be uploaded
> since there's no fs mounted yet where one could get the firmware
> from... and firmware uploading is done at probing-time)
>
> ...but I'm working on making the situation better (i.e. retry firmware
> uploading at device open() time... but I have to clean up some stuff
> before) :-
>
That explains a lot. Thank you.

Gerd.
-- 
Gerhard Weiss * Voice +49 621 7650 322 * Fax +49 621 7650 100
HERMSTEDT AG * Carl-Reuther-Str. 3 * 68305 Mannheim * Germany