Where to start...
Turns out the flash layout bigbrd created for me is slightly different from the previous standard. I have a 608K zImage (instead of 640K) which gives extra room for the root partition. The boot and params are the same size as before. Seems like the order they are on the chip is different too (boot, params, zImage, root).
This means that a kernel with parport built into it is unlikely to fit, without removing other equally important pieces. I am loathe to go in and start rearranging things though, for fear that if I break it I won't be able to get it back together again without a month's delay.
In the meantime, I happened across a tgz package on Vit's site which contains parport.o and parport_pc.o ( http://www.luban.org/Webpal/modules-bin-2.4.18-rmk-4.tgz ). I've unpackaged it, and insmod parport and insmod parport_pc io=0x278 irq=13. Unfortunately, parport_pc.o gives me an error that it cannot resolve the symbols "outsl" and "insl", so it won't load. A bit of searching around says these may be related to the non-existant PCI of the webpal, so those functions were probably not compiled into the diminutive zImage kernel bigbrd has provided for me.
I wonder if there is a workaround. One incomplete message I read suggested:
> How is that supposed to work on systems without PCI? For now I have
> defined
>
> #define insl(port,buf,len) isa_insb(port,buf,(len)<<2)
> #define outsl(port,buf,len) isa_outsb(port,buf,(len)<<2)
Apparently taking the parport_pc calls and moving them to an equivalent call for the ISA bus. Could this sort of remapping be put into its own small LKM, say PCI.o, and allow parport_pc.o to load? probably an entirely rhetorical question as I wouldn't even know where to begin to try it.
I wonder if the source code to the parport_pc.o file could have these two #defines added to it and recompiled, although I don't see the source on Vit's site. Maybe this is the source code to the module? http://www.iglu.org.il/lxr/source/drivers/parport/parport_pc.c?v=linux-2.4.18
Perhaps someone as clueless as I about the ways of the Linux was never meant to have printing nirvana, and I should be entirely content with my now QUIET (after a pointer to noflushd) Samba based file server.
