USB booting :: problems with new kernel and modules

hi there.
i need your help please!!!
i'll try to make it as short as possible:
i remastered dsl-2.0RC3b.iso, changed the windowmanager... and so on. i also compiled a new kernel. i used the sources from kernel.org (kernel 2.4.31) and a dsl-2.0-config. i stripped down the config because i don't need all the flags and modules. after that, i copied the minirt24.gz, gunzipped it, mounted it and replaced the necessary modules... cloop, usb-ohci.o and so on, which means that i copied the modules that i've compiled for the new kernel. in the end i created the new image, used the pendrive script (which i have modified to fit my needs) to write the image to an usb-stick and tried to boot it. now here's the problem:
when i try to boot the standard dsl-kernel everything works, but when i want to boot my own kernel, i get an error message and then i'm dropped to the grub-shell. the messages i get (after commenting out the ">/dev/null"s in the linuxrc) are like "/modules/scsi/usb**: init_module: ... insmod errors can be caused by incorrect modules parameters, including invalid IO or IRQ parameters"
so the reason that the system doesn't boot is: some modules aren't loaded correctly, especially the usb-ohci.o module. i wasn't able to figure out yet, why the hell it doesn't work. i downloaded the kernel, applied the knoppix-patch, copied over the dsl-config, made "oldconfig", made "menuconfig", and then made "dep" "bzImage" "modules" and "modules_install", so everything should be alright. and yes, i made sure that there are all modules that i need. and of course i enabled loop-support, zlib-compression (for cloop support), initrd-support (and so on) in the config file. and i changed the isolinux.cfg in the boot folder. but it's always the same: the modules aren't loaded correctly and i get messages about unresolved symbols.
can anyone give me a hint, why the modules aren't loaded correctly although they are compiled for the kernel i use?
(i made a symlink for gcc --> gcc-2.95 and g++ --> g++-2.95 so that i can be sure that always the right compiler is used)

for your information:
the kernel, modules and the new dsl-iso are created under debian3.1 and then written to a kingston usb-stick. the machine i want to boot has a VIA Epia ME6000 board (no hdd, no cdrom, no pci cards, just the plain board)

any ideas?

greetings

mark

hi again.

no one? no tipps, hints or anything else that could lead me into the right direction?
the strange thing is that sometimes (after compiling a new kernel+modules) the modules work and sometimes they don't work and i get symbol errors. the only module that doesn't want to load at all is usb-ohci.o.
i even tried to use the new kernel and the old inirtd. since it's the same kernelversion i thought, maybe the usb-ohci.o module would work. but it doesn't.

greetings

mark

your really remastered dsl 2.0RC3?. dsl 2.0 has a typo in the linuxrc script which prevents usb-ohci from getting loaded at all.

hello.

yes, i remastered dsl-2.0RC3. there's a typo in the linuxrc???
where? the only lines concerning the usb-ohci.o module are this ones:

[...]
240 if test -f /modules/scsi/usbcore.o; then
241 $INSMOD /modules/scsi/usbcore.o
242 FOUNDUSB=""
243 for i in usb-uhci.o usb-ohci.o ehci-hcd.o ; do
244 test -f /modules/scsi/$i && $INSMOD /modules/scsi/$i &&
FOUNDUSB="yes"
245 done
[...]

what i know is that the usbcore module has to be loaded before trying to load usb-uhci.o, usb-ohci.o or ehci-hcd.o.
and: if there is a typo, why are there no problems with the standard-kernel and its modules, because when booting the standard-kernel, the usb-ohci.o module loads without problems.
now i'm really confused.

any other tipps, concerning the symbol errors?

greetings

mark

oops - i did make a mistake here: the typo concerned ehci-hcd.o only (and your version doesn't contain that typo - must be in the 2.0 release version only).

have you remastered KNOPPIX (/lib/modules) also?

it seems that your kernel somehow 'doesn't like' the minirt ramdisk bootup phase. does any of your kernel changes touch the ramdisk subsystem?

Next Page...
original here.