Perhaps you should add a printk at the beginning of input's init
function. The delay could happen during the startup of the input layer
instead of the 1394 drivers.
ftp://ftp.kernel.org/pub/linux/kernel/people/akpm/patches/2.6/2.6.17/2.6.17-mm6/
and start bisecting within the set of 1394 patches. Among them, the
patches to sbp2, dv1394, and raw1394 are unrelated to the problem. Note
that "origin.patch" touches drivers/ieee1394/ too.
The VIA VT6306 OHCI controller which you have according to your lspci
output is known to work; yours doesn't even have the small quirk
mentioned in http://www.linux1394.org/view_device.php?id=713 .