essen@dev-extend.eu said:
> There's probably some differences in the detection while booting and
> after boot. 

In theory no, because USB exploration takes place after interrupts,
kernel threads etc are running and the general flow of code should
be the same as later at runtime.
In practice, ther can be subtle timing differences, and interactions
with BIOS / power management / ACPI which gets in-between.

I'd start with some try-and-error like apprach - increase all
the delays in USB code by a factor of 2 or so and see whether
it makes a difference.

What hardware do you have, an is it just the mouse or are
highspeed devices affected as well?
(This makes a serious difference because completely different
controller are involved on a USB2 system.)

best regards
Matthias