You should not expect any real-time behavior if you use a L4Linux task to serve interrupts. Instead of this, implement the time-critical code (e.g. I/O to VME board, computations of output values) in an L4 task. Then let this task communicate with L4Linux by IPC (usually by using an own thread).
Then you will get worst-case latencies of about 85 microseconds (P800).
A quick note: This relatively high number is not due to a bad implementation
of the kernel itself. It is the result of a bunch of impacts such as TLB misses, cache misses for data, code and cache misses on parsing page tables, etc...
Sebastian
l4-hackers@os.inf.tu-dresden.de