reading / writing multiple registers

[jtc@redback.com (J.T. Conklin)]

While debugging a problem in calling inferior functions on one of our
targets, I noticed that GDB fetched registers individually instead of
doing a bulk transfer.  There's a comment:

    Prior to the recent fixes to check for partial updates, both read
    and write_register_bytes always checked to see if any registers
    were stale, and then called target_fetch_registers (-1) to update
    the whole set.  This caused really slowed things down for remote
    targets.

This is interesting, since not fetching the whole set of registers
really slows down things as well.  

Perhaps there needs to be a heuristic where the number and/or size of
registers that need to be fetched/stored is compared with the size of
the entire register set.  If some threshold is reached, the whole set
would be fetched/stored, otherwise the registers would be handled
individually.

Thoughts?

        --jtc

-- 
J.T. Conklin
RedBack Networks