SUM query with non-unique ID from an INNER JOIN?

I have the following problem:

(this is a simplified setup)

I have a datatable
(id, date, value)
1, 1.1.2005, 300
1, 2.1.2005, 400
2, 1.1.2005, 300
2, 2.1.2005, 400
3, 1.1.2005, 300
3, 2.1.2005, 400


and a keytable
(parentid, id)
1, 1
1, 1
1, 2
2, 1
2, 2
2, 2
2, 3

Of course, "id" in my data table references "id" in my key table.
Now i want to query the sum of "value" group by "date" for a certain "parentid"

Unfortunately, as You can see, the "id" is not unique for a "parentid".

SELECT SUM( value ), date FROM
datatable INNER JOIN keytable ON datatable.id = keytable.id
WHERE keytable.parentid = 2
GROUP BY date ORDER BY date

will return the sum of these records:
2, 1.1.2005, 300
2, 2.1.2005, 400
2, 1.1.2005, 300 -- doublette!
2, 2.1.2005, 400 -- doublette!
3, 1.1.2005, 300
3, 2.1.2005, 400
because id=2 occurs twice for the keytable.

Has anybody an idea how i can avoid the doublettes without using an IN( ... ) subselect which will probably take about forever?

Thanks in advance,
Joachim