Apologies for the latency.

On Wed, 2009-06-03 at 11:27 +0000, Jarek Poplawski wrote:

After the second look I have some questions:
- if it's really aimed to skip checking by ->get() tp's before they're
configured in ->change(), maybe instead of using tp_c to check this it
would be simpler to generally skip calling ->get() for newly created
tp's?
- otherwise the current method probably needs adding a tp_c check for
NULL in u32_destroy()?
- it seems this method would also need adding a 'handle' lookup to
the u32_change(); otherwise its 'handle' parameter isn't controlled
for for uniqueness, unless I miss something?

Lets just ignore the need for these changes since the patch fixes them
for now. I would still like to make the changes i suggested but later
with more thought put into them.

cheers,
jamal