On 11/27/2020 3:19 PM, Tobias Waldekranz wrote:

quoted

The initial design of switchdev was transactions. First there was a
prepare call, where you validated the requested action is possible,
and allocate resources needed, but don't actually do it. This prepare
call is allowed to fail. Then there is a second call to actually do
it, and that call is not allowed to fail. This structure avoids most
of the complexity of the unwind, just free up some resources. If you
never had to allocate the resources in the first place, better still.

OK I think I finally see what you are saying. Sorry it took me this
long. I do not mean to be difficult, I just want to understand.

How about this:

- Add a `lags_max` field to `struct dsa_switch` to let each driver
  define the maximum number supported by the hardware. By default this
  would be zero, which would mean that LAG offloading is not supported.

- In dsa_tree_setup we allocate a static array of the minimum supported
  number across the entire tree.

- When joining a new LAG, we ensure that a slot is available in
  NETDEV_PRECHANGEUPPER, avoiding the issue you are describing.

- In NETDEV_CHANGEUPPER, we actually mark it as busy and start using it.

Sounds reasonable to me.
-- 
Florian