Re: [RFC 1/2] net/ethtool: Add new coalescing parameter for queue

[Florian Fainelli <f.fainelli@gmail.com>]

On 07/12/15 20:42, kan.liang@intel.com wrote:
> From: Kan Liang <kan.liang@intel.com>
> 
> Intrdouce "queue" option for coalesce getting and setting.
> For coalesce getting, only the coalescing parameters from specific
> queue will be passed to user space.
> For coalesce setting, the coalescing parameters will only be applied to
> specific queue.
> If the queue is set to -1, the coalescing parameters will apply to all
> queues.

This looks like a good start, but there are a few things that need to be
clarified, in particular:

- if the number of TX and RX queues differ, but the ethtool coalesce
structure contains parameters that affect both the RX and TX side, and
the queue number is invalid/non-existent for one of these sides, what is
the expected outcome? Same question with specifying a queue number, with
RX queue N not belonging to the same queue pair as TX queue N?

- from an user perspective do we want to iterate over all queues to set
their parameters, or should we have a queue bitmask parameter which
allows setting them with the same settings in one shot? What would be
the appropriate bitmask size then (32-bits with 16-bits for TX and
16-bits for RX might be too small)?

Thanks
-- 
Florian