Re: [PATCH v2 net-next 02/12] net: ethtool: add support for MAC Merge layer

[Vladimir Oltean]

On Mon, Jan 16, 2023 at 09:02:55AM +0000, Somisetty, Pranavi wrote:
> Does it make sense to have two separate functions, one to
> enable/disable pmac and tx preemption status and configure additional
> fragment size (both parameters can be configured after negotiating
> with the link partner), and another to modify the verification related
> parameters, i.e, enable/disable verification and verify time value?
> 
> For example, if the user wants to change the verify time value, Is
> there a way to change it without disturbing the value of the
> additional fragment size that has been set through LLDP after
> negotiating with the link partner?

Could you please clarify what you mean by "disturbing"? The handler for
the ETHTOOL_MSG_MM_SET message first calls get_mm(struct ethtool_mm_state),
then mm_state_to_cfg(struct ethtool_mm_state, struct ethtool_mm_cfg),
then set_mm(struct ethtool_mm_cfg). In other words, a SET ethtool
command translates into a read-modify-write in the kernel. If user space
did not request the modification of the additional fragment size, just
the verify time, then the set_mm() call will contain the old additional
fragment size and the new verify time.