On 8/25/26 3:28 PM, Ales Musil wrote:
>>> +void
>>> +evpn_mac_binding_sync_run(
>>> +    struct ovsdb_idl_txn *ovnsb_idl_txn,
>>> +    struct ovsdb_idl_index *sbrec_mac_binding_by_lport_ip,
>>> +    const struct sbrec_mac_binding_table *mb_table,
>>> +    const struct hmap *local_datapaths,
>>> +    const struct hmap *evpn_arps,
>>> +    struct mac_cache_data *mac_cache_data,
>>> +    struct ed_type_evpn_mac_binding_sync *data,
>>> +    struct evpn_mb_sync_waker *waker)
>>> +{
>>> +    if (!ovnsb_idl_txn) {
>>> +        data->sb_changes_pending = true;
>> Isn't this a bit too conservative?  We might not have any pending
>> changes, right?  Should we delay setting data->sb_changes_pending until
>> we actually need to update the SB somewhere later in this function?
>>
> So I had a design that actually deffered the sb_changed_pending
> for later, but it turned out to be a mess. So I have decided to make
> it easier. Let me know if I should invest time in trying to come up
> with better solution.

I guess that's probably fine.

Let's leave it as is.

Regards,
Dumitru

_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to