On Wed, Nov 02, 2022 at 05:32:57PM -0400, izunna.otiji...@renesas.com wrote:
> @@ -521,6 +521,12 @@ static int clock_management_set(struct clock *c, struct > port *p, > *changed = 1; > respond = 1; > break; > + case MID_DOMAIN: > + mtd = (struct management_tlv_datum *) tlv->data; > + c->dds.domainNumber = mtd->val; This naive approach of simply changing the value within the dynamic data structure is not going to work. A correct implementation must purge all of the live state that depends on domainNumber, for example the foreign clock lists and the unicast subscriptions. That will be a major change. I suspect that stopping the ptp4l program and re-starting with a new configuration will be just as fast in practice, and it is certainly easier. So, if you _really_ think supporting SET(domainNumber) is important, then you have to: 1. Ensure that the code correctly handles the multiple side effects of a change on domain number. 2. Provide analysis that explains why your changes for #1 are complete. 3. Prove that the new logic is significantly faster than simply re-starting the ptp4l service. The same goes for the other four patches in this series. Thanks, Richard _______________________________________________ Linuxptp-devel mailing list Linuxptp-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-devel
