On Sun, Dec 13, 2020 at 12:48:38PM +0100, Christophe JAILLET wrote:
> In case of error after calling 'ocelot_init()', it must be undone by a
> corresponding 'ocelot_deinit()' call, as already done in the remove
> function.
>
This changes the behavior slightly in another way as well, but it's
probably a bug fix.
drivers/net/ethernet/mscc/ocelot_vsc7514.c
1250 ports = of_get_child_by_name(np, "ethernet-ports");
1251 if (!ports) {
1252 dev_err(ocelot->dev, "no ethernet-ports child node found\n");
1253 return -ENODEV;
1254 }
1255
1256 ocelot->num_phys_ports = of_get_child_count(ports);
1257 ocelot->num_flooding_pgids = 1;
1258
1259 ocelot->vcap = vsc7514_vcap_props;
1260 ocelot->inj_prefix = OCELOT_TAG_PREFIX_NONE;
1261 ocelot->xtr_prefix = OCELOT_TAG_PREFIX_NONE;
1262 ocelot->npi = -1;
1263
1264 err = ocelot_init(ocelot);
1265 if (err)
1266 goto out_put_ports;
1267
1268 err = mscc_ocelot_init_ports(pdev, ports);
1269 if (err)
1270 goto out_put_ports;
1271
1272 if (ocelot->ptp) {
1273 err = ocelot_init_timestamp(ocelot, &ocelot_ptp_clock_info);
1274 if (err) {
1275 dev_err(ocelot->dev,
1276 "Timestamp initialization failed\n");
1277 ocelot->ptp = 0;
1278 }
In the original code, if ocelot_init_timestamp() failed we returned
a negative error code but now we return success. This probably is what
the original authors intended, though.
1279 }
1280
1281 register_netdevice_notifier(&ocelot_netdevice_nb);
1282 register_switchdev_notifier(&ocelot_switchdev_nb);
1283 register_switchdev_blocking_notifier(&ocelot_switchdev_blocking_nb);
1284
1285 dev_info(&pdev->dev, "Ocelot switch probed\n");
1286
1287 out_put_ports:
1288 of_node_put(ports);
1289 return err;
1290 }
regards,
dan carpenter
