cp: Fix 'REPLY_BUSY' handling logic

Looks like REPLY_BUSY from a PD was just ignored by the CP. Not sure
when this was broken (due to too many changes lately and blame isn't
helpful).

This was just accidentally caught during an unrelated work. Also
indicates we don't have sufficient test coverage.

Signed-off-by: Siddharth Chandrasekaran <[email protected]>

src/osdp_cp.c

@@ -868,12 +868,6 @@ static int cp_phy_state_update(struct osdp_pd *pd)
 			if (sc_is_active(pd)) {
 				pd->sc_tstamp = osdp_millis_now();
 			}
-			if (pd->reply_id == REPLY_BUSY) {
-				pd->phy_tstamp = osdp_millis_now();
-				pd->wait_ms = OSDP_CMD_RETRY_WAIT_MS;
-				pd->phy_state = OSDP_CP_PHY_STATE_WAIT;
-				return OSDP_CP_ERR_CAN_YIELD;
-			}
 			pd->phy_state = OSDP_CP_PHY_STATE_DONE;
 			return OSDP_CP_ERR_NONE;
 		}
@@ -888,6 +882,12 @@ static int cp_phy_state_update(struct osdp_pd *pd)
 		if (rc == OSDP_CP_ERR_GENERIC || rc == OSDP_CP_ERR_UNKNOWN) {
 			goto error;
 		}
+		if (rc == OSDP_CP_ERR_RETRY_CMD) {
+			pd->phy_tstamp = osdp_millis_now();
+			pd->wait_ms = OSDP_CMD_RETRY_WAIT_MS;
+			pd->phy_state = OSDP_CP_PHY_STATE_WAIT;
+			return OSDP_CP_ERR_CAN_YIELD;
+		}
 		if (osdp_millis_since(pd->phy_tstamp) > OSDP_RESP_TOUT_MS) {
 			if (pd->phy_retry_count < OSDP_CMD_MAX_RETRIES) {
 				pd->wait_ms = OSDP_CMD_RETRY_WAIT_MS;
@@ -902,10 +902,6 @@ static int cp_phy_state_update(struct osdp_pd *pd)
 				osdp_cmd_name(pd->cmd_id), pd->cmd_id);
 			goto error;
 		}
-		if (rc == OSDP_CP_ERR_RETRY_CMD) {
-			pd->phy_state = OSDP_CP_PHY_STATE_DONE;
-			return OSDP_CP_ERR_NONE;
-		}
 		ret = OSDP_CP_ERR_INPROG;
 		break;
 	}