i3c: master: svc: fix probe failure when no i3c device exist

commit 6e13d65 upstream.

I3C masters are expected to support hot-join. This means at initialization
time we might not yet discover any device and this should not be treated
as a fatal error.

During the DAA procedure which happens at probe time, if no device has
joined, all CCC will be NACKed (from a bus perspective). This leads to an
early return with an error code which fails the probe of the master.

Let's avoid this by just telling the core through an I3C_ERROR_M2
return command code that no device was discovered, which is a valid
situation. This way the master will no longer bail out and fail to probe
for a wrong reason.

Cc: [email protected]
Fixes: dd3c528 ("i3c: master: svc: Add Silvaco I3C master driver")
Signed-off-by: Frank Li <[email protected]>
Acked-by: Miquel Raynal <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Alexandre Belloni <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>

drivers/i3c/master/svc-i3c-master.c

@@ -789,6 +789,10 @@ static int svc_i3c_master_do_daa_locked(struct svc_i3c_master *master,
 				 */
 				break;
 			} else if (SVC_I3C_MSTATUS_NACKED(reg)) {
+				/* No I3C devices attached */
+				if (dev_nb == 0)
+					break;
+
 				/*
 				 * A slave device nacked the address, this is
 				 * allowed only once, DAA will be stopped and
@@ -1263,11 +1267,17 @@ static int svc_i3c_master_send_ccc_cmd(struct i3c_master_controller *m,
 {
 	struct svc_i3c_master *master = to_svc_i3c_master(m);
 	bool broadcast = cmd->id < 0x80;
+	int ret;
 
 	if (broadcast)
-		return svc_i3c_master_send_bdcast_ccc_cmd(master, cmd);
+		ret = svc_i3c_master_send_bdcast_ccc_cmd(master, cmd);
 	else
-		return svc_i3c_master_send_direct_ccc_cmd(master, cmd);
+		ret = svc_i3c_master_send_direct_ccc_cmd(master, cmd);
+
+	if (ret)
+		cmd->err = I3C_ERROR_M2;
+
+	return ret;
 }
 
 static int svc_i3c_master_priv_xfers(struct i3c_dev_desc *dev,