[Mellanox] wait SFP ready when receive an insert event with module host management mode enabled

platform/mellanox/mlnx-platform-api/sonic_platform/chassis.py

@@ -475,6 +475,8 @@ def get_change_event_for_module_host_management_mode(self, timeout):
                 if fd_type == 'hw_present':
                     # event could be EVENT_NOT_PRESENT or EVENT_PRESENT
                     event = sfp.EVENT_NOT_PRESENT if fd_value == 0 else sfp.EVENT_PRESENT
+                    if fd_value == 1:
+                        s.processing_insert_event = True
                     s.on_event(event)
                 elif fd_type == 'present':
                     if str(fd_value) == sfp.SFP_STATUS_ERROR:

platform/mellanox/mlnx-platform-api/sonic_platform/sfp.py

@@ -476,6 +476,7 @@ def __init__(self, sfp_index, sfp_type=None, slot_id=0, linecard_port_count=0, l
             self.state = STATE_DOWN
         else:
             self.state = STATE_FCP_DOWN
+        self.processing_insert_event = False
 
     def __str__(self):
         return f'SFP {self.sdk_index}'
@@ -1504,7 +1505,12 @@ def action_on_start(cls, sfp):
                 sfp.set_hw_reset(1)
                 sfp.on_event(EVENT_RESET)
             else:
-                sfp.on_event(EVENT_POWER_ON)
+                if not sfp.processing_insert_event:
+                    sfp.on_event(EVENT_POWER_ON)
+                else:
+                    sfp.processing_insert_event = False
+                    logger.log_info(f'SFP {sfp.sdk_index} is processing insert event and needs to wait module ready')
+                    sfp.on_event(EVENT_RESET)
 
     @classmethod
     def action_fcp_on_start(cls, sfp):

platform/mellanox/mlnx-platform-api/sonic_platform/wait_sfp_ready_task.py

@@ -62,8 +62,8 @@ def schedule_wait(self, sfp_index):
         """
         logger.log_debug(f'SFP {sfp_index} is scheduled for waiting reset done')
         with self.lock:
-            if len(self._wait_dict) == 0:
-                is_empty = True
+            is_empty = len(self._wait_dict) == 0
+
             # The item will be expired in 3 seconds
             self._wait_dict[sfp_index] = time.time() + self.WAIT_TIME
 

platform/mellanox/mlnx-platform-api/tests/test_change_event.py

@@ -83,15 +83,16 @@ def test_get_change_event_legacy(self, mock_status, mock_time, mock_create_poll,
         _, change_event = c.get_change_event(timeout)
         assert 'sfp' in change_event and sfp_index in change_event['sfp'] and change_event['sfp'][sfp_index] == '2'
         assert 'sfp_error' in change_event and sfp_index in change_event['sfp_error'] and change_event['sfp_error'][sfp_index] == 'some error'
-
+
+    @mock.patch('sonic_platform.wait_sfp_ready_task.WaitSfpReadyTask.get_ready_set')    
     @mock.patch('sonic_platform.sfp.SFP.get_fd')
     @mock.patch('select.poll')
     @mock.patch('time.time')
     @mock.patch('sonic_platform.device_data.DeviceDataManager.is_module_host_management_mode', mock.MagicMock(return_value=True))
     @mock.patch('sonic_platform.device_data.DeviceDataManager.get_sfp_count', mock.MagicMock(return_value=1))
     @mock.patch('sonic_platform.chassis.extract_RJ45_ports_index', mock.MagicMock(return_value=[]))
     @mock.patch('sonic_platform.module_host_mgmt_initializer.ModuleHostMgmtInitializer.initialize', mock.MagicMock())
-    def test_get_change_event_for_module_host_management_mode(self, mock_time, mock_create_poll, mock_get_fd):
+    def test_get_change_event_for_module_host_management_mode(self, mock_time, mock_create_poll, mock_get_fd, mock_ready):
         """Test steps:
             1. Simulate polling with no event
             2. Simulate polling the first dummy event. (SDK always return a event when first polling the fd even if there is no change)
@@ -163,6 +164,7 @@ def get_fd(fd_type):
         s.determine_control_type = mock.MagicMock(return_value=sfp.SFP_FW_CONTROL)
         s.set_control_type = mock.MagicMock()
         mock_time.side_effect = [0, timeout]
+        mock_ready.return_value = set([0])
         mock_hw_present_file.read.return_value = sfp.SFP_STATUS_INSERTED
         _, change_event = c.get_change_event(timeout)
         assert 'sfp' in change_event and sfp_index in change_event['sfp'] and change_event['sfp'][sfp_index] == sfp.SFP_STATUS_INSERTED
@@ -173,6 +175,7 @@ def get_fd(fd_type):
         print(c.registered_fds)
 
         # error event, expect returning error
+        mock_ready.return_value = []
         mock_time.side_effect = [0, timeout]
         mock_poll.poll.return_value = [(3, 10)]
         mock_present_file.read.return_value = sfp.SFP_STATUS_ERROR
@@ -193,6 +196,7 @@ def get_fd(fd_type):
 
         # plug in a software control cable, expect returning insert event
         mock_time.side_effect = [0, timeout]
+        mock_ready.return_value = set([0])
         mock_poll.poll.return_value = [(1, 10)]
         mock_hw_present_file.read.return_value = sfp.SFP_STATUS_INSERTED
         s.determine_control_type.return_value = sfp.SFP_SW_CONTROL