Update qos sai due to no lossless pg for some platforms (#16440)

JibinBao · web-flow · commit 485ec3a46ff8 · 2025-04-01T08:30:22.000+03:00
1. For spc4 and above, there is only the lossy buffer, so the buffer for the lossless buffer will be taken by the lossy buffer. If the packet size is too small, the packet number sent to occupy the shared buffer will increase a lot, which will lead to the descriptor being exhausted, so update testQosSaiPgSharedWatermark, testQosSaiQSharedWatermark, and testQosSaiLossyQueue accordingly.
2. Remove the test config of scheduler.block_data_plane, otherwise it might raise yang validation error when do config reload
3. When there is no lossless buffer, return a dump buffer lossless pg profile, and skip tests related to lossless buffer case dynamically
4. Skip fixture reaseAllports for mellanox device, because after qos test is finished, the teardown will do config reload, it will restore the config of ports, we don't need this fixture before running tests. Also it can save 2 minutes
5. list the relevant Prs:
[Mellanox] Update buffer calculations for Mellanox-SN5600-C224O8 SKU sonic-buildimage#20992
[Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056
[Mellanox] Update Mellanox-SN5600-C256S1 buffer calculations sonic-buildimage#20991
[Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056
[Mellanox] Update Mellanox-SN5600-C256S1, Mellanox-SN5600-C224O8 buffers and DSCP mapping sonic-buildimage#21427
[Mellanox] Add x86_64-nvidia_sn5610n-r0 new platform and SKUs sonic-buildimage#21056
[Mellanox] Update DSCP mapping for SN5600, SN5610 SKUs sonic-buildimage#21762
diff --git a/tests/qos/files/mellanox/qos_param_generator.py b/tests/qos/files/mellanox/qos_param_generator.py
@@ -236,6 +236,11 @@ def calculate_parameters(self):
         self.qos_params_mlnx['wm_pg_shared_lossy'].update(wm_shared_lossy)
         wm_shared_lossy["pkts_num_margin"] = 8
         self.qos_params_mlnx['wm_q_shared_lossy'].update(wm_shared_lossy)
+        if 'lossy_dscp' in self.egressLossyProfile:
+            lossy_queue['dscp'] = self.egressLossyProfile['lossy_dscp']
+            self.qos_params_mlnx['wm_pg_shared_lossy']['dscp'] = self.egressLossyProfile['lossy_dscp']
+            self.qos_params_mlnx['wm_q_shared_lossy']['dscp'] = self.egressLossyProfile['lossy_dscp']
+            self.qos_params_mlnx['wm_q_shared_lossy']['queue'] = self.egressLossyProfile['lossy_queue']
 
         wm_buf_pool_lossless = self.qos_params_mlnx['wm_buf_pool_lossless']
         wm_buf_pool_lossless['pkts_num_trig_pfc'] = pkts_num_trig_pfc
diff --git a/tests/qos/files/mellanox/special_qos_config.yml b/tests/qos/files/mellanox/special_qos_config.yml
@@ -27,12 +27,12 @@ qos_params:
         xon_4:
             packet_size: 800
         lossy_queue_1:
-            packet_size: 800
+            packet_size: 1200
         wm_pg_shared_lossless:
             packet_size: 800
             pkts_num_margin: 7
         wm_pg_shared_lossy:
-            packet_size: 800
-            pkts_num_margin: 5
+            packet_size: 1200
+            pkts_num_margin: 8
         wm_q_shared_lossy:
-            packet_size: 800
+            packet_size: 1200
diff --git a/tests/qos/qos_sai_base.py b/tests/qos/qos_sai_base.py
@@ -350,8 +350,24 @@ def __getBufferProfile(self, request, dut_asic, os_version, table, port, priorit
             else:
                 bufferProfileName = out.translate({ord(i): None for i in '[]'})
         else:
-            bufferProfileName = bufkeystr + dut_asic.run_redis_cmd(
-                argv=["redis-cli", "-n", db, "HGET", keystr, "profile"])[0]
+            profile_content = dut_asic.run_redis_cmd(argv=["redis-cli", "-n", db, "HGET", keystr, "profile"])
+            if profile_content:
+                bufferProfileName = bufkeystr + profile_content[0]
+            else:
+                logger.info("No lossless buffer. To compatible the existing case, return dump bufferProfilfe")
+                dump_buffer_profile = {
+                    "profileName": f"{bufkeystr}pg_lossless_0_0m_profile",
+                    "pool": "ingress_lossless_pool",
+                    "xon": "0",
+                    "xoff": "0",
+                    "size": "0",
+                    "dynamic_th": "0",
+                    "pg_q_alpha": "0",
+                    "port_alpha": "0",
+                    "pool_size": "0",
+                    "static_th": "0"
+                }
+                return dump_buffer_profile
 
         result = dut_asic.run_redis_cmd(
             argv=["redis-cli", "-n", db, "HGETALL", bufferProfileName]
@@ -1765,10 +1781,13 @@ def releaseAllPorts(
             Raises:
                 RunAnsibleModuleFail if ptf test fails
         """
-        self.runPtfTest(
-            ptfhost, testCase="sai_qos_tests.ReleaseAllPorts",
-            testParams=dutTestParams["basicParams"]
-        )
+        if isMellanoxDevice(duthosts[0]):
+            logger.info("skip reaseAllports fixture for mellanox device")
+        else:
+            self.runPtfTest(
+                ptfhost, testCase="sai_qos_tests.ReleaseAllPorts",
+                testParams=dutTestParams["basicParams"]
+            )
 
     def __loadSwssConfig(self, duthost):
         """
@@ -2111,12 +2130,25 @@ def egressLossyProfile(
 
         srcport = dutConfig["dutInterfaces"][dutConfig["testPorts"]["src_port_id"]]
 
+        is_lossy_queue_only = False
+
         if srcport in dualtor_ports_for_duts:
             queues = "0-1"
         else:
-            queues = "0-2"
+            if isMellanoxDevice(duthost):
+                cable_len = dut_asic.shell(f"redis-cli -n 4 hget 'CABLE_LENGTH|AZURE' {srcport}")['stdout']
+                if cable_len == '0m':
+                    is_lossy_queue_only = True
+                    logger.info(f"{srcport} has only lossy queue")
+            if is_lossy_queue_only:
+                is_lossy_queue_only = True
+                queue_table_postfix_list = ['1-3', '4', '5']
+                queue_to_dscp_map = {'1-3': '1', '4': '11', '5': '31'}
+                queues = random.choice(queue_table_postfix_list)
+            else:
+                queues = "0-2"
 
-        yield self.__getBufferProfile(
+        egress_lossy_profile = self.__getBufferProfile(
             request,
             dut_asic,
             duthost.os_version,
@@ -2125,6 +2157,12 @@ def egressLossyProfile(
             srcport,
             queues
         )
+        if is_lossy_queue_only:
+            egress_lossy_profile['lossy_dscp'] = queue_to_dscp_map[queues]
+            egress_lossy_profile['lossy_queue'] = '1' if queues == '1-3' else queues
+        logger.info(f"queues:{queues}, egressLossyProfile: {egress_lossy_profile}")
+
+        yield egress_lossy_profile
 
     @pytest.fixture(scope='class')
     def losslessSchedProfile(
diff --git a/tests/qos/test_qos_sai.py b/tests/qos/test_qos_sai.py
@@ -175,6 +175,11 @@ def check_skip_xon_hysteresis_test(xonHysteresisKey, dutQosConfig,
             " Pls see qos.yaml for the port idx's that are needed.")
 
 
+def skip_test_on_no_lossless_pg(portSpeedCableLength):
+    if portSpeedCableLength == "0_0m":
+        pytest.skip("skip the test due to no buffer lossless pg")
+
+
 class TestQosSai(QosSaiBase):
     """TestQosSai derives from QosSaiBase and contains collection of QoS SAI test cases.
 
@@ -411,6 +416,7 @@ def testQosSaiPfcXoffLimit(
                 "Additional DSCPs are not supported on non-dual ToR ports")
 
         portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        skip_test_on_no_lossless_pg(portSpeedCableLength)
         if dutTestParams['hwsku'] in self.BREAKOUT_SKUS and 'backend' not in dutTestParams['topo']:
             qosConfig = dutQosConfig["param"][portSpeedCableLength]["breakout"]
         else:
@@ -508,6 +514,7 @@ def testPfcStormWithSharedHeadroomOccupancy(
             pytest.skip("Shared Headroom has to be enabled for this test")
 
         portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        skip_test_on_no_lossless_pg(portSpeedCableLength)
         if xonProfile in list(dutQosConfig["param"][portSpeedCableLength].keys()):
             qosConfig = dutQosConfig["param"][portSpeedCableLength]
         else:
@@ -681,6 +688,7 @@ def testQosSaiPfcXonLimit(
                 "Additional DSCPs are not supported on non-dual ToR ports")
 
         portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        skip_test_on_no_lossless_pg(portSpeedCableLength)
         if xonProfile in list(dutQosConfig["param"][portSpeedCableLength].keys()):
             qosConfig = dutQosConfig["param"][portSpeedCableLength]
         else:
@@ -858,6 +866,7 @@ def testQosSaiHeadroomPoolSize(
         """
 
         portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        skip_test_on_no_lossless_pg(portSpeedCableLength)
         qosConfig = dutQosConfig["param"][portSpeedCableLength]
         testPortIps = dutConfig["testPortIps"]
 
@@ -1616,6 +1625,7 @@ def testQosSaiDwrr(
                 RunAnsibleModuleFail if ptf test fails
         """
         portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        skip_test_on_no_lossless_pg(portSpeedCableLength)
         qosConfig = dutQosConfig["param"]
         if "wrr" in qosConfig[portSpeedCableLength]:
             qosConfigWrr = qosConfig[portSpeedCableLength]["wrr"]
@@ -1695,6 +1705,9 @@ def testQosSaiPgSharedWatermark(
         """
 
         portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        if pgProfile == "wm_pg_shared_lossless":
+            skip_test_on_no_lossless_pg(portSpeedCableLength)
+
         if pgProfile in list(dutQosConfig["param"][portSpeedCableLength].keys()):
             qosConfig = dutQosConfig["param"][portSpeedCableLength]
         else:
@@ -1798,6 +1811,7 @@ def testQosSaiPgHeadroomWatermark(
                 RunAnsibleModuleFail if ptf test fails
         """
         portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        skip_test_on_no_lossless_pg(portSpeedCableLength)
         if dutTestParams['hwsku'] in self.BREAKOUT_SKUS and 'backend' not in dutTestParams['topo']:
             qosConfig = dutQosConfig["param"][portSpeedCableLength]["breakout"]
         else:
@@ -1910,6 +1924,8 @@ def testQosSaiQSharedWatermark(
                 RunAnsibleModuleFail if ptf test fails
         """
         portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        if queueProfile == "wm_q_shared_lossless":
+            skip_test_on_no_lossless_pg(portSpeedCableLength)
 
         if queueProfile == "wm_q_shared_lossless":
             if dutTestParams["basicParams"]["sonic_asic_type"] == 'cisco-8000':
@@ -1972,7 +1988,7 @@ def testQosSaiQSharedWatermark(
 
     def testQosSaiDscpToPgMapping(
         self, get_src_dst_asic_and_duts, duthost, request, ptfhost, dutTestParams, dutConfig, dut_qos_maps,  # noqa F811
-            change_lag_lacp_timer):
+            change_lag_lacp_timer, dutQosConfig):
         """
             Test QoS SAI DSCP to PG mapping ptf test
 
@@ -1990,6 +2006,8 @@ def testQosSaiDscpToPgMapping(
                 RunAnsibleModuleFail if ptf test fails
         """
         disableTest = request.config.getoption("--disable_test")
+        portSpeedCableLength = dutQosConfig["portSpeedCableLength"]
+        skip_test_on_no_lossless_pg(portSpeedCableLength)
         if dutTestParams["basicParams"]["sonic_asic_type"] == 'cisco-8000' or \
                 ('platform_asic' in dutTestParams["basicParams"] and
                  dutTestParams["basicParams"]["platform_asic"] in ["broadcom-dnx", "mellanox"]):
diff --git a/tests/saitests/py3/sai_base_test.py b/tests/saitests/py3/sai_base_test.py
@@ -44,6 +44,7 @@
                                   "5": "scheduler.0",
                                   "6": "scheduler.0",
                                   "7": ""}
+BLOCK_DATA_PLANE_SCHEDULER_NAME = 'scheduler.block_data_plane'
 
 
 class ThriftInterface(BaseTest):
@@ -238,16 +239,15 @@ def disable_mellanox_egress_data_plane(self, ptf_port_list):
             dut_port = self.get_dut_port(ptf_port)
             dut_port_list.append(dut_port)
         self.original_dut_port_queue_scheduler_map = self.get_queue_scheduler_name(dut_port_list)
-        block_data_plane_scheduler_name = 'scheduler.block_data_plane'
         cmd_set_block_data_plane_scheduler = \
-            f'sonic-db-cli CONFIG_DB hset "SCHEDULER|{block_data_plane_scheduler_name}" "type" DWRR "weight" 15 "pir" 1'
+            f'sonic-db-cli CONFIG_DB hset "SCHEDULER|{BLOCK_DATA_PLANE_SCHEDULER_NAME}" "type" DWRR "weight" 15 "pir" 1'
 
         self.exec_cmd_on_dut(self.server, self.test_params['dut_username'], self.test_params['dut_password'],
                              cmd_set_block_data_plane_scheduler)
         for dut_port in dut_port_list:
             for q in DATA_PLANE_QUEUE_LIST:
                 cmd_block_q = \
-                    f" sonic-db-cli CONFIG_DB hset 'QUEUE|{dut_port}|{q}' scheduler {block_data_plane_scheduler_name}"
+                    f" sonic-db-cli CONFIG_DB hset 'QUEUE|{dut_port}|{q}' scheduler {BLOCK_DATA_PLANE_SCHEDULER_NAME}"
                 self.exec_cmd_on_dut(
                     self.server, self.test_params['dut_username'], self.test_params['dut_password'], cmd_block_q)
 
@@ -276,6 +276,20 @@ def enable_mellanox_egress_data_plane(self, ptf_port_list):
                 self.exec_cmd_on_dut(
                     self.server, self.test_params['dut_username'],
                     self.test_params['dut_password'], cmd_recover_q_scheduler_config)
+        self.remove_block_data_plan_scheduler()
+
+    def remove_block_data_plan_scheduler(self):
+        get_block_data_plane_scheduler_name = \
+            f"sonic-db-cli CONFIG_DB keys 'SCHEDULER|{BLOCK_DATA_PLANE_SCHEDULER_NAME}'"
+        scheduler_name, _, _ = self.exec_cmd_on_dut(self.server,
+                                                    self.test_params['dut_username'],
+                                                    self.test_params['dut_password'],
+                                                    get_block_data_plane_scheduler_name)
+        if isinstance(scheduler_name, list) and BLOCK_DATA_PLANE_SCHEDULER_NAME in scheduler_name[0]:
+            cmd_del_block_data_plane_scheduler = \
+                f'sonic-db-cli CONFIG_DB del "SCHEDULER|{BLOCK_DATA_PLANE_SCHEDULER_NAME}"'
+            self.exec_cmd_on_dut(self.server, self.test_params['dut_username'], self.test_params['dut_password'],
+                                 cmd_del_block_data_plane_scheduler)
 
 
 class ThriftInterfaceDataPlane(ThriftInterface):
diff --git a/tests/saitests/py3/sai_qos_tests.py b/tests/saitests/py3/sai_qos_tests.py
@@ -5209,7 +5209,7 @@ def runTest(self):
                 assert (q_wm_res[queue] <= (margin + 1) * cell_size)
             elif pkts_num_fill_min:
                 assert (q_wm_res[queue] == 0)
-            elif 'cisco-8000' in asic_type or "SN5600" in hwsku or "SN5400" in hwsku:
+            elif 'cisco-8000' in asic_type or "SN56" in hwsku or "SN5400" in hwsku:
                 assert (q_wm_res[queue] <= (margin + 1) * cell_size)
             else:
                 if platform_asic and platform_asic == "broadcom-dnx":
