[dts][PATCH V2 1/2] test_plans/large_vf: modify script to adapt to changes in dpdk

[dts][PATCH V2 1/2] test_plans/large_vf: modify script to adapt to changes in dpdk

[Hongbo Li]

1.after dpdk is modified, if all ports fail to start, dpdk still allows testpmd to enter the command line, instead of unconditionally terminating testpmd.
According to dpdk commit 7e40372522c("app/testpmd: fix interactive mode with no ports").

2.after driver is modified, the default RX rings available decreased(now from SW4.2 default number of queues are created according to the single NUMA CPU cores), so modify script to adapt.

Signed-off-by: Hongbo Li <hongbox.li@intel.com>
---
V2: modify the comments

 test_plans/large_vf_test_plan.rst | 18 +++++++++++-------
 1 file changed, 11 insertions(+), 7 deletions(-)

diff --git a/test_plans/large_vf_test_plan.rst b/test_plans/large_vf_test_plan.rst
index e043a660..0f2318de 100644
--- a/test_plans/large_vf_test_plan.rst
+++ b/test_plans/large_vf_test_plan.rst
@@ -264,25 +264,29 @@ Subcase 5: PF fdir + large VF fdir co-exist
 
 Start testpmd on VF0 with 256 queues.
 
-Create 10 rules on PF0, queue from [54~63]::
+Get RX rings available on DUT ::
 
-    ethtool -N enp96s0f0 flow-type udp4 dst-ip 192.168.0.21 src-port 22 action 63
-    ethtool -N enp96s0f0 flow-type udp4 dst-ip 192.168.0.22 src-port 22 action 62
-    ethtool -N enp96s0f0 flow-type udp4 dst-ip 192.168.0.23 src-port 22 action 61
+    ethtool -n enp96s0f0
+
+Create 10 rules on PF0, and the actions decrease from the maximum value of RX rings available::
+
+    ethtool -N enp96s0f0 flow-type udp4 dst-ip 192.168.0.21 src-port 22 action 35
+    ethtool -N enp96s0f0 flow-type udp4 dst-ip 192.168.0.22 src-port 22 action 34
+    ethtool -N enp96s0f0 flow-type udp4 dst-ip 192.168.0.23 src-port 22 action 33
     ...
-    ethtool -N enp96s0f0 flow-type udp4 dst-ip 192.168.0.30 src-port 22 action 54
+    ethtool -N enp96s0f0 flow-type udp4 dst-ip 192.168.0.30 src-port 22 action 26
 
 Check rules on PF::
 
     ethtool -n enp96s0f0
 
-Send matched patches to PF::
+Send matched packet to PF::
 
     pkt1=Ether(dst="00:00:00:00:01:00")/IP(src=RandIP(),dst="192.168.0.21")/UDP(sport=22,dport=23)/Raw('x'*80)
     ......
     pkt10=Ether(dst="00:00:00:00:01:00")/IP(src=RandIP(),dst="192.168.0.30")/UDP(sport=22,dport=23)/Raw('x'*80)
 
-Check PF matched queue [54~63] could receive matched packet::
+Check PF matched queue could receive matched packet::
 
     ethtool -S enp96s0f0
 
-- 
2.17.1

[PATCH V2 2/2] tests/large_vf: modify script to adapt to changes in dpdk

[Hongbo Li]

1.after dpdk is modified, if all ports fail to start, dpdk still allows testpmd to enter the command line, instead of unconditionally terminating testpmd.
According to dpdk commit 7e40372522c("app/testpmd: fix interactive mode with no ports").

2.after driver is modified, the default RX rings available decreased(now from SW4.2 default number of queues are created according to the single NUMA CPU cores), so modify script to adapt.

Signed-off-by: Hongbo Li <hongbox.li@intel.com>
---
 tests/TestSuite_large_vf.py | 23 +++++++++++++++--------
 1 file changed, 15 insertions(+), 8 deletions(-)

diff --git a/tests/TestSuite_large_vf.py b/tests/TestSuite_large_vf.py
index 4e2ff1d6..fb03145e 100644
--- a/tests/TestSuite_large_vf.py
+++ b/tests/TestSuite_large_vf.py
@@ -110,8 +110,7 @@ different_queues_switch = {
 
 pf_large_vf_fdir_coexist = {
     "name": "test_pf_large_vf_fdir_coexist",
-    "param": [21, 63],
-    "check_param": (54, 63),
+    "param": [21],
     "count": 1,
 }
 
@@ -124,7 +123,7 @@ exceed_256_queues = {
 more_than_3_vfs_256_queues = {
     "name": "test_more_than_3_vfs_256_queues",
     "param": "--txq=256 --rxq=256",
-    "check_param": "Cause: Start ports failed",
+    "check_param": "Start ports failed",
 }
 
 max_vfs_256_queues_3 = [
@@ -296,6 +295,7 @@ class TestLargeVf(TestCase):
 
     def rte_flow_process(self, vectors):
         test_results = {}
+        self.rule_num = []
         for tv in vectors:
             try:
                 subcase_name = tv["name"]
@@ -322,14 +322,14 @@ class TestLargeVf(TestCase):
                         self.pmdout_list[0],
                         self.pf0_intf,
                         tv["param"][0],
-                        tv["param"][1],
+                        self.max_ring_num - 1,
                     )
                     self.send_pkts_pf_check(
                         self.pmdout_list[0],
                         self.pf0_intf,
                         self.pf0_mac,
                         tv["param"][0],
-                        tv["check_param"],
+                        self.rule_num,
                         tv["count"],
                     )
                     self.create_fdir_rule(vectors[0]["rule"])
@@ -373,11 +373,14 @@ class TestLargeVf(TestCase):
                                 + "-- -i "
                                 + tv["param"]
                             )
-                            out = self.pmd_output.execute_cmd(cmd, "#")
+                            out = self.pmd_output.execute_cmd(
+                                cmd, "testpmd> ", timeout=30
+                            )
                             self.verify(
                                 tv["check_param"] in out,
                                 "fail: testpmd start successfully",
                             )
+                            self.pmd_output.quit()
                             self.pmdout_list[0].execute_cmd("quit", "# ")
                             break
                         else:
@@ -399,11 +402,12 @@ class TestLargeVf(TestCase):
                                     + "-- -i "
                                     + tv["param"]
                                 )
-                                out = self.pmd_output.execute_cmd(cmd, "#")
+                                out = self.pmd_output.execute_cmd(cmd, "testpmd> ")
                                 self.verify(
                                     tv["check_param"] in out,
                                     "fail: testpmd start successfully",
                                 )
+                                self.pmd_output.quit()
                                 # quit all testpmd
                                 self.pmdout_list[0].execute_cmd("quit", "# ")
                                 self.pmdout_list[1].execute_cmd("quit", "# ")
@@ -515,6 +519,7 @@ class TestLargeVf(TestCase):
                 pf_intf, ip, action
             )
             pmdout.execute_cmd(cmd, "#")
+            self.rule_num.append(action)
             ip += 1
             action -= 1
         self.validation_pf_rule(pmdout, pf_intf, 10)
@@ -534,7 +539,7 @@ class TestLargeVf(TestCase):
             ip += 1
         time.sleep(1)
         out = pmdout.execute_cmd("ethtool -S %s" % pf_intf, "# ")
-        for queue in range(check_param[0], check_param[1] + 1):
+        for queue in check_param:
             packet_str = "rx_queue_%d_packets: (\d+)" % queue
             packet = re.search(packet_str, out).group(1)
             self.verify(
@@ -544,8 +549,10 @@ class TestLargeVf(TestCase):
 
     def destroy_pf_rule(self, pmdout, pf_intf):
         rule_str = "Filter:.*?(\d+)"
+        max_ring_str = "(\d+) RX rings available"
         out = pmdout.execute_cmd("ethtool -n %s" % pf_intf, "#")
         rule_list = re.findall(rule_str, out)
+        self.max_ring_num = int(re.search(max_ring_str, out).group(1))
         if rule_list:
             for rule in rule_list:
                 cmd = "ethtool -N {} delete {}".format(pf_intf, rule)
-- 
2.17.1

[PATCH V2 2/2] tests/large_vf: modify script to adapt to changes in dpdk

[lijuan.tu]

On Thu,  9 Feb 2023 18:46:33 +0800, Hongbo Li <hongbox.li@intel.com> wrote:
> 1.after dpdk is modified, if all ports fail to start, dpdk still allows testpmd to enter the command line, instead of unconditionally terminating testpmd.
> According to dpdk commit 7e40372522c("app/testpmd: fix interactive mode with no ports").
> 
> 2.after driver is modified, the default RX rings available decreased(now from SW4.2 default number of queues are created according to the single NUMA CPU cores), so modify script to adapt.
> 
> Signed-off-by: Hongbo Li <hongbox.li@intel.com>


Series applied, thanks