The dpdk 18.11 version is on the i40e network card and the overcurrent 1G execution results in a high occupation of 0 cores

???? Tue, 07 Dec 2021 07:41:32 -0800

Hello, I used the ip_pipeline program of 18.11 dpdk and made it flow 1GB/s. I 
found that the 0-core CPU occupies very high (usleep has been added in the for 
loop of the ip_pipeline main function, which should not occupy very high 
CPU).Use perf to view and find that it is occupied in dw_readl.



perf situation:
&nbsp; &nbsp; -&nbsp; &nbsp;61.60%&nbsp; &nbsp; 61.60%&nbsp; [kernel]&nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;[k] dw_readl&nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;
&nbsp;&nbsp;
&nbsp; &nbsp; + 5.89% ret_from_fork&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;
&nbsp; &nbsp; &nbsp; 1.57% ordered_events__flush&nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;
&nbsp; &nbsp; + 1.25% 0xe1&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;
&nbsp; &nbsp; + 1.13% x86_64_start_kernel&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;
&nbsp; &nbsp; &nbsp; 1.02%&nbsp; &nbsp; &nbsp;hist_entry_iter__add&nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;
&nbsp; &nbsp; + 1.02% 0xd1&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;
&nbsp; &nbsp; +&nbsp; &nbsp;22.14%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[kernel]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;[k] 
__irqentry_text_start&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp;
&nbsp; &nbsp; +&nbsp; &nbsp;22.11%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[kernel]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;[k] handle_irq&nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;
&nbsp; &nbsp; +&nbsp; &nbsp;22.08%&nbsp; &nbsp; &nbsp;0.01%&nbsp; 
[kernel]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;[k] 
handle_irq_event_percpu&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; &nbsp; &nbsp;
&nbsp; &nbsp; +&nbsp; &nbsp;19.31%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[unknown]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; [k] 0xffffffff8168dd6d&nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp;&nbsp;
&nbsp; &nbsp; +&nbsp; &nbsp;19.30%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[unknown]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; [k] 0xffffffff81698bef&nbsp; 
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp;&nbsp;
&nbsp; &nbsp; +&nbsp; &nbsp;19.28%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[unknown]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; [k] 0xffffffff8102d26
&nbsp; &nbsp; +&nbsp; &nbsp;19.28%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[unknown]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; [k] 0xffffffff811337d
&nbsp; &nbsp; +&nbsp; &nbsp;19.27%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[unknown]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; [k] 0xffffffff8113033
&nbsp; &nbsp; +&nbsp; &nbsp;19.25%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[unknown]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; [k] 0xffffffff8113015
&nbsp; &nbsp; +&nbsp; &nbsp;17.14%&nbsp; &nbsp; &nbsp;0.02%&nbsp; 
[kernel]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;[k] i2c_dw_isr&nbsp; 
&nbsp; &nbsp;&nbsp;
&nbsp; &nbsp; +&nbsp; &nbsp;14.50%&nbsp; &nbsp; &nbsp;0.00%&nbsp; 
[unknown]&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; [k] 0xffffffffa04f406


This is my hardware information:


&nbsp; &nbsp; CPU model : Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz
&nbsp; &nbsp; Network card model and version : i40e 1.5.10-k, two card
&nbsp; &nbsp; Network card firmware : 6.01 0x8000372b 0.0.0
&nbsp; &nbsp; lscpi view network card information:
&nbsp; &nbsp; &nbsp; &nbsp; 03:00.0 Ethernet controller: Intel Corporation 
Ethernet Controller X710 for 10GbE SFP+ (rev 02)
&nbsp; &nbsp; &nbsp; &nbsp; 03:00.1 Ethernet controller: Intel Corporation 
Ethernet Controller X710 for 10GbE SFP+ (rev 02)


&nbsp; &nbsp; Network card driver use: vfio-pci


Modified ip_pipeline main function :


&nbsp; &nbsp; &nbsp; &nbsp; /* Script */
&nbsp; &nbsp; &nbsp; &nbsp; if (app.script_name)
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; cli_script_process(app.script_name,
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; app.conn.msg_in_len_max,
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
app.conn.msg_out_len_max);
&nbsp; &nbsp; &nbsp; &nbsp; #include <unistd.h&gt;
&nbsp; &nbsp; &nbsp; &nbsp; /* Dispatch loop */
&nbsp; &nbsp; &nbsp; &nbsp; for ( ; ; ) {
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; conn_poll_for_conn(conn);
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; usleep(5000);
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; conn_poll_for_msg(conn);
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; kni_handle_request();
&nbsp; &nbsp; &nbsp; &nbsp; }


CLI startup parameters:
&nbsp; &nbsp; ./ip_pipeline -c f -- -s ./l2fwd.cli


CLI content:


&nbsp; &nbsp; mempool MEMPOOL0 buffer 2304 pool 32K cache 256 cpu 0


&nbsp; &nbsp; link LINK0 dev 0000:03:00.1 rxq 1 128 MEMPOOL0 txq 1 512 
promiscuous on
&nbsp; &nbsp; link LINK1 dev 0000:03:00.0 rxq 1 128 MEMPOOL0 txq 1 512 
promiscuous on
&nbsp; &nbsp; ;link LINK2 dev 0000:06:00.0 rxq 1 128 MEMPOOL0 txq 1 512 
promiscuous on
&nbsp; &nbsp; ;link LINK3 dev 0000:06:00.1 rxq 1 128 MEMPOOL0 txq 1 512 
promiscuous on


&nbsp; &nbsp; pipeline PIPELINE0 period 10 offset_port_id 0 cpu 0


&nbsp; &nbsp; pipeline PIPELINE0 port in bsz 32 link LINK0 rxq 0
&nbsp; &nbsp; pipeline PIPELINE0 port in bsz 32 link LINK1 rxq 0
&nbsp; &nbsp; ;pipeline PIPELINE0 port in bsz 32 link LINK2 rxq 0
&nbsp; &nbsp; ;pipeline PIPELINE0 port in bsz 32 link LINK3 rxq 0


&nbsp; &nbsp; pipeline PIPELINE0 port out bsz 32 link LINK0 txq 0
&nbsp; &nbsp; pipeline PIPELINE0 port out bsz 32 link LINK1 txq 0
&nbsp; &nbsp; ;pipeline PIPELINE0 port out bsz 32 link LINK2 txq 0
&nbsp; &nbsp; ;pipeline PIPELINE0 port out bsz 32 link LINK3 txq 0


&nbsp; &nbsp; pipeline PIPELINE0 table match stub
&nbsp; &nbsp; pipeline PIPELINE0 table match stub
&nbsp; &nbsp; ;pipeline PIPELINE0 table match stub
&nbsp; &nbsp; ;pipeline PIPELINE0 table match stub


&nbsp; &nbsp; pipeline PIPELINE0 port in 0 table 0
&nbsp; &nbsp; pipeline PIPELINE0 port in 1 table 1
&nbsp; &nbsp; ;pipeline PIPELINE0 port in 2 table 2
&nbsp; &nbsp; ;pipeline PIPELINE0 port in 3 table 3


&nbsp; &nbsp; thread 1 pipeline PIPELINE0 enable


&nbsp; &nbsp; pipeline PIPELINE0 table 0 rule add match default action fwd port 
1
&nbsp; &nbsp; pipeline PIPELINE0 table 1 rule add match default action fwd port 0
&nbsp; &nbsp; ;pipeline PIPELINE0 table 2 rule add match default action fwd 
port 3
&nbsp; &nbsp; ;pipeline PIPELINE0 table 3 rule add match default action fwd 
port 2


I have done these tests


&nbsp; &nbsp; 1. Using the i40e device of E5-2620 CPU to perform the same test, 
there is no such problem


&nbsp; &nbsp; 2. I still use the i7-6700 i40e device, and I rolled back to the 
18.05 version for testing, there is no such problem.


&nbsp; &nbsp; 3. The fallback code locates the code node that causes this 
problem as 4205c7ccec4fc2aeafe3e7ccf6b028d9476fccaf, this node had no problems 
before, and there were problems afterwards.


&nbsp; &nbsp; 4. I found that there are two kernel modules on this i7-6700 
device: idma64, i2c_designware, when I uninstall these two modules, this 
phenomenon will not appear.


&nbsp; &nbsp; 5. I found that when there is a problem, the call frequency of an 
interrupt reaches tens of thousands per second! When there is no problem, the 
call frequency of this interrupt is 0.


&nbsp; &nbsp; 6. On the basis of 18.11, in the i40e_dev_start function


&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if 
(dev-&gt;data-&gt;dev_conf.intr_conf.rxq == 0) {
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
rte_eal_alarm_set(I40E_ALARM_INTERVAL,
&nbsp; &nbsp;
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; i40e_dev_alarm_handler, dev);
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; } else {
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; /* enable uio intr 
after callback register */
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
rte_intr_enable(intr_handle);
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; }


&nbsp; &nbsp; &nbsp; &nbsp; Change to


&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; /* enable uio intr after callback 
register */
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; rte_intr_enable(intr_handle);


&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if 
(dev-&gt;data-&gt;dev_conf.intr_conf.rxq == 0) {
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
rte_eal_alarm_set(I40E_ALARM_INTERVAL,
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; 
&nbsp; i40e_dev_alarm_handler, dev);
&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; }


&nbsp; &nbsp; &nbsp; &nbsp; Then this problem will not occur.


From the above scene, if I want to use the 18.11 dpdk version normally, I will 
either uninstall the module or add rte_intr_enable, but I don??t know what will 
happen if I do this? Can you answer me? Is this a bug?

The dpdk 18.11 version is on the i40e network card and the overcurrent 1G execution results in a high occupation of 0 cores

Reply via email to