Bug 3954 - [ANCK 4.19] Anolis 8.6系统使用intel提供的ice第三方驱动,网卡为intel的E810网卡,执行配置网卡多队列命令后,网卡显示down的状态。
Summary: [ANCK 4.19] Anolis 8.6系统使用intel提供的ice第三方驱动,网卡为intel的E810网卡,执行配置网卡多队列命令后,网卡显示d...
Status: CONFIRMED
Alias: None
Product: ANCK 4.19 Dev
Classification: ANCK
Component: net (show other bugs) net
Version: 4.19-026.x
Hardware: All Linux
: P3-Medium S3-normal
Target Milestone: ---
Assignee: gumi
QA Contact: shuming
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2023-02-06 19:18 UTC by songkai
Modified: 2023-05-05 09:46 UTC (History)
5 users (show)

See Also:


Attachments
图片 (1001.17 KB, image/jpeg)
2023-02-06 19:18 UTC, songkai
Details
日志 (311.45 KB, image/jpeg)
2023-02-20 20:08 UTC, songkai
Details

Note You need to log in before you can comment on or make changes to this bug.
Description songkai inspur_group 2023-02-06 19:18:02 UTC
Created attachment 622 [details]
图片

问题描述:

Anolis 8.6系统使用intel提供的ice第三方驱动,网卡为intel的E810网卡,执行配置网卡多

队列命令后,网卡显示down的状态,再执行systemctl restart network后,网卡显示up状态

,如果执行rmmod ice ,随后执行modprobe ice,此现象无法复现。可见附件图片。
Comment 1 songkai inspur_group 2023-02-20 20:08:08 UTC
Created attachment 643 [details]
日志
Comment 2 songkai inspur_group 2023-02-20 20:08:51 UTC
OS:       Anolis 8.6
kernel:   4.19
网卡型号:Intel Corporation Ethernet Controller E810-XXV

复现步骤:
1、执行命令:ethtool -L enp33s0f0 combined 8
2、ip a查看网卡状态
发现enp33s0f0变为down状态
3、执行systemctl restart network
4、发现enp33s0f0变为up状态

以上步骤可以重复复现

5、执行rmmod ice & insmod ice

发现如上的问题不再复现。

dmesg日志如图
Comment 3 YvaYang intel_group 2023-02-21 10:58:31 UTC
@songkai, would you please help to upload the below information to IPS#00785343 for further support from Intel network engineer? Thanks.

os and version? (cat /etc/*release*)
driver version? (ethtool -i ethxxx)
nvm image e-track id? ( ethtool -i ethxxx)
configuration of the port in os? (/etc/systemconf/network-script/if-xxxx)
has NetworkManager run or not?
any logs in dmesg?
Comment 4 gumi alibaba_cloud_group 2023-03-16 20:18:22 UTC
patch ab4ab73fc1ec引入的bug,ice driver在设置link-down-on-close flag后在ice_down()函数中会误将端口down掉,
Fix patch:8ac7132704f3fbd2095abb9459e5303ce8c9e559, 打此patch测试95次ethtool 命令问题验证通过

另外,bugzilla log中warning “Affinity broker due to vetctor space exhaustion”与此问题无关,该warning的原因是
执行此问题的ethtool会重新分配queue以及queue对应的中断,在中断affinity流程中触发了apic的一个bug产生的warning,修复
此warning的patch为:190113b4c6531c8e09b31d5235f9b5175cbb0f72
Comment 5 gumi alibaba_cloud_group 2023-05-05 09:46:32 UTC
已修复