I have frequent link down trouble in our customer factory but can't re-produce in our bench.
The frequent link down port uses the 82579 nic and I found out the below issue.
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/drivers/net/ethernet/intel/e1... kernel/git/torvalds/linux.git - Linux kernel source tree
It seems the relaxation of link down condition, but can't understand detail work around.
The modifed registers lost in 82579's specification, So, please share the information.
* MSE
what of MSE? how condition leads link down?
* original valule
please share the threshold and counter's original values and meenings.
Thanks,
Link Copied
Hi Ksugiyama,
Thank you for the post. I will have to further check on this. For the meantime, can you share
-what is the exact Linux OS version use
- what brand and model is your board?
-driver version (from Intel or the board vendor)
Thanks,
wb
Hi wb,
Thank you for your reply. and I'm sorry to lack the environment information.
- OS : Windows7sp1
- PC : CONTEC Custom ( equivalet of EPC2000 series )
- driver : 11.8.84
So, our environment windows. Because I heard the driver code base is same between windows and linux, I investigated linux driver modification history.
Thanks,
Ksugiyama
Hi Ksugiyama,
If you are using Windows environment, can you provide Windows detail e.g the event log? As Linux is different from Windows.
rgds,
wb
Hi Ksugiyama,
Thank you for the information. I will check how this can be done if needed. The log file did not show much information.
I will update you.
Thanks,
wb
Hi Ksugiyama,
Good day. If you are not able to reproduce the issue at your end, you
may check the switch used by your client.
Please help provide if there is any recent changes made when the issue occured,
what are the recent changes done on the system?
We have known issue of 82579 with Netgear XSM7224s switch wherein the link
flap/disconnection occurs when EEE mode is enabled, the solution is to disable
that function. You may refer to the README.TXT for reference:
https://downloadmirror.intel.com/22283/eng/readme.txt
The MSE stands for Mean Square Error which setting it higher would enable the link to stay up when noise is high.
The driver version 21.1 comes with higher driver version 12.15.31.4 for 82579 which is dated 3/29/16. You can check with the system board vendor for similar driver version
from the board vendor to test.
Hope the above information help.
Rgds,
wb
Hi wb,
Thank you for your information.
* switch
Yes, we suspect our switch,too. Although we apply EEE disable setting to driver, It's not effect for link flapping.
Our Switch Type is CentreCom GS924L V2 and we think the trial of switch replacement after checking driver update result.
* recent chage or cause event
In our fab, it isn't reproduced and as far as I know, it isn't caused by any changed event.
* MSE & driver information
My understanding of above patch is relaxing the linkdown condition by noise level (MSE threshold), collect?
Thanks to share the driver update information. we confirm board vendor.
Thanks,
Best Regards,
Ksugiyama
Hi Ksugiyama,
Thank you for the information. Please feel free to update me the result if you have confirm with the board vendor driver.Thanks.
rgds,
wb
Hello wb,
Good news!
Our field engineers report me the result of driver updating (to 12.15.30.0).
Before updating, link flapping occurs a machine / day in hundreds machines. After updating, although tempolary link down occurs, any link flapping didn't occur in several weeks.
Thank you!
I think this issue's root cause is around relaxing MSE setting, So, how did you know the confirmation method of linkdown reason?
Thanks,
Best Regards.
Ksugiyama
Hi Ksugiyama,
This is indeed a good news and I am glad for you. We do not have specific information but the Mean Square Error is the setting in which setting it higher would enable the link to stay up when noise is high. Further, it is recommended to use latest driver for network connection. For the meantime, any further assistance needed? Thanks.
rgds,
wb
Hi wb,
I see. And I appreciate for your assistance. we'll have been checking customer's machine status for a while.
Thanks.
Best Regards,
Ksugiyama
Hi Ksugiyama,
You are welcome. Please feel free to update us in case you need further assistance from us.
rgds,
wb
For more complete information about compiler optimizations, see our Optimization Notice.