[dpdk-dev] [PATCH] net/ixgbe: fix busy polling while fiber link update

Zhao1, Wei wei.zhao1 at intel.com
Tue Oct 16 10:29:22 CEST 2018


Hi, Laurent Hardy

From: Laurent Hardy [mailto:laurent.hardy at 6wind.com]
Sent: Monday, October 15, 2018 6:43 PM
To: Zhao1, Wei <wei.zhao1 at intel.com>; Ilya Maximets <i.maximets at samsung.com>; Zhang, Qi Z <qi.z.zhang at intel.com>
Cc: Lu, Wenzhuo <wenzhuo.lu at intel.com>; Ananyev, Konstantin <konstantin.ananyev at intel.com>; stable at dpdk.org; dev at dpdk.org
Subject: Re: [dpdk-dev] [PATCH] net/ixgbe: fix busy polling while fiber link update


Hi Wei,
On 10/12/2018 09:36 AM, Zhao1, Wei wrote:

Hi, laurent.hardy

  You are the author for the patch (* net/ixgbe: ensure link status is

updated), why do you implement code that way?

Is that must that  set up link with wait?



ixgbe_setup_link(hw, speed, true);



The main issue which has lead to this patch has been reported through a test

case with the autoneg enabled (which has been also reported in the pmd

test provided along with the patch:

http://patches.dpdk.org/comment/46253/).

In this context, without the flag set the patch wasn't effective.



My question is whether we can change to:



ixgbe_setup_link(hw, speed, false);



in your patch, it is " ixgbe_setup_link(hw, speed, true);"

some user may  need flexible in wait for complete.




Yes, the parameter (autoneg_wait_to_complete) could be change to false.
I redo the test following test plan provided through http://patches.dpdk.org/comment/46253/, with flag set to false
and speed defined to 1Gb on switch side.
In this case both ports goes properly up (nic used is 82599ES 10-Gigabit) with a correct link speed.

Thank you for your test and feedback!


More information about the dev mailing list