Re: [dpdk-dev] [PATCH V2] doc: add known PHY link up issue for i40e

["Zhang, Qi Z" <qi.z.zhang@intel.com>]

> -----Original Message-----
> From: Guo, Jia
> Sent: Thursday, November 8, 2018 7:22 PM
> To: Zhang, Qi Z <qi.z.zhang@intel.com>; Xing, Beilei <beilei.xing@intel.com>;
> Kovacevic, Marko <marko.kovacevic@intel.com>; Mcnamara, John
> <john.mcnamara@intel.com>
> Cc: dev@dpdk.org; Zhang, Helin <helin.zhang@intel.com>; Wu, Jingjing
> <jingjing.wu@intel.com>
> Subject: Re: [PATCH V2] doc: add known PHY link up issue for i40e
> 
> 
> On 11/9/2018 1:48 AM, Zhang, Qi Z wrote:
> >
> >> -----Original Message-----
> >> From: Guo, Jia
> >> Sent: Thursday, November 8, 2018 12:18 AM
> >> To: Xing, Beilei <beilei.xing@intel.com>; Kovacevic, Marko
> >> <marko.kovacevic@intel.com>; Mcnamara, John
> >> <john.mcnamara@intel.com>; Zhang, Qi Z <qi.z.zhang@intel.com>
> >> Cc: dev@dpdk.org; Guo, Jia <jia.guo@intel.com>; Zhang, Helin
> >> <helin.zhang@intel.com>; Wu, Jingjing <jingjing.wu@intel.com>
> >> Subject: [PATCH V2] doc: add known PHY link up issue for i40e
> >>
> >> If bind i40e NICs with PMD and run application, after application
> >> quits, some kernel driver may not be able to set PHY to link up again
> >> by ``ifconfig`` after NICs be rebound to kernel driver.
> >>
> >> Signed-off-by: Jeff Guo <jia.guo@intel.com>
> >> ---
> >> v2->v1:
> >> refine some typo and word.
> >> ---
> >>   doc/guides/rel_notes/known_issues.rst | 27
> >> +++++++++++++++++++++++++++
> >>   1 file changed, 27 insertions(+)
> >>
> >> diff --git a/doc/guides/rel_notes/known_issues.rst
> >> b/doc/guides/rel_notes/known_issues.rst
> >> index 95e4ce6..f4e3519 100644
> >> --- a/doc/guides/rel_notes/known_issues.rst
> >> +++ b/doc/guides/rel_notes/known_issues.rst
> >> @@ -759,3 +759,30 @@ Netvsc driver and application restart
> >>
> >>   **Driver/Module**:
> >>      ``uio_hv_generic`` module.
> >> +
> >> +
> >> +PHY link up failed when rebind i40e NICs to kernel driver
> >> +---------------------------------------------------------
> >> +
> >> +**Description**:
> >> +   If bind i40e NICs with PMD and run application, after application
> >> +quits,
> >> some
> >> +   kernel driver may not be able to set PHY to link up again by
> ``ifconfig``
> >> +   after NICs be rebound to kernel driver.
> > Should we switch the content of "description" and "implication"?
> 
> 
> I think **Description** is aim to explain the meaning of the title and give the
> detail of the phenomenon to acknowledge it.

That's similar as what I understand.

> 
> And **Implication** is what is the true thing behind the phenomenon and
> show which version will be expected to fix. I believe

My understand for **implication** is, it should include the information about the result and impact due to the issue.

Here the issue is:
  Some kernel driver not able to handle link status correctly after DPDK app set PHY to down."

The implication is:
  Link can't be turn up after NIC be rebound to kernel driver.

So the word below looks like the result that user will suffer by this issue, so I think it should be included in implication section.

	If bind i40e NICs with PMD and run application, after application quits,
	some kernel driver may not be able to set PHY to link up again by `ifconfig`` after NICs be rebound to kernel driver
	
while the "true thing behind the phenomenon" could still be included in description.

Anyway, I'm not expert for this, I can be totally wrong, its good chance that someone can help to explain for this :)

Thanks
Qi

> 
> other part in document is the same and it will no have a big gap here.
> If you find the problem, please give your solid reason here.
> 
> Anyway, i think it should be fine and want you could accept my explanation.
> 
> 
> >> +
> >> +**Implication**:
> >> +   Before application quits, it will invokes the function
> ``i40e_dev_stop``,
> >> +   which will set PHY to link down. This is the situation that some kernel
> >> +   driver may not be able to handle link status correctly after it retakes
> >> +   control of the device. It is a known PHY link configure
> >> +synchronous issue
> >> in
> >> +   i40e kernel driver. The fix has been addressed in 2.7.4 rc version. So
> if
> >> +   the i40e kernel driver is < 2.7.4 and before the fix be
> >> + backported into
> >> the
> >> +   previous versions, it will encounter this issue.
> >> +
> >> +**Resolution/Workaround**:
> >> +   If it does not help by remove and reinsert i40e kernel driver,
> >> +please
> >> reboot system.
> >> +
> >> +**Affected Environment/Platform**:
> >> +   All.
> >> +
> >> +**Driver/Module**:
> >> +   Poll Mode Driver (PMD).
> >> --
> >> 2.7.4