[2/2] doc: add the description for option pflink_fullchk in ixgbevf
Checks
Commit Message
Add the detail description for this pflink_fullchk option, when it will
be used, and it is used for fixing what kind of issue.
Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
---
doc/guides/nics/ixgbe.rst | 18 ++++++++++++++++++
1 file changed, 18 insertions(+)
Comments
Acked-by: Scott Daniels daniels@research.att.com
On Thu, 6 Jun 2019, Haiyue Wang wrote:
> Add the detail description for this pflink_fullchk option, when it will
> be used, and it is used for fixing what kind of issue.
>
> Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
On 07/06/2019 04:10, Haiyue Wang wrote:
> Add the detail description for this pflink_fullchk option, when it will
> be used, and it is used for fixing what kind of issue.
>
You can squash the docs into the 1/2 patch.
> Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
> ---
> doc/guides/nics/ixgbe.rst | 18 ++++++++++++++++++
> 1 file changed, 18 insertions(+)
>
> diff --git a/doc/guides/nics/ixgbe.rst b/doc/guides/nics/ixgbe.rst
> index 975143f..33812a5 100644
> --- a/doc/guides/nics/ixgbe.rst
> +++ b/doc/guides/nics/ixgbe.rst
> @@ -82,6 +82,24 @@ To guarantee the constraint, capabilities in dev_conf.rxmode.offloads will be ch
>
> fdir_conf->mode will also be checked.
>
> +VF Runtime Options
> +^^^^^^^^^^^^^^^^^^
> +
> +The following ``devargs`` options can be enabled at runtime. They must
> +be passed as part of EAL arguments. For example,
> +
> +.. code-block:: console
> +
> + testpmd -w af:10.0,pflink_fullchk=1 -- -i
> +
> +- ``pflink_fullchk`` (default **0**)
> +
> + Toggle behavior to get ixgbevf link status quickly by checking the
> + mailbox status or not. If set, the VF will check the PF's link status
> + and the PF's mailbox running status, which will trigger PF's mailbox
> + interrupt generation. Otherwise, just check PF's link status, then no
> + mailbox interrupt in PF.
> +
wait_to_complete is also considered, so doesn't it mean that
pflink_fullck is not a toggle in all cases e.g. rte_eth_link_get() will
set wait_to_complete=1 and then pflink_fullchk is ignored. At least you
should mention if there is some user visible API like this where the
flag will not result in the behaviour being toggled.
> RX Burst Size
> ^^^^^^^^^^^^^
>
>
> -----Original Message-----
> From: Kevin Traynor [mailto:ktraynor@redhat.com]
> Sent: Friday, June 7, 2019 20:25
> To: Wang, Haiyue <haiyue.wang@intel.com>; dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>; Lu,
> Wenzhuo <wenzhuo.lu@intel.com>; Wang, Liang-min <liang-min.wang@intel.com>; daniels@research.att.com
> Subject: Re: [dpdk-dev] [PATCH 2/2] doc: add the description for option pflink_fullchk in ixgbevf
>
> On 07/06/2019 04:10, Haiyue Wang wrote:
> > Add the detail description for this pflink_fullchk option, when it will
> > be used, and it is used for fixing what kind of issue.
> >
>
> You can squash the docs into the 1/2 patch.
>
Yes, I've sent a single patch before: https://patches.dpdk.org/patch/54374/
After some quiet time, I found that most of commits about doc are a single
patch, so I split it into two.
> > Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
> > ---
> > doc/guides/nics/ixgbe.rst | 18 ++++++++++++++++++
> > 1 file changed, 18 insertions(+)
> >
> > diff --git a/doc/guides/nics/ixgbe.rst b/doc/guides/nics/ixgbe.rst
> > index 975143f..33812a5 100644
> > --- a/doc/guides/nics/ixgbe.rst
> > +++ b/doc/guides/nics/ixgbe.rst
> > @@ -82,6 +82,24 @@ To guarantee the constraint, capabilities in dev_conf.rxmode.offloads will be ch
> >
> > fdir_conf->mode will also be checked.
> >
> > +VF Runtime Options
> > +^^^^^^^^^^^^^^^^^^
> > +
> > +The following ``devargs`` options can be enabled at runtime. They must
> > +be passed as part of EAL arguments. For example,
> > +
> > +.. code-block:: console
> > +
> > + testpmd -w af:10.0,pflink_fullchk=1 -- -i
> > +
> > +- ``pflink_fullchk`` (default **0**)
> > +
> > + Toggle behavior to get ixgbevf link status quickly by checking the
> > + mailbox status or not. If set, the VF will check the PF's link status
> > + and the PF's mailbox running status, which will trigger PF's mailbox
> > + interrupt generation. Otherwise, just check PF's link status, then no
> > + mailbox interrupt in PF.
> > +
>
> wait_to_complete is also considered, so doesn't it mean that
> pflink_fullck is not a toggle in all cases e.g. rte_eth_link_get() will
> set wait_to_complete=1 and then pflink_fullchk is ignored. At least you
> should mention if there is some user visible API like this where the
> flag will not result in the behaviour being toggled.
>
I use "get ixgbevf link status -quickly-" to describe "wait_to_complete = 0",
maybe I need some code to make the information more clear.
> > RX Burst Size
> > ^^^^^^^^^^^^^
> >
> >
On 07/06/2019 14:55, Wang, Haiyue wrote:
>> -----Original Message-----
>> From: Kevin Traynor [mailto:ktraynor@redhat.com]
>> Sent: Friday, June 7, 2019 20:25
>> To: Wang, Haiyue <haiyue.wang@intel.com>; dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>; Lu,
>> Wenzhuo <wenzhuo.lu@intel.com>; Wang, Liang-min <liang-min.wang@intel.com>; daniels@research.att.com
>> Subject: Re: [dpdk-dev] [PATCH 2/2] doc: add the description for option pflink_fullchk in ixgbevf
>>
>> On 07/06/2019 04:10, Haiyue Wang wrote:
>>> Add the detail description for this pflink_fullchk option, when it will
>>> be used, and it is used for fixing what kind of issue.
>>>
>>
>> You can squash the docs into the 1/2 patch.
>>
> Yes, I've sent a single patch before: https://patches.dpdk.org/patch/54374/
> After some quiet time, I found that most of commits about doc are a single
> patch, so I split it into two.
>
Personally I don't have an issue either way, but docs with code is the
preferred method for DPDK so you could squash if you are doing a v2.
>>> Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
>>> ---
>>> doc/guides/nics/ixgbe.rst | 18 ++++++++++++++++++
>>> 1 file changed, 18 insertions(+)
>>>
>>> diff --git a/doc/guides/nics/ixgbe.rst b/doc/guides/nics/ixgbe.rst
>>> index 975143f..33812a5 100644
>>> --- a/doc/guides/nics/ixgbe.rst
>>> +++ b/doc/guides/nics/ixgbe.rst
>>> @@ -82,6 +82,24 @@ To guarantee the constraint, capabilities in dev_conf.rxmode.offloads will be ch
>>>
>>> fdir_conf->mode will also be checked.
>>>
>>> +VF Runtime Options
>>> +^^^^^^^^^^^^^^^^^^
>>> +
>>> +The following ``devargs`` options can be enabled at runtime. They must
>>> +be passed as part of EAL arguments. For example,
>>> +
>>> +.. code-block:: console
>>> +
>>> + testpmd -w af:10.0,pflink_fullchk=1 -- -i
>>> +
>>> +- ``pflink_fullchk`` (default **0**)
>>> +
>>> + Toggle behavior to get ixgbevf link status quickly by checking the
>>> + mailbox status or not. If set, the VF will check the PF's link status
>>> + and the PF's mailbox running status, which will trigger PF's mailbox
>>> + interrupt generation. Otherwise, just check PF's link status, then no
>>> + mailbox interrupt in PF.
>>> +
>>
>> wait_to_complete is also considered, so doesn't it mean that
>> pflink_fullck is not a toggle in all cases e.g. rte_eth_link_get() will
>> set wait_to_complete=1 and then pflink_fullchk is ignored. At least you
>> should mention if there is some user visible API like this where the
>> flag will not result in the behaviour being toggled.
>>
>
> I use "get ixgbevf link status -quickly-" to describe "wait_to_complete = 0",
> maybe I need some code to make the information more clear.
>
I think all you need to add is that rte_eth_link_get() will still use
the mailbox method regardless of the pflink_fullchk setting, but maybe
there's more I haven't considered wrt testpmd.
>>> RX Burst Size
>>> ^^^^^^^^^^^^^
>>>
>>>
>
> -----Original Message-----
> From: Kevin Traynor [mailto:ktraynor@redhat.com]
> Sent: Friday, June 7, 2019 22:38
> To: Wang, Haiyue <haiyue.wang@intel.com>; dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>; Lu,
> Wenzhuo <wenzhuo.lu@intel.com>; Wang, Liang-min <liang-min.wang@intel.com>; daniels@research.att.com
> Subject: Re: [dpdk-dev] [PATCH 2/2] doc: add the description for option pflink_fullchk in ixgbevf
>
> On 07/06/2019 14:55, Wang, Haiyue wrote:
> >> -----Original Message-----
> >> From: Kevin Traynor [mailto:ktraynor@redhat.com]
> >> Sent: Friday, June 7, 2019 20:25
> >> To: Wang, Haiyue <haiyue.wang@intel.com>; dev@dpdk.org; Zhang, Qi Z <qi.z.zhang@intel.com>; Lu,
> >> Wenzhuo <wenzhuo.lu@intel.com>; Wang, Liang-min <liang-min.wang@intel.com>;
> daniels@research.att.com
> >> Subject: Re: [dpdk-dev] [PATCH 2/2] doc: add the description for option pflink_fullchk in ixgbevf
> >>
> >> On 07/06/2019 04:10, Haiyue Wang wrote:
> >>> Add the detail description for this pflink_fullchk option, when it will
> >>> be used, and it is used for fixing what kind of issue.
> >>>
> >>
> >> You can squash the docs into the 1/2 patch.
> >>
> > Yes, I've sent a single patch before: https://patches.dpdk.org/patch/54374/
> > After some quiet time, I found that most of commits about doc are a single
> > patch, so I split it into two.
> >
>
> Personally I don't have an issue either way, but docs with code is the
> preferred method for DPDK so you could squash if you are doing a v2.
>
One patch sounds better now, thanks!
> >>> Signed-off-by: Haiyue Wang <haiyue.wang@intel.com>
> >>> ---
> >>> doc/guides/nics/ixgbe.rst | 18 ++++++++++++++++++
> >>> 1 file changed, 18 insertions(+)
> >>>
> >>> diff --git a/doc/guides/nics/ixgbe.rst b/doc/guides/nics/ixgbe.rst
> >>> index 975143f..33812a5 100644
> >>> --- a/doc/guides/nics/ixgbe.rst
> >>> +++ b/doc/guides/nics/ixgbe.rst
> >>> @@ -82,6 +82,24 @@ To guarantee the constraint, capabilities in dev_conf.rxmode.offloads will be
> ch
> >>>
> >>> fdir_conf->mode will also be checked.
> >>>
> >>> +VF Runtime Options
> >>> +^^^^^^^^^^^^^^^^^^
> >>> +
> >>> +The following ``devargs`` options can be enabled at runtime. They must
> >>> +be passed as part of EAL arguments. For example,
> >>> +
> >>> +.. code-block:: console
> >>> +
> >>> + testpmd -w af:10.0,pflink_fullchk=1 -- -i
> >>> +
> >>> +- ``pflink_fullchk`` (default **0**)
> >>> +
> >>> + Toggle behavior to get ixgbevf link status quickly by checking the
> >>> + mailbox status or not. If set, the VF will check the PF's link status
> >>> + and the PF's mailbox running status, which will trigger PF's mailbox
> >>> + interrupt generation. Otherwise, just check PF's link status, then no
> >>> + mailbox interrupt in PF.
> >>> +
> >>
> >> wait_to_complete is also considered, so doesn't it mean that
> >> pflink_fullck is not a toggle in all cases e.g. rte_eth_link_get() will
> >> set wait_to_complete=1 and then pflink_fullchk is ignored. At least you
> >> should mention if there is some user visible API like this where the
> >> flag will not result in the behaviour being toggled.
> >>
> >
> > I use "get ixgbevf link status -quickly-" to describe "wait_to_complete = 0",
> > maybe I need some code to make the information more clear.
> >
>
> I think all you need to add is that rte_eth_link_get() will still use
> the mailbox method regardless of the pflink_fullchk setting, but maybe
> there's more I haven't considered wrt testpmd.
>
Agree, will update a new patch to make this clear. Thanks!
> >>> RX Burst Size
> >>> ^^^^^^^^^^^^^
> >>>
> >>>
> >
@@ -82,6 +82,24 @@ To guarantee the constraint, capabilities in dev_conf.rxmode.offloads will be ch
fdir_conf->mode will also be checked.
+VF Runtime Options
+^^^^^^^^^^^^^^^^^^
+
+The following ``devargs`` options can be enabled at runtime. They must
+be passed as part of EAL arguments. For example,
+
+.. code-block:: console
+
+ testpmd -w af:10.0,pflink_fullchk=1 -- -i
+
+- ``pflink_fullchk`` (default **0**)
+
+ Toggle behavior to get ixgbevf link status quickly by checking the
+ mailbox status or not. If set, the VF will check the PF's link status
+ and the PF's mailbox running status, which will trigger PF's mailbox
+ interrupt generation. Otherwise, just check PF's link status, then no
+ mailbox interrupt in PF.
+
RX Burst Size
^^^^^^^^^^^^^