[dpdk-dev] [PATCH v1 00/15] preparing l2fwd for eventmode additions

[Joseph, Anoob]

Hi,

Gentle reminder!

Thanks,
Anoob

On 19-06-2018 19:37, Anoob Joseph wrote:
> External Email
>
> Hi Bruce,
>
>
> Thanks for the feedback. Please see inline.
>
>
> + Hemant, Nikhil, Sunil, Gage, Harry, Narender, Pavan, Thomas, Akhil
>
>
> On 19/06/18 15:39, Bruce Richardson wrote:
>> On Tue, Jun 19, 2018 at 03:34:29PM +0530, Anoob Joseph wrote:
>>> Hi Bruce, Pablo,
>>>
>>> Any comments on this series?
>>>
>>> Thanks,
>>> Anoob
>>>
>>> On 14/06/18 17:18, Anoob Joseph wrote:
>>>> This patchset modularizes l2fwd application to prepare it for 
>>>> eventmode
>>>> additions. This patchset doesn't change the code flow or logic, except
>>>> for few minor improvements. Some of the newly added functions are used
>>>> in just one place, but is added for efficient usage with eventmode.
>>>>
>>>> v1:
>>>> * Fix all checkpatch reported issues
>>>>
>> My main concern here is how much this eventmode addition is going to
>> complicate the l2fwd example. l2fwd has always been a pretty basic 
>> example
>> app to get users started on the basics of DPDK use, and I'm not sure how
>> much we want to move away from that. Is this eventmode-l2fwd better 
>> being a
>> separate app, to allow l2fwd to be kept as simple as it can be?
>>
>> Looking for more thoughts from others here, since it's a community 
>> decision
>> as to the scope of the examples.
>>
>> /Bruce
> The eventmode helper abstracts most of the changes required by the
> application to run in eventmode. This was taken up following the
> comments on a patch submitted by Sunil(sunil.kori at nxp.com).
> http://patches.dpdk.org/patch/37955/
>
> With eventmode helper, an application can be moved to eventmode with
> minimal changes. For l2fwd, the key patch which enables eventmode is,
>
> http://patches.dpdk.org/patch/40920/
> [The aforementioned patch is dependent on this patch series]
>
> The bulk of the code in this patch(40920) is adding multiple event mode
> worker functions.The existing init code and poll mode worker is barely
> touched. Multiple workers were introduced because a single event mode
> worker would not have made the best use of the varying capabilities of
> event devices.
>
> Single event mode worker could've demonstrated how minimal the changes
> can be. But the ability to register multiple workers, fine tuned for
> varying capabilities, is a good feature to have since it will enable
> applications to utilize the full potential of the hardware.
>
> Eventmode helper patch series:
> http://patches.dpdk.org/project/dpdk/list/?series=61
>
> The rules that were followed while drafting eventmode helper were very
> simple,
> 1. Move any code common to multiple applications to eventmode helper
> 2. Expose all capabilities of the devices involved (event & eth devs)
> 3. Minimize changes to the existing code
>
> For l2fwd we can opt for a new eventmode-l2fwd app, but this might not
> work for more complicated apps like l3fwd & ipsec-secgw. L2fwd app will
> stay the same even with the eventmode additions. It will still be a
> quick-start, easy-to-use app. In addition to demonstrating DPDK, it will
> also be able to demonstrate how easily an app can be made to run in
> eventmode, using the helper functions.
>
> With more event adapters getting added (tx adapter, crypto adapter,
> timer adapter etc), the helper will prove useful in abstracting the
> complex configuration options exposed by adapters. Similar changes would
> be required in other example apps, and the additions in l2fwd is to
> finalize on the approach.
>
> The current patch series just re-factors the code with couple of patches
> fixing preexisting checkpatch issues. The rest of the changes are split
> into individual patches for ease of review and testing. Hence the large
> number of patches.
>
> Thanks,
> Anoob