Re: [dpdk-ci] [dpdk-dev] [PATCH v1 0/4] add Intel DCF PMD support

[Kevin Traynor <ktraynor@redhat.com>]

On 09/03/2020 17:57, Thomas Monjalon wrote:
> 09/03/2020 17:20, Ye Xiaolong:
>> Hi, David
>>
>> On 03/09, David Marchand wrote:
>>> On Mon, Mar 9, 2020 at 3:22 PM Haiyue Wang <haiyue.wang@intel.com> wrote:
>>>>
>>>> A DCF (Device Config Function) based approach is proposed where a device
>>>> bound to the device's VF0 can act as a sole controlling entity to exercise
>>>> advance functionality (such as switch, ACL) for rest of the VFs.
>>>>
>>>> The DCF works as a standalone PMD to support this function, which shares the
>>>> ice PMD flow control core function and the iavf virtchnl mailbox core module.
>>>>
>>>> This patchset is based on:
>>>> [1] https://patchwork.dpdk.org/cover/66417/ update ice base code
>>>
>>> The problem is that the CI(s) won't handle this.
>>> Example for the robot: https://travis-ci.com/ovsrobot/dpdk/builds/152461907
>>>
>>> Maybe we could add something as an annotation to the cover letter or
>>> the first patch of a series so that the CI(s) can detect and try to be
>>> intelligent?
>>
>> Agree, It'd be helpful if the cover letter of the first patch contains some
>> base tree info including the base commit and dependency patchset info (if any), 
>> so the CI can determine the correct base on top of which the developer's
>> patchset applies to avoid any apply issue and potential false positive. 
>>
>> And I know there is one option '--base'' of `git format-patch` which is
>> dedicated for this kind of usage, it can help create the base tree info block
>> which can be easily consumed by the CI. Here is the simple intro of it.
>>
>> Imagine that on top of the public commit P (already in upstream), the developer
>> applied well-known (on-flight, in the mailing list but not merged yet) patches
>> X, Y and Z from somebody else or himself, and then built his three-patch series
>> A, B, C, the commit history would be like:
>>
>> ................................................
>> ---P---X---Y---Z---A---B---C
>> ................................................
>>
>> With `git format-patch --base=P -3 C`,
>>
>> where P could be the exact commit sha, or variants e.g. HEAD~6, we can also use
>> --base=auto for convenience, the base tree information block will be shown at
>> the end of the first message the command outputs (either the first patch, or
>> the cover letter), like this:
>>
>> ------------
>> base-commit: P
>> prerequisite-patch-id: X
>> prerequisite-patch-id: Y
>> prerequisite-patch-id: Z
>> ------------
>>
>> Here P is the commit sha, and X,Y,Z are the patch ids of the dependency patches.
>>
>>
>> With this info in place, I think CI should be able to setup the exact base for
>> the coming patchset, the missing part I can see is the mapping of 
>> (in-flight patch <-> patch id), since we have all the in-flight patches in
>> patchwork, creating and maintaining such mapping in DB is doable, what do you
>> think?
> 
> I think it would simpler to list dependencies as patchwork ids.
> Example:
> 	Depends-on: series-42, patch-12345
> 

+1. I don't think it should depend on a base-commit. If it doesn't
apply/build/work with the latest upstream code then it's a valid error.

>