Upstreaming plan

Upstreaming plan

[Miguel Ojeda]

Hi all,

I think we are in a reasonable position to start thinking about
upstreaming. While things are still experimental and will be for a
while; and while there are many bits to be solved (e.g. an example
driver and everything it may entail), the setup is closer to how
things will look like, which means maintainers can start giving
feedback. Plus things look fairly usable for other developers too.

Today v5.11 was released, and this is the approximate kernel release schedule:

  2021-04-18 v5.12 released
  2021-05-02 v5.13 end of merge window
  2021-06-13 v5.13 released
  2021-06-27 v5.14 end of merge window

I propose we target v5.13's merge window for the upstreaming. If we
don't make it (likely), that is fine, we can use it as a test drive to
solve whatever is left and be ready for v5.14.

To that end, the plan can be:

  - Up to around 2021-03-28 (6 weeks): finish whatever features we may
want to have in the initial PR; including cleanups and any example
module/driver, etc.

  - 2021-03-29 - 2021-04-04 (1 week): Then I will rebase and draft a
series for upstreaming. Here it is where we should do a full review of
the entire diff, just in case something got messed up in the different
syncs, plus allows us to spot any final touches.

  - 2021-04-05 - 2021-04-18 (2 weeks): Send the series as an RFC 2
weeks or so before v5.13 merge window to the LKML. This way we get
some general feedback and discussion, plus Linus et. al. can be aware
we might send it. Depending how things go here, we can decide whether
to go for v5.13 merge window or delay to v5.14 (then the same plan
would apply, shifted by 2 months).

Thoughts welcome! -- Specially if you have something baking that may
need more time than those initial ~6 weeks (e.g. you are busy etc.).

Cheers,
Miguel

Re: Upstreaming plan

[Alex Gaynor]

For folks who haven't committed them to memory,
https://github.com/Rust-for-Linux/linux/issues?q=is%3Aissue+is%3Aopen+label%3A%22upstream+-+required%22
is our list of issues that we currently think are shoul be done
pre-upstreaming. I'll commit to taking on #53, hopefully this week.

Thanks for putting this schedule together Miguel!

Alex

On Mon, Feb 15, 2021 at 12:37 AM Miguel Ojeda
<miguel.ojeda.sandonis@gmail.com> wrote:
>
> Hi all,
>
> I think we are in a reasonable position to start thinking about
> upstreaming. While things are still experimental and will be for a
> while; and while there are many bits to be solved (e.g. an example
> driver and everything it may entail), the setup is closer to how
> things will look like, which means maintainers can start giving
> feedback. Plus things look fairly usable for other developers too.
>
> Today v5.11 was released, and this is the approximate kernel release schedule:
>
>   2021-04-18 v5.12 released
>   2021-05-02 v5.13 end of merge window
>   2021-06-13 v5.13 released
>   2021-06-27 v5.14 end of merge window
>
> I propose we target v5.13's merge window for the upstreaming. If we
> don't make it (likely), that is fine, we can use it as a test drive to
> solve whatever is left and be ready for v5.14.
>
> To that end, the plan can be:
>
>   - Up to around 2021-03-28 (6 weeks): finish whatever features we may
> want to have in the initial PR; including cleanups and any example
> module/driver, etc.
>
>   - 2021-03-29 - 2021-04-04 (1 week): Then I will rebase and draft a
> series for upstreaming. Here it is where we should do a full review of
> the entire diff, just in case something got messed up in the different
> syncs, plus allows us to spot any final touches.
>
>   - 2021-04-05 - 2021-04-18 (2 weeks): Send the series as an RFC 2
> weeks or so before v5.13 merge window to the LKML. This way we get
> some general feedback and discussion, plus Linus et. al. can be aware
> we might send it. Depending how things go here, we can decide whether
> to go for v5.13 merge window or delay to v5.14 (then the same plan
> would apply, shifted by 2 months).
>
> Thoughts welcome! -- Specially if you have something baking that may
> need more time than those initial ~6 weeks (e.g. you are busy etc.).
>
> Cheers,
> Miguel



-- 
All that is necessary for evil to succeed is for good people to do nothing.

Re: Upstreaming plan

[Finn Behrens]

Hi,
Upstreaming sounds great.
I could hopefully try building the rust part with nix again next weekend (currently exams in school :-) ), as I think there should be a buildable derivation before upstreaming. It should be doable, and just some work because of how the kernel is build with nix.
Will write again when I have something that is building, great work so far.

Finn

> On 15. Feb 2021, at 06:35, Miguel Ojeda <miguel.ojeda.sandonis@gmail.com> wrote:
> 
> Hi all,
> 
> I think we are in a reasonable position to start thinking about
> upstreaming. While things are still experimental and will be for a
> while; and while there are many bits to be solved (e.g. an example
> driver and everything it may entail), the setup is closer to how
> things will look like, which means maintainers can start giving
> feedback. Plus things look fairly usable for other developers too.
> 
> Today v5.11 was released, and this is the approximate kernel release schedule:
> 
>  2021-04-18 v5.12 released
>  2021-05-02 v5.13 end of merge window
>  2021-06-13 v5.13 released
>  2021-06-27 v5.14 end of merge window
> 
> I propose we target v5.13's merge window for the upstreaming. If we
> don't make it (likely), that is fine, we can use it as a test drive to
> solve whatever is left and be ready for v5.14.
> 
> To that end, the plan can be:
> 
>  - Up to around 2021-03-28 (6 weeks): finish whatever features we may
> want to have in the initial PR; including cleanups and any example
> module/driver, etc.
> 
>  - 2021-03-29 - 2021-04-04 (1 week): Then I will rebase and draft a
> series for upstreaming. Here it is where we should do a full review of
> the entire diff, just in case something got messed up in the different
> syncs, plus allows us to spot any final touches.
> 
>  - 2021-04-05 - 2021-04-18 (2 weeks): Send the series as an RFC 2
> weeks or so before v5.13 merge window to the LKML. This way we get
> some general feedback and discussion, plus Linus et. al. can be aware
> we might send it. Depending how things go here, we can decide whether
> to go for v5.13 merge window or delay to v5.14 (then the same plan
> would apply, shifted by 2 months).
> 
> Thoughts welcome! -- Specially if you have something baking that may
> need more time than those initial ~6 weeks (e.g. you are busy etc.).
> 
> Cheers,
> Miguel

Re: Upstreaming plan

[Miguel Ojeda]

On Wed, Mar 3, 2021 at 9:27 AM Finn Behrens <finn@kloenk.dev> wrote:
>
> Hi,
> Upstreaming sounds great.
> I could hopefully try building the rust part with nix again next weekend (currently exams in school :-) ), as I think there should be a buildable derivation before upstreaming. It should be doable, and just some work because of how the kernel is build with nix.
> Will write again when I have something that is building, great work so far.

Thanks Finn, that would be great!

Good luck on your exams!

Cheers,
Miguel