Cherry-picking 5.16 commits into OpenBMC Linux?

Cherry-picking 5.16 commits into OpenBMC Linux?

[Oskar Senft]

Hi Joel

There are a few commit that I sent upstream that have been accepted by
the maintainers (i.e. have the appropriate "Reviewed-by" tags) that
are waiting for 5.16 to open.

I'm waiting for these patches to be available to be able to send a new
meta-tyan directory to OpenBMC for review.

Would it be possible to cherry-pick these commits into the OpenBMC
Linux tree to speed up this process? Or is this something you
generally prefer to not do?

Specific patches:
https://lore.kernel.org/lkml/20211020164213.174597-1-osk@google.com/t/#m8f7ac85809049fadcabf6e0bed1ebab12e71f094
https://lore.kernel.org/linux-arm-kernel/20210909004920.1634322-1-osk@google.com/

There will also be an update to the DTS from the second patch once the
driver change from the first patch has fully landed. Or should I
better send that DTS change now and we'll take it into OpenBMC Linux
together?

Thanks
Oskar.

Re: Cherry-picking 5.16 commits into OpenBMC Linux?

[Joel Stanley]

On Wed, 3 Nov 2021 at 16:29, Oskar Senft <osk@google.com> wrote:
>
> Hi Joel
>
> There are a few commit that I sent upstream that have been accepted by
> the maintainers (i.e. have the appropriate "Reviewed-by" tags) that
> are waiting for 5.16 to open.
>
> I'm waiting for these patches to be available to be able to send a new
> meta-tyan directory to OpenBMC for review.
>
> Would it be possible to cherry-pick these commits into the OpenBMC
> Linux tree to speed up this process? Or is this something you
> generally prefer to not do?

Yes, this is exactly what the tree is for.

> Specific patches:
> https://lore.kernel.org/lkml/20211020164213.174597-1-osk@google.com/t/#m8f7ac85809049fadcabf6e0bed1ebab12e71f094

I've applied this one.

> https://lore.kernel.org/linux-arm-kernel/20210909004920.1634322-1-osk@google.com/

If you take a look at the dev-5.15 branch, you can see that this one
is already applied.

> There will also be an update to the DTS from the second patch once the
> driver change from the first patch has fully landed. Or should I
> better send that DTS change now and we'll take it into OpenBMC Linux
> together?

Send the dts change to the upstream lists. I'll review and put it
straight in the openbmc tree.

In general I encourage developers to send patches straight to the
upstream lists, and once they've had a review there send me a note to
backport them and I will do that as soon as possible.

Cheers,

Joel

Re: Cherry-picking 5.16 commits into OpenBMC Linux?

[Thang Nguyen]

Hi Joel,
We have some commits for Mt.Jade device tree that have your Reviewed-by:
https://www.spinics.net/lists/kernel/msg4115571.html
https://www.spinics.net/lists/kernel/msg4115572.html
And another commit applied at 
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=next-20211029&id=0b32c1b4071c482f2cf98b717cfc7380423ec619.
Would it be possible to apply into OpenBMC kernel 5.15 tree? I need 
these commits to replace some u-boot patches in meta-ampere.

Thanks,
Thang Q. Nguyen

On 04/11/2021 06:22, Joel Stanley wrote:
> On Wed, 3 Nov 2021 at 16:29, Oskar Senft <osk@google.com> wrote:
>>
>> Hi Joel
>>
>> There are a few commit that I sent upstream that have been accepted by
>> the maintainers (i.e. have the appropriate "Reviewed-by" tags) that
>> are waiting for 5.16 to open.
>>
>> I'm waiting for these patches to be available to be able to send a new
>> meta-tyan directory to OpenBMC for review.
>>
>> Would it be possible to cherry-pick these commits into the OpenBMC
>> Linux tree to speed up this process? Or is this something you
>> generally prefer to not do?
> 
> Yes, this is exactly what the tree is for.
> 
>> Specific patches:
>> https://lore.kernel.org/lkml/20211020164213.174597-1-osk@google.com/t/#m8f7ac85809049fadcabf6e0bed1ebab12e71f094
> 
> I've applied this one.
> 
>> https://lore.kernel.org/linux-arm-kernel/20210909004920.1634322-1-osk@google.com/
> 
> If you take a look at the dev-5.15 branch, you can see that this one
> is already applied.
> 
>> There will also be an update to the DTS from the second patch once the
>> driver change from the first patch has fully landed. Or should I
>> better send that DTS change now and we'll take it into OpenBMC Linux
>> together?
> 
> Send the dts change to the upstream lists. I'll review and put it
> straight in the openbmc tree.
> 
> In general I encourage developers to send patches straight to the
> upstream lists, and once they've had a review there send me a note to
> backport them and I will do that as soon as possible.
> 
> Cheers,
> 
> Joel
> 

Re: Cherry-picking 5.16 commits into OpenBMC Linux?

[Oskar Senft]

Hi Joel

> > Would it be possible to cherry-pick these commits into the OpenBMC
> > Linux tree to speed up this process? Or is this something you
> > generally prefer to not do?
>
> Yes, this is exactly what the tree is for.
Aha, cool! I'm starting to get the picture :-)

> > https://lore.kernel.org/linux-arm-kernel/20210909004920.1634322-1-osk@google.com/
>
> If you take a look at the dev-5.15 branch, you can see that this one
> is already applied.
Awesome, thank you! I didn't even think of looking there. I thought
I'd have to wait for 5.16 to manifest itself.

I sent https://gerrit.openbmc-project.xyz/c/openbmc/openbmc/+/48632 to
forward the commit for linux-aspeed. I hope that was right? Or does
someone (you?) do that periodically anyway? I did test (build and run)
and it works as expected.

Oskar.

Re: Cherry-picking 5.16 commits into OpenBMC Linux?

[Joel Stanley]

On Fri, 5 Nov 2021 at 03:52, Oskar Senft <osk@google.com> wrote:
>
> Hi Joel
>
> > > Would it be possible to cherry-pick these commits into the OpenBMC
> > > Linux tree to speed up this process? Or is this something you
> > > generally prefer to not do?
> >
> > Yes, this is exactly what the tree is for.
> Aha, cool! I'm starting to get the picture :-)
>
> > > https://lore.kernel.org/linux-arm-kernel/20210909004920.1634322-1-osk@google.com/
> >
> > If you take a look at the dev-5.15 branch, you can see that this one
> > is already applied.
> Awesome, thank you! I didn't even think of looking there. I thought
> I'd have to wait for 5.16 to manifest itself.
>
> I sent https://gerrit.openbmc-project.xyz/c/openbmc/openbmc/+/48632 to
> forward the commit for linux-aspeed. I hope that was right? Or does
> someone (you?) do that periodically anyway? I did test (build and run)
> and it works as expected.

I normally do this, but you're welcome to do it too. I'll merge your
device tree update now, and if you could then update your bump to
include that commit I'll +2 it.

I have a slightly horrible script that does the work for me. It
assumes the openbmc kernel is on a remote called 'openbmc' (this check
also makes sure that I've actually pushed the commits that I'm bumping
the recipie to).

#!/bin/bash
#
# Bump kernel version in meta-aspeed recipe and create a commit
# message using git shortlog
#
# Customise location of KERNELGITDIR to point to your openbmc git tree.
# Run from root of openbmc checkout tree.
#
# Joel Stanley <joel@jms.id.au>
#

set -xe

SUMMARY=$1

KERNELGITDIR="$HOME/dev/kernels/openbmc/.git"

RECIPE="meta-aspeed/recipes-kernel/linux/linux-aspeed_git.bb"
BRANCH=dev-5.15

OLDSHA=$(grep -Po "(\w{40})" ${RECIPE})
SHA=$(GIT_DIR=${KERNELGITDIR} git rev-parse openbmc/${BRANCH})
SHORTLOG=$(GIT_DIR=${KERNELGITDIR} git shortlog "${OLDSHA}..${SHA}")
VERSION=$(GIT_DIR=${KERNELGITDIR} git describe --abbrev=0 | cut -c 2-)

cat << EOF > ${RECIPE}
KBRANCH ?= "${BRANCH}"
LINUX_VERSION ?= "${VERSION}"

SRCREV="${SHA}"

require linux-aspeed.inc
EOF

MESSAGE="
linux-aspeed: ${SUMMARY}

${SHORTLOG}
"

echo "$MESSAGE" | git commit -s -F - ${RECIPE}

Re: Cherry-picking 5.16 commits into OpenBMC Linux?

[Oskar Senft]

Hi Joel

> > I sent https://gerrit.openbmc-project.xyz/c/openbmc/openbmc/+/48632
>
> I normally do this, but you're welcome to do it too. I'll merge your
> device tree update now, and if you could then update your bump to
> include that commit I'll +2 it.
Done. Thank you!

> I have a slightly horrible script that does the work for me.
I created the commit message manually this time. I hope I didn't get
the order backwards :-)

Oskar.