Request to backport KERNEL_DTC_FLAGS to Dunfell

Request to backport KERNEL_DTC_FLAGS to Dunfell

[Mark Jonas]

Hi,

We are using DT overlays and have the need to pass the -@ parameter in DTC_FLAGS to Linux' make. The patch

"kernel-devicetree: Introduce KERNEL_DTC_FLAGS to pass dtc flags"
https://git.openembedded.org/openembedded-core/commit/?h=hardknott&id=063b5de86624a42b0aa784db6dddc7552a6dee7f

makes that possible in a nice way. Would it be reasonable to request backporting of this patch to Dunfell?

We were also wondering what could be an alternative approach for getting this or a similar functionality. Patching a bbclass using a bbappend is not possible, is it?

Cheers,
Mark

 Mark Jonas

Building Technologies, Panel Software Fire (BT-FIR/ENG1-Grb)
Bosch Sicherheitssysteme GmbH | Postfach 11 11 | 85626 Grasbrunn | GERMANY | www.boschsecurity.com
Tel. +49 89 6290-1105 | Telefax +49 89 6290-281105 | Mark.Jonas@de.bosch.com

Sitz: Stuttgart, Registergericht: Amtsgericht Stuttgart HRB 23118
Aufsichtsratsvorsitzender: Christian Fischer; Geschäftsführung: Thomas Quante, Andreas Bartz

Re: Request to backport KERNEL_DTC_FLAGS to Dunfell

[Steve Sakoman]

On Fri, Jun 4, 2021 at 2:45 AM Jonas Mark (BT-FIR/ENG1-Grb)
<Mark.Jonas@de.bosch.com> wrote:
>
> Hi,
>
> We are using DT overlays and have the need to pass the -@ parameter in DTC_FLAGS to Linux' make. The patch
>
> "kernel-devicetree: Introduce KERNEL_DTC_FLAGS to pass dtc flags"
> https://git.openembedded.org/openembedded-core/commit/?h=hardknott&id=063b5de86624a42b0aa784db6dddc7552a6dee7f
>
> makes that possible in a nice way. Would it be reasonable to request backporting of this patch to Dunfell?

This is sort of in the grey area between feature and bug fix.  It does
seem relatively safe, so I would consider taking the patch if there
are no objections from the community.

Added Bruce to the cc list to get his opinion.

> We were also wondering what could be an alternative approach for getting this or a similar functionality. Patching a bbclass using a bbappend is not possible, is it?

No, there is no equivalent of bbappend for classes.

Other solutions are possible, but taking this patch would be the
cleanest solution to your problem.  Let's see what others think before
we go there :-)

Steve


>
> Cheers,
> Mark
>
>  Mark Jonas
>
> Building Technologies, Panel Software Fire (BT-FIR/ENG1-Grb)
> Bosch Sicherheitssysteme GmbH | Postfach 11 11 | 85626 Grasbrunn | GERMANY | www.boschsecurity.com
> Tel. +49 89 6290-1105 | Telefax +49 89 6290-281105 | Mark.Jonas@de.bosch.com
>
> Sitz: Stuttgart, Registergericht: Amtsgericht Stuttgart HRB 23118
> Aufsichtsratsvorsitzender: Christian Fischer; Geschäftsführung: Thomas Quante, Andreas Bartz
>

Re: Request to backport KERNEL_DTC_FLAGS to Dunfell

[Bruce Ashfield]

On Fri, Jun 4, 2021 at 11:09 AM Steve Sakoman <steve@sakoman.com> wrote:
>
> On Fri, Jun 4, 2021 at 2:45 AM Jonas Mark (BT-FIR/ENG1-Grb)
> <Mark.Jonas@de.bosch.com> wrote:
> >
> > Hi,
> >
> > We are using DT overlays and have the need to pass the -@ parameter in DTC_FLAGS to Linux' make. The patch
> >
> > "kernel-devicetree: Introduce KERNEL_DTC_FLAGS to pass dtc flags"
> > https://git.openembedded.org/openembedded-core/commit/?h=hardknott&id=063b5de86624a42b0aa784db6dddc7552a6dee7f
> >
> > makes that possible in a nice way. Would it be reasonable to request backporting of this patch to Dunfell?
>
> This is sort of in the grey area between feature and bug fix.  It does
> seem relatively safe, so I would consider taking the patch if there
> are no objections from the community.
>
> Added Bruce to the cc list to get his opinion.

Good timing, I happened to notice this earlier and was thinking about it.

While I also wouldn't call it a bug, I agree it is a small/standalone
change that shouldn't break anything.

A quick scan of the dunfell tested kernels shows unchanged Makefile
handling of the option, so there's little risk there.

I'd say that it would be fine for backport.

Bruce

>
> > We were also wondering what could be an alternative approach for getting this or a similar functionality. Patching a bbclass using a bbappend is not possible, is it?
>
> No, there is no equivalent of bbappend for classes.
>
> Other solutions are possible, but taking this patch would be the
> cleanest solution to your problem.  Let's see what others think before
> we go there :-)
>
> Steve
>
>
> >
> > Cheers,
> > Mark
> >
> >  Mark Jonas
> >
> > Building Technologies, Panel Software Fire (BT-FIR/ENG1-Grb)
> > Bosch Sicherheitssysteme GmbH | Postfach 11 11 | 85626 Grasbrunn | GERMANY | www.boschsecurity.com
> > Tel. +49 89 6290-1105 | Telefax +49 89 6290-281105 | Mark.Jonas@de.bosch.com
> >
> > Sitz: Stuttgart, Registergericht: Amtsgericht Stuttgart HRB 23118
> > Aufsichtsratsvorsitzender: Christian Fischer; Geschäftsführung: Thomas Quante, Andreas Bartz
> >



-- 
- Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end
- "Use the force Harry" - Gandalf, Star Trek II

Re: [OE-core] Request to backport KERNEL_DTC_FLAGS to Dunfell

[Quentin Schulz]

Hi all,

On Fri, Jun 04, 2021 at 11:25:36AM -0400, Bruce Ashfield wrote:
> On Fri, Jun 4, 2021 at 11:09 AM Steve Sakoman <steve@sakoman.com> wrote:
> >
> > On Fri, Jun 4, 2021 at 2:45 AM Jonas Mark (BT-FIR/ENG1-Grb)
> > <Mark.Jonas@de.bosch.com> wrote:
> > >
> > > Hi,
> > >
> > > We are using DT overlays and have the need to pass the -@ parameter in DTC_FLAGS to Linux' make. The patch
> > >
> > > "kernel-devicetree: Introduce KERNEL_DTC_FLAGS to pass dtc flags"
> > > https://git.openembedded.org/openembedded-core/commit/?h=hardknott&id=063b5de86624a42b0aa784db6dddc7552a6dee7f
> > >
> > > makes that possible in a nice way. Would it be reasonable to request backporting of this patch to Dunfell?
> >
> > This is sort of in the grey area between feature and bug fix.  It does
> > seem relatively safe, so I would consider taking the patch if there
> > are no objections from the community.
> >
> > Added Bruce to the cc list to get his opinion.
> 
> Good timing, I happened to notice this earlier and was thinking about it.
> 
> While I also wouldn't call it a bug, I agree it is a small/standalone
> change that shouldn't break anything.
> 
> A quick scan of the dunfell tested kernels shows unchanged Makefile
> handling of the option, so there's little risk there.
> 
> I'd say that it would be fine for backport.
> 

Please backport the documentation patch as well if you do backport the
feature, see:
https://git.yoctoproject.org/cgit.cgi/yocto-docs/commit/documentation/ref-manual/variables.rst?id=b2d5bf48fe4e76446a38762839865176294a4a3c

Cheers,
Quentin

Re: Request to backport KERNEL_DTC_FLAGS to Dunfell

[Mark Jonas]

Hi Bruce and Steve,

> Von: Bruce Ashfield <bruce.ashfield@gmail.com>

> On Fri, Jun 4, 2021 at 11:09 AM Steve Sakoman <steve@sakoman.com> wrote:
> >
> > On Fri, Jun 4, 2021 at 2:45 AM Jonas Mark (BT-FIR/ENG1-Grb)
> > <Mark.Jonas@de.bosch.com> wrote:
> > >
> > > Hi,
> > >
> > > We are using DT overlays and have the need to pass the -@ parameter
> > > in DTC_FLAGS to Linux' make. The patch
> > >
> > > "kernel-devicetree: Introduce KERNEL_DTC_FLAGS to pass dtc flags"

[cut out web links messed up by Outlook]

> > > makes that possible in a nice way. Would it be reasonable to request
> backporting of this patch to Dunfell?
> >
> > This is sort of in the grey area between feature and bug fix.  It does
> > seem relatively safe, so I would consider taking the patch if there
> > are no objections from the community.
> >
> > Added Bruce to the cc list to get his opinion.
> 
> Good timing, I happened to notice this earlier and was thinking about it.
> 
> While I also wouldn't call it a bug, I agree it is a small/standalone change that
> shouldn't break anything.
> 
> A quick scan of the dunfell tested kernels shows unchanged Makefile handling of
> the option, so there's little risk there.
> 
> I'd say that it would be fine for backport.

Can I help with the backport in any way?

My colleague Tingquan gave it a try and the patch from Hardknott applies cleanly on Dunfell.

The documentation patch for Poky proposed by Quentin does not apply though. But it is just that ref-variables.rst was renamed to variables.rst.

> 
> Bruce
> 
> >
> > > We were also wondering what could be an alternative approach for getting
> this or a similar functionality. Patching a bbclass using a bbappend is not
> possible, is it?
> >
> > No, there is no equivalent of bbappend for classes.
> >
> > Other solutions are possible, but taking this patch would be the
> > cleanest solution to your problem.  Let's see what others think before
> > we go there :-)
> >
> > Steve

Cheers,
Mark

Re: Request to backport KERNEL_DTC_FLAGS to Dunfell

[Steve Sakoman]

On Wed, Jun 16, 2021 at 5:28 AM Jonas Mark (BT-FIR/ENG1-Grb)
<Mark.Jonas@de.bosch.com> wrote:
>
> Hi Bruce and Steve,
>
> > Von: Bruce Ashfield <bruce.ashfield@gmail.com>
>
> > On Fri, Jun 4, 2021 at 11:09 AM Steve Sakoman <steve@sakoman.com> wrote:
> > >
> > > On Fri, Jun 4, 2021 at 2:45 AM Jonas Mark (BT-FIR/ENG1-Grb)
> > > <Mark.Jonas@de.bosch.com> wrote:
> > > >
> > > > Hi,
> > > >
> > > > We are using DT overlays and have the need to pass the -@ parameter
> > > > in DTC_FLAGS to Linux' make. The patch
> > > >
> > > > "kernel-devicetree: Introduce KERNEL_DTC_FLAGS to pass dtc flags"
>
> [cut out web links messed up by Outlook]
>
> > > > makes that possible in a nice way. Would it be reasonable to request
> > backporting of this patch to Dunfell?
> > >
> > > This is sort of in the grey area between feature and bug fix.  It does
> > > seem relatively safe, so I would consider taking the patch if there
> > > are no objections from the community.
> > >
> > > Added Bruce to the cc list to get his opinion.
> >
> > Good timing, I happened to notice this earlier and was thinking about it.
> >
> > While I also wouldn't call it a bug, I agree it is a small/standalone change that
> > shouldn't break anything.
> >
> > A quick scan of the dunfell tested kernels shows unchanged Makefile handling of
> > the option, so there's little risk there.
> >
> > I'd say that it would be fine for backport.
>
> Can I help with the backport in any way?

The reference patch is already in dunfell:

https://git.openembedded.org/openembedded-core/commit/?h=dunfell&id=2246b0d7a71c69eb2e89c55991d1387069895466

> My colleague Tingquan gave it a try and the patch from Hardknott applies cleanly on Dunfell.
>
> The documentation patch for Poky proposed by Quentin does not apply though. But it is just that ref-variables.rst was renamed to variables.rst.

Please see the documentation patch currently on the list for review:

https://lists.yoctoproject.org/g/docs/message/1469

Both should be in the 3.1.9 release coming up in the next week or so.

Steve



> > > > We were also wondering what could be an alternative approach for getting
> > this or a similar functionality. Patching a bbclass using a bbappend is not
> > possible, is it?
> > >
> > > No, there is no equivalent of bbappend for classes.
> > >
> > > Other solutions are possible, but taking this patch would be the
> > > cleanest solution to your problem.  Let's see what others think before
> > > we go there :-)
> > >
> > > Steve
>
> Cheers,
> Mark