[QUERY]: Avoiding tree wide changes

[QUERY]: Avoiding tree wide changes

[Prabhakar Mahadev Lad]

Hi Pavel/Nobuhiro,

We intend to backport IRQC driver [0] to 5.10-cip but for doing this we need the patch [1] ("of/platform: Drop static setup of IRQ resource from DT core"). The problem with applying patch [1] means we will have to do tree wide changes to replace platform_get_resource() -> platform_get_irq().

What are your thoughts on how we can get the IRQC driver to 5.10-cip?

[0] https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=next-20221215&id=3fed09559cd8be79568d368ce02bf7f2d56259b6
[1] https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=next-20221215&id=a1a2b7125e1079cfcc13a116aa3af3df2f9e002b

Cheers,
Prabhakar

Re: [QUERY]: Avoiding tree wide changes

[Pavel Machek]

[-- Attachment #1: Type: text/plain, Size: 1428 bytes --]

Hi!

> We intend to backport IRQC driver [0] to 5.10-cip but for doing this
>we need the patch [1] ("of/platform: Drop static setup of IRQ
>resource from DT core"). The problem with applying patch [1] means we
>will have to do tree wide changes to replace platform_get_resource()
>-> platform_get_irq().

We would not want to do that.

But... your platforms are already coverted to platform_get_irq() in
5.10, no?

> What are your thoughts on how we can get the IRQC driver to 5.10-cip?
> 
> [0] https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=next-20221215&id=3fed09559cd8be79568d368ce02bf7f2d56259b6
> [1] https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=next-20221215&id=a1a2b7125e1079cfcc13a116aa3af3df2f9e002b
>

Is the change in [0] really neccessary for your driver to work, no? I
wonder what happens if you revert it on new kernel. Is there chance it
will not do harm on platforms with "platform_get_resource() ->
platform_get_irq()" conversion done?

Do you have an example of "platform_get_resource() -> platform_get_irq()" changes?

One way forward could be introducing of_device_alloc_new and using it
on your platform, but finding other solutions would be better.

								Pavel
-- 
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

RE: [QUERY]: Avoiding tree wide changes

[Prabhakar Mahadev Lad]

Hi Pavel,

> -----Original Message-----
> From: Pavel Machek <pavel@denx.de>
> Sent: 16 December 2022 09:38
> To: Prabhakar Mahadev Lad <prabhakar.mahadev-lad.rj@bp.renesas.com>
> Cc: Pavel Machek <pavel@denx.de>; nobuhiro1.iwamatsu@toshiba.co.jp; cip-dev@lists.cip-project.org
> Subject: Re: [QUERY]: Avoiding tree wide changes
> 
> Hi!
> 
> > We intend to backport IRQC driver [0] to 5.10-cip but for doing this
> >we need the patch [1] ("of/platform: Drop static setup of IRQ resource
> >from DT core"). The problem with applying patch [1] means we will have
> >to do tree wide changes to replace platform_get_resource()
> >-> platform_get_irq().
> 
> We would not want to do that.
> 
OK.

> But... your platforms are already coverted to platform_get_irq() in 5.10, no?
> 
Yes our platform uses platform_get_irq() in upstream and shall be backported in same manner.

> > What are your thoughts on how we can get the IRQC driver to 5.10-cip?
> >
> > [0]
> > https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/co
> > mmit/?h=next-20221215&id=3fed09559cd8be79568d368ce02bf7f2d56259b6
> > [1]
> > https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/co
> > mmit/?h=next-20221215&id=a1a2b7125e1079cfcc13a116aa3af3df2f9e002b
> >
> 
> Is the change in [0] really neccessary for your driver to work, no? I wonder what happens if you
s/[0]/[1]?
> revert it on new kernel. Is there chance it will not do harm on platforms with
> "platform_get_resource() -> platform_get_irq()" conversion done?
> 
If we revert [1] the IRQC driver won't work as the chaining from IRQC to GIC won't happen. It won't do any harm to platforms which have platform_get_irq() conversion done.

> Do you have an example of "platform_get_resource() -> platform_get_irq()" changes?
> 
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=next-20221216&id=228662b074149e16949ab73205ac93b5ca2c3552
https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=next-20221216&id=3277e022a1c31f5041753942b3d9006174ca7aa2

https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/commit/?h=next-20221216&id=9879c81b680795b45a11894e48dfa9e4be2f4426
^^ and some special handling like the above too.

> One way forward could be introducing of_device_alloc_new and using it on your platform, but finding
> other solutions would be better.
> 

One way is to rename interrupts property to something else for driver [0] and manually parsing it but that's no good as it will differ from upstream. We will have to come up with the better approach.

Cheers,
Prabhakar