Re: [RFC 1/4] ARM: imx: use device_initcall for imx_soc_device_init
From: Shawn Guo
Date: Thu Feb 13 2020 - 00:55:42 EST
On Thu, Feb 13, 2020 at 05:47:41AM +0000, Peng Fan wrote:
> > Subject: Re: [RFC 1/4] ARM: imx: use device_initcall for imx_soc_device_init
> >
> > On Fri, Jan 17, 2020 at 08:15:54AM +0000, Peng Fan wrote:
> > > > > Subject: [RFC 1/4] ARM: imx: use device_initcall for
> > > > > imx_soc_device_init
> > > > >
> > > > > From: Peng Fan <peng.fan@xxxxxxx>
> > > > >
> > > > > This is preparation to move imx_soc_device_init to
> > > > > drivers/soc/imx/
> > > > >
> > > > > There is no reason to must put dt devices under /sys/devices/soc0,
> > > > > they could also be under /sys/devices/platform, so we could pass
> > > > > NULL as parent when calling of_platform_default_populate.
> > > > >
> > > >
> > > > This change will impact various internal test case & userspace lib, I think.
> > > > Need to ask test team & other developer to double check the impact.
> > >
> > > /sys/devices/soc0 is still there, the patchset only moves the platform
> > > devices which under /sys/devices/soc0 to /sys/devices/platform
> >
> > Jacky's concern still stands, as there are many user spaces which will be
> > broken and need update.
>
> The soc device itself still under /sys/devices/soc0, the soc_id/revision still there.
> It is just the platform devices moved to /sys/devices/platform.
>
> When I confirm with Jacky before, his concern is soc_id/revision will be
> moved. But this is not true, they are still there as before.
>
> >
> > > In this way, we aligned with ARM64. And simplify arch code by moving
> > > the code to drivers/soc/imx. In future, considering more cleanup, we
> > > could merge the code to soc-imx8.c, since they share similar silicon
> > > rev ocotp logic.
> >
> > Though this is a good thing from maintenance point of view, we do not want
> > to break user spaces.
>
> Actually not break user spaces, since this is RFC, I not expect this be merged.
> If you agree, I could post normal V1 patchset.
Okay. You send formal patches, and we get them into linux-next to see
if people will complain any breakage.
Shawn