Re: [GIT PULL 6/6] i.MX defconfig update for 5.2

[Shawn Guo <shawnguo@kernel.org>]

On Tue, Apr 30, 2019 at 09:18:18PM +0000, Leonard Crestez wrote:
> On 4/30/2019 5:27 AM, Shawn Guo wrote:
> > On Mon, Apr 29, 2019 at 09:29:10AM -0700, Olof Johansson wrote:
> >> On Mon, Apr 22, 2019 at 04:28:36PM +0800, Shawn Guo wrote:
> >>> The following changes since commit 9e98c678c2d6ae3a17cb2de55d17f69dddaa231b:
> >>>
> >>>    Linux 5.1-rc1 (2019-03-17 14:22:26 -0700)
> >>>
> >>> are available in the Git repository at:
> >>>
> >>>    git://git.kernel.org/pub/scm/linux/kernel/git/shawnguo/linux.git tags/imx-defconfig-5.2
> >>>
> >>> for you to fetch changes up to a3f12aac21848796f5b5c6e9ce2f04e411ec2129:
> >>>
> >>>    arm64: defconfig: NVMEM_IMX_OCOTP=y for imx8m (2019-04-22 10:33:53 +0800)
> >>>
> >>> ----------------------------------------------------------------
> >>> i.MX defconfig update for 5.2:
> >>>   - Enable PCF857X GPIO expander and SIOX bus driver support for
> >>>     imx_v6_v7_defconfig.
> >>>   - Enable more drivers for i.MX8 platform support in arm64 defconfig:
> >>>     SNVS RTC, SCU Watchdog, SPI, i.MX8MM pinctrl and clock, LPI2C,
> >>>     ROHM_BD718XX PMIC and OCOTP NVMEM.
> >>
> >> All of these are enabled as =y. I don't think all of them are critical for
> >> booting the system up to a rootfs. Can you please switch over any non-critical
> >> drivers to =m?
> 
> > - CONFIG_MFD_ROHM_BD718XX=y, CONFIG_REGULATOR_BD718XX=y
> > 
> >    @Leonard, are they critical for booting imx8mm-evk board up? Or can
> >    they be installed as module after system is up?
> 
> This is the PMIC used on imx8mm-evk and also the purism devkit. Since 
> it's used for main chip power "y" makes sense but boards can boot 
> without it and I can't think of a corner case where we could fail to 
> reach rootfs because of no PMIC driver.
> 
> Looking through arm64 defconfig other PMIC regulators seem to be "y"

Okay.

> 
>  > - CONFIG_GPIO_MAX732X=y
>  > - CONFIG_I2C_IMX_LPI2C=y
> 
> Other I2C/GPIO drivers are "y" but again can't point to anything 
> specific that would fail.

I understand that the rootfs in Olof's context might be ramdisk, and
missing GPIO driver shouldn't be a problem.  However, I really would
like build a kernel image out of defconfig that can run into rootfs on
SD/MMC or NFS, which would be much more useful.  In that case, drivers
like GPIO should be required. 

> 
> > - CONFIG_NVMEM_IMX_OCOTP=y
> > 
> >    @Olof, this is the driver for fuse access on i.MX platform.  Drivers
> >    like cpufreq will need to access fuse for cpu grade data.  So we would
> >    take this driver as fundamental infrastructural.
> 
> Other than thermal and speed grading fuses also contain MAC address on 
> imx chips. Since FEC (imx ethernet controller) is "y" for easy netboots 
> it makes a lot of sense to keep this "y" as well.

Agreed.

> 
> These are all relatively small drivers. I don't think that RTC drivers 
> are worth excluding either.

I don't think that is a good argument.  The need of being 'y' is based
on necessity rather than driver size.  arm-soc folks are maintaining
one single defconfig for all arm64 platforms. There could be hundreds
of such small drivers to be 'y' for no good reason.

Shawn

_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel