Re: [PATCH] drivers/mfd: Convert to DEFINE_SIMPLE_DEV_PM_OPS()

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:

New issues:
- [Low] The commit message prefix uses 'drivers/mfd:' instead of the required 'mfd: ' format.

Pre-existing issues:
- [High] Device teardown sequence disables PM runtime and clocks before unregistering MFD child devices, leading to an external abort kernel panic.
--

--- Patch [1]: [PATCH] drivers/mfd: Convert to DEFINE_SIMPLE_DEV_PM_OPS() ---
commit f2c525c2ee807012f8f33d16a4bcbea747e75dcf
Author: Triet Hoang <[email protected]>

    drivers/mfd: Convert to DEFINE_SIMPLE_DEV_PM_OPS()
    
    Convert the deprecated SIMPLE_DEV_PM_OPS
    to DEFINE_SIMPLE_DEV_PM_OPS
    and pm_sleep_ptr().

>     drivers/mfd: Convert to DEFINE_SIMPLE_DEV_PM_OPS()

[Severity: Low]
This isn't a bug, but does this commit title violate the MFD subsystem naming
conventions? The standard format requires using mfd: <Driver>: for the
commit message prefix instead of drivers/mfd:.

> diff --git a/drivers/mfd/ti_am335x_tscadc.c b/drivers/mfd/ti_am335x_tscadc.c
> index 068c25401c6cd..81f07a1d780c5 100644
> --- a/drivers/mfd/ti_am335x_tscadc.c
> +++ b/drivers/mfd/ti_am335x_tscadc.c
[ ... ]
> @@ -373,7 +373,7 @@ MODULE_DEVICE_TABLE(of, ti_tscadc_dt_ids);
>  static struct platform_driver ti_tscadc_driver = {
>  	.driver = {
>  		.name   = "ti_am3359-tscadc",
> -		.pm	= &tscadc_pm_ops,
> +		.pm	= pm_sleep_ptr(&tscadc_pm_ops),
>  		.of_match_table = ti_tscadc_dt_ids,
>  	},
>  	.probe	= ti_tscadc_probe,

[Severity: High]
This is a pre-existing issue, but does ti_tscadc_remove() cause a kernel panic
by disabling PM runtime before unregistering child devices?

When the module is unloaded, ti_tscadc_remove() drops the clock reference and
disables PM runtime:

    pm_runtime_put_sync(&pdev->dev);
    pm_runtime_disable(&pdev->dev);
    mfd_remove_devices(tscadc->dev);

When mfd_remove_devices() unbinds the ti_am335x_adc child driver, the child's
tiadc_remove() calls am335x_tsc_se_clr(). This issues a regmap_write() to the
REG_SE MMIO register while the parent device's clock is already powered down,
which triggers an L3 interconnect error and kernel panic.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1