RE: [PATCH v2 2/3] power: sequencing: Add Renesas RZ/G3L Power Ready driver

[Biju Das]

Hi Philipp Zabel,

Thanks for the feedback.

> -----Original Message-----
> From: Philipp Zabel <p.zabel@xxxxxxxxxxxxxx>
> Sent: 29 July 2026 11:08
> Subject: Re: [PATCH v2 2/3] power: sequencing: Add Renesas RZ/G3L Power Ready driver
> 
> On Mi, 2026-07-29 at 10:31 +0100, Biju wrote:
> > From: Biju Das <biju.das.jz@xxxxxxxxxxxxxx>
> >
> > Add a power sequencing driver for the Renesas RZ/G3L PWRRDY module,
> > which signals power readiness for various IPs (USB, DSI, CSI etc.) on
> > the SoC. The driver binds as an auxiliary device to the parent SYSC
> > driver, using its regmap to toggle the SYS_PWRRDY_N register bits, and
> > exposes {usb,dsi,csi}-pwrrdy pwrseq targets.
> >
> > Signed-off-by: Biju Das <biju.das.jz@xxxxxxxxxxxxxx>
> > ---
> > v1->v2:
> >  * Added a comment in pwrseq_rzg3l_pwrrdy_match().
> >  * Dropped dev_get_regmap() from probe as regmap is now part of platform
> >    data.
> >  * Added a blank line before devm_pwrseq_device_register() in probe.
> >  * Dropped the error message devm_pwrseq_device_register() as probe prints
> >    failure message.
> >  * Dropped local variables dev and regmap from probe().
> > ---
> >  drivers/power/sequencing/Kconfig              |   8 +
> >  drivers/power/sequencing/Makefile             |   1 +
> >  .../power/sequencing/pwrseq-renesas-pwrrdy.c  | 139
> > ++++++++++++++++++
> >  3 files changed, 148 insertions(+)
> >  create mode 100644 drivers/power/sequencing/pwrseq-renesas-pwrrdy.c
> >
> [...]
> > diff --git a/drivers/power/sequencing/pwrseq-renesas-pwrrdy.c
> > b/drivers/power/sequencing/pwrseq-renesas-pwrrdy.c
> > new file mode 100644
> > index 000000000000..723d574c3f10
> > --- /dev/null
> > +++ b/drivers/power/sequencing/pwrseq-renesas-pwrrdy.c
> > @@ -0,0 +1,139 @@
> > +// SPDX-License-Identifier: GPL-2.0
> > +/*
> > + * Renesas RZ/G3L Power Ready driver
> > + *
> > + */
> > +
> > +#include <linux/auxiliary_bus.h>
> > +#include <linux/module.h>
> > +#include <linux/pwrseq/provider.h>
> > +#include <linux/regmap.h>
> > +
> > +#define SYS_PWRRDY_N		0xd70
> > +#define SYS_PWRRDY_N_USB_MASK	BIT(0)
> > +#define SYS_PWRRDY_N_DSI_MASK	BIT(1)
> > +#define SYS_PWRRDY_N_CSI_MASK	BIT(2)
> > +
> > +static int pwrseq_rzg3l_set_pwrrdy(struct pwrseq_device *pwrseq, u32
> > +mask, u32 val) {
> > +	struct regmap *regmap = pwrseq_device_get_drvdata(pwrseq);
> > +
> > +	return regmap_update_bits(regmap, SYS_PWRRDY_N, mask, val); }
> 
> Did you mean to use regmap_assign_bits() with a boolean val?
> See below.

No, Originally, I planned to use regmap_update_bits() but there was logical
mistake as you pointed out below.

I will use regmap_assign_bits() with boolean val in next version.

> 
> > +static int pwrseq_rzg3l_usb_pwrrdy_enable(struct pwrseq_device
> > +*pwrseq) {
> > +	return pwrseq_rzg3l_set_pwrrdy(pwrseq, SYS_PWRRDY_N_USB_MASK, 0); }
> > +
> > +static int pwrseq_rzg3l_usb_pwrrdy_disable(struct pwrseq_device
> > +*pwrseq) {
> > +	return pwrseq_rzg3l_set_pwrrdy(pwrseq, SYS_PWRRDY_N_USB_MASK, 1);
> 
> This happens to work correctly since SYS_PWRRDY_N_USB_MASK is 1.

OK.

> 
> > +}
> > +
> > +static const struct pwrseq_unit_data pwrseq_rzg3l_usb_pwrrdy_unit = {
> > +	.name = "usb-pwrrdy-power-sequence",
> > +	.enable = pwrseq_rzg3l_usb_pwrrdy_enable,
> > +	.disable = pwrseq_rzg3l_usb_pwrrdy_disable, };
> > +
> > +static int pwrseq_rzg3l_dsi_pwrrdy_enable(struct pwrseq_device
> > +*pwrseq) {
> > +	return pwrseq_rzg3l_set_pwrrdy(pwrseq, SYS_PWRRDY_N_DSI_MASK, 0); }
> > +
> > +static int pwrseq_rzg3l_dsi_pwrrdy_disable(struct pwrseq_device
> > +*pwrseq) {
> > +	return pwrseq_rzg3l_set_pwrrdy(pwrseq, SYS_PWRRDY_N_DSI_MASK, 1);
> 
> 1 is not in SYS_PWRRDY_N_DSI_MASK, so this will clear the bit. Either set val = mask or use
> regmap_assign_bits().

Thanks for pointing out the mistake.

> 
> > +}
> > +
> > +static const struct pwrseq_unit_data pwrseq_rzg3l_dsi_pwrrdy_unit = {
> > +	.name = "dsi-pwrrdy-sequence",
> > +	.enable = pwrseq_rzg3l_dsi_pwrrdy_enable,
> > +	.disable = pwrseq_rzg3l_dsi_pwrrdy_disable, };
> > +
> > +static int pwrseq_rzg3l_csi_pwrrdy_enable(struct pwrseq_device
> > +*pwrseq) {
> > +	return pwrseq_rzg3l_set_pwrrdy(pwrseq, SYS_PWRRDY_N_CSI_MASK, 0); }
> > +
> > +static int pwrseq_rzg3l_csi_pwrrdy_disable(struct pwrseq_device
> > +*pwrseq) {
> > +	return pwrseq_rzg3l_set_pwrrdy(pwrseq, SYS_PWRRDY_N_CSI_MASK, 1);
> 
> Same issue as above.

Will fix this in next version.
> 
> [...]
> > +static int pwrseq_rzg3l_pwrrdy_match(struct pwrseq_device *pwrseq,
> > +				     struct device *dev)
> > +{
> > +	/*
> > +	 * The pwrrdy provider does not rely on any device-tree phandle/property
> > +	 * matching, so it unconditionally matches every consumer device that
> > +	 * requests it.
> > +	 */
> > +	return PWRSEQ_MATCH_OK;
> > +}
> > +
> > +static int pwrseq_rzg3l_pwrrdy_probe(struct auxiliary_device *adev,
> > +				     const struct auxiliary_device_id *id) {
> > +	struct pwrseq_device *pwrseq;
> > +	struct pwrseq_config config = {
> > +		.parent = &adev->dev,
> > +		.owner = THIS_MODULE,
> > +		.drvdata = adev->dev.platform_data,
> > +		.match = pwrseq_rzg3l_pwrrdy_match,
> > +		.targets = pwrseq_rzg3l_pwrrdy_targets,
> > +	};
> > +
> > +	pwrseq = devm_pwrseq_device_register(&adev->dev, &config);
> > +
> > +	return IS_ERR(pwrseq) ? PTR_ERR(pwrseq) : 0;
> 
> 	return PTR_ERR_OR_ZERO(pwrseq);

Yes, it is much simpler.

Cheers,
Biju