RE: [PATCH 07/16] reset: rzg2l-usbphy-ctrl: Get reset control array

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Claudiu,

> -----Original Message-----
> From: claudiu beznea <claudiu.beznea@xxxxxxxxx>
> Sent: Friday, August 23, 2024 9:05 AM
> Subject: Re: [PATCH 07/16] reset: rzg2l-usbphy-ctrl: Get reset control array
> 
> Hi, Biju,
> 
> On 23.08.2024 10:25, Biju Das wrote:
> > Hi Claudiu,
> >
> >> -----Original Message-----
> >> From: Claudiu <claudiu.beznea@xxxxxxxxx>
> >> Sent: Thursday, August 22, 2024 4:28 PM
> >> Subject: [PATCH 07/16] reset: rzg2l-usbphy-ctrl: Get reset control
> >> array
> >>
> >> From: Claudiu Beznea <claudiu.beznea.uj@xxxxxxxxxxxxxx>
> >>
> >> Before accessing the USB area of the RZ/G3S SoC the PWRRDY bit of the
> >> SYS_USB_PWRRDY register need to be cleared. When USB area is not used the PWRRDY bit of the
> SYS_USB_PWRRDY register need to be set.
> >> This register is in the SYSC controller address space and the
> >> assert/de-assert of the signal handled by SYSC_USB_PWRRDY was implemented as a reset signal.
> >>
> >> The USB modules available on the RZ/G3S SoC that need this bit set are:
> >> - USB ch0 (supporting host and peripheral mode)
> >> - USB ch2 (supporting host mode)
> >> - USBPHY control
> >>
> >> As the USBPHY control is the root device for all the other USB
> >> channels (USB ch0, USB ch1) add support to set the PWRRDY for the USB
> >> area when initializing the USBPHY control. As this is done though reset signals get the reset array
> in the USBPHY control driver.
> >>
> >
> > Comment applicable, if the USB PWRRDY signal is modelled as reset signal.
> >
> > There is no user for this patch. The first user is RZ/G3S and is there is no support yet.
> > I think you should merge this patch with next one so that there is a
> > user(RZ/G3S) for this patch.
> 
> I have nothing against to squash it. I was seeing it differently:
> - this patch prepares the ground for the addition of RZ/G3S
> - the next patch just enables the support for RZ/G3S
> 
> It looks to me more modular like this, patches are simpler, easier for review and each issue is
> described differently in patch description.
> 
> I can do it either ways.

If it is complicated restructuring for supporting new SoC then it make sense to split into
a number of patches.

This is straight forward. You need a device quirk to support RZ/G3S.
So this kind of changes can go in a single patch, as on second patch
you are just filling device entry.

Previously I got feedback from mainline to add a patch, if there is a user.

Cheers,
Biju


> >> Signed-off-by: Claudiu Beznea <claudiu.beznea.uj@xxxxxxxxxxxxxx>
> >> ---
> >>  drivers/reset/reset-rzg2l-usbphy-ctrl.c | 2 +-
> >>  1 file changed, 1 insertion(+), 1 deletion(-)
> >>
> >> diff --git a/drivers/reset/reset-rzg2l-usbphy-ctrl.c
> >> b/drivers/reset/reset-rzg2l-usbphy-ctrl.c
> >> index 1cd157f4f03b..8b64c12f3bec 100644
> >> --- a/drivers/reset/reset-rzg2l-usbphy-ctrl.c
> >> +++ b/drivers/reset/reset-rzg2l-usbphy-ctrl.c
> >> @@ -132,7 +132,7 @@ static int rzg2l_usbphy_ctrl_probe(struct platform_device *pdev)
> >>  	if (IS_ERR(regmap))
> >>  		return PTR_ERR(regmap);
> >>
> >> -	priv->rstc = devm_reset_control_get_exclusive(&pdev->dev, NULL);
> >> +	priv->rstc = devm_reset_control_array_get_exclusive(&pdev->dev);
> >>  	if (IS_ERR(priv->rstc))
> >>  		return dev_err_probe(dev, PTR_ERR(priv->rstc),
> >>  				     "failed to get reset\n");
> >> --
> >> 2.39.2
> >




[Index of Archives]     [Linux Media]     [Linux Input]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Old Linux USB Devel Archive]

  Powered by Linux