> -----Original Message-----
> From: Felipe Balbi <balbi@xxxxxxxxxx>
> Sent: Wednesday, August 18, 2021 3:24 PM
> To: Shubhrajyoti Datta <shubhraj@xxxxxxxxxx>
> Cc: linux-usb@xxxxxxxxxxxxxxx; gregkh@xxxxxxxxxxxxxxxxxxx; Michal Simek
> <michals@xxxxxxxxxx>; git <git@xxxxxxxxxx>
> Subject: Re: [PATCH] usb: gadget: udc-xilinx: Add clock support
>
>
> Hi,
>
> Shubhrajyoti Datta <shubhrajyoti.datta@xxxxxxxxxx> writes:
> > Currently the driver depends on the bootloader to enable the clocks.
> > Add support for clocking. The patch enables the clock at probe and
> > disables them at remove.
> >
> > Signed-off-by: Shubhrajyoti Datta <shubhrajyoti.datta@xxxxxxxxxx>
> > ---
> > drivers/usb/gadget/udc/udc-xilinx.c | 24 ++++++++++++++++++++++++
> > 1 file changed, 24 insertions(+)
> >
> > diff --git a/drivers/usb/gadget/udc/udc-xilinx.c
> > b/drivers/usb/gadget/udc/udc-xilinx.c
> > index fb4ffedd6f0d..30070a488c87 100644
> > --- a/drivers/usb/gadget/udc/udc-xilinx.c
> > +++ b/drivers/usb/gadget/udc/udc-xilinx.c
> > @@ -11,6 +11,7 @@
> > * USB peripheral controller (at91_udc.c).
> > */
> >
> > +#include <linux/clk.h>
> > #include <linux/delay.h>
> > #include <linux/device.h>
> > #include <linux/dma-mapping.h>
> > @@ -171,6 +172,7 @@ struct xusb_ep {
> > * @addr: the usb device base address
> > * @lock: instance of spinlock
> > * @dma_enabled: flag indicating whether the dma is included in the
> > system
> > + * @clk: pointer to struct clk
> > * @read_fn: function pointer to read device registers
> > * @write_fn: function pointer to write to device registers
> > */
> > @@ -188,6 +190,7 @@ struct xusb_udc {
> > void __iomem *addr;
> > spinlock_t lock;
> > bool dma_enabled;
> > + struct clk *clk;
> >
> > unsigned int (*read_fn)(void __iomem *);
> > void (*write_fn)(void __iomem *, u32, u32); @@ -2092,6 +2095,26 @@
> > static int xudc_probe(struct platform_device *pdev)
> > udc->gadget.ep0 = &udc->ep[XUSB_EP_NUMBER_ZERO].ep_usb;
> > udc->gadget.name = driver_name;
> >
> > + udc->clk = devm_clk_get(&pdev->dev, "s_axi_aclk");
> > + if (IS_ERR(udc->clk)) {
> > + if (PTR_ERR(udc->clk) != -ENOENT) {
> > + ret = PTR_ERR(udc->clk);
> > + goto fail;
> > + }
> > +
> > + /*
> > + * Clock framework support is optional, continue on,
> > + * anyways if we don't find a matching clock
> > + */
> > + udc->clk = NULL;
>
> should it be, though? Might be a good idea to add fixed-clock instances to the
> boards still depending on clock framework. Maybe that can be done over time,
> but worth considering anyhow.
But for backward compatibility , I think it will be good to have the support instead of forcing the fixed-clock node.
>
> --
> balbi
