Hi Sergei,
Sergei Shtylyov <sergei.shtylyov@xxxxxxxxxxxxxxxxxx> wrote on Fri, 11
Jan 2019 21:03:01 +0300:
> Hello!
>
> On 01/11/2019 04:31 PM, Miquel Raynal wrote:
>
> > No need to initialize the PHY from the driver's probe. It is done by
> > the core automatically and doing it twice would increment the
> > phy->powercount counter to 2 instead of 1. During later suspend
> > operation, the counter will be decremented to one, no phy->power_off()
> > will occur and worst than that, the following phy->power_on() at
>
> Worse.
Noted
>
> > resume time will be also skipped, failing the whole S2RAM operation.
> >
> > Signed-off-by: Miquel Raynal <miquel.raynal@xxxxxxxxxxx>
> > ---
> > drivers/usb/host/ehci-orion.c | 26 +++-----------------------
> > 1 file changed, 3 insertions(+), 23 deletions(-)
> >
> > diff --git a/drivers/usb/host/ehci-orion.c b/drivers/usb/host/ehci-orion.c
> > index 1ad72647a069..3109f082949e 100644
> > --- a/drivers/usb/host/ehci-orion.c
> > +++ b/drivers/usb/host/ehci-orion.c
> > @@ -257,15 +257,7 @@ static int ehci_orion_drv_probe(struct platform_device *pdev)
> > if (IS_ERR(priv->phy)) {
> > err = PTR_ERR(priv->phy);
> > if (err != -ENOSYS)
> > - goto err_phy_get;
> > - } else {
> > - err = phy_init(priv->phy);
> > - if (err)
> > - goto err_phy_init;
> > -
> > - err = phy_power_on(priv->phy);
> > - if (err)
> > - goto err_phy_power_on;
> > + goto err_dis_clk;
>
> Familiar code in unfamiliar place. Somebody must have blindly copied it... :-)
Actually a git blame shows that this code is there since 2014, 4 years
before the HCD core supported PHYs. This driver was probably forgotten
during the process.
Thanks,
Miquèl
