Re: [PATCH 2/2] drivers: phy: qcom: ipq806x-usb: conver latch function to pool macro

[Vinod Koul]

On 23-01-22, 15:46, Ansuel Smith wrote:
> On Sun, Jan 23, 2022 at 05:22:18PM +0530, Vinod Koul wrote:
> > On 17-01-22, 01:26, Ansuel Smith wrote:
> > > Convert latch function to readl pool macro to tidy things up.
> > > 
> > > Signed-off-by: Ansuel Smith <ansuelsmth@xxxxxxxxx>
> > > ---
> > >  drivers/phy/qualcomm/phy-qcom-ipq806x-usb.c | 17 +++++------------
> > >  1 file changed, 5 insertions(+), 12 deletions(-)
> > > 
> > > diff --git a/drivers/phy/qualcomm/phy-qcom-ipq806x-usb.c b/drivers/phy/qualcomm/phy-qcom-ipq806x-usb.c
> > > index 6788e0e8272a..ab2d1431546d 100644
> > > --- a/drivers/phy/qualcomm/phy-qcom-ipq806x-usb.c
> > > +++ b/drivers/phy/qualcomm/phy-qcom-ipq806x-usb.c
> > > @@ -112,6 +112,9 @@
> > >  #define SS_CR_READ_REG				BIT(0)
> > >  #define SS_CR_WRITE_REG				BIT(0)
> > >  
> > > +#define LATCH_SLEEP				40
> > > +#define LATCH_TIMEOUT				100
> > > +
> > >  struct usb_phy {
> > >  	void __iomem		*base;
> > >  	struct device		*dev;
> > > @@ -156,19 +159,9 @@ static inline void usb_phy_write_readback(struct usb_phy *phy_dwc3,
> > >  
> > >  static int wait_for_latch(void __iomem *addr)
> > >  {
> > > -	u32 retry = 10;
> > > -
> > > -	while (true) {
> > > -		if (!readl(addr))
> > > -			break;
> > 
> > we break if read returns non zero value...
> > 
> > Do you know what is the value expected?
> >
> 
> If I understand the logic here, we write a value and we wait for it to
> get applied. To confirm that we execute a writel and then we readl the
> same address until it does return a value. That is the way used to
> understand that the write process has finished and that the value has
> been applied/we can write again.
> 
> > > -
> > > -		if (--retry == 0)
> > > -			return -ETIMEDOUT;
> > > -
> > > -		usleep_range(10, 20);
> > > -	}
> > > +	u32 val;
> > 
> > Okay this contains garbage..
> 
> I think I didn't understand, val value will get replaced by readl in
> the pool_timeout function.
> 
> > >  
> > > -	return 0;
> > > +	return readl_poll_timeout(addr, val, !val, LATCH_SLEEP, LATCH_TIMEOUT);
> > 
> > and we are waiting for it read a garbage value!
> > 
> 
> Again could be very confused and wrong but the pool_timeout macro does
> the exact same thing of the wait_for_latch function with th only
> difference of handling the sleep differently. We put in val the return
> of readl and the break condition as !val. Or I didn't understand the
> concern about garbage value.

Sorry I read the readl_poll_timeout wrongly, this seems correct, I will
pick these now.

-- 
~Vinod