Re: [PATCH 1/2] usb: xhci-mtk: rebuild xhci_mtk_setup()

[Chunfeng Yun]

On Tue, 2017-03-07 at 16:57 +0200, Mathias Nyman wrote:
> On 07.03.2017 08:57, Chunfeng Yun wrote:
> > simplify xhci_mtk_setup() and add xhci_mtk_start() for
> > xhci_driver_overrides struct
> >
> 
> Code itself looks fine, but it's bit unclear for me what the benefit of this is?
> 
> > Signed-off-by: Chunfeng Yun <chunfeng.yun@xxxxxxxxxxxx>
> > ---
> >   drivers/usb/host/xhci-mtk.c |   16 +++++++++++-----
> >   1 file changed, 11 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/usb/host/xhci-mtk.c b/drivers/usb/host/xhci-mtk.c
> > index 67d5dc7..9636884 100644
> > --- a/drivers/usb/host/xhci-mtk.c
> > +++ b/drivers/usb/host/xhci-mtk.c
> > @@ -381,8 +381,10 @@ static int usb_wakeup_of_property_parse(struct xhci_hcd_mtk *mtk,
> >   }
> >
> >   static int xhci_mtk_setup(struct usb_hcd *hcd);
> > +static int xhci_mtk_start(struct usb_hcd *hcd);
> >   static const struct xhci_driver_overrides xhci_mtk_overrides __initconst = {
> >   	.reset = xhci_mtk_setup,
> > +	.start = xhci_mtk_start,
> >   };
> >
> >   static struct hc_driver __read_mostly xhci_mtk_hc_driver;
> > @@ -492,7 +494,6 @@ static void xhci_mtk_quirks(struct device *dev, struct xhci_hcd *xhci)
> >   /* called during probe() after chip reset completes */
> >   static int xhci_mtk_setup(struct usb_hcd *hcd)
> >   {
> > -	struct xhci_hcd *xhci = hcd_to_xhci(hcd);
> >   	struct xhci_hcd_mtk *mtk = hcd_to_mtk(hcd);
> >   	int ret;
> >
> > @@ -502,9 +503,14 @@ static int xhci_mtk_setup(struct usb_hcd *hcd)
> >   			return ret;
> >   	}
> >
> > -	ret = xhci_gen_setup(hcd, xhci_mtk_quirks);
> > -	if (ret)
> > -		return ret;
> > +	return xhci_gen_setup(hcd, xhci_mtk_quirks);
> > +}
> > +
> > +static int xhci_mtk_start(struct usb_hcd *hcd)
> > +{
> > +	struct xhci_hcd *xhci = hcd_to_xhci(hcd);
> > +	struct xhci_hcd_mtk *mtk = hcd_to_mtk(hcd);
> > +	int ret;
> >
> >   	if (usb_hcd_is_primary_hcd(hcd)) {
> >   		mtk->num_u3_ports = xhci->num_usb3_ports;
> 
> In theory this causes xhci_mtk_sch_init() to allocate new memory
> for mtk->sch_array every time   .start callback is called.
> 
I agree with you, it indeed a problem no matter called in .reset
or .start in theory. Maybe it's better to call it in probe().

> In practice it should not matter as .start is only called
> when hcd is added in usb core.
> 
> Still wondering if there is a reason for this change?
Just make xhci_mtk_setup() look simple.
> 
> -Mathias
> 
>