On 10/14/2015 3:58 PM, Douglas Anderson wrote: > From code inspection, it appears to be unsafe to do a read-modify-write > of PCGCTL in dwc2_port_resume(). Let's make sure the spinlock is held > around this operation. > > Signed-off-by: Douglas Anderson <dianders at chromium.org> > --- > drivers/usb/dwc2/hcd.c | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/drivers/usb/dwc2/hcd.c b/drivers/usb/dwc2/hcd.c > index af4e4a2..e79baf7 100644 > --- a/drivers/usb/dwc2/hcd.c > +++ b/drivers/usb/dwc2/hcd.c > @@ -1500,6 +1500,8 @@ static void dwc2_port_resume(struct dwc2_hsotg *hsotg) > u32 hprt0; > u32 pcgctl; > > + spin_lock_irqsave(&hsotg->lock, flags); > + > /* > * If hibernation is supported, Phy clock is already resumed > * after registers restore. > @@ -1508,10 +1510,11 @@ static void dwc2_port_resume(struct dwc2_hsotg *hsotg) > pcgctl = dwc2_readl(hsotg->regs + PCGCTL); > pcgctl &= ~PCGCTL_STOPPCLK; > dwc2_writel(pcgctl, hsotg->regs + PCGCTL); > + spin_unlock_irqrestore(&hsotg->lock, flags); > usleep_range(20000, 40000); > + spin_lock_irqsave(&hsotg->lock, flags); > } > > - spin_lock_irqsave(&hsotg->lock, flags); > hprt0 = dwc2_read_hprt0(hsotg); > hprt0 |= HPRT0_RES; > hprt0 &= ~HPRT0_SUSP; > Acked-by: John Youn <johnyoun at synopsys.com> John
