On Thu, Feb 23, 2012 at 10:48:12AM +0100, Julia Lawall wrote:
> On Thu, 23 Feb 2012, Felipe Balbi wrote:
>
> >Hi,
> >
> >On Sat, Feb 18, 2012 at 06:04:39PM +0100, Julia Lawall wrote:
> >>The function at91udc_probe in the file drivers/usb/gadget/at91_udc.c
> >>contains the following code:
> >>
> >> udc->iclk = clk_get(dev, "udc_clk");
> >> udc->fclk = clk_get(dev, "udpck");
> >> if (IS_ERR(udc->iclk) || IS_ERR(udc->fclk)) {
> >> DBG("clocks missing\n");
> >> retval = -ENODEV;
> >> /* NOTE: we "know" here that refcounts on these are NOPs */
> >> goto fail0b;
> >> }
> >>
> >>Why is it known here that refcounts are NOPs. When does this
> >>property cease to hold? clk_put is present in the corresponding
> >>remove function.
> >
> >I guess we need to ask AT91 maintainers as I have never worked with that
> >architecture, but I believe that if clk_get() fails, the clk won't get
> >its reference counter incremented. It depends on how they have
> >implemented their clk API.
>
> Sure, but it's an ||, so it might be that only one of them has
> failed. Maybe someone else has an idea.
aha, now I see your point. Indeed, it smells fishy.
--
balbi
Attachment:
signature.asc
Description: Digital signature
