On Thu, Jul 26, 2012 at 08:29:06AM +0000, Gupta, Ajay Kumar wrote: > Hi, > > On Thu, Jul 26, 2012 at 08:22:35AM +0000, Gupta, Ajay Kumar wrote: > > > > > @@ -408,7 +408,7 @@ static int am35x_musb_exit(struct musb *musb) > > > > > data->set_phy_power(0); > > > > > > > > > > usb_put_phy(musb->xceiv); > > > > > - usb_nop_xceiv_unregister(); > > > > > + usb_nop_xceiv_unregister(musb->xceiv); > > > > > > > > Dude, this is so wrong... You don't know if there are other users of > > > > the nop xceiv in the system. If there are other users which already > > > > have claimed id 0, you will try to create duplicated directories under > > sysfs. > > > > MUSB's ID has nothing to do with NOP's ID. > > > > > > I could not understood this comment as I am not using id. Please clarify. > > > > sorry, my bad. I put the comment on the wrong hunk: > > > > > > > @@ -364,7 +364,7 @@ static int am35x_musb_init(struct musb *musb) > > > > > if (!rev) > > > > > return -ENODEV; > > > > > > > > > > - usb_nop_xceiv_register(); > > > > > + usb_nop_xceiv_register(musb->id); > > > > > musb->xceiv = usb_get_phy(USB_PHY_TYPE_USB2); > > > > > if (IS_ERR_OR_NULL(musb->xceiv)) > > > > > return -ENODEV; > > > > here. You pass musb->id to become NOP's ID. > > Got this, so something like musb_ida is needed even for NOP. Right? that's correct. But I guess you can avoid nop_get_id() and build that into nop_register and nop_unregister() calls, so that the changes are local to nop xceiv only. -- balbi
Attachment:
signature.asc
Description: Digital signature
