Hi,
On Tue, Jun 18, 2013 at 10:34:41AM +0200, Sebastian Andrzej Siewior wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
> On 06/18/2013 10:27 AM, Felipe Balbi wrote:
> >> diff --git a/drivers/usb/musb/musb_dsps.c
> >> b/drivers/usb/musb/musb_dsps.c index e1b661d..d9ff390 100644 ---
> >> a/drivers/usb/musb/musb_dsps.c +++
> >> b/drivers/usb/musb/musb_dsps.c @@ -415,9 +415,14 @@ static int
> >> dsps_musb_init(struct musb *musb) /* mentor core register starts
> >> at offset of 0x400 from musb base */ musb->mregs +=
> >> wrp->musb_core_offset;
> >>
> >> - /* NOP driver needs change if supporting dual instance */ -
> >> usb_nop_xceiv_register(); - musb->xceiv =
> >> usb_get_phy(USB_PHY_TYPE_USB2); + if (!glue->dev->of_node) { +
> >> /* This hack works only for a single instance. */ +
> >> usb_nop_xceiv_register(); + musb->xceiv =
> >> usb_get_phy(USB_PHY_TYPE_USB2);
> >
> > I think you can drop this altogether, am335x is DT-only anyway :-)
>
> Yes, but this is also used by:
>
> $ git grep "musb-ti81xx"
> arch/arm/mach-omap2/usb-musb.c: name = "musb-ti81xx";
> drivers/usb/musb/musb_dsps.c: .name = "musb-ti81xx",
>
> Is that one also am33xx?
my bad, it's not :-)
> >> + } else { + musb->xceiv =
> >> devm_usb_get_phy_by_phandle(glue->dev, "phys", +
> >> musb->config->instance); + }
> >
> > after doing all this, perhaps we should re-factor phy_get into
> > musb_core.c, so that we can remove this sort of support from all
> > glue layers.
>
> So keep this is as is and add later a phy_get into musb? I would have
> check if everyone has a phy instance but I guess so.
sure, let's keep it as is :-)
--
balbi
signature.asc
Description: Digital signature
