[PATCH libbsd 10/11] ehci_imx: Set/clear ENHOSTDISCONNECT in USB PHY.

Gedare Bloom gedare at rtems.org
Thu Apr 2 15:13:28 UTC 2020


Maybe file a ticket against this for the future.

On Thu, Apr 2, 2020 at 8:43 AM Christian Mauderer
<christian.mauderer at embedded-brains.de> wrote:
>
> This is not a nice solution but it should work on all chips. As soon as
> FreeBSD has a nice solution via the USB PHY driver, this should be
> replaced.
>
> Update #3869.
> ---
>  freebsd/sys/dev/usb/controller/ehci_imx.c | 59 +++++++++++++++++++++++++++++++
>  1 file changed, 59 insertions(+)
>
> diff --git a/freebsd/sys/dev/usb/controller/ehci_imx.c b/freebsd/sys/dev/usb/controller/ehci_imx.c
> index d158df1d..544a63b5 100644
> --- a/freebsd/sys/dev/usb/controller/ehci_imx.c
> +++ b/freebsd/sys/dev/usb/controller/ehci_imx.c
> @@ -50,6 +50,9 @@ __FBSDID("$FreeBSD$");
>
>  #include <dev/ofw/ofw_bus.h>
>  #include <dev/ofw/ofw_bus_subr.h>
> +#ifdef __rtems__
> +#include <dev/fdt/fdt_common.h>
> +#endif /* __rtems__ */
>
>  #include <dev/usb/usb.h>
>  #include <dev/usb/usbdi.h>
> @@ -270,6 +273,9 @@ struct imx_ehci_softc {
>         device_t        dev;
>         struct resource *ehci_mem_res;  /* EHCI core regs. */
>         struct resource *ehci_irq_res;  /* EHCI core IRQ. */
> +#ifdef __rtems__
> +       void            *phy_regs;
> +#endif /* __rtems__ */
>  };
>
>  static struct ofw_compat_data compat_data[] = {
> @@ -380,6 +386,32 @@ imx_ehci_disable_oc(struct imx_ehci_softc *sc)
>         index = usbmprops[1];
>         imx_usbmisc_set_ctrl(usbmdev, index, USBNC_OVER_CUR_DIS);
>  }
> +#ifdef __rtems__
> +#define BUS_SPACE_PHYSADDR(res, offs) \
> +       ((u_int)(rman_get_start(res)+(offs)))
> +
> +static uint16_t
> +imx_ehci_get_port_speed_portsc(struct ehci_softc *sc, uint16_t index)
> +{
> +       uint32_t v;
> +       struct imx_ehci_softc *isc = (struct imx_ehci_softc *)sc;
> +
> +       v = ehci_get_port_speed_portsc(sc, index);
> +
> +       /* Set/clear ENHOSTDISCONDETECT in USBPHY CTRL register */
> +       /* FIXME: This is a very hacky method. It should be done by the phy
> +        * driver instead. */
> +       volatile uint32_t *ctrl_set = isc->phy_regs + 0x34;
> +       volatile uint32_t *ctrl_clr = isc->phy_regs + 0x38;
> +       if (v == UPS_HIGH_SPEED) {
> +               *ctrl_set = 0x2;
> +       } else {
> +               *ctrl_clr = 0x2;
> +       }
> +
> +       return (v);
> +}
> +#endif /* __rtems__ */
>
>  static int
>  imx_ehci_attach(device_t dev)
> @@ -474,6 +506,33 @@ imx_ehci_attach(device_t dev)
>         esc->sc_flags |= EHCI_SCFLG_NORESTERM | EHCI_SCFLG_TT;
>         esc->sc_vendor_post_reset = imx_ehci_post_reset;
>         esc->sc_vendor_get_port_speed = ehci_get_port_speed_portsc;
> +#ifdef __rtems__
> +       /*
> +        * FIXME: This is a big hack to get the PHY regs and set one bit during
> +        * changes of the port speed.
> +        */
> +       sc->phy_regs = 0;
> +       if (OF_hasprop(ofw_bus_get_node(sc->dev), "fsl,usbphy")) {
> +               phandle_t *cells;
> +               int len;
> +               phandle_t phynode;
> +
> +               cells = NULL;
> +               len = OF_getencprop_alloc_multi(ofw_bus_get_node(sc->dev),
> +                   "fsl,usbphy", sizeof(*cells), (void **)&cells);
> +               if (len > 0
> +                   && (phynode = OF_node_from_xref(cells[0])) != -1 &&
> +                   ofw_bus_node_is_compatible(phynode, "fsl,imx6ul-usbphy")) {
> +                       u_long base, size;
> +                       if (fdt_regsize(phynode, &base, &size) == 0) {
> +                               sc->phy_regs = (void*)base;
> +                               esc->sc_vendor_get_port_speed =
> +                                   imx_ehci_get_port_speed_portsc;
> +                       }
> +               }
> +               OF_prop_free(cells);
> +       }
> +#endif /* __rtems__ */
>
>         err = ehci_init(esc);
>         if (err != 0) {
> --
> 2.16.4
>
> _______________________________________________
> devel mailing list
> devel at rtems.org
> http://lists.rtems.org/mailman/listinfo/devel


More information about the devel mailing list