Hi,
John Youn <johnyoun@xxxxxxxxxxxx> writes:
> ENDXFER polling is available on version 3.10a and later of the
> DWC_usb3 (USB 3.0) controller. With this feature, the software can poll
> the CMDACT bit in the DEPCMD register after issuing an ENDXFER command.
> This feature is enabled by writing GUCTL2[14].
>
> This feature is NOT available on the DWC_usb31 (USB 3.1) IP.
>
> Signed-off-by: John Youn <johnyoun@xxxxxxxxxxxx>
> ---
> drivers/usb/dwc3/core.c | 11 +++++++++++
> drivers/usb/dwc3/core.h | 4 ++++
> drivers/usb/dwc3/gadget.c | 31 ++++++++++++++++++++++++++++++-
> 3 files changed, 45 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c
> index 71ac7c2..057739d 100644
> --- a/drivers/usb/dwc3/core.c
> +++ b/drivers/usb/dwc3/core.c
> @@ -704,6 +704,17 @@ static int dwc3_core_init(struct dwc3 *dwc)
> break;
> }
>
> + /*
> + * ENDXFER polling is available on version 3.10a and later of
> + * the DWC_usb3 controller. It is NOT available in the
> + * DWC_usb31 controller.
> + */
> + if (!dwc3_is_usb31(dwc) && dwc->revision >= DWC3_REVISION_310A) {
> + reg = dwc3_readl(dwc->regs, DWC3_GUCTL2);
> + reg |= DWC3_GUCTL2_RST_ACTBITLATER;
> + dwc3_writel(dwc->regs, DWC3_GUCTL2, reg);
> + }
> +
> return 0;
>
> err4:
> diff --git a/drivers/usb/dwc3/core.h b/drivers/usb/dwc3/core.h
> index 002e647..b2317e7 100644
> --- a/drivers/usb/dwc3/core.h
> +++ b/drivers/usb/dwc3/core.h
> @@ -109,6 +109,7 @@
> #define DWC3_GPRTBIMAP_HS1 0xc184
> #define DWC3_GPRTBIMAP_FS0 0xc188
> #define DWC3_GPRTBIMAP_FS1 0xc18c
> +#define DWC3_GUCTL2 0xc19c
>
> #define DWC3_VER_NUMBER 0xc1a0
> #define DWC3_VER_TYPE 0xc1a4
> @@ -288,6 +289,9 @@
> #define DWC3_GFLADJ_30MHZ_SDBND_SEL (1 << 7)
> #define DWC3_GFLADJ_30MHZ_MASK 0x3f
>
> +/* Global User Control Register 2 */
> +#define DWC3_GUCTL2_RST_ACTBITLATER (1 << 14)
> +
> /* Device Configuration Register */
> #define DWC3_DCFG_DEVADDR(addr) ((addr) << 3)
> #define DWC3_DCFG_DEVADDR_MASK DWC3_DCFG_DEVADDR(0x7f)
> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
> index c6fe9a3..823e65d 100644
> --- a/drivers/usb/dwc3/gadget.c
> +++ b/drivers/usb/dwc3/gadget.c
> @@ -2202,6 +2202,7 @@ static void dwc3_stop_active_transfer(struct dwc3 *dwc, u32 epnum, bool force)
> struct dwc3_gadget_ep_cmd_params params;
> u32 cmd;
> int ret;
> + unsigned int timeout;
>
> dep = dwc->eps[epnum];
>
> @@ -2225,6 +2226,14 @@ static void dwc3_stop_active_transfer(struct dwc3 *dwc, u32 epnum, bool force)
> *
> * - Issue EndTransfer WITH CMDIOC bit set
> * - Wait 100us
> + *
> + * As of IP version 3.10a of the DWC_usb3 IP, the controller
> + * supports a mode to work around the above limitation. The
> + * software can poll the CMDACT bit in the DEPCMD register
> + * after issuing a EndTransfer command. This mode is enabled
> + * by writing GUCTL2[14].
> + *
> + * This mode is NOT available on the DWC_usb31 IP.
> */
>
> cmd = DWC3_DEPCMD_ENDTRANSFER;
> @@ -2236,7 +2245,27 @@ static void dwc3_stop_active_transfer(struct dwc3 *dwc, u32 epnum, bool force)
> WARN_ON_ONCE(ret);
> dep->resource_index = 0;
> dep->flags &= ~DWC3_EP_BUSY;
> - udelay(100);
> +
> + if (dwc3_is_usb31(dwc) || dwc->revision < DWC3_REVISION_310A) {
> + udelay(100);
> + return;
> + }
> +
> + timeout = 100;
> +
> + do {
> + cmd = dwc3_readl(dep->regs, DWC3_DEPCMD);
> + if (!(cmd & DWC3_DEPCMD_CMDACT))
> + break;
we're already polling CMDACT bit when we send the command. Do we _need_
to poll it again? Seems like the check above for running udelay(100) is
enough. Just remove the return call from there.
> + timeout--;
> + if (!timeout) {
> + dev_err(dwc->dev, "EndTransfer didn't complete\n");
> + break;
> + }
> +
> + udelay(1);
> + } while (1);
not into infinite loops, actually :-) See my recent patches removing
these infinite loops and udelay(1) calls.
--
balbi
Attachment:
signature.asc
Description: PGP signature
