[PATCH 1/2] USB: EHCI: ehci-mv: switch the HSIC HCI to HSIC mode

[PATCH 1/2] USB: EHCI: ehci-mv: switch the HSIC HCI to HSIC mode

[Lubomir Rintel]

Turns out the undocumented and reserved bits of port status/control
register of the root port need to be set to use the HCI in HSIC mode.

Typically the firmware does this, but that is not always good enough,
because the bits get lost if the HSIC clock is disabled (e.g. when
ehci-mv is build as a module).

This supplements commit 7b104f890ade ("USB: EHCI: ehci-mv: add HSIC
support").

Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
---
 drivers/usb/host/ehci-mv.c | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/drivers/usb/host/ehci-mv.c b/drivers/usb/host/ehci-mv.c
index bd4f6ef534d96..ddb668963955f 100644
--- a/drivers/usb/host/ehci-mv.c
+++ b/drivers/usb/host/ehci-mv.c
@@ -110,6 +110,7 @@ static int mv_ehci_probe(struct platform_device *pdev)
 	struct resource *r;
 	int retval = -ENODEV;
 	u32 offset;
+	u32 status;
 
 	if (usb_disabled())
 		return -ENODEV;
@@ -213,6 +214,14 @@ static int mv_ehci_probe(struct platform_device *pdev)
 		device_wakeup_enable(hcd->self.controller);
 	}
 
+	if (of_usb_get_phy_mode(pdev->dev.of_node) == USBPHY_INTERFACE_MODE_HSIC) {
+		status = ehci_readl(ehci, &ehci->regs->port_status[0]);
+		/* These "reserved" bits actually enable HSIC mode. */
+		status |= BIT(25);
+		status &= ~GENMASK(31, 30);
+		ehci_writel(ehci, status, &ehci->regs->port_status[0]);
+	}
+
 	dev_info(&pdev->dev,
 		 "successful find EHCI device with regs 0x%p irq %d"
 		 " working in %s mode\n", hcd->regs, hcd->irq,
-- 
2.25.1

[PATCH 2/2] USB: EHCI: ehci-mv: use a unique bus name

[Lubomir Rintel]

In case there are multiple Marvell EHCI blocks in system, we need a
different bus name for each one. Otherwise debugfs gets mildly upset about
a directory name in usb/ehci:

  debugfs: Directory 'mv ehci' with parent 'ehci' already present!

Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
---
 drivers/usb/host/ehci-mv.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/usb/host/ehci-mv.c b/drivers/usb/host/ehci-mv.c
index ddb668963955f..1300c457d9ed6 100644
--- a/drivers/usb/host/ehci-mv.c
+++ b/drivers/usb/host/ehci-mv.c
@@ -115,7 +115,7 @@ static int mv_ehci_probe(struct platform_device *pdev)
 	if (usb_disabled())
 		return -ENODEV;
 
-	hcd = usb_create_hcd(&ehci_platform_hc_driver, &pdev->dev, "mv ehci");
+	hcd = usb_create_hcd(&ehci_platform_hc_driver, &pdev->dev, dev_name(&pdev->dev));
 	if (!hcd)
 		return -ENOMEM;
 
-- 
2.25.1

Re: [PATCH 1/2] USB: EHCI: ehci-mv: switch the HSIC HCI to HSIC mode

[Alan Stern]

On Mon, 9 Mar 2020, Lubomir Rintel wrote:

> Turns out the undocumented and reserved bits of port status/control
> register of the root port need to be set to use the HCI in HSIC mode.
> 
> Typically the firmware does this, but that is not always good enough,
> because the bits get lost if the HSIC clock is disabled (e.g. when
> ehci-mv is build as a module).
> 
> This supplements commit 7b104f890ade ("USB: EHCI: ehci-mv: add HSIC
> support").
> 
> Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>

Acked-by: Alan Stern <stern@rowland.harvard.edu>

> ---
>  drivers/usb/host/ehci-mv.c | 9 +++++++++
>  1 file changed, 9 insertions(+)
> 
> diff --git a/drivers/usb/host/ehci-mv.c b/drivers/usb/host/ehci-mv.c
> index bd4f6ef534d96..ddb668963955f 100644
> --- a/drivers/usb/host/ehci-mv.c
> +++ b/drivers/usb/host/ehci-mv.c
> @@ -110,6 +110,7 @@ static int mv_ehci_probe(struct platform_device *pdev)
>  	struct resource *r;
>  	int retval = -ENODEV;
>  	u32 offset;
> +	u32 status;
>  
>  	if (usb_disabled())
>  		return -ENODEV;
> @@ -213,6 +214,14 @@ static int mv_ehci_probe(struct platform_device *pdev)
>  		device_wakeup_enable(hcd->self.controller);
>  	}
>  
> +	if (of_usb_get_phy_mode(pdev->dev.of_node) == USBPHY_INTERFACE_MODE_HSIC) {
> +		status = ehci_readl(ehci, &ehci->regs->port_status[0]);
> +		/* These "reserved" bits actually enable HSIC mode. */
> +		status |= BIT(25);
> +		status &= ~GENMASK(31, 30);
> +		ehci_writel(ehci, status, &ehci->regs->port_status[0]);
> +	}
> +
>  	dev_info(&pdev->dev,
>  		 "successful find EHCI device with regs 0x%p irq %d"
>  		 " working in %s mode\n", hcd->regs, hcd->irq,
> 

Re: [PATCH 2/2] USB: EHCI: ehci-mv: use a unique bus name

[Alan Stern]

On Mon, 9 Mar 2020, Lubomir Rintel wrote:

> In case there are multiple Marvell EHCI blocks in system, we need a
> different bus name for each one. Otherwise debugfs gets mildly upset about
> a directory name in usb/ehci:
> 
>   debugfs: Directory 'mv ehci' with parent 'ehci' already present!
> 
> Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>

Acked-by: Alan Stern <stern@rowland.harvard.edu>

> ---
>  drivers/usb/host/ehci-mv.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/usb/host/ehci-mv.c b/drivers/usb/host/ehci-mv.c
> index ddb668963955f..1300c457d9ed6 100644
> --- a/drivers/usb/host/ehci-mv.c
> +++ b/drivers/usb/host/ehci-mv.c
> @@ -115,7 +115,7 @@ static int mv_ehci_probe(struct platform_device *pdev)
>  	if (usb_disabled())
>  		return -ENODEV;
>  
> -	hcd = usb_create_hcd(&ehci_platform_hc_driver, &pdev->dev, "mv ehci");
> +	hcd = usb_create_hcd(&ehci_platform_hc_driver, &pdev->dev, dev_name(&pdev->dev));
>  	if (!hcd)
>  		return -ENOMEM;
>  
>