Message ID | 20181016045846.2345-1-peter.chen@nxp.com |
---|---|
Headers | show |
Series | usb: chipidea: imx: add HSIC support | expand |
Hi Peter, On 16.10.18 07:01, Peter Chen wrote: > Most of NXP (Freescale) i.mx USB part has HSIC support, in this series, > we add support for them, it should cover all imx6 and imx7d. I have > no HSIC interface board which is supported by upstream kernel, so this > patches are only compiled ok, Frieder Schrempf, would you please > help me test it on your board? Thanks. Thank you for providing the patch! I applied it to v4.19-rc8 and tested and the LAN9730 comes up and works just fine. Here is how my devicetree looks like: usbphy_dummy: usbphy_dummy@1 { compatible = "usb-nop-xceiv"; }; [...] &usbh2 { vbus-supply = <®_usb_h2_vbus>; pinctrl-names = "idle", "active"; pinctrl-0 = <&pinctrl_usbh2_idle>; pinctrl-1 = <&pinctrl_usbh2_active>; fsl,usbphy = <&usbphy_dummy>; phy_type = "hsic"; status = "okay"; #address-cells = <1>; #size-cells = <0>; usbnet: smsc@1 { compatible = "usb424,9730"; /* Filled in by U-Boot */ mac-address = [00 00 00 00 00 00]; reg = <1>; }; }; [...] pinctrl_usbh2_idle: usbh2grp-idle { fsl,pins = < MX6QDL_PAD_RGMII_TXC__USB_H2_DATA 0x40013030 MX6QDL_PAD_RGMII_TX_CTL__USB_H2_STROBE 0x40013030 >; }; pinctrl_usbh2_active: usbh2grp-active { fsl,pins = < MX6QDL_PAD_RGMII_TXC__USB_H2_DATA 0x40013030 MX6QDL_PAD_RGMII_TX_CTL__USB_H2_STROBE 0x40017030 >; }; Are there any test cases I should try? How can I test suspend/resume? I also have some suggestions for your patch. Please see the separate email. Thanks, Frieder > > Peter Chen (4): > usb: chipidea: add flag for imx hsic implementation > usb: chipidea: imx: add HSIC support > usb: chipidea: host: override ehci->hub_control > doc: usb: ci-hdrc-usb2: Add pinctrl properties for HSIC pin groups > > .../devicetree/bindings/usb/ci-hdrc-usb2.txt | 1 + > drivers/usb/chipidea/ci_hdrc_imx.c | 153 ++++++++++++++++++--- > drivers/usb/chipidea/ci_hdrc_imx.h | 9 +- > drivers/usb/chipidea/host.c | 98 +++++++++++++ > drivers/usb/chipidea/usbmisc_imx.c | 131 ++++++++++++++++++ > include/linux/usb/chipidea.h | 3 + > 6 files changed, 376 insertions(+), 19 deletions(-) >
Hi Peter, please see my comments below. For reference I also pushed the changes here: https://github.com/fschrempf/linux/commits/usb-hsic-test On 16.10.18 07:01, Peter Chen wrote: > To support imx HSIC, there are some special requirement: > - The HSIC pad is 1.2v, it may need to supply from external > - The data/strobe pin needs to be pulled down first, and after > host mode is initialized, the strobe pin needs to be pulled up > - During the USB suspend/resume, special setting is needed > > Signed-off-by: Peter Chen <peter.chen@nxp.com> > --- > drivers/usb/chipidea/ci_hdrc_imx.c | 153 +++++++++++++++++++++++++++++++++---- > drivers/usb/chipidea/ci_hdrc_imx.h | 9 ++- > drivers/usb/chipidea/usbmisc_imx.c | 131 +++++++++++++++++++++++++++++++ > 3 files changed, 274 insertions(+), 19 deletions(-) > > diff --git a/drivers/usb/chipidea/ci_hdrc_imx.c b/drivers/usb/chipidea/ci_hdrc_imx.c > index 09b37c0d075d..d566771fc77a 100644 > --- a/drivers/usb/chipidea/ci_hdrc_imx.c > +++ b/drivers/usb/chipidea/ci_hdrc_imx.c > @@ -85,6 +85,9 @@ struct ci_hdrc_imx_data { > bool supports_runtime_pm; > bool override_phy_control; > bool in_lpm; > + struct pinctrl *pinctrl; > + struct pinctrl_state *pinctrl_hsic_active; > + struct regulator *hsic_pad_regulator; > /* SoC before i.mx6 (except imx23/imx28) needs three clks */ > bool need_three_clks; > struct clk *clk_ipg; > @@ -245,19 +248,58 @@ static void imx_disable_unprepare_clks(struct device *dev) > } > } > > +static int ci_hdrc_imx_notify_event(struct ci_hdrc *ci, unsigned int event) > +{ > + struct device *dev = ci->dev->parent; > + struct ci_hdrc_imx_data *data = dev_get_drvdata(dev); > + int ret = 0; > + > + switch (event) { > + case CI_HDRC_IMX_HSIC_ACTIVE_EVENT: > + if (!IS_ERR(data->pinctrl) && > + !IS_ERR(data->pinctrl_hsic_active)) { If we make the pinctrl mandatory in case of HSIC as proposed below, we don't need the checks here. > + ret = pinctrl_select_state(data->pinctrl, > + data->pinctrl_hsic_active); > + if (ret) > + dev_err(dev, > + "hsic_active select failed, err=%d\n", > + ret); > + return ret; > + } > + break; > + case CI_HDRC_IMX_HSIC_SUSPEND_EVENT: > + if (data->usbmisc_data) { > + ret = imx_usbmisc_hsic_set_connect(data->usbmisc_data); > + if (ret) > + dev_err(dev, > + "hsic_set_connect failed, err=%d\n", > + ret); > + return ret; > + } > + break; > + default: > + break; > + } > + > + return ret; > +} > + > static int ci_hdrc_imx_probe(struct platform_device *pdev) > { > struct ci_hdrc_imx_data *data; > struct ci_hdrc_platform_data pdata = { > .name = dev_name(&pdev->dev), > .capoffset = DEF_CAPOFFSET, > + .notify_event = ci_hdrc_imx_notify_event, > }; > int ret; > const struct of_device_id *of_id; > const struct ci_hdrc_imx_platform_flag *imx_platform_flag; > struct device_node *np = pdev->dev.of_node; > + struct device *dev = &pdev->dev; > + struct pinctrl_state *pinctrl_hsic_idle; > > - of_id = of_match_device(ci_hdrc_imx_dt_ids, &pdev->dev); > + of_id = of_match_device(ci_hdrc_imx_dt_ids, dev); > if (!of_id) > return -ENODEV; > > @@ -268,19 +310,48 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev) > return -ENOMEM; > > platform_set_drvdata(pdev, data); > - data->usbmisc_data = usbmisc_get_init_data(&pdev->dev); > + data->usbmisc_data = usbmisc_get_init_data(dev); > if (IS_ERR(data->usbmisc_data)) > return PTR_ERR(data->usbmisc_data); > > - ret = imx_get_clks(&pdev->dev); > + data->pinctrl = devm_pinctrl_get(dev); > + if (IS_ERR(data->pinctrl)) { > + dev_dbg(dev, "pinctrl get failed, err=%ld\n", > + PTR_ERR(data->pinctrl)); > + } else { > + pinctrl_hsic_idle = pinctrl_lookup_state(data->pinctrl, "idle"); > + if (IS_ERR(pinctrl_hsic_idle)) { > + dev_dbg(dev, > + "pinctrl_hsic_idle lookup failed, err=%ld\n", > + PTR_ERR(pinctrl_hsic_idle)); > + } else { > + ret = pinctrl_select_state(data->pinctrl, > + pinctrl_hsic_idle); > + if (ret) { > + dev_err(dev, > + "hsic_idle select failed, err=%d\n", > + ret); > + return ret; > + } > + } > + > + data->pinctrl_hsic_active = pinctrl_lookup_state(data->pinctrl, > + "active"); > + if (IS_ERR(data->pinctrl_hsic_active)) > + dev_dbg(dev, > + "pinctrl_hsic_active lookup failed, err=%ld\n", > + PTR_ERR(data->pinctrl_hsic_active)); > + } In the paragraph above, I think we should make the pinctrl settings mandatory in case of HSIC and error out if one of them is missing. Also I think we could make the code more readable by removing the nested conditions. Maybe something like this would be better? if (of_usb_get_phy_mode(dev->of_node) == USBPHY_INTERFACE_MODE_HSIC) { data->pinctrl = devm_pinctrl_get(dev); if (IS_ERR(data->pinctrl)) { dev_err(dev, "failed to get HSIC pinctrl, err=%ld\n", PTR_ERR(data->pinctrl)); return PTR_ERR(data->pinctrl); } pinctrl_hsic_idle = pinctrl_lookup_state(data->pinctrl, "idle"); if (IS_ERR(pinctrl_hsic_idle)) { dev_err(dev, "failed to get HSIC idle pinctrl," "err=%ld\n", PTR_ERR(pinctrl_hsic_idle)); return PTR_ERR(pinctrl_hsic_idle); } ret = pinctrl_select_state(data->pinctrl, pinctrl_hsic_idle); if (ret) { dev_err(dev, "failed to select HSIC idle pinctrl," "err=%d\n", ret); return ret; } data->pinctrl_hsic_active = pinctrl_lookup_state(data->pinctrl, "active"); if (IS_ERR(data->pinctrl_hsic_active)) { dev_err(dev, "failed to get HSIC active pinctrl," "err=%ld\n", PTR_ERR(data->pinctrl_hsic_active)); return PTR_ERR(data->pinctrl_hsic_active); } } > + > + ret = imx_get_clks(dev); > if (ret) > return ret; > > - ret = imx_prepare_enable_clks(&pdev->dev); > + ret = imx_prepare_enable_clks(dev); > if (ret) > return ret; > > - data->phy = devm_usb_get_phy_by_phandle(&pdev->dev, "fsl,usbphy", 0); > + data->phy = devm_usb_get_phy_by_phandle(dev, "fsl,usbphy", 0); > if (IS_ERR(data->phy)) { > ret = PTR_ERR(data->phy); > /* Return -EINVAL if no usbphy is available */ > @@ -303,42 +374,72 @@ static int ci_hdrc_imx_probe(struct platform_device *pdev) > if (pdata.flags & CI_HDRC_SUPPORTS_RUNTIME_PM) > data->supports_runtime_pm = true; > > + if (of_usb_get_phy_mode(dev->of_node) == USBPHY_INTERFACE_MODE_HSIC) { > + pdata.flags |= CI_HDRC_IMX_IS_HSIC; > + data->usbmisc_data->hsic = 1; > + data->hsic_pad_regulator = devm_regulator_get(dev, "hsic"); > + if (PTR_ERR(data->hsic_pad_regulator) == -EPROBE_DEFER) { > + ret = -EPROBE_DEFER; > + goto err_clk; > + } else if (PTR_ERR(data->hsic_pad_regulator) == -ENODEV) { > + /* no pad regualator is needed */ > + data->hsic_pad_regulator = NULL; > + } else if (IS_ERR(data->hsic_pad_regulator)) { > + dev_err(dev, "Get hsic pad regulator error: %ld\n", > + PTR_ERR(data->hsic_pad_regulator)); > + ret = PTR_ERR(data->hsic_pad_regulator); > + goto err_clk; > + } > + > + if (data->hsic_pad_regulator) { > + ret = regulator_enable(data->hsic_pad_regulator); > + if (ret) { > + dev_err(dev, > + "Fail to enable hsic pad regulator\n"); > + goto err_clk; > + } > + } > + } > + > ret = imx_usbmisc_init(data->usbmisc_data); > if (ret) { > - dev_err(&pdev->dev, "usbmisc init failed, ret=%d\n", ret); > - goto err_clk; > + dev_err(dev, "usbmisc init failed, ret=%d\n", ret); > + goto disable_hsic_regulator; > } > > - data->ci_pdev = ci_hdrc_add_device(&pdev->dev, > + data->ci_pdev = ci_hdrc_add_device(dev, > pdev->resource, pdev->num_resources, > &pdata); > if (IS_ERR(data->ci_pdev)) { > ret = PTR_ERR(data->ci_pdev); > if (ret != -EPROBE_DEFER) > - dev_err(&pdev->dev, > - "ci_hdrc_add_device failed, err=%d\n", ret); > - goto err_clk; > + dev_err(dev, "ci_hdrc_add_device failed, err=%d\n", > + ret); > + goto disable_hsic_regulator; > } > > ret = imx_usbmisc_init_post(data->usbmisc_data); > if (ret) { > - dev_err(&pdev->dev, "usbmisc post failed, ret=%d\n", ret); > + dev_err(dev, "usbmisc post failed, ret=%d\n", ret); > goto disable_device; > } > > if (data->supports_runtime_pm) { > - pm_runtime_set_active(&pdev->dev); > - pm_runtime_enable(&pdev->dev); > + pm_runtime_set_active(dev); > + pm_runtime_enable(dev); > } > > - device_set_wakeup_capable(&pdev->dev, true); > + device_set_wakeup_capable(dev, true); > > return 0; > > disable_device: > ci_hdrc_remove_device(data->ci_pdev); > +disable_hsic_regulator: > + if (data->hsic_pad_regulator) > + ret = regulator_disable(data->hsic_pad_regulator); > err_clk: > - imx_disable_unprepare_clks(&pdev->dev); > + imx_disable_unprepare_clks(dev); > return ret; > } > > @@ -355,6 +456,8 @@ static int ci_hdrc_imx_remove(struct platform_device *pdev) > if (data->override_phy_control) > usb_phy_shutdown(data->phy); > imx_disable_unprepare_clks(&pdev->dev); > + if (data->hsic_pad_regulator) > + regulator_disable(data->hsic_pad_regulator); > > return 0; > } > @@ -367,9 +470,19 @@ static void ci_hdrc_imx_shutdown(struct platform_device *pdev) > static int __maybe_unused imx_controller_suspend(struct device *dev) > { > struct ci_hdrc_imx_data *data = dev_get_drvdata(dev); > + int ret = 0; > > dev_dbg(dev, "at %s\n", __func__); > > + if (data->usbmisc_data) { Why do we have this check here, but not in imx_controller_resume()? > + ret = imx_usbmisc_hsic_set_clk(data->usbmisc_data, false); > + if (ret) { > + dev_err(dev, > + "usbmisc hsic_set_clk failed, ret=%d\n", ret); > + return ret; > + } > + } > + > imx_disable_unprepare_clks(dev); > data->in_lpm = true; > > @@ -400,8 +513,16 @@ static int __maybe_unused imx_controller_resume(struct device *dev) > goto clk_disable; > } > Why don't we have a check for data->usbmisc_data here, but in imx_controller_suspend() we have one? > + ret = imx_usbmisc_hsic_set_clk(data->usbmisc_data, true); > + if (ret) { > + dev_err(dev, "usbmisc hsic_set_clk failed, ret=%d\n", ret); > + goto hsic_set_clk_fail; > + } > + > return 0; > > +hsic_set_clk_fail: > + imx_usbmisc_set_wakeup(data->usbmisc_data, true); > clk_disable: > imx_disable_unprepare_clks(dev); > return ret; > diff --git a/drivers/usb/chipidea/ci_hdrc_imx.h b/drivers/usb/chipidea/ci_hdrc_imx.h > index 204275f47573..fcecab478934 100644 > --- a/drivers/usb/chipidea/ci_hdrc_imx.h > +++ b/drivers/usb/chipidea/ci_hdrc_imx.h > @@ -14,10 +14,13 @@ struct imx_usbmisc_data { > unsigned int oc_polarity:1; /* over current polarity if oc enabled */ > unsigned int evdo:1; /* set external vbus divider option */ > unsigned int ulpi:1; /* connected to an ULPI phy */ > + unsigned int hsic:1; /* HSIC controlller */ > }; > > -int imx_usbmisc_init(struct imx_usbmisc_data *); > -int imx_usbmisc_init_post(struct imx_usbmisc_data *); > -int imx_usbmisc_set_wakeup(struct imx_usbmisc_data *, bool); > +int imx_usbmisc_init(struct imx_usbmisc_data *data); > +int imx_usbmisc_init_post(struct imx_usbmisc_data *data); > +int imx_usbmisc_set_wakeup(struct imx_usbmisc_data *data, bool enabled); > +int imx_usbmisc_hsic_set_connect(struct imx_usbmisc_data *data); > +int imx_usbmisc_hsic_set_clk(struct imx_usbmisc_data *data, bool on); > > #endif /* __DRIVER_USB_CHIPIDEA_CI_HDRC_IMX_H */ > diff --git a/drivers/usb/chipidea/usbmisc_imx.c b/drivers/usb/chipidea/usbmisc_imx.c > index def80ff547e4..a66a15075200 100644 > --- a/drivers/usb/chipidea/usbmisc_imx.c > +++ b/drivers/usb/chipidea/usbmisc_imx.c > @@ -64,10 +64,22 @@ > #define MX6_BM_OVER_CUR_DIS BIT(7) > #define MX6_BM_OVER_CUR_POLARITY BIT(8) > #define MX6_BM_WAKEUP_ENABLE BIT(10) > +#define MX6_BM_UTMI_ON_CLOCK BIT(13) > #define MX6_BM_ID_WAKEUP BIT(16) > #define MX6_BM_VBUS_WAKEUP BIT(17) > #define MX6SX_BM_DPDM_WAKEUP_EN BIT(29) > #define MX6_BM_WAKEUP_INTR BIT(31) > + > +#define MX6_USB_HSIC_CTRL_OFFSET 0x10 > +/* Send resume signal without 480Mhz PHY clock */ > +#define MX6SX_BM_HSIC_AUTO_RESUME BIT(23) > +/* set before portsc.suspendM = 1 */ > +#define MX6_BM_HSIC_DEV_CONN BIT(21) > +/* HSIC enable */ > +#define MX6_BM_HSIC_EN BIT(12) > +/* Force HSIC module 480M clock on, even when in Host is in suspend mode */ > +#define MX6_BM_HSIC_CLK_ON BIT(11) > + > #define MX6_USB_OTG1_PHY_CTRL 0x18 > /* For imx6dql, it is host-only controller, for later imx6, it is otg's */ > #define MX6_USB_OTG2_PHY_CTRL 0x1c > @@ -94,6 +106,10 @@ struct usbmisc_ops { > int (*post)(struct imx_usbmisc_data *data); > /* It's called when we need to enable/disable usb wakeup */ > int (*set_wakeup)(struct imx_usbmisc_data *data, bool enabled); > + /* It's called before setting portsc.suspendM */ > + int (*hsic_set_connect)(struct imx_usbmisc_data *data); > + /* It's called during suspend/resume */ > + int (*hsic_set_clk)(struct imx_usbmisc_data *data, bool enabled); > }; > > struct imx_usbmisc { > @@ -353,6 +369,18 @@ static int usbmisc_imx6q_init(struct imx_usbmisc_data *data) > writel(reg | MX6_BM_NON_BURST_SETTING, > usbmisc->base + data->index * 4); > > + /* For HSIC controller */ > + if (data->hsic) { > + reg = readl(usbmisc->base + data->index * 4); > + writel(reg | MX6_BM_UTMI_ON_CLOCK, > + usbmisc->base + data->index * 4); > + reg = readl(usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET > + + (data->index - 2) * 4); > + reg |= MX6_BM_HSIC_EN | MX6_BM_HSIC_CLK_ON; > + writel(reg, usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET > + + (data->index - 2) * 4); > + } > + > spin_unlock_irqrestore(&usbmisc->lock, flags); > > usbmisc_imx6q_set_wakeup(data, false); > @@ -360,6 +388,70 @@ static int usbmisc_imx6q_init(struct imx_usbmisc_data *data) > return 0; > } > > +static int usbmisc_imx6_hsic_set_connect(struct imx_usbmisc_data *data) > +{ > + unsigned long flags; > + u32 val, offset; > + struct imx_usbmisc *usbmisc = dev_get_drvdata(data->dev); > + int ret = 0; > + > + spin_lock_irqsave(&usbmisc->lock, flags); > + if (data->index == 2 || data->index == 3) { > + offset = (data->index - 2) * 4; > + } else if (data->index == 0) { > + /* > + * For controllers later than imx7d (imx7d is included), "For SOCs like i.MX7D and later, ..." > + * each controller has its own non core register region. > + * And the controllers before than imx7d, the 1st controller > + * is not HSIC controller. "For SOCs before i.MX7D, the first USB controller is non-HSIC." Thanks, Frieder > + */ > + offset = 0; > + } else { > + dev_err(data->dev, "index is error for usbmisc\n"); > + offset = 0; > + ret = -EINVAL; > + } > + > + val = readl(usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET + offset); > + if (!(val & MX6_BM_HSIC_DEV_CONN)) > + writel(val | MX6_BM_HSIC_DEV_CONN, > + usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET + offset); > + spin_unlock_irqrestore(&usbmisc->lock, flags); > + > + return ret; > +} > + > +static int usbmisc_imx6_hsic_set_clk(struct imx_usbmisc_data *data, bool on) > +{ > + unsigned long flags; > + u32 val, offset; > + struct imx_usbmisc *usbmisc = dev_get_drvdata(data->dev); > + int ret = 0; > + > + spin_lock_irqsave(&usbmisc->lock, flags); > + if (data->index == 2 || data->index == 3) { > + offset = (data->index - 2) * 4; > + } else if (data->index == 0) { > + offset = 0; > + } else { > + dev_err(data->dev, "index is error for usbmisc\n"); > + offset = 0; > + ret = -EINVAL; > + } > + > + val = readl(usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET + offset); > + val |= MX6_BM_HSIC_EN | MX6_BM_HSIC_CLK_ON; > + if (on) > + val |= MX6_BM_HSIC_CLK_ON; > + else > + val &= ~MX6_BM_HSIC_CLK_ON; > + writel(val, usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET + offset); > + spin_unlock_irqrestore(&usbmisc->lock, flags); > + > + return 0; > +} > + > + > static int usbmisc_imx6sx_init(struct imx_usbmisc_data *data) > { > void __iomem *reg = NULL; > @@ -385,6 +477,13 @@ static int usbmisc_imx6sx_init(struct imx_usbmisc_data *data) > spin_unlock_irqrestore(&usbmisc->lock, flags); > } > > + /* For HSIC controller */ > + if (data->hsic) { > + val = readl(usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET); > + val |= MX6SX_BM_HSIC_AUTO_RESUME; > + writel(val, usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET); > + } > + > return 0; > } > > @@ -454,6 +553,7 @@ static int usbmisc_imx7d_init(struct imx_usbmisc_data *data) > reg &= ~MX7D_USB_VBUS_WAKEUP_SOURCE_MASK; > writel(reg | MX7D_USB_VBUS_WAKEUP_SOURCE_BVALID, > usbmisc->base + MX7D_USBNC_USB_CTRL2); > + > spin_unlock_irqrestore(&usbmisc->lock, flags); > > usbmisc_imx7d_set_wakeup(data, false); > @@ -481,6 +581,8 @@ static const struct usbmisc_ops imx53_usbmisc_ops = { > static const struct usbmisc_ops imx6q_usbmisc_ops = { > .set_wakeup = usbmisc_imx6q_set_wakeup, > .init = usbmisc_imx6q_init, > + .hsic_set_connect = usbmisc_imx6_hsic_set_connect, > + .hsic_set_clk = usbmisc_imx6_hsic_set_clk, > }; > > static const struct usbmisc_ops vf610_usbmisc_ops = { > @@ -490,6 +592,8 @@ static const struct usbmisc_ops vf610_usbmisc_ops = { > static const struct usbmisc_ops imx6sx_usbmisc_ops = { > .set_wakeup = usbmisc_imx6q_set_wakeup, > .init = usbmisc_imx6sx_init, > + .hsic_set_connect = usbmisc_imx6_hsic_set_connect, > + .hsic_set_clk = usbmisc_imx6_hsic_set_clk, > }; > > static const struct usbmisc_ops imx7d_usbmisc_ops = { > @@ -546,6 +650,33 @@ int imx_usbmisc_set_wakeup(struct imx_usbmisc_data *data, bool enabled) > } > EXPORT_SYMBOL_GPL(imx_usbmisc_set_wakeup); > > +int imx_usbmisc_hsic_set_connect(struct imx_usbmisc_data *data) > +{ > + struct imx_usbmisc *usbmisc; > + > + if (!data) > + return 0; > + > + usbmisc = dev_get_drvdata(data->dev); > + if (!usbmisc->ops->hsic_set_connect || !data->hsic) > + return 0; > + return usbmisc->ops->hsic_set_connect(data); > +} > +EXPORT_SYMBOL_GPL(imx_usbmisc_hsic_set_connect); > + > +int imx_usbmisc_hsic_set_clk(struct imx_usbmisc_data *data, bool on) > +{ > + struct imx_usbmisc *usbmisc; > + > + if (!data) > + return 0; > + > + usbmisc = dev_get_drvdata(data->dev); > + if (!usbmisc->ops->hsic_set_clk || !data->hsic) > + return 0; > + return usbmisc->ops->hsic_set_clk(data, on); > +} > +EXPORT_SYMBOL_GPL(imx_usbmisc_hsic_set_clk); > static const struct of_device_id usbmisc_imx_dt_ids[] = { > { > .compatible = "fsl,imx25-usbmisc", >
> On 16.10.18 07:01, Peter Chen wrote: > > Most of NXP (Freescale) i.mx USB part has HSIC support, in this > > series, we add support for them, it should cover all imx6 and imx7d. I > > have no HSIC interface board which is supported by upstream kernel, so > > this patches are only compiled ok, Frieder Schrempf, would you please > > help me test it on your board? Thanks. > > Thank you for providing the patch! > I applied it to v4.19-rc8 and tested and the LAN9730 comes up and works just fine. > > Here is how my devicetree looks like: > > usbphy_dummy: usbphy_dummy@1 { > compatible = "usb-nop-xceiv"; > }; > > [...] > > &usbh2 { > vbus-supply = <®_usb_h2_vbus>; > pinctrl-names = "idle", "active"; > pinctrl-0 = <&pinctrl_usbh2_idle>; > pinctrl-1 = <&pinctrl_usbh2_active>; > fsl,usbphy = <&usbphy_dummy>; > phy_type = "hsic"; > status = "okay"; > #address-cells = <1>; > #size-cells = <0>; > > usbnet: smsc@1 { > compatible = "usb424,9730"; > /* Filled in by U-Boot */ > mac-address = [00 00 00 00 00 00]; > reg = <1>; > }; > }; > > [...] > > pinctrl_usbh2_idle: usbh2grp-idle { > fsl,pins = < > MX6QDL_PAD_RGMII_TXC__USB_H2_DATA 0x40013030 > MX6QDL_PAD_RGMII_TX_CTL__USB_H2_STROBE 0x40013030 > >; > }; > > pinctrl_usbh2_active: usbh2grp-active { > fsl,pins = < > MX6QDL_PAD_RGMII_TXC__USB_H2_DATA 0x40013030 > MX6QDL_PAD_RGMII_TX_CTL__USB_H2_STROBE 0x40017030 > >; > }; > > > Are there any test cases I should try? > How can I test suspend/resume? > - System suspend/resume 1. Enable USB wakeup for i in $(find /sys -name wakeup | grep usb);do echo enabled > $i;echo "echo enabled > $i";done; 2. Let the system enter suspend using below command echo mem > /sys/power/state 3. And see if there is a wakeup block system entering suspend, and check if USB HSIC works ok after system resume - Runtime suspend 1. Enable auto suspend for all USB devices, and check if USBOH3 clock is closed, make sure do not plug any ethernet cable on the RJ45 port. /* Enable auto suspend */ for i in $(find /sys -name control | grep usb);do echo auto > $i;echo "echo auto > $i";done; /* Scripts to see USB clock */ root@imx8qxpmek:~# cat uclk.sh /home/root/dump-clocks.sh | grep usb root@imx8qxpmek:~# cat /home/root/dump-clocks.sh #!/bin/bash if ! mount|grep -sq '/sys/kernel/debug'; then mount -t debugfs none /sys/kernel/debug fi cat /sys/kernel/debug/clk/clk_summary 2. If USB OH3 clock can close successfully, it means USB HSIC can enter suspend successfully. 3. Plug ethernet cable to see if the link can be on. The purpose of above two tests is to see if USB HSIC can be suspended. Thanks. Peter > I also have some suggestions for your patch. Please see the separate email. > > Thanks, > Frieder > > > > > Peter Chen (4): > > usb: chipidea: add flag for imx hsic implementation > > usb: chipidea: imx: add HSIC support > > usb: chipidea: host: override ehci->hub_control > > doc: usb: ci-hdrc-usb2: Add pinctrl properties for HSIC pin groups > > > > .../devicetree/bindings/usb/ci-hdrc-usb2.txt | 1 + > > drivers/usb/chipidea/ci_hdrc_imx.c | 153 ++++++++++++++++++--- > > drivers/usb/chipidea/ci_hdrc_imx.h | 9 +- > > drivers/usb/chipidea/host.c | 98 +++++++++++++ > > drivers/usb/chipidea/usbmisc_imx.c | 131 ++++++++++++++++++ > > include/linux/usb/chipidea.h | 3 + > > 6 files changed, 376 insertions(+), 19 deletions(-) > >
Hi Peter, On 17.10.18 09:23, Peter Chen wrote: > >> On 16.10.18 07:01, Peter Chen wrote: >>> Most of NXP (Freescale) i.mx USB part has HSIC support, in this >>> series, we add support for them, it should cover all imx6 and imx7d. I >>> have no HSIC interface board which is supported by upstream kernel, so >>> this patches are only compiled ok, Frieder Schrempf, would you please >>> help me test it on your board? Thanks. >> >> Thank you for providing the patch! >> I applied it to v4.19-rc8 and tested and the LAN9730 comes up and works just fine. >> >> Here is how my devicetree looks like: >> >> usbphy_dummy: usbphy_dummy@1 { >> compatible = "usb-nop-xceiv"; >> }; >> >> [...] >> >> &usbh2 { >> vbus-supply = <®_usb_h2_vbus>; >> pinctrl-names = "idle", "active"; >> pinctrl-0 = <&pinctrl_usbh2_idle>; >> pinctrl-1 = <&pinctrl_usbh2_active>; >> fsl,usbphy = <&usbphy_dummy>; >> phy_type = "hsic"; >> status = "okay"; >> #address-cells = <1>; >> #size-cells = <0>; >> >> usbnet: smsc@1 { >> compatible = "usb424,9730"; >> /* Filled in by U-Boot */ >> mac-address = [00 00 00 00 00 00]; >> reg = <1>; >> }; >> }; >> >> [...] >> >> pinctrl_usbh2_idle: usbh2grp-idle { >> fsl,pins = < >> MX6QDL_PAD_RGMII_TXC__USB_H2_DATA 0x40013030 >> MX6QDL_PAD_RGMII_TX_CTL__USB_H2_STROBE 0x40013030 >> >; >> }; >> >> pinctrl_usbh2_active: usbh2grp-active { >> fsl,pins = < >> MX6QDL_PAD_RGMII_TXC__USB_H2_DATA 0x40013030 >> MX6QDL_PAD_RGMII_TX_CTL__USB_H2_STROBE 0x40017030 >> >; >> }; >> >> >> Are there any test cases I should try? >> How can I test suspend/resume? >> > > - System suspend/resume > 1. Enable USB wakeup > for i in $(find /sys -name wakeup | grep usb);do echo enabled > $i;echo "echo enabled > $i";done; > 2. Let the system enter suspend using below command > echo mem > /sys/power/state > 3. And see if there is a wakeup block system entering suspend, and check if USB HSIC works ok > after system resume System suspend/resume seems to work fine. After resume the ethernet controller works. root@imxceet-solo-s-43:~# echo mem > /sys/power/state PM: suspend entry (deep) PM: Syncing filesystems ... done. Freezing user space processes ... (elapsed 0.001 seconds) done. OOM killer disabled. Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done. smsc95xx 3-1:1.0 eth1: entering SUSPEND2 mode PM: suspend devices took 0.050 seconds Disabling non-boot CPUs ... usb 3-1: reset high-speed USB device number 2 using ci_hdrc PM: resume devices took 0.590 seconds OOM killer enabled. Restarting tasks ... done. PM: suspend exit smsc95xx 3-1:1.0 eth1: link up, 100Mbps, full-duplex, lpa 0x4DE1 fec 2188000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx > > - Runtime suspend > 1. Enable auto suspend for all USB devices, and check if USBOH3 clock is closed, > make sure do not plug any ethernet cable on the RJ45 port. > > /* Enable auto suspend */ > for i in $(find /sys -name control | grep usb);do echo auto > $i;echo "echo auto > $i";done; This doesn't work. When the port is suspended it gets into a loop of suspending/resuming endlessly. If i put two logs in ci_hdrc_imx_runtime_suspend() and ci_hdrc_imx_runtime_resume(), I get this: [...] ci_hdrc_imx_runtime_resume:603 ci_hdrc_imx_runtime_suspend:574 ci_hdrc_imx_runtime_resume:603 ci_hdrc_imx_runtime_suspend:574 ci_hdrc_imx_runtime_resume:603 ci_hdrc_imx_runtime_suspend:574 ci_hdrc_imx_runtime_resume:603 ci_hdrc_imx_runtime_suspend:574 [...] Any ideas how to debug? Thanks, Frieder > > /* Scripts to see USB clock */ > root@imx8qxpmek:~# cat uclk.sh > /home/root/dump-clocks.sh | grep usb > root@imx8qxpmek:~# cat /home/root/dump-clocks.sh > #!/bin/bash > > if ! mount|grep -sq '/sys/kernel/debug'; then > mount -t debugfs none /sys/kernel/debug > fi > > cat /sys/kernel/debug/clk/clk_summary > > 2. If USB OH3 clock can close successfully, it means USB HSIC can enter suspend successfully. > > 3. Plug ethernet cable to see if the link can be on. > > The purpose of above two tests is to see if USB HSIC can be suspended. > > Thanks. > > Peter > > >> I also have some suggestions for your patch. Please see the separate email. >> >> Thanks, >> Frieder >> >>> >>> Peter Chen (4): >>> usb: chipidea: add flag for imx hsic implementation >>> usb: chipidea: imx: add HSIC support >>> usb: chipidea: host: override ehci->hub_control >>> doc: usb: ci-hdrc-usb2: Add pinctrl properties for HSIC pin groups >>> >>> .../devicetree/bindings/usb/ci-hdrc-usb2.txt | 1 + >>> drivers/usb/chipidea/ci_hdrc_imx.c | 153 ++++++++++++++++++--- >>> drivers/usb/chipidea/ci_hdrc_imx.h | 9 +- >>> drivers/usb/chipidea/host.c | 98 +++++++++++++ >>> drivers/usb/chipidea/usbmisc_imx.c | 131 ++++++++++++++++++ >>> include/linux/usb/chipidea.h | 3 + >>> 6 files changed, 376 insertions(+), 19 deletions(-) >>>
On 17.10.18 11:56, Frieder Schrempf wrote: > Hi Peter, > > On 17.10.18 09:23, Peter Chen wrote: >>> On 16.10.18 07:01, Peter Chen wrote: >>>> Most of NXP (Freescale) i.mx USB part has HSIC support, in this >>>> series, we add support for them, it should cover all imx6 and imx7d. I >>>> have no HSIC interface board which is supported by upstream kernel, so >>>> this patches are only compiled ok, Frieder Schrempf, would you please >>>> help me test it on your board? Thanks. >>> >>> Thank you for providing the patch! >>> I applied it to v4.19-rc8 and tested and the LAN9730 comes up and >>> works just fine. >>> >>> Here is how my devicetree looks like: >>> >>> usbphy_dummy: usbphy_dummy@1 { >>> compatible = "usb-nop-xceiv"; >>> }; >>> >>> [...] >>> >>> &usbh2 { >>> vbus-supply = <®_usb_h2_vbus>; >>> pinctrl-names = "idle", "active"; >>> pinctrl-0 = <&pinctrl_usbh2_idle>; >>> pinctrl-1 = <&pinctrl_usbh2_active>; >>> fsl,usbphy = <&usbphy_dummy>; >>> phy_type = "hsic"; >>> status = "okay"; >>> #address-cells = <1>; >>> #size-cells = <0>; >>> >>> usbnet: smsc@1 { >>> compatible = "usb424,9730"; >>> /* Filled in by U-Boot */ >>> mac-address = [00 00 00 00 00 00]; >>> reg = <1>; >>> }; >>> }; >>> >>> [...] >>> >>> pinctrl_usbh2_idle: usbh2grp-idle { >>> fsl,pins = < >>> MX6QDL_PAD_RGMII_TXC__USB_H2_DATA 0x40013030 >>> MX6QDL_PAD_RGMII_TX_CTL__USB_H2_STROBE 0x40013030 >>> >; >>> }; >>> >>> pinctrl_usbh2_active: usbh2grp-active { >>> fsl,pins = < >>> MX6QDL_PAD_RGMII_TXC__USB_H2_DATA 0x40013030 >>> MX6QDL_PAD_RGMII_TX_CTL__USB_H2_STROBE 0x40017030 >>> >; >>> }; >>> >>> >>> Are there any test cases I should try? >>> How can I test suspend/resume? >>> >> >> - System suspend/resume >> 1. Enable USB wakeup >> for i in $(find /sys -name wakeup | grep usb);do echo enabled > >> $i;echo "echo enabled > $i";done; >> 2. Let the system enter suspend using below command >> echo mem > /sys/power/state >> 3. And see if there is a wakeup block system entering suspend, and >> check if USB HSIC works ok >> after system resume > > System suspend/resume seems to work fine. After resume the ethernet > controller works. > > root@imxceet-solo-s-43:~# echo mem > /sys/power/state > PM: suspend entry (deep) > PM: Syncing filesystems ... done. > Freezing user space processes ... (elapsed 0.001 seconds) done. > OOM killer disabled. > Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done. > smsc95xx 3-1:1.0 eth1: entering SUSPEND2 mode > PM: suspend devices took 0.050 seconds > Disabling non-boot CPUs ... > usb 3-1: reset high-speed USB device number 2 using ci_hdrc > PM: resume devices took 0.590 seconds > OOM killer enabled. > Restarting tasks ... done. > PM: suspend exit > smsc95xx 3-1:1.0 eth1: link up, 100Mbps, full-duplex, lpa 0x4DE1 > fec 2188000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx > >> >> - Runtime suspend >> 1. Enable auto suspend for all USB devices, and check if USBOH3 clock >> is closed, >> make sure do not plug any ethernet cable on the RJ45 port. >> >> /* Enable auto suspend */ >> for i in $(find /sys -name control | grep usb);do echo auto > $i;echo >> "echo auto > $i";done; > > This doesn't work. When the port is suspended it gets into a loop of > suspending/resuming endlessly. If i put two logs in > ci_hdrc_imx_runtime_suspend() and ci_hdrc_imx_runtime_resume(), I get this: > > [...] > ci_hdrc_imx_runtime_resume:603 > ci_hdrc_imx_runtime_suspend:574 > ci_hdrc_imx_runtime_resume:603 > ci_hdrc_imx_runtime_suspend:574 > ci_hdrc_imx_runtime_resume:603 > ci_hdrc_imx_runtime_suspend:574 > ci_hdrc_imx_runtime_resume:603 > ci_hdrc_imx_runtime_suspend:574 > [...] Ok, forget about the loop, this was caused by one of the other ports, that had issues with overcurrent detection. But still it doesn't work for the HSIC port. The HSIC device is stuck in status "suspending" (note: "suspending" and not "suspended"): ~# cat /sys/bus/usb/devices/usb1/power/runtime_status suspended ~# cat /sys/bus/usb/devices/usb2/power/runtime_status suspended ~# cat /sys/bus/usb/devices/usb3/power/runtime_status active ~# cat /sys/bus/usb/devices/usb3/3-1/power/runtime_status suspending
On 17.10.18 13:04, Frieder Schrempf wrote: > On 17.10.18 11:56, Frieder Schrempf wrote: [...] >>> - System suspend/resume >>> 1. Enable USB wakeup >>> for i in $(find /sys -name wakeup | grep usb);do echo enabled > >>> $i;echo "echo enabled > $i";done; >>> 2. Let the system enter suspend using below command >>> echo mem > /sys/power/state >>> 3. And see if there is a wakeup block system entering suspend, and >>> check if USB HSIC works ok >>> after system resume >> >> System suspend/resume seems to work fine. After resume the ethernet >> controller works. >> >> root@imxceet-solo-s-43:~# echo mem > /sys/power/state >> PM: suspend entry (deep) >> PM: Syncing filesystems ... done. >> Freezing user space processes ... (elapsed 0.001 seconds) done. >> OOM killer disabled. >> Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done. >> smsc95xx 3-1:1.0 eth1: entering SUSPEND2 mode >> PM: suspend devices took 0.050 seconds >> Disabling non-boot CPUs ... >> usb 3-1: reset high-speed USB device number 2 using ci_hdrc >> PM: resume devices took 0.590 seconds >> OOM killer enabled. >> Restarting tasks ... done. >> PM: suspend exit >> smsc95xx 3-1:1.0 eth1: link up, 100Mbps, full-duplex, lpa 0x4DE1 >> fec 2188000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx >> >>> >>> - Runtime suspend >>> 1. Enable auto suspend for all USB devices, and check if USBOH3 clock >>> is closed, >>> make sure do not plug any ethernet cable on the RJ45 port. >>> >>> /* Enable auto suspend */ >>> for i in $(find /sys -name control | grep usb);do echo auto > >>> $i;echo "echo auto > $i";done; >> >> This doesn't work. When the port is suspended it gets into a loop of >> suspending/resuming endlessly. If i put two logs in >> ci_hdrc_imx_runtime_suspend() and ci_hdrc_imx_runtime_resume(), I get >> this: >> >> [...] >> ci_hdrc_imx_runtime_resume:603 >> ci_hdrc_imx_runtime_suspend:574 >> ci_hdrc_imx_runtime_resume:603 >> ci_hdrc_imx_runtime_suspend:574 >> ci_hdrc_imx_runtime_resume:603 >> ci_hdrc_imx_runtime_suspend:574 >> ci_hdrc_imx_runtime_resume:603 >> ci_hdrc_imx_runtime_suspend:574 >> [...] > > Ok, forget about the loop, this was caused by one of the other ports, > that had issues with overcurrent detection. > > But still it doesn't work for the HSIC port. > > The HSIC device is stuck in status "suspending" (note: "suspending" and > not "suspended"): > > ~# cat /sys/bus/usb/devices/usb1/power/runtime_status > suspended > ~# cat /sys/bus/usb/devices/usb2/power/runtime_status > suspended > ~# cat /sys/bus/usb/devices/usb3/power/runtime_status > active > ~# cat /sys/bus/usb/devices/usb3/3-1/power/runtime_status > suspending It seems like this is a problem with the smsc95xx driver. I fixed this and now the device is suspending properly, when the network interface is down and resumes when it comes up. I also checked the USBOH3 clock and its properly switched on and off. What didn't work is auto suspend and resume on ethernet link down/up, but this seems to be a restriction of the smsc95xx/usbnet driver. So all in all this looks good to me. Please let me know if you need any more information or tests. Thanks, Frieder
> > +static int ci_hdrc_imx_notify_event(struct ci_hdrc *ci, unsigned int > > +event) { > > + struct device *dev = ci->dev->parent; > > + struct ci_hdrc_imx_data *data = dev_get_drvdata(dev); > > + int ret = 0; > > + > > + switch (event) { > > + case CI_HDRC_IMX_HSIC_ACTIVE_EVENT: > > + if (!IS_ERR(data->pinctrl) && > > + !IS_ERR(data->pinctrl_hsic_active)) { > > If we make the pinctrl mandatory in case of HSIC as proposed below, we don't need > the checks here. > Will delete them > > + > > + data->pinctrl_hsic_active = pinctrl_lookup_state(data->pinctrl, > > + "active"); > > + if (IS_ERR(data->pinctrl_hsic_active)) > > + dev_dbg(dev, > > + "pinctrl_hsic_active lookup failed, err=%ld\n", > > + PTR_ERR(data->pinctrl_hsic_active)); > > + } > > In the paragraph above, I think we should make the pinctrl settings mandatory in > case of HSIC and error out if one of them is missing. > > Also I think we could make the code more readable by removing the nested > conditions. > > Maybe something like this would be better? > > if (of_usb_get_phy_mode(dev->of_node) == USBPHY_INTERFACE_MODE_HSIC) > { > data->pinctrl = devm_pinctrl_get(dev); > if (IS_ERR(data->pinctrl)) { > dev_err(dev, "failed to get HSIC pinctrl, err=%ld\n", > PTR_ERR(data->pinctrl)); > return PTR_ERR(data->pinctrl); > } > > pinctrl_hsic_idle = pinctrl_lookup_state(data->pinctrl, "idle"); > if (IS_ERR(pinctrl_hsic_idle)) { > dev_err(dev, "failed to get HSIC idle pinctrl," > "err=%ld\n", PTR_ERR(pinctrl_hsic_idle)); > return PTR_ERR(pinctrl_hsic_idle); > } > > ret = pinctrl_select_state(data->pinctrl, pinctrl_hsic_idle); > if (ret) { > dev_err(dev, "failed to select HSIC idle pinctrl," > "err=%d\n", ret); > return ret; > } > > data->pinctrl_hsic_active = pinctrl_lookup_state(data->pinctrl, > "active"); > if (IS_ERR(data->pinctrl_hsic_active)) { > dev_err(dev, "failed to get HSIC active pinctrl," > "err=%ld\n", > PTR_ERR(data->pinctrl_hsic_active)); > return PTR_ERR(data->pinctrl_hsic_active); > } > } > Good idea. > > } > > @@ -367,9 +470,19 @@ static void ci_hdrc_imx_shutdown(struct > platform_device *pdev) > > static int __maybe_unused imx_controller_suspend(struct device *dev) > > { > > struct ci_hdrc_imx_data *data = dev_get_drvdata(dev); > > + int ret = 0; > > > > dev_dbg(dev, "at %s\n", __func__); > > > > + if (data->usbmisc_data) { > > Why do we have this check here, but not in imx_controller_resume()? > I will delete check here since there is NULL point check at imx_usbmisc_hsic_set_clk too. > > + ret = imx_usbmisc_hsic_set_clk(data->usbmisc_data, false); > > + if (ret) { > > + dev_err(dev, > > + "usbmisc hsic_set_clk failed, ret=%d\n", ret); > > + return ret; > > + } > > + } > > + > > imx_disable_unprepare_clks(dev); > > data->in_lpm = true; > > > > @@ -400,8 +513,16 @@ static int __maybe_unused > imx_controller_resume(struct device *dev) > > goto clk_disable; > > } > > > > Why don't we have a check for data->usbmisc_data here, but in > imx_controller_suspend() we have one? > Yes, not need. > > + ret = imx_usbmisc_hsic_set_clk(data->usbmisc_data, true); > > + if (ret) { > > + dev_err(dev, "usbmisc hsic_set_clk failed, ret=%d\n", ret); > > + goto hsic_set_clk_fail; > > + } > > + > > return 0; > > > > +hsic_set_clk_fail: > > + imx_usbmisc_set_wakeup(data->usbmisc_data, true); > > clk_disable: > > imx_disable_unprepare_clks(dev); > > return ret; > > diff --git a/drivers/usb/chipidea/ci_hdrc_imx.h > > b/drivers/usb/chipidea/ci_hdrc_imx.h > > index 204275f47573..fcecab478934 100644 > > --- a/drivers/usb/chipidea/ci_hdrc_imx.h > > +++ b/drivers/usb/chipidea/ci_hdrc_imx.h > > @@ -14,10 +14,13 @@ struct imx_usbmisc_data { > > unsigned int oc_polarity:1; /* over current polarity if oc enabled */ > > unsigned int evdo:1; /* set external vbus divider option */ > > unsigned int ulpi:1; /* connected to an ULPI phy */ > > + unsigned int hsic:1; /* HSIC controlller */ > > }; > > > > -int imx_usbmisc_init(struct imx_usbmisc_data *); -int > > imx_usbmisc_init_post(struct imx_usbmisc_data *); -int > > imx_usbmisc_set_wakeup(struct imx_usbmisc_data *, bool); > > +int imx_usbmisc_init(struct imx_usbmisc_data *data); int > > +imx_usbmisc_init_post(struct imx_usbmisc_data *data); int > > +imx_usbmisc_set_wakeup(struct imx_usbmisc_data *data, bool enabled); > > +int imx_usbmisc_hsic_set_connect(struct imx_usbmisc_data *data); int > > +imx_usbmisc_hsic_set_clk(struct imx_usbmisc_data *data, bool on); > > > > #endif /* __DRIVER_USB_CHIPIDEA_CI_HDRC_IMX_H */ diff --git > > a/drivers/usb/chipidea/usbmisc_imx.c > > b/drivers/usb/chipidea/usbmisc_imx.c > > index def80ff547e4..a66a15075200 100644 > > --- a/drivers/usb/chipidea/usbmisc_imx.c > > +++ b/drivers/usb/chipidea/usbmisc_imx.c > > @@ -64,10 +64,22 @@ > > #define MX6_BM_OVER_CUR_DIS BIT(7) > > #define MX6_BM_OVER_CUR_POLARITY BIT(8) > > #define MX6_BM_WAKEUP_ENABLE BIT(10) > > +#define MX6_BM_UTMI_ON_CLOCK BIT(13) > > #define MX6_BM_ID_WAKEUP BIT(16) > > #define MX6_BM_VBUS_WAKEUP BIT(17) > > #define MX6SX_BM_DPDM_WAKEUP_EN BIT(29) > > #define MX6_BM_WAKEUP_INTR BIT(31) > > + > > +#define MX6_USB_HSIC_CTRL_OFFSET 0x10 > > +/* Send resume signal without 480Mhz PHY clock */ > > +#define MX6SX_BM_HSIC_AUTO_RESUME BIT(23) > > +/* set before portsc.suspendM = 1 */ > > +#define MX6_BM_HSIC_DEV_CONN BIT(21) > > +/* HSIC enable */ > > +#define MX6_BM_HSIC_EN BIT(12) > > +/* Force HSIC module 480M clock on, even when in Host is in suspend mode */ > > +#define MX6_BM_HSIC_CLK_ON BIT(11) > > + > > #define MX6_USB_OTG1_PHY_CTRL 0x18 > > /* For imx6dql, it is host-only controller, for later imx6, it is otg's */ > > #define MX6_USB_OTG2_PHY_CTRL 0x1c > > @@ -94,6 +106,10 @@ struct usbmisc_ops { > > int (*post)(struct imx_usbmisc_data *data); > > /* It's called when we need to enable/disable usb wakeup */ > > int (*set_wakeup)(struct imx_usbmisc_data *data, bool enabled); > > + /* It's called before setting portsc.suspendM */ > > + int (*hsic_set_connect)(struct imx_usbmisc_data *data); > > + /* It's called during suspend/resume */ > > + int (*hsic_set_clk)(struct imx_usbmisc_data *data, bool enabled); > > }; > > > > struct imx_usbmisc { > > @@ -353,6 +369,18 @@ static int usbmisc_imx6q_init(struct imx_usbmisc_data > *data) > > writel(reg | MX6_BM_NON_BURST_SETTING, > > usbmisc->base + data->index * 4); > > > > + /* For HSIC controller */ > > + if (data->hsic) { > > + reg = readl(usbmisc->base + data->index * 4); > > + writel(reg | MX6_BM_UTMI_ON_CLOCK, > > + usbmisc->base + data->index * 4); > > + reg = readl(usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET > > + + (data->index - 2) * 4); > > + reg |= MX6_BM_HSIC_EN | MX6_BM_HSIC_CLK_ON; > > + writel(reg, usbmisc->base + MX6_USB_HSIC_CTRL_OFFSET > > + + (data->index - 2) * 4); > > + } > > + > > spin_unlock_irqrestore(&usbmisc->lock, flags); > > > > usbmisc_imx6q_set_wakeup(data, false); @@ -360,6 +388,70 @@ static > > int usbmisc_imx6q_init(struct imx_usbmisc_data *data) > > return 0; > > } > > > > +static int usbmisc_imx6_hsic_set_connect(struct imx_usbmisc_data > > +*data) { > > + unsigned long flags; > > + u32 val, offset; > > + struct imx_usbmisc *usbmisc = dev_get_drvdata(data->dev); > > + int ret = 0; > > + > > + spin_lock_irqsave(&usbmisc->lock, flags); > > + if (data->index == 2 || data->index == 3) { > > + offset = (data->index - 2) * 4; > > + } else if (data->index == 0) { > > + /* > > + * For controllers later than imx7d (imx7d is included), > > "For SOCs like i.MX7D and later, ..." > > > + * each controller has its own non core register region. > > + * And the controllers before than imx7d, the 1st controller > > + * is not HSIC controller. > > "For SOCs before i.MX7D, the first USB controller is non-HSIC." > I will change both. Thanks. Peter
> >>> - System suspend/resume > >>> 1. Enable USB wakeup > >>> for i in $(find /sys -name wakeup | grep usb);do echo enabled > > >>> $i;echo "echo enabled > $i";done; 2. Let the system enter suspend > >>> using below command echo mem > /sys/power/state 3. And see if there > >>> is a wakeup block system entering suspend, and check if USB HSIC > >>> works ok after system resume > >> > >> System suspend/resume seems to work fine. After resume the ethernet > >> controller works. > >> > >> root@imxceet-solo-s-43:~# echo mem > /sys/power/state > >> PM: suspend entry (deep) > >> PM: Syncing filesystems ... done. > >> Freezing user space processes ... (elapsed 0.001 seconds) done. > >> OOM killer disabled. > >> Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done. > >> smsc95xx 3-1:1.0 eth1: entering SUSPEND2 mode > >> PM: suspend devices took 0.050 seconds Disabling non-boot CPUs ... > >> usb 3-1: reset high-speed USB device number 2 using ci_hdrc > >> PM: resume devices took 0.590 seconds OOM killer enabled. > >> Restarting tasks ... done. > >> PM: suspend exit > >> smsc95xx 3-1:1.0 eth1: link up, 100Mbps, full-duplex, lpa 0x4DE1 fec > >> 2188000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx > >> > >>> > >>> - Runtime suspend > >>> 1. Enable auto suspend for all USB devices, and check if USBOH3 > >>> clock is closed, make sure do not plug any ethernet cable on the > >>> RJ45 port. > >>> > >>> /* Enable auto suspend */ > >>> for i in $(find /sys -name control | grep usb);do echo auto > > >>> $i;echo "echo auto > $i";done; > >> > >> This doesn't work. When the port is suspended it gets into a loop of > >> suspending/resuming endlessly. If i put two logs in > >> ci_hdrc_imx_runtime_suspend() and ci_hdrc_imx_runtime_resume(), I get > >> this: > >> > >> [...] > >> ci_hdrc_imx_runtime_resume:603 > >> ci_hdrc_imx_runtime_suspend:574 > >> ci_hdrc_imx_runtime_resume:603 > >> ci_hdrc_imx_runtime_suspend:574 > >> ci_hdrc_imx_runtime_resume:603 > >> ci_hdrc_imx_runtime_suspend:574 > >> ci_hdrc_imx_runtime_resume:603 > >> ci_hdrc_imx_runtime_suspend:574 > >> [...] > > > > Ok, forget about the loop, this was caused by one of the other ports, > > that had issues with overcurrent detection. > > > > But still it doesn't work for the HSIC port. > > > > The HSIC device is stuck in status "suspending" (note: "suspending" > > and not "suspended"): > > > > ~# cat /sys/bus/usb/devices/usb1/power/runtime_status > > suspended > > ~# cat /sys/bus/usb/devices/usb2/power/runtime_status > > suspended > > ~# cat /sys/bus/usb/devices/usb3/power/runtime_status > > active > > ~# cat /sys/bus/usb/devices/usb3/3-1/power/runtime_status > > suspending > > It seems like this is a problem with the smsc95xx driver. I fixed this and now the > device is suspending properly, when the network interface is down and resumes > when it comes up. > > I also checked the USBOH3 clock and its properly switched on and off. > > What didn't work is auto suspend and resume on ethernet link down/up, but this > seems to be a restriction of the smsc95xx/usbnet driver. > > So all in all this looks good to me. Please let me know if you need any more > information or tests. > Thanks, Frieder, no more tests are needed. You could send me your dts changes as patches, I will append it at my v2 patch series. Peter
On 18.10.18 03:22, Peter Chen wrote: > >>>>> - System suspend/resume >>>>> 1. Enable USB wakeup >>>>> for i in $(find /sys -name wakeup | grep usb);do echo enabled > >>>>> $i;echo "echo enabled > $i";done; 2. Let the system enter suspend >>>>> using below command echo mem > /sys/power/state 3. And see if there >>>>> is a wakeup block system entering suspend, and check if USB HSIC >>>>> works ok after system resume >>>> >>>> System suspend/resume seems to work fine. After resume the ethernet >>>> controller works. >>>> >>>> root@imxceet-solo-s-43:~# echo mem > /sys/power/state >>>> PM: suspend entry (deep) >>>> PM: Syncing filesystems ... done. >>>> Freezing user space processes ... (elapsed 0.001 seconds) done. >>>> OOM killer disabled. >>>> Freezing remaining freezable tasks ... (elapsed 0.001 seconds) done. >>>> smsc95xx 3-1:1.0 eth1: entering SUSPEND2 mode >>>> PM: suspend devices took 0.050 seconds Disabling non-boot CPUs ... >>>> usb 3-1: reset high-speed USB device number 2 using ci_hdrc >>>> PM: resume devices took 0.590 seconds OOM killer enabled. >>>> Restarting tasks ... done. >>>> PM: suspend exit >>>> smsc95xx 3-1:1.0 eth1: link up, 100Mbps, full-duplex, lpa 0x4DE1 fec >>>> 2188000.ethernet eth0: Link is Up - 100Mbps/Full - flow control rx/tx >>>> >>>>> >>>>> - Runtime suspend >>>>> 1. Enable auto suspend for all USB devices, and check if USBOH3 >>>>> clock is closed, make sure do not plug any ethernet cable on the >>>>> RJ45 port. >>>>> >>>>> /* Enable auto suspend */ >>>>> for i in $(find /sys -name control | grep usb);do echo auto > >>>>> $i;echo "echo auto > $i";done; >>>> >>>> This doesn't work. When the port is suspended it gets into a loop of >>>> suspending/resuming endlessly. If i put two logs in >>>> ci_hdrc_imx_runtime_suspend() and ci_hdrc_imx_runtime_resume(), I get >>>> this: >>>> >>>> [...] >>>> ci_hdrc_imx_runtime_resume:603 >>>> ci_hdrc_imx_runtime_suspend:574 >>>> ci_hdrc_imx_runtime_resume:603 >>>> ci_hdrc_imx_runtime_suspend:574 >>>> ci_hdrc_imx_runtime_resume:603 >>>> ci_hdrc_imx_runtime_suspend:574 >>>> ci_hdrc_imx_runtime_resume:603 >>>> ci_hdrc_imx_runtime_suspend:574 >>>> [...] >>> >>> Ok, forget about the loop, this was caused by one of the other ports, >>> that had issues with overcurrent detection. >>> >>> But still it doesn't work for the HSIC port. >>> >>> The HSIC device is stuck in status "suspending" (note: "suspending" >>> and not "suspended"): >>> >>> ~# cat /sys/bus/usb/devices/usb1/power/runtime_status >>> suspended >>> ~# cat /sys/bus/usb/devices/usb2/power/runtime_status >>> suspended >>> ~# cat /sys/bus/usb/devices/usb3/power/runtime_status >>> active >>> ~# cat /sys/bus/usb/devices/usb3/3-1/power/runtime_status >>> suspending >> >> It seems like this is a problem with the smsc95xx driver. I fixed this and now the >> device is suspending properly, when the network interface is down and resumes >> when it comes up. >> >> I also checked the USBOH3 clock and its properly switched on and off. >> >> What didn't work is auto suspend and resume on ethernet link down/up, but this >> seems to be a restriction of the smsc95xx/usbnet driver. >> >> So all in all this looks good to me. Please let me know if you need any more >> information or tests. >> > > Thanks, Frieder, no more tests are needed. > You could send me your dts changes as patches, I will append it at my v2 patch series. My board is currently off-tree so I can't send any patch for the pinmux settings in devicetree. But I will send a patch with the changes that should go to imx6qdl.dtsi, imx6sl.dtsi and imx6sx.dtsi. Though I only tested on i.MX6S. Thanks, Frieder
> > > > Thanks, Frieder, no more tests are needed. > > You could send me your dts changes as patches, I will append it at my v2 patch > series. > > My board is currently off-tree so I can't send any patch for the pinmux settings in > devicetree. But I will send a patch with the changes that should go to imx6qdl.dtsi, > imx6sl.dtsi and imx6sx.dtsi. Though I only tested on i.MX6S. > No, the changes are board level specific, you could not add it into SoC file. Peter
On 18.10.18 10:48, Peter Chen wrote: > >>> >>> Thanks, Frieder, no more tests are needed. >>> You could send me your dts changes as patches, I will append it at my v2 patch >> series. >> >> My board is currently off-tree so I can't send any patch for the pinmux settings in >> devicetree. But I will send a patch with the changes that should go to imx6qdl.dtsi, >> imx6sl.dtsi and imx6sx.dtsi. Though I only tested on i.MX6S. >> > > No, the changes are board level specific, you could not add it into SoC file. Not all changes are board specific. The HSIC-only host controllers can only be used to interface with on-board chips via HSIC. So they need a "usb-nop-xceiv" dummy PHY in all use cases. So if I understand this correctly this should be done in the SoCs dtsi files like in this patch I just sent: [1]. Please also note that for i.MX7S this is already added in imx7s.dtsi: [2] Thanks, Frieder [1]: https://lkml.org/lkml/2018/10/18/414 [2]: https://github.com/torvalds/linux/blob/master/arch/arm/boot/dts/imx7s.dtsi#L88
> >> My board is currently off-tree so I can't send any patch for the > >> pinmux settings in devicetree. But I will send a patch with the > >> changes that should go to imx6qdl.dtsi, imx6sl.dtsi and imx6sx.dtsi. Though I > only tested on i.MX6S. > >> > > > > No, the changes are board level specific, you could not add it into SoC file. > > Not all changes are board specific. The HSIC-only host controllers can only be used > to interface with on-board chips via HSIC. So they need a "usb-nop-xceiv" dummy > PHY in all use cases. > > So if I understand this correctly this should be done in the SoCs dtsi files like in this > patch I just sent: [1]. > > Please also note that for i.MX7S this is already added in imx7s.dtsi: [2] > Oh, I have misunderstood your meaning, the [1] is needed since it is for SoC. I thought your board level changes (like pinctrl) would go into SoC dtsi file. If there are no more comments, I will send v2 next week, feel free to add your reviewed-by and tested-by tag, thanks. Peter > > [1]: > https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flkml.org%2 > Flkml%2F2018%2F10%2F18%2F414&data=02%7C01%7Cpeter.chen%40nxp. > com%7C3c73fdf4ec98477da49308d634d78b39%7C686ea1d3bc2b4c6fa92cd99c5c > 301635%7C0%7C0%7C636754497687793991&sdata=aHhMhaj%2FbdTFu0B > qXi0loiBh%2BHzl%2F%2BTSw3txyjBSNLQ%3D&reserved=0 > [2]: > https://emea01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com > %2Ftorvalds%2Flinux%2Fblob%2Fmaster%2Farch%2Farm%2Fboot%2Fdts%2Fim > x7s.dtsi%23L88&data=02%7C01%7Cpeter.chen%40nxp.com%7C3c73fdf4ec9 > 8477da49308d634d78b39%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0% > 7C636754497687793991&sdata=7cVhAMQpWWYjGHjY8Xh2g0CG3jPaeG8X > 9K368AooQcc%3D&reserved=0