[PATCH v4 0/4] usb: dwc3-am62: module removal and errata fixes

[PATCH v4 0/4] usb: dwc3-am62: module removal and errata fixes

[Roger Quadros]

Hi,

This series fixes errors during module removal. It also
implements PHY core voltage selection as per TI recommendation
and workaround for Errata i2409 [1].

The workaround needs PHY2 region to be present in device node.
The device tree patch will be sent later after the DT binding doc
is merged.

[1] - https://www.ti.com/lit/er/sprz487d/sprz487d.pdf

Signed-off-by: Roger Quadros <rogerq@kernel.org>
---
Changes in v4:
- re-arranged patches into first 2 bug-fixes and added Cc stable for them
- Added Acked-by
- Link to v3: https://lore.kernel.org/r/20240214-for-v6-9-am62-usb-errata-3-0-v3-0-147ec5eae18c@kernel.org

---
Roger Quadros (4):
      usb: dwc3-am62: fix module unload/reload behavior
      usb: dwc3-am62: Disable wakeup at remove
      usb: dwc3-am62: Fix PHY core voltage selection
      usb: dwc3-am62: add workaround for Errata i2409

 drivers/usb/dwc3/dwc3-am62.c | 42 ++++++++++++++++++++++++++++++------------
 1 file changed, 30 insertions(+), 12 deletions(-)
---
base-commit: 6613476e225e090cc9aad49be7fa504e290dd33d
change-id: 20240206-for-v6-9-am62-usb-errata-3-0-233024ea8e9d

Best regards,
-- 
Roger Quadros <rogerq@kernel.org>

[PATCH v4 1/4] usb: dwc3-am62: fix module unload/reload behavior

[Roger Quadros]

As runtime PM is enabled, the module can be runtime
suspended when .remove() is called.

Do a pm_runtime_get_sync() to make sure module is active
before doing any register operations.

Doing a pm_runtime_put_sync() should disable the refclk
so no need to disable it again.

Fixes the below warning at module removel.

[   39.705310] ------------[ cut here ]------------
[   39.710004] clk:162:3 already disabled
[   39.713941] WARNING: CPU: 0 PID: 921 at drivers/clk/clk.c:1090 clk_core_disable+0xb0/0xb8

We called of_platform_populate() in .probe() so call the
cleanup function of_platform_depopulate() in .remove().
Get rid of the now unnnecessary dwc3_ti_remove_core().
Without this, module re-load doesn't work properly.

Fixes: e8784c0aec03 ("drivers: usb: dwc3: Add AM62 USB wrapper driver")
Cc:  <stable@vger.kernel.org> # v5.19+
Signed-off-by: Roger Quadros <rogerq@kernel.org>
---
Changelog:
v4:
- added Fixes and Cc stable.
- added of_platform_depopulate() changes here.
- drop wakeup disable code.
v3: no change
v2: no change
---
 drivers/usb/dwc3/dwc3-am62.c | 12 ++----------
 1 file changed, 2 insertions(+), 10 deletions(-)

diff --git a/drivers/usb/dwc3/dwc3-am62.c b/drivers/usb/dwc3/dwc3-am62.c
index 90a587bc29b7..f85603b7f7c5 100644
--- a/drivers/usb/dwc3/dwc3-am62.c
+++ b/drivers/usb/dwc3/dwc3-am62.c
@@ -267,21 +267,14 @@ static int dwc3_ti_probe(struct platform_device *pdev)
 	return ret;
 }
 
-static int dwc3_ti_remove_core(struct device *dev, void *c)
-{
-	struct platform_device *pdev = to_platform_device(dev);
-
-	platform_device_unregister(pdev);
-	return 0;
-}
-
 static void dwc3_ti_remove(struct platform_device *pdev)
 {
 	struct device *dev = &pdev->dev;
 	struct dwc3_am62 *am62 = platform_get_drvdata(pdev);
 	u32 reg;
 
-	device_for_each_child(dev, NULL, dwc3_ti_remove_core);
+	pm_runtime_get_sync(dev);
+	of_platform_depopulate(dev);
 
 	/* Clear mode valid bit */
 	reg = dwc3_ti_readl(am62, USBSS_MODE_CONTROL);
@@ -289,7 +282,6 @@ static void dwc3_ti_remove(struct platform_device *pdev)
 	dwc3_ti_writel(am62, USBSS_MODE_CONTROL, reg);
 
 	pm_runtime_put_sync(dev);
-	clk_disable_unprepare(am62->usb2_refclk);
 	pm_runtime_disable(dev);
 	pm_runtime_set_suspended(dev);
 }

-- 
2.34.1

[PATCH v4 2/4] usb: dwc3-am62: Disable wakeup at remove

[Roger Quadros]

Disable wakeup at remove.
Fixes the below warnings on module unload and reload.

> dwc3-am62 f900000.dwc3-usb: couldn't enable device as a wakeup source: -17
> dwc3-am62 f910000.dwc3-usb: couldn't enable device as a wakeup source: -17

Fixes: 4e3972b589da ("usb: dwc3-am62: Enable as a wakeup source by default")
Cc:  <stable@vger.kernel.org> # v6.4+
Signed-off-by: Roger Quadros <rogerq@kernel.org>
---
Changelog:
v4 - new patch. split out from v3 patch 2.
---
 drivers/usb/dwc3/dwc3-am62.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/usb/dwc3/dwc3-am62.c b/drivers/usb/dwc3/dwc3-am62.c
index f85603b7f7c5..ea6e29091c0c 100644
--- a/drivers/usb/dwc3/dwc3-am62.c
+++ b/drivers/usb/dwc3/dwc3-am62.c
@@ -274,6 +274,7 @@ static void dwc3_ti_remove(struct platform_device *pdev)
 	u32 reg;
 
 	pm_runtime_get_sync(dev);
+	device_init_wakeup(dev, false);
 	of_platform_depopulate(dev);
 
 	/* Clear mode valid bit */

-- 
2.34.1

[PATCH v4 3/4] usb: dwc3-am62: Fix PHY core voltage selection

[Roger Quadros]

TRM information is outdated and design team has confirmed
that PHY_CORE_VOLTAGE should be 0 irrespective of
VDD_CORE voltage.

Signed-off-by: Roger Quadros <rogerq@kernel.org>
Acked-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
---
Changelog:
v4: Added Acked-by
v3: no change
v2: no change
v1: https://lore.kernel.org/all/20240201121220.5523-4-rogerq@kernel.org/
---
 drivers/usb/dwc3/dwc3-am62.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/drivers/usb/dwc3/dwc3-am62.c b/drivers/usb/dwc3/dwc3-am62.c
index ea6e29091c0c..09b1f5a1c4fb 100644
--- a/drivers/usb/dwc3/dwc3-am62.c
+++ b/drivers/usb/dwc3/dwc3-am62.c
@@ -97,7 +97,8 @@
 #define USBSS_VBUS_STAT_SESSVALID	BIT(2)
 #define USBSS_VBUS_STAT_VBUSVALID	BIT(0)
 
-/* Mask for PHY PLL REFCLK */
+/* USB_PHY_CTRL register bits in CTRL_MMR */
+#define PHY_CORE_VOLTAGE_MASK	BIT(31)
 #define PHY_PLL_REFCLK_MASK	GENMASK(3, 0)
 
 #define DWC3_AM62_AUTOSUSPEND_DELAY	100
@@ -162,6 +163,13 @@ static int phy_syscon_pll_refclk(struct dwc3_am62 *am62)
 
 	am62->offset = args.args[0];
 
+	/* Core voltage. PHY_CORE_VOLTAGE bit Recommended to be 0 always */
+	ret = regmap_update_bits(am62->syscon, am62->offset, PHY_CORE_VOLTAGE_MASK, 0);
+	if (ret) {
+		dev_err(dev, "failed to set phy core voltage\n");
+		return ret;
+	}
+
 	ret = regmap_update_bits(am62->syscon, am62->offset, PHY_PLL_REFCLK_MASK, am62->rate_code);
 	if (ret) {
 		dev_err(dev, "failed to set phy pll reference clock rate\n");

-- 
2.34.1

[PATCH v4 4/4] usb: dwc3-am62: add workaround for Errata i2409

[Roger Quadros]

All AM62 devices have Errata i2409 [1] due to which
USB2 PHY may lock up due to short suspend.

Workaround involves setting bit 5 and 4 PLL_REG12
in PHY2 register space after USB controller is brought
out of LPSC reset but before controller initialization.

Handle this workaround.

[1] - https://www.ti.com/lit/er/sprz487d/sprz487d.pdf

Signed-off-by: Roger Quadros <rogerq@kernel.org>
Acked-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
---
Changelog:
v4: Added Acked-by
v3: move requesting of PHY2 region to right where we need it
v2: don't add phy read/write helpers or add phy to private data
v1: https://lore.kernel.org/all/20240201121220.5523-5-rogerq@kernel.org/
---
 drivers/usb/dwc3/dwc3-am62.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)

diff --git a/drivers/usb/dwc3/dwc3-am62.c b/drivers/usb/dwc3/dwc3-am62.c
index 09b1f5a1c4fb..fad151e78fd6 100644
--- a/drivers/usb/dwc3/dwc3-am62.c
+++ b/drivers/usb/dwc3/dwc3-am62.c
@@ -101,6 +101,11 @@
 #define PHY_CORE_VOLTAGE_MASK	BIT(31)
 #define PHY_PLL_REFCLK_MASK	GENMASK(3, 0)
 
+/* USB PHY2 register offsets */
+#define	USB_PHY_PLL_REG12		0x130
+#define	USB_PHY_PLL_LDO_REF_EN		BIT(5)
+#define	USB_PHY_PLL_LDO_REF_EN_EN	BIT(4)
+
 #define DWC3_AM62_AUTOSUSPEND_DELAY	100
 
 struct dwc3_am62 {
@@ -184,8 +189,9 @@ static int dwc3_ti_probe(struct platform_device *pdev)
 	struct device *dev = &pdev->dev;
 	struct device_node *node = pdev->dev.of_node;
 	struct dwc3_am62 *am62;
-	int i, ret;
 	unsigned long rate;
+	void __iomem *phy;
+	int i, ret;
 	u32 reg;
 
 	am62 = devm_kzalloc(dev, sizeof(*am62), GFP_KERNEL);
@@ -227,6 +233,17 @@ static int dwc3_ti_probe(struct platform_device *pdev)
 	if (ret)
 		return ret;
 
+	/* Workaround Errata i2409 */
+	phy = devm_platform_ioremap_resource(pdev, 1);
+	if (IS_ERR(phy)) {
+		dev_err(dev, "can't map PHY IOMEM resource. Won't apply i2409 fix.\n");
+		phy = NULL;
+	} else {
+		reg = readl(phy + USB_PHY_PLL_REG12);
+		reg |= USB_PHY_PLL_LDO_REF_EN | USB_PHY_PLL_LDO_REF_EN_EN;
+		writel(reg, phy + USB_PHY_PLL_REG12);
+	}
+
 	/* VBUS divider select */
 	am62->vbus_divider = device_property_read_bool(dev, "ti,vbus-divider");
 	reg = dwc3_ti_readl(am62, USBSS_PHY_CONFIG);

-- 
2.34.1