[PATCH v7 0/2] Update designware pwm driver

[PATCH v7 0/2] Update designware pwm driver

[dongxuyang]

From: Xuyang Dong <dongxuyang@eswincomputing.com>

There is already a patch [1] for the DesignWare PWM driver,
which is posted by Ben and still under review.
Based on this patch, this series is a continuation of [1]
to add support for IP versions 2.11a and later, which
includes support for "Pulse Width Modulation with 0%
and 100% Duty Cycle".

Supported chips:
ESWIN EIC7700 series SoC.

Test:
Tested this patch on the Sifive HiFive Premier P550 (which uses the EIC7700
SoC).

[1] https://lore.kernel.org/lkml/20230907161242.67190-1-ben.dooks@codethink.co.uk/

Updates:
  Changes in v7:
  - YAML:
    - Dropped Conor's Acked-by due to significant schema changes.
    - Rename patch 1 from "dt-bindings: pwm: dwc: add optional reset" to
      "dt-bindings: pwm: dwc: Add eswin compatible and resets property".
    - Update the commit message to explain why the EIC7700 supports only
      one reset.
    - Add contstraints 'minItems: 1' and 'maxItems: 1' for the resets
      property of eswin,eic7700-pwm.
    - Add an example for eswin,eic7700-pwm.

  - Link to v6: https://lore.kernel.org/all/20260424094529.1691-1-dongxuyang@eswincomputing.com/

  Changes in v6:
  - YAML:
    - Drop properties resets and its items description for eswin,eic7700-pwm.

  - Link to v5: https://lore.kernel.org/all/20260423083644.1168-1-dongxuyang@eswincomputing.com/

  Changes in v5:
  - YAML:
    - Add 'eswin,eic7700-pwm' compatible string.
    - Add the items description for the resets property and set minItems to 1.
    - Require resets property with exactly 1 reset for eswin,eic7700-pwm compatible.
  - Driver:
    - Add support for 'eswin,eic7700-pwm' compatible.
    - Add structure dwc_pwm_plat_data to manage the API for obtaining resets.

  - Link to v4: https://lore.kernel.org/all/20260415094908.1539-1-dongxuyang@eswincomputing.com/

  Changes in v4:
  - YAML:
    - Change maxItems from 1 to 2. As there is a corresponding reset signal
      for each clock domain, the effective maxItems of the resets property
      is set to 2.
    - Update the YAML commit message to describe the hardware.
  - Driver:
    - Replace devm_reset_control_get_optional_exclusive() with
      devm_reset_control_array_get_optional_exclusive(). Since the number
      of reset signals has increased from one to two, we need to use the
      array API to acquire them.

  - Link to v3: https://lore.kernel.org/all/20260402091718.1608-1-dongxuyang@eswincomputing.com/

  Changes in v3:
  - YAML:
    - Added a clear justification for the optional resets property. It is
      required to support proper controller initialization when no PWM
      channel is active at boot time, while allowing the driver to skip
      reset deassertion if any channel is already enabled.
  - Driver:
    - Update the boundary value check of tmp in __dwc_pwm_configure_timer()
      for DWC_TIM_CTRL_0N100PWM_EN.
    - Replace 'sizeof(struct dwc_pwm_drvdata)' with
      'struct_size(data, chips, 1)'.
    - Drop devm_clk_get_enabled() in favor of devm_clk_get() with explicit
      clk_prepare_enable() and clk_disable_unprepare() allowing runtime PM
      to manage clock state.
    - Replace devm_reset_control_get_optional_exclusive_deasserted() with
      devm_reset_control_get_optional_exclusive() and issue a full reset via
      reset_control_reset() only when no PWM channel is active at probe time.
    - Detect bootloader-enabled PWM channels by reading the enable bit, and
      initialize runtime PM as active for those channels by calling
      pm_runtime_set_active() and pm_runtime_get_noresume().
    - Remove autosuspend as it is not required for this driver.
    - Use explicit pm_runtime_enable() and pm_runtime_disable() instead of
      the managed devm_pm_runtime_enable() variant to ensure correct cleanup.
    - On device removal, recheck the channel enable status. If any channel
      remains active, call pm_runtime_put_noidle() before disabling clocks
      via clk_disable_unprepare().
      Resume device before register access during removal if it is runtime
      suspended, and re-suspend it afterward.
    - If device is suspended, resume it before register access during system
      resume/suspend.
    - Use pm_ptr() instead of pm_sleep_ptr() for correct PM operation.

  - Link to v2: https://lore.kernel.org/all/20260306093000.2065-1-dongxuyang@eswincomputing.com/

  Changes in v2:
  - YAML:
    - Remove eswin,eic7700-pwm.yaml. Use snps,dw-apb-timers-pwm2.yaml.
      The description in snps,dw-apb-timers-pwm2.yaml is better.
    - Add the resets property as optional, as defined in the databook.
    - Remove snps,pwm-full-range-enable as no additional property is needed.
  - Driver:
    - Change the file from pwm-dwc-eic7700.c to pwm-dwc-of.c from [1].
    - Define DWC_TIM_VERSION_ID_2_11A 2.11a as the baseline version.
    - Enable the 0% and 100% duty cycle mode by setting dwc->feature if
      the version read from the TIMERS_COMP_VERSION register is later
      than or equal to DWC_TIM_VERSION_ID_2_11A.
    - Use the DIV_ROUND_UP_ULL() to calculate width in the .apply and
      .get_state.
    - Additionally, Power Management (PM) support has been added to the
      pwm-dwc-of.c driver.
    - Drop the headers that are not used.
    - Use devm_clk_get_enabled() instead of devm_clk_get().
    - Drop of_match_ptr.
    - Fix build error with 1ULL << 32.
      Reported-by: kernel test robot <lkp@intel.com>
      Closes: https://lore.kernel.org/oe-kbuild-all/202512061720.j31AsgM7-lkp@intel.com/

  - Link to v1: https://lore.kernel.org/all/20251205090411.1388-1-dongxuyang@eswincomputing.com/
  - Link to v9: https://lore.kernel.org/lkml/20230907161242.67190-1-ben.dooks@codethink.co.uk/

Xuyang Dong (2):
  dt-bindings: pwm: dwc: Add eswin compatible and resets property
  pwm: dwc: add of/platform support

 .../bindings/pwm/snps,dw-apb-timers-pwm2.yaml |  38 +-
 drivers/pwm/Kconfig                           |  10 +
 drivers/pwm/Makefile                          |   1 +
 drivers/pwm/pwm-dwc-core.c                    | 101 +++--
 drivers/pwm/pwm-dwc-of.c                      | 346 ++++++++++++++++++
 drivers/pwm/pwm-dwc.h                         |  25 +-
 6 files changed, 488 insertions(+), 33 deletions(-)
 create mode 100644 drivers/pwm/pwm-dwc-of.c

--
2.34.1

[PATCH v7 1/2] dt-bindings: pwm: dwc: Add eswin compatible and resets property

[dongxuyang]

From: Xuyang Dong <dongxuyang@eswincomputing.com>

EIC7700 use DesignWare IP for PWM controllers. Add ESWIN EIC7700 support
in snps,dw-apb-timers-pwm2.yaml.

The DesignWare PWM includes separate reset signals dedicated to each clock
domain:
The presetn signal resets logic in pclk domain.
The timer_N_resetn signal resets logic in the timer_N_clk domain.
The resets are active-low.

The generic snps,dw-apb-timers-pwm2 binding allows one or two optional
reset lines depending on SoC integration.

On EIC7700, the presetn and timer_N_resetn inputs are physically tied
to a single reset line, therefore exactly one reset is required.

Signed-off-by: Xuyang Dong <dongxuyang@eswincomputing.com>
---
 .../bindings/pwm/snps,dw-apb-timers-pwm2.yaml | 38 +++++++++++++++++--
 1 file changed, 34 insertions(+), 4 deletions(-)

diff --git a/Documentation/devicetree/bindings/pwm/snps,dw-apb-timers-pwm2.yaml b/Documentation/devicetree/bindings/pwm/snps,dw-apb-timers-pwm2.yaml
index 7523a89a1773..a4b7929f2e05 100644
--- a/Documentation/devicetree/bindings/pwm/snps,dw-apb-timers-pwm2.yaml
+++ b/Documentation/devicetree/bindings/pwm/snps,dw-apb-timers-pwm2.yaml
@@ -20,12 +20,11 @@ description:
   instead of having to encode the IP version number in the device tree
   compatible.
 
-allOf:
-  - $ref: pwm.yaml#
-
 properties:
   compatible:
-    const: snps,dw-apb-timers-pwm2
+    enum:
+      - snps,dw-apb-timers-pwm2
+      - eswin,eic7700-pwm
 
   reg:
     maxItems: 1
@@ -43,6 +42,12 @@ properties:
       - const: bus
       - const: timer
 
+  resets:
+    minItems: 1
+    items:
+      - description: Interface bus reset
+      - description: PWM timer logic reset
+
   snps,pwm-number:
     $ref: /schemas/types.yaml#/definitions/uint32
     description: The number of PWM channels configured for this instance
@@ -54,6 +59,22 @@ required:
   - clocks
   - clock-names
 
+allOf:
+  - $ref: pwm.yaml#
+
+  - if:
+      properties:
+        compatible:
+          contains:
+            const: eswin,eic7700-pwm
+    then:
+      properties:
+        resets:
+          minItems: 1
+          maxItems: 1
+      required:
+        - resets
+
 additionalProperties: false
 
 examples:
@@ -65,3 +86,12 @@ examples:
       clocks = <&bus>, <&timer>;
       clock-names = "bus", "timer";
     };
+  - |
+    pwm@50818000 {
+      compatible = "eswin,eic7700-pwm";
+      reg = <0x50818000 0x4000>;
+      #pwm-cells = <3>;
+      clocks = <&bus>, <&timer>;
+      clock-names = "bus", "timer";
+      resets = <&reset>;
+    };
-- 
2.34.1

[PATCH v7 2/2] pwm: dwc: add of/platform support

[Xuyang Dong]

The dwc pwm controller can be used in non-PCI systems, so allow
either platform or OF based probing.

The controller is reset only when no PWM channel is enabled.
Otherwise, clocks are enabled and the runtime PM state is updated
to reflect the active hardware configuration.

Co-developed-by: Ben Dooks <ben.dooks@codethink.co.uk>
Signed-off-by: Ben Dooks <ben.dooks@codethink.co.uk>
Signed-off-by: Xiang Xu <xuxiang@eswincomputing.com>
Signed-off-by: Guosheng Wang <wangguosheng@eswincomputing.com>
Signed-off-by: Xuyang Dong <dongxuyang@eswincomputing.com>
---
 drivers/pwm/Kconfig        |  10 ++
 drivers/pwm/Makefile       |   1 +
 drivers/pwm/pwm-dwc-core.c | 101 ++++++++---
 drivers/pwm/pwm-dwc-of.c   | 346 +++++++++++++++++++++++++++++++++++++
 drivers/pwm/pwm-dwc.h      |  25 ++-
 5 files changed, 454 insertions(+), 29 deletions(-)
 create mode 100644 drivers/pwm/pwm-dwc-of.c

diff --git a/drivers/pwm/Kconfig b/drivers/pwm/Kconfig
index 6f3147518376..50aea24b6168 100644
--- a/drivers/pwm/Kconfig
+++ b/drivers/pwm/Kconfig
@@ -249,6 +249,16 @@ config PWM_DWC
 	  To compile this driver as a module, choose M here: the module
 	  will be called pwm-dwc.
 
+config PWM_DWC_OF
+	tristate "DesignWare PWM Controller (OF bus)"
+	depends on HAS_IOMEM && (OF || COMPILE_TEST)
+	select PWM_DWC_CORE
+	help
+	  PWM driver for Synopsys DWC PWM Controller on an OF bus or
+	  a platform bus.
+	  To compile this driver as a module, choose M here: the module
+	  will be called pwm-dwc-of.
+
 config PWM_EP93XX
 	tristate "Cirrus Logic EP93xx PWM support"
 	depends on ARCH_EP93XX || COMPILE_TEST
diff --git a/drivers/pwm/Makefile b/drivers/pwm/Makefile
index 0dc0d2b69025..470411a7e5ea 100644
--- a/drivers/pwm/Makefile
+++ b/drivers/pwm/Makefile
@@ -20,6 +20,7 @@ obj-$(CONFIG_PWM_CRC)		+= pwm-crc.o
 obj-$(CONFIG_PWM_CROS_EC)	+= pwm-cros-ec.o
 obj-$(CONFIG_PWM_DWC_CORE)	+= pwm-dwc-core.o
 obj-$(CONFIG_PWM_DWC)		+= pwm-dwc.o
+obj-$(CONFIG_PWM_DWC_OF)	+= pwm-dwc-of.o
 obj-$(CONFIG_PWM_EP93XX)	+= pwm-ep93xx.o
 obj-$(CONFIG_PWM_FSL_FTM)	+= pwm-fsl-ftm.o
 obj-$(CONFIG_PWM_GPIO)		+= pwm-gpio.o
diff --git a/drivers/pwm/pwm-dwc-core.c b/drivers/pwm/pwm-dwc-core.c
index 6dabec93a3c6..a6de05e27321 100644
--- a/drivers/pwm/pwm-dwc-core.c
+++ b/drivers/pwm/pwm-dwc-core.c
@@ -12,6 +12,7 @@
 #define DEFAULT_SYMBOL_NAMESPACE "dwc_pwm"
 
 #include <linux/bitops.h>
+#include <linux/clk.h>
 #include <linux/export.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
@@ -44,21 +45,52 @@ static int __dwc_pwm_configure_timer(struct dwc_pwm *dwc,
 	u32 high;
 	u32 low;
 
-	/*
-	 * Calculate width of low and high period in terms of input clock
-	 * periods and check are the result within HW limits between 1 and
-	 * 2^32 periods.
-	 */
-	tmp = DIV_ROUND_CLOSEST_ULL(state->duty_cycle, dwc->clk_ns);
-	if (tmp < 1 || tmp > (1ULL << 32))
-		return -ERANGE;
-	low = tmp - 1;
-
-	tmp = DIV_ROUND_CLOSEST_ULL(state->period - state->duty_cycle,
-				    dwc->clk_ns);
-	if (tmp < 1 || tmp > (1ULL << 32))
-		return -ERANGE;
-	high = tmp - 1;
+	if (dwc->clk)
+		dwc->clk_rate = clk_get_rate(dwc->clk);
+
+	if (dwc->features & DWC_TIM_CTRL_0N100PWM_EN) {
+		/*
+		 * Calculate width of low and high period in terms of input
+		 * clock periods and check are the result within HW limits
+		 * between 0 and 2^32 periods.
+		 */
+		tmp = state->duty_cycle * dwc->clk_rate;
+		tmp = DIV_ROUND_UP_ULL(tmp, NSEC_PER_SEC);
+		if (tmp >= (1ULL << 32))
+			return -ERANGE;
+
+		if (pwm->args.polarity == PWM_POLARITY_INVERSED)
+			high = tmp;
+		else
+			low = tmp;
+
+		tmp = (state->period - state->duty_cycle) * dwc->clk_rate;
+		tmp = DIV_ROUND_UP_ULL(tmp, NSEC_PER_SEC);
+		if (tmp >= (1ULL << 32))
+			return -ERANGE;
+
+		if (pwm->args.polarity == PWM_POLARITY_INVERSED)
+			low = tmp;
+		else
+			high = tmp;
+	} else {
+		/*
+		 * Calculate width of low and high period in terms of input
+		 * clock periods and check are the result within HW limits
+		 * between 1 and 2^32 periods.
+		 */
+		tmp = state->duty_cycle * dwc->clk_rate;
+		tmp = DIV_ROUND_UP_ULL(tmp, NSEC_PER_SEC);
+		if (tmp < 1 || tmp > (1ULL << 32))
+			return -ERANGE;
+		low = tmp - 1;
+
+		tmp = (state->period - state->duty_cycle) * dwc->clk_rate;
+		tmp = DIV_ROUND_UP_ULL(tmp, NSEC_PER_SEC);
+		if (tmp < 1 || tmp > (1ULL << 32))
+			return -ERANGE;
+		high = tmp - 1;
+	}
 
 	/*
 	 * Specification says timer usage flow is to disable timer, then
@@ -74,6 +106,7 @@ static int __dwc_pwm_configure_timer(struct dwc_pwm *dwc,
 	 * width of low period and latter the width of high period in terms
 	 * multiple of input clock periods:
 	 * Width = ((Count + 1) * input clock period).
+	 * Width = (Count * input clock period) : supported 0% and 100%.
 	 */
 	dwc_pwm_writel(dwc, low, DWC_TIM_LD_CNT(pwm->hwpwm));
 	dwc_pwm_writel(dwc, high, DWC_TIM_LD_CNT2(pwm->hwpwm));
@@ -85,6 +118,9 @@ static int __dwc_pwm_configure_timer(struct dwc_pwm *dwc,
 	 * periods are set by Load Count registers.
 	 */
 	ctrl = DWC_TIM_CTRL_MODE_USER | DWC_TIM_CTRL_PWM;
+	if (dwc->features & DWC_TIM_CTRL_0N100PWM_EN)
+		ctrl |= DWC_TIM_CTRL_0N100PWM_EN;
+
 	dwc_pwm_writel(dwc, ctrl, DWC_TIM_CTRL(pwm->hwpwm));
 
 	/*
@@ -121,11 +157,17 @@ static int dwc_pwm_get_state(struct pwm_chip *chip, struct pwm_device *pwm,
 			     struct pwm_state *state)
 {
 	struct dwc_pwm *dwc = to_dwc_pwm(chip);
+	unsigned long clk_rate;
 	u64 duty, period;
 	u32 ctrl, ld, ld2;
 
 	pm_runtime_get_sync(pwmchip_parent(chip));
 
+	if (dwc->clk)
+		dwc->clk_rate = clk_get_rate(dwc->clk);
+
+	clk_rate = dwc->clk_rate;
+
 	ctrl = dwc_pwm_readl(dwc, DWC_TIM_CTRL(pwm->hwpwm));
 	ld = dwc_pwm_readl(dwc, DWC_TIM_LD_CNT(pwm->hwpwm));
 	ld2 = dwc_pwm_readl(dwc, DWC_TIM_LD_CNT2(pwm->hwpwm));
@@ -137,17 +179,32 @@ static int dwc_pwm_get_state(struct pwm_chip *chip, struct pwm_device *pwm,
 	 * based on the timer load-count only.
 	 */
 	if (ctrl & DWC_TIM_CTRL_PWM) {
-		duty = (ld + 1) * dwc->clk_ns;
-		period = (ld2 + 1)  * dwc->clk_ns;
-		period += duty;
+		if (dwc->features & DWC_TIM_CTRL_0N100PWM_EN) {
+			if (pwm->args.polarity == PWM_POLARITY_INVERSED)
+				duty = ld2;
+			else
+				duty = ld;
+			period = (u64)ld + ld2;
+		} else {
+			duty = ld + 1;
+			period = ld2 + 1;
+			period += duty;
+		}
 	} else {
-		duty = (ld + 1) * dwc->clk_ns;
+		duty = ld + 1;
 		period = duty * 2;
 	}
 
 	state->polarity = PWM_POLARITY_INVERSED;
-	state->period = period;
-	state->duty_cycle = duty;
+	/*
+	 * If the ld register is at its maximum value. The duty value is
+	 * 4,294,967,295 (0xFFFF FFFF). The product (duty * NSEC_PER_SEC)
+	 * is guaranteed to be less than 2^64.
+	 */
+	duty *= NSEC_PER_SEC;
+	period *= NSEC_PER_SEC;
+	state->period = DIV_ROUND_UP_ULL(period, clk_rate);
+	state->duty_cycle = DIV_ROUND_UP_ULL(duty, clk_rate);
 
 	pm_runtime_put_sync(pwmchip_parent(chip));
 
@@ -169,7 +226,7 @@ struct pwm_chip *dwc_pwm_alloc(struct device *dev)
 		return chip;
 	dwc = to_dwc_pwm(chip);
 
-	dwc->clk_ns = 10;
+	dwc->clk_rate = NSEC_PER_SEC / 10;
 	chip->ops = &dwc_pwm_ops;
 
 	return chip;
diff --git a/drivers/pwm/pwm-dwc-of.c b/drivers/pwm/pwm-dwc-of.c
new file mode 100644
index 000000000000..d8b1606e51af
--- /dev/null
+++ b/drivers/pwm/pwm-dwc-of.c
@@ -0,0 +1,346 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * DesignWare PWM Controller driver OF
+ *
+ * Copyright (C) 2026 SiFive, Inc.
+ */
+
+#define DEFAULT_SYMBOL_NAMESPACE "dwc_pwm_of"
+
+#include <linux/clk.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+#include <linux/pwm.h>
+#include <linux/reset.h>
+
+#include "pwm-dwc.h"
+
+struct dwc_pwm_plat_data {
+	bool reset_required;
+};
+
+static int dwc_pwm_plat_probe(struct platform_device *pdev)
+{
+	const struct dwc_pwm_plat_data *pdata;
+	struct device *dev = &pdev->dev;
+	struct dwc_pwm_drvdata *data;
+	u32 ctrl[DWC_TIMERS_TOTAL];
+	struct pwm_chip *chip;
+	struct dwc_pwm *dwc;
+	bool pwm_en = false;
+	u32 nr_pwm, tim_id;
+	unsigned int i;
+	int ret;
+
+	data = devm_kzalloc(dev, struct_size(data, chips, 1), GFP_KERNEL);
+	if (!data)
+		return -ENOMEM;
+
+	chip = dwc_pwm_alloc(dev);
+	if (IS_ERR(chip))
+		return dev_err_probe(dev, PTR_ERR(chip),
+				     "failed to alloc pwm\n");
+
+	dwc = to_dwc_pwm(chip);
+
+	dwc->base = devm_platform_ioremap_resource(pdev, 0);
+	if (IS_ERR(dwc->base))
+		return PTR_ERR(dwc->base);
+
+	if (!device_property_read_u32(dev, "snps,pwm-number", &nr_pwm)) {
+		if (nr_pwm > DWC_TIMERS_TOTAL)
+			dev_warn(dev, "too many PWMs (%d), capping at %d\n",
+				 nr_pwm, chip->npwm);
+		else
+			chip->npwm = nr_pwm;
+	}
+
+	dwc->bus_clk = devm_clk_get(dev, "bus");
+	if (IS_ERR(dwc->bus_clk))
+		return dev_err_probe(dev, PTR_ERR(dwc->bus_clk),
+				     "failed to get bus clock\n");
+
+	dwc->clk = devm_clk_get(dev, "timer");
+	if (IS_ERR(dwc->clk))
+		return dev_err_probe(dev, PTR_ERR(dwc->clk),
+				     "failed to get timer clock\n");
+
+	ret = devm_clk_rate_exclusive_get(dev, dwc->clk);
+	if (ret)
+		return dev_err_probe(dev, ret,
+				     "failed to get exclusive rate\n");
+
+	dwc->clk_rate = clk_get_rate(dwc->clk);
+
+	pdata = device_get_match_data(dev);
+	if (pdata && pdata->reset_required)
+		dwc->rst = devm_reset_control_get_exclusive(dev, NULL);
+	else
+		dwc->rst = devm_reset_control_array_get_optional_exclusive(dev);
+
+	if (IS_ERR(dwc->rst))
+		return dev_err_probe(dev, PTR_ERR(dwc->rst),
+				     "failed to get reset control\n");
+
+	ret = clk_prepare_enable(dwc->bus_clk);
+	if (ret)
+		return dev_err_probe(dev, ret,
+				     "failed to enable bus clock\n");
+
+	ret = clk_prepare_enable(dwc->clk);
+	if (ret) {
+		dev_err(dev, "failed to enable timer clock\n");
+		goto disable_busclk;
+	}
+
+	/*
+	 * Check all channels to see if any channel is enabled.
+	 * Read the control register of each channel and extract the enable bit
+	 */
+	for (i = 0; i < chip->npwm; i++) {
+		ctrl[i] = dwc_pwm_readl(dwc, DWC_TIM_CTRL(i)) & DWC_TIM_CTRL_EN;
+		if (ctrl[i])
+			pwm_en = true;
+	}
+
+	/* Only issue reset when all channels are disabled */
+	if (!pwm_en) {
+		ret = reset_control_reset(dwc->rst);
+		if (ret) {
+			dev_err(dev, "failed to reset\n");
+			goto disable_clk;
+		}
+	}
+
+	/* init PWM feature */
+	dwc->features = 0;
+	/*
+	 * Support for 0% and 100% duty cycle mode was added in version 2.11a
+	 * and later.
+	 */
+	tim_id = dwc_pwm_readl(dwc, DWC_TIMERS_COMP_VERSION);
+	if (tim_id >= DWC_TIM_VERSION_ID_2_11A)
+		dwc->features |= DWC_TIM_CTRL_0N100PWM_EN;
+
+	ret = devm_pwmchip_add(dev, chip);
+	if (ret) {
+		dev_err(dev, "failed to add pwm chip\n");
+		goto reset_assert;
+	}
+
+	data->chips[0] = chip;
+	dev_set_drvdata(dev, data);
+
+	/*
+	 * If any PWM channel is enabled, mark device active and hold runtime PM
+	 * references for each enabled channel. Otherwise, gate the clocks.
+	 */
+	if (pwm_en) {
+		pm_runtime_set_active(dev);
+		for (i = 0; i < chip->npwm; i++) {
+			if (ctrl[i])
+				pm_runtime_get_noresume(dev);
+		}
+	} else {
+		clk_disable_unprepare(dwc->clk);
+		clk_disable_unprepare(dwc->bus_clk);
+	}
+
+	pm_runtime_enable(dev);
+
+	return 0;
+
+reset_assert:
+	reset_control_assert(dwc->rst);
+disable_clk:
+	clk_disable_unprepare(dwc->clk);
+disable_busclk:
+	clk_disable_unprepare(dwc->bus_clk);
+
+	return ret;
+}
+
+static void dwc_pwm_plat_remove(struct platform_device *pdev)
+{
+	struct dwc_pwm_drvdata *data = platform_get_drvdata(pdev);
+	struct pwm_chip *chip = data->chips[0];
+	struct dwc_pwm *dwc = to_dwc_pwm(chip);
+	bool pwm_en = false;
+	unsigned int idx;
+	bool pm_flags;
+
+	/*
+	 * Resume the device if it is runtime suspended to allow
+	 * safe register access.
+	 */
+	pm_flags = pm_runtime_status_suspended(&pdev->dev);
+	if (pm_flags)
+		pm_runtime_get_sync(&pdev->dev);
+
+	for (idx = 0; idx < chip->npwm; idx++) {
+		if (dwc_pwm_readl(dwc, DWC_TIM_CTRL(idx)) & DWC_TIM_CTRL_EN) {
+			pwm_en = true;
+			pm_runtime_put_noidle(&pdev->dev);
+		}
+	}
+
+	/*
+	 * Re-suspend the device if it was runtime suspended prior to
+	 * the register access.
+	 */
+	if (pm_flags)
+		pm_runtime_put_sync(&pdev->dev);
+
+	if (pwm_en) {
+		clk_disable_unprepare(dwc->clk);
+		clk_disable_unprepare(dwc->bus_clk);
+	}
+
+	pm_runtime_disable(&pdev->dev);
+	reset_control_assert(dwc->rst);
+}
+
+static int dwc_pwm_runtime_suspend(struct device *dev)
+{
+	struct dwc_pwm_drvdata *data = dev_get_drvdata(dev);
+	struct pwm_chip *chip = data->chips[0];
+	struct dwc_pwm *dwc = to_dwc_pwm(chip);
+
+	clk_disable_unprepare(dwc->clk);
+	clk_disable_unprepare(dwc->bus_clk);
+
+	return 0;
+}
+
+static int dwc_pwm_runtime_resume(struct device *dev)
+{
+	struct dwc_pwm_drvdata *data = dev_get_drvdata(dev);
+	struct pwm_chip *chip = data->chips[0];
+	struct dwc_pwm *dwc = to_dwc_pwm(chip);
+	int ret;
+
+	ret = clk_prepare_enable(dwc->bus_clk);
+	if (ret) {
+		dev_err(dev, "failed to enable bus clock: %d\n", ret);
+		return ret;
+	}
+
+	ret = clk_prepare_enable(dwc->clk);
+	if (ret) {
+		dev_err(dev, "failed to enable timer clock: %d\n", ret);
+		clk_disable_unprepare(dwc->bus_clk);
+		return ret;
+	}
+
+	return 0;
+}
+
+static int dwc_pwm_suspend(struct device *dev)
+{
+	struct dwc_pwm_drvdata *data = dev_get_drvdata(dev);
+	struct pwm_chip *chip = data->chips[0];
+	struct dwc_pwm *dwc = to_dwc_pwm(chip);
+	unsigned int idx;
+	int ret;
+
+	if (pm_runtime_status_suspended(dev)) {
+		ret = dwc_pwm_runtime_resume(dev);
+		if (ret)
+			return ret;
+	}
+
+	for (idx = 0; idx < chip->npwm; idx++) {
+		if (chip->pwms[idx].state.enabled)
+			return -EBUSY;
+
+		dwc->ctx[idx].cnt = dwc_pwm_readl(dwc, DWC_TIM_LD_CNT(idx));
+		dwc->ctx[idx].cnt2 = dwc_pwm_readl(dwc, DWC_TIM_LD_CNT2(idx));
+		dwc->ctx[idx].ctrl = dwc_pwm_readl(dwc, DWC_TIM_CTRL(idx));
+	}
+
+	ret = dwc_pwm_runtime_suspend(dev);
+	if (ret)
+		return ret;
+
+	return 0;
+}
+
+static int dwc_pwm_resume(struct device *dev)
+{
+	struct dwc_pwm_drvdata *data = dev_get_drvdata(dev);
+	struct pwm_chip *chip = data->chips[0];
+	struct dwc_pwm *dwc = to_dwc_pwm(chip);
+	unsigned int idx;
+	bool pm_flags;
+	int ret;
+
+	/* Check if device was runtime suspended before system resume */
+	pm_flags = pm_runtime_status_suspended(dev);
+	if (pm_flags) {
+		/*
+		 * Use PM framework to resume device
+		 * (calls dwc_pwm_runtime_resume)
+		 */
+		ret = pm_runtime_get_sync(dev);
+		if (ret < 0)
+			return ret;
+	} else {
+		/*
+		 * Device was active, but clocks might be off after system sleep
+		 * Call runtime_resume directly to restore hardware state
+		 */
+		ret = dwc_pwm_runtime_resume(dev);
+		if (ret)
+			return ret;
+	}
+
+	for (idx = 0; idx < chip->npwm; idx++) {
+		dwc_pwm_writel(dwc, dwc->ctx[idx].cnt, DWC_TIM_LD_CNT(idx));
+		dwc_pwm_writel(dwc, dwc->ctx[idx].cnt2, DWC_TIM_LD_CNT2(idx));
+		dwc_pwm_writel(dwc, dwc->ctx[idx].ctrl, DWC_TIM_CTRL(idx));
+	}
+
+	if (pm_flags) {
+		/* Balance the refcount taken by pm_runtime_get_sync
+		 * if it was used
+		 */
+		ret = pm_runtime_put_sync(dev);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+static const struct dev_pm_ops dwc_pwm_pm_ops = {
+	RUNTIME_PM_OPS(dwc_pwm_runtime_suspend, dwc_pwm_runtime_resume, NULL)
+	SYSTEM_SLEEP_PM_OPS(dwc_pwm_suspend, dwc_pwm_resume)
+};
+
+static const struct dwc_pwm_plat_data pwm_eic7700_pdata = {
+	.reset_required = true,
+};
+
+static const struct of_device_id dwc_pwm_dt_ids[] = {
+	{ .compatible = "snps,dw-apb-timers-pwm2" },
+	{ .compatible = "eswin,eic7700-pwm", .data = &pwm_eic7700_pdata },
+	{ }
+};
+MODULE_DEVICE_TABLE(of, dwc_pwm_dt_ids);
+
+static struct platform_driver dwc_pwm_plat_driver = {
+	.driver = {
+		.name = "dwc-pwm",
+		.pm = pm_ptr(&dwc_pwm_pm_ops),
+		.of_match_table = dwc_pwm_dt_ids,
+	},
+	.probe = dwc_pwm_plat_probe,
+	.remove = dwc_pwm_plat_remove,
+};
+
+module_platform_driver(dwc_pwm_plat_driver);
+
+MODULE_ALIAS("platform:dwc-pwm-of");
+MODULE_AUTHOR("Ben Dooks <ben.dooks@codethink.co.uk>");
+MODULE_DESCRIPTION("DesignWare PWM Controller");
+MODULE_LICENSE("GPL");
diff --git a/drivers/pwm/pwm-dwc.h b/drivers/pwm/pwm-dwc.h
index 1562594e7f85..75f7c2d031c4 100644
--- a/drivers/pwm/pwm-dwc.h
+++ b/drivers/pwm/pwm-dwc.h
@@ -26,12 +26,19 @@ MODULE_IMPORT_NS("dwc_pwm");
 #define DWC_TIMERS_TOTAL	8
 
 /* Timer Control Register */
-#define DWC_TIM_CTRL_EN		BIT(0)
-#define DWC_TIM_CTRL_MODE	BIT(1)
-#define DWC_TIM_CTRL_MODE_FREE	(0 << 1)
-#define DWC_TIM_CTRL_MODE_USER	(1 << 1)
-#define DWC_TIM_CTRL_INT_MASK	BIT(2)
-#define DWC_TIM_CTRL_PWM	BIT(3)
+#define DWC_TIM_CTRL_EN			BIT(0)
+#define DWC_TIM_CTRL_MODE		BIT(1)
+#define DWC_TIM_CTRL_MODE_FREE		(0 << 1)
+#define DWC_TIM_CTRL_MODE_USER		BIT(1)
+#define DWC_TIM_CTRL_INT_MASK		BIT(2)
+#define DWC_TIM_CTRL_PWM		BIT(3)
+#define DWC_TIM_CTRL_0N100PWM_EN	BIT(4)
+
+/*
+ * The version 2.11a and later add "Pulse Width Modulation with
+ * 0% and 100% Duty Cycle".
+ */
+#define DWC_TIM_VERSION_ID_2_11A	0x3231312a
 
 struct dwc_pwm_info {
 	unsigned int nr;
@@ -52,8 +59,12 @@ struct dwc_pwm_ctx {
 
 struct dwc_pwm {
 	void __iomem *base;
-	unsigned int clk_ns;
+	struct clk *bus_clk;
+	struct clk *clk;
+	unsigned long clk_rate;
+	struct reset_control *rst;
 	struct dwc_pwm_ctx ctx[DWC_TIMERS_TOTAL];
+	u32 features;
 };
 
 static inline struct dwc_pwm *to_dwc_pwm(struct pwm_chip *chip)
-- 
2.34.1

Re: [PATCH v7 2/2] pwm: dwc: add of/platform support

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 11 potential issue(s) to consider:

New issues:
- [High] Registering the PWM chip before completing runtime PM initialization exposes the hardware to consumers while clocks might be turned off.
- [High] Manual hardware teardown in `.remove()` precedes `pwm_chip` deregistration, leading to potential bus faults.
- [High] TOCTOU race in `remove()` when conditionally acquiring a runtime PM reference allows concurrent runtime suspend while accessing registers.
- [High] System resume is incorrectly aborted if expected runtime PM idle operations return `-EAGAIN`.
- [High] Unconditional assertion of the reset control on the probe error path and during module removal permanently kills active hardware and unbalances the reset API.
- [High] `remove()` artificially drops runtime PM references that belong to active consumers, causing an underflow.
- [Medium] Integer overflow in timer configuration allows invalid large duty cycle/period values to bypass validation.
- [Medium] Clock references are leaked during module removal if the device is held in an active PM state but no PWM channels are enabled.
- [Medium] The driver rounds UP duty cycle and period calculations instead of rounding DOWN, violating PWM subsystem rules.
- [Medium] The driver evaluates the static device tree polarity instead of the dynamically requested runtime polarity, mapping the duty cycle incorrectly.

Pre-existing issues:
- [High] `dwc_pwm_get_state()` ignores the return value of `pm_runtime_get_sync()`, allowing unprotected hardware access.
--

commit 517966ff9e8e50806716056dd4bd3f2f88b832c0
Author: Xuyang Dong <dongxuyang@eswincomputing.com>

pwm: dwc: add of/platform support

The dwc pwm controller can be used in non-PCI systems, so allow
either platform or OF based probing.

> diff --git a/drivers/pwm/pwm-dwc-core.c b/drivers/pwm/pwm-dwc-core.c
> index 6dabec93a3c64..a6de05e273216 100644
> --- a/drivers/pwm/pwm-dwc-core.c
> +++ b/drivers/pwm/pwm-dwc-core.c
[ ... ]
> @@ -44,21 +45,52 @@ static int __dwc_pwm_configure_timer(struct dwc_pwm *dwc,
[ ... ]
> +	if (dwc->features & DWC_TIM_CTRL_0N100PWM_EN) {
> +		/*
> +		 * Calculate width of low and high period in terms of input
> +		 * clock periods and check are the result within HW limits
> +		 * between 0 and 2^32 periods.
> +		 */
> +		tmp = state->duty_cycle * dwc->clk_rate;

[Severity: Medium]
Could this multiplication overflow the 64-bit boundary before the hardware
limit check if a consumer requests an extremely large period? Both duty_cycle
and period are u64 nanoseconds, while clk_rate is an unsigned long. A
wrapped-around small value would evade the hardware limit check entirely.
Should this use mul_u64_u64_div_u64() to scale the values safely?

> +		tmp = DIV_ROUND_UP_ULL(tmp, NSEC_PER_SEC);

[Severity: Medium]
Does this violate the PWM subsystem rules by rounding up instead of down?
The PWM core framework strictly requires that the .apply callback rounds down
the requested period and duty cycle to the nearest achievable hardware value.

> +		if (tmp >= (1ULL << 32))
> +			return -ERANGE;
> +
> +		if (pwm->args.polarity == PWM_POLARITY_INVERSED)
> +			high = tmp;
> +		else
> +			low = tmp;

[Severity: Medium]
Should this evaluate the dynamically requested state->polarity instead of the
static device tree polarity in pwm->args.polarity? Since dwc_pwm_apply()
enforces PWM_POLARITY_INVERSED, if the device tree was configured with
PWM_POLARITY_NORMAL, this will program the hardware with the inverted duty
cycle relative to what the consumer actually requested.

[ ... ]
> @@ -121,11 +157,17 @@ static int dwc_pwm_get_state(struct pwm_chip *chip, struct pwm_device *pwm,
>  			     struct pwm_state *state)
>  {
>  	struct dwc_pwm *dwc = to_dwc_pwm(chip);
> +	unsigned long clk_rate;
>  	u64 duty, period;
>  	u32 ctrl, ld, ld2;
>  
>  	pm_runtime_get_sync(pwmchip_parent(chip));

[Severity: High]
This isn't a bug introduced by this patch, but the return value of
pm_runtime_get_sync() is ignored here. If the function fails (for example,
returning -EACCES when PM runtime is disabled during driver removal, or before
it is enabled in probe), the hardware might not be powered on and clocks could
remain gated. The function will then proceed to read the DWC_TIM_CTRL and
DWC_TIM_LD_CNT registers, which can cause a synchronous external abort when
accessing unclocked hardware. Can we add a check for the return value here?

> diff --git a/drivers/pwm/pwm-dwc-of.c b/drivers/pwm/pwm-dwc-of.c
> new file mode 100644
> index 0000000000000..d8b1606e51af4
> --- /dev/null
> +++ b/drivers/pwm/pwm-dwc-of.c
[ ... ]
> +static int dwc_pwm_plat_probe(struct platform_device *pdev)
> +{
[ ... ]
> +	ret = devm_pwmchip_add(dev, chip);
> +	if (ret) {
> +		dev_err(dev, "failed to add pwm chip\n");
> +		goto reset_assert;
> +	}
> +
> +	data->chips[0] = chip;
> +	dev_set_drvdata(dev, data);
> +
> +	/*
> +	 * If any PWM channel is enabled, mark device active and hold runtime PM
> +	 * references for each enabled channel. Otherwise, gate the clocks.
> +	 */
> +	if (pwm_en) {
> +		pm_runtime_set_active(dev);
> +		for (i = 0; i < chip->npwm; i++) {
> +			if (ctrl[i])
> +				pm_runtime_get_noresume(dev);
> +		}
> +	} else {
> +		clk_disable_unprepare(dwc->clk);
> +		clk_disable_unprepare(dwc->bus_clk);
> +	}
> +
> +	pm_runtime_enable(dev);

[Severity: High]
Does registering the PWM chip before enabling PM expose the hardware to
consumers prematurely? devm_pwmchip_add() is called before pm_runtime_enable().
If a consumer accesses the PWM during this window, the core driver's
pm_runtime_get_sync() call would fail with -EACCES, but still proceed to access
registers. At the same time, probe might gate the clocks if pwm_en is false,
leading to a bus fault.

> +
> +	return 0;
> +
> +reset_assert:
> +	reset_control_assert(dwc->rst);

[Severity: High]
Does this unconditionally assert the reset control on the error path? The
commit message mentions resetting only when no PWM channel is enabled. If
pwm_en was true, this forcibly kills actively running hardware. If pwm_en was
false, probe used reset_control_reset() (which is a pulse), so asserting it
here leaves the line permanently asserted in an unbalanced state.

> +disable_clk:
> +	clk_disable_unprepare(dwc->clk);
> +disable_busclk:
> +	clk_disable_unprepare(dwc->bus_clk);
> +
> +	return ret;
> +}
> +
> +static void dwc_pwm_plat_remove(struct platform_device *pdev)
> +{
[ ... ]
> +	/*
> +	 * Resume the device if it is runtime suspended to allow
> +	 * safe register access.
> +	 */
> +	pm_flags = pm_runtime_status_suspended(&pdev->dev);
> +	if (pm_flags)
> +		pm_runtime_get_sync(&pdev->dev);

[Severity: High]
Is there a race condition when conditionally acquiring this PM reference? If
pm_runtime_status_suspended() returns false, the code proceeds to read
registers without taking a reference. An auto-suspend timer or a concurrent
consumer could suspend the device immediately after this check, turning off the
clocks and causing the subsequent DWC_TIM_CTRL reads to trigger a bus fault.
Should pm_runtime_get_sync() be called unconditionally?

> +
> +	for (idx = 0; idx < chip->npwm; idx++) {
> +		if (dwc_pwm_readl(dwc, DWC_TIM_CTRL(idx)) & DWC_TIM_CTRL_EN) {
> +			pwm_en = true;
> +			pm_runtime_put_noidle(&pdev->dev);
> +		}
> +	}

[Severity: High]
Does dropping the PM reference here underflow the usage count for active
consumers? If a consumer has taken over the PWM channel, it logically owns that
reference via dwc_pwm_apply(). Because devres unregisters the PWM chip after
remove() finishes, the PWM core will eventually call
dwc_pwm_apply(..., enabled=false), which calls pm_runtime_put_sync(). Dropping
it here artificially might lead to a usage count underflow.

> +
> +	/*
> +	 * Re-suspend the device if it was runtime suspended prior to
> +	 * the register access.
> +	 */
> +	if (pm_flags)
> +		pm_runtime_put_sync(&pdev->dev);
> +
> +	if (pwm_en) {
> +		clk_disable_unprepare(dwc->clk);
> +		clk_disable_unprepare(dwc->bus_clk);
> +	}

[Severity: Medium]
Will this leak clock references if the device is active but no PWM channels are
enabled? If pwm_en is false but the device's PM runtime state is RPM_ACTIVE
(due to an open sysfs handle or a pending auto-suspend), skipping this block
leaves the clocks on. The subsequent pm_runtime_disable() does not turn off
clocks, permanently leaking them.

> +
> +	pm_runtime_disable(&pdev->dev);
> +	reset_control_assert(dwc->rst);
> +}

[Severity: High]
Does manually tearing down the hardware here lead to bus faults during devres
cleanup? Since the driver registers the chip using devm_pwmchip_add(), devres
will unregister the PWM chip after dwc_pwm_plat_remove() returns. During that
window, the device is fully accessible to consumers, but clocks are off and
reset is asserted. Additionally, unconditionally asserting the reset here
ignores the earlier logic to keep active hardware running if pwm_en was true.

[ ... ]
> +static int dwc_pwm_resume(struct device *dev)
> +{
[ ... ]
> +	for (idx = 0; idx < chip->npwm; idx++) {
> +		dwc_pwm_writel(dwc, dwc->ctx[idx].cnt, DWC_TIM_LD_CNT(idx));
> +		dwc_pwm_writel(dwc, dwc->ctx[idx].cnt2, DWC_TIM_LD_CNT2(idx));
> +		dwc_pwm_writel(dwc, dwc->ctx[idx].ctrl, DWC_TIM_CTRL(idx));
> +	}
> +
> +	if (pm_flags) {
> +		/* Balance the refcount taken by pm_runtime_get_sync
> +		 * if it was used
> +		 */
> +		ret = pm_runtime_put_sync(dev);
> +		if (ret < 0)
> +			return ret;
> +	}
> +
> +	return 0;
> +}

[Severity: High]
Should a legitimate -EAGAIN from pm_runtime_put_sync() abort the system resume?
According to PM subsystem rules, pm_runtime_put_sync() can return -EAGAIN if
another thread holds a usage reference. Returning this expected error code from
a system resume callback will unexpectedly abort the entire system resume
process for the device.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260605082242.1541-1-dongxuyang@eswincomputing.com?part=2