Skip to contentSkip to conversation

i2c: designware: Enable mode swapping

PATCH Superseded Heikki Krogerus 4 participants 16 messages Dec 18, 2025 – Jan 19

Subsystems (4)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

v2 cover i2c: designware: Enable mode swapping

Hi,

Changed since v1:
- The init cb is replaced with i2c_dw_init() also in i2c-designware-amdisp.c
- Checking I2C_FUNC_SLAVE in i2c_dw_reg_slave()

The original cover letter:

It's currently not possible to support MCTP or any other protocol that requires support for both modes with the DesignWare I2C because the driver can only be used in one mode. I'm assuming that the driver was designed this way because the DesignWare I2C can not be operated as I2C master and I2C slave simultaneously, however, that does not actually mean both modes could not be supported at the same time. See the patch 5/6 for more detailed explanation.

This series will enable support for both modes in the driver by utilising a simple mode swap method, and that way make it possible to support MCTP, IPMI, SMBus Host Notification Protocol, and any other protocol requires the support for both modes with the DesignWare I2C.

thanks,

Heikki Krogerus (6):
i2c: designware: Remove useless driver specific option for I2C target
i2c: designware: Remove unnecessary function exports
i2c: designware: Combine some of the common functions
i2c: designware: Combine the init functions
i2c: designware: Enable mode swapping
i2c: designware: Remove an unnecessary condition

 drivers/i2c/busses/Kconfig                 |  10 +-
 drivers/i2c/busses/Makefile                |   2 +-
 drivers/i2c/busses/i2c-designware-amdisp.c |   4 +-
 drivers/i2c/busses/i2c-designware-common.c | 211 +++++++++++++++++++--
 drivers/i2c/busses/i2c-designware-core.h   |  25 +--
 drivers/i2c/busses/i2c-designware-master.c | 174 ++---------------
 drivers/i2c/busses/i2c-designware-slave.c  | 127 ++-----------
 7 files changed, 254 insertions(+), 299 deletions(-)
Signature · 1 line
2.50.1
#1 lore raw

PATCH v2 1/6 i2c: designware: Remove useless driver specific option for I2C target

The generic option for I2C target is already user selectable, which makes the DesignWare specific option completely unnecessary. The DesignWare option also silently selected I2C_SLAVE instead of depending on it without any real need for it.

---
 drivers/i2c/busses/Kconfig               | 10 ++--------
 drivers/i2c/busses/Makefile              |  2 +-
 drivers/i2c/busses/i2c-designware-core.h |  2 +-
 3 files changed, 4 insertions(+), 10 deletions(-)
drivers/i2c/busses/Kconfig +2 -8
index 09ba55bae1fa..860812e224a0 100644
@@ -569,20 +569,14 @@ config I2C_DESIGNWARE_CORE
 	help
 	  This option enables support for the Synopsys DesignWare I2C adapter.
 	  This driver includes support for the I2C host on the Synopsys
-	  Designware I2C adapter.
+	  Designware I2C adapter, and the I2C slave when enabled (select
+	  I2C_SLAVE).
 
 	  To compile the driver as a module, choose M here: the module will be
 	  called i2c-designware-core.
 
 if I2C_DESIGNWARE_CORE
 
-config I2C_DESIGNWARE_SLAVE
-	bool "Synopsys DesignWare Slave"
-	select I2C_SLAVE
-	help
-	  If you say yes to this option, support will be included for the
-	  Synopsys DesignWare I2C slave adapter.
-
 config I2C_DESIGNWARE_PLATFORM
 	tristate "Synopsys DesignWare Platform driver"
 	depends on (ACPI && COMMON_CLK) || !ACPI
drivers/i2c/busses/Makefile +1 -1
index fb985769f5ff..547123ab351f 100644
@@ -53,7 +53,7 @@ obj-$(CONFIG_I2C_DAVINCI)	+= i2c-davinci.o
 obj-$(CONFIG_I2C_DESIGNWARE_CORE)			+= i2c-designware-core.o
 i2c-designware-core-y					:= i2c-designware-common.o
 i2c-designware-core-y					+= i2c-designware-master.o
-i2c-designware-core-$(CONFIG_I2C_DESIGNWARE_SLAVE) 	+= i2c-designware-slave.o
+i2c-designware-core-$(CONFIG_I2C_SLAVE) 		+= i2c-designware-slave.o
 obj-$(CONFIG_I2C_DESIGNWARE_PLATFORM)			+= i2c-designware-platform.o
 i2c-designware-platform-y 				:= i2c-designware-platdrv.o
 i2c-designware-platform-$(CONFIG_I2C_DESIGNWARE_AMDPSP)	+= i2c-designware-amdpsp.o
drivers/i2c/busses/i2c-designware-core.h +1 -1
index bb5ce0a382f9..2a7decc24931 100644
@@ -386,7 +386,7 @@ void i2c_dw_disable(struct dw_i2c_dev *dev);
 extern void i2c_dw_configure_master(struct dw_i2c_dev *dev);
 extern int i2c_dw_probe_master(struct dw_i2c_dev *dev);
 
-#if IS_ENABLED(CONFIG_I2C_DESIGNWARE_SLAVE)
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
 extern void i2c_dw_configure_slave(struct dw_i2c_dev *dev);
 extern int i2c_dw_probe_slave(struct dw_i2c_dev *dev);
 #else
Signature · 1 line
2.50.1
#2 lore raw

PATCH v2 2/6 i2c: designware: Remove unnecessary function exports

In reply to Heikki Krogerus

The master and slave probe functions are only called from the core.

---
 drivers/i2c/busses/i2c-designware-master.c | 1 -
 drivers/i2c/busses/i2c-designware-slave.c  | 1 -
 2 files changed, 2 deletions(-)
drivers/i2c/busses/i2c-designware-master.c +0 -1
index f247cf323207..15b3a46f0132 100644
@@ -1101,7 +1101,6 @@ int i2c_dw_probe_master(struct dw_i2c_dev *dev)
 
 	return ret;
 }
-EXPORT_SYMBOL_GPL(i2c_dw_probe_master);
 
 MODULE_DESCRIPTION("Synopsys DesignWare I2C bus master adapter");
 MODULE_LICENSE("GPL");
drivers/i2c/busses/i2c-designware-slave.c +0 -1
index 6eb16b7d75a6..1995be79544d 100644
@@ -277,7 +277,6 @@ int i2c_dw_probe_slave(struct dw_i2c_dev *dev)
 
 	return ret;
 }
-EXPORT_SYMBOL_GPL(i2c_dw_probe_slave);
 
 MODULE_AUTHOR("Luis Oliveira <[email protected]>");
 MODULE_DESCRIPTION("Synopsys DesignWare I2C bus slave adapter");
Signature · 1 line
2.50.1
#3 lore raw

PATCH v2 3/6 i2c: designware: Combine some of the common functions

In reply to Heikki Krogerus

The adapter can be registered just in the core instead of separately in the master and slave drivers. The same applies to the interrupt.

---
 drivers/i2c/busses/i2c-designware-common.c | 108 +++++++++++++++++++--
 drivers/i2c/busses/i2c-designware-core.h   |  11 ++-
 drivers/i2c/busses/i2c-designware-master.c |  97 +++---------------
 drivers/i2c/busses/i2c-designware-slave.c  |  53 ++--------
 4 files changed, 126 insertions(+), 143 deletions(-)
drivers/i2c/busses/i2c-designware-common.c +101 -7
index 5b1e8f74c4ac..1823e4b71004 100644
@@ -131,7 +131,7 @@ static int dw_reg_write_word(void *context, unsigned int reg, unsigned int val)
  *
  * Return: 0 on success, or negative errno otherwise.
  */
-int i2c_dw_init_regmap(struct dw_i2c_dev *dev)
+static int i2c_dw_init_regmap(struct dw_i2c_dev *dev)
 {
 	struct regmap_config map_cfg = {
 		.reg_bits = 32,
@@ -457,7 +457,7 @@ u32 i2c_dw_scl_lcnt(struct dw_i2c_dev *dev, unsigned int reg, u32 ic_clk,
 	return DIV_ROUND_CLOSEST_ULL((u64)ic_clk * (tLOW + tf), MICRO) - 1 + offset;
 }
 
-int i2c_dw_set_sda_hold(struct dw_i2c_dev *dev)
+static int i2c_dw_set_sda_hold(struct dw_i2c_dev *dev)
 {
 	unsigned int reg;
 	int ret;
@@ -672,7 +672,7 @@ int i2c_dw_handle_tx_abort(struct dw_i2c_dev *dev)
 	return -EIO;
 }
 
-int i2c_dw_set_fifo_size(struct dw_i2c_dev *dev)
+static int i2c_dw_set_fifo_size(struct dw_i2c_dev *dev)
 {
 	u32 tx_fifo_depth, rx_fifo_depth;
 	unsigned int param;
@@ -741,19 +741,113 @@ void i2c_dw_disable(struct dw_i2c_dev *dev)
 }
 EXPORT_SYMBOL_GPL(i2c_dw_disable);
 
+static irqreturn_t i2c_dw_isr(int this_irq, void *dev_id)
+{
+	struct dw_i2c_dev *dev = dev_id;
+
+	if (dev->mode == DW_IC_SLAVE)
+		return i2c_dw_isr_slave(dev);
+
+	return i2c_dw_isr_master(dev);
+}
+
+static const struct i2c_algorithm i2c_dw_algo = {
+	.xfer = i2c_dw_xfer,
+	.functionality = i2c_dw_func,
+#if IS_ENABLED(CONFIG_I2C_SLAVE)
+	.reg_slave = i2c_dw_reg_slave,
+	.unreg_slave = i2c_dw_unreg_slave,
+#endif
+};
+
+static const struct i2c_adapter_quirks i2c_dw_quirks = {
+	.flags = I2C_AQ_NO_ZERO_LEN,
+};
+
 int i2c_dw_probe(struct dw_i2c_dev *dev)
 {
+	struct i2c_adapter *adap = &dev->adapter;
+	unsigned long irq_flags;
+	int ret;
+
 	device_set_node(&dev->adapter.dev, dev_fwnode(dev->dev));
 
+	ret = i2c_dw_init_regmap(dev);
+	if (ret)
+		return ret;
+
+	ret = i2c_dw_set_sda_hold(dev);
+	if (ret)
+		return ret;
+
+	ret = i2c_dw_set_fifo_size(dev);
+	if (ret)
+		return ret;
+
 	switch (dev->mode) {
 	case DW_IC_SLAVE:
-		return i2c_dw_probe_slave(dev);
+		ret = i2c_dw_probe_slave(dev);
+		break;
 	case DW_IC_MASTER:
-		return i2c_dw_probe_master(dev);
+		ret = i2c_dw_probe_master(dev);
+		break;
 	default:
-		dev_err(dev->dev, "Wrong operation mode: %d\n", dev->mode);
-		return -EINVAL;
+		ret = -EINVAL;
+		break;
 	}
+	if (ret)
+		return ret;
+
+	ret = dev->init(dev);
+	if (ret)
+		return ret;
+
+	if (!adap->name[0])
+		strscpy(adap->name, "Synopsys DesignWare I2C adapter");
+
+	adap->retries = 3;
+	adap->algo = &i2c_dw_algo;
+	adap->quirks = &i2c_dw_quirks;
+	adap->dev.parent = dev->dev;
+	i2c_set_adapdata(adap, dev);
+
+	/*
+	 * REVISIT: The mode check may not be necessary.
+	 * For now keeping the flags as they were originally.
+	 */
+	if (dev->mode == DW_IC_SLAVE)
+		irq_flags = IRQF_SHARED;
+	else if (dev->flags & ACCESS_NO_IRQ_SUSPEND)
+		irq_flags = IRQF_NO_SUSPEND;
+	else
+		irq_flags = IRQF_SHARED | IRQF_COND_SUSPEND;
+
+	ret = i2c_dw_acquire_lock(dev);
+	if (ret)
+		return ret;
+
+	__i2c_dw_write_intr_mask(dev, 0);
+	i2c_dw_release_lock(dev);
+
+	if (!(dev->flags & ACCESS_POLLING)) {
+		ret = devm_request_irq(dev->dev, dev->irq, i2c_dw_isr,
+				       irq_flags, dev_name(dev->dev), dev);
+		if (ret)
+			return ret;
+	}
+
+	/*
+	 * Increment PM usage count during adapter registration in order to
+	 * avoid possible spurious runtime suspend when adapter device is
+	 * registered to the device core and immediate resume in case bus has
+	 * registered I2C slaves that do I2C transfers in their probe.
+	 */
+	ACQUIRE(pm_runtime_noresume, pm)(dev->dev);
+	ret = ACQUIRE_ERR(pm_runtime_noresume, &pm);
+	if (ret)
+		return ret;
+
+	return i2c_add_numbered_adapter(adap);
 }
 EXPORT_SYMBOL_GPL(i2c_dw_probe);
 
drivers/i2c/busses/i2c-designware-core.h +8 -3
index 2a7decc24931..0f58c4b50377 100644
@@ -13,6 +13,7 @@
 #include <linux/completion.h>
 #include <linux/errno.h>
 #include <linux/i2c.h>
+#include <linux/irqreturn.h>
 #include <linux/pm.h>
 #include <linux/regmap.h>
 #include <linux/types.h>
@@ -333,20 +334,18 @@ struct i2c_dw_semaphore_callbacks {
 	int	(*probe)(struct dw_i2c_dev *dev);
 };
 
-int i2c_dw_init_regmap(struct dw_i2c_dev *dev);
 u32 i2c_dw_scl_hcnt(struct dw_i2c_dev *dev, unsigned int reg, u32 ic_clk,
 		    u32 tSYMBOL, u32 tf, int offset);
 u32 i2c_dw_scl_lcnt(struct dw_i2c_dev *dev, unsigned int reg, u32 ic_clk,
 		    u32 tLOW, u32 tf, int offset);
-int i2c_dw_set_sda_hold(struct dw_i2c_dev *dev);
 u32 i2c_dw_clk_rate(struct dw_i2c_dev *dev);
 int i2c_dw_prepare_clk(struct dw_i2c_dev *dev, bool prepare);
 int i2c_dw_acquire_lock(struct dw_i2c_dev *dev);
 void i2c_dw_release_lock(struct dw_i2c_dev *dev);
 int i2c_dw_wait_bus_not_busy(struct dw_i2c_dev *dev);
 int i2c_dw_handle_tx_abort(struct dw_i2c_dev *dev);
-int i2c_dw_set_fifo_size(struct dw_i2c_dev *dev);
 u32 i2c_dw_func(struct i2c_adapter *adap);
+irqreturn_t i2c_dw_isr_master(struct dw_i2c_dev *dev);
 
 extern const struct dev_pm_ops i2c_dw_dev_pm_ops;
 
@@ -386,12 +385,18 @@ void i2c_dw_disable(struct dw_i2c_dev *dev);
 extern void i2c_dw_configure_master(struct dw_i2c_dev *dev);
 extern int i2c_dw_probe_master(struct dw_i2c_dev *dev);
 
+int i2c_dw_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs, int num);
+
 #if IS_ENABLED(CONFIG_I2C_SLAVE)
 extern void i2c_dw_configure_slave(struct dw_i2c_dev *dev);
 extern int i2c_dw_probe_slave(struct dw_i2c_dev *dev);
+irqreturn_t i2c_dw_isr_slave(struct dw_i2c_dev *dev);
+int i2c_dw_reg_slave(struct i2c_client *client);
+int i2c_dw_unreg_slave(struct i2c_client *client);
 #else
 static inline void i2c_dw_configure_slave(struct dw_i2c_dev *dev) { }
 static inline int i2c_dw_probe_slave(struct dw_i2c_dev *dev) { return -EINVAL; }
+static inline irqreturn_t i2c_dw_isr_slave(struct dw_i2c_dev *dev) { return IRQ_NONE; }
 #endif
 
 static inline void i2c_dw_configure(struct dw_i2c_dev *dev)
drivers/i2c/busses/i2c-designware-master.c +11 -86

154 lines not on this page · raw message

drivers/i2c/busses/i2c-designware-slave.c +6 -47

87 lines not on this page · raw message

Signature · 1 line
2.50.1
#4 lore raw

Re: [PATCH v2 3/6] i2c: designware: Combine some of the common functions

Quoted 3 lines · Heikki Krogerus

The adapter can be registered just in the core instead of separately in the master and slave drivers. The same applies to the interrupt.

...

Quoted 34 lines

int i2c_dw_probe(struct dw_i2c_dev *dev)
{

drivers/i2c/busses/i2c-designware-common.c +22 -4
+	struct i2c_adapter *adap = &dev->adapter;
+	unsigned long irq_flags;
+	int ret;
+
 	device_set_node(&dev->adapter.dev, dev_fwnode(dev->dev));
 
+	ret = i2c_dw_init_regmap(dev);
+	if (ret)
+		return ret;
+
+	ret = i2c_dw_set_sda_hold(dev);
+	if (ret)
+		return ret;
+
+	ret = i2c_dw_set_fifo_size(dev);
+	if (ret)
+		return ret;
+
 	switch (dev->mode) {
 	case DW_IC_SLAVE:
-		return i2c_dw_probe_slave(dev);
+		ret = i2c_dw_probe_slave(dev);
+		break;
 	case DW_IC_MASTER:
-		return i2c_dw_probe_master(dev);
+		ret = i2c_dw_probe_master(dev);
+		break;
 	default:
-		dev_err(dev->dev, "Wrong operation mode: %d\n", dev->mode);
-		return -EINVAL;
+		ret = -EINVAL;
+		break;

drivers/i2c/busses/i2c-designware-common.c void i2c_dw_disable(struct dw_i2c_dev *dev)

Quoted 10 lines
drivers/i2c/busses/i2c-designware-common.c +8 -0
 	}
+	if (ret)
+		return ret;
+
+	ret = dev->init(dev);
+	if (ret)
+		return ret;
 
+	if (!adap->name[0])
+		strscpy(adap->name, "Synopsys DesignWare I2C adapter");

Hmm... See below.

Quoted 44 lines
drivers/i2c/busses/i2c-designware-common.c +43 -0
+	adap->retries = 3;
+	adap->algo = &i2c_dw_algo;
+	adap->quirks = &i2c_dw_quirks;
+	adap->dev.parent = dev->dev;
+	i2c_set_adapdata(adap, dev);
+
+	/*
+	 * REVISIT: The mode check may not be necessary.
+	 * For now keeping the flags as they were originally.
+	 */
+	if (dev->mode == DW_IC_SLAVE)
+		irq_flags = IRQF_SHARED;
+	else if (dev->flags & ACCESS_NO_IRQ_SUSPEND)
+		irq_flags = IRQF_NO_SUSPEND;
+	else
+		irq_flags = IRQF_SHARED | IRQF_COND_SUSPEND;
+
+	ret = i2c_dw_acquire_lock(dev);
+	if (ret)
+		return ret;
+
+	__i2c_dw_write_intr_mask(dev, 0);
+	i2c_dw_release_lock(dev);
+
+	if (!(dev->flags & ACCESS_POLLING)) {
+		ret = devm_request_irq(dev->dev, dev->irq, i2c_dw_isr,
+				       irq_flags, dev_name(dev->dev), dev);
+		if (ret)
+			return ret;
+	}
+
+	/*
+	 * Increment PM usage count during adapter registration in order to
+	 * avoid possible spurious runtime suspend when adapter device is
+	 * registered to the device core and immediate resume in case bus has
+	 * registered I2C slaves that do I2C transfers in their probe.
+	 */
+	ACQUIRE(pm_runtime_noresume, pm)(dev->dev);
+	ret = ACQUIRE_ERR(pm_runtime_noresume, &pm);
+	if (ret)
+		return ret;
+
+	return i2c_add_numbered_adapter(adap);
 }

...

drivers/i2c/busses/i2c-designware-slave.c EXPORT_SYMBOL_GPL(i2c_dw_configure_slave); · line 261

Quoted 2 lines
drivers/i2c/busses/i2c-designware-slave.c +0 -2
-	snprintf(adap->name, sizeof(adap->name),
-		 "Synopsys DesignWare I2C Slave adapter");

This patch changes the user visible strings (via i2cdetect) or module names in case we want to find it by name (note, we already have such precedents for other adapters). Currently we have three variants if I not miss anything:
Generic for master (as in this change), Generic for slave, and AMD platform driver case. If you think this is okay change, then just drop the AMD case as well, and hence remove the no more needed conditional. Otherwise I would somehow group this naming in one place, if possible.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#5 lore raw

Re: [PATCH v2 3/6] i2c: designware: Combine some of the common functions

Hi Andy,

Quoted 10 lines

drivers/i2c/busses/i2c-designware-slave.c EXPORT_SYMBOL_GPL(i2c_dw_configure_slave); · line 261

Quoted 2 lines
drivers/i2c/busses/i2c-designware-slave.c +0 -2
-	snprintf(adap->name, sizeof(adap->name),
-		 "Synopsys DesignWare I2C Slave adapter");

This patch changes the user visible strings (via i2cdetect) or module names in case we want to find it by name (note, we already have such precedents for other adapters). Currently we have three variants if I not miss anything:
Generic for master (as in this change), Generic for slave, and AMD platform driver case. If you think this is okay change, then just drop the AMD case as well, and hence remove the no more needed conditional. Otherwise I would somehow group this naming in one place, if possible.

The only thing that this will change is, it removes the common slave/target only description, because after this that setup is no longer possible - master mode is now always supported. So this is the correct thing to do.

I don't think the user space should ever rely on a description like this except possibly with some customised/non-common systems that the user space really has to handle in some specific way, but if something really did rely on this common "target only" description, it could have only used it to determine that it basically can't use the device for anything as it's slave/target only - so basically to use it to check the functionality (same as i2cdetect -F). But as said, this is no longer a problem.

As for the AMD case, if I understood what you are proposing, I disagree with you. The glue drivers should always be allowed to assign the name (these would be the "non-common" systems that the user space may actually need to know about). I'm also against grouping the naming. The glue drivers must handle the platform specifics including the naming if needed, not the core.

thanks,

Signature · 1 line
heikki
#6 lore raw

Re: [PATCH v2 3/6] i2c: designware: Combine some of the common functions

Tue, Dec 30, 2025 at 04:14:32PM +0200, Heikki Krogerus kirjoitti:

Quoted 27 lines

Hi Andy,

Quoted 10 lines

drivers/i2c/busses/i2c-designware-slave.c EXPORT_SYMBOL_GPL(i2c_dw_configure_slave); · line 261

Quoted 2 lines
drivers/i2c/busses/i2c-designware-slave.c +0 -2
-	snprintf(adap->name, sizeof(adap->name),
-		 "Synopsys DesignWare I2C Slave adapter");

This patch changes the user visible strings (via i2cdetect) or module names in case we want to find it by name (note, we already have such precedents for other adapters). Currently we have three variants if I not miss anything:
Generic for master (as in this change), Generic for slave, and AMD platform driver case. If you think this is okay change, then just drop the AMD case as well, and hence remove the no more needed conditional. Otherwise I would somehow group this naming in one place, if possible.

The only thing that this will change is, it removes the common slave/target only description, because after this that setup is no longer possible - master mode is now always supported. So this is the correct thing to do.

I don't think the user space should ever rely on a description like this except possibly with some customised/non-common systems that the user space really has to handle in some specific way, but if something really did rely on this common "target only" description, it could have only used it to determine that it basically can't use the device for anything as it's slave/target only - so basically to use it to check the functionality (same as i2cdetect -F). But as said, this is no longer a problem.

Quoted 6 lines

As for the AMD case, if I understood what you are proposing, I disagree with you. The glue drivers should always be allowed to assign the name (these would be the "non-common" systems that the user space may actually need to know about). I'm also against grouping the naming. The glue drivers must handle the platform specifics including the naming if needed, not the core.

Ping.

Signature · 1 line
heikki
#7 lore raw

Re: [PATCH v2 3/6] i2c: designware: Combine some of the common functions

Quoted 1 line · Heikki Krogerus

Tue, Dec 30, 2025 at 04:14:32PM +0200, Heikki Krogerus kirjoitti:

...

Quoted 25 lines
Quoted 25 lines
Quoted 10 lines

drivers/i2c/busses/i2c-designware-slave.c EXPORT_SYMBOL_GPL(i2c_dw_configure_slave); · line 261

Quoted 2 lines
drivers/i2c/busses/i2c-designware-slave.c +0 -2
-	snprintf(adap->name, sizeof(adap->name),
-		 "Synopsys DesignWare I2C Slave adapter");

This patch changes the user visible strings (via i2cdetect) or module names in case we want to find it by name (note, we already have such precedents for other adapters). Currently we have three variants if I not miss anything:
Generic for master (as in this change), Generic for slave, and AMD platform driver case. If you think this is okay change, then just drop the AMD case as well, and hence remove the no more needed conditional. Otherwise I would somehow group this naming in one place, if possible.

The only thing that this will change is, it removes the common slave/target only description, because after this that setup is no longer possible - master mode is now always supported. So this is the correct thing to do.

I don't think the user space should ever rely on a description like this except possibly with some customised/non-common systems that the user space really has to handle in some specific way, but if something really did rely on this common "target only" description, it could have only used it to determine that it basically can't use the device for anything as it's slave/target only - so basically to use it to check the functionality (same as i2cdetect -F). But as said, this is no longer a problem.

Quoted 8 lines
Quoted 6 lines

As for the AMD case, if I understood what you are proposing, I disagree with you. The glue drivers should always be allowed to assign the name (these would be the "non-common" systems that the user space may actually need to know about). I'm also against grouping the naming. The glue drivers must handle the platform specifics including the naming if needed, not the core.

Ping.

Do you need my input on this? If you think this is good change, make sure the summary of the above goes to the commit message.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#8 lore raw

Re: [PATCH v2 3/6] i2c: designware: Combine some of the common functions

Mon, Jan 19, 2026 at 03:44:22PM +0200, Andy Shevchenko kirjoitti:

Quoted 37 lines · Heikki Krogerus
Quoted 1 line

Tue, Dec 30, 2025 at 04:14:32PM +0200, Heikki Krogerus kirjoitti:

...

Quoted 32 lines
Quoted 31 lines
Quoted 10 lines

drivers/i2c/busses/i2c-designware-slave.c EXPORT_SYMBOL_GPL(i2c_dw_configure_slave); · line 261

Quoted 2 lines
drivers/i2c/busses/i2c-designware-slave.c +0 -2
-	snprintf(adap->name, sizeof(adap->name),
-		 "Synopsys DesignWare I2C Slave adapter");

This patch changes the user visible strings (via i2cdetect) or module names in case we want to find it by name (note, we already have such precedents for other adapters). Currently we have three variants if I not miss anything:
Generic for master (as in this change), Generic for slave, and AMD platform driver case. If you think this is okay change, then just drop the AMD case as well, and hence remove the no more needed conditional. Otherwise I would somehow group this naming in one place, if possible.

The only thing that this will change is, it removes the common slave/target only description, because after this that setup is no longer possible - master mode is now always supported. So this is the correct thing to do.

I don't think the user space should ever rely on a description like this except possibly with some customised/non-common systems that the user space really has to handle in some specific way, but if something really did rely on this common "target only" description, it could have only used it to determine that it basically can't use the device for anything as it's slave/target only - so basically to use it to check the functionality (same as i2cdetect -F). But as said, this is no longer a problem.

As for the AMD case, if I understood what you are proposing, I disagree with you. The glue drivers should always be allowed to assign the name (these would be the "non-common" systems that the user space may actually need to know about). I'm also against grouping the naming. The glue drivers must handle the platform specifics including the naming if needed, not the core.

Quoted 4 lines
Quoted 1 line

Ping.

Do you need my input on this? If you think this is good change, make sure the summary of the above goes to the commit message.

Got it.

Thanks Andy,

Signature · 1 line
heikki
#9 lore raw

PATCH v2 4/6 i2c: designware: Combine the init functions

In reply to Heikki Krogerus

Providing a single function for controller initialisation. The controller initialisation has the same steps for master and slave modes, except the timing parameters are only needed in master mode.

---
 drivers/i2c/busses/i2c-designware-amdisp.c |  4 +-
 drivers/i2c/busses/i2c-designware-common.c | 81 +++++++++++++++++++++-
 drivers/i2c/busses/i2c-designware-core.h   |  3 +-
 drivers/i2c/busses/i2c-designware-master.c | 70 +------------------
 drivers/i2c/busses/i2c-designware-slave.c  | 44 ------------
 5 files changed, 85 insertions(+), 117 deletions(-)
drivers/i2c/busses/i2c-designware-amdisp.c +2 -2
index 450793d5f839..ec9259dd2a4f 100644
@@ -163,8 +163,8 @@ static int amd_isp_dw_i2c_plat_runtime_resume(struct device *dev)
 
 	if (!i_dev->shared_with_punit)
 		i2c_dw_prepare_clk(i_dev, true);
-	if (i_dev->init)
-		i_dev->init(i_dev);
+
+	i2c_dw_init(i_dev);
 
 	return 0;
 }
drivers/i2c/busses/i2c-designware-common.c +79 -2

99 lines not on this page · raw message

drivers/i2c/busses/i2c-designware-core.h +1 -2
index 0f58c4b50377..82465b134c34 100644
@@ -240,7 +240,6 @@ struct reset_control;
  * @semaphore_idx: Index of table with semaphore type attached to the bus. It's
  *	-1 if there is no semaphore.
  * @shared_with_punit: true if this bus is shared with the SoC's PUNIT
- * @init: function to initialize the I2C hardware
  * @set_sda_hold_time: callback to retrieve IP specific SDA hold timing
  * @mode: operation mode - DW_IC_MASTER or DW_IC_SLAVE
  * @rinfo: I²C GPIO recovery information
@@ -301,7 +300,6 @@ struct dw_i2c_dev {
 	void			(*release_lock)(void);
 	int			semaphore_idx;
 	bool			shared_with_punit;
-	int			(*init)(struct dw_i2c_dev *dev);
 	int			(*set_sda_hold_time)(struct dw_i2c_dev *dev);
 	int			mode;
 	struct i2c_bus_recovery_info rinfo;
@@ -408,6 +406,7 @@ static inline void i2c_dw_configure(struct dw_i2c_dev *dev)
 }
 
 int i2c_dw_probe(struct dw_i2c_dev *dev);
+int i2c_dw_init(struct dw_i2c_dev *dev);
 
 #if IS_ENABLED(CONFIG_I2C_DESIGNWARE_BAYTRAIL)
 int i2c_dw_baytrail_probe_lock_support(struct dw_i2c_dev *dev);
drivers/i2c/busses/i2c-designware-master.c +3 -67

101 lines not on this page · raw message

drivers/i2c/busses/i2c-designware-slave.c +0 -44

56 lines not on this page · raw message

Signature · 1 line
2.50.1
reviewer Reviewed-by on v2 4/6
#10 lore raw

Re: [PATCH v2 4/6] i2c: designware: Combine the init functions

Quoted 4 lines · Heikki Krogerus

Providing a single function for controller initialisation. The controller initialisation has the same steps for master and slave modes, except the timing parameters are only needed in master mode.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#11 lore raw

PATCH v2 5/6 i2c: designware: Enable mode swapping

In reply to Heikki Krogerus

The DesignWare I2C can not be operated as I2C master and I2C slave simultaneously, but that does not actually mean master and slave modes can not be supported at the same time. It just means an explicit mode swap needs to be executed when the mode is changed. The DesignWare I2C documentation actually describes a couple of cases where the mode is excepted to be changed.

The I2C master will now always be supported. Both modes are now always configured in i2c_dw_configure(), but the slave mode will continue to be available only when the Kconfig option I2C_SLAVE is enabled.

The driver will now start in master mode and then swap to slave mode when a slave device is registered. After a slave device is registered, the controller is swapped to master mode when a transfer in master mode is started and then back to slave mode again after the transfer is completed.

The DesignWare I2C can now be used with protocols such as MCTP (drivers/net/mctp/mctp-i2c.c) and IPMI (drivers/char/ipmi/) that require support for both I2C master and I2C slave. It is now also possible to support the SMBus Host Notification Protocol as I2C master if needed.

---
 drivers/i2c/busses/i2c-designware-common.c | 50 +++++++++++++++-------
 drivers/i2c/busses/i2c-designware-core.h   |  9 ++--
 drivers/i2c/busses/i2c-designware-master.c |  6 ++-
 drivers/i2c/busses/i2c-designware-slave.c  | 35 +++++++--------
 4 files changed, 57 insertions(+), 43 deletions(-)
drivers/i2c/busses/i2c-designware-common.c +34 -16
index 8e99549b37a3..aac7b1f4710f 100644
@@ -358,21 +358,25 @@ static inline u32 i2c_dw_acpi_round_bus_speed(struct device *device) { return 0;
 
 #endif	/* CONFIG_ACPI */
 
-static void i2c_dw_configure_mode(struct dw_i2c_dev *dev)
+static void i2c_dw_configure_mode(struct dw_i2c_dev *dev, int mode)
 {
-	switch (dev->mode) {
+	switch (mode) {
 	case DW_IC_MASTER:
 		regmap_write(dev->map, DW_IC_TX_TL, dev->tx_fifo_depth / 2);
 		regmap_write(dev->map, DW_IC_RX_TL, 0);
 		regmap_write(dev->map, DW_IC_CON, dev->master_cfg);
 		break;
 	case DW_IC_SLAVE:
+		dev->status = 0;
 		regmap_write(dev->map, DW_IC_TX_TL, 0);
 		regmap_write(dev->map, DW_IC_RX_TL, 0);
 		regmap_write(dev->map, DW_IC_CON, dev->slave_cfg);
+		regmap_write(dev->map, DW_IC_SAR, dev->slave->addr);
 		regmap_write(dev->map, DW_IC_INTR_MASK, DW_IC_INTR_SLAVE_MASK);
+		__i2c_dw_enable(dev);
 		break;
 	default:
+		WARN(1, "Invalid mode %d\n", mode);
 		return;
 	}
 }
@@ -394,6 +398,31 @@ static void i2c_dw_write_timings(struct dw_i2c_dev *dev)
 	}
 }
 
+/**
+ * i2c_dw_set_mode() - Select the controller mode of operation - master or slave
+ * @dev: device private data
+ * @mode: I2C mode of operation
+ *
+ * Configures the controller to operate in @mode. This function needs to be
+ * called when ever a mode swap is required.
+ *
+ * Setting the slave mode does not have an effect before a slave device is
+ * registered. So before the slave device is registered, the controller is kept
+ * in master mode regardless of @mode.
+ *
+ * The controller must be disabled before this function is called.
+ */
+void i2c_dw_set_mode(struct dw_i2c_dev *dev, int mode)
+{
+	if (mode == DW_IC_SLAVE && !dev->slave)
+		mode = DW_IC_MASTER;
+	if (dev->mode == mode)
+		return;
+
+	i2c_dw_configure_mode(dev, mode);
+	dev->mode = mode;
+}
+
 /**
  * i2c_dw_init() - Initialize the DesignWare I2C hardware
  * @dev: device private data
@@ -420,14 +449,13 @@ int i2c_dw_init(struct dw_i2c_dev *dev)
 	 */
 	regmap_write(dev->map, DW_IC_SMBUS_INTR_MASK, 0);
 
-	if (dev->mode == DW_IC_MASTER)
-		i2c_dw_write_timings(dev);
+	i2c_dw_write_timings(dev);
 
 	/* Write SDA hold time if supported */
 	if (dev->sda_hold_time)
 		regmap_write(dev->map, DW_IC_SDA_HOLD, dev->sda_hold_time);
 
-	i2c_dw_configure_mode(dev);
+	i2c_dw_configure_mode(dev, dev->mode);
 
 	i2c_dw_release_lock(dev);
 
@@ -861,17 +889,7 @@ int i2c_dw_probe(struct dw_i2c_dev *dev)
 	if (ret)
 		return ret;
 
-	switch (dev->mode) {
-	case DW_IC_SLAVE:
-		ret = i2c_dw_probe_slave(dev);
-		break;
-	case DW_IC_MASTER:
-		ret = i2c_dw_probe_master(dev);
-		break;
-	default:
-		ret = -EINVAL;
-		break;
-	}
+	ret = i2c_dw_probe_master(dev);
 	if (ret)
 		return ret;
 
drivers/i2c/busses/i2c-designware-core.h +3 -6
index 82465b134c34..5d783d585406 100644
@@ -387,26 +387,23 @@ int i2c_dw_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs, int num);
 
 #if IS_ENABLED(CONFIG_I2C_SLAVE)
 extern void i2c_dw_configure_slave(struct dw_i2c_dev *dev);
-extern int i2c_dw_probe_slave(struct dw_i2c_dev *dev);
 irqreturn_t i2c_dw_isr_slave(struct dw_i2c_dev *dev);
 int i2c_dw_reg_slave(struct i2c_client *client);
 int i2c_dw_unreg_slave(struct i2c_client *client);
 #else
 static inline void i2c_dw_configure_slave(struct dw_i2c_dev *dev) { }
-static inline int i2c_dw_probe_slave(struct dw_i2c_dev *dev) { return -EINVAL; }
 static inline irqreturn_t i2c_dw_isr_slave(struct dw_i2c_dev *dev) { return IRQ_NONE; }
 #endif
 
 static inline void i2c_dw_configure(struct dw_i2c_dev *dev)
 {
-	if (i2c_detect_slave_mode(dev->dev))
-		i2c_dw_configure_slave(dev);
-	else
-		i2c_dw_configure_master(dev);
+	i2c_dw_configure_slave(dev);
+	i2c_dw_configure_master(dev);
 }
 
 int i2c_dw_probe(struct dw_i2c_dev *dev);
 int i2c_dw_init(struct dw_i2c_dev *dev);
+void i2c_dw_set_mode(struct dw_i2c_dev *dev, int mode);
 
 #if IS_ENABLED(CONFIG_I2C_DESIGNWARE_BAYTRAIL)
 int i2c_dw_baytrail_probe_lock_support(struct dw_i2c_dev *dev);
drivers/i2c/busses/i2c-designware-master.c +5 -1
index 33432bbaec1f..ba2ee526ecc6 100644
@@ -194,6 +194,8 @@ static void i2c_dw_xfer_init(struct dw_i2c_dev *dev)
 	/* Disable the adapter */
 	__i2c_dw_disable(dev);
 
+	i2c_dw_set_mode(dev, DW_IC_MASTER);
+
 	/* If the slave address is ten bit address, enable 10BITADDR */
 	if (msgs[dev->msg_write_idx].flags & I2C_M_TEN) {
 		ic_con = DW_IC_CON_10BITADDR_MASTER;
@@ -831,6 +833,8 @@ i2c_dw_xfer_common(struct dw_i2c_dev *dev, struct i2c_msg msgs[], int num)
 	ret = -EIO;
 
 done:
+	i2c_dw_set_mode(dev, DW_IC_SLAVE);
+
 	i2c_dw_release_lock(dev);
 
 done_nolock:
@@ -853,7 +857,7 @@ void i2c_dw_configure_master(struct dw_i2c_dev *dev)
 {
 	struct i2c_timings *t = &dev->timings;
 
-	dev->functionality = I2C_FUNC_10BIT_ADDR | DW_IC_DEFAULT_FUNCTIONALITY;
+	dev->functionality |= I2C_FUNC_10BIT_ADDR | DW_IC_DEFAULT_FUNCTIONALITY;
 
 	dev->master_cfg = DW_IC_CON_MASTER | DW_IC_CON_SLAVE_DISABLE |
 			  DW_IC_CON_RESTART_EN;
drivers/i2c/busses/i2c-designware-slave.c +15 -20
index 9fc8faa33735..ad0d5fbfa6d5 100644
@@ -24,24 +24,25 @@
 int i2c_dw_reg_slave(struct i2c_client *slave)
 {
 	struct dw_i2c_dev *dev = i2c_get_adapdata(slave->adapter);
+	int ret;
 
+	if (!i2c_check_functionality(slave->adapter, I2C_FUNC_SLAVE))
+		return -EOPNOTSUPP;
 	if (dev->slave)
 		return -EBUSY;
 	if (slave->flags & I2C_CLIENT_TEN)
 		return -EAFNOSUPPORT;
-	pm_runtime_get_sync(dev->dev);
 
-	/*
-	 * Set slave address in the IC_SAR register,
-	 * the address to which the DW_apb_i2c responds.
-	 */
+	ret = i2c_dw_acquire_lock(dev);
+	if (ret)
+		return ret;
+
+	pm_runtime_get_sync(dev->dev);
 	__i2c_dw_disable_nowait(dev);
-	regmap_write(dev->map, DW_IC_SAR, slave->addr);
 	dev->slave = slave;
+	i2c_dw_set_mode(dev, DW_IC_SLAVE);
 
-	__i2c_dw_enable(dev);
-
-	dev->status = 0;
+	i2c_dw_release_lock(dev);
 
 	return 0;
 }
@@ -54,6 +55,7 @@ int i2c_dw_unreg_slave(struct i2c_client *slave)
 	i2c_dw_disable(dev);
 	synchronize_irq(dev->irq);
 	dev->slave = NULL;
+	i2c_dw_set_mode(dev, DW_IC_MASTER);
 	pm_runtime_put_sync_suspend(dev->dev);
 
 	return 0;
@@ -176,23 +178,16 @@ irqreturn_t i2c_dw_isr_slave(struct dw_i2c_dev *dev)
 
 void i2c_dw_configure_slave(struct dw_i2c_dev *dev)
 {
-	dev->functionality = I2C_FUNC_SLAVE;
+	if (dev->flags & ACCESS_POLLING)
+		return;
+
+	dev->functionality |= I2C_FUNC_SLAVE;
 
 	dev->slave_cfg = DW_IC_CON_RX_FIFO_FULL_HLD_CTRL |
 			 DW_IC_CON_RESTART_EN | DW_IC_CON_STOP_DET_IFADDRESSED;
-
-	dev->mode = DW_IC_SLAVE;
 }
 EXPORT_SYMBOL_GPL(i2c_dw_configure_slave);
 
-int i2c_dw_probe_slave(struct dw_i2c_dev *dev)
-{
-	if (dev->flags & ACCESS_POLLING)
-		return -EOPNOTSUPP;
-
-	return 0;
-}
-
 MODULE_AUTHOR("Luis Oliveira <[email protected]>");
 MODULE_DESCRIPTION("Synopsys DesignWare I2C bus slave adapter");
 MODULE_LICENSE("GPL v2");
Signature · 1 line
2.50.1
#12 lore raw

PATCH v2 6/6 i2c: designware: Remove an unnecessary condition

In reply to Heikki Krogerus

Writing also the high speed timing registers unconditionally. The reset value for these registers is 0, so this should always be safe.

---
 drivers/i2c/busses/i2c-designware-common.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
drivers/i2c/busses/i2c-designware-common.c +3 -5
index aac7b1f4710f..1981dbfa260f 100644
@@ -391,11 +391,9 @@ static void i2c_dw_write_timings(struct dw_i2c_dev *dev)
 	regmap_write(dev->map, DW_IC_FS_SCL_HCNT, dev->fs_hcnt);
 	regmap_write(dev->map, DW_IC_FS_SCL_LCNT, dev->fs_lcnt);
 
-	/* Write high speed timing parameters if supported */
-	if (dev->hs_hcnt && dev->hs_lcnt) {
-		regmap_write(dev->map, DW_IC_HS_SCL_HCNT, dev->hs_hcnt);
-		regmap_write(dev->map, DW_IC_HS_SCL_LCNT, dev->hs_lcnt);
-	}
+	/* Write high speed timing parameters */
+	regmap_write(dev->map, DW_IC_HS_SCL_HCNT, dev->hs_hcnt);
+	regmap_write(dev->map, DW_IC_HS_SCL_LCNT, dev->hs_lcnt);
 }
 
 /**
Signature · 1 line
2.50.1
reviewer Reviewed-by on v2 6/6
#13 lore raw

Re: [PATCH v2 6/6] i2c: designware: Remove an unnecessary condition

Quoted 3 lines · Heikki Krogerus

Writing also the high speed timing registers unconditionally. The reset value for these registers is 0, so this should always be safe.

Signature · 2 lines
With Best Regards,
Andy Shevchenko
maintainer Acked-by on the v2 series
#14 lore raw
In reply to Heikki Krogerus
Quoted 23 lines · Heikki Krogerus

Hi,

Changed since v1:
- The init cb is replaced with i2c_dw_init() also in i2c-designware-amdisp.c
- Checking I2C_FUNC_SLAVE in i2c_dw_reg_slave()

The original cover letter:

It's currently not possible to support MCTP or any other protocol that requires support for both modes with the DesignWare I2C because the driver can only be used in one mode. I'm assuming that the driver was designed this way because the DesignWare I2C can not be operated as I2C master and I2C slave simultaneously, however, that does not actually mean both modes could not be supported at the same time. See the patch 5/6 for more detailed explanation.

This series will enable support for both modes in the driver by utilising a simple mode swap method, and that way make it possible to support MCTP, IPMI, SMBus Host Notification Protocol, and any other protocol requires the support for both modes with the DesignWare I2C.

thanks,

Quoted 7 lines

Heikki Krogerus (6):
i2c: designware: Remove useless driver specific option for I2C target
i2c: designware: Remove unnecessary function exports
i2c: designware: Combine some of the common functions
i2c: designware: Combine the init functions
i2c: designware: Enable mode swapping
i2c: designware: Remove an unnecessary condition

For the whole series,

maintainer
#15 lore raw
In reply to Heikki Krogerus

Hi Heikki,

Quoted 3 lines

Heikki Krogerus (6):
i2c: designware: Remove useless driver specific option for I2C target
i2c: designware: Remove unnecessary function exports

to unburden you a little, I applied the first two patches.

Andi