Skip to contentSkip to conversation

i2c-octeon and i2c-thunderx drivers

PATCH Applied Jan Glauber 4 participants 38 messages Apr 25, 2016 – Apr 27, 2016

Subsystems (3)

v7 of 11 Applied Latest is v11 →

lore
Share & tools
lore
#0 lore raw

v7 cover i2c-octeon and i2c-thunderx drivers

Hi Wolfram,

v7 implements the changes from your review plus some comments from David. I've moved the flush writeq patch before the HLC patch, because it is already needed there and seems quite trivial and I've added a new patch to disable the SMBUS QUICK support as discussed.

This series for the Octeon i2c driver is an attempt to upstream some bug fixes and features that accumulated for some time.

On top of the Octeon changes a i2c driver for the ThunderX SOC is added which uses the same functional block as the Octeon driver.

Patches are on top of next-20160422 and were tested on OCTEON, OCTEON-78 and ThunderX.

Changes to v6:
- Fixed read_int kerneldoc
- Removed udelay after write-int in recovery
- Killed retries in recovery, use EAGAIN
- Disable SMBUS QUICK and remove unneeded length check
- Spell out enable/disable
- Switch to wait_event_timeout
- Removed superfluous status check in HLC write
- Optimize wait-queue also for HLC
- Use readq/writeq instead of __raw_* in some places
- Add STAT_IDLE to status check (valid after a write)

Changes to v5:
- Switch to i2c recovery framework
- Clean-up register access, introduce new helper functions
- Fixed ready bit check in combined write
- Fixed IFLG clear in hlc_enable
- Removed complicated last phase logic, not needed when we send START for every message part

Changes to v4:
- Splitted the High-Level Controller patch into several patches
- Reworded some commit messages

Changes to v3:
- Added more functionality flags for SMBUS
- Removed both module parameters
- Make xfer return also other errors than EGAIN
- Return EPROTO on invalid SMBUS block length
- Use devm_ioremap_resource
- Added rename-only patch
- Removed kerneldoc patch from series
- Improved defines

Changes to v2:
- Split clenaup patch into several patches
- Strictly moved functional changes to later patches
- Fixed do-while checkpatch errors
- Moved defines to the patches that use them
- Use BIT_ULL macro
- Split ThunderX patch into 2 patches

Changes to v1:
- Fixed compile error on x86_64
- Disabled thunderx driver on MIPS
- Re-ordered some thunderx probe functions for readability
- Fix missing of_irq.h and i2c-smbus.h includes
- Use IS_ENABLED for CONFIG options

Jan

-------------------------------------------------

David Daney (3):
i2c: octeon: Enable High-Level Controller
i2c: octeon: Add support for cn78xx chips
i2c: octeon: Add workaround for broken irqs on CN3860

Jan Glauber (10):
i2c: octeon: Improve error status checking
i2c: octeon: Use i2c recovery framework
i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support
dt-bindings: i2c: Add Octeon cn78xx TWSI
i2c: octeon: Move read function before write
i2c: octeon: Rename driver to prepare for split
i2c: octeon: Split the driver into two parts
i2c: thunderx: Add i2c driver for ThunderX SOC
i2c: octeon,thunderx: Move register offsets to struct
i2c: thunderx: Add smbus alert support

Peter Swain (2):
i2c: octeon: Add flush writeq helper function
i2c: octeon: Improve performance if interrupt is early

 .../devicetree/bindings/i2c/i2c-octeon.txt         |   6 +
 drivers/i2c/busses/Kconfig                         |  10 +
 drivers/i2c/busses/Makefile                        |   3 +
 drivers/i2c/busses/i2c-cavium.c                    | 799 +++++++++++++++++++++
 drivers/i2c/busses/i2c-cavium.h                    | 214 ++++++
 drivers/i2c/busses/i2c-octeon-core.c               | 288 ++++++++
 drivers/i2c/busses/i2c-octeon.c                    | 606 ----------------
 drivers/i2c/busses/i2c-thunderx-core.c             | 306 ++++++++
 8 files changed, 1626 insertions(+), 606 deletions(-)
 create mode 100644 drivers/i2c/busses/i2c-cavium.c
 create mode 100644 drivers/i2c/busses/i2c-cavium.h
 create mode 100644 drivers/i2c/busses/i2c-octeon-core.c
 delete mode 100644 drivers/i2c/busses/i2c-octeon.c
 create mode 100644 drivers/i2c/busses/i2c-thunderx-core.c
Signature · 1 line
1.9.1
#1 lore raw

PATCH v7 1/15 i2c: octeon: Improve error status checking

Introduce a function that checks for valid status codes depending on the phase of a transmit or receive. Also add all existing status codes and improve error handling for various states.

The Octeon TWSI has an "assert acknowledge" bit (TWSI_CTL_AAK) that is required to be set in master receive mode until the last byte is requested. The state check needs to consider if this bit was set.

---
 drivers/i2c/busses/i2c-octeon.c | 129 +++++++++++++++++++++++++++++++++-------
 1 file changed, 106 insertions(+), 23 deletions(-)
drivers/i2c/busses/i2c-octeon.c +106 -23
index 4275007..471d06e 100644
@@ -55,13 +55,35 @@
 #define TWSI_CTL_IFLG		0x08	/* HW event, SW writes 0 to ACK */
 #define TWSI_CTL_AAK		0x04	/* Assert ACK */
 
-/* Some status values */
+/* Status values */
+#define STAT_ERROR		0x00
 #define STAT_START		0x08
-#define STAT_RSTART		0x10
+#define STAT_REP_START		0x10
 #define STAT_TXADDR_ACK		0x18
+#define STAT_TXADDR_NAK		0x20
 #define STAT_TXDATA_ACK		0x28
+#define STAT_TXDATA_NAK		0x30
+#define STAT_LOST_ARB_38	0x38
 #define STAT_RXADDR_ACK		0x40
+#define STAT_RXADDR_NAK		0x48
 #define STAT_RXDATA_ACK		0x50
+#define STAT_RXDATA_NAK		0x58
+#define STAT_SLAVE_60		0x60
+#define STAT_LOST_ARB_68	0x68
+#define STAT_SLAVE_70		0x70
+#define STAT_LOST_ARB_78	0x78
+#define STAT_SLAVE_80		0x80
+#define STAT_SLAVE_88		0x88
+#define STAT_GENDATA_ACK	0x90
+#define STAT_GENDATA_NAK	0x98
+#define STAT_SLAVE_A0		0xA0
+#define STAT_SLAVE_A8		0xA8
+#define STAT_LOST_ARB_B0	0xB0
+#define STAT_SLAVE_LOST		0xB8
+#define STAT_SLAVE_NAK		0xC0
+#define STAT_SLAVE_ACK		0xC8
+#define STAT_AD2W_ACK		0xD0
+#define STAT_AD2W_NAK		0xD8
 #define STAT_IDLE		0xF8
 
 /* TWSI_INT values */
@@ -225,6 +247,67 @@ static int octeon_i2c_wait(struct octeon_i2c *i2c)
 	return 0;
 }
 
+static int octeon_i2c_check_status(struct octeon_i2c *i2c, int final_read)
+{
+	u8 stat = octeon_i2c_stat_read(i2c);
+
+	switch (stat) {
+	/* Everything is fine */
+	case STAT_IDLE:
+	case STAT_AD2W_ACK:
+	case STAT_RXADDR_ACK:
+	case STAT_TXADDR_ACK:
+	case STAT_TXDATA_ACK:
+		return 0;
+
+	/* ACK allowed on pre-terminal bytes only */
+	case STAT_RXDATA_ACK:
+		if (!final_read)
+			return 0;
+		return -EIO;
+
+	/* NAK allowed on terminal byte only */
+	case STAT_RXDATA_NAK:
+		if (final_read)
+			return 0;
+		return -EIO;
+
+	/* Arbitration lost */
+	case STAT_LOST_ARB_38:
+	case STAT_LOST_ARB_68:
+	case STAT_LOST_ARB_78:
+	case STAT_LOST_ARB_B0:
+		return -EAGAIN;
+
+	/* Being addressed as slave, should back off & listen */
+	case STAT_SLAVE_60:
+	case STAT_SLAVE_70:
+	case STAT_GENDATA_ACK:
+	case STAT_GENDATA_NAK:
+		return -EOPNOTSUPP;
+
+	/* Core busy as slave */
+	case STAT_SLAVE_80:
+	case STAT_SLAVE_88:
+	case STAT_SLAVE_A0:
+	case STAT_SLAVE_A8:
+	case STAT_SLAVE_LOST:
+	case STAT_SLAVE_NAK:
+	case STAT_SLAVE_ACK:
+		return -EOPNOTSUPP;
+
+	case STAT_TXDATA_NAK:
+		return -EIO;
+	case STAT_TXADDR_NAK:
+	case STAT_RXADDR_NAK:
+	case STAT_AD2W_NAK:
+		return -ENXIO;
+	default:
+		dev_err(i2c->dev, "unhandled state: %d\n", stat);
+		return -EIO;
+	}
+}
+
 /* calculate and set clock divisors */
 static void octeon_i2c_set_clock(struct octeon_i2c *i2c)
 {
@@ -318,7 +401,7 @@ static int octeon_i2c_start(struct octeon_i2c *i2c)
 	}
 
 	data = octeon_i2c_stat_read(i2c);
-	if ((data != STAT_START) && (data != STAT_RSTART)) {
+	if ((data != STAT_START) && (data != STAT_REP_START)) {
 		dev_err(i2c->dev, "%s: bad status (0x%x)\n", __func__, data);
 		return -EIO;
 	}
@@ -347,7 +430,6 @@ static int octeon_i2c_write(struct octeon_i2c *i2c, int target,
 			    const u8 *data, int length)
 {
 	int i, result;
-	u8 tmp;
 
 	result = octeon_i2c_start(i2c);
 	if (result)
@@ -361,14 +443,9 @@ static int octeon_i2c_write(struct octeon_i2c *i2c, int target,
 		return result;
 
 	for (i = 0; i < length; i++) {
-		tmp = octeon_i2c_stat_read(i2c);
-
-		if ((tmp != STAT_TXADDR_ACK) && (tmp != STAT_TXDATA_ACK)) {
-			dev_err(i2c->dev,
-				"%s: bad status before write (0x%x)\n",
-				__func__, tmp);
-			return -EIO;
-		}
+		result = octeon_i2c_check_status(i2c, false);
+		if (result)
+			return result;
 
 		octeon_i2c_data_write(i2c, data[i]);
 		octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
@@ -397,7 +474,7 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 			   u8 *data, u16 *rlength, bool recv_len)
 {
 	int i, result, length = *rlength;
-	u8 tmp;
+	bool final_read = false;
 
 	if (length < 1)
 		return -EINVAL;
@@ -413,19 +490,21 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 	if (result)
 		return result;
 
+	/* address OK ? */
+	result = octeon_i2c_check_status(i2c, false);
+	if (result)
+		return result;
+
 	for (i = 0; i < length; i++) {
-		tmp = octeon_i2c_stat_read(i2c);
-		if ((tmp != STAT_RXDATA_ACK) && (tmp != STAT_RXADDR_ACK)) {
-			dev_err(i2c->dev,
-				"%s: bad status before read (0x%x)\n",
-				__func__, tmp);
-			return -EIO;
-		}
+		/* for the last byte TWSI_CTL_AAK must not be set */
+		if (i + 1 == length)
+			final_read = true;
 
-		if (i + 1 < length)
-			octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB | TWSI_CTL_AAK);
-		else
+		/* clear iflg to allow next event */
+		if (final_read)
 			octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
+		else
+			octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB | TWSI_CTL_AAK);
 
 		result = octeon_i2c_wait(i2c);
 		if (result)
@@ -441,6 +520,10 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 			}
 			length += data[i];
 		}
+
+		result = octeon_i2c_check_status(i2c, final_read);
+		if (result)
+			return result;
 	}
 	*rlength = length;
 	return 0;
Signature · 1 line
1.9.1
Applied
#2 lore raw

Re: [PATCH v7 01/15] i2c: octeon: Improve error status checking

Quoted 9 lines · Jan Glauber

Introduce a function that checks for valid status codes depending on the phase of a transmit or receive. Also add all existing status codes and improve error handling for various states.

The Octeon TWSI has an "assert acknowledge" bit (TWSI_CTL_AAK) that is required to be set in master receive mode until the last byte is requested. The state check needs to consider if this bit was set.

Applied to for-next, thanks!

#3 lore raw

PATCH v7 2/15 i2c: octeon: Use i2c recovery framework

In reply to Jan Glauber

Switch to the i2c bus recovery framework using generic SCL recovery. If this fails try to reset the hardware. The recovery is triggered during START on timeout of the interrupt or failure to reach the START / repeated-START condition.

The START function is moved to xfer and while at it remove the xfer debug message (i2c core already provides a debug message for this).

---
 drivers/i2c/busses/i2c-octeon.c | 160 ++++++++++++++++++++++++----------------
 1 file changed, 97 insertions(+), 63 deletions(-)
drivers/i2c/busses/i2c-octeon.c +97 -63
index 471d06e..0f536a1 100644
@@ -90,6 +90,8 @@
 #define TWSI_INT_CORE_EN	BIT_ULL(6)
 #define TWSI_INT_SDA_OVR	BIT_ULL(8)
 #define TWSI_INT_SCL_OVR	BIT_ULL(9)
+#define TWSI_INT_SDA		BIT_ULL(10)
+#define TWSI_INT_SCL		BIT_ULL(11)
 
 struct octeon_i2c {
 	wait_queue_head_t queue;
@@ -152,6 +154,18 @@ static u8 octeon_i2c_reg_read(struct octeon_i2c *i2c, u64 eop_reg)
 #define octeon_i2c_stat_read(i2c)					\
 	octeon_i2c_reg_read(i2c, SW_TWSI_EOP_TWSI_STAT)
 
+
+/**
+ * octeon_i2c_read_int - read the TWSI_INT register
+ * @i2c: The struct octeon_i2c
+ *
+ * Returns the value of the register.
+ */
+static u64 octeon_i2c_read_int(struct octeon_i2c *i2c)
+{
+	return __raw_readq(i2c->twsi_base + TWSI_INT);
+}
+
 /**
  * octeon_i2c_write_int - write the TWSI_INT register
  * @i2c: The struct octeon_i2c
@@ -182,33 +196,6 @@ static void octeon_i2c_int_disable(struct octeon_i2c *i2c)
 	octeon_i2c_write_int(i2c, 0);
 }
 
-/**
- * octeon_i2c_unblock - unblock the bus
- * @i2c: The struct octeon_i2c
- *
- * If there was a reset while a device was driving 0 to bus, bus is blocked.
- * We toggle it free manually by some clock cycles and send a stop.
- */
-static void octeon_i2c_unblock(struct octeon_i2c *i2c)
-{
-	int i;
-
-	dev_dbg(i2c->dev, "%s\n", __func__);
-
-	for (i = 0; i < 9; i++) {
-		octeon_i2c_write_int(i2c, 0);
-		udelay(5);
-		octeon_i2c_write_int(i2c, TWSI_INT_SCL_OVR);
-		udelay(5);
-	}
-	/* hand-crank a STOP */
-	octeon_i2c_write_int(i2c, TWSI_INT_SDA_OVR | TWSI_INT_SCL_OVR);
-	udelay(5);
-	octeon_i2c_write_int(i2c, TWSI_INT_SDA_OVR);
-	udelay(5);
-	octeon_i2c_write_int(i2c, 0);
-}
-
 /* interrupt service routine */
 static irqreturn_t octeon_i2c_isr(int irq, void *dev_id)
 {
@@ -371,6 +358,17 @@ static int octeon_i2c_init_lowlevel(struct octeon_i2c *i2c)
 	return -EIO;
 }
 
+static int octeon_i2c_recovery(struct octeon_i2c *i2c)
+{
+	int ret;
+
+	ret = i2c_recover_bus(&i2c->adap);
+	if (ret)
+		/* recover failed, try hardware re-init */
+		ret = octeon_i2c_init_lowlevel(i2c);
+	return ret;
+}
+
 /**
  * octeon_i2c_start - send START to the bus
  * @i2c: The struct octeon_i2c
@@ -379,34 +377,23 @@ static int octeon_i2c_init_lowlevel(struct octeon_i2c *i2c)
  */
 static int octeon_i2c_start(struct octeon_i2c *i2c)
 {
-	int result;
-	u8 data;
+	int ret;
+	u8 stat;
 
 	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB | TWSI_CTL_STA);
+	ret = octeon_i2c_wait(i2c);
+	if (ret)
+		goto error;
 
-	result = octeon_i2c_wait(i2c);
-	if (result) {
-		if (octeon_i2c_stat_read(i2c) == STAT_IDLE) {
-			/*
-			 * Controller refused to send start flag May
-			 * be a client is holding SDA low - let's try
-			 * to free it.
-			 */
-			octeon_i2c_unblock(i2c);
-			octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB | TWSI_CTL_STA);
-			result = octeon_i2c_wait(i2c);
-		}
-		if (result)
-			return result;
-	}
-
-	data = octeon_i2c_stat_read(i2c);
-	if ((data != STAT_START) && (data != STAT_REP_START)) {
-		dev_err(i2c->dev, "%s: bad status (0x%x)\n", __func__, data);
-		return -EIO;
-	}
+	stat = octeon_i2c_stat_read(i2c);
+	if (stat == STAT_START || stat == STAT_REP_START)
+		/* START successful, bail out */
+		return 0;
 
-	return 0;
+error:
+	/* START failed, try to recover */
+	ret = octeon_i2c_recovery(i2c);
+	return (ret) ? ret : -EAGAIN;
 }
 
 /* send STOP to the bus */
@@ -431,10 +418,6 @@ static int octeon_i2c_write(struct octeon_i2c *i2c, int target,
 {
 	int i, result;
 
-	result = octeon_i2c_start(i2c);
-	if (result)
-		return result;
-
 	octeon_i2c_data_write(i2c, target << 1);
 	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
 
@@ -479,10 +462,6 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 	if (length < 1)
 		return -EINVAL;
 
-	result = octeon_i2c_start(i2c);
-	if (result)
-		return result;
-
 	octeon_i2c_data_write(i2c, (target << 1) | 1);
 	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
 
@@ -546,10 +525,10 @@ static int octeon_i2c_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs,
 	for (i = 0; ret == 0 && i < num; i++) {
 		struct i2c_msg *pmsg = &msgs[i];
 
-		dev_dbg(i2c->dev,
-			"Doing %s %d byte(s) to/from 0x%02x - %d of %d messages\n",
-			 pmsg->flags & I2C_M_RD ? "read" : "write",
-			 pmsg->len, pmsg->addr, i + 1, num);
+		ret = octeon_i2c_start(i2c);
+		if (ret)
+			return ret;
+
 		if (pmsg->flags & I2C_M_RD)
 			ret = octeon_i2c_read(i2c, pmsg->addr, pmsg->buf,
 					      &pmsg->len, pmsg->flags & I2C_M_RECV_LEN);
@@ -562,6 +541,60 @@ static int octeon_i2c_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs,
 	return (ret != 0) ? ret : num;
 }
 
+static int octeon_i2c_get_scl(struct i2c_adapter *adap)
+{
+	struct octeon_i2c *i2c = i2c_get_adapdata(adap);
+	u64 state;
+
+	state = octeon_i2c_read_int(i2c);
+	return state & TWSI_INT_SCL;
+}
+
+static void octeon_i2c_set_scl(struct i2c_adapter *adap, int val)
+{
+	struct octeon_i2c *i2c = i2c_get_adapdata(adap);
+
+	octeon_i2c_write_int(i2c, TWSI_INT_SCL_OVR);
+}
+
+static int octeon_i2c_get_sda(struct i2c_adapter *adap)
+{
+	struct octeon_i2c *i2c = i2c_get_adapdata(adap);
+	u64 state;
+
+	state = octeon_i2c_read_int(i2c);
+	return state & TWSI_INT_SDA;
+}
+
+static void octeon_i2c_prepare_recovery(struct i2c_adapter *adap)
+{
+	struct octeon_i2c *i2c = i2c_get_adapdata(adap);
+
+	/*
+	 * The stop resets the state machine, does not _transmit_ STOP unless
+	 * engine was active.
+	 */
+	octeon_i2c_stop(i2c);
+
+	octeon_i2c_write_int(i2c, 0);
+}
+
+static void octeon_i2c_unprepare_recovery(struct i2c_adapter *adap)
+{
+	struct octeon_i2c *i2c = i2c_get_adapdata(adap);
+
+	octeon_i2c_write_int(i2c, 0);
+}
+
+static struct i2c_bus_recovery_info octeon_i2c_recovery_info = {
+	.recover_bus = i2c_generic_scl_recovery,
+	.get_scl = octeon_i2c_get_scl,
+	.set_scl = octeon_i2c_set_scl,
+	.get_sda = octeon_i2c_get_sda,
+	.prepare_recovery = octeon_i2c_prepare_recovery,
+	.unprepare_recovery = octeon_i2c_unprepare_recovery,
+};
+
 static u32 octeon_i2c_functionality(struct i2c_adapter *adap)
 {
 	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL |
@@ -642,6 +675,7 @@ static int octeon_i2c_probe(struct platform_device *pdev)
 	i2c->adap = octeon_i2c_ops;
 	i2c->adap.timeout = msecs_to_jiffies(2);
 	i2c->adap.retries = 5;
+	i2c->adap.bus_recovery_info = &octeon_i2c_recovery_info;
 	i2c->adap.dev.parent = &pdev->dev;
 	i2c->adap.dev.of_node = node;
 	i2c_set_adapdata(&i2c->adap, i2c);
Signature · 1 line
1.9.1
Applied
#4 lore raw

Re: [PATCH v7 02/15] i2c: octeon: Use i2c recovery framework

Quoted 10 lines · Jan Glauber

Switch to the i2c bus recovery framework using generic SCL recovery. If this fails try to reset the hardware. The recovery is triggered during START on timeout of the interrupt or failure to reach the START / repeated-START condition.

The START function is moved to xfer and while at it remove the xfer debug message (i2c core already provides a debug message for this).

Applied to for-next, thanks!

Fixed this checkpatch warning:

CHECK: Please don't use multiple blank lines
#32: FILE: drivers/i2c/busses/i2c-octeon.c:157:

drivers/i2c/busses/i2c-octeon.c static int octeon_i2c_init_lowlevel(struct octeon_i2c *i2c) · line 396

Quoted 1 line
drivers/i2c/busses/i2c-octeon.c +1 -0
+	return (ret) ? ret : -EAGAIN;

I'd write 'return ret ?: -EAGAIN' but this is a 'mileage varies' thing.

#5 lore raw

PATCH v7 3/15 i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support

In reply to Jan Glauber

SMBUS QUICK never worked for the read case, because EINVAL was returned for a zero length message. The hardware does not support SMBUS QUICK messages so disable the support and remove the zero length check.

---
 drivers/i2c/busses/i2c-octeon.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
drivers/i2c/busses/i2c-octeon.c +1 -4
index 0f536a1..ad563cf 100644
@@ -459,9 +459,6 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 	int i, result, length = *rlength;
 	bool final_read = false;
 
-	if (length < 1)
-		return -EINVAL;
-
 	octeon_i2c_data_write(i2c, (target << 1) | 1);
 	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
 
@@ -597,7 +594,7 @@ static struct i2c_bus_recovery_info octeon_i2c_recovery_info = {
 
 static u32 octeon_i2c_functionality(struct i2c_adapter *adap)
 {
-	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL |
+	return I2C_FUNC_I2C | (I2C_FUNC_SMBUS_EMUL & ~I2C_FUNC_SMBUS_QUICK) |
 	       I2C_FUNC_SMBUS_READ_BLOCK_DATA | I2C_SMBUS_BLOCK_PROC_CALL;
 }
 
Signature · 1 line
1.9.1
#6 lore raw

Re: [PATCH v7 03/15] i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support

Quoted 5 lines · Jan Glauber

SMBUS QUICK never worked for the read case, because EINVAL was returned for a zero length message. The hardware does not support SMBUS QUICK messages so disable the support and remove the zero length check.

I see better now and I think we need to drop this patch. It looks like the driver supports only SMBUS_QUICK_WRITE which can be used for scanning devices. The driver probably works with 'i2cdetect -q'. This will regress if I apply the patch.

It seems it can't do SMBUS_QUICK_READ thus the extra check in the code.

I2C_FUNC_SMBUS_QUICK should have been split up in READ and WRITE to support this scenario.

Are my assumptions correct?

Quoted 30 lines
---
 drivers/i2c/busses/i2c-octeon.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
drivers/i2c/busses/i2c-octeon.c +1 -4
index 0f536a1..ad563cf 100644
@@ -459,9 +459,6 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 	int i, result, length = *rlength;
 	bool final_read = false;
 
-	if (length < 1)
-		return -EINVAL;
-
 	octeon_i2c_data_write(i2c, (target << 1) | 1);
 	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
 
@@ -597,7 +594,7 @@ static struct i2c_bus_recovery_info octeon_i2c_recovery_info = {
 
 static u32 octeon_i2c_functionality(struct i2c_adapter *adap)
 {
-	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL |
+	return I2C_FUNC_I2C | (I2C_FUNC_SMBUS_EMUL & ~I2C_FUNC_SMBUS_QUICK) |
 	       I2C_FUNC_SMBUS_READ_BLOCK_DATA | I2C_SMBUS_BLOCK_PROC_CALL;
 }
 
Signature · 1 line
1.9.1
#7 lore raw

Re: [PATCH v7 03/15] i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support

Quoted 12 lines · Wolfram Sang
Quoted 5 lines · Jan Glauber

SMBUS QUICK never worked for the read case, because EINVAL was returned for a zero length message. The hardware does not support SMBUS QUICK messages so disable the support and remove the zero length check.

I see better now and I think we need to drop this patch. It looks like the driver supports only SMBUS_QUICK_WRITE which can be used for scanning devices. The driver probably works with 'i2cdetect -q'. This will regress if I apply the patch.

Quoted 6 lines

It seems it can't do SMBUS_QUICK_READ thus the extra check in the code.

I2C_FUNC_SMBUS_QUICK should have been split up in READ and WRITE to support this scenario.

Are my assumptions correct?

I think you may be reading too much into the intent of the code.

I don't know that anyone has ever run "i2cdetect -q" on these buses, I know that I haven't. Since I2C is inherently unprobable, we specify the presence of all devices from the device tree passed by system firmware.
Runtime scanning for devices is not necessary.

That said, it is not a big deal one way or the other with respect to this patch. If someone verifies that "i2cdetect -q" does something sensible, then there is probably no harm in dropping this patch.

Quoted 32 lines
Quoted 32 lines

---
  drivers/i2c/busses/i2c-octeon.c | 5 +----
  1 file changed, 1 insertion(+), 4 deletions(-)
drivers/i2c/busses/i2c-octeon.c +1 -5
index 0f536a1..ad563cf 100644
@@ -459,9 +459,6 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
  	int i, result, length = *rlength;
  	bool final_read = false;
 
-	if (length < 1)
-		return -EINVAL;
-
  	octeon_i2c_data_write(i2c, (target << 1) | 1);
  	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
 
@@ -597,7 +594,7 @@ static struct i2c_bus_recovery_info octeon_i2c_recovery_info = {
 
  static u32 octeon_i2c_functionality(struct i2c_adapter *adap)
  {
-	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL |
+	return I2C_FUNC_I2C | (I2C_FUNC_SMBUS_EMUL & ~I2C_FUNC_SMBUS_QUICK) |
  	       I2C_FUNC_SMBUS_READ_BLOCK_DATA | I2C_SMBUS_BLOCK_PROC_CALL;
  }
 
--

1.9.1

#8 lore raw

Re: [PATCH v7 03/15] i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support

In reply to Wolfram Sang
Quoted 12 lines · Wolfram Sang
Quoted 5 lines · Jan Glauber

SMBUS QUICK never worked for the read case, because EINVAL was returned for a zero length message. The hardware does not support SMBUS QUICK messages so disable the support and remove the zero length check.

I see better now and I think we need to drop this patch. It looks like the driver supports only SMBUS_QUICK_WRITE which can be used for scanning devices. The driver probably works with 'i2cdetect -q'. This will regress if I apply the patch.

Quoted 6 lines

It seems it can't do SMBUS_QUICK_READ thus the extra check in the code.

I2C_FUNC_SMBUS_QUICK should have been split up in READ and WRITE to support this scenario.

Are my assumptions correct?

Yes, I thought briefly about splitting SMBUS_QUICK into read-write variants too. To me the question is if this feature is still used on modern devices or if this is more a relict of the past. I don't know enough about SMBUS to answer that.

Also, the ThunderX documentation does not mention that it is supported. The read case would be forbidden by the documentation (that was probably the reason for adding the read EINVAL). The write case is kind of a grey area, it might or might not work.

I'll see what i2cdetect does.

Thanks,
Jan

Quoted 31 lines
Quoted 31 lines
---
 drivers/i2c/busses/i2c-octeon.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
drivers/i2c/busses/i2c-octeon.c +1 -4
index 0f536a1..ad563cf 100644
@@ -459,9 +459,6 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 	int i, result, length = *rlength;
 	bool final_read = false;
 
-	if (length < 1)
-		return -EINVAL;
-
 	octeon_i2c_data_write(i2c, (target << 1) | 1);
 	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
 
@@ -597,7 +594,7 @@ static struct i2c_bus_recovery_info octeon_i2c_recovery_info = {
 
 static u32 octeon_i2c_functionality(struct i2c_adapter *adap)
 {
-	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL |
+	return I2C_FUNC_I2C | (I2C_FUNC_SMBUS_EMUL & ~I2C_FUNC_SMBUS_QUICK) |
 	       I2C_FUNC_SMBUS_READ_BLOCK_DATA | I2C_SMBUS_BLOCK_PROC_CALL;
 }
 
Signature · 1 line
1.9.1
#9 lore raw

Re: [PATCH v7 03/15] i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support

Quoted 25 lines · Jan Glauber
Quoted 12 lines · Wolfram Sang
Quoted 5 lines · Jan Glauber

SMBUS QUICK never worked for the read case, because EINVAL was returned for a zero length message. The hardware does not support SMBUS QUICK messages so disable the support and remove the zero length check.

I see better now and I think we need to drop this patch. It looks like the driver supports only SMBUS_QUICK_WRITE which can be used for scanning devices. The driver probably works with 'i2cdetect -q'. This will regress if I apply the patch.

Quoted 6 lines

It seems it can't do SMBUS_QUICK_READ thus the extra check in the code.

I2C_FUNC_SMBUS_QUICK should have been split up in READ and WRITE to support this scenario.

Are my assumptions correct?

Yes, I thought briefly about splitting SMBUS_QUICK into read-write variants too. To me the question is if this feature is still used on modern devices or if this is more a relict of the past. I don't know enough about SMBUS to answer that.

Quoted 6 lines

Also, the ThunderX documentation does not mention that it is supported. The read case would be forbidden by the documentation (that was probably the reason for adding the read EINVAL). The write case is kind of a grey area, it might or might not work.

I'll see what i2cdetect does.

Checking on ThunderX:

[root@localhost linux-aarch64]# i2cdetect -q 0
WARNING! This program can confuse your I2C bus, cause data loss and worse! I will probe file /dev/i2c-0 using quick write commands.
I will probe address range 0x03-0x77.
Continue? [Y/n]
0 1 2 3 4 5 6 7 8 9 a b c d e f
00: 03 04 05 06 07 08 09 0a 0b 0c 0d 0e 0f
10: 10 11 12 13 14 15 16 17 18 19 1a 1b 1c 1d 1e 1f
20: 20 21 22 23 24 25 26 27 28 29 2a 2b 2c 2d 2e 2f
30: 30 31 32 33 34 35 36 37 38 39 3a 3b 3c 3d 3e 3f
40: 40 41 42 43 44 45 46 47 48 49 4a 4b 4c 4d 4e 4f
50: 50 51 52 53 54 55 56 57 58 59 5a 5b 5c 5d 5e 5f
60: 60 61 62 63 64 65 66 67 UU 69 6a 6b 6c 6d 6e 6f
70: 70 71 72 73 74 75 76 77

Address 68 is the only device (rtc clock, module loaded).

Do all these other numbers make sense (although there are no devices)?

--Jan

#10 lore raw

Re: [PATCH v7 03/15] i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support

Quoted 5 lines
Quoted 5 lines

Yes, I thought briefly about splitting SMBUS_QUICK into read-write variants too. To me the question is if this feature is still used on modern devices or if this is more a relict of the past. I don't know enough about SMBUS to answer that.

Well, note that there are zero-length messages in I2C allowed as well. Not only in SMBUS. I mainly use the term SMBUS_QUICK because it covers both cases.

Quoted 1 line

Checking on ThunderX:

...

Quoted 2 lines

Do all these other numbers make sense (although there are no devices)?

It makes sense in a way that it shows SMBUS_QUICK_WRITE is broken :) It doesn't react to ACK/NACK properly. So, what needs to be done:

1) remove SMBUS_QUICK as you did in this patch 2) move the length check so it doesn't only check read messages but also write messages. That is to prevent handling custom setup I2C messages with a length of 0 (which is legal). I'd suggest to return -EOPNOTSUPP in this case.

Thanks,

Wolfram

#11 lore raw

Re: [PATCH v7 03/15] i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support

Quoted 16 lines · Wolfram Sang
Quoted 5 lines
Quoted 5 lines

Yes, I thought briefly about splitting SMBUS_QUICK into read-write variants too. To me the question is if this feature is still used on modern devices or if this is more a relict of the past. I don't know enough about SMBUS to answer that.

Well, note that there are zero-length messages in I2C allowed as well. Not only in SMBUS. I mainly use the term SMBUS_QUICK because it covers both cases.

Quoted 1 line

Checking on ThunderX:

...

Quoted 2 lines

Do all these other numbers make sense (although there are no devices)?

Quoted 7 lines

It makes sense in a way that it shows SMBUS_QUICK_WRITE is broken :) It doesn't react to ACK/NACK properly. So, what needs to be done:

1) remove SMBUS_QUICK as you did in this patch 2) move the length check so it doesn't only check read messages but also write messages. That is to prevent handling custom setup I2C messages with a length of 0 (which is legal). I'd suggest to return -EOPNOTSUPP in this case.

OK, I'll do that. Should I rebase the remaining patches or would you like to review them first ? :)

Thanks,
Jan

Quoted 4 lines

Thanks,

Wolfram

#12 lore raw

Re: [PATCH v7 03/15] i2c: octeon: Remove I2C_FUNC_SMBUS_QUICK support

Quoted 3 lines

OK, I'll do that. Should I rebase the remaining patches or would you like to review them first ? :)

Just resend this one, please. I probably pick up patches 8+9 for the next cycle, but with the i2c-mux locking rework planned for 4.7, too, I don't think I have the bandwidth to apply the rest for 4.7 as well.

#13 lore raw

PATCH 1/1 i2c: octeon: Remove zero-length message support

Zero-length message support (SMBUS QUICK or i2c) never worked with the Octeon hardware. Disable SMBUS QUICK support and bail out in case of a zero-length i2c request.

After this change 'i2c-detect -q' will return an error on Octeon but the previously reported results were wrong anyway.

---
 drivers/i2c/busses/i2c-octeon.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)
drivers/i2c/busses/i2c-octeon.c +7 -4
index 9f11e19..e8a65f5 100644
@@ -837,9 +837,6 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 	int i, result, length = *rlength;
 	bool final_read = false;
 
-	if (length < 1)
-		return -EINVAL;
-
 	octeon_i2c_data_write(i2c, (target << 1) | 1);
 	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
 
@@ -925,6 +922,12 @@ static int octeon_i2c_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs,
 	for (i = 0; ret == 0 && i < num; i++) {
 		struct i2c_msg *pmsg = &msgs[i];
 
+		/* zero-length messages are not supported */
+		if (!pmsg->len) {
+			ret = -EOPNOTSUPP;
+			break;
+		}
+
 		ret = octeon_i2c_start(i2c);
 		if (ret)
 			return ret;
@@ -998,7 +1001,7 @@ static struct i2c_bus_recovery_info octeon_i2c_recovery_info = {
 
 static u32 octeon_i2c_functionality(struct i2c_adapter *adap)
 {
-	return I2C_FUNC_I2C | I2C_FUNC_SMBUS_EMUL |
+	return I2C_FUNC_I2C | (I2C_FUNC_SMBUS_EMUL & ~I2C_FUNC_SMBUS_QUICK) |
 	       I2C_FUNC_SMBUS_READ_BLOCK_DATA | I2C_SMBUS_BLOCK_PROC_CALL;
 }
 
Signature · 1 line
1.9.1
Applied
#14 lore raw

Re: [PATCH] i2c: octeon: Remove zero-length message support

Quoted 8 lines · Jan Glauber

Zero-length message support (SMBUS QUICK or i2c) never worked with the Octeon hardware. Disable SMBUS QUICK support and bail out in case of a zero-length i2c request.

After this change 'i2c-detect -q' will return an error on Octeon but the previously reported results were wrong anyway.

Applied to for-next, thanks!

#15 lore raw

PATCH v7 4/15 i2c: octeon: Add flush writeq helper function

In reply to Jan Glauber

From: Peter Swain <[email protected]>

Add helper function that reads back a value after writing to make sure the write is finished and use it in octeon_i2c_write_int().

---
 drivers/i2c/busses/i2c-octeon.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/i2c/busses/i2c-octeon.c +7 -2
index ad563cf..60a02b2 100644
@@ -103,6 +103,12 @@ struct octeon_i2c {
 	struct device *dev;
 };
 
+static void octeon_i2c_writeq_flush(u64 val, void __iomem *addr)
+{
+	__raw_writeq(val, addr);
+	__raw_readq(addr);	/* wait for write to land */
+}
+
 /**
  * octeon_i2c_reg_write - write an I2C core register
  * @i2c: The struct octeon_i2c
@@ -173,8 +179,7 @@ static u64 octeon_i2c_read_int(struct octeon_i2c *i2c)
  */
 static void octeon_i2c_write_int(struct octeon_i2c *i2c, u64 data)
 {
-	__raw_writeq(data, i2c->twsi_base + TWSI_INT);
-	__raw_readq(i2c->twsi_base + TWSI_INT);
+	octeon_i2c_writeq_flush(data, i2c->twsi_base + TWSI_INT);
 }
 
 /**
Signature · 1 line
1.9.1
#17 lore raw

PATCH v7 5/15 i2c: octeon: Enable High-Level Controller

In reply to Jan Glauber

From: David Daney <[email protected]>

Use High-Level Controller (HLC) when possible. The HLC can read/write up to 8 bytes and is completely optional. The most important difference of the HLC is that it only requires one interrupt for a transfer (up to 8 bytes) where the low-level read/write requires 2 interrupts plus one interrupt per transferred byte. Since the interrupts are costly using the HLC improves the performance. Also, the HLC provides improved error handling.

---
 drivers/i2c/busses/i2c-octeon.c | 345 ++++++++++++++++++++++++++++++++++++++--
 1 file changed, 335 insertions(+), 10 deletions(-)
drivers/i2c/busses/i2c-octeon.c +335 -10
index 60a02b2..9520f78 100644
@@ -29,13 +29,23 @@
 /* Register offsets */
 #define SW_TWSI			0x00
 #define TWSI_INT		0x10
+#define SW_TWSI_EXT		0x18
 
 /* Controller command patterns */
 #define SW_TWSI_V		BIT_ULL(63)	/* Valid bit */
+#define SW_TWSI_EIA		BIT_ULL(61)	/* Extended internal address */
 #define SW_TWSI_R		BIT_ULL(56)	/* Result or read bit */
+#define SW_TWSI_SOVR		BIT_ULL(55)	/* Size override */
+#define SW_TWSI_SIZE_SHIFT	52
+#define SW_TWSI_ADDR_SHIFT	40
+#define SW_TWSI_IA_SHIFT	32		/* Internal address */
 
 /* Controller opcode word (bits 60:57) */
 #define SW_TWSI_OP_SHIFT	57
+#define SW_TWSI_OP_7		(0ULL << SW_TWSI_OP_SHIFT)
+#define SW_TWSI_OP_7_IA		(1ULL << SW_TWSI_OP_SHIFT)
+#define SW_TWSI_OP_10		(2ULL << SW_TWSI_OP_SHIFT)
+#define SW_TWSI_OP_10_IA	(3ULL << SW_TWSI_OP_SHIFT)
 #define SW_TWSI_OP_TWSI_CLK	(4ULL << SW_TWSI_OP_SHIFT)
 #define SW_TWSI_OP_EOP		(6ULL << SW_TWSI_OP_SHIFT) /* Extended opcode */
 
@@ -48,7 +58,7 @@
 #define SW_TWSI_EOP_TWSI_RST	(SW_TWSI_OP_EOP | 7ULL << SW_TWSI_EOP_SHIFT)
 
 /* Controller command and status bits */
-#define TWSI_CTL_CE		0x80
+#define TWSI_CTL_CE		0x80	/* High level controller enable */
 #define TWSI_CTL_ENAB		0x40	/* Bus enable */
 #define TWSI_CTL_STA		0x20	/* Master-mode start, HW clears when done */
 #define TWSI_CTL_STP		0x10	/* Master-mode stop, HW clears when done */
@@ -87,6 +97,11 @@
 #define STAT_IDLE		0xF8
 
 /* TWSI_INT values */
+#define TWSI_INT_ST_INT		BIT_ULL(0)
+#define TWSI_INT_TS_INT		BIT_ULL(1)
+#define TWSI_INT_CORE_INT	BIT_ULL(2)
+#define TWSI_INT_ST_EN		BIT_ULL(4)
+#define TWSI_INT_TS_EN		BIT_ULL(5)
 #define TWSI_INT_CORE_EN	BIT_ULL(6)
 #define TWSI_INT_SDA_OVR	BIT_ULL(8)
 #define TWSI_INT_SCL_OVR	BIT_ULL(9)
@@ -101,6 +116,7 @@ struct octeon_i2c {
 	int sys_freq;
 	void __iomem *twsi_base;
 	struct device *dev;
+	bool hlc_enabled;
 };
 
 static void octeon_i2c_writeq_flush(u64 val, void __iomem *addr)
@@ -201,6 +217,47 @@ static void octeon_i2c_int_disable(struct octeon_i2c *i2c)
 	octeon_i2c_write_int(i2c, 0);
 }
 
+/*
+ * Cleanup low-level state & enable high-level controller.
+ */
+static void octeon_i2c_hlc_enable(struct octeon_i2c *i2c)
+{
+	int try = 0;
+	u64 val;
+
+	if (i2c->hlc_enabled)
+		return;
+	i2c->hlc_enabled = true;
+
+	while (1) {
+		val = octeon_i2c_ctl_read(i2c);
+		if (!(val & (TWSI_CTL_STA | TWSI_CTL_STP)))
+			break;
+
+		/* clear IFLG event */
+		if (val & TWSI_CTL_IFLG)
+			octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
+
+		if (try++ > 100) {
+			pr_err("%s: giving up\n", __func__);
+			break;
+		}
+
+		/* spin until any start/stop has finished */
+		udelay(10);
+	}
+	octeon_i2c_ctl_write(i2c, TWSI_CTL_CE | TWSI_CTL_AAK | TWSI_CTL_ENAB);
+}
+
+static void octeon_i2c_hlc_disable(struct octeon_i2c *i2c)
+{
+	if (!i2c->hlc_enabled)
+		return;
+
+	i2c->hlc_enabled = false;
+	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
+}
+
 /* interrupt service routine */
 static irqreturn_t octeon_i2c_isr(int irq, void *dev_id)
 {
@@ -300,6 +357,243 @@ static int octeon_i2c_check_status(struct octeon_i2c *i2c, int final_read)
 	}
 }
 
+static bool octeon_i2c_hlc_test_ready(struct octeon_i2c *i2c)
+{
+	u64 val = __raw_readq(i2c->twsi_base + SW_TWSI);
+
+	return (val & SW_TWSI_V) == 0;
+}
+
+static void octeon_i2c_hlc_int_enable(struct octeon_i2c *i2c)
+{
+	octeon_i2c_write_int(i2c, TWSI_INT_ST_EN);
+}
+
+static void octeon_i2c_hlc_int_clear(struct octeon_i2c *i2c)
+{
+	/* clear ST/TS events, listen for neither */
+	octeon_i2c_write_int(i2c, TWSI_INT_ST_INT | TWSI_INT_TS_INT);
+}
+
+/**
+ * octeon_i2c_hlc_wait - wait for an HLC operation to complete
+ * @i2c: The struct octeon_i2c
+ *
+ * Returns 0 on success, otherwise -ETIMEDOUT.
+ */
+static int octeon_i2c_hlc_wait(struct octeon_i2c *i2c)
+{
+	int time_left;
+
+	octeon_i2c_hlc_int_enable(i2c);
+	time_left = wait_event_timeout(i2c->queue,
+				       octeon_i2c_hlc_test_ready(i2c),
+				       i2c->adap.timeout);
+	octeon_i2c_int_disable(i2c);
+	if (!time_left) {
+		octeon_i2c_hlc_int_clear(i2c);
+		return -ETIMEDOUT;
+	}
+	return 0;
+}
+
+/* high-level-controller pure read of up to 8 bytes */
+static int octeon_i2c_hlc_read(struct octeon_i2c *i2c, struct i2c_msg *msgs)
+{
+	int i, j, ret = 0;
+	u64 cmd;
+
+	octeon_i2c_hlc_enable(i2c);
+	octeon_i2c_hlc_int_clear(i2c);
+
+	cmd = SW_TWSI_V | SW_TWSI_R | SW_TWSI_SOVR;
+	/* SIZE */
+	cmd |= (u64) (msgs[0].len - 1) << SW_TWSI_SIZE_SHIFT;
+	/* A */
+	cmd |= (u64) (msgs[0].addr & 0x7full) << SW_TWSI_ADDR_SHIFT;
+
+	if (msgs[0].flags & I2C_M_TEN)
+		cmd |= SW_TWSI_OP_10;
+	else
+		cmd |= SW_TWSI_OP_7;
+
+	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI);
+	ret = octeon_i2c_hlc_wait(i2c);
+	if (ret)
+		goto err;
+
+	cmd = __raw_readq(i2c->twsi_base + SW_TWSI);
+	if ((cmd & SW_TWSI_R) == 0)
+		return -EAGAIN;
+
+	for (i = 0, j = msgs[0].len - 1; i  < msgs[0].len && i < 4; i++, j--)
+		msgs[0].buf[j] = (cmd >> (8 * i)) & 0xff;
+
+	if (msgs[0].len > 4) {
+		cmd = __raw_readq(i2c->twsi_base + SW_TWSI_EXT);
+		for (i = 0; i  < msgs[0].len - 4 && i < 4; i++, j--)
+			msgs[0].buf[j] = (cmd >> (8 * i)) & 0xff;
+	}
+
+err:
+	return ret;
+}
+
+/* high-level-controller pure write of up to 8 bytes */
+static int octeon_i2c_hlc_write(struct octeon_i2c *i2c, struct i2c_msg *msgs)
+{
+	int i, j, ret = 0;
+	u64 cmd;
+
+	octeon_i2c_hlc_enable(i2c);
+	octeon_i2c_hlc_int_clear(i2c);
+
+	cmd = SW_TWSI_V | SW_TWSI_SOVR;
+	/* SIZE */
+	cmd |= (u64) (msgs[0].len - 1) << SW_TWSI_SIZE_SHIFT;
+	/* A */
+	cmd |= (u64) (msgs[0].addr & 0x7full) << SW_TWSI_ADDR_SHIFT;
+
+	if (msgs[0].flags & I2C_M_TEN)
+		cmd |= SW_TWSI_OP_10;
+	else
+		cmd |= SW_TWSI_OP_7;
+
+	for (i = 0, j = msgs[0].len - 1; i  < msgs[0].len && i < 4; i++, j--)
+		cmd |= (u64)msgs[0].buf[j] << (8 * i);
+
+	if (msgs[0].len > 4) {
+		u64 ext = 0;
+
+		for (i = 0; i < msgs[0].len - 4 && i < 4; i++, j--)
+			ext |= (u64) msgs[0].buf[j] << (8 * i);
+		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT);
+	}
+
+	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI);
+	ret = octeon_i2c_hlc_wait(i2c);
+	if (ret)
+		goto err;
+
+	cmd = __raw_readq(i2c->twsi_base + SW_TWSI);
+	if ((cmd & SW_TWSI_R) == 0)
+		return -EAGAIN;
+
+	ret = octeon_i2c_check_status(i2c, false);
+
+err:
+	return ret;
+}
+
+/* high-level-controller composite write+read, msg0=addr, msg1=data */
+static int octeon_i2c_hlc_comp_read(struct octeon_i2c *i2c, struct i2c_msg *msgs)
+{
+	int i, j, ret = 0;
+	u64 cmd;
+
+	octeon_i2c_hlc_enable(i2c);
+
+	cmd = SW_TWSI_V | SW_TWSI_R | SW_TWSI_SOVR;
+	/* SIZE */
+	cmd |= (u64)(msgs[1].len - 1) << SW_TWSI_SIZE_SHIFT;
+	/* A */
+	cmd |= (u64)(msgs[0].addr & 0x7full) << SW_TWSI_ADDR_SHIFT;
+
+	if (msgs[0].flags & I2C_M_TEN)
+		cmd |= SW_TWSI_OP_10_IA;
+	else
+		cmd |= SW_TWSI_OP_7_IA;
+
+	if (msgs[0].len == 2) {
+		u64 ext = 0;
+
+		cmd |= SW_TWSI_EIA;
+		ext = (u64) msgs[0].buf[0] << SW_TWSI_IA_SHIFT;
+		cmd |= (u64) msgs[0].buf[1] << SW_TWSI_IA_SHIFT;
+		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT);
+	} else
+		cmd |= (u64) msgs[0].buf[0] << SW_TWSI_IA_SHIFT;
+
+	octeon_i2c_hlc_int_clear(i2c);
+	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI);
+
+	ret = octeon_i2c_hlc_wait(i2c);
+	if (ret)
+		goto err;
+
+	cmd = __raw_readq(i2c->twsi_base + SW_TWSI);
+	if ((cmd & SW_TWSI_R) == 0)
+		return -EAGAIN;
+
+	for (i = 0, j = msgs[1].len - 1; i  < msgs[1].len && i < 4; i++, j--)
+		msgs[1].buf[j] = (cmd >> (8 * i)) & 0xff;
+
+	if (msgs[1].len > 4) {
+		cmd = __raw_readq(i2c->twsi_base + SW_TWSI_EXT);
+		for (i = 0; i  < msgs[1].len - 4 && i < 4; i++, j--)
+			msgs[1].buf[j] = (cmd >> (8 * i)) & 0xff;
+	}
+
+err:
+	return ret;
+}
+
+/* high-level-controller composite write+write, m[0]len<=2, m[1]len<=8 */
+static int octeon_i2c_hlc_comp_write(struct octeon_i2c *i2c, struct i2c_msg *msgs)
+{
+	bool set_ext = false;
+	int i, j, ret = 0;
+	u64 cmd, ext = 0;
+
+	octeon_i2c_hlc_enable(i2c);
+
+	cmd = SW_TWSI_V | SW_TWSI_SOVR;
+	/* SIZE */
+	cmd |= (u64) (msgs[1].len - 1) << SW_TWSI_SIZE_SHIFT;
+	/* A */
+	cmd |= (u64) (msgs[0].addr & 0x7full) << SW_TWSI_ADDR_SHIFT;
+
+	if (msgs[0].flags & I2C_M_TEN)
+		cmd |= SW_TWSI_OP_10_IA;
+	else
+		cmd |= SW_TWSI_OP_7_IA;
+
+	if (msgs[0].len == 2) {
+		cmd |= SW_TWSI_EIA;
+		ext |= (u64) msgs[0].buf[0] << SW_TWSI_IA_SHIFT;
+		set_ext = true;
+		cmd |= (u64) msgs[0].buf[1] << SW_TWSI_IA_SHIFT;
+	} else
+		cmd |= (u64) msgs[0].buf[0] << SW_TWSI_IA_SHIFT;
+
+	for (i = 0, j = msgs[1].len - 1; i  < msgs[1].len && i < 4; i++, j--)
+		cmd |= (u64) msgs[1].buf[j] << (8 * i);
+
+	if (msgs[1].len > 4) {
+		for (i = 0; i < msgs[1].len - 4 && i < 4; i++, j--)
+			ext |= (u64)msgs[1].buf[j] << (8 * i);
+		set_ext = true;
+	}
+	if (set_ext)
+		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT);
+
+	octeon_i2c_hlc_int_clear(i2c);
+	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI);
+
+	ret = octeon_i2c_hlc_wait(i2c);
+	if (ret)
+		goto err;
+
+	cmd = __raw_readq(i2c->twsi_base + SW_TWSI);
+	if ((cmd & SW_TWSI_R) == 0)
+		return -EAGAIN;
+
+	ret = octeon_i2c_check_status(i2c, false);
+
+err:
+	return ret;
+}
+
 /* calculate and set clock divisors */
 static void octeon_i2c_set_clock(struct octeon_i2c *i2c)
 {
@@ -344,23 +638,29 @@ static void octeon_i2c_set_clock(struct octeon_i2c *i2c)
 
 static int octeon_i2c_init_lowlevel(struct octeon_i2c *i2c)
 {
-	u8 status;
+	u8 status = 0;
 	int tries;
 
-	/* disable high level controller, enable bus access */
-	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
-
 	/* reset controller */
 	octeon_i2c_reg_write(i2c, SW_TWSI_EOP_TWSI_RST, 0);
 
-	for (tries = 10; tries; tries--) {
+	for (tries = 10; tries && status != STAT_IDLE; tries--) {
 		udelay(1);
 		status = octeon_i2c_stat_read(i2c);
 		if (status == STAT_IDLE)
-			return 0;
+			break;
 	}
-	dev_err(i2c->dev, "%s: TWSI_RST failed! (0x%x)\n", __func__, status);
-	return -EIO;
+
+	if (status != STAT_IDLE) {
+		dev_err(i2c->dev, "%s: TWSI_RST failed! (0x%x)\n",
+			__func__, status);
+		return -EIO;
+	}
+
+	/* toggle twice to force both teardowns */
+	octeon_i2c_hlc_enable(i2c);
+	octeon_i2c_hlc_disable(i2c);
+	return 0;
 }
 
 static int octeon_i2c_recovery(struct octeon_i2c *i2c)
@@ -385,6 +685,8 @@ static int octeon_i2c_start(struct octeon_i2c *i2c)
 	int ret;
 	u8 stat;
 
+	octeon_i2c_hlc_disable(i2c);
+
 	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB | TWSI_CTL_STA);
 	ret = octeon_i2c_wait(i2c);
 	if (ret)
@@ -524,6 +826,28 @@ static int octeon_i2c_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs,
 	struct octeon_i2c *i2c = i2c_get_adapdata(adap);
 	int i, ret = 0;
 
+	if (num == 1) {
+		if (msgs[0].len > 0 && msgs[0].len <= 8) {
+			if (msgs[0].flags & I2C_M_RD)
+				ret = octeon_i2c_hlc_read(i2c, msgs);
+			else
+				ret = octeon_i2c_hlc_write(i2c, msgs);
+			goto out;
+		}
+	} else if (num == 2) {
+		if ((msgs[0].flags & I2C_M_RD) == 0 &&
+		    (msgs[1].flags & I2C_M_RECV_LEN) == 0 &&
+		    msgs[0].len > 0 && msgs[0].len <= 2 &&
+		    msgs[1].len > 0 && msgs[1].len <= 8 &&
+		    msgs[0].addr == msgs[1].addr) {
+			if (msgs[1].flags & I2C_M_RD)
+				ret = octeon_i2c_hlc_comp_read(i2c, msgs);
+			else
+				ret = octeon_i2c_hlc_comp_write(i2c, msgs);
+			goto out;
+		}
+	}
+
 	for (i = 0; ret == 0 && i < num; i++) {
 		struct i2c_msg *pmsg = &msgs[i];
 
@@ -539,7 +863,7 @@ static int octeon_i2c_xfer(struct i2c_adapter *adap, struct i2c_msg *msgs,
 					       pmsg->len);
 	}
 	octeon_i2c_stop(i2c);
-
+out:
 	return (ret != 0) ? ret : num;
 }
 
@@ -578,6 +902,7 @@ static void octeon_i2c_prepare_recovery(struct i2c_adapter *adap)
 	 */
 	octeon_i2c_stop(i2c);
 
+	octeon_i2c_hlc_disable(i2c);
 	octeon_i2c_write_int(i2c, 0);
 }
 
Signature · 1 line
1.9.1
Applied
#18 lore raw

Re: [PATCH v7 05/15] i2c: octeon: Enable High-Level Controller

Quoted 10 lines · Jan Glauber

From: David Daney <[email protected]>

Use High-Level Controller (HLC) when possible. The HLC can read/write up to 8 bytes and is completely optional. The most important difference of the HLC is that it only requires one interrupt for a transfer (up to 8 bytes) where the low-level read/write requires 2 interrupts plus one interrupt per transferred byte. Since the interrupts are costly using the HLC improves the performance. Also, the HLC provides improved error handling.

Quoted 2 lines

Applied to for-next, thanks!

I can recommend using --strict with checkpatch. It found some issues:

total: 0 errors, 0 warnings, 18 checks, 427 lines checked

CHECK: No space is necessary after a cast
#327: FILE: drivers/i2c/busses/i2c-octeon.c:563:
+ ext |= (u64) msgs[0].buf[0] << SW_TWSI_IA_SHIFT;

I fixed all 14 of those.

CHECK: braces {} should be used on all arms of this statement #325: FILE: drivers/i2c/busses/i2c-octeon.c:561:
+ if (msgs[0].len == 2) {
[...]
+ } else
[...]

and 2 of those.

#19 lore raw

Re: [PATCH v7 05/15] i2c: octeon: Enable High-Level Controller

Quoted 11 lines · Wolfram Sang
Quoted 10 lines · Jan Glauber

From: David Daney <[email protected]>

Use High-Level Controller (HLC) when possible. The HLC can read/write up to 8 bytes and is completely optional. The most important difference of the HLC is that it only requires one interrupt for a transfer (up to 8 bytes) where the low-level read/write requires 2 interrupts plus one interrupt per transferred byte. Since the interrupts are costly using the HLC improves the performance. Also, the HLC provides improved error handling.

Quoted 6 lines
Quoted 2 lines

Applied to for-next, thanks!

I can recommend using --strict with checkpatch. It found some issues:

I thought checkpatch is always strict, but seriously, I didn't know that option. Thanks for fixing that!

Jan

Quoted 17 lines

total: 0 errors, 0 warnings, 18 checks, 427 lines checked

CHECK: No space is necessary after a cast
#327: FILE: drivers/i2c/busses/i2c-octeon.c:563:

drivers/i2c/busses/i2c-octeon.c +1 -0
+		ext |= (u64) msgs[0].buf[0] << SW_TWSI_IA_SHIFT;

I fixed all 14 of those.

CHECK: braces {} should be used on all arms of this statement #325: FILE: drivers/i2c/busses/i2c-octeon.c:561:

(quoted diff) +2 -0
+	if (msgs[0].len == 2) {
[...]
+	} else
[...]

and 2 of those.

#20 lore raw

PATCH v7 6/15 dt-bindings: i2c: Add Octeon cn78xx TWSI

In reply to Jan Glauber

Add compatible string for Cavium Octeon cn78XX SOCs TWSI.

Cc Rob Herring <[email protected]>
Cc Pawel Moll <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Ian Campbell <[email protected]>
Cc Kumar Gala <[email protected]>
---
 Documentation/devicetree/bindings/i2c/i2c-octeon.txt | 6 ++++++
 1 file changed, 6 insertions(+)
Documentation/devicetree/bindings/i2c/i2c-octeon.txt +6 -0
index dced82e..872d485 100644
@@ -4,6 +4,12 @@
 
   Compatibility with all cn3XXX, cn5XXX and cn6XXX SOCs.
 
+  or
+
+  compatible: "cavium,octeon-7890-twsi"
+
+  Compatibility with cn78XX SOCs.
+
 - reg: The base address of the TWSI/I2C bus controller register bank.
 
 - #address-cells: Must be <1>.
Signature · 1 line
1.9.1
Applied
#21 lore raw

Re: [PATCH v7 06/15] dt-bindings: i2c: Add Octeon cn78xx TWSI

Quoted 10 lines · Jan Glauber

Add compatible string for Cavium Octeon cn78XX SOCs TWSI.

Cc Rob Herring <[email protected]>
Cc Pawel Moll <[email protected]>
Cc Mark Rutland <[email protected]>
Cc Ian Campbell <[email protected]>
Cc Kumar Gala <[email protected]>

Applied to for-next, but will squash it with the next one.

#22 lore raw

PATCH v7 7/15 i2c: octeon: Add support for cn78xx chips

In reply to Jan Glauber

From: David Daney <[email protected]>

cn78xx has a different interrupt architecture, so we have to manage the interrupts differently.

---
 drivers/i2c/busses/i2c-octeon.c | 131 ++++++++++++++++++++++++++++++++++++----
 1 file changed, 120 insertions(+), 11 deletions(-)
drivers/i2c/busses/i2c-octeon.c +120 -11
index 9520f78..76fa479 100644
@@ -11,6 +11,7 @@
  * warranty of any kind, whether express or implied.
  */
 
+#include <linux/atomic.h>
 #include <linux/platform_device.h>
 #include <linux/interrupt.h>
 #include <linux/kernel.h>
@@ -112,11 +113,18 @@ struct octeon_i2c {
 	wait_queue_head_t queue;
 	struct i2c_adapter adap;
 	int irq;
+	int hlc_irq;		/* For cn7890 only */
 	u32 twsi_freq;
 	int sys_freq;
 	void __iomem *twsi_base;
 	struct device *dev;
 	bool hlc_enabled;
+	void (*int_enable)(struct octeon_i2c *);
+	void (*int_disable)(struct octeon_i2c *);
+	void (*hlc_int_enable)(struct octeon_i2c *);
+	void (*hlc_int_disable)(struct octeon_i2c *);
+	atomic_t int_enable_cnt;
+	atomic_t hlc_int_enable_cnt;
 };
 
 static void octeon_i2c_writeq_flush(u64 val, void __iomem *addr)
@@ -217,6 +225,58 @@ static void octeon_i2c_int_disable(struct octeon_i2c *i2c)
 	octeon_i2c_write_int(i2c, 0);
 }
 
+/**
+ * octeon_i2c_int_enable78 - enable the CORE interrupt
+ * @i2c: The struct octeon_i2c
+ *
+ * The interrupt will be asserted when there is non-STAT_IDLE state in the
+ * SW_TWSI_EOP_TWSI_STAT register.
+ */
+static void octeon_i2c_int_enable78(struct octeon_i2c *i2c)
+{
+	atomic_inc_return(&i2c->int_enable_cnt);
+	enable_irq(i2c->irq);
+}
+
+static void __octeon_i2c_irq_disable(atomic_t *cnt, int irq)
+{
+	int count;
+
+	/*
+	 * The interrupt can be disabled in two places, but we only
+	 * want to make the disable_irq_nosync() call once, so keep
+	 * track with the atomic variable.
+	 */
+	count = atomic_dec_if_positive(cnt);
+	if (count >= 0)
+		disable_irq_nosync(irq);
+}
+
+/* disable the CORE interrupt */
+static void octeon_i2c_int_disable78(struct octeon_i2c *i2c)
+{
+	__octeon_i2c_irq_disable(&i2c->int_enable_cnt, i2c->irq);
+}
+
+/**
+ * octeon_i2c_hlc_int_enable78 - enable the ST interrupt
+ * @i2c: The struct octeon_i2c
+ *
+ * The interrupt will be asserted when there is non-STAT_IDLE state in
+ * the SW_TWSI_EOP_TWSI_STAT register.
+ */
+static void octeon_i2c_hlc_int_enable78(struct octeon_i2c *i2c)
+{
+	atomic_inc_return(&i2c->hlc_int_enable_cnt);
+	enable_irq(i2c->hlc_irq);
+}
+
+/* disable the ST interrupt */
+static void octeon_i2c_hlc_int_disable78(struct octeon_i2c *i2c)
+{
+	__octeon_i2c_irq_disable(&i2c->hlc_int_enable_cnt, i2c->hlc_irq);
+}
+
 /*
  * Cleanup low-level state & enable high-level controller.
  */
@@ -263,7 +323,18 @@ static irqreturn_t octeon_i2c_isr(int irq, void *dev_id)
 {
 	struct octeon_i2c *i2c = dev_id;
 
-	octeon_i2c_int_disable(i2c);
+	i2c->int_disable(i2c);
+	wake_up(&i2c->queue);
+
+	return IRQ_HANDLED;
+}
+
+/* HLC interrupt service routine */
+static irqreturn_t octeon_i2c_hlc_isr78(int irq, void *dev_id)
+{
+	struct octeon_i2c *i2c = dev_id;
+
+	i2c->hlc_int_disable(i2c);
 	wake_up(&i2c->queue);
 
 	return IRQ_HANDLED;
@@ -284,10 +355,10 @@ static int octeon_i2c_wait(struct octeon_i2c *i2c)
 {
 	long time_left;
 
-	octeon_i2c_int_enable(i2c);
+	i2c->int_enable(i2c);
 	time_left = wait_event_timeout(i2c->queue, octeon_i2c_test_iflg(i2c),
 				       i2c->adap.timeout);
-	octeon_i2c_int_disable(i2c);
+	i2c->int_disable(i2c);
 	if (!time_left) {
 		dev_dbg(i2c->dev, "%s: timeout\n", __func__);
 		return -ETIMEDOUT;
@@ -385,11 +456,11 @@ static int octeon_i2c_hlc_wait(struct octeon_i2c *i2c)
 {
 	int time_left;
 
-	octeon_i2c_hlc_int_enable(i2c);
+	i2c->hlc_int_enable(i2c);
 	time_left = wait_event_timeout(i2c->queue,
 				       octeon_i2c_hlc_test_ready(i2c),
 				       i2c->adap.timeout);
-	octeon_i2c_int_disable(i2c);
+	i2c->hlc_int_disable(i2c);
 	if (!time_left) {
 		octeon_i2c_hlc_int_clear(i2c);
 		return -ETIMEDOUT;
@@ -942,14 +1013,26 @@ static struct i2c_adapter octeon_i2c_ops = {
 static int octeon_i2c_probe(struct platform_device *pdev)
 {
 	struct device_node *node = pdev->dev.of_node;
+	int irq, result = 0, hlc_irq = 0;
 	struct resource *res_mem;
 	struct octeon_i2c *i2c;
-	int irq, result = 0;
-
-	/* All adaptors have an irq.  */
-	irq = platform_get_irq(pdev, 0);
-	if (irq < 0)
-		return irq;
+	bool cn78xx_style;
+
+	cn78xx_style = of_device_is_compatible(node, "cavium,octeon-7890-twsi");
+	if (cn78xx_style) {
+		hlc_irq = platform_get_irq(pdev, 0);
+		if (hlc_irq < 0)
+			return hlc_irq;
+
+		irq = platform_get_irq(pdev, 2);
+		if (irq < 0)
+			return irq;
+	} else {
+		/* All adaptors have an irq.  */
+		irq = platform_get_irq(pdev, 0);
+		if (irq < 0)
+			return irq;
+	}
 
 	i2c = devm_kzalloc(&pdev->dev, sizeof(*i2c), GFP_KERNEL);
 	if (!i2c) {
@@ -984,6 +1067,31 @@ static int octeon_i2c_probe(struct platform_device *pdev)
 
 	i2c->irq = irq;
 
+	if (cn78xx_style) {
+		i2c->hlc_irq = hlc_irq;
+
+		i2c->int_enable = octeon_i2c_int_enable78;
+		i2c->int_disable = octeon_i2c_int_disable78;
+		i2c->hlc_int_enable = octeon_i2c_hlc_int_enable78;
+		i2c->hlc_int_disable = octeon_i2c_hlc_int_disable78;
+
+		irq_set_status_flags(i2c->irq, IRQ_NOAUTOEN);
+		irq_set_status_flags(i2c->hlc_irq, IRQ_NOAUTOEN);
+
+		result = devm_request_irq(&pdev->dev, i2c->hlc_irq,
+					  octeon_i2c_hlc_isr78, 0,
+					  DRV_NAME, i2c);
+		if (result < 0) {
+			dev_err(i2c->dev, "failed to attach interrupt\n");
+			goto out;
+		}
+	} else {
+		i2c->int_enable = octeon_i2c_int_enable;
+		i2c->int_disable = octeon_i2c_int_disable;
+		i2c->hlc_int_enable = octeon_i2c_hlc_int_enable;
+		i2c->hlc_int_disable = octeon_i2c_int_disable;
+	}
+
 	result = devm_request_irq(&pdev->dev, i2c->irq,
 				  octeon_i2c_isr, 0, DRV_NAME, i2c);
 	if (result < 0) {
@@ -1030,6 +1138,7 @@ static int octeon_i2c_remove(struct platform_device *pdev)
 
 static const struct of_device_id octeon_i2c_match[] = {
 	{ .compatible = "cavium,octeon-3860-twsi", },
+	{ .compatible = "cavium,octeon-7890-twsi", },
 	{},
 };
 MODULE_DEVICE_TABLE(of, octeon_i2c_match);
Signature · 1 line
1.9.1
#24 lore raw

PATCH v7 8/15 i2c: octeon: Improve performance if interrupt is early

In reply to Jan Glauber

From: Peter Swain <[email protected]>

There is a race between the TWSI interrupt and the condition that is required before proceeding:

Low-level: interrupt flag bit must be set
High-level controller: valid bit must be clear

If the interrupt comes too early and the condition is not met the wait will time out, and the transfer is aborted leading to very poor performance.

To avoid this race retry for the condition ~80 µs later. The retry is avoided on the very first invocation of wait_event_timeout() (which tests the condition before entering the wait and is therefore always wrong in this case).

EEPROM reads on 100kHz i2c now measure ~5.2kB/s, about 1/2 what's achievable, and much better than the worst-case 100 bytes/sec before.

While at it remove the debug print from the low-level wait function.

---
 drivers/i2c/busses/i2c-octeon.c | 55 ++++++++++++++++++++++++++++++++++-------
 1 file changed, 46 insertions(+), 9 deletions(-)
drivers/i2c/busses/i2c-octeon.c +46 -9
index 76fa479..3144fe9 100644
@@ -109,6 +109,8 @@
 #define TWSI_INT_SDA		BIT_ULL(10)
 #define TWSI_INT_SCL		BIT_ULL(11)
 
+#define I2C_OCTEON_EVENT_WAIT 80 /* microseconds */
+
 struct octeon_i2c {
 	wait_queue_head_t queue;
 	struct i2c_adapter adap;
@@ -340,11 +342,29 @@ static irqreturn_t octeon_i2c_hlc_isr78(int irq, void *dev_id)
 	return IRQ_HANDLED;
 }
 
-static int octeon_i2c_test_iflg(struct octeon_i2c *i2c)
+static bool octeon_i2c_test_iflg(struct octeon_i2c *i2c)
 {
 	return (octeon_i2c_ctl_read(i2c) & TWSI_CTL_IFLG);
 }
 
+static bool octeon_i2c_test_ready(struct octeon_i2c *i2c, bool *first)
+{
+	if (octeon_i2c_test_iflg(i2c))
+		return true;
+
+	if (*first) {
+		*first = false;
+		return false;
+	}
+
+	/*
+	 * IRQ has signaled an event but IFLG hasn't changed.
+	 * Sleep and retry once.
+	 */
+	usleep_range(I2C_OCTEON_EVENT_WAIT, 2 * I2C_OCTEON_EVENT_WAIT);
+	return octeon_i2c_test_iflg(i2c);
+}
+
 /**
  * octeon_i2c_wait - wait for the IFLG to be set
  * @i2c: The struct octeon_i2c
@@ -354,15 +374,14 @@ static int octeon_i2c_test_iflg(struct octeon_i2c *i2c)
 static int octeon_i2c_wait(struct octeon_i2c *i2c)
 {
 	long time_left;
+	bool first = 1;
 
 	i2c->int_enable(i2c);
-	time_left = wait_event_timeout(i2c->queue, octeon_i2c_test_iflg(i2c),
+	time_left = wait_event_timeout(i2c->queue, octeon_i2c_test_ready(i2c, &first),
 				       i2c->adap.timeout);
 	i2c->int_disable(i2c);
-	if (!time_left) {
-		dev_dbg(i2c->dev, "%s: timeout\n", __func__);
+	if (!time_left)
 		return -ETIMEDOUT;
-	}
 
 	return 0;
 }
@@ -428,11 +447,28 @@ static int octeon_i2c_check_status(struct octeon_i2c *i2c, int final_read)
 	}
 }
 
-static bool octeon_i2c_hlc_test_ready(struct octeon_i2c *i2c)
+static bool octeon_i2c_hlc_test_valid(struct octeon_i2c *i2c)
+{
+	return (__raw_readq(i2c->twsi_base + SW_TWSI) & SW_TWSI_V) == 0;
+}
+
+static bool octeon_i2c_hlc_test_ready(struct octeon_i2c *i2c, bool *first)
 {
-	u64 val = __raw_readq(i2c->twsi_base + SW_TWSI);
+	/* check if valid bit is cleared */
+	if (octeon_i2c_hlc_test_valid(i2c))
+		return true;
 
-	return (val & SW_TWSI_V) == 0;
+	if (*first) {
+		*first = false;
+		return false;
+	}
+
+	/*
+	 * IRQ has signaled an event but valid bit isn't cleared.
+	 * Sleep and retry once.
+	 */
+	usleep_range(I2C_OCTEON_EVENT_WAIT, 2 * I2C_OCTEON_EVENT_WAIT);
+	return octeon_i2c_hlc_test_valid(i2c);
 }
 
 static void octeon_i2c_hlc_int_enable(struct octeon_i2c *i2c)
@@ -454,11 +490,12 @@ static void octeon_i2c_hlc_int_clear(struct octeon_i2c *i2c)
  */
 static int octeon_i2c_hlc_wait(struct octeon_i2c *i2c)
 {
+	bool first = 1;
 	int time_left;
 
 	i2c->hlc_int_enable(i2c);
 	time_left = wait_event_timeout(i2c->queue,
-				       octeon_i2c_hlc_test_ready(i2c),
+				       octeon_i2c_hlc_test_ready(i2c, &first),
 				       i2c->adap.timeout);
 	i2c->hlc_int_disable(i2c);
 	if (!time_left) {
Signature · 1 line
1.9.1
Applied
#25 lore raw

Re: [PATCH v7 08/15] i2c: octeon: Improve performance if interrupt is early

Quoted 17 lines · Jan Glauber

From: Peter Swain <[email protected]>

There is a race between the TWSI interrupt and the condition that is required before proceeding:

Low-level: interrupt flag bit must be set
High-level controller: valid bit must be clear

If the interrupt comes too early and the condition is not met the wait will time out, and the transfer is aborted leading to very poor performance.

To avoid this race retry for the condition ~80 µs later. The retry is avoided on the very first invocation of wait_event_timeout() (which tests the condition before entering the wait and is therefore always wrong in this case).

Quoted 7 lines

EEPROM reads on 100kHz i2c now measure ~5.2kB/s, about 1/2 what's achievable, and much better than the worst-case 100 bytes/sec before.

While at it remove the debug print from the low-level wait function.

Applied to for-next, thanks!

Assigned 'true' instead of '1' to bool...

#26 lore raw

Re: [PATCH v7 08/15] i2c: octeon: Improve performance if interrupt is early

Quoted 1 line

Assigned 'true' instead of '1' to bool...

To save you from annoying rebasing, I reverted my change. But it would be nice if it could be fixed somewhen later.

#27 lore raw

PATCH v7 9/15 i2c: octeon: Add workaround for broken irqs on CN3860

In reply to Jan Glauber

From: David Daney <[email protected]>

CN3860 does not interrupt the CPU when the i2c status changes. If we get a timeout, and see the status has in fact changed, we know we have this problem, and drop back to polling.

---
 drivers/i2c/busses/i2c-octeon.c | 53 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 51 insertions(+), 2 deletions(-)
drivers/i2c/busses/i2c-octeon.c +51 -2
index 3144fe9..009cc33 100644
@@ -121,6 +121,8 @@ struct octeon_i2c {
 	void __iomem *twsi_base;
 	struct device *dev;
 	bool hlc_enabled;
+	bool broken_irq_mode;
+	bool broken_irq_check;
 	void (*int_enable)(struct octeon_i2c *);
 	void (*int_disable)(struct octeon_i2c *);
 	void (*hlc_int_enable)(struct octeon_i2c *);
@@ -376,10 +378,32 @@ static int octeon_i2c_wait(struct octeon_i2c *i2c)
 	long time_left;
 	bool first = 1;
 
+	/*
+	 * Some chip revisions don't assert the irq in the interrupt
+	 * controller. So we must poll for the IFLG change.
+	 */
+	if (i2c->broken_irq_mode) {
+		u64 end = get_jiffies_64() + i2c->adap.timeout;
+
+		while (!octeon_i2c_test_iflg(i2c) &&
+		       time_before64(get_jiffies_64(), end))
+			udelay(50);
+
+		return octeon_i2c_test_iflg(i2c) ? 0 : -ETIMEDOUT;
+	}
+
 	i2c->int_enable(i2c);
 	time_left = wait_event_timeout(i2c->queue, octeon_i2c_test_ready(i2c, &first),
 				       i2c->adap.timeout);
 	i2c->int_disable(i2c);
+
+	if (i2c->broken_irq_check && !time_left &&
+	    octeon_i2c_test_iflg(i2c)) {
+		dev_err(i2c->dev, "broken irq connection detected, switching to polling mode.\n");
+		i2c->broken_irq_mode = true;
+		return 0;
+	}
+
 	if (!time_left)
 		return -ETIMEDOUT;
 
@@ -493,15 +517,37 @@ static int octeon_i2c_hlc_wait(struct octeon_i2c *i2c)
 	bool first = 1;
 	int time_left;
 
+	/*
+	 * Some cn38xx boards don't assert the irq in the interrupt
+	 * controller. So we must poll for the valid bit change.
+	 */
+	if (i2c->broken_irq_mode) {
+		u64 end = get_jiffies_64() + i2c->adap.timeout;
+
+		while (!octeon_i2c_hlc_test_valid(i2c) &&
+		       time_before64(get_jiffies_64(), end))
+			udelay(50);
+
+		return octeon_i2c_hlc_test_valid(i2c) ? 0 : -ETIMEDOUT;
+	}
+
 	i2c->hlc_int_enable(i2c);
 	time_left = wait_event_timeout(i2c->queue,
 				       octeon_i2c_hlc_test_ready(i2c, &first),
 				       i2c->adap.timeout);
 	i2c->hlc_int_disable(i2c);
-	if (!time_left) {
+	if (!time_left)
 		octeon_i2c_hlc_int_clear(i2c);
-		return -ETIMEDOUT;
+
+	if (i2c->broken_irq_check && !time_left &&
+	    octeon_i2c_hlc_test_valid(i2c)) {
+		dev_err(i2c->dev, "broken irq connection detected, switching to polling mode.\n");
+			i2c->broken_irq_mode = true;
+			return 0;
 	}
+
+	if (!time_left)
+		return -ETIMEDOUT;
 	return 0;
 }
 
@@ -1136,6 +1182,9 @@ static int octeon_i2c_probe(struct platform_device *pdev)
 		goto out;
 	}
 
+	if (OCTEON_IS_MODEL(OCTEON_CN38XX))
+		i2c->broken_irq_check = true;
+
 	result = octeon_i2c_init_lowlevel(i2c);
 	if (result) {
 		dev_err(i2c->dev, "init low level failed\n");
Signature · 1 line
1.9.1
#28 lore raw

Re: [PATCH v7 09/15] i2c: octeon: Add workaround for broken irqs on CN3860

Quoted 8 lines · Jan Glauber

From: David Daney <[email protected]>

CN3860 does not interrupt the CPU when the i2c status changes. If we get a timeout, and see the status has in fact changed, we know we have this problem, and drop back to polling.

My code checkers say something:

CHECKPATCH
CHECK: usleep_range is preferred over udelay; see Documentation/timers/timers-howto.txt #37: FILE: drivers/i2c/busses/i2c-octeon.c:390:
+ udelay(50);

CHECK: usleep_range is preferred over udelay; see Documentation/timers/timers-howto.txt #70: FILE: drivers/i2c/busses/i2c-octeon.c:529:
+ udelay(50);

Dunno if you want to change that? Seems reasonable to me. Also:

SMATCH
drivers/i2c/busses/i2c-octeon.c:544 octeon_i2c_hlc_wait() warn: inconsistent indenting

This is true as well.

#29 lore raw

Re: [PATCH v7 09/15] i2c: octeon: Add workaround for broken irqs on CN3860

Quoted 21 lines · Wolfram Sang
Quoted 8 lines · Jan Glauber

From: David Daney <[email protected]>

CN3860 does not interrupt the CPU when the i2c status changes. If we get a timeout, and see the status has in fact changed, we know we have this problem, and drop back to polling.

My code checkers say something:

CHECKPATCH
CHECK: usleep_range is preferred over udelay; see Documentation/timers/timers-howto.txt #37: FILE: drivers/i2c/busses/i2c-octeon.c:390:

(quoted diff) +1 -0
+			udelay(50);

CHECK: usleep_range is preferred over udelay; see Documentation/timers/timers-howto.txt #70: FILE: drivers/i2c/busses/i2c-octeon.c:529:

(quoted diff) +1 -0
+			udelay(50);
Quoted 1 line

Dunno if you want to change that? Seems reasonable to me. Also:

Yes, makes sense.

Quoted 5 lines

SMATCH
drivers/i2c/busses/i2c-octeon.c:544 octeon_i2c_hlc_wait() warn: inconsistent indenting

This is true as well.

OK, I'll reply with an updated patch.

#30 lore raw

PATCH 1/1 i2c: octeon: Add workaround for broken irqs on CN3860

In reply to Wolfram Sang

From: David Daney <[email protected]>

CN3860 does not interrupt the CPU when the i2c status changes. If we get a timeout, and see the status has in fact changed, we know we have this problem, and drop back to polling.

---
 drivers/i2c/busses/i2c-octeon.c | 53 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 51 insertions(+), 2 deletions(-)
drivers/i2c/busses/i2c-octeon.c +51 -2
index 3144fe9..5b32a02 100644
@@ -121,6 +121,8 @@ struct octeon_i2c {
 	void __iomem *twsi_base;
 	struct device *dev;
 	bool hlc_enabled;
+	bool broken_irq_mode;
+	bool broken_irq_check;
 	void (*int_enable)(struct octeon_i2c *);
 	void (*int_disable)(struct octeon_i2c *);
 	void (*hlc_int_enable)(struct octeon_i2c *);
@@ -376,10 +378,32 @@ static int octeon_i2c_wait(struct octeon_i2c *i2c)
 	long time_left;
 	bool first = 1;
 
+	/*
+	 * Some chip revisions don't assert the irq in the interrupt
+	 * controller. So we must poll for the IFLG change.
+	 */
+	if (i2c->broken_irq_mode) {
+		u64 end = get_jiffies_64() + i2c->adap.timeout;
+
+		while (!octeon_i2c_test_iflg(i2c) &&
+		       time_before64(get_jiffies_64(), end))
+			usleep_range(I2C_OCTEON_EVENT_WAIT / 2, I2C_OCTEON_EVENT_WAIT);
+
+		return octeon_i2c_test_iflg(i2c) ? 0 : -ETIMEDOUT;
+	}
+
 	i2c->int_enable(i2c);
 	time_left = wait_event_timeout(i2c->queue, octeon_i2c_test_ready(i2c, &first),
 				       i2c->adap.timeout);
 	i2c->int_disable(i2c);
+
+	if (i2c->broken_irq_check && !time_left &&
+	    octeon_i2c_test_iflg(i2c)) {
+		dev_err(i2c->dev, "broken irq connection detected, switching to polling mode.\n");
+		i2c->broken_irq_mode = true;
+		return 0;
+	}
+
 	if (!time_left)
 		return -ETIMEDOUT;
 
@@ -493,15 +517,37 @@ static int octeon_i2c_hlc_wait(struct octeon_i2c *i2c)
 	bool first = 1;
 	int time_left;
 
+	/*
+	 * Some cn38xx boards don't assert the irq in the interrupt
+	 * controller. So we must poll for the valid bit change.
+	 */
+	if (i2c->broken_irq_mode) {
+		u64 end = get_jiffies_64() + i2c->adap.timeout;
+
+		while (!octeon_i2c_hlc_test_valid(i2c) &&
+		       time_before64(get_jiffies_64(), end))
+			usleep_range(I2C_OCTEON_EVENT_WAIT / 2, I2C_OCTEON_EVENT_WAIT);
+
+		return octeon_i2c_hlc_test_valid(i2c) ? 0 : -ETIMEDOUT;
+	}
+
 	i2c->hlc_int_enable(i2c);
 	time_left = wait_event_timeout(i2c->queue,
 				       octeon_i2c_hlc_test_ready(i2c, &first),
 				       i2c->adap.timeout);
 	i2c->hlc_int_disable(i2c);
-	if (!time_left) {
+	if (!time_left)
 		octeon_i2c_hlc_int_clear(i2c);
-		return -ETIMEDOUT;
+
+	if (i2c->broken_irq_check && !time_left &&
+	    octeon_i2c_hlc_test_valid(i2c)) {
+		dev_err(i2c->dev, "broken irq connection detected, switching to polling mode.\n");
+		i2c->broken_irq_mode = true;
+		return 0;
 	}
+
+	if (!time_left)
+		return -ETIMEDOUT;
 	return 0;
 }
 
@@ -1136,6 +1182,9 @@ static int octeon_i2c_probe(struct platform_device *pdev)
 		goto out;
 	}
 
+	if (OCTEON_IS_MODEL(OCTEON_CN38XX))
+		i2c->broken_irq_check = true;
+
 	result = octeon_i2c_init_lowlevel(i2c);
 	if (result) {
 		dev_err(i2c->dev, "init low level failed\n");
Signature · 1 line
1.9.1
#32 lore raw

PATCH v7 10/15 i2c: octeon: Move read function before write

In reply to Jan Glauber

Just sorting the functions to be consistent with the other read/write variants.

---
 drivers/i2c/busses/i2c-octeon.c | 78 ++++++++++++++++++++---------------------
 1 file changed, 39 insertions(+), 39 deletions(-)
drivers/i2c/busses/i2c-octeon.c +39 -39
index 009cc33..2bc32b9 100644
@@ -864,45 +864,6 @@ static void octeon_i2c_stop(struct octeon_i2c *i2c)
 }
 
 /**
- * octeon_i2c_write - send data to the bus via low-level controller
- * @i2c: The struct octeon_i2c
- * @target: Target address
- * @data: Pointer to the data to be sent
- * @length: Length of the data
- *
- * The address is sent over the bus, then the data.
- *
- * Returns 0 on success, otherwise a negative errno.
- */
-static int octeon_i2c_write(struct octeon_i2c *i2c, int target,
-			    const u8 *data, int length)
-{
-	int i, result;
-
-	octeon_i2c_data_write(i2c, target << 1);
-	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
-
-	result = octeon_i2c_wait(i2c);
-	if (result)
-		return result;
-
-	for (i = 0; i < length; i++) {
-		result = octeon_i2c_check_status(i2c, false);
-		if (result)
-			return result;
-
-		octeon_i2c_data_write(i2c, data[i]);
-		octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
-
-		result = octeon_i2c_wait(i2c);
-		if (result)
-			return result;
-	}
-
-	return 0;
-}
-
-/**
  * octeon_i2c_read - receive data from the bus via low-level controller
  * @i2c: The struct octeon_i2c
  * @target: Target address
@@ -967,6 +928,45 @@ static int octeon_i2c_read(struct octeon_i2c *i2c, int target,
 }
 
 /**
+ * octeon_i2c_write - send data to the bus via low-level controller
+ * @i2c: The struct octeon_i2c
+ * @target: Target address
+ * @data: Pointer to the data to be sent
+ * @length: Length of the data
+ *
+ * The address is sent over the bus, then the data.
+ *
+ * Returns 0 on success, otherwise a negative errno.
+ */
+static int octeon_i2c_write(struct octeon_i2c *i2c, int target,
+			    const u8 *data, int length)
+{
+	int i, result;
+
+	octeon_i2c_data_write(i2c, target << 1);
+	octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
+
+	result = octeon_i2c_wait(i2c);
+	if (result)
+		return result;
+
+	for (i = 0; i < length; i++) {
+		result = octeon_i2c_check_status(i2c, false);
+		if (result)
+			return result;
+
+		octeon_i2c_data_write(i2c, data[i]);
+		octeon_i2c_ctl_write(i2c, TWSI_CTL_ENAB);
+
+		result = octeon_i2c_wait(i2c);
+		if (result)
+			return result;
+	}
+
+	return 0;
+}
+
+/**
  * octeon_i2c_xfer - The driver's master_xfer function
  * @adap: Pointer to the i2c_adapter structure
  * @msgs: Pointer to the messages to be processed
Signature · 1 line
1.9.1
#33 lore raw

PATCH v7 11/15 i2c: octeon: Rename driver to prepare for split

In reply to Jan Glauber

This is an intermediate commit in preparation of the driver split. The module rename in this commit will be reverted in the next patch, this is just done to make the series bisectible.

---
 drivers/i2c/busses/Makefile                            | 2 +-
 drivers/i2c/busses/{i2c-octeon.c => i2c-octeon-core.c} | 0
 2 files changed, 1 insertion(+), 1 deletion(-)
 rename drivers/i2c/busses/{i2c-octeon.c => i2c-octeon-core.c} (100%)
drivers/i2c/busses/Makefile +1 -1
index 37f2819..3405286 100644
@@ -91,7 +91,7 @@ obj-$(CONFIG_I2C_UNIPHIER)	+= i2c-uniphier.o
 obj-$(CONFIG_I2C_UNIPHIER_F)	+= i2c-uniphier-f.o
 obj-$(CONFIG_I2C_VERSATILE)	+= i2c-versatile.o
 obj-$(CONFIG_I2C_WMT)		+= i2c-wmt.o
-obj-$(CONFIG_I2C_OCTEON)	+= i2c-octeon.o
+obj-$(CONFIG_I2C_OCTEON)	+= i2c-octeon-core.o
 obj-$(CONFIG_I2C_XILINX)	+= i2c-xiic.o
 obj-$(CONFIG_I2C_XLR)		+= i2c-xlr.o
 obj-$(CONFIG_I2C_XLP9XX)	+= i2c-xlp9xx.o
drivers/i2c/busses/i2c-octeon-core.c renamed from drivers/i2c/busses/i2c-octeon.c +0 -0
similarity index 100%
rename from drivers/i2c/busses/i2c-octeon.c
rename to drivers/i2c/busses/i2c-octeon-core.c
Signature · 1 line
1.9.1
#34 lore raw

PATCH v7 12/15 i2c: octeon: Split the driver into two parts

In reply to Jan Glauber

Move common functionality into a separate file in preparation of the re-use from the ThunderX i2c driver.

---
 drivers/i2c/busses/Makefile          |   3 +-
 drivers/i2c/busses/i2c-cavium.c      | 799 +++++++++++++++++++++++++++++
 drivers/i2c/busses/i2c-cavium.h      | 197 +++++++
 drivers/i2c/busses/i2c-octeon-core.c | 963 +----------------------------------
 4 files changed, 999 insertions(+), 963 deletions(-)
 create mode 100644 drivers/i2c/busses/i2c-cavium.c
 create mode 100644 drivers/i2c/busses/i2c-cavium.h
drivers/i2c/busses/Makefile +2 -1
index 3405286..282f781 100644
@@ -91,7 +91,8 @@ obj-$(CONFIG_I2C_UNIPHIER)	+= i2c-uniphier.o
 obj-$(CONFIG_I2C_UNIPHIER_F)	+= i2c-uniphier-f.o
 obj-$(CONFIG_I2C_VERSATILE)	+= i2c-versatile.o
 obj-$(CONFIG_I2C_WMT)		+= i2c-wmt.o
-obj-$(CONFIG_I2C_OCTEON)	+= i2c-octeon-core.o
+i2c-octeon-objs := i2c-cavium.o i2c-octeon-core.o
+obj-$(CONFIG_I2C_OCTEON)	+= i2c-octeon.o
 obj-$(CONFIG_I2C_XILINX)	+= i2c-xiic.o
 obj-$(CONFIG_I2C_XLR)		+= i2c-xlr.o
 obj-$(CONFIG_I2C_XLP9XX)	+= i2c-xlp9xx.o
drivers/i2c/busses/i2c-cavium.c new file +799 -0

799 lines not on this page · raw message

drivers/i2c/busses/i2c-cavium.h new file +197 -0

197 lines not on this page · raw message

drivers/i2c/busses/i2c-octeon-core.c +1 -962

989 lines not on this page · raw message

Signature · 1 line
1.9.1
#35 lore raw

PATCH v7 13/15 i2c: thunderx: Add i2c driver for ThunderX SOC

In reply to Jan Glauber

The ThunderX SOC uses the same i2c block as the Octeon SOC. The main difference is that on ThunderX the device is a PCI device so device probing is done via PCI, interrupts are MSIX and the clock is taken from device tree.

---
 drivers/i2c/busses/Kconfig             |  10 ++
 drivers/i2c/busses/Makefile            |   2 +
 drivers/i2c/busses/i2c-cavium.h        |  16 +-
 drivers/i2c/busses/i2c-thunderx-core.c | 267 +++++++++++++++++++++++++++++++++
 4 files changed, 292 insertions(+), 3 deletions(-)
 create mode 100644 drivers/i2c/busses/i2c-thunderx-core.c
drivers/i2c/busses/Kconfig +10 -0
index faa8e68..92d23de 100644
@@ -953,6 +953,16 @@ config I2C_OCTEON
 	  This driver can also be built as a module.  If so, the module
 	  will be called i2c-octeon.
 
+config I2C_THUNDERX
+	tristate "Cavium ThunderX I2C bus support"
+	depends on 64BIT && PCI && !CAVIUM_OCTEON_SOC
+	help
+	  Say yes if you want to support the I2C serial bus on Cavium
+	  ThunderX SOC.
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-thunderx.
+
 config I2C_XILINX
 	tristate "Xilinx I2C Controller"
 	depends on HAS_IOMEM
drivers/i2c/busses/Makefile +2 -0
index 282f781..a32ff14 100644
@@ -93,6 +93,8 @@ obj-$(CONFIG_I2C_VERSATILE)	+= i2c-versatile.o
 obj-$(CONFIG_I2C_WMT)		+= i2c-wmt.o
 i2c-octeon-objs := i2c-cavium.o i2c-octeon-core.o
 obj-$(CONFIG_I2C_OCTEON)	+= i2c-octeon.o
+i2c-thunderx-objs := i2c-cavium.o i2c-thunderx-core.o
+obj-$(CONFIG_I2C_THUNDERX)	+= i2c-thunderx.o
 obj-$(CONFIG_I2C_XILINX)	+= i2c-xiic.o
 obj-$(CONFIG_I2C_XLR)		+= i2c-xlr.o
 obj-$(CONFIG_I2C_XLP9XX)	+= i2c-xlp9xx.o
drivers/i2c/busses/i2c-cavium.h +13 -3
index 81c6a81..33c7e1f 100644
@@ -8,9 +8,15 @@
 #include <linux/pci.h>
 
 /* Register offsets */
-#define SW_TWSI			0x00
-#define TWSI_INT		0x10
-#define SW_TWSI_EXT		0x18
+#if IS_ENABLED(CONFIG_I2C_THUNDERX)
+	#define SW_TWSI			0x1000
+	#define TWSI_INT		0x1010
+	#define SW_TWSI_EXT		0x1018
+#else
+	#define SW_TWSI			0x00
+	#define TWSI_INT		0x10
+	#define SW_TWSI_EXT		0x18
+#endif
 
 /* Controller command patterns */
 #define SW_TWSI_V		BIT_ULL(63)	/* Valid bit */
@@ -94,6 +100,7 @@
 struct octeon_i2c {
 	wait_queue_head_t queue;
 	struct i2c_adapter adap;
+	struct clk *clk;
 	int irq;
 	int hlc_irq;		/* For cn7890 only */
 	u32 twsi_freq;
@@ -109,6 +116,9 @@ struct octeon_i2c {
 	void (*hlc_int_disable)(struct octeon_i2c *);
 	atomic_t int_enable_cnt;
 	atomic_t hlc_int_enable_cnt;
+#if IS_ENABLED(CONFIG_I2C_THUNDERX)
+	struct msix_entry i2c_msix;
+#endif
 };
 
 static inline void octeon_i2c_writeq_flush(u64 val, void __iomem *addr)
drivers/i2c/busses/i2c-thunderx-core.c new file +267 -0

267 lines not on this page · raw message

Signature · 1 line
1.9.1
#36 lore raw

PATCH v7 14/15 i2c: octeon,thunderx: Move register offsets to struct

In reply to Jan Glauber

The register offsets are different between Octeon and ThunderX so move them into the algorithm struct and get rid of the define.

---
 drivers/i2c/busses/i2c-cavium.c        | 28 +++++++++++++--------------
 drivers/i2c/busses/i2c-cavium.h        | 35 +++++++++++++++++-----------------
 drivers/i2c/busses/i2c-octeon-core.c   |  4 ++++
 drivers/i2c/busses/i2c-thunderx-core.c |  4 ++++
 4 files changed, 40 insertions(+), 31 deletions(-)
drivers/i2c/busses/i2c-cavium.c +14 -14
index ac37ae1..6ac9d63 100644
@@ -99,7 +99,7 @@ static int octeon_i2c_wait(struct octeon_i2c *i2c)
 
 static bool octeon_i2c_hlc_test_valid(struct octeon_i2c *i2c)
 {
-	return (__raw_readq(i2c->twsi_base + SW_TWSI) & SW_TWSI_V) == 0;
+	return (__raw_readq(i2c->twsi_base + SW_TWSI(i2c)) & SW_TWSI_V) == 0;
 }
 
 static bool octeon_i2c_hlc_test_ready(struct octeon_i2c *i2c, bool *first)
@@ -443,12 +443,12 @@ static int octeon_i2c_hlc_read(struct octeon_i2c *i2c, struct i2c_msg *msgs)
 	else
 		cmd |= SW_TWSI_OP_7;
 
-	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI);
+	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI(i2c));
 	ret = octeon_i2c_hlc_wait(i2c);
 	if (ret)
 		goto err;
 
-	cmd = __raw_readq(i2c->twsi_base + SW_TWSI);
+	cmd = __raw_readq(i2c->twsi_base + SW_TWSI(i2c));
 	if ((cmd & SW_TWSI_R) == 0)
 		return -EAGAIN;
 
@@ -456,7 +456,7 @@ static int octeon_i2c_hlc_read(struct octeon_i2c *i2c, struct i2c_msg *msgs)
 		msgs[0].buf[j] = (cmd >> (8 * i)) & 0xff;
 
 	if (msgs[0].len > 4) {
-		cmd = __raw_readq(i2c->twsi_base + SW_TWSI_EXT);
+		cmd = __raw_readq(i2c->twsi_base + SW_TWSI_EXT(i2c));
 		for (i = 0; i  < msgs[0].len - 4 && i < 4; i++, j--)
 			msgs[0].buf[j] = (cmd >> (8 * i)) & 0xff;
 	}
@@ -493,15 +493,15 @@ static int octeon_i2c_hlc_write(struct octeon_i2c *i2c, struct i2c_msg *msgs)
 
 		for (i = 0; i < msgs[0].len - 4 && i < 4; i++, j--)
 			ext |= (u64) msgs[0].buf[j] << (8 * i);
-		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT);
+		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT(i2c));
 	}
 
-	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI);
+	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI(i2c));
 	ret = octeon_i2c_hlc_wait(i2c);
 	if (ret)
 		goto err;
 
-	cmd = __raw_readq(i2c->twsi_base + SW_TWSI);
+	cmd = __raw_readq(i2c->twsi_base + SW_TWSI(i2c));
 	if ((cmd & SW_TWSI_R) == 0)
 		return -EAGAIN;
 
@@ -536,18 +536,18 @@ static int octeon_i2c_hlc_comp_read(struct octeon_i2c *i2c, struct i2c_msg *msgs
 		cmd |= SW_TWSI_EIA;
 		ext = (u64) msgs[0].buf[0] << SW_TWSI_IA_SHIFT;
 		cmd |= (u64) msgs[0].buf[1] << SW_TWSI_IA_SHIFT;
-		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT);
+		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT(i2c));
 	} else
 		cmd |= (u64) msgs[0].buf[0] << SW_TWSI_IA_SHIFT;
 
 	octeon_i2c_hlc_int_clear(i2c);
-	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI);
+	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI(i2c));
 
 	ret = octeon_i2c_hlc_wait(i2c);
 	if (ret)
 		goto err;
 
-	cmd = __raw_readq(i2c->twsi_base + SW_TWSI);
+	cmd = __raw_readq(i2c->twsi_base + SW_TWSI(i2c));
 	if ((cmd & SW_TWSI_R) == 0)
 		return -EAGAIN;
 
@@ -555,7 +555,7 @@ static int octeon_i2c_hlc_comp_read(struct octeon_i2c *i2c, struct i2c_msg *msgs
 		msgs[1].buf[j] = (cmd >> (8 * i)) & 0xff;
 
 	if (msgs[1].len > 4) {
-		cmd = __raw_readq(i2c->twsi_base + SW_TWSI_EXT);
+		cmd = __raw_readq(i2c->twsi_base + SW_TWSI_EXT(i2c));
 		for (i = 0; i  < msgs[1].len - 4 && i < 4; i++, j--)
 			msgs[1].buf[j] = (cmd >> (8 * i)) & 0xff;
 	}
@@ -601,16 +601,16 @@ static int octeon_i2c_hlc_comp_write(struct octeon_i2c *i2c, struct i2c_msg *msg
 		set_ext = true;
 	}
 	if (set_ext)
-		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT);
+		octeon_i2c_writeq_flush(ext, i2c->twsi_base + SW_TWSI_EXT(i2c));
 
 	octeon_i2c_hlc_int_clear(i2c);
-	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI);
+	octeon_i2c_writeq_flush(cmd, i2c->twsi_base + SW_TWSI(i2c));
 
 	ret = octeon_i2c_hlc_wait(i2c);
 	if (ret)
 		goto err;
 
-	cmd = __raw_readq(i2c->twsi_base + SW_TWSI);
+	cmd = __raw_readq(i2c->twsi_base + SW_TWSI(i2c));
 	if ((cmd & SW_TWSI_R) == 0)
 		return -EAGAIN;
 
drivers/i2c/busses/i2c-cavium.h +18 -17
index 33c7e1f..ad57b03 100644
@@ -7,17 +7,6 @@
 #include <linux/kernel.h>
 #include <linux/pci.h>
 
-/* Register offsets */
-#if IS_ENABLED(CONFIG_I2C_THUNDERX)
-	#define SW_TWSI			0x1000
-	#define TWSI_INT		0x1010
-	#define SW_TWSI_EXT		0x1018
-#else
-	#define SW_TWSI			0x00
-	#define TWSI_INT		0x10
-	#define SW_TWSI_EXT		0x18
-#endif
-
 /* Controller command patterns */
 #define SW_TWSI_V		BIT_ULL(63)	/* Valid bit */
 #define SW_TWSI_EIA		BIT_ULL(61)	/* Extended internal address */
@@ -97,9 +86,21 @@
 
 #define I2C_OCTEON_EVENT_WAIT 80 /* microseconds */
 
+/* Register offsets */
+struct octeon_i2c_reg_offset {
+	unsigned int sw_twsi;
+	unsigned int twsi_int;
+	unsigned int sw_twsi_ext;
+};
+
+#define SW_TWSI(x)	(x->roff.sw_twsi)
+#define TWSI_INT(x)	(x->roff.twsi_int)
+#define SW_TWSI_EXT(x)	(x->roff.sw_twsi_ext)
+
 struct octeon_i2c {
 	wait_queue_head_t queue;
 	struct i2c_adapter adap;
+	struct octeon_i2c_reg_offset roff;
 	struct clk *clk;
 	int irq;
 	int hlc_irq;		/* For cn7890 only */
@@ -139,9 +140,9 @@ static inline void octeon_i2c_reg_write(struct octeon_i2c *i2c, u64 eop_reg, u8
 {
 	u64 tmp;
 
-	__raw_writeq(SW_TWSI_V | eop_reg | data, i2c->twsi_base + SW_TWSI);
+	__raw_writeq(SW_TWSI_V | eop_reg | data, i2c->twsi_base + SW_TWSI(i2c));
 	do {
-		tmp = __raw_readq(i2c->twsi_base + SW_TWSI);
+		tmp = __raw_readq(i2c->twsi_base + SW_TWSI(i2c));
 	} while ((tmp & SW_TWSI_V) != 0);
 }
 
@@ -163,9 +164,9 @@ static inline u8 octeon_i2c_reg_read(struct octeon_i2c *i2c, u64 eop_reg)
 {
 	u64 tmp;
 
-	__raw_writeq(SW_TWSI_V | eop_reg | SW_TWSI_R, i2c->twsi_base + SW_TWSI);
+	__raw_writeq(SW_TWSI_V | eop_reg | SW_TWSI_R, i2c->twsi_base + SW_TWSI(i2c));
 	do {
-		tmp = __raw_readq(i2c->twsi_base + SW_TWSI);
+		tmp = __raw_readq(i2c->twsi_base + SW_TWSI(i2c));
 	} while ((tmp & SW_TWSI_V) != 0);
 
 	return tmp & 0xFF;
@@ -186,7 +187,7 @@ static inline u8 octeon_i2c_reg_read(struct octeon_i2c *i2c, u64 eop_reg)
  */
 static inline u64 octeon_i2c_read_int(struct octeon_i2c *i2c)
 {
-	return __raw_readq(i2c->twsi_base + TWSI_INT);
+	return __raw_readq(i2c->twsi_base + TWSI_INT(i2c));
 }
 
 /**
@@ -196,7 +197,7 @@ static inline u64 octeon_i2c_read_int(struct octeon_i2c *i2c)
  */
 static inline void octeon_i2c_write_int(struct octeon_i2c *i2c, u64 data)
 {
-	octeon_i2c_writeq_flush(data, i2c->twsi_base + TWSI_INT);
+	octeon_i2c_writeq_flush(data, i2c->twsi_base + TWSI_INT(i2c));
 }
 
 /* Prototypes */
drivers/i2c/busses/i2c-octeon-core.c +4 -0
index fec4306..038007d 100644
@@ -163,6 +163,10 @@ static int octeon_i2c_probe(struct platform_device *pdev)
 	}
 	i2c->dev = &pdev->dev;
 
+	i2c->roff.sw_twsi = 0x00;
+	i2c->roff.twsi_int = 0x10;
+	i2c->roff.sw_twsi_ext = 0x18;
+
 	res_mem = platform_get_resource(pdev, IORESOURCE_MEM, 0);
 	i2c->twsi_base = devm_ioremap_resource(&pdev->dev, res_mem);
 	if (IS_ERR(i2c->twsi_base)) {
drivers/i2c/busses/i2c-thunderx-core.c +4 -0
index 656b405..8fcd26c 100644
@@ -134,6 +134,10 @@ static int thunder_i2c_probe_pci(struct pci_dev *pdev,
 	if (!i2c)
 		return -ENOMEM;
 
+	i2c->roff.sw_twsi = 0x1000;
+	i2c->roff.twsi_int = 0x1010;
+	i2c->roff.sw_twsi_ext = 0x1018;
+
 	i2c->dev = dev;
 	pci_set_drvdata(pdev, i2c);
 	ret = pci_enable_device(pdev);
Signature · 1 line
1.9.1
#37 lore raw

PATCH v7 15/15 i2c: thunderx: Add smbus alert support

In reply to Jan Glauber

Add smbus alert interrupt support.

---
 drivers/i2c/busses/i2c-cavium.h        |  6 ++++++
 drivers/i2c/busses/i2c-thunderx-core.c | 35 ++++++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+)
drivers/i2c/busses/i2c-cavium.h +6 -0
index ad57b03..749dffc 100644
@@ -3,6 +3,7 @@
 #include <linux/delay.h>
 #include <linux/device.h>
 #include <linux/i2c.h>
+#include <linux/i2c-smbus.h>
 #include <linux/io.h>
 #include <linux/kernel.h>
 #include <linux/pci.h>
@@ -120,6 +121,11 @@ struct octeon_i2c {
 #if IS_ENABLED(CONFIG_I2C_THUNDERX)
 	struct msix_entry i2c_msix;
 #endif
+
+#if IS_ENABLED(CONFIG_I2C_SMBUS)
+	struct i2c_smbus_alert_setup alert_data;
+	struct i2c_client *ara;
+#endif
 };
 
 static inline void octeon_i2c_writeq_flush(u64 val, void __iomem *addr)
drivers/i2c/busses/i2c-thunderx-core.c +35 -0
index 8fcd26c..e1a111d 100644
@@ -9,9 +9,11 @@
 #include <linux/clk.h>
 #include <linux/delay.h>
 #include <linux/i2c.h>
+#include <linux/i2c-smbus.h>
 #include <linux/interrupt.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
+#include <linux/of_irq.h>
 #include <linux/pci.h>
 
 #include "i2c-cavium.h"
@@ -106,6 +108,35 @@ static void thunder_i2c_clock_disable(struct device *dev, struct clk *clk)
 	devm_clk_put(dev, clk);
 }
 
+static int thunder_i2c_smbus_setup(struct octeon_i2c *i2c,
+				   struct device_node *node)
+{
+#if IS_ENABLED(CONFIG_I2C_SMBUS)
+	u32 type;
+
+	i2c->alert_data.irq = irq_of_parse_and_map(node, 0);
+	if (!i2c->alert_data.irq)
+		return -EINVAL;
+
+	type = irqd_get_trigger_type(irq_get_irq_data(i2c->alert_data.irq));
+	i2c->alert_data.alert_edge_triggered =
+		(type & IRQ_TYPE_LEVEL_MASK) ? 1 : 0;
+
+	i2c->ara = i2c_setup_smbus_alert(&i2c->adap, &i2c->alert_data);
+	if (!i2c->ara)
+		return -ENODEV;
+#endif
+	return 0;
+}
+
+static void thunder_i2c_smbus_remove(struct octeon_i2c *i2c)
+{
+#if IS_ENABLED(CONFIG_I2C_SMBUS)
+	if (i2c->ara)
+		i2c_unregister_device(i2c->ara);
+#endif
+}
+
 static void thunder_i2c_set_name(struct pci_dev *pdev, struct octeon_i2c *i2c,
 				 char *name)
 {
@@ -210,6 +241,9 @@ static int thunder_i2c_probe_pci(struct pci_dev *pdev,
 		goto out_irq;
 	}
 
+	ret = thunder_i2c_smbus_setup(i2c, node);
+	if (ret < 0)
+		dev_info(dev, "Failed to setup smbus alert\n");
 	dev_info(i2c->dev, "probed\n");
 	return 0;
 
@@ -240,6 +274,7 @@ static void thunder_i2c_remove_pci(struct pci_dev *pdev)
 
 	dev = i2c->dev;
 	thunder_i2c_clock_disable(dev, i2c->clk);
+	thunder_i2c_smbus_remove(i2c);
 	i2c_del_adapter(&i2c->adap);
 	devm_free_irq(dev, i2c->i2c_msix.vector, i2c);
 	pci_disable_msix(pdev);
Signature · 1 line
1.9.1