Skip to contentSkip to conversation

Blackfin: blackfin i2c driver

PATCH In review Wu, Bryan 7 participants 19 messages Mar 6, 2007 – Mar 9, 2007 I2C subsystem host drivers

lore
Share & tools
lore

Wu, Bryan

author
#0 lore raw

PATCH 1/1 Blackfin: blackfin i2c driver

Hi folks,

[PATCH] Blackfin: blackfin i2c driver

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

---
 drivers/i2c/busses/Kconfig         |   47 ++++
 drivers/i2c/busses/i2c-bfin-gpio.c |   98 ++++++++++
 drivers/i2c/busses/i2c-bfin-twi.c  |  530 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 675 insertions(+)
drivers/i2c/busses/Kconfig +47 -0
Index: linux-2.6/drivers/i2c/busses/Kconfig
===================================================================
@@ -5,6 +5,53 @@
 menu "I2C Hardware Bus support"
 	depends on I2C
 
+config I2C_BFIN_GPIO
+	tristate "Generic Blackfin and HHBF533/561 development board I2C support"
+	depends on I2C && EXPERIMENTAL
+	select I2C_ALGOBIT
+	help
+	--
+
+menu "BFIN I2C SDA/SCL Selection"
+	depends on I2C_BFIN_GPIO
+config BFIN_SDA
+	int "SDA is GPIO Number"
+	range 0 15 if (BF533 || BF532 || BF531) 
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 2 if (BF533 || BF532 || BF531) 
+
+config BFIN_SCL
+	int "SCL is GPIO Number"
+	range 0 15 if (BF533 || BF532 || BF531) 
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 3 
+endmenu
+
+config I2C_BFIN_GPIO_CYCLE_DELAY
+	int "Cycle Delay in usec"
+	depends on I2C_BFIN_GPIO
+	range 1 100 
+	default 40
+
+config I2C_BFIN_TWI
+	tristate "Blackfin TWI I2C support"
+	depends on I2C && (BF534 || BF536 || BF537)
+	help
+	  This the TWI I2C device driver for Blackfin 534/536/537.
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-bfin-twi.
+
+config TWICLK_KHZ
+	int "TWI clock (kHZ)"
+	depends on I2C_BFIN_TWI
+	default 50
+	help
+		The unit of the TWI clock is kilo HZ. Please divide the clock 
+		by 1024 if you count it in HZ. The value should be less than 400.
+
 config I2C_ALI1535
 	tristate "ALI 1535"
 	depends on I2C && PCI
drivers/i2c/busses/i2c-bfin-gpio.c new file +98 -0

98 lines not on this page · raw message

drivers/i2c/busses/i2c-bfin-twi.c new file +530 -0

530 lines not on this page · raw message

_

Thanks
-Bryan Wu

#1 lore raw
Quoted 7 lines · Bryan

Hi folks,

[PATCH] Blackfin: blackfin i2c driver

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

Little things...

drivers/i2c/busses/i2c-bfin-gpio.c line 76

Quoted 12 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +12 -0
+static int __init i2c_hhbf_init(void)
+{
+
+    if(gpio_request(CONFIG_BFIN_SCL, NULL)) {
+    	printk(KERN_ERR "%s: gpio_request GPIO %d failed \n",__FUNCTION__, CONFIG_BFIN_SCL);
+	return -1;
+	}
+
+    if(gpio_request(CONFIG_BFIN_SDA, NULL)) {
+    	printk(KERN_ERR "%s: gpio_request GPIO %d failed \n",__FUNCTION__, CONFIG_BFIN_SDA);
+	return -1;
+	}

whitespace breakage there

drivers/i2c/busses/i2c-bfin-gpio.c

Quoted 12 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +12 -0
+
+    gpio_direction_output(CONFIG_BFIN_SCL);
+    gpio_direction_input(CONFIG_BFIN_SDA);
+    gpio_set_value(CONFIG_BFIN_SCL, 1);    
+
+    return i2c_bit_add_bus(&hhbf_ops);
+}
+
+#define TWI_I2C_MODE_COMBINED		0x04
+
+struct bfin_twi_iface
+{
	struct bfin_twi_iface {

drivers/i2c/busses/i2c-bfin-twi.c line 38

Quoted 1 line
drivers/i2c/busses/i2c-bfin-twi.c new file +1 -0
+	struct semaphore 	twi_lock;

Can this be converted to a mutex?

drivers/i2c/busses/i2c-bfin-twi.c line 187

Quoted 9 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +6 -0
+	int			irq;
+	spinlock_t		lock;
+ */
 
...
 
+static int bfin_twi_master_xfer(struct i2c_adapter *adap, struct i2c_msg msgs[], int num)
+{
+	struct bfin_twi_iface* iface = (struct bfin_twi_iface*)adap->algo_data;

This code has zillions of unneeded casts of void*

Quoted 4 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +4 -0
+	struct i2c_msg *pmsg;
+	int i, ret;
+	int rc = 0;
+

drivers/i2c/busses/i2c-bfin-twi.c line 201

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+	if (!(bfin_read_TWI_CONTROL() & TWI_ENA))
+		return -ENXIO;
+
+	down(&iface->twi_lock);
+
+	while (bfin_read_TWI_MASTER_STAT() & BUSBUSY) {
+		up(&iface->twi_lock);
+		schedule();
+		down(&iface->twi_lock);
+	}

That's a busy loop until this task's timeslice has expired. It'll work, but it'll suck a bit. (Repeated in several places)

drivers/i2c/busses/i2c-bfin-twi.c line 205

Quoted 3 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +3 -0
+	ret = 0;
+	for (i = 0; rc >= 0 && i < num;) {
+		pmsg = &msgs[i++];

Strange. Why not do the i++ in the `for' statement?

drivers/i2c/busses/i2c-bfin-twi.c line 383

Quoted 13 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+		if (pmsg->flags & I2C_M_TEN) {
+			printk(KERN_ERR "i2c-bfin-twi: 10 bits addr not supported !\n");
+			rc = -EINVAL;
+			break;
+		}
+
 
...
 
+	switch (iface->cur_mode) {
+	case TWI_I2C_MODE_STANDARDSUB:
+		bfin_write_TWI_XMT_DATA8(iface->command);
+		bfin_write_TWI_INT_MASK(MCOMP | MERR | ((iface->read_write == I2C_SMBUS_READ) ? RCVSERV : XMTSERV));

It's preferred if code is readable in an 80-col display.

Quoted 4 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +4 -0
+		SSYNC();
+
+		if (iface->writeNum + 1 <= 255)
+			bfin_write_TWI_MASTER_CTL(((iface->writeNum+1) << 6));
#2 lore raw
Quoted 16 lines · Andrew Morton
Quoted 14 lines · Bryan

Hi folks,

[PATCH] Blackfin: blackfin i2c driver

drivers/i2c/busses/i2c-bfin-twi.c new file +9 -0
+     struct i2c_msg *pmsg;
+     int i, ret;
+     int rc = 0;
+
+     if (!(bfin_read_TWI_CONTROL() & TWI_ENA))
+             return -ENXIO;
+
+     down(&iface->twi_lock);
+
Quoted 9 lines

drivers/i2c/busses/i2c-bfin-twi.c line 201

Quoted 5 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +5 -0
+     while (bfin_read_TWI_MASTER_STAT() & BUSBUSY) {
+             up(&iface->twi_lock);
+             schedule();
+             down(&iface->twi_lock);
+     }

That's a busy loop until this task's timeslice has expired. It'll work, but it'll suck a bit. (Repeated in several places)

OK, I change it into yield(). So, current process will be move to the tail of the run queue. Is that OK with you?

#3 lore raw
Quoted 27 lines · On Wed, 7 Mar 2007 13:17:57 +0800 "Sonic Zhang" <[email protected]> wrote:
Quoted 16 lines · Andrew Morton
Quoted 14 lines · Bryan

Hi folks,

[PATCH] Blackfin: blackfin i2c driver

drivers/i2c/busses/i2c-bfin-twi.c new file +9 -0
+     struct i2c_msg *pmsg;
+     int i, ret;
+     int rc = 0;
+
+     if (!(bfin_read_TWI_CONTROL() & TWI_ENA))
+             return -ENXIO;
+
+     down(&iface->twi_lock);
+
Quoted 9 lines

drivers/i2c/busses/i2c-bfin-twi.c line 201

Quoted 5 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +5 -0
+     while (bfin_read_TWI_MASTER_STAT() & BUSBUSY) {
+             up(&iface->twi_lock);
+             schedule();
+             down(&iface->twi_lock);
+     }

That's a busy loop until this task's timeslice has expired. It'll work, but it'll suck a bit. (Repeated in several places)

Quoted 2 lines

OK, I change it into yield(). So, current process will be move to the tail of the run queue. Is that OK with you?

Nope, yield is terribly bad when there are busy processes running: it can stall for a very long time indeed,

Is this hardware not capable of generating an interrupt when BUSBUSY gets negated?

I guess not, in which case you're stuck with having to poll it - probably use a cond_resched() in the loop, and an angry comment.

author
#4 lore raw
Quoted 4 lines
Quoted 3 lines

OK, I change it into yield(). So, current process will be move to the tail of the run queue. Is that OK with you?

Quoted 8 lines

Nope, yield is terribly bad when there are busy processes running: it can stall for a very long time indeed,

Is this hardware not capable of generating an interrupt when BUSBUSY gets negated?

I guess not, in which case you're stuck with having to poll it - probably use a cond_resched() in the loop, and an angry comment.

Thanks, we fix it. please pick up this one.

[PATCH] Blackfin: blackfin i2c driver

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

---

 drivers/i2c/busses/Kconfig         |   47 ++++
 drivers/i2c/busses/i2c-bfin-gpio.c |   98 +++++++++
 drivers/i2c/busses/i2c-bfin-twi.c  |  589 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 734 insertions(+)
drivers/i2c/busses/Kconfig +47 -0
Index: linux-2.6/drivers/i2c/busses/Kconfig
===================================================================
@@ -5,6 +5,53 @@
 menu "I2C Hardware Bus support"
 	depends on I2C
 
+config I2C_BFIN_GPIO
+	tristate "Generic Blackfin and HHBF533/561 development board I2C support"
+	depends on I2C && EXPERIMENTAL
+	select I2C_ALGOBIT
+	help
+	--
+
+menu "BFIN I2C SDA/SCL Selection"
+	depends on I2C_BFIN_GPIO
+config BFIN_SDA
+	int "SDA is GPIO Number"
+	range 0 15 if (BF533 || BF532 || BF531) 
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 2 if (BF533 || BF532 || BF531) 
+
+config BFIN_SCL
+	int "SCL is GPIO Number"
+	range 0 15 if (BF533 || BF532 || BF531) 
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 3 
+endmenu
+
+config I2C_BFIN_GPIO_CYCLE_DELAY
+	int "Cycle Delay in usec"
+	depends on I2C_BFIN_GPIO
+	range 1 100 
+	default 40
+
+config I2C_BFIN_TWI
+	tristate "Blackfin TWI I2C support"
+	depends on I2C && (BF534 || BF536 || BF537)
+	help
+	  This the TWI I2C device driver for Blackfin 534/536/537.
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-bfin-twi.
+
+config TWICLK_KHZ
+	int "TWI clock (kHZ)"
+	depends on I2C_BFIN_TWI
+	default 50
+	help
+	  The unit of the TWI clock is kilo HZ. Please divide the clock 
+	  by 1024 if you count it in HZ. The value should be less than 400.
+
 config I2C_ALI1535
 	tristate "ALI 1535"
 	depends on I2C && PCI
drivers/i2c/busses/i2c-bfin-gpio.c new file +98 -0

98 lines not on this page · raw message

drivers/i2c/busses/i2c-bfin-twi.c new file +589 -0

589 lines not on this page · raw message

_

Best Regards,
-Bryan Wu

#5 lore raw
In reply to Wu, Bryan

drivers/i2c/busses/Kconfig line 53

Quoted 12 lines · Bryan

[PATCH] Blackfin: blackfin i2c driver

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

drivers/i2c/busses/Kconfig +7 -0
+config TWICLK_KHZ
+	int "TWI clock (kHZ)"
+	depends on I2C_BFIN_TWI
+	default 50
+	help
+		The unit of the TWI clock is kilo HZ. Please divide the clock 
+		by 1024 if you count it in HZ. The value should be less than 400.

Indentation in Kconfig files is Tab-Space-Space.

drivers/i2c/busses/i2c-bfin-gpio.c line 69

Quoted 8 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +5 -0
 
+static int __init i2c_hhbf_init(void)
+{
+
+    if(gpio_request(CONFIG_BFIN_SCL, NULL)) {
+    	printk(KERN_ERR "%s: gpio_request GPIO %d failed \n",__FUNCTION__, CONFIG_BFIN_SCL);

I may be first to ask this, but, please, use __func__ in all new code. __FUNCTION__ is gcc-2.95-ism, which is no longer supported.

drivers/i2c/busses/i2c-bfin-gpio.c line 74

Quoted 5 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +5 -0
+	return -1;
+	}
+
+    if(gpio_request(CONFIG_BFIN_SDA, NULL)) {
+    	printk(KERN_ERR "%s: gpio_request GPIO %d failed \n",__FUNCTION__, CONFIG_BFIN_SDA);

__func__, Do you need to gpio_free() ?

drivers/i2c/busses/i2c-bfin-gpio.c line 83

Quoted 9 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +9 -0
+	return -1;
+	}
+
+
+    gpio_direction_output(CONFIG_BFIN_SCL);
+    gpio_direction_input(CONFIG_BFIN_SDA);
+    gpio_set_value(CONFIG_BFIN_SCL, 1);    
+
+    return i2c_bit_add_bus(&hhbf_ops);

Do you need unwinding with gpio_free() like below?

drivers/i2c/busses/i2c-bfin-gpio.c line 89

Quoted 4 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +4 -0
+static void __exit i2c_hhbf_exit(void)
+{
+    gpio_free(CONFIG_BFIN_SCL);
+    gpio_free(CONFIG_BFIN_SDA);    

trailing whitespace.

drivers/i2c/busses/i2c-bfin-twi.c line 62

Quoted 12 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +2 -0
+    i2c_bit_del_bus(&hhbf_ops);
+}
drivers/i2c/busses/i2c-bfin-twi.c new file +6 -0
 
+static void bfin_twi_handle_interrupt(struct bfin_twi_iface *iface)
+{
+	unsigned short twi_int_stat = bfin_read_TWI_INT_STAT();
+	unsigned short mast_stat = bfin_read_TWI_MASTER_STAT();
+
+	if (XMTSERV & twi_int_stat) {

Please, put bitmasks on the right. Much easier to read. See below for other places.

if (twi_int_stat & XMTSERV)

Also, I'd rename variable to twi_intr_status, to not confuse with statistics or something.

drivers/i2c/busses/i2c-bfin-twi.c line 66

Quoted 4 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +4 -0
+		/* Transmit next data */
+		if (iface->writeNum > 0) {
+			bfin_write_TWI_XMT_DATA8(*(iface->transPtr++));
+			iface->writeNum--;

StudlyCaps.

Quoted 25 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +24 -0
+static int bfin_twi_master_xfer(struct i2c_adapter *adap, struct i2c_msg msgs[], int num)
+{
+	struct bfin_twi_iface* iface = (struct bfin_twi_iface*)adap->algo_data;
+	struct i2c_msg *pmsg;
+	int i, ret;
+	int rc = 0;
+
+	if (!(bfin_read_TWI_CONTROL() & TWI_ENA))
+		return -ENXIO;
+
+	down(&iface->twi_lock);
+
+	while (bfin_read_TWI_MASTER_STAT() & BUSBUSY) {
+		up(&iface->twi_lock);
+		schedule();
+		down(&iface->twi_lock);
+	}
 
+int bfin_twi_smbus_xfer(struct i2c_adapter *adap, u16 addr,
+			unsigned short flags, char read_write,
+			u8 command, int size, union i2c_smbus_data * data)
+{
+	struct bfin_twi_iface* iface = (struct bfin_twi_iface*)adap->algo_data;
+	int rc = 0;
+

drivers/i2c/busses/i2c-bfin-twi.c line 296

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+	if (!(bfin_read_TWI_CONTROL() & TWI_ENA))
+		return -ENXIO;
+
+	down(&iface->twi_lock);
+
+	while(bfin_read_TWI_MASTER_STAT() & BUSBUSY) {
+		up(&iface->twi_lock);
+		schedule();
+		down(&iface->twi_lock);
+	}

Smells like schedule() abuse for delays.

drivers/i2c/busses/i2c-bfin-twi.c line 387

Quoted 8 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +8 -0
+	switch (iface->cur_mode) {
+	case TWI_I2C_MODE_STANDARDSUB:
+		bfin_write_TWI_XMT_DATA8(iface->command);
+		bfin_write_TWI_INT_MASK(MCOMP | MERR | ((iface->read_write == I2C_SMBUS_READ) ? RCVSERV : XMTSERV));
+		SSYNC();
+
+		if (iface->writeNum + 1 <= 255)
+			bfin_write_TWI_MASTER_CTL(((iface->writeNum+1) << 6));

Too many ():

			bfin_write_TWI_MASTER_CTL((iface->writeNum + 1) << 6);

drivers/i2c/busses/i2c-bfin-twi.c

Quoted 2 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +2 -0
+		else {
+			bfin_write_TWI_MASTER_CTL(( 0xff << 6 ));

ditto

drivers/i2c/busses/i2c-bfin-twi.c line 401

Quoted 12 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +12 -0
+			iface->manual_stop = 1;
+		}
+		/* Master enable */
+		bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL() | MEN | ((CONFIG_TWICLK_KHZ>100) ? FAST : 0));
+		break;
+	case TWI_I2C_MODE_COMBINED:
+		bfin_write_TWI_XMT_DATA8(iface->command);
+		bfin_write_TWI_INT_MASK(MCOMP | MERR | RCVSERV | XMTSERV);
+		SSYNC();
+
+		if (iface->writeNum > 0)
+			bfin_write_TWI_MASTER_CTL(((iface->writeNum+1) << 6));

ditto

drivers/i2c/busses/i2c-bfin-twi.c line 403

Quoted 2 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +2 -0
+		else
+			bfin_write_TWI_MASTER_CTL(( 0x1 << 6 ));

ditto, etc

Quoted 13 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +12 -0
+static int __init i2c_bfin_twi_init(void)
+{
 
+	rc = request_irq(twi_iface.irq, bfin_twi_interrupt_entry, SA_INTERRUPT, "i2c-bfin-twi", &twi_iface);
+	if (rc) {
+		printk(KERN_ERR "i2c-bfin-twi: can't get IRQ %d !\n", twi_iface.irq);
+		return -ENODEV;
+	}
+
+	/* Set TWI internal clock as 10MHz */
+	bfin_write_TWI_CONTROL(((get_sclk() / 1024 / 1024 + 5) / 10) & 0x7F);
+
+	/* Set Twi interface clock as specified */

drivers/i2c/busses/i2c-bfin-twi.c line 516

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+	if (CONFIG_TWICLK_KHZ > 400)
+		bfin_write_TWI_CLKDIV((( 5*1024 / 400 ) << 8) | (( 5*1024 / 400 ) & 0xFF));
+	else
+		bfin_write_TWI_CLKDIV((( 5*1024 / CONFIG_TWICLK_KHZ ) << 8) | (( 5*1024 / CONFIG_TWICLK_KHZ ) & 0xFF));
+
+	/* Enable TWI */
+	bfin_write_TWI_CONTROL(bfin_read_TWI_CONTROL() | TWI_ENA);
+	SSYNC();
+
+	return i2c_add_adapter(p_adap);

free_irq on error?

#6 lore raw
Quoted 16 lines · Alexey Dobriyan
Quoted 16 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +16 -0
+static int __init i2c_bfin_twi_init(void)
+{
+     rc = request_irq(twi_iface.irq, bfin_twi_interrupt_entry, SA_INTERRUPT, "i2c-bfin-twi", &twi_iface);
+     if (rc) {
+             printk(KERN_ERR "i2c-bfin-twi: can't get IRQ %d !\n", twi_iface.irq);
+             return -ENODEV;
+     }
+
+     /* Set TWI internal clock as 10MHz */
+     bfin_write_TWI_CONTROL(((get_sclk() / 1024 / 1024 + 5) / 10) & 0x7F);
+
+     /* Set Twi interface clock as specified */
+     if (CONFIG_TWICLK_KHZ > 400)
+             bfin_write_TWI_CLKDIV((( 5*1024 / 400 ) << 8) | (( 5*1024 / 400 ) & 0xFF));
+     else
+             bfin_write_TWI_CLKDIV((( 5*1024 / CONFIG_TWICLK_KHZ ) << 8) | (( 5*1024 / CONFIG_TWICLK_KHZ ) & 0xFF));
Quoted 8 lines

drivers/i2c/busses/i2c-bfin-twi.c line 516

Quoted 6 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +6 -0
+
+     /* Enable TWI */
+     bfin_write_TWI_CONTROL(bfin_read_TWI_CONTROL() | TWI_ENA);
+     SSYNC();
+
+     return i2c_add_adapter(p_adap);

free_irq on error?

what error ? do you mean i2c_add_adapter() ?
-mike

author
#7 lore raw
In reply to Alexey Dobriyan

Dear Andrew and Alexey:

Thanks a lot for the review.

Here is the updated blackfin i2c driver.

[PATCH] Blackfin: blackfin i2c driver

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

---
 drivers/i2c/busses/Kconfig         |   47 ++++
 drivers/i2c/busses/i2c-bfin-gpio.c |   98 +++++++++
 drivers/i2c/busses/i2c-bfin-twi.c  |  589 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 734 insertions(+)
drivers/i2c/busses/Kconfig +47 -0
Index: linux-2.6/drivers/i2c/busses/Kconfig
===================================================================
@@ -5,6 +5,53 @@
 menu "I2C Hardware Bus support"
 	depends on I2C
 
+config I2C_BFIN_GPIO
+	tristate "Generic Blackfin and HHBF533/561 development board I2C support"
+	depends on I2C && EXPERIMENTAL
+	select I2C_ALGOBIT
+	help
+	--
+
+menu "BFIN I2C SDA/SCL Selection"
+	depends on I2C_BFIN_GPIO
+config BFIN_SDA
+	int "SDA is GPIO Number"
+	range 0 15 if (BF533 || BF532 || BF531) 
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 2 if (BF533 || BF532 || BF531) 
+
+config BFIN_SCL
+	int "SCL is GPIO Number"
+	range 0 15 if (BF533 || BF532 || BF531) 
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 3 
+endmenu
+
+config I2C_BFIN_GPIO_CYCLE_DELAY
+	int "Cycle Delay in usec"
+	depends on I2C_BFIN_GPIO
+	range 1 100 
+	default 40
+
+config I2C_BFIN_TWI
+	tristate "Blackfin TWI I2C support"
+	depends on I2C && (BF534 || BF536 || BF537)
+	help
+	  This the TWI I2C device driver for Blackfin 534/536/537.
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-bfin-twi.
+
+config TWICLK_KHZ
+	int "TWI clock (kHZ)"
+	depends on I2C_BFIN_TWI
+	default 50
+	help
+	  The unit of the TWI clock is kilo HZ. Please divide the clock 
+	  by 1024 if you count it in HZ. The value should be less than 400.
+
 config I2C_ALI1535
 	tristate "ALI 1535"
 	depends on I2C && PCI
drivers/i2c/busses/i2c-bfin-gpio.c new file +98 -0

98 lines not on this page · raw message

drivers/i2c/busses/i2c-bfin-twi.c new file +589 -0

589 lines not on this page · raw message

_

Thanks a lot
-Bryan Wu

#8 lore raw

Hi Bryan,

Quoted 4 lines · Bryan

Here is the updated blackfin i2c driver.

[PATCH] Blackfin: blackfin i2c driver

Quoted 8 lines

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

---
 drivers/i2c/busses/Kconfig         |   47 ++++
 drivers/i2c/busses/i2c-bfin-gpio.c |   98 +++++++++
 drivers/i2c/busses/i2c-bfin-twi.c  |  589 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++

I'd prefer i2c-blackfin-gpio and i2c-blackfin-twi. Abreviations tend to confuse newcomers.

drivers/i2c/busses/Kconfig line 8

Quoted 11 lines
 3 files changed, 734 insertions(+)

Index: linux-2.6/drivers/i2c/busses/Kconfig
===================================================================

drivers/i2c/busses/Kconfig +1 -0
@@ -5,6 +5,53 @@
 menu "I2C Hardware Bus support"
 	depends on I2C
 
+config I2C_BFIN_GPIO

I2C_BLACKFIN_GPIO

Please move the entries to the right location. The list is sorted alphabetically if you didn't notice.

drivers/i2c/busses/Kconfig line 9

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	tristate "Generic Blackfin and HHBF533/561 development board I2C support"

You can drop the trailing "I2C support", the user is in a menu named "I2C hardware bus support" so it's pretty clear what we're talking about.

drivers/i2c/busses/Kconfig line 17

Quoted 8 lines
drivers/i2c/busses/Kconfig +8 -0
+	depends on I2C && EXPERIMENTAL
+	select I2C_ALGOBIT
+	help
+	--
+
+menu "BFIN I2C SDA/SCL Selection"
+	depends on I2C_BFIN_GPIO
+config BFIN_SDA

I2C_BLACKFIN_SDA

drivers/i2c/busses/Kconfig line 18

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	int "SDA is GPIO Number"

"SDA GPIO pin number"

drivers/i2c/busses/Kconfig line 19

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	range 0 15 if (BF533 || BF532 || BF531) 

Trailing whitespace.

drivers/i2c/busses/Kconfig line 22

Quoted 3 lines
drivers/i2c/busses/Kconfig +3 -0
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 2 if (BF533 || BF532 || BF531) 

Trailing whitespace.

No default for the other cases?

drivers/i2c/busses/Kconfig line 24

Quoted 2 lines
drivers/i2c/busses/Kconfig +2 -0
+
+config BFIN_SCL

I2C_BLACKFIN_SCL
Etc etc, all the options should start with I2C_BLACKFIN.

drivers/i2c/busses/Kconfig line 25

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	int "SCL is GPIO Number"

"SCL GPIO pin number"

drivers/i2c/busses/Kconfig line 26

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	range 0 15 if (BF533 || BF532 || BF531) 

Trailing whitespace, and many more after that. Please fix them all!

drivers/i2c/busses/Kconfig line 36

Quoted 10 lines
drivers/i2c/busses/Kconfig +10 -0
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 3 
+endmenu
+
+config I2C_BFIN_GPIO_CYCLE_DELAY
+	int "Cycle Delay in usec"
+	depends on I2C_BFIN_GPIO
+	range 1 100 
+	default 40

This should really not be a kernel configuration option. Please turn it into a kernel module parameter or a sysfs attribute if you really need it. Also note that we already have an interface to change this value from user-space (using an ioctl on /dev/i2c-N) and that might be sufficient for your needs.

And allowing 1 usec delay is probably not a good idea, I don't recommend values below 6 usec with i2c-algo-bit.

drivers/i2c/busses/Kconfig line 48

Quoted 12 lines
drivers/i2c/busses/Kconfig +12 -0
+
+config I2C_BFIN_TWI
+	tristate "Blackfin TWI I2C support"
+	depends on I2C && (BF534 || BF536 || BF537)
+	help
+	  This the TWI I2C device driver for Blackfin 534/536/537.
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-bfin-twi.
+
+config TWICLK_KHZ
+	int "TWI clock (kHZ)"

kHz

drivers/i2c/busses/Kconfig line 53

Quoted 5 lines
drivers/i2c/busses/Kconfig +5 -0
+	depends on I2C_BFIN_TWI
+	default 50
+	help
+	  The unit of the TWI clock is kilo HZ. Please divide the clock 
+	  by 1024 if you count it in HZ. The value should be less than 400.

Why don't you use "range" here too to ensure that the value is actually less than 400? Either way, same as above, IMHO this should not be a compilation-time decision.

A kHz is really 1000 Hz, not 1024. And everybody skilled enough to configure a kernel should know that, I doubt it's worth reminding.

drivers/i2c/busses/Kconfig line 57

Quoted 4 lines
drivers/i2c/busses/Kconfig +1 -0
+
 config I2C_ALI1535
 	tristate "ALI 1535"
 	depends on I2C && PCI

All these options won't work really well until you also change drivers/i2c/busses/Makefile to make something useful with them...

Signature · 1 line
Jean Delvare
#9 lore raw
Quoted 3 lines · On Wed, 7 Mar 2007 07:58:22 +0100 Jean Delvare <[email protected]> wrote:
Quoted 1 line

+config BFIN_SDA

I2C_BLACKFIN_SDA

The blackfin architecture uses "bfin" pretty much universally, so this usage is consistent.

box:/usr/src/25> grep -i blackfin patches/blackfin*|wc -l
1608
box:/usr/src/25> grep -i bfin patches/blackfin*|wc -l
6198

Let's just hope nobody makes a bluefin.

Quoted 3 lines

Review

Quoted 1 line
(quoted diff) +1 -0
+	range 0 15 if (BF533 || BF532 || BF531) 

Trailing whitespace.

I always remove that when merging a patch.

author
#10 lore raw
Quoted 6 lines · Andrew Morton
Quoted 3 lines · On Wed, 7 Mar 2007 07:58:22 +0100 Jean Delvare <[email protected]> wrote:
Quoted 1 line

+config BFIN_SDA

I2C_BLACKFIN_SDA

Quoted 8 lines

The blackfin architecture uses "bfin" pretty much universally, so this usage is consistent.

box:/usr/src/25> grep -i blackfin patches/blackfin*|wc -l
1608
box:/usr/src/25> grep -i bfin patches/blackfin*|wc -l
6198

Thanks for you understanding, but now we want to move to use CONFIG_BLACKFIN options. There is a new task in our development plan to change things to CONFIG_BLACKFIN.

At this moment, we both provide CONFIG_BFIN and CONFIG_BLACKFIN. When all the things relied on CONFIG_BFIN/bfin are changed to CONFIG_BLACKFIN/blackfin, the CONFIG_BFIN will be removed.

So here I will follow Jean's comments.

Quoted 1 line

Let's just hope nobody makes a bluefin.

So it is ok for both blackfin and bluefin. But I think Black is cooler than Blue. -:)

Quoted 6 lines
Quoted 4 lines

Review

Quoted 2 lines
(quoted diff) +1 -0
+	range 0 15 if (BF533 || BF532 || BF531) 

Trailing whitespace.

I always remove that when merging a patch.

Thanks a lot, could you please give me a script just to kill this whitespace? So I can do it before sending you patches.

Thanks Jean and Andrew.
-Bryan

#11 lore raw
Quoted 2 lines · Bryan

Thanks a lot, could you please give me a script just to kill this whitespace? So I can do it before sending you patches.

Is pretty simple:

#!/bin/sh
#
# Strip any trailing whitespace which a unified diff adds. #

strip1()

{
	TMP=$(mktemp /tmp/XXXXXX)
	cp $1 $TMP
	sed -e '/^+/s/[ 	]*$//' < $TMP > $1
	rm $TMP
}

for i in $*
do
strip1 $i
done

that'll be in
http://www.zip.com.au/~akpm/linux/patches/patch-scripts-0.20/patch-scripts-0.20.tar.gz
too

#12 lore raw

trailing whitespace killing (Re: [PATCH -mm] Blackfin: blackfin i2c driver)

Quoted 4 lines

From: Andrew Morton
Newsgroups: gmane.linux.kernel
Subject: Re: [PATCH -mm] Blackfin: blackfin i2c driver
Date: Tue, 6 Mar 2007 23:45:29 -0800

[]

Quoted 27 lines · Bryan
Quoted 3 lines

Thanks a lot, could you please give me a script just to kill this whitespace? So I can do it before sending you patches.

Is pretty simple:

#!/bin/sh
#
# Strip any trailing whitespace which a unified diff adds. #

strip1()

{
	TMP=$(mktemp /tmp/XXXXXX)
	cp $1 $TMP
	sed -e '/^+/s/[ 	]*$//' < $TMP > $1
	rm $TMP
}

for i in $*
do
strip1 $i
done

Quoted 3 lines

that'll be in
http://www.zip.com.au/~akpm/linux/patches/patch-scripts-0.20/patch-scripts-0.20.tar.gz
too

It doesn't work for me. Maybe i can't understand what you are trying to do, anyway.

General suggestion is can be:

sed -e 's_[ \t]*$__'

(i.e any line on stdin with space/tab mixed tails is stripped on stdout)

You can use it as wrapper for diff, sending patch bombs, etc. (very nice with pipes):

shell$ diff -Npu2 old new | sed -e 's_[ \t]*$__' > patch.diff
shell$ < patch-set.mbox sed -e 's_[ \t]*$__' | formail -s /usr/sbin/sendmail -bm -t

similar in scripts; quilt (patch sets manager) notices about them.
____

#13 lore raw
In reply to Andrew Morton
Quoted 27 lines · Andrew Morton
Quoted 2 lines · Bryan

Thanks a lot, could you please give me a script just to kill this whitespace? So I can do it before sending you patches.

Is pretty simple:

#!/bin/sh
#
# Strip any trailing whitespace which a unified diff adds. #

strip1()

{
	TMP=$(mktemp /tmp/XXXXXX)
	cp $1 $TMP
	sed -e '/^+/s/[ 	]*$//' < $TMP > $1
	rm $TMP
}

for i in $*
do
strip1 $i
done

Quoted 3 lines

that'll be in
http://www.zip.com.au/~akpm/linux/patches/patch-scripts-0.20/patch-scripts-0.20.tar.gz
too

Alternatively, you can use quilt [1] to manage your patches and enable the --strip-trailing-whitespace option by default.

[1] http://savannah.nongnu.org/projects/quilt/

Signature · 1 line
Jean Delvare
author Reviewed-by Andrew Morton Reviewed-by Alexey Dobriyan Reviewed-by Jean Delvare
#14 lore raw
In reply to Jean Delvare
Quoted 10 lines
Quoted 6 lines
---
 drivers/i2c/busses/Kconfig         |   47 ++++
 drivers/i2c/busses/i2c-bfin-gpio.c |   98 +++++++++
 drivers/i2c/busses/i2c-bfin-twi.c  |  589 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++

I'd prefer i2c-blackfin-gpio and i2c-blackfin-twi. Abreviations tend to confuse newcomers.

There are tons of code using bfin abreviations for functions/files/variables names. So we prefer to follow the name scheme.

Quoted 10 lines
Quoted 10 lines
 3 files changed, 734 insertions(+)

Index: linux-2.6/drivers/i2c/busses/Kconfig
===================================================================

drivers/i2c/busses/Kconfig +0 -0
@@ -5,6 +5,53 @@
 menu "I2C Hardware Bus support"
 	depends on I2C
 
Quoted 4 lines

drivers/i2c/busses/Kconfig line 8

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+config I2C_BFIN_GPIO

I2C_BLACKFIN_GPIO

In the Kconfig, we are trying to using CONFIG_BLACKFIN. So, I changed all the BFIN to BLACKFIN as you mentioned.

Quoted 42 lines

Please move the entries to the right location. The list is sorted alphabetically if you didn't notice.

drivers/i2c/busses/Kconfig line 9

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	tristate "Generic Blackfin and HHBF533/561 development board I2C support"

You can drop the trailing "I2C support", the user is in a menu named "I2C hardware bus support" so it's pretty clear what we're talking about.

drivers/i2c/busses/Kconfig line 17

Quoted 8 lines
drivers/i2c/busses/Kconfig +8 -0
+	depends on I2C && EXPERIMENTAL
+	select I2C_ALGOBIT
+	help
+	--
+
+menu "BFIN I2C SDA/SCL Selection"
+	depends on I2C_BFIN_GPIO
+config BFIN_SDA

I2C_BLACKFIN_SDA

drivers/i2c/busses/Kconfig line 18

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	int "SDA is GPIO Number"

"SDA GPIO pin number"

drivers/i2c/busses/Kconfig line 19

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	range 0 15 if (BF533 || BF532 || BF531) 

Trailing whitespace.

drivers/i2c/busses/Kconfig line 22

Quoted 3 lines
drivers/i2c/busses/Kconfig +3 -0
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 2 if (BF533 || BF532 || BF531) 

Trailing whitespace.

No default for the other cases?

drivers/i2c/busses/Kconfig line 24

Quoted 2 lines
drivers/i2c/busses/Kconfig +2 -0
+
+config BFIN_SCL

I2C_BLACKFIN_SCL
Etc etc, all the options should start with I2C_BLACKFIN.

Quoted 8 lines

drivers/i2c/busses/Kconfig line 25

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	int "SCL is GPIO Number"

"SCL GPIO pin number"

drivers/i2c/busses/Kconfig line 26

Quoted 1 line
drivers/i2c/busses/Kconfig +1 -0
+	range 0 15 if (BF533 || BF532 || BF531) 

Trailing whitespace, and many more after that. Please fix them all!

All above fixed.

Quoted 8 lines
Quoted 8 lines
drivers/i2c/busses/Kconfig +8 -0
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 3 
+endmenu
+
+config I2C_BFIN_GPIO_CYCLE_DELAY
+	int "Cycle Delay in usec"
+	depends on I2C_BFIN_GPIO
Quoted 9 lines

drivers/i2c/busses/Kconfig line 36

Quoted 2 lines
drivers/i2c/busses/Kconfig +2 -0
+	range 1 100 
+	default 40

This should really not be a kernel configuration option. Please turn it into a kernel module parameter or a sysfs attribute if you really need it. Also note that we already have an interface to change this value from user-space (using an ioctl on /dev/i2c-N) and that might be sufficient for your needs.

Actually, for some customer's requirement, they just want to set the configuration in kernel config time not in the runtime.

Quoted 2 lines

And allowing 1 usec delay is probably not a good idea, I don't recommend values below 6 usec with i2c-algo-bit.

I add a range here from 5 to 100, default is 40.

Quoted 25 lines

drivers/i2c/busses/Kconfig line 48

Quoted 12 lines
drivers/i2c/busses/Kconfig +12 -0
+
+config I2C_BFIN_TWI
+	tristate "Blackfin TWI I2C support"
+	depends on I2C && (BF534 || BF536 || BF537)
+	help
+	  This the TWI I2C device driver for Blackfin 534/536/537.
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-bfin-twi.
+
+config TWICLK_KHZ
+	int "TWI clock (kHZ)"

kHz

drivers/i2c/busses/Kconfig line 53

Quoted 5 lines
drivers/i2c/busses/Kconfig +5 -0
+	depends on I2C_BFIN_TWI
+	default 50
+	help
+	  The unit of the TWI clock is kilo HZ. Please divide the clock 
+	  by 1024 if you count it in HZ. The value should be less than 400.

Why don't you use "range" here too to ensure that the value is actually less than 400? Either way, same as above, IMHO this should not be a compilation-time decision.

Quoted 3 lines

A kHz is really 1000 Hz, not 1024. And everybody skilled enough to configure a kernel should know that, I doubt it's worth reminding.

All above fixed.

Quoted 8 lines

drivers/i2c/busses/Kconfig line 57

Quoted 4 lines
drivers/i2c/busses/Kconfig +1 -0
+
 config I2C_ALI1535
 	tristate "ALI 1535"
 	depends on I2C && PCI

All these options won't work really well until you also change drivers/i2c/busses/Makefile to make something useful with them...

Sorry for missing the Makefile in this patch.

Thanks for your review and this is the latest one.

Andrew, I also fixed Kconfig bug in your
blackfin-blackfin-i2c-driver-fix.patch
Please add this one to -mm tree and remove old
one/update.patch/fix.patch from -mm tree.

[PATCH] Blackfin: blackfin i2c driver

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

---

 drivers/i2c/busses/Kconfig         |   47 ++++
 drivers/i2c/busses/Makefile        |    2
 drivers/i2c/busses/i2c-bfin-gpio.c |  100 +++++++++
 drivers/i2c/busses/i2c-bfin-twi.c  |  589 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 738 insertions(+)
drivers/i2c/busses/Kconfig +47 -0
Index: linux-2.6/drivers/i2c/busses/Kconfig
===================================================================
@@ -91,6 +91,53 @@
 	  This driver can also be built as a module.  If so, the module
 	  will be called i2c-au1550.
 
+config I2C_BLACKFIN_GPIO
+	tristate "Blackfin GPIO based I2C interface"
+	depends on I2C && BLACKFIN && EXPERIMENTAL
+	select I2C_ALGOBIT
+	help
+	  Say Y here if you have an Blackfin BF5xx based
+	  system and are using GPIO lines for an I2C bus.
+
+menu "Blackfin I2C SDA/SCL Selection"
+	depends on I2C_BLACKFIN_GPIO
+config I2C_BLACKFIN_GPIO_SDA
+	int "SDA GPIO pin number"
+	range 0 15 if (BF533 || BF532 || BF531)
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 2
+
+config I2C_BLACKFIN_GPIO_SCL
+	int "SCL GPIO pin number"
+	range 0 15 if (BF533 || BF532 || BF531)
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 3
+endmenu
+
+config I2C_BLACKFIN_GPIO_CYCLE_DELAY
+	int "Cycle Delay in uSec"
+	depends on I2C_BLACKFIN_GPIO
+	range 5 100
+	default 40
+
+config I2C_BLACKFIN_TWI
+	tristate "Blackfin TWI I2C support"
+	depends on I2C && (BF534 || BF536 || BF537)
+	help
+	  This is the TWI I2C device driver for Blackfin 534/536/537.
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-bfin-twi.
+
+config I2C_BLACKFIN_TWI_CLK_KHZ
+	int "Blackfin TWI I2C clock (kHz)"
+	depends on I2C_BLACKFIN_TWI
+	range 10 400
+	default 50
+	help
+	  The unit of the TWI clock is kilo HZ.
+
 config I2C_ELEKTOR
 	tristate "Elektor ISA card"
 	depends on I2C && ISA && BROKEN_ON_SMP
drivers/i2c/busses/i2c-bfin-gpio.c new file +100 -0

100 lines not on this page · raw message

drivers/i2c/busses/i2c-bfin-twi.c new file +589 -0

589 lines not on this page · raw message

drivers/i2c/busses/Makefile +2 -0
Index: linux-2.6/drivers/i2c/busses/Makefile
===================================================================
@@ -10,6 +10,8 @@
 obj-$(CONFIG_I2C_AMD8111)	+= i2c-amd8111.o
 obj-$(CONFIG_I2C_AT91)		+= i2c-at91.o
 obj-$(CONFIG_I2C_AU1550)	+= i2c-au1550.o
+obj-$(CONFIG_I2C_BLACKFIN_GPIO) += i2c-bfin-gpio.o
+obj-$(CONFIG_I2C_BLACKFIN_TWI)  += i2c-bfin-twi.o
 obj-$(CONFIG_I2C_ELEKTOR)	+= i2c-elektor.o
 obj-$(CONFIG_I2C_HYDRA)		+= i2c-hydra.o
 obj-$(CONFIG_I2C_I801)		+= i2c-i801.o

_

Thanks again
Best Regards,
-Bryan Wu

#15 lore raw

Hi Brian,

Thanks for the quick update.

Quoted 9 lines

[PATCH] Blackfin: blackfin i2c driver

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

Actually I had only reviewed the Kconfig part, not the driver itself. Here's a more complete review:

Quoted 33 lines

---

 drivers/i2c/busses/Kconfig         |   47 ++++
 drivers/i2c/busses/Makefile        |    2
 drivers/i2c/busses/i2c-bfin-gpio.c |  100 +++++++++
 drivers/i2c/busses/i2c-bfin-twi.c  |  589 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 4 files changed, 738 insertions(+)

Index: linux-2.6/drivers/i2c/busses/Kconfig
===================================================================

drivers/i2c/busses/Kconfig +17 -0
@@ -91,6 +91,53 @@
 	  This driver can also be built as a module.  If so, the module
 	  will be called i2c-au1550.
 
+config I2C_BLACKFIN_GPIO
+	tristate "Blackfin GPIO based I2C interface"
+	depends on I2C && BLACKFIN && EXPERIMENTAL
+	select I2C_ALGOBIT
+	help
+	  Say Y here if you have an Blackfin BF5xx based
+	  system and are using GPIO lines for an I2C bus.
+
+menu "Blackfin I2C SDA/SCL Selection"
+	depends on I2C_BLACKFIN_GPIO
+config I2C_BLACKFIN_GPIO_SDA
+	int "SDA GPIO pin number"
+	range 0 15 if (BF533 || BF532 || BF531)
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 2
+

drivers/i2c/busses/Kconfig line 120

Quoted 10 lines
drivers/i2c/busses/Kconfig +10 -0
+config I2C_BLACKFIN_GPIO_SCL
+	int "SCL GPIO pin number"
+	range 0 15 if (BF533 || BF532 || BF531)
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 3
+endmenu
+
+config I2C_BLACKFIN_GPIO_CYCLE_DELAY
+	int "Cycle Delay in uSec"

The proper symbol is "us".

Quoted 9 lines
drivers/i2c/busses/Kconfig +9 -0
+	depends on I2C_BLACKFIN_GPIO
+	range 5 100
+	default 40
+
+config I2C_BLACKFIN_TWI
+	tristate "Blackfin TWI I2C support"
+	depends on I2C && (BF534 || BF536 || BF537)
+	help
+	  This is the TWI I2C device driver for Blackfin 534/536/537.

drivers/i2c/busses/Kconfig line 139

Quoted 10 lines
drivers/i2c/busses/Kconfig +10 -0
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-bfin-twi.
+
+config I2C_BLACKFIN_TWI_CLK_KHZ
+	int "Blackfin TWI I2C clock (kHz)"
+	depends on I2C_BLACKFIN_TWI
+	range 10 400
+	default 50
+	help
+	  The unit of the TWI clock is kilo HZ.

kHz

drivers/i2c/busses/i2c-bfin-gpio.c line 4

Quoted 13 lines
drivers/i2c/busses/Kconfig +1 -0
+
 config I2C_ELEKTOR
 	tristate "Elektor ISA card"
 	depends on I2C && ISA && BROKEN_ON_SMP

Index: linux-2.6/drivers/i2c/busses/i2c-bfin-gpio.c
===================================================================

drivers/i2c/busses/i2c-bfin-gpio.c new file +4 -0
@@ -0,0 +1,100 @@
+/****************************************************************
+ * Description:                                                 *
+ *                                                              *
+ * Maintainer: Meihui Fan <[email protected]>		        *

Indentation problem. Also, maintainer information goes in MAINTAINERS, not individual drivers.

drivers/i2c/busses/i2c-bfin-gpio.c line 11

Quoted 7 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +7 -0
+ *                                                              *
+ * CopyRight (c)  2004  HHTech                                  *
+ *   www.hhcn.com, www.hhcn.org                                 *
+ *   All rights reserved.                                       *
+ *                                                              *
+ * This file is free software;                                  *
+ *   you are free to modify and/or redistribute it   	        *

Indentation problem here too.

Quoted 5 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +5 -0
+ *   under the terms of the GNU General Public Licence (GPL).   *
+ *                                                              *
+ ****************************************************************/
+
+#include <linux/module.h>

drivers/i2c/busses/i2c-bfin-gpio.c line 26

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +10 -0
+#include <linux/kernel.h>
+#include <linux/delay.h>
+#include <linux/init.h>
+#include <linux/i2c.h>
+#include <linux/i2c-algo-bit.h>
+
+#include <asm/blackfin.h>
+#include <asm/gpio.h>
+
+#define	I2C_HW_B_HHBF		    0x13

Holy god, no! This define goes in include/linux/i2c-id.h.

drivers/i2c/busses/i2c-bfin-gpio.c line 32

Quoted 6 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +6 -0
+
+static void hhbf_setsda(void *data, int state)
+{
+	if (state) {
+		gpio_direction_input(CONFIG_I2C_BLACKFIN_GPIO_SDA);
+

Extra blank line.

Quoted 11 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +11 -0
+	} else {
+		gpio_direction_output(CONFIG_I2C_BLACKFIN_GPIO_SDA);
+		gpio_set_value(CONFIG_I2C_BLACKFIN_GPIO_SDA, 0);
+	}
+}
+
+static void hhbf_setscl(void *data, int state)
+{
+	gpio_set_value(CONFIG_I2C_BLACKFIN_GPIO_SCL, state);
+}
+

drivers/i2c/busses/i2c-bfin-gpio.c line 53

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +10 -0
+static int hhbf_getsda(void *data)
+{
+	return (gpio_get_value(CONFIG_I2C_BLACKFIN_GPIO_SDA) != 0);
+}
+
+
+static struct i2c_algo_bit_data bit_hhbf_data = {
+	.setsda  = hhbf_setsda,
+	.setscl  = hhbf_setscl,
+	.getsda  = hhbf_getsda,

Is the hardware not capable of SCL sensing? If it is, you really want to implement it. Bit-banged I2C without SCL sensing doesn't completely implement the I2C standard, it lacks the SCL stretching feature.

drivers/i2c/busses/i2c-bfin-gpio.c line 55

Quoted 2 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +2 -0
+	.udelay  = CONFIG_I2C_BLACKFIN_GPIO_CYCLE_DELAY,
+	.timeout = HZ

Always add an extra comma at the end of struct declarations.

drivers/i2c/busses/i2c-bfin-gpio.c line 62

Quoted 7 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +7 -0
+};
+
+static struct i2c_adapter hhbf_ops = {
+	.owner 	= THIS_MODULE,
+	.id 	= I2C_HW_B_HHBF,
+	.algo_data 	= &bit_hhbf_data,
+	.name	= "Blackfin GPIO based I2C driver",

This structure describes a (virtual) device, not a driver.

drivers/i2c/busses/i2c-bfin-gpio.c line 67

Quoted 5 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +5 -0
+};
+
+static int __init i2c_hhbf_init(void)
+{
+

Extra blank line.

drivers/i2c/busses/i2c-bfin-gpio.c line 70

Quoted 3 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +3 -0
+	if (gpio_request(CONFIG_I2C_BLACKFIN_GPIO_SCL, NULL)) {
+		printk(KERN_ERR "%s: gpio_request GPIO %d failed \n", __func__,
+				CONFIG_I2C_BLACKFIN_GPIO_SCL);

The function name doesn't matter here, what is the end user going to do with the information? What matters is the driver name.

drivers/i2c/busses/i2c-bfin-gpio.c line 71

Quoted 1 line
drivers/i2c/busses/i2c-bfin-gpio.c new file +1 -0
+		return -1;

This -1 will be interpreted as -EPERM by insmod. Please use a proper error code.

drivers/i2c/busses/i2c-bfin-gpio.c line 77

Quoted 6 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +6 -0
+	}
+
+	if (gpio_request(CONFIG_I2C_BLACKFIN_GPIO_SDA, NULL)) {
+		printk(KERN_ERR "%s: gpio_request GPIO %d failed \n", __func__,
+				CONFIG_I2C_BLACKFIN_GPIO_SDA);
+		return -1;

Broken error path. You've successfully requested the GPIO for SCL, but return with an error without releasing it.

drivers/i2c/busses/i2c-bfin-gpio.c line 43

Quoted 3 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +3 -0
+	}
+
+

Extra blank line.

drivers/i2c/busses/i2c-bfin-gpio.c line 84

Quoted 4 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +4 -0
+	gpio_direction_output(CONFIG_I2C_BLACKFIN_GPIO_SCL);
+	gpio_direction_input(CONFIG_I2C_BLACKFIN_GPIO_SDA);
+	gpio_set_value(CONFIG_I2C_BLACKFIN_GPIO_SCL, 1);
+

At this point, i2c-core will complain that you created an i2c_adapter without a valid parent device. This needs to be fixed.

drivers/i2c/busses/i2c-bfin-gpio.c line 92

Quoted 8 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +8 -0
+	return i2c_bit_add_bus(&hhbf_ops);
+}
+
+static void __exit i2c_hhbf_exit(void)
+{
+	gpio_free(CONFIG_I2C_BLACKFIN_GPIO_SCL);
+	gpio_free(CONFIG_I2C_BLACKFIN_GPIO_SDA);
+	i2c_bit_del_bus(&hhbf_ops);

You're doing things the wrong way around here, you're freeing the resources while there might still be someone using the adapter.

BTW, i2c_bit_del_bus no longer exists, so this won't compile. Use i2c_del_adapter instead. You really want to test your code with a recent kernel before submitting it again.

Quoted 13 lines
drivers/i2c/busses/i2c-bfin-gpio.c new file +8 -0
+}
+
+MODULE_AUTHOR("Meihui Fan <[email protected]>");
+MODULE_DESCRIPTION("I2C-Bus adapter routines for Blackfin and HHBF Boards");
+MODULE_LICENSE("GPL");
+
+module_init(i2c_hhbf_init);
+module_exit(i2c_hhbf_exit);

Index: linux-2.6/drivers/i2c/busses/i2c-bfin-twi.c
===================================================================

drivers/i2c/busses/i2c-bfin-twi.c line 3

Quoted 3 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +3 -0
@@ -0,0 +1,589 @@
+/****************************************************************
+ * Description: Driver for Blackfin Two Wire Interface          *
+ * Maintainer:  sonicz  <[email protected]>                *

Same as above, this goes to MAINTAINERS.

drivers/i2c/busses/i2c-bfin-twi.c line 8

Quoted 5 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +5 -0
+ *                                                              *
+ * Copyright (c) 2005-2007 Analog Device                        *
+ *                                                              *
+ * This file is free software;                                  *
+ *   you are free to modify and/or redistribute it   	        *

Broken indentation.

Quoted 7 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +7 -0
+ *   under the terms of the GNU General Public Licence (GPL).   *
+ ****************************************************************/
+
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/i2c.h>

drivers/i2c/busses/i2c-bfin-twi.c line 25

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+#include <linux/mm.h>
+#include <linux/timer.h>
+#include <linux/spinlock.h>
+#include <linux/completion.h>
+#include <linux/interrupt.h>
+
+#include <asm/blackfin.h>
+#include <asm/irq.h>
+
+#define I2C_BLACKFIN_TWI       0x00

No, define a valid ID in linux/i2c-id.h if you need one, otherwise don't set an ID at all.

drivers/i2c/busses/i2c-bfin-twi.c line 29

Quoted 4 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +4 -0
+#define POLL_TIMEOUT       (2 * HZ)
+#ifndef CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ
+#define CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ  400
+#endif

How could this actually not be defined?

drivers/i2c/busses/i2c-bfin-twi.c line 37

Quoted 8 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +8 -0
+
+/* SMBus mode*/
+#define TWI_I2C_MODE_STANDARD		0x01
+#define TWI_I2C_MODE_STANDARDSUB	0x02
+#define TWI_I2C_MODE_COMBINED		0x04
+
+struct bfin_twi_iface
+{

The curly brace goes at the end of the previous line.

Quoted 131 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +131 -0
+	struct mutex		twi_lock;
+	int			irq;
+	spinlock_t		lock;
+	char			read_write;
+	u8			command;
+	u8			*transPtr;
+	int			readNum;
+	int			writeNum;
+	int			cur_mode;
+	int			manual_stop;
+	int			result;
+	int			timeout_count;
+	struct timer_list	timeout_timer;
+	struct i2c_adapter	adap;
+	struct completion	complete;
+};
+
+static struct bfin_twi_iface twi_iface;
+
+static void bfin_twi_handle_interrupt(struct bfin_twi_iface *iface)
+{
+	unsigned short twi_int_status = bfin_read_TWI_INT_STAT();
+	unsigned short mast_stat = bfin_read_TWI_MASTER_STAT();
+
+	if (twi_int_status & XMTSERV) {
+		/* Transmit next data */
+		if (iface->writeNum > 0) {
+			bfin_write_TWI_XMT_DATA8(*(iface->transPtr++));
+			iface->writeNum--;
+		}
+		/* start receive immediately after complete sending in
+		 * combine mode.
+		 */
+		else if (iface->cur_mode == TWI_I2C_MODE_COMBINED) {
+			bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL()
+				| MDIR | RSTART);
+		} else if (iface->manual_stop)
+			bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL()
+				| STOP);
+		SSYNC();
+		/* Clear status */
+		bfin_write_TWI_INT_STAT(XMTSERV);
+		SSYNC();
+	}
+	if (twi_int_status & RCVSERV) {
+		if (iface->readNum > 0) {
+			/* Receive next data */
+			*(iface->transPtr) = bfin_read_TWI_RCV_DATA8();
+			if (iface->cur_mode == TWI_I2C_MODE_COMBINED) {
+				/* Change combine mode into sub mode after
+				 * read first data.
+				 */
+				iface->cur_mode = TWI_I2C_MODE_STANDARDSUB;
+				/* Get read number from first byte in block
+				 * combine mode.
+				 */
+				if (iface->readNum == 1 && iface->manual_stop)
+					iface->readNum = *iface->transPtr+1;
+			}
+			iface->transPtr++;
+			iface->readNum--;
+		} else if (iface->manual_stop) {
+			bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL()
+				| STOP);
+			SSYNC();
+		}
+		/* Clear interrupt source */
+		bfin_write_TWI_INT_STAT(RCVSERV);
+		SSYNC();
+	}
+	if (twi_int_status & MERR) {
+		bfin_write_TWI_INT_STAT(MERR);
+		bfin_write_TWI_INT_MASK(0);
+		bfin_write_TWI_MASTER_STAT(0x3e);
+		bfin_write_TWI_MASTER_CTL(0);
+		SSYNC();
+		iface->result = -1;
+		/* if both err and complete int stats are set, return proper
+		 * results.
+		 */
+		if (twi_int_status & MCOMP) {
+			bfin_write_TWI_INT_STAT(MCOMP);
+			bfin_write_TWI_INT_MASK(0);
+			bfin_write_TWI_MASTER_CTL(0);
+			SSYNC();
+			/* If it is a quick transfer, only address bug no data,
+			 * not an err, return 1.
+			 */
+			if (iface->writeNum == 0 && (mast_stat & BUFRDERR))
+				iface->result = 1;
+			/* If address not acknowledged return -1,
+			 * else return 0.
+			 */
+			else if (!(mast_stat & ANAK))
+				iface->result = 0;
+		}
+		complete(&iface->complete);
+		return;
+	}
+	if (twi_int_status & MCOMP) {
+		bfin_write_TWI_INT_STAT(MCOMP);
+		SSYNC();
+		if (iface->cur_mode == TWI_I2C_MODE_COMBINED) {
+			if (iface->readNum == 0) {
+				/* set the read number to 1 and ask for manual
+				 * stop in block combine mode
+				 */
+				iface->readNum = 1;
+				iface->manual_stop = 1;
+				bfin_write_TWI_MASTER_CTL(
+					bfin_read_TWI_MASTER_CTL()
+					| (0xff << 6));
+			} else {
+				/* set the readd number in other
+				 * combine mode.
+				 */
+				bfin_write_TWI_MASTER_CTL(
+					(bfin_read_TWI_MASTER_CTL() &
+					(~(0xff << 6))) |
+					( iface->readNum << 6));
+			}
+			/* remove restart bit and enable master receive */
+			bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL() &
+				~RSTART);
+			bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL() |
+				MEN | MDIR);
+			SSYNC();
+		} else {
+			iface->result = 1;
+			bfin_write_TWI_INT_MASK(0);
+			bfin_write_TWI_MASTER_CTL(0);

drivers/i2c/busses/i2c-bfin-twi.c line 178

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+			SSYNC();
+			complete(&iface->complete);
+		}
+	}
+}
+
+/* Interrupt handler */
+static irqreturn_t bfin_twi_interrupt_entry(int irq, void *dev_id)
+{
+	struct bfin_twi_iface *iface = (struct bfin_twi_iface *)dev_id;

Cast isn't needed.

Quoted 24 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +24 -0
+	unsigned long flags;
+
+	spin_lock_irqsave(&iface->lock, flags);
+	del_timer(&iface->timeout_timer);
+	bfin_twi_handle_interrupt(iface);
+	spin_unlock_irqrestore(&iface->lock, flags);
+	return IRQ_HANDLED;
+}
+
+static void bfin_twi_timeout(unsigned long data)
+{
+	struct bfin_twi_iface *iface = (struct bfin_twi_iface *)data;
+	unsigned long flags;
+
+	spin_lock_irqsave(&iface->lock, flags);
+	bfin_twi_handle_interrupt(iface);
+	if (iface->result == 0) {
+		iface->timeout_count--;
+		if (iface->timeout_count > 0) {
+			iface->timeout_timer.expires = jiffies + POLL_TIMEOUT;
+			add_timer(&iface->timeout_timer);
+		} else {
+			iface->result = -1;
+			complete(&iface->complete);

drivers/i2c/busses/i2c-bfin-twi.c line 212

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+		}
+	}
+	spin_unlock_irqrestore(&iface->lock, flags);
+}
+
+/*
+ * Generic i2c master transfer entrypoint
+ */
+static int bfin_twi_master_xfer(struct i2c_adapter *adap,
+				struct i2c_msg msgs[], int num)

The prefered syntax is struct i2c_msg *msgs.

Quoted 13 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +13 -0
+{
+	struct bfin_twi_iface* iface = adap->algo_data;
+	struct i2c_msg *pmsg;
+	int i, ret;
+	int rc = 0;
+
+	if (!(bfin_read_TWI_CONTROL() & TWI_ENA))
+		return -ENXIO;
+
+	mutex_lock(&iface->twi_lock);
+
+	while (bfin_read_TWI_MASTER_STAT() & BUSBUSY) {
+		mutex_unlock(&iface->twi_lock);

drivers/i2c/busses/i2c-bfin-twi.c line 235

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+		cond_resched();
+		mutex_lock(&iface->twi_lock);
+	}
+
+	ret = 0;
+	for (i = 0; rc >= 0 && i < num; i++) {
+		pmsg = &msgs[i];
+		if (pmsg->flags & I2C_M_TEN) {
+			printk(KERN_ERR "i2c-bfin-twi: 10 bits addr \
+					not supported !\n");

This is not how you break a string in C. The right way is:

			printk(KERN_ERR "i2c-bfin-twi: 10 bits addr "
					"not supported!\n");

BTW you should be using dev_err() here instead.

Quoted 62 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +62 -0
+			rc = -EINVAL;
+			break;
+		}
+
+		iface->cur_mode = TWI_I2C_MODE_STANDARD;
+		iface->manual_stop = 0;
+		iface->transPtr = pmsg->buf;
+		iface->writeNum = iface->readNum = pmsg->len;
+		iface->result = 0;
+		iface->timeout_count = 10;
+		/* Set Transmit device address */
+		bfin_write_TWI_MASTER_ADDR(pmsg->addr);
+
+		/* FIFO Initiation. Data in FIFO should be
+		 *  discarded before start a new operation.
+		 */
+		bfin_write_TWI_FIFO_CTL(0x3);
+		SSYNC();
+		bfin_write_TWI_FIFO_CTL(0);
+		SSYNC();
+
+		if (pmsg->flags & I2C_M_RD)
+			iface->read_write = I2C_SMBUS_READ;
+		else {
+			iface->read_write = I2C_SMBUS_WRITE;
+			/* Transmit first data */
+			if (iface->writeNum > 0) {
+				bfin_write_TWI_XMT_DATA8(*(iface->transPtr++));
+				iface->writeNum--;
+				SSYNC();
+			}
+		}
+
+		/* clear int stat */
+		bfin_write_TWI_INT_STAT(MERR|MCOMP|XMTSERV|RCVSERV);
+
+		/* Interrupt mask . Enable XMT, RCV interrupt */
+		bfin_write_TWI_INT_MASK(MCOMP | MERR |
+			((iface->read_write == I2C_SMBUS_READ)?
+			RCVSERV : XMTSERV));
+		SSYNC();
+
+		if (pmsg->len > 0 && pmsg->len <= 255)
+			bfin_write_TWI_MASTER_CTL(pmsg->len << 6);
+		else if (pmsg->len > 255) {
+			bfin_write_TWI_MASTER_CTL(0xff << 6);
+			iface->manual_stop = 1;
+		} else
+			break;
+
+		iface->timeout_timer.expires = jiffies + POLL_TIMEOUT;
+		add_timer(&iface->timeout_timer);
+
+		/* Master enable */
+		bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL() | MEN |
+			((iface->read_write == I2C_SMBUS_READ) ? MDIR : 0) |
+			((CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ>100) ? FAST : 0));
+		SSYNC();
+
+		wait_for_completion(&iface->complete);
+
+		rc = iface->result;

drivers/i2c/busses/i2c-bfin-twi.c line 307

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+		if (rc == 1)
+			ret++;
+		else if (rc == -1)
+			break;
+	}
+
+	/* Release mutex */
+	mutex_unlock(&iface->twi_lock);
+
+	return ret;

ret always has the same value as i as far as I can see, so you don't need to separate variables.

Quoted 189 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +189 -0
+}
+
+/*
+ * SMBus type transfer entrypoint
+ */
+
+int bfin_twi_smbus_xfer(struct i2c_adapter *adap, u16 addr,
+			unsigned short flags, char read_write,
+			u8 command, int size, union i2c_smbus_data * data)
+{
+	struct bfin_twi_iface* iface = adap->algo_data;
+	int rc = 0;
+
+	if (!(bfin_read_TWI_CONTROL() & TWI_ENA))
+		return -ENXIO;
+
+	mutex_lock(&iface->twi_lock);
+
+	while(bfin_read_TWI_MASTER_STAT() & BUSBUSY) {
+		mutex_unlock(&iface->twi_lock);
+		cond_resched();
+		mutex_lock(&iface->twi_lock);
+	}
+
+	iface->writeNum = 0;
+	iface->readNum = 0;
+
+	/* Prepare datas & select mode */
+	switch (size) {
+	case I2C_SMBUS_QUICK:
+		iface->transPtr = NULL;
+		iface->cur_mode = TWI_I2C_MODE_STANDARD;
+		break;
+	case I2C_SMBUS_BYTE:
+		if (data == NULL)
+			iface->transPtr = NULL;
+		else {
+			if (read_write == I2C_SMBUS_READ)
+				iface->readNum = 1;
+			else
+				iface->writeNum = 1;
+			iface->transPtr = &data->byte;
+		}
+		iface->cur_mode = TWI_I2C_MODE_STANDARD;
+		break;
+	case I2C_SMBUS_BYTE_DATA:
+		if (read_write == I2C_SMBUS_READ) {
+			iface->readNum = 1;
+			iface->cur_mode = TWI_I2C_MODE_COMBINED;
+		} else {
+			iface->writeNum = 1;
+			iface->cur_mode = TWI_I2C_MODE_STANDARDSUB;
+		}
+		iface->transPtr = &data->byte;
+		break;
+	case I2C_SMBUS_WORD_DATA:
+		if (read_write == I2C_SMBUS_READ) {
+			iface->readNum = 2;
+			iface->cur_mode = TWI_I2C_MODE_COMBINED;
+		} else {
+			iface->writeNum = 2;
+			iface->cur_mode = TWI_I2C_MODE_STANDARDSUB;
+		}
+		iface->transPtr = (u8 *)&data->word;
+		break;
+	case I2C_SMBUS_PROC_CALL:
+		iface->writeNum = 2;
+		iface->readNum = 2;
+		iface->cur_mode = TWI_I2C_MODE_COMBINED;
+		iface->transPtr = (u8 *)&data->word;
+		break;
+	case I2C_SMBUS_BLOCK_DATA:
+		if (read_write == I2C_SMBUS_READ) {
+			iface->readNum = 0;
+			iface->cur_mode = TWI_I2C_MODE_COMBINED;
+		} else {
+			iface->writeNum = data->block[0]+1;
+			iface->cur_mode = TWI_I2C_MODE_STANDARDSUB;
+		}
+		iface->transPtr = data->block;
+		break;
+	default:
+		return -1;
+	}
+
+	iface->result = 0;
+	iface->manual_stop = 0;
+	iface->read_write = read_write;
+	iface->command = command;
+	iface->timeout_count = 10;
+
+	/* FIFO Initiation. Data in FIFO should be discarded before
+	 * start a new operation.
+	 */
+	bfin_write_TWI_FIFO_CTL(0x3);
+	SSYNC();
+	bfin_write_TWI_FIFO_CTL(0);
+
+	/* clear int stat */
+	bfin_write_TWI_INT_STAT(MERR|MCOMP|XMTSERV|RCVSERV);
+
+	/* Set Transmit device address */
+	bfin_write_TWI_MASTER_ADDR(addr);
+	SSYNC();
+
+	iface->timeout_timer.expires = jiffies + POLL_TIMEOUT;
+	add_timer(&iface->timeout_timer);
+
+	switch (iface->cur_mode) {
+	case TWI_I2C_MODE_STANDARDSUB:
+		bfin_write_TWI_XMT_DATA8(iface->command);
+		bfin_write_TWI_INT_MASK(MCOMP | MERR |
+			((iface->read_write == I2C_SMBUS_READ) ?
+			RCVSERV : XMTSERV));
+		SSYNC();
+
+		if (iface->writeNum + 1 <= 255)
+			bfin_write_TWI_MASTER_CTL((iface->writeNum+1) << 6);
+		else {
+			bfin_write_TWI_MASTER_CTL(0xff << 6);
+			iface->manual_stop = 1;
+		}
+		/* Master enable */
+		bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL() | MEN |
+			((CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ>100) ? FAST : 0));
+		break;
+	case TWI_I2C_MODE_COMBINED:
+		bfin_write_TWI_XMT_DATA8(iface->command);
+		bfin_write_TWI_INT_MASK(MCOMP | MERR | RCVSERV | XMTSERV);
+		SSYNC();
+
+		if (iface->writeNum > 0)
+			bfin_write_TWI_MASTER_CTL((iface->writeNum+1) << 6);
+		else
+			bfin_write_TWI_MASTER_CTL(0x1 << 6);
+		/* Master enable */
+		bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL() | MEN |
+			((CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ>100) ? FAST : 0));
+		break;
+	default:
+		bfin_write_TWI_MASTER_CTL(0);
+		if (size != I2C_SMBUS_QUICK) {
+			/* Don't access xmit data register when this is a
+			 * read operation.
+			 */
+			if (iface->read_write != I2C_SMBUS_READ) {
+				if (iface->writeNum > 0) {
+					bfin_write_TWI_XMT_DATA8
+						(*(iface->transPtr++));
+					if (iface->writeNum <= 255)
+						bfin_write_TWI_MASTER_CTL
+							(iface->writeNum << 6);
+					else {
+						bfin_write_TWI_MASTER_CTL
+							(0xff << 6);
+						iface->manual_stop = 1;
+					}
+					iface->writeNum--;
+				} else {
+					bfin_write_TWI_XMT_DATA8
+						(iface->command);
+					bfin_write_TWI_MASTER_CTL(1 << 6);
+				}
+			} else {
+				if (iface->readNum > 0
+					&& iface->readNum <= 255)
+					bfin_write_TWI_MASTER_CTL
+						(iface->readNum << 6);
+				else if (iface->readNum > 255) {
+					bfin_write_TWI_MASTER_CTL( 0xff << 6 );
+					iface->manual_stop = 1;
+				} else {
+					del_timer(&iface->timeout_timer);
+					break;
+				}
+			}
+		}
+		bfin_write_TWI_INT_MASK(MCOMP | MERR |
+			((iface->read_write == I2C_SMBUS_READ) ?
+			RCVSERV : XMTSERV));
+		SSYNC();
+
+		/* Master enable */
+		bfin_write_TWI_MASTER_CTL(bfin_read_TWI_MASTER_CTL() | MEN |
+			((iface->read_write == I2C_SMBUS_READ) ? MDIR : 0) |
+			((CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ > 100) ? FAST : 0));
+		break;
+	}
+	SSYNC();

drivers/i2c/busses/i2c-bfin-twi.c line 506

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+
+	wait_for_completion(&iface->complete);
+
+	rc = (iface->result >= 0) ? 0 : -1;
+
+	/* Release mutex */
+	mutex_unlock(&iface->twi_lock);
+
+	return rc;
+}

i2c-core can emulate SMBus transactions using master_xfer, so in general when you have a complete master_xfer implementation you do not need to define a separate smbus_xfer function. This would save a lot of code.

Quoted 26 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +26 -0
+
+/*
+ * Return what the adapter supports
+ */
+static u32 bfin_twi_functionality(struct i2c_adapter *adap)
+{
+	return I2C_FUNC_SMBUS_QUICK | I2C_FUNC_SMBUS_BYTE |
+	       I2C_FUNC_SMBUS_BYTE_DATA | I2C_FUNC_SMBUS_WORD_DATA |
+	       I2C_FUNC_SMBUS_BLOCK_DATA | I2C_FUNC_SMBUS_PROC_CALL |
+	       I2C_FUNC_I2C;
+}
+
+
+static struct i2c_algorithm bfin_twi_algorithm = {
+	.master_xfer   = bfin_twi_master_xfer,
+	.smbus_xfer    = bfin_twi_smbus_xfer,
+	.functionality = bfin_twi_functionality,
+};
+
+static int __init i2c_bfin_twi_init(void)
+{
+	struct i2c_adapter *p_adap;
+	int rc;
+
+	mutex_init(&(twi_iface.twi_lock));
+	spin_lock_init(&twi_iface.lock);

drivers/i2c/busses/i2c-bfin-twi.c line 542

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+	init_completion(&twi_iface.complete);
+	twi_iface.irq = IRQ_TWI;
+
+	init_timer(&twi_iface.timeout_timer);
+	twi_iface.timeout_timer.function = bfin_twi_timeout;
+	twi_iface.timeout_timer.data = (unsigned long)&twi_iface;
+
+	p_adap = &twi_iface.adap;
+	p_adap->id = I2C_BLACKFIN_TWI;
+	strcpy(p_adap->name, "i2c-bfin-twi");

Please use strlcpy intead.

drivers/i2c/busses/i2c-bfin-twi.c line 546

Quoted 4 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +4 -0
+	p_adap->algo = &bfin_twi_algorithm;
+	p_adap->algo_data = &twi_iface;
+	p_adap->client_register = NULL;
+	p_adap->client_unregister = NULL;

These are already NULL, no need to set them explicitly.

drivers/i2c/busses/i2c-bfin-twi.c line 552

Quoted 6 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +6 -0
+	p_adap->class = I2C_CLASS_ALL;
+
+	rc = request_irq(twi_iface.irq, bfin_twi_interrupt_entry,
+		SA_INTERRUPT, "i2c-bfin-twi", &twi_iface);
+	if (rc) {
+		printk(KERN_ERR "i2c-bfin-twi: can't get IRQ %d !\n",

No space before exclamation mark in english.

Quoted 4 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +4 -0
+			twi_iface.irq);
+		return -ENODEV;
+	}
+

drivers/i2c/busses/i2c-bfin-twi.c line 566

Quoted 10 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +10 -0
+	/* Set TWI internal clock as 10MHz */
+	bfin_write_TWI_CONTROL(((get_sclk() / 1024 / 1024 + 5) / 10) & 0x7F);
+
+	/* Set Twi interface clock as specified */
+	if (CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ > 400)
+		bfin_write_TWI_CLKDIV((( 5*1024 / 400 ) << 8) |
+			(( 5*1024 / 400 ) & 0xFF));
+	else
+		bfin_write_TWI_CLKDIV((( 5*1024 / CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ ) << 8) |
+			(( 5*1024 / CONFIG_I2C_BLACKFIN_TWI_CLK_KHZ ) & 0xFF));

Line too long, please fold more.

drivers/i2c/busses/i2c-bfin-twi.c line 572

Quoted 6 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +6 -0
+
+	/* Enable TWI */
+	bfin_write_TWI_CONTROL(bfin_read_TWI_CONTROL() | TWI_ENA);
+	SSYNC();
+
+	rc = i2c_add_adapter(p_adap);

Here too, i2c-core will complain that you failed to provide a valid parent device. You need to fix this, most probably by converting your driver to a platform driver.

Quoted 30 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +17 -0
+	if(rc < 0)
+		free_irq(twi_iface.irq, &twi_iface);
+	return rc;
+}
+
+static void __exit i2c_bfin_twi_exit(void)
+{
+	i2c_del_adapter(&twi_iface.adap);
+	free_irq(twi_iface.irq, &twi_iface);
+}
+
+MODULE_AUTHOR("Sonic Zhang <[email protected]>");
+MODULE_DESCRIPTION("I2C-Bus adapter routines for Blackfin TWI");
+MODULE_LICENSE("GPL");
+
+module_init(i2c_bfin_twi_init);
+module_exit(i2c_bfin_twi_exit);

Index: linux-2.6/drivers/i2c/busses/Makefile
===================================================================

drivers/i2c/busses/Makefile +2 -0
@@ -10,6 +10,8 @@
 obj-$(CONFIG_I2C_AMD8111)	+= i2c-amd8111.o
 obj-$(CONFIG_I2C_AT91)		+= i2c-at91.o
 obj-$(CONFIG_I2C_AU1550)	+= i2c-au1550.o
+obj-$(CONFIG_I2C_BLACKFIN_GPIO) += i2c-bfin-gpio.o
+obj-$(CONFIG_I2C_BLACKFIN_TWI)  += i2c-bfin-twi.o
 obj-$(CONFIG_I2C_ELEKTOR)	+= i2c-elektor.o
 obj-$(CONFIG_I2C_HYDRA)		+= i2c-hydra.o
 obj-$(CONFIG_I2C_I801)		+= i2c-i801.o
Signature · 1 line
Jean Delvare
#16 lore raw
Quoted 6 lines · Jean Delvare

Hi Brian,

Thanks for the quick update.

Quoted 14 lines

drivers/i2c/busses/i2c-bfin-twi.c line 506

Quoted 8 lines
drivers/i2c/busses/i2c-bfin-twi.c new file +8 -0
+
+     rc = (iface->result >= 0) ? 0 : -1;
+
+     /* Release mutex */
+     mutex_unlock(&iface->twi_lock);
+
+     return rc;
+}

i2c-core can emulate SMBus transactions using master_xfer, so in general when you have a complete master_xfer implementation you do not need to define a separate smbus_xfer function. This would save a lot of code.

Actually the i2c-core can't emulate SMBus transactions using the master_xfer function, because the blackfin TWI controller provide hardware support to the SMBus transactions and the combination of master_xfer operations can't generate proper signal for SMBus.

#17 lore raw
Quoted 10 lines · Sonic Zhang
Quoted 4 lines · Jean Delvare

i2c-core can emulate SMBus transactions using master_xfer, so in general when you have a complete master_xfer implementation you do not need to define a separate smbus_xfer function. This would save a lot of code.

Actually the i2c-core can't emulate SMBus transactions using the master_xfer function, because the blackfin TWI controller provide hardware support to the SMBus transactions and the combination of master_xfer operations can't generate proper signal for SMBus.

Did you try? I can't think of any valid reason why it wouldn't work. All SMBus transactions are compatible with I2C by definition.

Now performance might be better with dedicated code if the hardware accelerate SMBus transactions, that's a different issue. If you prefer to keep the smbus_xfer method for performance reasons, that's OK with me. After all you're the one who will have to maintain the driver ;)

Signature · 1 line
Jean Delvare
#18 lore raw
In reply to Wu, Bryan
Quoted 59 lines · Bryan

Here is the updated blackfin i2c driver.

[PATCH] Blackfin: blackfin i2c driver

The i2c linux driver for blackfin architecture which supports both GPIO i2c operation and blackfin on-chip TWI controller i2c operation.

---
 drivers/i2c/busses/Kconfig         |   47 ++++
 drivers/i2c/busses/i2c-bfin-gpio.c |   98 +++++++++
 drivers/i2c/busses/i2c-bfin-twi.c  |  589 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 734 insertions(+)

Index: linux-2.6/drivers/i2c/busses/Kconfig
===================================================================

drivers/i2c/busses/Kconfig +37 -0
@@ -5,6 +5,53 @@
 menu "I2C Hardware Bus support"
 	depends on I2C
 
+config I2C_BFIN_GPIO
+	tristate "Generic Blackfin and HHBF533/561 development board I2C support"
+	depends on I2C && EXPERIMENTAL
+	select I2C_ALGOBIT
+	help
+	--
+
+menu "BFIN I2C SDA/SCL Selection"
+	depends on I2C_BFIN_GPIO
+config BFIN_SDA
+	int "SDA is GPIO Number"
+	range 0 15 if (BF533 || BF532 || BF531) 
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 2 if (BF533 || BF532 || BF531) 
+
+config BFIN_SCL
+	int "SCL is GPIO Number"
+	range 0 15 if (BF533 || BF532 || BF531) 
+	range 0 47 if (BF534 || BF536 || BF537)
+	range 0 47 if BF561
+	default 3 
+endmenu
+
+config I2C_BFIN_GPIO_CYCLE_DELAY
+	int "Cycle Delay in usec"
+	depends on I2C_BFIN_GPIO
+	range 1 100 
+	default 40
+
+config I2C_BFIN_TWI
+	tristate "Blackfin TWI I2C support"
+	depends on I2C && (BF534 || BF536 || BF537)
+	help
+	  This the TWI I2C device driver for Blackfin 534/536/537.
+
+	  This driver can also be built as a module.  If so, the module

drivers/i2c/busses/Kconfig line 54

Quoted 10 lines
drivers/i2c/busses/Kconfig +10 -0
+	  will be called i2c-bfin-twi.
+
+config TWICLK_KHZ
+	int "TWI clock (kHZ)"
+	depends on I2C_BFIN_TWI
+	default 50
+	help
+	  The unit of the TWI clock is kilo HZ. Please divide the clock 
+	  by 1024 if you count it in HZ. The value should be less than 400.
+

Well that's cute. This patch causes an i386 `make allmodconfig' to spew these:

    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 
    SDA is GPIO Number (BFIN_SDA) [] (NEW) 

out at about 1,000,000/sec, infinitely.

I'll put a `depends on BFIN' in there to shut it up, but I think you've tickled a Kconfig bug.