Skip to contentSkip to conversation

Marvell mv64xxx i2c driver

PATCH In review Mark A. Greer 4 participants 19 messages Jan 26, 2005 – May 19, 2005

Subsystems (2)
I2C subsystem host drivers I2C MV64XXX Marvell and Allwinner driver
lore
Share & tools
lore

Mark A. Greer

author
#0 lore raw

PATCH 1/1 Marvell mv64xxx i2c driver

Greg, Philip,

Marvell makes a line of host bridge for PPC and MIPS systems. On those bridges is an i2c controller. This patch adds the driver for that i2c controller.

Please let me know if you see any problems with this patch.

Also, if you're not the correct person(s), please point me to who is.

Thanks,

Mark

Signed-off-by Mark A. Greer <[email protected]>

--

drivers/i2c/busses/Kconfig +10 -0
diff -Nru a/drivers/i2c/busses/Kconfig b/drivers/i2c/busses/Kconfig
@@ -476,4 +476,14 @@
 	  This driver can also be built as a module.  If so, the module
 	  will be called i2c-pca-isa.
 
+config I2C_MV64XXX
+	tristate "Marvell mv64xxx I2C Controller"
+	depends on I2C && MV64X60
+	help
+	  If you say yes to this option, support will be included for the
+	  built-in I2C interface on the Marvell 64xxx line of host bridges
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-mv64xxx.
+
 endmenu
drivers/i2c/busses/Makefile +1 -0
diff -Nru a/drivers/i2c/busses/Makefile b/drivers/i2c/busses/Makefile
@@ -20,6 +20,7 @@
 obj-$(CONFIG_I2C_IXP4XX)	+= i2c-ixp4xx.o
 obj-$(CONFIG_I2C_KEYWEST)	+= i2c-keywest.o
 obj-$(CONFIG_I2C_MPC)		+= i2c-mpc.o
+obj-$(CONFIG_I2C_MV64XXX)	+= i2c-mv64xxx.o
 obj-$(CONFIG_I2C_NFORCE2)	+= i2c-nforce2.o
 obj-$(CONFIG_I2C_PARPORT)	+= i2c-parport.o
 obj-$(CONFIG_I2C_PARPORT_LIGHT)	+= i2c-parport-light.o
drivers/i2c/busses/i2c-mv64xxx.c new file +550 -0

550 lines not on this page · raw message

drivers/i2c/busses/i2c-mv64xxx.h new file +99 -0

99 lines not on this page · raw message

include/linux/i2c-id.h +4 -0
diff -Nru a/include/linux/i2c-id.h b/include/linux/i2c-id.h
@@ -200,6 +200,7 @@
 
 #define I2C_ALGO_SIBYTE 0x150000	/* Broadcom SiByte SOCs		*/
 #define I2C_ALGO_SGI	0x160000        /* SGI algorithm                */
+#define I2C_ALGO_MV64XXX 0x170000       /* Marvell mv64xxx i2c ctlr     */
 
 #define I2C_ALGO_EXP	0x800000	/* experimental			*/
 

@@ -304,5 +305,8 @@
 
 /* --- MCP107 adapter */
 #define I2C_HW_MPC107 0x00
+
+/* --- Marvell mv64xxx i2c adapter */
+#define I2C_HW_MV64XXX 0x00
 
 #endif /* LINUX_I2C_ID_H */
include/linux/mv643xx.h +11 -6
diff -Nru a/include/linux/mv643xx.h b/include/linux/mv643xx.h
@@ -977,12 +977,9 @@
 /* I2C Registers                        */
 /****************************************/
 
-#define MV64340_I2C_SLAVE_ADDR                                      0xc000
-#define MV64340_I2C_EXTENDED_SLAVE_ADDR                             0xc010
-#define MV64340_I2C_DATA                                            0xc004
-#define MV64340_I2C_CONTROL                                         0xc008
-#define MV64340_I2C_STATUS_BAUDE_RATE                               0xc00C
-#define MV64340_I2C_SOFT_RESET                                      0xc01c
+#define	MV64XXX_I2C_CTLR_NAME					"mv64xxx i2c"
+#define MV64XXX_I2C_OFFSET					    0xc000
+#define MV64XXX_I2C_REG_BLOCK_SIZE				    0x0020
 
 /****************************************/
 /* GPP Interface Registers              */

@@ -1083,6 +1080,14 @@
 	u8	brg_can_tune;
 	u8	brg_clk_src;
 	u32	brg_clk_freq;
+};
+
+/* i2c Platform Device, Driver Data */
+struct mv64xxx_i2c_pdata {
+	u32	freq_m;
+	u32	freq_n;
+	u32	timeout;	/* In milliseconds */
+	u32	retries;
 };
 
 #endif /* __ASM_MV64340_H */
#1 lore raw

Hi Mark,

Quoted 5 lines

Marvell makes a line of host bridge for PPC and MIPS systems. On those bridges is an i2c controller. This patch adds the driver for that i2c controller.

Please let me know if you see any problems with this patch.

I do not feel qualified for a full review of this code. However, I noticed the following minor issues:

drivers/i2c/busses/Kconfig line 481

Quoted 3 lines
drivers/i2c/busses/Kconfig +3 -0
+config I2C_MV64XXX
+	tristate "Marvell mv64xxx I2C Controller"
+	depends on I2C && MV64X60

&& EXPERIMENTAL?

include/linux/i2c-id.h line 203

Quoted 8 lines

diff -Nru a/include/linux/i2c-id.h b/include/linux/i2c-id.h

include/linux/i2c-id.h +1 -0
@@ -200,6 +200,7 @@
 
 #define I2C_ALGO_SIBYTE 0x150000	/* Broadcom SiByte SOCs		*/
 #define I2C_ALGO_SGI	0x160000        /* SGI algorithm                */
+#define I2C_ALGO_MV64XXX 0x170000       /* Marvell mv64xxx i2c ctlr     */

0x170000 is reserved within the legacy i2c project for an USB algorithm, and 0x180000 for virtual busses. Could you please use 0x190000 instead, so as to avoid future collisions?

include/linux/i2c-id.h

Quoted 9 lines
include/linux/i2c-id.h +3 -6
-#define MV64340_I2C_SLAVE_ADDR                                      0xc000
-#define MV64340_I2C_EXTENDED_SLAVE_ADDR                             0xc010
-#define MV64340_I2C_DATA                                            0xc004
-#define MV64340_I2C_CONTROL                                         0xc008
-#define MV64340_I2C_STATUS_BAUDE_RATE                               0xc00C
-#define MV64340_I2C_SOFT_RESET                                      0xc01c
+#define	MV64XXX_I2C_CTLR_NAME					"mv64xxx i2c"
+#define MV64XXX_I2C_OFFSET					    0xc000
+#define MV64XXX_I2C_REG_BLOCK_SIZE				    0x0020

You have a tab instead of space before MV64XXX_I2C_CTLR_NAME, it seems. Also, you want to align the numerical values using only tabs, no space.

Thanks,

Signature · 2 lines
Jean Delvare
http://khali.linux-fr.org/
author
#2 lore raw
Quoted 8 lines · Jean Delvare

Review

Quoted 5 lines
(quoted diff) +3 -0
+config I2C_MV64XXX
+	tristate "Marvell mv64xxx I2C Controller"
+	depends on I2C && MV64X60

&& EXPERIMENTAL?

Done.

Quoted 8 lines
Quoted 3 lines

+#define I2C_ALGO_MV64XXX 0x170000 /* Marvell mv64xxx i2c ctlr */

0x170000 is reserved within the legacy i2c project for an USB algorithm, and 0x180000 for virtual busses. Could you please use 0x190000 instead, so as to avoid future collisions?

Done.

Quoted 11 lines
(quoted diff) +3 -0
+#define	MV64XXX_I2C_CTLR_NAME					"mv64xxx i2c"
+#define MV64XXX_I2C_OFFSET					    0xc000
+#define MV64XXX_I2C_REG_BLOCK_SIZE				    0x0020

You have a tab instead of space before MV64XXX_I2C_CTLR_NAME, it seems. Also, you want to align the numerical values using only tabs, no space.

Actually, I did use tabs but what was already there used spaces. So the patch won't mess up the alignment, I changed mine to spaces too.

The attached patch is a complete replacement for the one posted earlier.

Signed-off-by Mark A. Greer <[email protected]>
drivers/i2c/busses/Kconfig +10 -0
diff -Nru a/drivers/i2c/busses/Kconfig b/drivers/i2c/busses/Kconfig
@@ -476,4 +476,14 @@
 	  This driver can also be built as a module.  If so, the module
 	  will be called i2c-pca-isa.
 
+config I2C_MV64XXX
+	tristate "Marvell mv64xxx I2C Controller"
+	depends on I2C && MV64X60 && EXPERIMENTAL
+	help
+	  If you say yes to this option, support will be included for the
+	  built-in I2C interface on the Marvell 64xxx line of host bridges
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-mv64xxx.
+
 endmenu
drivers/i2c/busses/Makefile +1 -0
diff -Nru a/drivers/i2c/busses/Makefile b/drivers/i2c/busses/Makefile
@@ -20,6 +20,7 @@
 obj-$(CONFIG_I2C_IXP4XX)	+= i2c-ixp4xx.o
 obj-$(CONFIG_I2C_KEYWEST)	+= i2c-keywest.o
 obj-$(CONFIG_I2C_MPC)		+= i2c-mpc.o
+obj-$(CONFIG_I2C_MV64XXX)	+= i2c-mv64xxx.o
 obj-$(CONFIG_I2C_NFORCE2)	+= i2c-nforce2.o
 obj-$(CONFIG_I2C_PARPORT)	+= i2c-parport.o
 obj-$(CONFIG_I2C_PARPORT_LIGHT)	+= i2c-parport-light.o
drivers/i2c/busses/i2c-mv64xxx.c new file +550 -0

550 lines not on this page · raw message

drivers/i2c/busses/i2c-mv64xxx.h new file +99 -0

99 lines not on this page · raw message

include/linux/i2c-id.h +6 -0
diff -Nru a/include/linux/i2c-id.h b/include/linux/i2c-id.h
@@ -200,6 +200,9 @@
 
 #define I2C_ALGO_SIBYTE 0x150000	/* Broadcom SiByte SOCs		*/
 #define I2C_ALGO_SGI	0x160000        /* SGI algorithm                */
+					/* 0x170000 - USB		*/
+					/* 0x180000 - Virtual buses	*/
+#define I2C_ALGO_MV64XXX 0x190000       /* Marvell mv64xxx i2c ctlr	*/
 
 #define I2C_ALGO_EXP	0x800000	/* experimental			*/
 

@@ -304,5 +307,8 @@
 
 /* --- MCP107 adapter */
 #define I2C_HW_MPC107 0x00
+
+/* --- Marvell mv64xxx i2c adapter */
+#define I2C_HW_MV64XXX 0x00
 
 #endif /* LINUX_I2C_ID_H */
include/linux/mv643xx.h +11 -6
diff -Nru a/include/linux/mv643xx.h b/include/linux/mv643xx.h
@@ -977,12 +977,9 @@
 /* I2C Registers                        */
 /****************************************/
 
-#define MV64340_I2C_SLAVE_ADDR                                      0xc000
-#define MV64340_I2C_EXTENDED_SLAVE_ADDR                             0xc010
-#define MV64340_I2C_DATA                                            0xc004
-#define MV64340_I2C_CONTROL                                         0xc008
-#define MV64340_I2C_STATUS_BAUDE_RATE                               0xc00C
-#define MV64340_I2C_SOFT_RESET                                      0xc01c
+#define MV64XXX_I2C_CTLR_NAME					"mv64xxx i2c"
+#define MV64XXX_I2C_OFFSET                                          0xc000
+#define MV64XXX_I2C_REG_BLOCK_SIZE                                  0x0020
 
 /****************************************/
 /* GPP Interface Registers              */

@@ -1083,6 +1080,14 @@
 	u8	brg_can_tune;
 	u8	brg_clk_src;
 	u32	brg_clk_freq;
+};
+
+/* i2c Platform Device, Driver Data */
+struct mv64xxx_i2c_pdata {
+	u32	freq_m;
+	u32	freq_n;
+	u32	timeout;	/* In milliseconds */
+	u32	retries;
 };
 
 #endif /* __ASM_MV64340_H */
#3 lore raw
Quoted 5 lines · Mark A. Greer
drivers/i2c/busses/i2c-mv64xxx.c new file +5 -0
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/init.h>

drivers/i2c/busses/i2c-mv64xxx.c line 28

Quoted 10 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +10 -0
+#include <linux/pci.h>
+#include <linux/wait.h>
+#include <linux/spinlock.h>
+#include <asm/io.h>
+#include <asm/ocp.h>
+#include <linux/i2c.h>
+#include <linux/interrupt.h>
+#include <linux/delay.h>
+#include <linux/mv643xx.h>
+#include "i2c-mv64xxx.h"

Please put <asm/ after <linux/

drivers/i2c/busses/i2c-mv64xxx.c line 41

Quoted 5 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +5 -0
+static inline void
+mv64xxx_i2c_fsm(struct mv64xxx_i2c_data *drv_data, u32 status)
+{
+	pr_debug("mv64xxx_i2c_fsm: ENTER--state: %d, status: 0x%x\n",
+		drv_data->state, status);

You have a lot of pr_debug and other printk() for stuff in this driver. Please use dev_dbg(), dev_err() and friends instead. That way you get a consistant message, that points to the exact device that is causing the message.

drivers/i2c/busses/i2c-mv64xxx.c line 257

Quoted 3 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +3 -0
+static inline void
+mv64xxx_i2c_prepare_for_io(struct mv64xxx_i2c_data *drv_data,
+	struct i2c_msg *msg)

You have some big inline functions here. Should they really be inline? We aren't really worried about speed here, right? Size is probably a bigger issue.

drivers/i2c/busses/i2c-mv64xxx.h

Quoted 3 lines

diff -Nru a/drivers/i2c/busses/i2c-mv64xxx.h b/drivers/i2c/busses/i2c-mv64xxx.h

drivers/i2c/busses/i2c-mv64xxx.h new file +0 -0

Is this header file really needed? Does any other file other than this single driver ever include it? If not, please just put it into the driver itself.

thanks,

greg k-h

author
#4 lore raw
Quoted 2 lines · Greg KH

Please put <asm/ after <linux/

Done.

Quoted 5 lines

You have a lot of pr_debug and other printk() for stuff in this driver. Please use dev_dbg(), dev_err() and friends instead. That way you get a consistant message, that points to the exact device that is causing the message.

Cool. Done.

Quoted 4 lines

You have some big inline functions here. Should they really be inline? We aren't really worried about speed here, right? Size is probably a bigger issue.

No, no, and done.

Quoted 6 lines

Is this header file really needed? Does any other file other than this single driver ever include it? If not, please just put it into the driver itself.

No, no, and done.

Included is an *incremental* patch that I hope addresses your concerns.

Thanks Greg.

Mark
--

drivers/i2c/busses/i2c-mv64xxx.c +98 -34
diff -Nru a/drivers/i2c/busses/i2c-mv64xxx.c b/drivers/i2c/busses/i2c-mv64xxx.c
@@ -11,21 +11,94 @@
  * is licensed "as is" without any warranty of any kind, whether express
  * or implied.
  */
-#include <linux/config.h>
 #include <linux/kernel.h>
 #include <linux/module.h>
-#include <linux/sched.h>
-#include <linux/init.h>
-#include <linux/pci.h>
-#include <linux/wait.h>
 #include <linux/spinlock.h>
-#include <asm/io.h>
-#include <asm/ocp.h>
 #include <linux/i2c.h>
 #include <linux/interrupt.h>
-#include <linux/delay.h>
 #include <linux/mv643xx.h>
-#include "i2c-mv64xxx.h"
+#include <asm/io.h>
+
+/* Register defines */
+#define	MV64XXX_I2C_REG_SLAVE_ADDR			0x00
+#define	MV64XXX_I2C_REG_DATA				0x04
+#define	MV64XXX_I2C_REG_CONTROL				0x08
+#define	MV64XXX_I2C_REG_STATUS				0x0c
+#define	MV64XXX_I2C_REG_BAUD				0x0c
+#define	MV64XXX_I2C_REG_EXT_SLAVE_ADDR			0x10
+#define	MV64XXX_I2C_REG_SOFT_RESET			0x1c
+
+#define	MV64XXX_I2C_REG_CONTROL_ACK			0x00000004
+#define	MV64XXX_I2C_REG_CONTROL_IFLG			0x00000008
+#define	MV64XXX_I2C_REG_CONTROL_STOP			0x00000010
+#define	MV64XXX_I2C_REG_CONTROL_START			0x00000020
+#define	MV64XXX_I2C_REG_CONTROL_TWSIEN			0x00000040
+#define	MV64XXX_I2C_REG_CONTROL_INTEN			0x00000080
+
+/* Ctlr status values */
+#define	MV64XXX_I2C_STATUS_BUS_ERR			0x00
+#define	MV64XXX_I2C_STATUS_MAST_START			0x08
+#define	MV64XXX_I2C_STATUS_MAST_REPEAT_START		0x10
+#define	MV64XXX_I2C_STATUS_MAST_WR_ADDR_ACK		0x18
+#define	MV64XXX_I2C_STATUS_MAST_WR_ADDR_NO_ACK		0x20
+#define	MV64XXX_I2C_STATUS_MAST_WR_ACK			0x28
+#define	MV64XXX_I2C_STATUS_MAST_WR_NO_ACK		0x30
+#define	MV64XXX_I2C_STATUS_MAST_LOST_ARB		0x38
+#define	MV64XXX_I2C_STATUS_MAST_RD_ADDR_ACK		0x40
+#define	MV64XXX_I2C_STATUS_MAST_RD_ADDR_NO_ACK		0x48
+#define	MV64XXX_I2C_STATUS_MAST_RD_DATA_ACK		0x50
+#define	MV64XXX_I2C_STATUS_MAST_RD_DATA_NO_ACK		0x58
+#define	MV64XXX_I2C_STATUS_MAST_WR_ADDR_2_ACK		0xd0
+#define	MV64XXX_I2C_STATUS_MAST_WR_ADDR_2_NO_ACK	0xd8
+#define	MV64XXX_I2C_STATUS_MAST_RD_ADDR_2_ACK		0xe0
+#define	MV64XXX_I2C_STATUS_MAST_RD_ADDR_2_NO_ACK	0xe8
+#define	MV64XXX_I2C_STATUS_NO_STATUS			0xf8
+
+/* Driver states */
+enum {
+	MV64XXX_I2C_STATE_INVALID,
+	MV64XXX_I2C_STATE_IDLE,
+	MV64XXX_I2C_STATE_WAITING_FOR_START_COND,
+	MV64XXX_I2C_STATE_WAITING_FOR_ADDR_1_ACK,
+	MV64XXX_I2C_STATE_WAITING_FOR_ADDR_2_ACK,
+	MV64XXX_I2C_STATE_WAITING_FOR_SLAVE_ACK,
+	MV64XXX_I2C_STATE_WAITING_FOR_SLAVE_DATA,
+	MV64XXX_I2C_STATE_ABORTING,
+};
+
+/* Driver actions */
+enum {
+	MV64XXX_I2C_ACTION_INVALID,
+	MV64XXX_I2C_ACTION_CONTINUE,
+	MV64XXX_I2C_ACTION_SEND_START,
+	MV64XXX_I2C_ACTION_SEND_ADDR_1,
+	MV64XXX_I2C_ACTION_SEND_ADDR_2,
+	MV64XXX_I2C_ACTION_SEND_DATA,
+	MV64XXX_I2C_ACTION_RCV_DATA,
+	MV64XXX_I2C_ACTION_RCV_DATA_STOP,
+	MV64XXX_I2C_ACTION_SEND_STOP,
+};
+
+struct mv64xxx_i2c_data {
+	int			irq;
+	uint			state;
+	uint			action;
+	u32			cntl_bits;
+	void			*reg_base;
+	ulong			reg_base_p;
+	u32			addr1;
+	u32			addr2;
+	uint			bytes_left;
+	uint			byte_posn;
+	uint			block;
+	int			rc;
+	u32			freq_m;
+	u32			freq_n;
+	wait_queue_head_t	waitq;
+	spinlock_t		lock;
+	struct i2c_msg		*msg;
+	struct i2c_adapter	adapter;
+};
 
 /*
  *****************************************************************************

@@ -34,12 +107,9 @@
  *
  *****************************************************************************
  */
-static inline void
+static void
 mv64xxx_i2c_fsm(struct mv64xxx_i2c_data *drv_data, u32 status)
 {
-	pr_debug("mv64xxx_i2c_fsm: ENTER--state: %d, status: 0x%x\n",
-		drv_data->state, status);
-
 	/*
 	 * If state is idle, then this is likely the remnants of an old
 	 * operation that driver has given up on or the user has killed.

@@ -48,14 +118,12 @@
 	if (drv_data->state == MV64XXX_I2C_STATE_IDLE) {
 		drv_data->action = MV64XXX_I2C_ACTION_SEND_STOP;
 		drv_data->state = MV64XXX_I2C_STATE_IDLE;
-		pr_debug("mv64xxx_i2c_fsm: EXIT--Entered when in IDLE state\n");
 		return;
 	}
 
 	if (drv_data->state == MV64XXX_I2C_STATE_ABORTING) {
 		drv_data->action = MV64XXX_I2C_ACTION_SEND_STOP;
 		drv_data->state = MV64XXX_I2C_STATE_IDLE;
-		pr_debug("mv64xxx_i2c_fsm: EXIT--Aborting\n");
 		return;
 	}
 

@@ -135,27 +203,22 @@
 		break;
 
 	default:
-		printk(KERN_ERR "mv64xxx_i2c_fsm: Ctlr Error -- "
-			"state: 0x%x, status: 0x%x\n", drv_data->state, status);
-		printk(KERN_INFO "addr: 0x%x, flags: 0x%x\n",
-			drv_data->msg->addr, drv_data->msg->flags);
+		dev_err(&drv_data->adapter.dev,
+			"mv64xxx_i2c_fsm: Ctlr Error -- state: 0x%x, "
+			"status: 0x%x, addr: 0x%x, flags: 0x%x\n",
+			 drv_data->state, status, drv_data->msg->addr,
+			 drv_data->msg->flags);
 		drv_data->action = MV64XXX_I2C_ACTION_SEND_STOP;
 		drv_data->state = MV64XXX_I2C_STATE_IDLE;
 		drv_data->rc = -EIO;
 	}
 
-	pr_debug("mv64xxx_i2c_fsm: EXIT--action: %d, state: %d, rc: 0x%x\n",
-		drv_data->action, drv_data->state, drv_data->rc);
 	return;
 }
 
 static void
 mv64xxx_i2c_do_action(struct mv64xxx_i2c_data *drv_data)
 {
-	pr_debug("mv64xxx_i2c_do_action: ENTER--action: %d, state: %d, "
-		"cntl: 0x%x\n", drv_data->action, drv_data->state,
-		drv_data->cntl_bits);
-
 	switch(drv_data->action) {
 	case MV64XXX_I2C_ACTION_CONTINUE:
 		writel(drv_data->cntl_bits,

@@ -207,7 +270,8 @@
 
 	case MV64XXX_I2C_ACTION_INVALID:
 	default:
-		printk(KERN_ERR "mv64xxx_i2c_do_action: Invalid action: %d\n",
+		dev_err(&drv_data->adapter.dev,
+			"mv64xxx_i2c_do_action: Invalid action: %d\n",
 			drv_data->action);
 		drv_data->rc = -EIO;
 		/* FALLTHRU */

@@ -220,7 +284,6 @@
 		break;
 	}
 
-	pr_debug("mv64xxx_i2c_do_action: EXIT\n");
 	return;
 }
 

@@ -252,7 +315,7 @@
  *
  *****************************************************************************
  */
-static inline void
+static void
 mv64xxx_i2c_prepare_for_io(struct mv64xxx_i2c_data *drv_data,
 	struct i2c_msg *msg)
 {

@@ -283,7 +346,7 @@
 	return;
 }
 
-static inline void
+static void
 mv64xxx_i2c_wait_for_completion(struct mv64xxx_i2c_data *drv_data)
 {
 	long	flags, time_left;

@@ -312,7 +375,8 @@
 
 		if (!time_left <= 0) {
 			drv_data->state = MV64XXX_I2C_STATE_IDLE;
-			printk(KERN_WARNING "mv64xxx: I2C bus locked\n");
+			dev_err(&drv_data->adapter.dev,
+				"mv64xxx: I2C bus locked\n");
 		}
 	}
 	else

@@ -321,7 +385,7 @@
 	return;
 }
 
-static inline int
+static int
 mv64xxx_i2c_execute_msg(struct mv64xxx_i2c_data *drv_data, struct i2c_msg *msg)
 {
 	long	flags;

@@ -484,7 +548,7 @@
 		if (request_irq(drv_data->irq, mv64xxx_i2c_intr, 0,
 			MV64XXX_I2C_CTLR_NAME, drv_data)) {
 
-			printk(KERN_ERR "mv64xxx: Can't register intr handler "
+			dev_err(dev, "mv64xxx: Can't register intr handler "
 				"irq: %d\\n", drv_data->irq);
 
 			mv64xxx_i2c_unmap_regs(drv_data);

@@ -492,8 +556,8 @@
 			return -EINVAL;
 		}
 		else if ((rc = i2c_add_adapter(&drv_data->adapter)) != 0) {
-			printk(KERN_WARNING "mv64xxx: Can't add i2c adapter "
-				"rc: %d\n", -rc);
+			dev_err(dev, "mv64xxx: Can't add i2c adapter, rc: %d\n",
+				-rc);
 			free_irq(drv_data->irq, drv_data);
 			mv64xxx_i2c_unmap_regs(drv_data);
 			kfree(drv_data);
drivers/i2c/busses/i2c-mv64xxx.h deleted +0 -99

99 lines not on this page · raw message

author
#5 lore raw
In reply to Jean Delvare
Quoted 2 lines · Jean Delvare

Hi Mark,

Thanks for the commenting, Jean.

Quoted 4 lines

<snip>

Quoted 8 lines

drivers/i2c/busses/Kconfig line 481

Quoted 5 lines
drivers/i2c/busses/Kconfig +3 -0
+config I2C_MV64XXX
+	tristate "Marvell mv64xxx I2C Controller"
+	depends on I2C && MV64X60

&& EXPERIMENTAL?

Yes, I guess that's the correct thing to do. I'll add that.

Quoted 5 lines
Quoted 5 lines

diff -Nru a/include/linux/i2c-id.h b/include/linux/i2c-id.h

include/linux/i2c-id.h +0 -0
@@ -200,6 +200,7 @@
Quoted 10 lines

include/linux/i2c-id.h line 206

Quoted 5 lines
include/linux/i2c-id.h +1 -0
 #define I2C_ALGO_SIBYTE 0x150000	/* Broadcom SiByte SOCs		*/
 #define I2C_ALGO_SGI	0x160000        /* SGI algorithm                */
+#define I2C_ALGO_MV64XXX 0x170000       /* Marvell mv64xxx i2c ctlr     */

0x170000 is reserved within the legacy i2c project for an USB algorithm, and 0x180000 for virtual busses. Could you please use 0x190000 instead, so as to avoid future collisions?

Absolutely. I was unaware of the other uses.

Quoted 13 lines

include/linux/i2c-id.h

Quoted 8 lines
include/linux/i2c-id.h +3 -1
-#define MV64340_I2C_SOFT_RESET                                      0xc01c
+#define	MV64XXX_I2C_CTLR_NAME					"mv64xxx i2c"
+#define MV64XXX_I2C_OFFSET					    0xc000
+#define MV64XXX_I2C_REG_BLOCK_SIZE				    0x0020

You have a tab instead of space before MV64XXX_I2C_CTLR_NAME, it seems. Also, you want to align the numerical values using only tabs, no space.

Oops. I'll fix that too and repost later today.

Mark

#6 lore raw
In reply to Mark A. Greer

drivers/i2c/busses/i2c-mv64xxx.c line 38

Quoted 2 lines · Mark A. Greer
drivers/i2c/busses/i2c-mv64xxx.c new file +2 -0
+static inline void
+mv64xxx_i2c_fsm(struct mv64xxx_i2c_data *drv_data, u32 status)

This is a much too big of a function to be "inline". Please change it. Same for your other inline functions, that's not really needed, right?

drivers/i2c/busses/i2c-mv64xxx.c line 41

Quoted 3 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +3 -0
+{
+	pr_debug("mv64xxx_i2c_fsm: ENTER--state: %d, status: 0x%x\n",
+		drv_data->state, status);

Please use the dev_* calls instead. It gives you an accurate description of the specific device that emits the messages. Also use it for all of the printk() calls in the driver too.

thanks,

greg k-h

author
#7 lore raw
Quoted 23 lines · Greg KH

drivers/i2c/busses/i2c-mv64xxx.c line 38

Quoted 4 lines · Mark A. Greer
drivers/i2c/busses/i2c-mv64xxx.c new file +2 -0
+static inline void
+mv64xxx_i2c_fsm(struct mv64xxx_i2c_data *drv_data, u32 status)

This is a much too big of a function to be "inline". Please change it. Same for your other inline functions, that's not really needed, right?

drivers/i2c/busses/i2c-mv64xxx.c line 41

Quoted 5 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +3 -0
+{
+	pr_debug("mv64xxx_i2c_fsm: ENTER--state: %d, status: 0x%x\n",
+		drv_data->state, status);

Please use the dev_* calls instead. It gives you an accurate description of the specific device that emits the messages. Also use it for all of the printk() calls in the driver too.

Quoted 4 lines

thanks,

greg k-h

Certainly. I already posted this [incremental] patch based on your previous comments, http://www.ussg.iu.edu/hypermail/linux/kernel/0501.3/0941.html. Is that better?

Mark

#8 lore raw
Quoted 27 lines · Mark A. Greer
Quoted 25 lines · Greg KH

drivers/i2c/busses/i2c-mv64xxx.c line 38

Quoted 4 lines · Mark A. Greer
drivers/i2c/busses/i2c-mv64xxx.c new file +2 -0
+static inline void
+mv64xxx_i2c_fsm(struct mv64xxx_i2c_data *drv_data, u32 status)

This is a much too big of a function to be "inline". Please change it. Same for your other inline functions, that's not really needed, right?

drivers/i2c/busses/i2c-mv64xxx.c line 41

Quoted 5 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +3 -0
+{
+	pr_debug("mv64xxx_i2c_fsm: ENTER--state: %d, status: 0x%x\n",
+		drv_data->state, status);

Please use the dev_* calls instead. It gives you an accurate description of the specific device that emits the messages. Also use it for all of the printk() calls in the driver too.

thanks,

Quoted 7 lines
Quoted 2 lines

greg k-h

Certainly. I already posted this [incremental] patch based on your previous comments,
http://www.ussg.iu.edu/hypermail/linux/kernel/0501.3/0941.html. Is that better?

How about a whole new patch that I could apply? That would be better :)

thanks,

greg k-h

author
#9 lore raw
Quoted 3 lines · Greg KH

How about a whole new patch that I could apply? That would be better :)

Oops, sorry. :)

drivers/i2c/busses/Kconfig +10 -0
diff -Nru a/drivers/i2c/busses/Kconfig b/drivers/i2c/busses/Kconfig
@@ -476,4 +476,14 @@
 	  This driver can also be built as a module.  If so, the module
 	  will be called i2c-pca-isa.
 
+config I2C_MV64XXX
+	tristate "Marvell mv64xxx I2C Controller"
+	depends on I2C && MV64X60 && EXPERIMENTAL
+	help
+	  If you say yes to this option, support will be included for the
+	  built-in I2C interface on the Marvell 64xxx line of host bridges
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-mv64xxx.
+
 endmenu
drivers/i2c/busses/Makefile +1 -0
diff -Nru a/drivers/i2c/busses/Makefile b/drivers/i2c/busses/Makefile
@@ -20,6 +20,7 @@
 obj-$(CONFIG_I2C_IXP4XX)	+= i2c-ixp4xx.o
 obj-$(CONFIG_I2C_KEYWEST)	+= i2c-keywest.o
 obj-$(CONFIG_I2C_MPC)		+= i2c-mpc.o
+obj-$(CONFIG_I2C_MV64XXX)	+= i2c-mv64xxx.o
 obj-$(CONFIG_I2C_NFORCE2)	+= i2c-nforce2.o
 obj-$(CONFIG_I2C_PARPORT)	+= i2c-parport.o
 obj-$(CONFIG_I2C_PARPORT_LIGHT)	+= i2c-parport-light.o
drivers/i2c/busses/i2c-mv64xxx.c new file +614 -0

614 lines not on this page · raw message

include/linux/i2c-id.h +6 -0
diff -Nru a/include/linux/i2c-id.h b/include/linux/i2c-id.h
@@ -200,6 +200,9 @@
 
 #define I2C_ALGO_SIBYTE 0x150000	/* Broadcom SiByte SOCs		*/
 #define I2C_ALGO_SGI	0x160000        /* SGI algorithm                */
+					/* 0x170000 - USB		*/
+					/* 0x180000 - Virtual buses	*/
+#define I2C_ALGO_MV64XXX 0x190000       /* Marvell mv64xxx i2c ctlr	*/
 
 #define I2C_ALGO_EXP	0x800000	/* experimental			*/
 

@@ -305,5 +308,8 @@
 
 /* --- MCP107 adapter */
 #define I2C_HW_MPC107 0x00
+
+/* --- Marvell mv64xxx i2c adapter */
+#define I2C_HW_MV64XXX 0x00
 
 #endif /* LINUX_I2C_ID_H */
include/linux/mv643xx.h +11 -6
diff -Nru a/include/linux/mv643xx.h b/include/linux/mv643xx.h
@@ -977,12 +977,9 @@
 /* I2C Registers                        */
 /****************************************/
 
-#define MV64340_I2C_SLAVE_ADDR                                      0xc000
-#define MV64340_I2C_EXTENDED_SLAVE_ADDR                             0xc010
-#define MV64340_I2C_DATA                                            0xc004
-#define MV64340_I2C_CONTROL                                         0xc008
-#define MV64340_I2C_STATUS_BAUDE_RATE                               0xc00C
-#define MV64340_I2C_SOFT_RESET                                      0xc01c
+#define MV64XXX_I2C_CTLR_NAME					"mv64xxx i2c"
+#define MV64XXX_I2C_OFFSET                                          0xc000
+#define MV64XXX_I2C_REG_BLOCK_SIZE                                  0x0020
 
 /****************************************/
 /* GPP Interface Registers              */

@@ -1083,6 +1080,14 @@
 	u8	brg_can_tune;
 	u8	brg_clk_src;
 	u32	brg_clk_freq;
+};
+
+/* i2c Platform Device, Driver Data */
+struct mv64xxx_i2c_pdata {
+	u32	freq_m;
+	u32	freq_n;
+	u32	timeout;	/* In milliseconds */
+	u32	retries;
 };
 
 #endif /* __ASM_MV64340_H */
#10 lore raw

PATCH 1/1 Marvell mv64xxx i2c driver

In reply to Mark A. Greer

drivers/i2c/busses/i2c-mv64xxx.c

Quoted 5 lines · Mark A. Greer
drivers/i2c/busses/i2c-mv64xxx.c new file +2 -0
 
+struct mv64xxx_i2c_data {
+	void			*reg_base;

ioremap() returns "void __iomem *".

drivers/i2c/busses/i2c-mv64xxx.c line 437

Quoted 5 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +5 -0
+static void __devexit
+mv64xxx_i2c_unmap_regs(struct mv64xxx_i2c_data *drv_data)
+{
+	if (!drv_data->reg_base) {
+		iounmap(drv_data->reg_base);

A typo? You're unmapping known to be invalid address.

drivers/i2c/busses/i2c-mv64xxx.c line 442

Quoted 1 line
drivers/i2c/busses/i2c-mv64xxx.c new file +1 -0
+	drv_data->reg_base = 0;

Use NULL because drv_data->reg_base is a pointer.

drivers/i2c/busses/i2c-mv64xxx.c

Quoted 7 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +5 -0
+};
 
+static int __devinit
+mv64xxx_i2c_probe(struct device *dev)
+{
 
+	if ((pd->id = 0) && pdata) {

Rewrite this as:

	if ((pd->id != 0) || !pdata)
		return -ENODEV;

and save one level of indentation below.

drivers/i2c/busses/i2c-mv64xxx.c line 466

Quoted 9 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +9 -0
+		drv_data = kmalloc(sizeof(struct mv64xxx_i2c_data), GFP_KERNEL);
+		if (!drv_data)
+			return -ENOMEM;
+		memset(drv_data, 0, sizeof(struct mv64xxx_i2c_data));
+
+		if (mv64xxx_i2c_map_regs(pd, drv_data)) {
+			kfree(drv_data);
+			return -ENODEV;
+		}

[snip non-branching stuff]

drivers/i2c/busses/i2c-mv64xxx.c line 488

Quoted 5 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +5 -0
+		if (request_irq(drv_data->irq, mv64xxx_i2c_intr, 0,
+			MV64XXX_I2C_CTLR_NAME, drv_data)) {
+
+			dev_err(dev, "mv64xxx: Can't register intr handler "
+				"irq: %d\\n", drv_data->irq);

"\n".

drivers/i2c/busses/i2c-mv64xxx.c line 501

Quoted 13 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +13 -0
+
+			mv64xxx_i2c_unmap_regs(drv_data);
+			kfree(drv_data);
+			return -EINVAL;
+		}
+		else if ((rc = i2c_add_adapter(&drv_data->adapter)) != 0) {
+			dev_err(dev, "mv64xxx: Can't add i2c adapter, rc: %d\n",
+				-rc);
+			free_irq(drv_data->irq, drv_data);
+			mv64xxx_i2c_unmap_regs(drv_data);
+			kfree(drv_data);
+			return rc;
+		}

		[snip]

drivers/i2c/busses/i2c-mv64xxx.c line 60

Quoted 3 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +2 -0
+	}
 
+}

Use centralized exiting and goto's. Something like:

	if (mv64xxx_i2c_map_regs(pd, drv_data)) {
		rc = -ENODEV;
		goto exit_free;
	}
	if (request_irq) {
		rc = -EINVAL;
		dev_err
		goto exit_unmap_regs
	}
	if (rc = i2c_add_adapter) {
		dev_err
		goto exit_free_irq
	}
	mv64xxx_i2c_hw_init

	return 0;

	exit_free_irq:
		free_irq
	exit_unmap_regs:
		mv64xxx_i2c_unmap_regs
	exit_free:
		kfree
	return rc;

					Alexey

P. S.: struct mv64xxx_i2c_data revisited...

drivers/i2c/busses/i2c-mv64xxx.c

Quoted 2 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +2 -0
+	uint			state;
+	ulong			reg_base_p;

Silly request, but... Maybe this should be changed to plain old "unsigned int" and "unsigned long". Please. I just don't understand why people use "uint", "u_int", "uInt", "UINT", "uINT", "uint_t" which are always typedef'ed to "unsigned int".

#11 lore raw
Quoted 9 lines · Alexey Dobriyan

P. S.: struct mv64xxx_i2c_data revisited...

drivers/i2c/busses/i2c-mv64xxx.h line 84

Quoted 2 lines
drivers/i2c/busses/i2c-mv64xxx.h new file +2 -0
+	uint			state;
+	ulong			reg_base_p;

Silly request, but... Maybe this should be changed to plain old "unsigned int" and "unsigned long". Please. I just don't understand why people use "uint", "u_int", "uInt", "UINT", "uINT", "uint_t" which are always typedef'ed to "unsigned int".

Not a silly request at all. Please use the u32, u64 and so on values instead. That way we know what you mean, and it's portable.

thanks,

greg k-h

author
#12 lore raw
In reply to Alexey Dobriyan
Quoted 11 lines · Alexey Dobriyan

Review

Quoted 4 lines · Mark A. Greer
(quoted diff) +2 -0
+struct mv64xxx_i2c_data {
+	void			*reg_base;

ioremap() returns "void __iomem *".

Okay.

Quoted 12 lines

Review

Quoted 9 lines
(quoted diff) +5 -0
+static void __devexit
+mv64xxx_i2c_unmap_regs(struct mv64xxx_i2c_data *drv_data)
+{
+	if (!drv_data->reg_base) {
+		iounmap(drv_data->reg_base);

A typo? You're unmapping known to be invalid address.

Yes, a typo. Good catch.

Quoted 8 lines

Review

Quoted 5 lines
(quoted diff) +1 -0
+	drv_data->reg_base = 0;

Use NULL because drv_data->reg_base is a pointer.

Done.

Quoted 8 lines

Review

Quoted 5 lines
(quoted diff) +1 -0
+	if ((pd->id == 0) && pdata) {

Rewrite this as:

Quoted 5 lines
	if ((pd->id != 0) || !pdata)
		return -ENODEV;

and save one level of indentation below.

Done.

Quoted 3 lines

<snip of xxx_probe modifications>

Done.

Quoted 7 lines

Alexey

P. S.: struct mv64xxx_i2c_data revisited...

Quoted 12 lines

Review

Quoted 4 lines
(quoted diff) +2 -0
+	uint			state;
+	ulong			reg_base_p;

Silly request, but... Maybe this should be changed to plain old "unsigned int" and "unsigned long". Please. I just don't understand why people use "uint", "u_int", "uInt", "UINT", "uINT", "uint_t" which are always typedef'ed to "unsigned int".

Thanks Alexey & Greg. How's this (a complete replacement for previous patch)?

Signed-off-by Mark A. Greer <[email protected]>

--

drivers/i2c/busses/Kconfig +10 -0
diff -Nru a/drivers/i2c/busses/Kconfig b/drivers/i2c/busses/Kconfig
@@ -476,4 +476,14 @@
 	  This driver can also be built as a module.  If so, the module
 	  will be called i2c-pca-isa.
 
+config I2C_MV64XXX
+	tristate "Marvell mv64xxx I2C Controller"
+	depends on I2C && MV64X60 && EXPERIMENTAL
+	help
+	  If you say yes to this option, support will be included for the
+	  built-in I2C interface on the Marvell 64xxx line of host bridges
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-mv64xxx.
+
 endmenu
drivers/i2c/busses/Makefile +1 -0
diff -Nru a/drivers/i2c/busses/Makefile b/drivers/i2c/busses/Makefile
@@ -20,6 +20,7 @@
 obj-$(CONFIG_I2C_IXP4XX)	+= i2c-ixp4xx.o
 obj-$(CONFIG_I2C_KEYWEST)	+= i2c-keywest.o
 obj-$(CONFIG_I2C_MPC)		+= i2c-mpc.o
+obj-$(CONFIG_I2C_MV64XXX)	+= i2c-mv64xxx.o
 obj-$(CONFIG_I2C_NFORCE2)	+= i2c-nforce2.o
 obj-$(CONFIG_I2C_PARPORT)	+= i2c-parport.o
 obj-$(CONFIG_I2C_PARPORT_LIGHT)	+= i2c-parport-light.o
drivers/i2c/busses/i2c-mv64xxx.c new file +615 -0

615 lines not on this page · raw message

include/linux/i2c-id.h +6 -0
diff -Nru a/include/linux/i2c-id.h b/include/linux/i2c-id.h
@@ -200,6 +200,9 @@
 
 #define I2C_ALGO_SIBYTE 0x150000	/* Broadcom SiByte SOCs		*/
 #define I2C_ALGO_SGI	0x160000        /* SGI algorithm                */
+					/* 0x170000 - USB		*/
+					/* 0x180000 - Virtual buses	*/
+#define I2C_ALGO_MV64XXX 0x190000       /* Marvell mv64xxx i2c ctlr	*/
 
 #define I2C_ALGO_EXP	0x800000	/* experimental			*/
 

@@ -305,5 +308,8 @@
 
 /* --- MCP107 adapter */
 #define I2C_HW_MPC107 0x00
+
+/* --- Marvell mv64xxx i2c adapter */
+#define I2C_HW_MV64XXX 0x00
 
 #endif /* LINUX_I2C_ID_H */
include/linux/mv643xx.h +11 -6
diff -Nru a/include/linux/mv643xx.h b/include/linux/mv643xx.h
@@ -977,12 +977,9 @@
 /* I2C Registers                        */
 /****************************************/
 
-#define MV64340_I2C_SLAVE_ADDR                                      0xc000
-#define MV64340_I2C_EXTENDED_SLAVE_ADDR                             0xc010
-#define MV64340_I2C_DATA                                            0xc004
-#define MV64340_I2C_CONTROL                                         0xc008
-#define MV64340_I2C_STATUS_BAUDE_RATE                               0xc00C
-#define MV64340_I2C_SOFT_RESET                                      0xc01c
+#define MV64XXX_I2C_CTLR_NAME					"mv64xxx i2c"
+#define MV64XXX_I2C_OFFSET                                          0xc000
+#define MV64XXX_I2C_REG_BLOCK_SIZE                                  0x0020
 
 /****************************************/
 /* GPP Interface Registers              */

@@ -1083,6 +1080,14 @@
 	u8	brg_can_tune;
 	u8	brg_clk_src;
 	u32	brg_clk_freq;
+};
+
+/* i2c Platform Device, Driver Data */
+struct mv64xxx_i2c_pdata {
+	u32	freq_m;
+	u32	freq_n;
+	u32	timeout;	/* In milliseconds */
+	u32	retries;
 };
 
 #endif /* __ASM_MV64340_H */
#13 lore raw

PATCH 1/1 Marvell mv64xxx i2c driver

In reply to Mark A. Greer

drivers/i2c/busses/Kconfig line 484

Quoted 7 lines · Mark A. Greer

How's this (a complete replacement for previous patch)?

drivers/i2c/busses/Kconfig +2 -0
 
+	  If you say yes to this option, support will be included for the
+	  built-in I2C interface on the Marvell 64xxx line of host bridges

Dot at the end. Should have noticed it earlier...

drivers/i2c/busses/i2c-mv64xxx.c

Quoted 11 lines
drivers/i2c/busses/Kconfig +1 -0
+	  This driver can also be built as a module.  If so, the module
drivers/i2c/busses/i2c-mv64xxx.c new file +5 -0
 
+static void
+mv64xxx_i2c_fsm(struct mv64xxx_i2c_data *drv_data, u32 status)
 
+	if (drv_data->state = MV64XXX_I2C_STATE_IDLE) {
+		drv_data->action = MV64XXX_I2C_ACTION_SEND_STOP;
+		drv_data->state = MV64XXX_I2C_STATE_IDLE;

drv_data->state is already MV64XXX_I2C_STATE_IDLE. Gcc will probably optimize this line away, but...

drivers/i2c/busses/i2c-mv64xxx.c

Quoted 2 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +2 -0
+		}
+		else {

It should be "} else {" according to CodingStyle.

drivers/i2c/busses/i2c-mv64xxx.c line 239

Quoted 7 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +6 -0
+static int
+mv64xxx_i2c_intr(int irq, void *dev_id, struct pt_regs *regs)
 
+	while (readl(drv_data->reg_base + MV64XXX_I2C_REG_CONTROL) &
+						MV64XXX_I2C_REG_CONTROL_IFLG) {
+		status = readl(drv_data->reg_base + MV64XXX_I2C_REG_STATUS);
+		mv64xxx_i2c_fsm(drv_data, status);

It can set drv_data->rc to -ENODEV or -EIO. In both cases ->action goes to MV64XXX_I2C_ACTION_SEND_STOP and mv64xxx_i2c_do_action() will writel() something. Is it correct to _not_ check ->rc here? If it isn't then would it be better to make all functions that set it return -E___ instead and drop struct mv64xxx_i2c_data.rc altogether?

drivers/i2c/busses/i2c-mv64xxx.c line 313

Quoted 10 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +8 -0
+		mv64xxx_i2c_do_action(drv_data);
 
+static void
+mv64xxx_i2c_wait_for_completion(struct mv64xxx_i2c_data *drv_data)
 
+		time_left = wait_event_timeout(drv_data->waitq,
+			!drv_data->block,
+			msecs_to_jiffies(drv_data->adapter.timeout));
+
+		if (!time_left <= 0) {

Confusing. You meant "if (time_left)" or "if (time_left > 0)"?

drivers/i2c/busses/i2c-mv64xxx.c line 386

Quoted 7 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +6 -0
+			drv_data->state = MV64XXX_I2C_STATE_IDLE;
+			dev_err(&drv_data->adapter.dev,
+				"mv64xxx: I2C bus locked\n");
+		}
 
+static struct i2c_algorithm mv64xxx_i2c_algo = {
+	.name = MV64XXX_I2C_CTLR_NAME "algorithm",

MV64XXX_I2C_CTLR_NAME doesn't end with space. " algorithm" here.

drivers/i2c/busses/i2c-mv64xxx.c line 469

Quoted 5 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +4 -0
+static int __devinit
+mv64xxx_i2c_probe(struct device *dev)
 
+	strncpy(drv_data->adapter.name, MV64XXX_I2C_CTLR_NAME "adapter",
+		I2C_NAME_SIZE);

And " adapter" there.

drivers/i2c/busses/i2c-mv64xxx.c

Quoted 2 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +2 -0
+		dev_err(dev, "mv64xxx: Can't register intr handler "
+			"irq: %d\\n", drv_data->irq);

You snipped s# \\n # \n # suggestion in my previous email. ;-)

include/linux/mv643xx.h line 980

Quoted 4 lines
include/linux/mv643xx.h +1 -0
 
+#define MV64XXX_I2C_CTLR_NAME				"mv64xxx i2c"

Alexey

author
#14 lore raw

Hi, Alexey.
--

Quoted 17 lines · Alexey Dobriyan
Quoted 10 lines · Mark A. Greer

How's this (a complete replacement for previous patch)?

drivers/i2c/busses/Kconfig +0 -0
Quoted 7 lines

drivers/i2c/busses/Kconfig

Quoted 4 lines
drivers/i2c/busses/Kconfig +2 -0
+	  If you say yes to this option, support will be included for the
+	  built-in I2C interface on the Marvell 64xxx line of host bridges

Dot at the end. Should have noticed it earlier...

Gahh, I noticed that too a while back. I thought I'd fixed it... Done.

Quoted 11 lines

drivers/i2c/busses/Kconfig

Quoted 7 lines
drivers/i2c/busses/Kconfig +3 -0
+	if (drv_data->state == MV64XXX_I2C_STATE_IDLE) {
+		drv_data->action = MV64XXX_I2C_ACTION_SEND_STOP;
+		drv_data->state = MV64XXX_I2C_STATE_IDLE;

drv_data->state is already MV64XXX_I2C_STATE_IDLE. Gcc will probably optimize this line away, but...

Done away with.

Quoted 9 lines

drivers/i2c/busses/Kconfig

Quoted 6 lines
drivers/i2c/busses/Kconfig +2 -0
+		}
+		else {

It should be "} else {" according to CodingStyle.

Done.

Quoted 9 lines
Quoted 6 lines
drivers/i2c/busses/Kconfig +2 -0
+static int
+mv64xxx_i2c_intr(int irq, void *dev_id, struct pt_regs *regs)
Quoted 11 lines

drivers/i2c/busses/Kconfig

Quoted 6 lines
drivers/i2c/busses/Kconfig +4 -0
+	while (readl(drv_data->reg_base + MV64XXX_I2C_REG_CONTROL) &
+						MV64XXX_I2C_REG_CONTROL_IFLG) {
+		status = readl(drv_data->reg_base + MV64XXX_I2C_REG_STATUS);
+		mv64xxx_i2c_fsm(drv_data, status);

It can set drv_data->rc to -ENODEV or -EIO. In both cases ->action goes to MV64XXX_I2C_ACTION_SEND_STOP and mv64xxx_i2c_do_action() will writel() something. Is it correct to _not_ check ->rc here?

I think so. It still needs to go into do_action even when rc != 0 (in which case it'll do a STOP condition).

Quoted 4 lines

If it isn't then would it be
better to make all functions that set it return -E___ instead and drop struct mv64xxx_i2c_data.rc altogether?

I may not be understanding what you mean but I think I need something like mv64xxx_i2c_data.rc or a plain old global to hang onto the return code. I need that so when the wait_event_interruptible_timeout() returns, that thread can find out what happened while it was blocked. This is what I mean:

- calling thread enters i2c_xfer
- eventually the initial i2c action is executed and the thread blocks in wait_event_interruptible_timeout()
- other processes run
- several interrupts happen, last one causing an error which is stored in mv64xxx_i2c_data.rc
- do_action issues a stop on the i2c bus
- thread is unblocked and now has to dig out the rc from mv64xxx_i2c_data.rc

Or am I not understanding what you mean?

Quoted 7 lines
(quoted diff) +2 -0
+
+		if (!time_left <= 0) {

Confusing. You meant "if (time_left)" or "if (time_left > 0)"?

No, I'm blind. I meant "if (time_left <=0)". Should be fixed now. Good catch.

Quoted 9 lines

drivers/i2c/busses/Kconfig

Quoted 6 lines
drivers/i2c/busses/Kconfig +2 -0
+static struct i2c_algorithm mv64xxx_i2c_algo = {
+	.name = MV64XXX_I2C_CTLR_NAME "algorithm",

MV64XXX_I2C_CTLR_NAME doesn't end with space. " algorithm" here.

Space added.

Quoted 9 lines

drivers/i2c/busses/Kconfig

Quoted 6 lines
drivers/i2c/busses/Kconfig +2 -0
+		dev_err(dev, "mv64xxx: Can't register intr handler "
+			"irq: %d\\n", drv_data->irq);

You snipped s# \\n # \n # suggestion in my previous email. ;-)

Ah, got it this time. :)

This patch is a replacement patch that should address your concerns except maybe the mv64xxx_i2c_data.rc one.

Signed-off-by Mark A. Greer <[email protected]>

--

drivers/i2c/busses/Kconfig +10 -0
diff -Nru a/drivers/i2c/busses/Kconfig b/drivers/i2c/busses/Kconfig
@@ -476,4 +476,14 @@
 	  This driver can also be built as a module.  If so, the module
 	  will be called i2c-pca-isa.
 
+config I2C_MV64XXX
+	tristate "Marvell mv64xxx I2C Controller"
+	depends on I2C && MV64X60 && EXPERIMENTAL
+	help
+	  If you say yes to this option, support will be included for the
+	  built-in I2C interface on the Marvell 64xxx line of host bridges.
+
+	  This driver can also be built as a module.  If so, the module
+	  will be called i2c-mv64xxx.
+
 endmenu
drivers/i2c/busses/Makefile +1 -0
diff -Nru a/drivers/i2c/busses/Makefile b/drivers/i2c/busses/Makefile
@@ -20,6 +20,7 @@
 obj-$(CONFIG_I2C_IXP4XX)	+= i2c-ixp4xx.o
 obj-$(CONFIG_I2C_KEYWEST)	+= i2c-keywest.o
 obj-$(CONFIG_I2C_MPC)		+= i2c-mpc.o
+obj-$(CONFIG_I2C_MV64XXX)	+= i2c-mv64xxx.o
 obj-$(CONFIG_I2C_NFORCE2)	+= i2c-nforce2.o
 obj-$(CONFIG_I2C_PARPORT)	+= i2c-parport.o
 obj-$(CONFIG_I2C_PARPORT_LIGHT)	+= i2c-parport-light.o
drivers/i2c/busses/i2c-mv64xxx.c new file +606 -0

606 lines not on this page · raw message

include/linux/i2c-id.h +6 -0
diff -Nru a/include/linux/i2c-id.h b/include/linux/i2c-id.h
@@ -200,6 +200,9 @@
 
 #define I2C_ALGO_SIBYTE 0x150000	/* Broadcom SiByte SOCs		*/
 #define I2C_ALGO_SGI	0x160000        /* SGI algorithm                */
+					/* 0x170000 - USB		*/
+					/* 0x180000 - Virtual buses	*/
+#define I2C_ALGO_MV64XXX 0x190000       /* Marvell mv64xxx i2c ctlr	*/
 
 #define I2C_ALGO_EXP	0x800000	/* experimental			*/
 

@@ -305,5 +308,8 @@
 
 /* --- MCP107 adapter */
 #define I2C_HW_MPC107 0x00
+
+/* --- Marvell mv64xxx i2c adapter */
+#define I2C_HW_MV64XXX 0x00
 
 #endif /* LINUX_I2C_ID_H */
include/linux/mv643xx.h +11 -6
diff -Nru a/include/linux/mv643xx.h b/include/linux/mv643xx.h
@@ -977,12 +977,9 @@
 /* I2C Registers                        */
 /****************************************/
 
-#define MV64340_I2C_SLAVE_ADDR                                      0xc000
-#define MV64340_I2C_EXTENDED_SLAVE_ADDR                             0xc010
-#define MV64340_I2C_DATA                                            0xc004
-#define MV64340_I2C_CONTROL                                         0xc008
-#define MV64340_I2C_STATUS_BAUDE_RATE                               0xc00C
-#define MV64340_I2C_SOFT_RESET                                      0xc01c
+#define MV64XXX_I2C_CTLR_NAME					"mv64xxx i2c"
+#define MV64XXX_I2C_OFFSET                                          0xc000
+#define MV64XXX_I2C_REG_BLOCK_SIZE                                  0x0020
 
 /****************************************/
 /* GPP Interface Registers              */

@@ -1083,6 +1080,14 @@
 	u8	brg_can_tune;
 	u8	brg_clk_src;
 	u32	brg_clk_freq;
+};
+
+/* i2c Platform Device, Driver Data */
+struct mv64xxx_i2c_pdata {
+	u32	freq_m;
+	u32	freq_n;
+	u32	timeout;	/* In milliseconds */
+	u32	retries;
 };
 
 #endif /* __ASM_MV64340_H */
#15 lore raw
Quoted 8 lines · Mark A. Greer
Quoted 5 lines

drivers/i2c/busses/i2c-mv64xxx.c line 300

Quoted 1 line
drivers/i2c/busses/i2c-mv64xxx.c new file +1 -0
+		mv64xxx_i2c_fsm(drv_data, status);

It can set drv_data->rc to -ENODEV or -EIO. In both cases ->action goes to MV64XXX_I2C_ACTION_SEND_STOP and mv64xxx_i2c_do_action() will writel() something. Is it correct to _not_ check ->rc here?

I think so. It still needs to go into do_action even when rc != 0 (in which case it'll do a STOP condition).

Ok. Thanks for the explanation. Agree, ->rc should be left as is.

include/linux/i2c-id.h line 205

Quoted 9 lines

This patch is a replacement patch that should address your concerns except maybe the mv64xxx_i2c_data.rc one.

include/linux/i2c-id.h +3 -0
 
+					/* 0x170000 - USB		*/
+					/* 0x180000 - Virtual buses	*/
+#define I2C_ALGO_MV64XXX 0x190000       /* Marvell mv64xxx i2c ctlr	*/

While I searched for typos and you're fixing them, au1550 owned 0x170000. 2.6.11-rc3 says:

	#define I2C_ALGO_AU1550 0x170000 /* Au1550 PSC algorithm */

So, I'd remove first two comments.

Oh, and the last note: current sparse and gcc 4 don't produce any warnings.

Alexey

author
#16 lore raw
Quoted 12 lines · Alexey Dobriyan
Quoted 8 lines · Mark A. Greer
Quoted 8 lines

drivers/i2c/busses/i2c-mv64xxx.c line 300

Quoted 3 lines
drivers/i2c/busses/i2c-mv64xxx.c new file +1 -0
+		mv64xxx_i2c_fsm(drv_data, status);

It can set drv_data->rc to -ENODEV or -EIO. In both cases ->action goes to MV64XXX_I2C_ACTION_SEND_STOP and mv64xxx_i2c_do_action() will writel() something. Is it correct to _not_ check ->rc here?

Quoted 6 lines
Quoted 3 lines

I think so. It still needs to go into do_action even when rc != 0 (in which case it'll do a STOP condition).

Ok. Thanks for the explanation. Agree, ->rc should be left as is.

Okay.

Quoted 18 lines
Quoted 6 lines

include/linux/i2c-id.h line 205

Quoted 8 lines
include/linux/i2c-id.h +3 -0
 
 
 
 
 
+					/* 0x170000 - USB		*/
+					/* 0x180000 - Virtual buses	*/
+#define I2C_ALGO_MV64XXX 0x190000       /* Marvell mv64xxx i2c ctlr	*/

While I searched for typos and you're fixing them, au1550 owned 0x170000. 2.6.11-rc3 says:

Quoted 4 lines
	#define I2C_ALGO_AU1550 0x170000 /* Au1550 PSC algorithm */

So, I'd remove first two comments.

I added the comments b/c of this email from Jean Delvare, http://www.uwsg.iu.edu/hypermail/linux/kernel/0501.3/0977.html. The relevant part being:

"0x170000 is reserved within the legacy i2c project for an USB algorithm, and 0x180000 for virtual busses. Could you please use 0x190000 instead, so as to avoid future collisions?"

It looks like I2C_ALGO_AU1550 was just added so my guess is Jean is correct and I2C_ALGO_AU1550 should be made 0x1a0000 (or I move mine back one). Would someone confirm that 0x170000 is used by legacy i2c stuffs? I don't really know where to look. If so, I can easily make a patch moving it back.

Quoted 3 lines

Oh, and the last note: current sparse and gcc 4 don't produce any warnings.

Cool!

Mark

#17 lore raw

Hi Mark, Alexey,

Quoted 19 lines
Quoted 10 lines

include/linux/i2c-id.h line 205

Quoted 3 lines
include/linux/i2c-id.h +3 -0
+					/* 0x170000 - USB		*/
+					/* 0x180000 - Virtual buses	*/
+#define I2C_ALGO_MV64XXX 0x190000       /* Marvell mv64xxx i2c ctlr	*/

While I searched for typos and you're fixing them, au1550 owned 0x170000. 2.6.11-rc3 says:

	#define I2C_ALGO_AU1550 0x170000 /* Au1550 PSC algorithm */

So, I'd remove first two comments.

I added the comments b/c of this email from Jean Delvare,
http://www.uwsg.iu.edu/hypermail/linux/kernel/0501.3/0977.html. The relevant part being:

"0x170000 is reserved within the legacy i2c project for an USB algorithm, and 0x180000 for virtual busses. Could you please use 0x190000 instead, so as to avoid future collisions?"

Quoted 5 lines

It looks like I2C_ALGO_AU1550 was just added so my guess is Jean is correct and I2C_ALGO_AU1550 should be made 0x1a0000 (or I move mine back one). Would someone confirm that 0x170000 is used by legacy i2c stuffs? I don't really know where to look. If so, I can easily make a patch moving it back.

I am as surprised as you are to see this here. I2C_ALGO_AU1550 should really be made a different value. There is also a problem with I2C_ALGO_PCA and I2C_ALGO_SIBYTE having the same value, which was already reported to Greg some days ago if memory serves. I think I will send a patch against 2.6.10-rc3 to Linus this evening, which fixes the broken algo IDs. That way Mark can keep the algorithm ID he is using right now, and each algorithm will get its own, unique ID, as should be.

You can see the i2c-id.h list from the legacy i2c project here:
http://www2.lm-sensors.nu/~lm78/cvs/i2c/kernel/i2c-id.h
I am indeed trying to keep it in sync with the one in Linux 2.6 (and with the one in Linux 2.4 at times), in the hope it'll avoid confusion and increase portability.

Thanks,
--
Jean Delvare

#18 lore raw

Hi all,

Quoting myself:

Quoted 8 lines

I am as surprised as you are to see this here. I2C_ALGO_AU1550 should really be made a different value. There is also a problem with I2C_ALGO_PCA and I2C_ALGO_SIBYTE having the same value, which was already reported to Greg some days ago if memory serves. I think I will send a patch against 2.6.10-rc3 to Linus this evening, which fixes the broken algo IDs. That way Mark can keep the algorithm ID he is using right now, and each algorithm will get its own, unique ID, as should be.

I've changed my mind since. There isn't anything critical here and pushing random patches to Linus right before he releases 2.6.11 wouldn't have been particularly subtle. So I instead sent a patch to Greg, as can be seen here:
http://archives.andrew.net.au/lm-sensors/msg29405.html

So Mark, providing Greg accepts this patch, you can assume that I2C_ALGO_MV64XXX is already properly defined in Greg's tree (where your own patch will end up) so you don't need to add it yourself. You will still have to define I2C_HW_MV64XXX though.

Thanks,

Signature · 1 line
Jean Delvare