Revision History
Date Issue Description Author
<08/07/2010> <1.0> Msm7227平臺I2C驅動分析 滕景東
目錄
1. 摘要 3
2. 簡介 3
3. I2C架構 3
4. I2C總線初始化 4
5. I2C適配器驅動 5
6. I2C設備驅動 9
7. 用戶空間驅動支持 12
8. 數據傳輸框架 16
9. References 16
1. 摘要
主要介紹Msm7227平臺上I2C驅動原理，多數部分是29內核標準架構。
2. 簡介
I2C只有兩條線,一條串行數據線:SDA,一條是時鐘線SCL。I2C是一種多主機控制總線，同一總線上可允許多個master.
i2c總線適配器（adapter）就是一條i2c總線的控制器，在物理連接上若干i2c設備。在linux驅動中，每種處理器平臺有自己的適配器驅動。
3. I2C架構
內核中i2c相關代碼可以分爲三個層次：
i2c框架層：i2c.h和i2c-core.c爲其主體框架代碼，提供了核心數據結構的定義、i2c適配器驅動和設備驅動的註冊、註銷管理等；i2c-dev.c用於創建i2c適配器的/dev/i2c-%d設備節點，提供i2c設備的用戶空間訪問方法等。
i2c總線適配器驅動：i2c/busses/目錄下，如i2c-msm.c。定義描述具體i2c總線適配器的i2c_adapter數據結構、實現在具體i2c適配器上的i2c總線通信的具體實現，並由i2c_algorithm數據結構描述與i2c設備通信的方法。
i2c設備驅動：定義描述具體設備的i2c_client和可能的私有數據結構。
上圖展示了內核I2C結構大整體框架，以下根據內核加載順序介紹I2C總線初始化，I2C總線適配器驅動，I2C設備驅動和用戶空間驅動支持及數據傳輸框架五部分介紹。
4. I2C總線初始化
該過程主要完成了sysfs總線結構，最終形成如下結構：
/sys/bus/i2c/
|-- devices
|-- drivers
| |-- dummy
| |-- bind
| |-- uevent
| `-- unbind
|-- drivers_autoprobe
|-- drivers_probe
`-- uevent
和
/sys/class/i2c-adapter/
dummy_driver僅僅是註冊了一個空的設備驅動，註冊驅動時會遍歷加載/sys/class/i2c-adapter/中的所有設備，該過程在初始話總線過程中完成，/sys/class/i2c-adapter/基本爲空，所以我認爲這裏的驅動註冊只是驗證i2c總線結構的完整性考慮的。
5. I2C適配器驅動
Linux內核的所有適配器驅動程序都在driver/i2c/busses/目錄下，當前高通的驅動是i2c-msm.c,適配器驅動的註冊過程如下：
在kernel中提供了兩個adapter註冊接口,分別爲i2c_add_adapter()和i2c_add_numbered_adapter().由於在系統中可能存在多個adapter，因爲將每一條I2C總線對應一個編號,下文中稱爲I2C總線號。對於i2c_add_adapter()而言,它使用的是動態總線號,即由系統給其分配一個總線號，而i2c_add_numbered_adapter()則是自己指定總線號，如果這個總線號非法或者是被佔用，就會註冊失敗。高通的adapter驅動使用了i2c_add_numbered_adapter()註冊，總線號最初保存在platform_data中。
I2C adapter以platform_device方式註冊進系統，在proble函數中初始化了struct i2c_adapter結構：
struct i2c_adapter {
struct module *owner;
unsigned int id;
unsigned int class; /* classes to allow probing for */
const struct i2c_algorithm *algo; /* the algorithm to access the bus */
void *algo_data;
/* --- administration stuff. */
int (*client_register)(struct i2c_client *);
int (*client_unregister)(struct i2c_client *);
/* data fields that are valid for all devices */
u8 level; /* nesting level for lockdep */
struct mutex bus_lock;
struct mutex clist_lock;
int timeout; /* in jiffies */
int retries;
struct device dev; /* the adapter device */
int nr; /*該成員描述了總線號*/
struct list_head clients; /* i2c_client結構鏈表，該結構包含device，driver和
adapter結構*/
char name[48];
struct completion dev_released;
};
struct i2c_adapter {
struct module *owner;
unsigned int id;
unsigned int class; /* classes to allow probing for */
const struct i2c_algorithm *algo; /* the algorithm to access the bus */
void *algo_data;
/* --- administration stuff. */
int (*client_register)(struct i2c_client *);
int (*client_unregister)(struct i2c_client *);
/* data fields that are valid for all devices */
u8 level; /* nesting level for lockdep */
struct mutex bus_lock;
struct mutex clist_lock;
int timeout; /* in jiffies */
int retries;
struct device dev; /* the adapter device */
int nr; /*該成員描述了總線號*/
struct list_head clients; /* i2c_client結構鏈表，該結構包含device，driver和
adapter結構*/
char name[48];
struct completion dev_released;
};
其中nr的值是在arch/arm/mach-msm/devices.c中定義的：
struct platform_device msm_device_i2c = {
.name = "msm_i2c",
.id = 0,
.num_resources = ARRAY_SIZE(resources_i2c),
.resource = resources_i2c,
};
struct platform_device msm_device_i2c_2 = {
.name = "msm_i2c",
.id = 2,
.num_resources = ARRAY_SIZE(resources_i2c_2),
.resource = resources_i2c_2,
};
struct platform_device msm_device_i2c = {
.name = "msm_i2c",
.id = 0,
.num_resources = ARRAY_SIZE(resources_i2c),
.resource = resources_i2c,
};
struct platform_device msm_device_i2c_2 = {
.name = "msm_i2c",
.id = 2,
.num_resources = ARRAY_SIZE(resources_i2c_2),
.resource = resources_i2c_2,
};
該結構以參數形式傳進i2c_add_numbered_adapter()，下一步將進入
static int i2c_register_adapter(struct i2c_adapter *adap)
{
int res = 0, dummy;
/* Can't register until after driver model init */
if (unlikely(WARN_ON(!i2c_bus_type.p)))
return -EAGAIN;
mutex_init(&adap->bus_lock);
mutex_init(&adap->clist_lock);
INIT_LIST_HEAD(&adap->clients);/*初始化設備鏈表*/
mutex_lock(&core_lock);
/* Add the adapter to the driver core.
* If the parent pointer is not set up,
* we add this adapter to the host bus.
*/
if (adap->dev.parent == NULL) {
adap->dev.parent = &platform_bus;/*父設備是platform_bus*/
pr_debug("I2C adapter driver [%s] forgot to specify "
"physical device/n", adap->name);
}
dev_set_name(&adap->dev, "i2c-%d", adap->nr);/*設備節點名字*/
adap->dev.release = &i2c_adapter_dev_release;
adap->dev.class = &i2c_adapter_class;
res = device_register(&adap->dev); /*註冊adapter這個設備本身*/
if (res)
goto out_list;
dev_dbg(&adap->dev, "adapter [%s] registered/n", adap->name);
/*以下部分完成i2c設備和驅動的註冊*/
if (adap->nr < __i2c_first_dynamic_bus_num)/*主板初始化時的動態總線號，該值已導出符號表*/
i2c_scan_static_board_info(adap);/*完成新類型i2c設備的註冊，一般只在主板初始化時*/
/* Notify drivers */
dummy = bus_for_each_drv(&i2c_bus_type, NULL, adap,
i2c_do_add_adapter); /*探測總線上的所有i2c設備驅動，同時完成client、driver、device、adapter的綁定，但driver->address_data非空的情況下有用，而這又意味着只對舊的i2c機制有效*/
out_unlock:
mutex_unlock(&core_lock);
return res;
out_list:
idr_remove(&i2c_adapter_idr, adap->nr);
goto out_unlock;
}
static int i2c_register_adapter(struct i2c_adapter *adap)
{
int res = 0, dummy;
/* Can't register until after driver model init */
if (unlikely(WARN_ON(!i2c_bus_type.p)))
return -EAGAIN;
mutex_init(&adap->bus_lock);
mutex_init(&adap->clist_lock);
INIT_LIST_HEAD(&adap->clients);/*初始化設備鏈表*/
mutex_lock(&core_lock);
/* Add the adapter to the driver core.
* If the parent pointer is not set up,
* we add this adapter to the host bus.
*/
if (adap->dev.parent == NULL) {
adap->dev.parent = &platform_bus;/*父設備是platform_bus*/
pr_debug("I2C adapter driver [%s] forgot to specify "
"physical device/n", adap->name);
}
dev_set_name(&adap->dev, "i2c-%d", adap->nr);/*設備節點名字*/
adap->dev.release = &i2c_adapter_dev_release;
adap->dev.class = &i2c_adapter_class;
res = device_register(&adap->dev); /*註冊adapter這個設備本身*/
if (res)
goto out_list;
dev_dbg(&adap->dev, "adapter [%s] registered/n", adap->name);
/*以下部分完成i2c設備和驅動的註冊*/
if (adap->nr < __i2c_first_dynamic_bus_num)/*主板初始化時的動態總線號，該值已導出符號表*/
i2c_scan_static_board_info(adap);/*完成新類型i2c設備的註冊，一般只在主板初始化時*/
/* Notify drivers */
dummy = bus_for_each_drv(&i2c_bus_type, NULL, adap,
i2c_do_add_adapter); /*探測總線上的所有i2c設備驅動，同時完成client、driver、device、adapter的綁定，但driver->address_data非空的情況下有用，而這又意味着只對舊的i2c機制有效*/
out_unlock:
mutex_unlock(&core_lock);
return res;
out_list:
idr_remove(&i2c_adapter_idr, adap->nr);
goto out_unlock;
}
i2c_scan_static_board_info對應的初始化過程在board-msm7x27.c中完成，
i2c_register_board_info(0, i2c_devices, ARRAY_SIZE(i2c_devices));
6. I2C設備7. 驅動
驅動的編寫方法已在《msm7227-I2C設備驅動實現要點.doc》中介紹，此節分析驅動和設備的註冊過程。
本還想詳細分析代碼，但發現，這張圖已經足夠說明i2c驅動的註冊過程了，下面對我看代碼時碰到的一些問題簡要分析。
設備和驅動的關聯
大家知道，對於一個驅動程序有兩個元素不可或缺，即設備和驅動，一般驅動都是通過設備名和驅動名的匹配建立關係的，我從i2c/chips/裏看到的示例代碼了只能發現驅動的註冊，卻不見設備註冊的蹤影，令人疑惑，跟蹤發現，在i2c adapter註冊時會遍歷i2c_board_info這樣一個結構，而這個結構在29以前或更早的內核裏是不存在的，該數據結構在board-msm7x27.c中初始化了i2c設備名及設備地址，這便解決了驅動與設備的匹配問題，同時器件地址的提供也有所改變，舊的內核是在驅動中使用一個normal_i2c數組保存地址的。
名字匹配
一個i2c驅動是可以有多個名字的，即一個驅動程序可以支持多個設備，該機制是通過 struct i2c_device_id實現的，驅動中建立這麼一個結構體數組，i2c架構層便會掃描該數組，與設備名去匹配，匹配成功的都會進入相應probe函數。
進入probe
該過程困惑了我一段時間，其實要進入自己驅動的probe首先需要進入總線的probe，而進入總線probe的前提是與總線的match成功，具體實現大家可以根據上面的圖看一下相應代碼便知。
設備模型
I2C的架構充分利用的設備模型的原理及sysfs的實現，我認爲理解i2C架構前先了解一下設備模型是很有必要的。這裏將我的個人理解總結一下：
Kobject是設備模型的最小單位，kset是對kobject的集合，struct driver_private、struct device等結構都內嵌了kobject，kset也內嵌kobject用於表徵自己。相同特性的kset的合集又構成了subsys，舉個不太恰當的類比：
kobject之於設備或驅動；kset之於某一類設備，如i2c；subsys之於子系統，如輸入子系統。其實在29內核中subsys就是一個kset結構，貼兩張圖理解一下：
8. 用戶空間驅動支持
這部分在i2c-dev.c中實現，這部分內容簡單的說就是通過內嵌一個具有file_operations的標準字符設備驅動來虛擬i2c設備，這樣，就可以在用戶空間直接操作i2c設備了。
流程如下圖：
餘下的就是常規file_operation了，open操作：
static int i2cdev_open(struct inode *inode, struct file *file)
{
unsigned int minor = iminor(inode);
struct i2c_client *client;
struct i2c_adapter *adap;
struct i2c_dev *i2c_dev;
int ret = 0;
lock_kernel();/*內核上鎖，一般只在多cpu是有用*/
i2c_dev = i2c_dev_get_by_minor(minor);/*因爲有兩個adapter，同一個主設備號*/
if (!i2c_dev) {
ret = -ENODEV;
goto out;
}
adap = i2c_get_adapter(i2c_dev->adap->nr);
if (!adap) {
ret = -ENODEV;
goto out;
}
/* This creates an anonymous i2c_client, which may later be
* pointed to some address using I2C_SLAVE or I2C_SLAVE_FORCE.
*
* This client is ** NEVER REGISTERED ** with the driver model
* or I2C core code!! It just holds private copies of addressing
* information and maybe a PEC flag.
*/
client = kzalloc(sizeof(*client), GFP_KERNEL);
if (!client) {
i2c_put_adapter(adap);
ret = -ENOMEM;
goto out;
}
snprintf(client->name, I2C_NAME_SIZE, "i2c-dev %d", adap->nr);
client->driver = &i2cdev_driver;/*綁定字符設備驅動*/
client->adapter = adap;
file->private_data = client;
out:
unlock_kernel();
return ret;
}
static int i2cdev_open(struct inode *inode, struct file *file)
{
unsigned int minor = iminor(inode);
struct i2c_client *client;
struct i2c_adapter *adap;
struct i2c_dev *i2c_dev;
int ret = 0;
lock_kernel();/*內核上鎖，一般只在多cpu是有用*/
i2c_dev = i2c_dev_get_by_minor(minor);/*因爲有兩個adapter，同一個主設備號*/
if (!i2c_dev) {
ret = -ENODEV;
goto out;
}
adap = i2c_get_adapter(i2c_dev->adap->nr);
if (!adap) {
ret = -ENODEV;
goto out;
}
/* This creates an anonymous i2c_client, which may later be
* pointed to some address using I2C_SLAVE or I2C_SLAVE_FORCE.
*
* This client is ** NEVER REGISTERED ** with the driver model
* or I2C core code!! It just holds private copies of addressing
* information and maybe a PEC flag.
*/
client = kzalloc(sizeof(*client), GFP_KERNEL);
if (!client) {
i2c_put_adapter(adap);
ret = -ENOMEM;
goto out;
}
snprintf(client->name, I2C_NAME_SIZE, "i2c-dev %d", adap->nr);
client->driver = &i2cdev_driver;/*綁定字符設備驅動*/
client->adapter = adap;
file->private_data = client;
out:
unlock_kernel();
return ret;
}
注意這裏分配並初始化了一個struct i2c_client結構.但是沒有註冊這個clinet.此外,這個函數中還有一個比較奇怪的操作.不是在前面已經將i2c_dev->adap指向要操作的adapter麼?爲什麼還要以adapter->nr爲關鍵字從i2c_adapter_idr去找這個操作的adapter呢?注意了,調用i2c_get_adapter()從總線號nr找到操作的adapter的時候,還會增加module的引用計數.這樣可以防止模塊意外被釋放掉.也許有人會有這樣的疑問,那 i2c_dev->adap->nr操作,如果i2c_dev->adap被釋放掉的話,不是一樣會引起系統崩潰麼?這裏因爲,在i2cdev_attach_adapter()間接的增加了一次adapter的一次引用計數.如下:
view plaincopy to clipboardprint?
static int i2cdev_attach_adapter(struct i2c_adapter *adap)
{
．．．．．．
i2c_dev->dev = device_create(i2c_dev_class, &adap->dev,
MKDEV(I2C_MAJOR, adap->nr),
"i2c-%d", adap->nr);
．．．．．．
}
static int i2cdev_attach_adapter(struct i2c_adapter *adap)
{
．．．．．．
i2c_dev->dev = device_create(i2c_dev_class, &adap->dev,
MKDEV(I2C_MAJOR, adap->nr),
"i2c-%d", adap->nr);
．．．．．．
}
看到了麼，i2c_dev內嵌的device是以adap->dev爲父結點,在device_create()中會增次adap->dev的一次引用計數.
好了,open()操作到此就完成了.
使用方法：（參考kernel-test/i2c-msm-test.c）
1、構造struct i2c_msg
[讀] struct i2c_msg msgs[] = {
[0] = {
.addr = slave_address,
.flags = 0,
.buf = (void *)offset_data,
.len = ARRAY_SIZE(offset_data),
},
[1] = {
.addr = slave_address,
.flags = I2C_M_RD,
.buf = (void *)buf,
.len = count,
},
};
[寫] struct i2c_msg msgs[] = {
[0] = {
.addr = slave_address,
.flags = 0,
.buf = (void *)data,
.len = (2 + len) * sizeof(*data),
},
}
[讀] struct i2c_msg msgs[] = {
[0] = {
.addr = slave_address,
.flags = 0,
.buf = (void *)offset_data,
.len = ARRAY_SIZE(offset_data),
},
[1] = {
.addr = slave_address,
.flags = I2C_M_RD,
.buf = (void *)buf,
.len = count,
},
};
[寫] struct i2c_msg msgs[] = {
[0] = {
.addr = slave_address,
.flags = 0,
.buf = (void *)data,
.len = (2 + len) * sizeof(*data),
},
}
2、通過ioctl操作設備
view plaincopy to clipboardprint?
static int do_rdwr(int fd, struct i2c_msg *msgs, int nmsgs)
{
struct i2c_rdwr_ioctl_data msgset = {
.msgs = msgs,
.nmsgs = nmsgs, /* msgs 個數*/
};
if (msgs == NULL || nmsgs <= 0)
return -1;
if (ioctl(fd, I2C_RDWR, &msgset) < 0)
return -1;
return 0;
}
static int do_rdwr(int fd, struct i2c_msg *msgs, int nmsgs)
{
struct i2c_rdwr_ioctl_data msgset = {
.msgs = msgs,
.nmsgs = nmsgs, /* msgs 個數*/
};
if (msgs == NULL || nmsgs <= 0)
return -1;
if (ioctl(fd, I2C_RDWR, &msgset) < 0)
return -1;
return 0;
}
3、ioctl命令字：
#define I2C_SMBUS_READ 1
#define I2C_SMBUS_WRITE 0
#define I2C_SMBUS_QUICK 0
#define I2C_SMBUS_BYTE 1
#define I2C_SMBUS_BYTE_DATA 2
#define I2C_SMBUS_WORD_DATA 3
#define I2C_SMBUS_PROC_CALL 4
#define I2C_SMBUS_BLOCK_DATA 5
#define I2C_SMBUS_I2C_BLOCK_DATA 6
#define I2C_SMBUS_BLOCK_PROC_CALL 7
#define I2C_RETRIES 0x0701
#define I2C_TIMEOUT 0x0702
#define I2C_SLAVE 0x0703
#define I2C_SLAVE_FORCE 0x0706
#define I2C_TENBIT 0x0704
#define I2C_FUNCS 0x0705
#define I2C_RDWR 0x0707
#define I2C_PEC 0x0708
#define I2C_SMBUS 0x0720
#define I2C_SMBUS_READ 1
#define I2C_SMBUS_WRITE 0
#define I2C_SMBUS_QUICK 0
#define I2C_SMBUS_BYTE 1
#define I2C_SMBUS_BYTE_DATA 2
#define I2C_SMBUS_WORD_DATA 3
#define I2C_SMBUS_PROC_CALL 4
#define I2C_SMBUS_BLOCK_DATA 5
#define I2C_SMBUS_I2C_BLOCK_DATA 6
#define I2C_SMBUS_BLOCK_PROC_CALL 7
#define I2C_RETRIES 0x0701
#define I2C_TIMEOUT 0x0702
#define I2C_SLAVE 0x0703
#define I2C_SLAVE_FORCE 0x0706
#define I2C_TENBIT 0x0704
#define I2C_FUNCS 0x0705
#define I2C_RDWR 0x0707
#define I2C_PEC 0x0708
#define I2C_SMBUS 0x0720
9. 數據傳輸框架
I2C架構的讀寫支持兩種類型，默認實現的操作是smbus協議，該協議與i2c協議類似，如果控制器不支持smbus，框架層可以用i2c_transfer模擬smbus的實現，系統默認的i2c傳輸函數一般都是基於i2c模擬的smbus方法傳輸的，如i2c_smbus_write_byte_data，i2c_smbus_read_byte_data等。
I2C協議的總線實現應該是I2C控制器，而不是SMBUS控制器, I2C協議和SMBUS協議不完成等同,SMBUS是I2C的子集，smbus由I2C衍生而來。smbus總線上傳輸的數據一定是I2C的格式的，但是SMBUS上傳輸的數據不一定能滿足具體某個I2C從設備的通信要求(數據序列)。
下圖以i2c_smbus_write_byte_data介紹數據流程：
本文來自CSDN博客，轉載請標明出處：http://blog.csdn.net/zhenwenxian/archive/2010/09/14/5882265.aspx