Linux设备驱动之I2C架构分析--暂存

原文地址： http://dongyulong.blog.51cto.com/1451604/324943

 

Linux设备驱动之I2C架构分析

一:前言

I2c是philips提出的外设总线.I2C只有两条线,一条串行数据线:SDA,一条是时钟线SCL.正因为这样,它方便了工程人员的布线.另外,I2C是一种多主机控制总线.它和USB总线不同,USB是基于master-slave机制,任何设备的通信必须由主机发起才可以.而I2C是基于multi master机制.一同总线上可允许多个master.关于I2C协议的知识,这里不再赘述.可自行下载spec阅读即可.

二:I2C架构概述

在linux中,I2C驱动架构如下所示:

如上图所示,每一条I2C对应一个adapter.在kernel中,每一个adapter提供了一个描述的结构(struct i2c_adapter),也定义了adapter支持的操作(struct i2c_adapter).再通过i2c core层将i2c设备与i2c adapter关联起来.

这个图只是提供了一个大概的框架.在下面的代码分析中,从下至上的来分析这个框架图.以下的代码分析是基于linux 2.6.26.分析的代码基本位于: linux-2.6.26.3/drivers/i2c/位置.

三:adapter注册

在kernel中提供了两个adapter注册接口,分别为i2c_add_adapter()和i2c_add_numbered_adapter().由于在系统中可能存在多个adapter,因为将每一条I2C总线对应一个编号,下文中称为I2C总线号.这个总线号的PCI中的总线号不同.它和硬件无关,只是软件上便于区分而已.

对于i2c_add_adapter()而言,它使用的是动态总线号,即由系统给其分析一个总线号,而i2c_add_numbered_adapter()则是自己指定总线号,如果这个总线号非法或者是被占用,就会注册失败.

分别来看一下这两个函数的代码:

int i2c_add_adapter(struct i2c_adapter *adapter)

{

    int id, res = 0;

retry:

    if (idr_pre_get(&i2c_adapter_idr, GFP_KERNEL) == 0)

        return -ENOMEM;

    mutex_lock(&core_lock);

    /* "above" here means "above or equal to", sigh */

    res = idr_get_new_above(&i2c_adapter_idr, adapter,

                __i2c_first_dynamic_bus_num, &id);

    mutex_unlock(&core_lock);

    if (res < 0) {

        if (res == -EAGAIN)

            goto retry;

        return res;

    }

    adapter->nr = id;

    return i2c_register_adapter(adapter);

}

在这里涉及到一个idr结构.idr结构本来是为了配合page cache中的radix tree而设计的.在这里我们只需要知道,它是一种高效的搜索树,且这个树预先存放了一些内存.避免在内存不够的时候出现问题.所在,在往idr中插入结构的时候,首先要调用idr_pre_get()为它预留足够的空闲内存,然后再调用idr_get_new_above()将结构插入idr中,该函数以参数的形式返回一个id.以后凭这个id就可以在idr中找到相对应的结构了.对这个数据结构操作不太理解的可以查阅本站<< linux文件系统之文件的读写>>中有关radix tree的分析.

注意一下idr_get_new_above(&i2c_adapter_idr, adapter,__i2c_first_dynamic_bus_num, &id)的参数的含义,它是将adapter结构插入到i2c_adapter_idr中,存放位置的id必须要大于或者等于__i2c_first_dynamic_bus_num,

然后将对应的id号存放在adapter->nr中.调用i2c_register_adapter(adapter)对这个adapter进行进一步注册.

看一下另外一人注册函数: i2c_add_numbered_adapter( ),如下所示:

int i2c_add_numbered_adapter(struct i2c_adapter *adap)

{

    int id;

    int status;

    if (adap->nr & ~MAX_ID_MASK)

        return -EINVAL;

retry:

    if (idr_pre_get(&i2c_adapter_idr, GFP_KERNEL) == 0)

        return -ENOMEM;

    mutex_lock(&core_lock);

    /* "above" here means "above or equal to", sigh;

     * we need the "equal to" result to force the result

     */

    status = idr_get_new_above(&i2c_adapter_idr, adap, adap->nr, &id);

    if (status == 0 && id != adap->nr) {

        status = -EBUSY;

        idr_remove(&i2c_adapter_idr, id);

    }

    mutex_unlock(&core_lock);

    if (status == -EAGAIN)

        goto retry;

    if (status == 0)

        status = i2c_register_adapter(adap);

    return status;

}

对比一下就知道差别了,在这里它已经指定好了adapter->nr了.如果分配的id不和指定的相等,便返回错误.

过一步跟踪i2c_register_adapter().代码如下:

static int i2c_register_adapter(struct i2c_adapter *adap)

{

    int res = 0, dummy;

    mutex_init(&adap->bus_lock);

    mutex_init(&adap->clist_lock);

    INIT_LIST_HEAD(&adap->clients);

    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;

        pr_debug("I2C adapter driver [%s] forgot to specify "

             "physical device ", adap->name);

    }

    sprintf(adap->dev.bus_id, "i2c-%d", adap->nr);

    adap->dev.release = &i2c_adapter_dev_release;

    adap->dev.class = &i2c_adapter_class;

    res = device_register(&adap->dev);

    if (res)

        goto out_list;

    dev_dbg(&adap->dev, "adapter [%s] registered ", adap->name);

    /* create pre-declared device nodes for new-style drivers */

    if (adap->nr < __i2c_first_dynamic_bus_num)

        i2c_scan_static_board_info(adap);

    /* let legacy drivers scan this bus for matching devices */

    dummy = bus_for_each_drv(&i2c_bus_type, NULL, adap,

                 i2c_do_add_adapter);

out_unlock:

    mutex_unlock(&core_lock);

    return res;

out_list:

    idr_remove(&i2c_adapter_idr, adap->nr);

    goto out_unlock;

}

首先对adapter和adapter中内嵌的struct device结构进行必须的初始化.之后将adapter内嵌的struct device注册.

在这里注意一下adapter->dev的初始化.它的类别为i2c_adapter_class,如果没有父结点,则将其父结点设为platform_bus.adapter->dev的名字为i2c + 总线号.

测试一下:

[eric@mochow i2c]$ cd /sys/class/i2c-adapter/

[eric@mochow i2c-adapter]$ ls

i2c-0

可以看到,在我的PC上,有一个I2C adapter,看下详细信息:

[eric@mochow i2c-adapter]$ tree

.

`-- i2c-0

    |-- device -> ../../../devices/pci0000:00/0000:00:1f.3/i2c-0

    |-- name

    |-- subsystem -> ../../../class/i2c-adapter

    `-- uevent

3 directories, 2 files

可以看到,该adapter是一个PCI设备.

继续往下看:

之后,在注释中看到,有两种类型的driver,一种是new-style drivers,另外一种是legacy drivers

New-style drivers是在2.6近版的kernel加入的.它们最主要的区别是在adapter和i2c driver的匹配上.

3.1: new-style 形式的adapter注册

对于第一种,也就是new-style drivers,将相关代码再次列出如下:

    if (adap->nr < __i2c_first_dynamic_bus_num)

        i2c_scan_static_board_info(adap);

如果adap->nr 小于__i2c_first_dynamic_bus_num的话,就会进入到i2c_scan_static_board_info().

结合我们之前分析的adapter的两种注册分式: i2c_add_adapter()所分得的总线号肯会不会小于__i2c_first_dynamic_bus_num.只有i2c_add_numbered_adapter()才有可能满足:

(adap->nr < __i2c_first_dynamic_bus_num)

而且必须要调用i2c_register_board_info()将板子上的I2C设备信息预先注册时才会更改__i2c_first_dynamic_bus_num的值.在x86上只没有使用i2c_register_board_info()的.因此,x86平台上的分析可以忽略掉new-style driver的方式.不过,还是详细分析这种情况下.

首先看一下i2c_register_board_info(),如下:

int __init

i2c_register_board_info(int busnum,

    struct i2c_board_info const *info, unsigned len)

{

    int status;

    mutex_lock(&__i2c_board_lock);

    /* dynamic bus numbers will be assigned after the last static one */

    if (busnum >= __i2c_first_dynamic_bus_num)

        __i2c_first_dynamic_bus_num = busnum + 1;

    for (status = 0; len; len--, info++) {

        struct i2c_devinfo  *devinfo;

        devinfo = kzalloc(sizeof(*devinfo), GFP_KERNEL);

        if (!devinfo) {

            pr_debug("i2c-core: can't register boardinfo! ");

            status = -ENOMEM;

            break;

        }

        devinfo->busnum = busnum;

        devinfo->board_info = *info;

        list_add_tail(&devinfo->list, &__i2c_board_list);

    }

    mutex_unlock(&__i2c_board_lock);

    return status;

}

这个函数比较简单, struct i2c_board_info用来表示I2C设备的一些情况,比如所在的总线.名称,地址,中断号等.最后,这些信息会被存放到__i2c_board_list链表.

跟踪i2c_scan_static_board_info():代码如下:

static void i2c_scan_static_board_info(struct i2c_adapter *adapter)

{

    struct i2c_devinfo  *devinfo;

    mutex_lock(&__i2c_board_lock);

    list_for_each_entry(devinfo, &__i2c_board_list, list) {

        if (devinfo->busnum == adapter->nr

                && !i2c_new_device(adapter,

                        &devinfo->board_info))

            printk(KERN_ERR "i2c-core: can't create i2c%d-%04x ",

                i2c_adapter_id(adapter),

                devinfo->board_info.addr);

    }

    mutex_unlock(&__i2c_board_lock);

}

该函数遍历挂在__i2c_board_list链表上面的i2c设备的信息,也就是我们在启动的时候指出的i2c设备的信息.

如果指定设备是位于adapter所在的I2C总线上,那么,就调用i2c_new_device().代码如下:

struct i2c_client *

i2c_new_device(struct i2c_adapter *adap, struct i2c_board_info const *info)

{

    struct i2c_client   *client;

    int         status;

    client = kzalloc(sizeof *client, GFP_KERNEL);

    if (!client)

        return NULL;

    client->adapter = adap;

    client->dev.platform_data = info->platform_data;

    device_init_wakeup(&client->dev, info->flags & I2C_CLIENT_WAKE);

    client->flags = info->flags & ~I2C_CLIENT_WAKE;

    client->addr = info->addr;

    client->irq = info->irq;

    strlcpy(client->name, info->type, sizeof(client->name));

    /* a new style driver may be bound to this device when we

     * return from this function, or any later moment (e.g. maybe

     * hotplugging will load the driver module).  and the device

     * refcount model is the standard driver model one.

     */

    status = i2c_attach_client(client);

    if (status < 0) {

        kfree(client);

        client = NULL;

    }

    return client;

}

我们又遇到了一个新的结构:struct i2c_client,不要被这个结构吓倒了,其实它就是一个嵌入struct device的I2C设备的封装.它和我们之前遇到的struct usb_device结构的作用是一样的.

首先,在clinet里保存该设备的相关消息.特别的, client->adapter指向了它所在的adapter.

特别的,clinet->name为info->name.也是指定好了的.

一切初始化完成之后,便会调用i2c_attach_client( ).看这个函数的字面意思,是将clinet关联起来.到底怎么样关联呢?继续往下看:

int i2c_attach_client(struct i2c_client *client)

{

    struct i2c_adapter *adapter = client->adapter;

    int res = 0;

    //初始化client内嵌的dev结构

    //父结点为所在的adapter,所在bus为i2c_bus_type

    client->dev.parent = &client->adapter->dev;

    client->dev.bus = &i2c_bus_type;

    //如果client已经指定了driver,将driver和内嵌的dev关联起来

    if (client->driver)

        client->dev.driver = &client->driver->driver;

    //指定了driver, 但不是newstyle的

    if (client->driver && !is_newstyle_driver(client->driver)) {

        client->dev.release = i2c_client_release;

        client->dev.uevent_suppress = 1;

    } else

        client->dev.release = i2c_client_dev_release;

    //clinet->dev的名称

    snprintf(&client->dev.bus_id[0], sizeof(client->dev.bus_id),

        "%d-%04x", i2c_adapter_id(adapter), client->addr);

    //将内嵌的dev注册

    res = device_register(&client->dev);

    if (res)

        goto out_err;

    //将clinet链到adapter->clients中

    mutex_lock(&adapter->clist_lock);

    list_add_tail(&client->list, &adapter->clients);

    mutex_unlock(&adapter->clist_lock);

    dev_dbg(&adapter->dev, "client [%s] registered with bus id %s ",

        client->name, client->dev.bus_id);

    //如果adapter->cleinet_reqister存在,就调用它

    if (adapter->client_register)  {

        if (adapter->client_register(client)) {

            dev_dbg(&adapter->dev, "client_register "

                "failed for client [%s] at 0x%02x ",

                client->name, client->addr);

        }

    }

    return 0;

out_err:

    dev_err(&adapter->dev, "Failed to attach i2c client %s at 0x%02x "

        "(%d) ", client->name, client->addr, res);

    return res;

}

参考上面添加的注释,应该很容易理解这段代码了,就不加详细分析了.这个函数的名字不是i2c_attach_client()么?怎么没看到它的关系过程呢?

这是因为:在代码中设置了client->dev所在的bus为i2c_bus_type .以为只需要有bus为i2c_bus_type的driver注册,就会产生probe了.这个过程呆后面分析i2c driver的时候再来详细分析.

3.2: legacy形式的adapter注册

Legacy形式的adapter注册代码片段如下:

    dummy = bus_for_each_drv(&i2c_bus_type, NULL, adap,

                 i2c_do_add_adapter);

这段代码遍历挂在i2c_bus_type上的驱动,然后对每一个驱动和adapter调用i2c_do_add_adapter().

代码如下:

static int i2c_do_add_adapter(struct device_driver *d, void *data)

{

    struct i2c_driver *driver = to_i2c_driver(d);

    struct i2c_adapter *adap = data;

    if (driver->attach_adapter) {

        /* We ignore the return code; if it fails, too bad */

        driver->attach_adapter(adap);

    }

    return 0;

}

该函数很简单,就是调用driver的attach_adapter()接口.

到此为止,adapter的注册已经分析完了.

四:i2c driver注册

在分析i2c driver的时候,有必要先分析一下i2c架构的初始化

代码如下:

static int __init i2c_init(void)

{

    int retval;

    retval = bus_register(&i2c_bus_type);

    if (retval)

        return retval;

    retval = class_register(&i2c_adapter_class);

    if (retval)

        goto bus_err;

    retval = i2c_add_driver(&dummy_driver);

    if (retval)

        goto class_err;

    return 0;

class_err:

    class_unregister(&i2c_adapter_class);

bus_err:

    bus_unregister(&i2c_bus_type);

    return retval;

}

subsys_initcall(i2c_init);

很明显,i2c_init()会在系统初始化的时候被调用.

在i2c_init中,先注册了i2c_bus_type的bus,i2c_adapter_class的class.然后再调用i2c_add_driver()注册了一个i2c driver.

I2c_bus_type结构如下:

static struct bus_type i2c_bus_type = {

    .name       = "i2c",

    .dev_attrs  = i2c_dev_attrs,

    .match      = i2c_device_match,

    .uevent     = i2c_device_uevent,

    .probe      = i2c_device_probe,

    .remove     = i2c_device_remove,

    .shutdown   = i2c_device_shutdown,

    .suspend    = i2c_device_suspend,

    .resume     = i2c_device_resume,

};

这个结构先放在这里吧,以后还会用到里面的信息的.

从上面的初始化函数里也看到了,注册i2c driver的接口为i2c_add_driver().代码如下:

static inline int i2c_add_driver(struct i2c_driver *driver)

{

    return i2c_register_driver(THIS_MODULE, driver);

}

继续跟踪:

int i2c_register_driver(struct module *owner, struct i2c_driver *driver)

{

    int res;

    /* new style driver methods can't mix with legacy ones */

    //如果是一个newstyle的driver.但又定义了attach_adapter/detach_adapter.非法

    if (is_newstyle_driver(driver)) {

        if (driver->attach_adapter || driver->detach_adapter

                || driver->detach_client) {

            printk(KERN_WARNING

                    "i2c-core: driver [%s] is confused ",

                    driver->driver.name);

            return -EINVAL;

        }

    }

    /* add the driver to the list of i2c drivers in the driver core */

    //关联到i2c_bus_types

    driver->driver.owner = owner;

    driver->driver.bus = &i2c_bus_type;

    /* for new style drivers, when registration returns the driver core

     * will have called probe() for all matching-but-unbound devices.

     */

     //注册内嵌的driver

    res = driver_register(&driver->driver);

    if (res)

        return res;

    mutex_lock(&core_lock);

    pr_debug("i2c-core: driver [%s] registered ", driver->driver.name);

    /* legacy drivers scan i2c busses directly */

    //遍历所有的adapter,对其都调用driver->attach_adapter

    if (driver->attach_adapter) {

        struct i2c_adapter *adapter;

        down(&i2c_adapter_class.sem);

        list_for_each_entry(adapter, &i2c_adapter_class.devices,

                    dev.node) {

            driver->attach_adapter(adapter);

        }

        up(&i2c_adapter_class.sem);

    }

    mutex_unlock(&core_lock);

    return 0;

}

这里也有两种形式的区分,对于第一种,只需要将内嵌的driver注册就可以了,对于legacy的情况,对每一个adapter都调用driver->attach_adapter().

现在,我们可以将adapter和i2c driver关联起来考虑一下了:

1:如果是news style形式的,在注册adapter的时候,将它上面的i2c 设备转换成了struct client.struct client->dev->bus又指定了和i2c driver同一个bus.因为,它们可以发生probe.

2:如果是legacy形式,就直接找到对应的对象,调用driver->attach_adapter().