Android(安卓)电话系统框架介绍
16lz
2021-01-26
目录
1.简介
1.1模块组成
1.2系统框架图
1.3代码结构图
2.RILD框架设计
3.RILD源码分析
3.1RILD进程入口函数分析
3.2启动事件循环处理eventLoop工作线程
3.2.1添加事件
3.2.2触发事件
3.2.3处理事件
3.2.4超时事件查询
3.2.5可读事件查询
3.2.6事件处理
3.3RIL_Env定义
3.4RIL_RadioFunctions定义
1.onRequest
2.currentState
3.onSupports
4.onCancel
5.getVersion
3.5注册RIL_Env接口
3.6RIL_Init的主要任务：
3.6.1.打开AT模块
3.6.2.添加定时事件RIL_requestTimedCallback
3.6.3.readLoop工作线程
3.6.4.注册RIL_RadioFunctions接口
3.7.客户端连接处理
3.8.客户端通信处理
4.电话拨打流程
1.简介
在android系统中rild运行在AP上，AP上的应用通过rild发送AT指令给BP，BP接收到信息后又通过rild传送给AP。AP与BP之间有两种通信方式：
- Solicited Response:Ap向Bp发送请求，Bp给Ap发送回复,该类型的AT指令及其回调函数以数组的形式存放在Ril_commands.h文件中：{数组中的索引号，请求回调函数，响应回调函数}
-
unSolicited Response:Bp主动给Ap发送事件,该类型的AT指令及其回调函数以数组的形式存放在ril_unsol_commands.h文件中：{数组中的索引号，响应回调函数，类型}
不同手机厂商使用的AT命令不完全相同，为了保密，AP与BP之间通过各厂商自己的相关动态库来通信。
1.1模块组成
RIL模块由rild守护进程、libril.so、librefrence.so三部分组成：
1.rild模块被编译为一个可执行文件，实现一个main函数作为整个ril模块的入口点。在初始化时使用dlopen打开librefrence_ril.so，从中取出并执行RIL_Init函数，得到RIL_RadioFunctions指针，通过RIL_register()函数注册到libril.so库中，其源码结构如下：
2.libril.so是共享库，主要负责同上层的通信工作，接收ril的请求，并传递给librefrence_ril.so，同时将librefrence_ril.so返回的消息送给调用进程，源码结构如下所示：
3.librefrence_ril.so是由各手机厂商自己实现，在rild进程运行中通过dlopen方式加载，主要负责跟modem硬件通信，转换来自libril.so的请求为AT命令，同时监听Modem的反馈信息给libril.so
Android的电话系统主要分为三个部分，java层的各种电话相关应用，java层的Phone Service，主要为上层提供API，同时与native进行通信，可以看做为电话系统的客户端，native层的电话服务进程RILD，负责为上层提供各种电话功能服务，直接与modem进行交互：
1.2系统框架图
Android电话系统设计框架图：
由于Android 开发者使用的Modem 是不一样的，各种指令格式，初始化序列都可能不一样，所以为了消除这些差别，Android 设计者将ril 做了一个抽象，使用一个虚拟电话的概念，不同modem相关的AT指令或者通信协议编译成相应的动态链接库.so文件，Rild 是具体的AT 指令合成者和应答解析者。
1.3代码结构图
Android电话系统代码结构图：
2.RILD框架设计
在android的电话系统中，在native层实现了电话服务的服务端，由RILD服务与modem的交互，在java层实现电话的客户端，本文主要介绍电话系统的服务端RILD进程，以下是RILD的设计框架图：
3.RILD源码分析
在kernel启动完成后，将启动第一个应用进程Init进程，在android之Init进程启动过程源码分析一文中对init进程的启动流程进行了详细的介绍。init进程在启动过程中将读取init.rc文件来启动一些重量级的native服务，rild进程就是通过配置在init.rc中来启动的。
service ril-daemon /system/bin/rild class main socket rild stream 660 root radio socket rild-debug stream 660 radio system user root group radio cache inet misc audio sdcard_rw log 3.1RILD进程入口函数分析
接下来给出的是RILD进程启动的时序图：
hardware\ril\rild\rild.c
int main(int argc, char **argv) { const char * rilLibPath = NULL; char **rilArgv; void *dlHandle; const RIL_RadioFunctions *(*rilInit)(const struct RIL_Env *, int, char **); const RIL_RadioFunctions *funcs; char libPath[PROPERTY_VALUE_MAX]; unsigned char hasLibArgs = 0; int i; umask(S_IRGRP | S_IWGRP | S_IXGRP | S_IROTH | S_IWOTH | S_IXOTH); //rild启动无参数 for (i = 1; i < argc ;) { if (0 == strcmp(argv[i], "-l") && (argc - i > 1)) { rilLibPath = argv[i + 1]; i += 2; } else if (0 == strcmp(argv[i], "--")) { i++; hasLibArgs = 1; break; } else { usage(argv[0]); } } if (rilLibPath == NULL) { //通过Android属性系统读取属性"rild.libpath"的值，即lib库的存放路径 if ( 0 == property_get(LIB_PATH_PROPERTY, libPath, NULL)) { goto done; } else { rilLibPath = libPath; } } ################################################################################## 判断是否为模拟器 ################################################################################## #if 1 { static char* arg_overrides[3]; static char arg_device[32]; int done = 0; #define REFERENCE_RIL_PATH "/system/lib/libreference-ril.so" /* first, read /proc/cmdline into memory */ char buffer[1024], *p, *q; int len; int fd = open("/proc/cmdline",O_RDONLY); if (fd < 0) { LOGD("could not open /proc/cmdline:%s", strerror(errno)); goto OpenLib; } //读取/proc/cmdline文件中的内容 do { len = read(fd,buffer,sizeof(buffer)); } while (len == -1 && errno == EINTR); if (len < 0) { LOGD("could not read /proc/cmdline:%s", strerror(errno)); close(fd); goto OpenLib; } close(fd); //判断是否为模拟器，对于真机，此处条件为false if (strstr(buffer, "android.qemud=") != NULL) { int tries = 5; #define QEMUD_SOCKET_NAME "qemud" while (1) { int fd; sleep(1); fd = socket_local_client(QEMUD_SOCKET_NAME, ANDROID_SOCKET_NAMESPACE_RESERVED, SOCK_STREAM ); if (fd >= 0) { close(fd); snprintf( arg_device, sizeof(arg_device), "%s/%s", ANDROID_SOCKET_DIR, QEMUD_SOCKET_NAME ); arg_overrides[1] = "-s"; arg_overrides[2] = arg_device; done = 1; break; } LOGD("could not connect to %s socket: %s",QEMUD_SOCKET_NAME, strerror(errno)); if (--tries == 0) break; } if (!done) { LOGE("could not connect to %s socket (giving up): %s", QEMUD_SOCKET_NAME, strerror(errno)); while(1) sleep(0x00ffffff); } } /* otherwise, try to see if we passed a device name from the kernel */ if (!done) do { //true #define KERNEL_OPTION "android.ril=" #define DEV_PREFIX "/dev/" //判断/proc/cmdline中的内容是否包含"android.ril=" p = strstr( buffer, KERNEL_OPTION ); if (p == NULL) break; p += sizeof(KERNEL_OPTION)-1; q = strpbrk( p, " \t\n\r" ); if (q != NULL) *q = 0; snprintf( arg_device, sizeof(arg_device), DEV_PREFIX "%s", p ); arg_device[sizeof(arg_device)-1] = 0; arg_overrides[1] = "-d"; arg_overrides[2] = arg_device; done = 1; } while (0); if (done) { //false argv = arg_overrides; argc = 3; i = 1; hasLibArgs = 1; rilLibPath = REFERENCE_RIL_PATH; LOGD("overriding with %s %s", arg_overrides[1], arg_overrides[2]); } } OpenLib: #endif ################################################################################## 动态库装载 ################################################################################## switchUser();//设置Rild进程的组用户为radio //加载厂商自定义的库 ①dlHandle = dlopen(rilLibPath, RTLD_NOW); if (dlHandle == NULL) { fprintf(stderr, "dlopen failed: %s\n", dlerror()); exit(-1); } //创建客户端事件监听线程 ②RIL_startEventLoop(); //通过dlsym定位到RIL_Init函数的地址，并且强制转换为RIL_RadioFunctions的函数指针 ③rilInit = (const RIL_RadioFunctions *(*)(const struct RIL_Env *, int, char **))dlsym(dlHandle, "RIL_Init"); if (rilInit == NULL) { fprintf(stderr, "RIL_Init not defined or exported in %s\n", rilLibPath); exit(-1); } if (hasLibArgs) { //false rilArgv = argv + i - 1; argc = argc -i + 1; } else { static char * newArgv[MAX_LIB_ARGS]; static char args[PROPERTY_VALUE_MAX]; rilArgv = newArgv; property_get(LIB_ARGS_PROPERTY, args, "");//通过属性系统读取"rild.libargs"属性值 argc = make_argv(args, rilArgv); } // Make sure there's a reasonable argv[0] rilArgv[0] = argv[0]; //调用RIL_Init函数来初始化rild，传入参数s_rilEnv，返回RIL_RadioFunctions地址 ④funcs = rilInit(&s_rilEnv, argc, rilArgv); //注册客户端事件处理接口RIL_RadioFunctions，并创建socket监听事件 ⑤RIL_register(funcs); done: while(1) { // sleep(UINT32_MAX) seems to return immediately on bionic sleep(0x00ffffff); } } 在main函数中主要完成以下工作:
1.解析命令行参数，通过判断是否为模拟器采取不同的方式来读取libreference-ril.so库的存放路径；
2.使用dlopen手动装载libreference-ril.so库；
3.启动事件循环处理；
4.从libreference-ril.so库中取得RIL_Init函数地址，并使用该函数将libril.so库中的RIL_Env接口注册到libreference-ril.so库，同时将libreference-ril.so库中的RIL_RadioFunctions接口注册到到libril.so库中，建立起libril.so库与libreference-ril.so库通信桥梁；
3.2启动事件循环处理eventLoop工作线程
建立多路I/O驱动机制的消息队列，用来接收上层发出的命令以及往Modem发送AT指令的工作，时整个RIL系统的核心部分。创建一个事件分发线程s_tid_dispatch，线程执行体为eventLoop。
hardware\ril\libril\Ril.cpp
extern "C" void RIL_startEventLoop(void) { int ret; pthread_attr_t attr; /* spin up eventLoop thread and wait for it to get started */ s_started = 0; pthread_mutex_lock(&s_startupMutex); pthread_attr_init (&attr); pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED); //创建一个工作线程eventLoop ret = pthread_create(&s_tid_dispatch, &attr, eventLoop, NULL); //确保函数返回前eventLoop线程启动运行 while (s_started == 0) { pthread_cond_wait(&s_startupCond, &s_startupMutex); } pthread_mutex_unlock(&s_startupMutex); if (ret < 0) { LOGE("Failed to create dispatch thread errno:%d", errno); return; } } eventLoop执行时序图：
static void * eventLoop(void *param) { int ret; int filedes[2]; ril_event_init(); //初始化请求队列 pthread_mutex_lock(&s_startupMutex); s_started = 1; //eventLoop线程运行标志位 pthread_cond_broadcast(&s_startupCond); pthread_mutex_unlock(&s_startupMutex); //创建匿名管道 ret = pipe(filedes); if (ret < 0) { LOGE("Error in pipe() errno:%d", errno); return NULL; } //s_fdWakeupRead为管道读端 s_fdWakeupRead = filedes[0]; //s_fdWakeupWrite为管道写端 s_fdWakeupWrite = filedes[1]; //设置管道读端为O_NONBLOCK非阻塞 fcntl(s_fdWakeupRead, F_SETFL, O_NONBLOCK); //初始化s_wakeupfd_event结构体的内容，句柄为s_fdWakeupRead，回调函数为 processWakeupCallback ril_event_set (&s_wakeupfd_event, s_fdWakeupRead, true,processWakeupCallback, NULL); ①rilEventAddWakeup (&s_wakeupfd_event); // Only returns on error ②ril_event_loop(); LOGE ("error in event_loop_base errno:%d", errno); return NULL; }在rild中定义了event的概念，Rild支持两种类型的事件：
1. 定时事件：根据事件的执行时间来启动执行，通过ril_timer_add添加到time_list队列中
2. Wakeup事件：这些事件的句柄fd将加入的select IO多路复用的句柄池readFDs中，当对应的fd可读时将触发这些事件。对于处于listen端的socket，fd可读表示有个客户端连接，此时需要调用accept接受连接。
事件定义如下：
struct ril_event { struct ril_event *next; struct ril_event *prev; int fd; //文件句柄 int index; //该事件在监控表中的索引 bool persist; //如果是保持的，则不从watch_list 中删除 struct timeval timeout; //任务执行时间 ril_event_cb func; //回调事件处理函数 void *param; //回调时参数 }; 在Rild进程中的几个重要事件有
static struct ril_event s_commands_event; ril_event_set (&s_commands_event, s_fdCommand, 1,processCommandsCallback, p_rs) static struct ril_event s_wakeupfd_event; ril_event_set (&s_wakeupfd_event, s_fdWakeupRead, true,processWakeupCallback, NULL) static struct ril_event s_listen_event; ril_event_set (&s_listen_event, s_fdListen, false,listenCallback, NULL) static struct ril_event s_wake_timeout_event; ril_timer_add(&(p_info->event), &myRelativeTime)；static struct ril_event s_debug_event; ril_event_set (&s_debug_event, s_fdDebug, true,debugCallback, NULL) 在RILD中定义了三个事件队列，用于处理不同的事件：
/事件监控队列
static struct ril_event * watch_table[MAX_FD_EVENTS];
//定时事件队列
static struct ril_event timer_list;
//处理事件队列
static struct ril_event pending_list; //待处理事件队列，事件已经触发，需要所回调处理的事件
3.2.1添加事件
1.添加Wakeup 事件
static void rilEventAddWakeup(struct ril_event *ev) { ril_event_add(ev); //向监控表watch_table添加一个s_wakeupfd_event事件 triggerEvLoop(); //向管道s_fdWakeupWrite中写入之来触发事件循环 } void ril_event_add(struct ril_event * ev) { dlog("~~~~ +ril_event_add ~~~~"); MUTEX_ACQUIRE(); for (int i = 0; i < MAX_FD_EVENTS; i++) { //遍历监控表watch_table if (watch_table[i] == NULL) { //从监控表中查找空闲的索引，然后把该任务加入到监控表中 watch_table[i] = ev; //向监控表中添加事件 ev->index = i; //事件的索引设置为在监控表中的索引 dlog("~~~~ added at %d ~~~~", i); dump_event(ev); FD_SET(ev->fd, &readFds); //将添加的事件对应的句柄添加到句柄池readFds中 if (ev->fd >= nfds) nfds = ev->fd+1; //修改句柄最大值 dlog("~~~~ nfds = %d ~~~~", nfds); break; } } MUTEX_RELEASE(); dlog("~~~~ -ril_event_add ~~~~"); } 2.添加定时事件
void ril_timer_add(struct ril_event * ev, struct timeval * tv) { dlog("~~~~ +ril_timer_add ~~~~"); MUTEX_ACQUIRE(); struct ril_event * list; if (tv != NULL) { list = timer_list.next; ev->fd = -1; // make sure fd is invalid struct timeval now; getNow(&now); timeradd(&now, tv, &ev->timeout); // keep list sorted while (timercmp(&list->timeout, &ev->timeout, < ) && (list != &timer_list)) { list = list->next; } // list now points to the first event older than ev addToList(ev, list); } MUTEX_RELEASE(); dlog("~~~~ -ril_timer_add ~~~~"); } 3.2.2触发事件
static void triggerEvLoop() { int ret; if (!pthread_equal(pthread_self(), s_tid_dispatch)) { //如果当前线程ID不等于事件分发线程eventLoop的线程ID do { ret = write (s_fdWakeupWrite, " ", 1); //向管道写端写入值1来触发eventLoop事件循环 } while (ret < 0 && errno == EINTR); } } 3.2.3处理事件
void ril_event_loop() { int n; fd_set rfds; struct timeval tv; struct timeval * ptv; for (;;) { memcpy(&rfds, &readFds, sizeof(fd_set)); if (-1 == calcNextTimeout(&tv)) { dlog("~~~~ no timers; blocking indefinitely ~~~~"); ptv = NULL; } else { dlog("~~~~ blocking for %ds + %dus ~~~~", (int)tv.tv_sec, (int)tv.tv_usec); ptv = &tv; } //使用select 函数等待在FDS 上,只要FDS 中记录的设备有数据到来，select 就会设置相应的标志位并返回。readFDS 记录了所有的事件相关设备句柄。readFDS 中句柄是在在AddEvent 加入的。 printReadies(&rfds); n = select(nfds, &rfds, NULL, NULL, ptv); printReadies(&rfds); dlog("~~~~ %d events fired ~~~~", n); if (n < 0) { if (errno == EINTR) continue; LOGE("ril_event: select error (%d)", errno); return; } processTimeouts(); //从timer_list中查询执行时间已到的事件，并添加到pending_list中 processReadReadies(&rfds, n); //从watch_table中查询数据可读的事件，并添加到pending_list中去处理，如果该事件不是持久事件，则同时从watch_table中删除 //遍历pending_list，调用事件处理回调函数处理所有事件 firePending(); } } 在eventLoop工作线程中，循环处理到来的事件及定时结束事件，整个处理流程如下图所示：
首先通过Linux中的select多路I/O复用对句柄池中的所有句柄进行监控，当有事件到来时select返回，否则阻塞。当select返回时，表示有事件的到来，通过调用processTimeouts函数来处理超时事件，处理方式是遍历time_list链表以查询超时事件，并将超时事件移入到pending_list链表中，接着调用processReadReadies函数来处理触发的事件，处理方式为遍历watch_table列表以查询触发的事件，并将触发的事件移入到pending_list链表中，如果该事件不是持久事件，还需要从watch_table列表中移除，当查询完两种待处理的事件并放入到pending_list链表中后，调用firePending函数对待处理的事件进行集中处理，处理方式为遍历链表，调用每一个事件的回调函数。
3.2.4超时事件查询
static void processTimeouts() { dlog("~~~~ +processTimeouts ~~~~"); MUTEX_ACQUIRE(); struct timeval now; struct ril_event * tev = timer_list.next; struct ril_event * next; getNow(&now); //获取当前时间 dlog("~~~~ Looking for timers <= %ds + %dus ~~~~", (int)now.tv_sec, (int)now.tv_usec); //如果当前时间大于事件的超时时间，则将该事件从timer_list中移除，添加到pending_list while ((tev != &timer_list) && (timercmp(&now, &tev->timeout, >))) { dlog("~~~~ firing timer ~~~~"); next = tev->next; removeFromList(tev); //从timer_list中移除事件 addToList(tev, &pending_list); //将事件添加到pending_list tev = next; } MUTEX_RELEASE(); dlog("~~~~ -processTimeouts ~~~~"); } 3.2.5可读事件查询
static void processReadReadies(fd_set * rfds, int n) { dlog("~~~~ +processReadReadies (%d) ~~~~", n); MUTEX_ACQUIRE(); //遍历watch_table数组，根据select返回的句柄n查找对应的事件 for (int i = 0; (i < MAX_FD_EVENTS) && (n > 0); i++) { struct ril_event * rev = watch_table[i]; //得到相应的事件 if (rev != NULL && FD_ISSET(rev->fd, rfds)) { addToList(rev, &pending_list); //将该事件添加到pending_list中 if (rev->persist == false) { //如果该事件不是持久事件还要从watch_table中移除 removeWatch(rev, i); } n--; } } MUTEX_RELEASE(); dlog("~~~~ -processReadReadies (%d) ~~~~", n); } 3.2.6事件处理
static void firePending() { dlog("~~~~ +firePending ~~~~"); struct ril_event * ev = pending_list.next; while (ev != &pending_list) { //遍历pending_list链表，处理链表中的所有事件 struct ril_event * next = ev->next; removeFromList(ev); //将处理完的事件从pending_list中移除 ev->func(ev->fd, 0, ev->param); //调用事件处理的回调函数 ev = next; } dlog("~~~~ -firePending ~~~~"); } 3.3RIL_Env定义
hardware\ril\include\telephony\ril.h
struct RIL_Env { //动态库完成请求后通知处理结果的接口 void (*OnRequestComplete)(RIL_Token t, RIL_Errno e,void *response, size_t responselen); //动态库unSolicited Response通知接口 void (*OnUnsolicitedResponse)(int unsolResponse, const void *data,size_t datalen); //向Rild提交一个超时任务的接口 void (*RequestTimedCallback) (RIL_TimedCallback callback,void *param, const struct timeval *relativeTime); };hardware\ril\rild\rild.c
s_rilEnv变量定义：
static struct RIL_Env s_rilEnv = { RIL_onRequestComplete, RIL_onUnsolicitedResponse, RIL_requestTimedCallback }; 在hardware\ril\libril\ril.cpp中实现了RIL_Env的各个接口函数
1.RIL_onRequestComplete
extern "C" void RIL_onRequestComplete(RIL_Token t, RIL_Errno e, void *response, size_t responselen) { RequestInfo *pRI; int ret; size_t errorOffset; pRI = (RequestInfo *)t; if (!checkAndDequeueRequestInfo(pRI)) { LOGE ("RIL_onRequestComplete: invalid RIL_Token"); return; } if (pRI->local > 0) { // Locally issued command...void only! // response does not go back up the command socket LOGD("C[locl]< %s", requestToString(pRI->pCI->requestNumber)); goto done; } appendPrintBuf("[%04d]< %s",pRI->token, requestToString(pRI->pCI->requestNumber)); if (pRI->cancelled == 0) { Parcel p; p.writeInt32 (RESPONSE_SOLICITED); p.writeInt32 (pRI->token); errorOffset = p.dataPosition(); p.writeInt32 (e); if (response != NULL) { // there is a response payload, no matter success or not. ret = pRI->pCI->responseFunction(p, response, responselen); /* if an error occurred, rewind and mark it */ if (ret != 0) { p.setDataPosition(errorOffset); p.writeInt32 (ret); } } if (e != RIL_E_SUCCESS) { appendPrintBuf("%s fails by %s", printBuf, failCauseToString(e)); } if (s_fdCommand < 0) { LOGD ("RIL onRequestComplete: Command channel closed"); } sendResponse(p); } done: free(pRI); } 通过调用responseXXX将底层响应传给客户进程
2.RIL_onUnsolicitedResponse
extern "C" void RIL_onUnsolicitedResponse(int unsolResponse, void *data, size_t datalen) { int unsolResponseIndex; int ret; int64_t timeReceived = 0; bool shouldScheduleTimeout = false; if (s_registerCalled == 0) { // Ignore RIL_onUnsolicitedResponse before RIL_register LOGW("RIL_onUnsolicitedResponse called before RIL_register"); return; } unsolResponseIndex = unsolResponse - RIL_UNSOL_RESPONSE_BASE; if ((unsolResponseIndex < 0) || (unsolResponseIndex >= (int32_t)NUM_ELEMS(s_unsolResponses))) { LOGE("unsupported unsolicited response code %d", unsolResponse); return; } // Grab a wake lock if needed for this reponse, // as we exit we'll either release it immediately // or set a timer to release it later. switch (s_unsolResponses[unsolResponseIndex].wakeType) { case WAKE_PARTIAL: grabPartialWakeLock(); shouldScheduleTimeout = true; break; case DONT_WAKE: default: // No wake lock is grabed so don't set timeout shouldScheduleTimeout = false; break; } // Mark the time this was received, doing this // after grabing the wakelock incase getting // the elapsedRealTime might cause us to goto // sleep. if (unsolResponse == RIL_UNSOL_NITZ_TIME_RECEIVED) { timeReceived = elapsedRealtime(); } appendPrintBuf("[UNSL]< %s", requestToString(unsolResponse)); Parcel p; p.writeInt32 (RESPONSE_UNSOLICITED); p.writeInt32 (unsolResponse); ret = s_unsolResponses[unsolResponseIndex].responseFunction(p, data, datalen); if (ret != 0) { // Problem with the response. Don't continue; goto error_exit; } // some things get more payload switch(unsolResponse) { case RIL_UNSOL_RESPONSE_RADIO_STATE_CHANGED: p.writeInt32(s_callbacks.onStateRequest()); appendPrintBuf("%s {%s}", printBuf, radioStateToString(s_callbacks.onStateRequest())); break; case RIL_UNSOL_NITZ_TIME_RECEIVED: // Store the time that this was received so the // handler of this message can account for // the time it takes to arrive and process. In // particular the system has been known to sleep // before this message can be processed. p.writeInt64(timeReceived); break; } ret = sendResponse(p); if (ret != 0 && unsolResponse == RIL_UNSOL_NITZ_TIME_RECEIVED) { // Unfortunately, NITZ time is not poll/update like everything // else in the system. So, if the upstream client isn't connected, // keep a copy of the last NITZ response (with receive time noted // above) around so we can deliver it when it is connected if (s_lastNITZTimeData != NULL) { free (s_lastNITZTimeData); s_lastNITZTimeData = NULL; } s_lastNITZTimeData = malloc(p.dataSize()); s_lastNITZTimeDataSize = p.dataSize(); memcpy(s_lastNITZTimeData, p.data(), p.dataSize()); } // For now, we automatically go back to sleep after TIMEVAL_WAKE_TIMEOUT // FIXME The java code should handshake here to release wake lock if (shouldScheduleTimeout) { // Cancel the previous request if (s_last_wake_timeout_info != NULL) { s_last_wake_timeout_info->userParam = (void *)1; } s_last_wake_timeout_info= internalRequestTimedCallback(wakeTimeoutCallback, NULL, &TIMEVAL_WAKE_TIMEOUT); } return; error_exit: if (shouldScheduleTimeout) { releaseWakeLock(); } } 这个函数处理modem从网络端接收到的各种事件，如网络信号变化，拨入的电话，收到短信等。然后传给客户进程。
3.RIL_requestTimedCallback
extern "C" void RIL_requestTimedCallback (RIL_TimedCallback callback, void *param, const struct timeval *relativeTime) { internalRequestTimedCallback (callback, param, relativeTime); } static UserCallbackInfo *internalRequestTimedCallback (RIL_TimedCallback callback, void *param, const struct timeval *relativeTime) { struct timeval myRelativeTime; UserCallbackInfo *p_info; p_info = (UserCallbackInfo *) malloc (sizeof(UserCallbackInfo)); p_info->p_callback = callback; p_info->userParam = param; if (relativeTime == NULL) { /* treat null parameter as a 0 relative time */ memset (&myRelativeTime, 0, sizeof(myRelativeTime)); } else { /* FIXME I think event_add's tv param is really const anyway */ memcpy (&myRelativeTime, relativeTime, sizeof(myRelativeTime)); } ril_event_set(&(p_info->event), -1, false, userTimerCallback, p_info); ril_timer_add(&(p_info->event), &myRelativeTime); triggerEvLoop(); return p_info; } 3.4RIL_RadioFunctions定义
客户端向Rild发送请求的接口，由各手机厂商实现。
hardware\ril\include\telephony\Ril.h
typedef struct { int version; //Rild版本 RIL_RequestFunc onRequest; //AP请求接口 RIL_RadioStateRequest onStateRequest;//BP状态查询 RIL_Supports supports; RIL_Cancel onCancel; RIL_GetVersion getVersion;//动态库版本 } RIL_RadioFunctions; static const RIL_RadioFunctions s_callbacks = { RIL_VERSION, onRequest, currentState, onSupports, onCancel, getVersion }; 在hardware\ril\reference-ril\reference-ril.c中实现了RIL_RadioFunctions的各个接口函数
1.onRequest
static void onRequest (int request, void *data, size_t datalen, RIL_Token t) { ATResponse *p_response; int err; LOGD("onRequest: %s", requestToString(request)); /* Ignore all requests except RIL_REQUEST_GET_SIM_STATUS * when RADIO_STATE_UNAVAILABLE. */ if (sState == RADIO_STATE_UNAVAILABLE && request != RIL_REQUEST_GET_SIM_STATUS ) { RIL_onRequestComplete(t, RIL_E_RADIO_NOT_AVAILABLE, NULL, 0); return; } /* Ignore all non-power requests when RADIO_STATE_OFF * (except RIL_REQUEST_GET_SIM_STATUS) */ if (sState == RADIO_STATE_OFF&& !(request == RIL_REQUEST_RADIO_POWER || request == RIL_REQUEST_GET_SIM_STATUS) ) { RIL_onRequestComplete(t, RIL_E_RADIO_NOT_AVAILABLE, NULL, 0); return; } switch (request) { case RIL_REQUEST_GET_SIM_STATUS: { RIL_CardStatus *p_card_status; char *p_buffer; int buffer_size; int result = getCardStatus(&p_card_status); if (result == RIL_E_SUCCESS) { p_buffer = (char *)p_card_status; buffer_size = sizeof(*p_card_status); } else { p_buffer = NULL; buffer_size = 0; } RIL_onRequestComplete(t, result, p_buffer, buffer_size); freeCardStatus(p_card_status); break; } case RIL_REQUEST_GET_CURRENT_CALLS: requestGetCurrentCalls(data, datalen, t); break; case RIL_REQUEST_DIAL: requestDial(data, datalen, t); break; case RIL_REQUEST_HANGUP: requestHangup(data, datalen, t); break; case RIL_REQUEST_HANGUP_WAITING_OR_BACKGROUND: // 3GPP 22.030 6.5.5 // "Releases all held calls or sets User Determined User Busy // (UDUB) for a waiting call." at_send_command("AT+CHLD=0", NULL); /* success or failure is ignored by the upper layer here. it will call GET_CURRENT_CALLS and determine success that way */ RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); break; case RIL_REQUEST_HANGUP_FOREGROUND_RESUME_BACKGROUND: // 3GPP 22.030 6.5.5 // "Releases all active calls (if any exist) and accepts // the other (held or waiting) call." at_send_command("AT+CHLD=1", NULL); /* success or failure is ignored by the upper layer here. it will call GET_CURRENT_CALLS and determine success that way */ RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); break; case RIL_REQUEST_SWITCH_WAITING_OR_HOLDING_AND_ACTIVE: // 3GPP 22.030 6.5.5 // "Places all active calls (if any exist) on hold and accepts // the other (held or waiting) call." at_send_command("AT+CHLD=2", NULL); #ifdef WORKAROUND_ERRONEOUS_ANSWER s_expectAnswer = 1; #endif /* WORKAROUND_ERRONEOUS_ANSWER */ /* success or failure is ignored by the upper layer here. it will call GET_CURRENT_CALLS and determine success that way */ RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); break; case RIL_REQUEST_ANSWER: at_send_command("ATA", NULL); #ifdef WORKAROUND_ERRONEOUS_ANSWER s_expectAnswer = 1; #endif /* WORKAROUND_ERRONEOUS_ANSWER */ /* success or failure is ignored by the upper layer here. it will call GET_CURRENT_CALLS and determine success that way */ RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); break; case RIL_REQUEST_CONFERENCE: // 3GPP 22.030 6.5.5 // "Adds a held call to the conversation" at_send_command("AT+CHLD=3", NULL); /* success or failure is ignored by the upper layer here. it will call GET_CURRENT_CALLS and determine success that way */ RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); break; case RIL_REQUEST_UDUB: /* user determined user busy */ /* sometimes used: ATH */ at_send_command("ATH", NULL); /* success or failure is ignored by the upper layer here. it will call GET_CURRENT_CALLS and determine success that way */ RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); break; case RIL_REQUEST_SEPARATE_CONNECTION: { char cmd[12]; int party = ((int*)data)[0]; // Make sure that party is in a valid range. // (Note: The Telephony middle layer imposes a range of 1 to 7. // It's sufficient for us to just make sure it's single digit.) if (party > 0 && party < 10) { sprintf(cmd, "AT+CHLD=2%d", party); at_send_command(cmd, NULL); RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); } else { RIL_onRequestComplete(t, RIL_E_GENERIC_FAILURE, NULL, 0); } } break; case RIL_REQUEST_SIGNAL_STRENGTH: requestSignalStrength(data, datalen, t); break; case RIL_REQUEST_REGISTRATION_STATE: case RIL_REQUEST_GPRS_REGISTRATION_STATE: requestRegistrationState(request, data, datalen, t); break; case RIL_REQUEST_OPERATOR: requestOperator(data, datalen, t); break; case RIL_REQUEST_RADIO_POWER: requestRadioPower(data, datalen, t); break; case RIL_REQUEST_DTMF: { char c = ((char *)data)[0]; char *cmd; asprintf(&cmd, "AT+VTS=%c", (int)c); at_send_command(cmd, NULL); free(cmd); RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); break; } case RIL_REQUEST_SEND_SMS: requestSendSMS(data, datalen, t); break; case RIL_REQUEST_SETUP_DATA_CALL: requestSetupDataCall(data, datalen, t); break; case RIL_REQUEST_SMS_ACKNOWLEDGE: requestSMSAcknowledge(data, datalen, t); break; case RIL_REQUEST_GET_IMSI: p_response = NULL; err = at_send_command_numeric("AT+CIMI", &p_response); if (err < 0 || p_response->success == 0) { RIL_onRequestComplete(t, RIL_E_GENERIC_FAILURE, NULL, 0); } else { RIL_onRequestComplete(t, RIL_E_SUCCESS, p_response->p_intermediates->line, sizeof(char *)); } at_response_free(p_response); break; case RIL_REQUEST_GET_IMEI: p_response = NULL; err = at_send_command_numeric("AT+CGSN", &p_response); if (err < 0 || p_response->success == 0) { RIL_onRequestComplete(t, RIL_E_GENERIC_FAILURE, NULL, 0); } else { RIL_onRequestComplete(t, RIL_E_SUCCESS, p_response->p_intermediates->line, sizeof(char *)); } at_response_free(p_response); break; case RIL_REQUEST_SIM_IO: requestSIM_IO(data,datalen,t); break; case RIL_REQUEST_SEND_USSD: requestSendUSSD(data, datalen, t); break; case RIL_REQUEST_CANCEL_USSD: p_response = NULL; err = at_send_command_numeric("AT+CUSD=2", &p_response); if (err < 0 || p_response->success == 0) { RIL_onRequestComplete(t, RIL_E_GENERIC_FAILURE, NULL, 0); } else { RIL_onRequestComplete(t, RIL_E_SUCCESS, p_response->p_intermediates->line, sizeof(char *)); } at_response_free(p_response); break; case RIL_REQUEST_SET_NETWORK_SELECTION_AUTOMATIC: at_send_command("AT+COPS=0", NULL); break; case RIL_REQUEST_DATA_CALL_LIST: requestDataCallList(data, datalen, t); break; case RIL_REQUEST_QUERY_NETWORK_SELECTION_MODE: requestQueryNetworkSelectionMode(data, datalen, t); break; case RIL_REQUEST_OEM_HOOK_RAW: // echo back data RIL_onRequestComplete(t, RIL_E_SUCCESS, data, datalen); break; case RIL_REQUEST_OEM_HOOK_STRINGS: { int i; const char ** cur; LOGD("got OEM_HOOK_STRINGS: 0x%8p %lu", data, (long)datalen); for (i = (datalen / sizeof (char *)), cur = (const char **)data ; i > 0 ; cur++, i --) { LOGD("> '%s'", *cur); } // echo back strings RIL_onRequestComplete(t, RIL_E_SUCCESS, data, datalen); break; } case RIL_REQUEST_WRITE_SMS_TO_SIM: requestWriteSmsToSim(data, datalen, t); break; case RIL_REQUEST_DELETE_SMS_ON_SIM: { char * cmd; p_response = NULL; asprintf(&cmd, "AT+CMGD=%d", ((int *)data)[0]); err = at_send_command(cmd, &p_response); free(cmd); if (err < 0 || p_response->success == 0) { RIL_onRequestComplete(t, RIL_E_GENERIC_FAILURE, NULL, 0); } else { RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); } at_response_free(p_response); break; } case RIL_REQUEST_ENTER_SIM_PIN: case RIL_REQUEST_ENTER_SIM_PUK: case RIL_REQUEST_ENTER_SIM_PIN2: case RIL_REQUEST_ENTER_SIM_PUK2: case RIL_REQUEST_CHANGE_SIM_PIN: case RIL_REQUEST_CHANGE_SIM_PIN2: requestEnterSimPin(data, datalen, t); break; case RIL_REQUEST_GSM_SMS_BROADCAST_ACTIVATION: requestSmsBroadcastActivation(0,data, datalen, t); break; case RIL_REQUEST_GSM_SET_BROADCAST_SMS_CONFIG: LOGD("onRequest RIL_REQUEST_GSM_SET_BROADCAST_SMS_CONFIG"); requestSetSmsBroadcastConfig(0,data, datalen, t); break; case RIL_REQUEST_GSM_GET_BROADCAST_SMS_CONFIG: requestGetSmsBroadcastConfig(0,data, datalen, t); break; default: RIL_onRequestComplete(t, RIL_E_REQUEST_NOT_SUPPORTED, NULL, 0); break; } } 对每一个RIL_REQUEST_XXX请求转化成相应的ATcommand，发送给modem，然后睡眠等待，当收到ATcommand的最终响应后，线程被唤醒，将响应传给客户端进程。
2.currentState
static RIL_RadioState currentState() { return sState; } 3.onSupports
static int onSupports (int requestCode) { //@@@ todo return 1; } 4.onCancel
static void onCancel (RIL_Token t) { //@@@todo } 5.getVersion
static const char * getVersion(void) { return "android reference-ril 1.0"; } 3.5注册RIL_Env接口
由于各手机厂商的AT指令差异，因此与modem交互层需要各手机厂商实现，以动态库的形式提供。作为介于modem与上层的中间层，即要与底层交互也要与上层通信，因此就需要定义一个接口来衔接RILD与动态库，RIL_Env和RIL_RadioFunctions接口就是libril.so与librefrence.so通信的桥梁。是Rild架构中用于隔离通用代码和厂商代码的接口，RIL_Env由通用代码实现，而RIL_RadioFunctions则是由厂商代码实现。
3.6RIL_Init的主要任务：
1. 向librefrence.so注册libril.so提供的接口RIL_Env；
2. 创建一个mainLoop工作线程，用于初始化AT模块，并监控AT模块的状态，一旦AT被关闭，则重新打开并初始化AT；
3. 当AT被打开后，mainLoop工作线程将向Rild提交一个定时事件，并触发eventLoop来完成对modem的初始化；
4. 创建一个readLoop工作线程，用于从AT串口中读取数据；
5.返回librefrence.so提供的接口RIL_RadioFunctions；
hardware\ril\reference-ril\reference-ril.c
const RIL_RadioFunctions *RIL_Init(const struct RIL_Env *env, int argc, char **argv) { int ret; int fd = -1; int opt; pthread_attr_t attr; s_rilenv = env; //将ril.cpp中定义的RIL_Env注册到reference-ril.c中的s_rilenv while ( -1 != (opt = getopt(argc, argv, "p:d:s:"))) { switch (opt) { case 'p': s_port = atoi(optarg); if (s_port == 0) { usage(argv[0]); return NULL; } LOGI("Opening loopback port %d\n", s_port); break; case 'd': s_device_path = optarg; LOGI("Opening tty device %s\n", s_device_path); break; case 's': s_device_path = optarg; s_device_socket = 1; LOGI("Opening socket %s\n", s_device_path); break; default: usage(argv[0]); return NULL; } } if (s_port < 0 && s_device_path == NULL) { usage(argv[0]); return NULL; } pthread_attr_init (&attr); pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED); //创建一个mainLoop线程 ret = pthread_create(&s_tid_mainloop, &attr, mainLoop, NULL); //将reference-ril.c中定义的RIL_RadioFunctions返回并注册到ril.cpp中的s_callbacks return &s_callbacks; } mainLoop工作线程是用来初始化并监控AT模块的，一旦AT模块被关闭，就自动打开。
static void * mainLoop(void *param) { int fd; int ret; AT_DUMP("== ", "entering mainLoop()", -1 ); //为AT模块设置回调函数 at_set_on_reader_closed(onATReaderClosed); at_set_on_timeout(onATTimeout); for (;;) { fd = -1; while (fd < 0) { //获得串口AT模块的设备文件描述符 if (s_port > 0) { fd = socket_loopback_client(s_port, SOCK_STREAM); } else if (s_device_socket) { if (!strcmp(s_device_path, "/dev/socket/qemud")) { /* Qemu-specific control socket */ fd = socket_local_client( "qemud", ANDROID_SOCKET_NAMESPACE_RESERVED,SOCK_STREAM ); if (fd >= 0 ) { char answer[2]; if ( write(fd, "gsm", 3) != 3 ||read(fd, answer, 2) != 2 || memcmp(answer, "OK", 2) != 0) { close(fd); fd = -1; } } } else fd = socket_local_client( s_device_path, ANDROID_SOCKET_NAMESPACE_FILESYSTEM,SOCK_STREAM ); } else if (s_device_path != NULL) { fd = open (s_device_path, O_RDWR); if ( fd >= 0 && !memcmp( s_device_path, "/dev/ttyS", 9 ) ) { /* disable echo on serial ports */ struct termios ios; tcgetattr( fd, &ios ); ios.c_lflag = 0; /* disable ECHO, ICANON, etc... */ tcsetattr( fd, TCSANOW, &ios ); } } if (fd < 0) { perror ("opening AT interface. retrying..."); sleep(10); } } s_closed = 0; //打开AT模块，创建AT读取线程s_tid_reader，fd为modem设备文件句柄 ret = at_open(fd, onUnsolicited); if (ret < 0) { LOGE ("AT error %d on at_open\n", ret); return 0; } //向Rild提交超时任务 RIL_requestTimedCallback(initializeCallback, NULL, &TIMEVAL_0); sleep(1); //如果AT模块被关闭，则waitForClose返回，重新打开AT，如果AT已打开，则阻塞 waitForClose(); LOGI("Re-opening after close"); } } 3.6.1.打开AT模块
通过at_open打开文件描述符为fd的AT串口设备，并注册回调函数ATUnsolHandler
int at_open(int fd, ATUnsolHandler h) { int ret; pthread_t tid; pthread_attr_t attr; s_fd = fd; s_unsolHandler = h; s_readerClosed = 0; s_responsePrefix = NULL; s_smsPDU = NULL; sp_response = NULL; /* Android power control ioctl */ #ifdef HAVE_ANDROID_OS #ifdef OMAP_CSMI_POWER_CONTROL ret = ioctl(fd, OMAP_CSMI_TTY_ENABLE_ACK); if(ret == 0) { int ack_count; int read_count; int old_flags; char sync_buf[256]; old_flags = fcntl(fd, F_GETFL, 0); fcntl(fd, F_SETFL, old_flags | O_NONBLOCK); do { ioctl(fd, OMAP_CSMI_TTY_READ_UNACKED, &ack_count); read_count = 0; do { ret = read(fd, sync_buf, sizeof(sync_buf)); if(ret > 0) read_count += ret; } while(ret > 0 || (ret < 0 && errno == EINTR)); ioctl(fd, OMAP_CSMI_TTY_ACK, &ack_count); } while(ack_count > 0 || read_count > 0); fcntl(fd, F_SETFL, old_flags); s_readCount = 0; s_ackPowerIoctl = 1; } else s_ackPowerIoctl = 0; #else // OMAP_CSMI_POWER_CONTROL s_ackPowerIoctl = 0; #endif // OMAP_CSMI_POWER_CONTROL #endif /*HAVE_ANDROID_OS*/ pthread_attr_init (&attr); pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED); //创建readerLoop工作线程，该线程用于从串口读取数据 ret = pthread_create(&s_tid_reader, &attr, readerLoop, &attr); if (ret < 0) { perror ("pthread_create"); return -1; } return 0; } 3.6.2.添加定时事件RIL_requestTimedCallback
RIL_requestTimedCallback(initializeCallback, NULL, &TIMEVAL_0); #define RIL_requestTimedCallback(a,b,c) s_rilenv->RequestTimedCallback(a,b,c) 向定时事件队列中添加一个定时事件，该事件的处理函数为initializeCallback，用于发送一些AT指令来初始化BP的modem。
3.6.3.readLoop工作线程
Read loop 解析从Modem 发过来的回应。如果遇到URC 则通过handleUnsolicited 上报的RIL_JAVA。如果是命令的应答，则通过handleFinalResponse 通知send_at_command 有应答结果。
static void *readerLoop(void *arg) { for (;;) { const char * line; line = readline(); if (line == NULL) { break; } if(isSMSUnsolicited(line)) { //判断是否是SMS 通知 char *line1; const char *line2; line1 = strdup(line); line2 = readline(); if (line2 == NULL) { break; } if (s_unsolHandler != NULL) { s_unsolHandler (line1, line2); //回调通知SMS } free(line1); } else { processLine(line); //处理接收到的数据，根据line中的指令调用不同的回调函数 } #ifdef HAVE_ANDROID_OS if (s_ackPowerIoctl > 0) { /* acknowledge that bytes have been read and processed */ ioctl(s_fd, OMAP_CSMI_TTY_ACK, &s_readCount); s_readCount = 0; } #endif /*HAVE_ANDROID_OS*/ } onReaderClosed(); return NULL; } 3.6.4.注册RIL_RadioFunctions接口
hardware\ril\libril\ril.cpp
extern "C" void RIL_register (const RIL_RadioFunctions *callbacks) { int ret; int flags; //版本验证 if (callbacks == NULL || ((callbacks->version != RIL_VERSION)&& (callbacks->version < 2))) { return; } if (callbacks->version < RIL_VERSION) { LOGE ("RIL_register: upgrade RIL to version %d current version=%d", RIL_VERSION, callbacks->version); } if (s_registerCalled > 0) { LOGE("RIL_register has been called more than once. "Subsequent call ignored"); return; } //将reference-ril.c中定义的RIL_RadioFunctions注册到ril.cpp中 memcpy(&s_callbacks, callbacks, sizeof (RIL_RadioFunctions)); s_registerCalled = 1; for (int i = 0; i < (int)NUM_ELEMS(s_commands); i++) { assert(i == s_commands[i].requestNumber); //序号验证 } for (int i = 0; i < (int)NUM_ELEMS(s_unsolResponses); i++) { assert(i + RIL_UNSOL_RESPONSE_BASE== s_unsolResponses[i].requestNumber); } // old standalone impl wants it here. if (s_started == 0) { RIL_startEventLoop(); } // 得到名为rild的socket句柄 s_fdListen = android_get_control_socket(SOCKET_NAME_RIL); if (s_fdListen < 0) { LOGE("Failed to get socket '" SOCKET_NAME_RIL "'"); exit(-1); } // 监听该socket ret = listen(s_fdListen, 4); if (ret < 0) { LOGE("Failed to listen on control socket '%d': %s",s_fdListen, strerror(errno)); exit(-1); } /* 设置s_listen_event事件，一旦有客户端连接，即s_fdListen可读就会导致eventLoop工作线程中的select返回，因为该事件不是持久的，因此调用为listenCallback处理完后，将从watch_table移除该事件，所以Rild只支持一个客户端连接*/ ril_event_set (&s_listen_event, s_fdListen, false,listenCallback, NULL); /* 添加s_listen_event事件，并触发eventLoop工作线程 */ rilEventAddWakeup (&s_listen_event); #if 1 // 得到调试socket的句柄rild-debug s_fdDebug = android_get_control_socket(SOCKET_NAME_RIL_DEBUG); if (s_fdDebug < 0) { LOGE("Failed to get socket '" SOCKET_NAME_RIL_DEBUG "' errno:%d", errno); exit(-1); } //监听该socket ret = listen(s_fdDebug, 4); if (ret < 0) { LOGE("Failed to listen on ril debug socket '%d': %s",s_fdDebug, strerror(errno)); exit(-1); } /* 设置s_debug_event事件 */ ril_event_set (&s_debug_event, s_fdDebug, true,debugCallback, NULL); /* 添加s_debug_event事件，并触发eventLoop工作线程 */ rilEventAddWakeup (&s_debug_event); #endif } 打开监听端口，接收来自客户端进程的命令请求，当与客户进程连接建立时调用listenCallback函数，创建单独线程监视并处理所有事件源。
3.7.客户端连接处理
s_listen_event事件用于处理上层客户端的socket连接，当得到socket连接请求时，eventLoop工作线程里的select返回并自动调用listenCallback回调函数进行处理：
static void listenCallback (int fd, short flags, void *param) { int ret; int err; int is_phone_socket; RecordStream *p_rs; commthread_data_t *user_data = NULL; user_data =(commthread_data_t *)malloc(sizeof(commthread_data_t)); struct sockaddr_un peeraddr; socklen_t socklen = sizeof (peeraddr); struct ucred creds; socklen_t szCreds = sizeof(creds); struct passwd *pwd = NULL; assert (s_fdCommand < 0); assert (fd == s_fdListen); //接收一个客户端的连接，并将该socket连接保存在变量s_fdCommand中 s_fdCommand = accept(s_fdListen, (sockaddr *) &peeraddr, &socklen); if (s_fdCommand < 0 ) { LOGE("Error on accept() errno:%d", errno); /* start listening for new connections again */ rilEventAddWakeup(&s_listen_event); return; } /* 对客户端权限判断，判断是否是进程组ID为radio的进程发起的连接*/ errno = 0; is_phone_socket = 0; err = getsockopt(s_fdCommand, SOL_SOCKET, SO_PEERCRED, &creds, &szCreds); if (err == 0 && szCreds > 0) { errno = 0; pwd = getpwuid(creds.uid); if (pwd != NULL) { if (strcmp(pwd->pw_name, PHONE_PROCESS) == 0) { is_phone_socket = 1; } else { LOGE("RILD can't accept socket from process %s", pwd->pw_name); } } else { LOGE("Error on getpwuid() errno: %d", errno); } } else { LOGD("Error on getsockopt() errno: %d", errno); } if ( !is_phone_socket ) { LOGE("RILD must accept socket from %s", PHONE_PROCESS); close(s_fdCommand); s_fdCommand = -1; onCommandsSocketClosed(); /* start listening for new connections again */ rilEventAddWakeup(&s_listen_event); return; } #if 0 if(s_dualSimMode) { if(s_sim_num == 0) { property_get(SIM_POWER_PROPERTY, prop, "0"); if(!strcmp(prop, "0")) { property_set(SIM_POWER_PROPERTY, "1"); s_callbacks.powerSIM(NULL); } } else if(s_sim_num == 1) { property_get(SIM_POWER_PROPERTY1, prop, "0"); if(!strcmp(prop, "0")) { property_set(SIM_POWER_PROPERTY1, "1"); s_callbacks.powerSIM(NULL); } } } else { property_get(SIM_POWER_PROPERTY, prop, "0"); if(!strcmp(prop, "0")) { property_set(SIM_POWER_PROPERTY, "1"); s_callbacks.powerSIM(NULL); } } #endif //p_rs为RecordStream类型，它内部会分配一个缓冲区来存储客户端发送过来的数据 p_rs = record_stream_new(s_fdCommand, MAX_COMMAND_BYTES); //添加一个针对接收到的客户端连接的处理事件，从而在eventLoop工作线程中处理该客户端的各种请求 ril_event_set (&s_commands_event, s_fdCommand, 1,processCommandsCallback, p_rs); rilEventAddWakeup (&s_commands_event); onNewCommandConnect(); } 3.8.客户端通信处理
在listenCallback中首先接收客户端的连接请求，并验证客户端的权限，同时将该客户端以事件的形式添加到eventLoop工作线程中进行监控，当该客户端有数据请求时，eventLoop工作线程从select中返回，并自动调用processCommandsCallback回调函数：
static void processCommandsCallback(int fd, short flags, void *param) { RecordStream *p_rs; void *p_record; size_t recordlen; int ret; assert(fd == s_fdCommand); p_rs = (RecordStream *)param; for (;;) { //循环处理客户端发送过来的AT命令 //读取一条AT命令 ret = record_stream_get_next(p_rs, &p_record, &recordlen); if (ret == 0 && p_record == NULL) { break; } else if (ret < 0) { break; } else if (ret == 0) { /* && p_record != NULL */ //处理客户端发送过来的AT命令 processCommandBuffer(p_record, recordlen); } } if (ret == 0 || !(errno == EAGAIN || errno == EINTR)) { if (ret != 0) { LOGE("error on reading command socket errno:%d\n", errno); } else { LOGW("EOS. Closing command socket."); } close(s_fdCommand); s_fdCommand = -1; ril_event_del(&s_commands_event); record_stream_free(p_rs); rilEventAddWakeup(&s_listen_event); onCommandsSocketClosed(); } } 通过processCommandBuffer函数来处理每一条AT命令：
static int processCommandBuffer(void *buffer, size_t buflen) { Parcel p; status_t status; int32_t request; int32_t token; RequestInfo *pRI; int ret; p.setData((uint8_t *) buffer, buflen); // status checked at end status = p.readInt32(&request); status = p.readInt32 (&token); if (status != NO_ERROR) { LOGE("invalid request block"); return 0; } if (request < 1 || request >= (int32_t)NUM_ELEMS(s_commands)) { LOGE("unsupported request code %d token %d", request, token); return 0; } pRI = (RequestInfo *)calloc(1, sizeof(RequestInfo)); pRI->token = token; //AT命令标号 pRI->pCI = &(s_commands[request]); //根据request找到s_commands命令数组中的指定AT命令 ret = pthread_mutex_lock(&s_pendingRequestsMutex); assert (ret == 0); pRI->p_next = s_pendingRequests; s_pendingRequests = pRI; ret = pthread_mutex_unlock(&s_pendingRequestsMutex); assert (ret == 0); //调用指定AT命令的dispatch函数，根据接收来自客户进程的命令和参数，调用onRequest进行处理。 pRI->pCI->dispatchFunction(p, pRI); return 0; } 打电话的AT命令：{RIL_REQUEST_DIAL, dispatchDial, responseVoid},
发短信的AT命令：{RIL_REQUEST_SEND_SMS, dispatchStrings, responseSMS},
4.电话拨打流程
static void dispatchDial (Parcel &p, RequestInfo *pRI) { RIL_Dial dial; //RIL_Dial存储了打电话的所有信息 RIL_UUS_Info uusInfo; int32_t sizeOfDial; int32_t t; .................. //初始化dial变量 s_callbacks.onRequest(pRI->pCI->requestNumber, &dial, sizeOfDial, pRI); ................. return; } s_callbacks.onRequest其实就是调用RIL_RadioFunctions中的onRequest函数，该函数在前面已介绍过了。
static void onRequest (int request, void *data, size_t datalen, RIL_Token t) { switch (request) { case RIL_REQUEST_DIAL: requestDial(data, datalen, t); break; } }static void requestDial(void *data, size_t datalen, RIL_Token t) { RIL_Dial *p_dial; char *cmd; const char *clir; int ret; p_dial = (RIL_Dial *)data; switch (p_dial->clir) { case 1: clir = "I"; break; /*invocation*/ case 2: clir = "i"; break; /*suppression*/ default: case 0: clir = ""; break; /*subscription default*/ } //向串口发送AT指令 ret = at_send_command(cmd, NULL); free(cmd); //通知请求结果 RIL_onRequestComplete(t, RIL_E_SUCCESS, NULL, 0); } 向AT发送完拨号指令后，通过RIL_onRequestComplete返回处理结果，RIL_onRequestComplete实际上是RIL_Env中的OnRequestComplete函数，在前面我们也介绍过了
extern "C" void RIL_onRequestComplete(RIL_Token t, RIL_Errno e, void *response, size_t responselen) { RequestInfo *pRI; int ret; size_t errorOffset; pRI = (RequestInfo *)t; //该请求已经处理，需要从请求队列中移除该请求 if (!checkAndDequeueRequestInfo(pRI)) { LOGE ("RIL_onRequestComplete: invalid RIL_Token"); return; } if (pRI->local > 0) { ........... sendResponse(p); } done: free(pRI); } static int sendResponse (Parcel &p) { return sendResponseRaw(p.data(), p.dataSize()); //将结果发送给JAVA RIL客户端 } static int sendResponseRaw (const void *data, size_t dataSize) { int fd = s_fdCommand; int ret; uint32_t header; if (s_fdCommand < 0) { return -1; } if (dataSize > MAX_COMMAND_BYTES) { return -1; } pthread_mutex_lock(&s_writeMutex); header = htonl(dataSize); ret = blockingWrite(fd, (void *)&header, sizeof(header)); if (ret < 0) { pthread_mutex_unlock(&s_writeMutex); return ret; } ret = blockingWrite(fd, data, dataSize); if (ret < 0) { pthread_mutex_unlock(&s_writeMutex); return ret; } pthread_mutex_unlock(&s_writeMutex); return 0; } 拨打电话的时序图如下：
Rild通过onRequest向动态库提交一个请求，然后返回，动态库处理完请求后，处理结果通过回调接口通知客户端
更多相关文章
- SpringBoot 2.0 中 HikariCP 数据库连接池原理解析
- android自定义控件宽高的获取
- Android(安卓)| 如何给Fragment里的button添加监听事件
- runOnUiThread()方法
- Android(安卓)SurfaceView 双缓冲机制
- 异常：ava.lang. NoClassDefFoundError: android .os.AsyncTask
- 自定义ViewGroup实现瀑布流效果
- Android(安卓)7.0 ActivityManagerService(1) AMS的启动过程
- android input 命令总结