课程答疑和新课信息:QQ交流群:422901085进行课程讨论
FrameWork入门课视频链接:https://edu.csdn.net/course/detail/30298
FrameWork实战课1视频链接:https://edu.csdn.net/course/detail/30275
专题博客系列:
Android 8.1 zygote 启动过程源码
Android Framework实战视频–Zygote的fork进程篇
Android Framework实战视频–SystemServer启动篇
Android Framework实战视频–SystemServer启动FallbackHome篇
Android Framework实战视频–FallbackHome进程启动及Activity启动篇
Android Framework实战视频–FallbackHome结束启动Launcher篇
Android Framework实战视频–BootAnimation的启动源码分析(Android8.1)
Android Framework实战视频–init进程的bootanimation启动源码分析(补充Android 10部分的BootAnimation的启动源码分析)
提示:针对有的同学可能代码较新这里基于Android 10源码对init部分进行分析
针对Android init进程启动property service的启动部分流程如下:
init进程的入口main函数已经不在init.cpp而是在main.cpp
路径如下:
system/core/init/main.cpp
int main(int argc, char** argv) {
#if __has_feature(address_sanitizer)
__asan_set_error_report_callback(AsanReportCallback);
#endif
if (!strcmp(basename(argv[0]), "ueventd")) {
return ueventd_main(argc, argv);
}
if (argc > 1) {
if (!strcmp(argv[1], "subcontext")) {
android::base::InitLogging(argv, &android::base::KernelLogger);
const BuiltinFunctionMap function_map;
return SubcontextMain(argc, argv, &function_map);
}
if (!strcmp(argv[1], "selinux_setup")) {
return SetupSelinux(argv);
}
if (!strcmp(argv[1], "second_stage")) {
return SecondStageMain(argc, argv);//这里是关键,二阶段启动
}
}
return FirstStageMain(argc, argv);
}
这里主要看的是二阶段启动方法SecondStageMain,该方法又是回到init.cpp
int SecondStageMain(int argc, char** argv) {
//省略
StartPropertyService(&property_fd);
if (auto result = epoll.RegisterHandler(property_fd, HandlePropertyFd); !result) {
//监听fd的事件
LOG(FATAL) << "Could not register epoll handler for property fd: " << result.error();
}
//省略
return 0;
}
这里又看一个熟悉的方法StartPropertyService,而且传递进去了一个property_fd地址,然后对property_fd进行监听,该方法如下:
void StartPropertyService(int* epoll_socket) {
property_set("ro.property_service.version", "2");
int sockets[2];
if (socketpair(AF_UNIX, SOCK_SEQPACKET | SOCK_CLOEXEC, 0, sockets) != 0) {
//特别注意这里双工通信
PLOG(FATAL) << "Failed to socketpair() between property_service and init";
}
//特别注意这里双工通信,可以实现0端写入1端读取,1端写入0端读取
*epoll_socket = sockets[0];//把fd传递给了指针
init_socket = sockets[1];
property_set_fd = CreateSocket(PROP_SERVICE_NAME, SOCK_STREAM | SOCK_CLOEXEC | SOCK_NONBLOCK,false, 0666, 0, 0, nullptr);//创建PROP_SERVICE_NAME 的socket进行通信
if (property_set_fd == -1) {
PLOG(FATAL) << "start_property_service socket creation failed";
}
listen(property_set_fd, 8);
std::thread{
PropertyServiceThread}.detach();//启动线程监听数据
property_set = [](const std::string& key, const std::string& value) -> uint32_t {
android::base::SetProperty(key, value);
return 0;
};
}
这里其实和8.1差不多,主要创建socket,监听socket绑定和接收数据(特别注意这里双工通信,可以实现0端写入1端读取,1端写入0端读取),而且接收数据和对数据处理是在PropertyServiceThread方法中实现:
static void PropertyServiceThread() {
Epoll epoll;
if (auto result = epoll.Open(); !result) {
LOG(FATAL) << result.error();
}
if (auto result = epoll.RegisterHandler(property_set_fd, handle_property_set_fd); !result) {
LOG(FATAL) << result.error();
}
if (auto result = epoll.RegisterHandler(init_socket, HandleInitSocket); !result) {
LOG(FATAL) << result.error();
}
while (true) {
if (auto result = epoll.Wait(std::nullopt); !result) {
LOG(ERROR) << result.error();
}
}
}
这里主要使用了epoll来监听各个fd的变化,这里我们主要关心是epoll.RegisterHandler(property_set_fd, handle_property_set_fd); !result),这个property_set_fd才是我们关心的fd,有数据变化这里会触发handle_property_set_fd方法:
static void handle_property_set_fd() {
//省略
switch (cmd) {
case PROP_MSG_SETPROP: {
//省略
uint32_t result =HandlePropertySet(prop_name, prop_value, source_context, cr, nullptr, &error);
//省略
break;
}
case PROP_MSG_SETPROP2: {
//省略
uint32_t result = HandlePropertySet(name, value, source_context, cr, &socket, &error);
//省略
default:
//省略
break;
}
}
这里最后其实最后调用到了HandlePropertySet:
// This returns one of the enum of PROP_SUCCESS or PROP_ERROR*.
uint32_t HandlePropertySet(const std::string& name, const std::string& value,
const std::string& source_context, const ucred& cr,
SocketConnection* socket, std::string* error) {
//省略
if (StartsWith(name, "ctl.")) {
//开机动画就是符合这个
return SendControlMessage(name.c_str() + 4, value, cr.pid, socket, error);
}
//省略
return PropertySet(name, value, error);
}
代码可以看出开机动画就是符合这个“ctrl.”情况接下来执行SendControlMessage:
static uint32_t SendControlMessage(const std::string& msg, const std::string& name, pid_t pid,SocketConnection* socket, std::string* error) {
//省略
if (auto result = SendMessage(init_socket, property_msg); !result) {
// We've already released the fd above, so if we fail to send the message to init, we need
// to manually free it here.
if (fd != -1) {
close(fd);
}
*error = "Failed to send control message: " + result.error_string();
return PROP_ERROR_HANDLE_CONTROL_MESSAGE;
}
return PROP_SUCCESS;
}
这里其实就是调用SendMessage来发一个消息,但是这里大家注意发给了init_socket这个fd(就是前面说过的0端写入1端读取),那接下来在就会在init.cpp的HandlePropertyFd收到消息,故来分析方法:
static void HandlePropertyFd() {
auto message = ReadMessage(property_fd);
//省略
switch (property_message.msg_case()) {
case PropertyMessage::kControlMessage: {
auto& control_message = property_message.control_message();
//这里是关键,又会调用HandleControlMessage启动及判断是否成功
bool success = HandleControlMessage(control_message.msg(), control_message.name(),
control_message.pid());
uint32_t response = success ? PROP_SUCCESS : PROP_ERROR_HANDLE_CONTROL_MESSAGE;
if (control_message.has_fd()) {
int fd = control_message.fd();
TEMP_FAILURE_RETRY(send(fd, &response, sizeof(response), 0));
close(fd);
}
break;
}
case PropertyMessage::kChangedMessage: {
auto& changed_message = property_message.changed_message();
property_changed(changed_message.name(), changed_message.value());
break;
}
default:
LOG(ERROR) << "Unknown message type from property service: "
<< property_message.msg_case();
}
}
这里最后又调用到HandleControlMessage:
bool HandleControlMessage(const std::string& msg, const std::string& name, pid_t pid) {
//获取一个control msg的map
const auto& map = get_control_message_map();
const auto it = map.find(msg);
//省略
const ControlMessageFunction& function = it->second;//得到对应function
Service* svc = nullptr;
switch (function.target) {
case ControlTarget::SERVICE:
svc = ServiceList::GetInstance().FindService(name);//寻找出对于service
break;
case ControlTarget::INTERFACE:
svc = ServiceList::GetInstance().FindInterface(name);
break;
default:
LOG(ERROR) << "Invalid function target from static map key '" << msg << "': "
<< static_cast<std::underlying_type<ControlTarget>::type>(function.target);
return false;
}
if (svc == nullptr) {
LOG(ERROR) << "Could not find '" << name << "' for ctl." << msg;
return false;
}
if (auto result = function.action(svc); !result) {
//获取service后,会调用function进行执行
LOG(ERROR) << "Could not ctl." << msg << " for '" << name << "': " << result.error();
return false;
}
return true;
}
其实就是根据control msg获取对应的function,然后执行方法,那具体执行什么方法呢?
这里的get_control_message_map如下:
static const std::map<std::string, ControlMessageFunction>& get_control_message_map() {
// clang-format off
static const std::map<std::string, ControlMessageFunction> control_message_functions = {
{
"sigstop_on", {
ControlTarget::SERVICE,
[](auto* service) {
service->set_sigstop(true); return Success(); }}},
{
"sigstop_off", {
ControlTarget::SERVICE,
[](auto* service) {
service->set_sigstop(false); return Success(); }}},
{
"start", {
ControlTarget::SERVICE, DoControlStart}},
{
"stop", {
ControlTarget::SERVICE, DoControlStop}},
{
"restart", {
ControlTarget::SERVICE, DoControlRestart}},
{
"interface_start", {
ControlTarget::INTERFACE, DoControlStart}},
{
"interface_stop", {
ControlTarget::INTERFACE, DoControlStop}},
{
"interface_restart", {
ControlTarget::INTERFACE, DoControlRestart}},
};
// clang-format on
return control_message_functions;
}
所以这里我们bootanimation(“ctrl.start”)明显属于start,所以匹配的方法为DoControlStart:
static Result<Success> DoControlStart(Service* service) {
return service->Start();
}
看到service->Start();后就知道和以前8.1一模一样了,都是service启动,不再进行分析,看对于视频或者blog既可以