實(shí)現(xiàn)一個(gè)jstack
在聊Jstack得工作原理前呢展蒂，不如讓我們先寫一個(gè)簡(jiǎn)單的jstack玩玩呢簸。不用怕，很簡(jiǎn)單的,就幾行代碼的事漓概，看：
public class MyJstack {
public static void main(String[] args)throws Exception {
VirtualMachine virtualMachine = VirtualMachine.attach("6361");
HotSpotVirtualMachine hotSpotVirtualMachine = (HotSpotVirtualMachine)virtualMachine;
InputStream inputStream = hotSpotVirtualMachine.remoteDataDump(new String[]{});
byte[] buff = new byte[256];
int len;
do {
len = inputStream.read(buff);
if (len > 0) {
String respone = new String(buff, 0, len, "UTF-8");
System.out.print(respone);
}
} while(len > 0);
inputStream.close();
virtualMachine.detach();
}
}
很簡(jiǎn)單吧漾月，貼到你的開發(fā)環(huán)境里，運(yùn)行就好了胃珍，別忘了把6361這個(gè)進(jìn)程號(hào)換成你自己的Java進(jìn)程號(hào)哦梁肿。
實(shí)現(xiàn)原理
jstack有兩種實(shí)現(xiàn)方式，一種是基于attach api,其實(shí)現(xiàn)可以在tools.jar里找到觅彰；另一種是基于SA的實(shí)現(xiàn)吩蔑，它被放在了sa-jdi.jar里。如果你通過(guò)idea搜索Jstack類缔莲，你會(huì)看到tools.jar和sa-jdi.jar各有一個(gè)Jstack類哥纫。
本文呢峰伙，就通過(guò)分析attch api的源碼划址，來(lái)了解jstack的工作原理。
jstack本地源碼實(shí)現(xiàn)
我們來(lái)看一下HotSpotVirtualMachine的remoteDataDump方法：
public InputStream remoteDataDump(Object... var1) throws IOException {
return this.executeCommand("threaddump", var1);
}
他是在執(zhí)行一個(gè)叫threaddump的命令荧呐。沿著這個(gè)executeCommand方法繼續(xù)往里追读拆，會(huì)發(fā)現(xiàn)他是調(diào)用了如下方法：
InputStream execute(String var1, Object... var2) throws AgentLoadException, IOException {
assert var2.length <= 3;
String var3;
synchronized(this) {
if (this.path == null) {
throw new IOException("Detached from target VM");
}
var3 = this.path;
}
int var4 = socket();
try {
connect(var4, var3);
} catch (IOException var9) {
close(var4);
throw var9;
}
IOException var5 = null;
try {
this.writeString(var4, "1");
this.writeString(var4, var1);
var1參數(shù)就是我們的threaddump指令擅憔，不難看出，這個(gè)方法是建立了一個(gè)socket連接檐晕，然后將threaddump指令發(fā)送給另一端暑诸，即我們要檢查的jvm進(jìn)程蚌讼。
注意：限于篇幅我并沒有貼整個(gè)方法代碼。execute是HotSpotVirtualMachine的抽象方法个榕，不同平臺(tái)的jdk有不同的execute方法的實(shí)現(xiàn)篡石，我這里的代碼是mac下的execute實(shí)現(xiàn)，位于BsdVirtualMachine類中西采。
通過(guò)jtack本地源代碼凰萨，我們大致可以粗略的認(rèn)為：jstack就是通過(guò)與指定的jvm進(jìn)程建立socket連接，然后發(fā)送指令械馆，最后將jvm進(jìn)程返回的內(nèi)容打印出來(lái)胖眷。
JVM的源碼實(shí)現(xiàn)
了解了jstack的本地源碼，我們?cè)诳纯磈vm進(jìn)程是如何處理的霹崎。
當(dāng)我們使用Java命令啟動(dòng)jvm進(jìn)程時(shí)珊搀，Java命令會(huì)加載虛擬機(jī)共享庫(kù)，然后執(zhí)行共享庫(kù)里的JNI_CreateJavaVM方法完成虛擬機(jī)的創(chuàng)建尾菇，在JNI_CreateJavaVM方法里會(huì)調(diào)用如下代碼境析，完成具體的一個(gè)創(chuàng)建過(guò)程：
result = Threads::create_vm((JavaVMInitArgs*) args, &can_try_again);
如果你有心，或許會(huì)留意到派诬，在你啟動(dòng)一個(gè)jvm進(jìn)程時(shí)簿晓，即便你什么線程也沒創(chuàng)建，你用jstack查看還是有很多的線程千埃，如：Signal Dispatcher，VM Thread忆植，Attach Listener等等放可。當(dāng)過(guò)閱讀本文，你會(huì)了解到這三個(gè)線程的作用朝刊。
01 VM Thread線程
Threads::create_vm這個(gè)方法很長(zhǎng)耀里，接下來(lái)咱們跳出一些重要的段落，來(lái)分析分析拾氓。
// Create the VMThread
{ TraceTime timer("Start VMThread", TraceStartupTime);
VMThread::create();//創(chuàng)建Thread對(duì)象
Thread* vmthread = VMThread::vm_thread();
if (!os::create_thread(vmthread, os::vm_thread))//調(diào)用操作系統(tǒng)api創(chuàng)建線程
vm_exit_during_initialization("Cannot create VM thread. Out of system resources.");
// Wait for the VM thread to become ready, and VMThread::run to initialize
// Monitors can have spurious returns, must always check another state flag
{
MutexLocker ml(Notify_lock);
os::start_thread(vmthread);//啟動(dòng)線程
while (vmthread->active_handles() == NULL) {
Notify_lock->wait();
}
}
}
通過(guò)注釋冯挎，你也知道，這一段代碼是從來(lái)創(chuàng)建VM Thread線程的咙鞍。VMThread::create()完成了對(duì)現(xiàn)成的命名工作房官，代碼如下：
void VMThread::create() {
assert(vm_thread() == NULL, "we can only allocate one VMThread");
_vm_thread = new VMThread();
// Create VM operation queue
_vm_queue = new VMOperationQueue();
guarantee(_vm_queue != NULL, "just checking");
_terminate_lock = new Monitor(Mutex::safepoint, "VMThread::_terminate_lock", true);
if (UsePerfData) {
// jvmstat performance counters
Thread* THREAD = Thread::current();
_perf_accumulated_vm_operation_time =
PerfDataManager::create_counter(SUN_THREADS, "vmOperationTime",
PerfData::U_Ticks, CHECK);
}
}
VMThread::VMThread() : NamedThread() {
set_name("VM Thread");
}
通過(guò)new VMThread()創(chuàng)建線程對(duì)象，在VMThread的構(gòu)造方法里將線程命名成VM Thread续滋，這就是我們jstack看到的VM Thread線程,同時(shí)還為這個(gè)線程創(chuàng)建了一個(gè)叫VMOperationQueue的隊(duì)列翰守。
至于VM Thread線程的作用，我們留到最后再說(shuō)疲酌。
02 Signal Dispatcher線程
繼續(xù)沿著 Threads::create_vm方法往下看蜡峰，我們會(huì)看到如下代碼：
// Signal Dispatcher needs to be started before VMInit event is posted
os::signal_init();
這一句代碼實(shí)現(xiàn)了Signal Dispatcher線程的創(chuàng)建了袁，進(jìn)入到signal_init()方法看看:
void os::signal_init() {
if (!ReduceSignalUsage) {
// Setup JavaThread for processing signals
EXCEPTION_MARK;
Klass* k = SystemDictionary::resolve_or_fail(vmSymbols::java_lang_Thread(), true, CHECK);
instanceKlassHandle klass (THREAD, k);
instanceHandle thread_oop = klass->allocate_instance_handle(CHECK);
const char thread_name[] = "Signal Dispatcher";
Handle string = java_lang_String::create_from_str(thread_name, CHECK);
// Initialize thread_oop to put it into the system threadGroup
Handle thread_group (THREAD, Universe::system_thread_group());
JavaValue result(T_VOID);
JavaCalls::call_special(&result, thread_oop,
klass,
vmSymbols::object_initializer_name(),
vmSymbols::threadgroup_string_void_signature(),
thread_group,
string,
CHECK);
KlassHandle group(THREAD, SystemDictionary::ThreadGroup_klass());
JavaCalls::call_special(&result,
thread_group,
group,
vmSymbols::add_method_name(),
vmSymbols::thread_void_signature(),
thread_oop, // ARG 1
CHECK);
os::signal_init_pd();
{ MutexLocker mu(Threads_lock);
JavaThread* signal_thread = new JavaThread(&signal_thread_entry);
// At this point it may be possible that no osthread was created for the
// JavaThread due to lack of memory. We would have to throw an exception
// in that case. However, since this must work and we do not allow
// exceptions anyway, check and abort if this fails.
if (signal_thread == NULL || signal_thread->osthread() == NULL) {
vm_exit_during_initialization("java.lang.OutOfMemoryError",
"unable to create new native thread");
}
java_lang_Thread::set_thread(thread_oop(), signal_thread);
java_lang_Thread::set_priority(thread_oop(), NearMaxPriority);
java_lang_Thread::set_daemon(thread_oop());
signal_thread->set_threadObj(thread_oop());
Threads::add(signal_thread);
Thread::start(signal_thread);
}
// Handle ^BREAK
os::signal(SIGBREAK, os::user_handler());
}
}
在這個(gè)方法里，我們可以看到要?jiǎng)?chuàng)建的線程名字:Signal Dispatcher湿颅，以及線程啟動(dòng)后調(diào)用的方法signal_thread_entry载绿。（方法較長(zhǎng)，看重點(diǎn)就好油航，沒必要每句話都扣清楚）崭庸。
有了對(duì)上邊代碼的分析，我們只需要看看signal_thread_entry方法劝堪，就知道Signal Dispatcher線程的作用了冀自。
static void signal_thread_entry(JavaThread* thread, TRAPS) {
os::set_priority(thread, NearMaxPriority);
while (true) {
int sig;
{
// FIXME : Currently we have not decieded what should be the status
// for this java thread blocked here. Once we decide about
// that we should fix this.
sig = os::signal_wait();//等待獲取信號(hào)
}
if (sig == os::sigexitnum_pd()) {
// Terminate the signal thread
return;
}
switch (sig) {
case SIGBREAK: {
// Check if the signal is a trigger to start the Attach Listener - in that
// case don't print stack traces.
if (!DisableAttachMechanism && AttachListener::is_init_trigger()) {
continue;
}
// Print stack traces
// Any SIGBREAK operations added here should make sure to flush
// the output stream (e.g. tty->flush()) after output. See 4803766.
// Each module also prints an extra carriage return after its output.
VM_PrintThreads op;
VMThread::execute(&op);
VM_PrintJNI jni_op;
VMThread::execute(&jni_op);
VM_FindDeadlocks op1(tty);
VMThread::execute(&op1);
Universe::print_heap_at_SIGBREAK();
if (PrintClassHistogram) {
VM_GC_HeapInspection op1(gclog_or_tty, true /* force full GC before heap inspection */);
VMThread::execute(&op1);
}
if (JvmtiExport::should_post_data_dump()) {
JvmtiExport::post_data_dump();
}
break;
這個(gè)方法里調(diào)用os::signal_wait()獲取傳給該jvm進(jìn)程的信號(hào)，然后對(duì)信號(hào)進(jìn)行處理秒啦。
說(shuō)下case SIGBREAK里的處理邏輯熬粗，當(dāng)接收到SIGBREAK信號(hào)時(shí)，會(huì)先判斷是否禁止Attach機(jī)制余境，如果沒有禁止驻呐，會(huì)調(diào)用AttachListener::is_init_trigger()方法觸發(fā)Attach Listener線程的初始化.如果attach機(jī)制被禁用，則會(huì)創(chuàng)建VM_PrintThreads芳来、VM_PrintJNI含末、VM_FindDeadlocks等代表某一個(gè)操作的對(duì)象，通過(guò)VMThread::execute()方法扔到VM Thread線程的VMOperationQueue隊(duì)列即舌。
03 Attach Listener線程
繼續(xù)沿著 Threads::create_vm方法往下看佣盒，在緊挨著啟動(dòng)Signal Dispatcher線程的下邊，就是啟動(dòng)Attach Listener線程的語(yǔ)句：
// Start Attach Listener if +StartAttachListener or it can't be started lazily
if (!DisableAttachMechanism) {
AttachListener::vm_start();
if (StartAttachListener || AttachListener::init_at_startup()) {
AttachListener::init();
}
}
重點(diǎn)就在AttachListener::init()方法里：
// Starts the Attach Listener thread
void AttachListener::init() {
EXCEPTION_MARK;
Klass* k = SystemDictionary::resolve_or_fail(vmSymbols::java_lang_Thread(), true, CHECK);
instanceKlassHandle klass (THREAD, k);
instanceHandle thread_oop = klass->allocate_instance_handle(CHECK);
const char thread_name[] = "Attach Listener";
Handle string = java_lang_String::create_from_str(thread_name, CHECK);
// Initialize thread_oop to put it into the system threadGroup
Handle thread_group (THREAD, Universe::system_thread_group());
JavaValue result(T_VOID);
JavaCalls::call_special(&result, thread_oop,
klass,
vmSymbols::object_initializer_name(),
vmSymbols::threadgroup_string_void_signature(),
thread_group,
string,
THREAD);
if (HAS_PENDING_EXCEPTION) {
tty->print_cr("Exception in VM (AttachListener::init) : ");
java_lang_Throwable::print(PENDING_EXCEPTION, tty);
tty->cr();
CLEAR_PENDING_EXCEPTION;
return;
}
KlassHandle group(THREAD, SystemDictionary::ThreadGroup_klass());
JavaCalls::call_special(&result,
thread_group,
group,
vmSymbols::add_method_name(),
vmSymbols::thread_void_signature(),
thread_oop, // ARG 1
THREAD);
if (HAS_PENDING_EXCEPTION) {
tty->print_cr("Exception in VM (AttachListener::init) : ");
java_lang_Throwable::print(PENDING_EXCEPTION, tty);
tty->cr();
CLEAR_PENDING_EXCEPTION;
return;
}
{ MutexLocker mu(Threads_lock);
JavaThread* listener_thread = new JavaThread(&attach_listener_thread_entry);
// Check that thread and osthread were created
if (listener_thread == NULL || listener_thread->osthread() == NULL) {
vm_exit_during_initialization("java.lang.OutOfMemoryError",
"unable to create new native thread");
}
java_lang_Thread::set_thread(thread_oop(), listener_thread);
java_lang_Thread::set_daemon(thread_oop());
listener_thread->set_threadObj(thread_oop());
Threads::add(listener_thread);
Thread::start(listener_thread);
}
}
我們可以通過(guò)代碼看出其創(chuàng)建了一個(gè)叫Attach Listener的線程顽聂，線程執(zhí)行的邏輯封裝在了attach_listener_thread_entry方法里肥惭。
Attach Listener線程的作用，我們看看attach_listener_thread_entry方法便知:
static void attach_listener_thread_entry(JavaThread* thread, TRAPS) {
os::set_priority(thread, NearMaxPriority);
thread->record_stack_base_and_size();
if (AttachListener::pd_init() != 0) {
return;
}
AttachListener::set_initialized();
for (;;) {
AttachOperation* op = AttachListener::dequeue();//從隊(duì)列里獲取操作對(duì)象
if (op == NULL) {
return; // dequeue failed or shutdown
}
ResourceMark rm;
bufferedStream st;
jint res = JNI_OK;
// handle special detachall operation
if (strcmp(op->name(), AttachOperation::detachall_operation_name()) == 0) {
AttachListener::detachall();
} else {
// find the function to dispatch too
AttachOperationFunctionInfo* info = NULL;
for (int i=0; funcs[i].name != NULL; i++) {
const char* name = funcs[i].name;
assert(strlen(name) <= AttachOperation::name_length_max, "operation <= name_length_max");
if (strcmp(op->name(), name) == 0) {
info = &(funcs[i]);
break;
}
}
// check for platform dependent attach operation
if (info == NULL) {
info = AttachListener::pd_find_operation(op->name());
}
if (info != NULL) {
// dispatch to the function that implements this operation
res = (info->func)(op, &st);//執(zhí)行操作對(duì)象
} else {
st.print("Operation %s not recognized!", op->name());
res = JNI_ERR;
}
}
// operation complete - send result and output to client
op->complete(res, &st);
}
}
方法很長(zhǎng)紊搪，我把重點(diǎn)挑出來(lái)分析蜜葱。
首先我們看看調(diào)用AttachListener::pd_init()完了什么：
int AttachListener::pd_init() {
JavaThread* thread = JavaThread::current();
ThreadBlockInVM tbivm(thread);
thread->set_suspend_equivalent();
// cleared by handle_special_suspend_equivalent_condition() or
// java_suspend_self() via check_and_wait_while_suspended()
int ret_code = LinuxAttachListener::init();
// were we externally suspended while we were waiting?
thread->check_and_wait_while_suspended();
return ret_code;
}
int LinuxAttachListener::init() {
char path[UNIX_PATH_MAX]; // socket file
char initial_path[UNIX_PATH_MAX]; // socket file during setup
int listener; // listener socket (file descriptor)
// register function to cleanup
::atexit(listener_cleanup);
int n = snprintf(path, UNIX_PATH_MAX, "%s/.java_pid%d",
os::get_temp_directory(), os::current_process_id());
if (n < (int)UNIX_PATH_MAX) {
n = snprintf(initial_path, UNIX_PATH_MAX, "%s.tmp", path);
}
if (n >= (int)UNIX_PATH_MAX) {
return -1;
}
// create the listener socket
listener = ::socket(PF_UNIX, SOCK_STREAM, 0);//創(chuàng)建套接字
if (listener == -1) {
return -1;
}
// bind socket
struct sockaddr_un addr;
addr.sun_family = AF_UNIX;
strcpy(addr.sun_path, initial_path);
::unlink(initial_path);
int res = ::bind(listener, (struct sockaddr*)&addr, sizeof(addr));//綁定地址
if (res == -1) {
::close(listener);
return -1;
}
// put in listen mode, set permissions, and rename into place
res = ::listen(listener, 5);//發(fā)起監(jiān)聽
if (res == 0) {
RESTARTABLE(::chmod(initial_path, S_IREAD|S_IWRITE), res);
if (res == 0) {
res = ::rename(initial_path, path);
}
}
if (res == -1) {
::close(listener);
::unlink(initial_path);
return -1;
}
set_path(path);
set_listener(listener);
return 0;
}
不難發(fā)現(xiàn)，AttachListener::pd_init()方法又調(diào)用了LinuxAttachListener::init()方法耀石，完成了對(duì)套接字的創(chuàng)建和監(jiān)聽牵囤。這與jstack本地代碼建立socket連接發(fā)送命令，不謀而合滞伟。
再就是有一個(gè)for死循環(huán)揭鳞，不停地調(diào)用AttachOperation* op = AttachListener::dequeue();獲取操作對(duì)象。如果進(jìn)入到AttachListener::dequeue()方法看一看诗良，其實(shí)就是在讀上邊監(jiān)聽的套接字汹桦，我這里就不貼源碼了。
在這個(gè)死循環(huán)里鉴裹，我們重點(diǎn)看看如下代碼:
// find the function to dispatch too
AttachOperationFunctionInfo* info = NULL;
for (int i=0; funcs[i].name != NULL; i++) {
const char* name = funcs[i].name;
assert(strlen(name) <= AttachOperation::name_length_max, "operation <= name_length_max");
if (strcmp(op->name(), name) == 0) {
info = &(funcs[i]);
break;
}
}
// check for platform dependent attach operation
if (info == NULL) {
info = AttachListener::pd_find_operation(op->name());
}
if (info != NULL) {
// dispatch to the function that implements this operation
res = (info->func)(op, &st);//調(diào)動(dòng)方法
} else {
st.print("Operation %s not recognized!", op->name());
res = JNI_ERR;
}
}
// operation complete - send result and output to client
op->complete(res, &st);
這個(gè)for循環(huán)會(huì)遍歷funcs數(shù)組舞骆，然后根據(jù)從隊(duì)列里拿到的AttachOperation對(duì)象的name來(lái)找到一個(gè)匹配的AttachOperationFunctionInfo對(duì)象钥弯，然后調(diào)用其func方法。
看到這里你或許很多疑惑督禽，當(dāng)然看看funcs數(shù)組里的東西脆霎，就開朗了：
static AttachOperationFunctionInfo funcs[] = {
{ "agentProperties", get_agent_properties },
{ "datadump", data_dump },
{ "dumpheap", dump_heap },
{ "load", JvmtiExport::load_agent_library },
{ "properties", get_system_properties },
{ "threaddump", thread_dump },
{ "inspectheap", heap_inspection },
{ "setflag", set_flag },
{ "printflag", print_flag },
{ "jcmd", jcmd },
{ NULL, NULL }
};
有沒有看到上文中我們提到的threaddump命令。jstack通過(guò)與jvm進(jìn)程建立socket連接狈惫，然后向jvm進(jìn)程發(fā)送threaddump指令睛蛛。上文說(shuō)道調(diào)用AttachOperationFunctionInfo對(duì)象的func方法處理指令，其實(shí)就是調(diào)用了thread_dump方法胧谈，針對(duì)threaddump命令來(lái)說(shuō)忆肾。
堅(jiān)持，馬上就要說(shuō)完了菱肖。來(lái)看看thread_dump方法干了些啥吧:
// Implementation of "threaddump" command - essentially a remote ctrl-break
// See also: ThreadDumpDCmd class
//
static jint thread_dump(AttachOperation* op, outputStream* out) {
bool print_concurrent_locks = false;
if (op->arg(0) != NULL && strcmp(op->arg(0), "-l") == 0) {
print_concurrent_locks = true;
}
// thread stacks
VM_PrintThreads op1(out, print_concurrent_locks);
VMThread::execute(&op1);
// JNI global handles
VM_PrintJNI op2(out);
VMThread::execute(&op2);
// Deadlock detection
VM_FindDeadlocks op3(out);
VMThread::execute(&op3);
return JNI_OK;
}
很簡(jiǎn)單客冈，創(chuàng)建了VM_PrintThreads、VM_PrintJNI稳强、VM_FindDeadlocks三個(gè)對(duì)象场仲，扔給了VM Thread線程的隊(duì)列。
說(shuō)到這里退疫，VM Thread線程的作用渠缕，應(yīng)該真相大白了，就是讀取隊(duì)列褒繁，然后執(zhí)行相應(yīng)的操作亦鳞。有興趣你可以繼續(xù)追進(jìn)去看看源代碼，我這里就不追下去了棒坏。
總結(jié)
看了這么多代碼蚜迅，確實(shí)很頭疼，總結(jié)下吧俊抵。
jstack是通過(guò)與jvm進(jìn)程建立socket連接，然后發(fā)送指令來(lái)實(shí)現(xiàn)相關(guān)操作坐梯。
jvm的Attach Listener線程監(jiān)聽套接字徽诲，讀取jstack發(fā)來(lái)的指令，然后將相關(guān)的操作扔給VM Thread線程來(lái)執(zhí)行吵血，最后返回給jstack谎替。
在jvm啟動(dòng)的時(shí)候，如果沒有指定StartAttachListener蹋辅，Attach Listener線程是不會(huì)啟動(dòng)的钱贯，在Signal Dispatcher線程收到SIGBREAK信號(hào)時(shí)，會(huì)調(diào)用 AttachListener::is_init_trigger（）通過(guò)調(diào)用用AttachListener::init（）啟動(dòng)了Attach Listener 線程侦另。
