I was wondering what happens when an exception is thrown inside a handler function when using boost asio's run() function on the io_context from multiple threads. My thread function which calls the run operation on the io_context looks like this:
while(!io->stopped() && *stop == false) {
try {
auto cnt = io->run();
}catch(std::exception &e) {
}
if(io->stopped()) {
break;
}
}
The number of threads is 1..N. The documentation states that any subsequent calls to run() must call restart() first but restart() must not be called when there are still any active calls to run() which I can't know because there may be still threads calling run().
What is the solution for this when there is just one io_context and many threads calling run()
You're right that you need to catch/handle exceptions.
The solution to your conundrum is quite simple: run() should actually not be called in a loop. This realization helps because it let's you realize that once run(); exited normally, you can always do a break;.
More generically speaking, you could also inspect the return value of run/poll{*} member functions: if it returns zero then the service ran out of work.
See also Should the exception thrown by boost::asio::io_service::run() be caught?
The updated documentation is (mutatis mutandis) identical https://www.boost.org/doc/libs/1_77_0/doc/html/boost_asio/reference/io_service.html#boost_asio.reference.io_service.effect_of_exceptions_thrown_from_handlers.
A normal exit from run requires a call to restart in order for subsequent calls to run to not return immediately. Throwing an exception is not a normal exit so it is fine to call run again immediately.
Your code could just be:
while(true) {
try {
auto cnt = io->run();
break;
}catch(std::exception &e) {
}
}
Related
I have class which has a execute() function. Execution of execute()
function only stops when terminate() function is called. I want to test the execute() function.
class Process{
public:
void execute(){ // start execution until terminate() is called.. }
void terminate(){ //stop the processing of execute()... }
}
My unit test case is given below. I am using MSTest.
TEST_METHOD(StartTest)
{
Process p;
bool isRunning = true;
std::thread th([&](){
p.execute();
isRunning = false;
});
th.detach();
std::this_thread::sleep_for(std::chrono::milliseconds(300));
Assert::isTrue(isRunning);
}
If using thread a good practice should I close the thread inside the test case instead of detaching it from main thread?
Also better suggestion is appreciable.
First of all access to isRunning should be synchronized. In your example you can simply use std::atomic<bool> and be done with it.
Disclaimer: it's been a while since I've done any kind of serios multithreading so take this with a grain of salt. Also, I haven't tested to code, other than to check it compiles.
This is where I would start:
auto test()
{
std::condition_variable cv{};
std::mutex m{};
Process p{};
bool isRunning{true};
std::thread th([&] {
p.execute();
{
std::lock_guard<std::mutex> lk{m};
isRunning = false;
}
cv.notify_one();
});
{
std::unique_lock<std::mutex> lk{m};
// expect timeout
Assert::isFalse(cv.wait_for(lk, std::chrono::milliseconds(300),
[&] () { return !isRunning; }));
}
p.terminate();
{
std::unique_lock<std::mutex> lk{m};
// expect condition to change
Assert::isTrue(cv.wait_for(lk, std::chrono::milliseconds(300),
[&] () { return !isRunning; }));
}
th.join();
}
This way you check both for execute to block and for terminate to terminate and you have more flexibility. If the execute unblocks early you don't wait your full timeout and for terminate you have a wiggle to wait for the other thread to finish and you unblock as soon as it does.
If terminate() fails to stop the execution, will th thread continue
his execution after the end of this test case?
If terminate doesn't stop the execution then the 2nd wait_for ends after timeout returning false and the assert kicks in. I don't know what testing framework you use and what the Assert does.
if it returns the execution to test then the test will block on join until the thread finishes
if it throws an exception then the join is not called and at the destructor of th if the thread has still not ended std::terminate will be called. This can be changed with a try catch
if it forces an exit (e.g. calls std::terminate) then... well... your program ends regardless
This is indeed a problem you need to analyze. It all depends on what you want to do if terminate fails to stop execute within your wait interval.
if you are ok with waiting within test, then all you need to do is make sure join is called. As I've said this is solvable with a try catch.
if you want to end the current test but are ok with the thread still going on then you need to detach the thread if terminate failed to end it.
if you want to kill the thread then... that's not possible. You could instead kill the entire app via std::terminate.
Question:
Is using exceptions the proper way to terminate my program if all I want is to display an error message and close (accounting that I may be deep in the program)? Or can I just explicitly call something like exit(EXIT_FAILURE) instead?
What I'm Currently Doing:
I'm working on a game project and am trying to figure out the best way to terminate the program in the case of an error that calls for such an action. For example, in the case the textures can't be loaded I display an error message and terminate the program.
I'm currently doing this with exceptions like so:
int main()
{
Game game;
try
{
game.run();
}
catch (BadResolutionException & e)
{
Notification::showErrorMessage(e.what(), "ERROR: Resolution");
return 1;
}
catch (BadAssetException & e)
{
Notification::showErrorMessage(e.what(), "ERROR: Assets");
return 1;
}
catch (std::bad_alloc & e)
{
Notification::showErrorMessage(e.what(), "ERROR: Memory");
return 1;
}
return 0;
}
All but bad_alloc are my own defined exceptions derived from runtime_error.
I don't need any manual resource cleanup and I'm using std::unique_ptr for any dynamic allocation. I just need to display the error message and close the program.
Research/Alternatives to Exceptions:
I've looked up a lot of posts on SO and other places and have seen others say anything from don't use exceptions, to use exceptions but your using them wrong. I've also looked up explicitly calling something like exit().
Using exit() sounds nice but I read it won't go back through the call stack up to main cleaning everything up (if I can find this again I'll post the link). Additionally, according to http://www.cplusplus.com/reference/cstdlib/exit/ this should not be used if multiple threads are active. I do expect to be creating a second thread for a short time at least once, and an error could occur in that thread.
Not using exceptions was mentioned in some replies here in relation to games https://gamedev.stackexchange.com/questions/103285/how-industy-games-handle-their-code-errors-and-exceptions
Use exceptions was discussed here: http://www.quora.com/Why-do-some-people-recommend-not-using-exception-handling-in-C++
There are a number of other sources I've read but those were the most recent I looked at.
Personal Conclusion:
Due to my limited experience of working with error handling and using exceptions, I'm not sure if I'm on the right track. I've chosen the route of using exceptions based on the code I posted above. If you agree that I should tackle those cases with exceptions, am I using it correctly?
It's generally considered good practice to let all exceptions propagate through to main. This is primarily because you can be sure the stack is properly unwound and all destructors are called (see this answer). I also think it's more organised to do things this way; you always known where your program will terminate (unless the program crashes). It's also facilitates more consistent error reporting (a point often neglected in exception handling; if you can't handle the exception, you should make sure your user knows exactly why). If you always start with this basic layout
int main(int argc, const char **argv)
{
try {
// do stuff
return EXIT_SUCCESS;
} catch (...) {
std::cerr << "Error: unknown exception" << std::endl;
return EXIT_FAILURE;
}
}
then you won't go far wrong. You can (and should) add specific catch statements for better error reporting.
Exceptions when multithreading
There are two basic ways of executing code asynchronously in C++11 using standard library features: std::async and std::thread.
First the simple one. std::async will return a std::future which will capture and store any uncaught exceptions thrown in the given function. Calling std::future::get on the future will cause any exceptions to propagate into the calling thread.
auto fut = std::async(std::launch::async, [] () { throw std::runtime_error {"oh dear"}; });
fut.get(); // fine, throws exception
On the other hand, if an exception in a std::thread object is uncaught then std::terminate will be called:
try {
std::thread t {[] () { throw std::runtime_error {"oh dear"};}};
t.join();
} catch(...) {
// only get here if std::thread constructor throws
}
One solution to this could be to pass a std::exception_ptr into the std::thread object which it can pass the exception to:
void foo(std::exception_ptr& eptr)
{
try {
throw std::runtime_error {"oh dear"};
} catch (...) {
eptr = std::current_exception();
}
}
void bar()
{
std::exception_ptr eptr {};
std::thread t {foo, std::ref(eptr)};
try {
// do stuff
} catch(...) {
t.join(); // t may also have thrown
throw;
}
t.join();
if (eptr) {
std::rethrow_exception(eptr);
}
}
Although a better way is to use std::package_task:
void foo()
{
throw std::runtime_error {"oh dear"};
}
void bar()
{
std::packaged_task<void()> task {foo};
auto fut = task.get_future();
std::thread t {std::move(task)};
t.join();
auto result = fut.get(); // throws here
}
But unless you have good reason to use std::thread, prefer std::async.
There's nothing wrong with catching unrecoverable errors and shutting down your program this way. In fact, it's how exceptions should be used. However, be careful not to cross the line of using exceptions to control the flow of your program in ordinary circumstances. They should always represent an error which cannot be gracefully handled at the level the error occurred.
Calling exit() would not unwind the stack from wherever you called it. If you want to exit cleanly, what you're already doing is ideal.
You have already accepted an answer, but I wanted to add something about this:
Can I just explicitly call something like exit() instead?
You can call exit, but (probably) shouldn't.
std::exit should be reserved for situations where you want to express "exit right now!", not simply "application has nothing left to do".
As an example, if you were to write a controller for a laser used in cancer treatments, your first priority in case something went wrong would be to shut down the laser and call std::exit - or possibly std::terminate (to ensure any side effects of a hanging, slow or crashing application do not kill a patient).
Similar to how exceptions should not be used for controlling application flow, exit should not be used to stop the application in normal conditions.
Is using exceptions the proper way to terminate my program if all I want is to display an error message and close (accounting that I may be deep in the program)?
Yes. This is the reason for using exceptions. An error occurred deep down in the code and something at a higher level will handle it. In your case, at the highest level.
There are arguments for/against exceptions vs. error codes, and this is a good read:
Exceptions or error codes
Can I just explicitly call something like exit() instead?
You can, but you may end up duplicating your logging code. Also, if in the future you decide that you want to handle an exception differently you will have to change all your exit calls. Imagine you wanted a different message, or to fall back on an alternative process.
Another similar question:
Correct usage of exit() in c++?
You also have a flaw in your approach as you don't handle all (C++) exceptions. You want something like this:
int main()
{
Game game;
try
{
game.run();
}
catch (BadResolutionException & e)
{
Notification::showErrorMessage(e.what(), "ERROR: Resolution");
return 1;
}
catch (BadAssetException & e)
{
Notification::showErrorMessage(e.what(), "ERROR: Assets");
return 1;
}
catch (std::bad_alloc & e)
{
Notification::showErrorMessage(e.what(), "ERROR: Memory");
return 1;
}
catch (...)
{
// overload?
Notification::showErrorMessage("ERROR: Unhandled");
return 1;
}
return 0;
}
If you don't handle all* exceptions you could have the game terminate without telling you anything useful.
You can't handle ALL exceptions. See this link:
C++ catching all exceptions
From the documentation:
[[noreturn]] void exit (int status);
Terminate calling process
Terminates the process normally, performing the regular cleanup for terminating programs.
Normal program termination performs the following (in the same order):
Objects associated with the current thread with thread storage duration are destroyed (C++11 only).
Objects with static storage duration are destroyed (C++) and functions registered with atexit are called.
All C streams (open with functions in ) are closed (and flushed, if buffered), and all files created with tmpfile are removed.
Control is returned to the host environment.
Note that objects with automatic storage are not destroyed by calling exit (C++).
If status is zero or EXIT_SUCCESS, a successful termination status is returned to the host environment.
If status is EXIT_FAILURE, an unsuccessful termination status is returned to the host environment.
Otherwise, the status returned depends on the system and library implementation.
For a similar function that does not perform the cleanup described above, see quick_exit.
I've created a custom ThreadPool which starts a number of win32 threads with _beginthreadex(). The threads are running a simple loop that attempts to dequeue tasks from a blocking queue, but sometimes I need to stop the threads and if they're blocked on Dequeue then I don't know how to get the threads out of that blocking state.
void ThreadPool::Loop()
{
while(_running)
{
try
{
// Attempts to dequeue a task and run it
_taskQueue.Dequeue()->Run();
}
catch(BlockingQueueTerminate&)
{
// Eat the exception and check the running flag
continue;
}
}
}
My idea was to enqueue the same number of special tasks (let's call them "termination tasks") as there are threads in the pool and each "termination task" will call _endthreadex(0) in order to exit the thread. If there are other tasks in the blocking queue, then I won't really care because as soon as I dequeue a task, I will run it and I will check the _running flag to determine if the thread needs to dequeue any more tasks.
void TerminationTask::Run()
{
_endthreadex(0);
}
I have several concerns about this approach; mainly, if I processed a non-terminating task and the _running flag is set to false, then my thread will not call _endthreadex(0) when it exits the loop. I was wondering if I could call _endthreadex(0) at the end of the loop like this:
void ThreadPool::Loop()
{
while(_running)
{
try
{
// Attempts to dequeue a task and run it
_taskQueue.Dequeue()->Run();
}
catch(BlockingQueueTerminate&)
{
// Eat the exception and check the running flag
continue;
}
}
_endthreadex(0);
}
Will this cause a conflict with my TerminationTask or will the thread exit the loop directly after executing TerminationTask::Run() (i.e. it won't call _endthreadex(0) twice)? Furthermore, is there a better approach than this?
Calling _endthreadex(0) at the end of the thread method is fine. It is also optional. If you just leave the thread method normally, then _endthreadex(0) is called for you.
You can call _endthread or _endthreadex explicitly to terminate a thread; however, _endthread or _endthreadex is called automatically when the thread returns from the routine passed as a parameter to _beginthread or _beginthreadex.ref
Sending a termination task is the correct way to get a blocked thread pool thread to unblock and quit.
So, to summarise:
Your strategy is good and the implementation of TerminationTask::Run is correct.
You can remove the harmless call to _endthreadex(0) at the end of ThreadPool::Loop.
Putting a temination task in the Queue is correct. I would try a different approach to handling it, though:
class TerminateThreadNow {};
void TerminationTask::Run()
{
throw TerminateThreadNow();
}
void ThreadPool::Loop()
{
while(_running)
{
try
{
// Attempts to dequeue a task and run it
_taskQueue.Dequeue()->Run();
}
catch(TerminateThreadNow&)
{
_running = false;
}
catch(BlockingQueueTerminate&)
{
// Eat the exception and check the running flag
}
}
}
I am using boost library for threading and synchronization in my application.
First of all I must say exceptions within threads on synchronization is compilitey new thing for me.
In any case below is the pseudo code what I want to achieve. I want synchronized threads to throw same exception that MAY have been thrown from the thread doing notify. How can I achieve this?
Could not find any topics from Stack Overflow regarding exception throwing with cross thread interaction using boost threading model
Many thanks in advance!
// mutex and scondition variable for the problem
mutable boost::mutex conditionMutex;
mutable boost::condition_variable condition;
inline void doTheThing() const {
if (noone doing the thing) {
try {
doIt()
// I succeeded
failed = false;
condition.notify_all();
}
catch (...) {
// I failed to do it
failed = true;
condition.notify_all();
throw
}
else {
boost::mutex::scoped_lock lock(conditionMutex);
condition.wait(lock);
if (failed) {
// throw the same exception that was thrown from
// thread doing notify_all
}
}
}
So you want the first thread that hits doTheThing() to call doIt(), and all subsequent threads that hit doTheThing() to wait for the first thread to finish calling doIt() before they proceed.
I think this should do the trick:
boost::mutex conditionMutex; // mutable qualifier not needed
bool failed = false;
bool done = false;
inline void doTheThing() const {
boost::unique_lock uql(conditionMutex);
if (!done) {
done = true;
try {
doIt();
failed = false;
}
catch (...) {
failed = true;
throw
}
}
else if (failed)
{
uql.unlock();
// now this thread knows that another thread called doIt() and an exception
// was thrown in that thread.
}
}
Important notes:
Every thread that calls doTheThing() must take a lock. There is no way around this. You are synchronizing threads, and for a thread to know anything about what's happening in another thread, it must take a lock. (Or it can use atomic memory operations, but that's a more advanced technique.) The variables failed and done are protected by the conditionMutex.
C++ will call destructor of uql when the function exits normally or by throwing exception.
EDIT Oh, and as for throwing the exception to all the other threads, forget about that, it's almost impossible, and it isn't the way things are done in C++. Instead, each thread can check to see if the first thread successfully called doIt() in the place I've indicated above.
EDIT There is no language support for propagating an exception to another thread. You can generalize the problem of propagating exceptions to another thread to passing messages to another thread. There are lots of library solutions to the problem of passing messages between threads ( boost::asio::io_service::post() ), and you could pass a message that contains the exception, with instructions to throw that exception on receipt of message. It's a bad idea, though. Only throw exceptions when you have an error that prevents you from unwinding the call stack by ordinary function return. That's what an exception is--an alternative way to return from a function when returning the usual way doesn't make sense.
We have a function which a single thread calls into (we name this the main thread). Within the body of the function we spawn multiple worker threads to do CPU intensive work, wait for all threads to finish, then return the result on the main thread.
The result is that the caller can use the function naively, and internally it'll make use of multiple cores.
All good so far..
The problem we have is dealing with exceptions. We don't want exceptions on the worker threads to crash the application. We want the caller to the function to be able to catch them on the main thread. We must catch exceptions on the worker threads and propagate them across to the main thread to have them continue unwinding from there.
How can we do this?
The best I can think of is:
Catch a whole variety of exceptions on our worker threads (std::exception and a few of our own ones).
Record the type and message of the exception.
Have a corresponding switch statement on the main thread which rethrows exceptions of whatever type was recorded on the worker thread.
This has the obvious disadvantage of only supporting a limited set of exception types, and would need modification whenever new exception types were added.
C++11 introduced the exception_ptr type that allows to transport exceptions between threads:
#include<iostream>
#include<thread>
#include<exception>
#include<stdexcept>
static std::exception_ptr teptr = nullptr;
void f()
{
try
{
std::this_thread::sleep_for(std::chrono::seconds(1));
throw std::runtime_error("To be passed between threads");
}
catch(...)
{
teptr = std::current_exception();
}
}
int main(int argc, char **argv)
{
std::thread mythread(f);
mythread.join();
if (teptr) {
try{
std::rethrow_exception(teptr);
}
catch(const std::exception &ex)
{
std::cerr << "Thread exited with exception: " << ex.what() << "\n";
}
}
return 0;
}
Because in your case you have multiple worker threads, you will need to keep one exception_ptr for each of them.
Note that exception_ptr is a shared ptr-like pointer, so you will need to keep at least one exception_ptr pointing to each exception or they will be released.
Microsoft specific: if you use SEH Exceptions (/EHa), the example code will also transport SEH exceptions like access violations, which may not be what you want.
Currently, the only portable way is to write catch clauses for all the types of exceptions that you might like to transfer between threads, store the information somewhere from that catch clause and then use it later to rethrow an exception. This is the approach taken by Boost.Exception.
In C++0x, you will be able to catch an exception with catch(...) and then store it in an instance of std::exception_ptr using std::current_exception(). You can then rethrow it later from the same or a different thread with std::rethrow_exception().
If you are using Microsoft Visual Studio 2005 or later, then the just::thread C++0x thread library supports std::exception_ptr. (Disclaimer: this is my product).
If you're using C++11, then std::future might do exactly what you're looking for: it can automagically trap exceptions that make it to the top of the worker thread, and pass them through to the parent thread at the point that std::future::get is called. (Behind the scenes, this happens exactly as in #AnthonyWilliams' answer; it's just been implemented for you already.)
The down side is that there's no standard way to "stop caring about" a std::future; even its destructor will simply block until the task is done. [EDIT, 2017: The blocking-destructor behavior is a misfeature only of the pseudo-futures returned from std::async, which you should never use anyway. Normal futures don't block in their destructor. But you still can't "cancel" tasks if you're using std::future: the promise-fulfilling task(s) will continue running behind the scenes even if nobody is listening for the answer anymore.] Here's a toy example that might clarify what I mean:
#include <atomic>
#include <chrono>
#include <exception>
#include <future>
#include <thread>
#include <vector>
#include <stdio.h>
bool is_prime(int n)
{
if (n == 1010) {
puts("is_prime(1010) throws an exception");
throw std::logic_error("1010");
}
/* We actually want this loop to run slowly, for demonstration purposes. */
std::this_thread::sleep_for(std::chrono::milliseconds(100));
for (int i=2; i < n; ++i) { if (n % i == 0) return false; }
return (n >= 2);
}
int worker()
{
static std::atomic<int> hundreds(0);
const int start = 100 * hundreds++;
const int end = start + 100;
int sum = 0;
for (int i=start; i < end; ++i) {
if (is_prime(i)) { printf("%d is prime\n", i); sum += i; }
}
return sum;
}
int spawn_workers(int N)
{
std::vector<std::future<int>> waitables;
for (int i=0; i < N; ++i) {
std::future<int> f = std::async(std::launch::async, worker);
waitables.emplace_back(std::move(f));
}
int sum = 0;
for (std::future<int> &f : waitables) {
sum += f.get(); /* may throw an exception */
}
return sum;
/* But watch out! When f.get() throws an exception, we still need
* to unwind the stack, which means destructing "waitables" and each
* of its elements. The destructor of each std::future will block
* as if calling this->wait(). So in fact this may not do what you
* really want. */
}
int main()
{
try {
int sum = spawn_workers(100);
printf("sum is %d\n", sum);
} catch (std::exception &e) {
/* This line will be printed after all the prime-number output. */
printf("Caught %s\n", e.what());
}
}
I just tried to write a work-alike example using std::thread and std::exception_ptr, but something's going wrong with std::exception_ptr (using libc++) so I haven't gotten it to actually work yet. :(
[EDIT, 2017:
int main() {
std::exception_ptr e;
std::thread t1([&e](){
try {
::operator new(-1);
} catch (...) {
e = std::current_exception();
}
});
t1.join();
try {
std::rethrow_exception(e);
} catch (const std::bad_alloc&) {
puts("Success!");
}
}
I have no idea what I was doing wrong in 2013, but I'm sure it was my fault.]
You problem is that you could receive multiple exceptions, from multiple threads, as each could fail, perhaps from different reasons.
I am assuming the main thread is somehow waiting for the threads to end to retrieve the results, or checking regularly the other threads' progress, and that access to shared data is synchronized.
Simple solution
The simple solution would be to catch all exceptions in each thread, record them in a shared variable (in the main thread).
Once all threads finished, decide what to do with the exceptions. This means that all other threads continued their processing, which perhaps, is not what you want.
Complex solution
The more complex solution is have each of your threads check at strategic points of their execution, if an exception was thrown from another thread.
If a thread throws an exception, it is caught before exiting the thread, the exception object is copied into some container in the main thread (as in the simple solution), and some shared boolean variable is set to true.
And when another thread tests this boolean, it sees the execution is to be aborted, and aborts in a graceful way.
When all thread did abort, the main thread can handle the exception as needed.
An exception thrown from a thread will not be catchable in the parent thread. Threads have different contexts and stacks, and generally the parent thread is not required to stay there and wait for the children to finish, so that it could catch their exceptions. There is simply no place in code for that catch:
try
{
start thread();
wait_finish( thread );
}
catch(...)
{
// will catch exceptions generated within start and wait,
// but not from the thread itself
}
You will need to catch exceptions inside each thread and interpret exit status from threads in the main thread to re-throw any exceptions you might need.
BTW, in the absents of a catch in a thread it is implementation specific if stack unwinding will be done at all, i.e. your automatic variables' destructors may not even be called before terminate is called. Some compilers do that, but it's not required.
Could you serialize the exception in the worker thread, transmit that back to the main thread, deserialize, and throw it again? I expect that for this to work the exceptions would all have to derive from the same class (or at least a small set of classes with the switch statement thing again). Also, I'm not sure that they would be serializable, I'm just thinking out loud.
There is, indeed, no good and generic way to transmit exceptions from one thread to the next.
If, as it should, all your exceptions derive from std::exception, then you can have a top-level general exception catch that will somehow send the exception to the main thread where it will be thrown again. The problem being you loose the throwing point of the exception. You can probably write compiler-dependent code to get this information and transmit it though.
If not all your exception inherit std::exception, then you are in trouble and have to write a lot of top-level catch in your thread ... but the solution still hold.
You will need to do a generic catch for all exceptions in the worker (including non-std exceptions, like access violations), and send a message from the worker thread (i suppose you have some kind of messaging in place?) to the controlling thread, containing a live pointer to the exception, and rethrow there by creating a copy of the exception.
Then the worker can free the original object and exit.
See http://www.boost.org/doc/libs/release/libs/exception/doc/tutorial_exception_ptr.html. It is also possible to write a wrapper function of whatever function you call to join a child thread, which automatically re-throws (using boost::rethrow_exception) any exception emitted by a child thread.