Valgrind錯誤：系統調用參數epoll_pwait（sigmask）指向無法尋址的字節（s）

Question

我有以下最小的示例代碼。我想要做的是： 在一個單獨的線程中，啓動一個boost::asio::deadline_timer，以便每調用一個函數loop，就會完成一些操作，定時器會自行重新設置，直到無窮大。

#include <iostream> 
#include <boost/asio.hpp> 
#include <boost/thread/thread.hpp> 

void wait(int milliseconds) { 
    boost::asio::io_service ioService; 
    boost::asio::deadline_timer timer(ioService, boost::posix_time::milliseconds(milliseconds)); 
    timer.wait(); 
} 

class Foo { 
public: 
    boost::asio::io_service ioService; 
    boost::thread* thread = nullptr; 
    boost::posix_time::milliseconds duration; 
    boost::asio::deadline_timer timer; 
    bool shouldStop = false; 

    Foo() : ioService(), duration(10), timer(ioService, duration) { 
    } 

    ~Foo() { 
    if (thread != nullptr) { 
     shouldStop = true; 
     wait(15); 
     ioService.stop(); 
     thread->join(); 
     delete thread; 
     thread = nullptr; 
     } 
    } 

    void start() { 
    timer.async_wait(boost::bind(&Foo::loop, this)); 
    thread = new boost::thread([&](){ioService.run();}); 
    } 

    void loop() { 
    if (shouldStop) 
     return; 
    timer.expires_at(timer.expires_at() + duration); 
    timer.async_wait(boost::bind(&Foo::loop, this)); 
    } 
}; 

int main() { 
    Foo foo; 
    foo.start(); 
    wait(1000); 
    return 0; 
} 

您可以

g++ -std=c++11 -Wall -pedantic -pthread -g -I. -L/usr/local/lib -L/usr/lib -lboost_system -lpthread -lboost_thread threaded-io-service.cpp -o threaded-io-service 

它工作正常編譯它，我敢肯定，前一陣子valgrind沒有不抱怨。我最近更新，並valgrind-3.13.0現在抱怨：

==22212== Memcheck, a memory error detector 
==22212== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al. 
==22212== Using Valgrind-3.13.0 and LibVEX; rerun with -h for copyright info 
==22212== Command: ./threaded-io-service 
==22212== 
==22212== Thread 2: 
==22212== Syscall param epoll_pwait(sigmask) points to unaddressable byte(s) 
==22212== at 0x5E62326: epoll_pwait (in /usr/lib/libc-2.26.so) 
==22212== by 0x111D50: boost::asio::detail::epoll_reactor::run(bool, boost::asio::detail::op_queue<boost::asio::detail::task_io_service_operation>&) (epoll_reactor.ipp:438) 
==22212== by 0x112EE8: boost::asio::detail::task_io_service::do_run_one(boost::asio::detail::scoped_lock<boost::asio::detail::posix_mutex>&, boost::asio::detail::task_io_service_thread_info&, boost::system::error_code const&) (task_io_service.ipp:356) 
==22212== by 0x1129F0: boost::asio::detail::task_io_service::run(boost::system::error_code&) (task_io_service.ipp:149) 
==22212== by 0x11322B: boost::asio::io_service::run() (io_service.ipp:59) 
==22212== by 0x114C9C: Foo::start()::{lambda()#1}::operator()() const (threaded-io-service.cpp:35) 
==22212== by 0x11E8FF: boost::detail::thread_data<Foo::start()::{lambda()#1}>::run() (thread.hpp:116) 
==22212== by 0x526E44E: ??? (in /usr/lib/libboost_thread.so.1.65.1) 
==22212== by 0x5045089: start_thread (in /usr/lib/libpthread-2.26.so) 
==22212== by 0x5E621BE: clone (in /usr/lib/libc-2.26.so) 
==22212== Address 0x0 is not stack'd, malloc'd or (recently) free'd 
==22212== 
==22212== 
==22212== HEAP SUMMARY: 
==22212==  in use at exit: 0 bytes in 0 blocks 
==22212== total heap usage: 31 allocs, 31 frees, 81,701 bytes allocated 
==22212== 
==22212== All heap blocks were freed -- no leaks are possible 
==22212== 
==22212== For counts of detected and suppressed errors, rerun with: -v 
==22212== ERROR SUMMARY: 112 errors from 1 contexts (suppressed: 0 from 0) 

這是一個真正的錯誤？我發現this bug report人們認爲它可能是valgrind中的一個錯誤。

Answer 1

是的，這確實是一個錯誤，這是固定的上游（Valgrind的SVN r16451，2017年6月17日），經由鏈接找到您發佈的評論：

https://bugs.kde.org/show_bug.cgi?id=381289

Mark Wielaard 2017-06-16 15:44:30 UTC Originally reported against Fedora: https://bugzilla.redhat.com/show_bug.cgi?id=1462258

According to the epoll_pwait(2) man page:

The sigmask argument may be specified as NULL, in which case 
    epoll_pwait() is equivalent to epoll_wait(). 

But doing that under valgrind gives:

==13887== Syscall param epoll_pwait(sigmask) points to unaddressable byte(s) 
==13887== at 0x4F2B940: epoll_pwait (epoll_pwait.c:43) 
==13887== by 0x400ADE: main (syscalls-2007.c:89) 
==13887== Address 0x0 is not stack'd, malloc'd or (recently) free'd 

This is because the sys_epoll_pwait wrapper has:

if (ARG4) 
     PRE_MEM_READ("epoll_pwait(sigmask)", ARG5, sizeof(vki_sigset_t)); 

Which looks like a typo (ARGS4 is timeout and ARG5 is sigmask).