spdlog segfault

[m4ce]

I recently upgraded from spdlog 1.3.1 to 1.4.2. It seems that since then my applications crash on shutdown leaving a core file. It seems it's trying to log but the sink is gone?

Would anyone know why this happens?

Program terminated with signal SIGSEGV, Segmentation fault.
#0  0x00007f09fb77d837 in spdlog::async_logger::backend_sink_it_ (this=0x2799410, msg=...) at /usr/local/include/spdlog/async_logger-inl.h:63
63                      sink->log(msg);
[Current thread is 1 (Thread 0x7f09d1a07700 (LWP 58))]
(gdb)
(gdb) bt
#0  0x00007f09fb77d837 in spdlog::async_logger::backend_sink_it_ (this=0x2799410, msg=...) at /usr/local/include/spdlog/async_logger-inl.h:63
#1  0x00007f09fb77d6b9 in spdlog::details::thread_pool::process_next_msg_ (this=0x230d9d0) at /usr/local/include/spdlog/details/thread_pool-inl.h:103
#2  0x00007f09fb77d57a in spdlog::details::thread_pool::worker_loop_ (this=0x230d9d0) at /usr/local/include/spdlog/details/thread_pool-inl.h:84
#3  0x00007f09fb77d245 in spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<void ()>)::{lambda()#1}::operator()() const (__closure=0x2a026b8) at /usr/local/include/spdlog/details/thread_pool-inl.h:27
#4  0x00007f09fb7857c3 in std::__invoke_impl<void, spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<void ()>)::{lambda()#1}>(std::__invoke_other, spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<voi
d ()>)::{lambda()#1}&&) (__f=...) at /usr/local/include/c++/8.3.0/bits/invoke.h:60
#5  0x00007f09fb78416d in std::__invoke<spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<void ()>)::{lambda()#1}>(std::__invoke_result&&, (spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<void ()>):
:{lambda()#1}&&)...) (__fn=...) at /usr/local/include/c++/8.3.0/bits/invoke.h:95

[zh1an]

Do you call spdlog::shutdown(), when your application dead? | | 兰剑 | | 邮箱：lan_0032@163.com | On 11/28/2019 20:01, Matteo Cerutti wrote: I recently upgraded from spdlog 1.3.1 to 1.4.2. It seems that since then my applications crashes on shutdown leaving a core file. It seems it's trying to log but the sink is gone? Would anyone know why this happens? Program terminated with signal SIGSEGV, Segmentation fault. #0 0x00007f09fb77d837 in spdlog::async_logger::backend_sink_it_ (this=0x2799410, msg=...) at /usr/local/include/spdlog/async_logger-inl.h:63 63 sink->log(msg); [Current thread is 1 (Thread 0x7f09d1a07700 (LWP 58))] (gdb) (gdb) bt #0 0x00007f09fb77d837 in spdlog::async_logger::backend_sink_it_ (this=0x2799410, msg=...) at /usr/local/include/spdlog/async_logger-inl.h:63 #1 0x00007f09fb77d6b9 in spdlog::details::thread_pool::process_next_msg_ (this=0x230d9d0) at /usr/local/include/spdlog/details/thread_pool-inl.h:103 #2 0x00007f09fb77d57a in spdlog::details::thread_pool::worker_loop_ (this=0x230d9d0) at /usr/local/include/spdlog/details/thread_pool-inl.h:84 #3 0x00007f09fb77d245 in spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<void ()>)::{lambda()#1}::operator()() const (__closure=0x2a026b8) at /usr/local/include/spdlog/details/thread_pool-inl.h:27 #4 0x00007f09fb7857c3 in std::__invoke_impl<void, spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<void ()>)::{lambda()#1}>(std::__invoke_other, spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<voi d ()>)::{lambda()#1}&&) (__f=...) at /usr/local/include/c++/8.3.0/bits/invoke.h:60 #5 0x00007f09fb78416d in std::__invoke<spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<void ()>)::{lambda()#1}>(std::__invoke_result&&, (spdlog::details::thread_pool::thread_pool(unsigned long, unsigned long, std::function<void ()>): :{lambda()#1}&&)...) (__fn=...) at /usr/local/include/c++/8.3.0/bits/invoke.h:95 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[m4ce]

No - I don't do that. Should I? I see it's only an issue on windows (https://github.com/gabime/spdlog/wiki/6.-Asynchronous-logging) that leads to a deadlock. But here it's really a segfault..

[zh1an]

Sorry, I didn't notice gcc compiler.

[zh1an]

Can you provide your test code?

[m4ce]

@zh1an, eventually I found the issue. It has nothing to do with spdlog. I was clearing a buffer on shutdown but boost asio was still holding on to that and moving it. The thing is that the error was too generic and misleading. At some point, I even got errors like:

Fatal error: glibc detected an invalid stdio handle
Segmentation fault

or

Program terminated with signal SIGSEGV, Segmentation fault.
#0  __memmove_avx_unaligned_erms () at ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:499
499     ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S: No such file or directory.
[Current thread is 1 (Thread 0x7fcc15535740 (LWP 629))]
(gdb) bt
#0  __memmove_avx_unaligned_erms () at ../sysdeps/x86_64/multiarch/memmove-vec-unaligned-erms.S:499

So that led me to think that it wasn't a spdlog issue..