Re: [Boost-users] [stacktrace] Getting a stacktrace from unhandled exception in windows

-------- Original Message -------- Subject: Boost-users Digest, Vol 4770, Issue 1 From: boost-users-request@lists.boost.org To: boost-users@lists.boost.org Date: 9/10/2017 10:56 PM

Subject: Re: [Boost-users] [stacktrace] Getting a stacktrace from unhandled exception in windows From: Stian Zeljko Vrba Date: 9/10/2017 10:54 PM

Unless you need the stacktrace to be logged to a text file, I suggest you look at this system-provided facility: https://msdn.microsoft.com/en-us/library/windows/desktop/bb787181(v=vs.85).a... https://msdn.microsoft.com/en-us/library/windows/desktop/bb787181%28v=vs.85%...

...> IOW, you can get out more information than with boost.stacktrace with zero coding.

...

*Sent:* Monday, September 11, 2017 1:27:47 AM *To:* boost-users@lists.boost.org *Subject:* [Boost-users] [stacktrace] Getting a stacktrace from unhandled exception in windows

...

...

I am trying use boost stacktrace to get a stack dump when my program encounters an unhandled exception under windows. I have installed a signal handler as described in the stacktrace documentation, and it gets called. However, the resulting stack trace is after everything has been unwound.

For example, when I force a failure by dereferenceing a null pointer many levels down in my function hierarchy, the stack trace shows:

   0# boost::stacktrace::detail::this_thread_frames::safe_dump_to_impl at f:\home\tools\boost\boost\stacktrace\safe_dump_to.hpp:68    1# boost::stacktrace::safe_dump_to at f:\home\tools\boost\boost\stacktrace\safe_dump_to.hpp:120    2# my_signal_handler at f:\home\test\main.cpp:233    3# 0x5171407B    4# my_signal_handler at f:\home\test\main.cpp:235    5# 0x51713883    6# __tmainCRTStartup at f:\dd\vctools\crt_bld\self_x86\crt\src\crtexe.c:603    7# WinMainCRTStartup at f:\dd\vctools\crt_bld\self_x86\crt\src\crtexe.c:403    8# BaseThreadInitThunk in kernel32    9# RtlInitializeExceptionChain in ntdll 10# RtlInitializeExceptionChain in ntdll

This is not particularly helpful, of course -- I want to know the call stack at the point of the failure. Is there any way to get that? What am I missing?

Thanks for the suggestion, but yes, I want to log the data to a log file. It seems from the stacktrace documentation that this should work. Don