bombela / backward-cpp Goto Github PK

A chance this library will be split in a client that grabs a backtrace, that could be stored and sent to a server and be symbolized there?

This way you can run a striped binary, but before you strip you copy the debug information and store it on the back-end. When you get a crash, you do the symbolizing on the back-end using your stored debug data.

When including the BackwardConfig.cmake via find_package:

find_package(backward NO_MODULE REQUIRED)

it fails to properly detect backward if neither libdw nor libbfd are found as in this case BACKWARD_LIBRARIES is empty, hence the failure in find_package_handle_standard_args in https://github.com/bombela/backward-cpp/blob/master/BackwardConfig.cmake#L121 . Given that both libraries are optional, the failure should not be a hard error (even with REQUIRED)

SUMMARY

One of the ways of using backward-cpp is linking to the library with an imported target in CMake. However, it doesn't register any signal handlers in case program crashes.

The probable cause is that - according to my understanding - in C++ the moment of instantiation of a global variable in a different translation unit does not have to happen before the main() function is called and a signal is raised. In that case it's strange to suggest using it like that.

Also note that using backward-cpp with the add_subdirectory method prints stack traces correctly.

STEPS TO REPRODUCE

Download backward-cpp:

mkdir external
cd external
git clone https://github.com/bombela/backward-cpp.git

Create CMakeLists.txt in the main project folder:

cmake_minimum_required(VERSION 3.0)
project(backward-cpp-test)

#set(BACWARD_SHARED ON)  # note - uncommenting this doesn't change the outcome
list(APPEND CMAKE_PREFIX_PATH external/backward-cpp)
find_package(Backward)
add_executable(${PROJECT_NAME} main.cpp)
target_link_libraries(${PROJECT_NAME} PUBLIC Backward::Backward)

Create main.cpp:

#include <csignal>

int main()
{
    raise(SIGSEGV);
    return 0;
}

Build and run the project:

mkdir build
cd build
cmake ..
make
./backward-cpp-test

EXPECTED RESULT

Program should print:
Segmentation fault (core dumped)
followed by a stack trace.

ACTUAL RESULT

Program prints only:
Segmentation fault (core dumped)

ENVIRONMENT
Ubuntu 17.10
GCC 7.2
CMake 3.11
libbfd installed through binutils-dev
(all 64-bit)

It would be nice to be able to suppress the initial line (e.g., stack 0x7f697f84b010 - 0x7f698004b010) in the default output. It's not a problem if there is an exceptional termination with a stack trace, but it's unnecessarily noisy for a normal execution.

Hello!

Would you mind tagging a new release 1.2 ? So I can create the conda-forge packages with the latest commit.

Thanks!

Tinkering around and would like to get stack traces on osx. binutils is in homebrew and I found argp-standalone in homebrew (required by elfutils), then I got elfutils to not complain about GNU99 support, guessing that this is an issue in elfutils?

Just a small suggestion to add some kind of switch via a IFDEF to print from "super pretty" stack trace to the copypastable one

file:line sintax, so many program will just go to the right place.

If you deem that reasonable I will do a merge request, else in

void print_source_loc(std::ostream& os, const char* indent,

I now have

		auto fileNameClean = source_loc.filename.substr(3);
		os << indent
		   << "Source \""
		   << fileNameClean
		   << ":"
		   << source_loc.line
		   << "\"\n in "
		   << source_loc.function;

2. So double click and most shell will select what is inside the "
3. ??
4. profit

Hi, I saw that some commits were added after the 1.0 release.

I suggest that a 1.1 release be tagged.

Hello,
Do you know about Conan?
Conan is modern dependency manager for C++. And will be great if your library will be available via package manager for other developers.

Here you can find example, how you can create package for the library.

If you have any questions, just ask :-)

We needed -funwind-tables option on ARM processors to get backtrace. Could that be added to CMakeLists.txt conditionally for ARM ?

Semantically it seems like StackTrace::operator[](size_t idx) is a const operation, since it is simply returning by value a Trace object constructed out of the current state of the stack traces, but it is not currently declared const.

This impedes some uses, the obvious ones being that you have to pass around a non-const StackTrace object to methods that might use [], but I ran into the case where a StackTrace member of another class couldn't be accessed with operator[] in a const function of the containing class.

The current (Linux) implementation of operator[] doesn't have any problem being declared const. Adding const should be a non-breaking change, although you couldn't rule out that some other implementation would want to make this a physically non-const operation (e.g., caching values, whatever) - but that's where mutable comes in.

When I build using clang (from clion) on a 64-bit mac I get the following error message:

backward-cpp/backward.hpp:2093:3: warning: ":/ sorry, ain't know no nothing none not of your architecture!" [-W#warnings]`
#       warning ":/ sorry, ain't know no nothing none not of your architecture!"
        ^

Looking at the code, it looks like #ifdef REG_RIP // x86_64 is not defined for my architecture, and it probably shouldn't be. That's not a compiler-defined preprocessor value for x86 architecture, but __x86_64__ is

is it possible to use this library to examin local variables stored in the stack frame?

similar to what gdb does with "info locals" command.

If not can you recommend me something?

Best regards,

ramy gad

I noticed this comment on HN (https://news.ycombinator.com/item?id=5381414) and did not see an Issue filed. I'm experiencing a similar issue when compiling my project (FCEUX) with clang++ and backward-cpp. I'm using "libdw" and passing "-DBACKWARD_HAS_DW=1" to the compiler. I receive the following error when compiling with clang++:

clang++ -o src/utils/backward.o -c -Wall -Wno-write-strings -Wno-sign-compare -Isrc/lua/src -DBACKWARD_HAS_DW=1 -DHAVE_ASPRINTF -pthread -D_GTK -DLUA_USE_LINUX -DOPENGL -g -DHAVE_GD -D_GNU_SOURCE=1 -D_REENTRANT -D_GTK2 -D_S9XLUA_H -DPSS_STYLE=1 -DLSB_FIRST -DFRAMESKIP -D_DEBUG -DCREATE_AVI -I/usr/include/SDL -I/usr/include/gtk-2.0 -I/usr/lib/gtk-2.0/include -I/usr/include/pango-1.0 -I/usr/include/atk-1.0 -I/usr/include/cairo -I/usr/include/pixman-1 -I/usr/include/libdrm -I/usr/include/libpng15 -I/usr/include/gdk-pixbuf-2.0 -I/usr/include/harfbuzz -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include -I/usr/include/freetype2 src/utils/backward.cpp
In file included from src/utils/backward.cpp:26:
src/utils/backward.hpp:595:18: error: use of undeclared identifier '_Unwind_GetIPInfo'
                uintptr_t ip = _Unwind_GetIPInfo(ctx, &ip_before_instruction);
                               ^
1 error generated.
scons: *** [src/utils/backward.o] Error 1
scons: building terminated because of errors.

When compiling with "-DBACKWARD_HAS_BACKTRACE=1" in addition to "-DBACKWARD_HAS_DW=1", as suggested in your reply to the HN comment, everything works as per usual. In addition, when compiling with g++, the "-DBACKWARD_HAS_BACKTRACE=1" is not required and everything works as suspected.

I'm testing with Arch Linux (64 bit fwiw). If you need any additional information, feel free to let me know. Thanks again for this really great/useful open source contribution!

Here is the backtrace printed by gdb,

#0  0x000000000287790a in sys_futex (a=0x4087f80 <tcmalloc::Static::central_cache_+1216>, o=128, v=2, t=0x7febee032860) at /home/amos/repos/dsql/src/gutil/linux_syscall_support.h:2713
#1  0x0000000002877a4b in base::internal::SpinLockDelay (w=0x4087f80 <tcmalloc::Static::central_cache_+1216>, value=2, loop=1073) at /home/amos/repos/dsql/src/gutil/spinlock_linux-inl.h:88
#2  0x00000000028af9a9 in SpinLock::SlowLock() ()
#3  0x0000000002967cd9 in tcmalloc::CentralFreeList::RemoveRange(void**, void**, int) ()
#4  0x00000000029753b3 in tcmalloc::ThreadCache::FetchFromCentralCache(unsigned long, unsigned long) ()
#5  0x000000000298236b in tc_malloc ()
#6  0x00000000028dca86 in bfd_malloc ()
#7  0x00000000028dd2bb in bfd_follow_gnu_debuglink ()
#8  0x0000000002954d37 in find_line ()
#9  0x0000000002954e97 in _bfd_dwarf2_find_nearest_line ()
#10 0x00000000028efefa in _bfd_elf_find_nearest_line ()
#11 0x00000000011b2858 in backward::TraceResolverLinuxImpl<backward::trace_resolver_tag::libbfd>::find_in_section (this=0x7febee032ff8, addr=213107247788, base_addr=213106294784, fobj=..., section=0x12e02ce0, result=...) at /home/amos/repo
s/dsql/src/util/backward.hpp:1084
#12 0x00000000011b26f9 in backward::TraceResolverLinuxImpl<backward::trace_resolver_tag::libbfd>::find_in_section_trampoline (section=0x12e02ce0, data=0x7febee032d60) at /home/amos/repos/dsql/src/util/backward.hpp:1060
#13 0x00000000028de4dc in bfd_map_over_sections ()
#14 0x00000000011b266c in backward::TraceResolverLinuxImpl<backward::trace_resolver_tag::libbfd>::find_symbol_details (this=0x7febee032ff8, fobj=..., addr=0x319e2e8aac <clone+108>, base_addr=0x319e200000) at /home/amos/repos/dsql/src/util/backward.hpp:1048
#15 0x00000000011b203e in backward::TraceResolverLinuxImpl<backward::trace_resolver_tag::libbfd>::resolve (this=0x7febee032ff8, trace=...) at /home/amos/repos/dsql/src/util/backward.hpp:817
#16 0x00000000011b51d1 in backward::Printer::print<backward::StackTrace> (this=0x7febee032ff0, st=..., os=0x319e58f120 <_IO_2_1_stderr_>) at /home/amos/repos/dsql/src/util/backward.hpp:1747
#17 0x00000000011b3ee4 in backward::SignalHandling::sig_handler (info=0x7febee033470, _ctx=0x7febee033340) at /home/amos/repos/dsql/src/util/backward.hpp:1959
#18 0x00007fecddbe8852 in os::Linux::chained_handler(int, siginfo*, void*) () from /home/amos/repos/dsql/install/postgres/jdk/jre/lib/amd64/server/libjvm.so
#19 0x00007fecddbeec86 in JVM_handle_linux_signal () from /home/amos/repos/dsql/install/postgres/jdk/jre/lib/amd64/server/libjvm.so
#20 0x00007fecddbe54b3 in signalHandler(int, siginfo*, void*) () from /home/amos/repos/dsql/install/postgres/jdk/jre/lib/amd64/server/libjvm.so
#21 <signal handler called>

https://travis-ci.org/bombela/backward-cpp/jobs/319897290

conan test_package --build=outdated
ERROR: Unknown command 'test_package'

https://github.com/conan-io/conan/releases

Command test_package has been removed. Use conan create instead, and conan test for just running package tests.

Also on line 35 in the conanfile.py I think the new practice is to use CMake(self) instead of CMake(self.settings). Ref: https://github.com/conan-io/docs/blob/master/reference/build_helpers/cmake.rst

0.29.0: Deprecation: Removed old CMake helper methods (only valid constructor is CMake(self))

"You will need to install some dependencies to get the ultimate stack trace. Two libraries are currently supported, the only difference is which one is the easiest for you to install, so pick your poison:"

after which you mention libbfd, libdw, libdwarf, and libelf, which is four and not two to me. Can you clarify what library(s) are actually needed for this ultimate stack trace support ?

I am trying backward-cpp with following sample code:

test.cpp:

#include <assert.h>
#include "backward.hpp"
int main()
{
  assert(0);
}

build command:

g++ -g test.cpp backward.cpp

but stack is not providing any useful information:

Stack trace (most recent call last):
#7 Object "[0xffffffffffffffff]", at 0xffffffffffffffff, in
#6 Object "./a.out", at 0x4019d8, in
#5 Object "/lib64/libc.so.6", at 0x7f540b545af4, in __libc_start_main
#4 Object "./a.out", at 0x401ab9, in
#3 Object "/lib64/libc.so.6", at 0x7f540b5525f1, in __assert_fail
#2 Object "/lib64/libc.so.6", at 0x7f540b552545, in
#1 Object "/lib64/libc.so.6", at 0x7f540b55acc7, in abort
#0 Object "/lib64/libc.so.6", at 0x7f540b5595d7, in gsignal
Aborted (Signal sent by tkill() 85950 28120)
Aborted

can you please advice whats I am doing wrong ?

backward uses malloc in the signal handler, which cause a very nasty deadlock in case if original signal raised by malloc itself, for example:

#0  0x00007f561384912c in __lll_lock_wait_private () from /lib64/libc.so.6
#1  0x00007f56137c6f93 in _L_lock_14932 () from /lib64/libc.so.6
#2  0x00007f56137c4013 in malloc () from /lib64/libc.so.6
#3  0x00007f56140ab288 in operator new(unsigned long) () from /usr/lib/gcc/x86_64-redhat-linux/5.3.1/libstdc++.so.6
#4  0x00000000005af533 in allocate (this=0x7f555ffec950, __n=<optimized out>) at /usr/include/c++/4.9.2/ext/new_allocator.h:104
#5  allocate (__a=..., __n=<optimized out>) at /usr/include/c++/4.9.2/bits/alloc_traits.h:357
#6  _M_allocate (this=0x7f555ffec950, __n=<optimized out>) at /usr/include/c++/4.9.2/bits/stl_vector.h:170
#7  std::vector<void*, std::allocator<void*> >::_M_default_append (this=this@entry=0x7f555ffec950, __n=40) at /usr/include/c++/4.9.2/bits/vector.tcc:557
#8  0x00000000005af848 in _M_default_append (__n=<optimized out>, this=0x7f555ffec950) at /usr/include/c++/4.9.2/bits/vector.tcc:550
#9  resize (__new_size=40, this=0x7f555ffec950) at /usr/include/c++/4.9.2/bits/stl_vector.h:676
#10 backward::StackTraceImpl<backward::system_tag::linux_tag>::load_here (this=this@entry=0x7f555ffec940, depth=depth@entry=40) at /usr/src/debug/crow-cluster/third-party/backward-cpp/backward.hpp:616
#11 0x00000000005b1eca in load_from (depth=32, addr=0x7f56137bf823 <malloc_consolidate+307>, this=0x7f555ffec940) at /usr/src/debug/crow-cluster/third-party/backward-cpp/backward.hpp:623
#12 backward::SignalHandling::sig_handler (info=0x7f555ffecb30, _ctx=<optimized out>) at /usr/src/debug/crow-cluster/third-party/backward-cpp/backward.hpp:1960
#13 <signal handler called>
#14 0x00007f56137bf823 in malloc_consolidate () from /lib64/libc.so.6
#15 0x00007f56137c1385 in _int_malloc () from /lib64/libc.so.6
#16 0x00007f56137c3fbc in malloc () from /lib64/libc.so.6
#17 0x00007f56140ab288 in operator new(unsigned long) () from /usr/lib/gcc/x86_64-redhat-linux/5.3.1/libstdc++.so.6

I don't think malloc is safe to be called from signal handlers at all, at least POSIX doesn't state so

In BackwardConfig.cmake, macro add_backward does not necessarily get correctly expanded:
indeed it uses variable _BACKWARD_INCLUDE_DIRS (instead of BACKWARD_INCLUDE_DIRS), so when this macro is called elsewhere in another cmake scope, _BACKWARD_INCLUDE_DIRS is empty while BACKWARD_INCLUDE_DIRS is ok.

The simple fix is just to replace _BACKWARD_INCLUDE_DIRS by BACKWARD_INCLUDE_DIRS

Your member function "print" is not async-signal-safe.

I suggest also to reconsider the implementation detail about the use of "sig_handler" as a static member function for the intended purpose. Does the class "SignalHandling" need a different design for portable signal handling?

Does your code work on all platforms ? I want to use it on an embedded arm 64 bit system.

FILE * limits usage options.

Right now code snippets are of some hard-coded value. Lets expose them for customization.

Is it possible to enable the signal handlers from a python runtime? Its create from the documentation how one can add signal handlers to a C++ executable with its cmake target; however, it's unclear how one could register signal handlers for a python process using C++ extensions defined in a library containing backward-cpp.

SIGUSR1, SIGUSR2 and SIGPOLL should be excluded from signal handling.

I'm using nvcc_wrapper (a compiler wrapper provided by project
https://github.com/kokkos/kokkos ) which wraps calls to the actual nvidia nvcc compiler.

When using nvcc_wrapper in a cmake project, cmake recognizes it as GNU : CMAKE_COMPILER_IS_GNUCXX is true, but actually:

• nvcc does not support GNU cxx extensions, i.e. one must use "-std=c++11" instead of "-std=gnu++11"
• nvcc does not support flag "-pedantic-errors"

Would you mind if I provide a simple modification to the top level CMakeLists.txt that checks if compiler is nvcc/nvcc_wrapper and if true, turns "-std=gnu++11" into "-std=c++11" and turns off pedantic-errors flag ?

like this：
Stack trace (most recent call last):
#1 Object "/lib/libc.so.6", at 0xf6f499bf, in __default_rt_sa_restorer_v1
#0 | Source "backward.hpp", line 2254, in backward::SignalHandling::sig_handler(int, siginfo_t*, void*)
| Source "backward.hpp", line 2232, in backward::SignalHandling::handleSignal(int, siginfo_t*, void*)
Source "backward.hpp", line 710, in backward::StackTraceImplbackward::system_tag::linux_tag::load_from(void*, unsigned int) [0x17183]

Which line in my source file is not located? I have follow define：
#define BACKWARD_HAS_BFD 1
#define BACKWARD_HAS_UNWIND == 1

thanks！

Hi,

I have a project in which Python uses static C++ libraries. How do I include backward in the project? Do I have to include it in each CMake file for each library? An example would be great.

Thanks

See conan-io/conan#1008

It seems that openSUSE Tumbleweed is removing SIGUNUSED definition from /usr/include/bits/signum.h, indirectly messing class SignalHandling compilation at backward.hpp (line 1978).

Hi,

I was thinking if the project has merged enough fixes and changes to consider releasing a v1.4 version. What do you think?

In default configuration, there is not any function name info in the log. Is this right?

Stack trace (most recent call last):
#3 Object "/lib/libc.so.6", at 0x442d77f7, in __libc_start_main
#2 Object "/usr/bin/mobile", at 0xf70b, in
#1 Object "/lib/libc.so.6", at 0x442edb4f, in
#0 Object "/usr/bin/mobile", at 0xdd6b, in

Segmentation fault (Address not mapped to object [(nil)])

README suggests appending a path to CMAKE_MODULE_PATH to find Backward-cpp.

However that results in a following error:

CMake Warning at CMakeLists.txt:7 (find_package):
  By not providing "FindBackward.cmake" in CMAKE_MODULE_PATH this project has
  asked CMake to find a package configuration file provided by "Backward",
  but CMake did not find one.

  Could not find a package configuration file provided by "Backward" with any
  of the following names:

    BackwardConfig.cmake
    backward-config.cmake

  Add the installation prefix of "Backward" to CMAKE_PREFIX_PATH or set
  "Backward_DIR" to a directory containing one of the above files.  If
  "Backward" provides a separate development package or SDK, be sure it has
  been installed.


Configuring done
CMake Warning (dev) at CMakeLists.txt:4 (add_executable):
  Policy CMP0028 is not set: Double colon in target name means ALIAS or
  IMPORTED target.  Run "cmake --help-policy CMP0028" for policy details.
  Use the cmake_policy command to set the policy and suppress this warning.

  Target "backward-cpp-test" links to target "Backward::Backward" but the
  target was not found.  Perhaps a find_package() call is missing for an
  IMPORTED target, or an ALIAS target is missing?
This warning is for project developers.  Use -Wno-dev to suppress it.

Generating done

Trying to compile it results in this build error:

14:54:43: Running steps for project backward-cpp-test...
14:54:43: Starting: "/usr/bin/cmake" --build . --target all
Scanning dependencies of target backward-cpp-test
[ 50%] Building CXX object CMakeFiles/backward-cpp-test.dir/main.cpp.o
/home/arek/dev/backward-cpp-test/main.cpp:2:10: fatal error: backward.hpp: No such file or directory
 #include <backward.hpp>
          ^~~~~~~~~~~~~~
compilation terminated.

Replacing CMAKE_MODULE_PATH with CMAKE_PREFIX_PATH fixes the problem for me.
Note that I haven't checked older versions of CMake.

My setup:
Xubuntu 17.10 x64
CMake 3.9.1
GCC 7.2.0

My CMakeLists.txt:

cmake_minimum_required(VERSION 2.8)

project(backward-cpp-test)
add_executable(${PROJECT_NAME} "main.cpp")

list(APPEND CMAKE_MODULE_PATH ${CMAKE_SOURCE_DIR}/external/backward-cpp)
find_package(Backward)
target_link_libraries(${PROJECT_NAME} PUBLIC Backward::Backward)

My main.cpp file:

#include <backward.hpp>

using namespace backward;

int main()
{
    StackTrace st;
    st.load_here(32);
    Printer p;
    p.print(st, std::cerr);

    return 0;
}

I am trying to use the CMake integration using the imported target.

It does not work when none of the linux library (libbfd or libdw) is installed on the system.
In this case, the variable BACKWARD_LIBRARIES is not defined. However, you are looking for this variable: https://github.com/bombela/backward-cpp/blob/master/BackwardConfig.cmake#L131

Do you see a way to fix this configuration issue?

Thanks for your help

In file included from /home/user/ALPHABOX/alpha-app/app/recorder/main.cpp:27:0:
/home/user/ALPHABOX/alpha-app/app/recorder/backward.hpp:1938:3: warning: #warning ":/ sorry, ain't know no nothing none not of your architecture!"
/home/user/ALPHABOX/alpha-app/app/recorder/backward.hpp: In static member function 'static void backward::SignalHandling::sig_handler(int, siginfo_t_, void_)':
/home/user/ALPHABOX/alpha-app/app/recorder/backward.hpp:1929:15: warning: unused variable 'uctx'
In file included from /home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/bits/move.h:38:0,
from /home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/bits/stl_pair.h:60,
from /home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/bits/stl_algobase.h:66,
from /home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/bits/char_traits.h:41,
from /home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/ios:41,
from /home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/ostream:40,
from /home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/iostream:40,
from /home/user/ALPHABOX/alpha-app/app/recorder/main.cpp:4:
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/type_traits: At global scope:
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/type_traits: In instantiation of 'const bool std::__is_convertible_helper<backward::SourceFile, backward::SourceFile, false>::__value':
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/type_traits:312:5: instantiated from 'std::is_convertible<backward::SourceFile, backward::SourceFile>'
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/bits/stl_map.h:451:11: instantiated from 'mapped_type& std::map<_Key, _Tp, _Compare, _Alloc>::operator[](const key_type&) [with _Key = std::basic_string, _Tp = backward::SourceFile, _Compare = std::lessstd::basic_string, _Alloc = std::allocator<std::pair<const std::basic_string, backward::SourceFile> >, mapped_type = backward::SourceFile, key_type = std::basic_string]'
/home/user/ALPHABOX/alpha-app/app/recorder/backward.hpp:1647:49: instantiated from here
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/type_traits:302:71: error: no matching function for call to 'backward::SourceFile::SourceFile(backward::SourceFile)'
/home/user/ALPHABOX/alpha-app/app/recorder/backward.hpp:1481:2: note: candidates are: backward::SourceFile::SourceFile(const std::string&)
/home/user/ALPHABOX/alpha-app/app/recorder/backward.hpp:1480:2: note: backward::SourceFile::SourceFile()
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/type_traits:302:71: error: initializing argument 1 of 'static std::__sfinae_types::__one std::__is_convertible_helper<_From, _To, false>::__test(_To) [with _From = backward::SourceFile, _To = backward::SourceFile, std::__sfinae_types::__one = char]'
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/type_traits: In instantiation of 'std::is_convertible<backward::SourceFile, backward::SourceFile>':
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/bits/stl_map.h:451:11: instantiated from 'mapped_type& std::map<_Key, _Tp, _Compare, _Alloc>::operator[](const key_type&) [with _Key = std::basic_string, _Tp = backward::SourceFile, _Compare = std::lessstd::basic_string, _Alloc = std::allocator<std::pair<const std::basic_string, backward::SourceFile> >, mapped_type = backward::SourceFile, key_type = std::basic_string]'
/home/user/ALPHABOX/alpha-app/app/recorder/backward.hpp:1647:49: instantiated from here
/home/user/SDK/linux/linux-current/ti-sdk-am335x-evm-05.04.01.00/linux-devkit/bin/../lib/gcc/arm-arago-linux-gnueabi/4.5.3/../../../../arm-arago-linux-gnueabi/include/c++/4.5.3/type_traits:312:5: error: 'std::__is_convertible_helper<backward::SourceFile, backward::SourceFile, false>::__value' is not a valid template argument for type 'bool' because it is a non-constant expression

builds.sh cmake and builds.sh make work, but builds.sh test results in

make: Entering directory '~/workspace/backward-cpp/build_c++98_gcc-4.6'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++98_gcc-4.6'
make: Entering directory '~/workspace/backward-cpp/build_c++98_gcc-4.8'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++98_gcc-4.8'
make: Entering directory '~/workspace/backward-cpp/build_c++98_gcc-530'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++98_gcc-530'
make: Entering directory '~/workspace/backward-cpp/build_c++98_gcc-620'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++98_gcc-620'
make: Entering directory '~/workspace/backward-cpp/build_c++98_clang'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++98_clang'
make: Entering directory '~/workspace/backward-cpp/build_c++11_gcc-4.8'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++11_gcc-4.8'
make: Entering directory '~/workspace/backward-cpp/build_c++11_gcc-530'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++11_gcc-530'
make: Entering directory '~/workspace/backward-cpp/build_c++11_gcc-620'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++11_gcc-620'
make: Entering directory '~/workspace/backward-cpp/build_c++11_clang'
make: *** No rule to make target 'test'.  Stop.
make: Leaving directory '~/workspace/backward-cpp/build_c++11_clang'

I have an use case where I clone the backward repository as a dependency of my project at CMake configure time (i.e. when cmake is run). The project is cloned in a directory part of my buildtree, which is then included by my project CMakeLists.txt. Like:

git_clone(backward-cpp)

add_subdirectory(backward-cpp_srcdir)

For some reason I cannot figure out the find_path(BACKWARD_INCLUDE_DIR ...) command in BackwardConfig.cmake fails. When trying to debug this issue I started to think if find_path() is the "best" (of course best is completely subjective...) way to set the library include dir. I mean, if I'm not wrong with the current project layout a simple

set(BACKWARD_INCLUDE_DIR "${CMAKE_CURRENT_LIST_DIR}")

would do the work. The only advantage of using find_path (please correct me if I'm wrong) is that a change in the config script is not required if we change the project layout (the header would be found anyway).

If you don't mind I'm going to push a PR changing that find_path to the alternative explained above.

If an executable is run by e.g. being on the PATH (anything that results in argv[0] not being a valid path to the executable), problems ensue when backward-cpp tries to print a stack trace.

dladdr returns argv[0] in dli_fname for symbols contained in the main executable, which is not a valid path if the executable was found by a search of the PATH environment variable. Then, load_object_with_bfd is called on that non-existent filename and fails.

This would just result in not getting a pretty stack trace in this situation if not for another bug:

One of the handle objects somehow gets filled with a nullptr but with _empty = false, resulting in bfd_close eventually being called with nullptr as the argument.

That results in a segfault midway through the stack trace.

Test platform: Ubuntu 16.04.3 (xenial) amd64

Did the docs and code get out of sync at some point? The readme refers to the class as "Stacktrace" and "StackTrace" in different places (including example code) but it looks like it should only be "StackTrace".

Howto: http://docs.conan.io/en/latest/uploading_packages/bintray/conan_center_guide.html#conan-center-flow

@Manu343726 Do you have any experience with this? Is it better to go via bincrafters?

Hey,
while installing the library in my project i got 2 compiler warnings due to my pedantic settings ("/W4" under Windows, "-Wall -Wextra -Werror -Wno-long-long -Wpedantic" under linux).

It would be cool if you could fix them in the original code, this is the diff i used to fix both errors, however you should have a brief look the solution because i just started working with c++:

1st Error: compiler warning because of possible lost of data when converting size_t to unsigned
2nd Error: isspace returns an int, but the function should return a bool

 backward.hpp
@@ -435,7 +435,7 @@ struct Trace {
 		addr(0), idx(0) {}
 
 	explicit Trace(void* addr, size_t idx):
- 		addr(addr), idx(idx) {}
+ 		addr(addr), idx(unsigned(idx)) {}
 };
 
 struct ResolvedTrace: public Trace {
@@ -1520,7 +1520,7 @@ public:
 		// What a good boy am I.
 		struct isspace {
 			bool operator()(char c) {
- 				return std::isspace(c);
+ 				return std::isspace(c) != 0;
 			}
 		};

For BACKWARD_HAS_BFD == 1 use at least, it seems that libdl (i.e., -ldl on the link line) is needed in addition to libbfd.

Might be worth mentioning it in the readme.

As far I know backward-cpp doesn't currently support Windows. I've been checking the Windows API docs and it seems really simple to take at least a minimal stacktrace (Not sure about source code printing).

I will be glad to work on this.

Something like:
print(std::ostream& os, bool assume_color_support=false)

I am trying to use the first CMake method, and am totally a CMake novice. I got the project to build fine without the 2 external dependencies (so the less pretty stack traces).

add_subdirectory(/path/to/backward-cpp)

# This will add backward.cpp to your target
add_executable(mytarget mysource.cpp ${BACKWARD_ENABLE})
target_link_libraries(mytarget
[my libraries])
# This will add libraries, definitions and include directories needed by backward
# by setting each property on the target.
add_backward(mytarget)

I installed both of the suggested extra dependencies, and the BackwardConfig.cmake file is detecting one of them, so I get

-- BACKWARD_HAS_UNWIND=1
-- BACKWARD_HAS_BACKTRACE=0
-- BACKWARD_HAS_BACKTRACE_SYMBOL=0
-- BACKWARD_HAS_DW=1
-- BACKWARD_HAS_BFD=0

(As an aside, the logic located in BackwardConfig.cmake makes it seem like HAS_DW or HAS_BFD could only be true mutually exclusively, was this the intention?)

However, when I build the program, I get an error with a long list of 'undefined reference' e.g.

CMakeFiles/backward.dir/backward.cpp.o: In function `backward::TraceResolverLinuxImpl<backward::trace_resolver_tag::libdw>::resolve(backward::ResolvedTrace)':
/home/[my path]/backward-cpp/backward.hpp:1158: undefined reference to `dwfl_linux_proc_find_elf'
/home/[my path]/backward-cpp/backward.hpp:1159: undefined reference to `dwfl_standard_find_debuginfo'
etc.

which I think is a linker error? Any idea what I'm doing wrong? Do I need to add a line in target_link_libraries?

gcc 8 complains about switch without a default case and about old style cast ((type) var) as mentioned in anbox/anbox#809

Hi, i followed the Readme but i don't get to the pretty stacktrace, only the simple one:

i=0
i=1
i=2
i=3
i=4
i=5
i=6
i=7
i=8
i=9
i=10
i=11
i=12
i=13
i=14
i=15
i=16
i=17
i=18
i=19
i=20
i=21
i=22
i=23
i=24
i=25
i=26
i=27
i=28
i=29
i=30
i=31
i=32
i=33
i=34
i=35
i=36
i=37
i=38
i=39
i=40
i=41
#0 at unsigned long backward::details::unwind<backward::StackTraceImpl<backward::system_tag::linux_tag>::callback>(backward::StackTraceImpl<backward::system_tag::linux_tag>::callback, unsigned long)
#1 at backward::StackTraceImpl<backward::system_tag::linux_tag>::load_here(unsigned long)
#2 at TracedException::_get_trace[abi:cxx11]()
#3 at TracedException::TracedException()
#4 at f(int)
#5 at f(int)
#6 at f(int)
#7 at f(int)
#8 at f(int)
#9 at f(int)
#10 at f(int)
#11 at f(int)
#12 at f(int)
#13 at f(int)
#14 at f(int)
#15 at f(int)
#16 at f(int)
#17 at f(int)
#18 at f(int)
#19 at f(int)
#20 at f(int)
#21 at f(int)
#22 at f(int)
#23 at f(int)
#24 at f(int)
#25 at f(int)
#26 at f(int)
#27 at f(int)
#28 at f(int)
#29 at f(int)
#30 at f(int)
#31 at f(int)
Stack trace (most recent call last):

I copied the source from the _test_main.cpp file to my project

#include "backward.hpp"

using namespace backward;

class TracedException : public std::runtime_error
{
public:
	TracedException() :
		std::runtime_error(_get_trace())
	{}

private:
	std::string _get_trace()
	{
		std::ostringstream ss;

		StackTrace stackTrace;
		TraceResolver resolver;
		stackTrace.load_here();
		resolver.load_stacktrace(stackTrace);

		for (std::size_t i = 0; i < stackTrace.size(); ++i)
		{
			const ResolvedTrace trace = resolver.resolve(stackTrace[i]);

			ss << "#" << i << " at " << trace.object_function << "\n";
		}

		return ss.str();
	}
};

void f(int i)
{
	if (i >= 42)
	{
		throw TracedException();
	}
	else
	{
		std::cout << "i=" << i << "\n";
		f(i + 1);
	}
}

namespace {

	TEST(pfFiles, TestStackError)
	{
		backward::StackTrace st;
		try
		{
			f(0);
		}
		catch (const TracedException& ex)
		{
			std::cout << ex.what();
		}

		backward::Printer printer;
		//    printer.address = true;
		printer.object = true;
		printer.print(st, stdout);
	}
}

I use the following CMAKE settings.

  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wall -Wextra -Werror -Wno-long-long -Wpedantic -std=c++14 -g")
  set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -lbfd")

I have installed backward-cpp as subdirectory in my cmake script.
I installed the gnu binutils package under ubuntu 16.04LTS with gcc 5.4.
I neither set the define (-DBACKWARD_HAS_...=1) flag nor used #define BACKWARD_HAS_BFD 1 because it seems that your cmake script automaticly detects the library:

Looking for IceConnectionNumber in ICE - found
-- Found X11: /usr/lib/x86_64-linux-gnu/libX11.so
-- Could NOT find libdw (missing:  LIBDW_LIBRARY LIBDW_INCLUDE_DIR) 
-- Found libbfd: /usr/lib/x86_64-linux-gnu/libbfd.so  
-- BACKWARD_HAS_UNWIND=1
-- BACKWARD_HAS_BACKTRACE=0
-- BACKWARD_HAS_BACKTRACE_SYMBOL=0
-- BACKWARD_HAS_DW=0
-- BACKWARD_HAS_BFD=1
-- Found Backward: /home/gitlab-runner/builds/0c2717af/0/PfTTech/PfTEngine/3rdparty/backward-cpp  

i compile with:

add_executable(PfTEngineTest ${SOURCES} ${BACKWARD_ENABLE})
add_backward(PfTEngineTest)

So my question is: What do i have to do to get the full pretty code style?
Have i installed all dependencys correclty ?
Have i forgot something ?
The README doesn't specify what needs to be in code for pretty code ? is it enough just to compile backward.cpp with my target ?
Or do i have to create some classes ?
Do i need to throw special Exceptions or is everything catched ?
Do i need a TRY/CATCH block in main which covers everything ?

Thanks for any help :)

PS: I used the changed code from #51

Hello I think your library really cool and I am evaluating if we could use it in the project. So I have the following questions:

• How does the signal handling works since you are using classes like std::vector for in the printer class, which are not to be used while signal handling according to signals cpp?

Best Regards
Federico