No valid backtrace from EIPs

Tested with 3 different rpt/debug pairs compiled on Windows with the new makefile. GDB reads the symbols then spits out a rather useless "Breakpoint 1 at <eip>" which is pretty useless.

Sample rpt/debug pair:
https://cdn.discordapp.com/attachments/569169328386211883/888219734632460328/srb2-am-spin.RPT
https://cdn.discordapp.com/attachments/569169328386211883/888219788202090546/srb2-am-spin.exe.debug

added Ecosystem Regression labels

changed the description

changed milestone to %2.2.10

Tested c266a270 and 4e71ee7a I assume?

The former yes, the latter does not compile. Though upon closer investigation I may have been misled; since I lacked a proper crash from the pre-makefile exe I fed it an eip from vanilla 2.2.9. But using that same eip generated from 2.2.9 still produces a (most likely inaccurate but at least present) backtrace in latest next.

changed title from GDB backtrace doesn't work in builds compiled on Windows post-makefile change to No valid backtrace from EIPs

I should also note that Golden attempted addr2line with the suspect EIPs and got nothing.

changed the description

After additional testing I can confirm it is in fact the new makefile causing this. I abused the crushtacean bug to intentionally crash the debug exes of both c266a270 and 1721b5b0 and got the following results:

Pre-makefile:

Post-makefile:

Despite the actual crash log pointing to the correct line in both circumstances, the post-makefile build does not give any backtrace when fed to the debug exe via gdb.

What is your make line for both commits? make info may also be useful post-rewrite.

With the old makefile it is mingw32-make MINGW=1 WINDOWSHELL=1 SDL=1 NOOBJDUMP=1 CC=gcc

New makefile is just make

make info dump:

Detected a Windows system, compiling for 32-bit MinGW SDL...)
Your compiler version, GCC 10.3.0, is not supported by the Makefile.
The Makefile will assume GCC 10.2.
Using built-in specs.
COLLECT_GCC=C:\msys64\mingw32\bin\cc.exe
COLLECT_LTO_WRAPPER=C:/msys64/mingw32/bin/../lib/gcc/i686-w64-mingw32/10.3.0/lto-wrapper.exe
Target: i686-w64-mingw32
Configured with: ../gcc-10.3.0/configure --prefix=/mingw32 --with-local-prefix=/mingw32/local --build=i686-w64-mingw32 --host=i686-w64-mingw32 --target=i686-w64-mingw32 --with-native-system-header-dir=/mingw32/i686-w64-mingw32/include --libexecdir=/mingw32/lib --enable-bootstrap --enable-checking=release --with-arch=i686 --with-tune=generic --enable-languages=c,lto,c++,fortran,ada,objc,obj-c++,jit --enable-shared --enable-static --enable-libatomic --enable-threads=posix --enable-graphite --enable-fully-dynamic-string --enable-libstdcxx-filesystem-ts=yes --enable-libstdcxx-time=yes --disable-libstdcxx-pch --disable-libstdcxx-debug --enable-lto --enable-libgomp --disable-multilib --disable-rpath --disable-win32-registry --disable-nls --disable-werror --disable-symvers --with-libiconv --with-system-zlib --with-gmp=/mingw32 --with-mpfr=/mingw32 --with-mpc=/mingw32 --with-isl=/mingw32 --with-pkgversion='Rev5, Built by MSYS2 project' --with-bugurl=https://github.com/msys2/MINGW-packages/issues --with-gnu-as --with-gnu-ld --with-boot-ldflags='-pipe -Wl,--dynamicbase,--nxcompat,--no-seh -Wl,--large-address-aware -Wl,--disable-dynamicbase -static-libstdc++ -static-libgcc' 'LDFLAGS_FOR_TARGET=-pipe -Wl,--dynamicbase,--nxcompat,--no-seh -Wl,--large-address-aware' --enable-linker-plugin-flags='LDFLAGS=-static-libstdc++\ -static-libgcc\ -pipe\ -Wl,--dynamicbase,--nxcompat,--no-seh\ -Wl,--large-address-aware\ -Wl,--stack,12582912' --disable-sjlj-exceptions --with-dwarf2
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 10.3.0 (Rev5, Built by MSYS2 project)

SHELL ..... /bin/sh

CC ........ cc

CFLAGS .... -DCOMPVERSION -g -DSTDC_HEADERS -DUSE_WGL_SWAP -I../libs -DSTATIC_MINIUPNPC -DDIRECTFULLSCREEN -DHAVE_SDL -DHAVE_MIXER -DHAVE_MIXERX -DHAVE_THREADS -I../libs/SDL2/i686-w64-mingw32/include/SDL2 -I../libs/SDLMixerX/i686-w64-mingw32/include/SDL2 -Dmain=SDL_main -DSDLMAIN -DHWRENDER -DUSEASM -I../libs/libpng-src -DHAVE_PNG -I../libs/curl/include -DHAVE_CURL -I../libs/gme/include -DHAVE_GME -I../libs/libopenmpt/inc -DHAVE_OPENMPT -I../libs/zlib -DHAVE_ZLIB -msse3 -mfpmath=sse -O3 -DNDEBUG          -Wall -Wno-trigraphs -W -Wno-div-by-zero -Wfloat-equal -Wundef -Wendif-labels -Wdeclaration-after-statement -Wno-error=declaration-after-statement -Wshadow -Wpointer-arith -Wbad-function-cast -Wcast-qual -Wcast-align -Wwrite-strings -Wsign-compare -Wno-error=address-of-packed-member -Wlogical-op -Waggregate-return -Wold-style-definition -Wmissing-prototypes -Wmissing-declarations -Wmissing-field-initializers -Wmissing-noreturn -Wnested-externs -Winline -funit-at-a-time -Wlogical-op -Wdisabled-optimization -Wformat-y2k -Wno-error=format-overflow=2 -Wformat-security -Wno-suggest-attribute=noreturn -Wno-error=array-bounds -Wno-error=suggest-attribute=noreturn -Wno-logical-op -Wno-error=logical-op -Wno-tautological-compare -Wno-error=tautological-compare -Wimplicit-fallthrough=4 -Wno-error=format-overflow -Wno-error=stringop-truncation -Wno-error=stringop-overflow -Wno-format-overflow -Wno-stringop-truncation -Wno-stringop-overflow -Wno-error=multistatement-macros

LDFLAGS ... -ladvapi32 -lkernel32 -lmsvcrt -luser32 -lws2_32 -L../libs/miniupnpc/mingw -lws2_32 -liphlpapi -lSDL2_mixer_ext -L../libs/SDL2/i686-w64-mingw32/lib -L../libs/SDLMixerX/i686-w64-mingw32/lib -lmingw32 -lSDL2main -lSDL2 -mwindows -L../libs/libpng-src/projects -lpng32 -L../libs/curl/lib32 -lcurl -L../libs/gme/win32 -lgme -L../libs/libopenmpt/lib/x86/mingw -lopenmpt -L../libs/zlib/win32 -lz32

make[1]: Leaving directory '/c/SRB2/src'

Does gcc -v match make info?

i checked and they match
here's gcc -v anyway

COLLECT_GCC=C:\msys64\mingw32\bin\gcc.exe
COLLECT_LTO_WRAPPER=C:/msys64/mingw32/bin/../lib/gcc/i686-w64-mingw32/10.3.0/lto-wrapper.exe
Target: i686-w64-mingw32
Configured with: ../gcc-10.3.0/configure --prefix=/mingw32 --with-local-prefix=/mingw32/local --build=i686-w64-mingw32 --host=i686-w64-mingw32 --target=i686-w64-mingw32 --with-native-system-header-dir=/mingw32/i686-w64-mingw32/include --libexecdir=/mingw32/lib --enable-bootstrap --enable-checking=release --with-arch=i686 --with-tune=generic --enable-languages=c,lto,c++,fortran,ada,objc,obj-c++,jit --enable-shared --enable-static --enable-libatomic --enable-threads=posix --enable-graphite --enable-fully-dynamic-string --enable-libstdcxx-filesystem-ts=yes --enable-libstdcxx-time=yes --disable-libstdcxx-pch --disable-libstdcxx-debug --enable-lto --enable-libgomp --disable-multilib --disable-rpath --disable-win32-registry --disable-nls --disable-werror --disable-symvers --with-libiconv --with-system-zlib --with-gmp=/mingw32 --with-mpfr=/mingw32 --with-mpc=/mingw32 --with-isl=/mingw32 --with-pkgversion='Rev5, Built by MSYS2 project' --with-bugurl=https://github.com/msys2/MINGW-packages/issues --with-gnu-as --with-gnu-ld --with-boot-ldflags='-pipe -Wl,--dynamicbase,--nxcompat,--no-seh -Wl,--large-address-aware -Wl,--disable-dynamicbase -static-libstdc++ -static-libgcc' 'LDFLAGS_FOR_TARGET=-pipe -Wl,--dynamicbase,--nxcompat,--no-seh -Wl,--large-address-aware' --enable-linker-plugin-flags='LDFLAGS=-static-libstdc++\ -static-libgcc\ -pipe\ -Wl,--dynamicbase,--nxcompat,--no-seh\ -Wl,--large-address-aware\ -Wl,--stack,12582912' --disable-sjlj-exceptions --with-dwarf2
Thread model: posix
Supported LTO compression algorithms: zlib zstd
gcc version 10.3.0 (Rev5, Built by MSYS2 project)

Posting my findings as I discovered that typing in info sources within the GDB shell seems to give different outputs pre-makefile rewrite and post-makefile write, maybe relates in a way.

Output from build made from the v2.2.9 tag

(gdb) info sources
Source files for which symbols have been read in:

/e/GitRepos/STJr/SRB2/src/p_inter.c,
E:\GitRepos\STJr\SRB2\src/C:/msys64/mingw32/i686-w64-mingw32/include/stdio.h,
/e/GitRepos/STJr/SRB2/src/m_fixed.h,
E:\GitRepos\STJr\SRB2\src/C:/msys64/mingw32/i686-w64-mingw32/include/corecrt.h,
E:\GitRepos\STJr\SRB2\src/C:/msys64/mingw32/i686-w64-mingw32/include/minwindef.h,
E:\GitRepos\STJr\SRB2\src/C:/msys64/mingw32/i686-w64-mingw32/include/combaseapi.h,
E:\GitRepos\STJr\SRB2\src/C:/msys64/mingw32/i686-w64-mingw32/include/wtypes.h,
E:\GitRepos\STJr\SRB2\src/C:/msys64/mingw32/i686-w64-mingw32/include/stdint.h,
/e/GitRepos/STJr/SRB2/src/doomtype.h, /e/GitRepos/STJr/SRB2/src/doomdef.h,
/e/GitRepos/STJr/SRB2/src/g_state.h, /e/GitRepos/STJr/SRB2/src/d_ticcmd.h,
/e/GitRepos/STJr/SRB2/src/i_system.h, /e/GitRepos/STJr/SRB2/src/tables.h,
/e/GitRepos/STJr/SRB2/src/d_think.h, /e/GitRepos/STJr/SRB2/src/taglist.h,
/e/GitRepos/STJr/SRB2/src/doomdata.h, /e/GitRepos/STJr/SRB2/src/p_mobj.h,
/e/GitRepos/STJr/SRB2/src/sounds.h, /e/GitRepos/STJr/SRB2/src/info.h,
/e/GitRepos/STJr/SRB2/src/r_defs.h, /e/GitRepos/STJr/SRB2/src/d_player.h,
/e/GitRepos/STJr/SRB2/src/command.h, /e/GitRepos/STJr/SRB2/src/i_sound.h,
/e/GitRepos/STJr/SRB2/src/s_sound.h, /e/GitRepos/STJr/SRB2/src/am_map.h,
/e/GitRepos/STJr/SRB2/src/doomstat.h, /e/GitRepos/STJr/SRB2/src/d_netcmd.h,
/e/GitRepos/STJr/SRB2/src/d_clisrv.h, /e/GitRepos/STJr/SRB2/src/g_demo.h,
/e/GitRepos/STJr/SRB2/src/m_cheat.h, /e/GitRepos/STJr/SRB2/src/g_game.h,
/e/GitRepos/STJr/SRB2/src/p_tick.h, /e/GitRepos/STJr/SRB2/src/m_dllist.h,
/e/GitRepos/STJr/SRB2/src/p_polyobj.h, /e/GitRepos/STJr/SRB2/src/p_local.h,
/e/GitRepos/STJr/SRB2/src/p_spec.h, /e/GitRepos/STJr/SRB2/src/m_cond.h,
/e/GitRepos/STJr/SRB2/libs/libopenmpt/inc/libopenmpt/libopenmpt_config.h

Output from build post makefile rewrite

(gdb) info sources
Source files for which symbols have been read in:

Ignore the comment above as it turns out it has no relation to the issue at all.

for what it's worth i get wildly different outputs from objdump -g on a recent debug exe vs the 2.2.9 debug exe

dump from the new exe is very barren, more details at https://discord.com/channels/357945222111428609/569169328386211883/898833926762680320

should I ask why LDFLAGS does not have -g? We need to make debugging symbols while we are compiling, and well as during the linking step

I don't think -g was ever passed to the linker flags on the old Makefile. Adding it manually didn't change much either, as it would still not give valid backtrace from EIPs

and if you want see what EIP lines up with source code, you may want to compiling with -Og, not -O3... https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-Og

I compiled with -O0 back when I was first reporting the issue and it didn't make any difference.

A clean build? recompiling all the object files?

yes

4,5c4
< -DHAVE_LIBGME
< -DHAVE_MINIUPNPC
---
> -DHAVE_GME
21c20
< -I../libs/
---
> -I../libs
35a35
> -Wdeclaration-after-statement
40a41
> -Wimplicit-fallthrough=4
41a43,44
> -Wlogical-op
> -Wlogical-op
42a46
> -Wmissing-field-initializers
46a51,69
> -Wno-error=address-of-packed-member
> -Wno-error=array-bounds
> -Wno-error=declaration-after-statement
> -Wno-error=format-overflow
> -Wno-error=format-overflow=2
> -Wno-error=logical-op
> -Wno-error=multistatement-macros
> -Wno-error=stringop-overflow
> -Wno-error=stringop-truncation
> -Wno-error=suggest-attribute=noreturn
> -Wno-error=tautological-compare
> -Wno-format-overflow
> -Wno-logical-op
> -Wno-stringop-overflow
> -Wno-stringop-truncation
> -Wno-suggest-attribute=noreturn
> -Wno-tautological-compare
> -Wno-trigraphs
> -Wold-style-definition
47a71
> -Wshadow
52c76
< -fno-exceptions
---
> -funit-at-a-time
57c81
< ../objs/Mingw/SDL/Release/d_clisrv.o
---
> ../make/mingw/SDL/objs/./d_clisrv.o