You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Intention
I was reading through the build scripts for Dyninst 13.0.0 (porting to make a Meson wrap), and ran across a case in common/CMakeLists.txt where a source file is listed in the _private_headers array. This seemed like a typo, so I tried moving it to the _sources array:
ran across a case in common/CMakeLists.txt where a source file is listed in the _private_headers array
I sometimes impress myself with the stupid mistakes I make.
After applying the above patch, build Dyninst and attempt to load libcommon.so. The result is an FPE (divide by zero):
Oh, SIOF, you haunt my dreams.
For additional fun, there is an extra copy of MachRegister.h in the _public_headers list. I really made a mess here.
@kupsch I fixed this by moving the definitions of names and all_regs into dyn_regs.C so that they are in the same TU as the definitions of the registers and guaranteed to be initialized in order. It's a little ugly because it would make them global symbols in libcommon, but we can leave them unexported to keep them from escaping further. Thoughts?
Intention
I was reading through the build scripts for Dyninst 13.0.0 (porting to make a Meson wrap), and ran across a case in
common/CMakeLists.txt
where a source file is listed in the_private_headers
array. This seemed like a typo, so I tried moving it to the_sources
array:Describe the bug
Despite the seemingly innocuous nature of the above patch, the resulting libraries crash when used.
To Reproduce
After applying the above patch, build Dyninst and attempt to load
libcommon.so
. The result is an FPE (divide by zero):Expected behavior
No crash should happen.
System (please complete the following information):
The text was updated successfully, but these errors were encountered: