Patch to remove 'using namespace std' from headers

[ekilmer]

closes #139

[mrexodia]

My intention was to try this patch with remill (since that's where I first ran into the issue), but it looks like there are some additional patches there that break with this branch. I managed to fix the ArmThumb patch from remill and the only issue I ran into was that register is no longer supported in C++17. This patch seems to fix that.

Unfortunately now I run into a linker error from remill (which I think related to the fact that remill is on a different version this repo and the API changed?):

[100%] Linking CXX executable remill-lift-15
Undefined symbols for architecture arm64:
  "ContextInternal::restoreFromSpec(Element const*, AddrSpaceManager const*)", referenced from:
      remill::sleigh::SingleInstructionSleighContext::restoreEngineFromStorage() in libremill_arch_sleigh.a(Arch.cpp.o)
ld: symbol(s) not found for architecture arm64
clang: error: linker command failed with exit code 1 (use -v to see invocation)

Here is my current WIP repo if you want to try: https://github.com/mrexodia/remill/tree/sleigh-using-namespace

[ekilmer]

My intention was to try this patch with remill (since that's where I first ran into the issue), but it looks like there are some additional patches there that break with this branch. I managed to fix the ArmThumb patch from remill and the only issue I ran into was that register is no longer supported in C++17. This patch seems to fix that.

Unfortunately now I run into a linker error from remill (which I think related to the fact that remill is on a different version this repo and the API changed?):

[100%] Linking CXX executable remill-lift-15
Undefined symbols for architecture arm64:
  "ContextInternal::restoreFromSpec(Element const*, AddrSpaceManager const*)", referenced from:
      remill::sleigh::SingleInstructionSleighContext::restoreEngineFromStorage() in libremill_arch_sleigh.a(Arch.cpp.o)
ld: symbol(s) not found for architecture arm64
clang: error: linker command failed with exit code 1 (use -v to see invocation)

Here is my current WIP repo if you want to try: https://github.com/mrexodia/remill/tree/sleigh-using-namespace

Ah. Yeah, this patch is for the latest Ghidra 10.2 version, which has some internal sleigh API changes, and remill has not been updated yet. Remill also has some additional patches, which I have not looked into yet and seem to require updating, as you've found out.

I won't have time to look into this too much this week, but I will take a look when I am back.

There are a few places I will start to look based on updating to Ghidra 10.2 in other tools:

• f38eee4#diff-aed2f30b604fcd3c832b0255d6cd0ef5a7cca4cab0d8d2c77ee6d125d2fcf301
• https://github.com/trailofbits/maat/pull/160/files#diff-7e5860b8ac02e6dc2650d19d18a9e3fb9ab939518a85d7b0f99dd67fca124216 (although these tests are failing for some reason now 🤔)
• https://github.com/trailofbits/mishegos/pull/1462/files#diff-a0f9b268e448e17e90e505165fe08069883b1e3b5d0c7cac92065f501f923722 (to a lesser extent)

[mrexodia]

Thanks, I'll take a look to see if a lot of effort is required to update remill to the latest Ghidra version!

[pgoodman]

Ping @ekilmer

[mrexodia]

Any chance this could be merged? I tested and this does fix the issue. It's a blocker for building remill on Windows and also a blocker for trailofbits/maat#97

[tetsuo-cpp]

LGTM from my point of view.

[ekilmer]

Updated the patches for the latest commit on Ghidra's master branch. I assume that if things are compiling correctly and passing tests, then it works for you as well @mrexodia.

Would you be willing to write a test (or post some C++ code that I can integrate into CMake) that we could include in this repo to make sure we don't suffer regressions?