Peter Goodman

It's more than that. Here's an example: ``` pag$ echo | `which cc` -E -Wp,-v - clang -cc1 version 10.0.1 (clang-1001.0.46.4) default target x86_64-apple-darwin18.7.0 #include "..." search starts here: #include...

Really what we need is, instead of invoking `COMMAND "${CMAKE_BC_COMPILER}" - ...`, we should invoke a Python script that will do the right thing.

If you have IDA Pro on Windows, then you can invoke the mcsema-disass Python programs on Windows without building or installing mcsema. The way to do it is to do...

Also, the Windows Subsystem for Linux should in theory be able to build remill and mcsema. I tried it some time ago with WSL1 and it worked.

There is a `--keep_memops` option, that prevents McSema from lowering Remill memory access intrinsics into LLVM `load` and `store` instructions, is this what you're looking for?

It's possible/plausible that you also mean: don't even attempt to lift cross-references or data at all, and just lift the code, leaving all addresses as integers. That would be a...

That is very long :-P Do you have Binary Ninja? If you, you should try [anvill](https://github.com/lifting-bits/anvill).

I am not sure what is going on, but I strongly recommend migrating to [anvill](https://github.com/lifting-bits/anvill). I just compiled coremark for macOS, and opened up Binary Ninja on this specific function....

Otherwise, what it looks like mcsema is doing is seeing the indirect call, and then casting it to a generic `N`-integer-argument function. The `N` in this case is derived from...

mcsema hasn't been maintained in a long time, and is pinned on an older, less capable version of anvill and remill. Anvill has evolved substantially since then. However, Anvill still...