[Lldb-commits] [PATCH] D127048: [lldb] Set COFF and PDB module env from default target triple

[Martin Storsjö via Phabricator via lldb-commits]

mstorsjo added a comment.

In D127048#3559866 <https://reviews.llvm.org/D127048#3559866>, @DavidSpickett wrote:

>> This changes the PE/COFF and PDB plugins to set the module triple according to the default target triple used to build LLDB.
>
> I'm missing some context here but using triple used when *building* lldb seems to conflict with the idea that the same lldb (client at least) can be used to debug many architectures.

Yes, that's true. The idea here is that when doing local debugging in either the mingw or msvc ecosystems, the flavour of the lldb build itself could be a decent hint - but it's not indeed not a correct and faultproof fix.

> Or is this not a concern because remote debugging is not a thing for Windows? For instance, is it possible that I'm on Linux with some AArch64 linux targeted lldb trying to remote debug a windows system. If the default triple is the best signal you've got then fair enough I guess just wanted to check there wasn't some more dynamic way to get it.
>
> And I assume PE/COFF have no embedded flag in them to indicate this ABI?

No, there's no flag for that. Mingw and MSVC environment binaries are quite the same.

(Qt Creator tries to do some similar heuristics to guess which ABI a binary uses, by looking at e.g. linker version numbers in the binary, to deduce whether a binary has been linked with GNU ld.bfd or MS link.exe - but when it comes to LLD, that logic falls apart. A more complex heuristic that can be used is e.g. looking at the list of DLLs that a binary links against, but that's also much more complex and also not entirely faultproof.)

@alvinhochun has got another patch in progress - https://reviews.llvm.org/D127053 (which isn't yet submitted for review as it lacks tests), which tries to decude the same based on whether a binary contains dwarf debug info.

That's more fault proof, but is also not an entirely complete fix - while finding dwarf debug sections mostly can imply using the itanium C++ ABI, you can also use PDB files with itanium C++ ABI (i.e. in mingw setups). It's less common but still a viable scenario. And reversely, I think I've heard about special cases where people use dwarf debug info with MSVC-ecosystem builds too...

So in short, what would be needed is a good enough heuristic that works for the vast majority of cases, and an option that lets users specify it for the cases that can't easily be autodetected.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D127048/new/

https://reviews.llvm.org/D127048