[Lldb-commits] [PATCH] D49685: LLDB does not respect platform sysroot when loading core on Linux

[Greg Clayton via Phabricator via lldb-commits]

clayborg added a comment.

In https://reviews.llvm.org/D49685#1173720, @EugeneBi wrote:

> In https://reviews.llvm.org/D49685#1173640, @clayborg wrote:
>
> > We should never be loading the wrong shared libraries. The module spec we fill out must contain the UUID of the file are looking for. If it doesn't we have no chance of every really loading the right stuff.
>
>
> Well, that's what it does on my machine. So prepending sysroot *after* trying to load module without it will cause problems to me. Also, I assume that if you specified sysroot for a platform, we should not try to load solibs without it - these paths are just not a part of the platform.
>
> > I think doing this in the module list is not the right place. Why? Some platforms might have multiple sysroot to check. iOS for instance has a directory for each device that Xcode has connected to which can be checked. I am fine with adding this ability to>  lldb_private::Platform, but I would just do it in there. Try GetRemoteSharedModule with the spec, if it fails, try again after modifying the spec to prepend the sysroot path. Possible even just check the sysroot path + path first if m_sdk_sysroot is filled in. I don't see the need to change ModuleList itself.
>
> I do not see how this prepend sysroot could be done in the platform... Essentially, my fix is doing what you suggest: the platform supplies sysroot to the module list, and two different platforms (for two XCode devices, etc.) would supply different sysroots. What do I miss?


You would just move:

  auto resolved_module_spec(module_spec);
    if (sysroot != nullptr)
      resolved_module_spec.GetFileSpec().PrependPathComponent(sysroot);

into the code in Platform.cpp and do it all there.


Repository:
  rL LLVM

https://reviews.llvm.org/D49685