[Lldb-commits] [PATCH] D114675: [lldb] [Target] Support fallback to file address in ReadMemory()

[Pavel Labath via Phabricator via lldb-commits]

labath requested changes to this revision.
labath added a reviewer: jingham.
labath added a comment.
This revision now requires changes to proceed.

I don't think this is the right solution to this problem.

The way I see it, there are two separate questions here:

- why is the module containing `hz` not loaded? The core file contains a proper memory address for it, does it not? We need to figure out why is that not being applied. I took a brief look, but I couldn't actually find the place where elf file modules are created and where their load addresses are set. I think the first step should be to figure out how this actually works (or why does it not work). Without a load address, a lot of things will not operate properly. If it helps you can look at ProcessMinidump::ReadModuleList to see how it works there, but be aware that this uses a different (I am not saying that it's better or worse) approach to create and load modules -- it does it directly, whereas the elf class seems to rely on DynamicLoaderStatic to do the loading.

- should "memory read" work on unloaded modules. I think this was deliberately disabled because the concept of "memory" only really makes sense on a "running process" (a core file is considered a running process here). However, your example kinda makes sense even on an unloaded module (a not-yet-started process if the typical situation where you get those). So, we could consider making this work as well. I'm adding Jim to see what he thinks about this. However, **if** we do decide to do that, then the right solution is to use the Target version of `ReadMemory` which accepts an Address object directly. D113098 <https://reviews.llvm.org/D113098> might be relevant/interesting here.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D114675/new/

https://reviews.llvm.org/D114675