[Lldb-commits] [PATCH] D124198: [LLDB][Unwind] Add stack scanning as fallback unwind plan if no symbol file is available.

[Greg Clayton via Phabricator via lldb-commits]

clayborg added a comment.

If we do find what looks like a return address, is there any validation done on the instruction before the return address to see if it is a function call instruction? That would be the best way to validate that something on the stack just doesn't look like a return address and is actually just a random function pointer. I would like to see validation on anything that looks like a return address like:

- make sure it doesn't point to the first address of a function since if it actually is a return address is should be in the middle of a function
- make sure the previous instruction was a function call to the function the current function from which we are unwinding if we have function bounds for the current frame
- if we don't have function bounds for the current function, we can benefit from knowing the function start address in the current frame if we can find the call instruction since that will help us parse the correct function prologue


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D124198/new/

https://reviews.llvm.org/D124198