[PATCH] D74613: [DebugInfo] Fix printing CIE offset in EH FDEs.

[James Henderson via Phabricator via llvm-commits]

jhenderson added a comment.

Just making sure I understand: in .debug_frame, the cie_pointer field is always an absolute offset, whereas in .eh_frame, it is a relative value from the field, but we weren't making this distinction before, right? What does e.g. classic dwarfdump or GNU readelf display?



================
Comment at: llvm/include/llvm/DebugInfo/DWARF/DWARFDebugFrame.h:226
 public:
   // Each FDE has a CIE it's "linked to". Our FDE contains is constructed with
   // an offset to the CIE (provided by parsing the FDE header). The CIE itself
----------------
Could you fix the second sentence of this comment (possibly as a separate commit). I think the "contains" needs deleting.

It might also be worth saying whether the offset is relative or from the start of the section.


================
Comment at: llvm/include/llvm/DebugInfo/DWARF/DWARFDebugFrame.h:252
 private:
   /// The following fields are defined in section 6.4.1 of the DWARF standard v3
+  const uint64_t CIEPointer;
----------------
It may be worth making a note of the eh_frame/debug_frame difference here.


================
Comment at: llvm/lib/DebugInfo/DWARF/DWARFDebugFrame.cpp:317
+  OS << format("%08x %08x %08x", (uint32_t)Offset, (uint32_t)Length,
+               (uint32_t)CIEPointer)
+     << " FDE cie=";
----------------
I'm confused by the casting to 32-bits. I take it this code isn't DWARF64 ready yet?

Also, I'm not sure CIEPointer should be an unsigned type for .eh_frame at least. In the Ian Lance Taylor blog on the topic (https://www.airs.com/blog/archives/460), which I use as my .eh_frame "standard", it talks about how this value can be negative or positive. Do we have examples in tests which show the behaviour for positive and negative values?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D74613/new/

https://reviews.llvm.org/D74613