[PATCH] D50490: Restore correct x86_64 EH encodings in kernel code model

[Harald van Dijk via Phabricator via llvm-commits]

hvdijk added inline comments.


================
Comment at: llvm/trunk/lib/CodeGen/TargetLoweringObjectFileImpl.cpp:135
+        ((CM == CodeModel::Small || CM == CodeModel::Medium)
+         ? dwarf::DW_EH_PE_sdata8 : dwarf::DW_EH_PE_sdata4);
     } else {
----------------
hvdijk wrote:
> This looks like it's the wrong way around compared to the rest. I'm also not sure why the medium code model is treated like small in the PIC case, but like large in the non-PIC case, shouldn't it be the same for both?
Proposed to swap sdata4 and sdata8 in D102132. The different handling of the medium model between the PIC and non-PIC cases actually looks like it might be correct after all: in the PIC case, we do not store the typeinfo address directly.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D50490/new/

https://reviews.llvm.org/D50490