[DebugInfo] Change default value of FDEPointerEncoding

[maksfb]

Summary

If the encoding is not specified in CIE augmentation string, then it
should be DW_EH_PE_absptr instead of DW_EH_PE_omit.

[MaskRay]

The most relevant document about .eh_frame I can find is http://refspecs.linuxbase.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/ehframechpt.html , which does not say what the default encoding for FDE pointers should be.

But the http://www.nongnu.org/libunwind/ implementation provides a default depending on address size http://git.savannah.gnu.org/gitweb/?p=libunwind.git;a=blob;f=src/dwarf/Gfde.c;h=9250b895eabebdaa964ab3ddefdaece8100593c6;hb=HEAD#l64
DW_EH_PE_absptr seems to do the same thing.

[MaskRay]

Just curious, in which case the R (in augmentation string) is missing?

[rafauler]

@MaskRay this was originally reported here: https://github.com/facebookincubator/BOLT/issues/23

Apparently, this happens in binaries generated by icc.

llvm/trunk/lib/DebugInfo/DWARF/DWARFDebugFrame.cpp

if (IsCIE) {
       Data.setAddressSize(AddressSize);
       uint8_t SegmentDescriptorSize = Version < 4 ? 0 : Data.getU8(&Offset);
       uint64_t CodeAlignmentFactor = Data.getULEB128(&Offset);
       int64_t DataAlignmentFactor = Data.getSLEB128(&Offset);
       uint64_t ReturnAddressRegister = Data.getULEB128(&Offset);
 
       // Parse the augmentation data for EH CIEs
       StringRef AugmentationData("");
-      uint32_t FDEPointerEncoding = DW_EH_PE_omit;
+      uint32_t FDEPointerEncoding = DW_EH_PE_absptr;
       uint32_t LSDAPointerEncoding = DW_EH_PE_omit;
       Optional<uint64_t> Personality;
       Optional<uint32_t> PersonalityEncoding;
       if (IsEH) {
         Optional<uint64_t> AugmentationLength;
         uint32_t StartAugmentationOffset;
         uint32_t EndAugmentationOffset;