[llvm] r174574 - DWARFDebugFrame.cpp: Fix formatting on i686 hosts.

[Tim Northover]

> Thank you for the fix, Takumi. Semantically it's correct, since we
> don't currently support DWARF64 all such values are really 32-bits.

Aren't DWARF64 and ELF64 orthogonal? Most of the changes are probably
fine for DWARF32, but "InitialLocation" is a target address in both
DWARFs.

I've produced a binary linked with code above 32-bits (attached) where
this gives incorrect results (pc is  0x800000000 in the DWARF and
objdump). It's AArch64 because I couldn't convince x86_64 to produce a
.debug_frame section at all.

Cheers.

Tim.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: high_load
Type: application/octet-stream
Size: 67286 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130209/d27bcf5b/attachment.obj>