[llvm] [llvm-readobj] Resolve relocations to function symbols and addends in call-graph-info (PR #213156)

[Roland McGrath via llvm-commits]

================

----------------
frobtech wrote:

Does this account for REL-format in-band addends?  I didn't clearly see a test case exercising the REL case in all the permutations.  The nonzero addend cases are the most likely case where REL and RELA handling might produce different results if there is a bug, though it wouldn't hurt to just test all the cases in both REL and RELA incarnations.  Since the data is synthetic anyway, you can do that in lots of ways.  If you are trying to produce test data that is more representative of real-world cases, then arm(32) ELF is traditionally REL rather than RELA by default so you could use `EM_ARM` examples for `SHT_REL`; `EM_386` (x86-32) is also usually `SHT_REL`.

https://github.com/llvm/llvm-project/pull/213156