Extract Symbol address

[mbentin]

Is it possible to extract the symbol address from the DWARF?
It could be a really useful function for symbolicating.

[mattmassicotte]

Yes, it absolutely is. I have used CoreSymbolication do that in the past.

[mbentin]

@mattmassicotte Sorry for the dumb question, but is it visible in the swift wrapper?
Because in https://github.com/ChimeHQ/CoreSymbolication/blob/main/Sources/SwiftCoreSymbolication/Symbol.swift I cannot locate any address. And even in the C Headers I am not so sure if it is there?

[mattmassicotte]

You should be able to create a Symbolicator type by pointing it at the dwarf file within a dSYM and have it work. The Symbol type just wraps up the information about a particular symbol entry, and is independent of where the symbolic information comes from.

But you are bringing up a good point that perhaps some more documentation would be useful...

[mbentin]

My understanding was that when I get a crash report, in my case a MXCrashDiagnostic, I just read the load address "address", then I get the address of the symbol from "offsetIntoBinaryTextSegment".
Atos does the thing quite good, calculates the ASLR slide and search automagically in the dSYM for the symbol.
With CoreSymbolication, I am not so sure how I can do that. I can calculate the address I need to look for in the DWARF, but how do I match it?

So far so good, I loaded the dwarf in Symbolicator, going through, but there is no address?

for symbolicator in symbolicators {
         for owner in symbolicator.symbolOwners {
                owner.enumerateSymbols { symbol in
                       symbol.enumerateSoureInfo { info in
                            print("\(owner.baseAddress):\(symbol.name) -- \(info.sanitizedFilePath):\(info.lineNumber)")
                        }
               }
        }
}

And with a dwarfdump --arch arm64 myDSYMFile I also get the addresses inside the DWARF.
That's why I am a bit lost ^^

[mattmassicotte]

Ahhh yes, you need to consult the info.range, I believe! If you are looking for the range of the entire symbol, it is possible the library supports that, but we'd have to do some digging.

[mbentin]

Great thank you! I did not know it could be a range. After reviewing the DWARF doc it's clear https://dwarfstd.org/doc/DWARF5.pdf, page 51:

The base address of the scope for any of the debugging information entries listed above is given by either the DW_AT_low_pc attribute or the first address in the first range entry in the list of ranges given by the DW_AT_ranges attribute. If there is no such attribute, the base address is undefined.

So the first address of the range is the DW_AT_low_pc aka the address of the symbol

[mattmassicotte]

Glad you got it figured out!

I'm going to keep this issue open to cover adding a little documentation on how to use this.