Mismatch between IDs from minidump_stalkwalk and dump_syms

[Kalki70]

Hi,

I am trying to use google breakpad, but I am facing a strange issue. i am working in linux. I have my own library, my_lib.so, which I process with dump_syms and generates this:

$ dump_syms my_lib.so|head -2 MODULE Linux mips 3BB485681467218D36EB2FF02287096C0 my_lib.so INFO CODE_ID 6885B43B67148D2136EB2FF02287096C

I create the symbols directory with the appropiate subdirectories. I then generate a minidump for the program that uses a stripped version of my_lib.so, but when I try to process it with minidump_stackwalk:

0x77dce000 - 0x77e23fff my_lib.so ??? (WARNING: No symbols, my_lib.so, AC40136B433E5A68F66CCE8C2C2E6C250)

It is seaching for a differente ID, AC40136B433E5A68F66CCE8C2C2E6C250, so it does not find the symbols. Why the mismatch?

Knowing that it searches for AC40136B433E5A68F66CCE8C2C2E6C250 I manually changed the tree directory in symbols, to match that one, just to test. I also changed the id inside the my_lib.so.sym file, and then minidump_stalkwalk does not complain about not finding the symbols, but still I can't see the stack trace.

Any ideas about this mismatch?

by the way, if I run readelf -n over the original library and the stripped one, I get the same GNU BUILD ID.

I did not know about this group, so I also asked this yesterday in Stack Overflow :

https://stackoverflow.com/questions/72528236/mismatch-between-ids-from-minidump-stalkwalk-and-dump-syms

[eva DING]

Kalki70, 

I just met same issue and want to know if you found the root cause and solution. Appreciate your sharing, thanks in advance.

BTW, the issue happens occasionally not always on my side.

Eva