Chris Lamb: > [...] > In particular: > > $ helpers/coll/objdump-info-helper \ > ./usr/lib/gcc-cross/mips-linux-gnu/8/* > > […] > > 19097 Filename: ./usr/lib/gcc-cross/mips-linux-gnu/8/libgo.a(bzip2.o) > 19098 Section-Headers: > 19099 Dynamic-Section: > 19100 SONAME °^F¬<8f>²^F<90><8f>µ^F<8c>^P > 19101 SONAME <88>$^E > 19102 RPATH %<8f>·^F<94>¯µ^F¤¯²^F¨¯°^F¬<8f>²^F<90><8f>µ^F<8c>^P > 19103 RPATH >
That indeed looks crazy on several levels! > > [https://gist.githubusercontent.com/lamby/7d08d1cdeee9fed41d888ec6c85d1863/raw] > > Is this a bug in objdump? > > > Regards, > We are using readelf these days as it copes better with foreign architecture binaries. Although, as you have probably guessed by now, we have not to renamed our internal tooling to reflect the "objdump -> readelf" migration. Does it also produce non-sense when you run "readelf -a -W" directly on the binaries? Thanks, ~Niels
