https://sourceware.org/bugzilla/show_bug.cgi?id=33401
--- Comment #5 from Indu Bhagat <indu.bhagat at oracle dot com> --- I see R_*_NONE relocations intermingled with other relas in the relocatable object automa_module.o. $ readelf -r automa_module.o Relocation section '.rela.sframe' at offset 0xc07b8 contains 78 entries: Offset Info Type Sym. Value Sym. Name + Addend ... 000000000224 005400000002 R_X86_64_PC32 0000000000000000 .text._ZN11Player[...] + 0 000000000224 000000000000 R_X86_64_NONE 0 000000000224 000000000000 R_X86_64_NONE 0 000000000238 005500000002 R_X86_64_PC32 0000000000000000 .text._ZN11StateC[...] + 0 000000000238 000000000000 R_X86_64_NONE 0 00000000024c 005600000002 R_X86_64_PC32 0000000000000000 .text._ZN11StateC[...] + 0 ... The above R_*_NONE relas are corresponding to some discarded symbols from StateClient.o (and there are other R_*_NONE that follow which are corresponding to some .text.* from StatePlaying.o). While the fix may be as straightforward as: skip R_*_NONE relas, and dont trip over them, I would really like to have a testcase for this. I tried to create one with three object files, each with a group/COMDAT section for the same foo symbol, hoping to see R_*_NONE, but this doesnt manifest there. Is there a way to share the .s files from the build ? -- You are receiving this mail because: You are on the CC list for the bug.
