Jeffrey Walton <[email protected]> writes: > I'm seeing a fair amount of "ld: warning: relocation error: > R_AMD64_64" when linking from nettle-3.6 release tarball:
On what platform, and what version of gcc and linker? > gcc -g2 -O2 -m64 -march=native -fPIC -pthread -ggdb3 -Wall -W > -Wno-sign-compare -Wmissing-prototypes -Wmissing-declarations > -Wstrict-prototypes -Wpointer-arith -Wbad-function-cast > -Wnested-externs -L.. -L/opt/ssh/lib -m64 -Wl,-R,'RIGIN/../lib' > -Wl,-R,/opt/ssh/lib -Wl,-z,now rsa-sign.o io.o read_rsa_key.o \ > -lhogweed -lnettle -lgmp -ldl -lpthread -o rsa-sign > ld: warning: relocation error: R_AMD64_64: file ../getopt.o: symbol Is this command line correct? ../getopt.o doesn't appear on it, only in the error message. > optarg: external symbolic relocation against non-allocatable section optarg is a plain global variable (part of the traditional getopt interface). > .debug_info; cannot be processed at runtime: relocation ignored > > It looks like getopt was not built correctly. This all looks strange to me. For AMD64, Nettle is a plain C library, no linker or assembly magic. And command line options above look reasonable to me. Do your custom configure arguments matter, i.e., does it work better with a plain ./configure && make or ./configure CC='gcc -m64' && make? In general, if you have problems with bulding nettle, gmp, or any other package, please try configure with plain and documented options before bug reporting. If that still fails, it will be easier for developers to understand and reproduce the problem. And if works, it means it's something in your specific configure setting that triggers the breakage. Regards, /Niels -- Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677. Internet email is subject to wholesale government surveillance. _______________________________________________ nettle-bugs mailing list [email protected] http://lists.lysator.liu.se/mailman/listinfo/nettle-bugs
