Re: [libvirt] [PATCH v3 7/9] tests: Build virpcimock on non-Linux

On Thu, 2018-05-03 at 12:54 +0200, Andrea Bolognani wrote: > There are only a couple issues preventing it from working on > other platform such as FreeBSD. Let's fix them. > > With the mocking in place, qemumemlocktest and qemuxml2xmltest > can finally succeed on FreeBSD. Too bad it also happens to break compilation on macOS duplicate symbol _rpl_lstat in: .libs/virpcimock.o ../gnulib/lib/.libs/libgnu.a(lstat.o) duplicate symbol _rpl_open in: .libs/virpcimock.o ../gnulib/lib/.libs/libgnu.a(open.o) duplicate symbol _rpl_stat in: .libs/virpcimock.o ../gnulib/lib/.libs/libgnu.a(stat.o) ld: 3 duplicate symbols for architecture x86_64 clang: error: linker command failed with exit code 1 (use -v to see invocation) make[2]: *** [virpcimock.la] Error 1 and on MinGW ../../tests/virpcimock.c: In function 'make_symlink': ../../tests/virpcimock.c:203:9: error: implicit declaration of function 'symlink'; did you mean 'unlink'? [-Werror=implicit-function-declaration] if (symlink(target, filepath) < 0) ^~~~~~~ unlink ../../tests/virpcimock.c:203:9: error: nested extern declaration of 'symlink' [-Werror=nested-externs] ../../tests/virpcimock.c: In function 'pci_read_file': ../../tests/virpcimock.c:228:5: error: implicit declaration of function 'bzero' [-Werror=implicit-function-declaration] bzero(buf, buf_size); ^~~~~ ../../tests/virpcimock.c:228:5: error: incompatible implicit declaration of built-in function 'bzero' [-Werror] In file included from ../gnulib/lib/fcntl.h:58:0, from ../../tests/virpcimock.c:27: ../../tests/virpcimock.c: At top level: ../../tests/virpcimock.c:964:1: error: redefinition of 'rpl_stat' stat(const char *path, struct stat *sb) ^ ../../tests/virpcimock.c:924:1: note: previous definition of 'rpl_stat' was here lstat(const char *path, struct stat *sb) ^ cc1: all warnings being treated as errors gmake[2]: *** [Makefile:5505: virpcimock.lo] Error 1 I'll look into solving the latter, but I have no intention of spending time on the former because I don't have access to macOS machines and debugging this kind of failure through Travis would be just too painful and time consuming.