On Tue, Aug 23, 2016 at 4:27 PM, Jean Guyomarc'h <jean.guyoma...@openwide.fr> wrote: [...] > - if (efl_net_socket_fd_close_on_exec_get(o)) > - extra_flags |= SOCK_CLOEXEC; > - > - fd = socket(addr.ss_family, SOCK_STREAM | extra_flags, IPPROTO_TCP); > + fd = socket(addr.ss_family, SOCK_STREAM, IPPROTO_TCP);
sorry about the build break, but this is not a good solution for Linux. There is a race, if some other thread spawns a process now, the process will inherit fd. The best solution would be to #ifndef SOCK_CLOEXEC and keep both my code and your new code. Can you do that? > - if (efl_net_server_fd_close_on_exec_get(o)) > - flags |= SOCK_CLOEXEC; > - > addrlen = sizeof(addr); > -#ifdef HAVE_ACCEPT4 > - client = accept4(fd, (struct sockaddr *)&addr, &addrlen, flags); > -#else > client = accept(fd, (struct sockaddr *)&addr, &addrlen); > -#endif likewise, here we could leak the fd. That's why accept4() was introduced. > - fd = socket(addr.ss_family, SOCK_STREAM | extra_flags, IPPROTO_TCP); > + fd = socket(addr.ss_family, SOCK_STREAM, IPPROTO_TCP); same here. BTW, is there some build bot with such results or somewhere I could test MacOS to avoid breaking it? -- Gustavo Sverzut Barbieri -------------------------------------- Mobile: +55 (16) 99354-9890 ------------------------------------------------------------------------------ _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
