> On 4/24/2013 2:12 PM, Corinna Vinschen wrote: >> >> On Apr 24 19:53, Samuel Thibault wrote: >>> >>> Corinna Vinschen, le Wed 24 Apr 2013 19:51:07 +0200, a écrit : >>>>>> >>>>>> /usr/include/cygwin/in6.h:75:8: error: redefinition of ‘struct >>>>>> in6_addr’ >>>>>> >>>>>> /usr/lib/gcc/i686-pc-cygwin/4.5.3/../../../../include/w32api/in6addr.h:17:16: >>>>>> note: originally defined here >>>> >>>> >>>> Don't know this file but this is a bug in config.h. struct in6_addr >>>> is defined in a system header, so it should not be gratuitously defined >>>> in a project header, especially if the definition collides with the >>>> right one from the system header. >>> >>> >>> Mmm, but here it's w32api definition vs cygwin definition. >> >> >> Maybe I didn't read it exactly, but the bottom line here is, when >> building for Cygwin, either use the POSIX network headers (preferredly), >> or use winsock.h. Mixing them is not supported. > > > Yeah, though this case is a bit buried, so it's not obvious. Also the code > is from 2008, so I expect it has suffered some bit rot over time with > respect to Cygwin. In any case, it looks like removing iphlpapi.h from > sysdep.c solves the problem.
Hey, I got it! It's not cygwin fault... It's the inclusion of <netinet/in.h> which is done for the first time in sysdep.h, AFTER the inclusions of the winapi32, where struct in6_addr is declared. Actually sysdep.c includes its header at the end of the inclusion list and sysdep.h in its turn has an inclusion for <netinet/in.h> . The solution is in the following patch: --- //ioxp/cygwin/usr/src/vpnc-0.5.3/sysdep.c mer nov 19 21:02:39 2008 +++ //ioxp/cygwin/usr/src/vpnc-0.5.3.1/sysdep.c mer apr 24 21:53:29 2013 @@ -28,6 +28,7 @@ #include <sys/socket.h> #include <net/if.h> +#include <netinet/in.h> #ifdef __sun__ #include <ctype.h> @@ -38,7 +39,6 @@ #include <sys/sockio.h> #include <signal.h> #include <stropts.h> -#include <netinet/in.h> #include <netinet/in_systm.h> #include <netinet/ip.h> #include <netinet/tcp.h> Basically the #include <netinet/in.h> has to be moved up a few lines to get it out of the #ifdef __sun__ ... #endif directive. This way the winapi32 headers don't declare struct in6_addr, I didn't check, but I guess they do so only #ifndef _CYGWIN_IN6_H, am I right? For ease of reading, here are the inclusions in sysdep.c as patched by me: #include <unistd.h> #include <fcntl.h> #include <stdlib.h> #include <stdio.h> #include <string.h> #include <syslog.h> #include <sys/ioctl.h> #include <errno.h> #include <sys/socket.h> #include <net/if.h> #include <netinet/in.h> #ifdef __sun__ #include <ctype.h> #include <sys/time.h> #include <sys/wait.h> #include <sys/types.h> #include <sys/socket.h> #include <sys/sockio.h> #include <signal.h> #include <stropts.h> #include <netinet/in_systm.h> #include <netinet/ip.h> #include <netinet/tcp.h> #endif #if defined(__CYGWIN__) #include <io.h> #include <w32api/windef.h> #include <w32api/winbase.h> #include <w32api/winnt.h> #include <w32api/winioctl.h> #include <w32api/iphlpapi.h> #include <w32api/iptypes.h> #include <w32api/winreg.h> #include <sys/cygwin.h> #endif #if defined(__DragonFly__) #include <net/tun/if_tun.h> #elif defined(__linux__) #include <linux/if_tun.h> #elif defined(__APPLE__) /* no header for tun */ #elif defined(__CYGWIN__) #include "tap-win32.h" #else #include <net/if_tun.h> #endif #include "sysdep.h" #if !defined(HAVE_VASPRINTF) || !defined(HAVE_ASPRINTF) || !defined(HAVE_ERROR) #include <stdarg.h> #endif #if defined(__sun__) extern char **environ; static int ip_fd = -1, muxid; #endif OK, so that's it, vpnc-0.5.3 will then compile like a charm. Ah, by the way, you must have perl installed and run ./enum2debug.pl isakmp.h >vpnc-debug.c 2>vpnc-debug.h right before to make. Thanks to all for attention, and once again sorry if I dared to post in the wrong mailing list. You must be patient I'm just a newcomer... Didn't mean to bother!! :-) BR, Max -- Problem reports: http://cygwin.com/problems.html FAQ: http://cygwin.com/faq/ Documentation: http://cygwin.com/docs.html Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
