On Jan 19 15:18, Darren Tucker wrote: > On Sat, Jan 18, 2014 at 3:03 AM, Corinna Vinschen <vinschen at redhat.com> wrote: > [...] > > "signed - unsigned comparisons". The reason is that socklen_t > > is signed, but sizeof returns a size_t value (addrmatch.c), resp > > i was defined as unsigend type (canohost.c). > > committed upstream. > > [...]' > > GCC is unhappy: error: no return statement in function returning non-void > > [-Werror=return-type]: > [...] > > longjmp(kexjmp, 1); > > + return 0; /* Silence compiler warnings. */ > > This doesn't seem to be a problem on any other platforms. Should your > longjmp be marked __attribute__ ((__noreturn__)) ? If I'm looking at > the right thing it seems like it's not: > http://cygwin.com/cgi-bin/cvsweb.cgi/src/newlib/libc/include/setjmp.h?cvsroot=src Thanks, I applied a patch upstream. What about the patches from http://lists.mindrot.org/pipermail/openssh-unix-dev/2014-January/031999.html ? Thanks, Corinna -- Corinna Vinschen Cygwin Maintainer Red Hat -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 819 bytes Desc: not available URL: <http://lists.mindrot.org/pipermail/openssh-unix-dev/attachments/20140119/33122b2e/attachment.bin>