Re: [Libguestfs] [libnbd PATCH 3/6] macOS: Simple cloexec/nonblock fix

On Tue, Jul 13, 2021 at 11:26:05PM +0200, Martin Kletzander wrote: > diff --git a/fuzzing/libnbd-fuzz-wrapper.c b/fuzzing/libnbd-fuzz-wrapper.c > index 99a6d803258f..3d127e673e9e 100644 > --- a/fuzzing/libnbd-fuzz-wrapper.c > +++ b/fuzzing/libnbd-fuzz-wrapper.c > @@ -41,6 +41,34 @@ > static void client (int s); > static void server (int fd, int s); > > +static int > +get_socketpair (int domain, int type, int protocol, int *fds) > +{ > + int ret; > + > +#ifdef __APPLE__ > + size_t i; > +#else > + type |= SOCK_CLOEXEC; > +#endif > + > + ret = socketpair (domain, type, protocol, fds); > + > +#ifdef __APPLE__ > + if (ret == 0) { > + for (i = 0; i < 2; i++) { > + if (fcntl (fds[i], F_SETFD, FD_CLOEXEC) == -1) { > + close(fds[0]); > + close(fds[1]); > + return -1; > + } > + } > + } Is the fuzzer multithreaded? Or can we just blindly do the workaround ourselves and not worry about trying to use SOCK_CLOEXEC here? Also, it feels like some duplication; can we not reuse nbd_internal_socketpair from the fuzzer?