implement sendmmsg and recvmmsg

these are not pure syscall wrappers because they have to work around
kernel API bugs on 64-bit archs. the workarounds could probably be
made somewhat more efficient, but at the cost of more complexity. this
may be revisited later.

3 files changed, 55 insertions, 0 deletions

diff --git a/include/sys/socket.h b/include/sys/socket.h
index d7527911..b911d6ee 100644
--- a/include/sys/socket.h
+++ b/include/sys/socket.h
@@ -26,6 +26,17 @@ struct ucred
 	uid_t uid;
 	gid_t gid;
 };
+
+struct mmsghdr
+{
+	struct msghdr msg_hdr;
+	unsigned int  msg_len;
+};
+
+struct timespec;
+
+int sendmmsg (int, struct mmsghdr *, unsigned int, unsigned int);
+int recvmmsg (int, struct mmsghdr *, unsigned int, unsigned int, struct timespec *);
 #endif
 
 struct linger
diff --git a/src/network/recvmmsg.c b/src/network/recvmmsg.c
new file mode 100644
index 00000000..58b1b2f6
--- /dev/null
+++ b/src/network/recvmmsg.c
@@ -0,0 +1,15 @@
+#define _GNU_SOURCE
+#include <sys/socket.h>
+#include <limits.h>
+#include "syscall.h"
+
+int recvmmsg(int fd, struct mmsghdr *msgvec, unsigned int vlen, unsigned int flags, struct timespec *timeout)
+{
+#if LONG_MAX > INT_MAX
+	struct mmsghdr *mh = msgvec;
+	unsigned int i;
+	for (i = vlen; i; i--, mh++)
+		mh->msg_hdr.__pad1 = mh->msg_hdr.__pad2 = 0;
+#endif
+	return syscall_cp(SYS_recvmmsg, fd, msgvec, vlen, flags, timeout);
+}
diff --git a/src/network/sendmmsg.c b/src/network/sendmmsg.c
new file mode 100644
index 00000000..ff9f8618
--- /dev/null
+++ b/src/network/sendmmsg.c
@@ -0,0 +1,29 @@
+#define _GNU_SOURCE
+#include <sys/socket.h>
+#include <limits.h>
+#include <errno.h>
+#include "syscall.h"
+
+int sendmmsg(int fd, struct mmsghdr *msgvec, unsigned int vlen, unsigned int flags)
+{
+#if LONG_MAX > INT_MAX
+	/* Can't use the syscall directly because the kernel has the wrong
+	 * idea for the types of msg_iovlen, msg_controllen, and cmsg_len,
+	 * and the cmsg blocks cannot be modified in-place. */
+	int i;
+	if (vlen > IOV_MAX) vlen = IOV_MAX; /* This matches the kernel. */
+	for (i=0; i<vlen; i++) {
+		/* As an unfortunate inconsistency, the sendmmsg API uses
+		 * unsigned int for the resulting msg_len, despite sendmsg
+		 * returning ssize_t. However Linux limits the total bytes
+		 * sent by sendmsg to INT_MAX, so the assignment is safe. */
+		ssize_t r = sendmsg(fd, &msgvec[i].msg_hdr, flags);
+		if (r < 0) goto error;
+		msgvec[i].msg_len = r;
+	}
+error:
+	return i ? i : -1;
+#else
+	return syscall_cp(SYS_sendmmsg, fd, msgvec, vlen, flags);
+#endif
+}