nhrpd: implement next hop resolution protocol

This provides DMVPN support and integrates to strongSwan. Please read
README.nhrpd and README.kernel for more details.
diff --git a/nhrpd/zbuf.c b/nhrpd/zbuf.c
new file mode 100644
index 0000000..ead7cfd
--- /dev/null
+++ b/nhrpd/zbuf.c
@@ -0,0 +1,219 @@
+/* Stream/packet buffer API implementation
+ * Copyright (c) 2014-2015 Timo Teräs
+ *
+ * This file is free software: you may copy, redistribute and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 2 of the License, or
+ * (at your option) any later version.
+ */
+
+#define _GNU_SOURCE
+#include <string.h>
+#include <unistd.h>
+#include <errno.h>
+#include "zassert.h"
+#include "zbuf.h"
+#include "memory.h"
+#include "memtypes.h"
+#include "nhrpd.h"
+
+#define ERRNO_IO_RETRY(EN) (((EN) == EAGAIN) || ((EN) == EWOULDBLOCK) || ((EN) == EINTR))
+
+struct zbuf *zbuf_alloc(size_t size)
+{
+	struct zbuf *zb;
+
+	zb = XMALLOC(MTYPE_STREAM_DATA, sizeof(*zb) + size);
+	if (!zb)
+		return NULL;
+
+	zbuf_init(zb, zb+1, size, 0);
+	zb->allocated = 1;
+
+	return zb;
+}
+
+void zbuf_init(struct zbuf *zb, void *buf, size_t len, size_t datalen)
+{
+	*zb = (struct zbuf) {
+		.buf = buf,
+		.end = (uint8_t *)buf + len,
+		.head = buf,
+		.tail = (uint8_t *)buf + datalen,
+	};
+}
+
+void zbuf_free(struct zbuf *zb)
+{
+	if (zb->allocated)
+		XFREE(MTYPE_STREAM_DATA, zb);
+}
+
+void zbuf_reset(struct zbuf *zb)
+{
+	zb->head = zb->tail = zb->buf;
+	zb->error = 0;
+}
+
+void zbuf_reset_head(struct zbuf *zb, void *ptr)
+{
+	zassert((void*)zb->buf <= ptr && ptr <= (void*)zb->tail);
+	zb->head = ptr;
+}
+
+static void zbuf_remove_headroom(struct zbuf *zb)
+{
+	ssize_t headroom = zbuf_headroom(zb);
+	if (!headroom)
+		return;
+	memmove(zb->buf, zb->head, zbuf_used(zb));
+	zb->head -= headroom;
+	zb->tail -= headroom;
+}
+
+ssize_t zbuf_read(struct zbuf *zb, int fd, size_t maxlen)
+{
+	ssize_t r;
+
+	if (zb->error)
+		return -3;
+
+	zbuf_remove_headroom(zb);
+	if (maxlen > zbuf_tailroom(zb))
+		maxlen = zbuf_tailroom(zb);
+
+	r = read(fd, zb->tail, maxlen);
+	if (r > 0) zb->tail += r;
+	else if (r == 0) r = -2;
+	else if (r < 0 && ERRNO_IO_RETRY(errno)) r = 0;
+
+	return r;
+}
+
+ssize_t zbuf_write(struct zbuf *zb, int fd)
+{
+	ssize_t r;
+
+	if (zb->error)
+		return -3;
+
+	r = write(fd, zb->head, zbuf_used(zb));
+	if (r > 0) {
+		zb->head += r;
+		if (zb->head == zb->tail)
+			zbuf_reset(zb);
+	}
+	else if (r == 0) r = -2;
+	else if (r < 0 && ERRNO_IO_RETRY(errno)) r = 0;
+
+	return r;
+}
+
+ssize_t zbuf_recv(struct zbuf *zb, int fd)
+{
+	ssize_t r;
+
+	if (zb->error)
+		return -3;
+
+	zbuf_remove_headroom(zb);
+	r = recv(fd, zb->tail, zbuf_tailroom(zb), 0);
+	if (r > 0) zb->tail += r;
+	else if (r == 0) r = -2;
+	else if (r < 0 && ERRNO_IO_RETRY(errno)) r = 0;
+	return r;
+}
+
+ssize_t zbuf_send(struct zbuf *zb, int fd)
+{
+	ssize_t r;
+
+	if (zb->error)
+		return -3;
+
+	r = send(fd, zb->head, zbuf_used(zb), 0);
+	if (r >= 0)
+		zbuf_reset(zb);
+
+	return r;
+}
+
+void *zbuf_may_pull_until(struct zbuf *zb, const char *sep, struct zbuf *msg)
+{
+	size_t seplen = strlen(sep), len;
+	uint8_t *ptr;
+
+	ptr = memmem(zb->head, zbuf_used(zb), sep, seplen);
+	if (!ptr) return NULL;
+
+	len = ptr - zb->head + seplen;
+	zbuf_init(msg, zbuf_pulln(zb, len), len, len);
+	return msg->head;
+}
+
+void zbufq_init(struct zbuf_queue *zbq)
+{
+	*zbq = (struct zbuf_queue) {
+		.queue_head = LIST_INITIALIZER(zbq->queue_head),
+	};
+}
+
+void zbufq_reset(struct zbuf_queue *zbq)
+{
+	struct zbuf *buf, *bufn;
+
+	list_for_each_entry_safe(buf, bufn, &zbq->queue_head, queue_list) {
+		list_del(&buf->queue_list);
+		zbuf_free(buf);
+	}
+}
+
+void zbufq_queue(struct zbuf_queue *zbq, struct zbuf *zb)
+{
+	list_add_tail(&zb->queue_list, &zbq->queue_head);
+}
+
+int zbufq_write(struct zbuf_queue *zbq, int fd)
+{
+	struct iovec iov[16];
+	struct zbuf *zb, *zbn;
+	ssize_t r;
+	size_t iovcnt = 0;
+
+	list_for_each_entry_safe(zb, zbn, &zbq->queue_head, queue_list) {
+		iov[iovcnt++] = (struct iovec) {
+			.iov_base = zb->head,
+			.iov_len = zbuf_used(zb),
+		};
+		if (iovcnt >= ZEBRA_NUM_OF(iov))
+			break;
+	}
+
+	r = writev(fd, iov, iovcnt);
+	if (r < 0)
+		return r;
+
+	list_for_each_entry_safe(zb, zbn, &zbq->queue_head, queue_list) {
+		if (r < (ssize_t)zbuf_used(zb)) {
+			zb->head += r;
+			return 1;
+		}
+
+		r -= zbuf_used(zb);
+		list_del(&zb->queue_list);
+		zbuf_free(zb);
+	}
+
+	return 0;
+}
+
+void zbuf_copy(struct zbuf *zdst, struct zbuf *zsrc, size_t len)
+{
+	const void *src;
+	void *dst;
+
+	dst = zbuf_pushn(zdst, len);
+	src = zbuf_pulln(zsrc, len);
+	if (!dst || !src) return;
+	memcpy(dst, src, len);
+}