X-Git-Url: http://nsz.repo.hu/git/?a=blobdiff_plain;f=src%2Fnetwork%2Fres_msend.c;h=1e76886aa12277668a9979fadeb4a79b5cd76697;hb=51d4669fb97782f6a66606da852b5afd49a08001;hp=d0e8e4815d0d87cfa44be2d845d0367913059cea;hpb=2683e267fa6c20d2e7a498a85f79a1dfc4301f23;p=musl

diff --git a/src/network/res_msend.c b/src/network/res_msend.c
index d0e8e481..1e76886a 100644
--- a/src/network/res_msend.c
+++ b/src/network/res_msend.c
@@ -1,5 +1,6 @@
 #include <sys/socket.h>
 #include <netinet/in.h>
+#include <netinet/tcp.h>
 #include <netdb.h>
 #include <arpa/inet.h>
 #include <stdint.h>
@@ -16,7 +17,9 @@
 
 static void cleanup(void *p)
 {
-	__syscall(SYS_close, (intptr_t)p);
+	struct pollfd *pfd = p;
+	for (int i=0; pfd[i].fd >= -1; i++)
+		if (pfd[i].fd >= 0) __syscall(SYS_close, pfd[i].fd);
 }
 
 static unsigned long mtime()
@@ -27,18 +30,61 @@ static unsigned long mtime()
 		+ ts.tv_nsec / 1000000;
 }
 
-int __res_msend(int nqueries, const unsigned char *const *queries,
-	const int *qlens, unsigned char *const *answers, int *alens, int asize)
+static int start_tcp(struct pollfd *pfd, int family, const void *sa, socklen_t sl, const unsigned char *q, int ql)
+{
+	struct msghdr mh = {
+		.msg_name = (void *)sa,
+		.msg_namelen = sl,
+		.msg_iovlen = 2,
+		.msg_iov = (struct iovec [2]){
+			{ .iov_base = (uint8_t[]){ ql>>8, ql }, .iov_len = 2 },
+			{ .iov_base = (void *)q, .iov_len = ql } }
+	};
+	int r;
+	int fd = socket(family, SOCK_STREAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
+	pfd->fd = fd;
+	pfd->events = POLLOUT;
+	if (!setsockopt(fd, IPPROTO_TCP, TCP_FASTOPEN_CONNECT,
+	    &(int){1}, sizeof(int))) {
+		r = sendmsg(fd, &mh, MSG_FASTOPEN|MSG_NOSIGNAL);
+		if (r == ql+2) pfd->events = POLLIN;
+		if (r >= 0) return r;
+		if (errno == EINPROGRESS) return 0;
+	}
+	r = connect(fd, sa, sl);
+	if (!r || errno == EINPROGRESS) return 0;
+	close(fd);
+	pfd->fd = -1;
+	return -1;
+}
+
+static void step_mh(struct msghdr *mh, size_t n)
+{
+	/* Adjust iovec in msghdr to skip first n bytes. */
+	while (mh->msg_iovlen && n >= mh->msg_iov->iov_len) {
+		n -= mh->msg_iov->iov_len;
+		mh->msg_iov++;
+		mh->msg_iovlen--;
+	}
+	if (!mh->msg_iovlen) return;
+	mh->msg_iov->iov_base = (char *)mh->msg_iov->iov_base + n;
+	mh->msg_iov->iov_len -= n;
+}
+
+/* Internal contract for __res_msend[_rc]: asize must be >=512, nqueries
+ * must be sufficiently small to be safe as VLA size. In practice it's
+ * either 1 or 2, anyway. */
+
+int __res_msend_rc(int nqueries, const unsigned char *const *queries,
+	const int *qlens, unsigned char *const *answers, int *alens, int asize,
+	const struct resolvconf *conf)
 {
 	int fd;
-	FILE *f, _f;
-	unsigned char _buf[256];
-	char line[64], *s, *z;
-	int timeout = 5000, attempts = 2, retry_interval, servfail_retry;
+	int timeout, attempts, retry_interval, servfail_retry;
 	union {
 		struct sockaddr_in sin;
 		struct sockaddr_in6 sin6;
-	} sa = {0}, ns[3] = {{0}};
+	} sa = {0}, ns[MAXNS] = {{0}};
 	socklen_t sl = sizeof sa.sin;
 	int nns = 0;
 	int family = AF_INET;
@@ -46,85 +92,63 @@ int __res_msend(int nqueries, const unsigned char *const *queries,
 	int next;
 	int i, j;
 	int cs;
-	struct pollfd pfd;
+	struct pollfd pfd[nqueries+2];
+	int qpos[nqueries], apos[nqueries];
+	unsigned char alen_buf[nqueries][2];
+	int r;
 	unsigned long t0, t1, t2;
-	struct address iplit;
 
 	pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, &cs);
 
-	/* Get nameservers from resolv.conf, fallback to localhost */
-	f = __fopen_rb_ca("/etc/resolv.conf", &_f, _buf, sizeof _buf);
-	if (!f) switch (errno) {
-	case ENOENT:
-	case ENOTDIR:
-	case EACCES:
-		goto no_resolv_conf;
-	default:
-		return -1;
-	}
-	for (nns=0; nns<3 && fgets(line, sizeof line, f); ) {
-		if (!strncmp(line, "options", 7) && isspace(line[7])) {
-			unsigned long x;
-			char *p, *z;
-			p = strstr(line, "timeout:");
-			if (p && isdigit(p[8])) {
-				p += 8;
-				x = strtoul(p, &z, 10);
-				if (z != p) timeout = x < 30 ? x*1000 : 30000;
-			}
-			p = strstr(line, "attempts:");
-			if (p && isdigit(p[9])) {
-				p += 9;
-				x = strtoul(p, &z, 10);
-				if (z != p) attempts = x < 10 ? x : 10;
-				if (!attempts) attempts = 1;
-			}
-		}
-		if (strncmp(line, "nameserver", 10) || !isspace(line[10]))
-			continue;
-		for (s=line+11; isspace(*s); s++);
-		for (z=s; *z && !isspace(*z); z++);
-		*z=0;
-
-		if (__lookup_ipliteral(&iplit, s, AF_UNSPEC)>0) {
-			if (iplit.family == AF_INET) {
-				memcpy(&ns[nns].sin.sin_addr, iplit.addr, 4);
-				ns[nns].sin.sin_port = htons(53);
-				ns[nns++].sin.sin_family = AF_INET;
-			} else {
-				sl = sizeof sa.sin6;
-				memcpy(&ns[nns].sin6.sin6_addr, iplit.addr, 16);
-				ns[nns].sin6.sin6_port = htons(53);
-				ns[nns].sin6.sin6_scope_id = iplit.scopeid;
-				ns[nns++].sin6.sin6_family = family = AF_INET6;
-			}
+	timeout = 1000*conf->timeout;
+	attempts = conf->attempts;
+
+	for (nns=0; nns<conf->nns; nns++) {
+		const struct address *iplit = &conf->ns[nns];
+		if (iplit->family == AF_INET) {
+			memcpy(&ns[nns].sin.sin_addr, iplit->addr, 4);
+			ns[nns].sin.sin_port = htons(53);
+			ns[nns].sin.sin_family = AF_INET;
+		} else {
+			sl = sizeof sa.sin6;
+			memcpy(&ns[nns].sin6.sin6_addr, iplit->addr, 16);
+			ns[nns].sin6.sin6_port = htons(53);
+			ns[nns].sin6.sin6_scope_id = iplit->scopeid;
+			ns[nns].sin6.sin6_family = family = AF_INET6;
 		}
 	}
-	__fclose_ca(f);
-no_resolv_conf:
-	if (!nns) {
-		ns[0].sin.sin_family = AF_INET;
-		ns[0].sin.sin_port = htons(53);
-		ns[0].sin.sin_addr.s_addr = htonl(0x7f000001);
-		nns=1;
-	}
 
 	/* Get local address and open/bind a socket */
-	sa.sin.sin_family = family;
 	fd = socket(family, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
 
 	/* Handle case where system lacks IPv6 support */
 	if (fd < 0 && family == AF_INET6 && errno == EAFNOSUPPORT) {
+		for (i=0; i<nns && conf->ns[nns].family == AF_INET6; i++);
+		if (i==nns) {
+			pthread_setcancelstate(cs, 0);
+			return -1;
+		}
 		fd = socket(AF_INET, SOCK_DGRAM|SOCK_CLOEXEC|SOCK_NONBLOCK, 0);
 		family = AF_INET;
+		sl = sizeof sa.sin;
+	}
+	sa.sin.sin_family = family;
+	if (fd < 0 || bind(fd, (void *)&sa, sl) < 0) {
+		if (fd >= 0) close(fd);
+		pthread_setcancelstate(cs, 0);
+		return -1;
 	}
-	if (fd < 0 || bind(fd, (void *)&sa, sl) < 0) return -1;
 
 	/* Past this point, there are no errors. Each individual query will
 	 * yield either no reply (indicated by zero length) or an answer
 	 * packet which is up to the caller to interpret. */
 
-	pthread_cleanup_push(cleanup, (void *)(intptr_t)fd);
+	for (i=0; i<nqueries; i++) pfd[i].fd = -1;
+	pfd[nqueries].fd = fd;
+	pfd[nqueries].events = POLLIN;
+	pfd[nqueries+1].fd = -2;
+
+	pthread_cleanup_push(cleanup, pfd);
 	pthread_setcancelstate(cs, 0);
 
 	/* Convert any IPv4 addresses in a mixed environment to v4-mapped */
@@ -144,14 +168,17 @@ no_resolv_conf:
 
 	memset(alens, 0, sizeof *alens * nqueries);
 
-	pfd.fd = fd;
-	pfd.events = POLLIN;
 	retry_interval = timeout / attempts;
 	next = 0;
 	t0 = t2 = mtime();
 	t1 = t2 - retry_interval;
 
 	for (; t2-t0 < timeout; t2=mtime()) {
+		/* This is the loop exit condition: that all queries
+		 * have an accepted answer. */
+		for (i=0; i<nqueries && alens[i]>0; i++);
+		if (i==nqueries) break;
+
 		if (t2-t1 >= retry_interval) {
 			/* Query all configured namservers in parallel */
 			for (i=0; i<nqueries; i++)
@@ -165,9 +192,10 @@ no_resolv_conf:
 		}
 
 		/* Wait for a response, or until time to retry */
-		if (poll(&pfd, 1, t1+retry_interval-t2) <= 0) continue;
+		if (poll(pfd, nqueries+1, t1+retry_interval-t2) <= 0) continue;
 
-		while ((rlen = recvfrom(fd, answers[next], asize, 0,
+		while (next < nqueries &&
+		  (rlen = recvfrom(fd, answers[next], asize, 0,
 		  (void *)&sa, (socklen_t[1]){sl})) >= 0) {
 
 			/* Ignore non-identifiable packets */
@@ -208,11 +236,79 @@ no_resolv_conf:
 			else
 				memcpy(answers[i], answers[next], rlen);
 
-			if (next == nqueries) goto out;
+			/* Ignore further UDP if all slots full or TCP-mode */
+			if (next == nqueries) pfd[nqueries].events = 0;
+
+			/* If answer is truncated (TC bit), fallback to TCP */
+			if (answers[i][2] & 2) {
+				alens[i] = -1;
+				pthread_setcancelstate(PTHREAD_CANCEL_DISABLE, 0);
+				r = start_tcp(pfd+i, family, ns+j, sl, queries[i], qlens[i]);
+				pthread_setcancelstate(cs, 0);
+				if (r >= 0) {
+					qpos[i] = r;
+					apos[i] = 0;
+				}
+				continue;
+			}
+		}
+
+		for (i=0; i<nqueries; i++) if (pfd[i].revents & POLLOUT) {
+			struct msghdr mh = {
+				.msg_iovlen = 2,
+				.msg_iov = (struct iovec [2]){
+					{ .iov_base = (uint8_t[]){ qlens[i]>>8, qlens[i] }, .iov_len = 2 },
+					{ .iov_base = (void *)queries[i], .iov_len = qlens[i] } }
+			};
+			step_mh(&mh, qpos[i]);
+			r = sendmsg(pfd[i].fd, &mh, MSG_NOSIGNAL);
+			if (r < 0) goto out;
+			qpos[i] += r;
+			if (qpos[i] == qlens[i]+2)
+				pfd[i].events = POLLIN;
+		}
+
+		for (i=0; i<nqueries; i++) if (pfd[i].revents & POLLIN) {
+			struct msghdr mh = {
+				.msg_iovlen = 2,
+				.msg_iov = (struct iovec [2]){
+					{ .iov_base = alen_buf[i], .iov_len = 2 },
+					{ .iov_base = answers[i], .iov_len = asize } }
+			};
+			step_mh(&mh, apos[i]);
+			r = recvmsg(pfd[i].fd, &mh, 0);
+			if (r < 0) goto out;
+			apos[i] += r;
+			if (apos[i] < 2) continue;
+			int alen = alen_buf[i][0]*256 + alen_buf[i][1];
+			if (alen < 13) goto out;
+			if (apos[i] < alen+2 && apos[i] < asize+2)
+				continue;
+			int rcode = answers[i][3] & 15;
+			if (rcode != 0 && rcode != 3)
+				goto out;
+
+			/* Storing the length here commits the accepted answer.
+			 * Immediately close TCP socket so as not to consume
+			 * resources we no longer need. */
+			alens[i] = alen;
+			__syscall(SYS_close, pfd[i].fd);
+			pfd[i].fd = -1;
 		}
 	}
 out:
 	pthread_cleanup_pop(1);
 
+	/* Disregard any incomplete TCP results */
+	for (i=0; i<nqueries; i++) if (alens[i]<0) alens[i] = 0;
+
 	return 0;
 }
+
+int __res_msend(int nqueries, const unsigned char *const *queries,
+	const int *qlens, unsigned char *const *answers, int *alens, int asize)
+{
+	struct resolvconf conf;
+	if (__get_resolv_conf(&conf, 0, 0) < 0) return -1;
+	return __res_msend_rc(nqueries, queries, qlens, answers, alens, asize, &conf);
+}