From: mbuhl <mbuhl@openbsd.org>
Date: Fri, 2 Sep 2022 13:18:06 +0000 (+0000)
Subject: add the recvmmsg syscall that allows receiving multiple msghdrs at
X-Git-Url: http://artulab.com/gitweb/?a=commitdiff_plain;h=8945982be22ebdb35731babc914d40f1e79d8d3f;p=openbsd

add the recvmmsg syscall that allows receiving multiple msghdrs at
once. libc, man page, and regress parts to come.
With input from jca@, guenther@, bluhm@.
OK bluhm@
---

diff --git a/sys/kern/init_sysent.c b/sys/kern/init_sysent.c
index 7de53351014..ca0649b533a 100644
--- a/sys/kern/init_sysent.c
+++ b/sys/kern/init_sysent.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: init_sysent.c,v 1.243 2022/08/01 14:57:19 deraadt Exp $	*/
+/*	$OpenBSD: init_sysent.c,v 1.244 2022/09/02 13:18:06 mbuhl Exp $	*/
 
 /*
  * System call switch table.
@@ -264,10 +264,10 @@ const struct sysent sysent[] = {
 	    sys_unveil },			/* 114 = unveil */
 	{ 2, s(struct sys___realpath_args), 0,
 	    sys___realpath },			/* 115 = __realpath */
+	{ 5, s(struct sys_recvmmsg_args), SY_NOLOCK | 0,
+	    sys_recvmmsg },			/* 116 = recvmmsg */
 	{ 0, 0, 0,
-	    sys_nosys },			/* 116 = obsolete t32_gettimeofday */
-	{ 0, 0, 0,
-	    sys_nosys },			/* 117 = obsolete t32_getrusage */
+	    sys_nosys },			/* 117 = unimplemented sendmmsg */
 	{ 5, s(struct sys_getsockopt_args), 0,
 	    sys_getsockopt },			/* 118 = getsockopt */
 	{ 3, s(struct sys_thrkill_args), 0,
diff --git a/sys/kern/syscalls.c b/sys/kern/syscalls.c
index 7d1339fad1e..0749b50568d 100644
--- a/sys/kern/syscalls.c
+++ b/sys/kern/syscalls.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: syscalls.c,v 1.241 2022/08/01 14:57:19 deraadt Exp $	*/
+/*	$OpenBSD: syscalls.c,v 1.242 2022/09/02 13:18:06 mbuhl Exp $	*/
 
 /*
  * System call names.
@@ -136,8 +136,8 @@ const char *const syscallnames[] = {
 	"#113 (unimplemented fktrace)",		/* 113 = unimplemented fktrace */
 	"unveil",			/* 114 = unveil */
 	"__realpath",			/* 115 = __realpath */
-	"#116 (obsolete t32_gettimeofday)",		/* 116 = obsolete t32_gettimeofday */
-	"#117 (obsolete t32_getrusage)",		/* 117 = obsolete t32_getrusage */
+	"recvmmsg",			/* 116 = recvmmsg */
+	"#117 (unimplemented sendmmsg)",		/* 117 = unimplemented sendmmsg */
 	"getsockopt",			/* 118 = getsockopt */
 	"thrkill",			/* 119 = thrkill */
 	"readv",			/* 120 = readv */
diff --git a/sys/kern/syscalls.master b/sys/kern/syscalls.master
index cf4612e067b..eaf34dfabfe 100644
--- a/sys/kern/syscalls.master
+++ b/sys/kern/syscalls.master
@@ -1,4 +1,4 @@
-;	$OpenBSD: syscalls.master,v 1.229 2022/08/01 14:56:59 deraadt Exp $
+;	$OpenBSD: syscalls.master,v 1.230 2022/09/02 13:18:06 mbuhl Exp $
 ;	$NetBSD: syscalls.master,v 1.32 1996/04/23 10:24:21 mycroft Exp $
 
 ;	@(#)syscalls.master	8.2 (Berkeley) 1/13/94
@@ -244,8 +244,10 @@
 			    const char *permissions); }
 115	STD		{ int sys___realpath(const char *pathname, \
 			    char *resolved); }
-116	OBSOL		t32_gettimeofday
-117	OBSOL		t32_getrusage
+116	STD NOLOCK	{ int sys_recvmmsg(int s, struct mmsghdr *mmsg, \
+			    unsigned int vlen, unsigned int flags, \
+			    struct timespec *timeout); }
+117	UNIMPL		sendmmsg
 118	STD		{ int sys_getsockopt(int s, int level, int name, \
 			    void *val, socklen_t *avalsize); }
 119	STD		{ int sys_thrkill(pid_t tid, int signum, void *tcb); }
diff --git a/sys/kern/uipc_syscalls.c b/sys/kern/uipc_syscalls.c
index dbd6373285e..b68e8534048 100644
--- a/sys/kern/uipc_syscalls.c
+++ b/sys/kern/uipc_syscalls.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: uipc_syscalls.c,v 1.201 2022/08/14 01:58:28 jsg Exp $	*/
+/*	$OpenBSD: uipc_syscalls.c,v 1.202 2022/09/02 13:18:06 mbuhl Exp $	*/
 /*	$NetBSD: uipc_syscalls.c,v 1.19 1996/02/09 19:00:48 christos Exp $	*/
 
 /*
@@ -804,6 +804,140 @@ done:
 	return (error);
 }
 
+int
+sys_recvmmsg(struct proc *p, void *v, register_t *retval)
+{
+	struct sys_recvmmsg_args /* {
+		syscallarg(int)			s;
+		syscallarg(struct mmsghdr *)	mmsg;
+		syscallarg(unsigned int)	vlen;
+		syscallarg(unsigned int)	flags;
+		syscallarg(struct timespec *)	timeout;
+	} */ *uap = v;
+	struct mmsghdr mmsg, *mmsgp;
+	struct timespec ts, now;
+	struct iovec aiov[UIO_SMALLIOV], *uiov, *iov = aiov;
+	struct file *fp;
+	struct socket *so;
+	struct timespec *timeout;
+	size_t iovlen = UIO_SMALLIOV;
+	register_t retrec;
+	unsigned int vlen, dgrams;
+	int error = 0, flags, s;
+
+	s = SCARG(uap, s);
+	if ((error = getsock(p, s, &fp)))
+		return (error);
+	so = (struct socket *)fp->f_data;
+
+	timeout = SCARG(uap, timeout);
+	if (timeout != NULL) {
+		error = copyin(timeout, &ts, sizeof(ts));
+		if (error)
+			return error;
+#ifdef KTRACE
+		if (KTRPOINT(p, KTR_STRUCT))
+			ktrreltimespec(p, &ts);
+#endif
+		getnanotime(&now);
+		timespecadd(&now, &ts, &ts);
+	}
+
+	flags = SCARG(uap, flags);
+
+	/* Arbitrarily capped at 1024 datagrams. */
+	vlen = SCARG(uap, vlen);
+	if (vlen > 1024)
+		vlen = 1024;
+
+	mmsgp = SCARG(uap, mmsg);
+	for (dgrams = 0; dgrams < vlen;) {
+		error = copyin(&mmsgp[dgrams], &mmsg, sizeof(mmsg));
+		if (error)
+			break;
+
+		if (mmsg.msg_hdr.msg_iovlen > IOV_MAX) {
+			error = EMSGSIZE;
+			break;
+		}
+
+		if (mmsg.msg_hdr.msg_iovlen > iovlen) {
+			if (iov != aiov)
+				free(iov, M_IOV, iovlen *
+				    sizeof(struct iovec));
+
+			iovlen = mmsg.msg_hdr.msg_iovlen;
+			iov = mallocarray(iovlen, sizeof(struct iovec),
+			    M_IOV, M_WAITOK);
+		}
+
+		if (mmsg.msg_hdr.msg_iovlen > 0) {
+			error = copyin(mmsg.msg_hdr.msg_iov, iov,
+			    mmsg.msg_hdr.msg_iovlen * sizeof(struct iovec));
+			if (error)
+				break;
+		}
+
+		uiov = mmsg.msg_hdr.msg_iov;
+		mmsg.msg_hdr.msg_iov = iov;
+		mmsg.msg_hdr.msg_flags = flags;
+
+		error = recvit(p, s, &mmsg.msg_hdr, NULL, &retrec);
+		if (error) {
+			if (error == EAGAIN && dgrams > 0)
+				error = 0;
+			break;
+		}
+
+		if (dgrams == 0 && flags & MSG_WAITFORONE) {
+			flags &= ~MSG_WAITFORONE;
+			flags |= MSG_DONTWAIT;
+		}
+
+		mmsg.msg_hdr.msg_iov = uiov;
+		mmsg.msg_len = retrec;
+#ifdef KTRACE
+		if (KTRPOINT(p, KTR_STRUCT)) {
+			ktrmmsghdr(p, &mmsg);
+			if (mmsg.msg_hdr.msg_iovlen)
+				ktriovec(p, iov, mmsg.msg_hdr.msg_iovlen);
+		}
+#endif
+
+		error = copyout(&mmsg, &mmsgp[dgrams], sizeof(mmsg));
+		if (error)
+			break;
+
+		dgrams++;
+		if (mmsg.msg_hdr.msg_flags & MSG_OOB)
+			break;
+
+		if (timeout != NULL) {
+			getnanotime(&now);
+			timespecsub(&now, &ts, &now);
+			if (now.tv_sec > 0)
+				break;
+		}
+	}
+
+	if (iov != aiov)
+		free(iov, M_IOV, iovlen * sizeof(struct iovec));
+
+	*retval = dgrams;
+
+	/*
+	 * If we succeeded at least once, return 0, hopefully so->so_error
+	 * will catch it next time.
+	 */
+	if (error && dgrams > 0) {
+		so->so_error = error;
+		error = 0;
+	}
+
+	FRELE(fp, p);
+	return (error);
+}
+
 int
 recvit(struct proc *p, int s, struct msghdr *mp, caddr_t namelenp,
     register_t *retsize)
diff --git a/sys/sys/ktrace.h b/sys/sys/ktrace.h
index 90c8258de9c..d16ce8b8982 100644
--- a/sys/sys/ktrace.h
+++ b/sys/sys/ktrace.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: ktrace.h,v 1.41 2022/02/22 17:14:14 deraadt Exp $	*/
+/*	$OpenBSD: ktrace.h,v 1.42 2022/09/02 13:18:07 mbuhl Exp $	*/
 /*	$NetBSD: ktrace.h,v 1.12 1996/02/04 02:12:29 christos Exp $	*/
 
 /*
@@ -239,6 +239,8 @@ void    ktrstruct(struct proc *, const char *, const void *, size_t);
 	ktrstruct((p), "quota", (s), sizeof(struct dqblk))
 #define ktrmsghdr(p, s) \
 	ktrstruct(p, "msghdr", s, sizeof(struct msghdr))
+#define ktrmmsghdr(p, s) \
+	ktrstruct(p, "mmsghdr", s, sizeof(struct mmsghdr))
 #define ktriovec(p, s, count) \
 	ktrstruct(p, "iovec", s, (count) * sizeof(struct iovec))
 #define ktrcmsghdr(p, c, len) \
diff --git a/sys/sys/socket.h b/sys/sys/socket.h
index fe1b8f954c8..35758827b30 100644
--- a/sys/sys/socket.h
+++ b/sys/sys/socket.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: socket.h,v 1.102 2022/02/22 01:01:02 guenther Exp $	*/
+/*	$OpenBSD: socket.h,v 1.103 2022/09/02 13:18:07 mbuhl Exp $	*/
 /*	$NetBSD: socket.h,v 1.14 1996/02/09 18:25:36 christos Exp $	*/
 
 /*
@@ -490,6 +490,13 @@ struct msghdr {
 	int		msg_flags;	/* flags on received message */
 };
 
+struct mmsghdr {
+	struct msghdr msg_hdr;
+	unsigned int msg_len;
+};
+
+struct timespec;
+
 #define	MSG_OOB			0x1	/* process out-of-band data */
 #define	MSG_PEEK		0x2	/* peek at incoming message */
 #define	MSG_DONTROUTE		0x4	/* send without using routing tables */
@@ -502,6 +509,7 @@ struct msghdr {
 #define	MSG_MCAST		0x200	/* this message rec'd as multicast */
 #define	MSG_NOSIGNAL		0x400	/* do not send SIGPIPE */
 #define	MSG_CMSG_CLOEXEC	0x800	/* set FD_CLOEXEC on received fds */
+#define	MSG_WAITFORONE		0x1000	/* nonblocking but wait for one msg */
 
 /*
  * Header for ancillary data objects in msg_control buffer.
@@ -565,6 +573,8 @@ int	listen(int, int);
 ssize_t	recv(int, void *, size_t, int);
 ssize_t	recvfrom(int, void *, size_t, int, struct sockaddr *, socklen_t *);
 ssize_t	recvmsg(int, struct msghdr *, int);
+int	recvmmsg(int, struct mmsghdr *, unsigned int, unsigned int,
+	    struct timespec *);
 ssize_t	send(int, const void *, size_t, int);
 ssize_t	sendto(int, const void *,
 	    size_t, int, const struct sockaddr *, socklen_t);
diff --git a/sys/sys/syscall.h b/sys/sys/syscall.h
index ce5beabe24a..114853c3cae 100644
--- a/sys/sys/syscall.h
+++ b/sys/sys/syscall.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: syscall.h,v 1.240 2022/08/01 14:57:19 deraadt Exp $	*/
+/*	$OpenBSD: syscall.h,v 1.241 2022/09/02 13:18:07 mbuhl Exp $	*/
 
 /*
  * System call numbers.
@@ -352,8 +352,9 @@
 /* syscall: "__realpath" ret: "int" args: "const char *" "char *" */
 #define	SYS___realpath	115
 
-				/* 116 is obsolete t32_gettimeofday */
-				/* 117 is obsolete t32_getrusage */
+/* syscall: "recvmmsg" ret: "int" args: "int" "struct mmsghdr *" "unsigned int" "unsigned int" "struct timespec *" */
+#define	SYS_recvmmsg	116
+
 /* syscall: "getsockopt" ret: "int" args: "int" "int" "int" "void *" "socklen_t *" */
 #define	SYS_getsockopt	118
 
diff --git a/sys/sys/syscallargs.h b/sys/sys/syscallargs.h
index 244efc75638..18c22debe02 100644
--- a/sys/sys/syscallargs.h
+++ b/sys/sys/syscallargs.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: syscallargs.h,v 1.243 2022/08/01 14:57:19 deraadt Exp $	*/
+/*	$OpenBSD: syscallargs.h,v 1.244 2022/09/02 13:18:07 mbuhl Exp $	*/
 
 /*
  * System call argument lists.
@@ -595,6 +595,14 @@ struct sys___realpath_args {
 	syscallarg(char *) resolved;
 };
 
+struct sys_recvmmsg_args {
+	syscallarg(int) s;
+	syscallarg(struct mmsghdr *) mmsg;
+	syscallarg(unsigned int) vlen;
+	syscallarg(unsigned int) flags;
+	syscallarg(struct timespec *) timeout;
+};
+
 struct sys_getsockopt_args {
 	syscallarg(int) s;
 	syscallarg(int) level;
@@ -1302,6 +1310,7 @@ int	sys_sigsuspend(struct proc *, void *, register_t *);
 int	sys_sendsyslog(struct proc *, void *, register_t *);
 int	sys_unveil(struct proc *, void *, register_t *);
 int	sys___realpath(struct proc *, void *, register_t *);
+int	sys_recvmmsg(struct proc *, void *, register_t *);
 int	sys_getsockopt(struct proc *, void *, register_t *);
 int	sys_thrkill(struct proc *, void *, register_t *);
 int	sys_readv(struct proc *, void *, register_t *);