NetBSD/sys/compat/linux/common/linux_socket.c

1700 lines
38 KiB
C

/* $NetBSD: linux_socket.c,v 1.57 2005/09/24 15:51:03 christos Exp $ */
/*-
* Copyright (c) 1995, 1998 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Frank van der Linden and Eric Haszlakiewicz.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the NetBSD
* Foundation, Inc. and its contributors.
* 4. Neither the name of The NetBSD Foundation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
* TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
* POSSIBILITY OF SUCH DAMAGE.
*/
/*
* Functions in multiarch:
* linux_sys_socketcall : linux_socketcall.c
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: linux_socket.c,v 1.57 2005/09/24 15:51:03 christos Exp $");
#if defined(_KERNEL_OPT)
#include "opt_inet.h"
#endif
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/systm.h>
#include <sys/buf.h>
#include <sys/malloc.h>
#include <sys/ioctl.h>
#include <sys/tty.h>
#include <sys/file.h>
#include <sys/filedesc.h>
#include <sys/select.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/domain.h>
#include <net/if.h>
#include <net/if_dl.h>
#include <net/if_types.h>
#include <netinet/in.h>
#include <netinet/tcp.h>
#include <sys/mount.h>
#include <sys/proc.h>
#include <sys/vnode.h>
#include <sys/device.h>
#include <sys/protosw.h>
#include <sys/mbuf.h>
#include <sys/syslog.h>
#include <sys/exec.h>
#include <sys/sa.h>
#include <sys/syscallargs.h>
#ifdef INET6
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#endif
#include <compat/sys/socket.h>
#include <compat/linux/common/linux_types.h>
#include <compat/linux/common/linux_util.h>
#include <compat/linux/common/linux_signal.h>
#include <compat/linux/common/linux_ioctl.h>
#include <compat/linux/common/linux_socket.h>
#include <compat/linux/common/linux_socketcall.h>
#include <compat/linux/common/linux_sockio.h>
#include <compat/linux/linux_syscallargs.h>
#ifdef DEBUG_LINUX
#define DPRINTF(a) uprintf a
#else
#define DPRINTF(a)
#endif
/*
* The calls in this file are entered either via the linux_socketcall()
* interface or, on the Alpha, as individual syscalls. The
* linux_socketcall function does any massaging of arguments so that all
* the calls in here need not think that they are anything other
* than a normal syscall.
*/
static int linux_to_bsd_domain __P((int));
static int bsd_to_linux_domain __P((int));
int linux_to_bsd_sopt_level __P((int));
int linux_to_bsd_so_sockopt __P((int));
int linux_to_bsd_ip_sockopt __P((int));
int linux_to_bsd_tcp_sockopt __P((int));
int linux_to_bsd_udp_sockopt __P((int));
int linux_getifhwaddr __P((struct proc *, register_t *, u_int, void *));
static int linux_sa_get __P((struct proc *, int, caddr_t *, struct sockaddr **,
const struct osockaddr *, int *));
static int linux_sa_put __P((struct osockaddr *osa));
static int linux_to_bsd_msg_flags __P((int));
static int bsd_to_linux_msg_flags __P((int));
static const int linux_to_bsd_domain_[LINUX_AF_MAX] = {
AF_UNSPEC,
AF_UNIX,
AF_INET,
AF_CCITT, /* LINUX_AF_AX25 */
AF_IPX,
AF_APPLETALK,
-1, /* LINUX_AF_NETROM */
-1, /* LINUX_AF_BRIDGE */
-1, /* LINUX_AF_ATMPVC */
AF_CCITT, /* LINUX_AF_X25 */
AF_INET6,
-1, /* LINUX_AF_ROSE */
AF_DECnet,
-1, /* LINUX_AF_NETBEUI */
-1, /* LINUX_AF_SECURITY */
pseudo_AF_KEY,
AF_ROUTE, /* LINUX_AF_NETLINK */
-1, /* LINUX_AF_PACKET */
-1, /* LINUX_AF_ASH */
-1, /* LINUX_AF_ECONET */
-1, /* LINUX_AF_ATMSVC */
AF_SNA,
/* rest up to LINUX_AF_MAX-1 is not allocated */
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
};
static const int bsd_to_linux_domain_[AF_MAX] = {
LINUX_AF_UNSPEC,
LINUX_AF_UNIX,
LINUX_AF_INET,
-1, /* AF_IMPLINK */
-1, /* AF_PUP */
-1, /* AF_CHAOS */
-1, /* AF_NS */
-1, /* AF_ISO */
-1, /* AF_ECMA */
-1, /* AF_DATAKIT */
LINUX_AF_AX25, /* AF_CCITT */
LINUX_AF_SNA,
LINUX_AF_DECnet,
-1, /* AF_DLI */
-1, /* AF_LAT */
-1, /* AF_HYLINK */
LINUX_AF_APPLETALK,
LINUX_AF_NETLINK,
-1, /* AF_LINK */
-1, /* AF_XTP */
-1, /* AF_COIP */
-1, /* AF_CNT */
-1, /* pseudo_AF_RTIP */
LINUX_AF_IPX,
LINUX_AF_INET6,
-1, /* pseudo_AF_PIP */
-1, /* AF_ISDN */
-1, /* AF_NATM */
-1, /* AF_ARP */
LINUX_pseudo_AF_KEY,
-1, /* pseudo_AF_HDRCMPLT */
};
static const int bsd_to_linux_msg_flags_[] = {
MSG_OOB, LINUX_MSG_OOB,
MSG_PEEK, LINUX_MSG_PEEK,
MSG_DONTROUTE, LINUX_MSG_DONTROUTE,
MSG_EOR, LINUX_MSG_EOR,
MSG_TRUNC, LINUX_MSG_TRUNC,
MSG_CTRUNC, LINUX_MSG_CTRUNC,
MSG_WAITALL, LINUX_MSG_WAITALL,
MSG_DONTWAIT, LINUX_MSG_DONTWAIT,
MSG_BCAST, 0, /* not supported, clear */
MSG_MCAST, 0, /* not supported, clear */
-1, /* not supp */ LINUX_MSG_PROBE,
-1, /* not supp */ LINUX_MSG_FIN,
-1, /* not supp */ LINUX_MSG_SYN,
-1, /* not supp */ LINUX_MSG_CONFIRM,
-1, /* not supp */ LINUX_MSG_RST,
-1, /* not supp */ LINUX_MSG_ERRQUEUE,
-1, /* not supp */ LINUX_MSG_NOSIGNAL,
-1, /* not supp */ LINUX_MSG_MORE,
};
/*
* Convert between Linux and BSD socket domain values
*/
static int
linux_to_bsd_domain(ldom)
int ldom;
{
if (ldom < 0 || ldom >= LINUX_AF_MAX)
return (-1);
return linux_to_bsd_domain_[ldom];
}
/*
* Convert between BSD and Linux socket domain values
*/
static int
bsd_to_linux_domain(bdom)
int bdom;
{
if (bdom < 0 || bdom >= AF_MAX)
return (-1);
return bsd_to_linux_domain_[bdom];
}
static int
linux_to_bsd_msg_flags(lflag)
int lflag;
{
int i, lfl, bfl;
int bflag = 0;
if (lflag == 0)
return (0);
for(i=0; i < sizeof(bsd_to_linux_msg_flags_)/
sizeof(bsd_to_linux_msg_flags_[0])/2; i += 2) {
bfl = bsd_to_linux_msg_flags_[i];
lfl = bsd_to_linux_msg_flags_[i+1];
if (lfl == 0)
continue;
if (lflag & lfl) {
if (bfl < 0)
return (-1);
bflag |= bfl;
}
}
return (bflag);
}
static int
bsd_to_linux_msg_flags(bflag)
int bflag;
{
int i, lfl, bfl;
int lflag = 0;
if (bflag == 0)
return (0);
for(i=0; i < sizeof(bsd_to_linux_msg_flags_)/
sizeof(bsd_to_linux_msg_flags_[0])/2; i += 2) {
bfl = bsd_to_linux_msg_flags_[i];
lfl = bsd_to_linux_msg_flags_[i+1];
if (bfl <= 0)
continue;
if (bflag & bfl) {
if (lfl < 0)
return (-1);
lflag |= lfl;
}
}
return (lflag);
}
int
linux_sys_socket(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_socket_args /* {
syscallarg(int) domain;
syscallarg(int) type;
syscallarg(int) protocol;
} */ *uap = v;
struct sys_socket_args bsa;
int error;
SCARG(&bsa, protocol) = SCARG(uap, protocol);
SCARG(&bsa, type) = SCARG(uap, type);
SCARG(&bsa, domain) = linux_to_bsd_domain(SCARG(uap, domain));
if (SCARG(&bsa, domain) == -1)
return EINVAL;
error = sys_socket(l, &bsa, retval);
#ifdef INET6
/*
* Linux AF_INET6 socket has IPV6_V6ONLY setsockopt set to 0 by
* default and some apps depend on this. So, set V6ONLY to 0
* for Linux apps if the sysctl value is set to 1.
*/
if (!error && ip6_v6only && SCARG(&bsa, domain) == PF_INET6) {
struct proc *p = l->l_proc;
struct file *fp;
if (getsock(p->p_fd, *retval, &fp) == 0) {
struct mbuf *m;
m = m_get(M_WAIT, MT_SOOPTS);
m->m_len = sizeof(int);
*mtod(m, int *) = 0;
/* ignore error */
(void) sosetopt((struct socket *)fp->f_data,
IPPROTO_IPV6, IPV6_V6ONLY, m);
FILE_UNUSE(fp, p);
}
}
#endif
return (error);
}
int
linux_sys_socketpair(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_socketpair_args /* {
syscallarg(int) domain;
syscallarg(int) type;
syscallarg(int) protocol;
syscallarg(int *) rsv;
} */ *uap = v;
struct sys_socketpair_args bsa;
SCARG(&bsa, domain) = linux_to_bsd_domain(SCARG(uap, domain));
if (SCARG(&bsa, domain) == -1)
return EINVAL;
SCARG(&bsa, type) = SCARG(uap, type);
SCARG(&bsa, protocol) = SCARG(uap, protocol);
SCARG(&bsa, rsv) = SCARG(uap, rsv);
return sys_socketpair(l, &bsa, retval);
}
int
linux_sys_sendto(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_sendto_args /* {
syscallarg(int) s;
syscallarg(void *) msg;
syscallarg(int) len;
syscallarg(int) flags;
syscallarg(struct osockaddr *) to;
syscallarg(int) tolen;
} */ *uap = v;
struct proc *p = l->l_proc;
struct sys_sendto_args bsa;
int tolen;
SCARG(&bsa, s) = SCARG(uap, s);
SCARG(&bsa, buf) = SCARG(uap, msg);
SCARG(&bsa, len) = (size_t) SCARG(uap, len);
SCARG(&bsa, flags) = SCARG(uap, flags);
tolen = SCARG(uap, tolen);
if (SCARG(uap, to)) {
struct sockaddr *sa;
int error;
caddr_t sg = stackgap_init(p, 0);
error = linux_sa_get(p, SCARG(uap, s), &sg, &sa,
SCARG(uap, to), &tolen);
if (error)
return (error);
SCARG(&bsa, to) = sa;
} else
SCARG(&bsa, to) = NULL;
SCARG(&bsa, tolen) = tolen;
return (sys_sendto(l, &bsa, retval));
}
int
linux_sys_sendmsg(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_sendmsg_args /* {
syscallarg(int) s;
syscallarg(struct msghdr *) msg;
syscallarg(u_int) flags;
} */ *uap = v;
struct proc *p = l->l_proc;
struct msghdr msg;
int error;
struct iovec aiov[UIO_SMALLIOV], *iov;
caddr_t sg = 0;
int bflags;
u_int8_t *control=NULL;
error = copyin(SCARG(uap, msg), (caddr_t)&msg, sizeof(msg));
if (error)
return (error);
if ((unsigned int)msg.msg_iovlen > UIO_SMALLIOV) {
if ((unsigned int)msg.msg_iovlen > IOV_MAX)
return (EMSGSIZE);
iov = malloc(sizeof(struct iovec) * msg.msg_iovlen,
M_IOV, M_WAITOK);
} else
iov = aiov;
if ((unsigned int)msg.msg_iovlen > 0) {
error = copyin((caddr_t)msg.msg_iov, (caddr_t)iov,
(size_t)(msg.msg_iovlen * sizeof(struct iovec)));
if (error)
goto done;
}
msg.msg_iov = iov;
msg.msg_flags = 0;
/* Convert the sockaddr if necessary */
if (msg.msg_name) {
struct sockaddr *sa;
sg = stackgap_init(p, 0);
error = linux_sa_get(p, SCARG(uap, s), &sg, &sa,
(struct osockaddr *) msg.msg_name, &msg.msg_namelen);
if (error)
goto done;
msg.msg_name = sa;
}
/*
* Translate message flags.
*/
bflags = linux_to_bsd_msg_flags(SCARG(uap, flags));
if (bflags < 0) {
/* Some supported flag */
error = EINVAL;
goto done;
}
/*
* Handle cmsg if there is any.
*/
if (CMSG_FIRSTHDR(&msg)) {
struct cmsghdr cmsg, *cc;
int changed = 0;
ssize_t resid = msg.msg_controllen;
size_t clen, cidx = 0, cspace;
/*
* Limit the size even more than what sockargs() would do,
* We need to fit into stackgap space.
*/
if (msg.msg_controllen > (STACKGAPLEN / 2)) {
/* Sorry guys! */
error = EMSGSIZE;
goto done;
}
control = malloc((clen = msg.msg_controllen), M_TEMP, M_WAITOK);
if (!control) {
error = ENOMEM;
goto done;
}
cc = CMSG_FIRSTHDR(&msg);
do {
error = copyin(cc, &cmsg, sizeof(cmsg));
if (error)
goto done;
/*
* Sanity check the control message length.
*/
if (cmsg.cmsg_len > resid
|| cmsg.cmsg_len < sizeof(struct cmsghdr)) {
error = EINVAL;
goto done;
}
/*
* Refuse unsupported control messages, and
* translate fields as appropriate.
*/
switch (cmsg.cmsg_level) {
case LINUX_SOL_SOCKET:
/* It only differs on some archs */
if (LINUX_SOL_SOCKET != SOL_SOCKET) {
cmsg.cmsg_level = SOL_SOCKET;
changed = 1;
}
switch(cmsg.cmsg_type) {
case LINUX_SCM_RIGHTS:
/* Linux SCM_RIGHTS is same as NetBSD */
break;
default:
/* other types not supported */
error = EINVAL;
goto done;
}
break;
default:
/* pray and leave intact */
break;
}
cspace = CMSG_SPACE(cmsg.cmsg_len - sizeof(cmsg));
/* Check the buffer is big enough */
if (__predict_false(cidx + cspace > clen)) {
u_int8_t *nc;
clen = cidx + cspace;
nc = realloc(control, clen, M_TEMP, M_WAITOK);
if (!nc) {
error = ENOMEM;
goto done;
}
control = nc;
}
/* Copy header */
memcpy(&control[cidx], &cmsg, sizeof(cmsg));
/* Zero are between header and data */
memset(&control[cidx+sizeof(cmsg)], 0,
CMSG_ALIGN(sizeof(cmsg)) - sizeof(cmsg));
/* Copyin the data */
error = copyin(LINUX_CMSG_DATA(cc),
CMSG_DATA(control),
cmsg.cmsg_len - sizeof(cmsg));
if (error)
goto done;
/*
* If there is alignment difference, we changed
* layout of cmsg.
*/
if (LINUX_CMSG_ALIGNDIFF)
changed = 1;
resid -= cspace;
cidx += cspace;
} while ((cc = LINUX_CMSG_NXTHDR(&msg, cc)) && resid > 0);
/*
* If any of the passed control message needed
* a change, put the changed data into stackgap
* and adjust msg appropriately.
*/
if (changed) {
char *newc;
/*
* Check again the total len is maximum half of
* stackgap. The length might change if the
* alignment is different.
*/
if (clen > STACKGAPLEN/2) {
error = EMSGSIZE;
goto done;
}
/*
* Allocate space on stack within stackgap, and
* copy changed data there.
*/
if (!sg)
sg = stackgap_init(p, STACKGAPLEN/3);
newc = stackgap_alloc(p, &sg, clen);
if (!newc) {
error = ENOMEM;
goto done;
}
error = copyout(control, newc, clen);
if (error)
goto done;
msg.msg_control = newc;
msg.msg_controllen = clen;
}
free(control, M_TEMP);
control = NULL;
}
error = sendit(p, SCARG(uap, s), &msg, bflags, retval);
done:
if (control)
free(control, M_TEMP);
if (iov != aiov)
free(iov, M_IOV);
return (error);
}
int
linux_sys_recvfrom(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_recvfrom_args /* {
syscallarg(int) s;
syscallarg(void *) buf;
syscallarg(int) len;
syscallarg(int) flags;
syscallarg(struct osockaddr *) from;
syscallarg(int *) fromlenaddr;
} */ *uap = v;
int error;
struct sys_recvfrom_args bra;
SCARG(&bra, s) = SCARG(uap, s);
SCARG(&bra, buf) = SCARG(uap, buf);
SCARG(&bra, len) = SCARG(uap, len);
SCARG(&bra, flags) = SCARG(uap, flags);
SCARG(&bra, from) = (struct sockaddr *) SCARG(uap, from);
SCARG(&bra, fromlenaddr) = SCARG(uap, fromlenaddr);
if ((error = sys_recvfrom(l, &bra, retval)))
return (error);
if (SCARG(uap, from) && (error = linux_sa_put(SCARG(uap, from))))
return (error);
return (0);
}
int
linux_sys_recvmsg(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_recvmsg_args /* {
syscallarg(int) s;
syscallarg(struct msghdr *) msg;
syscallarg(u_int) flags;
} */ *uap = v;
struct msghdr msg;
int error;
struct sys_recvmsg_args bsa;
int lflags;
u_int8_t *ocontrol = NULL; /* XXX: gcc */
socklen_t ocontrollen = 0;
/*
* Data alignment is different on some architectures. If control
* message is expected, we must arrange for the control message
* to be initially put elsewhere, and copy to target place
* with Linux alignment.
*/
if (LINUX_CMSG_ALIGNDIFF) {
error = copyin(SCARG(uap, msg), &msg, sizeof(msg));
if (error)
return (error);
if (CMSG_FIRSTHDR(&msg)) {
caddr_t sg;
/* Need to fit within stackgap */
if (msg.msg_controllen > STACKGAPLEN/2) {
/* Sorry guys! */
return (EINVAL);
}
sg = stackgap_init(l->l_proc, STACKGAPLEN/3);
ocontrol = msg.msg_control;
ocontrollen = msg.msg_controllen;
/* space for at least one message's worth align */
msg.msg_controllen += CMSG_ALIGN(1);
msg.msg_control = stackgap_alloc(l->l_proc, &sg,
msg.msg_controllen);
if (!msg.msg_control)
return (ENOMEM);
/*
* Okay to overwrite the original structure, it's
* supposed to be writable.
*/
error = copyout(&msg, SCARG(uap, msg), sizeof(msg));
if (error)
return (error);
}
}
SCARG(&bsa, s) = SCARG(uap, s);
SCARG(&bsa, msg) = SCARG(uap, msg);
SCARG(&bsa, flags) = linux_to_bsd_msg_flags(SCARG(uap, flags));
if (SCARG(&bsa, flags) < 0) {
/* Some unsupported flag */
return (EINVAL);
}
if ((error = sys_recvmsg(l, &bsa, retval)))
goto done;
/* Fixup sockaddr */
error = copyin((caddr_t)SCARG(uap, msg), (caddr_t)&msg,
sizeof(msg));
if (error)
goto done;
if (msg.msg_name && msg.msg_namelen > 2) {
if ((error = linux_sa_put(msg.msg_name)))
goto done;
}
/* Fixup msg flags */
lflags = bsd_to_linux_msg_flags(msg.msg_flags);
if (lflags < 0) {
/* Some flag unsupported by Linux */
error = EINVAL;
goto done;
}
error = copyout(&lflags, (u_int8_t *) SCARG(uap, msg) +
offsetof(struct msghdr, msg_flags), sizeof(lflags));
if (error)
goto done;
/*
* Fixup cmsg. We handle two things:
* 1. different values for level/type on some archs
* 2. different alignment of CMSG_DATA on some archs
*/
if (CMSG_FIRSTHDR(&msg)) {
struct cmsghdr cmsg, *cc;
int changed = 0;
size_t resid = ocontrollen;
cc = CMSG_FIRSTHDR(&msg);
do {
error = copyin(cc, &cmsg, sizeof(cmsg));
if (error)
goto done;
switch (cmsg.cmsg_level) {
case SOL_SOCKET:
if (SOL_SOCKET != LINUX_SOL_SOCKET) {
cmsg.cmsg_level = LINUX_SOL_SOCKET;
changed = 1;
}
switch (cmsg.cmsg_type) {
case SCM_RIGHTS:
/* Linux SCM_RIGHTS is same as NetBSD */
break;
default:
/* other types not supported */
error = EINVAL;
goto done;
}
default:
/* pray and leave intact */
break;
}
if (LINUX_CMSG_ALIGNDIFF) {
int i;
u_int8_t d, *sd, *td;
/*
* Sanity check.
*/
if (cmsg.cmsg_len > resid
|| cmsg.cmsg_len < sizeof(cmsg)) {
error = EINVAL;
goto done;
}
/*
* Need to copy the cmsg from scratch area
* to the original place, converting data
* alignment from NetBSD to Linux one.
*/
error = copyout(&cmsg, ocontrol, sizeof(cmsg));
if (error)
goto done;
/* zero pad */
for(i=0; i < LINUX_CMSG_ALIGN(sizeof(cmsg)) - sizeof(cmsg); i++) {
copyout("",&ocontrol[sizeof(cmsg)+i],1);
}
sd = CMSG_DATA(cc);
td = LINUX_CMSG_DATA(ocontrol);
/* This is not particularily effective, but ..*/
d = '\0';
for(i=0; i < cmsg.cmsg_len - sizeof(cmsg); i++){
copyin(sd++, &d, 1);
copyout(&d, td++, 1);
}
resid -= (td - ocontrol);
ocontrol = td;
} else if (changed) {
/* Update cmsghdr in-place */
error = copyout(&cmsg, cc, sizeof(cmsg));
if (error)
goto done;
changed = 0;
}
} while((cc = CMSG_NXTHDR(&msg, cc)));
}
done:
return (error);
}
/*
* Convert socket option level from Linux to NetBSD value. Only SOL_SOCKET
* is different, the rest matches IPPROTO_* on both systems.
*/
int
linux_to_bsd_sopt_level(llevel)
int llevel;
{
switch (llevel) {
case LINUX_SOL_SOCKET:
return SOL_SOCKET;
case LINUX_SOL_IP:
return IPPROTO_IP;
case LINUX_SOL_TCP:
return IPPROTO_TCP;
case LINUX_SOL_UDP:
return IPPROTO_UDP;
default:
return -1;
}
}
/*
* Convert Linux socket level socket option numbers to NetBSD values.
*/
int
linux_to_bsd_so_sockopt(lopt)
int lopt;
{
switch (lopt) {
case LINUX_SO_DEBUG:
return SO_DEBUG;
case LINUX_SO_REUSEADDR:
/*
* Linux does not implement SO_REUSEPORT, but allows reuse of a
* host:port pair through SO_REUSEADDR even if the address is not a
* multicast-address. Effectively, this means that we should use
* SO_REUSEPORT to allow Linux applications to not exit with
* EADDRINUSE
*/
return SO_REUSEPORT;
case LINUX_SO_TYPE:
return SO_TYPE;
case LINUX_SO_ERROR:
return SO_ERROR;
case LINUX_SO_DONTROUTE:
return SO_DONTROUTE;
case LINUX_SO_BROADCAST:
return SO_BROADCAST;
case LINUX_SO_SNDBUF:
return SO_SNDBUF;
case LINUX_SO_RCVBUF:
return SO_RCVBUF;
case LINUX_SO_KEEPALIVE:
return SO_KEEPALIVE;
case LINUX_SO_OOBINLINE:
return SO_OOBINLINE;
case LINUX_SO_LINGER:
return SO_LINGER;
case LINUX_SO_PRIORITY:
case LINUX_SO_NO_CHECK:
default:
return -1;
}
}
/*
* Convert Linux IP level socket option number to NetBSD values.
*/
int
linux_to_bsd_ip_sockopt(lopt)
int lopt;
{
switch (lopt) {
case LINUX_IP_TOS:
return IP_TOS;
case LINUX_IP_TTL:
return IP_TTL;
case LINUX_IP_MULTICAST_TTL:
return IP_MULTICAST_TTL;
case LINUX_IP_MULTICAST_LOOP:
return IP_MULTICAST_LOOP;
case LINUX_IP_MULTICAST_IF:
return IP_MULTICAST_IF;
case LINUX_IP_ADD_MEMBERSHIP:
return IP_ADD_MEMBERSHIP;
case LINUX_IP_DROP_MEMBERSHIP:
return IP_DROP_MEMBERSHIP;
default:
return -1;
}
}
/*
* Convert Linux TCP level socket option number to NetBSD values.
*/
int
linux_to_bsd_tcp_sockopt(lopt)
int lopt;
{
switch (lopt) {
case LINUX_TCP_NODELAY:
return TCP_NODELAY;
case LINUX_TCP_MAXSEG:
return TCP_MAXSEG;
default:
return -1;
}
}
/*
* Convert Linux UDP level socket option number to NetBSD values.
*/
int
linux_to_bsd_udp_sockopt(lopt)
int lopt;
{
switch (lopt) {
default:
return -1;
}
}
/*
* Another reasonably straightforward function: setsockopt(2).
* The level and option numbers are converted; the values passed
* are not (yet) converted, the ones currently implemented don't
* need conversion, as they are the same on both systems.
*/
int
linux_sys_setsockopt(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_setsockopt_args /* {
syscallarg(int) s;
syscallarg(int) level;
syscallarg(int) optname;
syscallarg(void *) optval;
syscallarg(int) optlen;
} */ *uap = v;
struct sys_setsockopt_args bsa;
int name;
SCARG(&bsa, s) = SCARG(uap, s);
SCARG(&bsa, level) = linux_to_bsd_sopt_level(SCARG(uap, level));
SCARG(&bsa, val) = SCARG(uap, optval);
SCARG(&bsa, valsize) = SCARG(uap, optlen);
switch (SCARG(&bsa, level)) {
case SOL_SOCKET:
name = linux_to_bsd_so_sockopt(SCARG(uap, optname));
break;
case IPPROTO_IP:
name = linux_to_bsd_ip_sockopt(SCARG(uap, optname));
break;
case IPPROTO_TCP:
name = linux_to_bsd_tcp_sockopt(SCARG(uap, optname));
break;
case IPPROTO_UDP:
name = linux_to_bsd_udp_sockopt(SCARG(uap, optname));
break;
default:
return EINVAL;
}
if (name == -1)
return EINVAL;
SCARG(&bsa, name) = name;
return sys_setsockopt(l, &bsa, retval);
}
/*
* getsockopt(2) is very much the same as setsockopt(2) (see above)
*/
int
linux_sys_getsockopt(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_getsockopt_args /* {
syscallarg(int) s;
syscallarg(int) level;
syscallarg(int) optname;
syscallarg(void *) optval;
syscallarg(int *) optlen;
} */ *uap = v;
struct sys_getsockopt_args bga;
int name;
SCARG(&bga, s) = SCARG(uap, s);
SCARG(&bga, level) = linux_to_bsd_sopt_level(SCARG(uap, level));
SCARG(&bga, val) = SCARG(uap, optval);
SCARG(&bga, avalsize) = SCARG(uap, optlen);
switch (SCARG(&bga, level)) {
case SOL_SOCKET:
name = linux_to_bsd_so_sockopt(SCARG(uap, optname));
break;
case IPPROTO_IP:
name = linux_to_bsd_ip_sockopt(SCARG(uap, optname));
break;
case IPPROTO_TCP:
name = linux_to_bsd_tcp_sockopt(SCARG(uap, optname));
break;
case IPPROTO_UDP:
name = linux_to_bsd_udp_sockopt(SCARG(uap, optname));
break;
default:
return EINVAL;
}
if (name == -1)
return EINVAL;
SCARG(&bga, name) = name;
return sys_getsockopt(l, &bga, retval);
}
#define IF_NAME_LEN 16
int
linux_getifhwaddr(p, retval, fd, data)
struct proc *p;
register_t *retval;
u_int fd;
void *data;
{
/* Not the full structure, just enough to map what we do here */
struct linux_ifreq {
char if_name[IF_NAME_LEN];
struct osockaddr hwaddr;
} lreq;
struct filedesc *fdp;
struct file *fp;
struct ifaddr *ifa;
struct ifnet *ifp;
struct sockaddr_dl *sadl;
int error, found;
int index, ifnum;
/*
* We can't emulate this ioctl by calling sys_ioctl() to run
* SIOCGIFCONF, because the user buffer is not of the right
* type to take those results. We can't use kernel buffers to
* receive the results, as the implementation of sys_ioctl()
* and ifconf() [which implements SIOCGIFCONF] use
* copyin()/copyout() which will fail on kernel addresses.
*
* So, we must duplicate code from sys_ioctl() and ifconf(). Ugh.
*/
fdp = p->p_fd;
if ((fp = fd_getfile(fdp, fd)) == NULL)
return (EBADF);
FILE_USE(fp);
if ((fp->f_flag & (FREAD | FWRITE)) == 0) {
error = EBADF;
goto out;
}
error = copyin(data, (caddr_t)&lreq, sizeof(lreq));
if (error)
goto out;
lreq.if_name[IF_NAME_LEN-1] = '\0'; /* just in case */
/*
* Try real interface name first, then fake "ethX"
*/
for (ifp = ifnet.tqh_first, found = 0;
ifp != 0 && !found;
ifp = ifp->if_list.tqe_next) {
if (strcmp(lreq.if_name, ifp->if_xname))
/* not this interface */
continue;
found=1;
if ((ifa = ifp->if_addrlist.tqh_first) != 0) {
for (; ifa != 0; ifa = ifa->ifa_list.tqe_next) {
sadl = (struct sockaddr_dl *)ifa->ifa_addr;
/* only return ethernet addresses */
/* XXX what about FDDI, etc. ? */
if (sadl->sdl_family != AF_LINK ||
sadl->sdl_type != IFT_ETHER)
continue;
memcpy((caddr_t)&lreq.hwaddr.sa_data,
LLADDR(sadl),
MIN(sadl->sdl_alen,
sizeof(lreq.hwaddr.sa_data)));
lreq.hwaddr.sa_family =
sadl->sdl_family;
error = copyout((caddr_t)&lreq, data,
sizeof(lreq));
goto out;
}
} else {
error = ENODEV;
goto out;
}
}
if (strncmp(lreq.if_name, "eth", 3) == 0) {
for (ifnum = 0, index = 3;
lreq.if_name[index] != '\0' && index < IF_NAME_LEN;
index++) {
ifnum *= 10;
ifnum += lreq.if_name[index] - '0';
}
error = EINVAL; /* in case we don't find one */
for (ifp = ifnet.tqh_first, found = 0;
ifp != 0 && !found;
ifp = ifp->if_list.tqe_next) {
memcpy(lreq.if_name, ifp->if_xname,
MIN(IF_NAME_LEN, IFNAMSIZ));
if ((ifa = ifp->if_addrlist.tqh_first) == 0)
/* no addresses on this interface */
continue;
else
for (; ifa != 0; ifa = ifa->ifa_list.tqe_next) {
sadl = (struct sockaddr_dl *)ifa->ifa_addr;
/* only return ethernet addresses */
/* XXX what about FDDI, etc. ? */
if (sadl->sdl_family != AF_LINK ||
sadl->sdl_type != IFT_ETHER)
continue;
if (ifnum--)
/* not the reqested iface */
continue;
memcpy((caddr_t)&lreq.hwaddr.sa_data,
LLADDR(sadl),
MIN(sadl->sdl_alen,
sizeof(lreq.hwaddr.sa_data)));
lreq.hwaddr.sa_family =
sadl->sdl_family;
error = copyout((caddr_t)&lreq, data,
sizeof(lreq));
found = 1;
break;
}
}
} else {
/* unknown interface, not even an "eth*" name */
error = ENODEV;
}
out:
FILE_UNUSE(fp, p);
return error;
}
#undef IF_NAME_LEN
int
linux_ioctl_socket(p, uap, retval)
struct proc *p;
struct linux_sys_ioctl_args /* {
syscallarg(int) fd;
syscallarg(u_long) com;
syscallarg(caddr_t) data;
} */ *uap;
register_t *retval;
{
u_long com;
int error = 0, isdev = 0, dosys = 1;
struct sys_ioctl_args ia;
struct file *fp;
struct filedesc *fdp;
struct vnode *vp;
int (*ioctlf)(struct file *, u_long, void *, struct proc *);
struct ioctl_pt pt;
fdp = p->p_fd;
if ((fp = fd_getfile(fdp, SCARG(uap, fd))) == NULL)
return (EBADF);
FILE_USE(fp);
if (fp->f_type == DTYPE_VNODE) {
vp = (struct vnode *)fp->f_data;
isdev = vp->v_type == VCHR;
}
/*
* Don't try to interpret socket ioctl calls that are done
* on a device filedescriptor, just pass them through, to
* emulate Linux behaviour. Use PTIOCLINUX so that the
* device will only handle these if it's prepared to do
* so, to avoid unexpected things from happening.
*/
if (isdev) {
dosys = 0;
ioctlf = fp->f_ops->fo_ioctl;
pt.com = SCARG(uap, com);
pt.data = SCARG(uap, data);
error = ioctlf(fp, PTIOCLINUX, (caddr_t)&pt, p);
/*
* XXX hack: if the function returns EJUSTRETURN,
* it has stuffed a sysctl return value in pt.data.
*/
if (error == EJUSTRETURN) {
retval[0] = (register_t)pt.data;
error = 0;
}
goto out;
}
com = SCARG(uap, com);
retval[0] = 0;
switch (com) {
case LINUX_SIOCGIFCONF:
SCARG(&ia, com) = OSIOCGIFCONF;
break;
case LINUX_SIOCGIFFLAGS:
SCARG(&ia, com) = SIOCGIFFLAGS;
break;
case LINUX_SIOCSIFFLAGS:
SCARG(&ia, com) = SIOCSIFFLAGS;
break;
case LINUX_SIOCGIFADDR:
SCARG(&ia, com) = OSIOCGIFADDR;
break;
case LINUX_SIOCGIFDSTADDR:
SCARG(&ia, com) = OSIOCGIFDSTADDR;
break;
case LINUX_SIOCGIFBRDADDR:
SCARG(&ia, com) = OSIOCGIFBRDADDR;
break;
case LINUX_SIOCGIFNETMASK:
SCARG(&ia, com) = OSIOCGIFNETMASK;
break;
case LINUX_SIOCADDMULTI:
SCARG(&ia, com) = SIOCADDMULTI;
break;
case LINUX_SIOCDELMULTI:
SCARG(&ia, com) = SIOCDELMULTI;
break;
case LINUX_SIOCGIFHWADDR:
error = linux_getifhwaddr(p, retval, SCARG(uap, fd),
SCARG(uap, data));
dosys = 0;
break;
default:
error = EINVAL;
}
out:
FILE_UNUSE(fp, p);
if (error ==0 && dosys) {
SCARG(&ia, fd) = SCARG(uap, fd);
SCARG(&ia, data) = SCARG(uap, data);
/* XXX NJWLWP */
error = sys_ioctl(curlwp, &ia, retval);
}
return error;
}
int
linux_sys_connect(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_connect_args /* {
syscallarg(int) s;
syscallarg(const struct sockaddr *) name;
syscallarg(int) namelen;
} */ *uap = v;
struct proc *p = l->l_proc;
int error;
struct sockaddr *sa;
struct sys_connect_args bca;
caddr_t sg = stackgap_init(p, 0);
int namlen;
namlen = SCARG(uap, namelen);
error = linux_sa_get(p, SCARG(uap, s), &sg, &sa,
SCARG(uap, name), &namlen);
if (error)
return (error);
SCARG(&bca, s) = SCARG(uap, s);
SCARG(&bca, name) = sa;
SCARG(&bca, namelen) = (unsigned int) namlen;
error = sys_connect(l, &bca, retval);
if (error == EISCONN) {
struct file *fp;
struct socket *so;
int s, state, prflags;
/* getsock() will use the descriptor for us */
if (getsock(p->p_fd, SCARG(uap, s), &fp) != 0)
return EISCONN;
s = splsoftnet();
so = (struct socket *)fp->f_data;
state = so->so_state;
prflags = so->so_proto->pr_flags;
splx(s);
FILE_UNUSE(fp, p);
/*
* We should only let this call succeed once per
* non-blocking connect; however we don't have
* a convenient place to keep that state..
*/
if ((state & SS_NBIO) && (state & SS_ISCONNECTED) &&
(prflags & PR_CONNREQUIRED))
return 0;
}
return (error);
}
int
linux_sys_bind(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_bind_args /* {
syscallarg(int) s;
syscallarg(const struct osockaddr *) name;
syscallarg(int) namelen;
} */ *uap = v;
struct proc *p = l->l_proc;
int error, namlen;
struct sys_bind_args bsa;
namlen = SCARG(uap, namelen);
SCARG(&bsa, s) = SCARG(uap, s);
if (SCARG(uap, name)) {
struct sockaddr *sa;
caddr_t sg = stackgap_init(p, 0);
error = linux_sa_get(p, SCARG(uap, s), &sg, &sa,
SCARG(uap, name), &namlen);
if (error)
return (error);
SCARG(&bsa, name) = sa;
} else
SCARG(&bsa, name) = NULL;
SCARG(&bsa, namelen) = namlen;
return (sys_bind(l, &bsa, retval));
}
int
linux_sys_getsockname(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_getsockname_args /* {
syscallarg(int) fdes;
syscallarg(caddr_t) asa;
syscallarg(int *) alen;
} */ *uap = v;
int error;
if ((error = sys_getsockname(l, uap, retval)) != 0)
return (error);
if ((error = linux_sa_put((struct osockaddr *)SCARG(uap, asa))))
return (error);
return (0);
}
int
linux_sys_getpeername(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct sys_getpeername_args /* {
syscallarg(int) fdes;
syscallarg(caddr_t) asa;
syscallarg(int *) alen;
} */ *uap = v;
int error;
if ((error = sys_getpeername(l, uap, retval)) != 0)
return (error);
if ((error = linux_sa_put((struct osockaddr *)SCARG(uap, asa))))
return (error);
return (0);
}
/*
* Copy the osockaddr structure pointed to by osa to kernel, adjust
* family and convert to sockaddr, allocate stackgap and put the
* the converted structure there, address on stackgap returned in sap.
*/
static int
linux_sa_get(p, s, sgp, sap, osa, osalen)
struct proc *p;
int s;
caddr_t *sgp;
struct sockaddr **sap;
const struct osockaddr *osa;
int *osalen;
{
int error=0, bdom;
struct sockaddr *sa, *usa;
struct osockaddr *kosa = (struct osockaddr *) &sa;
int alloclen;
#ifdef INET6
int oldv6size;
struct sockaddr_in6 *sin6;
#endif
if (*osalen < 2 || *osalen > UCHAR_MAX || !osa) {
DPRINTF(("bad osa=%p osalen=%d\n", osa, *osalen));
return (EINVAL);
}
alloclen = *osalen;
#ifdef INET6
oldv6size = 0;
/*
* Check for old (pre-RFC2553) sockaddr_in6. We may accept it
* if it's a v4-mapped address, so reserve the proper space
* for it.
*/
if (alloclen == sizeof (struct sockaddr_in6) - sizeof (u_int32_t)) {
alloclen = sizeof (struct sockaddr_in6);
oldv6size = 1;
}
#endif
kosa = (struct osockaddr *) malloc(alloclen, M_TEMP, M_WAITOK);
if ((error = copyin(osa, (caddr_t) kosa, *osalen))) {
DPRINTF(("error copying osa %d\n", error));
goto out;
}
bdom = linux_to_bsd_domain(kosa->sa_family);
if (bdom == -1) {
DPRINTF(("bad linux family=%d\n", kosa->sa_family));
error = EINVAL;
goto out;
}
#ifdef INET6
/*
* Older Linux IPv6 code uses obsolete RFC2133 struct sockaddr_in6,
* which lacks the scope id compared with RFC2553 one. If we detect
* the situation, reject the address and write a message to system log.
*
* Still accept addresses for which the scope id is not used.
*/
if (oldv6size && bdom == AF_INET6) {
sin6 = (struct sockaddr_in6 *)kosa;
if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr) ||
(!IN6_IS_ADDR_LINKLOCAL(&sin6->sin6_addr) &&
!IN6_IS_ADDR_SITELOCAL(&sin6->sin6_addr) &&
!IN6_IS_ADDR_V4COMPAT(&sin6->sin6_addr) &&
!IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr) &&
!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))) {
sin6->sin6_scope_id = 0;
} else {
int uid = p->p_cred && p->p_ucred ?
p->p_ucred->cr_uid : -1;
log(LOG_DEBUG,
"pid %d (%s), uid %d: obsolete pre-RFC2553 "
"sockaddr_in6 rejected",
p->p_pid, p->p_comm, uid);
error = EINVAL;
goto out;
}
}
#endif
/*
* If the family is unspecified, use address family of the
* socket. This avoid triggering COMPAT_43 struct socket family check
* in sockargs() on little-endian machines, and strict family checks
* in netinet/in_pcb.c et.al.
*/
if (bdom == AF_UNSPEC) {
struct file *fp;
struct socket *so;
/* getsock() will use the descriptor for us */
if ((error = getsock(p->p_fd, s, &fp)) != 0)
goto out;
so = (struct socket *)fp->f_data;
bdom = so->so_proto->pr_domain->dom_family;
FILE_UNUSE(fp, p);
DPRINTF(("AF_UNSPEC family adjusted to %d\n", bdom));
}
if (bdom == AF_INET) {
alloclen = sizeof(struct sockaddr_in);
}
sa = (struct sockaddr *) kosa;
sa->sa_family = bdom;
sa->sa_len = alloclen;
#ifdef DEBUG_LINUX
DPRINTF(("family %d, len = %d [ ", sa->sa_family, sa->sa_len));
for (bdom = 0; bdom < sizeof(sa->sa_data); bdom++)
DPRINTF(("%02x ", (unsigned char) sa->sa_data[bdom]));
DPRINTF(("\n"));
#endif
usa = (struct sockaddr *) stackgap_alloc(p, sgp, alloclen);
if (!usa) {
error = ENOMEM;
goto out;
}
if ((error = copyout(sa, usa, alloclen))) {
DPRINTF(("error copying out socket %d\n", error));
goto out;
}
*sap = usa;
out:
*osalen = alloclen;
free(kosa, M_TEMP);
return (error);
}
static int
linux_sa_put(osa)
struct osockaddr *osa;
{
struct sockaddr sa;
struct osockaddr *kosa;
int error, bdom, len;
/*
* Only read/write the sockaddr family and length part, the rest is
* not changed.
*/
len = sizeof(sa.sa_len) + sizeof(sa.sa_family);
error = copyin((caddr_t) osa, (caddr_t) &sa, len);
if (error)
return (error);
bdom = bsd_to_linux_domain(sa.sa_family);
if (bdom == -1)
return (EINVAL);
/* Note: we convert from sockaddr to osockaddr here, too */
kosa = (struct osockaddr *) &sa;
kosa->sa_family = bdom;
error = copyout(kosa, osa, len);
if (error)
return (error);
return (0);
}
#ifndef __amd64__
int
linux_sys_recv(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_recv_args /* {
syscallarg(int) s;
syscallarg(void *) buf;
syscallarg(int) len;
syscallarg(int) flags;
} */ *uap = v;
struct sys_recvfrom_args bra;
SCARG(&bra, s) = SCARG(uap, s);
SCARG(&bra, buf) = SCARG(uap, buf);
SCARG(&bra, len) = (size_t) SCARG(uap, len);
SCARG(&bra, flags) = SCARG(uap, flags);
SCARG(&bra, from) = NULL;
SCARG(&bra, fromlenaddr) = NULL;
return (sys_recvfrom(l, &bra, retval));
}
int
linux_sys_send(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_send_args /* {
syscallarg(int) s;
syscallarg(caddr_t) buf;
syscallarg(int) len;
syscallarg(int) flags;
} */ *uap = v;
struct sys_sendto_args bsa;
SCARG(&bsa, s) = SCARG(uap, s);
SCARG(&bsa, buf) = SCARG(uap, buf);
SCARG(&bsa, len) = SCARG(uap, len);
SCARG(&bsa, flags) = SCARG(uap, flags);
SCARG(&bsa, to) = NULL;
SCARG(&bsa, tolen) = 0;
return (sys_sendto(l, &bsa, retval));
}
#endif /* __amd64__ */
int
linux_sys_accept(l, v, retval)
struct lwp *l;
void *v;
register_t *retval;
{
struct linux_sys_accept_args /* {
syscallarg(int) s;
syscallarg(struct osockaddr *) name;
syscallarg(int *) anamelen;
} */ *uap = v;
int error;
struct sys_accept_args baa;
SCARG(&baa, s) = SCARG(uap, s);
SCARG(&baa, name) = (struct sockaddr *) SCARG(uap, name);
SCARG(&baa, anamelen) = (unsigned int *) SCARG(uap, anamelen);
if ((error = sys_accept(l, &baa, retval)))
return (error);
if (SCARG(uap, name) && (error = linux_sa_put(SCARG(uap, name))))
return (error);
return (0);
}