/* ARGSUSED */
int
sys_getpeername(struct proc *p, void *v, register_t *retval)
{
struct sys_getpeername_args /* {
syscallarg(int) fdes;
syscallarg(struct sockaddr *) asa;
syscallarg(socklen_t *) alen;
} */ *uap = v;
struct file *fp;
struct socket *so;
struct mbuf *m = NULL;
socklen_t len;
int error;
if ((error = getsock(p->p_fd, SCARG(uap, fdes), &fp)) != 0)
return (error);
so = fp->f_data;
if ((so->so_state & (SS_ISCONNECTED|SS_ISCONFIRMING)) == 0) {
FRELE(fp, p);
return (ENOTCONN);
}
error = copyin(SCARG(uap, alen), &len, sizeof (len));
if (error)
goto bad;
m = m_getclr(M_WAIT, MT_SONAME);
error = (*so->so_proto->pr_usrreq)(so, PRU_PEERADDR, 0, m, 0, p);
if (error)
goto bad;
error = copyaddrout(p, m, SCARG(uap, asa), len, SCARG(uap, alen));
bad:
FRELE(fp, p);
m_freem(m);
return (error);
}
/* ARGSUSED */
int
sys_connect(struct proc *p, void *v, register_t *retval)
{
struct sys_connect_args /* {
syscallarg(int) s;
syscallarg(const struct sockaddr *) name;
syscallarg(socklen_t) namelen;
} */ *uap = v;
struct file *fp;
struct socket *so;
struct mbuf *nam = NULL;
int error, s;
if ((error = getsock(p->p_fd, SCARG(uap, s), &fp)) != 0)
return (error);
so = fp->f_data;
if ((so->so_state & SS_NBIO) && (so->so_state & SS_ISCONNECTING)) {
FRELE(fp, p);
return (EALREADY);
}
error = sockargs(&nam, SCARG(uap, name), SCARG(uap, namelen),
MT_SONAME);
if (error)
goto bad;
#ifdef KTRACE
if (KTRPOINT(p, KTR_STRUCT))
ktrsockaddr(p, mtod(nam, caddr_t), SCARG(uap, namelen));
#endif
error = soconnect(so, nam);
if (error)
goto bad;
if ((so->so_state & SS_NBIO) && (so->so_state & SS_ISCONNECTING)) {
FRELE(fp, p);
m_freem(nam);
return (EINPROGRESS);
}
s = splsoftnet();
while ((so->so_state & SS_ISCONNECTING) && so->so_error == 0) {
error = tsleep(&so->so_timeo, PSOCK | PCATCH, "netcon2", 0);
if (error)
break;
}
if (error == 0) {
error = so->so_error;
so->so_error = 0;
}
splx(s);
bad:
so->so_state &= ~SS_ISCONNECTING;
FRELE(fp, p);
if (nam)
m_freem(nam);
if (error == ERESTART)
error = EINTR;
return (error);
}
/* ARGSUSED */
int
sys_bind(struct proc *p, void *v, register_t *retval)
{
struct sys_bind_args /* {
syscallarg(int) s;
syscallarg(const struct sockaddr *) name;
syscallarg(socklen_t) namelen;
} */ *uap = v;
struct file *fp;
struct mbuf *nam;
int error;
if ((error = getsock(p->p_fd, SCARG(uap, s), &fp)) != 0)
return (error);
error = sockargs(&nam, SCARG(uap, name), SCARG(uap, namelen),
MT_SONAME);
if (error == 0) {
#ifdef KTRACE
if (KTRPOINT(p, KTR_STRUCT))
ktrsockaddr(p, mtod(nam, caddr_t), SCARG(uap, namelen));
#endif
error = sobind(fp->f_data, nam, p);
m_freem(nam);
}
FRELE(fp, p);
return (error);
}
/* ARGSUSED */
int
sys_setsockopt(struct proc *p, void *v, register_t *retval)
{
struct sys_setsockopt_args /* {
syscallarg(int) s;
syscallarg(int) level;
syscallarg(int) name;
syscallarg(const void *) val;
syscallarg(socklen_t) valsize;
} */ *uap = v;
struct file *fp;
struct mbuf *m = NULL;
int error;
if ((error = getsock(p->p_fd, SCARG(uap, s), &fp)) != 0)
return (error);
if (SCARG(uap, valsize) > MCLBYTES) {
error = EINVAL;
goto bad;
}
if (SCARG(uap, val)) {
m = m_get(M_WAIT, MT_SOOPTS);
if (SCARG(uap, valsize) > MLEN) {
MCLGET(m, M_DONTWAIT);
if ((m->m_flags & M_EXT) == 0) {
error = ENOBUFS;
goto bad;
}
}
if (m == NULL) {
error = ENOBUFS;
goto bad;
}
error = copyin(SCARG(uap, val), mtod(m, caddr_t),
SCARG(uap, valsize));
if (error) {
goto bad;
}
m->m_len = SCARG(uap, valsize);
}
error = sosetopt(fp->f_data, SCARG(uap, level), SCARG(uap, name), m);
m = NULL;
bad:
if (m)
m_freem(m);
FRELE(fp, p);
return (error);
}
/* ARGSUSED */
int
sys_shutdown(struct proc *p, void *v, register_t *retval)
{
struct sys_shutdown_args /* {
syscallarg(int) s;
syscallarg(int) how;
} */ *uap = v;
struct file *fp;
int error;
if ((error = getsock(p->p_fd, SCARG(uap, s), &fp)) != 0)
return (error);
error = soshutdown(fp->f_data, SCARG(uap, how));
FRELE(fp, p);
return (error);
}
/*
* Given a file descriptor, check an ACL for it.
*/
int
sys___acl_aclcheck_fd(struct proc *p, void *v, register_t *retval)
{
struct sys___acl_aclcheck_fd_args *uap = v;
struct vnode *vp;
struct file *fp;
int error;
error = getvnode(p->p_fd, SCARG(uap, filedes), &fp);
if (error)
return (error);
vp = (struct vnode *)fp->f_data;
vref(vp); /* vn_acl_check() will drop the reference. */
error = vn_acl_check(p, vp, SCARG(uap, type), SCARG(uap, aclp));
FRELE(fp);
return (error);
}
/* ARGSUSED */
int
sys_getsockopt(struct proc *p, void *v, register_t *retval)
{
struct sys_getsockopt_args /* {
syscallarg(int) s;
syscallarg(int) level;
syscallarg(int) name;
syscallarg(void *) val;
syscallarg(socklen_t *) avalsize;
} */ *uap = v;
struct file *fp;
struct mbuf *m = NULL;
socklen_t valsize;
int error;
if ((error = getsock(p->p_fd, SCARG(uap, s), &fp)) != 0)
return (error);
if (SCARG(uap, val)) {
error = copyin(SCARG(uap, avalsize),
&valsize, sizeof (valsize));
if (error)
goto out;
} else
valsize = 0;
if ((error = sogetopt(fp->f_data, SCARG(uap, level),
SCARG(uap, name), &m)) == 0 && SCARG(uap, val) && valsize &&
m != NULL) {
if (valsize > m->m_len)
valsize = m->m_len;
error = copyout(mtod(m, caddr_t), SCARG(uap, val), valsize);
if (error == 0)
error = copyout(&valsize,
SCARG(uap, avalsize), sizeof (valsize));
}
out:
FRELE(fp, p);
if (m != NULL)
(void)m_free(m);
return (error);
}
int
linux_ioctl_fdio(struct proc *p, struct linux_sys_ioctl_args *uap,
register_t *retval)
{
struct filedesc *fdp;
struct file *fp;
int error;
int (*ioctlf)(struct file *, u_long, caddr_t, struct proc *);
u_long com;
struct fd_type fparams;
struct linux_floppy_struct lflop;
struct linux_floppy_drive_struct ldrive;
com = (u_long)SCARG(uap, data);
fdp = p->p_fd;
if ((fp = fd_getfile(fdp, SCARG(uap, fd))) == NULL)
return (EBADF);
FREF(fp);
com = SCARG(uap, com);
ioctlf = fp->f_ops->fo_ioctl;
retval[0] = error = 0;
switch (com) {
case LINUX_FDMSGON:
case LINUX_FDMSGOFF:
case LINUX_FDTWADDLE:
case LINUX_FDCLRPRM:
/* whatever you say */
break;
case LINUX_FDPOLLDRVSTAT:
/*
* Just fill in some innocent defaults.
*/
memset(&ldrive, 0, sizeof ldrive);
ldrive.fd_ref = 1;
ldrive.maxblock = 2;
ldrive.maxtrack = ldrive.track = 1;
ldrive.flags = LINUX_FD_DISK_WRITABLE;
error = copyout(&ldrive, SCARG(uap, data), sizeof ldrive);
break;
case LINUX_FDGETPRM:
error = ioctlf(fp, FD_GTYPE, (caddr_t)&fparams, p);
if (error != 0)
break;
lflop.size = fparams.heads * fparams.sectrac * fparams.tracks;
lflop.sect = fparams.sectrac;
lflop.head = fparams.heads;
lflop.track = fparams.tracks;
lflop.stretch = fparams.step == 2 ? 1 : 0;
lflop.spec1 = fparams.steprate;
lflop.gap = fparams.gap1;
lflop.fmt_gap = fparams.gap2;
lflop.rate = fparams.rate;
error = copyout(&lflop, SCARG(uap, data), sizeof lflop);
break;
case LINUX_FDSETPRM:
/*
* Should use FDIOCSETFORMAT here, iff its interface
* is extended.
*/
case LINUX_FDDEFPRM:
case LINUX_FDFMTBEG:
case LINUX_FDFMTTRK:
case LINUX_FDFMTEND:
case LINUX_FDSETEMSGTRESH:
case LINUX_FDFLUSH:
case LINUX_FDSETMAXERRS:
case LINUX_FDGETMAXERRS:
case LINUX_FDGETDRVTYP:
case LINUX_FDSETDRVPRM:
case LINUX_FDGETDRVPRM:
case LINUX_FDGETDRVSTAT:
case LINUX_FDRESET:
case LINUX_FDGETFDCSTAT:
case LINUX_FDWERRORCLR:
case LINUX_FDWERRORGET:
case LINUX_FDRAWCMD:
case LINUX_FDEJECT:
default:
error = EINVAL;
}
FRELE(fp, p);
return 0;
}
/*
* We come here in a last attempt to satisfy a Linux ioctl() call
*/
int
linux_machdepioctl(struct proc *p, void *v, register_t *retval)
{
struct linux_sys_ioctl_args /* {
syscallarg(int) fd;
syscallarg(u_long) com;
syscallarg(caddr_t) data;
} */ *uap = v;
struct sys_ioctl_args bia;
u_long com;
int error;
#if (NWSDISPLAY > 0 && defined(WSDISPLAY_COMPAT_USL))
struct vt_mode lvt;
caddr_t bvtp, sg;
#endif
struct filedesc *fdp;
struct file *fp;
int fd;
int (*ioctlf)(struct file *, u_long, caddr_t, struct proc *);
struct ioctl_pt pt;
fd = SCARG(uap, fd);
SCARG(&bia, fd) = SCARG(uap, fd);
SCARG(&bia, data) = SCARG(uap, data);
com = SCARG(uap, com);
fdp = p->p_fd;
if ((fp = fd_getfile(fdp, fd)) == NULL)
return (EBADF);
switch (com) {
#if (NWSDISPLAY > 0 && defined(WSDISPLAY_COMPAT_USL))
case LINUX_KDGKBMODE:
com = KDGKBMODE;
break;
case LINUX_KDSKBMODE:
com = KDSKBMODE;
if ((unsigned)SCARG(uap, data) == LINUX_K_MEDIUMRAW)
SCARG(&bia, data) = (caddr_t)K_RAW;
break;
case LINUX_KIOCSOUND:
SCARG(&bia, data) =
(caddr_t)(((unsigned long)SCARG(&bia, data)) & 0xffff);
/* FALLTHROUGH */
case LINUX_KDMKTONE:
com = KDMKTONE;
break;
case LINUX_KDSETMODE:
com = KDSETMODE;
break;
case LINUX_KDGETMODE:
#if NWSDISPLAY > 0 && defined(WSDISPLAY_COMPAT_USL)
com = WSDISPLAYIO_GMODE;
#else
com = KDGETMODE;
#endif
break;
case LINUX_KDENABIO:
com = KDENABIO;
break;
case LINUX_KDDISABIO:
com = KDDISABIO;
break;
case LINUX_KDGETLED:
com = KDGETLED;
break;
case LINUX_KDSETLED:
com = KDSETLED;
break;
case LINUX_VT_OPENQRY:
com = VT_OPENQRY;
break;
case LINUX_VT_GETMODE: {
int sig;
SCARG(&bia, com) = VT_GETMODE;
if ((error = sys_ioctl(p, &bia, retval)))
return error;
if ((error = copyin(SCARG(uap, data), (caddr_t)&lvt,
sizeof (struct vt_mode))))
return error;
/* We need to bounds check here in case there
is a race with another thread */
if ((error = bsd_to_linux_signal(lvt.relsig, &sig)))
return error;
lvt.relsig = sig;
if ((error = bsd_to_linux_signal(lvt.acqsig, &sig)))
return error;
lvt.acqsig = sig;
if ((error = bsd_to_linux_signal(lvt.frsig, &sig)))
return error;
lvt.frsig = sig;
return copyout((caddr_t)&lvt, SCARG(uap, data),
sizeof (struct vt_mode));
}
case LINUX_VT_SETMODE: {
int sig;
com = VT_SETMODE;
if ((error = copyin(SCARG(uap, data), (caddr_t)&lvt,
sizeof (struct vt_mode))))
return error;
if ((error = linux_to_bsd_signal(lvt.relsig, &sig)))
return error;
lvt.relsig = sig;
if ((error = linux_to_bsd_signal(lvt.acqsig, &sig)))
return error;
lvt.acqsig = sig;
if ((error = linux_to_bsd_signal(lvt.frsig, &sig)))
return error;
lvt.frsig = sig;
sg = stackgap_init(p->p_emul);
bvtp = stackgap_alloc(&sg, sizeof (struct vt_mode));
if ((error = copyout(&lvt, bvtp, sizeof (struct vt_mode))))
return error;
SCARG(&bia, data) = bvtp;
break;
}
case LINUX_VT_DISALLOCATE:
/* XXX should use WSDISPLAYIO_DELSCREEN */
return 0;
case LINUX_VT_RELDISP:
com = VT_RELDISP;
break;
case LINUX_VT_ACTIVATE:
com = VT_ACTIVATE;
break;
case LINUX_VT_WAITACTIVE:
com = VT_WAITACTIVE;
break;
case LINUX_VT_GETSTATE:
com = VT_GETSTATE;
break;
case LINUX_KDGKBTYPE:
{
char tmp = KB_101;
/* This is what Linux does */
return copyout(&tmp, SCARG(uap, data), sizeof(char));
}
#endif
default:
/*
* Unknown to us. If it's on a device, just pass it through
* using PTIOCLINUX, the device itself might be able to
* make some sense of it.
* XXX hack: if the function returns EJUSTRETURN,
* it has stuffed a sysctl return value in pt.data.
*/
FREF(fp);
ioctlf = fp->f_ops->fo_ioctl;
pt.com = SCARG(uap, com);
pt.data = SCARG(uap, data);
error = ioctlf(fp, PTIOCLINUX, (caddr_t)&pt, p);
FRELE(fp);
if (error == EJUSTRETURN) {
retval[0] = (register_t)pt.data;
error = 0;
}
if (error == ENOTTY)
printf("linux_machdepioctl: invalid ioctl %08lx\n",
com);
return (error);
}
SCARG(&bia, com) = com;
return sys_ioctl(p, &bia, retval);
}
int
linux_ioctl_hdio(struct proc *p, struct linux_sys_ioctl_args *uap,
register_t *retval)
{
u_long com;
int error, error1;
caddr_t sg;
struct filedesc *fdp;
struct file *fp;
int (*ioctlf)(struct file *, u_long, caddr_t, struct proc *);
struct ataparams *atap, ata;
struct atareq req;
struct disklabel label, *labp;
struct partinfo partp;
struct linux_hd_geometry hdg;
struct linux_hd_big_geometry hdg_big;
fdp = p->p_fd;
if ((fp = fd_getfile(fdp, SCARG(uap, fd))) == NULL)
return (EBADF);
FREF(fp);
com = SCARG(uap, com);
ioctlf = fp->f_ops->fo_ioctl;
retval[0] = error = 0;
com = SCARG(uap, com);
switch (com) {
case LINUX_HDIO_OBSOLETE_IDENTITY:
case LINUX_HDIO_GET_IDENTITY:
sg = stackgap_init(p->p_emul);
atap = stackgap_alloc(&sg, DEV_BSIZE);
if (atap == NULL) {
error = ENOMEM;
break;
}
req.flags = ATACMD_READ;
req.command = WDCC_IDENTIFY;
req.databuf = (caddr_t)atap;
req.datalen = DEV_BSIZE;
req.timeout = 1000;
error = ioctlf(fp, ATAIOCCOMMAND, (caddr_t)&req, p);
if (error != 0)
break;
if (req.retsts != ATACMD_OK) {
error = EIO;
break;
}
error = copyin(atap, &ata, sizeof ata);
if (error != 0)
break;
/*
* 142 is the size of the old structure used by Linux,
* which doesn't seem to be defined anywhere anymore.
*/
error = copyout(&ata, SCARG(uap, data),
com == LINUX_HDIO_GET_IDENTITY ? sizeof ata : 142);
break;
case LINUX_HDIO_GETGEO:
error = linux_machdepioctl(p, uap, retval);
if (error == 0)
break;
error = ioctlf(fp, DIOCGDINFO, (caddr_t)&label, p);
error1 = ioctlf(fp, DIOCGPART, (caddr_t)&partp, p);
if (error != 0 && error1 != 0) {
error = error1;
break;
}
labp = error != 0 ? &label : partp.disklab;
hdg.start = error1 != 0 ? partp.part->p_offset : 0;
hdg.heads = labp->d_ntracks;
hdg.cylinders = labp->d_ncylinders;
hdg.sectors = labp->d_nsectors;
error = copyout(&hdg, SCARG(uap, data), sizeof hdg);
break;
case LINUX_HDIO_GETGEO_BIG:
error = linux_machdepioctl(p, uap, retval);
if (error == 0)
break;
case LINUX_HDIO_GETGEO_BIG_RAW:
error = ioctlf(fp, DIOCGDINFO, (caddr_t)&label, p);
error1 = ioctlf(fp, DIOCGPART, (caddr_t)&partp, p);
if (error != 0 && error1 != 0) {
error = error1;
break;
}
labp = error != 0 ? &label : partp.disklab;
hdg_big.start = error1 != 0 ? partp.part->p_offset : 0;
hdg_big.heads = labp->d_ntracks;
hdg_big.cylinders = labp->d_ncylinders;
hdg_big.sectors = labp->d_nsectors;
error = copyout(&hdg_big, SCARG(uap, data), sizeof hdg_big);
break;
case LINUX_HDIO_GET_UNMASKINTR:
case LINUX_HDIO_GET_MULTCOUNT:
case LINUX_HDIO_GET_KEEPSETTINGS:
case LINUX_HDIO_GET_32BIT:
case LINUX_HDIO_GET_NOWERR:
case LINUX_HDIO_GET_DMA:
case LINUX_HDIO_GET_NICE:
case LINUX_HDIO_DRIVE_RESET:
case LINUX_HDIO_TRISTATE_HWIF:
case LINUX_HDIO_DRIVE_TASK:
case LINUX_HDIO_DRIVE_CMD:
case LINUX_HDIO_SET_MULTCOUNT:
case LINUX_HDIO_SET_UNMASKINTR:
case LINUX_HDIO_SET_KEEPSETTINGS:
case LINUX_HDIO_SET_32BIT:
case LINUX_HDIO_SET_NOWERR:
case LINUX_HDIO_SET_DMA:
case LINUX_HDIO_SET_PIO_MODE:
case LINUX_HDIO_SCAN_HWIF:
case LINUX_HDIO_SET_NICE:
case LINUX_HDIO_UNREGISTER_HWIF:
error = EINVAL;
}
FRELE(fp);
return error;
}
int
recvit(struct proc *p, int s, struct msghdr *mp, caddr_t namelenp,
register_t *retsize)
{
struct file *fp;
struct uio auio;
struct iovec *iov;
int i;
size_t len;
int error;
struct mbuf *from = NULL, *control = NULL;
#ifdef KTRACE
struct iovec *ktriov = NULL;
#endif
if ((error = getsock(p->p_fd, s, &fp)) != 0)
return (error);
auio.uio_iov = mp->msg_iov;
auio.uio_iovcnt = mp->msg_iovlen;
auio.uio_segflg = UIO_USERSPACE;
auio.uio_rw = UIO_READ;
auio.uio_procp = p;
auio.uio_offset = 0; /* XXX */
auio.uio_resid = 0;
iov = mp->msg_iov;
for (i = 0; i < mp->msg_iovlen; i++, iov++) {
/* Don't allow sum > SSIZE_MAX */
if (iov->iov_len > SSIZE_MAX ||
(auio.uio_resid += iov->iov_len) > SSIZE_MAX) {
error = EINVAL;
goto out;
}
}
#ifdef KTRACE
if (KTRPOINT(p, KTR_GENIO)) {
int iovlen = auio.uio_iovcnt * sizeof (struct iovec);
ktriov = malloc(iovlen, M_TEMP, M_WAITOK);
bcopy(auio.uio_iov, ktriov, iovlen);
}
#endif
len = auio.uio_resid;
error = soreceive(fp->f_data, &from, &auio, NULL,
mp->msg_control ? &control : NULL,
&mp->msg_flags,
mp->msg_control ? mp->msg_controllen : 0);
if (error) {
if (auio.uio_resid != len && (error == ERESTART ||
error == EINTR || error == EWOULDBLOCK))
error = 0;
}
#ifdef KTRACE
if (ktriov != NULL) {
if (error == 0)
ktrgenio(p, s, UIO_READ, ktriov, len - auio.uio_resid);
free(ktriov, M_TEMP);
}
#endif
if (error)
goto out;
*retsize = len - auio.uio_resid;
if (mp->msg_name) {
socklen_t alen;
if (from == NULL)
alen = 0;
else {
alen = from->m_len;
error = copyout(mtod(from, caddr_t), mp->msg_name,
MIN(alen, mp->msg_namelen));
if (error)
goto out;
#ifdef KTRACE
if (KTRPOINT(p, KTR_STRUCT))
ktrsockaddr(p, mtod(from, caddr_t), alen);
#endif
}
mp->msg_namelen = alen;
if (namelenp &&
(error = copyout(&alen, namelenp, sizeof(alen)))) {
goto out;
}
}
if (mp->msg_control) {
len = mp->msg_controllen;
if (len <= 0 || control == NULL)
len = 0;
else {
struct mbuf *m = control;
caddr_t p = mp->msg_control;
do {
i = m->m_len;
if (len < i) {
mp->msg_flags |= MSG_CTRUNC;
i = len;
}
error = copyout(mtod(m, caddr_t), p, i);
if (m->m_next)
i = ALIGN(i);
p += i;
len -= i;
if (error != 0 || len <= 0)
break;
} while ((m = m->m_next) != NULL);
len = p - (caddr_t)mp->msg_control;
}
mp->msg_controllen = len;
}
if (!error) {
fp->f_rxfer++;
fp->f_rbytes += *retsize;
}
out:
FRELE(fp, p);
if (from)
m_freem(from);
if (control)
m_freem(control);
return (error);
}
int
sendit(struct proc *p, int s, struct msghdr *mp, int flags, register_t *retsize)
{
struct file *fp;
struct uio auio;
struct iovec *iov;
int i;
struct mbuf *to, *control;
size_t len;
int error;
#ifdef KTRACE
struct iovec *ktriov = NULL;
#endif
to = NULL;
if ((error = getsock(p->p_fd, s, &fp)) != 0)
return (error);
auio.uio_iov = mp->msg_iov;
auio.uio_iovcnt = mp->msg_iovlen;
auio.uio_segflg = UIO_USERSPACE;
auio.uio_rw = UIO_WRITE;
auio.uio_procp = p;
auio.uio_offset = 0; /* XXX */
auio.uio_resid = 0;
iov = mp->msg_iov;
for (i = 0; i < mp->msg_iovlen; i++, iov++) {
/* Don't allow sum > SSIZE_MAX */
if (iov->iov_len > SSIZE_MAX ||
(auio.uio_resid += iov->iov_len) > SSIZE_MAX) {
error = EINVAL;
goto bad;
}
}
if (mp->msg_name) {
error = sockargs(&to, mp->msg_name, mp->msg_namelen,
MT_SONAME);
if (error)
goto bad;
#ifdef KTRACE
if (KTRPOINT(p, KTR_STRUCT))
ktrsockaddr(p, mtod(to, caddr_t), mp->msg_namelen);
#endif
}
if (mp->msg_control) {
if (mp->msg_controllen < CMSG_ALIGN(sizeof(struct cmsghdr))) {
error = EINVAL;
goto bad;
}
error = sockargs(&control, mp->msg_control,
mp->msg_controllen, MT_CONTROL);
if (error)
goto bad;
} else
control = 0;
#ifdef KTRACE
if (KTRPOINT(p, KTR_GENIO)) {
int iovlen = auio.uio_iovcnt * sizeof (struct iovec);
ktriov = malloc(iovlen, M_TEMP, M_WAITOK);
bcopy(auio.uio_iov, ktriov, iovlen);
}
#endif
len = auio.uio_resid;
error = sosend(fp->f_data, to, &auio, NULL, control, flags);
if (error) {
if (auio.uio_resid != len && (error == ERESTART ||
error == EINTR || error == EWOULDBLOCK))
error = 0;
if (error == EPIPE && (flags & MSG_NOSIGNAL) == 0)
ptsignal(p, SIGPIPE, STHREAD);
}
if (error == 0) {
*retsize = len - auio.uio_resid;
fp->f_wxfer++;
fp->f_wbytes += *retsize;
}
#ifdef KTRACE
if (ktriov != NULL) {
if (error == 0)
ktrgenio(p, s, UIO_WRITE, ktriov, *retsize);
free(ktriov, M_TEMP);
}
#endif
bad:
FRELE(fp, p);
if (to)
m_freem(to);
return (error);
}
int
sys_accept(struct proc *p, void *v, register_t *retval)
{
struct sys_accept_args /* {
syscallarg(int) s;
syscallarg(struct sockaddr *) name;
syscallarg(socklen_t *) anamelen;
} */ *uap = v;
struct file *fp, *headfp;
struct mbuf *nam;
socklen_t namelen;
int error, s, tmpfd;
struct socket *head, *so;
int nflag;
if (SCARG(uap, name) && (error = copyin(SCARG(uap, anamelen),
&namelen, sizeof (namelen))))
return (error);
if ((error = getsock(p->p_fd, SCARG(uap, s), &fp)) != 0)
return (error);
headfp = fp;
s = splsoftnet();
head = fp->f_data;
redo:
if ((head->so_options & SO_ACCEPTCONN) == 0) {
error = EINVAL;
goto bad;
}
if ((head->so_state & SS_NBIO) && head->so_qlen == 0) {
if (head->so_state & SS_CANTRCVMORE)
error = ECONNABORTED;
else
error = EWOULDBLOCK;
goto bad;
}
while (head->so_qlen == 0 && head->so_error == 0) {
if (head->so_state & SS_CANTRCVMORE) {
head->so_error = ECONNABORTED;
break;
}
error = tsleep(&head->so_timeo, PSOCK | PCATCH, "netcon", 0);
if (error) {
goto bad;
}
}
if (head->so_error) {
error = head->so_error;
head->so_error = 0;
goto bad;
}
/* Take note if socket was non-blocking. */
nflag = (headfp->f_flag & FNONBLOCK);
fdplock(p->p_fd);
error = falloc(p, &fp, &tmpfd);
fdpunlock(p->p_fd);
if (error != 0) {
/*
* Probably ran out of file descriptors. Wakeup
* so some other process might have a chance at it.
*/
wakeup_one(&head->so_timeo);
goto bad;
}
nam = m_get(M_WAIT, MT_SONAME);
/*
* Check whether the queue emptied while we slept: falloc() or
* m_get() may have blocked, allowing the connection to be reset
* or another thread or process to accept it. If so, start over.
*/
if (head->so_qlen == 0) {
m_freem(nam);
fdplock(p->p_fd);
fdremove(p->p_fd, tmpfd);
closef(fp, p);
fdpunlock(p->p_fd);
goto redo;
}
/*
* Do not sleep after we have taken the socket out of the queue.
*/
so = TAILQ_FIRST(&head->so_q);
if (soqremque(so, 1) == 0)
panic("accept");
/* connection has been removed from the listen queue */
KNOTE(&head->so_rcv.sb_sel.si_note, 0);
fp->f_type = DTYPE_SOCKET;
fp->f_flag = FREAD | FWRITE | nflag;
fp->f_ops = &socketops;
fp->f_data = so;
error = soaccept(so, nam);
if (!error && SCARG(uap, name)) {
error = copyaddrout(p, nam, SCARG(uap, name), namelen,
SCARG(uap, anamelen));
}
if (error) {
/* if an error occurred, free the file descriptor */
fdplock(p->p_fd);
fdremove(p->p_fd, tmpfd);
closef(fp, p);
fdpunlock(p->p_fd);
} else {
FILE_SET_MATURE(fp, p);
*retval = tmpfd;
}
m_freem(nam);
bad:
splx(s);
FRELE(headfp, p);
return (error);
}
int
sys_mmap(struct proc *p, void *v, register_t *retval)
{
struct sys_mmap_args /* {
syscallarg(void *) addr;
syscallarg(size_t) len;
syscallarg(int) prot;
syscallarg(int) flags;
syscallarg(int) fd;
syscallarg(long) pad;
syscallarg(off_t) pos;
} */ *uap = v;
vaddr_t addr;
struct vattr va;
off_t pos;
vsize_t size, pageoff;
vm_prot_t prot, maxprot;
int flags, fd;
vaddr_t vm_min_address = VM_MIN_ADDRESS;
struct filedesc *fdp = p->p_fd;
struct file *fp = NULL;
struct vnode *vp;
caddr_t handle;
int error;
/*
* first, extract syscall args from the uap.
*/
addr = (vaddr_t) SCARG(uap, addr);
size = (vsize_t) SCARG(uap, len);
prot = SCARG(uap, prot);
flags = SCARG(uap, flags);
fd = SCARG(uap, fd);
pos = SCARG(uap, pos);
/*
* Fixup the old deprecated MAP_COPY into MAP_PRIVATE, and
* validate the flags.
*/
if ((prot & VM_PROT_ALL) != prot)
return (EINVAL);
if ((flags & MAP_FLAGMASK) != flags)
return (EINVAL);
if (flags & MAP_COPY)
flags = (flags & ~MAP_COPY) | MAP_PRIVATE;
if ((flags & (MAP_SHARED|MAP_PRIVATE)) == (MAP_SHARED|MAP_PRIVATE))
return (EINVAL);
if (flags & MAP_DENYWRITE)
return (EINVAL);
/*
* align file position and save offset. adjust size.
*/
ALIGN_ADDR(pos, size, pageoff);
/*
* now check (MAP_FIXED) or get (!MAP_FIXED) the "addr"
*/
if (flags & MAP_FIXED) {
/* adjust address by the same amount as we did the offset */
addr -= pageoff;
if (addr & PAGE_MASK)
return (EINVAL); /* not page aligned */
if (addr > SIZE_MAX - size)
return (EINVAL); /* no wrapping! */
if (VM_MAXUSER_ADDRESS > 0 &&
(addr + size) > VM_MAXUSER_ADDRESS)
return (EINVAL);
if (vm_min_address > 0 && addr < vm_min_address)
return (EINVAL);
} else {
/*
* not fixed: make sure we skip over the largest possible heap.
* we will refine our guess later (e.g. to account for VAC, etc)
*/
if (addr == 0)
addr = uvm_map_hint(p, prot);
else if (!(flags & MAP_TRYFIXED) &&
addr < (vaddr_t)p->p_vmspace->vm_daddr)
addr = uvm_map_hint(p, prot);
}
/*
* check for file mappings (i.e. not anonymous) and verify file.
*/
if ((flags & MAP_ANON) == 0) {
if ((fp = fd_getfile(fdp, fd)) == NULL)
return (EBADF);
FREF(fp);
if (fp->f_type != DTYPE_VNODE) {
error = ENODEV; /* only mmap vnodes! */
goto out;
}
vp = (struct vnode *)fp->f_data; /* convert to vnode */
if (vp->v_type != VREG && vp->v_type != VCHR &&
vp->v_type != VBLK) {
error = ENODEV; /* only REG/CHR/BLK support mmap */
goto out;
}
if (vp->v_type == VREG && (pos + size) < pos) {
error = EINVAL; /* no offset wrapping */
goto out;
}
/* special case: catch SunOS style /dev/zero */
if (vp->v_type == VCHR && iszerodev(vp->v_rdev)) {
flags |= MAP_ANON;
FRELE(fp);
fp = NULL;
goto is_anon;
}
/*
* Old programs may not select a specific sharing type, so
* default to an appropriate one.
*
* XXX: how does MAP_ANON fit in the picture?
*/
if ((flags & (MAP_SHARED|MAP_PRIVATE)) == 0) {
#if defined(DEBUG)
printf("WARNING: defaulted mmap() share type to "
"%s (pid %d comm %s)\n", vp->v_type == VCHR ?
"MAP_SHARED" : "MAP_PRIVATE", p->p_pid,
p->p_comm);
#endif
if (vp->v_type == VCHR)
flags |= MAP_SHARED; /* for a device */
else
flags |= MAP_PRIVATE; /* for a file */
}
/*
* MAP_PRIVATE device mappings don't make sense (and aren't
* supported anyway). However, some programs rely on this,
* so just change it to MAP_SHARED.
*/
if (vp->v_type == VCHR && (flags & MAP_PRIVATE) != 0) {
flags = (flags & ~MAP_PRIVATE) | MAP_SHARED;
}
#ifdef ANOUBIS
/* Force DENYWRITE mappings if file->denywrite is set. */
if (fp->denywrite)
flags |= MAP_DENYWRITE;
#endif
/*
* now check protection
*/
/*
* Don't allow the file to be mapped into executable memory if
* the underlying file system is marked as 'noexec'.
*/
if (prot & PROT_EXEC && vp->v_mount->mnt_flag & MNT_NOEXEC) {
error = EACCES;
goto out;
}
maxprot = VM_PROT_EXECUTE;
/* check read access */
if (fp->f_flag & FREAD)
maxprot |= VM_PROT_READ;
else if (prot & PROT_READ) {
error = EACCES;
goto out;
}
/* PROT_EXEC only makes sense if the descriptor is readable. */
if (!(fp->f_flag & FREAD) && prot & PROT_EXEC) {
error = EACCES;
goto out;
}
/* check write access, shared case first */
if (flags & MAP_SHARED) {
/*
* if the file is writable, only add PROT_WRITE to
* maxprot if the file is not immutable, append-only.
* otherwise, if we have asked for PROT_WRITE, return
* EPERM.
*/
if (fp->f_flag & FWRITE) {
if ((error =
VOP_GETATTR(vp, &va, p->p_ucred, p)))
goto out;
if ((va.va_flags & (IMMUTABLE|APPEND)) == 0)
maxprot |= VM_PROT_WRITE;
else if (prot & PROT_WRITE) {
error = EPERM;
goto out;
}
} else if (prot & PROT_WRITE) {
error = EACCES;
goto out;
}
} else {
/* MAP_PRIVATE mappings can always write to */
maxprot |= VM_PROT_WRITE;
}
#ifdef MAC
error = mac_vnode_check_mmap(p->p_ucred, vp, prot, flags);
if (error)
goto out;
#endif
vfs_mark_atime(vp, p->p_ucred);
/*
* set handle to vnode
*/
handle = (caddr_t)vp;
} else { /* MAP_ANON case */
/*
* XXX What do we do about (MAP_SHARED|MAP_PRIVATE) == 0?
*/
if (fd != -1) {
error = EINVAL;
goto out;
}
is_anon: /* label for SunOS style /dev/zero */
handle = NULL;
maxprot = VM_PROT_ALL;
pos = 0;
}
if ((flags & MAP_ANON) != 0 ||
((flags & MAP_PRIVATE) != 0 && (prot & PROT_WRITE) != 0)) {
if (size >
(p->p_rlimit[RLIMIT_DATA].rlim_cur - ptoa(p->p_vmspace->vm_dused))) {
error = ENOMEM;
goto out;
}
}
/*
* now let kernel internal function uvm_mmap do the work.
*/
error = uvm_mmap(&p->p_vmspace->vm_map, &addr, size, prot, maxprot,
flags, handle, pos, p->p_rlimit[RLIMIT_MEMLOCK].rlim_cur, p);
if (error == 0)
/* remember to add offset */
*retval = (register_t)(addr + pageoff);
out:
if (fp)
FRELE(fp);
return (error);
}
int
sys_mquery(struct proc *p, void *v, register_t *retval)
{
struct sys_mquery_args /* {
syscallarg(void *) addr;
syscallarg(size_t) len;
syscallarg(int) prot;
syscallarg(int) flags;
syscallarg(int) fd;
syscallarg(long) pad;
syscallarg(off_t) pos;
} */ *uap = v;
struct file *fp;
struct uvm_object *uobj;
voff_t uoff;
int error;
vaddr_t vaddr;
int flags = 0;
vsize_t size;
vm_prot_t prot;
int fd;
vaddr = (vaddr_t) SCARG(uap, addr);
prot = SCARG(uap, prot);
size = (vsize_t) SCARG(uap, len);
fd = SCARG(uap, fd);
if ((prot & VM_PROT_ALL) != prot)
return (EINVAL);
if (SCARG(uap, flags) & MAP_FIXED)
flags |= UVM_FLAG_FIXED;
if (fd >= 0) {
if ((error = getvnode(p->p_fd, fd, &fp)) != 0)
return (error);
uobj = &((struct vnode *)fp->f_data)->v_uvm.u_obj;
uoff = SCARG(uap, pos);
} else {
fp = NULL;
uobj = NULL;
uoff = 0;
}
if (vaddr == 0)
vaddr = uvm_map_hint(p, prot);
/* prevent a user requested address from falling in heap space */
if ((vaddr + size > (vaddr_t)p->p_vmspace->vm_daddr) &&
(vaddr < (vaddr_t)p->p_vmspace->vm_daddr + MAXDSIZ)) {
if (flags & UVM_FLAG_FIXED) {
error = EINVAL;
goto done;
}
vaddr = round_page((vaddr_t)p->p_vmspace->vm_daddr + MAXDSIZ);
}
vm_map_lock(&p->p_vmspace->vm_map);
again:
if (uvm_map_findspace(&p->p_vmspace->vm_map, vaddr, size,
&vaddr, uobj, uoff, 0, flags) == NULL) {
if (flags & UVM_FLAG_FIXED)
error = EINVAL;
else
error = ENOMEM;
} else {
/* prevent a returned address from falling in heap space */
if ((vaddr + size > (vaddr_t)p->p_vmspace->vm_daddr)
&& (vaddr < (vaddr_t)p->p_vmspace->vm_daddr + MAXDSIZ)) {
vaddr = round_page((vaddr_t)p->p_vmspace->vm_daddr +
MAXDSIZ);
goto again;
}
error = 0;
*retval = (register_t)(vaddr);
}
vm_map_unlock(&p->p_vmspace->vm_map);
done:
if (fp != NULL)
FRELE(fp);
return (error);
}
int
diskmapioctl(dev_t dev, u_long cmd, caddr_t addr, int flag, struct proc *p)
{
struct dk_diskmap *dm;
struct nameidata ndp;
struct filedesc *fdp;
struct file *fp = NULL;
struct vnode *vp = NULL, *ovp;
char *devname;
int fd, error = EINVAL;
if (cmd != DIOCMAP)
return EINVAL;
/*
* Map a request for a disk to the correct device. We should be
* supplied with either a diskname or a disklabel UID.
*/
dm = (struct dk_diskmap *)addr;
fd = dm->fd;
devname = malloc(PATH_MAX, M_DEVBUF, M_WAITOK);
if (copyinstr(dm->device, devname, PATH_MAX, NULL))
goto invalid;
if (disk_map(devname, devname, PATH_MAX, dm->flags) == 0)
if (copyoutstr(devname, dm->device, PATH_MAX, NULL))
goto invalid;
/* Attempt to open actual device. */
fdp = p->p_fd;
fdplock(fdp);
if ((error = getvnode(fdp, fd, &fp)) != 0)
goto bad;
ndp.ni_segflg = UIO_SYSSPACE;
ndp.ni_dirfd = AT_FDCWD;
ndp.ni_dirp = devname;
ndp.ni_cnd.cn_proc = p;
if ((error = vn_open(&ndp, fp->f_flag, 0)) != 0)
goto bad;
vp = ndp.ni_vp;
/* Close the original vnode. */
ovp = (struct vnode *)fp->f_data;
if (fp->f_flag & FWRITE)
ovp->v_writecount--;
if (ovp->v_writecount == 0) {
vn_lock(ovp, LK_EXCLUSIVE | LK_RETRY, p);
VOP_CLOSE(ovp, fp->f_flag, p->p_ucred);
vput(ovp);
}
fp->f_type = DTYPE_VNODE;
fp->f_ops = &vnops;
fp->f_data = (caddr_t)vp;
fp->f_offset = 0;
fp->f_rxfer = 0;
fp->f_wxfer = 0;
fp->f_seek = 0;
fp->f_rbytes = 0;
fp->f_wbytes = 0;
VOP_UNLOCK(vp, 0);
FRELE(fp, p);
fdpunlock(fdp);
free(devname, M_DEVBUF, 0);
return 0;
bad:
if (vp)
vput(vp);
if (fp)
FRELE(fp, p);
fdpunlock(fdp);
invalid:
free(devname, M_DEVBUF, 0);
return (error);
}
