int
at_control(struct socket *so, u_long cmd, caddr_t data,
struct ifnet *ifp, struct thread *td )
{
struct ifreq *ifr = (struct ifreq *)data;
struct sockaddr_at *sat;
struct netrange *nr;
struct at_aliasreq *ifra = (struct at_aliasreq *)data;
struct at_ifaddr *aa0;
struct at_ifaddr *aa = 0;
struct ifaddr *ifa, *ifa0;
int error;
/*
* If we have an ifp, then find the matching at_ifaddr if it exists
*/
if ( ifp ) {
for ( aa = at_ifaddr; aa; aa = aa->aa_next ) {
if ( aa->aa_ifp == ifp ) break;
}
}
/*
* In this first switch table we are basically getting ready for
* the second one, by getting the atalk-specific things set up
* so that they start to look more similar to other protocols etc.
*/
switch ( cmd ) {
case SIOCAIFADDR:
case SIOCDIFADDR:
/*
* If we have an appletalk sockaddr, scan forward of where
* we are now on the at_ifaddr list to find one with a matching
* address on this interface.
* This may leave aa pointing to the first address on the
* NEXT interface!
*/
if ( ifra->ifra_addr.sat_family == AF_APPLETALK ) {
for ( ; aa; aa = aa->aa_next ) {
if ( aa->aa_ifp == ifp &&
sateqaddr( &aa->aa_addr, &ifra->ifra_addr )) {
break;
}
}
}
/*
* If we a retrying to delete an addres but didn't find such,
* then rewurn with an error
*/
if ( cmd == SIOCDIFADDR && aa == 0 ) {
return( EADDRNOTAVAIL );
}
/*FALLTHROUGH*/
case SIOCSIFADDR:
/*
* If we are not superuser, then we don't get to do these ops.
*/
if (priv_check(td, PRIV_ROOT))
return(EPERM);
sat = satosat( &ifr->ifr_addr );
nr = (struct netrange *)sat->sat_zero;
if ( nr->nr_phase == 1 ) {
/*
* Look for a phase 1 address on this interface.
* This may leave aa pointing to the first address on the
* NEXT interface!
*/
for ( ; aa; aa = aa->aa_next ) {
if ( aa->aa_ifp == ifp &&
( aa->aa_flags & AFA_PHASE2 ) == 0 ) {
break;
}
}
} else { /* default to phase 2 */
/*
* Look for a phase 2 address on this interface.
* This may leave aa pointing to the first address on the
* NEXT interface!
*/
for ( ; aa; aa = aa->aa_next ) {
if ( aa->aa_ifp == ifp && ( aa->aa_flags & AFA_PHASE2 )) {
break;
}
}
}
if ( ifp == 0 )
panic( "at_control" );
/*
* If we failed to find an existing at_ifaddr entry, then we
* allocate a fresh one.
*/
if ( aa == NULL ) {
aa0 = ifa_create(sizeof(struct at_ifaddr), M_WAITOK);
callout_init(&aa0->aa_ch);
if (( aa = at_ifaddr ) != NULL ) {
/*
* Don't let the loopback be first, since the first
* address is the machine's default address for
* binding.
* If it is, stick ourself in front, otherwise
* go to the back of the list.
*/
if ( at_ifaddr->aa_ifp->if_flags & IFF_LOOPBACK ) {
aa = aa0;
aa->aa_next = at_ifaddr;
at_ifaddr = aa;
} else {
for ( ; aa->aa_next; aa = aa->aa_next )
;
aa->aa_next = aa0;
}
} else {
at_ifaddr = aa0;
}
aa = aa0;
/*
* Find the end of the interface's addresses
* and link our new one on the end
*/
ifa = (struct ifaddr *)aa;
ifa_iflink(ifa, ifp, 1);
/*
* As the at_ifaddr contains the actual sockaddrs,
* and the ifaddr itself, link them al together correctly.
*/
ifa->ifa_addr = (struct sockaddr *)&aa->aa_addr;
ifa->ifa_dstaddr = (struct sockaddr *)&aa->aa_addr;
ifa->ifa_netmask = (struct sockaddr *)&aa->aa_netmask;
/*
* Set/clear the phase 2 bit.
*/
if ( nr->nr_phase == 1 ) {
aa->aa_flags &= ~AFA_PHASE2;
} else {
aa->aa_flags |= AFA_PHASE2;
}
/*
* and link it all together
*/
aa->aa_ifp = ifp;
} else {
/*
* If we DID find one then we clobber any routes dependent on it..
*/
at_scrub( ifp, aa );
}
break;
case SIOCGIFADDR :
sat = satosat( &ifr->ifr_addr );
nr = (struct netrange *)sat->sat_zero;
if ( nr->nr_phase == 1 ) {
/*
* If the request is specifying phase 1, then
* only look at a phase one address
*/
for ( ; aa; aa = aa->aa_next ) {
if ( aa->aa_ifp == ifp &&
( aa->aa_flags & AFA_PHASE2 ) == 0 ) {
break;
}
}
} else {
/*
* default to phase 2
*/
for ( ; aa; aa = aa->aa_next ) {
if ( aa->aa_ifp == ifp && ( aa->aa_flags & AFA_PHASE2 )) {
break;
}
}
}
if ( aa == NULL )
return( EADDRNOTAVAIL );
break;
}
/*
* By the time this switch is run we should be able to assume that
* the "aa" pointer is valid when needed.
*/
switch ( cmd ) {
case SIOCGIFADDR:
/*
* copy the contents of the sockaddr blindly.
*/
sat = (struct sockaddr_at *)&ifr->ifr_addr;
*sat = aa->aa_addr;
/*
* and do some cleanups
*/
((struct netrange *)&sat->sat_zero)->nr_phase
= (aa->aa_flags & AFA_PHASE2) ? 2 : 1;
((struct netrange *)&sat->sat_zero)->nr_firstnet = aa->aa_firstnet;
((struct netrange *)&sat->sat_zero)->nr_lastnet = aa->aa_lastnet;
break;
case SIOCSIFADDR:
return( at_ifinit( ifp, aa, (struct sockaddr_at *)&ifr->ifr_addr ));
case SIOCAIFADDR:
if ( sateqaddr( &ifra->ifra_addr, &aa->aa_addr )) {
return( 0 );
}
return( at_ifinit( ifp, aa, (struct sockaddr_at *)&ifr->ifr_addr ));
case SIOCDIFADDR:
/*
* scrub all routes.. didn't we just DO this? XXX yes, del it
*/
at_scrub( ifp, aa );
/*
* remove the ifaddr from the interface
*/
ifa0 = (struct ifaddr *)aa;
ifa_ifunlink(ifa0, ifp);
/*
* Now remove the at_ifaddr from the parallel structure
* as well, or we'd be in deep trouble
*/
aa0 = aa;
if ( aa0 == ( aa = at_ifaddr )) {
at_ifaddr = aa->aa_next;
} else {
while ( aa->aa_next && ( aa->aa_next != aa0 )) {
aa = aa->aa_next;
}
/*
* if we found it, remove it, otherwise we screwed up.
*/
if ( aa->aa_next ) {
aa->aa_next = aa0->aa_next;
} else {
panic( "at_control" );
}
}
/*
* Now dump the memory we were using.
* Decrement the reference count.
* This should probably be the last reference
* as the count will go from 1 to 0.
* (unless there is still a route referencing this)
*/
ifa_destroy(ifa0);
break;
default:
if ( ifp == 0 || ifp->if_ioctl == 0 )
return( EOPNOTSUPP );
ifnet_serialize_all(ifp);
error = ifp->if_ioctl(ifp, cmd, data, td->td_proc->p_ucred);
ifnet_deserialize_all(ifp);
return (error);
}
return( 0 );
}
struct tty * tty_alloc(const char * drv_name, dev_t dev_id,
const char * dev_name, size_t data_size)
{
struct dev_info * dev;
struct tty * tty;
dev = kzalloc(sizeof(struct dev_info) + sizeof(struct tty) + data_size);
if (!dev)
return NULL;
tty = (struct tty *)((uintptr_t)dev + sizeof(struct dev_info));
dev->dev_id = dev_id;
dev->drv_name = drv_name;
strlcpy(dev->dev_name, dev_name, sizeof(dev->dev_name));
dev->flags = DEV_FLAGS_MB_READ | DEV_FLAGS_WR_BT_MASK;
dev->block_size = 1;
dev->read = tty_read;
dev->write = tty_write;
dev->lseek = tty_lseek;
dev->open_callback = tty_open_callback;
dev->close_callback = tty_close_callback;
dev->ioctl = tty_ioctl;
dev->opt_data = tty;
/*
* Linux defaults:
* tty->conf.c_lflag = ISIG | ICANON | ECHO | ECHOE | ECHOK | ECHOCTL |
* ECHOKE | IEXTEN;
*/
/*
* TODO termios support
* *supported now*
* iflags: -
* oflags: -
* cflags: -
* lfags: -
*/
/* TODO Better winsize support. */
tty->winsize = (struct winsize){
.ws_row = 24,
.ws_col = 80,
.ws_xpixel = 0,
.ws_ypixel = 0,
};
return tty;
}
void tty_free(struct tty * tty)
{
struct dev_info * dev;
dev = (struct dev_info *)((uintptr_t)tty - sizeof(struct dev_info));
KASSERT(dev->opt_data == tty, "opt_data changed or invalid tty");
kfree(dev);
}
struct dev_info * tty_get_dev(struct tty * tty)
{
return (struct dev_info *)((uintptr_t)tty - sizeof(struct dev_info));
}
int make_ttydev(struct tty * tty)
{
struct dev_info * dev;
vnode_t * vn;
dev = (struct dev_info *)((uintptr_t)tty - sizeof(struct dev_info));
KASSERT(dev->opt_data == tty, "opt_data changed or invalid tty");
if (tty->tty_vn != NULL) {
KERROR(KERROR_ERR, "A device file is already created for this tty\n");
return -EMLINK;
}
if (make_dev(dev, 0, 0, 0666, &vn)) {
KERROR(KERROR_ERR, "Failed to make a tty dev.\n");
return -ENODEV;
}
tty->tty_vn = vn;
return 0;
}
void destroy_ttydev(struct tty * tty)
{
struct dev_info * dev;
dev = (struct dev_info *)((uintptr_t)tty - sizeof(struct dev_info));
KASSERT(dev->opt_data == tty, "opt_data changed or invalid tty");
destroy_dev(tty->tty_vn);
}
static ssize_t tty_read(struct dev_info * devnfo, off_t blkno, uint8_t * buf,
size_t bcount, int oflags)
{
struct tty * tty = (struct tty *)devnfo->opt_data;
KASSERT(tty, "opt_data should have a tty");
return tty->read(tty, blkno, buf, bcount, oflags);
}
static ssize_t tty_write(struct dev_info * devnfo, off_t blkno, uint8_t * buf,
size_t bcount, int oflags)
{
struct tty * tty = (struct tty *)devnfo->opt_data;
ssize_t retval;
KASSERT(tty, "opt_data should have a tty");
retval = tty->write(tty, blkno, buf, bcount, oflags);
if (retval > 0) {
off_t n = tty->write_count + retval;
tty->write_count = (n < 0) ? -n : n;
}
return retval;
}
static off_t tty_lseek(file_t * file, struct dev_info * devnfo, off_t offset,
int whence)
{
struct tty * tty = (struct tty *)devnfo->opt_data;
/*
* Many unices will return the number of written characters if whence is
* SEEK_SET and the file is a tty, and some will return -ESPIPE. We support
* the write count.
*/
if (whence == SEEK_SET)
return tty->write_count;
/*
* Some drivers may use seek_pos as an index variable and on this kernel
* we promise to return it if lseek is called with offset zero and SEEK_CUR
* set as a whence.
*/
if (offset == 0 && whence == SEEK_CUR)
return file->seek_pos;
return -ESPIPE;
}
static void tty_open_callback(struct proc_info * p, file_t * file,
struct dev_info * devnfo)
{
struct tty * tty = (struct tty *)devnfo->opt_data;
KASSERT(tty, "opt_data should have a tty");
if (tty->open_callback)
tty->open_callback(file, tty);
}
static void tty_close_callback(struct proc_info * p, file_t * file,
struct dev_info * devnfo)
{
struct tty * tty = (struct tty *)devnfo->opt_data;
KASSERT(tty, "opt_data should have a tty");
if (tty->close_callback)
tty->close_callback(file, tty);
}
static int tty_ioctl(struct dev_info * devnfo, uint32_t request,
void * arg, size_t arg_len)
{
int err;
struct tty * tty = (struct tty *)(devnfo->opt_data);
if (!tty)
return -EINVAL;
/*
* First call ioctl of the device driver since it may override some ioctls
* defined here.
*/
if (tty->ioctl) {
err = tty->ioctl(devnfo, request, arg, arg_len);
if (err == 0 || err != -EINVAL)
return err;
} /* otherwise check if we can handle it here */
switch (request) {
case IOCTL_GTERMIOS:
if (arg_len < sizeof(struct termios))
return -EINVAL;
memcpy(arg, &(tty->conf), sizeof(struct termios));
break;
case IOCTL_STERMIOS:
if (arg_len < sizeof(struct termios))
return -EINVAL;
err = priv_check(&curproc->cred, PRIV_TTY_SETA);
if (err)
return err;
memcpy(&(tty->conf), arg, sizeof(struct termios));
tty->setconf(&tty->conf);
break;
case IOCTL_TIOCGWINSZ:
if (arg_len < sizeof(struct winsize))
return -EINVAL;
memcpy(arg, &(tty->winsize), sizeof(struct winsize));
break;
case IOCTL_TIOCSWINSZ:
if (arg_len < sizeof(struct winsize))
return -EINVAL;
memcpy(&(tty->winsize), arg, sizeof(struct winsize));
break;
/*
* This should be probably overriden and "optimized" in
* the low level driver. Also if there is any muxing on
* any lower level flush may do stupid things if done
* by this function.
*/
case IOCTL_TTYFLUSH:
if (arg_len < sizeof(int)) {
return -EINVAL;
} else {
int control = (int)arg;
uint8_t buf[5];
switch (control) {
case TCIFLUSH:
while (tty->read(tty, 0, buf, sizeof(buf), O_NONBLOCK) >= 0);
break;
default:
return -EINVAL;
}
}
break;
case IOCTL_TCSBRK:
/* NOP */
break;
default:
return -EINVAL;
}
return 0;
}
static int
ugen_ioctl_post(struct usb_fifo *f, u_long cmd, void *addr, int fflags)
{
union {
struct usb_interface_descriptor *idesc;
struct usb_alt_interface *ai;
struct usb_device_descriptor *ddesc;
struct usb_config_descriptor *cdesc;
struct usb_device_stats *stat;
struct usb_fs_init *pinit;
struct usb_fs_uninit *puninit;
uint32_t *ptime;
void *addr;
int *pint;
} u;
struct usb_device_descriptor *dtemp;
struct usb_config_descriptor *ctemp;
struct usb_interface *iface;
int error = 0;
uint8_t n;
u.addr = addr;
DPRINTFN(6, "cmd=0x%08lx\n", cmd);
switch (cmd) {
case USB_DISCOVER:
usb_needs_explore_all();
break;
case USB_SETDEBUG:
if (!(fflags & FWRITE)) {
error = EPERM;
break;
}
usb_debug = *(int *)addr;
break;
case USB_GET_CONFIG:
*(int *)addr = f->udev->curr_config_index;
break;
case USB_SET_CONFIG:
if (!(fflags & FWRITE)) {
error = EPERM;
break;
}
error = ugen_set_config(f, *(int *)addr);
break;
case USB_GET_ALTINTERFACE:
iface = usbd_get_iface(f->udev,
u.ai->uai_interface_index);
if (iface && iface->idesc) {
u.ai->uai_alt_index = iface->alt_index;
} else {
error = EINVAL;
}
break;
case USB_SET_ALTINTERFACE:
if (!(fflags & FWRITE)) {
error = EPERM;
break;
}
error = ugen_set_interface(f,
u.ai->uai_interface_index, u.ai->uai_alt_index);
break;
case USB_GET_DEVICE_DESC:
dtemp = usbd_get_device_descriptor(f->udev);
if (!dtemp) {
error = EIO;
break;
}
*u.ddesc = *dtemp;
break;
case USB_GET_CONFIG_DESC:
ctemp = usbd_get_config_descriptor(f->udev);
if (!ctemp) {
error = EIO;
break;
}
*u.cdesc = *ctemp;
break;
case USB_GET_FULL_DESC:
error = ugen_get_cdesc(f, addr);
break;
case USB_GET_STRING_DESC:
error = ugen_get_sdesc(f, addr);
break;
case USB_GET_IFACE_DRIVER:
error = ugen_get_iface_driver(f, addr);
break;
case USB_REQUEST:
case USB_DO_REQUEST:
if (!(fflags & FWRITE)) {
error = EPERM;
break;
}
error = ugen_do_request(f, addr);
break;
case USB_DEVICEINFO:
case USB_GET_DEVICEINFO:
error = usb_gen_fill_deviceinfo(f, addr);
break;
case USB_DEVICESTATS:
for (n = 0; n != 4; n++) {
u.stat->uds_requests_fail[n] =
f->udev->bus->stats_err.uds_requests[n];
u.stat->uds_requests_ok[n] =
f->udev->bus->stats_ok.uds_requests[n];
}
break;
case USB_DEVICEENUMERATE:
error = ugen_re_enumerate(f);
break;
case USB_GET_PLUGTIME:
*u.ptime = f->udev->plugtime;
break;
case USB_CLAIM_INTERFACE:
case USB_RELEASE_INTERFACE:
/* TODO */
break;
case USB_IFACE_DRIVER_ACTIVE:
n = *u.pint & 0xFF;
iface = usbd_get_iface(f->udev, n);
if (iface && iface->subdev)
error = 0;
else
error = ENXIO;
break;
case USB_IFACE_DRIVER_DETACH:
error = priv_check(curthread, PRIV_DRIVER);
if (error)
break;
n = *u.pint & 0xFF;
if (n == USB_IFACE_INDEX_ANY) {
error = EINVAL;
break;
}
/*
* Detach the currently attached driver.
*/
usb_detach_device(f->udev, n, 0);
/*
* Set parent to self, this should keep attach away
* until the next set configuration event.
*/
usbd_set_parent_iface(f->udev, n, n);
break;
case USB_SET_POWER_MODE:
error = ugen_set_power_mode(f, *u.pint);
break;
case USB_GET_POWER_MODE:
*u.pint = ugen_get_power_mode(f);
break;
case USB_GET_POWER_USAGE:
*u.pint = ugen_get_power_usage(f);
break;
case USB_SET_PORT_ENABLE:
error = ugen_do_port_feature(f,
*u.pint, 1, UHF_PORT_ENABLE);
break;
case USB_SET_PORT_DISABLE:
error = ugen_do_port_feature(f,
*u.pint, 0, UHF_PORT_ENABLE);
break;
case USB_FS_INIT:
/* verify input parameters */
if (u.pinit->pEndpoints == NULL) {
error = EINVAL;
break;
}
if (u.pinit->ep_index_max > 127) {
error = EINVAL;
break;
}
if (u.pinit->ep_index_max == 0) {
error = EINVAL;
break;
}
if (f->fs_xfer != NULL) {
error = EBUSY;
break;
}
if (f->dev_ep_index != 0) {
error = EINVAL;
break;
}
if (ugen_fifo_in_use(f, fflags)) {
error = EBUSY;
break;
}
error = usb_fifo_alloc_buffer(f, 1, u.pinit->ep_index_max);
if (error) {
break;
}
f->fs_xfer = malloc(sizeof(f->fs_xfer[0]) *
u.pinit->ep_index_max, M_USB, M_WAITOK | M_ZERO);
if (f->fs_xfer == NULL) {
usb_fifo_free_buffer(f);
error = ENOMEM;
break;
}
f->fs_ep_max = u.pinit->ep_index_max;
f->fs_ep_ptr = u.pinit->pEndpoints;
break;
case USB_FS_UNINIT:
if (u.puninit->dummy != 0) {
error = EINVAL;
break;
}
error = ugen_fs_uninit(f);
break;
default:
mtx_lock(f->priv_mtx);
error = ugen_iface_ioctl(f, cmd, addr, fflags);
mtx_unlock(f->priv_mtx);
break;
}
DPRINTFN(6, "error=%d\n", error);
return (error);
}
/* ARGSUSED */
int
sys_auditctl(struct thread *td, struct auditctl_args *uap)
{
struct nameidata nd;
struct ucred *cred;
struct vnode *vp;
int error = 0;
int flags, vfslocked;
if (jailed(td->td_ucred))
return (ENOSYS);
error = priv_check(td, PRIV_AUDIT_CONTROL);
if (error)
return (error);
vp = NULL;
cred = NULL;
/*
* If a path is specified, open the replacement vnode, perform
* validity checks, and grab another reference to the current
* credential.
*
* On Darwin, a NULL path argument is also used to disable audit.
*/
if (uap->path == NULL)
return (EINVAL);
NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | MPSAFE | AUDITVNODE1,
UIO_USERSPACE, uap->path, td);
flags = AUDIT_OPEN_FLAGS;
error = vn_open(&nd, &flags, 0, NULL);
if (error)
return (error);
vfslocked = NDHASGIANT(&nd);
vp = nd.ni_vp;
#ifdef MAC
error = mac_system_check_auditctl(td->td_ucred, vp);
VOP_UNLOCK(vp, 0);
if (error) {
vn_close(vp, AUDIT_CLOSE_FLAGS, td->td_ucred, td);
VFS_UNLOCK_GIANT(vfslocked);
return (error);
}
#else
VOP_UNLOCK(vp, 0);
#endif
NDFREE(&nd, NDF_ONLY_PNBUF);
if (vp->v_type != VREG) {
vn_close(vp, AUDIT_CLOSE_FLAGS, td->td_ucred, td);
VFS_UNLOCK_GIANT(vfslocked);
return (EINVAL);
}
VFS_UNLOCK_GIANT(vfslocked);
cred = td->td_ucred;
crhold(cred);
/*
* XXXAUDIT: Should audit_suspended actually be cleared by
* audit_worker?
*/
audit_suspended = 0;
audit_rotate_vnode(cred, vp);
return (error);
}
/*
* Q_QUOTAON - set up a quota file for a particular filesystem.
*/
int
quotaon(struct thread *td, struct mount *mp, int type, void *fname)
{
struct ufsmount *ump;
struct vnode *vp, **vpp;
struct vnode *mvp;
struct dquot *dq;
int error, flags;
struct nameidata nd;
error = priv_check(td, PRIV_UFS_QUOTAON);
if (error != 0) {
vfs_unbusy(mp);
return (error);
}
if ((mp->mnt_flag & MNT_RDONLY) != 0) {
vfs_unbusy(mp);
return (EROFS);
}
ump = VFSTOUFS(mp);
dq = NODQUOT;
NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, fname, td);
flags = FREAD | FWRITE;
vfs_ref(mp);
vfs_unbusy(mp);
error = vn_open(&nd, &flags, 0, NULL);
if (error != 0) {
vfs_rel(mp);
return (error);
}
NDFREE(&nd, NDF_ONLY_PNBUF);
vp = nd.ni_vp;
error = vfs_busy(mp, MBF_NOWAIT);
vfs_rel(mp);
if (error == 0) {
if (vp->v_type != VREG) {
error = EACCES;
vfs_unbusy(mp);
}
}
if (error != 0) {
VOP_UNLOCK(vp, 0);
(void) vn_close(vp, FREAD|FWRITE, td->td_ucred, td);
return (error);
}
UFS_LOCK(ump);
if ((ump->um_qflags[type] & (QTF_OPENING|QTF_CLOSING)) != 0) {
UFS_UNLOCK(ump);
VOP_UNLOCK(vp, 0);
(void) vn_close(vp, FREAD|FWRITE, td->td_ucred, td);
vfs_unbusy(mp);
return (EALREADY);
}
ump->um_qflags[type] |= QTF_OPENING|QTF_CLOSING;
UFS_UNLOCK(ump);
if ((error = dqopen(vp, ump, type)) != 0) {
VOP_UNLOCK(vp, 0);
UFS_LOCK(ump);
ump->um_qflags[type] &= ~(QTF_OPENING|QTF_CLOSING);
UFS_UNLOCK(ump);
(void) vn_close(vp, FREAD|FWRITE, td->td_ucred, td);
vfs_unbusy(mp);
return (error);
}
VOP_UNLOCK(vp, 0);
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_QUOTA;
MNT_IUNLOCK(mp);
vpp = &ump->um_quotas[type];
if (*vpp != vp)
quotaoff1(td, mp, type);
/*
* When the directory vnode containing the quota file is
* inactivated, due to the shared lookup of the quota file
* vput()ing the dvp, the qsyncvp() call for the containing
* directory would try to acquire the quota lock exclusive.
* At the same time, lookup already locked the quota vnode
* shared. Mark the quota vnode lock as allowing recursion
* and automatically converting shared locks to exclusive.
*
* Also mark quota vnode as system.
*/
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
vp->v_vflag |= VV_SYSTEM;
VN_LOCK_AREC(vp);
VN_LOCK_DSHARE(vp);
VOP_UNLOCK(vp, 0);
*vpp = vp;
/*
* Save the credential of the process that turned on quotas.
* Set up the time limits for this quota.
*/
ump->um_cred[type] = crhold(td->td_ucred);
ump->um_btime[type] = MAX_DQ_TIME;
ump->um_itime[type] = MAX_IQ_TIME;
if (dqget(NULLVP, 0, ump, type, &dq) == 0) {
if (dq->dq_btime > 0)
ump->um_btime[type] = dq->dq_btime;
if (dq->dq_itime > 0)
ump->um_itime[type] = dq->dq_itime;
dqrele(NULLVP, dq);
}
/*
* Allow the getdq from getinoquota below to read the quota
* from file.
*/
UFS_LOCK(ump);
ump->um_qflags[type] &= ~QTF_CLOSING;
UFS_UNLOCK(ump);
/*
* Search vnodes associated with this mount point,
* adding references to quota file being opened.
* NB: only need to add dquot's for inodes being modified.
*/
again:
MNT_VNODE_FOREACH_ALL(vp, mp, mvp) {
if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK, td)) {
MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
goto again;
}
if (vp->v_type == VNON || vp->v_writecount == 0) {
VOP_UNLOCK(vp, 0);
vrele(vp);
continue;
}
error = getinoquota(VTOI(vp));
VOP_UNLOCK(vp, 0);
vrele(vp);
if (error) {
MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
break;
}
}
if (error)
quotaoff_inchange(td, mp, type);
UFS_LOCK(ump);
ump->um_qflags[type] &= ~QTF_OPENING;
KASSERT((ump->um_qflags[type] & QTF_CLOSING) == 0,
("quotaon: leaking flags"));
UFS_UNLOCK(ump);
vfs_unbusy(mp);
return (error);
}
static int
cd9660_mount(struct mount *mp)
{
struct vnode *devvp;
struct thread *td;
char *fspec;
int error;
accmode_t accmode;
struct nameidata ndp;
struct iso_mnt *imp = NULL;
td = curthread;
/*
* Unconditionally mount as read-only.
*/
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_RDONLY;
MNT_IUNLOCK(mp);
fspec = vfs_getopts(mp->mnt_optnew, "from", &error);
if (error)
return (error);
imp = VFSTOISOFS(mp);
if (mp->mnt_flag & MNT_UPDATE) {
if (vfs_flagopt(mp->mnt_optnew, "export", NULL, 0))
return (0);
}
/*
* Not an update, or updating the name: look up the name
* and verify that it refers to a sensible block device.
*/
NDINIT(&ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, fspec, td);
if ((error = namei(&ndp)))
return (error);
NDFREE(&ndp, NDF_ONLY_PNBUF);
devvp = ndp.ni_vp;
if (!vn_isdisk(devvp, &error)) {
vput(devvp);
return (error);
}
/*
* Verify that user has necessary permissions on the device,
* or has superuser abilities
*/
accmode = VREAD;
error = VOP_ACCESS(devvp, accmode, td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
vput(devvp);
return (error);
}
if ((mp->mnt_flag & MNT_UPDATE) == 0) {
error = iso_mountfs(devvp, mp);
if (error)
vrele(devvp);
} else {
if (devvp != imp->im_devvp)
error = EINVAL; /* needs translation */
vput(devvp);
}
if (error)
return (error);
vfs_mountedfrom(mp, fspec);
return (0);
}
static int
fuse_vfsop_mount(struct mount *mp)
{
int err;
uint64_t mntopts, __mntopts;
int max_read_set;
uint32_t max_read;
int daemon_timeout;
int fd;
size_t len;
struct cdev *fdev;
struct fuse_data *data;
struct thread *td;
struct file *fp, *fptmp;
char *fspec, *subtype;
struct vfsoptlist *opts;
subtype = NULL;
max_read_set = 0;
max_read = ~0;
err = 0;
mntopts = 0;
__mntopts = 0;
td = curthread;
fuse_trace_printf_vfsop();
if (mp->mnt_flag & MNT_UPDATE)
return EOPNOTSUPP;
mp->mnt_flag |= MNT_SYNCHRONOUS;
mp->mnt_data = NULL;
/* Get the new options passed to mount */
opts = mp->mnt_optnew;
if (!opts)
return EINVAL;
/* `fspath' contains the mount point (eg. /mnt/fuse/sshfs); REQUIRED */
if (!vfs_getopts(opts, "fspath", &err))
return err;
/* `from' contains the device name (eg. /dev/fuse0); REQUIRED */
fspec = vfs_getopts(opts, "from", &err);
if (!fspec)
return err;
/* `fd' contains the filedescriptor for this session; REQUIRED */
if (vfs_scanopt(opts, "fd", "%d", &fd) != 1)
return EINVAL;
err = fuse_getdevice(fspec, td, &fdev);
if (err != 0)
return err;
/*
* With the help of underscored options the mount program
* can inform us from the flags it sets by default
*/
FUSE_FLAGOPT(allow_other, FSESS_DAEMON_CAN_SPY);
FUSE_FLAGOPT(push_symlinks_in, FSESS_PUSH_SYMLINKS_IN);
FUSE_FLAGOPT(default_permissions, FSESS_DEFAULT_PERMISSIONS);
FUSE_FLAGOPT(no_attrcache, FSESS_NO_ATTRCACHE);
FUSE_FLAGOPT(no_readahed, FSESS_NO_READAHEAD);
FUSE_FLAGOPT(no_datacache, FSESS_NO_DATACACHE);
FUSE_FLAGOPT(no_namecache, FSESS_NO_NAMECACHE);
FUSE_FLAGOPT(no_mmap, FSESS_NO_MMAP);
FUSE_FLAGOPT(brokenio, FSESS_BROKENIO);
if (vfs_scanopt(opts, "max_read=", "%u", &max_read) == 1)
max_read_set = 1;
if (vfs_scanopt(opts, "timeout=", "%u", &daemon_timeout) == 1) {
if (daemon_timeout < FUSE_MIN_DAEMON_TIMEOUT)
daemon_timeout = FUSE_MIN_DAEMON_TIMEOUT;
else if (daemon_timeout > FUSE_MAX_DAEMON_TIMEOUT)
daemon_timeout = FUSE_MAX_DAEMON_TIMEOUT;
} else {
daemon_timeout = FUSE_DEFAULT_DAEMON_TIMEOUT;
}
subtype = vfs_getopts(opts, "subtype=", &err);
FS_DEBUG2G("mntopts 0x%jx\n", (uintmax_t)mntopts);
err = fget(td, fd, CAP_READ, &fp);
if (err != 0) {
FS_DEBUG("invalid or not opened device: data=%p\n", data);
goto out;
}
fptmp = td->td_fpop;
td->td_fpop = fp;
err = devfs_get_cdevpriv((void **)&data);
td->td_fpop = fptmp;
fdrop(fp, td);
FUSE_LOCK();
if (err != 0 || data == NULL || data->mp != NULL) {
FS_DEBUG("invalid or not opened device: data=%p data.mp=%p\n",
data, data != NULL ? data->mp : NULL);
err = ENXIO;
FUSE_UNLOCK();
goto out;
}
if (fdata_get_dead(data)) {
FS_DEBUG("device is dead during mount: data=%p\n", data);
err = ENOTCONN;
FUSE_UNLOCK();
goto out;
}
/* Sanity + permission checks */
if (!data->daemoncred)
panic("fuse daemon found, but identity unknown");
if (mntopts & FSESS_DAEMON_CAN_SPY)
err = priv_check(td, PRIV_VFS_FUSE_ALLOWOTHER);
if (err == 0 && td->td_ucred->cr_uid != data->daemoncred->cr_uid)
/* are we allowed to do the first mount? */
err = priv_check(td, PRIV_VFS_FUSE_MOUNT_NONUSER);
if (err) {
FUSE_UNLOCK();
goto out;
}
/* We need this here as this slot is used by getnewvnode() */
mp->mnt_stat.f_iosize = PAGE_SIZE;
mp->mnt_data = data;
data->ref++;
data->mp = mp;
data->dataflags |= mntopts;
data->max_read = max_read;
data->daemon_timeout = daemon_timeout;
#ifdef XXXIP
if (!priv_check(td, PRIV_VFS_FUSE_SYNC_UNMOUNT))
data->dataflags |= FSESS_CAN_SYNC_UNMOUNT;
#endif
FUSE_UNLOCK();
vfs_getnewfsid(mp);
mp->mnt_flag |= MNT_LOCAL;
mp->mnt_kern_flag |= MNTK_MPSAFE;
if (subtype) {
strlcat(mp->mnt_stat.f_fstypename, ".", MFSNAMELEN);
strlcat(mp->mnt_stat.f_fstypename, subtype, MFSNAMELEN);
}
copystr(fspec, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, &len);
bzero(mp->mnt_stat.f_mntfromname + len, MNAMELEN - len);
FS_DEBUG2G("mp %p: %s\n", mp, mp->mnt_stat.f_mntfromname);
/* Now handshaking with daemon */
fuse_internal_send_init(data, td);
out:
if (err) {
FUSE_LOCK();
if (data->mp == mp) {
/*
* Destroy device only if we acquired reference to
* it
*/
FS_DEBUG("mount failed, destroy device: data=%p mp=%p"
" err=%d\n",
data, mp, err);
data->mp = NULL;
fdata_trydestroy(data);
}
FUSE_UNLOCK();
dev_rel(fdev);
}
return err;
}
/*
* mp - path - addr in user space of mount point (ie /usr or whatever)
* data - addr in user space of mount params including the name of the block
* special file to treat as a filesystem.
*/
static int
msdosfs_mount(struct mount *mp)
{
struct vnode *devvp; /* vnode for blk device to mount */
struct thread *td;
/* msdosfs specific mount control block */
struct msdosfsmount *pmp = NULL;
struct nameidata ndp;
int error, flags;
accmode_t accmode;
char *from;
td = curthread;
if (vfs_filteropt(mp->mnt_optnew, msdosfs_opts))
return (EINVAL);
/*
* If updating, check whether changing from read-only to
* read/write; if there is no device name, that's all we do.
*/
if (mp->mnt_flag & MNT_UPDATE) {
pmp = VFSTOMSDOSFS(mp);
if (vfs_flagopt(mp->mnt_optnew, "export", NULL, 0)) {
/*
* Forbid export requests if filesystem has
* MSDOSFS_LARGEFS flag set.
*/
if ((pmp->pm_flags & MSDOSFS_LARGEFS) != 0) {
vfs_mount_error(mp,
"MSDOSFS_LARGEFS flag set, cannot export");
return (EOPNOTSUPP);
}
}
if (!(pmp->pm_flags & MSDOSFSMNT_RONLY) &&
vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0)) {
error = VFS_SYNC(mp, MNT_WAIT);
if (error)
return (error);
flags = WRITECLOSE;
if (mp->mnt_flag & MNT_FORCE)
flags |= FORCECLOSE;
error = vflush(mp, 0, flags, td);
if (error)
return (error);
/*
* Now the volume is clean. Mark it so while the
* device is still rw.
*/
error = markvoldirty(pmp, 0);
if (error) {
(void)markvoldirty(pmp, 1);
return (error);
}
/* Downgrade the device from rw to ro. */
DROP_GIANT();
g_topology_lock();
error = g_access(pmp->pm_cp, 0, -1, 0);
g_topology_unlock();
PICKUP_GIANT();
if (error) {
(void)markvoldirty(pmp, 1);
return (error);
}
/*
* Backing out after an error was painful in the
* above. Now we are committed to succeeding.
*/
pmp->pm_fmod = 0;
pmp->pm_flags |= MSDOSFSMNT_RONLY;
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_RDONLY;
MNT_IUNLOCK(mp);
} else if ((pmp->pm_flags & MSDOSFSMNT_RONLY) &&
!vfs_flagopt(mp->mnt_optnew, "ro", NULL, 0)) {
/*
* If upgrade to read-write by non-root, then verify
* that user has necessary permissions on the device.
*/
devvp = pmp->pm_devvp;
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
error = VOP_ACCESS(devvp, VREAD | VWRITE,
td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
VOP_UNLOCK(devvp, 0);
return (error);
}
VOP_UNLOCK(devvp, 0);
DROP_GIANT();
g_topology_lock();
error = g_access(pmp->pm_cp, 0, 1, 0);
g_topology_unlock();
PICKUP_GIANT();
if (error)
return (error);
pmp->pm_fmod = 1;
pmp->pm_flags &= ~MSDOSFSMNT_RONLY;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_RDONLY;
MNT_IUNLOCK(mp);
/* Now that the volume is modifiable, mark it dirty. */
error = markvoldirty(pmp, 1);
if (error)
return (error);
}
}
/*
* Not an update, or updating the name: look up the name
* and verify that it refers to a sensible disk device.
*/
if (vfs_getopt(mp->mnt_optnew, "from", (void **)&from, NULL))
return (EINVAL);
NDINIT(&ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, from, td);
error = namei(&ndp);
if (error)
return (error);
devvp = ndp.ni_vp;
NDFREE(&ndp, NDF_ONLY_PNBUF);
if (!vn_isdisk(devvp, &error)) {
vput(devvp);
return (error);
}
/*
* If mount by non-root, then verify that user has necessary
* permissions on the device.
*/
accmode = VREAD;
if ((mp->mnt_flag & MNT_RDONLY) == 0)
accmode |= VWRITE;
error = VOP_ACCESS(devvp, accmode, td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
vput(devvp);
return (error);
}
if ((mp->mnt_flag & MNT_UPDATE) == 0) {
error = mountmsdosfs(devvp, mp);
#ifdef MSDOSFS_DEBUG /* only needed for the printf below */
pmp = VFSTOMSDOSFS(mp);
#endif
} else {
vput(devvp);
if (devvp != pmp->pm_devvp)
return (EINVAL); /* XXX needs translation */
}
if (error) {
vrele(devvp);
return (error);
}
error = update_mp(mp, td);
if (error) {
if ((mp->mnt_flag & MNT_UPDATE) == 0)
msdosfs_unmount(mp, MNT_FORCE);
return error;
}
if (devvp->v_type == VCHR && devvp->v_rdev != NULL)
devvp->v_rdev->si_mountpt = mp;
vfs_mountedfrom(mp, from);
#ifdef MSDOSFS_DEBUG
printf("msdosfs_mount(): mp %p, pmp %p, inusemap %p\n", mp, pmp, pmp->pm_inusemap);
#endif
return (0);
}
int
udp6_output(struct in6pcb *in6p, struct mbuf *m, struct sockaddr *addr6,
struct mbuf *control, struct thread *td)
{
u_int32_t ulen = m->m_pkthdr.len;
u_int32_t plen = sizeof(struct udphdr) + ulen;
struct ip6_hdr *ip6;
struct udphdr *udp6;
struct in6_addr *laddr, *faddr;
u_short fport;
int error = 0;
struct ip6_pktopts opt, *stickyopt = in6p->in6p_outputopts;
int priv;
int hlen = sizeof(struct ip6_hdr);
struct sockaddr_in6 tmp;
priv = !priv_check(td, PRIV_ROOT); /* 1 if privileged, 0 if not */
if (control) {
if ((error = ip6_setpktoptions(control, &opt,
in6p->in6p_outputopts,
IPPROTO_UDP, priv)) != 0)
goto release;
in6p->in6p_outputopts = &opt;
}
if (addr6) {
/*
* IPv4 version of udp_output calls in_pcbconnect in this case,
* which needs splnet and affects performance.
* Since we saw no essential reason for calling in_pcbconnect,
* we get rid of such kind of logic, and call in6_selectsrc
* and in6_pcbsetlport in order to fill in the local address
* and the local port.
*/
struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)addr6;
/* Caller should have rejected the v4-mapped address */
KASSERT(!IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr),
("v4-mapped address"));
if (sin6->sin6_port == 0) {
error = EADDRNOTAVAIL;
goto release;
}
if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
/* how about ::ffff:0.0.0.0 case? */
error = EISCONN;
goto release;
}
if (!prison_remote_ip(td, addr6)) {
error = EAFNOSUPPORT; /* IPv4 only jail */
goto release;
}
/* protect *sin6 from overwrites */
tmp = *sin6;
sin6 = &tmp;
faddr = &sin6->sin6_addr;
fport = sin6->sin6_port; /* allow 0 port */
/* KAME hack: embed scopeid */
if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, NULL) != 0) {
error = EINVAL;
goto release;
}
laddr = in6_selectsrc(sin6, in6p->in6p_outputopts,
in6p->in6p_moptions,
&in6p->in6p_route,
&in6p->in6p_laddr, &error, NULL);
if (laddr == NULL) {
if (error == 0)
error = EADDRNOTAVAIL;
goto release;
}
if (in6p->in6p_lport == 0 &&
(error = in6_pcbsetlport(laddr, in6p, td)) != 0)
goto release;
} else {
if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
error = ENOTCONN;
goto release;
}
/* Connection to v4-mapped address should have been rejected */
KASSERT(!IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr),
("bound to v4-mapped address"));
laddr = &in6p->in6p_laddr;
faddr = &in6p->in6p_faddr;
fport = in6p->in6p_fport;
}
/*
* Calculate data length and get a mbuf
* for UDP and IP6 headers.
*/
M_PREPEND(m, hlen + sizeof(struct udphdr), M_NOWAIT);
if (m == NULL) {
error = ENOBUFS;
goto release;
}
/*
* Stuff checksum and output datagram.
*/
udp6 = (struct udphdr *)(mtod(m, caddr_t) + hlen);
udp6->uh_sport = in6p->in6p_lport; /* lport is always set in the PCB */
udp6->uh_dport = fport;
if (plen <= 0xffff)
udp6->uh_ulen = htons((u_short)plen);
else
udp6->uh_ulen = 0;
udp6->uh_sum = 0;
ip6 = mtod(m, struct ip6_hdr *);
ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
ip6->ip6_vfc &= ~IPV6_VERSION_MASK;
ip6->ip6_vfc |= IPV6_VERSION;
#if 0 /* ip6_plen will be filled in ip6_output. */
ip6->ip6_plen = htons((u_short)plen);
#endif
ip6->ip6_nxt = IPPROTO_UDP;
ip6->ip6_hlim = in6_selecthlim(in6p,
in6p->in6p_route.ro_rt ?
in6p->in6p_route.ro_rt->rt_ifp : NULL);
ip6->ip6_src = *laddr;
ip6->ip6_dst = *faddr;
if ((udp6->uh_sum = in6_cksum(m, IPPROTO_UDP,
sizeof(struct ip6_hdr), plen)) == 0) {
udp6->uh_sum = 0xffff;
}
udp6stat.udp6s_opackets++;
error = ip6_output(m, in6p->in6p_outputopts, &in6p->in6p_route, 0,
in6p->in6p_moptions, NULL, in6p);
goto releaseopt;
release:
m_freem(m);
releaseopt:
if (control) {
ip6_clearpktopts(in6p->in6p_outputopts, -1);
in6p->in6p_outputopts = stickyopt;
m_freem(control);
}
return (error);
}
int
thread_create(struct thread *td, struct rtprio *rtp,
int (*initialize_thread)(struct thread *, void *), void *thunk)
{
struct thread *newtd;
struct proc *p;
int error;
p = td->td_proc;
if (rtp != NULL) {
switch(rtp->type) {
case RTP_PRIO_REALTIME:
case RTP_PRIO_FIFO:
/* Only root can set scheduler policy */
if (priv_check(td, PRIV_SCHED_SETPOLICY) != 0)
return (EPERM);
if (rtp->prio > RTP_PRIO_MAX)
return (EINVAL);
break;
case RTP_PRIO_NORMAL:
rtp->prio = 0;
break;
default:
return (EINVAL);
}
}
#ifdef RACCT
if (racct_enable) {
PROC_LOCK(p);
error = racct_add(p, RACCT_NTHR, 1);
PROC_UNLOCK(p);
if (error != 0)
return (EPROCLIM);
}
#endif
/* Initialize our td */
error = kern_thr_alloc(p, 0, &newtd);
if (error)
goto fail;
cpu_copy_thread(newtd, td);
bzero(&newtd->td_startzero,
__rangeof(struct thread, td_startzero, td_endzero));
bcopy(&td->td_startcopy, &newtd->td_startcopy,
__rangeof(struct thread, td_startcopy, td_endcopy));
newtd->td_proc = td->td_proc;
newtd->td_rb_list = newtd->td_rbp_list = newtd->td_rb_inact = 0;
thread_cow_get(newtd, td);
error = initialize_thread(newtd, thunk);
if (error != 0) {
thread_cow_free(newtd);
thread_free(newtd);
goto fail;
}
PROC_LOCK(p);
p->p_flag |= P_HADTHREADS;
thread_link(newtd, p);
bcopy(p->p_comm, newtd->td_name, sizeof(newtd->td_name));
thread_lock(td);
/* let the scheduler know about these things. */
sched_fork_thread(td, newtd);
thread_unlock(td);
if (P_SHOULDSTOP(p))
newtd->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK;
if (p->p_ptevents & PTRACE_LWP)
newtd->td_dbgflags |= TDB_BORN;
PROC_UNLOCK(p);
#ifdef HWPMC_HOOKS
if (PMC_PROC_IS_USING_PMCS(p))
PMC_CALL_HOOK(newtd, PMC_FN_THR_CREATE, NULL);
else if (PMC_SYSTEM_SAMPLING_ACTIVE())
PMC_CALL_HOOK_UNLOCKED(newtd, PMC_FN_THR_CREATE_LOG, NULL);
#endif
tidhash_add(newtd);
thread_lock(newtd);
if (rtp != NULL) {
if (!(td->td_pri_class == PRI_TIMESHARE &&
rtp->type == RTP_PRIO_NORMAL)) {
rtp_to_pri(rtp, newtd);
sched_prio(newtd, newtd->td_user_pri);
} /* ignore timesharing class */
}
TD_SET_CAN_RUN(newtd);
sched_add(newtd, SRQ_BORING);
thread_unlock(newtd);
return (0);
fail:
#ifdef RACCT
if (racct_enable) {
PROC_LOCK(p);
racct_sub(p, RACCT_NTHR, 1);
PROC_UNLOCK(p);
}
#endif
return (error);
}
/*
* Generic internet control operations (ioctl's).
*/
int
ipx_control_oncpu(struct socket *so, u_long cmd, caddr_t data,
struct ifnet *ifp, struct thread *td)
{
struct ifreq *ifr = (struct ifreq *)data;
struct ipx_aliasreq *ifra = (struct ipx_aliasreq *)data;
struct ipx_ifaddr *ia;
struct ifaddr *ifa;
struct ipx_ifaddr *oia;
int dstIsNew, hostIsNew;
int error = 0;
/*
* Find address for this interface, if it exists.
*/
if (ifp == NULL)
return (EADDRNOTAVAIL);
for (ia = ipx_ifaddr; ia != NULL; ia = ia->ia_next)
if (ia->ia_ifp == ifp)
break;
switch (cmd) {
case SIOCGIFADDR:
if (ia == NULL)
return (EADDRNOTAVAIL);
*(struct sockaddr_ipx *)&ifr->ifr_addr = ia->ia_addr;
return (0);
case SIOCGIFBRDADDR:
if (ia == NULL)
return (EADDRNOTAVAIL);
if ((ifp->if_flags & IFF_BROADCAST) == 0)
return (EINVAL);
*(struct sockaddr_ipx *)&ifr->ifr_dstaddr = ia->ia_broadaddr;
return (0);
case SIOCGIFDSTADDR:
if (ia == NULL)
return (EADDRNOTAVAIL);
if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
return (EINVAL);
*(struct sockaddr_ipx *)&ifr->ifr_dstaddr = ia->ia_dstaddr;
return (0);
}
if ((error = priv_check(td, PRIV_ROOT)) != 0)
return (error);
switch (cmd) {
case SIOCAIFADDR:
case SIOCDIFADDR:
if (ifra->ifra_addr.sipx_family == AF_IPX)
for (oia = ia; ia != NULL; ia = ia->ia_next) {
if (ia->ia_ifp == ifp &&
ipx_neteq(ia->ia_addr.sipx_addr,
ifra->ifra_addr.sipx_addr))
break;
}
if (cmd == SIOCDIFADDR && ia == NULL)
return (EADDRNOTAVAIL);
/* FALLTHROUGH */
case SIOCSIFADDR:
case SIOCSIFDSTADDR:
if (ia == NULL) {
oia = ifa_create(sizeof(*ia), M_WAITOK);
if ((ia = ipx_ifaddr) != NULL) {
for ( ; ia->ia_next != NULL; ia = ia->ia_next)
;
ia->ia_next = oia;
} else
ipx_ifaddr = oia;
ia = oia;
ifa = (struct ifaddr *)ia;
ifa_iflink(ifa, ifp, 1);
ia->ia_ifp = ifp;
ifa->ifa_addr = (struct sockaddr *)&ia->ia_addr;
ifa->ifa_netmask = (struct sockaddr *)&ipx_netmask;
ifa->ifa_dstaddr = (struct sockaddr *)&ia->ia_dstaddr;
if (ifp->if_flags & IFF_BROADCAST) {
ia->ia_broadaddr.sipx_family = AF_IPX;
ia->ia_broadaddr.sipx_len = sizeof(ia->ia_addr);
ia->ia_broadaddr.sipx_addr.x_host = ipx_broadhost;
}
}
}
switch (cmd) {
case SIOCSIFDSTADDR:
if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
return (EINVAL);
if (ia->ia_flags & IFA_ROUTE) {
rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST);
ia->ia_flags &= ~IFA_ROUTE;
}
if (ifp->if_ioctl) {
ifnet_serialize_all(ifp);
error = ifp->if_ioctl(ifp, SIOCSIFDSTADDR,
(void *)ia, td->td_proc->p_ucred);
ifnet_deserialize_all(ifp);
if (error)
return (error);
}
*(struct sockaddr *)&ia->ia_dstaddr = ifr->ifr_dstaddr;
return (0);
case SIOCSIFADDR:
return (ipx_ifinit(ifp, ia,
(struct sockaddr_ipx *)&ifr->ifr_addr, 1));
case SIOCDIFADDR:
ipx_ifscrub(ifp, ia);
ifa = (struct ifaddr *)ia;
ifa_ifunlink(ifa, ifp);
oia = ia;
if (oia == (ia = ipx_ifaddr)) {
ipx_ifaddr = ia->ia_next;
} else {
while (ia->ia_next && (ia->ia_next != oia)) {
ia = ia->ia_next;
}
if (ia->ia_next)
ia->ia_next = oia->ia_next;
else
kprintf("Didn't unlink ipxifadr from list\n");
}
ifa_destroy(&oia->ia_ifa);
return (0);
case SIOCAIFADDR:
dstIsNew = 0;
hostIsNew = 1;
if (ia->ia_addr.sipx_family == AF_IPX) {
if (ifra->ifra_addr.sipx_len == 0) {
ifra->ifra_addr = ia->ia_addr;
hostIsNew = 0;
} else if (ipx_neteq(ifra->ifra_addr.sipx_addr,
ia->ia_addr.sipx_addr))
hostIsNew = 0;
}
if ((ifp->if_flags & IFF_POINTOPOINT) &&
(ifra->ifra_dstaddr.sipx_family == AF_IPX)) {
if (hostIsNew == 0)
ipx_ifscrub(ifp, ia);
ia->ia_dstaddr = ifra->ifra_dstaddr;
dstIsNew = 1;
}
if (ifra->ifra_addr.sipx_family == AF_IPX &&
(hostIsNew || dstIsNew))
error = ipx_ifinit(ifp, ia, &ifra->ifra_addr, 0);
return (error);
default:
if (ifp->if_ioctl == NULL)
return (EOPNOTSUPP);
ifnet_serialize_all(ifp);
error = ifp->if_ioctl(ifp, cmd, data, td->td_proc->p_ucred);
ifnet_deserialize_all(ifp);
return (error);
}
}
static int
acd_geom_ioctl(struct g_provider *pp, u_long cmd, void *addr, int fflag, struct thread *td)
{
device_t dev = pp->geom->softc;
struct ata_device *atadev = device_get_softc(dev);
struct acd_softc *cdp = device_get_ivars(dev);
int error = 0, nocopyout = 0;
if (!cdp)
return ENXIO;
if (atadev->flags & ATA_D_MEDIA_CHANGED) {
switch (cmd) {
case CDIOCRESET:
acd_test_ready(dev);
break;
default:
acd_read_toc(dev);
acd_prevent_allow(dev, 1);
cdp->flags |= F_LOCKED;
break;
}
}
switch (cmd) {
case CDIOCRESUME:
error = acd_pause_resume(dev, 1);
break;
case CDIOCPAUSE:
error = acd_pause_resume(dev, 0);
break;
case CDIOCSTART:
error = acd_start_stop(dev, 1);
break;
case CDIOCSTOP:
error = acd_start_stop(dev, 0);
break;
case CDIOCALLOW:
error = acd_prevent_allow(dev, 0);
cdp->flags &= ~F_LOCKED;
break;
case CDIOCPREVENT:
error = acd_prevent_allow(dev, 1);
cdp->flags |= F_LOCKED;
break;
/*
* XXXRW: Why does this require privilege?
*/
case CDIOCRESET:
error = priv_check(td, PRIV_DRIVER);
if (error)
break;
error = acd_test_ready(dev);
break;
case CDIOCEJECT:
if (pp->acr != 1) {
error = EBUSY;
break;
}
error = acd_tray(dev, 0);
break;
case CDIOCCLOSE:
if (pp->acr != 1)
break;
error = acd_tray(dev, 1);
break;
case CDIOREADTOCHEADER:
if (!cdp->toc.hdr.ending_track) {
error = EIO;
break;
}
bcopy(&cdp->toc.hdr, addr, sizeof(cdp->toc.hdr));
break;
case CDIOREADTOCENTRYS:
{
struct ioc_read_toc_entry *te = (struct ioc_read_toc_entry *)addr;
struct toc *toc = &cdp->toc;
int starting_track = te->starting_track;
int len;
if (!toc->hdr.ending_track) {
error = EIO;
break;
}
if (te->data_len < sizeof(toc->tab[0]) ||
(te->data_len % sizeof(toc->tab[0])) != 0 ||
(te->address_format != CD_MSF_FORMAT &&
te->address_format != CD_LBA_FORMAT)) {
error = EINVAL;
break;
}
if (!starting_track)
starting_track = toc->hdr.starting_track;
else if (starting_track == 170)
starting_track = toc->hdr.ending_track + 1;
else if (starting_track < toc->hdr.starting_track ||
starting_track > toc->hdr.ending_track + 1) {
error = EINVAL;
break;
}
len = ((toc->hdr.ending_track + 1 - starting_track) + 1) *
sizeof(toc->tab[0]);
if (te->data_len < len)
len = te->data_len;
if (len > sizeof(toc->tab)) {
error = EINVAL;
break;
}
if (te->address_format == CD_MSF_FORMAT) {
struct cd_toc_entry *entry;
if (!(toc = malloc(sizeof(struct toc), M_ACD, M_NOWAIT))) {
error = ENOMEM;
break;
}
bcopy(&cdp->toc, toc, sizeof(struct toc));
entry = toc->tab + (toc->hdr.ending_track + 1 -
toc->hdr.starting_track) + 1;
while (--entry >= toc->tab) {
lba2msf(ntohl(entry->addr.lba), &entry->addr.msf.minute,
&entry->addr.msf.second, &entry->addr.msf.frame);
entry->addr_type = CD_MSF_FORMAT;
}
}
error = copyout(toc->tab + starting_track - toc->hdr.starting_track,
te->data, len);
if (te->address_format == CD_MSF_FORMAT)
free(toc, M_ACD);
}
break;
case CDIOREADTOCENTRY:
{
struct ioc_read_toc_single_entry *te =
(struct ioc_read_toc_single_entry *)addr;
struct toc *toc = &cdp->toc;
u_char track = te->track;
if (!toc->hdr.ending_track) {
error = EIO;
break;
}
if (te->address_format != CD_MSF_FORMAT &&
te->address_format != CD_LBA_FORMAT) {
error = EINVAL;
break;
}
if (!track)
track = toc->hdr.starting_track;
else if (track == 170)
track = toc->hdr.ending_track + 1;
else if (track < toc->hdr.starting_track ||
track > toc->hdr.ending_track + 1) {
error = EINVAL;
break;
}
if (te->address_format == CD_MSF_FORMAT) {
struct cd_toc_entry *entry;
if (!(toc = malloc(sizeof(struct toc), M_ACD, M_NOWAIT))) {
error = ENOMEM;
break;
}
bcopy(&cdp->toc, toc, sizeof(struct toc));
entry = toc->tab + (track - toc->hdr.starting_track);
lba2msf(ntohl(entry->addr.lba), &entry->addr.msf.minute,
&entry->addr.msf.second, &entry->addr.msf.frame);
}
bcopy(toc->tab + track - toc->hdr.starting_track,
&te->entry, sizeof(struct cd_toc_entry));
if (te->address_format == CD_MSF_FORMAT)
free(toc, M_ACD);
}
break;
#if __FreeBSD_version > 600008
case CDIOCREADSUBCHANNEL_SYSSPACE:
nocopyout = 1;
/* FALLTHROUGH */
#endif
case CDIOCREADSUBCHANNEL:
{
struct ioc_read_subchannel *args =
(struct ioc_read_subchannel *)addr;
u_int8_t format;
int8_t ccb[16] = { ATAPI_READ_SUBCHANNEL, 0, 0x40, 1, 0, 0, 0,
sizeof(cdp->subchan)>>8, sizeof(cdp->subchan),
0, 0, 0, 0, 0, 0, 0 };
if (args->data_len > sizeof(struct cd_sub_channel_info) ||
args->data_len < sizeof(struct cd_sub_channel_header)) {
error = EINVAL;
break;
}
format = args->data_format;
if ((format != CD_CURRENT_POSITION) &&
(format != CD_MEDIA_CATALOG) && (format != CD_TRACK_INFO)) {
error = EINVAL;
break;
}
ccb[1] = args->address_format & CD_MSF_FORMAT;
if ((error = ata_atapicmd(dev, ccb, (caddr_t)&cdp->subchan,
sizeof(cdp->subchan), ATA_R_READ, 10)))
break;
if ((format == CD_MEDIA_CATALOG) || (format == CD_TRACK_INFO)) {
if (cdp->subchan.header.audio_status == 0x11) {
error = EINVAL;
break;
}
ccb[3] = format;
if (format == CD_TRACK_INFO)
ccb[6] = args->track;
if ((error = ata_atapicmd(dev, ccb, (caddr_t)&cdp->subchan,
sizeof(cdp->subchan),ATA_R_READ,10))){
break;
}
}
if (nocopyout == 0) {
error = copyout(&cdp->subchan, args->data, args->data_len);
} else {
error = 0;
bcopy(&cdp->subchan, args->data, args->data_len);
}
}
break;
case CDIOCPLAYMSF:
{
struct ioc_play_msf *args = (struct ioc_play_msf *)addr;
error =
acd_play(dev,
msf2lba(args->start_m, args->start_s, args->start_f),
msf2lba(args->end_m, args->end_s, args->end_f));
}
break;
case CDIOCPLAYBLOCKS:
{
struct ioc_play_blocks *args = (struct ioc_play_blocks *)addr;
error = acd_play(dev, args->blk, args->blk + args->len);
}
break;
case CDIOCPLAYTRACKS:
{
struct ioc_play_track *args = (struct ioc_play_track *)addr;
int t1, t2;
if (!cdp->toc.hdr.ending_track) {
error = EIO;
break;
}
if (args->end_track < cdp->toc.hdr.ending_track + 1)
++args->end_track;
if (args->end_track > cdp->toc.hdr.ending_track + 1)
args->end_track = cdp->toc.hdr.ending_track + 1;
t1 = args->start_track - cdp->toc.hdr.starting_track;
t2 = args->end_track - cdp->toc.hdr.starting_track;
if (t1 < 0 || t2 < 0 ||
t1 > (cdp->toc.hdr.ending_track-cdp->toc.hdr.starting_track)) {
error = EINVAL;
break;
}
error = acd_play(dev, ntohl(cdp->toc.tab[t1].addr.lba),
ntohl(cdp->toc.tab[t2].addr.lba));
}
break;
case CDIOCGETVOL:
{
struct ioc_vol *arg = (struct ioc_vol *)addr;
if ((error = acd_mode_sense(dev, ATAPI_CDROM_AUDIO_PAGE,
(caddr_t)&cdp->au, sizeof(cdp->au))))
break;
if (cdp->au.page_code != ATAPI_CDROM_AUDIO_PAGE) {
error = EIO;
break;
}
arg->vol[0] = cdp->au.port[0].volume;
arg->vol[1] = cdp->au.port[1].volume;
arg->vol[2] = cdp->au.port[2].volume;
arg->vol[3] = cdp->au.port[3].volume;
}
break;
case CDIOCSETVOL:
{
struct ioc_vol *arg = (struct ioc_vol *)addr;
if ((error = acd_mode_sense(dev, ATAPI_CDROM_AUDIO_PAGE,
(caddr_t)&cdp->au, sizeof(cdp->au))))
break;
if (cdp->au.page_code != ATAPI_CDROM_AUDIO_PAGE) {
error = EIO;
break;
}
if ((error = acd_mode_sense(dev, ATAPI_CDROM_AUDIO_PAGE_MASK,
(caddr_t)&cdp->aumask,
sizeof(cdp->aumask))))
break;
cdp->au.data_length = 0;
cdp->au.port[0].channels = CHANNEL_0;
cdp->au.port[1].channels = CHANNEL_1;
cdp->au.port[0].volume = arg->vol[0] & cdp->aumask.port[0].volume;
cdp->au.port[1].volume = arg->vol[1] & cdp->aumask.port[1].volume;
cdp->au.port[2].volume = arg->vol[2] & cdp->aumask.port[2].volume;
cdp->au.port[3].volume = arg->vol[3] & cdp->aumask.port[3].volume;
error = acd_mode_select(dev, (caddr_t)&cdp->au, sizeof(cdp->au));
}
break;
case CDIOCSETPATCH:
{
struct ioc_patch *arg = (struct ioc_patch *)addr;
error = acd_setchan(dev, arg->patch[0], arg->patch[1],
arg->patch[2], arg->patch[3]);
}
break;
case CDIOCSETMONO:
error = acd_setchan(dev, CHANNEL_0|CHANNEL_1, CHANNEL_0|CHANNEL_1, 0,0);
break;
case CDIOCSETSTEREO:
error = acd_setchan(dev, CHANNEL_0, CHANNEL_1, 0, 0);
break;
case CDIOCSETMUTE:
error = acd_setchan(dev, 0, 0, 0, 0);
break;
case CDIOCSETLEFT:
error = acd_setchan(dev, CHANNEL_0, CHANNEL_0, 0, 0);
break;
case CDIOCSETRIGHT:
error = acd_setchan(dev, CHANNEL_1, CHANNEL_1, 0, 0);
break;
case CDRIOCBLANK:
error = acd_blank(dev, (*(int *)addr));
break;
case CDRIOCNEXTWRITEABLEADDR:
{
struct acd_track_info track_info;
if ((error = acd_read_track_info(dev, 0xff, &track_info)))
break;
if (!track_info.nwa_valid) {
error = EINVAL;
break;
}
*(int*)addr = track_info.next_writeable_addr;
}
break;
case CDRIOCINITWRITER:
error = acd_init_writer(dev, (*(int *)addr));
break;
case CDRIOCINITTRACK:
error = acd_init_track(dev, (struct cdr_track *)addr);
break;
case CDRIOCFLUSH:
error = acd_flush(dev);
break;
case CDRIOCFIXATE:
error = acd_fixate(dev, (*(int *)addr));
break;
case CDRIOCREADSPEED:
{
int speed = *(int *)addr;
/* Preserve old behavior: units in multiples of CDROM speed */
if (speed < 177)
speed *= 177;
error = acd_set_speed(dev, speed, CDR_MAX_SPEED);
}
break;
case CDRIOCWRITESPEED:
{
int speed = *(int *)addr;
if (speed < 177)
speed *= 177;
error = acd_set_speed(dev, CDR_MAX_SPEED, speed);
}
break;
case CDRIOCGETBLOCKSIZE:
*(int *)addr = cdp->block_size;
break;
case CDRIOCSETBLOCKSIZE:
cdp->block_size = *(int *)addr;
pp->sectorsize = cdp->block_size; /* hack for GEOM SOS */
acd_set_ioparm(dev);
break;
case CDRIOCGETPROGRESS:
error = acd_get_progress(dev, (int *)addr);
break;
case CDRIOCSENDCUE:
error = acd_send_cue(dev, (struct cdr_cuesheet *)addr);
break;
case CDRIOCREADFORMATCAPS:
error = acd_read_format_caps(dev, (struct cdr_format_capacities *)addr);
break;
case CDRIOCFORMAT:
error = acd_format(dev, (struct cdr_format_params *)addr);
break;
case DVDIOCREPORTKEY:
if (cdp->cap.media & MST_READ_DVDROM)
error = acd_report_key(dev, (struct dvd_authinfo *)addr);
else
error = EINVAL;
break;
case DVDIOCSENDKEY:
if (cdp->cap.media & MST_READ_DVDROM)
error = acd_send_key(dev, (struct dvd_authinfo *)addr);
else
error = EINVAL;
break;
case DVDIOCREADSTRUCTURE:
if (cdp->cap.media & MST_READ_DVDROM)
error = acd_read_structure(dev, (struct dvd_struct *)addr);
else
error = EINVAL;
break;
default:
error = ata_device_ioctl(dev, cmd, addr);
}
return error;
}
static int
acd_geom_access(struct g_provider *pp, int dr, int dw, int de)
{
device_t dev = pp->geom->softc;
struct acd_softc *cdp = device_get_ivars(dev);
struct ata_request *request;
int8_t ccb[16] = { ATAPI_TEST_UNIT_READY, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
int timeout = 60, track;
if (!(request = ata_alloc_request()))
return ENOMEM;
/* wait if drive is not finished loading the medium */
while (timeout--) {
request->dev = dev;
bcopy(ccb, request->u.atapi.ccb, 16);
request->flags = ATA_R_ATAPI;
request->timeout = ATA_REQUEST_TIMEOUT;
ata_queue_request(request);
if (!request->error &&
(request->u.atapi.sense.key == 2 ||
request->u.atapi.sense.key == 7) &&
request->u.atapi.sense.asc == 4 &&
request->u.atapi.sense.ascq == 1)
pause("acdld", hz / 2);
else
break;
}
ata_free_request(request);
if (pp->acr == 0) {
acd_prevent_allow(dev, 1);
cdp->flags |= F_LOCKED;
acd_read_toc(dev);
}
if (dr + pp->acr == 0) {
acd_prevent_allow(dev, 0);
cdp->flags &= ~F_LOCKED;
}
if ((track = pp->index)) {
pp->sectorsize = (cdp->toc.tab[track - 1].control & 4) ? 2048 : 2352;
pp->mediasize = ntohl(cdp->toc.tab[track].addr.lba) -
ntohl(cdp->toc.tab[track - 1].addr.lba);
}
else {
pp->sectorsize = cdp->block_size;
pp->mediasize = cdp->disk_size;
}
pp->mediasize *= pp->sectorsize;
return 0;
}
/*
* MPALMOSTSAFE
*/
int
sys_msgctl(struct msgctl_args *uap)
{
struct thread *td = curthread;
struct proc *p = td->td_proc;
int msqid = uap->msqid;
int cmd = uap->cmd;
struct msqid_ds *user_msqptr = uap->buf;
int rval, eval;
struct msqid_ds msqbuf;
struct msqid_ds *msqptr;
#ifdef MSG_DEBUG_OK
kprintf("call to msgctl(%d, %d, 0x%x)\n", msqid, cmd, user_msqptr);
#endif
if (!jail_sysvipc_allowed && td->td_ucred->cr_prison != NULL)
return (ENOSYS);
get_mplock();
msqid = IPCID_TO_IX(msqid);
if (msqid < 0 || msqid >= msginfo.msgmni) {
#ifdef MSG_DEBUG_OK
kprintf("msqid (%d) out of range (0<=msqid<%d)\n", msqid,
msginfo.msgmni);
#endif
eval = EINVAL;
goto done;
}
msqptr = &msqids[msqid];
if (msqptr->msg_qbytes == 0) {
#ifdef MSG_DEBUG_OK
kprintf("no such msqid\n");
#endif
eval = EINVAL;
goto done;
}
if (msqptr->msg_perm.seq != IPCID_TO_SEQ(uap->msqid)) {
#ifdef MSG_DEBUG_OK
kprintf("wrong sequence number\n");
#endif
eval = EINVAL;
goto done;
}
rval = 0;
switch (cmd) {
case IPC_RMID:
{
struct msg *msghdr;
if ((eval = ipcperm(p, &msqptr->msg_perm, IPC_M)) != 0)
break;
/* Free the message headers */
msghdr = msqptr->msg_first;
while (msghdr != NULL) {
struct msg *msghdr_tmp;
/* Free the segments of each message */
msqptr->msg_cbytes -= msghdr->msg_ts;
msqptr->msg_qnum--;
msghdr_tmp = msghdr;
msghdr = msghdr->msg_next;
msg_freehdr(msghdr_tmp);
}
if (msqptr->msg_cbytes != 0)
panic("msg_cbytes is screwed up");
if (msqptr->msg_qnum != 0)
panic("msg_qnum is screwed up");
msqptr->msg_qbytes = 0; /* Mark it as free */
wakeup((caddr_t)msqptr);
}
break;
case IPC_SET:
if ((eval = ipcperm(p, &msqptr->msg_perm, IPC_M)) != 0)
break;
if ((eval = copyin(user_msqptr, &msqbuf, sizeof(msqbuf))) != 0)
break;
if (msqbuf.msg_qbytes > msqptr->msg_qbytes) {
eval = priv_check(td, PRIV_ROOT);
if (eval)
break;
}
if (msqbuf.msg_qbytes > msginfo.msgmnb) {
#ifdef MSG_DEBUG_OK
kprintf("can't increase msg_qbytes beyond %d (truncating)\n",
msginfo.msgmnb);
#endif
msqbuf.msg_qbytes = msginfo.msgmnb; /* silently restrict qbytes to system limit */
}
if (msqbuf.msg_qbytes == 0) {
#ifdef MSG_DEBUG_OK
kprintf("can't reduce msg_qbytes to 0\n");
#endif
eval = EINVAL; /* non-standard errno! */
break;
}
msqptr->msg_perm.uid = msqbuf.msg_perm.uid; /* change the owner */
msqptr->msg_perm.gid = msqbuf.msg_perm.gid; /* change the owner */
msqptr->msg_perm.mode = (msqptr->msg_perm.mode & ~0777) |
(msqbuf.msg_perm.mode & 0777);
msqptr->msg_qbytes = msqbuf.msg_qbytes;
msqptr->msg_ctime = time_second;
break;
case IPC_STAT:
if ((eval = ipcperm(p, &msqptr->msg_perm, IPC_R))) {
#ifdef MSG_DEBUG_OK
kprintf("requester doesn't have read access\n");
#endif
eval = EINVAL;
break;
}
eval = copyout(msqptr, user_msqptr, sizeof(struct msqid_ds));
break;
default:
#ifdef MSG_DEBUG_OK
kprintf("invalid command %d\n", cmd);
#endif
eval = EINVAL;
break;
}
done:
rel_mplock();
if (eval == 0)
uap->sysmsg_result = rval;
return(eval);
}
int
thread_create(struct thread *td, struct rtprio *rtp,
int (*initialize_thread)(struct thread *, void *), void *thunk)
{
struct thread *newtd;
struct proc *p;
int error;
p = td->td_proc;
if (rtp != NULL) {
switch(rtp->type) {
case RTP_PRIO_REALTIME:
case RTP_PRIO_FIFO:
/* Only root can set scheduler policy */
if (priv_check(td, PRIV_SCHED_SETPOLICY) != 0)
return (EPERM);
if (rtp->prio > RTP_PRIO_MAX)
return (EINVAL);
break;
case RTP_PRIO_NORMAL:
rtp->prio = 0;
break;
default:
return (EINVAL);
}
}
#ifdef RACCT
if (racct_enable) {
PROC_LOCK(p);
error = racct_add(p, RACCT_NTHR, 1);
PROC_UNLOCK(p);
if (error != 0)
return (EPROCLIM);
}
#endif
/* Initialize our td */
error = kern_thr_alloc(p, 0, &newtd);
if (error)
goto fail;
cpu_set_upcall(newtd, td);
bzero(&newtd->td_startzero,
__rangeof(struct thread, td_startzero, td_endzero));
bcopy(&td->td_startcopy, &newtd->td_startcopy,
__rangeof(struct thread, td_startcopy, td_endcopy));
newtd->td_proc = td->td_proc;
thread_cow_get(newtd, td);
error = initialize_thread(newtd, thunk);
if (error != 0) {
thread_cow_free(newtd);
thread_free(newtd);
goto fail;
}
PROC_LOCK(p);
p->p_flag |= P_HADTHREADS;
thread_link(newtd, p);
bcopy(p->p_comm, newtd->td_name, sizeof(newtd->td_name));
newtd->td_pax = p->p_pax;
thread_lock(td);
/* let the scheduler know about these things. */
sched_fork_thread(td, newtd);
thread_unlock(td);
if (P_SHOULDSTOP(p))
newtd->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK;
if (p->p_flag2 & P2_LWP_EVENTS)
newtd->td_dbgflags |= TDB_BORN;
/*
* Copy the existing thread VM policy into the new thread.
*/
vm_domain_policy_localcopy(&newtd->td_vm_dom_policy,
&td->td_vm_dom_policy);
PROC_UNLOCK(p);
tidhash_add(newtd);
thread_lock(newtd);
if (rtp != NULL) {
if (!(td->td_pri_class == PRI_TIMESHARE &&
rtp->type == RTP_PRIO_NORMAL)) {
rtp_to_pri(rtp, newtd);
sched_prio(newtd, newtd->td_user_pri);
} /* ignore timesharing class */
}
TD_SET_CAN_RUN(newtd);
sched_add(newtd, SRQ_BORING);
thread_unlock(newtd);
return (0);
fail:
#ifdef RACCT
if (racct_enable) {
PROC_LOCK(p);
racct_sub(p, RACCT_NTHR, 1);
PROC_UNLOCK(p);
}
#endif
return (error);
}
static int
create_thread(struct thread *td, mcontext_t *ctx,
void (*start_func)(void *), void *arg,
char *stack_base, size_t stack_size,
char *tls_base,
long *child_tid, long *parent_tid,
int flags, struct rtprio *rtp)
{
stack_t stack;
struct thread *newtd;
struct proc *p;
int error;
p = td->td_proc;
/* Have race condition but it is cheap. */
if (p->p_numthreads >= max_threads_per_proc) {
++max_threads_hits;
return (EPROCLIM);
}
if (rtp != NULL) {
switch(rtp->type) {
case RTP_PRIO_REALTIME:
case RTP_PRIO_FIFO:
/* Only root can set scheduler policy */
if (priv_check(td, PRIV_SCHED_SETPOLICY) != 0)
return (EPERM);
if (rtp->prio > RTP_PRIO_MAX)
return (EINVAL);
break;
case RTP_PRIO_NORMAL:
rtp->prio = 0;
break;
default:
return (EINVAL);
}
}
#ifdef RACCT
PROC_LOCK(td->td_proc);
error = racct_add(p, RACCT_NTHR, 1);
PROC_UNLOCK(td->td_proc);
if (error != 0)
return (EPROCLIM);
#endif
/* Initialize our td */
newtd = thread_alloc(0);
if (newtd == NULL) {
error = ENOMEM;
goto fail;
}
/*
* Try the copyout as soon as we allocate the td so we don't
* have to tear things down in a failure case below.
* Here we copy out tid to two places, one for child and one
* for parent, because pthread can create a detached thread,
* if parent wants to safely access child tid, it has to provide
* its storage, because child thread may exit quickly and
* memory is freed before parent thread can access it.
*/
if ((child_tid != NULL &&
suword_lwpid(child_tid, newtd->td_tid)) ||
(parent_tid != NULL &&
suword_lwpid(parent_tid, newtd->td_tid))) {
thread_free(newtd);
error = EFAULT;
goto fail;
}
bzero(&newtd->td_startzero,
__rangeof(struct thread, td_startzero, td_endzero));
bcopy(&td->td_startcopy, &newtd->td_startcopy,
__rangeof(struct thread, td_startcopy, td_endcopy));
newtd->td_proc = td->td_proc;
newtd->td_ucred = crhold(td->td_ucred);
cpu_set_upcall(newtd, td);
if (ctx != NULL) { /* old way to set user context */
error = set_mcontext(newtd, ctx);
if (error != 0) {
thread_free(newtd);
crfree(td->td_ucred);
goto fail;
}
} else {
/* Set up our machine context. */
stack.ss_sp = stack_base;
stack.ss_size = stack_size;
/* Set upcall address to user thread entry function. */
cpu_set_upcall_kse(newtd, start_func, arg, &stack);
/* Setup user TLS address and TLS pointer register. */
error = cpu_set_user_tls(newtd, tls_base);
if (error != 0) {
thread_free(newtd);
crfree(td->td_ucred);
goto fail;
}
}
PROC_LOCK(td->td_proc);
td->td_proc->p_flag |= P_HADTHREADS;
newtd->td_sigmask = td->td_sigmask;
thread_link(newtd, p);
bcopy(p->p_comm, newtd->td_name, sizeof(newtd->td_name));
thread_lock(td);
/* let the scheduler know about these things. */
sched_fork_thread(td, newtd);
thread_unlock(td);
if (P_SHOULDSTOP(p))
newtd->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK;
PROC_UNLOCK(p);
tidhash_add(newtd);
thread_lock(newtd);
if (rtp != NULL) {
if (!(td->td_pri_class == PRI_TIMESHARE &&
rtp->type == RTP_PRIO_NORMAL)) {
rtp_to_pri(rtp, newtd);
sched_prio(newtd, newtd->td_user_pri);
} /* ignore timesharing class */
}
TD_SET_CAN_RUN(newtd);
sched_add(newtd, SRQ_BORING);
thread_unlock(newtd);
return (0);
fail:
#ifdef RACCT
PROC_LOCK(p);
racct_sub(p, RACCT_NTHR, 1);
PROC_UNLOCK(p);
#endif
return (error);
}
/*
* VFS Operations.
*
* mount system call
*/
static int
ext2_mount(struct mount *mp)
{
struct vfsoptlist *opts;
struct vnode *devvp;
struct thread *td;
struct ext2mount *ump = NULL;
struct m_ext2fs *fs;
struct nameidata nd, *ndp = &nd;
accmode_t accmode;
char *path, *fspec;
int error, flags, len;
td = curthread;
opts = mp->mnt_optnew;
if (vfs_filteropt(opts, ext2_opts))
return (EINVAL);
vfs_getopt(opts, "fspath", (void **)&path, NULL);
/* Double-check the length of path.. */
if (strlen(path) >= MAXMNTLEN - 1)
return (ENAMETOOLONG);
fspec = NULL;
error = vfs_getopt(opts, "from", (void **)&fspec, &len);
if (!error && fspec[len - 1] != '\0')
return (EINVAL);
/*
* If updating, check whether changing from read-only to
* read/write; if there is no device name, that's all we do.
*/
if (mp->mnt_flag & MNT_UPDATE) {
ump = VFSTOEXT2(mp);
fs = ump->um_e2fs;
error = 0;
if (fs->e2fs_ronly == 0 &&
vfs_flagopt(opts, "ro", NULL, 0)) {
error = VFS_SYNC(mp, MNT_WAIT);
if (error)
return (error);
flags = WRITECLOSE;
if (mp->mnt_flag & MNT_FORCE)
flags |= FORCECLOSE;
error = ext2_flushfiles(mp, flags, td);
if ( error == 0 && fs->e2fs_wasvalid && ext2_cgupdate(ump, MNT_WAIT) == 0) {
fs->e2fs->e2fs_state |= E2FS_ISCLEAN;
ext2_sbupdate(ump, MNT_WAIT);
}
fs->e2fs_ronly = 1;
vfs_flagopt(opts, "ro", &mp->mnt_flag, MNT_RDONLY);
DROP_GIANT();
g_topology_lock();
g_access(ump->um_cp, 0, -1, 0);
g_topology_unlock();
PICKUP_GIANT();
}
if (!error && (mp->mnt_flag & MNT_RELOAD))
error = ext2_reload(mp, td);
if (error)
return (error);
devvp = ump->um_devvp;
if (fs->e2fs_ronly && !vfs_flagopt(opts, "ro", NULL, 0)) {
if (ext2_check_sb_compat(fs->e2fs, devvp->v_rdev, 0))
return (EPERM);
/*
* If upgrade to read-write by non-root, then verify
* that user has necessary permissions on the device.
*/
vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
error = VOP_ACCESS(devvp, VREAD | VWRITE,
td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
VOP_UNLOCK(devvp, 0);
return (error);
}
VOP_UNLOCK(devvp, 0);
DROP_GIANT();
g_topology_lock();
error = g_access(ump->um_cp, 0, 1, 0);
g_topology_unlock();
PICKUP_GIANT();
if (error)
return (error);
if ((fs->e2fs->e2fs_state & E2FS_ISCLEAN) == 0 ||
(fs->e2fs->e2fs_state & E2FS_ERRORS)) {
if (mp->mnt_flag & MNT_FORCE) {
printf(
"WARNING: %s was not properly dismounted\n", fs->e2fs_fsmnt);
} else {
printf(
"WARNING: R/W mount of %s denied. Filesystem is not clean - run fsck\n",
fs->e2fs_fsmnt);
return (EPERM);
}
}
fs->e2fs->e2fs_state &= ~E2FS_ISCLEAN;
(void)ext2_cgupdate(ump, MNT_WAIT);
fs->e2fs_ronly = 0;
MNT_ILOCK(mp);
mp->mnt_flag &= ~MNT_RDONLY;
MNT_IUNLOCK(mp);
}
if (vfs_flagopt(opts, "export", NULL, 0)) {
/* Process export requests in vfs_mount.c. */
return (error);
}
}
/*
* Not an update, or updating the name: look up the name
* and verify that it refers to a sensible disk device.
*/
if (fspec == NULL)
return (EINVAL);
NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, fspec, td);
if ((error = namei(ndp)) != 0)
return (error);
NDFREE(ndp, NDF_ONLY_PNBUF);
devvp = ndp->ni_vp;
if (!vn_isdisk(devvp, &error)) {
vput(devvp);
return (error);
}
/*
* If mount by non-root, then verify that user has necessary
* permissions on the device.
*
* XXXRW: VOP_ACCESS() enough?
*/
accmode = VREAD;
if ((mp->mnt_flag & MNT_RDONLY) == 0)
accmode |= VWRITE;
error = VOP_ACCESS(devvp, accmode, td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
vput(devvp);
return (error);
}
if ((mp->mnt_flag & MNT_UPDATE) == 0) {
error = ext2_mountfs(devvp, mp);
} else {
if (devvp != ump->um_devvp) {
vput(devvp);
return (EINVAL); /* needs translation */
} else
vput(devvp);
}
if (error) {
vrele(devvp);
return (error);
}
ump = VFSTOEXT2(mp);
fs = ump->um_e2fs;
/*
* Note that this strncpy() is ok because of a check at the start
* of ext2_mount().
*/
strncpy(fs->e2fs_fsmnt, path, MAXMNTLEN);
fs->e2fs_fsmnt[MAXMNTLEN - 1] = '\0';
vfs_mountedfrom(mp, fspec);
return (0);
}
#include <machine/psl.h>
#include <machine/specialreg.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <machine/iodev.h>
/* ARGSUSED */
int
ioopen(struct cdev *dev __unused, int flags __unused, int fmt __unused,
struct thread *td)
{
int error;
error = priv_check(td, PRIV_IO);
if (error != 0)
return (error);
error = securelevel_gt(td->td_ucred, 0);
if (error != 0)
return (error);
td->td_frame->tf_eflags |= PSL_IOPL;
return (0);
}
/* ARGSUSED */
int
ioclose(struct cdev *dev __unused, int flags __unused, int fmt __unused,
struct thread *td)
static int
udf_mount(struct mount *mp)
{
struct vnode *devvp; /* vnode of the mount device */
struct thread *td;
struct udf_mnt *imp = NULL;
struct vfsoptlist *opts;
char *fspec, *cs_disk, *cs_local;
int error, len, *udf_flags;
struct nameidata nd, *ndp = &nd;
td = curthread;
opts = mp->mnt_optnew;
/*
* Unconditionally mount as read-only.
*/
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_RDONLY;
MNT_IUNLOCK(mp);
/*
* No root filesystem support. Probably not a big deal, since the
* bootloader doesn't understand UDF.
*/
if (mp->mnt_flag & MNT_ROOTFS)
return (ENOTSUP);
fspec = NULL;
error = vfs_getopt(opts, "from", (void **)&fspec, &len);
if (!error && fspec[len - 1] != '\0')
return (EINVAL);
if (mp->mnt_flag & MNT_UPDATE) {
return (0);
}
/* Check that the mount device exists */
if (fspec == NULL)
return (EINVAL);
NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, fspec, td);
if ((error = namei(ndp)))
return (error);
NDFREE(ndp, NDF_ONLY_PNBUF);
devvp = ndp->ni_vp;
if (vn_isdisk(devvp, &error) == 0) {
vput(devvp);
return (error);
}
/* Check the access rights on the mount device */
error = VOP_ACCESS(devvp, VREAD, td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
vput(devvp);
return (error);
}
if ((error = udf_mountfs(devvp, mp))) {
vrele(devvp);
return (error);
}
imp = VFSTOUDFFS(mp);
udf_flags = NULL;
error = vfs_getopt(opts, "flags", (void **)&udf_flags, &len);
if (error || len != sizeof(int))
return (EINVAL);
imp->im_flags = *udf_flags;
if (imp->im_flags & UDFMNT_KICONV && udf_iconv) {
cs_disk = NULL;
error = vfs_getopt(opts, "cs_disk", (void **)&cs_disk, &len);
if (!error && cs_disk[len - 1] != '\0')
return (EINVAL);
cs_local = NULL;
error = vfs_getopt(opts, "cs_local", (void **)&cs_local, &len);
if (!error && cs_local[len - 1] != '\0')
return (EINVAL);
udf_iconv->open(cs_local, cs_disk, &imp->im_d2l);
#if 0
udf_iconv->open(cs_disk, cs_local, &imp->im_l2d);
#endif
}
vfs_mountedfrom(mp, fspec);
return 0;
};
/*
* Process ioctls
*/
int
procfs_ioctl(PFS_IOCTL_ARGS)
{
struct procfs_status *ps;
#ifdef COMPAT_FREEBSD32
struct procfs_status32 *ps32;
#endif
int error, flags, sig;
#ifdef COMPAT_FREEBSD6
int ival;
#endif
KASSERT(p != NULL,
("%s() called without a process", __func__));
PROC_LOCK_ASSERT(p, MA_OWNED);
error = 0;
switch (cmd) {
#if defined(COMPAT_FREEBSD5) || defined(COMPAT_FREEBSD4) || defined(COMPAT_43)
case _IOC(IOC_IN, 'p', 1, 0):
#endif
#ifdef COMPAT_FREEBSD6
case _IO('p', 1):
ival = IOCPARM_IVAL(data);
data = &ival;
#endif
case PIOCBIS:
p->p_stops |= *(unsigned int *)data;
break;
#if defined(COMPAT_FREEBSD5) || defined(COMPAT_FREEBSD4) || defined(COMPAT_43)
case _IOC(IOC_IN, 'p', 2, 0):
#endif
#ifdef COMPAT_FREEBSD6
case _IO('p', 2):
ival = IOCPARM_IVAL(data);
data = &ival;
#endif
case PIOCBIC:
p->p_stops &= ~*(unsigned int *)data;
break;
#if defined(COMPAT_FREEBSD5) || defined(COMPAT_FREEBSD4) || defined(COMPAT_43)
case _IOC(IOC_IN, 'p', 3, 0):
#endif
#ifdef COMPAT_FREEBSD6
case _IO('p', 3):
ival = IOCPARM_IVAL(data);
data = &ival;
#endif
case PIOCSFL:
flags = *(unsigned int *)data;
if (flags & PF_ISUGID) {
/*
* XXXRW: Is this specific check required here, as
* p_candebug() should implement it, or other checks
* are missing.
*/
error = priv_check(td, PRIV_DEBUG_SUGID);
if (error)
break;
}
p->p_pfsflags = flags;
break;
case PIOCGFL:
*(unsigned int *)data = p->p_pfsflags;
break;
case PIOCWAIT:
while (p->p_step == 0 && (p->p_flag & P_WEXIT) == 0) {
/* sleep until p stops */
_PHOLD(p);
error = msleep(&p->p_stype, &p->p_mtx,
PWAIT|PCATCH, "pioctl", 0);
_PRELE(p);
if (error != 0)
break;
}
/* fall through to PIOCSTATUS */
case PIOCSTATUS:
ps = (struct procfs_status *)data;
ps->state = (p->p_step == 0);
ps->flags = 0; /* nope */
ps->events = p->p_stops;
ps->why = p->p_step ? p->p_stype : 0;
ps->val = p->p_step ? p->p_xstat : 0;
break;
#ifdef COMPAT_FREEBSD32
case PIOCWAIT32:
while (p->p_step == 0 && (p->p_flag & P_WEXIT) == 0) {
/* sleep until p stops */
_PHOLD(p);
error = msleep(&p->p_stype, &p->p_mtx,
PWAIT|PCATCH, "pioctl", 0);
_PRELE(p);
if (error != 0)
break;
}
/* fall through to PIOCSTATUS32 */
case PIOCSTATUS32:
ps32 = (struct procfs_status32 *)data;
ps32->state = (p->p_step == 0);
ps32->flags = 0; /* nope */
ps32->events = p->p_stops;
ps32->why = p->p_step ? p->p_stype : 0;
ps32->val = p->p_step ? p->p_xstat : 0;
break;
#endif
#if defined(COMPAT_FREEBSD5) || defined(COMPAT_FREEBSD4) || defined(COMPAT_43)
case _IOC(IOC_IN, 'p', 5, 0):
#endif
#ifdef COMPAT_FREEBSD6
case _IO('p', 5):
ival = IOCPARM_IVAL(data);
data = &ival;
#endif
case PIOCCONT:
if (p->p_step == 0)
break;
sig = *(unsigned int *)data;
if (sig != 0 && !_SIG_VALID(sig)) {
error = EINVAL;
break;
}
#if 0
p->p_step = 0;
if (P_SHOULDSTOP(p)) {
p->p_xstat = sig;
p->p_flag &= ~(P_STOPPED_TRACE|P_STOPPED_SIG);
PROC_SLOCK(p);
thread_unsuspend(p);
PROC_SUNLOCK(p);
} else if (sig)
kern_psignal(p, sig);
#else
if (sig)
kern_psignal(p, sig);
p->p_step = 0;
wakeup(&p->p_step);
#endif
break;
default:
error = (ENOTTY);
}
return (error);
}
/*
* audit_syscall_enter() is called on entry to each system call. It is
* responsible for deciding whether or not to audit the call (preselection),
* and if so, allocating a per-thread audit record. audit_new() will fill in
* basic thread/credential properties.
*/
void
audit_syscall_enter(unsigned short code, struct thread *td)
{
struct au_mask *aumask;
au_class_t class;
au_event_t event;
au_id_t auid;
KASSERT(td->td_ar == NULL, ("audit_syscall_enter: td->td_ar != NULL"));
KASSERT((td->td_pflags & TDP_AUDITREC) == 0,
("audit_syscall_enter: TDP_AUDITREC set"));
/*
* In FreeBSD, each ABI has its own system call table, and hence
* mapping of system call codes to audit events. Convert the code to
* an audit event identifier using the process system call table
* reference. In Darwin, there's only one, so we use the global
* symbol for the system call table. No audit record is generated
* for bad system calls, as no operation has been performed.
*/
if (code >= td->td_proc->p_sysent->sv_size)
return;
event = td->td_proc->p_sysent->sv_table[code].sy_auevent;
if (event == AUE_NULL)
return;
/*
* Check which audit mask to use; either the kernel non-attributable
* event mask or the process audit mask.
*/
auid = td->td_ucred->cr_audit.ai_auid;
if (auid == AU_DEFAUDITID)
aumask = &audit_nae_mask;
else
aumask = &td->td_ucred->cr_audit.ai_mask;
/*
* Allocate an audit record, if preselection allows it, and store in
* the thread for later use.
*/
class = au_event_class(event);
if (au_preselect(event, class, aumask, AU_PRS_BOTH)) {
/*
* If we're out of space and need to suspend unprivileged
* processes, do that here rather than trying to allocate
* another audit record.
*
* Note: we might wish to be able to continue here in the
* future, if the system recovers. That should be possible
* by means of checking the condition in a loop around
* cv_wait(). It might be desirable to reevaluate whether an
* audit record is still required for this event by
* re-calling au_preselect().
*/
if (audit_in_failure &&
priv_check(td, PRIV_AUDIT_FAILSTOP) != 0) {
cv_wait(&audit_fail_cv, &audit_mtx);
panic("audit_failing_stop: thread continued");
}
td->td_ar = audit_new(event, td);
if (td->td_ar != NULL)
td->td_pflags |= TDP_AUDITREC;
} else if (audit_pipe_preselect(auid, event, class, AU_PRS_BOTH, 0)) {
/*
* VFS call to manage extended attributes in UFS. If filename_vp is
* non-NULL, it must be passed in locked, and regardless of errors in
* processing, will be unlocked.
*/
int
ufs_extattrctl(struct mount *mp, int cmd, struct vnode *filename_vp,
int attrnamespace, const char *attrname)
{
struct ufsmount *ump = VFSTOUFS(mp);
struct thread *td = curthread;
int error;
/*
* Processes with privilege, but in jail, are not allowed to
* configure extended attributes.
*/
error = priv_check(td, PRIV_UFS_EXTATTRCTL);
if (error) {
if (filename_vp != NULL)
VOP_UNLOCK(filename_vp, 0);
return (error);
}
/*
* We only allow extattrctl(2) on UFS1 file systems, as UFS2 uses
* native extended attributes.
*/
if (ump->um_fstype != UFS1) {
if (filename_vp != NULL)
VOP_UNLOCK(filename_vp, 0);
return (EOPNOTSUPP);
}
switch(cmd) {
case UFS_EXTATTR_CMD_START:
if (filename_vp != NULL) {
VOP_UNLOCK(filename_vp, 0);
return (EINVAL);
}
if (attrname != NULL)
return (EINVAL);
error = ufs_extattr_start(mp, td);
return (error);
case UFS_EXTATTR_CMD_STOP:
if (filename_vp != NULL) {
VOP_UNLOCK(filename_vp, 0);
return (EINVAL);
}
if (attrname != NULL)
return (EINVAL);
error = ufs_extattr_stop(mp, td);
return (error);
case UFS_EXTATTR_CMD_ENABLE:
if (filename_vp == NULL)
return (EINVAL);
if (attrname == NULL) {
VOP_UNLOCK(filename_vp, 0);
return (EINVAL);
}
/*
* ufs_extattr_enable_with_open() will always unlock the
* vnode, regardless of failure.
*/
ufs_extattr_uepm_lock(ump);
error = ufs_extattr_enable_with_open(ump, filename_vp,
attrnamespace, attrname, td);
ufs_extattr_uepm_unlock(ump);
return (error);
case UFS_EXTATTR_CMD_DISABLE:
if (filename_vp != NULL) {
VOP_UNLOCK(filename_vp, 0);
return (EINVAL);
}
if (attrname == NULL)
return (EINVAL);
ufs_extattr_uepm_lock(ump);
error = ufs_extattr_disable(ump, attrnamespace, attrname,
td);
ufs_extattr_uepm_unlock(ump);
return (error);
default:
return (EINVAL);
}
}
static int
ugen_set_power_mode(struct usb_fifo *f, int mode)
{
struct usb_device *udev = f->udev;
int err;
uint8_t old_mode;
if ((udev == NULL) ||
(udev->parent_hub == NULL)) {
return (EINVAL);
}
err = priv_check(curthread, PRIV_DRIVER);
if (err)
return (err);
/* get old power mode */
old_mode = udev->power_mode;
/* if no change, then just return */
if (old_mode == mode)
return (0);
switch (mode) {
case USB_POWER_MODE_OFF:
/* get the device unconfigured */
err = ugen_set_config(f, USB_UNCONFIG_INDEX);
if (err) {
DPRINTFN(0, "Could not unconfigure "
"device (ignored)\n");
}
/* clear port enable */
err = usbd_req_clear_port_feature(udev->parent_hub,
NULL, udev->port_no, UHF_PORT_ENABLE);
break;
case USB_POWER_MODE_ON:
case USB_POWER_MODE_SAVE:
break;
case USB_POWER_MODE_RESUME:
#if USB_HAVE_POWERD
/* let USB-powerd handle resume */
USB_BUS_LOCK(udev->bus);
udev->pwr_save.write_refs++;
udev->pwr_save.last_xfer_time = ticks;
USB_BUS_UNLOCK(udev->bus);
/* set new power mode */
usbd_set_power_mode(udev, USB_POWER_MODE_SAVE);
/* wait for resume to complete */
usb_pause_mtx(NULL, hz / 4);
/* clear write reference */
USB_BUS_LOCK(udev->bus);
udev->pwr_save.write_refs--;
USB_BUS_UNLOCK(udev->bus);
#endif
mode = USB_POWER_MODE_SAVE;
break;
case USB_POWER_MODE_SUSPEND:
#if USB_HAVE_POWERD
/* let USB-powerd handle suspend */
USB_BUS_LOCK(udev->bus);
udev->pwr_save.last_xfer_time = ticks - (256 * hz);
USB_BUS_UNLOCK(udev->bus);
#endif
mode = USB_POWER_MODE_SAVE;
break;
default:
return (EINVAL);
}
if (err)
return (ENXIO); /* I/O failure */
/* if we are powered off we need to re-enumerate first */
if (old_mode == USB_POWER_MODE_OFF) {
if (udev->flags.usb_mode == USB_MODE_HOST) {
if (udev->re_enumerate_wait == 0)
udev->re_enumerate_wait = 1;
}
/* set power mode will wake up the explore thread */
}
/* set new power mode */
usbd_set_power_mode(udev, mode);
return (0); /* success */
}
/*
* Mount system call
*/
static int
reiserfs_mount(struct mount *mp)
{
size_t size;
int error, len;
accmode_t accmode;
char *path, *fspec;
struct vnode *devvp;
struct vfsoptlist *opts;
struct reiserfs_mount *rmp;
struct reiserfs_sb_info *sbi;
struct nameidata nd, *ndp = &nd;
struct thread *td;
td = curthread;
if (!(mp->mnt_flag & MNT_RDONLY))
return EROFS;
/* Get the new options passed to mount */
opts = mp->mnt_optnew;
/* `fspath' contains the mount point (eg. /mnt/linux); REQUIRED */
vfs_getopt(opts, "fspath", (void **)&path, NULL);
reiserfs_log(LOG_INFO, "mount point is `%s'\n", path);
/* `from' contains the device name (eg. /dev/ad0s1); REQUIRED */
fspec = NULL;
error = vfs_getopt(opts, "from", (void **)&fspec, &len);
if (!error && fspec[len - 1] != '\0')
return (EINVAL);
reiserfs_log(LOG_INFO, "device is `%s'\n", fspec);
/* Handle MNT_UPDATE (mp->mnt_flag) */
if (mp->mnt_flag & MNT_UPDATE) {
/* For now, only NFS export is supported. */
if (vfs_flagopt(opts, "export", NULL, 0))
return (0);
}
/* Not an update, or updating the name: look up the name
* and verify that it refers to a sensible disk device. */
if (fspec == NULL)
return (EINVAL);
NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, fspec, td);
if ((error = namei(ndp)) != 0)
return (error);
NDFREE(ndp, NDF_ONLY_PNBUF);
devvp = ndp->ni_vp;
if (!vn_isdisk(devvp, &error)) {
vput(devvp);
return (error);
}
/* If mount by non-root, then verify that user has necessary
* permissions on the device. */
accmode = VREAD;
if ((mp->mnt_flag & MNT_RDONLY) == 0)
accmode |= VWRITE;
error = VOP_ACCESS(devvp, accmode, td->td_ucred, td);
if (error)
error = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (error) {
vput(devvp);
return (error);
}
if ((mp->mnt_flag & MNT_UPDATE) == 0) {
error = reiserfs_mountfs(devvp, mp, td);
} else {
/* TODO Handle MNT_UPDATE */
vput(devvp);
return (EOPNOTSUPP);
}
if (error) {
vrele(devvp);
return (error);
}
rmp = VFSTOREISERFS(mp);
sbi = rmp->rm_reiserfs;
/*
* Note that this strncpy() is ok because of a check at the start
* of reiserfs_mount().
*/
reiserfs_log(LOG_DEBUG, "prepare statfs data\n");
(void)copystr(fspec, mp->mnt_stat.f_mntfromname, MNAMELEN - 1, &size);
bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size);
(void)reiserfs_statfs(mp, &mp->mnt_stat);
reiserfs_log(LOG_DEBUG, "done\n");
return (0);
}
/* ARGSUSED */
int
sys_audit(struct thread *td, struct audit_args *uap)
{
int error;
void * rec;
struct kaudit_record *ar;
if (jailed(td->td_ucred))
return (ENOSYS);
error = priv_check(td, PRIV_AUDIT_SUBMIT);
if (error)
return (error);
if ((uap->length <= 0) || (uap->length > audit_qctrl.aq_bufsz))
return (EINVAL);
ar = currecord();
/*
* If there's no current audit record (audit() itself not audited)
* commit the user audit record.
*/
if (ar == NULL) {
/*
* This is not very efficient; we're required to allocate a
* complete kernel audit record just so the user record can
* tag along.
*
* XXXAUDIT: Maybe AUE_AUDIT in the system call context and
* special pre-select handling?
*/
td->td_ar = audit_new(AUE_NULL, td);
if (td->td_ar == NULL)
return (ENOTSUP);
td->td_pflags |= TDP_AUDITREC;
ar = td->td_ar;
}
if (uap->length > MAX_AUDIT_RECORD_SIZE)
return (EINVAL);
rec = malloc(uap->length, M_AUDITDATA, M_WAITOK);
error = copyin(uap->record, rec, uap->length);
if (error)
goto free_out;
/* Verify the record. */
if (bsm_rec_verify(rec) == 0) {
error = EINVAL;
goto free_out;
}
#ifdef MAC
error = mac_system_check_audit(td->td_ucred, rec, uap->length);
if (error)
goto free_out;
#endif
/*
* Attach the user audit record to the kernel audit record. Because
* this system call is an auditable event, we will write the user
* record along with the record for this audit event.
*
* XXXAUDIT: KASSERT appropriate starting values of k_udata, k_ulen,
* k_ar_commit & AR_COMMIT_USER?
*/
ar->k_udata = rec;
ar->k_ulen = uap->length;
ar->k_ar_commit |= AR_COMMIT_USER;
/*
* Currently we assume that all preselection has been performed in
* userspace. We unconditionally set these masks so that the records
* get committed both to the trail and pipe. In the future we will
* want to setup kernel based preselection.
*/
ar->k_ar_commit |= (AR_PRESELECT_USER_TRAIL | AR_PRESELECT_USER_PIPE);
return (0);
free_out:
/*
* audit_syscall_exit() will free the audit record on the thread even
* if we allocated it above.
*/
free(rec, M_AUDITDATA);
return (error);
}
static int
ntfs_mount(struct mount *mp)
{
int err = 0, error;
struct vnode *devvp;
struct nameidata ndp;
struct thread *td;
char *from;
td = curthread;
if (vfs_filteropt(mp->mnt_optnew, ntfs_opts))
return (EINVAL);
/* Force mount as read-only. */
MNT_ILOCK(mp);
mp->mnt_flag |= MNT_RDONLY;
MNT_IUNLOCK(mp);
from = vfs_getopts(mp->mnt_optnew, "from", &error);
if (error)
return (error);
/*
* If updating, check whether changing from read-only to
* read/write.
*/
if (mp->mnt_flag & MNT_UPDATE) {
if (vfs_flagopt(mp->mnt_optnew, "export", NULL, 0)) {
/* Process export requests in vfs_mount.c */
return (0);
} else {
printf("ntfs_mount(): MNT_UPDATE not supported\n");
return (EINVAL);
}
}
/*
* Not an update, or updating the name: look up the name
* and verify that it refers to a sensible block device.
*/
NDINIT(&ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE, from, td);
err = namei(&ndp);
if (err)
return (err);
NDFREE(&ndp, NDF_ONLY_PNBUF);
devvp = ndp.ni_vp;
if (!vn_isdisk(devvp, &err)) {
vput(devvp);
return (err);
}
/*
* If mount by non-root, then verify that user has necessary
* permissions on the device.
*/
err = VOP_ACCESS(devvp, VREAD, td->td_ucred, td);
if (err)
err = priv_check(td, PRIV_VFS_MOUNT_PERM);
if (err) {
vput(devvp);
return (err);
}
/*
* Since this is a new mount, we want the names for the device and
* the mount point copied in. If an error occurs, the mountpoint is
* discarded by the upper level code. Note that vfs_mount() handles
* copying the mountpoint f_mntonname for us, so we don't have to do
* it here unless we want to set it to something other than "path"
* for some rason.
*/
err = ntfs_mountfs(devvp, mp, td);
if (err == 0) {
/* Save "mounted from" info for mount point. */
vfs_mountedfrom(mp, from);
} else
vrele(devvp);
return (err);
}
/* ARGSUSED */
int
auditon(struct thread *td, struct auditon_args *uap)
{
struct ucred *cred, *newcred, *oldcred;
int error;
union auditon_udata udata;
struct proc *tp;
if (jailed(td->td_ucred))
return (ENOSYS);
AUDIT_ARG_CMD(uap->cmd);
#ifdef MAC
error = mac_system_check_auditon(td->td_ucred, uap->cmd);
if (error)
return (error);
#endif
error = priv_check(td, PRIV_AUDIT_CONTROL);
if (error)
return (error);
if ((uap->length <= 0) || (uap->length > sizeof(union auditon_udata)))
return (EINVAL);
memset((void *)&udata, 0, sizeof(udata));
/*
* Some of the GET commands use the arguments too.
*/
switch (uap->cmd) {
case A_SETPOLICY:
case A_OLDSETPOLICY:
case A_SETKMASK:
case A_SETQCTRL:
case A_OLDSETQCTRL:
case A_SETSTAT:
case A_SETUMASK:
case A_SETSMASK:
case A_SETCOND:
case A_OLDSETCOND:
case A_SETCLASS:
case A_SETPMASK:
case A_SETFSIZE:
case A_SETKAUDIT:
case A_GETCLASS:
case A_GETPINFO:
case A_GETPINFO_ADDR:
case A_SENDTRIGGER:
error = copyin(uap->data, (void *)&udata, uap->length);
if (error)
return (error);
AUDIT_ARG_AUDITON(&udata);
break;
}
/*
* XXXAUDIT: Locking?
*/
switch (uap->cmd) {
case A_OLDGETPOLICY:
case A_GETPOLICY:
if (uap->length == sizeof(udata.au_policy64)) {
if (!audit_fail_stop)
udata.au_policy64 |= AUDIT_CNT;
if (audit_panic_on_write_fail)
udata.au_policy64 |= AUDIT_AHLT;
if (audit_argv)
udata.au_policy64 |= AUDIT_ARGV;
if (audit_arge)
udata.au_policy64 |= AUDIT_ARGE;
break;
}
if (uap->length != sizeof(udata.au_policy))
return (EINVAL);
if (!audit_fail_stop)
udata.au_policy |= AUDIT_CNT;
if (audit_panic_on_write_fail)
udata.au_policy |= AUDIT_AHLT;
if (audit_argv)
udata.au_policy |= AUDIT_ARGV;
if (audit_arge)
udata.au_policy |= AUDIT_ARGE;
break;
case A_OLDSETPOLICY:
case A_SETPOLICY:
if (uap->length == sizeof(udata.au_policy64)) {
if (udata.au_policy & (~AUDIT_CNT|AUDIT_AHLT|
AUDIT_ARGV|AUDIT_ARGE))
return (EINVAL);
audit_fail_stop = ((udata.au_policy64 & AUDIT_CNT) ==
0);
audit_panic_on_write_fail = (udata.au_policy64 &
AUDIT_AHLT);
audit_argv = (udata.au_policy64 & AUDIT_ARGV);
audit_arge = (udata.au_policy64 & AUDIT_ARGE);
break;
}
if (uap->length != sizeof(udata.au_policy))
return (EINVAL);
if (udata.au_policy & ~(AUDIT_CNT|AUDIT_AHLT|AUDIT_ARGV|
AUDIT_ARGE))
return (EINVAL);
/*
* XXX - Need to wake up waiters if the policy relaxes?
*/
audit_fail_stop = ((udata.au_policy & AUDIT_CNT) == 0);
audit_panic_on_write_fail = (udata.au_policy & AUDIT_AHLT);
audit_argv = (udata.au_policy & AUDIT_ARGV);
audit_arge = (udata.au_policy & AUDIT_ARGE);
break;
case A_GETKMASK:
if (uap->length != sizeof(udata.au_mask))
return (EINVAL);
udata.au_mask = audit_nae_mask;
break;
case A_SETKMASK:
if (uap->length != sizeof(udata.au_mask))
return (EINVAL);
audit_nae_mask = udata.au_mask;
break;
case A_OLDGETQCTRL:
case A_GETQCTRL:
if (uap->length == sizeof(udata.au_qctrl64)) {
udata.au_qctrl64.aq64_hiwater =
(u_int64_t)audit_qctrl.aq_hiwater;
udata.au_qctrl64.aq64_lowater =
(u_int64_t)audit_qctrl.aq_lowater;
udata.au_qctrl64.aq64_bufsz =
(u_int64_t)audit_qctrl.aq_bufsz;
udata.au_qctrl64.aq64_minfree =
(u_int64_t)audit_qctrl.aq_minfree;
break;
}
if (uap->length != sizeof(udata.au_qctrl))
return (EINVAL);
udata.au_qctrl = audit_qctrl;
break;
case A_OLDSETQCTRL:
case A_SETQCTRL:
if (uap->length == sizeof(udata.au_qctrl64)) {
if ((udata.au_qctrl64.aq64_hiwater > AQ_MAXHIGH) ||
(udata.au_qctrl64.aq64_lowater >=
udata.au_qctrl.aq_hiwater) ||
(udata.au_qctrl64.aq64_bufsz > AQ_MAXBUFSZ) ||
(udata.au_qctrl64.aq64_minfree < 0) ||
(udata.au_qctrl64.aq64_minfree > 100))
return (EINVAL);
audit_qctrl.aq_hiwater =
(int)udata.au_qctrl64.aq64_hiwater;
audit_qctrl.aq_lowater =
(int)udata.au_qctrl64.aq64_lowater;
audit_qctrl.aq_bufsz =
(int)udata.au_qctrl64.aq64_bufsz;
audit_qctrl.aq_minfree =
(int)udata.au_qctrl64.aq64_minfree;
audit_qctrl.aq_delay = -1; /* Not used. */
break;
}
if (uap->length != sizeof(udata.au_qctrl))
return (EINVAL);
if ((udata.au_qctrl.aq_hiwater > AQ_MAXHIGH) ||
(udata.au_qctrl.aq_lowater >= udata.au_qctrl.aq_hiwater) ||
(udata.au_qctrl.aq_bufsz > AQ_MAXBUFSZ) ||
(udata.au_qctrl.aq_minfree < 0) ||
(udata.au_qctrl.aq_minfree > 100))
return (EINVAL);
audit_qctrl = udata.au_qctrl;
/* XXX The queue delay value isn't used with the kernel. */
audit_qctrl.aq_delay = -1;
break;
case A_GETCWD:
return (ENOSYS);
break;
case A_GETCAR:
return (ENOSYS);
break;
case A_GETSTAT:
return (ENOSYS);
break;
case A_SETSTAT:
return (ENOSYS);
break;
case A_SETUMASK:
return (ENOSYS);
break;
case A_SETSMASK:
return (ENOSYS);
break;
case A_OLDGETCOND:
case A_GETCOND:
if (uap->length == sizeof(udata.au_cond64)) {
if (audit_enabled && !audit_suspended)
udata.au_cond64 = AUC_AUDITING;
else
udata.au_cond64 = AUC_NOAUDIT;
break;
}
if (uap->length != sizeof(udata.au_cond))
return (EINVAL);
if (audit_enabled && !audit_suspended)
udata.au_cond = AUC_AUDITING;
else
udata.au_cond = AUC_NOAUDIT;
break;
case A_OLDSETCOND:
case A_SETCOND:
if (uap->length == sizeof(udata.au_cond64)) {
if (udata.au_cond64 == AUC_NOAUDIT)
audit_suspended = 1;
if (udata.au_cond64 == AUC_AUDITING)
audit_suspended = 0;
if (udata.au_cond64 == AUC_DISABLED) {
audit_suspended = 1;
audit_shutdown(NULL, 0);
}
break;
}
if (uap->length != sizeof(udata.au_cond))
return (EINVAL);
if (udata.au_cond == AUC_NOAUDIT)
audit_suspended = 1;
if (udata.au_cond == AUC_AUDITING)
audit_suspended = 0;
if (udata.au_cond == AUC_DISABLED) {
audit_suspended = 1;
audit_shutdown(NULL, 0);
}
break;
case A_GETCLASS:
if (uap->length != sizeof(udata.au_evclass))
return (EINVAL);
udata.au_evclass.ec_class = au_event_class(
udata.au_evclass.ec_number);
break;
case A_SETCLASS:
if (uap->length != sizeof(udata.au_evclass))
return (EINVAL);
au_evclassmap_insert(udata.au_evclass.ec_number,
udata.au_evclass.ec_class);
break;
case A_GETPINFO:
if (uap->length != sizeof(udata.au_aupinfo))
return (EINVAL);
if (udata.au_aupinfo.ap_pid < 1)
return (ESRCH);
if ((tp = pfind(udata.au_aupinfo.ap_pid)) == NULL)
return (ESRCH);
if ((error = p_cansee(td, tp)) != 0) {
PROC_UNLOCK(tp);
return (error);
}
cred = tp->p_ucred;
if (cred->cr_audit.ai_termid.at_type == AU_IPv6) {
PROC_UNLOCK(tp);
return (EINVAL);
}
udata.au_aupinfo.ap_auid = cred->cr_audit.ai_auid;
udata.au_aupinfo.ap_mask.am_success =
cred->cr_audit.ai_mask.am_success;
udata.au_aupinfo.ap_mask.am_failure =
cred->cr_audit.ai_mask.am_failure;
udata.au_aupinfo.ap_termid.machine =
cred->cr_audit.ai_termid.at_addr[0];
udata.au_aupinfo.ap_termid.port =
(dev_t)cred->cr_audit.ai_termid.at_port;
udata.au_aupinfo.ap_asid = cred->cr_audit.ai_asid;
PROC_UNLOCK(tp);
break;
case A_SETPMASK:
if (uap->length != sizeof(udata.au_aupinfo))
return (EINVAL);
if (udata.au_aupinfo.ap_pid < 1)
return (ESRCH);
newcred = crget();
if ((tp = pfind(udata.au_aupinfo.ap_pid)) == NULL) {
crfree(newcred);
return (ESRCH);
}
if ((error = p_cansee(td, tp)) != 0) {
PROC_UNLOCK(tp);
crfree(newcred);
return (error);
}
oldcred = tp->p_ucred;
crcopy(newcred, oldcred);
newcred->cr_audit.ai_mask.am_success =
udata.au_aupinfo.ap_mask.am_success;
newcred->cr_audit.ai_mask.am_failure =
udata.au_aupinfo.ap_mask.am_failure;
td->td_proc->p_ucred = newcred;
PROC_UNLOCK(tp);
crfree(oldcred);
break;
case A_SETFSIZE:
if (uap->length != sizeof(udata.au_fstat))
return (EINVAL);
if ((udata.au_fstat.af_filesz != 0) &&
(udata.au_fstat.af_filesz < MIN_AUDIT_FILE_SIZE))
return (EINVAL);
audit_fstat.af_filesz = udata.au_fstat.af_filesz;
break;
case A_GETFSIZE:
if (uap->length != sizeof(udata.au_fstat))
return (EINVAL);
udata.au_fstat.af_filesz = audit_fstat.af_filesz;
udata.au_fstat.af_currsz = audit_fstat.af_currsz;
break;
case A_GETPINFO_ADDR:
if (uap->length != sizeof(udata.au_aupinfo_addr))
return (EINVAL);
if (udata.au_aupinfo_addr.ap_pid < 1)
return (ESRCH);
if ((tp = pfind(udata.au_aupinfo_addr.ap_pid)) == NULL)
return (ESRCH);
cred = tp->p_ucred;
udata.au_aupinfo_addr.ap_auid = cred->cr_audit.ai_auid;
udata.au_aupinfo_addr.ap_mask.am_success =
cred->cr_audit.ai_mask.am_success;
udata.au_aupinfo_addr.ap_mask.am_failure =
cred->cr_audit.ai_mask.am_failure;
udata.au_aupinfo_addr.ap_termid = cred->cr_audit.ai_termid;
udata.au_aupinfo_addr.ap_asid = cred->cr_audit.ai_asid;
PROC_UNLOCK(tp);
break;
case A_GETKAUDIT:
if (uap->length != sizeof(udata.au_kau_info))
return (EINVAL);
audit_get_kinfo(&udata.au_kau_info);
break;
case A_SETKAUDIT:
if (uap->length != sizeof(udata.au_kau_info))
return (EINVAL);
if (udata.au_kau_info.ai_termid.at_type != AU_IPv4 &&
udata.au_kau_info.ai_termid.at_type != AU_IPv6)
return (EINVAL);
audit_set_kinfo(&udata.au_kau_info);
break;
case A_SENDTRIGGER:
if (uap->length != sizeof(udata.au_trigger))
return (EINVAL);
if ((udata.au_trigger < AUDIT_TRIGGER_MIN) ||
(udata.au_trigger > AUDIT_TRIGGER_MAX))
return (EINVAL);
return (audit_send_trigger(udata.au_trigger));
default:
return (EINVAL);
}
/*
* Copy data back to userspace for the GET comands.
*/
switch (uap->cmd) {
case A_GETPOLICY:
case A_OLDGETPOLICY:
case A_GETKMASK:
case A_GETQCTRL:
case A_OLDGETQCTRL:
case A_GETCWD:
case A_GETCAR:
case A_GETSTAT:
case A_GETCOND:
case A_OLDGETCOND:
case A_GETCLASS:
case A_GETPINFO:
case A_GETFSIZE:
case A_GETPINFO_ADDR:
case A_GETKAUDIT:
error = copyout((void *)&udata, uap->data, uap->length);
if (error)
return (error);
break;
}
return (0);
}
int
udp6_output(struct in6pcb *in6p, struct mbuf *m, struct sockaddr *addr6,
struct mbuf *control, struct thread *td)
{
u_int32_t ulen = m->m_pkthdr.len;
u_int32_t plen = sizeof(struct udphdr) + ulen;
struct ip6_hdr *ip6;
struct udphdr *udp6;
struct in6_addr *laddr, *faddr;
u_short fport;
int error = 0;
struct ip6_pktopts opt, *stickyopt = in6p->in6p_outputopts;
int priv;
int af = AF_INET6, hlen = sizeof(struct ip6_hdr);
int flags;
struct sockaddr_in6 tmp;
priv = !priv_check(td, PRIV_ROOT); /* 1 if privileged, 0 if not */
if (control) {
if ((error = ip6_setpktoptions(control, &opt,
in6p->in6p_outputopts,
IPPROTO_UDP, priv)) != 0)
goto release;
in6p->in6p_outputopts = &opt;
}
if (addr6) {
/*
* IPv4 version of udp_output calls in_pcbconnect in this case,
* which needs splnet and affects performance.
* Since we saw no essential reason for calling in_pcbconnect,
* we get rid of such kind of logic, and call in6_selectsrc
* and in6_pcbsetport in order to fill in the local address
* and the local port.
*/
struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)addr6;
if (sin6->sin6_port == 0) {
error = EADDRNOTAVAIL;
goto release;
}
if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
/* how about ::ffff:0.0.0.0 case? */
error = EISCONN;
goto release;
}
if (!prison_remote_ip(td, addr6)) {
error = EAFNOSUPPORT; /* IPv4 only jail */
goto release;
}
/* protect *sin6 from overwrites */
tmp = *sin6;
sin6 = &tmp;
faddr = &sin6->sin6_addr;
fport = sin6->sin6_port; /* allow 0 port */
if (IN6_IS_ADDR_V4MAPPED(faddr)) {
if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY)) {
/*
* I believe we should explicitly discard the
* packet when mapped addresses are disabled,
* rather than send the packet as an IPv6 one.
* If we chose the latter approach, the packet
* might be sent out on the wire based on the
* default route, the situation which we'd
* probably want to avoid.
* (20010421 [email protected])
*/
error = EINVAL;
goto release;
} else
af = AF_INET;
}
/* KAME hack: embed scopeid */
if (in6_embedscope(&sin6->sin6_addr, sin6, in6p, NULL) != 0) {
error = EINVAL;
goto release;
}
if (!IN6_IS_ADDR_V4MAPPED(faddr)) {
laddr = in6_selectsrc(sin6, in6p->in6p_outputopts,
in6p->in6p_moptions,
&in6p->in6p_route,
&in6p->in6p_laddr, &error, NULL);
} else
laddr = &in6p->in6p_laddr; /* XXX */
if (laddr == NULL) {
if (error == 0)
error = EADDRNOTAVAIL;
goto release;
}
if (in6p->in6p_lport == 0 &&
(error = in6_pcbsetport(laddr, in6p, td)) != 0)
goto release;
} else {
if (IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) {
error = ENOTCONN;
goto release;
}
if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) {
if ((in6p->in6p_flags & IN6P_IPV6_V6ONLY)) {
/*
* XXX: this case would happen when the
* application sets the V6ONLY flag after
* connecting the foreign address.
* Such applications should be fixed,
* so we bark here.
*/
log(LOG_INFO, "udp6_output: IPV6_V6ONLY "
"option was set for a connected socket\n");
error = EINVAL;
goto release;
} else
af = AF_INET;
}
laddr = &in6p->in6p_laddr;
faddr = &in6p->in6p_faddr;
fport = in6p->in6p_fport;
}
if (af == AF_INET)
hlen = sizeof(struct ip);
/*
* Calculate data length and get a mbuf
* for UDP and IP6 headers.
*/
M_PREPEND(m, hlen + sizeof(struct udphdr), MB_DONTWAIT);
if (m == NULL) {
error = ENOBUFS;
goto release;
}
/*
* Stuff checksum and output datagram.
*/
udp6 = (struct udphdr *)(mtod(m, caddr_t) + hlen);
udp6->uh_sport = in6p->in6p_lport; /* lport is always set in the PCB */
udp6->uh_dport = fport;
if (plen <= 0xffff)
udp6->uh_ulen = htons((u_short)plen);
else
udp6->uh_ulen = 0;
udp6->uh_sum = 0;
switch (af) {
case AF_INET6:
ip6 = mtod(m, struct ip6_hdr *);
ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK;
ip6->ip6_vfc &= ~IPV6_VERSION_MASK;
ip6->ip6_vfc |= IPV6_VERSION;
#if 0 /* ip6_plen will be filled in ip6_output. */
ip6->ip6_plen = htons((u_short)plen);
#endif
ip6->ip6_nxt = IPPROTO_UDP;
ip6->ip6_hlim = in6_selecthlim(in6p,
in6p->in6p_route.ro_rt ?
in6p->in6p_route.ro_rt->rt_ifp : NULL);
ip6->ip6_src = *laddr;
ip6->ip6_dst = *faddr;
if ((udp6->uh_sum = in6_cksum(m, IPPROTO_UDP,
sizeof(struct ip6_hdr), plen)) == 0) {
udp6->uh_sum = 0xffff;
}
flags = 0;
udp6stat.udp6s_opackets++;
error = ip6_output(m, in6p->in6p_outputopts, &in6p->in6p_route,
flags, in6p->in6p_moptions, NULL, in6p);
break;
case AF_INET:
error = EAFNOSUPPORT;
goto release;
}
goto releaseopt;
release:
m_freem(m);
releaseopt:
if (control) {
ip6_clearpktopts(in6p->in6p_outputopts, -1);
in6p->in6p_outputopts = stickyopt;
m_freem(control);
}
return (error);
}
