/*
* XXX Liang:
*
* kern_file_*() are not safe for multi-threads now,
* however, we need them only for tracefiled, so it's
* not so important to implement for MT.
*/
int
kern_file_size(struct cfs_kern_file *fp, off_t *psize)
{
int error;
off_t size;
error = vnode_size(fp->f_vp, &size, fp->f_ctxt);
if (error)
return error;
if (psize)
*psize = size;
return 0;
}
static int
vniocattach_shadow(struct vn_softc *vn, struct vn_ioctl_64 *vniop,
__unused dev_t dev, int in_kernel, proc_t p)
{
vfs_context_t ctx = vfs_context_current();
struct nameidata nd;
int error, flags;
shadow_map_t * map;
off_t file_size;
flags = FREAD|FWRITE;
if (in_kernel) {
NDINIT(&nd, LOOKUP, OP_OPEN, FOLLOW, UIO_SYSSPACE, vniop->vn_file, ctx);
}
else {
NDINIT(&nd, LOOKUP, OP_OPEN, FOLLOW,
(IS_64BIT_PROCESS(p) ? UIO_USERSPACE64 : UIO_USERSPACE32),
vniop->vn_file, ctx);
}
/* vn_open gives both long- and short-term references */
error = vn_open(&nd, flags, 0);
if (error) {
/* shadow MUST be writable! */
return (error);
}
if (nd.ni_vp->v_type != VREG
|| (error = vnode_size(nd.ni_vp, &file_size, ctx))) {
(void)vn_close(nd.ni_vp, flags, ctx);
vnode_put(nd.ni_vp);
return (error ? error : EINVAL);
}
map = shadow_map_create(vn->sc_fsize, file_size,
0, vn->sc_secsize);
if (map == NULL) {
(void)vn_close(nd.ni_vp, flags, ctx);
vnode_put(nd.ni_vp);
vn->sc_shadow_vp = NULL;
return (ENOMEM);
}
vn->sc_shadow_vp = nd.ni_vp;
vn->sc_shadow_vid = vnode_vid(nd.ni_vp);
vn->sc_shadow_vp->v_flag |= VNOCACHE_DATA;
vn->sc_shadow_map = map;
vn->sc_flags &= ~VNF_READONLY; /* we're now read/write */
/* lose the short-term reference */
vnode_put(nd.ni_vp);
return(0);
}
int
vm_swapfile_preallocate(vnode_t vp, uint64_t *size, boolean_t *pin)
{
int error = 0;
uint64_t file_size = 0;
vfs_context_t ctx = NULL;
ctx = vfs_context_current();
error = vnode_setsize(vp, *size, IO_NOZEROFILL, ctx);
if (error) {
printf("vnode_setsize for swap files failed: %d\n", error);
goto done;
}
error = vnode_size(vp, (off_t*) &file_size, ctx);
if (error) {
printf("vnode_size (new file) for swap file failed: %d\n", error);
goto done;
}
assert(file_size == *size);
if (pin != NULL && *pin != FALSE) {
error = VNOP_IOCTL(vp, FIOPINSWAP, NULL, 0, ctx);
if (error) {
printf("pin for swap files failed: %d, file_size = %lld\n", error, file_size);
/* this is not fatal, carry on with files wherever they landed */
*pin = FALSE;
error = 0;
}
}
vnode_lock_spin(vp);
SET(vp->v_flag, VSWAP);
vnode_unlock(vp);
done:
return error;
}
static int
vniocattach_file(struct vn_softc *vn,
struct vn_ioctl_64 *vniop,
dev_t dev,
int in_kernel,
proc_t p)
{
dev_t cdev;
vfs_context_t ctx = vfs_context_current();
kauth_cred_t cred;
struct nameidata nd;
off_t file_size;
int error, flags;
flags = FREAD|FWRITE;
if (in_kernel) {
NDINIT(&nd, LOOKUP, OP_OPEN, FOLLOW, UIO_SYSSPACE, vniop->vn_file, ctx);
}
else {
NDINIT(&nd, LOOKUP, OP_OPEN, FOLLOW,
(IS_64BIT_PROCESS(p) ? UIO_USERSPACE64 : UIO_USERSPACE32),
vniop->vn_file, ctx);
}
/* vn_open gives both long- and short-term references */
error = vn_open(&nd, flags, 0);
if (error) {
if (error != EACCES && error != EPERM && error != EROFS) {
return (error);
}
flags &= ~FWRITE;
if (in_kernel) {
NDINIT(&nd, LOOKUP, OP_OPEN, FOLLOW, UIO_SYSSPACE,
vniop->vn_file, ctx);
}
else {
NDINIT(&nd, LOOKUP, OP_OPEN, FOLLOW,
(IS_64BIT_PROCESS(p) ? UIO_USERSPACE64 : UIO_USERSPACE32),
vniop->vn_file, ctx);
}
error = vn_open(&nd, flags, 0);
if (error) {
return (error);
}
}
if (nd.ni_vp->v_type != VREG) {
error = EINVAL;
}
else {
error = vnode_size(nd.ni_vp, &file_size, ctx);
}
if (error != 0) {
(void) vn_close(nd.ni_vp, flags, ctx);
vnode_put(nd.ni_vp);
return (error);
}
cred = kauth_cred_proc_ref(p);
nd.ni_vp->v_flag |= VNOCACHE_DATA;
error = setcred(nd.ni_vp, cred);
if (error) {
(void)vn_close(nd.ni_vp, flags, ctx);
vnode_put(nd.ni_vp);
kauth_cred_unref(&cred);
return(error);
}
vn->sc_secsize = DEV_BSIZE;
vn->sc_fsize = file_size;
vn->sc_size = file_size / vn->sc_secsize;
vn->sc_vp = nd.ni_vp;
vn->sc_vid = vnode_vid(nd.ni_vp);
vn->sc_open_flags = flags;
vn->sc_cred = cred;
cdev = makedev(vndevice_cdev_major, minor(dev));
vn->sc_cdev = devfs_make_node(cdev, DEVFS_CHAR,
UID_ROOT, GID_OPERATOR,
0600, "rvn%d",
minor(dev));
vn->sc_flags |= VNF_INITED;
if (flags == FREAD)
vn->sc_flags |= VNF_READONLY;
/* lose the short-term reference */
vnode_put(nd.ni_vp);
return(0);
}
/*
* Routine: macx_swapon
* Function:
* Syscall interface to add a file to backing store
*/
int
macx_swapon(
struct macx_swapon_args *args)
{
int size = args->size;
vnode_t vp = (vnode_t)NULL;
struct nameidata nd, *ndp;
register int error;
kern_return_t kr;
mach_port_t backing_store;
memory_object_default_t default_pager;
int i;
boolean_t funnel_state;
off_t file_size;
vfs_context_t ctx = vfs_context_current();
struct proc *p = current_proc();
int dp_cluster_size;
AUDIT_MACH_SYSCALL_ENTER(AUE_SWAPON);
AUDIT_ARG(value32, args->priority);
funnel_state = thread_funnel_set(kernel_flock, TRUE);
ndp = &nd;
if ((error = suser(kauth_cred_get(), 0)))
goto swapon_bailout;
/*
* Get a vnode for the paging area.
*/
NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNPATH1,
((IS_64BIT_PROCESS(p)) ? UIO_USERSPACE64 : UIO_USERSPACE32),
(user_addr_t) args->filename, ctx);
if ((error = namei(ndp)))
goto swapon_bailout;
nameidone(ndp);
vp = ndp->ni_vp;
if (vp->v_type != VREG) {
error = EINVAL;
goto swapon_bailout;
}
/* get file size */
if ((error = vnode_size(vp, &file_size, ctx)) != 0)
goto swapon_bailout;
#if CONFIG_MACF
vnode_lock(vp);
error = mac_system_check_swapon(vfs_context_ucred(ctx), vp);
vnode_unlock(vp);
if (error)
goto swapon_bailout;
#endif
/* resize to desired size if it's too small */
if ((file_size < (off_t)size) && ((error = vnode_setsize(vp, (off_t)size, 0, ctx)) != 0))
goto swapon_bailout;
if (default_pager_init_flag == 0) {
start_def_pager(NULL);
default_pager_init_flag = 1;
}
/* add new backing store to list */
i = 0;
while(bs_port_table[i].vp != 0) {
if(i == MAX_BACKING_STORE)
break;
i++;
}
if(i == MAX_BACKING_STORE) {
error = ENOMEM;
goto swapon_bailout;
}
/* remember the vnode. This vnode has namei() reference */
bs_port_table[i].vp = vp;
/*
* Look to see if we are already paging to this file.
*/
/* make certain the copy send of kernel call will work */
default_pager = MEMORY_OBJECT_DEFAULT_NULL;
kr = host_default_memory_manager(host_priv_self(), &default_pager, 0);
if(kr != KERN_SUCCESS) {
error = EAGAIN;
bs_port_table[i].vp = 0;
goto swapon_bailout;
}
if (vp->v_mount->mnt_kern_flag & MNTK_SSD) {
/*
* keep the cluster size small since the
* seek cost is effectively 0 which means
* we don't care much about fragmentation
*/
dp_isssd = TRUE;
dp_cluster_size = 2 * PAGE_SIZE;
} else {
/*
* use the default cluster size
*/
dp_isssd = FALSE;
dp_cluster_size = 0;
}
kr = default_pager_backing_store_create(default_pager,
-1, /* default priority */
dp_cluster_size,
&backing_store);
memory_object_default_deallocate(default_pager);
if(kr != KERN_SUCCESS) {
error = ENOMEM;
bs_port_table[i].vp = 0;
goto swapon_bailout;
}
/* Mark this vnode as being used for swapfile */
vnode_lock_spin(vp);
SET(vp->v_flag, VSWAP);
vnode_unlock(vp);
/*
* NOTE: we are able to supply PAGE_SIZE here instead of
* an actual record size or block number because:
* a: we do not support offsets from the beginning of the
* file (allowing for non page size/record modulo offsets.
* b: because allow paging will be done modulo page size
*/
kr = default_pager_add_file(backing_store, (vnode_ptr_t) vp,
PAGE_SIZE, (int)(file_size/PAGE_SIZE));
if(kr != KERN_SUCCESS) {
bs_port_table[i].vp = 0;
if(kr == KERN_INVALID_ARGUMENT)
error = EINVAL;
else
error = ENOMEM;
/* This vnode is not to be used for swapfile */
vnode_lock_spin(vp);
CLR(vp->v_flag, VSWAP);
vnode_unlock(vp);
goto swapon_bailout;
}
bs_port_table[i].bs = (void *)backing_store;
error = 0;
ubc_setthreadcred(vp, p, current_thread());
/*
* take a long term reference on the vnode to keep
* vnreclaim() away from this vnode.
*/
vnode_ref(vp);
swapon_bailout:
if (vp) {
vnode_put(vp);
}
(void) thread_funnel_set(kernel_flock, FALSE);
AUDIT_MACH_SYSCALL_EXIT(error);
return(error);
}
/*
* shared_region_map_np()
*
* This system call is intended for dyld.
*
* dyld uses this to map a shared cache file into a shared region.
* This is usually done only the first time a shared cache is needed.
* Subsequent processes will just use the populated shared region without
* requiring any further setup.
*/
int
shared_region_map_np(
struct proc *p,
struct shared_region_map_np_args *uap,
__unused int *retvalp)
{
int error;
kern_return_t kr;
int fd;
struct fileproc *fp;
struct vnode *vp, *root_vp;
struct vnode_attr va;
off_t fs;
memory_object_size_t file_size;
user_addr_t user_mappings;
struct shared_file_mapping_np *mappings;
#define SFM_MAX_STACK 8
struct shared_file_mapping_np stack_mappings[SFM_MAX_STACK];
unsigned int mappings_count;
vm_size_t mappings_size;
memory_object_control_t file_control;
struct vm_shared_region *shared_region;
SHARED_REGION_TRACE_DEBUG(
("shared_region: %p [%d(%s)] -> map\n",
current_thread(), p->p_pid, p->p_comm));
shared_region = NULL;
mappings_count = 0;
mappings_size = 0;
mappings = NULL;
fp = NULL;
vp = NULL;
/* get file descriptor for shared region cache file */
fd = uap->fd;
/* get file structure from file descriptor */
error = fp_lookup(p, fd, &fp, 0);
if (error) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map: "
"fd=%d lookup failed (error=%d)\n",
current_thread(), p->p_pid, p->p_comm, fd, error));
goto done;
}
/* make sure we're attempting to map a vnode */
if (fp->f_fglob->fg_type != DTYPE_VNODE) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map: "
"fd=%d not a vnode (type=%d)\n",
current_thread(), p->p_pid, p->p_comm,
fd, fp->f_fglob->fg_type));
error = EINVAL;
goto done;
}
/* we need at least read permission on the file */
if (! (fp->f_fglob->fg_flag & FREAD)) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map: "
"fd=%d not readable\n",
current_thread(), p->p_pid, p->p_comm, fd));
error = EPERM;
goto done;
}
/* get vnode from file structure */
error = vnode_getwithref((vnode_t) fp->f_fglob->fg_data);
if (error) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map: "
"fd=%d getwithref failed (error=%d)\n",
current_thread(), p->p_pid, p->p_comm, fd, error));
goto done;
}
vp = (struct vnode *) fp->f_fglob->fg_data;
/* make sure the vnode is a regular file */
if (vp->v_type != VREG) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"not a file (type=%d)\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name, vp->v_type));
error = EINVAL;
goto done;
}
/* make sure vnode is on the process's root volume */
root_vp = p->p_fd->fd_rdir;
if (root_vp == NULL) {
root_vp = rootvnode;
}
if (vp->v_mount != root_vp->v_mount) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"not on process's root volume\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name));
error = EPERM;
goto done;
}
/* make sure vnode is owned by "root" */
VATTR_INIT(&va);
VATTR_WANTED(&va, va_uid);
error = vnode_getattr(vp, &va, vfs_context_current());
if (error) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"vnode_getattr(%p) failed (error=%d)\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name, vp, error));
goto done;
}
if (va.va_uid != 0) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"owned by uid=%d instead of 0\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name, va.va_uid));
error = EPERM;
goto done;
}
/* get vnode size */
error = vnode_size(vp, &fs, vfs_context_current());
if (error) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"vnode_size(%p) failed (error=%d)\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name, vp, error));
goto done;
}
file_size = fs;
/* get the file's memory object handle */
file_control = ubc_getobject(vp, UBC_HOLDOBJECT);
if (file_control == MEMORY_OBJECT_CONTROL_NULL) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"no memory object\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name));
error = EINVAL;
goto done;
}
/* get the list of mappings the caller wants us to establish */
mappings_count = uap->count; /* number of mappings */
mappings_size = (vm_size_t) (mappings_count * sizeof (mappings[0]));
if (mappings_count == 0) {
SHARED_REGION_TRACE_INFO(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"no mappings\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name));
error = 0; /* no mappings: we're done ! */
goto done;
} else if (mappings_count <= SFM_MAX_STACK) {
mappings = &stack_mappings[0];
} else {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"too many mappings (%d)\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name, mappings_count));
error = EINVAL;
goto done;
}
user_mappings = uap->mappings; /* the mappings, in user space */
error = copyin(user_mappings,
mappings,
mappings_size);
if (error) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"copyin(0x%llx, %d) failed (error=%d)\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name, (uint64_t)user_mappings, mappings_count, error));
goto done;
}
/* get the process's shared region (setup in vm_map_exec()) */
shared_region = vm_shared_region_get(current_task());
if (shared_region == NULL) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"no shared region\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name));
goto done;
}
/* map the file into that shared region's submap */
kr = vm_shared_region_map_file(shared_region,
mappings_count,
mappings,
file_control,
file_size,
(void *) p->p_fd->fd_rdir);
if (kr != KERN_SUCCESS) {
SHARED_REGION_TRACE_ERROR(
("shared_region: %p [%d(%s)] map(%p:'%s'): "
"vm_shared_region_map_file() failed kr=0x%x\n",
current_thread(), p->p_pid, p->p_comm,
vp, vp->v_name, kr));
switch (kr) {
case KERN_INVALID_ADDRESS:
error = EFAULT;
break;
case KERN_PROTECTION_FAILURE:
error = EPERM;
break;
case KERN_NO_SPACE:
error = ENOMEM;
break;
case KERN_FAILURE:
case KERN_INVALID_ARGUMENT:
default:
error = EINVAL;
break;
}
goto done;
}
/*
* The mapping was successful. Let the buffer cache know
* that we've mapped that file with these protections. This
* prevents the vnode from getting recycled while it's mapped.
*/
(void) ubc_map(vp, VM_PROT_READ);
error = 0;
/* update the vnode's access time */
if (! (vnode_vfsvisflags(vp) & MNT_NOATIME)) {
VATTR_INIT(&va);
nanotime(&va.va_access_time);
VATTR_SET_ACTIVE(&va, va_access_time);
vnode_setattr(vp, &va, vfs_context_current());
}
if (p->p_flag & P_NOSHLIB) {
/* signal that this process is now using split libraries */
OSBitAndAtomic(~((uint32_t)P_NOSHLIB), (UInt32 *)&p->p_flag);
}
done:
if (vp != NULL) {
/*
* release the vnode...
* ubc_map() still holds it for us in the non-error case
*/
(void) vnode_put(vp);
vp = NULL;
}
if (fp != NULL) {
/* release the file descriptor */
fp_drop(p, fd, fp, 0);
fp = NULL;
}
if (shared_region != NULL) {
vm_shared_region_deallocate(shared_region);
}
SHARED_REGION_TRACE_DEBUG(
("shared_region: %p [%d(%s)] <- map\n",
current_thread(), p->p_pid, p->p_comm));
return error;
}
int
vm_swapfile_preallocate(vnode_t vp, uint64_t *size)
{
int error = 0;
uint64_t file_size = 0;
vfs_context_t ctx = NULL;
ctx = vfs_context_current();
#if CONFIG_PROTECT
{
#if 0 // <rdar://11771612>
if ((error = cp_vnode_setclass(vp, PROTECTION_CLASS_F))) {
if(config_protect_bug) {
printf("swap protection class set failed with %d\n", error);
} else {
panic("swap protection class set failed with %d\n", error);
}
}
#endif
/* initialize content protection keys manually */
if ((error = cp_handle_vnop(vp, CP_WRITE_ACCESS, 0)) != 0) {
printf("Content Protection key failure on swap: %d\n", error);
vnode_put(vp);
vp = NULL;
goto done;
}
}
#endif
/*
* This check exists because dynamic_pager creates the 1st swapfile,
* swapfile0, for us from user-space in a supported manner (with IO_NOZEROFILL etc).
*
* If dynamic_pager, in the future, discontinues creating that file,
* then we need to change this check to a panic / assert or return an error.
* That's because we can't be sure if the file has been created correctly.
*/
if ((error = vnode_size(vp, (off_t*) &file_size, ctx)) != 0) {
printf("vnode_size (existing files) for swap files failed: %d\n", error);
goto done;
} else {
if (file_size == 0) {
error = vnode_setsize(vp, *size, IO_NOZEROFILL, ctx);
if (error) {
printf("vnode_setsize for swap files failed: %d\n", error);
goto done;
}
} else {
*size = file_size;
}
}
vnode_lock_spin(vp);
SET(vp->v_flag, VSWAP);
vnode_unlock(vp);
done:
return error;
}
int
vm_swapfile_preallocate(vnode_t vp, uint64_t *size, boolean_t *pin)
{
int error = 0;
uint64_t file_size = 0;
vfs_context_t ctx = NULL;
ctx = vfs_context_current();
#if CONFIG_PROTECT
{
#if 0 // <rdar://11771612>
if ((error = cp_vnode_setclass(vp, PROTECTION_CLASS_F))) {
if(config_protect_bug) {
printf("swap protection class set failed with %d\n", error);
} else {
panic("swap protection class set failed with %d\n", error);
}
}
#endif
/* initialize content protection keys manually */
if ((error = cp_handle_vnop(vp, CP_WRITE_ACCESS, 0)) != 0) {
printf("Content Protection key failure on swap: %d\n", error);
vnode_put(vp);
vp = NULL;
goto done;
}
}
#endif
error = vnode_setsize(vp, *size, IO_NOZEROFILL, ctx);
if (error) {
printf("vnode_setsize for swap files failed: %d\n", error);
goto done;
}
error = vnode_size(vp, (off_t*) &file_size, ctx);
if (error) {
printf("vnode_size (new file) for swap file failed: %d\n", error);
goto done;
}
assert(file_size == *size);
if (pin != NULL && *pin != FALSE) {
assert(vnode_tag(vp) == VT_HFS);
error = hfs_pin_vnode(VTOHFS(vp), vp, HFS_PIN_IT | HFS_DATALESS_PIN, NULL, ctx);
if (error) {
printf("hfs_pin_vnode for swap files failed: %d\n", error);
/* this is not fatal, carry on with files wherever they landed */
*pin = FALSE;
error = 0;
}
}
vnode_lock_spin(vp);
SET(vp->v_flag, VSWAP);
vnode_unlock(vp);
done:
return error;
}
/*
* This routine exists to support the load_dylinker().
*
* This routine has its own, separate, understanding of the FAT file format,
* which is terrifically unfortunate.
*/
static
load_return_t
get_macho_vnode(
char *path,
integer_t archbits,
struct mach_header *mach_header,
off_t *file_offset,
off_t *macho_size,
struct vnode **vpp
)
{
struct vnode *vp;
vfs_context_t ctx = vfs_context_current();
proc_t p = vfs_context_proc(ctx);
kauth_cred_t kerncred;
struct nameidata nid, *ndp;
boolean_t is_fat;
struct fat_arch fat_arch;
int error = LOAD_SUCCESS;
int resid;
union {
struct mach_header mach_header;
struct fat_header fat_header;
char pad[512];
} header;
off_t fsize = (off_t)0;
int err2;
/*
* Capture the kernel credential for use in the actual read of the
* file, since the user doing the execution may have execute rights
* but not read rights, but to exec something, we have to either map
* or read it into the new process address space, which requires
* read rights. This is to deal with lack of common credential
* serialization code which would treat NOCRED as "serialize 'root'".
*/
kerncred = vfs_context_ucred(vfs_context_kernel());
ndp = &nid;
/* init the namei data to point the file user's program name */
NDINIT(ndp, LOOKUP, FOLLOW | LOCKLEAF, UIO_SYSSPACE32, CAST_USER_ADDR_T(path), ctx);
if ((error = namei(ndp)) != 0) {
if (error == ENOENT) {
error = LOAD_ENOENT;
} else {
error = LOAD_FAILURE;
}
return(error);
}
nameidone(ndp);
vp = ndp->ni_vp;
/* check for regular file */
if (vp->v_type != VREG) {
error = LOAD_PROTECT;
goto bad1;
}
/* get size */
if ((error = vnode_size(vp, &fsize, ctx)) != 0) {
error = LOAD_FAILURE;
goto bad1;
}
/* Check mount point */
if (vp->v_mount->mnt_flag & MNT_NOEXEC) {
error = LOAD_PROTECT;
goto bad1;
}
/* check access */
if ((error = vnode_authorize(vp, NULL, KAUTH_VNODE_EXECUTE, ctx)) != 0) {
error = LOAD_PROTECT;
goto bad1;
}
/* try to open it */
if ((error = VNOP_OPEN(vp, FREAD, ctx)) != 0) {
error = LOAD_PROTECT;
goto bad1;
}
if ((error = vn_rdwr(UIO_READ, vp, (caddr_t)&header, sizeof(header), 0,
UIO_SYSSPACE32, IO_NODELOCKED, kerncred, &resid, p)) != 0) {
error = LOAD_IOERROR;
goto bad2;
}
if (header.mach_header.magic == MH_MAGIC ||
header.mach_header.magic == MH_MAGIC_64)
is_fat = FALSE;
else if (header.fat_header.magic == FAT_MAGIC ||
header.fat_header.magic == FAT_CIGAM)
is_fat = TRUE;
else {
error = LOAD_BADMACHO;
goto bad2;
}
if (is_fat) {
/* Look up our architecture in the fat file. */
error = fatfile_getarch_with_bits(vp, archbits, (vm_offset_t)(&header.fat_header), &fat_arch);
if (error != LOAD_SUCCESS)
goto bad2;
/* Read the Mach-O header out of it */
error = vn_rdwr(UIO_READ, vp, (caddr_t)&header.mach_header,
sizeof(header.mach_header), fat_arch.offset,
UIO_SYSSPACE32, IO_NODELOCKED, kerncred, &resid, p);
if (error) {
error = LOAD_IOERROR;
goto bad2;
}
/* Is this really a Mach-O? */
if (header.mach_header.magic != MH_MAGIC &&
header.mach_header.magic != MH_MAGIC_64) {
error = LOAD_BADMACHO;
goto bad2;
}
*file_offset = fat_arch.offset;
*macho_size = fat_arch.size;
} else {
/*
* Force get_macho_vnode() to fail if the architecture bits
* do not match the expected architecture bits. This in
* turn causes load_dylinker() to fail for the same reason,
* so it ensures the dynamic linker and the binary are in
* lock-step. This is potentially bad, if we ever add to
* the CPU_ARCH_* bits any bits that are desirable but not
* required, since the dynamic linker might work, but we will
* refuse to load it because of this check.
*/
if ((cpu_type_t)(header.mach_header.cputype & CPU_ARCH_MASK) != archbits)
return(LOAD_BADARCH);
*file_offset = 0;
*macho_size = fsize;
}
*mach_header = header.mach_header;
*vpp = vp;
ubc_setsize(vp, fsize);
return (error);
bad2:
err2 = VNOP_CLOSE(vp, FREAD, ctx);
vnode_put(vp);
return (error);
bad1:
vnode_put(vp);
return(error);
}