gems-kernel/source/THIRDPARTY/xnu/bsd/kern/kern_guarded.c
2024-06-03 11:29:39 -05:00

1552 lines
39 KiB
C

/*
* Copyright (c) 2018 Apple Inc. All rights reserved.
*
* @APPLE_OSREFERENCE_LICENSE_HEADER_START@
*
* This file contains Original Code and/or Modifications of Original Code
* as defined in and that are subject to the Apple Public Source License
* Version 2.0 (the 'License'). You may not use this file except in
* compliance with the License. The rights granted to you under the License
* may not be used to create, or enable the creation or redistribution of,
* unlawful or unlicensed copies of an Apple operating system, or to
* circumvent, violate, or enable the circumvention or violation of, any
* terms of an Apple operating system software license agreement.
*
* Please obtain a copy of the License at
* http://www.opensource.apple.com/apsl/ and read it before using this file.
*
* The Original Code and all software distributed under the License are
* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
* Please see the License for the specific language governing rights and
* limitations under the License.
*
* @APPLE_OSREFERENCE_LICENSE_HEADER_END@
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/filedesc.h>
#include <sys/kernel.h>
#include <sys/file_internal.h>
#include <sys/guarded.h>
#include <sys/sysproto.h>
#include <sys/vnode.h>
#include <sys/vnode_internal.h>
#include <sys/uio_internal.h>
#include <sys/ubc_internal.h>
#include <vfs/vfs_support.h>
#include <security/audit/audit.h>
#include <sys/syscall.h>
#include <sys/kauth.h>
#include <sys/kdebug.h>
#include <stdbool.h>
#include <vm/vm_protos.h>
#include <libkern/section_keywords.h>
#include <kern/kalloc.h>
#include <kern/task.h>
#include <kern/exc_guard.h>
#if CONFIG_MACF && CONFIG_VNGUARD
#include <security/mac.h>
#include <security/mac_framework.h>
#include <security/mac_policy.h>
#include <pexpert/pexpert.h>
#include <sys/sysctl.h>
#include <sys/reason.h>
#endif
#define f_flag fp_glob->fg_flag
extern int writev_uio(struct proc *p, int fd, user_addr_t user_iovp,
int iovcnt, off_t offset, int flags, guardid_t *puguard,
user_ssize_t *retval);
extern int write_internal(struct proc *p, int fd, user_addr_t buf,
user_size_t nbyte, off_t offset, int flags, guardid_t *puguard,
user_ssize_t *retval);
extern int exit_with_guard_exception(void *p, mach_exception_data_type_t code,
mach_exception_data_type_t subcode);
/*
* Experimental guarded file descriptor support.
*/
kern_return_t task_exception_notify(exception_type_t exception,
mach_exception_data_type_t code, mach_exception_data_type_t subcode, const bool fatal);
#define GUARD_REQUIRED (GUARD_DUP)
#define GUARD_ALL (GUARD_REQUIRED | \
(GUARD_CLOSE | GUARD_SOCKET_IPC | GUARD_FILEPORT | GUARD_WRITE))
static KALLOC_TYPE_DEFINE(fp_guard_zone, struct fileproc_guard, KT_DEFAULT);
struct gfp_crarg {
guardid_t gca_guard;
uint16_t gca_attrs;
};
static struct fileproc_guard *
guarded_fileproc_alloc(guardid_t guard)
{
struct fileproc_guard *fpg;
fpg = zalloc_flags(fp_guard_zone, Z_WAITOK | Z_ZERO | Z_NOFAIL);
fpg->fpg_guard = guard;
return fpg;
}
static void
guarded_fileproc_init(struct fileproc *fp, void *initarg)
{
struct gfp_crarg *arg = initarg;
assert(arg->gca_attrs);
fp->fp_guard = guarded_fileproc_alloc(arg->gca_guard);
fp->fp_guard_attrs = arg->gca_attrs;
}
/*
* This is called from fdt_fork(),
* where it needs to copy a guarded
* fd to the new shadow proc.
*/
void
guarded_fileproc_copy_guard(struct fileproc *ofp, struct fileproc *nfp)
{
struct gfp_crarg arg = {
.gca_guard = ofp->fp_guard->fpg_guard,
.gca_attrs = ofp->fp_guard_attrs
};
guarded_fileproc_init(nfp, &arg);
}
/*
* This is called from fileproc_free(),
* which is why it is safe to call
* without holding the proc_fdlock.
*/
void
guarded_fileproc_unguard(struct fileproc *fp)
{
struct fileproc_guard *fpg = fp->fp_guard;
fp->fp_guard_attrs = 0;
fp->fp_wset = fpg->fpg_wset;
zfree(fp_guard_zone, fpg);
}
static int
fp_lookup_guarded_locked(proc_t p, int fd, guardid_t guard,
struct fileproc **fpp)
{
int error;
struct fileproc *fp;
if ((error = fp_lookup(p, fd, &fp, 1)) != 0) {
return error;
}
if (fp->fp_guard_attrs == 0) {
(void) fp_drop(p, fd, fp, 1);
return EINVAL;
}
if (guard != fp->fp_guard->fpg_guard) {
(void) fp_drop(p, fd, fp, 1);
return EPERM; /* *not* a mismatch exception */
}
*fpp = fp;
return 0;
}
int
fp_lookup_guarded(proc_t p, int fd, guardid_t guard,
struct fileproc **fpp, int locked)
{
int error;
if (!locked) {
proc_fdlock_spin(p);
}
error = fp_lookup_guarded_locked(p, fd, guard, fpp);
if (!locked) {
proc_fdunlock(p);
}
return error;
}
/*
* Expected use pattern:
*
* if (fp_isguarded(fp, GUARD_CLOSE)) {
* error = fp_guard_exception(p, fd, fp, kGUARD_EXC_CLOSE);
* proc_fdunlock(p);
* return error;
* }
*/
int
fp_isguarded(struct fileproc *fp, u_int attrs)
{
return fp->fp_guard_attrs && (fp->fp_guard_attrs & attrs) == attrs;
}
extern char *proc_name_address(void *p);
int
fp_guard_exception(proc_t p, int fd, struct fileproc *fp, u_int flavor)
{
/* all fp guard fields protected via proc_fdlock() */
proc_fdlock_assert(p, LCK_MTX_ASSERT_OWNED);
mach_exception_code_t code = 0;
EXC_GUARD_ENCODE_TYPE(code, GUARD_TYPE_FD);
EXC_GUARD_ENCODE_FLAVOR(code, flavor);
EXC_GUARD_ENCODE_TARGET(code, fd);
mach_exception_subcode_t subcode = fp->fp_guard->fpg_guard;
assert(fp->fp_guard_attrs);
thread_t t = current_thread();
thread_guard_violation(t, code, subcode, TRUE);
return EPERM;
}
/*
* (Invoked before returning to userland from the syscall handler.)
*/
void
fd_guard_ast(
thread_t __unused t,
mach_exception_code_t code,
mach_exception_subcode_t subcode)
{
const bool fatal = true;
/*
* Check if anyone has registered for Synchronous EXC_GUARD, if yes then,
* deliver it synchronously and then kill the process, else kill the process
* and deliver the exception via EXC_CORPSE_NOTIFY. Always kill the process if we are not in dev mode.
*/
if (task_exception_notify(EXC_GUARD, code, subcode, fatal) == KERN_SUCCESS) {
psignal(current_proc(), SIGKILL);
} else {
exit_with_guard_exception(current_proc(), code, subcode);
}
}
/*
* Experimental guarded file descriptor SPIs
*/
/*
* int guarded_open_np(const char *pathname, int flags,
* const guardid_t *guard, u_int guardflags, ...);
*
* In this initial implementation, GUARD_DUP must be specified.
* GUARD_CLOSE, GUARD_SOCKET_IPC and GUARD_FILEPORT are optional.
*
* If GUARD_DUP wasn't specified, then we'd have to do the (extra) work
* to allow dup-ing a descriptor to inherit the guard onto the new
* descriptor. (Perhaps GUARD_DUP behaviours should just always be true
* for a guarded fd? Or, more sanely, all the dup operations should
* just always propagate the guard?)
*
* Guarded descriptors are always close-on-exec, and GUARD_CLOSE
* requires close-on-fork; O_CLOEXEC must be set in flags.
* This setting is immutable; attempts to clear the flag will
* cause a guard exception.
*
* XXX It's somewhat broken that change_fdguard_np() can completely
* remove the guard and thus revoke down the immutability
* promises above. Ick.
*/
int
guarded_open_np(proc_t p, struct guarded_open_np_args *uap, int32_t *retval)
{
if ((uap->flags & O_CLOEXEC) == 0) {
return EINVAL;
}
if (((uap->guardflags & GUARD_REQUIRED) != GUARD_REQUIRED) ||
((uap->guardflags & ~GUARD_ALL) != 0)) {
return EINVAL;
}
int error;
struct gfp_crarg crarg = {
.gca_attrs = (uint16_t)uap->guardflags
};
if ((error = copyin(uap->guard,
&(crarg.gca_guard), sizeof(crarg.gca_guard))) != 0) {
return error;
}
/*
* Disallow certain guard values -- is zero enough?
*/
if (crarg.gca_guard == 0) {
return EINVAL;
}
struct vnode_attr va;
struct nameidata nd;
vfs_context_t ctx = vfs_context_current();
int cmode;
VATTR_INIT(&va);
cmode = ((uap->mode & ~p->p_fd.fd_cmask) & ALLPERMS) & ~S_ISTXT;
VATTR_SET(&va, va_mode, cmode & ACCESSPERMS);
NDINIT(&nd, LOOKUP, OP_OPEN, FOLLOW | AUDITVNPATH1, UIO_USERSPACE,
uap->path, ctx);
return open1(ctx, &nd, uap->flags | O_CLOFORK, &va,
guarded_fileproc_init, &crarg, retval, AUTH_OPEN_NOAUTHFD);
}
/*
* int guarded_open_dprotected_np(const char *pathname, int flags,
* const guardid_t *guard, u_int guardflags, int dpclass, int dpflags, ...);
*
* This SPI is extension of guarded_open_np() to include dataprotection class on creation
* in "dpclass" and dataprotection flags 'dpflags'. Otherwise behaviors are same as in
* guarded_open_np()
*/
int
guarded_open_dprotected_np(proc_t p, struct guarded_open_dprotected_np_args *uap, int32_t *retval)
{
if ((uap->flags & O_CLOEXEC) == 0) {
return EINVAL;
}
if (((uap->guardflags & GUARD_REQUIRED) != GUARD_REQUIRED) ||
((uap->guardflags & ~GUARD_ALL) != 0)) {
return EINVAL;
}
int error;
struct gfp_crarg crarg = {
.gca_attrs = (uint16_t)uap->guardflags
};
if ((error = copyin(uap->guard,
&(crarg.gca_guard), sizeof(crarg.gca_guard))) != 0) {
return error;
}
/*
* Disallow certain guard values -- is zero enough?
*/
if (crarg.gca_guard == 0) {
return EINVAL;
}
struct vnode_attr va;
struct nameidata nd;
vfs_context_t ctx = vfs_context_current();
int cmode;
VATTR_INIT(&va);
cmode = ((uap->mode & ~p->p_fd.fd_cmask) & ALLPERMS) & ~S_ISTXT;
VATTR_SET(&va, va_mode, cmode & ACCESSPERMS);
NDINIT(&nd, LOOKUP, OP_OPEN, FOLLOW | AUDITVNPATH1, UIO_USERSPACE,
uap->path, ctx);
/*
* Initialize the extra fields in vnode_attr to pass down dataprotection
* extra fields.
* 1. target cprotect class.
* 2. set a flag to mark it as requiring open-raw-encrypted semantics.
*/
if (uap->flags & O_CREAT) {
VATTR_SET(&va, va_dataprotect_class, uap->dpclass);
}
if (uap->dpflags & (O_DP_GETRAWENCRYPTED | O_DP_GETRAWUNENCRYPTED)) {
if (uap->flags & (O_RDWR | O_WRONLY)) {
/* Not allowed to write raw encrypted bytes */
return EINVAL;
}
if (uap->dpflags & O_DP_GETRAWENCRYPTED) {
VATTR_SET(&va, va_dataprotect_flags, VA_DP_RAWENCRYPTED);
}
if (uap->dpflags & O_DP_GETRAWUNENCRYPTED) {
VATTR_SET(&va, va_dataprotect_flags, VA_DP_RAWUNENCRYPTED);
}
}
return open1(ctx, &nd, uap->flags | O_CLOFORK, &va,
guarded_fileproc_init, &crarg, retval, AUTH_OPEN_NOAUTHFD);
}
/*
* int guarded_kqueue_np(const guardid_t *guard, u_int guardflags);
*
* Create a guarded kqueue descriptor with guardid and guardflags.
*
* Same restrictions on guardflags as for guarded_open_np().
* All kqueues are -always- close-on-exec and close-on-fork by themselves
* and are not sendable.
*/
int
guarded_kqueue_np(proc_t p, struct guarded_kqueue_np_args *uap, int32_t *retval)
{
if (((uap->guardflags & GUARD_REQUIRED) != GUARD_REQUIRED) ||
((uap->guardflags & ~GUARD_ALL) != 0)) {
return EINVAL;
}
int error;
struct gfp_crarg crarg = {
.gca_attrs = (uint16_t)uap->guardflags
};
if ((error = copyin(uap->guard,
&(crarg.gca_guard), sizeof(crarg.gca_guard))) != 0) {
return error;
}
if (crarg.gca_guard == 0) {
return EINVAL;
}
return kqueue_internal(p, guarded_fileproc_init, &crarg, retval);
}
/*
* int guarded_close_np(int fd, const guardid_t *guard);
*/
int
guarded_close_np(proc_t p, struct guarded_close_np_args *uap,
__unused int32_t *retval)
{
struct fileproc *fp;
kauth_cred_t p_cred;
int fd = uap->fd;
int error;
guardid_t uguard;
AUDIT_SYSCLOSE(p, fd);
if ((error = copyin(uap->guard, &uguard, sizeof(uguard))) != 0) {
return error;
}
proc_fdlock(p);
if ((error = fp_lookup_guarded(p, fd, uguard, &fp, 1)) != 0) {
proc_fdunlock(p);
return error;
}
fp_drop(p, fd, fp, 1);
p_cred = current_cached_proc_cred(p);
return fp_close_and_unlock(p, p_cred, fd, fp, 0);
}
/*
* int
* change_fdguard_np(int fd, const guardid_t *guard, u_int guardflags,
* const guardid_t *nguard, u_int nguardflags, int *fdflagsp);
*
* Given a file descriptor, atomically exchange <guard, guardflags> for
* a new guard <nguard, nguardflags>, returning the previous fd
* flags (see fcntl:F_SETFD) in *fdflagsp.
*
* This syscall can be used to either (a) add a new guard to an existing
* unguarded file descriptor (b) remove the old guard from an existing
* guarded file descriptor or (c) change the guard (guardid and/or
* guardflags) on a guarded file descriptor.
*
* If 'guard' is NULL, fd must be unguarded at entry. If the call completes
* successfully the fd will be guarded with <nguard, nguardflags>.
*
* Guarding a file descriptor has some side-effects on the "fp_flags"
* associated with the descriptor - in particular FD_CLOEXEC is
* forced ON unconditionally, and FD_CLOFORK is forced ON by GUARD_CLOSE.
* Callers who wish to subsequently restore the state of the fd should save
* the value of *fdflagsp after a successful invocation.
*
* If 'nguard' is NULL, fd must be guarded at entry, <guard, guardflags>
* must match with what's already guarding the descriptor, and the
* result will be to completely remove the guard.
*
* If the descriptor is guarded, and neither 'guard' nor 'nguard' is NULL
* and <guard, guardflags> matches what's already guarding the descriptor,
* then <nguard, nguardflags> becomes the new guard. In this case, even if
* the GUARD_CLOSE flag is being cleared, it is still possible to continue
* to keep FD_CLOFORK on the descriptor by passing FD_CLOFORK via fdflagsp.
*
* (File descriptors whose underlying fileglobs are marked FG_CONFINED are
* still close-on-fork, regardless of the setting of FD_CLOFORK.)
*
* Example 1: Guard an unguarded descriptor during a set of operations,
* then restore the original state of the descriptor.
*
* int sav_flags = 0;
* change_fdguard_np(fd, NULL, 0, &myguard, GUARD_CLOSE, &sav_flags);
* // do things with now guarded 'fd'
* change_fdguard_np(fd, &myguard, GUARD_CLOSE, NULL, 0, &sav_flags);
* // fd now unguarded.
*
* Example 2: Change the guard of a guarded descriptor during a set of
* operations, then restore the original state of the descriptor.
*
* int sav_flags = (gdflags & GUARD_CLOSE) ? FD_CLOFORK : 0;
* change_fdguard_np(fd, &gd, gdflags, &myguard, GUARD_CLOSE, &sav_flags);
* // do things with 'fd' with a different guard
* change_fdguard_np(fd, &myg, GUARD_CLOSE, &gd, gdflags, &sav_flags);
* // back to original guarded state
*
* XXX This SPI is too much of a chainsaw and should be revised.
*/
int
change_fdguard_np(proc_t p, struct change_fdguard_np_args *uap,
__unused int32_t *retval)
{
struct fileproc_guard *fpg = NULL;
struct fileproc *fp;
int fd = uap->fd;
int error;
guardid_t oldg = 0, newg = 0;
int nfdflags = 0;
if (0 != uap->guard &&
0 != (error = copyin(uap->guard, &oldg, sizeof(oldg)))) {
return error; /* can't copyin current guard */
}
if (0 != uap->nguard &&
0 != (error = copyin(uap->nguard, &newg, sizeof(newg)))) {
return error; /* can't copyin new guard */
}
if (0 != uap->fdflagsp &&
0 != (error = copyin(uap->fdflagsp, &nfdflags, sizeof(nfdflags)))) {
return error; /* can't copyin new fdflags */
}
if (oldg == 0 && newg) {
fpg = guarded_fileproc_alloc(newg);
}
proc_fdlock(p);
if ((error = fp_lookup(p, fd, &fp, 1)) != 0) {
proc_fdunlock(p);
return error;
}
if (0 != uap->fdflagsp) {
int ofl = 0;
if (fp->fp_flags & FP_CLOEXEC) {
ofl |= FD_CLOEXEC;
}
if (fp->fp_flags & FP_CLOFORK) {
ofl |= FD_CLOFORK;
}
proc_fdunlock(p);
if (0 != (error = copyout(&ofl, uap->fdflagsp, sizeof(ofl)))) {
proc_fdlock(p);
goto dropout; /* can't copyout old fdflags */
}
proc_fdlock(p);
}
if (fp->fp_guard_attrs) {
if (0 == uap->guard || 0 == uap->guardflags) {
error = EINVAL; /* missing guard! */
} else if (0 == oldg) {
error = EPERM; /* guardids cannot be zero */
}
} else {
if (0 != uap->guard || 0 != uap->guardflags) {
error = EINVAL; /* guard provided, but none needed! */
}
}
if (0 != error) {
goto dropout;
}
if (0 != uap->nguard) {
/*
* There's a new guard in town.
*/
if (0 == newg) {
error = EINVAL; /* guards cannot contain zero */
} else if (((uap->nguardflags & GUARD_REQUIRED) != GUARD_REQUIRED) ||
((uap->nguardflags & ~GUARD_ALL) != 0)) {
error = EINVAL; /* must have valid attributes too */
}
if (0 != error) {
goto dropout;
}
if (fp->fp_guard_attrs) {
/*
* Replace old guard with new guard
*/
if (oldg == fp->fp_guard->fpg_guard &&
uap->guardflags == fp->fp_guard_attrs) {
/*
* Must match existing guard + attributes
* before we'll swap them to new ones, managing
* fdflags "side-effects" as we go. Note that
* userland can request FD_CLOFORK semantics.
*/
if (fp->fp_guard_attrs & GUARD_CLOSE) {
fp->fp_flags &= ~FP_CLOFORK;
}
fp->fp_guard->fpg_guard = newg;
fp->fp_guard_attrs = (uint16_t)uap->nguardflags;
if ((fp->fp_guard_attrs & GUARD_CLOSE) ||
(nfdflags & FD_CLOFORK)) {
fp->fp_flags |= FP_CLOFORK;
}
/* FG_CONFINED enforced regardless */
} else {
error = EPERM;
}
} else {
/*
* Add a guard to a previously unguarded descriptor
*/
switch (FILEGLOB_DTYPE(fp->fp_glob)) {
case DTYPE_VNODE:
case DTYPE_PIPE:
case DTYPE_SOCKET:
case DTYPE_KQUEUE:
case DTYPE_NETPOLICY:
break;
default:
error = ENOTSUP;
goto dropout;
}
fp->fp_guard_attrs = (uint16_t)uap->nguardflags;
fpg->fpg_wset = fp->fp_wset;
fp->fp_guard = fpg;
fpg = NULL;
if (fp->fp_guard_attrs & GUARD_CLOSE) {
fp->fp_flags |= FP_CLOFORK;
}
fp->fp_flags |= FP_CLOEXEC;
}
} else {
if (fp->fp_guard_attrs) {
/*
* Remove the guard altogether.
*/
if (0 != uap->nguardflags) {
error = EINVAL;
goto dropout;
}
if (oldg != fp->fp_guard->fpg_guard ||
uap->guardflags != fp->fp_guard_attrs) {
error = EPERM;
goto dropout;
}
assert(fpg == NULL);
fp->fp_guard_attrs = 0;
fpg = fp->fp_guard;
fp->fp_wset = fpg->fpg_wset;
fp->fp_flags &= ~(FP_CLOEXEC | FP_CLOFORK);
if (nfdflags & FD_CLOFORK) {
fp->fp_flags |= FP_CLOFORK;
}
if (nfdflags & FD_CLOEXEC) {
fp->fp_flags |= FP_CLOEXEC;
}
} else {
/*
* Not already guarded, and no new guard?
*/
error = EINVAL;
}
}
dropout:
(void) fp_drop(p, fd, fp, 1);
proc_fdunlock(p);
if (fpg) {
zfree(fp_guard_zone, fpg);
}
return error;
}
/*
* user_ssize_t guarded_write_np(int fd, const guardid_t *guard,
* user_addr_t cbuf, user_ssize_t nbyte);
*
* Initial implementation of guarded writes.
*/
int
guarded_write_np(struct proc *p, struct guarded_write_np_args *uap, user_ssize_t *retval)
{
int error;
guardid_t uguard;
AUDIT_ARG(fd, uap->fd);
if ((error = copyin(uap->guard, &uguard, sizeof(uguard))) != 0) {
return error;
}
return write_internal(p, uap->fd, uap->cbuf, uap->nbyte, 0, 0, &uguard, retval);
}
/*
* user_ssize_t guarded_pwrite_np(int fd, const guardid_t *guard,
* user_addr_t buf, user_size_t nbyte, off_t offset);
*
* Initial implementation of guarded pwrites.
*/
int
guarded_pwrite_np(struct proc *p, struct guarded_pwrite_np_args *uap, user_ssize_t *retval)
{
int error;
guardid_t uguard;
AUDIT_ARG(fd, uap->fd);
if ((error = copyin(uap->guard, &uguard, sizeof(uguard))) != 0) {
return error;
}
KERNEL_DEBUG_CONSTANT((BSDDBG_CODE(DBG_BSD_SC_EXTENDED_INFO, SYS_guarded_pwrite_np) | DBG_FUNC_NONE),
uap->fd, uap->nbyte, (unsigned int)((uap->offset >> 32)), (unsigned int)(uap->offset), 0);
return write_internal(p, uap->fd, uap->buf, uap->nbyte, uap->offset, FOF_OFFSET,
&uguard, retval);
}
/*
* user_ssize_t guarded_writev_np(int fd, const guardid_t *guard,
* struct iovec *iovp, u_int iovcnt);
*
* Initial implementation of guarded writev.
*
*/
int
guarded_writev_np(struct proc *p, struct guarded_writev_np_args *uap, user_ssize_t *retval)
{
int error;
guardid_t uguard;
AUDIT_ARG(fd, uap->fd);
if ((error = copyin(uap->guard, &uguard, sizeof(uguard))) != 0) {
return error;
}
return writev_uio(p, uap->fd, uap->iovp, uap->iovcnt, 0, 0, &uguard, retval);
}
/*
* int falloc_guarded(struct proc *p, struct fileproc **fp, int *fd,
* vfs_context_t ctx, const guardid_t *guard, u_int attrs);
*
* This SPI is the guarded variant of falloc(). It borrows the same
* restrictions as those used by the rest of the guarded_* routines.
*/
int
falloc_guarded(struct proc *p, struct fileproc **fp, int *fd,
vfs_context_t ctx, const guardid_t *guard, u_int attrs)
{
kauth_cred_t p_cred = current_cached_proc_cred(p);
struct gfp_crarg crarg;
if (((attrs & GUARD_REQUIRED) != GUARD_REQUIRED) ||
((attrs & ~GUARD_ALL) != 0) || (*guard == 0)) {
return EINVAL;
}
bzero(&crarg, sizeof(crarg));
crarg.gca_guard = *guard;
crarg.gca_attrs = (uint16_t)attrs;
return falloc_withinit(p, p_cred, ctx, fp, fd, guarded_fileproc_init, &crarg);
}
#if CONFIG_MACF && CONFIG_VNGUARD
/*
* Guarded vnodes
*
* Uses MAC hooks to guard operations on vnodes in the system. Given an fd,
* add data to the label on the fileglob and the vnode it points at.
* The data contains a pointer to the fileglob, the set of attributes to
* guard, a guard value for uniquification, and the pid of the process
* who set the guard up in the first place.
*
* The fd must have been opened read/write, and the underlying
* fileglob is FG_CONFINED so that there's no ambiguity about the
* owning process.
*
* When there's a callback for a vnode operation of interest (rename, unlink,
* etc.) check to see if the guard permits that operation, and if not
* take an action e.g. log a message or generate a crash report.
*
* The label is removed from the vnode and the fileglob when the fileglob
* is closed.
*
* The initial action to be taken can be specified by a boot arg (vnguard=0x42)
* and change via the "kern.vnguard.flags" sysctl.
*/
struct vng_owner;
struct vng_info { /* lives on the vnode label */
guardid_t vgi_guard;
unsigned vgi_attrs;
TAILQ_HEAD(, vng_owner) vgi_owners;
};
struct vng_owner { /* lives on the fileglob label */
proc_t vgo_p;
struct vng_info *vgo_vgi;
TAILQ_ENTRY(vng_owner) vgo_link;
};
static struct vng_info *
new_vgi(unsigned attrs, guardid_t guard)
{
struct vng_info *vgi = kalloc_type(struct vng_info, Z_WAITOK);
vgi->vgi_guard = guard;
vgi->vgi_attrs = attrs;
TAILQ_INIT(&vgi->vgi_owners);
return vgi;
}
static struct vng_owner *
new_vgo(proc_t p)
{
struct vng_owner *vgo = kalloc_type(struct vng_owner, Z_WAITOK | Z_ZERO);
vgo->vgo_p = p;
return vgo;
}
static void
vgi_add_vgo(struct vng_info *vgi, struct vng_owner *vgo)
{
vgo->vgo_vgi = vgi;
TAILQ_INSERT_HEAD(&vgi->vgi_owners, vgo, vgo_link);
}
static boolean_t
vgi_remove_vgo(struct vng_info *vgi, struct vng_owner *vgo)
{
TAILQ_REMOVE(&vgi->vgi_owners, vgo, vgo_link);
vgo->vgo_vgi = NULL;
return TAILQ_EMPTY(&vgi->vgi_owners);
}
static void
free_vgi(struct vng_info *vgi)
{
assert(TAILQ_EMPTY(&vgi->vgi_owners));
#if DEVELOP || DEBUG
memset(vgi, 0xbeadfade, sizeof(*vgi));
#endif
kfree_type(struct vng_info, vgi);
}
static void
free_vgo(struct vng_owner *vgo)
{
#if DEVELOP || DEBUG
memset(vgo, 0x2bedf1d0, sizeof(*vgo));
#endif
kfree_type(struct vng_owner, vgo);
}
static int label_slot;
static LCK_GRP_DECLARE(llock_grp, VNG_POLICY_NAME);
static LCK_RW_DECLARE(llock, &llock_grp);
static __inline void *
vng_lbl_get(struct label *label)
{
lck_rw_assert(&llock, LCK_RW_ASSERT_HELD);
void *data;
if (NULL == label) {
data = NULL;
} else {
data = (void *)mac_label_get(label, label_slot);
}
return data;
}
static __inline struct vng_info *
vng_lbl_get_withattr(struct label *label, unsigned attrmask)
{
struct vng_info *vgi = vng_lbl_get(label);
assert(NULL == vgi || (vgi->vgi_attrs & ~VNG_ALL) == 0);
if (NULL != vgi && 0 == (vgi->vgi_attrs & attrmask)) {
vgi = NULL;
}
return vgi;
}
static __inline void
vng_lbl_set(struct label *label, void *data)
{
assert(NULL != label);
lck_rw_assert(&llock, LCK_RW_ASSERT_EXCLUSIVE);
mac_label_set(label, label_slot, (intptr_t)data);
}
static int
vnguard_sysc_getguardattr(proc_t p, struct vnguard_getattr *vga)
{
const int fd = vga->vga_fd;
if (0 == vga->vga_guard) {
return EINVAL;
}
int error;
struct fileproc *fp;
if (0 != (error = fp_lookup(p, fd, &fp, 0))) {
return error;
}
do {
struct fileglob *fg = fp->fp_glob;
if (FILEGLOB_DTYPE(fg) != DTYPE_VNODE) {
error = EBADF;
break;
}
struct vnode *vp = fg_get_data(fg);
if (!vnode_isreg(vp) || NULL == vp->v_mount) {
error = EBADF;
break;
}
error = vnode_getwithref(vp);
if (0 != error) {
break;
}
vga->vga_attrs = 0;
lck_rw_lock_shared(&llock);
if (NULL != mac_vnode_label(vp)) {
const struct vng_info *vgi = vng_lbl_get(mac_vnode_label(vp));
if (NULL != vgi) {
if (vgi->vgi_guard != vga->vga_guard) {
error = EPERM;
} else {
vga->vga_attrs = vgi->vgi_attrs;
}
}
}
lck_rw_unlock_shared(&llock);
vnode_put(vp);
} while (0);
fp_drop(p, fd, fp, 0);
return error;
}
static int
vnguard_sysc_setguard(proc_t p, const struct vnguard_set *vns)
{
const int fd = vns->vns_fd;
if ((vns->vns_attrs & ~VNG_ALL) != 0 ||
0 == vns->vns_attrs || 0 == vns->vns_guard) {
return EINVAL;
}
int error;
struct fileproc *fp;
if (0 != (error = fp_lookup(p, fd, &fp, 0))) {
return error;
}
do {
/*
* To avoid trivial DoS, insist that the caller
* has read/write access to the file.
*/
if ((FREAD | FWRITE) != (fp->f_flag & (FREAD | FWRITE))) {
error = EBADF;
break;
}
struct fileglob *fg = fp->fp_glob;
if (FILEGLOB_DTYPE(fg) != DTYPE_VNODE) {
error = EBADF;
break;
}
/*
* Confinement means there's only one fd pointing at
* this fileglob, and will always be associated with
* this pid.
*/
if (0 == (FG_CONFINED & fg->fg_lflags)) {
error = EBADF;
break;
}
struct vnode *vp = fg_get_data(fg);
if (!vnode_isreg(vp) || NULL == vp->v_mount) {
error = EBADF;
break;
}
error = vnode_getwithref(vp);
if (0 != error) {
break;
}
/* Ensure the target vnode -has- a label */
struct vfs_context *ctx = vfs_context_current();
mac_vnode_label_update(ctx, vp, NULL);
struct vng_info *nvgi = new_vgi(vns->vns_attrs, vns->vns_guard);
struct vng_owner *nvgo = new_vgo(p);
lck_rw_lock_exclusive(&llock);
do {
/*
* A vnode guard is associated with one or more
* fileglobs in one or more processes.
*/
struct vng_info *vgi = vng_lbl_get(mac_vnode_label(vp));
struct vng_owner *vgo = fg->fg_vgo;
if (NULL == vgi) {
/* vnode unguarded, add the first guard */
if (NULL != vgo) {
panic("vnguard label on fileglob "
"but not vnode");
}
/* add a kusecount so we can unlabel later */
error = vnode_ref_ext(vp, O_EVTONLY, 0);
if (0 == error) {
/* add the guard */
vgi_add_vgo(nvgi, nvgo);
vng_lbl_set(mac_vnode_label(vp), nvgi);
fg->fg_vgo = nvgo;
} else {
free_vgo(nvgo);
free_vgi(nvgi);
}
} else {
/* vnode already guarded */
free_vgi(nvgi);
if (vgi->vgi_guard != vns->vns_guard) {
error = EPERM; /* guard mismatch */
} else if (vgi->vgi_attrs != vns->vns_attrs) {
/*
* Temporary workaround for older versions of SQLite:
* allow newer guard attributes to be silently cleared.
*/
const unsigned mask = ~(VNG_WRITE_OTHER | VNG_TRUNC_OTHER);
if ((vgi->vgi_attrs & mask) == (vns->vns_attrs & mask)) {
vgi->vgi_attrs &= vns->vns_attrs;
} else {
error = EACCES; /* attr mismatch */
}
}
if (0 != error || NULL != vgo) {
free_vgo(nvgo);
break;
}
/* record shared ownership */
vgi_add_vgo(vgi, nvgo);
fg->fg_vgo = nvgo;
}
} while (0);
lck_rw_unlock_exclusive(&llock);
vnode_put(vp);
} while (0);
fp_drop(p, fd, fp, 0);
return error;
}
static int
vng_policy_syscall(proc_t p, int cmd, user_addr_t arg)
{
int error = EINVAL;
switch (cmd) {
case VNG_SYSC_PING:
if (0 == arg) {
error = 0;
}
break;
case VNG_SYSC_SET_GUARD: {
struct vnguard_set vns;
error = copyin(arg, (void *)&vns, sizeof(vns));
if (error) {
break;
}
error = vnguard_sysc_setguard(p, &vns);
break;
}
case VNG_SYSC_GET_ATTR: {
struct vnguard_getattr vga;
error = copyin(arg, (void *)&vga, sizeof(vga));
if (error) {
break;
}
error = vnguard_sysc_getguardattr(p, &vga);
if (error) {
break;
}
error = copyout((void *)&vga, arg, sizeof(vga));
break;
}
default:
break;
}
return error;
}
/*
* This is called just before the fileglob disappears in fg_free().
* Take the exclusive lock: no other thread can add or remove
* a vng_info to any vnode in the system.
*/
void
vng_file_label_destroy(struct fileglob *fg)
{
struct vng_owner *lvgo = fg->fg_vgo;
struct vng_info *vgi = NULL;
if (lvgo) {
lck_rw_lock_exclusive(&llock);
fg->fg_vgo = NULL;
vgi = lvgo->vgo_vgi;
assert(vgi);
if (vgi_remove_vgo(vgi, lvgo)) {
/* that was the last reference */
vgi->vgi_attrs = 0;
if (DTYPE_VNODE == FILEGLOB_DTYPE(fg)) {
struct vnode *vp = fg_get_data(fg);
int error = vnode_getwithref(vp);
if (0 == error) {
vng_lbl_set(mac_vnode_label(vp), 0);
lck_rw_unlock_exclusive(&llock);
/* may trigger VNOP_INACTIVE */
vnode_rele_ext(vp, O_EVTONLY, 0);
vnode_put(vp);
free_vgi(vgi);
free_vgo(lvgo);
return;
}
}
}
lck_rw_unlock_exclusive(&llock);
free_vgo(lvgo);
}
}
static os_reason_t
vng_reason_from_pathname(const char *path, uint32_t pathlen)
{
os_reason_t r = os_reason_create(OS_REASON_GUARD, GUARD_REASON_VNODE);
if (NULL == r) {
return r;
}
/*
* If the pathname is very long, just keep the trailing part
*/
const uint32_t pathmax = 3 * EXIT_REASON_USER_DESC_MAX_LEN / 4;
if (pathlen > pathmax) {
path += (pathlen - pathmax);
pathlen = pathmax;
}
uint32_t rsize = kcdata_estimate_required_buffer_size(1, pathlen);
if (0 == os_reason_alloc_buffer(r, rsize)) {
struct kcdata_descriptor *kcd = &r->osr_kcd_descriptor;
mach_vm_address_t addr;
if (kcdata_get_memory_addr(kcd,
EXIT_REASON_USER_DESC, pathlen, &addr) == KERN_SUCCESS) {
kcdata_memcpy(kcd, addr, path, pathlen);
return r;
}
}
os_reason_free(r);
return OS_REASON_NULL;
}
static int vng_policy_flags;
/*
* Note: if an EXC_GUARD is generated, llock will be dropped and
* subsequently reacquired by this routine. Data derived from
* any label in the caller should be regenerated.
*/
static int
vng_guard_violation(const struct vng_info *vgi,
unsigned opval, vnode_t vp)
{
int retval = 0;
if (vng_policy_flags & kVNG_POLICY_EPERM) {
/* deny the operation */
retval = EPERM;
}
if (vng_policy_flags & (kVNG_POLICY_LOGMSG | kVNG_POLICY_UPRINTMSG)) {
/* log a message */
const char *op;
switch (opval) {
case VNG_RENAME_FROM:
op = "rename-from";
break;
case VNG_RENAME_TO:
op = "rename-to";
break;
case VNG_UNLINK:
op = "unlink";
break;
case VNG_LINK:
op = "link";
break;
case VNG_EXCHDATA:
op = "exchdata";
break;
case VNG_WRITE_OTHER:
op = "write";
break;
case VNG_TRUNC_OTHER:
op = "truncate";
break;
default:
op = "(unknown)";
break;
}
const char *nm = vnode_getname(vp);
proc_t p = current_proc();
const struct vng_owner *vgo;
TAILQ_FOREACH(vgo, &vgi->vgi_owners, vgo_link) {
const char fmt[] =
"%s[%d]: %s%s: '%s' guarded by %s[%d] (0x%llx)\n";
if (vng_policy_flags & kVNG_POLICY_LOGMSG) {
printf(fmt,
proc_name_address(p), proc_pid(p), op,
0 != retval ? " denied" : "",
NULL != nm ? nm : "(unknown)",
proc_name_address(vgo->vgo_p),
proc_pid(vgo->vgo_p), vgi->vgi_guard);
}
if (vng_policy_flags & kVNG_POLICY_UPRINTMSG) {
uprintf(fmt,
proc_name_address(p), proc_pid(p), op,
0 != retval ? " denied" : "",
NULL != nm ? nm : "(unknown)",
proc_name_address(vgo->vgo_p),
proc_pid(vgo->vgo_p), vgi->vgi_guard);
}
}
if (NULL != nm) {
vnode_putname(nm);
}
}
if (vng_policy_flags & (kVNG_POLICY_EXC | kVNG_POLICY_EXC_CORPSE)) {
/* EXC_GUARD exception */
const struct vng_owner *vgo = TAILQ_FIRST(&vgi->vgi_owners);
pid_t pid = vgo ? proc_pid(vgo->vgo_p) : 0;
mach_exception_code_t code;
mach_exception_subcode_t subcode;
code = 0;
EXC_GUARD_ENCODE_TYPE(code, GUARD_TYPE_VN);
EXC_GUARD_ENCODE_FLAVOR(code, opval);
EXC_GUARD_ENCODE_TARGET(code, pid);
subcode = vgi->vgi_guard;
lck_rw_unlock_shared(&llock);
if (vng_policy_flags & kVNG_POLICY_EXC_CORPSE) {
char *path;
int len = MAXPATHLEN;
path = zalloc_flags(ZV_NAMEI, Z_WAITOK | Z_NOFAIL);
os_reason_t r = NULL;
vn_getpath(vp, path, &len);
if (*path && len) {
r = vng_reason_from_pathname(path, len);
}
task_violated_guard(code, subcode, r, TRUE); /* not fatal */
if (NULL != r) {
os_reason_free(r);
}
zfree(ZV_NAMEI, path);
} else {
thread_t t = current_thread();
thread_guard_violation(t, code, subcode, TRUE);
}
lck_rw_lock_shared(&llock);
} else if (vng_policy_flags & kVNG_POLICY_SIGKILL) {
proc_t p = current_proc();
psignal(p, SIGKILL);
}
return retval;
}
/*
* A fatal vnode guard was tripped on this thread.
*
* (Invoked before returning to userland from the syscall handler.)
*/
void
vn_guard_ast(thread_t __unused t,
mach_exception_data_type_t code, mach_exception_data_type_t subcode)
{
const bool fatal = true;
/*
* Check if anyone has registered for Synchronous EXC_GUARD, if yes then,
* deliver it synchronously and then kill the process, else kill the process
* and deliver the exception via EXC_CORPSE_NOTIFY. Always kill the process if we are not in dev mode.
*/
if (task_exception_notify(EXC_GUARD, code, subcode, fatal) == KERN_SUCCESS) {
psignal(current_proc(), SIGKILL);
} else {
exit_with_guard_exception(current_proc(), code, subcode);
}
}
/*
* vnode callbacks
*/
static int
vng_vnode_check_rename(kauth_cred_t __unused cred,
struct vnode *__unused dvp, struct label *__unused dlabel,
struct vnode *vp, struct label *label,
struct componentname *__unused cnp,
struct vnode *__unused tdvp, struct label *__unused tdlabel,
struct vnode *tvp, struct label *tlabel,
struct componentname *__unused tcnp)
{
int error = 0;
if (NULL != label || NULL != tlabel) {
lck_rw_lock_shared(&llock);
const struct vng_info *vgi =
vng_lbl_get_withattr(label, VNG_RENAME_FROM);
if (NULL != vgi) {
error = vng_guard_violation(vgi, VNG_RENAME_FROM, vp);
}
if (0 == error) {
vgi = vng_lbl_get_withattr(tlabel, VNG_RENAME_TO);
if (NULL != vgi) {
error = vng_guard_violation(vgi,
VNG_RENAME_TO, tvp);
}
}
lck_rw_unlock_shared(&llock);
}
return error;
}
static int
vng_vnode_check_link(kauth_cred_t __unused cred,
struct vnode *__unused dvp, struct label *__unused dlabel,
struct vnode *vp, struct label *label, struct componentname *__unused cnp)
{
int error = 0;
if (NULL != label) {
lck_rw_lock_shared(&llock);
const struct vng_info *vgi =
vng_lbl_get_withattr(label, VNG_LINK);
if (vgi) {
error = vng_guard_violation(vgi, VNG_LINK, vp);
}
lck_rw_unlock_shared(&llock);
}
return error;
}
static int
vng_vnode_check_unlink(kauth_cred_t __unused cred,
struct vnode *__unused dvp, struct label *__unused dlabel,
struct vnode *vp, struct label *label, struct componentname *__unused cnp)
{
int error = 0;
if (NULL != label) {
lck_rw_lock_shared(&llock);
const struct vng_info *vgi =
vng_lbl_get_withattr(label, VNG_UNLINK);
if (vgi) {
error = vng_guard_violation(vgi, VNG_UNLINK, vp);
}
lck_rw_unlock_shared(&llock);
}
return error;
}
/*
* Only check violations for writes performed by "other processes"
*/
static int
vng_vnode_check_write(kauth_cred_t __unused actv_cred,
kauth_cred_t __unused file_cred, struct vnode *vp, struct label *label)
{
int error = 0;
if (NULL != label) {
lck_rw_lock_shared(&llock);
const struct vng_info *vgi =
vng_lbl_get_withattr(label, VNG_WRITE_OTHER);
if (vgi) {
proc_t p = current_proc();
const struct vng_owner *vgo;
TAILQ_FOREACH(vgo, &vgi->vgi_owners, vgo_link) {
if (vgo->vgo_p == p) {
goto done;
}
}
error = vng_guard_violation(vgi, VNG_WRITE_OTHER, vp);
}
done:
lck_rw_unlock_shared(&llock);
}
return error;
}
/*
* Only check violations for truncates performed by "other processes"
*/
static int
vng_vnode_check_truncate(kauth_cred_t __unused actv_cred,
kauth_cred_t __unused file_cred, struct vnode *vp,
struct label *label)
{
int error = 0;
if (NULL != label) {
lck_rw_lock_shared(&llock);
const struct vng_info *vgi =
vng_lbl_get_withattr(label, VNG_TRUNC_OTHER);
if (vgi) {
proc_t p = current_proc();
const struct vng_owner *vgo;
TAILQ_FOREACH(vgo, &vgi->vgi_owners, vgo_link) {
if (vgo->vgo_p == p) {
goto done;
}
}
error = vng_guard_violation(vgi, VNG_TRUNC_OTHER, vp);
}
done:
lck_rw_unlock_shared(&llock);
}
return error;
}
static int
vng_vnode_check_exchangedata(kauth_cred_t __unused cred,
struct vnode *fvp, struct label *flabel,
struct vnode *svp, struct label *slabel)
{
int error = 0;
if (NULL != flabel || NULL != slabel) {
lck_rw_lock_shared(&llock);
const struct vng_info *vgi =
vng_lbl_get_withattr(flabel, VNG_EXCHDATA);
if (NULL != vgi) {
error = vng_guard_violation(vgi, VNG_EXCHDATA, fvp);
}
if (0 == error) {
vgi = vng_lbl_get_withattr(slabel, VNG_EXCHDATA);
if (NULL != vgi) {
error = vng_guard_violation(vgi,
VNG_EXCHDATA, svp);
}
}
lck_rw_unlock_shared(&llock);
}
return error;
}
/* Intercept open-time truncations (by "other") of a guarded vnode */
static int
vng_vnode_check_open(kauth_cred_t cred,
struct vnode *vp, struct label *label, int acc_mode)
{
if (0 == (acc_mode & O_TRUNC)) {
return 0;
}
return vng_vnode_check_truncate(cred, NULL, vp, label);
}
/*
* Configuration gorp
*/
SECURITY_READ_ONLY_EARLY(static struct mac_policy_ops) vng_policy_ops = {
.mpo_vnode_check_link = vng_vnode_check_link,
.mpo_vnode_check_unlink = vng_vnode_check_unlink,
.mpo_vnode_check_rename = vng_vnode_check_rename,
.mpo_vnode_check_write = vng_vnode_check_write,
.mpo_vnode_check_truncate = vng_vnode_check_truncate,
.mpo_vnode_check_exchangedata = vng_vnode_check_exchangedata,
.mpo_vnode_check_open = vng_vnode_check_open,
.mpo_policy_syscall = vng_policy_syscall,
};
static const char *vng_labelnames[] = {
"vnguard",
};
#define ACOUNT(arr) ((unsigned)(sizeof (arr) / sizeof (arr[0])))
SECURITY_READ_ONLY_LATE(static struct mac_policy_conf) vng_policy_conf = {
.mpc_name = VNG_POLICY_NAME,
.mpc_fullname = "Guarded vnode policy",
.mpc_field_off = &label_slot,
.mpc_labelnames = vng_labelnames,
.mpc_labelname_count = ACOUNT(vng_labelnames),
.mpc_ops = &vng_policy_ops,
.mpc_loadtime_flags = 0,
.mpc_runtime_flags = 0
};
SECURITY_READ_ONLY_LATE(static mac_policy_handle_t) vng_policy_handle;
void
vnguard_policy_init(void)
{
if (0 == PE_i_can_has_debugger(NULL)) {
return;
}
vng_policy_flags = kVNG_POLICY_LOGMSG |
kVNG_POLICY_EXC_CORPSE | kVNG_POLICY_UPRINTMSG;
PE_parse_boot_argn("vnguard", &vng_policy_flags, sizeof(vng_policy_flags));
if (vng_policy_flags) {
mac_policy_register(&vng_policy_conf, &vng_policy_handle, NULL);
}
}
#if DEBUG || DEVELOPMENT
#include <sys/sysctl.h>
SYSCTL_DECL(_kern_vnguard);
SYSCTL_NODE(_kern, OID_AUTO, vnguard, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "vnguard");
SYSCTL_INT(_kern_vnguard, OID_AUTO, flags, CTLFLAG_RW | CTLFLAG_LOCKED,
&vng_policy_flags, 0, "vnguard policy flags");
#endif
#endif /* CONFIG_MACF && CONFIG_VNGUARD */