xnu-6153.41.3.tar.gz

[apple/xnu.git] / bsd / kern / kern_descrip.c

diff --git a/bsd/kern/kern_descrip.c b/bsd/kern/kern_descrip.c
index 7f53765a201a13a0c62f04987056bec4bd2f295d..320c27b2c1be8fe18547a807098907c6df8f2038 100644 (file)
--- a/bsd/kern/kern_descrip.c
+++ b/bsd/kern/kern_descrip.c
@@ -1,8 +1,8 @@
 /*
 /*

- * Copyright (c) 2000-2013 Apple Inc. All rights reserved.
+ * Copyright (c) 2000-2016 Apple Inc. All rights reserved.

  *
  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
  *
  * @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
  * 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


@@ -11,10 +11,10 @@
  * 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.
  * 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.
  * 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,
  * 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,


@@ -22,7 +22,7 @@
  * 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.
  * 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@
  */
 /* Copyright (c) 1995, 1997 Apple Computer, Inc. All Rights Reserved */
  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  */
 /* Copyright (c) 1995, 1997 Apple Computer, Inc. All Rights Reserved */


@@ -81,11 +81,13 @@
 #include <sys/kauth.h>
 #include <sys/file_internal.h>
 #include <sys/guarded.h>
 #include <sys/kauth.h>
 #include <sys/file_internal.h>
 #include <sys/guarded.h>

+#include <sys/priv.h>

 #include <sys/socket.h>
 #include <sys/socketvar.h>
 #include <sys/stat.h>
 #include <sys/ioctl.h>
 #include <sys/fcntl.h>
 #include <sys/socket.h>
 #include <sys/socketvar.h>
 #include <sys/stat.h>
 #include <sys/ioctl.h>
 #include <sys/fcntl.h>

+#include <sys/fsctl.h>

 #include <sys/malloc.h>
 #include <sys/mman.h>
 #include <sys/syslog.h>
 #include <sys/malloc.h>
 #include <sys/mman.h>
 #include <sys/syslog.h>


@@ -93,9 +95,10 @@
 #include <sys/resourcevar.h>
 #include <sys/aio_kern.h>
 #include <sys/ev.h>
 #include <sys/resourcevar.h>
 #include <sys/aio_kern.h>
 #include <sys/ev.h>

-#include <kern/lock.h>
+#include <kern/locks.h>

 #include <sys/uio_internal.h>
 #include <sys/codesign.h>
 #include <sys/uio_internal.h>
 #include <sys/codesign.h>

+#include <sys/codedir_internal.h>

 
 #include <security/audit/audit.h>
 
 
 #include <security/audit/audit.h>
 


@@ -104,8 +107,10 @@
 #include <sys/sysproto.h>
 #include <sys/pipe.h>
 #include <sys/spawn.h>
 #include <sys/sysproto.h>
 #include <sys/pipe.h>
 #include <sys/spawn.h>

+#include <sys/cprotect.h>

 #include <kern/kern_types.h>
 #include <kern/kalloc.h>
 #include <kern/kern_types.h>
 #include <kern/kalloc.h>

+#include <kern/waitq.h>

 #include <libkern/OSAtomic.h>
 
 #include <sys/ubc_internal.h>
 #include <libkern/OSAtomic.h>
 
 #include <sys/ubc_internal.h>


@@ -114,14 +119,15 @@
 #include <vm/vm_protos.h>
 
 #include <mach/mach_port.h>
 #include <vm/vm_protos.h>
 
 #include <mach/mach_port.h>

+#include <stdbool.h>

 
 

-#if CONFIG_PROTECT
-#include <sys/cprotect.h>
+#if CONFIG_MACF
+#include <security/mac_framework.h>

 #endif
 #endif

-#include <hfs/hfs.h>

 
 

+#define IPC_KMSG_FLAGS_ALLOW_IMMOVABLE_SEND 0x1

 kern_return_t ipc_object_copyin(ipc_space_t, mach_port_name_t,
 kern_return_t ipc_object_copyin(ipc_space_t, mach_port_name_t,

-    mach_msg_type_name_t, ipc_port_t *);
+    mach_msg_type_name_t, ipc_port_t *, mach_port_context_t, mach_msg_guard_flags_t *, uint32_t);

 void ipc_port_release_send(ipc_port_t);
 
 struct psemnode;
 void ipc_port_release_send(ipc_port_t);
 
 struct psemnode;


@@ -142,18 +148,29 @@ void fileport_releasefg(struct fileglob *fg);
 /* We don't want these exported */
 
 __private_extern__
 /* We don't want these exported */
 
 __private_extern__

-int unlink1(vfs_context_t, struct nameidata *, int);
+int unlink1(vfs_context_t, vnode_t, user_addr_t, enum uio_seg, int);

 
 static void _fdrelse(struct proc * p, int fd);
 
 
 extern void file_lock_init(void);
 
 static void _fdrelse(struct proc * p, int fd);
 
 
 extern void file_lock_init(void);

-extern int kqueue_stat(struct fileproc *fp, void *ub, int isstat4, proc_t p);

 
 

-extern kauth_scope_t   kauth_scope_fileop;
+extern kauth_scope_t    kauth_scope_fileop;

 
 /* Conflict wait queue for when selects collide (opaque type) */
 
 /* Conflict wait queue for when selects collide (opaque type) */

-extern struct wait_queue select_conflict_queue;
+extern struct waitq select_conflict_queue;
+
+#ifndef HFS_GET_BOOT_INFO
+#define HFS_GET_BOOT_INFO   (FCNTL_FS_SPECIFIC_BASE + 0x00004)
+#endif
+
+#ifndef HFS_SET_BOOT_INFO
+#define HFS_SET_BOOT_INFO   (FCNTL_FS_SPECIFIC_BASE + 0x00005)
+#endif
+
+#ifndef APFSIOC_REVERT_TO_SNAPSHOT
+#define APFSIOC_REVERT_TO_SNAPSHOT  _IOW('J', 1, u_int64_t)
+#endif

 
 #define f_flag f_fglob->fg_flag
 #define f_type f_fglob->fg_ops->fo_type
 
 #define f_flag f_fglob->fg_flag
 #define f_type f_fglob->fg_ops->fo_type


@@ -163,15 +180,15 @@ extern struct wait_queue select_conflict_queue;
 #define f_offset f_fglob->fg_offset
 #define f_data f_fglob->fg_data
 #define CHECK_ADD_OVERFLOW_INT64L(x, y) \
 #define f_offset f_fglob->fg_offset
 #define f_data f_fglob->fg_data
 #define CHECK_ADD_OVERFLOW_INT64L(x, y) \

-               (((((x) > 0) && ((y) > 0) && ((x) > LLONG_MAX - (y))) || \
-               (((x) < 0) && ((y) < 0) && ((x) < LLONG_MIN - (y)))) \
-               ? 1 : 0)
+               (((((x) > 0) && ((y) > 0) && ((x) > LLONG_MAX - (y))) || \
+               (((x) < 0) && ((y) < 0) && ((x) < LLONG_MIN - (y)))) \
+               ? 1 : 0)

 /*
  * Descriptor management.
  */
 /*
  * Descriptor management.
  */

-struct fmsglist fmsghead;      /* head of list of open files */
-struct fmsglist fmsg_ithead;   /* head of list of open files */
-int nfiles;                    /* actual number of open files */
+struct fmsglist fmsghead;       /* head of list of open files */
+struct fmsglist fmsg_ithead;    /* head of list of open files */
+int nfiles;                     /* actual number of open files */

 
 
 lck_grp_attr_t * file_lck_grp_attr;
 
 
 lck_grp_attr_t * file_lck_grp_attr;


@@ -189,9 +206,9 @@ lck_mtx_t * uipc_lock;
  * Parameters:  fl             Flock structure.
  *             cur_file_offset Current offset in the file.
  *
  * Parameters:  fl             Flock structure.
  *             cur_file_offset Current offset in the file.
  *

- * Returns:    0               on Success.
+ * Returns:     0               on Success.

  *             EOVERFLOW       on overflow.
  *             EOVERFLOW       on overflow.

- *             EINVAL          on offset less than zero.
+ *             EINVAL          on offset less than zero.

  */
 
 static int
  */
 
 static int


@@ -211,13 +228,13 @@ check_file_seek_range(struct flock *fl, off_t cur_file_offset)
                        return EINVAL;
                }
                /* Check if end marker is beyond LLONG_MAX. */
                        return EINVAL;
                }
                /* Check if end marker is beyond LLONG_MAX. */

-               if ((fl->l_len > 0) && (CHECK_ADD_OVERFLOW_INT64L(fl->l_start + 
-                       cur_file_offset, fl->l_len - 1))) {
+               if ((fl->l_len > 0) && (CHECK_ADD_OVERFLOW_INT64L(fl->l_start +
+                   cur_file_offset, fl->l_len - 1))) {

                        return EOVERFLOW;
                }
                /* Check if the end marker is negative. */
                if ((fl->l_len <= 0) && (fl->l_start + cur_file_offset +
                        return EOVERFLOW;
                }
                /* Check if the end marker is negative. */
                if ((fl->l_len <= 0) && (fl->l_start + cur_file_offset +

-                       fl->l_len < 0)) {
+                   fl->l_len < 0)) {

                        return EINVAL;
                }
        } else if (fl->l_whence == SEEK_SET) {
                        return EINVAL;
                }
        } else if (fl->l_whence == SEEK_SET) {


@@ -226,12 +243,12 @@ check_file_seek_range(struct flock *fl, off_t cur_file_offset)
                        return EINVAL;
                }
                /* Check if the end marker is beyond LLONG_MAX. */
                        return EINVAL;
                }
                /* Check if the end marker is beyond LLONG_MAX. */

-               if ((fl->l_len > 0) && 
+               if ((fl->l_len > 0) &&

                    CHECK_ADD_OVERFLOW_INT64L(fl->l_start, fl->l_len - 1)) {
                        return EOVERFLOW;
                }
                /* Check if the end marker is negative. */
                    CHECK_ADD_OVERFLOW_INT64L(fl->l_start, fl->l_len - 1)) {
                        return EOVERFLOW;
                }
                /* Check if the end marker is negative. */

-               if ((fl->l_len < 0) &&  fl->l_start + fl->l_len < 0) {
+               if ((fl->l_len < 0) && fl->l_start + fl->l_len < 0) {

                        return EINVAL;
                }
        }
                        return EINVAL;
                }
        }


@@ -254,9 +271,9 @@ void
 file_lock_init(void)
 {
        /* allocate file lock group attribute and group */
 file_lock_init(void)
 {
        /* allocate file lock group attribute and group */

-       file_lck_grp_attr= lck_grp_attr_alloc_init();
+       file_lck_grp_attr = lck_grp_attr_alloc_init();

 
 

-       file_lck_grp = lck_grp_alloc_init("file",  file_lck_grp_attr);
+       file_lck_grp = lck_grp_alloc_init("file", file_lck_grp_attr);

 
        /* Allocate file lock attribute */
        file_lck_attr = lck_attr_alloc_init();
 
        /* Allocate file lock attribute */
        file_lck_attr = lck_attr_alloc_init();


@@ -338,7 +355,7 @@ getdtablesize(proc_t p, __unused struct getdtablesize_args *uap, int32_t *retval
        *retval = min((int)p->p_rlimit[RLIMIT_NOFILE].rlim_cur, maxfiles);
        proc_fdunlock(p);
 
        *retval = min((int)p->p_rlimit[RLIMIT_NOFILE].rlim_cur, maxfiles);
        proc_fdunlock(p);
 

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -346,31 +363,32 @@ void
 procfdtbl_reservefd(struct proc * p, int fd)
 {
        p->p_fd->fd_ofiles[fd] = NULL;
 procfdtbl_reservefd(struct proc * p, int fd)
 {
        p->p_fd->fd_ofiles[fd] = NULL;

-        p->p_fd->fd_ofileflags[fd] |= UF_RESERVED;
+       p->p_fd->fd_ofileflags[fd] |= UF_RESERVED;

 }
 
 void
 procfdtbl_markclosefd(struct proc * p, int fd)
 {
 }
 
 void
 procfdtbl_markclosefd(struct proc * p, int fd)
 {

-        p->p_fd->fd_ofileflags[fd] |= (UF_RESERVED | UF_CLOSING);
+       p->p_fd->fd_ofileflags[fd] |= (UF_RESERVED | UF_CLOSING);

 }
 
 void
 procfdtbl_releasefd(struct proc * p, int fd, struct fileproc * fp)
 {
 }
 
 void
 procfdtbl_releasefd(struct proc * p, int fd, struct fileproc * fp)
 {

-       if (fp != NULL)
-               p->p_fd->fd_ofiles[fd] = fp;
-        p->p_fd->fd_ofileflags[fd] &= ~UF_RESERVED;
+       if (fp != NULL) {
+               p->p_fd->fd_ofiles[fd] = fp;
+       }
+       p->p_fd->fd_ofileflags[fd] &= ~UF_RESERVED;

        if ((p->p_fd->fd_ofileflags[fd] & UF_RESVWAIT) == UF_RESVWAIT) {
                p->p_fd->fd_ofileflags[fd] &= ~UF_RESVWAIT;
                wakeup(&p->p_fd);
        }
 }
 
        if ((p->p_fd->fd_ofileflags[fd] & UF_RESVWAIT) == UF_RESVWAIT) {
                p->p_fd->fd_ofileflags[fd] &= ~UF_RESVWAIT;
                wakeup(&p->p_fd);
        }
 }
 

-void 
+void

 procfdtbl_waitfd(struct proc * p, int fd)
 {
 procfdtbl_waitfd(struct proc * p, int fd)
 {

-        p->p_fd->fd_ofileflags[fd] |= UF_RESVWAIT;
+       p->p_fd->fd_ofileflags[fd] |= UF_RESVWAIT;

        msleep(&p->p_fd, &p->p_fdmlock, PRIBIO, "ftbl_waitfd", NULL);
 }
 
        msleep(&p->p_fd, &p->p_fdmlock, PRIBIO, "ftbl_waitfd", NULL);
 }
 


@@ -381,9 +399,9 @@ procfdtbl_clearfd(struct proc * p, int fd)
        int waiting;
 
        waiting = (p->p_fd->fd_ofileflags[fd] & UF_RESVWAIT);
        int waiting;
 
        waiting = (p->p_fd->fd_ofileflags[fd] & UF_RESVWAIT);

-       p->p_fd->fd_ofiles[fd] = NULL;       
+       p->p_fd->fd_ofiles[fd] = NULL;

        p->p_fd->fd_ofileflags[fd] = 0;
        p->p_fd->fd_ofileflags[fd] = 0;

-       if ( waiting == UF_RESVWAIT) {
+       if (waiting == UF_RESVWAIT) {

                wakeup(&p->p_fd);
        }
 }
                wakeup(&p->p_fd);
        }
 }


@@ -408,18 +426,22 @@ _fdrelse(struct proc * p, int fd)
        struct filedesc *fdp = p->p_fd;
        int nfd = 0;
 
        struct filedesc *fdp = p->p_fd;
        int nfd = 0;
 

-       if (fd < fdp->fd_freefile)
+       if (fd < fdp->fd_freefile) {

                fdp->fd_freefile = fd;
                fdp->fd_freefile = fd;

+       }

 #if DIAGNOSTIC
 #if DIAGNOSTIC

-       if (fd > fdp->fd_lastfile)
-               panic("fdrelse: fd_lastfile inconsistent");
+       if (fd > fdp->fd_lastfile) {
+               panic("fdrelse: fd_lastfile inconsistent");
+       }

 #endif
        procfdtbl_clearfd(p, fd);
 
        while ((nfd = fdp->fd_lastfile) > 0 &&
 #endif
        procfdtbl_clearfd(p, fd);
 
        while ((nfd = fdp->fd_lastfile) > 0 &&

-                       fdp->fd_ofiles[nfd] == NULL &&
-                       !(fdp->fd_ofileflags[nfd] & UF_RESERVED))
+           fdp->fd_ofiles[nfd] == NULL &&
+           !(fdp->fd_ofileflags[nfd] & UF_RESERVED)) {
+               /* JMM - What about files with lingering EV_VANISHED knotes? */

                fdp->fd_lastfile--;
                fdp->fd_lastfile--;

+       }

 }
 
 
 }
 
 


@@ -430,69 +452,78 @@ fd_rdwr(
        uint64_t base,
        int64_t len,
        enum uio_seg segflg,
        uint64_t base,
        int64_t len,
        enum uio_seg segflg,

-       off_t   offset,
-       int     io_flg,
+       off_t   offset,
+       int     io_flg,

        int64_t *aresid)
 {
        int64_t *aresid)
 {

-        struct fileproc *fp;
-       proc_t  p;
-        int error = 0;
+       struct fileproc *fp;
+       proc_t  p;
+       int error = 0;

        int flags = 0;
        int spacetype;
        uio_t auio = NULL;
        int flags = 0;
        int spacetype;
        uio_t auio = NULL;

-       char uio_buf[ UIO_SIZEOF(1) ];
+       char uio_buf[UIO_SIZEOF(1)];

        struct vfs_context context = *(vfs_context_current());
        struct vfs_context context = *(vfs_context_current());

+       bool wrote_some = false;

 
        p = current_proc();
 
 
        p = current_proc();
 

-        error = fp_lookup(p, fd, &fp, 0);
-        if (error)
-                return(error);
+       error = fp_lookup(p, fd, &fp, 0);
+       if (error) {
+               return error;
+       }

 
        if (fp->f_type != DTYPE_VNODE && fp->f_type != DTYPE_PIPE && fp->f_type != DTYPE_SOCKET) {
                error = EINVAL;
                goto out;
        }
        if (rw == UIO_WRITE && !(fp->f_flag & FWRITE)) {
 
        if (fp->f_type != DTYPE_VNODE && fp->f_type != DTYPE_PIPE && fp->f_type != DTYPE_SOCKET) {
                error = EINVAL;
                goto out;
        }
        if (rw == UIO_WRITE && !(fp->f_flag & FWRITE)) {

-                error = EBADF;
+               error = EBADF;

                goto out;
        }
                goto out;
        }

-       
+

        if (rw == UIO_READ && !(fp->f_flag & FREAD)) {
        if (rw == UIO_READ && !(fp->f_flag & FREAD)) {

-               error = EBADF;
-               goto out;
+               error = EBADF;
+               goto out;

        }
        }

-       
+

        context.vc_ucred = fp->f_fglob->fg_cred;
 
        context.vc_ucred = fp->f_fglob->fg_cred;
 

-       if (UIO_SEG_IS_USER_SPACE(segflg))
+       if (UIO_SEG_IS_USER_SPACE(segflg)) {

                spacetype = proc_is64bit(p) ? UIO_USERSPACE64 : UIO_USERSPACE32;
                spacetype = proc_is64bit(p) ? UIO_USERSPACE64 : UIO_USERSPACE32;

-       else
+       } else {

                spacetype = UIO_SYSSPACE;
                spacetype = UIO_SYSSPACE;

+       }

 
        auio = uio_createwithbuffer(1, offset, spacetype, rw, &uio_buf[0], sizeof(uio_buf));
 
        uio_addiov(auio, base, len);
 
 
        auio = uio_createwithbuffer(1, offset, spacetype, rw, &uio_buf[0], sizeof(uio_buf));
 
        uio_addiov(auio, base, len);
 

-       if ( !(io_flg & IO_APPEND))
+       if (!(io_flg & IO_APPEND)) {

                flags = FOF_OFFSET;
                flags = FOF_OFFSET;

+       }

 
 

-       if (rw == UIO_WRITE)
+       if (rw == UIO_WRITE) {
+               user_ssize_t orig_resid = uio_resid(auio);

                error = fo_write(fp, auio, flags, &context);
                error = fo_write(fp, auio, flags, &context);

-       else
+               wrote_some = uio_resid(auio) < orig_resid;
+       } else {

                error = fo_read(fp, auio, flags, &context);
                error = fo_read(fp, auio, flags, &context);

+       }

 
 

-       if (aresid)
+       if (aresid) {

                *aresid = uio_resid(auio);
                *aresid = uio_resid(auio);

-       else {
-               if (uio_resid(auio) && error == 0)
+       } else {
+               if (uio_resid(auio) && error == 0) {

                        error = EIO;
                        error = EIO;

+               }

        }
 out:
        }
 out:

-        if (rw == UIO_WRITE && error == 0)
-                fp_drop_written(p, fd, fp);
-        else
-                fp_drop(p, fd, fp, 0);
+       if (wrote_some) {
+               fp_drop_written(p, fd, fp);
+       } else {
+               fp_drop(p, fd, fp, 0);
+       }

 
        return error;
 }
 
        return error;
 }


@@ -523,26 +554,31 @@ dup(proc_t p, struct dup_args *uap, int32_t *retval)
        struct fileproc *fp;
 
        proc_fdlock(p);
        struct fileproc *fp;
 
        proc_fdlock(p);

-       if ( (error = fp_lookup(p, old, &fp, 1)) ) {
+       if ((error = fp_lookup(p, old, &fp, 1))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(error);
+               return error;

        }
        if (FP_ISGUARDED(fp, GUARD_DUP)) {
                error = fp_guard_exception(p, old, fp, kGUARD_EXC_DUP);
                (void) fp_drop(p, old, fp, 1);
                proc_fdunlock(p);
        }
        if (FP_ISGUARDED(fp, GUARD_DUP)) {
                error = fp_guard_exception(p, old, fp, kGUARD_EXC_DUP);
                (void) fp_drop(p, old, fp, 1);
                proc_fdunlock(p);

-               return (error);
+               return error;

        }
        }

-       if ( (error = fdalloc(p, 0, &new)) ) {
+       if ((error = fdalloc(p, 0, &new))) {

                fp_drop(p, old, fp, 1);
                proc_fdunlock(p);
                fp_drop(p, old, fp, 1);
                proc_fdunlock(p);

-               return (error);
+               return error;

        }
        error = finishdup(p, fdp, old, new, 0, retval);
        fp_drop(p, old, fp, 1);
        proc_fdunlock(p);
 
        }
        error = finishdup(p, fdp, old, new, 0, retval);
        fp_drop(p, old, fp, 1);
        proc_fdunlock(p);
 

-       return (error);
+       if (ENTR_SHOULDTRACE && fp->f_type == DTYPE_SOCKET) {
+               KERNEL_ENERGYTRACE(kEnTrActKernSocket, DBG_FUNC_START,
+                   new, 0, (int64_t)VM_KERNEL_ADDRPERM(fp->f_data));
+       }
+
+       return error;

 }
 
 /*
 }
 
 /*


@@ -572,34 +608,34 @@ dup2(proc_t p, struct dup2_args *uap, int32_t *retval)
        proc_fdlock(p);
 
 startover:
        proc_fdlock(p);
 
 startover:

-       if ( (error = fp_lookup(p, old, &fp, 1)) ) {
+       if ((error = fp_lookup(p, old, &fp, 1))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(error);
+               return error;

        }
        if (FP_ISGUARDED(fp, GUARD_DUP)) {
                error = fp_guard_exception(p, old, fp, kGUARD_EXC_DUP);
                (void) fp_drop(p, old, fp, 1);
                proc_fdunlock(p);
        }
        if (FP_ISGUARDED(fp, GUARD_DUP)) {
                error = fp_guard_exception(p, old, fp, kGUARD_EXC_DUP);
                (void) fp_drop(p, old, fp, 1);
                proc_fdunlock(p);

-               return (error);
+               return error;

        }
        if (new < 0 ||
        }
        if (new < 0 ||

-               (rlim_t)new >= p->p_rlimit[RLIMIT_NOFILE].rlim_cur ||
+           (rlim_t)new >= p->p_rlimit[RLIMIT_NOFILE].rlim_cur ||

            new >= maxfiles) {
                fp_drop(p, old, fp, 1);
                proc_fdunlock(p);
            new >= maxfiles) {
                fp_drop(p, old, fp, 1);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        if (old == new) {
                fp_drop(p, old, fp, 1);
                *retval = new;
                proc_fdunlock(p);
        }
        if (old == new) {
                fp_drop(p, old, fp, 1);
                *retval = new;
                proc_fdunlock(p);

-               return (0);
+               return 0;

        }
        if (new < 0 || new >= fdp->fd_nfiles) {
        }
        if (new < 0 || new >= fdp->fd_nfiles) {

-               if ( (error = fdalloc(p, new, &i)) ) {
+               if ((error = fdalloc(p, new, &i))) {

                        fp_drop(p, old, fp, 1);
                        proc_fdunlock(p);
                        fp_drop(p, old, fp, 1);
                        proc_fdunlock(p);

-                       return (error);
+                       return error;

                }
                if (new != i) {
                        fdrelse(p, i);
                }
                if (new != i) {
                        fdrelse(p, i);


@@ -607,13 +643,13 @@ startover:
                }
        } else {
 closeit:
                }
        } else {
 closeit:

-               while ((fdp->fd_ofileflags[new] & UF_RESERVED) == UF_RESERVED)  {
-                               fp_drop(p, old, fp, 1);
-                               procfdtbl_waitfd(p, new);
+               while ((fdp->fd_ofileflags[new] & UF_RESERVED) == UF_RESERVED) {
+                       fp_drop(p, old, fp, 1);
+                       procfdtbl_waitfd(p, new);

 #if DIAGNOSTIC
 #if DIAGNOSTIC

-                               proc_fdlock_assert(p, LCK_MTX_ASSERT_OWNED);
+                       proc_fdlock_assert(p, LCK_MTX_ASSERT_OWNED);

 #endif
 #endif

-                               goto startover;
+                       goto startover;

                }
 
                if ((fdp->fd_ofiles[new] != NULL) &&
                }
 
                if ((fdp->fd_ofiles[new] != NULL) &&


@@ -624,7 +660,7 @@ closeit:
                                    new, nfp, kGUARD_EXC_CLOSE);
                                (void) fp_drop(p, new, nfp, 1);
                                proc_fdunlock(p);
                                    new, nfp, kGUARD_EXC_CLOSE);
                                (void) fp_drop(p, new, nfp, 1);
                                proc_fdunlock(p);

-                               return (error);
+                               return error;

                        }
                        (void)close_internal_locked(p, new, nfp, FD_DUP2RESV);
 #if DIAGNOSTIC
                        }
                        (void)close_internal_locked(p, new, nfp, FD_DUP2RESV);
 #if DIAGNOSTIC


@@ -632,10 +668,11 @@ closeit:
 #endif
                        procfdtbl_clearfd(p, new);
                        goto startover;
 #endif
                        procfdtbl_clearfd(p, new);
                        goto startover;

-               } else  {
+               } else {

 #if DIAGNOSTIC
 #if DIAGNOSTIC

-                       if (fdp->fd_ofiles[new] != NULL)
+                       if (fdp->fd_ofiles[new] != NULL) {

                                panic("dup2: no ref on fileproc %d", new);
                                panic("dup2: no ref on fileproc %d", new);

+                       }

 #endif
                        procfdtbl_reservefd(p, new);
                }
 #endif
                        procfdtbl_reservefd(p, new);
                }


@@ -643,19 +680,20 @@ closeit:
 #if DIAGNOSTIC
                proc_fdlock_assert(p, LCK_MTX_ASSERT_OWNED);
 #endif
 #if DIAGNOSTIC
                proc_fdlock_assert(p, LCK_MTX_ASSERT_OWNED);
 #endif

-

        }
 #if DIAGNOSTIC
        }
 #if DIAGNOSTIC

-       if (fdp->fd_ofiles[new] != 0)
+       if (fdp->fd_ofiles[new] != 0) {

                panic("dup2: overwriting fd_ofiles with new %d", new);
                panic("dup2: overwriting fd_ofiles with new %d", new);

-       if ((fdp->fd_ofileflags[new] & UF_RESERVED) == 0)
+       }
+       if ((fdp->fd_ofileflags[new] & UF_RESERVED) == 0) {

                panic("dup2: unreserved fileflags with new %d", new);
                panic("dup2: unreserved fileflags with new %d", new);

+       }

 #endif
        error = finishdup(p, fdp, old, new, 0, retval);
        fp_drop(p, old, fp, 1);
        proc_fdunlock(p);
 
 #endif
        error = finishdup(p, fdp, old, new, 0, retval);
        fp_drop(p, old, fp, 1);
        proc_fdunlock(p);
 

-       return(error);
+       return error;

 }
 
 
 }
 
 


@@ -684,7 +722,7 @@ int
 fcntl(proc_t p, struct fcntl_args *uap, int32_t *retval)
 {
        __pthread_testcancel(1);
 fcntl(proc_t p, struct fcntl_args *uap, int32_t *retval)
 {
        __pthread_testcancel(1);

-       return(fcntl_nocancel(p, (struct fcntl_nocancel_args *)uap, retval));
+       return fcntl_nocancel(p, (struct fcntl_nocancel_args *)uap, retval);

 }
 
 
 }
 
 


@@ -749,9 +787,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
        char *pop;
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
        char *pop;

-       struct vnode *vp = NULLVP;      /* for AUDIT_ARG() at end */
-       int i, tmp, error, error2, flg = F_POSIX;
-       struct flock fl;
+       struct vnode *vp = NULLVP;      /* for AUDIT_ARG() at end */
+       int i, tmp, error, error2, flg = 0;
+       struct flock fl = {};

        struct flocktimeout fltimeout;
        struct timespec *timeout = NULL;
        struct vfs_context context;
        struct flocktimeout fltimeout;
        struct timespec *timeout = NULL;
        struct vfs_context context;


@@ -766,9 +804,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
        AUDIT_ARG(cmd, uap->cmd);
 
        proc_fdlock(p);
        AUDIT_ARG(cmd, uap->cmd);
 
        proc_fdlock(p);

-       if ( (error = fp_lookup(p, fd, &fp, 1)) ) {
+       if ((error = fp_lookup(p, fd, &fp, 1))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(error);
+               return error;

        }
        context.vc_thread = current_thread();
        context.vc_ucred = fp->f_cred;
        }
        context.vc_thread = current_thread();
        context.vc_ucred = fp->f_cred;


@@ -776,8 +814,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
        is64bit = proc_is64bit(p);
        if (is64bit) {
                argp = uap->arg;
        is64bit = proc_is64bit(p);
        if (is64bit) {
                argp = uap->arg;

-       }
-       else {
+       } else {

                /*
                 * Since the arg parameter is defined as a long but may be
                 * either a long or a pointer we must take care to handle
                /*
                 * Since the arg parameter is defined as a long but may be
                 * either a long or a pointer we must take care to handle


@@ -795,12 +832,12 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 #if CONFIG_MACF
        error = mac_file_check_fcntl(proc_ucred(p), fp->f_fglob, uap->cmd,
            uap->arg);
 #if CONFIG_MACF
        error = mac_file_check_fcntl(proc_ucred(p), fp->f_fglob, uap->cmd,
            uap->arg);

-       if (error)
+       if (error) {

                goto out;
                goto out;

+       }

 #endif
 
        switch (uap->cmd) {
 #endif
 
        switch (uap->cmd) {

-

        case F_DUPFD:
        case F_DUPFD_CLOEXEC:
                if (FP_ISGUARDED(fp, GUARD_DUP)) {
        case F_DUPFD:
        case F_DUPFD_CLOEXEC:
                if (FP_ISGUARDED(fp, GUARD_DUP)) {


@@ -814,8 +851,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = EINVAL;
                        goto out;
                }
                        error = EINVAL;
                        goto out;
                }

-               if ( (error = fdalloc(p, newmin, &i)) )
+               if ((error = fdalloc(p, newmin, &i))) {

                        goto out;
                        goto out;

+               }

                error = finishdup(p, fdp, fd, i,
                    uap->cmd == F_DUPFD_CLOEXEC ? UF_EXCLOSE : 0, retval);
                goto out;
                error = finishdup(p, fdp, fd, i,
                    uap->cmd == F_DUPFD_CLOEXEC ? UF_EXCLOSE : 0, retval);
                goto out;


@@ -827,9 +865,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 
        case F_SETFD:
                AUDIT_ARG(value32, uap->arg);
 
        case F_SETFD:
                AUDIT_ARG(value32, uap->arg);

-               if (uap->arg & FD_CLOEXEC)
+               if (uap->arg & FD_CLOEXEC) {

                        *pop |= UF_EXCLOSE;
                        *pop |= UF_EXCLOSE;

-               else {
+               } else {

                        if (FILEPROC_TYPE(fp) == FTYPE_GUARDED) {
                                error = fp_guard_exception(p,
                                    fd, fp, kGUARD_EXC_NOCLOEXEC);
                        if (FILEPROC_TYPE(fp) == FTYPE_GUARDED) {
                                error = fp_guard_exception(p,
                                    fd, fp, kGUARD_EXC_NOCLOEXEC);


@@ -852,12 +890,14 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                fp->f_flag |= FFLAGS(tmp) & FCNTLFLAGS;
                tmp = fp->f_flag & FNONBLOCK;
                error = fo_ioctl(fp, FIONBIO, (caddr_t)&tmp, &context);
                fp->f_flag |= FFLAGS(tmp) & FCNTLFLAGS;
                tmp = fp->f_flag & FNONBLOCK;
                error = fo_ioctl(fp, FIONBIO, (caddr_t)&tmp, &context);

-               if (error)
+               if (error) {

                        goto out;
                        goto out;

+               }

                tmp = fp->f_flag & FASYNC;
                error = fo_ioctl(fp, FIOASYNC, (caddr_t)&tmp, &context);
                tmp = fp->f_flag & FASYNC;
                error = fo_ioctl(fp, FIOASYNC, (caddr_t)&tmp, &context);

-               if (!error)
+               if (!error) {

                        goto out;
                        goto out;

+               }

                fp->f_flag &= ~FNONBLOCK;
                tmp = 0;
                (void)fo_ioctl(fp, FIONBIO, (caddr_t)&tmp, &context);
                fp->f_flag &= ~FNONBLOCK;
                tmp = 0;
                (void)fo_ioctl(fp, FIONBIO, (caddr_t)&tmp, &context);


@@ -869,7 +909,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = 0;
                        goto out;
                }
                        error = 0;
                        goto out;
                }

-               error = fo_ioctl(fp, (int)TIOCGPGRP, (caddr_t)retval, &context);
+               error = fo_ioctl(fp, TIOCGPGRP, (caddr_t)retval, &context);

                *retval = -*retval;
                goto out;
 
                *retval = -*retval;
                goto out;
 


@@ -878,7 +918,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                AUDIT_ARG(value32, tmp);
                if (fp->f_type == DTYPE_SOCKET) {
                        ((struct socket *)fp->f_data)->so_pgid = tmp;
                AUDIT_ARG(value32, tmp);
                if (fp->f_type == DTYPE_SOCKET) {
                        ((struct socket *)fp->f_data)->so_pgid = tmp;

-                       error =0;
+                       error = 0;

                        goto out;
                }
                if (fp->f_type == DTYPE_PIPE) {
                        goto out;
                }
                if (fp->f_type == DTYPE_PIPE) {


@@ -897,7 +937,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        tmp = (int)p1->p_pgrpid;
                        proc_rele(p1);
                }
                        tmp = (int)p1->p_pgrpid;
                        proc_rele(p1);
                }

-               error =  fo_ioctl(fp, (int)TIOCSPGRP, (caddr_t)&tmp, &context);
+               error =  fo_ioctl(fp, TIOCSPGRP, (caddr_t)&tmp, &context);

                goto out;
 
        case F_SETNOSIGPIPE:
                goto out;
 
        case F_SETNOSIGPIPE:


@@ -905,7 +945,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                if (fp->f_type == DTYPE_SOCKET) {
 #if SOCKETS
                        error = sock_setsockopt((struct socket *)fp->f_data,
                if (fp->f_type == DTYPE_SOCKET) {
 #if SOCKETS
                        error = sock_setsockopt((struct socket *)fp->f_data,

-                           SOL_SOCKET, SO_NOSIGPIPE, &tmp, sizeof (tmp));
+                           SOL_SOCKET, SO_NOSIGPIPE, &tmp, sizeof(tmp));

 #else
                        error = EINVAL;
 #endif
 #else
                        error = EINVAL;
 #endif


@@ -913,10 +953,11 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        struct fileglob *fg = fp->f_fglob;
 
                        lck_mtx_lock_spin(&fg->fg_lock);
                        struct fileglob *fg = fp->f_fglob;
 
                        lck_mtx_lock_spin(&fg->fg_lock);

-                       if (tmp)
+                       if (tmp) {

                                fg->fg_lflags |= FG_NOSIGPIPE;
                                fg->fg_lflags |= FG_NOSIGPIPE;

-                       else
-                               fg->fg_lflags &= FG_NOSIGPIPE;
+                       } else {
+                               fg->fg_lflags &= ~FG_NOSIGPIPE;
+                       }

                        lck_mtx_unlock(&fg->fg_lock);
                        error = 0;
                }
                        lck_mtx_unlock(&fg->fg_lock);
                        error = 0;
                }


@@ -925,7 +966,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
        case F_GETNOSIGPIPE:
                if (fp->f_type == DTYPE_SOCKET) {
 #if SOCKETS
        case F_GETNOSIGPIPE:
                if (fp->f_type == DTYPE_SOCKET) {
 #if SOCKETS

-                       int retsize = sizeof (*retval);
+                       int retsize = sizeof(*retval);

                        error = sock_getsockopt((struct socket *)fp->f_data,
                            SOL_SOCKET, SO_NOSIGPIPE, retval, &retsize);
 #else
                        error = sock_getsockopt((struct socket *)fp->f_data,
                            SOL_SOCKET, SO_NOSIGPIPE, retval, &retsize);
 #else


@@ -933,17 +974,57 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 #endif
                } else {
                        *retval = (fp->f_fglob->fg_lflags & FG_NOSIGPIPE) ?
 #endif
                } else {
                        *retval = (fp->f_fglob->fg_lflags & FG_NOSIGPIPE) ?

-                               1 : 0;
+                           1 : 0;

                        error = 0;
                }
                goto out;
 
                        error = 0;
                }
                goto out;
 

+       case F_SETCONFINED:
+               /*
+                * If this is the only reference to this fglob in the process
+                * and it's already marked as close-on-fork then mark it as
+                * (immutably) "confined" i.e. any fd that points to it will
+                * forever be close-on-fork, and attempts to use an IPC
+                * mechanism to move the descriptor elsewhere will fail.
+                */
+               if (CAST_DOWN_EXPLICIT(int, uap->arg)) {
+                       struct fileglob *fg = fp->f_fglob;
+
+                       lck_mtx_lock_spin(&fg->fg_lock);
+                       if (fg->fg_lflags & FG_CONFINED) {
+                               error = 0;
+                       } else if (1 != fg->fg_count) {
+                               error = EAGAIN; /* go close the dup .. */
+                       } else if (UF_FORKCLOSE == (*pop & UF_FORKCLOSE)) {
+                               fg->fg_lflags |= FG_CONFINED;
+                               error = 0;
+                       } else {
+                               error = EBADF;  /* open without O_CLOFORK? */
+                       }
+                       lck_mtx_unlock(&fg->fg_lock);
+               } else {
+                       /*
+                        * Other subsystems may have built on the immutability
+                        * of FG_CONFINED; clearing it may be tricky.
+                        */
+                       error = EPERM;          /* immutable */
+               }
+               goto out;
+
+       case F_GETCONFINED:
+               *retval = (fp->f_fglob->fg_lflags & FG_CONFINED) ? 1 : 0;
+               error = 0;
+               goto out;
+

        case F_SETLKWTIMEOUT:
        case F_SETLKW:
        case F_SETLKWTIMEOUT:
        case F_SETLKW:

+       case F_OFD_SETLKWTIMEOUT:
+       case F_OFD_SETLKW:

                flg |= F_WAIT;
                flg |= F_WAIT;

-               /* Fall into F_SETLK */
+       /* Fall into F_SETLK */

 
        case F_SETLK:
 
        case F_SETLK:

+       case F_OFD_SETLK:

                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;


@@ -955,7 +1036,8 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                proc_fdunlock(p);
 
                /* Copy in the lock structure */
                proc_fdunlock(p);
 
                /* Copy in the lock structure */

-               if (uap->cmd == F_SETLKWTIMEOUT) {
+               if (F_SETLKWTIMEOUT == uap->cmd ||
+                   F_OFD_SETLKWTIMEOUT == uap->cmd) {

                        error = copyin(argp, (caddr_t) &fltimeout, sizeof(fltimeout));
                        if (error) {
                                goto outdrop;
                        error = copyin(argp, (caddr_t) &fltimeout, sizeof(fltimeout));
                        if (error) {
                                goto outdrop;


@@ -976,11 +1058,12 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        goto outdrop;
                }
 
                        goto outdrop;
                }
 

-               if ( (error = vnode_getwithref(vp)) ) {
+               if ((error = vnode_getwithref(vp))) {

                        goto outdrop;
                }
                        goto outdrop;
                }

-               if (fl.l_whence == SEEK_CUR)
+               if (fl.l_whence == SEEK_CUR) {

                        fl.l_start += offset;
                        fl.l_start += offset;

+               }

 
 #if CONFIG_MACF
                error = mac_file_check_lock(proc_ucred(p), fp->f_fglob,
 
 #if CONFIG_MACF
                error = mac_file_check_lock(proc_ucred(p), fp->f_fglob,


@@ -990,45 +1073,92 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        goto outdrop;
                }
 #endif
                        goto outdrop;
                }
 #endif

-               switch (fl.l_type) {
-
-               case F_RDLCK:
-                       if ((fflag & FREAD) == 0) {
-                               (void)vnode_put(vp);
-                               error = EBADF;
-                               goto outdrop;
+               switch (uap->cmd) {
+               case F_OFD_SETLK:
+               case F_OFD_SETLKW:
+               case F_OFD_SETLKWTIMEOUT:
+                       flg |= F_OFD_LOCK;
+                       switch (fl.l_type) {
+                       case F_RDLCK:
+                               if ((fflag & FREAD) == 0) {
+                                       error = EBADF;
+                                       break;
+                               }
+                               error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob,
+                                   F_SETLK, &fl, flg, &context, timeout);
+                               break;
+                       case F_WRLCK:
+                               if ((fflag & FWRITE) == 0) {
+                                       error = EBADF;
+                                       break;
+                               }
+                               error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob,
+                                   F_SETLK, &fl, flg, &context, timeout);
+                               break;
+                       case F_UNLCK:
+                               error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob,
+                                   F_UNLCK, &fl, F_OFD_LOCK, &context,
+                                   timeout);
+                               break;
+                       default:
+                               error = EINVAL;
+                               break;

                        }
                        }

-                       // XXX UInt32 unsafe for LP64 kernel
-                       OSBitOrAtomic(P_LADVLOCK, &p->p_ladvflag);
-                       error = VNOP_ADVLOCK(vp, (caddr_t)p, F_SETLK, &fl, flg, &context, timeout);
-                       (void)vnode_put(vp);
-                       goto outdrop;
+                       if (0 == error &&
+                           (F_RDLCK == fl.l_type || F_WRLCK == fl.l_type)) {
+                               struct fileglob *fg = fp->f_fglob;

 
 

-               case F_WRLCK:
-                       if ((fflag & FWRITE) == 0) {
-                               (void)vnode_put(vp);
-                               error = EBADF;
-                               goto outdrop;
+                               /*
+                                * arrange F_UNLCK on last close (once
+                                * set, FG_HAS_OFDLOCK is immutable)
+                                */
+                               if ((fg->fg_lflags & FG_HAS_OFDLOCK) == 0) {
+                                       lck_mtx_lock_spin(&fg->fg_lock);
+                                       fg->fg_lflags |= FG_HAS_OFDLOCK;
+                                       lck_mtx_unlock(&fg->fg_lock);
+                               }

                        }
                        }

-                       // XXX UInt32 unsafe for LP64 kernel
-                       OSBitOrAtomic(P_LADVLOCK, &p->p_ladvflag);
-                       error = VNOP_ADVLOCK(vp, (caddr_t)p, F_SETLK, &fl, flg, &context, timeout);
-                       (void)vnode_put(vp);
-                       goto outdrop;
-
-               case F_UNLCK:
-                       error = VNOP_ADVLOCK(vp, (caddr_t)p, F_UNLCK, &fl,
-                               F_POSIX, &context, timeout);
-                       (void)vnode_put(vp);
-                       goto outdrop;
-
+                       break;

                default:
                default:

-                       (void)vnode_put(vp);
-                       error = EINVAL;
-                       goto outdrop;
+                       flg |= F_POSIX;
+                       switch (fl.l_type) {
+                       case F_RDLCK:
+                               if ((fflag & FREAD) == 0) {
+                                       error = EBADF;
+                                       break;
+                               }
+                               // XXX UInt32 unsafe for LP64 kernel
+                               OSBitOrAtomic(P_LADVLOCK, &p->p_ladvflag);
+                               error = VNOP_ADVLOCK(vp, (caddr_t)p,
+                                   F_SETLK, &fl, flg, &context, timeout);
+                               break;
+                       case F_WRLCK:
+                               if ((fflag & FWRITE) == 0) {
+                                       error = EBADF;
+                                       break;
+                               }
+                               // XXX UInt32 unsafe for LP64 kernel
+                               OSBitOrAtomic(P_LADVLOCK, &p->p_ladvflag);
+                               error = VNOP_ADVLOCK(vp, (caddr_t)p,
+                                   F_SETLK, &fl, flg, &context, timeout);
+                               break;
+                       case F_UNLCK:
+                               error = VNOP_ADVLOCK(vp, (caddr_t)p,
+                                   F_UNLCK, &fl, F_POSIX, &context, timeout);
+                               break;
+                       default:
+                               error = EINVAL;
+                               break;
+                       }
+                       break;

                }
                }

+               (void) vnode_put(vp);
+               goto outdrop;

 
        case F_GETLK:
 
        case F_GETLK:

+       case F_OFD_GETLK:
+       case F_GETLKPID:
+       case F_OFD_GETLKPID:

                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;


@@ -1040,8 +1170,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 
                /* Copy in the lock structure */
                error = copyin(argp, (caddr_t)&fl, sizeof(fl));
 
                /* Copy in the lock structure */
                error = copyin(argp, (caddr_t)&fl, sizeof(fl));

-               if (error)
+               if (error) {

                        goto outdrop;
                        goto outdrop;

+               }

 
                /* Check starting byte and ending byte for EOVERFLOW in SEEK_CUR */
                /* and ending byte for EOVERFLOW in SEEK_SET */
 
                /* Check starting byte and ending byte for EOVERFLOW in SEEK_CUR */
                /* and ending byte for EOVERFLOW in SEEK_SET */


@@ -1075,21 +1206,36 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        goto outdrop;
                }
 
                        goto outdrop;
                }
 

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
-                       if (fl.l_whence == SEEK_CUR)
-                               fl.l_start += offset;
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       if (fl.l_whence == SEEK_CUR) {
+                               fl.l_start += offset;
+                       }

 
 #if CONFIG_MACF
                        error = mac_file_check_lock(proc_ucred(p), fp->f_fglob,
                            uap->cmd, &fl);
                        if (error == 0)
 #endif
 
 #if CONFIG_MACF
                        error = mac_file_check_lock(proc_ucred(p), fp->f_fglob,
                            uap->cmd, &fl);
                        if (error == 0)
 #endif

-                       error = VNOP_ADVLOCK(vp, (caddr_t)p, uap->cmd, &fl, F_POSIX, &context, NULL);
+                       switch (uap->cmd) {
+                       case F_OFD_GETLK:
+                               error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob,
+                                   F_GETLK, &fl, F_OFD_LOCK, &context, NULL);
+                               break;
+                       case F_OFD_GETLKPID:
+                               error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob,
+                                   F_GETLKPID, &fl, F_OFD_LOCK, &context, NULL);
+                               break;
+                       default:
+                               error = VNOP_ADVLOCK(vp, (caddr_t)p,
+                                   uap->cmd, &fl, F_POSIX, &context, NULL);
+                               break;
+                       }

 
                        (void)vnode_put(vp);
 
 
                        (void)vnode_put(vp);
 

-                       if (error == 0)
+                       if (error == 0) {

                                error = copyout((caddr_t)&fl, argp, sizeof(fl));
                                error = copyout((caddr_t)&fl, argp, sizeof(fl));

+                       }

                }
                goto outdrop;
 
                }
                goto outdrop;
 


@@ -1112,12 +1258,13 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                }
 
                error = copyin(argp, (caddr_t)&alloc_struct, sizeof(alloc_struct));
                }
 
                error = copyin(argp, (caddr_t)&alloc_struct, sizeof(alloc_struct));

-               if (error)
+               if (error) {

                        goto outdrop;
                        goto outdrop;

+               }

 
                /* now set the space allocated to 0 */
                alloc_struct.fst_bytesalloc = 0;
 
                /* now set the space allocated to 0 */
                alloc_struct.fst_bytesalloc = 0;

-               
+

                /*
                 * Do some simple parameter checking
                 */
                /*
                 * Do some simple parameter checking
                 */


@@ -1125,12 +1272,14 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                /* set up the flags */
 
                alloc_flags |= PREALLOCATE;
                /* set up the flags */
 
                alloc_flags |= PREALLOCATE;

-               
-               if (alloc_struct.fst_flags & F_ALLOCATECONTIG)
+
+               if (alloc_struct.fst_flags & F_ALLOCATECONTIG) {

                        alloc_flags |= ALLOCATECONTIG;
                        alloc_flags |= ALLOCATECONTIG;

+               }

 
 

-               if (alloc_struct.fst_flags & F_ALLOCATEALL)
+               if (alloc_struct.fst_flags & F_ALLOCATEALL) {

                        alloc_flags |= ALLOCATEALL;
                        alloc_flags |= ALLOCATEALL;

+               }

 
                /*
                 * Do any position mode specific stuff.  The only
 
                /*
                 * Do any position mode specific stuff.  The only


@@ -1138,7 +1287,6 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                 */
 
                switch (alloc_struct.fst_posmode) {
                 */
 
                switch (alloc_struct.fst_posmode) {

-       

                case F_PEOFPOSMODE:
                        if (alloc_struct.fst_offset != 0) {
                                error = EINVAL;
                case F_PEOFPOSMODE:
                        if (alloc_struct.fst_offset != 0) {
                                error = EINVAL;


@@ -1160,25 +1308,141 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                default: {
                        error = EINVAL;
                        goto outdrop;
                default: {
                        error = EINVAL;
                        goto outdrop;

-                       }

                }
                }

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
-                       /*
+               }
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       /*

                         * call allocate to get the space
                         */
                         * call allocate to get the space
                         */

-                       error = VNOP_ALLOCATE(vp,alloc_struct.fst_length,alloc_flags,
-                                             &alloc_struct.fst_bytesalloc, alloc_struct.fst_offset,
-                                             &context);
+                       error = VNOP_ALLOCATE(vp, alloc_struct.fst_length, alloc_flags,
+                           &alloc_struct.fst_bytesalloc, alloc_struct.fst_offset,
+                           &context);

                        (void)vnode_put(vp);
 
                        error2 = copyout((caddr_t)&alloc_struct, argp, sizeof(alloc_struct));
 
                        (void)vnode_put(vp);
 
                        error2 = copyout((caddr_t)&alloc_struct, argp, sizeof(alloc_struct));
 

-                       if (error == 0)
+                       if (error == 0) {

                                error = error2;
                                error = error2;

+                       }
+               }
+               goto outdrop;
+       }
+       case F_PUNCHHOLE: {
+               fpunchhole_t args;
+
+               if (fp->f_type != DTYPE_VNODE) {
+                       error = EBADF;
+                       goto out;
+               }
+
+               vp = (struct vnode *)fp->f_data;
+               proc_fdunlock(p);
+
+               /* need write permissions */
+               if ((fp->f_flag & FWRITE) == 0) {
+                       error = EPERM;
+                       goto outdrop;
+               }
+
+               if ((error = copyin(argp, (caddr_t)&args, sizeof(args)))) {
+                       goto outdrop;
+               }
+
+               if ((error = vnode_getwithref(vp))) {
+                       goto outdrop;
+               }
+
+#if CONFIG_MACF
+               if ((error = mac_vnode_check_write(&context, fp->f_fglob->fg_cred, vp))) {
+                       (void)vnode_put(vp);
+                       goto outdrop;
+               }
+#endif
+
+               error = VNOP_IOCTL(vp, F_PUNCHHOLE, (caddr_t)&args, 0, &context);
+               (void)vnode_put(vp);
+
+               goto outdrop;
+       }
+       case F_TRIM_ACTIVE_FILE: {
+               ftrimactivefile_t args;
+
+               if (priv_check_cred(kauth_cred_get(), PRIV_TRIM_ACTIVE_FILE, 0)) {
+                       error = EACCES;
+                       goto out;
+               }
+
+               if (fp->f_type != DTYPE_VNODE) {
+                       error = EBADF;
+                       goto out;
+               }
+
+               vp = (struct vnode *)fp->f_data;
+               proc_fdunlock(p);
+
+               /* need write permissions */
+               if ((fp->f_flag & FWRITE) == 0) {
+                       error = EPERM;
+                       goto outdrop;
+               }
+
+               if ((error = copyin(argp, (caddr_t)&args, sizeof(args)))) {
+                       goto outdrop;
+               }
+
+               if ((error = vnode_getwithref(vp))) {
+                       goto outdrop;

                }
                }

+
+               error = VNOP_IOCTL(vp, F_TRIM_ACTIVE_FILE, (caddr_t)&args, 0, &context);
+               (void)vnode_put(vp);
+

                goto outdrop;
                goto outdrop;

-               
+       }
+       case F_SPECULATIVE_READ: {
+               fspecread_t args;
+
+               if (fp->f_type != DTYPE_VNODE) {
+                       error = EBADF;
+                       goto out;
+               }
+
+               vp = (struct vnode *)fp->f_data;
+               proc_fdunlock(p);
+
+               if ((error = copyin(argp, (caddr_t)&args, sizeof(args)))) {
+                       goto outdrop;
+               }
+
+               /* Discard invalid offsets or lengths */
+               if ((args.fsr_offset < 0) || (args.fsr_length < 0)) {
+                       error = EINVAL;
+                       goto outdrop;
+               }
+
+               /*
+                * Round the file offset down to a page-size boundary (or to 0).
+                * The filesystem will need to round the length up to the end of the page boundary
+                * or to the EOF of the file.
+                */
+               uint64_t foff = (((uint64_t)args.fsr_offset) & ~((uint64_t)PAGE_MASK));
+               uint64_t foff_delta = args.fsr_offset - foff;
+               args.fsr_offset = (off_t) foff;
+
+               /*
+                * Now add in the delta to the supplied length. Since we may have adjusted the
+                * offset, increase it by the amount that we adjusted.
+                */
+               args.fsr_length += foff_delta;
+
+               if ((error = vnode_getwithref(vp))) {
+                       goto outdrop;

                }
                }

+               error = VNOP_IOCTL(vp, F_SPECULATIVE_READ, (caddr_t)&args, 0, &context);
+               (void)vnode_put(vp);
+
+               goto outdrop;
+       }

        case F_SETSIZE:
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
        case F_SETSIZE:
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;


@@ -1187,14 +1451,16 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 

-               error = copyin(argp, (caddr_t)&offset, sizeof (off_t));
-               if (error)
+               error = copyin(argp, (caddr_t)&offset, sizeof(off_t));
+               if (error) {

                        goto outdrop;
                        goto outdrop;

+               }

                AUDIT_ARG(value64, offset);
 
                error = vnode_getwithref(vp);
                AUDIT_ARG(value64, offset);
 
                error = vnode_getwithref(vp);

-               if (error)
+               if (error) {

                        goto outdrop;
                        goto outdrop;

+               }

 
 #if CONFIG_MACF
                error = mac_vnode_check_truncate(&context,
 
 #if CONFIG_MACF
                error = mac_vnode_check_truncate(&context,


@@ -1206,7 +1472,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 #endif
                /*
                 * Make sure that we are root.  Growing a file
 #endif
                /*
                 * Make sure that we are root.  Growing a file

-                * without zero filling the data is a security hole 
+                * without zero filling the data is a security hole

                 * root would have access anyway so we'll allow it
                 */
                if (!kauth_cred_issuser(kauth_cred_get())) {
                 * root would have access anyway so we'll allow it
                 */
                if (!kauth_cred_issuser(kauth_cred_get())) {


@@ -1217,6 +1483,12 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                         */
                        error = vnode_setsize(vp, offset, IO_NOZEROFILL,
                            &context);
                         */
                        error = vnode_setsize(vp, offset, IO_NOZEROFILL,
                            &context);

+
+#if CONFIG_MACF
+                       if (error == 0) {
+                               mac_vnode_notify_truncate(&context, fp->f_fglob->fg_cred, vp);
+                       }
+#endif

                }
 
                (void)vnode_put(vp);
                }
 
                (void)vnode_put(vp);


@@ -1227,10 +1499,11 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = EBADF;
                        goto out;
                }
                        error = EBADF;
                        goto out;
                }

-               if (uap->arg)
-                       fp->f_fglob->fg_flag &= ~FNORDAHEAD;
-               else
-                       fp->f_fglob->fg_flag |= FNORDAHEAD;
+               if (uap->arg) {
+                       fp->f_fglob->fg_flag &= ~FNORDAHEAD;
+               } else {
+                       fp->f_fglob->fg_flag |= FNORDAHEAD;
+               }

 
                goto out;
 
 
                goto out;
 


@@ -1239,10 +1512,11 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = EBADF;
                        goto out;
                }
                        error = EBADF;
                        goto out;
                }

-               if (uap->arg)
-                       fp->f_fglob->fg_flag |= FNOCACHE;
-               else
-                       fp->f_fglob->fg_flag &= ~FNOCACHE;
+               if (uap->arg) {
+                       fp->f_fglob->fg_flag |= FNOCACHE;
+               } else {
+                       fp->f_fglob->fg_flag &= ~FNOCACHE;
+               }

 
                goto out;
 
 
                goto out;
 


@@ -1251,10 +1525,11 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = EBADF;
                        goto out;
                }
                        error = EBADF;
                        goto out;
                }

-               if (uap->arg)
-                       fp->f_fglob->fg_flag |= FNODIRECT;
-               else
-                       fp->f_fglob->fg_flag &= ~FNODIRECT;
+               if (uap->arg) {
+                       fp->f_fglob->fg_flag |= FNODIRECT;
+               } else {
+                       fp->f_fglob->fg_flag &= ~FNODIRECT;
+               }

 
                goto out;
 
 
                goto out;
 


@@ -1263,50 +1538,51 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = EBADF;
                        goto out;
                }
                        error = EBADF;
                        goto out;
                }

-               if (uap->arg)
-                       fp->f_fglob->fg_flag |= FSINGLE_WRITER;
-               else
-                       fp->f_fglob->fg_flag &= ~FSINGLE_WRITER;
+               if (uap->arg) {
+                       fp->f_fglob->fg_flag |= FSINGLE_WRITER;
+               } else {
+                       fp->f_fglob->fg_flag &= ~FSINGLE_WRITER;
+               }

 
                goto out;
 
        case F_GLOBAL_NOCACHE:
 
                goto out;
 
        case F_GLOBAL_NOCACHE:

-               if (fp->f_type != DTYPE_VNODE) {
-                       error = EBADF;
-                       goto out;
+               if (fp->f_type != DTYPE_VNODE) {
+                       error = EBADF;
+                       goto out;

                }
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 
                }
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
-
-                       *retval = vnode_isnocache(vp);
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       *retval = vnode_isnocache(vp);

 
 

-                       if (uap->arg)
-                               vnode_setnocache(vp);
-                       else
-                               vnode_clearnocache(vp);
+                       if (uap->arg) {
+                               vnode_setnocache(vp);
+                       } else {
+                               vnode_clearnocache(vp);
+                       }

 
                        (void)vnode_put(vp);
                }
                goto outdrop;
 
        case F_CHECK_OPENEVT:
 
                        (void)vnode_put(vp);
                }
                goto outdrop;
 
        case F_CHECK_OPENEVT:

-               if (fp->f_type != DTYPE_VNODE) {
-                       error = EBADF;
+               if (fp->f_type != DTYPE_VNODE) {
+                       error = EBADF;

                        goto out;
                }
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 
                        goto out;
                }
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       *retval = vnode_is_openevt(vp);

 
 

-                       *retval = vnode_is_openevt(vp);
-
-                       if (uap->arg)
-                               vnode_set_openevt(vp);
-                       else
-                               vnode_clear_openevt(vp);
+                       if (uap->arg) {
+                               vnode_set_openevt(vp);
+                       } else {
+                               vnode_clear_openevt(vp);
+                       }

 
                        (void)vnode_put(vp);
                }
 
                        (void)vnode_put(vp);
                }


@@ -1322,35 +1598,36 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 

-               if ( (error = copyin(argp, (caddr_t)&ra_struct, sizeof(ra_struct))) )
+               if ((error = copyin(argp, (caddr_t)&ra_struct, sizeof(ra_struct)))) {

                        goto outdrop;
                        goto outdrop;

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
-                       error = VNOP_IOCTL(vp, F_RDADVISE, (caddr_t)&ra_struct, 0, &context);
+               }
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       error = VNOP_IOCTL(vp, F_RDADVISE, (caddr_t)&ra_struct, 0, &context);

 
                        (void)vnode_put(vp);
                }
                goto outdrop;
 
                        (void)vnode_put(vp);
                }
                goto outdrop;

-               }
+       }

 
 

-        case F_FLUSH_DATA:
+       case F_FLUSH_DATA:

 
 

-                if (fp->f_type != DTYPE_VNODE) {
-                        error = EBADF;
-                        goto out;
-                }
-                vp = (struct vnode *)fp->f_data;
-                proc_fdunlock(p);
+               if (fp->f_type != DTYPE_VNODE) {
+                       error = EBADF;
+                       goto out;
+               }
+               vp = (struct vnode *)fp->f_data;
+               proc_fdunlock(p);

 
 

-                if ( (error = vnode_getwithref(vp)) == 0 ) {
-                        error = cluster_push(vp, 0);
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       error = VNOP_FSYNC(vp, MNT_NOWAIT, &context);

 
 

-                        (void)vnode_put(vp);
-                }
-                goto outdrop;
+                       (void)vnode_put(vp);
+               }
+               goto outdrop;

 
        case F_LOG2PHYS:
        case F_LOG2PHYS_EXT: {
 
        case F_LOG2PHYS:
        case F_LOG2PHYS_EXT: {

-               struct log2phys l2p_struct;    /* structure for allocate command */
+               struct log2phys l2p_struct = {};    /* structure for allocate command */

                int devBlockSize;
 
                off_t file_offset = 0;
                int devBlockSize;
 
                off_t file_offset = 0;


@@ -1359,8 +1636,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 
                if (uap->cmd == F_LOG2PHYS_EXT) {
                        error = copyin(argp, (caddr_t)&l2p_struct, sizeof(l2p_struct));
 
                if (uap->cmd == F_LOG2PHYS_EXT) {
                        error = copyin(argp, (caddr_t)&l2p_struct, sizeof(l2p_struct));

-                       if (error)
+                       if (error) {

                                goto out;
                                goto out;

+                       }

                        file_offset = l2p_struct.l2p_devoffset;
                } else {
                        file_offset = fp->f_offset;
                        file_offset = l2p_struct.l2p_devoffset;
                } else {
                        file_offset = fp->f_offset;


@@ -1371,7 +1649,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                }
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
                }
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);

-               if ( (error = vnode_getwithref(vp)) ) {
+               if ((error = vnode_getwithref(vp))) {

                        goto outdrop;
                }
                error = VNOP_OFFTOBLK(vp, file_offset, &lbn);
                        goto outdrop;
                }
                error = VNOP_OFFTOBLK(vp, file_offset, &lbn);


@@ -1386,37 +1664,27 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                }
                devBlockSize = vfs_devblocksize(vnode_mount(vp));
                if (uap->cmd == F_LOG2PHYS_EXT) {
                }
                devBlockSize = vfs_devblocksize(vnode_mount(vp));
                if (uap->cmd == F_LOG2PHYS_EXT) {

-#if defined(__LP64__) 
-                       a_size = l2p_struct.l2p_contigbytes;
-#else
-                       if ((l2p_struct.l2p_contigbytes > SIZE_MAX) || (l2p_struct.l2p_contigbytes < 0)) {
-                               /* size_t is 32-bit on a 32-bit kernel, therefore 
-                                * assigning l2p_contigbytes to a_size may have 
-                                * caused integer overflow.  We, therefore, return 
-                                * an error here instead of calculating incorrect 
-                                * value. 
-                                */
-                               printf ("fcntl: F_LOG2PHYS_EXT: l2p_contigbytes=%lld will overflow, returning error\n", l2p_struct.l2p_contigbytes);
-                               error = EFBIG;
+                       if (l2p_struct.l2p_contigbytes < 0) {
+                               vnode_put(vp);
+                               error = EINVAL;

                                goto outdrop;
                                goto outdrop;

-                       } else {
-                               a_size = l2p_struct.l2p_contigbytes;

                        }
                        }

-#endif
+
+                       a_size = MIN((uint64_t)l2p_struct.l2p_contigbytes, SIZE_MAX);

                } else {
                        a_size = devBlockSize;
                }
                } else {
                        a_size = devBlockSize;
                }

-               
+

                error = VNOP_BLOCKMAP(vp, offset, a_size, &bn, &run, NULL, 0, &context);
 
                (void)vnode_put(vp);
 
                if (!error) {
                error = VNOP_BLOCKMAP(vp, offset, a_size, &bn, &run, NULL, 0, &context);
 
                (void)vnode_put(vp);
 
                if (!error) {

-                       l2p_struct.l2p_flags = 0;       /* for now */
+                       l2p_struct.l2p_flags = 0;       /* for now */

                        if (uap->cmd == F_LOG2PHYS_EXT) {
                                l2p_struct.l2p_contigbytes = run - (file_offset - offset);
                        } else {
                        if (uap->cmd == F_LOG2PHYS_EXT) {
                                l2p_struct.l2p_contigbytes = run - (file_offset - offset);
                        } else {

-                               l2p_struct.l2p_contigbytes = 0; /* for now */
+                               l2p_struct.l2p_contigbytes = 0; /* for now */

                        }
 
                        /*
                        }
 
                        /*


@@ -1426,16 +1694,16 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        if (bn == -1) {
                                /* Don't multiply it by the block size */
                                l2p_struct.l2p_devoffset = bn;
                        if (bn == -1) {
                                /* Don't multiply it by the block size */
                                l2p_struct.l2p_devoffset = bn;

-                       }
-                       else {
+                       } else {

                                l2p_struct.l2p_devoffset = bn * devBlockSize;
                                l2p_struct.l2p_devoffset += file_offset - offset;
                        }
                        error = copyout((caddr_t)&l2p_struct, argp, sizeof(l2p_struct));
                }
                goto outdrop;
                                l2p_struct.l2p_devoffset = bn * devBlockSize;
                                l2p_struct.l2p_devoffset += file_offset - offset;
                        }
                        error = copyout((caddr_t)&l2p_struct, argp, sizeof(l2p_struct));
                }
                goto outdrop;

-               }
-       case F_GETPATH: {
+       }
+       case F_GETPATH:
+       case F_GETPATH_NOFIRMLINK: {

                char *pathbufp;
                int pathlen;
 
                char *pathbufp;
                int pathlen;
 


@@ -1452,12 +1720,17 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = ENOMEM;
                        goto outdrop;
                }
                        error = ENOMEM;
                        goto outdrop;
                }

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
-                       error = vn_getpath(vp, pathbufp, &pathlen);
-                       (void)vnode_put(vp);
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       if (uap->cmd == F_GETPATH_NOFIRMLINK) {
+                               error = vn_getpath_ext(vp, NULL, pathbufp, &pathlen, VN_GETPATH_NO_FIRMLINK);
+                       } else {
+                               error = vn_getpath(vp, pathbufp, &pathlen);
+                       }
+                       (void)vnode_put(vp);

 
 

-                       if (error == 0)
-                               error = copyout((caddr_t)pathbufp, argp, pathlen);
+                       if (error == 0) {
+                               error = copyout((caddr_t)pathbufp, argp, pathlen);
+                       }

                }
                FREE(pathbufp, M_TEMP);
                goto outdrop;
                }
                FREE(pathbufp, M_TEMP);
                goto outdrop;


@@ -1468,7 +1741,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                size_t pathlen;
 
                if (fp->f_type != DTYPE_VNODE) {
                size_t pathlen;
 
                if (fp->f_type != DTYPE_VNODE) {

-                       error = EBADF;
+                       error = EBADF;

                        goto out;
                }
                vp = (struct vnode *)fp->f_data;
                        goto out;
                }
                vp = (struct vnode *)fp->f_data;


@@ -1477,10 +1750,10 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                pathlen = MAXPATHLEN;
                pathbufp = kalloc(MAXPATHLEN);
 
                pathlen = MAXPATHLEN;
                pathbufp = kalloc(MAXPATHLEN);
 

-               if ( (error = copyinstr(argp, pathbufp, MAXPATHLEN, &pathlen)) == 0 ) {
-                       if ( (error = vnode_getwithref(vp)) == 0 ) {
+               if ((error = copyinstr(argp, pathbufp, MAXPATHLEN, &pathlen)) == 0) {
+                       if ((error = vnode_getwithref(vp)) == 0) {

                                AUDIT_ARG(text, pathbufp);
                                AUDIT_ARG(text, pathbufp);

-                               error = vn_path_package_check(vp, pathbufp, pathlen, retval);
+                               error = vn_path_package_check(vp, pathbufp, pathlen, retval);

 
                                (void)vnode_put(vp);
                        }
 
                                (void)vnode_put(vp);
                        }


@@ -1490,7 +1763,8 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
        }
 
        case F_CHKCLEAN:   // used by regression tests to see if all dirty pages got cleaned by fsync()
        }
 
        case F_CHKCLEAN:   // used by regression tests to see if all dirty pages got cleaned by fsync()

-       case F_FULLFSYNC:  // fsync + flush the journal + DKIOCSYNCHRONIZECACHE
+       case F_FULLFSYNC:  // fsync + flush the journal + DKIOCSYNCHRONIZE
+       case F_BARRIERFSYNC:  // fsync + barrier

        case F_FREEZE_FS:  // freeze all other fs operations for the fs of this fd
        case F_THAW_FS: {  // thaw all frozen fs operations for the fs of this fd
                if (fp->f_type != DTYPE_VNODE) {
        case F_FREEZE_FS:  // freeze all other fs operations for the fs of this fd
        case F_THAW_FS: {  // thaw all frozen fs operations for the fs of this fd
                if (fp->f_type != DTYPE_VNODE) {


@@ -1500,8 +1774,8 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
-                       error = VNOP_IOCTL(vp, uap->cmd, (caddr_t)NULL, 0, &context);
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       error = VNOP_IOCTL(vp, uap->cmd, (caddr_t)NULL, 0, &context);

 
                        (void)vnode_put(vp);
                }
 
                        (void)vnode_put(vp);
                }


@@ -1530,7 +1804,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = ENOENT;
                        goto outdrop;
                }
                        error = ENOENT;
                        goto outdrop;
                }

-               
+

                /* Only valid for directories */
                if (vp->v_type != VDIR) {
                        vnode_put(vp);
                /* Only valid for directories */
                if (vp->v_type != VDIR) {
                        vnode_put(vp);


@@ -1557,16 +1831,16 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                AUDIT_ARG(mode, fopen.o_mode);
                VATTR_INIT(&va);
                /* Mask off all but regular access permissions */
                AUDIT_ARG(mode, fopen.o_mode);
                VATTR_INIT(&va);
                /* Mask off all but regular access permissions */

-               cmode = ((fopen.o_mode &~ fdp->fd_cmask) & ALLPERMS) & ~S_ISTXT;
+               cmode = ((fopen.o_mode & ~fdp->fd_cmask) & ALLPERMS) & ~S_ISTXT;

                VATTR_SET(&va, va_mode, cmode & ACCESSPERMS);
 
                /* Start the lookup relative to the file descriptor's vnode. */
                NDINIT(&nd, LOOKUP, OP_OPEN, USEDVP | FOLLOW | AUDITVNPATH1, UIO_USERSPACE,
                VATTR_SET(&va, va_mode, cmode & ACCESSPERMS);
 
                /* Start the lookup relative to the file descriptor's vnode. */
                NDINIT(&nd, LOOKUP, OP_OPEN, USEDVP | FOLLOW | AUDITVNPATH1, UIO_USERSPACE,

-                      fopen.o_pathname, &context);
+                   fopen.o_pathname, &context);

                nd.ni_dvp = vp;
 
                error = open1(&context, &nd, fopen.o_flags, &va,
                nd.ni_dvp = vp;
 
                error = open1(&context, &nd, fopen.o_flags, &va,

-                             fileproc_alloc_init, NULL, retval);
+                   fileproc_alloc_init, NULL, retval);

 
                vnode_put(vp);
                break;
 
                vnode_put(vp);
                break;


@@ -1575,7 +1849,6 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
         * SPI (private) for unlinking a file starting from a dir fd
         */
        case F_UNLINKFROM: {
         * SPI (private) for unlinking a file starting from a dir fd
         */
        case F_UNLINKFROM: {

-               struct nameidata nd;

                user_addr_t pathname;
 
                /* Check if this isn't a valid file descriptor */
                user_addr_t pathname;
 
                /* Check if this isn't a valid file descriptor */


@@ -1591,7 +1864,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = ENOENT;
                        goto outdrop;
                }
                        error = ENOENT;
                        goto outdrop;
                }

-               
+

                /* Only valid for directories */
                if (vp->v_type != VDIR) {
                        vnode_put(vp);
                /* Only valid for directories */
                if (vp->v_type != VDIR) {
                        vnode_put(vp);


@@ -1607,25 +1880,23 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                }
 
                /* Start the lookup relative to the file descriptor's vnode. */
                }
 
                /* Start the lookup relative to the file descriptor's vnode. */

-               NDINIT(&nd, DELETE, OP_UNLINK, USEDVP | AUDITVNPATH1, UIO_USERSPACE,
-                      pathname, &context);
-               nd.ni_dvp = vp;
+               error = unlink1(&context, vp, pathname, UIO_USERSPACE, 0);

 
 

-               error = unlink1(&context, &nd, 0);
-               

                vnode_put(vp);
                break;
                vnode_put(vp);
                break;

-

        }
 
        case F_ADDSIGS:
        case F_ADDFILESIGS:
        }
 
        case F_ADDSIGS:
        case F_ADDFILESIGS:

+       case F_ADDFILESIGS_FOR_DYLD_SIM:
+       case F_ADDFILESIGS_RETURN:

        {
        {

+               struct cs_blob *blob = NULL;

                struct user_fsignatures fs;
                kern_return_t kr;
                struct user_fsignatures fs;
                kern_return_t kr;

-               off_t kernel_blob_offset;

                vm_offset_t kernel_blob_addr;
                vm_size_t kernel_blob_size;
                vm_offset_t kernel_blob_addr;
                vm_size_t kernel_blob_size;

+               int blob_add_flags = 0;

 
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
 
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;


@@ -1633,16 +1904,27 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                }
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
                }
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);

+
+               if (uap->cmd == F_ADDFILESIGS_FOR_DYLD_SIM) {
+                       blob_add_flags |= MAC_VNODE_CHECK_DYLD_SIM;
+                       if ((p->p_csflags & CS_KILL) == 0) {
+                               proc_lock(p);
+                               p->p_csflags |= CS_KILL;
+                               proc_unlock(p);
+                       }
+               }
+

                error = vnode_getwithref(vp);
                error = vnode_getwithref(vp);

-               if (error)
+               if (error) {

                        goto outdrop;
                        goto outdrop;

+               }

 
                if (IS_64BIT_PROCESS(p)) {
 
                if (IS_64BIT_PROCESS(p)) {

-                       error = copyin(argp, &fs, sizeof (fs));
+                       error = copyin(argp, &fs, sizeof(fs));

                } else {
                        struct user32_fsignatures fs32;
 
                } else {
                        struct user32_fsignatures fs32;
 

-                       error = copyin(argp, &fs32, sizeof (fs32));
+                       error = copyin(argp, &fs32, sizeof(fs32));

                        fs.fs_file_start = fs32.fs_file_start;
                        fs.fs_blob_start = CAST_USER_ADDR_T(fs32.fs_blob_start);
                        fs.fs_blob_size = fs32.fs_blob_size;
                        fs.fs_file_start = fs32.fs_file_start;
                        fs.fs_blob_start = CAST_USER_ADDR_T(fs32.fs_blob_start);
                        fs.fs_blob_size = fs32.fs_blob_size;


@@ -1653,114 +1935,159 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        goto outdrop;
                }
 
                        goto outdrop;
                }
 

-               if(ubc_cs_blob_get(vp, CPU_TYPE_ANY, fs.fs_file_start))
-               {
-                       vnode_put(vp);
-                       goto outdrop;
-               }
-#if defined(__LP64__)
-#define CS_MAX_BLOB_SIZE (2560ULL * 1024ULL) /* max shared cache file XXX ? */
-#else     
-#define CS_MAX_BLOB_SIZE (1600ULL * 1024ULL) /* max shared cache file XXX ? */
-#endif
-               if (fs.fs_blob_size > CS_MAX_BLOB_SIZE) {
-                       error = E2BIG;
-                       vnode_put(vp);
-                       goto outdrop;
+               /*
+                * First check if we have something loaded a this offset
+                */
+               blob = ubc_cs_blob_get(vp, CPU_TYPE_ANY, fs.fs_file_start);
+               if (blob != NULL) {
+                       /* If this is for dyld_sim revalidate the blob */
+                       if (uap->cmd == F_ADDFILESIGS_FOR_DYLD_SIM) {
+                               error = ubc_cs_blob_revalidate(vp, blob, NULL, blob_add_flags);
+                               if (error) {
+                                       blob = NULL;
+                                       if (error != EAGAIN) {
+                                               vnode_put(vp);
+                                               goto outdrop;
+                                       }
+                               }
+                       }

                }
 
                }
 

-               kernel_blob_size = CAST_DOWN(vm_size_t, fs.fs_blob_size);
-               kr = ubc_cs_blob_allocate(&kernel_blob_addr, &kernel_blob_size);
-               if (kr != KERN_SUCCESS) {
-                       error = ENOMEM;
-                       vnode_put(vp);
-                       goto outdrop;
-               }
+               if (blob == NULL) {
+                       /*
+                        * An arbitrary limit, to prevent someone from mapping in a 20GB blob.  This should cover
+                        * our use cases for the immediate future, but note that at the time of this commit, some
+                        * platforms are nearing 2MB blob sizes (with a prior soft limit of 2.5MB).
+                        *
+                        * We should consider how we can manage this more effectively; the above means that some
+                        * platforms are using megabytes of memory for signing data; it merely hasn't crossed the
+                        * threshold considered ridiculous at the time of this change.
+                        */
+#define CS_MAX_BLOB_SIZE (40ULL * 1024ULL * 1024ULL)
+                       if (fs.fs_blob_size > CS_MAX_BLOB_SIZE) {
+                               error = E2BIG;
+                               vnode_put(vp);
+                               goto outdrop;
+                       }

 
 

-               if(uap->cmd == F_ADDSIGS) {
-                       kernel_blob_offset = 0;
-                       error = copyin(fs.fs_blob_start,
-                                      (void *) kernel_blob_addr,
-                                      kernel_blob_size);
-               } else /* F_ADDFILESIGS */ {
-                       kernel_blob_offset = fs.fs_blob_start;
-                       error = vn_rdwr(UIO_READ,
-                                       vp,
-                                       (caddr_t) kernel_blob_addr,
-                                       kernel_blob_size,
-                                        fs.fs_file_start + fs.fs_blob_start,
-                                       UIO_SYSSPACE,
-                                       0,
-                                       kauth_cred_get(),
-                                       0,
-                                       p);
-               }
-               
-               if (error) {
-                       ubc_cs_blob_deallocate(kernel_blob_addr,
-                                              kernel_blob_size);
-                       vnode_put(vp);
-                       goto outdrop;
-               }
+                       kernel_blob_size = CAST_DOWN(vm_size_t, fs.fs_blob_size);
+                       kr = ubc_cs_blob_allocate(&kernel_blob_addr, &kernel_blob_size);
+                       if (kr != KERN_SUCCESS) {
+                               error = ENOMEM;
+                               vnode_put(vp);
+                               goto outdrop;
+                       }

 
 

-               error = ubc_cs_blob_add(
-                       vp,
-                       CPU_TYPE_ANY,   /* not for a specific architecture */
-                       fs.fs_file_start,
-                       kernel_blob_addr,
-                       kernel_blob_offset,
-                       kernel_blob_size);
-               if (error) {
-                       ubc_cs_blob_deallocate(kernel_blob_addr,
-                                              kernel_blob_size);
-               } else {
-                       /* ubc_blob_add() has consumed "kernel_blob_addr" */
+                       if (uap->cmd == F_ADDSIGS) {
+                               error = copyin(fs.fs_blob_start,
+                                   (void *) kernel_blob_addr,
+                                   kernel_blob_size);
+                       } else { /* F_ADDFILESIGS || F_ADDFILESIGS_RETURN || F_ADDFILESIGS_FOR_DYLD_SIM */
+                               int resid;
+
+                               error = vn_rdwr(UIO_READ,
+                                   vp,
+                                   (caddr_t) kernel_blob_addr,
+                                   kernel_blob_size,
+                                   fs.fs_file_start + fs.fs_blob_start,
+                                   UIO_SYSSPACE,
+                                   0,
+                                   kauth_cred_get(),
+                                   &resid,
+                                   p);
+                               if ((error == 0) && resid) {
+                                       /* kernel_blob_size rounded to a page size, but signature may be at end of file */
+                                       memset((void *)(kernel_blob_addr + (kernel_blob_size - resid)), 0x0, resid);
+                               }
+                       }
+
+                       if (error) {
+                               ubc_cs_blob_deallocate(kernel_blob_addr,
+                                   kernel_blob_size);
+                               vnode_put(vp);
+                               goto outdrop;
+                       }
+
+                       blob = NULL;
+                       error = ubc_cs_blob_add(vp,
+                           CPU_TYPE_ANY,                       /* not for a specific architecture */
+                           fs.fs_file_start,
+                           &kernel_blob_addr,
+                           kernel_blob_size,
+                           NULL,
+                           blob_add_flags,
+                           &blob);
+
+                       /* ubc_blob_add() has consumed "kernel_blob_addr" if it is zeroed */
+                       if (error) {
+                               if (kernel_blob_addr) {
+                                       ubc_cs_blob_deallocate(kernel_blob_addr,
+                                           kernel_blob_size);
+                               }
+                               vnode_put(vp);
+                               goto outdrop;
+                       } else {

 #if CHECK_CS_VALIDATION_BITMAP
 #if CHECK_CS_VALIDATION_BITMAP

-                       ubc_cs_validation_bitmap_allocate( vp );
+                               ubc_cs_validation_bitmap_allocate( vp );

 #endif
 #endif

+                       }
+               }
+
+               if (uap->cmd == F_ADDFILESIGS_RETURN || uap->cmd == F_ADDFILESIGS_FOR_DYLD_SIM) {
+                       /*
+                        * The first element of the structure is a
+                        * off_t that happen to have the same size for
+                        * all archs. Lets overwrite that.
+                        */
+                       off_t end_offset = 0;
+                       if (blob) {
+                               end_offset = blob->csb_end_offset;
+                       }
+                       error = copyout(&end_offset, argp, sizeof(end_offset));

                }
 
                (void) vnode_put(vp);
                break;
        }
                }
 
                (void) vnode_put(vp);
                break;
        }

+       case F_GETCODEDIR:

        case F_FINDSIGS: {
        case F_FINDSIGS: {

-#ifdef SECURE_KERNEL

                error = ENOTSUP;
                error = ENOTSUP;

-#else /* !SECURE_KERNEL */
-               off_t offsetMacho;
+               goto out;
+       }
+       case F_CHECK_LV: {
+               struct fileglob *fg;
+               fchecklv_t lv = {};

 
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                }
 
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                }

-               vp = (struct vnode *)fp->f_data;
+               fg = fp->f_fglob;

                proc_fdunlock(p);
                proc_fdunlock(p);

-               error = vnode_getwithref(vp);
-               if (error)
-                       goto outdrop;

 
 

-               error = copyin(argp, &offsetMacho, sizeof(offsetMacho));
+               if (IS_64BIT_PROCESS(p)) {
+                       error = copyin(argp, &lv, sizeof(lv));
+               } else {
+                       struct user32_fchecklv lv32 = {};
+
+                       error = copyin(argp, &lv32, sizeof(lv32));
+                       lv.lv_file_start = lv32.lv_file_start;
+                       lv.lv_error_message = (void *)(uintptr_t)lv32.lv_error_message;
+                       lv.lv_error_message_size = lv32.lv_error_message_size;
+               }

                if (error) {
                if (error) {

-                       (void)vnode_put(vp);

                        goto outdrop;
                }
 
 #if CONFIG_MACF
                        goto outdrop;
                }
 
 #if CONFIG_MACF

-               error = mac_vnode_find_sigs(p, vp, offsetMacho);
-#else
-               error = EPERM;
+               error = mac_file_check_library_validation(p, fg, lv.lv_file_start,
+                   (user_long_t)lv.lv_error_message, lv.lv_error_message_size);

 #endif
 #endif

-               if (error) {
-                       (void)vnode_put(vp);
-                       goto outdrop;
-               }
-#endif /* SECURE_KERNEL */
+

                break;
        }
 #if CONFIG_PROTECT
        case F_GETPROTECTIONCLASS: {
                break;
        }
 #if CONFIG_PROTECT
        case F_GETPROTECTIONCLASS: {

-               int class = 0;
-               

                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;


@@ -1773,20 +2100,28 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = ENOENT;
                        goto outdrop;
                }
                        error = ENOENT;
                        goto outdrop;
                }

-       
-               error = cp_vnode_getclass (vp, &class);
-               if (error == 0) {
-                       *retval = class;
+
+               struct vnode_attr va;
+
+               VATTR_INIT(&va);
+               VATTR_WANTED(&va, va_dataprotect_class);
+               error = VNOP_GETATTR(vp, &va, &context);
+               if (!error) {
+                       if (VATTR_IS_SUPPORTED(&va, va_dataprotect_class)) {
+                               *retval = va.va_dataprotect_class;
+                       } else {
+                               error = ENOTSUP;
+                       }

                }
 
                vnode_put(vp);
                break;
        }
                }
 
                vnode_put(vp);
                break;
        }

-       
+

        case F_SETPROTECTIONCLASS: {
                /* tmp must be a valid PROTECTION_CLASS_* */
                tmp = CAST_DOWN_EXPLICIT(uint32_t, uap->arg);
        case F_SETPROTECTIONCLASS: {
                /* tmp must be a valid PROTECTION_CLASS_* */
                tmp = CAST_DOWN_EXPLICIT(uint32_t, uap->arg);

-               
+

                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;


@@ -1794,82 +2129,93 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                vp = (struct vnode *)fp->f_data;
 
                proc_fdunlock(p);
                vp = (struct vnode *)fp->f_data;
 
                proc_fdunlock(p);

-       
+

                if (vnode_getwithref(vp)) {
                        error = ENOENT;
                        goto outdrop;
                if (vnode_getwithref(vp)) {
                        error = ENOENT;
                        goto outdrop;

-               }       
-               
+               }
+

                /* Only go forward if you have write access */
                vfs_context_t ctx = vfs_context_current();
                /* Only go forward if you have write access */
                vfs_context_t ctx = vfs_context_current();

-               if(vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {
+               if (vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {

                        vnode_put(vp);
                        error = EBADF;
                        goto outdrop;
                }
                        vnode_put(vp);
                        error = EBADF;
                        goto outdrop;
                }

-               error = cp_vnode_setclass (vp, tmp);
+
+               struct vnode_attr va;
+
+               VATTR_INIT(&va);
+               VATTR_SET(&va, va_dataprotect_class, tmp);
+
+               error = VNOP_SETATTR(vp, &va, ctx);
+

                vnode_put(vp);
                break;
                vnode_put(vp);
                break;

-       }       
+       }

 
        case F_TRANSCODEKEY: {
 
        case F_TRANSCODEKEY: {

-

                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                }
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                }

-               
+

                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 
                if (vnode_getwithref(vp)) {
                        error = ENOENT;
                        goto outdrop;
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 
                if (vnode_getwithref(vp)) {
                        error = ENOENT;
                        goto outdrop;

-               }       
-               
-               error = cp_vnode_transcode (vp);
+               }
+
+               cp_key_t k = {
+                       .len = CP_MAX_WRAPPEDKEYSIZE,
+               };
+
+               MALLOC(k.key, char *, k.len, M_TEMP, M_WAITOK | M_ZERO);
+
+               error = VNOP_IOCTL(vp, F_TRANSCODEKEY, (caddr_t)&k, 1, &context);
+

                vnode_put(vp);
                vnode_put(vp);

+
+               if (error == 0) {
+                       error = copyout(k.key, argp, k.len);
+                       *retval = k.len;
+               }
+
+               FREE(k.key, M_TEMP);
+

                break;
                break;

-       }       
+       }

 
        case F_GETPROTECTIONLEVEL:  {
 
        case F_GETPROTECTIONLEVEL:  {

-               uint32_t cp_version = 0;
-

                if (fp->f_type != DTYPE_VNODE) {
                if (fp->f_type != DTYPE_VNODE) {

-                       error = EBADF; 
+                       error = EBADF;

                        goto out;
                }
 
                vp = (struct vnode*) fp->f_data;
                        goto out;
                }
 
                vp = (struct vnode*) fp->f_data;

-               proc_fdunlock (p);
+               proc_fdunlock(p);

 
                if (vnode_getwithref(vp)) {
                        error = ENOENT;
                        goto outdrop;
                }
 
 
                if (vnode_getwithref(vp)) {
                        error = ENOENT;
                        goto outdrop;
                }
 

-               /*
-                * if cp_get_major_vers fails, error will be set to proper errno 
-                * and cp_version will still be 0.
-                */
-
-               error = cp_get_root_major_vers (vp, &cp_version);
-               *retval = cp_version;
+               error = VNOP_IOCTL(vp, F_GETPROTECTIONLEVEL, (caddr_t)retval, 0, &context);

 
 

-               vnode_put (vp);
+               vnode_put(vp);

                break;
        }
 
        case F_GETDEFAULTPROTLEVEL:  {
                break;
        }
 
        case F_GETDEFAULTPROTLEVEL:  {

-               uint32_t cp_default = 0;
-

                if (fp->f_type != DTYPE_VNODE) {
                if (fp->f_type != DTYPE_VNODE) {

-                       error = EBADF; 
+                       error = EBADF;

                        goto out;
                }
 
                vp = (struct vnode*) fp->f_data;
                        goto out;
                }
 
                vp = (struct vnode*) fp->f_data;

-               proc_fdunlock (p);
+               proc_fdunlock(p);

 
                if (vnode_getwithref(vp)) {
                        error = ENOENT;
 
                if (vnode_getwithref(vp)) {
                        error = ENOENT;


@@ -1877,18 +2223,16 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                }
 
                /*
                }
 
                /*

-                * if cp_get_major_vers fails, error will be set to proper errno 
+                * if cp_get_major_vers fails, error will be set to proper errno

                 * and cp_version will still be 0.
                 */
 
                 * and cp_version will still be 0.
                 */
 

-               error = cp_get_default_level(vp, &cp_default);
-               *retval = cp_default;
+               error = VNOP_IOCTL(vp, F_GETDEFAULTPROTLEVEL, (caddr_t)retval, 0, &context);

 
 

-               vnode_put (vp);
+               vnode_put(vp);

                break;
        }
 
                break;
        }
 

-       

 #endif /* CONFIG_PROTECT */
 
        case F_MOVEDATAEXTENTS: {
 #endif /* CONFIG_PROTECT */
 
        case F_MOVEDATAEXTENTS: {


@@ -1897,24 +2241,32 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                struct vnode *dst_vp = NULLVP;
                /* We need to grab the 2nd FD out of the argments before moving on. */
                int fd2 = CAST_DOWN_EXPLICIT(int32_t, uap->arg);
                struct vnode *dst_vp = NULLVP;
                /* We need to grab the 2nd FD out of the argments before moving on. */
                int fd2 = CAST_DOWN_EXPLICIT(int32_t, uap->arg);

-               
+
+               error = priv_check_cred(kauth_cred_get(), PRIV_VFS_MOVE_DATA_EXTENTS, 0);
+               if (error) {
+                       goto out;
+               }
+

                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                }
 
                if (fp->f_type != DTYPE_VNODE) {
                        error = EBADF;
                        goto out;
                }
 

-               /* For now, special case HFS+ only, since this is SPI. */
+               /*
+                * For now, special case HFS+ and APFS only, since this
+                * is SPI.
+                */

                src_vp = (struct vnode *)fp->f_data;
                src_vp = (struct vnode *)fp->f_data;

-               if (src_vp->v_tag != VT_HFS) {
-                       error = EINVAL;
+               if (src_vp->v_tag != VT_HFS && src_vp->v_tag != VT_APFS) {
+                       error = ENOTSUP;

                        goto out;
                }
 
                /*
                        goto out;
                }
 
                /*

-                * Get the references before we start acquiring iocounts on the vnodes, 
+                * Get the references before we start acquiring iocounts on the vnodes,

                 * while we still hold the proc fd lock
                 */
                 * while we still hold the proc fd lock
                 */

-               if ( (error = fp_lookup(p, fd2, &fp2, 1)) ) {
+               if ((error = fp_lookup(p, fd2, &fp2, 1))) {

                        error = EBADF;
                        goto out;
                }
                        error = EBADF;
                        goto out;
                }


@@ -1924,9 +2276,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        goto out;
                }
                dst_vp = (struct vnode *)fp2->f_data;
                        goto out;
                }
                dst_vp = (struct vnode *)fp2->f_data;

-               if (dst_vp->v_tag != VT_HFS) {
+               if (dst_vp->v_tag != VT_HFS && dst_vp->v_tag != VT_APFS) {

                        fp_drop(p, fd2, fp2, 1);
                        fp_drop(p, fd2, fp2, 1);

-                       error = EINVAL;
+                       error = ENOTSUP;

                        goto out;
                }
 
                        goto out;
                }
 


@@ -1947,30 +2299,30 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        fp_drop(p, fd2, fp2, 0);
                        error = ENOENT;
                        goto outdrop;
                        fp_drop(p, fd2, fp2, 0);
                        error = ENOENT;
                        goto outdrop;

-               }       
+               }

                if (vnode_getwithref(dst_vp)) {
                if (vnode_getwithref(dst_vp)) {

-                       vnode_put (src_vp);
+                       vnode_put(src_vp);

                        fp_drop(p, fd2, fp2, 0);
                        error = ENOENT;
                        goto outdrop;
                        fp_drop(p, fd2, fp2, 0);
                        error = ENOENT;
                        goto outdrop;

-               }       
-               
-               /* 
+               }
+
+               /*

                 * Basic asserts; validate they are not the same and that
                 * both live on the same filesystem.
                 */
                if (dst_vp == src_vp) {
                 * Basic asserts; validate they are not the same and that
                 * both live on the same filesystem.
                 */
                if (dst_vp == src_vp) {

-                       vnode_put (src_vp);
-                       vnode_put (dst_vp);
-                       fp_drop (p, fd2, fp2, 0);
+                       vnode_put(src_vp);
+                       vnode_put(dst_vp);
+                       fp_drop(p, fd2, fp2, 0);

                        error = EINVAL;
                        goto outdrop;
                        error = EINVAL;
                        goto outdrop;

-               }       
+               }

 
                if (dst_vp->v_mount != src_vp->v_mount) {
 
                if (dst_vp->v_mount != src_vp->v_mount) {

-                       vnode_put (src_vp);
-                       vnode_put (dst_vp);
-                       fp_drop (p, fd2, fp2, 0);
+                       vnode_put(src_vp);
+                       vnode_put(dst_vp);
+                       fp_drop(p, fd2, fp2, 0);

                        error = EXDEV;
                        goto outdrop;
                }
                        error = EXDEV;
                        goto outdrop;
                }


@@ -1978,59 +2330,59 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                /* Now we have a legit pair of FDs.  Go to work */
 
                /* Now check for write access to the target files */
                /* Now we have a legit pair of FDs.  Go to work */
 
                /* Now check for write access to the target files */

-               if(vnode_authorize(src_vp, NULLVP, 
-                                                  (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), &context) != 0) {
+               if (vnode_authorize(src_vp, NULLVP,
+                   (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), &context) != 0) {

                        vnode_put(src_vp);
                        vnode_put(dst_vp);
                        fp_drop(p, fd2, fp2, 0);
                        error = EBADF;
                        goto outdrop;
                }
                        vnode_put(src_vp);
                        vnode_put(dst_vp);
                        fp_drop(p, fd2, fp2, 0);
                        error = EBADF;
                        goto outdrop;
                }

-               
-               if(vnode_authorize(dst_vp, NULLVP, 
-                                                  (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), &context) != 0) {
+
+               if (vnode_authorize(dst_vp, NULLVP,
+                   (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), &context) != 0) {

                        vnode_put(src_vp);
                        vnode_put(dst_vp);
                        fp_drop(p, fd2, fp2, 0);
                        error = EBADF;
                        goto outdrop;
                }
                        vnode_put(src_vp);
                        vnode_put(dst_vp);
                        fp_drop(p, fd2, fp2, 0);
                        error = EBADF;
                        goto outdrop;
                }

-                       
+

                /* Verify that both vps point to files and not directories */
                /* Verify that both vps point to files and not directories */

-               if ( !vnode_isreg(src_vp) || !vnode_isreg(dst_vp)) {
+               if (!vnode_isreg(src_vp) || !vnode_isreg(dst_vp)) {

                        error = EINVAL;
                        error = EINVAL;

-                       vnode_put (src_vp);
-                       vnode_put (dst_vp);
-                       fp_drop (p, fd2, fp2, 0);
+                       vnode_put(src_vp);
+                       vnode_put(dst_vp);
+                       fp_drop(p, fd2, fp2, 0);

                        goto outdrop;
                }
 
                        goto outdrop;
                }
 

-               /* 
+               /*

                 * The exchangedata syscall handler passes in 0 for the flags to VNOP_EXCHANGE.
                 * We'll pass in our special bit indicating that the new behavior is expected
                 */
                 * The exchangedata syscall handler passes in 0 for the flags to VNOP_EXCHANGE.
                 * We'll pass in our special bit indicating that the new behavior is expected
                 */

-               
+

                error = VNOP_EXCHANGE(src_vp, dst_vp, FSOPT_EXCHANGE_DATA_ONLY, &context);
                error = VNOP_EXCHANGE(src_vp, dst_vp, FSOPT_EXCHANGE_DATA_ONLY, &context);

-               
-               vnode_put (src_vp);
-               vnode_put (dst_vp);
+
+               vnode_put(src_vp);
+               vnode_put(dst_vp);

                fp_drop(p, fd2, fp2, 0);
                break;
        }
                fp_drop(p, fd2, fp2, 0);
                break;
        }

-       
-       /* 
+
+       /*

         * SPI for making a file compressed.
         */
        case F_MAKECOMPRESSED: {
                uint32_t gcounter = CAST_DOWN_EXPLICIT(uint32_t, uap->arg);
 
                if (fp->f_type != DTYPE_VNODE) {
         * SPI for making a file compressed.
         */
        case F_MAKECOMPRESSED: {
                uint32_t gcounter = CAST_DOWN_EXPLICIT(uint32_t, uap->arg);
 
                if (fp->f_type != DTYPE_VNODE) {

-                       error = EBADF; 
+                       error = EBADF;

                        goto out;
                }
 
                vp = (struct vnode*) fp->f_data;
                        goto out;
                }
 
                vp = (struct vnode*) fp->f_data;

-               proc_fdunlock (p);
+               proc_fdunlock(p);

 
                /* get the vnode */
                if (vnode_getwithref(vp)) {
 
                /* get the vnode */
                if (vnode_getwithref(vp)) {


@@ -2046,9 +2398,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                }
 
                /* invoke ioctl to pass off to FS */
                }
 
                /* invoke ioctl to pass off to FS */

-               /* Only go forward if you have write access */ 
+               /* Only go forward if you have write access */

                vfs_context_t ctx = vfs_context_current();
                vfs_context_t ctx = vfs_context_current();

-               if(vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {
+               if (vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {

                        vnode_put(vp);
                        error = EBADF;
                        goto outdrop;
                        vnode_put(vp);
                        error = EBADF;
                        goto outdrop;


@@ -2056,18 +2408,18 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 
                error = VNOP_IOCTL(vp, uap->cmd, (caddr_t)&gcounter, 0, &context);
 
 
                error = VNOP_IOCTL(vp, uap->cmd, (caddr_t)&gcounter, 0, &context);
 

-               vnode_put (vp);
-               break;                             
+               vnode_put(vp);
+               break;

        }
        }

-                       
+

        /*
         * SPI (private) for indicating to a filesystem that subsequent writes to
         * the open FD will written to the Fastflow.
         */
        case F_SET_GREEDY_MODE:
        /*
         * SPI (private) for indicating to a filesystem that subsequent writes to
         * the open FD will written to the Fastflow.
         */
        case F_SET_GREEDY_MODE:

-               /* intentionally drop through to the same handler as F_SETSTATIC.
-                * both fcntls should pass the argument and their selector into VNOP_IOCTL.
-                */
+       /* intentionally drop through to the same handler as F_SETSTATIC.
+        * both fcntls should pass the argument and their selector into VNOP_IOCTL.
+        */

 
        /*
         * SPI (private) for indicating to a filesystem that subsequent writes to
 
        /*
         * SPI (private) for indicating to a filesystem that subsequent writes to


@@ -2095,7 +2447,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 
                /* Only go forward if you have write access */
                vfs_context_t ctx = vfs_context_current();
 
                /* Only go forward if you have write access */
                vfs_context_t ctx = vfs_context_current();

-               if(vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {
+               if (vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {

                        vnode_put(vp);
                        error = EBADF;
                        goto outdrop;
                        vnode_put(vp);
                        error = EBADF;
                        goto outdrop;


@@ -2103,11 +2455,71 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 
                error = VNOP_IOCTL(vp, uap->cmd, ioctl_arg, 0, &context);
                (void)vnode_put(vp);
 
                error = VNOP_IOCTL(vp, uap->cmd, ioctl_arg, 0, &context);
                (void)vnode_put(vp);

-               
+
+               break;
+       }
+
+       /*
+        * SPI (private) for indicating to the lower level storage driver that the
+        * subsequent writes should be of a particular IO type (burst, greedy, static),
+        * or other flavors that may be necessary.
+        */
+       case F_SETIOTYPE: {
+               caddr_t param_ptr;
+               uint32_t param;
+
+               if (uap->arg) {
+                       /* extract 32 bits of flags from userland */
+                       param_ptr = (caddr_t) uap->arg;
+                       param = (uint32_t) param_ptr;
+               } else {
+                       /* If no argument is specified, error out */
+                       error = EINVAL;
+                       goto out;
+               }
+
+               /*
+                * Validate the different types of flags that can be specified:
+                * all of them are mutually exclusive for now.
+                */
+               switch (param) {
+               case F_IOTYPE_ISOCHRONOUS:
+                       break;
+
+               default:
+                       error = EINVAL;
+                       goto out;
+               }
+
+
+               if (fp->f_type != DTYPE_VNODE) {
+                       error = EBADF;
+                       goto out;
+               }
+               vp = (struct vnode *)fp->f_data;
+               proc_fdunlock(p);
+
+               error = vnode_getwithref(vp);
+               if (error) {
+                       error = ENOENT;
+                       goto outdrop;
+               }
+
+               /* Only go forward if you have write access */
+               vfs_context_t ctx = vfs_context_current();
+               if (vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {
+                       vnode_put(vp);
+                       error = EBADF;
+                       goto outdrop;
+               }
+
+               error = VNOP_IOCTL(vp, uap->cmd, param_ptr, 0, &context);
+               (void)vnode_put(vp);
+

                break;
        }
                break;
        }

-       
-       /* 
+
+       /*

         * Set the vnode pointed to by 'fd'
         * and tag it as the (potentially future) backing store
         * for another filesystem
         * Set the vnode pointed to by 'fd'
         * and tag it as the (potentially future) backing store
         * for another filesystem


@@ -2117,7 +2529,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = EBADF;
                        goto out;
                }
                        error = EBADF;
                        goto out;
                }

-               
+

                vp = (struct vnode *)fp->f_data;
 
                if (vp->v_tag != VT_HFS) {
                vp = (struct vnode *)fp->f_data;
 
                if (vp->v_tag != VT_HFS) {


@@ -2130,29 +2542,28 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = ENOENT;
                        goto outdrop;
                }
                        error = ENOENT;
                        goto outdrop;
                }

-               
+

                /* only proceed if you have write access */
                vfs_context_t ctx = vfs_context_current();
                /* only proceed if you have write access */
                vfs_context_t ctx = vfs_context_current();

-               if(vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {
+               if (vnode_authorize(vp, NULLVP, (KAUTH_VNODE_ACCESS | KAUTH_VNODE_WRITE_DATA), ctx) != 0) {

                        vnode_put(vp);
                        error = EBADF;
                        goto outdrop;
                }
 
                        vnode_put(vp);
                        error = EBADF;
                        goto outdrop;
                }
 

-               
+

                /* If arg != 0, set, otherwise unset */
                if (uap->arg) {
                /* If arg != 0, set, otherwise unset */
                if (uap->arg) {

-                       error = VNOP_IOCTL (vp, uap->cmd, (caddr_t)1, 0, &context);
-               }
-               else {
-                       error = VNOP_IOCTL (vp, uap->cmd, (caddr_t)NULL, 0, &context);
+                       error = VNOP_IOCTL(vp, uap->cmd, (caddr_t)1, 0, &context);
+               } else {
+                       error = VNOP_IOCTL(vp, uap->cmd, (caddr_t)NULL, 0, &context);

                }
                }

-               
+

                vnode_put(vp);
                break;
        }
 
                vnode_put(vp);
                break;
        }
 

-       /* 
+       /*

         * like F_GETPATH, but special semantics for
         * the mobile time machine handler.
         */
         * like F_GETPATH, but special semantics for
         * the mobile time machine handler.
         */


@@ -2173,13 +2584,13 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        error = ENOMEM;
                        goto outdrop;
                }
                        error = ENOMEM;
                        goto outdrop;
                }

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
+               if ((error = vnode_getwithref(vp)) == 0) {

                        int backingstore = 0;
                        int backingstore = 0;

-                       
+

                        /* Check for error from vn_getpath before moving on */
                        if ((error = vn_getpath(vp, pathbufp, &pathlen)) == 0) {
                                if (vp->v_tag == VT_HFS) {
                        /* Check for error from vn_getpath before moving on */
                        if ((error = vn_getpath(vp, pathbufp, &pathlen)) == 0) {
                                if (vp->v_tag == VT_HFS) {

-                                       error = VNOP_IOCTL (vp, uap->cmd, (caddr_t) &backingstore, 0, &context);
+                                       error = VNOP_IOCTL(vp, uap->cmd, (caddr_t) &backingstore, 0, &context);

                                }
                                (void)vnode_put(vp);
 
                                }
                                (void)vnode_put(vp);
 


@@ -2187,7 +2598,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                                        error = copyout((caddr_t)pathbufp, argp, pathlen);
                                }
                                if (error == 0) {
                                        error = copyout((caddr_t)pathbufp, argp, pathlen);
                                }
                                if (error == 0) {

-                                       /* 
+                                       /*

                                         * If the copyout was successful, now check to ensure
                                         * that this vnode is not a BACKINGSTORE vnode.  mtmd
                                         * wants the path regardless.
                                         * If the copyout was successful, now check to ensure
                                         * that this vnode is not a BACKINGSTORE vnode.  mtmd
                                         * wants the path regardless.


@@ -2196,13 +2607,27 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                                                error = EBUSY;
                                        }
                                }
                                                error = EBUSY;
                                        }
                                }

-                       } else
+                       } else {

                                (void)vnode_put(vp);
                                (void)vnode_put(vp);

+                       }

                }
                FREE(pathbufp, M_TEMP);
                goto outdrop;
        }
 
                }
                FREE(pathbufp, M_TEMP);
                goto outdrop;
        }
 

+#if DEBUG || DEVELOPMENT
+       case F_RECYCLE:
+               if (fp->f_type != DTYPE_VNODE) {
+                       error = EBADF;
+                       goto out;
+               }
+               vp = (struct vnode *)fp->f_data;
+               proc_fdunlock(p);
+
+               vnode_recycle(vp);
+               break;
+#endif
+

        default:
                /*
                 * This is an fcntl() that we d not recognize at this level;
        default:
                /*
                 * This is an fcntl() that we d not recognize at this level;


@@ -2210,18 +2635,26 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                 * for this vnode; this can include special devices, and will
                 * effectively overload fcntl() to send ioctl()'s.
                 */
                 * for this vnode; this can include special devices, and will
                 * effectively overload fcntl() to send ioctl()'s.
                 */

-               if((uap->cmd & IOC_VOID) && (uap->cmd & IOC_INOUT)){
-            error = EINVAL;
+               if ((uap->cmd & IOC_VOID) && (uap->cmd & IOC_INOUT)) {
+                       error = EINVAL;

                        goto out;
                }
                        goto out;
                }

-               
+

                /* Catch any now-invalid fcntl() selectors */
                switch (uap->cmd) {
                /* Catch any now-invalid fcntl() selectors */
                switch (uap->cmd) {

-                       case F_MARKDEPENDENCY:
-                               error = EINVAL;
-                               goto out;
-                       default:
-                               break;
+               case (int)APFSIOC_REVERT_TO_SNAPSHOT:
+               case (int)FSIOC_FIOSEEKHOLE:
+               case (int)FSIOC_FIOSEEKDATA:
+               case (int)FSIOC_CAS_BSDFLAGS:
+               case HFS_GET_BOOT_INFO:
+               case HFS_SET_BOOT_INFO:
+               case FIOPINSWAP:
+               case F_MARKDEPENDENCY:
+               case TIOCREVOKE:
+                       error = EINVAL;
+                       goto out;
+               default:
+                       break;

                }
 
                if (fp->f_type != DTYPE_VNODE) {
                }
 
                if (fp->f_type != DTYPE_VNODE) {


@@ -2231,9 +2664,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 
                vp = (struct vnode *)fp->f_data;
                proc_fdunlock(p);
 

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
+               if ((error = vnode_getwithref(vp)) == 0) {

 #define STK_PARAMS 128
 #define STK_PARAMS 128

-                       char stkbuf[STK_PARAMS];
+                       char stkbuf[STK_PARAMS] = {0};

                        unsigned int size;
                        caddr_t data, memp;
                        /*
                        unsigned int size;
                        caddr_t data, memp;
                        /*


@@ -2254,7 +2687,7 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        }
 
                        memp = NULL;
                        }
 
                        memp = NULL;

-                       if (size > sizeof (stkbuf)) {
+                       if (size > sizeof(stkbuf)) {

                                if ((memp = (caddr_t)kalloc(size)) == 0) {
                                        (void)vnode_put(vp);
                                        error = ENOMEM;
                                if ((memp = (caddr_t)kalloc(size)) == 0) {
                                        (void)vnode_put(vp);
                                        error = ENOMEM;


@@ -2264,21 +2697,22 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        } else {
                                data = &stkbuf[0];
                        }
                        } else {
                                data = &stkbuf[0];
                        }

-                       
+

                        if (uap->cmd & IOC_IN) {
                                if (size) {
                                        /* structure */
                                        error = copyin(argp, data, size);
                                        if (error) {
                                                (void)vnode_put(vp);
                        if (uap->cmd & IOC_IN) {
                                if (size) {
                                        /* structure */
                                        error = copyin(argp, data, size);
                                        if (error) {
                                                (void)vnode_put(vp);

-                                               if (memp)
+                                               if (memp) {

                                                        kfree(memp, size);
                                                        kfree(memp, size);

+                                               }

                                                goto outdrop;
                                        }
 
                                        /* Bzero the section beyond that which was needed */
                                        if (size <= sizeof(stkbuf)) {
                                                goto outdrop;
                                        }
 
                                        /* Bzero the section beyond that which was needed */
                                        if (size <= sizeof(stkbuf)) {

-                                               bzero ( (((uint8_t*)data) + size), (sizeof(stkbuf) - size));
+                                               bzero((((uint8_t*)data) + size), (sizeof(stkbuf) - size));

                                        }
                                } else {
                                        /* int */
                                        }
                                } else {
                                        /* int */


@@ -2296,9 +2730,9 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                                bzero(data, size);
                        } else if (uap->cmd & IOC_VOID) {
                                if (is64bit) {
                                bzero(data, size);
                        } else if (uap->cmd & IOC_VOID) {
                                if (is64bit) {

-                                   *(user_addr_t *)data = argp;
+                                       *(user_addr_t *)data = argp;

                                } else {
                                } else {

-                                   *(uint32_t *)data = (uint32_t)argp;
+                                       *(uint32_t *)data = (uint32_t)argp;

                                }
                        }
 
                                }
                        }
 


@@ -2307,10 +2741,12 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
                        (void)vnode_put(vp);
 
                        /* Copy any output data to user */
                        (void)vnode_put(vp);
 
                        /* Copy any output data to user */

-                       if (error == 0 && (uap->cmd & IOC_OUT) && size) 
+                       if (error == 0 && (uap->cmd & IOC_OUT) && size) {

                                error = copyout(data, argp, size);
                                error = copyout(data, argp, size);

-                       if (memp)
+                       }
+                       if (memp) {

                                kfree(memp, size);
                                kfree(memp, size);

+                       }

                }
                break;
        }
                }
                break;
        }


@@ -2318,11 +2754,11 @@ fcntl_nocancel(proc_t p, struct fcntl_nocancel_args *uap, int32_t *retval)
 outdrop:
        AUDIT_ARG(vnpath_withref, vp, ARG_VNODE1);
        fp_drop(p, fd, fp, 0);
 outdrop:
        AUDIT_ARG(vnpath_withref, vp, ARG_VNODE1);
        fp_drop(p, fd, fp, 0);

-       return(error);
+       return error;

 out:
        fp_drop(p, fd, fp, 1);
        proc_fdunlock(p);
 out:
        fp_drop(p, fd, fp, 1);
        proc_fdunlock(p);

-       return(error);
+       return error;

 }
 
 
 }
 
 


@@ -2367,7 +2803,7 @@ finishdup(proc_t p,
        if ((ofp = fdp->fd_ofiles[old]) == NULL ||
            (fdp->fd_ofileflags[old] & UF_RESERVED)) {
                fdrelse(p, new);
        if ((ofp = fdp->fd_ofiles[old]) == NULL ||
            (fdp->fd_ofileflags[old] & UF_RESERVED)) {
                fdrelse(p, new);

-               return (EBADF);
+               return EBADF;

        }
        fg_ref(ofp);
 
        }
        fg_ref(ofp);
 


@@ -2376,7 +2812,7 @@ finishdup(proc_t p,
        if (error) {
                fg_drop(ofp);
                fdrelse(p, new);
        if (error) {
                fg_drop(ofp);
                fdrelse(p, new);

-               return (error);
+               return error;

        }
 #endif
 
        }
 #endif
 


@@ -2389,24 +2825,27 @@ finishdup(proc_t p,
        if (nfp == NULL) {
                fg_drop(ofp);
                fdrelse(p, new);
        if (nfp == NULL) {
                fg_drop(ofp);
                fdrelse(p, new);

-               return (ENOMEM);
+               return ENOMEM;

        }
 
        nfp->f_fglob = ofp->f_fglob;
 
 #if DIAGNOSTIC
        }
 
        nfp->f_fglob = ofp->f_fglob;
 
 #if DIAGNOSTIC

-       if (fdp->fd_ofiles[new] != 0)
+       if (fdp->fd_ofiles[new] != 0) {

                panic("finishdup: overwriting fd_ofiles with new %d", new);
                panic("finishdup: overwriting fd_ofiles with new %d", new);

-       if ((fdp->fd_ofileflags[new] & UF_RESERVED) == 0)
+       }
+       if ((fdp->fd_ofileflags[new] & UF_RESERVED) == 0) {

                panic("finishdup: unreserved fileflags with new %d", new);
                panic("finishdup: unreserved fileflags with new %d", new);

+       }

 #endif
 
 #endif
 

-       if (new > fdp->fd_lastfile)
+       if (new > fdp->fd_lastfile) {

                fdp->fd_lastfile = new;
                fdp->fd_lastfile = new;

+       }

        *fdflags(p, new) |= fd_flags;
        procfdtbl_releasefd(p, new, nfp);
        *retval = new;
        *fdflags(p, new) |= fd_flags;
        procfdtbl_releasefd(p, new, nfp);
        *retval = new;

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -2424,14 +2863,14 @@ finishdup(proc_t p,
  *     fp_lookup:EBADF                 Bad file descriptor
  *      fp_guard_exception:???          Guarded file descriptor
  *     close_internal:EBADF
  *     fp_lookup:EBADF                 Bad file descriptor
  *      fp_guard_exception:???          Guarded file descriptor
  *     close_internal:EBADF

- *     close_internal:???              Anything returnable by a per-fileops
+ *     close_internal:???              Anything returnable by a per-fileops

  *                                     close function
  */
 int
 close(proc_t p, struct close_args *uap, int32_t *retval)
 {
        __pthread_testcancel(1);
  *                                     close function
  */
 int
 close(proc_t p, struct close_args *uap, int32_t *retval)
 {
        __pthread_testcancel(1);

-       return(close_nocancel(p, (struct close_nocancel_args *)uap, retval));
+       return close_nocancel(p, (struct close_nocancel_args *)uap, retval);

 }
 
 
 }
 
 


@@ -2446,23 +2885,23 @@ close_nocancel(proc_t p, struct close_nocancel_args *uap, __unused int32_t *retv
 
        proc_fdlock(p);
 
 
        proc_fdlock(p);
 

-       if ( (error = fp_lookup(p,fd,&fp, 1)) ) {
+       if ((error = fp_lookup(p, fd, &fp, 1))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(error);
+               return error;

        }
 
        if (FP_ISGUARDED(fp, GUARD_CLOSE)) {
                error = fp_guard_exception(p, fd, fp, kGUARD_EXC_CLOSE);
                (void) fp_drop(p, fd, fp, 1);
                proc_fdunlock(p);
        }
 
        if (FP_ISGUARDED(fp, GUARD_CLOSE)) {
                error = fp_guard_exception(p, fd, fp, kGUARD_EXC_CLOSE);
                (void) fp_drop(p, fd, fp, 1);
                proc_fdunlock(p);

-               return (error);
+               return error;

        }
 
        error = close_internal_locked(p, fd, fp, 0);
 
        proc_fdunlock(p);
 
        }
 
        error = close_internal_locked(p, fd, fp, 0);
 
        proc_fdunlock(p);
 

-       return (error);
+       return error;

 }
 
 
 }
 
 


@@ -2478,7 +2917,7 @@ close_nocancel(proc_t p, struct close_nocancel_args *uap, __unused int32_t *retv
  *
  * Returns:    0                       Success
  *             EBADF                   fd already in close wait state
  *
  * Returns:    0                       Success
  *             EBADF                   fd already in close wait state

- *     closef_locked:???               Anything returnable by a per-fileops
+ *     closef_locked:???               Anything returnable by a per-fileops

  *                                     close function
  *
  * Locks:      Assumes proc_fdlock for process is held by the caller and returns
  *                                     close function
  *
  * Locks:      Assumes proc_fdlock for process is held by the caller and returns


@@ -2492,7 +2931,7 @@ int
 close_internal_locked(proc_t p, int fd, struct fileproc *fp, int flags)
 {
        struct filedesc *fdp = p->p_fd;
 close_internal_locked(proc_t p, int fd, struct fileproc *fp, int flags)
 {
        struct filedesc *fdp = p->p_fd;

-       int error =0;
+       int error = 0;

        int resvfd = flags & FD_DUP2RESV;
 
 
        int resvfd = flags & FD_DUP2RESV;
 
 


@@ -2510,72 +2949,85 @@ close_internal_locked(proc_t p, int fd, struct fileproc *fp, int flags)
 
 
 #if DIAGNOSTIC
 
 
 #if DIAGNOSTIC

-       if ((fdp->fd_ofileflags[fd] & UF_RESERVED) == 0)
+       if ((fdp->fd_ofileflags[fd] & UF_RESERVED) == 0) {

                panic("close_internal: unreserved fileflags with fd %d", fd);
                panic("close_internal: unreserved fileflags with fd %d", fd);

+       }

 #endif
 
        fp->f_flags |= FP_CLOSING;
 
 #endif
 
        fp->f_flags |= FP_CLOSING;
 

-       if ( (fp->f_flags & FP_AIOISSUED) || kauth_authorize_fileop_has_listeners() ) {
-
-               proc_fdunlock(p);
+       if ((fp->f_flags & FP_AIOISSUED) || kauth_authorize_fileop_has_listeners()) {
+               proc_fdunlock(p);

 
 

-               if ( (fp->f_type == DTYPE_VNODE) && kauth_authorize_fileop_has_listeners() ) {
-                       /*
-                        * call out to allow 3rd party notification of close. 
+               if ((fp->f_type == DTYPE_VNODE) && kauth_authorize_fileop_has_listeners()) {
+                       /*
+                        * call out to allow 3rd party notification of close.

                         * Ignore result of kauth_authorize_fileop call.
                         */
                         * Ignore result of kauth_authorize_fileop call.
                         */

-                       if (vnode_getwithref((vnode_t)fp->f_data) == 0) {
-                               u_int   fileop_flags = 0;
-                               if ((fp->f_flags & FP_WRITTEN) != 0)
-                                       fileop_flags |= KAUTH_FILEOP_CLOSE_MODIFIED;
-                               kauth_authorize_fileop(fp->f_fglob->fg_cred, KAUTH_FILEOP_CLOSE, 
-                                                      (uintptr_t)fp->f_data, (uintptr_t)fileop_flags);
+                       if (vnode_getwithref((vnode_t)fp->f_data) == 0) {
+                               u_int   fileop_flags = 0;
+                               if ((fp->f_flags & FP_WRITTEN) != 0) {
+                                       fileop_flags |= KAUTH_FILEOP_CLOSE_MODIFIED;
+                               }
+                               kauth_authorize_fileop(fp->f_fglob->fg_cred, KAUTH_FILEOP_CLOSE,
+                                   (uintptr_t)fp->f_data, (uintptr_t)fileop_flags);

                                vnode_put((vnode_t)fp->f_data);
                        }
                }
                                vnode_put((vnode_t)fp->f_data);
                        }
                }

-               if (fp->f_flags & FP_AIOISSUED)
-                       /*
+               if (fp->f_flags & FP_AIOISSUED) {
+                       /*

                         * cancel all async IO requests that can be cancelled.
                         */
                         * cancel all async IO requests that can be cancelled.
                         */

-                       _aio_close( p, fd );
+                       _aio_close( p, fd );
+               }

 
                proc_fdlock(p);
        }
 
 
                proc_fdlock(p);
        }
 

-       if (fd < fdp->fd_knlistsize)
+       if (fd < fdp->fd_knlistsize) {

                knote_fdclose(p, fd);
                knote_fdclose(p, fd);

+       }

 
 

-       if (fp->f_flags & FP_WAITEVENT) 
-               (void)waitevent_close(p, fp);
-
+       /* release the ref returned from fp_lookup before calling drain */
+       (void) os_ref_release_locked(&fp->f_iocount);

        fileproc_drain(p, fp);
 
        fileproc_drain(p, fp);
 

+       if (fp->f_flags & FP_WAITEVENT) {
+               (void)waitevent_close(p, fp);
+       }
+

        if (resvfd == 0) {
                _fdrelse(p, fd);
        } else {
                procfdtbl_reservefd(p, fd);
        }
 
        if (resvfd == 0) {
                _fdrelse(p, fd);
        } else {
                procfdtbl_reservefd(p, fd);
        }
 

+       if (ENTR_SHOULDTRACE && fp->f_type == DTYPE_SOCKET) {
+               KERNEL_ENERGYTRACE(kEnTrActKernSocket, DBG_FUNC_END,
+                   fd, 0, (int64_t)VM_KERNEL_ADDRPERM(fp->f_data));
+       }
+

        error = closef_locked(fp, fp->f_fglob, p);
        error = closef_locked(fp, fp->f_fglob, p);

-       if ((fp->f_flags & FP_WAITCLOSE) == FP_WAITCLOSE)
+       if ((fp->f_flags & FP_WAITCLOSE) == FP_WAITCLOSE) {

                wakeup(&fp->f_flags);
                wakeup(&fp->f_flags);

+       }

        fp->f_flags &= ~(FP_WAITCLOSE | FP_CLOSING);
 
        proc_fdunlock(p);
 
        fp->f_flags &= ~(FP_WAITCLOSE | FP_CLOSING);
 
        proc_fdunlock(p);
 

-       fileproc_free(fp);      
+       fileproc_free(fp);

 
        proc_fdlock(p);
 
 #if DIAGNOSTIC
        if (resvfd != 0) {
 
        proc_fdlock(p);
 
 #if DIAGNOSTIC
        if (resvfd != 0) {

-               if ((fdp->fd_ofileflags[fd] & UF_RESERVED) == 0)
+               if ((fdp->fd_ofileflags[fd] & UF_RESERVED) == 0) {

                        panic("close with reserved fd returns with freed fd:%d: proc: %p", fd, p);
                        panic("close with reserved fd returns with freed fd:%d: proc: %p", fd, p);

+               }

        }
 #endif
 
        }
 #endif
 

-       return(error);
+       return error;

 }
 
 
 }
 
 


@@ -2628,7 +3080,6 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
                struct user32_stat64 user32_sb64;
        } dest;
        int error, my_size;
                struct user32_stat64 user32_sb64;
        } dest;
        int error, my_size;

-       int funnel_state;

        file_type_t type;
        caddr_t data;
        kauth_filesec_t fsec;
        file_type_t type;
        caddr_t data;
        kauth_filesec_t fsec;


@@ -2640,7 +3091,7 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
        AUDIT_ARG(fd, fd);
 
        if ((error = fp_lookup(p, fd, &fp, 0)) != 0) {
        AUDIT_ARG(fd, fd);
 
        if ((error = fp_lookup(p, fd, &fp, 0)) != 0) {

-               return(error);
+               return error;

        }
        type = fp->f_type;
        data = fp->f_data;
        }
        type = fp->f_type;
        data = fp->f_data;


@@ -2649,7 +3100,6 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
        sbptr = (void *)&source;
 
        switch (type) {
        sbptr = (void *)&source;
 
        switch (type) {

-

        case DTYPE_VNODE:
                if ((error = vnode_getwithref((vnode_t)data)) == 0) {
                        /*
        case DTYPE_VNODE:
                if ((error = vnode_getwithref((vnode_t)data)) == 0) {
                        /*


@@ -2658,9 +3108,10 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
                         * going to let them get the basic stat information.
                         */
                        if (xsecurity == USER_ADDR_NULL) {
                         * going to let them get the basic stat information.
                         */
                        if (xsecurity == USER_ADDR_NULL) {

-                               error = vn_stat_noauth((vnode_t)data, sbptr, NULL, isstat64, ctx);
+                               error = vn_stat_noauth((vnode_t)data, sbptr, NULL, isstat64, 0, ctx,
+                                   fp->f_fglob->fg_cred);

                        } else {
                        } else {

-                               error = vn_stat((vnode_t)data, sbptr, &fsec, isstat64, ctx);
+                               error = vn_stat((vnode_t)data, sbptr, &fsec, isstat64, 0, ctx);

                        }
 
                        AUDIT_ARG(vnpath, (struct vnode *)data, ARG_VNODE1);
                        }
 
                        AUDIT_ARG(vnpath, (struct vnode *)data, ARG_VNODE1);


@@ -2683,9 +3134,7 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
                break;
 
        case DTYPE_KQUEUE:
                break;
 
        case DTYPE_KQUEUE:

-               funnel_state = thread_funnel_set(kernel_flock, TRUE);
-               error = kqueue_stat(fp, sbptr, isstat64, p);
-               thread_funnel_set(kernel_flock, funnel_state);
+               error = kqueue_stat((void *)data, sbptr, isstat64, p);

                break;
 
        default:
                break;
 
        default:


@@ -2701,11 +3150,11 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
                        source.sb64.st_qspare[1] = 0LL;
 
                        if (IS_64BIT_PROCESS(current_proc())) {
                        source.sb64.st_qspare[1] = 0LL;
 
                        if (IS_64BIT_PROCESS(current_proc())) {

-                               munge_user64_stat64(&source.sb64, &dest.user64_sb64); 
+                               munge_user64_stat64(&source.sb64, &dest.user64_sb64);

                                my_size = sizeof(dest.user64_sb64);
                                sbp = (caddr_t)&dest.user64_sb64;
                        } else {
                                my_size = sizeof(dest.user64_sb64);
                                sbp = (caddr_t)&dest.user64_sb64;
                        } else {

-                               munge_user32_stat64(&source.sb64, &dest.user32_sb64); 
+                               munge_user32_stat64(&source.sb64, &dest.user32_sb64);

                                my_size = sizeof(dest.user32_sb64);
                                sbp = (caddr_t)&dest.user32_sb64;
                        }
                                my_size = sizeof(dest.user32_sb64);
                                sbp = (caddr_t)&dest.user32_sb64;
                        }


@@ -2714,11 +3163,11 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
                        source.sb.st_qspare[0] = 0LL;
                        source.sb.st_qspare[1] = 0LL;
                        if (IS_64BIT_PROCESS(current_proc())) {
                        source.sb.st_qspare[0] = 0LL;
                        source.sb.st_qspare[1] = 0LL;
                        if (IS_64BIT_PROCESS(current_proc())) {

-                               munge_user64_stat(&source.sb, &dest.user64_sb); 
+                               munge_user64_stat(&source.sb, &dest.user64_sb);

                                my_size = sizeof(dest.user64_sb);
                                sbp = (caddr_t)&dest.user64_sb;
                        } else {
                                my_size = sizeof(dest.user64_sb);
                                sbp = (caddr_t)&dest.user64_sb;
                        } else {

-                               munge_user32_stat(&source.sb, &dest.user32_sb); 
+                               munge_user32_stat(&source.sb, &dest.user32_sb);

                                my_size = sizeof(dest.user32_sb);
                                sbp = (caddr_t)&dest.user32_sb;
                        }
                                my_size = sizeof(dest.user32_sb);
                                sbp = (caddr_t)&dest.user32_sb;
                        }


@@ -2729,9 +3178,8 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
 
        /* caller wants extended security information? */
        if (xsecurity != USER_ADDR_NULL) {
 
        /* caller wants extended security information? */
        if (xsecurity != USER_ADDR_NULL) {

-

                /* did we get any? */
                /* did we get any? */

-                if (fsec == KAUTH_FILESEC_NONE) {
+               if (fsec == KAUTH_FILESEC_NONE) {

                        if (susize(xsecurity_size, 0) != 0) {
                                error = EFAULT;
                                goto out;
                        if (susize(xsecurity_size, 0) != 0) {
                                error = EFAULT;
                                goto out;


@@ -2747,15 +3195,17 @@ fstat1(proc_t p, int fd, user_addr_t ub, user_addr_t xsecurity, user_addr_t xsec
                        }
 
                        /* if the caller supplied enough room, copy out to it */
                        }
 
                        /* if the caller supplied enough room, copy out to it */

-                       if (xsecurity_bufsize >= KAUTH_FILESEC_COPYSIZE(fsec))
+                       if (xsecurity_bufsize >= KAUTH_FILESEC_COPYSIZE(fsec)) {

                                error = copyout(fsec, xsecurity, KAUTH_FILESEC_COPYSIZE(fsec));
                                error = copyout(fsec, xsecurity, KAUTH_FILESEC_COPYSIZE(fsec));

+                       }

                }
        }
 out:
        fp_drop(p, fd, fp, 0);
                }
        }
 out:
        fp_drop(p, fd, fp, 0);

-       if (fsec != NULL)
+       if (fsec != NULL) {

                kauth_filesec_free(fsec);
                kauth_filesec_free(fsec);

-       return (error);
+       }
+       return error;

 }
 
 
 }
 
 


@@ -2778,9 +3228,9 @@ out:
 int
 fstat_extended(proc_t p, struct fstat_extended_args *uap, __unused int32_t *retval)
 {
 int
 fstat_extended(proc_t p, struct fstat_extended_args *uap, __unused int32_t *retval)
 {

-       return(fstat1(p, uap->fd, uap->ub, uap->xsecurity, uap->xsecurity_size, 0));
+       return fstat1(p, uap->fd, uap->ub, uap->xsecurity, uap->xsecurity_size, 0);

 }
 }

- 
+

 
 /*
  * fstat
 
 /*
  * fstat


@@ -2795,9 +3245,9 @@ fstat_extended(proc_t p, struct fstat_extended_args *uap, __unused int32_t *retv
  *             !0                              Errno (see fstat1)
  */
 int
  *             !0                              Errno (see fstat1)
  */
 int

-fstat(proc_t p, register struct fstat_args *uap, __unused int32_t *retval)
+fstat(proc_t p, struct fstat_args *uap, __unused int32_t *retval)

 {
 {

-       return(fstat1(p, uap->fd, uap->ub, 0, 0, 0));
+       return fstat1(p, uap->fd, uap->ub, 0, 0, 0);

 }
 
 
 }
 
 


@@ -2820,9 +3270,9 @@ fstat(proc_t p, register struct fstat_args *uap, __unused int32_t *retval)
 int
 fstat64_extended(proc_t p, struct fstat64_extended_args *uap, __unused int32_t *retval)
 {
 int
 fstat64_extended(proc_t p, struct fstat64_extended_args *uap, __unused int32_t *retval)
 {

-       return(fstat1(p, uap->fd, uap->ub, uap->xsecurity, uap->xsecurity_size, 1));
+       return fstat1(p, uap->fd, uap->ub, uap->xsecurity, uap->xsecurity_size, 1);

 }
 }

- 
+

 
 /*
  * fstat64
 
 /*
  * fstat64


@@ -2838,9 +3288,9 @@ fstat64_extended(proc_t p, struct fstat64_extended_args *uap, __unused int32_t *
  *             !0                              Errno (see fstat1)
  */
 int
  *             !0                              Errno (see fstat1)
  */
 int

-fstat64(proc_t p, register struct fstat64_args *uap, __unused int32_t *retval)
+fstat64(proc_t p, struct fstat64_args *uap, __unused int32_t *retval)

 {
 {

-       return(fstat1(p, uap->fd, uap->ub, 0, 0, 1));
+       return fstat1(p, uap->fd, uap->ub, 0, 0, 1);

 }
 
 
 }
 
 


@@ -2875,13 +3325,13 @@ fpathconf(proc_t p, struct fpathconf_args *uap, int32_t *retval)
 
 
        AUDIT_ARG(fd, uap->fd);
 
 
        AUDIT_ARG(fd, uap->fd);

-       if ( (error = fp_lookup(p, fd, &fp, 0)) )
-               return(error);
+       if ((error = fp_lookup(p, fd, &fp, 0))) {
+               return error;
+       }

        type = fp->f_type;
        data = fp->f_data;
 
        switch (type) {
        type = fp->f_type;
        data = fp->f_data;
 
        switch (type) {

-

        case DTYPE_SOCKET:
                if (uap->name != _PC_PIPE_BUF) {
                        error = EINVAL;
        case DTYPE_SOCKET:
                if (uap->name != _PC_PIPE_BUF) {
                        error = EINVAL;


@@ -2903,8 +3353,8 @@ fpathconf(proc_t p, struct fpathconf_args *uap, int32_t *retval)
        case DTYPE_VNODE:
                vp = (struct vnode *)data;
 
        case DTYPE_VNODE:
                vp = (struct vnode *)data;
 

-               if ( (error = vnode_getwithref(vp)) == 0) {
-                       AUDIT_ARG(vnpath, vp, ARG_VNODE1);
+               if ((error = vnode_getwithref(vp)) == 0) {
+                       AUDIT_ARG(vnpath, vp, ARG_VNODE1);

 
                        error = vn_pathconf(vp, uap->name, retval, vfs_context_current());
 
 
                        error = vn_pathconf(vp, uap->name, retval, vfs_context_current());
 


@@ -2915,12 +3365,11 @@ fpathconf(proc_t p, struct fpathconf_args *uap, int32_t *retval)
        default:
                error = EINVAL;
                goto out;
        default:
                error = EINVAL;
                goto out;

-

        }
        /*NOTREACHED*/
 out:
        fp_drop(p, fd, fp, 0);
        }
        /*NOTREACHED*/
 out:
        fp_drop(p, fd, fp, 0);

-       return(error);
+       return error;

 }
 
 /*
 }
 
 /*


@@ -2969,38 +3418,44 @@ fdalloc(proc_t p, int want, int *result)
        lim = min((int)p->p_rlimit[RLIMIT_NOFILE].rlim_cur, maxfiles);
        for (;;) {
                last = min(fdp->fd_nfiles, lim);
        lim = min((int)p->p_rlimit[RLIMIT_NOFILE].rlim_cur, maxfiles);
        for (;;) {
                last = min(fdp->fd_nfiles, lim);

-               if ((i = want) < fdp->fd_freefile)
+               if ((i = want) < fdp->fd_freefile) {

                        i = fdp->fd_freefile;
                        i = fdp->fd_freefile;

+               }

                for (; i < last; i++) {
                        if (fdp->fd_ofiles[i] == NULL && !(fdp->fd_ofileflags[i] & UF_RESERVED)) {
                                procfdtbl_reservefd(p, i);
                for (; i < last; i++) {
                        if (fdp->fd_ofiles[i] == NULL && !(fdp->fd_ofileflags[i] & UF_RESERVED)) {
                                procfdtbl_reservefd(p, i);

-                               if (i > fdp->fd_lastfile)
+                               if (i > fdp->fd_lastfile) {

                                        fdp->fd_lastfile = i;
                                        fdp->fd_lastfile = i;

-                               if (want <= fdp->fd_freefile)
+                               }
+                               if (want <= fdp->fd_freefile) {

                                        fdp->fd_freefile = i;
                                        fdp->fd_freefile = i;

+                               }

                                *result = i;
                                *result = i;

-                               return (0);
+                               return 0;

                        }
                }
 
                /*
                 * No space in current array.  Expand?
                 */
                        }
                }
 
                /*
                 * No space in current array.  Expand?
                 */

-               if (fdp->fd_nfiles >= lim)
-                       return (EMFILE);
-               if (fdp->fd_nfiles < NDEXTENT)
+               if (fdp->fd_nfiles >= lim) {
+                       return EMFILE;
+               }
+               if (fdp->fd_nfiles < NDEXTENT) {

                        numfiles = NDEXTENT;
                        numfiles = NDEXTENT;

-               else
+               } else {

                        numfiles = 2 * fdp->fd_nfiles;
                        numfiles = 2 * fdp->fd_nfiles;

+               }

                /* Enforce lim */
                /* Enforce lim */

-               if (numfiles > lim)
+               if (numfiles > lim) {

                        numfiles = lim;
                        numfiles = lim;

+               }

                proc_fdunlock(p);
                MALLOC_ZONE(newofiles, struct fileproc **,
                proc_fdunlock(p);
                MALLOC_ZONE(newofiles, struct fileproc **,

-                               numfiles * OFILESIZE, M_OFILETABL, M_WAITOK);
+                   numfiles * OFILESIZE, M_OFILETABL, M_WAITOK);

                proc_fdlock(p);
                if (newofiles == NULL) {
                proc_fdlock(p);
                if (newofiles == NULL) {

-                       return (ENOMEM);
+                       return ENOMEM;

                }
                if (fdp->fd_nfiles >= numfiles) {
                        FREE_ZONE(newofiles, numfiles * OFILESIZE, M_OFILETABL);
                }
                if (fdp->fd_nfiles >= numfiles) {
                        FREE_ZONE(newofiles, numfiles * OFILESIZE, M_OFILETABL);


@@ -3013,15 +3468,15 @@ fdalloc(proc_t p, int want, int *result)
                 */
                oldnfiles = fdp->fd_nfiles;
                (void) memcpy(newofiles, fdp->fd_ofiles,
                 */
                oldnfiles = fdp->fd_nfiles;
                (void) memcpy(newofiles, fdp->fd_ofiles,

-                               oldnfiles * sizeof(*fdp->fd_ofiles));
+                   oldnfiles * sizeof(*fdp->fd_ofiles));

                (void) memset(&newofiles[oldnfiles], 0,
                (void) memset(&newofiles[oldnfiles], 0,

-                               (numfiles - oldnfiles) * sizeof(*fdp->fd_ofiles));
+                   (numfiles - oldnfiles) * sizeof(*fdp->fd_ofiles));

 
                (void) memcpy(newofileflags, fdp->fd_ofileflags,
 
                (void) memcpy(newofileflags, fdp->fd_ofileflags,

-                               oldnfiles * sizeof(*fdp->fd_ofileflags));
+                   oldnfiles * sizeof(*fdp->fd_ofileflags));

                (void) memset(&newofileflags[oldnfiles], 0,
                (void) memset(&newofileflags[oldnfiles], 0,

-                               (numfiles - oldnfiles) *
-                                               sizeof(*fdp->fd_ofileflags));
+                   (numfiles - oldnfiles) *
+                   sizeof(*fdp->fd_ofileflags));

                ofiles = fdp->fd_ofiles;
                fdp->fd_ofiles = newofiles;
                fdp->fd_ofileflags = newofileflags;
                ofiles = fdp->fd_ofiles;
                fdp->fd_ofiles = newofiles;
                fdp->fd_ofileflags = newofileflags;


@@ -3058,14 +3513,17 @@ fdavail(proc_t p, int n)
        int i, lim;
 
        lim = min((int)p->p_rlimit[RLIMIT_NOFILE].rlim_cur, maxfiles);
        int i, lim;
 
        lim = min((int)p->p_rlimit[RLIMIT_NOFILE].rlim_cur, maxfiles);

-       if ((i = lim - fdp->fd_nfiles) > 0 && (n -= i) <= 0)
-               return (1);
+       if ((i = lim - fdp->fd_nfiles) > 0 && (n -= i) <= 0) {
+               return 1;
+       }

        fpp = &fdp->fd_ofiles[fdp->fd_freefile];
        flags = &fdp->fd_ofileflags[fdp->fd_freefile];
        fpp = &fdp->fd_ofiles[fdp->fd_freefile];
        flags = &fdp->fd_ofileflags[fdp->fd_freefile];

-       for (i = fdp->fd_nfiles - fdp->fd_freefile; --i >= 0; fpp++, flags++)
-               if (*fpp == NULL && !(*flags & UF_RESERVED) && --n <= 0)
-                       return (1);
-       return (0);
+       for (i = fdp->fd_nfiles - fdp->fd_freefile; --i >= 0; fpp++, flags++) {
+               if (*fpp == NULL && !(*flags & UF_RESERVED) && --n <= 0) {
+                       return 1;
+               }
+       }
+       return 0;

 }
 
 
 }
 
 


@@ -3117,13 +3575,14 @@ fdgetf_noref(proc_t p, int fd, struct fileproc **resultfp)
        struct fileproc *fp;
 
        if (fd < 0 || fd >= fdp->fd_nfiles ||
        struct fileproc *fp;
 
        if (fd < 0 || fd >= fdp->fd_nfiles ||

-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
-               return (EBADF);
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+               return EBADF;

        }
        }

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       return (0);
+       }
+       return 0;

 }
 
 
 }
 
 


@@ -3162,24 +3621,26 @@ fp_getfvp(proc_t p, int fd, struct fileproc **resultfp, struct vnode **resultvp)
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||

-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        if (fp->f_type != DTYPE_VNODE) {
                proc_fdunlock(p);
        }
        if (fp->f_type != DTYPE_VNODE) {
                proc_fdunlock(p);

-               return(ENOTSUP);
+               return ENOTSUP;

        }
        }

-       fp->f_iocount++;
+       os_ref_retain_locked(&fp->f_iocount);

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (resultvp)
+       }
+       if (resultvp) {

                *resultvp = (struct vnode *)fp->f_data;
                *resultvp = (struct vnode *)fp->f_data;

+       }

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -3214,33 +3675,36 @@ fp_getfvp(proc_t p, int fd, struct fileproc **resultfp, struct vnode **resultvp)
  */
 int
 fp_getfvpandvid(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfvpandvid(proc_t p, int fd, struct fileproc **resultfp,

-               struct vnode **resultvp, uint32_t *vidp)
+    struct vnode **resultvp, uint32_t *vidp)

 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||
 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||

-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        if (fp->f_type != DTYPE_VNODE) {
                proc_fdunlock(p);
        }
        if (fp->f_type != DTYPE_VNODE) {
                proc_fdunlock(p);

-               return(ENOTSUP);
+               return ENOTSUP;

        }
        }

-       fp->f_iocount++;
+       os_ref_retain_locked(&fp->f_iocount);

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (resultvp)
+       }
+       if (resultvp) {

                *resultvp = (struct vnode *)fp->f_data;
                *resultvp = (struct vnode *)fp->f_data;

-       if (vidp)
+       }
+       if (vidp) {

                *vidp = (uint32_t)vnode_vid((struct vnode *)fp->f_data);
                *vidp = (uint32_t)vnode_vid((struct vnode *)fp->f_data);

+       }

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -3271,31 +3735,33 @@ fp_getfvpandvid(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfsock(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfsock(proc_t p, int fd, struct fileproc **resultfp,

-           struct socket **results)
+    struct socket **results)

 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||
 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||

-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        if (fp->f_type != DTYPE_SOCKET) {
                proc_fdunlock(p);
        }
        if (fp->f_type != DTYPE_SOCKET) {
                proc_fdunlock(p);

-               return(EOPNOTSUPP);
+               return EOPNOTSUPP;

        }
        }

-       fp->f_iocount++;
+       os_ref_retain_locked(&fp->f_iocount);

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (results)
+       }
+       if (results) {

                *results = (struct socket *)fp->f_data;
                *results = (struct socket *)fp->f_data;

+       }

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -3326,31 +3792,33 @@ fp_getfsock(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfkq(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfkq(proc_t p, int fd, struct fileproc **resultfp,

-         struct kqueue **resultkq)
+    struct kqueue **resultkq)

 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);

-       if ( fd < 0 || fd >= fdp->fd_nfiles ||
-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+       if (fd < 0 || fd >= fdp->fd_nfiles ||
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        if (fp->f_type != DTYPE_KQUEUE) {
                proc_fdunlock(p);
        }
        if (fp->f_type != DTYPE_KQUEUE) {
                proc_fdunlock(p);

-               return(EBADF);
+               return EBADF;

        }
        }

-       fp->f_iocount++;
+       os_ref_retain_locked(&fp->f_iocount);

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (resultkq)
+       }
+       if (resultkq) {

                *resultkq = (struct kqueue *)fp->f_data;
                *resultkq = (struct kqueue *)fp->f_data;

+       }

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -3383,32 +3851,33 @@ fp_getfkq(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfpshm(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfpshm(proc_t p, int fd, struct fileproc **resultfp,

-           struct pshmnode **resultpshm)
+    struct pshmnode **resultpshm)

 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||
 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||

-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        if (fp->f_type != DTYPE_PSXSHM) {
        }
        if (fp->f_type != DTYPE_PSXSHM) {

-

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(EBADF);
+               return EBADF;

        }
        }

-       fp->f_iocount++;
+       os_ref_retain_locked(&fp->f_iocount);

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (resultpshm)
+       }
+       if (resultpshm) {

                *resultpshm = (struct pshmnode *)fp->f_data;
                *resultpshm = (struct pshmnode *)fp->f_data;

+       }

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -3450,31 +3919,33 @@ fp_getfpshm(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfpsem(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfpsem(proc_t p, int fd, struct fileproc **resultfp,

-           struct psemnode **resultpsem)
+    struct psemnode **resultpsem)

 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||
 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||

-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        if (fp->f_type != DTYPE_PSXSEM) {
                proc_fdunlock(p);
        }
        if (fp->f_type != DTYPE_PSXSEM) {
                proc_fdunlock(p);

-               return(EBADF);
+               return EBADF;

        }
        }

-       fp->f_iocount++;
+       os_ref_retain_locked(&fp->f_iocount);

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (resultpsem)
+       }
+       if (resultpsem) {

                *resultpsem = (struct psemnode *)fp->f_data;
                *resultpsem = (struct psemnode *)fp->f_data;

+       }

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -3505,33 +3976,36 @@ fp_getfpsem(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfpipe(proc_t p, int fd, struct fileproc **resultfp,
  */
 int
 fp_getfpipe(proc_t p, int fd, struct fileproc **resultfp,

-           struct pipe **resultpipe)
+    struct pipe **resultpipe)

 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||
 {
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= fdp->fd_nfiles ||

-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        if (fp->f_type != DTYPE_PIPE) {
                proc_fdunlock(p);
        }
        if (fp->f_type != DTYPE_PIPE) {
                proc_fdunlock(p);

-               return(EBADF);
+               return EBADF;

        }
        }

-       fp->f_iocount++;
+       os_ref_retain_locked(&fp->f_iocount);

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (resultpipe)
+       }
+       if (resultpipe) {

                *resultpipe = (struct pipe *)fp->f_data;
                *resultpipe = (struct pipe *)fp->f_data;

+       }

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return (0);
+       return 0;

 }
 
 }
 

+

 /*
  * fp_lookup
  *
 /*
  * fp_lookup
  *


@@ -3562,32 +4036,37 @@ fp_lookup(proc_t p, int fd, struct fileproc **resultfp, int locked)
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 
        struct filedesc *fdp = p->p_fd;
        struct fileproc *fp;
 

-       if (!locked)
+       if (!locked) {

                proc_fdlock_spin(p);
                proc_fdlock_spin(p);

+       }

        if (fd < 0 || fdp == NULL || fd >= fdp->fd_nfiles ||
        if (fd < 0 || fdp == NULL || fd >= fdp->fd_nfiles ||

-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
-               if (!locked)
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           (fdp->fd_ofileflags[fd] & UF_RESERVED)) {
+               if (!locked) {

                        proc_fdunlock(p);
                        proc_fdunlock(p);

-               return (EBADF);
+               }
+               return EBADF;

        }
        }

-       fp->f_iocount++;
+       os_ref_retain_locked(&fp->f_iocount);

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (!locked)
+       }
+       if (!locked) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               
-       return (0);
+       }
+
+       return 0;

 }
 
 
 /*
  * fp_tryswap
 }
 
 
 /*
  * fp_tryswap

- * 
+ *

  * Description: Swap the fileproc pointer for a given fd with a new
  *             fileproc pointer in the per-process open file table of
  *             the specified process.  The fdlock must be held at entry.
  * Description: Swap the fileproc pointer for a given fd with a new
  *             fileproc pointer in the per-process open file table of
  *             the specified process.  The fdlock must be held at entry.

+ *             Iff the swap is successful, the old fileproc pointer is freed.

  *
  * Parameters:  p              Process containing the fd
  *             fd              The fd of interest
  *
  * Parameters:  p              Process containing the fd
  *             fd              The fd of interest


@@ -3596,7 +4075,7 @@ fp_lookup(proc_t p, int fd, struct fileproc **resultfp, int locked)
  * Returns:    0               Success
  *             EBADF           Bad file descriptor
  *             EINTR           Interrupted
  * Returns:    0               Success
  *             EBADF           Bad file descriptor
  *             EINTR           Interrupted

- *             EKEEPLOOKING    f_iocount changed while lock was dropped.
+ *             EKEEPLOOKING    Other references were active, try again.

  */
 int
 fp_tryswap(proc_t p, int fd, struct fileproc *nfp)
  */
 int
 fp_tryswap(proc_t p, int fd, struct fileproc *nfp)


@@ -3606,27 +4085,35 @@ fp_tryswap(proc_t p, int fd, struct fileproc *nfp)
 
        proc_fdlock_assert(p, LCK_MTX_ASSERT_OWNED);
 
 
        proc_fdlock_assert(p, LCK_MTX_ASSERT_OWNED);
 

-       if (0 != (error = fp_lookup(p, fd, &fp, 1)))
-               return (error);
+       if (0 != (error = fp_lookup(p, fd, &fp, 1))) {
+               return error;
+       }

        /*
         * At this point, our caller (change_guardedfd_np) has
         * one f_iocount reference, and we just took another
         * one to begin the replacement.
        /*
         * At this point, our caller (change_guardedfd_np) has
         * one f_iocount reference, and we just took another
         * one to begin the replacement.

+        * fp and nfp have a +1 reference from allocation.
+        * Thus if no-one else is looking, f_iocount should be 3.

         */
         */

-       if (fp->f_iocount < 2) {
-               panic("f_iocount too small %d", fp->f_iocount);
-       } else if (2 == fp->f_iocount) {
-
+       if (os_ref_get_count(&fp->f_iocount) < 3 ||
+           1 != os_ref_get_count(&nfp->f_iocount)) {
+               panic("%s: f_iocount", __func__);
+       } else if (3 == os_ref_get_count(&fp->f_iocount)) {

                /* Copy the contents of *fp, preserving the "type" of *nfp */
 
                nfp->f_flags = (nfp->f_flags & FP_TYPEMASK) |
                /* Copy the contents of *fp, preserving the "type" of *nfp */
 
                nfp->f_flags = (nfp->f_flags & FP_TYPEMASK) |

-                       (fp->f_flags & ~FP_TYPEMASK);
-               nfp->f_iocount = fp->f_iocount;
+                   (fp->f_flags & ~FP_TYPEMASK);
+               os_ref_retain_locked(&nfp->f_iocount);
+               os_ref_retain_locked(&nfp->f_iocount);

                nfp->f_fglob = fp->f_fglob;
                nfp->f_fglob = fp->f_fglob;

-               nfp->f_waddr = fp->f_waddr;
+               nfp->f_wset = fp->f_wset;

 
                p->p_fd->fd_ofiles[fd] = nfp;
 
                p->p_fd->fd_ofiles[fd] = nfp;

-               (void) fp_drop(p, fd, nfp, 1);
+               fp_drop(p, fd, nfp, 1);
+
+               os_ref_release_live(&fp->f_iocount);
+               os_ref_release_live(&fp->f_iocount);
+               fileproc_free(fp);

        } else {
                /*
                 * Wait for all other active references to evaporate.
        } else {
                /*
                 * Wait for all other active references to evaporate.


@@ -3640,11 +4127,10 @@ fp_tryswap(proc_t p, int fd, struct fileproc *nfp)
                         * reevaluation of the change-guard attempt.
                         */
                        error = EKEEPLOOKING;
                         * reevaluation of the change-guard attempt.
                         */
                        error = EKEEPLOOKING;

-                       printf("%s: lookup collision fd %d\n", __func__, fd);

                }
                (void) fp_drop(p, fd, fp, 1);
        }
                }
                (void) fp_drop(p, fd, fp, 1);
        }

-       return (error);
+       return error;

 }
 
 
 }
 
 


@@ -3670,17 +4156,17 @@ fp_tryswap(proc_t p, int fd, struct fileproc *nfp)
 int
 fp_drop_written(proc_t p, int fd, struct fileproc *fp)
 {
 int
 fp_drop_written(proc_t p, int fd, struct fileproc *fp)
 {

-        int error;
+       int error;

 
        proc_fdlock_spin(p);
 
        fp->f_flags |= FP_WRITTEN;
 
        proc_fdlock_spin(p);
 
        fp->f_flags |= FP_WRITTEN;

-       
+

        error = fp_drop(p, fd, fp, 1);
 
        proc_fdunlock(p);
        error = fp_drop(p, fd, fp, 1);
 
        proc_fdunlock(p);

-               
-       return (error);
+
+       return error;

 }
 
 
 }
 
 


@@ -3706,17 +4192,17 @@ fp_drop_written(proc_t p, int fd, struct fileproc *fp)
 int
 fp_drop_event(proc_t p, int fd, struct fileproc *fp)
 {
 int
 fp_drop_event(proc_t p, int fd, struct fileproc *fp)
 {

-        int error;
+       int error;

 
        proc_fdlock_spin(p);
 
        fp->f_flags |= FP_WAITEVENT;
 
        proc_fdlock_spin(p);
 
        fp->f_flags |= FP_WAITEVENT;

-       
+

        error = fp_drop(p, fd, fp, 1);
 
        proc_fdunlock(p);
        error = fp_drop(p, fd, fp, 1);
 
        proc_fdunlock(p);

-               
-       return (error);
+
+       return error;

 }
 
 
 }
 
 


@@ -3747,35 +4233,39 @@ int
 fp_drop(proc_t p, int fd, struct fileproc *fp, int locked)
 {
        struct filedesc *fdp = p->p_fd;
 fp_drop(proc_t p, int fd, struct fileproc *fp, int locked)
 {
        struct filedesc *fdp = p->p_fd;

-       int     needwakeup = 0;
+       int     needwakeup = 0;

 
 

-       if (!locked)
+       if (!locked) {

                proc_fdlock_spin(p);
                proc_fdlock_spin(p);

-        if ((fp == FILEPROC_NULL) && (fd < 0 || fd >= fdp->fd_nfiles ||
-                       (fp = fdp->fd_ofiles[fd]) == NULL ||
-                       ((fdp->fd_ofileflags[fd] & UF_RESERVED) &&
-                        !(fdp->fd_ofileflags[fd] & UF_CLOSING)))) {
-               if (!locked)
+       }
+       if ((fp == FILEPROC_NULL) && (fd < 0 || fd >= fdp->fd_nfiles ||
+           (fp = fdp->fd_ofiles[fd]) == NULL ||
+           ((fdp->fd_ofileflags[fd] & UF_RESERVED) &&
+           !(fdp->fd_ofileflags[fd] & UF_CLOSING)))) {
+               if (!locked) {

                        proc_fdunlock(p);
                        proc_fdunlock(p);

-               return (EBADF);
+               }
+               return EBADF;

        }
        }

-       fp->f_iocount--;

 
 

-       if (fp->f_iocount == 0) {
-               if (fp->f_flags & FP_SELCONFLICT)
+       if (1 == os_ref_release_locked(&fp->f_iocount)) {
+               if (fp->f_flags & FP_SELCONFLICT) {

                        fp->f_flags &= ~FP_SELCONFLICT;
                        fp->f_flags &= ~FP_SELCONFLICT;

+               }

 
                if (p->p_fpdrainwait) {
                        p->p_fpdrainwait = 0;
                        needwakeup = 1;
                }
        }
 
                if (p->p_fpdrainwait) {
                        p->p_fpdrainwait = 0;
                        needwakeup = 1;
                }
        }

-       if (!locked)
+       if (!locked) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-       if (needwakeup)
-               wakeup(&p->p_fpdrainwait);
-               
-       return (0);
+       }
+       if (needwakeup) {
+               wakeup(&p->p_fpdrainwait);
+       }
+
+       return 0;

 }
 
 
 }
 
 


@@ -3820,22 +4310,23 @@ file_vnode(int fd, struct vnode **vpp)
        proc_t p = current_proc();
        struct fileproc *fp;
        int error;
        proc_t p = current_proc();
        struct fileproc *fp;
        int error;

-       
+

        proc_fdlock_spin(p);
        proc_fdlock_spin(p);

-       if ( (error = fp_lookup(p, fd, &fp, 1)) ) {
+       if ((error = fp_lookup(p, fd, &fp, 1))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(error);
+               return error;

        }
        if (fp->f_type != DTYPE_VNODE) {
        }
        if (fp->f_type != DTYPE_VNODE) {

-               fp_drop(p, fd, fp,1);
+               fp_drop(p, fd, fp, 1);

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(EINVAL);
+               return EINVAL;

        }
        }

-       if (vpp != NULL)
+       if (vpp != NULL) {

                *vpp = (struct vnode *)fp->f_data;
                *vpp = (struct vnode *)fp->f_data;

+       }

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return(0);
+       return 0;

 }
 
 
 }
 
 


@@ -3882,27 +4373,29 @@ file_vnode_withvid(int fd, struct vnode **vpp, uint32_t * vidp)
        struct fileproc *fp;
        vnode_t vp;
        int error;
        struct fileproc *fp;
        vnode_t vp;
        int error;

-       
+

        proc_fdlock_spin(p);
        proc_fdlock_spin(p);

-       if ( (error = fp_lookup(p, fd, &fp, 1)) ) {
+       if ((error = fp_lookup(p, fd, &fp, 1))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(error);
+               return error;

        }
        if (fp->f_type != DTYPE_VNODE) {
        }
        if (fp->f_type != DTYPE_VNODE) {

-               fp_drop(p, fd, fp,1);
+               fp_drop(p, fd, fp, 1);

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(EINVAL);
+               return EINVAL;

        }
        vp = (struct vnode *)fp->f_data;
        }
        vp = (struct vnode *)fp->f_data;

-       if (vpp != NULL) 
+       if (vpp != NULL) {

                *vpp = vp;
                *vpp = vp;

+       }

 
 

-       if ((vidp != NULL) && (vp != NULLVP)) 
+       if ((vidp != NULL) && (vp != NULLVP)) {

                *vidp = (uint32_t)vp->v_id;
                *vidp = (uint32_t)vp->v_id;

+       }

 
        proc_fdunlock(p);
 
 
        proc_fdunlock(p);
 

-       return(0);
+       return 0;

 }
 
 
 }
 
 


@@ -3946,21 +4439,21 @@ file_socket(int fd, struct socket **sp)
        proc_t p = current_proc();
        struct fileproc *fp;
        int error;
        proc_t p = current_proc();
        struct fileproc *fp;
        int error;

-       
+

        proc_fdlock_spin(p);
        proc_fdlock_spin(p);

-       if ( (error = fp_lookup(p, fd, &fp, 1)) ) {
+       if ((error = fp_lookup(p, fd, &fp, 1))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(error);
+               return error;

        }
        if (fp->f_type != DTYPE_SOCKET) {
        }
        if (fp->f_type != DTYPE_SOCKET) {

-               fp_drop(p, fd, fp,1);
+               fp_drop(p, fd, fp, 1);

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(ENOTSOCK);
+               return ENOTSOCK;

        }
        *sp = (struct socket *)fp->f_data;
        proc_fdunlock(p);
 
        }
        *sp = (struct socket *)fp->f_data;
        proc_fdunlock(p);
 

-       return(0);
+       return 0;

 }
 
 
 }
 
 


@@ -3990,21 +4483,20 @@ file_socket(int fd, struct socket **sp)
 int
 file_flags(int fd, int *flags)
 {
 int
 file_flags(int fd, int *flags)
 {

-

        proc_t p = current_proc();
        struct fileproc *fp;
        int error;
        proc_t p = current_proc();
        struct fileproc *fp;
        int error;

-       
+

        proc_fdlock_spin(p);
        proc_fdlock_spin(p);

-       if ( (error = fp_lookup(p, fd, &fp, 1)) ) {
+       if ((error = fp_lookup(p, fd, &fp, 1))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return(error);
+               return error;

        }
        *flags = (int)fp->f_flag;
        }
        *flags = (int)fp->f_flag;

-       fp_drop(p, fd, fp,1);
+       fp_drop(p, fd, fp, 1);

        proc_fdunlock(p);
 
        proc_fdunlock(p);
 

-       return(0);
+       return 0;

 }
 
 
 }
 
 


@@ -4044,26 +4536,26 @@ file_flags(int fd, int *flags)
  *
  *             Use of this function is discouraged.
  */
  *
  *             Use of this function is discouraged.
  */

-int 
+int

 file_drop(int fd)
 {
        struct fileproc *fp;
        proc_t p = current_proc();
 file_drop(int fd)
 {
        struct fileproc *fp;
        proc_t p = current_proc();

-       int     needwakeup = 0;
+       int     needwakeup = 0;

 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= p->p_fd->fd_nfiles ||
 
        proc_fdlock_spin(p);
        if (fd < 0 || fd >= p->p_fd->fd_nfiles ||

-                       (fp = p->p_fd->fd_ofiles[fd]) == NULL ||
-                       ((p->p_fd->fd_ofileflags[fd] & UF_RESERVED) &&
-                        !(p->p_fd->fd_ofileflags[fd] & UF_CLOSING))) {
+           (fp = p->p_fd->fd_ofiles[fd]) == NULL ||
+           ((p->p_fd->fd_ofileflags[fd] & UF_RESERVED) &&
+           !(p->p_fd->fd_ofileflags[fd] & UF_CLOSING))) {

                proc_fdunlock(p);
                proc_fdunlock(p);

-               return (EBADF);
+               return EBADF;

        }
        }

-       fp->f_iocount --;

 
 

-       if (fp->f_iocount == 0) {
-               if (fp->f_flags & FP_SELCONFLICT)
+       if (1 == os_ref_release_locked(&fp->f_iocount)) {
+               if (fp->f_flags & FP_SELCONFLICT) {

                        fp->f_flags &= ~FP_SELCONFLICT;
                        fp->f_flags &= ~FP_SELCONFLICT;

+               }

 
                if (p->p_fpdrainwait) {
                        p->p_fpdrainwait = 0;
 
                if (p->p_fpdrainwait) {
                        p->p_fpdrainwait = 0;


@@ -4072,9 +4564,10 @@ file_drop(int fd)
        }
        proc_fdunlock(p);
 
        }
        proc_fdunlock(p);
 

-       if (needwakeup)
-               wakeup(&p->p_fpdrainwait);
-       return(0);
+       if (needwakeup) {
+               wakeup(&p->p_fpdrainwait);
+       }
+       return 0;

 }
 
 
 }
 
 


@@ -4115,8 +4608,8 @@ static int falloc_withalloc_locked(proc_t, struct fileproc **, int *,
 int
 falloc(proc_t p, struct fileproc **resultfp, int *resultfd, vfs_context_t ctx)
 {
 int
 falloc(proc_t p, struct fileproc **resultfp, int *resultfd, vfs_context_t ctx)
 {

-       return (falloc_withalloc(p, resultfp, resultfd, ctx,
-           fileproc_alloc_init, NULL));
+       return falloc_withalloc(p, resultfp, resultfd, ctx,
+                  fileproc_alloc_init, NULL);

 }
 
 /*
 }
 
 /*


@@ -4133,7 +4626,7 @@ falloc_withalloc(proc_t p, struct fileproc **resultfp, int *resultfd,
            resultfp, resultfd, ctx, fp_zalloc, arg, 1);
        proc_fdunlock(p);
 
            resultfp, resultfd, ctx, fp_zalloc, arg, 1);
        proc_fdunlock(p);
 

-       return (error);
+       return error;

 }
 
 /*
 }
 
 /*


@@ -4185,10 +4678,10 @@ static const struct fileops uninitops;
  */
 int
 falloc_locked(proc_t p, struct fileproc **resultfp, int *resultfd,
  */
 int
 falloc_locked(proc_t p, struct fileproc **resultfp, int *resultfd,

-             vfs_context_t ctx, int locked)
+    vfs_context_t ctx, int locked)

 {
 {

-       return (falloc_withalloc_locked(p, resultfp, resultfd, ctx,
-           fileproc_alloc_init, NULL, locked));
+       return falloc_withalloc_locked(p, resultfp, resultfd, ctx,
+                  fileproc_alloc_init, NULL, locked);

 }
 
 static int
 }
 
 static int


@@ -4200,25 +4693,29 @@ falloc_withalloc_locked(proc_t p, struct fileproc **resultfp, int *resultfd,
        struct fileglob *fg;
        int error, nfd;
 
        struct fileglob *fg;
        int error, nfd;
 

-       if (!locked)
+       if (nfiles >= maxfiles) {
+               tablefull("file");
+               return ENFILE;
+       }
+
+       if (!locked) {

                proc_fdlock(p);
                proc_fdlock(p);

-       if ( (error = fdalloc(p, 0, &nfd)) ) {
-               if (!locked)
-                       proc_fdunlock(p);
-               return (error);

        }
        }

-       if (nfiles >= maxfiles) {
-               if (!locked)
+
+       if ((error = fdalloc(p, 0, &nfd))) {
+               if (!locked) {

                        proc_fdunlock(p);
                        proc_fdunlock(p);

-               tablefull("file");
-               return (ENFILE);
+               }
+               return error;

        }
        }

+

 #if CONFIG_MACF
        error = mac_file_check_create(proc_ucred(p));
        if (error) {
 #if CONFIG_MACF
        error = mac_file_check_create(proc_ucred(p));
        if (error) {

-               if (!locked)
+               if (!locked) {

                        proc_fdunlock(p);
                        proc_fdunlock(p);

-               return (error);
+               }
+               return error;

        }
 #endif
 
        }
 #endif
 


@@ -4232,21 +4729,23 @@ falloc_withalloc_locked(proc_t p, struct fileproc **resultfp, int *resultfd,
 
        fp = (*fp_zalloc)(crarg);
        if (fp == NULL) {
 
        fp = (*fp_zalloc)(crarg);
        if (fp == NULL) {

-               if (locked)
+               if (locked) {

                        proc_fdlock(p);
                        proc_fdlock(p);

-               return (ENOMEM);
+               }
+               return ENOMEM;

        }
        MALLOC_ZONE(fg, struct fileglob *, sizeof(struct fileglob), M_FILEGLOB, M_WAITOK);
        if (fg == NULL) {
                fileproc_free(fp);
        }
        MALLOC_ZONE(fg, struct fileglob *, sizeof(struct fileglob), M_FILEGLOB, M_WAITOK);
        if (fg == NULL) {
                fileproc_free(fp);

-               if (locked)
+               if (locked) {

                        proc_fdlock(p);
                        proc_fdlock(p);

-               return (ENOMEM);
+               }
+               return ENOMEM;

        }
        bzero(fg, sizeof(struct fileglob));
        lck_mtx_init(&fg->fg_lock, file_lck_grp, file_lck_attr);
 
        }
        bzero(fg, sizeof(struct fileglob));
        lck_mtx_init(&fg->fg_lock, file_lck_grp, file_lck_attr);
 

-       fp->f_iocount = 1;
+       os_ref_retain_locked(&fp->f_iocount);

        fg->fg_count = 1;
        fg->fg_ops = &uninitops;
        fp->f_fglob = fg;
        fg->fg_count = 1;
        fg->fg_ops = &uninitops;
        fp->f_fglob = fg;


@@ -4268,15 +4767,18 @@ falloc_withalloc_locked(proc_t p, struct fileproc **resultfp, int *resultfd,
 
        p->p_fd->fd_ofiles[nfd] = fp;
 
 
        p->p_fd->fd_ofiles[nfd] = fp;
 

-       if (!locked)
+       if (!locked) {

                proc_fdunlock(p);
                proc_fdunlock(p);

+       }

 
 

-       if (resultfp)
+       if (resultfp) {

                *resultfp = fp;
                *resultfp = fp;

-       if (resultfd)
+       }
+       if (resultfd) {

                *resultfd = nfd;
                *resultfd = nfd;

+       }

 
 

-       return (0);
+       return 0;

 }
 
 
 }
 
 


@@ -4297,6 +4799,11 @@ fg_free(struct fileglob *fg)
 {
        OSAddAtomic(-1, &nfiles);
 
 {
        OSAddAtomic(-1, &nfiles);
 

+       if (fg->fg_vn_data) {
+               fg_vn_data_free(fg->fg_vn_data);
+               fg->fg_vn_data = NULL;
+       }
+

        if (IS_VALID_CRED(fg->fg_cred)) {
                kauth_cred_unref(&fg->fg_cred);
        }
        if (IS_VALID_CRED(fg->fg_cred)) {
                kauth_cred_unref(&fg->fg_cred);
        }


@@ -4309,6 +4816,28 @@ fg_free(struct fileglob *fg)
 }
 
 
 }
 
 

+/*
+ * fg_get_vnode
+ *
+ * Description:        Return vnode associated with the file structure, if
+ *             any.  The lifetime of the returned vnode is bound to
+ *             the lifetime of the file structure.
+ *
+ * Parameters: fg                              Pointer to fileglob to
+ *                                             inspect
+ *
+ * Returns:    vnode_t
+ */
+vnode_t
+fg_get_vnode(struct fileglob *fg)
+{
+       if (FILEGLOB_DTYPE(fg) == DTYPE_VNODE) {
+               return (vnode_t)fg->fg_data;
+       } else {
+               return NULL;
+       }
+}
+

 /*
  * fdexec
  *
 /*
  * fdexec
  *


@@ -4326,30 +4855,39 @@ fg_free(struct fileglob *fg)
  * Returns:    void
  *
  * Locks:      This function internally takes and drops proc_fdlock()
  * Returns:    void
  *
  * Locks:      This function internally takes and drops proc_fdlock()

+ *          But assumes tables don't grow/change while unlocked.

  *
  *

- * Notes:      This function drops and retakes the kernel funnel; this is
- *             inherently unsafe, since another thread may have the
- *             proc_fdlock.
- *
- * XXX:                We should likely reverse the lock and funnel drop/acquire
- *             order to avoid the small race window; it's also possible that
- *             if the program doing the exec has an outstanding listen socket
- *             and a network connection is completed asynchronously that we
- *             will end up with a "ghost" socket reference in the new process.
- *
- *             This needs reworking to make it safe to remove the funnel from
- *             the execve and posix_spawn system calls.

  */
 void
  */
 void

-fdexec(proc_t p, short flags)
+fdexec(proc_t p, short flags, int self_exec)

 {
        struct filedesc *fdp = p->p_fd;
        int i;
        boolean_t cloexec_default = (flags & POSIX_SPAWN_CLOEXEC_DEFAULT) != 0;
 {
        struct filedesc *fdp = p->p_fd;
        int i;
        boolean_t cloexec_default = (flags & POSIX_SPAWN_CLOEXEC_DEFAULT) != 0;

+       thread_t self = current_thread();
+       struct uthread *ut = get_bsdthread_info(self);
+       struct kqworkq *dealloc_kqwq = NULL;
+
+       /*
+        * If the current thread is bound as a workq/workloop
+        * servicing thread, we need to unbind it first.
+        */
+       if (ut->uu_kqr_bound && self_exec) {
+               kqueue_threadreq_unbind(p, ut->uu_kqr_bound);
+       }

 
        proc_fdlock(p);
 
        proc_fdlock(p);

-       for (i = fdp->fd_lastfile; i >= 0; i--) {

 
 

+       /*
+        * Deallocate the knotes for this process
+        * and mark the tables non-existent so
+        * subsequent kqueue closes go faster.
+        */
+       knotes_dealloc(p);
+       assert(fdp->fd_knlistsize == 0);
+       assert(fdp->fd_knhashmask == 0);
+
+       for (i = fdp->fd_lastfile; i >= 0; i--) {

                struct fileproc *fp = fdp->fd_ofiles[i];
                char *flagp = &fdp->fd_ofileflags[i];
 
                struct fileproc *fp = fdp->fd_ofiles[i];
                char *flagp = &fdp->fd_ofileflags[i];
 


@@ -4360,41 +4898,56 @@ fdexec(proc_t p, short flags)
                         * except files marked explicitly as "inherit" and
                         * not marked close-on-exec.
                         */
                         * except files marked explicitly as "inherit" and
                         * not marked close-on-exec.
                         */

-                       if ((*flagp & (UF_EXCLOSE|UF_INHERIT)) != UF_INHERIT)
+                       if ((*flagp & (UF_EXCLOSE | UF_INHERIT)) != UF_INHERIT) {

                                *flagp |= UF_EXCLOSE;
                                *flagp |= UF_EXCLOSE;

+                       }

                        *flagp &= ~UF_INHERIT;
                }
 
                if (
                        *flagp &= ~UF_INHERIT;
                }
 
                if (

-                   ((*flagp & (UF_RESERVED|UF_EXCLOSE)) == UF_EXCLOSE)
+                       ((*flagp & (UF_RESERVED | UF_EXCLOSE)) == UF_EXCLOSE)

 #if CONFIG_MACF
 #if CONFIG_MACF

-                   || (fp && mac_file_check_inherit(proc_ucred(p), fp->f_fglob))
+                       || (fp && mac_file_check_inherit(proc_ucred(p), fp->f_fglob))

 #endif
 #endif

-               ) {
-                        if (i < fdp->fd_knlistsize)
-                                knote_fdclose(p, i);
+                       ) {

                        procfdtbl_clearfd(p, i);
                        procfdtbl_clearfd(p, i);

-                       if (i == fdp->fd_lastfile && i > 0)
+                       if (i == fdp->fd_lastfile && i > 0) {

                                fdp->fd_lastfile--;
                                fdp->fd_lastfile--;

-                       if (i < fdp->fd_freefile)
+                       }
+                       if (i < fdp->fd_freefile) {

                                fdp->fd_freefile = i;
                                fdp->fd_freefile = i;

+                       }

 
                        /*
                         * Wait for any third party viewers (e.g., lsof)
                         * to release their references to this fileproc.
                         */
 
                        /*
                         * Wait for any third party viewers (e.g., lsof)
                         * to release their references to this fileproc.
                         */

-                       while (fp->f_iocount > 0) {
+                       while (os_ref_get_count(&fp->f_iocount) > 1) {

                                p->p_fpdrainwait = 1;
                                msleep(&p->p_fpdrainwait, &p->p_fdmlock, PRIBIO,
                                    "fpdrain", NULL);
                        }
                                p->p_fpdrainwait = 1;
                                msleep(&p->p_fpdrainwait, &p->p_fdmlock, PRIBIO,
                                    "fpdrain", NULL);
                        }

-
+                       if (fp->f_flags & FP_WAITEVENT) {
+                               (void)waitevent_close(p, fp);
+                       }

                        closef_locked(fp, fp->f_fglob, p);
 
                        fileproc_free(fp);
                }
        }
                        closef_locked(fp, fp->f_fglob, p);
 
                        fileproc_free(fp);
                }
        }

+
+       /* release the per-process workq kq */
+       if (fdp->fd_wqkqueue) {
+               dealloc_kqwq = fdp->fd_wqkqueue;
+               fdp->fd_wqkqueue = NULL;
+       }
+

        proc_fdunlock(p);
        proc_fdunlock(p);

+
+       /* Anything to free? */
+       if (dealloc_kqwq) {
+               kqworkq_dealloc(dealloc_kqwq);
+       }

 }
 
 
 }
 
 


@@ -4440,12 +4993,13 @@ fdcopy(proc_t p, vnode_t uth_cdir)
        struct filedesc *newfdp, *fdp = p->p_fd;
        int i;
        struct fileproc *ofp, *fp;
        struct filedesc *newfdp, *fdp = p->p_fd;
        int i;
        struct fileproc *ofp, *fp;

-       vnode_t v_dir;
+       vnode_t v_dir;

 
        MALLOC_ZONE(newfdp, struct filedesc *,
 
        MALLOC_ZONE(newfdp, struct filedesc *,

-                       sizeof(*newfdp), M_FILEDESC, M_WAITOK);
-       if (newfdp == NULL)
-               return(NULL);
+           sizeof(*newfdp), M_FILEDESC, M_WAITOK);
+       if (newfdp == NULL) {
+               return NULL;
+       }

 
        proc_fdlock(p);
 
 
        proc_fdlock(p);
 


@@ -4459,8 +5013,9 @@ fdcopy(proc_t p, vnode_t uth_cdir)
         * inherit the new current working directory from the current thread
         * instead, before we take our references.
         */
         * inherit the new current working directory from the current thread
         * instead, before we take our references.
         */

-       if (uth_cdir != NULLVP)
+       if (uth_cdir != NULLVP) {

                newfdp->fd_cdir = uth_cdir;
                newfdp->fd_cdir = uth_cdir;

+       }

 
        /*
         * For both fd_cdir and fd_rdir make sure we get
 
        /*
         * For both fd_cdir and fd_rdir make sure we get


@@ -4470,16 +5025,18 @@ fdcopy(proc_t p, vnode_t uth_cdir)
         * and allows us to do the vnode_rele only on
         * a properly referenced vp
         */
         * and allows us to do the vnode_rele only on
         * a properly referenced vp
         */

-       if ( (v_dir = newfdp->fd_cdir) ) {
-               if (vnode_getwithref(v_dir) == 0) {
-                       if ( (vnode_ref(v_dir)) )
-                               newfdp->fd_cdir = NULL;
+       if ((v_dir = newfdp->fd_cdir)) {
+               if (vnode_getwithref(v_dir) == 0) {
+                       if ((vnode_ref(v_dir))) {
+                               newfdp->fd_cdir = NULL;
+                       }

                        vnode_put(v_dir);
                        vnode_put(v_dir);

-               } else
-                       newfdp->fd_cdir = NULL;
+               } else {
+                       newfdp->fd_cdir = NULL;
+               }

        }
        if (newfdp->fd_cdir == NULL && fdp->fd_cdir) {
        }
        if (newfdp->fd_cdir == NULL && fdp->fd_cdir) {

-               /*
+               /*

                 * we couldn't get a new reference on
                 * the current working directory being
                 * inherited... we might as well drop
                 * we couldn't get a new reference on
                 * the current working directory being
                 * inherited... we might as well drop


@@ -4488,22 +5045,23 @@ fdcopy(proc_t p, vnode_t uth_cdir)
                 * it useless... by dropping it we'll
                 * be that much closer to recycling it
                 */
                 * it useless... by dropping it we'll
                 * be that much closer to recycling it
                 */

-               vnode_rele(fdp->fd_cdir);
+               vnode_rele(fdp->fd_cdir);

                fdp->fd_cdir = NULL;
        }
 
                fdp->fd_cdir = NULL;
        }
 

-       if ( (v_dir = newfdp->fd_rdir) ) {
+       if ((v_dir = newfdp->fd_rdir)) {

                if (vnode_getwithref(v_dir) == 0) {
                if (vnode_getwithref(v_dir) == 0) {

-                       if ( (vnode_ref(v_dir)) )
-                               newfdp->fd_rdir = NULL;
+                       if ((vnode_ref(v_dir))) {
+                               newfdp->fd_rdir = NULL;
+                       }

                        vnode_put(v_dir);
                } else {
                        vnode_put(v_dir);
                } else {

-                       newfdp->fd_rdir = NULL;
+                       newfdp->fd_rdir = NULL;

                }
        }
        /* Coming from a chroot environment and unable to get a reference... */
        if (newfdp->fd_rdir == NULL && fdp->fd_rdir) {
                }
        }
        /* Coming from a chroot environment and unable to get a reference... */
        if (newfdp->fd_rdir == NULL && fdp->fd_rdir) {

-               /*
+               /*

                 * We couldn't get a new reference on
                 * the chroot directory being
                 * inherited... this is fatal, since
                 * We couldn't get a new reference on
                 * the chroot directory being
                 * inherited... this is fatal, since


@@ -4511,10 +5069,11 @@ fdcopy(proc_t p, vnode_t uth_cdir)
                 * escape from a chroot environment by
                 * the new process.
                 */
                 * escape from a chroot environment by
                 * the new process.
                 */

-               if (newfdp->fd_cdir)
-                       vnode_rele(newfdp->fd_cdir);
+               if (newfdp->fd_cdir) {
+                       vnode_rele(newfdp->fd_cdir);
+               }

                FREE_ZONE(newfdp, sizeof *newfdp, M_FILEDESC);
                FREE_ZONE(newfdp, sizeof *newfdp, M_FILEDESC);

-               return(NULL);
+               return NULL;

        }
 
        /*
        }
 
        /*


@@ -4523,30 +5082,33 @@ fdcopy(proc_t p, vnode_t uth_cdir)
         * additional memory for the number of descriptors currently
         * in use.
         */
         * additional memory for the number of descriptors currently
         * in use.
         */

-       if (newfdp->fd_lastfile < NDFILE)
+       if (newfdp->fd_lastfile < NDFILE) {

                i = NDFILE;
                i = NDFILE;

-       else {
+       } else {

                /*
                 * Compute the smallest multiple of NDEXTENT needed
                 * for the file descriptors currently in use,
                 * allowing the table to shrink.
                 */
                i = newfdp->fd_nfiles;
                /*
                 * Compute the smallest multiple of NDEXTENT needed
                 * for the file descriptors currently in use,
                 * allowing the table to shrink.
                 */
                i = newfdp->fd_nfiles;

-               while (i > 2 * NDEXTENT && i > newfdp->fd_lastfile * 2)
+               while (i > 1 + 2 * NDEXTENT && i > 1 + newfdp->fd_lastfile * 2) {

                        i /= 2;
                        i /= 2;

+               }

        }
        proc_fdunlock(p);
 
        MALLOC_ZONE(newfdp->fd_ofiles, struct fileproc **,
        }
        proc_fdunlock(p);
 
        MALLOC_ZONE(newfdp->fd_ofiles, struct fileproc **,

-                               i * OFILESIZE, M_OFILETABL, M_WAITOK);
+           i * OFILESIZE, M_OFILETABL, M_WAITOK);

        if (newfdp->fd_ofiles == NULL) {
        if (newfdp->fd_ofiles == NULL) {

-               if (newfdp->fd_cdir)
-                       vnode_rele(newfdp->fd_cdir);
-               if (newfdp->fd_rdir)
+               if (newfdp->fd_cdir) {
+                       vnode_rele(newfdp->fd_cdir);
+               }
+               if (newfdp->fd_rdir) {

                        vnode_rele(newfdp->fd_rdir);
                        vnode_rele(newfdp->fd_rdir);

+               }

 
                FREE_ZONE(newfdp, sizeof(*newfdp), M_FILEDESC);
 
                FREE_ZONE(newfdp, sizeof(*newfdp), M_FILEDESC);

-               return(NULL);
+               return NULL;

        }
        (void) memset(newfdp->fd_ofiles, 0, i * OFILESIZE);
        proc_fdlock(p);
        }
        (void) memset(newfdp->fd_ofiles, 0, i * OFILESIZE);
        proc_fdlock(p);


@@ -4559,43 +5121,20 @@ fdcopy(proc_t p, vnode_t uth_cdir)
                char *flags;
 
                (void) memcpy(newfdp->fd_ofiles, fdp->fd_ofiles,
                char *flags;
 
                (void) memcpy(newfdp->fd_ofiles, fdp->fd_ofiles,

-                                       (newfdp->fd_lastfile + 1) * sizeof(*fdp->fd_ofiles));
+                   (newfdp->fd_lastfile + 1) * sizeof(*fdp->fd_ofiles));

                (void) memcpy(newfdp->fd_ofileflags, fdp->fd_ofileflags,
                (void) memcpy(newfdp->fd_ofileflags, fdp->fd_ofileflags,

-                                       (newfdp->fd_lastfile + 1) * sizeof(*fdp->fd_ofileflags));
-
-               /*
-                * kq descriptors cannot be copied.
-                */
-               if (newfdp->fd_knlistsize != -1) {
-                       fpp = &newfdp->fd_ofiles[newfdp->fd_lastfile];
-                       flags = &newfdp->fd_ofileflags[newfdp->fd_lastfile];
-                       for (i = newfdp->fd_lastfile;
-                           i >= 0; i--, fpp--, flags--) {
-                               if (*flags & UF_RESERVED)
-                                       continue;       /* (removed below) */
-                               if (*fpp != NULL && (*fpp)->f_type == DTYPE_KQUEUE) {
-                                       *fpp = NULL;
-                                       *flags = 0;
-                                       if (i < newfdp->fd_freefile)
-                                               newfdp->fd_freefile = i;
-                               }
-                               if (*fpp == NULL && i == newfdp->fd_lastfile && i > 0)
-                                       newfdp->fd_lastfile--;
-                       }
-                       newfdp->fd_knlist = NULL;
-                       newfdp->fd_knlistsize = -1;
-                       newfdp->fd_knhash = NULL;
-                       newfdp->fd_knhashmask = 0;
-               }
-               fpp = newfdp->fd_ofiles;
-               flags = newfdp->fd_ofileflags;
+                   (newfdp->fd_lastfile + 1) * sizeof(*fdp->fd_ofileflags));

 
 

-               for (i = newfdp->fd_lastfile + 1; --i >= 0; fpp++, flags++)
+               fpp = &newfdp->fd_ofiles[newfdp->fd_lastfile];
+               flags = &newfdp->fd_ofileflags[newfdp->fd_lastfile];
+               for (i = newfdp->fd_lastfile; i >= 0; i--, fpp--, flags--) {

                        if ((ofp = *fpp) != NULL &&
                        if ((ofp = *fpp) != NULL &&

-                           0 == (*flags & (UF_FORKCLOSE|UF_RESERVED))) {
+                           0 == (ofp->f_fglob->fg_lflags & FG_CONFINED) &&
+                           0 == (*flags & (UF_FORKCLOSE | UF_RESERVED))) {

 #if DEBUG
 #if DEBUG

-                               if (FILEPROC_TYPE(ofp) != FTYPE_SIMPLE)
+                               if (FILEPROC_TYPE(ofp) != FTYPE_SIMPLE) {

                                        panic("complex fileproc");
                                        panic("complex fileproc");

+                               }

 #endif
                                fp = fileproc_alloc_init(NULL);
                                if (fp == NULL) {
 #endif
                                fp = fileproc_alloc_init(NULL);
                                if (fp == NULL) {


@@ -4612,15 +5151,36 @@ fdcopy(proc_t p, vnode_t uth_cdir)
                                        *fpp = fp;
                                }
                        } else {
                                        *fpp = fp;
                                }
                        } else {

-                               if (i < newfdp->fd_freefile)
-                                       newfdp->fd_freefile = i;

                                *fpp = NULL;
                                *flags = 0;
                        }
                                *fpp = NULL;
                                *flags = 0;
                        }

+                       if (*fpp == NULL) {
+                               if (i == newfdp->fd_lastfile && i > 0) {
+                                       newfdp->fd_lastfile--;
+                               }
+                               if (i < newfdp->fd_freefile) {
+                                       newfdp->fd_freefile = i;
+                               }
+                       }
+               }

        }
 
        proc_fdunlock(p);
        }
 
        proc_fdunlock(p);

-       return (newfdp);
+
+       /*
+        * Initialize knote and kqueue tracking structs
+        */
+       newfdp->fd_knlist = NULL;
+       newfdp->fd_knlistsize = 0;
+       newfdp->fd_knhash = NULL;
+       newfdp->fd_knhashmask = 0;
+       newfdp->fd_kqhash = NULL;
+       newfdp->fd_kqhashmask = 0;
+       newfdp->fd_wqkqueue = NULL;
+       lck_mtx_init(&newfdp->fd_kqhashlock, proc_kqhashlock_grp, proc_lck_attr);
+       lck_mtx_init(&newfdp->fd_knhashlock, proc_knhashlock_grp, proc_lck_attr);
+
+       return newfdp;

 }
 
 
 }
 
 


@@ -4643,33 +5203,50 @@ fdfree(proc_t p)
 {
        struct filedesc *fdp;
        struct fileproc *fp;
 {
        struct filedesc *fdp;
        struct fileproc *fp;

+       struct kqworkq *dealloc_kqwq = NULL;

        int i;
 
        proc_fdlock(p);
 
        if (p == kernproc || NULL == (fdp = p->p_fd)) {
        int i;
 
        proc_fdlock(p);
 
        if (p == kernproc || NULL == (fdp = p->p_fd)) {

-               proc_fdunlock(p);
+               proc_fdunlock(p);

                return;
        }
 
        extern struct filedesc filedesc0;
 
                return;
        }
 
        extern struct filedesc filedesc0;
 

-       if (&filedesc0 == fdp)
+       if (&filedesc0 == fdp) {

                panic("filedesc0");
                panic("filedesc0");

+       }
+
+       /*
+        * deallocate all the knotes up front and claim empty
+        * tables to make any subsequent kqueue closes faster.
+        */
+       knotes_dealloc(p);
+       assert(fdp->fd_knlistsize == 0);
+       assert(fdp->fd_knhashmask == 0);

 
 

+       /*
+        * dealloc all workloops that have outstanding retains
+        * when created with scheduling parameters.
+        */
+       kqworkloops_dealloc(p);
+
+       /* close file descriptors */

        if (fdp->fd_nfiles > 0 && fdp->fd_ofiles) {
        if (fdp->fd_nfiles > 0 && fdp->fd_ofiles) {

-               for (i = fdp->fd_lastfile; i >= 0; i--) {
+               for (i = fdp->fd_lastfile; i >= 0; i--) {

                        if ((fp = fdp->fd_ofiles[i]) != NULL) {
                        if ((fp = fdp->fd_ofiles[i]) != NULL) {

-                         
-                         if (fdp->fd_ofileflags[i] & UF_RESERVED)
-                               panic("fdfree: found fp with UF_RESERVED");
+                               if (fdp->fd_ofileflags[i] & UF_RESERVED) {
+                                       panic("fdfree: found fp with UF_RESERVED");
+                               }

 
 

+                               fileproc_drain(p, fp);

                                procfdtbl_reservefd(p, i);
 
                                procfdtbl_reservefd(p, i);
 

-                               if (i < fdp->fd_knlistsize)
-                                       knote_fdclose(p, i);
-                               if (fp->f_flags & FP_WAITEVENT) 
+                               if (fp->f_flags & FP_WAITEVENT) {

                                        (void)waitevent_close(p, fp);
                                        (void)waitevent_close(p, fp);

+                               }

                                (void) closef_locked(fp, fp->f_fglob, p);
                                fileproc_free(fp);
                        }
                                (void) closef_locked(fp, fp->f_fglob, p);
                                fileproc_free(fp);
                        }


@@ -4677,23 +5254,38 @@ fdfree(proc_t p)
                FREE_ZONE(fdp->fd_ofiles, fdp->fd_nfiles * OFILESIZE, M_OFILETABL);
                fdp->fd_ofiles = NULL;
                fdp->fd_nfiles = 0;
                FREE_ZONE(fdp->fd_ofiles, fdp->fd_nfiles * OFILESIZE, M_OFILETABL);
                fdp->fd_ofiles = NULL;
                fdp->fd_nfiles = 0;

-       }        
+       }
+
+       if (fdp->fd_wqkqueue) {
+               dealloc_kqwq = fdp->fd_wqkqueue;
+               fdp->fd_wqkqueue = NULL;
+       }

 
        proc_fdunlock(p);
 
        proc_fdunlock(p);

-       
-       if (fdp->fd_cdir)
-               vnode_rele(fdp->fd_cdir);
-       if (fdp->fd_rdir)
+
+       if (dealloc_kqwq) {
+               kqworkq_dealloc(dealloc_kqwq);
+       }
+       if (fdp->fd_cdir) {
+               vnode_rele(fdp->fd_cdir);
+       }
+       if (fdp->fd_rdir) {

                vnode_rele(fdp->fd_rdir);
                vnode_rele(fdp->fd_rdir);

+       }

 
        proc_fdlock_spin(p);
        p->p_fd = NULL;
        proc_fdunlock(p);
 
 
        proc_fdlock_spin(p);
        p->p_fd = NULL;
        proc_fdunlock(p);
 

-       if (fdp->fd_knlist)
-               FREE(fdp->fd_knlist, M_KQUEUE);
-       if (fdp->fd_knhash)
-               FREE(fdp->fd_knhash, M_KQUEUE);
+       if (fdp->fd_kqhash) {
+               for (uint32_t j = 0; j <= fdp->fd_kqhashmask; j++) {
+                       assert(LIST_EMPTY(&fdp->fd_kqhash[j]));
+               }
+               FREE(fdp->fd_kqhash, M_KQUEUE);
+       }
+
+       lck_mtx_destroy(&fdp->fd_kqhashlock, proc_kqhashlock_grp);
+       lck_mtx_destroy(&fdp->fd_knhashlock, proc_knhashlock_grp);

 
        FREE_ZONE(fdp, sizeof(*fdp), M_FILEDESC);
 }
 
        FREE_ZONE(fdp, sizeof(*fdp), M_FILEDESC);
 }


@@ -4708,7 +5300,7 @@ fdfree(proc_t p)
  *             p                       Pointer to proc structure
  *
  * Returns:    0                       Success
  *             p                       Pointer to proc structure
  *
  * Returns:    0                       Success

- *     closef_finish:???               Anything returnable by a per-fileops
+ *     closef_finish:???               Anything returnable by a per-fileops

  *                                     close function
  *
  * Note:       Decrements reference count on file structure; if this was the
  *                                     close function
  *
  * Note:       Decrements reference count on file structure; if this was the


@@ -4727,14 +5319,15 @@ closef_locked(struct fileproc *fp, struct fileglob *fg, proc_t p)
        int error;
 
        if (fg == NULL) {
        int error;
 
        if (fg == NULL) {

-               return (0);
+               return 0;

        }
 
        /* Set up context with cred stashed in fg */
        }
 
        /* Set up context with cred stashed in fg */

-       if (p == current_proc())
+       if (p == current_proc()) {

                context.vc_thread = current_thread();
                context.vc_thread = current_thread();

-       else
+       } else {

                context.vc_thread = NULL;
                context.vc_thread = NULL;

+       }

        context.vc_ucred = fg->fg_cred;
 
        /*
        context.vc_ucred = fg->fg_cred;
 
        /*


@@ -4755,7 +5348,7 @@ closef_locked(struct fileproc *fp, struct fileglob *fg, proc_t p)
                lf.l_type = F_UNLCK;
                vp = (struct vnode *)fg->fg_data;
 
                lf.l_type = F_UNLCK;
                vp = (struct vnode *)fg->fg_data;
 

-               if ( (error = vnode_getwithref(vp)) == 0 ) {
+               if ((error = vnode_getwithref(vp)) == 0) {

                        (void) VNOP_ADVLOCK(vp, (caddr_t)p, F_UNLCK, &lf, F_POSIX, &context, NULL);
                        (void)vnode_put(vp);
                }
                        (void) VNOP_ADVLOCK(vp, (caddr_t)p, F_UNLCK, &lf, F_POSIX, &context, NULL);
                        (void)vnode_put(vp);
                }


@@ -4766,32 +5359,36 @@ closef_locked(struct fileproc *fp, struct fileglob *fg, proc_t p)
 
        if (fg->fg_count > 0) {
                lck_mtx_unlock(&fg->fg_lock);
 
        if (fg->fg_count > 0) {
                lck_mtx_unlock(&fg->fg_lock);

-               return (0);
+               return 0;

        }
 #if DIAGNOSTIC
        }
 #if DIAGNOSTIC

-       if (fg->fg_count != 0)
+       if (fg->fg_count != 0) {

                panic("fg %p: being freed with bad fg_count (%d)", fg, fg->fg_count);
                panic("fg %p: being freed with bad fg_count (%d)", fg, fg->fg_count);

+       }

 #endif
 
 #endif
 

-       if (fp && (fp->f_flags & FP_WRITTEN))
-               fg->fg_flag |= FWASWRITTEN;
+       if (fp && (fp->f_flags & FP_WRITTEN)) {
+               fg->fg_flag |= FWASWRITTEN;
+       }

 
        fg->fg_lflags |= FG_TERM;
        lck_mtx_unlock(&fg->fg_lock);
 
 
        fg->fg_lflags |= FG_TERM;
        lck_mtx_unlock(&fg->fg_lock);
 

-       if (p)
+       if (p) {

                proc_fdunlock(p);
                proc_fdunlock(p);

+       }

 
 

-       /* Since we ensure that fg->fg_ops is always initialized, 
+       /* Since we ensure that fg->fg_ops is always initialized,

         * it is safe to invoke fo_close on the fg */
        error = fo_close(fg, &context);
 
        fg_free(fg);
         * it is safe to invoke fo_close on the fg */
        error = fo_close(fg, &context);
 
        fg_free(fg);

-       
-       if (p)
+
+       if (p) {

                proc_fdlock(p);
                proc_fdlock(p);

+       }

 
 

-       return(error);
+       return error;

 }
 
 
 }
 
 


@@ -4813,7 +5410,7 @@ closef_locked(struct fileproc *fp, struct fileglob *fg, proc_t p)
  *             close to prevent fd's from being closed out from under
  *             operations currently in progress and blocked
  *
  *             close to prevent fd's from being closed out from under
  *             operations currently in progress and blocked
  *

- * See Also:   file_vnode(), file_socket(), file_drop(), and the cautions
+ * See Also:    file_vnode(), file_socket(), file_drop(), and the cautions

  *             regarding their use and interaction with this function.
  */
 void
  *             regarding their use and interaction with this function.
  */
 void


@@ -4821,37 +5418,40 @@ fileproc_drain(proc_t p, struct fileproc * fp)
 {
        struct vfs_context context;
 
 {
        struct vfs_context context;
 

-       context.vc_thread = proc_thread(p);     /* XXX */
+       context.vc_thread = proc_thread(p);     /* XXX */

        context.vc_ucred = fp->f_fglob->fg_cred;
 
        context.vc_ucred = fp->f_fglob->fg_cred;
 

-       fp->f_iocount-- ; /* (the one the close holds) */
-
-       while (fp->f_iocount) {
+       /* Set the vflag for drain */
+       fileproc_modify_vflags(fp, FPV_DRAIN, FALSE);

 
 

-               lck_mtx_convert_spin(&p->p_fdmlock);
+       while (os_ref_get_count(&fp->f_iocount) > 1) {
+               lck_mtx_convert_spin(&p->p_fdmlock);

 
 

-               if (fp->f_fglob->fg_ops->fo_drain) {
-                       (*fp->f_fglob->fg_ops->fo_drain)(fp, &context);
-               }
+               fo_drain(fp, &context);

                if ((fp->f_flags & FP_INSELECT) == FP_INSELECT) {
                if ((fp->f_flags & FP_INSELECT) == FP_INSELECT) {

-                       if (wait_queue_wakeup_all((wait_queue_t)fp->f_waddr, NULL, THREAD_INTERRUPTED) == KERN_INVALID_ARGUMENT)
-                               panic("bad wait queue for wait_queue_wakeup_all %p", fp->f_waddr);
-               } 
+                       if (waitq_wakeup64_all((struct waitq *)fp->f_wset, NO_EVENT64,
+                           THREAD_INTERRUPTED, WAITQ_ALL_PRIORITIES) == KERN_INVALID_ARGUMENT) {
+                               panic("bad wait queue for waitq_wakeup64_all %p (fp:%p)", fp->f_wset, fp);
+                       }
+               }

                if ((fp->f_flags & FP_SELCONFLICT) == FP_SELCONFLICT) {
                if ((fp->f_flags & FP_SELCONFLICT) == FP_SELCONFLICT) {

-                       if (wait_queue_wakeup_all(&select_conflict_queue, NULL, THREAD_INTERRUPTED) == KERN_INVALID_ARGUMENT)
+                       if (waitq_wakeup64_all(&select_conflict_queue, NO_EVENT64,
+                           THREAD_INTERRUPTED, WAITQ_ALL_PRIORITIES) == KERN_INVALID_ARGUMENT) {

                                panic("bad select_conflict_queue");
                                panic("bad select_conflict_queue");

+                       }

                }
                p->p_fpdrainwait = 1;
 
                msleep(&p->p_fpdrainwait, &p->p_fdmlock, PRIBIO, "fpdrain", NULL);
                }
                p->p_fpdrainwait = 1;
 
                msleep(&p->p_fpdrainwait, &p->p_fdmlock, PRIBIO, "fpdrain", NULL);

-

        }
 #if DIAGNOSTIC
        }
 #if DIAGNOSTIC

-       if ((fp->f_flags & FP_INSELECT) != 0)
+       if ((fp->f_flags & FP_INSELECT) != 0) {

                panic("FP_INSELECT set on drained fp");
                panic("FP_INSELECT set on drained fp");

+       }

 #endif
 #endif

-       if ((fp->f_flags & FP_SELCONFLICT) == FP_SELCONFLICT)
+       if ((fp->f_flags & FP_SELCONFLICT) == FP_SELCONFLICT) {

                fp->f_flags &= ~FP_SELCONFLICT;
                fp->f_flags &= ~FP_SELCONFLICT;

+       }

 }
 
 
 }
 
 


@@ -4865,22 +5465,17 @@ fileproc_drain(proc_t p, struct fileproc * fp)
  * Parameters: p                               Process containing fd
  *             fd                              fd to be released
  *             fp                              fileproc to be freed
  * Parameters: p                               Process containing fd
  *             fd                              fd to be released
  *             fp                              fileproc to be freed

- *
- * Returns:    0                               Success
- *
- * Notes:      XXX function should be void - no one interprets the returns
- *             XXX code

  */
  */

-int
+void

 fp_free(proc_t p, int fd, struct fileproc * fp)
 {
 fp_free(proc_t p, int fd, struct fileproc * fp)
 {

-        proc_fdlock_spin(p);
+       proc_fdlock_spin(p);

        fdrelse(p, fd);
        fdrelse(p, fd);

-        proc_fdunlock(p);
+       proc_fdunlock(p);

 
        fg_free(fp->f_fglob);
 
        fg_free(fp->f_fglob);

+       os_ref_release_live(&fp->f_iocount);

        fileproc_free(fp);
        fileproc_free(fp);

-       return(0);

 }
 
 
 }
 
 


@@ -4894,7 +5489,7 @@ fp_free(proc_t p, int fd, struct fileproc * fp)
  *                                             attempted
  *             uap->how                        (Un)Lock bits, including type
  *             retval                          Pointer to the call return area
  *                                             attempted
  *             uap->how                        (Un)Lock bits, including type
  *             retval                          Pointer to the call return area

- *             
+ *

  * Returns:    0                               Success
  *     fp_getfvp:EBADF                         Bad file descriptor
  *     fp_getfvp:ENOTSUP                       fd does not refer to a vnode
  * Returns:    0                               Success
  *     fp_getfvp:EBADF                         Bad file descriptor
  *     fp_getfvp:ENOTSUP                       fd does not refer to a vnode


@@ -4916,13 +5511,13 @@ flock(proc_t p, struct flock_args *uap, __unused int32_t *retval)
        struct vnode *vp;
        struct flock lf;
        vfs_context_t ctx = vfs_context_current();
        struct vnode *vp;
        struct flock lf;
        vfs_context_t ctx = vfs_context_current();

-       int error=0;
+       int error = 0;

 
        AUDIT_ARG(fd, uap->fd);
 
        AUDIT_ARG(fd, uap->fd);

-       if ( (error = fp_getfvp(p, fd, &fp, &vp)) ) {
-               return(error);
+       if ((error = fp_getfvp(p, fd, &fp, &vp))) {
+               return error;

        }
        }

-       if ( (error = vnode_getwithref(vp)) ) {
+       if ((error = vnode_getwithref(vp))) {

                goto out1;
        }
        AUDIT_ARG(vnpath, vp, ARG_VNODE1);
                goto out1;
        }
        AUDIT_ARG(vnpath, vp, ARG_VNODE1);


@@ -4936,31 +5531,31 @@ flock(proc_t p, struct flock_args *uap, __unused int32_t *retval)
                error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob, F_UNLCK, &lf, F_FLOCK, ctx, NULL);
                goto out;
        }
                error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob, F_UNLCK, &lf, F_FLOCK, ctx, NULL);
                goto out;
        }

-       if (how & LOCK_EX)
+       if (how & LOCK_EX) {

                lf.l_type = F_WRLCK;
                lf.l_type = F_WRLCK;

-       else if (how & LOCK_SH)
+       } else if (how & LOCK_SH) {

                lf.l_type = F_RDLCK;
                lf.l_type = F_RDLCK;

-       else {
-               error = EBADF;
+       } else {
+               error = EBADF;

                goto out;
        }
 #if CONFIG_MACF
        error = mac_file_check_lock(proc_ucred(p), fp->f_fglob, F_SETLK, &lf);
                goto out;
        }
 #if CONFIG_MACF
        error = mac_file_check_lock(proc_ucred(p), fp->f_fglob, F_SETLK, &lf);

-       if (error)
+       if (error) {

                goto out;
                goto out;

+       }

 #endif
 #endif

-       fp->f_flag |= FHASLOCK;
-       if (how & LOCK_NB) {
-               error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob, F_SETLK, &lf, F_FLOCK, ctx, NULL);
-               goto out;       
+       error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob, F_SETLK, &lf,
+           (how & LOCK_NB ? F_FLOCK : F_FLOCK | F_WAIT),
+           ctx, NULL);
+       if (!error) {
+               fp->f_flag |= FHASLOCK;

        }
        }

-       error = VNOP_ADVLOCK(vp, (caddr_t)fp->f_fglob, F_SETLK, &lf, F_FLOCK|F_WAIT, ctx, NULL);

 out:
        (void)vnode_put(vp);
 out1:
        fp_drop(p, fd, fp, 0);
 out:
        (void)vnode_put(vp);
 out1:
        fp_drop(p, fd, fp, 0);

-       return(error);
-
+       return error;

 }
 
 /*
 }
 
 /*


@@ -4969,17 +5564,17 @@ out1:
  * Description: Obtain a Mach send right for a given file descriptor.
  *
  * Parameters: p               Process calling fileport
  * Description: Obtain a Mach send right for a given file descriptor.
  *
  * Parameters: p               Process calling fileport

- *             uap->fd         The fd to reference
- *             uap->portnamep  User address at which to place port name.
+ *              uap->fd                The fd to reference
+ *              uap->portnamep  User address at which to place port name.

  *
  * Returns:    0               Success.
  *
  * Returns:    0               Success.

- *                     EBADF           Bad file descriptor.
- *                     EINVAL          File descriptor had type that cannot be sent, misc. other errors.
- *                     EFAULT          Address at which to store port name is not valid.
- *                     EAGAIN          Resource shortage.
+ *              EBADF          Bad file descriptor.
+ *              EINVAL         File descriptor had type that cannot be sent, misc. other errors.
+ *              EFAULT         Address at which to store port name is not valid.
+ *              EAGAIN         Resource shortage.

  *
  * Implicit returns:
  *
  * Implicit returns:

- *             On success, name of send right is stored at user-specified address.             
+ *             On success, name of send right is stored at user-specified address.

  */
 int
 fileport_makeport(proc_t p, struct fileport_makeport_args *uap,
  */
 int
 fileport_makeport(proc_t p, struct fileport_makeport_args *uap,


@@ -4993,27 +5588,28 @@ fileport_makeport(proc_t p, struct fileport_makeport_args *uap,
        ipc_port_t fileport;
        mach_port_name_t name = MACH_PORT_NULL;
 
        ipc_port_t fileport;
        mach_port_name_t name = MACH_PORT_NULL;
 

-       err = fp_lookup(p, fd, &fp, 0);
+       proc_fdlock(p);
+       err = fp_lookup(p, fd, &fp, 1);

        if (err != 0) {
        if (err != 0) {

-               goto out;
+               goto out_unlock;

        }
 
        }
 

-       if (!filetype_issendable(fp->f_type)) {
+       if (!file_issendable(p, fp)) {

                err = EINVAL;
                err = EINVAL;

-               goto out;
+               goto out_unlock;

        }
 
        if (FP_ISGUARDED(fp, GUARD_FILEPORT)) {
        }
 
        if (FP_ISGUARDED(fp, GUARD_FILEPORT)) {

-               proc_fdlock(p);

                err = fp_guard_exception(p, fd, fp, kGUARD_EXC_FILEPORT);
                err = fp_guard_exception(p, fd, fp, kGUARD_EXC_FILEPORT);

-               proc_fdunlock(p);
-               goto out;
+               goto out_unlock;

        }
 
        /* Dropped when port is deallocated */
        fg = fp->f_fglob;
        fg_ref(fp);
 
        }
 
        /* Dropped when port is deallocated */
        fg = fp->f_fglob;
        fg_ref(fp);
 

+       proc_fdunlock(p);
+

        /* Allocate and initialize a port */
        fileport = fileport_alloc(fg);
        if (fileport == IPC_PORT_NULL) {
        /* Allocate and initialize a port */
        fileport = fileport_alloc(fg);
        if (fileport == IPC_PORT_NULL) {


@@ -5021,14 +5617,14 @@ fileport_makeport(proc_t p, struct fileport_makeport_args *uap,
                fg_drop(fp);
                goto out;
        }
                fg_drop(fp);
                goto out;
        }

-       
+

        /* Add an entry.  Deallocates port on failure. */
        name = ipc_port_copyout_send(fileport, get_task_ipcspace(p->task));
        if (!MACH_PORT_VALID(name)) {
                err = EINVAL;
                goto out;
        /* Add an entry.  Deallocates port on failure. */
        name = ipc_port_copyout_send(fileport, get_task_ipcspace(p->task));
        if (!MACH_PORT_VALID(name)) {
                err = EINVAL;
                goto out;

-       } 
-       
+       }
+

        err = copyout(&name, user_portaddr, sizeof(mach_port_name_t));
        if (err != 0) {
                goto out;
        err = copyout(&name, user_portaddr, sizeof(mach_port_name_t));
        if (err != 0) {
                goto out;


@@ -5043,6 +5639,8 @@ fileport_makeport(proc_t p, struct fileport_makeport_args *uap,
 
        return 0;
 
 
        return 0;
 

+out_unlock:
+       proc_fdunlock(p);

 out:
        if (MACH_PORT_VALID(name)) {
                /* Don't care if another thread races us to deallocate the entry */
 out:
        if (MACH_PORT_VALID(name)) {
                /* Don't care if another thread races us to deallocate the entry */


@@ -5064,15 +5662,11 @@ fileport_releasefg(struct fileglob *fg)
        return;
 }
 
        return;
 }
 

-

 /*
 /*

- * fileport_makefd
+ * fileport_makefd_internal

  *
  * Description: Obtain the file descriptor for a given Mach send right.
  *
  *
  * Description: Obtain the file descriptor for a given Mach send right.
  *

- * Parameters: p               Process calling fileport
- *             uap->port       Name of send right to file port.
- *

  * Returns:    0               Success
  *             EINVAL          Invalid Mach port name, or port is not for a file.
  *     fdalloc:EMFILE
  * Returns:    0               Success
  *             EINVAL          Invalid Mach port name, or port is not for a file.
  *     fdalloc:EMFILE


@@ -5082,24 +5676,13 @@ fileport_releasefg(struct fileglob *fg)
  *             *retval (modified)              The new descriptor
  */
 int
  *             *retval (modified)              The new descriptor
  */
 int

-fileport_makefd(proc_t p, struct fileport_makefd_args *uap, int32_t *retval)
+fileport_makefd_internal(proc_t p, ipc_port_t port, int uf_flags, int *retval)

 {
        struct fileglob *fg;
 {
        struct fileglob *fg;

-       struct fileproc *fp = FILEPROC_NULL;
-       ipc_port_t port = IPC_PORT_NULL;
-       mach_port_name_t send = uap->port;
-       kern_return_t res;
+       struct fileproc *fp = FILEPROC_NULL;

        int fd;
        int err;
 
        int fd;
        int err;
 

-       res = ipc_object_copyin(get_task_ipcspace(p->task),
-                       send, MACH_MSG_TYPE_COPY_SEND, &port);
-
-       if (res != KERN_SUCCESS) {
-               err = EINVAL;
-               goto out;
-       }
-

        fg = fileport_port_to_fileglob(port);
        if (fg == NULL) {
                err = EINVAL;
        fg = fileport_port_to_fileglob(port);
        if (fg == NULL) {
                err = EINVAL;


@@ -5115,13 +5698,16 @@ fileport_makefd(proc_t p, struct fileport_makefd_args *uap, int32_t *retval)
        fp->f_fglob = fg;
        fg_ref(fp);
 
        fp->f_fglob = fg;
        fg_ref(fp);
 

-       proc_fdlock(p);
+       proc_fdlock(p);

        err = fdalloc(p, 0, &fd);
        if (err != 0) {
                proc_fdunlock(p);
        err = fdalloc(p, 0, &fd);
        if (err != 0) {
                proc_fdunlock(p);

+               fg_drop(fp);

                goto out;
        }
                goto out;
        }

-       *fdflags(p, fd) |= UF_EXCLOSE;
+       if (uf_flags) {
+               *fdflags(p, fd) |= uf_flags;
+       }

 
        procfdtbl_releasefd(p, fd, fp);
        proc_fdunlock(p);
 
        procfdtbl_releasefd(p, fd, fp);
        proc_fdunlock(p);


@@ -5131,7 +5717,43 @@ fileport_makefd(proc_t p, struct fileport_makefd_args *uap, int32_t *retval)
 out:
        if ((fp != NULL) && (0 != err)) {
                fileproc_free(fp);
 out:
        if ((fp != NULL) && (0 != err)) {
                fileproc_free(fp);

-       } 
+       }
+
+       return err;
+}
+
+/*
+ * fileport_makefd
+ *
+ * Description: Obtain the file descriptor for a given Mach send right.
+ *
+ * Parameters: p               Process calling fileport
+ *              uap->port      Name of send right to file port.
+ *
+ * Returns:    0               Success
+ *             EINVAL          Invalid Mach port name, or port is not for a file.
+ *     fdalloc:EMFILE
+ *     fdalloc:ENOMEM          Unable to allocate fileproc or extend file table.
+ *
+ * Implicit returns:
+ *             *retval (modified)              The new descriptor
+ */
+int
+fileport_makefd(proc_t p, struct fileport_makefd_args *uap, int32_t *retval)
+{
+       ipc_port_t port = IPC_PORT_NULL;
+       mach_port_name_t send = uap->port;
+       kern_return_t res;
+       int err;
+
+       res = ipc_object_copyin(get_task_ipcspace(p->task),
+           send, MACH_MSG_TYPE_COPY_SEND, &port, 0, NULL, IPC_KMSG_FLAGS_ALLOW_IMMOVABLE_SEND);
+
+       if (res == KERN_SUCCESS) {
+               err = fileport_makefd_internal(p, port, UF_EXCLOSE, retval);
+       } else {
+               err = EINVAL;
+       }

 
        if (IPC_PORT_NULL != port) {
                ipc_port_release_send(port);
 
        if (IPC_PORT_NULL != port) {
                ipc_port_release_send(port);


@@ -5182,17 +5804,16 @@ dupfdopen(struct filedesc *fdp, int indx, int dfd, int flags, int error)
 
        fp = fdp->fd_ofiles[indx];
        if (dfd < 0 || dfd >= fdp->fd_nfiles ||
 
        fp = fdp->fd_ofiles[indx];
        if (dfd < 0 || dfd >= fdp->fd_nfiles ||

-                       (wfp = fdp->fd_ofiles[dfd]) == NULL || wfp == fp ||
-                       (fdp->fd_ofileflags[dfd] & UF_RESERVED)) {
-
-               proc_fdunlock(p);
-               return (EBADF);
+           (wfp = fdp->fd_ofiles[dfd]) == NULL || wfp == fp ||
+           (fdp->fd_ofileflags[dfd] & UF_RESERVED)) {
+               proc_fdunlock(p);
+               return EBADF;

        }
 #if CONFIG_MACF
        myerror = mac_file_check_dup(proc_ucred(p), wfp->f_fglob, dfd);
        if (myerror) {
                proc_fdunlock(p);
        }
 #if CONFIG_MACF
        myerror = mac_file_check_dup(proc_ucred(p), wfp->f_fglob, dfd);
        if (myerror) {
                proc_fdunlock(p);

-               return (myerror);
+               return myerror;

        }
 #endif
        /*
        }
 #endif
        /*


@@ -5210,37 +5831,37 @@ dupfdopen(struct filedesc *fdp, int indx, int dfd, int flags, int error)
        switch (error) {
        case ENODEV:
                if (FP_ISGUARDED(wfp, GUARD_DUP)) {
        switch (error) {
        case ENODEV:
                if (FP_ISGUARDED(wfp, GUARD_DUP)) {

-                       int err = fp_guard_exception(p,
-                           dfd, wfp, kGUARD_EXC_DUP);

                        proc_fdunlock(p);
                        proc_fdunlock(p);

-                       return (err);
+                       return EPERM;

                }
 
                /*
                 * Check that the mode the file is being opened for is a
                 * subset of the mode of the existing descriptor.
                 */
                }
 
                /*
                 * Check that the mode the file is being opened for is a
                 * subset of the mode of the existing descriptor.
                 */

-               if (((flags & (FREAD|FWRITE)) | wfp->f_flag) != wfp->f_flag) {
-                       proc_fdunlock(p);
-                       return (EACCES);
+               if (((flags & (FREAD | FWRITE)) | wfp->f_flag) != wfp->f_flag) {
+                       proc_fdunlock(p);
+                       return EACCES;

                }
                }

-               if (indx > fdp->fd_lastfile)
+               if (indx > fdp->fd_lastfile) {

                        fdp->fd_lastfile = indx;
                        fdp->fd_lastfile = indx;

+               }

                (void)fg_ref(wfp);
 
                (void)fg_ref(wfp);
 

-               if (fp->f_fglob)
-                       fg_free(fp->f_fglob);
+               if (fp->f_fglob) {
+                       fg_free(fp->f_fglob);
+               }

                fp->f_fglob = wfp->f_fglob;
 
                fdp->fd_ofileflags[indx] = fdp->fd_ofileflags[dfd] |
                fp->f_fglob = wfp->f_fglob;
 
                fdp->fd_ofileflags[indx] = fdp->fd_ofileflags[dfd] |

-                       (flags & O_CLOEXEC) ? UF_EXCLOSE : 0;
+                   (flags & O_CLOEXEC) ? UF_EXCLOSE : 0;

 
 

-               proc_fdunlock(p);
-               return (0);
+               proc_fdunlock(p);
+               return 0;

 
        default:
 
        default:

-               proc_fdunlock(p);
-               return (error);
+               proc_fdunlock(p);
+               return error;

        }
        /* NOTREACHED */
 }
        }
        /* NOTREACHED */
 }


@@ -5268,12 +5889,14 @@ fg_ref(struct fileproc * fp)
        lck_mtx_lock_spin(&fg->fg_lock);
 
 #if DIAGNOSTIC
        lck_mtx_lock_spin(&fg->fg_lock);
 
 #if DIAGNOSTIC

-       if ((fp->f_flags & ~((unsigned int)FP_VALID_FLAGS)) != 0)
+       if ((fp->f_flags & ~((unsigned int)FP_VALID_FLAGS)) != 0) {

                panic("fg_ref: invalid bits on fp %p", fp);
                panic("fg_ref: invalid bits on fp %p", fp);

+       }

 
 

-       if (fg->fg_count == 0)
+       if (fg->fg_count == 0) {

                panic("fg_ref: adding fgcount to zeroed fg: fp %p fg %p",
                    fp, fg);
                panic("fg_ref: adding fgcount to zeroed fg: fp %p fg %p",
                    fp, fg);

+       }

 #endif
        fg->fg_count++;
        lck_mtx_unlock(&fg->fg_lock);
 #endif
        fg->fg_count++;
        lck_mtx_unlock(&fg->fg_lock);


@@ -5305,23 +5928,23 @@ fg_drop(struct fileproc * fp)
 
 #if SOCKETS
 /*
 
 #if SOCKETS
 /*

- * fg_insertuipc
+ * fg_insertuipc_mark

  *
  *

- * Description:        Insert fileglob onto message queue
+ * Description:        Mark fileglob for insertion onto message queue if needed
+ *             Also takes fileglob reference

  *
  *

- * Parameters: fg                              Fileglob pointer to insert
+ * Parameters: fg      Fileglob pointer to insert

  *
  *

- * Returns:    void
+ * Returns:    true, if the fileglob needs to be inserted onto msg queue

  *
  * Locks:      Takes and drops fg_lock, potentially many times
  */
  *
  * Locks:      Takes and drops fg_lock, potentially many times
  */

-void
-fg_insertuipc(struct fileglob * fg)
+boolean_t
+fg_insertuipc_mark(struct fileglob * fg)

 {
 {

-       int insertque = 0;
+       boolean_t insert = FALSE;

 
        lck_mtx_lock_spin(&fg->fg_lock);
 
        lck_mtx_lock_spin(&fg->fg_lock);

-

        while (fg->fg_lflags & FG_RMMSGQ) {
                lck_mtx_convert_spin(&fg->fg_lock);
 
        while (fg->fg_lflags & FG_RMMSGQ) {
                lck_mtx_convert_spin(&fg->fg_lock);
 


@@ -5333,11 +5956,30 @@ fg_insertuipc(struct fileglob * fg)
        fg->fg_msgcount++;
        if (fg->fg_msgcount == 1) {
                fg->fg_lflags |= FG_INSMSGQ;
        fg->fg_msgcount++;
        if (fg->fg_msgcount == 1) {
                fg->fg_lflags |= FG_INSMSGQ;

-               insertque=1;
+               insert = TRUE;

        }
        lck_mtx_unlock(&fg->fg_lock);
        }
        lck_mtx_unlock(&fg->fg_lock);

+       return insert;
+}

 
 

-       if (insertque) {
+/*
+ * fg_insertuipc
+ *
+ * Description:        Insert marked fileglob onto message queue
+ *
+ * Parameters: fg      Fileglob pointer to insert
+ *
+ * Returns:    void
+ *
+ * Locks:      Takes and drops fg_lock & uipc_lock
+ *             DO NOT call this function with proc_fdlock held as unp_gc()
+ *             can potentially try to acquire proc_fdlock, which can result
+ *             in a deadlock if this function is in unp_gc_wait().
+ */
+void
+fg_insertuipc(struct fileglob * fg)
+{
+       if (fg->fg_lflags & FG_INSMSGQ) {

                lck_mtx_lock_spin(uipc_lock);
                unp_gc_wait();
                LIST_INSERT_HEAD(&fmsghead, fg, f_msglist);
                lck_mtx_lock_spin(uipc_lock);
                unp_gc_wait();
                LIST_INSERT_HEAD(&fmsghead, fg, f_msglist);


@@ -5350,25 +5992,24 @@ fg_insertuipc(struct fileglob * fg)
                }
                lck_mtx_unlock(&fg->fg_lock);
        }
                }
                lck_mtx_unlock(&fg->fg_lock);
        }

-

 }
 
 }
 

-

 /*
 /*

- * fg_removeuipc
+ * fg_removeuipc_mark

  *
  *

- * Description:        Remove fileglob from message queue
+ * Description:        Mark the fileglob for removal from message queue if needed
+ *             Also releases fileglob message queue reference

  *
  *

- * Parameters: fg                              Fileglob pointer to remove
+ * Parameters: fg      Fileglob pointer to remove

  *
  *

- * Returns:    void
+ * Returns:    true, if the fileglob needs to be removed from msg queue

  *
  * Locks:      Takes and drops fg_lock, potentially many times
  */
  *
  * Locks:      Takes and drops fg_lock, potentially many times
  */

-void
-fg_removeuipc(struct fileglob * fg)
+boolean_t
+fg_removeuipc_mark(struct fileglob * fg)

 {
 {

-       int removeque = 0;
+       boolean_t remove = FALSE;

 
        lck_mtx_lock_spin(&fg->fg_lock);
        while (fg->fg_lflags & FG_INSMSGQ) {
 
        lck_mtx_lock_spin(&fg->fg_lock);
        while (fg->fg_lflags & FG_INSMSGQ) {


@@ -5380,11 +6021,30 @@ fg_removeuipc(struct fileglob * fg)
        fg->fg_msgcount--;
        if (fg->fg_msgcount == 0) {
                fg->fg_lflags |= FG_RMMSGQ;
        fg->fg_msgcount--;
        if (fg->fg_msgcount == 0) {
                fg->fg_lflags |= FG_RMMSGQ;

-               removeque=1;
+               remove = TRUE;

        }
        lck_mtx_unlock(&fg->fg_lock);
        }
        lck_mtx_unlock(&fg->fg_lock);

+       return remove;
+}

 
 

-       if (removeque) {
+/*
+ * fg_removeuipc
+ *
+ * Description:        Remove marked fileglob from message queue
+ *
+ * Parameters: fg      Fileglob pointer to remove
+ *
+ * Returns:    void
+ *
+ * Locks:      Takes and drops fg_lock & uipc_lock
+ *             DO NOT call this function with proc_fdlock held as unp_gc()
+ *             can potentially try to acquire proc_fdlock, which can result
+ *             in a deadlock if this function is in unp_gc_wait().
+ */
+void
+fg_removeuipc(struct fileglob * fg)
+{
+       if (fg->fg_lflags & FG_RMMSGQ) {

                lck_mtx_lock_spin(uipc_lock);
                unp_gc_wait();
                LIST_REMOVE(fg, f_msglist);
                lck_mtx_lock_spin(uipc_lock);
                unp_gc_wait();
                LIST_REMOVE(fg, f_msglist);


@@ -5417,7 +6077,14 @@ fg_removeuipc(struct fileglob * fg)
 int
 fo_read(struct fileproc *fp, struct uio *uio, int flags, vfs_context_t ctx)
 {
 int
 fo_read(struct fileproc *fp, struct uio *uio, int flags, vfs_context_t ctx)
 {

-       return ((*fp->f_ops->fo_read)(fp, uio, flags, ctx));
+       return (*fp->f_ops->fo_read)(fp, uio, flags, ctx);
+}
+
+int
+fo_no_read(struct fileproc *fp, struct uio *uio, int flags, vfs_context_t ctx)
+{
+#pragma unused(fp, uio, flags, ctx)
+       return ENXIO;

 }
 
 
 }
 
 


@@ -5438,7 +6105,14 @@ fo_read(struct fileproc *fp, struct uio *uio, int flags, vfs_context_t ctx)
 int
 fo_write(struct fileproc *fp, struct uio *uio, int flags, vfs_context_t ctx)
 {
 int
 fo_write(struct fileproc *fp, struct uio *uio, int flags, vfs_context_t ctx)
 {

-       return((*fp->f_ops->fo_write)(fp, uio, flags, ctx));
+       return (*fp->f_ops->fo_write)(fp, uio, flags, ctx);
+}
+
+int
+fo_no_write(struct fileproc *fp, struct uio *uio, int flags, vfs_context_t ctx)
+{
+#pragma unused(fp, uio, flags, ctx)
+       return ENXIO;

 }
 
 
 }
 
 


@@ -5464,7 +6138,7 @@ fo_write(struct fileproc *fp, struct uio *uio, int flags, vfs_context_t ctx)
  *             function, it will need to revalidate/reacquire any cached
  *             protected data obtained prior to the call.
  */
  *             function, it will need to revalidate/reacquire any cached
  *             protected data obtained prior to the call.
  */

-int 
+int

 fo_ioctl(struct fileproc *fp, u_long com, caddr_t data, vfs_context_t ctx)
 {
        int error;
 fo_ioctl(struct fileproc *fp, u_long com, caddr_t data, vfs_context_t ctx)
 {
        int error;


@@ -5472,8 +6146,15 @@ fo_ioctl(struct fileproc *fp, u_long com, caddr_t data, vfs_context_t ctx)
        proc_fdunlock(vfs_context_proc(ctx));
        error = (*fp->f_ops->fo_ioctl)(fp, com, data, ctx);
        proc_fdlock(vfs_context_proc(ctx));
        proc_fdunlock(vfs_context_proc(ctx));
        error = (*fp->f_ops->fo_ioctl)(fp, com, data, ctx);
        proc_fdlock(vfs_context_proc(ctx));

-       return(error);
-}       
+       return error;
+}
+
+int
+fo_no_ioctl(struct fileproc *fp, u_long com, caddr_t data, vfs_context_t ctx)
+{
+#pragma unused(fp, com, data, ctx)
+       return ENOTTY;
+}

 
 
 /*
 
 
 /*


@@ -5492,8 +6173,15 @@ fo_ioctl(struct fileproc *fp, u_long com, caddr_t data, vfs_context_t ctx)
  */
 int
 fo_select(struct fileproc *fp, int which, void *wql, vfs_context_t ctx)
  */
 int
 fo_select(struct fileproc *fp, int which, void *wql, vfs_context_t ctx)

-{       
-       return((*fp->f_ops->fo_select)(fp, which, wql, ctx));
+{
+       return (*fp->f_ops->fo_select)(fp, which, wql, ctx);
+}
+
+int
+fo_no_select(struct fileproc *fp, int which, void *wql, vfs_context_t ctx)
+{
+#pragma unused(fp, which, wql, ctx)
+       return ENOTSUP;

 }
 
 
 }
 
 


@@ -5512,68 +6200,117 @@ fo_select(struct fileproc *fp, int which, void *wql, vfs_context_t ctx)
  */
 int
 fo_close(struct fileglob *fg, vfs_context_t ctx)
  */
 int
 fo_close(struct fileglob *fg, vfs_context_t ctx)

-{       
-       return((*fg->fg_ops->fo_close)(fg, ctx));
+{
+       return (*fg->fg_ops->fo_close)(fg, ctx);

 }
 
 
 /*
 }
 
 
 /*

- * fo_kqfilter
+ * fo_drain

  *
  * Description:        Generic fileops kqueue filter indirected through the fileops
  *             pointer in the fileproc structure
  *
  * Parameters: fp                              fileproc structure pointer
  *
  * Description:        Generic fileops kqueue filter indirected through the fileops
  *             pointer in the fileproc structure
  *
  * Parameters: fp                              fileproc structure pointer

- *             kn                              pointer to knote to filter on

  *             ctx                             VFS context for operation
  *
  * Returns:    0                               Success
  *             ctx                             VFS context for operation
  *
  * Returns:    0                               Success

- *             !0                              Errno from kqueue filter
+ *             !0                              errno from drain
+ */
+int
+fo_drain(struct fileproc *fp, vfs_context_t ctx)
+{
+       return (*fp->f_ops->fo_drain)(fp, ctx);
+}
+
+int
+fo_no_drain(struct fileproc *fp, vfs_context_t ctx)
+{
+#pragma unused(fp, ctx)
+       return ENOTSUP;
+}
+
+
+/*
+ * fo_kqfilter
+ *
+ * Description:        Generic fileops kqueue filter indirected through the fileops
+ *             pointer in the fileproc structure
+ *
+ * Parameters: fp                              fileproc structure pointer
+ *             kn                              pointer to knote to filter on
+ *
+ * Returns:    (kn->kn_flags & EV_ERROR)       error in kn->kn_data
+ *             0                               Filter is not active
+ *             !0                              Filter is active

  */
 int
  */
 int

-fo_kqfilter(struct fileproc *fp, struct knote *kn, vfs_context_t ctx)
+fo_kqfilter(struct fileproc *fp, struct knote *kn, struct kevent_qos_s *kev)
+{
+       return (*fp->f_ops->fo_kqfilter)(fp, kn, kev);
+}
+
+int
+fo_no_kqfilter(struct fileproc *fp, struct knote *kn, struct kevent_qos_s *kev)

 {
 {

-        return ((*fp->f_ops->fo_kqfilter)(fp, kn, ctx));
+#pragma unused(fp, kev)
+       knote_set_error(kn, ENOTSUP);
+       return 0;

 }
 
 }
 

+

 /*
  * The ability to send a file descriptor to another
  * process is opt-in by file type.
  */
 boolean_t
 /*
  * The ability to send a file descriptor to another
  * process is opt-in by file type.
  */
 boolean_t

-filetype_issendable(file_type_t fdtype) 
+file_issendable(proc_t p, struct fileproc *fp)

 {
 {

-       switch (fdtype) {
-               case DTYPE_VNODE:
-               case DTYPE_SOCKET:
-               case DTYPE_PIPE:
-               case DTYPE_PSXSHM:
-                       return TRUE;
-               default:
-                       /* DTYPE_KQUEUE, DTYPE_FSEVENTS, DTYPE_PSXSEM */
-                       return FALSE;
+       proc_fdlock_assert(p, LCK_MTX_ASSERT_OWNED);
+
+       switch (fp->f_type) {
+       case DTYPE_VNODE:
+       case DTYPE_SOCKET:
+       case DTYPE_PIPE:
+       case DTYPE_PSXSHM:
+       case DTYPE_NETPOLICY:
+               return 0 == (fp->f_fglob->fg_lflags & FG_CONFINED);
+       default:
+               /* DTYPE_KQUEUE, DTYPE_FSEVENTS, DTYPE_PSXSEM */
+               return FALSE;

        }
 }
 
        }
 }
 

+os_refgrp_decl(, f_iocount_refgrp, "f_iocount", NULL);

 
 struct fileproc *
 fileproc_alloc_init(__unused void *arg)
 {
        struct fileproc *fp;
 
 
 struct fileproc *
 fileproc_alloc_init(__unused void *arg)
 {
        struct fileproc *fp;
 

-       MALLOC_ZONE(fp, struct fileproc *, sizeof (*fp), M_FILEPROC, M_WAITOK);
-       if (fp)
-               bzero(fp, sizeof (*fp));
+       MALLOC_ZONE(fp, struct fileproc *, sizeof(*fp), M_FILEPROC, M_WAITOK);
+       if (fp) {
+               bzero(fp, sizeof(*fp));
+               os_ref_init(&fp->f_iocount, &f_iocount_refgrp);
+       }

 
 

-       return (fp);
+       return fp;

 }
 
 }
 

+

 void
 fileproc_free(struct fileproc *fp)
 {
 void
 fileproc_free(struct fileproc *fp)
 {

+       os_ref_count_t __unused refc = os_ref_release(&fp->f_iocount);
+#if DEVELOPMENT || DEBUG
+       if (0 != refc) {
+               panic("%s: pid %d refc: %u != 0",
+                   __func__, proc_pid(current_proc()), refc);
+       }
+#endif

        switch (FILEPROC_TYPE(fp)) {
        case FTYPE_SIMPLE:
        switch (FILEPROC_TYPE(fp)) {
        case FTYPE_SIMPLE:

-               FREE_ZONE(fp, sizeof (*fp), M_FILEPROC);
+               FREE_ZONE(fp, sizeof(*fp), M_FILEPROC);

                break;
        case FTYPE_GUARDED:
                guarded_fileproc_free(fp);
                break;
        case FTYPE_GUARDED:
                guarded_fileproc_free(fp);


@@ -5582,3 +6319,19 @@ fileproc_free(struct fileproc *fp)
                panic("%s: corrupt fp %p flags %x", __func__, fp, fp->f_flags);
        }
 }
                panic("%s: corrupt fp %p flags %x", __func__, fp, fp->f_flags);
        }
 }

+
+void
+fileproc_modify_vflags(struct fileproc *fp, fileproc_vflags_t vflags, boolean_t clearflags)
+{
+       if (clearflags) {
+               os_atomic_andnot(&fp->f_vflags, vflags, relaxed);
+       } else {
+               os_atomic_or(&fp->f_vflags, vflags, relaxed);
+       }
+}
+
+fileproc_vflags_t
+fileproc_get_vflags(struct fileproc *fp)
+{
+       return os_atomic_load(&fp->f_vflags, relaxed);
+}