]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/kern/kern_exec.c
xnu-792.6.61.tar.gz
[apple/xnu.git] / bsd / kern / kern_exec.c
index 191e3f3965bb3efd2a03df1bbc5de35acf028848..e3ed77f2cdedff2bdb74283a4bc069a024efa651 100644 (file)
 #include <sys/systm.h>
 #include <sys/filedesc.h>
 #include <sys/kernel.h>
-#include <sys/proc.h>
+#include <sys/proc_internal.h>
+#include <sys/kauth.h>
 #include <sys/user.h>
-#include <sys/buf.h>
 #include <sys/socketvar.h>
 #include <sys/malloc.h>
 #include <sys/namei.h>
-#include <sys/mount.h>
-#include <sys/vnode.h>         
-#include <sys/file.h>
+#include <sys/mount_internal.h>
+#include <sys/vnode_internal.h>                
+#include <sys/file_internal.h>
 #include <sys/stat.h>
-#include <sys/uio.h>
+#include <sys/uio_internal.h>
 #include <sys/acct.h>
 #include <sys/exec.h>
 #include <sys/kdebug.h>
 #include <sys/signal.h>
 #include <sys/aio_kern.h>
+#include <sys/sysproto.h>
+#include <sys/shm_internal.h>          /* shmexec() */
+#include <sys/ubc_internal.h>          /* ubc_map() */
 
 #include <bsm/audit_kernel.h>
 
+#include <mach/mach_types.h>
+#include <mach/task.h>
+#include <mach/thread_act.h>
+#include <mach/vm_map.h>
+#include <mach/mach_vm.h>
 #include <mach/vm_param.h>
 
 #include <vm/vm_map.h>
-
-extern vm_map_t vm_map_switch(vm_map_t    map); /* XXX */
-
 #include <vm/vm_kern.h>
+#include <vm/vm_pager.h>
+#include <vm/vm_kern.h>
+#include <vm/task_working_set.h>
 #include <vm/vm_shared_memory_server.h>
 
+/*
+ * Mach things for which prototypes are unavailable from Mach headers
+ */
+void           ipc_task_reset(
+                       task_t          task);
+
+extern struct savearea *get_user_regs(thread_t);
+
+
 #include <kern/thread.h>
 #include <kern/task.h>
-
 #include <kern/ast.h>
 #include <kern/mach_loader.h>
 #include <mach-o/fat.h>
@@ -112,152 +128,723 @@ extern vm_map_t vm_map_switch(vm_map_t    map); /* XXX */
 #include <machine/vmparam.h>
 #if KTRACE   
 #include <sys/ktrace.h>
-#include <sys/ubc.h>
 #endif
+#include <sys/imgact.h>
+
+
+/*
+ * SIZE_MAXPTR         The maximum size of a user space pointer, in bytes
+ * SIZE_IMG_STRSPACE   The available string space, minus two pointers; we
+ *                     define it interms of the maximum, since we don't
+ *                     know the pointer size going in, until after we've
+ *                     parsed the executable image.
+ */
+#define        SIZE_MAXPTR             8                               /* 64 bits */
+#define        SIZE_IMG_STRSPACE       (NCARGS - 2 * SIZE_MAXPTR)
 
 int    app_profile = 0;
 
 extern vm_map_t bsd_pageable_map;
+extern struct fileops vnops;
 
 #define        ROUND_PTR(type, addr)   \
        (type *)( ( (unsigned)(addr) + 16 - 1) \
                  & ~(16 - 1) )
 
+struct image_params;   /* Forward */
+static int exec_copyout_strings(struct image_params *imgp, user_addr_t *stackp);
 static int load_return_to_errno(load_return_t lrtn);
-int execve(struct proc *p, struct execve_args *uap, register_t *retval);
-static int execargs_alloc(vm_offset_t *addrp);
-static int execargs_free(vm_offset_t addr);
+static int execargs_alloc(struct image_params *imgp);
+static int execargs_free(struct image_params *imgp);
+static int exec_check_permissions(struct image_params *imgp);
+static int exec_extract_strings(struct image_params *imgp);
+static int exec_handle_sugid(struct image_params *imgp);
 static int sugid_scripts = 0;
 SYSCTL_INT (_kern, OID_AUTO, sugid_scripts, CTLFLAG_RW, &sugid_scripts, 0, "");
+static kern_return_t create_unix_stack(vm_map_t map, user_addr_t user_stack,
+                                       int customstack, struct proc *p);
+static int copyoutptr(user_addr_t ua, user_addr_t ptr, int ptr_size);
+
+/* XXX forward; should be in headers, but can't be for one reason or another */
+extern int grade_binary(cpu_type_t exectype, cpu_subtype_t execsubtype);
+extern void vfork_return(thread_t th_act,
+                               struct proc * p,
+                               struct proc *p2,
+                               register_t *retval);
 
-int
-execv(p, args, retval)
-       struct proc *p;
-       void *args;
-       int *retval;
-{
-       ((struct execve_args *)args)->envp = NULL;
-       return (execve(p, args, retval));
-}
 
 extern char classichandler[32];
-extern long classichandler_fsid;
+extern uint32_t classichandler_fsid;
 extern long classichandler_fileid;
 
+
 /*
- * Helper routine to get rid of a loop in execve.  Given a pointer to
- * something for the arg list (which might be in kernel space or in user
- * space), copy it into the kernel buffer at the currentWritePt.  This code
- * does the proper thing to get the data transferred.
- * bytesWritten, currentWritePt, and bytesLeft are kept up-to-date.
+ * exec_add_string
+ *
+ * Add the requested string to the string space area.
+ *
+ * Parameters; struct image_params *           image parameter block
+ *             user_addr_t                     string to add to strings area
+ *             uio_seg                         segment where string is located
+ *
+ * Returns:    0                       Success
+ *             !0                      Failure errno from copyinstr()
+ *
+ * Implicit returns:
+ *             (imgp->ip_strendp)      updated location of next add, if any
+ *             (imgp->ip_strspace)     updated byte count of space remaining
  */
-
-static int copyArgument(char *argument, int pointerInKernel,
-                       int *bytesWritten,char **currentWritePt,
-                       int *bytesLeft){
+static int
+exec_add_string(struct image_params *imgp, user_addr_t str, /*uio_seg*/int seg)
+{
         int error = 0;
+
         do {
                 size_t len = 0;
-               if (*bytesLeft <= 0) {
+               if (imgp->ip_strspace <= 0) {
                        error = E2BIG;
                        break;
                }
-               if (pointerInKernel == UIO_SYSSPACE) {
-                       error = copystr(argument, *currentWritePt, (unsigned)*bytesLeft, &len);
+               if (IS_UIO_SYS_SPACE(seg)) {
+                       char *kstr = CAST_DOWN(char *,str);     /* SAFE */
+                       error = copystr(kstr, imgp->ip_strendp, imgp->ip_strspace, &len);
                } else  {
-              /*
-               * pointer in kernel == UIO_USERSPACE
-               * Copy in from user space.
-               */ 
-                 error = copyinstr((caddr_t)argument, *currentWritePt, (unsigned)*bytesLeft,
+                       error = copyinstr(str, imgp->ip_strendp, imgp->ip_strspace,
                            &len);
                }
-               *currentWritePt += len;
-               *bytesWritten += len;
-               *bytesLeft -= len;
+               imgp->ip_strendp += len;
+               imgp->ip_strspace -= len;
        } while (error == ENAMETOOLONG);
+
        return error;
 }
 
-/* ARGSUSED */
-int
-execve(p, uap, retval)
-       register struct proc *p;
-       register struct execve_args *uap;
-       register_t *retval;
+/*
+ * exec_save_path
+ *
+ * To support new app package launching for Mac OS X, the dyld needs the
+ * first argument to execve() stored on the user stack.
+ *
+ * Save the executable path name at the top of the strings area and set
+ * the argument vector pointer to the location following that to indicate
+ * the start of the argument and environment tuples, setting the remaining
+ * string space count to the size of the string area minus the path length
+ * and a reserve for two pointers.
+ *
+ * Parameters; struct image_params *           image parameter block
+ *             char *                          path used to invoke program
+ *             uio_seg                         segment where path is located
+ *
+ * Returns:    int                     0       Success
+ *                                     !0      Failure: error number
+ * Implicit returns:
+ *             (imgp->ip_strings)              saved path
+ *             (imgp->ip_strspace)             space remaining in ip_strings
+ *             (imgp->ip_argv)                 beginning of argument list
+ *             (imgp->ip_strendp)              start of remaining copy area
+ *
+ * Note:       We have to do this before the initial namei() since in the
+ *             path contains symbolic links, namei() will overwrite the
+ *             original path buffer contents.  If the last symbolic link
+ *             resolved was a relative pathname, we would lose the original
+ *             "path", which could be an absolute pathname. This might be
+ *             unacceptable for dyld.
+ */
+static int
+exec_save_path(struct image_params *imgp, user_addr_t path, /*uio_seg*/int seg)
 {
-       register struct ucred *cred = p->p_ucred;
-       register struct filedesc *fdp = p->p_fd;
-       int nc;
-       char *cp;
-       int na, ne, ucp, ap, cc;
-       unsigned len;
-       int executingInterpreter=0;
-
-       int executingClassic=0;
-       char binaryWithClassicName[sizeof(p->p_comm)] = {0};
-       char *execnamep;
-       struct vnode *vp;
-       struct vattr vattr;
-       struct vattr origvattr;
-       vm_offset_t execargs;
-       struct nameidata nd;
-       struct ps_strings ps;
-#define        SHSIZE  512
-       /* Argument(s) to an interpreter.  If we're executing a shell
-        * script, the name (#!/bin/csh) is allowed to be followed by
-        * arguments.  cfarg holds these arguments.
+       int error;
+       size_t  len;
+       char *kpath = CAST_DOWN(char *,path);   /* SAFE */
+
+       imgp->ip_strendp = imgp->ip_strings;
+       imgp->ip_strspace = SIZE_IMG_STRSPACE;
+
+       len = MIN(MAXPATHLEN, imgp->ip_strspace);
+
+       switch( seg) {
+       case UIO_USERSPACE32:
+       case UIO_USERSPACE64:   /* Same for copyin()... */
+               error = copyinstr(path, imgp->ip_strings, len, &len);
+               break;
+       case UIO_SYSSPACE32:
+               error = copystr(kpath, imgp->ip_strings, len, &len);
+               break;
+       default:
+               error = EFAULT;
+               break;
+       }
+
+       if (!error) {
+               imgp->ip_strendp += len;
+               imgp->ip_strspace -= len;
+               imgp->ip_argv = imgp->ip_strendp;
+       }
+
+       return(error);
+}
+
+
+
+/*
+ * exec_shell_imgact
+ *
+ * Image activator for interpreter scripts.  If the image begins with the
+ * characters "#!", then it is an interpreter script.  Verify that we are
+ * not already executing in Classic mode, and that the length of the script
+ * line indicating the interpreter is not in excess of the maximum allowed
+ * size.  If this is the case, then break out the arguments, if any, which
+ * are separated by white space, and copy them into the argument save area
+ * as if they were provided on the command line before all other arguments.
+ * The line ends when we encounter a comment character ('#') or newline.
+ *
+ * Parameters; struct image_params *   image parameter block
+ *
+ * Returns:    -1                      not an interpreter (keep looking)
+ *             -3                      Success: interpreter: relookup
+ *             >0                      Failure: interpreter: error number
+ *
+ * A return value other than -1 indicates subsequent image activators should
+ * not be given the opportunity to attempt to activate the image.
+ */
+static int
+exec_shell_imgact(struct image_params *imgp)
+{
+       char *vdata = imgp->ip_vdata;
+       char *ihp;
+       char *line_endp;
+       char *interp;
+
+       /*
+        * Make sure it's a shell script.  If we've already redirected
+        * from an interpreted file once, don't do it again.
+        *
+        * Note: We disallow Classic, since the expectation is that we
+        * may run a Classic interpreter, but not an interpret a Classic
+        * image.  This is consistent with historical behaviour.
         */
-       char cfarg[SHSIZE];
-       boolean_t               is_fat;
-       kern_return_t           ret;
-       struct mach_header      *mach_header;
-       struct fat_header       *fat_header;
-       struct fat_arch         fat_arch;
-       load_return_t           lret;
-       load_result_t           load_result;
+       if (vdata[0] != '#' ||
+           vdata[1] != '!' ||
+           (imgp->ip_flags & IMGPF_INTERPRET) != 0) {
+               return (-1);
+       }
+
+
+       imgp->ip_flags |= IMGPF_INTERPRET;
+
+        /* Check to see if SUGID scripts are permitted.  If they aren't then
+        * clear the SUGID bits.
+        * imgp->ip_vattr is known to be valid.
+         */
+        if (sugid_scripts == 0) {
+          imgp->ip_origvattr->va_mode &= ~(VSUID | VSGID);
+       }
+
+       /* Find the nominal end of the interpreter line */
+       for( ihp = &vdata[2]; *ihp != '\n' && *ihp != '#'; ihp++) {
+               if (ihp >= &vdata[IMG_SHSIZE])
+                       return (ENOEXEC);
+       }
+
+       line_endp = ihp;
+       ihp = &vdata[2];
+       /* Skip over leading spaces - until the interpreter name */
+       while ( ihp < line_endp && ((*ihp == ' ') || (*ihp == '\t')))
+               ihp++;
+
+       /*
+        * Find the last non-whitespace character before the end of line or
+        * the beginning of a comment; this is our new end of line.
+        */
+       for (;line_endp > ihp && ((*line_endp == ' ') || (*line_endp == '\t')); line_endp--)
+               continue;
+
+       /* Empty? */
+       if (line_endp == ihp)
+               return (ENOEXEC);
+
+       /* copy the interpreter name */
+       interp = imgp->ip_interp_name;
+       while ((ihp < line_endp) && (*ihp != ' ') && (*ihp != '\t'))
+               *interp++ = *ihp++;
+       *interp = '\0';
+
+       exec_save_path(imgp, CAST_USER_ADDR_T(imgp->ip_interp_name),
+                                                       UIO_SYSSPACE32);
+
+       ihp = &vdata[2];
+       while (ihp < line_endp) {
+               /* Skip leading whitespace before each argument */
+               while ((*ihp == ' ') || (*ihp == '\t'))
+                       ihp++;
+
+               if (ihp >= line_endp)
+                       break;
+
+               /* We have an argument; copy it */
+               while ((ihp < line_endp) && (*ihp != ' ') && (*ihp != '\t')) {  
+                       *imgp->ip_strendp++ = *ihp++;
+                       imgp->ip_strspace--;
+               }
+               *imgp->ip_strendp++ = 0;
+               imgp->ip_strspace--;
+               imgp->ip_argc++;
+       }
+
+       return (-3);
+}
+
+
+
+/*
+ * exec_fat_imgact
+ *
+ * Image activator for fat 1.0 binaries.  If the binary is fat, then we
+ * need to select an image from it internally, and make that the image
+ * we are going to attempt to execute.  At present, this consists of
+ * reloading the first page for the image with a first page from the
+ * offset location indicated by the fat header.
+ *
+ * Important:  This image activator is byte order neutral.
+ *
+ * Note:       If we find an encapsulated binary, we make no assertions
+ *             about its  validity; instead, we leave that up to a rescan
+ *             for an activator to claim it, and, if it is claimed by one,
+ *             that activator is responsible for determining validity.
+ */
+static int
+exec_fat_imgact(struct image_params *imgp)
+{
+       struct proc *p = vfs_context_proc(imgp->ip_vfs_context);
+       kauth_cred_t cred = p->p_ucred;
+       struct fat_header *fat_header = (struct fat_header *)imgp->ip_vdata;
+       struct fat_arch fat_arch;
+       int resid, error;
+       load_return_t lret;
+
+       /* Make sure it's a fat binary */
+       if ((fat_header->magic != FAT_MAGIC) &&
+            (fat_header->magic != FAT_CIGAM)) {
+               error = -1;
+               goto bad;
+       }
+
+       /* Look up our preferred architecture in the fat file. */
+       lret = fatfile_getarch_affinity(imgp->ip_vp,
+                                       (vm_offset_t)fat_header,
+                                       &fat_arch,
+                                       (p->p_flag & P_AFFINITY));
+       if (lret != LOAD_SUCCESS) {
+               error = load_return_to_errno(lret);
+               goto bad;
+       }
+
+       /* Read the Mach-O header out of it */
+       error = vn_rdwr(UIO_READ, imgp->ip_vp, imgp->ip_vdata,
+                       PAGE_SIZE, fat_arch.offset,
+                       UIO_SYSSPACE32, (IO_UNIT|IO_NODELOCKED),
+                       cred, &resid, p);
+       if (error) {
+               goto bad;
+       }
+
+       /* Did we read a complete header? */
+       if (resid) {
+               error = EBADEXEC;
+               goto bad;
+       }
+
+       /* Success.  Indicate we have identified an encapsulated binary */
+       error = -2;
+       imgp->ip_arch_offset = (user_size_t)fat_arch.offset;
+       imgp->ip_arch_size = (user_size_t)fat_arch.size;
+
+bad:
+       return (error);
+}
+
+/*
+ * exec_mach_imgact
+ *
+ * Image activator for mach-o 1.0 binaries.
+ *
+ * Important:  This image activator is NOT byte order neutral.
+ */
+static int
+exec_mach_imgact(struct image_params *imgp)
+{
+       struct mach_header *mach_header = (struct mach_header *)imgp->ip_vdata;
+       kauth_cred_t            cred = vfs_context_ucred(imgp->ip_vfs_context);
+       struct proc             *p = vfs_context_proc(imgp->ip_vfs_context);
+       int                     error = 0;
+       int                     vfexec = 0;
+       task_t                  task;
+       task_t                  new_task;
+       thread_t                thread;
        struct uthread          *uthread;
-       vm_map_t old_map;
+       vm_map_t old_map = VM_MAP_NULL;
        vm_map_t map;
-       int i;
        boolean_t                               clean_regions = FALSE;
-       shared_region_mapping_t shared_region = NULL;
     shared_region_mapping_t initial_region = NULL;
+       load_return_t           lret;
+       load_result_t           load_result;
+               
+       /*
+        * make sure it's a Mach-O 1.0 or Mach-O 2.0 binary; the difference
+        * is a reserved field on the end, so for the most part, we can
+        * treat them as if they were identical.
+        */
+       if ((mach_header->magic != MH_MAGIC) &&
+           (mach_header->magic != MH_MAGIC_64)) {
+       error = -1;
+               goto bad;
+       }
+
+       task = current_task();
+       thread = current_thread();
+       uthread = get_bsdthread_info(thread);
+
+       if (uthread->uu_flag & UT_VFORK)
+               vfexec = 1;      /* Mark in exec */
+
+       if ((mach_header->cputype & CPU_ARCH_ABI64) == CPU_ARCH_ABI64)
+               imgp->ip_flags |= IMGPF_IS_64BIT;
+
+       if (!grade_binary(mach_header->cputype, mach_header->cpusubtype)) {
+               error = EBADARCH;
+               goto bad;
+       }
+
+       /*
+        * Copy in arguments/environment from the old process, if the
+        * vector is non-NULL (i.e. exec is not being called from
+        * load_init_program(), as a special case, at system startup).
+        */
+       if (imgp->ip_user_argv != 0LL) {
+               error = exec_extract_strings(imgp);
+               if (error)
+                       goto bad;
+       }
+
+       /*
+        * Hack for binary compatability; put three NULs on the end of the
+        * string area, and round it up to the next word boundary.  This
+        * ensures padding with NULs to the boundary.
+        */
+       imgp->ip_strendp[0] = 0;
+       imgp->ip_strendp[1] = 0;
+       imgp->ip_strendp[2] = 0;
+       imgp->ip_strendp += (((imgp->ip_strendp - imgp->ip_strings) + NBPW-1) & ~(NBPW-1));
+
+
+       if (vfexec) {
+               kern_return_t   result;
+
+               result = task_create_internal(task, FALSE, &new_task);
+               if (result != KERN_SUCCESS)
+               printf("execve: task_create failed. Code: 0x%x\n", result);
+               p->task = new_task;
+               set_bsdtask_info(new_task, p);
+               if (p->p_nice != 0)
+                       resetpriority(p);
+               map = get_task_map(new_task);
+               result = thread_create(new_task, &imgp->ip_vfork_thread);
+               if (result != KERN_SUCCESS)
+               printf("execve: thread_create failed. Code: 0x%x\n", result);
+               /* reset local idea of task, thread, uthread */
+               task = new_task;
+               thread = imgp->ip_vfork_thread;
+               uthread = get_bsdthread_info(thread);
+       } else {
+               map = VM_MAP_NULL;
+       }
+
+       /*
+        * We set these flags here; this is OK, since if we fail after
+        * this point, we have already destroyed the parent process anyway.
+        */
+       if (imgp->ip_flags & IMGPF_IS_64BIT) {
+               task_set_64bit(task, TRUE);
+               p->p_flag |= P_LP64;
+       } else {
+               task_set_64bit(task, FALSE);
+               p->p_flag &= ~P_LP64;
+       }
+
+       /*
+        *      Load the Mach-O file.
+        */
+/* LP64 - remove following "if" statement after osfmk/vm/task_working_set.c */
+if((imgp->ip_flags & IMGPF_IS_64BIT) == 0)
+       if(imgp->ip_tws_cache_name) {
+               tws_handle_startup_file(task, kauth_cred_getuid(cred), 
+                       imgp->ip_tws_cache_name, imgp->ip_vp, &clean_regions);
+       }
+
+       vm_get_shared_region(task, &initial_region);
+
+       
+       /*
+        * NOTE: An error after this point  indicates we have potentially
+        * destroyed or overwrote some process state while attempting an
+        * execve() following a vfork(), which is an unrecoverable condition.
+        */
+
+       /*
+        * We reset the task to 64-bit (or not) here.  It may have picked up
+        * a new map, and we need that to reflect its true 64-bit nature.
+        */
+       task_set_64bit(task, 
+                      ((imgp->ip_flags & IMGPF_IS_64BIT) == IMGPF_IS_64BIT));
+
+       /*
+        * Actually load the image file we previously decided to load.
+        */
+       lret = load_machfile(imgp, mach_header, thread, map, clean_regions, &load_result);
+
+       if (lret != LOAD_SUCCESS) {
+               error = load_return_to_errno(lret);
+               goto badtoolate;
+       }
+
+       /* load_machfile() maps the vnode */
+       (void)ubc_map(imgp->ip_vp, PROT_EXEC);
+
+       /*
+        * deal with set[ug]id.
+        */
+       error = exec_handle_sugid(imgp);
+
+       KNOTE(&p->p_klist, NOTE_EXEC);
+
+       if (!vfexec && (p->p_flag & P_TRACED))
+               psignal(p, SIGTRAP);
+
+       if (error) {
+               goto badtoolate;
+       }
+       vnode_put(imgp->ip_vp);
+       imgp->ip_vp = NULL;
+       
+       if (load_result.unixproc &&
+               create_unix_stack(get_task_map(task),
+                                 load_result.user_stack, load_result.customstack, p)) {
+               error = load_return_to_errno(LOAD_NOSPACE);
+               goto badtoolate;
+       }
+
+       if (vfexec) {
+               uthread->uu_ar0 = (void *)get_user_regs(thread);
+               old_map = vm_map_switch(get_task_map(task));
+       }
+
+       if (load_result.unixproc) {
+               user_addr_t     ap;
+
+               /*
+                * Copy the strings area out into the new process address
+                * space.
+                */
+               ap = p->user_stack;
+               error = exec_copyout_strings(imgp, &ap);
+               if (error) {
+                       if (vfexec)
+                               vm_map_switch(old_map);
+                       goto badtoolate;
+               }
+               /* Set the stack */
+               thread_setuserstack(thread, ap);
+       }
+       
+       if (load_result.dynlinker) {
+               uint64_t        ap;
+
+               /* Adjust the stack */
+               if (imgp->ip_flags & IMGPF_IS_64BIT) {
+                       ap = thread_adjuserstack(thread, -8);
+                       (void)copyoutptr(load_result.mach_header, ap, 8);
+               } else {
+                       ap = thread_adjuserstack(thread, -4);
+                       (void)suword(ap, load_result.mach_header);
+               }
+       }
+
+       if (vfexec) {
+               vm_map_switch(old_map);
+       }
+       /* Set the entry point */
+       thread_setentrypoint(thread, load_result.entry_point);
+
+       /* Stop profiling */
+       stopprofclock(p);
+
+       /*
+        * Reset signal state.
+        */
+       execsigs(p, thread);
+
+       /*
+        * Close file descriptors
+        * which specify close-on-exec.
+        */
+       fdexec(p);
+
+       /*
+        * need to cancel async IO requests that can be cancelled and wait for those
+        * already active.  MAY BLOCK!
+        */
+       _aio_exec( p );
+
+       /* FIXME: Till vmspace inherit is fixed: */
+       if (!vfexec && p->vm_shm)
+               shmexec(p);
+       /* Clean up the semaphores */
+       semexit(p);
+
+       /*
+        * Remember file name for accounting.
+        */
+       p->p_acflag &= ~AFORK;
+       /* If the translated name isn't NULL, then we want to use
+        * that translated name as the name we show as the "real" name.
+        * Otherwise, use the name passed into exec.
+        */
+       if (0 != imgp->ip_p_comm[0]) {
+               bcopy((caddr_t)imgp->ip_p_comm, (caddr_t)p->p_comm,
+                       sizeof(p->p_comm));
+       } else {
+               if (imgp->ip_ndp->ni_cnd.cn_namelen > MAXCOMLEN)
+                       imgp->ip_ndp->ni_cnd.cn_namelen = MAXCOMLEN;
+               bcopy((caddr_t)imgp->ip_ndp->ni_cnd.cn_nameptr, (caddr_t)p->p_comm,
+                       (unsigned)imgp->ip_ndp->ni_cnd.cn_namelen);
+               p->p_comm[imgp->ip_ndp->ni_cnd.cn_namelen] = '\0';
+       }
 
-       union {
-               /* #! and name of interpreter */
-               char                    ex_shell[SHSIZE];
-               /* Mach-O executable */
-               struct mach_header      mach_header;
-               /* Fat executable */
-               struct fat_header       fat_header;
-               char    pad[512];
-       } exdata;
+       {
+         /* This is for kdebug */
+         long dbg_arg1, dbg_arg2, dbg_arg3, dbg_arg4;
+
+         /* Collect the pathname for tracing */
+         kdbg_trace_string(p, &dbg_arg1, &dbg_arg2, &dbg_arg3, &dbg_arg4);
+
+
+
+         if (vfexec)
+         {
+                 KERNEL_DEBUG_CONSTANT1((TRACEDBG_CODE(DBG_TRACE_DATA, 2)) | DBG_FUNC_NONE,
+                                       p->p_pid ,0,0,0, (unsigned int)thread);
+                 KERNEL_DEBUG_CONSTANT1((TRACEDBG_CODE(DBG_TRACE_STRING, 2)) | DBG_FUNC_NONE,
+                                       dbg_arg1, dbg_arg2, dbg_arg3, dbg_arg4, (unsigned int)thread);
+         }
+         else
+         {
+                 KERNEL_DEBUG_CONSTANT((TRACEDBG_CODE(DBG_TRACE_DATA, 2)) | DBG_FUNC_NONE,
+                                       p->p_pid ,0,0,0,0);
+                 KERNEL_DEBUG_CONSTANT((TRACEDBG_CODE(DBG_TRACE_STRING, 2)) | DBG_FUNC_NONE,
+                                       dbg_arg1, dbg_arg2, dbg_arg3, dbg_arg4, 0);
+         }
+       }
+
+               p->p_flag &= ~P_CLASSIC;
+
+       /*
+        * mark as execed, wakeup the process that vforked (if any) and tell
+        * it that it now has it's own resources back
+        */
+       p->p_flag |= P_EXEC;
+       if (p->p_pptr && (p->p_flag & P_PPWAIT)) {
+               p->p_flag &= ~P_PPWAIT;
+               wakeup((caddr_t)p->p_pptr);
+       }
+
+       if (vfexec && (p->p_flag & P_TRACED)) {
+               psignal_vfork(p, new_task, thread, SIGTRAP);
+       }
+
+badtoolate:
+       if (vfexec) {
+               task_deallocate(new_task);
+               thread_deallocate(thread);
+               if (error)
+                       error = 0;
+       }
+
+bad:
+       return(error);
+}
+
+
+
+
+/*
+ * Our image activator table; this is the table of the image types we are
+ * capable of loading.  We list them in order of preference to ensure the
+ * fastest image load speed.
+ *
+ * XXX hardcoded, for now; should use linker sets
+ */
+struct execsw {
+       int (*ex_imgact)(struct image_params *);
+       const char *ex_name;
+} execsw[] = {
+       { exec_mach_imgact,             "Mach-o Binary" },
+       { exec_fat_imgact,              "Fat Binary" },
+       { exec_shell_imgact,            "Interpreter Script" },
+       { NULL, NULL}
+};
+
+
+/*
+ * TODO:       Dynamic linker header address on stack is copied via suword()
+ */
+/* ARGSUSED */
+int
+execve(struct proc *p, struct execve_args *uap, register_t *retval)
+{
+       kauth_cred_t cred = p->p_ucred;
+       struct image_params image_params, *imgp;
+       struct vnode_attr va;
+       struct vnode_attr origva;
+       struct nameidata nd;
+       struct uthread          *uthread;
+       int i;
        int resid, error;
-       char *savedpath;
-       int savedpathlen = 0;
-       vm_offset_t *execargsp;
-       char *cpnospace;
        task_t  task;
-       task_t new_task;
-       thread_act_t thr_act;
        int numthreads;
        int vfexec=0;
-       unsigned long arch_offset =0;
-       unsigned long arch_size = 0;
-        char           *ws_cache_name = NULL;  /* used for pre-heat */
+       int once = 1;   /* save SGUID-ness for interpreted files */
+       char alt_p_comm[sizeof(p->p_comm)] = {0};       /* for Classic */
+       int is_64 = IS_64BIT_PROCESS(p);
+       int seg = (is_64 ? UIO_USERSPACE64 : UIO_USERSPACE32);
+       struct vfs_context context;
+
+       context.vc_proc = p;
+       context.vc_ucred = p->p_ucred;  /* XXX must NOT be kauth_cred_get() */
+
+
+       imgp = &image_params;
+
+       /* Initialize the common data in the image_params structure */
+       bzero(imgp, sizeof(*imgp));
+       imgp->ip_user_fname = uap->fname;
+       imgp->ip_user_argv = uap->argp;
+       imgp->ip_user_envv = uap->envp;
+       imgp->ip_vattr = &va;
+       imgp->ip_origvattr = &origva;
+       imgp->ip_vfs_context = &context;
+       imgp->ip_flags = (is_64 ? IMGPF_WAS_64BIT : IMGPF_NONE);
+       imgp->ip_tws_cache_name = NULL;
+       imgp->ip_p_comm = alt_p_comm;           /* for Classic */
 
-        /*
+       /*
          * XXXAUDIT: Currently, we only audit the pathname of the binary.
          * There may also be poor interaction with dyld.
          */
 
-       cfarg[0] = '\0'; /* initialize to null value. */
        task = current_task();
-       thr_act = current_act();
-       uthread = get_bsdthread_info(thr_act);
+       uthread = get_bsdthread_info(current_thread());
 
-       if (uthread->uu_flag & P_VFORK) {
+       if (uthread->uu_flag & UT_VFORK) {
                        vfexec = 1; /* Mark in exec */
        } else {
                if (task != kernel_task) { 
@@ -265,543 +852,556 @@ execve(p, uap, retval)
                        if (numthreads <= 0 )
                                return(EINVAL);
                        if (numthreads > 1) {
-                               return(EOPNOTSUPP);
+                               return(ENOTSUP);
                        }
                }
        }
 
-       error = execargs_alloc(&execargs);
+       error = execargs_alloc(imgp);
        if (error)
                return(error);
-
-       savedpath = (char *)execargs;
-
-       /*
-        * To support new app package launching for Mac OS X, the dyld
-        * needs the first argument to execve() stored on the user stack.
-        * Copyin the "path" at the begining of the "execargs" buffer
-        * allocated above.
-        *
-        * We have to do this before namei() because in case of
-        * symbolic links, namei() would overwrite the original "path".
-        * In case the last symbolic link resolved was a relative pathname
-        * we would lose the original "path", which could be an
-        * absolute pathname. This might be unacceptable for dyld.
-        */
-       /* XXX We could optimize to avoid copyinstr in the namei() */
-
+       
        /*
         * XXXAUDIT: Note: the double copyin introduces an audit
         * race.  To correct this race, we must use a single
-        * copyin().
+        * copyin(), e.g. by passing a flag to namei to indicate an
+        * external path buffer is being used.
         */
-       
-       error = copyinstr(uap->fname, savedpath,
-                               MAXPATHLEN, (size_t *)&savedpathlen);
+       error = exec_save_path(imgp, uap->fname, seg);
        if (error) {
-               execargs_free(execargs);
+               execargs_free(imgp);
                return(error);
        }
+
        /*
-        * copyinstr will put in savedpathlen, the count of
-        * characters (including NULL) in the path.
         * No app profiles under chroot
         */
-
-       if((fdp->fd_rdir == NULLVP) && (app_profile != 0)) {
+       if((p->p_fd->fd_rdir == NULLVP) && (app_profile != 0)) {
 
                /* grab the name of the file out of its path */
                /* we will need this for lookup within the   */
                /* name file */
-               ws_cache_name = savedpath + savedpathlen;
-                       while (ws_cache_name[0] != '/') {
-                               if(ws_cache_name == savedpath) {
-                                       ws_cache_name--;
+               /* Scan backwards for the first '/' or start of string */
+               imgp->ip_tws_cache_name = imgp->ip_strendp;
+                       while (imgp->ip_tws_cache_name[0] != '/') {
+                               if(imgp->ip_tws_cache_name == imgp->ip_strings) {
+                                       imgp->ip_tws_cache_name--;
                                        break;
                        }
-                               ws_cache_name--;
+                               imgp->ip_tws_cache_name--;
                        }
-                       ws_cache_name++;
+                       imgp->ip_tws_cache_name++;
        }
+       NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | AUDITVNPATH1,
+               seg, uap->fname, imgp->ip_vfs_context);
 
-       /* Save the name aside for future use */
-       execargsp = (vm_offset_t *)((char *)(execargs) + savedpathlen);
-       
-       NDINIT(&nd, LOOKUP, FOLLOW | LOCKLEAF | SAVENAME | AUDITVNPATH1,
-                                       UIO_USERSPACE, uap->fname, p);
+again:
        error = namei(&nd);
        if (error)
-               goto bad1;
-       vp = nd.ni_vp;
-       VOP_LEASE(vp, p, p->p_ucred, LEASE_READ);
-
-       if ((error = VOP_GETATTR(vp, &origvattr, p->p_ucred, p)))
                goto bad;
+       imgp->ip_ndp = &nd;     /* successful namei(); call nameidone() later */
+       imgp->ip_vp = nd.ni_vp; /* if set, need to vnode_put() at some point */
 
-       /* Check mount point */
-       if (vp->v_mount->mnt_flag & MNT_NOEXEC) {
-               error = EACCES;
+       error = exec_check_permissions(imgp);
+       if (error)
                goto bad;
-       }
 
-       if ((vp->v_mount->mnt_flag & MNT_NOSUID) || (p->p_flag & P_TRACED))
-               origvattr.va_mode &= ~(VSUID | VSGID);
-               
-       *(&vattr) = *(&origvattr);
+       /* Copy; avoid invocation of an interpreter overwriting the original */
+       if (once) {
+               once = 0;
+               origva = va;
+       }
 
-again:
-       error = check_exec_access(p, vp, &vattr);
+       error = vn_rdwr(UIO_READ, imgp->ip_vp, imgp->ip_vdata, PAGE_SIZE, 0,
+                       UIO_SYSSPACE32, IO_NODELOCKED, cred, &resid, p);
        if (error)
                goto bad;
+               
+encapsulated_binary:
+       error = -1;
+       for(i = 0; error == -1 && execsw[i].ex_imgact != NULL; i++) {
 
-       /*
-        * Read in first few bytes of file for segment sizes, magic number:
-        *      407 = plain executable
-        *      410 = RO text
-        *      413 = demand paged RO text
-        * Also an ASCII line beginning with #! is
-        * the file name of a ``shell'' and arguments may be prepended
-        * to the argument list if given here.
-        *
-        * SHELL NAMES ARE LIMITED IN LENGTH.
-        *
-        * ONLY ONE ARGUMENT MAY BE PASSED TO THE SHELL FROM
-        * THE ASCII LINE.
-        */
+               error = (*execsw[i].ex_imgact)(imgp);
 
-       exdata.ex_shell[0] = '\0';      /* for zero length files */
+               switch (error) {
+               /* case -1: not claimed: continue */
+               case -2:                /* Encapsulated binary */
+                       goto encapsulated_binary;
 
-       error = vn_rdwr(UIO_READ, vp, (caddr_t)&exdata, sizeof (exdata), 0,
-                       UIO_SYSSPACE, IO_NODELOCKED, p->p_ucred, &resid, p);
+               case -3:                /* Interpreter */
+                       vnode_put(imgp->ip_vp);
+                       imgp->ip_vp = NULL;     /* already put */
+                       nd.ni_cnd.cn_nameiop = LOOKUP;
+                       nd.ni_cnd.cn_flags = (nd.ni_cnd.cn_flags & HASBUF) |
+                                               (FOLLOW | LOCKLEAF);
 
-       if (error)
-               goto bad;
 
-#ifndef lint
-       if (resid > sizeof(exdata) - min(sizeof(exdata.mach_header),
-                                        sizeof(exdata.fat_header))
-           && exdata.ex_shell[0] != '#') {
+                       nd.ni_segflg = UIO_SYSSPACE32;
+                       nd.ni_dirp = CAST_USER_ADDR_T(imgp->ip_interp_name);
+                       goto again;
+
+               default:
+                       break;
+               }
+       }
+       
+       /* call out to allow 3rd party notification of exec. 
+        * Ignore result of kauth_authorize_fileop call.
+        */
+       if (error == 0 && kauth_authorize_fileop_has_listeners()) {
+               kauth_authorize_fileop(vfs_context_ucred(&context), KAUTH_FILEOP_EXEC, 
+                                                          (uintptr_t)nd.ni_vp, 0);
+       }
+       
+       /* Image not claimed by any activator? */
+       if (error == -1)
                error = ENOEXEC;
-               goto bad;
+
+bad:
+       if (imgp->ip_ndp)
+               nameidone(imgp->ip_ndp);
+       if (imgp->ip_vp)
+               vnode_put(imgp->ip_vp);
+       if (imgp->ip_strings)
+               execargs_free(imgp);
+       if (!error && vfexec) {
+                       vfork_return(current_thread(), p->p_pptr, p, retval);
+                       (void)thread_resume(imgp->ip_vfork_thread);
+                       return(0);
        }
-#endif /* lint */
-       mach_header = &exdata.mach_header;
-       fat_header = &exdata.fat_header;
-       if ((mach_header->magic == MH_CIGAM) &&
-           (classichandler[0] == 0)) {
-               error = EBADARCH;
-               goto bad;
-       } else if ((mach_header->magic == MH_MAGIC) || 
-               (mach_header->magic == MH_CIGAM)) {
-           is_fat = FALSE;
-       } else if ((fat_header->magic == FAT_MAGIC) ||
-                      (fat_header->magic == FAT_CIGAM)) {
-           is_fat = TRUE;
+       return(error);
+}
+
+
+static int
+copyinptr(user_addr_t froma, user_addr_t *toptr, int ptr_size)
+{
+       int error;
+
+       if (ptr_size == 4) {
+               /* 64 bit value containing 32 bit address */
+               unsigned int i;
+
+               error = copyin(froma, &i, 4);
+               *toptr = CAST_USER_ADDR_T(i);   /* SAFE */
        } else {
-         /* If we've already redirected once from an interpreted file
-          * to an interpreter, don't permit the second time.
-          */
-               if (exdata.ex_shell[0] != '#' ||
-                   exdata.ex_shell[1] != '!' ||
-                   executingInterpreter) {
-                       error = ENOEXEC;
-                       goto bad;
-               }
-               if (executingClassic == 1) {
-                 error = EBADARCH;
-                 goto bad;
-               }
+               error = copyin(froma, toptr, 8);
+       }
+       return (error);
+}
 
-               /* Check to see if SUGID scripts are permitted.  If they aren't then
-                * clear the SUGID bits.
-                */
-               if (sugid_scripts == 0) {
-                  origvattr.va_mode &= ~(VSUID | VSGID);
-               }
-
-               cp = &exdata.ex_shell[2];               /* skip "#!" */
-               while (cp < &exdata.ex_shell[SHSIZE]) {
-                       if (*cp == '\t')                /* convert all tabs to spaces */
-                               *cp = ' ';
-                       else if (*cp == '\n' || *cp == '#') {
-                               *cp = '\0';                     /* trunc the line at nl or comment */
-
-                               /* go back and remove the spaces before the /n or # */
-                               /* todo: do we have to do this if we fix the passing of args to shells ? */
-                               if ( cp != &exdata.ex_shell[2] ) {
-                                       do {
-                                               if ( *(cp-1) != ' ')
-                                                       break;
-                                               *(--cp) = '\0';
-                                       } while ( cp != &exdata.ex_shell[2] );
-                               }
-                               break;
-                       }
-                       cp++;
-               }
-               if (*cp != '\0') {
-                       error = ENOEXEC;
-                       goto bad;
-               }
-               cp = &exdata.ex_shell[2];
-               while (*cp == ' ')
-                       cp++;
-               execnamep = cp;
-               while (*cp && *cp != ' ')
-                       cp++;
-               cfarg[0] = '\0';
-               cpnospace = cp;
-               if (*cp) {
-                       *cp++ = '\0';
-                       while (*cp == ' ')
-                               cp++;
-                       if (*cp)
-                               bcopy((caddr_t)cp, (caddr_t)cfarg, SHSIZE);
-               }
 
-               /*
-                * Support for new app package launching for Mac OS X.
-                * We are about to retry the execve() by changing the path to the
-                * interpreter name. Need to re-initialize the savedpath and
-                * savedpathlen. +1 for NULL.
-                */
-               savedpathlen = (cpnospace - execnamep + 1);
-               error = copystr(execnamep, savedpath,
-                                       savedpathlen, (size_t *)&savedpathlen);
-               if (error)
-                       goto bad;
+static int
+copyoutptr(user_addr_t ua, user_addr_t ptr, int ptr_size)
+{
+       int error;
 
-               /* Save the name aside for future use */
-               execargsp = (vm_offset_t *)((char *)(execargs) + savedpathlen);
-
-               executingInterpreter= 1;
-               vput(vp);
-               nd.ni_cnd.cn_nameiop = LOOKUP;
-               nd.ni_cnd.cn_flags = (nd.ni_cnd.cn_flags & HASBUF) |
-                                               (FOLLOW | LOCKLEAF | SAVENAME);
-               nd.ni_segflg = UIO_SYSSPACE;
-               nd.ni_dirp = execnamep;
-               if ((error = namei(&nd)))
-                       goto bad1;
-               vp = nd.ni_vp;
-               VOP_LEASE(vp, p, cred, LEASE_READ);
-               if ((error = VOP_GETATTR(vp, &vattr, p->p_ucred, p)))
-                       goto bad;
-               goto again;
+       if (ptr_size == 4) {
+               /* 64 bit value containing 32 bit address */
+               unsigned int i = CAST_DOWN(unsigned int,ua);    /* SAFE */
+
+               error = copyout(&i, ptr, 4);
+       } else {
+               error = copyout(&ua, ptr, 8);
        }
+       return (error);
+}
+
+
+/*
+ * exec_copyout_strings
+ *
+ * Copy out the strings segment to user space.  The strings segment is put
+ * on a preinitialized stack frame.
+ *
+ * Parameters: struct image_params *   the image parameter block
+ *             int *                   a pointer to the stack offset variable
+ *
+ * Returns:    0                       Success
+ *             !0                      Faiure: errno
+ *
+ * Implicit returns:
+ *             (*stackp)               The stack offset, modified
+ *
+ * Note:       The strings segment layout is backward, from the beginning
+ *             of the top of the stack to consume the minimal amount of
+ *             space possible; the returned stack pointer points to the
+ *             end of the area consumed (stacks grow upward).
+ *
+ *             argc is an int; arg[i] are pointers; env[i] are pointers;
+ *             exec_path is a pointer; the 0's are (void *)NULL's
+ *
+ * The stack frame layout is:
+ *
+ *     +-------------+
+ * sp->        |     argc    |
+ *     +-------------+
+ *     |    arg[0]   |
+ *     +-------------+
+ *            :
+ *            :
+ *     +-------------+
+ *     | arg[argc-1] |
+ *     +-------------+
+ *     |      0      |
+ *     +-------------+
+ *     |    env[0]   |
+ *     +-------------+
+ *            :
+ *            :
+ *     +-------------+
+ *     |    env[n]   |
+ *     +-------------+
+ *     |      0      |
+ *     +-------------+
+ *     |  exec_path  | In MacOS X PR2 Beaker2E the path passed to exec() is
+ *     +-------------+ passed on the stack just after the trailing 0 of the
+ *     |      0      | the envp[] array as a pointer to a string.
+ *     +-------------+
+ *     |  PATH AREA  |
+ *     +-------------+
+ *     | STRING AREA |
+ *            :
+ *            :
+ *     |             | <- p->user_stack
+ *     +-------------+
+ *
+ * Although technically a part of the STRING AREA, we treat the PATH AREA as
+ * a separate entity.  This allows us to align the beginning of the PATH AREA
+ * to a pointer boundary so that the exec_path, env[i], and argv[i] pointers
+ * which preceed it on the stack are properly aligned.
+ *
+ * TODO:       argc copied with suword(), which takes a 64 bit address
+ */
+static int
+exec_copyout_strings(struct image_params *imgp, user_addr_t *stackp)
+{
+       struct proc *p = vfs_context_proc(imgp->ip_vfs_context);
+       int     ptr_size = (imgp->ip_flags & IMGPF_IS_64BIT) ? 8 : 4;
+       char    *argv = imgp->ip_argv;  /* modifiable copy of argv */
+       user_addr_t     string_area;    /* *argv[], *env[] */
+       user_addr_t     path_area;      /* package launch path */
+       user_addr_t     ptr_area;       /* argv[], env[], exec_path */
+       user_addr_t     stack;
+       int     stringc = imgp->ip_argc + imgp->ip_envc;
+       int len;
+       int error;
+       int strspace;
+
+       stack = *stackp;
+
+       /*
+        * Set up pointers to the beginning of the string area, the beginning
+        * of the path area, and the beginning of the pointer area (actually,
+        * the location of argc, an int, which may be smaller than a pointer,
+        * but we use ptr_size worth of space for it, for alignment).
+        */
+       string_area = stack - (((imgp->ip_strendp - imgp->ip_strings) + ptr_size-1) & ~(ptr_size-1)) - ptr_size;
+       path_area = string_area - (((imgp->ip_argv - imgp->ip_strings) + ptr_size-1) & ~(ptr_size-1));
+       ptr_area = path_area - ((imgp->ip_argc + imgp->ip_envc + 4) * ptr_size) - ptr_size /*argc*/;
+
+       /* Return the initial stack address: the location of argc */
+       *stackp = ptr_area;
 
        /*
-        * Collect arguments on "file" in swap space.
+        * Record the size of the arguments area so that sysctl_procargs()
+        * can return the argument area without having to parse the arguments.
         */
-       na = 0;
-       ne = 0;
-       nc = 0;
-       cc = 0;
+       p->p_argc = imgp->ip_argc;
+       p->p_argslen = (int)(stack - path_area);
+
+
        /*
         * Support for new app package launching for Mac OS X allocates
-        * the "path" at the begining.
-        * execargs get allocated after that
+        * the "path" at the begining of the imgp->ip_strings buffer.
+        * copy it just before the string area.
         */
-       cp = (char *) execargsp;        /* running pointer for copy */
+       len = 0;
+       error = copyoutstr(imgp->ip_strings, path_area,
+                               (unsigned)(imgp->ip_argv - imgp->ip_strings),
+                               (size_t *)&len);
+       if (error)
+               goto bad;
+
+
+       /* Save a NULL pointer below it */
+       (void)copyoutptr(0LL, path_area - ptr_size, ptr_size);
+
+       /* Save the pointer to "path" just below it */
+       (void)copyoutptr(path_area, path_area - 2*ptr_size, ptr_size);
+
        /*
-        * size of execargs less sizeof "path",
-        * a pointer to "path" and a NULL poiter
+        * ptr_size for 2 NULL one each ofter arg[argc -1] and env[n]
+        * ptr_size for argc
+        * skip over saved path, ptr_size for pointer to path,
+        * and ptr_size for the NULL after pointer to path.
         */
-       cc = NCARGS - savedpathlen - 2*NBPW;
+
+       /* argc (int32, stored in a ptr_size area) */
+       (void)suword(ptr_area, imgp->ip_argc);
+       ptr_area += sizeof(int);
+       /* pad to ptr_size, if 64 bit image, to ensure user stack alignment */
+       if (imgp->ip_flags & IMGPF_IS_64BIT) {
+               (void)suword(ptr_area, 0);      /* int, not long: ignored */
+               ptr_area += sizeof(int);
+       }
+
+
        /*
-        * Copy arguments into file in argdev area.
+        * We use (string_area - path_area) here rather than the more
+        * intuitive (imgp->ip_argv - imgp->ip_strings) because we are
+        * interested in the length of the PATH_AREA in user space,
+        * rather than the actual length of the execution path, since
+        * it includes alignment padding of the PATH_AREA + STRING_AREA
+        * to a ptr_size boundary.
         */
+       strspace = SIZE_IMG_STRSPACE - (string_area - path_area);
+       for (;;) {
+               if (stringc == imgp->ip_envc) {
+                       /* argv[n] = NULL */
+                       (void)copyoutptr(0LL, ptr_area, ptr_size);
+                       ptr_area += ptr_size;
+               }
+               if (--stringc < 0)
+                       break;
+
+               /* pointer: argv[n]/env[n] */
+               (void)copyoutptr(string_area, ptr_area, ptr_size);
 
+               /* string : argv[n][]/env[n][] */
+               do {
+                       if (strspace <= 0) {
+                               error = E2BIG;
+                               break;
+                       }
+                       error = copyoutstr(argv, string_area,
+                                               (unsigned)strspace,
+                                               (size_t *)&len);
+                       string_area += len;
+                       argv += len;
+                       strspace -= len;
+               } while (error == ENAMETOOLONG);
+               if (error == EFAULT || error == E2BIG)
+                       break;  /* bad stack - user's problem */
+               ptr_area += ptr_size;
+       }
+       /* env[n] = NULL */
+       (void)copyoutptr(0LL, ptr_area, ptr_size);
+
+bad:
+       return(error);
+}
+
+
+/*
+ * exec_extract_strings
+ *
+ * Copy arguments and environment from user space into work area; we may
+ * have already copied some early arguments into the work area, and if
+ * so, any arguments opied in are appended to those already there.
+ *
+ * Parameters: struct image_params *   the image parameter block
+ *
+ * Returns:    0                       Success
+ *             !0                      Failure: errno
+ *
+ * Implicit returns;
+ *             (imgp->ip_argc)         Count of arguments, updated
+ *             (imgp->ip_envc)         Count of environment strings, updated
+ *
+ *
+ * Notes:      The argument and environment vectors are user space pointers
+ *             to arrays of user space pointers.
+ */
+static int
+exec_extract_strings(struct image_params *imgp)
+{
+       int error = 0;
+       struct proc *p = vfs_context_proc(imgp->ip_vfs_context);
+       int seg = (IS_64BIT_PROCESS(p) ? UIO_USERSPACE64 : UIO_USERSPACE32);
+       int     ptr_size = (imgp->ip_flags & IMGPF_WAS_64BIT) ? 8 : 4;
+       user_addr_t     argv = imgp->ip_user_argv;
+       user_addr_t     envv = imgp->ip_user_envv;
+
+       /* Now, get rest of arguments */
 
        /*
-        * If we have a fat file, find "our" executable.
+        * If we are running an interpreter, replace the av[0] that was
+        * passed to execve() with the fully qualified path name that was
+        * passed to execve() for interpreters which do not use the PATH
+        * to locate their script arguments.
         */
-       if (is_fat) {
-               /*
-                * Look up our architecture in the fat file.
-                */
-               lret = fatfile_getarch_affinity(vp,(vm_offset_t)fat_header, &fat_arch,
-                                               (p->p_flag & P_AFFINITY));
-               if (lret != LOAD_SUCCESS) {
-                       error = load_return_to_errno(lret);
+       if((imgp->ip_flags & IMGPF_INTERPRET) != 0 && argv != 0LL) {
+               user_addr_t     arg;
+
+               error = copyinptr(argv, &arg, ptr_size);
+               if (error)
                        goto bad;
+               if (arg != 0LL && arg != (user_addr_t)-1) {
+                       argv += ptr_size;
+                       error = exec_add_string(imgp, imgp->ip_user_fname, seg);
+                       if (error)
+                               goto bad;
+                       imgp->ip_argc++;
                }
-               /* Read the Mach-O header out of it */
-               error = vn_rdwr(UIO_READ, vp, (caddr_t)&exdata.mach_header,
-                               sizeof (exdata.mach_header),
-                               fat_arch.offset,
-                               UIO_SYSSPACE, (IO_UNIT|IO_NODELOCKED), cred, &resid, p);
+       }
 
-               if (error) {
+       while (argv != 0LL) {
+               user_addr_t     arg;
+
+               error = copyinptr(argv, &arg, ptr_size);
+               if (error)
                        goto bad;
-               }
 
-               /* Did we read a complete header? */
-               if (resid) {
-                       error = EBADEXEC;
+               argv += ptr_size;
+               if (arg == 0LL) {
+                       break;
+               } else if (arg == (user_addr_t)-1) {
+                       /* Um... why would it be -1? */
+                       error = EFAULT;
                        goto bad;
                }
+               /*
+               * av[n...] = arg[n]
+               */
+               error = exec_add_string(imgp, arg, seg);
+               if (error)
+                       goto bad;
+               imgp->ip_argc++;
+       }        
+
+       /* Now, get the environment */
+       while (envv != 0LL) {
+               user_addr_t     env;
+
+               error = copyinptr(envv, &env, ptr_size);
+               if (error)
+                       goto bad;
 
-               /* Is what we found a Mach-O executable */
-               if ((mach_header->magic != MH_MAGIC) &&
-                   (mach_header->magic != MH_CIGAM)) {
-                       error = ENOEXEC;
+               envv += ptr_size;
+               if (env == 0LL) {
+                       break;
+               } else if (env == (user_addr_t)-1) {
+                       error = EFAULT;
                        goto bad;
                }
-
-               arch_offset = fat_arch.offset;
-               arch_size = fat_arch.size;
-       } else {
                /*
-                *      Load the Mach-O file.
-                */
-               arch_offset = 0;
-               arch_size = (u_long)vattr.va_size;
-       }
-
-       if ( ! check_cpu_subtype(mach_header->cpusubtype) ) {
-               error = EBADARCH;
-               goto bad;
-       }
-
-       if (mach_header->magic == MH_CIGAM) {
-
-               int classicBinaryLen = nd.ni_cnd.cn_namelen;
-               if (classicBinaryLen > MAXCOMLEN)
-               classicBinaryLen = MAXCOMLEN;
-               bcopy((caddr_t)nd.ni_cnd.cn_nameptr,
-                               (caddr_t)binaryWithClassicName, 
-                               (unsigned)classicBinaryLen);
-               binaryWithClassicName[classicBinaryLen] = '\0';
-               executingClassic = 1;
-
-               vput(vp); /* cleanup? */
-               nd.ni_cnd.cn_nameiop = LOOKUP;
-
-               nd.ni_cnd.cn_flags = (nd.ni_cnd.cn_flags & HASBUF) |
-               /*      (FOLLOW | LOCKLEAF | SAVENAME) */
-               (LOCKLEAF | SAVENAME);
-                nd.ni_segflg = UIO_SYSSPACE;
-
-                       nd.ni_dirp = classichandler;
-                       if ((error = namei(&nd)) != 0) {
-                       error = EBADARCH;
-                               goto bad1;
-               }
-               vp = nd.ni_vp;
-
-               VOP_LEASE(vp,p,cred,LEASE_READ);
-               if ((error = VOP_GETATTR(vp,&vattr,p->p_ucred,p))) {
+               * av[n...] = env[n]
+               */
+               error = exec_add_string(imgp, env, seg);
+               if (error)
                        goto bad;
-               }
-               goto again;
+               imgp->ip_envc++;
        }
+bad:
+       return error;
+}
 
-       if (uap->argp != NULL) {
-         /* geez -- why would argp ever be NULL, and why would we proceed? */
-         
-         /* First, handle any argument massaging */
-         if (executingInterpreter && executingClassic) {
-           error = copyArgument(classichandler,UIO_SYSSPACE,&nc,&cp,&cc);
-           na++;
-           if (error) goto bad;
-           
-           /* Now name the interpreter. */
-           error = copyArgument(savedpath,UIO_SYSSPACE,&nc,&cp,&cc);
-           na++;
-           if (error) goto bad;
 
-           /*
-            * if we're running an interpreter, as we'd be passing the
-            * command line executable as an argument to the interpreter already.
-            * Doing "execve("myShellScript","bogusName",arg1,arg2,...)
-            * probably shouldn't ever let bogusName be seen by the shell
-            * script.
-            */
-
-           if (cfarg[0]) {
-             error = copyArgument(cfarg,UIO_SYSSPACE,&nc,&cp,&cc);
-             na++;
-             if (error) goto bad;
-           }
-
-           char* originalExecutable = uap->fname;
-           error = copyArgument(originalExecutable,UIO_USERSPACE,&nc,&cp,&cc);
-           na++;
-           /* remove argv[0] b/c we've already placed it at */
-           /* this point */
-           uap->argp++;
-           if (error) goto bad;
-
-           /* and continue with rest of the arguments. */
-         } else if (executingClassic) {
-           error = copyArgument(classichandler,UIO_SYSSPACE,&nc,&cp,&cc);
-           na++;
-           if (error) goto bad;
-           
-           char* originalExecutable = uap->fname;
-           error = copyArgument(originalExecutable,UIO_USERSPACE,&nc,&cp,&cc);
-           if (error) goto bad;
-           uap->argp++;
-           na++;
-
-           /* and rest of arguments continue as before. */
-         } else if (executingInterpreter) {
-           char *actualExecutable = nd.ni_cnd.cn_nameptr;
-           error = copyArgument(actualExecutable,UIO_SYSSPACE,&nc,&cp,&cc);
-           na++;
-           /* remove argv[0] b/c we just placed it in the arg list. */
-           uap->argp++;
-           if (error) goto bad;
-           /* Copy the argument in the interpreter first line if there
-            * was one. 
-            */
-           if (cfarg[0]) {
-             error = copyArgument(cfarg,UIO_SYSSPACE,&nc,&cp,&cc);
-             na++;
-             if (error) goto bad;
-           }
-           
-           /* copy the name of the file being interpreted, gotten from
-            * the structures passed in to execve.
-            */
-           error = copyArgument(uap->fname,UIO_USERSPACE,&nc,&cp,&cc);
-           na++;
-         }
-         /* Now, get rest of arguments */
-         while (uap->argp != NULL) {
-           char* userArgument = (char*)fuword((caddr_t) uap->argp);
-           uap->argp++;
-           if (userArgument == NULL) {
-             break;
-           } else if ((int)userArgument == -1) {
-             /* Um... why would it be -1? */
-             error = EFAULT;
-             goto bad;
-           }
-           error = copyArgument(userArgument, UIO_USERSPACE,&nc,&cp,&cc);
-           if (error) goto bad;
-           na++;
-         }      
-         /* Now, get the environment */
-         while (uap->envp != NULL) {
-           char *userEnv = (char*) fuword((caddr_t) uap->envp);
-           uap->envp++;
-           if (userEnv == NULL) {
-             break;
-           } else if ((int)userEnv == -1) {
-             error = EFAULT;
-             goto bad;
-           }
-           error = copyArgument(userEnv,UIO_USERSPACE,&nc,&cp,&cc);
-           if (error) goto bad;
-           na++;
-           ne++;
-         }
-       }
+#define        unix_stack_size(p)      (p->p_rlimit[RLIMIT_STACK].rlim_cur)
 
-       /* make sure there are nulls are the end!! */
-       {
-               int     cnt = 3;
-               char *mp = cp;
+static int
+exec_check_permissions(struct image_params *imgp)
+{
+       struct vnode *vp = imgp->ip_vp;
+       struct vnode_attr *vap = imgp->ip_vattr;
+       struct proc *p = vfs_context_proc(imgp->ip_vfs_context);
+       int error;
+       kauth_action_t action;
 
-               while ( cnt-- )
-                       *mp++ = '\0';   
-       }
+       /* Only allow execution of regular files */
+       if (!vnode_isreg(vp))
+               return (EACCES);
+       
+       /* Get the file attributes that we will be using here and elsewhere */
+       VATTR_INIT(vap);
+       VATTR_WANTED(vap, va_uid);
+       VATTR_WANTED(vap, va_gid);
+       VATTR_WANTED(vap, va_mode);
+       VATTR_WANTED(vap, va_fsid);
+       VATTR_WANTED(vap, va_fileid);
+       VATTR_WANTED(vap, va_data_size);
+       if ((error = vnode_getattr(vp, vap, imgp->ip_vfs_context)) != 0)
+               return (error);
 
-       /* and round up count of bytes written to next word. */
-       nc = (nc + NBPW-1) & ~(NBPW-1);
+       /*
+        * Ensure that at least one execute bit is on - otherwise root
+        * will always succeed, and we don't want to happen unless the
+        * file really is executable.
+        */
+       if ((vap->va_mode & (S_IXUSR | S_IXGRP | S_IXOTH)) == 0)
+               return (EACCES);
 
-       if (vattr.va_fsid == classichandler_fsid &&
-               vattr.va_fileid == classichandler_fileid) {
-               executingClassic = 1;
-       }
+       /* Disallow zero length files */
+       if (vap->va_data_size == 0)
+               return (ENOEXEC);
 
-       if (vfexec) {
-               kern_return_t   result;
+       imgp->ip_arch_offset = (user_size_t)0;
+       imgp->ip_arch_size = vap->va_data_size;
 
-               result = task_create_internal(task, FALSE, &new_task);
-               if (result != KERN_SUCCESS)
-               printf("execve: task_create failed. Code: 0x%x\n", result);
-               p->task = new_task;
-               set_bsdtask_info(new_task, p);
-               if (p->p_nice != 0)
-                       resetpriority(p);
-               task = new_task;
-               map = get_task_map(new_task);
-               result = thread_create(new_task, &thr_act);
-               if (result != KERN_SUCCESS)
-               printf("execve: thread_create failed. Code: 0x%x\n", result);
-               uthread = get_bsdthread_info(thr_act);
-       } else {
-               map = VM_MAP_NULL;
-       }
+       /* Disable setuid-ness for traced programs or if MNT_NOSUID */
+       if ((vp->v_mount->mnt_flag & MNT_NOSUID) || (p->p_flag & P_TRACED))
+               vap->va_mode &= ~(VSUID | VSGID);
+
+       /* Check for execute permission */
+       action = KAUTH_VNODE_EXECUTE;
+       /* Traced images must also be readable */
+       if (p->p_flag & P_TRACED)
+               action |= KAUTH_VNODE_READ_DATA;
+       if ((error = vnode_authorize(vp, NULL, action, imgp->ip_vfs_context)) != 0)
+               return (error);
 
-       /*
-        *      Load the Mach-O file.
-        */
-       VOP_UNLOCK(vp, 0, p);   /* XXX */
-       if(ws_cache_name) {
-               tws_handle_startup_file(task, cred->cr_uid, 
-                       ws_cache_name, vp, &clean_regions);
-       }
+       /* Don't let it run if anyone had it open for writing */
+       if (vp->v_writecount)
+               return (ETXTBSY);
 
-       vm_get_shared_region(task, &initial_region);
-    int parentIsClassic = (p->p_flag & P_CLASSIC);
-       struct vnode *rootDir = p->p_fd->fd_rdir;
-
-       if ((parentIsClassic && !executingClassic) ||
-               (!parentIsClassic && executingClassic)) {
-               shared_region = lookup_default_shared_region(
-                               (int)rootDir,
-                               (executingClassic ?
-                               CPU_TYPE_POWERPC :
-                               machine_slot[cpu_number()].cpu_type));
-               if (shared_region == NULL) {
-                       shared_region_mapping_t old_region;
-                       shared_region_mapping_t new_region;
-                       vm_get_shared_region(current_task(), &old_region);
-                       /* grrrr... this sets current_task(), not task
-                       * -- they're different (usually)
-                       */
-                       shared_file_boot_time_init(
-                               (int)rootDir,
-                               (executingClassic ?
-                               CPU_TYPE_POWERPC :
-                               machine_slot[cpu_number()].cpu_type));
-                       if ( current_task() != task ) {
-                               vm_get_shared_region(current_task(),&new_region);
-                               vm_set_shared_region(task,new_region);
-                               vm_set_shared_region(current_task(),old_region);
-                       }
-               } else {
-                       vm_set_shared_region(task, shared_region);
-               }
-               shared_region_mapping_dealloc(initial_region);
-       }
-       
-       lret = load_machfile(vp, mach_header, arch_offset,
-               arch_size, &load_result, thr_act, map, clean_regions);
 
-       if (lret != LOAD_SUCCESS) {
-               error = load_return_to_errno(lret);
-               vrele(vp);
-               vp = NULL;
-               goto badtoolate;
-       }
+       /* XXX May want to indicate to underlying FS that vnode is open */
 
-       /* load_machfile() maps the vnode */
-       ubc_map(vp);
+       return (error);
+}
+
+/*
+ * exec_handle_sugid
+ *
+ * Initially clear the P_SUGID in the process flags; if an SUGID process is
+ * exec'ing a non-SUGID image, then  this is the point of no return.
+ *
+ * If the image being activated is SUGI, then replace the credential with a
+ * copy, disable tracing (unless the tracing process is root), reset the
+ * mach task port to revoke it, set the P_SUGID bit,
+ *
+ * If the saved user and group ID will be changing, then make sure it happens
+ * to a new credential, rather than a shared one.
+ *
+ * Set the security token (this is probably obsolete, given that the token
+ * should not technically be separate from the credential itself).
+ *
+ * Parameters: struct image_params *   the image parameter block
+ *
+ * Returns:    void                    No failure indication
+ *
+ * Implicit returns:
+ *             <process credential>    Potentially modified/replaced
+ *             <task port>             Potentially revoked
+ *             <process flags>         P_SUGID bit potentially modified
+ *             <security token>        Potentially modified
+ */
+static int
+exec_handle_sugid(struct image_params *imgp)
+{
+       kauth_cred_t            cred = vfs_context_ucred(imgp->ip_vfs_context);
+       struct proc             *p = vfs_context_proc(imgp->ip_vfs_context);
+       int                     i;
+       int                     error = 0;
+       static struct vnode     *dev_null = NULLVP;
 
-       /*
-        * deal with set[ug]id.
-        */
        p->p_flag &= ~P_SUGID;
-       if (((origvattr.va_mode & VSUID) != 0 &&
-           p->p_ucred->cr_uid != origvattr.va_uid)
-           || (origvattr.va_mode & VSGID) != 0 &&
-           p->p_ucred->cr_gid != origvattr.va_gid) {
-               p->p_ucred = crcopy(cred);
+
+       if (((imgp->ip_origvattr->va_mode & VSUID) != 0 &&
+            kauth_cred_getuid(cred) != imgp->ip_origvattr->va_uid) ||
+           ((imgp->ip_origvattr->va_mode & VSGID) != 0 &&
+            cred->cr_gid != imgp->ip_origvattr->va_gid)) {
 #if KTRACE
                /*
                 * If process is being ktraced, turn off - unless
@@ -811,16 +1411,18 @@ again:
                        struct vnode *tvp = p->p_tracep;
                        p->p_tracep = NULL;
                        p->p_traceflag = 0;
-
-                       if (UBCINFOEXISTS(tvp))
-                               ubc_rele(tvp);
-                       vrele(tvp);
+                       vnode_rele(tvp);
                }
 #endif
-               if (origvattr.va_mode & VSUID)
-                       p->p_ucred->cr_uid = origvattr.va_uid;
-               if (origvattr.va_mode & VSGID)
-                       p->p_ucred->cr_gid = origvattr.va_gid;
+           /*
+                * Replace the credential with a copy of itself if euid or egid change.
+                */
+               if (imgp->ip_origvattr->va_mode & VSUID) {
+                       p->p_ucred = kauth_cred_seteuid(p->p_ucred, imgp->ip_origvattr->va_uid);
+               }
+               if (imgp->ip_origvattr->va_mode & VSGID) {
+                       p->p_ucred = kauth_cred_setegid(p->p_ucred, imgp->ip_origvattr->va_gid);
+               }
 
                /*
                 * Have mach reset the task port.  We don't want
@@ -828,10 +1430,30 @@ again:
                 * exec to be able to access/control the task
                 * after.
                 */
-               ipc_task_reset(task);
+               if (current_task() == p->task)
+                       ipc_task_reset(p->task);
 
                p->p_flag |= P_SUGID;
 
+               /* Cache the vnode for /dev/null the first time around */
+               if (dev_null == NULLVP) {
+                       struct nameidata nd1;
+
+                       NDINIT(&nd1, LOOKUP, FOLLOW, UIO_SYSSPACE32,
+                           CAST_USER_ADDR_T("/dev/null"),
+                           imgp->ip_vfs_context);
+
+                       if ((error = vn_open(&nd1, FREAD, 0)) == 0) {
+                               dev_null = nd1.ni_vp;
+                               /*
+                                * vn_open returns with both a use_count
+                                * and an io_count on the found vnode
+                                * drop the io_count, but keep the use_count
+                                */
+                               vnode_put(nd1.ni_vp);
+                       }
+               }
+
                /* Radar 2261856; setuid security hole fix */
                /* Patch from OpenBSD: A. Ramesh */
                /*
@@ -841,307 +1463,77 @@ again:
                 * descriptors in this range which has implied meaning
                 * to libc.
                 */
-               for (i = 0; i < 3; i++) {
-                       extern struct fileops vnops;
-                       struct nameidata nd1;
-                       struct file *fp;
-                       int indx;
+               if (dev_null != NULLVP) {
+                       for (i = 0; i < 3; i++) {
+                               struct fileproc *fp;
+                               int indx;
+
+                               if (p->p_fd->fd_ofiles[i] != NULL)
+                                       continue;
 
-                       if (p->p_fd->fd_ofiles[i] == NULL) {
                                if ((error = falloc(p, &fp, &indx)) != 0)
                                        continue;
-                               NDINIT(&nd1, LOOKUP, FOLLOW, UIO_SYSSPACE,
-                                   "/dev/null", p);
-                               if ((error = vn_open(&nd1, FREAD, 0)) != 0) {
-                                       ffree(fp);
-                                       p->p_fd->fd_ofiles[indx] = NULL;
+
+                               if ((error = vnode_ref_ext(dev_null, FREAD)) != 0) {
+                                       fp_free(p, indx, fp);
                                        break;
                                }
-                               fp->f_flag = FREAD;
-                               fp->f_type = DTYPE_VNODE;
-                               fp->f_ops = &vnops;
-                               fp->f_data = (caddr_t)nd1.ni_vp;
-                               VOP_UNLOCK(nd1.ni_vp, 0, p);
-                       }
-               }
-       }
-       p->p_cred->p_svuid = p->p_ucred->cr_uid;
-       p->p_cred->p_svgid = p->p_ucred->cr_gid;
-       set_security_token(p);
-
-       KNOTE(&p->p_klist, NOTE_EXEC);
-
-       if (!vfexec && (p->p_flag & P_TRACED))
-               psignal(p, SIGTRAP);
-
-       if (error) {
-               vrele(vp);
-               vp = NULL;
-               goto badtoolate;
-       }
-       VOP_LOCK(vp,  LK_EXCLUSIVE | LK_RETRY, p); /* XXX */
-       vput(vp);
-       vp = NULL;
-       
-       if (load_result.unixproc &&
-               create_unix_stack(get_task_map(task),
-                                 load_result.user_stack, load_result.customstack, p)) {
-               error = load_return_to_errno(LOAD_NOSPACE);
-               goto badtoolate;
-       }
-
-       if (vfexec) {
-               uthread->uu_ar0 = (void *)get_user_regs(thr_act);
-       }
-
-       /*
-        * Copy back arglist if necessary.
-        */
-
-
-       ucp = (int)p->user_stack;
-       if (vfexec) {
-               old_map = vm_map_switch(get_task_map(task));
-       }
-       if (load_result.unixproc) {
-               int pathptr;
-               
-               ucp = ucp - nc - NBPW;  /* begining of the STRING AREA */
-
-               /*
-                * Support for new app package launching for Mac OS X allocates
-                * the "path" at the begining of the execargs buffer.
-                * copy it just before the string area.
-                */
-               len = 0;
-               pathptr = ucp - ((savedpathlen + NBPW-1) & ~(NBPW-1));
-               error = copyoutstr(savedpath, (caddr_t)pathptr,
-                                       (unsigned)savedpathlen, (size_t *)&len);
-               savedpathlen = (savedpathlen + NBPW-1) & ~(NBPW-1);
-
-               if (error) {
-                       if (vfexec)
-                               vm_map_switch(old_map);
-                       goto badtoolate;
-               }
-
-               /*
-                * Record the size of the arguments area so that
-                * sysctl_procargs() can return the argument area without having
-                * to parse the arguments.
-                */
-               p->p_argslen = (int)p->user_stack - pathptr;
-               p->p_argc = na - ne;    /* save argc for sysctl_procargs() */
-
-               /* Save a NULL pointer below it */
-               (void) suword((caddr_t)(pathptr - NBPW), 0);
 
-               /* Save the pointer to "path" just below it */
-               (void) suword((caddr_t)(pathptr - 2*NBPW), pathptr);
-
-               /*
-                * na includes arg[] and env[].
-                * NBPW for 2 NULL one each ofter arg[argc -1] and env[n]
-                * NBPW for argc
-                * skip over saved path, NBPW for pointer to path,
-                * and NBPW for the NULL after pointer to path.
-                */
-               ap = ucp - na*NBPW - 3*NBPW - savedpathlen - 2*NBPW;
-#if defined(ppc)
-               thread_setuserstack(thr_act, ap);       /* Set the stack */
-#else
-               uthread->uu_ar0[SP] = ap;
-#endif
-               (void) suword((caddr_t)ap, na-ne); /* argc */
-               nc = 0;
-               cc = 0;
-
-               cp = (char *) execargsp;
-               cc = NCARGS - savedpathlen - 2*NBPW;
-               ps.ps_argvstr = (char *)ucp;    /* first argv string */
-               ps.ps_nargvstr = na - ne;               /* argc */
-               for (;;) {
-                       ap += NBPW;
-                       if (na == ne) {
-                               (void) suword((caddr_t)ap, 0);
-                               ap += NBPW;
-                               ps.ps_envstr = (char *)ucp;
-                               ps.ps_nenvstr = ne;
+                               fp->f_fglob->fg_flag = FREAD;
+                               fp->f_fglob->fg_type = DTYPE_VNODE;
+                               fp->f_fglob->fg_ops = &vnops;
+                               fp->f_fglob->fg_data = (caddr_t)dev_null;
+                               
+                               proc_fdlock(p);
+                               *fdflags(p, indx) &= ~UF_RESERVED;
+                               fp_drop(p, indx, fp, 1);
+                               proc_fdunlock(p);
                        }
-                       if (--na < 0)
-                               break;
-                       (void) suword((caddr_t)ap, ucp);
-                       do {
-                               error = copyoutstr(cp, (caddr_t)ucp,
-                                                  (unsigned)cc, (size_t *)&len);
-                               ucp += len;
-                               cp += len;
-                               nc += len;
-                               cc -= len;
-                       } while (error == ENAMETOOLONG);
-                       if (error == EFAULT)
-                               break;  /* bad stack - user's problem */
+                       /*
+                        * for now we need to drop the reference immediately
+                        * since we don't have any mechanism in place to
+                        * release it before starting to unmount "/dev"
+                        * during a reboot/shutdown
+                        */
+                       vnode_rele(dev_null);
+                       dev_null = NULLVP;
                }
-               (void) suword((caddr_t)ap, 0);
-       }
-       
-       if (load_result.dynlinker) {
-#if defined(ppc)
-               ap = thread_adjuserstack(thr_act, -4);  /* Adjust the stack */
-#else
-               ap = uthread->uu_ar0[SP] -= 4;
-#endif
-               (void) suword((caddr_t)ap, load_result.mach_header);
-       }
-
-       if (vfexec) {
-               vm_map_switch(old_map);
-       }
-#if defined(ppc)
-       thread_setentrypoint(thr_act, load_result.entry_point); /* Set the entry point */
-#elif defined(i386) 
-       uthread->uu_ar0[PC] = load_result.entry_point;
-#else
-#error architecture not implemented!
-#endif 
-
-       /* Stop profiling */
-       stopprofclock(p);
-
-       /*
-        * Reset signal state.
-        */
-       execsigs(p, thr_act);
-
-       /*
-        * Close file descriptors
-        * which specify close-on-exec.
-        */
-       fdexec(p);
-
-       /*
-        * need to cancel async IO requests that can be cancelled and wait for those
-        * already active.  MAY BLOCK!
-        */
-       _aio_exec( p );
-
-       /* FIXME: Till vmspace inherit is fixed: */
-       if (!vfexec && p->vm_shm)
-               shmexec(p);
-       /* Clean up the semaphores */
-       semexit(p);
-
-       /*
-        * Remember file name for accounting.
-        */
-       p->p_acflag &= ~AFORK;
-       /* If the translated name isn't NULL, then we want to use
-        * that translated name as the name we show as the "real" name.
-        * Otherwise, use the name passed into exec.
-        */
-       if (0 != binaryWithClassicName[0]) {
-               bcopy((caddr_t)binaryWithClassicName, (caddr_t)p->p_comm,
-                       sizeof(binaryWithClassicName));
-       } else {
-               if (nd.ni_cnd.cn_namelen > MAXCOMLEN)
-                       nd.ni_cnd.cn_namelen = MAXCOMLEN;
-               bcopy((caddr_t)nd.ni_cnd.cn_nameptr, (caddr_t)p->p_comm,
-                       (unsigned)nd.ni_cnd.cn_namelen);
-               p->p_comm[nd.ni_cnd.cn_namelen] = '\0';
-       }
-
-       {
-         /* This is for kdebug */
-         long dbg_arg1, dbg_arg2, dbg_arg3, dbg_arg4;
-
-         /* Collect the pathname for tracing */
-         kdbg_trace_string(p, &dbg_arg1, &dbg_arg2, &dbg_arg3, &dbg_arg4);
-
-
-
-         if (vfexec)
-         {
-                 KERNEL_DEBUG_CONSTANT1((TRACEDBG_CODE(DBG_TRACE_DATA, 2)) | DBG_FUNC_NONE,
-                                       p->p_pid ,0,0,0, (unsigned int)thr_act);
-                 KERNEL_DEBUG_CONSTANT1((TRACEDBG_CODE(DBG_TRACE_STRING, 2)) | DBG_FUNC_NONE,
-                                       dbg_arg1, dbg_arg2, dbg_arg3, dbg_arg4, (unsigned int)thr_act);
-         }
-         else
-         {
-                 KERNEL_DEBUG_CONSTANT((TRACEDBG_CODE(DBG_TRACE_DATA, 2)) | DBG_FUNC_NONE,
-                                       p->p_pid ,0,0,0,0);
-                 KERNEL_DEBUG_CONSTANT((TRACEDBG_CODE(DBG_TRACE_STRING, 2)) | DBG_FUNC_NONE,
-                                       dbg_arg1, dbg_arg2, dbg_arg3, dbg_arg4, 0);
-         }
        }
 
-       if (executingClassic)
-               p->p_flag |= P_CLASSIC | P_AFFINITY;
-       else
-               p->p_flag &= ~P_CLASSIC;
-
        /*
-        * mark as execed, wakeup the process that vforked (if any) and tell
-        * it that it now has it's own resources back
+        * Implement the semantic where the effective user and group become
+        * the saved user and group in exec'ed programs.
         */
-       p->p_flag |= P_EXEC;
-       if (p->p_pptr && (p->p_flag & P_PPWAIT)) {
-               p->p_flag &= ~P_PPWAIT;
-               wakeup((caddr_t)p->p_pptr);
-       }
-
-       if (vfexec && (p->p_flag & P_TRACED)) {
-                       psignal_vfork(p, new_task, thr_act, SIGTRAP);
-       }
+       p->p_ucred = kauth_cred_setsvuidgid(p->p_ucred, kauth_cred_getuid(p->p_ucred),  p->p_ucred->cr_gid);
+       
+       /* XXX Obsolete; security token should not be separate from cred */
+       set_security_token(p);
 
-badtoolate:
-       if (vfexec) {
-               task_deallocate(new_task);
-               act_deallocate(thr_act);
-               if (error)
-                       error = 0;
-       }
-bad:
-       FREE_ZONE(nd.ni_cnd.cn_pnbuf, nd.ni_cnd.cn_pnlen, M_NAMEI);
-       if (vp)
-               vput(vp);
-bad1:
-       if (execargs)
-               execargs_free(execargs);
-       if (!error && vfexec) {
-                       vfork_return(current_act(), p->p_pptr, p, retval);
-                       (void) thread_resume(thr_act);
-                       return(0);
-       }
        return(error);
 }
 
-
-#define        unix_stack_size(p)      (p->p_rlimit[RLIMIT_STACK].rlim_cur)
-
-kern_return_t
-create_unix_stack(map, user_stack, customstack, p)
-       vm_map_t        map;
-       vm_offset_t     user_stack;
-       int                     customstack;
-       struct proc     *p;
+static kern_return_t
+create_unix_stack(vm_map_t map, user_addr_t user_stack, int customstack,
+                       struct proc *p)
 {
-       vm_size_t       size;
-       vm_offset_t     addr;
+       mach_vm_size_t  size;
+       mach_vm_offset_t addr;
 
-       p->user_stack = (caddr_t)user_stack;
+       p->user_stack = user_stack;
        if (!customstack) {
-               size = round_page_64(unix_stack_size(p));
-               addr = trunc_page_32(user_stack - size);
-               return (vm_allocate(map, &addr, size,
-                                       VM_MAKE_TAG(VM_MEMORY_STACK) | FALSE));
+               size = mach_vm_round_page(unix_stack_size(p));
+               addr = mach_vm_trunc_page(user_stack - size);
+               return (mach_vm_allocate(map, &addr, size,
+                                       VM_MAKE_TAG(VM_MEMORY_STACK) |
+                                       VM_FLAGS_FIXED));
        } else
                return(KERN_SUCCESS);
 }
 
 #include <sys/reboot.h>
 
-char           init_program_name[128] = "/sbin/mach_init\0";
+static char            init_program_name[128] = "/sbin/launchd";
+static const char *    other_init = "/sbin/mach_init";
 
 char           init_args[128] = "";
 
@@ -1150,15 +1542,12 @@ int             init_attempts = 0;
 
 
 void
-load_init_program(p)
-       struct proc *p;
+load_init_program(struct proc *p)
 {
        vm_offset_t     init_addr;
-       int             *old_ap;
        char            *argv[3];
-       int             error;
-       register_t retval[2];
-       struct uthread * ut;
+       int                     error;
+       register_t      retval[2];
 
        error = 0;
 
@@ -1174,7 +1563,6 @@ load_init_program(p)
 
                if (error && ((boothowto & RB_INITNAME) == 0) &&
                                        (init_attempts == 1)) {
-                       static char other_init[] = "/etc/mach_init";
                        printf("Load of %s, errno %d, trying %s\n",
                                init_program_name, error, other_init);
                        error = 0;
@@ -1198,11 +1586,12 @@ load_init_program(p)
 
                init_addr = VM_MIN_ADDRESS;
                (void) vm_allocate(current_map(), &init_addr,
-                                  PAGE_SIZE, TRUE);
+                                  PAGE_SIZE, VM_FLAGS_ANYWHERE);
                if (init_addr == 0)
                        init_addr++;
+
                (void) copyout((caddr_t) init_program_name,
-                               (caddr_t) (init_addr),
+                               CAST_USER_ADDR_T(init_addr),
                                (unsigned) sizeof(init_program_name)+1);
 
                argv[0] = (char *) init_addr;
@@ -1216,7 +1605,7 @@ load_init_program(p)
                 */
 
                (void) copyout((caddr_t) init_args,
-                               (caddr_t) (init_addr),
+                               CAST_USER_ADDR_T(init_addr),
                                (unsigned) sizeof(init_args));
 
                argv[1] = (char *) init_addr;
@@ -1234,16 +1623,16 @@ load_init_program(p)
                 */
                
                (void) copyout((caddr_t) argv,
-                               (caddr_t) (init_addr),
+                               CAST_USER_ADDR_T(init_addr),
                                (unsigned) sizeof(argv));
 
                /*
                 *      Set up argument block for fake call to execve.
                 */
 
-               init_exec_args.fname = argv[0];
-               init_exec_args.argp = (char **) init_addr;
-               init_exec_args.envp = 0;
+               init_exec_args.fname = CAST_USER_ADDR_T(argv[0]);
+               init_exec_args.argp = CAST_USER_ADDR_T((char **)init_addr);
+               init_exec_args.envp = CAST_USER_ADDR_T(0);
                
                /* So that mach_init task 
                 * is set with uid,gid 0 token 
@@ -1284,31 +1673,6 @@ load_return_to_errno(load_return_t lrtn)
        }
 }
 
-/*
- * exec_check_access()
- */
-int
-check_exec_access(p, vp, vap)
-       struct proc  *p;
-       struct vnode *vp;
-       struct vattr *vap;
-{
-       int flag;
-       int error;
-
-       if (error = VOP_ACCESS(vp, VEXEC, p->p_ucred, p))
-               return (error);
-       flag = p->p_flag;
-       if (flag & P_TRACED) {
-               if (error = VOP_ACCESS(vp, VREAD, p->p_ucred, p))
-                       return (error);
-       }
-       if (vp->v_type != VREG ||
-           (vap->va_mode & (S_IXUSR | S_IXGRP | S_IXOTH)) == 0)
-               return (EACCES);
-       return (0);
-}
-
 #include <mach/mach_types.h>
 #include <mach/vm_prot.h>
 #include <mach/semaphore.h>
@@ -1318,9 +1682,12 @@ check_exec_access(p, vp, vap)
 
 extern semaphore_t execve_semaphore;
 
+/*
+ * The block of memory used by the execve arguments.  At the same time,
+ * we allocate a page so that we can read in the first page of the image.
+ */
 static int
-execargs_alloc(addrp)
-       vm_offset_t     *addrp;
+execargs_alloc(struct image_params *imgp)
 {
        kern_return_t kret;
 
@@ -1337,7 +1704,8 @@ execargs_alloc(addrp)
                        return (EINTR);
                }
 
-       kret = kmem_alloc_pageable(bsd_pageable_map, addrp, NCARGS);
+       kret = kmem_alloc_pageable(bsd_pageable_map, (vm_offset_t *)&imgp->ip_strings, NCARGS + PAGE_SIZE);
+       imgp->ip_vdata = imgp->ip_strings + NCARGS;
        if (kret != KERN_SUCCESS) {
                semaphore_signal(execve_semaphore);
                return (ENOMEM);
@@ -1346,12 +1714,12 @@ execargs_alloc(addrp)
 }
 
 static int
-execargs_free(addr)
-       vm_offset_t     addr;
+execargs_free(struct image_params *imgp)
 {
        kern_return_t kret;
 
-       kmem_free(bsd_pageable_map, addr, NCARGS);
+       kmem_free(bsd_pageable_map, (vm_offset_t)imgp->ip_strings, NCARGS + PAGE_SIZE);
+       imgp->ip_strings = NULL;
 
        kret = semaphore_signal(execve_semaphore);
        switch (kret) {