]> git.saurik.com Git - apple/xnu.git/blobdiff - bsd/kern/kern_resource.c
xnu-3789.1.32.tar.gz
[apple/xnu.git] / bsd / kern / kern_resource.c
index 5e2f8c17158a37f4d4619451fdb8a5a662bd9363..6259167150b53f6f8b7cf84c04568c0423456093 100644 (file)
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
+ * Copyright (c) 2000-2008 Apple Inc. All rights reserved.
  *
  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
  * 
@@ -86,7 +86,7 @@
 #include <sys/mount_internal.h>
 #include <sys/sysproto.h>
 
-#include <bsm/audit_kernel.h>
+#include <security/audit/audit.h>
 
 #include <machine/vmparam.h>
 
 #include <mach/vm_map.h>
 #include <mach/mach_vm.h>
 #include <mach/thread_act.h>  /* for thread_policy_set( ) */
-#include <kern/lock.h>
 #include <kern/thread.h>
+#include <kern/policy_internal.h>
 
 #include <kern/task.h>
 #include <kern/clock.h>                /* for absolutetime_to_microtime() */
-#include <netinet/in.h>                /* for TRAFFIC_MGT_SO_BACKGROUND */
+#include <netinet/in.h>                /* for TRAFFIC_MGT_SO_* */
 #include <sys/socketvar.h>     /* for struct socket */
 
 #include <vm/vm_map.h>
 
+#include <kern/assert.h>
+#include <sys/resource.h>
+#include <sys/priv.h>
+#include <IOKit/IOBSD.h>
+
 int    donice(struct proc *curp, struct proc *chgp, int n);
 int    dosetrlimit(struct proc *p, u_int which, struct rlimit *limp);
-static int do_background_thread(struct proc *curp, int priority);
+int    uthread_get_background_state(uthread_t);
+static void do_background_socket(struct proc *p, thread_t thread);
+static int do_background_thread(thread_t thread, int priority);
+static int do_background_proc(struct proc *curp, struct proc *targetp, int priority);
+static int set_gpudeny_proc(struct proc *curp, struct proc *targetp, int priority);
+static int proc_set_darwin_role(proc_t curp, proc_t targetp, int priority);
+static int proc_get_darwin_role(proc_t curp, proc_t targetp, int *priority);
+static int get_background_proc(struct proc *curp, struct proc *targetp, int *priority);
+int proc_pid_rusage(int pid, int flavor, user_addr_t buf, int32_t *retval);
+void gather_rusage_info(proc_t p, rusage_info_current *ru, int flavor);
+int fill_task_rusage(task_t task, rusage_info_current *ri);
+void fill_task_billed_usage(task_t task, rusage_info_current *ri);
+int fill_task_io_rusage(task_t task, rusage_info_current *ri);
+int fill_task_qos_rusage(task_t task, rusage_info_current *ri);
+static void rusage_info_conversion(rusage_info_t ri_info, rusage_info_current *ri_current, int flavor);
+
+int proc_get_rusage(proc_t p, int flavor, user_addr_t buffer, __unused int is_zombie);
 
 rlim_t maxdmap = MAXDSIZ;      /* XXX */ 
-rlim_t maxsmap = MAXSSIZ - PAGE_SIZE;  /* XXX */ 
+rlim_t maxsmap = MAXSSIZ - PAGE_MAX_SIZE;      /* XXX */ 
 
 /*
  * Limits on the number of open files per process, and the number
@@ -120,12 +141,12 @@ rlim_t maxsmap = MAXSSIZ - PAGE_SIZE;     /* XXX */
  *
  * Note: would be in kern/subr_param.c in FreeBSD.
  */
-int maxfilesperproc = OPEN_MAX;                /* per-proc open files limit */
+__private_extern__ int maxfilesperproc = OPEN_MAX;             /* per-proc open files limit */
 
-SYSCTL_INT( _kern, KERN_MAXPROCPERUID, maxprocperuid, CTLFLAG_RW,
+SYSCTL_INT(_kern, KERN_MAXPROCPERUID, maxprocperuid, CTLFLAG_RW | CTLFLAG_LOCKED,
                &maxprocperuid, 0, "Maximum processes allowed per userid" );
 
-SYSCTL_INT( _kern, KERN_MAXFILESPERPROC, maxfilesperproc, CTLFLAG_RW,       
+SYSCTL_INT(_kern, KERN_MAXFILESPERPROC, maxfilesperproc, CTLFLAG_RW | CTLFLAG_LOCKED,
                &maxfilesperproc, 0, "Maximum files allowed open per process" );
 
 /* Args and fn for proc_iteration callback used in setpriority */
@@ -152,11 +173,13 @@ static int ppgrp_donice_callback(proc_t p, void * arg);
  * Resource controls and accounting.
  */
 int
-getpriority(struct proc *curp, struct getpriority_args *uap, register_t *retval)
+getpriority(struct proc *curp, struct getpriority_args *uap, int32_t *retval)
 {
        struct proc *p;
        int low = PRIO_MAX + 1;
        kauth_cred_t my_cred;
+       int refheld = 0;
+       int error = 0;
 
        /* would also test (uap->who < 0), but id_t is unsigned */
        if (uap->who > 0x7fffffff)
@@ -189,7 +212,7 @@ getpriority(struct proc *curp, struct getpriority_args *uap, register_t *retval)
                }
                /* No need for iteration as it is a simple scan */
                pgrp_lock(pg);
-               for (p = pg->pg_members.lh_first; p != 0; p = p->p_pglist.le_next) {
+               PGMEMBERS_FOREACH(pg, p) {
                        if (p->p_nice < low)
                                low = p->p_nice;
                }
@@ -216,24 +239,50 @@ getpriority(struct proc *curp, struct getpriority_args *uap, register_t *retval)
 
                break;
 
-       case PRIO_DARWIN_THREAD: {
-               thread_t                        thread;
-               struct uthread          *ut;
-
+       case PRIO_DARWIN_THREAD:
                /* we currently only support the current thread */
-               if (uap->who != 0) {
+               if (uap->who != 0)
                        return (EINVAL);
+
+               low = proc_get_thread_policy(current_thread(), TASK_POLICY_INTERNAL, TASK_POLICY_DARWIN_BG);
+
+               break;
+
+       case PRIO_DARWIN_PROCESS:
+               if (uap->who == 0) {
+                       p = curp;
+               } else {
+                       p = proc_find(uap->who);
+                       if (p == PROC_NULL)
+                               break;
+                       refheld = 1;
                }
-       
-               thread = current_thread();
-               ut = get_bsdthread_info(thread);
 
-               low = 0;
-               if ( (ut->uu_flag & UT_BACKGROUND) != 0 ) {
-                       low = 1;
+               error = get_background_proc(curp, p, &low);
+
+               if (refheld)
+                       proc_rele(p);
+               if (error)
+                       return (error);
+               break;
+
+       case PRIO_DARWIN_ROLE:
+               if (uap->who == 0) {
+                       p = curp;
+               } else {
+                       p = proc_find(uap->who);
+                       if (p == PROC_NULL)
+                               break;
+                       refheld = 1;
                }
+
+               error = proc_get_darwin_role(curp, p, &low);
+
+               if (refheld)
+                       proc_rele(p);
+               if (error)
+                       return (error);
                break;
-       }
 
        default:
                return (EINVAL);
@@ -295,7 +344,7 @@ ppgrp_donice_callback(proc_t p, void * arg)
  */
 /* ARGSUSED */
 int
-setpriority(struct proc *curp, struct setpriority_args *uap, __unused register_t *retval)
+setpriority(struct proc *curp, struct setpriority_args *uap, int32_t *retval)
 {
        struct proc *p;
        int found = 0, error = 0;
@@ -303,7 +352,7 @@ setpriority(struct proc *curp, struct setpriority_args *uap, __unused register_t
 
        AUDIT_ARG(cmd, uap->which);
        AUDIT_ARG(owner, uap->who, 0);
-       AUDIT_ARG(value, uap->prio);
+       AUDIT_ARG(value32, uap->prio);
 
        /* would also test (uap->who < 0), but id_t is unsigned */
        if (uap->who > 0x7fffffff)
@@ -365,11 +414,62 @@ setpriority(struct proc *curp, struct setpriority_args *uap, __unused register_t
 
        case PRIO_DARWIN_THREAD: {
                /* we currently only support the current thread */
-               if (uap->who != 0) {
+               if (uap->who != 0)
                        return (EINVAL);
+
+               error = do_background_thread(current_thread(), uap->prio);
+               found++;
+               break;
+       }
+
+       case PRIO_DARWIN_PROCESS: {
+               if (uap->who == 0)
+                       p = curp;
+               else {
+                       p = proc_find(uap->who);
+                       if (p == 0)
+                               break;
+                       refheld = 1;
                }
-               error = do_background_thread(curp, uap->prio);
+
+               error = do_background_proc(curp, p, uap->prio);
+
+               found++;
+               if (refheld != 0)
+                       proc_rele(p);
+               break;
+       }
+
+       case PRIO_DARWIN_GPU: {
+               if (uap->who == 0)
+                       return (EINVAL);
+
+               p = proc_find(uap->who);
+               if (p == PROC_NULL)
+                       break;
+
+               error = set_gpudeny_proc(curp, p, uap->prio);
+
                found++;
+               proc_rele(p);
+               break;
+       }
+
+       case PRIO_DARWIN_ROLE: {
+               if (uap->who == 0) {
+                       p = curp;
+               } else {
+                       p = proc_find(uap->who);
+                       if (p == PROC_NULL)
+                               break;
+                       refheld = 1;
+               }
+
+               error = proc_set_darwin_role(curp, p, uap->prio);
+
+               found++;
+               if (refheld != 0)
+                       proc_rele(p);
                break;
        }
 
@@ -378,6 +478,10 @@ setpriority(struct proc *curp, struct setpriority_args *uap, __unused register_t
        }
        if (found == 0)
                return (ESRCH);
+       if (error == EIDRM) {
+               *retval = -2;
+               error = 0;
+       }
        return (error);
 }
 
@@ -398,9 +502,9 @@ donice(struct proc *curp, struct proc *chgp, int n)
        ucred = kauth_cred_proc_ref(curp);
        my_cred = kauth_cred_proc_ref(chgp);
 
-       if (suser(ucred, NULL) && ucred->cr_ruid &&
+       if (suser(ucred, NULL) && kauth_cred_getruid(ucred) &&
            kauth_cred_getuid(ucred) != kauth_cred_getuid(my_cred) &&
-           ucred->cr_ruid != kauth_cred_getuid(my_cred)) {
+           kauth_cred_getruid(ucred) != kauth_cred_getuid(my_cred)) {
                error = EPERM;
                goto out;
        }
@@ -427,88 +531,321 @@ out:
        return (error);
 }
 
-/*
- * do_background_thread
- * Returns:    0                       Success
- * XXX - todo - does this need a MACF hook?
- */
 static int
-do_background_thread(struct proc *curp, int priority)
+set_gpudeny_proc(struct proc *curp, struct proc *targetp, int priority)
 {
-       int                                                                     i;
-       thread_t                                                        thread;
-       struct uthread                                          *ut;
-       thread_precedence_policy_data_t         policy;
-       struct filedesc                                         *fdp;
-       struct fileproc                                         *fp;
-       
-       thread = current_thread();
-       ut = get_bsdthread_info(thread);
+       int error = 0;
+       kauth_cred_t ucred;
+       kauth_cred_t target_cred;
+
+       ucred = kauth_cred_get();
+       target_cred = kauth_cred_proc_ref(targetp);
+
+       /* TODO: Entitlement instead of uid check */
+
+       if (!kauth_cred_issuser(ucred) && kauth_cred_getruid(ucred) &&
+           kauth_cred_getuid(ucred)   != kauth_cred_getuid(target_cred) &&
+           kauth_cred_getruid(ucred)  != kauth_cred_getuid(target_cred)) {
+               error = EPERM;
+               goto out;
+       }
+
+       if (curp == targetp) {
+               error = EPERM;
+               goto out;
+       }
 
-       if ( (priority & PRIO_DARWIN_BG) == 0 ) {
-               /* turn off backgrounding of thread */
-               if ( (ut->uu_flag & UT_BACKGROUND) == 0 ) {
-                       /* already off */
-                       return(0);
+#if CONFIG_MACF
+       error = mac_proc_check_sched(curp, targetp);
+       if (error)
+               goto out;
+#endif
+
+       switch (priority) {
+               case PRIO_DARWIN_GPU_DENY:
+                       task_set_gpu_denied(proc_task(targetp), TRUE);
+                       break;
+               case PRIO_DARWIN_GPU_ALLOW:
+                       task_set_gpu_denied(proc_task(targetp), FALSE);
+                       break;
+               default:
+                       error = EINVAL;
+                       goto out;
+       }
+
+out:
+       kauth_cred_unref(&target_cred);
+       return (error);
+
+}
+
+static int
+proc_set_darwin_role(proc_t curp, proc_t targetp, int priority)
+{
+       int error = 0;
+       uint32_t flagsp;
+
+       kauth_cred_t ucred, target_cred;
+
+       ucred = kauth_cred_get();
+       target_cred = kauth_cred_proc_ref(targetp);
+
+       if (!kauth_cred_issuser(ucred) && kauth_cred_getruid(ucred) &&
+           kauth_cred_getuid(ucred)  != kauth_cred_getuid(target_cred) &&
+           kauth_cred_getruid(ucred) != kauth_cred_getuid(target_cred)) {
+               if (priv_check_cred(ucred, PRIV_SETPRIORITY_DARWIN_ROLE, 0) != 0) {
+                       error = EPERM;
+                       goto out;
                }
+       }
 
-               /* clear background bit in thread and disable disk IO throttle */
-               ut->uu_flag &= ~UT_BACKGROUND;
-               ut->uu_iopol_disk = IOPOL_NORMAL;
+       if (curp != targetp) {
+#if CONFIG_MACF
+               if ((error = mac_proc_check_sched(curp, targetp)))
+                       goto out;
+#endif
+       }
+
+       proc_get_darwinbgstate(proc_task(targetp), &flagsp);
+       if ((flagsp & PROC_FLAG_APPLICATION) != PROC_FLAG_APPLICATION) {
+               error = ENOTSUP;
+               goto out;
+       }
+
+       integer_t role = 0;
+
+       if ((error = proc_darwin_role_to_task_role(priority, &role)))
+               goto out;
+
+       proc_set_task_policy(proc_task(targetp), TASK_POLICY_ATTRIBUTE,
+                            TASK_POLICY_ROLE, role);
+
+out:
+       kauth_cred_unref(&target_cred);
+       return (error);
+}
+
+static int
+proc_get_darwin_role(proc_t curp, proc_t targetp, int *priority)
+{
+       int error = 0;
+       int role = 0;
+
+       kauth_cred_t ucred, target_cred;
+
+       ucred = kauth_cred_get();
+       target_cred = kauth_cred_proc_ref(targetp);
+
+       if (!kauth_cred_issuser(ucred) && kauth_cred_getruid(ucred) &&
+           kauth_cred_getuid(ucred)  != kauth_cred_getuid(target_cred) &&
+           kauth_cred_getruid(ucred) != kauth_cred_getuid(target_cred)) {
+               error = EPERM;
+               goto out;
+       }
+
+       if (curp != targetp) {
+#if CONFIG_MACF
+               if ((error = mac_proc_check_sched(curp, targetp)))
+                       goto out;
+#endif
+       }
+
+       role = proc_get_task_policy(proc_task(targetp), TASK_POLICY_ATTRIBUTE, TASK_POLICY_ROLE);
+
+       *priority = proc_task_role_to_darwin_role(role);
+
+out:
+       kauth_cred_unref(&target_cred);
+       return (error);
+}
+
+
+static int
+get_background_proc(struct proc *curp, struct proc *targetp, int *priority)
+{
+       int external = 0;
+       int error = 0;
+       kauth_cred_t ucred, target_cred;
+
+       ucred = kauth_cred_get();
+       target_cred = kauth_cred_proc_ref(targetp);
+
+       if (!kauth_cred_issuser(ucred) && kauth_cred_getruid(ucred) &&
+           kauth_cred_getuid(ucred) != kauth_cred_getuid(target_cred) &&
+           kauth_cred_getruid(ucred) != kauth_cred_getuid(target_cred)) {
+               error = EPERM;
+               goto out;
+       }
+
+       external = (curp == targetp) ? TASK_POLICY_INTERNAL : TASK_POLICY_EXTERNAL;
+
+       *priority = proc_get_task_policy(current_task(), external, TASK_POLICY_DARWIN_BG);
+
+out:
+       kauth_cred_unref(&target_cred);
+       return (error);
+}
+
+static int
+do_background_proc(struct proc *curp, struct proc *targetp, int priority)
+{
+#if !CONFIG_MACF
+#pragma unused(curp)
+#endif
+       int error = 0;
+       kauth_cred_t ucred;
+       kauth_cred_t target_cred;
+       int external;
+       int enable;
+
+       ucred = kauth_cred_get();
+       target_cred = kauth_cred_proc_ref(targetp);
+
+       if (!kauth_cred_issuser(ucred) && kauth_cred_getruid(ucred) &&
+               kauth_cred_getuid(ucred) != kauth_cred_getuid(target_cred) &&
+               kauth_cred_getruid(ucred) != kauth_cred_getuid(target_cred))
+       {
+               error = EPERM;
+               goto out;
+       }
+
+#if CONFIG_MACF
+       error = mac_proc_check_sched(curp, targetp);
+       if (error) 
+               goto out;
+#endif
 
-               /* reset thread priority (we did not save previous value) */
-               policy.importance = 0;
-               thread_policy_set( thread, THREAD_PRECEDENCE_POLICY,
-                                                  (thread_policy_t)&policy,
-                                                  THREAD_PRECEDENCE_POLICY_COUNT );
+       external = (curp == targetp) ? TASK_POLICY_INTERNAL : TASK_POLICY_EXTERNAL;
 
+       switch (priority) {
+               case PRIO_DARWIN_BG:
+                       enable = TASK_POLICY_ENABLE;
+                       break;
+               case PRIO_DARWIN_NONUI:
+                       /* ignored for compatibility */
+                       goto out;
+               default:
+                       /* TODO: EINVAL if priority != 0 */
+                       enable = TASK_POLICY_DISABLE;
+                       break;
+       }
+
+       proc_set_task_policy(proc_task(targetp), external, TASK_POLICY_DARWIN_BG, enable);
+
+out:
+       kauth_cred_unref(&target_cred);
+       return (error);
+}
+
+static void 
+do_background_socket(struct proc *p, thread_t thread)
+{
+#if SOCKETS
+       struct filedesc                     *fdp;
+       struct fileproc                     *fp;
+       int                                 i, background;
+
+       proc_fdlock(p);
+
+       if (thread != THREAD_NULL)
+               background = proc_get_effective_thread_policy(thread, TASK_POLICY_ALL_SOCKETS_BG);
+       else
+               background = proc_get_effective_task_policy(proc_task(p), TASK_POLICY_ALL_SOCKETS_BG);
+
+       if (background) {
+               /*
+                * For PRIO_DARWIN_PROCESS (thread is NULL), simply mark
+                * the sockets with the background flag.  There's nothing
+                * to do here for the PRIO_DARWIN_THREAD case.
+                */
+               if (thread == THREAD_NULL) {
+                       fdp = p->p_fd;
+
+                       for (i = 0; i < fdp->fd_nfiles; i++) {
+                               struct socket       *sockp;
+
+                               fp = fdp->fd_ofiles[i];
+                               if (fp == NULL || (fdp->fd_ofileflags[i] & UF_RESERVED) != 0 ||
+                                   FILEGLOB_DTYPE(fp->f_fglob) != DTYPE_SOCKET) {
+                                       continue;
+                               }
+                               sockp = (struct socket *)fp->f_fglob->fg_data;
+                               socket_set_traffic_mgt_flags(sockp, TRAFFIC_MGT_SO_BACKGROUND);
+                               sockp->so_background_thread = NULL;
+                       }
+               }
+       } else {
                /* disable networking IO throttle.
                 * NOTE - It is a known limitation of the current design that we 
                 * could potentially clear TRAFFIC_MGT_SO_BACKGROUND bit for 
                 * sockets created by other threads within this process.  
                 */
-               proc_fdlock(curp);
-               fdp = curp->p_fd;
+               fdp = p->p_fd;
                for ( i = 0; i < fdp->fd_nfiles; i++ ) {
-                       struct socket           *sockp;
-                       
+                       struct socket       *sockp;
+
                        fp = fdp->fd_ofiles[ i ];
-                       if ( fp == NULL || (fdp->fd_ofileflags[ i ] & UF_RESERVED) != 0 || 
-                                fp->f_fglob->fg_type != DTYPE_SOCKET ) {
+                       if ( fp == NULL || (fdp->fd_ofileflags[ i ] & UF_RESERVED) != 0 ||
+                           FILEGLOB_DTYPE(fp->f_fglob) != DTYPE_SOCKET ) {
                                continue;
                        }
                        sockp = (struct socket *)fp->f_fglob->fg_data;
-                       if ( sockp->so_background_thread != thread ) {
+                       /* skip if only clearing this thread's sockets */
+                       if ((thread) && (sockp->so_background_thread != thread)) {
                                continue;
                        }
-                       sockp->so_traffic_mgt_flags &= ~TRAFFIC_MGT_SO_BACKGROUND;
+                       socket_clear_traffic_mgt_flags(sockp, TRAFFIC_MGT_SO_BACKGROUND);
                        sockp->so_background_thread = NULL;
                }
-               proc_fdunlock(curp);
+       }
 
-               return(0);
+       proc_fdunlock(p);
+#else
+#pragma unused(p, thread)
+#endif
+}
+
+
+/*
+ * do_background_thread
+ *
+ * Requires: thread reference
+ *
+ * Returns:     0                       Success
+ *              EPERM                   Tried to background while in vfork
+ * XXX - todo - does this need a MACF hook?
+ */
+static int
+do_background_thread(thread_t thread, int priority)
+{
+       struct uthread *ut;
+       int enable, external;
+       int rv = 0;
+
+       ut = get_bsdthread_info(thread);
+
+       /* Backgrounding is unsupported for threads in vfork */
+       if ((ut->uu_flag & UT_VFORK) != 0)
+               return(EPERM);
+
+       /* Backgrounding is unsupported for workq threads */
+       if (thread_is_static_param(thread)) {
+               return(EPERM);
        }
-       
-       /* background this thread */
-       if ( (ut->uu_flag & UT_BACKGROUND) != 0 ) {
-               /* already backgrounded */
-               return(0);
+
+       /* Not allowed to combine QoS and DARWIN_BG, doing so strips the QoS */
+       if (thread_has_qos_policy(thread)) {
+               thread_remove_qos_policy(thread);
+               rv = EIDRM;
        }
 
-       /* tag thread as background and throttle disk IO */
-       ut->uu_flag |= UT_BACKGROUND;
-       ut->uu_iopol_disk = IOPOL_THROTTLE;
+       /* TODO: Fail if someone passes something besides 0 or PRIO_DARWIN_BG */
+       enable   = (priority == PRIO_DARWIN_BG) ? TASK_POLICY_ENABLE   : TASK_POLICY_DISABLE;
+       external = (current_thread() == thread) ? TASK_POLICY_INTERNAL : TASK_POLICY_EXTERNAL;
 
-       policy.importance = INT_MIN;
-       thread_policy_set( thread, THREAD_PRECEDENCE_POLICY,
-                                          (thread_policy_t)&policy,
-                                          THREAD_PRECEDENCE_POLICY_COUNT );
-       
-       /* throttle networking IO happens in socket( ) syscall.
-        * If UT_BACKGROUND is set in the current thread then
-        * TRAFFIC_MGT_SO_BACKGROUND socket option is set.
-        */
-       return(0);
+       proc_set_thread_policy(thread, external, TASK_POLICY_DARWIN_BG, enable);
+
+       return rv;
 }
 
 
@@ -519,7 +856,7 @@ do_background_thread(struct proc *curp, int priority)
  */
 /* ARGSUSED */
 int
-setrlimit(struct proc *p, struct setrlimit_args *uap, __unused register_t *retval)
+setrlimit(struct proc *p, struct setrlimit_args *uap, __unused int32_t *retval)
 {
        struct rlimit alim;
        int error;
@@ -585,12 +922,16 @@ dosetrlimit(struct proc *p, u_int which, struct rlimit *limp)
                        task_absolutetime_info_data_t   tinfo;
                        mach_msg_type_number_t                  count;
                        struct timeval                                  ttv, tv;
+                       clock_sec_t                                             tv_sec;
+                       clock_usec_t                                    tv_usec;
 
                        count = TASK_ABSOLUTETIME_INFO_COUNT;
                        task_info(p->task, TASK_ABSOLUTETIME_INFO,
                                                                (task_info_t)&tinfo, &count);
                        absolutetime_to_microtime(tinfo.total_user + tinfo.total_system,
-                                                               (uint32_t *) &ttv.tv_sec, (uint32_t *) &ttv.tv_usec);
+                                                                         &tv_sec, &tv_usec);
+                       ttv.tv_sec = tv_sec;
+                       ttv.tv_usec = tv_usec;
 
                        tv.tv_sec = (limp->rlim_cur > __INT_MAX__ ? __INT_MAX__ : limp->rlim_cur);
                        tv.tv_usec = 0;
@@ -650,12 +991,7 @@ dosetrlimit(struct proc *p, u_int which, struct rlimit *limp)
                                size = round_page_64(limp->rlim_cur);
                                size -= round_page_64(alimp->rlim_cur);
 
-#if STACK_GROWTH_UP
-                               /* go to top of current stack */
-                       addr = p->user_stack + round_page_64(alimp->rlim_cur);
-#else  /* STACK_GROWTH_UP */
                        addr = p->user_stack - round_page_64(limp->rlim_cur);
-#endif /* STACK_GROWTH_UP */
                        kr = mach_vm_protect(current_map(), 
                                             addr, size,
                                             FALSE, VM_PROT_DEFAULT);
@@ -677,28 +1013,6 @@ dosetrlimit(struct proc *p, u_int which, struct rlimit *limp)
                         */
                        cur_sp = thread_adjuserstack(current_thread(),
                                                     0);
-#if STACK_GROWTH_UP
-                       if (cur_sp >= p->user_stack &&
-                           cur_sp < (p->user_stack +
-                                     round_page_64(alimp->rlim_cur))) {
-                               /* current stack pointer is in main stack */
-                               if (cur_sp >= (p->user_stack +
-                                              round_page_64(limp->rlim_cur))) {
-                                       /*
-                                        * New limit would cause
-                                        * current usage to be invalid:
-                                        * reject new limit.
-                                        */
-                                       error =  EINVAL;
-                                       goto out;
-                       }
-                       } else {
-                               /* not on the main stack: reject */
-                               error =  EINVAL;
-                               goto out;
-               }
-                                
-#else  /* STACK_GROWTH_UP */
                        if (cur_sp <= p->user_stack &&
                            cur_sp > (p->user_stack -
                                      round_page_64(alimp->rlim_cur))) {
@@ -718,16 +1032,11 @@ dosetrlimit(struct proc *p, u_int which, struct rlimit *limp)
                                error =  EINVAL;
                                goto out;
                        }
-#endif /* STACK_GROWTH_UP */
                                
                        size = round_page_64(alimp->rlim_cur);
                        size -= round_page_64(limp->rlim_cur);
 
-#if STACK_GROWTH_UP
-                       addr = p->user_stack + round_page_64(limp->rlim_cur);
-#else  /* STACK_GROWTH_UP */
                        addr = p->user_stack - round_page_64(alimp->rlim_cur);
-#endif /* STACK_GROWTH_UP */
 
                        kr = mach_vm_protect(current_map(),
                                             addr, size,
@@ -749,7 +1058,7 @@ dosetrlimit(struct proc *p, u_int which, struct rlimit *limp)
                 * because historically, people have been able to attempt to
                 * set RLIM_INFINITY to get "whatever the maximum is".
                */
-               if ( is_suser() ) {
+               if ( kauth_cred_issuser(kauth_cred_get()) ) {
                        if (limp->rlim_cur != alimp->rlim_cur &&
                            limp->rlim_cur > (rlim_t)maxfiles) {
                                if (posix) {
@@ -783,7 +1092,7 @@ dosetrlimit(struct proc *p, u_int which, struct rlimit *limp)
                 * systemwide resource; all others are limited to
                 * maxprocperuid (presumably less than maxproc).
                 */
-               if ( is_suser() ) {
+               if ( kauth_cred_issuser(kauth_cred_get()) ) {
                        if (limp->rlim_cur > (rlim_t)maxproc)
                                limp->rlim_cur = maxproc;
                        if (limp->rlim_max > (rlim_t)maxproc)
@@ -817,7 +1126,7 @@ out:
 
 /* ARGSUSED */
 int
-getrlimit(struct proc *p, struct getrlimit_args *uap, __unused register_t *retval)
+getrlimit(struct proc *p, struct getrlimit_args *uap, __unused int32_t *retval)
 {
        struct rlimit lim;
 
@@ -851,14 +1160,16 @@ calcru(struct proc *p, struct timeval *up, struct timeval *sp, struct timeval *i
 
        task = p->task;
        if (task) {
-               task_basic_info_data_t tinfo;
+               mach_task_basic_info_data_t tinfo;
                task_thread_times_info_data_t ttimesinfo;
-               mach_msg_type_number_t task_info_stuff, task_ttimes_stuff;
+               task_events_info_data_t teventsinfo;
+               mach_msg_type_number_t task_info_count, task_ttimes_count;
+               mach_msg_type_number_t task_events_count;
                struct timeval ut,st;
 
-               task_info_stuff = TASK_BASIC_INFO_COUNT;
-               task_info(task, TASK_BASIC_INFO,
-                         (task_info_t)&tinfo, &task_info_stuff);
+               task_info_count = MACH_TASK_BASIC_INFO_COUNT;
+               task_info(task, MACH_TASK_BASIC_INFO,
+                         (task_info_t)&tinfo, &task_info_count);
                ut.tv_sec = tinfo.user_time.seconds;
                ut.tv_usec = tinfo.user_time.microseconds;
                st.tv_sec = tinfo.system_time.seconds;
@@ -866,9 +1177,9 @@ calcru(struct proc *p, struct timeval *up, struct timeval *sp, struct timeval *i
                timeradd(&ut, up, up);
                timeradd(&st, sp, sp);
 
-               task_ttimes_stuff = TASK_THREAD_TIMES_INFO_COUNT;
+               task_ttimes_count = TASK_THREAD_TIMES_INFO_COUNT;
                task_info(task, TASK_THREAD_TIMES_INFO,
-                         (task_info_t)&ttimesinfo, &task_ttimes_stuff);
+                         (task_info_t)&ttimesinfo, &task_ttimes_count);
 
                ut.tv_sec = ttimesinfo.user_time.seconds;
                ut.tv_usec = ttimesinfo.user_time.microseconds;
@@ -876,17 +1187,37 @@ calcru(struct proc *p, struct timeval *up, struct timeval *sp, struct timeval *i
                st.tv_usec = ttimesinfo.system_time.microseconds;
                timeradd(&ut, up, up);
                timeradd(&st, sp, sp);
+
+               task_events_count = TASK_EVENTS_INFO_COUNT;
+               task_info(task, TASK_EVENTS_INFO,
+                         (task_info_t)&teventsinfo, &task_events_count);
+
+               /*
+                * No need to lock "p":  this does not need to be
+                * completely consistent, right ?
+                */
+               p->p_stats->p_ru.ru_minflt = (teventsinfo.faults -
+                                             teventsinfo.pageins);
+               p->p_stats->p_ru.ru_majflt = teventsinfo.pageins;
+               p->p_stats->p_ru.ru_nivcsw = (teventsinfo.csw -
+                                             p->p_stats->p_ru.ru_nvcsw);
+               if (p->p_stats->p_ru.ru_nivcsw < 0)
+                       p->p_stats->p_ru.ru_nivcsw = 0;
+
+               p->p_stats->p_ru.ru_maxrss = tinfo.resident_size_max;
        }
 }
 
-__private_extern__ void munge_rusage(struct rusage *a_rusage_p, struct user_rusage *a_user_rusage_p);
+__private_extern__ void munge_user64_rusage(struct rusage *a_rusage_p, struct user64_rusage *a_user_rusage_p);
+__private_extern__ void munge_user32_rusage(struct rusage *a_rusage_p, struct user32_rusage *a_user_rusage_p);
 
 /* ARGSUSED */
 int
-getrusage(struct proc *p, struct getrusage_args *uap, __unused register_t *retval)
+getrusage(struct proc *p, struct getrusage_args *uap, __unused int32_t *retval)
 {
        struct rusage *rup, rubuf;
-       struct user_rusage rubuf64;
+       struct user64_rusage rubuf64;
+       struct user32_rusage rubuf32;
        size_t retsize = sizeof(rubuf);                 /* default: 32 bits */
        caddr_t retbuf = (caddr_t)&rubuf;               /* default: 32 bits */
        struct timeval utime;
@@ -896,7 +1227,6 @@ getrusage(struct proc *p, struct getrusage_args *uap, __unused register_t *retva
        switch (uap->who) {
        case RUSAGE_SELF:
                calcru(p, &utime, &stime, NULL);
-               // LP64todo: proc struct should have 64 bit version of struct
                proc_lock(p);
                rup = &p->p_stats->p_ru;
                rup->ru_utime = utime;
@@ -920,8 +1250,13 @@ getrusage(struct proc *p, struct getrusage_args *uap, __unused register_t *retva
        if (IS_64BIT_PROCESS(p)) {
                retsize = sizeof(rubuf64);
                retbuf = (caddr_t)&rubuf64;
-               munge_rusage(&rubuf, &rubuf64);
+               munge_user64_rusage(&rubuf, &rubuf64);
+       } else {
+               retsize = sizeof(rubuf32);
+               retbuf = (caddr_t)&rubuf32;
+               munge_user32_rusage(&rubuf, &rubuf32);
        }
+
        return (copyout(retbuf, uap->rusage, retsize));
 }
 
@@ -929,7 +1264,7 @@ void
 ruadd(struct rusage *ru, struct rusage *ru2)
 {
        long *ip, *ip2;
-       int i;
+       long i;
 
        timeradd(&ru->ru_utime, &ru2->ru_utime, &ru->ru_utime);
        timeradd(&ru->ru_stime, &ru2->ru_stime, &ru->ru_stime);
@@ -940,6 +1275,31 @@ ruadd(struct rusage *ru, struct rusage *ru2)
                *ip++ += *ip2++;
 }
 
+/*
+ * Add the rusage stats of child in parent.
+ * 
+ * It adds rusage statistics of child process and statistics of all its
+ * children to its parent.
+ *
+ * Note: proc lock of parent should be held while calling this function.
+ */
+void
+update_rusage_info_child(struct rusage_info_child *ri, rusage_info_current *ri_current)
+{
+       ri->ri_child_user_time += (ri_current->ri_user_time +
+                                       ri_current->ri_child_user_time);
+       ri->ri_child_system_time += (ri_current->ri_system_time +
+                                       ri_current->ri_child_system_time);
+       ri->ri_child_pkg_idle_wkups += (ri_current->ri_pkg_idle_wkups +
+                                       ri_current->ri_child_pkg_idle_wkups);
+       ri->ri_child_interrupt_wkups += (ri_current->ri_interrupt_wkups +
+                                       ri_current->ri_child_interrupt_wkups);
+       ri->ri_child_pageins += (ri_current->ri_pageins +
+                                       ri_current->ri_child_pageins);
+       ri->ri_child_elapsed_abstime += ((ri_current->ri_proc_exit_abstime -
+               ri_current->ri_proc_start_abstime) + ri_current->ri_child_elapsed_abstime);
+}
+
 void
 proc_limitget(proc_t p, int which, struct rlimit * limp)
 {
@@ -1045,7 +1405,6 @@ proc_limitreplace(proc_t p)
        return(0);
 }
 
-
 /*
  * iopolicysys
  *
@@ -1058,77 +1417,541 @@ proc_limitreplace(proc_t p)
  *             EINVAL                          Invalid command or invalid policy arguments
  *
  */
+
+static int
+iopolicysys_disk(struct proc *p, int cmd, int scope, int policy, struct _iopol_param_t *iop_param);
+static int
+iopolicysys_vfs(struct proc *p, int cmd, int scope, int policy, struct _iopol_param_t *iop_param);
+
 int
-iopolicysys(__unused struct proc *p, __unused struct iopolicysys_args *uap, __unused register_t *retval)
+iopolicysys(struct proc *p, struct iopolicysys_args *uap, int32_t *retval)
 {
-       int     error = 0;
-       thread_t thread = THREAD_NULL;
-       int *policy;
-       struct uthread  *ut = NULL;
+       int     error = 0;
        struct _iopol_param_t iop_param;
 
        if ((error = copyin(uap->arg, &iop_param, sizeof(iop_param))) != 0)
-               goto exit;
+               goto out;
 
-       if (iop_param.iop_iotype != IOPOL_TYPE_DISK) {
-               error = EINVAL;
-               goto exit;
+       switch (iop_param.iop_iotype) {
+               case IOPOL_TYPE_DISK:
+                       error = iopolicysys_disk(p, uap->cmd, iop_param.iop_scope, iop_param.iop_policy, &iop_param);
+                       if (error == EIDRM) {
+                               *retval = -2;
+                               error = 0;
+                       }
+                       if (error)
+                               goto out;
+                       break;
+               case IOPOL_TYPE_VFS_HFS_CASE_SENSITIVITY:
+                       error = iopolicysys_vfs(p, uap->cmd, iop_param.iop_scope, iop_param.iop_policy, &iop_param);
+                       if (error)
+                               goto out;
+                       break;
+               default:
+                       error = EINVAL;
+                       goto out;
+       }
+
+       /* Individual iotype handlers are expected to update iop_param, if requested with a GET command */
+       if (uap->cmd == IOPOL_CMD_GET) {
+               error = copyout((caddr_t)&iop_param, uap->arg, sizeof(iop_param));
+               if (error)
+                       goto out;
+       }
+
+out:
+       return (error);
+}
+
+static int
+iopolicysys_disk(struct proc *p __unused, int cmd, int scope, int policy, struct _iopol_param_t *iop_param)
+{
+       int                     error = 0;
+       thread_t        thread;
+       int                     policy_flavor;
+
+       /* Validate scope */
+       switch (scope) {
+               case IOPOL_SCOPE_PROCESS:
+                       thread = THREAD_NULL;
+                       policy_flavor = TASK_POLICY_IOPOL;
+                       break;
+
+               case IOPOL_SCOPE_THREAD:
+                       thread = current_thread();
+                       policy_flavor = TASK_POLICY_IOPOL;
+
+                       /* Not allowed to combine QoS and (non-PASSIVE) IO policy, doing so strips the QoS */
+                       if (cmd == IOPOL_CMD_SET && thread_has_qos_policy(thread)) {
+                               switch (policy) {
+                                       case IOPOL_DEFAULT:
+                                       case IOPOL_PASSIVE:
+                                               break;
+                                       case IOPOL_UTILITY:
+                                       case IOPOL_THROTTLE:
+                                       case IOPOL_IMPORTANT:
+                                       case IOPOL_STANDARD:
+                                               if (!thread_is_static_param(thread)) {
+                                                       thread_remove_qos_policy(thread);
+                                                       /*
+                                                        * This is not an error case, this is to return a marker to user-space that
+                                                        * we stripped the thread of its QoS class.
+                                                        */
+                                                       error = EIDRM;
+                                                       break;
+                                               }
+                                               /* otherwise, fall through to the error case. */
+                                       default:
+                                               error = EINVAL;
+                                               goto out;
+                               }
+                       }
+                       break;
+
+               case IOPOL_SCOPE_DARWIN_BG:
+                       thread = THREAD_NULL;
+                       policy_flavor = TASK_POLICY_DARWIN_BG_IOPOL;
+                       break;
+
+               default:
+                       error = EINVAL;
+                       goto out;
+       }
+
+       /* Validate policy */
+       if (cmd == IOPOL_CMD_SET) {
+               switch (policy) {
+                       case IOPOL_DEFAULT:
+                               if (scope == IOPOL_SCOPE_DARWIN_BG) {
+                                       /* the current default BG throttle level is UTILITY */
+                                       policy = IOPOL_UTILITY;
+                               } else {
+                                       policy = IOPOL_IMPORTANT;
+                               }
+                               break;
+                       case IOPOL_UTILITY:
+                               /* fall-through */
+                       case IOPOL_THROTTLE:
+                               /* These levels are OK */
+                               break;
+                       case IOPOL_IMPORTANT:
+                               /* fall-through */
+                       case IOPOL_STANDARD:
+                               /* fall-through */
+                       case IOPOL_PASSIVE:
+                               if (scope == IOPOL_SCOPE_DARWIN_BG) {
+                                       /* These levels are invalid for BG */
+                                       error = EINVAL;
+                                       goto out;
+                               } else {
+                                       /* OK for other scopes */
+                               }
+                               break;
+                       default:
+                               error = EINVAL;
+                               goto out;
+               }
+       }
+
+       /* Perform command */
+       switch(cmd) {
+               case IOPOL_CMD_SET:
+                       if (thread != THREAD_NULL)
+                               proc_set_thread_policy(thread, TASK_POLICY_INTERNAL, policy_flavor, policy);
+                       else
+                               proc_set_task_policy(current_task(), TASK_POLICY_INTERNAL, policy_flavor, policy);
+                       break;
+               case IOPOL_CMD_GET:
+                       if (thread != THREAD_NULL)
+                               policy = proc_get_thread_policy(thread, TASK_POLICY_INTERNAL, policy_flavor);
+                       else
+                               policy = proc_get_task_policy(current_task(), TASK_POLICY_INTERNAL, policy_flavor);
+                       iop_param->iop_policy = policy;
+                       break;
+               default:
+                       error = EINVAL; /* unknown command */
+                       break;
+       }
+
+out:
+       return (error);
+}
+
+static int
+iopolicysys_vfs(struct proc *p, int cmd, int scope, int policy, struct _iopol_param_t *iop_param)
+{
+       int                     error = 0;
+
+       /* Validate scope */
+       switch (scope) {
+               case IOPOL_SCOPE_PROCESS:
+                       /* Only process OK */
+                       break;
+               default:
+                       error = EINVAL;
+                       goto out;
+       }
+
+       /* Validate policy */
+       if (cmd == IOPOL_CMD_SET) {
+               switch (policy) {
+                       case IOPOL_VFS_HFS_CASE_SENSITIVITY_DEFAULT:
+                               /* fall-through */
+                       case IOPOL_VFS_HFS_CASE_SENSITIVITY_FORCE_CASE_SENSITIVE:
+                               /* These policies are OK */
+                               break;
+                       default:
+                               error = EINVAL;
+                               goto out;
+               }
+       }
+
+       /* Perform command */
+       switch(cmd) {
+               case IOPOL_CMD_SET:
+                       if (0 == kauth_cred_issuser(kauth_cred_get())) {
+                               /* If it's a non-root process, it needs to have the entitlement to set the policy */
+                               boolean_t entitled = FALSE;
+                               entitled = IOTaskHasEntitlement(current_task(), "com.apple.private.iopol.case_sensitivity");
+                               if (!entitled) {
+                                       error = EPERM;
+                                       goto out;
+                               }
+                       }
+
+                       switch (policy) {
+                               case IOPOL_VFS_HFS_CASE_SENSITIVITY_DEFAULT:
+                                       OSBitAndAtomic16(~((uint32_t)P_VFS_IOPOLICY_FORCE_HFS_CASE_SENSITIVITY), &p->p_vfs_iopolicy);
+                                       break;
+                               case IOPOL_VFS_HFS_CASE_SENSITIVITY_FORCE_CASE_SENSITIVE:
+                                       OSBitOrAtomic16((uint32_t)P_VFS_IOPOLICY_FORCE_HFS_CASE_SENSITIVITY, &p->p_vfs_iopolicy);
+                                       break;
+                               default:
+                                       error = EINVAL;
+                                       goto out;
+                       }
+                       
+                       break;
+               case IOPOL_CMD_GET:
+                       iop_param->iop_policy = (p->p_vfs_iopolicy & P_VFS_IOPOLICY_FORCE_HFS_CASE_SENSITIVITY)
+                               ? IOPOL_VFS_HFS_CASE_SENSITIVITY_FORCE_CASE_SENSITIVE
+                               : IOPOL_VFS_HFS_CASE_SENSITIVITY_DEFAULT;
+                       break;
+               default:
+                       error = EINVAL; /* unknown command */
+                       break;
+       }
+
+out:
+       return (error);
+}
+
+/* BSD call back function for task_policy networking changes */
+void
+proc_apply_task_networkbg(void * bsd_info, thread_t thread)
+{
+       assert(bsd_info != PROC_NULL);
+
+       pid_t pid = proc_pid((proc_t)bsd_info);
+
+       proc_t p = proc_find(pid);
+
+       if (p != PROC_NULL) {
+               assert(p == (proc_t)bsd_info);
+
+               do_background_socket(p, thread);
+               proc_rele(p);
+       }
+}
+
+void
+gather_rusage_info(proc_t p, rusage_info_current *ru, int flavor)
+{
+       struct rusage_info_child *ri_child;
+
+       assert(p->p_stats != NULL);
+       switch(flavor) {
+
+       case RUSAGE_INFO_V3:
+               fill_task_qos_rusage(p->task, ru);
+               fill_task_billed_usage(p->task, ru);
+               /* fall through */
+
+       case RUSAGE_INFO_V2:
+               fill_task_io_rusage(p->task, ru);
+               /* fall through */
+
+       case RUSAGE_INFO_V1:
+               /*
+                * p->p_stats->ri_child statistics are protected under proc lock.
+                */
+               proc_lock(p);
+               
+               ri_child = &(p->p_stats->ri_child);
+               ru->ri_child_user_time = ri_child->ri_child_user_time;
+               ru->ri_child_system_time = ri_child->ri_child_system_time;
+               ru->ri_child_pkg_idle_wkups = ri_child->ri_child_pkg_idle_wkups;
+               ru->ri_child_interrupt_wkups = ri_child->ri_child_interrupt_wkups;
+               ru->ri_child_pageins = ri_child->ri_child_pageins;
+               ru->ri_child_elapsed_abstime = ri_child->ri_child_elapsed_abstime;
+
+               proc_unlock(p);
+               /* fall through */
+
+       case RUSAGE_INFO_V0:
+               proc_getexecutableuuid(p, (unsigned char *)&ru->ri_uuid, sizeof (ru->ri_uuid));
+               fill_task_rusage(p->task, ru);
+               ru->ri_proc_start_abstime = p->p_stats->ps_start;
+       }
+}
+
+static void
+rusage_info_conversion(rusage_info_t ri_info, rusage_info_current *ri_current, int flavor)
+{
+       struct rusage_info_v0 *ri_v0;
+       struct rusage_info_v1 *ri_v1;
+       struct rusage_info_v2 *ri_v2;
+
+       switch (flavor) {
+
+       case RUSAGE_INFO_V2:
+               ri_v2 = (struct rusage_info_v2 *)ri_info;
+               ri_v2->ri_diskio_bytesread = ri_current->ri_diskio_bytesread;
+               ri_v2->ri_diskio_byteswritten = ri_current->ri_diskio_byteswritten;
+               /* fall through */
+
+       case RUSAGE_INFO_V1:
+               ri_v1 = (struct rusage_info_v1 *)ri_info;
+               ri_v1->ri_child_user_time = ri_current->ri_child_user_time;
+               ri_v1->ri_child_system_time = ri_current->ri_child_system_time;
+               ri_v1->ri_child_pkg_idle_wkups = ri_current->ri_child_pkg_idle_wkups;
+               ri_v1->ri_child_interrupt_wkups = ri_current->ri_child_interrupt_wkups; 
+               ri_v1->ri_child_pageins = ri_current->ri_child_pageins;
+               ri_v1->ri_child_elapsed_abstime = ri_current->ri_child_elapsed_abstime;
+               /* fall through */
+
+       case RUSAGE_INFO_V0:
+               ri_v0 = (struct rusage_info_v0 *)ri_info;
+               memcpy(&ri_v0->ri_uuid[0], &ri_current->ri_uuid[0], sizeof(ri_v0->ri_uuid));    
+               ri_v0->ri_user_time = ri_current->ri_user_time;
+               ri_v0->ri_system_time = ri_current->ri_system_time;
+               ri_v0->ri_pkg_idle_wkups = ri_current->ri_pkg_idle_wkups;
+               ri_v0->ri_interrupt_wkups = ri_current->ri_interrupt_wkups;
+               ri_v0->ri_pageins = ri_current->ri_pageins;
+               ri_v0->ri_wired_size = ri_current->ri_wired_size;
+               ri_v0->ri_resident_size = ri_current->ri_resident_size;
+               ri_v0->ri_phys_footprint = ri_current->ri_phys_footprint;
+               ri_v0->ri_proc_start_abstime = ri_current->ri_proc_start_abstime;
+               ri_v0->ri_proc_exit_abstime = ri_current->ri_proc_exit_abstime;
+
+               break;
+       
+       default:
+               break;
        }
+}
+
+
+int
+proc_get_rusage(proc_t p, int flavor, user_addr_t buffer, __unused int is_zombie)
+{
+       struct rusage_info_v0 ri_v0;
+       struct rusage_info_v1 ri_v1;
+       struct rusage_info_v2 ri_v2;
+       struct rusage_info_v3 ri_v3;
+
+       rusage_info_current ri_current;
+
+       int error = 0;
 
-       switch (iop_param.iop_scope) {
-       case IOPOL_SCOPE_PROCESS:
-               policy = &p->p_iopol_disk;
+       switch (flavor) {
+       case RUSAGE_INFO_V0:
+               /*
+                * If task is still alive, collect info from the live task itself.
+                * Otherwise, look to the cached info in the zombie proc.
+                */
+               if (p->p_ru == NULL) {
+                       gather_rusage_info(p, &ri_current, flavor);
+                       ri_current.ri_proc_exit_abstime = 0;
+                       rusage_info_conversion(&ri_v0, &ri_current, flavor);
+               } else {
+                       rusage_info_conversion(&ri_v0, &p->p_ru->ri, flavor);
+               }
+               error = copyout(&ri_v0, buffer, sizeof (ri_v0));
                break;
-       case IOPOL_SCOPE_THREAD:
-               thread = current_thread();
-               ut = get_bsdthread_info(thread);
-               policy = &ut->uu_iopol_disk;
+
+       case RUSAGE_INFO_V1:
+               /*
+                * If task is still alive, collect info from the live task itself.
+                * Otherwise, look to the cached info in the zombie proc.
+                */
+               if (p->p_ru == NULL) {
+                       gather_rusage_info(p, &ri_current, flavor);
+                       ri_current.ri_proc_exit_abstime = 0;
+                       rusage_info_conversion(&ri_v1, &ri_current, flavor);
+               } else {
+                       rusage_info_conversion(&ri_v1, &p->p_ru->ri, flavor);
+               }
+               error = copyout(&ri_v1, buffer, sizeof (ri_v1));
+               break;
+
+       case RUSAGE_INFO_V2:
+               /*
+                * If task is still alive, collect info from the live task itself.
+                * Otherwise, look to the cached info in the zombie proc.
+                */
+               if (p->p_ru == NULL) {
+                       gather_rusage_info(p, &ri_current, flavor);
+                       ri_current.ri_proc_exit_abstime = 0;
+                       rusage_info_conversion(&ri_v2, &ri_current, flavor);
+               } else {
+                       rusage_info_conversion(&ri_v2, &p->p_ru->ri, flavor);
+               }
+               error = copyout(&ri_v2, buffer, sizeof (ri_v2));
+               break;
+
+       case RUSAGE_INFO_V3:
+               /*
+                * If task is still alive, collect info from the live task itself.
+                * Otherwise, look to the cached info in the zombie proc.
+                */
+               if (p->p_ru == NULL) {
+                       gather_rusage_info(p, &ri_v3, flavor);
+                       ri_v3.ri_proc_exit_abstime = 0;
+               } else {
+                       ri_v3 = p->p_ru->ri;
+               }
+               error = copyout(&ri_v3, buffer, sizeof (ri_v3));
                break;
+
        default:
                error = EINVAL;
-               goto exit;
+               break;
        }
-               
-       switch(uap->cmd) {
-       case IOPOL_CMD_SET:
-               switch (iop_param.iop_policy) {
-               case IOPOL_DEFAULT:
-               case IOPOL_NORMAL:
-               case IOPOL_THROTTLE:
-               case IOPOL_PASSIVE:
-                       proc_lock(p);
-                       *policy = iop_param.iop_policy;
-                       proc_unlock(p);
+
+       return (error); 
+}
+
+static int
+mach_to_bsd_rv(int mach_rv)
+{
+       int bsd_rv = 0;
+
+       switch (mach_rv) {
+       case KERN_SUCCESS:
+               bsd_rv = 0;
+               break;
+       case KERN_INVALID_ARGUMENT:
+               bsd_rv = EINVAL;
+               break;
+       default:
+               panic("unknown error %#x", mach_rv);
+       }
+
+       return bsd_rv;
+}
+
+/*
+ * Resource limit controls
+ *
+ * uap->flavor available flavors:
+ *
+ *     RLIMIT_WAKEUPS_MONITOR
+ */
+int
+proc_rlimit_control(__unused struct proc *p, struct proc_rlimit_control_args *uap, __unused int32_t *retval)
+{
+       proc_t  targetp;
+       int     error = 0;
+       struct  proc_rlimit_control_wakeupmon wakeupmon_args;
+       uint32_t cpumon_flags;
+       uint32_t cpulimits_flags;
+       kauth_cred_t my_cred, target_cred;
+
+       /* -1 implicitly means our own process (perhaps even the current thread for per-thread attributes) */
+       if (uap->pid == -1) {
+               targetp = proc_self();
+       } else {
+               targetp = proc_find(uap->pid);
+       }
+
+       /* proc_self() can return NULL for an exiting process */
+       if (targetp == PROC_NULL) {
+               return (ESRCH);
+       }
+
+       my_cred = kauth_cred_get();
+       target_cred = kauth_cred_proc_ref(targetp);
+
+       if (!kauth_cred_issuser(my_cred) && kauth_cred_getruid(my_cred) &&
+           kauth_cred_getuid(my_cred) != kauth_cred_getuid(target_cred) &&
+           kauth_cred_getruid(my_cred) != kauth_cred_getuid(target_cred)) {
+               proc_rele(targetp);
+               kauth_cred_unref(&target_cred);
+               return (EACCES);
+       }
+
+       switch (uap->flavor) {
+       case RLIMIT_WAKEUPS_MONITOR:
+               if ((error = copyin(uap->arg, &wakeupmon_args, sizeof (wakeupmon_args))) != 0) {
+                       break;
+               }
+               if ((error = mach_to_bsd_rv(task_wakeups_monitor_ctl(targetp->task, &wakeupmon_args.wm_flags,
+                    &wakeupmon_args.wm_rate))) != 0) {
                        break;
-               default:
-                       error = EINVAL;
-                       goto exit;
                }
+               error = copyout(&wakeupmon_args, uap->arg, sizeof (wakeupmon_args));
+               break;
+       case RLIMIT_CPU_USAGE_MONITOR:
+               cpumon_flags = uap->arg; // XXX temporarily stashing flags in argp (12592127)
+               error = mach_to_bsd_rv(task_cpu_usage_monitor_ctl(targetp->task, &cpumon_flags));
                break;
-       case IOPOL_CMD_GET:
-               switch (*policy) {
-               case IOPOL_DEFAULT:
-               case IOPOL_NORMAL:
-               case IOPOL_THROTTLE:
-               case IOPOL_PASSIVE:
-                       iop_param.iop_policy = *policy;
+       case RLIMIT_THREAD_CPULIMITS:
+               cpulimits_flags = (uint32_t)uap->arg; // only need a limited set of bits, pass in void * argument
+
+               if (uap->pid != -1) {
+                       error = EINVAL;
                        break;
-               default: // in-kernel 
-                       // this should never happen
-                       printf("%s: unknown I/O policy %d\n", __func__, *policy);
-                       // restore to default value
-                       *policy = IOPOL_DEFAULT;
-                       iop_param.iop_policy = *policy;
                }
-               
-               error = copyout((caddr_t)&iop_param, uap->arg, sizeof(iop_param));
+
+               uint8_t percent = 0;
+               uint32_t ms_refill = 0;
+               uint64_t ns_refill;
+
+               percent = (uint8_t)(cpulimits_flags & 0xffU);                                   /* low 8 bits for percent */
+               ms_refill = (cpulimits_flags >> 8) & 0xffffff;          /* next 24 bits represent ms refill value */
+               if (percent >= 100) {
+                       error = EINVAL;
+                       break;
+               }
+
+               ns_refill = ((uint64_t)ms_refill) * NSEC_PER_MSEC;
+
+               error = mach_to_bsd_rv(thread_set_cpulimit(THREAD_CPULIMIT_BLOCK, percent, ns_refill));
                break;
        default:
-               error = EINVAL; // unknown command
+               error = EINVAL;
                break;
        }
 
-  exit:
-       *retval = error;
+       proc_rele(targetp);
+       kauth_cred_unref(&target_cred);
+
+       /*
+        * Return value from this function becomes errno to userland caller.
+        */
        return (error);
 }
+
+/*
+ * Return the current amount of CPU consumed by this thread (in either user or kernel mode)
+ */
+int thread_selfusage(struct proc *p __unused, struct thread_selfusage_args *uap __unused, uint64_t *retval)
+{
+       uint64_t runtime;
+
+       runtime = thread_get_runtime_self();
+       *retval = runtime;
+
+       return (0);
+}