X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/55e303ae13a4cf49d70f2294092726f2fffb9ef2..7ddcb079202367355dddccdfa4318e57d50318be:/bsd/kern/bsd_init.c diff --git a/bsd/kern/bsd_init.c b/bsd/kern/bsd_init.c index cc54ea10b..dac2c94d8 100644 --- a/bsd/kern/bsd_init.c +++ b/bsd/kern/bsd_init.c @@ -1,16 +1,19 @@ /* - * Copyright (c) 2000-2003 Apple Computer, Inc. All rights reserved. + * Copyright (c) 2000-2009 Apple Inc. All rights reserved. * - * @APPLE_LICENSE_HEADER_START@ - * - * Copyright (c) 1999-2003 Apple Computer, Inc. All Rights Reserved. + * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * * This file contains Original Code and/or Modifications of Original Code * as defined in and that are subject to the Apple Public Source License * Version 2.0 (the 'License'). You may not use this file except in - * compliance with the License. Please obtain a copy of the License at - * http://www.opensource.apple.com/apsl/ and read it before using this - * file. + * compliance with the License. The rights granted to you under the License + * may not be used to create, or enable the creation or redistribution of, + * unlawful or unlicensed copies of an Apple operating system, or to + * circumvent, violate, or enable the circumvention or violation of, any + * terms of an Apple operating system software license agreement. + * + * Please obtain a copy of the License at + * http://www.opensource.apple.com/apsl/ and read it before using this file. * * The Original Code and all software distributed under the License are * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER @@ -20,9 +23,9 @@ * Please see the License for the specific language governing rights and * limitations under the License. * - * @APPLE_LICENSE_HEADER_END@ - */ -/* + * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ + * + * * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993 * The Regents of the University of California. All rights reserved. * (c) UNIX System Laboratories, Inc. @@ -69,30 +72,41 @@ * All rights reserved. The CMU software License Agreement specifies * the terms and conditions for use and redistribution. */ +/* + * NOTICE: This file was modified by McAfee Research in 2004 to introduce + * support for mandatory and extensible security protections. This notice + * is included in support of clause 2.2 (b) of the Apple Public License, + * Version 2.0. + */ #include #include #include -#include -#include +#include +#include +#include #include -#include +#include #include -#include +#include #include #include #include #include +#include -#include +#include #include #include +#include -#include +#include #include #include #include +#include +#include #include @@ -112,35 +126,89 @@ #include #include #include +#include /* for thread_resume() */ +#include /* for task_set_exception_ports() */ +#include /* for ux_handler() */ +#include /* for ubc_init() */ +#include /* for mcache_init() */ +#include /* for mbinit() */ +#include /* for knote_init() */ +#include /* for kern_memorystatus_init() */ +#include /* for aio_init() */ +#include /* for psem_cache_init() */ +#include /* for dlil_init() */ +#include /* for proto_kpi_init() */ +#include /* for pipeinit() */ +#include /* for socketinit() */ +#include /* for domaininit() */ +#include /* for thread_wakeup() */ +#include /* for ether_family_init() */ +#include /* for vnode_pager_bootstrap() */ +#include /* for devfs_kernel_mount() */ +#include /* for host_set_exception_ports() */ +#include /* for host_priv_self() */ +#include /* for kmem_suballoc() */ +#include /* for psem_lock_init() */ +#include /* for log_setsize() */ +#include /* for tty_init() */ +#include /* for utun_register_control() */ +#include /* for net_str_id_init() */ +#include /* for netsrc_init() */ +#include /* for assert() */ + +#include + +#if CONFIG_MACF +#include +#include /* mac_init_bsd() */ +#include /* mac_update_task_label() */ +#endif + +#include -#include -#include +#if NFSCLIENT +#include +#endif -extern int app_profile; /* on/off switch for pre-heat cache */ +#if CONFIG_IMAGEBOOT +#include +#endif -char copyright[] = +#if PFLOG +#include +#endif + +#include +#include + +void * get_user_regs(thread_t); /* XXX kludge for */ +void IOKitInitializeTime(void); /* XXX */ +void IOSleep(unsigned int); /* XXX */ +void loopattach(void); /* XXX */ +void vc_progress_set(boolean_t, uint32_t); /* XXX */ + +const char copyright[] = "Copyright (c) 1982, 1986, 1989, 1991, 1993\n\t" "The Regents of the University of California. " "All rights reserved.\n\n"; -extern void ux_handler(); - /* Components of the first process -- never freed. */ struct proc proc0; struct session session0; struct pgrp pgrp0; -struct pcred cred0; struct filedesc filedesc0; struct plimit limit0; struct pstats pstats0; struct sigacts sigacts0; -struct proc *kernproc, *initproc; +proc_t kernproc; +proc_t initproc; long tk_cancc; long tk_nin; long tk_nout; long tk_rawcc; +int lock_trace = 0; /* Global variables to make pstat happy. We do swapping differently */ int nswdev, nswap; int nswapmap; @@ -155,14 +223,17 @@ char hostname[MAXHOSTNAMELEN]; int hostnamelen; char domainname[MAXDOMNAMELEN]; int domainnamelen; -char classichandler[32] = {0}; -long classichandler_fsid = -1L; -long classichandler_fileid = -1L; +#if defined(__i386__) || defined(__x86_64__) +struct exec_archhandler exec_archhandler_ppc = { + .path = "/usr/libexec/oah/RosettaNonGrata", +}; +#else /* __i386__ */ +struct exec_archhandler exec_archhandler_ppc; +#endif /* __i386__ */ char rootdevice[16]; /* hfs device names have at least 9 chars */ -struct timeval boottime; /* GRODY! This has to go... */ -#ifdef KMEMSTATS +#if KMEMSTATS struct kmemstats kmemstats[M_LAST]; #endif @@ -170,23 +241,95 @@ int lbolt; /* awoken once a second */ struct vnode *rootvp; int boothowto = RB_DEBUG; -#define BSD_PAGABLE_MAP_SIZE (16 * 512 * 1024) +void lightning_bolt(void *); +extern kern_return_t IOFindBSDRoot(char *, unsigned int, dev_t *, u_int32_t *); +extern void IOSecureBSDRoot(const char * rootName); +extern kern_return_t IOKitBSDInit(void ); +extern void kminit(void); +extern void klogwakeup(void); +extern void file_lock_init(void); +extern void kmeminit(void); +extern void bsd_bufferinit(void); + +extern int serverperfmode; +extern int ncl; + vm_map_t bsd_pageable_map; vm_map_t mb_map; -semaphore_t execve_semaphore; + +static int bsd_simul_execs; +static int bsd_pageable_map_size; +__private_extern__ int execargs_cache_size = 0; +__private_extern__ int execargs_free_count = 0; +__private_extern__ vm_offset_t * execargs_cache = NULL; + +void bsd_exec_setup(int); + +/* + * Set to disable grading 64 bit Mach-o binaries as executable, for testing; + * Intel only. + */ +__private_extern__ int bootarg_no64exec = 0; +__private_extern__ int bootarg_vnode_cache_defeat = 0; + +/* + * Prevent kernel-based ASLR from being used, for testing. + */ +#if DEVELOPMENT || DEBUG +__private_extern__ int bootarg_disable_aslr = 0; +#endif int cmask = CMASK; +extern int customnbuf; + +void bsd_init(void) __attribute__((section("__TEXT, initcode"))); +kern_return_t bsd_autoconf(void) __attribute__((section("__TEXT, initcode"))); +void bsd_utaskbootstrap(void) __attribute__((section("__TEXT, initcode"))); -int parse_bsd_args(void); -extern int bsd_hardclockinit; +static void parse_bsd_args(void); extern task_t bsd_init_task; extern char init_task_failure_data[]; extern void time_zone_slock_init(void); +extern void select_wait_queue_init(void); +static void process_name(const char *, proc_t); + +static void setconf(void); funnel_t *kernel_flock; -funnel_t *network_flock; -int disable_funnel = 0; /* disables split funnel */ -int enable_funnel = 0; /* disables split funnel */ + +#if SYSV_SHM +extern void sysv_shm_lock_init(void); +#endif +#if SYSV_SEM +extern void sysv_sem_lock_init(void); +#endif +#if SYSV_MSG +extern void sysv_msg_lock_init(void); +#endif + +#if !defined(SECURE_KERNEL) +/* kmem access not enabled by default; can be changed with boot-args */ +/* We don't need to keep this symbol around in RELEASE kernel */ +int setup_kmem = 0; +#endif + +#if CONFIG_MACF +#if defined (__i386__) || defined (__x86_64__) +/* MACF policy_check configuration flags; see policy_check.c for details */ +int policy_check_flags = 0; + +extern int check_policy_init(int); +#endif +#endif /* CONFIG_MACF */ + +extern void stackshot_lock_init(void); + + +/* If we are using CONFIG_DTRACE */ +#if CONFIG_DTRACE + extern void dtrace_postinit(void); +#endif + /* * Initialization code. @@ -203,12 +346,10 @@ int enable_funnel = 0; /* disables split funnel */ /* * Sets the name for the given task. */ -void -proc_name(s, p) - char *s; - struct proc *p; +static void +process_name(const char *s, proc_t p) { - int length = strlen(s); + size_t length = strlen(s); bcopy(s, p->p_comm, length >= sizeof(p->p_comm) ? sizeof(p->p_comm) : @@ -217,126 +358,258 @@ proc_name(s, p) /* To allow these values to be patched, they're globals here */ #include -struct rlimit vm_initial_limit_stack = { DFLSSIZ, MAXSSIZ }; +struct rlimit vm_initial_limit_stack = { DFLSSIZ, MAXSSIZ - PAGE_SIZE }; struct rlimit vm_initial_limit_data = { DFLDSIZ, MAXDSIZ }; struct rlimit vm_initial_limit_core = { DFLCSIZ, MAXCSIZ }; -extern thread_t first_thread; -extern thread_act_t cloneproc(struct proc *, int); -extern int (*mountroot) __P((void)); -extern int netboot_mountroot(); /* netboot.c */ -extern int netboot_setup(struct proc * p); +extern thread_t cloneproc(task_t, proc_t, int); +extern int (*mountroot)(void); + +lck_grp_t * proc_lck_grp; +lck_grp_t * proc_slock_grp; +lck_grp_t * proc_fdmlock_grp; +lck_grp_t * proc_mlock_grp; +lck_grp_attr_t * proc_lck_grp_attr; +lck_attr_t * proc_lck_attr; +lck_mtx_t * proc_list_mlock; +lck_mtx_t * proc_klist_mlock; + +extern lck_mtx_t * execargs_cache_lock; /* hook called after root is mounted XXX temporary hack */ void (*mountroot_post_hook)(void); +void (*unmountroot_pre_hook)(void); +/* + * This function is called very early on in the Mach startup, from the + * function start_kernel_threads() in osfmk/kern/startup.c. It's called + * in the context of the current (startup) task using a call to the + * function kernel_thread_create() to jump into start_kernel_threads(). + * Internally, kernel_thread_create() calls thread_create_internal(), + * which calls uthread_alloc(). The function of uthread_alloc() is + * normally to allocate a uthread structure, and fill out the uu_sigmask, + * uu_context fields. It skips filling these out in the case of the "task" + * being "kernel_task", because the order of operation is inverted. To + * account for that, we need to manually fill in at least the contents + * of the uu_context.vc_ucred field so that the uthread structure can be + * used like any other. + */ void -bsd_init() +bsd_init(void) { - register struct proc *p; - extern struct ucred *rootcred; - register int i; - int s; - thread_t th; - void lightning_bolt(void ); + struct uthread *ut; + unsigned int i; +#if __i386__ || __x86_64__ + int error; +#endif + struct vfs_context context; kern_return_t ret; - boolean_t funnel_state; - extern void uthread_zone_init(); - + struct ucred temp_cred; + struct posix_cred temp_pcred; +#if NFSCLIENT || CONFIG_IMAGEBOOT + boolean_t netboot = FALSE; +#endif - /* split funnel is enabled by default */ - PE_parse_boot_arg("dfnl", &disable_funnel); +#define bsd_init_kprintf(x...) /* kprintf("bsd_init: " x) */ kernel_flock = funnel_alloc(KERNEL_FUNNEL); if (kernel_flock == (funnel_t *)0 ) { panic("bsd_init: Failed to allocate kernel funnel"); } - funnel_state = thread_funnel_set(kernel_flock, TRUE); - - if (!disable_funnel) { - network_flock = funnel_alloc(NETWORK_FUNNEL); - if (network_flock == (funnel_t *)0 ) { - panic("bsd_init: Failed to allocate network funnel"); - } - } else { - network_flock = kernel_flock; - } - printf(copyright); - + + bsd_init_kprintf("calling kmeminit\n"); kmeminit(); + bsd_init_kprintf("calling parse_bsd_args\n"); parse_bsd_args(); - bsd_bufferinit(); - - /* Initialize the uthread zone */ - uthread_zone_init(); + /* Initialize kauth subsystem before instancing the first credential */ + bsd_init_kprintf("calling kauth_init\n"); + kauth_init(); - /* - * Initialize process and pgrp structures. - */ + /* Initialize process and pgrp structures. */ + bsd_init_kprintf("calling procinit\n"); procinit(); - kernproc = &proc0; + /* Initialize the ttys (MUST be before kminit()/bsd_autoconf()!)*/ + tty_init(); - p = kernproc; + kernproc = &proc0; /* implicitly bzero'ed */ /* kernel_task->proc = kernproc; */ - set_bsdtask_info(kernel_task,(void *)p); - p->p_pid = 0; + set_bsdtask_info(kernel_task,(void *)kernproc); /* give kernproc a name */ - proc_name("kernel_task", p); + bsd_init_kprintf("calling process_name\n"); + process_name("kernel_task", kernproc); + + /* allocate proc lock group attribute and group */ + bsd_init_kprintf("calling lck_grp_attr_alloc_init\n"); + proc_lck_grp_attr= lck_grp_attr_alloc_init(); + + proc_lck_grp = lck_grp_alloc_init("proc", proc_lck_grp_attr); +#if CONFIG_FINE_LOCK_GROUPS + proc_slock_grp = lck_grp_alloc_init("proc-slock", proc_lck_grp_attr); + proc_fdmlock_grp = lck_grp_alloc_init("proc-fdmlock", proc_lck_grp_attr); + proc_mlock_grp = lck_grp_alloc_init("proc-mlock", proc_lck_grp_attr); +#endif + /* Allocate proc lock attribute */ + proc_lck_attr = lck_attr_alloc_init(); +#if 0 +#if __PROC_INTERNAL_DEBUG + lck_attr_setdebug(proc_lck_attr); +#endif +#endif + +#if CONFIG_FINE_LOCK_GROUPS + proc_list_mlock = lck_mtx_alloc_init(proc_mlock_grp, proc_lck_attr); + proc_klist_mlock = lck_mtx_alloc_init(proc_mlock_grp, proc_lck_attr); + lck_mtx_init(&kernproc->p_mlock, proc_mlock_grp, proc_lck_attr); + lck_mtx_init(&kernproc->p_fdmlock, proc_fdmlock_grp, proc_lck_attr); + lck_spin_init(&kernproc->p_slock, proc_slock_grp, proc_lck_attr); +#else + proc_list_mlock = lck_mtx_alloc_init(proc_lck_grp, proc_lck_attr); + proc_klist_mlock = lck_mtx_alloc_init(proc_lck_grp, proc_lck_attr); + lck_mtx_init(&kernproc->p_mlock, proc_lck_grp, proc_lck_attr); + lck_mtx_init(&kernproc->p_fdmlock, proc_lck_grp, proc_lck_attr); + lck_spin_init(&kernproc->p_slock, proc_lck_grp, proc_lck_attr); +#endif + assert(bsd_simul_execs != 0); + execargs_cache_lock = lck_mtx_alloc_init(proc_lck_grp, proc_lck_attr); + execargs_cache_size = bsd_simul_execs; + execargs_free_count = bsd_simul_execs; + execargs_cache = (vm_offset_t *)kalloc(bsd_simul_execs * sizeof(vm_offset_t)); + bzero(execargs_cache, bsd_simul_execs * sizeof(vm_offset_t)); + if (current_task() != kernel_task) printf("bsd_init: We have a problem, " "current task is not kernel task\n"); + bsd_init_kprintf("calling get_bsdthread_info\n"); + ut = (uthread_t)get_bsdthread_info(current_thread()); + +#if CONFIG_MACF + /* + * Initialize the MAC Framework + */ + mac_policy_initbsd(); + kernproc->p_mac_enforce = 0; + +#if defined (__i386__) || defined (__x86_64__) + /* + * We currently only support this on i386/x86_64, as that is the + * only lock code we have instrumented so far. + */ + check_policy_init(policy_check_flags); +#endif +#endif /* MAC */ + /* * Create process 0. */ - LIST_INSERT_HEAD(&allproc, p, p_list); - p->p_pgrp = &pgrp0; + proc_list_lock(); + LIST_INSERT_HEAD(&allproc, kernproc, p_list); + kernproc->p_pgrp = &pgrp0; LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); LIST_INIT(&pgrp0.pg_members); - LIST_INSERT_HEAD(&pgrp0.pg_members, p, p_pglist); +#ifdef CONFIG_FINE_LOCK_GROUPS + lck_mtx_init(&pgrp0.pg_mlock, proc_mlock_grp, proc_lck_attr); +#else + lck_mtx_init(&pgrp0.pg_mlock, proc_lck_grp, proc_lck_attr); +#endif + /* There is no other bsd thread this point and is safe without pgrp lock */ + LIST_INSERT_HEAD(&pgrp0.pg_members, kernproc, p_pglist); + kernproc->p_listflag |= P_LIST_INPGRP; + kernproc->p_pgrpid = 0; + kernproc->p_uniqueid = 0; pgrp0.pg_session = &session0; + pgrp0.pg_membercnt = 1; + session0.s_count = 1; - session0.s_leader = p; + session0.s_leader = kernproc; + session0.s_listflags = 0; +#ifdef CONFIG_FINE_LOCK_GROUPS + lck_mtx_init(&session0.s_mlock, proc_mlock_grp, proc_lck_attr); +#else + lck_mtx_init(&session0.s_mlock, proc_lck_grp, proc_lck_attr); +#endif + LIST_INSERT_HEAD(SESSHASH(0), &session0, s_hash); + proc_list_unlock(); - p->task = kernel_task; +#if CONFIG_LCTX + kernproc->p_lctx = NULL; +#endif + + kernproc->task = kernel_task; + + kernproc->p_stat = SRUN; + kernproc->p_flag = P_SYSTEM; + kernproc->p_lflag = 0; + kernproc->p_ladvflag = 0; - p->p_stat = SRUN; - p->p_flag = P_INMEM|P_SYSTEM; - p->p_nice = NZERO; - p->p_pptr = p; - lockinit(&p->signal_lock, PVM, "signal", 0, 0); - TAILQ_INIT(&p->p_uthlist); - p->sigwait = FALSE; - p->sigwait_thread = THREAD_NULL; - p->exit_thread = THREAD_NULL; - - /* Create credentials. */ - lockinit(&cred0.pc_lock, PLOCK, "proc0 cred", 0, 0); - cred0.p_refcnt = 1; - p->p_cred = &cred0; - p->p_ucred = crget(); - p->p_ucred->cr_ngroups = 1; /* group 0 */ +#if DEVELOPMENT || DEBUG + if (bootarg_disable_aslr) + kernproc->p_flag |= P_DISABLE_ASLR; +#endif + + kernproc->p_nice = NZERO; + kernproc->p_pptr = kernproc; + + TAILQ_INIT(&kernproc->p_uthlist); + TAILQ_INSERT_TAIL(&kernproc->p_uthlist, ut, uu_list); - TAILQ_INIT(&p->aio_activeq); - TAILQ_INIT(&p->aio_doneq); - p->aio_active_count = 0; - p->aio_done_count = 0; + kernproc->sigwait = FALSE; + kernproc->sigwait_thread = THREAD_NULL; + kernproc->exit_thread = THREAD_NULL; + kernproc->p_csflags = CS_VALID; - /* Set the audit info for this process */ - audit_proc_init(p); + /* + * Create credential. This also Initializes the audit information. + */ + bsd_init_kprintf("calling bzero\n"); + bzero(&temp_cred, sizeof(temp_cred)); + bzero(&temp_pcred, sizeof(temp_pcred)); + temp_pcred.cr_ngroups = 1; + + temp_cred.cr_audit.as_aia_p = audit_default_aia_p; + + bsd_init_kprintf("calling kauth_cred_create\n"); + /* + * We have to label the temp cred before we create from it to + * properly set cr_ngroups, or the create will fail. + */ + posix_cred_label(&temp_cred, &temp_pcred); + kernproc->p_ucred = kauth_cred_create(&temp_cred); + + /* update cred on proc */ + PROC_UPDATE_CREDS_ONPROC(kernproc); + + /* give the (already exisiting) initial thread a reference on it */ + bsd_init_kprintf("calling kauth_cred_ref\n"); + kauth_cred_ref(kernproc->p_ucred); + ut->uu_context.vc_ucred = kernproc->p_ucred; + ut->uu_context.vc_thread = current_thread(); + + TAILQ_INIT(&kernproc->p_aio_activeq); + TAILQ_INIT(&kernproc->p_aio_doneq); + kernproc->p_aio_total_count = 0; + kernproc->p_aio_active_count = 0; + + bsd_init_kprintf("calling file_lock_init\n"); + file_lock_init(); + +#if CONFIG_MACF + mac_cred_label_associate_kernel(kernproc->p_ucred); + mac_task_label_update_cred (kernproc->p_ucred, (struct task *) kernproc->task); +#endif /* Create the file descriptor table. */ filedesc0.fd_refcnt = 1+1; /* +1 so shutdown will not _FREE_ZONE */ - p->p_fd = &filedesc0; + kernproc->p_fd = &filedesc0; filedesc0.fd_cmask = cmask; filedesc0.fd_knlistsize = -1; filedesc0.fd_knlist = NULL; @@ -344,24 +617,25 @@ bsd_init() filedesc0.fd_knhashmask = 0; /* Create the limits structures. */ - p->p_limit = &limit0; - for (i = 0; i < sizeof(p->p_rlimit)/sizeof(p->p_rlimit[0]); i++) + kernproc->p_limit = &limit0; + for (i = 0; i < sizeof(kernproc->p_rlimit)/sizeof(kernproc->p_rlimit[0]); i++) limit0.pl_rlimit[i].rlim_cur = limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY; limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = NOFILE; - limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = MAXUPRC; + limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = maxprocperuid; limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc; limit0.pl_rlimit[RLIMIT_STACK] = vm_initial_limit_stack; limit0.pl_rlimit[RLIMIT_DATA] = vm_initial_limit_data; limit0.pl_rlimit[RLIMIT_CORE] = vm_initial_limit_core; - limit0.p_refcnt = 1; + limit0.pl_refcnt = 1; - p->p_stats = &pstats0; - p->p_sigacts = &sigacts0; + kernproc->p_stats = &pstats0; + kernproc->p_sigacts = &sigacts0; /* - * Charge root for one process. + * Charge root for two processes: init and mach_init. */ + bsd_init_kprintf("calling chgproccnt\n"); (void)chgproccnt(0, 1); /* @@ -369,71 +643,154 @@ bsd_init() * for temporary copying (execve()). */ { - vm_offset_t min; + vm_offset_t minimum; + bsd_init_kprintf("calling kmem_suballoc\n"); + assert(bsd_pageable_map_size != 0); ret = kmem_suballoc(kernel_map, - &min, - (vm_size_t)BSD_PAGABLE_MAP_SIZE, - TRUE, + &minimum, + (vm_size_t)bsd_pageable_map_size, TRUE, + VM_FLAGS_ANYWHERE, &bsd_pageable_map); if (ret != KERN_SUCCESS) panic("bsd_init: Failed to allocate bsd pageable map"); } + /* + * Initialize buffers and hash links for buffers + * + * SIDE EFFECT: Starts a thread for bcleanbuf_thread(), so must + * happen after a credential has been associated with + * the kernel task. + */ + bsd_init_kprintf("calling bsd_bufferinit\n"); + bsd_bufferinit(); + /* Initialize the execve() semaphore */ - ret = semaphore_create(kernel_task, &execve_semaphore, - SYNC_POLICY_FIFO, (BSD_PAGABLE_MAP_SIZE / NCARGS)); + bsd_init_kprintf("calling semaphore_create\n"); + if (ret != KERN_SUCCESS) panic("bsd_init: Failed to create execve semaphore"); /* * Initialize the calendar. */ - IOKitResetTime(); + bsd_init_kprintf("calling IOKitInitializeTime\n"); + IOKitInitializeTime(); + bsd_init_kprintf("calling ubc_init\n"); ubc_init(); + /* + * Initialize device-switches. + */ + bsd_init_kprintf("calling devsw_init() \n"); + devsw_init(); + /* Initialize the file systems. */ + bsd_init_kprintf("calling vfsinit\n"); vfsinit(); +#if SOCKETS + /* Initialize per-CPU cache allocator */ + mcache_init(); + /* Initialize mbuf's. */ + bsd_init_kprintf("calling mbinit\n"); mbinit(); - - /* Initialize syslog */ - log_init(); + net_str_id_init(); /* for mbuf tags */ +#endif /* SOCKETS */ /* * Initializes security event auditing. * XXX: Should/could this occur later? */ - audit_init(); +#if CONFIG_AUDIT + bsd_init_kprintf("calling audit_init\n"); + audit_init(); +#endif /* Initialize kqueues */ + bsd_init_kprintf("calling knote_init\n"); knote_init(); /* Initialize for async IO */ + bsd_init_kprintf("calling aio_init\n"); aio_init(); + /* Initialize pipes */ + bsd_init_kprintf("calling pipeinit\n"); + pipeinit(); + + /* Initialize SysV shm subsystem locks; the subsystem proper is + * initialized through a sysctl. + */ +#if SYSV_SHM + bsd_init_kprintf("calling sysv_shm_lock_init\n"); + sysv_shm_lock_init(); +#endif +#if SYSV_SEM + bsd_init_kprintf("calling sysv_sem_lock_init\n"); + sysv_sem_lock_init(); +#endif +#if SYSV_MSG + bsd_init_kprintf("sysv_msg_lock_init\n"); + sysv_msg_lock_init(); +#endif + bsd_init_kprintf("calling pshm_lock_init\n"); + pshm_lock_init(); + bsd_init_kprintf("calling psem_lock_init\n"); + psem_lock_init(); + + pthread_init(); /* POSIX Shm and Sem */ + bsd_init_kprintf("calling pshm_cache_init\n"); pshm_cache_init(); + bsd_init_kprintf("calling psem_cache_init\n"); psem_cache_init(); + bsd_init_kprintf("calling time_zone_slock_init\n"); time_zone_slock_init(); + bsd_init_kprintf("calling select_wait_queue_init\n"); + select_wait_queue_init(); + /* Stack snapshot facility lock */ + stackshot_lock_init(); /* * Initialize protocols. Block reception of incoming packets * until everything is ready. */ - s = splimp(); + bsd_init_kprintf("calling sysctl_register_fixed\n"); sysctl_register_fixed(); + bsd_init_kprintf("calling sysctl_mib_init\n"); sysctl_mib_init(); +#if NETWORKING + bsd_init_kprintf("calling dlil_init\n"); dlil_init(); + bsd_init_kprintf("calling proto_kpi_init\n"); + proto_kpi_init(); +#endif /* NETWORKING */ +#if SOCKETS + bsd_init_kprintf("calling socketinit\n"); socketinit(); + bsd_init_kprintf("calling domaininit\n"); domaininit(); - splx(s); +#endif /* SOCKETS */ + + kernproc->p_fd->fd_cdir = NULL; + kernproc->p_fd->fd_rdir = NULL; - p->p_fd->fd_cdir = NULL; - p->p_fd->fd_rdir = NULL; +#if CONFIG_FREEZE + /* Initialise background hibernation */ + bsd_init_kprintf("calling kern_hibernation_init\n"); + kern_hibernation_init(); +#endif + +#if CONFIG_EMBEDDED + /* Initialize kernel memory status notifications */ + bsd_init_kprintf("calling kern_memorystatus_init\n"); + kern_memorystatus_init(); +#endif #ifdef GPROF /* Initialize kernel profiling. */ @@ -442,10 +799,15 @@ bsd_init() /* kick off timeout driven events by calling first time */ thread_wakeup(&lbolt); - timeout((void (*)(void *))lightning_bolt, 0, hz); + timeout(lightning_bolt, 0, hz); + bsd_init_kprintf("calling bsd_autoconf\n"); bsd_autoconf(); +#if CONFIG_DTRACE + dtrace_postinit(); +#endif + /* * We attach the loopback interface *way* down here to ensure * it happens after autoconf(), otherwise it becomes the @@ -453,141 +815,210 @@ bsd_init() */ #include #if NLOOP > 0 + bsd_init_kprintf("calling loopattach\n"); loopattach(); /* XXX */ #endif - - /* Register the built-in dlil ethernet interface family */ + +#if PFLOG + /* Initialize packet filter log interface */ + pfloginit(); +#endif /* PFLOG */ + +#if NETHER > 0 + /* Register the built-in dlil ethernet interface family */ + bsd_init_kprintf("calling ether_family_init\n"); ether_family_init(); +#endif /* ETHER */ + +#if NETWORKING + /* Call any kext code that wants to run just after network init */ + bsd_init_kprintf("calling net_init_run\n"); + net_init_run(); + + /* register user tunnel kernel control handler */ + utun_register_control(); + netsrc_init(); + + /* wait for network domain to finish */ + domainfin(); +#endif /* NETWORKING */ + bsd_init_kprintf("calling vnode_pager_bootstrap\n"); vnode_pager_bootstrap(); +#if 0 + /* XXX Hack for early debug stop */ + printf("\nabout to sleep for 10 seconds\n"); + IOSleep( 10 * 1000 ); + /* Debugger("hello"); */ +#endif + + bsd_init_kprintf("calling inittodr\n"); + inittodr(0); /* Mount the root file system. */ while( TRUE) { int err; + bsd_init_kprintf("calling setconf\n"); setconf(); - /* - * read the time after clock_initialize_calendar() - * and before nfs mount - */ - microtime((struct timeval *)&time); - - bsd_hardclockinit = -1; /* start ticking */ +#if NFSCLIENT + netboot = (mountroot == netboot_mountroot); +#endif + bsd_init_kprintf("vfs_mountroot\n"); if (0 == (err = vfs_mountroot())) break; - if (mountroot == netboot_mountroot) { - printf("cannot mount network root, errno = %d\n", err); - mountroot = NULL; - if (0 == (err = vfs_mountroot())) - break; + rootdevice[0] = '\0'; +#if NFSCLIENT + if (netboot) { + PE_display_icon( 0, "noroot"); /* XXX a netboot-specific icon would be nicer */ + vc_progress_set(FALSE, 0); + for (i=1; 1; i*=2) { + printf("bsd_init: failed to mount network root, error %d, %s\n", + err, PE_boot_args()); + printf("We are hanging here...\n"); + IOSleep(i*60*1000); + } + /*NOTREACHED*/ } +#endif printf("cannot mount root, errno = %d\n", err); boothowto |= RB_ASKNAME; } - mountlist.cqh_first->mnt_flag |= MNT_ROOTFS; + IOSecureBSDRoot(rootdevice); + + context.vc_thread = current_thread(); + context.vc_ucred = kernproc->p_ucred; + mountlist.tqh_first->mnt_flag |= MNT_ROOTFS; + bsd_init_kprintf("calling VFS_ROOT\n"); /* Get the vnode for '/'. Set fdp->fd_fd.fd_cdir to reference it. */ - if (VFS_ROOT(mountlist.cqh_first, &rootvnode)) - panic("bsd_init: cannot find root vnode"); - VREF(rootvnode); + if (VFS_ROOT(mountlist.tqh_first, &rootvnode, &context)) + panic("bsd_init: cannot find root vnode: %s", PE_boot_args()); + rootvnode->v_flag |= VROOT; + (void)vnode_ref(rootvnode); + (void)vnode_put(rootvnode); filedesc0.fd_cdir = rootvnode; - VOP_UNLOCK(rootvnode, 0, p); - if (mountroot == netboot_mountroot) { +#if NFSCLIENT + if (netboot) { int err; + + netboot = TRUE; /* post mount setup */ - if (err = netboot_setup(p)) { - panic("bsd_init: NetBoot could not find root, %d", err); + if ((err = netboot_setup()) != 0) { + PE_display_icon( 0, "noroot"); /* XXX a netboot-specific icon would be nicer */ + vc_progress_set(FALSE, 0); + for (i=1; 1; i*=2) { + printf("bsd_init: NetBoot could not find root, error %d: %s\n", + err, PE_boot_args()); + printf("We are hanging here...\n"); + IOSleep(i*60*1000); + } + /*NOTREACHED*/ } } +#endif +#if CONFIG_IMAGEBOOT /* - * Now can look at time, having had a chance to verify the time - * from the file system. Reset p->p_rtime as it may have been - * munched in mi_switch() after the time got set. - */ - p->p_stats->p_start = boottime = time; - p->p_rtime.tv_sec = p->p_rtime.tv_usec = 0; + * See if a system disk image is present. If so, mount it and + * switch the root vnode to point to it + */ + if (netboot == FALSE && imageboot_needed()) { + /* + * An image was found. No turning back: we're booted + * with a kernel from the disk image. + */ + imageboot_setup(); + } +#endif /* CONFIG_IMAGEBOOT */ + + /* set initial time; all other resource data is already zero'ed */ + microtime(&kernproc->p_start); + kernproc->p_stats->p_start = kernproc->p_start; /* for compat */ #if DEVFS { - extern void devfs_kernel_mount(char * str); - - devfs_kernel_mount("/dev"); + char mounthere[] = "/dev"; /* !const because of internal casting */ + + bsd_init_kprintf("calling devfs_kernel_mount\n"); + devfs_kernel_mount(mounthere); } #endif /* DEVFS */ /* Initialize signal state for process 0. */ - siginit(p); + bsd_init_kprintf("calling siginit\n"); + siginit(kernproc); + bsd_init_kprintf("calling bsd_utaskbootstrap\n"); bsd_utaskbootstrap(); +#if defined(__LP64__) + kernproc->p_flag |= P_LP64; + printf("Kernel is LP64\n"); +#endif + + pal_kernel_announce(); + +#if __i386__ || __x86_64__ + /* this should be done after the root filesystem is mounted */ + error = set_archhandler(kernproc, CPU_TYPE_POWERPC); + if (error) /* XXX make more generic */ + exec_archhandler_ppc.path[0] = 0; +#endif + + bsd_init_kprintf("calling mountroot_post_hook\n"); + /* invoke post-root-mount hook */ if (mountroot_post_hook != NULL) mountroot_post_hook(); - - (void) thread_funnel_set(kernel_flock, funnel_state); + +#if 0 /* not yet */ + consider_zone_gc(FALSE); +#endif + + bsd_init_kprintf("done\n"); } /* Called with kernel funnel held */ void bsdinit_task(void) { - struct proc *p = current_proc(); + proc_t p = current_proc(); struct uthread *ut; - kern_return_t kr; - thread_act_t th_act; - shared_region_mapping_t system_region; + thread_t thread; - proc_name("init", p); + process_name("init", p); ux_handler_init(); - th_act = current_act(); + thread = current_thread(); (void) host_set_exception_ports(host_priv_self(), - EXC_MASK_ALL & ~(EXC_MASK_SYSCALL | - EXC_MASK_MACH_SYSCALL | - EXC_MASK_RPC_ALERT), - ux_exception_port, - EXCEPTION_DEFAULT, 0); - - (void) task_set_exception_ports(get_threadtask(th_act), - EXC_MASK_ALL & ~(EXC_MASK_SYSCALL | - EXC_MASK_MACH_SYSCALL | - EXC_MASK_RPC_ALERT), - ux_exception_port, - EXCEPTION_DEFAULT, 0); - - - + EXC_MASK_ALL & ~(EXC_MASK_RPC_ALERT),//pilotfish (shark) needs this port + (mach_port_t) ux_exception_port, + EXCEPTION_DEFAULT| MACH_EXCEPTION_CODES, + 0); - ut = (uthread_t)get_bsdthread_info(th_act); - ut->uu_ar0 = (void *)get_user_regs(th_act); + ut = (uthread_t)get_bsdthread_info(thread); - bsd_hardclockinit = 1; /* Start bsd hardclock */ - bsd_init_task = get_threadtask(th_act); + bsd_init_task = get_threadtask(thread); init_task_failure_data[0] = 0; - system_region = lookup_default_shared_region(ENV_DEFAULT_ROOT, - machine_slot[cpu_number()].cpu_type); - if (system_region == NULL) { - shared_file_boot_time_init(ENV_DEFAULT_ROOT, - machine_slot[cpu_number()].cpu_type); - } else { - vm_set_shared_region(get_threadtask(th_act), system_region); - } + +#if CONFIG_MACF + mac_cred_label_associate_user(p->p_ucred); + mac_task_label_update_cred (p->p_ucred, (struct task *) p->task); +#endif load_init_program(p); - /* turn on app-profiling i.e. pre-heating */ - app_profile = 1; + lock_trace = 1; } void -lightning_bolt() +lightning_bolt(__unused void *dummy) { boolean_t funnel_state; - extern void klogwakeup(void); funnel_state = thread_funnel_set(kernel_flock, TRUE); @@ -598,10 +1029,10 @@ lightning_bolt() (void) thread_funnel_set(kernel_flock, FALSE); } -bsd_autoconf() +kern_return_t +bsd_autoconf(void) { - extern kern_return_t IOKitBSDInit( void ); - + kprintf("bsd_autoconf: calling kminit\n"); kminit(); /* @@ -620,10 +1051,9 @@ bsd_autoconf() #include /* for MAXPARTITIONS */ -setconf() +static void +setconf(void) { - extern kern_return_t IOFindBSDRoot( char * rootName, - dev_t * root, u_int32_t * flags ); u_int32_t flags; kern_return_t err; @@ -632,160 +1062,148 @@ setconf() * which needs to be under network funnel. Right thing to do * here is to drop the funnel alltogether and regrab it afterwards */ - thread_funnel_set(kernel_flock, FALSE); - err = IOFindBSDRoot( rootdevice, &rootdev, &flags ); - thread_funnel_set(kernel_flock, TRUE); + err = IOFindBSDRoot(rootdevice, sizeof(rootdevice), &rootdev, &flags); if( err) { printf("setconf: IOFindBSDRoot returned an error (%d);" "setting rootdevice to 'sd0a'.\n", err); /* XXX DEBUG TEMP */ rootdev = makedev( 6, 0 ); - strcpy( rootdevice, "sd0a" ); + strlcpy(rootdevice, "sd0a", sizeof(rootdevice)); flags = 0; } +#if NFSCLIENT if( flags & 1 ) { /* network device */ mountroot = netboot_mountroot; } else { +#endif /* otherwise have vfs determine root filesystem */ mountroot = NULL; +#if NFSCLIENT } +#endif } -bsd_utaskbootstrap() +void +bsd_utaskbootstrap(void) { - thread_act_t th_act; + thread_t thread; struct uthread *ut; - th_act = cloneproc(kernproc, 0); - initproc = pfind(1); - /* Set the launch time for init */ - initproc->p_stats->p_start = time; + /* + * Clone the bootstrap process from the kernel process, without + * inheriting either task characteristics or memory from the kernel; + */ + thread = cloneproc(TASK_NULL, kernproc, FALSE); + + /* Hold the reference as it will be dropped during shutdown */ + initproc = proc_find(1); +#if __PROC_INTERNAL_DEBUG + if (initproc == PROC_NULL) + panic("bsd_utaskbootstrap: initproc not set\n"); +#endif + /* + * Since we aren't going back out the normal way to our parent, + * we have to drop the transition locks explicitly. + */ + proc_signalend(initproc, 0); + proc_transend(initproc, 0); - ut = (struct uthread *)get_bsdthread_info(th_act); + ut = (struct uthread *)get_bsdthread_info(thread); ut->uu_sigmask = 0; - act_set_astbsd(th_act); - (void) thread_resume(th_act); + act_set_astbsd(thread); + (void) thread_resume(thread); } -parse_bsd_args() +static void +parse_bsd_args(void) { - extern char init_args[]; - char namep[16]; - extern int boothowto; - extern int srv; - extern int ncl; - - int len; + char namep[16]; + int msgbuf; - if (PE_parse_boot_arg("-s", namep)) { + if (PE_parse_boot_argn("-s", namep, sizeof (namep))) boothowto |= RB_SINGLE; - len = strlen(init_args); - if(len != 0) - strcat(init_args," -s"); - else - strcat(init_args,"-s"); - } - if (PE_parse_boot_arg("-b", namep)) { + if (PE_parse_boot_argn("-b", namep, sizeof (namep))) boothowto |= RB_NOBOOTRC; - len = strlen(init_args); - if(len != 0) - strcat(init_args," -b"); - else - strcat(init_args,"-b"); - } - if (PE_parse_boot_arg("-F", namep)) { - len = strlen(init_args); - if(len != 0) - strcat(init_args," -F"); - else - strcat(init_args,"-F"); - } + if (PE_parse_boot_argn("-x", namep, sizeof (namep))) /* safe boot */ + boothowto |= RB_SAFEBOOT; - if (PE_parse_boot_arg("-v", namep)) { - len = strlen(init_args); - if(len != 0) - strcat(init_args," -v"); - else - strcat(init_args,"-v"); - } + /* disable 64 bit grading */ + if (PE_parse_boot_argn("-no64exec", namep, sizeof (namep))) + bootarg_no64exec = 1; - if (PE_parse_boot_arg("-x", namep)) { /* safe boot */ - len = strlen(init_args); - if(len != 0) - strcat(init_args," -x"); - else - strcat(init_args,"-x"); - } + /* disable vnode_cache_is_authorized() by setting vnode_cache_defeat */ + if (PE_parse_boot_argn("-vnode_cache_defeat", namep, sizeof (namep))) + bootarg_vnode_cache_defeat = 1; + +#if DEVELOPMENT || DEBUG + if (PE_parse_boot_argn("-disable_aslr", namep, sizeof (namep))) + bootarg_disable_aslr = 1; +#endif - if (PE_parse_boot_arg("-d", namep)) { - len = strlen(init_args); - if(len != 0) - strcat(init_args," -d"); - else - strcat(init_args,"-d"); + PE_parse_boot_argn("ncl", &ncl, sizeof (ncl)); + if (PE_parse_boot_argn("nbuf", &max_nbuf_headers, + sizeof (max_nbuf_headers))) { + customnbuf = 1; } +#if !defined(SECURE_KERNEL) + PE_parse_boot_argn("kmem", &setup_kmem, sizeof (setup_kmem)); +#endif + +#if CONFIG_MACF +#if defined (__i386__) || defined (__x86_64__) + PE_parse_boot_argn("policy_check", &policy_check_flags, sizeof (policy_check_flags)); +#endif +#endif /* CONFIG_MACF */ - PE_parse_boot_arg("srv", &srv); - PE_parse_boot_arg("ncl", &ncl); - PE_parse_boot_arg("nbuf", &nbuf); + if (PE_parse_boot_argn("msgbuf", &msgbuf, sizeof (msgbuf))) { + log_setsize(msgbuf); + } - return 0; + if (PE_parse_boot_argn("-novfscache", namep, sizeof(namep))) { + nc_disabled = 1; + } } -boolean_t -thread_funnel_switch( - int oldfnl, - int newfnl) +void +bsd_exec_setup(int scale) { - boolean_t funnel_state_prev; - int curfnl; - funnel_t * curflock; - funnel_t * oldflock; - funnel_t * newflock; - funnel_t * exist_funnel; - extern int disable_funnel; - - - if (disable_funnel) - return(TRUE); - if(oldfnl == newfnl) { - panic("thread_funnel_switch: can't switch to same funnel"); - } - - if ((oldfnl != NETWORK_FUNNEL) && (oldfnl != KERNEL_FUNNEL)) { - panic("thread_funnel_switch: invalid oldfunnel"); - } - if ((newfnl != NETWORK_FUNNEL) && (newfnl != KERNEL_FUNNEL)) { - panic("thread_funnel_switch: invalid newfunnel"); - } - - if((curflock = thread_funnel_get()) == THR_FUNNEL_NULL) { - panic("thread_funnel_switch: no funnel held"); + switch (scale) { + case 0: + case 1: + bsd_simul_execs = BSD_SIMUL_EXECS; + break; + case 2: + case 3: + bsd_simul_execs = 65; + break; + case 4: + case 5: + bsd_simul_execs = 129; + break; + case 6: + case 7: + bsd_simul_execs = 257; + break; + default: + bsd_simul_execs = 513; + break; + } - - if ((oldfnl == NETWORK_FUNNEL) && (curflock != network_flock)) - panic("thread_funnel_switch: network funnel not held"); - - if ((oldfnl == KERNEL_FUNNEL) && (curflock != kernel_flock)) - panic("thread_funnel_switch: kernel funnel not held"); - - if(oldfnl == NETWORK_FUNNEL) { - oldflock = network_flock; - newflock = kernel_flock; - } else { - oldflock = kernel_flock; - newflock = network_flock; - } - KERNEL_DEBUG(0x603242c | DBG_FUNC_NONE, oldflock, 1, 0, 0, 0); - thread_funnel_set(oldflock, FALSE); - KERNEL_DEBUG(0x6032428 | DBG_FUNC_NONE, newflock, 1, 0, 0, 0); - thread_funnel_set(newflock, TRUE); - KERNEL_DEBUG(0x6032434 | DBG_FUNC_NONE, newflock, 1, 0, 0, 0); - - return(TRUE); + bsd_pageable_map_size = (bsd_simul_execs * BSD_PAGEABLE_SIZE_PER_EXEC); +} + +#if !NFSCLIENT +int +netboot_root(void); + +int +netboot_root(void) +{ + return(0); } +#endif