X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/593a1d5fd87cdf5b46dd5fcb84467b432cea0f91..d9a64523371fa019c4575bb400cbbc3a50ac9903:/bsd/kern/bsd_init.c diff --git a/bsd/kern/bsd_init.c b/bsd/kern/bsd_init.c index cf4ee656a..23e115db0 100644 --- a/bsd/kern/bsd_init.c +++ b/bsd/kern/bsd_init.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2000-2007 Apple Inc. All rights reserved. + * Copyright (c) 2000-2015 Apple Inc. All rights reserved. * * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * @@ -24,8 +24,8 @@ * limitations under the License. * * @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. @@ -94,8 +94,9 @@ #include #include #include +#include -#include +#include #include #include @@ -105,18 +106,19 @@ #include #include #include +#include +#include /* for ux_handler_setup() */ #include #include #include -#include /* for ux_exception_port */ - #include -#include #include /* for pseudo_inits */ #include +#include +#include #include #include @@ -125,28 +127,45 @@ #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 eventhandler_init() */ +#include /* for memorystatus_init() */ #include /* for aio_init() */ #include /* for psem_cache_init() */ #include /* for dlil_init() */ #include /* for proto_kpi_init() */ +#include /* for iptap_init() */ #include /* for pipeinit() */ #include /* for socketinit() */ #include /* for domaininit() */ #include /* for thread_wakeup() */ #include /* for ether_family_init() */ +#include /* for gif_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 /* proc_uuid_policy_init() */ +#include /* flow_divert_init() */ +#include /* for cfil_init() */ +#include /* for necp_init() */ +#include /* for netagent_init() */ +#include /* for pkt_mnglr_init() */ +#include /* for utun_register_control() */ +#include /* for ipsec_register_control() */ +#include /* for net_str_id_init() */ +#include /* for netsrc_init() */ +#include /* for nstat_init() */ +#include /* for tcp_cc_init() */ +#include /* for mptcp_control_register() */ +#include /* for nwk_wq_init */ +#include /* for assert() */ +#include /* for init_system_override() */ #include @@ -158,17 +177,33 @@ #include +#if NFSCLIENT +#include +#endif + #if CONFIG_IMAGEBOOT #include #endif +#if PFLOG +#include +#endif + + #include +#include +#include + +#if CONFIG_XNUPOST +#include +#endif void * get_user_regs(thread_t); /* XXX kludge for */ void IOKitInitializeTime(void); /* XXX */ +void IOSleep(unsigned int); /* XXX */ void loopattach(void); /* XXX */ -char copyright[] = +const char copyright[] = "Copyright (c) 1982, 1986, 1989, 1991, 1993\n\t" "The Regents of the University of California. " "All rights reserved.\n\n"; @@ -204,62 +239,97 @@ char hostname[MAXHOSTNAMELEN]; int hostnamelen; char domainname[MAXDOMNAMELEN]; int domainnamelen; -#if __i386__ -struct exec_archhandler exec_archhandler_ppc = { - .path = "/usr/libexec/oah/translate", -}; -#else /* __i386__ */ -struct exec_archhandler exec_archhandler_ppc; -#endif /* __i386__ */ -char rootdevice[16]; /* hfs device names have at least 9 chars */ +char rootdevice[DEVMAXNAMESIZE]; #if KMEMSTATS struct kmemstats kmemstats[M_LAST]; #endif -int lbolt; /* awoken once a second */ struct vnode *rootvp; int boothowto = RB_DEBUG; +int minimalboot = 0; +#if CONFIG_EMBEDDED +int darkboot = 0; +#endif + +#if PROC_REF_DEBUG +__private_extern__ int proc_ref_tracking_disabled = 0; /* disable panics on leaked proc refs across syscall boundary */ +#endif + +#if OS_REASON_DEBUG +__private_extern__ int os_reason_debug_disabled = 0; /* disable asserts for when we fail to allocate OS reasons */ +#endif -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 void oslog_setsize(int size); +extern void throttle_init(void); +extern void acct_init(void); -extern int srv; +extern int serverperfmode; extern int ncl; -#define BSD_SIMUL_EXECS 33 /* 32 , allow for rounding */ -#define BSD_PAGABLE_MAP_SIZE (BSD_SIMUL_EXECS * (NCARGS + PAGE_SIZE)) 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); + +#if __arm64__ +__private_extern__ int bootarg_no64exec = 0; +#endif +#if __x86_64__ +__private_extern__ int bootarg_no32exec = 0; +#endif +__private_extern__ int bootarg_vnode_cache_defeat = 0; + +#if CONFIG_JETSAM && (DEVELOPMENT || DEBUG) +__private_extern__ int bootarg_no_vnode_jetsam = 0; +#endif /* CONFIG_JETSAM && (DEVELOPMENT || DEBUG) */ + +/* + * Prevent kernel-based ASLR from being used, for testing. + */ +#if DEVELOPMENT || DEBUG +__private_extern__ int bootarg_disable_aslr = 0; +#endif + +/* + * Allow an alternate dyld to be used for testing. + */ + +#if DEVELOPMENT || DEBUG +char dyld_alt_path[MAXPATHLEN]; +int use_alt_dyld = 0; +#endif int cmask = CMASK; extern int customnbuf; -void bsd_init(void) __attribute__((section("__TEXT, initcode"))); -__private_extern__ void ubc_init(void ) __attribute__((section("__TEXT, initcode"))); -void vfsinit(void) __attribute__((section("__TEXT, initcode"))); -kern_return_t bsd_autoconf(void) __attribute__((section("__TEXT, initcode"))); -void bsd_utaskbootstrap(void) __attribute__((section("__TEXT, initcode"))); +kern_return_t bsd_autoconf(void); +void bsd_utaskbootstrap(void); static void parse_bsd_args(void); -extern task_t bsd_init_task; -extern char init_task_failure_data[]; +#if CONFIG_DEV_KMEM +extern void dev_kmem_init(void); +#endif extern void time_zone_slock_init(void); +extern void select_waitq_init(void); static void process_name(const char *, proc_t); static void setconf(void); -funnel_t *kernel_flock; - #if SYSV_SHM extern void sysv_shm_lock_init(void); #endif @@ -269,19 +339,22 @@ extern void sysv_sem_lock_init(void); #if SYSV_MSG extern void sysv_msg_lock_init(void); #endif -extern void pthread_init(void); -/* kmem access not enabled by default; can be changed with boot-args */ -int setup_kmem = 0; +extern void ulock_initialize(void); -/* size of kernel trace buffer, disabled by default */ -unsigned int new_nkdbufs = 0; +#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; -/* mach leak logging */ -int log_leaks = 0; -int turn_on_log_leaks = 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. @@ -301,32 +374,52 @@ extern void stackshot_lock_init(void); static void process_name(const char *s, proc_t p) { - size_t length = strlen(s); - - bcopy(s, p->p_comm, - length >= sizeof(p->p_comm) ? sizeof(p->p_comm) : - length + 1); + strlcpy(p->p_comm, s, sizeof(p->p_comm)); + strlcpy(p->p_name, s, sizeof(p->p_name)); } /* To allow these values to be patched, they're globals here */ #include -struct rlimit vm_initial_limit_stack = { DFLSSIZ, MAXSSIZ - PAGE_SIZE }; +struct rlimit vm_initial_limit_stack = { DFLSSIZ, MAXSSIZ - PAGE_MAX_SIZE }; struct rlimit vm_initial_limit_data = { DFLDSIZ, MAXDSIZ }; struct rlimit vm_initial_limit_core = { DFLCSIZ, MAXCSIZ }; -extern thread_t cloneproc(proc_t, int); +extern thread_t cloneproc(task_t, coalition_t, proc_t, int, int); extern int (*mountroot)(void); -extern int netboot_mountroot(void); /* netboot.c */ -extern int netboot_setup(void); lck_grp_t * proc_lck_grp; +lck_grp_t * proc_slock_grp; +lck_grp_t * proc_fdmlock_grp; +lck_grp_t * proc_kqhashlock_grp; +lck_grp_t * proc_knhashlock_grp; +lck_grp_t * proc_ucred_mlock_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; +#if CONFIG_XNUPOST +lck_grp_t * sysctl_debug_test_stackshot_owner_grp; +lck_mtx_t * sysctl_debug_test_stackshot_owner_init_mtx; +#endif /* !CONFIG_XNUPOST */ + +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 before IOKit initialization, so that globals + * like the sysctl tree are initialized before kernel extensions + * are started (since they may want to register sysctls + */ +void +bsd_early_init(void) +{ + sysctl_early_init(); +} /* * This function is called very early on in the Mach startup, from the @@ -342,26 +435,24 @@ void (*mountroot_post_hook)(void); * of the uu_context.vc_ucred field so that the uthread structure can be * used like any other. */ + void bsd_init(void) { - proc_t p; struct uthread *ut; unsigned int i; -#if __i386__ - int error; -#endif struct vfs_context context; kern_return_t ret; struct ucred temp_cred; + struct posix_cred temp_pcred; +#if NFSCLIENT || CONFIG_IMAGEBOOT + boolean_t netboot = FALSE; +#endif #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"); - } - + throttle_init(); + printf(copyright); bsd_init_kprintf("calling kmeminit\n"); @@ -370,6 +461,11 @@ bsd_init(void) bsd_init_kprintf("calling parse_bsd_args\n"); parse_bsd_args(); +#if CONFIG_DEV_KMEM + bsd_init_kprintf("calling dev_kmem_init\n"); + dev_kmem_init(); +#endif + /* Initialize kauth subsystem before instancing the first credential */ bsd_init_kprintf("calling kauth_init\n"); kauth_init(); @@ -378,25 +474,38 @@ bsd_init(void) 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; - p->p_ppid = 0; + set_bsdtask_info(kernel_task,(void *)kernproc); /* give kernproc a name */ bsd_init_kprintf("calling process_name\n"); - process_name("kernel_task", p); + 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_ucred_mlock_grp = lck_grp_alloc_init("proc-ucred-mlock", proc_lck_grp_attr); + proc_mlock_grp = lck_grp_alloc_init("proc-mlock", proc_lck_grp_attr); + proc_fdmlock_grp = lck_grp_alloc_init("proc-fdmlock", proc_lck_grp_attr); +#endif + proc_kqhashlock_grp = lck_grp_alloc_init("proc-kqhashlock", proc_lck_grp_attr); + proc_knhashlock_grp = lck_grp_alloc_init("proc-knhashlock", proc_lck_grp_attr); +#if CONFIG_XNUPOST + sysctl_debug_test_stackshot_owner_grp = lck_grp_alloc_init("test-stackshot-owner-grp", LCK_GRP_ATTR_NULL); + sysctl_debug_test_stackshot_owner_init_mtx = lck_mtx_alloc_init( + sysctl_debug_test_stackshot_owner_grp, + LCK_ATTR_NULL); +#endif /* !CONFIG_XNUPOST */ /* Allocate proc lock attribute */ proc_lck_attr = lck_attr_alloc_init(); #if 0 @@ -405,12 +514,29 @@ bsd_init(void) #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_mtx_init(&kernproc->p_ucred_mlock, proc_ucred_mlock_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(&p->p_mlock, proc_lck_grp, proc_lck_attr); - lck_mtx_init(&p->p_fdmlock, proc_lck_grp, proc_lck_attr); - lck_spin_init(&p->p_slock, 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_mtx_init(&kernproc->p_ucred_mlock, 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"); @@ -423,95 +549,135 @@ bsd_init(void) * Initialize the MAC Framework */ mac_policy_initbsd(); - p->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 */ + ulock_initialize(); + /* * Create process 0. */ proc_list_lock(); - LIST_INSERT_HEAD(&allproc, p, p_list); - p->p_pgrp = &pgrp0; + LIST_INSERT_HEAD(&allproc, kernproc, p_list); + kernproc->p_pgrp = &pgrp0; LIST_INSERT_HEAD(PGRPHASH(0), &pgrp0, pg_hash); LIST_INIT(&pgrp0.pg_members); +#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, p, p_pglist); - p->p_listflag |= P_LIST_INPGRP; - p->p_pgrpid = 0; + 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(); -#if CONFIG_LCTX - p->p_lctx = NULL; +#if CONFIG_PERSONAS + kernproc->p_persona = NULL; #endif - p->task = kernel_task; + kernproc->task = kernel_task; - p->p_stat = SRUN; - p->p_flag = P_SYSTEM; - p->p_nice = NZERO; - p->p_pptr = p; + kernproc->p_stat = SRUN; + kernproc->p_flag = P_SYSTEM; + kernproc->p_lflag = 0; + kernproc->p_ladvflag = 0; - TAILQ_INIT(&p->p_uthlist); - TAILQ_INSERT_TAIL(&p->p_uthlist, ut, uu_list); +#if defined(__LP64__) + kernproc->p_flag |= P_LP64; +#endif + +#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); - p->sigwait = FALSE; - p->sigwait_thread = THREAD_NULL; - p->exit_thread = THREAD_NULL; - p->p_csflags = CS_VALID; + kernproc->sigwait = FALSE; + kernproc->sigwait_thread = THREAD_NULL; + kernproc->exit_thread = THREAD_NULL; + kernproc->p_csflags = CS_VALID; /* * Create credential. This also Initializes the audit information. - * XXX It is not clear what the initial values should be for audit ID, - * XXX session ID, etc.. */ bsd_init_kprintf("calling bzero\n"); bzero(&temp_cred, sizeof(temp_cred)); - temp_cred.cr_ngroups = 1; - + bzero(&temp_pcred, sizeof(temp_pcred)); + temp_pcred.cr_ngroups = 1; + /* kern_proc, shouldn't call up to DS for group membership */ + temp_pcred.cr_flags = CRF_NOMEMBERD; + temp_cred.cr_audit.as_aia_p = audit_default_aia_p; + bsd_init_kprintf("calling kauth_cred_create\n"); - p->p_ucred = kauth_cred_create(&temp_cred); + /* + * 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(p->p_ucred); - ut->uu_context.vc_ucred = p->p_ucred; + kauth_cred_ref(kernproc->p_ucred); + ut->uu_context.vc_ucred = kernproc->p_ucred; ut->uu_context.vc_thread = current_thread(); - TAILQ_INIT(&p->aio_activeq); - TAILQ_INIT(&p->aio_doneq); - p->aio_active_count = 0; - p->aio_done_count = 0; + 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(p->p_ucred); - mac_task_label_update_cred (p->p_ucred, (struct task *) p->task); + mac_cred_label_associate_kernel(kernproc->p_ucred); #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; filedesc0.fd_knhash = NULL; filedesc0.fd_knhashmask = 0; + lck_mtx_init(&filedesc0.fd_kqhashlock, proc_kqhashlock_grp, proc_lck_attr); + lck_mtx_init(&filedesc0.fd_knhashlock, proc_knhashlock_grp, proc_lck_attr); /* 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; @@ -522,11 +688,11 @@ bsd_init(void) limit0.pl_rlimit[RLIMIT_CORE] = vm_initial_limit_core; limit0.pl_refcnt = 1; - p->p_stats = &pstats0; - p->p_sigacts = &sigacts0; + kernproc->p_stats = &pstats0; + kernproc->p_sigacts = &sigacts0; /* - * Charge root for two processes: init and mach_init. + * Charge root for one process: launchd. */ bsd_init_kprintf("calling chgproccnt\n"); (void)chgproccnt(0, 1); @@ -539,16 +705,22 @@ bsd_init(void) vm_offset_t minimum; bsd_init_kprintf("calling kmem_suballoc\n"); + assert(bsd_pageable_map_size != 0); ret = kmem_suballoc(kernel_map, &minimum, - (vm_size_t)BSD_PAGABLE_MAP_SIZE, + (vm_size_t)bsd_pageable_map_size, TRUE, VM_FLAGS_ANYWHERE, + VM_MAP_KERNEL_FLAGS_NONE, + VM_KERN_MEMORY_BSD, &bsd_pageable_map); if (ret != KERN_SUCCESS) panic("bsd_init: Failed to allocate bsd pageable map"); } + bsd_init_kprintf("calling fpxlog_init\n"); + fpxlog_init(); + /* * Initialize buffers and hash links for buffers * @@ -559,25 +731,12 @@ bsd_init(void) bsd_init_kprintf("calling bsd_bufferinit\n"); bsd_bufferinit(); - /* Initialize the execve() semaphore */ - bsd_init_kprintf("calling semaphore_create\n"); - ret = semaphore_create(kernel_task, &execve_semaphore, - SYNC_POLICY_FIFO, BSD_SIMUL_EXECS -1); - if (ret != KERN_SUCCESS) - panic("bsd_init: Failed to create execve semaphore"); - /* * Initialize the calendar. */ bsd_init_kprintf("calling IOKitInitializeTime\n"); IOKitInitializeTime(); - if (turn_on_log_leaks && !new_nkdbufs) - new_nkdbufs = 200000; - start_kern_tracing(new_nkdbufs); - if (turn_on_log_leaks) - log_leaks = 1; - bsd_init_kprintf("calling ubc_init\n"); ubc_init(); @@ -585,6 +744,12 @@ bsd_init(void) bsd_init_kprintf("calling vfsinit\n"); vfsinit(); +#if CONFIG_PROC_UUID_POLICY + /* Initial proc_uuid_policy subsystem */ + bsd_init_kprintf("calling proc_uuid_policy_init()\n"); + proc_uuid_policy_init(); +#endif + #if SOCKETS /* Initialize per-CPU cache allocator */ mcache_init(); @@ -592,13 +757,14 @@ bsd_init(void) /* Initialize mbuf's. */ bsd_init_kprintf("calling mbinit\n"); mbinit(); + net_str_id_init(); /* for mbuf tags */ #endif /* SOCKETS */ /* * Initializes security event auditing. * XXX: Should/could this occur later? */ -#if AUDIT +#if CONFIG_AUDIT bsd_init_kprintf("calling audit_init\n"); audit_init(); #endif @@ -607,11 +773,9 @@ bsd_init(void) bsd_init_kprintf("calling knote_init\n"); knote_init(); -#if CONFIG_EMBEDDED - /* Initialize kernel memory status notifications */ - bsd_init_kprintf("calling kern_memorystatus_init\n"); - kern_memorystatus_init(); -#endif + /* Initialize event handler */ + bsd_init_kprintf("calling eventhandler_init\n"); + eventhandler_init(); /* Initialize for async IO */ bsd_init_kprintf("calling aio_init\n"); @@ -649,18 +813,16 @@ bsd_init(void) psem_cache_init(); bsd_init_kprintf("calling time_zone_slock_init\n"); time_zone_slock_init(); + bsd_init_kprintf("calling select_waitq_init\n"); + select_waitq_init(); - /* Stack snapshot facility lock */ - stackshot_lock_init(); /* * Initialize protocols. Block reception of incoming packets * until everything is ready. */ - 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 nwk_wq_init\n"); + nwk_wq_init(); bsd_init_kprintf("calling dlil_init\n"); dlil_init(); bsd_init_kprintf("calling proto_kpi_init\n"); @@ -671,25 +833,47 @@ bsd_init(void) socketinit(); bsd_init_kprintf("calling domaininit\n"); domaininit(); + iptap_init(); +#if FLOW_DIVERT + flow_divert_init(); +#endif /* FLOW_DIVERT */ #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 +#ifndef CONFIG_MEMORYSTATUS + #error "CONFIG_FREEZE defined without matching CONFIG_MEMORYSTATUS" +#endif + /* Initialise background freezing */ + bsd_init_kprintf("calling memorystatus_freeze_init\n"); + memorystatus_freeze_init(); +#endif + +#if CONFIG_MEMORYSTATUS + /* Initialize kernel memory status notifications */ + bsd_init_kprintf("calling memorystatus_init\n"); + memorystatus_init(); +#endif /* CONFIG_MEMORYSTATUS */ + + bsd_init_kprintf("calling acct_init\n"); + acct_init(); #ifdef GPROF /* Initialize kernel profiling. */ kmstartup(); #endif - /* kick off timeout driven events by calling first time */ - thread_wakeup(&lbolt); - timeout(lightning_bolt, 0, hz); + bsd_init_kprintf("calling sysctl_mib_init\n"); + sysctl_mib_init() bsd_init_kprintf("calling bsd_autoconf\n"); bsd_autoconf(); + bsd_init_kprintf("calling os_reason_init\n"); + os_reason_init(); + #if CONFIG_DTRACE - extern void dtrace_postinit(void); dtrace_postinit(); #endif @@ -703,7 +887,16 @@ bsd_init(void) bsd_init_kprintf("calling loopattach\n"); loopattach(); /* XXX */ #endif - +#if NGIF + /* Initialize gif interface (after lo0) */ + gif_init(); +#endif + +#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"); @@ -714,78 +907,66 @@ bsd_init(void) /* Call any kext code that wants to run just after network init */ bsd_init_kprintf("calling net_init_run\n"); net_init_run(); + +#if CONTENT_FILTER + cfil_init(); +#endif + +#if PACKET_MANGLER + pkt_mnglr_init(); +#endif + +#if NECP + /* Initialize Network Extension Control Policies */ + necp_init(); +#endif + + netagent_init(); + + /* register user tunnel kernel control handler */ + utun_register_control(); +#if IPSEC + ipsec_register_control(); +#endif /* IPSEC */ + netsrc_init(); + nstat_init(); + tcp_cc_init(); +#if MPTCP + mptcp_control_register(); +#endif /* MPTCP */ #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); -#if CONFIG_EMBEDDED - { - /* print out early VM statistics */ - kern_return_t kr1; - vm_statistics_data_t stat; - mach_msg_type_number_t count; - - count = HOST_VM_INFO_COUNT; - kr1 = host_statistics(host_self(), - HOST_VM_INFO, - (host_info_t)&stat, - &count); - kprintf("Mach Virtual Memory Statistics (page size of 4096) bytes\n" - "Pages free:\t\t\t%u.\n" - "Pages active:\t\t\t%u.\n" - "Pages inactive:\t\t\t%u.\n" - "Pages wired down:\t\t%u.\n" - "\"Translation faults\":\t\t%u.\n" - "Pages copy-on-write:\t\t%u.\n" - "Pages zero filled:\t\t%u.\n" - "Pages reactivated:\t\t%u.\n" - "Pageins:\t\t\t%u.\n" - "Pageouts:\t\t\t\%u.\n" - "Object cache: %u hits of %u lookups (%d%% hit rate)\n", - - stat.free_count, - stat.active_count, - stat.inactive_count, - stat.wire_count, - stat.faults, - stat.cow_faults, - stat.zero_fill_count, - stat.reactivations, - stat.pageins, - stat.pageouts, - stat.hits, - stat.lookups, - (stat.hits == 0) ? 100 : - ((stat.lookups * 100) / stat.hits)); - } -#endif /* CONFIG_EMBEDDED */ - /* Mount the root file system. */ while( TRUE) { int err; bsd_init_kprintf("calling setconf\n"); setconf(); +#if NFSCLIENT + netboot = (mountroot == netboot_mountroot); +#endif bsd_init_kprintf("vfs_mountroot\n"); if (0 == (err = vfs_mountroot())) break; rootdevice[0] = '\0'; #if NFSCLIENT - if (mountroot == netboot_mountroot) { - printf("bsd_init: netboot_mountroot failed," - " errno = %d\n", err); - panic("bsd_init: failed to mount network root: %s", PE_boot_args()); + 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); @@ -795,7 +976,7 @@ bsd_init(void) IOSecureBSDRoot(rootdevice); context.vc_thread = current_thread(); - context.vc_ucred = p->p_ucred; + context.vc_ucred = kernproc->p_ucred; mountlist.tqh_first->mnt_flag |= MNT_ROOTFS; bsd_init_kprintf("calling VFS_ROOT\n"); @@ -808,11 +989,21 @@ bsd_init(void) filedesc0.fd_cdir = rootvnode; #if NFSCLIENT - if (mountroot == netboot_mountroot) { + if (netboot) { int err; + + netboot = TRUE; /* post mount setup */ if ((err = netboot_setup()) != 0) { - panic("bsd_init: NetBoot could not find root, %d: %s", err, PE_boot_args()); + 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 @@ -823,25 +1014,17 @@ bsd_init(void) * See if a system disk image is present. If so, mount it and * switch the root vnode to point to it */ - - if(imageboot_needed()) { - int err; - - /* An image was found */ - if((err = imageboot_setup())) { - /* - * this is not fatal. Keep trying to root - * off the original media - */ - printf("%s: imageboot could not find root, %d\n", - __FUNCTION__, err); - } + 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 */ - microtime(&p->p_stats->p_start); /* for compat sake */ - microtime(&p->p_start); - p->p_rtime.tv_sec = p->p_rtime.tv_usec = 0; + /* set initial time; all other resource data is already zero'ed */ + microtime_with_abstime(&kernproc->p_start, &kernproc->p_stats->ps_start); #if DEVFS { @@ -851,20 +1034,15 @@ bsd_init(void) devfs_kernel_mount(mounthere); } #endif /* DEVFS */ - + /* Initialize signal state for process 0. */ bsd_init_kprintf("calling siginit\n"); - siginit(p); + siginit(kernproc); bsd_init_kprintf("calling bsd_utaskbootstrap\n"); bsd_utaskbootstrap(); -#if __i386__ - /* this should be done after the root filesystem is mounted */ - error = set_archhandler(p, CPU_TYPE_POWERPC); - if (error) /* XXX make more generic */ - exec_archhandler_ppc.path[0] = 0; -#endif + pal_kernel_announce(); bsd_init_kprintf("calling mountroot_post_hook\n"); @@ -873,60 +1051,43 @@ bsd_init(void) mountroot_post_hook(); #if 0 /* not yet */ - IOKitJettisonKLD(); - consider_zone_gc(); + consider_zone_gc(FALSE); #endif + + /* Initialize System Override call */ + init_system_override(); bsd_init_kprintf("done\n"); } -/* Called with kernel funnel held */ void bsdinit_task(void) { proc_t p = current_proc(); - struct uthread *ut; - thread_t thread; process_name("init", p); - ux_handler_init(); - - 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 | - EXC_MASK_CRASH), - (mach_port_t)ux_exception_port, - EXCEPTION_DEFAULT| MACH_EXCEPTION_CODES, - 0); - - ut = (uthread_t)get_bsdthread_info(thread); - - bsd_init_task = get_threadtask(thread); - init_task_failure_data[0] = 0; + /* Set up exception-to-signal reflection */ + ux_handler_setup(); #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); - lock_trace = 1; -} -void -lightning_bolt(__unused void *dummy) -{ - boolean_t funnel_state; + vm_init_before_launchd(); - funnel_state = thread_funnel_set(kernel_flock, TRUE); +#if CONFIG_XNUPOST + int result = bsd_list_tests(); + result = bsd_do_post(); + if (result != 0) { + panic("bsd_do_post: Tests failed with result = 0x%08x\n", result); + } +#endif - thread_wakeup(&lbolt); - timeout(lightning_bolt,0,hz); - klogwakeup(); + bsd_init_kprintf("bsd_do_post - done"); - (void) thread_funnel_set(kernel_flock, FALSE); + load_init_program(p); + lock_trace = 1; } kern_return_t @@ -957,11 +1118,6 @@ setconf(void) u_int32_t flags; kern_return_t err; - /* - * calls into IOKit can generate networking registrations - * which needs to be under network funnel. Right thing to do - * here is to drop the funnel alltogether and regrab it afterwards - */ err = IOFindBSDRoot(rootdevice, sizeof(rootdevice), &rootdev, &flags); if( err) { printf("setconf: IOFindBSDRoot returned an error (%d);" @@ -991,22 +1147,29 @@ bsd_utaskbootstrap(void) thread_t thread; struct uthread *ut; - thread = cloneproc(kernproc, 0); + /* + * Clone the bootstrap process from the kernel process, without + * inheriting either task characteristics or memory from the kernel; + */ + thread = cloneproc(TASK_NULL, COALITION_NULL, kernproc, FALSE, TRUE); + /* 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 - /* Set the launch time for init */ - microtime(&initproc->p_start); - microtime(&initproc->p_stats->p_start); /* for compat sake */ - + /* + * 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(thread); ut->uu_sigmask = 0; act_set_astbsd(thread); - (void) thread_resume(thread); + task_clear_return_wait(get_threadtask(thread)); } static void @@ -1015,7 +1178,7 @@ parse_bsd_args(void) char namep[16]; int msgbuf; - if (PE_parse_boot_argn("-s", namep, sizeof (namep))) + if ( PE_parse_boot_argn("-s", namep, sizeof (namep))) boothowto |= RB_SINGLE; if (PE_parse_boot_argn("-b", namep, sizeof (namep))) @@ -1024,25 +1187,137 @@ parse_bsd_args(void) if (PE_parse_boot_argn("-x", namep, sizeof (namep))) /* safe boot */ boothowto |= RB_SAFEBOOT; - if (PE_parse_boot_argn("-l", namep, sizeof (namep))) /* leaks logging */ - turn_on_log_leaks = 1; + if (PE_parse_boot_argn("-minimalboot", namep, sizeof(namep))) { + /* + * -minimalboot indicates that we want userspace to be bootstrapped to a + * minimal environment. What constitutes minimal is up to the bootstrap + * process. + */ + minimalboot = 1; + } + +#if __arm64__ + /* disable 64 bit grading */ + if (PE_parse_boot_argn("-no64exec", namep, sizeof (namep))) + bootarg_no64exec = 1; +#endif +#if __x86_64__ + /* disable 32 bit grading */ + if (PE_parse_boot_argn("-no32exec", namep, sizeof (namep))) + bootarg_no32exec = 1; +#endif + + /* 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 - PE_parse_boot_argn("srv", &srv, sizeof (srv)); PE_parse_boot_argn("ncl", &ncl, sizeof (ncl)); - if (PE_parse_boot_argn("nbuf", &max_nbuf_headers, sizeof (max_nbuf_headers))) { + 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)); + +#if CONFIG_MACF +#if defined (__i386__) || defined (__x86_64__) + PE_parse_boot_argn("policy_check", &policy_check_flags, sizeof (policy_check_flags)); #endif - PE_parse_boot_argn("trace", &new_nkdbufs, sizeof (new_nkdbufs)); +#endif /* CONFIG_MACF */ if (PE_parse_boot_argn("msgbuf", &msgbuf, sizeof (msgbuf))) { log_setsize(msgbuf); + oslog_setsize(msgbuf); + } + + if (PE_parse_boot_argn("-novfscache", namep, sizeof(namep))) { + nc_disabled = 1; } + +#if CONFIG_JETSAM && (DEVELOPMENT || DEBUG) + if (PE_parse_boot_argn("-no_vnode_jetsam", namep, sizeof(namep))) + bootarg_no_vnode_jetsam = 1; +#endif /* CONFIG_JETSAM && (DEVELOPMENT || DEBUG) */ + + +#if CONFIG_EMBEDDED + /* + * The darkboot flag is specified by the bootloader and is stored in + * boot_args->bootFlags. This flag is available starting revision 2. + */ + boot_args *args = (boot_args *) PE_state.bootArgs; + if ((args != NULL) && (args->Revision >= kBootArgsRevision2)) { + darkboot = (args->bootFlags & kBootFlagsDarkBoot) ? 1 : 0; + } else { + darkboot = 0; + } +#endif + +#if PROC_REF_DEBUG + if (PE_parse_boot_argn("-disable_procref_tracking", namep, sizeof(namep))) { + proc_ref_tracking_disabled = 1; + } +#endif + +#if OS_REASON_DEBUG + if (PE_parse_boot_argn("-disable_osreason_debug", namep, sizeof(namep))) { + os_reason_debug_disabled = 1; + } +#endif + + PE_parse_boot_argn("sigrestrict", &sigrestrict_arg, sizeof(sigrestrict_arg)); + +#if DEVELOPMENT|| DEBUG + if (PE_parse_boot_argn("-no_sigsys", namep, sizeof(namep))) { + send_sigsys = false; + } +#endif + +#if (DEVELOPMENT|| DEBUG) + if (PE_parse_boot_argn("alt-dyld", dyld_alt_path, sizeof(dyld_alt_path))) { + if (strlen(dyld_alt_path) > 0) { + use_alt_dyld = 1; + } + } +#endif +} + +void +bsd_exec_setup(int scale) +{ + + 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; + + } + bsd_pageable_map_size = (bsd_simul_execs * BSD_PAGEABLE_SIZE_PER_EXEC); } #if !NFSCLIENT +int +netboot_root(void); + int netboot_root(void) {