X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/13f56ec4e58bf8687e2a68032c093c0213dd519b..ecc0ceb4089d506a0b8d16686a95817b331af9cb:/bsd/kern/bsd_init.c?ds=sidebyside diff --git a/bsd/kern/bsd_init.c b/bsd/kern/bsd_init.c index e60df12e0..8cf33e20a 100644 --- a/bsd/kern/bsd_init.c +++ b/bsd/kern/bsd_init.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2000-2009 Apple Inc. All rights reserved. + * Copyright (c) 2000-2015 Apple Inc. All rights reserved. * * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * @@ -133,16 +133,18 @@ #include /* for mcache_init() */ #include /* for mbinit() */ #include /* for knote_init() */ -#include /* for kern_memorystatus_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() */ @@ -151,10 +153,21 @@ #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 assert() */ +#include /* for init_system_override() */ #include @@ -182,6 +195,7 @@ #include #include + void * get_user_regs(thread_t); /* XXX kludge for */ void IOKitInitializeTime(void); /* XXX */ void IOSleep(unsigned int); /* XXX */ @@ -223,13 +237,6 @@ char hostname[MAXHOSTNAMELEN]; int hostnamelen; char domainname[MAXDOMNAMELEN]; int domainnamelen; -#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 */ @@ -237,19 +244,24 @@ char rootdevice[16]; /* hfs device names have at least 9 chars */ struct kmemstats kmemstats[M_LAST]; #endif -int lbolt; /* awoken once a second */ struct vnode *rootvp; int boothowto = RB_DEBUG; +int minimalboot = 0; + +#if PROC_REF_DEBUG +__private_extern__ int proc_ref_tracking_disabled = 0; /* disable panics on leaked proc refs across syscall boundary */ +#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 throttle_init(void); +extern void macx_init(void); +extern void acct_init(void); extern int serverperfmode; extern int ncl; @@ -265,13 +277,12 @@ __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; +#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. */ @@ -282,21 +293,22 @@ __private_extern__ int bootarg_disable_aslr = 0; 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"))); +void bsd_init(void); +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[]; +extern boolean_t init_task_died; +#if CONFIG_DEV_KMEM +extern void dev_kmem_init(void); +#endif extern void time_zone_slock_init(void); -extern void select_wait_queue_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 @@ -307,12 +319,6 @@ extern void sysv_sem_lock_init(void); 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 */ @@ -322,15 +328,11 @@ 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. * Called from cold start routine as @@ -349,25 +351,23 @@ 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(task_t, proc_t, int); +extern thread_t cloneproc(task_t, coalition_t, proc_t, int, 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_ucred_mlock_grp; lck_grp_t * proc_mlock_grp; lck_grp_attr_t * proc_lck_grp_attr; lck_attr_t * proc_lck_attr; @@ -394,14 +394,12 @@ void (*unmountroot_pre_hook)(void); * of the uu_context.vc_ucred field so that the uthread structure can be * used like any other. */ + void bsd_init(void) { struct uthread *ut; unsigned int i; -#if __i386__ || __x86_64__ - int error; -#endif struct vfs_context context; kern_return_t ret; struct ucred temp_cred; @@ -412,11 +410,8 @@ bsd_init(void) #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"); @@ -425,6 +420,11 @@ 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(); @@ -453,6 +453,7 @@ bsd_init(void) #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_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); #endif /* Allocate proc lock attribute */ @@ -468,12 +469,14 @@ bsd_init(void) 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(&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 @@ -507,6 +510,9 @@ bsd_init(void) #endif #endif /* MAC */ + /* Initialize System Override call */ + init_system_override(); + /* * Create process 0. */ @@ -540,8 +546,8 @@ bsd_init(void) LIST_INSERT_HEAD(SESSHASH(0), &session0, s_hash); proc_list_unlock(); -#if CONFIG_LCTX - kernproc->p_lctx = NULL; +#if CONFIG_PERSONAS + kernproc->p_persona = NULL; #endif kernproc->task = kernel_task; @@ -574,9 +580,10 @@ bsd_init(void) bzero(&temp_cred, sizeof(temp_cred)); 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"); /* * We have to label the temp cred before we create from it to @@ -604,11 +611,9 @@ bsd_init(void) #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 */ kernproc->p_fd = &filedesc0; filedesc0.fd_cmask = cmask; filedesc0.fd_knlistsize = -1; @@ -633,7 +638,7 @@ bsd_init(void) 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); @@ -651,7 +656,7 @@ bsd_init(void) &minimum, (vm_size_t)bsd_pageable_map_size, TRUE, - VM_FLAGS_ANYWHERE, + VM_FLAGS_ANYWHERE | VM_MAKE_TAG(VM_KERN_MEMORY_BSD), &bsd_pageable_map); if (ret != KERN_SUCCESS) panic("bsd_init: Failed to allocate bsd pageable map"); @@ -692,6 +697,12 @@ 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(); @@ -751,11 +762,9 @@ bsd_init(void) 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(); + 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. @@ -775,32 +784,41 @@ 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; #if CONFIG_FREEZE - /* Initialise background hibernation */ - bsd_init_kprintf("calling kern_hibernation_init\n"); - kern_hibernation_init(); +#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_EMBEDDED +#if CONFIG_MEMORYSTATUS /* Initialize kernel memory status notifications */ - bsd_init_kprintf("calling kern_memorystatus_init\n"); - kern_memorystatus_init(); -#endif + bsd_init_kprintf("calling memorystatus_init\n"); + memorystatus_init(); +#endif /* CONFIG_MEMORYSTATUS */ + + bsd_init_kprintf("calling macx_init\n"); + macx_init(); + + 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 bsd_autoconf\n"); bsd_autoconf(); @@ -818,6 +836,10 @@ 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 */ @@ -835,22 +857,36 @@ bsd_init(void) 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(); - netsrc_init(); - - /* wait for network domain to finish */ - domainfin(); +#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); @@ -937,8 +973,7 @@ bsd_init(void) #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 */ + microtime_with_abstime(&kernproc->p_start, &kernproc->p_stats->ps_start); #if DEVFS { @@ -948,7 +983,7 @@ bsd_init(void) devfs_kernel_mount(mounthere); } #endif /* DEVFS */ - + /* Initialize signal state for process 0. */ bsd_init_kprintf("calling siginit\n"); siginit(kernproc); @@ -958,18 +993,10 @@ bsd_init(void) #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 */ @@ -980,10 +1007,10 @@ bsd_init(void) consider_zone_gc(FALSE); #endif + bsd_init_kprintf("done\n"); } -/* Called with kernel funnel held */ void bsdinit_task(void) { @@ -1005,30 +1032,15 @@ bsdinit_task(void) ut = (uthread_t)get_bsdthread_info(thread); bsd_init_task = get_threadtask(thread); - init_task_failure_data[0] = 0; + init_task_died = FALSE; #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; - - funnel_state = thread_funnel_set(kernel_flock, TRUE); - - thread_wakeup(&lbolt); - timeout(lightning_bolt,0,hz); - klogwakeup(); - - (void) thread_funnel_set(kernel_flock, FALSE); -} - kern_return_t bsd_autoconf(void) { @@ -1057,11 +1069,6 @@ 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);" @@ -1095,7 +1102,7 @@ bsd_utaskbootstrap(void) * Clone the bootstrap process from the kernel process, without * inheriting either task characteristics or memory from the kernel; */ - thread = cloneproc(TASK_NULL, kernproc, FALSE); + thread = cloneproc(TASK_NULL, COALITION_NULL, kernproc, FALSE, TRUE); /* Hold the reference as it will be dropped during shutdown */ initproc = proc_find(1); @@ -1113,7 +1120,7 @@ bsd_utaskbootstrap(void) ut = (struct uthread *)get_bsdthread_info(thread); ut->uu_sigmask = 0; act_set_astbsd(thread); - (void) thread_resume(thread); + proc_clear_return_wait(initproc, thread); } static void @@ -1122,7 +1129,7 @@ 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))) @@ -1131,9 +1138,15 @@ parse_bsd_args(void) if (PE_parse_boot_argn("-x", namep, sizeof (namep))) /* safe boot */ boothowto |= RB_SAFEBOOT; - /* disable 64 bit grading */ - if (PE_parse_boot_argn("-no64exec", namep, sizeof (namep))) - bootarg_no64exec = 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; + } + /* disable vnode_cache_is_authorized() by setting vnode_cache_defeat */ if (PE_parse_boot_argn("-vnode_cache_defeat", namep, sizeof (namep))) @@ -1149,9 +1162,6 @@ parse_bsd_args(void) 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__) @@ -1166,6 +1176,21 @@ parse_bsd_args(void) 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 PROC_REF_DEBUG + if (PE_parse_boot_argn("-disable_procref_tracking", namep, sizeof(namep))) { + proc_ref_tracking_disabled = 1; + } +#endif + + PE_parse_boot_argn("sigrestrict", &sigrestrict_arg, sizeof(sigrestrict_arg)); } void