#include <debug.h>
#include <xpr_debug.h>
#include <mach_kdp.h>
#include <mach/boolean.h>
#include <mach/machine.h>
#include <mach/thread_act.h>
#include <mach/task_special_ports.h>
#include <mach/vm_param.h>
#include <ipc/ipc_init.h>
#include <kern/assert.h>
#include <kern/mach_param.h>
#include <kern/misc_protos.h>
#include <kern/clock.h>
#include <kern/coalition.h>
#include <kern/cpu_number.h>
#include <kern/cpu_quiesce.h>
#include <kern/ledger.h>
#include <kern/machine.h>
#include <kern/processor.h>
#include <kern/sched_prim.h>
#include <kern/turnstile.h>
#if CONFIG_SCHED_SFI
#include <kern/sfi.h>
#endif
#include <kern/startup.h>
#include <kern/task.h>
#include <kern/thread.h>
#include <kern/timer.h>
#if CONFIG_TELEMETRY
#include <kern/telemetry.h>
#endif
#include <kern/xpr.h>
#include <kern/zalloc.h>
#include <kern/locks.h>
#include <kern/debug.h>
#include <corpses/task_corpse.h>
#include <prng/random.h>
#include <console/serial_protos.h>
#include <vm/vm_kern.h>
#include <vm/vm_init.h>
#include <vm/vm_map.h>
#include <vm/vm_object.h>
#include <vm/vm_page.h>
#include <vm/vm_pageout.h>
#include <vm/vm_shared_region.h>
#include <machine/pmap.h>
#include <machine/commpage.h>
#include <libkern/version.h>
#include <sys/codesign.h>
#include <sys/kdebug.h>
#include <sys/random.h>
#include <sys/ktrace.h>
#include <libkern/section_keywords.h>
#include <kern/ltable.h>
#include <kern/waitq.h>
#include <ipc/ipc_voucher.h>
#include <voucher/ipc_pthread_priority_internal.h>
#include <mach/host_info.h>
#include <pthread/workqueue_internal.h>
#if CONFIG_XNUPOST
#include <tests/ktest.h>
#include <tests/xnupost.h>
#endif
#if CONFIG_ATM
#include <atm/atm_internal.h>
#endif
#if CONFIG_CSR
#include <sys/csr.h>
#endif
#include <bank/bank_internal.h>
#if ALTERNATE_DEBUGGER
#include <arm64/alternate_debugger.h>
#endif
#if MACH_KDP
#include <kdp/kdp.h>
#endif
#if CONFIG_MACF
#include <security/mac_mach_internal.h>
#if CONFIG_VNGUARD
extern void vnguard_policy_init(void);
#endif
#endif
#if KPC
#include <kern/kpc.h>
#endif
#if HYPERVISOR
#include <kern/hv_support.h>
#endif
#include <san/kasan.h>
#if defined(__arm__) || defined(__arm64__)
#include <arm/misc_protos.h> // for arm_vm_prot_finalize
#endif
#include <i386/pmCPU.h>
static void kernel_bootstrap_thread(void);
static void load_context(
thread_t thread);
#if (defined(__i386__) || defined(__x86_64__)) && NCOPY_WINDOWS > 0
extern void cpu_userwindow_init(int);
extern void cpu_physwindow_init(int);
#endif
#if CONFIG_ECC_LOGGING
#include <kern/ecc.h>
#endif
#if (defined(__i386__) || defined(__x86_64__)) && CONFIG_VMX
#include <i386/vmx/vmx_cpu.h>
#endif
#if CONFIG_DTRACE
extern void dtrace_early_init(void);
extern void sdt_early_init(void);
#endif
extern void OSKextRemoveKextBootstrap(void);
void scale_setup(void);
extern void bsd_scale_setup(int);
extern unsigned int semaphore_max;
extern void stackshot_init(void);
extern void ktrace_init(void);
extern void oslog_init(void);
extern int serverperfmode;
unsigned int new_nkdbufs = 0;
unsigned int wake_nkdbufs = 0;
unsigned int write_trace_on_panic = 0;
static char trace_typefilter[64] = { 0 };
unsigned int trace_wrap = 0;
boolean_t trace_serial = FALSE;
boolean_t early_boot_complete = FALSE;
SECURITY_READ_ONLY_LATE(uintptr_t) phys_carveout = 0;
SECURITY_READ_ONLY_LATE(uintptr_t) phys_carveout_pa = 0;
SECURITY_READ_ONLY_LATE(size_t) phys_carveout_size = 0;
int log_leaks = 0;
static inline void
kernel_bootstrap_log(const char *message)
{
kernel_debug_string_early(message);
}
static inline void
kernel_bootstrap_thread_log(const char *message)
{
kernel_debug_string_early(message);
}
void
kernel_early_bootstrap(void)
{
if (PE_parse_boot_argn("serverperfmode", &serverperfmode, sizeof (serverperfmode))) {
serverperfmode = 1;
}
lck_mod_init();
timer_call_init();
#if CONFIG_SCHED_SFI
sfi_early_init();
#endif
}
extern boolean_t IORamDiskBSDRoot(void);
extern kern_return_t cpm_preallocate_early(void);
void
kernel_bootstrap(void)
{
kern_return_t result;
thread_t thread;
char namep[16];
printf("%s\n", version);
if (PE_parse_boot_argn("-l", namep, sizeof (namep)))
log_leaks = 1;
PE_parse_boot_argn("trace", &new_nkdbufs, sizeof (new_nkdbufs));
PE_parse_boot_argn("trace_wake", &wake_nkdbufs, sizeof (wake_nkdbufs));
PE_parse_boot_argn("trace_panic", &write_trace_on_panic, sizeof(write_trace_on_panic));
PE_parse_boot_arg_str("trace_typefilter", trace_typefilter, sizeof(trace_typefilter));
PE_parse_boot_argn("trace_wrap", &trace_wrap, sizeof(trace_wrap));
scale_setup();
kernel_bootstrap_log("vm_mem_bootstrap");
vm_mem_bootstrap();
kernel_bootstrap_log("cs_init");
cs_init();
kernel_bootstrap_log("vm_mem_init");
vm_mem_init();
machine_info.memory_size = (uint32_t)mem_size;
machine_info.max_mem = max_mem;
machine_info.major_version = version_major;
machine_info.minor_version = version_minor;
oslog_init();
#if KASAN
kernel_bootstrap_log("kasan_late_init");
kasan_late_init();
#endif
#if CONFIG_TELEMETRY
kernel_bootstrap_log("telemetry_init");
telemetry_init();
#endif
#if CONFIG_CSR
kernel_bootstrap_log("csr_init");
csr_init();
#endif
if (PE_i_can_has_debugger(NULL)) {
if (PE_parse_boot_argn("-show_pointers", &namep, sizeof(namep))) {
doprnt_hide_pointers = FALSE;
}
if (PE_parse_boot_argn("-no_slto_panic", &namep, sizeof(namep))) {
extern boolean_t spinlock_timeout_panic;
spinlock_timeout_panic = FALSE;
}
}
kernel_bootstrap_log("console_init");
console_init();
kernel_bootstrap_log("stackshot_init");
stackshot_init();
kernel_bootstrap_log("sched_init");
sched_init();
kernel_bootstrap_log("ltable_bootstrap");
ltable_bootstrap();
kernel_bootstrap_log("waitq_bootstrap");
waitq_bootstrap();
kernel_bootstrap_log("ipc_bootstrap");
ipc_bootstrap();
#if CONFIG_MACF
kernel_bootstrap_log("mac_policy_init");
mac_policy_init();
#endif
kernel_bootstrap_log("ipc_init");
ipc_init();
kernel_bootstrap_log("PMAP_ACTIVATE_KERNEL");
PMAP_ACTIVATE_KERNEL(master_cpu);
kernel_bootstrap_log("mapping_free_prime");
mapping_free_prime();
kernel_bootstrap_log("machine_init");
machine_init();
kernel_bootstrap_log("clock_init");
clock_init();
ledger_init();
#if CONFIG_COALITIONS
kernel_bootstrap_log("coalitions_init");
coalitions_init();
#endif
kernel_bootstrap_log("task_init");
task_init();
kernel_bootstrap_log("thread_init");
thread_init();
kernel_bootstrap_log("workq_init");
workq_init();
kernel_bootstrap_log("turnstiles_init");
turnstiles_init();
#if CONFIG_ATM
kernel_bootstrap_log("atm_init");
atm_init();
#endif
kernel_bootstrap_log("mach_init_activity_id");
mach_init_activity_id();
kernel_bootstrap_log("bank_init");
bank_init();
kernel_bootstrap_log("ipc_pthread_priority_init");
ipc_pthread_priority_init();
corpses_init();
host_statistics_init();
kernel_bootstrap_log("kernel_thread_create");
result = kernel_thread_create((thread_continue_t)kernel_bootstrap_thread, NULL, MAXPRI_KERNEL, &thread);
if (result != KERN_SUCCESS) panic("kernel_bootstrap: result = %08X\n", result);
thread->state = TH_RUN;
thread->last_made_runnable_time = mach_absolute_time();
thread_deallocate(thread);
kernel_bootstrap_log("load_context - done");
load_context(thread);
}
int kth_started = 0;
vm_offset_t vm_kernel_addrperm;
vm_offset_t buf_kernel_addrperm;
vm_offset_t vm_kernel_addrperm_ext;
uint64_t vm_kernel_addrhash_salt;
uint64_t vm_kernel_addrhash_salt_ext;
static void
kernel_bootstrap_thread(void)
{
processor_t processor = current_processor();
#define kernel_bootstrap_thread_kprintf(x...)
kernel_bootstrap_thread_log("idle_thread_create");
idle_thread_create(processor);
kernel_bootstrap_thread_log("sched_startup");
sched_startup();
kernel_bootstrap_thread_log("thread_daemon_init");
thread_daemon_init();
vm_kernel_reserved_entry_init();
kernel_bootstrap_thread_log("thread_call_initialize");
thread_call_initialize();
kernel_bootstrap_thread_log("thread_bind");
thread_bind(processor);
#if __arm64__
if (IORamDiskBSDRoot()) {
cpm_preallocate_early();
}
#endif
kernel_bootstrap_thread_log("ipc_thread_call_init");
ipc_thread_call_init();
kernel_bootstrap_thread_log("mapping_adjust");
mapping_adjust();
kernel_bootstrap_thread_log("clock_service_create");
clock_service_create();
device_service_create();
kth_started = 1;
#if (defined(__i386__) || defined(__x86_64__)) && NCOPY_WINDOWS > 0
cpu_physwindow_init(0);
#endif
if (PE_i_can_has_debugger(NULL)) {
unsigned int phys_carveout_mb = 0;
if (PE_parse_boot_argn("phys_carveout_mb", &phys_carveout_mb,
sizeof(phys_carveout_mb)) && phys_carveout_mb > 0) {
phys_carveout_size = phys_carveout_mb * 1024 * 1024;
kern_return_t kr = kmem_alloc_contig(kernel_map,
(vm_offset_t *)&phys_carveout, phys_carveout_size,
VM_MAP_PAGE_MASK(kernel_map), 0, 0, KMA_NOPAGEWAIT,
VM_KERN_MEMORY_DIAG);
if (kr != KERN_SUCCESS) {
kprintf("failed to allocate %uMB for phys_carveout_mb: %u\n",
phys_carveout_mb, (unsigned int)kr);
} else {
phys_carveout_pa = kvtophys((vm_offset_t)phys_carveout);
}
}
}
#if MACH_KDP
kernel_bootstrap_log("kdp_init");
kdp_init();
#endif
#if ALTERNATE_DEBUGGER
alternate_debugger_init();
#endif
#if KPC
kpc_init();
#endif
#if CONFIG_ECC_LOGGING
ecc_log_init();
#endif
#if HYPERVISOR
hv_support_init();
#endif
#if CONFIG_TELEMETRY
kernel_bootstrap_log("bootprofile_init");
bootprofile_init();
#endif
#if (defined(__i386__) || defined(__x86_64__)) && CONFIG_VMX
vmx_init();
#endif
kernel_bootstrap_thread_log("ktrace_init");
ktrace_init();
kdebug_init(new_nkdbufs, trace_typefilter, trace_wrap);
#ifdef MACH_BSD
kernel_bootstrap_log("bsd_early_init");
bsd_early_init();
#endif
#if defined(__arm64__)
ml_lockdown_init();
#endif
#ifdef IOKIT
kernel_bootstrap_log("PE_init_iokit");
PE_init_iokit();
#endif
assert(ml_get_interrupts_enabled() == FALSE);
early_boot_complete = TRUE;
#if INTERRUPT_MASKED_DEBUG
ml_spin_debug_clear_self();
#endif
(void) spllo();
#if (defined(__i386__) || defined(__x86_64__)) && NCOPY_WINDOWS > 0
cpu_userwindow_init(0);
#endif
vm_shared_region_init();
vm_commpage_init();
vm_commpage_text_init();
#if CONFIG_MACF
kernel_bootstrap_log("mac_policy_initmach");
mac_policy_initmach();
#if CONFIG_VNGUARD
vnguard_policy_init();
#endif
#endif
#if CONFIG_DTRACE
dtrace_early_init();
sdt_early_init();
#endif
OSKextRemoveKextBootstrap();
#if defined(__arm__) || defined(__arm64__)
#if CONFIG_KERNEL_INTEGRITY
machine_lockdown_preflight();
#endif
arm_vm_prot_finalize(PE_state.bootArgs);
#endif
#if CONFIG_SCHED_SFI
kernel_bootstrap_log("sfi_init");
sfi_init();
#endif
read_random(&vm_kernel_addrperm, sizeof(vm_kernel_addrperm));
vm_kernel_addrperm |= 1;
read_random(&buf_kernel_addrperm, sizeof(buf_kernel_addrperm));
buf_kernel_addrperm |= 1;
read_random(&vm_kernel_addrperm_ext, sizeof(vm_kernel_addrperm_ext));
vm_kernel_addrperm_ext |= 1;
read_random(&vm_kernel_addrhash_salt, sizeof(vm_kernel_addrhash_salt));
read_random(&vm_kernel_addrhash_salt_ext, sizeof(vm_kernel_addrhash_salt_ext));
vm_set_restrictions();
#ifdef CONFIG_XNUPOST
kern_return_t result = kernel_list_tests();
result = kernel_do_post();
if (result != KERN_SUCCESS) {
panic("kernel_do_post: Tests failed with result = 0x%08x\n", result);
}
kernel_bootstrap_log("kernel_do_post - done");
#endif
#ifdef MACH_BSD
bsd_init();
#endif
kdebug_free_early_buf();
serial_keyboard_init();
vm_page_init_local_q();
thread_bind(PROCESSOR_NULL);
vm_pageout();
}
void
slave_main(void *machine_param)
{
processor_t processor = current_processor();
thread_t thread;
if (processor->next_thread == THREAD_NULL) {
thread = processor->idle_thread;
thread->continuation = (thread_continue_t)processor_start_thread;
thread->parameter = machine_param;
}
else {
thread = processor->next_thread;
processor->next_thread = THREAD_NULL;
}
load_context(thread);
}
void
processor_start_thread(void *machine_param)
{
processor_t processor = current_processor();
thread_t self = current_thread();
slave_machine_init(machine_param);
if (self == processor->idle_thread)
thread_block((thread_continue_t)idle_thread);
thread_terminate(self);
}
static void __attribute__((noreturn))
load_context(
thread_t thread)
{
processor_t processor = current_processor();
#define load_context_kprintf(x...)
load_context_kprintf("machine_set_current_thread\n");
machine_set_current_thread(thread);
load_context_kprintf("processor_up\n");
processor_up(processor);
PMAP_ACTIVATE_KERNEL(processor->cpu_id);
load_context_kprintf("thread %p, stack %lx, stackptr %lx\n", thread,
thread->kernel_stack, thread->machine.kstackptr);
if (!thread->kernel_stack) {
load_context_kprintf("stack_alloc_try\n");
if (!stack_alloc_try(thread))
panic("load_context");
}
if (!(thread->state & TH_IDLE))
sched_run_incr(thread);
processor->active_thread = thread;
processor_state_update_explicit(processor, thread->sched_pri,
SFI_CLASS_KERNEL, PSET_SMP, thread_get_perfcontrol_class(thread));
processor->starting_pri = thread->sched_pri;
processor->deadline = UINT64_MAX;
thread->last_processor = processor;
processor->last_dispatch = mach_absolute_time();
timer_start(&thread->system_timer, processor->last_dispatch);
PROCESSOR_DATA(processor, thread_timer) = PROCESSOR_DATA(processor, kernel_timer) = &thread->system_timer;
timer_start(&PROCESSOR_DATA(processor, system_state), processor->last_dispatch);
PROCESSOR_DATA(processor, current_state) = &PROCESSOR_DATA(processor, system_state);
cpu_quiescent_counter_join(processor->last_dispatch);
PMAP_ACTIVATE_USER(thread, processor->cpu_id);
load_context_kprintf("machine_load_context\n");
machine_load_context(thread);
}
void
scale_setup()
{
int scale = 0;
#if defined(__LP64__)
typeof(task_max) task_max_base = task_max;
if ((serverperfmode != 0) && ((uint64_t)sane_size >= (uint64_t)(16 * 1024 * 1024 *1024ULL))) {
scale = (int)((uint64_t)sane_size / (uint64_t)(8 * 1024 * 1024 *1024ULL));
if (scale > 16)
scale = 16;
task_max_base = 2500;
} else if ((uint64_t)sane_size >= (uint64_t)(3 * 1024 * 1024 *1024ULL)) {
if ((uint64_t)sane_size < (uint64_t)(8 * 1024 * 1024 *1024ULL)) {
scale = 2;
} else {
scale = MIN(16, (int)((uint64_t)sane_size / (uint64_t)(4 * 1024 * 1024 *1024ULL)));
}
}
task_max = MAX(task_max, task_max_base * scale);
if (scale != 0) {
task_threadmax = task_max;
thread_max = task_max * 5;
}
#endif
bsd_scale_setup(scale);
ipc_space_max = SPACE_MAX;
ipc_port_max = PORT_MAX;
ipc_pset_max = SET_MAX;
semaphore_max = SEMAPHORE_MAX;
}