]> git.saurik.com Git - apple/xnu.git/blobdiff - osfmk/kern/startup.c
xnu-3247.10.11.tar.gz
[apple/xnu.git] / osfmk / kern / startup.c
index cfcef75424f42eb89a62d352c713bbb7b86b4a75..53013fa79ef7dfa820f3459e6254ebfc29764777 100644 (file)
@@ -1,31 +1,29 @@
 /*
- * Copyright (c) 2000-2005 Apple Computer, Inc. All rights reserved.
+ * Copyright (c) 2000-2010 Apple Inc. All rights reserved.
  *
- * @APPLE_LICENSE_OSREFERENCE_HEADER_START@
+ * @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.  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 
- * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, 
- * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, 
- * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. 
- * Please see the License for the specific language governing rights and 
+ * 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. 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
+ * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
+ * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
+ * Please see the License for the specific language governing rights and
  * limitations under the License.
- *
- * @APPLE_LICENSE_OSREFERENCE_HEADER_END@
+ * 
+ * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  */
 /*
  * @OSF_COPYRIGHT@
  * any improvements or extensions that they make and grant Carnegie Mellon
  * the rights to redistribute these changes.
  */
+/*
+ * 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.
+ */
 /*
  */
 
@@ -65,8 +69,6 @@
 #include <debug.h>
 #include <xpr_debug.h>
 #include <mach_kdp.h>
-#include <mach_host.h>
-#include <norma_vm.h>
 
 #include <mach/boolean.h>
 #include <mach/machine.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/ledger.h>
 #include <kern/machine.h>
 #include <kern/processor.h>
 #include <kern/sched_prim.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_shared_memory_server.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 <kern/waitq.h>
+
+
+#if CONFIG_ATM
+#include <atm/atm_internal.h>
+#endif
+
+#if CONFIG_CSR
+#include <sys/csr.h>
+#endif
+
+#if CONFIG_BANK
+#include <bank/bank_internal.h>
+#endif
+
+#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>
+#endif
 
-#ifdef __ppc__
-#include <ppc/Firmware.h>
-#include <ppc/mappings.h>
+#if KPC
+#include <kern/kpc.h>
 #endif
 
+#if KPERF
+#include <kperf/kperf.h>
+#endif
+
+#if HYPERVISOR
+#include <kern/hv_support.h>
+#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
 
-#ifdef i386
-extern void    cpu_window_init(int);
+#if CONFIG_ECC_LOGGING
+#include <kern/ecc.h>
+#endif 
+
+#if (defined(__i386__) || defined(__x86_64__)) && CONFIG_VMX
+#include <i386/vmx/vmx_cpu.h>
 #endif
 
+// libkern/OSKextLib.cpp
+extern void    OSKextRemoveKextBootstrap(void);
+
+void scale_setup(void);
+extern void bsd_scale_setup(int);
+extern unsigned int semaphore_max;
+extern void stackshot_lock_init(void);
 
 /*
  *     Running in virtual memory, on the interrupt stack.
  */
 
+extern int serverperfmode;
+
+/* size of kernel trace buffer, disabled by default */
+unsigned int new_nkdbufs = 0;
+unsigned int wake_nkdbufs = 0;
+unsigned int write_trace_on_panic = 0;
+unsigned int trace_typefilter = 0;
+boolean_t    trace_serial = FALSE;
+
+/* mach leak logging */
+int log_leaks = 0;
+int turn_on_log_leaks = 0;
+
+static inline void
+kernel_bootstrap_log(const char *message)
+{
+//     kprintf("kernel_bootstrap: %s\n", message);
+       kernel_debug_string_simple(message);
+}
+
+static inline void
+kernel_bootstrap_thread_log(const char *message)
+{
+//     kprintf("kernel_bootstrap_thread: %s\n", message);
+       kernel_debug_string_simple(message);
+}
+
+void
+kernel_early_bootstrap(void)
+{
+       /* serverperfmode is needed by timer setup */
+        if (PE_parse_boot_argn("serverperfmode", &serverperfmode, sizeof (serverperfmode))) {
+                serverperfmode = 1;
+        }
+
+       lck_mod_init();
+
+       /*
+        * Initialize the timer callout world
+        */
+       timer_call_init();
+
+#if CONFIG_SCHED_SFI
+       /*
+        * Configure SFI classes
+        */
+       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;
+       thread_t        thread;
+       char            namep[16];
 
-       lck_mod_init();
-       sched_init();
+       printf("%s\n", version); /* log kernel version */
+
+       if (PE_parse_boot_argn("-l", namep, sizeof (namep))) /* leaks logging */
+               turn_on_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_argn("trace_typefilter", &trace_typefilter, sizeof(trace_typefilter));
+
+       scale_setup();
+
+       kernel_bootstrap_log("vm_mem_bootstrap");
        vm_mem_bootstrap();
-       ipc_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;
+
+
+#if CONFIG_TELEMETRY
+       kernel_bootstrap_log("telemetry_init");
+       telemetry_init();
+#endif
+
+#if CONFIG_CSR
+       kernel_bootstrap_log("csr_init");
+       csr_init();
+#endif
+
+       kernel_bootstrap_log("stackshot_lock_init");    
+       stackshot_lock_init();
+
+       kernel_bootstrap_log("sched_init");
+       sched_init();
+
+       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();
 
        /*
         * As soon as the virtual memory system is up, we record
         * that this CPU is using the kernel pmap.
         */
+       kernel_bootstrap_log("PMAP_ACTIVATE_KERNEL");
        PMAP_ACTIVATE_KERNEL(master_cpu);
 
+       kernel_bootstrap_log("mapping_free_prime");
        mapping_free_prime();                                           /* Load up with temporary mapping blocks */
 
+       kernel_bootstrap_log("machine_init");
        machine_init();
-       kmod_init();
+
+       kernel_bootstrap_log("clock_init");
        clock_init();
 
-       machine_info.memory_size = mem_size;
-       machine_info.max_mem = max_mem;
-       machine_info.major_version = version_major;
-       machine_info.minor_version = version_minor;
+       ledger_init();
 
        /*
         *      Initialize the IPC, task, and thread subsystems.
         */
-       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();
+
+#if CONFIG_ATM
+       /* Initialize the Activity Trace Resource Manager. */
+       kernel_bootstrap_log("atm_init");
+       atm_init();
+#endif
+
+#if CONFIG_BANK
+       /* Initialize the BANK Manager. */
+       kernel_bootstrap_log("bank_init");
+       bank_init();
+#endif
        
+       /* initialize the corpse config based on boot-args */
+       corpses_init();
+
        /*
         *      Create a kernel thread to execute the kernel bootstrap.
         */
+       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);
        /*NOTREACHED*/
 }
 
+int kth_started = 0;
+
+vm_offset_t vm_kernel_addrperm;
+vm_offset_t buf_kernel_addrperm;
+vm_offset_t vm_kernel_addrperm_ext;
+
 /*
  * Now running in a thread.  Kick off other services,
  * invoke user bootstrap, enter pageout loop.
@@ -179,8 +382,9 @@ static void
 kernel_bootstrap_thread(void)
 {
        processor_t             processor = current_processor();
-       thread_t                self = current_thread();
 
+#define kernel_bootstrap_thread_kprintf(x...) /* kprintf("kernel_bootstrap_thread: " x) */
+       kernel_bootstrap_thread_log("idle_thread_create");
        /*
         * Create the idle processor thread.
         */
@@ -192,22 +396,47 @@ kernel_bootstrap_thread(void)
         *
         * Start up the scheduler services.
         */
+       kernel_bootstrap_thread_log("sched_startup");
        sched_startup();
 
+       /*
+        * Thread lifecycle maintenance (teardown, stack allocation)
+        */
+       kernel_bootstrap_thread_log("thread_daemon_init");
+       thread_daemon_init();
+
+       /* Create kernel map entry reserve */
+       vm_kernel_reserved_entry_init();
+
+       /*
+        * Thread callout service.
+        */
+       kernel_bootstrap_thread_log("thread_call_initialize");
+       thread_call_initialize();
+
        /*
         * Remain on current processor as
         * additional processors come online.
         */
-       thread_bind(self, processor);
+       kernel_bootstrap_thread_log("thread_bind");
+       thread_bind(processor);
+
+       /*
+        * Initialize ipc thread call support.
+        */
+       kernel_bootstrap_thread_log("ipc_thread_call_init");
+       ipc_thread_call_init();
 
        /*
         * Kick off memory mapping adjustments.
         */
+       kernel_bootstrap_thread_log("mapping_adjust");
        mapping_adjust();
 
        /*
         *      Create the clock service.
         */
+       kernel_bootstrap_thread_log("clock_service_create");
        clock_service_create();
 
        /*
@@ -215,41 +444,159 @@ kernel_bootstrap_thread(void)
         */
        device_service_create();
 
-       shared_file_boot_time_init(ENV_DEFAULT_ROOT, cpu_type());
+       kth_started = 1;
+               
+#if (defined(__i386__) || defined(__x86_64__)) && NCOPY_WINDOWS > 0
+       /*
+        * Create and initialize the physical copy window for processor 0
+        * This is required before starting kicking off  IOKit.
+        */
+       cpu_physwindow_init(0);
+#endif
 
-#ifdef IOKIT
-       {
-               PE_init_iokit();
+
+       
+#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 KPERF
+       kperf_bootstrap();
+#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
+
+#if (defined(__i386__) || defined(__x86_64__))
+       if (kdebug_serial) {
+               new_nkdbufs = 1;
+               if (trace_typefilter == 0)
+                       trace_typefilter = 1;
        }
+       if (turn_on_log_leaks && !new_nkdbufs)
+               new_nkdbufs = 200000;
+       if (trace_typefilter)
+               start_kern_tracing_with_typefilter(new_nkdbufs,
+                                                  FALSE,
+                                                  trace_typefilter);
+       else
+               start_kern_tracing(new_nkdbufs, FALSE);
+       if (turn_on_log_leaks)
+               log_leaks = 1;
+
 #endif
-       
+
+       kernel_bootstrap_log("prng_init");
+       prng_cpu_init(master_cpu);
+
+#ifdef IOKIT
+       PE_init_iokit();
+#endif
+
+       assert(ml_get_interrupts_enabled() == FALSE);
        (void) spllo();         /* Allow interruptions */
 
+#if (defined(__i386__) || defined(__x86_64__)) && NCOPY_WINDOWS > 0
        /*
-        *      Fill in the comm area (mapped into every task address space.)
+        * Create and initialize the copy window for processor 0
+        * This also allocates window space for all other processors.
+        * However, this is dependent on the number of processors - so this call
+        * must be after IOKit has been started because IOKit performs processor
+        * discovery.
         */
-       commpage_populate();
+       cpu_userwindow_init(0);
+#endif
+
+#if (!defined(__i386__) && !defined(__x86_64__))
+       if (turn_on_log_leaks && !new_nkdbufs)
+               new_nkdbufs = 200000;
+       if (trace_typefilter)
+               start_kern_tracing_with_typefilter(new_nkdbufs, FALSE, trace_typefilter);
+       else
+               start_kern_tracing(new_nkdbufs, FALSE);
+       if (turn_on_log_leaks)
+               log_leaks = 1;
+#endif
 
-#ifdef i386
        /*
-        *      create and initialize a copy window
-        *      for processor 0
+        *      Initialize the shared region module.
         */
-       cpu_window_init(0);
+       vm_shared_region_init();
+       vm_commpage_init();
+       vm_commpage_text_init();
+
+
+#if CONFIG_MACF
+       kernel_bootstrap_log("mac_policy_initmach");
+       mac_policy_initmach();
+#endif
+
+#if CONFIG_SCHED_SFI
+       kernel_bootstrap_log("sfi_init");
+       sfi_init();
 #endif
 
+       /*
+        * Initialize the globals used for permuting kernel
+        * addresses that may be exported to userland as tokens
+        * using VM_KERNEL_ADDRPERM()/VM_KERNEL_ADDRPERM_EXTERNAL().
+        * Force the random number to be odd to avoid mapping a non-zero
+        * word-aligned address to zero via addition.
+        * Note: at this stage we can use the cryptographically secure PRNG
+        * rather than early_random().
+        */
+       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;
+
+       vm_set_restrictions();
+
+
+
        /*
         *      Start the user bootstrap.
         */
 #ifdef MACH_BSD
-       { 
-               bsd_init();
-       }
+       bsd_init();
 #endif
 
+    /*
+     * Get rid of segments used to bootstrap kext loading. This removes
+     * the KLD, PRELINK symtab, LINKEDIT, and symtab segments/load commands.
+     */
+       OSKextRemoveKextBootstrap();
+
        serial_keyboard_init();         /* Start serial keyboard if wanted */
 
-       thread_bind(self, PROCESSOR_NULL);
+       vm_page_init_local_q();
+
+       thread_bind(PROCESSOR_NULL);
 
        /*
         *      Become the pageout daemon.
@@ -264,7 +611,7 @@ kernel_bootstrap_thread(void)
  *     Load the first thread to start a processor.
  */
 void
-slave_main(void)
+slave_main(void *machine_param)
 {
        processor_t             processor = current_processor();
        thread_t                thread;
@@ -276,7 +623,7 @@ slave_main(void)
        if (processor->next_thread == THREAD_NULL) {
                thread = processor->idle_thread;
                thread->continuation = (thread_continue_t)processor_start_thread;
-               thread->parameter = NULL;
+               thread->parameter = machine_param;
        }
        else {
                thread = processor->next_thread;
@@ -295,12 +642,12 @@ slave_main(void)
  *     Called at splsched.
  */
 void
-processor_start_thread(void)
+processor_start_thread(void *machine_param)
 {
        processor_t             processor = current_processor();
        thread_t                self = current_thread();
 
-       slave_machine_init();
+       slave_machine_init(machine_param);
 
        /*
         *      If running the idle processor thread,
@@ -324,17 +671,26 @@ load_context(
 {
        processor_t             processor = current_processor();
 
+
+#define load_context_kprintf(x...) /* kprintf("load_context: " 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_DATA(processor, slot_num));
+       PMAP_ACTIVATE_KERNEL(processor->cpu_id);
 
        /*
         * Acquire a stack if none attached.  The panic
         * should never occur since the thread is expected
         * to have reserved stack.
         */
+       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");
        }
@@ -344,19 +700,59 @@ load_context(
         * running for load calculations.
         */
        if (!(thread->state & TH_IDLE))
-               pset_run_incr(thread->processor_set);
+               sched_run_incr(thread);
 
        processor->active_thread = thread;
        processor->current_pri = thread->sched_pri;
+       processor->current_thmode = thread->sched_mode;
        processor->deadline = UINT64_MAX;
        thread->last_processor = processor;
 
        processor->last_dispatch = mach_absolute_time();
-       timer_switch((uint32_t)processor->last_dispatch,
-                                                       &PROCESSOR_DATA(processor, offline_timer));
+       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);
 
-       PMAP_ACTIVATE_USER(thread, PROCESSOR_DATA(processor, slot_num));
+       PMAP_ACTIVATE_USER(thread, processor->cpu_id);
 
+       load_context_kprintf("machine_load_context\n");
        machine_load_context(thread);
        /*NOTREACHED*/
 }
+
+void
+scale_setup()
+{
+       int scale = 0;
+#if defined(__LP64__)
+       typeof(task_max) task_max_base = task_max;
+
+       /* Raise limits for servers with >= 16G */
+       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));
+               /* limit to 128 G */
+               if (scale > 16)
+                       scale = 16;
+               task_max_base = 2500;
+       } else if ((uint64_t)sane_size >= (uint64_t)(3 * 1024 * 1024 *1024ULL))
+               scale = 2;
+
+       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;
+}
+