zhangyang-variable-monitor/source/module/monitor_kernel_lib.c

#include "monitor_kernel.h"

#include <linux/interrupt.h>
#include <linux/sched.h>
#include <linux/stacktrace.h>
// #include <linux/sched/task.h>
// #include <linux/sched/mm.h>

#define __task_contributes_to_load(task)                                       \
  ((READ_ONCE(task->__state) & TASK_UNINTERRUPTIBLE) != 0 &&                   \
   (task->flags & PF_FROZEN) == 0 &&                                           \
   (READ_ONCE(task->__state) & TASK_NOLOAD) == 0)

/**
 * @brief watch_arg to kernel_watch_arg
 *
 * @param ptr: kernel space address
 * @param warg: watch_arg
 * @param k_watch_arg: kernel_watch_arg
 * @return unsigned char
 */
static unsigned char w_arg2k_w_arg(void *kptr, watch_arg warg,
                                   kernel_watch_arg *k_watch_arg) {
  // k_watch_arg init
  k_watch_arg->task_id = warg.task_id;
  strncpy(k_watch_arg->name, warg.name, MAX_NAME_LEN + 1); // name
  k_watch_arg->name[MAX_NAME_LEN + 1] = '\0';              // just in case
  k_watch_arg->ptr = warg.ptr;
  k_watch_arg->kptr = kptr;
  k_watch_arg->length_byte = warg.length_byte;
  k_watch_arg->threshold = warg.threshold;
  k_watch_arg->is_unsigned = warg.is_unsigned;
  k_watch_arg->above_threshold = warg.above_threshold;
  k_watch_arg->true_value = 0;
  return 0;
}

static long long convert_to_longlong(void *ptr, int size, char isUnsigned) {
  long long ret = 0;
  // ptr is null
  if (!ptr) {
    return 0;
  }
  switch (size) {
  case 1: // 8-bit integer.
    ret = isUnsigned ? (*(unsigned char *)ptr) : (*(char *)ptr);
    break;
  case 2: // 16-bit integer.
    ret = isUnsigned ? (*(unsigned short *)ptr) : (*(short *)ptr);
    break;
  case 4: // 32-bit integer.
    ret = isUnsigned ? (*(unsigned int *)ptr) : (*(int *)ptr);
    break;
  case 8:
    ret = isUnsigned ? (*(unsigned long long *)ptr) : (*(long long *)ptr);
    break;
  default:
    ret = 0;
    break;
  }
  return ret;
}

/**
 * @brief kernel_watch_arg to threshold
 *
 * @param k_watch_arg
 * @param threshold
 */
static void k_w_arg2threshold(kernel_watch_arg *k_watch_arg,
                              threshold *threshold) {
  threshold->task_id = k_watch_arg->task_id;
  strncpy(threshold->name, k_watch_arg->name, MAX_NAME_LEN + 1);
  threshold->name[MAX_NAME_LEN + 1] = '\0';
  threshold->ptr = k_watch_arg->ptr;
  threshold->threshold = k_watch_arg->threshold;
  // read true value
  threshold->true_value = k_watch_arg->true_value;
}

static void init_mm_tree(mm_tree *mm_tree) {
  INIT_RADIX_TREE(&mm_tree->mm_tree, GFP_ATOMIC);
  spin_lock_init(&mm_tree->mm_tree_lock);
}

/**
 * @brief init buffer
 *
 * @param buf_size
 * @param buffer
 * @return int
 */
static int init_buffer(unsigned int buf_size,
                       struct diag_variant_buffer *buffer) {
  init_mm_tree(&mm_tree_struct); // init mm_tree
  init_diag_variant_buffer(buffer, buf_size);
  int ret = 0;
  ret = alloc_diag_variant_buffer(buffer);
  return ret;
}

// init load_monitor_variant_buffer
static int init_global_buffer(void) {
  return init_buffer(VARIABLE_MONITOR_BUFFER_SIZE,
                     &load_monitor_variant_buffer);
}
// init stand_alone_buffer
static int init_sa_buffer(void) {
  return init_buffer(STAND_ALONE_BUFFER_SIZE, &stand_alone_buffer);
}

/**
 * @brief diag task info | brief | kernel stack | proc chains | raw
 * stack
 *
 * @param p
 * @param tsk_info
 */
static void diag_tsk(struct task_struct *p, variable_monitor_task *tsk_info) {
  unsigned int nr_bt;
  // printk(KERN_INFO "diag_tsk\n");
  diag_task_brief(p, &tsk_info->task);                      // task brief
  if (tsk_info->task.sys_task == 1) {                       // system task
    nr_bt = diag_task_kern_stack(p, &tsk_info->kern_stack); // kernel stack
    dump_proc_chains_argv(1, p, &mm_tree_struct,
                          &tsk_info->proc_chains);          // proc chains
  } else {                                                  // other task
    nr_bt = diag_task_kern_stack(p, &tsk_info->kern_stack); // kernel stack
    dump_proc_chains_argv(1, p, &mm_tree_struct,
                          &tsk_info->proc_chains); // proc chains
    diag_task_raw_stack(p, &tsk_info->raw_stack);  // raw stack
  }
}

static void push_tskinfo_2_buffer_orig(variable_monitor_task *tsk_info,
                                       unsigned long *flags,
                                       struct diag_variant_buffer *buffer) {
  // printk(KERN_INFO "push_tsk_info\n");
  diag_variant_buffer_spin_lock(buffer, *flags);
  diag_variant_buffer_reserve(buffer, sizeof(variable_monitor_task));
  diag_variant_buffer_write_nolock(buffer, tsk_info,
                                   sizeof(variable_monitor_task));
  diag_variant_buffer_seal(buffer);
  diag_variant_buffer_spin_unlock(buffer, *flags);
}

/**
 * @brief push task info to global buffer
 *
 * @param tsk_info
 * @param flags
 */
// static void push_tskinfo_2_buffer(variable_monitor_task *tsk_info,
//                                   unsigned long *flags) {
//   push_tskinfo_2_buffer_orig(tsk_info, flags, &load_monitor_variant_buffer);
// }

/**
 * @brief push task info to stand_alone_buffer
 *
 * @param tsk_info
 * @param flags
 */
static void push_tskinfo_2_sa_buffer(variable_monitor_task *tsk_info,
                                     unsigned long *flags) {
  push_tskinfo_2_buffer_orig(tsk_info, flags, &stand_alone_buffer);
}

/**
 * @brief push user/sys task info to global buffer
 *
 * @param tsk_info
 * @param flags
 */
static void push_tskinfo_22_buffer_orig(variable_monitor_task *tsk_info,
                                   unsigned long *flags,
                                   struct diag_variant_buffer *buffer) {
  variable_monitor_task_system *tsk_info_system;
  if (tsk_info->task.sys_task == 1) // system task
  {
    tsk_info_system = (variable_monitor_task_system *)tsk_info;
    tsk_info_system->et_type = VARIABLE_MONITOR_TASK_TYPE_SYSTEM;
    diag_variant_buffer_reserve(buffer, sizeof(variable_monitor_task_system));
    diag_variant_buffer_write_nolock(buffer, tsk_info_system,
                                     sizeof(variable_monitor_task_system));
    diag_variant_buffer_seal(buffer);
  } else {
    diag_variant_buffer_reserve(buffer, sizeof(variable_monitor_task));
    diag_variant_buffer_write_nolock(buffer, tsk_info,
                                     sizeof(variable_monitor_task));
    diag_variant_buffer_seal(buffer);
  }
}

static void push_tskinfo_22_buffer(variable_monitor_task *tsk_info,
                                   unsigned long *flags) {
  push_tskinfo_22_buffer_orig(tsk_info, flags, &load_monitor_variant_buffer);
}

// static void push_tskinfo_22_sa_buffer(variable_monitor_task *tsk_info,
//                                       unsigned long *flags) {
//   push_tskinfo_22_buffer_orig(tsk_info, flags, &stand_alone_buffer);
// }

/// @brief clear all watch and reset kernel_wtimer_list/kernel_wtimer_num
/// @param
static void clear_all_watch(void) {
  printk(KERN_INFO "clear all watch variable\n");
  // cancel timer
  cancel_all_hrTimer();
  // stop and destory work
  cancel_destory_all_work();
  // unmap and release the page
  free_all_page_list();
  // clear timer
  kernel_wtimer_num = 0;
  memset(kernel_wtimer_list, 0, sizeof(kernel_wtimer_list));
}

/**
 * @brief all threshold reached info
 *
 * @param k_watch_timer
 * @param is_print
 */
static void diag_vm_record(kernel_watch_timer *k_watch_timer,
                           unsigned char is_print) {

  unsigned long long start_time = ktime_get_real();

  static variable_monitor_record vm_record;

  kernel_watch_arg *kwarg;
  int i;
  unsigned long flags;

  unsigned long event_id = get_cycles();

  vm_record.id = event_id;
  vm_record.et_type = VARIABLE_MONITOR_RECORD_TYPE;
  vm_record.tv = k_watch_timer->tv;
  vm_record.threshold_over_count = k_watch_timer->threshold_over_count;

  for (i = 0; i < vm_record.threshold_over_count; i++) {
    kwarg = &k_watch_timer->k_watch_args[k_watch_timer->threshold_buffer[i]];
    k_w_arg2threshold(kwarg, &vm_record.threshold_record[i]);
  }
  rcu_read_lock();
  diag_variant_buffer_spin_lock(&load_monitor_variant_buffer, flags);

  diag_variant_buffer_reserve(&load_monitor_variant_buffer,
                              sizeof(variable_monitor_record));
  diag_variant_buffer_write_nolock(&load_monitor_variant_buffer, &vm_record,
                                   sizeof(variable_monitor_record));
  diag_variant_buffer_seal(&load_monitor_variant_buffer);

  diag_variant_buffer_spin_unlock(&load_monitor_variant_buffer, flags);
  rcu_read_unlock();

  if (is_print) {
    printk(KERN_INFO "-----------variable monitor----------\n");
    printk(KERN_INFO
           "threshold exceeded, Timestamp %lld, Stack finish Delay %lld:\n",
           vm_record.tv, start_time - vm_record.tv);

    for (i = 0; i < vm_record.threshold_over_count; i++) {
      printk(
          KERN_INFO
          "\t: pid: %d, name: %s, ptr: %p, threshold:%lld, true_value:%lld\n",
          vm_record.threshold_record[i].task_id,
          vm_record.threshold_record[i]
              .name, // Assuming name is a null-terminated string
          vm_record.threshold_record[i].ptr,
          vm_record.threshold_record[i].threshold,
          vm_record.threshold_record[i].true_value);
    }

    printk(KERN_INFO "-------------------------------------\n");
  }
}

/**
 * @brief diag task by tgid
 *
 * @param tgid
 */
void diag_task_by_tgid_orig(pid_t tgid, struct diag_variant_buffer *buffer) {

  struct task_struct *tsk;
  int ret;
  unsigned long flags;

  static variable_monitor_task tsk_info;

  tsk = NULL;
  rcu_read_lock();
  tsk = NULL;
  if (orig_find_task_by_vpid)
    tsk = orig_find_task_by_vpid(tgid);
  if (!tsk) {
    ret = -EINVAL;
    rcu_read_unlock();
    return;
  }

  diag_variant_buffer_spin_lock(buffer, flags);

  struct task_struct *thread = tsk;
  unsigned long event_id = get_cycles();

  while_each_thread(tsk, thread) {

    tsk_info.et_type = VARIABLE_MONITOR_TASK_TYPE;
    tsk_info.id = event_id;
    tsk_info.tv = ktime_get_real();
    diag_tsk(tsk, &tsk_info);

    push_tskinfo_22_buffer_orig(&tsk_info, &flags, buffer); // push to buffer
  }
  diag_variant_buffer_spin_unlock(buffer, flags);
  rcu_read_unlock();
}

void diag_task_by_tgid(pid_t tgid){
  diag_task_by_tgid_orig(tgid, &load_monitor_variant_buffer);
}

void diag_task_sa_by_tgid(pid_t tgid){
  diag_task_by_tgid_orig(tgid, &stand_alone_buffer);
}

/**
 * @brief diag all task info
 *
 */
void diag_task_all(void) {
  static variable_monitor_task tsk_info;

  unsigned long event_id = get_cycles();
  struct task_struct *g, *p; // g: task group; p: task
  unsigned long flags;

  rcu_read_lock();
  diag_variant_buffer_spin_lock(&load_monitor_variant_buffer, flags);

  do_each_thread(g, p) {
    if (p->__state == TASK_RUNNING || __task_contributes_to_load(p) ||
        ((READ_ONCE(p->__state) & TASK_IDLE) != 0)) {

      // get_task_struct(p); // count +1

      tsk_info.et_type = VARIABLE_MONITOR_TASK_TYPE;
      tsk_info.id = event_id;
      tsk_info.tv = ktime_get_real();
      diag_tsk(p, &tsk_info);

      // put_task_struct(p); // count -1

      push_tskinfo_22_buffer(&tsk_info, &flags); // push to buffer
    }
  }
  while_each_thread(g, p);

  diag_variant_buffer_spin_unlock(&load_monitor_variant_buffer, flags);
  rcu_read_unlock();
}

/**
 * @brief diag task entry
 *
 * @param k_watch_timer
 */
void diag_task(kernel_watch_timer *k_watch_timer) {
  if (k_watch_timer->threshold_over_count <= 0) // if no threshold reached
    return;
  pr_info("diag_stack, tv %lld\n", ktime_get_real());
  if (sample_all) { // sample all task
    diag_task_all();
  } else { // only care threshold reached task
    int i;
    kernel_watch_arg *kwarg;
    for (i = 0; i < k_watch_timer->threshold_over_count; i++) {
      kwarg = &k_watch_timer->k_watch_args[k_watch_timer->threshold_buffer[i]];
      diag_task_by_tgid_orig(kwarg->task_id, &load_monitor_variant_buffer);
    }
  }
  pr_info("diag_stack, finish tv %lld\n", ktime_get_real());
  diag_vm_record(k_watch_timer, 1);
}

/**
 * @brief diag task, for work queue
 *
 * @param work
 */
void diag_task_info_work(struct work_struct *work) {
  kernel_watch_timer *k_watch_timer =
      container_of(work, kernel_watch_timer, wk);
  diag_task(k_watch_timer);
}

/**
 * @brief all module function init. orig_X | buffer
 *
 * @return int
 */
int monitor_init(void) {
  int ret = 0;

  ret = init_orig_fun(); // init orig_X
  if (ret)
    return ret;
  ret = init_global_buffer(); // 256M
  if (ret)
    return -1;

  ret = init_sa_buffer(); // 50M
  if (ret)
    return -1;
  return 0;
}

/**
 * @brief monitor exit: clear all watch and free buffer
 *
 */
void monitor_exit(void) {
  // clear all watch
  clear_all_watch();
  // free buffer
  destroy_diag_variant_buffer(&load_monitor_variant_buffer);
  printk(KERN_INFO "clear all buffer\n");
}

/**
 * @brief start watch variable
 *
 * @param warg: uapi watch_arg
 * @return int 0 is success
 */
int start_watch_variable(watch_arg warg) {
  void *kptr;
  kernel_watch_timer *timer = NULL;
  kernel_watch_arg k_watch_arg;

  // user space address to kernel space address
  kptr = convert_user_space_ptr(warg.task_id, (unsigned long)warg.ptr);
  if (kptr == NULL) {
    printk(KERN_ERR "Cannot access user space\n");
    return -EACCES;
  }
  // check length
  if (warg.length_byte != 1 && warg.length_byte != 2 && warg.length_byte != 4 &&
      warg.length_byte != 8) {
    printk(KERN_ERR "Invalid length %d\n", warg.length_byte);
    return -EINVAL;
  }
  // k_watch_arg init
  w_arg2k_w_arg(kptr, warg, &k_watch_arg);
  timer = get_timer(warg.time_ns); // get a valuable timer

  if (timer == NULL) {
    printk(KERN_ERR "No timer available, ALL timer is full\n");
    return -1;
  }

  INIT_WORK(&timer->wk, diag_task_info_work);

  printk(KERN_INFO "Convert ptr to kptr: %p\n", kptr);
  printk(KERN_INFO "Associated timer: %p , there are already %d variables, "
                   "timer period %lld.\n",
         timer, timer->sentinel, timer->time_ns);
  // printk(KERN_INFO "timer->hr_timer: %p\n", &timer->hr_timer);

  TIMER_CANCEL(timer); // just in case
  timer_add_watch(timer, k_watch_arg);
  TIMER_START(timer);

  printk(KERN_INFO "Start watching var: %s\n", warg.name);
  return 0;
}

/**
 * @brief reinit all timer's work
 *
 */
void init_work_all_hrTimer(void) {
  int i = 0;
  kernel_watch_timer *timer = NULL;
  for (i = 0; i < kernel_wtimer_num; i++) {
    timer = &(kernel_wtimer_list[i]);
    // init work
    INIT_WORK(&timer->wk, diag_task_info_work);
  }
}

/**
 * @brief clear watch with pid
 *
 * @param pid
 */
void clear_watch(pid_t pid) {
  printk(KERN_INFO "Clear pid: %d's watch variable\n", pid);
  cancel_all_hrTimer();      // just in case
  cancel_all_work();         //
  del_all_kwarg_by_pid(pid); // delete all kwarg with pid
  free_page_list(pid);       // free page with pid
  init_work_all_hrTimer();
  start_all_hrTimer(); // restart timer
}

#include <linux/module.h>
#include <linux/smp.h>

static void ipi_test(void *info) {
  // pr_info("CPU%d do task\n", smp_processor_id());
  struct  task_struct *tsk = current;
  // pr_info("CPU%d pid: %d, name: %s\n", smp_processor_id(), tsk->pid, tsk->comm);

  struct diag_variant_buffer *buffer = &stand_alone_buffer;
  unsigned long flags;

  static variable_monitor_task tsk_info;

  rcu_read_lock();
  diag_variant_buffer_spin_lock(buffer, flags);

  unsigned long event_id = get_cycles();

  tsk_info.et_type = VARIABLE_MONITOR_TASK_TYPE;
  tsk_info.id = event_id;
  tsk_info.tv = ktime_get_real();
  pr_info("diag_tsk tv %lld\n", tsk_info.tv);
  diag_tsk(tsk, &tsk_info);

  push_tskinfo_22_buffer_orig(&tsk_info, &flags, buffer); // push to buffer
  
  diag_variant_buffer_spin_unlock(buffer, flags);
  rcu_read_unlock();
}

/**
 * @brief main callback function
 *
 * @param timer
 * @return enum hrtimer_restart
 */
enum hrtimer_restart check_variable_cb(struct hrtimer *timer) {
  kernel_watch_timer *k_watch_timer =
      container_of(timer, kernel_watch_timer, hr_timer);
  int i = 0, j = 0;
  kernel_watch_arg *kwarg;

  // check all watched kernel_watch_arg
  for (i = 0; i < k_watch_timer->sentinel; i++) {
    kwarg = &k_watch_timer->k_watch_args[i];
    if (read_and_compare(kwarg->kptr, kwarg->length_byte,
                         kwarg->above_threshold, kwarg->is_unsigned,
                         kwarg->threshold)) {
      // printk(KERN_INFO "threshold reached\n");
      kwarg->true_value = convert_to_longlong(kwarg->kptr, kwarg->length_byte,
                                              kwarg->is_unsigned);
      k_watch_timer->threshold_buffer[j] = i;
      j++;
    }
  }
  if (j > 0) // if any threshold reached
  {
    k_watch_timer->threshold_over_count = j;
    k_watch_timer->tv = ktime_get_real();
    pr_info("threshold reached, tv %lld\n", k_watch_timer->tv);
    // highpri_wq
    // queue_work(system_highpri_wq, &k_watch_timer->wk);

    pid_t pid = (pid_t)2636;
    struct task_struct *tsk;
    tsk = NULL;

    rcu_read_lock();
    if (orig_find_task_by_vpid)
      tsk = orig_find_task_by_vpid(pid);
    rcu_read_unlock();
    if (tsk) {
      int cpu = task_cpu(tsk);
      // pr_info("diag_pid: %d, cpu %d\n", tsk->pid, cpu);
      smp_call_function_single(cpu, ipi_test, NULL, 1);
    }

    // diag_task(k_watch_timer);
    // orig_raise_softirq(MY_SOFTIRQ); // for swirq test

    // restart timer after dump_reset_sec sec
    hrtimer_forward(timer, timer->base->get_time(),
                    ktime_set(dump_reset_sec, 0));
  } else {
    // keep frequency
    hrtimer_forward(timer, timer->base->get_time(), k_watch_timer->kt);
  }
  return HRTIMER_RESTART; // restart timer
}

/**
 * @brief for test only
 *
 * @param id
 * @return int
 */
int diag_pid(int id) {
  pr_info("diag_pid\n");
  struct task_struct *tsk;
  int ret;
  // unsigned long flags;
  // unsigned long event_id = get_cycles();

  // static variable_monitor_task tsk_info = {0};
  // static variable_monitor_record vm_record = {0};

  pid_t pid = (pid_t)id;

  rcu_read_lock();
  tsk = NULL;
  if (orig_find_task_by_vpid)
    tsk = orig_find_task_by_vpid(pid);
  if (!tsk) {
    ret = -EINVAL;
    rcu_read_unlock();
    return ret;
  }
  rcu_read_unlock();
  int cpu = task_cpu(tsk);
  pr_info("diag_pid: %d, cpu %d\n", tsk->pid, cpu);

  // smp_call_function_single(cpu, ipi_test, NULL, 1); // 让 CPU2 执行 print_str()

  // get_task_struct(tsk); // count +1

  // tsk_info.et_type = VARIABLE_MONITOR_TASK_TYPE;
  // tsk_info.id = event_id;
  // tsk_info.tv = vm_record.tv;
  // diag_tsk(tsk, &tsk_info);
  // printk(KERN_INFO "pid: %d, name: %s\n", tsk->pid, tsk->comm);
  // setup_perf_event_for_task(tsk); // setup perf event for task

  // put_task_struct(tsk); // count -1

  // push_tskinfo_2_sa_buffer(&tsk_info, &flags); // push to buffer

  return 0;
}

/**
 * @brief for test only
 *
 * @param id
 * @return int
 */
int diag_tgid(int id) {
  struct task_struct *tsk;
  int ret;
  unsigned long flags;
  unsigned long event_id = get_cycles();

  static variable_monitor_task tsk_info = {0};
  static variable_monitor_record vm_record = {0};

  pid_t tgid = (pid_t)id;

  rcu_read_lock();
  tsk = NULL;
  if (orig_find_task_by_vpid)
    tsk = orig_find_task_by_vpid(tgid);
  if (!tsk) {
    ret = -EINVAL;
    rcu_read_unlock();
    return ret;
  }
  rcu_read_unlock();

  struct task_struct *thread = tsk;

  while_each_thread(tsk, thread) {

    get_task_struct(thread); // count +1

    tsk_info.et_type = VARIABLE_MONITOR_TASK_TYPE;
    tsk_info.id = event_id;
    tsk_info.tv = vm_record.tv;
    diag_tsk(thread, &tsk_info);

    put_task_struct(thread); // count -1

    push_tskinfo_2_sa_buffer(&tsk_info, &flags); // push to buffer
  }

  return 0;
}