softlockup: fix potential race in hung_task when resetting timeout
authorMandeep Singh Baines <msb@google.com>
Sat, 17 Jan 2009 18:31:48 +0000 (10:31 -0800)
committerIngo Molnar <mingo@elte.hu>
Sun, 18 Jan 2009 18:20:17 +0000 (19:20 +0100)
Impact: fix potential false panic

A potential race exists if sysctl_hung_task_timeout_secs is reset to 0
while inside check_hung_uniterruptible_tasks(). If check_task() is
entered, a comparison with 0 will result in a false hung_task being
detected.

If sysctl_hung_task_panic is set, the system will panic.

Signed-off-by: Mandeep Singh Baines <msb@google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
kernel/hung_task.c

index ba5a77c..ba8ccd4 100644 (file)
@@ -72,7 +72,8 @@ static unsigned long get_timestamp(void)
        return cpu_clock(this_cpu) >> 30LL;  /* 2^30 ~= 10^9 */
 }
 
-static void check_hung_task(struct task_struct *t, unsigned long now)
+static void check_hung_task(struct task_struct *t, unsigned long now,
+                           unsigned long timeout)
 {
        unsigned long switch_count = t->nvcsw + t->nivcsw;
 
@@ -84,8 +85,7 @@ static void check_hung_task(struct task_struct *t, unsigned long now)
                t->last_switch_timestamp = now;
                return;
        }
-       if ((long)(now - t->last_switch_timestamp) <
-                                       sysctl_hung_task_timeout_secs)
+       if ((long)(now - t->last_switch_timestamp) < timeout)
                return;
        if (!sysctl_hung_task_warnings)
                return;
@@ -96,8 +96,7 @@ static void check_hung_task(struct task_struct *t, unsigned long now)
         * complain:
         */
        printk(KERN_ERR "INFO: task %s:%d blocked for more than "
-                       "%ld seconds.\n", t->comm, t->pid,
-                       sysctl_hung_task_timeout_secs);
+                       "%ld seconds.\n", t->comm, t->pid, timeout);
        printk(KERN_ERR "\"echo 0 > /proc/sys/kernel/hung_task_timeout_secs\""
                        " disables this message.\n");
        sched_show_task(t);
@@ -115,7 +114,7 @@ static void check_hung_task(struct task_struct *t, unsigned long now)
  * a really long time (120 seconds). If that happens, print out
  * a warning.
  */
-static void check_hung_uninterruptible_tasks(void)
+static void check_hung_uninterruptible_tasks(unsigned long timeout)
 {
        int max_count = sysctl_hung_task_check_count;
        unsigned long now = get_timestamp();
@@ -134,7 +133,7 @@ static void check_hung_uninterruptible_tasks(void)
                        goto unlock;
                /* use "==" to skip the TASK_KILLABLE tasks waiting on NFS */
                if (t->state == TASK_UNINTERRUPTIBLE)
-                       check_hung_task(t, now);
+                       check_hung_task(t, now, timeout);
        } while_each_thread(g, t);
  unlock:
        read_unlock(&tasklist_lock);
@@ -180,8 +179,17 @@ static int watchdog(void *dummy)
        update_poll_jiffies();
 
        for ( ; ; ) {
+               unsigned long timeout;
+
                while (schedule_timeout_interruptible(hung_task_poll_jiffies));
-               check_hung_uninterruptible_tasks();
+
+               /*
+                * Need to cache timeout here to avoid timeout being set
+                * to 0 via sysctl while inside check_hung_*_tasks().
+                */
+               timeout = sysctl_hung_task_timeout_secs;
+               if (timeout)
+                       check_hung_uninterruptible_tasks(timeout);
        }
 
        return 0;