mm/pdflush.c

   1 /*
   2  * mm/pdflush.c - worker threads for writing back filesystem data
   3  *
   4  * Copyright (C) 2002, Linus Torvalds.
   5  *
   6  * 09Apr2002    akpm@zip.com.au
   7  *              Initial version
   8  * 29Feb2004    kaos@sgi.com
   9  *              Move worker thread creation to kthread to avoid chewing
  10  *              up stack space with nested calls to kernel_thread.
  11  */
  12
  13 #include <linux/sched.h>
  14 #include <linux/list.h>
  15 #include <linux/signal.h>
  16 #include <linux/spinlock.h>
  17 #include <linux/gfp.h>
  18 #include <linux/init.h>
  19 #include <linux/module.h>
  20 #include <linux/fs.h>           // Needed by writeback.h
  21 #include <linux/writeback.h>    // Prototypes pdflush_operation()
  22 #include <linux/kthread.h>
  23 #include <linux/cpuset.h>
  24
  25
  26 /*
  27  * Minimum and maximum number of pdflush instances
  28  */
  29 #define MIN_PDFLUSH_THREADS     2
  30 #define MAX_PDFLUSH_THREADS     8
  31
  32 static void start_one_pdflush_thread(void);
  33
  34
  35 /*
  36  * The pdflush threads are worker threads for writing back dirty data.
  37  * Ideally, we'd like one thread per active disk spindle.  But the disk
  38  * topology is very hard to divine at this level.   Instead, we take
  39  * care in various places to prevent more than one pdflush thread from
  40  * performing writeback against a single filesystem.  pdflush threads
  41  * have the PF_FLUSHER flag set in current->flags to aid in this.
  42  */
  43
  44 /*
  45  * All the pdflush threads.  Protected by pdflush_lock
  46  */
  47 static LIST_HEAD(pdflush_list);
  48 static DEFINE_SPINLOCK(pdflush_lock);
  49
  50 /*
  51  * The count of currently-running pdflush threads.  Protected
  52  * by pdflush_lock.
  53  *
  54  * Readable by sysctl, but not writable.  Published to userspace at
  55  * /proc/sys/vm/nr_pdflush_threads.
  56  */
  57 int nr_pdflush_threads = 0;
  58
  59 /*
  60  * The time at which the pdflush thread pool last went empty
  61  */
  62 static unsigned long last_empty_jifs;
  63
  64 /*
  65  * The pdflush thread.
  66  *
  67  * Thread pool management algorithm:
  68  *
  69  * - The minimum and maximum number of pdflush instances are bound
  70  *   by MIN_PDFLUSH_THREADS and MAX_PDFLUSH_THREADS.
  71  *
  72  * - If there have been no idle pdflush instances for 1 second, create
  73  *   a new one.
  74  *
  75  * - If the least-recently-went-to-sleep pdflush thread has been asleep
  76  *   for more than one second, terminate a thread.
  77  */
  78
  79 /*
  80  * A structure for passing work to a pdflush thread.  Also for passing
  81  * state information between pdflush threads.  Protected by pdflush_lock.
  82  */
  83 struct pdflush_work {
  84         struct task_struct *who;        /* The thread */
  85         void (*fn)(unsigned long);      /* A callback function */
  86         unsigned long arg0;             /* An argument to the callback */
  87         struct list_head list;          /* On pdflush_list, when idle */
  88         unsigned long when_i_went_to_sleep;
  89 };
  90
  91 static int __pdflush(struct pdflush_work *my_work)
  92 {
  93         current->flags |= PF_FLUSHER | PF_SWAPWRITE;
  94         my_work->fn = NULL;
  95         my_work->who = current;
  96         INIT_LIST_HEAD(&my_work->list);
  97
  98         spin_lock_irq(&pdflush_lock);
  99         nr_pdflush_threads++;
 100         for ( ; ; ) {
 101                 struct pdflush_work *pdf;
 102
 103                 set_current_state(TASK_INTERRUPTIBLE);
 104                 list_move(&my_work->list, &pdflush_list);
 105                 my_work->when_i_went_to_sleep = jiffies;
 106                 spin_unlock_irq(&pdflush_lock);
 107                 schedule();
 108                 try_to_freeze();
 109                 spin_lock_irq(&pdflush_lock);
 110                 if (!list_empty(&my_work->list)) {
 111                         /*
 112                          * Someone woke us up, but without removing our control
 113                          * structure from the global list.  swsusp will do this
 114                          * in try_to_freeze()->refrigerator().  Handle it.
 115                          */
 116                         my_work->fn = NULL;
 117                         continue;
 118                 }
 119                 if (my_work->fn == NULL) {
 120                         printk("pdflush: bogus wakeup\n");
 121                         continue;
 122                 }
 123                 spin_unlock_irq(&pdflush_lock);
 124
 125                 (*my_work->fn)(my_work->arg0);
 126
 127                 /*
 128                  * Thread creation: For how long have there been zero
 129                  * available threads?
 130                  */
 131                 if (jiffies - last_empty_jifs > 1 * HZ) {
 132                         /* unlocked list_empty() test is OK here */
 133                         if (list_empty(&pdflush_list)) {
 134                                 /* unlocked test is OK here */
 135                                 if (nr_pdflush_threads < MAX_PDFLUSH_THREADS)
 136                                         start_one_pdflush_thread();
 137                         }
 138                 }
 139
 140                 spin_lock_irq(&pdflush_lock);
 141                 my_work->fn = NULL;
 142
 143                 /*
 144                  * Thread destruction: For how long has the sleepiest
 145                  * thread slept?
 146                  */
 147                 if (list_empty(&pdflush_list))
 148                         continue;
 149                 if (nr_pdflush_threads <= MIN_PDFLUSH_THREADS)
 150                         continue;
 151                 pdf = list_entry(pdflush_list.prev, struct pdflush_work, list);
 152                 if (jiffies - pdf->when_i_went_to_sleep > 1 * HZ) {
 153                         /* Limit exit rate */
 154                         pdf->when_i_went_to_sleep = jiffies;
 155                         break;                                  /* exeunt */
 156                 }
 157         }
 158         nr_pdflush_threads--;
 159         spin_unlock_irq(&pdflush_lock);
 160         return 0;
 161 }
 162
 163 /*
 164  * Of course, my_work wants to be just a local in __pdflush().  It is
 165  * separated out in this manner to hopefully prevent the compiler from
 166  * performing unfortunate optimisations against the auto variables.  Because
 167  * these are visible to other tasks and CPUs.  (No problem has actually
 168  * been observed.  This is just paranoia).
 169  */
 170 static int pdflush(void *dummy)
 171 {
 172         struct pdflush_work my_work;
 173         cpumask_t cpus_allowed;
 174
 175         /*
 176          * pdflush can spend a lot of time doing encryption via dm-crypt.  We
 177          * don't want to do that at keventd's priority.
 178          */
 179         set_user_nice(current, 0);
 180
 181         /*
 182          * Some configs put our parent kthread in a limited cpuset,
 183          * which kthread() overrides, forcing cpus_allowed == CPU_MASK_ALL.
 184          * Our needs are more modest - cut back to our cpusets cpus_allowed.
 185          * This is needed as pdflush's are dynamically created and destroyed.
 186          * The boottime pdflush's are easily placed w/o these 2 lines.
 187          */
 188         cpus_allowed = cpuset_cpus_allowed(current);
 189         set_cpus_allowed(current, cpus_allowed);
 190
 191         return __pdflush(&my_work);
 192 }
 193
 194 /*
 195  * Attempt to wake up a pdflush thread, and get it to do some work for you.
 196  * Returns zero if it indeed managed to find a worker thread, and passed your
 197  * payload to it.
 198  */
 199 int pdflush_operation(void (*fn)(unsigned long), unsigned long arg0)
 200 {
 201         unsigned long flags;
 202         int ret = 0;
 203
 204         if (fn == NULL)
 205                 BUG();          /* Hard to diagnose if it's deferred */
 206
 207         spin_lock_irqsave(&pdflush_lock, flags);
 208         if (list_empty(&pdflush_list)) {
 209                 spin_unlock_irqrestore(&pdflush_lock, flags);
 210                 ret = -1;
 211         } else {
 212                 struct pdflush_work *pdf;
 213
 214                 pdf = list_entry(pdflush_list.next, struct pdflush_work, list);
 215                 list_del_init(&pdf->list);
 216                 if (list_empty(&pdflush_list))
 217                         last_empty_jifs = jiffies;
 218                 pdf->fn = fn;
 219                 pdf->arg0 = arg0;
 220                 wake_up_process(pdf->who);
 221                 spin_unlock_irqrestore(&pdflush_lock, flags);
 222         }
 223         return ret;
 224 }
 225
 226 static void start_one_pdflush_thread(void)
 227 {
 228         kthread_run(pdflush, NULL, "pdflush");
 229 }
 230
 231 static int __init pdflush_init(void)
 232 {
 233         int i;
 234
 235         for (i = 0; i < MIN_PDFLUSH_THREADS; i++)
 236                 start_one_pdflush_thread();
 237         return 0;
 238 }
 239
 240 module_init(pdflush_init);