writeback: report dirty thresholds in /proc/vmstat
Michael Rubin [Tue, 26 Oct 2010 21:21:36 +0000 (14:21 -0700)]
The kernel already exposes the user desired thresholds in /proc/sys/vm
with dirty_background_ratio and background_ratio.  But the kernel may
alter the number requested without giving the user any indication that is
the case.

Knowing the actual ratios the kernel is honoring can help app developers
understand how their buffered IO will be sent to the disk.

        $ grep threshold /proc/vmstat
        nr_dirty_threshold 409111
        nr_dirty_background_threshold 818223

Signed-off-by: Michael Rubin <mrubin@google.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/vmstat.c

index 44e7ac0..baa4ab3 100644 (file)
@@ -17,6 +17,7 @@
 #include <linux/vmstat.h>
 #include <linux/sched.h>
 #include <linux/math64.h>
+#include <linux/writeback.h>
 
 #ifdef CONFIG_VM_EVENT_COUNTERS
 DEFINE_PER_CPU(struct vm_event_state, vm_event_states) = {{0}};
@@ -747,6 +748,8 @@ static const char * const vmstat_text[] = {
        "nr_shmem",
        "nr_dirtied",
        "nr_written",
+       "nr_dirty_threshold",
+       "nr_dirty_background_threshold",
 
 #ifdef CONFIG_NUMA
        "numa_hit",
@@ -907,36 +910,44 @@ static const struct file_operations proc_zoneinfo_file_operations = {
        .release        = seq_release,
 };
 
+enum writeback_stat_item {
+       NR_DIRTY_THRESHOLD,
+       NR_DIRTY_BG_THRESHOLD,
+       NR_VM_WRITEBACK_STAT_ITEMS,
+};
+
 static void *vmstat_start(struct seq_file *m, loff_t *pos)
 {
        unsigned long *v;
-#ifdef CONFIG_VM_EVENT_COUNTERS
-       unsigned long *e;
-#endif
-       int i;
+       int i, stat_items_size;
 
        if (*pos >= ARRAY_SIZE(vmstat_text))
                return NULL;
+       stat_items_size = NR_VM_ZONE_STAT_ITEMS * sizeof(unsigned long) +
+                         NR_VM_WRITEBACK_STAT_ITEMS * sizeof(unsigned long);
 
 #ifdef CONFIG_VM_EVENT_COUNTERS
-       v = kmalloc(NR_VM_ZONE_STAT_ITEMS * sizeof(unsigned long)
-                       + sizeof(struct vm_event_state), GFP_KERNEL);
-#else
-       v = kmalloc(NR_VM_ZONE_STAT_ITEMS * sizeof(unsigned long),
-                       GFP_KERNEL);
+       stat_items_size += sizeof(struct vm_event_state);
 #endif
+
+       v = kmalloc(stat_items_size, GFP_KERNEL);
        m->private = v;
        if (!v)
                return ERR_PTR(-ENOMEM);
        for (i = 0; i < NR_VM_ZONE_STAT_ITEMS; i++)
                v[i] = global_page_state(i);
+       v += NR_VM_ZONE_STAT_ITEMS;
+
+       global_dirty_limits(v + NR_DIRTY_BG_THRESHOLD,
+                           v + NR_DIRTY_THRESHOLD);
+       v += NR_VM_WRITEBACK_STAT_ITEMS;
+
 #ifdef CONFIG_VM_EVENT_COUNTERS
-       e = v + NR_VM_ZONE_STAT_ITEMS;
-       all_vm_events(e);
-       e[PGPGIN] /= 2;         /* sectors -> kbytes */
-       e[PGPGOUT] /= 2;
+       all_vm_events(v);
+       v[PGPGIN] /= 2;         /* sectors -> kbytes */
+       v[PGPGOUT] /= 2;
 #endif
-       return v + *pos;
+       return m->private + *pos;
 }
 
 static void *vmstat_next(struct seq_file *m, void *arg, loff_t *pos)