[PATCH] md: define ->congested_fn for raid1, raid10, and multipath

raid1, raid10 and multipath don't report their 'congested' status through
bdi_*_congested, but should.

This patch adds the appropriate functions which just check the 'congested'
status of all active members (with appropriate locking).

raid1 read_balance should be modified to prefer devices where
bdi_read_congested returns false.  Then we could use the '&' branch rather
than the '|' branch.  However that should would need some benchmarking first
to make sure it is actually a good idea.

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

3 files changed, 74 insertions, 0 deletions

diff --git a/drivers/md/multipath.c b/drivers/md/multipath.c
index e4f168d063db..171ff41b52b0 100644
--- a/drivers/md/multipath.c
+++ b/drivers/md/multipath.c
@@ -228,6 +228,28 @@ static int multipath_issue_flush(request_queue_t *q, struct gendisk *disk,
         rcu_read_unlock();
         return ret;
 }
+static int multipath_congested(void *data, int bits)
+{
+        mddev_t *mddev = data;
+        multipath_conf_t *conf = mddev_to_conf(mddev);
+        int i, ret = 0;
+
+        rcu_read_lock();
+        for (i = 0; i < mddev->raid_disks ; i++) {
+                mdk_rdev_t *rdev = rcu_dereference(conf->multipaths[i].rdev);
+                if (rdev && !test_bit(Faulty, &rdev->flags)) {
+                        request_queue_t *q = bdev_get_queue(rdev->bdev);
+
+                        ret |= bdi_congested(&q->backing_dev_info, bits);
+                        /* Just like multipath_map, we just check the
+                         * first available device
+                         */
+                        break;
+                }
+        }
+        rcu_read_unlock();
+        return ret;
+}
 
 /*
  * Careful, this can execute in IRQ contexts as well!
@@ -509,6 +531,8 @@ static int multipath_run (mddev_t *mddev)
 
         mddev->queue->unplug_fn = multipath_unplug;
         mddev->queue->issue_flush_fn = multipath_issue_flush;
+        mddev->queue->backing_dev_info.congested_fn = multipath_congested;
+        mddev->queue->backing_dev_info.congested_data = mddev;
 
         return 0;
 
diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index 99c4e031c7f1..dc9d2def0270 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -601,6 +601,32 @@ static int raid1_issue_flush(request_queue_t *q, struct gendisk *disk,
         return ret;
 }
 
+static int raid1_congested(void *data, int bits)
+{
+        mddev_t *mddev = data;
+        conf_t *conf = mddev_to_conf(mddev);
+        int i, ret = 0;
+
+        rcu_read_lock();
+        for (i = 0; i < mddev->raid_disks; i++) {
+                mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
+                if (rdev && !test_bit(Faulty, &rdev->flags)) {
+                        request_queue_t *q = bdev_get_queue(rdev->bdev);
+
+                        /* Note the '|| 1' - when read_balance prefers
+                         * non-congested targets, it can be removed
+                         */
+                        if ((bits & (1<<BDI_write_congested)) || 1)
+                                ret |= bdi_congested(&q->backing_dev_info, bits);
+                        else
+                                ret &= bdi_congested(&q->backing_dev_info, bits);
+                }
+        }
+        rcu_read_unlock();
+        return ret;
+}
+
+
 /* Barriers....
  * Sometimes we need to suspend IO while we do something else,
  * either some resync/recovery, or reconfigure the array.
@@ -1965,6 +1991,8 @@ static int run(mddev_t *mddev)
 
         mddev->queue->unplug_fn = raid1_unplug;
         mddev->queue->issue_flush_fn = raid1_issue_flush;
+        mddev->queue->backing_dev_info.congested_fn = raid1_congested;
+        mddev->queue->backing_dev_info.congested_data = mddev;
 
         return 0;
 
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 64f8016ab740..1250f0eab4af 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -648,6 +648,26 @@ static int raid10_issue_flush(request_queue_t *q, struct gendisk *disk,
         return ret;
 }
 
+static int raid10_congested(void *data, int bits)
+{
+        mddev_t *mddev = data;
+        conf_t *conf = mddev_to_conf(mddev);
+        int i, ret = 0;
+
+        rcu_read_lock();
+        for (i = 0; i < mddev->raid_disks && ret == 0; i++) {
+                mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
+                if (rdev && !test_bit(Faulty, &rdev->flags)) {
+                        request_queue_t *q = bdev_get_queue(rdev->bdev);
+
+                        ret |= bdi_congested(&q->backing_dev_info, bits);
+                }
+        }
+        rcu_read_unlock();
+        return ret;
+}
+
+
 /* Barriers....
  * Sometimes we need to suspend IO while we do something else,
  * either some resync/recovery, or reconfigure the array.
@@ -2094,6 +2114,8 @@ static int run(mddev_t *mddev)
 
         mddev->queue->unplug_fn = raid10_unplug;
         mddev->queue->issue_flush_fn = raid10_issue_flush;
+        mddev->queue->backing_dev_info.congested_fn = raid10_congested;
+        mddev->queue->backing_dev_info.congested_data = mddev;
 
         /* Calculate max read-ahead size.
          * We need to readahead at least twice a whole stripe....