From be4d3280b17bc51f23ec6ebb345728f302f80a0c Mon Sep 17 00:00:00 2001
From: Shaohua Li <shli@kernel.org>
Date: Tue, 31 Jul 2012 10:03:53 +1000
Subject: [PATCH] md/raid1: make sequential read detection per disk based

Currently the sequential read detection is global wide. It's natural to make it
per disk based, which can improve the detection for concurrent multiple
sequential reads. And next patch will make SSD read balance not use distance
based algorithm, where this change help detect truly sequential read for SSD.

Signed-off-by: Shaohua Li <shli@fusionio.com>
Signed-off-by: NeilBrown <neilb@suse.de>
---
 drivers/md/raid1.c | 35 +++++++----------------------------
 drivers/md/raid1.h | 11 +++++------
 2 files changed, 12 insertions(+), 34 deletions(-)

diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index d3d3568b4fb1..fb96c0c2db40 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -497,9 +497,8 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect
 	const sector_t this_sector = r1_bio->sector;
 	int sectors;
 	int best_good_sectors;
-	int start_disk;
 	int best_disk;
-	int i;
+	int disk;
 	sector_t best_dist;
 	struct md_rdev *rdev;
 	int choose_first;
@@ -517,23 +516,16 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect
 	best_good_sectors = 0;
 
 	if (conf->mddev->recovery_cp < MaxSector &&
-	    (this_sector + sectors >= conf->next_resync)) {
+	    (this_sector + sectors >= conf->next_resync))
 		choose_first = 1;
-		start_disk = 0;
-	} else {
+	else
 		choose_first = 0;
-		start_disk = conf->last_used;
-	}
 
-	for (i = 0 ; i < conf->raid_disks * 2 ; i++) {
+	for (disk = 0 ; disk < conf->raid_disks * 2 ; disk++) {
 		sector_t dist;
 		sector_t first_bad;
 		int bad_sectors;
 
-		int disk = start_disk + i;
-		if (disk >= conf->raid_disks * 2)
-			disk -= conf->raid_disks * 2;
-
 		rdev = rcu_dereference(conf->mirrors[disk].rdev);
 		if (r1_bio->bios[disk] == IO_BLOCKED
 		    || rdev == NULL
@@ -594,7 +586,7 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect
 		dist = abs(this_sector - conf->mirrors[disk].head_position);
 		if (choose_first
 		    /* Don't change to another disk for sequential reads */
-		    || conf->next_seq_sect == this_sector
+		    || conf->mirrors[disk].next_seq_sect == this_sector
 		    || dist == 0
 		    /* If device is idle, use it */
 		    || atomic_read(&rdev->nr_pending) == 0) {
@@ -620,8 +612,7 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect
 			goto retry;
 		}
 		sectors = best_good_sectors;
-		conf->next_seq_sect = this_sector + sectors;
-		conf->last_used = best_disk;
+		conf->mirrors[best_disk].next_seq_sect = this_sector + sectors;
 	}
 	rcu_read_unlock();
 	*max_sectors = sectors;
@@ -2599,7 +2590,6 @@ static struct r1conf *setup_conf(struct mddev *mddev)
 	conf->recovery_disabled = mddev->recovery_disabled - 1;
 
 	err = -EIO;
-	conf->last_used = -1;
 	for (i = 0; i < conf->raid_disks * 2; i++) {
 
 		disk = conf->mirrors + i;
@@ -2625,19 +2615,9 @@ static struct r1conf *setup_conf(struct mddev *mddev)
 			if (disk->rdev &&
 			    (disk->rdev->saved_raid_disk < 0))
 				conf->fullsync = 1;
-		} else if (conf->last_used < 0)
-			/*
-			 * The first working device is used as a
-			 * starting point to read balancing.
-			 */
-			conf->last_used = i;
+		}
 	}
 
-	if (conf->last_used < 0) {
-		printk(KERN_ERR "md/raid1:%s: no operational mirrors\n",
-		       mdname(mddev));
-		goto abort;
-	}
 	err = -ENOMEM;
 	conf->thread = md_register_thread(raid1d, mddev, "raid1");
 	if (!conf->thread) {
@@ -2894,7 +2874,6 @@ static int raid1_reshape(struct mddev *mddev)
 	conf->raid_disks = mddev->raid_disks = raid_disks;
 	mddev->delta_disks = 0;
 
-	conf->last_used = 0; /* just make sure it is in-range */
 	lower_barrier(conf);
 
 	set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
diff --git a/drivers/md/raid1.h b/drivers/md/raid1.h
index 4e3613daaea2..3770b4a27662 100644
--- a/drivers/md/raid1.h
+++ b/drivers/md/raid1.h
@@ -4,6 +4,11 @@
 struct raid1_info {
 	struct md_rdev	*rdev;
 	sector_t	head_position;
+
+	/* When choose the best device for a read (read_balance())
+	 * we try to keep sequential reads one the same device
+	 */
+	sector_t	next_seq_sect;
 };
 
 /*
@@ -29,12 +34,6 @@ struct r1conf {
 						 */
 	int			raid_disks;
 
-	/* When choose the best device for a read (read_balance())
-	 * we try to keep sequential reads one the same device
-	 * using 'last_used' and 'next_seq_sect'
-	 */
-	int			last_used;
-	sector_t		next_seq_sect;
 	/* During resync, read_balancing is only allowed on the part
 	 * of the array that has been resynced.  'next_resync' tells us
 	 * where that is.
-- 
2.30.2