raid6-serialize-ovelapping-reqs.patch

来自「lustre 1.6.5 source code」· PATCH 代码 · 共 151 行

diff -pur linux-2.6.9.orig/drivers/md/raid6main.c linux-2.6.9/drivers/md/raid6main.c
--- linux-2.6.9.orig/drivers/md/raid6main.c	2008-01-10 13:55:37.000000000 +0800
+++ linux-2.6.9/drivers/md/raid6main.c	2008-01-10 13:55:56.000000000 +0800
@@ -749,6 +749,10 @@ static void compute_parity(struct stripe
 			if ( i != pd_idx && i != qd_idx && sh->dev[i].towrite ) {
 				chosen = sh->dev[i].towrite;
 				sh->dev[i].towrite = NULL;
+
+				if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
+					wake_up(&conf->wait_for_overlap);
+
 				if (sh->dev[i].written) BUG();
 				sh->dev[i].written = chosen;
 			}
@@ -907,7 +911,7 @@ static void compute_block_2(struct strip
  * toread/towrite point to the first in a chain.
  * The bi_next chain must be in order.
  */
-static void add_stripe_bio (struct stripe_head *sh, struct bio *bi, int dd_idx, int forwrite)
+static int add_stripe_bio (struct stripe_head *sh, struct bio *bi, int dd_idx, int forwrite)
 {
 	struct bio **bip;
 	raid6_conf_t *conf = sh->raid_conf;
@@ -924,10 +928,13 @@ static void add_stripe_bio (struct strip
 	else
 		bip = &sh->dev[dd_idx].toread;
 	while (*bip && (*bip)->bi_sector < bi->bi_sector) {
-		BUG_ON((*bip)->bi_sector + ((*bip)->bi_size >> 9) > bi->bi_sector);
+		if((*bip)->bi_sector + ((*bip)->bi_size >> 9) > bi->bi_sector)
+			goto overlap;
 		bip = & (*bip)->bi_next;
 	}
-/* FIXME do I need to worry about overlapping bion */
+	if (*bip && (*bip)->bi_sector < bi->bi_sector + ((bi->bi_size)>>9))
+		goto overlap;
+
 	if (*bip && bi->bi_next && (*bip) != bi->bi_next)
 		BUG();
 	if (*bip)
@@ -954,6 +961,14 @@ static void add_stripe_bio (struct strip
 		if (sector >= sh->dev[dd_idx].sector + STRIPE_SECTORS)
 			set_bit(R5_OVERWRITE, &sh->dev[dd_idx].flags);
 	}
+
+	return 1;
+
+overlap:
+	set_bit(R5_Overlap, &sh->dev[dd_idx].flags);
+	spin_unlock_irq(&conf->device_lock);
+	spin_unlock(&sh->lock);
+	return 0;
 }
 
 /*
@@ -1038,6 +1053,9 @@ static void handle_stripe(struct stripe_
 			spin_lock_irq(&conf->device_lock);
 			rbi = dev->toread;
 			dev->toread = NULL;
+
+			if (test_and_clear_bit(R5_Overlap, &dev->flags))
+				wake_up(&conf->wait_for_overlap);
 			spin_unlock_irq(&conf->device_lock);
 			while (rbi && rbi->bi_sector < dev->sector + STRIPE_SECTORS) {
 				copy_data(0, rbi, dev->page, dev->sector);
@@ -1087,6 +1105,9 @@ static void handle_stripe(struct stripe_
 			sh->dev[i].towrite = NULL;
 			if (bi) to_write--;
 
+			if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
+				wake_up(&conf->wait_for_overlap);
+
 			while (bi && bi->bi_sector < sh->dev[i].sector + STRIPE_SECTORS){
 				struct bio *nextbi = r5_next_bio(bi, sh->dev[i].sector);
 				clear_bit(BIO_UPTODATE, &bi->bi_flags);
@@ -1115,6 +1136,8 @@ static void handle_stripe(struct stripe_
 			if (!test_bit(R5_Insync, &sh->dev[i].flags)) {
 				bi = sh->dev[i].toread;
 				sh->dev[i].toread = NULL;
+				if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
+					wake_up(&conf->wait_for_overlap);
 				if (bi) to_read--;
 				while (bi && bi->bi_sector < sh->dev[i].sector + STRIPE_SECTORS){
 					struct bio *nextbi = r5_next_bio(bi, sh->dev[i].sector);
@@ -1648,6 +1671,8 @@ static int make_request (request_queue_t
 		sh = NULL;
 		/* iterrate through all pages in the stripe */
 		for (j = 0; j < data_disks && sectors > 0; j++) {
+			DEFINE_WAIT(w);
+
 			if (r_sector + STRIPE_SECTORS <= bi->bi_sector ||
 			    r_sector >= last_sector) {
 				r_sector += sectors_per_chunk;
@@ -1656,6 +1681,9 @@ static int make_request (request_queue_t
 			new_sector = raid6_compute_sector(r_sector, raid_disks,
 							data_disks, &dd_idx, 
 							&pd_idx, conf);
+
+retry:
+			prepare_to_wait(&conf->wait_for_overlap, &w, TASK_UNINTERRUPTIBLE);
 			if (sh == NULL) {
 				/* first, try to get stripe w/o blocking
 				 * if we can't, then it's time to submit
@@ -1668,10 +1696,18 @@ static int make_request (request_queue_t
 				}
 			}
 			if (sh) {
-				add_stripe_bio(sh, bi, dd_idx, (bi->bi_rw&RW_MASK));
+				if(!add_stripe_bio(sh, bi, dd_idx, (bi->bi_rw&RW_MASK))) {
+					/* Failed to be added due to overlapped. */
+					raid6_unplug_device(mddev->queue);
+					release_stripe(sh);
+					schedule();
+					goto retry;
+				}
+				finish_wait(&conf->wait_for_overlap, &w);
 			} else {
 				/* cannot get stripe for read-ahead, just give-up */
 				clear_bit(BIO_UPTODATE, &bi->bi_flags);
+				finish_wait(&conf->wait_for_overlap, &w);
 				sectors = 0;
 				break;
 			}
@@ -1847,6 +1883,7 @@ static int run (mddev_t *mddev)
 
 	conf->device_lock = SPIN_LOCK_UNLOCKED;
 	init_waitqueue_head(&conf->wait_for_stripe);
+	init_waitqueue_head(&conf->wait_for_overlap);
 	INIT_LIST_HEAD(&conf->handle_list);
 	INIT_LIST_HEAD(&conf->delayed_list);
 	INIT_LIST_HEAD(&conf->inactive_list);
diff -pur linux-2.6.9.orig/include/linux/raid/raid5.h linux-2.6.9/include/linux/raid/raid5.h
--- linux-2.6.9.orig/include/linux/raid/raid5.h	2008-01-10 13:46:05.000000000 +0800
+++ linux-2.6.9/include/linux/raid/raid5.h	2008-01-10 13:55:56.000000000 +0800
@@ -154,6 +154,8 @@ struct stripe_head {
 #define	R5_Wantwrite	5
 #define	R5_Syncio	6	/* this io need to be accounted as resync io */
 #define	R5_Direct	7	/* use page from passed bio to avoid memcpy */
+#define	R5_Overlap      8	/* There is a pending overlapping request 
+					 * on this block */
 
 /*
  * Write method
@@ -221,6 +223,7 @@ struct raid5_private_data {
 	atomic_t		active_stripes;
 	struct list_head	inactive_list;
 	wait_queue_head_t	wait_for_stripe;
+	wait_queue_head_t	wait_for_overlap;
 	int			inactive_blocked;	/* release of inactive stripes blocked,
 							 * waiting for 25% to be free
 							 */