[patch] raid-2.5.1-I8

[Ingo Molnar <mingo@elte.hu>]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1799 bytes --]


the -I7 RAID patch did not apply to 2.5.1 cleanly - the attached -I8
version does.

	Ingo

-I7's Changelog:

 - cleaned up the resync engine. It got much simpler and easier to
   maintain, while still saturating the disks. Resync doesnt get stuck
   under heavy load anymore. (this code can be switched to use explicit IO
   barrier requests in the future.)

 - rewrote the read balancing code to use three estimators: a per-array
   'next expected sequential IO' position, plus an IRQ-driven 'estimated
   disk head' position. The head position is now updated from all the IO
   completion routines: end of READ, end of WRITE, end of resync-READ, end
   of resync-WRITE. I've added per-disk tracking of pending requests,
   and the read balancer now detects idle disks and utilizes them before
   trying to read-balance between busy disks. I've also removed the
   sector_count limit that artificially switched the current disk. These
   changes make read balancing more accurate and more effective.

 - the old raid1 code used to have a limitation: it has always read from
   the first disk until the resync finished. Now the code will
   read-balance READ requests up to the resync boundary. This should
   further improve performance during resyncs.

 - added the 'idle IO resync' feature which we used to have in the 2.2
   patches, but via a different implementation that does not touch the
   generic block IO code. Resync happens only when there is no normal IO
   pending on the array. This feature should make resync a more seemless
   operation. Resync behavior can be tuned via the speed_limit_min and
   speed_limit_max sysctl tunables. Default for the minimum resync speed
   is 500 KB/sec, the maximum is 200 MB/sec.

 - fixed a number of sector_t <=> unsigned long bugs still left.

[-- Attachment #2: Type: TEXT/PLAIN, Size: 5676 bytes --]

--- linux/include/linux/raid/raid1.h.orig	Mon Dec 17 14:22:38 2001
+++ linux/include/linux/raid/raid1.h	Wed Dec 19 13:56:30 2001
@@ -9,8 +9,8 @@
 	int		number;
 	int		raid_disk;
 	kdev_t		dev;
-	int		sect_limit;
-	int		head_position;
+	sector_t	head_position;
+	atomic_t	nr_pending;
 
 	/*
 	 * State bits:
@@ -31,23 +31,21 @@
 	int			raid_disks;
 	int			working_disks;
 	int			last_used;
-	sector_t		next_sect;
-	int			sect_count;
+	sector_t		next_seq_sect;
 	mdk_thread_t		*thread, *resync_thread;
 	int			resync_mirrors;
 	mirror_info_t		*spare;
 	spinlock_t		device_lock;
 
 	/* for use when syncing mirrors: */
-	unsigned long	start_active, start_ready,
-		start_pending, start_future;
-	int	cnt_done, cnt_active, cnt_ready,
-		cnt_pending, cnt_future;
-	int	phase;
-	int	window;
-	wait_queue_head_t	wait_done;
-	wait_queue_head_t	wait_ready;
-	spinlock_t		segment_lock;
+
+	spinlock_t		resync_lock;
+	int nr_pending;
+	int barrier;
+	sector_t		next_resync;
+
+	wait_queue_head_t	wait_idle;
+	wait_queue_head_t	wait_resume;
 
 	mempool_t *r1bio_pool;
 	mempool_t *r1buf_pool;
@@ -62,7 +60,8 @@
 #define mddev_to_conf(mddev) ((conf_t *) mddev->private)
 
 /*
- * this is our 'private' 'collective' RAID1 buffer head.
+ * this is our 'private' RAID1 bio.
+ *
  * it contains information about what kind of IO operations were started
  * for this RAID1 operation, and about their status:
  */
@@ -83,6 +82,7 @@
 	 * if the IO is in READ direction, then this bio is used:
 	 */
 	struct bio		*read_bio;
+	int			read_disk;
 	/*
 	 * if the IO is in WRITE direction, then multiple bios are used:
 	 */
@@ -94,5 +94,5 @@
 
 /* bits for r1bio.state */
 #define	R1BIO_Uptodate	1
-#define	R1BIO_SyncPhase	2
+
 #endif
--- linux/include/linux/raid/md_k.h.orig	Mon Dec 17 22:19:02 2001
+++ linux/include/linux/raid/md_k.h	Wed Dec 19 13:56:30 2001
@@ -240,7 +240,7 @@
 
 	int (*stop_resync)(mddev_t *mddev);
 	int (*restart_resync)(mddev_t *mddev);
-	int (*sync_request)(mddev_t *mddev, sector_t sector_nr);
+	int (*sync_request)(mddev_t *mddev, sector_t sector_nr, int go_faster);
 };
 
 
--- linux/drivers/md/raid1.c.orig	Wed Dec 19 13:56:23 2001
+++ linux/drivers/md/raid1.c	Wed Dec 19 13:56:30 2001
@@ -935,9 +935,9 @@
 	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
 	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
 
-	check_all_w_bios_empty(r1_bio);
 	if (r1_bio->read_bio != bio)
 		BUG();
+	update_head_pos(r1_bio->read_disk, r1_bio);
 	/*
 	 * we have read a block, now it needs to be re-written,
 	 * or re-read if the read failed.
@@ -957,13 +957,21 @@
 	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
 	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
 	mddev_t *mddev = r1_bio->mddev;
+	int i;
 
 	if (!uptodate)
 		md_error(mddev, bio->bi_dev);
 
+	for (i = 0; i < MD_SB_DISKS; i++)
+		if (r1_bio->write_bios[i] == bio) {
+			update_head_pos(i, r1_bio);
+			break;
+		}
+
 	if (atomic_dec_and_test(&r1_bio->remaining)) {
-		sync_request_done(r1_bio->sector, mddev_to_conf(mddev));
+		conf_t *conf = mddev_to_conf(mddev);
 		md_done_sync(mddev, r1_bio->master_bio->bi_size >> 9, uptodate);
+		resume_device(conf);
 		put_buf(r1_bio);
 	}
 	return 0;
@@ -1073,9 +1081,9 @@
 		r1_bio = list_entry(head->prev, r1bio_t, retry_list);
 		list_del(head->prev);
 		spin_unlock_irqrestore(&retry_list_lock, flags);
-		check_all_w_bios_empty(r1_bio);
 
 		mddev = r1_bio->mddev;
+		conf = mddev_to_conf(mddev);
 		if (mddev->sb_dirty) {
 			printk(KERN_INFO "raid1: dirty sb detected, updating.\n");
 			mddev->sb_dirty = 0;
--- linux/drivers/md/md.c.orig	Mon Dec 17 22:18:41 2001
+++ linux/drivers/md/md.c	Wed Dec 19 13:56:30 2001
@@ -66,7 +66,7 @@
 
 /*
  * Current RAID-1,4,5 parallel reconstruction 'guaranteed speed limit'
- * is 100 KB/sec, so the extra system load does not show up that much.
+ * is 1000 KB/sec, so the extra system load does not show up that much.
  * Increase it if you want to have more _guaranteed_ speed. Note that
  * the RAID driver will use the maximum available bandwith if the IO
  * subsystem is idle. There is also an 'absolute maximum' reconstruction
@@ -76,8 +76,8 @@
  * you can change it via /proc/sys/dev/raid/speed_limit_min and _max.
  */
 
-static int sysctl_speed_limit_min = 100;
-static int sysctl_speed_limit_max = 100000;
+static int sysctl_speed_limit_min = 1000;
+static int sysctl_speed_limit_max = 200000;
 
 static struct ctl_table_header *raid_table_header;
 
@@ -3336,7 +3336,7 @@
 int md_do_sync(mddev_t *mddev, mdp_disk_t *spare)
 {
 	mddev_t *mddev2;
-	unsigned int max_sectors, currspeed,
+	unsigned int max_sectors, currspeed = 0,
 		j, window, err, serialize;
 	unsigned long mark[SYNC_MARKS];
 	unsigned long mark_cnt[SYNC_MARKS];
@@ -3376,8 +3376,7 @@
 	max_sectors = mddev->sb->size << 1;
 
 	printk(KERN_INFO "md: syncing RAID array md%d\n", mdidx(mddev));
-	printk(KERN_INFO "md: minimum _guaranteed_ reconstruction speed: %d KB/sec/disc.\n",
-						sysctl_speed_limit_min);
+	printk(KERN_INFO "md: minimum _guaranteed_ reconstruction speed: %d KB/sec/disc.\n", sysctl_speed_limit_min);
 	printk(KERN_INFO "md: using maximum available idle IO bandwith "
 	       "(but not more than %d KB/sec) for reconstruction.\n",
 	       sysctl_speed_limit_max);
@@ -3409,7 +3408,7 @@
 	for (j = 0; j < max_sectors;) {
 		int sectors;
 
-		sectors = mddev->pers->sync_request(mddev, j);
+		sectors = mddev->pers->sync_request(mddev, j, currspeed < sysctl_speed_limit_min);
 		if (sectors < 0) {
 			err = sectors;
 			goto out;