Merge branch 'for-linus' of git://neil.brown.name/md

* 'for-linus' of git://neil.brown.name/md:
  md: use interruptible wait when duration is controlled by userspace.
  md/raid5: suspend shouldn't affect read requests.
  md: tidy up error paths in md_alloc
  md: fix error path when duplicate name is found on md device creation.
  md: avoid dereferencing NULL pointer when accessing suspend_* sysfs attributes.
  md: Use new topology calls to indicate alignment and I/O sizes

drivers/md/linear.c

@@ -166,8 +166,8 @@ static linear_conf_t *linear_conf(mddev_t *mddev, int raid_disks)
 			rdev->sectors = sectors * mddev->chunk_sectors;
 		}
 
-		blk_queue_stack_limits(mddev->queue,
-				       rdev->bdev->bd_disk->queue);
+		disk_stack_limits(mddev->gendisk, rdev->bdev,
+				  rdev->data_offset << 9);
 		/* as we don't honour merge_bvec_fn, we must never risk
 		 * violating it, so limit ->max_sector to one PAGE, as
 		 * a one page request is never in violation.

drivers/md/md.c

@@ -3573,7 +3573,8 @@ suspend_lo_store(mddev_t *mddev, const char *buf, size_t len)
 	char *e;
 	unsigned long long new = simple_strtoull(buf, &e, 10);
 
-	if (mddev->pers->quiesce == NULL)
+	if (mddev->pers == NULL || 
+	    mddev->pers->quiesce == NULL)
 		return -EINVAL;
 	if (buf == e || (*e && *e != '\n'))
 		return -EINVAL;
@@ -3601,7 +3602,8 @@ suspend_hi_store(mddev_t *mddev, const char *buf, size_t len)
 	char *e;
 	unsigned long long new = simple_strtoull(buf, &e, 10);
 
-	if (mddev->pers->quiesce == NULL)
+	if (mddev->pers == NULL ||
+	    mddev->pers->quiesce == NULL)
 		return -EINVAL;
 	if (buf == e || (*e && *e != '\n'))
 		return -EINVAL;
@@ -3844,11 +3846,9 @@ static int md_alloc(dev_t dev, char *name)
 	flush_scheduled_work();
 
 	mutex_lock(&disks_mutex);
-	if (mddev->gendisk) {
-		mutex_unlock(&disks_mutex);
-		mddev_put(mddev);
-		return -EEXIST;
-	}
+	error = -EEXIST;
+	if (mddev->gendisk)
+		goto abort;
 
 	if (name) {
 		/* Need to ensure that 'name' is not a duplicate.
@@ -3860,17 +3860,15 @@ static int md_alloc(dev_t dev, char *name)
 			if (mddev2->gendisk &&
 			    strcmp(mddev2->gendisk->disk_name, name) == 0) {
 				spin_unlock(&all_mddevs_lock);
-				return -EEXIST;
+				goto abort;
 			}
 		spin_unlock(&all_mddevs_lock);
 	}
 
+	error = -ENOMEM;
 	mddev->queue = blk_alloc_queue(GFP_KERNEL);
-	if (!mddev->queue) {
-		mutex_unlock(&disks_mutex);
-		mddev_put(mddev);
-		return -ENOMEM;
-	}
+	if (!mddev->queue)
+		goto abort;
 	mddev->queue->queuedata = mddev;
 
 	/* Can be unlocked because the queue is new: no concurrency */
@@ -3880,11 +3878,9 @@ static int md_alloc(dev_t dev, char *name)
 
 	disk = alloc_disk(1 << shift);
 	if (!disk) {
-		mutex_unlock(&disks_mutex);
 		blk_cleanup_queue(mddev->queue);
 		mddev->queue = NULL;
-		mddev_put(mddev);
-		return -ENOMEM;
+		goto abort;
 	}
 	disk->major = MAJOR(mddev->unit);
 	disk->first_minor = unit << shift;
@@ -3906,16 +3902,22 @@ static int md_alloc(dev_t dev, char *name)
 	mddev->gendisk = disk;
 	error = kobject_init_and_add(&mddev->kobj, &md_ktype,
 				     &disk_to_dev(disk)->kobj, "%s", "md");
-	mutex_unlock(&disks_mutex);
-	if (error)
+	if (error) {
+		/* This isn't possible, but as kobject_init_and_add is marked
+		 * __must_check, we must do something with the result
+		 */
 		printk(KERN_WARNING "md: cannot register %s/md - name in use\n",
 		       disk->disk_name);
-	else {
+		error = 0;
+	}
+ abort:
+	mutex_unlock(&disks_mutex);
+	if (!error) {
 		kobject_uevent(&mddev->kobj, KOBJ_ADD);
 		mddev->sysfs_state = sysfs_get_dirent(mddev->kobj.sd, "array_state");
 	}
 	mddev_put(mddev);
-	return 0;
+	return error;
 }
 
 static struct kobject *md_probe(dev_t dev, int *part, void *data)
@@ -6334,10 +6336,16 @@ void md_do_sync(mddev_t *mddev)
 			sysfs_notify(&mddev->kobj, NULL, "sync_completed");
 		}
 
-		if (j >= mddev->resync_max)
-			wait_event(mddev->recovery_wait,
-				   mddev->resync_max > j
-				   || kthread_should_stop());
+		while (j >= mddev->resync_max && !kthread_should_stop()) {
+			/* As this condition is controlled by user-space,
+			 * we can block indefinitely, so use '_interruptible'
+			 * to avoid triggering warnings.
+			 */
+			flush_signals(current); /* just in case */
+			wait_event_interruptible(mddev->recovery_wait,
+						 mddev->resync_max > j
+						 || kthread_should_stop());
+		}
 
 		if (kthread_should_stop())
 			goto interrupted;

drivers/md/multipath.c

@@ -294,7 +294,8 @@ static int multipath_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
 	for (path = first; path <= last; path++)
 		if ((p=conf->multipaths+path)->rdev == NULL) {
 			q = rdev->bdev->bd_disk->queue;
-			blk_queue_stack_limits(mddev->queue, q);
+			disk_stack_limits(mddev->gendisk, rdev->bdev,
+					  rdev->data_offset << 9);
 
 		/* as we don't honour merge_bvec_fn, we must never risk
 		 * violating it, so limit ->max_sector to one PAGE, as
@@ -463,9 +464,9 @@ static int multipath_run (mddev_t *mddev)
 
 		disk = conf->multipaths + disk_idx;
 		disk->rdev = rdev;
+		disk_stack_limits(mddev->gendisk, rdev->bdev,
+				  rdev->data_offset << 9);
 
-		blk_queue_stack_limits(mddev->queue,
-				       rdev->bdev->bd_disk->queue);
 		/* as we don't honour merge_bvec_fn, we must never risk
 		 * violating it, not that we ever expect a device with
 		 * a merge_bvec_fn to be involved in multipath */

drivers/md/raid0.c

@@ -170,8 +170,8 @@ static int create_strip_zones(mddev_t *mddev)
 		}
 		dev[j] = rdev1;
 
-		blk_queue_stack_limits(mddev->queue,
-				       rdev1->bdev->bd_disk->queue);
+		disk_stack_limits(mddev->gendisk, rdev1->bdev,
+				  rdev1->data_offset << 9);
 		/* as we don't honour merge_bvec_fn, we must never risk
 		 * violating it, so limit ->max_sector to one PAGE, as
 		 * a one page request is never in violation.
@@ -250,6 +250,11 @@ static int create_strip_zones(mddev_t *mddev)
 		       mddev->chunk_sectors << 9);
 		goto abort;
 	}
+
+	blk_queue_io_min(mddev->queue, mddev->chunk_sectors << 9);
+	blk_queue_io_opt(mddev->queue,
+			 (mddev->chunk_sectors << 9) * mddev->raid_disks);
+
 	printk(KERN_INFO "raid0: done.\n");
 	mddev->private = conf;
 	return 0;

drivers/md/raid1.c

@@ -1123,8 +1123,8 @@ static int raid1_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
 	for (mirror = first; mirror <= last; mirror++)
 		if ( !(p=conf->mirrors+mirror)->rdev) {
 
-			blk_queue_stack_limits(mddev->queue,
-					       rdev->bdev->bd_disk->queue);
+			disk_stack_limits(mddev->gendisk, rdev->bdev,
+					  rdev->data_offset << 9);
 			/* as we don't honour merge_bvec_fn, we must never risk
 			 * violating it, so limit ->max_sector to one PAGE, as
 			 * a one page request is never in violation.
@@ -1988,9 +1988,8 @@ static int run(mddev_t *mddev)
 		disk = conf->mirrors + disk_idx;
 
 		disk->rdev = rdev;
-
-		blk_queue_stack_limits(mddev->queue,
-				       rdev->bdev->bd_disk->queue);
+		disk_stack_limits(mddev->gendisk, rdev->bdev,
+				  rdev->data_offset << 9);
 		/* as we don't honour merge_bvec_fn, we must never risk
 		 * violating it, so limit ->max_sector to one PAGE, as
 		 * a one page request is never in violation.

drivers/md/raid10.c

@@ -1151,8 +1151,8 @@ static int raid10_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
 	for ( ; mirror <= last ; mirror++)
 		if ( !(p=conf->mirrors+mirror)->rdev) {
 
-			blk_queue_stack_limits(mddev->queue,
-					       rdev->bdev->bd_disk->queue);
+			disk_stack_limits(mddev->gendisk, rdev->bdev,
+					  rdev->data_offset << 9);
 			/* as we don't honour merge_bvec_fn, we must never risk
 			 * violating it, so limit ->max_sector to one PAGE, as
 			 * a one page request is never in violation.
@@ -2044,7 +2044,7 @@ raid10_size(mddev_t *mddev, sector_t sectors, int raid_disks)
 static int run(mddev_t *mddev)
 {
 	conf_t *conf;
-	int i, disk_idx;
+	int i, disk_idx, chunk_size;
 	mirror_info_t *disk;
 	mdk_rdev_t *rdev;
 	int nc, fc, fo;
@@ -2130,6 +2130,14 @@ static int run(mddev_t *mddev)
 	spin_lock_init(&conf->device_lock);
 	mddev->queue->queue_lock = &conf->device_lock;
 
+	chunk_size = mddev->chunk_sectors << 9;
+	blk_queue_io_min(mddev->queue, chunk_size);
+	if (conf->raid_disks % conf->near_copies)
+		blk_queue_io_opt(mddev->queue, chunk_size * conf->raid_disks);
+	else
+		blk_queue_io_opt(mddev->queue, chunk_size *
+				 (conf->raid_disks / conf->near_copies));
+
 	list_for_each_entry(rdev, &mddev->disks, same_set) {
 		disk_idx = rdev->raid_disk;
 		if (disk_idx >= mddev->raid_disks
@@ -2138,9 +2146,8 @@ static int run(mddev_t *mddev)
 		disk = conf->mirrors + disk_idx;
 
 		disk->rdev = rdev;
-
-		blk_queue_stack_limits(mddev->queue,
-				       rdev->bdev->bd_disk->queue);
+		disk_stack_limits(mddev->gendisk, rdev->bdev,
+				  rdev->data_offset << 9);
 		/* as we don't honour merge_bvec_fn, we must never risk
 		 * violating it, so limit ->max_sector to one PAGE, as
 		 * a one page request is never in violation.

drivers/md/raid5.c

@@ -3699,13 +3699,21 @@ static int make_request(struct request_queue *q, struct bio * bi)
 					goto retry;
 				}
 			}
-			/* FIXME what if we get a false positive because these
-			 * are being updated.
-			 */
-			if (logical_sector >= mddev->suspend_lo &&
+
+			if (bio_data_dir(bi) == WRITE &&
+			    logical_sector >= mddev->suspend_lo &&
 			    logical_sector < mddev->suspend_hi) {
 				release_stripe(sh);
-				schedule();
+				/* As the suspend_* range is controlled by
+				 * userspace, we want an interruptible
+				 * wait.
+				 */
+				flush_signals(current);
+				prepare_to_wait(&conf->wait_for_overlap,
+						&w, TASK_INTERRUPTIBLE);
+				if (logical_sector >= mddev->suspend_lo &&
+				    logical_sector < mddev->suspend_hi)
+					schedule();
 				goto retry;
 			}
 
@@ -4452,7 +4460,7 @@ static raid5_conf_t *setup_conf(mddev_t *mddev)
 static int run(mddev_t *mddev)
 {
 	raid5_conf_t *conf;
-	int working_disks = 0;
+	int working_disks = 0, chunk_size;
 	mdk_rdev_t *rdev;
 
 	if (mddev->recovery_cp != MaxSector)
@@ -4607,6 +4615,14 @@ static int run(mddev_t *mddev)
 	md_set_array_sectors(mddev, raid5_size(mddev, 0, 0));
 
 	blk_queue_merge_bvec(mddev->queue, raid5_mergeable_bvec);
+	chunk_size = mddev->chunk_sectors << 9;
+	blk_queue_io_min(mddev->queue, chunk_size);
+	blk_queue_io_opt(mddev->queue, chunk_size *
+			 (conf->raid_disks - conf->max_degraded));
+
+	list_for_each_entry(rdev, &mddev->disks, same_set)
+		disk_stack_limits(mddev->gendisk, rdev->bdev,
+				  rdev->data_offset << 9);
 
 	return 0;
 abort: