md: fix a potential deadlock of raid5/raid10 reshape

[sfrench/cifs-2.6.git] / drivers / md / md.c

diff --git a/drivers/md/md.c b/drivers/md/md.c
index 0081ace39a6494e8a168c310788348505aa6ca3b..254e44e44668f5fff8cc2d95bdb3b682450a204f 100644 (file)
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -801,6 +801,9 @@ void md_super_write(struct mddev *mddev, struct md_rdev *rdev,
        struct bio *bio;
        int ff = 0;
 
+       if (!page)
+               return;
+
        if (test_bit(Faulty, &rdev->flags))
                return;
 
@@ -5452,6 +5455,7 @@ int md_run(struct mddev *mddev)
         * the only valid external interface is through the md
         * device.
         */
+       mddev->has_superblocks = false;
        rdev_for_each(rdev, mddev) {
                if (test_bit(Faulty, &rdev->flags))
                        continue;
@@ -5465,6 +5469,9 @@ int md_run(struct mddev *mddev)
                                set_disk_ro(mddev->gendisk, 1);
                }
 
+               if (rdev->sb_page)
+                       mddev->has_superblocks = true;
+
                /* perform some consistency tests on the device.
                 * We don't want the data to overlap the metadata,
                 * Internal Bitmap issues have been handled elsewhere.
@@ -5497,8 +5504,10 @@ int md_run(struct mddev *mddev)
        }
        if (mddev->sync_set == NULL) {
                mddev->sync_set = bioset_create(BIO_POOL_SIZE, 0, BIOSET_NEED_BVECS);
-               if (!mddev->sync_set)
-                       return -ENOMEM;
+               if (!mddev->sync_set) {
+                       err = -ENOMEM;
+                       goto abort;
+               }
        }
 
        spin_lock(&pers_lock);
@@ -5511,7 +5520,8 @@ int md_run(struct mddev *mddev)
                else
                        pr_warn("md: personality for level %s is not loaded!\n",
                                mddev->clevel);
-               return -EINVAL;
+               err = -EINVAL;
+               goto abort;
        }
        spin_unlock(&pers_lock);
        if (mddev->level != pers->level) {
@@ -5524,7 +5534,8 @@ int md_run(struct mddev *mddev)
            pers->start_reshape == NULL) {
                /* This personality cannot handle reshaping... */
                module_put(pers->owner);
-               return -EINVAL;
+               err = -EINVAL;
+               goto abort;
        }
 
        if (pers->sync_request) {
@@ -5593,7 +5604,7 @@ int md_run(struct mddev *mddev)
                mddev->private = NULL;
                module_put(pers->owner);
                bitmap_destroy(mddev);
-               return err;
+               goto abort;
        }
        if (mddev->queue) {
                bool nonrot = true;
@@ -5655,6 +5666,18 @@ int md_run(struct mddev *mddev)
        sysfs_notify_dirent_safe(mddev->sysfs_action);
        sysfs_notify(&mddev->kobj, NULL, "degraded");
        return 0;
+
+abort:
+       if (mddev->bio_set) {
+               bioset_free(mddev->bio_set);
+               mddev->bio_set = NULL;
+       }
+       if (mddev->sync_set) {
+               bioset_free(mddev->sync_set);
+               mddev->sync_set = NULL;
+       }
+
+       return err;
 }
 EXPORT_SYMBOL_GPL(md_run);
 
@@ -7891,14 +7914,14 @@ static __poll_t mdstat_poll(struct file *filp, poll_table *wait)
        __poll_t mask;
 
        if (md_unloading)
-               return POLLIN|POLLRDNORM|POLLERR|POLLPRI;
+               return EPOLLIN|EPOLLRDNORM|EPOLLERR|EPOLLPRI;
        poll_wait(filp, &md_event_waiters, wait);
 
        /* always allow read */
-       mask = POLLIN | POLLRDNORM;
+       mask = EPOLLIN | EPOLLRDNORM;
 
        if (seq->poll_event != atomic_read(&md_event_count))
-               mask |= POLLERR | POLLPRI;
+               mask |= EPOLLERR | EPOLLPRI;
        return mask;
 }
 
@@ -8049,6 +8072,7 @@ EXPORT_SYMBOL(md_done_sync);
 bool md_write_start(struct mddev *mddev, struct bio *bi)
 {
        int did_change = 0;
+
        if (bio_data_dir(bi) != WRITE)
                return true;
 
@@ -8081,6 +8105,8 @@ bool md_write_start(struct mddev *mddev, struct bio *bi)
        rcu_read_unlock();
        if (did_change)
                sysfs_notify_dirent_safe(mddev->sysfs_state);
+       if (!mddev->has_superblocks)
+               return true;
        wait_event(mddev->sb_wait,
                   !test_bit(MD_SB_CHANGE_PENDING, &mddev->sb_flags) ||
                   mddev->suspended);
@@ -8543,6 +8569,19 @@ void md_do_sync(struct md_thread *thread)
        set_mask_bits(&mddev->sb_flags, 0,
                      BIT(MD_SB_CHANGE_PENDING) | BIT(MD_SB_CHANGE_DEVS));
 
+       if (test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery) &&
+                       !test_bit(MD_RECOVERY_INTR, &mddev->recovery) &&
+                       mddev->delta_disks > 0 &&
+                       mddev->pers->finish_reshape &&
+                       mddev->pers->size &&
+                       mddev->queue) {
+               mddev_lock_nointr(mddev);
+               md_set_array_sectors(mddev, mddev->pers->size(mddev, 0, 0));
+               mddev_unlock(mddev);
+               set_capacity(mddev->gendisk, mddev->array_sectors);
+               revalidate_disk(mddev->gendisk);
+       }
+
        spin_lock(&mddev->lock);
        if (!test_bit(MD_RECOVERY_INTR, &mddev->recovery)) {
                /* We completed so min/max setting can be forgotten if used. */
@@ -8569,6 +8608,10 @@ static int remove_and_add_spares(struct mddev *mddev,
        int removed = 0;
        bool remove_some = false;
 
+       if (this && test_bit(MD_RECOVERY_RUNNING, &mddev->recovery))
+               /* Mustn't remove devices when resync thread is running */
+               return 0;
+
        rdev_for_each(rdev, mddev) {
                if ((this == NULL || rdev == this) &&
                    rdev->raid_disk >= 0 &&