md: use MD_RECOVERY_INTR instead of kthread_should_stop in resync thread.

We currently use kthread_should_stop() in various places in the
sync/reshape code to abort early.
However some places set MD_RECOVERY_INTR but don't immediately call
md_reap_sync_thread() (and we will shortly get another one).
When this happens we are relying on md_check_recovery() to reap the
thread and that only happen when it finishes normally.
So MD_RECOVERY_INTR must lead to a normal finish without the
kthread_should_stop() test.

So replace all relevant tests, and be more careful when the thread is
interrupted not to acknowledge that latest step in a reshape as it may
not be fully committed yet.

Also add a test on MD_RECOVERY_INTR in the 'is_mddev_idle' loop
so we don't wait have to wait for the speed to drop before we can abort.

Signed-off-by: NeilBrown <neilb@suse.de>

3 files changed, 34 insertions, 31 deletions

diff --git a/drivers/md/md.c b/drivers/md/md.c
index 1ca47b0f4779..a74045df7bab 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -7410,9 +7410,6 @@ void md_do_sync(struct md_thread *thread)
                 mddev->curr_resync = 2;
 
         try_again:
-                if (kthread_should_stop())
-                        set_bit(MD_RECOVERY_INTR, &mddev->recovery);
-
                 if (test_bit(MD_RECOVERY_INTR, &mddev->recovery))
                         goto skip;
                 for_each_mddev(mddev2, tmp) {
@@ -7437,7 +7434,7 @@ void md_do_sync(struct md_thread *thread)
                                  * be caught by 'softlockup'
                                  */
                                 prepare_to_wait(&resync_wait, &wq, TASK_INTERRUPTIBLE);
-                                if (!kthread_should_stop() &&
+                                if (!test_bit(MD_RECOVERY_INTR, &mddev->recovery) &&
                                     mddev2->curr_resync >= mddev->curr_resync) {
                                         printk(KERN_INFO "md: delaying %s of %s"
                                                " until %s has finished (they"
@@ -7513,7 +7510,7 @@ void md_do_sync(struct md_thread *thread)
         last_check = 0;
 
         if (j>2) {
-                printk(KERN_INFO 
+                printk(KERN_INFO
                        "md: resuming %s of %s from checkpoint.\n",
                        desc, mdname(mddev));
                 mddev->curr_resync = j;
@@ -7550,7 +7547,8 @@ void md_do_sync(struct md_thread *thread)
                         sysfs_notify(&mddev->kobj, NULL, "sync_completed");
                 }
 
-                while (j >= mddev->resync_max && !kthread_should_stop()) {
+                while (j >= mddev->resync_max &&
+                       !test_bit(MD_RECOVERY_INTR, &mddev->recovery)) {
                         /* As this condition is controlled by user-space,
                          * we can block indefinitely, so use '_interruptible'
                          * to avoid triggering warnings.
@@ -7558,17 +7556,18 @@ void md_do_sync(struct md_thread *thread)
                         flush_signals(current); /* just in case */
                         wait_event_interruptible(mddev->recovery_wait,
                                                  mddev->resync_max > j
-                                                 || kthread_should_stop());
+                                                 || test_bit(MD_RECOVERY_INTR,
+                                                             &mddev->recovery));
                 }
 
-                if (kthread_should_stop())
-                        goto interrupted;
+                if (test_bit(MD_RECOVERY_INTR, &mddev->recovery))
+                        break;
 
                 sectors = mddev->pers->sync_request(mddev, j, &skipped,
                                                   currspeed < speed_min(mddev));
                 if (sectors == 0) {
                         set_bit(MD_RECOVERY_INTR, &mddev->recovery);
-                        goto out;
+                        break;
                 }
 
                 if (!skipped) { /* actual IO requested */
@@ -7605,10 +7604,8 @@ void md_do_sync(struct md_thread *thread)
                         last_mark = next;
                 }
 
-
-                if (kthread_should_stop())
-                        goto interrupted;
-
+                if (test_bit(MD_RECOVERY_INTR, &mddev->recovery))
+                        break;
 
                 /*
                  * this loop exits only if either when we are slower than
@@ -7631,11 +7628,12 @@ void md_do_sync(struct md_thread *thread)
                         }
                 }
         }
-        printk(KERN_INFO "md: %s: %s done.\n",mdname(mddev), desc);
+        printk(KERN_INFO "md: %s: %s %s.\n",mdname(mddev), desc,
+               test_bit(MD_RECOVERY_INTR, &mddev->recovery)
+               ? "interrupted" : "done");
         /*
          * this also signals 'finished resyncing' to md_stop
          */
- out:
         blk_finish_plug(&plug);
         wait_event(mddev->recovery_wait, !atomic_read(&mddev->recovery_active));
 
@@ -7689,16 +7687,6 @@ void md_do_sync(struct md_thread *thread)
         set_bit(MD_RECOVERY_DONE, &mddev->recovery);
         md_wakeup_thread(mddev->thread);
         return;
-
- interrupted:
-        /*
-         * got a signal, exit.
-         */
-        printk(KERN_INFO
-               "md: md_do_sync() got signal ... exiting\n");
-        set_bit(MD_RECOVERY_INTR, &mddev->recovery);
-        goto out;
-
 }
 EXPORT_SYMBOL_GPL(md_do_sync);
 
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 73dc8a377522..4f87ba5f3a66 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -4386,7 +4386,11 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr,
                 set_bit(MD_CHANGE_DEVS, &mddev->flags);
                 md_wakeup_thread(mddev->thread);
                 wait_event(mddev->sb_wait, mddev->flags == 0 ||
-                           kthread_should_stop());
+                           test_bit(MD_RECOVERY_INTR, &mddev->recovery));
+                if (test_bit(MD_RECOVERY_INTR, &mddev->recovery)) {
+                        allow_barrier(conf);
+                        return sectors_done;
+                }
                 conf->reshape_safe = mddev->reshape_position;
                 allow_barrier(conf);
         }
diff --git a/drivers/md/raid5.c b/drivers/md/raid5.c
index 2c8907706109..02f6bc2ac2db 100644
--- a/drivers/md/raid5.c
+++ b/drivers/md/raid5.c
@@ -4842,14 +4842,19 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk
             time_after(jiffies, conf->reshape_checkpoint + 10*HZ)) {
                 /* Cannot proceed until we've updated the superblock... */
                 wait_event(conf->wait_for_overlap,
-                           atomic_read(&conf->reshape_stripes)==0);
+                           atomic_read(&conf->reshape_stripes)==0
+                           || test_bit(MD_RECOVERY_INTR, &mddev->recovery));
+                if (atomic_read(&conf->reshape_stripes) != 0)
+                        return 0;
                 mddev->reshape_position = conf->reshape_progress;
                 mddev->curr_resync_completed = sector_nr;
                 conf->reshape_checkpoint = jiffies;
                 set_bit(MD_CHANGE_DEVS, &mddev->flags);
                 md_wakeup_thread(mddev->thread);
                 wait_event(mddev->sb_wait, mddev->flags == 0 ||
-                           kthread_should_stop());
+                           test_bit(MD_RECOVERY_INTR, &mddev->recovery));
+                if (test_bit(MD_RECOVERY_INTR, &mddev->recovery))
+                        return 0;
                 spin_lock_irq(&conf->device_lock);
                 conf->reshape_safe = mddev->reshape_position;
                 spin_unlock_irq(&conf->device_lock);
@@ -4932,7 +4937,10 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk
             >= mddev->resync_max - mddev->curr_resync_completed) {
                 /* Cannot proceed until we've updated the superblock... */
                 wait_event(conf->wait_for_overlap,
-                           atomic_read(&conf->reshape_stripes) == 0);
+                           atomic_read(&conf->reshape_stripes) == 0
+                           || test_bit(MD_RECOVERY_INTR, &mddev->recovery));
+                if (atomic_read(&conf->reshape_stripes) != 0)
+                        goto ret;
                 mddev->reshape_position = conf->reshape_progress;
                 mddev->curr_resync_completed = sector_nr;
                 conf->reshape_checkpoint = jiffies;
@@ -4940,13 +4948,16 @@ static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *sk
                 md_wakeup_thread(mddev->thread);
                 wait_event(mddev->sb_wait,
                            !test_bit(MD_CHANGE_DEVS, &mddev->flags)
-                           || kthread_should_stop());
+                           || test_bit(MD_RECOVERY_INTR, &mddev->recovery));
+                if (test_bit(MD_RECOVERY_INTR, &mddev->recovery))
+                        goto ret;
                 spin_lock_irq(&conf->device_lock);
                 conf->reshape_safe = mddev->reshape_position;
                 spin_unlock_irq(&conf->device_lock);
                 wake_up(&conf->wait_for_overlap);
                 sysfs_notify(&mddev->kobj, NULL, "sync_completed");
         }
+ret:
         return reshape_sectors;
 }