md/raid10: fix io loss while replacement replace rdev

[ Upstream commit 2ae6aaf76912bae53c74b191569d2ab484f24bf3 ]

When removing a disk with replacement, the replacement will be used to
replace rdev. During this process, there is a brief window in which both
rdev and replacement are read as NULL in raid10_write_request(). This
will result in io not being submitted but it should be.

  //remove				//write
  raid10_remove_disk			raid10_write_request
   mirror->rdev = NULL
					 read rdev -> NULL
   mirror->rdev = mirror->replacement
   mirror->replacement = NULL
					 read replacement -> NULL

Fix it by reading replacement first and rdev later, meanwhile, use smp_mb()
to prevent memory reordering.

Fixes: 475b0321a4df ("md/raid10: writes should get directed to replacement as well as original.")
Signed-off-by: Li Nan <linan122@huawei.com>
Reviewed-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20230602091839.743798-3-linan666@huaweicloud.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index 0eb2eb8..386c16c 100644 (file)
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -779,8 +779,16 @@ static struct md_rdev *read_balance(struct r10conf *conf,
                disk = r10_bio->devs[slot].devnum;
                rdev = rcu_dereference(conf->mirrors[disk].replacement);
                if (rdev == NULL || test_bit(Faulty, &rdev->flags) ||
-                   r10_bio->devs[slot].addr + sectors > rdev->recovery_offset)
+                   r10_bio->devs[slot].addr + sectors >
+                   rdev->recovery_offset) {
+                       /*
+                        * Read replacement first to prevent reading both rdev
+                        * and replacement as NULL during replacement replace
+                        * rdev.
+                        */
+                       smp_mb();
                        rdev = rcu_dereference(conf->mirrors[disk].rdev);
+               }
                if (rdev == NULL ||
                    test_bit(Faulty, &rdev->flags))
                        continue;
@@ -1477,9 +1485,15 @@ static void raid10_write_request(struct mddev *mddev, struct bio *bio,
 
        for (i = 0;  i < conf->copies; i++) {
                int d = r10_bio->devs[i].devnum;
-               struct md_rdev *rdev = rcu_dereference(conf->mirrors[d].rdev);
-               struct md_rdev *rrdev = rcu_dereference(
-                       conf->mirrors[d].replacement);
+               struct md_rdev *rdev, *rrdev;
+
+               rrdev = rcu_dereference(conf->mirrors[d].replacement);
+               /*
+                * Read replacement first to prevent reading both rdev and
+                * replacement as NULL during replacement replace rdev.
+                */
+               smp_mb();
+               rdev = rcu_dereference(conf->mirrors[d].rdev);
                if (rdev == rrdev)
                        rrdev = NULL;
                if (rdev && (test_bit(Faulty, &rdev->flags)))