]> www.infradead.org Git - users/jedix/linux-maple.git/commitdiff
md/raid1-10: factor out a new helper raid1_should_read_first()
authorYu Kuai <yukuai3@huawei.com>
Thu, 29 Feb 2024 09:57:09 +0000 (17:57 +0800)
committerSong Liu <song@kernel.org>
Fri, 1 Mar 2024 06:49:46 +0000 (22:49 -0800)
If resync is in progress, read_balance() should find the first usable
disk, otherwise, data could be inconsistent after resync is done. raid1
and raid10 implement the same checking, hence factor out the checking
to make code cleaner.

Noted that raid1 is using 'mddev->recovery_cp', which is updated after
all resync IO is done, while raid10 is using 'conf->next_resync', which
is inaccurate because raid10 update it before submitting resync IO.
Fortunately, raid10 read IO can't concurrent with resync IO, hence there
is no problem. And this patch also switch raid10 to use
'mddev->recovery_cp'.

Co-developed-by: Paul Luse <paul.e.luse@linux.intel.com>
Signed-off-by: Paul Luse <paul.e.luse@linux.intel.com>
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Xiao Ni <xni@redhat.com>
Signed-off-by: Song Liu <song@kernel.org>
Link: https://lore.kernel.org/r/20240229095714.926789-7-yukuai1@huaweicloud.com
drivers/md/raid1-10.c
drivers/md/raid1.c
drivers/md/raid10.c

index 9bc0f0022a6c11a0432c549806d3d64d2d549509..2ea1710a3b705e9c68d7ed4a05fcd0bc16e32e2b 100644 (file)
@@ -276,3 +276,23 @@ static inline int raid1_check_read_range(struct md_rdev *rdev,
        *len = first_bad + bad_sectors - this_sector;
        return 0;
 }
+
+/*
+ * Check if read should choose the first rdev.
+ *
+ * Balance on the whole device if no resync is going on (recovery is ok) or
+ * below the resync window. Otherwise, take the first readable disk.
+ */
+static inline bool raid1_should_read_first(struct mddev *mddev,
+                                          sector_t this_sector, int len)
+{
+       if ((mddev->recovery_cp < this_sector + len))
+               return true;
+
+       if (mddev_is_clustered(mddev) &&
+           md_cluster_ops->area_resyncing(mddev, READ, this_sector,
+                                          this_sector + len))
+               return true;
+
+       return false;
+}
index fa86d9fdb16ff44fec87673dcb830ee9374695d3..30f467bb48fdc2bd777735f51e7d7d31f7f1dd22 100644 (file)
@@ -605,11 +605,6 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect
        struct md_rdev *rdev;
        int choose_first;
 
-       /*
-        * Check if we can balance. We can balance on the whole
-        * device if no resync is going on, or below the resync window.
-        * We take the first readable disk when above the resync window.
-        */
  retry:
        sectors = r1_bio->sectors;
        best_disk = -1;
@@ -619,16 +614,10 @@ static int read_balance(struct r1conf *conf, struct r1bio *r1_bio, int *max_sect
        best_pending_disk = -1;
        min_pending = UINT_MAX;
        best_good_sectors = 0;
+       choose_first = raid1_should_read_first(conf->mddev, this_sector,
+                                              sectors);
        clear_bit(R1BIO_FailFast, &r1_bio->state);
 
-       if ((conf->mddev->recovery_cp < this_sector + sectors) ||
-           (mddev_is_clustered(conf->mddev) &&
-           md_cluster_ops->area_resyncing(conf->mddev, READ, this_sector,
-                   this_sector + sectors)))
-               choose_first = 1;
-       else
-               choose_first = 0;
-
        for (disk = 0 ; disk < conf->raid_disks * 2 ; disk++) {
                sector_t dist;
                sector_t first_bad;
index d5a7a621f0f0b429786e6446263b5823d81af17f..8aecdb1ccc169aeb55219b28ff5a94f9078fcfdf 100644 (file)
@@ -748,17 +748,8 @@ static struct md_rdev *read_balance(struct r10conf *conf,
        best_good_sectors = 0;
        do_balance = 1;
        clear_bit(R10BIO_FailFast, &r10_bio->state);
-       /*
-        * Check if we can balance. We can balance on the whole
-        * device if no resync is going on (recovery is ok), or below
-        * the resync window. We take the first readable disk when
-        * above the resync window.
-        */
-       if ((conf->mddev->recovery_cp < MaxSector
-            && (this_sector + sectors >= conf->next_resync)) ||
-           (mddev_is_clustered(conf->mddev) &&
-            md_cluster_ops->area_resyncing(conf->mddev, READ, this_sector,
-                                           this_sector + sectors)))
+
+       if (raid1_should_read_first(conf->mddev, this_sector, sectors))
                do_balance = 0;
 
        for (slot = 0; slot < conf->copies ; slot++) {