btrfs: auto enable discard=async when possible

There's a request to automatically enable async discard for capable
devices. We can do that, the async mode is designed to wait for larger
freed extents and is not intrusive, with limits to iops, kbps or latency.

The status and tunables will be exported in /sys/fs/btrfs/FSID/discard .

The automatic selection is done if there's at least one discard capable
device in the filesystem (not capable devices are skipped). Mounting
with any other discard option will honor that option, notably mounting
with nodiscard will keep it disabled.

Link: https://lore.kernel.org/linux-btrfs/CAEg-Je_b1YtdsCR0zS5XZ_SbvJgN70ezwvRwLiCZgDGLbeMB=w@mail.gmail.com/
Reviewed-by: Boris Burkov <boris@bur.io>
Signed-off-by: David Sterba <dsterba@suse.com>

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 5adf76d..3485f60 100644 (file)
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -1396,6 +1396,7 @@ enum {
        BTRFS_MOUNT_DISCARD_ASYNC               = (1UL << 28),
        BTRFS_MOUNT_IGNOREBADROOTS              = (1UL << 29),
        BTRFS_MOUNT_IGNOREDATACSUMS             = (1UL << 30),
+       BTRFS_MOUNT_NODISCARD                   = (1UL << 31),
 };
 
 #define BTRFS_DEFAULT_COMMIT_INTERVAL  (30)

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 5705335..cf59d8f 100644 (file)
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -3747,6 +3747,20 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
        }
 
        /*
+        * For devices supporting discard turn on discard=async automatically,
+        * unless it's already set or disabled. This could be turned off by
+        * nodiscard for the same mount.
+        */
+       if (!(btrfs_test_opt(fs_info, DISCARD_SYNC) ||
+             btrfs_test_opt(fs_info, DISCARD_ASYNC) ||
+             btrfs_test_opt(fs_info, NODISCARD)) &&
+           fs_info->fs_devices->discardable) {
+               btrfs_set_and_info(fs_info, DISCARD_ASYNC,
+                                  "auto enabling async discard");
+               btrfs_clear_opt(fs_info->mount_opt, NODISCARD);
+       }
+
+       /*
         * Mount does not set all options immediately, we can do it now and do
         * not have to wait for transaction commit
         */

diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 3a33dd9..42e0d2f 100644 (file)
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -918,12 +918,14 @@ int btrfs_parse_options(struct btrfs_fs_info *info, char *options,
                                ret = -EINVAL;
                                goto out;
                        }
+                       btrfs_clear_opt(info->mount_opt, NODISCARD);
                        break;
                case Opt_nodiscard:
                        btrfs_clear_and_info(info, DISCARD_SYNC,
                                             "turning off discard");
                        btrfs_clear_and_info(info, DISCARD_ASYNC,
                                             "turning off async discard");
+                       btrfs_set_opt(info->mount_opt, NODISCARD);
                        break;
                case Opt_space_cache:
                case Opt_space_cache_version:

diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 635f45f..796e9f5 100644 (file)
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -641,6 +641,9 @@ static int btrfs_open_one_device(struct btrfs_fs_devices *fs_devices,
        if (!bdev_nonrot(bdev))
                fs_devices->rotating = true;
 
+       if (bdev_max_discard_sectors(bdev))
+               fs_devices->discardable = true;
+
        device->bdev = bdev;
        clear_bit(BTRFS_DEV_STATE_IN_FS_METADATA, &device->dev_state);
        device->mode = flags;

diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index 099def5..a20ee7d 100644 (file)
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -354,6 +354,8 @@ struct btrfs_fs_devices {
         * nonrot flag set
         */
        bool rotating;
+       /* Devices support TRIM/discard commands */
+       bool discardable;
 
        struct btrfs_fs_info *fs_info;
        /* sysfs kobjects */