In some contexts, like in sysfs handlers, we don't want to trigger a
transaction commit. It's a heavy operation, we don't know what external
locks may be taken. Instead, make it possible to finish the operation
through sync syscall or SYNC_FS ioctl.
Signed-off-by: David Sterba <dsterba@xxxxxxx>
---
fs/btrfs/ctree.h | 1 +
fs/btrfs/transaction.c | 5 +++++
2 files changed, 6 insertions(+)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 1c9157e4ab0c..817fc19b8159 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -2132,6 +2132,7 @@ struct btrfs_ioctl_defrag_range_args {
#define BTRFS_PENDING_SET_INODE_MAP_CACHE (0)
#define BTRFS_PENDING_CLEAR_INODE_MAP_CACHE (1)
+#define BTRFS_PENDING_COMMIT (2)
#define btrfs_test_pending(info, opt) \
test_bit(BTRFS_PENDING_##opt, &(info)->pending_changes)
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 703238ed7337..9d8e2b8d12b4 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -2036,6 +2036,11 @@ void btrfs_apply_pending_changes(struct btrfs_fs_info *fs_info)
btrfs_clear_opt(fs_info->mount_opt, INODE_MAP_CACHE);
prev &= ~bit;
+ bit = 1 << BTRFS_PENDING_COMMIT;
+ if (prev & bit)
+ btrfs_debug(fs_info, "pending commit done");
+ prev &= ~bit;
+
if (prev)
btrfs_warn(fs_info,
"unknown pending changes left 0x%lx, ignoring", prev);
--
1.8.4.5
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
