btrfs: qgroup: allow user to clear the limitation on qgroup
authorYang Dongsheng <yangds.fnst@cn.fujitsu.com>
Wed, 3 Jun 2015 06:57:32 +0000 (14:57 +0800)
committerChris Mason <clm@fb.com>
Tue, 30 Jun 2015 20:20:00 +0000 (13:20 -0700)
Currently, we can only set a limitation on a qgroup, but we
can not clear it.

This patch provide a choice to user to clear a limitation on
qgroup by passing a value of CLEAR_VALUE(-1) to kernel.

Reported-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Dongsheng Yang <yangds.fnst@cn.fujitsu.com>
Tested-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
fs/btrfs/qgroup.c

index d5f1f033b7a00f3c2e2826e336ae5ca5b3e31d4a..e9ace099162ce14d73d04523962465ee163b30c7 100644 (file)
@@ -1349,6 +1349,11 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
        struct btrfs_root *quota_root;
        struct btrfs_qgroup *qgroup;
        int ret = 0;
+       /* Sometimes we would want to clear the limit on this qgroup.
+        * To meet this requirement, we treat the -1 as a special value
+        * which tell kernel to clear the limit on this qgroup.
+        */
+       const u64 CLEAR_VALUE = -1;
 
        mutex_lock(&fs_info->qgroup_ioctl_lock);
        quota_root = fs_info->quota_root;
@@ -1364,14 +1369,42 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
        }
 
        spin_lock(&fs_info->qgroup_lock);
-       if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_RFER)
-               qgroup->max_rfer = limit->max_rfer;
-       if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_EXCL)
-               qgroup->max_excl = limit->max_excl;
-       if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_RFER)
-               qgroup->rsv_rfer = limit->rsv_rfer;
-       if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_EXCL)
-               qgroup->rsv_excl = limit->rsv_excl;
+       if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_RFER) {
+               if (limit->max_rfer == CLEAR_VALUE) {
+                       qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_MAX_RFER;
+                       limit->flags &= ~BTRFS_QGROUP_LIMIT_MAX_RFER;
+                       qgroup->max_rfer = 0;
+               } else {
+                       qgroup->max_rfer = limit->max_rfer;
+               }
+       }
+       if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_EXCL) {
+               if (limit->max_excl == CLEAR_VALUE) {
+                       qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_MAX_EXCL;
+                       limit->flags &= ~BTRFS_QGROUP_LIMIT_MAX_EXCL;
+                       qgroup->max_excl = 0;
+               } else {
+                       qgroup->max_excl = limit->max_excl;
+               }
+       }
+       if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_RFER) {
+               if (limit->rsv_rfer == CLEAR_VALUE) {
+                       qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_RSV_RFER;
+                       limit->flags &= ~BTRFS_QGROUP_LIMIT_RSV_RFER;
+                       qgroup->rsv_rfer = 0;
+               } else {
+                       qgroup->rsv_rfer = limit->rsv_rfer;
+               }
+       }
+       if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_EXCL) {
+               if (limit->rsv_excl == CLEAR_VALUE) {
+                       qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_RSV_EXCL;
+                       limit->flags &= ~BTRFS_QGROUP_LIMIT_RSV_EXCL;
+                       qgroup->rsv_excl = 0;
+               } else {
+                       qgroup->rsv_excl = limit->rsv_excl;
+               }
+       }
        qgroup->lim_flags |= limit->flags;
 
        spin_unlock(&fs_info->qgroup_lock);