Btrfs: rescan for qgroups
authorJan Schmidt <list.btrfs@jan-o-sch.net>
Thu, 25 Apr 2013 16:04:51 +0000 (16:04 +0000)
committerJosef Bacik <jbacik@fusionio.com>
Mon, 6 May 2013 19:55:19 +0000 (15:55 -0400)
If qgroup tracking is out of sync, a rescan operation can be started. It
iterates the complete extent tree and recalculates all qgroup tracking data.
This is an expensive operation and should not be used unless required.

A filesystem under rescan can still be umounted. The rescan continues on the
next mount.  Status information is provided with a separate ioctl while a
rescan operation is in progress.

Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
Signed-off-by: Josef Bacik <jbacik@fusionio.com>
fs/btrfs/ctree.h
fs/btrfs/disk-io.c
fs/btrfs/ioctl.c
fs/btrfs/qgroup.c
include/uapi/linux/btrfs.h

index 2c48f52..d9bed5f 100644 (file)
@@ -1021,9 +1021,9 @@ struct btrfs_block_group_item {
  */
 #define BTRFS_QGROUP_STATUS_FLAG_ON            (1ULL << 0)
 /*
- * SCANNING is set during the initialization phase
+ * RESCAN is set during the initialization phase
  */
-#define BTRFS_QGROUP_STATUS_FLAG_SCANNING      (1ULL << 1)
+#define BTRFS_QGROUP_STATUS_FLAG_RESCAN                (1ULL << 1)
 /*
  * Some qgroup entries are known to be out of date,
  * either because the configuration has changed in a way that
@@ -1052,7 +1052,7 @@ struct btrfs_qgroup_status_item {
         * only used during scanning to record the progress
         * of the scan. It contains a logical address
         */
-       __le64 scan;
+       __le64 rescan;
 } __attribute__ ((__packed__));
 
 struct btrfs_qgroup_info_item {
@@ -1603,6 +1603,11 @@ struct btrfs_fs_info {
        /* used by btrfs_qgroup_record_ref for an efficient tree traversal */
        u64 qgroup_seq;
 
+       /* qgroup rescan items */
+       struct mutex qgroup_rescan_lock; /* protects the progress item */
+       struct btrfs_key qgroup_rescan_progress;
+       struct btrfs_workers qgroup_rescan_workers;
+
        /* filesystem state */
        unsigned long fs_state;
 
@@ -2886,8 +2891,8 @@ BTRFS_SETGET_FUNCS(qgroup_status_version, struct btrfs_qgroup_status_item,
                   version, 64);
 BTRFS_SETGET_FUNCS(qgroup_status_flags, struct btrfs_qgroup_status_item,
                   flags, 64);
-BTRFS_SETGET_FUNCS(qgroup_status_scan, struct btrfs_qgroup_status_item,
-                  scan, 64);
+BTRFS_SETGET_FUNCS(qgroup_status_rescan, struct btrfs_qgroup_status_item,
+                  rescan, 64);
 
 /* btrfs_qgroup_info_item */
 BTRFS_SETGET_FUNCS(qgroup_info_generation, struct btrfs_qgroup_info_item,
@@ -3828,7 +3833,7 @@ int btrfs_quota_enable(struct btrfs_trans_handle *trans,
                       struct btrfs_fs_info *fs_info);
 int btrfs_quota_disable(struct btrfs_trans_handle *trans,
                        struct btrfs_fs_info *fs_info);
-int btrfs_quota_rescan(struct btrfs_fs_info *fs_info);
+int btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info);
 int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans,
                              struct btrfs_fs_info *fs_info, u64 src, u64 dst);
 int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans,
index 92c44ed..d96305e 100644 (file)
@@ -1976,6 +1976,7 @@ static void btrfs_stop_all_workers(struct btrfs_fs_info *fs_info)
        btrfs_stop_workers(&fs_info->caching_workers);
        btrfs_stop_workers(&fs_info->readahead_workers);
        btrfs_stop_workers(&fs_info->flush_workers);
+       btrfs_stop_workers(&fs_info->qgroup_rescan_workers);
 }
 
 /* helper to cleanup tree roots */
@@ -2267,6 +2268,7 @@ int open_ctree(struct super_block *sb,
        fs_info->qgroup_seq = 1;
        fs_info->quota_enabled = 0;
        fs_info->pending_quota_state = 0;
+       mutex_init(&fs_info->qgroup_rescan_lock);
 
        btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
        btrfs_init_free_cluster(&fs_info->data_alloc_cluster);
@@ -2476,6 +2478,8 @@ int open_ctree(struct super_block *sb,
        btrfs_init_workers(&fs_info->readahead_workers, "readahead",
                           fs_info->thread_pool_size,
                           &fs_info->generic_worker);
+       btrfs_init_workers(&fs_info->qgroup_rescan_workers, "qgroup-rescan", 1,
+                          &fs_info->generic_worker);
 
        /*
         * endios are largely parallel and should have a very
@@ -2510,6 +2514,7 @@ int open_ctree(struct super_block *sb,
        ret |= btrfs_start_workers(&fs_info->caching_workers);
        ret |= btrfs_start_workers(&fs_info->readahead_workers);
        ret |= btrfs_start_workers(&fs_info->flush_workers);
+       ret |= btrfs_start_workers(&fs_info->qgroup_rescan_workers);
        if (ret) {
                err = -ENOMEM;
                goto fail_sb_buffer;
index a74edc7..f5f6af3 100644 (file)
@@ -3701,12 +3701,10 @@ static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)
        }
 
        down_write(&root->fs_info->subvol_sem);
-       if (sa->cmd != BTRFS_QUOTA_CTL_RESCAN) {
-               trans = btrfs_start_transaction(root->fs_info->tree_root, 2);
-               if (IS_ERR(trans)) {
-                       ret = PTR_ERR(trans);
-                       goto out;
-               }
+       trans = btrfs_start_transaction(root->fs_info->tree_root, 2);
+       if (IS_ERR(trans)) {
+               ret = PTR_ERR(trans);
+               goto out;
        }
 
        switch (sa->cmd) {
@@ -3716,9 +3714,6 @@ static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)
        case BTRFS_QUOTA_CTL_DISABLE:
                ret = btrfs_quota_disable(trans, root->fs_info);
                break;
-       case BTRFS_QUOTA_CTL_RESCAN:
-               ret = btrfs_quota_rescan(root->fs_info);
-               break;
        default:
                ret = -EINVAL;
                break;
@@ -3727,11 +3722,9 @@ static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)
        if (copy_to_user(arg, sa, sizeof(*sa)))
                ret = -EFAULT;
 
-       if (trans) {
-               err = btrfs_commit_transaction(trans, root->fs_info->tree_root);
-               if (err && !ret)
-                       ret = err;
-       }
+       err = btrfs_commit_transaction(trans, root->fs_info->tree_root);
+       if (err && !ret)
+               ret = err;
 out:
        kfree(sa);
        up_write(&root->fs_info->subvol_sem);
@@ -3886,6 +3879,64 @@ drop_write:
        return ret;
 }
 
+static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg)
+{
+       struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
+       struct btrfs_ioctl_quota_rescan_args *qsa;
+       int ret;
+
+       if (!capable(CAP_SYS_ADMIN))
+               return -EPERM;
+
+       ret = mnt_want_write_file(file);
+       if (ret)
+               return ret;
+
+       qsa = memdup_user(arg, sizeof(*qsa));
+       if (IS_ERR(qsa)) {
+               ret = PTR_ERR(qsa);
+               goto drop_write;
+       }
+
+       if (qsa->flags) {
+               ret = -EINVAL;
+               goto out;
+       }
+
+       ret = btrfs_qgroup_rescan(root->fs_info);
+
+out:
+       kfree(qsa);
+drop_write:
+       mnt_drop_write_file(file);
+       return ret;
+}
+
+static long btrfs_ioctl_quota_rescan_status(struct file *file, void __user *arg)
+{
+       struct btrfs_root *root = BTRFS_I(fdentry(file)->d_inode)->root;
+       struct btrfs_ioctl_quota_rescan_args *qsa;
+       int ret = 0;
+
+       if (!capable(CAP_SYS_ADMIN))
+               return -EPERM;
+
+       qsa = kzalloc(sizeof(*qsa), GFP_NOFS);
+       if (!qsa)
+               return -ENOMEM;
+
+       if (root->fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
+               qsa->flags = 1;
+               qsa->progress = root->fs_info->qgroup_rescan_progress.objectid;
+       }
+
+       if (copy_to_user(arg, qsa, sizeof(*qsa)))
+               ret = -EFAULT;
+
+       kfree(qsa);
+       return ret;
+}
+
 static long btrfs_ioctl_set_received_subvol(struct file *file,
                                            void __user *arg)
 {
@@ -4124,6 +4175,10 @@ long btrfs_ioctl(struct file *file, unsigned int
                return btrfs_ioctl_qgroup_create(file, argp);
        case BTRFS_IOC_QGROUP_LIMIT:
                return btrfs_ioctl_qgroup_limit(file, argp);
+       case BTRFS_IOC_QUOTA_RESCAN:
+               return btrfs_ioctl_quota_rescan(file, argp);
+       case BTRFS_IOC_QUOTA_RESCAN_STATUS:
+               return btrfs_ioctl_quota_rescan_status(file, argp);
        case BTRFS_IOC_DEV_REPLACE:
                return btrfs_ioctl_dev_replace(root, argp);
        case BTRFS_IOC_GET_FSLABEL:
index 1fb7d8d..da84583 100644 (file)
 #include "locking.h"
 #include "ulist.h"
 #include "backref.h"
+#include "extent_io.h"
 
 /* TODO XXX FIXME
  *  - subvol delete -> delete when ref goes to 0? delete limits also?
  *  - reorganize keys
  *  - compressed
  *  - sync
- *  - rescan
  *  - copy also limits on subvol creation
  *  - limit
  *  - caches fuer ulists
@@ -98,6 +98,14 @@ struct btrfs_qgroup_list {
        struct btrfs_qgroup *member;
 };
 
+struct qgroup_rescan {
+       struct btrfs_work       work;
+       struct btrfs_fs_info    *fs_info;
+};
+
+static void qgroup_rescan_start(struct btrfs_fs_info *fs_info,
+                               struct qgroup_rescan *qscan);
+
 /* must be called with qgroup_ioctl_lock held */
 static struct btrfs_qgroup *find_qgroup_rb(struct btrfs_fs_info *fs_info,
                                           u64 qgroupid)
@@ -298,7 +306,20 @@ int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
                        }
                        fs_info->qgroup_flags = btrfs_qgroup_status_flags(l,
                                                                          ptr);
-                       /* FIXME read scan element */
+                       fs_info->qgroup_rescan_progress.objectid =
+                                       btrfs_qgroup_status_rescan(l, ptr);
+                       if (fs_info->qgroup_flags &
+                           BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
+                               struct qgroup_rescan *qscan =
+                                       kmalloc(sizeof(*qscan), GFP_NOFS);
+                               if (!qscan) {
+                                       ret = -ENOMEM;
+                                       goto out;
+                               }
+                               fs_info->qgroup_rescan_progress.type = 0;
+                               fs_info->qgroup_rescan_progress.offset = 0;
+                               qgroup_rescan_start(fs_info, qscan);
+                       }
                        goto next1;
                }
 
@@ -719,7 +740,8 @@ static int update_qgroup_status_item(struct btrfs_trans_handle *trans,
        ptr = btrfs_item_ptr(l, slot, struct btrfs_qgroup_status_item);
        btrfs_set_qgroup_status_flags(l, ptr, fs_info->qgroup_flags);
        btrfs_set_qgroup_status_generation(l, ptr, trans->transid);
-       /* XXX scan */
+       btrfs_set_qgroup_status_rescan(l, ptr,
+                               fs_info->qgroup_rescan_progress.objectid);
 
        btrfs_mark_buffer_dirty(l);
 
@@ -830,7 +852,7 @@ int btrfs_quota_enable(struct btrfs_trans_handle *trans,
        fs_info->qgroup_flags = BTRFS_QGROUP_STATUS_FLAG_ON |
                                BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
        btrfs_set_qgroup_status_flags(leaf, ptr, fs_info->qgroup_flags);
-       btrfs_set_qgroup_status_scan(leaf, ptr, 0);
+       btrfs_set_qgroup_status_rescan(leaf, ptr, 0);
 
        btrfs_mark_buffer_dirty(leaf);
 
@@ -944,10 +966,11 @@ out:
        return ret;
 }
 
-int btrfs_quota_rescan(struct btrfs_fs_info *fs_info)
+static void qgroup_dirty(struct btrfs_fs_info *fs_info,
+                        struct btrfs_qgroup *qgroup)
 {
-       /* FIXME */
-       return 0;
+       if (list_empty(&qgroup->dirty))
+               list_add(&qgroup->dirty, &fs_info->dirty_qgroups);
 }
 
 int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans,
@@ -1155,13 +1178,6 @@ out:
        return ret;
 }
 
-static void qgroup_dirty(struct btrfs_fs_info *fs_info,
-                        struct btrfs_qgroup *qgroup)
-{
-       if (list_empty(&qgroup->dirty))
-               list_add(&qgroup->dirty, &fs_info->dirty_qgroups);
-}
-
 /*
  * btrfs_qgroup_record_ref is called when the ref is added or deleted. it puts
  * the modification into a list that's later used by btrfs_end_transaction to
@@ -1390,6 +1406,15 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
                BUG();
        }
 
+       mutex_lock(&fs_info->qgroup_rescan_lock);
+       if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
+               if (fs_info->qgroup_rescan_progress.objectid <= node->bytenr) {
+                       mutex_unlock(&fs_info->qgroup_rescan_lock);
+                       return 0;
+               }
+       }
+       mutex_unlock(&fs_info->qgroup_rescan_lock);
+
        /*
         * the delayed ref sequence number we pass depends on the direction of
         * the operation. for add operations, we pass
@@ -1403,7 +1428,15 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
        if (ret < 0)
                return ret;
 
+       mutex_lock(&fs_info->qgroup_rescan_lock);
        spin_lock(&fs_info->qgroup_lock);
+       if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
+               if (fs_info->qgroup_rescan_progress.objectid <= node->bytenr) {
+                       ret = 0;
+                       goto unlock;
+               }
+       }
+
        quota_root = fs_info->quota_root;
        if (!quota_root)
                goto unlock;
@@ -1445,6 +1478,7 @@ int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
 
 unlock:
        spin_unlock(&fs_info->qgroup_lock);
+       mutex_unlock(&fs_info->qgroup_rescan_lock);
        ulist_free(roots);
        ulist_free(tmp);
 
@@ -1823,3 +1857,259 @@ void assert_qgroups_uptodate(struct btrfs_trans_handle *trans)
                (u32)trans->delayed_ref_elem.seq);
        BUG();
 }
+
+/*
+ * returns < 0 on error, 0 when more leafs are to be scanned.
+ * returns 1 when done, 2 when done and FLAG_INCONSISTENT was cleared.
+ */
+static int
+qgroup_rescan_leaf(struct qgroup_rescan *qscan, struct btrfs_path *path,
+                  struct btrfs_trans_handle *trans, struct ulist *tmp,
+                  struct extent_buffer *scratch_leaf)
+{
+       struct btrfs_key found;
+       struct btrfs_fs_info *fs_info = qscan->fs_info;
+       struct ulist *roots = NULL;
+       struct ulist_node *unode;
+       struct ulist_iterator uiter;
+       struct seq_list tree_mod_seq_elem = {};
+       u64 seq;
+       int slot;
+       int ret;
+
+       path->leave_spinning = 1;
+       mutex_lock(&fs_info->qgroup_rescan_lock);
+       ret = btrfs_search_slot_for_read(fs_info->extent_root,
+                                        &fs_info->qgroup_rescan_progress,
+                                        path, 1, 0);
+
+       pr_debug("current progress key (%llu %u %llu), search_slot ret %d\n",
+                (unsigned long long)fs_info->qgroup_rescan_progress.objectid,
+                fs_info->qgroup_rescan_progress.type,
+                (unsigned long long)fs_info->qgroup_rescan_progress.offset,
+                ret);
+
+       if (ret) {
+               /*
+                * The rescan is about to end, we will not be scanning any
+                * further blocks. We cannot unset the RESCAN flag here, because
+                * we want to commit the transaction if everything went well.
+                * To make the live accounting work in this phase, we set our
+                * scan progress pointer such that every real extent objectid
+                * will be smaller.
+                */
+               fs_info->qgroup_rescan_progress.objectid = (u64)-1;
+               btrfs_release_path(path);
+               mutex_unlock(&fs_info->qgroup_rescan_lock);
+               return ret;
+       }
+
+       btrfs_item_key_to_cpu(path->nodes[0], &found,
+                             btrfs_header_nritems(path->nodes[0]) - 1);
+       fs_info->qgroup_rescan_progress.objectid = found.objectid + 1;
+
+       btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
+       memcpy(scratch_leaf, path->nodes[0], sizeof(*scratch_leaf));
+       slot = path->slots[0];
+       btrfs_release_path(path);
+       mutex_unlock(&fs_info->qgroup_rescan_lock);
+
+       for (; slot < btrfs_header_nritems(scratch_leaf); ++slot) {
+               btrfs_item_key_to_cpu(scratch_leaf, &found, slot);
+               if (found.type != BTRFS_EXTENT_ITEM_KEY)
+                       continue;
+               ret = btrfs_find_all_roots(trans, fs_info, found.objectid,
+                                          tree_mod_seq_elem.seq, &roots);
+               if (ret < 0)
+                       goto out;
+               spin_lock(&fs_info->qgroup_lock);
+               seq = fs_info->qgroup_seq;
+               fs_info->qgroup_seq += roots->nnodes + 1; /* max refcnt */
+
+               ret = qgroup_account_ref_step1(fs_info, roots, tmp, seq);
+               if (ret) {
+                       spin_unlock(&fs_info->qgroup_lock);
+                       ulist_free(roots);
+                       goto out;
+               }
+
+               /*
+                * step2 of btrfs_qgroup_account_ref works from a single root,
+                * we're doing all at once here.
+                */
+               ulist_reinit(tmp);
+               ULIST_ITER_INIT(&uiter);
+               while ((unode = ulist_next(roots, &uiter))) {
+                       struct btrfs_qgroup *qg;
+
+                       qg = find_qgroup_rb(fs_info, unode->val);
+                       if (!qg)
+                               continue;
+
+                       ret = ulist_add(tmp, qg->qgroupid, (uintptr_t)qg,
+                                       GFP_ATOMIC);
+                       if (ret < 0) {
+                               spin_unlock(&fs_info->qgroup_lock);
+                               ulist_free(roots);
+                               goto out;
+                       }
+               }
+
+               /* this loop is similar to step 2 of btrfs_qgroup_account_ref */
+               ULIST_ITER_INIT(&uiter);
+               while ((unode = ulist_next(tmp, &uiter))) {
+                       struct btrfs_qgroup *qg;
+                       struct btrfs_qgroup_list *glist;
+
+                       qg = (struct btrfs_qgroup *)(uintptr_t) unode->aux;
+                       qg->rfer += found.offset;
+                       qg->rfer_cmpr += found.offset;
+                       WARN_ON(qg->tag >= seq);
+                       if (qg->refcnt - seq == roots->nnodes) {
+                               qg->excl += found.offset;
+                               qg->excl_cmpr += found.offset;
+                       }
+                       qgroup_dirty(fs_info, qg);
+
+                       list_for_each_entry(glist, &qg->groups, next_group) {
+                               ret = ulist_add(tmp, glist->group->qgroupid,
+                                               (uintptr_t)glist->group,
+                                               GFP_ATOMIC);
+                               if (ret < 0) {
+                                       spin_unlock(&fs_info->qgroup_lock);
+                                       ulist_free(roots);
+                                       goto out;
+                               }
+                       }
+               }
+
+               spin_unlock(&fs_info->qgroup_lock);
+               ulist_free(roots);
+               ret = 0;
+       }
+
+out:
+       btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
+
+       return ret;
+}
+
+static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
+{
+       struct qgroup_rescan *qscan = container_of(work, struct qgroup_rescan,
+                                                  work);
+       struct btrfs_path *path;
+       struct btrfs_trans_handle *trans = NULL;
+       struct btrfs_fs_info *fs_info = qscan->fs_info;
+       struct ulist *tmp = NULL;
+       struct extent_buffer *scratch_leaf = NULL;
+       int err = -ENOMEM;
+
+       path = btrfs_alloc_path();
+       if (!path)
+               goto out;
+       tmp = ulist_alloc(GFP_NOFS);
+       if (!tmp)
+               goto out;
+       scratch_leaf = kmalloc(sizeof(*scratch_leaf), GFP_NOFS);
+       if (!scratch_leaf)
+               goto out;
+
+       err = 0;
+       while (!err) {
+               trans = btrfs_start_transaction(fs_info->fs_root, 0);
+               if (IS_ERR(trans)) {
+                       err = PTR_ERR(trans);
+                       break;
+               }
+               if (!fs_info->quota_enabled) {
+                       err = -EINTR;
+               } else {
+                       err = qgroup_rescan_leaf(qscan, path, trans,
+                                                tmp, scratch_leaf);
+               }
+               if (err > 0)
+                       btrfs_commit_transaction(trans, fs_info->fs_root);
+               else
+                       btrfs_end_transaction(trans, fs_info->fs_root);
+       }
+
+out:
+       kfree(scratch_leaf);
+       ulist_free(tmp);
+       btrfs_free_path(path);
+       kfree(qscan);
+
+       mutex_lock(&fs_info->qgroup_rescan_lock);
+       fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
+
+       if (err == 2 &&
+           fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT) {
+               fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+       } else if (err < 0) {
+               fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+       }
+       mutex_unlock(&fs_info->qgroup_rescan_lock);
+
+       if (err >= 0) {
+               pr_info("btrfs: qgroup scan completed%s\n",
+                       err == 2 ? " (inconsistency flag cleared)" : "");
+       } else {
+               pr_err("btrfs: qgroup scan failed with %d\n", err);
+       }
+}
+
+static void
+qgroup_rescan_start(struct btrfs_fs_info *fs_info, struct qgroup_rescan *qscan)
+{
+       memset(&qscan->work, 0, sizeof(qscan->work));
+       qscan->work.func = btrfs_qgroup_rescan_worker;
+       qscan->fs_info = fs_info;
+
+       pr_info("btrfs: qgroup scan started\n");
+       btrfs_queue_worker(&fs_info->qgroup_rescan_workers, &qscan->work);
+}
+
+int
+btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info)
+{
+       int ret = 0;
+       struct rb_node *n;
+       struct btrfs_qgroup *qgroup;
+       struct qgroup_rescan *qscan = kmalloc(sizeof(*qscan), GFP_NOFS);
+
+       if (!qscan)
+               return -ENOMEM;
+
+       mutex_lock(&fs_info->qgroup_rescan_lock);
+       spin_lock(&fs_info->qgroup_lock);
+       if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN)
+               ret = -EINPROGRESS;
+       else if (!(fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON))
+               ret = -EINVAL;
+       if (ret) {
+               spin_unlock(&fs_info->qgroup_lock);
+               mutex_unlock(&fs_info->qgroup_rescan_lock);
+               kfree(qscan);
+               return ret;
+       }
+
+       fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_RESCAN;
+       memset(&fs_info->qgroup_rescan_progress, 0,
+               sizeof(fs_info->qgroup_rescan_progress));
+
+       /* clear all current qgroup tracking information */
+       for (n = rb_first(&fs_info->qgroup_tree); n; n = rb_next(n)) {
+               qgroup = rb_entry(n, struct btrfs_qgroup, node);
+               qgroup->rfer = 0;
+               qgroup->rfer_cmpr = 0;
+               qgroup->excl = 0;
+               qgroup->excl_cmpr = 0;
+       }
+       spin_unlock(&fs_info->qgroup_lock);
+       mutex_unlock(&fs_info->qgroup_rescan_lock);
+
+       qgroup_rescan_start(fs_info, qscan);
+
+       return 0;
+}
index 5e39e85..5ef0df5 100644 (file)
@@ -376,12 +376,18 @@ struct btrfs_ioctl_get_dev_stats {
 
 #define BTRFS_QUOTA_CTL_ENABLE 1
 #define BTRFS_QUOTA_CTL_DISABLE        2
-#define BTRFS_QUOTA_CTL_RESCAN 3
+#define BTRFS_QUOTA_CTL_RESCAN__NOTUSED        3
 struct btrfs_ioctl_quota_ctl_args {
        __u64 cmd;
        __u64 status;
 };
 
+struct btrfs_ioctl_quota_rescan_args {
+       __u64   flags;
+       __u64   progress;
+       __u64   reserved[6];
+};
+
 struct btrfs_ioctl_qgroup_assign_args {
        __u64 assign;
        __u64 src;
@@ -520,6 +526,10 @@ struct btrfs_ioctl_send_args {
                               struct btrfs_ioctl_qgroup_create_args)
 #define BTRFS_IOC_QGROUP_LIMIT _IOR(BTRFS_IOCTL_MAGIC, 43, \
                               struct btrfs_ioctl_qgroup_limit_args)
+#define BTRFS_IOC_QUOTA_RESCAN _IOW(BTRFS_IOCTL_MAGIC, 44, \
+                              struct btrfs_ioctl_quota_rescan_args)
+#define BTRFS_IOC_QUOTA_RESCAN_STATUS _IOR(BTRFS_IOCTL_MAGIC, 45, \
+                              struct btrfs_ioctl_quota_rescan_args)
 #define BTRFS_IOC_GET_FSLABEL _IOR(BTRFS_IOCTL_MAGIC, 49, \
                                   char[BTRFS_LABEL_SIZE])
 #define BTRFS_IOC_SET_FSLABEL _IOW(BTRFS_IOCTL_MAGIC, 50, \