btrfs: qgroup: Add trace point for qgroup reserved space
authorQu Wenruo <quwenruo@cn.fujitsu.com>
Mon, 13 Mar 2017 07:52:08 +0000 (15:52 +0800)
committerDavid Sterba <dsterba@suse.com>
Tue, 18 Apr 2017 12:07:26 +0000 (14:07 +0200)
Introduce the following trace points:
qgroup_update_reserve
qgroup_meta_reserve

These trace points are handy to trace qgroup reserve space related
problems.

Also export btrfs_qgroup structure, as now we directly pass btrfs_qgroup
structure to trace points, so that structure needs to be exported.

Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
fs/btrfs/qgroup.c
fs/btrfs/qgroup.h
include/trace/events/btrfs.h

index 1d7942c..27a2296 100644 (file)
  *  - check all ioctl parameters
  */
 
-/*
- * one struct for each qgroup, organized in fs_info->qgroup_tree.
- */
-struct btrfs_qgroup {
-       u64 qgroupid;
-
-       /*
-        * state
-        */
-       u64 rfer;       /* referenced */
-       u64 rfer_cmpr;  /* referenced compressed */
-       u64 excl;       /* exclusive */
-       u64 excl_cmpr;  /* exclusive compressed */
-
-       /*
-        * limits
-        */
-       u64 lim_flags;  /* which limits are set */
-       u64 max_rfer;
-       u64 max_excl;
-       u64 rsv_rfer;
-       u64 rsv_excl;
-
-       /*
-        * reservation tracking
-        */
-       u64 reserved;
-
-       /*
-        * lists
-        */
-       struct list_head groups;  /* groups this group is member of */
-       struct list_head members; /* groups that are members of this group */
-       struct list_head dirty;   /* dirty groups */
-       struct rb_node node;      /* tree of qgroups */
-
-       /*
-        * temp variables for accounting operations
-        * Refer to qgroup_shared_accounting() for details.
-        */
-       u64 old_refcnt;
-       u64 new_refcnt;
-};
-
 static void btrfs_qgroup_update_old_refcnt(struct btrfs_qgroup *qg, u64 seq,
                                           int mod)
 {
@@ -1075,6 +1031,7 @@ static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
        qgroup->excl += sign * num_bytes;
        qgroup->excl_cmpr += sign * num_bytes;
        if (sign > 0) {
+               trace_qgroup_update_reserve(fs_info, qgroup, -(s64)num_bytes);
                if (WARN_ON(qgroup->reserved < num_bytes))
                        report_reserved_underflow(fs_info, qgroup, num_bytes);
                else
@@ -1100,6 +1057,8 @@ static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
                WARN_ON(sign < 0 && qgroup->excl < num_bytes);
                qgroup->excl += sign * num_bytes;
                if (sign > 0) {
+                       trace_qgroup_update_reserve(fs_info, qgroup,
+                                                   -(s64)num_bytes);
                        if (WARN_ON(qgroup->reserved < num_bytes))
                                report_reserved_underflow(fs_info, qgroup,
                                                          num_bytes);
@@ -2446,6 +2405,7 @@ retry:
 
                qg = unode_aux_to_qgroup(unode);
 
+               trace_qgroup_update_reserve(fs_info, qg, num_bytes);
                qg->reserved += num_bytes;
        }
 
@@ -2491,6 +2451,7 @@ void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
 
                qg = unode_aux_to_qgroup(unode);
 
+               trace_qgroup_update_reserve(fs_info, qg, -(s64)num_bytes);
                if (WARN_ON(qg->reserved < num_bytes))
                        report_reserved_underflow(fs_info, qg, num_bytes);
                else
@@ -2955,6 +2916,7 @@ int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
                return 0;
 
        BUG_ON(num_bytes != round_down(num_bytes, fs_info->nodesize));
+       trace_qgroup_meta_reserve(root, (s64)num_bytes);
        ret = qgroup_reserve(root, num_bytes, enforce);
        if (ret < 0)
                return ret;
@@ -2974,6 +2936,7 @@ void btrfs_qgroup_free_meta_all(struct btrfs_root *root)
        reserved = atomic64_xchg(&root->qgroup_meta_rsv, 0);
        if (reserved == 0)
                return;
+       trace_qgroup_meta_reserve(root, -(s64)reserved);
        btrfs_qgroup_free_refroot(fs_info, root->objectid, reserved);
 }
 
@@ -2988,6 +2951,7 @@ void btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes)
        BUG_ON(num_bytes != round_down(num_bytes, fs_info->nodesize));
        WARN_ON(atomic64_read(&root->qgroup_meta_rsv) < num_bytes);
        atomic64_sub(num_bytes, &root->qgroup_meta_rsv);
+       trace_qgroup_meta_reserve(root, -(s64)num_bytes);
        btrfs_qgroup_free_refroot(fs_info, root->objectid, num_bytes);
 }
 
index 96fc56e..31e468b 100644 (file)
@@ -62,6 +62,50 @@ struct btrfs_qgroup_extent_record {
 };
 
 /*
+ * one struct for each qgroup, organized in fs_info->qgroup_tree.
+ */
+struct btrfs_qgroup {
+       u64 qgroupid;
+
+       /*
+        * state
+        */
+       u64 rfer;       /* referenced */
+       u64 rfer_cmpr;  /* referenced compressed */
+       u64 excl;       /* exclusive */
+       u64 excl_cmpr;  /* exclusive compressed */
+
+       /*
+        * limits
+        */
+       u64 lim_flags;  /* which limits are set */
+       u64 max_rfer;
+       u64 max_excl;
+       u64 rsv_rfer;
+       u64 rsv_excl;
+
+       /*
+        * reservation tracking
+        */
+       u64 reserved;
+
+       /*
+        * lists
+        */
+       struct list_head groups;  /* groups this group is member of */
+       struct list_head members; /* groups that are members of this group */
+       struct list_head dirty;   /* dirty groups */
+       struct rb_node node;      /* tree of qgroups */
+
+       /*
+        * temp variables for accounting operations
+        * Refer to qgroup_shared_accounting() for details.
+        */
+       u64 old_refcnt;
+       u64 new_refcnt;
+};
+
+/*
  * For qgroup event trace points only
  */
 #define QGROUP_RESERVE         (1<<0)
index cef39e2..e379735 100644 (file)
@@ -25,6 +25,7 @@ struct extent_buffer;
 struct btrfs_work;
 struct __btrfs_workqueue;
 struct btrfs_qgroup_extent_record;
+struct btrfs_qgroup;
 
 #define show_ref_type(type)                                            \
        __print_symbolic(type,                                          \
@@ -1614,6 +1615,49 @@ TRACE_EVENT(qgroup_update_counters,
                  __entry->cur_new_count)
 );
 
+TRACE_EVENT(qgroup_update_reserve,
+
+       TP_PROTO(struct btrfs_fs_info *fs_info, struct btrfs_qgroup *qgroup,
+                s64 diff),
+
+       TP_ARGS(fs_info, qgroup, diff),
+
+       TP_STRUCT__entry_btrfs(
+               __field(        u64,    qgid                    )
+               __field(        u64,    cur_reserved            )
+               __field(        s64,    diff                    )
+       ),
+
+       TP_fast_assign_btrfs(fs_info,
+               __entry->qgid           = qgroup->qgroupid;
+               __entry->cur_reserved   = qgroup->reserved;
+               __entry->diff           = diff;
+       ),
+
+       TP_printk_btrfs("qgid=%llu cur_reserved=%llu diff=%lld",
+               __entry->qgid, __entry->cur_reserved, __entry->diff)
+);
+
+TRACE_EVENT(qgroup_meta_reserve,
+
+       TP_PROTO(struct btrfs_root *root, s64 diff),
+
+       TP_ARGS(root, diff),
+
+       TP_STRUCT__entry_btrfs(
+               __field(        u64,    refroot                 )
+               __field(        s64,    diff                    )
+       ),
+
+       TP_fast_assign_btrfs(root->fs_info,
+               __entry->refroot        = root->objectid;
+               __entry->diff           = diff;
+       ),
+
+       TP_printk_btrfs("refroot=%llu(%s) diff=%lld",
+               show_root_type(__entry->refroot), __entry->diff)
+);
+
 #endif /* _TRACE_BTRFS_H */
 
 /* This part must be outside protection */