ext4: Use tracepoints for mb_history trace file

The /proc/fs/ext4/<dev>/mb_history was maintained manually, and had a
number of problems: it required a largish amount of memory to be
allocated for each ext4 filesystem, and the s_mb_history_lock
introduced a CPU contention problem.  

By ripping out the mb_history code and replacing it with ftrace
tracepoints, and we get more functionality: timestamps, event
filtering, the ability to correlate mballoc history with other ext4
tracepoints, etc.
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>

Documentation/filesystems/proc.txt

@@ -1113,7 +1113,6 @@ Table 1-12: Files in /proc/fs/ext4/<devname>
 ..............................................................................
  File            Content                                        
  mb_groups       details of multiblock allocator buddy cache of free blocks
- mb_history      multiblock allocation history
 ..............................................................................
 
 

fs/ext4/ext4.h

@@ -65,6 +65,12 @@ typedef __u32 ext4_lblk_t;
 /* data type for block group number */
 typedef unsigned int ext4_group_t;
 
+/*
+ * Flags used in mballoc's allocation_context flags field.  
+ *
+ * Also used to show what's going on for debugging purposes when the
+ * flag field is exported via the traceport interface
+ */
 
 /* prefer goal again. length */
 #define EXT4_MB_HINT_MERGE		0x0001
@@ -971,14 +977,6 @@ struct ext4_sb_info {
 	unsigned long s_mb_last_group;
 	unsigned long s_mb_last_start;
 
-	/* history to debug policy */
-	struct ext4_mb_history *s_mb_history;
-	int s_mb_history_cur;
-	int s_mb_history_max;
-	int s_mb_history_num;
-	spinlock_t s_mb_history_lock;
-	int s_mb_history_filter;
-
 	/* stats for buddy allocator */
 	spinlock_t s_mb_pa_lock;
 	atomic_t s_bal_reqs;	/* number of reqs with len > 1 */

fs/ext4/mballoc.h

@@ -52,18 +52,8 @@ extern u8 mb_enable_debug;
 #define mb_debug(n, fmt, a...)
 #endif
 
-/*
- * with EXT4_MB_HISTORY mballoc stores last N allocations in memory
- * and you can monitor it in /proc/fs/ext4/<dev>/mb_history
- */
-#define EXT4_MB_HISTORY
 #define EXT4_MB_HISTORY_ALLOC		1	/* allocation */
 #define EXT4_MB_HISTORY_PREALLOC	2	/* preallocated blocks used */
-#define EXT4_MB_HISTORY_DISCARD		4	/* preallocation discarded */
-#define EXT4_MB_HISTORY_FREE		8	/* free */
-
-#define EXT4_MB_HISTORY_DEFAULT		(EXT4_MB_HISTORY_ALLOC | \
-					 EXT4_MB_HISTORY_PREALLOC)
 
 /*
  * How long mballoc can look for a best extent (in found extents)
@@ -217,22 +207,6 @@ struct ext4_allocation_context {
 #define AC_STATUS_FOUND		2
 #define AC_STATUS_BREAK		3
 
-struct ext4_mb_history {
-	struct ext4_free_extent orig;	/* orig allocation */
-	struct ext4_free_extent goal;	/* goal allocation */
-	struct ext4_free_extent result;	/* result allocation */
-	unsigned pid;
-	unsigned ino;
-	__u16 found;	/* how many extents have been found */
-	__u16 groups;	/* how many groups have been scanned */
-	__u16 tail;	/* what tail broke some buddy */
-	__u16 buddy;	/* buddy the tail ^^^ broke */
-	__u16 flags;
-	__u8 cr:3;	/* which phase the result extent was found at */
-	__u8 op:4;
-	__u8 merged:1;
-};
-
 struct ext4_buddy {
 	struct page *bd_buddy_page;
 	void *bd_buddy;
@@ -247,13 +221,6 @@ struct ext4_buddy {
 #define EXT4_MB_BITMAP(e4b)	((e4b)->bd_bitmap)
 #define EXT4_MB_BUDDY(e4b)	((e4b)->bd_buddy)
 
-#ifndef EXT4_MB_HISTORY
-static inline void ext4_mb_store_history(struct ext4_allocation_context *ac)
-{
-	return;
-}
-#endif
-
 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
 
 static inline ext4_fsblk_t ext4_grp_offs_to_block(struct super_block *sb,

fs/ext4/super.c

@@ -50,13 +50,6 @@
 #define CREATE_TRACE_POINTS
 #include <trace/events/ext4.h>
 
-static int default_mb_history_length = 1000;
-
-module_param_named(default_mb_history_length, default_mb_history_length,
-		   int, 0644);
-MODULE_PARM_DESC(default_mb_history_length,
-		 "Default number of entries saved for mb_history");
-
 struct proc_dir_entry *ext4_proc_root;
 static struct kset *ext4_kset;
 
@@ -1079,7 +1072,7 @@ enum {
 	Opt_journal_update, Opt_journal_dev,
 	Opt_journal_checksum, Opt_journal_async_commit,
 	Opt_abort, Opt_data_journal, Opt_data_ordered, Opt_data_writeback,
-	Opt_data_err_abort, Opt_data_err_ignore, Opt_mb_history_length,
+	Opt_data_err_abort, Opt_data_err_ignore,
 	Opt_usrjquota, Opt_grpjquota, Opt_offusrjquota, Opt_offgrpjquota,
 	Opt_jqfmt_vfsold, Opt_jqfmt_vfsv0, Opt_quota, Opt_noquota,
 	Opt_ignore, Opt_barrier, Opt_nobarrier, Opt_err, Opt_resize,
@@ -1126,7 +1119,6 @@ static const match_table_t tokens = {
 	{Opt_data_writeback, "data=writeback"},
 	{Opt_data_err_abort, "data_err=abort"},
 	{Opt_data_err_ignore, "data_err=ignore"},
-	{Opt_mb_history_length, "mb_history_length=%u"},
 	{Opt_offusrjquota, "usrjquota="},
 	{Opt_usrjquota, "usrjquota=%s"},
 	{Opt_offgrpjquota, "grpjquota="},
@@ -1367,13 +1359,6 @@ static int parse_options(char *options, struct super_block *sb,
 		case Opt_data_err_ignore:
 			clear_opt(sbi->s_mount_opt, DATA_ERR_ABORT);
 			break;
-		case Opt_mb_history_length:
-			if (match_int(&args[0], &option))
-				return 0;
-			if (option < 0)
-				return 0;
-			sbi->s_mb_history_max = option;
-			break;
 #ifdef CONFIG_QUOTA
 		case Opt_usrjquota:
 			qtype = USRQUOTA;
@@ -2435,7 +2420,6 @@ static int ext4_fill_super(struct super_block *sb, void *data, int silent)
 	sbi->s_commit_interval = JBD2_DEFAULT_MAX_COMMIT_AGE * HZ;
 	sbi->s_min_batch_time = EXT4_DEF_MIN_BATCH_TIME;
 	sbi->s_max_batch_time = EXT4_DEF_MAX_BATCH_TIME;
-	sbi->s_mb_history_max = default_mb_history_length;
 
 	set_opt(sbi->s_mount_opt, BARRIER);
 

include/trace/events/ext4.h

@@ -743,6 +743,169 @@ TRACE_EVENT(ext4_alloc_da_blocks,
 		  __entry->data_blocks, __entry->meta_blocks)
 );
 
+TRACE_EVENT(ext4_mballoc_alloc,
+	TP_PROTO(struct ext4_allocation_context *ac),
+
+	TP_ARGS(ac),
+
+	TP_STRUCT__entry(
+		__field(	dev_t,	dev			)
+		__field(	ino_t,	ino			)
+		__field(	__u16,	found			)
+		__field(	__u16,	groups			)
+		__field(	__u16,	buddy			)
+		__field(	__u16,	flags			)
+		__field(	__u16,	tail			)
+		__field(	__u8,	cr			)
+		__field(	__u32, 	orig_logical		)
+		__field(	  int,	orig_start		)
+		__field(	__u32, 	orig_group		)
+		__field(	  int,	orig_len		)
+		__field(	__u32, 	goal_logical		)
+		__field(	  int,	goal_start		)
+		__field(	__u32, 	goal_group		)
+		__field(	  int,	goal_len		)
+		__field(	__u32, 	result_logical		)
+		__field(	  int,	result_start		)
+		__field(	__u32, 	result_group		)
+		__field(	  int,	result_len		)
+	),
+
+	TP_fast_assign(
+		__entry->dev		= ac->ac_inode->i_sb->s_dev;
+		__entry->ino		= ac->ac_inode->i_ino;
+		__entry->found		= ac->ac_found;
+		__entry->flags		= ac->ac_flags;
+		__entry->groups		= ac->ac_groups_scanned;
+		__entry->buddy		= ac->ac_buddy;
+		__entry->tail		= ac->ac_tail;
+		__entry->cr		= ac->ac_criteria;
+		__entry->orig_logical	= ac->ac_o_ex.fe_logical;
+		__entry->orig_start	= ac->ac_o_ex.fe_start;
+		__entry->orig_group	= ac->ac_o_ex.fe_group;
+		__entry->orig_len	= ac->ac_o_ex.fe_len;
+		__entry->goal_logical	= ac->ac_g_ex.fe_logical;
+		__entry->goal_start	= ac->ac_g_ex.fe_start;
+		__entry->goal_group	= ac->ac_g_ex.fe_group;
+		__entry->goal_len	= ac->ac_g_ex.fe_len;
+		__entry->result_logical	= ac->ac_f_ex.fe_logical;
+		__entry->result_start	= ac->ac_f_ex.fe_start;
+		__entry->result_group	= ac->ac_f_ex.fe_group;
+		__entry->result_len	= ac->ac_f_ex.fe_len;
+	),
+
+	TP_printk("dev %s inode %lu orig %u/%d/%u@%u goal %u/%d/%u@%u "
+		  "result %u/%d/%u@%u blks %u grps %u cr %u flags 0x%04x "
+		  "tail %u broken %u",
+		  jbd2_dev_to_name(__entry->dev), (unsigned long) __entry->ino,
+		  __entry->orig_group, __entry->orig_start,
+		  __entry->orig_len, __entry->orig_logical,
+		  __entry->goal_group, __entry->goal_start,
+		  __entry->goal_len, __entry->goal_logical,
+		  __entry->result_group, __entry->result_start,
+		  __entry->result_len, __entry->result_logical,
+		  __entry->found, __entry->groups, __entry->cr,
+		  __entry->flags, __entry->tail,
+		  __entry->buddy ? 1 << __entry->buddy : 0)
+);
+
+TRACE_EVENT(ext4_mballoc_prealloc,
+	TP_PROTO(struct ext4_allocation_context *ac),
+
+	TP_ARGS(ac),
+
+	TP_STRUCT__entry(
+		__field(	dev_t,	dev			)
+		__field(	ino_t,	ino			)
+		__field(	__u32, 	orig_logical		)
+		__field(	  int,	orig_start		)
+		__field(	__u32, 	orig_group		)
+		__field(	  int,	orig_len		)
+		__field(	__u32, 	result_logical		)
+		__field(	  int,	result_start		)
+		__field(	__u32, 	result_group		)
+		__field(	  int,	result_len		)
+	),
+
+	TP_fast_assign(
+		__entry->dev		= ac->ac_inode->i_sb->s_dev;
+		__entry->ino		= ac->ac_inode->i_ino;
+		__entry->orig_logical	= ac->ac_o_ex.fe_logical;
+		__entry->orig_start	= ac->ac_o_ex.fe_start;
+		__entry->orig_group	= ac->ac_o_ex.fe_group;
+		__entry->orig_len	= ac->ac_o_ex.fe_len;
+		__entry->result_logical	= ac->ac_b_ex.fe_logical;
+		__entry->result_start	= ac->ac_b_ex.fe_start;
+		__entry->result_group	= ac->ac_b_ex.fe_group;
+		__entry->result_len	= ac->ac_b_ex.fe_len;
+	),
+
+	TP_printk("dev %s inode %lu orig %u/%d/%u@%u result %u/%d/%u@%u",
+		  jbd2_dev_to_name(__entry->dev), (unsigned long) __entry->ino,
+		  __entry->orig_group, __entry->orig_start,
+		  __entry->orig_len, __entry->orig_logical,
+		  __entry->result_group, __entry->result_start,
+		  __entry->result_len, __entry->result_logical)
+);
+
+TRACE_EVENT(ext4_mballoc_discard,
+	TP_PROTO(struct ext4_allocation_context *ac),
+
+	TP_ARGS(ac),
+
+	TP_STRUCT__entry(
+		__field(	dev_t,	dev			)
+		__field(	ino_t,	ino			)
+		__field(	__u32, 	result_logical		)
+		__field(	  int,	result_start		)
+		__field(	__u32, 	result_group		)
+		__field(	  int,	result_len		)
+	),
+
+	TP_fast_assign(
+		__entry->dev		= ac->ac_inode->i_sb->s_dev;
+		__entry->ino		= ac->ac_inode->i_ino;
+		__entry->result_logical	= ac->ac_b_ex.fe_logical;
+		__entry->result_start	= ac->ac_b_ex.fe_start;
+		__entry->result_group	= ac->ac_b_ex.fe_group;
+		__entry->result_len	= ac->ac_b_ex.fe_len;
+	),
+
+	TP_printk("dev %s inode %lu extent %u/%d/%u@%u ",
+		  jbd2_dev_to_name(__entry->dev), (unsigned long) __entry->ino,
+		  __entry->result_group, __entry->result_start,
+		  __entry->result_len, __entry->result_logical)
+);
+
+TRACE_EVENT(ext4_mballoc_free,
+	TP_PROTO(struct ext4_allocation_context *ac),
+
+	TP_ARGS(ac),
+
+	TP_STRUCT__entry(
+		__field(	dev_t,	dev			)
+		__field(	ino_t,	ino			)
+		__field(	__u32, 	result_logical		)
+		__field(	  int,	result_start		)
+		__field(	__u32, 	result_group		)
+		__field(	  int,	result_len		)
+	),
+
+	TP_fast_assign(
+		__entry->dev		= ac->ac_inode->i_sb->s_dev;
+		__entry->ino		= ac->ac_inode->i_ino;
+		__entry->result_logical	= ac->ac_b_ex.fe_logical;
+		__entry->result_start	= ac->ac_b_ex.fe_start;
+		__entry->result_group	= ac->ac_b_ex.fe_group;
+		__entry->result_len	= ac->ac_b_ex.fe_len;
+	),
+
+	TP_printk("dev %s inode %lu extent %u/%d/%u@%u ",
+		  jbd2_dev_to_name(__entry->dev), (unsigned long) __entry->ino,
+		  __entry->result_group, __entry->result_start,
+		  __entry->result_len, __entry->result_logical)
+);
+
 #endif /* _TRACE_EXT4_H */
 
 /* This part must be outside protection */