f2fs: add block_age-based extent cache

This patch introduces a runtime hot/cold data separation method
for f2fs, in order to improve the accuracy for data temperature
classification, reduce the garbage collection overhead after
long-term data updates.

Enhanced hot/cold data separation can record data block update
frequency as "age" of the extent per inode, and take use of the age
info to indicate better temperature type for data block allocation:
 - It records total data blocks allocated since mount;
 - When file extent has been updated, it calculate the count of data
blocks allocated since last update as the age of the extent;
 - Before the data block allocated, it searches for the age info and
chooses the suitable segment for allocation.

Test and result:
 - Prepare: create about 30000 files
  * 3% for cold files (with cold file extension like .apk, from 3M to 10M)
  * 50% for warm files (with random file extension like .FcDxq, from 1K
to 4M)
  * 47% for hot files (with hot file extension like .db, from 1K to 256K)
 - create(5%)/random update(90%)/delete(5%) the files
  * total write amount is about 70G
  * fsync will be called for .db files, and buffered write will be used
for other files

The storage of test device is large enough(128G) so that it will not
switch to SSR mode during the test.

Benefit: dirty segment count increment reduce about 14%
 - before: Dirty +21110
 - after:  Dirty +18286

Signed-off-by: qixiaoyu1 <qixiaoyu1@xiaomi.com>
Signed-off-by: xiongping1 <xiongping1@xiaomi.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>

1 files changed, 83 insertions, 3 deletions

diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h
index 2bb37892d2ba..31d994e6b4ca 100644
--- a/include/trace/events/f2fs.h
+++ b/include/trace/events/f2fs.h
@@ -49,6 +49,7 @@ TRACE_DEFINE_ENUM(CP_TRIMMED);
 TRACE_DEFINE_ENUM(CP_PAUSE);
 TRACE_DEFINE_ENUM(CP_RESIZE);
 TRACE_DEFINE_ENUM(EX_READ);
+TRACE_DEFINE_ENUM(EX_BLOCK_AGE);
 
 #define show_block_type(type)						\
 	__print_symbolic(type,						\
@@ -155,6 +156,11 @@ TRACE_DEFINE_ENUM(EX_READ);
 		{ COMPRESS_ZSTD,	"ZSTD" },			\
 		{ COMPRESS_LZORLE,	"LZO-RLE" })
 
+#define show_extent_type(type)						\
+	__print_symbolic(type,						\
+		{ EX_READ,	"Read" },				\
+		{ EX_BLOCK_AGE,	"Block Age" })
+
 struct f2fs_sb_info;
 struct f2fs_io_info;
 struct extent_info;
@@ -1544,7 +1550,7 @@ TRACE_EVENT(f2fs_lookup_extent_tree_start,
 	TP_printk("dev = (%d,%d), ino = %lu, pgofs = %u, type = %s",
 		show_dev_ino(__entry),
 		__entry->pgofs,
-		__entry->type == EX_READ ? "Read" : "N/A")
+		show_extent_type(__entry->type))
 );
 
 TRACE_EVENT_CONDITION(f2fs_lookup_read_extent_tree_end,
@@ -1583,6 +1589,45 @@ TRACE_EVENT_CONDITION(f2fs_lookup_read_extent_tree_end,
 		__entry->blk)
 );
 
+TRACE_EVENT_CONDITION(f2fs_lookup_age_extent_tree_end,
+
+	TP_PROTO(struct inode *inode, unsigned int pgofs,
+						struct extent_info *ei),
+
+	TP_ARGS(inode, pgofs, ei),
+
+	TP_CONDITION(ei),
+
+	TP_STRUCT__entry(
+		__field(dev_t,	dev)
+		__field(ino_t,	ino)
+		__field(unsigned int, pgofs)
+		__field(unsigned int, fofs)
+		__field(unsigned int, len)
+		__field(unsigned long long, age)
+		__field(unsigned long long, blocks)
+	),
+
+	TP_fast_assign(
+		__entry->dev = inode->i_sb->s_dev;
+		__entry->ino = inode->i_ino;
+		__entry->pgofs = pgofs;
+		__entry->fofs = ei->fofs;
+		__entry->len = ei->len;
+		__entry->age = ei->age;
+		__entry->blocks = ei->last_blocks;
+	),
+
+	TP_printk("dev = (%d,%d), ino = %lu, pgofs = %u, "
+		"age_ext_info(fofs: %u, len: %u, age: %llu, blocks: %llu)",
+		show_dev_ino(__entry),
+		__entry->pgofs,
+		__entry->fofs,
+		__entry->len,
+		__entry->age,
+		__entry->blocks)
+);
+
 TRACE_EVENT(f2fs_update_read_extent_tree_range,
 
 	TP_PROTO(struct inode *inode, unsigned int pgofs, unsigned int len,
@@ -1618,6 +1663,41 @@ TRACE_EVENT(f2fs_update_read_extent_tree_range,
 		__entry->c_len)
 );
 
+TRACE_EVENT(f2fs_update_age_extent_tree_range,
+
+	TP_PROTO(struct inode *inode, unsigned int pgofs, unsigned int len,
+					unsigned long long age,
+					unsigned long long last_blks),
+
+	TP_ARGS(inode, pgofs, len, age, last_blks),
+
+	TP_STRUCT__entry(
+		__field(dev_t,	dev)
+		__field(ino_t,	ino)
+		__field(unsigned int, pgofs)
+		__field(unsigned int, len)
+		__field(unsigned long long, age)
+		__field(unsigned long long, blocks)
+	),
+
+	TP_fast_assign(
+		__entry->dev = inode->i_sb->s_dev;
+		__entry->ino = inode->i_ino;
+		__entry->pgofs = pgofs;
+		__entry->len = len;
+		__entry->age = age;
+		__entry->blocks = last_blks;
+	),
+
+	TP_printk("dev = (%d,%d), ino = %lu, pgofs = %u, "
+				"len = %u, age = %llu, blocks = %llu",
+		show_dev_ino(__entry),
+		__entry->pgofs,
+		__entry->len,
+		__entry->age,
+		__entry->blocks)
+);
+
 TRACE_EVENT(f2fs_shrink_extent_tree,
 
 	TP_PROTO(struct f2fs_sb_info *sbi, unsigned int node_cnt,
@@ -1643,7 +1723,7 @@ TRACE_EVENT(f2fs_shrink_extent_tree,
 		show_dev(__entry->dev),
 		__entry->node_cnt,
 		__entry->tree_cnt,
-		__entry->type == EX_READ ? "Read" : "N/A")
+		show_extent_type(__entry->type))
 );
 
 TRACE_EVENT(f2fs_destroy_extent_tree,
@@ -1670,7 +1750,7 @@ TRACE_EVENT(f2fs_destroy_extent_tree,
 	TP_printk("dev = (%d,%d), ino = %lu, destroyed: node_cnt = %u, type = %s",
 		show_dev_ino(__entry),
 		__entry->node_cnt,
-		__entry->type == EX_READ ? "Read" : "N/A")
+		show_extent_type(__entry->type))
 );
 
 DECLARE_EVENT_CLASS(f2fs_sync_dirty_inodes,