f2fs: add page-order information for large folio reads in iostat

Track read folio counts by order in F2FS iostat sysfs and tracepoints.

Signed-off-by: Daniel Lee <[email protected]>
Reviewed-by: Chao Yu <[email protected]>
Signed-off-by: Jaegeuk Kim <[email protected]>

Author: Daniel Lee

fs/f2fs/data.c

@@ -2508,6 +2508,8 @@ static int f2fs_read_data_large_folio(struct inode *inode,
 	if (!folio)
 		goto out;
 
+	f2fs_update_read_folio_count(F2FS_I_SB(inode), folio);
+
 	folio_in_bio = false;
 	index = folio->index;
 	offset = 0;
@@ -2682,6 +2684,8 @@ static int f2fs_mpage_readpages(struct inode *inode, struct fsverity_info *vi,
 			prefetchw(&folio->flags);
 		}
 
+		f2fs_update_read_folio_count(F2FS_I_SB(inode), folio);
+
 #ifdef CONFIG_F2FS_FS_COMPRESSION
 		index = folio->index;
 

fs/f2fs/f2fs.h

@@ -10,6 +10,7 @@
 
 #include <linux/uio.h>
 #include <linux/types.h>
+#include <linux/mmzone.h>
 #include <linux/page-flags.h>
 #include <linux/slab.h>
 #include <linux/crc32.h>
@@ -2034,6 +2035,8 @@ struct f2fs_sb_info {
 	unsigned long long iostat_count[NR_IO_TYPE];
 	unsigned long long iostat_bytes[NR_IO_TYPE];
 	unsigned long long prev_iostat_bytes[NR_IO_TYPE];
+	unsigned long long iostat_read_folio_count[NR_PAGE_ORDERS];
+	unsigned long long prev_iostat_read_folio_count[NR_PAGE_ORDERS];
 	bool iostat_enable;
 	unsigned long iostat_next_period;
 	unsigned int iostat_period_ms;

fs/f2fs/iostat.c

@@ -34,6 +34,7 @@ int __maybe_unused iostat_info_seq_show(struct seq_file *seq, void *offset)
 {
 	struct super_block *sb = seq->private;
 	struct f2fs_sb_info *sbi = F2FS_SB(sb);
+	int i;
 
 	if (!sbi->iostat_enable)
 		return 0;
@@ -76,6 +77,12 @@ int __maybe_unused iostat_info_seq_show(struct seq_file *seq, void *offset)
 	IOSTAT_INFO_SHOW("fs node", FS_NODE_READ_IO);
 	IOSTAT_INFO_SHOW("fs meta", FS_META_READ_IO);
 
+	/* print read folio order stats */
+	seq_printf(seq, "%-23s", "fs read folio order:");
+	for (i = 0; i < NR_PAGE_ORDERS; i++)
+		seq_printf(seq, " %llu", sbi->iostat_read_folio_count[i]);
+	seq_putc(seq, '\n');
+
 	/* print other IOs */
 	seq_puts(seq, "[OTHER]\n");
 	IOSTAT_INFO_SHOW("fs discard", FS_DISCARD_IO);
@@ -113,6 +120,7 @@ static inline void __record_iostat_latency(struct f2fs_sb_info *sbi)
 static inline void f2fs_record_iostat(struct f2fs_sb_info *sbi)
 {
 	unsigned long long iostat_diff[NR_IO_TYPE];
+	unsigned long long read_folio_count_diff[NR_PAGE_ORDERS];
 	int i;
 	unsigned long flags;
 
@@ -133,9 +141,15 @@ static inline void f2fs_record_iostat(struct f2fs_sb_info *sbi)
 				sbi->prev_iostat_bytes[i];
 		sbi->prev_iostat_bytes[i] = sbi->iostat_bytes[i];
 	}
+
+	for (i = 0; i < NR_PAGE_ORDERS; i++) {
+		read_folio_count_diff[i] = sbi->iostat_read_folio_count[i] -
+					sbi->prev_iostat_read_folio_count[i];
+		sbi->prev_iostat_read_folio_count[i] = sbi->iostat_read_folio_count[i];
+	}
 	spin_unlock_irqrestore(&sbi->iostat_lock, flags);
 
-	trace_f2fs_iostat(sbi, iostat_diff);
+	trace_f2fs_iostat(sbi, iostat_diff, read_folio_count_diff);
 
 	__record_iostat_latency(sbi);
 }
@@ -151,6 +165,10 @@ void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
 		sbi->iostat_bytes[i] = 0;
 		sbi->prev_iostat_bytes[i] = 0;
 	}
+	for (i = 0; i < NR_PAGE_ORDERS; i++) {
+		sbi->iostat_read_folio_count[i] = 0;
+		sbi->prev_iostat_read_folio_count[i] = 0;
+	}
 	spin_unlock_irq(&sbi->iostat_lock);
 
 	spin_lock_irq(&sbi->iostat_lat_lock);
@@ -165,6 +183,24 @@ static inline void __f2fs_update_iostat(struct f2fs_sb_info *sbi,
 	sbi->iostat_count[type]++;
 }
 
+void f2fs_update_read_folio_count(struct f2fs_sb_info *sbi, struct folio *folio)
+{
+	unsigned int order = folio_order(folio);
+	unsigned long flags;
+
+	if (!sbi->iostat_enable)
+		return;
+
+	if (order >= NR_PAGE_ORDERS)
+		order = NR_PAGE_ORDERS - 1;
+
+	spin_lock_irqsave(&sbi->iostat_lock, flags);
+	sbi->iostat_read_folio_count[order]++;
+	spin_unlock_irqrestore(&sbi->iostat_lock, flags);
+
+	f2fs_record_iostat(sbi);
+}
+
 void f2fs_update_iostat(struct f2fs_sb_info *sbi, struct inode *inode,
 			enum iostat_type type, unsigned long long io_bytes)
 {

fs/f2fs/iostat.h

@@ -34,6 +34,8 @@ extern int __maybe_unused iostat_info_seq_show(struct seq_file *seq,
 extern void f2fs_reset_iostat(struct f2fs_sb_info *sbi);
 extern void f2fs_update_iostat(struct f2fs_sb_info *sbi, struct inode *inode,
 			enum iostat_type type, unsigned long long io_bytes);
+extern void f2fs_update_read_folio_count(struct f2fs_sb_info *sbi,
+			struct folio *folio);
 
 struct bio_iostat_ctx {
 	struct f2fs_sb_info *sbi;
@@ -68,6 +70,8 @@ extern void f2fs_destroy_iostat(struct f2fs_sb_info *sbi);
 #else
 static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, struct inode *inode,
 		enum iostat_type type, unsigned long long io_bytes) {}
+static inline void f2fs_update_read_folio_count(struct f2fs_sb_info *sbi,
+		struct folio *folio) {}
 static inline void iostat_update_and_unbind_ctx(struct bio *bio) {}
 static inline void iostat_alloc_and_bind_ctx(struct f2fs_sb_info *sbi,
 		struct bio *bio, struct bio_post_read_ctx *ctx) {}

include/trace/events/f2fs.h

@@ -2116,9 +2116,10 @@ DEFINE_EVENT(f2fs_zip_end, f2fs_decompress_pages_end,
 #ifdef CONFIG_F2FS_IOSTAT
 TRACE_EVENT(f2fs_iostat,
 
-	TP_PROTO(struct f2fs_sb_info *sbi, unsigned long long *iostat),
+	TP_PROTO(struct f2fs_sb_info *sbi, unsigned long long *iostat,
+			unsigned long long *read_folio_count),
 
-	TP_ARGS(sbi, iostat),
+	TP_ARGS(sbi, iostat, read_folio_count),
 
 	TP_STRUCT__entry(
 		__field(dev_t,	dev)
@@ -2150,6 +2151,7 @@ TRACE_EVENT(f2fs_iostat,
 		__field(unsigned long long,	fs_mrio)
 		__field(unsigned long long,	fs_discard)
 		__field(unsigned long long,	fs_reset_zone)
+		__array(unsigned long long,	read_folio_count, 11)
 	),
 
 	TP_fast_assign(
@@ -2182,6 +2184,9 @@ TRACE_EVENT(f2fs_iostat,
 		__entry->fs_mrio	= iostat[FS_META_READ_IO];
 		__entry->fs_discard	= iostat[FS_DISCARD_IO];
 		__entry->fs_reset_zone	= iostat[FS_ZONE_RESET_IO];
+		memset(__entry->read_folio_count, 0, sizeof(__entry->read_folio_count));
+		memcpy(__entry->read_folio_count, read_folio_count,
+				sizeof(unsigned long long) * min_t(int, NR_PAGE_ORDERS, 11));
 	),
 
 	TP_printk("dev = (%d,%d), "
@@ -2194,7 +2199,9 @@ TRACE_EVENT(f2fs_iostat,
 		"app [read=%llu (direct=%llu, buffered=%llu), mapped=%llu], "
 		"compr(buffered=%llu, mapped=%llu)], "
 		"fs [data=%llu, (gc_data=%llu, cdata=%llu), "
-		"node=%llu, meta=%llu]",
+		"node=%llu, meta=%llu], "
+		"read_folio_count [0=%llu, 1=%llu, 2=%llu, 3=%llu, 4=%llu, "
+		"5=%llu, 6=%llu, 7=%llu, 8=%llu, 9=%llu, 10=%llu]",
 		show_dev(__entry->dev), __entry->app_wio, __entry->app_dio,
 		__entry->app_bio, __entry->app_mio, __entry->app_bcdio,
 		__entry->app_mcdio, __entry->fs_dio, __entry->fs_cdio,
@@ -2205,7 +2212,13 @@ TRACE_EVENT(f2fs_iostat,
 		__entry->app_rio, __entry->app_drio, __entry->app_brio,
 		__entry->app_mrio, __entry->app_bcrio, __entry->app_mcrio,
 		__entry->fs_drio, __entry->fs_gdrio,
-		__entry->fs_cdrio, __entry->fs_nrio, __entry->fs_mrio)
+		__entry->fs_cdrio, __entry->fs_nrio, __entry->fs_mrio,
+		__entry->read_folio_count[0], __entry->read_folio_count[1],
+		__entry->read_folio_count[2], __entry->read_folio_count[3],
+		__entry->read_folio_count[4], __entry->read_folio_count[5],
+		__entry->read_folio_count[6], __entry->read_folio_count[7],
+		__entry->read_folio_count[8], __entry->read_folio_count[9],
+		__entry->read_folio_count[10])
 );
 
 #ifndef __F2FS_IOSTAT_LATENCY_TYPE