summaryrefslogtreecommitdiff
path: root/fs/f2fs/iostat.c
blob: 3d5bfb1ad585d9be86a5fd527615d3896f939230 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
// SPDX-License-Identifier: GPL-2.0
/*
 * f2fs iostat support
 *
 * Copyright 2021 Google LLC
 * Author: Daeho Jeong <daehojeong@google.com>
 */

#include <linux/fs.h>
#include <linux/f2fs_fs.h>
#include <linux/seq_file.h>

#include "f2fs.h"
#include "iostat.h"
#include <trace/events/f2fs.h>

static struct kmem_cache *bio_iostat_ctx_cache;
static mempool_t *bio_iostat_ctx_pool;

static inline unsigned long long iostat_get_avg_bytes(struct f2fs_sb_info *sbi,
	enum iostat_type type)
{
	return sbi->iostat_count[type] ? div64_u64(sbi->iostat_bytes[type],
		sbi->iostat_count[type]) : 0;
}

#define IOSTAT_INFO_SHOW(name, type)					\
	seq_printf(seq, "%-23s %-16llu %-16llu %-16llu\n",		\
			name":", sbi->iostat_bytes[type],		\
			sbi->iostat_count[type],			\
			iostat_get_avg_bytes(sbi, type))

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);

	if (!sbi->iostat_enable)
		return 0;

	seq_printf(seq, "time:		%-16llu\n", ktime_get_real_seconds());
	seq_printf(seq, "\t\t\t%-16s %-16s %-16s\n",
				"io_bytes", "count", "avg_bytes");

	/* print app write IOs */
	seq_puts(seq, "[WRITE]\n");
	IOSTAT_INFO_SHOW("app buffered data", APP_BUFFERED_IO);
	IOSTAT_INFO_SHOW("app direct data", APP_DIRECT_IO);
	IOSTAT_INFO_SHOW("app mapped data", APP_MAPPED_IO);
	IOSTAT_INFO_SHOW("app buffered cdata", APP_BUFFERED_CDATA_IO);
	IOSTAT_INFO_SHOW("app mapped cdata", APP_MAPPED_CDATA_IO);

	/* print fs write IOs */
	IOSTAT_INFO_SHOW("fs data", FS_DATA_IO);
	IOSTAT_INFO_SHOW("fs cdata", FS_CDATA_IO);
	IOSTAT_INFO_SHOW("fs node", FS_NODE_IO);
	IOSTAT_INFO_SHOW("fs meta", FS_META_IO);
	IOSTAT_INFO_SHOW("fs gc data", FS_GC_DATA_IO);
	IOSTAT_INFO_SHOW("fs gc node", FS_GC_NODE_IO);
	IOSTAT_INFO_SHOW("fs cp data", FS_CP_DATA_IO);
	IOSTAT_INFO_SHOW("fs cp node", FS_CP_NODE_IO);
	IOSTAT_INFO_SHOW("fs cp meta", FS_CP_META_IO);

	/* print app read IOs */
	seq_puts(seq, "[READ]\n");
	IOSTAT_INFO_SHOW("app buffered data", APP_BUFFERED_READ_IO);
	IOSTAT_INFO_SHOW("app direct data", APP_DIRECT_READ_IO);
	IOSTAT_INFO_SHOW("app mapped data", APP_MAPPED_READ_IO);
	IOSTAT_INFO_SHOW("app buffered cdata", APP_BUFFERED_CDATA_READ_IO);
	IOSTAT_INFO_SHOW("app mapped cdata", APP_MAPPED_CDATA_READ_IO);

	/* print fs read IOs */
	IOSTAT_INFO_SHOW("fs data", FS_DATA_READ_IO);
	IOSTAT_INFO_SHOW("fs gc data", FS_GDATA_READ_IO);
	IOSTAT_INFO_SHOW("fs cdata", FS_CDATA_READ_IO);
	IOSTAT_INFO_SHOW("fs node", FS_NODE_READ_IO);
	IOSTAT_INFO_SHOW("fs meta", FS_META_READ_IO);

	/* print other IOs */
	seq_puts(seq, "[OTHER]\n");
	IOSTAT_INFO_SHOW("fs discard", FS_DISCARD_IO);
	IOSTAT_INFO_SHOW("fs flush", FS_FLUSH_IO);

	return 0;
}

static inline void __record_iostat_latency(struct f2fs_sb_info *sbi)
{
	int io, idx;
	struct f2fs_iostat_latency iostat_lat[MAX_IO_TYPE][NR_PAGE_TYPE];
	struct iostat_lat_info *io_lat = sbi->iostat_io_lat;
	unsigned long flags;

	spin_lock_irqsave(&sbi->iostat_lat_lock, flags);
	for (idx = 0; idx < MAX_IO_TYPE; idx++) {
		for (io = 0; io < NR_PAGE_TYPE; io++) {
			iostat_lat[idx][io].peak_lat =
			   jiffies_to_msecs(io_lat->peak_lat[idx][io]);
			iostat_lat[idx][io].cnt = io_lat->bio_cnt[idx][io];
			iostat_lat[idx][io].avg_lat = iostat_lat[idx][io].cnt ?
			   jiffies_to_msecs(io_lat->sum_lat[idx][io]) / iostat_lat[idx][io].cnt : 0;
			io_lat->sum_lat[idx][io] = 0;
			io_lat->peak_lat[idx][io] = 0;
			io_lat->bio_cnt[idx][io] = 0;
		}
	}
	spin_unlock_irqrestore(&sbi->iostat_lat_lock, flags);

	trace_f2fs_iostat_latency(sbi, iostat_lat);
}

static inline void f2fs_record_iostat(struct f2fs_sb_info *sbi)
{
	unsigned long long iostat_diff[NR_IO_TYPE];
	int i;
	unsigned long flags;

	if (time_is_after_jiffies(sbi->iostat_next_period))
		return;

	/* Need double check under the lock */
	spin_lock_irqsave(&sbi->iostat_lock, flags);
	if (time_is_after_jiffies(sbi->iostat_next_period)) {
		spin_unlock_irqrestore(&sbi->iostat_lock, flags);
		return;
	}
	sbi->iostat_next_period = jiffies +
				msecs_to_jiffies(sbi->iostat_period_ms);

	for (i = 0; i < NR_IO_TYPE; i++) {
		iostat_diff[i] = sbi->iostat_bytes[i] -
				sbi->prev_iostat_bytes[i];
		sbi->prev_iostat_bytes[i] = sbi->iostat_bytes[i];
	}
	spin_unlock_irqrestore(&sbi->iostat_lock, flags);

	trace_f2fs_iostat(sbi, iostat_diff);

	__record_iostat_latency(sbi);
}

void f2fs_reset_iostat(struct f2fs_sb_info *sbi)
{
	struct iostat_lat_info *io_lat = sbi->iostat_io_lat;
	int i;

	spin_lock_irq(&sbi->iostat_lock);
	for (i = 0; i < NR_IO_TYPE; i++) {
		sbi->iostat_count[i] = 0;
		sbi->iostat_bytes[i] = 0;
		sbi->prev_iostat_bytes[i] = 0;
	}
	spin_unlock_irq(&sbi->iostat_lock);

	spin_lock_irq(&sbi->iostat_lat_lock);
	memset(io_lat, 0, sizeof(struct iostat_lat_info));
	spin_unlock_irq(&sbi->iostat_lat_lock);
}

static inline void __f2fs_update_iostat(struct f2fs_sb_info *sbi,
			enum iostat_type type, unsigned long long io_bytes)
{
	sbi->iostat_bytes[type] += io_bytes;
	sbi->iostat_count[type]++;
}

void f2fs_update_iostat(struct f2fs_sb_info *sbi, struct inode *inode,
			enum iostat_type type, unsigned long long io_bytes)
{
	unsigned long flags;

	if (!sbi->iostat_enable)
		return;

	spin_lock_irqsave(&sbi->iostat_lock, flags);
	__f2fs_update_iostat(sbi, type, io_bytes);

	if (type == APP_BUFFERED_IO || type == APP_DIRECT_IO)
		__f2fs_update_iostat(sbi, APP_WRITE_IO, io_bytes);

	if (type == APP_BUFFERED_READ_IO || type == APP_DIRECT_READ_IO)
		__f2fs_update_iostat(sbi, APP_READ_IO, io_bytes);

#ifdef CONFIG_F2FS_FS_COMPRESSION
	if (inode && f2fs_compressed_file(inode)) {
		if (type == APP_BUFFERED_IO)
			__f2fs_update_iostat(sbi, APP_BUFFERED_CDATA_IO, io_bytes);

		if (type == APP_BUFFERED_READ_IO)
			__f2fs_update_iostat(sbi, APP_BUFFERED_CDATA_READ_IO, io_bytes);

		if (type == APP_MAPPED_READ_IO)
			__f2fs_update_iostat(sbi, APP_MAPPED_CDATA_READ_IO, io_bytes);

		if (type == APP_MAPPED_IO)
			__f2fs_update_iostat(sbi, APP_MAPPED_CDATA_IO, io_bytes);

		if (type == FS_DATA_READ_IO)
			__f2fs_update_iostat(sbi, FS_CDATA_READ_IO, io_bytes);

		if (type == FS_DATA_IO)
			__f2fs_update_iostat(sbi, FS_CDATA_IO, io_bytes);
	}
#endif

	spin_unlock_irqrestore(&sbi->iostat_lock, flags);

	f2fs_record_iostat(sbi);
}

static inline void __update_iostat_latency(struct bio_iostat_ctx *iostat_ctx,
				enum iostat_lat_type lat_type)
{
	unsigned long ts_diff;
	unsigned int page_type = iostat_ctx->type;
	struct f2fs_sb_info *sbi = iostat_ctx->sbi;
	struct iostat_lat_info *io_lat = sbi->iostat_io_lat;
	unsigned long flags;

	if (!sbi->iostat_enable)
		return;

	ts_diff = jiffies - iostat_ctx->submit_ts;
	if (page_type == META_FLUSH) {
		page_type = META;
	} else if (page_type >= NR_PAGE_TYPE) {
		f2fs_warn(sbi, "%s: %d over NR_PAGE_TYPE", __func__, page_type);
		return;
	}

	spin_lock_irqsave(&sbi->iostat_lat_lock, flags);
	io_lat->sum_lat[lat_type][page_type] += ts_diff;
	io_lat->bio_cnt[lat_type][page_type]++;
	if (ts_diff > io_lat->peak_lat[lat_type][page_type])
		io_lat->peak_lat[lat_type][page_type] = ts_diff;
	spin_unlock_irqrestore(&sbi->iostat_lat_lock, flags);
}

void iostat_update_and_unbind_ctx(struct bio *bio)
{
	struct bio_iostat_ctx *iostat_ctx = bio->bi_private;
	enum iostat_lat_type lat_type;

	if (op_is_write(bio_op(bio))) {
		lat_type = bio->bi_opf & REQ_SYNC ?
				WRITE_SYNC_IO : WRITE_ASYNC_IO;
		bio->bi_private = iostat_ctx->sbi;
	} else {
		lat_type = READ_IO;
		bio->bi_private = iostat_ctx->post_read_ctx;
	}

	__update_iostat_latency(iostat_ctx, lat_type);
	mempool_free(iostat_ctx, bio_iostat_ctx_pool);
}

void iostat_alloc_and_bind_ctx(struct f2fs_sb_info *sbi,
		struct bio *bio, struct bio_post_read_ctx *ctx)
{
	struct bio_iostat_ctx *iostat_ctx;
	/* Due to the mempool, this never fails. */
	iostat_ctx = mempool_alloc(bio_iostat_ctx_pool, GFP_NOFS);
	iostat_ctx->sbi = sbi;
	iostat_ctx->submit_ts = 0;
	iostat_ctx->type = 0;
	iostat_ctx->post_read_ctx = ctx;
	bio->bi_private = iostat_ctx;
}

int __init f2fs_init_iostat_processing(void)
{
	bio_iostat_ctx_cache =
		kmem_cache_create("f2fs_bio_iostat_ctx",
				  sizeof(struct bio_iostat_ctx), 0, 0, NULL);
	if (!bio_iostat_ctx_cache)
		goto fail;
	bio_iostat_ctx_pool =
		mempool_create_slab_pool(NUM_PREALLOC_IOSTAT_CTXS,
					 bio_iostat_ctx_cache);
	if (!bio_iostat_ctx_pool)
		goto fail_free_cache;
	return 0;

fail_free_cache:
	kmem_cache_destroy(bio_iostat_ctx_cache);
fail:
	return -ENOMEM;
}

void f2fs_destroy_iostat_processing(void)
{
	mempool_destroy(bio_iostat_ctx_pool);
	kmem_cache_destroy(bio_iostat_ctx_cache);
}

int f2fs_init_iostat(struct f2fs_sb_info *sbi)
{
	/* init iostat info */
	spin_lock_init(&sbi->iostat_lock);
	spin_lock_init(&sbi->iostat_lat_lock);
	sbi->iostat_enable = false;
	sbi->iostat_period_ms = DEFAULT_IOSTAT_PERIOD_MS;
	sbi->iostat_io_lat = f2fs_kzalloc(sbi, sizeof(struct iostat_lat_info),
					GFP_KERNEL);
	if (!sbi->iostat_io_lat)
		return -ENOMEM;

	return 0;
}

void f2fs_destroy_iostat(struct f2fs_sb_info *sbi)
{
	kfree(sbi->iostat_io_lat);
}