[PATCH] f2fs: add iostat latency statistics support for discard

From: Yangtao Li
Date: Thu Feb 16 2023 - 06:29:47 EST

Next message: Peter Zijlstra: "Re: [PATCH 2/2] x86/entry: Fix unwinding from kprobe on PUSH/POP instruction"
Previous message: Youling Tang: "Re: [PATCH v4 2/5] LoongArch: Use la.pcrel instead of la.abs for exception handlers"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

In this patch, it adds to account discard latency.

Signed-off-by: Yangtao Li <frank.li@xxxxxxxx>
---
fs/f2fs/data.c | 2 +-
fs/f2fs/f2fs.h | 1 +
fs/f2fs/iostat.c | 83 +++++++++++++++++++++----------------
fs/f2fs/iostat.h | 49 +++++++++++-----------
fs/f2fs/segment.c | 8 +++-
include/trace/events/f2fs.h | 68 ++++++++++++++++--------------
6 files changed, 118 insertions(+), 93 deletions(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 28e09682b056..b67c0b41ee13 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -2312,7 +2312,7 @@ int f2fs_read_multi_pages(struct compress_ctx *cc, struct bio **bio_ret,
if (bio_add_page(bio, page, blocksize, 0) < blocksize)
goto submit_and_realloc;

- ctx = get_post_read_ctx(bio);
+ ctx = get_bio_iostat_private(bio);
ctx->enabled_steps |= STEP_DECOMPRESS;
refcount_inc(&dic->refcnt);

diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
index 21596e0266ba..29e973f67c3c 100644
--- a/fs/f2fs/f2fs.h
+++ b/fs/f2fs/f2fs.h
@@ -1131,6 +1131,7 @@ enum page_type {
META_FLUSH,
IPU, /* the below types are used by tracepoints only. */
OPU,
+ DISCARD, /* used by iostat */
};

enum temp_type {
diff --git a/fs/f2fs/iostat.c b/fs/f2fs/iostat.c
index 3d5bfb1ad585..304235b4f543 100644
--- a/fs/f2fs/iostat.c
+++ b/fs/f2fs/iostat.c
@@ -86,23 +86,21 @@ int __maybe_unused iostat_info_seq_show(struct seq_file *seq, void *offset)

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 f2fs_iostat_latency iostat_lat[MAX_LAT_TYPE];
struct iostat_lat_info *io_lat = sbi->iostat_io_lat;
+ unsigned int lat_type;
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;
- }
+ for (lat_type = 0; lat_type < MAX_LAT_TYPE; lat_type++) {
+ iostat_lat[lat_type].peak_lat =
+ jiffies_to_msecs(io_lat->peak_lat[lat_type]);
+ iostat_lat[lat_type].cnt = io_lat->bio_cnt[lat_type];
+ iostat_lat[lat_type].avg_lat = iostat_lat[lat_type].cnt ?
+ jiffies_to_msecs(io_lat->sum_lat[lat_type]) / iostat_lat[lat_type].cnt : 0;
+ io_lat->sum_lat[lat_type] = 0;
+ io_lat->peak_lat[lat_type] = 0;
+ io_lat->bio_cnt[lat_type] = 0;
}
spin_unlock_irqrestore(&sbi->iostat_lat_lock, flags);

@@ -208,62 +206,75 @@ void f2fs_update_iostat(struct f2fs_sb_info *sbi, struct inode *inode,
f2fs_record_iostat(sbi);
}

-static inline void __update_iostat_latency(struct bio_iostat_ctx *iostat_ctx,
- enum iostat_lat_type lat_type)
+static inline void __update_iostat_latency(struct bio_iostat_ctx *iostat_ctx)
{
- 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 int lat_type = iostat_ctx->lat_type;
+ unsigned long ts_diff;
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);
+ if (lat_type >= MAX_LAT_TYPE) {
+ f2fs_warn(sbi, "%s: %d over MAX_LAT_TYPE", __func__, lat_type);
return;
}
+ ts_diff = jiffies - iostat_ctx->submit_ts;

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;
+ io_lat->sum_lat[lat_type] += ts_diff;
+ io_lat->bio_cnt[lat_type]++;
+ if (ts_diff > io_lat->peak_lat[lat_type])
+ io_lat->peak_lat[lat_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;
+
+ if (op_is_write(bio_op(bio)) && !op_is_discard(bio_op(bio)))
+ bio->bi_private = iostat_ctx->sbi;
+ else
+ bio->bi_private = iostat_ctx->iostat_private;
+
+ __update_iostat_latency(iostat_ctx);
+ mempool_free(iostat_ctx, bio_iostat_ctx_pool);
+}
+
+void iostat_update_submit_ctx(struct bio *bio, enum page_type type)
{
struct bio_iostat_ctx *iostat_ctx = bio->bi_private;
enum iostat_lat_type lat_type;

- if (op_is_write(bio_op(bio))) {
+ iostat_ctx->submit_ts = jiffies;
+
+ if (type == DISCARD) {
+ lat_type = DISCARD_LAT;
+ } else 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;
+ WRITE_SYNC_DATA_LAT : WRITE_ASYNC_DATA_LAT;
+ lat_type += type;
} else {
- lat_type = READ_IO;
- bio->bi_private = iostat_ctx->post_read_ctx;
+ lat_type = READ_DATA_LAT;
+ lat_type += type;
}

- __update_iostat_latency(iostat_ctx, lat_type);
- mempool_free(iostat_ctx, bio_iostat_ctx_pool);
+ iostat_ctx->lat_type = lat_type;
}

void iostat_alloc_and_bind_ctx(struct f2fs_sb_info *sbi,
- struct bio *bio, struct bio_post_read_ctx *ctx)
+ struct bio *bio, void *private)
{
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;
+ iostat_ctx->lat_type = 0;
+ iostat_ctx->iostat_private = private;
bio->bi_private = iostat_ctx;
}

diff --git a/fs/f2fs/iostat.h b/fs/f2fs/iostat.h
index eb99d05cf272..824d5a23bc19 100644
--- a/fs/f2fs/iostat.h
+++ b/fs/f2fs/iostat.h
@@ -6,17 +6,24 @@
#ifndef __F2FS_IOSTAT_H__
#define __F2FS_IOSTAT_H__

+#ifdef CONFIG_F2FS_IOSTAT
+
struct bio_post_read_ctx;

enum iostat_lat_type {
- READ_IO = 0,
- WRITE_SYNC_IO,
- WRITE_ASYNC_IO,
- MAX_IO_TYPE,
+ READ_DATA_LAT = 0,
+ READ_NODE_LAT,
+ READ_META_LAT,
+ WRITE_SYNC_DATA_LAT,
+ WRITE_SYNC_NODE_LAT,
+ WRITE_SYNC_META_LAT,
+ WRITE_ASYNC_DATA_LAT,
+ WRITE_ASYNC_NODE_LAT,
+ WRITE_ASYNC_META_LAT,
+ DISCARD_LAT,
+ MAX_LAT_TYPE,
};

-#ifdef CONFIG_F2FS_IOSTAT
-
#define NUM_PREALLOC_IOSTAT_CTXS 128
#define DEFAULT_IOSTAT_PERIOD_MS 3000
#define MIN_IOSTAT_PERIOD_MS 100
@@ -24,9 +31,9 @@ enum iostat_lat_type {
#define MAX_IOSTAT_PERIOD_MS 8640000

struct iostat_lat_info {
- unsigned long sum_lat[MAX_IO_TYPE][NR_PAGE_TYPE]; /* sum of io latencies */
- unsigned long peak_lat[MAX_IO_TYPE][NR_PAGE_TYPE]; /* peak io latency */
- unsigned int bio_cnt[MAX_IO_TYPE][NR_PAGE_TYPE]; /* bio count */
+ unsigned long sum_lat[MAX_LAT_TYPE]; /* sum of io latencies */
+ unsigned long peak_lat[MAX_LAT_TYPE]; /* peak io latency */
+ unsigned int bio_cnt[MAX_LAT_TYPE]; /* bio count */
};

extern int __maybe_unused iostat_info_seq_show(struct seq_file *seq,
@@ -38,29 +45,21 @@ extern void f2fs_update_iostat(struct f2fs_sb_info *sbi, struct inode *inode,
struct bio_iostat_ctx {
struct f2fs_sb_info *sbi;
unsigned long submit_ts;
- enum page_type type;
- struct bio_post_read_ctx *post_read_ctx;
+ enum iostat_lat_type lat_type;
+ void *iostat_private;
};

-static inline void iostat_update_submit_ctx(struct bio *bio,
- enum page_type type)
-{
- struct bio_iostat_ctx *iostat_ctx = bio->bi_private;
-
- iostat_ctx->submit_ts = jiffies;
- iostat_ctx->type = type;
-}
-
-static inline struct bio_post_read_ctx *get_post_read_ctx(struct bio *bio)
+static inline struct bio_post_read_ctx *get_bio_iostat_private(struct bio *bio)
{
struct bio_iostat_ctx *iostat_ctx = bio->bi_private;

- return iostat_ctx->post_read_ctx;
+ return iostat_ctx->iostat_private;
}

+extern void iostat_update_submit_ctx(struct bio *bio, enum page_type type);
extern void iostat_update_and_unbind_ctx(struct bio *bio);
extern void iostat_alloc_and_bind_ctx(struct f2fs_sb_info *sbi,
- struct bio *bio, struct bio_post_read_ctx *ctx);
+ struct bio *bio, void *private);
extern int f2fs_init_iostat_processing(void);
extern void f2fs_destroy_iostat_processing(void);
extern int f2fs_init_iostat(struct f2fs_sb_info *sbi);
@@ -70,10 +69,10 @@ static inline void f2fs_update_iostat(struct f2fs_sb_info *sbi, struct inode *in
enum iostat_type type, unsigned long long io_bytes) {}
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) {}
+ struct bio *bio, void *private) {}
static inline void iostat_update_submit_ctx(struct bio *bio,
enum page_type type) {}
-static inline struct bio_post_read_ctx *get_post_read_ctx(struct bio *bio)
+static inline struct bio_post_read_ctx *get_bio_iostat_private(struct bio *bio)
{
return bio->bi_private;
}
diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
index 227e25836173..9a6c45a62e8b 100644
--- a/fs/f2fs/segment.c
+++ b/fs/f2fs/segment.c
@@ -1012,9 +1012,12 @@ static void __remove_discard_cmd(struct f2fs_sb_info *sbi,

static void f2fs_submit_discard_endio(struct bio *bio)
{
- struct discard_cmd *dc = (struct discard_cmd *)bio->bi_private;
+ struct discard_cmd *dc;
unsigned long flags;

+ iostat_update_and_unbind_ctx(bio);
+ dc = bio->bi_private;
+
spin_lock_irqsave(&dc->lock, flags);
if (!dc->error)
dc->error = blk_status_to_errno(bio->bi_status);
@@ -1188,6 +1191,9 @@ static int __submit_discard_cmd(struct f2fs_sb_info *sbi,
bio->bi_private = dc;
bio->bi_end_io = f2fs_submit_discard_endio;
bio->bi_opf |= flag;
+
+ iostat_alloc_and_bind_ctx(sbi, bio, dc);
+ iostat_update_submit_ctx(bio, DISCARD);
submit_bio(bio);

atomic_inc(&dcc->issued_discard);
diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h
index 1322d34a5dfc..a0b115cf505c 100644
--- a/include/trace/events/f2fs.h
+++ b/include/trace/events/f2fs.h
@@ -2045,7 +2045,7 @@ struct f2fs_iostat_latency {

TRACE_EVENT(f2fs_iostat_latency,

- TP_PROTO(struct f2fs_sb_info *sbi, struct f2fs_iostat_latency (*iostat_lat)[NR_PAGE_TYPE]),
+ TP_PROTO(struct f2fs_sb_info *sbi, struct f2fs_iostat_latency *iostat_lat),

TP_ARGS(sbi, iostat_lat),

@@ -2078,37 +2078,43 @@ TRACE_EVENT(f2fs_iostat_latency,
__field(unsigned int, m_wr_as_peak)
__field(unsigned int, m_wr_as_avg)
__field(unsigned int, m_wr_as_cnt)
+ __field(unsigned int, discard_peak)
+ __field(unsigned int, discard_avg)
+ __field(unsigned int, discard_cnt)
),

TP_fast_assign(
__entry->dev = sbi->sb->s_dev;
- __entry->d_rd_peak = iostat_lat[READ_IO][DATA].peak_lat;
- __entry->d_rd_avg = iostat_lat[READ_IO][DATA].avg_lat;
- __entry->d_rd_cnt = iostat_lat[READ_IO][DATA].cnt;
- __entry->n_rd_peak = iostat_lat[READ_IO][NODE].peak_lat;
- __entry->n_rd_avg = iostat_lat[READ_IO][NODE].avg_lat;
- __entry->n_rd_cnt = iostat_lat[READ_IO][NODE].cnt;
- __entry->m_rd_peak = iostat_lat[READ_IO][META].peak_lat;
- __entry->m_rd_avg = iostat_lat[READ_IO][META].avg_lat;
- __entry->m_rd_cnt = iostat_lat[READ_IO][META].cnt;
- __entry->d_wr_s_peak = iostat_lat[WRITE_SYNC_IO][DATA].peak_lat;
- __entry->d_wr_s_avg = iostat_lat[WRITE_SYNC_IO][DATA].avg_lat;
- __entry->d_wr_s_cnt = iostat_lat[WRITE_SYNC_IO][DATA].cnt;
- __entry->n_wr_s_peak = iostat_lat[WRITE_SYNC_IO][NODE].peak_lat;
- __entry->n_wr_s_avg = iostat_lat[WRITE_SYNC_IO][NODE].avg_lat;
- __entry->n_wr_s_cnt = iostat_lat[WRITE_SYNC_IO][NODE].cnt;
- __entry->m_wr_s_peak = iostat_lat[WRITE_SYNC_IO][META].peak_lat;
- __entry->m_wr_s_avg = iostat_lat[WRITE_SYNC_IO][META].avg_lat;
- __entry->m_wr_s_cnt = iostat_lat[WRITE_SYNC_IO][META].cnt;
- __entry->d_wr_as_peak = iostat_lat[WRITE_ASYNC_IO][DATA].peak_lat;
- __entry->d_wr_as_avg = iostat_lat[WRITE_ASYNC_IO][DATA].avg_lat;
- __entry->d_wr_as_cnt = iostat_lat[WRITE_ASYNC_IO][DATA].cnt;
- __entry->n_wr_as_peak = iostat_lat[WRITE_ASYNC_IO][NODE].peak_lat;
- __entry->n_wr_as_avg = iostat_lat[WRITE_ASYNC_IO][NODE].avg_lat;
- __entry->n_wr_as_cnt = iostat_lat[WRITE_ASYNC_IO][NODE].cnt;
- __entry->m_wr_as_peak = iostat_lat[WRITE_ASYNC_IO][META].peak_lat;
- __entry->m_wr_as_avg = iostat_lat[WRITE_ASYNC_IO][META].avg_lat;
- __entry->m_wr_as_cnt = iostat_lat[WRITE_ASYNC_IO][META].cnt;
+ __entry->d_rd_peak = iostat_lat[READ_DATA_LAT].peak_lat;
+ __entry->d_rd_avg = iostat_lat[READ_DATA_LAT].avg_lat;
+ __entry->d_rd_cnt = iostat_lat[READ_DATA_LAT].cnt;
+ __entry->n_rd_peak = iostat_lat[READ_NODE_LAT].peak_lat;
+ __entry->n_rd_avg = iostat_lat[READ_NODE_LAT].avg_lat;
+ __entry->n_rd_cnt = iostat_lat[READ_NODE_LAT].cnt;
+ __entry->m_rd_peak = iostat_lat[READ_META_LAT].peak_lat;
+ __entry->m_rd_avg = iostat_lat[READ_META_LAT].avg_lat;
+ __entry->m_rd_cnt = iostat_lat[READ_META_LAT].cnt;
+ __entry->d_wr_s_peak = iostat_lat[WRITE_SYNC_DATA_LAT].peak_lat;
+ __entry->d_wr_s_avg = iostat_lat[WRITE_SYNC_DATA_LAT].avg_lat;
+ __entry->d_wr_s_cnt = iostat_lat[WRITE_SYNC_DATA_LAT].cnt;
+ __entry->n_wr_s_peak = iostat_lat[WRITE_SYNC_NODE_LAT].peak_lat;
+ __entry->n_wr_s_avg = iostat_lat[WRITE_SYNC_NODE_LAT].avg_lat;
+ __entry->n_wr_s_cnt = iostat_lat[WRITE_SYNC_NODE_LAT].cnt;
+ __entry->m_wr_s_peak = iostat_lat[WRITE_SYNC_META_LAT].peak_lat;
+ __entry->m_wr_s_avg = iostat_lat[WRITE_SYNC_META_LAT].avg_lat;
+ __entry->m_wr_s_cnt = iostat_lat[WRITE_SYNC_META_LAT].cnt;
+ __entry->d_wr_as_peak = iostat_lat[WRITE_ASYNC_DATA_LAT].peak_lat;
+ __entry->d_wr_as_avg = iostat_lat[WRITE_ASYNC_DATA_LAT].avg_lat;
+ __entry->d_wr_as_cnt = iostat_lat[WRITE_ASYNC_DATA_LAT].cnt;
+ __entry->n_wr_as_peak = iostat_lat[WRITE_ASYNC_NODE_LAT].peak_lat;
+ __entry->n_wr_as_avg = iostat_lat[WRITE_ASYNC_NODE_LAT].avg_lat;
+ __entry->n_wr_as_cnt = iostat_lat[WRITE_ASYNC_NODE_LAT].cnt;
+ __entry->m_wr_as_peak = iostat_lat[WRITE_ASYNC_META_LAT].peak_lat;
+ __entry->m_wr_as_avg = iostat_lat[WRITE_ASYNC_META_LAT].avg_lat;
+ __entry->m_wr_as_cnt = iostat_lat[WRITE_ASYNC_META_LAT].cnt;
+ __entry->discard_peak = iostat_lat[DISCARD_LAT].peak_lat;
+ __entry->discard_avg = iostat_lat[DISCARD_LAT].avg_lat;
+ __entry->discard_cnt = iostat_lat[DISCARD_LAT].cnt;
),

TP_printk("dev = (%d,%d), "
@@ -2116,7 +2122,8 @@ TRACE_EVENT(f2fs_iostat_latency,
"rd_data [%u/%u/%u], rd_node [%u/%u/%u], rd_meta [%u/%u/%u], "
"wr_sync_data [%u/%u/%u], wr_sync_node [%u/%u/%u], "
"wr_sync_meta [%u/%u/%u], wr_async_data [%u/%u/%u], "
- "wr_async_node [%u/%u/%u], wr_async_meta [%u/%u/%u]",
+ "wr_async_node [%u/%u/%u], wr_async_meta [%u/%u/%u], "
+ "discard [%u/%u/%u]",
show_dev(__entry->dev),
__entry->d_rd_peak, __entry->d_rd_avg, __entry->d_rd_cnt,
__entry->n_rd_peak, __entry->n_rd_avg, __entry->n_rd_cnt,
@@ -2126,7 +2133,8 @@ TRACE_EVENT(f2fs_iostat_latency,
__entry->m_wr_s_peak, __entry->m_wr_s_avg, __entry->m_wr_s_cnt,
__entry->d_wr_as_peak, __entry->d_wr_as_avg, __entry->d_wr_as_cnt,
__entry->n_wr_as_peak, __entry->n_wr_as_avg, __entry->n_wr_as_cnt,
- __entry->m_wr_as_peak, __entry->m_wr_as_avg, __entry->m_wr_as_cnt)
+ __entry->m_wr_as_peak, __entry->m_wr_as_avg, __entry->m_wr_as_cnt,
+ __entry->discard_peak, __entry->discard_avg, __entry->discard_cnt)
);
#endif

--
2.25.1

Next message: Peter Zijlstra: "Re: [PATCH 2/2] x86/entry: Fix unwinding from kprobe on PUSH/POP instruction"
Previous message: Youling Tang: "Re: [PATCH v4 2/5] LoongArch: Use la.pcrel instead of la.abs for exception handlers"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]