RE:(2) (2) [PATCH v6] f2fs: add async reset zone command support
From: Daejun Park
Date: Wed May 17 2023 - 22:09:13 EST
> On 05/11, Daejun Park wrote:
> > > Sender : Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
> > > Date : 2023-05-10 03:53 (GMT+9)
> > > Title : Re: [PATCH v6] f2fs: add async reset zone command support
> > > To : 박대준<daejun7.park@xxxxxxxxxxx>
> > > CC : chao@xxxxxxxxxx<chao@xxxxxxxxxx>, rostedt@xxxxxxxxxxx<rostedt@xxxxxxxxxxx>, mhiramat@xxxxxxxxxx<mhiramat@xxxxxxxxxx>, linux-f2fs-devel@xxxxxxxxxxxxxxxxxxxxx<linux-f2fs-devel@xxxxxxxxxxxxxxxxxxxxx>, linux-kernel@xxxxxxxxxxxxxxx<linux-kernel@xxxxxxxxxxxxxxx>, linux-trace-kernel@xxxxxxxxxxxxxxx<linux-trace-kernel@xxxxxxxxxxxxxxx>, 김석환<sukka.kim@xxxxxxxxxxx>, 송용길<yonggil.song@xxxxxxxxxxx>, 김범수<beomsu7.kim@xxxxxxxxxxx>
> > >
> > > On 05/08, Daejun Park wrote:
> > > > v5 -> v6
> > > > Added trace_f2fs_iostat support for zone reset command.
> > > >
> > > > v4 -> v5
> > > > Added f2fs iostat for zone reset command.
> > > >
> > > > v3 -> v4
> > > > Fixed build error caused by unused function.
> > > >
> > > > v2 -> v3
> > > > Modified arguments to be correct for ftrace parameter.
> > > > Changed __submit_zone_reset_cmd to void return.
> > > > Refactored the f2fs_wait_discard_bio function.
> > > > Fixed code that was previously incorrectly merged.
> > > >
> > > > v1 -> v2
> > > > Changed to apply the optional async reset write pointer by default.
> > >
> > > Don't add the history in the patch description.
> >
> > OK, I will do.
> >
> > > >
> > > > This patch enables submit reset zone command asynchornously. It helps
> > > > decrease average latency of write IOs in high utilization scenario by
> > > > faster checkpointing.
> > > >
> > > > Signed-off-by: Daejun Park <daejun7.park@xxxxxxxxxxx>
> > > > ---
> > > > fs/f2fs/f2fs.h | 1 +
> > > > fs/f2fs/iostat.c | 1 +
> > > > fs/f2fs/segment.c | 84 +++++++++++++++++++++++++++++++++++--
> > > > include/trace/events/f2fs.h | 24 +++++++++--
> > > > 4 files changed, 104 insertions(+), 6 deletions(-)
> > > >
> > > > diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h
> > > > index d211ee89c158..51b68a629814 100644
> > > > --- a/fs/f2fs/f2fs.h
> > > > +++ b/fs/f2fs/f2fs.h
> > > > @@ -1175,6 +1175,7 @@ enum iostat_type {
> > > > /* other */
> > > > FS_DISCARD_IO, /* discard */
> > > > FS_FLUSH_IO, /* flush */
> > > > + FS_ZONE_RESET_IO, /* zone reset */
> > > > NR_IO_TYPE,
> > > > };
> > > >
> > > > diff --git a/fs/f2fs/iostat.c b/fs/f2fs/iostat.c
> > > > index 3d5bfb1ad585..f8703038e1d8 100644
> > > > --- a/fs/f2fs/iostat.c
> > > > +++ b/fs/f2fs/iostat.c
> > > > @@ -80,6 +80,7 @@ int __maybe_unused iostat_info_seq_show(struct seq_file *seq, void *offset)
> > > > seq_puts(seq, "[OTHER]\n");
> > > > IOSTAT_INFO_SHOW("fs discard", FS_DISCARD_IO);
> > > > IOSTAT_INFO_SHOW("fs flush", FS_FLUSH_IO);
> > > > + IOSTAT_INFO_SHOW("fs zone reset", FS_ZONE_RESET_IO);
> > > >
> > > > return 0;
> > > > }
> > > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> > > > index 6db410f1bb8c..4802b05a795b 100644
> > > > --- a/fs/f2fs/segment.c
> > > > +++ b/fs/f2fs/segment.c
> > > > @@ -1196,6 +1196,45 @@ static void __init_discard_policy(struct f2fs_sb_info *sbi,
> > > > static void __update_discard_tree_range(struct f2fs_sb_info *sbi,
> > > > struct block_device *bdev, block_t lstart,
> > > > block_t start, block_t len);
> > > > +
> > > > +#ifdef CONFIG_BLK_DEV_ZONED
> > > > +static void __submit_zone_reset_cmd(struct f2fs_sb_info *sbi,
> > >
> > > Why can't we use __f2fs_issue_discard_zone()?
> >
> > In my understanding, __f2fs_issue_discard_zone is used to queue the zone reset command.
> > Which point do you think is the issue?
>
> That actually calls blkdev_zone_mgmt() which is doing the same thing.
f2fs_clear_prefree_segments -> f2fs_issue_discard -> __issue_discard_async -> __f2fs_issue_discard_zone
In the case of discard in the existing code, discard can be queued by __queue_discard_cmd(). However the reset command is issued immediately.
In this patch, I changed the zone reset command can be queued like discard command.
Therefore, I changed __f2fs_issue_discard_zone() to queue the reset command, and to issue the reset command through __submit_discard_cmd() -> __submit_zone_reset_cmd().
Also, in blkdev_zone_mgmt(), bi_end_io cannot be specified, so waiting in f2fs_wait_discard_bio() is not possible.
That's why I created a new __submit_zone_reset_cmd().
>
> >
> > >
> > > > + struct discard_cmd *dc, blk_opf_t flag,
> > > > + struct list_head *wait_list,
> > > > + unsigned int *issued)
> > > > +{
> > > > + struct discard_cmd_control *dcc = SM_I(sbi)->dcc_info;
> > > > + struct block_device *bdev = dc->bdev;
> > > > + struct bio *bio = bio_alloc(bdev, 0, REQ_OP_ZONE_RESET | flag, GFP_NOFS);
> > > > + unsigned long flags;
> > > > +
> > > > + trace_f2fs_issue_reset_zone(bdev, dc->di.start);
> > > > +
> > > > + spin_lock_irqsave(&dc->lock, flags);
> > > > + dc->state = D_SUBMIT;
> > > > + dc->bio_ref++;
> > > > + spin_unlock_irqrestore(&dc->lock, flags);
> > > > +
> > > > + if (issued)
> > > > + (*issued)++;
> > > > +
> > > > + atomic_inc(&dcc->queued_discard);
> > > > + dc->queued++;
> > > > + list_move_tail(&dc->list, wait_list);
> > > > +
> > > > + /* sanity check on discard range */
> > > > + __check_sit_bitmap(sbi, dc->di.lstart, dc->di.lstart + dc->di.len);
> > > > +
> > > > + bio->bi_iter.bi_sector = SECTOR_FROM_BLOCK(dc->di.start);
> > > > + bio->bi_private = dc;
> > > > + bio->bi_end_io = f2fs_submit_discard_endio;
> > > > + submit_bio(bio);
> > > > +
> > > > + atomic_inc(&dcc->issued_discard);
> > > > + f2fs_update_iostat(sbi, NULL, FS_ZONE_RESET_IO, dc->di.len * F2FS_BLKSIZE);
> > > > +}
> > > > +#endif
> > > > +
> > > > /* this function is copied from blkdev_issue_discard from block/blk-lib.c */
> > > > static int __submit_discard_cmd(struct f2fs_sb_info *sbi,
> > > > struct discard_policy *dpolicy,
> > > > @@ -1217,6 +1256,13 @@ static int __submit_discard_cmd(struct f2fs_sb_info *sbi,
> > > > if (is_sbi_flag_set(sbi, SBI_NEED_FSCK))
> > > > return 0;
> > > >
> > > > +#ifdef CONFIG_BLK_DEV_ZONED
> > > > + if (f2fs_sb_has_blkzoned(sbi) && bdev_is_zoned(bdev)) {
> > > > + __submit_zone_reset_cmd(sbi, dc, flag, wait_list, issued);
> > > > + return 0;
> > > > + }
> > > > +#endif
> > > > +
> > > > trace_f2fs_issue_discard(bdev, dc->di.start, dc->di.len);
> > > >
> > > > lstart = dc->di.lstart;
> > > > @@ -1461,6 +1507,19 @@ static void __update_discard_tree_range(struct f2fs_sb_info *sbi,
> > > > }
> > > > }
> > > >
> > > > +#ifdef CONFIG_BLK_DEV_ZONED
> > > > +static void __queue_zone_reset_cmd(struct f2fs_sb_info *sbi,
> > > > + struct block_device *bdev, block_t blkstart, block_t lblkstart,
> > > > + block_t blklen)
> > > > +{
> > > > + trace_f2fs_queue_reset_zone(bdev, blkstart);
> > > > +
> > > > + mutex_lock(&SM_I(sbi)->dcc_info->cmd_lock);
> > > > + __insert_discard_cmd(sbi, bdev, lblkstart, blkstart, blklen);
> > > > + mutex_unlock(&SM_I(sbi)->dcc_info->cmd_lock);
> > > > +}
> > > > +#endif
> > > > +
> > > > static void __queue_discard_cmd(struct f2fs_sb_info *sbi,
> > > > struct block_device *bdev, block_t blkstart, block_t blklen)
> > > > {
> > > > @@ -1724,6 +1783,19 @@ static void f2fs_wait_discard_bio(struct f2fs_sb_info *sbi, block_t blkaddr)
> > > >
> > > > mutex_lock(&dcc->cmd_lock);
> > > > dc = __lookup_discard_cmd(sbi, blkaddr);
> > > > +#ifdef CONFIG_BLK_DEV_ZONED
> > > > + if (dc && f2fs_sb_has_blkzoned(sbi) && bdev_is_zoned(dc->bdev)) {
> > > > + /* force submit zone reset */
> > > > + if (dc->state == D_PREP)
> > > > + __submit_zone_reset_cmd(sbi, dc, REQ_SYNC,
> > > > + &dcc->wait_list, NULL);
> > > > + dc->ref++;
> > > > + mutex_unlock(&dcc->cmd_lock);
> > > > + /* wait zone reset */
> > > > + __wait_one_discard_bio(sbi, dc);
> > > > + return;
> > > > + }
> > > > +#endif
> > > > if (dc) {
> > > > if (dc->state == D_PREP) {
> > > > __punch_discard_cmd(sbi, dc, blkaddr);
> > > > @@ -1876,9 +1948,15 @@ static int __f2fs_issue_discard_zone(struct f2fs_sb_info *sbi,
> > > > blkstart, blklen);
> > > > return -EIO;
> > > > }
> > > > - trace_f2fs_issue_reset_zone(bdev, blkstart);
> > > > - return blkdev_zone_mgmt(bdev, REQ_OP_ZONE_RESET,
> > > > - sector, nr_sects, GFP_NOFS);
> > > > +
> > > > + if (unlikely(is_sbi_flag_set(sbi, SBI_POR_DOING))) {
> > > > + trace_f2fs_issue_reset_zone(bdev, blkstart);
> > > > + return blkdev_zone_mgmt(bdev, REQ_OP_ZONE_RESET,
> > > > + sector, nr_sects, GFP_NOFS);
> > > > + }
> > > > +
> > > > + __queue_zone_reset_cmd(sbi, bdev, blkstart, lblkstart, blklen);
> > > > + return 0;
> > > > }
> > > >
> > > > /* For conventional zones, use regular discard if supported */
> > > > diff --git a/include/trace/events/f2fs.h b/include/trace/events/f2fs.h
> > > > index 99cbc5949e3c..793f82cc1515 100644
> > > > --- a/include/trace/events/f2fs.h
> > > > +++ b/include/trace/events/f2fs.h
> > > > @@ -1512,7 +1512,7 @@ DEFINE_EVENT(f2fs_discard, f2fs_remove_discard,
> > > > TP_ARGS(dev, blkstart, blklen)
> > > > );
> > > >
> > > > -TRACE_EVENT(f2fs_issue_reset_zone,
> > > > +DECLARE_EVENT_CLASS(f2fs_reset_zone,
> > > >
> > > > TP_PROTO(struct block_device *dev, block_t blkstart),
> > > >
> > > > @@ -1528,11 +1528,25 @@ TRACE_EVENT(f2fs_issue_reset_zone,
> > > > __entry->blkstart = blkstart;
> > > > ),
> > > >
> > > > - TP_printk("dev = (%d,%d), reset zone at block = 0x%llx",
> > > > + TP_printk("dev = (%d,%d), zone at block = 0x%llx",
> > > > show_dev(__entry->dev),
> > > > (unsigned long long)__entry->blkstart)
> > > > );
> > > >
> > > > +DEFINE_EVENT(f2fs_reset_zone, f2fs_queue_reset_zone,
> > > > +
> > > > + TP_PROTO(struct block_device *dev, block_t blkstart),
> > > > +
> > > > + TP_ARGS(dev, blkstart)
> > > > +);
> > > > +
> > > > +DEFINE_EVENT(f2fs_reset_zone, f2fs_issue_reset_zone,
> > > > +
> > > > + TP_PROTO(struct block_device *dev, block_t blkstart),
> > > > +
> > > > + TP_ARGS(dev, blkstart)
> > > > +);
> > > > +
> > > > TRACE_EVENT(f2fs_issue_flush,
> > > >
> > > > TP_PROTO(struct block_device *dev, unsigned int nobarrier,
> > > > @@ -1979,6 +1993,7 @@ TRACE_EVENT(f2fs_iostat,
> > > > __field(unsigned long long, fs_nrio)
> > > > __field(unsigned long long, fs_mrio)
> > > > __field(unsigned long long, fs_discard)
> > > > + __field(unsigned long long, fs_reset_zone)
> > > > ),
> > > >
> > > > TP_fast_assign(
> > > > @@ -2010,12 +2025,14 @@ TRACE_EVENT(f2fs_iostat,
> > > > __entry->fs_nrio = iostat[FS_NODE_READ_IO];
> > > > __entry->fs_mrio = iostat[FS_META_READ_IO];
> > > > __entry->fs_discard = iostat[FS_DISCARD_IO];
> > > > + __entry->fs_reset_zone = iostat[FS_ZONE_RESET_IO];
> > > > ),
> > > >
> > > > TP_printk("dev = (%d,%d), "
> > > > "app [write=%llu (direct=%llu, buffered=%llu), mapped=%llu, "
> > > > "compr(buffered=%llu, mapped=%llu)], "
> > > > - "fs [data=%llu, cdata=%llu, node=%llu, meta=%llu, discard=%llu], "
> > > > + "fs [data=%llu, cdata=%llu, node=%llu, meta=%llu, discard=%llu, "
> > > > + "reset_zone=%llu], "
> > > > "gc [data=%llu, node=%llu], "
> > > > "cp [data=%llu, node=%llu, meta=%llu], "
> > > > "app [read=%llu (direct=%llu, buffered=%llu), mapped=%llu], "
> > > > @@ -2026,6 +2043,7 @@ TRACE_EVENT(f2fs_iostat,
> > > > __entry->app_bio, __entry->app_mio, __entry->app_bcdio,
> > > > __entry->app_mcdio, __entry->fs_dio, __entry->fs_cdio,
> > > > __entry->fs_nio, __entry->fs_mio, __entry->fs_discard,
> > > > + __entry->fs_reset_zone,
> > > > __entry->fs_gc_dio, __entry->fs_gc_nio, __entry->fs_cp_dio,
> > > > __entry->fs_cp_nio, __entry->fs_cp_mio,
> > > > __entry->app_rio, __entry->app_drio, __entry->app_brio,
> > > > --
> > > > 2.25.1