All of lore.kernel.org
 help / color / mirror / Atom feed
From: Chao Yu <chao@kernel.org>
To: Jaegeuk Kim <jaegeuk@kernel.org>,
	linux-kernel@vger.kernel.org,
	linux-f2fs-devel@lists.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH 1/5] f2fs: check number of blocks in a current section
Date: Mon, 26 Feb 2024 10:40:41 +0800	[thread overview]
Message-ID: <afa444c5-7ac1-457e-9559-6d7c6623c426@kernel.org> (raw)
In-Reply-To: <20240223205535.307307-1-jaegeuk@kernel.org>

On 2024/2/24 4:55, Jaegeuk Kim wrote:
> In cfd66bb715fd ("f2fs: fix deadloop in foreground GC"), we needed to check
> the number of blocks in a section instead of the segment.
> 
> In addtion, let's check the entire node sections when checking the avaiable
> node block space. It does not match one to one per temperature, but currently

I tested this patch w/ testcase in [1], it doesn't complain.

[1] https://bugzilla.kernel.org/show_bug.cgi?id=215914

> we don't have exact dirty page count per temperature. Hence, use a rough
> estimation.

Do we need more accurate per-temperature dirty node count for extreme
corner case?

e.g. node_blocks is counted from hot dirty nodes, and current hot_node
log has no enough free space for it.

Thanks,

> 
> Fixes: cfd66bb715fd ("f2fs: fix deadloop in foreground GC")
> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> ---
>   fs/f2fs/segment.h | 22 +++++++++++-----------
>   1 file changed, 11 insertions(+), 11 deletions(-)
> 
> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
> index 3edd3809b6b5..15bf5edd9b3c 100644
> --- a/fs/f2fs/segment.h
> +++ b/fs/f2fs/segment.h
> @@ -561,23 +561,23 @@ static inline bool has_curseg_enough_space(struct f2fs_sb_info *sbi,
>   			unsigned int node_blocks, unsigned int dent_blocks)
>   {
>   
> -	unsigned int segno, left_blocks;
> +	unsigned segno, left_blocks;
>   	int i;
>   
> -	/* check current node segment */
> +	/* check current node sections, which counts very roughly */
> +	left_blocks = 0;
>   	for (i = CURSEG_HOT_NODE; i <= CURSEG_COLD_NODE; i++) {
>   		segno = CURSEG_I(sbi, i)->segno;
> -		left_blocks = f2fs_usable_blks_in_seg(sbi, segno) -
> -				get_seg_entry(sbi, segno)->ckpt_valid_blocks;
> -
> -		if (node_blocks > left_blocks)
> -			return false;
> +		left_blocks += CAP_BLKS_PER_SEC(sbi) -
> +				get_ckpt_valid_blocks(sbi, segno, true);
>   	}
> +	if (node_blocks > left_blocks)
> +		return false;
>   
> -	/* check current data segment */
> +	/* check current data section for dentry blocks. */
>   	segno = CURSEG_I(sbi, CURSEG_HOT_DATA)->segno;
> -	left_blocks = f2fs_usable_blks_in_seg(sbi, segno) -
> -			get_seg_entry(sbi, segno)->ckpt_valid_blocks;
> +	left_blocks = CAP_BLKS_PER_SEC(sbi) -
> +			get_ckpt_valid_blocks(sbi, segno, true);
>   	if (dent_blocks > left_blocks)
>   		return false;
>   	return true;
> @@ -626,7 +626,7 @@ static inline bool has_not_enough_free_secs(struct f2fs_sb_info *sbi,
>   
>   	if (free_secs > upper_secs)
>   		return false;
> -	else if (free_secs <= lower_secs)
> +	if (free_secs <= lower_secs)
>   		return true;
>   	return !curseg_space;
>   }

WARNING: multiple messages have this Message-ID (diff)
From: Chao Yu <chao@kernel.org>
To: Jaegeuk Kim <jaegeuk@kernel.org>,
	linux-kernel@vger.kernel.org,
	linux-f2fs-devel@lists.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH 1/5] f2fs: check number of blocks in a current section
Date: Mon, 26 Feb 2024 10:40:41 +0800	[thread overview]
Message-ID: <afa444c5-7ac1-457e-9559-6d7c6623c426@kernel.org> (raw)
In-Reply-To: <20240223205535.307307-1-jaegeuk@kernel.org>

On 2024/2/24 4:55, Jaegeuk Kim wrote:
> In cfd66bb715fd ("f2fs: fix deadloop in foreground GC"), we needed to check
> the number of blocks in a section instead of the segment.
> 
> In addtion, let's check the entire node sections when checking the avaiable
> node block space. It does not match one to one per temperature, but currently

I tested this patch w/ testcase in [1], it doesn't complain.

[1] https://bugzilla.kernel.org/show_bug.cgi?id=215914

> we don't have exact dirty page count per temperature. Hence, use a rough
> estimation.

Do we need more accurate per-temperature dirty node count for extreme
corner case?

e.g. node_blocks is counted from hot dirty nodes, and current hot_node
log has no enough free space for it.

Thanks,

> 
> Fixes: cfd66bb715fd ("f2fs: fix deadloop in foreground GC")
> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
> ---
>   fs/f2fs/segment.h | 22 +++++++++++-----------
>   1 file changed, 11 insertions(+), 11 deletions(-)
> 
> diff --git a/fs/f2fs/segment.h b/fs/f2fs/segment.h
> index 3edd3809b6b5..15bf5edd9b3c 100644
> --- a/fs/f2fs/segment.h
> +++ b/fs/f2fs/segment.h
> @@ -561,23 +561,23 @@ static inline bool has_curseg_enough_space(struct f2fs_sb_info *sbi,
>   			unsigned int node_blocks, unsigned int dent_blocks)
>   {
>   
> -	unsigned int segno, left_blocks;
> +	unsigned segno, left_blocks;
>   	int i;
>   
> -	/* check current node segment */
> +	/* check current node sections, which counts very roughly */
> +	left_blocks = 0;
>   	for (i = CURSEG_HOT_NODE; i <= CURSEG_COLD_NODE; i++) {
>   		segno = CURSEG_I(sbi, i)->segno;
> -		left_blocks = f2fs_usable_blks_in_seg(sbi, segno) -
> -				get_seg_entry(sbi, segno)->ckpt_valid_blocks;
> -
> -		if (node_blocks > left_blocks)
> -			return false;
> +		left_blocks += CAP_BLKS_PER_SEC(sbi) -
> +				get_ckpt_valid_blocks(sbi, segno, true);
>   	}
> +	if (node_blocks > left_blocks)
> +		return false;
>   
> -	/* check current data segment */
> +	/* check current data section for dentry blocks. */
>   	segno = CURSEG_I(sbi, CURSEG_HOT_DATA)->segno;
> -	left_blocks = f2fs_usable_blks_in_seg(sbi, segno) -
> -			get_seg_entry(sbi, segno)->ckpt_valid_blocks;
> +	left_blocks = CAP_BLKS_PER_SEC(sbi) -
> +			get_ckpt_valid_blocks(sbi, segno, true);
>   	if (dent_blocks > left_blocks)
>   		return false;
>   	return true;
> @@ -626,7 +626,7 @@ static inline bool has_not_enough_free_secs(struct f2fs_sb_info *sbi,
>   
>   	if (free_secs > upper_secs)
>   		return false;
> -	else if (free_secs <= lower_secs)
> +	if (free_secs <= lower_secs)
>   		return true;
>   	return !curseg_space;
>   }


_______________________________________________
Linux-f2fs-devel mailing list
Linux-f2fs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

  parent reply	other threads:[~2024-02-26  2:40 UTC|newest]

Thread overview: 52+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-02-23 20:55 [PATCH 1/5] f2fs: check number of blocks in a current section Jaegeuk Kim
2024-02-23 20:55 ` [f2fs-dev] " Jaegeuk Kim
2024-02-23 20:55 ` [PATCH 2/5] f2fs: fix write pointers all the time Jaegeuk Kim
2024-02-23 20:55   ` [f2fs-dev] " Jaegeuk Kim
2024-02-26  2:48   ` Chao Yu
2024-02-26  2:48     ` Chao Yu
2024-02-26 17:33     ` Daeho Jeong
2024-02-26 17:33       ` Daeho Jeong
2024-02-27  1:00     ` Jaegeuk Kim
2024-02-27  1:00       ` Jaegeuk Kim
2024-02-27  0:59   ` [PATCH 2/5 v2] " Jaegeuk Kim
2024-02-27  0:59     ` [f2fs-dev] " Jaegeuk Kim
2024-02-27  6:19     ` Chao Yu
2024-02-27  6:19       ` Chao Yu
2024-02-23 20:55 ` [PATCH 3/5] f2fs: print zone status in string and some log Jaegeuk Kim
2024-02-23 20:55   ` [f2fs-dev] " Jaegeuk Kim
2024-02-26  2:54   ` Chao Yu
2024-02-26  2:54     ` Chao Yu
2024-02-26 19:03   ` Daeho Jeong
2024-02-26 19:03     ` Daeho Jeong
2024-02-26 22:52   ` [PATCH 3/5 v2] " Jaegeuk Kim
2024-02-26 22:52     ` [f2fs-dev] " Jaegeuk Kim
2024-02-27  6:26     ` Chao Yu
2024-02-27  6:26       ` Chao Yu
2024-03-04 17:47   ` [PATCH 3/5] " Jaegeuk Kim
2024-03-04 17:47     ` [f2fs-dev] " Jaegeuk Kim
2024-02-23 20:55 ` [PATCH 4/5] f2fs: prevent an f2fs_gc loop during disable_checkpoint Jaegeuk Kim
2024-02-23 20:55   ` [f2fs-dev] " Jaegeuk Kim
2024-02-26  2:58   ` Chao Yu
2024-02-26  2:58     ` Chao Yu
2024-02-26 19:14     ` Daeho Jeong
2024-02-26 19:14       ` Daeho Jeong
2024-02-23 20:55 ` [PATCH 5/5] f2fs: allow to mount if cap is 100 Jaegeuk Kim
2024-02-23 20:55   ` [f2fs-dev] " Jaegeuk Kim
2024-02-26  2:59   ` Chao Yu
2024-02-26  2:59     ` Chao Yu
2024-02-26 19:34   ` Daeho Jeong
2024-02-26 19:34     ` Daeho Jeong
2024-02-26 22:47     ` Jaegeuk Kim
2024-02-26 22:47       ` Jaegeuk Kim
2024-02-28  3:47       ` Daeho Jeong
2024-02-28  3:47         ` Daeho Jeong
2024-02-26  2:40 ` Chao Yu [this message]
2024-02-26  2:40   ` [f2fs-dev] [PATCH 1/5] f2fs: check number of blocks in a current section Chao Yu
2024-02-26 17:17   ` Daeho Jeong
2024-02-26 17:17     ` Daeho Jeong
2024-02-26 23:14 ` [PATCH 1/5 v2] " Jaegeuk Kim
2024-02-26 23:14   ` [f2fs-dev] " Jaegeuk Kim
2024-02-27  6:08   ` Chao Yu
2024-02-27  6:08     ` Chao Yu
2024-02-28 22:50 ` [f2fs-dev] [PATCH 1/5] " patchwork-bot+f2fs
2024-02-28 22:50   ` patchwork-bot+f2fs

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=afa444c5-7ac1-457e-9559-6d7c6623c426@kernel.org \
    --to=chao@kernel.org \
    --cc=jaegeuk@kernel.org \
    --cc=linux-f2fs-devel@lists.sourceforge.net \
    --cc=linux-kernel@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.