Patchwork [6/6] ext4: update extent status tree after an extent is zeroed out

login
register
mail settings
Submitter Dmitri Monakho
Date Feb. 28, 2013, 6:45 p.m.
Message ID <87621cxoom.fsf@openvz.org>
Download mbox | patch
Permalink /patch/224134/
State New
Headers show

Comments

Dmitri Monakho - Feb. 28, 2013, 6:45 p.m.
On Fri,  1 Mar 2013 02:26:44 +0800, Zheng Liu <gnehzuil.liu@gmail.com> wrote:
> From: Zheng Liu <wenqing.lz@taobao.com>
> 
> When we try to split an extent, this extent could be zeroed out and mark
> as initialized.  But we don't know this in ext4_map_blocks because it
> only returns a allocated length of extent.  Meanwhile we will mark this
> extent as uninitialized because we only check m_flags.
> 
> This commit update extent status tree when we try to split an unwritten
> extent.  We don't need to worry about the status of this extent because
> we always mark it as initialized.
Your solution seems too complex, why not just update es_cache inside 
ext4_ext_zeroout() like this:

From e0565336c9ea540b0297781196453b358c3c3313 Mon Sep 17 00:00:00 2001
From: Dmitry Monakhov <dmonakhov@openvz.org>
Date: Thu, 28 Feb 2013 22:40:14 +0400
Subject: [PATCH] ext4: update status tree on extent zeroout


Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
---
 fs/ext4/extents.c           |    4 ++++
 include/trace/events/ext4.h |   27 +++++++++++++++++++++++++++
 2 files changed, 31 insertions(+), 0 deletions(-)
Zheng Liu - March 1, 2013, 2:37 a.m.
On Thu, Feb 28, 2013 at 10:45:29PM +0400, Dmitry Monakhov wrote:
> On Fri,  1 Mar 2013 02:26:44 +0800, Zheng Liu <gnehzuil.liu@gmail.com> wrote:
> > From: Zheng Liu <wenqing.lz@taobao.com>
> > 
> > When we try to split an extent, this extent could be zeroed out and mark
> > as initialized.  But we don't know this in ext4_map_blocks because it
> > only returns a allocated length of extent.  Meanwhile we will mark this
> > extent as uninitialized because we only check m_flags.
> > 
> > This commit update extent status tree when we try to split an unwritten
> > extent.  We don't need to worry about the status of this extent because
> > we always mark it as initialized.
> Your solution seems too complex, why not just update es_cache inside 
> ext4_ext_zeroout() like this:

Honestly this is my first solution.  But I don't think we can update
es_cache in ext4_ext_zeroout().  The reason is that when
ext4_ext_zerouout is called ext4_ext_dirty hasn't been called.  That
means that extent tree on disk is not changed.  We can't update es_cache
before metadata is changed.  Otherwise the assertation in
ext4_es_insert_extent_check() will fail.  Am I miss something?

Thanks
                                                - Zheng

> 
> From e0565336c9ea540b0297781196453b358c3c3313 Mon Sep 17 00:00:00 2001
> From: Dmitry Monakhov <dmonakhov@openvz.org>
> Date: Thu, 28 Feb 2013 22:40:14 +0400
> Subject: [PATCH] ext4: update status tree on extent zeroout
> 
> 
> Signed-off-by: Dmitry Monakhov <dmonakhov@openvz.org>
> ---
>  fs/ext4/extents.c           |    4 ++++
>  include/trace/events/ext4.h |   27 +++++++++++++++++++++++++++
>  2 files changed, 31 insertions(+), 0 deletions(-)
> 
> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
> index c89e850..15cc324 100644
> --- a/fs/ext4/extents.c
> +++ b/fs/ext4/extents.c
> @@ -2885,12 +2885,16 @@ static int ext4_ext_zeroout(struct inode *inode, struct ext4_extent *ex)
>  	unsigned int ee_len;
>  	int ret;
>  
> +	trace_ext4_ext_zeroout(inode, ex);
>  	ee_len    = ext4_ext_get_actual_len(ex);
>  	ee_pblock = ext4_ext_pblock(ex);
>  
>  	ret = sb_issue_zeroout(inode->i_sb, ee_pblock, ee_len, GFP_NOFS);
>  	if (ret > 0)
>  		ret = 0;
> +	if (!ret)
> +		ret = ext4_es_insert_extent(inode, le32_to_cpu(ex->ee_block),
> +					    ee_len, ee_pblock, EXTENT_STATUS_WRITTEN);
>  
>  	return ret;
>  }
> diff --git a/include/trace/events/ext4.h b/include/trace/events/ext4.h
> index c0457c0..cd750a6 100644
> --- a/include/trace/events/ext4.h
> +++ b/include/trace/events/ext4.h
> @@ -2092,6 +2092,33 @@ TRACE_EVENT(ext4_ext_remove_space_done,
>  		  (unsigned short) __entry->eh_entries)
>  );
>  
> +TRACE_EVENT(ext4_ext_zeroout,
> +	TP_PROTO(struct inode *inode, struct ext4_extent *ex),
> +
> +	TP_ARGS(inode, ex),
> +
> +	TP_STRUCT__entry(
> +		__field(	dev_t,		dev	)
> +		__field(	ino_t,		ino	)
> +		__field(	ext4_lblk_t,	u_lblk	)
> +		__field(	unsigned,	u_len	)
> +		__field(	ext4_fsblk_t,	u_pblk	)
> +	),
> +
> +	TP_fast_assign(
> +		__entry->dev		= inode->i_sb->s_dev;
> +		__entry->ino		= inode->i_ino;
> +		__entry->u_lblk		= le32_to_cpu(ex->ee_block);
> +		__entry->u_len		= ext4_ext_get_actual_len(ex);
> +		__entry->u_pblk		= ext4_ext_pblock(ex);
> +	),
> +
> +	TP_printk("dev %d,%d ino %lu ee_lblk %u ee_len %u ee_pblk %llu",
> +		  MAJOR(__entry->dev), MINOR(__entry->dev),
> +		  (unsigned long) __entry->ino,
> +		  __entry->u_lblk, __entry->u_len, __entry->u_pblk)
> +);
> +
>  TRACE_EVENT(ext4_es_insert_extent,
>  	TP_PROTO(struct inode *inode, struct extent_status *es),
>  
> -- 
> 1.7.1
> 
> 
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index c89e850..15cc324 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -2885,12 +2885,16 @@  static int ext4_ext_zeroout(struct inode *inode, struct ext4_extent *ex)
 	unsigned int ee_len;
 	int ret;
 
+	trace_ext4_ext_zeroout(inode, ex);
 	ee_len    = ext4_ext_get_actual_len(ex);
 	ee_pblock = ext4_ext_pblock(ex);
 
 	ret = sb_issue_zeroout(inode->i_sb, ee_pblock, ee_len, GFP_NOFS);
 	if (ret > 0)
 		ret = 0;
+	if (!ret)
+		ret = ext4_es_insert_extent(inode, le32_to_cpu(ex->ee_block),
+					    ee_len, ee_pblock, EXTENT_STATUS_WRITTEN);
 
 	return ret;
 }
diff --git a/include/trace/events/ext4.h b/include/trace/events/ext4.h
index c0457c0..cd750a6 100644
--- a/include/trace/events/ext4.h
+++ b/include/trace/events/ext4.h
@@ -2092,6 +2092,33 @@  TRACE_EVENT(ext4_ext_remove_space_done,
 		  (unsigned short) __entry->eh_entries)
 );
 
+TRACE_EVENT(ext4_ext_zeroout,
+	TP_PROTO(struct inode *inode, struct ext4_extent *ex),
+
+	TP_ARGS(inode, ex),
+
+	TP_STRUCT__entry(
+		__field(	dev_t,		dev	)
+		__field(	ino_t,		ino	)
+		__field(	ext4_lblk_t,	u_lblk	)
+		__field(	unsigned,	u_len	)
+		__field(	ext4_fsblk_t,	u_pblk	)
+	),
+
+	TP_fast_assign(
+		__entry->dev		= inode->i_sb->s_dev;
+		__entry->ino		= inode->i_ino;
+		__entry->u_lblk		= le32_to_cpu(ex->ee_block);
+		__entry->u_len		= ext4_ext_get_actual_len(ex);
+		__entry->u_pblk		= ext4_ext_pblock(ex);
+	),
+
+	TP_printk("dev %d,%d ino %lu ee_lblk %u ee_len %u ee_pblk %llu",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  (unsigned long) __entry->ino,
+		  __entry->u_lblk, __entry->u_len, __entry->u_pblk)
+);
+
 TRACE_EVENT(ext4_es_insert_extent,
 	TP_PROTO(struct inode *inode, struct extent_status *es),