[PATCH] writeback: clear auto_free in initializaiton

[PATCH] writeback: clear auto_free in initializaiton

[trix]

From: Tom Rix <trix@redhat.com>

Review fs/fs-writeback.c bdi_split_work_to_wbs
The CONFIG_CGROUP_WRITEBACK version contains this line
	base_work->auto_free = 0;
Which seems like a strange place to set auto_free as
it is not where the rest of base_work is initialized.

In the default version of bdi_split_work_to_wbs, if a
successful malloc happens, base_work is copied and
auto_free is set to 1, else the base_work is
copied to another local valarible and its auto_free
is set to 0.

So move the clearing of auto_free to the
initialization of the local base_work structures.

Signed-off-by: Tom Rix <trix@redhat.com>
---
 fs/fs-writeback.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
index a605c3dddabc..fa1106de2ab0 100644
--- a/fs/fs-writeback.c
+++ b/fs/fs-writeback.c
@@ -881,7 +881,6 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
 		work = &fallback_work;
 		*work = *base_work;
 		work->nr_pages = nr_pages;
-		work->auto_free = 0;
 		work->done = &fallback_work_done;
 
 		wb_queue_work(wb, work);
@@ -1055,10 +1054,8 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
 {
 	might_sleep();
 
-	if (!skip_if_busy || !writeback_in_progress(&bdi->wb)) {
-		base_work->auto_free = 0;
+	if (!skip_if_busy || !writeback_in_progress(&bdi->wb))
 		wb_queue_work(&bdi->wb, base_work);
-	}
 }
 
 #endif	/* CONFIG_CGROUP_WRITEBACK */
@@ -2459,6 +2456,7 @@ static void __writeback_inodes_sb_nr(struct super_block *sb, unsigned long nr,
 		.done			= &done,
 		.nr_pages		= nr,
 		.reason			= reason,
+		.auto_free		= 0,
 	};
 
 	if (!bdi_has_dirty_io(bdi) || bdi == &noop_backing_dev_info)
@@ -2538,6 +2536,7 @@ void sync_inodes_sb(struct super_block *sb)
 		.done		= &done,
 		.reason		= WB_REASON_SYNC,
 		.for_sync	= 1,
+		.auto_free	= 0,
 	};
 
 	/*
-- 
2.18.1

Re: [PATCH] writeback: clear auto_free in initializaiton

[Jan Kara]

On Tue 18-08-20 07:13:30, trix@redhat.com wrote:
> From: Tom Rix <trix@redhat.com>
> 
> Review fs/fs-writeback.c bdi_split_work_to_wbs
> The CONFIG_CGROUP_WRITEBACK version contains this line
> 	base_work->auto_free = 0;

It is actually the !CONFIG_CGROUP_WRITEBACK version...

> Which seems like a strange place to set auto_free as
> it is not where the rest of base_work is initialized.

Otherwise I agree it's a strange place. I've added Tejun to CC just in case
he remembers why he's added that.

> In the default version of bdi_split_work_to_wbs, if a
> successful malloc happens, base_work is copied and
> auto_free is set to 1, else the base_work is
> copied to another local valarible and its auto_free
> is set to 0.
> 
> So move the clearing of auto_free to the
> initialization of the local base_work structures.
> 
> Signed-off-by: Tom Rix <trix@redhat.com>

Some more comments below.

> ---
>  fs/fs-writeback.c | 7 +++----
>  1 file changed, 3 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/fs-writeback.c b/fs/fs-writeback.c
> index a605c3dddabc..fa1106de2ab0 100644
> --- a/fs/fs-writeback.c
> +++ b/fs/fs-writeback.c
> @@ -881,7 +881,6 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
>  		work = &fallback_work;
>  		*work = *base_work;
>  		work->nr_pages = nr_pages;
> -		work->auto_free = 0;
>  		work->done = &fallback_work_done;

Honestly, I'd leave this alone. Although base_work should have auto_free ==
0, this assignment IMO helps readability.

> @@ -1055,10 +1054,8 @@ static void bdi_split_work_to_wbs(struct backing_dev_info *bdi,
>  {
>  	might_sleep();
>  
> -	if (!skip_if_busy || !writeback_in_progress(&bdi->wb)) {
> -		base_work->auto_free = 0;
> +	if (!skip_if_busy || !writeback_in_progress(&bdi->wb))
>  		wb_queue_work(&bdi->wb, base_work);
> -	}

Agreed with this.


> @@ -2459,6 +2456,7 @@ static void __writeback_inodes_sb_nr(struct super_block *sb, unsigned long nr,
>  		.done			= &done,
>  		.nr_pages		= nr,
>  		.reason			= reason,
> +		.auto_free		= 0,
>  	};
>  
>  	if (!bdi_has_dirty_io(bdi) || bdi == &noop_backing_dev_info)
> @@ -2538,6 +2536,7 @@ void sync_inodes_sb(struct super_block *sb)
>  		.done		= &done,
>  		.reason		= WB_REASON_SYNC,
>  		.for_sync	= 1,
> +		.auto_free	= 0,
>  	};

No need for explicit initialization to 0 - that is implicit with the
initializers.

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH] writeback: clear auto_free in initializaiton

[Tejun Heo]

Hello,

On Wed, Aug 26, 2020 at 06:11:26PM +0200, Jan Kara wrote:
> > Which seems like a strange place to set auto_free as
> > it is not where the rest of base_work is initialized.
> 
> Otherwise I agree it's a strange place. I've added Tejun to CC just in case
> he remembers why he's added that.

I don't remember exactly but maybe I was trying to mirror the
CGROUP_WRITEBACK counterpart without actually thinking about it? The patch
looks good to me.

Thanks.

-- 
tejun