Re: [f2fs-dev] [PATCH] f2fs: Avoid double lock for cp_rwsem

Re: [f2fs-dev] [PATCH] f2fs: Avoid double lock for cp_rwsem

[Markus Elfring]

> Call stack :
> f2fs_write_checkpoint()
> -> block_operations(sbi)
…

Will another imperative wording become helpful besides the provided information
for this change description?
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/process/submitting-patches.rst?id=c578ddb39e565139897124e74e5a43e56538cb33#n151

How do you think about to add the tag “Fixes” because of adjustments
for the data synchronisation?

Regards,
Markus

Re: [f2fs-dev] [PATCH] f2fs: Avoid double lock for cp_rwsem

[Jaegeuk Kim]

On 04/23, Jaegeuk Kim wrote:
> On 04/23, Sayali Lokhande wrote:
> > Call stack :
> > f2fs_write_checkpoint()
> > -> block_operations(sbi)
> >     f2fs_lock_all(sbi);
> >      down_write(&sbi->cp_rwsem); => write lock held
> > <>
> > -> f2fs_sync_node_pages()
> >     if (is_inline_node(page))
> >      flush_inline_data()
> > 	page = f2fs_pagecache_get_page()
> >          if (!page)
> >            goto iput_out;
> > 	iput_out:
> > 	 iput(inode);
> >           -> f2fs_evict_inode()
> > 	      f2fs_truncate_blocks()
> > 	       f2fs_lock_op()
> > 	        down_read(&sbi->cp_rwsem); => read lock fail
> 
> How about this, since we don't actually need to flush inline_data?

Hmm, nvm. This causes no space panic regressed during xfstests.

> 
> diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
> index 4da0d8713df5c..9af8d5319fdd3 100644
> --- a/fs/f2fs/node.c
> +++ b/fs/f2fs/node.c
> @@ -1871,7 +1871,7 @@ int f2fs_sync_node_pages(struct f2fs_sb_info *sbi,
>                         }
> 
>                         /* flush inline_data */
> -                       if (is_inline_node(page)) {
> +                       if (is_inline_node(page) && io_type != FS_CP_NODE_IO) {
>                                 clear_inline_node(page);
>                                 unlock_page(page);
>                                 flush_inline_data(sbi, ino_of_node(page));
> 
> > 
> > Signed-off-by: Sayali Lokhande <sayalil@codeaurora.org>
> > ---
> >  fs/f2fs/checkpoint.c | 10 ++++------
> >  1 file changed, 4 insertions(+), 6 deletions(-)
> > 
> > diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
> > index 5ba649e..5c504cf 100644
> > --- a/fs/f2fs/checkpoint.c
> > +++ b/fs/f2fs/checkpoint.c
> > @@ -1219,21 +1219,19 @@ static int block_operations(struct f2fs_sb_info *sbi)
> >  		goto retry_flush_quotas;
> >  	}
> >  
> > -retry_flush_nodes:
> >  	down_write(&sbi->node_write);
> >  
> >  	if (get_pages(sbi, F2FS_DIRTY_NODES)) {
> >  		up_write(&sbi->node_write);
> > +		f2fs_unlock_all(sbi);
> >  		atomic_inc(&sbi->wb_sync_req[NODE]);
> >  		err = f2fs_sync_node_pages(sbi, &wbc, false, FS_CP_NODE_IO);
> >  		atomic_dec(&sbi->wb_sync_req[NODE]);
> > -		if (err) {
> > -			up_write(&sbi->node_change);
> > -			f2fs_unlock_all(sbi);
> > +		up_write(&sbi->node_change);
> > +		if (err)
> >  			goto out;
> > -		}
> >  		cond_resched();
> > -		goto retry_flush_nodes;
> > +		goto retry_flush_quotas;
> >  	}
> >  
> >  	/*
> > -- 
> > The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
> > a Linux Foundation Collaborative Project
> 
> 
> _______________________________________________
> Linux-f2fs-devel mailing list
> Linux-f2fs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

Re: [f2fs-dev] [PATCH] f2fs: Avoid double lock for cp_rwsem

[Chao Yu]

Hi Sayali,

On 2020-4-23 18:13, Sayali Lokhande wrote:
> Call stack :
> f2fs_write_checkpoint()
> -> block_operations(sbi)
>     f2fs_lock_all(sbi);
>      down_write(&sbi->cp_rwsem); => write lock held
> <>

It looks the full race condition should be?

						Thread B
						- open()
						 - igrab()
						- write() write inline data
						- unlink()
> -> f2fs_sync_node_pages()
>     if (is_inline_node(page))
>      flush_inline_data()

	ilookup()

> 	page = f2fs_pagecache_get_page()
>          if (!page)
>            goto iput_out;
> 	iput_out:
						- close()
						 - iput()
> 	 iput(inode);
>           -> f2fs_evict_inode()
> 	      f2fs_truncate_blocks()
> 	       f2fs_lock_op()
> 	        down_read(&sbi->cp_rwsem); => read lock fail
>
> Signed-off-by: Sayali Lokhande <sayalil@codeaurora.org>
> ---
>  fs/f2fs/checkpoint.c | 10 ++++------
>  1 file changed, 4 insertions(+), 6 deletions(-)
>
> diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
> index 5ba649e..5c504cf 100644
> --- a/fs/f2fs/checkpoint.c
> +++ b/fs/f2fs/checkpoint.c
> @@ -1219,21 +1219,19 @@ static int block_operations(struct f2fs_sb_info *sbi)
>  		goto retry_flush_quotas;
>  	}
>
> -retry_flush_nodes:
>  	down_write(&sbi->node_write);
>
>  	if (get_pages(sbi, F2FS_DIRTY_NODES)) {
>  		up_write(&sbi->node_write);

Call up_write(&sbi->node_change) here could wake up threads earlier who hang on
this lock, how do you think?

Thanks,

> +		f2fs_unlock_all(sbi);
>  		atomic_inc(&sbi->wb_sync_req[NODE]);
>  		err = f2fs_sync_node_pages(sbi, &wbc, false, FS_CP_NODE_IO);
>  		atomic_dec(&sbi->wb_sync_req[NODE]);
> -		if (err) {
> -			up_write(&sbi->node_change);
> -			f2fs_unlock_all(sbi);
> +		up_write(&sbi->node_change);
> +		if (err)
>  			goto out;
> -		}
>  		cond_resched();
> -		goto retry_flush_nodes;
> +		goto retry_flush_quotas;
>  	}
>
>  	/*
>