[PATCH 1/2] f2fs: fix inconsistency of block count during recovery

[PATCH 1/2] f2fs: fix inconsistency of block count during recovery

[Jaegeuk Kim]

Currently f2fs recovers the dentry of fsynced files.
When power-off-recovery is conducted, this newly recovered inode should increase
node block count as well as inode block count.

This patch resolves this inconsistency that results in:

1. create a file
2. write data
3. fsync
4. reboot without sync
5. mount and recover the file
6. node block count is 1 and inode block count is 2
 : fall into the inconsistent state
7. unlink the file
 : trigger the following BUG_ON

------------[ cut here ]------------
kernel BUG at /home/zeus/f2fs_test/src/fs/f2fs/f2fs.h:716!
Call Trace:
 [<ffffffffa0344100>] ? get_node_page+0x50/0x1a0 [f2fs]
 [<ffffffffa0344bfc>] remove_inode_page+0x8c/0x100 [f2fs]
 [<ffffffffa03380f0>] ? f2fs_evict_inode+0x180/0x2d0 [f2fs]
 [<ffffffffa033812e>] f2fs_evict_inode+0x1be/0x2d0 [f2fs]
 [<ffffffff811c7a67>] evict+0xa7/0x1a0
 [<ffffffff811c82b5>] iput+0x105/0x190
 [<ffffffff811c2b30>] d_kill+0xe0/0x120
 [<ffffffff811c2c57>] dput+0xe7/0x1e0
 [<ffffffff811acc3d>] __fput+0x19d/0x2d0
 [<ffffffff811acd7e>] ____fput+0xe/0x10
 [<ffffffff81070645>] task_work_run+0xb5/0xe0
 [<ffffffff81002941>] do_notify_resume+0x71/0xb0
 [<ffffffff8175f14a>] int_signal+0x12/0x17

Reported-by: Chris Fries <C.Fries@motorola.com>
Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
---
 fs/f2fs/node.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index 3df43b4..9641534 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1492,6 +1492,8 @@ int recover_inode_page(struct f2fs_sb_info *sbi, struct page *page)
 	new_ni = old_ni;
 	new_ni.ino = ino;
 
+	if (!inc_valid_node_count(sbi, NULL, 1))
+		WARN_ON(1);
 	set_node_addr(sbi, &new_ni, NEW_ADDR);
 	inc_valid_inode_count(sbi);
 
-- 
1.8.1.3.566.gaa39828

[PATCH 2/2] f2fs: fix the inconsistent state of data pages

[Jaegeuk Kim]

In get_lock_data_page, if there is a data race between get_dnode_of_data for
node and grab_cache_page for data, f2fs is able to face with the following
BUG_ON(dn.data_blkaddr == NEW_ADDR).

kernel BUG at /home/zeus/f2fs_test/src/fs/f2fs/data.c:251!
 [<ffffffffa044966c>] get_lock_data_page+0x1ec/0x210 [f2fs]
Call Trace:
 [<ffffffffa043b089>] f2fs_readdir+0x89/0x210 [f2fs]
 [<ffffffff811a0920>] ? fillonedir+0x100/0x100
 [<ffffffff811a0920>] ? fillonedir+0x100/0x100
 [<ffffffff811a07f8>] vfs_readdir+0xb8/0xe0
 [<ffffffff811a0b4f>] sys_getdents+0x8f/0x110
 [<ffffffff816d7999>] system_call_fastpath+0x16/0x1b

This bug is able to be occurred when the block address of the data block is
changed after f2fs_put_dnode().
In order to avoid that, this patch fixes the lock order of node and data
blocks in which the node block lock is covered by the data block lock.

Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
---
 fs/f2fs/data.c | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 91ff93b..05fb5c6 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -233,18 +233,23 @@ struct page *get_lock_data_page(struct inode *inode, pgoff_t index)
 	struct page *page;
 	int err;
 
+repeat:
+	page = grab_cache_page(mapping, index);
+	if (!page)
+		return ERR_PTR(-ENOMEM);
+
 	set_new_dnode(&dn, inode, NULL, NULL, 0);
 	err = get_dnode_of_data(&dn, index, LOOKUP_NODE);
-	if (err)
+	if (err) {
+		f2fs_put_page(page, 1);
 		return ERR_PTR(err);
+	}
 	f2fs_put_dnode(&dn);
 
-	if (dn.data_blkaddr == NULL_ADDR)
+	if (dn.data_blkaddr == NULL_ADDR) {
+		f2fs_put_page(page, 1);
 		return ERR_PTR(-ENOENT);
-repeat:
-	page = grab_cache_page(mapping, index);
-	if (!page)
-		return ERR_PTR(-ENOMEM);
+	}
 
 	if (PageUptodate(page))
 		return page;
-- 
1.8.1.3.566.gaa39828

Re: [PATCH 1/2] f2fs: fix inconsistency of block count during recovery

[Namjae Jeon]

>
> diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
> index 3df43b4..9641534 100644
> --- a/fs/f2fs/node.c
> +++ b/fs/f2fs/node.c
> @@ -1492,6 +1492,8 @@ int recover_inode_page(struct f2fs_sb_info *sbi,
> struct page *page)
>  	new_ni = old_ni;
>  	new_ni.ino = ino;
>
Hi. Jaegeuk.

I have a minor comment.
> +	if (!inc_valid_node_count(sbi, NULL, 1))
> +		WARN_ON(1);
How about change WARN_ON(!inc_valid_node_count(sbi, NULL, 1)); ?

Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com>
Thanks.
>  	set_node_addr(sbi, &new_ni, NEW_ADDR);
>  	inc_valid_inode_count(sbi);
>
> --
> 1.8.1.3.566.gaa39828
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: [PATCH 2/2] f2fs: fix the inconsistent state of data pages

[Namjae Jeon]

2013/5/15, Jaegeuk Kim <jaegeuk.kim@samsung.com>:
> In get_lock_data_page, if there is a data race between get_dnode_of_data
> for
> node and grab_cache_page for data, f2fs is able to face with the following
> BUG_ON(dn.data_blkaddr == NEW_ADDR).
>
> kernel BUG at /home/zeus/f2fs_test/src/fs/f2fs/data.c:251!
>  [<ffffffffa044966c>] get_lock_data_page+0x1ec/0x210 [f2fs]
> Call Trace:
>  [<ffffffffa043b089>] f2fs_readdir+0x89/0x210 [f2fs]
>  [<ffffffff811a0920>] ? fillonedir+0x100/0x100
>  [<ffffffff811a0920>] ? fillonedir+0x100/0x100
>  [<ffffffff811a07f8>] vfs_readdir+0xb8/0xe0
>  [<ffffffff811a0b4f>] sys_getdents+0x8f/0x110
>  [<ffffffff816d7999>] system_call_fastpath+0x16/0x1b
>
> This bug is able to be occurred when the block address of the data block is
> changed after f2fs_put_dnode().
> In order to avoid that, this patch fixes the lock order of node and data
> blocks in which the node block lock is covered by the data block lock.
>
> Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
> ---
>  fs/f2fs/data.c | 17 +++++++++++------
>  1 file changed, 11 insertions(+), 6 deletions(-)
>
> diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
> index 91ff93b..05fb5c6 100644
> --- a/fs/f2fs/data.c
> +++ b/fs/f2fs/data.c
> @@ -233,18 +233,23 @@ struct page *get_lock_data_page(struct inode *inode,
> pgoff_t index)
>  	struct page *page;
>  	int err;
>
> +repeat:
> +	page = grab_cache_page(mapping, index);
> +	if (!page)
> +		return ERR_PTR(-ENOMEM);
> +
>  	set_new_dnode(&dn, inode, NULL, NULL, 0);
>  	err = get_dnode_of_data(&dn, index, LOOKUP_NODE);
> -	if (err)
> +	if (err) {
> +		f2fs_put_page(page, 1);
>  		return ERR_PTR(err);
> +	}
>  	f2fs_put_dnode(&dn);
>
> -	if (dn.data_blkaddr == NULL_ADDR)
> +	if (dn.data_blkaddr == NULL_ADDR) {
> +		f2fs_put_page(page, 1);
>  		return ERR_PTR(-ENOENT);
> -repeat:
> -	page = grab_cache_page(mapping, index);
> -	if (!page)
> -		return ERR_PTR(-ENOMEM);
> +	}
>
>  	if (PageUptodate(page))
>  		return page;
Is there no need to move PageUptodate condition checking to
grab_cache_page next ?

Thanks.
> --
> 1.8.1.3.566.gaa39828
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
>

Re: [PATCH 1/2] f2fs: fix inconsistency of block count during recovery

[Jaegeuk Kim]

[-- Attachment #1: Type: text/plain, Size: 1160 bytes --]

2013-05-15 (수), 12:47 +0900, Namjae Jeon:
> >
> > diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
> > index 3df43b4..9641534 100644
> > --- a/fs/f2fs/node.c
> > +++ b/fs/f2fs/node.c
> > @@ -1492,6 +1492,8 @@ int recover_inode_page(struct f2fs_sb_info *sbi,
> > struct page *page)
> >  	new_ni = old_ni;
> >  	new_ni.ino = ino;
> >
> Hi. Jaegeuk.
> 
> I have a minor comment.
> > +	if (!inc_valid_node_count(sbi, NULL, 1))
> > +		WARN_ON(1);

Hi Namjae,

Negative since inc_valid_node_count() is not for debugging.
IMO, for readability, we need to make clear what the function call is
used for.
Thank you for the review. :)

> How about change WARN_ON(!inc_valid_node_count(sbi, NULL, 1)); ?
> 
> Reviewed-by: Namjae Jeon <namjae.jeon@samsung.com>
> Thanks.
> >  	set_node_addr(sbi, &new_ni, NEW_ADDR);
> >  	inc_valid_inode_count(sbi);
> >
> > --
> > 1.8.1.3.566.gaa39828
> >
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >

-- 
Jaegeuk Kim
Samsung

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH 2/2] f2fs: fix the inconsistent state of data pages

[Jaegeuk Kim]

[-- Attachment #1: Type: text/plain, Size: 2749 bytes --]

2013-05-15 (수), 13:04 +0900, Namjae Jeon:
> 2013/5/15, Jaegeuk Kim <jaegeuk.kim@samsung.com>:
> > In get_lock_data_page, if there is a data race between get_dnode_of_data
> > for
> > node and grab_cache_page for data, f2fs is able to face with the following
> > BUG_ON(dn.data_blkaddr == NEW_ADDR).
> >
> > kernel BUG at /home/zeus/f2fs_test/src/fs/f2fs/data.c:251!
> >  [<ffffffffa044966c>] get_lock_data_page+0x1ec/0x210 [f2fs]
> > Call Trace:
> >  [<ffffffffa043b089>] f2fs_readdir+0x89/0x210 [f2fs]
> >  [<ffffffff811a0920>] ? fillonedir+0x100/0x100
> >  [<ffffffff811a0920>] ? fillonedir+0x100/0x100
> >  [<ffffffff811a07f8>] vfs_readdir+0xb8/0xe0
> >  [<ffffffff811a0b4f>] sys_getdents+0x8f/0x110
> >  [<ffffffff816d7999>] system_call_fastpath+0x16/0x1b
> >
> > This bug is able to be occurred when the block address of the data block is
> > changed after f2fs_put_dnode().
> > In order to avoid that, this patch fixes the lock order of node and data
> > blocks in which the node block lock is covered by the data block lock.
> >
> > Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
> > ---
> >  fs/f2fs/data.c | 17 +++++++++++------
> >  1 file changed, 11 insertions(+), 6 deletions(-)
> >
> > diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
> > index 91ff93b..05fb5c6 100644
> > --- a/fs/f2fs/data.c
> > +++ b/fs/f2fs/data.c
> > @@ -233,18 +233,23 @@ struct page *get_lock_data_page(struct inode *inode,
> > pgoff_t index)
> >  	struct page *page;
> >  	int err;
> >
> > +repeat:
> > +	page = grab_cache_page(mapping, index);
> > +	if (!page)
> > +		return ERR_PTR(-ENOMEM);
> > +
> >  	set_new_dnode(&dn, inode, NULL, NULL, 0);
> >  	err = get_dnode_of_data(&dn, index, LOOKUP_NODE);
> > -	if (err)
> > +	if (err) {
> > +		f2fs_put_page(page, 1);
> >  		return ERR_PTR(err);
> > +	}
> >  	f2fs_put_dnode(&dn);
> >
> > -	if (dn.data_blkaddr == NULL_ADDR)
> > +	if (dn.data_blkaddr == NULL_ADDR) {
> > +		f2fs_put_page(page, 1);
> >  		return ERR_PTR(-ENOENT);
> > -repeat:
> > -	page = grab_cache_page(mapping, index);
> > -	if (!page)
> > -		return ERR_PTR(-ENOMEM);
> > +	}
> >
> >  	if (PageUptodate(page))
> >  		return page;
> Is there no need to move PageUptodate condition checking to
> grab_cache_page next ?

For the data consistency, I'd like to check index in its node block
prior to this.
Thanks,

> 
> Thanks.
> > --
> > 1.8.1.3.566.gaa39828
> >
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at  http://www.tux.org/lkml/
> >

-- 
Jaegeuk Kim
Samsung

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]