[PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Darrick J. Wong]

Use the new error code passed to the directio end_io function to
decide if we're going to remap the blocks.  If an IO error happened
during an AIO DIO, we must skip the unwritten extent conversion to
avoid exposing stale blocks.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/ext4/ext4.h    |    2 ++
 fs/ext4/inode.c   |   23 +++++++++++++++--------
 fs/ext4/page-io.c |    2 +-
 3 files changed, 18 insertions(+), 9 deletions(-)

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 0662b28..363c701 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1523,6 +1523,8 @@ static inline void ext4_inode_aio_set(struct inode *inode, ext4_io_end_t *io)
 	inode->i_private = io;
 }
 
+extern void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end);
+
 /*
  * Inode dynamic state flags
  */
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 9db04dd..a302094 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -3166,9 +3166,6 @@ static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
 {
         ext4_io_end_t *io_end = iocb->private;
 
-	if (size <= 0)
-		return 0;
-
 	/* if not async direct IO just return */
 	if (!io_end)
 		return 0;
@@ -3179,11 +3176,21 @@ static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
 		  size);
 
 	iocb->private = NULL;
-	io_end->offset = offset;
-	io_end->size = size;
-	ext4_put_io_end(io_end);
-
-	return 0;
+	/*
+	 * If an IO error happened, skip the unwritten extent conversion
+	 * to avoid disclosing old disk contents.
+	 */
+	if (size <= 0) {
+		if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
+			ext4_clear_io_unwritten_flag(io_end);
+			if (ext4_handle_valid(io_end->handle))
+				ext4_journal_stop(io_end->handle);
+		}
+	} else {
+		io_end->offset = offset;
+		io_end->size = size;
+	}
+	return ext4_put_io_end(io_end);
 }
 
 /*
diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c
index 090b349..a0bc182b 100644
--- a/fs/ext4/page-io.c
+++ b/fs/ext4/page-io.c
@@ -139,7 +139,7 @@ static void ext4_release_io_end(ext4_io_end_t *io_end)
 	kmem_cache_free(io_end_cachep, io_end);
 }
 
-static void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
+void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
 {
 	struct inode *inode = io_end->inode;
 

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Christoph Hellwig]

Might help to tell that this is on top of a direct-io.c patch from the
XFS tree.

I don't think clearing any flags is the right thing - now that we
always call ->end_io the code dealing with it in ext4_ext_direct_IO
can simply be moved to the ->end_io handler.

Something like the untested patch below:

diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 9db04dd..b741c79 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -3166,23 +3166,25 @@ static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
 {
         ext4_io_end_t *io_end = iocb->private;
 
-	if (size <= 0)
-		return 0;
-
 	/* if not async direct IO just return */
 	if (!io_end)
 		return 0;
 
+	if (size <= 0) {
+		WARN_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
+		goto out;
+	}
+
 	ext_debug("ext4_end_io_dio(): io_end 0x%p "
 		  "for inode %lu, iocb 0x%p, offset %llu, size %zd\n",
  		  iocb->private, io_end->inode->i_ino, iocb, offset,
 		  size);
 
-	iocb->private = NULL;
 	io_end->offset = offset;
 	io_end->size = size;
+out:
 	ext4_put_io_end(io_end);

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Jan Kara]

On Wed 17-02-16 22:01:48, Christoph Hellwig wrote:
> Might help to tell that this is on top of a direct-io.c patch from the
> XFS tree.
> 
> I don't think clearing any flags is the right thing - now that we
> always call ->end_io the code dealing with it in ext4_ext_direct_IO
> can simply be moved to the ->end_io handler.
> 
> Something like the untested patch below:

Yeah, this looks good to me.

								Honza
> 
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 9db04dd..b741c79 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -3166,23 +3166,25 @@ static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
>  {
>          ext4_io_end_t *io_end = iocb->private;
>  
> -	if (size <= 0)
> -		return 0;
> -
>  	/* if not async direct IO just return */
>  	if (!io_end)
>  		return 0;
>  
> +	if (size <= 0) {
> +		WARN_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
> +		goto out;
> +	}
> +
>  	ext_debug("ext4_end_io_dio(): io_end 0x%p "
>  		  "for inode %lu, iocb 0x%p, offset %llu, size %zd\n",
>   		  iocb->private, io_end->inode->i_ino, iocb, offset,
>  		  size);
>  
> -	iocb->private = NULL;
>  	io_end->offset = offset;
>  	io_end->size = size;
> +out:
>  	ext4_put_io_end(io_end);
> -
> +	iocb->private = NULL;
>  	return 0;
>  }
>  
> @@ -3306,16 +3308,6 @@ static ssize_t ext4_ext_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
>  	if (io_end) {
>  		ext4_inode_aio_set(inode, NULL);
>  		ext4_put_io_end(io_end);
> -		/*
> -		 * When no IO was submitted ext4_end_io_dio() was not
> -		 * called so we have to put iocb's reference.
> -		 */
> -		if (ret <= 0 && ret != -EIOCBQUEUED && iocb->private) {
> -			WARN_ON(iocb->private != io_end);
> -			WARN_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
> -			ext4_put_io_end(io_end);
> -			iocb->private = NULL;
> -		}
>  	}
>  	if (ret > 0 && !overwrite && ext4_test_inode_state(inode,
>  						EXT4_STATE_DIO_UNWRITTEN)) {
> --
> 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
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Dave Chinner]

On Wed, Feb 17, 2016 at 10:01:48PM -0800, Christoph Hellwig wrote:
> Might help to tell that this is on top of a direct-io.c patch from the
> XFS tree.
> 
> I don't think clearing any flags is the right thing - now that we
> always call ->end_io the code dealing with it in ext4_ext_direct_IO
> can simply be moved to the ->end_io handler.
> 
> Something like the untested patch below:
> 
> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> index 9db04dd..b741c79 100644
> --- a/fs/ext4/inode.c
> +++ b/fs/ext4/inode.c
> @@ -3166,23 +3166,25 @@ static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
>  {
>          ext4_io_end_t *io_end = iocb->private;
>  
> -	if (size <= 0)
> -		return 0;
> -
>  	/* if not async direct IO just return */
>  	if (!io_end)
>  		return 0;
>  
> +	if (size <= 0) {
> +		WARN_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
> +		goto out;
> +	}

That will still issue a warning when an I/O error occurs on an
unwritten extent.

> +
>  	ext_debug("ext4_end_io_dio(): io_end 0x%p "
>  		  "for inode %lu, iocb 0x%p, offset %llu, size %zd\n",
>   		  iocb->private, io_end->inode->i_ino, iocb, offset,
>  		  size);
>  
> -	iocb->private = NULL;
>  	io_end->offset = offset;
>  	io_end->size = size;
> +out:
>  	ext4_put_io_end(io_end);

Won't that now call ext4_put_io_end() ->
ext4_convert_unwritten_extents() with an uninitialised offset and
size?

i.e. I don't think this prevents warnings, and may make things
worse when real errors occur....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Jan Kara]

[-- Attachment #1: Type: text/plain, Size: 2196 bytes --]

On Fri 19-02-16 09:02:32, Dave Chinner wrote:
> On Wed, Feb 17, 2016 at 10:01:48PM -0800, Christoph Hellwig wrote:
> > Might help to tell that this is on top of a direct-io.c patch from the
> > XFS tree.
> > 
> > I don't think clearing any flags is the right thing - now that we
> > always call ->end_io the code dealing with it in ext4_ext_direct_IO
> > can simply be moved to the ->end_io handler.
> > 
> > Something like the untested patch below:
> > 
> > diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
> > index 9db04dd..b741c79 100644
> > --- a/fs/ext4/inode.c
> > +++ b/fs/ext4/inode.c
> > @@ -3166,23 +3166,25 @@ static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
> >  {
> >          ext4_io_end_t *io_end = iocb->private;
> >  
> > -	if (size <= 0)
> > -		return 0;
> > -
> >  	/* if not async direct IO just return */
> >  	if (!io_end)
> >  		return 0;
> >  
> > +	if (size <= 0) {
> > +		WARN_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
> > +		goto out;
> > +	}
> 
> That will still issue a warning when an I/O error occurs on an
> unwritten extent.

Ah, correct.

> > +
> >  	ext_debug("ext4_end_io_dio(): io_end 0x%p "
> >  		  "for inode %lu, iocb 0x%p, offset %llu, size %zd\n",
> >   		  iocb->private, io_end->inode->i_ino, iocb, offset,
> >  		  size);
> >  
> > -	iocb->private = NULL;
> >  	io_end->offset = offset;
> >  	io_end->size = size;
> > +out:
> >  	ext4_put_io_end(io_end);
> 
> Won't that now call ext4_put_io_end() ->
> ext4_convert_unwritten_extents() with an uninitialised offset and
> size?
> 
> i.e. I don't think this prevents warnings, and may make things
> worse when real errors occur....

Yeah, if IO error occurs while writing to unwritten extent we need to just
destroy the IO end without doing the extent conversion (since we don't know
how much got written). Attached patch should fix the issue - full xfstests
run is in progress but a quick check using generic/299 has passed.

How do we merge this? It depends on the changes in Dave's tree so do we
merge it via that? I have other ext4 changes pending in this area so Ted
would then have to pull some branch from Dave's tree. Guys?

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

[-- Attachment #2: 0001-ext4-Fix-data-exposure-after-failed-AIO-DIO.patch --]
[-- Type: text/x-patch, Size: 4185 bytes --]

>From fe96f559b86e609b8d98da03b5291a9a0da1d9a8 Mon Sep 17 00:00:00 2001
From: Jan Kara <jack@suse.cz>
Date: Fri, 19 Feb 2016 13:53:11 +0100
Subject: [PATCH] ext4: Fix data exposure after failed AIO DIO

When AIO DIO fails e.g. due to IO error, we must not convert unwritten
extents as that will expose uninitialized data. Handle this case
by clearing unwritten flag from io_end in case of error and thus
preventing extent conversion.

Signed-off-by: Jan Kara <jack@suse.cz>
---
 fs/ext4/ext4.h    | 30 +++++++++++++++++++++---------
 fs/ext4/inode.c   | 21 ++++++++-------------
 fs/ext4/page-io.c | 10 ----------
 3 files changed, 29 insertions(+), 32 deletions(-)

diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index 0662b285dc8a..56c12df107ab 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1504,15 +1504,6 @@ static inline int ext4_valid_inum(struct super_block *sb, unsigned long ino)
 		 ino <= le32_to_cpu(EXT4_SB(sb)->s_es->s_inodes_count));
 }
 
-static inline void ext4_set_io_unwritten_flag(struct inode *inode,
-					      struct ext4_io_end *io_end)
-{
-	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
-		io_end->flag |= EXT4_IO_END_UNWRITTEN;
-		atomic_inc(&EXT4_I(inode)->i_unwritten);
-	}
-}
-
 static inline ext4_io_end_t *ext4_inode_aio(struct inode *inode)
 {
 	return inode->i_private;
@@ -3293,6 +3284,27 @@ extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
 extern int ext4_resize_begin(struct super_block *sb);
 extern void ext4_resize_end(struct super_block *sb);
 
+static inline void ext4_set_io_unwritten_flag(struct inode *inode,
+					      struct ext4_io_end *io_end)
+{
+	if (!(io_end->flag & EXT4_IO_END_UNWRITTEN)) {
+		io_end->flag |= EXT4_IO_END_UNWRITTEN;
+		atomic_inc(&EXT4_I(inode)->i_unwritten);
+	}
+}
+
+static inline void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
+{
+	struct inode *inode = io_end->inode;
+
+	if (io_end->flag & EXT4_IO_END_UNWRITTEN) {
+		io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
+		/* Wake up anyone waiting on unwritten extent conversion */
+		if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
+			wake_up_all(ext4_ioend_wq(inode));
+	}
+}
+
 #endif	/* __KERNEL__ */
 
 #define EFSBADCRC	EBADMSG		/* Bad CRC detected */
diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c
index 9db04dd9b88a..2b98171a9432 100644
--- a/fs/ext4/inode.c
+++ b/fs/ext4/inode.c
@@ -3166,9 +3166,6 @@ static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
 {
         ext4_io_end_t *io_end = iocb->private;
 
-	if (size <= 0)
-		return 0;
-
 	/* if not async direct IO just return */
 	if (!io_end)
 		return 0;
@@ -3179,6 +3176,14 @@ static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
 		  size);
 
 	iocb->private = NULL;
+	/*
+	 * Error during AIO DIO. We cannot convert unwritten extents as the
+	 * data was not written. Just clear the unwritten flag and drop io_end.
+	 */
+	if (size <= 0) {
+		ext4_clear_io_unwritten_flag(io_end);
+		size = 0;
+	}
 	io_end->offset = offset;
 	io_end->size = size;
 	ext4_put_io_end(io_end);
@@ -3306,16 +3311,6 @@ static ssize_t ext4_ext_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
 	if (io_end) {
 		ext4_inode_aio_set(inode, NULL);
 		ext4_put_io_end(io_end);
-		/*
-		 * When no IO was submitted ext4_end_io_dio() was not
-		 * called so we have to put iocb's reference.
-		 */
-		if (ret <= 0 && ret != -EIOCBQUEUED && iocb->private) {
-			WARN_ON(iocb->private != io_end);
-			WARN_ON(io_end->flag & EXT4_IO_END_UNWRITTEN);
-			ext4_put_io_end(io_end);
-			iocb->private = NULL;
-		}
 	}
 	if (ret > 0 && !overwrite && ext4_test_inode_state(inode,
 						EXT4_STATE_DIO_UNWRITTEN)) {
diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c
index 090b3498638e..f49a87c4fb63 100644
--- a/fs/ext4/page-io.c
+++ b/fs/ext4/page-io.c
@@ -139,16 +139,6 @@ static void ext4_release_io_end(ext4_io_end_t *io_end)
 	kmem_cache_free(io_end_cachep, io_end);
 }
 
-static void ext4_clear_io_unwritten_flag(ext4_io_end_t *io_end)
-{
-	struct inode *inode = io_end->inode;
-
-	io_end->flag &= ~EXT4_IO_END_UNWRITTEN;
-	/* Wake up anyone waiting on unwritten extent conversion */
-	if (atomic_dec_and_test(&EXT4_I(inode)->i_unwritten))
-		wake_up_all(ext4_ioend_wq(inode));
-}

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Theodore Ts'o]

On Fri, Feb 19, 2016 at 02:18:29PM +0100, Jan Kara wrote:
> Yeah, if IO error occurs while writing to unwritten extent we need to just
> destroy the IO end without doing the extent conversion (since we don't know
> how much got written). Attached patch should fix the issue - full xfstests
> run is in progress but a quick check using generic/299 has passed.
> 
> How do we merge this? It depends on the changes in Dave's tree so do we
> merge it via that? I have other ext4 changes pending in this area so Ted
> would then have to pull some branch from Dave's tree. Guys?

I've been asking Darrick this on the weekly ext4 teleconference for a
while.  If DIO changes can be put this on a separate git branch, we
can merge it that way.  Worse case, the DIO fix can go in during the
next merge window, and then the ext4 fixup can go in post rc2.  After
all, we've been living this for a while...

					- Ted

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Dave Chinner]

On Fri, Feb 19, 2016 at 02:18:29PM +0100, Jan Kara wrote:
> On Fri 19-02-16 09:02:32, Dave Chinner wrote:
> > Won't that now call ext4_put_io_end() ->
> > ext4_convert_unwritten_extents() with an uninitialised offset and
> > size?
> > 
> > i.e. I don't think this prevents warnings, and may make things
> > worse when real errors occur....
> 
> Yeah, if IO error occurs while writing to unwritten extent we need to just
> destroy the IO end without doing the extent conversion (since we don't know
> how much got written). Attached patch should fix the issue - full xfstests
> run is in progress but a quick check using generic/299 has passed.
> 
> How do we merge this? It depends on the changes in Dave's tree so do we
> merge it via that? I have other ext4 changes pending in this area so Ted
> would then have to pull some branch from Dave's tree. Guys?

The xfs-dio-fix-4.6 branch in the XFS tree is stable, so feel free
to pull it into other trees. However, it might be better for me to
append this patch to that branch once it is revewed and tested to
keep them all together in a stable branch. It can still be pulled
into other trees if needed...

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Jan Kara]

On Sun 21-02-16 17:28:26, Dave Chinner wrote:
> On Fri, Feb 19, 2016 at 02:18:29PM +0100, Jan Kara wrote:
> > On Fri 19-02-16 09:02:32, Dave Chinner wrote:
> > > Won't that now call ext4_put_io_end() ->
> > > ext4_convert_unwritten_extents() with an uninitialised offset and
> > > size?
> > > 
> > > i.e. I don't think this prevents warnings, and may make things
> > > worse when real errors occur....
> > 
> > Yeah, if IO error occurs while writing to unwritten extent we need to just
> > destroy the IO end without doing the extent conversion (since we don't know
> > how much got written). Attached patch should fix the issue - full xfstests
> > run is in progress but a quick check using generic/299 has passed.
> > 
> > How do we merge this? It depends on the changes in Dave's tree so do we
> > merge it via that? I have other ext4 changes pending in this area so Ted
> > would then have to pull some branch from Dave's tree. Guys?
> 
> The xfs-dio-fix-4.6 branch in the XFS tree is stable, so feel free
> to pull it into other trees. However, it might be better for me to
> append this patch to that branch once it is revewed and tested to
> keep them all together in a stable branch. It can still be pulled
> into other trees if needed...

Full xfstests run completed for me fine with the patch so it should be
good to go in your tree in that regard. So additional review would be fine
though. Darrick, can you have a look?

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

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Christoph Hellwig]

Thanks Jan,

this looks fine to me:

Reviewed-by: Christoph Hellwig <hch@lst.de>

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Darrick J. Wong]

On Mon, Feb 22, 2016 at 09:19:14AM +0100, Jan Kara wrote:
> On Sun 21-02-16 17:28:26, Dave Chinner wrote:
> > On Fri, Feb 19, 2016 at 02:18:29PM +0100, Jan Kara wrote:
> > > On Fri 19-02-16 09:02:32, Dave Chinner wrote:
> > > > Won't that now call ext4_put_io_end() ->
> > > > ext4_convert_unwritten_extents() with an uninitialised offset and
> > > > size?
> > > > 
> > > > i.e. I don't think this prevents warnings, and may make things
> > > > worse when real errors occur....
> > > 
> > > Yeah, if IO error occurs while writing to unwritten extent we need to just
> > > destroy the IO end without doing the extent conversion (since we don't know
> > > how much got written). Attached patch should fix the issue - full xfstests
> > > run is in progress but a quick check using generic/299 has passed.
> > > 
> > > How do we merge this? It depends on the changes in Dave's tree so do we
> > > merge it via that? I have other ext4 changes pending in this area so Ted
> > > would then have to pull some branch from Dave's tree. Guys?
> > 
> > The xfs-dio-fix-4.6 branch in the XFS tree is stable, so feel free
> > to pull it into other trees. However, it might be better for me to
> > append this patch to that branch once it is revewed and tested to
> > keep them all together in a stable branch. It can still be pulled
> > into other trees if needed...
> 
> Full xfstests run completed for me fine with the patch so it should be
> good to go in your tree in that regard. So additional review would be fine
> though. Darrick, can you have a look?

Looks good to me and passes generic/25[02], so
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>

--D

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

Re: [PATCH] ext4: use directio end_io error status to finish unwritten aio dio correctly

[Dave Chinner]

On Fri, Feb 19, 2016 at 02:18:29PM +0100, Jan Kara wrote:
> How do we merge this? It depends on the changes in Dave's tree so do we
> merge it via that? I have other ext4 changes pending in this area so Ted
> would then have to pull some branch from Dave's tree. Guys?
...
> Subject: [PATCH] ext4: Fix data exposure after failed AIO DIO

Now committed and pushed to the xfs-dio-fixes-4.6 branch, and merged
back into the XFS for-next branch so it will be picked up by
linux-next builds.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com