[PATCH v4 0/3] xfs: fix inode use-after-free during log recovery

[PATCH v4 0/3] xfs: fix inode use-after-free during log recovery

[Darrick J. Wong]

Hi all,

In this second series, I try to fix a use-after-free that I discovered
during development of the dfops freezer, where BUI recovery releases the
inode even if it requeues itself.  If the inode gets reclaimed, the fs
corrupts memory and explodes.  The fix is to make the dfops capture
struct take over ownership of the inodes if there's any more work to be
done.  This is a bit clunky, but it's a simpler mechanism than saving
inode pointers and inode numbers and introducing tagged structures so
that we can distinguish one from the other.

v2: rebase atop the new defer capture code
v3: only capture one inode, move as much of the defer capture code to
xfs_defer.c as we can
v4: make defer capture ihold the inode, and the caller still gets to
iunlock and irele it

If you're going to start using this mess, you probably ought to just
pull from my git trees, which are linked below.

This is an extraordinary way to destroy everything.  Enjoy!
Comments and questions are, as always, welcome.

--D

kernel git tree:
https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=fix-bmap-intent-recovery-5.10
---
 fs/xfs/libxfs/xfs_defer.c  |   43 ++++++++++++++++++++---
 fs/xfs/libxfs/xfs_defer.h  |   11 +++++-
 fs/xfs/xfs_bmap_item.c     |   84 ++++++++++++++++++++------------------------
 fs/xfs/xfs_extfree_item.c  |    2 +
 fs/xfs/xfs_log_recover.c   |    7 +++-
 fs/xfs/xfs_refcount_item.c |    2 +
 fs/xfs/xfs_rmap_item.c     |    2 +
 7 files changed, 95 insertions(+), 56 deletions(-)

[PATCH 1/3] xfs: clean up bmap intent item recovery checking

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

The bmap intent item checking code in xfs_bui_item_recover is spread all
over the function.  We should check the recovered log item at the top
before we allocate any resources or do anything else, so do that.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
---
 fs/xfs/xfs_bmap_item.c |   38 ++++++++++++--------------------------
 1 file changed, 12 insertions(+), 26 deletions(-)


diff --git a/fs/xfs/xfs_bmap_item.c b/fs/xfs/xfs_bmap_item.c
index 126df48dae5f..c1f2cc3c42cb 100644
--- a/fs/xfs/xfs_bmap_item.c
+++ b/fs/xfs/xfs_bmap_item.c
@@ -437,8 +437,6 @@ xfs_bui_item_recover(
 	xfs_fsblock_t			inode_fsb;
 	xfs_filblks_t			count;
 	xfs_exntst_t			state;
-	enum xfs_bmap_intent_type	type;
-	bool				op_ok;
 	unsigned int			bui_type;
 	int				whichfork;
 	int				error = 0;
@@ -456,16 +454,19 @@ xfs_bui_item_recover(
 			   XFS_FSB_TO_DADDR(mp, bmap->me_startblock));
 	inode_fsb = XFS_BB_TO_FSB(mp, XFS_FSB_TO_DADDR(mp,
 			XFS_INO_TO_FSB(mp, bmap->me_owner)));
-	switch (bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK) {
+	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
+			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
+	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
+			XFS_ATTR_FORK : XFS_DATA_FORK;
+	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
+	switch (bui_type) {
 	case XFS_BMAP_MAP:
 	case XFS_BMAP_UNMAP:
-		op_ok = true;
 		break;
 	default:
-		op_ok = false;
-		break;
+		return -EFSCORRUPTED;
 	}
-	if (!op_ok || startblock_fsb == 0 ||
+	if (startblock_fsb == 0 ||
 	    bmap->me_len == 0 ||
 	    inode_fsb == 0 ||
 	    startblock_fsb >= mp->m_sb.sb_dblocks ||
@@ -493,32 +494,17 @@ xfs_bui_item_recover(
 	if (VFS_I(ip)->i_nlink == 0)
 		xfs_iflags_set(ip, XFS_IRECOVERY);
 
-	/* Process deferred bmap item. */
-	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
-			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
-	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
-			XFS_ATTR_FORK : XFS_DATA_FORK;
-	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
-	switch (bui_type) {
-	case XFS_BMAP_MAP:
-	case XFS_BMAP_UNMAP:
-		type = bui_type;
-		break;
-	default:
-		XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
-		error = -EFSCORRUPTED;
-		goto err_inode;
-	}
 	xfs_trans_ijoin(tp, ip, 0);
 
 	count = bmap->me_len;
-	error = xfs_trans_log_finish_bmap_update(tp, budp, type, ip, whichfork,
-			bmap->me_startoff, bmap->me_startblock, &count, state);
+	error = xfs_trans_log_finish_bmap_update(tp, budp, bui_type, ip,
+			whichfork, bmap->me_startoff, bmap->me_startblock,
+			&count, state);
 	if (error)
 		goto err_inode;
 
 	if (count > 0) {
-		ASSERT(type == XFS_BMAP_UNMAP);
+		ASSERT(bui_type == XFS_BMAP_UNMAP);
 		irec.br_startblock = bmap->me_startblock;
 		irec.br_blockcount = count;
 		irec.br_startoff = bmap->me_startoff;

[PATCH 2/3] xfs: clean up xfs_bui_item_recover iget/trans_alloc/ilock ordering

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

In most places in XFS, we have a specific order in which we gather
resources: grab the inode, allocate a transaction, then lock the inode.
xfs_bui_item_recover doesn't do it in that order, so fix it to be more
consistent.  This also makes the error bailout code a bit less weird.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Brian Foster <bfoster@redhat.com>
---
 fs/xfs/xfs_bmap_item.c |   41 +++++++++++++++++++++++------------------
 1 file changed, 23 insertions(+), 18 deletions(-)


diff --git a/fs/xfs/xfs_bmap_item.c b/fs/xfs/xfs_bmap_item.c
index c1f2cc3c42cb..852411568d14 100644
--- a/fs/xfs/xfs_bmap_item.c
+++ b/fs/xfs/xfs_bmap_item.c
@@ -475,25 +475,26 @@ xfs_bui_item_recover(
 	    (bmap->me_flags & ~XFS_BMAP_EXTENT_FLAGS))
 		return -EFSCORRUPTED;
 
-	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate,
-			XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK), 0, 0, &tp);
-	if (error)
-		return error;
-
-	budp = xfs_trans_get_bud(tp, buip);
-
 	/* Grab the inode. */
-	error = xfs_iget(mp, tp, bmap->me_owner, 0, XFS_ILOCK_EXCL, &ip);
+	error = xfs_iget(mp, NULL, bmap->me_owner, 0, 0, &ip);
 	if (error)
-		goto err_inode;
+		return error;
 
-	error = xfs_qm_dqattach_locked(ip, false);
+	error = xfs_qm_dqattach(ip);
 	if (error)
-		goto err_inode;
+		goto err_rele;
 
 	if (VFS_I(ip)->i_nlink == 0)
 		xfs_iflags_set(ip, XFS_IRECOVERY);
 
+	/* Allocate transaction and do the work. */
+	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate,
+			XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK), 0, 0, &tp);
+	if (error)
+		goto err_rele;
+
+	budp = xfs_trans_get_bud(tp, buip);
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
 	xfs_trans_ijoin(tp, ip, 0);
 
 	count = bmap->me_len;
@@ -501,7 +502,7 @@ xfs_bui_item_recover(
 			whichfork, bmap->me_startoff, bmap->me_startblock,
 			&count, state);
 	if (error)
-		goto err_inode;
+		goto err_cancel;
 
 	if (count > 0) {
 		ASSERT(bui_type == XFS_BMAP_UNMAP);
@@ -512,17 +513,21 @@ xfs_bui_item_recover(
 		xfs_bmap_unmap_extent(tp, ip, &irec);
 	}
 
+	/* Commit transaction, which frees the transaction. */
 	error = xfs_defer_ops_capture_and_commit(tp, capture_list);
+	if (error)
+		goto err_unlock;
+
 	xfs_iunlock(ip, XFS_ILOCK_EXCL);
 	xfs_irele(ip);
-	return error;
+	return 0;
 
-err_inode:
+err_cancel:
 	xfs_trans_cancel(tp);
-	if (ip) {
-		xfs_iunlock(ip, XFS_ILOCK_EXCL);
-		xfs_irele(ip);
-	}
+err_unlock:
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+err_rele:
+	xfs_irele(ip);
 	return error;
 }
 

[PATCH 3/3] xfs: fix an incore inode UAF in xfs_bui_recover

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

In xfs_bui_item_recover, there exists a use-after-free bug with regards
to the inode that is involved in the bmap replay operation.  If the
mapping operation does not complete, we call xfs_bmap_unmap_extent to
create a deferred op to finish the unmapping work, and we retain a
pointer to the incore inode.

Unfortunately, the very next thing we do is commit the transaction and
drop the inode.  If reclaim tears down the inode before we try to finish
the defer ops, we dereference garbage and blow up.  Therefore, create a
way to join inodes to the defer ops freezer so that we can maintain the
xfs_inode reference until we're done with the inode.

Note: This imposes the requirement that there be enough memory to keep
every incore inode in memory throughout recovery.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
---
 fs/xfs/libxfs/xfs_defer.c  |   43 ++++++++++++++++++++++++++++++++++++++-----
 fs/xfs/libxfs/xfs_defer.h  |   11 +++++++++--
 fs/xfs/xfs_bmap_item.c     |    7 +++++--
 fs/xfs/xfs_extfree_item.c  |    2 +-
 fs/xfs/xfs_log_recover.c   |    7 ++++++-
 fs/xfs/xfs_refcount_item.c |    2 +-
 fs/xfs/xfs_rmap_item.c     |    2 +-
 7 files changed, 61 insertions(+), 13 deletions(-)


diff --git a/fs/xfs/libxfs/xfs_defer.c b/fs/xfs/libxfs/xfs_defer.c
index e19dc1ced7e6..4d2583758f3d 100644
--- a/fs/xfs/libxfs/xfs_defer.c
+++ b/fs/xfs/libxfs/xfs_defer.c
@@ -16,6 +16,7 @@
 #include "xfs_inode.h"
 #include "xfs_inode_item.h"
 #include "xfs_trace.h"
+#include "xfs_icache.h"
 
 /*
  * Deferred Operations in XFS
@@ -553,10 +554,14 @@ xfs_defer_move(
  * deferred ops state is transferred to the capture structure and the
  * transaction is then ready for the caller to commit it.  If there are no
  * intent items to capture, this function returns NULL.
+ *
+ * If capture_ip is not NULL, the capture structure will obtain an extra
+ * reference to the inode.
  */
 static struct xfs_defer_capture *
 xfs_defer_ops_capture(
-	struct xfs_trans		*tp)
+	struct xfs_trans		*tp,
+	struct xfs_inode		*capture_ip)
 {
 	struct xfs_defer_capture	*dfc;
 
@@ -582,6 +587,15 @@ xfs_defer_ops_capture(
 	/* Preserve the log reservation size. */
 	dfc->dfc_logres = tp->t_log_res;
 
+	/*
+	 * Grab an extra reference to this inode and attach it to the capture
+	 * structure.
+	 */
+	if (capture_ip) {
+		ihold(VFS_I(capture_ip));
+		dfc->dfc_capture_ip = capture_ip;
+	}
+
 	return dfc;
 }
 
@@ -592,24 +606,33 @@ xfs_defer_ops_release(
 	struct xfs_defer_capture	*dfc)
 {
 	xfs_defer_cancel_list(mp, &dfc->dfc_dfops);
+	if (dfc->dfc_capture_ip)
+		xfs_irele(dfc->dfc_capture_ip);
 	kmem_free(dfc);
 }
 
 /*
  * Capture any deferred ops and commit the transaction.  This is the last step
- * needed to finish a log intent item that we recovered from the log.
+ * needed to finish a log intent item that we recovered from the log.  If any
+ * of the deferred ops operate on an inode, the caller must pass in that inode
+ * so that the reference can be transferred to the capture structure.  The
+ * caller must hold ILOCK_EXCL on the inode, and must unlock it before calling
+ * xfs_defer_ops_continue.
  */
 int
 xfs_defer_ops_capture_and_commit(
 	struct xfs_trans		*tp,
+	struct xfs_inode		*capture_ip,
 	struct list_head		*capture_list)
 {
 	struct xfs_mount		*mp = tp->t_mountp;
 	struct xfs_defer_capture	*dfc;
 	int				error;
 
+	ASSERT(!capture_ip || xfs_isilocked(capture_ip, XFS_ILOCK_EXCL));
+
 	/* If we don't capture anything, commit transaction and exit. */
-	dfc = xfs_defer_ops_capture(tp);
+	dfc = xfs_defer_ops_capture(tp, capture_ip);
 	if (!dfc)
 		return xfs_trans_commit(tp);
 
@@ -626,16 +649,26 @@ xfs_defer_ops_capture_and_commit(
 
 /*
  * Attach a chain of captured deferred ops to a new transaction and free the
- * capture structure.
+ * capture structure.  If an inode was captured, it will be passed back to the
+ * caller with ILOCK_EXCL held and joined to the transaction with lockflags==0.
+ * The caller now owns the inode reference.
  */
 void
 xfs_defer_ops_continue(
 	struct xfs_defer_capture	*dfc,
-	struct xfs_trans		*tp)
+	struct xfs_trans		*tp,
+	struct xfs_inode		**captured_ipp)
 {
 	ASSERT(tp->t_flags & XFS_TRANS_PERM_LOG_RES);
 	ASSERT(!(tp->t_flags & XFS_TRANS_DIRTY));
 
+	/* Lock and join the captured inode to the new transaction. */
+	if (dfc->dfc_capture_ip) {
+		xfs_ilock(dfc->dfc_capture_ip, XFS_ILOCK_EXCL);
+		xfs_trans_ijoin(tp, dfc->dfc_capture_ip, 0);
+	}
+	*captured_ipp = dfc->dfc_capture_ip;
+
 	/* Move captured dfops chain and state to the transaction. */
 	list_splice_init(&dfc->dfc_dfops, &tp->t_dfops);
 	tp->t_flags |= dfc->dfc_tpflags;
diff --git a/fs/xfs/libxfs/xfs_defer.h b/fs/xfs/libxfs/xfs_defer.h
index 6cde6f0713f7..05472f71fffe 100644
--- a/fs/xfs/libxfs/xfs_defer.h
+++ b/fs/xfs/libxfs/xfs_defer.h
@@ -82,6 +82,12 @@ struct xfs_defer_capture {
 
 	/* Log reservation saved from the transaction. */
 	unsigned int		dfc_logres;
+
+	/*
+	 * An inode reference that must be maintained to complete the deferred
+	 * work.
+	 */
+	struct xfs_inode	*dfc_capture_ip;
 };
 
 /*
@@ -89,8 +95,9 @@ struct xfs_defer_capture {
  * This doesn't normally happen except log recovery.
  */
 int xfs_defer_ops_capture_and_commit(struct xfs_trans *tp,
-		struct list_head *capture_list);
-void xfs_defer_ops_continue(struct xfs_defer_capture *d, struct xfs_trans *tp);
+		struct xfs_inode *capture_ip, struct list_head *capture_list);
+void xfs_defer_ops_continue(struct xfs_defer_capture *d, struct xfs_trans *tp,
+		struct xfs_inode **captured_ipp);
 void xfs_defer_ops_release(struct xfs_mount *mp, struct xfs_defer_capture *d);
 
 #endif /* __XFS_DEFER_H__ */
diff --git a/fs/xfs/xfs_bmap_item.c b/fs/xfs/xfs_bmap_item.c
index 852411568d14..4570da07eb06 100644
--- a/fs/xfs/xfs_bmap_item.c
+++ b/fs/xfs/xfs_bmap_item.c
@@ -513,8 +513,11 @@ xfs_bui_item_recover(
 		xfs_bmap_unmap_extent(tp, ip, &irec);
 	}
 
-	/* Commit transaction, which frees the transaction. */
-	error = xfs_defer_ops_capture_and_commit(tp, capture_list);
+	/*
+	 * Commit transaction, which frees the transaction and saves the inode
+	 * for later replay activities.
+	 */
+	error = xfs_defer_ops_capture_and_commit(tp, ip, capture_list);
 	if (error)
 		goto err_unlock;
 
diff --git a/fs/xfs/xfs_extfree_item.c b/fs/xfs/xfs_extfree_item.c
index 17d36fe5cfd0..3920542f5736 100644
--- a/fs/xfs/xfs_extfree_item.c
+++ b/fs/xfs/xfs_extfree_item.c
@@ -627,7 +627,7 @@ xfs_efi_item_recover(
 
 	}
 
-	return xfs_defer_ops_capture_and_commit(tp, capture_list);
+	return xfs_defer_ops_capture_and_commit(tp, NULL, capture_list);
 
 abort_error:
 	xfs_trans_cancel(tp);
diff --git a/fs/xfs/xfs_log_recover.c b/fs/xfs/xfs_log_recover.c
index 001e1585ddc6..a8289adc1b29 100644
--- a/fs/xfs/xfs_log_recover.c
+++ b/fs/xfs/xfs_log_recover.c
@@ -2439,6 +2439,7 @@ xlog_finish_defer_ops(
 {
 	struct xfs_defer_capture *dfc, *next;
 	struct xfs_trans	*tp;
+	struct xfs_inode	*ip;
 	int			error = 0;
 
 	list_for_each_entry_safe(dfc, next, capture_list, dfc_list) {
@@ -2464,9 +2465,13 @@ xlog_finish_defer_ops(
 		 * from recovering a single intent item.
 		 */
 		list_del_init(&dfc->dfc_list);
-		xfs_defer_ops_continue(dfc, tp);
+		xfs_defer_ops_continue(dfc, tp, &ip);
 
 		error = xfs_trans_commit(tp);
+		if (ip) {
+			xfs_iunlock(ip, XFS_ILOCK_EXCL);
+			xfs_irele(ip);
+		}
 		if (error)
 			return error;
 	}
diff --git a/fs/xfs/xfs_refcount_item.c b/fs/xfs/xfs_refcount_item.c
index 0478374add64..ad895b48f365 100644
--- a/fs/xfs/xfs_refcount_item.c
+++ b/fs/xfs/xfs_refcount_item.c
@@ -544,7 +544,7 @@ xfs_cui_item_recover(
 	}
 
 	xfs_refcount_finish_one_cleanup(tp, rcur, error);
-	return xfs_defer_ops_capture_and_commit(tp, capture_list);
+	return xfs_defer_ops_capture_and_commit(tp, NULL, capture_list);
 
 abort_error:
 	xfs_refcount_finish_one_cleanup(tp, rcur, error);
diff --git a/fs/xfs/xfs_rmap_item.c b/fs/xfs/xfs_rmap_item.c
index 0d8fa707f079..1163f32c3e62 100644
--- a/fs/xfs/xfs_rmap_item.c
+++ b/fs/xfs/xfs_rmap_item.c
@@ -567,7 +567,7 @@ xfs_rui_item_recover(
 	}
 
 	xfs_rmap_finish_one_cleanup(tp, rcur, error);
-	return xfs_defer_ops_capture_and_commit(tp, capture_list);
+	return xfs_defer_ops_capture_and_commit(tp, NULL, capture_list);
 
 abort_error:
 	xfs_rmap_finish_one_cleanup(tp, rcur, error);

Re: [PATCH 3/3] xfs: fix an incore inode UAF in xfs_bui_recover

[Christoph Hellwig]

On Mon, Oct 05, 2020 at 11:20:50AM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> In xfs_bui_item_recover, there exists a use-after-free bug with regards
> to the inode that is involved in the bmap replay operation.  If the
> mapping operation does not complete, we call xfs_bmap_unmap_extent to
> create a deferred op to finish the unmapping work, and we retain a
> pointer to the incore inode.
> 
> Unfortunately, the very next thing we do is commit the transaction and
> drop the inode.  If reclaim tears down the inode before we try to finish
> the defer ops, we dereference garbage and blow up.  Therefore, create a
> way to join inodes to the defer ops freezer so that we can maintain the
> xfs_inode reference until we're done with the inode.
> 
> Note: This imposes the requirement that there be enough memory to keep
> every incore inode in memory throughout recovery.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> Reviewed-by: Brian Foster <bfoster@redhat.com>

Looks good,

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

[PATCH 1/3] xfs: clean up bmap intent item recovery checking

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

The bmap intent item checking code in xfs_bui_item_recover is spread all
over the function.  We should check the recovered log item at the top
before we allocate any resources or do anything else, so do that.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
---
 fs/xfs/xfs_bmap_item.c |   38 ++++++++++++--------------------------
 1 file changed, 12 insertions(+), 26 deletions(-)


diff --git a/fs/xfs/xfs_bmap_item.c b/fs/xfs/xfs_bmap_item.c
index 126df48dae5f..c1f2cc3c42cb 100644
--- a/fs/xfs/xfs_bmap_item.c
+++ b/fs/xfs/xfs_bmap_item.c
@@ -437,8 +437,6 @@ xfs_bui_item_recover(
 	xfs_fsblock_t			inode_fsb;
 	xfs_filblks_t			count;
 	xfs_exntst_t			state;
-	enum xfs_bmap_intent_type	type;
-	bool				op_ok;
 	unsigned int			bui_type;
 	int				whichfork;
 	int				error = 0;
@@ -456,16 +454,19 @@ xfs_bui_item_recover(
 			   XFS_FSB_TO_DADDR(mp, bmap->me_startblock));
 	inode_fsb = XFS_BB_TO_FSB(mp, XFS_FSB_TO_DADDR(mp,
 			XFS_INO_TO_FSB(mp, bmap->me_owner)));
-	switch (bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK) {
+	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
+			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
+	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
+			XFS_ATTR_FORK : XFS_DATA_FORK;
+	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
+	switch (bui_type) {
 	case XFS_BMAP_MAP:
 	case XFS_BMAP_UNMAP:
-		op_ok = true;
 		break;
 	default:
-		op_ok = false;
-		break;
+		return -EFSCORRUPTED;
 	}
-	if (!op_ok || startblock_fsb == 0 ||
+	if (startblock_fsb == 0 ||
 	    bmap->me_len == 0 ||
 	    inode_fsb == 0 ||
 	    startblock_fsb >= mp->m_sb.sb_dblocks ||
@@ -493,32 +494,17 @@ xfs_bui_item_recover(
 	if (VFS_I(ip)->i_nlink == 0)
 		xfs_iflags_set(ip, XFS_IRECOVERY);
 
-	/* Process deferred bmap item. */
-	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
-			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
-	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
-			XFS_ATTR_FORK : XFS_DATA_FORK;
-	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
-	switch (bui_type) {
-	case XFS_BMAP_MAP:
-	case XFS_BMAP_UNMAP:
-		type = bui_type;
-		break;
-	default:
-		XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
-		error = -EFSCORRUPTED;
-		goto err_inode;
-	}
 	xfs_trans_ijoin(tp, ip, 0);
 
 	count = bmap->me_len;
-	error = xfs_trans_log_finish_bmap_update(tp, budp, type, ip, whichfork,
-			bmap->me_startoff, bmap->me_startblock, &count, state);
+	error = xfs_trans_log_finish_bmap_update(tp, budp, bui_type, ip,
+			whichfork, bmap->me_startoff, bmap->me_startblock,
+			&count, state);
 	if (error)
 		goto err_inode;
 
 	if (count > 0) {
-		ASSERT(type == XFS_BMAP_UNMAP);
+		ASSERT(bui_type == XFS_BMAP_UNMAP);
 		irec.br_startblock = bmap->me_startblock;
 		irec.br_blockcount = count;
 		irec.br_startoff = bmap->me_startoff;

[PATCH 1/3] xfs: clean up bmap intent item recovery checking

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

The bmap intent item checking code in xfs_bui_item_recover is spread all
over the function.  We should check the recovered log item at the top
before we allocate any resources or do anything else, so do that.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
---
 fs/xfs/xfs_bmap_item.c |   38 ++++++++++++--------------------------
 1 file changed, 12 insertions(+), 26 deletions(-)


diff --git a/fs/xfs/xfs_bmap_item.c b/fs/xfs/xfs_bmap_item.c
index b73f0a0890a2..8f20eac72287 100644
--- a/fs/xfs/xfs_bmap_item.c
+++ b/fs/xfs/xfs_bmap_item.c
@@ -437,8 +437,6 @@ xfs_bui_item_recover(
 	xfs_fsblock_t			inode_fsb;
 	xfs_filblks_t			count;
 	xfs_exntst_t			state;
-	enum xfs_bmap_intent_type	type;
-	bool				op_ok;
 	unsigned int			bui_type;
 	int				whichfork;
 	int				error = 0;
@@ -456,16 +454,19 @@ xfs_bui_item_recover(
 			   XFS_FSB_TO_DADDR(mp, bmap->me_startblock));
 	inode_fsb = XFS_BB_TO_FSB(mp, XFS_FSB_TO_DADDR(mp,
 			XFS_INO_TO_FSB(mp, bmap->me_owner)));
-	switch (bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK) {
+	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
+			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
+	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
+			XFS_ATTR_FORK : XFS_DATA_FORK;
+	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
+	switch (bui_type) {
 	case XFS_BMAP_MAP:
 	case XFS_BMAP_UNMAP:
-		op_ok = true;
 		break;
 	default:
-		op_ok = false;
-		break;
+		return -EFSCORRUPTED;
 	}
-	if (!op_ok || startblock_fsb == 0 ||
+	if (startblock_fsb == 0 ||
 	    bmap->me_len == 0 ||
 	    inode_fsb == 0 ||
 	    startblock_fsb >= mp->m_sb.sb_dblocks ||
@@ -493,32 +494,17 @@ xfs_bui_item_recover(
 	if (VFS_I(ip)->i_nlink == 0)
 		xfs_iflags_set(ip, XFS_IRECOVERY);
 
-	/* Process deferred bmap item. */
-	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
-			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
-	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
-			XFS_ATTR_FORK : XFS_DATA_FORK;
-	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
-	switch (bui_type) {
-	case XFS_BMAP_MAP:
-	case XFS_BMAP_UNMAP:
-		type = bui_type;
-		break;
-	default:
-		XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
-		error = -EFSCORRUPTED;
-		goto err_inode;
-	}
 	xfs_trans_ijoin(tp, ip, 0);
 
 	count = bmap->me_len;
-	error = xfs_trans_log_finish_bmap_update(tp, budp, type, ip, whichfork,
-			bmap->me_startoff, bmap->me_startblock, &count, state);
+	error = xfs_trans_log_finish_bmap_update(tp, budp, bui_type, ip,
+			whichfork, bmap->me_startoff, bmap->me_startblock,
+			&count, state);
 	if (error)
 		goto err_inode;
 
 	if (count > 0) {
-		ASSERT(type == XFS_BMAP_UNMAP);
+		ASSERT(bui_type == XFS_BMAP_UNMAP);
 		irec.br_startblock = bmap->me_startblock;
 		irec.br_blockcount = count;
 		irec.br_startoff = bmap->me_startoff;

Re: [PATCH 1/3] xfs: clean up bmap intent item recovery checking

[Darrick J. Wong]

On Wed, Sep 23, 2020 at 08:16:14AM +0100, Christoph Hellwig wrote:
> On Wed, Sep 16, 2020 at 08:29:30PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > The bmap intent item checking code in xfs_bui_item_recover is spread all
> > over the function.  We should check the recovered log item at the top
> > before we allocate any resources or do anything else, so do that.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> >  fs/xfs/xfs_bmap_item.c |   57 ++++++++++++++++--------------------------------
> >  1 file changed, 19 insertions(+), 38 deletions(-)
> > 
> > 
> > diff --git a/fs/xfs/xfs_bmap_item.c b/fs/xfs/xfs_bmap_item.c
> > index 0a0904a7650a..877afe76d76a 100644
> > --- a/fs/xfs/xfs_bmap_item.c
> > +++ b/fs/xfs/xfs_bmap_item.c
> > @@ -437,17 +437,13 @@ xfs_bui_item_recover(
> >  	xfs_fsblock_t			inode_fsb;
> >  	xfs_filblks_t			count;
> >  	xfs_exntst_t			state;
> > -	enum xfs_bmap_intent_type	type;
> > -	bool				op_ok;
> >  	unsigned int			bui_type;
> >  	int				whichfork;
> >  	int				error = 0;
> >  
> >  	/* Only one mapping operation per BUI... */
> > -	if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS) {
> > -		xfs_bui_release(buip);
> > -		return -EFSCORRUPTED;
> > -	}
> > +	if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS)
> > +		goto garbage;
> 
> We don't really need the xfs_bui_release any more, and can stick to
> plain "return -EFSCORRUPTED" instead of the goto, but I suspect the
> previous patch has taken care of that and you've rebased already?

Yep.

--D

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

Re: [PATCH 1/3] xfs: clean up bmap intent item recovery checking

[Christoph Hellwig]

On Wed, Sep 16, 2020 at 08:29:30PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> The bmap intent item checking code in xfs_bui_item_recover is spread all
> over the function.  We should check the recovered log item at the top
> before we allocate any resources or do anything else, so do that.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  fs/xfs/xfs_bmap_item.c |   57 ++++++++++++++++--------------------------------
>  1 file changed, 19 insertions(+), 38 deletions(-)
> 
> 
> diff --git a/fs/xfs/xfs_bmap_item.c b/fs/xfs/xfs_bmap_item.c
> index 0a0904a7650a..877afe76d76a 100644
> --- a/fs/xfs/xfs_bmap_item.c
> +++ b/fs/xfs/xfs_bmap_item.c
> @@ -437,17 +437,13 @@ xfs_bui_item_recover(
>  	xfs_fsblock_t			inode_fsb;
>  	xfs_filblks_t			count;
>  	xfs_exntst_t			state;
> -	enum xfs_bmap_intent_type	type;
> -	bool				op_ok;
>  	unsigned int			bui_type;
>  	int				whichfork;
>  	int				error = 0;
>  
>  	/* Only one mapping operation per BUI... */
> -	if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS) {
> -		xfs_bui_release(buip);
> -		return -EFSCORRUPTED;
> -	}
> +	if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS)
> +		goto garbage;

We don't really need the xfs_bui_release any more, and can stick to
plain "return -EFSCORRUPTED" instead of the goto, but I suspect the
previous patch has taken care of that and you've rebased already?

Otherwise looks good:

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

Re: [PATCH 1/3] xfs: clean up bmap intent item recovery checking

[Dave Chinner]

On Wed, Sep 16, 2020 at 08:29:30PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> The bmap intent item checking code in xfs_bui_item_recover is spread all
> over the function.  We should check the recovered log item at the top
> before we allocate any resources or do anything else, so do that.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  fs/xfs/xfs_bmap_item.c |   57 ++++++++++++++++--------------------------------
>  1 file changed, 19 insertions(+), 38 deletions(-)

Looks good - I have a very similar patch here....

Reviewed-by: Dave Chinner <dchinner@redhat.com>
-- 
Dave Chinner
david@fromorbit.com

[PATCH 1/3] xfs: clean up bmap intent item recovery checking

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

The bmap intent item checking code in xfs_bui_item_recover is spread all
over the function.  We should check the recovered log item at the top
before we allocate any resources or do anything else, so do that.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/xfs_bmap_item.c |   57 ++++++++++++++++--------------------------------
 1 file changed, 19 insertions(+), 38 deletions(-)


diff --git a/fs/xfs/xfs_bmap_item.c b/fs/xfs/xfs_bmap_item.c
index 0a0904a7650a..877afe76d76a 100644
--- a/fs/xfs/xfs_bmap_item.c
+++ b/fs/xfs/xfs_bmap_item.c
@@ -437,17 +437,13 @@ xfs_bui_item_recover(
 	xfs_fsblock_t			inode_fsb;
 	xfs_filblks_t			count;
 	xfs_exntst_t			state;
-	enum xfs_bmap_intent_type	type;
-	bool				op_ok;
 	unsigned int			bui_type;
 	int				whichfork;
 	int				error = 0;
 
 	/* Only one mapping operation per BUI... */
-	if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS) {
-		xfs_bui_release(buip);
-		return -EFSCORRUPTED;
-	}
+	if (buip->bui_format.bui_nextents != XFS_BUI_MAX_FAST_EXTENTS)
+		goto garbage;
 
 	/*
 	 * First check the validity of the extent described by the
@@ -458,29 +454,26 @@ xfs_bui_item_recover(
 			   XFS_FSB_TO_DADDR(mp, bmap->me_startblock));
 	inode_fsb = XFS_BB_TO_FSB(mp, XFS_FSB_TO_DADDR(mp,
 			XFS_INO_TO_FSB(mp, bmap->me_owner)));
-	switch (bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK) {
+	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
+			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
+	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
+			XFS_ATTR_FORK : XFS_DATA_FORK;
+	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
+	switch (bui_type) {
 	case XFS_BMAP_MAP:
 	case XFS_BMAP_UNMAP:
-		op_ok = true;
 		break;
 	default:
-		op_ok = false;
-		break;
+		goto garbage;
 	}
-	if (!op_ok || startblock_fsb == 0 ||
+	if (startblock_fsb == 0 ||
 	    bmap->me_len == 0 ||
 	    inode_fsb == 0 ||
 	    startblock_fsb >= mp->m_sb.sb_dblocks ||
 	    bmap->me_len >= mp->m_sb.sb_agblocks ||
 	    inode_fsb >= mp->m_sb.sb_dblocks ||
-	    (bmap->me_flags & ~XFS_BMAP_EXTENT_FLAGS)) {
-		/*
-		 * This will pull the BUI from the AIL and
-		 * free the memory associated with it.
-		 */
-		xfs_bui_release(buip);
-		return -EFSCORRUPTED;
-	}
+	    (bmap->me_flags & ~XFS_BMAP_EXTENT_FLAGS))
+		goto garbage;
 
 	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate,
 			XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK), 0, 0, &tp);
@@ -501,32 +494,17 @@ xfs_bui_item_recover(
 	if (VFS_I(ip)->i_nlink == 0)
 		xfs_iflags_set(ip, XFS_IRECOVERY);
 
-	/* Process deferred bmap item. */
-	state = (bmap->me_flags & XFS_BMAP_EXTENT_UNWRITTEN) ?
-			XFS_EXT_UNWRITTEN : XFS_EXT_NORM;
-	whichfork = (bmap->me_flags & XFS_BMAP_EXTENT_ATTR_FORK) ?
-			XFS_ATTR_FORK : XFS_DATA_FORK;
-	bui_type = bmap->me_flags & XFS_BMAP_EXTENT_TYPE_MASK;
-	switch (bui_type) {
-	case XFS_BMAP_MAP:
-	case XFS_BMAP_UNMAP:
-		type = bui_type;
-		break;
-	default:
-		XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
-		error = -EFSCORRUPTED;
-		goto err_inode;
-	}
 	xfs_trans_ijoin(tp, ip, 0);
 
 	count = bmap->me_len;
-	error = xfs_trans_log_finish_bmap_update(tp, budp, type, ip, whichfork,
-			bmap->me_startoff, bmap->me_startblock, &count, state);
+	error = xfs_trans_log_finish_bmap_update(tp, budp, bui_type, ip,
+			whichfork, bmap->me_startoff, bmap->me_startblock,
+			&count, state);
 	if (error)
 		goto err_inode;
 
 	if (count > 0) {
-		ASSERT(type == XFS_BMAP_UNMAP);
+		ASSERT(bui_type == XFS_BMAP_UNMAP);
 		irec.br_startblock = bmap->me_startblock;
 		irec.br_blockcount = count;
 		irec.br_startoff = bmap->me_startoff;
@@ -546,6 +524,9 @@ xfs_bui_item_recover(
 		xfs_irele(ip);
 	}
 	return error;
+garbage:
+	xfs_bui_release(buip);
+	return -EFSCORRUPTED;
 }
 
 STATIC bool