Merge tag 'xfs-for-linus-v3.12-rc1-2' of git://oss.sgi.com/xfs/xfs

Pull xfs update #2 from Ben Myers:
"Here we have defrag support for v5 superblock, a number of bugfixes
and a cleanup or two.

- defrag support for CRC filesystems
- fix endian worning in xlog_recover_get_buf_lsn
- fixes for sparse warnings
- fix for assert in xfs_dir3_leaf_hdr_from_disk
- fix for log recovery of remote symlinks
- fix for log recovery of btree root splits
- fixes formemory allocation failures with ACLs
- fix for assert in xfs_buf_item_relse
- fix for assert in xfs_inode_buf_verify
- fix an assignment in an assert that should be a test in
xfs_bmbt_change_owner
- remove dead code in xlog_recover_inode_pass2"

* tag 'xfs-for-linus-v3.12-rc1-2' of git://oss.sgi.com/xfs/xfs:
xfs: remove dead code from xlog_recover_inode_pass2
xfs: = vs == typo in ASSERT()
xfs: don't assert fail on bad inode numbers
xfs: aborted buf items can be in the AIL.
xfs: factor all the kmalloc-or-vmalloc fallback allocations
xfs: fix memory allocation failures with ACLs
xfs: ensure we copy buffer type in da btree root splits
xfs: set remote symlink buffer type for recovery
xfs: recovery of swap extents operations for CRC filesystems
xfs: swap extents operations for CRC filesystems
xfs: check magic numbers in dir3 leaf verifier first
xfs: fix some minor sparse warnings
xfs: fix endian warning in xlog_recover_get_buf_lsn()

fs/xfs/kmem.c

···
 
 /*
  * Greedy allocation.  May fail and may return vmalloced memory.
- *
- * Must be freed using kmem_free_large.
  */
 void *
 kmem_zalloc_greedy(size_t *size, size_t minsize, size_t maxsize)
···
 	void		*ptr;
 	size_t		kmsize = maxsize;
 
-	while (!(ptr = kmem_zalloc_large(kmsize))) {
+	while (!(ptr = vzalloc(kmsize))) {
 		if ((kmsize >>= 1) <= minsize)
 			kmsize = minsize;
 	}
···
 	if (ptr)
 		memset((char *)ptr, 0, (int)size);
 	return ptr;
+}
+
+void *
+kmem_zalloc_large(size_t size, xfs_km_flags_t flags)
+{
+	void	*ptr;
+
+	ptr = kmem_zalloc(size, flags | KM_MAYFAIL);
+	if (ptr)
+		return ptr;
+	return vzalloc(size);
 }
 
 void

fs/xfs/kmem.h

···
 
 extern void *kmem_alloc(size_t, xfs_km_flags_t);
 extern void *kmem_zalloc(size_t, xfs_km_flags_t);
+extern void *kmem_zalloc_large(size_t size, xfs_km_flags_t);
 extern void *kmem_realloc(const void *, size_t, size_t, xfs_km_flags_t);
 extern void  kmem_free(const void *);
 
-static inline void *kmem_zalloc_large(size_t size)
-{
-	return vzalloc(size);
-}
-static inline void kmem_free_large(void *ptr)
-{
-	vfree(ptr);
-}
 
 extern void *kmem_zalloc_greedy(size_t *, size_t, size_t);
 

fs/xfs/xfs_acl.c

···
 	 * go out to the disk.
 	 */
 	len = XFS_ACL_MAX_SIZE(ip->i_mount);
-	xfs_acl = kzalloc(len, GFP_KERNEL);
+	xfs_acl = kmem_zalloc_large(len, KM_SLEEP);
 	if (!xfs_acl)
 		return ERR_PTR(-ENOMEM);
 
···
 	if (IS_ERR(acl))
 		goto out;
 
- out_update_cache:
+out_update_cache:
 	set_cached_acl(inode, type, acl);
- out:
-	kfree(xfs_acl);
+out:
+	kmem_free(xfs_acl);
 	return acl;
 }
 
···
 		struct xfs_acl *xfs_acl;
 		int len = XFS_ACL_MAX_SIZE(ip->i_mount);
 
-		xfs_acl = kzalloc(len, GFP_KERNEL);
+		xfs_acl = kmem_zalloc_large(len, KM_SLEEP);
 		if (!xfs_acl)
 			return -ENOMEM;
 
···
 		error = -xfs_attr_set(ip, ea_name, (unsigned char *)xfs_acl,
 				len, ATTR_ROOT);
 
-		kfree(xfs_acl);
+		kmem_free(xfs_acl);
 	} else {
 		/*
 		 * A NULL ACL argument means we want to remove the ACL.

fs/xfs/xfs_bmap.c

···
 {
 	struct xfs_mount	*mp = ip->i_mount;
 	struct xfs_ifork	*ifp;
-	struct xfs_bmalloca	bma = { 0 };	/* args for xfs_bmap_alloc */
+	struct xfs_bmalloca	bma = { NULL };	/* args for xfs_bmap_alloc */
 	xfs_fileoff_t		end;		/* end of mapped file region */
 	int			eof;		/* after the end of extents */
 	int			error;		/* error return */

fs/xfs/xfs_bmap_btree.c

···
 		return blocklen / sizeof(xfs_bmdr_rec_t);
 	return blocklen / (sizeof(xfs_bmdr_key_t) + sizeof(xfs_bmdr_ptr_t));
 }
+
+/*
+ * Change the owner of a btree format fork fo the inode passed in. Change it to
+ * the owner of that is passed in so that we can change owners before or after
+ * we switch forks between inodes. The operation that the caller is doing will
+ * determine whether is needs to change owner before or after the switch.
+ *
+ * For demand paged transactional modification, the fork switch should be done
+ * after reading in all the blocks, modifying them and pinning them in the
+ * transaction. For modification when the buffers are already pinned in memory,
+ * the fork switch can be done before changing the owner as we won't need to
+ * validate the owner until the btree buffers are unpinned and writes can occur
+ * again.
+ *
+ * For recovery based ownership change, there is no transactional context and
+ * so a buffer list must be supplied so that we can record the buffers that we
+ * modified for the caller to issue IO on.
+ */
+int
+xfs_bmbt_change_owner(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	int			whichfork,
+	xfs_ino_t		new_owner,
+	struct list_head	*buffer_list)
+{
+	struct xfs_btree_cur	*cur;
+	int			error;
+
+	ASSERT(tp || buffer_list);
+	ASSERT(!(tp && buffer_list));
+	if (whichfork == XFS_DATA_FORK)
+		ASSERT(ip->i_d.di_format == XFS_DINODE_FMT_BTREE);
+	else
+		ASSERT(ip->i_d.di_aformat == XFS_DINODE_FMT_BTREE);
+
+	cur = xfs_bmbt_init_cursor(ip->i_mount, tp, ip, whichfork);
+	if (!cur)
+		return ENOMEM;
+
+	error = xfs_btree_change_owner(cur, new_owner, buffer_list);
+	xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
+	return error;
+}

fs/xfs/xfs_bmap_btree.h

···
 extern int xfs_bmdr_maxrecs(struct xfs_mount *, int blocklen, int leaf);
 extern int xfs_bmbt_maxrecs(struct xfs_mount *, int blocklen, int leaf);
 
+extern int xfs_bmbt_change_owner(struct xfs_trans *tp, struct xfs_inode *ip,
+				 int whichfork, xfs_ino_t new_owner,
+				 struct list_head *buffer_list);
+
 extern struct xfs_btree_cur *xfs_bmbt_init_cursor(struct xfs_mount *,
 		struct xfs_trans *, struct xfs_inode *, int);
 

fs/xfs/xfs_bmap_util.c

···
 
 	if (bmv->bmv_count > ULONG_MAX / sizeof(struct getbmapx))
 		return XFS_ERROR(ENOMEM);
-	out = kmem_zalloc(bmv->bmv_count * sizeof(struct getbmapx), KM_MAYFAIL);
-	if (!out) {
-		out = kmem_zalloc_large(bmv->bmv_count *
-					sizeof(struct getbmapx));
-		if (!out)
-			return XFS_ERROR(ENOMEM);
-	}
+	out = kmem_zalloc_large(bmv->bmv_count * sizeof(struct getbmapx), 0);
+	if (!out)
+		return XFS_ERROR(ENOMEM);
 
 	xfs_ilock(ip, XFS_IOLOCK_SHARED);
 	if (whichfork == XFS_DATA_FORK && !(iflags & BMV_IF_DELALLOC)) {
···
 			break;
 	}
 
-	if (is_vmalloc_addr(out))
-		kmem_free_large(out);
-	else
-		kmem_free(out);
+	kmem_free(out);
 	return error;
 }
 
···
 	int		taforkblks = 0;
 	__uint64_t	tmp;
 
-	/*
-	 * We have no way of updating owner information in the BMBT blocks for
-	 * each inode on CRC enabled filesystems, so to avoid corrupting the
-	 * this metadata we simply don't allow extent swaps to occur.
-	 */
-	if (xfs_sb_version_hascrc(&mp->m_sb))
-		return XFS_ERROR(EINVAL);
-
 	tempifp = kmem_alloc(sizeof(xfs_ifork_t), KM_MAYFAIL);
 	if (!tempifp) {
 		error = XFS_ERROR(ENOMEM);
···
 			goto out_trans_cancel;
 	}
 
+	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+	xfs_trans_ijoin(tp, tip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
+
+	/*
+	 * Before we've swapped the forks, lets set the owners of the forks
+	 * appropriately. We have to do this as we are demand paging the btree
+	 * buffers, and so the validation done on read will expect the owner
+	 * field to be correctly set. Once we change the owners, we can swap the
+	 * inode forks.
+	 *
+	 * Note the trickiness in setting the log flags - we set the owner log
+	 * flag on the opposite inode (i.e. the inode we are setting the new
+	 * owner to be) because once we swap the forks and log that, log
+	 * recovery is going to see the fork as owned by the swapped inode,
+	 * not the pre-swapped inodes.
+	 */
+	src_log_flags = XFS_ILOG_CORE;
+	target_log_flags = XFS_ILOG_CORE;
+	if (ip->i_d.di_version == 3 &&
+	    ip->i_d.di_format == XFS_DINODE_FMT_BTREE) {
+		target_log_flags |= XFS_ILOG_DOWNER;
+		error = xfs_bmbt_change_owner(tp, ip, XFS_DATA_FORK,
+					      tip->i_ino, NULL);
+		if (error)
+			goto out_trans_cancel;
+	}
+
+	if (tip->i_d.di_version == 3 &&
+	    tip->i_d.di_format == XFS_DINODE_FMT_BTREE) {
+		src_log_flags |= XFS_ILOG_DOWNER;
+		error = xfs_bmbt_change_owner(tp, tip, XFS_DATA_FORK,
+					      ip->i_ino, NULL);
+		if (error)
+			goto out_trans_cancel;
+	}
+
 	/*
 	 * Swap the data forks of the inodes
 	 */
···
 	tip->i_delayed_blks = ip->i_delayed_blks;
 	ip->i_delayed_blks = 0;
 
-	src_log_flags = XFS_ILOG_CORE;
 	switch (ip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
 		/* If the extents fit in the inode, fix the
···
 		src_log_flags |= XFS_ILOG_DEXT;
 		break;
 	case XFS_DINODE_FMT_BTREE:
+		ASSERT(ip->i_d.di_version < 3 ||
+		       (src_log_flags & XFS_ILOG_DOWNER));
 		src_log_flags |= XFS_ILOG_DBROOT;
 		break;
 	}
 
-	target_log_flags = XFS_ILOG_CORE;
 	switch (tip->i_d.di_format) {
 	case XFS_DINODE_FMT_EXTENTS:
 		/* If the extents fit in the inode, fix the
···
 		break;
 	case XFS_DINODE_FMT_BTREE:
 		target_log_flags |= XFS_ILOG_DBROOT;
+		ASSERT(tip->i_d.di_version < 3 ||
+		       (target_log_flags & XFS_ILOG_DOWNER));
 		break;
 	}
-
-
-	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
-	xfs_trans_ijoin(tp, tip, XFS_ILOCK_EXCL | XFS_IOLOCK_EXCL);
 
 	xfs_trans_log_inode(tp, ip,  src_log_flags);
 	xfs_trans_log_inode(tp, tip, target_log_flags);

fs/xfs/xfs_btree.c

···
 	return xfs_btree_readahead_sblock(cur, lr, block);
 }
 
+STATIC xfs_daddr_t
+xfs_btree_ptr_to_daddr(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr)
+{
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+		ASSERT(ptr->l != cpu_to_be64(NULLDFSBNO));
+
+		return XFS_FSB_TO_DADDR(cur->bc_mp, be64_to_cpu(ptr->l));
+	} else {
+		ASSERT(cur->bc_private.a.agno != NULLAGNUMBER);
+		ASSERT(ptr->s != cpu_to_be32(NULLAGBLOCK));
+
+		return XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_private.a.agno,
+					be32_to_cpu(ptr->s));
+	}
+}
+
+/*
+ * Readahead @count btree blocks at the given @ptr location.
+ *
+ * We don't need to care about long or short form btrees here as we have a
+ * method of converting the ptr directly to a daddr available to us.
+ */
+STATIC void
+xfs_btree_readahead_ptr(
+	struct xfs_btree_cur	*cur,
+	union xfs_btree_ptr	*ptr,
+	xfs_extlen_t		count)
+{
+	xfs_buf_readahead(cur->bc_mp->m_ddev_targp,
+			  xfs_btree_ptr_to_daddr(cur, ptr),
+			  cur->bc_mp->m_bsize * count, cur->bc_ops->buf_ops);
+}
+
 /*
  * Set the buffer for level "lev" in the cursor to bp, releasing
  * any previous buffer.
···
 	else {
 		ptr->s = cpu_to_be32(xfs_daddr_to_agbno(cur->bc_mp,
 					XFS_BUF_ADDR(bp)));
-	}
-}
-
-STATIC xfs_daddr_t
-xfs_btree_ptr_to_daddr(
-	struct xfs_btree_cur	*cur,
-	union xfs_btree_ptr	*ptr)
-{
-	if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
-		ASSERT(ptr->l != cpu_to_be64(NULLDFSBNO));
-
-		return XFS_FSB_TO_DADDR(cur->bc_mp, be64_to_cpu(ptr->l));
-	} else {
-		ASSERT(cur->bc_private.a.agno != NULLAGNUMBER);
-		ASSERT(ptr->s != cpu_to_be32(NULLAGBLOCK));
-
-		return XFS_AGB_TO_DADDR(cur->bc_mp, cur->bc_private.a.agno,
-					be32_to_cpu(ptr->s));
 	}
 }
 
···
 	 */
 	*recp = xfs_btree_rec_addr(cur, ptr, block);
 	*stat = 1;
+	return 0;
+}
+
+/*
+ * Change the owner of a btree.
+ *
+ * The mechanism we use here is ordered buffer logging. Because we don't know
+ * how many buffers were are going to need to modify, we don't really want to
+ * have to make transaction reservations for the worst case of every buffer in a
+ * full size btree as that may be more space that we can fit in the log....
+ *
+ * We do the btree walk in the most optimal manner possible - we have sibling
+ * pointers so we can just walk all the blocks on each level from left to right
+ * in a single pass, and then move to the next level and do the same. We can
+ * also do readahead on the sibling pointers to get IO moving more quickly,
+ * though for slow disks this is unlikely to make much difference to performance
+ * as the amount of CPU work we have to do before moving to the next block is
+ * relatively small.
+ *
+ * For each btree block that we load, modify the owner appropriately, set the
+ * buffer as an ordered buffer and log it appropriately. We need to ensure that
+ * we mark the region we change dirty so that if the buffer is relogged in
+ * a subsequent transaction the changes we make here as an ordered buffer are
+ * correctly relogged in that transaction.  If we are in recovery context, then
+ * just queue the modified buffer as delayed write buffer so the transaction
+ * recovery completion writes the changes to disk.
+ */
+static int
+xfs_btree_block_change_owner(
+	struct xfs_btree_cur	*cur,
+	int			level,
+	__uint64_t		new_owner,
+	struct list_head	*buffer_list)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+	union xfs_btree_ptr     rptr;
+
+	/* do right sibling readahead */
+	xfs_btree_readahead(cur, level, XFS_BTCUR_RIGHTRA);
+
+	/* modify the owner */
+	block = xfs_btree_get_block(cur, level, &bp);
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		block->bb_u.l.bb_owner = cpu_to_be64(new_owner);
+	else
+		block->bb_u.s.bb_owner = cpu_to_be32(new_owner);
+
+	/*
+	 * If the block is a root block hosted in an inode, we might not have a
+	 * buffer pointer here and we shouldn't attempt to log the change as the
+	 * information is already held in the inode and discarded when the root
+	 * block is formatted into the on-disk inode fork. We still change it,
+	 * though, so everything is consistent in memory.
+	 */
+	if (bp) {
+		if (cur->bc_tp) {
+			xfs_trans_ordered_buf(cur->bc_tp, bp);
+			xfs_btree_log_block(cur, bp, XFS_BB_OWNER);
+		} else {
+			xfs_buf_delwri_queue(bp, buffer_list);
+		}
+	} else {
+		ASSERT(cur->bc_flags & XFS_BTREE_ROOT_IN_INODE);
+		ASSERT(level == cur->bc_nlevels - 1);
+	}
+
+	/* now read rh sibling block for next iteration */
+	xfs_btree_get_sibling(cur, block, &rptr, XFS_BB_RIGHTSIB);
+	if (xfs_btree_ptr_is_null(cur, &rptr))
+		return ENOENT;
+
+	return xfs_btree_lookup_get_block(cur, level, &rptr, &block);
+}
+
+int
+xfs_btree_change_owner(
+	struct xfs_btree_cur	*cur,
+	__uint64_t		new_owner,
+	struct list_head	*buffer_list)
+{
+	union xfs_btree_ptr     lptr;
+	int			level;
+	struct xfs_btree_block	*block = NULL;
+	int			error = 0;
+
+	cur->bc_ops->init_ptr_from_cur(cur, &lptr);
+
+	/* for each level */
+	for (level = cur->bc_nlevels - 1; level >= 0; level--) {
+		/* grab the left hand block */
+		error = xfs_btree_lookup_get_block(cur, level, &lptr, &block);
+		if (error)
+			return error;
+
+		/* readahead the left most block for the next level down */
+		if (level > 0) {
+			union xfs_btree_ptr     *ptr;
+
+			ptr = xfs_btree_ptr_addr(cur, 1, block);
+			xfs_btree_readahead_ptr(cur, ptr, 1);
+
+			/* save for the next iteration of the loop */
+			lptr = *ptr;
+		}
+
+		/* for each buffer in the level */
+		do {
+			error = xfs_btree_block_change_owner(cur, level,
+							     new_owner,
+							     buffer_list);
+		} while (!error);
+
+		if (error != ENOENT)
+			return error;
+	}
+
 	return 0;
 }

fs/xfs/xfs_btree.h

···
 /*
  * For logging record fields.
  */
-#define	XFS_BB_MAGIC		0x01
-#define	XFS_BB_LEVEL		0x02
-#define	XFS_BB_NUMRECS		0x04
-#define	XFS_BB_LEFTSIB		0x08
-#define	XFS_BB_RIGHTSIB		0x10
-#define	XFS_BB_BLKNO		0x20
+#define	XFS_BB_MAGIC		(1 << 0)
+#define	XFS_BB_LEVEL		(1 << 1)
+#define	XFS_BB_NUMRECS		(1 << 2)
+#define	XFS_BB_LEFTSIB		(1 << 3)
+#define	XFS_BB_RIGHTSIB		(1 << 4)
+#define	XFS_BB_BLKNO		(1 << 5)
+#define	XFS_BB_LSN		(1 << 6)
+#define	XFS_BB_UUID		(1 << 7)
+#define	XFS_BB_OWNER		(1 << 8)
 #define	XFS_BB_NUM_BITS		5
 #define	XFS_BB_ALL_BITS		((1 << XFS_BB_NUM_BITS) - 1)
-#define	XFS_BB_NUM_BITS_CRC	8
+#define	XFS_BB_NUM_BITS_CRC	9
 #define	XFS_BB_ALL_BITS_CRC	((1 << XFS_BB_NUM_BITS_CRC) - 1)
 
 /*
···
 int xfs_btree_insert(struct xfs_btree_cur *, int *);
 int xfs_btree_delete(struct xfs_btree_cur *, int *);
 int xfs_btree_get_rec(struct xfs_btree_cur *, union xfs_btree_rec **, int *);
+int xfs_btree_change_owner(struct xfs_btree_cur *cur, __uint64_t new_owner,
+			   struct list_head *buffer_list);
 
 /*
  * btree block CRC helpers

fs/xfs/xfs_buf_item.c

···
 			}
 		}
 	}
-	if (clean || aborted) {
-		if (atomic_dec_and_test(&bip->bli_refcount)) {
-			ASSERT(!aborted || XFS_FORCED_SHUTDOWN(lip->li_mountp));
+
+	/*
+	 * Clean buffers, by definition, cannot be in the AIL. However, aborted
+	 * buffers may be dirty and hence in the AIL. Therefore if we are
+	 * aborting a buffer and we've just taken the last refernce away, we
+	 * have to check if it is in the AIL before freeing it. We need to free
+	 * it in this case, because an aborted transaction has already shut the
+	 * filesystem down and this is the last chance we will have to do so.
+	 */
+	if (atomic_dec_and_test(&bip->bli_refcount)) {
+		if (clean)
+			xfs_buf_item_relse(bp);
+		else if (aborted) {
+			ASSERT(XFS_FORCED_SHUTDOWN(lip->li_mountp));
+			if (lip->li_flags & XFS_LI_IN_AIL) {
+				xfs_trans_ail_delete(lip->li_ailp, lip,
+						     SHUTDOWN_LOG_IO_ERROR);
+			}
 			xfs_buf_item_relse(bp);
 		}
-	} else
-		atomic_dec(&bip->bli_refcount);
+	}
 
 	if (!(flags & XFS_BLI_HOLD))
 		xfs_buf_relse(bp);

fs/xfs/xfs_da_btree.c

···
 	xfs_trans_log_buf(tp, bp, 0, size - 1);
 
 	bp->b_ops = blk1->bp->b_ops;
+	xfs_trans_buf_copy_type(bp, blk1->bp);
 	blk1->bp = bp;
 	blk1->blkno = blkno;
 

fs/xfs/xfs_dir2_leaf.c

···
 	return true;
 }
 
+/*
+ * We verify the magic numbers before decoding the leaf header so that on debug
+ * kernels we don't get assertion failures in xfs_dir3_leaf_hdr_from_disk() due
+ * to incorrect magic numbers.
+ */
 static bool
 xfs_dir3_leaf_verify(
 	struct xfs_buf		*bp,
···
 
 	ASSERT(magic == XFS_DIR2_LEAF1_MAGIC || magic == XFS_DIR2_LEAFN_MAGIC);
 
-	xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
 	if (xfs_sb_version_hascrc(&mp->m_sb)) {
 		struct xfs_dir3_leaf_hdr *leaf3 = bp->b_addr;
+		__uint16_t		magic3;
 
-		if ((magic == XFS_DIR2_LEAF1_MAGIC &&
-		     leafhdr.magic != XFS_DIR3_LEAF1_MAGIC) ||
-		    (magic == XFS_DIR2_LEAFN_MAGIC &&
-		     leafhdr.magic != XFS_DIR3_LEAFN_MAGIC))
+		magic3 = (magic == XFS_DIR2_LEAF1_MAGIC) ? XFS_DIR3_LEAF1_MAGIC
+							 : XFS_DIR3_LEAFN_MAGIC;
+
+		if (leaf3->info.hdr.magic != cpu_to_be16(magic3))
 			return false;
-
 		if (!uuid_equal(&leaf3->info.uuid, &mp->m_sb.sb_uuid))
 			return false;
 		if (be64_to_cpu(leaf3->info.blkno) != bp->b_bn)
 			return false;
 	} else {
-		if (leafhdr.magic != magic)
+		if (leaf->hdr.info.magic != cpu_to_be16(magic))
 			return false;
 	}
+
+	xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
 	return xfs_dir3_leaf_check_int(mp, &leafhdr, leaf);
 }
 

fs/xfs/xfs_dquot_item.c

···
 STATIC uint
 xfs_qm_dquot_logitem_push(
 	struct xfs_log_item	*lip,
-	struct list_head	*buffer_list)
+	struct list_head	*buffer_list) __releases(&lip->li_ailp->xa_lock)
+					      __acquires(&lip->li_ailp->xa_lock)
 {
 	struct xfs_dquot	*dqp = DQUOT_ITEM(lip)->qli_dquot;
 	struct xfs_buf		*bp = NULL;

fs/xfs/xfs_extent_busy.c

···
 	struct xfs_extent_busy	*busyp,
 	xfs_agblock_t		fbno,
 	xfs_extlen_t		flen,
-	bool			userdata)
+	bool			userdata) __releases(&pag->pagb_lock)
+					  __acquires(&pag->pagb_lock)
 {
 	xfs_agblock_t		fend = fbno + flen;
 	xfs_agblock_t		bbno = busyp->bno;

fs/xfs/xfs_icache.c

···
 /*
  * Allocate and initialise an xfs_inode.
  */
-STATIC struct xfs_inode *
+struct xfs_inode *
 xfs_inode_alloc(
 	struct xfs_mount	*mp,
 	xfs_ino_t		ino)
···
 	kmem_zone_free(xfs_inode_zone, ip);
 }
 
-STATIC void
+void
 xfs_inode_free(
 	struct xfs_inode	*ip)
 {

fs/xfs/xfs_icache.h

···
 int xfs_iget(struct xfs_mount *mp, struct xfs_trans *tp, xfs_ino_t ino,
 	     uint flags, uint lock_flags, xfs_inode_t **ipp);
 
+/* recovery needs direct inode allocation capability */
+struct xfs_inode * xfs_inode_alloc(struct xfs_mount *mp, xfs_ino_t ino);
+void xfs_inode_free(struct xfs_inode *ip);
+
 void xfs_reclaim_worker(struct work_struct *work);
 
 int xfs_reclaim_inodes(struct xfs_mount *mp, int mode);

fs/xfs/xfs_inode_buf.c

···
 					i * mp->m_sb.sb_inodesize);
 		if (!dip->di_next_unlinked)  {
 			xfs_alert(mp,
-	"Detected bogus zero next_unlinked field in incore inode buffer 0x%p.",
-				bp);
-			ASSERT(dip->di_next_unlinked);
+	"Detected bogus zero next_unlinked field in inode %d buffer 0x%llx.",
+				i, (long long)bp->b_bn);
 		}
 	}
 }
···
 			XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_HIGH,
 					     mp, dip);
 #ifdef DEBUG
-			xfs_emerg(mp,
+			xfs_alert(mp,
 				"bad inode magic/vsn daddr %lld #%d (magic=%x)",
 				(unsigned long long)bp->b_bn, i,
 				be16_to_cpu(dip->di_magic));
-			ASSERT(0);
 #endif
 		}
 	}
···
 	return 0;
 }
 
-STATIC void
+void
 xfs_dinode_from_disk(
 	xfs_icdinode_t		*to,
 	xfs_dinode_t		*from)

fs/xfs/xfs_inode_buf.h

···
 	ushort		im_boffset;	/* inode offset in block in bytes */
 };
 
-int		xfs_imap_to_bp(struct xfs_mount *, struct xfs_trans *,
-			       struct xfs_imap *, struct xfs_dinode **,
-			       struct xfs_buf **, uint, uint);
-int		xfs_iread(struct xfs_mount *, struct xfs_trans *,
-			  struct xfs_inode *, uint);
-void		xfs_dinode_calc_crc(struct xfs_mount *, struct xfs_dinode *);
-void		xfs_dinode_to_disk(struct xfs_dinode *,
-				   struct xfs_icdinode *);
+int	xfs_imap_to_bp(struct xfs_mount *, struct xfs_trans *,
+		       struct xfs_imap *, struct xfs_dinode **,
+		       struct xfs_buf **, uint, uint);
+int	xfs_iread(struct xfs_mount *, struct xfs_trans *,
+		  struct xfs_inode *, uint);
+void	xfs_dinode_calc_crc(struct xfs_mount *, struct xfs_dinode *);
+void	xfs_dinode_to_disk(struct xfs_dinode *to, struct xfs_icdinode *from);
+void	xfs_dinode_from_disk(struct xfs_icdinode *to, struct xfs_dinode *from);
 
 #if defined(DEBUG)
-void		xfs_inobp_check(struct xfs_mount *, struct xfs_buf *);
+void	xfs_inobp_check(struct xfs_mount *, struct xfs_buf *);
 #else
 #define	xfs_inobp_check(mp, bp)
 #endif /* DEBUG */

fs/xfs/xfs_ioctl.c

···
 	int			hsize;
 	xfs_handle_t		handle;
 	struct inode		*inode;
-	struct fd		f = {0};
+	struct fd		f = {NULL};
 	struct path		path;
 	int			error;
 	struct xfs_inode	*ip;
···
 	if (IS_ERR(dentry))
 		return PTR_ERR(dentry);
 
-	kbuf = kmem_zalloc(al_hreq.buflen, KM_SLEEP | KM_MAYFAIL);
-	if (!kbuf) {
-		kbuf = kmem_zalloc_large(al_hreq.buflen);
-		if (!kbuf)
-			goto out_dput;
-	}
+	kbuf = kmem_zalloc_large(al_hreq.buflen, KM_SLEEP);
+	if (!kbuf)
+		goto out_dput;
 
 	cursor = (attrlist_cursor_kern_t *)&al_hreq.pos;
 	error = -xfs_attr_list(XFS_I(dentry->d_inode), kbuf, al_hreq.buflen,
···
 	if (copy_to_user(al_hreq.buffer, kbuf, al_hreq.buflen))
 		error = -EFAULT;
 
- out_kfree:
-	if (is_vmalloc_addr(kbuf))
-		kmem_free_large(kbuf);
-	else
-		kmem_free(kbuf);
- out_dput:
+out_kfree:
+	kmem_free(kbuf);
+out_dput:
 	dput(dentry);
 	return error;
 }
···
 
 	if (*len > XATTR_SIZE_MAX)
 		return EINVAL;
-	kbuf = kmem_zalloc(*len, KM_SLEEP | KM_MAYFAIL);
-	if (!kbuf) {
-		kbuf = kmem_zalloc_large(*len);
-		if (!kbuf)
-			return ENOMEM;
-	}
+	kbuf = kmem_zalloc_large(*len, KM_SLEEP);
+	if (!kbuf)
+		return ENOMEM;
 
 	error = xfs_attr_get(XFS_I(inode), name, kbuf, (int *)len, flags);
 	if (error)
···
 	if (copy_to_user(ubuf, kbuf, *len))
 		error = EFAULT;
 
- out_kfree:
-	if (is_vmalloc_addr(kbuf))
-		kmem_free_large(kbuf);
-	else
-		kmem_free(kbuf);
+out_kfree:
+	kmem_free(kbuf);
 	return error;
 }
 

fs/xfs/xfs_ioctl32.c

···
 		return PTR_ERR(dentry);
 
 	error = -ENOMEM;
-	kbuf = kmem_zalloc(al_hreq.buflen, KM_SLEEP | KM_MAYFAIL);
-	if (!kbuf) {
-		kbuf = kmem_zalloc_large(al_hreq.buflen);
-		if (!kbuf)
-			goto out_dput;
-	}
+	kbuf = kmem_zalloc_large(al_hreq.buflen, KM_SLEEP);
+	if (!kbuf)
+		goto out_dput;
 
 	cursor = (attrlist_cursor_kern_t *)&al_hreq.pos;
 	error = -xfs_attr_list(XFS_I(dentry->d_inode), kbuf, al_hreq.buflen,
···
 	if (copy_to_user(compat_ptr(al_hreq.buffer), kbuf, al_hreq.buflen))
 		error = -EFAULT;
 
- out_kfree:
-	if (is_vmalloc_addr(kbuf))
-		kmem_free_large(kbuf);
-	else
-		kmem_free(kbuf);
- out_dput:
+out_kfree:
+	kmem_free(kbuf);
+out_dput:
 	dput(dentry);
 	return error;
 }

fs/xfs/xfs_itable.c

···
 	/*
 	 * Done, we're either out of filesystem or space to put the data.
 	 */
-	kmem_free_large(irbuf);
+	kmem_free(irbuf);
 	*ubcountp = ubelem;
 	/*
 	 * Found some inodes, return them now and return the error next time.
···
 	 * at the expense of the error case.
 	 */
 
-	ino = (xfs_ino_t)*lastinop;
-	error = xfs_bulkstat_one(mp, ino, buffer, sizeof(xfs_bstat_t), 0, &res);
+	ino = *lastinop;
+	error = xfs_bulkstat_one(mp, ino, buffer, sizeof(xfs_bstat_t),
+				 NULL, &res);
 	if (error) {
 		/*
 		 * Special case way failed, do it the "long" way

fs/xfs/xfs_log.c

···
 	struct xlog		*log,
 	struct xlog_grant_head	*head,
 	struct xlog_ticket	*tic,
-	int			need_bytes)
+	int			need_bytes) __releases(&head->lock)
+					    __acquires(&head->lock)
 {
 	list_add_tail(&tic->t_queue, &head->waiters);
 

fs/xfs/xfs_log_format.h

···
 #define	XFS_ILOG_ADATA	0x040	/* log i_af.if_data */
 #define	XFS_ILOG_AEXT	0x080	/* log i_af.if_extents */
 #define	XFS_ILOG_ABROOT	0x100	/* log i_af.i_broot */
+#define XFS_ILOG_DOWNER	0x200	/* change the data fork owner on replay */
+#define XFS_ILOG_AOWNER	0x400	/* change the attr fork owner on replay */
 
 
 /*
···
 #define	XFS_ILOG_NONCORE	(XFS_ILOG_DDATA | XFS_ILOG_DEXT | \
 				 XFS_ILOG_DBROOT | XFS_ILOG_DEV | \
 				 XFS_ILOG_UUID | XFS_ILOG_ADATA | \
-				 XFS_ILOG_AEXT | XFS_ILOG_ABROOT)
+				 XFS_ILOG_AEXT | XFS_ILOG_ABROOT | \
+				 XFS_ILOG_DOWNER | XFS_ILOG_AOWNER)
 
 #define	XFS_ILOG_DFORK		(XFS_ILOG_DDATA | XFS_ILOG_DEXT | \
 				 XFS_ILOG_DBROOT)
···
 				 XFS_ILOG_DEXT | XFS_ILOG_DBROOT | \
 				 XFS_ILOG_DEV | XFS_ILOG_UUID | \
 				 XFS_ILOG_ADATA | XFS_ILOG_AEXT | \
-				 XFS_ILOG_ABROOT | XFS_ILOG_TIMESTAMP)
+				 XFS_ILOG_ABROOT | XFS_ILOG_TIMESTAMP | \
+				 XFS_ILOG_DOWNER | XFS_ILOG_AOWNER)
 
 static inline int xfs_ilog_fbroot(int w)
 {

fs/xfs/xfs_log_recover.c

···
 	case XFS_ATTR3_RMT_MAGIC:
 		return be64_to_cpu(((struct xfs_attr3_rmt_hdr *)blk)->rm_lsn);
 	case XFS_SB_MAGIC:
-		return be64_to_cpu(((struct xfs_sb *)blk)->sb_lsn);
+		return be64_to_cpu(((struct xfs_dsb *)blk)->sb_lsn);
 	default:
 		break;
 	}
···
 	return error;
 }
 
+/*
+ * Inode fork owner changes
+ *
+ * If we have been told that we have to reparent the inode fork, it's because an
+ * extent swap operation on a CRC enabled filesystem has been done and we are
+ * replaying it. We need to walk the BMBT of the appropriate fork and change the
+ * owners of it.
+ *
+ * The complexity here is that we don't have an inode context to work with, so
+ * after we've replayed the inode we need to instantiate one.  This is where the
+ * fun begins.
+ *
+ * We are in the middle of log recovery, so we can't run transactions. That
+ * means we cannot use cache coherent inode instantiation via xfs_iget(), as
+ * that will result in the corresponding iput() running the inode through
+ * xfs_inactive(). If we've just replayed an inode core that changes the link
+ * count to zero (i.e. it's been unlinked), then xfs_inactive() will run
+ * transactions (bad!).
+ *
+ * So, to avoid this, we instantiate an inode directly from the inode core we've
+ * just recovered. We have the buffer still locked, and all we really need to
+ * instantiate is the inode core and the forks being modified. We can do this
+ * manually, then run the inode btree owner change, and then tear down the
+ * xfs_inode without having to run any transactions at all.
+ *
+ * Also, because we don't have a transaction context available here but need to
+ * gather all the buffers we modify for writeback so we pass the buffer_list
+ * instead for the operation to use.
+ */
+
+STATIC int
+xfs_recover_inode_owner_change(
+	struct xfs_mount	*mp,
+	struct xfs_dinode	*dip,
+	struct xfs_inode_log_format *in_f,
+	struct list_head	*buffer_list)
+{
+	struct xfs_inode	*ip;
+	int			error;
+
+	ASSERT(in_f->ilf_fields & (XFS_ILOG_DOWNER|XFS_ILOG_AOWNER));
+
+	ip = xfs_inode_alloc(mp, in_f->ilf_ino);
+	if (!ip)
+		return ENOMEM;
+
+	/* instantiate the inode */
+	xfs_dinode_from_disk(&ip->i_d, dip);
+	ASSERT(ip->i_d.di_version >= 3);
+
+	error = xfs_iformat_fork(ip, dip);
+	if (error)
+		goto out_free_ip;
+
+
+	if (in_f->ilf_fields & XFS_ILOG_DOWNER) {
+		ASSERT(in_f->ilf_fields & XFS_ILOG_DBROOT);
+		error = xfs_bmbt_change_owner(NULL, ip, XFS_DATA_FORK,
+					      ip->i_ino, buffer_list);
+		if (error)
+			goto out_free_ip;
+	}
+
+	if (in_f->ilf_fields & XFS_ILOG_AOWNER) {
+		ASSERT(in_f->ilf_fields & XFS_ILOG_ABROOT);
+		error = xfs_bmbt_change_owner(NULL, ip, XFS_ATTR_FORK,
+					      ip->i_ino, buffer_list);
+		if (error)
+			goto out_free_ip;
+	}
+
+out_free_ip:
+	xfs_inode_free(ip);
+	return error;
+}
+
 STATIC int
 xlog_recover_inode_pass2(
 	struct xlog			*log,
···
 	error = bp->b_error;
 	if (error) {
 		xfs_buf_ioerror_alert(bp, "xlog_recover_do..(read#2)");
-		xfs_buf_relse(bp);
-		goto error;
+		goto out_release;
 	}
 	ASSERT(in_f->ilf_fields & XFS_ILOG_CORE);
 	dip = (xfs_dinode_t *)xfs_buf_offset(bp, in_f->ilf_boffset);
···
 	 * like an inode!
 	 */
 	if (unlikely(dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC))) {
-		xfs_buf_relse(bp);
 		xfs_alert(mp,
 	"%s: Bad inode magic number, dip = 0x%p, dino bp = 0x%p, ino = %Ld",
 			__func__, dip, bp, in_f->ilf_ino);
 		XFS_ERROR_REPORT("xlog_recover_inode_pass2(1)",
 				 XFS_ERRLEVEL_LOW, mp);
 		error = EFSCORRUPTED;
-		goto error;
+		goto out_release;
 	}
 	dicp = item->ri_buf[1].i_addr;
 	if (unlikely(dicp->di_magic != XFS_DINODE_MAGIC)) {
-		xfs_buf_relse(bp);
 		xfs_alert(mp,
 			"%s: Bad inode log record, rec ptr 0x%p, ino %Ld",
 			__func__, item, in_f->ilf_ino);
 		XFS_ERROR_REPORT("xlog_recover_inode_pass2(2)",
 				 XFS_ERRLEVEL_LOW, mp);
 		error = EFSCORRUPTED;
-		goto error;
+		goto out_release;
 	}
 
 	/*
 	 * If the inode has an LSN in it, recover the inode only if it's less
-	 * than the lsn of the transaction we are replaying.
+	 * than the lsn of the transaction we are replaying. Note: we still
+	 * need to replay an owner change even though the inode is more recent
+	 * than the transaction as there is no guarantee that all the btree
+	 * blocks are more recent than this transaction, too.
 	 */
 	if (dip->di_version >= 3) {
 		xfs_lsn_t	lsn = be64_to_cpu(dip->di_lsn);
···
 		if (lsn && lsn != -1 && XFS_LSN_CMP(lsn, current_lsn) >= 0) {
 			trace_xfs_log_recover_inode_skip(log, in_f);
 			error = 0;
-			goto out_release;
+			goto out_owner_change;
 		}
 	}
 
···
 		    dicp->di_flushiter < (DI_MAX_FLUSH >> 1)) {
 			/* do nothing */
 		} else {
-			xfs_buf_relse(bp);
 			trace_xfs_log_recover_inode_skip(log, in_f);
 			error = 0;
-			goto error;
+			goto out_release;
 		}
 	}
 
···
 		    (dicp->di_format != XFS_DINODE_FMT_BTREE)) {
 			XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(3)",
 					 XFS_ERRLEVEL_LOW, mp, dicp);
-			xfs_buf_relse(bp);
 			xfs_alert(mp,
 		"%s: Bad regular inode log record, rec ptr 0x%p, "
 		"ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
 				__func__, item, dip, bp, in_f->ilf_ino);
 			error = EFSCORRUPTED;
-			goto error;
+			goto out_release;
 		}
 	} else if (unlikely(S_ISDIR(dicp->di_mode))) {
 		if ((dicp->di_format != XFS_DINODE_FMT_EXTENTS) &&
···
 		    (dicp->di_format != XFS_DINODE_FMT_LOCAL)) {
 			XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(4)",
 					     XFS_ERRLEVEL_LOW, mp, dicp);
-			xfs_buf_relse(bp);
 			xfs_alert(mp,
 		"%s: Bad dir inode log record, rec ptr 0x%p, "
 		"ino ptr = 0x%p, ino bp = 0x%p, ino %Ld",
 				__func__, item, dip, bp, in_f->ilf_ino);
 			error = EFSCORRUPTED;
-			goto error;
+			goto out_release;
 		}
 	}
 	if (unlikely(dicp->di_nextents + dicp->di_anextents > dicp->di_nblocks)){
 		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(5)",
 				     XFS_ERRLEVEL_LOW, mp, dicp);
-		xfs_buf_relse(bp);
 		xfs_alert(mp,
 	"%s: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, "
 	"dino bp 0x%p, ino %Ld, total extents = %d, nblocks = %Ld",
···
 			dicp->di_nextents + dicp->di_anextents,
 			dicp->di_nblocks);
 		error = EFSCORRUPTED;
-		goto error;
+		goto out_release;
 	}
 	if (unlikely(dicp->di_forkoff > mp->m_sb.sb_inodesize)) {
 		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(6)",
 				     XFS_ERRLEVEL_LOW, mp, dicp);
-		xfs_buf_relse(bp);
 		xfs_alert(mp,
 	"%s: Bad inode log record, rec ptr 0x%p, dino ptr 0x%p, "
 	"dino bp 0x%p, ino %Ld, forkoff 0x%x", __func__,
 			item, dip, bp, in_f->ilf_ino, dicp->di_forkoff);
 		error = EFSCORRUPTED;
-		goto error;
+		goto out_release;
 	}
 	isize = xfs_icdinode_size(dicp->di_version);
 	if (unlikely(item->ri_buf[1].i_len > isize)) {
 		XFS_CORRUPTION_ERROR("xlog_recover_inode_pass2(7)",
 				     XFS_ERRLEVEL_LOW, mp, dicp);
-		xfs_buf_relse(bp);
 		xfs_alert(mp,
 			"%s: Bad inode log record length %d, rec ptr 0x%p",
 			__func__, item->ri_buf[1].i_len, item);
 		error = EFSCORRUPTED;
-		goto error;
+		goto out_release;
 	}
 
 	/* The core is in in-core format */
···
 	}
 
 	if (in_f->ilf_size == 2)
-		goto write_inode_buffer;
+		goto out_owner_change;
 	len = item->ri_buf[2].i_len;
 	src = item->ri_buf[2].i_addr;
 	ASSERT(in_f->ilf_size <= 4);
···
 		default:
 			xfs_warn(log->l_mp, "%s: Invalid flag", __func__);
 			ASSERT(0);
-			xfs_buf_relse(bp);
 			error = EIO;
-			goto error;
+			goto out_release;
 		}
 	}
 
-write_inode_buffer:
+out_owner_change:
+	if (in_f->ilf_fields & (XFS_ILOG_DOWNER|XFS_ILOG_AOWNER))
+		error = xfs_recover_inode_owner_change(mp, dip, in_f,
+						       buffer_list);
 	/* re-generate the checksum. */
 	xfs_dinode_calc_crc(log->l_mp, dip);
 

fs/xfs/xfs_symlink.c

···
 #include "xfs_trans_space.h"
 #include "xfs_trace.h"
 #include "xfs_symlink.h"
+#include "xfs_buf_item.h"
 
 /* ----- Kernel only functions below ----- */
 STATIC int
···
 			pathlen -= byte_cnt;
 			offset += byte_cnt;
 
+			xfs_trans_buf_set_type(tp, bp, XFS_BLFT_SYMLINK_BUF);
 			xfs_trans_log_buf(tp, bp, 0, (buf + byte_cnt - 1) -
 							(char *)bp->b_addr);
 		}