[PATCH v2 0/9] xfs-4.17: online scrub fixes

@@ -173,7 +173,7 @@ extern void xfs_dir2_data_log_unused(struct xfs_da_args *args,
 extern void xfs_dir2_data_make_free(struct xfs_da_args *args,
 		struct xfs_buf *bp, xfs_dir2_data_aoff_t offset,
 		xfs_dir2_data_aoff_t len, int *needlogp, int *needscanp);
-extern void xfs_dir2_data_use_free(struct xfs_da_args *args,
+extern int xfs_dir2_data_use_free(struct xfs_da_args *args,
 		struct xfs_buf *bp, struct xfs_dir2_data_unused *dup,
 		xfs_dir2_data_aoff_t offset, xfs_dir2_data_aoff_t len,
 		int *needlogp, int *needscanp);

@@ -454,12 +454,15 @@ xfs_dir2_block_addname(
 		/*
 		 * Mark the space needed for the new leaf entry, now in use.
 		 */
-		xfs_dir2_data_use_free(args, bp, enddup,
+		error = xfs_dir2_data_use_free(args, bp, enddup,
 			(xfs_dir2_data_aoff_t)
 			((char *)enddup - (char *)hdr + be16_to_cpu(enddup->length) -
 			 sizeof(*blp)),
 			(xfs_dir2_data_aoff_t)sizeof(*blp),
 			&needlog, &needscan);
+		if (error)
+			return error;
+
 		/*
 		 * Update the tail (entry count).
 		 */

@@ -541,9 +544,11 @@ xfs_dir2_block_addname(
 	/*
 	 * Mark space for the data entry used.
 	 */
-	xfs_dir2_data_use_free(args, bp, dup,
+	error = xfs_dir2_data_use_free(args, bp, dup,
 		(xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr),
 		(xfs_dir2_data_aoff_t)len, &needlog, &needscan);
+	if (error)
+		return error;
 	/*
 	 * Create the new data entry.
 	 */

@@ -997,8 +1002,10 @@ xfs_dir2_leaf_to_block(
 	/*
 	 * Use up the space at the end of the block (blp/btp).
 	 */
-	xfs_dir2_data_use_free(args, dbp, dup, args->geo->blksize - size, size,
-		&needlog, &needscan);
+	error = xfs_dir2_data_use_free(args, dbp, dup,
+		args->geo->blksize - size, size, &needlog, &needscan);
+	if (error)
+		return error;
 	/*
 	 * Initialize the block tail.
 	 */

@@ -1110,18 +1117,14 @@ xfs_dir2_sf_to_block(
 	 * Add block 0 to the inode.
 	 */
 	error = xfs_dir2_grow_inode(args, XFS_DIR2_DATA_SPACE, &blkno);
-	if (error) {
-		kmem_free(sfp);
-		return error;
-	}
+	if (error)
+		goto out_free;
 	/*
 	 * Initialize the data block, then convert it to block format.
 	 */
 	error = xfs_dir3_data_init(args, blkno, &bp);
-	if (error) {
-		kmem_free(sfp);
-		return error;
-	}
+	if (error)
+		goto out_free;
 	xfs_dir3_block_init(mp, tp, bp, dp);
 	hdr = bp->b_addr;
 

@@ -1136,8 +1139,10 @@ xfs_dir2_sf_to_block(
 	 */
 	dup = dp->d_ops->data_unused_p(hdr);
 	needlog = needscan = 0;
-	xfs_dir2_data_use_free(args, bp, dup, args->geo->blksize - i,
+	error = xfs_dir2_data_use_free(args, bp, dup, args->geo->blksize - i,
 			       i, &needlog, &needscan);
+	if (error)
+		goto out_free;
 	ASSERT(needscan == 0);
 	/*
 	 * Fill in the tail.

@@ -1150,9 +1155,11 @@ xfs_dir2_sf_to_block(
 	/*
 	 * Remove the freespace, we'll manage it.
 	 */
-	xfs_dir2_data_use_free(args, bp, dup,
+	error = xfs_dir2_data_use_free(args, bp, dup,
 		(xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr),
 		be16_to_cpu(dup->length), &needlog, &needscan);
+	if (error)
+		goto out_free;
 	/*
 	 * Create entry for .
 	 */

@@ -1256,4 +1263,7 @@ xfs_dir2_sf_to_block(
 	xfs_dir2_block_log_tail(tp, bp);
 	xfs_dir3_data_check(dp, bp);
 	return 0;
+out_free:
+	kmem_free(sfp);
+	return error;
 }

@@ -932,10 +932,51 @@ xfs_dir2_data_make_free(
 	*needscanp = needscan;
 }
 
+/* Check our free data for obvious signs of corruption. */
+static inline xfs_failaddr_t
+xfs_dir2_data_check_free(
+	struct xfs_dir2_data_hdr	*hdr,
+	struct xfs_dir2_data_unused	*dup,
+	xfs_dir2_data_aoff_t		offset,
+	xfs_dir2_data_aoff_t		len)
+{
+	if (hdr->magic != cpu_to_be32(XFS_DIR2_DATA_MAGIC) &&
+	    hdr->magic != cpu_to_be32(XFS_DIR3_DATA_MAGIC) &&
+	    hdr->magic != cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) &&
+	    hdr->magic != cpu_to_be32(XFS_DIR3_BLOCK_MAGIC))
+		return __this_address;
+	if (be16_to_cpu(dup->freetag) != XFS_DIR2_DATA_FREE_TAG)
+		return __this_address;
+	if (offset < (char *)dup - (char *)hdr)
+		return __this_address;
+	if (offset + len > (char *)dup + be16_to_cpu(dup->length) - (char *)hdr)
+		return __this_address;
+	if ((char *)dup - (char *)hdr !=
+	    be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)))
+		return __this_address;
+	return NULL;
+}
+
+/* Sanity-check a new bestfree entry. */
+static inline xfs_failaddr_t
+xfs_dir2_data_check_new_free(
+	struct xfs_dir2_data_hdr	*hdr,
+	struct xfs_dir2_data_free	*dfp,
+	struct xfs_dir2_data_unused	*newdup)
+{
+	if (dfp == NULL)
+		return __this_address;
+	if (dfp->length != newdup->length)
+		return __this_address;
+	if (be16_to_cpu(dfp->offset) != (char *)newdup - (char *)hdr)
+		return __this_address;
+	return NULL;
+}
+
 /*
  * Take a byte range out of an existing unused space and make it un-free.
  */
-void
+int
 xfs_dir2_data_use_free(
 	struct xfs_da_args	*args,
 	struct xfs_buf		*bp,

@@ -947,23 +988,19 @@ xfs_dir2_data_use_free(
 {
 	xfs_dir2_data_hdr_t	*hdr;		/* data block header */
 	xfs_dir2_data_free_t	*dfp;		/* bestfree pointer */
+	xfs_dir2_data_unused_t	*newdup;	/* new unused entry */
+	xfs_dir2_data_unused_t	*newdup2;	/* another new unused entry */
+	struct xfs_dir2_data_free *bf;
+	xfs_failaddr_t		fa;
 	int			matchback;	/* matches end of freespace */
 	int			matchfront;	/* matches start of freespace */
 	int			needscan;	/* need to regen bestfree */
-	xfs_dir2_data_unused_t	*newdup;	/* new unused entry */
-	xfs_dir2_data_unused_t	*newdup2;	/* another new unused entry */
 	int			oldlen;		/* old unused entry's length */
-	struct xfs_dir2_data_free *bf;
 
 	hdr = bp->b_addr;
-	ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
-	       hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
-	       hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
-	       hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
-	ASSERT(be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG);
-	ASSERT(offset >= (char *)dup - (char *)hdr);
-	ASSERT(offset + len <= (char *)dup + be16_to_cpu(dup->length) - (char *)hdr);
-	ASSERT((char *)dup - (char *)hdr == be16_to_cpu(*xfs_dir2_data_unused_tag_p(dup)));
+	fa = xfs_dir2_data_check_free(hdr, dup, offset, len);
+	if (fa)
+		goto corrupt;
 	/*
 	 * Look up the entry in the bestfree table.
 	 */

@@ -1008,9 +1045,9 @@ xfs_dir2_data_use_free(
 			xfs_dir2_data_freeremove(hdr, bf, dfp, needlogp);
 			dfp = xfs_dir2_data_freeinsert(hdr, bf, newdup,
 						       needlogp);
-			ASSERT(dfp != NULL);
-			ASSERT(dfp->length == newdup->length);
-			ASSERT(be16_to_cpu(dfp->offset) == (char *)newdup - (char *)hdr);
+			fa = xfs_dir2_data_check_new_free(hdr, dfp, newdup);
+			if (fa)
+				goto corrupt;
 			/*
 			 * If we got inserted at the last slot,
 			 * that means we don't know if there was a better

@@ -1036,9 +1073,9 @@ xfs_dir2_data_use_free(
 			xfs_dir2_data_freeremove(hdr, bf, dfp, needlogp);
 			dfp = xfs_dir2_data_freeinsert(hdr, bf, newdup,
 						       needlogp);
-			ASSERT(dfp != NULL);
-			ASSERT(dfp->length == newdup->length);
-			ASSERT(be16_to_cpu(dfp->offset) == (char *)newdup - (char *)hdr);
+			fa = xfs_dir2_data_check_new_free(hdr, dfp, newdup);
+			if (fa)
+				goto corrupt;
 			/*
 			 * If we got inserted at the last slot,
 			 * that means we don't know if there was a better

@@ -1084,6 +1121,11 @@ xfs_dir2_data_use_free(
 		}
 	}
 	*needscanp = needscan;
+	return 0;
+corrupt:
+	xfs_corruption_error(__func__, XFS_ERRLEVEL_LOW, args->dp->i_mount,
+			hdr, __FILE__, __LINE__, fa);
+	return -EFSCORRUPTED;
 }
 
 /* Find the end of the entry data in a data/block format dir block. */

@@ -877,9 +877,13 @@ xfs_dir2_leaf_addname(
 	/*
 	 * Mark the initial part of our freespace in use for the new entry.
 	 */
-	xfs_dir2_data_use_free(args, dbp, dup,
+	error = xfs_dir2_data_use_free(args, dbp, dup,
 		(xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr), length,
 		&needlog, &needscan);
+	if (error) {
+		xfs_trans_brelse(tp, lbp);
+		return error;
+	}
 	/*
 	 * Initialize our new entry (at last).
 	 */

@@ -2023,9 +2023,11 @@ xfs_dir2_node_addname_int(
 	/*
 	 * Mark the first part of the unused space, inuse for us.
 	 */
-	xfs_dir2_data_use_free(args, dbp, dup,
+	error = xfs_dir2_data_use_free(args, dbp, dup,
 		(xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr), length,
 		&needlog, &needscan);
+	if (error)
+		return error;
 	/*
 	 * Fill in the new entry and log it.
 	 */

@@ -1261,11 +1261,15 @@ xfs_iread_extents(
 		 */
 		frp = XFS_BMBT_REC_ADDR(mp, block, 1);
 		for (j = 0; j < num_recs; j++, frp++, i++) {
+			xfs_failaddr_t	fa;
+
 			xfs_bmbt_disk_get_all(frp, &new);
-			if (!xfs_bmbt_validate_extent(mp, whichfork, &new)) {
-				XFS_ERROR_REPORT("xfs_bmap_read_extents(2)",
-						 XFS_ERRLEVEL_LOW, mp);
+			fa = xfs_bmap_validate_extent(ip, whichfork, &new);
+			if (fa) {
 				error = -EFSCORRUPTED;
+				xfs_inode_verifier_error(ip, error,
+						"xfs_bmap_read_extents(2)",
+						frp, sizeof(*frp), fa);
 				goto out_brelse;
 			}
 			xfs_iext_insert(ip, &icur, &new, state);

@@ -6154,3 +6158,48 @@ xfs_bmap_finish_one(
 
 	return error;
 }
+
+/* Check that an extent does not have invalid flags or bad ranges. */
+xfs_failaddr_t
+xfs_bmbt_validate_extent(
+	struct xfs_mount	*mp,
+	bool			isrt,
+	int			whichfork,
+	struct xfs_bmbt_irec	*irec)
+{
+	xfs_fsblock_t		endfsb;
+
+	endfsb = irec->br_startblock + irec->br_blockcount - 1;
+	if (isrt) {
+		if (!xfs_verify_rtbno(mp, irec->br_startblock))
+			return __this_address;
+		if (!xfs_verify_rtbno(mp, endfsb))
+			return __this_address;
+	} else {
+		if (!xfs_verify_fsbno(mp, irec->br_startblock))
+			return __this_address;
+		if (!xfs_verify_fsbno(mp, endfsb))
+			return __this_address;
+		if (XFS_FSB_TO_AGNO(mp, irec->br_startblock) !=
+		    XFS_FSB_TO_AGNO(mp, endfsb))
+			return __this_address;
+	}
+	if (irec->br_state != XFS_EXT_NORM) {
+		if (whichfork != XFS_DATA_FORK)
+			return __this_address;
+		if (!xfs_sb_version_hasextflgbit(&mp->m_sb))
+			return __this_address;
+	}
+	return NULL;
+}
+
+/* Check that an inode's extent does not have invalid flags or bad ranges. */
+xfs_failaddr_t
+xfs_bmap_validate_extent(
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_bmbt_irec	*irec)
+{
+	return xfs_bmbt_validate_extent(ip->i_mount, XFS_IS_REALTIME_INODE(ip),
+			whichfork, irec);
+}

@@ -274,4 +274,9 @@ static inline int xfs_bmap_fork_to_state(int whichfork)
 	}
 }
 
+xfs_failaddr_t xfs_bmbt_validate_extent(struct xfs_mount *mp, bool isrt,
+		int whichfork, struct xfs_bmbt_irec *irec);
+xfs_failaddr_t xfs_bmap_validate_extent(struct xfs_inode *ip, int whichfork,
+		struct xfs_bmbt_irec *irec);
+
 #endif	/* __XFS_BMAP_H__ */

@@ -118,18 +118,4 @@ extern int xfs_bmbt_change_owner(struct xfs_trans *tp, struct xfs_inode *ip,
 extern struct xfs_btree_cur *xfs_bmbt_init_cursor(struct xfs_mount *,
 		struct xfs_trans *, struct xfs_inode *, int);
 
-/*
- * Check that the extent does not contain an invalid unwritten extent flag.
- */
-static inline bool xfs_bmbt_validate_extent(struct xfs_mount *mp, int whichfork,
-		struct xfs_bmbt_irec *irec)
-{
-	if (irec->br_state == XFS_EXT_NORM)
-		return true;
-	if (whichfork == XFS_DATA_FORK &&
-	    xfs_sb_version_hasextflgbit(&mp->m_sb))
-		return true;
-	return false;
-}
-
 #endif	/* __XFS_BMAP_BTREE_H__ */

@@ -245,10 +245,14 @@ xfs_iformat_extents(
 
 		xfs_iext_first(ifp, &icur);
 		for (i = 0; i < nex; i++, dp++) {
+			xfs_failaddr_t	fa;
+
 			xfs_bmbt_disk_get_all(dp, &new);
-			if (!xfs_bmbt_validate_extent(mp, whichfork, &new)) {
-				XFS_ERROR_REPORT("xfs_iformat_extents(2)",
-						 XFS_ERRLEVEL_LOW, mp);
+			fa = xfs_bmap_validate_extent(ip, whichfork, &new);
+			if (fa) {
+				xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+						"xfs_iformat_extents(2)",
+						dp, sizeof(*dp), fa);
 				return -EFSCORRUPTED;
 			}
 

@@ -595,7 +599,7 @@ xfs_iextents_copy(
 	for_each_xfs_iext(ifp, &icur, &rec) {
 		if (isnullstartblock(rec.br_startblock))
 			continue;
-		ASSERT(xfs_bmbt_validate_extent(ip->i_mount, whichfork, &rec));
+		ASSERT(xfs_bmap_validate_extent(ip, whichfork, &rec) == NULL);
 		xfs_bmbt_disk_set_all(dp, &rec);
 		trace_xfs_write_extent(ip, &icur, state, _RET_IP_);
 		copied += sizeof(struct xfs_bmbt_rec);

@@ -1244,8 +1244,9 @@ xfs_iread_extents(
 			xfs_warn(ip->i_mount,
 				"corrupt dinode %Lu, (btree extents).",
 				(unsigned long long) ip->i_ino);
-			XFS_CORRUPTION_ERROR(__func__,
-				XFS_ERRLEVEL_LOW, ip->i_mount, block);
+			xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+					__func__, block, sizeof(*block),
+					__this_address);
 			error = -EFSCORRUPTED;
 			goto out_brelse;
 		}

@@ -195,8 +195,9 @@ xfs_iformat_local(
 	"corrupt inode %Lu (bad size %d for local fork, size = %d).",
 			(unsigned long long) ip->i_ino, size,
 			XFS_DFORK_SIZE(dip, ip->i_mount, whichfork));
-		XFS_CORRUPTION_ERROR("xfs_iformat_local", XFS_ERRLEVEL_LOW,
-				     ip->i_mount, dip);
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+				"xfs_iformat_local", dip, sizeof(*dip),
+				__this_address);
 		return -EFSCORRUPTED;
 	}
 

@@ -231,8 +232,9 @@ xfs_iformat_extents(
 	if (unlikely(size < 0 || size > XFS_DFORK_SIZE(dip, mp, whichfork))) {
 		xfs_warn(ip->i_mount, "corrupt inode %Lu ((a)extents = %d).",
 			(unsigned long long) ip->i_ino, nex);
-		XFS_CORRUPTION_ERROR("xfs_iformat_extents(1)", XFS_ERRLEVEL_LOW,
-				     mp, dip);
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+				"xfs_iformat_extents(1)", dip, sizeof(*dip),
+				__this_address);
 		return -EFSCORRUPTED;
 	}
 

@@ -309,8 +311,9 @@ xfs_iformat_btree(
 		     level == 0 || level > XFS_BTREE_MAXLEVELS) {
 		xfs_warn(mp, "corrupt inode %Lu (btree).",
 					(unsigned long long) ip->i_ino);
-		XFS_CORRUPTION_ERROR("xfs_iformat_btree", XFS_ERRLEVEL_LOW,
-					 mp, dip);
+		xfs_inode_verifier_error(ip, -EFSCORRUPTED,
+				"xfs_iformat_btree", dfp, size,
+				__this_address);
 		return -EFSCORRUPTED;
 	}
 

@@ -115,13 +115,15 @@ xfs_inode_buf_verify(
 				return;
 			}
 
-			xfs_verifier_error(bp, -EFSCORRUPTED, __this_address);
 #ifdef DEBUG
 			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));
 #endif
+			xfs_buf_verifier_error(bp, -EFSCORRUPTED,
+					__func__, dip, sizeof(*dip),
+					NULL);
 		}
 	}
 	xfs_inobp_check(mp, bp);

@@ -347,27 +347,32 @@ xfs_corruption_error(
  * values, and omit the stack trace unless the error level is tuned high.
  */
 void
-xfs_verifier_error(
+xfs_buf_verifier_error(
 	struct xfs_buf		*bp,
 	int			error,
+	const char		*name,
+	void			*buf,
+	size_t			bufsz,
 	xfs_failaddr_t		failaddr)
 {
 	struct xfs_mount	*mp = bp->b_target->bt_mount;
 	xfs_failaddr_t		fa;
+	int			sz;
 
 	fa = failaddr ? failaddr : __return_address;
 	__xfs_buf_ioerror(bp, error, fa);
 
-	xfs_alert(mp, "Metadata %s detected at %pS, %s block 0x%llx",
+	xfs_alert(mp, "Metadata %s detected at %pS, %s block 0x%llx %s",
 		  bp->b_error == -EFSBADCRC ? "CRC error" : "corruption",
-		  fa, bp->b_ops->name, bp->b_bn);
+		  fa, bp->b_ops->name, bp->b_bn, name);
 
 	xfs_alert(mp, "Unmount and run xfs_repair");
 
 	if (xfs_error_level >= XFS_ERRLEVEL_LOW) {
+		sz = min_t(size_t, XFS_CORRUPTION_DUMP_LEN, bufsz);
 		xfs_alert(mp, "First %d bytes of corrupted metadata buffer:",
-				XFS_CORRUPTION_DUMP_LEN);
-		xfs_hex_dump(xfs_buf_offset(bp, 0), XFS_CORRUPTION_DUMP_LEN);
+				sz);
+		xfs_hex_dump(buf, sz);
 	}
 
 	if (xfs_error_level >= XFS_ERRLEVEL_HIGH)

@@ -375,6 +380,20 @@ xfs_verifier_error(
 }
 
 /*
+ * Warnings specifically for verifier errors.  Differentiate CRC vs. invalid
+ * values, and omit the stack trace unless the error level is tuned high.
+ */
+void
+xfs_verifier_error(
+	struct xfs_buf		*bp,
+	int			error,
+	xfs_failaddr_t		failaddr)
+{
+	return xfs_buf_verifier_error(bp, error, "", xfs_buf_offset(bp, 0),
+			XFS_CORRUPTION_DUMP_LEN, failaddr);
+}
+
+/*
  * Warnings for inode corruption problems.  Don't bother with the stack
  * trace unless the error level is turned up high.
  */

@@ -26,6 +26,9 @@ extern void xfs_error_report(const char *tag, int level, struct xfs_mount *mp,
 extern void xfs_corruption_error(const char *tag, int level,
 			struct xfs_mount *mp, void *p, const char *filename,
 			int linenum, xfs_failaddr_t failaddr);
+extern void xfs_buf_verifier_error(struct xfs_buf *bp, int error,
+			const char *name, void *buf, size_t bufsz,
+			xfs_failaddr_t failaddr);
 extern void xfs_verifier_error(struct xfs_buf *bp, int error,
 			xfs_failaddr_t failaddr);
 extern void xfs_inode_verifier_error(struct xfs_inode *ip, int error,

@@ -37,6 +37,7 @@
 #include "xfs_bmap_util.h"
 #include "xfs_bmap_btree.h"
 #include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
 #include "xfs_refcount.h"
 #include "scrub/xfs_scrub.h"
 #include "scrub/scrub.h"

@@ -423,6 +424,163 @@ xfs_scrub_bmap_btree(
 	return error;
 }
 
+struct xfs_scrub_bmap_check_rmap_info {
+	struct xfs_scrub_context	*sc;
+	int				whichfork;
+	struct xfs_iext_cursor		icur;
+};
+
+/* Can we find bmaps that fit this rmap? */
+STATIC int
+xfs_scrub_bmap_check_rmap(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_bmbt_irec		irec;
+	struct xfs_scrub_bmap_check_rmap_info	*sbcri = priv;
+	struct xfs_ifork		*ifp;
+	struct xfs_scrub_context	*sc = sbcri->sc;
+	bool				have_map;
+
+	/* Is this even the right fork? */
+	if (rec->rm_owner != sc->ip->i_ino)
+		return 0;
+	if ((sbcri->whichfork == XFS_ATTR_FORK) ^
+	    !!(rec->rm_flags & XFS_RMAP_ATTR_FORK))
+		return 0;
+	if (rec->rm_flags & XFS_RMAP_BMBT_BLOCK)
+		return 0;
+
+	/* Now look up the bmbt record. */
+	ifp = XFS_IFORK_PTR(sc->ip, sbcri->whichfork);
+	if (!ifp) {
+		xfs_scrub_fblock_set_corrupt(sc, sbcri->whichfork,
+				rec->rm_offset);
+		goto out;
+	}
+	have_map = xfs_iext_lookup_extent(sc->ip, ifp, rec->rm_offset,
+			&sbcri->icur, &irec);
+	if (!have_map)
+		xfs_scrub_fblock_set_corrupt(sc, sbcri->whichfork,
+				rec->rm_offset);
+
+	while (have_map) {
+		if (irec.br_startoff != rec->rm_offset)
+			xfs_scrub_fblock_set_corrupt(sc, sbcri->whichfork,
+					rec->rm_offset);
+		if (irec.br_startblock != XFS_AGB_TO_FSB(sc->mp,
+				cur->bc_private.a.agno, rec->rm_startblock))
+			xfs_scrub_fblock_set_corrupt(sc, sbcri->whichfork,
+					rec->rm_offset);
+		if (irec.br_blockcount > rec->rm_blockcount)
+			xfs_scrub_fblock_set_corrupt(sc, sbcri->whichfork,
+					rec->rm_offset);
+		if (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
+			break;
+		rec->rm_startblock += irec.br_blockcount;
+		rec->rm_offset += irec.br_blockcount;
+		rec->rm_blockcount -= irec.br_blockcount;
+		if (rec->rm_blockcount == 0)
+			break;
+		have_map = xfs_iext_next_extent(ifp, &sbcri->icur, &irec);
+		if (!have_map)
+			xfs_scrub_fblock_set_corrupt(sc, sbcri->whichfork,
+					rec->rm_offset);
+	}
+
+out:
+	if (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
+		return XFS_BTREE_QUERY_RANGE_ABORT;
+	return 0;
+}
+
+/* Make sure each rmap has a corresponding bmbt entry. */
+STATIC int
+xfs_scrub_bmap_check_ag_rmaps(
+	struct xfs_scrub_context	*sc,
+	int				whichfork,
+	xfs_agnumber_t			agno)
+{
+	struct xfs_scrub_bmap_check_rmap_info	sbcri;
+	struct xfs_btree_cur		*cur;
+	struct xfs_buf			*agf;
+	int				error;
+
+	error = xfs_alloc_read_agf(sc->mp, sc->tp, agno, 0, &agf);
+	if (error)
+		return error;
+
+	cur = xfs_rmapbt_init_cursor(sc->mp, sc->tp, agf, agno);
+	if (!cur) {
+		error = -ENOMEM;
+		goto out_agf;
+	}
+
+	sbcri.sc = sc;
+	sbcri.whichfork = whichfork;
+	error = xfs_rmap_query_all(cur, xfs_scrub_bmap_check_rmap, &sbcri);
+	if (error == XFS_BTREE_QUERY_RANGE_ABORT)
+		error = 0;
+
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+out_agf:
+	xfs_trans_brelse(sc->tp, agf);
+	return error;
+}
+
+/* Make sure each rmap has a corresponding bmbt entry. */
+STATIC int
+xfs_scrub_bmap_check_rmaps(
+	struct xfs_scrub_context	*sc,
+	int				whichfork)
+{
+	loff_t				size;
+	xfs_agnumber_t			agno;
+	int				error;
+
+	if (!xfs_sb_version_hasrmapbt(&sc->mp->m_sb) ||
+	    whichfork == XFS_COW_FORK ||
+	    (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT))
+		return 0;
+
+	/* Don't support realtime rmap checks yet. */
+	if (XFS_IS_REALTIME_INODE(sc->ip) && whichfork == XFS_DATA_FORK)
+		return 0;
+
+	/*
+	 * Only do this for complex maps that are in btree format, or for
+	 * situations where we would seem to have a size but zero extents.
+	 * The inode repair code can zap broken iforks, which means we have
+	 * to flag this bmap as corrupt if there are rmaps that need to be
+	 * reattached.
+	 */
+	switch (whichfork) {
+	case XFS_DATA_FORK:
+		size = i_size_read(VFS_I(sc->ip));
+		break;
+	case XFS_ATTR_FORK:
+		size = XFS_IFORK_Q(sc->ip);
+		break;
+	default:
+		size = 0;
+		break;
+	}
+	if (XFS_IFORK_FORMAT(sc->ip, whichfork) != XFS_DINODE_FMT_BTREE &&
+	    (size == 0 || XFS_IFORK_NEXTENTS(sc->ip, whichfork) > 0))
+		return 0;
+
+	for (agno = 0; agno < sc->mp->m_sb.sb_agcount; agno++) {
+		error = xfs_scrub_bmap_check_ag_rmaps(sc, whichfork, agno);
+		if (error)
+			return error;
+		if (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
+			break;
+	}
+
+	return 0;
+}
+
 /*
  * Scrub an inode fork's block mappings.
  *

@@ -463,7 +621,7 @@ xfs_scrub_bmap(
 		break;
 	case XFS_ATTR_FORK:
 		if (!ifp)
-			goto out;
+			goto out_check_rmap;
 		if (!xfs_sb_version_hasattr(&mp->m_sb) &&
 		    !xfs_sb_version_hasattr2(&mp->m_sb))
 			xfs_scrub_ino_set_corrupt(sc, sc->ip->i_ino, NULL);

@@ -534,6 +692,10 @@ xfs_scrub_bmap(
 			goto out;
 	}
 
+out_check_rmap:
+	error = xfs_scrub_bmap_check_rmaps(sc, whichfork);
+	if (!xfs_scrub_fblock_xref_process_error(sc, whichfork, 0, &error))
+		goto out;
 out:
 	return error;
 }

@@ -515,72 +515,6 @@ xfs_scrub_dinode(
 	}
 }
 
-/* Map and read a raw inode. */
-STATIC int
-xfs_scrub_inode_map_raw(
-	struct xfs_scrub_context	*sc,
-	xfs_ino_t			ino,
-	struct xfs_buf			**bpp,
-	struct xfs_dinode		**dipp)
-{
-	struct xfs_imap			imap;
-	struct xfs_mount		*mp = sc->mp;
-	struct xfs_buf			*bp = NULL;
-	struct xfs_dinode		*dip;
-	int				error;
-
-	error = xfs_imap(mp, sc->tp, ino, &imap, XFS_IGET_UNTRUSTED);
-	if (error == -EINVAL) {
-		/*
-		 * Inode could have gotten deleted out from under us;
-		 * just forget about it.
-		 */
-		error = -ENOENT;
-		goto out;
-	}
-	if (!xfs_scrub_process_error(sc, XFS_INO_TO_AGNO(mp, ino),
-			XFS_INO_TO_AGBNO(mp, ino), &error))
-		goto out;
-
-	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
-			imap.im_blkno, imap.im_len, XBF_UNMAPPED, &bp,
-			NULL);
-	if (!xfs_scrub_process_error(sc, XFS_INO_TO_AGNO(mp, ino),
-			XFS_INO_TO_AGBNO(mp, ino), &error))
-		goto out;
-
-	/*
-	 * Is this really an inode?  We disabled verifiers in the above
-	 * xfs_trans_read_buf call because the inode buffer verifier
-	 * fails on /any/ inode record in the inode cluster with a bad
-	 * magic or version number, not just the one that we're
-	 * checking.  Therefore, grab the buffer unconditionally, attach
-	 * the inode verifiers by hand, and run the inode verifier only
-	 * on the one inode we want.
-	 */
-	bp->b_ops = &xfs_inode_buf_ops;
-	dip = xfs_buf_offset(bp, imap.im_boffset);
-	if (xfs_dinode_verify(mp, ino, dip) != NULL ||
-	    !xfs_dinode_good_version(mp, dip->di_version)) {
-		xfs_scrub_ino_set_corrupt(sc, ino, bp);
-		goto out_buf;
-	}
-
-	/* ...and is it the one we asked for? */
-	if (be32_to_cpu(dip->di_gen) != sc->sm->sm_gen) {
-		error = -ENOENT;
-		goto out_buf;
-	}
-
-	*dipp = dip;
-	*bpp = bp;
-out:
-	return error;
-out_buf:
-	xfs_trans_brelse(sc->tp, bp);
-	return error;
-}
-
 /*
  * Make sure the finobt doesn't think this inode is free.
  * We don't have to check the inobt ourselves because we got the inode via

@@ -727,43 +661,29 @@ xfs_scrub_inode(
 	struct xfs_scrub_context	*sc)
 {
 	struct xfs_dinode		di;
-	struct xfs_buf			*bp = NULL;
-	struct xfs_dinode		*dip;
-	xfs_ino_t			ino;
 	int				error = 0;
 
-	/* Did we get the in-core inode, or are we doing this manually? */
-	if (sc->ip) {
-		ino = sc->ip->i_ino;
-		xfs_inode_to_disk(sc->ip, &di, 0);
-		dip = &di;
-	} else {
-		/* Map & read inode. */
-		ino = sc->sm->sm_ino;
-		error = xfs_scrub_inode_map_raw(sc, ino, &bp, &dip);
-		if (error || !bp)
-			goto out;
+	/* iget failed means automatic fail. */
+	if (!sc->ip) {
+		xfs_scrub_ino_set_corrupt(sc, sc->sm->sm_ino, NULL);
+		return 0;
 	}
 
-	xfs_scrub_dinode(sc, bp, dip, ino);
+	/* Scrub the inode core. */
+	xfs_inode_to_disk(sc->ip, &di, 0);
+	xfs_scrub_dinode(sc, NULL, &di, sc->ip->i_ino);
 	if (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
 		goto out;
 
-	/* Now let's do the things that require a live inode. */
-	if (!sc->ip)
-		goto out;
-
 	/*
 	 * Look for discrepancies between file's data blocks and the reflink
 	 * iflag.  We already checked the iflag against the file mode when
 	 * we scrubbed the dinode.
 	 */
 	if (S_ISREG(VFS_I(sc->ip)->i_mode))
-		xfs_scrub_inode_check_reflink_iflag(sc, ino, bp);
+		xfs_scrub_inode_check_reflink_iflag(sc, sc->ip->i_ino, NULL);
 
-	xfs_scrub_inode_xref(sc, ino, dip);
+	xfs_scrub_inode_xref(sc, sc->ip->i_ino, &di);
 out:
-	if (bp)
-		xfs_trans_brelse(sc->tp, bp);
 	return error;
 }