xfs: implement live quotacheck inode scan

Create a new trio of scrub functions to check quota counters.  While the
dquots themselves are filesystem metadata and should be checked early,
the dquot counter values are computed from other metadata and are
therefore summary counters.  We don't plug these into the scrub dispatch
just yet, because we still need to be able to watch quota updates while
doing our scan.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>

13 files changed, 676 insertions, 4 deletions

diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index db0f822c58c8..958a3244c808 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -164,6 +164,7 @@ xfs-$(CONFIG_XFS_RT)		+= $(addprefix scrub/, \
 xfs-$(CONFIG_XFS_QUOTA)		+= $(addprefix scrub/, \
 				   iscan.o \
 				   quota.o \
+				   quotacheck.o \
 				   )
 
 # online repair
diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
index f0560861c94e..6a7aba52089f 100644
--- a/fs/xfs/libxfs/xfs_fs.h
+++ b/fs/xfs/libxfs/xfs_fs.h
@@ -716,9 +716,10 @@ struct xfs_scrub_metadata {
 #define XFS_SCRUB_TYPE_GQUOTA	22	/* group quotas */
 #define XFS_SCRUB_TYPE_PQUOTA	23	/* project quotas */
 #define XFS_SCRUB_TYPE_FSCOUNTERS 24	/* fs summary counters */
+#define XFS_SCRUB_TYPE_QUOTACHECK 25	/* quota counters */
 
 /* Number of scrub subcommands. */
-#define XFS_SCRUB_TYPE_NR	25
+#define XFS_SCRUB_TYPE_NR	26
 
 /* i: Repair this metadata. */
 #define XFS_SCRUB_IFLAG_REPAIR		(1 << 0)
diff --git a/fs/xfs/scrub/common.c b/fs/xfs/scrub/common.c
index b37b3d07aeed..d1273d4dd79e 100644
--- a/fs/xfs/scrub/common.c
+++ b/fs/xfs/scrub/common.c
@@ -27,6 +27,7 @@
 #include "xfs_attr.h"
 #include "xfs_reflink.h"
 #include "xfs_ag.h"
+#include "xfs_error.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/trace.h"
@@ -79,6 +80,15 @@ __xchk_process_error(
 				sc->ip ? sc->ip : XFS_I(file_inode(sc->file)),
 				sc->sm, *error);
 		break;
+	case -ECANCELED:
+		/*
+		 * ECANCELED here means that the caller set one of the scrub
+		 * outcome flags (corrupt, xfail, xcorrupt) and wants to exit
+		 * quickly.  Set error to zero and do not continue.
+		 */
+		trace_xchk_op_error(sc, agno, bno, *error, ret_ip);
+		*error = 0;
+		break;
 	case -EFSBADCRC:
 	case -EFSCORRUPTED:
 		/* Note the badness but don't abort. */
@@ -86,8 +96,7 @@ __xchk_process_error(
 		*error = 0;
 		fallthrough;
 	default:
-		trace_xchk_op_error(sc, agno, bno, *error,
-				ret_ip);
+		trace_xchk_op_error(sc, agno, bno, *error, ret_ip);
 		break;
 	}
 	return false;
@@ -132,6 +141,16 @@ __xchk_fblock_process_error(
 		/* Used to restart an op with deadlock avoidance. */
 		trace_xchk_deadlock_retry(sc->ip, sc->sm, *error);
 		break;
+	case -ECANCELED:
+		/*
+		 * ECANCELED here means that the caller set one of the scrub
+		 * outcome flags (corrupt, xfail, xcorrupt) and wants to exit
+		 * quickly.  Set error to zero and do not continue.
+		 */
+		trace_xchk_file_op_error(sc, whichfork, offset, *error,
+				ret_ip);
+		*error = 0;
+		break;
 	case -EFSBADCRC:
 	case -EFSCORRUPTED:
 		/* Note the badness but don't abort. */
@@ -223,6 +242,17 @@ xchk_block_set_corrupt(
 	trace_xchk_block_error(sc, xfs_buf_daddr(bp), __return_address);
 }
 
+/* Record a corrupt quota counter. */
+void
+xchk_qcheck_set_corrupt(
+	struct xfs_scrub	*sc,
+	unsigned int		dqtype,
+	xfs_dqid_t		id)
+{
+	sc->sm->sm_flags |= XFS_SCRUB_OFLAG_CORRUPT;
+	trace_xchk_qcheck_error(sc, dqtype, id, __return_address);
+}
+
 /* Record a corruption while cross-referencing. */
 void
 xchk_block_xref_set_corrupt(
@@ -558,6 +588,21 @@ xchk_ag_init(
 
 /* Per-scrubber setup functions */
 
+void
+xchk_trans_cancel(
+	struct xfs_scrub	*sc)
+{
+	xfs_trans_cancel(sc->tp);
+	sc->tp = NULL;
+}
+
+int
+xchk_trans_alloc_empty(
+	struct xfs_scrub	*sc)
+{
+	return xfs_trans_alloc_empty(sc->mp, &sc->tp);
+}
+
 /*
  * Grab an empty transaction so that we can re-grab locked buffers if
  * one of our btrees turns out to be cyclic.
diff --git a/fs/xfs/scrub/common.h b/fs/xfs/scrub/common.h
index 2045857a947b..385b030206f7 100644
--- a/fs/xfs/scrub/common.h
+++ b/fs/xfs/scrub/common.h
@@ -32,6 +32,9 @@ xchk_should_terminate(
 }
 
 int xchk_trans_alloc(struct xfs_scrub *sc, uint resblks);
+int xchk_trans_alloc_empty(struct xfs_scrub *sc);
+void xchk_trans_cancel(struct xfs_scrub *sc);
+
 bool xchk_process_error(struct xfs_scrub *sc, xfs_agnumber_t agno,
 		xfs_agblock_t bno, int *error);
 bool xchk_fblock_process_error(struct xfs_scrub *sc, int whichfork,
@@ -52,6 +55,8 @@ void xchk_block_set_corrupt(struct xfs_scrub *sc,
 void xchk_ino_set_corrupt(struct xfs_scrub *sc, xfs_ino_t ino);
 void xchk_fblock_set_corrupt(struct xfs_scrub *sc, int whichfork,
 		xfs_fileoff_t offset);
+void xchk_qcheck_set_corrupt(struct xfs_scrub *sc, unsigned int dqtype,
+		xfs_dqid_t id);
 
 void xchk_block_xref_set_corrupt(struct xfs_scrub *sc,
 		struct xfs_buf *bp);
@@ -101,12 +106,18 @@ xchk_setup_rtsummary(struct xfs_scrub *sc)
 #endif
 #ifdef CONFIG_XFS_QUOTA
 int xchk_setup_quota(struct xfs_scrub *sc);
+int xchk_setup_quotacheck(struct xfs_scrub *sc);
 #else
 static inline int
 xchk_setup_quota(struct xfs_scrub *sc)
 {
 	return -ENOENT;
 }
+static inline int
+xchk_setup_quotacheck(struct xfs_scrub *sc)
+{
+	return -ENOENT;
+}
 #endif
 int xchk_setup_fscounters(struct xfs_scrub *sc);
 
diff --git a/fs/xfs/scrub/health.c b/fs/xfs/scrub/health.c
index 54b457482934..3130f7c4ee14 100644
--- a/fs/xfs/scrub/health.c
+++ b/fs/xfs/scrub/health.c
@@ -103,6 +103,7 @@ static const struct xchk_health_map type_to_health_flag[XFS_SCRUB_TYPE_NR] = {
 	[XFS_SCRUB_TYPE_GQUOTA]		= { XHG_FS,  XFS_SICK_FS_GQUOTA },
 	[XFS_SCRUB_TYPE_PQUOTA]		= { XHG_FS,  XFS_SICK_FS_PQUOTA },
 	[XFS_SCRUB_TYPE_FSCOUNTERS]	= { XHG_FS,  XFS_SICK_FS_COUNTERS },
+	[XFS_SCRUB_TYPE_QUOTACHECK]	= { XHG_FS,  XFS_SICK_FS_QUOTACHECK },
 };
 
 /* Return the health status mask for this scrub type. */
diff --git a/fs/xfs/scrub/quotacheck.c b/fs/xfs/scrub/quotacheck.c
new file mode 100644
index 000000000000..53bb2f9ad9dd
--- /dev/null
+++ b/fs/xfs/scrub/quotacheck.c
@@ -0,0 +1,461 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2021 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <djwong@kernel.org>
+ */
+#include "xfs.h"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_inode.h"
+#include "xfs_quota.h"
+#include "xfs_qm.h"
+#include "xfs_icache.h"
+#include "xfs_bmap_util.h"
+#include "xfs_ialloc.h"
+#include "xfs_ag.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/repair.h"
+#include "scrub/xfarray.h"
+#include "scrub/iscan.h"
+#include "scrub/quotacheck.h"
+#include "scrub/trace.h"
+
+/*
+ * Live Quotacheck
+ * ===============
+ *
+ * Quota counters are "summary" metadata, in the sense that they are computed
+ * as the summation of the block usage counts for every file on the filesystem.
+ * Therefore, we compute the correct icount, bcount, and rtbcount values by
+ * creating a shadow quota counter structure and walking every inode.
+ */
+
+/* Set us up to scrub quota counters. */
+int
+xchk_setup_quotacheck(
+	struct xfs_scrub	*sc)
+{
+	/* Not ready for general consumption yet. */
+	return -EOPNOTSUPP;
+
+	if (!XFS_IS_QUOTA_ON(sc->mp))
+		return -ENOENT;
+
+	sc->buf = kmem_zalloc(sizeof(struct xqcheck), KM_NOFS | KM_MAYFAIL);
+	if (!sc->buf)
+		return -ENOMEM;
+
+	return xchk_setup_fs(sc);
+}
+
+/* Update an incore dquot counter information from a live update. */
+static int
+xqcheck_update_incore_counts(
+	struct xqcheck		*xqc,
+	struct xfarray		*counts,
+	xfs_dqid_t		id,
+	int64_t			inodes,
+	int64_t			nblks,
+	int64_t			rtblks)
+{
+	struct xqcheck_dquot	xcdq;
+	int			error;
+
+	error = xfarray_load_sparse(counts, id, &xcdq);
+	if (error)
+		return error;
+
+	xcdq.icount += inodes;
+	xcdq.bcount += nblks;
+	xcdq.rtbcount += rtblks;
+
+	error = xfarray_store(counts, id, &xcdq);
+	if (error == -EFBIG) {
+		/*
+		 * EFBIG means we tried to store data at too high a byte offset
+		 * in the sparse array.  IOWs, we cannot complete the check and
+		 * must notify userspace that the check was incomplete.
+		 */
+		xchk_set_incomplete(xqc->sc);
+		error = -ECANCELED;
+	}
+	return error;
+}
+
+/* Record this inode's quota usage in our shadow quota counter data. */
+STATIC int
+xqcheck_inode(
+	struct xqcheck		*xqc,
+	struct xfs_inode	*ip)
+{
+	struct xfs_trans	*tp = xqc->sc->tp;
+	xfs_filblks_t		nblks, rtblks;
+	uint			ilock_flags = 0;
+	xfs_dqid_t		id;
+	int			error;
+
+	if (xfs_is_quota_inode(&tp->t_mountp->m_sb, ip->i_ino)) {
+		/*
+		 * Quota inode blocks are never counted towards quota, so we
+		 * do not need to take the lock.
+		 */
+		xchk_iscan_mark_visited(&xqc->iscan, ip);
+		return 0;
+	}
+
+	/* Figure out the data / rt device block counts. */
+	xfs_ilock(ip, XFS_IOLOCK_SHARED | XFS_MMAPLOCK_SHARED);
+	ilock_flags = xfs_ilock_data_map_shared(ip);
+	if (XFS_IS_REALTIME_INODE(ip)) {
+		error = xfs_iread_extents(tp, ip, XFS_DATA_FORK);
+		if (error)
+			goto out_ilock;
+	}
+	xfs_inode_count_blocks(tp, ip, &nblks, &rtblks);
+
+	/* Update the shadow dquot counters. */
+	mutex_lock(&xqc->lock);
+	id = xfs_qm_id_for_quotatype(ip, XFS_DQTYPE_USER);
+	if (xqc->ucounts) {
+		error = xqcheck_update_incore_counts(xqc, xqc->ucounts, id, 1,
+				nblks, rtblks);
+		if (error)
+			goto out_incomplete;
+	}
+
+	id = xfs_qm_id_for_quotatype(ip, XFS_DQTYPE_GROUP);
+	if (xqc->gcounts) {
+		error = xqcheck_update_incore_counts(xqc, xqc->gcounts, id, 1,
+				nblks, rtblks);
+		if (error)
+			goto out_incomplete;
+	}
+
+	id = xfs_qm_id_for_quotatype(ip, XFS_DQTYPE_PROJ);
+	if (xqc->pcounts) {
+		error = xqcheck_update_incore_counts(xqc, xqc->pcounts, id, 1,
+				nblks, rtblks);
+		if (error)
+			goto out_incomplete;
+	}
+	mutex_unlock(&xqc->lock);
+
+	xchk_iscan_mark_visited(&xqc->iscan, ip);
+	goto out_ilock;
+
+out_incomplete:
+	mutex_unlock(&xqc->lock);
+	xchk_set_incomplete(xqc->sc);
+out_ilock:
+	xfs_iunlock(ip, XFS_IOLOCK_SHARED | XFS_MMAPLOCK_SHARED | ilock_flags);
+	return error;
+}
+
+/* Walk all the allocated inodes and run a quota scan on them. */
+STATIC int
+xqcheck_collect_counts(
+	struct xqcheck		*xqc)
+{
+	struct xfs_scrub	*sc = xqc->sc;
+	struct xchk_iscan	*iscan = &xqc->iscan;
+	int			error;
+
+	/*
+	 * Set up for a potentially lengthy filesystem scan by reducing our
+	 * transaction resource usage for the duration.  Specifically:
+	 *
+	 * Cancel the transaction to release the log grant space while we scan
+	 * the filesystem.
+	 *
+	 * Create a new empty transaction to eliminate the possibility of the
+	 * inode scan deadlocking on cyclical metadata.
+	 *
+	 * We pass the empty transaction to the file scanning function to avoid
+	 * repeatedly cycling empty transactions.  This can be done without
+	 * risk of deadlock between sb_internal and the IOLOCK (we take the
+	 * IOLOCK to quiesce the file before scanning) because empty
+	 * transactions do not take sb_internal.
+	 */
+	xchk_trans_cancel(sc);
+	error = xchk_trans_alloc_empty(sc);
+	if (error)
+		return error;
+
+	while ((error = xchk_iscan_advance(sc, iscan)) == 1) {
+		struct xfs_inode	*ip;
+
+		error = xchk_iscan_iget(sc, iscan, &ip);
+		if (error == -EAGAIN)
+			continue;
+		if (error)
+			break;
+
+		error = xqcheck_inode(xqc, ip);
+		xfs_irele(ip);
+		if (error)
+			break;
+
+		if (xchk_should_terminate(sc, &error))
+			break;
+	}
+
+	if (error == -ECANCELED)
+		xchk_set_incomplete(sc);
+	if (error)
+		return error;
+
+	/*
+	 * Switch out for a real transaction in preparation for building a new
+	 * tree.
+	 */
+	xchk_trans_cancel(sc);
+	return xchk_setup_fs(sc);
+}
+
+/*
+ * Check the dquot data against what we observed.  Caller must hold the dquot
+ * lock.
+ */
+STATIC int
+xqcheck_compare_dquot(
+	struct xfs_dquot	*dqp,
+	xfs_dqtype_t		dqtype,
+	void			*priv)
+{
+	struct xqcheck_dquot	xcdq;
+	struct xqcheck		*xqc = priv;
+	struct xfarray		*counts = xqcheck_counters_for(xqc, dqtype);
+	int			error;
+
+	mutex_lock(&xqc->lock);
+	error = xfarray_load_sparse(counts, dqp->q_id, &xcdq);
+	if (error)
+		goto out_unlock;
+
+	if (xcdq.icount != dqp->q_ino.count)
+		xchk_qcheck_set_corrupt(xqc->sc, dqtype, dqp->q_id);
+
+	if (xcdq.bcount != dqp->q_blk.count)
+		xchk_qcheck_set_corrupt(xqc->sc, dqtype, dqp->q_id);
+
+	if (xcdq.rtbcount != dqp->q_rtb.count)
+		xchk_qcheck_set_corrupt(xqc->sc, dqtype, dqp->q_id);
+
+	xcdq.flags |= XQCHECK_DQUOT_COMPARE_SCANNED;
+	error = xfarray_store(counts, dqp->q_id, &xcdq);
+	if (error == -EFBIG) {
+		/*
+		 * EFBIG means we tried to store data at too high a byte offset
+		 * in the sparse array.  IOWs, we cannot complete the check and
+		 * must notify userspace that the check was incomplete.
+		 */
+		xchk_set_incomplete(xqc->sc);
+		error = -ECANCELED;
+	}
+	mutex_unlock(&xqc->lock);
+	if (error)
+		return error;
+
+	if (xqc->sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
+		return -ECANCELED;
+
+	return 0;
+
+out_unlock:
+	mutex_unlock(&xqc->lock);
+	return error;
+}
+
+/*
+ * Walk all the observed dquots, and make sure there's a matching incore
+ * dquot and that its counts match ours.
+ */
+STATIC int
+xqcheck_walk_observations(
+	struct xqcheck		*xqc,
+	xfs_dqtype_t		dqtype)
+{
+	struct xqcheck_dquot	xcdq;
+	struct xfs_dquot	*dqp;
+	struct xfarray		*counts = xqcheck_counters_for(xqc, dqtype);
+	uint64_t		nr = 0;
+	int			error;
+
+	mutex_lock(&xqc->lock);
+	while ((error = xfarray_iter(counts, &nr, &xcdq)) == 1) {
+		xfs_dqid_t	id = nr - 1;
+
+		if (xcdq.flags & XQCHECK_DQUOT_COMPARE_SCANNED)
+			continue;
+
+		mutex_unlock(&xqc->lock);
+
+		error = xfs_qm_dqget(xqc->sc->mp, id, dqtype, false, &dqp);
+		if (error == -ENOENT) {
+			xchk_qcheck_set_corrupt(xqc->sc, dqtype, id);
+			return 0;
+		}
+		if (error)
+			return error;
+
+		error = xqcheck_compare_dquot(dqp, dqtype, xqc);
+		xfs_qm_dqput(dqp);
+		if (error)
+			return error;
+
+		if (xchk_should_terminate(xqc->sc, &error))
+			return error;
+
+		mutex_lock(&xqc->lock);
+	}
+	mutex_unlock(&xqc->lock);
+
+	return error;
+}
+
+/* Compare the quota counters we observed against the live dquots. */
+STATIC int
+xqcheck_compare_dqtype(
+	struct xqcheck		*xqc,
+	xfs_dqtype_t		dqtype)
+{
+	struct xfs_scrub	*sc = xqc->sc;
+	int			error;
+
+	if (sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT)
+		return 0;
+
+	/* If the quota CHKD flag is cleared, we need to repair this quota. */
+	if (!(xfs_quota_chkd_flag(dqtype) & sc->mp->m_qflags)) {
+		xchk_qcheck_set_corrupt(xqc->sc, dqtype, 0);
+		return 0;
+	}
+
+	/* Compare what we observed against the actual dquots. */
+	error = xfs_qm_dqiterate(sc->mp, dqtype, xqcheck_compare_dquot, xqc);
+	if (error)
+		return error;
+
+	/* Walk all the observed dquots and compare to the incore ones. */
+	return xqcheck_walk_observations(xqc, dqtype);
+}
+
+/* Tear down everything associated with a quotacheck. */
+static void
+xqcheck_teardown_scan(
+	struct xqcheck		*xqc)
+{
+	if (xqc->pcounts) {
+		xfarray_destroy(xqc->pcounts);
+		xqc->pcounts = NULL;
+	}
+
+	if (xqc->gcounts) {
+		xfarray_destroy(xqc->gcounts);
+		xqc->gcounts = NULL;
+	}
+
+	if (xqc->ucounts) {
+		xfarray_destroy(xqc->ucounts);
+		xqc->ucounts = NULL;
+	}
+
+	xchk_iscan_finish(&xqc->iscan);
+	mutex_destroy(&xqc->lock);
+	xqc->sc = NULL;
+}
+
+/*
+ * Scan all inodes in the entire filesystem to generate quota counter data.
+ * If the scan is successful, the quota data will be left alive for a repair.
+ * If any error occurs, we'll tear everything down.
+ */
+STATIC int
+xqcheck_setup_scan(
+	struct xfs_scrub	*sc,
+	struct xqcheck		*xqc)
+{
+	int			error;
+
+	ASSERT(xqc->sc == NULL);
+	xqc->sc = sc;
+
+	mutex_init(&xqc->lock);
+	xqc->iscan.iget_tries = 20;
+	xqc->iscan.iget_retry_delay = HZ / 10;
+	xchk_iscan_start(&xqc->iscan);
+
+	error = -ENOMEM;
+	if (xfs_this_quota_on(sc->mp, XFS_DQTYPE_USER)) {
+		error = xfarray_create(sc->mp, "user dquots",
+				sizeof(struct xqcheck_dquot), &xqc->ucounts);
+		if (error)
+			goto out_teardown;
+	}
+
+	if (xfs_this_quota_on(sc->mp, XFS_DQTYPE_GROUP)) {
+		error = xfarray_create(sc->mp, "group dquots",
+				sizeof(struct xqcheck_dquot), &xqc->gcounts);
+		if (error)
+			goto out_teardown;
+	}
+
+	if (xfs_this_quota_on(sc->mp, XFS_DQTYPE_PROJ)) {
+		error = xfarray_create(sc->mp, "project dquots",
+				sizeof(struct xqcheck_dquot), &xqc->pcounts);
+		if (error)
+			goto out_teardown;
+	}
+
+	/* Use deferred cleanup to pass the quota count data to repair. */
+	sc->buf_cleanup = (void (*)(void *))xqcheck_teardown_scan;
+	return 0;
+
+out_teardown:
+	xqcheck_teardown_scan(xqc);
+	return error;
+}
+
+/* Scrub all counters for a given quota type. */
+int
+xchk_quotacheck(
+	struct xfs_scrub	*sc)
+{
+	struct xqcheck		*xqc = sc->buf;
+	int			error = 0;
+
+	/* Check quota counters on the live filesystem. */
+	error = xqcheck_setup_scan(sc, xqc);
+	if (error)
+		return error;
+
+	/* Walk all inodes, picking up quota information. */
+	error = xqcheck_collect_counts(xqc);
+	if (!xchk_xref_process_error(sc, 0, 0, &error))
+		return error;
+
+	/* Compare quota counters. */
+	if (xqc->ucounts) {
+		error = xqcheck_compare_dqtype(xqc, XFS_DQTYPE_USER);
+		if (!xchk_xref_process_error(sc, 0, 0, &error))
+			return error;
+	}
+	if (xqc->gcounts) {
+		error = xqcheck_compare_dqtype(xqc, XFS_DQTYPE_GROUP);
+		if (!xchk_xref_process_error(sc, 0, 0, &error))
+			return error;
+	}
+	if (xqc->pcounts) {
+		error = xqcheck_compare_dqtype(xqc, XFS_DQTYPE_PROJ);
+		if (!xchk_xref_process_error(sc, 0, 0, &error))
+			return error;
+	}
+
+	return 0;
+}
diff --git a/fs/xfs/scrub/quotacheck.h b/fs/xfs/scrub/quotacheck.h
new file mode 100644
index 000000000000..415f7a63418f
--- /dev/null
+++ b/fs/xfs/scrub/quotacheck.h
@@ -0,0 +1,61 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Copyright (C) 2021 Oracle.  All Rights Reserved.
+ * Author: Darrick J. Wong <djwong@kernel.org>
+ */
+#ifndef __XFS_SCRUB_QUOTACHECK_H__
+#define __XFS_SCRUB_QUOTACHECK_H__
+
+/* Quota counters for live quotacheck. */
+struct xqcheck_dquot {
+	/* block usage count */
+	int64_t			bcount;
+
+	/* inode usage count */
+	int64_t			icount;
+
+	/* realtime block usage count */
+	int64_t			rtbcount;
+
+	/* Record state */
+	unsigned int		flags;
+};
+
+/* Already checked this dquot */
+#define XQCHECK_DQUOT_COMPARE_SCANNED	(1U << 0)
+
+/* Live quotacheck control structure. */
+struct xqcheck {
+	struct xfs_scrub	*sc;
+
+	/* Shadow dquot counter data. */
+	struct xfarray		*ucounts;
+	struct xfarray		*gcounts;
+	struct xfarray		*pcounts;
+
+	/* Lock protecting quotacheck count observations */
+	struct mutex		lock;
+
+	struct xchk_iscan	iscan;
+};
+
+/* Return the incore counter array for a given quota type. */
+static inline struct xfarray *
+xqcheck_counters_for(
+	struct xqcheck		*xqc,
+	xfs_dqtype_t		dqtype)
+{
+	switch (dqtype) {
+	case XFS_DQTYPE_USER:
+		return xqc->ucounts;
+	case XFS_DQTYPE_GROUP:
+		return xqc->gcounts;
+	case XFS_DQTYPE_PROJ:
+		return xqc->pcounts;
+	}
+
+	ASSERT(0);
+	return NULL;
+}
+
+#endif /* __XFS_SCRUB_QUOTACHECK_H__ */
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 4bbb717ed262..ce72e66d925d 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -177,7 +177,10 @@ xchk_teardown(
 		sc->xfile = NULL;
 	}
 	if (sc->buf) {
+		if (sc->buf_cleanup)
+			sc->buf_cleanup(sc->buf);
 		kmem_free(sc->buf);
+		sc->buf_cleanup = NULL;
 		sc->buf = NULL;
 	}
 	return error;
@@ -345,6 +348,12 @@ static const struct xchk_meta_ops meta_scrub_ops[] = {
 		.scrub	= xchk_fscounters,
 		.repair	= xrep_notsupported,
 	},
+	[XFS_SCRUB_TYPE_QUOTACHECK] = {	/* quota counters */
+		.type	= ST_FS,
+		.setup	= xchk_setup_quotacheck,
+		.scrub	= xchk_quotacheck,
+		.repair	= xrep_notsupported,
+	},
 };
 
 /* This isn't a stable feature, warn once per day. */
diff --git a/fs/xfs/scrub/scrub.h b/fs/xfs/scrub/scrub.h
index 4bab7d8ce185..5071534324de 100644
--- a/fs/xfs/scrub/scrub.h
+++ b/fs/xfs/scrub/scrub.h
@@ -80,6 +80,14 @@ struct xfs_scrub {
 	/* Kernel memory buffer used by scrubbers; freed at teardown. */
 	void				*buf;
 
+	/*
+	 * Clean up resources owned by whatever is in the buffer.  Cleanup can
+	 * be deferred with this hook as a means for scrub functions to pass
+	 * data to repair functions.  This function must not free the buffer
+	 * itself.
+	 */
+	void				(*buf_cleanup)(void *buf);
+
 	/* xfile used by the scrubbers; freed at teardown. */
 	struct xfile			*xfile;
 
@@ -143,12 +151,18 @@ xchk_rtsummary(struct xfs_scrub *sc)
 #endif
 #ifdef CONFIG_XFS_QUOTA
 int xchk_quota(struct xfs_scrub *sc);
+int xchk_quotacheck(struct xfs_scrub *sc);
 #else
 static inline int
 xchk_quota(struct xfs_scrub *sc)
 {
 	return -ENOENT;
 }
+static inline int
+xchk_quotacheck(struct xfs_scrub *sc)
+{
+	return -ENOENT;
+}
 #endif
 int xchk_fscounters(struct xfs_scrub *sc);
 
diff --git a/fs/xfs/scrub/trace.h b/fs/xfs/scrub/trace.h
index e320cac10136..550fee84e542 100644
--- a/fs/xfs/scrub/trace.h
+++ b/fs/xfs/scrub/trace.h
@@ -15,6 +15,7 @@
 
 #include <linux/tracepoint.h>
 #include "xfs_bit.h"
+#include "xfs_quota_defs.h"
 
 struct xfs_scrub;
 struct xfile;
@@ -86,7 +87,8 @@ TRACE_DEFINE_ENUM(XFS_SCRUB_TYPE_FSCOUNTERS);
 	{ XFS_SCRUB_TYPE_UQUOTA,	"usrquota" }, \
 	{ XFS_SCRUB_TYPE_GQUOTA,	"grpquota" }, \
 	{ XFS_SCRUB_TYPE_PQUOTA,	"prjquota" }, \
-	{ XFS_SCRUB_TYPE_FSCOUNTERS,	"fscounters" }
+	{ XFS_SCRUB_TYPE_FSCOUNTERS,	"fscounters" }, \
+	{ XFS_SCRUB_TYPE_QUOTACHECK,	"quotacheck" }
 
 #define XFS_SCRUB_FLAG_STRINGS \
 	{ XFS_SCRUB_IFLAG_REPAIR,		"repair" }, \
@@ -310,6 +312,29 @@ DEFINE_EVENT(xchk_fblock_error_class, name, \
 DEFINE_SCRUB_FBLOCK_ERROR_EVENT(xchk_fblock_error);
 DEFINE_SCRUB_FBLOCK_ERROR_EVENT(xchk_fblock_warning);
 
+TRACE_EVENT(xchk_qcheck_error,
+	TP_PROTO(struct xfs_scrub *sc, xfs_dqtype_t dqtype, xfs_dqid_t id,
+		 void *ret_ip),
+	TP_ARGS(sc, dqtype, id, ret_ip),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_dqtype_t, dqtype)
+		__field(xfs_dqid_t, id)
+		__field(void *, ret_ip)
+	),
+	TP_fast_assign(
+		__entry->dev = sc->mp->m_super->s_dev;
+		__entry->dqtype = dqtype;
+		__entry->id = id;
+		__entry->ret_ip = ret_ip;
+	),
+	TP_printk("dev %d:%d dquot type %s id 0x%x ret_ip %pS",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __print_symbolic(__entry->dqtype, XFS_DQTYPE_STRINGS),
+		  __entry->id,
+		  __entry->ret_ip)
+);
+
 TRACE_EVENT(xchk_incomplete,
 	TP_PROTO(struct xfs_scrub *sc, void *ret_ip),
 	TP_ARGS(sc, ret_ip),
diff --git a/fs/xfs/scrub/xfarray.h b/fs/xfs/scrub/xfarray.h
index d034ba9ca97f..56c0a1f66ca6 100644
--- a/fs/xfs/scrub/xfarray.h
+++ b/fs/xfs/scrub/xfarray.h
@@ -29,6 +29,25 @@ int xfarray_store_anywhere(struct xfarray *array, void *ptr);
 bool xfarray_is_null(struct xfarray *array, void *ptr);
 int xfarray_nullify(struct xfarray *array, uint64_t idx);
 
+/*
+ * Load an array element, but zero the buffer if there's no data because we
+ * haven't stored to that array element yet.
+ */
+static inline int
+xfarray_load_sparse(
+	struct xfarray	*array,
+	uint64_t	idx,
+	void		*rec)
+{
+	int		error = xfarray_load(array, idx, rec);
+
+	if (error == -ENODATA) {
+		memset(rec, 0, array->obj_size);
+		return 0;
+	}
+	return error;
+}
+
 /* Append an element to the array. */
 static inline int xfarray_append(struct xfarray *array, void *ptr)
 {
diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c
index 84ac64b41184..30f4b782d1e9 100644
--- a/fs/xfs/xfs_inode.c
+++ b/fs/xfs/xfs_inode.c
@@ -3798,3 +3798,24 @@ xfs_iunlock2_io_mmap(
 	if (ip1 != ip2)
 		inode_unlock(VFS_I(ip1));
 }
+
+/* Compute the number of data and realtime blocks used by a file. */
+void
+xfs_inode_count_blocks(
+	struct xfs_trans	*tp,
+	struct xfs_inode	*ip,
+	xfs_filblks_t		*dblocks,
+	xfs_filblks_t		*rblocks)
+{
+	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+
+	if (!XFS_IS_REALTIME_INODE(ip)) {
+		*dblocks = ip->i_nblocks;
+		*rblocks = 0;
+		return;
+	}
+
+	*rblocks = 0;
+	xfs_bmap_count_leaves(ifp, rblocks);
+	*dblocks = ip->i_nblocks - *rblocks;
+}
diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h
index c447bf04205a..d5edb563b77d 100644
--- a/fs/xfs/xfs_inode.h
+++ b/fs/xfs/xfs_inode.h
@@ -517,4 +517,7 @@ void xfs_end_io(struct work_struct *work);
 int xfs_ilock2_io_mmap(struct xfs_inode *ip1, struct xfs_inode *ip2);
 void xfs_iunlock2_io_mmap(struct xfs_inode *ip1, struct xfs_inode *ip2);
 
+void xfs_inode_count_blocks(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_filblks_t *dblocks, xfs_filblks_t *rblocks);
+
 #endif	/* __XFS_INODE_H__ */