Skip to contentSkip to conversation

xfs: count free space btree blocks when scrubbing pre-lazysbcount fses

PATCH In review Darrick J. Wong 3 participants 3 messages Apr 27, 2021 XFS filesystem

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 xfs: count free space btree blocks when scrubbing pre-lazysbcount fses

From: Darrick J. Wong <[email protected]>

Since agf_btreeblks didn't exist before the lazysbcount feature, the fs summary count scrubber needs to walk the free space btrees to determine the amount of space being used by those btrees.

---
Note: I /think/ the four patches on the list right now fix all the obvious problems with !lazysbcount filesystems, except for xfs/49[12] which fuzz the summary counters.

---
 fs/xfs/scrub/fscounters.c |   37 +++++++++++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
fs/xfs/scrub/fscounters.c +37 -0
index 318b81c0f90d..87476a00de9d 100644
@@ -13,6 +13,7 @@
 #include "xfs_alloc.h"
 #include "xfs_ialloc.h"
 #include "xfs_health.h"
+#include "xfs_btree.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/trace.h"
@@ -143,6 +144,35 @@ xchk_setup_fscounters(
 	return xchk_trans_alloc(sc, 0);
 }
 
+/* Count free space btree blocks manually for pre-lazysbcount filesystems. */
+static int
+xchk_fscount_btreeblks(
+	struct xfs_scrub	*sc,
+	struct xchk_fscounters	*fsc,
+	xfs_agnumber_t		agno)
+{
+	xfs_extlen_t		blocks;
+	int			error;
+
+	error = xchk_ag_init(sc, agno, &sc->sa);
+	if (error)
+		return error;
+
+	error = xfs_btree_count_blocks(sc->sa.bno_cur, &blocks);
+	if (error)
+		goto out_free;
+	fsc->fdblocks += blocks - 1;
+
+	error = xfs_btree_count_blocks(sc->sa.cnt_cur, &blocks);
+	if (error)
+		goto out_free;
+	fsc->fdblocks += blocks - 1;
+
+out_free:
+	xchk_ag_free(sc, &sc->sa);
+	return error;
+}
+
 /*
  * Calculate what the global in-core counters ought to be from the incore
  * per-AG structure.  Callers can compare this to the actual in-core counters
@@ -184,6 +214,13 @@ xchk_fscount_aggregate_agcounts(
 		fsc->fdblocks += pag->pagf_flcount;
 		if (xfs_sb_version_haslazysbcount(&sc->mp->m_sb))
 			fsc->fdblocks += pag->pagf_btreeblks;
+		else {
+			error = xchk_fscount_btreeblks(sc, fsc, agno);
+			if (error) {
+				xfs_perag_put(pag);
+				break;
+			}
+		}
 
 		/*
 		 * Per-AG reservations are taken out of the incore counters,
Reviewed-by on v1 1/1
#1 lore raw
Quoted 8 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

Since agf_btreeblks didn't exist before the lazysbcount feature, the fs summary count scrubber needs to walk the free space btrees to determine the amount of space being used by those btrees.

---

Quoted 70 lines

Note: I /think/ the four patches on the list right now fix all the obvious problems with !lazysbcount filesystems, except for xfs/49[12] which fuzz the summary counters.

---
 fs/xfs/scrub/fscounters.c |   37 +++++++++++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
fs/xfs/scrub/fscounters.c +37 -0
index 318b81c0f90d..87476a00de9d 100644
@@ -13,6 +13,7 @@
 #include "xfs_alloc.h"
 #include "xfs_ialloc.h"
 #include "xfs_health.h"
+#include "xfs_btree.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/trace.h"
@@ -143,6 +144,35 @@ xchk_setup_fscounters(
 	return xchk_trans_alloc(sc, 0);
 }
 
+/* Count free space btree blocks manually for pre-lazysbcount filesystems. */
+static int
+xchk_fscount_btreeblks(
+	struct xfs_scrub	*sc,
+	struct xchk_fscounters	*fsc,
+	xfs_agnumber_t		agno)
+{
+	xfs_extlen_t		blocks;
+	int			error;
+
+	error = xchk_ag_init(sc, agno, &sc->sa);
+	if (error)
+		return error;
+
+	error = xfs_btree_count_blocks(sc->sa.bno_cur, &blocks);
+	if (error)
+		goto out_free;
+	fsc->fdblocks += blocks - 1;
+
+	error = xfs_btree_count_blocks(sc->sa.cnt_cur, &blocks);
+	if (error)
+		goto out_free;
+	fsc->fdblocks += blocks - 1;
+
+out_free:
+	xchk_ag_free(sc, &sc->sa);
+	return error;
+}
+
 /*
  * Calculate what the global in-core counters ought to be from the incore
  * per-AG structure.  Callers can compare this to the actual in-core counters
@@ -184,6 +214,13 @@ xchk_fscount_aggregate_agcounts(
 		fsc->fdblocks += pag->pagf_flcount;
 		if (xfs_sb_version_haslazysbcount(&sc->mp->m_sb))
 			fsc->fdblocks += pag->pagf_btreeblks;
+		else {
+			error = xchk_fscount_btreeblks(sc, fsc, agno);
+			if (error) {
+				xfs_perag_put(pag);
+				break;
+			}
+		}
 
 		/*
 		 * Per-AG reservations are taken out of the incore counters,
Reviewed-by on v1 1/1
#2 lore raw
In reply to Darrick J. Wong
Quoted 7 lines · Darrick J. Wong

From: Darrick J. Wong <[email protected]>

Since agf_btreeblks didn't exist before the lazysbcount feature, the fs summary count scrubber needs to walk the free space btrees to determine the amount of space being used by those btrees.

Don't look into scrubbing code too much, but the code here looks fine to me:

Thanks,
Gao Xiang