Skip to contentSkip to conversation

ocfs2: consolidate extent list validation into block read callbacks

PATCH Applied Joseph Qi 3 participants 11 messages Apr 3 – Apr 7 OCFS2

lore
Share & tools
lore

Joseph Qi

author maintainer
#0 lore raw

cover ocfs2: consolidate extent list validation into block read callbacks

ocfs2 validates extent list fields (l_count, l_next_free_rec) at various points during extent tree traversal. This is fragile because each caller must remember to check for corrupted on-disk data before using it.

This series moves those checks into the block read validation callbacks (ocfs2_validate_dx_root and ocfs2_validate_extent_block), so corrupted fields are caught early at block read time. Redundant post-read checks are then removed.

Patch 1-2 handle the dx_root extent list:
- Move l_count and l_next_free_rec validation into ocfs2_validate_dx_root(), skipping inline dx roots.
- Remove the now-redundant l_next_free_rec == 0 check in ocfs2_dx_dir_lookup_rec() and rework the error message to avoid a potential NULL pointer dereference of rec.

Patch 3-4 handle extent blocks:
- Add l_count and l_next_free_rec validation to ocfs2_validate_extent_block().
- Remove the duplicate l_next_free_rec > l_count check in

    __ocfs2_find_path().

Joseph Qi (4):
ocfs2: validate dx_root extent list fields during block read
ocfs2: remove empty extent list check in ocfs2_dx_dir_lookup_rec()
ocfs2: validate extent block list fields during block read
ocfs2: remove redundant l_next_free_rec check in __ocfs2_find_path()

 fs/ocfs2/alloc.c | 35 +++++++++++++++++++++-------------
 fs/ocfs2/dir.c   | 49 +++++++++++++++++++++++++++---------------------
 2 files changed, 50 insertions(+), 34 deletions(-)
Signature · 1 line
2.39.3
author maintainer
#1 lore raw

PATCH 1/4 ocfs2: validate dx_root extent list fields during block read

Move the extent list l_count validation from ocfs2_dx_dir_lookup_rec() into ocfs2_validate_dx_root(), so that corrupted on-disk fields are caught early at block read time rather than during directory lookups.

Additionally, add a l_next_free_rec <= l_count check to prevent out-of-bounds access when iterating over extent records.

Both checks are skipped for inline dx roots (OCFS2_DX_FLAG_INLINE), which use dr_entries instead of dr_list.

---
 fs/ocfs2/dir.c | 34 +++++++++++++++++++++++++---------
 1 file changed, 25 insertions(+), 9 deletions(-)
fs/ocfs2/dir.c +25 -9
index 1c8abf2c592c..82e720c8ba32 100644
@@ -593,7 +593,7 @@ static int ocfs2_validate_dx_root(struct super_block *sb,
 		mlog(ML_ERROR,
 		     "Checksum failed for dir index root block %llu\n",
 		     (unsigned long long)bh->b_blocknr);
-		return ret;
+		goto bail;
 	}
 
 	if (!OCFS2_IS_VALID_DX_ROOT(dx_root)) {
@@ -601,8 +601,32 @@ static int ocfs2_validate_dx_root(struct super_block *sb,
 				  "Dir Index Root # %llu has bad signature %.*s\n",
 				  (unsigned long long)le64_to_cpu(dx_root->dr_blkno),
 				  7, dx_root->dr_signature);
+		goto bail;
+	}
+
+	if (!(dx_root->dr_flags & OCFS2_DX_FLAG_INLINE)) {
+		struct ocfs2_extent_list *el = &dx_root->dr_list;
+
+		if (le16_to_cpu(el->l_count) != ocfs2_extent_recs_per_dx_root(sb)) {
+			ret = ocfs2_error(sb,
+					  "Dir Index Root # %llu has invalid l_count %u (expected %u)\n",
+					  (unsigned long long)le64_to_cpu(dx_root->dr_blkno),
+					  le16_to_cpu(el->l_count),
+					  ocfs2_extent_recs_per_dx_root(sb));
+			goto bail;
+		}
+
+		if (le16_to_cpu(el->l_next_free_rec) > le16_to_cpu(el->l_count)) {
+			ret = ocfs2_error(sb,
+					  "Dir Index Root # %llu has invalid l_next_free_rec %u (l_count %u)\n",
+					  (unsigned long long)le64_to_cpu(dx_root->dr_blkno),
+					  le16_to_cpu(el->l_next_free_rec),
+					  le16_to_cpu(el->l_count));
+			goto bail;
+		}
 	}
 
+bail:
 	return ret;
 }
 
@@ -791,14 +815,6 @@ static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
 	struct ocfs2_extent_block *eb;
 	struct ocfs2_extent_rec *rec = NULL;
 
-	if (le16_to_cpu(el->l_count) !=
-	    ocfs2_extent_recs_per_dx_root(inode->i_sb)) {
-		ret = ocfs2_error(inode->i_sb,
-				  "Inode %lu has invalid extent list length %u\n",
-				  inode->i_ino, le16_to_cpu(el->l_count));
-		goto out;
-	}
-
 	if (el->l_tree_depth) {
 		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, major_hash,
 				      &eb_bh);
Signature · 1 line
2.39.3
reviewer Reviewed-by on v1 1/4
#2 lore raw

Re: [PATCH 1/4] ocfs2: validate dx_root extent list fields during block read

Quoted 11 lines · Joseph Qi

Move the extent list l_count validation from ocfs2_dx_dir_lookup_rec() into ocfs2_validate_dx_root(), so that corrupted on-disk fields are caught early at block read time rather than during directory lookups.

Additionally, add a l_next_free_rec <= l_count check to prevent out-of-bounds access when iterating over extent records.

Both checks are skipped for inline dx roots (OCFS2_DX_FLAG_INLINE), which use dr_entries instead of dr_list.

LGTM

Quoted 68 lines
---
 fs/ocfs2/dir.c | 34 +++++++++++++++++++++++++---------
 1 file changed, 25 insertions(+), 9 deletions(-)
fs/ocfs2/dir.c +25 -9
index 1c8abf2c592c..82e720c8ba32 100644
@@ -593,7 +593,7 @@ static int ocfs2_validate_dx_root(struct super_block *sb,
 		mlog(ML_ERROR,
 		     "Checksum failed for dir index root block %llu\n",
 		     (unsigned long long)bh->b_blocknr);
-		return ret;
+		goto bail;
 	}
 
 	if (!OCFS2_IS_VALID_DX_ROOT(dx_root)) {
@@ -601,8 +601,32 @@ static int ocfs2_validate_dx_root(struct super_block *sb,
 				  "Dir Index Root # %llu has bad signature %.*s\n",
 				  (unsigned long long)le64_to_cpu(dx_root->dr_blkno),
 				  7, dx_root->dr_signature);
+		goto bail;
+	}
+
+	if (!(dx_root->dr_flags & OCFS2_DX_FLAG_INLINE)) {
+		struct ocfs2_extent_list *el = &dx_root->dr_list;
+
+		if (le16_to_cpu(el->l_count) != ocfs2_extent_recs_per_dx_root(sb)) {
+			ret = ocfs2_error(sb,
+					  "Dir Index Root # %llu has invalid l_count %u (expected %u)\n",
+					  (unsigned long long)le64_to_cpu(dx_root->dr_blkno),
+					  le16_to_cpu(el->l_count),
+					  ocfs2_extent_recs_per_dx_root(sb));
+			goto bail;
+		}
+
+		if (le16_to_cpu(el->l_next_free_rec) > le16_to_cpu(el->l_count)) {
+			ret = ocfs2_error(sb,
+					  "Dir Index Root # %llu has invalid l_next_free_rec %u (l_count %u)\n",
+					  (unsigned long long)le64_to_cpu(dx_root->dr_blkno),
+					  le16_to_cpu(el->l_next_free_rec),
+					  le16_to_cpu(el->l_count));
+			goto bail;
+		}
 	}
 
+bail:
 	return ret;
 }
 
@@ -791,14 +815,6 @@ static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
 	struct ocfs2_extent_block *eb;
 	struct ocfs2_extent_rec *rec = NULL;
 
-	if (le16_to_cpu(el->l_count) !=
-	    ocfs2_extent_recs_per_dx_root(inode->i_sb)) {
-		ret = ocfs2_error(inode->i_sb,
-				  "Inode %lu has invalid extent list length %u\n",
-				  inode->i_ino, le16_to_cpu(el->l_count));
-		goto out;
-	}
-
 	if (el->l_tree_depth) {
 		ret = ocfs2_find_leaf(INODE_CACHE(inode), el, major_hash,
 				      &eb_bh);
Signature · 1 line
2.39.3
author maintainer
#3 lore raw

PATCH 2/4 ocfs2: remove empty extent list check in ocfs2_dx_dir_lookup_rec()

In reply to Joseph Qi

The full extent list check is introduced by commit 44acc46d182f, which is to avoid NULL pointer dereference if a dirent is not found.

Reworking the error message to not reference rec. Instead, report major_hash being looked up and l_next_free_rec, which naturally covers both failure cases (empty extent list and no matching record) without needing a separate l_next_free_rec == 0 guard.

---
 fs/ocfs2/dir.c | 15 +++------------
 1 file changed, 3 insertions(+), 12 deletions(-)
fs/ocfs2/dir.c +3 -12
index 82e720c8ba32..d94595a49923 100644
@@ -835,14 +835,6 @@ static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
 		}
 	}
 
-	if (le16_to_cpu(el->l_next_free_rec) == 0) {
-		ret = ocfs2_error(inode->i_sb,
-				  "Inode %lu has empty extent list at depth %u\n",
-				  inode->i_ino,
-				  le16_to_cpu(el->l_tree_depth));
-		goto out;
-	}
-
 	found = 0;
 	for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
 		rec = &el->l_recs[i];
@@ -855,10 +847,9 @@ static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
 
 	if (!found) {
 		ret = ocfs2_error(inode->i_sb,
-				  "Inode %lu has bad extent record (%u, %u, 0) in btree\n",
-				  inode->i_ino,
-				  le32_to_cpu(rec->e_cpos),
-				  ocfs2_rec_clusters(el, rec));
+				  "Inode %lu has no extent record for hash %u in btree (next_free_rec %u)\n",
+				  inode->i_ino, major_hash,
+				  le16_to_cpu(el->l_next_free_rec));
 		goto out;
 	}
 
Signature · 1 line
2.39.3
reviewer Reviewed-by on v1 2/4
#4 lore raw

Re: [PATCH 2/4] ocfs2: remove empty extent list check in ocfs2_dx_dir_lookup_rec()

Quoted 9 lines · Joseph Qi

The full extent list check is introduced by commit 44acc46d182f, which is to avoid NULL pointer dereference if a dirent is not found.

Reworking the error message to not reference rec. Instead, report major_hash being looked up and l_next_free_rec, which naturally covers both failure cases (empty extent list and no matching record) without needing a separate l_next_free_rec == 0 guard.

LGTM

Quoted 40 lines
---
 fs/ocfs2/dir.c | 15 +++------------
 1 file changed, 3 insertions(+), 12 deletions(-)
fs/ocfs2/dir.c +3 -12
index 82e720c8ba32..d94595a49923 100644
@@ -835,14 +835,6 @@ static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
 		}
 	}
 
-	if (le16_to_cpu(el->l_next_free_rec) == 0) {
-		ret = ocfs2_error(inode->i_sb,
-				  "Inode %lu has empty extent list at depth %u\n",
-				  inode->i_ino,
-				  le16_to_cpu(el->l_tree_depth));
-		goto out;
-	}
-
 	found = 0;
 	for (i = le16_to_cpu(el->l_next_free_rec) - 1; i >= 0; i--) {
 		rec = &el->l_recs[i];
@@ -855,10 +847,9 @@ static int ocfs2_dx_dir_lookup_rec(struct inode *inode,
 
 	if (!found) {
 		ret = ocfs2_error(inode->i_sb,
-				  "Inode %lu has bad extent record (%u, %u, 0) in btree\n",
-				  inode->i_ino,
-				  le32_to_cpu(rec->e_cpos),
-				  ocfs2_rec_clusters(el, rec));
+				  "Inode %lu has no extent record for hash %u in btree (next_free_rec %u)\n",
+				  inode->i_ino, major_hash,
+				  le16_to_cpu(el->l_next_free_rec));
 		goto out;
 	}
 
Signature · 1 line
2.39.3
author maintainer
#5 lore raw

PATCH 3/4 ocfs2: validate extent block list fields during block read

In reply to Joseph Qi

Add extent list validation to ocfs2_validate_extent_block() so that corrupted on-disk fields are caught early at block read time rather than during extent tree traversal.

Two checks are added:

- l_count must equal the expected value from ocfs2_extent_recs_per_eb(), catching blocks with a corrupted record count before any array iteration.

- l_next_free_rec must not exceed l_count, preventing out-of-bounds access when iterating over extent records.

---
 fs/ocfs2/alloc.c | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)
fs/ocfs2/alloc.c +22 -1
index 344fd4d95fbc..8639806bcbb8 100644
@@ -917,11 +917,32 @@ static int ocfs2_validate_extent_block(struct super_block *sb,
 		goto bail;
 	}
 
-	if (le32_to_cpu(eb->h_fs_generation) != OCFS2_SB(sb)->fs_generation)
+	if (le32_to_cpu(eb->h_fs_generation) != OCFS2_SB(sb)->fs_generation) {
 		rc = ocfs2_error(sb,
 				 "Extent block #%llu has an invalid h_fs_generation of #%u\n",
 				 (unsigned long long)bh->b_blocknr,
 				 le32_to_cpu(eb->h_fs_generation));
+		goto bail;
+	}
+
+	if (le16_to_cpu(eb->h_list.l_count) != ocfs2_extent_recs_per_eb(sb)) {
+		rc = ocfs2_error(sb,
+				 "Extent block #%llu has invalid l_count %u (expected %u)\n",
+				 (unsigned long long)bh->b_blocknr,
+				 le16_to_cpu(eb->h_list.l_count),
+				 ocfs2_extent_recs_per_eb(sb));
+		goto bail;
+	}
+
+	if (le16_to_cpu(eb->h_list.l_next_free_rec) > le16_to_cpu(eb->h_list.l_count)) {
+		rc = ocfs2_error(sb,
+				 "Extent block #%llu has invalid l_next_free_rec %u (l_count %u)\n",
+				 (unsigned long long)bh->b_blocknr,
+				 le16_to_cpu(eb->h_list.l_next_free_rec),
+				 le16_to_cpu(eb->h_list.l_count));
+		goto bail;
+	}
+
 bail:
 	return rc;
 }
Signature · 1 line
2.39.3
reviewer Reviewed-by on v1 3/4
#6 lore raw

Re: [PATCH 3/4] ocfs2: validate extent block list fields during block read

Quoted 6 lines · Joseph Qi

Add extent list validation to ocfs2_validate_extent_block() so that corrupted on-disk fields are caught early at block read time rather than during extent tree traversal.

Two checks are added:

Quoted 8 lines

- l_count must equal the expected value from ocfs2_extent_recs_per_eb(), catching blocks with a corrupted record count before any array iteration.

- l_next_free_rec must not exceed l_count, preventing out-of-bounds access when iterating over extent records.

LGTM

Quoted 45 lines
---
 fs/ocfs2/alloc.c | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)
fs/ocfs2/alloc.c +22 -1
index 344fd4d95fbc..8639806bcbb8 100644
@@ -917,11 +917,32 @@ static int ocfs2_validate_extent_block(struct super_block *sb,
 		goto bail;
 	}
 
-	if (le32_to_cpu(eb->h_fs_generation) != OCFS2_SB(sb)->fs_generation)
+	if (le32_to_cpu(eb->h_fs_generation) != OCFS2_SB(sb)->fs_generation) {
 		rc = ocfs2_error(sb,
 				 "Extent block #%llu has an invalid h_fs_generation of #%u\n",
 				 (unsigned long long)bh->b_blocknr,
 				 le32_to_cpu(eb->h_fs_generation));
+		goto bail;
+	}
+
+	if (le16_to_cpu(eb->h_list.l_count) != ocfs2_extent_recs_per_eb(sb)) {
+		rc = ocfs2_error(sb,
+				 "Extent block #%llu has invalid l_count %u (expected %u)\n",
+				 (unsigned long long)bh->b_blocknr,
+				 le16_to_cpu(eb->h_list.l_count),
+				 ocfs2_extent_recs_per_eb(sb));
+		goto bail;
+	}
+
+	if (le16_to_cpu(eb->h_list.l_next_free_rec) > le16_to_cpu(eb->h_list.l_count)) {
+		rc = ocfs2_error(sb,
+				 "Extent block #%llu has invalid l_next_free_rec %u (l_count %u)\n",
+				 (unsigned long long)bh->b_blocknr,
+				 le16_to_cpu(eb->h_list.l_next_free_rec),
+				 le16_to_cpu(eb->h_list.l_count));
+		goto bail;
+	}
+
 bail:
 	return rc;
 }
Signature · 1 line
2.39.3
author maintainer
#7 lore raw

PATCH 4/4 ocfs2: remove redundant l_next_free_rec check in __ocfs2_find_path()

In reply to Joseph Qi

The l_next_free_rec > l_count check after ocfs2_read_extent_block() in __ocfs2_find_path() is now redundant, as ocfs2_validate_extent_block() already performs this validation at block read time.

Remove the duplicate check to avoid maintaining the same validation in two places.

---
 fs/ocfs2/alloc.c | 12 ------------
 1 file changed, 12 deletions(-)
fs/ocfs2/alloc.c +0 -12
index 8639806bcbb8..08e137928080 100644
@@ -1878,18 +1878,6 @@ static int __ocfs2_find_path(struct ocfs2_caching_info *ci,
 		eb = (struct ocfs2_extent_block *) bh->b_data;
 		el = &eb->h_list;
 
-		if (le16_to_cpu(el->l_next_free_rec) >
-		    le16_to_cpu(el->l_count)) {
-			ocfs2_error(ocfs2_metadata_cache_get_super(ci),
-				    "Owner %llu has bad count in extent list at block %llu (next free=%u, count=%u)\n",
-				    (unsigned long long)ocfs2_metadata_cache_owner(ci),
-				    (unsigned long long)bh->b_blocknr,
-				    le16_to_cpu(el->l_next_free_rec),
-				    le16_to_cpu(el->l_count));
-			ret = -EROFS;
-			goto out;
-		}
-
 		if (func)
 			func(data, bh);
 	}
Signature · 1 line
2.39.3
reviewer Reviewed-by on v1 4/4
#8 lore raw

Re: [PATCH 4/4] ocfs2: remove redundant l_next_free_rec check in __ocfs2_find_path()

Quoted 8 lines · Joseph Qi

The l_next_free_rec > l_count check after ocfs2_read_extent_block() in __ocfs2_find_path() is now redundant, as ocfs2_validate_extent_block() already performs this validation at block read time.

Remove the duplicate check to avoid maintaining the same validation in two places.

LGTM

Quoted 30 lines
---
 fs/ocfs2/alloc.c | 12 ------------
 1 file changed, 12 deletions(-)
fs/ocfs2/alloc.c +0 -12
index 8639806bcbb8..08e137928080 100644
@@ -1878,18 +1878,6 @@ static int __ocfs2_find_path(struct ocfs2_caching_info *ci,
 		eb = (struct ocfs2_extent_block *) bh->b_data;
 		el = &eb->h_list;
 
-		if (le16_to_cpu(el->l_next_free_rec) >
-		    le16_to_cpu(el->l_count)) {
-			ocfs2_error(ocfs2_metadata_cache_get_super(ci),
-				    "Owner %llu has bad count in extent list at block %llu (next free=%u, count=%u)\n",
-				    (unsigned long long)ocfs2_metadata_cache_owner(ci),
-				    (unsigned long long)bh->b_blocknr,
-				    le16_to_cpu(el->l_next_free_rec),
-				    le16_to_cpu(el->l_count));
-			ret = -EROFS;
-			goto out;
-		}
-
 		if (func)
 			func(data, bh);
 	}
Signature · 1 line
2.39.3
#9 lore raw
In reply to Joseph Qi
Quoted 8 lines · On Fri, 3 Apr 2026 17:07:59 +0800 Joseph Qi <[email protected]>…

ocfs2 validates extent list fields (l_count, l_next_free_rec) at various points during extent tree traversal. This is fragile because each caller must remember to check for corrupted on-disk data before using it.

This series moves those checks into the block read validation callbacks (ocfs2_validate_dx_root and ocfs2_validate_extent_block), so corrupted fields are caught early at block read time. Redundant post-read checks are then removed.

AI review asked a few questions - mainly suggesting additional changes.
https://sashiko.dev/#/patchset/[email protected]

author maintainer
#10 lore raw
Quoted 6 lines · Andrew Morton
Quoted 4 lines · On Fri, 3 Apr 2026 17:07:59 +0800 Joseph Qi <[email protected]>…

ocfs2 validates extent list fields (l_count, l_next_free_rec) at various points during extent tree traversal. This is fragile because each caller must remember to check for corrupted on-disk data before using it.

Quoted 7 lines
Quoted 4 lines

This series moves those checks into the block read validation callbacks (ocfs2_validate_dx_root and ocfs2_validate_extent_block), so corrupted fields are caught early at block read time. Redundant post-read checks are then removed.

AI review asked a few questions - mainly suggesting additional changes.
https://sashiko.dev/#/patchset/[email protected]

Yes, not regressions, actually some improvements.

Thanks,
Joseph