btrfs: don't BUG_ON() in btrfs_drop_extents()
PATCH v2 1/1 btrfs: don't BUG_ON() in btrfs_drop_extents()
From: Johannes Thumshirn <[email protected]>
btrfs_drop_extents() calls BUG_ON() in case the counter of to be deleted extents is greater than 0. But all of these code paths can handle errors, so there's no need to crash the kernel. Instead WARN() that the condition has been met and gracefully bail out.
---
Changes to v1:
- Fix spelling error in commit message
- Change ASSERT() to WARN_ON()
- Take care of the other BUG_ON() cases as well
Link to v1:
- https://lore.kernel.org/linux-btrfs/[email protected]
--- fs/btrfs/file.c | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-)
fs/btrfs/file.c +16 -4
@@ -245,7 +245,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans, next_slot: leaf = path->nodes[0 ];if (path->slots[0 ] >=btrfs_header_nritems (leaf)) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ; +break ; + } ret =btrfs_next_leaf (root, path);if (ret <0 )break ;
@@ -321,7 +324,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans,* | -------- extent -------- | */ if (args->start > key.offset && args->end < extent_end) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ; +break ; + }if (extent_type ==BTRFS_FILE_EXTENT_INLINE ) { ret = -EOPNOTSUPP ;break ;
@@ -409,7 +415,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans,* | -------- extent -------- | */ if (args->start > key.offset && args->end >= extent_end) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ; +break ; + }if (extent_type ==BTRFS_FILE_EXTENT_INLINE ) { ret = -EOPNOTSUPP ;break ;
@@ -437,7 +446,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans, del_slot = path->slots[0 ]; del_nr =1 ; }else { -BUG_ON (del_slot + del_nr != path->slots[0 ]); +if (WARN_ON (del_slot + del_nr != path->slots[0 ])) { + ret = -EINVAL ; +break ; + } del_nr++; }
Signature · 1 line
2.43.0
Quoted 9 lines · Johannes Thumshirn
From: Johannes Thumshirn <[email protected]>
btrfs_drop_extents()callsBUG_ON()in case the counter of to be deleted extents is greater than 0. But all of these code paths can handle errors, so there's no need to crash the kernel. InsteadWARN()that the condition has been met and gracefully bail out.Signed-off-by Johannes Thumshirn <[email protected]>
Looks good, thanks.
Quoted 68 lines
---
Changes to v1:
- Fix spelling error in commit message
- ChangeASSERT()toWARN_ON()
- Take care of the otherBUG_ON()cases as wellLink to v1:
- https://lore.kernel.org/linux-btrfs/[email protected]--- fs/btrfs/file.c | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-)fs/btrfs/file.c +16
-5@@ -245,7 +245,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans, next_slot: leaf = path->nodes[0 ];if (path->slots[0 ] >=btrfs_header_nritems (leaf)) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ; +break ; + } ret =btrfs_next_leaf (root, path);if (ret <0 )break ;@@ -321,7 +324,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans,* | -------- extent -------- | */ if (args->start > key.offset && args->end < extent_end) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ; +break ; + }if (extent_type ==BTRFS_FILE_EXTENT_INLINE ) { ret = -EOPNOTSUPP ;break ;@@ -409,7 +415,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans,* | -------- extent -------- | */ if (args->start > key.offset && args->end >= extent_end) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ; +break ; + }if (extent_type ==BTRFS_FILE_EXTENT_INLINE ) { ret = -EOPNOTSUPP ;break ;@@ -437,7 +446,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans, del_slot = path->slots[0 ]; del_nr =1 ; }else { -BUG_ON (del_slot + del_nr != path->slots[0 ]); +if (WARN_ON (del_slot + del_nr != path->slots[0 ])) { + ret = -EINVAL ; +break ; + } del_nr++; } --2.43.0
Quoted 37 lines · Johannes Thumshirn
From: Johannes Thumshirn <[email protected]>
btrfs_drop_extents()callsBUG_ON()in case the counter of to be deleted extents is greater than 0. But all of these code paths can handle errors, so there's no need to crash the kernel. InsteadWARN()that the condition has been met and gracefully bail out.Signed-off-by Johannes Thumshirn <[email protected]>---
Changes to v1:
- Fix spelling error in commit message
- ChangeASSERT()toWARN_ON()
- Take care of the otherBUG_ON()cases as wellLink to v1:
- https://lore.kernel.org/linux-btrfs/[email protected]--- fs/btrfs/file.c | 20 ++++++++++++++++---- 1 file changed, 16 insertions(+), 4 deletions(-)fs/btrfs/file.c +4
-1@@ -245,7 +245,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans, next_slot: leaf = path->nodes[0 ];if (path->slots[0 ] >=btrfs_header_nritems (leaf)) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ; +break ; + } ret =btrfs_next_leaf (root, path);if (ret <0 )break ;
fs/btrfs/file.c int btrfs_drop_extents(struct btrfs_trans_handle *trans, · line 328
Quoted 6 lines
fs/btrfs/file.c +2
-1@@ -321,7 +324,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans,* | -------- extent -------- | */ if (args->start > key.offset && args->end < extent_end) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ;
Do we also want to dump the leaf and output a more readable error message?
Just like what we did inside lookup_inlin_extent_backref()/update_inline_extent_backref()/insert_inline_extent_backref()/__btrfs_free_extent()/...?
Thanks,
Qu
Quoted 29 lines
fs/btrfs/file.c +10
-2+break ; + }if (extent_type ==BTRFS_FILE_EXTENT_INLINE ) { ret = -EOPNOTSUPP ;break ;@@ -409,7 +415,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans,* | -------- extent -------- | */ if (args->start > key.offset && args->end >= extent_end) { -BUG_ON (del_nr >0 ); +if (WARN_ON (del_nr >0 )) { + ret = -EINVAL ; +break ; + }if (extent_type ==BTRFS_FILE_EXTENT_INLINE ) { ret = -EOPNOTSUPP ;break ;@@ -437,7 +446,10 @@ int btrfs_drop_extents(struct btrfs_trans_handle *trans, del_slot = path->slots[0 ]; del_nr =1 ; }else { -BUG_ON (del_slot + del_nr != path->slots[0 ]); +if (WARN_ON (del_slot + del_nr != path->slots[0 ])) { + ret = -EINVAL ; +break ; + } del_nr++; }