Now, ext4_free_blocks() doesn't revoke data blocks of per-file data
journalled inode and it can cause file data inconsistency problems.
Even though data blocks of per-file data journalled inode are already
forgotten by jbd2_journal_invalidatepage() in advance of invoking
ext4_free_blocks(), we still need to revoke the data blocks here.
Moreover some of the metadata blocks, which are not found by
sb_find_get_block(), are still needed to be revoked, but this is also
missing here.
Signed-off-by: Daeho Jeong <[email protected]>
---
fs/ext4/mballoc.c | 29 ++++++++++++++---------------
1 file changed, 14 insertions(+), 15 deletions(-)
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 61eaf74..e40a2df 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -4695,16 +4695,6 @@ void ext4_free_blocks(handle_t *handle, struct inode *inode,
}
/*
- * We need to make sure we don't reuse the freed block until
- * after the transaction is committed, which we can do by
- * treating the block as metadata, below. We make an
- * exception if the inode is to be written in writeback mode
- * since writeback mode has weak data consistency guarantees.
- */
- if (!ext4_should_writeback_data(inode))
- flags |= EXT4_FREE_BLOCKS_METADATA;
-
- /*
* If the extent to be freed does not begin on a cluster
* boundary, we need to deal with partial clusters at the
* beginning and end of the extent. Normally we will free
@@ -4738,17 +4728,26 @@ void ext4_free_blocks(handle_t *handle, struct inode *inode,
if (!bh && (flags & EXT4_FREE_BLOCKS_FORGET)) {
int i;
+ int is_metadata = flags & EXT4_FREE_BLOCKS_METADATA;
for (i = 0; i < count; i++) {
cond_resched();
- bh = sb_find_get_block(inode->i_sb, block + i);
- if (!bh)
- continue;
- ext4_forget(handle, flags & EXT4_FREE_BLOCKS_METADATA,
- inode, bh, block + i);
+ if (is_metadata)
+ bh = sb_find_get_block(inode->i_sb, block + i);
+ ext4_forget(handle, is_metadata, inode, bh, block + i);
}
}
+ /*
+ * We need to make sure we don't reuse the freed block until
+ * after the transaction is committed, which we can do by
+ * treating the block as metadata, below. We make an
+ * exception if the inode is to be written in writeback mode
+ * since writeback mode has weak data consistency guarantees.
+ */
+ if (!ext4_should_writeback_data(inode))
+ flags |= EXT4_FREE_BLOCKS_METADATA;
+
do_more:
overflow = 0;
ext4_get_group_no_and_offset(sb, block, &block_group, &bit);
--
1.7.9.5
On Wed 16-12-15 09:20:43, Daeho Jeong wrote:
> Now, ext4_free_blocks() doesn't revoke data blocks of per-file data
> journalled inode and it can cause file data inconsistency problems.
> Even though data blocks of per-file data journalled inode are already
> forgotten by jbd2_journal_invalidatepage() in advance of invoking
> ext4_free_blocks(), we still need to revoke the data blocks here.
> Moreover some of the metadata blocks, which are not found by
> sb_find_get_block(), are still needed to be revoked, but this is also
> missing here.
Thanks for looking into the issue! Just one small nit below.
> Signed-off-by: Daeho Jeong <[email protected]>
> ---
> fs/ext4/mballoc.c | 29 ++++++++++++++---------------
> 1 file changed, 14 insertions(+), 15 deletions(-)
...
> + /*
> + * We need to make sure we don't reuse the freed block until
> + * after the transaction is committed, which we can do by
> + * treating the block as metadata, below. We make an
> + * exception if the inode is to be written in writeback mode
> + * since writeback mode has weak data consistency guarantees.
> + */
> + if (!ext4_should_writeback_data(inode))
> + flags |= EXT4_FREE_BLOCKS_METADATA;
> +
I think it would be clearer what's going on if your just modified the
condition checking whether EFD entry needs to be created. Like:
- if ((flags & EXT4_FREE_BLOCKS_METADATA) && ext4_handle_valid(handle)) {
+ /*
+ * We need to make sure we don't reuse the freed block until after the
+ * transaction is committed. We make an exception if the inode is to be
+ * written in writeback mode since writeback mode has weak data
+ * consistency guarantees.
+ */
+ if (ext4_handle_valid(handle) &&
+ ((flags & EXT4_FREE_BLOCKS_METADATA) ||
+ !ext4_should_writeback_data(inode))) {
struct ext4_free_data *new_entry;
/*
- * blocks being freed are metadata. these blocks shouldn't
- * be used until this transaction is committed
- *
* We use __GFP_NOFAIL because ext4_free_blocks() is not
* allowed
* to fail.
Otherwise the patch looks fine so after changing this feel free to add:
Reviewed-by: Jan Kara <[email protected]>
Honza
--
Jan Kara <[email protected]>
SUSE Labs, CR
It looks clearer. I will modify the condition as you mentioned.
Thank you for your review. :-)