Grab the current mapping->wb_err when linking a transaction to the list
and stash it in the journal inode. Then we can use that as a "since"
value when committing it to ensure that there were no writeback errors
since the transaction was started.
We do still need to perform old-style error handling too for now in
journal_finish_inode_data_buffers. jbd2 is shared infrastructure between
several filesystems. Eventually we should be able to remove the flag check
and simplify this function again.
For journal recovery, sample the wb_err early on and then pass that as
the since value to sync_blockdev_since.
Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxx>
---
fs/jbd2/commit.c | 29 +++++++++++++++++++----------
fs/jbd2/recovery.c | 5 +++--
fs/jbd2/transaction.c | 1 +
include/linux/jbd2.h | 3 +++
4 files changed, 26 insertions(+), 12 deletions(-)
diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index b6b194ec1b4f..aea71e4bc9be 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -259,21 +259,30 @@ static int journal_finish_inode_data_buffers(journal_t *journal,
/* For locking, see the comment in journal_submit_data_buffers() */
spin_lock(&journal->j_list_lock);
list_for_each_entry(jinode, &commit_transaction->t_inode_list, i_list) {
+ struct inode *inode = jinode->i_vfs_inode;
+
if (!(jinode->i_flags & JI_WAIT_DATA))
continue;
jinode->i_flags |= JI_COMMIT_RUNNING;
spin_unlock(&journal->j_list_lock);
- err = filemap_fdatawait(jinode->i_vfs_inode->i_mapping);
- if (err) {
- /*
- * Because AS_EIO is cleared by
- * filemap_fdatawait_range(), set it again so
- * that user process can get -EIO from fsync().
- */
- mapping_set_error(jinode->i_vfs_inode->i_mapping, -EIO);
-
- if (!ret)
+ if (inode->i_sb->s_type->fs_flags & FS_WB_ERRSEQ) {
+ err = filemap_fdatawait_since(inode->i_mapping,
+ jinode->i_since);
+ if (err && !ret)
ret = err;
+ } else {
+ err = filemap_fdatawait(inode->i_mapping);
+ if (err) {
+ /*
+ * Because AS_EIO is cleared by
+ * filemap_fdatawait_range(), we must set it again so
+ * that user process can get -EIO from fsync() if
+ * non-errseq_t based error tracking is in play.
+ */
+ mapping_set_error(inode->i_mapping, -EIO);
+ if (!ret)
+ ret = err;
+ }
}
spin_lock(&journal->j_list_lock);
jinode->i_flags &= ~JI_COMMIT_RUNNING;
diff --git a/fs/jbd2/recovery.c b/fs/jbd2/recovery.c
index 02dd3360cb20..06a8ee71848c 100644
--- a/fs/jbd2/recovery.c
+++ b/fs/jbd2/recovery.c
@@ -248,11 +248,12 @@ int jbd2_journal_recover(journal_t *journal)
{
int err, err2;
journal_superblock_t * sb;
-
struct recovery_info info;
+ errseq_t since;
memset(&info, 0, sizeof(info));
sb = journal->j_superblock;
+ since = filemap_sample_wb_err(journal->j_fs_dev->bd_inode->i_mapping);
/*
* The journal superblock's s_start field (the current log head)
@@ -284,7 +285,7 @@ int jbd2_journal_recover(journal_t *journal)
journal->j_transaction_sequence = ++info.end_transaction;
jbd2_journal_clear_revoke(journal);
- err2 = sync_blockdev(journal->j_fs_dev);
+ err2 = sync_blockdev_since(journal->j_fs_dev, since);
if (!err)
err = err2;
/* Make sure all replayed data is on permanent storage */
diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index 9ee4832b6f8b..e9e6af20a087 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -2535,6 +2535,7 @@ static int jbd2_journal_file_inode(handle_t *handle, struct jbd2_inode *jinode,
/* Not on any transaction list... */
J_ASSERT(!jinode->i_next_transaction);
jinode->i_transaction = transaction;
+ jinode->i_since = filemap_sample_wb_err(jinode->i_vfs_inode->i_mapping);
list_add(&jinode->i_list, &transaction->t_inode_list);
done:
spin_unlock(&journal->j_list_lock);
diff --git a/include/linux/jbd2.h b/include/linux/jbd2.h
index 606b6bce3a5b..b6901eac2d8e 100644
--- a/include/linux/jbd2.h
+++ b/include/linux/jbd2.h
@@ -439,6 +439,9 @@ struct jbd2_inode {
/* Flags of inode [j_list_lock] */
unsigned long i_flags;
+
+ /* Sampled writeback error at the time of transaction start */
+ errseq_t i_since;
};
struct jbd2_revoke_table_s;
--
2.9.4
--
To unsubscribe from this list: send the line "unsubscribe linux-doc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
