f2fs: give a chance to mount again when encountering errors
This patch gives another chance to try mount process when we encounter an error.
This makes an effect on the roll-forward recovery failures as well.
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c
index 7a54779..e161e13 100644
--- a/fs/f2fs/super.c
+++ b/fs/f2fs/super.c
@@ -902,8 +902,10 @@
struct buffer_head *raw_super_buf;
struct inode *root;
long err = -EINVAL;
+ bool retry = true;
int i;
+try_onemore:
/* allocate memory for f2fs-specific super block info */
sbi = kzalloc(sizeof(struct f2fs_sb_info), GFP_KERNEL);
if (!sbi)
@@ -1083,9 +1085,11 @@
/* recover fsynced data */
if (!test_opt(sbi, DISABLE_ROLL_FORWARD)) {
err = recover_fsync_data(sbi);
- if (err)
+ if (err) {
f2fs_msg(sb, KERN_ERR,
"Cannot recover all fsync data errno=%ld", err);
+ goto free_kobj;
+ }
}
/*
@@ -1126,6 +1130,13 @@
brelse(raw_super_buf);
free_sbi:
kfree(sbi);
+
+ /* give only one another chance */
+ if (retry) {
+ retry = !retry;
+ shrink_dcache_sb(sb);
+ goto try_onemore;
+ }
return err;
}