The BUG_ON check performed on the return value of __getblk() in
nilfs_finish_roll_forward() assumes that a buffer that has been
successfully read once is retrieved with the same parameters and does
not fail (__getblk() does not return an error due to memory allocation
failure). Also, nilfs_finish_roll_forward() is called at most once
during mount.
Taking these into consideration, rewrite the check to use WARN_ON() to
avoid using BUG_ON().
Signed-off-by: Ryusuke Konishi <[email protected]>
---
Andrew, please add this to the queue for the next cycle (if it's not late).
This is a change to reduce BUG_ONs in legacy usage.
fs/nilfs2/recovery.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/fs/nilfs2/recovery.c b/fs/nilfs2/recovery.c
index e48372618ac4..020f304c600e 100644
--- a/fs/nilfs2/recovery.c
+++ b/fs/nilfs2/recovery.c
@@ -699,7 +699,9 @@ static void nilfs_finish_roll_forward(struct the_nilfs *nilfs,
return;
bh = __getblk(nilfs->ns_bdev, ri->ri_lsegs_start, nilfs->ns_blocksize);
- BUG_ON(!bh);
+ if (WARN_ON(!bh))
+ return; /* should never happen */
+
memset(bh->b_data, 0, bh->b_size);
set_buffer_dirty(bh);
err = sync_dirty_buffer(bh);
--
2.34.1