[PATCH] f2fs: fix termination of f2fs_sync_meta_caches()

[PATCH] f2fs: fix termination of f2fs_sync_meta_caches()

[Seongjae Jeong]

When __f2fs_write_meta_cache() returns false in f2fs_sync_meta_caches(),
the function only exits the inner loop instead of the entire
while loop. This leads to an unnecessary lookup of the next batch.

Also, when nwritten >= nr_to_write, only the inner loop terminates.
This allows the function to continue processing subsequent batches
after the writeback budget has been reached.

Terminate the entire while loop in both cases to avoid unnecessary
batch lookups and respect the writeback budget.

Signed-off-by: Seongjae Jeong <jsjlee1020@gmail.com>
---
based on the f2fs dev-test at commit ce439a2b0690

 fs/f2fs/checkpoint.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c
index 2e5e0ee5feb2..4cafc96056ab 100644
--- a/fs/f2fs/checkpoint.c
+++ b/fs/f2fs/checkpoint.c
@@ -654,7 +654,8 @@ long f2fs_sync_meta_caches(struct f2fs_sb_info *sbi, long nr_to_write,
 
 			if (!__f2fs_write_meta_cache(entry, io_type)) {
 				f2fs_unlock_cache(entry);
-				break;
+				f2fs_cache_gang_release(entries, nr);
+				goto stop;
 			}
 			nwritten++;
 			next = entry->index + 1;
@@ -663,6 +664,9 @@ long f2fs_sync_meta_caches(struct f2fs_sb_info *sbi, long nr_to_write,
 		}
 		f2fs_cache_gang_release(entries, nr);
 		cond_resched();
+
+		if (unlikely(nwritten >= nr_to_write))
+			break;
 	}
 stop:
 	if (nwritten)
-- 
2.53.0