From: Chao Yu <[email protected]> [ Upstream commit 4b270a8cc5047682f0a3f3f9af3b498408dbd2bc ]
In synchronous scenario, like in checkpoint(), we are going to flush dirty node pages to device synchronously, we can easily failed writebacking node page due to trylock_page() failure, especially in condition of intensive lock competition, which can cause long latency of checkpoint(). So let's use lock_page() in synchronous scenario to avoid this issue. Signed-off-by: Yunlei He <[email protected]> Signed-off-by: Chao Yu <[email protected]> Signed-off-by: Jaegeuk Kim <[email protected]> Signed-off-by: Sasha Levin <[email protected]> --- fs/f2fs/node.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index f623da26159f..712505ec5de4 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1610,7 +1610,9 @@ int sync_node_pages(struct f2fs_sb_info *sbi, struct writeback_control *wbc, !is_cold_node(page))) continue; lock_node: - if (!trylock_page(page)) + if (wbc->sync_mode == WB_SYNC_ALL) + lock_page(page); + else if (!trylock_page(page)) continue; if (unlikely(page->mapping != NODE_MAPPING(sbi))) { -- 2.17.1
