bcachefs: Don't kick journal reclaim unless low on space

We shouldn't kick journal reclaim unnecessarily, it's got its own timer
for that.

Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>

diff --git a/fs/bcachefs/journal_io.c b/fs/bcachefs/journal_io.c
index a6a8737e92ad5e524b006ad215a21f8060b0544b..351d5d9d822513e3f8578b5fab5a198b5a205194 100644 (file)
--- a/fs/bcachefs/journal_io.c
+++ b/fs/bcachefs/journal_io.c
@@ -1442,7 +1442,8 @@ static void journal_write_done(struct closure *cl)
         * Must come before signaling write completion, for
         * bch2_fs_journal_stop():
         */
-       journal_reclaim_kick(&c->journal);
+       if (j->watermark)
+               journal_reclaim_kick(&c->journal);
 
        /* also must come before signalling write completion: */
        closure_debug_destroy(cl);