Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::3:7 as permitted sender) client-ip=2620:137:e000::3:7;
From:   Kent Overstreet <kent.overstreet@linux.dev>
To:     linux-kernel@vger.kernel.org, linux-bcachefs@vger.kernel.org,
        peterz@infradead.org
Cc:     boqun.feng@gmail.com, longman@redhat.com, will@kernel.org,
        mingo@redhat.com, Kent Overstreet <kent.overstreet@gmail.com>,
        Kent Overstreet <kent.overstreet@linux.dev>
Subject: [PATCH 5/6] bcachefs: Check for btree locks held on transaction init
Date:   Wed, 22 Nov 2023 18:51:12 -0500
Message-ID: <20231122235113.180132-6-kent.overstreet@linux.dev>
In-Reply-To: <20231122235113.180132-1-kent.overstreet@linux.dev>
References: <20231122235113.180132-1-kent.overstreet@linux.dev>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: Kent Overstreet <kent.overstreet@gmail.com>

Ideally we would disallow multiple btree_trans being initialized within
the same process - and hopefully we will at some point, the stack usage
is excessive - but for now there are a couple places where we do this:

 - transaction commit error path -> journal reclaim - btree key cache
   flush
 - move data path -> do_pending_writes -> write path -> bucket
   allocation (in the near future when bucket allocation switches to
   using a freespace btree)

In order to avoid deadlocking the first btree_trans must have been
unlocked with bch2_trans_unlock() before using the second btree_trans -
this patch adds an assertion to bch2_trans_init() that verifies that
this has been done when lockdep is enabled.

Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com>
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
---
 fs/bcachefs/btree_iter.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/bcachefs/btree_iter.c b/fs/bcachefs/btree_iter.c
index 1d734e297eb4..a52fd206f822 100644
--- a/fs/bcachefs/btree_iter.c
+++ b/fs/bcachefs/btree_iter.c
@@ -2854,6 +2854,8 @@ struct btree_trans *__bch2_trans_get(struct bch_fs *c, unsigned fn_idx)
 	struct btree_trans *trans;
 	struct btree_transaction_stats *s;
 
+	bch2_assert_btree_nodes_not_locked();
+
 	trans = bch2_trans_alloc(c);
 
 	memset(trans, 0, sizeof(*trans));
-- 
2.42.0