From 06b3e6ae408ad93d9ff00699a275a3678e50bec1 Mon Sep 17 00:00:00 2001
From: Qu Wenruo <wqu@suse.com>
Date: Fri, 4 Dec 2020 09:24:47 +0800
Subject: [PATCH] btrfs: qgroup: don't try to wait flushing if we're already
 holding a transaction
References: bsc#1179575
Git-commit: ae5e070eaca9dbebde3459dd8f4c2756f8c097d0
Patch-mainline: v5.11-rc3

There is a chance of racing for qgroup flushing which may lead to
deadlock:

	Thread A		|	Thread B
   (not holding trans handle)	|  (holding a trans handle)
--------------------------------+--------------------------------
__btrfs_qgroup_reserve_meta()   | __btrfs_qgroup_reserve_meta()
|- try_flush_qgroup()		| |- try_flush_qgroup()
   |- QGROUP_FLUSHING bit set   |    |
   |				|    |- test_and_set_bit()
   |				|    |- wait_event()
   |- btrfs_join_transaction()	|
   |- btrfs_commit_transaction()|

			!!! DEAD LOCK !!!

Since thread A wants to commit transaction, but thread B is holding a
transaction handle, blocking the commit.
At the same time, thread B is waiting for thread A to finish its commit.

This is just a hot fix, and would lead to more EDQUOT when we're near
the qgroup limit.

The proper fix would be to make all metadata/data reservations happen
without holding a transaction handle.

CC: stable@vger.kernel.org # 5.9+
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
---
 fs/btrfs/qgroup.c | 30 ++++++++++++++++++++----------
 1 file changed, 20 insertions(+), 10 deletions(-)

diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index fe3046007f52..47f27658eac1 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -3530,16 +3530,6 @@ static int try_flush_qgroup(struct btrfs_root *root)
 	int ret;
 	bool can_commit = true;
 
-	/*
-	 * We don't want to run flush again and again, so if there is a running
-	 * one, we won't try to start a new flush, but exit directly.
-	 */
-	if (test_and_set_bit(BTRFS_ROOT_QGROUP_FLUSHING, &root->state)) {
-		wait_event(root->qgroup_flush_wait,
-			!test_bit(BTRFS_ROOT_QGROUP_FLUSHING, &root->state));
-		return 0;
-	}
-
 	/*
 	 * If current process holds a transaction, we shouldn't flush, as we
 	 * assume all space reservation happens before a transaction handle is
@@ -3554,6 +3544,26 @@ static int try_flush_qgroup(struct btrfs_root *root)
 	    current->journal_info != BTRFS_SEND_TRANS_STUB)
 		can_commit = false;
 
+	/*
+	 * We don't want to run flush again and again, so if there is a running
+	 * one, we won't try to start a new flush, but exit directly.
+	 */
+	if (test_and_set_bit(BTRFS_ROOT_QGROUP_FLUSHING, &root->state)) {
+		/*
+		 * We are already holding a transaction, thus we can block other
+		 * threads from flushing.  So exit right now. This increases
+		 * the chance of EDQUOT for heavy load and near limit cases.
+		 * But we can argue that if we're already near limit, EDQUOT is
+		 * unavoidable anyway.
+		 */
+		if (!can_commit)
+			return 0;
+
+		wait_event(root->qgroup_flush_wait,
+			!test_bit(BTRFS_ROOT_QGROUP_FLUSHING, &root->state));
+		return 0;
+	}
+
 	ret = btrfs_start_delalloc_snapshot(root);
 	if (ret < 0)
 		goto out;
-- 
2.26.2