[PATCH 3.12 099/142] iscsi-target: Fix potential dead-lock during node acl delete

From: Jiri Slaby
Date: Wed Feb 24 2016 - 05:29:54 EST


From: Nicholas Bellinger <nab@xxxxxxxxxxxxxxx>

3.12-stable review patch. If anyone has any objections, please let me know.

===============

commit 26a99c19f810b2593410899a5b304b21b47428a6 upstream.

This patch is a iscsi-target specific bug-fix for a dead-lock
that can occur during explicit struct se_node_acl->acl_group
se_session deletion via configfs rmdir(2), when iscsi-target
time2retain timer is still active.

It changes iscsi-target to obtain se_portal_group->session_lock
internally using spin_in_locked() to check for the specific
se_node_acl configfs shutdown rmdir(2) case.

Note this patch is intended for stable, and the subsequent
v4.5-rc patch converts target_core_tpg.c to use proper
se_sess->sess_kref reference counting for both se_node_acl
deletion + se_node_acl->queue_depth se_session restart.

Reported-by:: Sagi Grimberg <sagig@xxxxxxxxxxxx>
Cc: Christoph Hellwig <hch@xxxxxx>
Cc: Hannes Reinecke <hare@xxxxxxx>
Cc: Andy Grover <agrover@xxxxxxxxxx>
Cc: Mike Christie <michaelc@xxxxxxxxxxx>
Signed-off-by: Nicholas Bellinger <nab@xxxxxxxxxxxxxxx>
Signed-off-by: Jiri Slaby <jslaby@xxxxxxx>
---
drivers/target/iscsi/iscsi_target_configfs.c | 16 +++++++++++++++-
1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/drivers/target/iscsi/iscsi_target_configfs.c b/drivers/target/iscsi/iscsi_target_configfs.c
index 8a1bd1af414b..dcebe96d2b23 100644
--- a/drivers/target/iscsi/iscsi_target_configfs.c
+++ b/drivers/target/iscsi/iscsi_target_configfs.c
@@ -1863,7 +1863,8 @@ static void lio_tpg_release_fabric_acl(
}

/*
- * Called with spin_lock_bh(struct se_portal_group->session_lock) held..
+ * Called with spin_lock_irq(struct se_portal_group->session_lock) held
+ * or not held.
*
* Also, this function calls iscsit_inc_session_usage_count() on the
* struct iscsi_session in question.
@@ -1871,19 +1872,32 @@ static void lio_tpg_release_fabric_acl(
static int lio_tpg_shutdown_session(struct se_session *se_sess)
{
struct iscsi_session *sess = se_sess->fabric_sess_ptr;
+ struct se_portal_group *se_tpg = se_sess->se_tpg;
+ bool local_lock = false;
+
+ if (!spin_is_locked(&se_tpg->session_lock)) {
+ spin_lock_irq(&se_tpg->session_lock);
+ local_lock = true;
+ }

spin_lock(&sess->conn_lock);
if (atomic_read(&sess->session_fall_back_to_erl0) ||
atomic_read(&sess->session_logout) ||
(sess->time2retain_timer_flags & ISCSI_TF_EXPIRED)) {
spin_unlock(&sess->conn_lock);
+ if (local_lock)
+ spin_unlock_irq(&sess->conn_lock);
return 0;
}
atomic_set(&sess->session_reinstatement, 1);
spin_unlock(&sess->conn_lock);

iscsit_stop_time2retain_timer(sess);
+ spin_unlock_irq(&se_tpg->session_lock);
+
iscsit_stop_session(sess, 1, 1);
+ if (!local_lock)
+ spin_lock_irq(&se_tpg->session_lock);

return 1;
}
--
2.7.1