This patch fixes a generate_node_acls = 1 + cache_dynamic_acls = 0
regression, that was introduced by
  commit 
01d4d673558985d9a118e1e05026633c3e2ade9b
  Author: Nicholas Bellinger <nab@linux-iscsi.org>
  Date:   Wed Dec 7 12:55:54 2016 -0800
which originally had the proper list_del_init() usage, but was
dropped during list review as it was thought unnecessary by HCH.
However, list_del_init() usage is required during the special
generate_node_acls = 1 + cache_dynamic_acls = 0 case when
transport_free_session() does a list_del(&se_nacl->acl_list),
followed by target_complete_nacl() doing the same thing.
This was manifesting as a general protection fault as reported
by Justin:
kernel: general protection fault: 0000 [#1] SMP
kernel: Modules linked in:
kernel: CPU: 0 PID: 11047 Comm: iscsi_ttx Not tainted 4.13.0-rc2.x86_64.1+ #20
kernel: Hardware name: Intel Corporation S5500BC/S5500BC, BIOS S5500.86B.01.00.0064.
050520141428 05/05/2014
kernel: task: 
ffff88026939e800 task.stack: 
ffffc90007884000
kernel: RIP: 0010:target_put_nacl+0x49/0xb0
kernel: RSP: 0018:
ffffc90007887d70 EFLAGS: 
00010246
kernel: RAX: 
dead000000000200 RBX: 
ffff8802556ca000 RCX: 
0000000000000000
kernel: RDX: 
dead000000000100 RSI: 
0000000000000246 RDI: 
ffff8802556ce028
kernel: RBP: 
ffffc90007887d88 R08: 
0000000000000001 R09: 
0000000000000000
kernel: R10: 
ffffc90007887df8 R11: 
ffffea0009986900 R12: 
ffff8802556ce020
kernel: R13: 
ffff8802556ce028 R14: 
ffff8802556ce028 R15: 
ffffffff88d85540
kernel: FS:  
0000000000000000(0000) GS:
ffff88027fc00000(0000) knlGS:
0000000000000000
kernel: CS:  0010 DS: 0000 ES: 0000 CR0: 
0000000080050033
kernel: CR2: 
00007fffe36f5f94 CR3: 
0000000009209000 CR4: 
00000000003406f0
kernel: DR0: 
0000000000000000 DR1: 
0000000000000000 DR2: 
0000000000000000
kernel: DR3: 
0000000000000000 DR6: 
00000000fffe0ff0 DR7: 
0000000000000400
kernel: Call Trace:
kernel:  transport_free_session+0x67/0x140
kernel:  transport_deregister_session+0x7a/0xc0
kernel:  iscsit_close_session+0x92/0x210
kernel:  iscsit_close_connection+0x5f9/0x840
kernel:  iscsit_take_action_for_connection_exit+0xfe/0x110
kernel:  iscsi_target_tx_thread+0x140/0x1e0
kernel:  ? wait_woken+0x90/0x90
kernel:  kthread+0x124/0x160
kernel:  ? iscsit_thread_get_cpumask+0x90/0x90
kernel:  ? kthread_create_on_node+0x40/0x40
kernel:  ret_from_fork+0x22/0x30
kernel: Code: 00 48 89 fb 4c 8b a7 48 01 00 00 74 68 4d 8d 6c 24 08 4c
89 ef e8 e8 28 43 00 48 8b 93 20 04 00 00 48 8b 83 28 04 00 00 4c 89
ef <48> 89 42 08 48 89 10 48 b8 00 01 00 00 00 00 ad de 48 89 83 20
kernel: RIP: target_put_nacl+0x49/0xb0 RSP: 
ffffc90007887d70
kernel: ---[ end trace 
f12821adbfd46fed ]---
To address this, go ahead and use proper list_del_list() for all
cases of se_nacl->acl_list deletion.
Reported-by: Justin Maggard <jmaggard01@gmail.com>
Tested-by: Justin Maggard <jmaggard01@gmail.com>
Cc: Justin Maggard <jmaggard01@gmail.com>
Cc: stable@vger.kernel.org # 4.1+
Signed-off-by: Nicholas Bellinger <nab@linux-iscsi.org>
        mutex_lock(&tpg->acl_node_mutex);
        if (acl->dynamic_node_acl)
                acl->dynamic_node_acl = 0;
-       list_del(&acl->acl_list);
+       list_del_init(&acl->acl_list);
        mutex_unlock(&tpg->acl_node_mutex);
 
        target_shutdown_sessions(acl);
         * in transport_deregister_session().
         */
        list_for_each_entry_safe(nacl, nacl_tmp, &node_list, acl_list) {
-               list_del(&nacl->acl_list);
+               list_del_init(&nacl->acl_list);
 
                core_tpg_wait_for_nacl_pr_ref(nacl);
                core_free_device_list_for_node(nacl, se_tpg);
 
        }
 
        mutex_lock(&se_tpg->acl_node_mutex);
-       list_del(&nacl->acl_list);
+       list_del_init(&nacl->acl_list);
        mutex_unlock(&se_tpg->acl_node_mutex);
 
        core_tpg_wait_for_nacl_pr_ref(nacl);
                        spin_unlock_irqrestore(&se_nacl->nacl_sess_lock, flags);
 
                        if (se_nacl->dynamic_stop)
-                               list_del(&se_nacl->acl_list);
+                               list_del_init(&se_nacl->acl_list);
                }
                mutex_unlock(&se_tpg->acl_node_mutex);