From becd9be6069e7b183c084f460f0eb363e43cc487 Mon Sep 17 00:00:00 2001
From: Mike Christie <michael.christie@oracle.com>
Date: Sat, 18 Mar 2023 20:56:12 -0500
Subject: [PATCH 1/9] scsi: target: Move sess cmd counter to new struct

iSCSI needs to wait on outstanding commands like how SRP and the FC/FCoE
drivers do. It can't use target_stop_session() because for MCS support we
can't stop the entire session during recovery because if other connections
are OK then we want to be able to continue to execute I/O on them.

Move the per session cmd counters to a new struct so iSCSI can allocate
them per connection. The xcopy code can also just not allocate in the
future since it doesn't need to track commands.

Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-2-michael.christie@oracle.com
Reviewed-by: Maurizio Lombardi <mlombard@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/target/target_core_tpg.c         |   2 +-
 drivers/target/target_core_transport.c   | 135 ++++++++++++++++-------
 include/target/iscsi/iscsi_target_core.h |   1 +
 include/target/target_core_base.h        |  13 ++-
 4 files changed, 107 insertions(+), 44 deletions(-)

diff --git a/drivers/target/target_core_tpg.c b/drivers/target/target_core_tpg.c
index 736847c933e5..8ebccdbd94f0 100644
--- a/drivers/target/target_core_tpg.c
+++ b/drivers/target/target_core_tpg.c
@@ -328,7 +328,7 @@ static void target_shutdown_sessions(struct se_node_acl *acl)
 restart:
 	spin_lock_irqsave(&acl->nacl_sess_lock, flags);
 	list_for_each_entry(sess, &acl->acl_sess_list, sess_acl_list) {
-		if (atomic_read(&sess->stopped))
+		if (sess->cmd_cnt && atomic_read(&sess->cmd_cnt->stopped))
 			continue;
 
 		list_del_init(&sess->sess_acl_list);
diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index 5926316252eb..3d6034f00dcd 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -220,11 +220,49 @@ void transport_subsystem_check_init(void)
 	sub_api_initialized = 1;
 }
 
-static void target_release_sess_cmd_refcnt(struct percpu_ref *ref)
+static void target_release_cmd_refcnt(struct percpu_ref *ref)
 {
-	struct se_session *sess = container_of(ref, typeof(*sess), cmd_count);
+	struct target_cmd_counter *cmd_cnt  = container_of(ref,
+							   typeof(*cmd_cnt),
+							   refcnt);
+	wake_up(&cmd_cnt->refcnt_wq);
+}
 
-	wake_up(&sess->cmd_count_wq);
+static struct target_cmd_counter *target_alloc_cmd_counter(void)
+{
+	struct target_cmd_counter *cmd_cnt;
+	int rc;
+
+	cmd_cnt = kzalloc(sizeof(*cmd_cnt), GFP_KERNEL);
+	if (!cmd_cnt)
+		return NULL;
+
+	init_completion(&cmd_cnt->stop_done);
+	init_waitqueue_head(&cmd_cnt->refcnt_wq);
+	atomic_set(&cmd_cnt->stopped, 0);
+
+	rc = percpu_ref_init(&cmd_cnt->refcnt, target_release_cmd_refcnt, 0,
+			     GFP_KERNEL);
+	if (rc)
+		goto free_cmd_cnt;
+
+	return cmd_cnt;
+
+free_cmd_cnt:
+	kfree(cmd_cnt);
+	return NULL;
+}
+
+static void target_free_cmd_counter(struct target_cmd_counter *cmd_cnt)
+{
+	/*
+	 * Drivers like loop do not call target_stop_session during session
+	 * shutdown so we have to drop the ref taken at init time here.
+	 */
+	if (!atomic_read(&cmd_cnt->stopped))
+		percpu_ref_put(&cmd_cnt->refcnt);
+
+	percpu_ref_exit(&cmd_cnt->refcnt);
 }
 
 /**
@@ -238,25 +276,17 @@ int transport_init_session(struct se_session *se_sess)
 	INIT_LIST_HEAD(&se_sess->sess_list);
 	INIT_LIST_HEAD(&se_sess->sess_acl_list);
 	spin_lock_init(&se_sess->sess_cmd_lock);
-	init_waitqueue_head(&se_sess->cmd_count_wq);
-	init_completion(&se_sess->stop_done);
-	atomic_set(&se_sess->stopped, 0);
-	return percpu_ref_init(&se_sess->cmd_count,
-			       target_release_sess_cmd_refcnt, 0, GFP_KERNEL);
+	se_sess->cmd_cnt = target_alloc_cmd_counter();
+	if (!se_sess->cmd_cnt)
+		return -ENOMEM;
+
+	return  0;
 }
 EXPORT_SYMBOL(transport_init_session);
 
 void transport_uninit_session(struct se_session *se_sess)
 {
-	/*
-	 * Drivers like iscsi and loop do not call target_stop_session
-	 * during session shutdown so we have to drop the ref taken at init
-	 * time here.
-	 */
-	if (!atomic_read(&se_sess->stopped))
-		percpu_ref_put(&se_sess->cmd_count);
-
-	percpu_ref_exit(&se_sess->cmd_count);
+	target_free_cmd_counter(se_sess->cmd_cnt);
 }
 
 /**
@@ -2970,9 +3000,16 @@ int target_get_sess_cmd(struct se_cmd *se_cmd, bool ack_kref)
 		se_cmd->se_cmd_flags |= SCF_ACK_KREF;
 	}
 
-	if (!percpu_ref_tryget_live(&se_sess->cmd_count))
-		ret = -ESHUTDOWN;
-
+	/*
+	 * Users like xcopy do not use counters since they never do a stop
+	 * and wait.
+	 */
+	if (se_sess->cmd_cnt) {
+		if (!percpu_ref_tryget_live(&se_sess->cmd_cnt->refcnt))
+			ret = -ESHUTDOWN;
+		else
+			se_cmd->cmd_cnt = se_sess->cmd_cnt;
+	}
 	if (ret && ack_kref)
 		target_put_sess_cmd(se_cmd);
 
@@ -2993,7 +3030,7 @@ static void target_free_cmd_mem(struct se_cmd *cmd)
 static void target_release_cmd_kref(struct kref *kref)
 {
 	struct se_cmd *se_cmd = container_of(kref, struct se_cmd, cmd_kref);
-	struct se_session *se_sess = se_cmd->se_sess;
+	struct target_cmd_counter *cmd_cnt = se_cmd->cmd_cnt;
 	struct completion *free_compl = se_cmd->free_compl;
 	struct completion *abrt_compl = se_cmd->abrt_compl;
 
@@ -3004,7 +3041,8 @@ static void target_release_cmd_kref(struct kref *kref)
 	if (abrt_compl)
 		complete(abrt_compl);
 
-	percpu_ref_put(&se_sess->cmd_count);
+	if (cmd_cnt)
+		percpu_ref_put(&cmd_cnt->refcnt);
 }
 
 /**
@@ -3123,46 +3161,65 @@ void target_show_cmd(const char *pfx, struct se_cmd *cmd)
 }
 EXPORT_SYMBOL(target_show_cmd);
 
-static void target_stop_session_confirm(struct percpu_ref *ref)
+static void target_stop_cmd_counter_confirm(struct percpu_ref *ref)
 {
-	struct se_session *se_sess = container_of(ref, struct se_session,
-						  cmd_count);
-	complete_all(&se_sess->stop_done);
+	struct target_cmd_counter *cmd_cnt = container_of(ref,
+						struct target_cmd_counter,
+						refcnt);
+	complete_all(&cmd_cnt->stop_done);
+}
+
+/**
+ * target_stop_cmd_counter - Stop new IO from being added to the counter.
+ * @cmd_cnt: counter to stop
+ */
+static void target_stop_cmd_counter(struct target_cmd_counter *cmd_cnt)
+{
+	pr_debug("Stopping command counter.\n");
+	if (!atomic_cmpxchg(&cmd_cnt->stopped, 0, 1))
+		percpu_ref_kill_and_confirm(&cmd_cnt->refcnt,
+					    target_stop_cmd_counter_confirm);
 }
 
 /**
  * target_stop_session - Stop new IO from being queued on the session.
- * @se_sess:    session to stop
+ * @se_sess: session to stop
  */
 void target_stop_session(struct se_session *se_sess)
 {
-	pr_debug("Stopping session queue.\n");
-	if (atomic_cmpxchg(&se_sess->stopped, 0, 1) == 0)
-		percpu_ref_kill_and_confirm(&se_sess->cmd_count,
-					    target_stop_session_confirm);
+	target_stop_cmd_counter(se_sess->cmd_cnt);
 }
 EXPORT_SYMBOL(target_stop_session);
 
 /**
- * target_wait_for_sess_cmds - Wait for outstanding commands
- * @se_sess:    session to wait for active I/O
+ * target_wait_for_cmds - Wait for outstanding cmds.
+ * @cmd_cnt: counter to wait for active I/O for.
  */
-void target_wait_for_sess_cmds(struct se_session *se_sess)
+static void target_wait_for_cmds(struct target_cmd_counter *cmd_cnt)
 {
 	int ret;
 
-	WARN_ON_ONCE(!atomic_read(&se_sess->stopped));
+	WARN_ON_ONCE(!atomic_read(&cmd_cnt->stopped));
 
 	do {
 		pr_debug("Waiting for running cmds to complete.\n");
-		ret = wait_event_timeout(se_sess->cmd_count_wq,
-				percpu_ref_is_zero(&se_sess->cmd_count),
-				180 * HZ);
+		ret = wait_event_timeout(cmd_cnt->refcnt_wq,
+					 percpu_ref_is_zero(&cmd_cnt->refcnt),
+					 180 * HZ);
 	} while (ret <= 0);
 
-	wait_for_completion(&se_sess->stop_done);
+	wait_for_completion(&cmd_cnt->stop_done);
 	pr_debug("Waiting for cmds done.\n");
 }
+
+/**
+ * target_wait_for_sess_cmds - Wait for outstanding commands
+ * @se_sess: session to wait for active I/O
+ */
+void target_wait_for_sess_cmds(struct se_session *se_sess)
+{
+	target_wait_for_cmds(se_sess->cmd_cnt);
+}
 EXPORT_SYMBOL(target_wait_for_sess_cmds);
 
 /*
diff --git a/include/target/iscsi/iscsi_target_core.h b/include/target/iscsi/iscsi_target_core.h
index 94d06ddfd80a..229118156a1f 100644
--- a/include/target/iscsi/iscsi_target_core.h
+++ b/include/target/iscsi/iscsi_target_core.h
@@ -600,6 +600,7 @@ struct iscsit_conn {
 	struct iscsi_tpg_np	*tpg_np;
 	/* Pointer to parent session */
 	struct iscsit_session	*sess;
+	struct target_cmd_counter *cmd_cnt;
 	int			bitmap_id;
 	int			rx_thread_active;
 	struct task_struct	*rx_thread;
diff --git a/include/target/target_core_base.h b/include/target/target_core_base.h
index 12c9ba16217e..bd299790e99c 100644
--- a/include/target/target_core_base.h
+++ b/include/target/target_core_base.h
@@ -494,6 +494,7 @@ struct se_cmd {
 	struct se_lun		*se_lun;
 	/* Only used for internal passthrough and legacy TCM fabric modules */
 	struct se_session	*se_sess;
+	struct target_cmd_counter *cmd_cnt;
 	struct se_tmr_req	*se_tmr_req;
 	struct llist_node	se_cmd_list;
 	struct completion	*free_compl;
@@ -619,22 +620,26 @@ static inline struct se_node_acl *fabric_stat_to_nacl(struct config_item *item)
 			acl_fabric_stat_group);
 }
 
-struct se_session {
+struct target_cmd_counter {
+	struct percpu_ref	refcnt;
+	wait_queue_head_t	refcnt_wq;
+	struct completion	stop_done;
 	atomic_t		stopped;
+};
+
+struct se_session {
 	u64			sess_bin_isid;
 	enum target_prot_op	sup_prot_ops;
 	enum target_prot_type	sess_prot_type;
 	struct se_node_acl	*se_node_acl;
 	struct se_portal_group *se_tpg;
 	void			*fabric_sess_ptr;
-	struct percpu_ref	cmd_count;
 	struct list_head	sess_list;
 	struct list_head	sess_acl_list;
 	spinlock_t		sess_cmd_lock;
-	wait_queue_head_t	cmd_count_wq;
-	struct completion	stop_done;
 	void			*sess_cmd_map;
 	struct sbitmap_queue	sess_tag_pool;
+	struct target_cmd_counter *cmd_cnt;
 };
 
 struct se_device;

From 4edba7e4a8f39112398d3cda94128a8e13a7d527 Mon Sep 17 00:00:00 2001
From: Mike Christie <michael.christie@oracle.com>
Date: Sat, 18 Mar 2023 20:56:13 -0500
Subject: [PATCH 2/9] scsi: target: Move cmd counter allocation

iSCSI needs to allocate its cmd counter per connection for MCS support
where we need to stop and wait on commands running on a connection instead
of per session. This moves the cmd counter allocation to
target_setup_session() which is used by drivers that need the stop+wait
behavior per session.

xcopy doesn't need stop+wait at all, so we will be OK moving the cmd
counter allocation outside of transport_init_session().

Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-3-michael.christie@oracle.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/target/iscsi/iscsi_target_login.c | 10 +++++
 drivers/target/target_core_internal.h     |  1 -
 drivers/target/target_core_transport.c    | 55 +++++++++++------------
 drivers/target/target_core_xcopy.c        | 15 +------
 include/target/target_core_fabric.h       |  4 +-
 5 files changed, 42 insertions(+), 43 deletions(-)

diff --git a/drivers/target/iscsi/iscsi_target_login.c b/drivers/target/iscsi/iscsi_target_login.c
index 27e448c2d066..8ab6c0107d89 100644
--- a/drivers/target/iscsi/iscsi_target_login.c
+++ b/drivers/target/iscsi/iscsi_target_login.c
@@ -324,8 +324,18 @@ static int iscsi_login_zero_tsih_s1(
 		goto free_ops;
 	}
 
+	/*
+	 * This is temp for iser. It will be moved to per conn in later
+	 * patches for iscsi.
+	 */
+	sess->se_sess->cmd_cnt = target_alloc_cmd_counter();
+	if (!sess->se_sess->cmd_cnt)
+		goto free_se_sess;
+
 	return 0;
 
+free_se_sess:
+	transport_free_session(sess->se_sess);
 free_ops:
 	kfree(sess->sess_ops);
 free_id:
diff --git a/drivers/target/target_core_internal.h b/drivers/target/target_core_internal.h
index 38a6d08f75b3..85e35cf582e5 100644
--- a/drivers/target/target_core_internal.h
+++ b/drivers/target/target_core_internal.h
@@ -138,7 +138,6 @@ int	init_se_kmem_caches(void);
 void	release_se_kmem_caches(void);
 u32	scsi_get_new_index(scsi_index_t);
 void	transport_subsystem_check_init(void);
-void	transport_uninit_session(struct se_session *);
 unsigned char *transport_dump_cmd_direction(struct se_cmd *);
 void	transport_dump_dev_state(struct se_device *, char *, int *);
 void	transport_dump_dev_info(struct se_device *, struct se_lun *,
diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index 3d6034f00dcd..60647a49a1d3 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -228,7 +228,7 @@ static void target_release_cmd_refcnt(struct percpu_ref *ref)
 	wake_up(&cmd_cnt->refcnt_wq);
 }
 
-static struct target_cmd_counter *target_alloc_cmd_counter(void)
+struct target_cmd_counter *target_alloc_cmd_counter(void)
 {
 	struct target_cmd_counter *cmd_cnt;
 	int rc;
@@ -252,6 +252,7 @@ static struct target_cmd_counter *target_alloc_cmd_counter(void)
 	kfree(cmd_cnt);
 	return NULL;
 }
+EXPORT_SYMBOL_GPL(target_alloc_cmd_counter);
 
 static void target_free_cmd_counter(struct target_cmd_counter *cmd_cnt)
 {
@@ -271,24 +272,14 @@ static void target_free_cmd_counter(struct target_cmd_counter *cmd_cnt)
  *
  * The caller must have zero-initialized @se_sess before calling this function.
  */
-int transport_init_session(struct se_session *se_sess)
+void transport_init_session(struct se_session *se_sess)
 {
 	INIT_LIST_HEAD(&se_sess->sess_list);
 	INIT_LIST_HEAD(&se_sess->sess_acl_list);
 	spin_lock_init(&se_sess->sess_cmd_lock);
-	se_sess->cmd_cnt = target_alloc_cmd_counter();
-	if (!se_sess->cmd_cnt)
-		return -ENOMEM;
-
-	return  0;
 }
 EXPORT_SYMBOL(transport_init_session);
 
-void transport_uninit_session(struct se_session *se_sess)
-{
-	target_free_cmd_counter(se_sess->cmd_cnt);
-}
-
 /**
  * transport_alloc_session - allocate a session object and initialize it
  * @sup_prot_ops: bitmask that defines which T10-PI modes are supported.
@@ -296,7 +287,6 @@ void transport_uninit_session(struct se_session *se_sess)
 struct se_session *transport_alloc_session(enum target_prot_op sup_prot_ops)
 {
 	struct se_session *se_sess;
-	int ret;
 
 	se_sess = kmem_cache_zalloc(se_sess_cache, GFP_KERNEL);
 	if (!se_sess) {
@@ -304,11 +294,7 @@ struct se_session *transport_alloc_session(enum target_prot_op sup_prot_ops)
 				" se_sess_cache\n");
 		return ERR_PTR(-ENOMEM);
 	}
-	ret = transport_init_session(se_sess);
-	if (ret < 0) {
-		kmem_cache_free(se_sess_cache, se_sess);
-		return ERR_PTR(ret);
-	}
+	transport_init_session(se_sess);
 	se_sess->sup_prot_ops = sup_prot_ops;
 
 	return se_sess;
@@ -474,8 +460,13 @@ target_setup_session(struct se_portal_group *tpg,
 		     int (*callback)(struct se_portal_group *,
 				     struct se_session *, void *))
 {
+	struct target_cmd_counter *cmd_cnt;
 	struct se_session *sess;
+	int rc;
 
+	cmd_cnt = target_alloc_cmd_counter();
+	if (!cmd_cnt)
+		return ERR_PTR(-ENOMEM);
 	/*
 	 * If the fabric driver is using percpu-ida based pre allocation
 	 * of I/O descriptor tags, go ahead and perform that setup now..
@@ -485,29 +476,36 @@ target_setup_session(struct se_portal_group *tpg,
 	else
 		sess = transport_alloc_session(prot_op);
 
-	if (IS_ERR(sess))
-		return sess;
+	if (IS_ERR(sess)) {
+		rc = PTR_ERR(sess);
+		goto free_cnt;
+	}
+	sess->cmd_cnt = cmd_cnt;
 
 	sess->se_node_acl = core_tpg_check_initiator_node_acl(tpg,
 					(unsigned char *)initiatorname);
 	if (!sess->se_node_acl) {
-		transport_free_session(sess);
-		return ERR_PTR(-EACCES);
+		rc = -EACCES;
+		goto free_sess;
 	}
 	/*
 	 * Go ahead and perform any remaining fabric setup that is
 	 * required before transport_register_session().
 	 */
 	if (callback != NULL) {
-		int rc = callback(tpg, sess, private);
-		if (rc) {
-			transport_free_session(sess);
-			return ERR_PTR(rc);
-		}
+		rc = callback(tpg, sess, private);
+		if (rc)
+			goto free_sess;
 	}
 
 	transport_register_session(tpg, sess->se_node_acl, sess, private);
 	return sess;
+
+free_sess:
+	transport_free_session(sess);
+free_cnt:
+	target_free_cmd_counter(cmd_cnt);
+	return ERR_PTR(rc);
 }
 EXPORT_SYMBOL(target_setup_session);
 
@@ -632,7 +630,8 @@ void transport_free_session(struct se_session *se_sess)
 		sbitmap_queue_free(&se_sess->sess_tag_pool);
 		kvfree(se_sess->sess_cmd_map);
 	}
-	transport_uninit_session(se_sess);
+	if (se_sess->cmd_cnt)
+		target_free_cmd_counter(se_sess->cmd_cnt);
 	kmem_cache_free(se_sess_cache, se_sess);
 }
 EXPORT_SYMBOL(transport_free_session);
diff --git a/drivers/target/target_core_xcopy.c b/drivers/target/target_core_xcopy.c
index 49eaee022ef1..49a83500c8b7 100644
--- a/drivers/target/target_core_xcopy.c
+++ b/drivers/target/target_core_xcopy.c
@@ -461,8 +461,6 @@ static const struct target_core_fabric_ops xcopy_pt_tfo = {
 
 int target_xcopy_setup_pt(void)
 {
-	int ret;
-
 	xcopy_wq = alloc_workqueue("xcopy_wq", WQ_MEM_RECLAIM, 0);
 	if (!xcopy_wq) {
 		pr_err("Unable to allocate xcopy_wq\n");
@@ -479,9 +477,7 @@ int target_xcopy_setup_pt(void)
 	INIT_LIST_HEAD(&xcopy_pt_nacl.acl_list);
 	INIT_LIST_HEAD(&xcopy_pt_nacl.acl_sess_list);
 	memset(&xcopy_pt_sess, 0, sizeof(struct se_session));
-	ret = transport_init_session(&xcopy_pt_sess);
-	if (ret < 0)
-		goto destroy_wq;
+	transport_init_session(&xcopy_pt_sess);
 
 	xcopy_pt_nacl.se_tpg = &xcopy_pt_tpg;
 	xcopy_pt_nacl.nacl_sess = &xcopy_pt_sess;
@@ -490,19 +486,12 @@ int target_xcopy_setup_pt(void)
 	xcopy_pt_sess.se_node_acl = &xcopy_pt_nacl;
 
 	return 0;
-
-destroy_wq:
-	destroy_workqueue(xcopy_wq);
-	xcopy_wq = NULL;
-	return ret;
 }
 
 void target_xcopy_release_pt(void)
 {
-	if (xcopy_wq) {
+	if (xcopy_wq)
 		destroy_workqueue(xcopy_wq);
-		transport_uninit_session(&xcopy_pt_sess);
-	}
 }
 
 /*
diff --git a/include/target/target_core_fabric.h b/include/target/target_core_fabric.h
index 38f0662476d1..65527174b8bc 100644
--- a/include/target/target_core_fabric.h
+++ b/include/target/target_core_fabric.h
@@ -133,7 +133,9 @@ struct se_session *target_setup_session(struct se_portal_group *,
 				struct se_session *, void *));
 void target_remove_session(struct se_session *);
 
-int transport_init_session(struct se_session *se_sess);
+struct target_cmd_counter *target_alloc_cmd_counter(void);
+
+void transport_init_session(struct se_session *se_sess);
 struct se_session *transport_alloc_session(enum target_prot_op);
 int transport_alloc_session_tags(struct se_session *, unsigned int,
 		unsigned int);

From 8e288be8606ad87c1726618eacfb8fbd3ab4b806 Mon Sep 17 00:00:00 2001
From: Mike Christie <michael.christie@oracle.com>
Date: Sat, 18 Mar 2023 20:56:14 -0500
Subject: [PATCH 3/9] scsi: target: Pass in cmd counter to use during cmd setup

Allow target_get_sess_cmd() users to pass in the cmd counter they want to
use. Right now we pass in the session's cmd counter but in a subsequent
commit iSCSI will switch from per session to per conn.

Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-4-michael.christie@oracle.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/target/iscsi/iscsi_target.c    | 10 +++++----
 drivers/target/target_core_transport.c | 28 ++++++++++++--------------
 drivers/target/target_core_xcopy.c     |  8 ++++----
 drivers/usb/gadget/function/f_tcm.c    |  4 ++--
 include/target/target_core_fabric.h    |  8 +++++---
 5 files changed, 30 insertions(+), 28 deletions(-)

diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
index baf4da7bb3b4..87927a36f90d 100644
--- a/drivers/target/iscsi/iscsi_target.c
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -1190,9 +1190,10 @@ int iscsit_setup_scsi_cmd(struct iscsit_conn *conn, struct iscsit_cmd *cmd,
 	 * Initialize struct se_cmd descriptor from target_core_mod infrastructure
 	 */
 	__target_init_cmd(&cmd->se_cmd, &iscsi_ops,
-			 conn->sess->se_sess, be32_to_cpu(hdr->data_length),
-			 cmd->data_direction, sam_task_attr,
-			 cmd->sense_buffer + 2, scsilun_to_int(&hdr->lun));
+			  conn->sess->se_sess, be32_to_cpu(hdr->data_length),
+			  cmd->data_direction, sam_task_attr,
+			  cmd->sense_buffer + 2, scsilun_to_int(&hdr->lun),
+			  conn->sess->se_sess->cmd_cnt);
 
 	pr_debug("Got SCSI Command, ITT: 0x%08x, CmdSN: 0x%08x,"
 		" ExpXferLen: %u, Length: %u, CID: %hu\n", hdr->itt,
@@ -2055,7 +2056,8 @@ iscsit_handle_task_mgt_cmd(struct iscsit_conn *conn, struct iscsit_cmd *cmd,
 	__target_init_cmd(&cmd->se_cmd, &iscsi_ops,
 			  conn->sess->se_sess, 0, DMA_NONE,
 			  TCM_SIMPLE_TAG, cmd->sense_buffer + 2,
-			  scsilun_to_int(&hdr->lun));
+			  scsilun_to_int(&hdr->lun),
+			  conn->sess->se_sess->cmd_cnt);
 
 	target_get_sess_cmd(&cmd->se_cmd, true);
 
diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index 60647a49a1d3..c395606ab1a9 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -1441,14 +1441,12 @@ target_cmd_size_check(struct se_cmd *cmd, unsigned int size)
  *
  * Preserves the value of @cmd->tag.
  */
-void __target_init_cmd(
-	struct se_cmd *cmd,
-	const struct target_core_fabric_ops *tfo,
-	struct se_session *se_sess,
-	u32 data_length,
-	int data_direction,
-	int task_attr,
-	unsigned char *sense_buffer, u64 unpacked_lun)
+void __target_init_cmd(struct se_cmd *cmd,
+		       const struct target_core_fabric_ops *tfo,
+		       struct se_session *se_sess, u32 data_length,
+		       int data_direction, int task_attr,
+		       unsigned char *sense_buffer, u64 unpacked_lun,
+		       struct target_cmd_counter *cmd_cnt)
 {
 	INIT_LIST_HEAD(&cmd->se_delayed_node);
 	INIT_LIST_HEAD(&cmd->se_qf_node);
@@ -1468,6 +1466,7 @@ void __target_init_cmd(
 	cmd->sam_task_attr = task_attr;
 	cmd->sense_buffer = sense_buffer;
 	cmd->orig_fe_lun = unpacked_lun;
+	cmd->cmd_cnt = cmd_cnt;
 
 	if (!(cmd->se_cmd_flags & SCF_USE_CPUID))
 		cmd->cpuid = raw_smp_processor_id();
@@ -1687,7 +1686,8 @@ int target_init_cmd(struct se_cmd *se_cmd, struct se_session *se_sess,
 	 * target_core_fabric_ops->queue_status() callback
 	 */
 	__target_init_cmd(se_cmd, se_tpg->se_tpg_tfo, se_sess, data_length,
-			  data_dir, task_attr, sense, unpacked_lun);
+			  data_dir, task_attr, sense, unpacked_lun,
+			  se_sess->cmd_cnt);
 
 	/*
 	 * Obtain struct se_cmd->cmd_kref reference. A second kref_get here is
@@ -1982,7 +1982,8 @@ int target_submit_tmr(struct se_cmd *se_cmd, struct se_session *se_sess,
 	BUG_ON(!se_tpg);
 
 	__target_init_cmd(se_cmd, se_tpg->se_tpg_tfo, se_sess,
-			  0, DMA_NONE, TCM_SIMPLE_TAG, sense, unpacked_lun);
+			  0, DMA_NONE, TCM_SIMPLE_TAG, sense, unpacked_lun,
+			  se_sess->cmd_cnt);
 	/*
 	 * FIXME: Currently expect caller to handle se_cmd->se_tmr_req
 	 * allocation failure.
@@ -2986,7 +2987,6 @@ EXPORT_SYMBOL(transport_generic_free_cmd);
  */
 int target_get_sess_cmd(struct se_cmd *se_cmd, bool ack_kref)
 {
-	struct se_session *se_sess = se_cmd->se_sess;
 	int ret = 0;
 
 	/*
@@ -3003,11 +3003,9 @@ int target_get_sess_cmd(struct se_cmd *se_cmd, bool ack_kref)
 	 * Users like xcopy do not use counters since they never do a stop
 	 * and wait.
 	 */
-	if (se_sess->cmd_cnt) {
-		if (!percpu_ref_tryget_live(&se_sess->cmd_cnt->refcnt))
+	if (se_cmd->cmd_cnt) {
+		if (!percpu_ref_tryget_live(&se_cmd->cmd_cnt->refcnt))
 			ret = -ESHUTDOWN;
-		else
-			se_cmd->cmd_cnt = se_sess->cmd_cnt;
 	}
 	if (ret && ack_kref)
 		target_put_sess_cmd(se_cmd);
diff --git a/drivers/target/target_core_xcopy.c b/drivers/target/target_core_xcopy.c
index 49a83500c8b7..91ed015b588c 100644
--- a/drivers/target/target_core_xcopy.c
+++ b/drivers/target/target_core_xcopy.c
@@ -591,8 +591,8 @@ static int target_xcopy_read_source(
 		(unsigned long long)src_lba, transfer_length_block, src_bytes);
 
 	__target_init_cmd(se_cmd, &xcopy_pt_tfo, &xcopy_pt_sess, src_bytes,
-			  DMA_FROM_DEVICE, 0, &xpt_cmd.sense_buffer[0], 0);
-
+			  DMA_FROM_DEVICE, 0, &xpt_cmd.sense_buffer[0], 0,
+			  NULL);
 	rc = target_xcopy_setup_pt_cmd(&xpt_cmd, xop, src_dev, &cdb[0],
 				remote_port);
 	if (rc < 0) {
@@ -636,8 +636,8 @@ static int target_xcopy_write_destination(
 		(unsigned long long)dst_lba, transfer_length_block, dst_bytes);
 
 	__target_init_cmd(se_cmd, &xcopy_pt_tfo, &xcopy_pt_sess, dst_bytes,
-			  DMA_TO_DEVICE, 0, &xpt_cmd.sense_buffer[0], 0);
-
+			  DMA_TO_DEVICE, 0, &xpt_cmd.sense_buffer[0], 0,
+			  NULL);
 	rc = target_xcopy_setup_pt_cmd(&xpt_cmd, xop, dst_dev, &cdb[0],
 				remote_port);
 	if (rc < 0) {
diff --git a/drivers/usb/gadget/function/f_tcm.c b/drivers/usb/gadget/function/f_tcm.c
index 658e2e21fdd0..c21acebe8aae 100644
--- a/drivers/usb/gadget/function/f_tcm.c
+++ b/drivers/usb/gadget/function/f_tcm.c
@@ -1054,7 +1054,7 @@ static void usbg_cmd_work(struct work_struct *work)
 				  tv_nexus->tvn_se_sess->se_tpg->se_tpg_tfo,
 				  tv_nexus->tvn_se_sess, cmd->data_len, DMA_NONE,
 				  cmd->prio_attr, cmd->sense_iu.sense,
-				  cmd->unpacked_lun);
+				  cmd->unpacked_lun, NULL);
 		goto out;
 	}
 
@@ -1183,7 +1183,7 @@ static void bot_cmd_work(struct work_struct *work)
 				  tv_nexus->tvn_se_sess->se_tpg->se_tpg_tfo,
 				  tv_nexus->tvn_se_sess, cmd->data_len, DMA_NONE,
 				  cmd->prio_attr, cmd->sense_iu.sense,
-				  cmd->unpacked_lun);
+				  cmd->unpacked_lun, NULL);
 		goto out;
 	}
 
diff --git a/include/target/target_core_fabric.h b/include/target/target_core_fabric.h
index 65527174b8bc..d507e7885f17 100644
--- a/include/target/target_core_fabric.h
+++ b/include/target/target_core_fabric.h
@@ -151,9 +151,11 @@ void	transport_deregister_session_configfs(struct se_session *);
 void	transport_deregister_session(struct se_session *);
 
 
-void	__target_init_cmd(struct se_cmd *,
-		const struct target_core_fabric_ops *,
-		struct se_session *, u32, int, int, unsigned char *, u64);
+void	__target_init_cmd(struct se_cmd *cmd,
+		const struct target_core_fabric_ops *tfo,
+		struct se_session *sess, u32 data_length, int data_direction,
+		int task_attr, unsigned char *sense_buffer, u64 unpacked_lun,
+		struct target_cmd_counter *cmd_cnt);
 int	target_init_cmd(struct se_cmd *se_cmd, struct se_session *se_sess,
 		unsigned char *sense, u64 unpacked_lun, u32 data_length,
 		int task_attr, int data_dir, int flags);

From 6d256bee602b131bd4fbc92863b6a1210bcf6325 Mon Sep 17 00:00:00 2001
From: Mike Christie <michael.christie@oracle.com>
Date: Sat, 18 Mar 2023 20:56:15 -0500
Subject: [PATCH 4/9] scsi: target: iscsit: isert: Alloc per conn cmd counter

This has iscsit allocate a per conn cmd counter and converts iscsit/isert
to use it instead of the per session one.

Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-5-michael.christie@oracle.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/infiniband/ulp/isert/ib_isert.c   |  4 ++--
 drivers/target/iscsi/iscsi_target.c       |  4 ++--
 drivers/target/iscsi/iscsi_target_login.c | 17 +++++++----------
 drivers/target/target_core_transport.c    |  9 ++++++---
 include/target/target_core_fabric.h       |  3 +++
 5 files changed, 20 insertions(+), 17 deletions(-)

diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c
index 75404885cf98..f290cd49698e 100644
--- a/drivers/infiniband/ulp/isert/ib_isert.c
+++ b/drivers/infiniband/ulp/isert/ib_isert.c
@@ -2506,8 +2506,8 @@ isert_wait4cmds(struct iscsit_conn *conn)
 	isert_info("iscsit_conn %p\n", conn);
 
 	if (conn->sess) {
-		target_stop_session(conn->sess->se_sess);
-		target_wait_for_sess_cmds(conn->sess->se_sess);
+		target_stop_cmd_counter(conn->cmd_cnt);
+		target_wait_for_cmds(conn->cmd_cnt);
 	}
 }
 
diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
index 87927a36f90d..11115c207844 100644
--- a/drivers/target/iscsi/iscsi_target.c
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -1193,7 +1193,7 @@ int iscsit_setup_scsi_cmd(struct iscsit_conn *conn, struct iscsit_cmd *cmd,
 			  conn->sess->se_sess, be32_to_cpu(hdr->data_length),
 			  cmd->data_direction, sam_task_attr,
 			  cmd->sense_buffer + 2, scsilun_to_int(&hdr->lun),
-			  conn->sess->se_sess->cmd_cnt);
+			  conn->cmd_cnt);
 
 	pr_debug("Got SCSI Command, ITT: 0x%08x, CmdSN: 0x%08x,"
 		" ExpXferLen: %u, Length: %u, CID: %hu\n", hdr->itt,
@@ -2057,7 +2057,7 @@ iscsit_handle_task_mgt_cmd(struct iscsit_conn *conn, struct iscsit_cmd *cmd,
 			  conn->sess->se_sess, 0, DMA_NONE,
 			  TCM_SIMPLE_TAG, cmd->sense_buffer + 2,
 			  scsilun_to_int(&hdr->lun),
-			  conn->sess->se_sess->cmd_cnt);
+			  conn->cmd_cnt);
 
 	target_get_sess_cmd(&cmd->se_cmd, true);
 
diff --git a/drivers/target/iscsi/iscsi_target_login.c b/drivers/target/iscsi/iscsi_target_login.c
index 8ab6c0107d89..274bdd7845ca 100644
--- a/drivers/target/iscsi/iscsi_target_login.c
+++ b/drivers/target/iscsi/iscsi_target_login.c
@@ -324,18 +324,8 @@ static int iscsi_login_zero_tsih_s1(
 		goto free_ops;
 	}
 
-	/*
-	 * This is temp for iser. It will be moved to per conn in later
-	 * patches for iscsi.
-	 */
-	sess->se_sess->cmd_cnt = target_alloc_cmd_counter();
-	if (!sess->se_sess->cmd_cnt)
-		goto free_se_sess;
-
 	return 0;
 
-free_se_sess:
-	transport_free_session(sess->se_sess);
 free_ops:
 	kfree(sess->sess_ops);
 free_id:
@@ -1157,8 +1147,14 @@ static struct iscsit_conn *iscsit_alloc_conn(struct iscsi_np *np)
 		goto free_conn_cpumask;
 	}
 
+	conn->cmd_cnt = target_alloc_cmd_counter();
+	if (!conn->cmd_cnt)
+		goto free_conn_allowed_cpumask;
+
 	return conn;
 
+free_conn_allowed_cpumask:
+	free_cpumask_var(conn->allowed_cpumask);
 free_conn_cpumask:
 	free_cpumask_var(conn->conn_cpumask);
 free_conn_ops:
@@ -1172,6 +1168,7 @@ static struct iscsit_conn *iscsit_alloc_conn(struct iscsi_np *np)
 
 void iscsit_free_conn(struct iscsit_conn *conn)
 {
+	target_free_cmd_counter(conn->cmd_cnt);
 	free_cpumask_var(conn->allowed_cpumask);
 	free_cpumask_var(conn->conn_cpumask);
 	kfree(conn->conn_ops);
diff --git a/drivers/target/target_core_transport.c b/drivers/target/target_core_transport.c
index c395606ab1a9..86adff2a86ed 100644
--- a/drivers/target/target_core_transport.c
+++ b/drivers/target/target_core_transport.c
@@ -254,7 +254,7 @@ struct target_cmd_counter *target_alloc_cmd_counter(void)
 }
 EXPORT_SYMBOL_GPL(target_alloc_cmd_counter);
 
-static void target_free_cmd_counter(struct target_cmd_counter *cmd_cnt)
+void target_free_cmd_counter(struct target_cmd_counter *cmd_cnt)
 {
 	/*
 	 * Drivers like loop do not call target_stop_session during session
@@ -265,6 +265,7 @@ static void target_free_cmd_counter(struct target_cmd_counter *cmd_cnt)
 
 	percpu_ref_exit(&cmd_cnt->refcnt);
 }
+EXPORT_SYMBOL_GPL(target_free_cmd_counter);
 
 /**
  * transport_init_session - initialize a session object
@@ -3170,13 +3171,14 @@ static void target_stop_cmd_counter_confirm(struct percpu_ref *ref)
  * target_stop_cmd_counter - Stop new IO from being added to the counter.
  * @cmd_cnt: counter to stop
  */
-static void target_stop_cmd_counter(struct target_cmd_counter *cmd_cnt)
+void target_stop_cmd_counter(struct target_cmd_counter *cmd_cnt)
 {
 	pr_debug("Stopping command counter.\n");
 	if (!atomic_cmpxchg(&cmd_cnt->stopped, 0, 1))
 		percpu_ref_kill_and_confirm(&cmd_cnt->refcnt,
 					    target_stop_cmd_counter_confirm);
 }
+EXPORT_SYMBOL_GPL(target_stop_cmd_counter);
 
 /**
  * target_stop_session - Stop new IO from being queued on the session.
@@ -3192,7 +3194,7 @@ EXPORT_SYMBOL(target_stop_session);
  * target_wait_for_cmds - Wait for outstanding cmds.
  * @cmd_cnt: counter to wait for active I/O for.
  */
-static void target_wait_for_cmds(struct target_cmd_counter *cmd_cnt)
+void target_wait_for_cmds(struct target_cmd_counter *cmd_cnt)
 {
 	int ret;
 
@@ -3208,6 +3210,7 @@ static void target_wait_for_cmds(struct target_cmd_counter *cmd_cnt)
 	wait_for_completion(&cmd_cnt->stop_done);
 	pr_debug("Waiting for cmds done.\n");
 }
+EXPORT_SYMBOL_GPL(target_wait_for_cmds);
 
 /**
  * target_wait_for_sess_cmds - Wait for outstanding commands
diff --git a/include/target/target_core_fabric.h b/include/target/target_core_fabric.h
index d507e7885f17..b188b1e90e1e 100644
--- a/include/target/target_core_fabric.h
+++ b/include/target/target_core_fabric.h
@@ -133,7 +133,10 @@ struct se_session *target_setup_session(struct se_portal_group *,
 				struct se_session *, void *));
 void target_remove_session(struct se_session *);
 
+void target_stop_cmd_counter(struct target_cmd_counter *cmd_cnt);
+void target_wait_for_cmds(struct target_cmd_counter *cmd_cnt);
 struct target_cmd_counter *target_alloc_cmd_counter(void);
+void target_free_cmd_counter(struct target_cmd_counter *cmd_cnt);
 
 void transport_init_session(struct se_session *se_sess);
 struct se_session *transport_alloc_session(enum target_prot_op);

From 395cee83d02de3073211b04fc85724f4abc663ad Mon Sep 17 00:00:00 2001
From: Mike Christie <michael.christie@oracle.com>
Date: Sat, 18 Mar 2023 20:56:16 -0500
Subject: [PATCH 5/9] scsi: target: iscsit: Stop/wait on cmds during conn close

This fixes a bug added in commit f36199355c64 ("scsi: target: iscsi: Fix
cmd abort fabric stop race").

If we have multiple sessions to the same se_device we can hit a race where
a LUN_RESET on one session cleans up the se_cmds from under another
session which is being closed. This results in the closing session freeing
its conn/session structs while they are still in use.

The bug is:

 1. Session1 has IO se_cmd1.

 2. Session2 can also have se_cmds for I/O and optionally TMRs for ABORTS
    but then gets a LUN_RESET.

 3. The LUN_RESET on session2 sees the se_cmds on session1 and during the
    drain stages marks them all with CMD_T_ABORTED.

 4. session1 is now closed so iscsit_release_commands_from_conn() only sees
    se_cmds with the CMD_T_ABORTED bit set and returns immediately even
    though we have outstanding commands.

 5. session1's connection and session are freed.

 6. The backend request for se_cmd1 completes and it accesses the freed
    connection/session.

This hooks the iscsit layer into the cmd counter code, so we can wait for
all outstanding se_cmds before freeing the connection.

Fixes: f36199355c64 ("scsi: target: iscsi: Fix cmd abort fabric stop race")
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-6-michael.christie@oracle.com
Reviewed-by: Maurizio Lombardi <mlombard@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/target/iscsi/iscsi_target.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
index 11115c207844..83b007141229 100644
--- a/drivers/target/iscsi/iscsi_target.c
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -4245,6 +4245,16 @@ static void iscsit_release_commands_from_conn(struct iscsit_conn *conn)
 		iscsit_free_cmd(cmd, true);
 
 	}
+
+	/*
+	 * Wait on commands that were cleaned up via the aborted_task path.
+	 * LLDs that implement iscsit_wait_conn will already have waited for
+	 * commands.
+	 */
+	if (!conn->conn_transport->iscsit_wait_conn) {
+		target_stop_cmd_counter(conn->cmd_cnt);
+		target_wait_for_cmds(conn->cmd_cnt);
+	}
 }
 
 static void iscsit_stop_timers_for_cmds(

From d8990b5a4d065f38f35d69bcd627ec5a7f8330ca Mon Sep 17 00:00:00 2001
From: Dmitry Bogdanov <d.bogdanov@yadro.com>
Date: Sat, 18 Mar 2023 20:56:17 -0500
Subject: [PATCH 6/9] scsi: target: iscsit: Free cmds before session free

Commands from recovery entries are freed after session has been closed.
That leads to use-after-free at command free or NPE with such call trace:

Time2Retain timer expired for SID: 1, cleaning up iSCSI session.
BUG: kernel NULL pointer dereference, address: 0000000000000140
RIP: 0010:sbitmap_queue_clear+0x3a/0xa0
Call Trace:
 target_release_cmd_kref+0xd1/0x1f0 [target_core_mod]
 transport_generic_free_cmd+0xd1/0x180 [target_core_mod]
 iscsit_free_cmd+0x53/0xd0 [iscsi_target_mod]
 iscsit_free_connection_recovery_entries+0x29d/0x320 [iscsi_target_mod]
 iscsit_close_session+0x13a/0x140 [iscsi_target_mod]
 iscsit_check_post_dataout+0x440/0x440 [iscsi_target_mod]
 call_timer_fn+0x24/0x140

Move cleanup of recovery enrties to before session freeing.

Reported-by: Forza <forza@tnonline.net>
Signed-off-by: Dmitry Bogdanov <d.bogdanov@yadro.com>
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-7-michael.christie@oracle.com
Reviewed-by: Maurizio Lombardi <mlombard@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/target/iscsi/iscsi_target.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
index 83b007141229..26634dbea9b8 100644
--- a/drivers/target/iscsi/iscsi_target.c
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -4529,6 +4529,9 @@ int iscsit_close_session(struct iscsit_session *sess, bool can_sleep)
 	iscsit_stop_time2retain_timer(sess);
 	spin_unlock_bh(&se_tpg->session_lock);
 
+	if (sess->sess_ops->ErrorRecoveryLevel == 2)
+		iscsit_free_connection_recovery_entries(sess);
+
 	/*
 	 * transport_deregister_session_configfs() will clear the
 	 * struct se_node_acl->nacl_sess pointer now as a iscsi_np process context
@@ -4552,9 +4555,6 @@ int iscsit_close_session(struct iscsit_session *sess, bool can_sleep)
 
 	transport_deregister_session(sess->se_sess);
 
-	if (sess->sess_ops->ErrorRecoveryLevel == 2)
-		iscsit_free_connection_recovery_entries(sess);
-
 	iscsit_free_all_ooo_cmdsns(sess);
 
 	spin_lock_bh(&se_tpg->session_lock);

From 673db054d7a2b5a470d7a25baf65956d005ad729 Mon Sep 17 00:00:00 2001
From: Mike Christie <michael.christie@oracle.com>
Date: Sat, 18 Mar 2023 20:56:18 -0500
Subject: [PATCH 7/9] scsi: target: Fix multiple LUN_RESET handling

This fixes a bug where an initiator thinks a LUN_RESET has cleaned up
running commands when it hasn't. The bug was added in commit 51ec502a3266
("target: Delete tmr from list before processing").

The problem occurs when:

 1. We have N I/O cmds running in the target layer spread over 2 sessions.

 2. The initiator sends a LUN_RESET for each session.

 3. session1's LUN_RESET loops over all the running commands from both
    sessions and moves them to its local drain_task_list.

 4. session2's LUN_RESET does not see the LUN_RESET from session1 because
    the commit above has it remove itself. session2 also does not see any
    commands since the other reset moved them off the state lists.

 5. sessions2's LUN_RESET will then complete with a successful response.

 6. sessions2's inititor believes the running commands on its session are
    now cleaned up due to the successful response and cleans up the running
    commands from its side. It then restarts them.

 7. The commands do eventually complete on the backend and the target
    starts to return aborted task statuses for them. The initiator will
    either throw a invalid ITT error or might accidentally lookup a new
    task if the ITT has been reallocated already.

Fix the bug by reverting the patch, and serialize the execution of
LUN_RESETs and Preempt and Aborts.

Also prevent us from waiting on LUN_RESETs in core_tmr_drain_tmr_list,
because it turns out the original patch fixed a bug that was not
mentioned. For LUN_RESET1 core_tmr_drain_tmr_list can see a second
LUN_RESET and wait on it. Then the second reset will run
core_tmr_drain_tmr_list and see the first reset and wait on it resulting in
a deadlock.

Fixes: 51ec502a3266 ("target: Delete tmr from list before processing")
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-8-michael.christie@oracle.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/target/target_core_device.c |  1 +
 drivers/target/target_core_tmr.c    | 26 +++++++++++++++++++++++---
 include/target/target_core_base.h   |  1 +
 3 files changed, 25 insertions(+), 3 deletions(-)

diff --git a/drivers/target/target_core_device.c b/drivers/target/target_core_device.c
index f6e58410ec3f..aeb03136773d 100644
--- a/drivers/target/target_core_device.c
+++ b/drivers/target/target_core_device.c
@@ -782,6 +782,7 @@ struct se_device *target_alloc_device(struct se_hba *hba, const char *name)
 	spin_lock_init(&dev->t10_alua.lba_map_lock);
 
 	INIT_WORK(&dev->delayed_cmd_work, target_do_delayed_work);
+	mutex_init(&dev->lun_reset_mutex);
 
 	dev->t10_wwn.t10_dev = dev;
 	/*
diff --git a/drivers/target/target_core_tmr.c b/drivers/target/target_core_tmr.c
index 2b95b4550a63..4718db628222 100644
--- a/drivers/target/target_core_tmr.c
+++ b/drivers/target/target_core_tmr.c
@@ -188,14 +188,23 @@ static void core_tmr_drain_tmr_list(
 	 * LUN_RESET tmr..
 	 */
 	spin_lock_irqsave(&dev->se_tmr_lock, flags);
-	if (tmr)
-		list_del_init(&tmr->tmr_list);
 	list_for_each_entry_safe(tmr_p, tmr_pp, &dev->dev_tmr_list, tmr_list) {
+		if (tmr_p == tmr)
+			continue;
+
 		cmd = tmr_p->task_cmd;
 		if (!cmd) {
 			pr_err("Unable to locate struct se_cmd for TMR\n");
 			continue;
 		}
+
+		/*
+		 * We only execute one LUN_RESET at a time so we can't wait
+		 * on them below.
+		 */
+		if (tmr_p->function == TMR_LUN_RESET)
+			continue;
+
 		/*
 		 * If this function was called with a valid pr_res_key
 		 * parameter (eg: for PROUT PREEMPT_AND_ABORT service action
@@ -379,14 +388,25 @@ int core_tmr_lun_reset(
 				tmr_nacl->initiatorname);
 		}
 	}
+
+
+	/*
+	 * We only allow one reset or preempt and abort to execute at a time
+	 * to prevent one call from claiming all the cmds causing a second
+	 * call from returning while cmds it should have waited on are still
+	 * running.
+	 */
+	mutex_lock(&dev->lun_reset_mutex);
+
 	pr_debug("LUN_RESET: %s starting for [%s], tas: %d\n",
 		(preempt_and_abort_list) ? "Preempt" : "TMR",
 		dev->transport->name, tas);
-
 	core_tmr_drain_tmr_list(dev, tmr, preempt_and_abort_list);
 	core_tmr_drain_state_list(dev, prout_cmd, tmr_sess, tas,
 				preempt_and_abort_list);
 
+	mutex_unlock(&dev->lun_reset_mutex);
+
 	/*
 	 * Clear any legacy SPC-2 reservation when called during
 	 * LOGICAL UNIT RESET
diff --git a/include/target/target_core_base.h b/include/target/target_core_base.h
index bd299790e99c..8cc42ad65c92 100644
--- a/include/target/target_core_base.h
+++ b/include/target/target_core_base.h
@@ -872,6 +872,7 @@ struct se_device {
 	struct rcu_head		rcu_head;
 	int			queue_cnt;
 	struct se_device_queue	*queues;
+	struct mutex		lun_reset_mutex;
 };
 
 struct target_opcode_descriptor {

From cc79da306ebb2edb700c3816b90219223182ac3c Mon Sep 17 00:00:00 2001
From: Mike Christie <michael.christie@oracle.com>
Date: Sat, 18 Mar 2023 20:56:19 -0500
Subject: [PATCH 8/9] scsi: target: iscsit: Fix TAS handling during conn
 cleanup

Fix a bug added in commit f36199355c64 ("scsi: target: iscsi: Fix cmd abort
fabric stop race").

If CMD_T_TAS is set on the se_cmd we must call iscsit_free_cmd() to do the
last put on the cmd and free it, because the connection is down and we will
not up sending the response and doing the put from the normal I/O
path.

Add a check for CMD_T_TAS in iscsit_release_commands_from_conn() so we now
detect this case and run iscsit_free_cmd().

Fixes: f36199355c64 ("scsi: target: iscsi: Fix cmd abort fabric stop race")
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-9-michael.christie@oracle.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/target/iscsi/iscsi_target.c | 16 +++++++++-------
 1 file changed, 9 insertions(+), 7 deletions(-)

diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
index 26634dbea9b8..07e196b44b91 100644
--- a/drivers/target/iscsi/iscsi_target.c
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -4220,9 +4220,12 @@ static void iscsit_release_commands_from_conn(struct iscsit_conn *conn)
 	list_for_each_entry_safe(cmd, cmd_tmp, &tmp_list, i_conn_node) {
 		struct se_cmd *se_cmd = &cmd->se_cmd;
 
-		if (se_cmd->se_tfo != NULL) {
-			spin_lock_irq(&se_cmd->t_state_lock);
-			if (se_cmd->transport_state & CMD_T_ABORTED) {
+		if (!se_cmd->se_tfo)
+			continue;
+
+		spin_lock_irq(&se_cmd->t_state_lock);
+		if (se_cmd->transport_state & CMD_T_ABORTED) {
+			if (!(se_cmd->transport_state & CMD_T_TAS))
 				/*
 				 * LIO's abort path owns the cleanup for this,
 				 * so put it back on the list and let
@@ -4230,11 +4233,10 @@ static void iscsit_release_commands_from_conn(struct iscsit_conn *conn)
 				 */
 				list_move_tail(&cmd->i_conn_node,
 					       &conn->conn_cmd_list);
-			} else {
-				se_cmd->transport_state |= CMD_T_FABRIC_STOP;
-			}
-			spin_unlock_irq(&se_cmd->t_state_lock);
+		} else {
+			se_cmd->transport_state |= CMD_T_FABRIC_STOP;
 		}
+		spin_unlock_irq(&se_cmd->t_state_lock);
 	}
 	spin_unlock_bh(&conn->cmd_lock);
 

From ea87981a0ee8fb8ced1c87d004a541b60623ff97 Mon Sep 17 00:00:00 2001
From: Dmitry Bogdanov <d.bogdanov@yadro.com>
Date: Sat, 18 Mar 2023 20:56:20 -0500
Subject: [PATCH 9/9] scsi: target: iscsi: Handle abort for WRITE_PENDING cmds

Sometimes an initiator does not send data for a WRITE command and tries to
abort it. The abort hangs waiting for frontend driver completion. iSCSI
driver waits for data and that timeout eventually initiates connection
reinstatment. The connection closing releases the commands in the
connection, but those aborted commands still did not handle the abort and
did not decrease a command ref counter. Because of that the connection
reinstatement hangs indefinitely and prevents re-login for that initiator.

Add handling in TCM of the abort for the WRITE_PENDING commands at
connection closing moment to make it possible to release them.

Signed-off-by: Dmitry Bogdanov <d.bogdanov@yadro.com>
[mnc: Rebase and expand comment]
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Link: https://lore.kernel.org/r/20230319015620.96006-10-michael.christie@oracle.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
---
 drivers/target/iscsi/iscsi_target.c | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/drivers/target/iscsi/iscsi_target.c b/drivers/target/iscsi/iscsi_target.c
index 07e196b44b91..834cce50f9b0 100644
--- a/drivers/target/iscsi/iscsi_target.c
+++ b/drivers/target/iscsi/iscsi_target.c
@@ -26,6 +26,7 @@
 #include <target/target_core_base.h>
 #include <target/target_core_fabric.h>
 
+#include <target/target_core_backend.h>
 #include <target/iscsi/iscsi_target_core.h>
 #include "iscsi_target_parameters.h"
 #include "iscsi_target_seq_pdu_list.h"
@@ -4236,6 +4237,16 @@ static void iscsit_release_commands_from_conn(struct iscsit_conn *conn)
 		} else {
 			se_cmd->transport_state |= CMD_T_FABRIC_STOP;
 		}
+
+		if (cmd->se_cmd.t_state == TRANSPORT_WRITE_PENDING) {
+			/*
+			 * We never submitted the cmd to LIO core, so we have
+			 * to tell LIO to perform the completion process.
+			 */
+			spin_unlock_irq(&se_cmd->t_state_lock);
+			target_complete_cmd(&cmd->se_cmd, SAM_STAT_TASK_ABORTED);
+			continue;
+		}
 		spin_unlock_irq(&se_cmd->t_state_lock);
 	}
 	spin_unlock_bh(&conn->cmd_lock);