[PATCH v3] scsi: iscsi: Fix multiple iscsi session unbind event sent to userspace

[Wenchao Hao]

I found an issue that kernel would send ISCSI_KEVENT_UNBIND_SESSION
for multiple times which should be fixed.

This patch introduce target_state in iscsi_cls_session to make
sure session would send only one ISCSI_KEVENT_UNBIND_SESSION.

But this would break issue fixed in commit 13e60d3ba287 ("scsi: iscsi:
Report unbind session event when the target has been removed"). The issue
is iscsid died for any reason after it send unbind session to kernel, once
iscsid restart again, it loss kernel's ISCSI_KEVENT_UNBIND_SESSION event.

Now kernel think iscsi_cls_session has already sent an
ISCSI_KEVENT_UNBIND_SESSION event and would not send it any more. Which
would cause userspace unable to logout. Actually the session is in
invalid state(it's target_id is INVALID), iscsid should not sync this
session in it's restart.

So we need to check session's target state during iscsid restart,
if session is in unbound state, do not sync this session and perform
session teardown. It's reasonable because once a session is unbound, we
can not recover it any more(mainly because it's target id is INVALID)

Changes from V2:
- Make target bind state to a state kind rather than a bool.

Changes from V1:
- Using target_unbound rather than state to indicate session has been
unbound

Signed-off-by: Wenchao Hao <haowe...@huawei.com>
---
drivers/scsi/scsi_transport_iscsi.c | 49 +++++++++++++++++++++++++++++
include/scsi/scsi_transport_iscsi.h | 7 +++++
2 files changed, 56 insertions(+)

diff --git a/drivers/scsi/scsi_transport_iscsi.c b/drivers/scsi/scsi_transport_iscsi.c
index 5d21f07456c6..43d48ddfea07 100644
--- a/drivers/scsi/scsi_transport_iscsi.c
+++ b/drivers/scsi/scsi_transport_iscsi.c
@@ -1676,6 +1676,29 @@ static const char *iscsi_session_state_name(int state)
return name;
}

+static struct {
+ int value;
+ char *name;
+} iscsi_session_target_state_names[] = {
+ { ISCSI_SESSION_TARGET_UNBOUND, "UNBOUND" },
+ { ISCSI_SESSION_TARGET_BOUND, "BOUND" },
+ { ISCSI_SESSION_TARGET_UNBINDING, "UNBINDING" },
+};
+
+static const char *iscsi_session_target_state_name(int state)
+{
+ int i;
+ char *name = NULL;
+
+ for (i = 0; i < ARRAY_SIZE(iscsi_session_target_state_names); i++) {
+ if (iscsi_session_target_state_names[i].value == state) {
+ name = iscsi_session_target_state_names[i].name;
+ break;
+ }
+ }
+ return name;
+}
+
int iscsi_session_chkready(struct iscsi_cls_session *session)
{
int err;
@@ -1899,6 +1922,7 @@ static void __iscsi_unblock_session(struct work_struct *work)
cancel_delayed_work_sync(&session->recovery_work);
spin_lock_irqsave(&session->lock, flags);
session->state = ISCSI_SESSION_LOGGED_IN;
+ session->target_state = ISCSI_SESSION_TARGET_BOUND;
spin_unlock_irqrestore(&session->lock, flags);
/* start IO */
scsi_target_unblock(&session->dev, SDEV_RUNNING);
@@ -1961,6 +1985,15 @@ static void __iscsi_unbind_session(struct work_struct *work)
unsigned long flags;
unsigned int target_id;

+ spin_lock_irqsave(&session->lock, flags);
+ if (session->target_state != ISCSI_SESSION_TARGET_BOUND) {
+ ISCSI_DBG_TRANS_SESSION(session, "Abort unbind sesison\n");
+ spin_unlock_irqrestore(&session->lock, flags);
+ return;
+ }
+ session->target_state = ISCSI_SESSION_TARGET_UNBINDING;
+ spin_unlock_irqrestore(&session->lock, flags);
+
ISCSI_DBG_TRANS_SESSION(session, "Unbinding session\n");

/* Prevent new scans and make sure scanning is not in progress */
@@ -1984,6 +2017,9 @@ static void __iscsi_unbind_session(struct work_struct *work)

unbind_session_exit:
iscsi_session_event(session, ISCSI_KEVENT_UNBIND_SESSION);
+ spin_lock_irqsave(&session->lock, flags);
+ session->target_state = ISCSI_SESSION_TARGET_UNBOUND;
+ spin_unlock_irqrestore(&session->lock, flags);
ISCSI_DBG_TRANS_SESSION(session, "Completed target removal\n");
}

@@ -4324,6 +4360,16 @@ iscsi_session_attr(def_taskmgmt_tmo, ISCSI_PARAM_DEF_TASKMGMT_TMO, 0);
iscsi_session_attr(discovery_parent_idx, ISCSI_PARAM_DISCOVERY_PARENT_IDX, 0);
iscsi_session_attr(discovery_parent_type, ISCSI_PARAM_DISCOVERY_PARENT_TYPE, 0);

+static ssize_t
+show_priv_session_target_state(struct device *dev, struct device_attribute *attr,
+ char *buf)
+{
+ struct iscsi_cls_session *session = iscsi_dev_to_session(dev->parent);
+ return sysfs_emit(buf, "%s\n",
+ iscsi_session_target_state_name(session->target_state));
+}
+static ISCSI_CLASS_ATTR(priv_sess, target_state, S_IRUGO,
+ show_priv_session_target_state, NULL);
static ssize_t
show_priv_session_state(struct device *dev, struct device_attribute *attr,
char *buf)
@@ -4426,6 +4472,7 @@ static struct attribute *iscsi_session_attrs[] = {
&dev_attr_sess_boot_target.attr,
&dev_attr_priv_sess_recovery_tmo.attr,
&dev_attr_priv_sess_state.attr,
+ &dev_attr_priv_sess_target_state.attr,
&dev_attr_priv_sess_creator.attr,
&dev_attr_sess_chap_out_idx.attr,
&dev_attr_sess_chap_in_idx.attr,
@@ -4539,6 +4586,8 @@ static umode_t iscsi_session_attr_is_visible(struct kobject *kobj,
return S_IRUGO | S_IWUSR;
else if (attr == &dev_attr_priv_sess_state.attr)
return S_IRUGO;
+ else if (attr == &dev_attr_priv_sess_target_state.attr)
+ return S_IRUGO;
else if (attr == &dev_attr_priv_sess_creator.attr)
return S_IRUGO;
else if (attr == &dev_attr_priv_sess_target_id.attr)
diff --git a/include/scsi/scsi_transport_iscsi.h b/include/scsi/scsi_transport_iscsi.h
index 9acb8422f680..377b039eefdd 100644
--- a/include/scsi/scsi_transport_iscsi.h
+++ b/include/scsi/scsi_transport_iscsi.h
@@ -236,6 +236,12 @@ enum {
ISCSI_SESSION_FREE,
};

+enum {
+ ISCSI_SESSION_TARGET_UNBOUND,
+ ISCSI_SESSION_TARGET_BOUND,
+ ISCSI_SESSION_TARGET_UNBINDING,
+};
+
#define ISCSI_MAX_TARGET -1

struct iscsi_cls_session {
@@ -264,6 +270,7 @@ struct iscsi_cls_session {
*/
pid_t creator;
int state;
+ int target_state; /* session target bind state */
int sid; /* session id */
void *dd_data; /* LLD private data */
struct device dev; /* sysfs transport/container device */
--
2.32.0

[Mike Christie]

On 8/1/22 10:47 PM, Wenchao Hao wrote:
> I found an issue that kernel would send ISCSI_KEVENT_UNBIND_SESSION
> for multiple times which should be fixed.
>
> This patch introduce target_state in iscsi_cls_session to make
> sure session would send only one ISCSI_KEVENT_UNBIND_SESSION.
>
> But this would break issue fixed in commit 13e60d3ba287 ("scsi: iscsi:
> Report unbind session event when the target has been removed"). The issue
> is iscsid died for any reason after it send unbind session to kernel, once
> iscsid restart again, it loss kernel's ISCSI_KEVENT_UNBIND_SESSION event.
>
> Now kernel think iscsi_cls_session has already sent an
> ISCSI_KEVENT_UNBIND_SESSION event and would not send it any more. Which
> would cause userspace unable to logout. Actually the session is in
> invalid state(it's target_id is INVALID), iscsid should not sync this
> session in it's restart.
>
> So we need to check session's target state during iscsid restart,
> if session is in unbound state, do not sync this session and perform
> session teardown. It's reasonable because once a session is unbound, we
> can not recover it any more(mainly because it's target id is INVALID)
>

I think we are ok now. Do you have a link to the userspace parts so
I can make sure we have everything covered now?

[wenchao hao]

I updated the PR on github: https://github.com/open-iscsi/open-iscsi/pull/339