------- Comment From dnban...@us.ibm.com 2018-04-23 23:49 EDT-------
I decided to take a look at qla2xxx driver's free and delete paths a little
more since my gut feeling was that these kinds of issues have to be
encountered by others too. Looking a little deeper I discovered these:
(Note this was from a quick perusal)
########################################
commit d8630bb95f46ea118dede63bd75533faa64f9612
Author: Quinn Tran <quinn.t...@cavium.com>
Date: Thu Dec 28 12:33:43 2017 -0800
scsi: qla2xxx: Serialize session deletion by using work_lock
for session deletion, replace sess_lock with work_lock.
Under certain case sess_lock is not feasiable to acquire.
The lock is needed temporarily to make sure a single
call to schedule of the work element.
########################################
commit 9cd883f07a54e5301d51e259acd250bb035996be
<does as part of its work>
+ /* use cancel to push work element through before re-queue */
+ cancel_work_sync(&sess->del_work);
INIT_WORK(&sess->del_work, qla24xx_delete_sess_fn);
queue_work(sess->vha->hw->wq, &sess->del_work);
########################################
commit 1ae634eb28533b82f9777a47c1ade44cb8c0182b
Author: Quinn Tran <quinn.t...@cavium.com>
Date: Thu Dec 28 12:33:44 2017 -0800
scsi: qla2xxx: Serialize session free in qlt_free_session_done
Add free_pending flag to serialize queueing of
free_work element onto the work queue
Signed-off-by: Quinn Tran <quinn.t...@cavium.com>
Signed-off-by: Himanshu Madhani <himanshu.madh...@cavium.com>
Signed-off-by: Martin K. Petersen <martin.peter...@oracle.com>
diff --git a/drivers/scsi/qla2xxx/qla_target.c
b/drivers/scsi/qla2xxx/qla_target.c
index 72b452d..0d3c3f6 100644
--- a/drivers/scsi/qla2xxx/qla_target.c
+++ b/drivers/scsi/qla2xxx/qla_target.c
@@ -1105,6 +1105,7 @@ static void qlt_free_session_done(struct work_struct
*work)
sess->plogi_link[QLT_PLOGI_LINK_SAME_WWN] = NULL;
}
}
+
spin_unlock_irqrestore(&ha->tgt.sess_lock, flags);
ql_dbg(ql_dbg_tgt_mgt, vha, 0xf001,
@@ -1118,6 +1119,9 @@ static void qlt_free_session_done(struct work_struct
*work)
wake_up_all(&vha->fcport_waitQ);
base_vha = pci_get_drvdata(ha->pdev);
+
+ sess->free_pending = 0;
+
if (test_bit(PFLG_DRIVER_REMOVING, &base_vha->pci_flags))
return;
@@ -1140,11 +1144,20 @@ static void qlt_free_session_done(struct work_struct
*work)
void qlt_unreg_sess(struct fc_port *sess)
{
struct scsi_qla_host *vha = sess->vha;
+ unsigned long flags;
ql_dbg(ql_dbg_disc, sess->vha, 0x210a,
"%s sess %p for deletion %8phC\n",
__func__, sess, sess->port_name);
+ spin_lock_irqsave(&sess->vha->work_lock, flags);
+ if (sess->free_pending) {
+ spin_unlock_irqrestore(&sess->vha->work_lock, flags);
+ return;
+ }
+ sess->free_pending = 1;
+ spin_unlock_irqrestore(&sess->vha->work_lock, flags);
+
if (sess->se_sess)
vha->hw->tgt.tgt_ops->clear_nacl_from_fcport_map(sess);
########################################
The last one is obviously a much more refined and well-thought
version of the free_work changes...
Obviously the code given in #132 was an attempt to move the testing/debug
forward and validate the cause analysis for the crashes.
Going forward, these changes (the relevant set -I just did a very quick
walk through ...the changes need to be picked carefully) need to be
selected with extreme diligence and pulled into the kernel (Canonical?).
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1762844
Title:
ISST-LTE:KVM:Ubuntu1804:BostonLC:boslcp3: Host crashed & enters into
xmon after moving to 4.15.0-15.16 kernel
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu-power-systems/+bug/1762844/+subscriptions
--
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
