From: mchristi@redhat.com <mchristi@redhat.com>
Date: Mon, 13 Jul 2009 13:06:19 -0500
Subject: [scsi] bnx2i: fix host setup and libiscsi abort locking
Message-id: 1247508379-2860-1-git-send-email-mchristi@redhat.com
O-Subject: [PATCH] RHEL 5.4: bnx2i - fix bnx2i host setup and libiscsi abort locking
Bugzilla: 511096

From: Mike Christie <mchristi@redhat.com>

This is for BZ 511096.

The first part of the patch (the bnx2i parts) fixes a bug where
the host is not properly registered so the iscsi connection
setup will fail. This is not yet upstream. It was sent last
week here:
http://www.spinics.net/lists/linux-scsi/msg37301.html

The second part of the patch fixes a bug in the iscsi eh
abort locking. This code is taking the sesison lock from the
scsi eh thread, but not disabling bottom halves. This is needed
because the session lock can also be taken from a softirq or timer
handler. This part of the patch is not upstream and has not been set
yet. I just found the bug while testing the bnx2i part of the patch.

Broadcom is doing the testing for the bnx2i driver. They made
the patch and tested it. I also verified it and ran some
iscsi/scsi eh tests (started and stopped sessions while running dd
and ran dd and set the scsi cmd timer to 1 second) while I was at it.

This patch was made over these patches:
http://post-office.corp.redhat.com/archives/rhkernel-list/2009-July/msg00221.html

diff --git a/drivers/scsi/bnx2i/bnx2i_init.c b/drivers/scsi/bnx2i/bnx2i_init.c
index 1f16439..a35e9fe 100644
--- a/drivers/scsi/bnx2i/bnx2i_init.c
+++ b/drivers/scsi/bnx2i/bnx2i_init.c
@@ -186,14 +186,17 @@ void bnx2i_stop(void *handle)
  */
 void bnx2i_register_device(struct bnx2i_hba *hba)
 {
+	int rc;
+
 	if (test_bit(ADAPTER_STATE_GOING_DOWN, &hba->adapter_state) ||
 	    test_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic)) {
 		return;
 	}
 
-	hba->cnic->register_device(hba->cnic, CNIC_ULP_ISCSI, hba);
+	rc = hba->cnic->register_device(hba->cnic, CNIC_ULP_ISCSI, hba);
 
-	set_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic);
+	if (!rc)
+		set_bit(BNX2I_CNIC_REGISTERED, &hba->reg_with_cnic);
 }
 
 
diff --git a/drivers/scsi/bnx2i/bnx2i_iscsi.c b/drivers/scsi/bnx2i/bnx2i_iscsi.c
index 4b8c949..5e93aaa 100644
--- a/drivers/scsi/bnx2i/bnx2i_iscsi.c
+++ b/drivers/scsi/bnx2i/bnx2i_iscsi.c
@@ -1664,15 +1664,18 @@ static struct iscsi_endpoint *bnx2i_ep_connect(struct Scsi_Host *shost,
 	struct iscsi_endpoint *ep;
 	int rc = 0;
 
-	if (shost)
+	if (shost) {
 		/* driver is given scsi host to work with */
 		hba = iscsi_host_priv(shost);
-	else
+		/* Register the device with cnic if not already done so */
+		bnx2i_register_device(hba);
+	} else
 		/*
 		 * check if the given destination can be reached through
 		 * a iscsi capable NetXtreme2 device
 		 */
 		hba = bnx2i_check_route(dst_addr);
+
 	if (!hba) {
 		rc = -ENOMEM;
 		goto check_busy;
diff --git a/drivers/scsi/libiscsi2.c b/drivers/scsi/libiscsi2.c
index 1b56432..9092a10 100644
--- a/drivers/scsi/libiscsi2.c
+++ b/drivers/scsi/libiscsi2.c
@@ -1911,10 +1911,10 @@ int iscsi2_eh_abort(struct scsi_cmnd *sc)
 		 * good and have never sent us a successful tmf response
 		 * then sent more data for the cmd.
 		 */
-		spin_lock(&session->lock);
+		spin_lock_bh(&session->lock);
 		fail_scsi_task(task, DID_ABORT);
 		conn->tmf_state = TMF_INITIAL;
-		spin_unlock(&session->lock);
+		spin_unlock_bh(&session->lock);
 		iscsi_start_tx(conn);
 		goto success_unlocked;
 	case TMF_TIMEDOUT: