scsi: qla2xxx: Fix unsafe removal from linked list

On NPIV delete, the VPort is taken off a linked list in an unsafe manner.
The check for VPort refcount should be done behind lock before taking off
the element.

[ 2733.016907] general protection fault: 0000 [#1] SMP NOPTI
[ 2733.016908] qla2xxx [0000:22:00.1]-7088:27: VP[4] deleted.
[ 2733.016912] CPU: 22 PID: 23481 Comm: qla2xxx_15_dpc Kdump: loaded Tainted:
G           OE KX    5.3.18-47-default #1 SLE15-SP3
[ 2733.016914] Hardware name: Dell Inc. PowerEdge R7525/0PYVT1, BIOS 2.1.4 02/17/2021
[ 2733.016929] RIP: 0010:qla2x00_abort_isp+0x90/0x850 [qla2xxx]
[ 2733.016933] RSP: 0018:ffffb9cfc91efe98 EFLAGS: 00010087
[ 2733.016935] RAX: 0000000000000292 RBX: dead000000000100 RCX: 0000000000000000
[ 2733.016936] RDX: 0000000000000001 RSI: ffff944bfeb99558 RDI: ffff944bfc4b4488
[ 2733.016937] RBP: ffff944bfc4b2868 R08: 00000000000187a2 R09: 0000000000000001
[ 2733.016937] R10: ffffb9cfc91efcc8 R11: 0000000000000001 R12: ffff944bfc4b4000
[ 2733.016938] R13: ffff944bfc4b4870 R14: ffff944bfc4b4488 R15: ffff944bda895c80
[ 2733.016939] FS:  0000000000000000(0000) GS:ffff944bfeb80000(0000) knlGS:0000000000000000
[ 2733.016940] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 2733.016940] CR2: 00007fc173e74458 CR3: 0000001ff57de000 CR4: 0000000000350ee0
[ 2733.016941] Call Trace:
[ 2733.016951]   qla2xxx_pci_error_detected+0x190/0x190 [qla2xxx]
[ 2733.016957]   qla2x00_do_dpc+0x560/0xa10 [qla2xxx]
[ 2733.016962]   kthread+0x10d/0x130
[ 2733.016963]   kthread_park+0xa0/0xa0
[ 2733.016966]   ret_from_fork+0x22/0x40

Link: https://lore.kernel.org/r/20210810043720.1137-9-njavali@marvell.com
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

1 files changed, 25 insertions, 17 deletions

diff --git a/drivers/scsi/qla2xxx/qla_mid.c b/drivers/scsi/qla2xxx/qla_mid.c
index 078d596dbd49..1c024055f8c5 100644
--- a/drivers/scsi/qla2xxx/qla_mid.c
+++ b/drivers/scsi/qla2xxx/qla_mid.c
@@ -65,7 +65,7 @@ qla24xx_deallocate_vp_id(scsi_qla_host_t *vha)
 	uint16_t vp_id;
 	struct qla_hw_data *ha = vha->hw;
 	unsigned long flags = 0;
-	u8 i;
+	u32 i, bailout;
 
 	mutex_lock(&ha->vport_lock);
 	/*
@@ -75,21 +75,29 @@ qla24xx_deallocate_vp_id(scsi_qla_host_t *vha)
 	 * ensures no active vp_list traversal while the vport is removed
 	 * from the queue)
 	 */
-	for (i = 0; i < 10; i++) {
-		if (wait_event_timeout(vha->vref_waitq,
-		    !atomic_read(&vha->vref_count), HZ) > 0)
+	bailout = 0;
+	for (i = 0; i < 500; i++) {
+		spin_lock_irqsave(&ha->vport_slock, flags);
+		if (atomic_read(&vha->vref_count) == 0) {
+			list_del(&vha->list);
+			qlt_update_vp_map(vha, RESET_VP_IDX);
+			bailout = 1;
+		}
+		spin_unlock_irqrestore(&ha->vport_slock, flags);
+
+		if (bailout)
 			break;
+		else
+			msleep(20);
 	}
-
-	spin_lock_irqsave(&ha->vport_slock, flags);
-	if (atomic_read(&vha->vref_count)) {
-		ql_dbg(ql_dbg_vport, vha, 0xfffa,
-		    "vha->vref_count=%u timeout\n", vha->vref_count.counter);
-		vha->vref_count = (atomic_t)ATOMIC_INIT(0);
+	if (!bailout) {
+		ql_log(ql_log_info, vha, 0xfffa,
+			"vha->vref_count=%u timeout\n", vha->vref_count.counter);
+		spin_lock_irqsave(&ha->vport_slock, flags);
+		list_del(&vha->list);
+		qlt_update_vp_map(vha, RESET_VP_IDX);
+		spin_unlock_irqrestore(&ha->vport_slock, flags);
 	}
-	list_del(&vha->list);
-	qlt_update_vp_map(vha, RESET_VP_IDX);
-	spin_unlock_irqrestore(&ha->vport_slock, flags);
 
 	vp_id = vha->vp_idx;
 	ha->num_vhosts--;
@@ -262,13 +270,13 @@ qla24xx_configure_vp(scsi_qla_host_t *vha)
 void
 qla2x00_alert_all_vps(struct rsp_que *rsp, uint16_t *mb)
 {
-	scsi_qla_host_t *vha;
+	scsi_qla_host_t *vha, *tvp;
 	struct qla_hw_data *ha = rsp->hw;
 	int i = 0;
 	unsigned long flags;
 
 	spin_lock_irqsave(&ha->vport_slock, flags);
-	list_for_each_entry(vha, &ha->vp_list, list) {
+	list_for_each_entry_safe(vha, tvp, &ha->vp_list, list) {
 		if (vha->vp_idx) {
 			if (test_bit(VPORT_DELETE, &vha->dpc_flags))
 				continue;
@@ -421,7 +429,7 @@ void
 qla2x00_do_dpc_all_vps(scsi_qla_host_t *vha)
 {
 	struct qla_hw_data *ha = vha->hw;
-	scsi_qla_host_t *vp;
+	scsi_qla_host_t *vp, *tvp;
 	unsigned long flags = 0;
 
 	if (vha->vp_idx)
@@ -435,7 +443,7 @@ qla2x00_do_dpc_all_vps(scsi_qla_host_t *vha)
 		return;
 
 	spin_lock_irqsave(&ha->vport_slock, flags);
-	list_for_each_entry(vp, &ha->vp_list, list) {
+	list_for_each_entry_safe(vp, tvp, &ha->vp_list, list) {
 		if (vp->vp_idx) {
 			atomic_inc(&vp->vref_count);
 			spin_unlock_irqrestore(&ha->vport_slock, flags);