[v2,05/37] qla2xxx: Chip reset uses wrong lock during IO flush.

Commit Message

Madhani, Himanshu Dec. 28, 2017, 8:33 p.m. UTC

From: Quinn Tran <[e-mail]>

As part of chip reset, all commands from all QPairs are
flushed. This patch fixes code to use Q Pair lock for flush
instead of using old hardware_lock.

Signed-off-by: Quinn Tran <[e-mail]>
Signed-off-by: Himanshu Madhani <[e-mail]>
---
 drivers/scsi/qla2xxx/qla_os.c | 158 ++++++++++++++++++++++--------------------
 1 file changed, 84 insertions(+), 74 deletions(-)

Patch

diff --git a/drivers/scsi/qla2xxx/qla_os.c b/drivers/scsi/qla2xxx/qla_os.c
index 029b95b2bd8a..2e4bfb7aabf7 100644
--- a/drivers/scsi/qla2xxx/qla_os.c
+++ b/drivers/scsi/qla2xxx/qla_os.c
@@ -1704,93 +1704,103 @@  qla2x00_loop_reset(scsi_qla_host_t *vha)
 	return QLA_SUCCESS;
 }
 
-void
-qla2x00_abort_all_cmds(scsi_qla_host_t *vha, int res)
+static void
+__qla2x00_abort_all_cmds(struct qla_qpair *qp, int res)
 {
-	int que, cnt, status;
+	int cnt, status;
 	unsigned long flags;
 	srb_t *sp;
+	scsi_qla_host_t *vha = qp->vha;
 	struct qla_hw_data *ha = vha->hw;
 	struct req_que *req;
 	struct qla_tgt *tgt = vha->vha_tgt.qla_tgt;
 	struct qla_tgt_cmd *cmd;
 	uint8_t trace = 0;
 
-	spin_lock_irqsave(&ha->hardware_lock, flags);
-	for (que = 0; que < ha->max_req_queues; que++) {
-		req = ha->req_q_map[que];
-		if (!req)
-			continue;
-		if (!req->outstanding_cmds)
-			continue;
-		for (cnt = 1; cnt < req->num_outstanding_cmds; cnt++) {
-			sp = req->outstanding_cmds[cnt];
-			if (sp) {
-				req->outstanding_cmds[cnt] = NULL;
-				if (sp->cmd_type == TYPE_SRB) {
-					if (sp->type == SRB_NVME_CMD ||
-					    sp->type == SRB_NVME_LS) {
-						sp_get(sp);
-						spin_unlock_irqrestore(
-						    &ha->hardware_lock, flags);
-						qla_nvme_abort(ha, sp);
-						spin_lock_irqsave(
-						    &ha->hardware_lock, flags);
-					} else if (GET_CMD_SP(sp) &&
-					    !ha->flags.eeh_busy &&
-					    (!test_bit(ABORT_ISP_ACTIVE,
-						&vha->dpc_flags)) &&
-					    (sp->type == SRB_SCSI_CMD)) {
-						/*
-						 * Don't abort commands in
-						 * adapter during EEH
-						 * recovery as it's not
-						 * accessible/responding.
-						 *
-						 * Get a reference to the sp
-						 * and drop the lock. The
-						 * reference ensures this
-						 * sp->done() call and not the
-						 * call in qla2xxx_eh_abort()
-						 * ends the SCSI command (with
-						 * result 'res').
-						 */
-						sp_get(sp);
-						spin_unlock_irqrestore(
-						    &ha->hardware_lock, flags);
-						status = qla2xxx_eh_abort(
-						    GET_CMD_SP(sp));
-						spin_lock_irqsave(
-						    &ha->hardware_lock, flags);
-						/*
-						 * Get rid of extra reference
-						 * if immediate exit from
-						 * ql2xxx_eh_abort
-						 */
-						if (status == FAILED &&
-						    (qla2x00_isp_reg_stat(ha)))
-							atomic_dec(
-							    &sp->ref_count);
-					}
-					sp->done(sp, res);
-				} else {
-					if (!vha->hw->tgt.tgt_ops || !tgt ||
-					    qla_ini_mode_enabled(vha)) {
-						if (!trace)
-							ql_dbg(ql_dbg_tgt_mgt,
-							    vha, 0xf003,
-							    "HOST-ABORT-HNDLR: dpc_flags=%lx. Target mode disabled\n",
-							    vha->dpc_flags);
-						continue;
-					}
-					cmd = (struct qla_tgt_cmd *)sp;
-					qlt_abort_cmd_on_host_reset(cmd->vha,
-					    cmd);
+	spin_lock_irqsave(qp->qp_lock_ptr, flags);
+	req = qp->req;
+	for (cnt = 1; cnt < req->num_outstanding_cmds; cnt++) {
+		sp = req->outstanding_cmds[cnt];
+		if (sp) {
+			req->outstanding_cmds[cnt] = NULL;
+			if (sp->cmd_type == TYPE_SRB) {
+				if (sp->type == SRB_NVME_CMD ||
+				    sp->type == SRB_NVME_LS) {
+					sp_get(sp);
+					spin_unlock_irqrestore(qp->qp_lock_ptr,
+					    flags);
+					qla_nvme_abort(ha, sp);
+					spin_lock_irqsave(qp->qp_lock_ptr,
+					    flags);
+				} else if (GET_CMD_SP(sp) &&
+				    !ha->flags.eeh_busy &&
+				    (!test_bit(ABORT_ISP_ACTIVE,
+					&vha->dpc_flags)) &&
+				    (sp->type == SRB_SCSI_CMD)) {
+					/*
+					 * Don't abort commands in
+					 * adapter during EEH
+					 * recovery as it's not
+					 * accessible/responding.
+					 *
+					 * Get a reference to the sp
+					 * and drop the lock. The
+					 * reference ensures this
+					 * sp->done() call and not the
+					 * call in qla2xxx_eh_abort()
+					 * ends the SCSI command (with
+					 * result 'res').
+					 */
+					sp_get(sp);
+					spin_unlock_irqrestore(qp->qp_lock_ptr,
+					    flags);
+					status = qla2xxx_eh_abort(
+					    GET_CMD_SP(sp));
+					spin_lock_irqsave(qp->qp_lock_ptr,
+					    flags);
+					/*
+					 * Get rid of extra reference
+					 * if immediate exit from
+					 * ql2xxx_eh_abort
+					 */
+					if (status == FAILED &&
+					    (qla2x00_isp_reg_stat(ha)))
+						atomic_dec(
+						    &sp->ref_count);
+				}
+				sp->done(sp, res);
+			} else {
+				if (!vha->hw->tgt.tgt_ops || !tgt ||
+				    qla_ini_mode_enabled(vha)) {
+					if (!trace)
+						ql_dbg(ql_dbg_tgt_mgt,
+						    vha, 0xf003,
+						    "HOST-ABORT-HNDLR: dpc_flags=%lx. Target mode disabled\n",
+						    vha->dpc_flags);
+					continue;
 				}
+				cmd = (struct qla_tgt_cmd *)sp;
+				qlt_abort_cmd_on_host_reset(cmd->vha, cmd);
 			}
 		}
 	}
-	spin_unlock_irqrestore(&ha->hardware_lock, flags);
+	spin_unlock_irqrestore(qp->qp_lock_ptr, flags);
+}
+
+void
+qla2x00_abort_all_cmds(scsi_qla_host_t *vha, int res)
+{
+	int que;
+	struct qla_hw_data *ha = vha->hw;
+
+	__qla2x00_abort_all_cmds(ha->base_qpair, res);
+
+	for (que = 0; que < ha->max_qpairs; que++) {
+		if (!ha->queue_pair_map[que])
+			continue;
+
+		__qla2x00_abort_all_cmds(ha->queue_pair_map[que], res);
+	}
 }
 
 static int