scsi: qla2xxx: Fix hang on NVMe command timeouts

The abort callback gets called only when it gets posted to firmware. The
refcounting is done properly in the callback. On internal errors, the
callback is not invoked leading to a hung I/O. Fix this by having separate
error code when command gets returned from firmware.

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arun Easi <[email protected]>
Signed-off-by: Nilesh Javali <[email protected]>
Signed-off-by: Martin K. Petersen <[email protected]>

Author: Arun Easi

drivers/scsi/qla2xxx/qla_def.h

@@ -5166,6 +5166,9 @@ struct secure_flash_update_block_pk {
 #define QLA_BUSY			0x107
 #define QLA_ALREADY_REGISTERED		0x109
 #define QLA_OS_TIMER_EXPIRED		0x10a
+#define QLA_ERR_NO_QPAIR		0x10b
+#define QLA_ERR_NOT_FOUND		0x10c
+#define QLA_ERR_FROM_FW			0x10d
 
 #define NVRAM_DELAY()		udelay(10)
 

drivers/scsi/qla2xxx/qla_init.c

@@ -157,7 +157,7 @@ int qla24xx_async_abort_cmd(srb_t *cmd_sp, bool wait)
 	sp = qla2xxx_get_qpair_sp(cmd_sp->vha, cmd_sp->qpair, cmd_sp->fcport,
 				  GFP_ATOMIC);
 	if (!sp)
-		return rval;
+		return QLA_MEMORY_ALLOC_FAILED;
 
 	abt_iocb = &sp->u.iocb_cmd;
 	sp->type = SRB_ABT_CMD;
@@ -190,7 +190,7 @@ int qla24xx_async_abort_cmd(srb_t *cmd_sp, bool wait)
 	if (wait) {
 		wait_for_completion(&abt_iocb->u.abt.comp);
 		rval = abt_iocb->u.abt.comp_status == CS_COMPLETE ?
-			QLA_SUCCESS : QLA_FUNCTION_FAILED;
+			QLA_SUCCESS : QLA_ERR_FROM_FW;
 		sp->free(sp);
 	}
 
@@ -1988,7 +1988,7 @@ qla24xx_async_abort_command(srb_t *sp)
 
 	if (handle == req->num_outstanding_cmds) {
 		/* Command not found. */
-		return QLA_FUNCTION_FAILED;
+		return QLA_ERR_NOT_FOUND;
 	}
 	if (sp->type == SRB_FXIOCB_DCMD)
 		return qlafx00_fx_disc(vha, &vha->hw->mr.fcport,

drivers/scsi/qla2xxx/qla_mbx.c

@@ -3245,7 +3245,7 @@ qla24xx_abort_command(srb_t *sp)
 	if (sp->qpair)
 		req = sp->qpair->req;
 	else
-		return QLA_FUNCTION_FAILED;
+		return QLA_ERR_NO_QPAIR;
 
 	if (ql2xasynctmfenable)
 		return qla24xx_async_abort_command(sp);
@@ -3258,7 +3258,7 @@ qla24xx_abort_command(srb_t *sp)
 	spin_unlock_irqrestore(qpair->qp_lock_ptr, flags);
 	if (handle == req->num_outstanding_cmds) {
 		/* Command not found. */
-		return QLA_FUNCTION_FAILED;
+		return QLA_ERR_NOT_FOUND;
 	}
 
 	abt = dma_pool_zalloc(ha->s_dma_pool, GFP_KERNEL, &abt_dma);

drivers/scsi/qla2xxx/qla_nvme.c

@@ -227,11 +227,11 @@ static void qla_nvme_abort_work(struct work_struct *work)
 	srb_t *sp = priv->sp;
 	fc_port_t *fcport = sp->fcport;
 	struct qla_hw_data *ha = fcport->vha->hw;
-	int rval;
+	int rval, abts_done_called = 1;
 
 	ql_dbg(ql_dbg_io, fcport->vha, 0xffff,
-	       "%s called for sp=%p, hndl=%x on fcport=%p deleted=%d\n",
-	       __func__, sp, sp->handle, fcport, fcport->deleted);
+	       "%s called for sp=%p, hndl=%x on fcport=%p desc=%p deleted=%d\n",
+	       __func__, sp, sp->handle, fcport, sp->u.iocb_cmd.u.nvme.desc, fcport->deleted);
 
 	if (!ha->flags.fw_started || fcport->deleted == QLA_SESS_DELETED)
 		goto out;
@@ -251,12 +251,20 @@ static void qla_nvme_abort_work(struct work_struct *work)
 	    __func__, (rval != QLA_SUCCESS) ? "Failed to abort" : "Aborted",
 	    sp, sp->handle, fcport, rval);
 
+	/*
+	 * If async tmf is enabled, the abort callback is called only on
+	 * return codes QLA_SUCCESS and QLA_ERR_FROM_FW.
+	 */
+	if (ql2xasynctmfenable &&
+	    rval != QLA_SUCCESS && rval != QLA_ERR_FROM_FW)
+		abts_done_called = 0;
+
 	/*
 	 * Returned before decreasing kref so that I/O requests
 	 * are waited until ABTS complete. This kref is decreased
 	 * at qla24xx_abort_sp_done function.
 	 */
-	if (ql2xabts_wait_nvme && QLA_ABTS_WAIT_ENABLED(sp))
+	if (abts_done_called && ql2xabts_wait_nvme && QLA_ABTS_WAIT_ENABLED(sp))
 		return;
 out:
 	/* kref_get was done before work was schedule. */
@@ -804,14 +812,14 @@ void qla_nvme_abort_process_comp_status(struct abort_entry_24xx *abt, srb_t *ori
 	case CS_PORT_LOGGED_OUT:
 	/* BA_RJT was received for the ABTS */
 	case CS_PORT_CONFIG_CHG:
-		ql_dbg(ql_dbg_async + ql_dbg_mbx, vha, 0xf09d,
+		ql_dbg(ql_dbg_async, vha, 0xf09d,
 		       "Abort I/O IOCB completed with error, comp_status=%x\n",
 		comp_status);
 		break;
 
 	/* BA_RJT was received for the ABTS */
 	case CS_REJECT_RECEIVED:
-		ql_dbg(ql_dbg_async + ql_dbg_mbx, vha, 0xf09e,
+		ql_dbg(ql_dbg_async, vha, 0xf09e,
 		       "BA_RJT was received for the ABTS rjt_vendorUnique = %u",
 			abt->fw.ba_rjt_vendorUnique);
 		ql_dbg(ql_dbg_async + ql_dbg_mbx, vha, 0xf09e,
@@ -820,18 +828,18 @@ void qla_nvme_abort_process_comp_status(struct abort_entry_24xx *abt, srb_t *ori
 		break;
 
 	case CS_COMPLETE:
-		ql_dbg(ql_dbg_async + ql_dbg_mbx, vha, 0xf09f,
+		ql_dbg(ql_dbg_async + ql_dbg_verbose, vha, 0xf09f,
 		       "IOCB request is completed successfully comp_status=%x\n",
 		comp_status);
 		break;
 
 	case CS_IOCB_ERROR:
-		ql_dbg(ql_dbg_async + ql_dbg_mbx, vha, 0xf0a0,
+		ql_dbg(ql_dbg_async, vha, 0xf0a0,
 		       "IOCB request is failed, comp_status=%x\n", comp_status);
 		break;
 
 	default:
-		ql_dbg(ql_dbg_async + ql_dbg_mbx, vha, 0xf0a1,
+		ql_dbg(ql_dbg_async, vha, 0xf0a1,
 		       "Invalid Abort IO IOCB Completion Status %x\n",
 		comp_status);
 		break;