diff --git a/drivers/scsi/lpfc/lpfc_hw4.h b/drivers/scsi/lpfc/lpfc_hw4.h
index 37c547b4bc78..a2f372d14eaa 100644
--- a/drivers/scsi/lpfc/lpfc_hw4.h
+++ b/drivers/scsi/lpfc/lpfc_hw4.h
@@ -4601,11 +4601,18 @@ union lpfc_wqe128 {
 	struct fcp_icmnd64_wqe fcp_icmd;
 	struct fcp_iread64_wqe fcp_iread;
 	struct fcp_iwrite64_wqe fcp_iwrite;
+	struct abort_cmd_wqe abort_cmd;
+	struct create_xri_wqe create_xri;
+	struct xmit_bcast64_wqe xmit_bcast64;
+	struct xmit_seq64_wqe xmit_sequence;
+	struct xmit_bls_rsp64_wqe xmit_bls_rsp;
+	struct xmit_els_rsp64_wqe xmit_els_rsp;
+	struct els_request64_wqe els_req;
+	struct gen_req64_wqe gen_req;
 	struct fcp_trsp64_wqe fcp_trsp;
 	struct fcp_tsend64_wqe fcp_tsend;
 	struct fcp_treceive64_wqe fcp_treceive;
-	struct xmit_seq64_wqe xmit_sequence;
-	struct gen_req64_wqe gen_req;
+	struct send_frame_wqe send_frame;
 };
 
 struct lpfc_grp_hdr {
diff --git a/drivers/scsi/lpfc/lpfc_nvme.c b/drivers/scsi/lpfc/lpfc_nvme.c
index 6ea6cc372647..6d215f27448f 100644
--- a/drivers/scsi/lpfc/lpfc_nvme.c
+++ b/drivers/scsi/lpfc/lpfc_nvme.c
@@ -275,14 +275,14 @@ lpfc_nvme_cmpl_gen_req(struct lpfc_hba *phba, struct lpfc_iocbq *cmdwqe,
 static int
 lpfc_nvme_gen_req(struct lpfc_vport *vport, struct lpfc_dmabuf *bmp,
 		  struct lpfc_dmabuf *inp,
-		 struct nvmefc_ls_req *pnvme_lsreq,
-	     void (*cmpl)(struct lpfc_hba *, struct lpfc_iocbq *,
-			   struct lpfc_wcqe_complete *),
-	     struct lpfc_nodelist *ndlp, uint32_t num_entry,
-	     uint32_t tmo, uint8_t retry)
+		  struct nvmefc_ls_req *pnvme_lsreq,
+		  void (*cmpl)(struct lpfc_hba *, struct lpfc_iocbq *,
+			       struct lpfc_wcqe_complete *),
+		  struct lpfc_nodelist *ndlp, uint32_t num_entry,
+		  uint32_t tmo, uint8_t retry)
 {
-	struct lpfc_hba  *phba = vport->phba;
-	union lpfc_wqe *wqe;
+	struct lpfc_hba *phba = vport->phba;
+	union lpfc_wqe128 *wqe;
 	struct lpfc_iocbq *genwqe;
 	struct ulp_bde64 *bpl;
 	struct ulp_bde64 bde;
@@ -628,8 +628,7 @@ lpfc_nvme_adj_fcp_sgls(struct lpfc_vport *vport,
 	 * the dma address.
 	 */
 
-	/* 128 byte wqe support here */
-	wqe = (union lpfc_wqe128 *)&lpfc_ncmd->cur_iocbq.wqe;
+	wqe = &lpfc_ncmd->cur_iocbq.wqe;
 
 	/*
 	 * Adjust the FCP_CMD and FCP_RSP DMA data and sge_len to
@@ -1048,7 +1047,7 @@ lpfc_nvme_prep_io_cmd(struct lpfc_vport *vport,
 	struct lpfc_hba *phba = vport->phba;
 	struct nvmefc_fcp_req *nCmd = lpfc_ncmd->nvmeCmd;
 	struct lpfc_iocbq *pwqeq = &(lpfc_ncmd->cur_iocbq);
-	union lpfc_wqe128 *wqe = (union lpfc_wqe128 *)&pwqeq->wqe;
+	union lpfc_wqe128 *wqe = &pwqeq->wqe;
 	uint32_t req_len;
 
 	if (!pnode || !NLP_CHK_NODE_ACT(pnode))
@@ -1187,7 +1186,7 @@ lpfc_nvme_prep_io_dma(struct lpfc_vport *vport,
 {
 	struct lpfc_hba *phba = vport->phba;
 	struct nvmefc_fcp_req *nCmd = lpfc_ncmd->nvmeCmd;
-	union lpfc_wqe128 *wqe = (union lpfc_wqe128 *)&lpfc_ncmd->cur_iocbq.wqe;
+	union lpfc_wqe128 *wqe = &lpfc_ncmd->cur_iocbq.wqe;
 	struct sli4_sge *sgl = lpfc_ncmd->nvme_sgl;
 	struct scatterlist *data_sg;
 	struct sli4_sge *first_data_sgl;
@@ -1595,7 +1594,7 @@ lpfc_nvme_fcp_abort(struct nvme_fc_local_port *pnvme_lport,
 	struct lpfc_iocbq *abts_buf;
 	struct lpfc_iocbq *nvmereq_wqe;
 	struct lpfc_nvme_fcpreq_priv *freqpriv;
-	union lpfc_wqe *abts_wqe;
+	union lpfc_wqe128 *abts_wqe;
 	unsigned long flags;
 	int ret_val;
 
@@ -2139,7 +2138,7 @@ lpfc_new_nvme_buf(struct lpfc_vport *vport, int num_to_alloc)
 			break;
 		}
 		pwqeq = &(lpfc_ncmd->cur_iocbq);
-		wqe = (union lpfc_wqe128 *)&pwqeq->wqe;
+		wqe = &pwqeq->wqe;
 
 		/* Allocate iotag for lpfc_ncmd->cur_iocbq. */
 		iotag = lpfc_sli_next_iotag(phba, pwqeq);
diff --git a/drivers/scsi/lpfc/lpfc_nvmet.c b/drivers/scsi/lpfc/lpfc_nvmet.c
index a1ad18153ca9..0cd45904aaa3 100644
--- a/drivers/scsi/lpfc/lpfc_nvmet.c
+++ b/drivers/scsi/lpfc/lpfc_nvmet.c
@@ -1151,7 +1151,7 @@ lpfc_nvmet_setup_io_context(struct lpfc_hba *phba)
 		}
 		ctx_buf->iocbq->iocb_flag = LPFC_IO_NVMET;
 		nvmewqe = ctx_buf->iocbq;
-		wqe = (union lpfc_wqe128 *)&nvmewqe->wqe;
+		wqe = &nvmewqe->wqe;
 		/* Initialize WQE */
 		memset(wqe, 0, sizeof(union lpfc_wqe));
 		/* Word 7 */
@@ -2025,7 +2025,7 @@ lpfc_nvmet_prep_ls_wqe(struct lpfc_hba *phba,
 {
 	struct lpfc_nodelist *ndlp;
 	struct lpfc_iocbq *nvmewqe;
-	union lpfc_wqe *wqe;
+	union lpfc_wqe128 *wqe;
 
 	if (!lpfc_is_link_up(phba)) {
 		lpfc_printf_log(phba, KERN_ERR, LOG_NVME_DISC,
@@ -2208,7 +2208,7 @@ lpfc_nvmet_prep_fcp_wqe(struct lpfc_hba *phba,
 	if (((ctxp->state == LPFC_NVMET_STE_RCV) &&
 	    (ctxp->entry_cnt == 1)) ||
 	    (ctxp->state == LPFC_NVMET_STE_DATA)) {
-		wqe = (union lpfc_wqe128 *)&nvmewqe->wqe;
+		wqe = &nvmewqe->wqe;
 	} else {
 		lpfc_printf_log(phba, KERN_ERR, LOG_NVME_IOERR,
 				"6111 Wrong state NVMET FCP: %d  cnt %d\n",
@@ -2750,7 +2750,7 @@ lpfc_nvmet_unsol_issue_abort(struct lpfc_hba *phba,
 {
 	struct lpfc_nvmet_tgtport *tgtp;
 	struct lpfc_iocbq *abts_wqeq;
-	union lpfc_wqe *wqe_abts;
+	union lpfc_wqe128 *wqe_abts;
 	struct lpfc_nodelist *ndlp;
 
 	lpfc_printf_log(phba, KERN_INFO, LOG_NVME_ABTS,
@@ -2845,7 +2845,7 @@ lpfc_nvmet_sol_fcp_issue_abort(struct lpfc_hba *phba,
 {
 	struct lpfc_nvmet_tgtport *tgtp;
 	struct lpfc_iocbq *abts_wqeq;
-	union lpfc_wqe *abts_wqe;
+	union lpfc_wqe128 *abts_wqe;
 	struct lpfc_nodelist *ndlp;
 	unsigned long flags;
 	int rc;
@@ -3035,7 +3035,7 @@ lpfc_nvmet_unsol_ls_issue_abort(struct lpfc_hba *phba,
 {
 	struct lpfc_nvmet_tgtport *tgtp;
 	struct lpfc_iocbq *abts_wqeq;
-	union lpfc_wqe *wqe_abts;
+	union lpfc_wqe128 *wqe_abts;
 	unsigned long flags;
 	int rc;
 
diff --git a/drivers/scsi/lpfc/lpfc_sli.c b/drivers/scsi/lpfc/lpfc_sli.c
index d20cf51ca15d..7999a40e1370 100644
--- a/drivers/scsi/lpfc/lpfc_sli.c
+++ b/drivers/scsi/lpfc/lpfc_sli.c
@@ -109,7 +109,7 @@ lpfc_get_iocb_from_iocbq(struct lpfc_iocbq *iocbq)
  * The caller is expected to hold the hbalock when calling this routine.
  **/
 static int
-lpfc_sli4_wq_put(struct lpfc_queue *q, union lpfc_wqe *wqe)
+lpfc_sli4_wq_put(struct lpfc_queue *q, union lpfc_wqe128 *wqe)
 {
 	union lpfc_wqe *temp_wqe;
 	struct lpfc_register doorbell;
@@ -8881,7 +8881,7 @@ lpfc_sli4_bpl2sgl(struct lpfc_hba *phba, struct lpfc_iocbq *piocbq,
  **/
 static int
 lpfc_sli4_iocb2wqe(struct lpfc_hba *phba, struct lpfc_iocbq *iocbq,
-		union lpfc_wqe *wqe)
+		union lpfc_wqe128 *wqe)
 {
 	uint32_t xmit_len = 0, total_len = 0;
 	uint8_t ct = 0;
@@ -9096,29 +9096,27 @@ lpfc_sli4_iocb2wqe(struct lpfc_hba *phba, struct lpfc_iocbq *iocbq,
 		if (phba->fcp_embed_io) {
 			struct lpfc_scsi_buf *lpfc_cmd;
 			struct sli4_sge *sgl;
-			union lpfc_wqe128 *wqe128;
 			struct fcp_cmnd *fcp_cmnd;
 			uint32_t *ptr;
 
 			/* 128 byte wqe support here */
-			wqe128 = (union lpfc_wqe128 *)wqe;
 
 			lpfc_cmd = iocbq->context1;
 			sgl = (struct sli4_sge *)lpfc_cmd->fcp_bpl;
 			fcp_cmnd = lpfc_cmd->fcp_cmnd;
 
 			/* Word 0-2 - FCP_CMND */
-			wqe128->generic.bde.tus.f.bdeFlags =
+			wqe->generic.bde.tus.f.bdeFlags =
 				BUFF_TYPE_BDE_IMMED;
-			wqe128->generic.bde.tus.f.bdeSize = sgl->sge_len;
-			wqe128->generic.bde.addrHigh = 0;
-			wqe128->generic.bde.addrLow =  88;  /* Word 22 */
+			wqe->generic.bde.tus.f.bdeSize = sgl->sge_len;
+			wqe->generic.bde.addrHigh = 0;
+			wqe->generic.bde.addrLow =  88;  /* Word 22 */
 
-			bf_set(wqe_wqes, &wqe128->fcp_iwrite.wqe_com, 1);
-			bf_set(wqe_dbde, &wqe128->fcp_iwrite.wqe_com, 0);
+			bf_set(wqe_wqes, &wqe->fcp_iwrite.wqe_com, 1);
+			bf_set(wqe_dbde, &wqe->fcp_iwrite.wqe_com, 0);
 
 			/* Word 22-29  FCP CMND Payload */
-			ptr = &wqe128->words[22];
+			ptr = &wqe->words[22];
 			memcpy(ptr, fcp_cmnd, sizeof(struct fcp_cmnd));
 		}
 		break;
@@ -9162,29 +9160,27 @@ lpfc_sli4_iocb2wqe(struct lpfc_hba *phba, struct lpfc_iocbq *iocbq,
 		if (phba->fcp_embed_io) {
 			struct lpfc_scsi_buf *lpfc_cmd;
 			struct sli4_sge *sgl;
-			union lpfc_wqe128 *wqe128;
 			struct fcp_cmnd *fcp_cmnd;
 			uint32_t *ptr;
 
 			/* 128 byte wqe support here */
-			wqe128 = (union lpfc_wqe128 *)wqe;
 
 			lpfc_cmd = iocbq->context1;
 			sgl = (struct sli4_sge *)lpfc_cmd->fcp_bpl;
 			fcp_cmnd = lpfc_cmd->fcp_cmnd;
 
 			/* Word 0-2 - FCP_CMND */
-			wqe128->generic.bde.tus.f.bdeFlags =
+			wqe->generic.bde.tus.f.bdeFlags =
 				BUFF_TYPE_BDE_IMMED;
-			wqe128->generic.bde.tus.f.bdeSize = sgl->sge_len;
-			wqe128->generic.bde.addrHigh = 0;
-			wqe128->generic.bde.addrLow =  88;  /* Word 22 */
+			wqe->generic.bde.tus.f.bdeSize = sgl->sge_len;
+			wqe->generic.bde.addrHigh = 0;
+			wqe->generic.bde.addrLow =  88;  /* Word 22 */
 
-			bf_set(wqe_wqes, &wqe128->fcp_iread.wqe_com, 1);
-			bf_set(wqe_dbde, &wqe128->fcp_iread.wqe_com, 0);
+			bf_set(wqe_wqes, &wqe->fcp_iread.wqe_com, 1);
+			bf_set(wqe_dbde, &wqe->fcp_iread.wqe_com, 0);
 
 			/* Word 22-29  FCP CMND Payload */
-			ptr = &wqe128->words[22];
+			ptr = &wqe->words[22];
 			memcpy(ptr, fcp_cmnd, sizeof(struct fcp_cmnd));
 		}
 		break;
@@ -9221,29 +9217,27 @@ lpfc_sli4_iocb2wqe(struct lpfc_hba *phba, struct lpfc_iocbq *iocbq,
 		if (phba->fcp_embed_io) {
 			struct lpfc_scsi_buf *lpfc_cmd;
 			struct sli4_sge *sgl;
-			union lpfc_wqe128 *wqe128;
 			struct fcp_cmnd *fcp_cmnd;
 			uint32_t *ptr;
 
 			/* 128 byte wqe support here */
-			wqe128 = (union lpfc_wqe128 *)wqe;
 
 			lpfc_cmd = iocbq->context1;
 			sgl = (struct sli4_sge *)lpfc_cmd->fcp_bpl;
 			fcp_cmnd = lpfc_cmd->fcp_cmnd;
 
 			/* Word 0-2 - FCP_CMND */
-			wqe128->generic.bde.tus.f.bdeFlags =
+			wqe->generic.bde.tus.f.bdeFlags =
 				BUFF_TYPE_BDE_IMMED;
-			wqe128->generic.bde.tus.f.bdeSize = sgl->sge_len;
-			wqe128->generic.bde.addrHigh = 0;
-			wqe128->generic.bde.addrLow =  88;  /* Word 22 */
+			wqe->generic.bde.tus.f.bdeSize = sgl->sge_len;
+			wqe->generic.bde.addrHigh = 0;
+			wqe->generic.bde.addrLow =  88;  /* Word 22 */
 
-			bf_set(wqe_wqes, &wqe128->fcp_icmd.wqe_com, 1);
-			bf_set(wqe_dbde, &wqe128->fcp_icmd.wqe_com, 0);
+			bf_set(wqe_wqes, &wqe->fcp_icmd.wqe_com, 1);
+			bf_set(wqe_dbde, &wqe->fcp_icmd.wqe_com, 0);
 
 			/* Word 22-29  FCP CMND Payload */
-			ptr = &wqe128->words[22];
+			ptr = &wqe->words[22];
 			memcpy(ptr, fcp_cmnd, sizeof(struct fcp_cmnd));
 		}
 		break;
@@ -9481,8 +9475,7 @@ __lpfc_sli_issue_iocb_s4(struct lpfc_hba *phba, uint32_t ring_number,
 			 struct lpfc_iocbq *piocb, uint32_t flag)
 {
 	struct lpfc_sglq *sglq;
-	union lpfc_wqe *wqe;
-	union lpfc_wqe128 wqe128;
+	union lpfc_wqe128 wqe;
 	struct lpfc_queue *wq;
 	struct lpfc_sli_ring *pring;
 
@@ -9502,9 +9495,7 @@ __lpfc_sli_issue_iocb_s4(struct lpfc_hba *phba, uint32_t ring_number,
 
 	/*
 	 * The WQE can be either 64 or 128 bytes,
-	 * so allocate space on the stack assuming the largest.
 	 */
-	wqe = (union lpfc_wqe *)&wqe128;
 
 	lockdep_assert_held(&phba->hbalock);
 
@@ -9554,10 +9545,10 @@ __lpfc_sli_issue_iocb_s4(struct lpfc_hba *phba, uint32_t ring_number,
 			return IOCB_ERROR;
 	}
 
-	if (lpfc_sli4_iocb2wqe(phba, piocb, wqe))
+	if (lpfc_sli4_iocb2wqe(phba, piocb, &wqe))
 		return IOCB_ERROR;
 
-	if (lpfc_sli4_wq_put(wq, wqe))
+	if (lpfc_sli4_wq_put(wq, &wqe))
 		return IOCB_ERROR;
 	lpfc_sli_ringtxcmpl_put(phba, pring, piocb);
 
@@ -10927,7 +10918,7 @@ lpfc_sli4_abort_nvme_io(struct lpfc_hba *phba, struct lpfc_sli_ring *pring,
 {
 	struct lpfc_vport *vport = cmdiocb->vport;
 	struct lpfc_iocbq *abtsiocbp;
-	union lpfc_wqe *abts_wqe;
+	union lpfc_wqe128 *abts_wqe;
 	int retval;
 
 	/*
@@ -19048,8 +19039,7 @@ lpfc_drain_txq(struct lpfc_hba *phba)
 	unsigned long iflags = 0;
 	char *fail_msg = NULL;
 	struct lpfc_sglq *sglq;
-	union lpfc_wqe128 wqe128;
-	union lpfc_wqe *wqe = (union lpfc_wqe *) &wqe128;
+	union lpfc_wqe128 wqe;
 	uint32_t txq_cnt = 0;
 
 	pring = lpfc_phba_elsring(phba);
@@ -19092,9 +19082,9 @@ lpfc_drain_txq(struct lpfc_hba *phba)
 		piocbq->sli4_xritag = sglq->sli4_xritag;
 		if (NO_XRI == lpfc_sli4_bpl2sgl(phba, piocbq, sglq))
 			fail_msg = "to convert bpl to sgl";
-		else if (lpfc_sli4_iocb2wqe(phba, piocbq, wqe))
+		else if (lpfc_sli4_iocb2wqe(phba, piocbq, &wqe))
 			fail_msg = "to convert iocb to wqe";
-		else if (lpfc_sli4_wq_put(phba->sli4_hba.els_wq, wqe))
+		else if (lpfc_sli4_wq_put(phba->sli4_hba.els_wq, &wqe))
 			fail_msg = " - Wq is full";
 		else
 			lpfc_sli_ringtxcmpl_put(phba, pring, piocbq);
@@ -19144,7 +19134,7 @@ lpfc_wqe_bpl2sgl(struct lpfc_hba *phba, struct lpfc_iocbq *pwqeq,
 	struct ulp_bde64 bde;
 	struct sli4_sge *sgl  = NULL;
 	struct lpfc_dmabuf *dmabuf;
-	union lpfc_wqe *wqe;
+	union lpfc_wqe128 *wqe;
 	int numBdes = 0;
 	int i = 0;
 	uint32_t offset = 0; /* accumulated offset in the sg request list */
@@ -19253,7 +19243,7 @@ int
 lpfc_sli4_issue_wqe(struct lpfc_hba *phba, uint32_t ring_number,
 		    struct lpfc_iocbq *pwqe)
 {
-	union lpfc_wqe *wqe = &pwqe->wqe;
+	union lpfc_wqe128 *wqe = &pwqe->wqe;
 	struct lpfc_nvmet_rcv_ctx *ctxp;
 	struct lpfc_queue *wq;
 	struct lpfc_sglq *sglq;
diff --git a/drivers/scsi/lpfc/lpfc_sli.h b/drivers/scsi/lpfc/lpfc_sli.h
index a3b1b5145d2b..ad7b2e0a2018 100644
--- a/drivers/scsi/lpfc/lpfc_sli.h
+++ b/drivers/scsi/lpfc/lpfc_sli.h
@@ -61,9 +61,8 @@ struct lpfc_iocbq {
 	struct lpfc_wcqe_complete wcqe_cmpl;	/* WQE cmpl */
 	uint64_t isr_timestamp;
 
-	/* Be careful here */
-	union lpfc_wqe wqe;	/* WQE cmd */
-	IOCB_t iocb;		/* For IOCB cmd or if we want 128 byte WQE */
+	union lpfc_wqe128 wqe;	/* SLI-4 */
+	IOCB_t iocb;		/* SLI-3 */
 
 	uint8_t rsvd2;
 	uint8_t priority;	/* OAS priority */