[SCSI] lpfc 8.3.29: T10 Diff fixes and enhancements

T10 Diff fixes and enhancements:

- Add SLI4 Lancer support for T10 DIF / BlockGuard (121980)
- Fix SLI4 BlockGuard behavior when protection data is generated by HBA (121980)
- Enhance debugfs for injecting T10 DIF errors (123966, 132966)
- Fix Incorrect usage of bghm for BlockGuard errors (127022)
Signed-off-by: Alex Iannicelli <alex.iannicelli@emulex.com>
Signed-off-by: James Smart <james.smart@emulex.com>
Signed-off-by: James Bottomley <JBottomley@Parallels.com>

drivers/scsi/lpfc/lpfc.h

@@ -534,6 +534,7 @@ struct lpfc_hba {
 	void (*lpfc_scsi_prep_cmnd)
 		(struct lpfc_vport *, struct lpfc_scsi_buf *,
 		 struct lpfc_nodelist *);
+
 	/* IOCB interface function jump table entries */
 	int (*__lpfc_sli_issue_iocb)
 		(struct lpfc_hba *, uint32_t,
@@ -541,8 +542,6 @@ struct lpfc_hba {
 	void (*__lpfc_sli_release_iocbq)(struct lpfc_hba *,
 			 struct lpfc_iocbq *);
 	int (*lpfc_hba_down_post)(struct lpfc_hba *phba);
-
-
 	IOCB_t * (*lpfc_get_iocb_from_iocbq)
 		(struct lpfc_iocbq *);
 	void (*lpfc_scsi_cmd_iocb_cmpl)
@@ -551,10 +550,12 @@ struct lpfc_hba {
 	/* MBOX interface function jump table entries */
 	int (*lpfc_sli_issue_mbox)
 		(struct lpfc_hba *, LPFC_MBOXQ_t *, uint32_t);
+
 	/* Slow-path IOCB process function jump table entries */
 	void (*lpfc_sli_handle_slow_ring_event)
 		(struct lpfc_hba *phba, struct lpfc_sli_ring *pring,
 		 uint32_t mask);
+
 	/* INIT device interface function jump table entries */
 	int (*lpfc_sli_hbq_to_firmware)
 		(struct lpfc_hba *, uint32_t, struct hbq_dmabuf *);
@@ -573,6 +574,10 @@ struct lpfc_hba {
 	int (*lpfc_selective_reset)
 		(struct lpfc_hba *);
 
+	int (*lpfc_bg_scsi_prep_dma_buf)
+		(struct lpfc_hba *, struct lpfc_scsi_buf *);
+	/* Add new entries here */
+
 	/* SLI4 specific HBA data structure */
 	struct lpfc_sli4_hba sli4_hba;
 
@@ -838,6 +843,7 @@ struct lpfc_hba {
 	struct dentry *debug_writeGuard; /* inject write guard_tag errors */
 	struct dentry *debug_writeApp;   /* inject write app_tag errors */
 	struct dentry *debug_writeRef;   /* inject write ref_tag errors */
+	struct dentry *debug_readGuard;  /* inject read guard_tag errors */
 	struct dentry *debug_readApp;    /* inject read app_tag errors */
 	struct dentry *debug_readRef;    /* inject read ref_tag errors */
 
@@ -845,10 +851,11 @@ struct lpfc_hba {
 	uint32_t lpfc_injerr_wgrd_cnt;
 	uint32_t lpfc_injerr_wapp_cnt;
 	uint32_t lpfc_injerr_wref_cnt;
+	uint32_t lpfc_injerr_rgrd_cnt;
 	uint32_t lpfc_injerr_rapp_cnt;
 	uint32_t lpfc_injerr_rref_cnt;
 	sector_t lpfc_injerr_lba;
-#define LPFC_INJERR_LBA_OFF	(sector_t)0xffffffffffffffff
+#define LPFC_INJERR_LBA_OFF	(sector_t)(-1)
 
 	struct dentry *debug_slow_ring_trc;
 	struct lpfc_debugfs_trc *slow_ring_trc;

drivers/scsi/lpfc/lpfc_debugfs.c

@@ -1019,6 +1019,8 @@ lpfc_debugfs_dif_err_read(struct file *file, char __user *buf,
 		cnt = snprintf(cbuf, 16, "%u\n", phba->lpfc_injerr_wapp_cnt);
 	else if (dent == phba->debug_writeRef)
 		cnt = snprintf(cbuf, 16, "%u\n", phba->lpfc_injerr_wref_cnt);
+	else if (dent == phba->debug_readGuard)
+		cnt = snprintf(cbuf, 16, "%u\n", phba->lpfc_injerr_rgrd_cnt);
 	else if (dent == phba->debug_readApp)
 		cnt = snprintf(cbuf, 16, "%u\n", phba->lpfc_injerr_rapp_cnt);
 	else if (dent == phba->debug_readRef)
@@ -1057,6 +1059,8 @@ lpfc_debugfs_dif_err_write(struct file *file, const char __user *buf,
 		phba->lpfc_injerr_wapp_cnt = (uint32_t)tmp;
 	else if (dent == phba->debug_writeRef)
 		phba->lpfc_injerr_wref_cnt = (uint32_t)tmp;
+	else if (dent == phba->debug_readGuard)
+		phba->lpfc_injerr_rgrd_cnt = (uint32_t)tmp;
 	else if (dent == phba->debug_readApp)
 		phba->lpfc_injerr_rapp_cnt = (uint32_t)tmp;
 	else if (dent == phba->debug_readRef)
@@ -3978,6 +3982,17 @@ lpfc_debugfs_initialize(struct lpfc_vport *vport)
 			goto debug_failed;
 		}
 
+		snprintf(name, sizeof(name), "readGuardInjErr");
+		phba->debug_readGuard =
+			debugfs_create_file(name, S_IFREG|S_IRUGO|S_IWUSR,
+			phba->hba_debugfs_root,
+			phba, &lpfc_debugfs_op_dif_err);
+		if (!phba->debug_readGuard) {
+			lpfc_printf_vlog(vport, KERN_ERR, LOG_INIT,
+				"0808 Cannot create debugfs readGuard\n");
+			goto debug_failed;
+		}
+
 		snprintf(name, sizeof(name), "readAppInjErr");
 		phba->debug_readApp =
 			debugfs_create_file(name, S_IFREG|S_IRUGO|S_IWUSR,
@@ -4318,6 +4333,10 @@ lpfc_debugfs_terminate(struct lpfc_vport *vport)
 			debugfs_remove(phba->debug_writeRef); /* writeRef */
 			phba->debug_writeRef = NULL;
 		}
+		if (phba->debug_readGuard) {
+			debugfs_remove(phba->debug_readGuard); /* readGuard */
+			phba->debug_readGuard = NULL;
+		}
 		if (phba->debug_readApp) {
 			debugfs_remove(phba->debug_readApp); /* readApp */
 			phba->debug_readApp = NULL;

drivers/scsi/lpfc/lpfc_hw4.h

@@ -321,6 +321,10 @@ struct lpfc_cqe {
 #define CQE_STATUS_CMD_REJECT		0xb
 #define CQE_STATUS_FCP_TGT_LENCHECK	0xc
 #define CQE_STATUS_NEED_BUFF_ENTRY	0xf
+#define CQE_STATUS_DI_ERROR		0x16
+
+/* Used when mapping CQE status to IOCB */
+#define LPFC_IOCB_STATUS_MASK		0xf
 
 /* Status returned by hardware (valid only if status = CQE_STATUS_SUCCESS). */
 #define CQE_HW_STATUS_NO_ERR		0x0
@@ -348,6 +352,21 @@ struct lpfc_wcqe_complete {
 #define lpfc_wcqe_c_hw_status_WORD	word0
 	uint32_t total_data_placed;
 	uint32_t parameter;
+#define lpfc_wcqe_c_bg_edir_SHIFT	5
+#define lpfc_wcqe_c_bg_edir_MASK	0x00000001
+#define lpfc_wcqe_c_bg_edir_WORD	parameter
+#define lpfc_wcqe_c_bg_tdpv_SHIFT	3
+#define lpfc_wcqe_c_bg_tdpv_MASK	0x00000001
+#define lpfc_wcqe_c_bg_tdpv_WORD	parameter
+#define lpfc_wcqe_c_bg_re_SHIFT		2
+#define lpfc_wcqe_c_bg_re_MASK		0x00000001
+#define lpfc_wcqe_c_bg_re_WORD		parameter
+#define lpfc_wcqe_c_bg_ae_SHIFT		1
+#define lpfc_wcqe_c_bg_ae_MASK		0x00000001
+#define lpfc_wcqe_c_bg_ae_WORD		parameter
+#define lpfc_wcqe_c_bg_ge_SHIFT		0
+#define lpfc_wcqe_c_bg_ge_MASK		0x00000001
+#define lpfc_wcqe_c_bg_ge_WORD		parameter
 	uint32_t word3;
 #define lpfc_wcqe_c_valid_SHIFT		lpfc_cqe_valid_SHIFT
 #define lpfc_wcqe_c_valid_MASK		lpfc_cqe_valid_MASK

drivers/scsi/lpfc/lpfc_init.c

@@ -4380,6 +4380,7 @@ lpfc_sli4_driver_resource_setup(struct lpfc_hba *phba)
 	uint8_t pn_page[LPFC_MAX_SUPPORTED_PAGES] = {0};
 	struct lpfc_mqe *mqe;
 	int longs, sli_family;
+	int sges_per_segment;
 
 	/* Before proceed, wait for POST done and device ready */
 	rc = lpfc_sli4_post_status_check(phba);
@@ -4443,6 +4444,11 @@ lpfc_sli4_driver_resource_setup(struct lpfc_hba *phba)
 	phba->fc_map[1] = LPFC_FCOE_FCF_MAP1;
 	phba->fc_map[2] = LPFC_FCOE_FCF_MAP2;
 
+	/* With BlockGuard we can have multiple SGEs per Data Segemnt */
+	sges_per_segment = 1;
+	if (phba->cfg_enable_bg)
+		sges_per_segment = 2;
+
 	/*
 	 * Since the sg_tablesize is module parameter, the sg_dma_buf_size
 	 * used to create the sg_dma_buf_pool must be dynamically calculated.
@@ -4451,7 +4457,8 @@ lpfc_sli4_driver_resource_setup(struct lpfc_hba *phba)
 	 * sgl sizes of must be a power of 2.
 	 */
 	buf_size = (sizeof(struct fcp_cmnd) + sizeof(struct fcp_rsp) +
-		    ((phba->cfg_sg_seg_cnt + 2) * sizeof(struct sli4_sge)));
+		    (((phba->cfg_sg_seg_cnt * sges_per_segment) + 2) *
+		    sizeof(struct sli4_sge)));
 
 	sli_family = bf_get(lpfc_sli_intf_sli_family, &phba->sli4_hba.sli_intf);
 	max_buf_size = LPFC_SLI4_MAX_BUF_SIZE;
@@ -4468,6 +4475,7 @@ lpfc_sli4_driver_resource_setup(struct lpfc_hba *phba)
 	default:
 		break;
 	}
+
 	for (dma_buf_size = LPFC_SLI4_MIN_BUF_SIZE;
 	     dma_buf_size < max_buf_size && buf_size > dma_buf_size;
 	     dma_buf_size = dma_buf_size << 1)

drivers/scsi/lpfc/lpfc_sli.c

@@ -7839,12 +7839,16 @@ lpfc_sli4_iocb2wqe(struct lpfc_hba *phba, struct lpfc_iocbq *iocbq,
 		bf_set(wqe_lnk, &wqe->fcp_iwrite.wqe_com, iocbq->iocb.ulpXS);
 		/* Always open the exchange */
 		bf_set(wqe_xc, &wqe->fcp_iwrite.wqe_com, 0);
-		bf_set(wqe_dbde, &wqe->fcp_iwrite.wqe_com, 1);
 		bf_set(wqe_iod, &wqe->fcp_iwrite.wqe_com, LPFC_WQE_IOD_WRITE);
 		bf_set(wqe_lenloc, &wqe->fcp_iwrite.wqe_com,
 		       LPFC_WQE_LENLOC_WORD4);
 		bf_set(wqe_ebde_cnt, &wqe->fcp_iwrite.wqe_com, 0);
 		bf_set(wqe_pu, &wqe->fcp_iwrite.wqe_com, iocbq->iocb.ulpPU);
+		if (iocbq->iocb_flag & LPFC_IO_DIF) {
+			iocbq->iocb_flag &= ~LPFC_IO_DIF;
+			bf_set(wqe_dif, &wqe->generic.wqe_com, 1);
+		}
+		bf_set(wqe_dbde, &wqe->fcp_iwrite.wqe_com, 1);
 		break;
 	case CMD_FCP_IREAD64_CR:
 		/* word3 iocb=iotag wqe=payload_offset_len */
@@ -7858,12 +7862,16 @@ lpfc_sli4_iocb2wqe(struct lpfc_hba *phba, struct lpfc_iocbq *iocbq,
 		bf_set(wqe_lnk, &wqe->fcp_iread.wqe_com, iocbq->iocb.ulpXS);
 		/* Always open the exchange */
 		bf_set(wqe_xc, &wqe->fcp_iread.wqe_com, 0);
-		bf_set(wqe_dbde, &wqe->fcp_iread.wqe_com, 1);
 		bf_set(wqe_iod, &wqe->fcp_iread.wqe_com, LPFC_WQE_IOD_READ);
 		bf_set(wqe_lenloc, &wqe->fcp_iread.wqe_com,
 		       LPFC_WQE_LENLOC_WORD4);
 		bf_set(wqe_ebde_cnt, &wqe->fcp_iread.wqe_com, 0);
 		bf_set(wqe_pu, &wqe->fcp_iread.wqe_com, iocbq->iocb.ulpPU);
+		if (iocbq->iocb_flag & LPFC_IO_DIF) {
+			iocbq->iocb_flag &= ~LPFC_IO_DIF;
+			bf_set(wqe_dif, &wqe->generic.wqe_com, 1);
+		}
+		bf_set(wqe_dbde, &wqe->fcp_iread.wqe_com, 1);
 		break;
 	case CMD_FCP_ICMND64_CR:
 		/* word3 iocb=IO_TAG wqe=reserved */
@@ -10672,12 +10680,14 @@ lpfc_sli4_iocb_param_transfer(struct lpfc_hba *phba,
 			      struct lpfc_wcqe_complete *wcqe)
 {
 	unsigned long iflags;
+	uint32_t status;
 	size_t offset = offsetof(struct lpfc_iocbq, iocb);
 
 	memcpy((char *)pIocbIn + offset, (char *)pIocbOut + offset,
 	       sizeof(struct lpfc_iocbq) - offset);
 	/* Map WCQE parameters into irspiocb parameters */
-	pIocbIn->iocb.ulpStatus = bf_get(lpfc_wcqe_c_status, wcqe);
+	status = bf_get(lpfc_wcqe_c_status, wcqe);
+	pIocbIn->iocb.ulpStatus = (status & LPFC_IOCB_STATUS_MASK);
 	if (pIocbOut->iocb_flag & LPFC_IO_FCP)
 		if (pIocbIn->iocb.ulpStatus == IOSTAT_FCP_RSP_ERROR)
 			pIocbIn->iocb.un.fcpi.fcpi_parm =
@@ -10690,6 +10700,44 @@ lpfc_sli4_iocb_param_transfer(struct lpfc_hba *phba,
 		pIocbIn->iocb.un.genreq64.bdl.bdeSize = wcqe->total_data_placed;
 	}
 
+	/* Convert BG errors for completion status */
+	if (status == CQE_STATUS_DI_ERROR) {
+		pIocbIn->iocb.ulpStatus = IOSTAT_LOCAL_REJECT;
+
+		if (bf_get(lpfc_wcqe_c_bg_edir, wcqe))
+			pIocbIn->iocb.un.ulpWord[4] = IOERR_RX_DMA_FAILED;
+		else
+			pIocbIn->iocb.un.ulpWord[4] = IOERR_TX_DMA_FAILED;
+
+		pIocbIn->iocb.unsli3.sli3_bg.bgstat = 0;
+		if (bf_get(lpfc_wcqe_c_bg_ge, wcqe)) /* Guard Check failed */
+			pIocbIn->iocb.unsli3.sli3_bg.bgstat |=
+				BGS_GUARD_ERR_MASK;
+		if (bf_get(lpfc_wcqe_c_bg_ae, wcqe)) /* App Tag Check failed */
+			pIocbIn->iocb.unsli3.sli3_bg.bgstat |=
+				BGS_APPTAG_ERR_MASK;
+		if (bf_get(lpfc_wcqe_c_bg_re, wcqe)) /* Ref Tag Check failed */
+			pIocbIn->iocb.unsli3.sli3_bg.bgstat |=
+				BGS_REFTAG_ERR_MASK;
+
+		/* Check to see if there was any good data before the error */
+		if (bf_get(lpfc_wcqe_c_bg_tdpv, wcqe)) {
+			pIocbIn->iocb.unsli3.sli3_bg.bgstat |=
+				BGS_HI_WATER_MARK_PRESENT_MASK;
+			pIocbIn->iocb.unsli3.sli3_bg.bghm =
+				wcqe->total_data_placed;
+		}
+
+		/*
+		* Set ALL the error bits to indicate we don't know what
+		* type of error it is.
+		*/
+		if (!pIocbIn->iocb.unsli3.sli3_bg.bgstat)
+			pIocbIn->iocb.unsli3.sli3_bg.bgstat |=
+				(BGS_REFTAG_ERR_MASK | BGS_APPTAG_ERR_MASK |
+				BGS_GUARD_ERR_MASK);
+	}
+
 	/* Pick up HBA exchange busy condition */
 	if (bf_get(lpfc_wcqe_c_xb, wcqe)) {
 		spin_lock_irqsave(&phba->hbalock, iflags);

drivers/scsi/lpfc/lpfc_sli.h

@@ -69,6 +69,7 @@ struct lpfc_iocbq {
 #define LPFC_USE_FCPWQIDX	0x80    /* Submit to specified FCPWQ index */
 #define DSS_SECURITY_OP		0x100	/* security IO */
 #define LPFC_IO_ON_Q		0x200	/* The IO is still on the TXCMPLQ */
+#define LPFC_IO_DIF		0x400	/* T10 DIF IO */
 
 #define LPFC_FIP_ELS_ID_MASK	0xc000	/* ELS_ID range 0-3, non-shifted mask */
 #define LPFC_FIP_ELS_ID_SHIFT	14