Merge tag 'ntb-4.16' of git://github.com/jonmason/ntb

Pull NTB updates from Jon Mason:
 "Bug fixes galore, removal of the ntb atom driver, and updates to the
  ntb tools and tests to support the multi-port interface"

* tag 'ntb-4.16' of git://github.com/jonmason/ntb: (37 commits)
  NTB: ntb_perf: fix cast to restricted __le32
  ntb_perf: Fix an error code in perf_copy_chunk()
  ntb_hw_switchtec: Make function switchtec_ntb_remove() static
  NTB: ntb_tool: fix memory leak on 'buf' on error exit path
  NTB: ntb_perf: fix printing of resource_size_t
  NTB: ntb_hw_idt: Set NTB_TOPO_SWITCH topology
  NTB: ntb_test: Update ntb_perf tests
  NTB: ntb_test: Update ntb_tool MW tests
  NTB: ntb_test: Add ntb_tool Message tests
  NTB: ntb_test: Update ntb_tool Scratchpad tests
  NTB: ntb_test: Update ntb_tool DB tests
  NTB: ntb_test: Update ntb_tool link tests
  NTB: ntb_test: Add ntb_tool port tests
  NTB: ntb_test: Safely use paths with whitespace
  NTB: ntb_perf: Add full multi-port NTB API support
  NTB: ntb_tool: Add full multi-port NTB API support
  NTB: ntb_pp: Add full multi-port NTB API support
  NTB: Fix UB/bug in ntb_mw_get_align()
  NTB: Set dma mask and dma coherent mask to NTB devices
  NTB: Rename NTB messaging API methods
  ...

10 files changed, 3269 insertions, 1852 deletions

diff --git a/drivers/ntb/hw/amd/ntb_hw_amd.c b/drivers/ntb/hw/amd/ntb_hw_amd.c
index f0788aae05c9..3cfa46876239 100644
--- a/drivers/ntb/hw/amd/ntb_hw_amd.c
+++ b/drivers/ntb/hw/amd/ntb_hw_amd.c
@@ -1020,6 +1020,10 @@ static int amd_ntb_init_pci(struct amd_ntb_dev *ndev,
 			goto err_dma_mask;
 		dev_warn(&pdev->dev, "Cannot DMA consistent highmem\n");
 	}
+	rc = dma_coerce_mask_and_coherent(&ndev->ntb.dev,
+					  dma_get_mask(&pdev->dev));
+	if (rc)
+		goto err_dma_mask;
 
 	ndev->self_mmio = pci_iomap(pdev, 0, 0);
 	if (!ndev->self_mmio) {
diff --git a/drivers/ntb/hw/idt/ntb_hw_idt.c b/drivers/ntb/hw/idt/ntb_hw_idt.c
index 0cd79f367f7c..8d98872d0983 100644
--- a/drivers/ntb/hw/idt/ntb_hw_idt.c
+++ b/drivers/ntb/hw/idt/ntb_hw_idt.c
@@ -1744,20 +1744,19 @@ static int idt_ntb_msg_clear_mask(struct ntb_dev *ntb, u64 mask_bits)
  * idt_ntb_msg_read() - read message register with specified index
  *			(NTB API callback)
  * @ntb:	NTB device context.
- * @midx:	Message register index
  * @pidx:	OUT - Port index of peer device a message retrieved from
- * @msg:	OUT - Data
+ * @midx:	Message register index
  *
  * Read data from the specified message register and source register.
  *
- * Return: zero on success, negative error if invalid argument passed.
+ * Return: inbound message register value.
  */
-static int idt_ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx, u32 *msg)
+static u32 idt_ntb_msg_read(struct ntb_dev *ntb, int *pidx, int midx)
 {
 	struct idt_ntb_dev *ndev = to_ndev_ntb(ntb);
 
 	if (midx < 0 || IDT_MSG_CNT <= midx)
-		return -EINVAL;
+		return ~(u32)0;
 
 	/* Retrieve source port index of the message */
 	if (pidx != NULL) {
@@ -1772,18 +1771,15 @@ static int idt_ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx, u32 *msg)
 	}
 
 	/* Retrieve data of the corresponding message register */
-	if (msg != NULL)
-		*msg = idt_nt_read(ndev, ntdata_tbl.msgs[midx].in);
-
-	return 0;
+	return idt_nt_read(ndev, ntdata_tbl.msgs[midx].in);
 }
 
 /*
- * idt_ntb_msg_write() - write data to the specified message register
- *			 (NTB API callback)
+ * idt_ntb_peer_msg_write() - write data to the specified message register
+ *			      (NTB API callback)
  * @ntb:	NTB device context.
- * @midx:	Message register index
  * @pidx:	Port index of peer device a message being sent to
+ * @midx:	Message register index
  * @msg:	Data to send
  *
  * Just try to send data to a peer. Message status register should be
@@ -1791,7 +1787,8 @@ static int idt_ntb_msg_read(struct ntb_dev *ntb, int midx, int *pidx, u32 *msg)
  *
  * Return: zero on success, negative error if invalid argument passed.
  */
-static int idt_ntb_msg_write(struct ntb_dev *ntb, int midx, int pidx, u32 msg)
+static int idt_ntb_peer_msg_write(struct ntb_dev *ntb, int pidx, int midx,
+				  u32 msg)
 {
 	struct idt_ntb_dev *ndev = to_ndev_ntb(ntb);
 	unsigned long irqflags;
@@ -2058,7 +2055,7 @@ static const struct ntb_dev_ops idt_ntb_ops = {
 	.msg_set_mask		= idt_ntb_msg_set_mask,
 	.msg_clear_mask		= idt_ntb_msg_clear_mask,
 	.msg_read		= idt_ntb_msg_read,
-	.msg_write		= idt_ntb_msg_write
+	.peer_msg_write		= idt_ntb_peer_msg_write
 };
 
 /*
@@ -2073,7 +2070,7 @@ static int idt_register_device(struct idt_ntb_dev *ndev)
 
 	/* Initialize the rest of NTB device structure and register it */
 	ndev->ntb.ops = &idt_ntb_ops;
-	ndev->ntb.topo = NTB_TOPO_PRI;
+	ndev->ntb.topo = NTB_TOPO_SWITCH;
 
 	ret = ntb_register_device(&ndev->ntb);
 	if (ret != 0) {
@@ -2269,7 +2266,7 @@ static ssize_t idt_dbgfs_info_read(struct file *filp, char __user *ubuf,
 		 "Message data:\n");
 	for (idx = 0; idx < IDT_MSG_CNT; idx++) {
 		int src;
-		(void)idt_ntb_msg_read(&ndev->ntb, idx, &src, &data);
+		data = idt_ntb_msg_read(&ndev->ntb, &src, idx);
 		off += scnprintf(strbuf + off, size - off,
 			"\t%hhu. 0x%08x from peer %hhu (Port %hhu)\n",
 			idx, data, src, ndev->peers[src].port);
@@ -2429,7 +2426,7 @@ static int idt_init_pci(struct idt_ntb_dev *ndev)
 	struct pci_dev *pdev = ndev->ntb.pdev;
 	int ret;
 
-	/* Initialize the bit mask of DMA */
+	/* Initialize the bit mask of PCI/NTB DMA */
 	ret = pci_set_dma_mask(pdev, DMA_BIT_MASK(64));
 	if (ret != 0) {
 		ret = pci_set_dma_mask(pdev, DMA_BIT_MASK(32));
@@ -2450,6 +2447,12 @@ static int idt_init_pci(struct idt_ntb_dev *ndev)
 		dev_warn(&pdev->dev,
 			"Cannot set consistent DMA highmem bit mask\n");
 	}
+	ret = dma_coerce_mask_and_coherent(&ndev->ntb.dev,
+					   dma_get_mask(&pdev->dev));
+	if (ret != 0) {
+		dev_err(&pdev->dev, "Failed to set NTB device DMA bit mask\n");
+		return ret;
+	}
 
 	/*
 	 * Enable the device advanced error reporting. It's not critical to
diff --git a/drivers/ntb/hw/intel/ntb_hw_intel.c b/drivers/ntb/hw/intel/ntb_hw_intel.c
index 4de074a86073..156b45cd4a19 100644
--- a/drivers/ntb/hw/intel/ntb_hw_intel.c
+++ b/drivers/ntb/hw/intel/ntb_hw_intel.c
@@ -74,12 +74,6 @@ MODULE_AUTHOR("Intel Corporation");
 #define bar0_off(base, bar) ((base) + ((bar) << 2))
 #define bar2_off(base, bar) bar0_off(base, (bar) - 2)
 
-static const struct intel_ntb_reg atom_reg;
-static const struct intel_ntb_alt_reg atom_pri_reg;
-static const struct intel_ntb_alt_reg atom_sec_reg;
-static const struct intel_ntb_alt_reg atom_b2b_reg;
-static const struct intel_ntb_xlat_reg atom_pri_xlat;
-static const struct intel_ntb_xlat_reg atom_sec_xlat;
 static const struct intel_ntb_reg xeon_reg;
 static const struct intel_ntb_alt_reg xeon_pri_reg;
 static const struct intel_ntb_alt_reg xeon_sec_reg;
@@ -184,15 +178,6 @@ static inline void _iowrite64(u64 val, void __iomem *mmio)
 #endif
 #endif
 
-static inline int pdev_is_atom(struct pci_dev *pdev)
-{
-	switch (pdev->device) {
-	case PCI_DEVICE_ID_INTEL_NTB_B2B_BWD:
-		return 1;
-	}
-	return 0;
-}
-
 static inline int pdev_is_xeon(struct pci_dev *pdev)
 {
 	switch (pdev->device) {
@@ -1006,8 +991,7 @@ static ssize_t ndev_debugfs_read(struct file *filp, char __user *ubuf,
 {
 	struct intel_ntb_dev *ndev = filp->private_data;
 
-	if (pdev_is_xeon(ndev->ntb.pdev) ||
-	    pdev_is_atom(ndev->ntb.pdev))
+	if (pdev_is_xeon(ndev->ntb.pdev))
 		return ndev_ntb_debugfs_read(filp, ubuf, count, offp);
 	else if (pdev_is_skx_xeon(ndev->ntb.pdev))
 		return ndev_ntb3_debugfs_read(filp, ubuf, count, offp);
@@ -1439,242 +1423,6 @@ static int intel_ntb_peer_spad_write(struct ntb_dev *ntb, int pidx,
 			       ndev->peer_reg->spad);
 }
 
-/* ATOM */
-
-static u64 atom_db_ioread(void __iomem *mmio)
-{
-	return ioread64(mmio);
-}
-
-static void atom_db_iowrite(u64 bits, void __iomem *mmio)
-{
-	iowrite64(bits, mmio);
-}
-
-static int atom_poll_link(struct intel_ntb_dev *ndev)
-{
-	u32 ntb_ctl;
-
-	ntb_ctl = ioread32(ndev->self_mmio + ATOM_NTBCNTL_OFFSET);
-
-	if (ntb_ctl == ndev->ntb_ctl)
-		return 0;
-
-	ndev->ntb_ctl = ntb_ctl;
-
-	ndev->lnk_sta = ioread32(ndev->self_mmio + ATOM_LINK_STATUS_OFFSET);
-
-	return 1;
-}
-
-static int atom_link_is_up(struct intel_ntb_dev *ndev)
-{
-	return ATOM_NTB_CTL_ACTIVE(ndev->ntb_ctl);
-}
-
-static int atom_link_is_err(struct intel_ntb_dev *ndev)
-{
-	if (ioread32(ndev->self_mmio + ATOM_LTSSMSTATEJMP_OFFSET)
-	    & ATOM_LTSSMSTATEJMP_FORCEDETECT)
-		return 1;
-
-	if (ioread32(ndev->self_mmio + ATOM_IBSTERRRCRVSTS0_OFFSET)
-	    & ATOM_IBIST_ERR_OFLOW)
-		return 1;
-
-	return 0;
-}
-
-static inline enum ntb_topo atom_ppd_topo(struct intel_ntb_dev *ndev, u32 ppd)
-{
-	struct device *dev = &ndev->ntb.pdev->dev;
-
-	switch (ppd & ATOM_PPD_TOPO_MASK) {
-	case ATOM_PPD_TOPO_B2B_USD:
-		dev_dbg(dev, "PPD %d B2B USD\n", ppd);
-		return NTB_TOPO_B2B_USD;
-
-	case ATOM_PPD_TOPO_B2B_DSD:
-		dev_dbg(dev, "PPD %d B2B DSD\n", ppd);
-		return NTB_TOPO_B2B_DSD;
-
-	case ATOM_PPD_TOPO_PRI_USD:
-	case ATOM_PPD_TOPO_PRI_DSD: /* accept bogus PRI_DSD */
-	case ATOM_PPD_TOPO_SEC_USD:
-	case ATOM_PPD_TOPO_SEC_DSD: /* accept bogus SEC_DSD */
-		dev_dbg(dev, "PPD %d non B2B disabled\n", ppd);
-		return NTB_TOPO_NONE;
-	}
-
-	dev_dbg(dev, "PPD %d invalid\n", ppd);
-	return NTB_TOPO_NONE;
-}
-
-static void atom_link_hb(struct work_struct *work)
-{
-	struct intel_ntb_dev *ndev = hb_ndev(work);
-	struct device *dev = &ndev->ntb.pdev->dev;
-	unsigned long poll_ts;
-	void __iomem *mmio;
-	u32 status32;
-
-	poll_ts = ndev->last_ts + ATOM_LINK_HB_TIMEOUT;
-
-	/* Delay polling the link status if an interrupt was received,
-	 * unless the cached link status says the link is down.
-	 */
-	if (time_after(poll_ts, jiffies) && atom_link_is_up(ndev)) {
-		schedule_delayed_work(&ndev->hb_timer, poll_ts - jiffies);
-		return;
-	}
-
-	if (atom_poll_link(ndev))
-		ntb_link_event(&ndev->ntb);
-
-	if (atom_link_is_up(ndev) || !atom_link_is_err(ndev)) {
-		schedule_delayed_work(&ndev->hb_timer, ATOM_LINK_HB_TIMEOUT);
-		return;
-	}
-
-	/* Link is down with error: recover the link! */
-
-	mmio = ndev->self_mmio;
-
-	/* Driver resets the NTB ModPhy lanes - magic! */
-	iowrite8(0xe0, mmio + ATOM_MODPHY_PCSREG6);
-	iowrite8(0x40, mmio + ATOM_MODPHY_PCSREG4);
-	iowrite8(0x60, mmio + ATOM_MODPHY_PCSREG4);
-	iowrite8(0x60, mmio + ATOM_MODPHY_PCSREG6);
-
-	/* Driver waits 100ms to allow the NTB ModPhy to settle */
-	msleep(100);
-
-	/* Clear AER Errors, write to clear */
-	status32 = ioread32(mmio + ATOM_ERRCORSTS_OFFSET);
-	dev_dbg(dev, "ERRCORSTS = %x\n", status32);
-	status32 &= PCI_ERR_COR_REP_ROLL;
-	iowrite32(status32, mmio + ATOM_ERRCORSTS_OFFSET);
-
-	/* Clear unexpected electrical idle event in LTSSM, write to clear */
-	status32 = ioread32(mmio + ATOM_LTSSMERRSTS0_OFFSET);
-	dev_dbg(dev, "LTSSMERRSTS0 = %x\n", status32);
-	status32 |= ATOM_LTSSMERRSTS0_UNEXPECTEDEI;
-	iowrite32(status32, mmio + ATOM_LTSSMERRSTS0_OFFSET);
-
-	/* Clear DeSkew Buffer error, write to clear */
-	status32 = ioread32(mmio + ATOM_DESKEWSTS_OFFSET);
-	dev_dbg(dev, "DESKEWSTS = %x\n", status32);
-	status32 |= ATOM_DESKEWSTS_DBERR;
-	iowrite32(status32, mmio + ATOM_DESKEWSTS_OFFSET);
-
-	status32 = ioread32(mmio + ATOM_IBSTERRRCRVSTS0_OFFSET);
-	dev_dbg(dev, "IBSTERRRCRVSTS0 = %x\n", status32);
-	status32 &= ATOM_IBIST_ERR_OFLOW;
-	iowrite32(status32, mmio + ATOM_IBSTERRRCRVSTS0_OFFSET);
-
-	/* Releases the NTB state machine to allow the link to retrain */
-	status32 = ioread32(mmio + ATOM_LTSSMSTATEJMP_OFFSET);
-	dev_dbg(dev, "LTSSMSTATEJMP = %x\n", status32);
-	status32 &= ~ATOM_LTSSMSTATEJMP_FORCEDETECT;
-	iowrite32(status32, mmio + ATOM_LTSSMSTATEJMP_OFFSET);
-
-	/* There is a potential race between the 2 NTB devices recovering at the
-	 * same time.  If the times are the same, the link will not recover and
-	 * the driver will be stuck in this loop forever.  Add a random interval
-	 * to the recovery time to prevent this race.
-	 */
-	schedule_delayed_work(&ndev->hb_timer, ATOM_LINK_RECOVERY_TIME
-			      + prandom_u32() % ATOM_LINK_RECOVERY_TIME);
-}
-
-static int atom_init_isr(struct intel_ntb_dev *ndev)
-{
-	int rc;
-
-	rc = ndev_init_isr(ndev, 1, ATOM_DB_MSIX_VECTOR_COUNT,
-			   ATOM_DB_MSIX_VECTOR_SHIFT, ATOM_DB_TOTAL_SHIFT);
-	if (rc)
-		return rc;
-
-	/* ATOM doesn't have link status interrupt, poll on that platform */
-	ndev->last_ts = jiffies;
-	INIT_DELAYED_WORK(&ndev->hb_timer, atom_link_hb);
-	schedule_delayed_work(&ndev->hb_timer, ATOM_LINK_HB_TIMEOUT);
-
-	return 0;
-}
-
-static void atom_deinit_isr(struct intel_ntb_dev *ndev)
-{
-	cancel_delayed_work_sync(&ndev->hb_timer);
-	ndev_deinit_isr(ndev);
-}
-
-static int atom_init_ntb(struct intel_ntb_dev *ndev)
-{
-	ndev->mw_count = ATOM_MW_COUNT;
-	ndev->spad_count = ATOM_SPAD_COUNT;
-	ndev->db_count = ATOM_DB_COUNT;
-
-	switch (ndev->ntb.topo) {
-	case NTB_TOPO_B2B_USD:
-	case NTB_TOPO_B2B_DSD:
-		ndev->self_reg = &atom_pri_reg;
-		ndev->peer_reg = &atom_b2b_reg;
-		ndev->xlat_reg = &atom_sec_xlat;
-
-		/* Enable Bus Master and Memory Space on the secondary side */
-		iowrite16(PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER,
-			  ndev->self_mmio + ATOM_SPCICMD_OFFSET);
-
-		break;
-
-	default:
-		return -EINVAL;
-	}
-
-	ndev->db_valid_mask = BIT_ULL(ndev->db_count) - 1;
-
-	return 0;
-}
-
-static int atom_init_dev(struct intel_ntb_dev *ndev)
-{
-	u32 ppd;
-	int rc;
-
-	rc = pci_read_config_dword(ndev->ntb.pdev, ATOM_PPD_OFFSET, &ppd);
-	if (rc)
-		return -EIO;
-
-	ndev->ntb.topo = atom_ppd_topo(ndev, ppd);
-	if (ndev->ntb.topo == NTB_TOPO_NONE)
-		return -EINVAL;
-
-	rc = atom_init_ntb(ndev);
-	if (rc)
-		return rc;
-
-	rc = atom_init_isr(ndev);
-	if (rc)
-		return rc;
-
-	if (ndev->ntb.topo != NTB_TOPO_SEC) {
-		/* Initiate PCI-E link training */
-		rc = pci_write_config_dword(ndev->ntb.pdev, ATOM_PPD_OFFSET,
-					    ppd | ATOM_PPD_INIT_LINK);
-		if (rc)
-			return rc;
-	}
-
-	return 0;
-}
-
-static void atom_deinit_dev(struct intel_ntb_dev *ndev)
-{
-	atom_deinit_isr(ndev);
-}
-
 /* Skylake Xeon NTB */
 
 static int skx_poll_link(struct intel_ntb_dev *ndev)
@@ -2586,6 +2334,10 @@ static int intel_ntb_init_pci(struct intel_ntb_dev *ndev, struct pci_dev *pdev)
 			goto err_dma_mask;
 		dev_warn(&pdev->dev, "Cannot DMA consistent highmem\n");
 	}
+	rc = dma_coerce_mask_and_coherent(&ndev->ntb.dev,
+					  dma_get_mask(&pdev->dev));
+	if (rc)
+		goto err_dma_mask;
 
 	ndev->self_mmio = pci_iomap(pdev, 0, 0);
 	if (!ndev->self_mmio) {
@@ -2658,24 +2410,7 @@ static int intel_ntb_pci_probe(struct pci_dev *pdev,
 
 	node = dev_to_node(&pdev->dev);
 
-	if (pdev_is_atom(pdev)) {
-		ndev = kzalloc_node(sizeof(*ndev), GFP_KERNEL, node);
-		if (!ndev) {
-			rc = -ENOMEM;
-			goto err_ndev;
-		}
-
-		ndev_init_struct(ndev, pdev);
-
-		rc = intel_ntb_init_pci(ndev, pdev);
-		if (rc)
-			goto err_init_pci;
-
-		rc = atom_init_dev(ndev);
-		if (rc)
-			goto err_init_dev;
-
-	} else if (pdev_is_xeon(pdev)) {
+	if (pdev_is_xeon(pdev)) {
 		ndev = kzalloc_node(sizeof(*ndev), GFP_KERNEL, node);
 		if (!ndev) {
 			rc = -ENOMEM;
@@ -2731,9 +2466,7 @@ static int intel_ntb_pci_probe(struct pci_dev *pdev,
 
 err_register:
 	ndev_deinit_debugfs(ndev);
-	if (pdev_is_atom(pdev))
-		atom_deinit_dev(ndev);
-	else if (pdev_is_xeon(pdev) || pdev_is_skx_xeon(pdev))
+	if (pdev_is_xeon(pdev) || pdev_is_skx_xeon(pdev))
 		xeon_deinit_dev(ndev);
 err_init_dev:
 	intel_ntb_deinit_pci(ndev);
@@ -2749,41 +2482,12 @@ static void intel_ntb_pci_remove(struct pci_dev *pdev)
 
 	ntb_unregister_device(&ndev->ntb);
 	ndev_deinit_debugfs(ndev);
-	if (pdev_is_atom(pdev))
-		atom_deinit_dev(ndev);
-	else if (pdev_is_xeon(pdev) || pdev_is_skx_xeon(pdev))
+	if (pdev_is_xeon(pdev) || pdev_is_skx_xeon(pdev))
 		xeon_deinit_dev(ndev);
 	intel_ntb_deinit_pci(ndev);
 	kfree(ndev);
 }
 
-static const struct intel_ntb_reg atom_reg = {
-	.poll_link		= atom_poll_link,
-	.link_is_up		= atom_link_is_up,
-	.db_ioread		= atom_db_ioread,
-	.db_iowrite		= atom_db_iowrite,
-	.db_size		= sizeof(u64),
-	.ntb_ctl		= ATOM_NTBCNTL_OFFSET,
-	.mw_bar			= {2, 4},
-};
-
-static const struct intel_ntb_alt_reg atom_pri_reg = {
-	.db_bell		= ATOM_PDOORBELL_OFFSET,
-	.db_mask		= ATOM_PDBMSK_OFFSET,
-	.spad			= ATOM_SPAD_OFFSET,
-};
-
-static const struct intel_ntb_alt_reg atom_b2b_reg = {
-	.db_bell		= ATOM_B2B_DOORBELL_OFFSET,
-	.spad			= ATOM_B2B_SPAD_OFFSET,
-};
-
-static const struct intel_ntb_xlat_reg atom_sec_xlat = {
-	/* FIXME : .bar0_base	= ATOM_SBAR0BASE_OFFSET, */
-	/* FIXME : .bar2_limit	= ATOM_SBAR2LMT_OFFSET, */
-	.bar2_xlat		= ATOM_SBAR2XLAT_OFFSET,
-};
-
 static const struct intel_ntb_reg xeon_reg = {
 	.poll_link		= xeon_poll_link,
 	.link_is_up		= xeon_link_is_up,
@@ -2940,7 +2644,6 @@ static const struct file_operations intel_ntb_debugfs_info = {
 };
 
 static const struct pci_device_id intel_ntb_pci_tbl[] = {
-	{PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_NTB_B2B_BWD)},
 	{PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_NTB_B2B_JSF)},
 	{PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_NTB_B2B_SNB)},
 	{PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_NTB_B2B_IVT)},
diff --git a/drivers/ntb/hw/intel/ntb_hw_intel.h b/drivers/ntb/hw/intel/ntb_hw_intel.h
index 2d6c38afb128..4415aa7ea775 100644
--- a/drivers/ntb/hw/intel/ntb_hw_intel.h
+++ b/drivers/ntb/hw/intel/ntb_hw_intel.h
@@ -66,7 +66,6 @@
 #define PCI_DEVICE_ID_INTEL_NTB_B2B_HSX	0x2F0D
 #define PCI_DEVICE_ID_INTEL_NTB_PS_HSX	0x2F0E
 #define PCI_DEVICE_ID_INTEL_NTB_SS_HSX	0x2F0F
-#define PCI_DEVICE_ID_INTEL_NTB_B2B_BWD	0x0C4E
 #define PCI_DEVICE_ID_INTEL_NTB_B2B_BDX	0x6F0D
 #define PCI_DEVICE_ID_INTEL_NTB_PS_BDX	0x6F0E
 #define PCI_DEVICE_ID_INTEL_NTB_SS_BDX	0x6F0F
@@ -196,63 +195,6 @@
 #define SKX_DB_TOTAL_SHIFT		33
 #define SKX_SPAD_COUNT			16
 
-/* Intel Atom hardware */
-
-#define ATOM_SBAR2XLAT_OFFSET		0x0008
-#define ATOM_PDOORBELL_OFFSET		0x0020
-#define ATOM_PDBMSK_OFFSET		0x0028
-#define ATOM_NTBCNTL_OFFSET		0x0060
-#define ATOM_SPAD_OFFSET			0x0080
-#define ATOM_PPD_OFFSET			0x00d4
-#define ATOM_PBAR2XLAT_OFFSET		0x8008
-#define ATOM_B2B_DOORBELL_OFFSET		0x8020
-#define ATOM_B2B_SPAD_OFFSET		0x8080
-#define ATOM_SPCICMD_OFFSET		0xb004
-#define ATOM_LINK_STATUS_OFFSET		0xb052
-#define ATOM_ERRCORSTS_OFFSET		0xb110
-#define ATOM_IP_BASE			0xc000
-#define ATOM_DESKEWSTS_OFFSET		(ATOM_IP_BASE + 0x3024)
-#define ATOM_LTSSMERRSTS0_OFFSET		(ATOM_IP_BASE + 0x3180)
-#define ATOM_LTSSMSTATEJMP_OFFSET	(ATOM_IP_BASE + 0x3040)
-#define ATOM_IBSTERRRCRVSTS0_OFFSET	(ATOM_IP_BASE + 0x3324)
-#define ATOM_MODPHY_PCSREG4		0x1c004
-#define ATOM_MODPHY_PCSREG6		0x1c006
-
-#define ATOM_PPD_INIT_LINK		0x0008
-#define ATOM_PPD_CONN_MASK		0x0300
-#define ATOM_PPD_CONN_TRANSPARENT	0x0000
-#define ATOM_PPD_CONN_B2B		0x0100
-#define ATOM_PPD_CONN_RP			0x0200
-#define ATOM_PPD_DEV_MASK		0x1000
-#define ATOM_PPD_DEV_USD			0x0000
-#define ATOM_PPD_DEV_DSD			0x1000
-#define ATOM_PPD_TOPO_MASK	(ATOM_PPD_CONN_MASK | ATOM_PPD_DEV_MASK)
-#define ATOM_PPD_TOPO_PRI_USD	(ATOM_PPD_CONN_TRANSPARENT | ATOM_PPD_DEV_USD)
-#define ATOM_PPD_TOPO_PRI_DSD	(ATOM_PPD_CONN_TRANSPARENT | ATOM_PPD_DEV_DSD)
-#define ATOM_PPD_TOPO_SEC_USD	(ATOM_PPD_CONN_RP | ATOM_PPD_DEV_USD)
-#define ATOM_PPD_TOPO_SEC_DSD	(ATOM_PPD_CONN_RP | ATOM_PPD_DEV_DSD)
-#define ATOM_PPD_TOPO_B2B_USD	(ATOM_PPD_CONN_B2B | ATOM_PPD_DEV_USD)
-#define ATOM_PPD_TOPO_B2B_DSD	(ATOM_PPD_CONN_B2B | ATOM_PPD_DEV_DSD)
-
-#define ATOM_MW_COUNT			2
-#define ATOM_DB_COUNT			34
-#define ATOM_DB_VALID_MASK		(BIT_ULL(ATOM_DB_COUNT) - 1)
-#define ATOM_DB_MSIX_VECTOR_COUNT	34
-#define ATOM_DB_MSIX_VECTOR_SHIFT	1
-#define ATOM_DB_TOTAL_SHIFT		34
-#define ATOM_SPAD_COUNT			16
-
-#define ATOM_NTB_CTL_DOWN_BIT		BIT(16)
-#define ATOM_NTB_CTL_ACTIVE(x)		!(x & ATOM_NTB_CTL_DOWN_BIT)
-
-#define ATOM_DESKEWSTS_DBERR		BIT(15)
-#define ATOM_LTSSMERRSTS0_UNEXPECTEDEI	BIT(20)
-#define ATOM_LTSSMSTATEJMP_FORCEDETECT	BIT(2)
-#define ATOM_IBIST_ERR_OFLOW		0x7FFF7FFF
-
-#define ATOM_LINK_HB_TIMEOUT		msecs_to_jiffies(1000)
-#define ATOM_LINK_RECOVERY_TIME		msecs_to_jiffies(500)
-
 /* Ntb control and link status */
 
 #define NTB_CTL_CFG_LOCK		BIT(0)
diff --git a/drivers/ntb/hw/mscc/ntb_hw_switchtec.c b/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
index afe8ed6f3b23..f624ae27eabe 100644
--- a/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
+++ b/drivers/ntb/hw/mscc/ntb_hw_switchtec.c
@@ -94,6 +94,9 @@ struct switchtec_ntb {
 	struct ntb_ctrl_regs __iomem *mmio_self_ctrl;
 	struct ntb_ctrl_regs __iomem *mmio_peer_ctrl;
 	struct ntb_dbmsg_regs __iomem *mmio_self_dbmsg;
+	struct ntb_dbmsg_regs __iomem *mmio_peer_dbmsg;
+
+	void __iomem *mmio_xlink_win;
 
 	struct shared_mw *self_shared;
 	struct shared_mw __iomem *peer_shared;
@@ -109,6 +112,7 @@ struct switchtec_ntb {
 
 	int nr_direct_mw;
 	int nr_lut_mw;
+	int nr_rsvd_luts;
 	int direct_mw_to_bar[MAX_DIRECT_MW];
 
 	int peer_nr_direct_mw;
@@ -118,6 +122,7 @@ struct switchtec_ntb {
 	bool link_is_up;
 	enum ntb_speed link_speed;
 	enum ntb_width link_width;
+	struct work_struct link_reinit_work;
 };
 
 static struct switchtec_ntb *ntb_sndev(struct ntb_dev *ntb)
@@ -172,7 +177,7 @@ static int switchtec_ntb_part_op(struct switchtec_ntb *sndev,
 
 	if (ps == status) {
 		dev_err(&sndev->stdev->dev,
-			"Timed out while peforming %s (%d). (%08x)",
+			"Timed out while performing %s (%d). (%08x)\n",
 			op_text[op], op,
 			ioread32(&ctl->partition_status));
 
@@ -185,10 +190,10 @@ static int switchtec_ntb_part_op(struct switchtec_ntb *sndev,
 static int switchtec_ntb_send_msg(struct switchtec_ntb *sndev, int idx,
 				  u32 val)
 {
-	if (idx < 0 || idx >= ARRAY_SIZE(sndev->mmio_self_dbmsg->omsg))
+	if (idx < 0 || idx >= ARRAY_SIZE(sndev->mmio_peer_dbmsg->omsg))
 		return -EINVAL;
 
-	iowrite32(val, &sndev->mmio_self_dbmsg->omsg[idx].msg);
+	iowrite32(val, &sndev->mmio_peer_dbmsg->omsg[idx].msg);
 
 	return 0;
 }
@@ -197,7 +202,7 @@ static int switchtec_ntb_mw_count(struct ntb_dev *ntb, int pidx)
 {
 	struct switchtec_ntb *sndev = ntb_sndev(ntb);
 	int nr_direct_mw = sndev->peer_nr_direct_mw;
-	int nr_lut_mw = sndev->peer_nr_lut_mw - 1;
+	int nr_lut_mw = sndev->peer_nr_lut_mw - sndev->nr_rsvd_luts;
 
 	if (pidx != NTB_DEF_PEER_IDX)
 		return -EINVAL;
@@ -210,12 +215,12 @@ static int switchtec_ntb_mw_count(struct ntb_dev *ntb, int pidx)
 
 static int lut_index(struct switchtec_ntb *sndev, int mw_idx)
 {
-	return mw_idx - sndev->nr_direct_mw + 1;
+	return mw_idx - sndev->nr_direct_mw + sndev->nr_rsvd_luts;
 }
 
 static int peer_lut_index(struct switchtec_ntb *sndev, int mw_idx)
 {
-	return mw_idx - sndev->peer_nr_direct_mw + 1;
+	return mw_idx - sndev->peer_nr_direct_mw + sndev->nr_rsvd_luts;
 }
 
 static int switchtec_ntb_mw_get_align(struct ntb_dev *ntb, int pidx,
@@ -306,7 +311,7 @@ static int switchtec_ntb_mw_set_trans(struct ntb_dev *ntb, int pidx, int widx,
 	if (pidx != NTB_DEF_PEER_IDX)
 		return -EINVAL;
 
-	dev_dbg(&sndev->stdev->dev, "MW %d: part %d addr %pad size %pap",
+	dev_dbg(&sndev->stdev->dev, "MW %d: part %d addr %pad size %pap\n",
 		widx, pidx, &addr, &size);
 
 	if (widx >= switchtec_ntb_mw_count(ntb, pidx))
@@ -315,6 +320,19 @@ static int switchtec_ntb_mw_set_trans(struct ntb_dev *ntb, int pidx, int widx,
 	if (xlate_pos < 12)
 		return -EINVAL;
 
+	if (!IS_ALIGNED(addr, BIT_ULL(xlate_pos))) {
+		/*
+		 * In certain circumstances we can get a buffer that is
+		 * not aligned to its size. (Most of the time
+		 * dma_alloc_coherent ensures this). This can happen when
+		 * using large buffers allocated by the CMA
+		 * (see CMA_CONFIG_ALIGNMENT)
+		 */
+		dev_err(&sndev->stdev->dev,
+			"ERROR: Memory window address is not aligned to it's size!\n");
+		return -EINVAL;
+	}
+
 	rc = switchtec_ntb_part_op(sndev, ctl, NTB_CTRL_PART_OP_LOCK,
 				   NTB_CTRL_PART_STATUS_LOCKED);
 	if (rc)
@@ -337,7 +355,7 @@ static int switchtec_ntb_mw_set_trans(struct ntb_dev *ntb, int pidx, int widx,
 
 	if (rc == -EIO) {
 		dev_err(&sndev->stdev->dev,
-			"Hardware reported an error configuring mw %d: %08x",
+			"Hardware reported an error configuring mw %d: %08x\n",
 			widx, ioread32(&ctl->bar_error));
 
 		if (widx < nr_direct_mw)
@@ -355,8 +373,9 @@ static int switchtec_ntb_mw_set_trans(struct ntb_dev *ntb, int pidx, int widx,
 static int switchtec_ntb_peer_mw_count(struct ntb_dev *ntb)
 {
 	struct switchtec_ntb *sndev = ntb_sndev(ntb);
+	int nr_lut_mw = sndev->nr_lut_mw - sndev->nr_rsvd_luts;
 
-	return sndev->nr_direct_mw + (use_lut_mws ? sndev->nr_lut_mw - 1 : 0);
+	return sndev->nr_direct_mw + (use_lut_mws ? nr_lut_mw : 0);
 }
 
 static int switchtec_ntb_direct_get_addr(struct switchtec_ntb *sndev,
@@ -463,18 +482,69 @@ static void switchtec_ntb_set_link_speed(struct switchtec_ntb *sndev)
 	sndev->link_width = min(self_width, peer_width);
 }
 
-enum {
+static int crosslink_is_enabled(struct switchtec_ntb *sndev)
+{
+	struct ntb_info_regs __iomem *inf = sndev->mmio_ntb;
+
+	return ioread8(&inf->ntp_info[sndev->peer_partition].xlink_enabled);
+}
+
+static void crosslink_init_dbmsgs(struct switchtec_ntb *sndev)
+{
+	int i;
+	u32 msg_map = 0;
+
+	if (!crosslink_is_enabled(sndev))
+		return;
+
+	for (i = 0; i < ARRAY_SIZE(sndev->mmio_peer_dbmsg->imsg); i++) {
+		int m = i | sndev->self_partition << 2;
+
+		msg_map |= m << i * 8;
+	}
+
+	iowrite32(msg_map, &sndev->mmio_peer_dbmsg->msg_map);
+	iowrite64(sndev->db_valid_mask << sndev->db_peer_shift,
+		  &sndev->mmio_peer_dbmsg->odb_mask);
+}
+
+enum switchtec_msg {
 	LINK_MESSAGE = 0,
 	MSG_LINK_UP = 1,
 	MSG_LINK_DOWN = 2,
 	MSG_CHECK_LINK = 3,
+	MSG_LINK_FORCE_DOWN = 4,
 };
 
-static void switchtec_ntb_check_link(struct switchtec_ntb *sndev)
+static int switchtec_ntb_reinit_peer(struct switchtec_ntb *sndev);
+
+static void link_reinit_work(struct work_struct *work)
+{
+	struct switchtec_ntb *sndev;
+
+	sndev = container_of(work, struct switchtec_ntb, link_reinit_work);
+
+	switchtec_ntb_reinit_peer(sndev);
+}
+
+static void switchtec_ntb_check_link(struct switchtec_ntb *sndev,
+				     enum switchtec_msg msg)
 {
 	int link_sta;
 	int old = sndev->link_is_up;
 
+	if (msg == MSG_LINK_FORCE_DOWN) {
+		schedule_work(&sndev->link_reinit_work);
+
+		if (sndev->link_is_up) {
+			sndev->link_is_up = 0;
+			ntb_link_event(&sndev->ntb);
+			dev_info(&sndev->stdev->dev, "ntb link forced down\n");
+		}
+
+		return;
+	}
+
 	link_sta = sndev->self_shared->link_sta;
 	if (link_sta) {
 		u64 peer = ioread64(&sndev->peer_shared->magic);
@@ -491,8 +561,11 @@ static void switchtec_ntb_check_link(struct switchtec_ntb *sndev)
 	if (link_sta != old) {
 		switchtec_ntb_send_msg(sndev, LINK_MESSAGE, MSG_CHECK_LINK);
 		ntb_link_event(&sndev->ntb);
-		dev_info(&sndev->stdev->dev, "ntb link %s",
+		dev_info(&sndev->stdev->dev, "ntb link %s\n",
 			 link_sta ? "up" : "down");
+
+		if (link_sta)
+			crosslink_init_dbmsgs(sndev);
 	}
 }
 
@@ -500,7 +573,7 @@ static void switchtec_ntb_link_notification(struct switchtec_dev *stdev)
 {
 	struct switchtec_ntb *sndev = stdev->sndev;
 
-	switchtec_ntb_check_link(sndev);
+	switchtec_ntb_check_link(sndev, MSG_CHECK_LINK);
 }
 
 static u64 switchtec_ntb_link_is_up(struct ntb_dev *ntb,
@@ -523,12 +596,12 @@ static int switchtec_ntb_link_enable(struct ntb_dev *ntb,
 {
 	struct switchtec_ntb *sndev = ntb_sndev(ntb);
 
-	dev_dbg(&sndev->stdev->dev, "enabling link");
+	dev_dbg(&sndev->stdev->dev, "enabling link\n");
 
 	sndev->self_shared->link_sta = 1;
 	switchtec_ntb_send_msg(sndev, LINK_MESSAGE, MSG_LINK_UP);
 
-	switchtec_ntb_check_link(sndev);
+	switchtec_ntb_check_link(sndev, MSG_CHECK_LINK);
 
 	return 0;
 }
@@ -537,12 +610,12 @@ static int switchtec_ntb_link_disable(struct ntb_dev *ntb)
 {
 	struct switchtec_ntb *sndev = ntb_sndev(ntb);
 
-	dev_dbg(&sndev->stdev->dev, "disabling link");
+	dev_dbg(&sndev->stdev->dev, "disabling link\n");
 
 	sndev->self_shared->link_sta = 0;
-	switchtec_ntb_send_msg(sndev, LINK_MESSAGE, MSG_LINK_UP);
+	switchtec_ntb_send_msg(sndev, LINK_MESSAGE, MSG_LINK_DOWN);
 
-	switchtec_ntb_check_link(sndev);
+	switchtec_ntb_check_link(sndev, MSG_CHECK_LINK);
 
 	return 0;
 }
@@ -638,7 +711,7 @@ static int switchtec_ntb_peer_db_addr(struct ntb_dev *ntb,
 	struct switchtec_ntb *sndev = ntb_sndev(ntb);
 	unsigned long offset;
 
-	offset = (unsigned long)sndev->mmio_self_dbmsg->odb -
+	offset = (unsigned long)sndev->mmio_peer_dbmsg->odb -
 		(unsigned long)sndev->stdev->mmio;
 
 	offset += sndev->db_shift / 8;
@@ -656,7 +729,7 @@ static int switchtec_ntb_peer_db_set(struct ntb_dev *ntb, u64 db_bits)
 	struct switchtec_ntb *sndev = ntb_sndev(ntb);
 
 	iowrite64(db_bits << sndev->db_peer_shift,
-		  &sndev->mmio_self_dbmsg->odb);
+		  &sndev->mmio_peer_dbmsg->odb);
 
 	return 0;
 }
@@ -777,24 +850,63 @@ static const struct ntb_dev_ops switchtec_ntb_ops = {
 	.peer_spad_addr		= switchtec_ntb_peer_spad_addr,
 };
 
-static void switchtec_ntb_init_sndev(struct switchtec_ntb *sndev)
+static int switchtec_ntb_init_sndev(struct switchtec_ntb *sndev)
 {
+	u64 tpart_vec;
+	int self;
 	u64 part_map;
+	int bit;
 
 	sndev->ntb.pdev = sndev->stdev->pdev;
 	sndev->ntb.topo = NTB_TOPO_SWITCH;
 	sndev->ntb.ops = &switchtec_ntb_ops;
 
+	INIT_WORK(&sndev->link_reinit_work, link_reinit_work);
+
 	sndev->self_partition = sndev->stdev->partition;
 
 	sndev->mmio_ntb = sndev->stdev->mmio_ntb;
+
+	self = sndev->self_partition;
+	tpart_vec = ioread32(&sndev->mmio_ntb->ntp_info[self].target_part_high);
+	tpart_vec <<= 32;
+	tpart_vec |= ioread32(&sndev->mmio_ntb->ntp_info[self].target_part_low);
+
 	part_map = ioread64(&sndev->mmio_ntb->ep_map);
 	part_map &= ~(1 << sndev->self_partition);
-	sndev->peer_partition = ffs(part_map) - 1;
 
-	dev_dbg(&sndev->stdev->dev, "Partition ID %d of %d (%llx)",
-		sndev->self_partition, sndev->stdev->partition_count,
-		part_map);
+	if (!ffs(tpart_vec)) {
+		if (sndev->stdev->partition_count != 2) {
+			dev_err(&sndev->stdev->dev,
+				"ntb target partition not defined\n");
+			return -ENODEV;
+		}
+
+		bit = ffs(part_map);
+		if (!bit) {
+			dev_err(&sndev->stdev->dev,
+				"peer partition is not NT partition\n");
+			return -ENODEV;
+		}
+
+		sndev->peer_partition = bit - 1;
+	} else {
+		if (ffs(tpart_vec) != fls(tpart_vec)) {
+			dev_err(&sndev->stdev->dev,
+				"ntb driver only supports 1 pair of 1-1 ntb mapping\n");
+			return -ENODEV;
+		}
+
+		sndev->peer_partition = ffs(tpart_vec) - 1;
+		if (!(part_map & (1 << sndev->peer_partition))) {
+			dev_err(&sndev->stdev->dev,
+				"ntb target partition is not NT partition\n");
+			return -ENODEV;
+		}
+	}
+
+	dev_dbg(&sndev->stdev->dev, "Partition ID %d of %d\n",
+		sndev->self_partition, sndev->stdev->partition_count);
 
 	sndev->mmio_ctrl = (void * __iomem)sndev->mmio_ntb +
 		SWITCHTEC_NTB_REG_CTRL_OFFSET;
@@ -804,6 +916,283 @@ static void switchtec_ntb_init_sndev(struct switchtec_ntb *sndev)
 	sndev->mmio_self_ctrl = &sndev->mmio_ctrl[sndev->self_partition];
 	sndev->mmio_peer_ctrl = &sndev->mmio_ctrl[sndev->peer_partition];
 	sndev->mmio_self_dbmsg = &sndev->mmio_dbmsg[sndev->self_partition];
+	sndev->mmio_peer_dbmsg = sndev->mmio_self_dbmsg;
+
+	return 0;
+}
+
+static int config_rsvd_lut_win(struct switchtec_ntb *sndev,
+			       struct ntb_ctrl_regs __iomem *ctl,
+			       int lut_idx, int partition, u64 addr)
+{
+	int peer_bar = sndev->peer_direct_mw_to_bar[0];
+	u32 ctl_val;
+	int rc;
+
+	rc = switchtec_ntb_part_op(sndev, ctl, NTB_CTRL_PART_OP_LOCK,
+				   NTB_CTRL_PART_STATUS_LOCKED);
+	if (rc)
+		return rc;
+
+	ctl_val = ioread32(&ctl->bar_entry[peer_bar].ctl);
+	ctl_val &= 0xFF;
+	ctl_val |= NTB_CTRL_BAR_LUT_WIN_EN;
+	ctl_val |= ilog2(LUT_SIZE) << 8;
+	ctl_val |= (sndev->nr_lut_mw - 1) << 14;
+	iowrite32(ctl_val, &ctl->bar_entry[peer_bar].ctl);
+
+	iowrite64((NTB_CTRL_LUT_EN | (part