Skip to contentSkip to conversation

bnxt patchset with bug fixes

PATCH Superseded Kalesh A P 4 participants 29 messages Jan 13, 2020 – Jan 16, 2020 Network drivers

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Kalesh A P

author
#0 lore raw

v2 cover bnxt patchset with bug fixes

From: Kalesh AP <[email protected]>

Please apply.

V1->V2: fixed the correct the commit ids in Fixes tag.

Kalesh AP (8):
net/bnxt: handle flow create failure
net/bnxt: fix probe failure in FreeBSD net/bnxt: fix to use correct IOVA mapping net/bnxt: fix enable/disable VLAN filtering net/bnxt: fix enable/disable vlan strip net/bnxt: handle hw filter setting when port is stopped net/bnxt: fix a memory leak in port stop net/bnxt: use macro for PCI log format

Somnath Kotur (2):
net/bnxt: release port upon close
net/bnxt: fix to cap max rings to minimum of compl rings and stat
contexts

 drivers/net/bnxt/bnxt.h        |  21 +-
 drivers/net/bnxt/bnxt_ethdev.c | 422 ++++++++++++++++++++++-------------------
 drivers/net/bnxt/bnxt_filter.c |   5 +-
 drivers/net/bnxt/bnxt_flow.c   |   4 +-
 drivers/net/bnxt/bnxt_hwrm.c   |  28 ++-
 drivers/net/bnxt/bnxt_irq.c    |   8 +
 drivers/net/bnxt/bnxt_ring.c   |  17 --
 drivers/net/bnxt/bnxt_vnic.c   |  11 --
 8 files changed, 273 insertions(+), 243 deletions(-)
Signature · 1 line
2.10.1
author
#1 lore raw

PATCH v2 1/10 net/bnxt: handle flow create failure

From: Kalesh AP <[email protected]>

If flow create fails due to not enough filter resources, driver does not populate the rte_flow_error using rte_flow_error_set().

Since "rte_errno" could have garbage value and is not relaiable, it could cause a segfault in the stack in port_flow_complain().

Fix it to set rte_flow_error using rte_flow_error_set() when flow create fails due to not enough filter resources.

---
 drivers/net/bnxt/bnxt_flow.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/net/bnxt/bnxt_flow.c +3 -1
index cde1fa4..5564c53 100644
@@ -1702,7 +1702,9 @@ bnxt_flow_create(struct rte_eth_dev *dev,
 
 	filter = bnxt_get_unused_filter(bp);
 	if (filter == NULL) {
-		PMD_DRV_LOG(ERR, "Not enough resources for a new flow.\n");
+		rte_flow_error_set(error, ENOSPC,
+				   RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+				   "Not enough resources for a new flow");
 		goto free_flow;
 	}
 
Signature · 1 line
2.10.1
author
#2 lore raw

PATCH v2 2/10 net/bnxt: fix probe failure in FreeBSD

In reply to Kalesh A P

From: Kalesh AP <[email protected]>

In FreeBSD environment, nic_uio driver does not support interrupts and rte_intr_callback_register() will fail to register interrupts which in turn causes bnxt driver probe failure.

Fixed driver to ignore interrupt callback failures in FreeBSD. Also fixed to not use a dedicated completion ring for async events from FW and process these events on RXQ0 in FreeBSD.

Fixes 6de4c538b393 ("net/bnxt: fix error handling in port start")
Fixes 43f78b380f89 ("net/bnxt: retry IRQ callback deregistration")
---
 drivers/net/bnxt/bnxt.h        | 8 ++++++++
 drivers/net/bnxt/bnxt_ethdev.c | 3 +++
 drivers/net/bnxt/bnxt_irq.c    | 8 ++++++++
 3 files changed, 19 insertions(+)
drivers/net/bnxt/bnxt.h +8 -0
index 21ca059..3487b91 100644
@@ -117,6 +117,14 @@
 #define BNXT_NUM_ASYNC_CPR(bp) 1
 #endif
 
+/* In FreeBSD OS, nic_uio driver does not support interrupts */
+#ifdef RTE_EXEC_ENV_FREEBSD
+#ifdef BNXT_NUM_ASYNC_CPR
+#undef BNXT_NUM_ASYNC_CPR
+#endif
+#define BNXT_NUM_ASYNC_CPR(bp)	0
+#endif
+
 #define BNXT_MISC_VEC_ID               RTE_INTR_VEC_ZERO_OFFSET
 #define BNXT_RX_VEC_START              RTE_INTR_VEC_RXTX_OFFSET
 
drivers/net/bnxt/bnxt_ethdev.c +3 -0
index 7b5df9a..879ea58 100644
@@ -439,8 +439,11 @@ static int bnxt_init_chip(struct bnxt *bp)
 
 	/* enable uio/vfio intr/eventfd mapping */
 	rc = rte_intr_enable(intr_handle);
+#ifndef RTE_EXEC_ENV_FREEBSD
+	/* In FreeBSD OS, nic_uio driver does not support interrupts */
 	if (rc)
 		goto err_free;
+#endif
 
 	rc = bnxt_get_hwrm_link_config(bp, &new);
 	if (rc) {
drivers/net/bnxt/bnxt_irq.c +8 -0
index 846325e..40e1b0c 100644
@@ -181,5 +181,13 @@ int bnxt_request_int(struct bnxt *bp)
 			irq->requested = 1;
 	}
 
+#ifdef RTE_EXEC_ENV_FREEBSD
+	/**
+	 * In FreeBSD OS, nic_uio does not support interrupts and
+	 * interrupt register callback will fail.
+	 */
+	rc = 0;
+#endif
+
 	return rc;
 }
Signature · 1 line
2.10.1
author
#3 lore raw

PATCH v2 3/10 net/bnxt: fix to use correct IOVA mapping

In reply to Kalesh A P

From: Kalesh AP <[email protected]>

Use rte_malloc_virt2iova() to obtain the IO address of a virtual address obtained through rte_malloc().

Fixed to use the iova address returned by rte_memzone_reserve_aligned() as the call always returns with populating "mz->iova" with rte_malloc_virt2iova(mz->addr).

Removed redundant rte_mem_lock_page() call to lock the pages.

Fixes 96b0931d51e7 ("fix extended port counter statistics")
---
 drivers/net/bnxt/bnxt_ethdev.c | 49 ------------------------------------------
 drivers/net/bnxt/bnxt_hwrm.c   | 28 ++++++++++--------------
 drivers/net/bnxt/bnxt_ring.c   | 17 ---------------
 drivers/net/bnxt/bnxt_vnic.c   | 11 ----------
 4 files changed, 11 insertions(+), 94 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +0 -49
index 879ea58..a948c78 100644
@@ -4233,18 +4233,6 @@ static int bnxt_alloc_ctx_mem_blk(struct bnxt *bp,
 
 		memset(mz->addr, 0, mz->len);
 		mz_phys_addr = mz->iova;
-		if ((unsigned long)mz->addr == mz_phys_addr) {
-			PMD_DRV_LOG(DEBUG,
-				    "physical address same as virtual\n");
-			PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-			mz_phys_addr = rte_mem_virt2iova(mz->addr);
-			if (mz_phys_addr == RTE_BAD_IOVA) {
-				PMD_DRV_LOG(ERR,
-					"unable to map addr to phys memory\n");
-				return -ENOMEM;
-			}
-		}
-		rte_mem_lock_page(((char *)mz->addr));
 
 		rmem->pg_tbl = mz->addr;
 		rmem->pg_tbl_map = mz_phys_addr;
@@ -4268,22 +4256,8 @@ static int bnxt_alloc_ctx_mem_blk(struct bnxt *bp,
 
 	memset(mz->addr, 0, mz->len);
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual.\n");
-		PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-		for (sz = 0; sz < mem_size; sz += BNXT_PAGE_SIZE)
-			rte_mem_lock_page(((char *)mz->addr) + sz);
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-				    "unable to map addr to phys memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	for (sz = 0, i = 0; sz < mem_size; sz += BNXT_PAGE_SIZE, i++) {
-		rte_mem_lock_page(((char *)mz->addr) + sz);
 		rmem->pg_arr[i] = ((char *)mz->addr) + sz;
 		rmem->dma_arr[i] = mz_phys_addr + sz;
 
@@ -4460,18 +4434,6 @@ static int bnxt_alloc_stats_mem(struct bnxt *bp)
 	}
 	memset(mz->addr, 0, mz->len);
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual.\n");
-		PMD_DRV_LOG(DEBUG,
-			    "Using rte_mem_virt2iova()\n");
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-				    "Can't map address to physical memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	bp->rx_mem_zone = (const void *)mz;
 	bp->hw_rx_port_stats = mz->addr;
@@ -4498,17 +4460,6 @@ static int bnxt_alloc_stats_mem(struct bnxt *bp)
 	}
 	memset(mz->addr, 0, mz->len);
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual\n");
-		PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-				    "Can't map address to physical memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	bp->tx_mem_zone = (const void *)mz;
 	bp->hw_tx_port_stats = mz->addr;
drivers/net/bnxt/bnxt_hwrm.c +11 -17
index 50272dc..3b01339 100644
@@ -309,8 +309,8 @@ int bnxt_hwrm_cfa_l2_set_rx_mask(struct bnxt *bp,
 	if (vlan_table) {
 		if (!(mask & HWRM_CFA_L2_SET_RX_MASK_INPUT_MASK_VLAN_NONVLAN))
 			mask |= HWRM_CFA_L2_SET_RX_MASK_INPUT_MASK_VLANONLY;
-		req.vlan_tag_tbl_addr = rte_cpu_to_le_64(
-			 rte_mem_virt2iova(vlan_table));
+		req.vlan_tag_tbl_addr =
+			rte_cpu_to_le_64(rte_malloc_virt2iova(vlan_table));
 		req.num_vlan_tags = rte_cpu_to_le_32((uint32_t)vlan_count);
 	}
 	req.mask = rte_cpu_to_le_32(mask);
@@ -351,7 +351,7 @@ int bnxt_hwrm_cfa_vlan_antispoof_cfg(struct bnxt *bp, uint16_t fid,
 	req.fid = rte_cpu_to_le_16(fid);
 
 	req.vlan_tag_mask_tbl_addr =
-		rte_cpu_to_le_64(rte_mem_virt2iova(vlan_table));
+		rte_cpu_to_le_64(rte_malloc_virt2iova(vlan_table));
 	req.num_vlan_entries = rte_cpu_to_le_32((uint32_t)vlan_count);
 
 	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_CHIMP_MB);
@@ -1024,9 +1024,8 @@ int bnxt_hwrm_ver_get(struct bnxt *bp)
 			rc = -ENOMEM;
 			goto error;
 		}
-		rte_mem_lock_page(bp->hwrm_cmd_resp_addr);
 		bp->hwrm_cmd_resp_dma_addr =
-			rte_mem_virt2iova(bp->hwrm_cmd_resp_addr);
+			rte_malloc_virt2iova(bp->hwrm_cmd_resp_addr);
 		if (bp->hwrm_cmd_resp_dma_addr == RTE_BAD_IOVA) {
 			PMD_DRV_LOG(ERR,
 			"Unable to map response buffer to physical memory.\n");
@@ -1061,9 +1060,8 @@ int bnxt_hwrm_ver_get(struct bnxt *bp)
 			rc = -ENOMEM;
 			goto error;
 		}
-		rte_mem_lock_page(bp->hwrm_short_cmd_req_addr);
 		bp->hwrm_short_cmd_req_dma_addr =
-			rte_mem_virt2iova(bp->hwrm_short_cmd_req_addr);
+			rte_malloc_virt2iova(bp->hwrm_short_cmd_req_addr);
 		if (bp->hwrm_short_cmd_req_dma_addr == RTE_BAD_IOVA) {
 			rte_free(bp->hwrm_short_cmd_req_addr);
 			PMD_DRV_LOG(ERR,
@@ -2471,11 +2469,10 @@ int bnxt_alloc_hwrm_resources(struct bnxt *bp)
 		pdev->addr.bus, pdev->addr.devid, pdev->addr.function);
 	bp->max_resp_len = HWRM_MAX_RESP_LEN;
 	bp->hwrm_cmd_resp_addr = rte_malloc(type, bp->max_resp_len, 0);
-	rte_mem_lock_page(bp->hwrm_cmd_resp_addr);
 	if (bp->hwrm_cmd_resp_addr == NULL)
 		return -ENOMEM;
 	bp->hwrm_cmd_resp_dma_addr =
-		rte_mem_virt2iova(bp->hwrm_cmd_resp_addr);
+		rte_malloc_virt2iova(bp->hwrm_cmd_resp_addr);
 	if (bp->hwrm_cmd_resp_dma_addr == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map response address to physical memory\n");
@@ -3421,7 +3418,7 @@ int bnxt_hwrm_func_buf_rgtr(struct bnxt *bp)
 			 page_getenum(bp->pf.active_vfs * HWRM_MAX_REQ_LEN));
 	req.req_buf_len = rte_cpu_to_le_16(HWRM_MAX_REQ_LEN);
 	req.req_buf_page_addr0 =
-		rte_cpu_to_le_64(rte_mem_virt2iova(bp->pf.vf_req_buf));
+		rte_cpu_to_le_64(rte_malloc_virt2iova(bp->pf.vf_req_buf));
 	if (req.req_buf_page_addr0 == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map buffer address to physical memory\n");
@@ -3851,10 +3848,9 @@ int bnxt_get_nvram_directory(struct bnxt *bp, uint32_t len, uint8_t *data)
 
 	buflen = dir_entries * entry_length;
 	buf = rte_malloc("nvm_dir", buflen, 0);
-	rte_mem_lock_page(buf);
 	if (buf == NULL)
 		return -ENOMEM;
-	dma_handle = rte_mem_virt2iova(buf);
+	dma_handle = rte_malloc_virt2iova(buf);
 	if (dma_handle == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map response address to physical memory\n");
@@ -3885,11 +3881,10 @@ int bnxt_hwrm_get_nvram_item(struct bnxt *bp, uint32_t index,
 	struct hwrm_nvm_read_output *resp = bp->hwrm_cmd_resp_addr;
 
 	buf = rte_malloc("nvm_item", length, 0);
-	rte_mem_lock_page(buf);
 	if (!buf)
 		return -ENOMEM;
 
-	dma_handle = rte_mem_virt2iova(buf);
+	dma_handle = rte_malloc_virt2iova(buf);
 	if (dma_handle == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map response address to physical memory\n");
@@ -3939,11 +3934,10 @@ int bnxt_hwrm_flash_nvram(struct bnxt *bp, uint16_t dir_type,
 	uint8_t *buf;
 
 	buf = rte_malloc("nvm_write", data_len, 0);
-	rte_mem_lock_page(buf);
 	if (!buf)
 		return -ENOMEM;
 
-	dma_handle = rte_mem_virt2iova(buf);
+	dma_handle = rte_malloc_virt2iova(buf);
 	if (dma_handle == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map response address to physical memory\n");
@@ -4006,7 +4000,7 @@ static int bnxt_hwrm_func_vf_vnic_query(struct bnxt *bp, uint16_t vf,
 
 	req.vf_id = rte_cpu_to_le_16(bp->pf.first_vf_id + vf);
 	req.max_vnic_id_cnt = rte_cpu_to_le_32(bp->pf.total_vnics);
-	req.vnic_id_tbl_addr = rte_cpu_to_le_64(rte_mem_virt2iova(vnic_ids));
+	req.vnic_id_tbl_addr = rte_cpu_to_le_64(rte_malloc_virt2iova(vnic_ids));
 
 	if (req.vnic_id_tbl_addr == RTE_BAD_IOVA) {
 		HWRM_UNLOCK();
drivers/net/bnxt/bnxt_ring.c +0 -17
index ea46fa9..d6e4e8a 100644
@@ -110,9 +110,7 @@ int bnxt_alloc_rings(struct bnxt *bp, uint16_t qidx,
 	uint64_t rx_offloads = bp->eth_dev->data->dev_conf.rxmode.offloads;
 	const struct rte_memzone *mz = NULL;
 	char mz_name[RTE_MEMZONE_NAMESIZE];
-	rte_iova_t mz_phys_addr_base;
 	rte_iova_t mz_phys_addr;
-	int sz;
 
 	int stats_len = (tx_ring_info || rx_ring_info) ?
 	    RTE_CACHE_LINE_ROUNDUP(sizeof(struct hwrm_stat_ctx_query_output) -
@@ -214,22 +212,7 @@ int bnxt_alloc_rings(struct bnxt *bp, uint16_t qidx,
 			return -ENOMEM;
 	}
 	memset(mz->addr, 0, mz->len);
-	mz_phys_addr_base = mz->iova;
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr_base) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual.\n");
-		PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-		for (sz = 0; sz < total_alloc_len; sz += getpagesize())
-			rte_mem_lock_page(((char *)mz->addr) + sz);
-		mz_phys_addr_base = rte_mem_virt2iova(mz->addr);
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-			"unable to map ring address to physical memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	if (tx_ring_info) {
 		txq->mz = mz;
drivers/net/bnxt/bnxt_vnic.c +0 -11
index 104342e..bc054a8 100644
@@ -150,17 +150,6 @@ int bnxt_alloc_vnic_attributes(struct bnxt *bp)
 			return -ENOMEM;
 	}
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual.\n");
-		PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-				    "unable to map to physical memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	for (i = 0; i < max_vnics; i++) {
 		vnic = &bp->vnic_info[i];
Signature · 1 line
2.10.1
author
#4 lore raw

PATCH v2 4/10 net/bnxt: fix enable/disable VLAN filtering

In reply to Kalesh A P

From: Kalesh AP <[email protected]>

There is no condition check for the user requested operation for VLAN filtering. As a result, VLAN filtering is getting disabled when the user enables/disables VLAN stripping on same port.

The function bnxt_hwrm_clear_l2_filter() didn't actually free L2 filter in HW if the reference count of filter is zero.

Fixed it by incrementing the reference count of filter in bnxt_alloc_filter() routine.

Because of the recent changes in bnxt_hwrm_clear_l2_filter(), change was needed in the routine bnxt_set_default_mac_addr_op() to destroy and re-create the default filter when the user changes the default MAC of the port.

Fixes 51fda51e7b70 ("net/bnxt: fix to keep the L2 filter intact so it can be reused")
Fixes 6118503d8071 ("net/bnxt: fix VLAN filtering")
---
 drivers/net/bnxt/bnxt_ethdev.c | 66 +++++++++++++++++++++++-------------------
 drivers/net/bnxt/bnxt_filter.c |  5 ++--
 2 files changed, 39 insertions(+), 32 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +36 -30
index a948c78..83f475d 100644
@@ -1860,18 +1860,12 @@ static int bnxt_del_dflt_mac_filter(struct bnxt *bp,
 }
 
 static int
-bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
+bnxt_config_vlan_hw_filter(struct bnxt *bp, uint64_t rx_offloads)
 {
-	struct bnxt *bp = dev->data->dev_private;
-	uint64_t rx_offloads = dev->data->dev_conf.rxmode.offloads;
 	struct bnxt_vnic_info *vnic;
 	unsigned int i;
 	int rc;
 
-	rc = is_bnxt_in_error(bp);
-	if (rc)
-		return rc;
-
 	vnic = BNXT_GET_DEFAULT_VNIC(bp);
 	if (!(rx_offloads & DEV_RX_OFFLOAD_VLAN_FILTER)) {
 		/* Remove any VLAN filters programmed */
@@ -1895,6 +1889,28 @@ bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
 	PMD_DRV_LOG(DEBUG, "VLAN Filtering: %d\n",
 		    !!(rx_offloads & DEV_RX_OFFLOAD_VLAN_FILTER));
 
+	return 0;
+}
+
+static int
+bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
+{
+	uint64_t rx_offloads = dev->data->dev_conf.rxmode.offloads;
+	struct bnxt *bp = dev->data->dev_private;
+	unsigned int i;
+	int rc;
+
+	rc = is_bnxt_in_error(bp);
+	if (rc)
+		return rc;
+
+	if (mask & ETH_VLAN_FILTER_MASK) {
+		/* Enable or disable VLAN filtering */
+		rc = bnxt_config_vlan_hw_filter(bp, rx_offloads);
+		if (rc)
+			return rc;
+	}
+
 	if (mask & ETH_VLAN_STRIP_MASK) {
 		/* Enable or disable VLAN stripping */
 		for (i = 0; i < bp->nr_vnics; i++) {
@@ -1984,7 +2000,6 @@ bnxt_set_default_mac_addr_op(struct rte_eth_dev *dev,
 	struct bnxt *bp = dev->data->dev_private;
 	/* Default Filter is tied to VNIC 0 */
 	struct bnxt_vnic_info *vnic = BNXT_GET_DEFAULT_VNIC(bp);
-	struct bnxt_filter_info *filter;
 	int rc;
 
 	rc = is_bnxt_in_error(bp);
@@ -1997,32 +2012,23 @@ bnxt_set_default_mac_addr_op(struct rte_eth_dev *dev,
 	if (rte_is_zero_ether_addr(addr))
 		return -EINVAL;
 
-	STAILQ_FOREACH(filter, &vnic->filter, next) {
-		/* Default Filter is at Index 0 */
-		if (filter->mac_index != 0)
-			continue;
-
-		memcpy(filter->l2_addr, addr, RTE_ETHER_ADDR_LEN);
-		memset(filter->l2_addr_mask, 0xff, RTE_ETHER_ADDR_LEN);
-		filter->flags |= HWRM_CFA_L2_FILTER_ALLOC_INPUT_FLAGS_PATH_RX |
-			HWRM_CFA_L2_FILTER_ALLOC_INPUT_FLAGS_OUTERMOST;
-		filter->enables |=
-			HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_ADDR |
-			HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_ADDR_MASK;
+	/* Check if the requested MAC is already added */
+	if (memcmp(addr, bp->mac_addr, RTE_ETHER_ADDR_LEN) == 0)
+		return 0;
 
-		rc = bnxt_hwrm_set_l2_filter(bp, vnic->fw_vnic_id, filter);
-		if (rc) {
-			memcpy(filter->l2_addr, bp->mac_addr,
-			       RTE_ETHER_ADDR_LEN);
-			return rc;
-		}
+	/* Destroy filter and re-create it */
+	bnxt_del_dflt_mac_filter(bp, vnic);
 
-		memcpy(bp->mac_addr, addr, RTE_ETHER_ADDR_LEN);
-		PMD_DRV_LOG(DEBUG, "Set MAC addr\n");
-		return 0;
+	memcpy(bp->mac_addr, addr, RTE_ETHER_ADDR_LEN);
+	if (dev->data->dev_conf.rxmode.offloads & DEV_RX_OFFLOAD_VLAN_FILTER) {
+		/* This filter will allow only untagged packets */
+		rc = bnxt_add_vlan_filter(bp, 0);
+	} else {
+		rc = bnxt_add_mac_filter(bp, vnic, addr, 0, 0);
 	}
 
-	return 0;
+	PMD_DRV_LOG(DEBUG, "Set MAC addr\n");
+	return rc;
 }
 
 static int
drivers/net/bnxt/bnxt_filter.c +3 -2
index 0c410f8..b31f104 100644
@@ -39,9 +39,10 @@ struct bnxt_filter_info *bnxt_alloc_filter(struct bnxt *bp)
 	filter->flags = HWRM_CFA_L2_FILTER_ALLOC_INPUT_FLAGS_PATH_RX;
 	filter->enables = HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_ADDR |
 			HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_ADDR_MASK;
-	memcpy(filter->l2_addr, bp->eth_dev->data->mac_addrs->addr_bytes,
-	       RTE_ETHER_ADDR_LEN);
+	memcpy(filter->l2_addr, bp->mac_addr, RTE_ETHER_ADDR_LEN);
 	memset(filter->l2_addr_mask, 0xff, RTE_ETHER_ADDR_LEN);
+	/* bump up the reference count of filter */
+	filter->l2_ref_cnt++;
 	return filter;
 }
 
Signature · 1 line
2.10.1
author
#5 lore raw

PATCH v2 5/10 net/bnxt: fix enable/disable vlan strip

In reply to Kalesh A P

From: Kalesh AP <[email protected]>

HWRM_VNIC_CFG command to configure vnic dynamically with traffic running is not working. Driver has to free and recreate the vnic and then reconfigure the vnic filters.

Fixes 7fe5668d2ea3 ("net/bnxt: support VLAN filter and strip")
---
 drivers/net/bnxt/bnxt_ethdev.c | 248 +++++++++++++++++++++++++----------------
 1 file changed, 150 insertions(+), 98 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +150 -98
index 83f475d..0b2c29b 100644
@@ -132,6 +132,7 @@ static int bnxt_dev_uninit(struct rte_eth_dev *eth_dev);
 static int bnxt_init_resources(struct bnxt *bp, bool reconfig_dev);
 static int bnxt_uninit_resources(struct bnxt *bp, bool reconfig_dev);
 static void bnxt_cancel_fw_health_check(struct bnxt *bp);
+static int bnxt_restore_vlan_filters(struct bnxt *bp);
 
 int is_bnxt_in_error(struct bnxt *bp)
 {
@@ -228,14 +229,97 @@ static int bnxt_alloc_mem(struct bnxt *bp, bool reconfig)
 	return rc;
 }
 
-static int bnxt_init_chip(struct bnxt *bp)
+static int bnxt_setup_one_vnic(struct bnxt *bp, uint16_t vnic_id)
 {
+	struct rte_eth_conf *dev_conf = &bp->eth_dev->data->dev_conf;
+	struct bnxt_vnic_info *vnic = &bp->vnic_info[vnic_id];
+	uint64_t rx_offloads = dev_conf->rxmode.offloads;
 	struct bnxt_rx_queue *rxq;
+	unsigned int j;
+	int rc;
+
+	rc = bnxt_vnic_grp_alloc(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	PMD_DRV_LOG(DEBUG, "vnic[%d] = %p vnic->fw_grp_ids = %p\n",
+		    vnic_id, vnic, vnic->fw_grp_ids);
+
+	rc = bnxt_hwrm_vnic_alloc(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	/* Alloc RSS context only if RSS mode is enabled */
+	if (dev_conf->rxmode.mq_mode & ETH_MQ_RX_RSS) {
+		int j, nr_ctxs = bnxt_rss_ctxts(bp);
+
+		rc = 0;
+		for (j = 0; j < nr_ctxs; j++) {
+			rc = bnxt_hwrm_vnic_ctx_alloc(bp, vnic, j);
+			if (rc)
+				break;
+		}
+		if (rc) {
+			PMD_DRV_LOG(ERR,
+				    "HWRM vnic %d ctx %d alloc failure rc: %x\n",
+				    vnic_id, j, rc);
+			goto err_out;
+		}
+		vnic->num_lb_ctxts = nr_ctxs;
+	}
+
+	/*
+	 * Firmware sets pf pair in default vnic cfg. If the VLAN strip
+	 * setting is not available at this time, it will not be
+	 * configured correctly in the CFA.
+	 */
+	if (rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
+		vnic->vlan_strip = true;
+	else
+		vnic->vlan_strip = false;
+
+	rc = bnxt_hwrm_vnic_cfg(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	rc = bnxt_set_hwrm_vnic_filters(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	for (j = 0; j < bp->rx_num_qs_per_vnic; j++) {
+		rxq = bp->eth_dev->data->rx_queues[j];
+
+		PMD_DRV_LOG(DEBUG,
+			    "rxq[%d]->vnic=%p vnic->fw_grp_ids=%p\n",
+			    j, rxq->vnic, rxq->vnic->fw_grp_ids);
+
+		if (BNXT_HAS_RING_GRPS(bp) && rxq->rx_deferred_start)
+			rxq->vnic->fw_grp_ids[j] = INVALID_HW_RING_ID;
+	}
+
+	rc = bnxt_vnic_rss_configure(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	bnxt_hwrm_vnic_plcmode_cfg(bp, vnic);
+
+	if (rx_offloads & DEV_RX_OFFLOAD_TCP_LRO)
+		bnxt_hwrm_vnic_tpa_cfg(bp, vnic, 1);
+	else
+		bnxt_hwrm_vnic_tpa_cfg(bp, vnic, 0);
+
+	return 0;
+err_out:
+	PMD_DRV_LOG(ERR, "HWRM vnic %d cfg failure rc: %x\n",
+		    vnic_id, rc);
+	return rc;
+}
+
+static int bnxt_init_chip(struct bnxt *bp)
+{
 	struct rte_eth_link new;
 	struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(bp->eth_dev);
-	struct rte_eth_conf *dev_conf = &bp->eth_dev->data->dev_conf;
 	struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
-	uint64_t rx_offloads = dev_conf->rxmode.offloads;
 	uint32_t intr_vector = 0;
 	uint32_t queue_id, base = BNXT_MISC_VEC_ID;
 	uint32_t vec = BNXT_MISC_VEC_ID;
@@ -303,93 +387,11 @@ static int bnxt_init_chip(struct bnxt *bp)
 
 	/* VNIC configuration */
 	for (i = 0; i < bp->nr_vnics; i++) {
-		struct rte_eth_conf *dev_conf = &bp->eth_dev->data->dev_conf;
-		struct bnxt_vnic_info *vnic = &bp->vnic_info[i];
-
-		rc = bnxt_vnic_grp_alloc(bp, vnic);
+		rc = bnxt_setup_one_vnic(bp, i);
 		if (rc)
 			goto err_out;
-
-		PMD_DRV_LOG(DEBUG, "vnic[%d] = %p vnic->fw_grp_ids = %p\n",
-			    i, vnic, vnic->fw_grp_ids);
-
-		rc = bnxt_hwrm_vnic_alloc(bp, vnic);
-		if (rc) {
-			PMD_DRV_LOG(ERR, "HWRM vnic %d alloc failure rc: %x\n",
-				i, rc);
-			goto err_out;
-		}
-
-		/* Alloc RSS context only if RSS mode is enabled */
-		if (dev_conf->rxmode.mq_mode & ETH_MQ_RX_RSS) {
-			int j, nr_ctxs = bnxt_rss_ctxts(bp);
-
-			rc = 0;
-			for (j = 0; j < nr_ctxs; j++) {
-				rc = bnxt_hwrm_vnic_ctx_alloc(bp, vnic, j);
-				if (rc)
-					break;
-			}
-			if (rc) {
-				PMD_DRV_LOG(ERR,
-				  "HWRM vnic %d ctx %d alloc failure rc: %x\n",
-				  i, j, rc);
-				goto err_out;
-			}
-			vnic->num_lb_ctxts = nr_ctxs;
-		}
-
-		/*
-		 * Firmware sets pf pair in default vnic cfg. If the VLAN strip
-		 * setting is not available at this time, it will not be
-		 * configured correctly in the CFA.
-		 */
-		if (rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
-			vnic->vlan_strip = true;
-		else
-			vnic->vlan_strip = false;
-
-		rc = bnxt_hwrm_vnic_cfg(bp, vnic);
-		if (rc) {
-			PMD_DRV_LOG(ERR, "HWRM vnic %d cfg failure rc: %x\n",
-				i, rc);
-			goto err_out;
-		}
-
-		rc = bnxt_set_hwrm_vnic_filters(bp, vnic);
-		if (rc) {
-			PMD_DRV_LOG(ERR,
-				"HWRM vnic %d filter failure rc: %x\n",
-				i, rc);
-			goto err_out;
-		}
-
-		for (j = 0; j < bp->rx_num_qs_per_vnic; j++) {
-			rxq = bp->eth_dev->data->rx_queues[j];
-
-			PMD_DRV_LOG(DEBUG,
-				    "rxq[%d]->vnic=%p vnic->fw_grp_ids=%p\n",
-				    j, rxq->vnic, rxq->vnic->fw_grp_ids);
-
-			if (BNXT_HAS_RING_GRPS(bp) && rxq->rx_deferred_start)
-				rxq->vnic->fw_grp_ids[j] = INVALID_HW_RING_ID;
-		}
-
-		rc = bnxt_vnic_rss_configure(bp, vnic);
-		if (rc) {
-			PMD_DRV_LOG(ERR,
-				    "HWRM vnic set RSS failure rc: %x\n", rc);
-			goto err_out;
-		}
-
-		bnxt_hwrm_vnic_plcmode_cfg(bp, vnic);
-
-		if (bp->eth_dev->data->dev_conf.rxmode.offloads &
-		    DEV_RX_OFFLOAD_TCP_LRO)
-			bnxt_hwrm_vnic_tpa_cfg(bp, vnic, 1);
-		else
-			bnxt_hwrm_vnic_tpa_cfg(bp, vnic, 0);
 	}
+
 	rc = bnxt_hwrm_cfa_l2_set_rx_mask(bp, &bp->vnic_info[0], 0, NULL);
 	if (rc) {
 		PMD_DRV_LOG(ERR,
@@ -1892,12 +1894,69 @@ bnxt_config_vlan_hw_filter(struct bnxt *bp, uint64_t rx_offloads)
 	return 0;
 }
 
+static int bnxt_free_one_vnic(struct bnxt *bp, uint16_t vnic_id)
+{
+	struct bnxt_vnic_info *vnic = &bp->vnic_info[vnic_id];
+	unsigned int i;
+	int rc;
+
+	/* Destroy vnic filters and vnic */
+	if (bp->eth_dev->data->dev_conf.rxmode.offloads &
+	    DEV_RX_OFFLOAD_VLAN_FILTER) {
+		for (i = 0; i < RTE_ETHER_MAX_VLAN_ID; i++)
+			bnxt_del_vlan_filter(bp, i);
+	}
+	bnxt_del_dflt_mac_filter(bp, vnic);
+
+	rc = bnxt_hwrm_vnic_free(bp, vnic);
+	if (rc)
+		return rc;
+
+	rte_free(vnic->fw_grp_ids);
+	vnic->fw_grp_ids = NULL;
+
+	return 0;
+}
+
+static int
+bnxt_config_vlan_hw_stripping(struct bnxt *bp, uint64_t rx_offloads)
+{
+	struct bnxt_vnic_info *vnic = BNXT_GET_DEFAULT_VNIC(bp);
+	int rc;
+
+	/* Destroy, recreate and reconfigure the default vnic */
+	rc = bnxt_free_one_vnic(bp, 0);
+	if (rc)
+		return rc;
+
+	/* default vnic 0 */
+	rc = bnxt_setup_one_vnic(bp, 0);
+	if (rc)
+		return rc;
+
+	if (bp->eth_dev->data->dev_conf.rxmode.offloads &
+	    DEV_RX_OFFLOAD_VLAN_FILTER) {
+		rc = bnxt_add_vlan_filter(bp, 0);
+		bnxt_restore_vlan_filters(bp);
+	} else {
+		rc = bnxt_add_mac_filter(bp, vnic, NULL, 0, 0);
+	}
+
+	rc = bnxt_hwrm_cfa_l2_set_rx_mask(bp, vnic, 0, NULL);
+	if (rc)
+		return rc;
+
+	PMD_DRV_LOG(DEBUG, "VLAN Strip Offload: %d\n",
+		    !!(rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP));
+
+	return rc;
+}
+
 static int
 bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
 {
 	uint64_t rx_offloads = dev->data->dev_conf.rxmode.offloads;
 	struct bnxt *bp = dev->data->dev_private;
-	unsigned int i;
 	int rc;
 
 	rc = is_bnxt_in_error(bp);
@@ -1913,16 +1972,9 @@ bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
 
 	if (mask & ETH_VLAN_STRIP_MASK) {
 		/* Enable or disable VLAN stripping */
-		for (i = 0; i < bp->nr_vnics; i++) {
-			struct bnxt_vnic_info *vnic = &bp->vnic_info[i];
-			if (rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
-				vnic->vlan_strip = true;
-			else
-				vnic->vlan_strip = false;
-			bnxt_hwrm_vnic_cfg(bp, vnic);
-		}
-		PMD_DRV_LOG(DEBUG, "VLAN Strip Offload: %d\n",
-			!!(rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP));
+		rc = bnxt_config_vlan_hw_stripping(bp, rx_offloads);
+		if (rc)
+			return rc;
 	}
 
 	if (mask & ETH_VLAN_EXTEND_MASK) {
Signature · 1 line
2.10.1
author
#6 lore raw

PATCH v2 6/10 net/bnxt: handle hw filter setting when port is stopped

In reply to Kalesh A P

From: Kalesh AP <[email protected]>

Driver destroy the vnic when the port is brought down. Port hw filter setting such as promiscuos, allmulti and vlan filtering will be applied when port is started.

Fixed to return success silently for these callbacks when port is stopped. Also fixed to clear "bp->dev_stopped" before invoking bnxt_vlan_offload_set_op() in bnxt_dev_start_op().

---
 drivers/net/bnxt/bnxt_ethdev.c | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)
drivers/net/bnxt/bnxt_ethdev.c +22 -1
index 0b2c29b..436ecbb 100644
@@ -870,6 +870,7 @@ static int bnxt_dev_start_op(struct rte_eth_dev *eth_dev)
 	eth_dev->data->scattered_rx = bnxt_scattered_rx(eth_dev);
 
 	bnxt_link_update(eth_dev, 1, ETH_LINK_UP);
+	bp->dev_stopped = 0;
 
 	if (rx_offloads & DEV_RX_OFFLOAD_VLAN_FILTER)
 		vlan_mask |= ETH_VLAN_FILTER_MASK;
@@ -884,7 +885,6 @@ static int bnxt_dev_start_op(struct rte_eth_dev *eth_dev)
 
 	bp->flags |= BNXT_FLAG_INIT_DONE;
 	eth_dev->data->dev_started = 1;
-	bp->dev_stopped = 0;
 	pthread_mutex_lock(&bp->def_cp_lock);
 	bnxt_schedule_fw_health_check(bp);
 	pthread_mutex_unlock(&bp->def_cp_lock);
@@ -895,6 +895,7 @@ static int bnxt_dev_start_op(struct rte_eth_dev *eth_dev)
 	bnxt_shutdown_nic(bp);
 	bnxt_free_tx_mbufs(bp);
 	bnxt_free_rx_mbufs(bp);
+	bp->dev_stopped = 1;
 	return rc;
 }
 
@@ -1168,6 +1169,10 @@ static int bnxt_promiscuous_enable_op(struct rte_eth_dev *eth_dev)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (bp->vnic_info == NULL)
 		return 0;
 
@@ -1193,6 +1198,10 @@ static int bnxt_promiscuous_disable_op(struct rte_eth_dev *eth_dev)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (bp->vnic_info == NULL)
 		return 0;
 
@@ -1218,6 +1227,10 @@ static int bnxt_allmulticast_enable_op(struct rte_eth_dev *eth_dev)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (bp->vnic_info == NULL)
 		return 0;
 
@@ -1243,6 +1256,10 @@ static int bnxt_allmulticast_disable_op(struct rte_eth_dev *eth_dev)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (bp->vnic_info == NULL)
 		return 0;
 
@@ -1963,6 +1980,10 @@ bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (mask & ETH_VLAN_FILTER_MASK) {
 		/* Enable or disable VLAN filtering */
 		rc = bnxt_config_vlan_hw_filter(bp, rx_offloads);
Signature · 1 line
2.10.1
author
#7 lore raw

PATCH v2 7/10 net/bnxt: fix a memory leak in port stop

In reply to Kalesh A P

From: Kalesh AP <[email protected]>

The memory for mark table is allocated during port start. But the allocated memory is freed only during port close or driver unload which in turn causes a memory leakage on each port start/stop.

Fixed it by moving the memory free to port stop.

Fixes a968a9f5456d ("net/bnxt: add support for flow mark action")
---
 drivers/net/bnxt/bnxt_ethdev.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +4 -4
index 436ecbb..2661418 100644
@@ -970,7 +970,10 @@ static void bnxt_dev_stop_op(struct rte_eth_dev *eth_dev)
 	bnxt_int_handler(eth_dev);
 	bnxt_shutdown_nic(bp);
 	bnxt_hwrm_if_change(bp, 0);
-	memset(bp->mark_table, 0, BNXT_MARK_TABLE_SZ);
+
+	rte_free(bp->mark_table);
+	bp->mark_table = NULL;
+
 	bp->flags &= ~BNXT_FLAG_RX_VECTOR_PKT_MODE;
 	bp->dev_stopped = 1;
 	bp->rx_cosq_cnt = 0;
@@ -992,9 +995,6 @@ static void bnxt_dev_close_op(struct rte_eth_dev *eth_dev)
 		bp->grp_info = NULL;
 	}
 
-	rte_free(bp->mark_table);
-	bp->mark_table = NULL;
-
 	bnxt_dev_uninit(eth_dev);
 }
 
Signature · 1 line
2.10.1
author
#8 lore raw

PATCH v2 8/10 net/bnxt: use macro for PCI log format

In reply to Kalesh A P

From: Kalesh AP <[email protected]>

Fixes 19e6af01bb36 ("net/bnxt: support get/set EEPROM")
---
 drivers/net/bnxt/bnxt_ethdev.c | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +11 -11
index 2661418..ba3f0a7 100644
@@ -3593,9 +3593,9 @@ bnxt_get_eeprom_length_op(struct rte_eth_dev *dev)
 	if (rc)
 		return rc;
 
-	PMD_DRV_LOG(INFO, "%04x:%02x:%02x:%02x\n",
-		bp->pdev->addr.domain, bp->pdev->addr.bus,
-		bp->pdev->addr.devid, bp->pdev->addr.function);
+	PMD_DRV_LOG(INFO, PCI_PRI_FMT "\n",
+		    bp->pdev->addr.domain, bp->pdev->addr.bus,
+		    bp->pdev->addr.devid, bp->pdev->addr.function);
 
 	rc = bnxt_hwrm_nvm_get_dir_info(bp, &dir_entries, &entry_length);
 	if (rc != 0)
@@ -3617,10 +3617,10 @@ bnxt_get_eeprom_op(struct rte_eth_dev *dev,
 	if (rc)
 		return rc;
 
-	PMD_DRV_LOG(INFO, "%04x:%02x:%02x:%02x in_eeprom->offset = %d "
-		"len = %d\n", bp->pdev->addr.domain,
-		bp->pdev->addr.bus, bp->pdev->addr.devid,
-		bp->pdev->addr.function, in_eeprom->offset, in_eeprom->length);
+	PMD_DRV_LOG(INFO, PCI_PRI_FMT " in_eeprom->offset = %d len = %d\n",
+		    bp->pdev->addr.domain, bp->pdev->addr.bus,
+		    bp->pdev->addr.devid, bp->pdev->addr.function,
+		    in_eeprom->offset, in_eeprom->length);
 
 	if (in_eeprom->offset == 0) /* special offset value to get directory */
 		return bnxt_get_nvram_directory(bp, in_eeprom->length,
@@ -3693,10 +3693,10 @@ bnxt_set_eeprom_op(struct rte_eth_dev *dev,
 	if (rc)
 		return rc;
 
-	PMD_DRV_LOG(INFO, "%04x:%02x:%02x:%02x in_eeprom->offset = %d "
-		"len = %d\n", bp->pdev->addr.domain,
-		bp->pdev->addr.bus, bp->pdev->addr.devid,
-		bp->pdev->addr.function, in_eeprom->offset, in_eeprom->length);
+	PMD_DRV_LOG(INFO, PCI_PRI_FMT " in_eeprom->offset = %d len = %d\n",
+		    bp->pdev->addr.domain, bp->pdev->addr.bus,
+		    bp->pdev->addr.devid, bp->pdev->addr.function,
+		    in_eeprom->offset, in_eeprom->length);
 
 	if (!BNXT_PF(bp)) {
 		PMD_DRV_LOG(ERR, "NVM write not supported from a VF\n");
Signature · 1 line
2.10.1
author
#9 lore raw

PATCH v2 9/10 net/bnxt: release port upon close

In reply to Kalesh A P

From: Somnath Kotur <[email protected]>

Set RTE_ETH_DEV_CLOSE_REMOVE upon probe so all the private resources for the port can be freed by rte_eth_dev_close().

---
 drivers/net/bnxt/bnxt_ethdev.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/bnxt/bnxt_ethdev.c +5 -0
index ba3f0a7..8acfade 100644
@@ -4856,6 +4856,11 @@ bnxt_dev_init(struct rte_eth_dev *eth_dev)
 	if (rc)
 		goto error_free;
 
+	/* Pass the information to the rte_eth_dev_close() that it should also
+	 * release the private port resources.
+	 */
+	eth_dev->data->dev_flags |= RTE_ETH_DEV_CLOSE_REMOVE;
+
 	PMD_DRV_LOG(INFO,
 		    DRV_MODULE_NAME "found at mem %" PRIX64 ", node addr %pM\n",
 		    pci_dev->mem_resource[0].phys_addr,
Signature · 1 line
2.10.1
author
#10 lore raw

PATCH v2 10/10 net/bnxt: fix to cap max rings to minimum of compl rings and stat contexts

In reply to Kalesh A P

From: Somnath Kotur <[email protected]>

Max Tx rings count could be lesser than max Rx rings in some cases, so take this into account as well.

Account for stat contexts available(one for each ring) along with
no: of completion rings(one for each ring) to cap the max no: of Tx /Rx rings that can be possibly created.

Fixes f03e66cb ("net/bnxt: limit queue count for NS3/Stingray devices")
---
 drivers/net/bnxt/bnxt.h | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)
drivers/net/bnxt/bnxt.h +10 -3
index 3487b91..ddb2681 100644
@@ -622,12 +622,19 @@ struct bnxt {
 	uint16_t		max_tx_rings;
 	uint16_t		max_rx_rings;
 #define MAX_STINGRAY_RINGS		128U
-#define BNXT_MAX_RINGS(bp) \
+/* For sake of symmetry, max Tx rings == max Rx rings, one stat ctx for each */
+#define BNXT_MAX_RX_RINGS(bp) \
 	(BNXT_STINGRAY(bp) ? RTE_MIN(RTE_MIN(bp->max_rx_rings, \
 					     MAX_STINGRAY_RINGS), \
-				     bp->max_stat_ctx) : \
-				RTE_MIN(bp->max_rx_rings, bp->max_stat_ctx))
+				     bp->max_stat_ctx / 2U) : \
+				RTE_MIN(bp->max_rx_rings, \
+					bp->max_stat_ctx / 2U))
+#define BNXT_MAX_TX_RINGS(bp) \
+	(RTE_MIN((bp)->max_tx_rings, BNXT_MAX_RX_RINGS(bp)))
 
+#define BNXT_MAX_RINGS(bp) \
+	(RTE_MIN((((bp)->max_cp_rings - BNXT_NUM_ASYNC_CPR(bp)) / 2U), \
+		 BNXT_MAX_TX_RINGS(bp)))
 	uint16_t		max_nq_rings;
 	uint16_t		max_l2_ctx;
 	uint16_t		max_rx_em_flows;
Signature · 1 line
2.10.1
#11 lore raw

v3 cover bnxt patchset with bug fixes

In reply to Kalesh A P

v1->v2: fixed the correct the commit ids in Fixes tag. v2->v3: shorten commit log and update commit ids for Fixes tag.

Note that Fixes tag in patch 07/10 is correct.

Kalesh AP (8):
net/bnxt: handle flow create failure
net/bnxt: fix probe failure in FreeBSD net/bnxt: fix to use correct IOVA mapping net/bnxt: fix enable/disable VLAN filtering net/bnxt: fix VLAN strip support
net/bnxt: handle HW filter setting when port is stopped net/bnxt: fix a memory leak in port stop net/bnxt: use macro for PCI log format

Somnath Kotur (2):
net/bnxt: release port upon close
net/bnxt: fix calculation of max rings

 drivers/net/bnxt/bnxt.h        |  21 +-
 drivers/net/bnxt/bnxt_ethdev.c | 422 ++++++++++++++++++---------------
 drivers/net/bnxt/bnxt_filter.c |   5 +-
 drivers/net/bnxt/bnxt_flow.c   |   4 +-
 drivers/net/bnxt/bnxt_hwrm.c   |  28 +--
 drivers/net/bnxt/bnxt_irq.c    |   8 +
 drivers/net/bnxt/bnxt_ring.c   |  17 --
 drivers/net/bnxt/bnxt_vnic.c   |  11 -
 8 files changed, 273 insertions(+), 243 deletions(-)
Signature · 1 line
2.21.0 (Apple Git-122.2)
#12 lore raw

PATCH v3 1/10 net/bnxt: handle flow create failure

From: Kalesh AP <[email protected]>

If flow create fails due to not enough filter resources, driver does not populate the rte_flow_error using rte_flow_error_set().

Since "rte_errno" could have garbage value and is not relaiable, it could cause a segfault in the stack in port_flow_complain().

Fix it to set rte_flow_error using rte_flow_error_set() when flow create fails due to not enough filter resources.

---
 drivers/net/bnxt/bnxt_flow.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/net/bnxt/bnxt_flow.c +3 -1
index cde1fa41c..5564c5363 100644
@@ -1702,7 +1702,9 @@ bnxt_flow_create(struct rte_eth_dev *dev,
 
 	filter = bnxt_get_unused_filter(bp);
 	if (filter == NULL) {
-		PMD_DRV_LOG(ERR, "Not enough resources for a new flow.\n");
+		rte_flow_error_set(error, ENOSPC,
+				   RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+				   "Not enough resources for a new flow");
 		goto free_flow;
 	}
 
Signature · 1 line
2.21.0 (Apple Git-122.2)
#13 lore raw

Re: [dpdk-dev] [PATCH v3 01/10] net/bnxt: handle flow create failure

Quoted 11 lines · Ajit Khaparde

From: Kalesh AP <[email protected]>

If flow create fails due to not enough filter resources, driver does not populate the rte_flow_error using rte_flow_error_set().

Since "rte_errno" could have garbage value and is not relaiable, it could cause a segfault in the stack in port_flow_complain().

Fix it to set rte_flow_error using rte_flow_error_set() when flow create fails due to not enough filter resources.

Hi Ajit, Kalesh,

In patch title, 'handle' seems means fix, can you please prefer the 'fix' since it become kind of keyword to define the patch content. This also helps us missing the "Fixes: " tag as it has been here.

Thanks,
ferruh

Quoted 23 lines
---
 drivers/net/bnxt/bnxt_flow.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/net/bnxt/bnxt_flow.c +3 -1
index cde1fa41c..5564c5363 100644
@@ -1702,7 +1702,9 @@ bnxt_flow_create(struct rte_eth_dev *dev,
 
 	filter = bnxt_get_unused_filter(bp);
 	if (filter == NULL) {
-		PMD_DRV_LOG(ERR, "Not enough resources for a new flow.\n");
+		rte_flow_error_set(error, ENOSPC,
+				   RTE_FLOW_ERROR_TYPE_HANDLE, NULL,
+				   "Not enough resources for a new flow");
 		goto free_flow;
 	}
 
#14 lore raw

PATCH v3 2/10 net/bnxt: fix probe failure in FreeBSD

In reply to Ajit Khaparde

From: Kalesh AP <[email protected]>

In FreeBSD environment, nic_uio driver does not support interrupts and rte_intr_callback_register() will fail to register interrupts which in turn causes bnxt driver probe failure.

Fixed driver to ignore interrupt callback failures in FreeBSD. Also fixed to not use a dedicated completion ring for async events from FW and process these events on RXQ0 in FreeBSD.

Fixes 6de4c538b393 ("net/bnxt: fix error handling in port start")
Fixes 43f78b380f89 ("net/bnxt: retry IRQ callback deregistration")
---
 drivers/net/bnxt/bnxt.h        | 8 ++++++++
 drivers/net/bnxt/bnxt_ethdev.c | 3 +++
 drivers/net/bnxt/bnxt_irq.c    | 8 ++++++++
 3 files changed, 19 insertions(+)
drivers/net/bnxt/bnxt.h +8 -0
index 21ca059b8..3487b917e 100644
@@ -117,6 +117,14 @@
 #define BNXT_NUM_ASYNC_CPR(bp) 1
 #endif
 
+/* In FreeBSD OS, nic_uio driver does not support interrupts */
+#ifdef RTE_EXEC_ENV_FREEBSD
+#ifdef BNXT_NUM_ASYNC_CPR
+#undef BNXT_NUM_ASYNC_CPR
+#endif
+#define BNXT_NUM_ASYNC_CPR(bp)	0
+#endif
+
 #define BNXT_MISC_VEC_ID               RTE_INTR_VEC_ZERO_OFFSET
 #define BNXT_RX_VEC_START              RTE_INTR_VEC_RXTX_OFFSET
 
drivers/net/bnxt/bnxt_ethdev.c +3 -0
index 7b5df9ac1..879ea580f 100644
@@ -439,8 +439,11 @@ static int bnxt_init_chip(struct bnxt *bp)
 
 	/* enable uio/vfio intr/eventfd mapping */
 	rc = rte_intr_enable(intr_handle);
+#ifndef RTE_EXEC_ENV_FREEBSD
+	/* In FreeBSD OS, nic_uio driver does not support interrupts */
 	if (rc)
 		goto err_free;
+#endif
 
 	rc = bnxt_get_hwrm_link_config(bp, &new);
 	if (rc) {
drivers/net/bnxt/bnxt_irq.c +8 -0
index 846325ea9..40e1b0c98 100644
@@ -181,5 +181,13 @@ int bnxt_request_int(struct bnxt *bp)
 			irq->requested = 1;
 	}
 
+#ifdef RTE_EXEC_ENV_FREEBSD
+	/**
+	 * In FreeBSD OS, nic_uio does not support interrupts and
+	 * interrupt register callback will fail.
+	 */
+	rc = 0;
+#endif
+
 	return rc;
 }
Signature · 1 line
2.21.0 (Apple Git-122.2)
#15 lore raw

PATCH v3 3/10 net/bnxt: fix to use correct IOVA mapping

In reply to Ajit Khaparde

From: Kalesh AP <[email protected]>

Use rte_malloc_virt2iova() to obtain the IO address of a virtual address obtained through rte_malloc().

Fixed to use the iova address returned by rte_memzone_reserve_aligned() as the call always returns with populating "mz->iova" with rte_malloc_virt2iova(mz->addr).

Removed redundant rte_mem_lock_page() call to lock the pages.

Fixes f55e12f33416 ("net/bnxt: support extended port counters")
---
 drivers/net/bnxt/bnxt_ethdev.c | 49 ----------------------------------
 drivers/net/bnxt/bnxt_hwrm.c   | 28 ++++++++-----------
 drivers/net/bnxt/bnxt_ring.c   | 17 ------------
 drivers/net/bnxt/bnxt_vnic.c   | 11 --------
 4 files changed, 11 insertions(+), 94 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +0 -49
index 879ea580f..a948c78cb 100644
@@ -4233,18 +4233,6 @@ static int bnxt_alloc_ctx_mem_blk(struct bnxt *bp,
 
 		memset(mz->addr, 0, mz->len);
 		mz_phys_addr = mz->iova;
-		if ((unsigned long)mz->addr == mz_phys_addr) {
-			PMD_DRV_LOG(DEBUG,
-				    "physical address same as virtual\n");
-			PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-			mz_phys_addr = rte_mem_virt2iova(mz->addr);
-			if (mz_phys_addr == RTE_BAD_IOVA) {
-				PMD_DRV_LOG(ERR,
-					"unable to map addr to phys memory\n");
-				return -ENOMEM;
-			}
-		}
-		rte_mem_lock_page(((char *)mz->addr));
 
 		rmem->pg_tbl = mz->addr;
 		rmem->pg_tbl_map = mz_phys_addr;
@@ -4268,22 +4256,8 @@ static int bnxt_alloc_ctx_mem_blk(struct bnxt *bp,
 
 	memset(mz->addr, 0, mz->len);
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual.\n");
-		PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-		for (sz = 0; sz < mem_size; sz += BNXT_PAGE_SIZE)
-			rte_mem_lock_page(((char *)mz->addr) + sz);
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-				    "unable to map addr to phys memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	for (sz = 0, i = 0; sz < mem_size; sz += BNXT_PAGE_SIZE, i++) {
-		rte_mem_lock_page(((char *)mz->addr) + sz);
 		rmem->pg_arr[i] = ((char *)mz->addr) + sz;
 		rmem->dma_arr[i] = mz_phys_addr + sz;
 
@@ -4460,18 +4434,6 @@ static int bnxt_alloc_stats_mem(struct bnxt *bp)
 	}
 	memset(mz->addr, 0, mz->len);
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual.\n");
-		PMD_DRV_LOG(DEBUG,
-			    "Using rte_mem_virt2iova()\n");
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-				    "Can't map address to physical memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	bp->rx_mem_zone = (const void *)mz;
 	bp->hw_rx_port_stats = mz->addr;
@@ -4498,17 +4460,6 @@ static int bnxt_alloc_stats_mem(struct bnxt *bp)
 	}
 	memset(mz->addr, 0, mz->len);
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual\n");
-		PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-				    "Can't map address to physical memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	bp->tx_mem_zone = (const void *)mz;
 	bp->hw_tx_port_stats = mz->addr;
drivers/net/bnxt/bnxt_hwrm.c +11 -17
index 50272dcf7..3b013396b 100644
@@ -309,8 +309,8 @@ int bnxt_hwrm_cfa_l2_set_rx_mask(struct bnxt *bp,
 	if (vlan_table) {
 		if (!(mask & HWRM_CFA_L2_SET_RX_MASK_INPUT_MASK_VLAN_NONVLAN))
 			mask |= HWRM_CFA_L2_SET_RX_MASK_INPUT_MASK_VLANONLY;
-		req.vlan_tag_tbl_addr = rte_cpu_to_le_64(
-			 rte_mem_virt2iova(vlan_table));
+		req.vlan_tag_tbl_addr =
+			rte_cpu_to_le_64(rte_malloc_virt2iova(vlan_table));
 		req.num_vlan_tags = rte_cpu_to_le_32((uint32_t)vlan_count);
 	}
 	req.mask = rte_cpu_to_le_32(mask);
@@ -351,7 +351,7 @@ int bnxt_hwrm_cfa_vlan_antispoof_cfg(struct bnxt *bp, uint16_t fid,
 	req.fid = rte_cpu_to_le_16(fid);
 
 	req.vlan_tag_mask_tbl_addr =
-		rte_cpu_to_le_64(rte_mem_virt2iova(vlan_table));
+		rte_cpu_to_le_64(rte_malloc_virt2iova(vlan_table));
 	req.num_vlan_entries = rte_cpu_to_le_32((uint32_t)vlan_count);
 
 	rc = bnxt_hwrm_send_message(bp, &req, sizeof(req), BNXT_USE_CHIMP_MB);
@@ -1024,9 +1024,8 @@ int bnxt_hwrm_ver_get(struct bnxt *bp)
 			rc = -ENOMEM;
 			goto error;
 		}
-		rte_mem_lock_page(bp->hwrm_cmd_resp_addr);
 		bp->hwrm_cmd_resp_dma_addr =
-			rte_mem_virt2iova(bp->hwrm_cmd_resp_addr);
+			rte_malloc_virt2iova(bp->hwrm_cmd_resp_addr);
 		if (bp->hwrm_cmd_resp_dma_addr == RTE_BAD_IOVA) {
 			PMD_DRV_LOG(ERR,
 			"Unable to map response buffer to physical memory.\n");
@@ -1061,9 +1060,8 @@ int bnxt_hwrm_ver_get(struct bnxt *bp)
 			rc = -ENOMEM;
 			goto error;
 		}
-		rte_mem_lock_page(bp->hwrm_short_cmd_req_addr);
 		bp->hwrm_short_cmd_req_dma_addr =
-			rte_mem_virt2iova(bp->hwrm_short_cmd_req_addr);
+			rte_malloc_virt2iova(bp->hwrm_short_cmd_req_addr);
 		if (bp->hwrm_short_cmd_req_dma_addr == RTE_BAD_IOVA) {
 			rte_free(bp->hwrm_short_cmd_req_addr);
 			PMD_DRV_LOG(ERR,
@@ -2471,11 +2469,10 @@ int bnxt_alloc_hwrm_resources(struct bnxt *bp)
 		pdev->addr.bus, pdev->addr.devid, pdev->addr.function);
 	bp->max_resp_len = HWRM_MAX_RESP_LEN;
 	bp->hwrm_cmd_resp_addr = rte_malloc(type, bp->max_resp_len, 0);
-	rte_mem_lock_page(bp->hwrm_cmd_resp_addr);
 	if (bp->hwrm_cmd_resp_addr == NULL)
 		return -ENOMEM;
 	bp->hwrm_cmd_resp_dma_addr =
-		rte_mem_virt2iova(bp->hwrm_cmd_resp_addr);
+		rte_malloc_virt2iova(bp->hwrm_cmd_resp_addr);
 	if (bp->hwrm_cmd_resp_dma_addr == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map response address to physical memory\n");
@@ -3421,7 +3418,7 @@ int bnxt_hwrm_func_buf_rgtr(struct bnxt *bp)
 			 page_getenum(bp->pf.active_vfs * HWRM_MAX_REQ_LEN));
 	req.req_buf_len = rte_cpu_to_le_16(HWRM_MAX_REQ_LEN);
 	req.req_buf_page_addr0 =
-		rte_cpu_to_le_64(rte_mem_virt2iova(bp->pf.vf_req_buf));
+		rte_cpu_to_le_64(rte_malloc_virt2iova(bp->pf.vf_req_buf));
 	if (req.req_buf_page_addr0 == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map buffer address to physical memory\n");
@@ -3851,10 +3848,9 @@ int bnxt_get_nvram_directory(struct bnxt *bp, uint32_t len, uint8_t *data)
 
 	buflen = dir_entries * entry_length;
 	buf = rte_malloc("nvm_dir", buflen, 0);
-	rte_mem_lock_page(buf);
 	if (buf == NULL)
 		return -ENOMEM;
-	dma_handle = rte_mem_virt2iova(buf);
+	dma_handle = rte_malloc_virt2iova(buf);
 	if (dma_handle == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map response address to physical memory\n");
@@ -3885,11 +3881,10 @@ int bnxt_hwrm_get_nvram_item(struct bnxt *bp, uint32_t index,
 	struct hwrm_nvm_read_output *resp = bp->hwrm_cmd_resp_addr;
 
 	buf = rte_malloc("nvm_item", length, 0);
-	rte_mem_lock_page(buf);
 	if (!buf)
 		return -ENOMEM;
 
-	dma_handle = rte_mem_virt2iova(buf);
+	dma_handle = rte_malloc_virt2iova(buf);
 	if (dma_handle == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map response address to physical memory\n");
@@ -3939,11 +3934,10 @@ int bnxt_hwrm_flash_nvram(struct bnxt *bp, uint16_t dir_type,
 	uint8_t *buf;
 
 	buf = rte_malloc("nvm_write", data_len, 0);
-	rte_mem_lock_page(buf);
 	if (!buf)
 		return -ENOMEM;
 
-	dma_handle = rte_mem_virt2iova(buf);
+	dma_handle = rte_malloc_virt2iova(buf);
 	if (dma_handle == RTE_BAD_IOVA) {
 		PMD_DRV_LOG(ERR,
 			"unable to map response address to physical memory\n");
@@ -4006,7 +4000,7 @@ static int bnxt_hwrm_func_vf_vnic_query(struct bnxt *bp, uint16_t vf,
 
 	req.vf_id = rte_cpu_to_le_16(bp->pf.first_vf_id + vf);
 	req.max_vnic_id_cnt = rte_cpu_to_le_32(bp->pf.total_vnics);
-	req.vnic_id_tbl_addr = rte_cpu_to_le_64(rte_mem_virt2iova(vnic_ids));
+	req.vnic_id_tbl_addr = rte_cpu_to_le_64(rte_malloc_virt2iova(vnic_ids));
 
 	if (req.vnic_id_tbl_addr == RTE_BAD_IOVA) {
 		HWRM_UNLOCK();
drivers/net/bnxt/bnxt_ring.c +0 -17
index ea46fa9bc..d6e4e8a28 100644
@@ -110,9 +110,7 @@ int bnxt_alloc_rings(struct bnxt *bp, uint16_t qidx,
 	uint64_t rx_offloads = bp->eth_dev->data->dev_conf.rxmode.offloads;
 	const struct rte_memzone *mz = NULL;
 	char mz_name[RTE_MEMZONE_NAMESIZE];
-	rte_iova_t mz_phys_addr_base;
 	rte_iova_t mz_phys_addr;
-	int sz;
 
 	int stats_len = (tx_ring_info || rx_ring_info) ?
 	    RTE_CACHE_LINE_ROUNDUP(sizeof(struct hwrm_stat_ctx_query_output) -
@@ -214,22 +212,7 @@ int bnxt_alloc_rings(struct bnxt *bp, uint16_t qidx,
 			return -ENOMEM;
 	}
 	memset(mz->addr, 0, mz->len);
-	mz_phys_addr_base = mz->iova;
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr_base) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual.\n");
-		PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-		for (sz = 0; sz < total_alloc_len; sz += getpagesize())
-			rte_mem_lock_page(((char *)mz->addr) + sz);
-		mz_phys_addr_base = rte_mem_virt2iova(mz->addr);
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-			"unable to map ring address to physical memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	if (tx_ring_info) {
 		txq->mz = mz;
drivers/net/bnxt/bnxt_vnic.c +0 -11
index 104342e13..bc054a8e0 100644
@@ -150,17 +150,6 @@ int bnxt_alloc_vnic_attributes(struct bnxt *bp)
 			return -ENOMEM;
 	}
 	mz_phys_addr = mz->iova;
-	if ((unsigned long)mz->addr == mz_phys_addr) {
-		PMD_DRV_LOG(DEBUG,
-			    "Memzone physical address same as virtual.\n");
-		PMD_DRV_LOG(DEBUG, "Using rte_mem_virt2iova()\n");
-		mz_phys_addr = rte_mem_virt2iova(mz->addr);
-		if (mz_phys_addr == RTE_BAD_IOVA) {
-			PMD_DRV_LOG(ERR,
-				    "unable to map to physical memory\n");
-			return -ENOMEM;
-		}
-	}
 
 	for (i = 0; i < max_vnics; i++) {
 		vnic = &bp->vnic_info[i];
Signature · 1 line
2.21.0 (Apple Git-122.2)
#16 lore raw

PATCH v3 4/10 net/bnxt: fix enable/disable VLAN filtering

In reply to Ajit Khaparde

From: Kalesh AP <[email protected]>

There is no condition check for the user requested operation for VLAN filtering. As a result, VLAN filtering is getting disabled when the user enables/disables VLAN stripping on same port.

The function bnxt_hwrm_clear_l2_filter() didn't actually free L2 filter in HW if the reference count of filter is zero.

Fixed it by incrementing the reference count of filter in bnxt_alloc_filter() routine.

Because of the recent changes in bnxt_hwrm_clear_l2_filter(), change was needed in the routine bnxt_set_default_mac_addr_op() to destroy and re-create the default filter when the user changes the default MAC of the port.

Fixes 5c1171c97216 ("net/bnxt: refactor filter/flow")
Fixes 6118503d8071 ("net/bnxt: fix VLAN filtering")
---
 drivers/net/bnxt/bnxt_ethdev.c | 66 ++++++++++++++++++----------------
 drivers/net/bnxt/bnxt_filter.c |  5 +--
 2 files changed, 39 insertions(+), 32 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +36 -30
index a948c78cb..83f475d49 100644
@@ -1860,18 +1860,12 @@ static int bnxt_del_dflt_mac_filter(struct bnxt *bp,
 }
 
 static int
-bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
+bnxt_config_vlan_hw_filter(struct bnxt *bp, uint64_t rx_offloads)
 {
-	struct bnxt *bp = dev->data->dev_private;
-	uint64_t rx_offloads = dev->data->dev_conf.rxmode.offloads;
 	struct bnxt_vnic_info *vnic;
 	unsigned int i;
 	int rc;
 
-	rc = is_bnxt_in_error(bp);
-	if (rc)
-		return rc;
-
 	vnic = BNXT_GET_DEFAULT_VNIC(bp);
 	if (!(rx_offloads & DEV_RX_OFFLOAD_VLAN_FILTER)) {
 		/* Remove any VLAN filters programmed */
@@ -1895,6 +1889,28 @@ bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
 	PMD_DRV_LOG(DEBUG, "VLAN Filtering: %d\n",
 		    !!(rx_offloads & DEV_RX_OFFLOAD_VLAN_FILTER));
 
+	return 0;
+}
+
+static int
+bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
+{
+	uint64_t rx_offloads = dev->data->dev_conf.rxmode.offloads;
+	struct bnxt *bp = dev->data->dev_private;
+	unsigned int i;
+	int rc;
+
+	rc = is_bnxt_in_error(bp);
+	if (rc)
+		return rc;
+
+	if (mask & ETH_VLAN_FILTER_MASK) {
+		/* Enable or disable VLAN filtering */
+		rc = bnxt_config_vlan_hw_filter(bp, rx_offloads);
+		if (rc)
+			return rc;
+	}
+
 	if (mask & ETH_VLAN_STRIP_MASK) {
 		/* Enable or disable VLAN stripping */
 		for (i = 0; i < bp->nr_vnics; i++) {
@@ -1984,7 +2000,6 @@ bnxt_set_default_mac_addr_op(struct rte_eth_dev *dev,
 	struct bnxt *bp = dev->data->dev_private;
 	/* Default Filter is tied to VNIC 0 */
 	struct bnxt_vnic_info *vnic = BNXT_GET_DEFAULT_VNIC(bp);
-	struct bnxt_filter_info *filter;
 	int rc;
 
 	rc = is_bnxt_in_error(bp);
@@ -1997,32 +2012,23 @@ bnxt_set_default_mac_addr_op(struct rte_eth_dev *dev,
 	if (rte_is_zero_ether_addr(addr))
 		return -EINVAL;
 
-	STAILQ_FOREACH(filter, &vnic->filter, next) {
-		/* Default Filter is at Index 0 */
-		if (filter->mac_index != 0)
-			continue;
-
-		memcpy(filter->l2_addr, addr, RTE_ETHER_ADDR_LEN);
-		memset(filter->l2_addr_mask, 0xff, RTE_ETHER_ADDR_LEN);
-		filter->flags |= HWRM_CFA_L2_FILTER_ALLOC_INPUT_FLAGS_PATH_RX |
-			HWRM_CFA_L2_FILTER_ALLOC_INPUT_FLAGS_OUTERMOST;
-		filter->enables |=
-			HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_ADDR |
-			HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_ADDR_MASK;
+	/* Check if the requested MAC is already added */
+	if (memcmp(addr, bp->mac_addr, RTE_ETHER_ADDR_LEN) == 0)
+		return 0;
 
-		rc = bnxt_hwrm_set_l2_filter(bp, vnic->fw_vnic_id, filter);
-		if (rc) {
-			memcpy(filter->l2_addr, bp->mac_addr,
-			       RTE_ETHER_ADDR_LEN);
-			return rc;
-		}
+	/* Destroy filter and re-create it */
+	bnxt_del_dflt_mac_filter(bp, vnic);
 
-		memcpy(bp->mac_addr, addr, RTE_ETHER_ADDR_LEN);
-		PMD_DRV_LOG(DEBUG, "Set MAC addr\n");
-		return 0;
+	memcpy(bp->mac_addr, addr, RTE_ETHER_ADDR_LEN);
+	if (dev->data->dev_conf.rxmode.offloads & DEV_RX_OFFLOAD_VLAN_FILTER) {
+		/* This filter will allow only untagged packets */
+		rc = bnxt_add_vlan_filter(bp, 0);
+	} else {
+		rc = bnxt_add_mac_filter(bp, vnic, addr, 0, 0);
 	}
 
-	return 0;
+	PMD_DRV_LOG(DEBUG, "Set MAC addr\n");
+	return rc;
 }
 
 static int
drivers/net/bnxt/bnxt_filter.c +3 -2
index 0c410f8ba..b31f10479 100644
@@ -39,9 +39,10 @@ struct bnxt_filter_info *bnxt_alloc_filter(struct bnxt *bp)
 	filter->flags = HWRM_CFA_L2_FILTER_ALLOC_INPUT_FLAGS_PATH_RX;
 	filter->enables = HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_ADDR |
 			HWRM_CFA_L2_FILTER_ALLOC_INPUT_ENABLES_L2_ADDR_MASK;
-	memcpy(filter->l2_addr, bp->eth_dev->data->mac_addrs->addr_bytes,
-	       RTE_ETHER_ADDR_LEN);
+	memcpy(filter->l2_addr, bp->mac_addr, RTE_ETHER_ADDR_LEN);
 	memset(filter->l2_addr_mask, 0xff, RTE_ETHER_ADDR_LEN);
+	/* bump up the reference count of filter */
+	filter->l2_ref_cnt++;
 	return filter;
 }
 
Signature · 1 line
2.21.0 (Apple Git-122.2)
#17 lore raw

PATCH v3 5/10 net/bnxt: fix VLAN strip support

In reply to Ajit Khaparde

From: Kalesh AP <[email protected]>

HWRM_VNIC_CFG command to configure vnic dynamically with traffic running is not working. Driver has to free and recreate the vnic and then reconfigure the vnic filters.

Fixes 7fe5668d2ea3 ("net/bnxt: support VLAN filter and strip")
---
 drivers/net/bnxt/bnxt_ethdev.c | 248 ++++++++++++++++++++-------------
 1 file changed, 150 insertions(+), 98 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +150 -98
index 83f475d49..0b2c29bc8 100644
@@ -132,6 +132,7 @@ static int bnxt_dev_uninit(struct rte_eth_dev *eth_dev);
 static int bnxt_init_resources(struct bnxt *bp, bool reconfig_dev);
 static int bnxt_uninit_resources(struct bnxt *bp, bool reconfig_dev);
 static void bnxt_cancel_fw_health_check(struct bnxt *bp);
+static int bnxt_restore_vlan_filters(struct bnxt *bp);
 
 int is_bnxt_in_error(struct bnxt *bp)
 {
@@ -228,14 +229,97 @@ static int bnxt_alloc_mem(struct bnxt *bp, bool reconfig)
 	return rc;
 }
 
-static int bnxt_init_chip(struct bnxt *bp)
+static int bnxt_setup_one_vnic(struct bnxt *bp, uint16_t vnic_id)
 {
+	struct rte_eth_conf *dev_conf = &bp->eth_dev->data->dev_conf;
+	struct bnxt_vnic_info *vnic = &bp->vnic_info[vnic_id];
+	uint64_t rx_offloads = dev_conf->rxmode.offloads;
 	struct bnxt_rx_queue *rxq;
+	unsigned int j;
+	int rc;
+
+	rc = bnxt_vnic_grp_alloc(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	PMD_DRV_LOG(DEBUG, "vnic[%d] = %p vnic->fw_grp_ids = %p\n",
+		    vnic_id, vnic, vnic->fw_grp_ids);
+
+	rc = bnxt_hwrm_vnic_alloc(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	/* Alloc RSS context only if RSS mode is enabled */
+	if (dev_conf->rxmode.mq_mode & ETH_MQ_RX_RSS) {
+		int j, nr_ctxs = bnxt_rss_ctxts(bp);
+
+		rc = 0;
+		for (j = 0; j < nr_ctxs; j++) {
+			rc = bnxt_hwrm_vnic_ctx_alloc(bp, vnic, j);
+			if (rc)
+				break;
+		}
+		if (rc) {
+			PMD_DRV_LOG(ERR,
+				    "HWRM vnic %d ctx %d alloc failure rc: %x\n",
+				    vnic_id, j, rc);
+			goto err_out;
+		}
+		vnic->num_lb_ctxts = nr_ctxs;
+	}
+
+	/*
+	 * Firmware sets pf pair in default vnic cfg. If the VLAN strip
+	 * setting is not available at this time, it will not be
+	 * configured correctly in the CFA.
+	 */
+	if (rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
+		vnic->vlan_strip = true;
+	else
+		vnic->vlan_strip = false;
+
+	rc = bnxt_hwrm_vnic_cfg(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	rc = bnxt_set_hwrm_vnic_filters(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	for (j = 0; j < bp->rx_num_qs_per_vnic; j++) {
+		rxq = bp->eth_dev->data->rx_queues[j];
+
+		PMD_DRV_LOG(DEBUG,
+			    "rxq[%d]->vnic=%p vnic->fw_grp_ids=%p\n",
+			    j, rxq->vnic, rxq->vnic->fw_grp_ids);
+
+		if (BNXT_HAS_RING_GRPS(bp) && rxq->rx_deferred_start)
+			rxq->vnic->fw_grp_ids[j] = INVALID_HW_RING_ID;
+	}
+
+	rc = bnxt_vnic_rss_configure(bp, vnic);
+	if (rc)
+		goto err_out;
+
+	bnxt_hwrm_vnic_plcmode_cfg(bp, vnic);
+
+	if (rx_offloads & DEV_RX_OFFLOAD_TCP_LRO)
+		bnxt_hwrm_vnic_tpa_cfg(bp, vnic, 1);
+	else
+		bnxt_hwrm_vnic_tpa_cfg(bp, vnic, 0);
+
+	return 0;
+err_out:
+	PMD_DRV_LOG(ERR, "HWRM vnic %d cfg failure rc: %x\n",
+		    vnic_id, rc);
+	return rc;
+}
+
+static int bnxt_init_chip(struct bnxt *bp)
+{
 	struct rte_eth_link new;
 	struct rte_pci_device *pci_dev = RTE_ETH_DEV_TO_PCI(bp->eth_dev);
-	struct rte_eth_conf *dev_conf = &bp->eth_dev->data->dev_conf;
 	struct rte_intr_handle *intr_handle = &pci_dev->intr_handle;
-	uint64_t rx_offloads = dev_conf->rxmode.offloads;
 	uint32_t intr_vector = 0;
 	uint32_t queue_id, base = BNXT_MISC_VEC_ID;
 	uint32_t vec = BNXT_MISC_VEC_ID;
@@ -303,93 +387,11 @@ static int bnxt_init_chip(struct bnxt *bp)
 
 	/* VNIC configuration */
 	for (i = 0; i < bp->nr_vnics; i++) {
-		struct rte_eth_conf *dev_conf = &bp->eth_dev->data->dev_conf;
-		struct bnxt_vnic_info *vnic = &bp->vnic_info[i];
-
-		rc = bnxt_vnic_grp_alloc(bp, vnic);
+		rc = bnxt_setup_one_vnic(bp, i);
 		if (rc)
 			goto err_out;
-
-		PMD_DRV_LOG(DEBUG, "vnic[%d] = %p vnic->fw_grp_ids = %p\n",
-			    i, vnic, vnic->fw_grp_ids);
-
-		rc = bnxt_hwrm_vnic_alloc(bp, vnic);
-		if (rc) {
-			PMD_DRV_LOG(ERR, "HWRM vnic %d alloc failure rc: %x\n",
-				i, rc);
-			goto err_out;
-		}
-
-		/* Alloc RSS context only if RSS mode is enabled */
-		if (dev_conf->rxmode.mq_mode & ETH_MQ_RX_RSS) {
-			int j, nr_ctxs = bnxt_rss_ctxts(bp);
-
-			rc = 0;
-			for (j = 0; j < nr_ctxs; j++) {
-				rc = bnxt_hwrm_vnic_ctx_alloc(bp, vnic, j);
-				if (rc)
-					break;
-			}
-			if (rc) {
-				PMD_DRV_LOG(ERR,
-				  "HWRM vnic %d ctx %d alloc failure rc: %x\n",
-				  i, j, rc);
-				goto err_out;
-			}
-			vnic->num_lb_ctxts = nr_ctxs;
-		}
-
-		/*
-		 * Firmware sets pf pair in default vnic cfg. If the VLAN strip
-		 * setting is not available at this time, it will not be
-		 * configured correctly in the CFA.
-		 */
-		if (rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
-			vnic->vlan_strip = true;
-		else
-			vnic->vlan_strip = false;
-
-		rc = bnxt_hwrm_vnic_cfg(bp, vnic);
-		if (rc) {
-			PMD_DRV_LOG(ERR, "HWRM vnic %d cfg failure rc: %x\n",
-				i, rc);
-			goto err_out;
-		}
-
-		rc = bnxt_set_hwrm_vnic_filters(bp, vnic);
-		if (rc) {
-			PMD_DRV_LOG(ERR,
-				"HWRM vnic %d filter failure rc: %x\n",
-				i, rc);
-			goto err_out;
-		}
-
-		for (j = 0; j < bp->rx_num_qs_per_vnic; j++) {
-			rxq = bp->eth_dev->data->rx_queues[j];
-
-			PMD_DRV_LOG(DEBUG,
-				    "rxq[%d]->vnic=%p vnic->fw_grp_ids=%p\n",
-				    j, rxq->vnic, rxq->vnic->fw_grp_ids);
-
-			if (BNXT_HAS_RING_GRPS(bp) && rxq->rx_deferred_start)
-				rxq->vnic->fw_grp_ids[j] = INVALID_HW_RING_ID;
-		}
-
-		rc = bnxt_vnic_rss_configure(bp, vnic);
-		if (rc) {
-			PMD_DRV_LOG(ERR,
-				    "HWRM vnic set RSS failure rc: %x\n", rc);
-			goto err_out;
-		}
-
-		bnxt_hwrm_vnic_plcmode_cfg(bp, vnic);
-
-		if (bp->eth_dev->data->dev_conf.rxmode.offloads &
-		    DEV_RX_OFFLOAD_TCP_LRO)
-			bnxt_hwrm_vnic_tpa_cfg(bp, vnic, 1);
-		else
-			bnxt_hwrm_vnic_tpa_cfg(bp, vnic, 0);
 	}
+
 	rc = bnxt_hwrm_cfa_l2_set_rx_mask(bp, &bp->vnic_info[0], 0, NULL);
 	if (rc) {
 		PMD_DRV_LOG(ERR,
@@ -1892,12 +1894,69 @@ bnxt_config_vlan_hw_filter(struct bnxt *bp, uint64_t rx_offloads)
 	return 0;
 }
 
+static int bnxt_free_one_vnic(struct bnxt *bp, uint16_t vnic_id)
+{
+	struct bnxt_vnic_info *vnic = &bp->vnic_info[vnic_id];
+	unsigned int i;
+	int rc;
+
+	/* Destroy vnic filters and vnic */
+	if (bp->eth_dev->data->dev_conf.rxmode.offloads &
+	    DEV_RX_OFFLOAD_VLAN_FILTER) {
+		for (i = 0; i < RTE_ETHER_MAX_VLAN_ID; i++)
+			bnxt_del_vlan_filter(bp, i);
+	}
+	bnxt_del_dflt_mac_filter(bp, vnic);
+
+	rc = bnxt_hwrm_vnic_free(bp, vnic);
+	if (rc)
+		return rc;
+
+	rte_free(vnic->fw_grp_ids);
+	vnic->fw_grp_ids = NULL;
+
+	return 0;
+}
+
+static int
+bnxt_config_vlan_hw_stripping(struct bnxt *bp, uint64_t rx_offloads)
+{
+	struct bnxt_vnic_info *vnic = BNXT_GET_DEFAULT_VNIC(bp);
+	int rc;
+
+	/* Destroy, recreate and reconfigure the default vnic */
+	rc = bnxt_free_one_vnic(bp, 0);
+	if (rc)
+		return rc;
+
+	/* default vnic 0 */
+	rc = bnxt_setup_one_vnic(bp, 0);
+	if (rc)
+		return rc;
+
+	if (bp->eth_dev->data->dev_conf.rxmode.offloads &
+	    DEV_RX_OFFLOAD_VLAN_FILTER) {
+		rc = bnxt_add_vlan_filter(bp, 0);
+		bnxt_restore_vlan_filters(bp);
+	} else {
+		rc = bnxt_add_mac_filter(bp, vnic, NULL, 0, 0);
+	}
+
+	rc = bnxt_hwrm_cfa_l2_set_rx_mask(bp, vnic, 0, NULL);
+	if (rc)
+		return rc;
+
+	PMD_DRV_LOG(DEBUG, "VLAN Strip Offload: %d\n",
+		    !!(rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP));
+
+	return rc;
+}
+
 static int
 bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
 {
 	uint64_t rx_offloads = dev->data->dev_conf.rxmode.offloads;
 	struct bnxt *bp = dev->data->dev_private;
-	unsigned int i;
 	int rc;
 
 	rc = is_bnxt_in_error(bp);
@@ -1913,16 +1972,9 @@ bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
 
 	if (mask & ETH_VLAN_STRIP_MASK) {
 		/* Enable or disable VLAN stripping */
-		for (i = 0; i < bp->nr_vnics; i++) {
-			struct bnxt_vnic_info *vnic = &bp->vnic_info[i];
-			if (rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP)
-				vnic->vlan_strip = true;
-			else
-				vnic->vlan_strip = false;
-			bnxt_hwrm_vnic_cfg(bp, vnic);
-		}
-		PMD_DRV_LOG(DEBUG, "VLAN Strip Offload: %d\n",
-			!!(rx_offloads & DEV_RX_OFFLOAD_VLAN_STRIP));
+		rc = bnxt_config_vlan_hw_stripping(bp, rx_offloads);
+		if (rc)
+			return rc;
 	}
 
 	if (mask & ETH_VLAN_EXTEND_MASK) {
Signature · 1 line
2.21.0 (Apple Git-122.2)
#18 lore raw

PATCH v3 6/10 net/bnxt: handle HW filter setting when port is stopped

In reply to Ajit Khaparde

From: Kalesh AP <[email protected]>

Driver destroy the vnic when the port is brought down. Port hw filter setting such as promiscuos, allmulti and vlan filtering will be applied when port is started.

Fixed to return success silently for these callbacks when port is stopped. Also fixed to clear "bp->dev_stopped" before invoking bnxt_vlan_offload_set_op() in bnxt_dev_start_op().

---
 drivers/net/bnxt/bnxt_ethdev.c | 23 ++++++++++++++++++++++-
 1 file changed, 22 insertions(+), 1 deletion(-)
drivers/net/bnxt/bnxt_ethdev.c +22 -1
index 0b2c29bc8..436ecbb84 100644
@@ -870,6 +870,7 @@ static int bnxt_dev_start_op(struct rte_eth_dev *eth_dev)
 	eth_dev->data->scattered_rx = bnxt_scattered_rx(eth_dev);
 
 	bnxt_link_update(eth_dev, 1, ETH_LINK_UP);
+	bp->dev_stopped = 0;
 
 	if (rx_offloads & DEV_RX_OFFLOAD_VLAN_FILTER)
 		vlan_mask |= ETH_VLAN_FILTER_MASK;
@@ -884,7 +885,6 @@ static int bnxt_dev_start_op(struct rte_eth_dev *eth_dev)
 
 	bp->flags |= BNXT_FLAG_INIT_DONE;
 	eth_dev->data->dev_started = 1;
-	bp->dev_stopped = 0;
 	pthread_mutex_lock(&bp->def_cp_lock);
 	bnxt_schedule_fw_health_check(bp);
 	pthread_mutex_unlock(&bp->def_cp_lock);
@@ -895,6 +895,7 @@ static int bnxt_dev_start_op(struct rte_eth_dev *eth_dev)
 	bnxt_shutdown_nic(bp);
 	bnxt_free_tx_mbufs(bp);
 	bnxt_free_rx_mbufs(bp);
+	bp->dev_stopped = 1;
 	return rc;
 }
 
@@ -1168,6 +1169,10 @@ static int bnxt_promiscuous_enable_op(struct rte_eth_dev *eth_dev)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (bp->vnic_info == NULL)
 		return 0;
 
@@ -1193,6 +1198,10 @@ static int bnxt_promiscuous_disable_op(struct rte_eth_dev *eth_dev)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (bp->vnic_info == NULL)
 		return 0;
 
@@ -1218,6 +1227,10 @@ static int bnxt_allmulticast_enable_op(struct rte_eth_dev *eth_dev)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (bp->vnic_info == NULL)
 		return 0;
 
@@ -1243,6 +1256,10 @@ static int bnxt_allmulticast_disable_op(struct rte_eth_dev *eth_dev)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (bp->vnic_info == NULL)
 		return 0;
 
@@ -1963,6 +1980,10 @@ bnxt_vlan_offload_set_op(struct rte_eth_dev *dev, int mask)
 	if (rc)
 		return rc;
 
+	/* Filter settings will get applied when port is started */
+	if (bp->dev_stopped == 1)
+		return 0;
+
 	if (mask & ETH_VLAN_FILTER_MASK) {
 		/* Enable or disable VLAN filtering */
 		rc = bnxt_config_vlan_hw_filter(bp, rx_offloads);
Signature · 1 line
2.21.0 (Apple Git-122.2)
#19 lore raw

PATCH v3 7/10 net/bnxt: fix a memory leak in port stop

In reply to Ajit Khaparde

From: Kalesh AP <[email protected]>

The memory for mark table is allocated during port start. But the allocated memory is freed only during port close or driver unload which in turn causes a memory leakage on each port start/stop.

Fixed it by moving the memory free to port stop.

Fixes a968a9f5456d ("net/bnxt: add support for flow mark action")
---
 drivers/net/bnxt/bnxt_ethdev.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +4 -4
index 436ecbb84..26614186f 100644
@@ -970,7 +970,10 @@ static void bnxt_dev_stop_op(struct rte_eth_dev *eth_dev)
 	bnxt_int_handler(eth_dev);
 	bnxt_shutdown_nic(bp);
 	bnxt_hwrm_if_change(bp, 0);
-	memset(bp->mark_table, 0, BNXT_MARK_TABLE_SZ);
+
+	rte_free(bp->mark_table);
+	bp->mark_table = NULL;
+
 	bp->flags &= ~BNXT_FLAG_RX_VECTOR_PKT_MODE;
 	bp->dev_stopped = 1;
 	bp->rx_cosq_cnt = 0;
@@ -992,9 +995,6 @@ static void bnxt_dev_close_op(struct rte_eth_dev *eth_dev)
 		bp->grp_info = NULL;
 	}
 
-	rte_free(bp->mark_table);
-	bp->mark_table = NULL;
-
 	bnxt_dev_uninit(eth_dev);
 }
 
Signature · 1 line
2.21.0 (Apple Git-122.2)
#20 lore raw

PATCH v3 8/10 net/bnxt: use macro for PCI log format

In reply to Ajit Khaparde

From: Kalesh AP <[email protected]>

Fixes 19e6af01bb36 ("net/bnxt: support get/set EEPROM")
---
 drivers/net/bnxt/bnxt_ethdev.c | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)
drivers/net/bnxt/bnxt_ethdev.c +11 -11
index 26614186f..ba3f0a7d9 100644
@@ -3593,9 +3593,9 @@ bnxt_get_eeprom_length_op(struct rte_eth_dev *dev)
 	if (rc)
 		return rc;
 
-	PMD_DRV_LOG(INFO, "%04x:%02x:%02x:%02x\n",
-		bp->pdev->addr.domain, bp->pdev->addr.bus,
-		bp->pdev->addr.devid, bp->pdev->addr.function);
+	PMD_DRV_LOG(INFO, PCI_PRI_FMT "\n",
+		    bp->pdev->addr.domain, bp->pdev->addr.bus,
+		    bp->pdev->addr.devid, bp->pdev->addr.function);
 
 	rc = bnxt_hwrm_nvm_get_dir_info(bp, &dir_entries, &entry_length);
 	if (rc != 0)
@@ -3617,10 +3617,10 @@ bnxt_get_eeprom_op(struct rte_eth_dev *dev,
 	if (rc)
 		return rc;
 
-	PMD_DRV_LOG(INFO, "%04x:%02x:%02x:%02x in_eeprom->offset = %d "
-		"len = %d\n", bp->pdev->addr.domain,
-		bp->pdev->addr.bus, bp->pdev->addr.devid,
-		bp->pdev->addr.function, in_eeprom->offset, in_eeprom->length);
+	PMD_DRV_LOG(INFO, PCI_PRI_FMT " in_eeprom->offset = %d len = %d\n",
+		    bp->pdev->addr.domain, bp->pdev->addr.bus,
+		    bp->pdev->addr.devid, bp->pdev->addr.function,
+		    in_eeprom->offset, in_eeprom->length);
 
 	if (in_eeprom->offset == 0) /* special offset value to get directory */
 		return bnxt_get_nvram_directory(bp, in_eeprom->length,
@@ -3693,10 +3693,10 @@ bnxt_set_eeprom_op(struct rte_eth_dev *dev,
 	if (rc)
 		return rc;
 
-	PMD_DRV_LOG(INFO, "%04x:%02x:%02x:%02x in_eeprom->offset = %d "
-		"len = %d\n", bp->pdev->addr.domain,
-		bp->pdev->addr.bus, bp->pdev->addr.devid,
-		bp->pdev->addr.function, in_eeprom->offset, in_eeprom->length);
+	PMD_DRV_LOG(INFO, PCI_PRI_FMT " in_eeprom->offset = %d len = %d\n",
+		    bp->pdev->addr.domain, bp->pdev->addr.bus,
+		    bp->pdev->addr.devid, bp->pdev->addr.function,
+		    in_eeprom->offset, in_eeprom->length);
 
 	if (!BNXT_PF(bp)) {
 		PMD_DRV_LOG(ERR, "NVM write not supported from a VF\n");
Signature · 1 line
2.21.0 (Apple Git-122.2)
#21 lore raw

PATCH v3 9/10 net/bnxt: release port upon close

In reply to Ajit Khaparde

From: Somnath Kotur <[email protected]>

Set RTE_ETH_DEV_CLOSE_REMOVE upon probe so all the private resources for the port can be freed by rte_eth_dev_close().

---
 drivers/net/bnxt/bnxt_ethdev.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/bnxt/bnxt_ethdev.c +5 -0
index ba3f0a7d9..8acfade5f 100644
@@ -4856,6 +4856,11 @@ bnxt_dev_init(struct rte_eth_dev *eth_dev)
 	if (rc)
 		goto error_free;
 
+	/* Pass the information to the rte_eth_dev_close() that it should also
+	 * release the private port resources.
+	 */
+	eth_dev->data->dev_flags |= RTE_ETH_DEV_CLOSE_REMOVE;
+
 	PMD_DRV_LOG(INFO,
 		    DRV_MODULE_NAME "found at mem %" PRIX64 ", node addr %pM\n",
 		    pci_dev->mem_resource[0].phys_addr,
Signature · 1 line
2.21.0 (Apple Git-122.2)
#22 lore raw

Re: [dpdk-dev] [PATCH v3 09/10] net/bnxt: release port upon close

Quoted 17 lines · Ajit Khaparde

From: Somnath Kotur <[email protected]>

Set RTE_ETH_DEV_CLOSE_REMOVE upon probe so all the private resources for the port can be freed by rte_eth_dev_close().

---
 drivers/net/bnxt/bnxt_ethdev.c | 5 +++++
 1 file changed, 5 insertions(+)
drivers/net/bnxt/bnxt_ethdev.c +0 -0
index ba3f0a7d9..8acfade5f 100644

drivers/net/bnxt/bnxt_ethdev.c bnxt_dev_init(struct rte_eth_dev *eth_dev) · line 4863

Quoted 8 lines
drivers/net/bnxt/bnxt_ethdev.c +5 -0
@@ -4856,6 +4856,11 @@ bnxt_dev_init(struct rte_eth_dev *eth_dev)
 	if (rc)
 		goto error_free;
 
+	/* Pass the information to the rte_eth_dev_close() that it should also
+	 * release the private port resources.
+	 */
+	eth_dev->data->dev_flags |= RTE_ETH_DEV_CLOSE_REMOVE;
+

When you give this flag, 'rte_eth_dev_close()' will release port resources, it will memset the 'eth_dev->data'.

In case user first closed the port, later removed it, remove path should be safe for it. I can see 'bnxt_dev_uninit()' operates on the values of 'eth_dev->data' which should crash in this case.

Can you please double check the return path, all following should be safe:
- close the port
- remove the port
- close and remove the port

#23 lore raw

Re: [dpdk-dev] [PATCH v3 09/10] net/bnxt: release port upon close

Ferruh,
Will recheck and revert on

Thanks
Som

Quoted 45 lines · On Tue, 14 Jan 2020, 18:26 Ferruh Yigit, <[email protected]> wrote:
Quoted 8 lines · Ajit Khaparde

From: Somnath Kotur <[email protected]>

Set RTE_ETH_DEV_CLOSE_REMOVE upon probe so all the private resources for the port can be freed by rte_eth_dev_close().

Review

Quoted 6 lines
---
 drivers/net/bnxt/bnxt_ethdev.c | 5 +++++
 1 file changed, 5 insertions(+)
(quoted diff) +0 -0

b/drivers/net/bnxt/bnxt_ethdev.c

drivers/net/bnxt/bnxt_ethdev.c bnxt_dev_init(struct rte_eth_dev *eth_dev) · line 4859

Quoted 8 lines

index ba3f0a7d9..8acfade5f 100644

drivers/net/bnxt/bnxt_ethdev.c +1 -0
@@ -4856,6 +4856,11 @@ bnxt_dev_init(struct rte_eth_dev *eth_dev)
      if (rc)
              goto error_free;
 
+     /* Pass the information to the rte_eth_dev_close() that it should

also

drivers/net/bnxt/bnxt_ethdev.c bnxt_dev_init(struct rte_eth_dev *eth_dev) · line 4863

Quoted 4 lines
drivers/net/bnxt/bnxt_ethdev.c +4 -0
+      * release the private port resources.
+      */
+     eth_dev->data->dev_flags |= RTE_ETH_DEV_CLOSE_REMOVE;
+

When you give this flag, 'rte_eth_dev_close()' will release port resources, it
will memset the 'eth_dev->data'.

In case user first closed the port, later removed it, remove path should be safe
for it. I can see 'bnxt_dev_uninit()' operates on the values of 'eth_dev->data'
which should crash in this case.

Can you please double check the return path, all following should be safe:
- close the port
- remove the port
- close and remove the port

#24 lore raw

Re: [dpdk-dev] [PATCH v3 09/10] net/bnxt: release port upon close

Quoted 2 lines · Somnath Kotur

Ferruh,
Will recheck and revert on

Hi Ajit,

this is blocking the next-net-brcm tree to be pulled. Would you want to drop this patch from your tree and continue with merge and deal with this patch later? Or if this patch is the critical for the series I can keep waiting.

Thanks,
ferruh

Quoted 50 lines

Thanks
Som

On Tue, 14 Jan 2020, 18:26 Ferruh Yigit, <[email protected] <mailto:[email protected]>> wrote:

    On 1/14/2020 5:14 AM, Ajit Khaparde wrote:
    > From: Somnath Kotur <[email protected]
    <mailto:[email protected]>>
    >
    > Set RTE_ETH_DEV_CLOSE_REMOVE upon probe so all the private resources
    > for the port can be freed by rte_eth_dev_close().
    >
    > Signed-off-by: Somnath Kotur <[email protected]
    <mailto:[email protected]>>
    > Reviewed-by: Kalesh Anakkur Purayil <[email protected]
    <mailto:[email protected]>>
    > Reviewed-by: Ajit Khaparde <[email protected]
    <mailto:[email protected]>>
    > ---
    >  drivers/net/bnxt/bnxt_ethdev.c | 5 +++++
    >  1 file changed, 5 insertions(+)
    >
    > diff --git a/drivers/net/bnxt/bnxt_ethdev.c b/drivers/net/bnxt/bnxt_ethdev.c
    > index ba3f0a7d9..8acfade5f 100644
    > --- a/drivers/net/bnxt/bnxt_ethdev.c
    > +++ b/drivers/net/bnxt/bnxt_ethdev.c
    > @@ -4856,6 +4856,11 @@ bnxt_dev_init(struct rte_eth_dev *eth_dev)
    >       if (rc)
    >               goto error_free;
    > 
    > +     /* Pass the information to the rte_eth_dev_close() that it should also
    > +      * release the private port resources.
    > +      */
    > +     eth_dev->data->dev_flags |= RTE_ETH_DEV_CLOSE_REMOVE;
    > +

    When you give this flag, 'rte_eth_dev_close()' will release port resources, it
    will memset the 'eth_dev->data'.

    In case user first closed the port, later removed it, remove path should be safe
    for it. I can see 'bnxt_dev_uninit()' operates on the values of 'eth_dev->data'
    which should crash in this case.

    Can you please double check the return path, all following should be safe:
    - close the port
    - remove the port
    - close and remove the port
#25 lore raw

Re: [dpdk-dev] [PATCH v3 09/10] net/bnxt: release port upon close

Ferruh,

Quoted 4 lines · On Thu, 16 Jan 2020, 16:50 Ferruh Yigit, <[email protected]> wrote:
Quoted 2 lines · Somnath Kotur

Ferruh,
Will recheck and revert on

Quoted 8 lines

Hi Ajit,

this is blocking the next-net-brcm tree to be pulled. Would you want to drop
this patch from your tree and continue with merge and deal with this patch later?
Or if this patch is the critical for the series I can keep waiting.

Kalesh will be sending v4 of the this patch set addressing this patch as well shortly ..few hours full now

Thanks
Som

Quoted 64 lines

Thanks,
ferruh

Quoted 4 lines

Thanks
Som

Quoted 8 lines

On Tue, 14 Jan 2020, 18:26 Ferruh Yigit, <[email protected] <mailto:[email protected]>> wrote:

On 1/14/2020 5:14 AM, Ajit Khaparde wrote:
> From: Somnath Kotur <[email protected] <mailto:[email protected]>>
>
> Set RTE_ETH_DEV_CLOSE_REMOVE upon probe so all the private

resources

Quoted 5 lines
Quoted 3 lines

for the port can be freed by rte_eth_dev_close().

<mailto:[email protected]>>
> Reviewed-by: Kalesh Anakkur Purayil <

[email protected]

Quoted 8 lines
    <mailto:[email protected]>>
    > Reviewed-by: Ajit Khaparde <[email protected]
    <mailto:[email protected]>>
    > ---
    >  drivers/net/bnxt/bnxt_ethdev.c | 5 +++++
    >  1 file changed, 5 insertions(+)
    >
    > diff --git a/drivers/net/bnxt/bnxt_ethdev.c

b/drivers/net/bnxt/bnxt_ethdev.c

Quoted 8 lines

drivers/net/bnxt/bnxt_ethdev.c bnxt_dev_init(struct rte_eth_dev *eth_dev) · line 4859

Quoted 8 lines

index ba3f0a7d9..8acfade5f 100644

drivers/net/bnxt/bnxt_ethdev.c +1 -0
@@ -4856,6 +4856,11 @@ bnxt_dev_init(struct rte_eth_dev *eth_dev)
      if (rc)
              goto error_free;
 
+     /* Pass the information to the rte_eth_dev_close() that it

should also

Quoted 6 lines

drivers/net/bnxt/bnxt_ethdev.c bnxt_dev_init(struct rte_eth_dev *eth_dev) · line 4863

Quoted 4 lines
drivers/net/bnxt/bnxt_ethdev.c +4 -0
+      * release the private port resources.
+      */
+     eth_dev->data->dev_flags |= RTE_ETH_DEV_CLOSE_REMOVE;
+

When you give this flag, 'rte_eth_dev_close()' will release port

resources, it

Quoted 3 lines

will memset the 'eth_dev->data'.

In case user first closed the port, later removed it, remove path

should be safe

Quoted 1 line

for it. I can see 'bnxt_dev_uninit()' operates on the values of

'eth_dev->data'

Quoted 3 lines

which should crash in this case.

Can you please double check the return path, all following should be

safe:

Quoted 4 lines

- close the port
- remove the port
- close and remove the port

#26 lore raw

PATCH v3 10/10 net/bnxt: fix calculation of max rings

In reply to Ajit Khaparde

From: Somnath Kotur <[email protected]>

Max Tx rings count could be lesser than max Rx rings in some cases, so take this into account as well.

Account for stat contexts available(one for each ring) along with
no: of completion rings(one for each ring) to cap the max no: of Tx /Rx rings that can be possibly created.

Fixes f03e66cb ("net/bnxt: limit queue count for NS3/Stingray devices")
---
 drivers/net/bnxt/bnxt.h | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)
drivers/net/bnxt/bnxt.h +10 -3
index 3487b917e..ddb26814c 100644
@@ -622,12 +622,19 @@ struct bnxt {
 	uint16_t		max_tx_rings;
 	uint16_t		max_rx_rings;
 #define MAX_STINGRAY_RINGS		128U
-#define BNXT_MAX_RINGS(bp) \
+/* For sake of symmetry, max Tx rings == max Rx rings, one stat ctx for each */
+#define BNXT_MAX_RX_RINGS(bp) \
 	(BNXT_STINGRAY(bp) ? RTE_MIN(RTE_MIN(bp->max_rx_rings, \
 					     MAX_STINGRAY_RINGS), \
-				     bp->max_stat_ctx) : \
-				RTE_MIN(bp->max_rx_rings, bp->max_stat_ctx))
+				     bp->max_stat_ctx / 2U) : \
+				RTE_MIN(bp->max_rx_rings, \
+					bp->max_stat_ctx / 2U))
+#define BNXT_MAX_TX_RINGS(bp) \
+	(RTE_MIN((bp)->max_tx_rings, BNXT_MAX_RX_RINGS(bp)))
 
+#define BNXT_MAX_RINGS(bp) \
+	(RTE_MIN((((bp)->max_cp_rings - BNXT_NUM_ASYNC_CPR(bp)) / 2U), \
+		 BNXT_MAX_TX_RINGS(bp)))
 	uint16_t		max_nq_rings;
 	uint16_t		max_l2_ctx;
 	uint16_t		max_rx_em_flows;
Signature · 1 line
2.21.0 (Apple Git-122.2)
#27 lore raw
In reply to Ajit Khaparde
Quoted 5 lines · Ajit Khaparde

v1->v2: fixed the correct the commit ids in Fixes tag. v2->v3: shorten commit log and update commit ids for Fixes tag.

Note that Fixes tag in patch 07/10 is correct.

Patches applied to dpdk-next-net-brcm. Thanks

Quoted 29 lines

Kalesh AP (8):
net/bnxt: handle flow create failure
net/bnxt: fix probe failure in FreeBSD net/bnxt: fix to use correct IOVA mapping net/bnxt: fix enable/disable VLAN filtering net/bnxt: fix VLAN strip support
net/bnxt: handle HW filter setting when port is stopped net/bnxt: fix a memory leak in port stop net/bnxt: use macro for PCI log format

Somnath Kotur (2):
net/bnxt: release port upon close
net/bnxt: fix calculation of max rings

 drivers/net/bnxt/bnxt.h        |  21 +-
 drivers/net/bnxt/bnxt_ethdev.c | 422 ++++++++++++++++++---------------
 drivers/net/bnxt/bnxt_filter.c |   5 +-
 drivers/net/bnxt/bnxt_flow.c   |   4 +-
 drivers/net/bnxt/bnxt_hwrm.c   |  28 +--
 drivers/net/bnxt/bnxt_irq.c    |   8 +
 drivers/net/bnxt/bnxt_ring.c   |  17 --
 drivers/net/bnxt/bnxt_vnic.c   |  11 -
 8 files changed, 273 insertions(+), 243 deletions(-)

--
2.21.0 (Apple Git-122.2)

#28 lore raw
Quoted 11 lines · Ajit Khaparde

On Mon, Jan 13, 2020 at 9:16 PM Ajit Khaparde <[email protected] <mailto:[email protected]>> wrote:

    v1->v2: fixed the correct the commit ids in Fixes tag.
    v2->v3: shorten commit log and update commit ids for Fixes tag.

    Note that Fixes tag in patch 07/10 is correct.

Patches applied to dpdk-next-net-brcm. Thanks

Hi Ajit,

There are some patches with fixes lines but not requested the backport [1], are they explicitly not wanted to be backported or missed the tag?
To prevent this questions, for the cases the commit explicitly not wanted to be backported can you please put the reasoning for it into the commit log as comment.

Thanks,
ferruh

[1]
Is it candidate for Cc: [email protected] backport?
net/bnxt: fix probe failure in FreeBSD
net/bnxt: fix VLAN strip support
net/bnxt: use macro for PCI log format
net/bnxt: fix calculation of max rings

Quoted 30 lines

Kalesh AP (8):
  net/bnxt: handle flow create failure net/bnxt: fix probe failure in FreeBSD net/bnxt: fix to use correct IOVA mapping net/bnxt: fix enable/disable VLAN filtering net/bnxt: fix VLAN strip support
  net/bnxt: handle HW filter setting when port is stopped net/bnxt: fix a memory leak in port stop net/bnxt: use macro for PCI log format

Somnath Kotur (2):
  net/bnxt: release port upon close
  net/bnxt: fix calculation of max rings

 drivers/net/bnxt/bnxt.h        |  21 +- drivers/net/bnxt/bnxt_ethdev.c | 422 ++++++++++++++++++--------------- drivers/net/bnxt/bnxt_filter.c |   5 +- drivers/net/bnxt/bnxt_flow.c   |   4 +- drivers/net/bnxt/bnxt_hwrm.c   |  28 +-- drivers/net/bnxt/bnxt_irq.c    |   8 + drivers/net/bnxt/bnxt_ring.c   |  17 -- drivers/net/bnxt/bnxt_vnic.c   |  11 -
 8 files changed, 273 insertions(+), 243 deletions(-)

--
2.21.0 (Apple Git-122.2)