[PATCH net-next v2 0/5] nfp: support FEC mode reporting and auto-neg

[PATCH net-next v2 0/5] nfp: support FEC mode reporting and auto-neg

[Simon Horman]

Hi,

this series adds support for the following features to the nfp driver:

* Patch 1/5: Support active FEC mode
* Patch 2/5: Don't halt driver on non-fatal error when interacting with fw
* Patch 3/5: Treat port independence as a firmware rather than port property
* Patch 4/5: Support link auto negotiation
* Patch 5/5: Support restart of link auto negotiation

Key changes since v1:
* Treat port independence as a firmware rather than port property
* remove the enforcement of FEC auto mode when link auto-neg is enabled
* Adjust the link reset function so that it can take effect
  in port force-up scenario

Fei Qin (1):
  nfp: add support restart of link auto-negotiation

Yinjun Zhang (4):
  nfp: add support for reporting active FEC mode
  nfp: avoid halt of driver init process when non-fatal error happens
  nfp: refine the ABI of getting `sp_indiff` info
  nfp: add support for link auto negotiation

 drivers/net/ethernet/netronome/nfp/nfp_main.c | 74 ++++++++++++++++++-
 drivers/net/ethernet/netronome/nfp/nfp_main.h |  3 +-
 .../net/ethernet/netronome/nfp/nfp_net_ctrl.c |  8 --
 .../net/ethernet/netronome/nfp/nfp_net_ctrl.h | 10 +--
 .../ethernet/netronome/nfp/nfp_net_ethtool.c  | 61 +++++++++++++--
 .../net/ethernet/netronome/nfp/nfp_net_main.c | 49 +-----------
 .../ethernet/netronome/nfp/nfpcore/nfp_nsp.h  |  3 +
 .../netronome/nfp/nfpcore/nfp_nsp_eth.c       | 11 ++-
 8 files changed, 147 insertions(+), 72 deletions(-)

-- 
2.30.2

[PATCH net-next v2 1/5] nfp: add support for reporting active FEC mode

[Simon Horman]

From: Yinjun Zhang <yinjun.zhang@corigine.com>

The latest management firmware can now report the active FEC
mode. Adapt driver accordingly so that user can get the active
FEC mode by running command:

  # ethtool --show-fec <intf>

Also correct use of `fec` field.

Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c     | 2 +-
 drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h     | 2 ++
 drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c | 9 ++++++++-
 3 files changed, 11 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
index db58532364b6..d50af23642a2 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
@@ -996,7 +996,7 @@ nfp_port_get_fecparam(struct net_device *netdev,
 		return 0;
 
 	param->fec = nfp_port_fec_nsp_to_ethtool(eth_port->fec_modes_supported);
-	param->active_fec = nfp_port_fec_nsp_to_ethtool(eth_port->fec);
+	param->active_fec = nfp_port_fec_nsp_to_ethtool(BIT(eth_port->act_fec));
 
 	return 0;
 }
diff --git a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h
index 77d66855be42..52465670a01e 100644
--- a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h
+++ b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h
@@ -132,6 +132,7 @@ enum nfp_eth_fec {
  * @ports.interface:	interface (module) plugged in
  * @ports.media:	media type of the @interface
  * @ports.fec:		forward error correction mode
+ * @ports.act_fec:	active forward error correction mode
  * @ports.aneg:		auto negotiation mode
  * @ports.mac_addr:	interface MAC address
  * @ports.label_port:	port id
@@ -162,6 +163,7 @@ struct nfp_eth_table {
 		enum nfp_eth_media media;
 
 		enum nfp_eth_fec fec;
+		enum nfp_eth_fec act_fec;
 		enum nfp_eth_aneg aneg;
 
 		u8 mac_addr[ETH_ALEN];
diff --git a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
index 4cc38799eabc..18ba7629cdc2 100644
--- a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
+++ b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
@@ -40,6 +40,7 @@
 #define NSP_ETH_STATE_OVRD_CHNG		BIT_ULL(22)
 #define NSP_ETH_STATE_ANEG		GENMASK_ULL(25, 23)
 #define NSP_ETH_STATE_FEC		GENMASK_ULL(27, 26)
+#define NSP_ETH_STATE_ACT_FEC		GENMASK_ULL(29, 28)
 
 #define NSP_ETH_CTRL_CONFIGURED		BIT_ULL(0)
 #define NSP_ETH_CTRL_ENABLED		BIT_ULL(1)
@@ -170,7 +171,13 @@ nfp_eth_port_translate(struct nfp_nsp *nsp, const union eth_table_entry *src,
 	if (dst->fec_modes_supported)
 		dst->fec_modes_supported |= NFP_FEC_AUTO | NFP_FEC_DISABLED;
 
-	dst->fec = 1 << FIELD_GET(NSP_ETH_STATE_FEC, state);
+	dst->fec = FIELD_GET(NSP_ETH_STATE_FEC, state);
+	dst->act_fec = dst->fec;
+
+	if (nfp_nsp_get_abi_ver_minor(nsp) < 33)
+		return;
+
+	dst->act_fec = FIELD_GET(NSP_ETH_STATE_ACT_FEC, state);
 }
 
 static void
-- 
2.30.2

[PATCH net-next v2 2/5] nfp: avoid halt of driver init process when non-fatal error happens

[Simon Horman]

From: Yinjun Zhang <yinjun.zhang@corigine.com>

It's not a fatal error when setting `hwinfo` into management firmware
fails, no need to halt the whole driver initialization process.

Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 drivers/net/ethernet/netronome/nfp/nfp_net_main.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_main.c b/drivers/net/ethernet/netronome/nfp/nfp_net_main.c
index e2d4c487e8de..f3852ba8099a 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_main.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_main.c
@@ -315,18 +315,17 @@ static int nfp_net_pf_cfg_nsp(struct nfp_pf *pf, bool sp_indiff)
 	int err;
 
 	nsp = nfp_nsp_open(pf->cpp);
-	if (IS_ERR(nsp)) {
-		err = PTR_ERR(nsp);
-		return err;
-	}
+	if (IS_ERR(nsp))
+		return PTR_ERR(nsp);
 
 	snprintf(hwinfo, sizeof(hwinfo), "sp_indiff=%d", sp_indiff);
 	err = nfp_nsp_hwinfo_set(nsp, hwinfo, sizeof(hwinfo));
+	/* Not a fatal error, no need to return error to stop driver from loading */
 	if (err)
 		nfp_warn(pf->cpp, "HWinfo(sp_indiff=%d) set failed: %d\n", sp_indiff, err);
 
 	nfp_nsp_close(nsp);
-	return err;
+	return 0;
 }
 
 static int nfp_net_pf_init_nsp(struct nfp_pf *pf)
-- 
2.30.2

[PATCH net-next v2 3/5] nfp: refine the ABI of getting `sp_indiff` info

[Simon Horman]

From: Yinjun Zhang <yinjun.zhang@corigine.com>

Considering that whether application firmware is indifferent to
port speed is a firmware property instead of port property, now use
a new rtsym to get the property instead of parsing per-port tlv caps.
With this change, relevant code is moved to `nfp_main` layer.

Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 drivers/net/ethernet/netronome/nfp/nfp_main.c | 67 ++++++++++++++++++-
 drivers/net/ethernet/netronome/nfp/nfp_main.h |  3 +-
 .../net/ethernet/netronome/nfp/nfp_net_ctrl.c |  8 ---
 .../net/ethernet/netronome/nfp/nfp_net_ctrl.h | 10 +--
 .../net/ethernet/netronome/nfp/nfp_net_main.c | 48 +------------
 5 files changed, 71 insertions(+), 65 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/nfp_main.c b/drivers/net/ethernet/netronome/nfp/nfp_main.c
index 873429f7a6da..91063f19c97d 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_main.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_main.c
@@ -691,6 +691,64 @@ static int nfp_pf_find_rtsyms(struct nfp_pf *pf)
 	return 0;
 }
 
+int nfp_net_pf_get_app_id(struct nfp_pf *pf)
+{
+	return nfp_pf_rtsym_read_optional(pf, "_pf%u_net_app_id",
+					  NFP_APP_CORE_NIC);
+}
+
+static u64 nfp_net_pf_get_app_cap(struct nfp_pf *pf)
+{
+	char name[32];
+	int err = 0;
+	u64 val;
+
+	snprintf(name, sizeof(name), "_pf%u_net_app_cap", nfp_cppcore_pcie_unit(pf->cpp));
+
+	val = nfp_rtsym_read_le(pf->rtbl, name, &err);
+	if (err) {
+		if (err != -ENOENT)
+			nfp_err(pf->cpp, "Unable to read symbol %s\n", name);
+
+		return 0;
+	}
+
+	return val;
+}
+
+static int nfp_pf_cfg_hwinfo(struct nfp_pf *pf, bool sp_indiff)
+{
+	struct nfp_nsp *nsp;
+	char hwinfo[32];
+	int err;
+
+	nsp = nfp_nsp_open(pf->cpp);
+	if (IS_ERR(nsp))
+		return PTR_ERR(nsp);
+
+	snprintf(hwinfo, sizeof(hwinfo), "sp_indiff=%d", sp_indiff);
+	err = nfp_nsp_hwinfo_set(nsp, hwinfo, sizeof(hwinfo));
+	/* Not a fatal error, no need to return error to stop driver from loading */
+	if (err)
+		nfp_warn(pf->cpp, "HWinfo(sp_indiff=%d) set failed: %d\n", sp_indiff, err);
+
+	nfp_nsp_close(nsp);
+	return 0;
+}
+
+static int nfp_pf_nsp_cfg(struct nfp_pf *pf)
+{
+	bool sp_indiff = (nfp_net_pf_get_app_id(pf) == NFP_APP_FLOWER_NIC) ||
+			 (nfp_net_pf_get_app_cap(pf) & NFP_NET_APP_CAP_SP_INDIFF);
+
+	return nfp_pf_cfg_hwinfo(pf, sp_indiff);
+}
+
+static void nfp_pf_nsp_clean(struct nfp_pf *pf)
+{
+	nfp_pf_cfg_hwinfo(pf, false);
+}
+
 static int nfp_pci_probe(struct pci_dev *pdev,
 			 const struct pci_device_id *pci_id)
 {
@@ -791,10 +849,14 @@ static int nfp_pci_probe(struct pci_dev *pdev,
 		goto err_fw_unload;
 	}
 
-	err = nfp_net_pci_probe(pf);
+	err = nfp_pf_nsp_cfg(pf);
 	if (err)
 		goto err_fw_unload;
 
+	err = nfp_net_pci_probe(pf);
+	if (err)
+		goto err_nsp_clean;
+
 	err = nfp_hwmon_register(pf);
 	if (err) {
 		dev_err(&pdev->dev, "Failed to register hwmon info\n");
@@ -805,6 +867,8 @@ static int nfp_pci_probe(struct pci_dev *pdev,
 
 err_net_remove:
 	nfp_net_pci_remove(pf);
+err_nsp_clean:
+	nfp_pf_nsp_clean(pf);
 err_fw_unload:
 	kfree(pf->rtbl);
 	nfp_mip_close(pf->mip);
@@ -844,6 +908,7 @@ static void __nfp_pci_shutdown(struct pci_dev *pdev, bool unload_fw)
 
 	nfp_net_pci_remove(pf);
 
+	nfp_pf_nsp_clean(pf);
 	vfree(pf->dumpspec);
 	kfree(pf->rtbl);
 	nfp_mip_close(pf->mip);
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_main.h b/drivers/net/ethernet/netronome/nfp/nfp_main.h
index 6805af186f1b..afd3edfa2428 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_main.h
+++ b/drivers/net/ethernet/netronome/nfp/nfp_main.h
@@ -65,7 +65,6 @@ struct nfp_dumpspec {
  * @num_vfs:		Number of SR-IOV VFs enabled
  * @fw_loaded:		Is the firmware loaded?
  * @unload_fw_on_remove:Do we need to unload firmware on driver removal?
- * @sp_indiff:		Is the firmware indifferent to physical port speed?
  * @ctrl_vnic:		Pointer to the control vNIC if available
  * @mip:		MIP handle
  * @rtbl:		RTsym table
@@ -115,7 +114,6 @@ struct nfp_pf {
 
 	bool fw_loaded;
 	bool unload_fw_on_remove;
-	bool sp_indiff;
 
 	struct nfp_net *ctrl_vnic;
 
@@ -163,6 +161,7 @@ bool nfp_ctrl_tx(struct nfp_net *nn, struct sk_buff *skb);
 
 int nfp_pf_rtsym_read_optional(struct nfp_pf *pf, const char *format,
 			       unsigned int default_val);
+int nfp_net_pf_get_app_id(struct nfp_pf *pf);
 u8 __iomem *
 nfp_pf_map_rtsym(struct nfp_pf *pf, const char *name, const char *sym_fmt,
 		 unsigned int min_size, struct nfp_cpp_area **area);
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.c
index d81bd8697047..c3a763134e79 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.c
@@ -148,14 +148,6 @@ int nfp_net_tlv_caps_parse(struct device *dev, u8 __iomem *ctrl_mem,
 							  true))
 				return -EINVAL;
 			break;
-		case NFP_NET_CFG_TLV_TYPE_SP_INDIFF:
-			if (length) {
-				dev_err(dev, "Unexpected len of SP_INDIFF TLV:%u\n", length);
-				return -EINVAL;
-			}
-
-			caps->sp_indiff = true;
-			break;
 		default:
 			if (!FIELD_GET(NFP_NET_CFG_TLV_HEADER_REQUIRED, hdr))
 				break;
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.h b/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.h
index 1d53f721a1c8..6714d5e8fdab 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.h
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ctrl.h
@@ -14,6 +14,9 @@
 
 #include <linux/types.h>
 
+/* 64-bit per app capabilities */
+#define NFP_NET_APP_CAP_SP_INDIFF	BIT_ULL(0) /* indifferent to port speed */
+
 /* Configuration BAR size.
  *
  * The configuration BAR is 8K in size, but due to
@@ -492,10 +495,6 @@
  * %NFP_NET_CFG_TLV_TYPE_CRYPTO_OPS_RX_SCAN:
  * Same as %NFP_NET_CFG_TLV_TYPE_CRYPTO_OPS, but crypto TLS does stream scan
  * RX sync, rather than kernel-assisted sync.
- *
- * %NFP_NET_CFG_TLV_TYPE_SP_INDIFF:
- * Empty, indicate the firmware is indifferent to port speed. Then no need to
- * reload driver and firmware when port speed is changed.
  */
 #define NFP_NET_CFG_TLV_TYPE_UNKNOWN		0
 #define NFP_NET_CFG_TLV_TYPE_RESERVED		1
@@ -509,7 +508,6 @@
 #define NFP_NET_CFG_TLV_TYPE_CRYPTO_OPS		11 /* see crypto/fw.h */
 #define NFP_NET_CFG_TLV_TYPE_VNIC_STATS		12
 #define NFP_NET_CFG_TLV_TYPE_CRYPTO_OPS_RX_SCAN	13
-#define NFP_NET_CFG_TLV_TYPE_SP_INDIFF		14
 
 struct device;
 
@@ -524,7 +522,6 @@ struct device;
  * @vnic_stats_off:	offset of vNIC stats area
  * @vnic_stats_cnt:	number of vNIC stats
  * @tls_resync_ss:	TLS resync will be performed via stream scan
- * @sp_indiff:		Firmware is indifferent to port speed
  */
 struct nfp_net_tlv_caps {
 	u32 me_freq_mhz;
@@ -537,7 +534,6 @@ struct nfp_net_tlv_caps {
 	unsigned int vnic_stats_off;
 	unsigned int vnic_stats_cnt;
 	unsigned int tls_resync_ss:1;
-	unsigned int sp_indiff:1;
 };
 
 int nfp_net_tlv_caps_parse(struct device *dev, u8 __iomem *ctrl_mem,
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_main.c b/drivers/net/ethernet/netronome/nfp/nfp_net_main.c
index f3852ba8099a..3bae92dc899e 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_main.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_main.c
@@ -77,12 +77,6 @@ static int nfp_net_pf_get_num_ports(struct nfp_pf *pf)
 	return nfp_pf_rtsym_read_optional(pf, "nfd_cfg_pf%u_num_ports", 1);
 }
 
-static int nfp_net_pf_get_app_id(struct nfp_pf *pf)
-{
-	return nfp_pf_rtsym_read_optional(pf, "_pf%u_net_app_id",
-					  NFP_APP_CORE_NIC);
-}
-
 static void nfp_net_pf_free_vnic(struct nfp_pf *pf, struct nfp_net *nn)
 {
 	if (nfp_net_is_data_vnic(nn))
@@ -206,7 +200,6 @@ nfp_net_pf_alloc_vnics(struct nfp_pf *pf, void __iomem *ctrl_bar,
 			nn->port->link_cb = nfp_net_refresh_port_table;
 
 		ctrl_bar += NFP_PF_CSR_SLICE_SIZE;
-		pf->sp_indiff |= nn->tlv_caps.sp_indiff;
 
 		/* Kill the vNIC if app init marked it as invalid */
 		if (nn->port && nn->port->type == NFP_PORT_INVALID)
@@ -308,36 +301,6 @@ static int nfp_net_pf_init_vnics(struct nfp_pf *pf)
 	return err;
 }
 
-static int nfp_net_pf_cfg_nsp(struct nfp_pf *pf, bool sp_indiff)
-{
-	struct nfp_nsp *nsp;
-	char hwinfo[32];
-	int err;
-
-	nsp = nfp_nsp_open(pf->cpp);
-	if (IS_ERR(nsp))
-		return PTR_ERR(nsp);
-
-	snprintf(hwinfo, sizeof(hwinfo), "sp_indiff=%d", sp_indiff);
-	err = nfp_nsp_hwinfo_set(nsp, hwinfo, sizeof(hwinfo));
-	/* Not a fatal error, no need to return error to stop driver from loading */
-	if (err)
-		nfp_warn(pf->cpp, "HWinfo(sp_indiff=%d) set failed: %d\n", sp_indiff, err);
-
-	nfp_nsp_close(nsp);
-	return 0;
-}
-
-static int nfp_net_pf_init_nsp(struct nfp_pf *pf)
-{
-	return nfp_net_pf_cfg_nsp(pf, pf->sp_indiff);
-}
-
-static void nfp_net_pf_clean_nsp(struct nfp_pf *pf)
-{
-	(void)nfp_net_pf_cfg_nsp(pf, false);
-}
-
 static int
 nfp_net_pf_app_init(struct nfp_pf *pf, u8 __iomem *qc_bar, unsigned int stride)
 {
@@ -349,8 +312,6 @@ nfp_net_pf_app_init(struct nfp_pf *pf, u8 __iomem *qc_bar, unsigned int stride)
 	if (IS_ERR(pf->app))
 		return PTR_ERR(pf->app);
 
-	pf->sp_indiff |= pf->app->type->id == NFP_APP_FLOWER_NIC;
-
 	devl_lock(devlink);
 	err = nfp_app_init(pf->app);
 	devl_unlock(devlink);
@@ -813,13 +774,9 @@ int nfp_net_pci_probe(struct nfp_pf *pf)
 	if (err)
 		goto err_clean_ddir;
 
-	err = nfp_net_pf_init_nsp(pf);
-	if (err)
-		goto err_free_vnics;
-
 	err = nfp_net_pf_alloc_irqs(pf);
 	if (err)
-		goto err_clean_nsp;
+		goto err_free_vnics;
 
 	err = nfp_net_pf_app_start(pf);
 	if (err)
@@ -838,8 +795,6 @@ int nfp_net_pci_probe(struct nfp_pf *pf)
 	nfp_net_pf_app_stop(pf);
 err_free_irqs:
 	nfp_net_pf_free_irqs(pf);
-err_clean_nsp:
-	nfp_net_pf_clean_nsp(pf);
 err_free_vnics:
 	nfp_net_pf_free_vnics(pf);
 err_clean_ddir:
@@ -870,7 +825,6 @@ void nfp_net_pci_remove(struct nfp_pf *pf)
 		nfp_net_pf_free_vnic(pf, nn);
 	}
 
-	nfp_net_pf_clean_nsp(pf);
 	nfp_net_pf_app_stop(pf);
 	/* stop app first, to avoid double free of ctrl vNIC's ddir */
 	nfp_net_debugfs_dir_clean(&pf->ddir);
-- 
2.30.2

[PATCH net-next v2 4/5] nfp: add support for link auto negotiation

[Simon Horman]

From: Yinjun Zhang <yinjun.zhang@corigine.com>

Report the auto negotiation capability if it's supported
in management firmware, and advertise it if it's enabled.
Changing port speed is not allowed when autoneg is enabled.

The ethtool <intf> command displays the auto-neg capability:

  # ethtool enp1s0np0
  Settings for enp1s0np0:
          Supported ports: [ FIBRE ]
          Supported link modes:   Not reported
          Supported pause frame use: Symmetric
          Supports auto-negotiation: Yes
          Supported FEC modes: None        RS      BASER
          Advertised link modes:  Not reported
          Advertised pause frame use: Symmetric
          Advertised auto-negotiation: Yes
          Advertised FEC modes: None       RS      BASER
          Speed: 25000Mb/s
          Duplex: Full
          Auto-negotiation: on
          Port: FIBRE
          PHYAD: 0
          Transceiver: internal
          Link detected: yes

Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 drivers/net/ethernet/netronome/nfp/nfp_main.c |  9 ++++++-
 .../ethernet/netronome/nfp/nfp_net_ethtool.c  | 26 ++++++++++++++++---
 .../ethernet/netronome/nfp/nfpcore/nfp_nsp.h  |  1 +
 .../netronome/nfp/nfpcore/nfp_nsp_eth.c       |  2 ++
 4 files changed, 33 insertions(+), 5 deletions(-)

diff --git a/drivers/net/ethernet/netronome/nfp/nfp_main.c b/drivers/net/ethernet/netronome/nfp/nfp_main.c
index 91063f19c97d..e66e548919d4 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_main.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_main.c
@@ -729,8 +729,15 @@ static int nfp_pf_cfg_hwinfo(struct nfp_pf *pf, bool sp_indiff)
 	snprintf(hwinfo, sizeof(hwinfo), "sp_indiff=%d", sp_indiff);
 	err = nfp_nsp_hwinfo_set(nsp, hwinfo, sizeof(hwinfo));
 	/* Not a fatal error, no need to return error to stop driver from loading */
-	if (err)
+	if (err) {
 		nfp_warn(pf->cpp, "HWinfo(sp_indiff=%d) set failed: %d\n", sp_indiff, err);
+	} else {
+		/* Need reinit eth_tbl since the eth table state may change
+		 * after sp_indiff is configured.
+		 */
+		kfree(pf->eth_tbl);
+		pf->eth_tbl = __nfp_eth_read_ports(pf->cpp, nsp);
+	}
 
 	nfp_nsp_close(nsp);
 	return 0;
diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
index d50af23642a2..678cea0fd274 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
@@ -290,8 +290,13 @@ nfp_net_get_link_ksettings(struct net_device *netdev,
 	if (eth_port) {
 		ethtool_link_ksettings_add_link_mode(cmd, supported, Pause);
 		ethtool_link_ksettings_add_link_mode(cmd, advertising, Pause);
-		cmd->base.autoneg = eth_port->aneg != NFP_ANEG_DISABLED ?
-			AUTONEG_ENABLE : AUTONEG_DISABLE;
+		if (eth_port->supp_aneg) {
+			ethtool_link_ksettings_add_link_mode(cmd, supported, Autoneg);
+			if (eth_port->aneg == NFP_ANEG_AUTO) {
+				ethtool_link_ksettings_add_link_mode(cmd, advertising, Autoneg);
+				cmd->base.autoneg = AUTONEG_ENABLE;
+			}
+		}
 		nfp_net_set_fec_link_mode(eth_port, cmd);
 	}
 
@@ -327,6 +332,7 @@ static int
 nfp_net_set_link_ksettings(struct net_device *netdev,
 			   const struct ethtool_link_ksettings *cmd)
 {
+	bool req_aneg = (cmd->base.autoneg == AUTONEG_ENABLE);
 	struct nfp_eth_table_port *eth_port;
 	struct nfp_port *port;
 	struct nfp_nsp *nsp;
@@ -346,13 +352,25 @@ nfp_net_set_link_ksettings(struct net_device *netdev,
 	if (IS_ERR(nsp))
 		return PTR_ERR(nsp);
 
-	err = __nfp_eth_set_aneg(nsp, cmd->base.autoneg == AUTONEG_ENABLE ?
-				 NFP_ANEG_AUTO : NFP_ANEG_DISABLED);
+	if (req_aneg && !eth_port->supp_aneg) {
+		netdev_warn(netdev, "Autoneg is not supported.\n");
+		err = -EOPNOTSUPP;
+		goto err_bad_set;
+	}
+
+	err = __nfp_eth_set_aneg(nsp, req_aneg ? NFP_ANEG_AUTO : NFP_ANEG_DISABLED);
 	if (err)
 		goto err_bad_set;
+
 	if (cmd->base.speed != SPEED_UNKNOWN) {
 		u32 speed = cmd->base.speed / eth_port->lanes;
 
+		if (req_aneg) {
+			netdev_err(netdev, "Speed changing is not allowed when working on autoneg mode.\n");
+			err = -EINVAL;
+			goto err_bad_set;
+		}
+
 		err = __nfp_eth_set_speed(nsp, speed);
 		if (err)
 			goto err_bad_set;
diff --git a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h
index 52465670a01e..992d72ac98d3 100644
--- a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h
+++ b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp.h
@@ -174,6 +174,7 @@ struct nfp_eth_table {
 		bool enabled;
 		bool tx_enabled;
 		bool rx_enabled;
+		bool supp_aneg;
 
 		bool override_changed;
 
diff --git a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
index 18ba7629cdc2..bb64efec4c46 100644
--- a/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
+++ b/drivers/net/ethernet/netronome/nfp/nfpcore/nfp_nsp_eth.c
@@ -27,6 +27,7 @@
 #define NSP_ETH_PORT_PHYLABEL		GENMASK_ULL(59, 54)
 #define NSP_ETH_PORT_FEC_SUPP_BASER	BIT_ULL(60)
 #define NSP_ETH_PORT_FEC_SUPP_RS	BIT_ULL(61)
+#define NSP_ETH_PORT_SUPP_ANEG		BIT_ULL(63)
 
 #define NSP_ETH_PORT_LANES_MASK		cpu_to_le64(NSP_ETH_PORT_LANES)
 
@@ -178,6 +179,7 @@ nfp_eth_port_translate(struct nfp_nsp *nsp, const union eth_table_entry *src,
 		return;
 
 	dst->act_fec = FIELD_GET(NSP_ETH_STATE_ACT_FEC, state);
+	dst->supp_aneg = FIELD_GET(NSP_ETH_PORT_SUPP_ANEG, port);
 }
 
 static void
-- 
2.30.2

[PATCH net-next v2 5/5] nfp: add support restart of link auto-negotiation

[Simon Horman]

From: Fei Qin <fei.qin@corigine.com>

Add support restart of link auto-negotiation.
This may be initiated using:

  # ethtool -r <intf>

Signed-off-by: Fei Qin <fei.qin@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
---
 .../ethernet/netronome/nfp/nfp_net_ethtool.c  | 33 +++++++++++++++++++
 1 file changed, 33 insertions(+)

diff --git a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
index 678cea0fd274..22a5d2419084 100644
--- a/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
+++ b/drivers/net/ethernet/netronome/nfp/nfp_net_ethtool.c
@@ -228,6 +228,37 @@ nfp_net_get_drvinfo(struct net_device *netdev, struct ethtool_drvinfo *drvinfo)
 	nfp_get_drvinfo(nn->app, nn->pdev, vnic_version, drvinfo);
 }
 
+static int
+nfp_net_nway_reset(struct net_device *netdev)
+{
+	struct nfp_eth_table_port *eth_port;
+	struct nfp_port *port;
+	int err;
+
+	port = nfp_port_from_netdev(netdev);
+	eth_port = nfp_port_get_eth_port(port);
+	if (!eth_port)
+		return -EOPNOTSUPP;
+
+	if (!netif_running(netdev))
+		return 0;
+
+	err = nfp_eth_set_configured(port->app->cpp, eth_port->index, false);
+	if (err) {
+		netdev_info(netdev, "Link down failed: %d\n", err);
+		return err;
+	}
+
+	err = nfp_eth_set_configured(port->app->cpp, eth_port->index, true);
+	if (err) {
+		netdev_info(netdev, "Link up failed: %d\n", err);
+		return err;
+	}
+
+	netdev_info(netdev, "Link reset succeeded\n");
+	return 0;
+}
+
 static void
 nfp_app_get_drvinfo(struct net_device *netdev, struct ethtool_drvinfo *drvinfo)
 {
@@ -1841,6 +1872,7 @@ static const struct ethtool_ops nfp_net_ethtool_ops = {
 				     ETHTOOL_COALESCE_MAX_FRAMES |
 				     ETHTOOL_COALESCE_USE_ADAPTIVE,
 	.get_drvinfo		= nfp_net_get_drvinfo,
+	.nway_reset             = nfp_net_nway_reset,
 	.get_link		= ethtool_op_get_link,
 	.get_ringparam		= nfp_net_get_ringparam,
 	.set_ringparam		= nfp_net_set_ringparam,
@@ -1878,6 +1910,7 @@ static const struct ethtool_ops nfp_net_ethtool_ops = {
 
 const struct ethtool_ops nfp_port_ethtool_ops = {
 	.get_drvinfo		= nfp_app_get_drvinfo,
+	.nway_reset             = nfp_net_nway_reset,
 	.get_link		= ethtool_op_get_link,
 	.get_strings		= nfp_port_get_strings,
 	.get_ethtool_stats	= nfp_port_get_stats,
-- 
2.30.2

Re: [PATCH net-next v2 0/5] nfp: support FEC mode reporting and auto-neg

[Jakub Kicinski]

On Thu, 29 Sep 2022 10:58:27 +0200 Simon Horman wrote:
> this series adds support for the following features to the nfp driver:
> 
> * Patch 1/5: Support active FEC mode
> * Patch 2/5: Don't halt driver on non-fatal error when interacting with fw
> * Patch 3/5: Treat port independence as a firmware rather than port property
> * Patch 4/5: Support link auto negotiation
> * Patch 5/5: Support restart of link auto negotiation

Looks better, thanks for the changes.

BTW shouldn't the sp_indiff symbol be prefixed by _pf%u ?
That's not really introduced by this series tho.

Re: [PATCH net-next v2 0/5] nfp: support FEC mode reporting and auto-neg

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net-next.git (master)
by Jakub Kicinski <kuba@kernel.org>:

On Thu, 29 Sep 2022 10:58:27 +0200 you wrote:
> Hi,
> 
> this series adds support for the following features to the nfp driver:
> 
> * Patch 1/5: Support active FEC mode
> * Patch 2/5: Don't halt driver on non-fatal error when interacting with fw
> * Patch 3/5: Treat port independence as a firmware rather than port property
> * Patch 4/5: Support link auto negotiation
> * Patch 5/5: Support restart of link auto negotiation
> 
> [...]

Here is the summary with links:
  - [net-next,v2,1/5] nfp: add support for reporting active FEC mode
    https://git.kernel.org/netdev/net-next/c/fc26e70f8aca
  - [net-next,v2,2/5] nfp: avoid halt of driver init process when non-fatal error happens
    https://git.kernel.org/netdev/net-next/c/965dd27d9893
  - [net-next,v2,3/5] nfp: refine the ABI of getting `sp_indiff` info
    https://git.kernel.org/netdev/net-next/c/b1e4f11e426d
  - [net-next,v2,4/5] nfp: add support for link auto negotiation
    https://git.kernel.org/netdev/net-next/c/8d545385bf26
  - [net-next,v2,5/5] nfp: add support restart of link auto-negotiation
    https://git.kernel.org/netdev/net-next/c/2820a400dfd3

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

Re: [PATCH net-next v2 0/5] nfp: support FEC mode reporting and auto-neg

[Yinjun Zhang]

On Fri, Sep 30, 2022 at 06:47:35PM -0700, Jakub Kicinski wrote:
> On Thu, 29 Sep 2022 10:58:27 +0200 Simon Horman wrote:
> > this series adds support for the following features to the nfp driver:
> > 
> > * Patch 1/5: Support active FEC mode
> > * Patch 2/5: Don't halt driver on non-fatal error when interacting with fw
> > * Patch 3/5: Treat port independence as a firmware rather than port property
> > * Patch 4/5: Support link auto negotiation
> > * Patch 5/5: Support restart of link auto negotiation
> 
> Looks better, thanks for the changes.
> 
> BTW shouldn't the sp_indiff symbol be prefixed by _pf%u ?
> That's not really introduced by this series tho.

Thanks for your advice. Although sp_indiff is exposed by per-PF rtsym
_pf%u_net_app_cap, which can be used for per-PF capabilities in future,
I think sp_indiff won't be inconsistent among PFs. We'll adjust it if
it happens.

Re: [PATCH net-next v2 0/5] nfp: support FEC mode reporting and auto-neg

[Jakub Kicinski]

On Mon, 3 Oct 2022 16:41:11 +0800 Yinjun Zhang wrote:
> > Looks better, thanks for the changes.
> > 
> > BTW shouldn't the sp_indiff symbol be prefixed by _pf%u ?
> > That's not really introduced by this series tho.  
> 
> Thanks for your advice. Although sp_indiff is exposed by per-PF rtsym
> _pf%u_net_app_cap, which can be used for per-PF capabilities in future,
> I think sp_indiff won't be inconsistent among PFs. We'll adjust it if
> it happens.

It's not about inconsistencies but about the fact that in multi-host
systems there are multiple driver instances which come and go.
The driver seems to set sp_indiff to one at load and to zero when it
unloads. IDK what that actually does to the FW but if it does anything
it's not gonna work reliably in MH.

Re: [PATCH net-next v2 0/5] nfp: support FEC mode reporting and auto-neg

[Yinjun Zhang]

On Mon, Oct 03, 2022 at 01:01:49PM -0700, Jakub Kicinski wrote:
> On Mon, 3 Oct 2022 16:41:11 +0800 Yinjun Zhang wrote:
> > Thanks for your advice. Although sp_indiff is exposed by per-PF rtsym
> > _pf%u_net_app_cap, which can be used for per-PF capabilities in future,
> > I think sp_indiff won't be inconsistent among PFs. We'll adjust it if
> > it happens.
> 
> It's not about inconsistencies but about the fact that in multi-host
> systems there are multiple driver instances which come and go.
> The driver seems to set sp_indiff to one at load and to zero when it
> unloads. IDK what that actually does to the FW but if it does anything
> it's not gonna work reliably in MH.

I see, it's indeed a problem, not only in MH case, but also in
single-host-multi-PF case. Instead of renaming sp_indiff, I'll use
`unload_fw_on_remove` to decide if need clean sp_indiff when unloading.
I think it makes more sense. Will fix it.