Skip to contentSkip to conversation

Add PTP support for Octeontx2

PATCH In review Subbaraya Sundeep 5 participants 11 messages Jul 15, 2020 – Jul 18, 2020

Subsystems (4)
lore
Share & tools
lore

Subbaraya Sundeep

author maintainer
#0 lore raw

v4 cover Add PTP support for Octeontx2

Hi,

This patchset adds PTP support for Octeontx2 platform. PTP is an independent coprocessor block from which CGX block fetches timestamp and prepends it to the packet before sending to NIX block. Patches are as follows:

Patch 1: Patch to enable/disable packet timstamping
in CGX upon mailbox request. It also adjusts packet parser (NPC) for the 8 bytes timestamp appearing before the packet.

Patch 2: Patch adding PTP pci driver which configures
the PTP block and hooks up to RVU AF driver. It also exposes a mailbox call to adjust PTP hardware clock.

Patch 3: Patch adding PTP clock driver for PF netdev.

Aleksey Makarov (2):
octeontx2-af: Add support for Marvell PTP coprocessor
octeontx2-pf: Add support for PTP clock

Zyta Szpak (1):
octeontx2-af: Support to enable/disable HW timestamping

 drivers/net/ethernet/marvell/octeontx2/af/Makefile |   2 +-
 drivers/net/ethernet/marvell/octeontx2/af/cgx.c    |  29 +++
 drivers/net/ethernet/marvell/octeontx2/af/cgx.h    |   4 +
 drivers/net/ethernet/marvell/octeontx2/af/mbox.h   |  21 ++
 drivers/net/ethernet/marvell/octeontx2/af/ptp.c    | 244 +++++++++++++++++++++
 drivers/net/ethernet/marvell/octeontx2/af/ptp.h    |  22 ++
 drivers/net/ethernet/marvell/octeontx2/af/rvu.c    |  29 ++-
 drivers/net/ethernet/marvell/octeontx2/af/rvu.h    |   5 +
 .../net/ethernet/marvell/octeontx2/af/rvu_cgx.c    |  54 +++++
 .../net/ethernet/marvell/octeontx2/af/rvu_nix.c    |  52 +++++
 .../net/ethernet/marvell/octeontx2/af/rvu_npc.c    |  27 +++
 .../net/ethernet/marvell/octeontx2/nic/Makefile    |   3 +-
 .../ethernet/marvell/octeontx2/nic/otx2_common.c   |   7 +
 .../ethernet/marvell/octeontx2/nic/otx2_common.h   |  19 ++
 .../ethernet/marvell/octeontx2/nic/otx2_ethtool.c  |  28 +++
 .../net/ethernet/marvell/octeontx2/nic/otx2_pf.c   | 170 +++++++++++++-
 .../net/ethernet/marvell/octeontx2/nic/otx2_ptp.c  | 209 ++++++++++++++++++
 .../net/ethernet/marvell/octeontx2/nic/otx2_ptp.h  |  13 ++
 .../net/ethernet/marvell/octeontx2/nic/otx2_txrx.c |  87 +++++++-
 .../net/ethernet/marvell/octeontx2/nic/otx2_txrx.h |   1 +
 20 files changed, 1016 insertions(+), 10 deletions(-)
 create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/ptp.c
 create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/ptp.h
 create mode 100644 drivers/net/ethernet/marvell/octeontx2/nic/otx2_ptp.c
 create mode 100644 drivers/net/ethernet/marvell/octeontx2/nic/otx2_ptp.h
Signature · 1 line
2.7.4
author maintainer
#1 lore raw

PATCH v4 1/3 octeontx2-af: Support to enable/disable HW timestamping

From: Zyta Szpak <[email protected]>

Four new mbox messages ids and handler are added in order to enable or disable timestamping procedure on tx and rx side. Additionally when PTP is enabled, the packet parser must skip over 8 bytes and start analyzing packet data there. To make NPC profiles work seemlesly PTR_ADVANCE of IKPU is set so that parsing can be done as before when all data pointers are shifted by 8 bytes automatically.

---

v4:
No changes
v3:
No changes
v2:
No changes

 drivers/net/ethernet/marvell/octeontx2/af/cgx.c    | 29 ++++++++++++
 drivers/net/ethernet/marvell/octeontx2/af/cgx.h    |  4 ++
 drivers/net/ethernet/marvell/octeontx2/af/mbox.h   |  4 ++
 drivers/net/ethernet/marvell/octeontx2/af/rvu.h    |  1 +
 .../net/ethernet/marvell/octeontx2/af/rvu_cgx.c    | 54 ++++++++++++++++++++++
 .../net/ethernet/marvell/octeontx2/af/rvu_nix.c    | 52 +++++++++++++++++++++
 .../net/ethernet/marvell/octeontx2/af/rvu_npc.c    | 27 +++++++++++
 7 files changed, 171 insertions(+)
drivers/net/ethernet/marvell/octeontx2/af/cgx.c +29 -0
index a4e65da..8f17e26 100644
@@ -468,6 +468,35 @@ static void cgx_lmac_pause_frm_config(struct cgx *cgx, int lmac_id, bool enable)
 	}
 }
 
+void cgx_lmac_ptp_config(void *cgxd, int lmac_id, bool enable)
+{
+	struct cgx *cgx = cgxd;
+	u64 cfg;
+
+	if (!cgx)
+		return;
+
+	if (enable) {
+		/* Enable inbound PTP timestamping */
+		cfg = cgx_read(cgx, lmac_id, CGXX_GMP_GMI_RXX_FRM_CTL);
+		cfg |= CGX_GMP_GMI_RXX_FRM_CTL_PTP_MODE;
+		cgx_write(cgx, lmac_id, CGXX_GMP_GMI_RXX_FRM_CTL, cfg);
+
+		cfg = cgx_read(cgx, lmac_id, CGXX_SMUX_RX_FRM_CTL);
+		cfg |= CGX_SMUX_RX_FRM_CTL_PTP_MODE;
+		cgx_write(cgx, lmac_id,	CGXX_SMUX_RX_FRM_CTL, cfg);
+	} else {
+		/* Disable inbound PTP stamping */
+		cfg = cgx_read(cgx, lmac_id, CGXX_GMP_GMI_RXX_FRM_CTL);
+		cfg &= ~CGX_GMP_GMI_RXX_FRM_CTL_PTP_MODE;
+		cgx_write(cgx, lmac_id, CGXX_GMP_GMI_RXX_FRM_CTL, cfg);
+
+		cfg = cgx_read(cgx, lmac_id, CGXX_SMUX_RX_FRM_CTL);
+		cfg &= ~CGX_SMUX_RX_FRM_CTL_PTP_MODE;
+		cgx_write(cgx, lmac_id, CGXX_SMUX_RX_FRM_CTL, cfg);
+	}
+}
+
 /* CGX Firmware interface low level support */
 static int cgx_fwi_cmd_send(u64 req, u64 *resp, struct lmac *lmac)
 {
drivers/net/ethernet/marvell/octeontx2/af/cgx.h +4 -0
index 394f965..27ca329 100644
@@ -58,8 +58,10 @@
 
 #define CGXX_SMUX_RX_FRM_CTL		0x20020
 #define CGX_SMUX_RX_FRM_CTL_CTL_BCK	BIT_ULL(3)
+#define CGX_SMUX_RX_FRM_CTL_PTP_MODE	BIT_ULL(12)
 #define CGXX_GMP_GMI_RXX_FRM_CTL	0x38028
 #define CGX_GMP_GMI_RXX_FRM_CTL_CTL_BCK	BIT_ULL(3)
+#define CGX_GMP_GMI_RXX_FRM_CTL_PTP_MODE BIT_ULL(12)
 #define CGXX_SMUX_TX_CTL		0x20178
 #define CGXX_SMUX_TX_PAUSE_PKT_TIME	0x20110
 #define CGXX_SMUX_TX_PAUSE_PKT_INTERVAL	0x20120
@@ -139,4 +141,6 @@ int cgx_lmac_get_pause_frm(void *cgxd, int lmac_id,
 			   u8 *tx_pause, u8 *rx_pause);
 int cgx_lmac_set_pause_frm(void *cgxd, int lmac_id,
 			   u8 tx_pause, u8 rx_pause);
+void cgx_lmac_ptp_config(void *cgxd, int lmac_id, bool enable);
+
 #endif /* CGX_H */
drivers/net/ethernet/marvell/octeontx2/af/mbox.h +4 -0
index 6dfd0f9..c89b098 100644
@@ -143,6 +143,8 @@ M(CGX_STOP_LINKEVENTS,	0x208, cgx_stop_linkevents, msg_req, msg_rsp)	\
 M(CGX_GET_LINKINFO,	0x209, cgx_get_linkinfo, msg_req, cgx_link_info_msg) \
 M(CGX_INTLBK_ENABLE,	0x20A, cgx_intlbk_enable, msg_req, msg_rsp)	\
 M(CGX_INTLBK_DISABLE,	0x20B, cgx_intlbk_disable, msg_req, msg_rsp)	\
+M(CGX_PTP_RX_ENABLE,	0x20C, cgx_ptp_rx_enable, msg_req, msg_rsp)	\
+M(CGX_PTP_RX_DISABLE,	0x20D, cgx_ptp_rx_disable, msg_req, msg_rsp)	\
 M(CGX_CFG_PAUSE_FRM,	0x20E, cgx_cfg_pause_frm, cgx_pause_frm_cfg,	\
 			       cgx_pause_frm_cfg)			\
 /* NPA mbox IDs (range 0x400 - 0x5FF) */				\
@@ -213,6 +215,8 @@ M(NIX_LSO_FORMAT_CFG,	0x8011, nix_lso_format_cfg,			\
 				 nix_lso_format_cfg,			\
 				 nix_lso_format_cfg_rsp)		\
 M(NIX_RXVLAN_ALLOC,	0x8012, nix_rxvlan_alloc, msg_req, msg_rsp)	\
+M(NIX_LF_PTP_TX_ENABLE, 0x8013, nix_lf_ptp_tx_enable, msg_req, msg_rsp)	\
+M(NIX_LF_PTP_TX_DISABLE, 0x8014, nix_lf_ptp_tx_disable, msg_req, msg_rsp) \
 M(NIX_BP_ENABLE,	0x8016, nix_bp_enable, nix_bp_cfg_req,	\
 				nix_bp_cfg_rsp)	\
 M(NIX_BP_DISABLE,	0x8017, nix_bp_disable, nix_bp_cfg_req, msg_rsp) \
drivers/net/ethernet/marvell/octeontx2/af/rvu.h +1 -0
index dcf25a0..62c3ed2 100644
@@ -469,6 +469,7 @@ int rvu_npc_init(struct rvu *rvu);
 void rvu_npc_freemem(struct rvu *rvu);
 int rvu_npc_get_pkind(struct rvu *rvu, u16 pf);
 void rvu_npc_set_pkind(struct rvu *rvu, int pkind, struct rvu_pfvf *pfvf);
+int npc_config_ts_kpuaction(struct rvu *rvu, int pf, u16 pcifunc, bool en);
 void rvu_npc_install_ucast_entry(struct rvu *rvu, u16 pcifunc,
 				 int nixlf, u64 chan, u8 *mac_addr);
 void rvu_npc_install_promisc_entry(struct rvu *rvu, u16 pcifunc,
drivers/net/ethernet/marvell/octeontx2/af/rvu_cgx.c +54 -0

60 lines not on this page · raw message

drivers/net/ethernet/marvell/octeontx2/af/rvu_nix.c +52 -0

58 lines not on this page · raw message

drivers/net/ethernet/marvell/octeontx2/af/rvu_npc.c +27 -0
index 0a21408..8179bbe 100644
@@ -27,6 +27,7 @@
 #define NIXLF_PROMISC_ENTRY	2
 
 #define NPC_PARSE_RESULT_DMAC_OFFSET	8
+#define NPC_HW_TSTAMP_OFFSET		8
 
 static void npc_mcam_free_all_entries(struct rvu *rvu, struct npc_mcam *mcam,
 				      int blkaddr, u16 pcifunc);
@@ -61,6 +62,32 @@ int rvu_npc_get_pkind(struct rvu *rvu, u16 pf)
 	return -1;
 }
 
+int npc_config_ts_kpuaction(struct rvu *rvu, int pf, u16 pcifunc, bool en)
+{
+	int pkind, blkaddr;
+	u64 val;
+
+	pkind = rvu_npc_get_pkind(rvu, pf);
+	if (pkind < 0) {
+		dev_err(rvu->dev, "%s: pkind not mapped\n", __func__);
+		return -EINVAL;
+	}
+
+	blkaddr = rvu_get_blkaddr(rvu, BLKTYPE_NPC, pcifunc);
+	if (blkaddr < 0) {
+		dev_err(rvu->dev, "%s: NPC block not implemented\n", __func__);
+		return -EINVAL;
+	}
+	val = rvu_read64(rvu, blkaddr, NPC_AF_PKINDX_ACTION0(pkind));
+	val &= ~0xff00000ULL; /* Zero ptr advance field */
+	if (en)
+		/* Set to timestamp offset */
+		val |= (NPC_HW_TSTAMP_OFFSET << 20);
+	rvu_write64(rvu, blkaddr, NPC_AF_PKINDX_ACTION0(pkind), val);
+
+	return 0;
+}
+
 static int npc_get_nixlf_mcam_index(struct npc_mcam *mcam,
 				    u16 pcifunc, int nixlf, int type)
 {
Signature · 1 line
2.7.4
author maintainer
#2 lore raw

PATCH v4 2/3 octeontx2-af: Add support for Marvell PTP coprocessor

In reply to Subbaraya Sundeep

From: Aleksey Makarov <[email protected]>

This patch adds driver for Precision Time
Protocol Clock and Timestamping block found on
Octeontx2 platform. The driver does initial
configuration and exposes a function to adjust
PTP hardware clock.

---

v4:
No changes
v3:
No changes
v2:
No changes

 drivers/net/ethernet/marvell/octeontx2/af/Makefile |   2 +-
 drivers/net/ethernet/marvell/octeontx2/af/mbox.h   |  17 ++
 drivers/net/ethernet/marvell/octeontx2/af/ptp.c    | 244 +++++++++++++++++++++
 drivers/net/ethernet/marvell/octeontx2/af/ptp.h    |  22 ++
 drivers/net/ethernet/marvell/octeontx2/af/rvu.c    |  29 ++-
 drivers/net/ethernet/marvell/octeontx2/af/rvu.h    |   4 +
 6 files changed, 314 insertions(+), 4 deletions(-)
 create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/ptp.c
 create mode 100644 drivers/net/ethernet/marvell/octeontx2/af/ptp.h
drivers/net/ethernet/marvell/octeontx2/af/Makefile +1 -1
index 1b25948..0bc2410 100644
@@ -8,4 +8,4 @@ obj-$(CONFIG_OCTEONTX2_AF) += octeontx2_af.o
 
 octeontx2_mbox-y := mbox.o
 octeontx2_af-y := cgx.o rvu.o rvu_cgx.o rvu_npa.o rvu_nix.o \
-		  rvu_reg.o rvu_npc.o rvu_debugfs.o
+		  rvu_reg.o rvu_npc.o rvu_debugfs.o ptp.o
drivers/net/ethernet/marvell/octeontx2/af/mbox.h +17 -0
index c89b098..4aaef0a 100644
@@ -127,6 +127,7 @@ M(ATTACH_RESOURCES,	0x002, attach_resources, rsrc_attach, msg_rsp)	\
 M(DETACH_RESOURCES,	0x003, detach_resources, rsrc_detach, msg_rsp)	\
 M(MSIX_OFFSET,		0x005, msix_offset, msg_req, msix_offset_rsp)	\
 M(VF_FLR,		0x006, vf_flr, msg_req, msg_rsp)		\
+M(PTP_OP,		0x007, ptp_op, ptp_req, ptp_rsp)		\
 M(GET_HW_CAP,		0x008, get_hw_cap, msg_req, get_hw_cap_rsp)	\
 /* CGX mbox IDs (range 0x200 - 0x3FF) */				\
 M(CGX_START_RXTX,	0x200, cgx_start_rxtx, msg_req, msg_rsp)	\
@@ -862,4 +863,20 @@ struct npc_get_kex_cfg_rsp {
 	u8 mkex_pfl_name[MKEX_NAME_LEN];
 };
 
+enum ptp_op {
+	PTP_OP_ADJFINE = 0,
+	PTP_OP_GET_CLOCK = 1,
+};
+
+struct ptp_req {
+	struct mbox_msghdr hdr;
+	u8 op;
+	s64 scaled_ppm;
+};
+
+struct ptp_rsp {
+	struct mbox_msghdr hdr;
+	u64 clk;
+};
+
 #endif /* MBOX_H */
drivers/net/ethernet/marvell/octeontx2/af/ptp.c new file +244 -0

244 lines not on this page · raw message

drivers/net/ethernet/marvell/octeontx2/af/ptp.h new file +22 -0
new file mode 100644
index 0000000..a344722
@@ -0,0 +1,22 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Marvell PTP driver */
+
+#ifndef PTP_H
+#define PTP_H
+
+#include <linux/timecounter.h>
+#include <linux/time64.h>
+#include <linux/spinlock.h>
+
+struct ptp {
+	struct pci_dev *pdev;
+	void __iomem *reg_base;
+	u32 clock_rate;
+};
+
+struct ptp *ptp_get(void);
+void ptp_put(struct ptp *ptp);
+
+extern struct pci_driver ptp_driver;
+
+#endif
drivers/net/ethernet/marvell/octeontx2/af/rvu.c +26 -3
index 557e429..c3ef73a 100644
@@ -18,6 +18,7 @@
 #include "cgx.h"
 #include "rvu.h"
 #include "rvu_reg.h"
+#include "ptp.h"
 
 #define DRV_NAME	"octeontx2-af"
 #define DRV_STRING      "Marvell OcteonTX2 RVU Admin Function Driver"
@@ -2565,13 +2566,21 @@ static int rvu_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 
 	pci_set_master(pdev);
 
+	rvu->ptp = ptp_get();
+	if (IS_ERR(rvu->ptp)) {
+		err = PTR_ERR(rvu->ptp);
+		if (err == -EPROBE_DEFER)
+			goto err_release_regions;
+		rvu->ptp = NULL;
+	}
+
 	/* Map Admin function CSRs */
 	rvu->afreg_base = pcim_iomap(pdev, PCI_AF_REG_BAR_NUM, 0);
 	rvu->pfreg_base = pcim_iomap(pdev, PCI_PF_REG_BAR_NUM, 0);
 	if (!rvu->afreg_base || !rvu->pfreg_base) {
 		dev_err(dev, "Unable to map admin function CSRs, aborting\n");
 		err = -ENOMEM;
-		goto err_release_regions;
+		goto err_put_ptp;
 	}
 
 	/* Store module params in rvu structure */
@@ -2586,7 +2595,7 @@ static int rvu_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 
 	err = rvu_setup_hw_resources(rvu);
 	if (err)
-		goto err_release_regions;
+		goto err_put_ptp;
 
 	/* Init mailbox btw AF and PFs */
 	err = rvu_mbox_init(rvu, &rvu->afpf_wq_info, TYPE_AFPF,
@@ -2626,6 +2635,8 @@ static int rvu_probe(struct pci_dev *pdev, const struct pci_device_id *id)
 	rvu_reset_all_blocks(rvu);
 	rvu_free_hw_resources(rvu);
 	rvu_clear_rvum_blk_revid(rvu);
+err_put_ptp:
+	ptp_put(rvu->ptp);
 err_release_regions:
 	pci_release_regions(pdev);
 err_disable_device:
@@ -2651,6 +2662,7 @@ static void rvu_remove(struct pci_dev *pdev)
 	rvu_reset_all_blocks(rvu);
 	rvu_free_hw_resources(rvu);
 	rvu_clear_rvum_blk_revid(rvu);
+	ptp_put(rvu->ptp);
 	pci_release_regions(pdev);
 	pci_disable_device(pdev);
 	pci_set_drvdata(pdev, NULL);
@@ -2676,9 +2688,19 @@ static int __init rvu_init_module(void)
 	if (err < 0)
 		return err;
 
+	err = pci_register_driver(&ptp_driver);
+	if (err < 0)
+		goto ptp_err;
+
 	err =  pci_register_driver(&rvu_driver);
 	if (err < 0)
-		pci_unregister_driver(&cgx_driver);
+		goto rvu_err;
+
+	return 0;
+rvu_err:
+	pci_unregister_driver(&ptp_driver);
+ptp_err:
+	pci_unregister_driver(&cgx_driver);
 
 	return err;
 }
@@ -2686,6 +2708,7 @@ static int __init rvu_init_module(void)
 static void __exit rvu_cleanup_module(void)
 {
 	pci_unregister_driver(&rvu_driver);
+	pci_unregister_driver(&ptp_driver);
 	pci_unregister_driver(&cgx_driver);
 }
 
drivers/net/ethernet/marvell/octeontx2/af/rvu.h +4 -0
index 62c3ed2..05da7a9 100644
@@ -289,6 +289,8 @@ struct rvu_fwdata {
 	u64 reserved[FWDATA_RESERVED_MEM];
 };
 
+struct ptp;
+
 struct rvu {
 	void __iomem		*afreg_base;
 	void __iomem		*pfreg_base;
@@ -337,6 +339,8 @@ struct rvu {
 	/* Firmware data */
 	struct rvu_fwdata	*fwdata;
 
+	struct ptp		*ptp;
+
 #ifdef CONFIG_DEBUG_FS
 	struct rvu_debugfs	rvu_dbg;
 #endif
Signature · 1 line
2.7.4
author maintainer
#3 lore raw

PATCH v4 3/3 octeontx2-pf: Add support for PTP clock

In reply to Subbaraya Sundeep

From: Aleksey Makarov <[email protected]>

This patch adds PTP clock and uses it in Octeontx2 network device. PTP clock uses mailbox calls to access the hardware counter on the RVU side.

---

v4:
Added rtnl_lock/unlock in otx2_reset to protect against network stack ndo_open and close calls
Added NULL check after ptp_clock_register in otx2_ptp.c
v3:
Fixed sparse error in otx2_txrx.c
Removed static inlines in otx2_txrx.c
v2:
Fixed kernel build robot reported error by adding timecounter.h to otx2_common.h

 .../net/ethernet/marvell/octeontx2/nic/Makefile    |   3 +-
 .../ethernet/marvell/octeontx2/nic/otx2_common.c   |   7 +
 .../ethernet/marvell/octeontx2/nic/otx2_common.h   |  19 ++
 .../ethernet/marvell/octeontx2/nic/otx2_ethtool.c  |  28 +++
 .../net/ethernet/marvell/octeontx2/nic/otx2_pf.c   | 170 ++++++++++++++++-
 .../net/ethernet/marvell/octeontx2/nic/otx2_ptp.c  | 209 +++++++++++++++++++++
 .../net/ethernet/marvell/octeontx2/nic/otx2_ptp.h  |  13 ++
 .../net/ethernet/marvell/octeontx2/nic/otx2_txrx.c |  87 ++++++++-
 .../net/ethernet/marvell/octeontx2/nic/otx2_txrx.h |   1 +
 9 files changed, 531 insertions(+), 6 deletions(-)
 create mode 100644 drivers/net/ethernet/marvell/octeontx2/nic/otx2_ptp.c
 create mode 100644 drivers/net/ethernet/marvell/octeontx2/nic/otx2_ptp.h
drivers/net/ethernet/marvell/octeontx2/nic/Makefile +2 -1
index 778df33..b2c6385 100644
@@ -6,7 +6,8 @@
 obj-$(CONFIG_OCTEONTX2_PF) += octeontx2_nicpf.o
 obj-$(CONFIG_OCTEONTX2_VF) += octeontx2_nicvf.o
 
-octeontx2_nicpf-y := otx2_pf.o otx2_common.o otx2_txrx.o otx2_ethtool.o
+octeontx2_nicpf-y := otx2_pf.o otx2_common.o otx2_txrx.o otx2_ethtool.o \
+		     otx2_ptp.o
 octeontx2_nicvf-y := otx2_vf.o
 
 ccflags-y += -I$(srctree)/drivers/net/ethernet/marvell/octeontx2/af
drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.c +7 -0
index 5975521..9cb3dc8 100644
@@ -671,6 +671,13 @@ static int otx2_sq_init(struct otx2_nic *pfvf, u16 qidx, u16 sqb_aura)
 	if (!sq->sg)
 		return -ENOMEM;
 
+	if (pfvf->ptp) {
+		err = qmem_alloc(pfvf->dev, &sq->timestamps, qset->sqe_cnt,
+				 sizeof(*sq->timestamps));
+		if (err)
+			return err;
+	}
+
 	sq->head = 0;
 	sq->sqe_per_sqb = (pfvf->hw.sqb_size / sq->sqe_size) - 1;
 	sq->num_sqbs = (qset->sqe_cnt + sq->sqe_per_sqb) / sq->sqe_per_sqb;
drivers/net/ethernet/marvell/octeontx2/nic/otx2_common.h +19 -0
index 2fa2988..689925b 100644
@@ -13,6 +13,9 @@
 
 #include <linux/pci.h>
 #include <linux/iommu.h>
+#include <linux/net_tstamp.h>
+#include <linux/ptp_clock_kernel.h>
+#include <linux/timecounter.h>
 
 #include <mbox.h>
 #include "otx2_reg.h"
@@ -209,6 +212,17 @@ struct refill_work {
 	struct otx2_nic *pf;
 };
 
+struct otx2_ptp {
+	struct ptp_clock_info ptp_info;
+	struct ptp_clock *ptp_clock;
+	struct otx2_nic *nic;
+
+	struct cyclecounter cycle_counter;
+	struct timecounter time_counter;
+};
+
+#define OTX2_HW_TIMESTAMP_LEN	8
+
 struct otx2_nic {
 	void __iomem		*reg_base;
 	struct net_device	*netdev;
@@ -216,6 +230,8 @@ struct otx2_nic {
 	u16			max_frs;
 	u16			rbsize; /* Receive buffer size */
 
+#define OTX2_FLAG_RX_TSTAMP_ENABLED		BIT_ULL(0)
+#define OTX2_FLAG_TX_TSTAMP_ENABLED		BIT_ULL(1)
 #define OTX2_FLAG_INTF_DOWN			BIT_ULL(2)
 #define OTX2_FLAG_RX_PAUSE_ENABLED		BIT_ULL(9)
 #define OTX2_FLAG_TX_PAUSE_ENABLED		BIT_ULL(10)
@@ -251,6 +267,9 @@ struct otx2_nic {
 
 	/* Block address of NIX either BLKADDR_NIX0 or BLKADDR_NIX1 */
 	int			nix_blkaddr;
+
+	struct otx2_ptp		*ptp;
+	struct hwtstamp_config	tstamp;
 };
 
 static inline bool is_otx2_lbkvf(struct pci_dev *pdev)
drivers/net/ethernet/marvell/octeontx2/nic/otx2_ethtool.c +28 -0
index d59f5a9..0341d969 100644
@@ -13,8 +13,10 @@
 #include <linux/stddef.h>
 #include <linux/etherdevice.h>
 #include <linux/log2.h>
+#include <linux/net_tstamp.h>
 
 #include "otx2_common.h"
+#include "otx2_ptp.h"
 
 #define DRV_NAME	"octeontx2-nicpf"
 #define DRV_VF_NAME	"octeontx2-nicvf"
@@ -663,6 +665,31 @@ static u32 otx2_get_link(struct net_device *netdev)
 	return pfvf->linfo.link_up;
 }
 
+static int otx2_get_ts_info(struct net_device *netdev,
+			    struct ethtool_ts_info *info)
+{
+	struct otx2_nic *pfvf = netdev_priv(netdev);
+
+	if (!pfvf->ptp)
+		return ethtool_op_get_ts_info(netdev, info);
+
+	info->so_timestamping = SOF_TIMESTAMPING_TX_SOFTWARE |
+				SOF_TIMESTAMPING_RX_SOFTWARE |
+				SOF_TIMESTAMPING_SOFTWARE |
+				SOF_TIMESTAMPING_TX_HARDWARE |
+				SOF_TIMESTAMPING_RX_HARDWARE |
+				SOF_TIMESTAMPING_RAW_HARDWARE;
+
+	info->phc_index = otx2_ptp_clock_index(pfvf);
+
+	info->tx_types = (1 << HWTSTAMP_TX_OFF) | (1 << HWTSTAMP_TX_ON);
+
+	info->rx_filters = (1 << HWTSTAMP_FILTER_NONE) |
+			   (1 << HWTSTAMP_FILTER_ALL);
+
+	return 0;
+}
+
 static const struct ethtool_ops otx2_ethtool_ops = {
 	.supported_coalesce_params = ETHTOOL_COALESCE_USECS |
 				     ETHTOOL_COALESCE_MAX_FRAMES,
@@ -687,6 +714,7 @@ static const struct ethtool_ops otx2_ethtool_ops = {
 	.set_msglevel		= otx2_set_msglevel,
 	.get_pauseparam		= otx2_get_pauseparam,
 	.set_pauseparam		= otx2_set_pauseparam,
+	.get_ts_info		= otx2_get_ts_info,
 };
 
 void otx2_set_ethtool_ops(struct net_device *netdev)
drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +168 -2

240 lines not on this page · raw message

drivers/net/ethernet/marvell/octeontx2/nic/otx2_ptp.c new file +209 -0

209 lines not on this page · raw message

drivers/net/ethernet/marvell/octeontx2/nic/otx2_ptp.h new file +13 -0
new file mode 100644
index 0000000..706d63a
@@ -0,0 +1,13 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/* Marvell OcteonTx2 PTP support for ethernet driver */
+
+#ifndef OTX2_PTP_H
+#define OTX2_PTP_H
+
+int otx2_ptp_init(struct otx2_nic *pfvf);
+void otx2_ptp_destroy(struct otx2_nic *pfvf);
+
+int otx2_ptp_clock_index(struct otx2_nic *pfvf);
+int otx2_ptp_tstamp2time(struct otx2_nic *pfvf, u64 tstamp, u64 *tsns);
+
+#endif
drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c +84 -3

149 lines not on this page · raw message

drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.h +1 -0
index da97f2d4..73af156 100644
@@ -91,6 +91,7 @@ struct otx2_snd_queue {
 	struct qmem		*sqe;
 	struct qmem		*tso_hdrs;
 	struct sg_list		*sg;
+	struct qmem		*timestamps;
 	struct queue_stats	stats;
 	u16			sqb_count;
 	u64			*sqb_ptrs;
Signature · 1 line
2.7.4
maintainer Acked-by on v4 3/3
#4 lore raw

Re: [PATCH v4 net-next 3/3] octeontx2-pf: Add support for PTP clock

drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c static void otx2_reset_task(struct work_struct *work) · line 1755

Quoted 12 lines · On Wed, 15 Jul 2020 18:08:09 +0530 Subbaraya Sundeep wrote:
drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +4 -0
@@ -1730,10 +1745,149 @@ static void otx2_reset_task(struct work_struct *work)
 	if (!netif_running(pf->netdev))
 		return;
 
+	rtnl_lock();
 	otx2_stop(pf->netdev);
 	pf->reset_count++;
 	otx2_open(pf->netdev);
 	netif_trans_update(pf->netdev);
+	rtnl_unlock();
+}
+

This looks unrelated, otherwise for the patches:

#5 lore raw

Re: [PATCH v4 net-next 3/3] octeontx2-pf: Add support for PTP clock

Hi Jakub,

Quoted 6 lines · Jakub Kicinski
Quoted 4 lines · On Wed, 15 Jul 2020 18:08:09 +0530 Subbaraya Sundeep wrote:
drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +0 -0
@@ -1730,10 +1745,149 @@ static void otx2_reset_task(struct work_struct *work)
      if (!netif_running(pf->netdev))
              return;
 
Quoted 10 lines

drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c static void otx2_reset_task(struct work_struct *work) · line 1755

Quoted 8 lines
drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +4 -0
+     rtnl_lock();
      otx2_stop(pf->netdev);
      pf->reset_count++;
      otx2_open(pf->netdev);
      netif_trans_update(pf->netdev);
+     rtnl_unlock();
+}
+

This looks unrelated, otherwise for the patches:

You mean the lock/unlock logic with this patch?
I can separate this out and put in another patch #4 if you insist.

Thanks,
Sundeep

Quoted 2 lines
maintainer
#6 lore raw

Re: [PATCH v4 net-next 3/3] octeontx2-pf: Add support for PTP clock

Quoted 16 lines · On Fri, 17 Jul 2020 10:41:49 +0530 sundeep subbaraya wrote:
Quoted 5 lines · Jakub Kicinski
Quoted 4 lines · On Wed, 15 Jul 2020 18:08:09 +0530 Subbaraya Sundeep wrote:
drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +0 -0
@@ -1730,10 +1745,149 @@ static void otx2_reset_task(struct work_struct *work)
      if (!netif_running(pf->netdev))
              return;
 
Quoted 9 lines

drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c static void otx2_reset_task(struct work_struct *work) · line 1754

Quoted 7 lines
drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +3 -0
+     rtnl_lock();
      otx2_stop(pf->netdev);
      pf->reset_count++;
      otx2_open(pf->netdev);
      netif_trans_update(pf->netdev);
+     rtnl_unlock();
+}

This looks unrelated, otherwise for the patches:

Quoted 1 line

You mean the lock/unlock logic with this patch?

Looks very much like a bug independent of the PTP support.

Also

$ git grep reset_task drivers/net/ethernet/marvell/octeontx2/
Doesn't reveal any place where you would flush or cancel that work.
Quoted 1 line

I can separate this out and put in another patch #4 if you insist.

Does someone need to insist for you to fix your bugs in the current release cycle? That's a basic part of the kernel release process :/

maintainer
#7 lore raw

Re: [PATCH v4 net-next 3/3] octeontx2-pf: Add support for PTP clock

From: Jakub Kicinski <[email protected]>
Date: Fri, 17 Jul 2020 10:48:12 -0700

Quoted 5 lines · On Fri, 17 Jul 2020 10:41:49 +0530 sundeep subbaraya wrote:
Quoted 1 line

I can separate this out and put in another patch #4 if you insist.

Does someone need to insist for you to fix your bugs in the current release cycle? That's a basic part of the kernel release process :/

Please submit the bug fix for the 'net' tree.

Wait for the net tree to eventually get merged into net-next.

Then you can resubmit this series on top.

That's how we do development, and we would appreciate it if you would submit bug fixes and new features properly.

Thank you.

#8 lore raw

Re: [PATCH v4 net-next 3/3] octeontx2-pf: Add support for PTP clock

Hi David,

Quoted 12 lines · David Miller

From: Jakub Kicinski <[email protected]>
Date: Fri, 17 Jul 2020 10:48:12 -0700

Quoted 5 lines · On Fri, 17 Jul 2020 10:41:49 +0530 sundeep subbaraya wrote:
Quoted 1 line

I can separate this out and put in another patch #4 if you insist.

Does someone need to insist for you to fix your bugs in the current release cycle? That's a basic part of the kernel release process :/

Please submit the bug fix for the 'net' tree.

Quoted 8 lines

Wait for the net tree to eventually get merged into net-next.

Then you can resubmit this series on top.

That's how we do development, and we would appreciate it if you would submit bug fixes and new features properly.

Thank you.

Sure will submit as you said.

Thanks,
Sundeep

#9 lore raw

Re: [PATCH v4 net-next 3/3] octeontx2-pf: Add support for PTP clock

In reply to Jakub Kicinski

Hi Jakub,

Quoted 22 lines · Jakub Kicinski
Quoted 16 lines · On Fri, 17 Jul 2020 10:41:49 +0530 sundeep subbaraya wrote:
Quoted 5 lines · Jakub Kicinski
Quoted 4 lines · On Wed, 15 Jul 2020 18:08:09 +0530 Subbaraya Sundeep wrote:
drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +0 -0
@@ -1730,10 +1745,149 @@ static void otx2_reset_task(struct work_struct *work)
      if (!netif_running(pf->netdev))
              return;
 
Quoted 9 lines

drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c static void otx2_reset_task(struct work_struct *work) · line 1754

Quoted 7 lines
drivers/net/ethernet/marvell/octeontx2/nic/otx2_pf.c +3 -0
+     rtnl_lock();
      otx2_stop(pf->netdev);
      pf->reset_count++;
      otx2_open(pf->netdev);
      netif_trans_update(pf->netdev);
+     rtnl_unlock();
+}

This looks unrelated, otherwise for the patches:

Quoted 1 line

You mean the lock/unlock logic with this patch?

Looks very much like a bug independent of the PTP support.

Quoted 8 lines

Also

$ git grep reset_task drivers/net/ethernet/marvell/octeontx2/
Doesn't reveal any place where you would flush or cancel that work.
Quoted 1 line

I can separate this out and put in another patch #4 if you insist.

Does someone need to insist for you to fix your bugs in the current release cycle? That's a basic part of the kernel release process :/

My confusion was whether it is okay to modify patches after Acked-by because reviewers have to review again. It is clear now.

Thanks,
Sundeep

Acked-by on the v4 series
#10 lore raw
In reply to Subbaraya Sundeep
Quoted 21 lines · Subbaraya Sundeep

Hi,

This patchset adds PTP support for Octeontx2 platform. PTP is an independent coprocessor block from which CGX block fetches timestamp and prepends it to the packet before sending to NIX block. Patches are as follows:

Patch 1: Patch to enable/disable packet timstamping
in CGX upon mailbox request. It also adjusts packet parser (NPC) for the 8 bytes timestamp appearing before the packet.

Patch 2: Patch adding PTP pci driver which configures
the PTP block and hooks up to RVU AF driver. It also exposes a mailbox call to adjust PTP hardware clock.

Patch 3: Patch adding PTP clock driver for PF netdev.

Quoted 6 lines

Aleksey Makarov (2):
octeontx2-af: Add support for Marvell PTP coprocessor
octeontx2-pf: Add support for PTP clock

Zyta Szpak (1):
octeontx2-af: Support to enable/disable HW timestamping