Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Hubert Feurstein <h.feurstein@gmail.com>
To:     netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Cc:     Hubert Feurstein <h.feurstein@gmail.com>,
        Andrew Lunn <andrew@lunn.ch>,
        Richard Cochran <richardcochran@gmail.com>,
        Florian Fainelli <f.fainelli@gmail.com>,
        Heiner Kallweit <hkallweit1@gmail.com>,
        Vladimir Oltean <olteanv@gmail.com>,
        "David S. Miller" <davem@davemloft.net>
Subject: [PATCH net-next v2 2/4] net: mdio: add PTP offset compensation to mdiobus_write_sts
Date:   Mon, 19 Aug 2019 19:28:25 +0200
Message-Id: <20190819172827.9550-3-hubert.feurstein@vahle.at>
In-Reply-To: <20190819172827.9550-1-hubert.feurstein@vahle.at>
References: <20190819172827.9550-1-hubert.feurstein@vahle.at>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

From: Hubert Feurstein <h.feurstein@gmail.com>

The slow MDIO access introduces quite a big offset (~13us) to the PTP
system time synchronisation. With this patch the driver has the possibility
to set the correct offset which can then be compensated.

Signed-off-by: Hubert Feurstein <h.feurstein@gmail.com>
---
 drivers/net/phy/mdio_bus.c | 12 ++++++++++++
 include/linux/phy.h        |  8 ++++++++
 2 files changed, 20 insertions(+)

diff --git a/drivers/net/phy/mdio_bus.c b/drivers/net/phy/mdio_bus.c
index 4dba2714495e..50a37cf46f96 100644
--- a/drivers/net/phy/mdio_bus.c
+++ b/drivers/net/phy/mdio_bus.c
@@ -739,6 +739,18 @@ int __mdiobus_write_sts(struct mii_bus *bus, int addr, u32 regnum, u16 val,
 	if (!(bus->flags & MII_BUS_F_PTP_STS_SUPPORTED))
 		ptp_read_system_postts(sts);
 
+	/* PTP offset compensation:
+	 * After the MDIO access is completed (from the chip perspective), the
+	 * switch chip will snapshot the PHC timestamp. To make sure our system
+	 * timestamp corresponds to the PHC timestamp, we have to add the
+	 * duration of this MDIO access to sts->post_ts. Linuxptp's phc2sys
+	 * takes the average of pre_ts and post_ts to calculate the final
+	 * system timestamp. With this in mind, we have to add ptp_sts_offset
+	 * twice to post_ts, in order to not introduce an constant time offset.
+	 */
+	if (sts)
+		timespec64_add_ns(&sts->post_ts, 2 * bus->ptp_sts_offset);
+
 	return retval;
 }
 EXPORT_SYMBOL(__mdiobus_write_sts);
diff --git a/include/linux/phy.h b/include/linux/phy.h
index 0b33662e0320..615df9c7f2c3 100644
--- a/include/linux/phy.h
+++ b/include/linux/phy.h
@@ -283,8 +283,16 @@ struct mii_bus {
 	 * The ptp_read_system_*ts functions already check the ptp_sts pointer.
 	 * The MII_BUS_F_PTP_STS_SUPPORTED-bit must be set in flags, when the
 	 * MDIO bus driver takes the timestamps as described above.
+	 *
+	 * @ptp_sts_offset: This is the compensation offset for the system
+	 * timestamp which is introduced by the slow MDIO access duration. An
+	 * MDIO access consists of 32 clock cycles. Usually the MDIO bus runs
+	 * at ~2.5MHz, so we have to compensate ~12800ns offset.
+	 * Set the ptp_sts_offset to the exact duration of one MDIO frame
+	 * (= 32 * clock-period) in nano-seconds.
 	 */
 	struct ptp_system_timestamp *ptp_sts;
+	u32 ptp_sts_offset;
 };
 
 #define to_mii_bus(d) container_of(d, struct mii_bus, dev)
-- 
2.22.1