Intel-Wired-Lan Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Jacob Keller <jacob.e.keller@intel.com>
To: Intel Wired LAN <intel-wired-lan@lists.osuosl.org>
Subject: [Intel-wired-lan] [PATCH net-next 07/15] ice: check Tx timestamp memory register for ready timestamps
Date: Tue,  1 Nov 2022 15:52:32 -0700	[thread overview]
Message-ID: <20221101225240.421525-8-jacob.e.keller@intel.com> (raw)
In-Reply-To: <20221101225240.421525-1-jacob.e.keller@intel.com>

The PHY for e822 based hardware has a register which indicates which
timestamps are valid in the PHY timestamp memory block. Each bit in the
register indicates whether the associated index in the timestamp memory is
valid.

Hardware sets this bit when the timestamp is captured, and clears the bit
when the timestamp is read. Use of this register is important as reading
timestamp registers can impact the way that hardware generates timestamp
interrupts.

This occurs because the PHY has an internal value which is incremented
when hardware captures a timestamp and decremented when software reads a
timestamp. Reading timestamps which are not marked as valid still decrement
the internal value and can result in the Tx timestamp interrupt not
triggering in the future.

To prevent this, use the timestamp memory value to determine which
timestamps are ready to be read. The ice_get_phy_tx_tstamp_ready function
reads this value. For e810 devices, this just always returns with all bits
set.

Skip any timestamp which is not set in this bitmap, avoiding reading extra
timestamps on e822 devices.

Modify the stale check to apply only to e810 devices. It is not necessary
for e822 devices because the timestamp memory register will prevent us from
reading a stale timestamp.

Modify the ICE_PTP_TS_VALID check to simply drop the timestamp immediately
so that in an event of getting such an invalid timestamp the driver does
not attempt to re-read the timestamp again in a future poll of the
register.

With these changes, the driver now reads each timestamp register exactly
once, and does not attempt any re-reads. This ensures the interrupt
tracking logic in the PHY will not get stuck.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
---
 drivers/net/ethernet/intel/ice/ice_ptp.c    | 41 ++++++++++--
 drivers/net/ethernet/intel/ice/ice_ptp_hw.c | 72 +++++++++++++++++++++
 drivers/net/ethernet/intel/ice/ice_ptp_hw.h |  1 +
 3 files changed, 108 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/intel/ice/ice_ptp.c b/drivers/net/ethernet/intel/ice/ice_ptp.c
index ebe910326963..39c68a272c6a 100644
--- a/drivers/net/ethernet/intel/ice/ice_ptp.c
+++ b/drivers/net/ethernet/intel/ice/ice_ptp.c
@@ -655,6 +655,9 @@ static bool ice_ptp_tx_tstamp(struct ice_ptp_tx *tx)
 	struct ice_ptp_port *ptp_port;
 	bool ts_handled = true;
 	struct ice_pf *pf;
+	struct ice_hw *hw;
+	u64 tstamp_ready;
+	int err;
 	u8 idx;
 
 	if (!tx->init)
@@ -662,6 +665,12 @@ static bool ice_ptp_tx_tstamp(struct ice_ptp_tx *tx)
 
 	ptp_port = container_of(tx, struct ice_ptp_port, tx);
 	pf = ptp_port_to_pf(ptp_port);
+	hw = &pf->hw;
+
+	/* Read the Tx ready status first */
+	err = ice_get_phy_tx_tstamp_ready(hw, tx->block, &tstamp_ready);
+	if (err)
+		return false;
 
 	for_each_set_bit(idx, tx->in_use, tx->len) {
 		struct skb_shared_hwtstamps shhwtstamps = {};
@@ -669,7 +678,6 @@ static bool ice_ptp_tx_tstamp(struct ice_ptp_tx *tx)
 		u64 raw_tstamp, tstamp;
 		bool drop_ts = false;
 		struct sk_buff *skb;
-		int err;
 
 		/* Drop packets which have waited for more than 2 seconds */
 		if (time_is_before_jiffies(tx->tstamps[idx].start + 2 * HZ)) {
@@ -677,24 +685,45 @@ static bool ice_ptp_tx_tstamp(struct ice_ptp_tx *tx)
 
 			/* Count the number of Tx timestamps that timed out */
 			pf->ptp.tx_hwtstamp_timeouts++;
+		}
 
-			goto skip_ts_read;
+		/* Only read a timestamp from the PHY if its marked as ready
+		 * by the tstamp_ready register. This avoids unnecessary
+		 * reading of timestamps which are not yet valid. This is
+		 * important as we must read all timestamps which are valid
+		 * and only timestamps which are valid during each interrupt.
+		 * If we do not, the hardware logic for generating a new
+		 * interrupt can get stuck on some devices.
+		 */
+		if (!(tstamp_ready & BIT_ULL(phy_idx))) {
+			if (drop_ts)
+				goto skip_ts_read;
+			else
+				continue;
 		}
 
 		ice_trace(tx_tstamp_fw_req, tx->tstamps[idx].skb, idx);
 
-		err = ice_read_phy_tstamp(&pf->hw, tx->block, phy_idx,
-					  &raw_tstamp);
+		err = ice_read_phy_tstamp(hw, tx->block, phy_idx, &raw_tstamp);
 		if (err)
 			continue;
 
 		ice_trace(tx_tstamp_fw_done, tx->tstamps[idx].skb, idx);
 
-		/* Check if the timestamp is invalid or stale */
-		if (!(raw_tstamp & ICE_PTP_TS_VALID) ||
+		/* For e810 hardware, the tstamp_ready bitmask does not
+		 * indicate whether a timestamp is ready. Instead, we check to
+		 * make sure the timestamp is different from the previous
+		 * cached value. If it is not, we need to re-read the
+		 * timestamp later until we get a valid value.
+		 */
+		if (!drop_ts && ice_is_e810(hw) &&
 		    raw_tstamp == tx->tstamps[idx].cached_tstamp)
 			continue;
 
+		/* Discard any timestamp value without the valid bit set */
+		if (!(raw_tstamp & ICE_PTP_TS_VALID))
+			drop_ts = true;
+
 skip_ts_read:
 		spin_lock(&tx->lock);
 		tx->tstamps[idx].cached_tstamp = raw_tstamp;
diff --git a/drivers/net/ethernet/intel/ice/ice_ptp_hw.c b/drivers/net/ethernet/intel/ice/ice_ptp_hw.c
index 6c149b88c235..55bbe76ce0bd 100644
--- a/drivers/net/ethernet/intel/ice/ice_ptp_hw.c
+++ b/drivers/net/ethernet/intel/ice/ice_ptp_hw.c
@@ -2417,6 +2417,43 @@ int ice_phy_calc_vernier_e822(struct ice_hw *hw, u8 port)
 	return 0;
 }
 
+/**
+ * ice_get_phy_tx_tstamp_ready_e822 - Read Tx memory status register
+ * @hw: pointer to the HW struct
+ * @quad: the timestamp quad to read from
+ * @tstamp_ready: contents of the Tx memory status register
+ *
+ * Read the Q_REG_TX_MEMORY_STATUS register indicating which timestamps in
+ * the PHY are ready. A set bit means the corresponding timestamp is valid and
+ * ready to be captured from the PHY timestamp block.
+ */
+static int
+ice_get_phy_tx_tstamp_ready_e822(struct ice_hw *hw, u8 quad, u64 *tstamp_ready)
+{
+	u32 hi, lo;
+	int err;
+
+	err = ice_read_quad_reg_e822(hw, quad, Q_REG_TX_MEMORY_STATUS_U,
+					&hi);
+	if (err) {
+		ice_debug(hw, ICE_DBG_PTP, "Failed to read TX_MEMORY_STATUS_U for quad %u, err %d\n",
+			  quad, err);
+		return err;
+	}
+
+	err = ice_read_quad_reg_e822(hw, quad, Q_REG_TX_MEMORY_STATUS_L,
+					&lo);
+	if (err) {
+		ice_debug(hw, ICE_DBG_PTP, "Failed to read TX_MEMORY_STATUS_L for quad %u, err %d\n",
+			  quad, err);
+		return err;
+	}
+
+	*tstamp_ready = (u64)hi << 32 | (u64)lo;
+
+	return 0;
+}
+
 /* E810 functions
  *
  * The following functions operate on the E810 series devices which use
@@ -3091,6 +3128,21 @@ int ice_clear_phy_tstamp(struct ice_hw *hw, u8 block, u8 idx)
 		return ice_clear_phy_tstamp_e822(hw, block, idx);
 }
 
+/* ice_get_phy_tx_tstamp_ready_e810 - Read Tx memory status register
+ * @hw: pointer to the HW struct
+ * @port: the PHY port to read
+ * @tstamp_ready: contents of the Tx memory status register
+ *
+ * E810 devices do not use a Tx memory status register. Instead simply
+ * indicate that all timestamps are currently ready.
+ */
+static int
+ice_get_phy_tx_tstamp_ready_e810(struct ice_hw *hw, u8 port, u64 *tstamp_ready)
+{
+	*tstamp_ready = 0xFFFFFFFFFFFFFFFF;
+	return 0;
+}
+
 /* E810T SMA functions
  *
  * The following functions operate specifically on E810T hardware and are used
@@ -3306,3 +3358,23 @@ int ice_ptp_init_phc(struct ice_hw *hw)
 	else
 		return ice_ptp_init_phc_e822(hw);
 }
+
+/* ice_get_phy_tx_tstamp_ready - Read PHY Tx memory status indication
+ * @hw: pointer to the HW struct
+ * @block: the timestamp block to check
+ * @tstamp_ready: storage for the PHY Tx memory status information
+ *
+ * Check the PHY for Tx timestamp memory status. This reports a 64 bit value
+ * which indicates which timestamps in the block may be captured. A set bit
+ * means the timestamp can be read. An unset bit means the timestamp is not
+ * ready and software should avoid reading the register.
+ */
+int ice_get_phy_tx_tstamp_ready(struct ice_hw *hw, u8 block, u64 *tstamp_ready)
+{
+	if (ice_is_e810(hw))
+		return ice_get_phy_tx_tstamp_ready_e810(hw, block,
+							tstamp_ready);
+	else
+		return ice_get_phy_tx_tstamp_ready_e822(hw, block,
+							tstamp_ready);
+}
diff --git a/drivers/net/ethernet/intel/ice/ice_ptp_hw.h b/drivers/net/ethernet/intel/ice/ice_ptp_hw.h
index b0cd73aaac6b..b781dadf5a39 100644
--- a/drivers/net/ethernet/intel/ice/ice_ptp_hw.h
+++ b/drivers/net/ethernet/intel/ice/ice_ptp_hw.h
@@ -135,6 +135,7 @@ int ice_read_phy_tstamp(struct ice_hw *hw, u8 block, u8 idx, u64 *tstamp);
 int ice_clear_phy_tstamp(struct ice_hw *hw, u8 block, u8 idx);
 void ice_ptp_reset_ts_memory(struct ice_hw *hw);
 int ice_ptp_init_phc(struct ice_hw *hw);
+int ice_get_phy_tx_tstamp_ready(struct ice_hw *hw, u8 block, u64 *tstamp_ready);
 
 /* E822 family functions */
 int ice_read_phy_reg_e822(struct ice_hw *hw, u8 port, u16 offset, u32 *val);
-- 
2.38.0.83.gd420dda05763

_______________________________________________
Intel-wired-lan mailing list
Intel-wired-lan@osuosl.org
https://lists.osuosl.org/mailman/listinfo/intel-wired-lan

  parent reply	other threads:[~2022-11-01 22:54 UTC|newest]

Thread overview: 31+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-01 22:52 [Intel-wired-lan] [PATCH net-next 00/15] ice: improve Tx timestamp corner cases Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 01/15] ice: Use more generic names for ice_ptp_tx fields Jacob Keller
2022-11-03 20:44   ` Tony Nguyen
2022-11-03 20:50     ` Temerkhanov, Sergey
2022-11-03 20:54       ` Jacob Keller
2022-11-03 20:53     ` Jacob Keller
2022-11-17 10:40   ` G, GurucharanX
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 02/15] ice: Remove the E822 vernier "bypass" logic Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 03/15] ice: Reset TS memory for all quads Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 04/15] ice: fix misuse of "link err" with "link status" Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 05/15] ice: always call ice_ptp_link_change and make it void Jacob Keller
2022-11-02  0:01   ` Ertman, David M
2022-11-03 20:44   ` Tony Nguyen
2022-11-03 20:48     ` Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 06/15] ice: handle discarding old Tx requests in ice_ptp_tx_tstamp Jacob Keller
2022-11-01 22:52 ` Jacob Keller [this message]
2022-11-03 19:23   ` [Intel-wired-lan] [PATCH net-next 07/15] ice: check Tx timestamp memory register for ready timestamps Jacob Keller
2022-11-03 20:44   ` Tony Nguyen
2022-11-03 20:52     ` Jacob Keller
2022-11-03 20:45   ` Stillwell Jr, Paul M
2022-11-03 20:52     ` Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 08/15] ice: protect init and calibrating fields with spinlock Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 09/15] ice: return true if Tx tracker is uninitialized Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 10/15] ice: disable Tx timestamps while link is down Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 11/15] ice: cleanup allocations in ice_ptp_alloc_tx_tracker Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 12/15] ice: handle flushing stale Tx timestamps in ice_ptp_tx_tstamp Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 13/15] ice: only check set bits in ice_ptp_flush_tx_tracker Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 14/15] ice: make Tx and Rx vernier offset calibration independent Jacob Keller
2022-11-03 20:45   ` Tony Nguyen
2022-11-03 20:52     ` Jacob Keller
2022-11-01 22:52 ` [Intel-wired-lan] [PATCH net-next 15/15] ice: reschedule ice_ptp_wait_for_offset_valid during reset Jacob Keller

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20221101225240.421525-8-jacob.e.keller@intel.com \
    --to=jacob.e.keller@intel.com \
    --cc=intel-wired-lan@lists.osuosl.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox