Skip to contentSkip to conversation

bnxt_en: Bug fixes

PATCH Applied Michael Chan 4 participants 14 messages Jul 18, 2021 – Jul 20, 2021

Subsystems (3)
lore
Share & tools
lore

Michael Chan

author maintainer
#0 lore raw

cover bnxt_en: Bug fixes

Most of the fixes in this series have to do with error recovery. They include error path handling when the error recovery has to abort, and the rediscovery of capabilities (PTP and RoCE) after firmware reset that may result in capability changes.

Two other fixes are to reject invalid ETS settings and to validate VLAN protocol in the RX path.

Edwin Peer (1):
bnxt_en: reject ETS settings that will starve a TC

Kalesh AP (1):
bnxt_en: don't disable an already disabled PCI device

Michael Chan (5):
bnxt_en: Refresh RoCE capabilities in bnxt_ulp_probe()
bnxt_en: Add missing check for BNXT_STATE_ABORT_ERR in

    bnxt_fw_rset_task()

bnxt_en: Validate vlan protocol ID on RX packets
bnxt_en: Move bnxt_ptp_init() to bnxt_open()
bnxt_en: Fix PTP capability discovery

Somnath Kotur (2):
bnxt_en: fix error path of FW reset
bnxt_en: Check abort error state in bnxt_half_open_nic()

 drivers/net/ethernet/broadcom/bnxt/bnxt.c     | 85 ++++++++++++++-----
 drivers/net/ethernet/broadcom/bnxt/bnxt_dcb.c | 10 ++-
 drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c | 24 ++----
 drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.h |  1 -
 drivers/net/ethernet/broadcom/bnxt/bnxt_ulp.c |  9 +-
 5 files changed, 83 insertions(+), 46 deletions(-)
Signature · 1 line
2.18.1
author maintainer
#1 lore raw

PATCH 1/9 bnxt_en: don't disable an already disabled PCI device

From: Kalesh AP <[email protected]>

If device is already disabled in reset path and PCI io error is detected before the device could be enabled, driver could call pci_disable_device() for already disabled device. Fix this problem by calling pci_disable_device() only if the device is already enabled.

Fixes 6316ea6db93d ("bnxt_en: Enable AER support.")
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +2 -1
index f56245eeef7b..fdfb75a1608d 100644
@@ -13436,7 +13436,8 @@ static pci_ers_result_t bnxt_io_error_detected(struct pci_dev *pdev,
 	if (netif_running(netdev))
 		bnxt_close(netdev);
 
-	pci_disable_device(pdev);
+	if (pci_is_enabled(pdev))
+		pci_disable_device(pdev);
 	bnxt_free_ctx_mem(bp);
 	kfree(bp->ctx);
 	bp->ctx = NULL;
Signature · 1 line
2.18.1
author maintainer
#2 lore raw

PATCH 2/9 bnxt_en: reject ETS settings that will starve a TC

In reply to Michael Chan

From: Edwin Peer <[email protected]>

ETS proportions are presented to HWRM_QUEUE_COS2BW_CFG as minimum bandwidth constraints. Thus, zero is a legal value for a given TC. However, if all the other TCs sum up to 100%, then at least one hardware queue will starve, resulting in guaranteed TX timeouts. Reject such nonsensical configurations.

---
 drivers/net/ethernet/broadcom/bnxt/bnxt_dcb.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
drivers/net/ethernet/broadcom/bnxt/bnxt_dcb.c +9 -1
index 8e90224c43a2..8a68df4d9e59 100644
@@ -433,6 +433,7 @@ static int bnxt_hwrm_queue_dscp2pri_cfg(struct bnxt *bp, struct dcb_app *app,
 static int bnxt_ets_validate(struct bnxt *bp, struct ieee_ets *ets, u8 *tc)
 {
 	int total_ets_bw = 0;
+	bool zero = false;
 	u8 max_tc = 0;
 	int i;
 
@@ -453,13 +454,20 @@ static int bnxt_ets_validate(struct bnxt *bp, struct ieee_ets *ets, u8 *tc)
 			break;
 		case IEEE_8021QAZ_TSA_ETS:
 			total_ets_bw += ets->tc_tx_bw[i];
+			zero = zero || !ets->tc_tx_bw[i];
 			break;
 		default:
 			return -ENOTSUPP;
 		}
 	}
-	if (total_ets_bw > 100)
+	if (total_ets_bw > 100) {
+		netdev_warn(bp->dev, "rejecting ETS config exceeding available bandwidth\n");
 		return -EINVAL;
+	}
+	if (zero && total_ets_bw == 100) {
+		netdev_warn(bp->dev, "rejecting ETS config starving a TC\n");
+		return -EINVAL;
+	}
 
 	if (max_tc >= bp->max_tc)
 		*tc = bp->max_tc;
Signature · 1 line
2.18.1
author maintainer
#3 lore raw

PATCH 3/9 bnxt_en: Refresh RoCE capabilities in bnxt_ulp_probe()

In reply to Michael Chan

The capabilities can change after firmware upgrade/downgrade, so we should get the up-to-date RoCE capabilities everytime bnxt_ulp_probe() is called.

Fixes 2151fe0830fd ("bnxt_en: Handle RESET_NOTIFY async event from firmware.")
---
 drivers/net/ethernet/broadcom/bnxt/bnxt_ulp.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt_ulp.c +5 -4
index a918e374f3c5..187ff643ad2a 100644
@@ -479,16 +479,17 @@ struct bnxt_en_dev *bnxt_ulp_probe(struct net_device *dev)
 		if (!edev)
 			return ERR_PTR(-ENOMEM);
 		edev->en_ops = &bnxt_en_ops_tbl;
-		if (bp->flags & BNXT_FLAG_ROCEV1_CAP)
-			edev->flags |= BNXT_EN_FLAG_ROCEV1_CAP;
-		if (bp->flags & BNXT_FLAG_ROCEV2_CAP)
-			edev->flags |= BNXT_EN_FLAG_ROCEV2_CAP;
 		edev->net = dev;
 		edev->pdev = bp->pdev;
 		edev->l2_db_size = bp->db_size;
 		edev->l2_db_size_nc = bp->db_size;
 		bp->edev = edev;
 	}
+	edev->flags &= ~BNXT_EN_FLAG_ROCE_CAP;
+	if (bp->flags & BNXT_FLAG_ROCEV1_CAP)
+		edev->flags |= BNXT_EN_FLAG_ROCEV1_CAP;
+	if (bp->flags & BNXT_FLAG_ROCEV2_CAP)
+		edev->flags |= BNXT_EN_FLAG_ROCEV2_CAP;
 	return bp->edev;
 }
 EXPORT_SYMBOL(bnxt_ulp_probe);
Signature · 1 line
2.18.1
author maintainer
#4 lore raw

PATCH 4/9 bnxt_en: Add missing check for BNXT_STATE_ABORT_ERR in bnxt_fw_rset_task()

In reply to Michael Chan

In the BNXT_FW_RESET_STATE_POLL_VF state in bnxt_fw_reset_task() after all VFs have unregistered, we need to check for BNXT_STATE_ABORT_ERR after we acquire the rtnl_lock. If the flag is set, we need to abort.

Fixes 230d1f0de754 ("bnxt_en: Handle firmware reset.")
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 4 ++++
 1 file changed, 4 insertions(+)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +4 -0
index fdfb75a1608d..39908a3d9460 100644
@@ -11992,6 +11992,10 @@ static void bnxt_fw_reset_task(struct work_struct *work)
 		}
 		bp->fw_reset_timestamp = jiffies;
 		rtnl_lock();
+		if (test_bit(BNXT_STATE_ABORT_ERR, &bp->state)) {
+			rtnl_unlock();
+			goto fw_reset_abort;
+		}
 		bnxt_fw_reset_close(bp);
 		if (bp->fw_cap & BNXT_FW_CAP_ERR_RECOVER_RELOAD) {
 			bp->fw_reset_state = BNXT_FW_RESET_STATE_POLL_FW_DOWN;
Signature · 1 line
2.18.1
author maintainer
#5 lore raw

PATCH 5/9 bnxt_en: fix error path of FW reset

In reply to Michael Chan

From: Somnath Kotur <[email protected]>

When bnxt_open() fails in the firmware reset path, the driver needs to gracefully abort, but it is executing code that should be invoked only in the success path. Define a function to abort FW reset and consolidate all error paths to call this new function.

Fixes dab62e7c2de7 ("bnxt_en: Implement faster recovery for firmware fatal error.")
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 31 +++++++++++++++--------
 1 file changed, 21 insertions(+), 10 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +21 -10
index 39908a3d9460..f2f1136fd492 100644
@@ -11959,10 +11959,21 @@ static bool bnxt_fw_reset_timeout(struct bnxt *bp)
 			  (bp->fw_reset_max_dsecs * HZ / 10));
 }
 
+static void bnxt_fw_reset_abort(struct bnxt *bp, int rc)
+{
+	clear_bit(BNXT_STATE_IN_FW_RESET, &bp->state);
+	if (bp->fw_reset_state != BNXT_FW_RESET_STATE_POLL_VF) {
+		bnxt_ulp_start(bp, rc);
+		bnxt_dl_health_status_update(bp, false);
+	}
+	bp->fw_reset_state = 0;
+	dev_close(bp->dev);
+}
+
 static void bnxt_fw_reset_task(struct work_struct *work)
 {
 	struct bnxt *bp = container_of(work, struct bnxt, fw_reset_task.work);
-	int rc;
+	int rc = 0;
 
 	if (!test_bit(BNXT_STATE_IN_FW_RESET, &bp->state)) {
 		netdev_err(bp->dev, "bnxt_fw_reset_task() called when not in fw reset mode!\n");
@@ -11993,8 +12004,9 @@ static void bnxt_fw_reset_task(struct work_struct *work)
 		bp->fw_reset_timestamp = jiffies;
 		rtnl_lock();
 		if (test_bit(BNXT_STATE_ABORT_ERR, &bp->state)) {
+			bnxt_fw_reset_abort(bp, rc);
 			rtnl_unlock();
-			goto fw_reset_abort;
+			return;
 		}
 		bnxt_fw_reset_close(bp);
 		if (bp->fw_cap & BNXT_FW_CAP_ERR_RECOVER_RELOAD) {
@@ -12043,6 +12055,7 @@ static void bnxt_fw_reset_task(struct work_struct *work)
 			if (val == 0xffff) {
 				if (bnxt_fw_reset_timeout(bp)) {
 					netdev_err(bp->dev, "Firmware reset aborted, PCI config space invalid\n");
+					rc = -ETIMEDOUT;
 					goto fw_reset_abort;
 				}
 				bnxt_queue_fw_reset_work(bp, HZ / 1000);
@@ -12052,6 +12065,7 @@ static void bnxt_fw_reset_task(struct work_struct *work)
 		clear_bit(BNXT_STATE_FW_FATAL_COND, &bp->state);
 		if (pci_enable_device(bp->pdev)) {
 			netdev_err(bp->dev, "Cannot re-enable PCI device\n");
+			rc = -ENODEV;
 			goto fw_reset_abort;
 		}
 		pci_set_master(bp->pdev);
@@ -12078,9 +12092,10 @@ static void bnxt_fw_reset_task(struct work_struct *work)
 		}
 		rc = bnxt_open(bp->dev);
 		if (rc) {
-			netdev_err(bp->dev, "bnxt_open_nic() failed\n");
-			clear_bit(BNXT_STATE_IN_FW_RESET, &bp->state);
-			dev_close(bp->dev);
+			netdev_err(bp->dev, "bnxt_open() failed during FW reset\n");
+			bnxt_fw_reset_abort(bp, rc);
+			rtnl_unlock();
+			return;
 		}
 
 		bp->fw_reset_state = 0;
@@ -12107,12 +12122,8 @@ static void bnxt_fw_reset_task(struct work_struct *work)
 		netdev_err(bp->dev, "fw_health_status 0x%x\n", sts);
 	}
 fw_reset_abort:
-	clear_bit(BNXT_STATE_IN_FW_RESET, &bp->state);
-	if (bp->fw_reset_state != BNXT_FW_RESET_STATE_POLL_VF)
-		bnxt_dl_health_status_update(bp, false);
-	bp->fw_reset_state = 0;
 	rtnl_lock();
-	dev_close(bp->dev);
+	bnxt_fw_reset_abort(bp, rc);
 	rtnl_unlock();
 }
 
Signature · 1 line
2.18.1
author maintainer
#6 lore raw

PATCH 6/9 bnxt_en: Validate vlan protocol ID on RX packets

In reply to Michael Chan

Only pass supported VLAN protocol IDs for stripped VLAN tags to the stack. The stack will hit WARN() if the protocol ID is unsupported.

Existing firmware sets up the chip to strip 0x8100, 0x88a8, 0x9100. Only the 1st two protocols are supported by the kernel.

Fixes a196e96bb68f ("bnxt_en: clean up VLAN feature bit handling")
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 21 ++++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +16 -5
index f2f1136fd492..169f093e01de 100644
@@ -1671,11 +1671,16 @@ static inline struct sk_buff *bnxt_tpa_end(struct bnxt *bp,
 
 	if ((tpa_info->flags2 & RX_CMP_FLAGS2_META_FORMAT_VLAN) &&
 	    (skb->dev->features & BNXT_HW_FEATURE_VLAN_ALL_RX)) {
-		u16 vlan_proto = tpa_info->metadata >>
-			RX_CMP_FLAGS2_METADATA_TPID_SFT;
+		__be16 vlan_proto = htons(tpa_info->metadata >>
+					  RX_CMP_FLAGS2_METADATA_TPID_SFT);
 		u16 vtag = tpa_info->metadata & RX_CMP_FLAGS2_METADATA_TCI_MASK;
 
-		__vlan_hwaccel_put_tag(skb, htons(vlan_proto), vtag);
+		if (eth_type_vlan(vlan_proto)) {
+			__vlan_hwaccel_put_tag(skb, vlan_proto, vtag);
+		} else {
+			dev_kfree_skb(skb);
+			return NULL;
+		}
 	}
 
 	skb_checksum_none_assert(skb);
@@ -1897,9 +1902,15 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 	    (skb->dev->features & BNXT_HW_FEATURE_VLAN_ALL_RX)) {
 		u32 meta_data = le32_to_cpu(rxcmp1->rx_cmp_meta_data);
 		u16 vtag = meta_data & RX_CMP_FLAGS2_METADATA_TCI_MASK;
-		u16 vlan_proto = meta_data >> RX_CMP_FLAGS2_METADATA_TPID_SFT;
+		__be16 vlan_proto = htons(meta_data >>
+					  RX_CMP_FLAGS2_METADATA_TPID_SFT);
 
-		__vlan_hwaccel_put_tag(skb, htons(vlan_proto), vtag);
+		if (eth_type_vlan(vlan_proto)) {
+			__vlan_hwaccel_put_tag(skb, vlan_proto, vtag);
+		} else {
+			dev_kfree_skb(skb);
+			goto next_rx;
+		}
 	}
 
 	skb_checksum_none_assert(skb);
Signature · 1 line
2.18.1
author maintainer
#7 lore raw

PATCH 7/9 bnxt_en: Check abort error state in bnxt_half_open_nic()

In reply to Michael Chan

From: Somnath Kotur <[email protected]>

bnxt_half_open_nic() is called during during ethtool self test and is protected by rtnl_lock. Firmware reset can be happening at the same time. Only critical portions of the entire firmware reset sequence are protected by the rtnl_lock. It is possible that bnxt_half_open_nic() can be called when the firmware reset sequence is aborting. In that case, bnxt_half_open_nic() needs to check if the ABORT_ERR flag is set and abort if it is. The ethtool self test will fail but the NIC will be brought to a consistent IF_DOWN state.

Without this patch, if bnxt_half_open_nic() were to continue in this error state, it may crash like this:

bnxt_en 0000:82:00.1 enp130s0f1np1: FW reset in progress during close, FW reset will be aborted

  Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
  ...
  Process ethtool (pid: 333327, stack limit = 0x0000000046476577)
  Call trace:
  bnxt_alloc_mem+0x444/0xef0 [bnxt_en]
  bnxt_half_open_nic+0x24/0xb8 [bnxt_en]
  bnxt_self_test+0x2dc/0x390 [bnxt_en]
  ethtool_self_test+0xe0/0x1f8
  dev_ethtool+0x1744/0x22d0
  dev_ioctl+0x190/0x3e0
  sock_ioctl+0x238/0x480
  do_vfs_ioctl+0xc4/0x758
  ksys_ioctl+0x84/0xb8
  __arm64_sys_ioctl+0x28/0x38
  el0_svc_handler+0xb0/0x180
  el0_svc+0x8/0xc
Fixes a1301f08c5ac ("bnxt_en: Check abort error state in bnxt_open_nic().")
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 6 ++++++
 1 file changed, 6 insertions(+)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +6 -0
index 169f093e01de..31eb3c00851a 100644
@@ -10208,6 +10208,12 @@ int bnxt_half_open_nic(struct bnxt *bp)
 {
 	int rc = 0;
 
+	if (test_bit(BNXT_STATE_ABORT_ERR, &bp->state)) {
+		netdev_err(bp->dev, "A previous firmware reset has not completed, aborting half open\n");
+		rc = -ENODEV;
+		goto half_open_err;
+	}
+
 	rc = bnxt_alloc_mem(bp, false);
 	if (rc) {
 		netdev_err(bp->dev, "bnxt_alloc_mem err: %x\n", rc);
Signature · 1 line
2.18.1
author maintainer
#8 lore raw

PATCH 8/9 bnxt_en: Move bnxt_ptp_init() to bnxt_open()

In reply to Michael Chan

The device needs to be in ifup state for PTP to function, so move bnxt_ptp_init() to bnxt_open(). This means that the PHC will be registered during bnxt_open().

This also makes firmware reset work correctly. PTP configurations may change after firmware upgrade or downgrade. bnxt_open() will be called after firmware reset, so it will work properly.

bnxt_ptp_start() is now incorporated into bnxt_ptp_init(). We now also need to call bnxt_ptp_clear() in bnxt_close().

Fixes 93cb62d98e9c ("bnxt_en: Enable hardware PTP support")
Cc Richard Cochran <[email protected]>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c     | 16 +++++++------
 drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c | 24 ++++++-------------
 drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.h |  1 -
 3 files changed, 16 insertions(+), 25 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +9 -7
index 31eb3c00851a..b8b73c210995 100644
@@ -10134,7 +10134,6 @@ static int __bnxt_open_nic(struct bnxt *bp, bool irq_re_init, bool link_re_init)
 		}
 	}
 
-	bnxt_ptp_start(bp);
 	rc = bnxt_init_nic(bp, irq_re_init);
 	if (rc) {
 		netdev_err(bp->dev, "bnxt_init_nic err: %x\n", rc);
@@ -10273,9 +10272,16 @@ static int bnxt_open(struct net_device *dev)
 	rc = bnxt_hwrm_if_change(bp, true);
 	if (rc)
 		return rc;
+
+	if (bnxt_ptp_init(bp)) {
+		netdev_warn(dev, "PTP initialization failed.\n");
+		kfree(bp->ptp_cfg);
+		bp->ptp_cfg = NULL;
+	}
 	rc = __bnxt_open_nic(bp, true, true);
 	if (rc) {
 		bnxt_hwrm_if_change(bp, false);
+		bnxt_ptp_clear(bp);
 	} else {
 		if (test_and_clear_bit(BNXT_STATE_FW_RESET_DET, &bp->state)) {
 			if (!test_bit(BNXT_STATE_IN_FW_RESET, &bp->state)) {
@@ -10366,6 +10372,7 @@ static int bnxt_close(struct net_device *dev)
 {
 	struct bnxt *bp = netdev_priv(dev);
 
+	bnxt_ptp_clear(bp);
 	bnxt_hwmon_close(bp);
 	bnxt_close_nic(bp, true, true);
 	bnxt_hwrm_shutdown_link(bp);
@@ -11352,6 +11359,7 @@ static void bnxt_fw_reset_close(struct bnxt *bp)
 		bnxt_clear_int_mode(bp);
 		pci_disable_device(bp->pdev);
 	}
+	bnxt_ptp_clear(bp);
 	__bnxt_close_nic(bp, true, false);
 	bnxt_vf_reps_free(bp);
 	bnxt_clear_int_mode(bp);
@@ -12694,7 +12702,6 @@ static void bnxt_remove_one(struct pci_dev *pdev)
 	if (BNXT_PF(bp))
 		devlink_port_type_clear(&bp->dl_port);
 
-	bnxt_ptp_clear(bp);
 	pci_disable_pcie_error_reporting(pdev);
 	unregister_netdev(dev);
 	clear_bit(BNXT_STATE_IN_FW_RESET, &bp->state);
@@ -13278,11 +13285,6 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 				   rc);
 	}
 
-	if (bnxt_ptp_init(bp)) {
-		netdev_warn(dev, "PTP initialization failed.\n");
-		kfree(bp->ptp_cfg);
-		bp->ptp_cfg = NULL;
-	}
 	bnxt_inv_fw_health_reg(bp);
 	bnxt_dl_register(bp);
 
drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.c +7 -17
index f698b6bd4ff8..9089e7f3fbd4 100644
@@ -385,22 +385,6 @@ int bnxt_get_rx_ts_p5(struct bnxt *bp, u64 *ts, u32 pkt_ts)
 	return 0;
 }
 
-void bnxt_ptp_start(struct bnxt *bp)
-{
-	struct bnxt_ptp_cfg *ptp = bp->ptp_cfg;
-
-	if (!ptp)
-		return;
-
-	if (bp->flags & BNXT_FLAG_CHIP_P5) {
-		spin_lock_bh(&ptp->ptp_lock);
-		ptp->current_time = bnxt_refclk_read(bp, NULL);
-		WRITE_ONCE(ptp->old_time, ptp->current_time);
-		spin_unlock_bh(&ptp->ptp_lock);
-		ptp_schedule_worker(ptp->ptp_clock, 0);
-	}
-}
-
 static const struct ptp_clock_info bnxt_ptp_caps = {
 	.owner		= THIS_MODULE,
 	.name		= "bnxt clock",
@@ -450,7 +434,13 @@ int bnxt_ptp_init(struct bnxt *bp)
 		bnxt_unmap_ptp_regs(bp);
 		return err;
 	}
-
+	if (bp->flags & BNXT_FLAG_CHIP_P5) {
+		spin_lock_bh(&ptp->ptp_lock);
+		ptp->current_time = bnxt_refclk_read(bp, NULL);
+		WRITE_ONCE(ptp->old_time, ptp->current_time);
+		spin_unlock_bh(&ptp->ptp_lock);
+		ptp_schedule_worker(ptp->ptp_clock, 0);
+	}
 	return 0;
 }
 
drivers/net/ethernet/broadcom/bnxt/bnxt_ptp.h +0 -1
index 6b6245750e20..4135ea3ec788 100644
@@ -75,7 +75,6 @@ int bnxt_hwtstamp_set(struct net_device *dev, struct ifreq *ifr);
 int bnxt_hwtstamp_get(struct net_device *dev, struct ifreq *ifr);
 int bnxt_get_tx_ts_p5(struct bnxt *bp, struct sk_buff *skb);
 int bnxt_get_rx_ts_p5(struct bnxt *bp, u64 *ts, u32 pkt_ts);
-void bnxt_ptp_start(struct bnxt *bp);
 int bnxt_ptp_init(struct bnxt *bp);
 void bnxt_ptp_clear(struct bnxt *bp);
 #endif
Signature · 1 line
2.18.1
maintainer
#9 lore raw

Re: [PATCH net 8/9] bnxt_en: Move bnxt_ptp_init() to bnxt_open()

Quoted 3 lines · Michael Chan

The device needs to be in ifup state for PTP to function, so move bnxt_ptp_init() to bnxt_open(). This means that the PHC will be registered during bnxt_open().

I think it's an anti-pattern to have the clock registered only when the device is up. Right, Richard?

IIRC Intel parts did it in the past because they had the clock hooked up to the MAC/PHY so the clock was not actually ticking. But seems like a wrong trade off to unreg PTP for SW convenience. Or maybe I'm biased against the live FW reset :) Let's see if Richard agrees.

Quoted 6 lines

This also makes firmware reset work correctly. PTP configurations may change after firmware upgrade or downgrade. bnxt_open() will be called after firmware reset, so it will work properly.

bnxt_ptp_start() is now incorporated into bnxt_ptp_init(). We now also need to call bnxt_ptp_clear() in bnxt_close().

#10 lore raw

Re: [PATCH net 8/9] bnxt_en: Move bnxt_ptp_init() to bnxt_open()

Quoted 7 lines · Jakub Kicinski
Quoted 3 lines · Michael Chan

The device needs to be in ifup state for PTP to function, so move bnxt_ptp_init() to bnxt_open(). This means that the PHC will be registered during bnxt_open().

I think it's an anti-pattern to have the clock registered only when the device is up. Right, Richard?

Yes, indeed.

Quoted 4 lines

IIRC Intel parts did it in the past because they had the clock hooked up to the MAC/PHY so the clock was not actually ticking. But seems like a wrong trade off to unreg PTP for SW convenience. Or maybe I'm biased against the live FW reset :) Let's see if Richard agrees.

Totally agree!

Ideally the PHC appears as soon as the driver instantiates for a HW device. Telling time from the clock is independent from the network interface being up or down.

Some drivers are lazy and fail to decouple these two orthogonal issues. In some (all?) cases, there is no HW limitation involved, just sloppy driver work.

Thanks,
Richard

author maintainer
#11 lore raw

Re: [PATCH net 8/9] bnxt_en: Move bnxt_ptp_init() to bnxt_open()

Quoted 4 lines · Richard Cochran

Ideally the PHC appears as soon as the driver instantiates for a HW device. Telling time from the clock is independent from the network interface being up or down.

OK. Will improve this in follow on patches to keep the PHC registered unless the firmware after reset no longer supports PTP. Thanks.

author maintainer
#12 lore raw

PATCH 9/9 bnxt_en: Fix PTP capability discovery

In reply to Michael Chan

The current PTP initialization logic does not account for firmware reset that may cause PTP capability to change. The valid pointer bp->ptp_cfg is used to indicate that the device is capable of PTP and that it has been initialized. So we must clean up bp->ptp_cfg and free it if the firmware after reset does not support PTP.

Fixes 93cb62d98e9c ("bnxt_en: Enable hardware PTP support")
Cc Richard Cochran <[email protected]>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 6 +++++-
 1 file changed, 5 insertions(+), 1 deletion(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +5 -1
index b8b73c210995..4db162cee911 100644
@@ -7574,8 +7574,12 @@ static int __bnxt_hwrm_func_qcaps(struct bnxt *bp)
 		bp->flags &= ~BNXT_FLAG_WOL_CAP;
 		if (flags & FUNC_QCAPS_RESP_FLAGS_WOL_MAGICPKT_SUPPORTED)
 			bp->flags |= BNXT_FLAG_WOL_CAP;
-		if (flags & FUNC_QCAPS_RESP_FLAGS_PTP_SUPPORTED)
+		if (flags & FUNC_QCAPS_RESP_FLAGS_PTP_SUPPORTED) {
 			__bnxt_hwrm_ptp_qcfg(bp);
+		} else {
+			kfree(bp->ptp_cfg);
+			bp->ptp_cfg = NULL;
+		}
 	} else {
 #ifdef CONFIG_BNXT_SRIOV
 		struct bnxt_vf_info *vf = &bp->vf;
Signature · 1 line
2.18.1
#13 lore raw
In reply to Michael Chan
This series was applied to netdev/net.git (refs/heads/master):

Hello:

This series was applied to netdev/net.git (refs/heads/master):

Quoted 9 lines · On Sun, 18 Jul 2021 15:36:24 -0400 you wrote:

Most of the fixes in this series have to do with error recovery. They include error path handling when the error recovery has to abort, and the rediscovery of capabilities (PTP and RoCE) after firmware reset that may result in capability changes.

Two other fixes are to reject invalid ETS settings and to validate VLAN protocol in the RX path.

[...]

Here is the summary with links:
- [net,1/9] bnxt_en: don't disable an already disabled PCI device

    https://git.kernel.org/netdev/net/c/c81cfb6256d9
  - [net,2/9] bnxt_en: reject ETS settings that will starve a TC
    https://git.kernel.org/netdev/net/c/c08c59653415
  - [net,3/9] bnxt_en: Refresh RoCE capabilities in bnxt_ulp_probe()
    https://git.kernel.org/netdev/net/c/2c9f046bc377
  - [net,4/9] bnxt_en: Add missing check for BNXT_STATE_ABORT_ERR in bnxt_fw_rset_task()
    https://git.kernel.org/netdev/net/c/6cd657cb3ee6
  - [net,5/9] bnxt_en: fix error path of FW reset
    https://git.kernel.org/netdev/net/c/3958b1da725a
  - [net,6/9] bnxt_en: Validate vlan protocol ID on RX packets
    https://git.kernel.org/netdev/net/c/96bdd4b9ea7e
  - [net,7/9] bnxt_en: Check abort error state in bnxt_half_open_nic()
    https://git.kernel.org/netdev/net/c/11a39259ff79
  - [net,8/9] bnxt_en: Move bnxt_ptp_init() to bnxt_open()
    https://git.kernel.org/netdev/net/c/d7859afb6880
  - [net,9/9] bnxt_en: Fix PTP capability discovery
    https://git.kernel.org/netdev/net/c/de5bf19414fe

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