Skip to contentSkip to conversation

i40e/i40evf: Enable NVMUpdate to retrieve AdminQ and add preservation flags for NVM update

PATCH New Alice Michael 1 participant 9 messages Dec 15, 2017

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/9 i40e/i40evf: Enable NVMUpdate to retrieve AdminQ and add preservation flags for NVM update

From: Pawel Jablonski <[email protected]>

This patch adds new I40E_NVMUPD_GET_AQ_EVENT state to allow retrieval of AdminQ events as a result of AdminQ commands sent to firmware.

Add preservation flags support on X722 devices for NVM update AdminQ function wrapper. Add new parameter and handling to nvm update admin queue function intended to allow nvmupdate tool to configure the preservation flags in the AdminQ command.

This is required to implement FlatNVM on X722 devices.

---
 drivers/net/ethernet/intel/i40e/i40e_adminq.c      |   2 +-
 drivers/net/ethernet/intel/i40e/i40e_adminq_cmd.h  |   8 +-
 drivers/net/ethernet/intel/i40e/i40e_common.c      |  13 +-
 drivers/net/ethernet/intel/i40e/i40e_main.c        |   2 +-
 drivers/net/ethernet/intel/i40e/i40e_nvm.c         | 141 ++++++++++++++++-----
 drivers/net/ethernet/intel/i40e/i40e_prototype.h   |   6 +-
 drivers/net/ethernet/intel/i40e/i40e_type.h        |  26 ++--
 .../net/ethernet/intel/i40evf/i40e_adminq_cmd.h    |   8 +-
 drivers/net/ethernet/intel/i40evf/i40e_type.h      |  26 ++--
 9 files changed, 173 insertions(+), 59 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_adminq.c +1 -1
index 9af7425..d9670cd 100644
@@ -1027,7 +1027,7 @@ i40e_status i40e_clean_arq_element(struct i40e_hw *hw,
 	hw->aq.arq.next_to_clean = ntc;
 	hw->aq.arq.next_to_use = ntu;
 
-	i40e_nvmupd_check_wait_event(hw, le16_to_cpu(e->desc.opcode));
+	i40e_nvmupd_check_wait_event(hw, le16_to_cpu(e->desc.opcode), &e->desc);
 clean_arq_element_out:
 	/* Set pending if needed, unlock and return */
 	if (pending)
drivers/net/ethernet/intel/i40e/i40e_adminq_cmd.h +6 -2
index c577634..0d471b0 100644
@@ -2231,8 +2231,12 @@ I40E_CHECK_CMD_LENGTH(i40e_aqc_phy_register_access);
  */
 struct i40e_aqc_nvm_update {
 	u8	command_flags;
-#define I40E_AQ_NVM_LAST_CMD	0x01
-#define I40E_AQ_NVM_FLASH_ONLY	0x80
+#define I40E_AQ_NVM_LAST_CMD			0x01
+#define I40E_AQ_NVM_FLASH_ONLY			0x80
+#define I40E_AQ_NVM_PRESERVATION_FLAGS_SHIFT	1
+#define I40E_AQ_NVM_PRESERVATION_FLAGS_MASK	0x03
+#define I40E_AQ_NVM_PRESERVATION_FLAGS_SELECTED	0x03
+#define I40E_AQ_NVM_PRESERVATION_FLAGS_ALL	0x01
 	u8	module_pointer;
 	__le16	length;
 	__le32	offset;
drivers/net/ethernet/intel/i40e/i40e_common.c +12 -1
index 40c5f76..cf929cc 100644
@@ -3465,13 +3465,14 @@ i40e_status i40e_aq_discover_capabilities(struct i40e_hw *hw,
  * @length: length of the section to be written (in bytes from the offset)
  * @data: command buffer (size [bytes] = length)
  * @last_command: tells if this is the last command in a series
+ * @preservation_flags: Preservation mode flags
  * @cmd_details: pointer to command details structure or NULL
  *
  * Update the NVM using the admin queue commands
  **/
 i40e_status i40e_aq_update_nvm(struct i40e_hw *hw, u8 module_pointer,
 			       u32 offset, u16 length, void *data,
-			       bool last_command,
+				bool last_command, u8 preservation_flags,
 			       struct i40e_asq_cmd_details *cmd_details)
 {
 	struct i40e_aq_desc desc;
@@ -3490,6 +3491,16 @@ i40e_status i40e_aq_update_nvm(struct i40e_hw *hw, u8 module_pointer,
 	/* If this is the last command in a series, set the proper flag. */
 	if (last_command)
 		cmd->command_flags |= I40E_AQ_NVM_LAST_CMD;
+	if (hw->mac.type == I40E_MAC_X722) {
+		if (preservation_flags == I40E_NVM_PRESERVATION_FLAGS_SELECTED)
+			cmd->command_flags |=
+				(I40E_AQ_NVM_PRESERVATION_FLAGS_SELECTED <<
+				 I40E_AQ_NVM_PRESERVATION_FLAGS_SHIFT);
+		else if (preservation_flags == I40E_NVM_PRESERVATION_FLAGS_ALL)
+			cmd->command_flags |=
+				(I40E_AQ_NVM_PRESERVATION_FLAGS_ALL <<
+				 I40E_AQ_NVM_PRESERVATION_FLAGS_SHIFT);
+	}
 	cmd->module_pointer = module_pointer;
 	cmd->offset = cpu_to_le32(offset);
 	cmd->length = cpu_to_le16(length);
drivers/net/ethernet/intel/i40e/i40e_main.c +1 -1
index e63a48f..4f74156 100644
@@ -11028,7 +11028,7 @@ i40e_status i40e_commit_partition_bw_setting(struct i40e_pf *pf)
 	ret = i40e_aq_update_nvm(&pf->hw,
 				 I40E_SR_NVM_CONTROL_WORD,
 				 0x10, sizeof(nvm_word),
-				 &nvm_word, true, NULL);
+				 &nvm_word, true, 0, NULL);
 	/* Save off last admin queue command status before releasing
 	 * the NVM
 	 */
drivers/net/ethernet/intel/i40e/i40e_nvm.c +109 -32

250 lines not on this page · raw message

drivers/net/ethernet/intel/i40e/i40e_prototype.h +4 -2
index b3cc89c..187dd53 100644
@@ -214,7 +214,7 @@ i40e_status i40e_aq_discover_capabilities(struct i40e_hw *hw,
 				struct i40e_asq_cmd_details *cmd_details);
 i40e_status i40e_aq_update_nvm(struct i40e_hw *hw, u8 module_pointer,
 				u32 offset, u16 length, void *data,
-				bool last_command,
+				bool last_command, u8 preservation_flags,
 				struct i40e_asq_cmd_details *cmd_details);
 i40e_status i40e_aq_get_lldp_mib(struct i40e_hw *hw, u8 bridge_type,
 				u8 mib_type, void *buff, u16 buff_size,
@@ -333,7 +333,9 @@ i40e_status i40e_validate_nvm_checksum(struct i40e_hw *hw,
 i40e_status i40e_nvmupd_command(struct i40e_hw *hw,
 				struct i40e_nvm_access *cmd,
 				u8 *bytes, int *);
-void i40e_nvmupd_check_wait_event(struct i40e_hw *hw, u16 opcode);
+void i40e_nvmupd_check_wait_event(struct i40e_hw *hw, u16 opcode,
+				  struct i40e_aq_desc *desc);
+void i40e_nvmupd_clear_wait_state(struct i40e_hw *hw);
 void i40e_set_pci_config_data(struct i40e_hw *hw, u16 link_status);
 
 extern struct i40e_rx_ptype_decoded i40e_ptype_lookup[];
drivers/net/ethernet/intel/i40e/i40e_type.h +17 -9
index 5a708c3..cd294e6 100644
@@ -402,6 +402,7 @@ enum i40e_nvmupd_cmd {
 	I40E_NVMUPD_STATUS,
 	I40E_NVMUPD_EXEC_AQ,
 	I40E_NVMUPD_GET_AQ_RESULT,
+	I40E_NVMUPD_GET_AQ_EVENT,
 };
 
 enum i40e_nvmupd_state {
@@ -421,15 +422,21 @@ enum i40e_nvmupd_state {
 
 #define I40E_NVM_MOD_PNT_MASK 0xFF
 
-#define I40E_NVM_TRANS_SHIFT	8
-#define I40E_NVM_TRANS_MASK	(0xf << I40E_NVM_TRANS_SHIFT)
-#define I40E_NVM_CON		0x0
-#define I40E_NVM_SNT		0x1
-#define I40E_NVM_LCB		0x2
-#define I40E_NVM_SA		(I40E_NVM_SNT | I40E_NVM_LCB)
-#define I40E_NVM_ERA		0x4
-#define I40E_NVM_CSUM		0x8
-#define I40E_NVM_EXEC		0xf
+#define I40E_NVM_TRANS_SHIFT			8
+#define I40E_NVM_TRANS_MASK			(0xf << I40E_NVM_TRANS_SHIFT)
+#define I40E_NVM_PRESERVATION_FLAGS_SHIFT	12
+#define I40E_NVM_PRESERVATION_FLAGS_MASK \
+				(0x3 << I40E_NVM_PRESERVATION_FLAGS_SHIFT)
+#define I40E_NVM_PRESERVATION_FLAGS_SELECTED	0x01
+#define I40E_NVM_PRESERVATION_FLAGS_ALL		0x02
+#define I40E_NVM_CON				0x0
+#define I40E_NVM_SNT				0x1
+#define I40E_NVM_LCB				0x2
+#define I40E_NVM_SA				(I40E_NVM_SNT | I40E_NVM_LCB)
+#define I40E_NVM_ERA				0x4
+#define I40E_NVM_CSUM				0x8
+#define I40E_NVM_AQE				0xe
+#define I40E_NVM_EXEC				0xf
 
 #define I40E_NVM_ADAPT_SHIFT	16
 #define I40E_NVM_ADAPT_MASK	(0xffff << I40E_NVM_ADAPT_SHIFT)
@@ -611,6 +618,7 @@ struct i40e_hw {
 	/* state of nvm update process */
 	enum i40e_nvmupd_state nvmupd_state;
 	struct i40e_aq_desc nvm_wb_desc;
+	struct i40e_aq_desc nvm_aq_event_desc;
 	struct i40e_virt_mem nvm_buff;
 	bool nvm_release_on_done;
 	u16 nvm_wait_opcode;
drivers/net/ethernet/intel/i40evf/i40e_adminq_cmd.h +6 -2
index 435a112..b0e6454 100644
@@ -2196,8 +2196,12 @@ I40E_CHECK_CMD_LENGTH(i40e_aqc_phy_register_access);
  */
 struct i40e_aqc_nvm_update {
 	u8	command_flags;
-#define I40E_AQ_NVM_LAST_CMD	0x01
-#define I40E_AQ_NVM_FLASH_ONLY	0x80
+#define I40E_AQ_NVM_LAST_CMD			0x01
+#define I40E_AQ_NVM_FLASH_ONLY			0x80
+#define I40E_AQ_NVM_PRESERVATION_FLAGS_SHIFT	1
+#define I40E_AQ_NVM_PRESERVATION_FLAGS_MASK	0x03
+#define I40E_AQ_NVM_PRESERVATION_FLAGS_SELECTED	0x03
+#define I40E_AQ_NVM_PRESERVATION_FLAGS_ALL	0x01
 	u8	module_pointer;
 	__le16	length;
 	__le32	offset;
drivers/net/ethernet/intel/i40evf/i40e_type.h +17 -9
index 6afc316..54951c8 100644
@@ -361,6 +361,7 @@ enum i40e_nvmupd_cmd {
 	I40E_NVMUPD_STATUS,
 	I40E_NVMUPD_EXEC_AQ,
 	I40E_NVMUPD_GET_AQ_RESULT,
+	I40E_NVMUPD_GET_AQ_EVENT,
 };
 
 enum i40e_nvmupd_state {
@@ -380,15 +381,21 @@ enum i40e_nvmupd_state {
 
 #define I40E_NVM_MOD_PNT_MASK 0xFF
 
-#define I40E_NVM_TRANS_SHIFT	8
-#define I40E_NVM_TRANS_MASK	(0xf << I40E_NVM_TRANS_SHIFT)
-#define I40E_NVM_CON		0x0
-#define I40E_NVM_SNT		0x1
-#define I40E_NVM_LCB		0x2
-#define I40E_NVM_SA		(I40E_NVM_SNT | I40E_NVM_LCB)
-#define I40E_NVM_ERA		0x4
-#define I40E_NVM_CSUM		0x8
-#define I40E_NVM_EXEC		0xf
+#define I40E_NVM_TRANS_SHIFT			8
+#define I40E_NVM_TRANS_MASK			(0xf << I40E_NVM_TRANS_SHIFT)
+#define I40E_NVM_PRESERVATION_FLAGS_SHIFT	12
+#define I40E_NVM_PRESERVATION_FLAGS_MASK \
+				(0x3 << I40E_NVM_PRESERVATION_FLAGS_SHIFT)
+#define I40E_NVM_PRESERVATION_FLAGS_SELECTED	0x01
+#define I40E_NVM_PRESERVATION_FLAGS_ALL		0x02
+#define I40E_NVM_CON				0x0
+#define I40E_NVM_SNT				0x1
+#define I40E_NVM_LCB				0x2
+#define I40E_NVM_SA				(I40E_NVM_SNT | I40E_NVM_LCB)
+#define I40E_NVM_ERA				0x4
+#define I40E_NVM_CSUM				0x8
+#define I40E_NVM_AQE				0xe
+#define I40E_NVM_EXEC				0xf
 
 #define I40E_NVM_ADAPT_SHIFT	16
 #define I40E_NVM_ADAPT_MASK	(0xffff << I40E_NVM_ADAPT_SHIFT)
@@ -561,6 +568,7 @@ struct i40e_hw {
 	/* state of nvm update process */
 	enum i40e_nvmupd_state nvmupd_state;
 	struct i40e_aq_desc nvm_wb_desc;
+	struct i40e_aq_desc nvm_aq_event_desc;
 	struct i40e_virt_mem nvm_buff;
 	bool nvm_release_on_done;
 	u16 nvm_wait_opcode;
Signature · 1 line
2.9.5
#1 lore raw

PATCH 2/9 i40e: avoid divide by zero

From: Mitch Williams <[email protected]>

In some weird circumstances with DCB enabled, the firmware can fail to configure the VSI, leaving us with zero traffic classes. Check for this state when we configure RSS to avoid a panic.

---
 drivers/net/ethernet/intel/i40e/i40e_main.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_main.c +7 -2
index 4f74156..7a882d7 100644
@@ -10839,8 +10839,13 @@ static int i40e_pf_config_rss(struct i40e_pf *pf)
 	/* Determine the RSS size of the VSI */
 	if (!vsi->rss_size) {
 		u16 qcount;
-
-		qcount = vsi->num_queue_pairs / vsi->tc_config.numtc;
+		/* If the firmware does something weird during VSI init, we
+		 * could end up with zero TCs. Check for that to avoid
+		 * divide-by-zero. It probably won't pass traffic, but it also
+		 * won't panic.
+		 */
+		qcount = vsi->num_queue_pairs /
+			 (vsi->tc_config.numtc ? vsi->tc_config.numtc : 1);
 		vsi->rss_size = min_t(int, pf->alloc_rss_size, qcount);
 	}
 	if (!vsi->rss_size)
Signature · 1 line
2.9.5
#2 lore raw

PATCH 3/9 i40evf: remove flush_scheduled_work call in i40evf_remove

In reply to Alice Michael

From: Sudheer Mogilappagari <[email protected]>

flush_schedule_work blocks until completion of all scheduled work items in global work-queue. This can cause deadlock in some cases. i40evf_remove() cleans up necessary work items with cancel_delayed_work_sync and cancel_work_sync. This fix removes flush_schedule_work call inside i40evf_remove().

---
 drivers/net/ethernet/intel/i40evf/i40evf_main.c | 2 --
 1 file changed, 2 deletions(-)
drivers/net/ethernet/intel/i40evf/i40evf_main.c +0 -2
index f92587a..6472116 100644
@@ -3088,8 +3088,6 @@ static void i40evf_remove(struct pci_dev *pdev)
 	if (adapter->watchdog_timer.function)
 		del_timer_sync(&adapter->watchdog_timer);
 
-	flush_scheduled_work();
-
 	i40evf_free_rss(adapter);
 
 	if (hw->aq.asq.count)
Signature · 1 line
2.9.5
#3 lore raw

PATCH 4/9 i40evf: Don't schedule reset_task when device is being removed

In reply to Alice Michael

From: Avinash Dayanand <[email protected]>

When a host disables and enables a PF device, all the associated VFs are removed and added back in. It also generates a PFR which in turn resets all the connected VFs. This behaviour is different from that of Linux guest on Linux host. Hence we end up in a situation where there's a PFR and device removal at the same time. And watchdog doesn't have a clue about this and schedules a reset_task. This patch adds code to send signal to reset_task that the device is currently being removed.

---
 drivers/net/ethernet/intel/i40evf/i40evf.h      | 1 +
 drivers/net/ethernet/intel/i40evf/i40evf_main.c | 9 ++++++++-
 2 files changed, 9 insertions(+), 1 deletion(-)
drivers/net/ethernet/intel/i40evf/i40evf.h +1 -0
index 47040ab..33c0ffc 100644
@@ -187,6 +187,7 @@ enum i40evf_state_t {
 enum i40evf_critical_section_t {
 	__I40EVF_IN_CRITICAL_TASK,	/* cannot be interrupted */
 	__I40EVF_IN_CLIENT_TASK,
+	__I40EVF_IN_REMOVE_TASK,	/* device being removed */
 };
 
 /* board specific private data structure */
drivers/net/ethernet/intel/i40evf/i40evf_main.c +8 -1
index 6472116..0e5df19 100644
@@ -1803,6 +1803,12 @@ static void i40evf_reset_task(struct work_struct *work)
 	int i = 0, err;
 	bool running;
 
+	/* When device is being removed it doesn't make sense to run the reset
+	 * task, just return in such a case.
+	 */
+	if (test_bit(__I40EVF_IN_REMOVE_TASK, &adapter->crit_section))
+		return;
+
 	while (test_and_set_bit(__I40EVF_IN_CLIENT_TASK,
 				&adapter->crit_section))
 		usleep_range(500, 1000);
@@ -3053,7 +3059,8 @@ static void i40evf_remove(struct pci_dev *pdev)
 	struct i40evf_mac_filter *f, *ftmp;
 	struct i40e_hw *hw = &adapter->hw;
 	int err;
-
+	/* Indicate we are in remove and not to run reset_task */
+	set_bit(__I40EVF_IN_REMOVE_TASK, &adapter->crit_section);
 	cancel_delayed_work_sync(&adapter->init_task);
 	cancel_work_sync(&adapter->reset_task);
 	cancel_delayed_work_sync(&adapter->client_task);
Signature · 1 line
2.9.5
#4 lore raw

PATCH 5/9 i40e: Fix for blinking activity instead of link LEDs

In reply to Alice Michael

From: Michal Kuchta <[email protected]>

This fix solves an issue occurring while calling i40e_led_set function from the driver with "blink" parameter set as TRUE. This call resulted in Activity LED blinking instead of Link LED, which may lead to errors in physically identyfying the port, since Activity LED may be blinking for different reasons aswell.

---
 drivers/net/ethernet/intel/i40e/i40e_common.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_common.c +2 -3
index cf929cc..ee6052e 100644
@@ -1486,6 +1486,7 @@ u32 i40e_led_get(struct i40e_hw *hw)
 		case I40E_COMBINED_ACTIVITY:
 		case I40E_FILTER_ACTIVITY:
 		case I40E_MAC_ACTIVITY:
+		case I40E_LINK_ACTIVITY:
 			continue;
 		default:
 			break;
@@ -1534,6 +1535,7 @@ void i40e_led_set(struct i40e_hw *hw, u32 mode, bool blink)
 		case I40E_COMBINED_ACTIVITY:
 		case I40E_FILTER_ACTIVITY:
 		case I40E_MAC_ACTIVITY:
+		case I40E_LINK_ACTIVITY:
 			continue;
 		default:
 			break;
@@ -1544,9 +1546,6 @@ void i40e_led_set(struct i40e_hw *hw, u32 mode, bool blink)
 		gpio_val |= ((mode << I40E_GLGEN_GPIO_CTL_LED_MODE_SHIFT) &
 			     I40E_GLGEN_GPIO_CTL_LED_MODE_MASK);
 
-		if (mode == I40E_LINK_ACTIVITY)
-			blink = false;
-
 		if (blink)
 			gpio_val |= BIT(I40E_GLGEN_GPIO_CTL_LED_BLINK_SHIFT);
 		else
Signature · 1 line
2.9.5
#5 lore raw

PATCH 6/9 i40e/i40evf: Detect and recover hung queue scenario

In reply to Alice Michael

From: Sudheer Mogilappagari <[email protected]>

In VFs, there is a known issue which can cause writebacks to not occur when interrupts are disabled and there are less than 4 descriptors resulting in TX timeout. Timeout can also occur due to lost interrupt.

The current implementation for detecting and recovering from hung queues in the PF is problematic because it actually actively encourages lost interrupts. By triggering a SW interrupt, interrupts are forced on. If we are already in napi_poll and an interrupt fires, napi_poll will not be rescheduled and the interrupt is effectively lost; thereby potentially *causing* hung queues.

This patch checks whether packets are being processed between every watchdog cycle and determine potential hung queue and fires triggers SW interrupt only for that particular queue.

---
 drivers/net/ethernet/intel/i40e/i40e_main.c     | 100 +-----------------------
 drivers/net/ethernet/intel/i40e/i40e_txrx.c     |  54 +++++++++++++
 drivers/net/ethernet/intel/i40e/i40e_txrx.h     |   2 +
 drivers/net/ethernet/intel/i40evf/i40e_txrx.c   |  54 +++++++++++++
 drivers/net/ethernet/intel/i40evf/i40e_txrx.h   |   2 +
 drivers/net/ethernet/intel/i40evf/i40evf_main.c |   2 +
 6 files changed, 115 insertions(+), 99 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_main.c +1 -99
index 7a882d7..a53f76f 100644
@@ -4862,104 +4862,6 @@ static int i40e_pf_wait_queues_disabled(struct i40e_pf *pf)
 #endif
 
 /**
- * i40e_detect_recover_hung_queue - Function to detect and recover hung_queue
- * @q_idx: TX queue number
- * @vsi: Pointer to VSI struct
- *
- * This function checks specified queue for given VSI. Detects hung condition.
- * We proactively detect hung TX queues by checking if interrupts are disabled
- * but there are pending descriptors.  If it appears hung, attempt to recover
- * by triggering a SW interrupt.
- **/
-static void i40e_detect_recover_hung_queue(int q_idx, struct i40e_vsi *vsi)
-{
-	struct i40e_ring *tx_ring = NULL;
-	struct i40e_pf	*pf;
-	u32 val, tx_pending;
-	int i;
-
-	pf = vsi->back;
-
-	/* now that we have an index, find the tx_ring struct */
-	for (i = 0; i < vsi->num_queue_pairs; i++) {
-		if (vsi->tx_rings[i] && vsi->tx_rings[i]->desc) {
-			if (q_idx == vsi->tx_rings[i]->queue_index) {
-				tx_ring = vsi->tx_rings[i];
-				break;
-			}
-		}
-	}
-
-	if (!tx_ring)
-		return;
-
-	/* Read interrupt register */
-	if (pf->flags & I40E_FLAG_MSIX_ENABLED)
-		val = rd32(&pf->hw,
-			   I40E_PFINT_DYN_CTLN(tx_ring->q_vector->v_idx +
-					       tx_ring->vsi->base_vector - 1));
-	else
-		val = rd32(&pf->hw, I40E_PFINT_DYN_CTL0);
-
-	tx_pending = i40e_get_tx_pending(tx_ring);
-
-	/* Interrupts are disabled and TX pending is non-zero,
-	 * trigger the SW interrupt (don't wait). Worst case
-	 * there will be one extra interrupt which may result
-	 * into not cleaning any queues because queues are cleaned.
-	 */
-	if (tx_pending && (!(val & I40E_PFINT_DYN_CTLN_INTENA_MASK)))
-		i40e_force_wb(vsi, tx_ring->q_vector);
-}
-
-/**
- * i40e_detect_recover_hung - Function to detect and recover hung_queues
- * @pf:  pointer to PF struct
- *
- * LAN VSI has netdev and netdev has TX queues. This function is to check
- * each of those TX queues if they are hung, trigger recovery by issuing
- * SW interrupt.
- **/
-static void i40e_detect_recover_hung(struct i40e_pf *pf)
-{
-	struct net_device *netdev;
-	struct i40e_vsi *vsi;
-	unsigned int i;
-
-	/* Only for LAN VSI */
-	vsi = pf->vsi[pf->lan_vsi];
-
-	if (!vsi)
-		return;
-
-	/* Make sure, VSI state is not DOWN/RECOVERY_PENDING */
-	if (test_bit(__I40E_VSI_DOWN, vsi->back->state) ||
-	    test_bit(__I40E_RESET_RECOVERY_PENDING, vsi->back->state))
-		return;
-
-	/* Make sure type is MAIN VSI */
-	if (vsi->type != I40E_VSI_MAIN)
-		return;
-
-	netdev = vsi->netdev;
-	if (!netdev)
-		return;
-
-	/* Bail out if netif_carrier is not OK */
-	if (!netif_carrier_ok(netdev))
-		return;
-
-	/* Go thru' TX queues for netdev */
-	for (i = 0; i < netdev->num_tx_queues; i++) {
-		struct netdev_queue *q;
-
-		q = netdev_get_tx_queue(netdev, i);
-		if (q)
-			i40e_detect_recover_hung_queue(i, vsi);
-	}
-}
-
-/**
  * i40e_get_iscsi_tc_map - Return TC map for iSCSI APP
  * @pf: pointer to PF
  *
@@ -9751,7 +9653,7 @@ static void i40e_service_task(struct work_struct *work)
 	if (test_and_set_bit(__I40E_SERVICE_SCHED, pf->state))
 		return;
 
-	i40e_detect_recover_hung(pf);
+	i40e_detect_recover_hung(pf->vsi[pf->lan_vsi]);
 	i40e_sync_filters_subtask(pf);
 	i40e_reset_subtask(pf);
 	i40e_handle_mdd_event(pf);
drivers/net/ethernet/intel/i40e/i40e_txrx.c +54 -0

66 lines not on this page · raw message

drivers/net/ethernet/intel/i40e/i40e_txrx.h +2 -0
index fbae118..f1c13be 100644
@@ -331,6 +331,7 @@ struct i40e_tx_queue_stats {
 	u64 tx_done_old;
 	u64 tx_linearize;
 	u64 tx_force_wb;
+	int prev_pkt_ctr;
 };
 
 struct i40e_rx_queue_stats {
@@ -498,6 +499,7 @@ void i40e_free_rx_resources(struct i40e_ring *rx_ring);
 int i40e_napi_poll(struct napi_struct *napi, int budget);
 void i40e_force_wb(struct i40e_vsi *vsi, struct i40e_q_vector *q_vector);
 u32 i40e_get_tx_pending(struct i40e_ring *ring);
+void i40e_detect_recover_hung(struct i40e_vsi *vsi);
 int __i40e_maybe_stop_tx(struct i40e_ring *tx_ring, int size);
 bool __i40e_chk_linearize(struct sk_buff *skb);
 
drivers/net/ethernet/intel/i40evf/i40e_txrx.c +54 -0

66 lines not on this page · raw message

drivers/net/ethernet/intel/i40evf/i40e_txrx.h +2 -0
index 8d26c85..e72f16b 100644
@@ -313,6 +313,7 @@ struct i40e_tx_queue_stats {
 	u64 tx_done_old;
 	u64 tx_linearize;
 	u64 tx_force_wb;
+	int prev_pkt_ctr;
 	u64 tx_lost_interrupt;
 };
 
@@ -467,6 +468,7 @@ void i40evf_free_rx_resources(struct i40e_ring *rx_ring);
 int i40evf_napi_poll(struct napi_struct *napi, int budget);
 void i40evf_force_wb(struct i40e_vsi *vsi, struct i40e_q_vector *q_vector);
 u32 i40evf_get_tx_pending(struct i40e_ring *ring, bool in_sw);
+void i40evf_detect_recover_hung(struct i40e_vsi *vsi);
 int __i40evf_maybe_stop_tx(struct i40e_ring *tx_ring, int size);
 bool __i40evf_chk_linearize(struct sk_buff *skb);
 
drivers/net/ethernet/intel/i40evf/i40evf_main.c +2 -0
index 0e5df19..8934f78 100644
@@ -1716,6 +1716,8 @@ static void i40evf_watchdog_task(struct work_struct *work)
 	if (adapter->state == __I40EVF_RUNNING)
 		i40evf_request_stats(adapter);
 watchdog_done:
+	if (adapter->state == __I40EVF_RUNNING)
+		i40evf_detect_recover_hung(&adapter->vsi);
 	clear_bit(__I40EVF_IN_CRITICAL_TASK, &adapter->crit_section);
 restart_watchdog:
 	if (adapter->state == __I40EVF_REMOVE)
Signature · 1 line
2.9.5
#6 lore raw

PATCH 7/9 i40e: check for invalid DCB config

In reply to Alice Michael

From: Mitch Williams <[email protected]>

The driver (and the entire netdev layer for that matter) assumes that TC0 will always be present in our DCB configuration. Unfortunately, this isn't always the case. Rather than fail to configure the VSI, let's go ahead and try to make it work, even though DCB will end up being disabled by the kernel.

If the driver fails to configure DCB, the driver queries what's valid, then writes that back to the hardware, always forcing TC0.

This fixes a bug where the driver could fail to adhere to ETS BW allocations if 8 TCs were configured on the switch.

---
 drivers/net/ethernet/intel/i40e/i40e_main.c | 34 ++++++++++++++++++++++++++++-
 1 file changed, 33 insertions(+), 1 deletion(-)
drivers/net/ethernet/intel/i40e/i40e_main.c +33 -1
index a53f76f..1ce24e2 100644
@@ -5229,6 +5229,8 @@ static void i40e_vsi_update_queue_map(struct i40e_vsi *vsi,
 static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
 {
 	u8 bw_share[I40E_MAX_TRAFFIC_CLASS] = {0};
+	struct i40e_pf *pf = vsi->back;
+	struct i40e_hw *hw = &pf->hw;
 	struct i40e_vsi_context ctxt;
 	int ret = 0;
 	int i;
@@ -5246,10 +5248,40 @@ static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
 
 	ret = i40e_vsi_configure_bw_alloc(vsi, enabled_tc, bw_share);
 	if (ret) {
+		struct i40e_aqc_query_vsi_bw_config_resp bw_config = {0};
+
 		dev_info(&vsi->back->pdev->dev,
 			 "Failed configuring TC map %d for VSI %d\n",
 			 enabled_tc, vsi->seid);
-		goto out;
+		ret = i40e_aq_query_vsi_bw_config(hw, vsi->seid,
+						  &bw_config, NULL);
+		if (ret) {
+			dev_info(&pf->pdev->dev,
+				 "Failed querying vsi bw info, err %s aq_err %s\n",
+				 i40e_stat_str(hw, ret),
+				 i40e_aq_str(hw, hw->aq.asq_last_status));
+			goto out;
+		}
+		if ((bw_config.tc_valid_bits & enabled_tc) != enabled_tc) {
+			u8 valid_tc = bw_config.tc_valid_bits & enabled_tc;
+
+			if (!valid_tc)
+				valid_tc = bw_config.tc_valid_bits;
+			/* Always enable TC0, no matter what */
+			valid_tc |= 1;
+			dev_info(&pf->pdev->dev,
+				 "Requested tc 0x%x, but FW reports 0x%x as valid. Attempting to use 0x%x.\n",
+				 enabled_tc, bw_config.tc_valid_bits, valid_tc);
+			enabled_tc = valid_tc;
+		}
+
+		ret = i40e_vsi_configure_bw_alloc(vsi, enabled_tc, bw_share);
+		if (ret) {
+			dev_err(&pf->pdev->dev,
+				"Unable to  configure TC map %d for VSI %d\n",
+				enabled_tc, vsi->seid);
+			goto out;
+		}
 	}
 
 	/* Update Queue Pairs Mapping for currently enabled UPs */
Signature · 1 line
2.9.5
#7 lore raw

PATCH 8/9 i40e: simplify pointer dereferences

In reply to Alice Michael

From: Mitch Williams <[email protected]>

Now that i40e_vsi_config_tc() has the pf and hw variable defined, use them, instead of dereferencing vsi->back. Much easier to read.

---
 drivers/net/ethernet/intel/i40e/i40e_main.c | 18 ++++++++----------
 1 file changed, 8 insertions(+), 10 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_main.c +8 -10
index 1ce24e2..5c5296e 100644
@@ -5250,7 +5250,7 @@ static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
 	if (ret) {
 		struct i40e_aqc_query_vsi_bw_config_resp bw_config = {0};
 
-		dev_info(&vsi->back->pdev->dev,
+		dev_info(&pf->pdev->dev,
 			 "Failed configuring TC map %d for VSI %d\n",
 			 enabled_tc, vsi->seid);
 		ret = i40e_aq_query_vsi_bw_config(hw, vsi->seid,
@@ -5321,13 +5321,12 @@ static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
 	/* Update the VSI after updating the VSI queue-mapping
 	 * information
 	 */
-	ret = i40e_aq_update_vsi_params(&vsi->back->hw, &ctxt, NULL);
+	ret = i40e_aq_update_vsi_params(hw, &ctxt, NULL);
 	if (ret) {
-		dev_info(&vsi->back->pdev->dev,
+		dev_info(&pf->pdev->dev,
 			 "Update vsi tc config failed, err %s aq_err %s\n",
-			 i40e_stat_str(&vsi->back->hw, ret),
-			 i40e_aq_str(&vsi->back->hw,
-				     vsi->back->hw.aq.asq_last_status));
+			 i40e_stat_str(hw, ret),
+			 i40e_aq_str(hw, hw->aq.asq_last_status));
 		goto out;
 	}
 	/* update the local VSI info with updated queue map */
@@ -5337,11 +5336,10 @@ static int i40e_vsi_config_tc(struct i40e_vsi *vsi, u8 enabled_tc)
 	/* Update current VSI BW information */
 	ret = i40e_vsi_get_bw_info(vsi);
 	if (ret) {
-		dev_info(&vsi->back->pdev->dev,
+		dev_info(&pf->pdev->dev,
 			 "Failed updating vsi bw info, err %s aq_err %s\n",
-			 i40e_stat_str(&vsi->back->hw, ret),
-			 i40e_aq_str(&vsi->back->hw,
-				     vsi->back->hw.aq.asq_last_status));
+			 i40e_stat_str(hw, ret),
+			 i40e_aq_str(hw, hw->aq.asq_last_status));
 		goto out;
 	}
 
Signature · 1 line
2.9.5
#8 lore raw

PATCH 9/9 i40e: Disable iWARP VSI PETCP_ENA flag on netdev down events

In reply to Alice Michael

From: Shiraz Saleem <[email protected]>

Client close is overloaded to handle both un-registration and netdev down event. On netdev down, i40iw client close is called which unregisters the RDMA dev and and this is too destructive since the netdev is still registered.

Do not call client close/open on netdev down/up events. Instead disable the PE TCP_ENA flag during a netdev down event. This blocks all TCP traffic to the RDMA Protocol Engine. On netdev up, re-enable the flag.

---
 drivers/net/ethernet/intel/i40e/i40e_client.c | 36 ++++++++++++++-------------
 drivers/net/ethernet/intel/i40e/i40e_client.h |  2 +-
 2 files changed, 20 insertions(+), 18 deletions(-)
drivers/net/ethernet/intel/i40e/i40e_client.c +19 -17
index 1b1e2ac..0de9610 100644
@@ -378,11 +378,11 @@ void i40e_client_subtask(struct i40e_pf *pf)
 	if (!client || !cdev)
 		return;
 
-	/* Here we handle client opens. If the client is down, but
-	 * the netdev is up, then open the client.
+	/* Here we handle client opens. If the client is down, and
+	 * the netdev is registered, then open the client.
 	 */
 	if (!test_bit(__I40E_CLIENT_INSTANCE_OPENED, &cdev->state)) {
-		if (!test_bit(__I40E_VSI_DOWN, vsi->state) &&
+		if (vsi->netdev_registered &&
 		    client->ops && client->ops->open) {
 			set_bit(__I40E_CLIENT_INSTANCE_OPENED, &cdev->state);
 			ret = client->ops->open(&cdev->lan_info, client);
@@ -393,17 +393,19 @@ void i40e_client_subtask(struct i40e_pf *pf)
 				i40e_client_del_instance(pf);
 			}
 		}
-	} else {
-	/* Likewise for client close. If the client is up, but the netdev
-	 * is down, then close the client.
-	 */
-		if (test_bit(__I40E_VSI_DOWN, vsi->state) &&
-		    client->ops && client->ops->close) {
-			clear_bit(__I40E_CLIENT_INSTANCE_OPENED, &cdev->state);
-			client->ops->close(&cdev->lan_info, client, false);
-			i40e_client_release_qvlist(&cdev->lan_info);
-		}
 	}
+
+	/* enable/disable PE TCP_ENA flag based on netdev down/up
+	 */
+	if (test_bit(__I40E_VSI_DOWN, vsi->state))
+		i40e_client_update_vsi_ctxt(&cdev->lan_info, client,
+					    0, 0, 0,
+					    I40E_CLIENT_VSI_FLAG_TCP_ENABLE);
+	else
+		i40e_client_update_vsi_ctxt(&cdev->lan_info, client,
+					    0, 0,
+					    I40E_CLIENT_VSI_FLAG_TCP_ENABLE,
+					    I40E_CLIENT_VSI_FLAG_TCP_ENABLE);
 }
 
 /**
@@ -717,13 +719,13 @@ static int i40e_client_update_vsi_ctxt(struct i40e_info *ldev,
 		return -ENOENT;
 	}
 
-	if ((valid_flag & I40E_CLIENT_VSI_FLAG_TCP_PACKET_ENABLE) &&
-	    (flag & I40E_CLIENT_VSI_FLAG_TCP_PACKET_ENABLE)) {
+	if ((valid_flag & I40E_CLIENT_VSI_FLAG_TCP_ENABLE) &&
+	    (flag & I40E_CLIENT_VSI_FLAG_TCP_ENABLE)) {
 		ctxt.info.valid_sections =
 			cpu_to_le16(I40E_AQ_VSI_PROP_QUEUE_OPT_VALID);
 		ctxt.info.queueing_opt_flags |= I40E_AQ_VSI_QUE_OPT_TCP_ENA;
-	} else if ((valid_flag & I40E_CLIENT_VSI_FLAG_TCP_PACKET_ENABLE) &&
-		  !(flag & I40E_CLIENT_VSI_FLAG_TCP_PACKET_ENABLE)) {
+	} else if ((valid_flag & I40E_CLIENT_VSI_FLAG_TCP_ENABLE) &&
+		  !(flag & I40E_CLIENT_VSI_FLAG_TCP_ENABLE)) {
 		ctxt.info.valid_sections =
 			cpu_to_le16(I40E_AQ_VSI_PROP_QUEUE_OPT_VALID);
 		ctxt.info.queueing_opt_flags &= ~I40E_AQ_VSI_QUE_OPT_TCP_ENA;
drivers/net/ethernet/intel/i40e/i40e_client.h +1 -1
index 15b21a5..ba55c88 100644
@@ -132,7 +132,7 @@ struct i40e_info {
 
 #define I40E_CLIENT_RESET_LEVEL_PF   1
 #define I40E_CLIENT_RESET_LEVEL_CORE 2
-#define I40E_CLIENT_VSI_FLAG_TCP_PACKET_ENABLE  BIT(1)
+#define I40E_CLIENT_VSI_FLAG_TCP_ENABLE  BIT(1)
 
 struct i40e_ops {
 	/* setup_q_vector_list enables queues with a particular vector */
Signature · 1 line
2.9.5