[dpdk-dev] MSI-X vector #1 seems to be stalled sometimes after VF reset (ixgbe)

Ruslan Nikolaev ruslan at purestorage.com
Tue Jan 10 03:48:25 CET 2017


dpdk_vfreset.patch:

diff -urN dpdk-16.07/doc/guides/nics/overview.rst dpdk-16.07-new/doc/guides/nics/overview.rst
--- dpdk-16.07/doc/guides/nics/overview.rst	2016-07-28 11:48:41.000000000 -0700
+++ dpdk-16.07-new/doc/guides/nics/overview.rst	2016-12-15 17:32:27.436425563 -0800
@@ -89,6 +89,7 @@
    Speed capabilities
    Link status            Y Y Y   Y Y   Y Y Y     Y   Y Y Y Y         Y Y         Y Y   Y Y Y Y Y
    Link status event      Y Y       Y     Y Y     Y   Y Y             Y Y         Y Y     Y Y
+   Link reset                               Y Y   Y     Y Y
    Queue status event                                                                       Y
    Rx interrupt                     Y     Y Y Y Y Y Y Y Y Y Y Y Y Y Y
    Queue start/stop           Y   Y   Y Y Y Y Y Y     Y Y     Y Y Y Y Y Y               Y Y   Y Y
diff -urN dpdk-16.07/doc/guides/rel_notes/release_16_07.rst dpdk-16.07-new/doc/guides/rel_notes/release_16_07.rst
--- dpdk-16.07/doc/guides/rel_notes/release_16_07.rst	2016-07-28 11:48:41.000000000 -0700
+++ dpdk-16.07-new/doc/guides/rel_notes/release_16_07.rst	2016-12-15 17:32:27.436425563 -0800
@@ -15,6 +15,15 @@
 
       firefox build/doc/html/guides/rel_notes/release_16_07.html
 
+* **Added device reset support for ixgbe VF.**
+
+  Added the device reset API. APP can call this API to reset the VF port
+  when it's not working.
+  Based on the mailbox interruption support, when VF reseives the control
+  message from PF, it means the PF link state changes, VF uses the reset
+  callback in the message handler to notice the APP. APP need call the device
+  reset API to reset the VF port.
+
 
 New Features
 ------------
diff -urN dpdk-16.07/drivers/net/ixgbe/ixgbe_ethdev.c dpdk-16.07-new/drivers/net/ixgbe/ixgbe_ethdev.c
--- dpdk-16.07/drivers/net/ixgbe/ixgbe_ethdev.c	2016-07-28 11:48:41.000000000 -0700
+++ dpdk-16.07-new/drivers/net/ixgbe/ixgbe_ethdev.c	2016-12-15 17:47:51.212425563 -0800
@@ -388,6 +388,10 @@
 static int ixgbe_dev_udp_tunnel_port_del(struct rte_eth_dev *dev,
 					 struct rte_eth_udp_tunnel *udp_tunnel);
 
+static int ixgbevf_dev_reset(struct rte_eth_dev *dev,
+			     struct rte_eth_reset_state *state,
+			     uint32_t nonce);
+
 /*
  * Define VF Stats MACRO for Non "cleared on read" register
  */
@@ -593,6 +597,7 @@
 	.reta_query           = ixgbe_dev_rss_reta_query,
 	.rss_hash_update      = ixgbe_dev_rss_hash_update,
 	.rss_hash_conf_get    = ixgbe_dev_rss_hash_conf_get,
+	.dev_reset            = ixgbevf_dev_reset,
 };
 
 /* store statistics names and its offset in stats structure */
@@ -4157,7 +4162,9 @@
 		ETH_VLAN_EXTEND_MASK;
 	ixgbevf_vlan_offload_set(dev, mask);
 
-	ixgbevf_dev_rxtx_start(dev);
+	err = ixgbevf_dev_rxtx_start(dev);
+	if (err)
+		return err;
 
 	/* check and configure queue intr-vector mapping */
 	if (dev->data->dev_conf.intr_conf.rxq != 0) {
@@ -7305,6 +7312,75 @@
 }
 
 static int
+ixgbevf_dev_reset(struct rte_eth_dev *dev, struct rte_eth_reset_state *state, uint32_t nonce)
+{
+	struct ixgbe_hw *hw = IXGBE_DEV_PRIVATE_TO_HW(dev->data->dev_private);
+	int diag = 0;
+	uint32_t vteiam;
+	struct timespec ts;
+
+	/* STATES: [0] initial, [2] stopped, and [1] started (reset complete). */
+	if (state->state <= 1)
+	{
+		/* Nothing needs to be done if reset is complete and the nonce is the same. */
+		if (state->state == 1 && state->nonce == nonce)
+			return 0;
+
+		state->nonce = nonce;
+		state->state = 2;
+		/* Performance VF reset. */
+		dev->data->dev_started = 0;
+		ixgbevf_dev_stop(dev);
+		if (dev->data->dev_conf.intr_conf.lsc == 0)
+			diag = ixgbe_dev_link_update(dev, 0);
+		if (diag) {
+			PMD_INIT_LOG(INFO, "Ixgbe VF reset: "
+				     "Failed to update link.");
+		}
+		clock_gettime(CLOCK_MONOTONIC_COARSE, &ts);
+		state->time = ts.tv_sec;
+		return 1;
+	}
+
+	/* Delay of 1s */
+	clock_gettime(CLOCK_MONOTONIC_COARSE, &ts);
+	if (ts.tv_sec - state->time < 1) {
+		return 1;
+	}
+
+	state->state = 0;
+	diag = ixgbevf_dev_start(dev);
+	/* If fail to start the device, need to stop/start it again. */
+	if (diag) {
+		PMD_INIT_LOG(ERR, "Ixgbe VF reset: "
+			     "Failed to start device.");
+		return 1;
+	}
+	dev->data->dev_started = 1;
+	ixgbevf_dev_stats_reset(dev);
+	if (dev->data->dev_conf.intr_conf.lsc == 0)
+		diag = ixgbe_dev_link_update(dev, 0);
+	if (diag) {
+		PMD_INIT_LOG(INFO, "Ixgbe VF reset: "
+			     "Failed to update link.");
+	}
+
+	/**
+	 * When the PF link is down, there has chance
+	 * that VF cannot operate its registers. Will
+	 * check if the registers is written
+	 * successfully. If not, repeat stop/start until
+	 * the PF link is up, in other words, until the
+	 * registers can be written.
+	 */
+	vteiam = IXGBE_READ_REG(hw, IXGBE_VTEIAM);
+	/* Reference ixgbevf_intr_enable when checking. */
+	state->state = (vteiam == IXGBE_VF_IRQ_ENABLE_MASK && state->nonce == nonce);
+	/* The state is going to be 1 if successful and 0 -- otherwise. */
+	return !state->state;
+}
+
+static int
 ixgbevf_dev_interrupt_get_status(struct rte_eth_dev *dev)
 {
 	uint32_t eicr;
diff -urN dpdk-16.07/drivers/net/ixgbe/ixgbe_ethdev.h dpdk-16.07-new/drivers/net/ixgbe/ixgbe_ethdev.h
--- dpdk-16.07/drivers/net/ixgbe/ixgbe_ethdev.h	2016-07-28 11:48:41.000000000 -0700
+++ dpdk-16.07-new/drivers/net/ixgbe/ixgbe_ethdev.h	2016-12-15 17:32:27.436425563 -0800
@@ -377,7 +377,7 @@
 
 void ixgbevf_dev_tx_init(struct rte_eth_dev *dev);
 
-void ixgbevf_dev_rxtx_start(struct rte_eth_dev *dev);
+int ixgbevf_dev_rxtx_start(struct rte_eth_dev *dev);
 
 uint16_t ixgbe_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts,
 		uint16_t nb_pkts);
diff -urN dpdk-16.07/drivers/net/ixgbe/ixgbe_rxtx.c dpdk-16.07-new/drivers/net/ixgbe/ixgbe_rxtx.c
--- dpdk-16.07/drivers/net/ixgbe/ixgbe_rxtx.c	2016-07-28 11:48:41.000000000 -0700
+++ dpdk-16.07-new/drivers/net/ixgbe/ixgbe_rxtx.c	2016-12-15 17:32:27.440425563 -0800
@@ -5247,7 +5247,7 @@
 /*
  * [VF] Start Transmit and Receive Units.
  */
-void __attribute__((cold))
+int __attribute__((cold))
 ixgbevf_dev_rxtx_start(struct rte_eth_dev *dev)
 {
 	struct ixgbe_hw     *hw;
@@ -5283,8 +5283,10 @@
 			rte_delay_ms(1);
 			txdctl = IXGBE_READ_REG(hw, IXGBE_VFTXDCTL(i));
 		} while (--poll_ms && !(txdctl & IXGBE_TXDCTL_ENABLE));
-		if (!poll_ms)
+		if (!poll_ms) {
 			PMD_INIT_LOG(ERR, "Could not enable Tx Queue %d", i);
+			return -1;
+		}
 	}
 	for (i = 0; i < dev->data->nb_rx_queues; i++) {
 
@@ -5300,12 +5302,16 @@
 			rte_delay_ms(1);
 			rxdctl = IXGBE_READ_REG(hw, IXGBE_VFRXDCTL(i));
 		} while (--poll_ms && !(rxdctl & IXGBE_RXDCTL_ENABLE));
-		if (!poll_ms)
+		if (!poll_ms) {
 			PMD_INIT_LOG(ERR, "Could not enable Rx Queue %d", i);
+			return -1;
+		}
 		rte_wmb();
 		IXGBE_WRITE_REG(hw, IXGBE_VFRDT(i), rxq->nb_rx_desc - 1);
 
 	}
+
+	return 0;
 }
 
 /* Stubs needed for linkage when CONFIG_RTE_IXGBE_INC_VECTOR is set to 'n' */
diff -urN dpdk-16.07/lib/librte_ether/rte_ethdev.c dpdk-16.07-new/lib/librte_ether/rte_ethdev.c
--- dpdk-16.07/lib/librte_ether/rte_ethdev.c	2016-07-28 11:48:41.000000000 -0700
+++ dpdk-16.07-new/lib/librte_ether/rte_ethdev.c	2016-12-15 17:32:27.440425563 -0800
@@ -3446,3 +3446,20 @@
 				-ENOTSUP);
 	return (*dev->dev_ops->l2_tunnel_offload_set)(dev, l2_tunnel, mask, en);
 }
+
+int
+rte_eth_dev_reset(uint8_t port_id, struct rte_eth_reset_state *state, uint32_t nonce)
+{
+	struct rte_eth_dev *dev;
+	int diag;
+
+	RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
+
+	dev = &rte_eth_devices[port_id];
+
+	RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->dev_reset, -ENOTSUP);
+
+	diag = (*dev->dev_ops->dev_reset)(dev, state, nonce);
+
+	return diag;
+}
diff -urN dpdk-16.07/lib/librte_ether/rte_ethdev.h dpdk-16.07-new/lib/librte_ether/rte_ethdev.h
--- dpdk-16.07/lib/librte_ether/rte_ethdev.h	2016-07-28 11:48:41.000000000 -0700
+++ dpdk-16.07-new/lib/librte_ether/rte_ethdev.h	2016-12-15 17:40:44.936425563 -0800
@@ -980,6 +980,15 @@
 };
 
 /**
+ * A structure to maintain reset state.
+ */
+struct rte_eth_reset_state {
+	time_t time;
+	uint32_t nonce;
+	uint8_t state;
+};
+
+/**
  * RX/TX queue states
  */
 #define RTE_ETH_QUEUE_STATE_STOPPED 0
@@ -1347,6 +1356,11 @@
 	 uint8_t en);
 /**< @internal enable/disable the l2 tunnel offload functions */
 
+typedef int  (*eth_dev_reset_t)(struct rte_eth_dev *dev,
+				struct rte_eth_reset_state *state,
+				uint32_t nonce);
+/**< @internal Function used to reset a configured Ethernet device. */
+
 #ifdef RTE_NIC_BYPASS
 
 enum {
@@ -1537,6 +1551,8 @@
 	eth_l2_tunnel_eth_type_conf_t l2_tunnel_eth_type_conf;
 	/** Enable/disable l2 tunnel offload functions */
 	eth_l2_tunnel_offload_set_t l2_tunnel_offload_set;
+	/** Reset device. */
+	eth_dev_reset_t dev_reset;
 };
 
 /**
@@ -4368,6 +4384,30 @@
 int
 rte_eth_dev_get_name_by_port(uint8_t port_id, char *name);
 
+/**
+ * Reset an ethernet device when it's not working. One scenario is, after PF
+ * port is down and up, the related VF port should be reset.
+ * The API will stop the port, clear the rx/tx queues, re-setup the rx/tx
+ * queues, restart the port.
+ * Before calling this API, APP should stop the rx/tx. When tx is being stopped,
+ * APP can drop the packets and release the buffer instead of sending them.
+ *
+ * @param port_id
+ *   The port identifier of the Ethernet device.
+ * @param state
+ *   The reset state (must be initialized to 0 initially).
+ * @param nonce
+ *   A 32-bit number indicating the current reset attempt.
+ *
+ * @return
+ *   - (0) if successful.
+ *   - (1) needs to be called again.
+ *   - (-ENODEV) if port identifier is invalid.
+ *   - (-ENOTSUP) if hardware doesn't support this function.
+ */
+int
+rte_eth_dev_reset(uint8_t port_id, struct rte_eth_reset_state *state, uint32_t nonce);
+
 #ifdef __cplusplus
 }
 #endif
diff -urN dpdk-16.07/lib/librte_ether/rte_ether_version.map dpdk-16.07-new/lib/librte_ether/rte_ether_version.map
--- dpdk-16.07/lib/librte_ether/rte_ether_version.map	2016-07-28 11:48:41.000000000 -0700
+++ dpdk-16.07-new/lib/librte_ether/rte_ether_version.map	2016-12-15 17:32:27.440425563 -0800
@@ -138,4 +138,5 @@
 	rte_eth_dev_get_name_by_port;
 	rte_eth_dev_get_port_by_name;
 	rte_eth_xstats_get_names;
+	rte_eth_dev_reset;
 } DPDK_16.04;



More information about the dev mailing list