diff options
author | Kuo Zhao <kuozhao@google.com> | 2020-09-11 10:38:47 -0700 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2020-09-11 14:31:54 -0700 |
commit | 24aeb56f2d38edf1b324bdb4f8bc6faf9f0f540c (patch) | |
tree | ae0520d098cebf82ce9fed455bb62c9e01c7301c /drivers/net/ethernet/google/gve/gve_main.c | |
parent | gve: Use dev_info/err instead of netif_info/err. (diff) | |
download | linux-dev-24aeb56f2d38edf1b324bdb4f8bc6faf9f0f540c.tar.xz linux-dev-24aeb56f2d38edf1b324bdb4f8bc6faf9f0f540c.zip |
gve: Add Gvnic stats AQ command and ethtool show/set-priv-flags.
This adds functionality to report driver stats to Hypervisor.
(Users may want to turn this feature off as a matter of privacy
so a "report-stats" flag is added as an ethtool priv option.
It is also disabled by default.)
The hypervisor would trigger a stats report in case "too many"
packets dropped; the stats would be useful in debugging stuck
queues.
A "stats_report_trigger_cnt" stat is added to count the number of times
the hypervisor attempts to trigger stats report.
A timer is also added so that when report-stats is enabled, stat are
updated once every 20 seconds.
Reviewed-by: Yangchun Fu <yangchun@google.com>
Signed-off-by: Kuo Zhao <kuozhao@google.com>
Signed-off-by: David Awogbemila <awogbemila@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'drivers/net/ethernet/google/gve/gve_main.c')
-rw-r--r-- | drivers/net/ethernet/google/gve/gve_main.c | 158 |
1 files changed, 156 insertions, 2 deletions
diff --git a/drivers/net/ethernet/google/gve/gve_main.c b/drivers/net/ethernet/google/gve/gve_main.c index 0fc68f844edf..936ff4c9d250 100644 --- a/drivers/net/ethernet/google/gve/gve_main.c +++ b/drivers/net/ethernet/google/gve/gve_main.c @@ -78,6 +78,66 @@ static void gve_free_counter_array(struct gve_priv *priv) priv->counter_array = NULL; } +/* NIC requests to report stats */ +static void gve_stats_report_task(struct work_struct *work) +{ + struct gve_priv *priv = container_of(work, struct gve_priv, + stats_report_task); + if (gve_get_do_report_stats(priv)) { + gve_handle_report_stats(priv); + gve_clear_do_report_stats(priv); + } +} + +static void gve_stats_report_schedule(struct gve_priv *priv) +{ + if (!gve_get_probe_in_progress(priv) && + !gve_get_reset_in_progress(priv)) { + gve_set_do_report_stats(priv); + queue_work(priv->gve_wq, &priv->stats_report_task); + } +} + +static void gve_stats_report_timer(struct timer_list *t) +{ + struct gve_priv *priv = from_timer(priv, t, stats_report_timer); + + mod_timer(&priv->stats_report_timer, + round_jiffies(jiffies + + msecs_to_jiffies(priv->stats_report_timer_period))); + gve_stats_report_schedule(priv); +} + +static int gve_alloc_stats_report(struct gve_priv *priv) +{ + int tx_stats_num, rx_stats_num; + + tx_stats_num = (GVE_TX_STATS_REPORT_NUM) * + priv->tx_cfg.num_queues; + rx_stats_num = (GVE_RX_STATS_REPORT_NUM) * + priv->rx_cfg.num_queues; + priv->stats_report_len = sizeof(struct gve_stats_report) + + (tx_stats_num + rx_stats_num) * + sizeof(struct stats); + priv->stats_report = + dma_alloc_coherent(&priv->pdev->dev, priv->stats_report_len, + &priv->stats_report_bus, GFP_KERNEL); + if (!priv->stats_report) + return -ENOMEM; + /* Set up timer for the report-stats task */ + timer_setup(&priv->stats_report_timer, gve_stats_report_timer, 0); + priv->stats_report_timer_period = GVE_STATS_REPORT_TIMER_PERIOD; + return 0; +} + +static void gve_free_stats_report(struct gve_priv *priv) +{ + del_timer_sync(&priv->stats_report_timer); + dma_free_coherent(&priv->pdev->dev, priv->stats_report_len, + priv->stats_report, priv->stats_report_bus); + priv->stats_report = NULL; +} + static irqreturn_t gve_mgmnt_intr(int irq, void *arg) { struct gve_priv *priv = arg; @@ -270,6 +330,9 @@ static int gve_setup_device_resources(struct gve_priv *priv) err = gve_alloc_notify_blocks(priv); if (err) goto abort_with_counter; + err = gve_alloc_stats_report(priv); + if (err) + goto abort_with_ntfy_blocks; err = gve_adminq_configure_device_resources(priv, priv->counter_array_bus, priv->num_event_counters, @@ -279,10 +342,18 @@ static int gve_setup_device_resources(struct gve_priv *priv) dev_err(&priv->pdev->dev, "could not setup device_resources: err=%d\n", err); err = -ENXIO; - goto abort_with_ntfy_blocks; + goto abort_with_stats_report; } + err = gve_adminq_report_stats(priv, priv->stats_report_len, + priv->stats_report_bus, + GVE_STATS_REPORT_TIMER_PERIOD); + if (err) + dev_err(&priv->pdev->dev, + "Failed to report stats: err=%d\n", err); gve_set_device_resources_ok(priv); return 0; +abort_with_stats_report: + gve_free_stats_report(priv); abort_with_ntfy_blocks: gve_free_notify_blocks(priv); abort_with_counter: @@ -298,6 +369,13 @@ static void gve_teardown_device_resources(struct gve_priv *priv) /* Tell device its resources are being freed */ if (gve_get_device_resources_ok(priv)) { + /* detach the stats report */ + err = gve_adminq_report_stats(priv, 0, 0x0, GVE_STATS_REPORT_TIMER_PERIOD); + if (err) { + dev_err(&priv->pdev->dev, + "Failed to detach stats report: err=%d\n", err); + gve_trigger_reset(priv); + } err = gve_adminq_deconfigure_device_resources(priv); if (err) { dev_err(&priv->pdev->dev, @@ -308,6 +386,7 @@ static void gve_teardown_device_resources(struct gve_priv *priv) } gve_free_counter_array(priv); gve_free_notify_blocks(priv); + gve_free_stats_report(priv); gve_clear_device_resources_ok(priv); } @@ -699,6 +778,11 @@ static int gve_open(struct net_device *dev) goto reset; gve_set_device_rings_ok(priv); + if (gve_get_report_stats(priv)) + mod_timer(&priv->stats_report_timer, + round_jiffies(jiffies + + msecs_to_jiffies(priv->stats_report_timer_period))); + gve_turnup(priv); netif_carrier_on(dev); priv->interface_up_cnt++; @@ -740,6 +824,7 @@ static int gve_close(struct net_device *dev) goto err; gve_clear_device_rings_ok(priv); } + del_timer_sync(&priv->stats_report_timer); gve_free_rings(priv); gve_free_qpls(priv); @@ -823,6 +908,7 @@ static void gve_turndown(struct gve_priv *priv) netif_tx_disable(priv->dev); gve_clear_napi_enabled(priv); + gve_clear_report_stats(priv); } static void gve_turnup(struct gve_priv *priv) @@ -873,6 +959,10 @@ static void gve_handle_status(struct gve_priv *priv, u32 status) dev_info(&priv->pdev->dev, "Device requested reset.\n"); gve_set_do_reset(priv); } + if (GVE_DEVICE_STATUS_REPORT_STATS_MASK & status) { + priv->stats_report_trigger_cnt++; + gve_set_do_report_stats(priv); + } } static void gve_handle_reset(struct gve_priv *priv) @@ -891,7 +981,68 @@ static void gve_handle_reset(struct gve_priv *priv) } } -/* Handle NIC status register changes and reset requests */ +void gve_handle_report_stats(struct gve_priv *priv) +{ + int idx, stats_idx = 0, tx_bytes; + unsigned int start = 0; + struct stats *stats = priv->stats_report->stats; + + if (!gve_get_report_stats(priv)) + return; + + be64_add_cpu(&priv->stats_report->written_count, 1); + /* tx stats */ + if (priv->tx) { + for (idx = 0; idx < priv->tx_cfg.num_queues; idx++) { + do { + start = u64_stats_fetch_begin(&priv->tx[idx].statss); + tx_bytes = priv->tx[idx].bytes_done; + } while (u64_stats_fetch_retry(&priv->tx[idx].statss, start)); + stats[stats_idx++] = (struct stats) { + .stat_name = cpu_to_be32(TX_WAKE_CNT), + .value = cpu_to_be64(priv->tx[idx].wake_queue), + .queue_id = cpu_to_be32(idx), + }; + stats[stats_idx++] = (struct stats) { + .stat_name = cpu_to_be32(TX_STOP_CNT), + .value = cpu_to_be64(priv->tx[idx].stop_queue), + .queue_id = cpu_to_be32(idx), + }; + stats[stats_idx++] = (struct stats) { + .stat_name = cpu_to_be32(TX_FRAMES_SENT), + .value = cpu_to_be64(priv->tx[idx].req), + .queue_id = cpu_to_be32(idx), + }; + stats[stats_idx++] = (struct stats) { + .stat_name = cpu_to_be32(TX_BYTES_SENT), + .value = cpu_to_be64(tx_bytes), + .queue_id = cpu_to_be32(idx), + }; + stats[stats_idx++] = (struct stats) { + .stat_name = cpu_to_be32(TX_LAST_COMPLETION_PROCESSED), + .value = cpu_to_be64(priv->tx[idx].done), + .queue_id = cpu_to_be32(idx), + }; + } + } + /* rx stats */ + if (priv->rx) { + for (idx = 0; idx < priv->rx_cfg.num_queues; idx++) { + stats[stats_idx++] = (struct stats) { + .stat_name = cpu_to_be32(RX_NEXT_EXPECTED_SEQUENCE), + .value = cpu_to_be64(priv->rx[idx].desc.seqno), + .queue_id = cpu_to_be32(idx), + }; + stats[stats_idx++] = (struct stats) { + .stat_name = cpu_to_be32(RX_BUFFERS_POSTED), + .value = cpu_to_be64(priv->rx[0].fill_cnt), + .queue_id = cpu_to_be32(idx), + }; + } + } +} + +/* Handle NIC status register changes, reset requests and report stats */ static void gve_service_task(struct work_struct *work) { struct gve_priv *priv = container_of(work, struct gve_priv, @@ -1056,6 +1207,7 @@ int gve_reset(struct gve_priv *priv, bool attempt_teardown) priv->reset_cnt++; priv->interface_up_cnt = 0; priv->interface_down_cnt = 0; + priv->stats_report_trigger_cnt = 0; return err; } @@ -1158,6 +1310,7 @@ static int gve_probe(struct pci_dev *pdev, const struct pci_device_id *ent) priv->db_bar2 = db_bar; priv->service_task_flags = 0x0; priv->state_flags = 0x0; + priv->ethtool_flags = 0x0; gve_set_probe_in_progress(priv); priv->gve_wq = alloc_ordered_workqueue("gve", 0); @@ -1167,6 +1320,7 @@ static int gve_probe(struct pci_dev *pdev, const struct pci_device_id *ent) goto abort_with_netdev; } INIT_WORK(&priv->service_task, gve_service_task); + INIT_WORK(&priv->stats_report_task, gve_stats_report_task); priv->tx_cfg.max_queues = max_tx_queues; priv->rx_cfg.max_queues = max_rx_queues; |