From 05e59e5fb0064c97a1c10921ecd549f2d4a58565 Mon Sep 17 00:00:00 2001
From: hc <hc@nodka.com>
Date: Wed, 09 Oct 2024 06:14:40 +0000
Subject: [PATCH] add REDIRECT
---
kernel/drivers/net/ethernet/amazon/ena/ena_netdev.c | 2288 ++++++++++++++++++++++++++++++++++++++++++----------------
1 files changed, 1,650 insertions(+), 638 deletions(-)
diff --git a/kernel/drivers/net/ethernet/amazon/ena/ena_netdev.c b/kernel/drivers/net/ethernet/amazon/ena/ena_netdev.c
index 9aea4cf..1722d40 100644
--- a/kernel/drivers/net/ethernet/amazon/ena/ena_netdev.c
+++ b/kernel/drivers/net/ethernet/amazon/ena/ena_netdev.c
@@ -1,33 +1,6 @@
+// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB
/*
- * Copyright 2015 Amazon.com, Inc. or its affiliates.
- *
- * This software is available to you under a choice of one of two
- * licenses. You may choose to be licensed under the terms of the GNU
- * General Public License (GPL) Version 2, available from the file
- * COPYING in the main directory of this source tree, or the
- * BSD license below:
- *
- * Redistribution and use in source and binary forms, with or
- * without modification, are permitted provided that the following
- * conditions are met:
- *
- * - Redistributions of source code must retain the above
- * copyright notice, this list of conditions and the following
- * disclaimer.
- *
- * - Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following
- * disclaimer in the documentation and/or other materials
- * provided with the distribution.
- *
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
- * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
- * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
- * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
- * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
- * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
- * SOFTWARE.
+ * Copyright 2015-2020 Amazon.com, Inc. or its affiliates. All rights reserved.
*/
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
@@ -36,10 +9,8 @@
#include <linux/cpu_rmap.h>
#endif /* CONFIG_RFS_ACCEL */
#include <linux/ethtool.h>
-#include <linux/if_vlan.h>
#include <linux/kernel.h>
#include <linux/module.h>
-#include <linux/moduleparam.h>
#include <linux/numa.h>
#include <linux/pci.h>
#include <linux/utsname.h>
@@ -48,14 +19,12 @@
#include <net/ip.h>
#include "ena_netdev.h"
+#include <linux/bpf_trace.h>
#include "ena_pci_id_tbl.h"
-
-static char version[] = DEVICE_NAME " v" DRV_MODULE_VERSION "\n";
MODULE_AUTHOR("Amazon.com, Inc. or its affiliates");
MODULE_DESCRIPTION(DEVICE_NAME);
MODULE_LICENSE("GPL");
-MODULE_VERSION(DRV_MODULE_VERSION);
/* Time in jiffies before concluding the transmitter is hung. */
#define TX_TIMEOUT (5 * HZ)
@@ -79,7 +48,37 @@
static void ena_destroy_device(struct ena_adapter *adapter, bool graceful);
static int ena_restore_device(struct ena_adapter *adapter);
-static void ena_tx_timeout(struct net_device *dev)
+static void ena_init_io_rings(struct ena_adapter *adapter,
+ int first_index, int count);
+static void ena_init_napi_in_range(struct ena_adapter *adapter, int first_index,
+ int count);
+static void ena_del_napi_in_range(struct ena_adapter *adapter, int first_index,
+ int count);
+static int ena_setup_tx_resources(struct ena_adapter *adapter, int qid);
+static int ena_setup_tx_resources_in_range(struct ena_adapter *adapter,
+ int first_index,
+ int count);
+static int ena_create_io_tx_queue(struct ena_adapter *adapter, int qid);
+static void ena_free_tx_resources(struct ena_adapter *adapter, int qid);
+static int ena_clean_xdp_irq(struct ena_ring *xdp_ring, u32 budget);
+static void ena_destroy_all_tx_queues(struct ena_adapter *adapter);
+static void ena_free_all_io_tx_resources(struct ena_adapter *adapter);
+static void ena_napi_disable_in_range(struct ena_adapter *adapter,
+ int first_index, int count);
+static void ena_napi_enable_in_range(struct ena_adapter *adapter,
+ int first_index, int count);
+static int ena_up(struct ena_adapter *adapter);
+static void ena_down(struct ena_adapter *adapter);
+static void ena_unmask_interrupt(struct ena_ring *tx_ring,
+ struct ena_ring *rx_ring);
+static void ena_update_ring_numa_node(struct ena_ring *tx_ring,
+ struct ena_ring *rx_ring);
+static void ena_unmap_tx_buff(struct ena_ring *tx_ring,
+ struct ena_tx_buffer *tx_info);
+static int ena_create_io_tx_queues_in_range(struct ena_adapter *adapter,
+ int first_index, int count);
+
+static void ena_tx_timeout(struct net_device *dev, unsigned int txqueue)
{
struct ena_adapter *adapter = netdev_priv(dev);
@@ -102,7 +101,7 @@
{
int i;
- for (i = 0; i < adapter->num_queues; i++)
+ for (i = 0; i < adapter->num_io_queues; i++)
adapter->rx_ring[i].mtu = mtu;
}
@@ -113,7 +112,7 @@
ret = ena_com_set_dev_mtu(adapter->ena_dev, new_mtu);
if (!ret) {
- netif_dbg(adapter, drv, dev, "set MTU to %d\n", new_mtu);
+ netif_dbg(adapter, drv, dev, "Set MTU to %d\n", new_mtu);
update_rx_ring_mtu(adapter, new_mtu);
dev->mtu = new_mtu;
} else {
@@ -124,16 +123,466 @@
return ret;
}
+static int ena_xmit_common(struct net_device *dev,
+ struct ena_ring *ring,
+ struct ena_tx_buffer *tx_info,
+ struct ena_com_tx_ctx *ena_tx_ctx,
+ u16 next_to_use,
+ u32 bytes)
+{
+ struct ena_adapter *adapter = netdev_priv(dev);
+ int rc, nb_hw_desc;
+
+ if (unlikely(ena_com_is_doorbell_needed(ring->ena_com_io_sq,
+ ena_tx_ctx))) {
+ netif_dbg(adapter, tx_queued, dev,
+ "llq tx max burst size of queue %d achieved, writing doorbell to send burst\n",
+ ring->qid);
+ ena_com_write_sq_doorbell(ring->ena_com_io_sq);
+ }
+
+ /* prepare the packet's descriptors to dma engine */
+ rc = ena_com_prepare_tx(ring->ena_com_io_sq, ena_tx_ctx,
+ &nb_hw_desc);
+
+ /* In case there isn't enough space in the queue for the packet,
+ * we simply drop it. All other failure reasons of
+ * ena_com_prepare_tx() are fatal and therefore require a device reset.
+ */
+ if (unlikely(rc)) {
+ netif_err(adapter, tx_queued, dev,
+ "Failed to prepare tx bufs\n");
+ u64_stats_update_begin(&ring->syncp);
+ ring->tx_stats.prepare_ctx_err++;
+ u64_stats_update_end(&ring->syncp);
+ if (rc != -ENOMEM) {
+ adapter->reset_reason =
+ ENA_REGS_RESET_DRIVER_INVALID_STATE;
+ set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags);
+ }
+ return rc;
+ }
+
+ u64_stats_update_begin(&ring->syncp);
+ ring->tx_stats.cnt++;
+ ring->tx_stats.bytes += bytes;
+ u64_stats_update_end(&ring->syncp);
+
+ tx_info->tx_descs = nb_hw_desc;
+ tx_info->last_jiffies = jiffies;
+ tx_info->print_once = 0;
+
+ ring->next_to_use = ENA_TX_RING_IDX_NEXT(next_to_use,
+ ring->ring_size);
+ return 0;
+}
+
+/* This is the XDP napi callback. XDP queues use a separate napi callback
+ * than Rx/Tx queues.
+ */
+static int ena_xdp_io_poll(struct napi_struct *napi, int budget)
+{
+ struct ena_napi *ena_napi = container_of(napi, struct ena_napi, napi);
+ u32 xdp_work_done, xdp_budget;
+ struct ena_ring *xdp_ring;
+ int napi_comp_call = 0;
+ int ret;
+
+ xdp_ring = ena_napi->xdp_ring;
+ xdp_ring->first_interrupt = ena_napi->first_interrupt;
+
+ xdp_budget = budget;
+
+ if (!test_bit(ENA_FLAG_DEV_UP, &xdp_ring->adapter->flags) ||
+ test_bit(ENA_FLAG_TRIGGER_RESET, &xdp_ring->adapter->flags)) {
+ napi_complete_done(napi, 0);
+ return 0;
+ }
+
+ xdp_work_done = ena_clean_xdp_irq(xdp_ring, xdp_budget);
+
+ /* If the device is about to reset or down, avoid unmask
+ * the interrupt and return 0 so NAPI won't reschedule
+ */
+ if (unlikely(!test_bit(ENA_FLAG_DEV_UP, &xdp_ring->adapter->flags))) {
+ napi_complete_done(napi, 0);
+ ret = 0;
+ } else if (xdp_budget > xdp_work_done) {
+ napi_comp_call = 1;
+ if (napi_complete_done(napi, xdp_work_done))
+ ena_unmask_interrupt(xdp_ring, NULL);
+ ena_update_ring_numa_node(xdp_ring, NULL);
+ ret = xdp_work_done;
+ } else {
+ ret = xdp_budget;
+ }
+
+ u64_stats_update_begin(&xdp_ring->syncp);
+ xdp_ring->tx_stats.napi_comp += napi_comp_call;
+ xdp_ring->tx_stats.tx_poll++;
+ u64_stats_update_end(&xdp_ring->syncp);
+
+ return ret;
+}
+
+static int ena_xdp_tx_map_buff(struct ena_ring *xdp_ring,
+ struct ena_tx_buffer *tx_info,
+ struct xdp_buff *xdp,
+ void **push_hdr,
+ u32 *push_len)
+{
+ struct ena_adapter *adapter = xdp_ring->adapter;
+ struct ena_com_buf *ena_buf;
+ dma_addr_t dma = 0;
+ u32 size;
+
+ tx_info->xdpf = xdp_convert_buff_to_frame(xdp);
+ size = tx_info->xdpf->len;
+ ena_buf = tx_info->bufs;
+
+ /* llq push buffer */
+ *push_len = min_t(u32, size, xdp_ring->tx_max_header_size);
+ *push_hdr = tx_info->xdpf->data;
+
+ if (size - *push_len > 0) {
+ dma = dma_map_single(xdp_ring->dev,
+ *push_hdr + *push_len,
+ size - *push_len,
+ DMA_TO_DEVICE);
+ if (unlikely(dma_mapping_error(xdp_ring->dev, dma)))
+ goto error_report_dma_error;
+
+ tx_info->map_linear_data = 1;
+ tx_info->num_of_bufs = 1;
+ }
+
+ ena_buf->paddr = dma;
+ ena_buf->len = size;
+
+ return 0;
+
+error_report_dma_error:
+ u64_stats_update_begin(&xdp_ring->syncp);
+ xdp_ring->tx_stats.dma_mapping_err++;
+ u64_stats_update_end(&xdp_ring->syncp);
+ netif_warn(adapter, tx_queued, adapter->netdev, "Failed to map xdp buff\n");
+
+ xdp_return_frame_rx_napi(tx_info->xdpf);
+ tx_info->xdpf = NULL;
+ tx_info->num_of_bufs = 0;
+
+ return -EINVAL;
+}
+
+static int ena_xdp_xmit_buff(struct net_device *dev,
+ struct xdp_buff *xdp,
+ int qid,
+ struct ena_rx_buffer *rx_info)
+{
+ struct ena_adapter *adapter = netdev_priv(dev);
+ struct ena_com_tx_ctx ena_tx_ctx = {};
+ struct ena_tx_buffer *tx_info;
+ struct ena_ring *xdp_ring;
+ u16 next_to_use, req_id;
+ int rc;
+ void *push_hdr;
+ u32 push_len;
+
+ xdp_ring = &adapter->tx_ring[qid];
+ next_to_use = xdp_ring->next_to_use;
+ req_id = xdp_ring->free_ids[next_to_use];
+ tx_info = &xdp_ring->tx_buffer_info[req_id];
+ tx_info->num_of_bufs = 0;
+ page_ref_inc(rx_info->page);
+ tx_info->xdp_rx_page = rx_info->page;
+
+ rc = ena_xdp_tx_map_buff(xdp_ring, tx_info, xdp, &push_hdr, &push_len);
+ if (unlikely(rc))
+ goto error_drop_packet;
+
+ ena_tx_ctx.ena_bufs = tx_info->bufs;
+ ena_tx_ctx.push_header = push_hdr;
+ ena_tx_ctx.num_bufs = tx_info->num_of_bufs;
+ ena_tx_ctx.req_id = req_id;
+ ena_tx_ctx.header_len = push_len;
+
+ rc = ena_xmit_common(dev,
+ xdp_ring,
+ tx_info,
+ &ena_tx_ctx,
+ next_to_use,
+ xdp->data_end - xdp->data);
+ if (rc)
+ goto error_unmap_dma;
+ /* trigger the dma engine. ena_com_write_sq_doorbell()
+ * has a mb
+ */
+ ena_com_write_sq_doorbell(xdp_ring->ena_com_io_sq);
+ u64_stats_update_begin(&xdp_ring->syncp);
+ xdp_ring->tx_stats.doorbells++;
+ u64_stats_update_end(&xdp_ring->syncp);
+
+ return NETDEV_TX_OK;
+
+error_unmap_dma:
+ ena_unmap_tx_buff(xdp_ring, tx_info);
+ tx_info->xdpf = NULL;
+error_drop_packet:
+ __free_page(tx_info->xdp_rx_page);
+ return NETDEV_TX_OK;
+}
+
+static int ena_xdp_execute(struct ena_ring *rx_ring,
+ struct xdp_buff *xdp,
+ struct ena_rx_buffer *rx_info)
+{
+ struct bpf_prog *xdp_prog;
+ u32 verdict = XDP_PASS;
+ u64 *xdp_stat;
+
+ rcu_read_lock();
+ xdp_prog = READ_ONCE(rx_ring->xdp_bpf_prog);
+
+ if (!xdp_prog)
+ goto out;
+
+ verdict = bpf_prog_run_xdp(xdp_prog, xdp);
+
+ if (verdict == XDP_TX) {
+ ena_xdp_xmit_buff(rx_ring->netdev,
+ xdp,
+ rx_ring->qid + rx_ring->adapter->num_io_queues,
+ rx_info);
+
+ xdp_stat = &rx_ring->rx_stats.xdp_tx;
+ } else if (unlikely(verdict == XDP_ABORTED)) {
+ trace_xdp_exception(rx_ring->netdev, xdp_prog, verdict);
+ xdp_stat = &rx_ring->rx_stats.xdp_aborted;
+ } else if (unlikely(verdict == XDP_DROP)) {
+ xdp_stat = &rx_ring->rx_stats.xdp_drop;
+ } else if (unlikely(verdict == XDP_PASS)) {
+ xdp_stat = &rx_ring->rx_stats.xdp_pass;
+ } else {
+ bpf_warn_invalid_xdp_action(verdict);
+ xdp_stat = &rx_ring->rx_stats.xdp_invalid;
+ }
+
+ u64_stats_update_begin(&rx_ring->syncp);
+ (*xdp_stat)++;
+ u64_stats_update_end(&rx_ring->syncp);
+out:
+ rcu_read_unlock();
+
+ return verdict;
+}
+
+static void ena_init_all_xdp_queues(struct ena_adapter *adapter)
+{
+ adapter->xdp_first_ring = adapter->num_io_queues;
+ adapter->xdp_num_queues = adapter->num_io_queues;
+
+ ena_init_io_rings(adapter,
+ adapter->xdp_first_ring,
+ adapter->xdp_num_queues);
+}
+
+static int ena_setup_and_create_all_xdp_queues(struct ena_adapter *adapter)
+{
+ int rc = 0;
+
+ rc = ena_setup_tx_resources_in_range(adapter, adapter->xdp_first_ring,
+ adapter->xdp_num_queues);
+ if (rc)
+ goto setup_err;
+
+ rc = ena_create_io_tx_queues_in_range(adapter,
+ adapter->xdp_first_ring,
+ adapter->xdp_num_queues);
+ if (rc)
+ goto create_err;
+
+ return 0;
+
+create_err:
+ ena_free_all_io_tx_resources(adapter);
+setup_err:
+ return rc;
+}
+
+/* Provides a way for both kernel and bpf-prog to know
+ * more about the RX-queue a given XDP frame arrived on.
+ */
+static int ena_xdp_register_rxq_info(struct ena_ring *rx_ring)
+{
+ int rc;
+
+ rc = xdp_rxq_info_reg(&rx_ring->xdp_rxq, rx_ring->netdev, rx_ring->qid);
+
+ if (rc) {
+ netif_err(rx_ring->adapter, ifup, rx_ring->netdev,
+ "Failed to register xdp rx queue info. RX queue num %d rc: %d\n",
+ rx_ring->qid, rc);
+ goto err;
+ }
+
+ rc = xdp_rxq_info_reg_mem_model(&rx_ring->xdp_rxq, MEM_TYPE_PAGE_SHARED,
+ NULL);
+
+ if (rc) {
+ netif_err(rx_ring->adapter, ifup, rx_ring->netdev,
+ "Failed to register xdp rx queue info memory model. RX queue num %d rc: %d\n",
+ rx_ring->qid, rc);
+ xdp_rxq_info_unreg(&rx_ring->xdp_rxq);
+ }
+
+err:
+ return rc;
+}
+
+static void ena_xdp_unregister_rxq_info(struct ena_ring *rx_ring)
+{
+ xdp_rxq_info_unreg_mem_model(&rx_ring->xdp_rxq);
+ xdp_rxq_info_unreg(&rx_ring->xdp_rxq);
+}
+
+static void ena_xdp_exchange_program_rx_in_range(struct ena_adapter *adapter,
+ struct bpf_prog *prog,
+ int first, int count)
+{
+ struct ena_ring *rx_ring;
+ int i = 0;
+
+ for (i = first; i < count; i++) {
+ rx_ring = &adapter->rx_ring[i];
+ xchg(&rx_ring->xdp_bpf_prog, prog);
+ if (prog) {
+ ena_xdp_register_rxq_info(rx_ring);
+ rx_ring->rx_headroom = XDP_PACKET_HEADROOM;
+ } else {
+ ena_xdp_unregister_rxq_info(rx_ring);
+ rx_ring->rx_headroom = 0;
+ }
+ }
+}
+
+static void ena_xdp_exchange_program(struct ena_adapter *adapter,
+ struct bpf_prog *prog)
+{
+ struct bpf_prog *old_bpf_prog = xchg(&adapter->xdp_bpf_prog, prog);
+
+ ena_xdp_exchange_program_rx_in_range(adapter,
+ prog,
+ 0,
+ adapter->num_io_queues);
+
+ if (old_bpf_prog)
+ bpf_prog_put(old_bpf_prog);
+}
+
+static int ena_destroy_and_free_all_xdp_queues(struct ena_adapter *adapter)
+{
+ bool was_up;
+ int rc;
+
+ was_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags);
+
+ if (was_up)
+ ena_down(adapter);
+
+ adapter->xdp_first_ring = 0;
+ adapter->xdp_num_queues = 0;
+ ena_xdp_exchange_program(adapter, NULL);
+ if (was_up) {
+ rc = ena_up(adapter);
+ if (rc)
+ return rc;
+ }
+ return 0;
+}
+
+static int ena_xdp_set(struct net_device *netdev, struct netdev_bpf *bpf)
+{
+ struct ena_adapter *adapter = netdev_priv(netdev);
+ struct bpf_prog *prog = bpf->prog;
+ struct bpf_prog *old_bpf_prog;
+ int rc, prev_mtu;
+ bool is_up;
+
+ is_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags);
+ rc = ena_xdp_allowed(adapter);
+ if (rc == ENA_XDP_ALLOWED) {
+ old_bpf_prog = adapter->xdp_bpf_prog;
+ if (prog) {
+ if (!is_up) {
+ ena_init_all_xdp_queues(adapter);
+ } else if (!old_bpf_prog) {
+ ena_down(adapter);
+ ena_init_all_xdp_queues(adapter);
+ }
+ ena_xdp_exchange_program(adapter, prog);
+
+ if (is_up && !old_bpf_prog) {
+ rc = ena_up(adapter);
+ if (rc)
+ return rc;
+ }
+ } else if (old_bpf_prog) {
+ rc = ena_destroy_and_free_all_xdp_queues(adapter);
+ if (rc)
+ return rc;
+ }
+
+ prev_mtu = netdev->max_mtu;
+ netdev->max_mtu = prog ? ENA_XDP_MAX_MTU : adapter->max_mtu;
+
+ if (!old_bpf_prog)
+ netif_info(adapter, drv, adapter->netdev,
+ "XDP program is set, changing the max_mtu from %d to %d",
+ prev_mtu, netdev->max_mtu);
+
+ } else if (rc == ENA_XDP_CURRENT_MTU_TOO_LARGE) {
+ netif_err(adapter, drv, adapter->netdev,
+ "Failed to set xdp program, the current MTU (%d) is larger than the maximum allowed MTU (%lu) while xdp is on",
+ netdev->mtu, ENA_XDP_MAX_MTU);
+ NL_SET_ERR_MSG_MOD(bpf->extack,
+ "Failed to set xdp program, the current MTU is larger than the maximum allowed MTU. Check the dmesg for more info");
+ return -EINVAL;
+ } else if (rc == ENA_XDP_NO_ENOUGH_QUEUES) {
+ netif_err(adapter, drv, adapter->netdev,
+ "Failed to set xdp program, the Rx/Tx channel count should be at most half of the maximum allowed channel count. The current queue count (%d), the maximal queue count (%d)\n",
+ adapter->num_io_queues, adapter->max_num_io_queues);
+ NL_SET_ERR_MSG_MOD(bpf->extack,
+ "Failed to set xdp program, there is no enough space for allocating XDP queues, Check the dmesg for more info");
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+/* This is the main xdp callback, it's used by the kernel to set/unset the xdp
+ * program as well as to query the current xdp program id.
+ */
+static int ena_xdp(struct net_device *netdev, struct netdev_bpf *bpf)
+{
+ switch (bpf->command) {
+ case XDP_SETUP_PROG:
+ return ena_xdp_set(netdev, bpf);
+ default:
+ return -EINVAL;
+ }
+ return 0;
+}
+
static int ena_init_rx_cpu_rmap(struct ena_adapter *adapter)
{
#ifdef CONFIG_RFS_ACCEL
u32 i;
int rc;
- adapter->netdev->rx_cpu_rmap = alloc_irq_cpu_rmap(adapter->num_queues);
+ adapter->netdev->rx_cpu_rmap = alloc_irq_cpu_rmap(adapter->num_io_queues);
if (!adapter->netdev->rx_cpu_rmap)
return -ENOMEM;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues; i++) {
int irq_idx = ENA_IO_IRQ_IDX(i);
rc = irq_cpu_rmap_add(adapter->netdev->rx_cpu_rmap,
@@ -159,14 +608,14 @@
ring->adapter = adapter;
ring->ena_dev = adapter->ena_dev;
ring->per_napi_packets = 0;
- ring->per_napi_bytes = 0;
ring->cpu = 0;
ring->first_interrupt = false;
ring->no_interrupt_event_cnt = 0;
u64_stats_init(&ring->syncp);
}
-static void ena_init_io_rings(struct ena_adapter *adapter)
+static void ena_init_io_rings(struct ena_adapter *adapter,
+ int first_index, int count)
{
struct ena_com_dev *ena_dev;
struct ena_ring *txr, *rxr;
@@ -174,29 +623,36 @@
ena_dev = adapter->ena_dev;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = first_index; i < first_index + count; i++) {
txr = &adapter->tx_ring[i];
rxr = &adapter->rx_ring[i];
- /* TX/RX common ring state */
+ /* TX common ring state */
ena_init_io_rings_common(adapter, txr, i);
- ena_init_io_rings_common(adapter, rxr, i);
/* TX specific ring state */
- txr->ring_size = adapter->tx_ring_size;
+ txr->ring_size = adapter->requested_tx_ring_size;
txr->tx_max_header_size = ena_dev->tx_max_header_size;
txr->tx_mem_queue_type = ena_dev->tx_mem_queue_type;
txr->sgl_size = adapter->max_tx_sgl_size;
txr->smoothed_interval =
ena_com_get_nonadaptive_moderation_interval_tx(ena_dev);
+ txr->disable_meta_caching = adapter->disable_meta_caching;
- /* RX specific ring state */
- rxr->ring_size = adapter->rx_ring_size;
- rxr->rx_copybreak = adapter->rx_copybreak;
- rxr->sgl_size = adapter->max_rx_sgl_size;
- rxr->smoothed_interval =
- ena_com_get_nonadaptive_moderation_interval_rx(ena_dev);
- rxr->empty_rx_queue = 0;
+ /* Don't init RX queues for xdp queues */
+ if (!ENA_IS_XDP_INDEX(adapter, i)) {
+ /* RX common ring state */
+ ena_init_io_rings_common(adapter, rxr, i);
+
+ /* RX specific ring state */
+ rxr->ring_size = adapter->requested_rx_ring_size;
+ rxr->rx_copybreak = adapter->rx_copybreak;
+ rxr->sgl_size = adapter->max_rx_sgl_size;
+ rxr->smoothed_interval =
+ ena_com_get_nonadaptive_moderation_interval_rx(ena_dev);
+ rxr->empty_rx_queue = 0;
+ adapter->ena_napi[i].dim.mode = DIM_CQ_PERIOD_MODE_START_FROM_EQE;
+ }
}
}
@@ -225,22 +681,28 @@
if (!tx_ring->tx_buffer_info) {
tx_ring->tx_buffer_info = vzalloc(size);
if (!tx_ring->tx_buffer_info)
- return -ENOMEM;
+ goto err_tx_buffer_info;
}
size = sizeof(u16) * tx_ring->ring_size;
- tx_ring->free_tx_ids = vzalloc_node(size, node);
- if (!tx_ring->free_tx_ids) {
- tx_ring->free_tx_ids = vzalloc(size);
- if (!tx_ring->free_tx_ids) {
- vfree(tx_ring->tx_buffer_info);
- return -ENOMEM;
- }
+ tx_ring->free_ids = vzalloc_node(size, node);
+ if (!tx_ring->free_ids) {
+ tx_ring->free_ids = vzalloc(size);
+ if (!tx_ring->free_ids)
+ goto err_tx_free_ids;
+ }
+
+ size = tx_ring->tx_max_header_size;
+ tx_ring->push_buf_intermediate_buf = vzalloc_node(size, node);
+ if (!tx_ring->push_buf_intermediate_buf) {
+ tx_ring->push_buf_intermediate_buf = vzalloc(size);
+ if (!tx_ring->push_buf_intermediate_buf)
+ goto err_push_buf_intermediate_buf;
}
/* Req id ring for TX out of order completions */
for (i = 0; i < tx_ring->ring_size; i++)
- tx_ring->free_tx_ids[i] = i;
+ tx_ring->free_ids[i] = i;
/* Reset tx statistics */
memset(&tx_ring->tx_stats, 0x0, sizeof(tx_ring->tx_stats));
@@ -249,6 +711,15 @@
tx_ring->next_to_clean = 0;
tx_ring->cpu = ena_irq->cpu;
return 0;
+
+err_push_buf_intermediate_buf:
+ vfree(tx_ring->free_ids);
+ tx_ring->free_ids = NULL;
+err_tx_free_ids:
+ vfree(tx_ring->tx_buffer_info);
+ tx_ring->tx_buffer_info = NULL;
+err_tx_buffer_info:
+ return -ENOMEM;
}
/* ena_free_tx_resources - Free I/O Tx Resources per Queue
@@ -264,20 +735,20 @@
vfree(tx_ring->tx_buffer_info);
tx_ring->tx_buffer_info = NULL;
- vfree(tx_ring->free_tx_ids);
- tx_ring->free_tx_ids = NULL;
+ vfree(tx_ring->free_ids);
+ tx_ring->free_ids = NULL;
+
+ vfree(tx_ring->push_buf_intermediate_buf);
+ tx_ring->push_buf_intermediate_buf = NULL;
}
-/* ena_setup_all_tx_resources - allocate I/O Tx queues resources for All queues
- * @adapter: private structure
- *
- * Return 0 on success, negative on failure
- */
-static int ena_setup_all_tx_resources(struct ena_adapter *adapter)
+static int ena_setup_tx_resources_in_range(struct ena_adapter *adapter,
+ int first_index,
+ int count)
{
int i, rc = 0;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = first_index; i < first_index + count; i++) {
rc = ena_setup_tx_resources(adapter, i);
if (rc)
goto err_setup_tx;
@@ -291,9 +762,18 @@
"Tx queue %d: allocation failed\n", i);
/* rewind the index freeing the rings as we go */
- while (i--)
+ while (first_index < i--)
ena_free_tx_resources(adapter, i);
return rc;
+}
+
+static void ena_free_all_io_tx_resources_in_range(struct ena_adapter *adapter,
+ int first_index, int count)
+{
+ int i;
+
+ for (i = first_index; i < first_index + count; i++)
+ ena_free_tx_resources(adapter, i);
}
/* ena_free_all_io_tx_resources - Free I/O Tx Resources for All Queues
@@ -303,28 +783,10 @@
*/
static void ena_free_all_io_tx_resources(struct ena_adapter *adapter)
{
- int i;
-
- for (i = 0; i < adapter->num_queues; i++)
- ena_free_tx_resources(adapter, i);
-}
-
-static inline int validate_rx_req_id(struct ena_ring *rx_ring, u16 req_id)
-{
- if (likely(req_id < rx_ring->ring_size))
- return 0;
-
- netif_err(rx_ring->adapter, rx_err, rx_ring->netdev,
- "Invalid rx req_id: %hu\n", req_id);
-
- u64_stats_update_begin(&rx_ring->syncp);
- rx_ring->rx_stats.bad_req_id++;
- u64_stats_update_end(&rx_ring->syncp);
-
- /* Trigger device reset */
- rx_ring->adapter->reset_reason = ENA_REGS_RESET_INV_RX_REQ_ID;
- set_bit(ENA_FLAG_TRIGGER_RESET, &rx_ring->adapter->flags);
- return -EFAULT;
+ ena_free_all_io_tx_resources_in_range(adapter,
+ 0,
+ adapter->xdp_num_queues +
+ adapter->num_io_queues);
}
/* ena_setup_rx_resources - allocate I/O Rx resources (Descriptors)
@@ -360,18 +822,19 @@
}
size = sizeof(u16) * rx_ring->ring_size;
- rx_ring->free_rx_ids = vzalloc_node(size, node);
- if (!rx_ring->free_rx_ids) {
- rx_ring->free_rx_ids = vzalloc(size);
- if (!rx_ring->free_rx_ids) {
+ rx_ring->free_ids = vzalloc_node(size, node);
+ if (!rx_ring->free_ids) {
+ rx_ring->free_ids = vzalloc(size);
+ if (!rx_ring->free_ids) {
vfree(rx_ring->rx_buffer_info);
+ rx_ring->rx_buffer_info = NULL;
return -ENOMEM;
}
}
/* Req id ring for receiving RX pkts out of order */
for (i = 0; i < rx_ring->ring_size; i++)
- rx_ring->free_rx_ids[i] = i;
+ rx_ring->free_ids[i] = i;
/* Reset rx statistics */
memset(&rx_ring->rx_stats, 0x0, sizeof(rx_ring->rx_stats));
@@ -397,8 +860,8 @@
vfree(rx_ring->rx_buffer_info);
rx_ring->rx_buffer_info = NULL;
- vfree(rx_ring->free_rx_ids);
- rx_ring->free_rx_ids = NULL;
+ vfree(rx_ring->free_ids);
+ rx_ring->free_ids = NULL;
}
/* ena_setup_all_rx_resources - allocate I/O Rx queues resources for all queues
@@ -410,7 +873,7 @@
{
int i, rc = 0;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues; i++) {
rc = ena_setup_rx_resources(adapter, i);
if (rc)
goto err_setup_rx;
@@ -438,16 +901,20 @@
{
int i;
- for (i = 0; i < adapter->num_queues; i++)
+ for (i = 0; i < adapter->num_io_queues; i++)
ena_free_rx_resources(adapter, i);
}
-static inline int ena_alloc_rx_page(struct ena_ring *rx_ring,
+static int ena_alloc_rx_page(struct ena_ring *rx_ring,
struct ena_rx_buffer *rx_info, gfp_t gfp)
{
+ int headroom = rx_ring->rx_headroom;
struct ena_com_buf *ena_buf;
struct page *page;
dma_addr_t dma;
+
+ /* restore page offset value in case it has been changed by device */
+ rx_info->page_offset = headroom;
/* if previous allocated page is not used */
if (unlikely(rx_info->page))
@@ -461,8 +928,11 @@
return -ENOMEM;
}
+ /* To enable NIC-side port-mirroring, AKA SPAN port,
+ * we make the buffer readable from the nic as well
+ */
dma = dma_map_page(rx_ring->dev, page, 0, ENA_PAGE_SIZE,
- DMA_FROM_DEVICE);
+ DMA_BIDIRECTIONAL);
if (unlikely(dma_mapping_error(rx_ring->dev, dma))) {
u64_stats_update_begin(&rx_ring->syncp);
rx_ring->rx_stats.dma_mapping_err++;
@@ -472,13 +942,12 @@
return -EIO;
}
netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev,
- "alloc page %p, rx_info %p\n", page, rx_info);
+ "Allocate page %p, rx_info %p\n", page, rx_info);
rx_info->page = page;
- rx_info->page_offset = 0;
ena_buf = &rx_info->ena_buf;
- ena_buf->paddr = dma;
- ena_buf->len = ENA_PAGE_SIZE;
+ ena_buf->paddr = dma + headroom;
+ ena_buf->len = ENA_PAGE_SIZE - headroom;
return 0;
}
@@ -495,8 +964,9 @@
return;
}
- dma_unmap_page(rx_ring->dev, ena_buf->paddr, ENA_PAGE_SIZE,
- DMA_FROM_DEVICE);
+ dma_unmap_page(rx_ring->dev, ena_buf->paddr - rx_ring->rx_headroom,
+ ENA_PAGE_SIZE,
+ DMA_BIDIRECTIONAL);
__free_page(page);
rx_info->page = NULL;
@@ -513,19 +983,15 @@
for (i = 0; i < num; i++) {
struct ena_rx_buffer *rx_info;
- req_id = rx_ring->free_rx_ids[next_to_use];
- rc = validate_rx_req_id(rx_ring, req_id);
- if (unlikely(rc < 0))
- break;
+ req_id = rx_ring->free_ids[next_to_use];
rx_info = &rx_ring->rx_buffer_info[req_id];
-
rc = ena_alloc_rx_page(rx_ring, rx_info,
GFP_ATOMIC | __GFP_COMP);
if (unlikely(rc < 0)) {
netif_warn(rx_ring->adapter, rx_err, rx_ring->netdev,
- "failed to alloc buffer for rx queue %d\n",
+ "Failed to allocate buffer for rx queue %d\n",
rx_ring->qid);
break;
}
@@ -534,7 +1000,7 @@
req_id);
if (unlikely(rc)) {
netif_warn(rx_ring->adapter, rx_status, rx_ring->netdev,
- "failed to add buffer for rx queue %d\n",
+ "Failed to add buffer for rx queue %d\n",
rx_ring->qid);
break;
}
@@ -546,9 +1012,9 @@
u64_stats_update_begin(&rx_ring->syncp);
rx_ring->rx_stats.refil_partial++;
u64_stats_update_end(&rx_ring->syncp);
- netdev_warn(rx_ring->netdev,
- "refilled rx qid %d with only %d buffers (from %d)\n",
- rx_ring->qid, i, num);
+ netif_warn(rx_ring->adapter, rx_err, rx_ring->netdev,
+ "Refilled rx qid %d with only %d buffers (from %d)\n",
+ rx_ring->qid, i, num);
}
/* ena_com_write_sq_doorbell issues a wmb() */
@@ -576,21 +1042,20 @@
/* ena_refill_all_rx_bufs - allocate all queues Rx buffers
* @adapter: board private structure
- *
*/
static void ena_refill_all_rx_bufs(struct ena_adapter *adapter)
{
struct ena_ring *rx_ring;
int i, rc, bufs_num;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues; i++) {
rx_ring = &adapter->rx_ring[i];
bufs_num = rx_ring->ring_size - 1;
rc = ena_refill_rx_bufs(rx_ring, bufs_num);
if (unlikely(rc != bufs_num))
netif_warn(rx_ring->adapter, rx_status, rx_ring->netdev,
- "refilling Queue %d failed. allocated %d buffers from: %d\n",
+ "Refilling Queue %d failed. allocated %d buffers from: %d\n",
i, rc, bufs_num);
}
}
@@ -599,8 +1064,38 @@
{
int i;
- for (i = 0; i < adapter->num_queues; i++)
+ for (i = 0; i < adapter->num_io_queues; i++)
ena_free_rx_bufs(adapter, i);
+}
+
+static void ena_unmap_tx_buff(struct ena_ring *tx_ring,
+ struct ena_tx_buffer *tx_info)
+{
+ struct ena_com_buf *ena_buf;
+ u32 cnt;
+ int i;
+
+ ena_buf = tx_info->bufs;
+ cnt = tx_info->num_of_bufs;
+
+ if (unlikely(!cnt))
+ return;
+
+ if (tx_info->map_linear_data) {
+ dma_unmap_single(tx_ring->dev,
+ dma_unmap_addr(ena_buf, paddr),
+ dma_unmap_len(ena_buf, len),
+ DMA_TO_DEVICE);
+ ena_buf++;
+ cnt--;
+ }
+
+ /* unmap remaining mapped pages */
+ for (i = 0; i < cnt; i++) {
+ dma_unmap_page(tx_ring->dev, dma_unmap_addr(ena_buf, paddr),
+ dma_unmap_len(ena_buf, len), DMA_TO_DEVICE);
+ ena_buf++;
+ }
}
/* ena_free_tx_bufs - Free Tx Buffers per Queue
@@ -613,39 +1108,22 @@
for (i = 0; i < tx_ring->ring_size; i++) {
struct ena_tx_buffer *tx_info = &tx_ring->tx_buffer_info[i];
- struct ena_com_buf *ena_buf;
- int nr_frags;
- int j;
if (!tx_info->skb)
continue;
if (print_once) {
- netdev_notice(tx_ring->netdev,
- "free uncompleted tx skb qid %d idx 0x%x\n",
- tx_ring->qid, i);
+ netif_notice(tx_ring->adapter, ifdown, tx_ring->netdev,
+ "Free uncompleted tx skb qid %d idx 0x%x\n",
+ tx_ring->qid, i);
print_once = false;
} else {
- netdev_dbg(tx_ring->netdev,
- "free uncompleted tx skb qid %d idx 0x%x\n",
- tx_ring->qid, i);
+ netif_dbg(tx_ring->adapter, ifdown, tx_ring->netdev,
+ "Free uncompleted tx skb qid %d idx 0x%x\n",
+ tx_ring->qid, i);
}
- ena_buf = tx_info->bufs;
- dma_unmap_single(tx_ring->dev,
- ena_buf->paddr,
- ena_buf->len,
- DMA_TO_DEVICE);
-
- /* unmap remaining mapped pages */
- nr_frags = tx_info->num_of_bufs - 1;
- for (j = 0; j < nr_frags; j++) {
- ena_buf++;
- dma_unmap_page(tx_ring->dev,
- ena_buf->paddr,
- ena_buf->len,
- DMA_TO_DEVICE);
- }
+ ena_unmap_tx_buff(tx_ring, tx_info);
dev_kfree_skb_any(tx_info->skb);
}
@@ -658,7 +1136,7 @@
struct ena_ring *tx_ring;
int i;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues + adapter->xdp_num_queues; i++) {
tx_ring = &adapter->tx_ring[i];
ena_free_tx_bufs(tx_ring);
}
@@ -669,7 +1147,7 @@
u16 ena_qid;
int i;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues + adapter->xdp_num_queues; i++) {
ena_qid = ENA_IO_TXQ_IDX(i);
ena_com_destroy_io_queue(adapter->ena_dev, ena_qid);
}
@@ -680,8 +1158,9 @@
u16 ena_qid;
int i;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues; i++) {
ena_qid = ENA_IO_RXQ_IDX(i);
+ cancel_work_sync(&adapter->ena_napi[i].dim.work);
ena_com_destroy_io_queue(adapter->ena_dev, ena_qid);
}
}
@@ -692,31 +1171,52 @@
ena_destroy_all_rx_queues(adapter);
}
-static int validate_tx_req_id(struct ena_ring *tx_ring, u16 req_id)
+static int handle_invalid_req_id(struct ena_ring *ring, u16 req_id,
+ struct ena_tx_buffer *tx_info, bool is_xdp)
{
- struct ena_tx_buffer *tx_info = NULL;
-
- if (likely(req_id < tx_ring->ring_size)) {
- tx_info = &tx_ring->tx_buffer_info[req_id];
- if (likely(tx_info->skb))
- return 0;
- }
-
if (tx_info)
- netif_err(tx_ring->adapter, tx_done, tx_ring->netdev,
- "tx_info doesn't have valid skb\n");
+ netif_err(ring->adapter,
+ tx_done,
+ ring->netdev,
+ "tx_info doesn't have valid %s",
+ is_xdp ? "xdp frame" : "skb");
else
- netif_err(tx_ring->adapter, tx_done, tx_ring->netdev,
- "Invalid req_id: %hu\n", req_id);
+ netif_err(ring->adapter,
+ tx_done,
+ ring->netdev,
+ "Invalid req_id: %hu\n",
+ req_id);
- u64_stats_update_begin(&tx_ring->syncp);
- tx_ring->tx_stats.bad_req_id++;
- u64_stats_update_end(&tx_ring->syncp);
+ u64_stats_update_begin(&ring->syncp);
+ ring->tx_stats.bad_req_id++;
+ u64_stats_update_end(&ring->syncp);
/* Trigger device reset */
- tx_ring->adapter->reset_reason = ENA_REGS_RESET_INV_TX_REQ_ID;
- set_bit(ENA_FLAG_TRIGGER_RESET, &tx_ring->adapter->flags);
+ ring->adapter->reset_reason = ENA_REGS_RESET_INV_TX_REQ_ID;
+ set_bit(ENA_FLAG_TRIGGER_RESET, &ring->adapter->flags);
return -EFAULT;
+}
+
+static int validate_tx_req_id(struct ena_ring *tx_ring, u16 req_id)
+{
+ struct ena_tx_buffer *tx_info;
+
+ tx_info = &tx_ring->tx_buffer_info[req_id];
+ if (likely(tx_info->skb))
+ return 0;
+
+ return handle_invalid_req_id(tx_ring, req_id, tx_info, false);
+}
+
+static int validate_xdp_req_id(struct ena_ring *xdp_ring, u16 req_id)
+{
+ struct ena_tx_buffer *tx_info;
+
+ tx_info = &xdp_ring->tx_buffer_info[req_id];
+ if (likely(tx_info->xdpf))
+ return 0;
+
+ return handle_invalid_req_id(xdp_ring, req_id, tx_info, true);
}
static int ena_clean_tx_irq(struct ena_ring *tx_ring, u32 budget)
@@ -736,14 +1236,17 @@
while (tx_pkts < budget) {
struct ena_tx_buffer *tx_info;
struct sk_buff *skb;
- struct ena_com_buf *ena_buf;
- int i, nr_frags;
rc = ena_com_tx_comp_req_id_get(tx_ring->ena_com_io_cq,
&req_id);
- if (rc)
+ if (rc) {
+ if (unlikely(rc == -EINVAL))
+ handle_invalid_req_id(tx_ring, req_id, NULL,
+ false);
break;
+ }
+ /* validate that the request id points to a valid skb */
rc = validate_tx_req_id(tx_ring, req_id);
if (rc)
break;
@@ -757,24 +1260,7 @@
tx_info->skb = NULL;
tx_info->last_jiffies = 0;
- if (likely(tx_info->num_of_bufs != 0)) {
- ena_buf = tx_info->bufs;
-
- dma_unmap_single(tx_ring->dev,
- dma_unmap_addr(ena_buf, paddr),
- dma_unmap_len(ena_buf, len),
- DMA_TO_DEVICE);
-
- /* unmap remaining mapped pages */
- nr_frags = tx_info->num_of_bufs - 1;
- for (i = 0; i < nr_frags; i++) {
- ena_buf++;
- dma_unmap_page(tx_ring->dev,
- dma_unmap_addr(ena_buf, paddr),
- dma_unmap_len(ena_buf, len),
- DMA_TO_DEVICE);
- }
- }
+ ena_unmap_tx_buff(tx_ring, tx_info);
netif_dbg(tx_ring->adapter, tx_done, tx_ring->netdev,
"tx_poll: q %d skb %p completed\n", tx_ring->qid,
@@ -785,7 +1271,7 @@
tx_pkts++;
total_done += tx_info->tx_descs;
- tx_ring->free_tx_ids[next_to_clean] = req_id;
+ tx_ring->free_ids[next_to_clean] = req_id;
next_to_clean = ENA_TX_RING_IDX_NEXT(next_to_clean,
tx_ring->ring_size);
}
@@ -805,13 +1291,15 @@
*/
smp_mb();
- above_thresh = ena_com_sq_empty_space(tx_ring->ena_com_io_sq) >
- ENA_TX_WAKEUP_THRESH;
+ above_thresh = ena_com_sq_have_enough_space(tx_ring->ena_com_io_sq,
+ ENA_TX_WAKEUP_THRESH);
if (unlikely(netif_tx_queue_stopped(txq) && above_thresh)) {
__netif_tx_lock(txq, smp_processor_id());
- above_thresh = ena_com_sq_empty_space(tx_ring->ena_com_io_sq) >
- ENA_TX_WAKEUP_THRESH;
- if (netif_tx_queue_stopped(txq) && above_thresh) {
+ above_thresh =
+ ena_com_sq_have_enough_space(tx_ring->ena_com_io_sq,
+ ENA_TX_WAKEUP_THRESH);
+ if (netif_tx_queue_stopped(txq) && above_thresh &&
+ test_bit(ENA_FLAG_DEV_UP, &tx_ring->adapter->flags)) {
netif_tx_wake_queue(txq);
u64_stats_update_begin(&tx_ring->syncp);
tx_ring->tx_stats.queue_wakeup++;
@@ -819,9 +1307,6 @@
}
__netif_tx_unlock(txq);
}
-
- tx_ring->per_napi_bytes += tx_bytes;
- tx_ring->per_napi_packets += tx_pkts;
return tx_pkts;
}
@@ -860,6 +1345,7 @@
len = ena_bufs[buf].len;
req_id = ena_bufs[buf].req_id;
+
rx_info = &rx_ring->rx_buffer_info[req_id];
if (unlikely(!rx_info->page)) {
@@ -874,7 +1360,8 @@
/* save virt address of first buffer */
va = page_address(rx_info->page) + rx_info->page_offset;
- prefetch(va + NET_IP_ALIGN);
+
+ prefetch(va);
if (len <= rx_ring->rx_copybreak) {
skb = ena_alloc_skb(rx_ring, false);
@@ -882,7 +1369,7 @@
return NULL;
netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev,
- "rx allocated small packet. len %d. data_len %d\n",
+ "RX allocated small packet. len %d. data_len %d\n",
skb->len, skb->data_len);
/* sync this buffer for CPU use */
@@ -898,7 +1385,7 @@
skb_put(skb, len);
skb->protocol = eth_type_trans(skb, rx_ring->netdev);
- rx_ring->free_rx_ids[*next_to_clean] = req_id;
+ rx_ring->free_ids[*next_to_clean] = req_id;
*next_to_clean = ENA_RX_RING_IDX_ADD(*next_to_clean, descs,
rx_ring->ring_size);
return skb;
@@ -911,18 +1398,18 @@
do {
dma_unmap_page(rx_ring->dev,
dma_unmap_addr(&rx_info->ena_buf, paddr),
- ENA_PAGE_SIZE, DMA_FROM_DEVICE);
+ ENA_PAGE_SIZE, DMA_BIDIRECTIONAL);
skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_info->page,
rx_info->page_offset, len, ENA_PAGE_SIZE);
netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev,
- "rx skb updated. len %d. data_len %d\n",
+ "RX skb updated. len %d. data_len %d\n",
skb->len, skb->data_len);
rx_info->page = NULL;
- rx_ring->free_rx_ids[*next_to_clean] = req_id;
+ rx_ring->free_ids[*next_to_clean] = req_id;
*next_to_clean =
ENA_RX_RING_IDX_NEXT(*next_to_clean,
rx_ring->ring_size);
@@ -932,6 +1419,7 @@
buf++;
len = ena_bufs[buf].len;
req_id = ena_bufs[buf].req_id;
+
rx_info = &rx_ring->rx_buffer_info[req_id];
} while (1);
@@ -943,7 +1431,7 @@
* @ena_rx_ctx: received packet context/metadata
* @skb: skb currently being received and modified
*/
-static inline void ena_rx_checksum(struct ena_ring *rx_ring,
+static void ena_rx_checksum(struct ena_ring *rx_ring,
struct ena_com_rx_ctx *ena_rx_ctx,
struct sk_buff *skb)
{
@@ -986,8 +1474,22 @@
return;
}
- skb->ip_summed = CHECKSUM_UNNECESSARY;
+ if (likely(ena_rx_ctx->l4_csum_checked)) {
+ skb->ip_summed = CHECKSUM_UNNECESSARY;
+ u64_stats_update_begin(&rx_ring->syncp);
+ rx_ring->rx_stats.csum_good++;
+ u64_stats_update_end(&rx_ring->syncp);
+ } else {
+ u64_stats_update_begin(&rx_ring->syncp);
+ rx_ring->rx_stats.csum_unchecked++;
+ u64_stats_update_end(&rx_ring->syncp);
+ skb->ip_summed = CHECKSUM_NONE;
+ }
+ } else {
+ skb->ip_summed = CHECKSUM_NONE;
+ return;
}
+
}
static void ena_set_rx_hash(struct ena_ring *rx_ring,
@@ -1012,6 +1514,32 @@
}
}
+static int ena_xdp_handle_buff(struct ena_ring *rx_ring, struct xdp_buff *xdp)
+{
+ struct ena_rx_buffer *rx_info;
+ int ret;
+
+ rx_info = &rx_ring->rx_buffer_info[rx_ring->ena_bufs[0].req_id];
+ xdp->data = page_address(rx_info->page) + rx_info->page_offset;
+ xdp_set_data_meta_invalid(xdp);
+ xdp->data_hard_start = page_address(rx_info->page);
+ xdp->data_end = xdp->data + rx_ring->ena_bufs[0].len;
+ /* If for some reason we received a bigger packet than
+ * we expect, then we simply drop it
+ */
+ if (unlikely(rx_ring->ena_bufs[0].len > ENA_XDP_MAX_MTU))
+ return XDP_DROP;
+
+ ret = ena_xdp_execute(rx_ring, xdp, rx_info);
+
+ /* The xdp program might expand the headers */
+ if (ret == XDP_PASS) {
+ rx_info->page_offset = xdp->data - xdp->data_hard_start;
+ rx_ring->ena_bufs[0].len = xdp->data_end - xdp->data;
+ }
+
+ return ret;
+}
/* ena_clean_rx_irq - Cleanup RX irq
* @rx_ring: RX ring to clean
* @napi: napi handler
@@ -1023,26 +1551,33 @@
u32 budget)
{
u16 next_to_clean = rx_ring->next_to_clean;
- u32 res_budget, work_done;
-
struct ena_com_rx_ctx ena_rx_ctx;
+ struct ena_rx_buffer *rx_info;
struct ena_adapter *adapter;
+ u32 res_budget, work_done;
+ int rx_copybreak_pkt = 0;
+ int refill_threshold;
struct sk_buff *skb;
int refill_required;
- int refill_threshold;
- int rc = 0;
+ struct xdp_buff xdp;
int total_len = 0;
- int rx_copybreak_pkt = 0;
+ int xdp_verdict;
+ int rc = 0;
int i;
netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev,
"%s qid %d\n", __func__, rx_ring->qid);
res_budget = budget;
+ xdp.rxq = &rx_ring->xdp_rxq;
+ xdp.frame_sz = ENA_PAGE_SIZE;
do {
+ xdp_verdict = XDP_PASS;
+ skb = NULL;
ena_rx_ctx.ena_bufs = rx_ring->ena_bufs;
ena_rx_ctx.max_bufs = rx_ring->sgl_size;
ena_rx_ctx.descs = 0;
+ ena_rx_ctx.pkt_offset = 0;
rc = ena_com_rx_pkt(rx_ring->ena_com_io_cq,
rx_ring->ena_com_io_sq,
&ena_rx_ctx);
@@ -1052,23 +1587,44 @@
if (unlikely(ena_rx_ctx.descs == 0))
break;
+ /* First descriptor might have an offset set by the device */
+ rx_info = &rx_ring->rx_buffer_info[rx_ring->ena_bufs[0].req_id];
+ rx_info->page_offset += ena_rx_ctx.pkt_offset;
+
netif_dbg(rx_ring->adapter, rx_status, rx_ring->netdev,
"rx_poll: q %d got packet from ena. descs #: %d l3 proto %d l4 proto %d hash: %x\n",
rx_ring->qid, ena_rx_ctx.descs, ena_rx_ctx.l3_proto,
ena_rx_ctx.l4_proto, ena_rx_ctx.hash);
- /* allocate skb and fill it */
- skb = ena_rx_skb(rx_ring, rx_ring->ena_bufs, ena_rx_ctx.descs,
- &next_to_clean);
+ if (ena_xdp_present_ring(rx_ring))
+ xdp_verdict = ena_xdp_handle_buff(rx_ring, &xdp);
- /* exit if we failed to retrieve a buffer */
+ /* allocate skb and fill it */
+ if (xdp_verdict == XDP_PASS)
+ skb = ena_rx_skb(rx_ring,
+ rx_ring->ena_bufs,
+ ena_rx_ctx.descs,
+ &next_to_clean);
+
if (unlikely(!skb)) {
+ /* The page might not actually be freed here since the
+ * page reference count is incremented in
+ * ena_xdp_xmit_buff(), and it will be decreased only
+ * when send completion was received from the device
+ */
+ if (xdp_verdict == XDP_TX)
+ ena_free_rx_page(rx_ring,
+ &rx_ring->rx_buffer_info[rx_ring->ena_bufs[0].req_id]);
for (i = 0; i < ena_rx_ctx.descs; i++) {
- rx_ring->free_tx_ids[next_to_clean] =
+ rx_ring->free_ids[next_to_clean] =
rx_ring->ena_bufs[i].req_id;
next_to_clean =
ENA_RX_RING_IDX_NEXT(next_to_clean,
rx_ring->ring_size);
+ }
+ if (xdp_verdict != XDP_PASS) {
+ res_budget--;
+ continue;
}
break;
}
@@ -1092,7 +1648,6 @@
} while (likely(res_budget));
work_done = budget - res_budget;
- rx_ring->per_napi_bytes += total_len;
rx_ring->per_napi_packets += work_done;
u64_stats_update_begin(&rx_ring->syncp);
rx_ring->rx_stats.bytes += total_len;
@@ -1102,8 +1657,10 @@
rx_ring->next_to_clean = next_to_clean;
- refill_required = ena_com_sq_empty_space(rx_ring->ena_com_io_sq);
- refill_threshold = rx_ring->ring_size / ENA_RX_REFILL_THRESH_DIVIDER;
+ refill_required = ena_com_free_q_entries(rx_ring->ena_com_io_sq);
+ refill_threshold =
+ min_t(int, rx_ring->ring_size / ENA_RX_REFILL_THRESH_DIVIDER,
+ ENA_RX_REFILL_THRESH_PACKET);
/* Optimization, try to batch new rx buffers */
if (refill_required > refill_threshold) {
@@ -1116,57 +1673,88 @@
error:
adapter = netdev_priv(rx_ring->netdev);
- u64_stats_update_begin(&rx_ring->syncp);
- rx_ring->rx_stats.bad_desc_num++;
- u64_stats_update_end(&rx_ring->syncp);
+ if (rc == -ENOSPC) {
+ u64_stats_update_begin(&rx_ring->syncp);
+ rx_ring->rx_stats.bad_desc_num++;
+ u64_stats_update_end(&rx_ring->syncp);
+ adapter->reset_reason = ENA_REGS_RESET_TOO_MANY_RX_DESCS;
+ } else {
+ u64_stats_update_begin(&rx_ring->syncp);
+ rx_ring->rx_stats.bad_req_id++;
+ u64_stats_update_end(&rx_ring->syncp);
+ adapter->reset_reason = ENA_REGS_RESET_INV_RX_REQ_ID;
+ }
- /* Too many desc from the device. Trigger reset */
- adapter->reset_reason = ENA_REGS_RESET_TOO_MANY_RX_DESCS;
set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags);
return 0;
}
-inline void ena_adjust_intr_moderation(struct ena_ring *rx_ring,
- struct ena_ring *tx_ring)
+static void ena_dim_work(struct work_struct *w)
{
- /* We apply adaptive moderation on Rx path only.
- * Tx uses static interrupt moderation.
- */
- ena_com_calculate_interrupt_delay(rx_ring->ena_dev,
- rx_ring->per_napi_packets,
- rx_ring->per_napi_bytes,
- &rx_ring->smoothed_interval,
- &rx_ring->moder_tbl_idx);
+ struct dim *dim = container_of(w, struct dim, work);
+ struct dim_cq_moder cur_moder =
+ net_dim_get_rx_moderation(dim->mode, dim->profile_ix);
+ struct ena_napi *ena_napi = container_of(dim, struct ena_napi, dim);
- /* Reset per napi packets/bytes */
- tx_ring->per_napi_packets = 0;
- tx_ring->per_napi_bytes = 0;
- rx_ring->per_napi_packets = 0;
- rx_ring->per_napi_bytes = 0;
+ ena_napi->rx_ring->smoothed_interval = cur_moder.usec;
+ dim->state = DIM_START_MEASURE;
}
-static inline void ena_unmask_interrupt(struct ena_ring *tx_ring,
+static void ena_adjust_adaptive_rx_intr_moderation(struct ena_napi *ena_napi)
+{
+ struct dim_sample dim_sample;
+ struct ena_ring *rx_ring = ena_napi->rx_ring;
+
+ if (!rx_ring->per_napi_packets)
+ return;
+
+ rx_ring->non_empty_napi_events++;
+
+ dim_update_sample(rx_ring->non_empty_napi_events,
+ rx_ring->rx_stats.cnt,
+ rx_ring->rx_stats.bytes,
+ &dim_sample);
+
+ net_dim(&ena_napi->dim, dim_sample);
+
+ rx_ring->per_napi_packets = 0;
+}
+
+static void ena_unmask_interrupt(struct ena_ring *tx_ring,
struct ena_ring *rx_ring)
{
struct ena_eth_io_intr_reg intr_reg;
+ u32 rx_interval = 0;
+ /* Rx ring can be NULL when for XDP tx queues which don't have an
+ * accompanying rx_ring pair.
+ */
+ if (rx_ring)
+ rx_interval = ena_com_get_adaptive_moderation_enabled(rx_ring->ena_dev) ?
+ rx_ring->smoothed_interval :
+ ena_com_get_nonadaptive_moderation_interval_rx(rx_ring->ena_dev);
/* Update intr register: rx intr delay,
* tx intr delay and interrupt unmask
*/
ena_com_update_intr_reg(&intr_reg,
- rx_ring->smoothed_interval,
+ rx_interval,
tx_ring->smoothed_interval,
true);
+
+ u64_stats_update_begin(&tx_ring->syncp);
+ tx_ring->tx_stats.unmask_interrupt++;
+ u64_stats_update_end(&tx_ring->syncp);
/* It is a shared MSI-X.
* Tx and Rx CQ have pointer to it.
* So we use one of them to reach the intr reg
+ * The Tx ring is used because the rx_ring is NULL for XDP queues
*/
- ena_com_unmask_intr(rx_ring->ena_com_io_cq, &intr_reg);
+ ena_com_unmask_intr(tx_ring->ena_com_io_cq, &intr_reg);
}
-static inline void ena_update_ring_numa_node(struct ena_ring *tx_ring,
+static void ena_update_ring_numa_node(struct ena_ring *tx_ring,
struct ena_ring *rx_ring)
{
int cpu = get_cpu();
@@ -1181,22 +1769,87 @@
if (numa_node != NUMA_NO_NODE) {
ena_com_update_numa_node(tx_ring->ena_com_io_cq, numa_node);
- ena_com_update_numa_node(rx_ring->ena_com_io_cq, numa_node);
+ if (rx_ring)
+ ena_com_update_numa_node(rx_ring->ena_com_io_cq,
+ numa_node);
}
tx_ring->cpu = cpu;
- rx_ring->cpu = cpu;
+ if (rx_ring)
+ rx_ring->cpu = cpu;
return;
out:
put_cpu();
}
+static int ena_clean_xdp_irq(struct ena_ring *xdp_ring, u32 budget)
+{
+ u32 total_done = 0;
+ u16 next_to_clean;
+ u32 tx_bytes = 0;
+ int tx_pkts = 0;
+ u16 req_id;
+ int rc;
+
+ if (unlikely(!xdp_ring))
+ return 0;
+ next_to_clean = xdp_ring->next_to_clean;
+
+ while (tx_pkts < budget) {
+ struct ena_tx_buffer *tx_info;
+ struct xdp_frame *xdpf;
+
+ rc = ena_com_tx_comp_req_id_get(xdp_ring->ena_com_io_cq,
+ &req_id);
+ if (rc) {
+ if (unlikely(rc == -EINVAL))
+ handle_invalid_req_id(xdp_ring, req_id, NULL,
+ true);
+ break;
+ }
+
+ /* validate that the request id points to a valid xdp_frame */
+ rc = validate_xdp_req_id(xdp_ring, req_id);
+ if (rc)
+ break;
+
+ tx_info = &xdp_ring->tx_buffer_info[req_id];
+ xdpf = tx_info->xdpf;
+
+ tx_info->xdpf = NULL;
+ tx_info->last_jiffies = 0;
+ ena_unmap_tx_buff(xdp_ring, tx_info);
+
+ netif_dbg(xdp_ring->adapter, tx_done, xdp_ring->netdev,
+ "tx_poll: q %d skb %p completed\n", xdp_ring->qid,
+ xdpf);
+
+ tx_bytes += xdpf->len;
+ tx_pkts++;
+ total_done += tx_info->tx_descs;
+
+ __free_page(tx_info->xdp_rx_page);
+ xdp_ring->free_ids[next_to_clean] = req_id;
+ next_to_clean = ENA_TX_RING_IDX_NEXT(next_to_clean,
+ xdp_ring->ring_size);
+ }
+
+ xdp_ring->next_to_clean = next_to_clean;
+ ena_com_comp_ack(xdp_ring->ena_com_io_sq, total_done);
+ ena_com_update_dev_comp_head(xdp_ring->ena_com_io_cq);
+
+ netif_dbg(xdp_ring->adapter, tx_done, xdp_ring->netdev,
+ "tx_poll: q %d done. total pkts: %d\n",
+ xdp_ring->qid, tx_pkts);
+
+ return tx_pkts;
+}
+
static int ena_io_poll(struct napi_struct *napi, int budget)
{
struct ena_napi *ena_napi = container_of(napi, struct ena_napi, napi);
struct ena_ring *tx_ring, *rx_ring;
-
int tx_work_done;
int rx_work_done = 0;
int tx_budget;
@@ -1205,6 +1858,9 @@
tx_ring = ena_napi->tx_ring;
rx_ring = ena_napi->rx_ring;
+
+ tx_ring->first_interrupt = ena_napi->first_interrupt;
+ rx_ring->first_interrupt = ena_napi->first_interrupt;
tx_budget = tx_ring->ring_size / ENA_TX_POLL_BUDGET_DIVIDER;
@@ -1235,10 +1891,15 @@
/* Update numa and unmask the interrupt only when schedule
* from the interrupt context (vs from sk_busy_loop)
*/
- if (napi_complete_done(napi, rx_work_done)) {
- /* Tx and Rx share the same interrupt vector */
+ if (napi_complete_done(napi, rx_work_done) &&
+ READ_ONCE(ena_napi->interrupts_masked)) {
+ smp_rmb(); /* make sure interrupts_masked is read */
+ WRITE_ONCE(ena_napi->interrupts_masked, false);
+ /* We apply adaptive moderation on Rx path only.
+ * Tx uses static interrupt moderation.
+ */
if (ena_com_get_adaptive_moderation_enabled(rx_ring->ena_dev))
- ena_adjust_intr_moderation(rx_ring, tx_ring);
+ ena_adjust_adaptive_rx_intr_moderation(ena_napi);
ena_unmask_interrupt(tx_ring, rx_ring);
}
@@ -1279,8 +1940,10 @@
{
struct ena_napi *ena_napi = data;
- ena_napi->tx_ring->first_interrupt = true;
- ena_napi->rx_ring->first_interrupt = true;
+ ena_napi->first_interrupt = true;
+
+ WRITE_ONCE(ena_napi->interrupts_masked, true);
+ smp_wmb(); /* write interrupts_masked before calling napi */
napi_schedule_irqoff(&ena_napi->napi);
@@ -1292,7 +1955,7 @@
* the number of potential io queues is the minimum of what the device
* supports and the number of vCPUs.
*/
-static int ena_enable_msix(struct ena_adapter *adapter, int num_queues)
+static int ena_enable_msix(struct ena_adapter *adapter)
{
int msix_vecs, irq_cnt;
@@ -1303,10 +1966,9 @@
}
/* Reserved the max msix vectors we might need */
- msix_vecs = ENA_MAX_MSIX_VEC(num_queues);
-
+ msix_vecs = ENA_MAX_MSIX_VEC(adapter->max_num_io_queues);
netif_dbg(adapter, probe, adapter->netdev,
- "trying to enable MSI-X, vectors %d\n", msix_vecs);
+ "Trying to enable MSI-X, vectors %d\n", msix_vecs);
irq_cnt = pci_alloc_irq_vectors(adapter->pdev, ENA_MIN_MSIX_VEC,
msix_vecs, PCI_IRQ_MSIX);
@@ -1319,9 +1981,9 @@
if (irq_cnt != msix_vecs) {
netif_notice(adapter, probe, adapter->netdev,
- "enable only %d MSI-X (out of %d), reduce the number of queues\n",
+ "Enable only %d MSI-X (out of %d), reduce the number of queues\n",
irq_cnt, msix_vecs);
- adapter->num_queues = irq_cnt - ENA_ADMIN_MSIX_VEC;
+ adapter->num_io_queues = irq_cnt - ENA_ADMIN_MSIX_VEC;
}
if (ena_init_rx_cpu_rmap(adapter))
@@ -1356,10 +2018,12 @@
{
struct net_device *netdev;
int irq_idx, i, cpu;
+ int io_queue_count;
netdev = adapter->netdev;
+ io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < io_queue_count; i++) {
irq_idx = ENA_IO_IRQ_IDX(i);
cpu = i % num_online_cpus();
@@ -1387,12 +2051,12 @@
irq->data);
if (rc) {
netif_err(adapter, probe, adapter->netdev,
- "failed to request admin irq\n");
+ "Failed to request admin irq\n");
return rc;
}
netif_dbg(adapter, probe, adapter->netdev,
- "set affinity hint of mgmnt irq.to 0x%lx (irq vector: %d)\n",
+ "Set affinity hint of mgmnt irq.to 0x%lx (irq vector: %d)\n",
irq->affinity_hint_mask.bits[0], irq->vector);
irq_set_affinity_hint(irq->vector, &irq->affinity_hint_mask);
@@ -1402,6 +2066,7 @@
static int ena_request_io_irq(struct ena_adapter *adapter)
{
+ u32 io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues;
unsigned long flags = 0;
struct ena_irq *irq;
int rc = 0, i, k;
@@ -1412,7 +2077,7 @@
return -EINVAL;
}
- for (i = ENA_IO_IRQ_FIRST_IDX; i < adapter->msix_vecs; i++) {
+ for (i = ENA_IO_IRQ_FIRST_IDX; i < ENA_MAX_MSIX_VEC(io_queue_count); i++) {
irq = &adapter->irq_tbl[i];
rc = request_irq(irq->vector, irq->handler, flags, irq->name,
irq->data);
@@ -1424,7 +2089,7 @@
}
netif_dbg(adapter, ifup, adapter->netdev,
- "set affinity hint of irq. index %d to 0x%lx (irq vector: %d)\n",
+ "Set affinity hint of irq. index %d to 0x%lx (irq vector: %d)\n",
i, irq->affinity_hint_mask.bits[0], irq->vector);
irq_set_affinity_hint(irq->vector, &irq->affinity_hint_mask);
@@ -1453,6 +2118,7 @@
static void ena_free_io_irq(struct ena_adapter *adapter)
{
+ u32 io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues;
struct ena_irq *irq;
int i;
@@ -1463,7 +2129,7 @@
}
#endif /* CONFIG_RFS_ACCEL */
- for (i = ENA_IO_IRQ_FIRST_IDX; i < adapter->msix_vecs; i++) {
+ for (i = ENA_IO_IRQ_FIRST_IDX; i < ENA_MAX_MSIX_VEC(io_queue_count); i++) {
irq = &adapter->irq_tbl[i];
irq_set_affinity_hint(irq->vector, NULL);
free_irq(irq->vector, irq->data);
@@ -1478,63 +2144,71 @@
static void ena_disable_io_intr_sync(struct ena_adapter *adapter)
{
+ u32 io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues;
int i;
if (!netif_running(adapter->netdev))
return;
- for (i = ENA_IO_IRQ_FIRST_IDX; i < adapter->msix_vecs; i++)
+ for (i = ENA_IO_IRQ_FIRST_IDX; i < ENA_MAX_MSIX_VEC(io_queue_count); i++)
synchronize_irq(adapter->irq_tbl[i].vector);
}
-static void ena_del_napi(struct ena_adapter *adapter)
+static void ena_del_napi_in_range(struct ena_adapter *adapter,
+ int first_index,
+ int count)
{
int i;
- for (i = 0; i < adapter->num_queues; i++)
+ for (i = first_index; i < first_index + count; i++) {
netif_napi_del(&adapter->ena_napi[i].napi);
+
+ WARN_ON(!ENA_IS_XDP_INDEX(adapter, i) &&
+ adapter->ena_napi[i].xdp_ring);
+ }
}
-static void ena_init_napi(struct ena_adapter *adapter)
+static void ena_init_napi_in_range(struct ena_adapter *adapter,
+ int first_index, int count)
{
- struct ena_napi *napi;
int i;
- for (i = 0; i < adapter->num_queues; i++) {
- napi = &adapter->ena_napi[i];
+ for (i = first_index; i < first_index + count; i++) {
+ struct ena_napi *napi = &adapter->ena_napi[i];
netif_napi_add(adapter->netdev,
- &adapter->ena_napi[i].napi,
- ena_io_poll,
+ &napi->napi,
+ ENA_IS_XDP_INDEX(adapter, i) ? ena_xdp_io_poll : ena_io_poll,
ENA_NAPI_BUDGET);
- napi->rx_ring = &adapter->rx_ring[i];
- napi->tx_ring = &adapter->tx_ring[i];
+
+ if (!ENA_IS_XDP_INDEX(adapter, i)) {
+ napi->rx_ring = &adapter->rx_ring[i];
+ napi->tx_ring = &adapter->tx_ring[i];
+ } else {
+ napi->xdp_ring = &adapter->tx_ring[i];
+ }
napi->qid = i;
}
}
-static void ena_napi_disable_all(struct ena_adapter *adapter)
+static void ena_napi_disable_in_range(struct ena_adapter *adapter,
+ int first_index,
+ int count)
{
int i;
- for (i = 0; i < adapter->num_queues; i++)
+ for (i = first_index; i < first_index + count; i++)
napi_disable(&adapter->ena_napi[i].napi);
}
-static void ena_napi_enable_all(struct ena_adapter *adapter)
+static void ena_napi_enable_in_range(struct ena_adapter *adapter,
+ int first_index,
+ int count)
{
int i;
- for (i = 0; i < adapter->num_queues; i++)
+ for (i = first_index; i < first_index + count; i++)
napi_enable(&adapter->ena_napi[i].napi);
-}
-
-static void ena_restore_ethtool_params(struct ena_adapter *adapter)
-{
- adapter->tx_usecs = 0;
- adapter->rx_usecs = 0;
- adapter->tx_frames = 1;
- adapter->rx_frames = 1;
}
/* Configure the Rx forwarding */
@@ -1586,16 +2260,16 @@
/* enable transmits */
netif_tx_start_all_queues(adapter->netdev);
- ena_restore_ethtool_params(adapter);
-
- ena_napi_enable_all(adapter);
+ ena_napi_enable_in_range(adapter,
+ 0,
+ adapter->xdp_num_queues + adapter->num_io_queues);
return 0;
}
static int ena_create_io_tx_queue(struct ena_adapter *adapter, int qid)
{
- struct ena_com_create_io_ctx ctx = { 0 };
+ struct ena_com_create_io_ctx ctx;
struct ena_com_dev *ena_dev;
struct ena_ring *tx_ring;
u32 msix_vector;
@@ -1608,11 +2282,13 @@
msix_vector = ENA_IO_IRQ_IDX(qid);
ena_qid = ENA_IO_TXQ_IDX(qid);
+ memset(&ctx, 0x0, sizeof(ctx));
+
ctx.direction = ENA_COM_IO_QUEUE_DIRECTION_TX;
ctx.qid = ena_qid;
ctx.mem_queue_type = ena_dev->tx_mem_queue_type;
ctx.msix_vector = msix_vector;
- ctx.queue_size = adapter->tx_ring_size;
+ ctx.queue_size = tx_ring->ring_size;
ctx.numa_node = cpu_to_node(tx_ring->cpu);
rc = ena_com_create_io_queue(ena_dev, &ctx);
@@ -1638,12 +2314,13 @@
return rc;
}
-static int ena_create_all_io_tx_queues(struct ena_adapter *adapter)
+static int ena_create_io_tx_queues_in_range(struct ena_adapter *adapter,
+ int first_index, int count)
{
struct ena_com_dev *ena_dev = adapter->ena_dev;
int rc, i;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = first_index; i < first_index + count; i++) {
rc = ena_create_io_tx_queue(adapter, i);
if (rc)
goto create_err;
@@ -1652,7 +2329,7 @@
return 0;
create_err:
- while (i--)
+ while (i-- > first_index)
ena_com_destroy_io_queue(ena_dev, ENA_IO_TXQ_IDX(i));
return rc;
@@ -1661,7 +2338,7 @@
static int ena_create_io_rx_queue(struct ena_adapter *adapter, int qid)
{
struct ena_com_dev *ena_dev;
- struct ena_com_create_io_ctx ctx = { 0 };
+ struct ena_com_create_io_ctx ctx;
struct ena_ring *rx_ring;
u32 msix_vector;
u16 ena_qid;
@@ -1673,11 +2350,13 @@
msix_vector = ENA_IO_IRQ_IDX(qid);
ena_qid = ENA_IO_RXQ_IDX(qid);
+ memset(&ctx, 0x0, sizeof(ctx));
+
ctx.qid = ena_qid;
ctx.direction = ENA_COM_IO_QUEUE_DIRECTION_RX;
ctx.mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
ctx.msix_vector = msix_vector;
- ctx.queue_size = adapter->rx_ring_size;
+ ctx.queue_size = rx_ring->ring_size;
ctx.numa_node = cpu_to_node(rx_ring->cpu);
rc = ena_com_create_io_queue(ena_dev, &ctx);
@@ -1695,12 +2374,14 @@
netif_err(adapter, ifup, adapter->netdev,
"Failed to get RX queue handlers. RX queue num %d rc: %d\n",
qid, rc);
- ena_com_destroy_io_queue(ena_dev, ena_qid);
- return rc;
+ goto err;
}
ena_com_update_numa_node(rx_ring->ena_com_io_cq, ctx.numa_node);
+ return rc;
+err:
+ ena_com_destroy_io_queue(ena_dev, ena_qid);
return rc;
}
@@ -1709,27 +2390,148 @@
struct ena_com_dev *ena_dev = adapter->ena_dev;
int rc, i;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues; i++) {
rc = ena_create_io_rx_queue(adapter, i);
if (rc)
goto create_err;
+ INIT_WORK(&adapter->ena_napi[i].dim.work, ena_dim_work);
}
return 0;
create_err:
- while (i--)
+ while (i--) {
+ cancel_work_sync(&adapter->ena_napi[i].dim.work);
ena_com_destroy_io_queue(ena_dev, ENA_IO_RXQ_IDX(i));
+ }
return rc;
}
+static void set_io_rings_size(struct ena_adapter *adapter,
+ int new_tx_size,
+ int new_rx_size)
+{
+ int i;
+
+ for (i = 0; i < adapter->num_io_queues; i++) {
+ adapter->tx_ring[i].ring_size = new_tx_size;
+ adapter->rx_ring[i].ring_size = new_rx_size;
+ }
+}
+
+/* This function allows queue allocation to backoff when the system is
+ * low on memory. If there is not enough memory to allocate io queues
+ * the driver will try to allocate smaller queues.
+ *
+ * The backoff algorithm is as follows:
+ * 1. Try to allocate TX and RX and if successful.
+ * 1.1. return success
+ *
+ * 2. Divide by 2 the size of the larger of RX and TX queues (or both if their size is the same).
+ *
+ * 3. If TX or RX is smaller than 256
+ * 3.1. return failure.
+ * 4. else
+ * 4.1. go back to 1.
+ */
+static int create_queues_with_size_backoff(struct ena_adapter *adapter)
+{
+ int rc, cur_rx_ring_size, cur_tx_ring_size;
+ int new_rx_ring_size, new_tx_ring_size;
+
+ /* current queue sizes might be set to smaller than the requested
+ * ones due to past queue allocation failures.
+ */
+ set_io_rings_size(adapter, adapter->requested_tx_ring_size,
+ adapter->requested_rx_ring_size);
+
+ while (1) {
+ if (ena_xdp_present(adapter)) {
+ rc = ena_setup_and_create_all_xdp_queues(adapter);
+
+ if (rc)
+ goto err_setup_tx;
+ }
+ rc = ena_setup_tx_resources_in_range(adapter,
+ 0,
+ adapter->num_io_queues);
+ if (rc)
+ goto err_setup_tx;
+
+ rc = ena_create_io_tx_queues_in_range(adapter,
+ 0,
+ adapter->num_io_queues);
+ if (rc)
+ goto err_create_tx_queues;
+
+ rc = ena_setup_all_rx_resources(adapter);
+ if (rc)
+ goto err_setup_rx;
+
+ rc = ena_create_all_io_rx_queues(adapter);
+ if (rc)
+ goto err_create_rx_queues;
+
+ return 0;
+
+err_create_rx_queues:
+ ena_free_all_io_rx_resources(adapter);
+err_setup_rx:
+ ena_destroy_all_tx_queues(adapter);
+err_create_tx_queues:
+ ena_free_all_io_tx_resources(adapter);
+err_setup_tx:
+ if (rc != -ENOMEM) {
+ netif_err(adapter, ifup, adapter->netdev,
+ "Queue creation failed with error code %d\n",
+ rc);
+ return rc;
+ }
+
+ cur_tx_ring_size = adapter->tx_ring[0].ring_size;
+ cur_rx_ring_size = adapter->rx_ring[0].ring_size;
+
+ netif_err(adapter, ifup, adapter->netdev,
+ "Not enough memory to create queues with sizes TX=%d, RX=%d\n",
+ cur_tx_ring_size, cur_rx_ring_size);
+
+ new_tx_ring_size = cur_tx_ring_size;
+ new_rx_ring_size = cur_rx_ring_size;
+
+ /* Decrease the size of the larger queue, or
+ * decrease both if they are the same size.
+ */
+ if (cur_rx_ring_size <= cur_tx_ring_size)
+ new_tx_ring_size = cur_tx_ring_size / 2;
+ if (cur_rx_ring_size >= cur_tx_ring_size)
+ new_rx_ring_size = cur_rx_ring_size / 2;
+
+ if (new_tx_ring_size < ENA_MIN_RING_SIZE ||
+ new_rx_ring_size < ENA_MIN_RING_SIZE) {
+ netif_err(adapter, ifup, adapter->netdev,
+ "Queue creation failed with the smallest possible queue size of %d for both queues. Not retrying with smaller queues\n",
+ ENA_MIN_RING_SIZE);
+ return rc;
+ }
+
+ netif_err(adapter, ifup, adapter->netdev,
+ "Retrying queue creation with sizes TX=%d, RX=%d\n",
+ new_tx_ring_size,
+ new_rx_ring_size);
+
+ set_io_rings_size(adapter, new_tx_ring_size,
+ new_rx_ring_size);
+ }
+}
+
static int ena_up(struct ena_adapter *adapter)
{
- int rc, i;
+ int io_queue_count, rc, i;
- netdev_dbg(adapter->netdev, "%s\n", __func__);
+ netif_dbg(adapter, ifup, adapter->netdev, "%s\n", __func__);
+ io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues;
ena_setup_io_intr(adapter);
/* napi poll functions should be initialized before running
@@ -1737,31 +2539,15 @@
* interrupt, causing the ISR to fire immediately while the poll
* function wasn't set yet, causing a null dereference
*/
- ena_init_napi(adapter);
+ ena_init_napi_in_range(adapter, 0, io_queue_count);
rc = ena_request_io_irq(adapter);
if (rc)
goto err_req_irq;
- /* allocate transmit descriptors */
- rc = ena_setup_all_tx_resources(adapter);
+ rc = create_queues_with_size_backoff(adapter);
if (rc)
- goto err_setup_tx;
-
- /* allocate receive descriptors */
- rc = ena_setup_all_rx_resources(adapter);
- if (rc)
- goto err_setup_rx;
-
- /* Create TX queues */
- rc = ena_create_all_io_tx_queues(adapter);
- if (rc)
- goto err_create_tx_queues;
-
- /* Create RX queues */
- rc = ena_create_all_io_rx_queues(adapter);
- if (rc)
- goto err_create_rx_queues;
+ goto err_create_queues_with_backoff;
rc = ena_up_complete(adapter);
if (rc)
@@ -1777,36 +2563,35 @@
set_bit(ENA_FLAG_DEV_UP, &adapter->flags);
/* Enable completion queues interrupt */
- for (i = 0; i < adapter->num_queues; i++)
+ for (i = 0; i < adapter->num_io_queues; i++)
ena_unmask_interrupt(&adapter->tx_ring[i],
&adapter->rx_ring[i]);
/* schedule napi in case we had pending packets
* from the last time we disable napi
*/
- for (i = 0; i < adapter->num_queues; i++)
+ for (i = 0; i < io_queue_count; i++)
napi_schedule(&adapter->ena_napi[i].napi);
return rc;
err_up:
- ena_destroy_all_rx_queues(adapter);
-err_create_rx_queues:
ena_destroy_all_tx_queues(adapter);
-err_create_tx_queues:
- ena_free_all_io_rx_resources(adapter);
-err_setup_rx:
ena_free_all_io_tx_resources(adapter);
-err_setup_tx:
+ ena_destroy_all_rx_queues(adapter);
+ ena_free_all_io_rx_resources(adapter);
+err_create_queues_with_backoff:
ena_free_io_irq(adapter);
err_req_irq:
- ena_del_napi(adapter);
+ ena_del_napi_in_range(adapter, 0, io_queue_count);
return rc;
}
static void ena_down(struct ena_adapter *adapter)
{
+ int io_queue_count = adapter->num_io_queues + adapter->xdp_num_queues;
+
netif_info(adapter, ifdown, adapter->netdev, "%s\n", __func__);
clear_bit(ENA_FLAG_DEV_UP, &adapter->flags);
@@ -1819,7 +2604,7 @@
netif_tx_disable(adapter->netdev);
/* After this point the napi handler won't enable the tx queue */
- ena_napi_disable_all(adapter);
+ ena_napi_disable_in_range(adapter, 0, io_queue_count);
/* After destroy the queue there won't be any new interrupts */
@@ -1828,14 +2613,17 @@
rc = ena_com_dev_reset(adapter->ena_dev, adapter->reset_reason);
if (rc)
- dev_err(&adapter->pdev->dev, "Device reset failed\n");
+ netif_err(adapter, ifdown, adapter->netdev,
+ "Device reset failed\n");
+ /* stop submitting admin commands on a device that was reset */
+ ena_com_set_admin_running_state(adapter->ena_dev, false);
}
ena_destroy_all_io_queues(adapter);
ena_disable_io_intr_sync(adapter);
ena_free_io_irq(adapter);
- ena_del_napi(adapter);
+ ena_del_napi_in_range(adapter, 0, io_queue_count);
ena_free_all_tx_bufs(adapter);
ena_free_all_rx_bufs(adapter);
@@ -1860,13 +2648,13 @@
int rc;
/* Notify the stack of the actual queue counts. */
- rc = netif_set_real_num_tx_queues(netdev, adapter->num_queues);
+ rc = netif_set_real_num_tx_queues(netdev, adapter->num_io_queues);
if (rc) {
netif_err(adapter, ifup, netdev, "Can't set num tx queues\n");
return rc;
}
- rc = netif_set_real_num_rx_queues(netdev, adapter->num_queues);
+ rc = netif_set_real_num_rx_queues(netdev, adapter->num_io_queues);
if (rc) {
netif_err(adapter, ifup, netdev, "Can't set num rx queues\n");
return rc;
@@ -1895,6 +2683,9 @@
netif_dbg(adapter, ifdown, netdev, "%s\n", __func__);
+ if (!test_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags))
+ return 0;
+
if (test_bit(ENA_FLAG_DEV_UP, &adapter->flags))
ena_down(adapter);
@@ -1912,7 +2703,63 @@
return 0;
}
-static void ena_tx_csum(struct ena_com_tx_ctx *ena_tx_ctx, struct sk_buff *skb)
+int ena_update_queue_sizes(struct ena_adapter *adapter,
+ u32 new_tx_size,
+ u32 new_rx_size)
+{
+ bool dev_was_up;
+
+ dev_was_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags);
+ ena_close(adapter->netdev);
+ adapter->requested_tx_ring_size = new_tx_size;
+ adapter->requested_rx_ring_size = new_rx_size;
+ ena_init_io_rings(adapter,
+ 0,
+ adapter->xdp_num_queues +
+ adapter->num_io_queues);
+ return dev_was_up ? ena_up(adapter) : 0;
+}
+
+int ena_update_queue_count(struct ena_adapter *adapter, u32 new_channel_count)
+{
+ struct ena_com_dev *ena_dev = adapter->ena_dev;
+ int prev_channel_count;
+ bool dev_was_up;
+
+ dev_was_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags);
+ ena_close(adapter->netdev);
+ prev_channel_count = adapter->num_io_queues;
+ adapter->num_io_queues = new_channel_count;
+ if (ena_xdp_present(adapter) &&
+ ena_xdp_allowed(adapter) == ENA_XDP_ALLOWED) {
+ adapter->xdp_first_ring = new_channel_count;
+ adapter->xdp_num_queues = new_channel_count;
+ if (prev_channel_count > new_channel_count)
+ ena_xdp_exchange_program_rx_in_range(adapter,
+ NULL,
+ new_channel_count,
+ prev_channel_count);
+ else
+ ena_xdp_exchange_program_rx_in_range(adapter,
+ adapter->xdp_bpf_prog,
+ prev_channel_count,
+ new_channel_count);
+ }
+
+ /* We need to destroy the rss table so that the indirection
+ * table will be reinitialized by ena_up()
+ */
+ ena_com_rss_destroy(ena_dev);
+ ena_init_io_rings(adapter,
+ 0,
+ adapter->xdp_num_queues +
+ adapter->num_io_queues);
+ return dev_was_up ? ena_open(adapter->netdev) : 0;
+}
+
+static void ena_tx_csum(struct ena_com_tx_ctx *ena_tx_ctx,
+ struct sk_buff *skb,
+ bool disable_meta_caching)
{
u32 mss = skb_shinfo(skb)->gso_size;
struct ena_com_tx_meta *ena_meta = &ena_tx_ctx->ena_meta;
@@ -1956,7 +2803,9 @@
ena_meta->l3_hdr_len = skb_network_header_len(skb);
ena_meta->l3_hdr_offset = skb_network_offset(skb);
ena_tx_ctx->meta_valid = 1;
-
+ } else if (disable_meta_caching) {
+ memset(ena_meta, 0, sizeof(*ena_meta));
+ ena_tx_ctx->meta_valid = 1;
} else {
ena_tx_ctx->meta_valid = 0;
}
@@ -1991,6 +2840,112 @@
return rc;
}
+static int ena_tx_map_skb(struct ena_ring *tx_ring,
+ struct ena_tx_buffer *tx_info,
+ struct sk_buff *skb,
+ void **push_hdr,
+ u16 *header_len)
+{
+ struct ena_adapter *adapter = tx_ring->adapter;
+ struct ena_com_buf *ena_buf;
+ dma_addr_t dma;
+ u32 skb_head_len, frag_len, last_frag;
+ u16 push_len = 0;
+ u16 delta = 0;
+ int i = 0;
+
+ skb_head_len = skb_headlen(skb);
+ tx_info->skb = skb;
+ ena_buf = tx_info->bufs;
+
+ if (tx_ring->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) {
+ /* When the device is LLQ mode, the driver will copy
+ * the header into the device memory space.
+ * the ena_com layer assume the header is in a linear
+ * memory space.
+ * This assumption might be wrong since part of the header
+ * can be in the fragmented buffers.
+ * Use skb_header_pointer to make sure the header is in a
+ * linear memory space.
+ */
+
+ push_len = min_t(u32, skb->len, tx_ring->tx_max_header_size);
+ *push_hdr = skb_header_pointer(skb, 0, push_len,
+ tx_ring->push_buf_intermediate_buf);
+ *header_len = push_len;
+ if (unlikely(skb->data != *push_hdr)) {
+ u64_stats_update_begin(&tx_ring->syncp);
+ tx_ring->tx_stats.llq_buffer_copy++;
+ u64_stats_update_end(&tx_ring->syncp);
+
+ delta = push_len - skb_head_len;
+ }
+ } else {
+ *push_hdr = NULL;
+ *header_len = min_t(u32, skb_head_len,
+ tx_ring->tx_max_header_size);
+ }
+
+ netif_dbg(adapter, tx_queued, adapter->netdev,
+ "skb: %p header_buf->vaddr: %p push_len: %d\n", skb,
+ *push_hdr, push_len);
+
+ if (skb_head_len > push_len) {
+ dma = dma_map_single(tx_ring->dev, skb->data + push_len,
+ skb_head_len - push_len, DMA_TO_DEVICE);
+ if (unlikely(dma_mapping_error(tx_ring->dev, dma)))
+ goto error_report_dma_error;
+
+ ena_buf->paddr = dma;
+ ena_buf->len = skb_head_len - push_len;
+
+ ena_buf++;
+ tx_info->num_of_bufs++;
+ tx_info->map_linear_data = 1;
+ } else {
+ tx_info->map_linear_data = 0;
+ }
+
+ last_frag = skb_shinfo(skb)->nr_frags;
+
+ for (i = 0; i < last_frag; i++) {
+ const skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
+
+ frag_len = skb_frag_size(frag);
+
+ if (unlikely(delta >= frag_len)) {
+ delta -= frag_len;
+ continue;
+ }
+
+ dma = skb_frag_dma_map(tx_ring->dev, frag, delta,
+ frag_len - delta, DMA_TO_DEVICE);
+ if (unlikely(dma_mapping_error(tx_ring->dev, dma)))
+ goto error_report_dma_error;
+
+ ena_buf->paddr = dma;
+ ena_buf->len = frag_len - delta;
+ ena_buf++;
+ tx_info->num_of_bufs++;
+ delta = 0;
+ }
+
+ return 0;
+
+error_report_dma_error:
+ u64_stats_update_begin(&tx_ring->syncp);
+ tx_ring->tx_stats.dma_mapping_err++;
+ u64_stats_update_end(&tx_ring->syncp);
+ netif_warn(adapter, tx_queued, adapter->netdev, "Failed to map skb\n");
+
+ tx_info->skb = NULL;
+
+ tx_info->num_of_bufs += i;
+ ena_unmap_tx_buff(tx_ring, tx_info);
+
+ return -EINVAL;
+}
+
/* Called with netif_tx_lock. */
static netdev_tx_t ena_start_xmit(struct sk_buff *skb, struct net_device *dev)
{
@@ -1999,16 +2954,9 @@
struct ena_com_tx_ctx ena_tx_ctx;
struct ena_ring *tx_ring;
struct netdev_queue *txq;
- struct ena_com_buf *ena_buf;
void *push_hdr;
- u32 len, last_frag;
- u16 next_to_use;
- u16 req_id;
- u16 push_len;
- u16 header_len;
- dma_addr_t dma;
- int qid, rc, nb_hw_desc;
- int i = -1;
+ u16 next_to_use, req_id, header_len;
+ int qid, rc;
netif_dbg(adapter, tx_queued, dev, "%s skb %p\n", __func__, skb);
/* Determine which tx ring we will be placed on */
@@ -2021,62 +2969,17 @@
goto error_drop_packet;
skb_tx_timestamp(skb);
- len = skb_headlen(skb);
next_to_use = tx_ring->next_to_use;
- req_id = tx_ring->free_tx_ids[next_to_use];
+ req_id = tx_ring->free_ids[next_to_use];
tx_info = &tx_ring->tx_buffer_info[req_id];
tx_info->num_of_bufs = 0;
WARN(tx_info->skb, "SKB isn't NULL req_id %d\n", req_id);
- ena_buf = tx_info->bufs;
- tx_info->skb = skb;
- if (tx_ring->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) {
- /* prepared the push buffer */
- push_len = min_t(u32, len, tx_ring->tx_max_header_size);
- header_len = push_len;
- push_hdr = skb->data;
- } else {
- push_len = 0;
- header_len = min_t(u32, len, tx_ring->tx_max_header_size);
- push_hdr = NULL;
- }
-
- netif_dbg(adapter, tx_queued, dev,
- "skb: %p header_buf->vaddr: %p push_len: %d\n", skb,
- push_hdr, push_len);
-
- if (len > push_len) {
- dma = dma_map_single(tx_ring->dev, skb->data + push_len,
- len - push_len, DMA_TO_DEVICE);
- if (dma_mapping_error(tx_ring->dev, dma))
- goto error_report_dma_error;
-
- ena_buf->paddr = dma;
- ena_buf->len = len - push_len;
-
- ena_buf++;
- tx_info->num_of_bufs++;
- }
-
- last_frag = skb_shinfo(skb)->nr_frags;
-
- for (i = 0; i < last_frag; i++) {
- const skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
-
- len = skb_frag_size(frag);
- dma = skb_frag_dma_map(tx_ring->dev, frag, 0, len,
- DMA_TO_DEVICE);
- if (dma_mapping_error(tx_ring->dev, dma))
- goto error_report_dma_error;
-
- ena_buf->paddr = dma;
- ena_buf->len = len;
- ena_buf++;
- }
-
- tx_info->num_of_bufs += last_frag;
+ rc = ena_tx_map_skb(tx_ring, tx_info, skb, &push_hdr, &header_len);
+ if (unlikely(rc))
+ goto error_drop_packet;
memset(&ena_tx_ctx, 0x0, sizeof(struct ena_com_tx_ctx));
ena_tx_ctx.ena_bufs = tx_info->bufs;
@@ -2086,43 +2989,25 @@
ena_tx_ctx.header_len = header_len;
/* set flags and meta data */
- ena_tx_csum(&ena_tx_ctx, skb);
+ ena_tx_csum(&ena_tx_ctx, skb, tx_ring->disable_meta_caching);
- /* prepare the packet's descriptors to dma engine */
- rc = ena_com_prepare_tx(tx_ring->ena_com_io_sq, &ena_tx_ctx,
- &nb_hw_desc);
-
- if (unlikely(rc)) {
- netif_err(adapter, tx_queued, dev,
- "failed to prepare tx bufs\n");
- u64_stats_update_begin(&tx_ring->syncp);
- tx_ring->tx_stats.queue_stop++;
- tx_ring->tx_stats.prepare_ctx_err++;
- u64_stats_update_end(&tx_ring->syncp);
- netif_tx_stop_queue(txq);
+ rc = ena_xmit_common(dev,
+ tx_ring,
+ tx_info,
+ &ena_tx_ctx,
+ next_to_use,
+ skb->len);
+ if (rc)
goto error_unmap_dma;
- }
netdev_tx_sent_queue(txq, skb->len);
-
- u64_stats_update_begin(&tx_ring->syncp);
- tx_ring->tx_stats.cnt++;
- tx_ring->tx_stats.bytes += skb->len;
- u64_stats_update_end(&tx_ring->syncp);
-
- tx_info->tx_descs = nb_hw_desc;
- tx_info->last_jiffies = jiffies;
- tx_info->print_once = 0;
-
- tx_ring->next_to_use = ENA_TX_RING_IDX_NEXT(next_to_use,
- tx_ring->ring_size);
/* stop the queue when no more space available, the packet can have up
* to sgl_size + 2. one for the meta descriptor and one for header
* (if the header is larger than tx_max_header_size).
*/
- if (unlikely(ena_com_sq_empty_space(tx_ring->ena_com_io_sq) <
- (tx_ring->sgl_size + 2))) {
+ if (unlikely(!ena_com_sq_have_enough_space(tx_ring->ena_com_io_sq,
+ tx_ring->sgl_size + 2))) {
netif_dbg(adapter, tx_queued, dev, "%s stop queue %d\n",
__func__, qid);
@@ -2141,8 +3026,8 @@
*/
smp_mb();
- if (ena_com_sq_empty_space(tx_ring->ena_com_io_sq)
- > ENA_TX_WAKEUP_THRESH) {
+ if (ena_com_sq_have_enough_space(tx_ring->ena_com_io_sq,
+ ENA_TX_WAKEUP_THRESH)) {
netif_tx_wake_queue(txq);
u64_stats_update_begin(&tx_ring->syncp);
tx_ring->tx_stats.queue_wakeup++;
@@ -2150,7 +3035,7 @@
}
}
- if (netif_xmit_stopped(txq) || !skb->xmit_more) {
+ if (netif_xmit_stopped(txq) || !netdev_xmit_more()) {
/* trigger the dma engine. ena_com_write_sq_doorbell()
* has a mb
*/
@@ -2162,42 +3047,17 @@
return NETDEV_TX_OK;
-error_report_dma_error:
- u64_stats_update_begin(&tx_ring->syncp);
- tx_ring->tx_stats.dma_mapping_err++;
- u64_stats_update_end(&tx_ring->syncp);
- netdev_warn(adapter->netdev, "failed to map skb\n");
-
+error_unmap_dma:
+ ena_unmap_tx_buff(tx_ring, tx_info);
tx_info->skb = NULL;
-error_unmap_dma:
- if (i >= 0) {
- /* save value of frag that failed */
- last_frag = i;
-
- /* start back at beginning and unmap skb */
- tx_info->skb = NULL;
- ena_buf = tx_info->bufs;
- dma_unmap_single(tx_ring->dev, dma_unmap_addr(ena_buf, paddr),
- dma_unmap_len(ena_buf, len), DMA_TO_DEVICE);
-
- /* unmap remaining mapped pages */
- for (i = 0; i < last_frag; i++) {
- ena_buf++;
- dma_unmap_page(tx_ring->dev, dma_unmap_addr(ena_buf, paddr),
- dma_unmap_len(ena_buf, len), DMA_TO_DEVICE);
- }
- }
-
error_drop_packet:
-
dev_kfree_skb(skb);
return NETDEV_TX_OK;
}
static u16 ena_select_queue(struct net_device *dev, struct sk_buff *skb,
- struct net_device *sb_dev,
- select_queue_fallback_t fallback)
+ struct net_device *sb_dev)
{
u16 qid;
/* we suspect that this is good for in--kernel network services that
@@ -2207,25 +3067,27 @@
if (skb_rx_queue_recorded(skb))
qid = skb_get_rx_queue(skb);
else
- qid = fallback(dev, skb, NULL);
+ qid = netdev_pick_tx(dev, skb, NULL);
return qid;
}
-static void ena_config_host_info(struct ena_com_dev *ena_dev)
+static void ena_config_host_info(struct ena_com_dev *ena_dev, struct pci_dev *pdev)
{
+ struct device *dev = &pdev->dev;
struct ena_admin_host_info *host_info;
int rc;
/* Allocate only the host info */
rc = ena_com_allocate_host_info(ena_dev);
if (rc) {
- pr_err("Cannot allocate host info\n");
+ dev_err(dev, "Cannot allocate host info\n");
return;
}
host_info = ena_dev->host_attr.host_info;
+ host_info->bdf = (pdev->bus->number << 8) | pdev->devfn;
host_info->os_type = ENA_ADMIN_OS_LINUX;
host_info->kernel_ver = LINUX_VERSION_CODE;
strlcpy(host_info->kernel_ver_str, utsname()->version,
@@ -2234,16 +3096,24 @@
strncpy(host_info->os_dist_str, utsname()->release,
sizeof(host_info->os_dist_str) - 1);
host_info->driver_version =
- (DRV_MODULE_VER_MAJOR) |
- (DRV_MODULE_VER_MINOR << ENA_ADMIN_HOST_INFO_MINOR_SHIFT) |
- (DRV_MODULE_VER_SUBMINOR << ENA_ADMIN_HOST_INFO_SUB_MINOR_SHIFT);
+ (DRV_MODULE_GEN_MAJOR) |
+ (DRV_MODULE_GEN_MINOR << ENA_ADMIN_HOST_INFO_MINOR_SHIFT) |
+ (DRV_MODULE_GEN_SUBMINOR << ENA_ADMIN_HOST_INFO_SUB_MINOR_SHIFT) |
+ ("K"[0] << ENA_ADMIN_HOST_INFO_MODULE_TYPE_SHIFT);
+ host_info->num_cpus = num_online_cpus();
+
+ host_info->driver_supported_features =
+ ENA_ADMIN_HOST_INFO_RX_OFFSET_MASK |
+ ENA_ADMIN_HOST_INFO_INTERRUPT_MODERATION_MASK |
+ ENA_ADMIN_HOST_INFO_RX_BUF_MIRRORING_MASK |
+ ENA_ADMIN_HOST_INFO_RSS_CONFIGURABLE_FUNCTION_KEY_MASK;
rc = ena_com_set_host_attributes(ena_dev);
if (rc) {
if (rc == -EOPNOTSUPP)
- pr_warn("Cannot set host attributes\n");
+ dev_warn(dev, "Cannot set host attributes\n");
else
- pr_err("Cannot set host attributes\n");
+ dev_err(dev, "Cannot set host attributes\n");
goto err;
}
@@ -2271,7 +3141,8 @@
rc = ena_com_allocate_debug_area(adapter->ena_dev, debug_area_size);
if (rc) {
- pr_err("Cannot allocate debug area\n");
+ netif_err(adapter, drv, adapter->netdev,
+ "Cannot allocate debug area\n");
return;
}
@@ -2291,6 +3162,19 @@
ena_com_delete_debug_area(adapter->ena_dev);
}
+int ena_update_hw_stats(struct ena_adapter *adapter)
+{
+ int rc = 0;
+
+ rc = ena_com_get_eni_stats(adapter->ena_dev, &adapter->eni_stats);
+ if (rc) {
+ dev_info_once(&adapter->pdev->dev, "Failed to get ENI stats\n");
+ return rc;
+ }
+
+ return 0;
+}
+
static void ena_get_stats64(struct net_device *netdev,
struct rtnl_link_stats64 *stats)
{
@@ -2298,12 +3182,13 @@
struct ena_ring *rx_ring, *tx_ring;
unsigned int start;
u64 rx_drops;
+ u64 tx_drops;
int i;
if (!test_bit(ENA_FLAG_DEV_UP, &adapter->flags))
return;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues; i++) {
u64 bytes, packets;
tx_ring = &adapter->tx_ring[i];
@@ -2332,9 +3217,11 @@
do {
start = u64_stats_fetch_begin_irq(&adapter->syncp);
rx_drops = adapter->dev_stats.rx_drops;
+ tx_drops = adapter->dev_stats.tx_drops;
} while (u64_stats_fetch_retry_irq(&adapter->syncp, start));
stats->rx_dropped = rx_drops;
+ stats->tx_dropped = tx_drops;
stats->multicast = 0;
stats->collisions = 0;
@@ -2360,6 +3247,7 @@
.ndo_change_mtu = ena_change_mtu,
.ndo_set_mac_address = NULL,
.ndo_validate_addr = eth_validate_addr,
+ .ndo_bpf = ena_xdp,
};
static int ena_device_validate_params(struct ena_adapter *adapter,
@@ -2376,13 +3264,6 @@
return -EINVAL;
}
- if ((get_feat_ctx->max_queues.max_cq_num < adapter->num_queues) ||
- (get_feat_ctx->max_queues.max_sq_num < adapter->num_queues)) {
- netif_err(adapter, drv, netdev,
- "Error, device doesn't support enough queues\n");
- return -EINVAL;
- }
-
if (get_feat_ctx->dev_attr.max_mtu < netdev->mtu) {
netif_err(adapter, drv, netdev,
"Error, device max mtu is smaller than netdev MTU\n");
@@ -2392,10 +3273,71 @@
return 0;
}
+static void set_default_llq_configurations(struct ena_llq_configurations *llq_config)
+{
+ llq_config->llq_header_location = ENA_ADMIN_INLINE_HEADER;
+ llq_config->llq_stride_ctrl = ENA_ADMIN_MULTIPLE_DESCS_PER_ENTRY;
+ llq_config->llq_num_decs_before_header = ENA_ADMIN_LLQ_NUM_DESCS_BEFORE_HEADER_2;
+ llq_config->llq_ring_entry_size = ENA_ADMIN_LIST_ENTRY_SIZE_128B;
+ llq_config->llq_ring_entry_size_value = 128;
+}
+
+static int ena_set_queues_placement_policy(struct pci_dev *pdev,
+ struct ena_com_dev *ena_dev,
+ struct ena_admin_feature_llq_desc *llq,
+ struct ena_llq_configurations *llq_default_configurations)
+{
+ int rc;
+ u32 llq_feature_mask;
+
+ llq_feature_mask = 1 << ENA_ADMIN_LLQ;
+ if (!(ena_dev->supported_features & llq_feature_mask)) {
+ dev_err(&pdev->dev,
+ "LLQ is not supported Fallback to host mode policy.\n");
+ ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
+ return 0;
+ }
+
+ rc = ena_com_config_dev_mode(ena_dev, llq, llq_default_configurations);
+ if (unlikely(rc)) {
+ dev_err(&pdev->dev,
+ "Failed to configure the device mode. Fallback to host mode policy.\n");
+ ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
+ }
+
+ return 0;
+}
+
+static int ena_map_llq_mem_bar(struct pci_dev *pdev, struct ena_com_dev *ena_dev,
+ int bars)
+{
+ bool has_mem_bar = !!(bars & BIT(ENA_MEM_BAR));
+
+ if (!has_mem_bar) {
+ if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) {
+ dev_err(&pdev->dev,
+ "ENA device does not expose LLQ bar. Fallback to host mode policy.\n");
+ ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
+ }
+
+ return 0;
+ }
+
+ ena_dev->mem_bar = devm_ioremap_wc(&pdev->dev,
+ pci_resource_start(pdev, ENA_MEM_BAR),
+ pci_resource_len(pdev, ENA_MEM_BAR));
+
+ if (!ena_dev->mem_bar)
+ return -EFAULT;
+
+ return 0;
+}
+
static int ena_device_init(struct ena_com_dev *ena_dev, struct pci_dev *pdev,
struct ena_com_dev_get_features_ctx *get_feat_ctx,
bool *wd_state)
{
+ struct ena_llq_configurations llq_config;
struct device *dev = &pdev->dev;
bool readless_supported;
u32 aenq_groups;
@@ -2404,7 +3346,7 @@
rc = ena_com_mmio_reg_read_request_init(ena_dev);
if (rc) {
- dev_err(dev, "failed to init mmio read less\n");
+ dev_err(dev, "Failed to init mmio read less\n");
return rc;
}
@@ -2422,7 +3364,7 @@
rc = ena_com_validate_version(ena_dev);
if (rc) {
- dev_err(dev, "device version is too low\n");
+ dev_err(dev, "Device version is too low\n");
goto err_mmio_read_less;
}
@@ -2440,7 +3382,7 @@
}
/* ENA admin level init */
- rc = ena_com_admin_init(ena_dev, &aenq_handlers, true);
+ rc = ena_com_admin_init(ena_dev, &aenq_handlers);
if (rc) {
dev_err(dev,
"Can not initialize ena admin queue with device\n");
@@ -2453,7 +3395,7 @@
*/
ena_com_set_admin_polling_mode(ena_dev, true);
- ena_config_host_info(ena_dev);
+ ena_config_host_info(ena_dev, pdev);
/* Get Device Attributes*/
rc = ena_com_get_dev_attr_feat(ena_dev, get_feat_ctx);
@@ -2479,6 +3421,15 @@
*wd_state = !!(aenq_groups & BIT(ENA_ADMIN_KEEP_ALIVE));
+ set_default_llq_configurations(&llq_config);
+
+ rc = ena_set_queues_placement_policy(pdev, ena_dev, &get_feat_ctx->llq,
+ &llq_config);
+ if (rc) {
+ dev_err(dev, "ENA device init failed\n");
+ goto err_admin_init;
+ }
+
return 0;
err_admin_init:
@@ -2490,14 +3441,13 @@
return rc;
}
-static int ena_enable_msix_and_set_admin_interrupts(struct ena_adapter *adapter,
- int io_vectors)
+static int ena_enable_msix_and_set_admin_interrupts(struct ena_adapter *adapter)
{
struct ena_com_dev *ena_dev = adapter->ena_dev;
struct device *dev = &adapter->pdev->dev;
int rc;
- rc = ena_enable_msix(adapter, io_vectors);
+ rc = ena_enable_msix(adapter);
if (rc) {
dev_err(dev, "Can not reserve msix vectors\n");
return rc;
@@ -2538,17 +3488,14 @@
dev_up = test_bit(ENA_FLAG_DEV_UP, &adapter->flags);
adapter->dev_up_before_reset = dev_up;
-
if (!graceful)
ena_com_set_admin_running_state(ena_dev, false);
if (test_bit(ENA_FLAG_DEV_UP, &adapter->flags))
ena_down(adapter);
- /* Before releasing the ENA resources, a device reset is required.
- * (to prevent the device from accessing them).
- * In case the reset flag is set and the device is up, ena_down()
- * already perform the reset, so it can be skipped.
+ /* Stop the device from sending AENQ events (in case reset flag is set
+ * and device is up, ena_down() already reset the device.
*/
if (!(test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags) && dev_up))
ena_com_dev_reset(adapter->ena_dev, adapter->reset_reason);
@@ -2565,6 +3512,7 @@
ena_com_mmio_reg_read_request_destroy(ena_dev);
+ /* return reset reason to default value */
adapter->reset_reason = ENA_REGS_RESET_NORMAL;
clear_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags);
@@ -2593,8 +3541,7 @@
goto err_device_destroy;
}
- rc = ena_enable_msix_and_set_admin_interrupts(adapter,
- adapter->num_queues);
+ rc = ena_enable_msix_and_set_admin_interrupts(adapter);
if (rc) {
dev_err(&pdev->dev, "Enable MSI-X failed\n");
goto err_device_destroy;
@@ -2615,6 +3562,8 @@
netif_carrier_on(adapter->netdev);
mod_timer(&adapter->timer_service, round_jiffies(jiffies + HZ));
+ adapter->last_keep_alive_jiffies = jiffies;
+
dev_err(&pdev->dev, "Device reset completed successfully\n");
return rc;
@@ -2740,7 +3689,9 @@
struct ena_ring *tx_ring;
struct ena_ring *rx_ring;
int i, budget, rc;
+ int io_queue_count;
+ io_queue_count = adapter->xdp_num_queues + adapter->num_io_queues;
/* Make sure the driver doesn't turn the device in other process */
smp_rmb();
@@ -2755,7 +3706,7 @@
budget = ENA_MONITORED_TX_QUEUES;
- for (i = adapter->last_monitored_tx_qid; i < adapter->num_queues; i++) {
+ for (i = adapter->last_monitored_tx_qid; i < io_queue_count; i++) {
tx_ring = &adapter->tx_ring[i];
rx_ring = &adapter->rx_ring[i];
@@ -2763,7 +3714,8 @@
if (unlikely(rc))
return;
- rc = check_for_rx_interrupt_queue(adapter, rx_ring);
+ rc = !ENA_IS_XDP_INDEX(adapter, i) ?
+ check_for_rx_interrupt_queue(adapter, rx_ring) : 0;
if (unlikely(rc))
return;
@@ -2772,7 +3724,7 @@
break;
}
- adapter->last_monitored_tx_qid = i % adapter->num_queues;
+ adapter->last_monitored_tx_qid = i % io_queue_count;
}
/* trigger napi schedule after 2 consecutive detections */
@@ -2802,11 +3754,10 @@
if (test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags))
return;
- for (i = 0; i < adapter->num_queues; i++) {
+ for (i = 0; i < adapter->num_io_queues; i++) {
rx_ring = &adapter->rx_ring[i];
- refill_required =
- ena_com_sq_empty_space(rx_ring->ena_com_io_sq);
+ refill_required = ena_com_free_q_entries(rx_ring->ena_com_io_sq);
if (unlikely(refill_required == (rx_ring->ring_size - 1))) {
rx_ring->empty_rx_queue++;
@@ -2816,7 +3767,7 @@
u64_stats_update_end(&rx_ring->syncp);
netif_err(adapter, drv, adapter->netdev,
- "trigger refill for ring %d\n", i);
+ "Trigger refill for ring %d\n", i);
napi_schedule(rx_ring->napi);
rx_ring->empty_rx_queue = 0;
@@ -2944,54 +3895,40 @@
mod_timer(&adapter->timer_service, round_jiffies(jiffies + HZ));
}
-static int ena_calc_io_queue_num(struct pci_dev *pdev,
- struct ena_com_dev *ena_dev,
- struct ena_com_dev_get_features_ctx *get_feat_ctx)
+static u32 ena_calc_max_io_queue_num(struct pci_dev *pdev,
+ struct ena_com_dev *ena_dev,
+ struct ena_com_dev_get_features_ctx *get_feat_ctx)
{
- int io_sq_num, io_queue_num;
+ u32 io_tx_sq_num, io_tx_cq_num, io_rx_num, max_num_io_queues;
- /* In case of LLQ use the llq number in the get feature cmd */
- if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) {
- io_sq_num = get_feat_ctx->max_queues.max_llq_num;
+ if (ena_dev->supported_features & BIT(ENA_ADMIN_MAX_QUEUES_EXT)) {
+ struct ena_admin_queue_ext_feature_fields *max_queue_ext =
+ &get_feat_ctx->max_queue_ext.max_queue_ext;
+ io_rx_num = min_t(u32, max_queue_ext->max_rx_sq_num,
+ max_queue_ext->max_rx_cq_num);
- if (io_sq_num == 0) {
- dev_err(&pdev->dev,
- "Trying to use LLQ but llq_num is 0. Fall back into regular queues\n");
-
- ena_dev->tx_mem_queue_type =
- ENA_ADMIN_PLACEMENT_POLICY_HOST;
- io_sq_num = get_feat_ctx->max_queues.max_sq_num;
- }
+ io_tx_sq_num = max_queue_ext->max_tx_sq_num;
+ io_tx_cq_num = max_queue_ext->max_tx_cq_num;
} else {
- io_sq_num = get_feat_ctx->max_queues.max_sq_num;
+ struct ena_admin_queue_feature_desc *max_queues =
+ &get_feat_ctx->max_queues;
+ io_tx_sq_num = max_queues->max_sq_num;
+ io_tx_cq_num = max_queues->max_cq_num;
+ io_rx_num = min_t(u32, io_tx_sq_num, io_tx_cq_num);
}
- io_queue_num = min_t(int, num_online_cpus(), ENA_MAX_NUM_IO_QUEUES);
- io_queue_num = min_t(int, io_queue_num, io_sq_num);
- io_queue_num = min_t(int, io_queue_num,
- get_feat_ctx->max_queues.max_cq_num);
+ /* In case of LLQ use the llq fields for the tx SQ/CQ */
+ if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV)
+ io_tx_sq_num = get_feat_ctx->llq.max_llq_num;
+
+ max_num_io_queues = min_t(u32, num_online_cpus(), ENA_MAX_NUM_IO_QUEUES);
+ max_num_io_queues = min_t(u32, max_num_io_queues, io_rx_num);
+ max_num_io_queues = min_t(u32, max_num_io_queues, io_tx_sq_num);
+ max_num_io_queues = min_t(u32, max_num_io_queues, io_tx_cq_num);
/* 1 IRQ for for mgmnt and 1 IRQs for each IO direction */
- io_queue_num = min_t(int, io_queue_num, pci_msix_vec_count(pdev) - 1);
- if (unlikely(!io_queue_num)) {
- dev_err(&pdev->dev, "The device doesn't have io queues\n");
- return -EFAULT;
- }
+ max_num_io_queues = min_t(u32, max_num_io_queues, pci_msix_vec_count(pdev) - 1);
- return io_queue_num;
-}
-
-static void ena_set_push_mode(struct pci_dev *pdev, struct ena_com_dev *ena_dev,
- struct ena_com_dev_get_features_ctx *get_feat_ctx)
-{
- bool has_mem_bar;
-
- has_mem_bar = pci_select_bars(pdev, IORESOURCE_MEM) & BIT(ENA_MEM_BAR);
-
- /* Enable push mode if device supports LLQ */
- if (has_mem_bar && (get_feat_ctx->max_queues.max_llq_num > 0))
- ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_DEV;
- else
- ena_dev->tx_mem_queue_type = ENA_ADMIN_PLACEMENT_POLICY_HOST;
+ return max_num_io_queues;
}
static void ena_set_dev_offloads(struct ena_com_dev_get_features_ctx *feat,
@@ -3071,7 +4008,7 @@
}
for (i = 0; i < ENA_RX_RSS_TABLE_SIZE; i++) {
- val = ethtool_rxfh_indir_default(i, adapter->num_queues);
+ val = ethtool_rxfh_indir_default(i, adapter->num_io_queues);
rc = ena_com_indirect_table_fill_entry(ena_dev, i,
ENA_IO_RXQ_IDX(val));
if (unlikely(rc && (rc != -EOPNOTSUPP))) {
@@ -3080,7 +4017,7 @@
}
}
- rc = ena_com_fill_hash_function(ena_dev, ENA_ADMIN_CRC32, NULL,
+ rc = ena_com_fill_hash_function(ena_dev, ENA_ADMIN_TOEPLITZ, NULL,
ENA_HASH_KEY_SIZE, 0xFFFFFFFF);
if (unlikely(rc && (rc != -EOPNOTSUPP))) {
dev_err(dev, "Cannot fill hash function\n");
@@ -3109,36 +4046,71 @@
pci_release_selected_regions(pdev, release_bars);
}
-static int ena_calc_queue_size(struct pci_dev *pdev,
- struct ena_com_dev *ena_dev,
- u16 *max_tx_sgl_size,
- u16 *max_rx_sgl_size,
- struct ena_com_dev_get_features_ctx *get_feat_ctx)
+
+static int ena_calc_io_queue_size(struct ena_calc_queue_size_ctx *ctx)
{
- u32 queue_size = ENA_DEFAULT_RING_SIZE;
+ struct ena_admin_feature_llq_desc *llq = &ctx->get_feat_ctx->llq;
+ struct ena_com_dev *ena_dev = ctx->ena_dev;
+ u32 tx_queue_size = ENA_DEFAULT_RING_SIZE;
+ u32 rx_queue_size = ENA_DEFAULT_RING_SIZE;
+ u32 max_tx_queue_size;
+ u32 max_rx_queue_size;
- queue_size = min_t(u32, queue_size,
- get_feat_ctx->max_queues.max_cq_depth);
- queue_size = min_t(u32, queue_size,
- get_feat_ctx->max_queues.max_sq_depth);
+ if (ena_dev->supported_features & BIT(ENA_ADMIN_MAX_QUEUES_EXT)) {
+ struct ena_admin_queue_ext_feature_fields *max_queue_ext =
+ &ctx->get_feat_ctx->max_queue_ext.max_queue_ext;
+ max_rx_queue_size = min_t(u32, max_queue_ext->max_rx_cq_depth,
+ max_queue_ext->max_rx_sq_depth);
+ max_tx_queue_size = max_queue_ext->max_tx_cq_depth;
- if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV)
- queue_size = min_t(u32, queue_size,
- get_feat_ctx->max_queues.max_llq_depth);
+ if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV)
+ max_tx_queue_size = min_t(u32, max_tx_queue_size,
+ llq->max_llq_depth);
+ else
+ max_tx_queue_size = min_t(u32, max_tx_queue_size,
+ max_queue_ext->max_tx_sq_depth);
- queue_size = rounddown_pow_of_two(queue_size);
+ ctx->max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS,
+ max_queue_ext->max_per_packet_tx_descs);
+ ctx->max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS,
+ max_queue_ext->max_per_packet_rx_descs);
+ } else {
+ struct ena_admin_queue_feature_desc *max_queues =
+ &ctx->get_feat_ctx->max_queues;
+ max_rx_queue_size = min_t(u32, max_queues->max_cq_depth,
+ max_queues->max_sq_depth);
+ max_tx_queue_size = max_queues->max_cq_depth;
- if (unlikely(!queue_size)) {
- dev_err(&pdev->dev, "Invalid queue size\n");
- return -EFAULT;
+ if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV)
+ max_tx_queue_size = min_t(u32, max_tx_queue_size,
+ llq->max_llq_depth);
+ else
+ max_tx_queue_size = min_t(u32, max_tx_queue_size,
+ max_queues->max_sq_depth);
+
+ ctx->max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS,
+ max_queues->max_packet_tx_descs);
+ ctx->max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS,
+ max_queues->max_packet_rx_descs);
}
- *max_tx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS,
- get_feat_ctx->max_queues.max_packet_tx_descs);
- *max_rx_sgl_size = min_t(u16, ENA_PKT_MAX_BUFS,
- get_feat_ctx->max_queues.max_packet_rx_descs);
+ max_tx_queue_size = rounddown_pow_of_two(max_tx_queue_size);
+ max_rx_queue_size = rounddown_pow_of_two(max_rx_queue_size);
- return queue_size;
+ tx_queue_size = clamp_val(tx_queue_size, ENA_MIN_RING_SIZE,
+ max_tx_queue_size);
+ rx_queue_size = clamp_val(rx_queue_size, ENA_MIN_RING_SIZE,
+ max_rx_queue_size);
+
+ tx_queue_size = rounddown_pow_of_two(tx_queue_size);
+ rx_queue_size = rounddown_pow_of_two(rx_queue_size);
+
+ ctx->max_tx_queue_size = max_tx_queue_size;
+ ctx->max_rx_queue_size = max_rx_queue_size;
+ ctx->tx_queue_size = tx_queue_size;
+ ctx->rx_queue_size = rx_queue_size;
+
+ return 0;
}
/* ena_probe - Device Initialization Routine
@@ -3153,22 +4125,17 @@
*/
static int ena_probe(struct pci_dev *pdev, const struct pci_device_id *ent)
{
+ struct ena_calc_queue_size_ctx calc_queue_ctx = {};
struct ena_com_dev_get_features_ctx get_feat_ctx;
- static int version_printed;
- struct net_device *netdev;
- struct ena_adapter *adapter;
struct ena_com_dev *ena_dev = NULL;
+ struct ena_adapter *adapter;
+ struct net_device *netdev;
static int adapters_found;
- int io_queue_num, bars, rc;
- int queue_size;
- u16 tx_sgl_size = 0;
- u16 rx_sgl_size = 0;
+ u32 max_num_io_queues;
bool wd_state;
+ int bars, rc;
dev_dbg(&pdev->dev, "%s\n", __func__);
-
- if (version_printed++ == 0)
- dev_info(&pdev->dev, "%s", version);
rc = pci_enable_device_mem(pdev);
if (rc) {
@@ -3202,50 +4169,48 @@
pci_resource_start(pdev, ENA_REG_BAR),
pci_resource_len(pdev, ENA_REG_BAR));
if (!ena_dev->reg_bar) {
- dev_err(&pdev->dev, "failed to remap regs bar\n");
+ dev_err(&pdev->dev, "Failed to remap regs bar\n");
rc = -EFAULT;
goto err_free_region;
}
+
+ ena_dev->ena_min_poll_delay_us = ENA_ADMIN_POLL_DELAY_US;
ena_dev->dmadev = &pdev->dev;
rc = ena_device_init(ena_dev, pdev, &get_feat_ctx, &wd_state);
if (rc) {
- dev_err(&pdev->dev, "ena device init failed\n");
+ dev_err(&pdev->dev, "ENA device init failed\n");
if (rc == -ETIME)
rc = -EPROBE_DEFER;
goto err_free_region;
}
- ena_set_push_mode(pdev, ena_dev, &get_feat_ctx);
-
- if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV) {
- ena_dev->mem_bar = devm_ioremap_wc(&pdev->dev,
- pci_resource_start(pdev, ENA_MEM_BAR),
- pci_resource_len(pdev, ENA_MEM_BAR));
- if (!ena_dev->mem_bar) {
- rc = -EFAULT;
- goto err_device_destroy;
- }
+ rc = ena_map_llq_mem_bar(pdev, ena_dev, bars);
+ if (rc) {
+ dev_err(&pdev->dev, "ENA llq bar mapping failed\n");
+ goto err_free_ena_dev;
}
- /* initial Tx interrupt delay, Assumes 1 usec granularity.
- * Updated during device initialization with the real granularity
- */
+ calc_queue_ctx.ena_dev = ena_dev;
+ calc_queue_ctx.get_feat_ctx = &get_feat_ctx;
+ calc_queue_ctx.pdev = pdev;
+
+ /* Initial TX and RX interrupt delay. Assumes 1 usec granularity.
+ * Updated during device initialization with the real granularity
+ */
ena_dev->intr_moder_tx_interval = ENA_INTR_INITIAL_TX_INTERVAL_USECS;
- io_queue_num = ena_calc_io_queue_num(pdev, ena_dev, &get_feat_ctx);
- queue_size = ena_calc_queue_size(pdev, ena_dev, &tx_sgl_size,
- &rx_sgl_size, &get_feat_ctx);
- if ((queue_size <= 0) || (io_queue_num <= 0)) {
+ ena_dev->intr_moder_rx_interval = ENA_INTR_INITIAL_RX_INTERVAL_USECS;
+ ena_dev->intr_delay_resolution = ENA_DEFAULT_INTR_DELAY_RESOLUTION;
+ max_num_io_queues = ena_calc_max_io_queue_num(pdev, ena_dev, &get_feat_ctx);
+ rc = ena_calc_io_queue_size(&calc_queue_ctx);
+ if (rc || !max_num_io_queues) {
rc = -EFAULT;
goto err_device_destroy;
}
- dev_info(&pdev->dev, "creating %d io queues. queue size: %d\n",
- io_queue_num, queue_size);
-
/* dev zeroed in init_etherdev */
- netdev = alloc_etherdev_mq(sizeof(struct ena_adapter), io_queue_num);
+ netdev = alloc_etherdev_mq(sizeof(struct ena_adapter), max_num_io_queues);
if (!netdev) {
dev_err(&pdev->dev, "alloc_etherdev_mq failed\n");
rc = -ENOMEM;
@@ -3266,16 +4231,26 @@
adapter->msg_enable = netif_msg_init(debug, DEFAULT_MSG_ENABLE);
adapter->reset_reason = ENA_REGS_RESET_NORMAL;
- adapter->tx_ring_size = queue_size;
- adapter->rx_ring_size = queue_size;
+ adapter->requested_tx_ring_size = calc_queue_ctx.tx_queue_size;
+ adapter->requested_rx_ring_size = calc_queue_ctx.rx_queue_size;
+ adapter->max_tx_ring_size = calc_queue_ctx.max_tx_queue_size;
+ adapter->max_rx_ring_size = calc_queue_ctx.max_rx_queue_size;
+ adapter->max_tx_sgl_size = calc_queue_ctx.max_tx_sgl_size;
+ adapter->max_rx_sgl_size = calc_queue_ctx.max_rx_sgl_size;
- adapter->max_tx_sgl_size = tx_sgl_size;
- adapter->max_rx_sgl_size = rx_sgl_size;
-
- adapter->num_queues = io_queue_num;
+ adapter->num_io_queues = max_num_io_queues;
+ adapter->max_num_io_queues = max_num_io_queues;
adapter->last_monitored_tx_qid = 0;
+ adapter->xdp_first_ring = 0;
+ adapter->xdp_num_queues = 0;
+
adapter->rx_copybreak = ENA_DEFAULT_RX_COPYBREAK;
+ if (ena_dev->tx_mem_queue_type == ENA_ADMIN_PLACEMENT_POLICY_DEV)
+ adapter->disable_meta_caching =
+ !!(get_feat_ctx.llq.accel_mode.u.get.supported_flags &
+ BIT(ENA_ADMIN_DISABLE_META_CACHING));
+
adapter->wd_state = wd_state;
snprintf(adapter->name, ENA_NAME_MAX_LEN, "ena_%d", adapters_found);
@@ -3286,7 +4261,10 @@
"Failed to query interrupt moderation feature\n");
goto err_netdev_destroy;
}
- ena_init_io_rings(adapter);
+ ena_init_io_rings(adapter,
+ 0,
+ adapter->xdp_num_queues +
+ adapter->num_io_queues);
netdev->netdev_ops = &ena_netdev_ops;
netdev->watchdog_timeo = TX_TIMEOUT;
@@ -3296,7 +4274,7 @@
u64_stats_init(&adapter->syncp);
- rc = ena_enable_msix_and_set_admin_interrupts(adapter, io_queue_num);
+ rc = ena_enable_msix_and_set_admin_interrupts(adapter);
if (rc) {
dev_err(&pdev->dev,
"Failed to enable and set the admin interrupts\n");
@@ -3309,6 +4287,11 @@
}
ena_config_debug_area(adapter);
+
+ if (!ena_update_hw_stats(adapter))
+ adapter->eni_stats_supported = true;
+ else
+ adapter->eni_stats_supported = false;
memcpy(adapter->netdev->perm_addr, adapter->mac_addr, netdev->addr_len);
@@ -3332,9 +4315,10 @@
timer_setup(&adapter->timer_service, ena_timer_service, 0);
mod_timer(&adapter->timer_service, round_jiffies(jiffies + HZ));
- dev_info(&pdev->dev, "%s found at mem %lx, mac addr %pM Queues %d\n",
+ dev_info(&pdev->dev,
+ "%s found at mem %lx, mac addr %pM\n",
DEVICE_NAME, (long)pci_resource_start(pdev, 0),
- netdev->dev_addr, io_queue_num);
+ netdev->dev_addr);
set_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags);
@@ -3347,10 +4331,11 @@
ena_com_rss_destroy(ena_dev);
err_free_msix:
ena_com_dev_reset(ena_dev, ENA_REGS_RESET_INIT_ERR);
+ /* stop submitting admin commands on a device that was reset */
+ ena_com_set_admin_running_state(ena_dev, false);
ena_free_mgmnt_irq(adapter);
ena_disable_msix(adapter);
err_worker_destroy:
- ena_com_destroy_interrupt_moderation(ena_dev);
del_timer(&adapter->timer_service);
err_netdev_destroy:
free_netdev(netdev);
@@ -3368,13 +4353,15 @@
/*****************************************************************************/
-/* ena_remove - Device Removal Routine
+/* __ena_shutoff - Helper used in both PCI remove/shutdown routines
* @pdev: PCI device information struct
+ * @shutdown: Is it a shutdown operation? If false, means it is a removal
*
- * ena_remove is called by the PCI subsystem to alert the driver
- * that it should release a PCI device.
+ * __ena_shutoff is a helper routine that does the real work on shutdown and
+ * removal paths; the difference between those paths is with regards to whether
+ * dettach or unregister the netdevice.
*/
-static void ena_remove(struct pci_dev *pdev)
+static void __ena_shutoff(struct pci_dev *pdev, bool shutdown)
{
struct ena_adapter *adapter = pci_get_drvdata(pdev);
struct ena_com_dev *ena_dev;
@@ -3396,19 +4383,18 @@
del_timer_sync(&adapter->timer_service);
cancel_work_sync(&adapter->reset_task);
- unregister_netdev(netdev);
-
- /* If the device is running then we want to make sure the device will be
- * reset to make sure no more events will be issued by the device.
- */
- if (test_bit(ENA_FLAG_DEVICE_RUNNING, &adapter->flags))
- set_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags);
-
- rtnl_lock();
+ rtnl_lock(); /* lock released inside the below if-else block */
+ adapter->reset_reason = ENA_REGS_RESET_SHUTDOWN;
ena_destroy_device(adapter, true);
- rtnl_unlock();
-
- free_netdev(netdev);
+ if (shutdown) {
+ netif_device_detach(netdev);
+ dev_close(netdev);
+ rtnl_unlock();
+ } else {
+ rtnl_unlock();
+ unregister_netdev(netdev);
+ free_netdev(netdev);
+ }
ena_com_rss_destroy(ena_dev);
@@ -3420,18 +4406,39 @@
pci_disable_device(pdev);
- ena_com_destroy_interrupt_moderation(ena_dev);
-
vfree(ena_dev);
}
-#ifdef CONFIG_PM
-/* ena_suspend - PM suspend callback
+/* ena_remove - Device Removal Routine
* @pdev: PCI device information struct
- * @state:power state
+ *
+ * ena_remove is called by the PCI subsystem to alert the driver
+ * that it should release a PCI device.
*/
-static int ena_suspend(struct pci_dev *pdev, pm_message_t state)
+
+static void ena_remove(struct pci_dev *pdev)
{
+ __ena_shutoff(pdev, false);
+}
+
+/* ena_shutdown - Device Shutdown Routine
+ * @pdev: PCI device information struct
+ *
+ * ena_shutdown is called by the PCI subsystem to alert the driver that
+ * a shutdown/reboot (or kexec) is happening and device must be disabled.
+ */
+
+static void ena_shutdown(struct pci_dev *pdev)
+{
+ __ena_shutoff(pdev, true);
+}
+
+/* ena_suspend - PM suspend callback
+ * @dev_d: Device information struct
+ */
+static int __maybe_unused ena_suspend(struct device *dev_d)
+{
+ struct pci_dev *pdev = to_pci_dev(dev_d);
struct ena_adapter *adapter = pci_get_drvdata(pdev);
u64_stats_update_begin(&adapter->syncp);
@@ -3441,7 +4448,7 @@
rtnl_lock();
if (unlikely(test_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags))) {
dev_err(&pdev->dev,
- "ignoring device reset request as the device is being suspended\n");
+ "Ignoring device reset request as the device is being suspended\n");
clear_bit(ENA_FLAG_TRIGGER_RESET, &adapter->flags);
}
ena_destroy_device(adapter, true);
@@ -3450,12 +4457,11 @@
}
/* ena_resume - PM resume callback
- * @pdev: PCI device information struct
- *
+ * @dev_d: Device information struct
*/
-static int ena_resume(struct pci_dev *pdev)
+static int __maybe_unused ena_resume(struct device *dev_d)
{
- struct ena_adapter *adapter = pci_get_drvdata(pdev);
+ struct ena_adapter *adapter = dev_get_drvdata(dev_d);
int rc;
u64_stats_update_begin(&adapter->syncp);
@@ -3467,23 +4473,22 @@
rtnl_unlock();
return rc;
}
-#endif
+
+static SIMPLE_DEV_PM_OPS(ena_pm_ops, ena_suspend, ena_resume);
static struct pci_driver ena_pci_driver = {
.name = DRV_MODULE_NAME,
.id_table = ena_pci_tbl,
.probe = ena_probe,
.remove = ena_remove,
-#ifdef CONFIG_PM
- .suspend = ena_suspend,
- .resume = ena_resume,
-#endif
+ .shutdown = ena_shutdown,
+ .driver.pm = &ena_pm_ops,
.sriov_configure = pci_sriov_configure_simple,
};
static int __init ena_init(void)
{
- pr_info("%s", version);
+ int ret;
ena_wq = create_singlethread_workqueue(DRV_MODULE_NAME);
if (!ena_wq) {
@@ -3491,7 +4496,11 @@
return -ENOMEM;
}
- return pci_register_driver(&ena_pci_driver);
+ ret = pci_register_driver(&ena_pci_driver);
+ if (ret)
+ destroy_workqueue(ena_wq);
+
+ return ret;
}
static void __exit ena_cleanup(void)
@@ -3520,7 +4529,7 @@
ENA_ADMIN_AENQ_LINK_CHANGE_DESC_LINK_STATUS_MASK;
if (status) {
- netdev_dbg(adapter->netdev, "%s\n", __func__);
+ netif_dbg(adapter, ifup, adapter->netdev, "%s\n", __func__);
set_bit(ENA_FLAG_LINK_UP, &adapter->flags);
if (!test_bit(ENA_FLAG_ONGOING_RESET, &adapter->flags))
netif_carrier_on(adapter->netdev);
@@ -3536,17 +4545,20 @@
struct ena_adapter *adapter = (struct ena_adapter *)adapter_data;
struct ena_admin_aenq_keep_alive_desc *desc;
u64 rx_drops;
+ u64 tx_drops;
desc = (struct ena_admin_aenq_keep_alive_desc *)aenq_e;
adapter->last_keep_alive_jiffies = jiffies;
rx_drops = ((u64)desc->rx_drops_high << 32) | desc->rx_drops_low;
+ tx_drops = ((u64)desc->tx_drops_high << 32) | desc->tx_drops_low;
u64_stats_update_begin(&adapter->syncp);
/* These stats are accumulated by the device, so the counters indicate
* all drops since last reset.
*/
adapter->dev_stats.rx_drops = rx_drops;
+ adapter->dev_stats.tx_drops = tx_drops;
u64_stats_update_end(&adapter->syncp);
}
@@ -3561,7 +4573,7 @@
aenq_e->aenq_common_desc.group,
ENA_ADMIN_NOTIFICATION);
- switch (aenq_e->aenq_common_desc.syndrom) {
+ switch (aenq_e->aenq_common_desc.syndrome) {
case ENA_ADMIN_UPDATE_HINTS:
hints = (struct ena_admin_ena_hw_hints *)
(&aenq_e->inline_data_w4);
@@ -3570,7 +4582,7 @@
default:
netif_err(adapter, drv, adapter->netdev,
"Invalid aenq notification link state %d\n",
- aenq_e->aenq_common_desc.syndrom);
+ aenq_e->aenq_common_desc.syndrome);
}
}
--
Gitblit v1.6.2