2008-10-23 05:47:49 +07:00
|
|
|
/*
|
|
|
|
* Copyright (c) 2007 Mellanox Technologies. All rights reserved.
|
|
|
|
*
|
|
|
|
* 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
|
|
|
|
* OpenIB.org 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.
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef _MLX4_EN_H_
|
|
|
|
#define _MLX4_EN_H_
|
|
|
|
|
2011-07-20 11:54:22 +07:00
|
|
|
#include <linux/bitops.h>
|
2008-10-23 05:47:49 +07:00
|
|
|
#include <linux/compiler.h>
|
|
|
|
#include <linux/list.h>
|
|
|
|
#include <linux/mutex.h>
|
|
|
|
#include <linux/netdevice.h>
|
2011-07-20 11:54:22 +07:00
|
|
|
#include <linux/if_vlan.h>
|
2013-04-23 13:06:49 +07:00
|
|
|
#include <linux/net_tstamp.h>
|
2012-04-05 04:33:26 +07:00
|
|
|
#ifdef CONFIG_MLX4_EN_DCB
|
|
|
|
#include <linux/dcbnl.h>
|
|
|
|
#endif
|
2012-07-19 05:33:52 +07:00
|
|
|
#include <linux/cpu_rmap.h>
|
2014-01-01 00:39:39 +07:00
|
|
|
#include <linux/ptp_clock_kernel.h>
|
2018-01-03 17:25:39 +07:00
|
|
|
#include <net/xdp.h>
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
#include <linux/mlx4/device.h>
|
|
|
|
#include <linux/mlx4/qp.h>
|
|
|
|
#include <linux/mlx4/cq.h>
|
|
|
|
#include <linux/mlx4/srq.h>
|
|
|
|
#include <linux/mlx4/doorbell.h>
|
2010-08-24 10:46:18 +07:00
|
|
|
#include <linux/mlx4/cmd.h>
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
#include "en_port.h"
|
2015-03-30 21:45:21 +07:00
|
|
|
#include "mlx4_stats.h"
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
#define DRV_NAME "mlx4_en"
|
2017-06-07 20:26:14 +07:00
|
|
|
#define DRV_VERSION "4.0-0"
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
#define MLX4_EN_MSG_LEVEL (NETIF_MSG_LINK | NETIF_MSG_IFDOWN)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Device constants
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
|
#define MLX4_EN_PAGE_SHIFT 12
|
|
|
|
#define MLX4_EN_PAGE_SIZE (1 << MLX4_EN_PAGE_SHIFT)
|
2012-12-02 10:49:23 +07:00
|
|
|
#define DEF_RX_RINGS 16
|
|
|
|
#define MAX_RX_RINGS 128
|
2011-03-23 05:37:52 +07:00
|
|
|
#define MIN_RX_RINGS 4
|
2017-06-15 18:35:39 +07:00
|
|
|
#define LOG_TXBB_SIZE 6
|
|
|
|
#define TXBB_SIZE BIT(LOG_TXBB_SIZE)
|
2008-10-23 05:47:49 +07:00
|
|
|
#define HEADROOM (2048 / TXBB_SIZE + 1)
|
|
|
|
#define STAMP_STRIDE 64
|
|
|
|
#define STAMP_DWORDS (STAMP_STRIDE / 4)
|
|
|
|
#define STAMP_SHIFT 31
|
|
|
|
#define STAMP_VAL 0x7fffffff
|
|
|
|
#define STATS_DELAY (HZ / 4)
|
2013-04-23 13:06:51 +07:00
|
|
|
#define SERVICE_TASK_DELAY (HZ / 4)
|
2012-07-05 11:03:49 +07:00
|
|
|
#define MAX_NUM_OF_FS_RULES 256
|
2008-10-23 05:47:49 +07:00
|
|
|
|
2012-07-19 05:33:52 +07:00
|
|
|
#define MLX4_EN_FILTER_HASH_SHIFT 4
|
|
|
|
#define MLX4_EN_FILTER_EXPIRY_QUOTA 60
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
/* Typical TSO descriptor with 16 gather entries is 352 bytes... */
|
|
|
|
#define MAX_DESC_SIZE 512
|
|
|
|
#define MAX_DESC_TXBBS (MAX_DESC_SIZE / TXBB_SIZE)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* OS related constants and tunables
|
|
|
|
*/
|
|
|
|
|
2014-07-22 19:44:10 +07:00
|
|
|
#define MLX4_EN_PRIV_FLAGS_BLUEFLAME 1
|
2015-07-27 18:46:34 +07:00
|
|
|
#define MLX4_EN_PRIV_FLAGS_PHV 2
|
2014-07-22 19:44:10 +07:00
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
#define MLX4_EN_WATCHDOG_TIMEOUT (15 * HZ)
|
|
|
|
|
mlx4: allocate just enough pages instead of always 4 pages
The driver uses a 2-order allocation, which is too much on architectures
like ppc64, which has a 64KiB page. This particular allocation is used
for large packet fragments that may have a size of 512, 1024, 4096 or
fill the whole allocation. So, a minimum size of 16384 is good enough
and will be the same size that is used in architectures of 4KiB sized
pages.
This will avoid allocation failures that we see when the system is under
stress, but still has plenty of memory, like the one below.
This will also allow us to set the interface MTU to higher values like
9000, which was not possible on ppc64 without this patch.
Node 1 DMA: 737*64kB 37*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB 0*8192kB 0*16384kB = 51904kB
83137 total pagecache pages
0 pages in swap cache
Swap cache stats: add 0, delete 0, find 0/0
Free swap = 10420096kB
Total swap = 10420096kB
107776 pages RAM
1184 pages reserved
147343 pages shared
28152 pages non-shared
netstat: page allocation failure. order:2, mode:0x4020
Call Trace:
[c0000001a4fa3770] [c000000000012f04] .show_stack+0x74/0x1c0 (unreliable)
[c0000001a4fa3820] [c00000000016af38] .__alloc_pages_nodemask+0x618/0x930
[c0000001a4fa39a0] [c0000000001a71a0] .alloc_pages_current+0xb0/0x170
[c0000001a4fa3a40] [d00000000dcc3e00] .mlx4_en_alloc_frag+0x200/0x240 [mlx4_en]
[c0000001a4fa3b10] [d00000000dcc3f8c] .mlx4_en_complete_rx_desc+0x14c/0x250 [mlx4_en]
[c0000001a4fa3be0] [d00000000dcc4eec] .mlx4_en_process_rx_cq+0x62c/0x850 [mlx4_en]
[c0000001a4fa3d20] [d00000000dcc5150] .mlx4_en_poll_rx_cq+0x40/0x90 [mlx4_en]
[c0000001a4fa3dc0] [c0000000004e2bb8] .net_rx_action+0x178/0x450
[c0000001a4fa3eb0] [c00000000009c9b8] .__do_softirq+0x118/0x290
[c0000001a4fa3f90] [c000000000031df8] .call_do_softirq+0x14/0x24
[c000000184c3b520] [c00000000000e700] .do_softirq+0xf0/0x110
[c000000184c3b5c0] [c00000000009c6d4] .irq_exit+0xb4/0xc0
[c000000184c3b640] [c00000000000e964] .do_IRQ+0x144/0x230
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@linux.vnet.ibm.com>
Signed-off-by: Kleber Sacilotto de Souza <klebers@linux.vnet.ibm.com>
Tested-by: Kleber Sacilotto de Souza <klebers@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-04-04 16:40:40 +07:00
|
|
|
/* Use the maximum between 16384 and a single page */
|
|
|
|
#define MLX4_EN_ALLOC_SIZE PAGE_ALIGN(16384)
|
2013-06-23 22:17:56 +07:00
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
#define MLX4_EN_MAX_RX_FRAGS 4
|
|
|
|
|
2009-01-09 01:57:37 +07:00
|
|
|
/* Maximum ring sizes */
|
|
|
|
#define MLX4_EN_MAX_TX_SIZE 8192
|
|
|
|
#define MLX4_EN_MAX_RX_SIZE 8192
|
|
|
|
|
2012-07-16 14:01:53 +07:00
|
|
|
/* Minimum ring size for our page-allocation scheme to work */
|
2008-10-23 05:47:49 +07:00
|
|
|
#define MLX4_EN_MIN_RX_SIZE (MLX4_EN_ALLOC_SIZE / SMP_CACHE_BYTES)
|
|
|
|
#define MLX4_EN_MIN_TX_SIZE (4096 / TXBB_SIZE)
|
|
|
|
|
2009-06-02 06:24:07 +07:00
|
|
|
#define MLX4_EN_SMALL_PKT_SIZE 64
|
2014-07-22 19:44:12 +07:00
|
|
|
#define MLX4_EN_MIN_TX_RING_P_UP 1
|
2012-05-17 07:58:10 +07:00
|
|
|
#define MLX4_EN_MAX_TX_RING_P_UP 32
|
2017-06-29 18:07:57 +07:00
|
|
|
#define MLX4_EN_NUM_UP_LOW 1
|
2017-06-29 18:07:56 +07:00
|
|
|
#define MLX4_EN_NUM_UP_HIGH 8
|
2008-10-23 05:47:49 +07:00
|
|
|
#define MLX4_EN_DEF_RX_RING_SIZE 1024
|
2017-06-15 18:35:38 +07:00
|
|
|
#define MLX4_EN_DEF_TX_RING_SIZE MLX4_EN_DEF_RX_RING_SIZE
|
2012-12-02 10:49:23 +07:00
|
|
|
#define MAX_TX_RINGS (MLX4_EN_MAX_TX_RING_P_UP * \
|
2017-06-29 18:07:56 +07:00
|
|
|
MLX4_EN_NUM_UP_HIGH)
|
2008-10-23 05:47:49 +07:00
|
|
|
|
2014-07-08 15:28:12 +07:00
|
|
|
#define MLX4_EN_DEFAULT_TX_WORK 256
|
|
|
|
|
2009-06-02 06:23:13 +07:00
|
|
|
/* Target number of packets to coalesce with interrupt moderation */
|
|
|
|
#define MLX4_EN_RX_COAL_TARGET 44
|
2008-10-23 05:47:49 +07:00
|
|
|
#define MLX4_EN_RX_COAL_TIME 0x10
|
|
|
|
|
2012-04-23 09:18:39 +07:00
|
|
|
#define MLX4_EN_TX_COAL_PKTS 16
|
2012-11-05 23:20:42 +07:00
|
|
|
#define MLX4_EN_TX_COAL_TIME 0x10
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
#define MLX4_EN_RX_RATE_LOW 400000
|
|
|
|
#define MLX4_EN_RX_COAL_TIME_LOW 0
|
|
|
|
#define MLX4_EN_RX_RATE_HIGH 450000
|
|
|
|
#define MLX4_EN_RX_COAL_TIME_HIGH 128
|
|
|
|
#define MLX4_EN_RX_SIZE_THRESH 1024
|
|
|
|
#define MLX4_EN_RX_RATE_THRESH (1000000 / MLX4_EN_RX_COAL_TIME_HIGH)
|
|
|
|
#define MLX4_EN_SAMPLE_INTERVAL 0
|
2011-03-23 05:37:36 +07:00
|
|
|
#define MLX4_EN_AVG_PKT_SMALL 256
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
#define MLX4_EN_AUTO_CONF 0xffff
|
|
|
|
|
|
|
|
#define MLX4_EN_DEF_RX_PAUSE 1
|
|
|
|
#define MLX4_EN_DEF_TX_PAUSE 1
|
|
|
|
|
tree-wide: fix assorted typos all over the place
That is "success", "unknown", "through", "performance", "[re|un]mapping"
, "access", "default", "reasonable", "[con]currently", "temperature"
, "channel", "[un]used", "application", "example","hierarchy", "therefore"
, "[over|under]flow", "contiguous", "threshold", "enough" and others.
Signed-off-by: André Goddard Rosa <andre.goddard@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
2009-11-14 22:09:05 +07:00
|
|
|
/* Interval between successive polls in the Tx routine when polling is used
|
2008-10-23 05:47:49 +07:00
|
|
|
instead of interrupts (in per-core Tx rings) - should be power of 2 */
|
|
|
|
#define MLX4_EN_TX_POLL_MODER 16
|
|
|
|
#define MLX4_EN_TX_POLL_TIMEOUT (HZ / 4)
|
|
|
|
|
|
|
|
#define SMALL_PACKET_SIZE (256 - NET_IP_ALIGN)
|
|
|
|
#define HEADER_COPY_SIZE (128 - NET_IP_ALIGN)
|
2010-08-24 10:46:18 +07:00
|
|
|
#define MLX4_LOOPBACK_TEST_PAYLOAD (HEADER_COPY_SIZE - ETH_HLEN)
|
2017-12-13 23:12:09 +07:00
|
|
|
#define PREAMBLE_LEN 8
|
|
|
|
#define MLX4_SELFTEST_LB_MIN_MTU (MLX4_LOOPBACK_TEST_PAYLOAD + NET_IP_ALIGN + \
|
|
|
|
ETH_HLEN + PREAMBLE_LEN)
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
#define MLX4_EN_MIN_MTU 46
|
2016-07-20 02:16:50 +07:00
|
|
|
/* VLAN_HLEN is added twice,to support skb vlan tagged with multiple
|
|
|
|
* headers. (For example: ETH_P_8021Q and ETH_P_8021AD).
|
|
|
|
*/
|
|
|
|
#define MLX4_EN_EFF_MTU(mtu) ((mtu) + ETH_HLEN + (2 * VLAN_HLEN))
|
2008-10-23 05:47:49 +07:00
|
|
|
#define ETH_BCAST 0xffffffffffffULL
|
|
|
|
|
2010-08-24 10:46:18 +07:00
|
|
|
#define MLX4_EN_LOOPBACK_RETRIES 5
|
|
|
|
#define MLX4_EN_LOOPBACK_TIMEOUT 100
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
#ifdef MLX4_EN_PERF_STAT
|
|
|
|
/* Number of samples to 'average' */
|
|
|
|
#define AVG_SIZE 128
|
|
|
|
#define AVG_FACTOR 1024
|
|
|
|
|
|
|
|
#define INC_PERF_COUNTER(cnt) (++(cnt))
|
|
|
|
#define ADD_PERF_COUNTER(cnt, add) ((cnt) += (add))
|
|
|
|
#define AVG_PERF_COUNTER(cnt, sample) \
|
|
|
|
((cnt) = ((cnt) * (AVG_SIZE - 1) + (sample) * AVG_FACTOR) / AVG_SIZE)
|
|
|
|
#define GET_PERF_COUNTER(cnt) (cnt)
|
|
|
|
#define GET_AVG_PERF_COUNTER(cnt) ((cnt) / AVG_FACTOR)
|
|
|
|
|
|
|
|
#else
|
|
|
|
|
|
|
|
#define INC_PERF_COUNTER(cnt) do {} while (0)
|
|
|
|
#define ADD_PERF_COUNTER(cnt, add) do {} while (0)
|
|
|
|
#define AVG_PERF_COUNTER(cnt, sample) do {} while (0)
|
|
|
|
#define GET_PERF_COUNTER(cnt) (0)
|
|
|
|
#define GET_AVG_PERF_COUNTER(cnt) (0)
|
|
|
|
#endif /* MLX4_EN_PERF_STAT */
|
|
|
|
|
2014-03-02 15:24:58 +07:00
|
|
|
/* Constants for TX flow */
|
|
|
|
enum {
|
|
|
|
MAX_INLINE = 104, /* 128 - 16 - 4 - 4 */
|
|
|
|
MAX_BF = 256,
|
|
|
|
MIN_PKT_LEN = 17,
|
|
|
|
};
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
/*
|
|
|
|
* Configurables
|
|
|
|
*/
|
|
|
|
|
|
|
|
enum cq_type {
|
2016-11-02 22:12:24 +07:00
|
|
|
/* keep tx types first */
|
2016-11-02 22:12:23 +07:00
|
|
|
TX,
|
|
|
|
TX_XDP,
|
2016-11-02 22:12:24 +07:00
|
|
|
#define MLX4_EN_NUM_TX_TYPES (TX_XDP + 1)
|
2016-11-02 22:12:23 +07:00
|
|
|
RX,
|
2008-10-23 05:47:49 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Useful macros
|
|
|
|
*/
|
|
|
|
#define ROUNDUP_LOG2(x) ilog2(roundup_pow_of_two(x))
|
|
|
|
#define XNOR(x, y) (!(x) == !(y))
|
|
|
|
|
|
|
|
|
|
|
|
struct mlx4_en_tx_info {
|
2016-07-20 02:16:55 +07:00
|
|
|
union {
|
|
|
|
struct sk_buff *skb;
|
|
|
|
struct page *page;
|
|
|
|
};
|
2014-10-05 16:35:13 +07:00
|
|
|
dma_addr_t map0_dma;
|
|
|
|
u32 map0_byte_count;
|
2014-10-05 16:35:10 +07:00
|
|
|
u32 nr_txbb;
|
|
|
|
u32 nr_bytes;
|
|
|
|
u8 linear;
|
|
|
|
u8 data_offset;
|
|
|
|
u8 inl;
|
|
|
|
u8 ts_requested;
|
2014-10-05 16:35:13 +07:00
|
|
|
u8 nr_maps;
|
2014-10-05 16:35:10 +07:00
|
|
|
} ____cacheline_aligned_in_smp;
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
|
|
|
|
#define MLX4_EN_BIT_DESC_OWN 0x80000000
|
|
|
|
#define CTRL_SIZE sizeof(struct mlx4_wqe_ctrl_seg)
|
|
|
|
#define MLX4_EN_MEMTYPE_PAD 0x100
|
|
|
|
#define DS_SIZE sizeof(struct mlx4_wqe_data_seg)
|
|
|
|
|
|
|
|
|
|
|
|
struct mlx4_en_tx_desc {
|
|
|
|
struct mlx4_wqe_ctrl_seg ctrl;
|
|
|
|
union {
|
|
|
|
struct mlx4_wqe_data_seg data; /* at least one data segment */
|
|
|
|
struct mlx4_wqe_lso_seg lso;
|
|
|
|
struct mlx4_wqe_inline_seg inl;
|
|
|
|
};
|
|
|
|
};
|
|
|
|
|
|
|
|
#define MLX4_EN_USE_SRQ 0x01000000
|
|
|
|
|
2011-03-23 05:38:07 +07:00
|
|
|
#define MLX4_EN_CX3_LOW_ID 0x1000
|
|
|
|
#define MLX4_EN_CX3_HIGH_ID 0x1005
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
struct mlx4_en_rx_alloc {
|
2013-06-23 22:17:56 +07:00
|
|
|
struct page *page;
|
|
|
|
dma_addr_t dma;
|
2013-10-07 18:38:12 +07:00
|
|
|
u32 page_offset;
|
2008-10-23 05:47:49 +07:00
|
|
|
};
|
|
|
|
|
2016-07-20 02:16:52 +07:00
|
|
|
#define MLX4_EN_CACHE_SIZE (2 * NAPI_POLL_WEIGHT)
|
2017-03-08 23:17:10 +07:00
|
|
|
|
2016-07-20 02:16:52 +07:00
|
|
|
struct mlx4_en_page_cache {
|
|
|
|
u32 index;
|
2017-03-08 23:17:10 +07:00
|
|
|
struct {
|
|
|
|
struct page *page;
|
|
|
|
dma_addr_t dma;
|
|
|
|
} buf[MLX4_EN_CACHE_SIZE];
|
2016-07-20 02:16:52 +07:00
|
|
|
};
|
|
|
|
|
2016-07-20 02:16:55 +07:00
|
|
|
struct mlx4_en_priv;
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
struct mlx4_en_tx_ring {
|
2014-10-05 16:35:10 +07:00
|
|
|
/* cache line used and dirtied in tx completion
|
|
|
|
* (mlx4_en_free_tx_buf())
|
|
|
|
*/
|
|
|
|
u32 last_nr_txbb;
|
|
|
|
u32 cons;
|
|
|
|
unsigned long wake_queue;
|
mlx4: reorganize struct mlx4_en_tx_ring
Goal is to reorganize this critical structure to increase performance.
ndo_start_xmit() should only dirty one cache line, and access as few
cache lines as possible.
Add sp_ (Slow Path) prefix to fields that are not used in fast path,
to make clear what is going on.
After this patch pahole reports something much better, as all
ndo_start_xmit() needed fields are packed into two cache lines instead
of seven or eight
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
struct netdev_queue * tx_queue; /* 0x10 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x18 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x20 0x8 */
/* XXX 24 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
unsigned int tx_dropped; /* 0x44 0x4 */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
struct mlx4_bf bf; /* 0x70 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 8 bytes ago --- */
__be32 doorbell_qpn; /* 0x88 0x4 */
__be32 mr_key; /* 0x8c 0x4 */
u32 size; /* 0x90 0x4 */
u32 size_mask; /* 0x94 0x4 */
u32 full_size; /* 0x98 0x4 */
u32 buf_size; /* 0x9c 0x4 */
void * buf; /* 0xa0 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0xa8 0x8 */
int qpn; /* 0xb0 0x4 */
u8 queue_index; /* 0xb4 0x1 */
bool bf_enabled; /* 0xb5 0x1 */
bool bf_alloced; /* 0xb6 0x1 */
u8 hwtstamp_tx_type; /* 0xb7 0x1 */
u8 * bounce_buf; /* 0xb8 0x8 */
/* --- cacheline 3 boundary (192 bytes) --- */
long unsigned int queue_stopped; /* 0xc0 0x8 */
struct mlx4_hwq_resources sp_wqres; /* 0xc8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 32 bytes ago --- */
struct mlx4_qp sp_qp; /* 0x120 0x30 */
/* --- cacheline 5 boundary (320 bytes) was 16 bytes ago --- */
struct mlx4_qp_context sp_context; /* 0x150 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 8 bytes ago --- */
cpumask_t sp_affinity_mask; /* 0x248 0x20 */
enum mlx4_qp_state sp_qp_state; /* 0x268 0x4 */
u16 sp_stride; /* 0x26c 0x2 */
u16 sp_cqn; /* 0x26e 0x2 */
/* size: 640, cachelines: 10, members: 36 */
/* sum members: 600, holes: 1, sum holes: 24 */
/* padding: 16 */
};
Instead of this silly placement :
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
/* XXX 48 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
/* XXX 4 bytes hole, try to pack */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
unsigned int tx_dropped; /* 0x70 0x4 */
/* XXX 4 bytes hole, try to pack */
struct mlx4_bf bf; /* 0x78 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 16 bytes ago --- */
long unsigned int queue_stopped; /* 0x90 0x8 */
cpumask_t affinity_mask; /* 0x98 0x10 */
struct mlx4_qp qp; /* 0xa8 0x30 */
/* --- cacheline 3 boundary (192 bytes) was 24 bytes ago --- */
struct mlx4_hwq_resources wqres; /* 0xd8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 48 bytes ago --- */
u32 size; /* 0x130 0x4 */
u32 size_mask; /* 0x134 0x4 */
u16 stride; /* 0x138 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 full_size; /* 0x13c 0x4 */
/* --- cacheline 5 boundary (320 bytes) --- */
u16 cqn; /* 0x140 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 buf_size; /* 0x144 0x4 */
__be32 doorbell_qpn; /* 0x148 0x4 */
__be32 mr_key; /* 0x14c 0x4 */
void * buf; /* 0x150 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0x158 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x160 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x168 0x8 */
u8 * bounce_buf; /* 0x170 0x8 */
struct mlx4_qp_context context; /* 0x178 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 48 bytes ago --- */
int qpn; /* 0x270 0x4 */
enum mlx4_qp_state qp_state; /* 0x274 0x4 */
u8 queue_index; /* 0x278 0x1 */
bool bf_enabled; /* 0x279 0x1 */
bool bf_alloced; /* 0x27a 0x1 */
/* XXX 5 bytes hole, try to pack */
/* --- cacheline 10 boundary (640 bytes) --- */
struct netdev_queue * tx_queue; /* 0x280 0x8 */
int hwtstamp_tx_type; /* 0x288 0x4 */
/* size: 704, cachelines: 11, members: 36 */
/* sum members: 587, holes: 6, sum holes: 65 */
/* padding: 52 */
};
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-23 06:56:10 +07:00
|
|
|
struct netdev_queue *tx_queue;
|
|
|
|
u32 (*free_tx_desc)(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_tx_ring *ring,
|
2017-06-15 18:35:31 +07:00
|
|
|
int index,
|
mlx4: reorganize struct mlx4_en_tx_ring
Goal is to reorganize this critical structure to increase performance.
ndo_start_xmit() should only dirty one cache line, and access as few
cache lines as possible.
Add sp_ (Slow Path) prefix to fields that are not used in fast path,
to make clear what is going on.
After this patch pahole reports something much better, as all
ndo_start_xmit() needed fields are packed into two cache lines instead
of seven or eight
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
struct netdev_queue * tx_queue; /* 0x10 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x18 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x20 0x8 */
/* XXX 24 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
unsigned int tx_dropped; /* 0x44 0x4 */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
struct mlx4_bf bf; /* 0x70 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 8 bytes ago --- */
__be32 doorbell_qpn; /* 0x88 0x4 */
__be32 mr_key; /* 0x8c 0x4 */
u32 size; /* 0x90 0x4 */
u32 size_mask; /* 0x94 0x4 */
u32 full_size; /* 0x98 0x4 */
u32 buf_size; /* 0x9c 0x4 */
void * buf; /* 0xa0 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0xa8 0x8 */
int qpn; /* 0xb0 0x4 */
u8 queue_index; /* 0xb4 0x1 */
bool bf_enabled; /* 0xb5 0x1 */
bool bf_alloced; /* 0xb6 0x1 */
u8 hwtstamp_tx_type; /* 0xb7 0x1 */
u8 * bounce_buf; /* 0xb8 0x8 */
/* --- cacheline 3 boundary (192 bytes) --- */
long unsigned int queue_stopped; /* 0xc0 0x8 */
struct mlx4_hwq_resources sp_wqres; /* 0xc8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 32 bytes ago --- */
struct mlx4_qp sp_qp; /* 0x120 0x30 */
/* --- cacheline 5 boundary (320 bytes) was 16 bytes ago --- */
struct mlx4_qp_context sp_context; /* 0x150 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 8 bytes ago --- */
cpumask_t sp_affinity_mask; /* 0x248 0x20 */
enum mlx4_qp_state sp_qp_state; /* 0x268 0x4 */
u16 sp_stride; /* 0x26c 0x2 */
u16 sp_cqn; /* 0x26e 0x2 */
/* size: 640, cachelines: 10, members: 36 */
/* sum members: 600, holes: 1, sum holes: 24 */
/* padding: 16 */
};
Instead of this silly placement :
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
/* XXX 48 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
/* XXX 4 bytes hole, try to pack */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
unsigned int tx_dropped; /* 0x70 0x4 */
/* XXX 4 bytes hole, try to pack */
struct mlx4_bf bf; /* 0x78 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 16 bytes ago --- */
long unsigned int queue_stopped; /* 0x90 0x8 */
cpumask_t affinity_mask; /* 0x98 0x10 */
struct mlx4_qp qp; /* 0xa8 0x30 */
/* --- cacheline 3 boundary (192 bytes) was 24 bytes ago --- */
struct mlx4_hwq_resources wqres; /* 0xd8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 48 bytes ago --- */
u32 size; /* 0x130 0x4 */
u32 size_mask; /* 0x134 0x4 */
u16 stride; /* 0x138 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 full_size; /* 0x13c 0x4 */
/* --- cacheline 5 boundary (320 bytes) --- */
u16 cqn; /* 0x140 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 buf_size; /* 0x144 0x4 */
__be32 doorbell_qpn; /* 0x148 0x4 */
__be32 mr_key; /* 0x14c 0x4 */
void * buf; /* 0x150 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0x158 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x160 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x168 0x8 */
u8 * bounce_buf; /* 0x170 0x8 */
struct mlx4_qp_context context; /* 0x178 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 48 bytes ago --- */
int qpn; /* 0x270 0x4 */
enum mlx4_qp_state qp_state; /* 0x274 0x4 */
u8 queue_index; /* 0x278 0x1 */
bool bf_enabled; /* 0x279 0x1 */
bool bf_alloced; /* 0x27a 0x1 */
/* XXX 5 bytes hole, try to pack */
/* --- cacheline 10 boundary (640 bytes) --- */
struct netdev_queue * tx_queue; /* 0x280 0x8 */
int hwtstamp_tx_type; /* 0x288 0x4 */
/* size: 704, cachelines: 11, members: 36 */
/* sum members: 587, holes: 6, sum holes: 65 */
/* padding: 52 */
};
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-23 06:56:10 +07:00
|
|
|
u64 timestamp, int napi_mode);
|
|
|
|
struct mlx4_en_rx_ring *recycle_ring;
|
2014-10-05 16:35:10 +07:00
|
|
|
|
|
|
|
/* cache line used and dirtied in mlx4_en_xmit() */
|
|
|
|
u32 prod ____cacheline_aligned_in_smp;
|
mlx4: reorganize struct mlx4_en_tx_ring
Goal is to reorganize this critical structure to increase performance.
ndo_start_xmit() should only dirty one cache line, and access as few
cache lines as possible.
Add sp_ (Slow Path) prefix to fields that are not used in fast path,
to make clear what is going on.
After this patch pahole reports something much better, as all
ndo_start_xmit() needed fields are packed into two cache lines instead
of seven or eight
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
struct netdev_queue * tx_queue; /* 0x10 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x18 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x20 0x8 */
/* XXX 24 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
unsigned int tx_dropped; /* 0x44 0x4 */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
struct mlx4_bf bf; /* 0x70 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 8 bytes ago --- */
__be32 doorbell_qpn; /* 0x88 0x4 */
__be32 mr_key; /* 0x8c 0x4 */
u32 size; /* 0x90 0x4 */
u32 size_mask; /* 0x94 0x4 */
u32 full_size; /* 0x98 0x4 */
u32 buf_size; /* 0x9c 0x4 */
void * buf; /* 0xa0 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0xa8 0x8 */
int qpn; /* 0xb0 0x4 */
u8 queue_index; /* 0xb4 0x1 */
bool bf_enabled; /* 0xb5 0x1 */
bool bf_alloced; /* 0xb6 0x1 */
u8 hwtstamp_tx_type; /* 0xb7 0x1 */
u8 * bounce_buf; /* 0xb8 0x8 */
/* --- cacheline 3 boundary (192 bytes) --- */
long unsigned int queue_stopped; /* 0xc0 0x8 */
struct mlx4_hwq_resources sp_wqres; /* 0xc8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 32 bytes ago --- */
struct mlx4_qp sp_qp; /* 0x120 0x30 */
/* --- cacheline 5 boundary (320 bytes) was 16 bytes ago --- */
struct mlx4_qp_context sp_context; /* 0x150 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 8 bytes ago --- */
cpumask_t sp_affinity_mask; /* 0x248 0x20 */
enum mlx4_qp_state sp_qp_state; /* 0x268 0x4 */
u16 sp_stride; /* 0x26c 0x2 */
u16 sp_cqn; /* 0x26e 0x2 */
/* size: 640, cachelines: 10, members: 36 */
/* sum members: 600, holes: 1, sum holes: 24 */
/* padding: 16 */
};
Instead of this silly placement :
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
/* XXX 48 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
/* XXX 4 bytes hole, try to pack */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
unsigned int tx_dropped; /* 0x70 0x4 */
/* XXX 4 bytes hole, try to pack */
struct mlx4_bf bf; /* 0x78 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 16 bytes ago --- */
long unsigned int queue_stopped; /* 0x90 0x8 */
cpumask_t affinity_mask; /* 0x98 0x10 */
struct mlx4_qp qp; /* 0xa8 0x30 */
/* --- cacheline 3 boundary (192 bytes) was 24 bytes ago --- */
struct mlx4_hwq_resources wqres; /* 0xd8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 48 bytes ago --- */
u32 size; /* 0x130 0x4 */
u32 size_mask; /* 0x134 0x4 */
u16 stride; /* 0x138 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 full_size; /* 0x13c 0x4 */
/* --- cacheline 5 boundary (320 bytes) --- */
u16 cqn; /* 0x140 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 buf_size; /* 0x144 0x4 */
__be32 doorbell_qpn; /* 0x148 0x4 */
__be32 mr_key; /* 0x14c 0x4 */
void * buf; /* 0x150 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0x158 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x160 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x168 0x8 */
u8 * bounce_buf; /* 0x170 0x8 */
struct mlx4_qp_context context; /* 0x178 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 48 bytes ago --- */
int qpn; /* 0x270 0x4 */
enum mlx4_qp_state qp_state; /* 0x274 0x4 */
u8 queue_index; /* 0x278 0x1 */
bool bf_enabled; /* 0x279 0x1 */
bool bf_alloced; /* 0x27a 0x1 */
/* XXX 5 bytes hole, try to pack */
/* --- cacheline 10 boundary (640 bytes) --- */
struct netdev_queue * tx_queue; /* 0x280 0x8 */
int hwtstamp_tx_type; /* 0x288 0x4 */
/* size: 704, cachelines: 11, members: 36 */
/* sum members: 587, holes: 6, sum holes: 65 */
/* padding: 52 */
};
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-23 06:56:10 +07:00
|
|
|
unsigned int tx_dropped;
|
2014-10-05 16:35:10 +07:00
|
|
|
unsigned long bytes;
|
|
|
|
unsigned long packets;
|
|
|
|
unsigned long tx_csum;
|
|
|
|
unsigned long tso_packets;
|
|
|
|
unsigned long xmit_more;
|
|
|
|
struct mlx4_bf bf;
|
|
|
|
|
|
|
|
/* Following part should be mostly read */
|
mlx4: reorganize struct mlx4_en_tx_ring
Goal is to reorganize this critical structure to increase performance.
ndo_start_xmit() should only dirty one cache line, and access as few
cache lines as possible.
Add sp_ (Slow Path) prefix to fields that are not used in fast path,
to make clear what is going on.
After this patch pahole reports something much better, as all
ndo_start_xmit() needed fields are packed into two cache lines instead
of seven or eight
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
struct netdev_queue * tx_queue; /* 0x10 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x18 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x20 0x8 */
/* XXX 24 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
unsigned int tx_dropped; /* 0x44 0x4 */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
struct mlx4_bf bf; /* 0x70 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 8 bytes ago --- */
__be32 doorbell_qpn; /* 0x88 0x4 */
__be32 mr_key; /* 0x8c 0x4 */
u32 size; /* 0x90 0x4 */
u32 size_mask; /* 0x94 0x4 */
u32 full_size; /* 0x98 0x4 */
u32 buf_size; /* 0x9c 0x4 */
void * buf; /* 0xa0 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0xa8 0x8 */
int qpn; /* 0xb0 0x4 */
u8 queue_index; /* 0xb4 0x1 */
bool bf_enabled; /* 0xb5 0x1 */
bool bf_alloced; /* 0xb6 0x1 */
u8 hwtstamp_tx_type; /* 0xb7 0x1 */
u8 * bounce_buf; /* 0xb8 0x8 */
/* --- cacheline 3 boundary (192 bytes) --- */
long unsigned int queue_stopped; /* 0xc0 0x8 */
struct mlx4_hwq_resources sp_wqres; /* 0xc8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 32 bytes ago --- */
struct mlx4_qp sp_qp; /* 0x120 0x30 */
/* --- cacheline 5 boundary (320 bytes) was 16 bytes ago --- */
struct mlx4_qp_context sp_context; /* 0x150 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 8 bytes ago --- */
cpumask_t sp_affinity_mask; /* 0x248 0x20 */
enum mlx4_qp_state sp_qp_state; /* 0x268 0x4 */
u16 sp_stride; /* 0x26c 0x2 */
u16 sp_cqn; /* 0x26e 0x2 */
/* size: 640, cachelines: 10, members: 36 */
/* sum members: 600, holes: 1, sum holes: 24 */
/* padding: 16 */
};
Instead of this silly placement :
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
/* XXX 48 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
/* XXX 4 bytes hole, try to pack */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
unsigned int tx_dropped; /* 0x70 0x4 */
/* XXX 4 bytes hole, try to pack */
struct mlx4_bf bf; /* 0x78 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 16 bytes ago --- */
long unsigned int queue_stopped; /* 0x90 0x8 */
cpumask_t affinity_mask; /* 0x98 0x10 */
struct mlx4_qp qp; /* 0xa8 0x30 */
/* --- cacheline 3 boundary (192 bytes) was 24 bytes ago --- */
struct mlx4_hwq_resources wqres; /* 0xd8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 48 bytes ago --- */
u32 size; /* 0x130 0x4 */
u32 size_mask; /* 0x134 0x4 */
u16 stride; /* 0x138 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 full_size; /* 0x13c 0x4 */
/* --- cacheline 5 boundary (320 bytes) --- */
u16 cqn; /* 0x140 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 buf_size; /* 0x144 0x4 */
__be32 doorbell_qpn; /* 0x148 0x4 */
__be32 mr_key; /* 0x14c 0x4 */
void * buf; /* 0x150 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0x158 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x160 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x168 0x8 */
u8 * bounce_buf; /* 0x170 0x8 */
struct mlx4_qp_context context; /* 0x178 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 48 bytes ago --- */
int qpn; /* 0x270 0x4 */
enum mlx4_qp_state qp_state; /* 0x274 0x4 */
u8 queue_index; /* 0x278 0x1 */
bool bf_enabled; /* 0x279 0x1 */
bool bf_alloced; /* 0x27a 0x1 */
/* XXX 5 bytes hole, try to pack */
/* --- cacheline 10 boundary (640 bytes) --- */
struct netdev_queue * tx_queue; /* 0x280 0x8 */
int hwtstamp_tx_type; /* 0x288 0x4 */
/* size: 704, cachelines: 11, members: 36 */
/* sum members: 587, holes: 6, sum holes: 65 */
/* padding: 52 */
};
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-23 06:56:10 +07:00
|
|
|
__be32 doorbell_qpn;
|
|
|
|
__be32 mr_key;
|
2014-10-05 16:35:10 +07:00
|
|
|
u32 size; /* number of TXBBs */
|
|
|
|
u32 size_mask;
|
2015-06-25 15:29:42 +07:00
|
|
|
u32 full_size;
|
2014-10-05 16:35:10 +07:00
|
|
|
u32 buf_size;
|
|
|
|
void *buf;
|
|
|
|
struct mlx4_en_tx_info *tx_info;
|
|
|
|
int qpn;
|
|
|
|
u8 queue_index;
|
|
|
|
bool bf_enabled;
|
|
|
|
bool bf_alloced;
|
mlx4: reorganize struct mlx4_en_tx_ring
Goal is to reorganize this critical structure to increase performance.
ndo_start_xmit() should only dirty one cache line, and access as few
cache lines as possible.
Add sp_ (Slow Path) prefix to fields that are not used in fast path,
to make clear what is going on.
After this patch pahole reports something much better, as all
ndo_start_xmit() needed fields are packed into two cache lines instead
of seven or eight
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
struct netdev_queue * tx_queue; /* 0x10 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x18 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x20 0x8 */
/* XXX 24 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
unsigned int tx_dropped; /* 0x44 0x4 */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
struct mlx4_bf bf; /* 0x70 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 8 bytes ago --- */
__be32 doorbell_qpn; /* 0x88 0x4 */
__be32 mr_key; /* 0x8c 0x4 */
u32 size; /* 0x90 0x4 */
u32 size_mask; /* 0x94 0x4 */
u32 full_size; /* 0x98 0x4 */
u32 buf_size; /* 0x9c 0x4 */
void * buf; /* 0xa0 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0xa8 0x8 */
int qpn; /* 0xb0 0x4 */
u8 queue_index; /* 0xb4 0x1 */
bool bf_enabled; /* 0xb5 0x1 */
bool bf_alloced; /* 0xb6 0x1 */
u8 hwtstamp_tx_type; /* 0xb7 0x1 */
u8 * bounce_buf; /* 0xb8 0x8 */
/* --- cacheline 3 boundary (192 bytes) --- */
long unsigned int queue_stopped; /* 0xc0 0x8 */
struct mlx4_hwq_resources sp_wqres; /* 0xc8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 32 bytes ago --- */
struct mlx4_qp sp_qp; /* 0x120 0x30 */
/* --- cacheline 5 boundary (320 bytes) was 16 bytes ago --- */
struct mlx4_qp_context sp_context; /* 0x150 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 8 bytes ago --- */
cpumask_t sp_affinity_mask; /* 0x248 0x20 */
enum mlx4_qp_state sp_qp_state; /* 0x268 0x4 */
u16 sp_stride; /* 0x26c 0x2 */
u16 sp_cqn; /* 0x26e 0x2 */
/* size: 640, cachelines: 10, members: 36 */
/* sum members: 600, holes: 1, sum holes: 24 */
/* padding: 16 */
};
Instead of this silly placement :
struct mlx4_en_tx_ring {
u32 last_nr_txbb; /* 0 0x4 */
u32 cons; /* 0x4 0x4 */
long unsigned int wake_queue; /* 0x8 0x8 */
/* XXX 48 bytes hole, try to pack */
/* --- cacheline 1 boundary (64 bytes) --- */
u32 prod; /* 0x40 0x4 */
/* XXX 4 bytes hole, try to pack */
long unsigned int bytes; /* 0x48 0x8 */
long unsigned int packets; /* 0x50 0x8 */
long unsigned int tx_csum; /* 0x58 0x8 */
long unsigned int tso_packets; /* 0x60 0x8 */
long unsigned int xmit_more; /* 0x68 0x8 */
unsigned int tx_dropped; /* 0x70 0x4 */
/* XXX 4 bytes hole, try to pack */
struct mlx4_bf bf; /* 0x78 0x18 */
/* --- cacheline 2 boundary (128 bytes) was 16 bytes ago --- */
long unsigned int queue_stopped; /* 0x90 0x8 */
cpumask_t affinity_mask; /* 0x98 0x10 */
struct mlx4_qp qp; /* 0xa8 0x30 */
/* --- cacheline 3 boundary (192 bytes) was 24 bytes ago --- */
struct mlx4_hwq_resources wqres; /* 0xd8 0x58 */
/* --- cacheline 4 boundary (256 bytes) was 48 bytes ago --- */
u32 size; /* 0x130 0x4 */
u32 size_mask; /* 0x134 0x4 */
u16 stride; /* 0x138 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 full_size; /* 0x13c 0x4 */
/* --- cacheline 5 boundary (320 bytes) --- */
u16 cqn; /* 0x140 0x2 */
/* XXX 2 bytes hole, try to pack */
u32 buf_size; /* 0x144 0x4 */
__be32 doorbell_qpn; /* 0x148 0x4 */
__be32 mr_key; /* 0x14c 0x4 */
void * buf; /* 0x150 0x8 */
struct mlx4_en_tx_info * tx_info; /* 0x158 0x8 */
struct mlx4_en_rx_ring * recycle_ring; /* 0x160 0x8 */
u32 (*free_tx_desc)(struct mlx4_en_priv *, struct mlx4_en_tx_ring *, int, u8, u64, int); /* 0x168 0x8 */
u8 * bounce_buf; /* 0x170 0x8 */
struct mlx4_qp_context context; /* 0x178 0xf8 */
/* --- cacheline 9 boundary (576 bytes) was 48 bytes ago --- */
int qpn; /* 0x270 0x4 */
enum mlx4_qp_state qp_state; /* 0x274 0x4 */
u8 queue_index; /* 0x278 0x1 */
bool bf_enabled; /* 0x279 0x1 */
bool bf_alloced; /* 0x27a 0x1 */
/* XXX 5 bytes hole, try to pack */
/* --- cacheline 10 boundary (640 bytes) --- */
struct netdev_queue * tx_queue; /* 0x280 0x8 */
int hwtstamp_tx_type; /* 0x288 0x4 */
/* size: 704, cachelines: 11, members: 36 */
/* sum members: 587, holes: 6, sum holes: 65 */
/* padding: 52 */
};
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-11-23 06:56:10 +07:00
|
|
|
u8 hwtstamp_tx_type;
|
|
|
|
u8 *bounce_buf;
|
|
|
|
|
|
|
|
/* Not used in fast path
|
|
|
|
* Only queue_stopped might be used if BQL is not properly working.
|
|
|
|
*/
|
|
|
|
unsigned long queue_stopped;
|
|
|
|
struct mlx4_hwq_resources sp_wqres;
|
|
|
|
struct mlx4_qp sp_qp;
|
|
|
|
struct mlx4_qp_context sp_context;
|
|
|
|
cpumask_t sp_affinity_mask;
|
|
|
|
enum mlx4_qp_state sp_qp_state;
|
|
|
|
u16 sp_stride;
|
|
|
|
u16 sp_cqn; /* index of port CQ associated with this ring */
|
2014-10-05 16:35:10 +07:00
|
|
|
} ____cacheline_aligned_in_smp;
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
struct mlx4_en_rx_desc {
|
|
|
|
/* actual number of entries depends on rx ring stride */
|
|
|
|
struct mlx4_wqe_data_seg data[0];
|
|
|
|
};
|
|
|
|
|
|
|
|
struct mlx4_en_rx_ring {
|
|
|
|
struct mlx4_hwq_resources wqres;
|
|
|
|
u32 size ; /* number of Rx descs*/
|
|
|
|
u32 actual_size;
|
|
|
|
u32 size_mask;
|
|
|
|
u16 stride;
|
|
|
|
u16 log_stride;
|
|
|
|
u16 cqn; /* index of port CQ associated with this ring */
|
|
|
|
u32 prod;
|
|
|
|
u32 cons;
|
|
|
|
u32 buf_size;
|
2011-11-15 02:25:36 +07:00
|
|
|
u8 fcs_del;
|
2008-10-23 05:47:49 +07:00
|
|
|
void *buf;
|
|
|
|
void *rx_info;
|
net/mlx4_en: protect ring->xdp_prog with rcu_read_lock
Depending on the preempt mode, the bpf_prog stored in xdp_prog may be
freed despite the use of call_rcu inside bpf_prog_put. The situation is
possible when running in PREEMPT_RCU=y mode, for instance, since the rcu
callback for destroying the bpf prog can run even during the bh handling
in the mlx4 rx path.
Several options were considered before this patch was settled on:
Add a napi_synchronize loop in mlx4_xdp_set, which would occur after all
of the rings are updated with the new program.
This approach has the disadvantage that as the number of rings
increases, the speed of update will slow down significantly due to
napi_synchronize's msleep(1).
Add a new rcu_head in bpf_prog_aux, to be used by a new bpf_prog_put_bh.
The action of the bpf_prog_put_bh would be to then call bpf_prog_put
later. Those drivers that consume a bpf prog in a bh context (like mlx4)
would then use the bpf_prog_put_bh instead when the ring is up. This has
the problem of complexity, in maintaining proper refcnts and rcu lists,
and would likely be harder to review. In addition, this approach to
freeing must be exclusive with other frees of the bpf prog, for instance
a _bh prog must not be referenced from a prog array that is consumed by
a non-_bh prog.
The placement of rcu_read_lock in this patch is functionally the same as
putting an rcu_read_lock in napi_poll. Actually doing so could be a
potentially controversial change, but would bring the implementation in
line with sk_busy_loop (though of course the nature of those two paths
is substantially different), and would also avoid future copy/paste
problems with future supporters of XDP. Still, this patch does not take
that opinionated option.
Testing was done with kernels in either PREEMPT_RCU=y or
CONFIG_PREEMPT_VOLUNTARY=y+PREEMPT_RCU=n modes, with neither exhibiting
any drawback. With PREEMPT_RCU=n, the extra call to rcu_read_lock did
not show up in the perf report whatsoever, and with PREEMPT_RCU=y the
overhead of rcu_read_lock (according to perf) was the same before/after.
In the rx path, rcu_read_lock is eventually called for every packet
from netif_receive_skb_internal, so the napi poll call's rcu_read_lock
is easily amortized.
v2:
Remove extra rcu_read_lock in mlx4_en_process_rx_cq body
Annotate xdp_prog with __rcu, and convert all usages to rcu_assign or
rcu_dereference[_protected] as appropriate.
Add explicit mutex lock around rcu_assign instead of xchg loop.
Fixes: d576acf0a22 ("net/mlx4_en: add page recycle to prepare rx ring for tx support")
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <alexei.starovoitov@gmail.com>
Signed-off-by: Brenden Blanco <bblanco@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2016-09-04 11:29:58 +07:00
|
|
|
struct bpf_prog __rcu *xdp_prog;
|
2016-07-20 02:16:52 +07:00
|
|
|
struct mlx4_en_page_cache page_cache;
|
2008-10-23 05:47:49 +07:00
|
|
|
unsigned long bytes;
|
|
|
|
unsigned long packets;
|
2011-10-18 08:50:56 +07:00
|
|
|
unsigned long csum_ok;
|
|
|
|
unsigned long csum_none;
|
2014-11-09 18:51:53 +07:00
|
|
|
unsigned long csum_complete;
|
2017-03-08 23:17:14 +07:00
|
|
|
unsigned long rx_alloc_pages;
|
2016-11-02 22:12:25 +07:00
|
|
|
unsigned long xdp_drop;
|
|
|
|
unsigned long xdp_tx;
|
|
|
|
unsigned long xdp_tx_full;
|
2016-04-20 20:01:18 +07:00
|
|
|
unsigned long dropped;
|
2013-04-23 13:06:49 +07:00
|
|
|
int hwtstamp_rx_filter;
|
2014-06-09 14:24:39 +07:00
|
|
|
cpumask_var_t affinity_mask;
|
2018-01-03 17:25:39 +07:00
|
|
|
struct xdp_rxq_info xdp_rxq;
|
2008-10-23 05:47:49 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
struct mlx4_en_cq {
|
|
|
|
struct mlx4_cq mcq;
|
|
|
|
struct mlx4_hwq_resources wqres;
|
|
|
|
int ring;
|
|
|
|
struct net_device *dev;
|
2017-06-15 18:35:37 +07:00
|
|
|
union {
|
|
|
|
struct napi_struct napi;
|
|
|
|
bool xdp_busy;
|
|
|
|
};
|
2008-10-23 05:47:49 +07:00
|
|
|
int size;
|
|
|
|
int buf_size;
|
2015-05-31 13:30:16 +07:00
|
|
|
int vector;
|
2016-11-02 22:12:23 +07:00
|
|
|
enum cq_type type;
|
2008-10-23 05:47:49 +07:00
|
|
|
u16 moder_time;
|
|
|
|
u16 moder_cnt;
|
|
|
|
struct mlx4_cqe *buf;
|
|
|
|
#define MLX4_EN_OPCODE_ERROR 0x1e
|
2013-06-18 20:18:27 +07:00
|
|
|
|
2014-06-29 15:54:55 +07:00
|
|
|
struct irq_desc *irq_desc;
|
2008-10-23 05:47:49 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
struct mlx4_en_port_profile {
|
|
|
|
u32 flags;
|
2016-11-02 22:12:24 +07:00
|
|
|
u32 tx_ring_num[MLX4_EN_NUM_TX_TYPES];
|
2008-10-23 05:47:49 +07:00
|
|
|
u32 rx_ring_num;
|
|
|
|
u32 tx_ring_size;
|
|
|
|
u32 rx_ring_size;
|
2016-07-18 22:35:12 +07:00
|
|
|
u8 num_tx_rings_p_up;
|
2008-11-05 11:48:36 +07:00
|
|
|
u8 rx_pause;
|
|
|
|
u8 rx_ppp;
|
|
|
|
u8 tx_pause;
|
|
|
|
u8 tx_ppp;
|
2017-06-29 18:07:56 +07:00
|
|
|
u8 num_up;
|
2012-01-18 05:54:55 +07:00
|
|
|
int rss_rings;
|
2014-03-02 15:24:58 +07:00
|
|
|
int inline_thold;
|
2016-07-18 22:35:12 +07:00
|
|
|
struct hwtstamp_config hwtstamp_config;
|
2008-10-23 05:47:49 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
struct mlx4_en_profile {
|
2010-08-24 10:46:42 +07:00
|
|
|
int udp_rss;
|
2008-10-23 05:47:49 +07:00
|
|
|
u8 rss_mask;
|
|
|
|
u32 active_ports;
|
|
|
|
u32 small_pkt_int;
|
|
|
|
u8 no_reset;
|
2017-10-10 16:28:33 +07:00
|
|
|
u8 max_num_tx_rings_p_up;
|
2008-10-23 05:47:49 +07:00
|
|
|
struct mlx4_en_port_profile prof[MLX4_MAX_PORTS + 1];
|
|
|
|
};
|
|
|
|
|
|
|
|
struct mlx4_en_dev {
|
|
|
|
struct mlx4_dev *dev;
|
|
|
|
struct pci_dev *pdev;
|
|
|
|
struct mutex state_lock;
|
|
|
|
struct net_device *pndev[MLX4_MAX_PORTS + 1];
|
2015-02-03 21:48:34 +07:00
|
|
|
struct net_device *upper[MLX4_MAX_PORTS + 1];
|
2008-10-23 05:47:49 +07:00
|
|
|
u32 port_cnt;
|
|
|
|
bool device_up;
|
|
|
|
struct mlx4_en_profile profile;
|
|
|
|
u32 LSO_support;
|
|
|
|
struct workqueue_struct *workqueue;
|
|
|
|
struct device *dma_device;
|
|
|
|
void __iomem *uar_map;
|
|
|
|
struct mlx4_uar priv_uar;
|
|
|
|
struct mlx4_mr mr;
|
|
|
|
u32 priv_pdn;
|
|
|
|
spinlock_t uar_lock;
|
2010-08-24 10:46:38 +07:00
|
|
|
u8 mac_removed[MLX4_MAX_PORTS + 1];
|
2014-01-01 00:39:39 +07:00
|
|
|
u32 nominal_c_mult;
|
2013-04-23 13:06:49 +07:00
|
|
|
struct cyclecounter cycles;
|
2017-02-10 00:10:04 +07:00
|
|
|
seqlock_t clock_lock;
|
2013-04-23 13:06:49 +07:00
|
|
|
struct timecounter clock;
|
|
|
|
unsigned long last_overflow_check;
|
2014-01-01 00:39:39 +07:00
|
|
|
struct ptp_clock *ptp_clock;
|
|
|
|
struct ptp_clock_info ptp_clock_info;
|
2015-02-03 21:48:34 +07:00
|
|
|
struct notifier_block nb;
|
2008-10-23 05:47:49 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
struct mlx4_en_rss_map {
|
|
|
|
int base_qpn;
|
2009-08-07 09:27:51 +07:00
|
|
|
struct mlx4_qp qps[MAX_RX_RINGS];
|
|
|
|
enum mlx4_qp_state state[MAX_RX_RINGS];
|
2017-06-15 18:35:32 +07:00
|
|
|
struct mlx4_qp *indir_qp;
|
2008-10-23 05:47:49 +07:00
|
|
|
enum mlx4_qp_state indir_state;
|
|
|
|
};
|
|
|
|
|
2014-10-27 16:37:40 +07:00
|
|
|
enum mlx4_en_port_flag {
|
|
|
|
MLX4_EN_PORT_ANC = 1<<0, /* Auto-negotiation complete */
|
|
|
|
MLX4_EN_PORT_ANE = 1<<1, /* Auto-negotiation enabled */
|
|
|
|
};
|
|
|
|
|
2010-08-24 10:46:18 +07:00
|
|
|
struct mlx4_en_port_state {
|
|
|
|
int link_state;
|
|
|
|
int link_speed;
|
2014-10-27 16:37:40 +07:00
|
|
|
int transceiver;
|
|
|
|
u32 flags;
|
2010-08-24 10:46:18 +07:00
|
|
|
};
|
|
|
|
|
2012-07-05 11:03:43 +07:00
|
|
|
enum mlx4_en_mclist_act {
|
|
|
|
MCLIST_NONE,
|
|
|
|
MCLIST_REM,
|
|
|
|
MCLIST_ADD,
|
|
|
|
};
|
|
|
|
|
|
|
|
struct mlx4_en_mc_list {
|
|
|
|
struct list_head list;
|
|
|
|
enum mlx4_en_mclist_act action;
|
|
|
|
u8 addr[ETH_ALEN];
|
{NET, IB}/mlx4: Add device managed flow steering firmware API
The driver is modified to support three operation modes.
If supported by firmware use the device managed flow steering
API, that which we call device managed steering mode. Else, if
the firmware supports the B0 steering mode use it, and finally,
if none of the above, use the A0 steering mode.
When the steering mode is device managed, the code is modified
such that L2 based rules set by the mlx4_en driver for Ethernet
unicast and multicast, and the IB stack multicast attach calls
done through the mlx4_ib driver are all routed to use the device
managed API.
When attaching rule using device managed flow steering API,
the firmware returns a 64 bit registration id, which is to be
provided during detach.
Currently the firmware is always programmed during HCA initialization
to use standard L2 hashing. Future work should be done to allow
configuring the flow-steering hash function with common, non
proprietary means.
Signed-off-by: Hadar Hen Zion <hadarh@mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-07-05 11:03:46 +07:00
|
|
|
u64 reg_id;
|
2013-12-23 21:09:44 +07:00
|
|
|
u64 tunnel_reg_id;
|
2012-07-05 11:03:43 +07:00
|
|
|
};
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
struct mlx4_en_frag_info {
|
|
|
|
u16 frag_size;
|
2017-03-08 23:17:08 +07:00
|
|
|
u32 frag_stride;
|
2008-10-23 05:47:49 +07:00
|
|
|
};
|
|
|
|
|
2012-04-05 04:33:26 +07:00
|
|
|
#ifdef CONFIG_MLX4_EN_DCB
|
|
|
|
/* Minimal TC BW - setting to 0 will block traffic */
|
|
|
|
#define MLX4_EN_BW_MIN 1
|
|
|
|
#define MLX4_EN_BW_MAX 100 /* Utilize 100% of the line */
|
|
|
|
|
2017-12-28 21:26:11 +07:00
|
|
|
#define MLX4_EN_TC_VENDOR 0
|
2012-04-05 04:33:26 +07:00
|
|
|
#define MLX4_EN_TC_ETS 7
|
|
|
|
|
2016-06-21 16:43:59 +07:00
|
|
|
enum dcb_pfc_type {
|
|
|
|
pfc_disabled = 0,
|
|
|
|
pfc_enabled_full,
|
|
|
|
pfc_enabled_tx,
|
|
|
|
pfc_enabled_rx
|
|
|
|
};
|
|
|
|
|
|
|
|
struct mlx4_en_cee_config {
|
|
|
|
bool pfc_state;
|
2017-06-29 18:07:56 +07:00
|
|
|
enum dcb_pfc_type dcb_pfc[MLX4_EN_NUM_UP_HIGH];
|
2016-06-21 16:43:59 +07:00
|
|
|
};
|
2012-04-05 04:33:26 +07:00
|
|
|
#endif
|
|
|
|
|
2012-07-05 11:03:49 +07:00
|
|
|
struct ethtool_flow_id {
|
2013-01-31 06:07:08 +07:00
|
|
|
struct list_head list;
|
2012-07-05 11:03:49 +07:00
|
|
|
struct ethtool_rx_flow_spec flow_spec;
|
|
|
|
u64 id;
|
|
|
|
};
|
|
|
|
|
2013-02-07 09:25:19 +07:00
|
|
|
enum {
|
|
|
|
MLX4_EN_FLAG_PROMISC = (1 << 0),
|
|
|
|
MLX4_EN_FLAG_MC_PROMISC = (1 << 1),
|
|
|
|
/* whether we need to enable hardware loopback by putting dmac
|
|
|
|
* in Tx WQE
|
|
|
|
*/
|
|
|
|
MLX4_EN_FLAG_ENABLE_HW_LOOPBACK = (1 << 2),
|
|
|
|
/* whether we need to drop packets that hardware loopback-ed */
|
2013-02-07 09:25:26 +07:00
|
|
|
MLX4_EN_FLAG_RX_FILTER_NEEDED = (1 << 3),
|
2014-11-09 18:51:53 +07:00
|
|
|
MLX4_EN_FLAG_FORCE_PROMISC = (1 << 4),
|
|
|
|
MLX4_EN_FLAG_RX_CSUM_NON_TCP_UDP = (1 << 5),
|
2016-06-21 16:43:59 +07:00
|
|
|
#ifdef CONFIG_MLX4_EN_DCB
|
|
|
|
MLX4_EN_FLAG_DCB_ENABLED = (1 << 6),
|
|
|
|
#endif
|
2013-02-07 09:25:19 +07:00
|
|
|
};
|
|
|
|
|
2015-04-02 20:31:20 +07:00
|
|
|
#define PORT_BEACON_MAX_LIMIT (65535)
|
2013-02-07 09:25:25 +07:00
|
|
|
#define MLX4_EN_MAC_HASH_SIZE (1 << BITS_PER_BYTE)
|
|
|
|
#define MLX4_EN_MAC_HASH_IDX 5
|
|
|
|
|
2015-03-30 21:45:24 +07:00
|
|
|
struct mlx4_en_stats_bitmap {
|
|
|
|
DECLARE_BITMAP(bitmap, NUM_ALL_STATS);
|
|
|
|
struct mutex mutex; /* for mutual access to stats bitmap */
|
|
|
|
};
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
struct mlx4_en_priv {
|
|
|
|
struct mlx4_en_dev *mdev;
|
|
|
|
struct mlx4_en_port_profile *prof;
|
|
|
|
struct net_device *dev;
|
2011-07-20 11:54:22 +07:00
|
|
|
unsigned long active_vlans[BITS_TO_LONGS(VLAN_N_VID)];
|
2010-08-24 10:46:18 +07:00
|
|
|
struct mlx4_en_port_state port_state;
|
2008-10-23 05:47:49 +07:00
|
|
|
spinlock_t stats_lock;
|
2012-07-05 11:03:49 +07:00
|
|
|
struct ethtool_flow_id ethtool_rules[MAX_NUM_OF_FS_RULES];
|
2013-01-31 06:07:08 +07:00
|
|
|
/* To allow rules removal while port is going down */
|
|
|
|
struct list_head ethtool_list;
|
2008-10-23 05:47:49 +07:00
|
|
|
|
2011-10-09 12:38:23 +07:00
|
|
|
unsigned long last_moder_packets[MAX_RX_RINGS];
|
2008-10-23 05:47:49 +07:00
|
|
|
unsigned long last_moder_tx_packets;
|
2011-10-09 12:38:23 +07:00
|
|
|
unsigned long last_moder_bytes[MAX_RX_RINGS];
|
2008-10-23 05:47:49 +07:00
|
|
|
unsigned long last_moder_jiffies;
|
2011-10-09 12:38:23 +07:00
|
|
|
int last_moder_time[MAX_RX_RINGS];
|
2008-10-23 05:47:49 +07:00
|
|
|
u16 rx_usecs;
|
|
|
|
u16 rx_frames;
|
|
|
|
u16 tx_usecs;
|
|
|
|
u16 tx_frames;
|
|
|
|
u32 pkt_rate_low;
|
|
|
|
u16 rx_usecs_low;
|
|
|
|
u32 pkt_rate_high;
|
|
|
|
u16 rx_usecs_high;
|
|
|
|
u16 sample_interval;
|
|
|
|
u16 adaptive_rx_coal;
|
|
|
|
u32 msg_enable;
|
2010-08-24 10:46:18 +07:00
|
|
|
u32 loopback_ok;
|
|
|
|
u32 validate_loopback;
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
struct mlx4_hwq_resources res;
|
|
|
|
int link_state;
|
|
|
|
int last_link_state;
|
|
|
|
bool port_up;
|
|
|
|
int port;
|
|
|
|
int registered;
|
|
|
|
int allocated;
|
|
|
|
int stride;
|
2014-07-08 15:25:24 +07:00
|
|
|
unsigned char current_mac[ETH_ALEN + 2];
|
2008-10-23 05:47:49 +07:00
|
|
|
int mac_index;
|
|
|
|
unsigned max_mtu;
|
|
|
|
int base_qpn;
|
2012-10-21 21:59:24 +07:00
|
|
|
int cqe_factor;
|
2014-09-18 15:51:01 +07:00
|
|
|
int cqe_size;
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
struct mlx4_en_rss_map rss_map;
|
2012-03-06 11:03:41 +07:00
|
|
|
__be32 ctrl_flags;
|
2008-10-23 05:47:49 +07:00
|
|
|
u32 flags;
|
2012-12-02 10:49:23 +07:00
|
|
|
u8 num_tx_rings_p_up;
|
2014-07-08 15:28:12 +07:00
|
|
|
u32 tx_work_limit;
|
2016-11-02 22:12:24 +07:00
|
|
|
u32 tx_ring_num[MLX4_EN_NUM_TX_TYPES];
|
2008-10-23 05:47:49 +07:00
|
|
|
u32 rx_ring_num;
|
|
|
|
u32 rx_skb_size;
|
|
|
|
struct mlx4_en_frag_info frag_info[MLX4_EN_MAX_RX_FRAGS];
|
2017-03-08 23:17:06 +07:00
|
|
|
u8 num_frags;
|
|
|
|
u8 log_rx_info;
|
|
|
|
u8 dma_dir;
|
2017-03-08 23:17:09 +07:00
|
|
|
u16 rx_headroom;
|
2008-10-23 05:47:49 +07:00
|
|
|
|
2016-11-02 22:12:24 +07:00
|
|
|
struct mlx4_en_tx_ring **tx_ring[MLX4_EN_NUM_TX_TYPES];
|
2013-11-07 17:19:52 +07:00
|
|
|
struct mlx4_en_rx_ring *rx_ring[MAX_RX_RINGS];
|
2016-11-02 22:12:24 +07:00
|
|
|
struct mlx4_en_cq **tx_cq[MLX4_EN_NUM_TX_TYPES];
|
2013-11-07 17:19:52 +07:00
|
|
|
struct mlx4_en_cq *rx_cq[MAX_RX_RINGS];
|
2012-07-05 11:03:50 +07:00
|
|
|
struct mlx4_qp drop_qp;
|
2013-02-07 09:25:23 +07:00
|
|
|
struct work_struct rx_mode_task;
|
2008-10-23 05:47:49 +07:00
|
|
|
struct work_struct watchdog_task;
|
|
|
|
struct work_struct linkstate_task;
|
|
|
|
struct delayed_work stats_task;
|
2013-04-23 13:06:51 +07:00
|
|
|
struct delayed_work service_task;
|
2014-03-27 19:02:04 +07:00
|
|
|
struct work_struct vxlan_add_task;
|
|
|
|
struct work_struct vxlan_del_task;
|
2008-10-23 05:47:49 +07:00
|
|
|
struct mlx4_en_perf_stats pstats;
|
|
|
|
struct mlx4_en_pkt_stats pkstats;
|
2015-06-15 21:59:06 +07:00
|
|
|
struct mlx4_en_counter_stats pf_stats;
|
2015-03-30 21:45:25 +07:00
|
|
|
struct mlx4_en_flow_stats_rx rx_priority_flowstats[MLX4_NUM_PRIORITIES];
|
|
|
|
struct mlx4_en_flow_stats_tx tx_priority_flowstats[MLX4_NUM_PRIORITIES];
|
|
|
|
struct mlx4_en_flow_stats_rx rx_flowstats;
|
|
|
|
struct mlx4_en_flow_stats_tx tx_flowstats;
|
2008-10-23 05:47:49 +07:00
|
|
|
struct mlx4_en_port_stats port_stats;
|
2016-11-02 22:12:25 +07:00
|
|
|
struct mlx4_en_xdp_stats xdp_stats;
|
2018-02-27 21:17:19 +07:00
|
|
|
struct mlx4_en_phy_stats phy_stats;
|
2015-03-30 21:45:24 +07:00
|
|
|
struct mlx4_en_stats_bitmap stats_bitmap;
|
2012-07-05 11:03:43 +07:00
|
|
|
struct list_head mc_list;
|
|
|
|
struct list_head curr_list;
|
{NET, IB}/mlx4: Add device managed flow steering firmware API
The driver is modified to support three operation modes.
If supported by firmware use the device managed flow steering
API, that which we call device managed steering mode. Else, if
the firmware supports the B0 steering mode use it, and finally,
if none of the above, use the A0 steering mode.
When the steering mode is device managed, the code is modified
such that L2 based rules set by the mlx4_en driver for Ethernet
unicast and multicast, and the IB stack multicast attach calls
done through the mlx4_ib driver are all routed to use the device
managed API.
When attaching rule using device managed flow steering API,
the firmware returns a 64 bit registration id, which is to be
provided during detach.
Currently the firmware is always programmed during HCA initialization
to use standard L2 hashing. Future work should be done to allow
configuring the flow-steering hash function with common, non
proprietary means.
Signed-off-by: Hadar Hen Zion <hadarh@mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2012-07-05 11:03:46 +07:00
|
|
|
u64 broadcast_id;
|
2008-10-23 05:47:49 +07:00
|
|
|
struct mlx4_en_stat_out_mbox hw_stats;
|
2010-08-26 21:19:22 +07:00
|
|
|
int vids[128];
|
2011-03-23 05:37:59 +07:00
|
|
|
bool wol;
|
2012-03-06 11:03:34 +07:00
|
|
|
struct device *ddev;
|
2013-02-07 09:25:25 +07:00
|
|
|
struct hlist_head mac_hash[MLX4_EN_MAC_HASH_SIZE];
|
2013-04-23 13:06:49 +07:00
|
|
|
struct hwtstamp_config hwtstamp_config;
|
2015-06-15 21:59:02 +07:00
|
|
|
u32 counter_index;
|
2012-04-05 04:33:26 +07:00
|
|
|
|
|
|
|
#ifdef CONFIG_MLX4_EN_DCB
|
2016-06-21 16:43:59 +07:00
|
|
|
#define MLX4_EN_DCB_ENABLED 0x3
|
2012-04-05 04:33:26 +07:00
|
|
|
struct ieee_ets ets;
|
2012-04-05 04:33:31 +07:00
|
|
|
u16 maxrate[IEEE_8021QAZ_MAX_TCS];
|
2015-03-06 01:16:13 +07:00
|
|
|
enum dcbnl_cndd_states cndd_state[IEEE_8021QAZ_MAX_TCS];
|
2016-09-11 14:56:19 +07:00
|
|
|
struct mlx4_en_cee_config cee_config;
|
|
|
|
u8 dcbx_cap;
|
2012-04-05 04:33:26 +07:00
|
|
|
#endif
|
2012-07-19 05:33:52 +07:00
|
|
|
#ifdef CONFIG_RFS_ACCEL
|
|
|
|
spinlock_t filters_lock;
|
|
|
|
int last_filter_id;
|
|
|
|
struct list_head filters;
|
|
|
|
struct hlist_head filter_hash[1 << MLX4_EN_FILTER_HASH_SHIFT];
|
|
|
|
#endif
|
2013-12-23 21:09:44 +07:00
|
|
|
u64 tunnel_reg_id;
|
2014-03-27 19:02:04 +07:00
|
|
|
__be16 vxlan_port;
|
2014-07-22 19:44:10 +07:00
|
|
|
|
|
|
|
u32 pflags;
|
2014-11-23 08:24:19 +07:00
|
|
|
u8 rss_key[MLX4_EN_RSS_KEY_SIZE];
|
2014-12-02 23:12:11 +07:00
|
|
|
u8 rss_hash_fn;
|
2011-03-23 05:37:59 +07:00
|
|
|
};
|
|
|
|
|
|
|
|
enum mlx4_en_wol {
|
|
|
|
MLX4_EN_WOL_MAGIC = (1ULL << 61),
|
|
|
|
MLX4_EN_WOL_ENABLED = (1ULL << 62),
|
2008-10-23 05:47:49 +07:00
|
|
|
};
|
|
|
|
|
2013-02-07 09:25:22 +07:00
|
|
|
struct mlx4_mac_entry {
|
2013-02-07 09:25:25 +07:00
|
|
|
struct hlist_node hlist;
|
2013-02-07 09:25:22 +07:00
|
|
|
unsigned char mac[ETH_ALEN + 2];
|
|
|
|
u64 reg_id;
|
2013-02-07 09:25:25 +07:00
|
|
|
struct rcu_head rcu;
|
2013-02-07 09:25:22 +07:00
|
|
|
};
|
|
|
|
|
2014-09-18 15:51:01 +07:00
|
|
|
static inline struct mlx4_cqe *mlx4_en_get_cqe(void *buf, int idx, int cqe_sz)
|
|
|
|
{
|
|
|
|
return buf + idx * cqe_sz;
|
|
|
|
}
|
|
|
|
|
2011-11-27 02:55:06 +07:00
|
|
|
#define MLX4_EN_WOL_DO_MODIFY (1ULL << 63)
|
2008-10-23 05:47:49 +07:00
|
|
|
|
2016-02-25 01:58:12 +07:00
|
|
|
void mlx4_en_init_ptys2ethtool_map(void);
|
2013-02-07 09:25:19 +07:00
|
|
|
void mlx4_en_update_loopback_state(struct net_device *dev,
|
|
|
|
netdev_features_t features);
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
void mlx4_en_destroy_netdev(struct net_device *dev);
|
|
|
|
int mlx4_en_init_netdev(struct mlx4_en_dev *mdev, int port,
|
|
|
|
struct mlx4_en_port_profile *prof);
|
|
|
|
|
2008-12-30 09:39:20 +07:00
|
|
|
int mlx4_en_start_port(struct net_device *dev);
|
2013-01-31 06:07:11 +07:00
|
|
|
void mlx4_en_stop_port(struct net_device *dev, int detach);
|
2008-12-30 09:39:20 +07:00
|
|
|
|
2015-03-30 21:45:23 +07:00
|
|
|
void mlx4_en_set_stats_bitmap(struct mlx4_dev *dev,
|
2015-03-30 21:45:25 +07:00
|
|
|
struct mlx4_en_stats_bitmap *stats_bitmap,
|
|
|
|
u8 rx_ppp, u8 rx_pause,
|
|
|
|
u8 tx_ppp, u8 tx_pause);
|
2015-03-30 21:45:22 +07:00
|
|
|
|
2016-07-18 22:35:12 +07:00
|
|
|
int mlx4_en_try_alloc_resources(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_priv *tmp,
|
2017-02-01 13:35:33 +07:00
|
|
|
struct mlx4_en_port_profile *prof,
|
|
|
|
bool carry_xdp_prog);
|
2016-07-18 22:35:12 +07:00
|
|
|
void mlx4_en_safe_replace_resources(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_priv *tmp);
|
2008-12-30 09:39:20 +07:00
|
|
|
|
2013-11-07 17:19:52 +07:00
|
|
|
int mlx4_en_create_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq,
|
2013-11-07 17:19:54 +07:00
|
|
|
int entries, int ring, enum cq_type mode, int node);
|
2013-11-07 17:19:52 +07:00
|
|
|
void mlx4_en_destroy_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq **pcq);
|
2011-10-09 12:26:31 +07:00
|
|
|
int mlx4_en_activate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq,
|
|
|
|
int cq_idx);
|
2008-10-23 05:47:49 +07:00
|
|
|
void mlx4_en_deactivate_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
|
|
|
|
int mlx4_en_set_cq_moder(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
|
2017-07-24 10:02:53 +07:00
|
|
|
void mlx4_en_arm_cq(struct mlx4_en_priv *priv, struct mlx4_en_cq *cq);
|
2008-10-23 05:47:49 +07:00
|
|
|
|
|
|
|
void mlx4_en_tx_irq(struct mlx4_cq *mcq);
|
2014-01-10 15:18:26 +07:00
|
|
|
u16 mlx4_en_select_queue(struct net_device *dev, struct sk_buff *skb,
|
2014-02-16 21:55:20 +07:00
|
|
|
void *accel_priv, select_queue_fallback_t fallback);
|
2009-09-01 02:50:58 +07:00
|
|
|
netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev);
|
2016-11-02 22:12:25 +07:00
|
|
|
netdev_tx_t mlx4_en_xmit_frame(struct mlx4_en_rx_ring *rx_ring,
|
|
|
|
struct mlx4_en_rx_alloc *frame,
|
2017-10-11 17:17:25 +07:00
|
|
|
struct mlx4_en_priv *priv, unsigned int length,
|
2017-06-15 18:35:36 +07:00
|
|
|
int tx_ind, bool *doorbell_pending);
|
2016-07-20 02:16:55 +07:00
|
|
|
void mlx4_en_xmit_doorbell(struct mlx4_en_tx_ring *ring);
|
|
|
|
bool mlx4_en_rx_recycle(struct mlx4_en_rx_ring *ring,
|
|
|
|
struct mlx4_en_rx_alloc *frame);
|
2008-10-23 05:47:49 +07:00
|
|
|
|
2013-11-07 17:19:52 +07:00
|
|
|
int mlx4_en_create_tx_ring(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_tx_ring **pring,
|
net/mlx4: Change QP allocation scheme
When using BF (Blue-Flame), the QPN overrides the VLAN, CV, and SV fields
in the WQE. Thus, BF may only be used for QPNs with bits 6,7 unset.
The current Ethernet driver code reserves a Tx QP range with 256b alignment.
This is wrong because if there are more than 64 Tx QPs in use,
QPNs >= base + 65 will have bits 6/7 set.
This problem is not specific for the Ethernet driver, any entity that
tries to reserve more than 64 BF-enabled QPs should fail. Also, using
ranges is not necessary here and is wasteful.
The new mechanism introduced here will support reservation for
"Eth QPs eligible for BF" for all drivers: bare-metal, multi-PF, and VFs
(when hypervisors support WC in VMs). The flow we use is:
1. In mlx4_en, allocate Tx QPs one by one instead of a range allocation,
and request "BF enabled QPs" if BF is supported for the function
2. In the ALLOC_RES FW command, change param1 to:
a. param1[23:0] - number of QPs
b. param1[31-24] - flags controlling QPs reservation
Bit 31 refers to Eth blueflame supported QPs. Those QPs must have
bits 6 and 7 unset in order to be used in Ethernet.
Bits 24-30 of the flags are currently reserved.
When a function tries to allocate a QP, it states the required attributes
for this QP. Those attributes are considered "best-effort". If an attribute,
such as Ethernet BF enabled QP, is a must-have attribute, the function has
to check that attribute is supported before trying to do the allocation.
In a lower layer of the code, mlx4_qp_reserve_range masks out the bits
which are unsupported. If SRIOV is used, the PF validates those attributes
and masks out unsupported attributes as well. In order to notify VFs which
attributes are supported, the VF uses QUERY_FUNC_CAP command. This command's
mailbox is filled by the PF, which notifies which QP allocation attributes
it supports.
Signed-off-by: Eugenia Emantayev <eugenia@mellanox.co.il>
Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2014-12-11 15:57:54 +07:00
|
|
|
u32 size, u16 stride,
|
2013-12-20 02:20:14 +07:00
|
|
|
int node, int queue_index);
|
2013-11-07 17:19:52 +07:00
|
|
|
void mlx4_en_destroy_tx_ring(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_tx_ring **pring);
|
2017-10-11 17:17:27 +07:00
|
|
|
void mlx4_en_init_tx_xdp_ring_descs(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_tx_ring *ring);
|
2008-10-23 05:47:49 +07:00
|
|
|
int mlx4_en_activate_tx_ring(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_tx_ring *ring,
|
2012-04-05 04:33:24 +07:00
|
|
|
int cq, int user_prio);
|
2008-10-23 05:47:49 +07:00
|
|
|
void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_tx_ring *ring);
|
2014-02-21 17:39:17 +07:00
|
|
|
void mlx4_en_set_num_rx_rings(struct mlx4_en_dev *mdev);
|
2015-04-30 21:32:46 +07:00
|
|
|
void mlx4_en_recover_from_oom(struct mlx4_en_priv *priv);
|
2008-10-23 05:47:49 +07:00
|
|
|
int mlx4_en_create_rx_ring(struct mlx4_en_priv *priv,
|
2013-11-07 17:19:52 +07:00
|
|
|
struct mlx4_en_rx_ring **pring,
|
2018-01-03 17:25:39 +07:00
|
|
|
u32 size, u16 stride, int node, int queue_index);
|
2008-10-23 05:47:49 +07:00
|
|
|
void mlx4_en_destroy_rx_ring(struct mlx4_en_priv *priv,
|
2013-11-07 17:19:52 +07:00
|
|
|
struct mlx4_en_rx_ring **pring,
|
2012-02-06 15:39:49 +07:00
|
|
|
u32 size, u16 stride);
|
2008-10-23 05:47:49 +07:00
|
|
|
int mlx4_en_activate_rx_rings(struct mlx4_en_priv *priv);
|
|
|
|
void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_rx_ring *ring);
|
|
|
|
int mlx4_en_process_rx_cq(struct net_device *dev,
|
|
|
|
struct mlx4_en_cq *cq,
|
|
|
|
int budget);
|
|
|
|
int mlx4_en_poll_rx_cq(struct napi_struct *napi, int budget);
|
2013-12-20 02:20:17 +07:00
|
|
|
int mlx4_en_poll_tx_cq(struct napi_struct *napi, int budget);
|
2017-06-15 18:35:37 +07:00
|
|
|
bool mlx4_en_process_tx_cq(struct net_device *dev,
|
|
|
|
struct mlx4_en_cq *cq, int napi_budget);
|
2016-07-20 02:16:55 +07:00
|
|
|
u32 mlx4_en_free_tx_desc(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_tx_ring *ring,
|
2017-06-15 18:35:31 +07:00
|
|
|
int index, u64 timestamp,
|
2016-07-20 02:16:55 +07:00
|
|
|
int napi_mode);
|
|
|
|
u32 mlx4_en_recycle_tx_desc(struct mlx4_en_priv *priv,
|
|
|
|
struct mlx4_en_tx_ring *ring,
|
2017-06-15 18:35:31 +07:00
|
|
|
int index, u64 timestamp,
|
2016-07-20 02:16:55 +07:00
|
|
|
int napi_mode);
|
2008-10-23 05:47:49 +07:00
|
|
|
void mlx4_en_fill_qp_context(struct mlx4_en_priv *priv, int size, int stride,
|
2012-04-05 04:33:24 +07:00
|
|
|
int is_tx, int rss, int qpn, int cqn, int user_prio,
|
|
|
|
struct mlx4_qp_context *context);
|
2009-04-20 11:30:03 +07:00
|
|
|
void mlx4_en_sqp_event(struct mlx4_qp *qp, enum mlx4_event event);
|
2015-10-15 18:44:39 +07:00
|
|
|
int mlx4_en_change_mcast_lb(struct mlx4_en_priv *priv, struct mlx4_qp *qp,
|
|
|
|
int loopback);
|
2008-10-23 05:47:49 +07:00
|
|
|
void mlx4_en_calc_rx_buf(struct net_device *dev);
|
|
|
|
int mlx4_en_config_rss_steer(struct mlx4_en_priv *priv);
|
|
|
|
void mlx4_en_release_rss_steer(struct mlx4_en_priv *priv);
|
2012-07-05 11:03:50 +07:00
|
|
|
int mlx4_en_create_drop_qp(struct mlx4_en_priv *priv);
|
|
|
|
void mlx4_en_destroy_drop_qp(struct mlx4_en_priv *priv);
|
2008-10-23 05:47:49 +07:00
|
|
|
int mlx4_en_free_tx_buf(struct net_device *dev, struct mlx4_en_tx_ring *ring);
|
|
|
|
void mlx4_en_rx_irq(struct mlx4_cq *mcq);
|
|
|
|
|
|
|
|
int mlx4_SET_MCAST_FLTR(struct mlx4_dev *dev, u8 port, u64 mac, u64 clear, u8 mode);
|
2011-07-20 11:54:22 +07:00
|
|
|
int mlx4_SET_VLAN_FLTR(struct mlx4_dev *dev, struct mlx4_en_priv *priv);
|
2008-10-23 05:47:49 +07:00
|
|
|
|
2016-11-25 22:46:20 +07:00
|
|
|
void mlx4_en_fold_software_stats(struct net_device *dev);
|
2008-10-23 05:47:49 +07:00
|
|
|
int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev *mdev, u8 port, u8 reset);
|
2010-08-24 10:46:18 +07:00
|
|
|
int mlx4_en_QUERY_PORT(struct mlx4_en_dev *mdev, u8 port);
|
|
|
|
|
2012-04-05 04:33:26 +07:00
|
|
|
#ifdef CONFIG_MLX4_EN_DCB
|
|
|
|
extern const struct dcbnl_rtnl_ops mlx4_en_dcbnl_ops;
|
2013-04-07 10:44:07 +07:00
|
|
|
extern const struct dcbnl_rtnl_ops mlx4_en_dcbnl_pfc_ops;
|
2012-04-05 04:33:26 +07:00
|
|
|
#endif
|
|
|
|
|
2012-12-02 10:49:23 +07:00
|
|
|
int mlx4_en_setup_tc(struct net_device *dev, u8 up);
|
2017-06-29 18:07:57 +07:00
|
|
|
int mlx4_en_alloc_tx_queue_per_tc(struct net_device *dev, u8 tc);
|
2012-12-02 10:49:23 +07:00
|
|
|
|
2012-07-19 05:33:52 +07:00
|
|
|
#ifdef CONFIG_RFS_ACCEL
|
2013-11-07 17:19:52 +07:00
|
|
|
void mlx4_en_cleanup_filters(struct mlx4_en_priv *priv);
|
2012-07-19 05:33:52 +07:00
|
|
|
#endif
|
|
|
|
|
2010-08-24 10:46:18 +07:00
|
|
|
#define MLX4_EN_NUM_SELF_TEST 5
|
|
|
|
void mlx4_en_ex_selftest(struct net_device *dev, u32 *flags, u64 *buf);
|
2013-04-23 13:06:51 +07:00
|
|
|
void mlx4_en_ptp_overflow_check(struct mlx4_en_dev *mdev);
|
2008-10-23 05:47:49 +07:00
|
|
|
|
2014-10-27 16:37:42 +07:00
|
|
|
#define DEV_FEATURE_CHANGED(dev, new_features, feature) \
|
|
|
|
((dev->features & feature) ^ (new_features & feature))
|
|
|
|
|
|
|
|
int mlx4_en_reset_config(struct net_device *dev,
|
|
|
|
struct hwtstamp_config ts_config,
|
|
|
|
netdev_features_t new_features);
|
2015-03-30 21:45:25 +07:00
|
|
|
void mlx4_en_update_pfc_stats_bitmap(struct mlx4_dev *dev,
|
|
|
|
struct mlx4_en_stats_bitmap *stats_bitmap,
|
|
|
|
u8 rx_ppp, u8 rx_pause,
|
|
|
|
u8 tx_ppp, u8 tx_pause);
|
2015-02-03 21:48:34 +07:00
|
|
|
int mlx4_en_netdev_event(struct notifier_block *this,
|
|
|
|
unsigned long event, void *ptr);
|
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
/*
|
2013-04-23 13:06:49 +07:00
|
|
|
* Functions for time stamping
|
|
|
|
*/
|
|
|
|
u64 mlx4_en_get_cqe_ts(struct mlx4_cqe *cqe);
|
|
|
|
void mlx4_en_fill_hwtstamps(struct mlx4_en_dev *mdev,
|
|
|
|
struct skb_shared_hwtstamps *hwts,
|
|
|
|
u64 timestamp);
|
|
|
|
void mlx4_en_init_timestamp(struct mlx4_en_dev *mdev);
|
2014-01-01 00:39:39 +07:00
|
|
|
void mlx4_en_remove_timestamp(struct mlx4_en_dev *mdev);
|
2013-04-23 13:06:49 +07:00
|
|
|
|
|
|
|
/* Globals
|
2008-10-23 05:47:49 +07:00
|
|
|
*/
|
|
|
|
extern const struct ethtool_ops mlx4_en_ethtool_ops;
|
2010-07-10 14:22:46 +07:00
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
* printk / logging functions
|
|
|
|
*/
|
|
|
|
|
2011-11-01 07:11:33 +07:00
|
|
|
__printf(3, 4)
|
2014-09-23 00:40:22 +07:00
|
|
|
void en_print(const char *level, const struct mlx4_en_priv *priv,
|
|
|
|
const char *format, ...);
|
2010-07-10 14:22:46 +07:00
|
|
|
|
2014-05-08 02:52:57 +07:00
|
|
|
#define en_dbg(mlevel, priv, format, ...) \
|
|
|
|
do { \
|
|
|
|
if (NETIF_MSG_##mlevel & (priv)->msg_enable) \
|
|
|
|
en_print(KERN_DEBUG, priv, format, ##__VA_ARGS__); \
|
2010-07-10 14:22:46 +07:00
|
|
|
} while (0)
|
2014-05-08 02:52:57 +07:00
|
|
|
#define en_warn(priv, format, ...) \
|
|
|
|
en_print(KERN_WARNING, priv, format, ##__VA_ARGS__)
|
|
|
|
#define en_err(priv, format, ...) \
|
|
|
|
en_print(KERN_ERR, priv, format, ##__VA_ARGS__)
|
|
|
|
#define en_info(priv, format, ...) \
|
|
|
|
en_print(KERN_INFO, priv, format, ##__VA_ARGS__)
|
|
|
|
|
|
|
|
#define mlx4_err(mdev, format, ...) \
|
|
|
|
pr_err(DRV_NAME " %s: " format, \
|
|
|
|
dev_name(&(mdev)->pdev->dev), ##__VA_ARGS__)
|
|
|
|
#define mlx4_info(mdev, format, ...) \
|
|
|
|
pr_info(DRV_NAME " %s: " format, \
|
|
|
|
dev_name(&(mdev)->pdev->dev), ##__VA_ARGS__)
|
|
|
|
#define mlx4_warn(mdev, format, ...) \
|
|
|
|
pr_warn(DRV_NAME " %s: " format, \
|
|
|
|
dev_name(&(mdev)->pdev->dev), ##__VA_ARGS__)
|
2010-07-10 14:22:46 +07:00
|
|
|
|
2008-10-23 05:47:49 +07:00
|
|
|
#endif
|