2 * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
4 * This software is available to you under a choice of one of two
5 * licenses. You may choose to be licensed under the terms of the GNU
6 * General Public License (GPL) Version 2, available from the file
7 * COPYING in the main directory of this source tree, or the
8 * OpenIB.org BSD license below:
10 * Redistribution and use in source and binary forms, with or
11 * without modification, are permitted provided that the following
14 * - Redistributions of source code must retain the above
15 * copyright notice, this list of conditions and the following
18 * - Redistributions in binary form must reproduce the above
19 * copyright notice, this list of conditions and the following
20 * disclaimer in the documentation and/or other materials
21 * provided with the distribution.
23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
37 #include <linux/compiler.h>
38 #include <linux/list.h>
39 #include <linux/mutex.h>
40 #include <linux/netdevice.h>
41 #include <linux/inet_lro.h>
43 #include <linux/mlx4/device.h>
44 #include <linux/mlx4/qp.h>
45 #include <linux/mlx4/cq.h>
46 #include <linux/mlx4/srq.h>
47 #include <linux/mlx4/doorbell.h>
51 #define DRV_NAME "mlx4_en"
52 #define DRV_VERSION "1.4.1.1"
53 #define DRV_RELDATE "June 2009"
55 #define MLX4_EN_MSG_LEVEL (NETIF_MSG_LINK | NETIF_MSG_IFDOWN)
62 #define MLX4_EN_PAGE_SHIFT 12
63 #define MLX4_EN_PAGE_SIZE (1 << MLX4_EN_PAGE_SHIFT)
64 #define MAX_TX_RINGS 16
65 #define MAX_RX_RINGS 16
67 #define HEADROOM (2048 / TXBB_SIZE + 1)
68 #define STAMP_STRIDE 64
69 #define STAMP_DWORDS (STAMP_STRIDE / 4)
70 #define STAMP_SHIFT 31
71 #define STAMP_VAL 0x7fffffff
72 #define STATS_DELAY (HZ / 4)
74 /* Typical TSO descriptor with 16 gather entries is 352 bytes... */
75 #define MAX_DESC_SIZE 512
76 #define MAX_DESC_TXBBS (MAX_DESC_SIZE / TXBB_SIZE)
79 * OS related constants and tunables
82 #define MLX4_EN_WATCHDOG_TIMEOUT (15 * HZ)
84 #define MLX4_EN_ALLOC_ORDER 2
85 #define MLX4_EN_ALLOC_SIZE (PAGE_SIZE << MLX4_EN_ALLOC_ORDER)
87 #define MLX4_EN_MAX_LRO_DESCRIPTORS 32
89 /* Receive fragment sizes; we use at most 4 fragments (for 9600 byte MTU
90 * and 4K allocations) */
92 FRAG_SZ0
= 512 - NET_IP_ALIGN
,
95 FRAG_SZ3
= MLX4_EN_ALLOC_SIZE
97 #define MLX4_EN_MAX_RX_FRAGS 4
99 /* Maximum ring sizes */
100 #define MLX4_EN_MAX_TX_SIZE 8192
101 #define MLX4_EN_MAX_RX_SIZE 8192
103 /* Minimum ring size for our page-allocation sceme to work */
104 #define MLX4_EN_MIN_RX_SIZE (MLX4_EN_ALLOC_SIZE / SMP_CACHE_BYTES)
105 #define MLX4_EN_MIN_TX_SIZE (4096 / TXBB_SIZE)
107 #define MLX4_EN_SMALL_PKT_SIZE 64
108 #define MLX4_EN_NUM_TX_RINGS 8
109 #define MLX4_EN_NUM_PPP_RINGS 8
110 #define MLX4_EN_DEF_TX_RING_SIZE 512
111 #define MLX4_EN_DEF_RX_RING_SIZE 1024
113 /* Target number of packets to coalesce with interrupt moderation */
114 #define MLX4_EN_RX_COAL_TARGET 44
115 #define MLX4_EN_RX_COAL_TIME 0x10
117 #define MLX4_EN_TX_COAL_PKTS 5
118 #define MLX4_EN_TX_COAL_TIME 0x80
120 #define MLX4_EN_RX_RATE_LOW 400000
121 #define MLX4_EN_RX_COAL_TIME_LOW 0
122 #define MLX4_EN_RX_RATE_HIGH 450000
123 #define MLX4_EN_RX_COAL_TIME_HIGH 128
124 #define MLX4_EN_RX_SIZE_THRESH 1024
125 #define MLX4_EN_RX_RATE_THRESH (1000000 / MLX4_EN_RX_COAL_TIME_HIGH)
126 #define MLX4_EN_SAMPLE_INTERVAL 0
128 #define MLX4_EN_AUTO_CONF 0xffff
130 #define MLX4_EN_DEF_RX_PAUSE 1
131 #define MLX4_EN_DEF_TX_PAUSE 1
133 /* Interval between successive polls in the Tx routine when polling is used
134 instead of interrupts (in per-core Tx rings) - should be power of 2 */
135 #define MLX4_EN_TX_POLL_MODER 16
136 #define MLX4_EN_TX_POLL_TIMEOUT (HZ / 4)
138 #define ETH_LLC_SNAP_SIZE 8
140 #define SMALL_PACKET_SIZE (256 - NET_IP_ALIGN)
141 #define HEADER_COPY_SIZE (128 - NET_IP_ALIGN)
143 #define MLX4_EN_MIN_MTU 46
144 #define ETH_BCAST 0xffffffffffffULL
146 #ifdef MLX4_EN_PERF_STAT
147 /* Number of samples to 'average' */
149 #define AVG_FACTOR 1024
150 #define NUM_PERF_STATS NUM_PERF_COUNTERS
152 #define INC_PERF_COUNTER(cnt) (++(cnt))
153 #define ADD_PERF_COUNTER(cnt, add) ((cnt) += (add))
154 #define AVG_PERF_COUNTER(cnt, sample) \
155 ((cnt) = ((cnt) * (AVG_SIZE - 1) + (sample) * AVG_FACTOR) / AVG_SIZE)
156 #define GET_PERF_COUNTER(cnt) (cnt)
157 #define GET_AVG_PERF_COUNTER(cnt) ((cnt) / AVG_FACTOR)
161 #define NUM_PERF_STATS 0
162 #define INC_PERF_COUNTER(cnt) do {} while (0)
163 #define ADD_PERF_COUNTER(cnt, add) do {} while (0)
164 #define AVG_PERF_COUNTER(cnt, sample) do {} while (0)
165 #define GET_PERF_COUNTER(cnt) (0)
166 #define GET_AVG_PERF_COUNTER(cnt) (0)
167 #endif /* MLX4_EN_PERF_STAT */
182 #define ROUNDUP_LOG2(x) ilog2(roundup_pow_of_two(x))
183 #define XNOR(x, y) (!(x) == !(y))
184 #define ILLEGAL_MAC(addr) (addr == 0xffffffffffffULL || addr == 0x0)
187 struct mlx4_en_tx_info
{
196 #define MLX4_EN_BIT_DESC_OWN 0x80000000
197 #define CTRL_SIZE sizeof(struct mlx4_wqe_ctrl_seg)
198 #define MLX4_EN_MEMTYPE_PAD 0x100
199 #define DS_SIZE sizeof(struct mlx4_wqe_data_seg)
202 struct mlx4_en_tx_desc
{
203 struct mlx4_wqe_ctrl_seg ctrl
;
205 struct mlx4_wqe_data_seg data
; /* at least one data segment */
206 struct mlx4_wqe_lso_seg lso
;
207 struct mlx4_wqe_inline_seg inl
;
211 #define MLX4_EN_USE_SRQ 0x01000000
213 struct mlx4_en_rx_alloc
{
218 struct mlx4_en_tx_ring
{
219 struct mlx4_hwq_resources wqres
;
220 u32 size
; /* number of TXBBs */
223 u16 cqn
; /* index of port CQ associated with this ring */
231 struct mlx4_en_tx_info
*tx_info
;
235 struct mlx4_qp_context context
;
237 enum mlx4_qp_state qp_state
;
238 struct mlx4_srq dummy
;
240 unsigned long packets
;
241 spinlock_t comp_lock
;
244 struct mlx4_en_rx_desc
{
245 /* actual number of entries depends on rx ring stride */
246 struct mlx4_wqe_data_seg data
[0];
249 struct mlx4_en_rx_ring
{
250 struct mlx4_hwq_resources wqres
;
251 struct mlx4_en_rx_alloc page_alloc
[MLX4_EN_MAX_RX_FRAGS
];
252 struct net_lro_mgr lro
;
253 u32 size
; /* number of Rx descs*/
258 u16 cqn
; /* index of port CQ associated with this ring */
265 unsigned long packets
;
269 static inline int mlx4_en_can_lro(__be16 status
)
271 return (status
& cpu_to_be16(MLX4_CQE_STATUS_IPV4
|
272 MLX4_CQE_STATUS_IPV4F
|
273 MLX4_CQE_STATUS_IPV6
|
274 MLX4_CQE_STATUS_IPV4OPT
|
275 MLX4_CQE_STATUS_TCP
|
276 MLX4_CQE_STATUS_UDP
|
277 MLX4_CQE_STATUS_IPOK
)) ==
278 cpu_to_be16(MLX4_CQE_STATUS_IPV4
|
279 MLX4_CQE_STATUS_IPOK
|
280 MLX4_CQE_STATUS_TCP
);
285 struct mlx4_hwq_resources wqres
;
288 struct net_device
*dev
;
289 struct napi_struct napi
;
290 /* Per-core Tx cq processing support */
291 struct timer_list timer
;
298 struct mlx4_cqe
*buf
;
299 #define MLX4_EN_OPCODE_ERROR 0x1e
302 struct mlx4_en_port_profile
{
314 struct mlx4_en_profile
{
321 struct mlx4_en_port_profile prof
[MLX4_MAX_PORTS
+ 1];
325 struct mlx4_dev
*dev
;
326 struct pci_dev
*pdev
;
327 struct mutex state_lock
;
328 struct net_device
*pndev
[MLX4_MAX_PORTS
+ 1];
331 struct mlx4_en_profile profile
;
333 struct workqueue_struct
*workqueue
;
334 struct device
*dma_device
;
335 void __iomem
*uar_map
;
336 struct mlx4_uar priv_uar
;
343 struct mlx4_en_rss_map
{
345 struct mlx4_qp qps
[MAX_RX_RINGS
];
346 enum mlx4_qp_state state
[MAX_RX_RINGS
];
347 struct mlx4_qp indir_qp
;
348 enum mlx4_qp_state indir_state
;
351 struct mlx4_en_rss_context
{
360 struct mlx4_en_pkt_stats
{
361 unsigned long broadcast
;
362 unsigned long rx_prio
[8];
363 unsigned long tx_prio
[8];
364 #define NUM_PKT_STATS 17
367 struct mlx4_en_port_stats
{
368 unsigned long lro_aggregated
;
369 unsigned long lro_flushed
;
370 unsigned long lro_no_desc
;
371 unsigned long tso_packets
;
372 unsigned long queue_stopped
;
373 unsigned long wake_queue
;
374 unsigned long tx_timeout
;
375 unsigned long rx_alloc_failed
;
376 unsigned long rx_chksum_good
;
377 unsigned long rx_chksum_none
;
378 unsigned long tx_chksum_offload
;
379 #define NUM_PORT_STATS 11
382 struct mlx4_en_perf_stats
{
389 #define NUM_PERF_COUNTERS 6
392 struct mlx4_en_frag_info
{
394 u16 frag_prefix_size
;
401 struct mlx4_en_priv
{
402 struct mlx4_en_dev
*mdev
;
403 struct mlx4_en_port_profile
*prof
;
404 struct net_device
*dev
;
405 struct vlan_group
*vlgrp
;
406 struct net_device_stats stats
;
407 struct net_device_stats ret_stats
;
408 spinlock_t stats_lock
;
410 unsigned long last_moder_packets
;
411 unsigned long last_moder_tx_packets
;
412 unsigned long last_moder_bytes
;
413 unsigned long last_moder_jiffies
;
424 u16 adaptive_rx_coal
;
427 struct mlx4_hwq_resources res
;
441 struct mlx4_en_rss_map rss_map
;
443 #define MLX4_EN_FLAG_PROMISC 0x1
447 struct mlx4_en_frag_info frag_info
[MLX4_EN_MAX_RX_FRAGS
];
451 struct mlx4_en_tx_ring tx_ring
[MAX_TX_RINGS
];
452 struct mlx4_en_rx_ring rx_ring
[MAX_RX_RINGS
];
453 struct mlx4_en_cq tx_cq
[MAX_TX_RINGS
];
454 struct mlx4_en_cq rx_cq
[MAX_RX_RINGS
];
455 struct work_struct mcast_task
;
456 struct work_struct mac_task
;
457 struct work_struct watchdog_task
;
458 struct work_struct linkstate_task
;
459 struct delayed_work stats_task
;
460 struct mlx4_en_perf_stats pstats
;
461 struct mlx4_en_pkt_stats pkstats
;
462 struct mlx4_en_port_stats port_stats
;
465 struct mlx4_en_stat_out_mbox hw_stats
;
469 void mlx4_en_destroy_netdev(struct net_device
*dev
);
470 int mlx4_en_init_netdev(struct mlx4_en_dev
*mdev
, int port
,
471 struct mlx4_en_port_profile
*prof
);
473 int mlx4_en_start_port(struct net_device
*dev
);
474 void mlx4_en_stop_port(struct net_device
*dev
);
476 void mlx4_en_free_resources(struct mlx4_en_priv
*priv
);
477 int mlx4_en_alloc_resources(struct mlx4_en_priv
*priv
);
479 int mlx4_en_create_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
,
480 int entries
, int ring
, enum cq_type mode
);
481 void mlx4_en_destroy_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
482 int mlx4_en_activate_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
483 void mlx4_en_deactivate_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
484 int mlx4_en_set_cq_moder(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
485 int mlx4_en_arm_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
487 void mlx4_en_poll_tx_cq(unsigned long data
);
488 void mlx4_en_tx_irq(struct mlx4_cq
*mcq
);
489 u16
mlx4_en_select_queue(struct net_device
*dev
, struct sk_buff
*skb
);
490 netdev_tx_t
mlx4_en_xmit(struct sk_buff
*skb
, struct net_device
*dev
);
492 int mlx4_en_create_tx_ring(struct mlx4_en_priv
*priv
, struct mlx4_en_tx_ring
*ring
,
493 u32 size
, u16 stride
);
494 void mlx4_en_destroy_tx_ring(struct mlx4_en_priv
*priv
, struct mlx4_en_tx_ring
*ring
);
495 int mlx4_en_activate_tx_ring(struct mlx4_en_priv
*priv
,
496 struct mlx4_en_tx_ring
*ring
,
498 void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv
*priv
,
499 struct mlx4_en_tx_ring
*ring
);
501 int mlx4_en_create_rx_ring(struct mlx4_en_priv
*priv
,
502 struct mlx4_en_rx_ring
*ring
,
503 u32 size
, u16 stride
);
504 void mlx4_en_destroy_rx_ring(struct mlx4_en_priv
*priv
,
505 struct mlx4_en_rx_ring
*ring
);
506 int mlx4_en_activate_rx_rings(struct mlx4_en_priv
*priv
);
507 void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv
*priv
,
508 struct mlx4_en_rx_ring
*ring
);
509 int mlx4_en_process_rx_cq(struct net_device
*dev
,
510 struct mlx4_en_cq
*cq
,
512 int mlx4_en_poll_rx_cq(struct napi_struct
*napi
, int budget
);
513 void mlx4_en_fill_qp_context(struct mlx4_en_priv
*priv
, int size
, int stride
,
514 int is_tx
, int rss
, int qpn
, int cqn
,
515 struct mlx4_qp_context
*context
);
516 void mlx4_en_sqp_event(struct mlx4_qp
*qp
, enum mlx4_event event
);
517 int mlx4_en_map_buffer(struct mlx4_buf
*buf
);
518 void mlx4_en_unmap_buffer(struct mlx4_buf
*buf
);
520 void mlx4_en_calc_rx_buf(struct net_device
*dev
);
521 int mlx4_en_config_rss_steer(struct mlx4_en_priv
*priv
);
522 void mlx4_en_release_rss_steer(struct mlx4_en_priv
*priv
);
523 int mlx4_en_free_tx_buf(struct net_device
*dev
, struct mlx4_en_tx_ring
*ring
);
524 void mlx4_en_rx_irq(struct mlx4_cq
*mcq
);
526 int mlx4_SET_MCAST_FLTR(struct mlx4_dev
*dev
, u8 port
, u64 mac
, u64 clear
, u8 mode
);
527 int mlx4_SET_VLAN_FLTR(struct mlx4_dev
*dev
, u8 port
, struct vlan_group
*grp
);
528 int mlx4_SET_PORT_general(struct mlx4_dev
*dev
, u8 port
, int mtu
,
529 u8 pptx
, u8 pfctx
, u8 pprx
, u8 pfcrx
);
530 int mlx4_SET_PORT_qpn_calc(struct mlx4_dev
*dev
, u8 port
, u32 base_qpn
,
533 int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev
*mdev
, u8 port
, u8 reset
);
538 extern const struct ethtool_ops mlx4_en_ethtool_ops
;
543 * printk / logging functions
546 int en_print(const char *level
, const struct mlx4_en_priv
*priv
,
547 const char *format
, ...) __attribute__ ((format (printf
, 3, 4)));
549 #define en_dbg(mlevel, priv, format, arg...) \
551 if (NETIF_MSG_##mlevel & priv->msg_enable) \
552 en_print(KERN_DEBUG, priv, format, ##arg); \
554 #define en_warn(priv, format, arg...) \
555 en_print(KERN_WARNING, priv, format, ##arg)
556 #define en_err(priv, format, arg...) \
557 en_print(KERN_ERR, priv, format, ##arg)
559 #define mlx4_err(mdev, format, arg...) \
560 pr_err("%s %s: " format, DRV_NAME, \
561 dev_name(&mdev->pdev->dev), ##arg)
562 #define mlx4_info(mdev, format, arg...) \
563 pr_info("%s %s: " format, DRV_NAME, \
564 dev_name(&mdev->pdev->dev), ##arg)
565 #define mlx4_warn(mdev, format, arg...) \
566 pr_warning("%s %s: " format, DRV_NAME, \
567 dev_name(&mdev->pdev->dev), ##arg)