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"
56 #define MLX4_EN_MSG_LEVEL (NETIF_MSG_LINK | NETIF_MSG_IFDOWN)
58 #define en_print(level, priv, format, arg...) \
60 if ((priv)->registered) \
61 printk(level "%s: %s: " format, DRV_NAME, \
62 (priv->dev)->name, ## arg); \
64 printk(level "%s: %s: Port %d: " format, \
65 DRV_NAME, dev_name(&priv->mdev->pdev->dev), \
66 (priv)->port, ## arg); \
69 #define en_dbg(mlevel, priv, format, arg...) \
71 if (NETIF_MSG_##mlevel & priv->msg_enable) \
72 en_print(KERN_DEBUG, priv, format, ## arg) \
74 #define en_warn(priv, format, arg...) \
75 en_print(KERN_WARNING, priv, format, ## arg)
76 #define en_err(priv, format, arg...) \
77 en_print(KERN_ERR, priv, format, ## arg)
79 #define mlx4_err(mdev, format, arg...) \
80 printk(KERN_ERR "%s %s: " format , DRV_NAME ,\
81 dev_name(&mdev->pdev->dev) , ## arg)
82 #define mlx4_info(mdev, format, arg...) \
83 printk(KERN_INFO "%s %s: " format , DRV_NAME ,\
84 dev_name(&mdev->pdev->dev) , ## arg)
85 #define mlx4_warn(mdev, format, arg...) \
86 printk(KERN_WARNING "%s %s: " format , DRV_NAME ,\
87 dev_name(&mdev->pdev->dev) , ## arg)
94 #define MLX4_EN_PAGE_SHIFT 12
95 #define MLX4_EN_PAGE_SIZE (1 << MLX4_EN_PAGE_SHIFT)
96 #define MAX_TX_RINGS 16
97 #define MAX_RX_RINGS 16
98 #define MAX_RSS_MAP_SIZE 64
101 #define HEADROOM (2048 / TXBB_SIZE + 1)
102 #define MAX_LSO_HDR_SIZE 92
103 #define STAMP_STRIDE 64
104 #define STAMP_DWORDS (STAMP_STRIDE / 4)
105 #define STAMP_SHIFT 31
106 #define STAMP_VAL 0x7fffffff
107 #define STATS_DELAY (HZ / 4)
109 /* Typical TSO descriptor with 16 gather entries is 352 bytes... */
110 #define MAX_DESC_SIZE 512
111 #define MAX_DESC_TXBBS (MAX_DESC_SIZE / TXBB_SIZE)
114 * OS related constants and tunables
117 #define MLX4_EN_WATCHDOG_TIMEOUT (15 * HZ)
119 #define MLX4_EN_ALLOC_ORDER 2
120 #define MLX4_EN_ALLOC_SIZE (PAGE_SIZE << MLX4_EN_ALLOC_ORDER)
122 #define MLX4_EN_MAX_LRO_DESCRIPTORS 32
124 /* Receive fragment sizes; we use at most 4 fragments (for 9600 byte MTU
125 * and 4K allocations) */
127 FRAG_SZ0
= 512 - NET_IP_ALIGN
,
130 FRAG_SZ3
= MLX4_EN_ALLOC_SIZE
132 #define MLX4_EN_MAX_RX_FRAGS 4
134 /* Maximum ring sizes */
135 #define MLX4_EN_MAX_TX_SIZE 8192
136 #define MLX4_EN_MAX_RX_SIZE 8192
138 /* Minimum ring size for our page-allocation sceme to work */
139 #define MLX4_EN_MIN_RX_SIZE (MLX4_EN_ALLOC_SIZE / SMP_CACHE_BYTES)
140 #define MLX4_EN_MIN_TX_SIZE (4096 / TXBB_SIZE)
142 #define MLX4_EN_SMALL_PKT_SIZE 64
143 #define MLX4_EN_NUM_TX_RINGS 8
144 #define MLX4_EN_NUM_PPP_RINGS 8
145 #define MLX4_EN_DEF_TX_RING_SIZE 512
146 #define MLX4_EN_DEF_RX_RING_SIZE 1024
148 /* Target number of packets to coalesce with interrupt moderation */
149 #define MLX4_EN_RX_COAL_TARGET 44
150 #define MLX4_EN_RX_COAL_TIME 0x10
152 #define MLX4_EN_TX_COAL_PKTS 5
153 #define MLX4_EN_TX_COAL_TIME 0x80
155 #define MLX4_EN_RX_RATE_LOW 400000
156 #define MLX4_EN_RX_COAL_TIME_LOW 0
157 #define MLX4_EN_RX_RATE_HIGH 450000
158 #define MLX4_EN_RX_COAL_TIME_HIGH 128
159 #define MLX4_EN_RX_SIZE_THRESH 1024
160 #define MLX4_EN_RX_RATE_THRESH (1000000 / MLX4_EN_RX_COAL_TIME_HIGH)
161 #define MLX4_EN_SAMPLE_INTERVAL 0
163 #define MLX4_EN_AUTO_CONF 0xffff
165 #define MLX4_EN_DEF_RX_PAUSE 1
166 #define MLX4_EN_DEF_TX_PAUSE 1
168 /* Interval between sucessive polls in the Tx routine when polling is used
169 instead of interrupts (in per-core Tx rings) - should be power of 2 */
170 #define MLX4_EN_TX_POLL_MODER 16
171 #define MLX4_EN_TX_POLL_TIMEOUT (HZ / 4)
173 #define ETH_LLC_SNAP_SIZE 8
175 #define SMALL_PACKET_SIZE (256 - NET_IP_ALIGN)
176 #define HEADER_COPY_SIZE (128 - NET_IP_ALIGN)
178 #define MLX4_EN_MIN_MTU 46
179 #define ETH_BCAST 0xffffffffffffULL
181 #ifdef MLX4_EN_PERF_STAT
182 /* Number of samples to 'average' */
184 #define AVG_FACTOR 1024
185 #define NUM_PERF_STATS NUM_PERF_COUNTERS
187 #define INC_PERF_COUNTER(cnt) (++(cnt))
188 #define ADD_PERF_COUNTER(cnt, add) ((cnt) += (add))
189 #define AVG_PERF_COUNTER(cnt, sample) \
190 ((cnt) = ((cnt) * (AVG_SIZE - 1) + (sample) * AVG_FACTOR) / AVG_SIZE)
191 #define GET_PERF_COUNTER(cnt) (cnt)
192 #define GET_AVG_PERF_COUNTER(cnt) ((cnt) / AVG_FACTOR)
196 #define NUM_PERF_STATS 0
197 #define INC_PERF_COUNTER(cnt) do {} while (0)
198 #define ADD_PERF_COUNTER(cnt, add) do {} while (0)
199 #define AVG_PERF_COUNTER(cnt, sample) do {} while (0)
200 #define GET_PERF_COUNTER(cnt) (0)
201 #define GET_AVG_PERF_COUNTER(cnt) (0)
202 #endif /* MLX4_EN_PERF_STAT */
217 #define ROUNDUP_LOG2(x) ilog2(roundup_pow_of_two(x))
218 #define XNOR(x, y) (!(x) == !(y))
219 #define ILLEGAL_MAC(addr) (addr == 0xffffffffffffULL || addr == 0x0)
222 struct mlx4_en_tx_info
{
231 #define MLX4_EN_BIT_DESC_OWN 0x80000000
232 #define CTRL_SIZE sizeof(struct mlx4_wqe_ctrl_seg)
233 #define MLX4_EN_MEMTYPE_PAD 0x100
234 #define DS_SIZE sizeof(struct mlx4_wqe_data_seg)
237 struct mlx4_en_tx_desc
{
238 struct mlx4_wqe_ctrl_seg ctrl
;
240 struct mlx4_wqe_data_seg data
; /* at least one data segment */
241 struct mlx4_wqe_lso_seg lso
;
242 struct mlx4_wqe_inline_seg inl
;
246 #define MLX4_EN_USE_SRQ 0x01000000
248 struct mlx4_en_rx_alloc
{
253 struct mlx4_en_tx_ring
{
254 struct mlx4_hwq_resources wqres
;
255 u32 size
; /* number of TXBBs */
258 u16 cqn
; /* index of port CQ associated with this ring */
266 struct mlx4_en_tx_info
*tx_info
;
270 struct mlx4_qp_context context
;
272 enum mlx4_qp_state qp_state
;
273 struct mlx4_srq dummy
;
275 unsigned long packets
;
276 spinlock_t comp_lock
;
279 struct mlx4_en_rx_desc
{
280 struct mlx4_wqe_srq_next_seg next
;
281 /* actual number of entries depends on rx ring stride */
282 struct mlx4_wqe_data_seg data
[0];
285 struct mlx4_en_rx_ring
{
287 struct mlx4_hwq_resources wqres
;
288 struct mlx4_en_rx_alloc page_alloc
[MLX4_EN_MAX_RX_FRAGS
];
289 struct net_lro_mgr lro
;
290 u32 size
; /* number of Rx descs*/
295 u16 cqn
; /* index of port CQ associated with this ring */
304 unsigned long packets
;
308 static inline int mlx4_en_can_lro(__be16 status
)
310 return (status
& cpu_to_be16(MLX4_CQE_STATUS_IPV4
|
311 MLX4_CQE_STATUS_IPV4F
|
312 MLX4_CQE_STATUS_IPV6
|
313 MLX4_CQE_STATUS_IPV4OPT
|
314 MLX4_CQE_STATUS_TCP
|
315 MLX4_CQE_STATUS_UDP
|
316 MLX4_CQE_STATUS_IPOK
)) ==
317 cpu_to_be16(MLX4_CQE_STATUS_IPV4
|
318 MLX4_CQE_STATUS_IPOK
|
319 MLX4_CQE_STATUS_TCP
);
324 struct mlx4_hwq_resources wqres
;
327 struct net_device
*dev
;
328 struct napi_struct napi
;
329 /* Per-core Tx cq processing support */
330 struct timer_list timer
;
337 struct mlx4_cqe
*buf
;
338 #define MLX4_EN_OPCODE_ERROR 0x1e
341 struct mlx4_en_port_profile
{
353 struct mlx4_en_profile
{
360 struct mlx4_en_port_profile prof
[MLX4_MAX_PORTS
+ 1];
364 struct mlx4_dev
*dev
;
365 struct pci_dev
*pdev
;
366 struct mutex state_lock
;
367 struct net_device
*pndev
[MLX4_MAX_PORTS
+ 1];
370 struct mlx4_en_profile profile
;
372 struct workqueue_struct
*workqueue
;
373 struct device
*dma_device
;
374 void __iomem
*uar_map
;
375 struct mlx4_uar priv_uar
;
382 struct mlx4_en_rss_map
{
385 u16 map
[MAX_RSS_MAP_SIZE
];
386 struct mlx4_qp qps
[MAX_RSS_MAP_SIZE
];
387 enum mlx4_qp_state state
[MAX_RSS_MAP_SIZE
];
388 struct mlx4_qp indir_qp
;
389 enum mlx4_qp_state indir_state
;
392 struct mlx4_en_rss_context
{
401 struct mlx4_en_pkt_stats
{
402 unsigned long broadcast
;
403 unsigned long rx_prio
[8];
404 unsigned long tx_prio
[8];
405 #define NUM_PKT_STATS 17
408 struct mlx4_en_port_stats
{
409 unsigned long lro_aggregated
;
410 unsigned long lro_flushed
;
411 unsigned long lro_no_desc
;
412 unsigned long tso_packets
;
413 unsigned long queue_stopped
;
414 unsigned long wake_queue
;
415 unsigned long tx_timeout
;
416 unsigned long rx_alloc_failed
;
417 unsigned long rx_chksum_good
;
418 unsigned long rx_chksum_none
;
419 unsigned long tx_chksum_offload
;
420 #define NUM_PORT_STATS 11
423 struct mlx4_en_perf_stats
{
430 #define NUM_PERF_COUNTERS 6
433 struct mlx4_en_frag_info
{
435 u16 frag_prefix_size
;
442 struct mlx4_en_priv
{
443 struct mlx4_en_dev
*mdev
;
444 struct mlx4_en_port_profile
*prof
;
445 struct net_device
*dev
;
446 struct vlan_group
*vlgrp
;
447 struct net_device_stats stats
;
448 struct net_device_stats ret_stats
;
449 spinlock_t stats_lock
;
451 unsigned long last_moder_packets
;
452 unsigned long last_moder_tx_packets
;
453 unsigned long last_moder_bytes
;
454 unsigned long last_moder_jiffies
;
465 u16 adaptive_rx_coal
;
468 struct mlx4_hwq_resources res
;
482 struct mlx4_en_rss_map rss_map
;
484 #define MLX4_EN_FLAG_PROMISC 0x1
488 struct mlx4_en_frag_info frag_info
[MLX4_EN_MAX_RX_FRAGS
];
492 struct mlx4_en_tx_ring tx_ring
[MAX_TX_RINGS
];
493 struct mlx4_en_rx_ring rx_ring
[MAX_RX_RINGS
];
494 struct mlx4_en_cq tx_cq
[MAX_TX_RINGS
];
495 struct mlx4_en_cq rx_cq
[MAX_RX_RINGS
];
496 struct work_struct mcast_task
;
497 struct work_struct mac_task
;
498 struct delayed_work refill_task
;
499 struct work_struct watchdog_task
;
500 struct work_struct linkstate_task
;
501 struct delayed_work stats_task
;
502 struct mlx4_en_perf_stats pstats
;
503 struct mlx4_en_pkt_stats pkstats
;
504 struct mlx4_en_port_stats port_stats
;
505 struct dev_mc_list
*mc_list
;
506 struct mlx4_en_stat_out_mbox hw_stats
;
510 void mlx4_en_destroy_netdev(struct net_device
*dev
);
511 int mlx4_en_init_netdev(struct mlx4_en_dev
*mdev
, int port
,
512 struct mlx4_en_port_profile
*prof
);
514 int mlx4_en_start_port(struct net_device
*dev
);
515 void mlx4_en_stop_port(struct net_device
*dev
);
517 void mlx4_en_free_resources(struct mlx4_en_priv
*priv
);
518 int mlx4_en_alloc_resources(struct mlx4_en_priv
*priv
);
520 int mlx4_en_create_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
,
521 int entries
, int ring
, enum cq_type mode
);
522 void mlx4_en_destroy_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
523 int mlx4_en_activate_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
524 void mlx4_en_deactivate_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
525 int mlx4_en_set_cq_moder(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
526 int mlx4_en_arm_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
528 void mlx4_en_poll_tx_cq(unsigned long data
);
529 void mlx4_en_tx_irq(struct mlx4_cq
*mcq
);
530 u16
mlx4_en_select_queue(struct net_device
*dev
, struct sk_buff
*skb
);
531 int mlx4_en_xmit(struct sk_buff
*skb
, struct net_device
*dev
);
533 int mlx4_en_create_tx_ring(struct mlx4_en_priv
*priv
, struct mlx4_en_tx_ring
*ring
,
534 u32 size
, u16 stride
);
535 void mlx4_en_destroy_tx_ring(struct mlx4_en_priv
*priv
, struct mlx4_en_tx_ring
*ring
);
536 int mlx4_en_activate_tx_ring(struct mlx4_en_priv
*priv
,
537 struct mlx4_en_tx_ring
*ring
,
539 void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv
*priv
,
540 struct mlx4_en_tx_ring
*ring
);
542 int mlx4_en_create_rx_ring(struct mlx4_en_priv
*priv
,
543 struct mlx4_en_rx_ring
*ring
,
544 u32 size
, u16 stride
);
545 void mlx4_en_destroy_rx_ring(struct mlx4_en_priv
*priv
,
546 struct mlx4_en_rx_ring
*ring
);
547 int mlx4_en_activate_rx_rings(struct mlx4_en_priv
*priv
);
548 void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv
*priv
,
549 struct mlx4_en_rx_ring
*ring
);
550 int mlx4_en_process_rx_cq(struct net_device
*dev
,
551 struct mlx4_en_cq
*cq
,
553 int mlx4_en_poll_rx_cq(struct napi_struct
*napi
, int budget
);
554 void mlx4_en_fill_qp_context(struct mlx4_en_priv
*priv
, int size
, int stride
,
555 int is_tx
, int rss
, int qpn
, int cqn
, int srqn
,
556 struct mlx4_qp_context
*context
);
557 void mlx4_en_sqp_event(struct mlx4_qp
*qp
, enum mlx4_event event
);
558 int mlx4_en_map_buffer(struct mlx4_buf
*buf
);
559 void mlx4_en_unmap_buffer(struct mlx4_buf
*buf
);
561 void mlx4_en_calc_rx_buf(struct net_device
*dev
);
562 void mlx4_en_set_default_rss_map(struct mlx4_en_priv
*priv
,
563 struct mlx4_en_rss_map
*rss_map
,
564 int num_entries
, int num_rings
);
565 int mlx4_en_config_rss_steer(struct mlx4_en_priv
*priv
);
566 void mlx4_en_release_rss_steer(struct mlx4_en_priv
*priv
);
567 int mlx4_en_free_tx_buf(struct net_device
*dev
, struct mlx4_en_tx_ring
*ring
);
568 void mlx4_en_rx_refill(struct work_struct
*work
);
569 void mlx4_en_rx_irq(struct mlx4_cq
*mcq
);
571 int mlx4_SET_MCAST_FLTR(struct mlx4_dev
*dev
, u8 port
, u64 mac
, u64 clear
, u8 mode
);
572 int mlx4_SET_VLAN_FLTR(struct mlx4_dev
*dev
, u8 port
, struct vlan_group
*grp
);
573 int mlx4_SET_PORT_general(struct mlx4_dev
*dev
, u8 port
, int mtu
,
574 u8 pptx
, u8 pfctx
, u8 pprx
, u8 pfcrx
);
575 int mlx4_SET_PORT_qpn_calc(struct mlx4_dev
*dev
, u8 port
, u32 base_qpn
,
578 int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev
*mdev
, u8 port
, u8 reset
);
583 extern const struct ethtool_ops mlx4_en_ethtool_ops
;