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
99 #define HEADROOM (2048 / TXBB_SIZE + 1)
100 #define STAMP_STRIDE 64
101 #define STAMP_DWORDS (STAMP_STRIDE / 4)
102 #define STAMP_SHIFT 31
103 #define STAMP_VAL 0x7fffffff
104 #define STATS_DELAY (HZ / 4)
106 /* Typical TSO descriptor with 16 gather entries is 352 bytes... */
107 #define MAX_DESC_SIZE 512
108 #define MAX_DESC_TXBBS (MAX_DESC_SIZE / TXBB_SIZE)
111 * OS related constants and tunables
114 #define MLX4_EN_WATCHDOG_TIMEOUT (15 * HZ)
116 #define MLX4_EN_ALLOC_ORDER 2
117 #define MLX4_EN_ALLOC_SIZE (PAGE_SIZE << MLX4_EN_ALLOC_ORDER)
119 #define MLX4_EN_MAX_LRO_DESCRIPTORS 32
121 /* Receive fragment sizes; we use at most 4 fragments (for 9600 byte MTU
122 * and 4K allocations) */
124 FRAG_SZ0
= 512 - NET_IP_ALIGN
,
127 FRAG_SZ3
= MLX4_EN_ALLOC_SIZE
129 #define MLX4_EN_MAX_RX_FRAGS 4
131 /* Maximum ring sizes */
132 #define MLX4_EN_MAX_TX_SIZE 8192
133 #define MLX4_EN_MAX_RX_SIZE 8192
135 /* Minimum ring size for our page-allocation sceme to work */
136 #define MLX4_EN_MIN_RX_SIZE (MLX4_EN_ALLOC_SIZE / SMP_CACHE_BYTES)
137 #define MLX4_EN_MIN_TX_SIZE (4096 / TXBB_SIZE)
139 #define MLX4_EN_SMALL_PKT_SIZE 64
140 #define MLX4_EN_NUM_TX_RINGS 8
141 #define MLX4_EN_NUM_PPP_RINGS 8
142 #define MLX4_EN_DEF_TX_RING_SIZE 512
143 #define MLX4_EN_DEF_RX_RING_SIZE 1024
145 /* Target number of packets to coalesce with interrupt moderation */
146 #define MLX4_EN_RX_COAL_TARGET 44
147 #define MLX4_EN_RX_COAL_TIME 0x10
149 #define MLX4_EN_TX_COAL_PKTS 5
150 #define MLX4_EN_TX_COAL_TIME 0x80
152 #define MLX4_EN_RX_RATE_LOW 400000
153 #define MLX4_EN_RX_COAL_TIME_LOW 0
154 #define MLX4_EN_RX_RATE_HIGH 450000
155 #define MLX4_EN_RX_COAL_TIME_HIGH 128
156 #define MLX4_EN_RX_SIZE_THRESH 1024
157 #define MLX4_EN_RX_RATE_THRESH (1000000 / MLX4_EN_RX_COAL_TIME_HIGH)
158 #define MLX4_EN_SAMPLE_INTERVAL 0
160 #define MLX4_EN_AUTO_CONF 0xffff
162 #define MLX4_EN_DEF_RX_PAUSE 1
163 #define MLX4_EN_DEF_TX_PAUSE 1
165 /* Interval between sucessive polls in the Tx routine when polling is used
166 instead of interrupts (in per-core Tx rings) - should be power of 2 */
167 #define MLX4_EN_TX_POLL_MODER 16
168 #define MLX4_EN_TX_POLL_TIMEOUT (HZ / 4)
170 #define ETH_LLC_SNAP_SIZE 8
172 #define SMALL_PACKET_SIZE (256 - NET_IP_ALIGN)
173 #define HEADER_COPY_SIZE (128 - NET_IP_ALIGN)
175 #define MLX4_EN_MIN_MTU 46
176 #define ETH_BCAST 0xffffffffffffULL
178 #ifdef MLX4_EN_PERF_STAT
179 /* Number of samples to 'average' */
181 #define AVG_FACTOR 1024
182 #define NUM_PERF_STATS NUM_PERF_COUNTERS
184 #define INC_PERF_COUNTER(cnt) (++(cnt))
185 #define ADD_PERF_COUNTER(cnt, add) ((cnt) += (add))
186 #define AVG_PERF_COUNTER(cnt, sample) \
187 ((cnt) = ((cnt) * (AVG_SIZE - 1) + (sample) * AVG_FACTOR) / AVG_SIZE)
188 #define GET_PERF_COUNTER(cnt) (cnt)
189 #define GET_AVG_PERF_COUNTER(cnt) ((cnt) / AVG_FACTOR)
193 #define NUM_PERF_STATS 0
194 #define INC_PERF_COUNTER(cnt) do {} while (0)
195 #define ADD_PERF_COUNTER(cnt, add) do {} while (0)
196 #define AVG_PERF_COUNTER(cnt, sample) do {} while (0)
197 #define GET_PERF_COUNTER(cnt) (0)
198 #define GET_AVG_PERF_COUNTER(cnt) (0)
199 #endif /* MLX4_EN_PERF_STAT */
214 #define ROUNDUP_LOG2(x) ilog2(roundup_pow_of_two(x))
215 #define XNOR(x, y) (!(x) == !(y))
216 #define ILLEGAL_MAC(addr) (addr == 0xffffffffffffULL || addr == 0x0)
219 struct mlx4_en_tx_info
{
228 #define MLX4_EN_BIT_DESC_OWN 0x80000000
229 #define CTRL_SIZE sizeof(struct mlx4_wqe_ctrl_seg)
230 #define MLX4_EN_MEMTYPE_PAD 0x100
231 #define DS_SIZE sizeof(struct mlx4_wqe_data_seg)
234 struct mlx4_en_tx_desc
{
235 struct mlx4_wqe_ctrl_seg ctrl
;
237 struct mlx4_wqe_data_seg data
; /* at least one data segment */
238 struct mlx4_wqe_lso_seg lso
;
239 struct mlx4_wqe_inline_seg inl
;
243 #define MLX4_EN_USE_SRQ 0x01000000
245 struct mlx4_en_rx_alloc
{
250 struct mlx4_en_tx_ring
{
251 struct mlx4_hwq_resources wqres
;
252 u32 size
; /* number of TXBBs */
255 u16 cqn
; /* index of port CQ associated with this ring */
263 struct mlx4_en_tx_info
*tx_info
;
267 struct mlx4_qp_context context
;
269 enum mlx4_qp_state qp_state
;
270 struct mlx4_srq dummy
;
272 unsigned long packets
;
273 spinlock_t comp_lock
;
276 struct mlx4_en_rx_desc
{
277 /* actual number of entries depends on rx ring stride */
278 struct mlx4_wqe_data_seg data
[0];
281 struct mlx4_en_rx_ring
{
282 struct mlx4_hwq_resources wqres
;
283 struct mlx4_en_rx_alloc page_alloc
[MLX4_EN_MAX_RX_FRAGS
];
284 struct net_lro_mgr lro
;
285 u32 size
; /* number of Rx descs*/
290 u16 cqn
; /* index of port CQ associated with this ring */
297 unsigned long packets
;
301 static inline int mlx4_en_can_lro(__be16 status
)
303 return (status
& cpu_to_be16(MLX4_CQE_STATUS_IPV4
|
304 MLX4_CQE_STATUS_IPV4F
|
305 MLX4_CQE_STATUS_IPV6
|
306 MLX4_CQE_STATUS_IPV4OPT
|
307 MLX4_CQE_STATUS_TCP
|
308 MLX4_CQE_STATUS_UDP
|
309 MLX4_CQE_STATUS_IPOK
)) ==
310 cpu_to_be16(MLX4_CQE_STATUS_IPV4
|
311 MLX4_CQE_STATUS_IPOK
|
312 MLX4_CQE_STATUS_TCP
);
317 struct mlx4_hwq_resources wqres
;
320 struct net_device
*dev
;
321 struct napi_struct napi
;
322 /* Per-core Tx cq processing support */
323 struct timer_list timer
;
330 struct mlx4_cqe
*buf
;
331 #define MLX4_EN_OPCODE_ERROR 0x1e
334 struct mlx4_en_port_profile
{
346 struct mlx4_en_profile
{
353 struct mlx4_en_port_profile prof
[MLX4_MAX_PORTS
+ 1];
357 struct mlx4_dev
*dev
;
358 struct pci_dev
*pdev
;
359 struct mutex state_lock
;
360 struct net_device
*pndev
[MLX4_MAX_PORTS
+ 1];
363 struct mlx4_en_profile profile
;
365 struct workqueue_struct
*workqueue
;
366 struct device
*dma_device
;
367 void __iomem
*uar_map
;
368 struct mlx4_uar priv_uar
;
375 struct mlx4_en_rss_map
{
377 struct mlx4_qp qps
[MAX_RX_RINGS
];
378 enum mlx4_qp_state state
[MAX_RX_RINGS
];
379 struct mlx4_qp indir_qp
;
380 enum mlx4_qp_state indir_state
;
383 struct mlx4_en_rss_context
{
392 struct mlx4_en_pkt_stats
{
393 unsigned long broadcast
;
394 unsigned long rx_prio
[8];
395 unsigned long tx_prio
[8];
396 #define NUM_PKT_STATS 17
399 struct mlx4_en_port_stats
{
400 unsigned long lro_aggregated
;
401 unsigned long lro_flushed
;
402 unsigned long lro_no_desc
;
403 unsigned long tso_packets
;
404 unsigned long queue_stopped
;
405 unsigned long wake_queue
;
406 unsigned long tx_timeout
;
407 unsigned long rx_alloc_failed
;
408 unsigned long rx_chksum_good
;
409 unsigned long rx_chksum_none
;
410 unsigned long tx_chksum_offload
;
411 #define NUM_PORT_STATS 11
414 struct mlx4_en_perf_stats
{
421 #define NUM_PERF_COUNTERS 6
424 struct mlx4_en_frag_info
{
426 u16 frag_prefix_size
;
433 struct mlx4_en_priv
{
434 struct mlx4_en_dev
*mdev
;
435 struct mlx4_en_port_profile
*prof
;
436 struct net_device
*dev
;
437 struct vlan_group
*vlgrp
;
438 struct net_device_stats stats
;
439 struct net_device_stats ret_stats
;
440 spinlock_t stats_lock
;
442 unsigned long last_moder_packets
;
443 unsigned long last_moder_tx_packets
;
444 unsigned long last_moder_bytes
;
445 unsigned long last_moder_jiffies
;
456 u16 adaptive_rx_coal
;
459 struct mlx4_hwq_resources res
;
473 struct mlx4_en_rss_map rss_map
;
475 #define MLX4_EN_FLAG_PROMISC 0x1
479 struct mlx4_en_frag_info frag_info
[MLX4_EN_MAX_RX_FRAGS
];
483 struct mlx4_en_tx_ring tx_ring
[MAX_TX_RINGS
];
484 struct mlx4_en_rx_ring rx_ring
[MAX_RX_RINGS
];
485 struct mlx4_en_cq tx_cq
[MAX_TX_RINGS
];
486 struct mlx4_en_cq rx_cq
[MAX_RX_RINGS
];
487 struct work_struct mcast_task
;
488 struct work_struct mac_task
;
489 struct work_struct watchdog_task
;
490 struct work_struct linkstate_task
;
491 struct delayed_work stats_task
;
492 struct mlx4_en_perf_stats pstats
;
493 struct mlx4_en_pkt_stats pkstats
;
494 struct mlx4_en_port_stats port_stats
;
495 struct dev_mc_list
*mc_list
;
496 struct mlx4_en_stat_out_mbox hw_stats
;
500 void mlx4_en_destroy_netdev(struct net_device
*dev
);
501 int mlx4_en_init_netdev(struct mlx4_en_dev
*mdev
, int port
,
502 struct mlx4_en_port_profile
*prof
);
504 int mlx4_en_start_port(struct net_device
*dev
);
505 void mlx4_en_stop_port(struct net_device
*dev
);
507 void mlx4_en_free_resources(struct mlx4_en_priv
*priv
);
508 int mlx4_en_alloc_resources(struct mlx4_en_priv
*priv
);
510 int mlx4_en_create_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
,
511 int entries
, int ring
, enum cq_type mode
);
512 void mlx4_en_destroy_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
513 int mlx4_en_activate_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
514 void mlx4_en_deactivate_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
515 int mlx4_en_set_cq_moder(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
516 int mlx4_en_arm_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
518 void mlx4_en_poll_tx_cq(unsigned long data
);
519 void mlx4_en_tx_irq(struct mlx4_cq
*mcq
);
520 u16
mlx4_en_select_queue(struct net_device
*dev
, struct sk_buff
*skb
);
521 netdev_tx_t
mlx4_en_xmit(struct sk_buff
*skb
, struct net_device
*dev
);
523 int mlx4_en_create_tx_ring(struct mlx4_en_priv
*priv
, struct mlx4_en_tx_ring
*ring
,
524 u32 size
, u16 stride
);
525 void mlx4_en_destroy_tx_ring(struct mlx4_en_priv
*priv
, struct mlx4_en_tx_ring
*ring
);
526 int mlx4_en_activate_tx_ring(struct mlx4_en_priv
*priv
,
527 struct mlx4_en_tx_ring
*ring
,
529 void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv
*priv
,
530 struct mlx4_en_tx_ring
*ring
);
532 int mlx4_en_create_rx_ring(struct mlx4_en_priv
*priv
,
533 struct mlx4_en_rx_ring
*ring
,
534 u32 size
, u16 stride
);
535 void mlx4_en_destroy_rx_ring(struct mlx4_en_priv
*priv
,
536 struct mlx4_en_rx_ring
*ring
);
537 int mlx4_en_activate_rx_rings(struct mlx4_en_priv
*priv
);
538 void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv
*priv
,
539 struct mlx4_en_rx_ring
*ring
);
540 int mlx4_en_process_rx_cq(struct net_device
*dev
,
541 struct mlx4_en_cq
*cq
,
543 int mlx4_en_poll_rx_cq(struct napi_struct
*napi
, int budget
);
544 void mlx4_en_fill_qp_context(struct mlx4_en_priv
*priv
, int size
, int stride
,
545 int is_tx
, int rss
, int qpn
, int cqn
,
546 struct mlx4_qp_context
*context
);
547 void mlx4_en_sqp_event(struct mlx4_qp
*qp
, enum mlx4_event event
);
548 int mlx4_en_map_buffer(struct mlx4_buf
*buf
);
549 void mlx4_en_unmap_buffer(struct mlx4_buf
*buf
);
551 void mlx4_en_calc_rx_buf(struct net_device
*dev
);
552 int mlx4_en_config_rss_steer(struct mlx4_en_priv
*priv
);
553 void mlx4_en_release_rss_steer(struct mlx4_en_priv
*priv
);
554 int mlx4_en_free_tx_buf(struct net_device
*dev
, struct mlx4_en_tx_ring
*ring
);
555 void mlx4_en_rx_irq(struct mlx4_cq
*mcq
);
557 int mlx4_SET_MCAST_FLTR(struct mlx4_dev
*dev
, u8 port
, u64 mac
, u64 clear
, u8 mode
);
558 int mlx4_SET_VLAN_FLTR(struct mlx4_dev
*dev
, u8 port
, struct vlan_group
*grp
);
559 int mlx4_SET_PORT_general(struct mlx4_dev
*dev
, u8 port
, int mtu
,
560 u8 pptx
, u8 pfctx
, u8 pprx
, u8 pfcrx
);
561 int mlx4_SET_PORT_qpn_calc(struct mlx4_dev
*dev
, u8 port
, u32 base_qpn
,
564 int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev
*mdev
, u8 port
, u8 reset
);
569 extern const struct ethtool_ops mlx4_en_ethtool_ops
;