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/bitops.h>
38 #include <linux/compiler.h>
39 #include <linux/list.h>
40 #include <linux/mutex.h>
41 #include <linux/netdevice.h>
42 #include <linux/if_vlan.h>
43 #ifdef CONFIG_MLX4_EN_DCB
44 #include <linux/dcbnl.h>
46 #include <linux/cpu_rmap.h>
48 #include <linux/mlx4/device.h>
49 #include <linux/mlx4/qp.h>
50 #include <linux/mlx4/cq.h>
51 #include <linux/mlx4/srq.h>
52 #include <linux/mlx4/doorbell.h>
53 #include <linux/mlx4/cmd.h>
57 #define DRV_NAME "mlx4_en"
58 #define DRV_VERSION "2.0"
59 #define DRV_RELDATE "Dec 2011"
61 #define MLX4_EN_MSG_LEVEL (NETIF_MSG_LINK | NETIF_MSG_IFDOWN)
68 #define MLX4_EN_PAGE_SHIFT 12
69 #define MLX4_EN_PAGE_SIZE (1 << MLX4_EN_PAGE_SHIFT)
70 #define MAX_RX_RINGS 16
71 #define MIN_RX_RINGS 4
73 #define HEADROOM (2048 / TXBB_SIZE + 1)
74 #define STAMP_STRIDE 64
75 #define STAMP_DWORDS (STAMP_STRIDE / 4)
76 #define STAMP_SHIFT 31
77 #define STAMP_VAL 0x7fffffff
78 #define STATS_DELAY (HZ / 4)
79 #define MAX_NUM_OF_FS_RULES 256
81 #define MLX4_EN_FILTER_HASH_SHIFT 4
82 #define MLX4_EN_FILTER_EXPIRY_QUOTA 60
84 /* Typical TSO descriptor with 16 gather entries is 352 bytes... */
85 #define MAX_DESC_SIZE 512
86 #define MAX_DESC_TXBBS (MAX_DESC_SIZE / TXBB_SIZE)
89 * OS related constants and tunables
92 #define MLX4_EN_WATCHDOG_TIMEOUT (15 * HZ)
94 /* Use the maximum between 16384 and a single page */
95 #define MLX4_EN_ALLOC_SIZE PAGE_ALIGN(16384)
96 #define MLX4_EN_ALLOC_ORDER get_order(MLX4_EN_ALLOC_SIZE)
98 /* Receive fragment sizes; we use at most 4 fragments (for 9600 byte MTU
99 * and 4K allocations) */
101 FRAG_SZ0
= 512 - NET_IP_ALIGN
,
104 FRAG_SZ3
= MLX4_EN_ALLOC_SIZE
106 #define MLX4_EN_MAX_RX_FRAGS 4
108 /* Maximum ring sizes */
109 #define MLX4_EN_MAX_TX_SIZE 8192
110 #define MLX4_EN_MAX_RX_SIZE 8192
112 /* Minimum ring size for our page-allocation scheme to work */
113 #define MLX4_EN_MIN_RX_SIZE (MLX4_EN_ALLOC_SIZE / SMP_CACHE_BYTES)
114 #define MLX4_EN_MIN_TX_SIZE (4096 / TXBB_SIZE)
116 #define MLX4_EN_SMALL_PKT_SIZE 64
117 #define MLX4_EN_MAX_TX_RING_P_UP 32
118 #define MLX4_EN_NUM_UP 8
119 #define MLX4_EN_DEF_TX_RING_SIZE 512
120 #define MLX4_EN_DEF_RX_RING_SIZE 1024
122 /* Target number of packets to coalesce with interrupt moderation */
123 #define MLX4_EN_RX_COAL_TARGET 44
124 #define MLX4_EN_RX_COAL_TIME 0x10
126 #define MLX4_EN_TX_COAL_PKTS 16
127 #define MLX4_EN_TX_COAL_TIME 0x10
129 #define MLX4_EN_RX_RATE_LOW 400000
130 #define MLX4_EN_RX_COAL_TIME_LOW 0
131 #define MLX4_EN_RX_RATE_HIGH 450000
132 #define MLX4_EN_RX_COAL_TIME_HIGH 128
133 #define MLX4_EN_RX_SIZE_THRESH 1024
134 #define MLX4_EN_RX_RATE_THRESH (1000000 / MLX4_EN_RX_COAL_TIME_HIGH)
135 #define MLX4_EN_SAMPLE_INTERVAL 0
136 #define MLX4_EN_AVG_PKT_SMALL 256
138 #define MLX4_EN_AUTO_CONF 0xffff
140 #define MLX4_EN_DEF_RX_PAUSE 1
141 #define MLX4_EN_DEF_TX_PAUSE 1
143 /* Interval between successive polls in the Tx routine when polling is used
144 instead of interrupts (in per-core Tx rings) - should be power of 2 */
145 #define MLX4_EN_TX_POLL_MODER 16
146 #define MLX4_EN_TX_POLL_TIMEOUT (HZ / 4)
148 #define ETH_LLC_SNAP_SIZE 8
150 #define SMALL_PACKET_SIZE (256 - NET_IP_ALIGN)
151 #define HEADER_COPY_SIZE (128 - NET_IP_ALIGN)
152 #define MLX4_LOOPBACK_TEST_PAYLOAD (HEADER_COPY_SIZE - ETH_HLEN)
154 #define MLX4_EN_MIN_MTU 46
155 #define ETH_BCAST 0xffffffffffffULL
157 #define MLX4_EN_LOOPBACK_RETRIES 5
158 #define MLX4_EN_LOOPBACK_TIMEOUT 100
160 #ifdef MLX4_EN_PERF_STAT
161 /* Number of samples to 'average' */
163 #define AVG_FACTOR 1024
164 #define NUM_PERF_STATS NUM_PERF_COUNTERS
166 #define INC_PERF_COUNTER(cnt) (++(cnt))
167 #define ADD_PERF_COUNTER(cnt, add) ((cnt) += (add))
168 #define AVG_PERF_COUNTER(cnt, sample) \
169 ((cnt) = ((cnt) * (AVG_SIZE - 1) + (sample) * AVG_FACTOR) / AVG_SIZE)
170 #define GET_PERF_COUNTER(cnt) (cnt)
171 #define GET_AVG_PERF_COUNTER(cnt) ((cnt) / AVG_FACTOR)
175 #define NUM_PERF_STATS 0
176 #define INC_PERF_COUNTER(cnt) do {} while (0)
177 #define ADD_PERF_COUNTER(cnt, add) do {} while (0)
178 #define AVG_PERF_COUNTER(cnt, sample) do {} while (0)
179 #define GET_PERF_COUNTER(cnt) (0)
180 #define GET_AVG_PERF_COUNTER(cnt) (0)
181 #endif /* MLX4_EN_PERF_STAT */
196 #define ROUNDUP_LOG2(x) ilog2(roundup_pow_of_two(x))
197 #define XNOR(x, y) (!(x) == !(y))
198 #define ILLEGAL_MAC(addr) (addr == 0xffffffffffffULL || addr == 0x0)
201 struct mlx4_en_tx_info
{
211 #define MLX4_EN_BIT_DESC_OWN 0x80000000
212 #define CTRL_SIZE sizeof(struct mlx4_wqe_ctrl_seg)
213 #define MLX4_EN_MEMTYPE_PAD 0x100
214 #define DS_SIZE sizeof(struct mlx4_wqe_data_seg)
217 struct mlx4_en_tx_desc
{
218 struct mlx4_wqe_ctrl_seg ctrl
;
220 struct mlx4_wqe_data_seg data
; /* at least one data segment */
221 struct mlx4_wqe_lso_seg lso
;
222 struct mlx4_wqe_inline_seg inl
;
226 #define MLX4_EN_USE_SRQ 0x01000000
228 #define MLX4_EN_CX3_LOW_ID 0x1000
229 #define MLX4_EN_CX3_HIGH_ID 0x1005
231 struct mlx4_en_rx_alloc
{
237 struct mlx4_en_tx_ring
{
238 struct mlx4_hwq_resources wqres
;
239 u32 size
; /* number of TXBBs */
242 u16 cqn
; /* index of port CQ associated with this ring */
249 struct mlx4_en_tx_info
*tx_info
;
253 struct mlx4_qp_context context
;
255 enum mlx4_qp_state qp_state
;
256 struct mlx4_srq dummy
;
258 unsigned long packets
;
259 unsigned long tx_csum
;
262 struct netdev_queue
*tx_queue
;
265 struct mlx4_en_rx_desc
{
266 /* actual number of entries depends on rx ring stride */
267 struct mlx4_wqe_data_seg data
[0];
270 struct mlx4_en_rx_ring
{
271 struct mlx4_hwq_resources wqres
;
272 struct mlx4_en_rx_alloc page_alloc
[MLX4_EN_MAX_RX_FRAGS
];
273 u32 size
; /* number of Rx descs*/
278 u16 cqn
; /* index of port CQ associated with this ring */
286 unsigned long packets
;
287 unsigned long csum_ok
;
288 unsigned long csum_none
;
293 struct mlx4_hwq_resources wqres
;
296 struct net_device
*dev
;
297 struct napi_struct napi
;
304 struct mlx4_cqe
*buf
;
305 #define MLX4_EN_OPCODE_ERROR 0x1e
308 struct mlx4_en_port_profile
{
321 struct mlx4_en_profile
{
328 u8 num_tx_rings_p_up
;
329 struct mlx4_en_port_profile prof
[MLX4_MAX_PORTS
+ 1];
333 struct mlx4_dev
*dev
;
334 struct pci_dev
*pdev
;
335 struct mutex state_lock
;
336 struct net_device
*pndev
[MLX4_MAX_PORTS
+ 1];
339 struct mlx4_en_profile profile
;
341 struct workqueue_struct
*workqueue
;
342 struct device
*dma_device
;
343 void __iomem
*uar_map
;
344 struct mlx4_uar priv_uar
;
348 u8 mac_removed
[MLX4_MAX_PORTS
+ 1];
352 struct mlx4_en_rss_map
{
354 struct mlx4_qp qps
[MAX_RX_RINGS
];
355 enum mlx4_qp_state state
[MAX_RX_RINGS
];
356 struct mlx4_qp indir_qp
;
357 enum mlx4_qp_state indir_state
;
360 struct mlx4_en_port_state
{
366 struct mlx4_en_pkt_stats
{
367 unsigned long broadcast
;
368 unsigned long rx_prio
[8];
369 unsigned long tx_prio
[8];
370 #define NUM_PKT_STATS 17
373 struct mlx4_en_port_stats
{
374 unsigned long tso_packets
;
375 unsigned long queue_stopped
;
376 unsigned long wake_queue
;
377 unsigned long tx_timeout
;
378 unsigned long rx_alloc_failed
;
379 unsigned long rx_chksum_good
;
380 unsigned long rx_chksum_none
;
381 unsigned long tx_chksum_offload
;
382 #define NUM_PORT_STATS 8
385 struct mlx4_en_perf_stats
{
392 #define NUM_PERF_COUNTERS 6
395 enum mlx4_en_mclist_act
{
401 struct mlx4_en_mc_list
{
402 struct list_head list
;
403 enum mlx4_en_mclist_act action
;
408 struct mlx4_en_frag_info
{
410 u16 frag_prefix_size
;
417 #ifdef CONFIG_MLX4_EN_DCB
418 /* Minimal TC BW - setting to 0 will block traffic */
419 #define MLX4_EN_BW_MIN 1
420 #define MLX4_EN_BW_MAX 100 /* Utilize 100% of the line */
422 #define MLX4_EN_TC_ETS 7
426 struct ethtool_flow_id
{
427 struct ethtool_rx_flow_spec flow_spec
;
431 struct mlx4_en_priv
{
432 struct mlx4_en_dev
*mdev
;
433 struct mlx4_en_port_profile
*prof
;
434 struct net_device
*dev
;
435 unsigned long active_vlans
[BITS_TO_LONGS(VLAN_N_VID
)];
436 struct net_device_stats stats
;
437 struct net_device_stats ret_stats
;
438 struct mlx4_en_port_state port_state
;
439 spinlock_t stats_lock
;
440 struct ethtool_flow_id ethtool_rules
[MAX_NUM_OF_FS_RULES
];
442 unsigned long last_moder_packets
[MAX_RX_RINGS
];
443 unsigned long last_moder_tx_packets
;
444 unsigned long last_moder_bytes
[MAX_RX_RINGS
];
445 unsigned long last_moder_jiffies
;
446 int last_moder_time
[MAX_RX_RINGS
];
456 u16 adaptive_rx_coal
;
459 u32 validate_loopback
;
461 struct mlx4_hwq_resources res
;
474 struct mlx4_en_rss_map rss_map
;
477 #define MLX4_EN_FLAG_PROMISC 0x1
478 #define MLX4_EN_FLAG_MC_PROMISC 0x2
482 struct mlx4_en_frag_info frag_info
[MLX4_EN_MAX_RX_FRAGS
];
486 struct mlx4_en_tx_ring
*tx_ring
;
487 struct mlx4_en_rx_ring rx_ring
[MAX_RX_RINGS
];
488 struct mlx4_en_cq
*tx_cq
;
489 struct mlx4_en_cq rx_cq
[MAX_RX_RINGS
];
490 struct mlx4_qp drop_qp
;
491 struct work_struct mcast_task
;
492 struct work_struct mac_task
;
493 struct work_struct watchdog_task
;
494 struct work_struct linkstate_task
;
495 struct delayed_work stats_task
;
496 struct mlx4_en_perf_stats pstats
;
497 struct mlx4_en_pkt_stats pkstats
;
498 struct mlx4_en_port_stats port_stats
;
500 struct list_head mc_list
;
501 struct list_head curr_list
;
503 struct mlx4_en_stat_out_mbox hw_stats
;
509 #ifdef CONFIG_MLX4_EN_DCB
511 u16 maxrate
[IEEE_8021QAZ_MAX_TCS
];
513 #ifdef CONFIG_RFS_ACCEL
514 spinlock_t filters_lock
;
516 struct list_head filters
;
517 struct hlist_head filter_hash
[1 << MLX4_EN_FILTER_HASH_SHIFT
];
523 MLX4_EN_WOL_MAGIC
= (1ULL << 61),
524 MLX4_EN_WOL_ENABLED
= (1ULL << 62),
527 #define MLX4_EN_WOL_DO_MODIFY (1ULL << 63)
529 void mlx4_en_destroy_netdev(struct net_device
*dev
);
530 int mlx4_en_init_netdev(struct mlx4_en_dev
*mdev
, int port
,
531 struct mlx4_en_port_profile
*prof
);
533 int mlx4_en_start_port(struct net_device
*dev
);
534 void mlx4_en_stop_port(struct net_device
*dev
);
536 void mlx4_en_free_resources(struct mlx4_en_priv
*priv
);
537 int mlx4_en_alloc_resources(struct mlx4_en_priv
*priv
);
539 int mlx4_en_create_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
,
540 int entries
, int ring
, enum cq_type mode
);
541 void mlx4_en_destroy_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
542 int mlx4_en_activate_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
,
544 void mlx4_en_deactivate_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
545 int mlx4_en_set_cq_moder(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
546 int mlx4_en_arm_cq(struct mlx4_en_priv
*priv
, struct mlx4_en_cq
*cq
);
548 void mlx4_en_tx_irq(struct mlx4_cq
*mcq
);
549 u16
mlx4_en_select_queue(struct net_device
*dev
, struct sk_buff
*skb
);
550 netdev_tx_t
mlx4_en_xmit(struct sk_buff
*skb
, struct net_device
*dev
);
552 int mlx4_en_create_tx_ring(struct mlx4_en_priv
*priv
, struct mlx4_en_tx_ring
*ring
,
553 int qpn
, u32 size
, u16 stride
);
554 void mlx4_en_destroy_tx_ring(struct mlx4_en_priv
*priv
, struct mlx4_en_tx_ring
*ring
);
555 int mlx4_en_activate_tx_ring(struct mlx4_en_priv
*priv
,
556 struct mlx4_en_tx_ring
*ring
,
557 int cq
, int user_prio
);
558 void mlx4_en_deactivate_tx_ring(struct mlx4_en_priv
*priv
,
559 struct mlx4_en_tx_ring
*ring
);
561 int mlx4_en_create_rx_ring(struct mlx4_en_priv
*priv
,
562 struct mlx4_en_rx_ring
*ring
,
563 u32 size
, u16 stride
);
564 void mlx4_en_destroy_rx_ring(struct mlx4_en_priv
*priv
,
565 struct mlx4_en_rx_ring
*ring
,
566 u32 size
, u16 stride
);
567 int mlx4_en_activate_rx_rings(struct mlx4_en_priv
*priv
);
568 void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv
*priv
,
569 struct mlx4_en_rx_ring
*ring
);
570 int mlx4_en_process_rx_cq(struct net_device
*dev
,
571 struct mlx4_en_cq
*cq
,
573 int mlx4_en_poll_rx_cq(struct napi_struct
*napi
, int budget
);
574 void mlx4_en_fill_qp_context(struct mlx4_en_priv
*priv
, int size
, int stride
,
575 int is_tx
, int rss
, int qpn
, int cqn
, int user_prio
,
576 struct mlx4_qp_context
*context
);
577 void mlx4_en_sqp_event(struct mlx4_qp
*qp
, enum mlx4_event event
);
578 int mlx4_en_map_buffer(struct mlx4_buf
*buf
);
579 void mlx4_en_unmap_buffer(struct mlx4_buf
*buf
);
581 void mlx4_en_calc_rx_buf(struct net_device
*dev
);
582 int mlx4_en_config_rss_steer(struct mlx4_en_priv
*priv
);
583 void mlx4_en_release_rss_steer(struct mlx4_en_priv
*priv
);
584 int mlx4_en_create_drop_qp(struct mlx4_en_priv
*priv
);
585 void mlx4_en_destroy_drop_qp(struct mlx4_en_priv
*priv
);
586 int mlx4_en_free_tx_buf(struct net_device
*dev
, struct mlx4_en_tx_ring
*ring
);
587 void mlx4_en_rx_irq(struct mlx4_cq
*mcq
);
589 int mlx4_SET_MCAST_FLTR(struct mlx4_dev
*dev
, u8 port
, u64 mac
, u64 clear
, u8 mode
);
590 int mlx4_SET_VLAN_FLTR(struct mlx4_dev
*dev
, struct mlx4_en_priv
*priv
);
592 int mlx4_en_DUMP_ETH_STATS(struct mlx4_en_dev
*mdev
, u8 port
, u8 reset
);
593 int mlx4_en_QUERY_PORT(struct mlx4_en_dev
*mdev
, u8 port
);
595 #ifdef CONFIG_MLX4_EN_DCB
596 extern const struct dcbnl_rtnl_ops mlx4_en_dcbnl_ops
;
599 #ifdef CONFIG_RFS_ACCEL
600 void mlx4_en_cleanup_filters(struct mlx4_en_priv
*priv
,
601 struct mlx4_en_rx_ring
*rx_ring
);
604 #define MLX4_EN_NUM_SELF_TEST 5
605 void mlx4_en_ex_selftest(struct net_device
*dev
, u32
*flags
, u64
*buf
);
606 u64
mlx4_en_mac_to_u64(u8
*addr
);
611 extern const struct ethtool_ops mlx4_en_ethtool_ops
;
616 * printk / logging functions
620 int en_print(const char *level
, const struct mlx4_en_priv
*priv
,
621 const char *format
, ...);
623 #define en_dbg(mlevel, priv, format, arg...) \
625 if (NETIF_MSG_##mlevel & priv->msg_enable) \
626 en_print(KERN_DEBUG, priv, format, ##arg); \
628 #define en_warn(priv, format, arg...) \
629 en_print(KERN_WARNING, priv, format, ##arg)
630 #define en_err(priv, format, arg...) \
631 en_print(KERN_ERR, priv, format, ##arg)
632 #define en_info(priv, format, arg...) \
633 en_print(KERN_INFO, priv, format, ## arg)
635 #define mlx4_err(mdev, format, arg...) \
636 pr_err("%s %s: " format, DRV_NAME, \
637 dev_name(&mdev->pdev->dev), ##arg)
638 #define mlx4_info(mdev, format, arg...) \
639 pr_info("%s %s: " format, DRV_NAME, \
640 dev_name(&mdev->pdev->dev), ##arg)
641 #define mlx4_warn(mdev, format, arg...) \
642 pr_warning("%s %s: " format, DRV_NAME, \
643 dev_name(&mdev->pdev->dev), ##arg)