summaryrefslogtreecommitdiff
path: root/drivers/staging/octeon/ethernet-tx.c
blob: 05b58f8b58fd97bfb5dd20d7048612920e9d572b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
/*********************************************************************
 * Author: Cavium Networks
 *
 * Contact: support@caviumnetworks.com
 * This file is part of the OCTEON SDK
 *
 * Copyright (c) 2003-2010 Cavium Networks
 *
 * This file is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License, Version 2, as
 * published by the Free Software Foundation.
 *
 * This file is distributed in the hope that it will be useful, but
 * AS-IS and WITHOUT ANY WARRANTY; without even the implied warranty
 * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE, TITLE, or
 * NONINFRINGEMENT.  See the GNU General Public License for more
 * details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this file; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
 * or visit http://www.gnu.org/licenses/.
 *
 * This file may also be available under a different license from Cavium.
 * Contact Cavium Networks for more information
*********************************************************************/
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/netdevice.h>
#include <linux/init.h>
#include <linux/etherdevice.h>
#include <linux/ip.h>
#include <linux/string.h>
#include <net/dst.h>
#ifdef CONFIG_XFRM
#include <linux/xfrm.h>
#include <net/xfrm.h>
#endif /* CONFIG_XFRM */

#include <asm/atomic.h>

#include <asm/octeon/octeon.h>

#include "ethernet-defines.h"
#include "octeon-ethernet.h"
#include "ethernet-tx.h"
#include "ethernet-util.h"

#include "cvmx-wqe.h"
#include "cvmx-fau.h"
#include "cvmx-pko.h"
#include "cvmx-helper.h"

#include "cvmx-gmxx-defs.h"

/*
 * You can define GET_SKBUFF_QOS() to override how the skbuff output
 * function determines which output queue is used. The default
 * implementation always uses the base queue for the port. If, for
 * example, you wanted to use the skb->priority fieid, define
 * GET_SKBUFF_QOS as: #define GET_SKBUFF_QOS(skb) ((skb)->priority)
 */
#ifndef GET_SKBUFF_QOS
#define GET_SKBUFF_QOS(skb) 0
#endif


static inline int32_t cvm_oct_adjust_skb_to_free(int32_t skb_to_free, int fau)
{
	int32_t undo;
	undo = skb_to_free > 0 ? MAX_SKB_TO_FREE : skb_to_free + MAX_SKB_TO_FREE;
	if (undo > 0)
		cvmx_fau_atomic_add32(fau, -undo);
	skb_to_free = -skb_to_free > MAX_SKB_TO_FREE ? MAX_SKB_TO_FREE : -skb_to_free;
	return skb_to_free;
}

void cvm_oct_free_tx_skbs(struct octeon_ethernet *priv)
{
	int32_t skb_to_free;
	int qos, queues_per_port;
	queues_per_port = cvmx_pko_get_num_queues(priv->port);
	/* Drain any pending packets in the free list */
	for (qos = 0; qos < queues_per_port; qos++) {
		if (skb_queue_len(&priv->tx_free_list[qos]) == 0)
			continue;
		skb_to_free = cvmx_fau_fetch_and_add32(priv->fau+qos*4, MAX_SKB_TO_FREE);
		skb_to_free = cvm_oct_adjust_skb_to_free(skb_to_free, priv->fau+qos*4);

		while (skb_to_free > 0) {
			dev_kfree_skb_any(skb_dequeue(&priv->tx_free_list[qos]));
			skb_to_free--;
		}
	}
}

enum hrtimer_restart cvm_oct_restart_tx(struct hrtimer *timer)
{
	struct octeon_ethernet *priv = container_of(timer, struct octeon_ethernet, tx_restart_timer);
	struct net_device *dev = cvm_oct_device[priv->port];

	cvm_oct_free_tx_skbs(priv);

	if (netif_queue_stopped(dev))
		netif_wake_queue(dev);

	return HRTIMER_NORESTART;
}

/**
 * Packet transmit
 *
 * @skb:    Packet to send
 * @dev:    Device info structure
 * Returns Always returns zero
 */
int cvm_oct_xmit(struct sk_buff *skb, struct net_device *dev)
{
	cvmx_pko_command_word0_t pko_command;
	union cvmx_buf_ptr hw_buffer;
	uint64_t old_scratch;
	uint64_t old_scratch2;
	int qos;
	enum {QUEUE_CORE, QUEUE_HW, QUEUE_DROP} queue_type;
	struct octeon_ethernet *priv = netdev_priv(dev);
	struct sk_buff *to_free_list;
	int32_t skb_to_free;
	int32_t buffers_to_free;
	unsigned long flags;
#if REUSE_SKBUFFS_WITHOUT_FREE
	unsigned char *fpa_head;
#endif

	/*
	 * Prefetch the private data structure.  It is larger that one
	 * cache line.
	 */
	prefetch(priv);

	/*
	 * The check on CVMX_PKO_QUEUES_PER_PORT_* is designed to
	 * completely remove "qos" in the event neither interface
	 * supports multiple queues per port.
	 */
	if ((CVMX_PKO_QUEUES_PER_PORT_INTERFACE0 > 1) ||
	    (CVMX_PKO_QUEUES_PER_PORT_INTERFACE1 > 1)) {
		qos = GET_SKBUFF_QOS(skb);
		if (qos <= 0)
			qos = 0;
		else if (qos >= cvmx_pko_get_num_queues(priv->port))
			qos = 0;
	} else
		qos = 0;

	if (USE_ASYNC_IOBDMA) {
		/* Save scratch in case userspace is using it */
		CVMX_SYNCIOBDMA;
		old_scratch = cvmx_scratch_read64(CVMX_SCR_SCRATCH);
		old_scratch2 = cvmx_scratch_read64(CVMX_SCR_SCRATCH + 8);

		/*
		 * Fetch and increment the number of packets to be
		 * freed.
		 */
		cvmx_fau_async_fetch_and_add32(CVMX_SCR_SCRATCH + 8,
					       FAU_NUM_PACKET_BUFFERS_TO_FREE,
					       0);
		cvmx_fau_async_fetch_and_add32(CVMX_SCR_SCRATCH,
					       priv->fau + qos * 4,
					       MAX_SKB_TO_FREE);
	}

	/*
	 * The CN3XXX series of parts has an errata (GMX-401) which
	 * causes the GMX block to hang if a collision occurs towards
	 * the end of a <68 byte packet. As a workaround for this, we
	 * pad packets to be 68 bytes whenever we are in half duplex
	 * mode. We don't handle the case of having a small packet but
	 * no room to add the padding.  The kernel should always give
	 * us at least a cache line
	 */
	if ((skb->len < 64) && OCTEON_IS_MODEL(OCTEON_CN3XXX)) {
		union cvmx_gmxx_prtx_cfg gmx_prt_cfg;
		int interface = INTERFACE(priv->port);
		int index = INDEX(priv->port);

		if (interface < 2) {
			/* We only need to pad packet in half duplex mode */
			gmx_prt_cfg.u64 =
			    cvmx_read_csr(CVMX_GMXX_PRTX_CFG(index, interface));
			if (gmx_prt_cfg.s.duplex == 0) {
				int add_bytes = 64 - skb->len;
				if ((skb_tail_pointer(skb) + add_bytes) <=
				    skb_end_pointer(skb))
					memset(__skb_put(skb, add_bytes), 0,
					       add_bytes);
			}
		}
	}

	/* Build the PKO buffer pointer */
	hw_buffer.u64 = 0;
	hw_buffer.s.addr = cvmx_ptr_to_phys(skb->data);
	hw_buffer.s.pool = 0;
	hw_buffer.s.size =
	    (unsigned long)skb_end_pointer(skb) - (unsigned long)skb->head;

	/* Build the PKO command */
	pko_command.u64 = 0;
	pko_command.s.n2 = 1;	/* Don't pollute L2 with the outgoing packet */
	pko_command.s.segs = 1;
	pko_command.s.total_bytes = skb->len;
	pko_command.s.size0 = CVMX_FAU_OP_SIZE_32;
	pko_command.s.subone0 = 1;

	pko_command.s.dontfree = 1;
	pko_command.s.reg0 = priv->fau + qos * 4;
	/*
	 * See if we can put this skb in the FPA pool. Any strange
	 * behavior from the Linux networking stack will most likely
	 * be caused by a bug in the following code. If some field is
	 * in use by the network stack and get carried over when a
	 * buffer is reused, bad thing may happen.  If in doubt and
	 * you dont need the absolute best performance, disable the
	 * define REUSE_SKBUFFS_WITHOUT_FREE. The reuse of buffers has
	 * shown a 25% increase in performance under some loads.
	 */
#if REUSE_SKBUFFS_WITHOUT_FREE
	fpa_head = skb->head + 256 - ((unsigned long)skb->head & 0x7f);
	if (unlikely(skb->data < fpa_head)) {
		/*
		 * printk("TX buffer beginning can't meet FPA
		 * alignment constraints\n");
		 */
		goto dont_put_skbuff_in_hw;
	}
	if (unlikely
	    ((skb_end_pointer(skb) - fpa_head) < CVMX_FPA_PACKET_POOL_SIZE)) {
		/*
		   printk("TX buffer isn't large enough for the FPA\n");
		 */
		goto dont_put_skbuff_in_hw;
	}
	if (unlikely(skb_shared(skb))) {
		/*
		   printk("TX buffer sharing data with someone else\n");
		 */
		goto dont_put_skbuff_in_hw;
	}
	if (unlikely(skb_cloned(skb))) {
		/*
		   printk("TX buffer has been cloned\n");
		 */
		goto dont_put_skbuff_in_hw;
	}
	if (unlikely(skb_header_cloned(skb))) {
		/*
		   printk("TX buffer header has been cloned\n");
		 */
		goto dont_put_skbuff_in_hw;
	}
	if (unlikely(skb->destructor)) {
		/*
		   printk("TX buffer has a destructor\n");
		 */
		goto dont_put_skbuff_in_hw;
	}
	if (unlikely(skb_shinfo(skb)->nr_frags)) {
		/*
		   printk("TX buffer has fragments\n");
		 */
		goto dont_put_skbuff_in_hw;
	}
	if (unlikely
	    (skb->truesize !=
	     sizeof(*skb) + skb_end_pointer(skb) - skb->head)) {
		/*
		   printk("TX buffer truesize has been changed\n");
		 */
		goto dont_put_skbuff_in_hw;
	}

	/*
	 * We can use this buffer in the FPA.  We don't need the FAU
	 * update anymore
	 */
	pko_command.s.reg0 = 0;
	pko_command.s.dontfree = 0;

	hw_buffer.s.back = ((unsigned long)skb->data >> 7) - ((unsigned long)fpa_head >> 7);
	*(struct sk_buff **)(fpa_head - sizeof(void *)) = skb;

	/*
	 * The skbuff will be reused without ever being freed. We must
	 * cleanup a bunch of core things.
	 */
	dst_release(skb_dst(skb));
	skb_dst_set(skb, NULL);
#ifdef CONFIG_XFRM
	secpath_put(skb->sp);
	skb->sp = NULL;
#endif
	nf_reset(skb);

#ifdef CONFIG_NET_SCHED
	skb->tc_index = 0;
#ifdef CONFIG_NET_CLS_ACT
	skb->tc_verd = 0;
#endif /* CONFIG_NET_CLS_ACT */
#endif /* CONFIG_NET_SCHED */
#endif /* REUSE_SKBUFFS_WITHOUT_FREE */

dont_put_skbuff_in_hw:

	/* Check if we can use the hardware checksumming */
	if (USE_HW_TCPUDP_CHECKSUM && (skb->protocol == htons(ETH_P_IP)) &&
	    (ip_hdr(skb)->version == 4) && (ip_hdr(skb)->ihl == 5) &&
	    ((ip_hdr(skb)->frag_off == 0) || (ip_hdr(skb)->frag_off == 1 << 14))
	    && ((ip_hdr(skb)->protocol == IP_PROTOCOL_TCP)
		|| (ip_hdr(skb)->protocol == IP_PROTOCOL_UDP))) {
		/* Use hardware checksum calc */
		pko_command.s.ipoffp1 = sizeof(struct ethhdr) + 1;
	}

	if (USE_ASYNC_IOBDMA) {
		/* Get the number of skbuffs in use by the hardware */
		CVMX_SYNCIOBDMA;
		skb_to_free = cvmx_scratch_read64(CVMX_SCR_SCRATCH);
		buffers_to_free = cvmx_scratch_read64(CVMX_SCR_SCRATCH + 8);
	} else {
		/* Get the number of skbuffs in use by the hardware */
		skb_to_free = cvmx_fau_fetch_and_add32(priv->fau + qos * 4,
						       MAX_SKB_TO_FREE);
		buffers_to_free =
		    cvmx_fau_fetch_and_add32(FAU_NUM_PACKET_BUFFERS_TO_FREE, 0);
	}

	skb_to_free = cvm_oct_adjust_skb_to_free(skb_to_free, priv->fau+qos*4);

	/*
	 * If we're sending faster than the receive can free them then
	 * don't do the HW free.
	 */
	if ((buffers_to_free < -100) && !pko_command.s.dontfree) {
		pko_command.s.dontfree = 1;
		pko_command.s.reg0 = priv->fau + qos * 4;
	}

	if (pko_command.s.dontfree)
		queue_type = QUEUE_CORE;
	else
		queue_type = QUEUE_HW;

	spin_lock_irqsave(&priv->tx_free_list[qos].lock, flags);

	/* Drop this packet if we have too many already queued to the HW */
	if (unlikely(skb_queue_len(&priv->tx_free_list[qos]) >= MAX_OUT_QUEUE_DEPTH)) {
		if (dev->tx_queue_len != 0) {
			/* Drop the lock when notifying the core.  */
			spin_unlock_irqrestore(&priv->tx_free_list[qos].lock, flags);
			netif_stop_queue(dev);
			hrtimer_start(&priv->tx_restart_timer,
				      priv->tx_restart_interval, HRTIMER_MODE_REL);
			spin_lock_irqsave(&priv->tx_free_list[qos].lock, flags);

		} else {
			/* If not using normal queueing.  */
			queue_type = QUEUE_DROP;
			goto skip_xmit;
		}
	}

	cvmx_pko_send_packet_prepare(priv->port, priv->queue + qos,
				     CVMX_PKO_LOCK_NONE);

	/* Send the packet to the output queue */
	if (unlikely(cvmx_pko_send_packet_finish(priv->port,
						 priv->queue + qos,
						 pko_command, hw_buffer,
						 CVMX_PKO_LOCK_NONE))) {
		DEBUGPRINT("%s: Failed to send the packet\n", dev->name);
		queue_type = QUEUE_DROP;
	}
skip_xmit:
	to_free_list = NULL;

	switch (queue_type) {
	case QUEUE_DROP:
		skb->next = to_free_list;
		to_free_list = skb;
		priv->stats.tx_dropped++;
		break;
	case QUEUE_HW:
		cvmx_fau_atomic_add32(FAU_NUM_PACKET_BUFFERS_TO_FREE, -1);
		break;
	case QUEUE_CORE:
		__skb_queue_tail(&priv->tx_free_list[qos], skb);
		break;
	default:
		BUG();
	}

	while (skb_to_free > 0) {
		struct sk_buff *t = __skb_dequeue(&priv->tx_free_list[qos]);
		t->next = to_free_list;
		to_free_list = t;
		skb_to_free--;
	}

	spin_unlock_irqrestore(&priv->tx_free_list[qos].lock, flags);

	/* Do the actual freeing outside of the lock. */
	while (to_free_list) {
		struct sk_buff *t = to_free_list;
		to_free_list = to_free_list->next;
		dev_kfree_skb_any(t);
	}

	if (USE_ASYNC_IOBDMA) {
		/* Restore the scratch area */
		cvmx_scratch_write64(CVMX_SCR_SCRATCH, old_scratch);
		cvmx_scratch_write64(CVMX_SCR_SCRATCH + 8, old_scratch2);
	}

	return NETDEV_TX_OK;
}

/**
 * Packet transmit to the POW
 *
 * @skb:    Packet to send
 * @dev:    Device info structure
 * Returns Always returns zero
 */
int cvm_oct_xmit_pow(struct sk_buff *skb, struct net_device *dev)
{
	struct octeon_ethernet *priv = netdev_priv(dev);
	void *packet_buffer;
	void *copy_location;

	/* Get a work queue entry */
	cvmx_wqe_t *work = cvmx_fpa_alloc(CVMX_FPA_WQE_POOL);
	if (unlikely(work == NULL)) {
		DEBUGPRINT("%s: Failed to allocate a work queue entry\n",
			   dev->name);
		priv->stats.tx_dropped++;
		dev_kfree_skb(skb);
		return 0;
	}

	/* Get a packet buffer */
	packet_buffer = cvmx_fpa_alloc(CVMX_FPA_PACKET_POOL);
	if (unlikely(packet_buffer == NULL)) {
		DEBUGPRINT("%s: Failed to allocate a packet buffer\n",
			   dev->name);
		cvmx_fpa_free(work, CVMX_FPA_WQE_POOL, DONT_WRITEBACK(1));
		priv->stats.tx_dropped++;
		dev_kfree_skb(skb);
		return 0;
	}

	/*
	 * Calculate where we need to copy the data to. We need to
	 * leave 8 bytes for a next pointer (unused). We also need to
	 * include any configure skip. Then we need to align the IP
	 * packet src and dest into the same 64bit word. The below
	 * calculation may add a little extra, but that doesn't
	 * hurt.
	 */
	copy_location = packet_buffer + sizeof(uint64_t);
	copy_location += ((CVMX_HELPER_FIRST_MBUFF_SKIP + 7) & 0xfff8) + 6;

	/*
	 * We have to copy the packet since whoever processes this
	 * packet will free it to a hardware pool. We can't use the
	 * trick of counting outstanding packets like in
	 * cvm_oct_xmit.
	 */
	memcpy(copy_location, skb->data, skb->len);

	/*
	 * Fill in some of the work queue fields. We may need to add
	 * more if the software at the other end needs them.
	 */
	work->hw_chksum = skb->csum;
	work->len = skb->len;
	work->ipprt = priv->port;
	work->qos = priv->port & 0x7;
	work->grp = pow_send_group;
	work->tag_type = CVMX_HELPER_INPUT_TAG_TYPE;
	work->tag = pow_send_group;	/* FIXME */
	/* Default to zero. Sets of zero later are commented out */
	work->word2.u64 = 0;
	work->word2.s.bufs = 1;
	work->packet_ptr.u64 = 0;
	work->packet_ptr.s.addr = cvmx_ptr_to_phys(copy_location);
	work->packet_ptr.s.pool = CVMX_FPA_PACKET_POOL;
	work->packet_ptr.s.size = CVMX_FPA_PACKET_POOL_SIZE;
	work->packet_ptr.s.back = (copy_location - packet_buffer) >> 7;

	if (skb->protocol == htons(ETH_P_IP)) {
		work->word2.s.ip_offset = 14;
#if 0
		work->word2.s.vlan_valid = 0;	/* FIXME */
		work->word2.s.vlan_cfi = 0;	/* FIXME */
		work->word2.s.vlan_id = 0;	/* FIXME */
		work->word2.s.dec_ipcomp = 0;	/* FIXME */
#endif
		work->word2.s.tcp_or_udp =
		    (ip_hdr(skb)->protocol == IP_PROTOCOL_TCP)
		    || (ip_hdr(skb)->protocol == IP_PROTOCOL_UDP);
#if 0
		/* FIXME */
		work->word2.s.dec_ipsec = 0;
		/* We only support IPv4 right now */
		work->word2.s.is_v6 = 0;
		/* Hardware would set to zero */
		work->word2.s.software = 0;
		/* No error, packet is internal */
		work->word2.s.L4_error = 0;
#endif
		work->word2.s.is_frag = !((ip_hdr(skb)->frag_off == 0)
					  || (ip_hdr(skb)->frag_off ==
					      1 << 14));
#if 0
		/* Assume Linux is sending a good packet */
		work->word2.s.IP_exc = 0;
#endif
		work->word2.s.is_bcast = (skb->pkt_type == PACKET_BROADCAST);
		work->word2.s.is_mcast = (skb->pkt_type == PACKET_MULTICAST);
#if 0
		/* This is an IP packet */
		work->word2.s.not_IP = 0;
		/* No error, packet is internal */
		work->word2.s.rcv_error = 0;
		/* No error, packet is internal */
		work->word2.s.err_code = 0;
#endif

		/*
		 * When copying the data, include 4 bytes of the
		 * ethernet header to align the same way hardware
		 * does.
		 */
		memcpy(work->packet_data, skb->data + 10,
		       sizeof(work->packet_data));
	} else {
#if 0
		work->word2.snoip.vlan_valid = 0;	/* FIXME */
		work->word2.snoip.vlan_cfi = 0;	/* FIXME */
		work->word2.snoip.vlan_id = 0;	/* FIXME */
		work->word2.snoip.software = 0;	/* Hardware would set to zero */
#endif
		work->word2.snoip.is_rarp = skb->protocol == htons(ETH_P_RARP);
		work->word2.snoip.is_arp = skb->protocol == htons(ETH_P_ARP);
		work->word2.snoip.is_bcast =
		    (skb->pkt_type == PACKET_BROADCAST);
		work->word2.snoip.is_mcast =
		    (skb->pkt_type == PACKET_MULTICAST);
		work->word2.snoip.not_IP = 1;	/* IP was done up above */
#if 0
		/* No error, packet is internal */
		work->word2.snoip.rcv_error = 0;
		/* No error, packet is internal */
		work->word2.snoip.err_code = 0;
#endif
		memcpy(work->packet_data, skb->data, sizeof(work->packet_data));
	}

	/* Submit the packet to the POW */
	cvmx_pow_work_submit(work, work->tag, work->tag_type, work->qos,
			     work->grp);
	priv->stats.tx_packets++;
	priv->stats.tx_bytes += skb->len;
	dev_kfree_skb(skb);
	return 0;
}

/**
 * This function frees all skb that are currently queued for TX.
 *
 * @dev:    Device being shutdown
 */
void cvm_oct_tx_shutdown(struct net_device *dev)
{
	struct octeon_ethernet *priv = netdev_priv(dev);
	unsigned long flags;
	int qos;

	for (qos = 0; qos < 16; qos++) {
		spin_lock_irqsave(&priv->tx_free_list[qos].lock, flags);
		while (skb_queue_len(&priv->tx_free_list[qos]))
			dev_kfree_skb_any(__skb_dequeue
					  (&priv->tx_free_list[qos]));
		spin_unlock_irqrestore(&priv->tx_free_list[qos].lock, flags);
	}
}