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
598
599
600
601
602
603
604
605
606
607
608
609
610
|
/***********************license start***************
* Author: Cavium Networks
*
* Contact: support@caviumnetworks.com
* This file is part of the OCTEON SDK
*
* Copyright (c) 2003-2008 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
***********************license end**************************************/
/**
*
* Interface to the hardware Packet Output unit.
*
* Starting with SDK 1.7.0, the PKO output functions now support
* two types of locking. CVMX_PKO_LOCK_ATOMIC_TAG continues to
* function similarly to previous SDKs by using POW atomic tags
* to preserve ordering and exclusivity. As a new option, you
* can now pass CVMX_PKO_LOCK_CMD_QUEUE which uses a ll/sc
* memory based locking instead. This locking has the advantage
* of not affecting the tag state but doesn't preserve packet
* ordering. CVMX_PKO_LOCK_CMD_QUEUE is appropriate in most
* generic code while CVMX_PKO_LOCK_CMD_QUEUE should be used
* with hand tuned fast path code.
*
* Some of other SDK differences visible to the command command
* queuing:
* - PKO indexes are no longer stored in the FAU. A large
* percentage of the FAU register block used to be tied up
* maintaining PKO queue pointers. These are now stored in a
* global named block.
* - The PKO <b>use_locking</b> parameter can now have a global
* effect. Since all application use the same named block,
* queue locking correctly applies across all operating
* systems when using CVMX_PKO_LOCK_CMD_QUEUE.
* - PKO 3 word commands are now supported. Use
* cvmx_pko_send_packet_finish3().
*
*/
#ifndef __CVMX_PKO_H__
#define __CVMX_PKO_H__
#include "cvmx-fpa.h"
#include "cvmx-pow.h"
#include "cvmx-cmd-queue.h"
#include "cvmx-pko-defs.h"
/* Adjust the command buffer size by 1 word so that in the case of using only
* two word PKO commands no command words stradle buffers. The useful values
* for this are 0 and 1. */
#define CVMX_PKO_COMMAND_BUFFER_SIZE_ADJUST (1)
#define CVMX_PKO_MAX_OUTPUT_QUEUES_STATIC 256
#define CVMX_PKO_MAX_OUTPUT_QUEUES ((OCTEON_IS_MODEL(OCTEON_CN31XX) || \
OCTEON_IS_MODEL(OCTEON_CN3010) || OCTEON_IS_MODEL(OCTEON_CN3005) || \
OCTEON_IS_MODEL(OCTEON_CN50XX)) ? 32 : \
(OCTEON_IS_MODEL(OCTEON_CN58XX) || \
OCTEON_IS_MODEL(OCTEON_CN56XX)) ? 256 : 128)
#define CVMX_PKO_NUM_OUTPUT_PORTS 40
/* use this for queues that are not used */
#define CVMX_PKO_MEM_QUEUE_PTRS_ILLEGAL_PID 63
#define CVMX_PKO_QUEUE_STATIC_PRIORITY 9
#define CVMX_PKO_ILLEGAL_QUEUE 0xFFFF
#define CVMX_PKO_MAX_QUEUE_DEPTH 0
typedef enum {
CVMX_PKO_SUCCESS,
CVMX_PKO_INVALID_PORT,
CVMX_PKO_INVALID_QUEUE,
CVMX_PKO_INVALID_PRIORITY,
CVMX_PKO_NO_MEMORY,
CVMX_PKO_PORT_ALREADY_SETUP,
CVMX_PKO_CMD_QUEUE_INIT_ERROR
} cvmx_pko_status_t;
/**
* This enumeration represents the differnet locking modes supported by PKO.
*/
typedef enum {
/*
* PKO doesn't do any locking. It is the responsibility of the
* application to make sure that no other core is accessing
* the same queue at the smae time
*/
CVMX_PKO_LOCK_NONE = 0,
/*
* PKO performs an atomic tagswitch to insure exclusive access
* to the output queue. This will maintain packet ordering on
* output.
*/
CVMX_PKO_LOCK_ATOMIC_TAG = 1,
/*
* PKO uses the common command queue locks to insure exclusive
* access to the output queue. This is a memory based
* ll/sc. This is the most portable locking mechanism.
*/
CVMX_PKO_LOCK_CMD_QUEUE = 2,
} cvmx_pko_lock_t;
typedef struct {
uint32_t packets;
uint64_t octets;
uint64_t doorbell;
} cvmx_pko_port_status_t;
/**
* This structure defines the address to use on a packet enqueue
*/
typedef union {
uint64_t u64;
struct {
/* Must CVMX_IO_SEG */
uint64_t mem_space:2;
/* Must be zero */
uint64_t reserved:13;
/* Must be one */
uint64_t is_io:1;
/* The ID of the device on the non-coherent bus */
uint64_t did:8;
/* Must be zero */
uint64_t reserved2:4;
/* Must be zero */
uint64_t reserved3:18;
/*
* The hardware likes to have the output port in
* addition to the output queue,
*/
uint64_t port:6;
/*
* The output queue to send the packet to (0-127 are
* legal)
*/
uint64_t queue:9;
/* Must be zero */
uint64_t reserved4:3;
} s;
} cvmx_pko_doorbell_address_t;
/**
* Structure of the first packet output command word.
*/
typedef union {
uint64_t u64;
struct {
/*
* The size of the reg1 operation - could be 8, 16,
* 32, or 64 bits.
*/
uint64_t size1:2;
/*
* The size of the reg0 operation - could be 8, 16,
* 32, or 64 bits.
*/
uint64_t size0:2;
/*
* If set, subtract 1, if clear, subtract packet
* size.
*/
uint64_t subone1:1;
/*
* The register, subtract will be done if reg1 is
* non-zero.
*/
uint64_t reg1:11;
/* If set, subtract 1, if clear, subtract packet size */
uint64_t subone0:1;
/* The register, subtract will be done if reg0 is non-zero */
uint64_t reg0:11;
/*
* When set, interpret segment pointer and segment
* bytes in little endian order.
*/
uint64_t le:1;
/*
* When set, packet data not allocated in L2 cache by
* PKO.
*/
uint64_t n2:1;
/*
* If set and rsp is set, word3 contains a pointer to
* a work queue entry.
*/
uint64_t wqp:1;
/* If set, the hardware will send a response when done */
uint64_t rsp:1;
/*
* If set, the supplied pkt_ptr is really a pointer to
* a list of pkt_ptr's.
*/
uint64_t gather:1;
/*
* If ipoffp1 is non zero, (ipoffp1-1) is the number
* of bytes to IP header, and the hardware will
* calculate and insert the UDP/TCP checksum.
*/
uint64_t ipoffp1:7;
/*
* If set, ignore the I bit (force to zero) from all
* pointer structures.
*/
uint64_t ignore_i:1;
/*
* If clear, the hardware will attempt to free the
* buffers containing the packet.
*/
uint64_t dontfree:1;
/*
* The total number of segs in the packet, if gather
* set, also gather list length.
*/
uint64_t segs:6;
/* Including L2, but no trailing CRC */
uint64_t total_bytes:16;
} s;
} cvmx_pko_command_word0_t;
/* CSR typedefs have been moved to cvmx-csr-*.h */
/**
* Definition of internal state for Packet output processing
*/
typedef struct {
/* ptr to start of buffer, offset kept in FAU reg */
uint64_t *start_ptr;
} cvmx_pko_state_elem_t;
/**
* Call before any other calls to initialize the packet
* output system.
*/
extern void cvmx_pko_initialize_global(void);
extern int cvmx_pko_initialize_local(void);
/**
* Enables the packet output hardware. It must already be
* configured.
*/
extern void cvmx_pko_enable(void);
/**
* Disables the packet output. Does not affect any configuration.
*/
extern void cvmx_pko_disable(void);
/**
* Shutdown and free resources required by packet output.
*/
extern void cvmx_pko_shutdown(void);
/**
* Configure a output port and the associated queues for use.
*
* @port: Port to configure.
* @base_queue: First queue number to associate with this port.
* @num_queues: Number of queues t oassociate with this port
* @priority: Array of priority levels for each queue. Values are
* allowed to be 1-8. A value of 8 get 8 times the traffic
* of a value of 1. There must be num_queues elements in the
* array.
*/
extern cvmx_pko_status_t cvmx_pko_config_port(uint64_t port,
uint64_t base_queue,
uint64_t num_queues,
const uint64_t priority[]);
/**
* Ring the packet output doorbell. This tells the packet
* output hardware that "len" command words have been added
* to its pending list. This command includes the required
* CVMX_SYNCWS before the doorbell ring.
*
* @port: Port the packet is for
* @queue: Queue the packet is for
* @len: Length of the command in 64 bit words
*/
static inline void cvmx_pko_doorbell(uint64_t port, uint64_t queue,
uint64_t len)
{
cvmx_pko_doorbell_address_t ptr;
ptr.u64 = 0;
ptr.s.mem_space = CVMX_IO_SEG;
ptr.s.did = CVMX_OCT_DID_PKT_SEND;
ptr.s.is_io = 1;
ptr.s.port = port;
ptr.s.queue = queue;
/*
* Need to make sure output queue data is in DRAM before
* doorbell write.
*/
CVMX_SYNCWS;
cvmx_write_io(ptr.u64, len);
}
/**
* Prepare to send a packet. This may initiate a tag switch to
* get exclusive access to the output queue structure, and
* performs other prep work for the packet send operation.
*
* cvmx_pko_send_packet_finish() MUST be called after this function is called,
* and must be called with the same port/queue/use_locking arguments.
*
* The use_locking parameter allows the caller to use three
* possible locking modes.
* - CVMX_PKO_LOCK_NONE
* - PKO doesn't do any locking. It is the responsibility
* of the application to make sure that no other core
* is accessing the same queue at the smae time.
* - CVMX_PKO_LOCK_ATOMIC_TAG
* - PKO performs an atomic tagswitch to insure exclusive
* access to the output queue. This will maintain
* packet ordering on output.
* - CVMX_PKO_LOCK_CMD_QUEUE
* - PKO uses the common command queue locks to insure
* exclusive access to the output queue. This is a
* memory based ll/sc. This is the most portable
* locking mechanism.
*
* NOTE: If atomic locking is used, the POW entry CANNOT be
* descheduled, as it does not contain a valid WQE pointer.
*
* @port: Port to send it on
* @queue: Queue to use
* @use_locking: CVMX_PKO_LOCK_NONE, CVMX_PKO_LOCK_ATOMIC_TAG, or
* CVMX_PKO_LOCK_CMD_QUEUE
*/
static inline void cvmx_pko_send_packet_prepare(uint64_t port, uint64_t queue,
cvmx_pko_lock_t use_locking)
{
if (use_locking == CVMX_PKO_LOCK_ATOMIC_TAG) {
/*
* Must do a full switch here to handle all cases. We
* use a fake WQE pointer, as the POW does not access
* this memory. The WQE pointer and group are only
* used if this work is descheduled, which is not
* supported by the
* cvmx_pko_send_packet_prepare/cvmx_pko_send_packet_finish
* combination. Note that this is a special case in
* which these fake values can be used - this is not a
* general technique.
*/
uint32_t tag =
CVMX_TAG_SW_BITS_INTERNAL << CVMX_TAG_SW_SHIFT |
CVMX_TAG_SUBGROUP_PKO << CVMX_TAG_SUBGROUP_SHIFT |
(CVMX_TAG_SUBGROUP_MASK & queue);
cvmx_pow_tag_sw_full((cvmx_wqe_t *) cvmx_phys_to_ptr(0x80), tag,
CVMX_POW_TAG_TYPE_ATOMIC, 0);
}
}
/**
* Complete packet output. cvmx_pko_send_packet_prepare() must be
* called exactly once before this, and the same parameters must be
* passed to both cvmx_pko_send_packet_prepare() and
* cvmx_pko_send_packet_finish().
*
* @port: Port to send it on
* @queue: Queue to use
* @pko_command:
* PKO HW command word
* @packet: Packet to send
* @use_locking: CVMX_PKO_LOCK_NONE, CVMX_PKO_LOCK_ATOMIC_TAG, or
* CVMX_PKO_LOCK_CMD_QUEUE
*
* Returns returns CVMX_PKO_SUCCESS on success, or error code on
* failure of output
*/
static inline cvmx_pko_status_t cvmx_pko_send_packet_finish(
uint64_t port,
uint64_t queue,
cvmx_pko_command_word0_t pko_command,
union cvmx_buf_ptr packet,
cvmx_pko_lock_t use_locking)
{
cvmx_cmd_queue_result_t result;
if (use_locking == CVMX_PKO_LOCK_ATOMIC_TAG)
cvmx_pow_tag_sw_wait();
result = cvmx_cmd_queue_write2(CVMX_CMD_QUEUE_PKO(queue),
(use_locking == CVMX_PKO_LOCK_CMD_QUEUE),
pko_command.u64, packet.u64);
if (likely(result == CVMX_CMD_QUEUE_SUCCESS)) {
cvmx_pko_doorbell(port, queue, 2);
return CVMX_PKO_SUCCESS;
}
|