summaryrefslogtreecommitdiff
path: root/net/xdp/xsk_queue.h
blob: 4a9663aa7afe6dc6c5e9ea3223263a42951ba9b3 (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
/* SPDX-License-Identifier: GPL-2.0 */
/* XDP user-space ring structure
 * Copyright(c) 2018 Intel Corporation.
 */

#ifndef _LINUX_XSK_QUEUE_H
#define _LINUX_XSK_QUEUE_H

#include <linux/types.h>
#include <linux/if_xdp.h>
#include <net/xdp_sock.h>
#include <net/xsk_buff_pool.h>

#include "xsk.h"

struct xdp_ring {
	u32 producer ____cacheline_aligned_in_smp;
	/* Hinder the adjacent cache prefetcher to prefetch the consumer
	 * pointer if the producer pointer is touched and vice versa.
	 */
	u32 pad1 ____cacheline_aligned_in_smp;
	u32 consumer ____cacheline_aligned_in_smp;
	u32 pad2 ____cacheline_aligned_in_smp;
	u32 flags;
	u32 pad3 ____cacheline_aligned_in_smp;
};

/* Used for the RX and TX queues for packets */
struct xdp_rxtx_ring {
	struct xdp_ring ptrs;
	struct xdp_desc desc[] ____cacheline_aligned_in_smp;
};

/* Used for the fill and completion queues for buffers */
struct xdp_umem_ring {
	struct xdp_ring ptrs;
	u64 desc[] ____cacheline_aligned_in_smp;
};

struct xsk_queue {
	u32 ring_mask;
	u32 nentries;
	u32 cached_prod;
	u32 cached_cons;
	struct xdp_ring *ring;
	u64 invalid_descs;
	u64 queue_empty_descs;
};

/* The structure of the shared state of the rings are the same as the
 * ring buffer in kernel/events/ring_buffer.c. For the Rx and completion
 * ring, the kernel is the producer and user space is the consumer. For
 * the Tx and fill rings, the kernel is the consumer and user space is
 * the producer.
 *
 * producer                         consumer
 *
 * if (LOAD ->consumer) {           LOAD ->producer
 *                    (A)           smp_rmb()       (C)
 *    STORE $data                   LOAD $data
 *    smp_wmb()       (B)           smp_mb()        (D)
 *    STORE ->producer              STORE ->consumer
 * }
 *
 * (A) pairs with (D), and (B) pairs with (C).
 *
 * Starting with (B), it protects the data from being written after
 * the producer pointer. If this barrier was missing, the consumer
 * could observe the producer pointer being set and thus load the data
 * before the producer has written the new data. The consumer would in
 * this case load the old data.
 *
 * (C) protects the consumer from speculatively loading the data before
 * the producer pointer actually has been read. If we do not have this
 * barrier, some architectures could load old data as speculative loads
 * are not discarded as the CPU does not know there is a dependency
 * between ->producer and data.
 *
 * (A) is a control dependency that separates the load of ->consumer
 * from the stores of $data. In case ->consumer indicates there is no
 * room in the buffer to store $data we do not. So no barrier is needed.
 *
 * (D) protects the load of the data to be observed to happen after the
 * store of the consumer pointer. If we did not have this memory
 * barrier, the producer could observe the consumer pointer being set
 * and overwrite the data with a new value before the consumer got the
 * chance to read the old value. The consumer would thus miss reading
 * the old entry and very likely read the new entry twice, once right
 * now and again after circling through the ring.
 */

/* The operations on the rings are the following:
 *
 * producer                           consumer
 *
 * RESERVE entries                    PEEK in the ring for entries
 * WRITE data into the ring           READ data from the ring
 * SUBMIT entries                     RELEASE entries
 *
 * The producer reserves one or more entries in the ring. It can then
 * fill in these entries and finally submit them so that they can be
 * seen and read by the consumer.
 *
 * The consumer peeks into the ring to see if the producer has written
 * any new entries. If so, the consumer can then read these entries
 * and when it is done reading them release them back to the producer
 * so that the producer can use these slots to fill in new entries.
 *
 * The function names below reflect these operations.
 */

/* Functions that read and validate content from consumer rings. */

static inline bool xskq_cons_read_addr_unchecked(struct xsk_queue *q, u64 *addr)
{
	struct xdp_umem_ring *ring = (struct xdp_umem_ring *)q->ring;

	if (q->cached_cons != q->cached_prod) {
		u32 idx = q->cached_cons & q->ring_mask;

		*addr = ring->desc[idx];
		return true;
	}

	return false;
}

static inline bool xp_aligned_validate_desc(struct xsk_buff_pool *pool,
					    struct xdp_desc *desc)
{
	u64 chunk, chunk_end;

	chunk = xp_aligned_extract_addr(pool, desc->addr);
	chunk_end = xp_aligned_extract_addr(pool, desc->addr + desc->len);
	if (chunk != chunk_end)
		return false;

	if (chunk >= pool->addrs_cnt)
		return false;

	if (desc->options)
		return false;
	return true;
}

static inline bool xp_unaligned_validate_desc(struct xsk_buff_pool *pool,
					      struct xdp_desc *desc)
{
	u64 addr, base_addr;

	base_addr = xp_unaligned_extract_addr(desc->addr);
	addr = xp_unaligned_add_offset_to_addr(desc->addr);

	if (desc->len > pool->chunk_size)
		return false;

	if (base_addr >= pool->addrs_cnt || addr >= pool->addrs_cnt ||
	    xp_desc_crosses_non_contig_pg(pool, addr, desc->len))
		return false;

	if (desc->options)
		return false;
	return true;
}

static inline bool xp_validate_desc(struct xsk_buff_pool *pool,
				    struct xdp_desc *desc)
{
	return pool->unaligned ? xp_unaligned_validate_desc(pool, desc) :
		xp_aligned_validate_desc(pool, desc);
}

static inline bool xskq_cons_is_valid_desc(struct xsk_queue *q,
					   struct xdp_desc *d,
					   struct xsk_buff_pool *pool)
{
	if (!xp_validate_desc(pool, d)) {
		q->invalid_descs++;
		return false;
	}
	return true;
}

static inline bool xskq_cons_read_desc(struct xsk_queue *q,
				       struct xdp_desc *desc,
				       struct xsk_buff_pool *pool)
{
	while (q->cached_cons != q->cached_prod) {
		struct xdp_rxtx_ring *ring = (struct xdp_rxtx_ring *)q->ring;
		u32 idx = q->cached_cons & q->ring_mask;

		*desc = ring->desc[idx];
		if (xskq_cons_is_valid_desc(q, desc, pool))
			return true;

		q->cached_cons++;
	}

	return false;
}

static inline u32 xskq_cons_read_desc_batch(struct xsk_queue *q,
					    struct xdp_desc *descs,
					    struct xsk_buff_pool *pool, u32 max)
{
	u32 cached_cons = q->cached_cons, nb_entries = 0;

	while (cached_cons != q->cached_prod && nb_entries < max) {
		struct xdp_rxtx_ring *ring = (struct xdp_rxtx_ring *)q->ring;
		u32 idx = cached_cons & q->ring_mask;

		descs[nb_entries] = ring->desc[idx];
		if (unlikely(!xskq_cons_is_valid_desc(q, &descs[nb_entries], pool))) {
			/* Skip the entry */
			cached_cons++;
			continue;
		}

		nb_entries++;
		cached_cons++;
	}

	return nb_entries;
}

/* Functions for consumers */

static inline void __xskq_cons_release(struct xsk_queue *q)
{
	smp_mb(); /* D, matches A */
	WRITE_ONCE(q->ring->consumer, q->cached_cons);
}

static inline void __xskq_cons_peek(struct xsk_queue *q)
{
	/* Refresh the local pointer */
	q->cached_prod = READ_ONCE(q->ring->producer);
	smp_rmb(); /* C, matches B */
}

static inline void xskq_cons_get_entries(struct xsk_queue *q)
{
	__xskq_cons_release(q);
	__xskq_cons_peek(q);
}

static inline u32 xskq_cons_nb_entries(struct xsk_queue *q, u32 max)
{
	u32 entries = q->cached_prod - q->cached_cons;

	if (entries >= max)
		return max;

	__xskq_cons_peek(q);
	entries = q->cached_prod - q->cached_cons;

	return entries >= max ? max : entries;
}

static inline bool xskq_cons_has_entries(struct xsk_queue *q, u32 cnt)
{
	return xskq_cons_nb_entries(q, cnt) >= cnt ? true : false;
}

static inline bool xskq_cons_peek_addr_unchecked(struct xsk_queue *q, u64 *addr)
{
	if (q->cached_prod == q->cached_cons)
		xskq_cons_get_entries(q);
	return xskq_cons_read_addr_unchecked(q, addr);
}

static inline bool xskq_cons_peek_desc(struct xsk_queue *q,
				       struct xdp_desc *desc,
				       struct xsk_buff_pool *pool)
{
	if (q->cached_prod == q->cached_cons)
		xskq_cons_get_entries(q);
	return xskq_cons_read_desc(q, desc, pool);
}

static inline u32 xskq_cons_peek_desc_batch(struct xsk_queue *q, struct xdp_desc *descs,
					    struct xsk_buff_pool *pool, u32 max)
{
	u32 entries = xskq_cons_nb_entries(q, max);

	return xskq_cons_read_desc_batch(q, descs, pool, entries);
}

/* To improve performance in the xskq_cons_release functions, only update local state here.
 * Reflect this to global state when we get new entries from the ring in
 * xskq_cons_get_entries() and whenever Rx or Tx processing are completed in the NAPI loop.
 */
static inline void xskq_cons_release(struct xsk_queue *q)
{
	q->cached_cons++;
}

static inline void xskq_cons_release_n(struct xsk_queue *q, u32 cnt)
{
	q->cached_cons += cnt;
}

static inline bool xskq_cons_is_full(struct xsk_queue *q)
{
	/* No barriers needed since data is not accessed */
	return READ_ONCE(q->ring->producer) - READ_ONCE(q->ring->consumer) ==
		q->nentries;
}

static inline u32 xskq_cons_present_entries(struct xsk_queue *q)
{
	/* No barriers needed since data is not accessed */
	return READ_ONCE(q->ring->producer) - READ_ONCE(q->ring->consumer);
}

/* Functions for producers */

static inline u32 xskq_prod_nb_free(struct xsk_queue *q, u32 max)
{
	u32 free_entries = q->nentries - (q->cached_prod - q->cached_cons);

	if (free_entries >= max)
		return max;

	/* Refresh the local tail pointer */
	q->cached_cons = READ_ONCE(q->ring->consumer);
	free_entries = q->nentries - (q->cached_prod - q->cached_cons);

	return free_entries >= max ? max : free_entries;
}

static inline bool xskq_prod_is_full(struct xsk_queue *q)
{
	return xskq_prod_nb_free(q, 1) ? false : true;
}

static inline int xskq_prod_reserve(struct xsk_queue *q)
{
	if (xskq_prod_is_full(q))
		return -ENOSPC;

	/* A, matches D */
	q->cached_prod++;
	return 0;
}

static inline int xskq_prod_reserve_addr(struct xsk_queue *q, u64 addr)
{
	struct xdp_umem_ring *ring = (struct xdp_umem_ring *)q->ring;

	if (xskq_prod_is_full(q))
		return -ENOSPC;

	/* A, matches D */
	ring->desc[q->cached_prod++ & q->ring_mask] = addr;
	return 0;
}

static inline u32 xskq_prod_reserve_addr_batch(struct xsk_queue *q, struct xdp_desc *descs,
					       u32 max)
{
	struct xdp_umem_ring *ring = (struct xdp_umem_ring *)q->ring;
	u32 nb_entries, i, cached_prod;

	nb_entries = xskq_prod_nb_free(q, max);

	/* A, matches D */
	cached_prod = q->cached_prod;
	for (i = 0; i < nb_entries; i++)
		ring->desc[cached_prod++ & q->ring_mask] = descs[i].addr;
	q->cached_prod = cached_prod;

	return nb_entries;
}

static inline int xskq_prod_reserve_desc(struct xsk_queue *q,
					 u64 addr, u32 len)
{
	struct xdp_rxtx_ring *ring = (struct xdp_rxtx_ring *)q->ring;
	u32 idx;

	if (xskq_prod_is_full(q))
		return -ENOSPC;

	/* A, matches D */
	idx = q->cached_prod++ & q->ring_mask;
	ring->desc[idx].addr = addr;
	ring->desc[idx].len = len;

	return 0;
}

static inline void __xskq_prod_submit(struct xsk_queue *q, u32 idx)
{
	smp_wmb(); /* B, matches C */

	WRITE_ONCE(q->ring->producer, idx);
}

static inline void xskq_prod_submit(struct xsk_queue *q)
{
	__xskq_prod_submit(q, q->cached_prod);
}

static inline void xskq_prod_submit_addr(struct xsk_queue *q, u64 addr)
{
	struct xdp_umem_ring *ring = (struct xdp_umem_ring *)q->ring;
	u32 idx = q->ring->producer;

	ring->desc[idx++ & q->ring_mask] = addr;

	__xskq_prod_submit(q, idx);
}

static inline void xskq_prod_submit_n(struct xsk_queue *q, u32 nb_entries)
{
	__xskq_prod_submit(q, q->ring->producer + nb_entries);
}

static inline bool xskq_prod_is_empty(struct xsk_queue *q)
{
	/* No barriers needed since data is not accessed */
	return READ_ONCE(q->ring->consumer) == READ_ONCE(q->ring->producer);
}

/* For both producers and consumers */

static inline u64 xskq_nb_invalid_descs(struct xsk_queue *q)
{
	return q ? q->invalid_descs : 0;
}

static inline u64 xskq_nb_queue_empty_descs(struct xsk_queue *q)
{
	return q ? q->queue_empty_descs : 0;
}

struct xsk_queue *xskq_create(u32 nentries, bool umem_queue);
void xskq_destroy(struct xsk_queue *q_ops);

#endif /* _LINUX_XSK_QUEUE_H */