xref: /linux/tools/lib/bpf/ringbuf.c (revision 9e56ff53b4115875667760445b028357848b4748)
1 // SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause)
2 /*
3  * Ring buffer operations.
4  *
5  * Copyright (C) 2020 Facebook, Inc.
6  */
7 #ifndef _GNU_SOURCE
8 #define _GNU_SOURCE
9 #endif
10 #include <stdlib.h>
11 #include <stdio.h>
12 #include <errno.h>
13 #include <unistd.h>
14 #include <linux/err.h>
15 #include <linux/bpf.h>
16 #include <asm/barrier.h>
17 #include <sys/mman.h>
18 #include <sys/epoll.h>
19 #include <time.h>
20 
21 #include "libbpf.h"
22 #include "libbpf_internal.h"
23 #include "bpf.h"
24 
25 struct ring {
26 	ring_buffer_sample_fn sample_cb;
27 	void *ctx;
28 	void *data;
29 	unsigned long *consumer_pos;
30 	unsigned long *producer_pos;
31 	unsigned long mask;
32 	int map_fd;
33 };
34 
35 struct ring_buffer {
36 	struct epoll_event *events;
37 	struct ring **rings;
38 	size_t page_size;
39 	int epoll_fd;
40 	int ring_cnt;
41 };
42 
43 struct user_ring_buffer {
44 	struct epoll_event event;
45 	unsigned long *consumer_pos;
46 	unsigned long *producer_pos;
47 	void *data;
48 	unsigned long mask;
49 	size_t page_size;
50 	int map_fd;
51 	int epoll_fd;
52 };
53 
54 /* 8-byte ring buffer header structure */
55 struct ringbuf_hdr {
56 	__u32 len;
57 	__u32 pad;
58 };
59 
60 static void ringbuf_free_ring(struct ring_buffer *rb, struct ring *r)
61 {
62 	if (r->consumer_pos) {
63 		munmap(r->consumer_pos, rb->page_size);
64 		r->consumer_pos = NULL;
65 	}
66 	if (r->producer_pos) {
67 		munmap(r->producer_pos, rb->page_size + 2 * (r->mask + 1));
68 		r->producer_pos = NULL;
69 	}
70 
71 	free(r);
72 }
73 
74 /* Add extra RINGBUF maps to this ring buffer manager */
75 int ring_buffer__add(struct ring_buffer *rb, int map_fd,
76 		     ring_buffer_sample_fn sample_cb, void *ctx)
77 {
78 	struct bpf_map_info info;
79 	__u32 len = sizeof(info);
80 	struct epoll_event *e;
81 	struct ring *r;
82 	__u64 mmap_sz;
83 	void *tmp;
84 	int err;
85 
86 	memset(&info, 0, sizeof(info));
87 
88 	err = bpf_map_get_info_by_fd(map_fd, &info, &len);
89 	if (err) {
90 		err = -errno;
91 		pr_warn("ringbuf: failed to get map info for fd=%d: %d\n",
92 			map_fd, err);
93 		return libbpf_err(err);
94 	}
95 
96 	if (info.type != BPF_MAP_TYPE_RINGBUF) {
97 		pr_warn("ringbuf: map fd=%d is not BPF_MAP_TYPE_RINGBUF\n",
98 			map_fd);
99 		return libbpf_err(-EINVAL);
100 	}
101 
102 	tmp = libbpf_reallocarray(rb->rings, rb->ring_cnt + 1, sizeof(*rb->rings));
103 	if (!tmp)
104 		return libbpf_err(-ENOMEM);
105 	rb->rings = tmp;
106 
107 	tmp = libbpf_reallocarray(rb->events, rb->ring_cnt + 1, sizeof(*rb->events));
108 	if (!tmp)
109 		return libbpf_err(-ENOMEM);
110 	rb->events = tmp;
111 
112 	r = calloc(1, sizeof(*r));
113 	if (!r)
114 		return libbpf_err(-ENOMEM);
115 	rb->rings[rb->ring_cnt] = r;
116 
117 	r->map_fd = map_fd;
118 	r->sample_cb = sample_cb;
119 	r->ctx = ctx;
120 	r->mask = info.max_entries - 1;
121 
122 	/* Map writable consumer page */
123 	tmp = mmap(NULL, rb->page_size, PROT_READ | PROT_WRITE, MAP_SHARED, map_fd, 0);
124 	if (tmp == MAP_FAILED) {
125 		err = -errno;
126 		pr_warn("ringbuf: failed to mmap consumer page for map fd=%d: %d\n",
127 			map_fd, err);
128 		goto err_out;
129 	}
130 	r->consumer_pos = tmp;
131 
132 	/* Map read-only producer page and data pages. We map twice as big
133 	 * data size to allow simple reading of samples that wrap around the
134 	 * end of a ring buffer. See kernel implementation for details.
135 	 */
136 	mmap_sz = rb->page_size + 2 * (__u64)info.max_entries;
137 	if (mmap_sz != (__u64)(size_t)mmap_sz) {
138 		err = -E2BIG;
139 		pr_warn("ringbuf: ring buffer size (%u) is too big\n", info.max_entries);
140 		goto err_out;
141 	}
142 	tmp = mmap(NULL, (size_t)mmap_sz, PROT_READ, MAP_SHARED, map_fd, rb->page_size);
143 	if (tmp == MAP_FAILED) {
144 		err = -errno;
145 		pr_warn("ringbuf: failed to mmap data pages for map fd=%d: %d\n",
146 			map_fd, err);
147 		goto err_out;
148 	}
149 	r->producer_pos = tmp;
150 	r->data = tmp + rb->page_size;
151 
152 	e = &rb->events[rb->ring_cnt];
153 	memset(e, 0, sizeof(*e));
154 
155 	e->events = EPOLLIN;
156 	e->data.fd = rb->ring_cnt;
157 	if (epoll_ctl(rb->epoll_fd, EPOLL_CTL_ADD, map_fd, e) < 0) {
158 		err = -errno;
159 		pr_warn("ringbuf: failed to epoll add map fd=%d: %d\n",
160 			map_fd, err);
161 		goto err_out;
162 	}
163 
164 	rb->ring_cnt++;
165 	return 0;
166 
167 err_out:
168 	ringbuf_free_ring(rb, r);
169 	return libbpf_err(err);
170 }
171 
172 void ring_buffer__free(struct ring_buffer *rb)
173 {
174 	int i;
175 
176 	if (!rb)
177 		return;
178 
179 	for (i = 0; i < rb->ring_cnt; ++i)
180 		ringbuf_free_ring(rb, rb->rings[i]);
181 	if (rb->epoll_fd >= 0)
182 		close(rb->epoll_fd);
183 
184 	free(rb->events);
185 	free(rb->rings);
186 	free(rb);
187 }
188 
189 struct ring_buffer *
190 ring_buffer__new(int map_fd, ring_buffer_sample_fn sample_cb, void *ctx,
191 		 const struct ring_buffer_opts *opts)
192 {
193 	struct ring_buffer *rb;
194 	int err;
195 
196 	if (!OPTS_VALID(opts, ring_buffer_opts))
197 		return errno = EINVAL, NULL;
198 
199 	rb = calloc(1, sizeof(*rb));
200 	if (!rb)
201 		return errno = ENOMEM, NULL;
202 
203 	rb->page_size = getpagesize();
204 
205 	rb->epoll_fd = epoll_create1(EPOLL_CLOEXEC);
206 	if (rb->epoll_fd < 0) {
207 		err = -errno;
208 		pr_warn("ringbuf: failed to create epoll instance: %d\n", err);
209 		goto err_out;
210 	}
211 
212 	err = ring_buffer__add(rb, map_fd, sample_cb, ctx);
213 	if (err)
214 		goto err_out;
215 
216 	return rb;
217 
218 err_out:
219 	ring_buffer__free(rb);
220 	return errno = -err, NULL;
221 }
222 
223 static inline int roundup_len(__u32 len)
224 {
225 	/* clear out top 2 bits (discard and busy, if set) */
226 	len <<= 2;
227 	len >>= 2;
228 	/* add length prefix */
229 	len += BPF_RINGBUF_HDR_SZ;
230 	/* round up to 8 byte alignment */
231 	return (len + 7) / 8 * 8;
232 }
233 
234 static int64_t ringbuf_process_ring(struct ring *r)
235 {
236 	int *len_ptr, len, err;
237 	/* 64-bit to avoid overflow in case of extreme application behavior */
238 	int64_t cnt = 0;
239 	unsigned long cons_pos, prod_pos;
240 	bool got_new_data;
241 	void *sample;
242 
243 	cons_pos = smp_load_acquire(r->consumer_pos);
244 	do {
245 		got_new_data = false;
246 		prod_pos = smp_load_acquire(r->producer_pos);
247 		while (cons_pos < prod_pos) {
248 			len_ptr = r->data + (cons_pos & r->mask);
249 			len = smp_load_acquire(len_ptr);
250 
251 			/* sample not committed yet, bail out for now */
252 			if (len & BPF_RINGBUF_BUSY_BIT)
253 				goto done;
254 
255 			got_new_data = true;
256 			cons_pos += roundup_len(len);
257 
258 			if ((len & BPF_RINGBUF_DISCARD_BIT) == 0) {
259 				sample = (void *)len_ptr + BPF_RINGBUF_HDR_SZ;
260 				err = r->sample_cb(r->ctx, sample, len);
261 				if (err < 0) {
262 					/* update consumer pos and bail out */
263 					smp_store_release(r->consumer_pos,
264 							  cons_pos);
265 					return err;
266 				}
267 				cnt++;
268 			}
269 
270 			smp_store_release(r->consumer_pos, cons_pos);
271 		}
272 	} while (got_new_data);
273 done:
274 	return cnt;
275 }
276 
277 /* Consume available ring buffer(s) data without event polling.
278  * Returns number of records consumed across all registered ring buffers (or
279  * INT_MAX, whichever is less), or negative number if any of the callbacks
280  * return error.
281  */
282 int ring_buffer__consume(struct ring_buffer *rb)
283 {
284 	int64_t err, res = 0;
285 	int i;
286 
287 	for (i = 0; i < rb->ring_cnt; i++) {
288 		struct ring *ring = rb->rings[i];
289 
290 		err = ringbuf_process_ring(ring);
291 		if (err < 0)
292 			return libbpf_err(err);
293 		res += err;
294 	}
295 	if (res > INT_MAX)
296 		return INT_MAX;
297 	return res;
298 }
299 
300 /* Poll for available data and consume records, if any are available.
301  * Returns number of records consumed (or INT_MAX, whichever is less), or
302  * negative number, if any of the registered callbacks returned error.
303  */
304 int ring_buffer__poll(struct ring_buffer *rb, int timeout_ms)
305 {
306 	int i, cnt;
307 	int64_t err, res = 0;
308 
309 	cnt = epoll_wait(rb->epoll_fd, rb->events, rb->ring_cnt, timeout_ms);
310 	if (cnt < 0)
311 		return libbpf_err(-errno);
312 
313 	for (i = 0; i < cnt; i++) {
314 		__u32 ring_id = rb->events[i].data.fd;
315 		struct ring *ring = rb->rings[ring_id];
316 
317 		err = ringbuf_process_ring(ring);
318 		if (err < 0)
319 			return libbpf_err(err);
320 		res += err;
321 	}
322 	if (res > INT_MAX)
323 		return INT_MAX;
324 	return res;
325 }
326 
327 /* Get an fd that can be used to sleep until data is available in the ring(s) */
328 int ring_buffer__epoll_fd(const struct ring_buffer *rb)
329 {
330 	return rb->epoll_fd;
331 }
332 
333 struct ring *ring_buffer__ring(struct ring_buffer *rb, unsigned int idx)
334 {
335 	if (idx >= rb->ring_cnt)
336 		return errno = ERANGE, NULL;
337 
338 	return rb->rings[idx];
339 }
340 
341 unsigned long ring__consumer_pos(const struct ring *r)
342 {
343 	/* Synchronizes with smp_store_release() in ringbuf_process_ring(). */
344 	return smp_load_acquire(r->consumer_pos);
345 }
346 
347 unsigned long ring__producer_pos(const struct ring *r)
348 {
349 	/* Synchronizes with smp_store_release() in __bpf_ringbuf_reserve() in
350 	 * the kernel.
351 	 */
352 	return smp_load_acquire(r->producer_pos);
353 }
354 
355 size_t ring__avail_data_size(const struct ring *r)
356 {
357 	unsigned long cons_pos, prod_pos;
358 
359 	cons_pos = ring__consumer_pos(r);
360 	prod_pos = ring__producer_pos(r);
361 	return prod_pos - cons_pos;
362 }
363 
364 size_t ring__size(const struct ring *r)
365 {
366 	return r->mask + 1;
367 }
368 
369 int ring__map_fd(const struct ring *r)
370 {
371 	return r->map_fd;
372 }
373 
374 int ring__consume(struct ring *r)
375 {
376 	int64_t res;
377 
378 	res = ringbuf_process_ring(r);
379 	if (res < 0)
380 		return libbpf_err(res);
381 
382 	return res > INT_MAX ? INT_MAX : res;
383 }
384 
385 static void user_ringbuf_unmap_ring(struct user_ring_buffer *rb)
386 {
387 	if (rb->consumer_pos) {
388 		munmap(rb->consumer_pos, rb->page_size);
389 		rb->consumer_pos = NULL;
390 	}
391 	if (rb->producer_pos) {
392 		munmap(rb->producer_pos, rb->page_size + 2 * (rb->mask + 1));
393 		rb->producer_pos = NULL;
394 	}
395 }
396 
397 void user_ring_buffer__free(struct user_ring_buffer *rb)
398 {
399 	if (!rb)
400 		return;
401 
402 	user_ringbuf_unmap_ring(rb);
403 
404 	if (rb->epoll_fd >= 0)
405 		close(rb->epoll_fd);
406 
407 	free(rb);
408 }
409 
410 static int user_ringbuf_map(struct user_ring_buffer *rb, int map_fd)
411 {
412 	struct bpf_map_info info;
413 	__u32 len = sizeof(info);
414 	__u64 mmap_sz;
415 	void *tmp;
416 	struct epoll_event *rb_epoll;
417 	int err;
418 
419 	memset(&info, 0, sizeof(info));
420 
421 	err = bpf_map_get_info_by_fd(map_fd, &info, &len);
422 	if (err) {
423 		err = -errno;
424 		pr_warn("user ringbuf: failed to get map info for fd=%d: %d\n", map_fd, err);
425 		return err;
426 	}
427 
428 	if (info.type != BPF_MAP_TYPE_USER_RINGBUF) {
429 		pr_warn("user ringbuf: map fd=%d is not BPF_MAP_TYPE_USER_RINGBUF\n", map_fd);
430 		return -EINVAL;
431 	}
432 
433 	rb->map_fd = map_fd;
434 	rb->mask = info.max_entries - 1;
435 
436 	/* Map read-only consumer page */
437 	tmp = mmap(NULL, rb->page_size, PROT_READ, MAP_SHARED, map_fd, 0);
438 	if (tmp == MAP_FAILED) {
439 		err = -errno;
440 		pr_warn("user ringbuf: failed to mmap consumer page for map fd=%d: %d\n",
441 			map_fd, err);
442 		return err;
443 	}
444 	rb->consumer_pos = tmp;
445 
446 	/* Map read-write the producer page and data pages. We map the data
447 	 * region as twice the total size of the ring buffer to allow the
448 	 * simple reading and writing of samples that wrap around the end of
449 	 * the buffer.  See the kernel implementation for details.
450 	 */
451 	mmap_sz = rb->page_size + 2 * (__u64)info.max_entries;
452 	if (mmap_sz != (__u64)(size_t)mmap_sz) {
453 		pr_warn("user ringbuf: ring buf size (%u) is too big\n", info.max_entries);
454 		return -E2BIG;
455 	}
456 	tmp = mmap(NULL, (size_t)mmap_sz, PROT_READ | PROT_WRITE, MAP_SHARED,
457 		   map_fd, rb->page_size);
458 	if (tmp == MAP_FAILED) {
459 		err = -errno;
460 		pr_warn("user ringbuf: failed to mmap data pages for map fd=%d: %d\n",
461 			map_fd, err);
462 		return err;
463 	}
464 
465 	rb->producer_pos = tmp;
466 	rb->data = tmp + rb->page_size;
467 
468 	rb_epoll = &rb->event;
469 	rb_epoll->events = EPOLLOUT;
470 	if (epoll_ctl(rb->epoll_fd, EPOLL_CTL_ADD, map_fd, rb_epoll) < 0) {
471 		err = -errno;
472 		pr_warn("user ringbuf: failed to epoll add map fd=%d: %d\n", map_fd, err);
473 		return err;
474 	}
475 
476 	return 0;
477 }
478 
479 struct user_ring_buffer *
480 user_ring_buffer__new(int map_fd, const struct user_ring_buffer_opts *opts)
481 {
482 	struct user_ring_buffer *rb;
483 	int err;
484 
485 	if (!OPTS_VALID(opts, user_ring_buffer_opts))
486 		return errno = EINVAL, NULL;
487 
488 	rb = calloc(1, sizeof(*rb));
489 	if (!rb)
490 		return errno = ENOMEM, NULL;
491 
492 	rb->page_size = getpagesize();
493 
494 	rb->epoll_fd = epoll_create1(EPOLL_CLOEXEC);
495 	if (rb->epoll_fd < 0) {
496 		err = -errno;
497 		pr_warn("user ringbuf: failed to create epoll instance: %d\n", err);
498 		goto err_out;
499 	}
500 
501 	err = user_ringbuf_map(rb, map_fd);
502 	if (err)
503 		goto err_out;
504 
505 	return rb;
506 
507 err_out:
508 	user_ring_buffer__free(rb);
509 	return errno = -err, NULL;
510 }
511 
512 static void user_ringbuf_commit(struct user_ring_buffer *rb, void *sample, bool discard)
513 {
514 	__u32 new_len;
515 	struct ringbuf_hdr *hdr;
516 	uintptr_t hdr_offset;
517 
518 	hdr_offset = rb->mask + 1 + (sample - rb->data) - BPF_RINGBUF_HDR_SZ;
519 	hdr = rb->data + (hdr_offset & rb->mask);
520 
521 	new_len = hdr->len & ~BPF_RINGBUF_BUSY_BIT;
522 	if (discard)
523 		new_len |= BPF_RINGBUF_DISCARD_BIT;
524 
525 	/* Synchronizes with smp_load_acquire() in __bpf_user_ringbuf_peek() in
526 	 * the kernel.
527 	 */
528 	__atomic_exchange_n(&hdr->len, new_len, __ATOMIC_ACQ_REL);
529 }
530 
531 void user_ring_buffer__discard(struct user_ring_buffer *rb, void *sample)
532 {
533 	user_ringbuf_commit(rb, sample, true);
534 }
535 
536 void user_ring_buffer__submit(struct user_ring_buffer *rb, void *sample)
537 {
538 	user_ringbuf_commit(rb, sample, false);
539 }
540 
541 void *user_ring_buffer__reserve(struct user_ring_buffer *rb, __u32 size)
542 {
543 	__u32 avail_size, total_size, max_size;
544 	/* 64-bit to avoid overflow in case of extreme application behavior */
545 	__u64 cons_pos, prod_pos;
546 	struct ringbuf_hdr *hdr;
547 
548 	/* The top two bits are used as special flags */
549 	if (size & (BPF_RINGBUF_BUSY_BIT | BPF_RINGBUF_DISCARD_BIT))
550 		return errno = E2BIG, NULL;
551 
552 	/* Synchronizes with smp_store_release() in __bpf_user_ringbuf_peek() in
553 	 * the kernel.
554 	 */
555 	cons_pos = smp_load_acquire(rb->consumer_pos);
556 	/* Synchronizes with smp_store_release() in user_ringbuf_commit() */
557 	prod_pos = smp_load_acquire(rb->producer_pos);
558 
559 	max_size = rb->mask + 1;
560 	avail_size = max_size - (prod_pos - cons_pos);
561 	/* Round up total size to a multiple of 8. */
562 	total_size = (size + BPF_RINGBUF_HDR_SZ + 7) / 8 * 8;
563 
564 	if (total_size > max_size)
565 		return errno = E2BIG, NULL;
566 
567 	if (avail_size < total_size)
568 		return errno = ENOSPC, NULL;
569 
570 	hdr = rb->data + (prod_pos & rb->mask);
571 	hdr->len = size | BPF_RINGBUF_BUSY_BIT;
572 	hdr->pad = 0;
573 
574 	/* Synchronizes with smp_load_acquire() in __bpf_user_ringbuf_peek() in
575 	 * the kernel.
576 	 */
577 	smp_store_release(rb->producer_pos, prod_pos + total_size);
578 
579 	return (void *)rb->data + ((prod_pos + BPF_RINGBUF_HDR_SZ) & rb->mask);
580 }
581 
582 static __u64 ns_elapsed_timespec(const struct timespec *start, const struct timespec *end)
583 {
584 	__u64 start_ns, end_ns, ns_per_s = 1000000000;
585 
586 	start_ns = (__u64)start->tv_sec * ns_per_s + start->tv_nsec;
587 	end_ns = (__u64)end->tv_sec * ns_per_s + end->tv_nsec;
588 
589 	return end_ns - start_ns;
590 }
591 
592 void *user_ring_buffer__reserve_blocking(struct user_ring_buffer *rb, __u32 size, int timeout_ms)
593 {
594 	void *sample;
595 	int err, ms_remaining = timeout_ms;
596 	struct timespec start;
597 
598 	if (timeout_ms < 0 && timeout_ms != -1)
599 		return errno = EINVAL, NULL;
600 
601 	if (timeout_ms != -1) {
602 		err = clock_gettime(CLOCK_MONOTONIC, &start);
603 		if (err)
604 			return NULL;
605 	}
606 
607 	do {
608 		int cnt, ms_elapsed;
609 		struct timespec curr;
610 		__u64 ns_per_ms = 1000000;
611 
612 		sample = user_ring_buffer__reserve(rb, size);
613 		if (sample)
614 			return sample;
615 		else if (errno != ENOSPC)
616 			return NULL;
617 
618 		/* The kernel guarantees at least one event notification
619 		 * delivery whenever at least one sample is drained from the
620 		 * ring buffer in an invocation to bpf_ringbuf_drain(). Other
621 		 * additional events may be delivered at any time, but only one
622 		 * event is guaranteed per bpf_ringbuf_drain() invocation,
623 		 * provided that a sample is drained, and the BPF program did
624 		 * not pass BPF_RB_NO_WAKEUP to bpf_ringbuf_drain(). If
625 		 * BPF_RB_FORCE_WAKEUP is passed to bpf_ringbuf_drain(), a
626 		 * wakeup event will be delivered even if no samples are
627 		 * drained.
628 		 */
629 		cnt = epoll_wait(rb->epoll_fd, &rb->event, 1, ms_remaining);
630 		if (cnt < 0)
631 			return NULL;
632 
633 		if (timeout_ms == -1)
634 			continue;
635 
636 		err = clock_gettime(CLOCK_MONOTONIC, &curr);
637 		if (err)
638 			return NULL;
639 
640 		ms_elapsed = ns_elapsed_timespec(&start, &curr) / ns_per_ms;
641 		ms_remaining = timeout_ms - ms_elapsed;
642 	} while (ms_remaining > 0);
643 
644 	/* Try one more time to reserve a sample after the specified timeout has elapsed. */
645 	return user_ring_buffer__reserve(rb, size);
646 }
647