1 #include <linux/bpf.h>
2 #include <linux/btf.h>
3 #include <linux/err.h>
4 #include <linux/irq_work.h>
5 #include <linux/slab.h>
6 #include <linux/filter.h>
7 #include <linux/mm.h>
8 #include <linux/vmalloc.h>
9 #include <linux/wait.h>
10 #include <linux/poll.h>
11 #include <linux/kmemleak.h>
12 #include <uapi/linux/btf.h>
13 #include <linux/btf_ids.h>
14
15 #define RINGBUF_CREATE_FLAG_MASK (BPF_F_NUMA_NODE)
16
17 /* non-mmap()'able part of bpf_ringbuf (everything up to consumer page) */
18 #define RINGBUF_PGOFF \
19 (offsetof(struct bpf_ringbuf, consumer_pos) >> PAGE_SHIFT)
20 /* consumer page and producer page */
21 #define RINGBUF_POS_PAGES 2
22 #define RINGBUF_NR_META_PAGES (RINGBUF_PGOFF + RINGBUF_POS_PAGES)
23
24 #define RINGBUF_MAX_RECORD_SZ (UINT_MAX/4)
25
26 struct bpf_ringbuf {
27 wait_queue_head_t waitq;
28 struct irq_work work;
29 u64 mask;
30 struct page **pages;
31 int nr_pages;
32 spinlock_t spinlock ____cacheline_aligned_in_smp;
33 /* For user-space producer ring buffers, an atomic_t busy bit is used
34 * to synchronize access to the ring buffers in the kernel, rather than
35 * the spinlock that is used for kernel-producer ring buffers. This is
36 * done because the ring buffer must hold a lock across a BPF program's
37 * callback:
38 *
39 * __bpf_user_ringbuf_peek() // lock acquired
40 * -> program callback_fn()
41 * -> __bpf_user_ringbuf_sample_release() // lock released
42 *
43 * It is unsafe and incorrect to hold an IRQ spinlock across what could
44 * be a long execution window, so we instead simply disallow concurrent
45 * access to the ring buffer by kernel consumers, and return -EBUSY from
46 * __bpf_user_ringbuf_peek() if the busy bit is held by another task.
47 */
48 atomic_t busy ____cacheline_aligned_in_smp;
49 /* Consumer and producer counters are put into separate pages to
50 * allow each position to be mapped with different permissions.
51 * This prevents a user-space application from modifying the
52 * position and ruining in-kernel tracking. The permissions of the
53 * pages depend on who is producing samples: user-space or the
54 * kernel. Note that the pending counter is placed in the same
55 * page as the producer, so that it shares the same cache line.
56 *
57 * Kernel-producer
58 * ---------------
59 * The producer position and data pages are mapped as r/o in
60 * userspace. For this approach, bits in the header of samples are
61 * used to signal to user-space, and to other producers, whether a
62 * sample is currently being written.
63 *
64 * User-space producer
65 * -------------------
66 * Only the page containing the consumer position is mapped r/o in
67 * user-space. User-space producers also use bits of the header to
68 * communicate to the kernel, but the kernel must carefully check and
69 * validate each sample to ensure that they're correctly formatted, and
70 * fully contained within the ring buffer.
71 */
72 unsigned long consumer_pos __aligned(PAGE_SIZE);
73 unsigned long producer_pos __aligned(PAGE_SIZE);
74 unsigned long pending_pos;
75 char data[] __aligned(PAGE_SIZE);
76 };
77
78 struct bpf_ringbuf_map {
79 struct bpf_map map;
80 struct bpf_ringbuf *rb;
81 };
82
83 /* 8-byte ring buffer record header structure */
84 struct bpf_ringbuf_hdr {
85 u32 len;
86 u32 pg_off;
87 };
88
89 static struct bpf_ringbuf *bpf_ringbuf_area_alloc(size_t data_sz, int numa_node)
90 {
91 const gfp_t flags = GFP_KERNEL_ACCOUNT | __GFP_RETRY_MAYFAIL |
92 __GFP_NOWARN | __GFP_ZERO;
93 int nr_meta_pages = RINGBUF_NR_META_PAGES;
94 int nr_data_pages = data_sz >> PAGE_SHIFT;
95 int nr_pages = nr_meta_pages + nr_data_pages;
96 struct page **pages, *page;
97 struct bpf_ringbuf *rb;
98 size_t array_size;
99 int i;
100
101 /* Each data page is mapped twice to allow "virtual"
102 * continuous read of samples wrapping around the end of ring
103 * buffer area:
104 * ------------------------------------------------------
105 * | meta pages | real data pages | same data pages |
106 * ------------------------------------------------------
107 * | | 1 2 3 4 5 6 7 8 9 | 1 2 3 4 5 6 7 8 9 |
108 * ------------------------------------------------------
109 * | | TA DA | TA DA |
110 * ------------------------------------------------------
111 * ^^^^^^^
112 * |
113 * Here, no need to worry about special handling of wrapped-around
114 * data due to double-mapped data pages. This works both in kernel and
115 * when mmap()'ed in user-space, simplifying both kernel and
116 * user-space implementations significantly.
117 */
118 array_size = (nr_meta_pages + 2 * nr_data_pages) * sizeof(*pages);
119 pages = bpf_map_area_alloc(array_size, numa_node);
120 if (!pages)
121 return NULL;
122
123 for (i = 0; i < nr_pages; i++) {
124 page = alloc_pages_node(numa_node, flags, 0);
125 if (!page) {
126 nr_pages = i;
127 goto err_free_pages;
128 }
129 pages[i] = page;
130 if (i >= nr_meta_pages)
131 pages[nr_data_pages + i] = page;
132 }
133
134 rb = vmap(pages, nr_meta_pages + 2 * nr_data_pages,
135 VM_MAP | VM_USERMAP, PAGE_KERNEL);
136 if (rb) {
137 kmemleak_not_leak(pages);
138 rb->pages = pages;
139 rb->nr_pages = nr_pages;
140 return rb;
141 }
142
143 err_free_pages:
144 for (i = 0; i < nr_pages; i++)
145 __free_page(pages[i]);
146 bpf_map_area_free(pages);
147 return NULL;
148 }
149
150 static void bpf_ringbuf_notify(struct irq_work *work)
151 {
152 struct bpf_ringbuf *rb = container_of(work, struct bpf_ringbuf, work);
153
154 wake_up_all(&rb->waitq);
155 }
156
157 /* Maximum size of ring buffer area is limited by 32-bit page offset within
158 * record header, counted in pages. Reserve 8 bits for extensibility, and
159 * take into account few extra pages for consumer/producer pages and
160 * non-mmap()'able parts, the current maximum size would be:
161 *
162 * (((1ULL << 24) - RINGBUF_POS_PAGES - RINGBUF_PGOFF) * PAGE_SIZE)
163 *
164 * This gives 64GB limit, which seems plenty for single ring buffer. Now
165 * considering that the maximum value of data_sz is (4GB - 1), there
166 * will be no overflow, so just note the size limit in the comments.
167 */
168 static struct bpf_ringbuf *bpf_ringbuf_alloc(size_t data_sz, int numa_node)
169 {
170 struct bpf_ringbuf *rb;
171
172 rb = bpf_ringbuf_area_alloc(data_sz, numa_node);
173 if (!rb)
174 return NULL;
175
176 spin_lock_init(&rb->spinlock);
177 atomic_set(&rb->busy, 0);
178 init_waitqueue_head(&rb->waitq);
179 init_irq_work(&rb->work, bpf_ringbuf_notify);
180
181 rb->mask = data_sz - 1;
182 rb->consumer_pos = 0;
183 rb->producer_pos = 0;
184 rb->pending_pos = 0;
185
186 return rb;
187 }
188
189 static struct bpf_map *ringbuf_map_alloc(union bpf_attr *attr)
190 {
191 struct bpf_ringbuf_map *rb_map;
192
193 if (attr->map_flags & ~RINGBUF_CREATE_FLAG_MASK)
194 return ERR_PTR(-EINVAL);
195
196 if (attr->key_size || attr->value_size ||
197 !is_power_of_2(attr->max_entries) ||
198 !PAGE_ALIGNED(attr->max_entries))
199 return ERR_PTR(-EINVAL);
200
201 rb_map = bpf_map_area_alloc(sizeof(*rb_map), NUMA_NO_NODE);
202 if (!rb_map)
203 return ERR_PTR(-ENOMEM);
204
205 bpf_map_init_from_attr(&rb_map->map, attr);
206
207 rb_map->rb = bpf_ringbuf_alloc(attr->max_entries, rb_map->map.numa_node);
208 if (!rb_map->rb) {
209 bpf_map_area_free(rb_map);
210 return ERR_PTR(-ENOMEM);
211 }
212
213 return &rb_map->map;
214 }
215
216 static void bpf_ringbuf_free(struct bpf_ringbuf *rb)
217 {
218 /* copy pages pointer and nr_pages to local variable, as we are going
219 * to unmap rb itself with vunmap() below
220 */
221 struct page **pages = rb->pages;
222 int i, nr_pages = rb->nr_pages;
223
224 vunmap(rb);
225 for (i = 0; i < nr_pages; i++)
226 __free_page(pages[i]);
227 bpf_map_area_free(pages);
228 }
229
230 static void ringbuf_map_free(struct bpf_map *map)
231 {
232 struct bpf_ringbuf_map *rb_map;
233
234 rb_map = container_of(map, struct bpf_ringbuf_map, map);
235 bpf_ringbuf_free(rb_map->rb);
236 bpf_map_area_free(rb_map);
237 }
238
239 static void *ringbuf_map_lookup_elem(struct bpf_map *map, void *key)
240 {
241 return ERR_PTR(-ENOTSUPP);
242 }
243
244 static long ringbuf_map_update_elem(struct bpf_map *map, void *key, void *value,
245 u64 flags)
246 {
247 return -ENOTSUPP;
248 }
249
250 static long ringbuf_map_delete_elem(struct bpf_map *map, void *key)
251 {
252 return -ENOTSUPP;
253 }
254
255 static int ringbuf_map_get_next_key(struct bpf_map *map, void *key,
256 void *next_key)
257 {
258 return -ENOTSUPP;
259 }
260
261 static int ringbuf_map_mmap_kern(struct bpf_map *map, struct vm_area_struct *vma)
262 {
263 struct bpf_ringbuf_map *rb_map;
264
265 rb_map = container_of(map, struct bpf_ringbuf_map, map);
266
267 if (vma->vm_flags & VM_WRITE) {
268 /* allow writable mapping for the consumer_pos only */
269 if (vma->vm_pgoff != 0 || vma->vm_end - vma->vm_start != PAGE_SIZE)
270 return -EPERM;
271 } else {
272 vm_flags_clear(vma, VM_MAYWRITE);
273 }
274 /* remap_vmalloc_range() checks size and offset constraints */
275 return remap_vmalloc_range(vma, rb_map->rb,
276 vma->vm_pgoff + RINGBUF_PGOFF);
277 }
278
279 static int ringbuf_map_mmap_user(struct bpf_map *map, struct vm_area_struct *vma)
280 {
281 struct bpf_ringbuf_map *rb_map;
282
283 rb_map = container_of(map, struct bpf_ringbuf_map, map);
284
285 if (vma->vm_flags & VM_WRITE) {
286 if (vma->vm_pgoff == 0)
287 /* Disallow writable mappings to the consumer pointer,
288 * and allow writable mappings to both the producer
289 * position, and the ring buffer data itself.
290 */
291 return -EPERM;
292 } else {
293 vm_flags_clear(vma, VM_MAYWRITE);
294 }
295 /* remap_vmalloc_range() checks size and offset constraints */
296 return remap_vmalloc_range(vma, rb_map->rb, vma->vm_pgoff + RINGBUF_PGOFF);
297 }
298
299 static unsigned long ringbuf_avail_data_sz(struct bpf_ringbuf *rb)
300 {
301 unsigned long cons_pos, prod_pos;
302
303 cons_pos = smp_load_acquire(&rb->consumer_pos);
304 prod_pos = smp_load_acquire(&rb->producer_pos);
305 return prod_pos - cons_pos;
306 }
307
308 static u32 ringbuf_total_data_sz(const struct bpf_ringbuf *rb)
309 {
310 return rb->mask + 1;
311 }
312
313 static __poll_t ringbuf_map_poll_kern(struct bpf_map *map, struct file *filp,
314 struct poll_table_struct *pts)
315 {
316 struct bpf_ringbuf_map *rb_map;
317
318 rb_map = container_of(map, struct bpf_ringbuf_map, map);
319 poll_wait(filp, &rb_map->rb->waitq, pts);
320
321 if (ringbuf_avail_data_sz(rb_map->rb))
322 return EPOLLIN | EPOLLRDNORM;
323 return 0;
324 }
325
326 static __poll_t ringbuf_map_poll_user(struct bpf_map *map, struct file *filp,
327 struct poll_table_struct *pts)
328 {
329 struct bpf_ringbuf_map *rb_map;
330
331 rb_map = container_of(map, struct bpf_ringbuf_map, map);
332 poll_wait(filp, &rb_map->rb->waitq, pts);
333
334 if (ringbuf_avail_data_sz(rb_map->rb) < ringbuf_total_data_sz(rb_map->rb))
335 return EPOLLOUT | EPOLLWRNORM;
336 return 0;
337 }
338
339 static u64 ringbuf_map_mem_usage(const struct bpf_map *map)
340 {
341 struct bpf_ringbuf *rb;
342 int nr_data_pages;
343 int nr_meta_pages;
344 u64 usage = sizeof(struct bpf_ringbuf_map);
345
346 rb = container_of(map, struct bpf_ringbuf_map, map)->rb;
347 usage += (u64)rb->nr_pages << PAGE_SHIFT;
348 nr_meta_pages = RINGBUF_NR_META_PAGES;
349 nr_data_pages = map->max_entries >> PAGE_SHIFT;
350 usage += (nr_meta_pages + 2 * nr_data_pages) * sizeof(struct page *);
351 return usage;
352 }
353
354 BTF_ID_LIST_SINGLE(ringbuf_map_btf_ids, struct, bpf_ringbuf_map)
355 const struct bpf_map_ops ringbuf_map_ops = {
356 .map_meta_equal = bpf_map_meta_equal,
357 .map_alloc = ringbuf_map_alloc,
358 .map_free = ringbuf_map_free,
359 .map_mmap = ringbuf_map_mmap_kern,
360 .map_poll = ringbuf_map_poll_kern,
361 .map_lookup_elem = ringbuf_map_lookup_elem,
362 .map_update_elem = ringbuf_map_update_elem,
363 .map_delete_elem = ringbuf_map_delete_elem,
364 .map_get_next_key = ringbuf_map_get_next_key,
365 .map_mem_usage = ringbuf_map_mem_usage,
366 .map_btf_id = &ringbuf_map_btf_ids[0],
367 };
368
369 BTF_ID_LIST_SINGLE(user_ringbuf_map_btf_ids, struct, bpf_ringbuf_map)
370 const struct bpf_map_ops user_ringbuf_map_ops = {
371 .map_meta_equal = bpf_map_meta_equal,
372 .map_alloc = ringbuf_map_alloc,
373 .map_free = ringbuf_map_free,
374 .map_mmap = ringbuf_map_mmap_user,
375 .map_poll = ringbuf_map_poll_user,
376 .map_lookup_elem = ringbuf_map_lookup_elem,
377 .map_update_elem = ringbuf_map_update_elem,
378 .map_delete_elem = ringbuf_map_delete_elem,
379 .map_get_next_key = ringbuf_map_get_next_key,
380 .map_mem_usage = ringbuf_map_mem_usage,
381 .map_btf_id = &user_ringbuf_map_btf_ids[0],
382 };
383
384 /* Given pointer to ring buffer record metadata and struct bpf_ringbuf itself,
385 * calculate offset from record metadata to ring buffer in pages, rounded
386 * down. This page offset is stored as part of record metadata and allows to
387 * restore struct bpf_ringbuf * from record pointer. This page offset is
388 * stored at offset 4 of record metadata header.
389 */
390 static size_t bpf_ringbuf_rec_pg_off(struct bpf_ringbuf *rb,
391 struct bpf_ringbuf_hdr *hdr)
392 {
393 return ((void *)hdr - (void *)rb) >> PAGE_SHIFT;
394 }
395
396 /* Given pointer to ring buffer record header, restore pointer to struct
397 * bpf_ringbuf itself by using page offset stored at offset 4
398 */
399 static struct bpf_ringbuf *
400 bpf_ringbuf_restore_from_rec(struct bpf_ringbuf_hdr *hdr)
401 {
402 unsigned long addr = (unsigned long)(void *)hdr;
403 unsigned long off = (unsigned long)hdr->pg_off << PAGE_SHIFT;
404
405 return (void*)((addr & PAGE_MASK) - off);
406 }
407
408 static void *__bpf_ringbuf_reserve(struct bpf_ringbuf *rb, u64 size)
409 {
410 unsigned long cons_pos, prod_pos, new_prod_pos, pend_pos, flags;
411 struct bpf_ringbuf_hdr *hdr;
412 u32 len, pg_off, tmp_size, hdr_len;
413
414 if (unlikely(size > RINGBUF_MAX_RECORD_SZ))
415 return NULL;
416
417 len = round_up(size + BPF_RINGBUF_HDR_SZ, 8);
418 if (len > ringbuf_total_data_sz(rb))
419 return NULL;
420
421 cons_pos = smp_load_acquire(&rb->consumer_pos);
422
423 if (in_nmi()) {
424 if (!spin_trylock_irqsave(&rb->spinlock, flags))
425 return NULL;
426 } else {
427 spin_lock_irqsave(&rb->spinlock, flags);
428 }
429
430 pend_pos = rb->pending_pos;
431 prod_pos = rb->producer_pos;
432 new_prod_pos = prod_pos + len;
433
434 while (pend_pos < prod_pos) {
435 hdr = (void *)rb->data + (pend_pos & rb->mask);
436 hdr_len = READ_ONCE(hdr->len);
437 if (hdr_len & BPF_RINGBUF_BUSY_BIT)
438 break;
439 tmp_size = hdr_len & ~BPF_RINGBUF_DISCARD_BIT;
440 tmp_size = round_up(tmp_size + BPF_RINGBUF_HDR_SZ, 8);
441 pend_pos += tmp_size;
442 }
443 rb->pending_pos = pend_pos;
444
445 /* check for out of ringbuf space:
446 * - by ensuring producer position doesn't advance more than
447 * (ringbuf_size - 1) ahead
448 * - by ensuring oldest not yet committed record until newest
449 * record does not span more than (ringbuf_size - 1)
450 */
451 if (new_prod_pos - cons_pos > rb->mask ||
452 new_prod_pos - pend_pos > rb->mask) {
453 spin_unlock_irqrestore(&rb->spinlock, flags);
454 return NULL;
455 }
456
457 hdr = (void *)rb->data + (prod_pos & rb->mask);
458 pg_off = bpf_ringbuf_rec_pg_off(rb, hdr);
459 hdr->len = size | BPF_RINGBUF_BUSY_BIT;
460 hdr->pg_off = pg_off;
461
462 /* pairs with consumer's smp_load_acquire() */
463 smp_store_release(&rb->producer_pos, new_prod_pos);
464
465 spin_unlock_irqrestore(&rb->spinlock, flags);
466
467 return (void *)hdr + BPF_RINGBUF_HDR_SZ;
468 }
469
470 BPF_CALL_3(bpf_ringbuf_reserve, struct bpf_map *, map, u64, size, u64, flags)
471 {
472 struct bpf_ringbuf_map *rb_map;
473
474 if (unlikely(flags))
475 return 0;
476
477 rb_map = container_of(map, struct bpf_ringbuf_map, map);
478 return (unsigned long)__bpf_ringbuf_reserve(rb_map->rb, size);
479 }
480
481 const struct bpf_func_proto bpf_ringbuf_reserve_proto = {
482 .func = bpf_ringbuf_reserve,
483 .ret_type = RET_PTR_TO_RINGBUF_MEM_OR_NULL,
484 .arg1_type = ARG_CONST_MAP_PTR,
485 .arg2_type = ARG_CONST_ALLOC_SIZE_OR_ZERO,
486 .arg3_type = ARG_ANYTHING,
487 };
488
489 static void bpf_ringbuf_commit(void *sample, u64 flags, bool discard)
490 {
491 unsigned long rec_pos, cons_pos;
492 struct bpf_ringbuf_hdr *hdr;
493 struct bpf_ringbuf *rb;
494 u32 new_len;
495
496 hdr = sample - BPF_RINGBUF_HDR_SZ;
497 rb = bpf_ringbuf_restore_from_rec(hdr);
498 new_len = hdr->len ^ BPF_RINGBUF_BUSY_BIT;
499 if (discard)
500 new_len |= BPF_RINGBUF_DISCARD_BIT;
501
502 /* update record header with correct final size prefix */
503 xchg(&hdr->len, new_len);
504
505 /* if consumer caught up and is waiting for our record, notify about
506 * new data availability
507 */
508 rec_pos = (void *)hdr - (void *)rb->data;
509 cons_pos = smp_load_acquire(&rb->consumer_pos) & rb->mask;
510
511 if (flags & BPF_RB_FORCE_WAKEUP)
512 irq_work_queue(&rb->work);
513 else if (cons_pos == rec_pos && !(flags & BPF_RB_NO_WAKEUP))
514 irq_work_queue(&rb->work);
515 }
516
517 BPF_CALL_2(bpf_ringbuf_submit, void *, sample, u64, flags)
518 {
519 bpf_ringbuf_commit(sample, flags, false /* discard */);
520 return 0;
521 }
522
523 const struct bpf_func_proto bpf_ringbuf_submit_proto = {
524 .func = bpf_ringbuf_submit,
525 .ret_type = RET_VOID,
526 .arg1_type = ARG_PTR_TO_RINGBUF_MEM | OBJ_RELEASE,
527 .arg2_type = ARG_ANYTHING,
528 };
529
530 BPF_CALL_2(bpf_ringbuf_discard, void *, sample, u64, flags)
531 {
532 bpf_ringbuf_commit(sample, flags, true /* discard */);
533 return 0;
534 }
535
536 const struct bpf_func_proto bpf_ringbuf_discard_proto = {
537 .func = bpf_ringbuf_discard,
538 .ret_type = RET_VOID,
539 .arg1_type = ARG_PTR_TO_RINGBUF_MEM | OBJ_RELEASE,
540 .arg2_type = ARG_ANYTHING,
541 };
542
543 BPF_CALL_4(bpf_ringbuf_output, struct bpf_map *, map, void *, data, u64, size,
544 u64, flags)
545 {
546 struct bpf_ringbuf_map *rb_map;
547 void *rec;
548
549 if (unlikely(flags & ~(BPF_RB_NO_WAKEUP | BPF_RB_FORCE_WAKEUP)))
550 return -EINVAL;
551
552 rb_map = container_of(map, struct bpf_ringbuf_map, map);
553 rec = __bpf_ringbuf_reserve(rb_map->rb, size);
554 if (!rec)
555 return -EAGAIN;
556
557 memcpy(rec, data, size);
558 bpf_ringbuf_commit(rec, flags, false /* discard */);
559 return 0;
560 }
561
562 const struct bpf_func_proto bpf_ringbuf_output_proto = {
563 .func = bpf_ringbuf_output,
564 .ret_type = RET_INTEGER,
565 .arg1_type = ARG_CONST_MAP_PTR,
566 .arg2_type = ARG_PTR_TO_MEM | MEM_RDONLY,
567 .arg3_type = ARG_CONST_SIZE_OR_ZERO,
568 .arg4_type = ARG_ANYTHING,
569 };
570
571 BPF_CALL_2(bpf_ringbuf_query, struct bpf_map *, map, u64, flags)
572 {
573 struct bpf_ringbuf *rb;
574
575 rb = container_of(map, struct bpf_ringbuf_map, map)->rb;
576
577 switch (flags) {
578 case BPF_RB_AVAIL_DATA:
579 return ringbuf_avail_data_sz(rb);
580 case BPF_RB_RING_SIZE:
581 return ringbuf_total_data_sz(rb);
582 case BPF_RB_CONS_POS:
583 return smp_load_acquire(&rb->consumer_pos);
584 case BPF_RB_PROD_POS:
585 return smp_load_acquire(&rb->producer_pos);
586 default:
587 return 0;
588 }
589 }
590
591 const struct bpf_func_proto bpf_ringbuf_query_proto = {
592 .func = bpf_ringbuf_query,
593 .ret_type = RET_INTEGER,
594 .arg1_type = ARG_CONST_MAP_PTR,
595 .arg2_type = ARG_ANYTHING,
596 };
597
598 BPF_CALL_4(bpf_ringbuf_reserve_dynptr, struct bpf_map *, map, u32, size, u64, flags,
599 struct bpf_dynptr_kern *, ptr)
600 {
601 struct bpf_ringbuf_map *rb_map;
602 void *sample;
603 int err;
604
605 if (unlikely(flags)) {
606 bpf_dynptr_set_null(ptr);
607 return -EINVAL;
608 }
609
610 err = bpf_dynptr_check_size(size);
611 if (err) {
612 bpf_dynptr_set_null(ptr);
613 return err;
614 }
615
616 rb_map = container_of(map, struct bpf_ringbuf_map, map);
617
618 sample = __bpf_ringbuf_reserve(rb_map->rb, size);
619 if (!sample) {
620 bpf_dynptr_set_null(ptr);
621 return -EINVAL;
622 }
623
624 bpf_dynptr_init(ptr, sample, BPF_DYNPTR_TYPE_RINGBUF, 0, size);
625
626 return 0;
627 }
628
629 const struct bpf_func_proto bpf_ringbuf_reserve_dynptr_proto = {
630 .func = bpf_ringbuf_reserve_dynptr,
631 .ret_type = RET_INTEGER,
632 .arg1_type = ARG_CONST_MAP_PTR,
633 .arg2_type = ARG_ANYTHING,
634 .arg3_type = ARG_ANYTHING,
635 .arg4_type = ARG_PTR_TO_DYNPTR | DYNPTR_TYPE_RINGBUF | MEM_UNINIT,
636 };
637
638 BPF_CALL_2(bpf_ringbuf_submit_dynptr, struct bpf_dynptr_kern *, ptr, u64, flags)
639 {
640 if (!ptr->data)
641 return 0;
642
643 bpf_ringbuf_commit(ptr->data, flags, false /* discard */);
644
645 bpf_dynptr_set_null(ptr);
646
647 return 0;
648 }
649
650 const struct bpf_func_proto bpf_ringbuf_submit_dynptr_proto = {
651 .func = bpf_ringbuf_submit_dynptr,
652 .ret_type = RET_VOID,
653 .arg1_type = ARG_PTR_TO_DYNPTR | DYNPTR_TYPE_RINGBUF | OBJ_RELEASE,
654 .arg2_type = ARG_ANYTHING,
655 };
656
657 BPF_CALL_2(bpf_ringbuf_discard_dynptr, struct bpf_dynptr_kern *, ptr, u64, flags)
658 {
659 if (!ptr->data)
660 return 0;
661
662 bpf_ringbuf_commit(ptr->data, flags, true /* discard */);
663
664 bpf_dynptr_set_null(ptr);
665
666 return 0;
667 }
668
669 const struct bpf_func_proto bpf_ringbuf_discard_dynptr_proto = {
670 .func = bpf_ringbuf_discard_dynptr,
671 .ret_type = RET_VOID,
672 .arg1_type = ARG_PTR_TO_DYNPTR | DYNPTR_TYPE_RINGBUF | OBJ_RELEASE,
673 .arg2_type = ARG_ANYTHING,
674 };
675
676 static int __bpf_user_ringbuf_peek(struct bpf_ringbuf *rb, void **sample, u32 *size)
677 {
678 int err;
679 u32 hdr_len, sample_len, total_len, flags, *hdr;
680 u64 cons_pos, prod_pos;
681
682 /* Synchronizes with smp_store_release() in user-space producer. */
683 prod_pos = smp_load_acquire(&rb->producer_pos);
684 if (prod_pos % 8)
685 return -EINVAL;
686
687 /* Synchronizes with smp_store_release() in __bpf_user_ringbuf_sample_release() */
688 cons_pos = smp_load_acquire(&rb->consumer_pos);
689 if (cons_pos >= prod_pos)
690 return -ENODATA;
691
692 hdr = (u32 *)((uintptr_t)rb->data + (uintptr_t)(cons_pos & rb->mask));
693 /* Synchronizes with smp_store_release() in user-space producer. */
694 hdr_len = smp_load_acquire(hdr);
695 flags = hdr_len & (BPF_RINGBUF_BUSY_BIT | BPF_RINGBUF_DISCARD_BIT);
696 sample_len = hdr_len & ~flags;
697 total_len = round_up(sample_len + BPF_RINGBUF_HDR_SZ, 8);
698
699 /* The sample must fit within the region advertised by the producer position. */
700 if (total_len > prod_pos - cons_pos)
701 return -EINVAL;
702
703 /* The sample must fit within the data region of the ring buffer. */
704 if (total_len > ringbuf_total_data_sz(rb))
705 return -E2BIG;
706
707 /* The sample must fit into a struct bpf_dynptr. */
708 err = bpf_dynptr_check_size(sample_len);
709 if (err)
710 return -E2BIG;
711
712 if (flags & BPF_RINGBUF_DISCARD_BIT) {
713 /* If the discard bit is set, the sample should be skipped.
714 *
715 * Update the consumer pos, and return -EAGAIN so the caller
716 * knows to skip this sample and try to read the next one.
717 */
718 smp_store_release(&rb->consumer_pos, cons_pos + total_len);
719 return -EAGAIN;
720 }
721
722 if (flags & BPF_RINGBUF_BUSY_BIT)
723 return -ENODATA;
724
725 *sample = (void *)((uintptr_t)rb->data +
726 (uintptr_t)((cons_pos + BPF_RINGBUF_HDR_SZ) & rb->mask));
727 *size = sample_len;
728 return 0;
729 }
730
731 static void __bpf_user_ringbuf_sample_release(struct bpf_ringbuf *rb, size_t size, u64 flags)
732 {
733 u64 consumer_pos;
734 u32 rounded_size = round_up(size + BPF_RINGBUF_HDR_SZ, 8);
735
736 /* Using smp_load_acquire() is unnecessary here, as the busy-bit
737 * prevents another task from writing to consumer_pos after it was read
738 * by this task with smp_load_acquire() in __bpf_user_ringbuf_peek().
739 */
740 consumer_pos = rb->consumer_pos;
741 /* Synchronizes with smp_load_acquire() in user-space producer. */
742 smp_store_release(&rb->consumer_pos, consumer_pos + rounded_size);
743 }
744
745 BPF_CALL_4(bpf_user_ringbuf_drain, struct bpf_map *, map,
746 void *, callback_fn, void *, callback_ctx, u64, flags)
747 {
748 struct bpf_ringbuf *rb;
749 long samples, discarded_samples = 0, ret = 0;
750 bpf_callback_t callback = (bpf_callback_t)callback_fn;
751 u64 wakeup_flags = BPF_RB_NO_WAKEUP | BPF_RB_FORCE_WAKEUP;
752 int busy = 0;
753
754 if (unlikely(flags & ~wakeup_flags))
755 return -EINVAL;
756
757 rb = container_of(map, struct bpf_ringbuf_map, map)->rb;
758
759 /* If another consumer is already consuming a sample, wait for them to finish. */
760 if (!atomic_try_cmpxchg(&rb->busy, &busy, 1))
761 return -EBUSY;
762
763 for (samples = 0; samples < BPF_MAX_USER_RINGBUF_SAMPLES && ret == 0; samples++) {
764 int err;
765 u32 size;
766 void *sample;
767 struct bpf_dynptr_kern dynptr;
768
769 err = __bpf_user_ringbuf_peek(rb, &sample, &size);
770 if (err) {
771 if (err == -ENODATA) {
772 break;
773 } else if (err == -EAGAIN) {
774 discarded_samples++;
775 continue;
776 } else {
777 ret = err;
778 goto schedule_work_return;
779 }
780 }
781
782 bpf_dynptr_init(&dynptr, sample, BPF_DYNPTR_TYPE_LOCAL, 0, size);
783 ret = callback((uintptr_t)&dynptr, (uintptr_t)callback_ctx, 0, 0, 0);
784 __bpf_user_ringbuf_sample_release(rb, size, flags);
785 }
786 ret = samples - discarded_samples;
787
788 schedule_work_return:
789 /* Prevent the clearing of the busy-bit from being reordered before the
790 * storing of any rb consumer or producer positions.
791 */
792 atomic_set_release(&rb->busy, 0);
793
794 if (flags & BPF_RB_FORCE_WAKEUP)
795 irq_work_queue(&rb->work);
796 else if (!(flags & BPF_RB_NO_WAKEUP) && samples > 0)
797 irq_work_queue(&rb->work);
798 return ret;
799 }
800
801 const struct bpf_func_proto bpf_user_ringbuf_drain_proto = {
802 .func = bpf_user_ringbuf_drain,
803 .ret_type = RET_INTEGER,
804 .arg1_type = ARG_CONST_MAP_PTR,
805 .arg2_type = ARG_PTR_TO_FUNC,
806 .arg3_type = ARG_PTR_TO_STACK_OR_NULL,
807 .arg4_type = ARG_ANYTHING,
808 };
809
Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.