kernel/bpf/arraymap.c

   1 /* Copyright (c) 2011-2014 PLUMgrid, http://plumgrid.com
   2  *
   3  * This program is free software; you can redistribute it and/or
   4  * modify it under the terms of version 2 of the GNU General Public
   5  * License as published by the Free Software Foundation.
   6  *
   7  * This program is distributed in the hope that it will be useful, but
   8  * WITHOUT ANY WARRANTY; without even the implied warranty of
   9  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
  10  * General Public License for more details.
  11  */
  12 #include <linux/bpf.h>
  13 #include <linux/err.h>
  14 #include <linux/vmalloc.h>
  15 #include <linux/slab.h>
  16 #include <linux/mm.h>
  17 #include <linux/filter.h>
  18 #include <linux/perf_event.h>
  19
  20 static void bpf_array_free_percpu(struct bpf_array *array)
  21 {
  22         int i;
  23
  24         for (i = 0; i < array->map.max_entries; i++)
  25                 free_percpu(array->pptrs[i]);
  26 }
  27
  28 static int bpf_array_alloc_percpu(struct bpf_array *array)
  29 {
  30         void __percpu *ptr;
  31         int i;
  32
  33         for (i = 0; i < array->map.max_entries; i++) {
  34                 ptr = __alloc_percpu_gfp(array->elem_size, 8,
  35                                          GFP_USER | __GFP_NOWARN);
  36                 if (!ptr) {
  37                         bpf_array_free_percpu(array);
  38                         return -ENOMEM;
  39                 }
  40                 array->pptrs[i] = ptr;
  41         }
  42
  43         return 0;
  44 }
  45
  46 /* Called from syscall */
  47 static struct bpf_map *array_map_alloc(union bpf_attr *attr)
  48 {
  49         bool percpu = attr->map_type == BPF_MAP_TYPE_PERCPU_ARRAY;
  50         struct bpf_array *array;
  51         u64 array_size;
  52         u32 elem_size;
  53
  54         /* check sanity of attributes */
  55         if (attr->max_entries == 0 || attr->key_size != 4 ||
  56             attr->value_size == 0 || attr->map_flags)
  57                 return ERR_PTR(-EINVAL);
  58
  59         if (attr->value_size >= 1 << (KMALLOC_SHIFT_MAX - 1))
  60                 /* if value_size is bigger, the user space won't be able to
  61                  * access the elements.
  62                  */
  63                 return ERR_PTR(-E2BIG);
  64
  65         elem_size = round_up(attr->value_size, 8);
  66
  67         array_size = sizeof(*array);
  68         if (percpu)
  69                 array_size += (u64) attr->max_entries * sizeof(void *);
  70         else
  71                 array_size += (u64) attr->max_entries * elem_size;
  72
  73         /* make sure there is no u32 overflow later in round_up() */
  74         if (array_size >= U32_MAX - PAGE_SIZE)
  75                 return ERR_PTR(-ENOMEM);
  76
  77
  78         /* allocate all map elements and zero-initialize them */
  79         array = kzalloc(array_size, GFP_USER | __GFP_NOWARN);
  80         if (!array) {
  81                 array = vzalloc(array_size);
  82                 if (!array)
  83                         return ERR_PTR(-ENOMEM);
  84         }
  85
  86         /* copy mandatory map attributes */
  87         array->map.map_type = attr->map_type;
  88         array->map.key_size = attr->key_size;
  89         array->map.value_size = attr->value_size;
  90         array->map.max_entries = attr->max_entries;
  91         array->elem_size = elem_size;
  92
  93         if (!percpu)
  94                 goto out;
  95
  96         array_size += (u64) attr->max_entries * elem_size * num_possible_cpus();
  97
  98         if (array_size >= U32_MAX - PAGE_SIZE ||
  99             elem_size > PCPU_MIN_UNIT_SIZE || bpf_array_alloc_percpu(array)) {
 100                 kvfree(array);
 101                 return ERR_PTR(-ENOMEM);
 102         }
 103 out:
 104         array->map.pages = round_up(array_size, PAGE_SIZE) >> PAGE_SHIFT;
 105
 106         return &array->map;
 107 }
 108
 109 /* Called from syscall or from eBPF program */
 110 static void *array_map_lookup_elem(struct bpf_map *map, void *key)
 111 {
 112         struct bpf_array *array = container_of(map, struct bpf_array, map);
 113         u32 index = *(u32 *)key;
 114
 115         if (unlikely(index >= array->map.max_entries))
 116                 return NULL;
 117
 118         return array->value + array->elem_size * index;
 119 }
 120
 121 /* Called from eBPF program */
 122 static void *percpu_array_map_lookup_elem(struct bpf_map *map, void *key)
 123 {
 124         struct bpf_array *array = container_of(map, struct bpf_array, map);
 125         u32 index = *(u32 *)key;
 126
 127         if (unlikely(index >= array->map.max_entries))
 128                 return NULL;
 129
 130         return this_cpu_ptr(array->pptrs[index]);
 131 }
 132
 133 int bpf_percpu_array_copy(struct bpf_map *map, void *key, void *value)
 134 {
 135         struct bpf_array *array = container_of(map, struct bpf_array, map);
 136         u32 index = *(u32 *)key;
 137         void __percpu *pptr;
 138         int cpu, off = 0;
 139         u32 size;
 140
 141         if (unlikely(index >= array->map.max_entries))
 142                 return -ENOENT;
 143
 144         /* per_cpu areas are zero-filled and bpf programs can only
 145          * access 'value_size' of them, so copying rounded areas
 146          * will not leak any kernel data
 147          */
 148         size = round_up(map->value_size, 8);
 149         rcu_read_lock();
 150         pptr = array->pptrs[index];
 151         for_each_possible_cpu(cpu) {
 152                 bpf_long_memcpy(value + off, per_cpu_ptr(pptr, cpu), size);
 153                 off += size;
 154         }
 155         rcu_read_unlock();
 156         return 0;
 157 }
 158
 159 /* Called from syscall */
 160 static int array_map_get_next_key(struct bpf_map *map, void *key, void *next_key)
 161 {
 162         struct bpf_array *array = container_of(map, struct bpf_array, map);
 163         u32 index = *(u32 *)key;
 164         u32 *next = (u32 *)next_key;
 165
 166         if (index >= array->map.max_entries) {
 167                 *next = 0;
 168                 return 0;
 169         }
 170
 171         if (index == array->map.max_entries - 1)
 172                 return -ENOENT;
 173
 174         *next = index + 1;
 175         return 0;
 176 }
 177
 178 /* Called from syscall or from eBPF program */
 179 static int array_map_update_elem(struct bpf_map *map, void *key, void *value,
 180                                  u64 map_flags)
 181 {
 182         struct bpf_array *array = container_of(map, struct bpf_array, map);
 183         u32 index = *(u32 *)key;
 184
 185         if (unlikely(map_flags > BPF_EXIST))
 186                 /* unknown flags */
 187                 return -EINVAL;
 188
 189         if (unlikely(index >= array->map.max_entries))
 190                 /* all elements were pre-allocated, cannot insert a new one */
 191                 return -E2BIG;
 192
 193         if (unlikely(map_flags == BPF_NOEXIST))
 194                 /* all elements already exist */
 195                 return -EEXIST;
 196
 197         if (array->map.map_type == BPF_MAP_TYPE_PERCPU_ARRAY)
 198                 memcpy(this_cpu_ptr(array->pptrs[index]),
 199                        value, map->value_size);
 200         else
 201                 memcpy(array->value + array->elem_size * index,
 202                        value, map->value_size);
 203         return 0;
 204 }
 205
 206 int bpf_percpu_array_update(struct bpf_map *map, void *key, void *value,
 207                             u64 map_flags)
 208 {
 209         struct bpf_array *array = container_of(map, struct bpf_array, map);
 210         u32 index = *(u32 *)key;
 211         void __percpu *pptr;
 212         int cpu, off = 0;
 213         u32 size;
 214
 215         if (unlikely(map_flags > BPF_EXIST))
 216                 /* unknown flags */
 217                 return -EINVAL;
 218
 219         if (unlikely(index >= array->map.max_entries))
 220                 /* all elements were pre-allocated, cannot insert a new one */
 221                 return -E2BIG;
 222
 223         if (unlikely(map_flags == BPF_NOEXIST))
 224                 /* all elements already exist */
 225                 return -EEXIST;
 226
 227         /* the user space will provide round_up(value_size, 8) bytes that
 228          * will be copied into per-cpu area. bpf programs can only access
 229          * value_size of it. During lookup the same extra bytes will be
 230          * returned or zeros which were zero-filled by percpu_alloc,
 231          * so no kernel data leaks possible
 232          */
 233         size = round_up(map->value_size, 8);
 234         rcu_read_lock();
 235         pptr = array->pptrs[index];
 236         for_each_possible_cpu(cpu) {
 237                 bpf_long_memcpy(per_cpu_ptr(pptr, cpu), value + off, size);
 238                 off += size;
 239         }
 240         rcu_read_unlock();
 241         return 0;
 242 }
 243
 244 /* Called from syscall or from eBPF program */
 245 static int array_map_delete_elem(struct bpf_map *map, void *key)
 246 {
 247         return -EINVAL;
 248 }
 249
 250 /* Called when map->refcnt goes to zero, either from workqueue or from syscall */
 251 static void array_map_free(struct bpf_map *map)
 252 {
 253         struct bpf_array *array = container_of(map, struct bpf_array, map);
 254
 255         /* at this point bpf_prog->aux->refcnt == 0 and this map->refcnt == 0,
 256          * so the programs (can be more than one that used this map) were
 257          * disconnected from events. Wait for outstanding programs to complete
 258          * and free the array
 259          */
 260         synchronize_rcu();
 261
 262         if (array->map.map_type == BPF_MAP_TYPE_PERCPU_ARRAY)
 263                 bpf_array_free_percpu(array);
 264
 265         kvfree(array);
 266 }
 267
 268 static const struct bpf_map_ops array_ops = {
 269         .map_alloc = array_map_alloc,
 270         .map_free = array_map_free,
 271         .map_get_next_key = array_map_get_next_key,
 272         .map_lookup_elem = array_map_lookup_elem,
 273         .map_update_elem = array_map_update_elem,
 274         .map_delete_elem = array_map_delete_elem,
 275 };
 276
 277 static struct bpf_map_type_list array_type __read_mostly = {
 278         .ops = &array_ops,
 279         .type = BPF_MAP_TYPE_ARRAY,
 280 };
 281
 282 static const struct bpf_map_ops percpu_array_ops = {
 283         .map_alloc = array_map_alloc,
 284         .map_free = array_map_free,
 285         .map_get_next_key = array_map_get_next_key,
 286         .map_lookup_elem = percpu_array_map_lookup_elem,
 287         .map_update_elem = array_map_update_elem,
 288         .map_delete_elem = array_map_delete_elem,
 289 };
 290
 291 static struct bpf_map_type_list percpu_array_type __read_mostly = {
 292         .ops = &percpu_array_ops,
 293         .type = BPF_MAP_TYPE_PERCPU_ARRAY,
 294 };
 295
 296 static int __init register_array_map(void)
 297 {
 298         bpf_register_map_type(&array_type);
 299         bpf_register_map_type(&percpu_array_type);
 300         return 0;
 301 }
 302 late_initcall(register_array_map);
 303
 304 static struct bpf_map *fd_array_map_alloc(union bpf_attr *attr)
 305 {
 306         /* only file descriptors can be stored in this type of map */
 307         if (attr->value_size != sizeof(u32))
 308                 return ERR_PTR(-EINVAL);
 309         return array_map_alloc(attr);
 310 }
 311
 312 static void fd_array_map_free(struct bpf_map *map)
 313 {
 314         struct bpf_array *array = container_of(map, struct bpf_array, map);
 315         int i;
 316
 317         synchronize_rcu();
 318
 319         /* make sure it's empty */
 320         for (i = 0; i < array->map.max_entries; i++)
 321                 BUG_ON(array->ptrs[i] != NULL);
 322         kvfree(array);
 323 }
 324
 325 static void *fd_array_map_lookup_elem(struct bpf_map *map, void *key)
 326 {
 327         return NULL;
 328 }
 329
 330 /* only called from syscall */
 331 int bpf_fd_array_map_update_elem(struct bpf_map *map, struct file *map_file,
 332                                  void *key, void *value, u64 map_flags)
 333 {
 334         struct bpf_array *array = container_of(map, struct bpf_array, map);
 335         void *new_ptr, *old_ptr;
 336         u32 index = *(u32 *)key, ufd;
 337
 338         if (map_flags != BPF_ANY)
 339                 return -EINVAL;
 340
 341         if (index >= array->map.max_entries)
 342                 return -E2BIG;
 343
 344         ufd = *(u32 *)value;
 345         new_ptr = map->ops->map_fd_get_ptr(map, map_file, ufd);
 346         if (IS_ERR(new_ptr))
 347                 return PTR_ERR(new_ptr);
 348
 349         old_ptr = xchg(array->ptrs + index, new_ptr);
 350         if (old_ptr)
 351                 map->ops->map_fd_put_ptr(old_ptr);
 352
 353         return 0;
 354 }
 355
 356 static int fd_array_map_delete_elem(struct bpf_map *map, void *key)
 357 {
 358         struct bpf_array *array = container_of(map, struct bpf_array, map);
 359         void *old_ptr;
 360         u32 index = *(u32 *)key;
 361
 362         if (index >= array->map.max_entries)
 363                 return -E2BIG;
 364
 365         old_ptr = xchg(array->ptrs + index, NULL);
 366         if (old_ptr) {
 367                 map->ops->map_fd_put_ptr(old_ptr);
 368                 return 0;
 369         } else {
 370                 return -ENOENT;
 371         }
 372 }
 373
 374 static void *prog_fd_array_get_ptr(struct bpf_map *map,
 375                                    struct file *map_file, int fd)
 376 {
 377         struct bpf_array *array = container_of(map, struct bpf_array, map);
 378         struct bpf_prog *prog = bpf_prog_get(fd);
 379
 380         if (IS_ERR(prog))
 381                 return prog;
 382
 383         if (!bpf_prog_array_compatible(array, prog)) {
 384                 bpf_prog_put(prog);
 385                 return ERR_PTR(-EINVAL);
 386         }
 387
 388         return prog;
 389 }
 390
 391 static void prog_fd_array_put_ptr(void *ptr)
 392 {
 393         bpf_prog_put(ptr);
 394 }
 395
 396 /* decrement refcnt of all bpf_progs that are stored in this map */
 397 void bpf_fd_array_map_clear(struct bpf_map *map)
 398 {
 399         struct bpf_array *array = container_of(map, struct bpf_array, map);
 400         int i;
 401
 402         for (i = 0; i < array->map.max_entries; i++)
 403                 fd_array_map_delete_elem(map, &i);
 404 }
 405
 406 static const struct bpf_map_ops prog_array_ops = {
 407         .map_alloc = fd_array_map_alloc,
 408         .map_free = fd_array_map_free,
 409         .map_get_next_key = array_map_get_next_key,
 410         .map_lookup_elem = fd_array_map_lookup_elem,
 411         .map_delete_elem = fd_array_map_delete_elem,
 412         .map_fd_get_ptr = prog_fd_array_get_ptr,
 413         .map_fd_put_ptr = prog_fd_array_put_ptr,
 414 };
 415
 416 static struct bpf_map_type_list prog_array_type __read_mostly = {
 417         .ops = &prog_array_ops,
 418         .type = BPF_MAP_TYPE_PROG_ARRAY,
 419 };
 420
 421 static int __init register_prog_array_map(void)
 422 {
 423         bpf_register_map_type(&prog_array_type);
 424         return 0;
 425 }
 426 late_initcall(register_prog_array_map);
 427
 428 static struct bpf_event_entry *bpf_event_entry_gen(struct file *perf_file,
 429                                                    struct file *map_file)
 430 {
 431         struct bpf_event_entry *ee;
 432
 433         ee = kzalloc(sizeof(*ee), GFP_ATOMIC);
 434         if (ee) {
 435                 ee->event = perf_file->private_data;
 436                 ee->perf_file = perf_file;
 437                 ee->map_file = map_file;
 438         }
 439
 440         return ee;
 441 }
 442
 443 static void __bpf_event_entry_free(struct rcu_head *rcu)
 444 {
 445         struct bpf_event_entry *ee;
 446
 447         ee = container_of(rcu, struct bpf_event_entry, rcu);
 448         fput(ee->perf_file);
 449         kfree(ee);
 450 }
 451
 452 static void bpf_event_entry_free_rcu(struct bpf_event_entry *ee)
 453 {
 454         call_rcu(&ee->rcu, __bpf_event_entry_free);
 455 }
 456
 457 static void *perf_event_fd_array_get_ptr(struct bpf_map *map,
 458                                          struct file *map_file, int fd)
 459 {
 460         const struct perf_event_attr *attr;
 461         struct bpf_event_entry *ee;
 462         struct perf_event *event;
 463         struct file *perf_file;
 464
 465         perf_file = perf_event_get(fd);
 466         if (IS_ERR(perf_file))
 467                 return perf_file;
 468
 469         event = perf_file->private_data;
 470         ee = ERR_PTR(-EINVAL);
 471
 472         attr = perf_event_attrs(event);
 473         if (IS_ERR(attr) || attr->inherit)
 474                 goto err_out;
 475
 476         switch (attr->type) {
 477         case PERF_TYPE_SOFTWARE:
 478                 if (attr->config != PERF_COUNT_SW_BPF_OUTPUT)
 479                         goto err_out;
 480                 /* fall-through */
 481         case PERF_TYPE_RAW:
 482         case PERF_TYPE_HARDWARE:
 483                 ee = bpf_event_entry_gen(perf_file, map_file);
 484                 if (ee)
 485                         return ee;
 486                 ee = ERR_PTR(-ENOMEM);
 487                 /* fall-through */
 488         default:
 489                 break;
 490         }
 491
 492 err_out:
 493         fput(perf_file);
 494         return ee;
 495 }
 496
 497 static void perf_event_fd_array_put_ptr(void *ptr)
 498 {
 499         bpf_event_entry_free_rcu(ptr);
 500 }
 501
 502 static void perf_event_fd_array_release(struct bpf_map *map,
 503                                         struct file *map_file)
 504 {
 505         struct bpf_array *array = container_of(map, struct bpf_array, map);
 506         struct bpf_event_entry *ee;
 507         int i;
 508
 509         rcu_read_lock();
 510         for (i = 0; i < array->map.max_entries; i++) {
 511                 ee = READ_ONCE(array->ptrs[i]);
 512                 if (ee && ee->map_file == map_file)
 513                         fd_array_map_delete_elem(map, &i);
 514         }
 515         rcu_read_unlock();
 516 }
 517
 518 static const struct bpf_map_ops perf_event_array_ops = {
 519         .map_alloc = fd_array_map_alloc,
 520         .map_free = fd_array_map_free,
 521         .map_get_next_key = array_map_get_next_key,
 522         .map_lookup_elem = fd_array_map_lookup_elem,
 523         .map_delete_elem = fd_array_map_delete_elem,
 524         .map_fd_get_ptr = perf_event_fd_array_get_ptr,
 525         .map_fd_put_ptr = perf_event_fd_array_put_ptr,
 526         .map_release = perf_event_fd_array_release,
 527 };
 528
 529 static struct bpf_map_type_list perf_event_array_type __read_mostly = {
 530         .ops = &perf_event_array_ops,
 531         .type = BPF_MAP_TYPE_PERF_EVENT_ARRAY,
 532 };
 533
 534 static int __init register_perf_event_array_map(void)
 535 {
 536         bpf_register_map_type(&perf_event_array_type);
 537         return 0;
 538 }
 539 late_initcall(register_perf_event_array_map);
 540
 541 #ifdef CONFIG_CGROUPS
 542 static void *cgroup_fd_array_get_ptr(struct bpf_map *map,
 543                                      struct file *map_file /* not used */,
 544                                      int fd)
 545 {
 546         return cgroup_get_from_fd(fd);
 547 }
 548
 549 static void cgroup_fd_array_put_ptr(void *ptr)
 550 {
 551         /* cgroup_put free cgrp after a rcu grace period */
 552         cgroup_put(ptr);
 553 }
 554
 555 static void cgroup_fd_array_free(struct bpf_map *map)
 556 {
 557         bpf_fd_array_map_clear(map);
 558         fd_array_map_free(map);
 559 }
 560
 561 static const struct bpf_map_ops cgroup_array_ops = {
 562         .map_alloc = fd_array_map_alloc,
 563         .map_free = cgroup_fd_array_free,
 564         .map_get_next_key = array_map_get_next_key,
 565         .map_lookup_elem = fd_array_map_lookup_elem,
 566         .map_delete_elem = fd_array_map_delete_elem,
 567         .map_fd_get_ptr = cgroup_fd_array_get_ptr,
 568         .map_fd_put_ptr = cgroup_fd_array_put_ptr,
 569 };
 570
 571 static struct bpf_map_type_list cgroup_array_type __read_mostly = {
 572         .ops = &cgroup_array_ops,
 573         .type = BPF_MAP_TYPE_CGROUP_ARRAY,
 574 };
 575
 576 static int __init register_cgroup_array_map(void)
 577 {
 578         bpf_register_map_type(&cgroup_array_type);
 579         return 0;
 580 }
 581 late_initcall(register_cgroup_array_map);
 582 #endif