summaryrefslogtreecommitdiffstats
path: root/src/vppinfra/freebsd/mem.c
blob: 7d27a0dc169184ae342f812ed00b93f0ace27aca (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
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
/* SPDX-License-Identifier: Apache-2.0
 * Copyright(c) 2021 Cisco Systems, Inc.
 * Copyright(c) 2024 Tom Jones <thj@freebsd.org>
 */

#define _GNU_SOURCE
#include <stdlib.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <unistd.h>
#include <sys/memrange.h>
#include <sys/mount.h>
#include <sys/mman.h>
#include <fcntl.h>
#include <vppinfra/clib.h>
#include <vppinfra/mem.h>
#include <vppinfra/lock.h>
#include <vppinfra/time.h>
#include <vppinfra/bitmap.h>
#include <vppinfra/format.h>
#include <vppinfra/clib_error.h>

#ifndef F_FBSD_SPECIFIC_BASE
#define F_FBSD_SPECIFIC_BASE 1024
#endif

#ifndef F_ADD_SEALS
#define F_ADD_SEALS (F_FBSD_SPECIFIC_BASE + 9)
#define F_GET_SEALS (F_FBSD_SPECIFIC_BASE + 10)

#define F_SEAL_SEAL   0x0001 /* prevent further seals from being set */
#define F_SEAL_SHRINK 0x0002 /* prevent file from shrinking */
#define F_SEAL_GROW   0x0004 /* prevent file from growing */
#define F_SEAL_WRITE  0x0008 /* prevent writes */
#endif

#ifndef MFD_HUGETLB
#define MFD_HUGETLB 0x0004U
#endif

#ifndef MAP_HUGE_SHIFT
#define MAP_HUGE_SHIFT 26
#endif

#ifndef MFD_HUGE_SHIFT
#define MFD_HUGE_SHIFT 26
#endif

#ifndef MAP_FIXED_NOREPLACE
#define MAP_FIXED_NOREPLACE MAP_FIXED
#endif

static void
map_lock ()
{
  while (clib_atomic_test_and_set (&clib_mem_main.map_lock))
    CLIB_PAUSE ();
}

static void
map_unlock ()
{
  clib_atomic_release (&clib_mem_main.map_lock);
}

void
clib_mem_main_init (void)
{
  clib_mem_main_t *mm = &clib_mem_main;
  long sysconf_page_size;
  uword page_size;
  void *va;

  if (mm->log2_page_sz != CLIB_MEM_PAGE_SZ_UNKNOWN)
    return;

  /* system page size */
  sysconf_page_size = sysconf (_SC_PAGESIZE);
  if (sysconf_page_size < 0)
    {
      clib_panic ("Could not determine the page size");
    }
  page_size = sysconf_page_size;
  mm->log2_page_sz = min_log2 (page_size);

  mm->log2_default_hugepage_sz = min_log2 (page_size);
  mm->log2_sys_default_hugepage_sz = mm->log2_default_hugepage_sz;

  /* numa nodes */
  va = mmap (0, page_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS,
	     -1, 0);
  if (va == MAP_FAILED)
    return;

  if (mlock (va, page_size))
    goto done;

  /*
   * TODO: In linux/mem.c we can move pages to numa domains, this isn't an
   * option in FreeBSD yet.
   */

done:
  munmap (va, page_size);
}

__clib_export u64
clib_mem_get_fd_page_size (int fd)
{
  struct stat st = { 0 };
  if (fstat (fd, &st) == -1)
    return 0;
  return st.st_blksize;
}

__clib_export clib_mem_page_sz_t
clib_mem_get_fd_log2_page_size (int fd)
{
  uword page_size = clib_mem_get_fd_page_size (fd);
  return page_size ? min_log2 (page_size) : CLIB_MEM_PAGE_SZ_UNKNOWN;
}

__clib_export void
clib_mem_vm_randomize_va (uword *requested_va,
			  clib_mem_page_sz_t log2_page_size)
{
  /* TODO: Not yet implemented */
}

__clib_export int
clib_mem_vm_create_fd (clib_mem_page_sz_t log2_page_size, char *fmt, ...)
{
  clib_mem_main_t *mm = &clib_mem_main;
  int fd;
  unsigned int memfd_flags;
  va_list va;
  u8 *s = 0;

  if (log2_page_size == mm->log2_page_sz)
    log2_page_size = CLIB_MEM_PAGE_SZ_DEFAULT;
  else if (log2_page_size == mm->log2_sys_default_hugepage_sz)
    log2_page_size = CLIB_MEM_PAGE_SZ_DEFAULT_HUGE;

  switch (log2_page_size)
    {
    case CLIB_MEM_PAGE_SZ_UNKNOWN:
      return CLIB_MEM_ERROR;
    case CLIB_MEM_PAGE_SZ_DEFAULT:
      memfd_flags = MFD_ALLOW_SEALING;
      break;
    case CLIB_MEM_PAGE_SZ_DEFAULT_HUGE:
      memfd_flags = MFD_HUGETLB;
      break;
    default:
      memfd_flags = MFD_HUGETLB | log2_page_size << MFD_HUGE_SHIFT;
    }

  va_start (va, fmt);
  s = va_format (0, fmt, &va);
  va_end (va);

  /* memfd_create maximum string size is 249 chars without trailing zero */
  if (vec_len (s) > 249)
    vec_set_len (s, 249);
  vec_add1 (s, 0);

  fd = memfd_create ((char *) s, memfd_flags);
  if (fd == -1)
    {
      vec_reset_length (mm->error);
      mm->error = clib_error_return_unix (mm->error, "memfd_create");
      vec_free (s);
      return CLIB_MEM_ERROR;
    }

  vec_free (s);

  if ((memfd_flags & MFD_ALLOW_SEALING) &&
      ((fcntl (fd, F_ADD_SEALS, F_SEAL_SHRINK)) == -1))
    {
      vec_reset_length (mm->error);
      mm->error = clib_error_return_unix (mm->error, "fcntl (F_ADD_SEALS)");
      close (fd);
      return CLIB_MEM_ERROR;
    }

  return fd;
}

uword
clib_mem_vm_reserve (uword start, uword size, clib_mem_page_sz_t log2_page_sz)
{
  clib_mem_main_t *mm = &clib_mem_main;
  uword pagesize = 1ULL << log2_page_sz;
  uword sys_page_sz = 1ULL << mm->log2_page_sz;
  uword n_bytes;
  void *base = 0, *p;

  size = round_pow2 (size, pagesize);

  /* in adition of requested reservation, we also rserve one system page
   * (typically 4K) adjacent to the start off reservation */

  if (start)
    {
      /* start address is provided, so we just need to make sure we are not
       * replacing existing map */
      if (start & pow2_mask (log2_page_sz))
	return ~0;
      base = (void *) start - sys_page_sz;
      base = mmap (base, size + sys_page_sz, PROT_NONE,
		   MAP_PRIVATE | MAP_ANONYMOUS | MAP_FIXED_NOREPLACE, -1, 0);

      return (base == MAP_FAILED) ? ~0 : start;
    }

  /* to make sure that we get reservation aligned to page_size we need to
   * request one additional page as mmap will return us address which is
   * aligned only to system page size */
  base =
    mmap (0, size + pagesize, PROT_NONE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);

  if (base == MAP_FAILED)
    return ~0;

  /* return additional space at the end of allocation */
  p = base + size + pagesize;
  n_bytes = (uword) p & pow2_mask (log2_page_sz);
  if (n_bytes)
    {
      p -= n_bytes;
      munmap (p, n_bytes);
    }

  /* return additional space at the start of allocation */
  n_bytes = pagesize - sys_page_sz - n_bytes;
  if (n_bytes)
    {
      munmap (base, n_bytes);
      base += n_bytes;
    }

  return (uword) base + sys_page_sz;
}

__clib_export clib_mem_vm_map_hdr_t *
clib_mem_vm_get_next_map_hdr (clib_mem_vm_map_hdr_t *hdr)
{
  /* TODO: Not yet implemented */
  return NULL;
}

void *
clib_mem_vm_map_internal (void *base, clib_mem_page_sz_t log2_page_sz,
			  uword size, int fd, uword offset, char *name)
{
  clib_mem_main_t *mm = &clib_mem_main;
  clib_mem_vm_map_hdr_t *hdr;
  uword sys_page_sz = 1ULL << mm->log2_page_sz;
  int mmap_flags = MAP_FIXED, is_huge = 0;

  if (fd != -1)
    {
      mmap_flags |= MAP_SHARED;
      log2_page_sz = clib_mem_get_fd_log2_page_size (fd);
      if (log2_page_sz > mm->log2_page_sz)
	is_huge = 1;
    }
  else
    {
      mmap_flags |= MAP_PRIVATE | MAP_ANONYMOUS;

      if (log2_page_sz == mm->log2_page_sz)
	log2_page_sz = CLIB_MEM_PAGE_SZ_DEFAULT;

      switch (log2_page_sz)
	{
	case CLIB_MEM_PAGE_SZ_UNKNOWN:
	  /* will fail later */
	  break;
	case CLIB_MEM_PAGE_SZ_DEFAULT:
	  log2_page_sz = mm->log2_page_sz;
	  break;
	case CLIB_MEM_PAGE_SZ_DEFAULT_HUGE:
	  /* We shouldn't be selecting HUGETLB on FreeBSD */
	  log2_page_sz = CLIB_MEM_PAGE_SZ_UNKNOWN;
	  break;
	default:
	  log2_page_sz = mm->log2_page_sz;
	  break;
	}
    }

  size = round_pow2 (size, 1ULL << log2_page_sz);

  base = (void *) clib_mem_vm_reserve ((uword) base, size, log2_page_sz);

  if (base == (void *) ~0)
    return CLIB_MEM_VM_MAP_FAILED;

  base = mmap (base, size, PROT_READ | PROT_WRITE, mmap_flags, fd, offset);

  if (base == MAP_FAILED)
    return CLIB_MEM_VM_MAP_FAILED;

  if (is_huge && (mlock (base, size) != 0))
    {
      munmap (base, size);
      return CLIB_MEM_VM_MAP_FAILED;
    }

  hdr = mmap (base - sys_page_sz, sys_page_sz, PROT_READ | PROT_WRITE,
	      MAP_ANONYMOUS | MAP_PRIVATE | MAP_FIXED, -1, 0);

  if (hdr != base - sys_page_sz)
    {
      munmap (base, size);
      return CLIB_MEM_VM_MAP_FAILED;
    }

  map_lock ();

  if (mm->last_map)
    {
      mprotect (mm->last_map, sys_page_sz, PROT_READ | PROT_WRITE);
      mm->last_map->next = hdr;
      mprotect (mm->last_map, sys_page_sz, PROT_NONE);
    }
  else
    mm->first_map = hdr;

  clib_mem_unpoison (hdr, sys_page_sz);
  hdr->next = 0;
  hdr->prev = mm->last_map;
  snprintf (hdr->name, CLIB_VM_MAP_HDR_NAME_MAX_LEN - 1, "%s", (char *) name);
  mm->last_map = hdr;

  hdr->base_addr = (uword) base;
  hdr->log2_page_sz = log2_page_sz;
  hdr->num_pages = size >> log2_page_sz;
  hdr->fd = fd;
  hdr->name[CLIB_VM_MAP_HDR_NAME_MAX_LEN - 1] = 0;
  mprotect (hdr, sys_page_sz, PROT_NONE);

  map_unlock ();

  clib_mem_unpoison (base, size);
  return base;
}

__clib_export int
clib_mem_vm_unmap (void *base)
{
  clib_mem_main_t *mm = &clib_mem_main;
  uword size, sys_page_sz = 1ULL << mm->log2_page_sz;
  clib_mem_vm_map_hdr_t *hdr = base - sys_page_sz;
  ;

  map_lock ();
  if (mprotect (hdr, sys_page_sz, PROT_READ | PROT_WRITE) != 0)
    goto out;

  size = hdr->num_pages << hdr->log2_page_sz;
  if (munmap ((void *) hdr->base_addr, size) != 0)
    goto out;

  if (hdr->next)
    {
      mprotect (hdr->next, sys_page_sz, PROT_READ | PROT_WRITE);
      hdr->next->prev = hdr->prev;
      mprotect (hdr->next, sys_page_sz, PROT_NONE);
    }
  else
    mm->last_map = hdr->prev;

  if (hdr->prev)
    {
      mprotect (hdr->prev, sys_page_sz, PROT_READ | PROT_WRITE);
      hdr->prev->next = hdr->next;
      mprotect (hdr->prev, sys_page_sz, PROT_NONE);
    }
  else
    mm->first_map = hdr->next;

  map_unlock ();

  if (munmap (hdr, sys_page_sz) != 0)
    return CLIB_MEM_ERROR;

  return 0;
out:
  map_unlock ();
  return CLIB_MEM_ERROR;
}

__clib_export void
clib_mem_get_page_stats (void *start, clib_mem_page_sz_t log2_page_size,
			 uword n_pages, clib_mem_page_stats_t *stats)
{
  int i, *status = 0;
  void **ptr = 0;

  log2_page_size = clib_mem_log2_page_size_validate (log2_page_size);

  vec_validate (status, n_pages - 1);
  vec_validate (ptr, n_pages - 1);

  for (i = 0; i < n_pages; i++)
    ptr[i] = start + (i << log2_page_size);

  clib_memset (stats, 0, sizeof (clib_mem_page_stats_t));
  stats->total = n_pages;
  stats->log2_page_sz = log2_page_size;

  /*
   * TODO: Until FreeBSD has support for tracking pages in NUMA domains just
   * return that all are unknown for the statsistics.
   */
  stats->unknown = n_pages;

  vec_free (status);
  vec_free (ptr);
}

__clib_export u64 *
clib_mem_vm_get_paddr (void *mem, clib_mem_page_sz_t log2_page_size,
		       int n_pages)
{
  struct mem_extract meme;
  int pagesize = sysconf (_SC_PAGESIZE);
  int fd;
  int i;
  u64 *r = 0;

  log2_page_size = clib_mem_log2_page_size_validate (log2_page_size);

  if ((fd = open ((char *) "/dev/mem", O_RDONLY)) == -1)
    return 0;

  for (i = 0; i < n_pages; i++)
    {
      meme.me_vaddr = pointer_to_uword (mem) + (((u64) i) << log2_page_size);

      if (ioctl (fd, MEM_EXTRACT_PADDR, &meme) == -1)
	goto done;
      vec_add1 (r, meme.me_paddr * pagesize);
    }

done:
  close (fd);
  if (vec_len (r) != n_pages)
    {
      vec_free (r);
      return 0;
    }
  return r;
}

__clib_export int
clib_mem_set_numa_affinity (u8 numa_node, int force)
{
  /* TODO: Not yet implemented */
  return CLIB_MEM_ERROR;
}

__clib_export int
clib_mem_set_default_numa_affinity ()
{
  /* TODO: Not yet implemented */
  return 0;
}