summary refs log tree commit diff
path: root/nscd/hstcache.c
blob: d4dd51f11aff7c2ee17526c1388c8f33674e3268 (plain) (blame)
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
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
/* Cache handling for host lookup.
   Copyright (C) 1998-2005, 2006, 2007, 2008 Free Software Foundation, Inc.
   This file is part of the GNU C Library.
   Contributed by Ulrich Drepper <drepper@cygnus.com>, 1998.

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published
   by the Free Software Foundation; version 2 of the License, or
   (at your option) any later version.

   This program is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
   GNU General Public License for more details.

   You should have received a copy of the GNU General Public License
   along with this program; if not, write to the Free Software Foundation,
   Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.  */

#include <alloca.h>
#include <assert.h>
#include <errno.h>
#include <error.h>
#include <libintl.h>
#include <netdb.h>
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <time.h>
#include <unistd.h>
#include <arpa/inet.h>
#include <arpa/nameser.h>
#include <sys/mman.h>
#include <stackinfo.h>

#include "nscd.h"
#include "dbg_log.h"
#ifdef HAVE_SENDFILE
# include <kernel-features.h>
#endif


/* This is the standard reply in case the service is disabled.  */
static const hst_response_header disabled =
{
  .version = NSCD_VERSION,
  .found = -1,
  .h_name_len = 0,
  .h_aliases_cnt = 0,
  .h_addrtype = -1,
  .h_length = -1,
  .h_addr_list_cnt = 0,
  .error = NETDB_INTERNAL
};

/* This is the struct describing how to write this record.  */
const struct iovec hst_iov_disabled =
{
  .iov_base = (void *) &disabled,
  .iov_len = sizeof (disabled)
};


/* This is the standard reply in case we haven't found the dataset.  */
static const hst_response_header notfound =
{
  .version = NSCD_VERSION,
  .found = 0,
  .h_name_len = 0,
  .h_aliases_cnt = 0,
  .h_addrtype = -1,
  .h_length = -1,
  .h_addr_list_cnt = 0,
  .error = HOST_NOT_FOUND
};


static void
cache_addhst (struct database_dyn *db, int fd, request_header *req,
	      const void *key, struct hostent *hst, uid_t owner,
	      struct hashentry *he, struct datahead *dh, int errval,
	      int32_t ttl)
{
  ssize_t total;
  ssize_t written;
  time_t t = time (NULL);

  /* We allocate all data in one memory block: the iov vector,
     the response header and the dataset itself.  */
  struct dataset
  {
    struct datahead head;
    hst_response_header resp;
    char strdata[0];
  } *dataset;

  assert (offsetof (struct dataset, resp) == offsetof (struct datahead, data));

  if (hst == NULL)
    {
      if (he != NULL && errval == EAGAIN)
	{
	  /* If we have an old record available but cannot find one
	     now because the service is not available we keep the old
	     record and make sure it does not get removed.  */
	  if (reload_count != UINT_MAX)
	    /* Do not reset the value if we never not reload the record.  */
	    dh->nreloads = reload_count - 1;

	  written = total = 0;
	}
      else
	{
	  /* We have no data.  This means we send the standard reply for this
	     case.  */
	  written = total = sizeof (notfound);

	  if (fd != -1)
	    written = TEMP_FAILURE_RETRY (send (fd, &notfound, total,
						MSG_NOSIGNAL));

	  dataset = mempool_alloc (db, sizeof (struct dataset) + req->key_len,
				   IDX_result_data);
	  /* If we cannot permanently store the result, so be it.  */
	  if (dataset != NULL)
	    {
	      dataset->head.allocsize = sizeof (struct dataset) + req->key_len;
	      dataset->head.recsize = total;
	      dataset->head.notfound = true;
	      dataset->head.nreloads = 0;
	      dataset->head.usable = true;

	      /* Compute the timeout time.  */
	      dataset->head.timeout = t + (ttl == INT32_MAX
					   ? db->negtimeout : ttl);

	      /* This is the reply.  */
	      memcpy (&dataset->resp, &notfound, total);

	      /* Copy the key data.  */
	      memcpy (dataset->strdata, key, req->key_len);

	      /* If necessary, we also propagate the data to disk.  */
	      if (db->persistent)
		{
		  // XXX async OK?
		  uintptr_t pval = (uintptr_t) dataset & ~pagesize_m1;
		  msync ((void *) pval,
			 ((uintptr_t) dataset & pagesize_m1)
			 + sizeof (struct dataset) + req->key_len, MS_ASYNC);
		}

	      /* Now get the lock to safely insert the records.  */
	      pthread_rwlock_rdlock (&db->lock);

	      (void) cache_add (req->type, &dataset->strdata, req->key_len,
				&dataset->head, true, db, owner, he == NULL);

	      pthread_rwlock_unlock (&db->lock);

	      /* Mark the old entry as obsolete.  */
	      if (dh != NULL)
		dh->usable = false;
	    }
	  else
	    ++db->head->addfailed;
	}
    }
  else
    {
      /* Determine the I/O structure.  */
      size_t h_name_len = strlen (hst->h_name) + 1;
      size_t h_aliases_cnt;
      uint32_t *h_aliases_len;
      size_t h_addr_list_cnt;
      int addr_list_type;
      char *addresses;
      char *aliases;
      char *key_copy = NULL;
      char *cp;
      size_t cnt;

      /* Determine the number of aliases.  */
      h_aliases_cnt = 0;
      for (cnt = 0; hst->h_aliases[cnt] != NULL; ++cnt)
	++h_aliases_cnt;
      /* Determine the length of all aliases.  */
      h_aliases_len = (uint32_t *) alloca (h_aliases_cnt * sizeof (uint32_t));
      total = 0;
      for (cnt = 0; cnt < h_aliases_cnt; ++cnt)
	{
	  h_aliases_len[cnt] = strlen (hst->h_aliases[cnt]) + 1;
	  total += h_aliases_len[cnt];
	}

      /* Determine the number of addresses.  */
      h_addr_list_cnt = 0;
      while (hst->h_addr_list[h_addr_list_cnt] != NULL)
	++h_addr_list_cnt;

      if (h_addr_list_cnt == 0)
	/* Invalid entry.  */
	return;

      total += (sizeof (struct dataset)
		+ h_name_len
		+ h_aliases_cnt * sizeof (uint32_t)
		+ h_addr_list_cnt * hst->h_length);
      written = total;

      /* If we refill the cache, first assume the reconrd did not
	 change.  Allocate memory on the cache since it is likely
	 discarded anyway.  If it turns out to be necessary to have a
	 new record we can still allocate real memory.  */
      bool alloca_used = false;
      dataset = NULL;

      /* If the record contains more than one IP address (used for
	 load balancing etc) don't cache the entry.  This is something
	 the current cache handling cannot handle and it is more than
	 questionable whether it is worthwhile complicating the cache
	 handling just for handling such a special case. */
      if (he == NULL && h_addr_list_cnt == 1)
	{
	  dataset = (struct dataset *) mempool_alloc (db,
						      total + req->key_len,
						      IDX_result_data);
	  if (dataset == NULL)
	    ++db->head->addfailed;
	}

      if (dataset == NULL)
	{
	  /* We cannot permanently add the result in the moment.  But
	     we can provide the result as is.  Store the data in some
	     temporary memory.  */
	  dataset = (struct dataset *) alloca (total + req->key_len);

	  /* We cannot add this record to the permanent database.  */
	  alloca_used = true;
	}

      dataset->head.allocsize = total + req->key_len;
      dataset->head.recsize = total - offsetof (struct dataset, resp);
      dataset->head.notfound = false;
      dataset->head.nreloads = he == NULL ? 0 : (dh->nreloads + 1);
      dataset->head.usable = true;

      /* Compute the timeout time.  */
      dataset->head.timeout = t + (ttl == INT32_MAX ? db->postimeout : ttl);

      dataset->resp.version = NSCD_VERSION;
      dataset->resp.found = 1;
      dataset->resp.h_name_len = h_name_len;
      dataset->resp.h_aliases_cnt = h_aliases_cnt;
      dataset->resp.h_addrtype = hst->h_addrtype;
      dataset->resp.h_length = hst->h_length;
      dataset->resp.h_addr_list_cnt = h_addr_list_cnt;
      dataset->resp.error = NETDB_SUCCESS;

      cp = dataset->strdata;

      cp = mempcpy (cp, hst->h_name, h_name_len);
      cp = mempcpy (cp, h_aliases_len, h_aliases_cnt * sizeof (uint32_t));

      /* The normal addresses first.  */
      addresses = cp;
      for (cnt = 0; cnt < h_addr_list_cnt; ++cnt)
	cp = mempcpy (cp, hst->h_addr_list[cnt], hst->h_length);

      /* Then the aliases.  */
      aliases = cp;
      for (cnt = 0; cnt < h_aliases_cnt; ++cnt)
	cp = mempcpy (cp, hst->h_aliases[cnt], h_aliases_len[cnt]);

      assert (cp
	      == dataset->strdata + total - offsetof (struct dataset,
						      strdata));

      /* If we are adding a GETHOSTBYNAME{,v6} entry we must be prepared
	 that the answer we get from the NSS does not contain the key
	 itself.  This is the case if the resolver is used and the name
	 is extended by the domainnames from /etc/resolv.conf.  Therefore
	 we explicitly add the name here.  */
      key_copy = memcpy (cp, key, req->key_len);

      /* Now we can determine whether on refill we have to create a new
	 record or not.  */
      if (he != NULL)
	{
	  assert (fd == -1);

	  if (total + req->key_len == dh->allocsize
	      && total - offsetof (struct dataset, resp) == dh->recsize
	      && memcmp (&dataset->resp, dh->data,
			 dh->allocsize - offsetof (struct dataset, resp)) == 0)
	    {
	      /* The data has not changed.  We will just bump the
		 timeout value.  Note that the new record has been
		 allocated on the stack and need not be freed.  */
	      assert (h_addr_list_cnt == 1);
	      dh->timeout = dataset->head.timeout;
	      ++dh->nreloads;
	    }
	  else
	    {
	      if (h_addr_list_cnt == 1)
		{
		  /* We have to create a new record.  Just allocate
		     appropriate memory and copy it.  */
		  struct dataset *newp
		    = (struct dataset *) mempool_alloc (db,
							total + req->key_len,
							IDX_result_data);
		  if (newp != NULL)
		    {
		      /* Adjust pointers into the memory block.  */
		      addresses = (char *) newp + (addresses
						   - (char *) dataset);
		      aliases = (char *) newp + (aliases - (char *) dataset);
		      assert (key_copy != NULL);
		      key_copy = (char *) newp + (key_copy - (char *) dataset);

		      dataset = memcpy (newp, dataset, total + req->key_len);
		      alloca_used = false;
		    }
		  else
		    ++db->head->addfailed;
		}

	      /* Mark the old record as obsolete.  */
	      dh->usable = false;
	    }
	}
      else
	{
	  /* We write the dataset before inserting it to the database
	     since while inserting this thread might block and so would
	     unnecessarily keep the receiver waiting.  */
	  assert (fd != -1);

#ifdef HAVE_SENDFILE
	  if (__builtin_expect (db->mmap_used, 1) && !alloca_used)
	    {
	      assert (db->wr_fd != -1);
	      assert ((char *) &dataset->resp > (char *) db->data);
	      assert ((char *) &dataset->resp - (char *) db->head
		      + total
		      <= (sizeof (struct database_pers_head)
			  + db->head->module * sizeof (ref_t)
			  + db->head->data_size));
	      written = sendfileall (fd, db->wr_fd,
				     (char *) &dataset->resp
				     - (char *) db->head, total);
# ifndef __ASSUME_SENDFILE
	      if (written == -1 && errno == ENOSYS)
		goto use_write;
# endif
	    }
	  else
# ifndef __ASSUME_SENDFILE
	  use_write:
# endif
#endif
	    written = writeall (fd, &dataset->resp, total);
	}

      /* Add the record to the database.  But only if it has not been
	 stored on the stack.

	 If the record contains more than one IP address (used for
	 load balancing etc) don't cache the entry.  This is something
	 the current cache handling cannot handle and it is more than
	 questionable whether it is worthwhile complicating the cache
	 handling just for handling such a special case. */
      if (! alloca_used)
	{
	  /* If necessary, we also propagate the data to disk.  */
	  if (db->persistent)
	    {
	      // XXX async OK?
	      uintptr_t pval = (uintptr_t) dataset & ~pagesize_m1;
	      msync ((void *) pval,
		     ((uintptr_t) dataset & pagesize_m1)
		     + total + req->key_len, MS_ASYNC);
	    }

	  addr_list_type = (hst->h_length == NS_INADDRSZ
			    ? GETHOSTBYADDR : GETHOSTBYADDRv6);

	  /* Now get the lock to safely insert the records.  */
	  pthread_rwlock_rdlock (&db->lock);

	  /* NB: the following code is really complicated.  It has
	     seemlingly duplicated code paths which do the same.  The
	     problem is that we always must add the hash table entry
	     with the FIRST flag set first.  Otherwise we get dangling
	     pointers in case memory allocation fails.  */
	  assert (hst->h_addr_list[1] == NULL);

	  /* Avoid adding names if more than one address is available.  See
	     above for more info.  */
	  assert (req->type == GETHOSTBYNAME
		  || req->type == GETHOSTBYNAMEv6
		  || req->type == GETHOSTBYADDR
		  || req->type == GETHOSTBYADDRv6);

	  (void) cache_add (req->type, key_copy, req->key_len,
			    &dataset->head, true, db, owner, he == NULL);

	  pthread_rwlock_unlock (&db->lock);
	}
    }

  if (__builtin_expect (written != total, 0) && debug_level > 0)
    {
      char buf[256];
      dbg_log (_("short write in %s: %s"),  __FUNCTION__,
	       strerror_r (errno, buf, sizeof (buf)));
    }
}


static int
lookup (int type, void *key, struct hostent *resultbufp, char *buffer,
	size_t buflen, struct hostent **hst, int32_t *ttlp)
{
  if (type == GETHOSTBYNAME)
    return __gethostbyname3_r (key, AF_INET, resultbufp, buffer, buflen, hst,
			       &h_errno, ttlp, NULL);
  if (type == GETHOSTBYNAMEv6)
    return __gethostbyname3_r (key, AF_INET6, resultbufp, buffer, buflen, hst,
			       &h_errno, ttlp, NULL);
  if (type == GETHOSTBYADDR)
    return __gethostbyaddr2_r (key, NS_INADDRSZ, AF_INET, resultbufp, buffer,
			       buflen, hst, &h_errno, ttlp);
  return __gethostbyaddr2_r (key, NS_IN6ADDRSZ, AF_INET6, resultbufp, buffer,
			     buflen, hst, &h_errno, ttlp);
}


static void
addhstbyX (struct database_dyn *db, int fd, request_header *req,
	   void *key, uid_t uid, struct hashentry *he, struct datahead *dh)
{
  /* Search for the entry matching the key.  Please note that we don't
     look again in the table whether the dataset is now available.  We
     simply insert it.  It does not matter if it is in there twice.  The
     pruning function only will look at the timestamp.  */
  int buflen = 1024;
  char *buffer = (char *) alloca (buflen);
  struct hostent resultbuf;
  struct hostent *hst;
  bool use_malloc = false;
  int errval = 0;
  int32_t ttl = INT32_MAX;

  if (__builtin_expect (debug_level > 0, 0))
    {
      const char *str;
      char buf[INET6_ADDRSTRLEN + 1];
      if (req->type == GETHOSTBYNAME || req->type == GETHOSTBYNAMEv6)
	str = key;
      else
	str = inet_ntop (req->type == GETHOSTBYADDR ? AF_INET : AF_INET6,
			 key, buf, sizeof (buf));

      if (he == NULL)
	dbg_log (_("Haven't found \"%s\" in hosts cache!"), (char *) str);
      else
	dbg_log (_("Reloading \"%s\" in hosts cache!"), (char *) str);
    }

  while (lookup (req->type, key, &resultbuf, buffer, buflen, &hst, &ttl) != 0
	 && h_errno == NETDB_INTERNAL
	 && (errval = errno) == ERANGE)
    {
      errno = 0;

      if (__builtin_expect (buflen > 32768, 0))
	{
	  char *old_buffer = buffer;
	  buflen *= 2;
	  buffer = (char *) realloc (use_malloc ? buffer : NULL, buflen);
	  if (buffer == NULL)
	    {
	      /* We ran out of memory.  We cannot do anything but
		 sending a negative response.  In reality this should
		 never happen.  */
	      hst = NULL;
	      buffer = old_buffer;

	      /* We set the error to indicate this is (possibly) a
		 temporary error and that it does not mean the entry
		 is not available at all.  */
	      errval = EAGAIN;
	      break;
	    }
	  use_malloc = true;
	}
      else
	/* Allocate a new buffer on the stack.  If possible combine it
	   with the previously allocated buffer.  */
	buffer = (char *) extend_alloca (buffer, buflen, 2 * buflen);
    }

  cache_addhst (db, fd, req, key, hst, uid, he, dh,
		h_errno == TRY_AGAIN ? errval : 0, ttl);

  if (use_malloc)
    free (buffer);
}


void
addhstbyname (struct database_dyn *db, int fd, request_header *req,
	      void *key, uid_t uid)
{
  addhstbyX (db, fd, req, key, uid, NULL, NULL);
}


void
readdhstbyname (struct database_dyn *db, struct hashentry *he,
		struct datahead *dh)
{
  request_header req =
    {
      .type = GETHOSTBYNAME,
      .key_len = he->len
    };

  addhstbyX (db, -1, &req, db->data + he->key, he->owner, he, dh);
}


void
addhstbyaddr (struct database_dyn *db, int fd, request_header *req,
	      void *key, uid_t uid)
{
  addhstbyX (db, fd, req, key, uid, NULL, NULL);
}


void
readdhstbyaddr (struct database_dyn *db, struct hashentry *he,
		struct datahead *dh)
{
  request_header req =
    {
      .type = GETHOSTBYADDR,
      .key_len = he->len
    };

  addhstbyX (db, -1, &req, db->data + he->key, he->owner, he, dh);
}


void
addhstbynamev6 (struct database_dyn *db, int fd, request_header *req,
		void *key, uid_t uid)
{
  addhstbyX (db, fd, req, key, uid, NULL, NULL);
}


void
readdhstbynamev6 (struct database_dyn *db, struct hashentry *he,
		  struct datahead *dh)
{
  request_header req =
    {
      .type = GETHOSTBYNAMEv6,
      .key_len = he->len
    };

  addhstbyX (db, -1, &req, db->data + he->key, he->owner, he, dh);
}


void
addhstbyaddrv6 (struct database_dyn *db, int fd, request_header *req,
		void *key, uid_t uid)
{
  addhstbyX (db, fd, req, key, uid, NULL, NULL);
}


void
readdhstbyaddrv6 (struct database_dyn *db, struct hashentry *he,
		  struct datahead *dh)
{
  request_header req =
    {
      .type = GETHOSTBYADDRv6,
      .key_len = he->len
    };

  addhstbyX (db, -1, &req, db->data + he->key, he->owner, he, dh);
}