summaryrefslogtreecommitdiffstats
path: root/src/vppinfra/valloc.c
blob: 927802bfeb16db7aefa7355c7171f19b21cc7d85 (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
/*
 * Copyright (c) 2018 Cisco and/or its affiliates.
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at:
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

#include <vppinfra/valloc.h>

/** @file
    @brief Simple first-fit virtual space allocator
*/

/** Add a chunk of memory to a virtual allocation arena
    @param vam - clib_valloc_main_t * pointer to the allocation arena
    @param template - clib_valloc_chunk_t * pointer to a template chunk which
    describes the virtual address range to add

    @note only the baseva and size member of the template chunk are significant
    It's perfectly OK for the new chunk to be discontinuous with previous
    chunks, the chunk fusion algorithm won't merge them.
 */

void
clib_valloc_add_chunk (clib_valloc_main_t * vam,
		       clib_valloc_chunk_t * template)
{
  clib_valloc_chunk_t *ch, *new_ch;
  u32 index;

  ASSERT (vam->flags & CLIB_VALLOC_INITIALIZED);

  clib_spinlock_lock_if_init (&vam->lock);

  /* Add at the beginning, or at the end... */
  index = vam->first_index;

  /*
   * Make sure we're not trying to add an overlapping chunk..
   * It's worth checking, because someone will eventually do that.
   */
  if (CLIB_DEBUG > 0 && index != ~0)
    {
      while (index != ~0)
	{
	  ch = pool_elt_at_index (vam->chunks, index);
	  ASSERT (template->baseva < ch->baseva || template->baseva >=
		  (ch->baseva + ch->size));
	  ASSERT (template->baseva + template->size < ch->baseva ||
		  template->baseva + template->size >=
		  (ch->baseva + ch->size));
	  index = ch->next;
	}
      index = vam->first_index;
    }

  if (index != ~0)
    ch = pool_elt_at_index (vam->chunks, index);

  if (index == ~0 || template->baseva < ch->baseva)
    {
      pool_get (vam->chunks, new_ch);
      clib_memset (new_ch, 0, sizeof (*new_ch));

      if (index != ~0)
	{
	  ch = pool_elt_at_index (vam->chunks, index);

	  new_ch->next = index;
	  new_ch->prev = ~0;
	  ch->prev = new_ch - vam->chunks;
	}
      else
	{
	  new_ch->next = new_ch->prev = ~0;
	}

      new_ch->baseva = template->baseva;
      new_ch->size = template->size;

      vam->first_index = new_ch - vam->chunks;

      hash_set (vam->chunk_index_by_baseva, new_ch->baseva, vam->first_index);
    }
  else
    {
      /* Walk to the end of the chunk chain */
      while (index != ~0)
	{
	  ch = pool_elt_at_index (vam->chunks, index);
	  index = ch->next;
	}
      /* we want the last chunk index */
      index = ch - vam->chunks;

      pool_get (vam->chunks, new_ch);
      clib_memset (new_ch, 0, sizeof (*new_ch));

      ch = pool_elt_at_index (vam->chunks, index);

      new_ch->next = ~0;
      new_ch->prev = index;
      ch->next = new_ch - vam->chunks;

      new_ch->baseva = template->baseva;
      new_ch->size = template->size;

      hash_set (vam->chunk_index_by_baseva, new_ch->baseva,
		new_ch - vam->chunks);
    }

  clib_spinlock_unlock_if_init (&vam->lock);
}

/** Initialize a virtual memory allocation arena
    @param vam - clib_valloc_main_t * pointer to the arena to initialize
    @param template - clib_valloc_chunk_t * pointer to a template chunk which
    describes the initial virtual address range
*/
void
clib_valloc_init (clib_valloc_main_t * vam, clib_valloc_chunk_t * template,
		  int need_lock)
{
  ASSERT (template && template->baseva && template->size);
  clib_memset (vam, 0, sizeof (*vam));
  if (need_lock)
    clib_spinlock_init (&vam->lock);

  vam->chunk_index_by_baseva = hash_create (0, sizeof (uword));
  vam->first_index = ~0;
  vam->flags |= CLIB_VALLOC_INITIALIZED;

  clib_valloc_add_chunk (vam, template);
}

/** Allocate virtual space
    @param vam - clib_valloc_main_t * pointer to the allocation arena
    @param size - u64 number of bytes to allocate
    @os_out_of_memory_on_failure - 1=> panic on allocation failure
    @return uword allocated space, 0=> failure
*/
uword
clib_valloc_alloc (clib_valloc_main_t * vam, uword size,
		   int os_out_of_memory_on_failure)
{
  clib_valloc_chunk_t *ch, *new_ch, *next_ch;
  u32 index;

  clib_spinlock_lock_if_init (&vam->lock);

  index = vam->first_index;

  while (index != ~0)
    {
      ch = pool_elt_at_index (vam->chunks, index);
      /* If the chunk is free... */
      if ((ch->flags & CLIB_VALLOC_BUSY) == 0)
	{
	  /* Too small? */
	  if (ch->size < size)
	    goto next_chunk;
	  /* Exact match? */
	  if (ch->size == size)
	    {
	      ch->flags |= CLIB_VALLOC_BUSY;
	      clib_spinlock_unlock_if_init (&vam->lock);
	      return ch->baseva;
	    }
	  /*
	   * The current free chunk is larger than necessary, split the block.
	   */
	  pool_get (vam->chunks, new_ch);
	  /* ch might have just moved */
	  ch = pool_elt_at_index (vam->chunks, index);
	  clib_memset (new_ch, 0, sizeof (*new_ch));
	  new_ch->next = new_ch->prev = ~0;
	  new_ch->baseva = ch->baseva + size;
	  new_ch->size = ch->size - size;
	  ch->size = size;

	  /* Insert into doubly-linked list */
	  new_ch->next = ch->next;
	  new_ch->prev = ch - vam->chunks;

	  if (ch->next != ~0)
	    {
	      next_ch = pool_elt_at_index (vam->chunks, ch->next);
	      next_ch->prev = new_ch - vam->chunks;
	    }
	  ch->next = new_ch - vam->chunks;

	  hash_set (vam->chunk_index_by_baseva, new_ch->baseva,
		    new_ch - vam->chunks);

	  ch->flags |= CLIB_VALLOC_BUSY;
	  clib_spinlock_unlock_if_init (&vam->lock);
	  return ch->baseva;
	}

    next_chunk:
      index = ch->next;
    }
  clib_spinlock_unlock_if_init (&vam->lock);

  if (os_out_of_memory_on_failure)
    os_out_of_memory ();

  return 0;
}


/** Free virtual space
    @param vam - clib_valloc_main_t * pointer to the allocation arena
    @param baseva - uword base virtual address of the returned space
    @return uword - size of the freed virtual chunk
    @note the size is returned since we know it / in case the caller
    doesn't memorize chunk sizes
*/
uword
clib_valloc_free (clib_valloc_main_t * vam, uword baseva)
{
  clib_valloc_chunk_t *ch, *prev_ch, *next_ch, *n2_ch;
  u32 index;
  uword return_size = 0;
  uword *p;

  clib_spinlock_lock_if_init (&vam->lock);

  p = hash_get (vam->chunk_index_by_baseva, baseva);

  /* Check even in production images */
  if (p == 0)
    os_panic ();

  index = p[0];

  ch = pool_elt_at_index (vam->chunks, index);

  return_size = ch->size;

  ASSERT (ch->baseva == baseva);
  ASSERT ((ch->flags & CLIB_VALLOC_BUSY) != 0);

  ch->flags &= ~CLIB_VALLOC_BUSY;

  /* combine with previous entry? */
  if (ch->prev != ~0)
    {
      prev_ch = pool_elt_at_index (vam->chunks, ch->prev);
      /*
       * Previous item must be free as well, and
       * tangent to this block.
       */
      if ((prev_ch->flags & CLIB_VALLOC_BUSY) == 0
	  && ((prev_ch->baseva + prev_ch->size) == ch->baseva))
	{
	  hash_unset (vam->chunk_index_by_baseva, baseva);
	  prev_ch->size += ch->size;
	  prev_ch->next = ch->next;
	  if (ch->next != ~0)
	    {
	      next_ch = pool_elt_at_index (vam->chunks, ch->next);
	      next_ch->prev = ch->prev;
	    }
	  ASSERT (ch - vam->chunks != vam->first_index);
	  clib_memset (ch, 0xfe, sizeof (*ch));
	  pool_put (vam->chunks, ch);
	  /* See about combining with next elt */
	  ch = prev_ch;
	}
    }

  /* Combine with next entry? */
  if (ch->next != ~0)
    {
      next_ch = pool_elt_at_index (vam->chunks, ch->next);

      if ((next_ch->flags & CLIB_VALLOC_BUSY) == 0
	  && ((ch->baseva + ch->size) == next_ch->baseva))
	{
	  hash_unset (vam->chunk_index_by_baseva, next_ch->baseva);
	  ch->size += next_ch->size;
	  ch->next = next_ch->next;
	  if (ch->next != ~0)
	    {
	      n2_ch = pool_elt_at_index (vam->chunks, next_ch->next);
	      n2_ch->prev = ch - vam->chunks;
	    }
	  ASSERT (next_ch - vam->chunks != vam->first_index);
	  clib_memset (next_ch, 0xfe, sizeof (*ch));
	  pool_put (vam->chunks, next_ch);
	}
    }

  clib_spinlock_unlock_if_init (&vam->lock);
  return return_size;
}

/** format a virtual allocation arena (varargs)
    @param vam - clib_valloc_main_t pointer to the arena to format
    @param verbose - int - verbosity level
    @return u8 vector
*/
u8 *
format_valloc (u8 * s, va_list * va)
{
  clib_valloc_main_t *vam = va_arg (*va, clib_valloc_main_t *);
  int verbose = va_arg (*va, int);
  clib_valloc_chunk_t *ch;
  u32 index;
  uword *p;

  clib_spinlock_lock_if_init (&vam->lock);

  s = format (s, "%d chunks, first index %d\n",
	      pool_elts (vam->chunks), vam->first_index);

  if (verbose)
    {
      index = vam->first_index;
      while (index != ~0)
	{
	  ch = pool_elt_at_index (vam->chunks, index);

	  s = format (s, "[%d] base %llx size %llx (%lld) prev %d %s\n",
		      index, ch->baseva, ch->size, ch->size, ch->prev,
		      (ch->flags & CLIB_VALLOC_BUSY) ? "busy" : "free");

	  p = hash_get (vam->chunk_index_by_baseva, ch->baseva);
	  if (p == 0)
	    {
	      s = format (s, "   BUG: baseva not in hash table!\n");
	    }
	  else if (p[0] != index)
	    {
	      s = format (s, "   BUG: baseva in hash table %d not %d!\n",
			  p[0], index);
	    }
	  index = ch->next;
	}
    }

  clib_spinlock_unlock_if_init (&vam->lock);

  return s;
}

/*
 * fd.io coding-style-patch-verification: ON
 *
 * Local Variables:
 * eval: (c-set-style "gnu")
 * End:
 */
u16 ip6_tcp_udp_icmp_compute_checksum (vlib_main_t * vm, vlib_buffer_t * p0, ip6_header_t * ip0, int *bogus_lengthp); void ip6_register_protocol (u32 protocol, u32 node_index); void ip6_unregister_protocol (u32 protocol); void ip6_local_hop_by_hop_register_protocol (u32 protocol, u32 node_index); serialize_function_t serialize_vnet_ip6_main, unserialize_vnet_ip6_main; void ip6_ethernet_update_adjacency (vnet_main_t * vnm, u32 sw_if_index, u32 ai); always_inline void ip6_link_local_address_from_ethernet_mac_address (ip6_address_t * ip, u8 * mac) { ip->as_u64[0] = clib_host_to_net_u64 (0xFE80000000000000ULL); /* Invert the "u" bit */ ip->as_u8[8] = mac[0] ^ (1 << 1); ip->as_u8[9] = mac[1]; ip->as_u8[10] = mac[2]; ip->as_u8[11] = 0xFF; ip->as_u8[12] = 0xFE; ip->as_u8[13] = mac[3]; ip->as_u8[14] = mac[4]; ip->as_u8[15] = mac[5]; } always_inline void ip6_ethernet_mac_address_from_link_local_address (u8 * mac, ip6_address_t * ip) { /* Invert the previously inverted "u" bit */ mac[0] = ip->as_u8[8] ^ (1 << 1); mac[1] = ip->as_u8[9]; mac[2] = ip->as_u8[10]; mac[3] = ip->as_u8[13]; mac[4] = ip->as_u8[14]; mac[5] = ip->as_u8[15]; } int vnet_set_ip6_flow_hash (u32 table_id, flow_hash_config_t flow_hash_config); clib_error_t *enable_ip6_interface (vlib_main_t * vm, u32 sw_if_index); clib_error_t *disable_ip6_interface (vlib_main_t * vm, u32 sw_if_index); int ip6_interface_enabled (vlib_main_t * vm, u32 sw_if_index); clib_error_t *set_ip6_link_local_address (vlib_main_t * vm, u32 sw_if_index, ip6_address_t * address); typedef int (*ip6_nd_change_event_cb_t) (u32 pool_index, const mac_address_t * new_mac, u32 sw_if_index, const ip6_address_t * address); int vnet_add_del_ip6_nd_change_event (vnet_main_t * vnm, ip6_nd_change_event_cb_t data_callback, u32 pid, void *address_arg, uword node_index, uword type_opaque, uword data, int is_add); int vnet_ip6_nd_term (vlib_main_t * vm, vlib_node_runtime_t * node, vlib_buffer_t * p0, ethernet_header_t * eth, ip6_header_t * ip, u32 sw_if_index, u16 bd_index); void send_ip6_na (vlib_main_t * vm, u32 sw_if_index); void send_ip6_na_w_addr (vlib_main_t * vm, const ip6_address_t * addr, u32 sw_if_index); u8 *format_ip6_forward_next_trace (u8 * s, va_list * args); u32 ip6_tcp_udp_icmp_validate_checksum (vlib_main_t * vm, vlib_buffer_t * p0); void ip6_punt_policer_add_del (u8 is_add, u32 policer_index); void ip6_punt_redirect_add (u32 rx_sw_if_index, u32 tx_sw_if_index, ip46_address_t * nh); void ip6_punt_redirect_add_paths (u32 rx_sw_if_index, fib_route_path_t * paths); void ip6_punt_redirect_del (u32 rx_sw_if_index); int vnet_set_ip6_classify_intfc (vlib_main_t * vm, u32 sw_if_index, u32 table_index); extern vlib_node_registration_t ip6_lookup_node; /* Compute flow hash. We'll use it to select which Sponge to use for this flow. And other things. */ always_inline u32 ip6_compute_flow_hash (const ip6_header_t * ip, flow_hash_config_t flow_hash_config) { tcp_header_t *tcp; u64 a, b, c; u64 t1, t2; uword is_tcp_udp = 0; u8 protocol = ip->protocol; if (PREDICT_TRUE ((ip->protocol == IP_PROTOCOL_TCP) || (ip->protocol == IP_PROTOCOL_UDP))) { is_tcp_udp = 1; tcp = (void *) (ip + 1); } else if (ip->protocol == IP_PROTOCOL_IP6_HOP_BY_HOP_OPTIONS) { ip6_hop_by_hop_header_t *hbh = (ip6_hop_by_hop_header_t *) (ip + 1); if ((hbh->protocol == IP_PROTOCOL_TCP) || (hbh->protocol == IP_PROTOCOL_UDP)) { is_tcp_udp = 1; tcp = (tcp_header_t *) ((u8 *) hbh + ((hbh->length + 1) << 3)); } protocol = hbh->protocol; } t1 = (ip->src_address.as_u64[0] ^ ip->src_address.as_u64[1]); t1 = (flow_hash_config & IP_FLOW_HASH_SRC_ADDR) ? t1 : 0; t2 = (ip->dst_address.as_u64[0] ^ ip->dst_address.as_u64[1]); t2 = (flow_hash_config & IP_FLOW_HASH_DST_ADDR) ? t2 : 0; a = (flow_hash_config & IP_FLOW_HASH_REVERSE_SRC_DST) ? t2 : t1; b = (flow_hash_config & IP_FLOW_HASH_REVERSE_SRC_DST) ? t1 : t2; t1 = is_tcp_udp ? tcp->src : 0; t2 = is_tcp_udp ? tcp->dst : 0; t1 = (flow_hash_config & IP_FLOW_HASH_SRC_PORT) ? t1 : 0; t2 = (flow_hash_config & IP_FLOW_HASH_DST_PORT) ? t2 : 0; if (flow_hash_config & IP_FLOW_HASH_SYMMETRIC) { if (b < a) { c = a; a = b; b = c; } if (t2 < t1) { t2 += t1; t1 = t2 - t1; t2 = t2 - t1; } } b ^= (flow_hash_config & IP_FLOW_HASH_PROTO) ? protocol : 0; c = (flow_hash_config & IP_FLOW_HASH_REVERSE_SRC_DST) ? ((t1 << 16) | t2) : ((t2 << 16) | t1); hash_mix64 (a, b, c); return (u32) c; } /* ip6_locate_header * * This function is to search for the header specified by the protocol number * in find_hdr_type. * This is used to locate a specific IPv6 extension header * or to find transport layer header. * 1. If the find_hdr_type < 0 then it finds and returns the protocol number and * offset stored in *offset of the transport or ESP header in the chain if * found. * 2. If a header with find_hdr_type > 0 protocol number is found then the * offset is stored in *offset and protocol number of the header is * returned. * 3. If find_hdr_type is not found or packet is malformed or * it is a non-first fragment -1 is returned. */ always_inline int ip6_locate_header (vlib_buffer_t * p0, ip6_header_t * ip0, int find_hdr_type, u32 * offset) { u8 next_proto = ip0->protocol; u8 *next_header; u8 done = 0; u32 cur_offset; u8 *temp_nxthdr = 0; u32 exthdr_len = 0; next_header = ip6_next_header (ip0); cur_offset = sizeof (ip6_header_t); while (1) { done = (next_proto == find_hdr_type); if (PREDICT_FALSE (next_header >= (u8 *) vlib_buffer_get_current (p0) + p0->current_length)) { //A malicious packet could set an extension header with a too big size return (-1); } if (done) break; if ((!ip6_ext_hdr (next_proto)) || next_proto == IP_PROTOCOL_IP6_NONXT) { if (find_hdr_type < 0) break; return -1; } if (next_proto == IP_PROTOCOL_IPV6_FRAGMENTATION) { ip6_frag_hdr_t *frag_hdr = (ip6_frag_hdr_t *) next_header; u16 frag_off = ip6_frag_hdr_offset (frag_hdr); /* Non first fragment return -1 */ if (frag_off) return (-1); exthdr_len = sizeof (ip6_frag_hdr_t); temp_nxthdr = next_header + exthdr_len; } else if (next_proto == IP_PROTOCOL_IPSEC_AH) { exthdr_len = ip6_ext_authhdr_len (((ip6_ext_header_t *) next_header)); temp_nxthdr = next_header + exthdr_len; } else { exthdr_len = ip6_ext_header_len (((ip6_ext_header_t *) next_header)); temp_nxthdr = next_header + exthdr_len; } next_proto = ((ip6_ext_header_t *) next_header)->next_hdr; next_header = temp_nxthdr; cur_offset += exthdr_len; } *offset = cur_offset; return (next_proto); } u8 *format_ip6_hop_by_hop_ext_hdr (u8 * s, va_list * args); /* * Hop-by-Hop handling */ typedef struct { /* Array of function pointers to HBH option handling routines */ int (*options[256]) (vlib_buffer_t * b, ip6_header_t * ip, ip6_hop_by_hop_option_t * opt); u8 *(*trace[256]) (u8 * s, ip6_hop_by_hop_option_t * opt); uword next_override; } ip6_hop_by_hop_main_t; extern ip6_hop_by_hop_main_t ip6_hop_by_hop_main; int ip6_hbh_register_option (u8 option, int options (vlib_buffer_t * b, ip6_header_t * ip, ip6_hop_by_hop_option_t * opt), u8 * trace (u8 * s, ip6_hop_by_hop_option_t * opt)); int ip6_hbh_unregister_option (u8 option); void ip6_hbh_set_next_override (uword next); /** * Push IPv6 header to buffer * * @param vm - vlib_main * @param b - buffer to write the header to * @param src - source IP * @param dst - destination IP * @param prot - payload proto * * @return - pointer to start of IP header */ always_inline void * vlib_buffer_push_ip6 (vlib_main_t * vm, vlib_buffer_t * b, ip6_address_t * src, ip6_address_t * dst, int proto) { ip6_header_t *ip6h; u16 payload_length; /* make some room */ ip6h = vlib_buffer_push_uninit (b, sizeof (ip6_header_t)); ip6h->ip_version_traffic_class_and_flow_label = clib_host_to_net_u32 (0x6 << 28); /* calculate ip6 payload length */ payload_length = vlib_buffer_length_in_chain (vm, b); payload_length -= sizeof (*ip6h); ip6h->payload_length = clib_host_to_net_u16 (payload_length); ip6h->hop_limit = 0xff; ip6h->protocol = proto; clib_memcpy_fast (ip6h->src_address.as_u8, src->as_u8, sizeof (ip6h->src_address)); clib_memcpy_fast (ip6h->dst_address.as_u8, dst->as_u8, sizeof (ip6h->src_address)); b->flags |= VNET_BUFFER_F_IS_IP6; return ip6h; } #endif /* included_ip_ip6_h */ /* * fd.io coding-style-patch-verification: ON * * Local Variables: * eval: (c-set-style "gnu") * End: */