| 1 | /* -*- mode: C++; c-basic-offset: 4; indent-tabs-mode: nil -*- */ |
| 2 | // vim: ft=cpp:expandtab:ts=8:sw=4:softtabstop=4: |
| 3 | #ident "$Id$" |
| 4 | /*====== |
| 5 | This file is part of PerconaFT. |
| 6 | |
| 7 | |
| 8 | Copyright (c) 2006, 2015, Percona and/or its affiliates. All rights reserved. |
| 9 | |
| 10 | PerconaFT is free software: you can redistribute it and/or modify |
| 11 | it under the terms of the GNU General Public License, version 2, |
| 12 | as published by the Free Software Foundation. |
| 13 | |
| 14 | PerconaFT is distributed in the hope that it will be useful, |
| 15 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 16 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 17 | GNU General Public License for more details. |
| 18 | |
| 19 | You should have received a copy of the GNU General Public License |
| 20 | along with PerconaFT. If not, see <http://www.gnu.org/licenses/>. |
| 21 | |
| 22 | ---------------------------------------- |
| 23 | |
| 24 | PerconaFT is free software: you can redistribute it and/or modify |
| 25 | it under the terms of the GNU Affero General Public License, version 3, |
| 26 | as published by the Free Software Foundation. |
| 27 | |
| 28 | PerconaFT is distributed in the hope that it will be useful, |
| 29 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 30 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 31 | GNU Affero General Public License for more details. |
| 32 | |
| 33 | You should have received a copy of the GNU Affero General Public License |
| 34 | along with PerconaFT. If not, see <http://www.gnu.org/licenses/>. |
| 35 | ======= */ |
| 36 | |
| 37 | #ident "Copyright (c) 2006, 2015, Percona and/or its affiliates. All rights reserved." |
| 38 | |
| 39 | /* Verify an FT. */ |
| 40 | /* Check: |
| 41 | * The tree is of uniform depth (and the height is correct at every node) |
| 42 | * For each pivot key: the max of the stuff to the left is <= the pivot key < the min of the stuff to the right. |
| 43 | * For each leaf node: All the keys are in strictly increasing order. |
| 44 | * For each nonleaf node: All the messages have keys that are between the associated pivot keys ( left_pivot_key < message <= right_pivot_key) |
| 45 | */ |
| 46 | |
| 47 | #include <my_global.h> |
| 48 | #include "ft/serialize/block_table.h" |
| 49 | #include "ft/ft.h" |
| 50 | #include "ft/ft-cachetable-wrappers.h" |
| 51 | #include "ft/ft-internal.h" |
| 52 | #include "ft/node.h" |
| 53 | |
| 54 | static int |
| 55 | compare_pairs (FT_HANDLE ft_handle, const DBT *a, const DBT *b) { |
| 56 | return ft_handle->ft->cmp(a, b); |
| 57 | } |
| 58 | |
| 59 | static int |
| 60 | compare_pair_to_key (FT_HANDLE ft_handle, const DBT *a, const void *key, uint32_t keylen) { |
| 61 | DBT y; |
| 62 | return ft_handle->ft->cmp(a, toku_fill_dbt(&y, key, keylen)); |
| 63 | } |
| 64 | |
| 65 | static int |
| 66 | verify_msg_in_child_buffer(FT_HANDLE ft_handle, enum ft_msg_type type, MSN msn, const void *key, uint32_t keylen, const void *UU(data), uint32_t UU(datalen), XIDS UU(xids), const DBT *lesser_pivot, const DBT *greatereq_pivot) |
| 67 | __attribute__((warn_unused_result)); |
| 68 | |
| 69 | UU() |
| 70 | static int |
| 71 | verify_msg_in_child_buffer(FT_HANDLE ft_handle, enum ft_msg_type type, MSN msn, const void *key, uint32_t keylen, const void *UU(data), uint32_t UU(datalen), XIDS UU(xids), const DBT *lesser_pivot, const DBT *greatereq_pivot) { |
| 72 | int result = 0; |
| 73 | if (msn.msn == ZERO_MSN.msn) |
| 74 | result = EINVAL; |
| 75 | switch (type) { |
| 76 | default: |
| 77 | break; |
| 78 | case FT_INSERT: |
| 79 | case FT_INSERT_NO_OVERWRITE: |
| 80 | case FT_DELETE_ANY: |
| 81 | case FT_ABORT_ANY: |
| 82 | case FT_COMMIT_ANY: |
| 83 | // verify key in bounds |
| 84 | if (lesser_pivot) { |
| 85 | int compare = compare_pair_to_key(ft_handle, lesser_pivot, key, keylen); |
| 86 | if (compare >= 0) |
| 87 | result = EINVAL; |
| 88 | } |
| 89 | if (result == 0 && greatereq_pivot) { |
| 90 | int compare = compare_pair_to_key(ft_handle, greatereq_pivot, key, keylen); |
| 91 | if (compare < 0) |
| 92 | result = EINVAL; |
| 93 | } |
| 94 | break; |
| 95 | } |
| 96 | return result; |
| 97 | } |
| 98 | |
| 99 | static DBT |
| 100 | get_ith_key_dbt (BASEMENTNODE bn, int i) { |
| 101 | DBT kdbt; |
| 102 | int r = bn->data_buffer.fetch_key_and_len(i, &kdbt.size, &kdbt.data); |
| 103 | invariant_zero(r); // this is a bad failure if it happens. |
| 104 | return kdbt; |
| 105 | } |
| 106 | |
| 107 | #define VERIFY_ASSERTION(predicate, i, string) ({ \ |
| 108 | if(!(predicate)) { \ |
| 109 | fprintf(stderr, "%s:%d: Looking at child %d of block %" PRId64 ": %s\n", __FILE__, __LINE__, i, blocknum.b, string); \ |
| 110 | result = TOKUDB_NEEDS_REPAIR; \ |
| 111 | if (!keep_going_on_failure) goto done; \ |
| 112 | }}) |
| 113 | |
| 114 | #define VERIFY_ASSERTION_BASEMENT(predicate, bn, entry, string) ({ \ |
| 115 | if(!(predicate)) { \ |
| 116 | fprintf(stderr, "%s:%d: Looking at block %" PRId64 " bn %d entry %d: %s\n", __FILE__, __LINE__, blocknum.b, bn, entry, string); \ |
| 117 | result = TOKUDB_NEEDS_REPAIR; \ |
| 118 | if (!keep_going_on_failure) goto done; \ |
| 119 | }}) |
| 120 | |
| 121 | struct { |
| 122 | int ; |
| 123 | MSN ; |
| 124 | message_buffer *; |
| 125 | }; |
| 126 | |
| 127 | // template-only function, but must be extern |
| 128 | int count_msgs(const int32_t &offset, const uint32_t UU(idx), struct count_msgs_extra *const e) |
| 129 | __attribute__((nonnull(3))); |
| 130 | int (const int32_t &offset, const uint32_t UU(idx), struct count_msgs_extra *const e) |
| 131 | { |
| 132 | MSN msn; |
| 133 | e->msg_buffer->get_message_key_msn(offset, nullptr, &msn); |
| 134 | if (msn.msn == e->msn.msn) { |
| 135 | e->count++; |
| 136 | } |
| 137 | return 0; |
| 138 | } |
| 139 | |
| 140 | struct { |
| 141 | message_buffer *; |
| 142 | bool ; |
| 143 | bool ; |
| 144 | int ; |
| 145 | int ; |
| 146 | BLOCKNUM ; |
| 147 | int ; |
| 148 | bool ; |
| 149 | }; |
| 150 | |
| 151 | int verify_message_tree(const int32_t &offset, const uint32_t UU(idx), struct verify_message_tree_extra *const e) __attribute__((nonnull(3))); |
| 152 | int (const int32_t &offset, const uint32_t UU(idx), struct verify_message_tree_extra *const e) |
| 153 | { |
| 154 | BLOCKNUM blocknum = e->blocknum; |
| 155 | int keep_going_on_failure = e->keep_going_on_failure; |
| 156 | int result = 0; |
| 157 | DBT k, v; |
| 158 | ft_msg msg = e->msg_buffer->get_message(offset, &k, &v); |
| 159 | bool is_fresh = e->msg_buffer->get_freshness(offset); |
| 160 | if (e->broadcast) { |
| 161 | VERIFY_ASSERTION(ft_msg_type_applies_all((enum ft_msg_type) msg.type()) || ft_msg_type_does_nothing((enum ft_msg_type) msg.type()), |
| 162 | e->i, "message found in broadcast list that is not a broadcast" ); |
| 163 | } else { |
| 164 | VERIFY_ASSERTION(ft_msg_type_applies_once((enum ft_msg_type) msg.type()), |
| 165 | e->i, "message found in fresh or stale message tree that does not apply once" ); |
| 166 | if (e->is_fresh) { |
| 167 | if (e->messages_have_been_moved) { |
| 168 | VERIFY_ASSERTION(is_fresh, |
| 169 | e->i, "message found in fresh message tree that is not fresh" ); |
| 170 | } |
| 171 | } else { |
| 172 | VERIFY_ASSERTION(!is_fresh, |
| 173 | e->i, "message found in stale message tree that is fresh" ); |
| 174 | } |
| 175 | } |
| 176 | done: |
| 177 | return result; |
| 178 | } |
| 179 | |
| 180 | int error_on_iter(const int32_t &UU(offset), const uint32_t UU(idx), void *UU(e)); |
| 181 | int error_on_iter(const int32_t &UU(offset), const uint32_t UU(idx), void *UU(e)) { |
| 182 | return TOKUDB_NEEDS_REPAIR; |
| 183 | } |
| 184 | |
| 185 | int verify_marked_messages(const int32_t &offset, const uint32_t UU(idx), struct verify_message_tree_extra *const e) __attribute__((nonnull(3))); |
| 186 | int (const int32_t &offset, const uint32_t UU(idx), struct verify_message_tree_extra *const e) |
| 187 | { |
| 188 | BLOCKNUM blocknum = e->blocknum; |
| 189 | int keep_going_on_failure = e->keep_going_on_failure; |
| 190 | int result = 0; |
| 191 | bool is_fresh = e->msg_buffer->get_freshness(offset); |
| 192 | VERIFY_ASSERTION(!is_fresh, e->i, "marked message found in the fresh message tree that is fresh" ); |
| 193 | done: |
| 194 | return result; |
| 195 | } |
| 196 | |
| 197 | template<typename verify_omt_t> |
| 198 | static int |
| 199 | verify_sorted_by_key_msn(FT_HANDLE ft_handle, message_buffer *msg_buffer, const verify_omt_t &mt) { |
| 200 | int result = 0; |
| 201 | size_t last_offset = 0; |
| 202 | for (uint32_t i = 0; i < mt.size(); i++) { |
| 203 | int32_t offset; |
| 204 | int r = mt.fetch(i, &offset); |
| 205 | assert_zero(r); |
| 206 | if (i > 0) { |
| 207 | struct toku_msg_buffer_key_msn_cmp_extra (ft_handle->ft->cmp, msg_buffer); |
| 208 | if (toku_msg_buffer_key_msn_cmp(extra, last_offset, offset) >= 0) { |
| 209 | result = TOKUDB_NEEDS_REPAIR; |
| 210 | break; |
| 211 | } |
| 212 | } |
| 213 | last_offset = offset; |
| 214 | } |
| 215 | return result; |
| 216 | } |
| 217 | |
| 218 | template<typename count_omt_t> |
| 219 | static int |
| 220 | count_eq_key_msn(FT_HANDLE ft_handle, message_buffer *msg_buffer, const count_omt_t &mt, const DBT *key, MSN msn) { |
| 221 | struct toku_msg_buffer_key_msn_heaviside_extra (ft_handle->ft->cmp, msg_buffer, key, msn); |
| 222 | int r = mt.template find_zero<struct toku_msg_buffer_key_msn_heaviside_extra, toku_msg_buffer_key_msn_heaviside>(extra, nullptr, nullptr); |
| 223 | int count; |
| 224 | if (r == 0) { |
| 225 | count = 1; |
| 226 | } else { |
| 227 | assert(r == DB_NOTFOUND); |
| 228 | count = 0; |
| 229 | } |
| 230 | return count; |
| 231 | } |
| 232 | |
| 233 | void |
| 234 | toku_get_node_for_verify( |
| 235 | BLOCKNUM blocknum, |
| 236 | FT_HANDLE ft_handle, |
| 237 | FTNODE* nodep |
| 238 | ) |
| 239 | { |
| 240 | uint32_t fullhash = toku_cachetable_hash(ft_handle->ft->cf, blocknum); |
| 241 | ftnode_fetch_extra bfe; |
| 242 | bfe.create_for_full_read(ft_handle->ft); |
| 243 | toku_pin_ftnode( |
| 244 | ft_handle->ft, |
| 245 | blocknum, |
| 246 | fullhash, |
| 247 | &bfe, |
| 248 | PL_WRITE_EXPENSIVE, // may_modify_node |
| 249 | nodep, |
| 250 | false |
| 251 | ); |
| 252 | } |
| 253 | |
| 254 | struct verify_msg_fn { |
| 255 | FT_HANDLE ft_handle; |
| 256 | NONLEAF_CHILDINFO bnc; |
| 257 | const DBT *curr_less_pivot; |
| 258 | const DBT *curr_geq_pivot; |
| 259 | BLOCKNUM blocknum; |
| 260 | MSN this_msn; |
| 261 | int verbose; |
| 262 | int keep_going_on_failure; |
| 263 | bool messages_have_been_moved; |
| 264 | |
| 265 | MSN last_msn; |
| 266 | int msg_i; |
| 267 | int result = 0; // needed by VERIFY_ASSERTION |
| 268 | |
| 269 | verify_msg_fn(FT_HANDLE handle, NONLEAF_CHILDINFO nl, const DBT *less, const DBT *geq, |
| 270 | BLOCKNUM b, MSN tmsn, int v, int k, bool m) : |
| 271 | ft_handle(handle), bnc(nl), curr_less_pivot(less), curr_geq_pivot(geq), |
| 272 | blocknum(b), this_msn(tmsn), verbose(v), keep_going_on_failure(k), messages_have_been_moved(m), last_msn(ZERO_MSN), msg_i(0) { |
| 273 | } |
| 274 | |
| 275 | int operator()(const ft_msg &msg, bool is_fresh) { |
| 276 | enum ft_msg_type type = (enum ft_msg_type) msg.type(); |
| 277 | MSN msn = msg.msn(); |
| 278 | XIDS xid = msg.xids(); |
| 279 | const void *key = msg.kdbt()->data; |
| 280 | const void *data = msg.vdbt()->data; |
| 281 | uint32_t keylen = msg.kdbt()->size; |
| 282 | uint32_t datalen = msg.vdbt()->size; |
| 283 | |
| 284 | int r = verify_msg_in_child_buffer(ft_handle, type, msn, key, keylen, data, datalen, xid, |
| 285 | curr_less_pivot, |
| 286 | curr_geq_pivot); |
| 287 | VERIFY_ASSERTION(r == 0, msg_i, "A message in the buffer is out of place" ); |
| 288 | VERIFY_ASSERTION((msn.msn > last_msn.msn), msg_i, "msn per msg must be monotonically increasing toward newer messages in buffer" ); |
| 289 | VERIFY_ASSERTION((msn.msn <= this_msn.msn), msg_i, "all messages must have msn within limit of this node's max_msn_applied_to_node_in_memory" ); |
| 290 | if (ft_msg_type_applies_once(type)) { |
| 291 | int count; |
| 292 | DBT keydbt; |
| 293 | toku_fill_dbt(&keydbt, key, keylen); |
| 294 | int total_count = 0; |
| 295 | count = count_eq_key_msn(ft_handle, &bnc->msg_buffer, bnc->fresh_message_tree, toku_fill_dbt(&keydbt, key, keylen), msn); |
| 296 | total_count += count; |
| 297 | if (is_fresh) { |
| 298 | VERIFY_ASSERTION(count == 1, msg_i, "a fresh message was not found in the fresh message tree" ); |
| 299 | } else if (messages_have_been_moved) { |
| 300 | VERIFY_ASSERTION(count == 0, msg_i, "a stale message was found in the fresh message tree" ); |
| 301 | } |
| 302 | VERIFY_ASSERTION(count <= 1, msg_i, "a message was found multiple times in the fresh message tree" ); |
| 303 | count = count_eq_key_msn(ft_handle, &bnc->msg_buffer, bnc->stale_message_tree, &keydbt, msn); |
| 304 | |
| 305 | total_count += count; |
| 306 | if (is_fresh) { |
| 307 | VERIFY_ASSERTION(count == 0, msg_i, "a fresh message was found in the stale message tree" ); |
| 308 | } else if (messages_have_been_moved) { |
| 309 | VERIFY_ASSERTION(count == 1, msg_i, "a stale message was not found in the stale message tree" ); |
| 310 | } |
| 311 | VERIFY_ASSERTION(count <= 1, msg_i, "a message was found multiple times in the stale message tree" ); |
| 312 | |
| 313 | VERIFY_ASSERTION(total_count <= 1, msg_i, "a message was found in both message trees (or more than once in a single tree)" ); |
| 314 | VERIFY_ASSERTION(total_count >= 1, msg_i, "a message was not found in either message tree" ); |
| 315 | } else { |
| 316 | VERIFY_ASSERTION(ft_msg_type_applies_all(type) || ft_msg_type_does_nothing(type), msg_i, "a message was found that does not apply either to all or to only one key" ); |
| 317 | struct count_msgs_extra = { .count = 0, .msn = msn, .msg_buffer = &bnc->msg_buffer }; |
| 318 | bnc->broadcast_list.iterate<struct count_msgs_extra, count_msgs>(&extra); |
| 319 | VERIFY_ASSERTION(extra.count == 1, msg_i, "a broadcast message was not found in the broadcast list" ); |
| 320 | } |
| 321 | last_msn = msn; |
| 322 | msg_i++; |
| 323 | done: |
| 324 | return result; |
| 325 | } |
| 326 | }; |
| 327 | |
| 328 | static int |
| 329 | toku_verify_ftnode_internal(FT_HANDLE ft_handle, |
| 330 | MSN rootmsn, MSN parentmsn_with_messages, bool messages_exist_above, |
| 331 | FTNODE node, int height, |
| 332 | const DBT *lesser_pivot, // Everything in the subtree should be > lesser_pivot. (lesser_pivot==NULL if there is no lesser pivot.) |
| 333 | const DBT *greatereq_pivot, // Everything in the subtree should be <= lesser_pivot. (lesser_pivot==NULL if there is no lesser pivot.) |
| 334 | int verbose, int keep_going_on_failure, bool messages_have_been_moved) |
| 335 | { |
| 336 | int result=0; |
| 337 | MSN this_msn; |
| 338 | BLOCKNUM blocknum = node->blocknum; |
| 339 | |
| 340 | //printf("%s:%d pin %p\n", __FILE__, __LINE__, node_v); |
| 341 | toku_ftnode_assert_fully_in_memory(node); |
| 342 | this_msn = node->max_msn_applied_to_node_on_disk; |
| 343 | |
| 344 | if (height >= 0) { |
| 345 | invariant(height == node->height); // this is a bad failure if wrong |
| 346 | } |
| 347 | if (node->height > 0 && messages_exist_above) { |
| 348 | VERIFY_ASSERTION((parentmsn_with_messages.msn >= this_msn.msn), 0, "node msn must be descending down tree, newest messages at top" ); |
| 349 | } |
| 350 | // Verify that all the pivot keys are in order. |
| 351 | for (int i = 0; i < node->n_children-2; i++) { |
| 352 | DBT x, y; |
| 353 | int compare = compare_pairs(ft_handle, node->pivotkeys.fill_pivot(i, &x), node->pivotkeys.fill_pivot(i + 1, &y)); |
| 354 | VERIFY_ASSERTION(compare < 0, i, "Value is >= the next value" ); |
| 355 | } |
| 356 | // Verify that all the pivot keys are lesser_pivot < pivot <= greatereq_pivot |
| 357 | for (int i = 0; i < node->n_children-1; i++) { |
| 358 | DBT x; |
| 359 | if (lesser_pivot) { |
| 360 | int compare = compare_pairs(ft_handle, lesser_pivot, node->pivotkeys.fill_pivot(i, &x)); |
| 361 | VERIFY_ASSERTION(compare < 0, i, "Pivot is >= the lower-bound pivot" ); |
| 362 | } |
| 363 | if (greatereq_pivot) { |
| 364 | int compare = compare_pairs(ft_handle, greatereq_pivot, node->pivotkeys.fill_pivot(i, &x)); |
| 365 | VERIFY_ASSERTION(compare >= 0, i, "Pivot is < the upper-bound pivot" ); |
| 366 | } |
| 367 | } |
| 368 | |
| 369 | for (int i = 0; i < node->n_children; i++) { |
| 370 | DBT x, y; |
| 371 | const DBT *curr_less_pivot = (i==0) ? lesser_pivot : node->pivotkeys.fill_pivot(i - 1, &x); |
| 372 | const DBT *curr_geq_pivot = (i==node->n_children-1) ? greatereq_pivot : node->pivotkeys.fill_pivot(i, &y); |
| 373 | if (node->height > 0) { |
| 374 | NONLEAF_CHILDINFO bnc = BNC(node, i); |
| 375 | // Verify that messages in the buffers are in the right place. |
| 376 | VERIFY_ASSERTION(verify_sorted_by_key_msn(ft_handle, &bnc->msg_buffer, bnc->fresh_message_tree) == 0, i, "fresh_message_tree" ); |
| 377 | VERIFY_ASSERTION(verify_sorted_by_key_msn(ft_handle, &bnc->msg_buffer, bnc->stale_message_tree) == 0, i, "stale_message_tree" ); |
| 378 | |
| 379 | verify_msg_fn verify_msg(ft_handle, bnc, curr_less_pivot, curr_geq_pivot, |
| 380 | blocknum, this_msn, verbose, keep_going_on_failure, messages_have_been_moved); |
| 381 | int r = bnc->msg_buffer.iterate(verify_msg); |
| 382 | if (r != 0) { result = r; goto done; } |
| 383 | |
| 384 | struct verify_message_tree_extra = { .msg_buffer = &bnc->msg_buffer, .broadcast = false, .is_fresh = true, .i = i, .verbose = verbose, .blocknum = node->blocknum, .keep_going_on_failure = keep_going_on_failure, .messages_have_been_moved = messages_have_been_moved }; |
| 385 | r = bnc->fresh_message_tree.iterate<struct verify_message_tree_extra, verify_message_tree>(&extra); |
| 386 | if (r != 0) { result = r; goto done; } |
| 387 | extra.is_fresh = false; |
| 388 | r = bnc->stale_message_tree.iterate<struct verify_message_tree_extra, verify_message_tree>(&extra); |
| 389 | if (r != 0) { result = r; goto done; } |
| 390 | |
| 391 | bnc->fresh_message_tree.verify_marks_consistent(); |
| 392 | if (messages_have_been_moved) { |
| 393 | VERIFY_ASSERTION(!bnc->fresh_message_tree.has_marks(), i, "fresh message tree still has marks after moving messages" ); |
| 394 | r = bnc->fresh_message_tree.iterate_over_marked<void, error_on_iter>(nullptr); |
| 395 | if (r != 0) { result = r; goto done; } |
| 396 | } |
| 397 | else { |
| 398 | r = bnc->fresh_message_tree.iterate_over_marked<struct verify_message_tree_extra, verify_marked_messages>(&extra); |
| 399 | if (r != 0) { result = r; goto done; } |
| 400 | } |
| 401 | |
| 402 | extra.broadcast = true; |
| 403 | r = bnc->broadcast_list.iterate<struct verify_message_tree_extra, verify_message_tree>(&extra); |
| 404 | if (r != 0) { result = r; goto done; } |
| 405 | } |
| 406 | else { |
| 407 | BASEMENTNODE bn = BLB(node, i); |
| 408 | for (uint32_t j = 0; j < bn->data_buffer.num_klpairs(); j++) { |
| 409 | VERIFY_ASSERTION((rootmsn.msn >= this_msn.msn), 0, "leaf may have latest msn, but cannot be greater than root msn" ); |
| 410 | DBT kdbt = get_ith_key_dbt(bn, j); |
| 411 | if (curr_less_pivot) { |
| 412 | int compare = compare_pairs(ft_handle, curr_less_pivot, &kdbt); |
| 413 | VERIFY_ASSERTION_BASEMENT(compare < 0, i, j, "The leafentry is >= the lower-bound pivot" ); |
| 414 | } |
| 415 | if (curr_geq_pivot) { |
| 416 | int compare = compare_pairs(ft_handle, curr_geq_pivot, &kdbt); |
| 417 | VERIFY_ASSERTION_BASEMENT(compare >= 0, i, j, "The leafentry is < the upper-bound pivot" ); |
| 418 | } |
| 419 | if (0 < j) { |
| 420 | DBT prev_key_dbt = get_ith_key_dbt(bn, j-1); |
| 421 | int compare = compare_pairs(ft_handle, &prev_key_dbt, &kdbt); |
| 422 | VERIFY_ASSERTION_BASEMENT(compare < 0, i, j, "Adjacent leafentries are out of order" ); |
| 423 | } |
| 424 | } |
| 425 | } |
| 426 | } |
| 427 | |
| 428 | done: |
| 429 | return result; |
| 430 | } |
| 431 | |
| 432 | |
| 433 | // input is a pinned node, on exit, node is unpinned |
| 434 | int |
| 435 | toku_verify_ftnode (FT_HANDLE ft_handle, |
| 436 | MSN rootmsn, MSN parentmsn_with_messages, bool messages_exist_above, |
| 437 | FTNODE node, int height, |
| 438 | const DBT *lesser_pivot, // Everything in the subtree should be > lesser_pivot. (lesser_pivot==NULL if there is no lesser pivot.) |
| 439 | const DBT *greatereq_pivot, // Everything in the subtree should be <= lesser_pivot. (lesser_pivot==NULL if there is no lesser pivot.) |
| 440 | int (*progress_callback)(void *, float progress), void *, |
| 441 | int recurse, int verbose, int keep_going_on_failure) |
| 442 | { |
| 443 | MSN this_msn; |
| 444 | |
| 445 | //printf("%s:%d pin %p\n", __FILE__, __LINE__, node_v); |
| 446 | toku_ftnode_assert_fully_in_memory(node); |
| 447 | this_msn = node->max_msn_applied_to_node_on_disk; |
| 448 | |
| 449 | int result = 0; |
| 450 | int result2 = 0; |
| 451 | if (node->height > 0) { |
| 452 | // Otherwise we'll just do the next call |
| 453 | |
| 454 | result = toku_verify_ftnode_internal( |
| 455 | ft_handle, rootmsn, parentmsn_with_messages, messages_exist_above, node, height, lesser_pivot, greatereq_pivot, |
| 456 | verbose, keep_going_on_failure, false); |
| 457 | if (result != 0 && (!keep_going_on_failure || result != TOKUDB_NEEDS_REPAIR)) goto done; |
| 458 | } |
| 459 | if (node->height > 0) { |
| 460 | toku_move_ftnode_messages_to_stale(ft_handle->ft, node); |
| 461 | } |
| 462 | result2 = toku_verify_ftnode_internal( |
| 463 | ft_handle, rootmsn, parentmsn_with_messages, messages_exist_above, node, height, lesser_pivot, greatereq_pivot, |
| 464 | verbose, keep_going_on_failure, true); |
| 465 | if (result == 0) { |
| 466 | result = result2; |
| 467 | if (result != 0 && (!keep_going_on_failure || result != TOKUDB_NEEDS_REPAIR)) goto done; |
| 468 | } |
| 469 | |
| 470 | // Verify that the subtrees have the right properties. |
| 471 | if (recurse && node->height > 0) { |
| 472 | for (int i = 0; i < node->n_children; i++) { |
| 473 | FTNODE child_node; |
| 474 | toku_get_node_for_verify(BP_BLOCKNUM(node, i), ft_handle, &child_node); |
| 475 | DBT x, y; |
| 476 | int r = toku_verify_ftnode(ft_handle, rootmsn, |
| 477 | (toku_bnc_n_entries(BNC(node, i)) > 0 |
| 478 | ? this_msn |
| 479 | : parentmsn_with_messages), |
| 480 | messages_exist_above || toku_bnc_n_entries(BNC(node, i)) > 0, |
| 481 | child_node, node->height-1, |
| 482 | (i==0) ? lesser_pivot : node->pivotkeys.fill_pivot(i - 1, &x), |
| 483 | (i==node->n_children-1) ? greatereq_pivot : node->pivotkeys.fill_pivot(i, &y), |
| 484 | progress_callback, progress_extra, |
| 485 | recurse, verbose, keep_going_on_failure); |
| 486 | if (r) { |
| 487 | result = r; |
| 488 | if (!keep_going_on_failure || result != TOKUDB_NEEDS_REPAIR) goto done; |
| 489 | } |
| 490 | } |
| 491 | } |
| 492 | done: |
| 493 | toku_unpin_ftnode(ft_handle->ft, node); |
| 494 | |
| 495 | if (result == 0 && progress_callback) |
| 496 | result = progress_callback(progress_extra, 0.0); |
| 497 | |
| 498 | return result; |
| 499 | } |
| 500 | |
| 501 | int |
| 502 | toku_verify_ft_with_progress (FT_HANDLE ft_handle, int (*progress_callback)(void *, float progress), void *, int verbose, int keep_on_going) { |
| 503 | assert(ft_handle->ft); |
| 504 | FTNODE root_node = NULL; |
| 505 | { |
| 506 | uint32_t root_hash; |
| 507 | CACHEKEY root_key; |
| 508 | toku_calculate_root_offset_pointer(ft_handle->ft, &root_key, &root_hash); |
| 509 | toku_get_node_for_verify(root_key, ft_handle, &root_node); |
| 510 | } |
| 511 | int r = toku_verify_ftnode(ft_handle, ft_handle->ft->h->max_msn_in_ft, ft_handle->ft->h->max_msn_in_ft, false, root_node, -1, NULL, NULL, progress_callback, progress_extra, 1, verbose, keep_on_going); |
| 512 | if (r == 0) { |
| 513 | toku_ft_lock(ft_handle->ft); |
| 514 | ft_handle->ft->h->time_of_last_verification = time(NULL); |
| 515 | ft_handle->ft->h->dirty = 1; |
| 516 | toku_ft_unlock(ft_handle->ft); |
| 517 | } |
| 518 | return r; |
| 519 | } |
| 520 | |
| 521 | int |
| 522 | toku_verify_ft (FT_HANDLE ft_handle) { |
| 523 | return toku_verify_ft_with_progress(ft_handle, NULL, NULL, 0, 0); |
| 524 | } |
| 525 | |