GeezerVisonThe World Wide Web, restored to 800 pixels of glory. |
| Home|About|Gaming cabinet|Archives|How it works|Starter sites |
GeezerVison is showing github.com — cpython/Objects/dictobject.c at 3.13 · python/cpython Expand file tree/ Copy pathdictobject.cMore file actions 7538 lines (6700 loc) · 221 KB / Copy pathdictobject.cFile metadata and controls7538 lines (6700 loc) · 221 KB 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 /* Dictionary object implementation using a hash table */ /* The distribution includes a separate file, Objects/dictnotes.txt, describing explorations into dictionary design and optimization. It covers typical dictionary use patterns, the parameters for tuning dictionaries, and several ideas for possible optimizations. */ /* PyDictKeysObject This implements the dictionary's hashtable. As of Python 3.6, this is compact and ordered. Basic idea is described here: * https://mail.python.org/pipermail/python-dev/2012-December/123028.html * https://morepypy.blogspot.com/2015/01/faster-more-memory-efficient-and-more.html layout: +---------------------+ | dk_refcnt | | dk_log2_size | | dk_log2_index_bytes | | dk_kind | | dk_version | | dk_usable | | dk_nentries | +---------------------+ | dk_indices[] | | | +---------------------+ | dk_entries[] | | | +---------------------+ dk_indices is actual hashtable. It holds index in entries, or DKIX_EMPTY(-1) or DKIX_DUMMY(-2). Size of indices is dk_size. Type of each index in indices is vary on dk_size: * int8 for dk_size <= 128 * int16 for 256 <= dk_size <= 2**15 * int32 for 2**16 <= dk_size <= 2**31 * int64 for 2**32 <= dk_size dk_entries is array of PyDictKeyEntry when dk_kind == DICT_KEYS_GENERAL or PyDictUnicodeEntry otherwise. Its length is USABLE_FRACTION(dk_size). NOTE: Since negative value is used for DKIX_EMPTY and DKIX_DUMMY, type of dk_indices entry is signed integer and int16 is used for table which dk_size == 256. */ /* The DictObject can be in one of two forms. Either: A combined table: ma_values == NULL, dk_refcnt == 1. Values are stored in the me_value field of the PyDictKeyEntry. Or: A split table: ma_values != NULL, dk_refcnt >= 1 Values are stored in the ma_values array. Only string (unicode) keys are allowed. There are four kinds of slots in the table (slot is index, and DK_ENTRIES(keys)[index] if index >= 0): 1. Unused. index == DKIX_EMPTY Does not hold an active (key, value) pair now and never did. Unused can transition to Active upon key insertion. This is each slot's initial state. 2. Active. index >= 0, me_key != NULL and me_value != NULL Holds an active (key, value) pair. Active can transition to Dummy or Pending upon key deletion (for combined and split tables respectively). This is the only case in which me_value != NULL. 3. Dummy. index == DKIX_DUMMY (combined only) Previously held an active (key, value) pair, but that was deleted and an active pair has not yet overwritten the slot. Dummy can transition to Active upon key insertion. Dummy slots cannot be made Unused again else the probe sequence in case of collision would have no way to know they were once active. In free-threaded builds dummy slots are not re-used to allow lock-free lookups to proceed safely. 4. Pending. index >= 0, key != NULL, and value == NULL (split only) Not yet inserted in split-table. */ /* Preserving insertion order It's simple for combined table. Since dk_entries is mostly append only, we can get insertion order by just iterating dk_entries. One exception is .popitem(). It removes last item in dk_entries and decrement dk_nentries to achieve amortized O(1). Since there are DKIX_DUMMY remains in dk_indices, we can't increment dk_usable even though dk_nentries is decremented. To preserve the order in a split table, a bit vector is used to record the insertion order. When a key is inserted the bit vector is shifted up by 4 bits and the index of the key is stored in the low 4 bits. As a consequence of this, split keys have a maximum size of 16. */ /* PyDict_MINSIZE is the starting size for any new dict. * 8 allows dicts with no more than 5 active entries; experiments suggested * this suffices for the majority of dicts (consisting mostly of usually-small * dicts created to pass keyword arguments). * Making this 8, rather than 4 reduces the number of resizes for most * dictionaries, without any significant extra memory use. */ #define PyDict_LOG_MINSIZE 3 #define PyDict_MINSIZE 8 #include "Python.h" #include "pycore_bitutils.h" // _Py_bit_length #include "pycore_call.h" // _PyObject_CallNoArgs() #include "pycore_ceval.h" // _PyEval_GetBuiltin() #include "pycore_code.h" // stats #include "pycore_critical_section.h" // Py_BEGIN_CRITICAL_SECTION, Py_END_CRITICAL_SECTION #include "pycore_dict.h" // export _PyDict_SizeOf() #include "pycore_freelist.h" // _PyFreeListState_GET() #include "pycore_gc.h" // _PyObject_GC_IS_TRACKED() #include "pycore_object.h" // _PyObject_GC_TRACK(), _PyDebugAllocatorStats() #include "pycore_pyatomic_ft_wrappers.h" // FT_ATOMIC_LOAD_SSIZE_RELAXED #include "pycore_pyerrors.h" // _PyErr_GetRaisedException() #include "pycore_pystate.h" // _PyThreadState_GET() #include "pycore_setobject.h" // _PySet_NextEntry() #include "stringlib/eq.h" // unicode_eq() #include <stdbool.h> /*[clinic input] class dict "PyDictObject *" "&PyDict_Type" [clinic start generated code]*/ /*[clinic end generated code: output=da39a3ee5e6b4b0d input=f157a5a0ce9589d6]*/ /* To ensure the lookup algorithm terminates, there must be at least one Unused slot (NULL key) in the table. To avoid slowing down lookups on a near-full table, we resize the table when it's USABLE_FRACTION (currently two-thirds) full. */ #ifdef Py_GIL_DISABLED static inline void ASSERT_DICT_LOCKED(PyObject *op) { _Py_CRITICAL_SECTION_ASSERT_OBJECT_LOCKED(op); } #define ASSERT_DICT_LOCKED(op) ASSERT_DICT_LOCKED(_Py_CAST(PyObject*, op)) #define ASSERT_WORLD_STOPPED_OR_DICT_LOCKED(op) \ if (!_PyInterpreterState_GET()->stoptheworld.world_stopped) { \ ASSERT_DICT_LOCKED(op); \ } #define ASSERT_WORLD_STOPPED_OR_OBJ_LOCKED(op) \ if (!_PyInterpreterState_GET()->stoptheworld.world_stopped) { \ _Py_CRITICAL_SECTION_ASSERT_OBJECT_LOCKED(op); \ } #define IS_DICT_SHARED(mp) _PyObject_GC_IS_SHARED(mp) #define SET_DICT_SHARED(mp) _PyObject_GC_SET_SHARED(mp) #define LOAD_INDEX(keys, size, idx) _Py_atomic_load_int##size##_relaxed(&((const int##size##_t*)keys->dk_indices)[idx]); #define STORE_INDEX(keys, size, idx, value) _Py_atomic_store_int##size##_relaxed(&((int##size##_t*)keys->dk_indices)[idx], (int##size##_t)value); #define ASSERT_OWNED_OR_SHARED(mp) \ assert(_Py_IsOwnedByCurrentThread((PyObject *)mp) || IS_DICT_SHARED(mp)); #define LOCK_KEYS_IF_SPLIT(keys, kind) \ if (kind == DICT_KEYS_SPLIT) { \ LOCK_KEYS(keys); \ } #define UNLOCK_KEYS_IF_SPLIT(keys, kind) \ if (kind == DICT_KEYS_SPLIT) { \ UNLOCK_KEYS(keys); \ } static inline Py_ssize_t load_keys_nentries(PyDictObject *mp) { PyDictKeysObject *keys = _Py_atomic_load_ptr(&mp->ma_keys); return _Py_atomic_load_ssize(&keys->dk_nentries); } static inline void set_keys(PyDictObject *mp, PyDictKeysObject *keys) { ASSERT_OWNED_OR_SHARED(mp); _Py_atomic_store_ptr_release(&mp->ma_keys, keys); } static inline void set_values(PyDictObject *mp, PyDictValues *values) { ASSERT_OWNED_OR_SHARED(mp); _Py_atomic_store_ptr_release(&mp->ma_values, values); } #define LOCK_KEYS(keys) PyMutex_LockFlags(&keys->dk_mutex, _Py_LOCK_DONT_DETACH) #define UNLOCK_KEYS(keys) PyMutex_Unlock(&keys->dk_mutex) #define ASSERT_KEYS_LOCKED(keys) assert(PyMutex_IsLocked(&keys->dk_mutex)) #define LOAD_SHARED_KEY(key) _Py_atomic_load_ptr_acquire(&key) #define STORE_SHARED_KEY(key, value) _Py_atomic_store_ptr_release(&key, value) // Inc refs the keys object, giving the previous value #define INCREF_KEYS(dk) _Py_atomic_add_ssize(&dk->dk_refcnt, 1) // Dec refs the keys object, giving the previous value #define DECREF_KEYS(dk) _Py_atomic_add_ssize(&dk->dk_refcnt, -1) #define LOAD_KEYS_NENTRIES(keys) _Py_atomic_load_ssize_relaxed(&keys->dk_nentries) #define INCREF_KEYS_FT(dk) dictkeys_incref(dk) #define DECREF_KEYS_FT(dk, shared) dictkeys_decref(_PyInterpreterState_GET(), dk, shared) static inline void split_keys_entry_added(PyDictKeysObject *keys) { ASSERT_KEYS_LOCKED(keys); // We increase before we decrease so we never get too small of a value // when we're racing with reads _Py_atomic_store_ssize_relaxed(&keys->dk_nentries, keys->dk_nentries + 1); _Py_atomic_store_ssize_release(&keys->dk_usable, keys->dk_usable - 1); } #else /* Py_GIL_DISABLED */ #define ASSERT_DICT_LOCKED(op) #define ASSERT_WORLD_STOPPED_OR_DICT_LOCKED(op) #define ASSERT_WORLD_STOPPED_OR_OBJ_LOCKED(op) #define LOCK_KEYS(keys) #define UNLOCK_KEYS(keys) #define ASSERT_KEYS_LOCKED(keys) #define LOAD_SHARED_KEY(key) key #define STORE_SHARED_KEY(key, value) key = value #define INCREF_KEYS(dk) dk->dk_refcnt++ #define DECREF_KEYS(dk) dk->dk_refcnt-- #define LOAD_KEYS_NENTRIES(keys) keys->dk_nentries #define INCREF_KEYS_FT(dk) #define DECREF_KEYS_FT(dk, shared) #define LOCK_KEYS_IF_SPLIT(keys, kind) #define UNLOCK_KEYS_IF_SPLIT(keys, kind) #define IS_DICT_SHARED(mp) (false) #define SET_DICT_SHARED(mp) #define LOAD_INDEX(keys, size, idx) ((const int##size##_t*)(keys->dk_indices))[idx] #define STORE_INDEX(keys, size, idx, value) ((int##size##_t*)(keys->dk_indices))[idx] = (int##size##_t)value static inline void split_keys_entry_added(PyDictKeysObject *keys) { keys->dk_usable--; keys->dk_nentries++; } static inline void set_keys(PyDictObject *mp, PyDictKeysObject *keys) { mp->ma_keys = keys; } static inline void set_values(PyDictObject *mp, PyDictValues *values) { mp->ma_values = values; } static inline Py_ssize_t load_keys_nentries(PyDictObject *mp) { return mp->ma_keys->dk_nentries; } #endif #define STORE_KEY(ep, key) FT_ATOMIC_STORE_PTR_RELEASE((ep)->me_key, key) #define STORE_VALUE(ep, value) FT_ATOMIC_STORE_PTR_RELEASE((ep)->me_value, value) #define STORE_SPLIT_VALUE(mp, idx, value) FT_ATOMIC_STORE_PTR_RELEASE(mp->ma_values->values[idx], value) #define STORE_HASH(ep, hash) FT_ATOMIC_STORE_SSIZE_RELAXED((ep)->me_hash, hash) #define STORE_KEYS_USABLE(keys, usable) FT_ATOMIC_STORE_SSIZE_RELAXED(keys->dk_usable, usable) #define STORE_KEYS_NENTRIES(keys, nentries) FT_ATOMIC_STORE_SSIZE_RELAXED(keys->dk_nentries, nentries) #define STORE_USED(mp, used) FT_ATOMIC_STORE_SSIZE_RELAXED(mp->ma_used, used) #define PERTURB_SHIFT 5 /* Major subtleties ahead: Most hash schemes depend on having a "good" hash function, in the sense of simulating randomness. Python doesn't: its most important hash functions (for ints) are very regular in common cases: >>>[hash(i) for i in range(4)] [0, 1, 2, 3] This isn't necessarily bad! To the contrary, in a table of size 2**i, taking the low-order i bits as the initial table index is extremely fast, and there are no collisions at all for dicts indexed by a contiguous range of ints. So this gives better-than-random behavior in common cases, and that's very desirable. OTOH, when collisions occur, the tendency to fill contiguous slices of the hash table makes a good collision resolution strategy crucial. Taking only the last i bits of the hash code is also vulnerable: for example, consider the list [i << 16 for i in range(20000)] as a set of keys. Since ints are their own hash codes, and this fits in a dict of size 2**15, the last 15 bits of every hash code are all 0: they *all* map to the same table index. But catering to unusual cases should not slow the usual ones, so we just take the last i bits anyway. It's up to collision resolution to do the rest. If we *usually* find the key we're looking for on the first try (and, it turns out, we usually do -- the table load factor is kept under 2/3, so the odds are solidly in our favor), then it makes best sense to keep the initial index computation dirt cheap. The first half of collision resolution is to visit table indices via this recurrence: j = ((5*j) + 1) mod 2**i For any initial j in range(2**i), repeating that 2**i times generates each int in range(2**i) exactly once (see any text on random-number generation for proof). By itself, this doesn't help much: like linear probing (setting j += 1, or j -= 1, on each loop trip), it scans the table entries in a fixed order. This would be bad, except that's not the only thing we do, and it's actually *good* in the common cases where hash keys are consecutive. In an example that's really too small to make this entirely clear, for a table of size 2**3 the order of indices is: 0 -> 1 -> 6 -> 7 -> 4 -> 5 -> 2 -> 3 -> 0 [and here it's repeating] If two things come in at index 5, the first place we look after is index 2, not 6, so if another comes in at index 6 the collision at 5 didn't hurt it. Linear probing is deadly in this case because there the fixed probe order is the *same* as the order consecutive keys are likely to arrive. But it's extremely unlikely hash codes will follow a 5*j+1 recurrence by accident, and certain that consecutive hash codes do not. The other half of the strategy is to get the other bits of the hash code into play. This is done by initializing a (unsigned) vrbl "perturb" to the full hash code, and changing the recurrence to: perturb >>= PERTURB_SHIFT; j = (5*j) + 1 + perturb; use j % 2**i as the next table index; Now the probe sequence depends (eventually) on every bit in the hash code, and the pseudo-scrambling property of recurring on 5*j+1 is more valuable, because it quickly magnifies small differences in the bits that didn't affect the initial index. Note that because perturb is unsigned, if the recurrence is executed often enough perturb eventually becomes and remains 0. At that point (very rarely reached) the recurrence is on (just) 5*j+1 again, and that's certain to find an empty slot eventually (since it generates every int in range(2**i), and we make sure there's always at least one empty slot). Selecting a good value for PERTURB_SHIFT is a balancing act. You want it small so that the high bits of the hash code continue to affect the probe sequence across iterations; but you want it large so that in really bad cases the high-order hash bits have an effect on early iterations. 5 was "the best" in minimizing total collisions across experiments Tim Peters ran (on both normal and pathological cases), but 4 and 6 weren't significantly worse. Historical: Reimer Behrends contributed the idea of using a polynomial-based approach, using repeated multiplication by x in GF(2**n) where an irreducible polynomial for each table size was chosen such that x was a primitive root. Christian Tismer later extended that to use division by x instead, as an efficient way to get the high bits of the hash code into play. This scheme also gave excellent collision statistics, but was more expensive: two if-tests were required inside the loop; computing "the next" index took about the same number of operations but without as much potential parallelism (e.g., computing 5*j can go on at the same time as computing 1+perturb in the above, and then shifting perturb can be done while the table index is being masked); and the PyDictObject struct required a member to hold the table's polynomial. In Tim's experiments the current scheme ran faster, produced equally good collision statistics, needed less code & used less memory. */ static int dictresize(PyInterpreterState *interp, PyDictObject *mp, uint8_t log_newsize, int unicode); static PyObject* dict_iter(PyObject *dict); static int setitem_lock_held(PyDictObject *mp, PyObject *key, PyObject *value); static int dict_setdefault_ref_lock_held(PyObject *d, PyObject *key, PyObject *default_value, PyObject **result, int incref_result); #ifndef NDEBUG static int _PyObject_InlineValuesConsistencyCheck(PyObject *obj); #endif #include "clinic/dictobject.c.h" #ifdef WITH_FREELISTS static struct _Py_dict_freelist * get_dict_freelist(void) { struct _Py_object_freelists *freelists = _Py_object_freelists_GET(); return &freelists->dicts; } static struct _Py_dictkeys_freelist * get_dictkeys_freelist(void) { struct _Py_object_freelists *freelists = _Py_object_freelists_GET(); return &freelists->dictkeys; } #endif void _PyDict_ClearFreeList(struct _Py_object_freelists *freelists, int is_finalization) { #ifdef WITH_FREELISTS struct _Py_dict_freelist *freelist = &freelists->dicts; while (freelist->numfree > 0) { PyDictObject *op = freelist->items[--freelist->numfree]; assert(PyDict_CheckExact(op)); PyObject_GC_Del(op); } struct _Py_dictkeys_freelist *keys_freelist = &freelists->dictkeys; while (keys_freelist->numfree > 0) { PyMem_Free(keys_freelist->items[--keys_freelist->numfree]); } if (is_finalization) { freelist->numfree = -1; keys_freelist->numfree = -1; } #endif } static inline Py_hash_t unicode_get_hash(PyObject *o) { assert(PyUnicode_CheckExact(o)); return FT_ATOMIC_LOAD_SSIZE_RELAXED(_PyASCIIObject_CAST(o)->hash); } /* Print summary info about the state of the optimized allocator */ void _PyDict_DebugMallocStats(FILE *out) { #ifdef WITH_FREELISTS struct _Py_dict_freelist *dict_freelist = get_dict_freelist(); _PyDebugAllocatorStats(out, "free PyDictObject", dict_freelist->numfree, sizeof(PyDictObject)); struct _Py_dictkeys_freelist *dictkeys_freelist = get_dictkeys_freelist(); _PyDebugAllocatorStats(out, "free PyDictKeysObject", dictkeys_freelist->numfree, sizeof(PyDictKeysObject)); #endif } #define DK_MASK(dk) (DK_SIZE(dk)-1) static void free_keys_object(PyDictKeysObject *keys, bool use_qsbr); /* PyDictKeysObject has refcounts like PyObject does, so we have the following two functions to mirror what Py_INCREF() and Py_DECREF() do. (Keep in mind that PyDictKeysObject isn't actually a PyObject.) Likewise a PyDictKeysObject can be immortal (e.g. Py_EMPTY_KEYS), so we apply a naive version of what Py_INCREF() and Py_DECREF() do for immortal objects. */ static inline void dictkeys_incref(PyDictKeysObject *dk) { if (FT_ATOMIC_LOAD_SSIZE_RELAXED(dk->dk_refcnt) == _Py_IMMORTAL_REFCNT) { return; } #ifdef Py_REF_DEBUG _Py_IncRefTotal(_PyThreadState_GET()); #endif INCREF_KEYS(dk); } static inline void dictkeys_decref(PyInterpreterState *interp, PyDictKeysObject *dk, bool use_qsbr) { if (FT_ATOMIC_LOAD_SSIZE_RELAXED(dk->dk_refcnt) == _Py_IMMORTAL_REFCNT) { return; } assert(FT_ATOMIC_LOAD_SSIZE(dk->dk_refcnt) > 0); #ifdef Py_REF_DEBUG _Py_DecRefTotal(_PyThreadState_GET()); #endif if (DECREF_KEYS(dk) == 1) { if (DK_IS_UNICODE(dk)) { PyDictUnicodeEntry *entries = DK_UNICODE_ENTRIES(dk); Py_ssize_t i, n; for (i = 0, n = dk->dk_nentries; i < n; i++) { Py_XDECREF(entries[i].me_key); Py_XDECREF(entries[i].me_value); } } else { PyDictKeyEntry *entries = DK_ENTRIES(dk); Py_ssize_t i, n; for (i = 0, n = dk->dk_nentries; i < n; i++) { Py_XDECREF(entries[i].me_key); Py_XDECREF(entries[i].me_value); } } free_keys_object(dk, use_qsbr); } } /* lookup indices. returns DKIX_EMPTY, DKIX_DUMMY, or ix >=0 */ static inline Py_ssize_t dictkeys_get_index(const PyDictKeysObject *keys, Py_ssize_t i) { int log2size = DK_LOG_SIZE(keys); Py_ssize_t ix; if (log2size < 8) { ix = LOAD_INDEX(keys, 8, i); } else if (log2size < 16) { ix = LOAD_INDEX(keys, 16, i); } #if SIZEOF_VOID_P > 4 else if (log2size >= 32) { ix = LOAD_INDEX(keys, 64, i); } #endif else { ix = LOAD_INDEX(keys, 32, i); } assert(ix >= DKIX_DUMMY); return ix; } /* write to indices. */ static inline void dictkeys_set_index(PyDictKeysObject *keys, Py_ssize_t i, Py_ssize_t ix) { int log2size = DK_LOG_SIZE(keys); assert(ix >= DKIX_DUMMY); assert(keys->dk_version == 0); if (log2size < 8) { assert(ix <= 0x7f); STORE_INDEX(keys, 8, i, ix); } else if (log2size < 16) { assert(ix <= 0x7fff); STORE_INDEX(keys, 16, i, ix); } #if SIZEOF_VOID_P > 4 else if (log2size >= 32) { STORE_INDEX(keys, 64, i, ix); } #endif else { assert(ix <= 0x7fffffff); STORE_INDEX(keys, 32, i, ix); } } /* USABLE_FRACTION is the maximum dictionary load. * Increasing this ratio makes dictionaries more dense resulting in more * collisions. Decreasing it improves sparseness at the expense of spreading * indices over more cache lines and at the cost of total memory consumed. * * USABLE_FRACTION must obey the following: * (0 < USABLE_FRACTION(n) < n) for all n >= 2 * * USABLE_FRACTION should be quick to calculate. * Fractions around 1/2 to 2/3 seem to work well in practice. */ #define USABLE_FRACTION(n) (((n) << 1)/3) /* Find the smallest dk_size >= minsize. */ static inline uint8_t calculate_log2_keysize(Py_ssize_t minsize) { #if SIZEOF_LONG == SIZEOF_SIZE_T minsize = Py_MAX(minsize, PyDict_MINSIZE); return _Py_bit_length(minsize - 1); #elif defined(_MSC_VER) // On 64bit Windows, sizeof(long) == 4. We cannot use _Py_bit_length. minsize = Py_MAX(minsize, PyDict_MINSIZE); unsigned long msb; _BitScanReverse64(&msb, (uint64_t)minsize - 1); return (uint8_t)(msb + 1); #else uint8_t log2_size; for (log2_size = PyDict_LOG_MINSIZE; (((Py_ssize_t)1) << log2_size) < minsize; log2_size++) ; return log2_size; #endif } /* estimate_keysize is reverse function of USABLE_FRACTION. * * This can be used to reserve enough size to insert n entries without * resizing. */ static inline uint8_t estimate_log2_keysize(Py_ssize_t n) { return calculate_log2_keysize((n*3 + 1) / 2); } /* GROWTH_RATE. Growth rate upon hitting maximum load. * Currently set to used*3. * This means that dicts double in size when growing without deletions, * but have more head room when the number of deletions is on a par with the * number of insertions. See also bpo-17563 and bpo-33205. * * GROWTH_RATE was set to used*4 up to version 3.2. * GROWTH_RATE was set to used*2 in version 3.3.0 * GROWTH_RATE was set to used*2 + capacity/2 in 3.4.0-3.6.0. */ #define GROWTH_RATE(d) ((d)->ma_used*3) /* This immutable, empty PyDictKeysObject is used for PyDict_Clear() * (which cannot fail and thus can do no allocation). * * See https://github.com/python/cpython/pull/127568#discussion_r1868070614 * for the rationale of using dk_log2_index_bytes=3 instead of 0. */ static PyDictKeysObject empty_keys_struct = { _Py_IMMORTAL_REFCNT, /* dk_refcnt */ 0, /* dk_log2_size */ 3, /* dk_log2_index_bytes */ DICT_KEYS_UNICODE, /* dk_kind */ #ifdef Py_GIL_DISABLED {0}, /* dk_mutex */ #endif 1, /* dk_version */ 0, /* dk_usable (immutable) */ 0, /* dk_nentries */ {DKIX_EMPTY, DKIX_EMPTY, DKIX_EMPTY, DKIX_EMPTY, DKIX_EMPTY, DKIX_EMPTY, DKIX_EMPTY, DKIX_EMPTY}, /* dk_indices */ }; #define Py_EMPTY_KEYS &empty_keys_struct /* Uncomment to check the dict content in _PyDict_CheckConsistency() */ // #define DEBUG_PYDICT #ifdef DEBUG_PYDICT # define ASSERT_CONSISTENT(op) assert(_PyDict_CheckConsistency((PyObject *)(op), 1)) #else # define ASSERT_CONSISTENT(op) assert(_PyDict_CheckConsistency((PyObject *)(op), 0)) #endif static inline int get_index_from_order(PyDictObject *mp, Py_ssize_t i) { assert(mp->ma_used <= SHARED_KEYS_MAX_SIZE); assert(i < mp->ma_values->size); uint8_t *array = get_insertion_order_array(mp->ma_values); return array[i]; } #ifdef DEBUG_PYDICT static void dump_entries(PyDictKeysObject *dk) { for (Py_ssize_t i = 0; i < dk->dk_nentries; i++) { if (DK_IS_UNICODE(dk)) { PyDictUnicodeEntry *ep = &DK_UNICODE_ENTRIES(dk)[i]; printf("key=%p value=%p\n", ep->me_key, ep->me_value); } else { PyDictKeyEntry *ep = &DK_ENTRIES(dk)[i]; printf("key=%p hash=%lx value=%p\n", ep->me_key, ep->me_hash, ep->me_value); } } } #endif int _PyDict_CheckConsistency(PyObject *op, int check_content) { ASSERT_WORLD_STOPPED_OR_DICT_LOCKED(op); #define CHECK(expr) \ do { if (!(expr)) { _PyObject_ASSERT_FAILED_MSG(op, Py_STRINGIFY(expr)); } } while (0) assert(op != NULL); CHECK(PyDict_Check(op)); PyDictObject *mp = (PyDictObject *)op; PyDictKeysObject *keys = mp->ma_keys; int splitted = _PyDict_HasSplitTable(mp); Py_ssize_t usable = USABLE_FRACTION(DK_SIZE(keys)); // In the free-threaded build, shared keys may be concurrently modified, // so use atomic loads. Py_ssize_t dk_usable = FT_ATOMIC_LOAD_SSIZE_ACQUIRE(keys->dk_usable); Py_ssize_t dk_nentries = FT_ATOMIC_LOAD_SSIZE_ACQUIRE(keys->dk_nentries); CHECK(0 <= mp->ma_used && mp->ma_used <= usable); CHECK(0 <= dk_usable && dk_usable <= usable); CHECK(0 <= dk_nentries && dk_nentries <= usable); CHECK(dk_usable + dk_nentries <= usable); if (!splitted) { /* combined table */ CHECK(keys->dk_kind != DICT_KEYS_SPLIT); CHECK(keys->dk_refcnt == 1 || keys == Py_EMPTY_KEYS); } else { CHECK(keys->dk_kind == DICT_KEYS_SPLIT); CHECK(mp->ma_used <= SHARED_KEYS_MAX_SIZE); if (mp->ma_values->embedded) { CHECK(mp->ma_values->embedded == 1); CHECK(mp->ma_values->valid == 1); } } if (check_content) { LOCK_KEYS_IF_SPLIT(keys, keys->dk_kind); for (Py_ssize_t i=0; i < DK_SIZE(keys); i++) { Py_ssize_t ix = dictkeys_get_index(keys, i); CHECK(DKIX_DUMMY <= ix && ix <= usable); } if (keys->dk_kind == DICT_KEYS_GENERAL) { PyDictKeyEntry *entries = DK_ENTRIES(keys); for (Py_ssize_t i=0; i < usable; i++) { PyDictKeyEntry *entry = &entries[i]; PyObject *key = entry->me_key; if (key != NULL) { /* test_dict fails if PyObject_Hash() is called again */ CHECK(entry->me_hash != -1); CHECK(entry->me_value != NULL); if (PyUnicode_CheckExact(key)) { Py_hash_t hash = unicode_get_hash(key); CHECK(entry->me_hash == hash); } } } } else { PyDictUnicodeEntry *entries = DK_UNICODE_ENTRIES(keys); for (Py_ssize_t i=0; i < usable; i++) { PyDictUnicodeEntry *entry = &entries[i]; PyObject *key = entry->me_key; if (key != NULL) { CHECK(PyUnicode_CheckExact(key)); Py_hash_t hash = unicode_get_hash(key); CHECK(hash != -1); if (!splitted) { CHECK(entry->me_value != NULL); } } if (splitted) { CHECK(entry->me_value == NULL); } } } if (splitted) { CHECK(mp->ma_used <= SHARED_KEYS_MAX_SIZE); /* splitted table */ int duplicate_check = 0; for (Py_ssize_t i=0; i < mp->ma_used; i++) { int index = get_index_from_order(mp, i); CHECK((duplicate_check & (1<<index)) == 0); duplicate_check |= (1<<index); CHECK(mp->ma_values->values[index] != NULL); } } UNLOCK_KEYS_IF_SPLIT(keys, keys->dk_kind); } return 1; #undef CHECK } static PyDictKeysObject* new_keys_object(PyInterpreterState *interp, uint8_t log2_size, bool unicode) { PyDictKeysObject *dk; Py_ssize_t usable; int log2_bytes; size_t entry_size = unicode ? sizeof(PyDictUnicodeEntry) : sizeof(PyDictKeyEntry); assert(log2_size >= PyDict_LOG_MINSIZE); usable = USABLE_FRACTION((size_t)1<<log2_size); if (log2_size < 8) { log2_bytes = log2_size; } else if (log2_size < 16) { log2_bytes = log2_size + 1; } #if SIZEOF_VOID_P > 4 else if (log2_size >= 32) { log2_bytes = log2_size + 3; } #endif else { log2_bytes = log2_size + 2; } #ifdef WITH_FREELISTS struct _Py_dictkeys_freelist *freelist = get_dictkeys_freelist(); if (log2_size == PyDict_LOG_MINSIZE && unicode && freelist->numfree > 0) { dk = freelist->items[--freelist->numfree]; OBJECT_STAT_INC(from_freelist); } else #endif { dk = PyMem_Malloc(sizeof(PyDictKeysObject) + ((size_t)1 << log2_bytes) + entry_size * usable); if (dk == NULL) { PyErr_NoMemory(); return NULL; } } #ifdef Py_REF_DEBUG _Py_IncRefTotal(_PyThreadState_GET()); #endif dk->dk_refcnt = 1; dk->dk_log2_size = log2_size; dk->dk_log2_index_bytes = log2_bytes; dk->dk_kind = unicode ? DICT_KEYS_UNICODE : DICT_KEYS_GENERAL; #ifdef Py_GIL_DISABLED dk->dk_mutex = (PyMutex){0}; #endif dk->dk_nentries = 0; dk->dk_usable = usable; dk->dk_version = 0; memset(&dk->dk_indices[0], 0xff, ((size_t)1 << log2_bytes)); memset(&dk->dk_indices[(size_t)1 << log2_bytes], 0, entry_size * usable); return dk; } static void free_keys_object(PyDictKeysObject *keys, bool use_qsbr) { #ifdef Py_GIL_DISABLED if (use_qsbr) { _PyMem_FreeDelayed(keys, _PyDict_KeysSize(keys)); return; } #endif #ifdef WITH_FREELISTS struct _Py_dictkeys_freelist *freelist = get_dictkeys_freelist(); if (DK_LOG_SIZE(keys) == PyDict_LOG_MINSIZE && freelist->numfree < PyDict_MAXFREELIST && freelist->numfree >= 0 && DK_IS_UNICODE(keys)) { freelist->items[freelist->numfree++] = keys; OBJECT_STAT_INC(to_freelist); return; } #endif PyMem_Free(keys); } static size_t values_size_from_count(size_t count) { assert(count >= 1); size_t suffix_size = _Py_SIZE_ROUND_UP(count, sizeof(PyObject *)); assert(suffix_size < 128); assert(suffix_size % sizeof(PyObject *) == 0); return (count + 1) * sizeof(PyObject *) + suffix_size; } #define CACHED_KEYS(tp) (((PyHeapTypeObject*)tp)->ht_cached_keys) static inline PyDictValues* new_values(size_t size) { size_t n = values_size_from_count(size); PyDictValues *res = (PyDictValues *)PyMem_Malloc(n); if (res == NULL) { return NULL; } res->embedded = 0; res->size = 0; assert(size < 256); res->capacity = (uint8_t)size; return res; } static inline void free_values(PyDictValues *values, bool use_qsbr) { assert(values->embedded == 0); #ifdef Py_GIL_DISABLED if (use_qsbr) { _PyMem_FreeDelayed(values, values_size_from_count(values->capacity)); return; } #endif PyMem_Free(values); } /* Consumes a reference to the keys object */ static PyObject * new_dict(PyInterpreterState *interp, PyDictKeysObject *keys, PyDictValues *values, Py_ssize_t used, int free_values_on_failure) { PyDictObject *mp; assert(keys != NULL); #ifdef WITH_FREELISTS struct _Py_dict_freelist *freelist = get_dict_freelist(); if (freelist->numfree > 0) { mp = freelist->items[--freelist->numfree]; assert (mp != NULL); assert (Py_IS_TYPE(mp, &PyDict_Type)); OBJECT_STAT_INC(from_freelist); _Py_NewReference((PyObject *)mp); } else #endif { mp = PyObject_GC_New(PyDictObject, &PyDict_Type); if (mp == NULL) { dictkeys_decref(interp, keys, false); if (free_values_on_failure) { free_values(values, false); } return NULL; } } mp->ma_keys = keys; mp->ma_values = values; mp->ma_used = used; mp->ma_version_tag = DICT_NEXT_VERSION(interp); ASSERT_CONSISTENT(mp); return (PyObject *)mp; } static PyObject * new_dict_with_shared_keys(PyInterpreterState *interp, PyDictKeysObject *keys) { size_t size = shared_keys_usable_size(keys); PyDictValues *values = new_values(size); if (values == NULL) { return PyErr_NoMemory(); } dictkeys_incref(keys); for (size_t i = 0; i < size; i++) { values->values[i] = NULL; } return new_dict(interp, keys, values, 0, 1); } static PyDictKeysObject * clone_combined_dict_keys(PyDictObject *orig) { assert(PyDict_Check(orig)); assert(Py_TYPE(orig)->tp_iter == dict_iter); assert(orig->ma_values == NULL); assert(orig->ma_keys != Py_EMPTY_KEYS); assert(orig->ma_keys->dk_refcnt == 1); ASSERT_DICT_LOCKED(orig); size_t keys_size = _PyDict_KeysSize(orig->ma_keys); PyDictKeysObject *keys = PyMem_Malloc(keys_size); if (keys == NULL) { PyErr_NoMemory(); return NULL; } memcpy(keys, orig->ma_keys, keys_size); /* After copying key/value pairs, we need to incref all keys and values and they are about to be co-owned by a new dict object. */ PyObject **pkey, **pvalue; size_t offs; if (DK_IS_UNICODE(orig->ma_keys)) { PyDictUnicodeEntry *ep0 = DK_UNICODE_ENTRIES(keys); pkey = &ep0->me_key; pvalue = &ep0->me_value; offs = sizeof(PyDictUnicodeEntry) / sizeof(PyObject*); } else { PyDictKeyEntry *ep0 = DK_ENTRIES(keys); pkey = &ep0->me_key; pvalue = &ep0->me_value; offs = sizeof(PyDictKeyEntry) / sizeof(PyObject*); |
| You are visitor 01402520 — GeezerVison does not store the pages you look up. Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800×600. No cookies required. |