ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/CBOR-XS/XS.xs
(Generate patch)

Comparing CBOR-XS/XS.xs (file contents):
Revision 1.51 by root, Sun Apr 24 13:15:19 2016 UTC vs.
Revision 1.65 by root, Thu Nov 15 19:52:41 2018 UTC

… …
6#include <string.h> 6#include <string.h>
7#include <stdlib.h> 7#include <stdlib.h>
8#include <stdio.h> 8#include <stdio.h>
9#include <limits.h> 9#include <limits.h>
10#include <float.h> 10#include <float.h>
11#include <inttypes.h>
11 12
12#define ECB_NO_THREADS 1 13#define ECB_NO_THREADS 1
13#include "ecb.h" 14#include "ecb.h"
14 15
15// compatibility with perl <5.18 16// compatibility with perl <5.18
… …
20# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
21#endif 22#endif
22#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
23# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
24#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
25#ifndef SvREFCNT_dec_NN 29#ifndef SvREFCNT_dec_NN
26# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
27#endif 31#endif
28 32
29// known major and minor types 33// known major and minor types
… …
99 103
100#define F_SHRINK 0x00000001UL 104#define F_SHRINK 0x00000001UL
101#define F_ALLOW_UNKNOWN 0x00000002UL 105#define F_ALLOW_UNKNOWN 0x00000002UL
102#define F_ALLOW_SHARING 0x00000004UL 106#define F_ALLOW_SHARING 0x00000004UL
103#define F_ALLOW_CYCLES 0x00000008UL 107#define F_ALLOW_CYCLES 0x00000008UL
108#define F_FORBID_OBJECTS 0x00000010UL
104#define F_PACK_STRINGS 0x00000010UL 109#define F_PACK_STRINGS 0x00000020UL
110#define F_TEXT_KEYS 0x00000040UL
105#define F_UTF8_STRINGS 0x00000020UL 111#define F_TEXT_STRINGS 0x00000080UL
106#define F_VALIDATE_UTF8 0x00000040UL 112#define F_VALIDATE_UTF8 0x00000100UL
107 113
108#define INIT_SIZE 32 // initial scalar size to be allocated 114#define INIT_SIZE 32 // initial scalar size to be allocated
109 115
110#define SB do { 116#define SB do {
111#define SE } while (0) 117#define SE } while (0)
… …
184 190
185// minimum length of a string to be registered for stringref 191// minimum length of a string to be registered for stringref
186ecb_inline int 192ecb_inline int
187minimum_string_length (UV idx) 193minimum_string_length (UV idx)
188{ 194{
189 return idx > 23 195 return idx <= 23 ? 3
190 ? idx > 0xffU 196 : idx <= 0xffU ? 4
191 ? idx > 0xffffU 197 : idx <= 0xffffU ? 5
192 ? idx > 0xffffffffU 198 : idx <= 0xffffffffU ? 7
193 ? 11 199 : 11;
194 : 7
195 : 5
196 : 4
197 : 3;
198} 200}
199 201
200///////////////////////////////////////////////////////////////////////////// 202/////////////////////////////////////////////////////////////////////////////
201// encoder 203// encoder
202 204
… …
215} enc_t; 217} enc_t;
216 218
217ecb_inline void 219ecb_inline void
218need (enc_t *enc, STRLEN len) 220need (enc_t *enc, STRLEN len)
219{ 221{
220 if (ecb_expect_false (enc->cur + len >= enc->end)) 222 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
221 { 223 {
222 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 224 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
223 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 225 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
224 enc->cur = SvPVX (enc->sv) + cur; 226 enc->cur = SvPVX (enc->sv) + cur;
225 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 227 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
277encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
278{ 280{
279 encode_uint (enc, MAJOR_TAG, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
280} 282}
281 283
284// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded
285ecb_noinline static void
286encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len)
287{
288 STRLEN ulen = len;
289 U8 *p, *pend = (U8 *)str + len;
290
291 for (p = (U8 *)str; p < pend; ++p)
292 ulen += *p >> 7; // count set high bits
293
294 encode_uint (enc, MAJOR_TEXT, ulen);
295
296 need (enc, ulen);
297 for (p = (U8 *)str; p < pend; ++p)
298 if (*p < 0x80)
299 *enc->cur++ = *p;
300 else
301 {
302 *enc->cur++ = 0xc0 + (*p >> 6);
303 *enc->cur++ = 0x80 + (*p & 63);
304 }
305}
306
282ecb_inline void 307ecb_inline void
283encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
284{ 309{
285 if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) 310 if (ecb_expect_false (upgrade_utf8))
286 if (!utf8) 311 if (!utf8)
287 { 312 {
288 SV *sv = sv_newmortal ();
289 char *s; STRLEN l;
290
291 sv_setpvn (sv, str, len);
292
293 s = SvPVutf8 (sv, l);
294 encode_str (enc, 1, s, l); 313 encode_str_utf8 (enc, utf8, str, len);
295 return; 314 return;
296 } 315 }
297 316
298 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
299 need (enc, len); 318 need (enc, len);
300 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
301 enc->cur += len; 320 enc->cur += len;
302} 321}
303 322
304static void 323ecb_inline void
305encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
306{ 325{
307 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
308 { 327 {
309 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
310 329
… …
321 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
322 ++enc->stringref_idx; 341 ++enc->stringref_idx;
323 } 342 }
324 } 343 }
325 344
326 encode_str (enc, utf8, str, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
327} 346}
328 347
329static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
330 349
331static void 350static void
… …
338 357
339 ++enc->depth; 358 ++enc->depth;
340 359
341 encode_uint (enc, MAJOR_ARRAY, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
342 361
343 if (SvMAGICAL (av)) 362 if (ecb_expect_false (SvMAGICAL (av)))
344 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
345 { 364 {
346 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
347 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
348 } 367 }
… …
367 ++enc->depth; 386 ++enc->depth;
368 387
369 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
370 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
371 390
372 if (mg) 391 if (ecb_expect_false (mg))
373 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
374 else 393 else
375 encode_uint (enc, MAJOR_MAP, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
376 395
377 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
378 { 397 {
379 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
380 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
381 else 400 else
382 encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); 401 encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he));
383 402
384 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); 403 encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he));
385 } 404 }
386 405
387 if (mg) 406 if (ecb_expect_false (mg))
388 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
389 408
390 --enc->depth; 409 --enc->depth;
391} 410}
392 411
… …
435 } 454 }
436 455
437 if (ecb_expect_false (SvREFCNT (sv) > 1) 456 if (ecb_expect_false (SvREFCNT (sv) > 1)
438 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
439 { 458 {
440 if (!enc->shareable) 459 if (ecb_expect_false (!enc->shareable))
441 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
442 461
443 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
444 463
445 if (SvOK (*svp)) 464 if (SvOK (*svp))
… …
459 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
460 { 479 {
461 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
462 GV *method; 481 GV *method;
463 482
483 if (enc->cbor.flags & F_FORBID_OBJECTS)
484 croak ("encountered object '%s', but forbid_objects is enabled",
485 SvPV_nolen (sv_2mortal (newRV_inc (sv))));
464 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
465 { 487 {
466 dSP; 488 dSP;
467 489
468 ENTER; SAVETMPS; 490 ENTER; SAVETMPS;
469 PUSHMARK (SP); 491 PUSHMARK (SP);
… …
488 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
489 { 511 {
490 dSP; 512 dSP;
491 513
492 ENTER; SAVETMPS; 514 ENTER; SAVETMPS;
493 SAVESTACK_POS ();
494 PUSHMARK (SP); 515 PUSHMARK (SP);
495 EXTEND (SP, 2); 516 EXTEND (SP, 2);
496 // we re-bless the reference to get overload and other niceties right 517 // we re-bless the reference to get overload and other niceties right
497 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 518 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
498 PUSHs (sv_cbor); 519 PUSHs (sv_cbor);
… …
505 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 526 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
506 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash)); 527 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
507 528
508 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 529 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
509 encode_uint (enc, MAJOR_ARRAY, count + 1); 530 encode_uint (enc, MAJOR_ARRAY, count + 1);
510 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 531 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
511 532
512 while (count) 533 {
534 int i;
535
536 for (i = 0; i < count; ++i)
513 encode_sv (enc, SP[1 - count--]); 537 encode_sv (enc, SP[i + 1 - count]);
538
539 SP -= count;
540 }
514 541
515 PUTBACK; 542 PUTBACK;
516 543
517 FREETMPS; LEAVE; 544 FREETMPS; LEAVE;
518 } 545 }
… …
541 if (ecb_expect_false (nv == (NV)(U32)nv)) 568 if (ecb_expect_false (nv == (NV)(U32)nv))
542 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 569 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
543 //TODO: maybe I32? 570 //TODO: maybe I32?
544 else if (ecb_expect_false (nv == (float)nv)) 571 else if (ecb_expect_false (nv == (float)nv))
545 { 572 {
573 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
574
546 uint32_t fp = ecb_float_to_binary32 (nv); 575 uint32_t fp = ecb_float_to_binary32 (nv);
547
548 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
549 576
550 if (!ecb_big_endian ()) 577 if (!ecb_big_endian ())
551 fp = ecb_bswap32 (fp); 578 fp = ecb_bswap32 (fp);
552 579
553 memcpy (enc->cur, &fp, 4); 580 memcpy (enc->cur, &fp, 4);
554 enc->cur += 4; 581 enc->cur += 4;
555 } 582 }
556 else 583 else
557 { 584 {
585 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
586
558 uint64_t fp = ecb_double_to_binary64 (nv); 587 uint64_t fp = ecb_double_to_binary64 (nv);
559
560 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
561 588
562 if (!ecb_big_endian ()) 589 if (!ecb_big_endian ())
563 fp = ecb_bswap64 (fp); 590 fp = ecb_bswap64 (fp);
564 591
565 memcpy (enc->cur, &fp, 8); 592 memcpy (enc->cur, &fp, 8);
… …
574 601
575 if (SvPOKp (sv)) 602 if (SvPOKp (sv))
576 { 603 {
577 STRLEN len; 604 STRLEN len;
578 char *str = SvPV (sv, len); 605 char *str = SvPV (sv, len);
579 encode_strref (enc, SvUTF8 (sv), str, len); 606 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
580 } 607 }
581 else if (SvNOKp (sv)) 608 else if (SvNOKp (sv))
582 encode_nv (enc, sv); 609 encode_nv (enc, sv);
583 else if (SvIOKp (sv)) 610 else if (SvIOKp (sv))
584 { 611 {
… …
603static SV * 630static SV *
604encode_cbor (SV *scalar, CBOR *cbor) 631encode_cbor (SV *scalar, CBOR *cbor)
605{ 632{
606 enc_t enc = { 0 }; 633 enc_t enc = { 0 };
607 634
608 enc.cbor = *cbor; 635 enc.cbor = *cbor;
609 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 636 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
610 enc.cur = SvPVX (enc.sv); 637 enc.cur = SvPVX (enc.sv);
611 enc.end = SvEND (enc.sv); 638 enc.end = SvEND (enc.sv);
612 639
613 SvPOK_only (enc.sv); 640 SvPOK_only (enc.sv);
614 641
615 if (cbor->flags & F_PACK_STRINGS) 642 if (cbor->flags & F_PACK_STRINGS)
616 { 643 {
… …
643 U32 depth; // recursion depth 670 U32 depth; // recursion depth
644 U32 maxdepth; // recursion depth limit 671 U32 maxdepth; // recursion depth limit
645 AV *shareable; 672 AV *shareable;
646 AV *stringref; 673 AV *stringref;
647 SV *decode_tagged; 674 SV *decode_tagged;
675 SV *err_sv; // optional sv for error, needs to be freed
648} dec_t; 676} dec_t;
649 677
650#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 678// set dec->err to ERRSV
679ecb_cold static void
680err_errsv (dec_t *dec)
681{
682 if (!dec->err)
683 {
684 dec->err_sv = newSVsv (ERRSV);
651 685
652#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 686 // chop off the trailing \n
687 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
688 *SvEND (dec->err_sv) = 0;
653 689
690 dec->err = SvPVutf8_nolen (dec->err_sv);
691 }
692}
693
694// the following functions are used to reduce code size and help the compiler to optimise
695ecb_cold static void
696err_set (dec_t *dec, const char *reason)
697{
698 if (!dec->err)
699 dec->err = reason;
700}
701
702ecb_cold static void
703err_unexpected_end (dec_t *dec)
704{
705 err_set (dec, "unexpected end of CBOR data");
706}
707
708#define ERR_DO(do) SB do; goto fail; SE
709#define ERR(reason) ERR_DO (err_set (dec, reason))
710#define ERR_ERRSV ERR_DO (err_errsv (dec))
711
712#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
713
654#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 714#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED)
655#define DEC_DEC_DEPTH --dec->depth 715#define DEC_DEC_DEPTH --dec->depth
656 716
657static UV 717static UV
658decode_uint (dec_t *dec) 718decode_uint (dec_t *dec)
659{ 719{
… …
725 785
726 for (;;) 786 for (;;)
727 { 787 {
728 WANT (1); 788 WANT (1);
729 789
730 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 790 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
731 { 791 {
732 ++dec->cur; 792 ++dec->cur;
733 break; 793 break;
734 } 794 }
735 795
736 av_push (av, decode_sv (dec)); 796 av_push (av, decode_sv (dec));
737 } 797 }
738 } 798 }
739 else 799 else
740 { 800 {
741 int i, len = decode_uint (dec); 801 UV i, len = decode_uint (dec);
742 802
743 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 803 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays
744 av_fill (av, len - 1); 804 av_fill (av, len - 1);
745 805
746 for (i = 0; i < len; ++i) 806 for (i = 0; i < len; ++i)
… …
749 809
750 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
751 return newRV_noinc ((SV *)av); 811 return newRV_noinc ((SV *)av);
752 812
753fail: 813fail:
754 SvREFCNT_dec (av); 814 SvREFCNT_dec_NN (av);
755 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
756 return &PL_sv_undef; 816 return &PL_sv_undef;
757} 817}
758 818
759static void 819static void
… …
763 // byte or utf-8 strings as keys, but only when !stringref 823 // byte or utf-8 strings as keys, but only when !stringref
764 824
765 if (ecb_expect_true (!dec->stringref)) 825 if (ecb_expect_true (!dec->stringref))
766 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 826 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
767 { 827 {
768 I32 len = decode_uint (dec); 828 STRLEN len = decode_uint (dec);
769 char *key = (char *)dec->cur; 829 char *key = (char *)dec->cur;
770 830
771 WANT (len); 831 WANT (len);
772 dec->cur += len; 832 dec->cur += len;
773 833
… …
775 835
776 return; 836 return;
777 } 837 }
778 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 838 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
779 { 839 {
780 I32 len = decode_uint (dec); 840 STRLEN len = decode_uint (dec);
781 char *key = (char *)dec->cur; 841 char *key = (char *)dec->cur;
782 842
783 WANT (len); 843 WANT (len);
784 dec->cur += len; 844 dec->cur += len;
785 845
… …
793 } 853 }
794 854
795 SV *k = decode_sv (dec); 855 SV *k = decode_sv (dec);
796 SV *v = decode_sv (dec); 856 SV *v = decode_sv (dec);
797 857
858 // we leak memory if uncaught exceptions are thrown by random magical
859 // methods, and this is hopefully the only place where it can happen,
860 // so if there is a chance of an exception, take the very slow path.
861 // since catching exceptions is "undocumented/internal/forbidden" by
862 // the new p5p powers, we need to call out to a perl function :/
863 if (ecb_expect_false (SvAMAGIC (k)))
864 {
865 dSP;
866
867 ENTER; SAVETMPS;
868 PUSHMARK (SP);
869 EXTEND (SP, 3);
870 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
871 PUSHs (sv_2mortal (k));
872 PUSHs (sv_2mortal (v));
873
874 PUTBACK;
875 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
876 SPAGAIN;
877
878 FREETMPS; LEAVE;
879
880 if (SvTRUE (ERRSV))
881 ERR_ERRSV;
882
883 return;
884 }
885
798 hv_store_ent (hv, k, v, 0); 886 hv_store_ent (hv, k, v, 0);
799 SvREFCNT_dec (k); 887 SvREFCNT_dec_NN (k);
800 888
801fail: 889fail:
802 ; 890 ;
803} 891}
804 892
… …
815 903
816 for (;;) 904 for (;;)
817 { 905 {
818 WANT (1); 906 WANT (1);
819 907
820 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 908 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
821 { 909 {
822 ++dec->cur; 910 ++dec->cur;
823 break; 911 break;
824 } 912 }
825 913
826 decode_he (dec, hv); 914 decode_he (dec, hv);
827 } 915 }
828 } 916 }
829 else 917 else
830 { 918 {
831 int pairs = decode_uint (dec); 919 UV pairs = decode_uint (dec);
920
921 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
832 922
833 while (pairs--) 923 while (pairs--)
834 decode_he (dec, hv); 924 decode_he (dec, hv);
835 } 925 }
836 926
837 DEC_DEC_DEPTH; 927 DEC_DEC_DEPTH;
838 return newRV_noinc ((SV *)hv); 928 return newRV_noinc ((SV *)hv);
839 929
840fail: 930fail:
841 SvREFCNT_dec (hv); 931 SvREFCNT_dec_NN (hv);
842 DEC_DEC_DEPTH; 932 DEC_DEC_DEPTH;
843 return &PL_sv_undef; 933 return &PL_sv_undef;
844} 934}
845 935
846static SV * 936static SV *
847decode_str (dec_t *dec, int utf8) 937decode_str (dec_t *dec, int utf8)
848{ 938{
849 SV *sv = 0; 939 SV *sv = 0;
850 940
851 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 941 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
852 { 942 {
853 // indefinite length strings 943 // indefinite length strings
854 ++dec->cur; 944 ++dec->cur;
855 945
856 U8 major = *dec->cur & MAJOR_MISC; 946 U8 major = *dec->cur & MAJOR_MISC;
… …
924 sv = newRV_noinc (decode_sv (dec)); 1014 sv = newRV_noinc (decode_sv (dec));
925 break; 1015 break;
926 1016
927 case CBOR_TAG_STRINGREF_NAMESPACE: 1017 case CBOR_TAG_STRINGREF_NAMESPACE:
928 { 1018 {
1019 // do not use SAVETMPS/FREETMPS, as these will
1020 // erase mortalised caches, e.g. "shareable"
929 ENTER; SAVETMPS; 1021 ENTER;
930 1022
931 SAVESPTR (dec->stringref); 1023 SAVESPTR (dec->stringref);
932 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1024 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
933 1025
934 sv = decode_sv (dec); 1026 sv = decode_sv (dec);
935 1027
936 FREETMPS; LEAVE; 1028 LEAVE;
937 } 1029 }
938 break; 1030 break;
939 1031
940 case CBOR_TAG_STRINGREF: 1032 case CBOR_TAG_STRINGREF:
941 { 1033 {
942 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1034 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
943 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1035 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
944 1036
945 UV idx = decode_uint (dec); 1037 UV idx = decode_uint (dec);
946 1038
947 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1039 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
948 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1040 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
949 1041
950 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1042 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
951 } 1043 }
952 break; 1044 break;
… …
980 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1072 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
981 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1073 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
982 1074
983 UV idx = decode_uint (dec); 1075 UV idx = decode_uint (dec);
984 1076
985 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1077 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
986 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1078 ERR ("corrupted CBOR data (sharedref index out of bounds)");
987 1079
988 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1080 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
989 1081
990 if (sv == &PL_sv_undef) 1082 if (sv == &PL_sv_undef)
… …
992 } 1084 }
993 break; 1085 break;
994 1086
995 case CBOR_TAG_PERL_OBJECT: 1087 case CBOR_TAG_PERL_OBJECT:
996 { 1088 {
1089 if (dec->cbor.flags & F_FORBID_OBJECTS)
1090 goto filter;
1091
997 sv = decode_sv (dec); 1092 sv = decode_sv (dec);
998 1093
999 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1094 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
1000 ERR ("corrupted CBOR data (non-array perl object)"); 1095 ERR ("corrupted CBOR data (non-array perl object)");
1001 1096
… …
1030 SPAGAIN; 1125 SPAGAIN;
1031 1126
1032 if (SvTRUE (ERRSV)) 1127 if (SvTRUE (ERRSV))
1033 { 1128 {
1034 FREETMPS; LEAVE; 1129 FREETMPS; LEAVE;
1035 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1130 ERR_ERRSV;
1036 } 1131 }
1037 1132
1038 SvREFCNT_dec (sv); 1133 SvREFCNT_dec_NN (sv);
1039 sv = SvREFCNT_inc (POPs); 1134 sv = SvREFCNT_inc (POPs);
1040 1135
1041 PUTBACK; 1136 PUTBACK;
1042 1137
1043 FREETMPS; LEAVE; 1138 FREETMPS; LEAVE;
1044 } 1139 }
1045 break; 1140 break;
1046 1141
1047 default: 1142 default:
1143 filter:
1048 { 1144 {
1145 SV *tag_sv = newSVuv (tag);
1146
1049 sv = decode_sv (dec); 1147 sv = decode_sv (dec);
1050 1148
1051 dSP; 1149 dSP;
1052 ENTER; SAVETMPS; 1150 ENTER; SAVETMPS;
1053 SAVESTACK_POS ();
1054 PUSHMARK (SP); 1151 PUSHMARK (SP);
1055 EXTEND (SP, 2); 1152 EXTEND (SP, 2);
1056 PUSHs (newSVuv (tag)); 1153 PUSHs (tag_sv);
1057 PUSHs (sv); 1154 PUSHs (sv);
1058 1155
1059 PUTBACK; 1156 PUTBACK;
1060 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1157 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1061 SPAGAIN; 1158 SPAGAIN;
1062 1159
1063 if (SvTRUE (ERRSV)) 1160 if (SvTRUE (ERRSV))
1064 { 1161 {
1162 SvREFCNT_dec_NN (tag_sv);
1065 FREETMPS; LEAVE; 1163 FREETMPS; LEAVE;
1066 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1164 ERR_ERRSV;
1067 } 1165 }
1068 1166
1069 if (count) 1167 if (count)
1070 { 1168 {
1169 SvREFCNT_dec_NN (tag_sv);
1071 SvREFCNT_dec (sv); 1170 SvREFCNT_dec_NN (sv);
1072 sv = SvREFCNT_inc (POPs); 1171 sv = SvREFCNT_inc_NN (TOPs);
1172 SP -= count;
1073 } 1173 }
1074 else 1174 else
1075 { 1175 {
1076 AV *av = newAV (); 1176 AV *av = newAV ();
1077 av_push (av, newSVuv (tag)); 1177 av_push (av, tag_sv);
1078 av_push (av, sv); 1178 av_push (av, sv);
1079 1179
1080 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1180 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1081 ? cbor_tagged_stash 1181 ? cbor_tagged_stash
1082 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1182 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1211 1311
1212 if (dec.err) 1312 if (dec.err)
1213 { 1313 {
1214 if (dec.shareable) 1314 if (dec.shareable)
1215 { 1315 {
1216 // need to break cyclic links, which whould all be in shareable 1316 // need to break cyclic links, which would all be in shareable
1217 int i; 1317 int i;
1218 SV **svp; 1318 SV **svp;
1219 1319
1220 for (i = av_len (dec.shareable) + 1; i--; ) 1320 for (i = av_len (dec.shareable) + 1; i--; )
1221 if ((svp = av_fetch (dec.shareable, i, 0))) 1321 if ((svp = av_fetch (dec.shareable, i, 0)))
1222 sv_setsv (*svp, &PL_sv_undef); 1322 sv_setsv (*svp, &PL_sv_undef);
1223 } 1323 }
1224 1324
1225 SvREFCNT_dec (sv); 1325 SvREFCNT_dec_NN (sv);
1326
1327 if (dec.err_sv)
1328 sv_2mortal (dec.err_sv);
1329
1226 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1330 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1227 } 1331 }
1228 1332
1229 sv = sv_2mortal (sv); 1333 sv = sv_2mortal (sv);
1230 1334
… …
1374 1478
1375 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1479 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1376 1480
1377 sv_cbor = newSVpv ("CBOR", 0); 1481 sv_cbor = newSVpv ("CBOR", 0);
1378 SvREADONLY_on (sv_cbor); 1482 SvREADONLY_on (sv_cbor);
1483
1484 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1379} 1485}
1380 1486
1381PROTOTYPES: DISABLE 1487PROTOTYPES: DISABLE
1382 1488
1383void CLONE (...) 1489void CLONE (...)
… …
1403 ALIAS: 1509 ALIAS:
1404 shrink = F_SHRINK 1510 shrink = F_SHRINK
1405 allow_unknown = F_ALLOW_UNKNOWN 1511 allow_unknown = F_ALLOW_UNKNOWN
1406 allow_sharing = F_ALLOW_SHARING 1512 allow_sharing = F_ALLOW_SHARING
1407 allow_cycles = F_ALLOW_CYCLES 1513 allow_cycles = F_ALLOW_CYCLES
1514 forbid_objects = F_FORBID_OBJECTS
1408 pack_strings = F_PACK_STRINGS 1515 pack_strings = F_PACK_STRINGS
1516 text_keys = F_TEXT_KEYS
1409 utf8_strings = F_UTF8_STRINGS 1517 text_strings = F_TEXT_STRINGS
1410 validate_utf8 = F_VALIDATE_UTF8 1518 validate_utf8 = F_VALIDATE_UTF8
1411 PPCODE: 1519 PPCODE:
1412{ 1520{
1413 if (enable) 1521 if (enable)
1414 self->flags |= ix; 1522 self->flags |= ix;
… …
1422 ALIAS: 1530 ALIAS:
1423 get_shrink = F_SHRINK 1531 get_shrink = F_SHRINK
1424 get_allow_unknown = F_ALLOW_UNKNOWN 1532 get_allow_unknown = F_ALLOW_UNKNOWN
1425 get_allow_sharing = F_ALLOW_SHARING 1533 get_allow_sharing = F_ALLOW_SHARING
1426 get_allow_cycles = F_ALLOW_CYCLES 1534 get_allow_cycles = F_ALLOW_CYCLES
1535 get_forbid_objects = F_FORBID_OBJECTS
1427 get_pack_strings = F_PACK_STRINGS 1536 get_pack_strings = F_PACK_STRINGS
1537 get_text_keys = F_TEXT_KEYS
1538 get_text_strings = F_TEXT_STRINGS
1428 get_validate_utf8 = F_VALIDATE_UTF8 1539 get_validate_utf8 = F_VALIDATE_UTF8
1429 PPCODE: 1540 PPCODE:
1430 XPUSHs (boolSV (self->flags & ix)); 1541 XPUSHs (boolSV (self->flags & ix));
1431 1542
1432void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1543void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1562 cbor_init (&cbor); 1673 cbor_init (&cbor);
1563 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1674 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1564 XPUSHs (cborstr); 1675 XPUSHs (cborstr);
1565} 1676}
1566 1677
1678#ifdef __AFL_COMPILER
1679
1680void
1681afl_init ()
1682 CODE:
1683 __AFL_INIT ();
1684
1685int
1686afl_loop (unsigned int count = 10000)
1687 CODE:
1688 RETVAL = __AFL_LOOP (count);
1689 OUTPUT:
1690 RETVAL
1691
1692#endif
1693

Diff Legend

– Removed lines
+ Added lines
< Changed lines
> Changed lines