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 1.48 by root, Sun Sep 20 23:25:53 2015 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
111#define F_TEXT_STRINGS 0x00000080UL
105#define F_VALIDATE_UTF8 0x00000020UL 112#define F_VALIDATE_UTF8 0x00000100UL
106 113
107#define INIT_SIZE 32 // initial scalar size to be allocated 114#define INIT_SIZE 32 // initial scalar size to be allocated
108 115
109#define SB do { 116#define SB do {
110#define SE } while (0) 117#define SE } while (0)
… …
183 190
184// minimum length of a string to be registered for stringref 191// minimum length of a string to be registered for stringref
185ecb_inline int 192ecb_inline int
186minimum_string_length (UV idx) 193minimum_string_length (UV idx)
187{ 194{
188 return idx > 23 195 return idx <= 23 ? 3
189 ? idx > 0xffU 196 : idx <= 0xffU ? 4
190 ? idx > 0xffffU 197 : idx <= 0xffffU ? 5
191 ? idx > 0xffffffffU 198 : idx <= 0xffffffffU ? 7
192 ? 11 199 : 11;
193 : 7
194 : 5
195 : 4
196 : 3;
197} 200}
198 201
199///////////////////////////////////////////////////////////////////////////// 202/////////////////////////////////////////////////////////////////////////////
200// encoder 203// encoder
201 204
… …
214} enc_t; 217} enc_t;
215 218
216ecb_inline void 219ecb_inline void
217need (enc_t *enc, STRLEN len) 220need (enc_t *enc, STRLEN len)
218{ 221{
219 if (ecb_expect_false (enc->cur + len >= enc->end)) 222 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
220 { 223 {
221 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 224 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
222 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 225 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
223 enc->cur = SvPVX (enc->sv) + cur; 226 enc->cur = SvPVX (enc->sv) + cur;
224 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 227 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
276encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
277{ 280{
278 encode_uint (enc, MAJOR_TAG, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
279} 282}
280 283
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
281ecb_inline void 307ecb_inline void
282encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
283{ 309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
284 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
285 need (enc, len); 318 need (enc, len);
286 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
287 enc->cur += len; 320 enc->cur += len;
288} 321}
289 322
290static void 323ecb_inline void
291encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
292{ 325{
293 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
294 { 327 {
295 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
296 329
… …
307 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
308 ++enc->stringref_idx; 341 ++enc->stringref_idx;
309 } 342 }
310 } 343 }
311 344
312 encode_str (enc, utf8, str, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
313} 346}
314 347
315static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
316 349
317static void 350static void
… …
324 357
325 ++enc->depth; 358 ++enc->depth;
326 359
327 encode_uint (enc, MAJOR_ARRAY, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
328 361
329 if (SvMAGICAL (av)) 362 if (ecb_expect_false (SvMAGICAL (av)))
330 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
331 { 364 {
332 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
333 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
334 } 367 }
… …
353 ++enc->depth; 386 ++enc->depth;
354 387
355 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
356 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
357 390
358 if (mg) 391 if (ecb_expect_false (mg))
359 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
360 else 393 else
361 encode_uint (enc, MAJOR_MAP, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
362 395
363 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
364 { 397 {
365 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
366 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
367 else 400 else
368 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));
369 402
370 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));
371 } 404 }
372 405
373 if (mg) 406 if (ecb_expect_false (mg))
374 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
375 408
376 --enc->depth; 409 --enc->depth;
377} 410}
378 411
… …
421 } 454 }
422 455
423 if (ecb_expect_false (SvREFCNT (sv) > 1) 456 if (ecb_expect_false (SvREFCNT (sv) > 1)
424 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
425 { 458 {
426 if (!enc->shareable) 459 if (ecb_expect_false (!enc->shareable))
427 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
428 461
429 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
430 463
431 if (SvOK (*svp)) 464 if (SvOK (*svp))
… …
445 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
446 { 479 {
447 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
448 GV *method; 481 GV *method;
449 482
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))));
450 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
451 { 487 {
452 dSP; 488 dSP;
453 489
454 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
455 // we re-bless the reference to get overload and other niceties right 492 // we re-bless the reference to get overload and other niceties right
456 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
457 494
458 PUTBACK; 495 PUTBACK;
459 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
… …
472 } 509 }
473 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
474 { 511 {
475 dSP; 512 dSP;
476 513
477 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 PUSHMARK (SP);
478 EXTEND (SP, 2); 516 EXTEND (SP, 2);
479 // 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
480 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 518 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
481 PUSHs (sv_cbor); 519 PUSHs (sv_cbor);
482 520
… …
488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 526 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
489 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));
490 528
491 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 529 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
492 encode_uint (enc, MAJOR_ARRAY, count + 1); 530 encode_uint (enc, MAJOR_ARRAY, count + 1);
493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 531 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
494 532
495 while (count) 533 {
534 int i;
535
536 for (i = 0; i < count; ++i)
496 encode_sv (enc, SP[1 - count--]); 537 encode_sv (enc, SP[i + 1 - count]);
538
539 SP -= count;
540 }
497 541
498 PUTBACK; 542 PUTBACK;
499 543
500 FREETMPS; LEAVE; 544 FREETMPS; LEAVE;
501 } 545 }
… …
524 if (ecb_expect_false (nv == (NV)(U32)nv)) 568 if (ecb_expect_false (nv == (NV)(U32)nv))
525 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 569 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
526 //TODO: maybe I32? 570 //TODO: maybe I32?
527 else if (ecb_expect_false (nv == (float)nv)) 571 else if (ecb_expect_false (nv == (float)nv))
528 { 572 {
573 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
574
529 uint32_t fp = ecb_float_to_binary32 (nv); 575 uint32_t fp = ecb_float_to_binary32 (nv);
530
531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
532 576
533 if (!ecb_big_endian ()) 577 if (!ecb_big_endian ())
534 fp = ecb_bswap32 (fp); 578 fp = ecb_bswap32 (fp);
535 579
536 memcpy (enc->cur, &fp, 4); 580 memcpy (enc->cur, &fp, 4);
537 enc->cur += 4; 581 enc->cur += 4;
538 } 582 }
539 else 583 else
540 { 584 {
585 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
586
541 uint64_t fp = ecb_double_to_binary64 (nv); 587 uint64_t fp = ecb_double_to_binary64 (nv);
542
543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
544 588
545 if (!ecb_big_endian ()) 589 if (!ecb_big_endian ())
546 fp = ecb_bswap64 (fp); 590 fp = ecb_bswap64 (fp);
547 591
548 memcpy (enc->cur, &fp, 8); 592 memcpy (enc->cur, &fp, 8);
… …
557 601
558 if (SvPOKp (sv)) 602 if (SvPOKp (sv))
559 { 603 {
560 STRLEN len; 604 STRLEN len;
561 char *str = SvPV (sv, len); 605 char *str = SvPV (sv, len);
562 encode_strref (enc, SvUTF8 (sv), str, len); 606 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
563 } 607 }
564 else if (SvNOKp (sv)) 608 else if (SvNOKp (sv))
565 encode_nv (enc, sv); 609 encode_nv (enc, sv);
566 else if (SvIOKp (sv)) 610 else if (SvIOKp (sv))
567 { 611 {
… …
586static SV * 630static SV *
587encode_cbor (SV *scalar, CBOR *cbor) 631encode_cbor (SV *scalar, CBOR *cbor)
588{ 632{
589 enc_t enc = { 0 }; 633 enc_t enc = { 0 };
590 634
591 enc.cbor = *cbor; 635 enc.cbor = *cbor;
592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 636 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
593 enc.cur = SvPVX (enc.sv); 637 enc.cur = SvPVX (enc.sv);
594 enc.end = SvEND (enc.sv); 638 enc.end = SvEND (enc.sv);
595 639
596 SvPOK_only (enc.sv); 640 SvPOK_only (enc.sv);
597 641
598 if (cbor->flags & F_PACK_STRINGS) 642 if (cbor->flags & F_PACK_STRINGS)
599 { 643 {
… …
626 U32 depth; // recursion depth 670 U32 depth; // recursion depth
627 U32 maxdepth; // recursion depth limit 671 U32 maxdepth; // recursion depth limit
628 AV *shareable; 672 AV *shareable;
629 AV *stringref; 673 AV *stringref;
630 SV *decode_tagged; 674 SV *decode_tagged;
675 SV *err_sv; // optional sv for error, needs to be freed
631} dec_t; 676} dec_t;
632 677
633#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);
634 685
635#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;
636 689
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
637#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)
638#define DEC_DEC_DEPTH --dec->depth 715#define DEC_DEC_DEPTH --dec->depth
639 716
640static UV 717static UV
641decode_uint (dec_t *dec) 718decode_uint (dec_t *dec)
642{ 719{
… …
708 785
709 for (;;) 786 for (;;)
710 { 787 {
711 WANT (1); 788 WANT (1);
712 789
713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 790 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
714 { 791 {
715 ++dec->cur; 792 ++dec->cur;
716 break; 793 break;
717 } 794 }
718 795
719 av_push (av, decode_sv (dec)); 796 av_push (av, decode_sv (dec));
720 } 797 }
721 } 798 }
722 else 799 else
723 { 800 {
724 int i, len = decode_uint (dec); 801 UV i, len = decode_uint (dec);
725 802
726 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
727 av_fill (av, len - 1); 804 av_fill (av, len - 1);
728 805
729 for (i = 0; i < len; ++i) 806 for (i = 0; i < len; ++i)
… …
732 809
733 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
734 return newRV_noinc ((SV *)av); 811 return newRV_noinc ((SV *)av);
735 812
736fail: 813fail:
737 SvREFCNT_dec (av); 814 SvREFCNT_dec_NN (av);
738 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
739 return &PL_sv_undef; 816 return &PL_sv_undef;
740} 817}
741 818
742static void 819static void
… …
746 // byte or utf-8 strings as keys, but only when !stringref 823 // byte or utf-8 strings as keys, but only when !stringref
747 824
748 if (ecb_expect_true (!dec->stringref)) 825 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 826 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
750 { 827 {
751 I32 len = decode_uint (dec); 828 STRLEN len = decode_uint (dec);
752 char *key = (char *)dec->cur; 829 char *key = (char *)dec->cur;
753 830
831 WANT (len);
754 dec->cur += len; 832 dec->cur += len;
755 833
756 hv_store (hv, key, len, decode_sv (dec), 0); 834 hv_store (hv, key, len, decode_sv (dec), 0);
757 835
758 return; 836 return;
759 } 837 }
760 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))
761 { 839 {
762 I32 len = decode_uint (dec); 840 STRLEN len = decode_uint (dec);
763 char *key = (char *)dec->cur; 841 char *key = (char *)dec->cur;
764 842
843 WANT (len);
765 dec->cur += len; 844 dec->cur += len;
766 845
767 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 846 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
768 if (!is_utf8_string (key, len)) 847 if (!is_utf8_string (key, len))
769 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 848 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
774 } 853 }
775 854
776 SV *k = decode_sv (dec); 855 SV *k = decode_sv (dec);
777 SV *v = decode_sv (dec); 856 SV *v = decode_sv (dec);
778 857
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
779 hv_store_ent (hv, k, v, 0); 886 hv_store_ent (hv, k, v, 0);
780 SvREFCNT_dec (k); 887 SvREFCNT_dec_NN (k);
781 888
782fail: 889fail:
783 ; 890 ;
784} 891}
785 892
… …
796 903
797 for (;;) 904 for (;;)
798 { 905 {
799 WANT (1); 906 WANT (1);
800 907
801 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 908 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
802 { 909 {
803 ++dec->cur; 910 ++dec->cur;
804 break; 911 break;
805 } 912 }
806 913
807 decode_he (dec, hv); 914 decode_he (dec, hv);
808 } 915 }
809 } 916 }
810 else 917 else
811 { 918 {
812 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
813 922
814 while (pairs--) 923 while (pairs--)
815 decode_he (dec, hv); 924 decode_he (dec, hv);
816 } 925 }
817 926
818 DEC_DEC_DEPTH; 927 DEC_DEC_DEPTH;
819 return newRV_noinc ((SV *)hv); 928 return newRV_noinc ((SV *)hv);
820 929
821fail: 930fail:
822 SvREFCNT_dec (hv); 931 SvREFCNT_dec_NN (hv);
823 DEC_DEC_DEPTH; 932 DEC_DEC_DEPTH;
824 return &PL_sv_undef; 933 return &PL_sv_undef;
825} 934}
826 935
827static SV * 936static SV *
828decode_str (dec_t *dec, int utf8) 937decode_str (dec_t *dec, int utf8)
829{ 938{
830 SV *sv = 0; 939 SV *sv = 0;
831 940
832 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 941 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
833 { 942 {
834 // indefinite length strings 943 // indefinite length strings
835 ++dec->cur; 944 ++dec->cur;
836 945
837 U8 major = *dec->cur & MAJOR_MISC; 946 U8 major = *dec->cur & MAJOR_MISC;
… …
905 sv = newRV_noinc (decode_sv (dec)); 1014 sv = newRV_noinc (decode_sv (dec));
906 break; 1015 break;
907 1016
908 case CBOR_TAG_STRINGREF_NAMESPACE: 1017 case CBOR_TAG_STRINGREF_NAMESPACE:
909 { 1018 {
1019 // do not use SAVETMPS/FREETMPS, as these will
1020 // erase mortalised caches, e.g. "shareable"
910 ENTER; SAVETMPS; 1021 ENTER;
911 1022
912 SAVESPTR (dec->stringref); 1023 SAVESPTR (dec->stringref);
913 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1024 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
914 1025
915 sv = decode_sv (dec); 1026 sv = decode_sv (dec);
916 1027
917 FREETMPS; LEAVE; 1028 LEAVE;
918 } 1029 }
919 break; 1030 break;
920 1031
921 case CBOR_TAG_STRINGREF: 1032 case CBOR_TAG_STRINGREF:
922 { 1033 {
923 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1034 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
924 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1035 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
925 1036
926 UV idx = decode_uint (dec); 1037 UV idx = decode_uint (dec);
927 1038
928 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1039 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
929 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)");
930 1041
931 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1042 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
932 } 1043 }
933 break; 1044 break;
… …
961 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1072 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
962 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1073 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
963 1074
964 UV idx = decode_uint (dec); 1075 UV idx = decode_uint (dec);
965 1076
966 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1077 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
967 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1078 ERR ("corrupted CBOR data (sharedref index out of bounds)");
968 1079
969 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1080 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
970 1081
971 if (sv == &PL_sv_undef) 1082 if (sv == &PL_sv_undef)
… …
973 } 1084 }
974 break; 1085 break;
975 1086
976 case CBOR_TAG_PERL_OBJECT: 1087 case CBOR_TAG_PERL_OBJECT:
977 { 1088 {
1089 if (dec->cbor.flags & F_FORBID_OBJECTS)
1090 goto filter;
1091
978 sv = decode_sv (dec); 1092 sv = decode_sv (dec);
979 1093
980 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1094 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
981 ERR ("corrupted CBOR data (non-array perl object)"); 1095 ERR ("corrupted CBOR data (non-array perl object)");
982 1096
… …
992 if (!method) 1106 if (!method)
993 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1107 ERR ("cannot decode perl-object (package does not have a THAW method)");
994 1108
995 dSP; 1109 dSP;
996 1110
997 ENTER; SAVETMPS; PUSHMARK (SP); 1111 ENTER; SAVETMPS;
1112 PUSHMARK (SP);
998 EXTEND (SP, len + 1); 1113 EXTEND (SP, len + 1);
999 // we re-bless the reference to get overload and other niceties right 1114 // we re-bless the reference to get overload and other niceties right
1000 PUSHs (*av_fetch (av, 0, 1)); 1115 PUSHs (*av_fetch (av, 0, 1));
1001 PUSHs (sv_cbor); 1116 PUSHs (sv_cbor);
1002 1117
… …
1010 SPAGAIN; 1125 SPAGAIN;
1011 1126
1012 if (SvTRUE (ERRSV)) 1127 if (SvTRUE (ERRSV))
1013 { 1128 {
1014 FREETMPS; LEAVE; 1129 FREETMPS; LEAVE;
1015 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1130 ERR_ERRSV;
1016 } 1131 }
1017 1132
1018 SvREFCNT_dec (sv); 1133 SvREFCNT_dec_NN (sv);
1019 sv = SvREFCNT_inc (POPs); 1134 sv = SvREFCNT_inc (POPs);
1020 1135
1021 PUTBACK; 1136 PUTBACK;
1022 1137
1023 FREETMPS; LEAVE; 1138 FREETMPS; LEAVE;
1024 } 1139 }
1025 break; 1140 break;
1026 1141
1027 default: 1142 default:
1143 filter:
1028 { 1144 {
1145 SV *tag_sv = newSVuv (tag);
1146
1029 sv = decode_sv (dec); 1147 sv = decode_sv (dec);
1030 1148
1031 dSP; 1149 dSP;
1032 ENTER; SAVETMPS; PUSHMARK (SP); 1150 ENTER; SAVETMPS;
1151 PUSHMARK (SP);
1033 EXTEND (SP, 2); 1152 EXTEND (SP, 2);
1034 PUSHs (newSVuv (tag)); 1153 PUSHs (tag_sv);
1035 PUSHs (sv); 1154 PUSHs (sv);
1036 1155
1037 PUTBACK; 1156 PUTBACK;
1038 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);
1039 SPAGAIN; 1158 SPAGAIN;
1040 1159
1041 if (SvTRUE (ERRSV)) 1160 if (SvTRUE (ERRSV))
1042 { 1161 {
1162 SvREFCNT_dec_NN (tag_sv);
1043 FREETMPS; LEAVE; 1163 FREETMPS; LEAVE;
1044 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1164 ERR_ERRSV;
1045 } 1165 }
1046 1166
1047 if (count) 1167 if (count)
1048 { 1168 {
1169 SvREFCNT_dec_NN (tag_sv);
1049 SvREFCNT_dec (sv); 1170 SvREFCNT_dec_NN (sv);
1050 sv = SvREFCNT_inc (POPs); 1171 sv = SvREFCNT_inc_NN (TOPs);
1172 SP -= count;
1051 } 1173 }
1052 else 1174 else
1053 { 1175 {
1054 AV *av = newAV (); 1176 AV *av = newAV ();
1055 av_push (av, newSVuv (tag)); 1177 av_push (av, tag_sv);
1056 av_push (av, sv); 1178 av_push (av, sv);
1057 1179
1058 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1180 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1059 ? cbor_tagged_stash 1181 ? cbor_tagged_stash
1060 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1182 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1189 1311
1190 if (dec.err) 1312 if (dec.err)
1191 { 1313 {
1192 if (dec.shareable) 1314 if (dec.shareable)
1193 { 1315 {
1194 // need to break cyclic links, which whould all be in shareable 1316 // need to break cyclic links, which would all be in shareable
1195 int i; 1317 int i;
1196 SV **svp; 1318 SV **svp;
1197 1319
1198 for (i = av_len (dec.shareable) + 1; i--; ) 1320 for (i = av_len (dec.shareable) + 1; i--; )
1199 if ((svp = av_fetch (dec.shareable, i, 0))) 1321 if ((svp = av_fetch (dec.shareable, i, 0)))
1200 sv_setsv (*svp, &PL_sv_undef); 1322 sv_setsv (*svp, &PL_sv_undef);
1201 } 1323 }
1202 1324
1203 SvREFCNT_dec (sv); 1325 SvREFCNT_dec_NN (sv);
1326
1327 if (dec.err_sv)
1328 sv_2mortal (dec.err_sv);
1329
1204 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);
1205 } 1331 }
1206 1332
1207 sv = sv_2mortal (sv); 1333 sv = sv_2mortal (sv);
1208 1334
… …
1352 1478
1353 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1479 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1354 1480
1355 sv_cbor = newSVpv ("CBOR", 0); 1481 sv_cbor = newSVpv ("CBOR", 0);
1356 SvREADONLY_on (sv_cbor); 1482 SvREADONLY_on (sv_cbor);
1483
1484 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1357} 1485}
1358 1486
1359PROTOTYPES: DISABLE 1487PROTOTYPES: DISABLE
1360 1488
1361void CLONE (...) 1489void CLONE (...)
… …
1381 ALIAS: 1509 ALIAS:
1382 shrink = F_SHRINK 1510 shrink = F_SHRINK
1383 allow_unknown = F_ALLOW_UNKNOWN 1511 allow_unknown = F_ALLOW_UNKNOWN
1384 allow_sharing = F_ALLOW_SHARING 1512 allow_sharing = F_ALLOW_SHARING
1385 allow_cycles = F_ALLOW_CYCLES 1513 allow_cycles = F_ALLOW_CYCLES
1514 forbid_objects = F_FORBID_OBJECTS
1386 pack_strings = F_PACK_STRINGS 1515 pack_strings = F_PACK_STRINGS
1516 text_keys = F_TEXT_KEYS
1517 text_strings = F_TEXT_STRINGS
1387 validate_utf8 = F_VALIDATE_UTF8 1518 validate_utf8 = F_VALIDATE_UTF8
1388 PPCODE: 1519 PPCODE:
1389{ 1520{
1390 if (enable) 1521 if (enable)
1391 self->flags |= ix; 1522 self->flags |= ix;
… …
1399 ALIAS: 1530 ALIAS:
1400 get_shrink = F_SHRINK 1531 get_shrink = F_SHRINK
1401 get_allow_unknown = F_ALLOW_UNKNOWN 1532 get_allow_unknown = F_ALLOW_UNKNOWN
1402 get_allow_sharing = F_ALLOW_SHARING 1533 get_allow_sharing = F_ALLOW_SHARING
1403 get_allow_cycles = F_ALLOW_CYCLES 1534 get_allow_cycles = F_ALLOW_CYCLES
1535 get_forbid_objects = F_FORBID_OBJECTS
1404 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
1405 get_validate_utf8 = F_VALIDATE_UTF8 1539 get_validate_utf8 = F_VALIDATE_UTF8
1406 PPCODE: 1540 PPCODE:
1407 XPUSHs (boolSV (self->flags & ix)); 1541 XPUSHs (boolSV (self->flags & ix));
1408 1542
1409void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1543void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1539 cbor_init (&cbor); 1673 cbor_init (&cbor);
1540 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1674 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1541 XPUSHs (cborstr); 1675 XPUSHs (cborstr);
1542} 1676}
1543 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