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 1.64 by root, Tue Jun 27 02:03:24 2017 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 SAVESTACK_POS ();
516 PUSHMARK (SP);
478 EXTEND (SP, 2); 517 EXTEND (SP, 2);
479 // we re-bless the reference to get overload and other niceties right 518 // we re-bless the reference to get overload and other niceties right
480 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 519 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
481 PUSHs (sv_cbor); 520 PUSHs (sv_cbor);
482 521
… …
488 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 527 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)); 528 croak ("%s::FREEZE(CBOR) method returned same object as was passed instead of a new one", HvNAME (stash));
490 529
491 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 530 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
492 encode_uint (enc, MAJOR_ARRAY, count + 1); 531 encode_uint (enc, MAJOR_ARRAY, count + 1);
493 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 532 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
494 533
495 while (count) 534 while (count)
496 encode_sv (enc, SP[1 - count--]); 535 encode_sv (enc, SP[1 - count--]);
497 536
498 PUTBACK; 537 PUTBACK;
… …
524 if (ecb_expect_false (nv == (NV)(U32)nv)) 563 if (ecb_expect_false (nv == (NV)(U32)nv))
525 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 564 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
526 //TODO: maybe I32? 565 //TODO: maybe I32?
527 else if (ecb_expect_false (nv == (float)nv)) 566 else if (ecb_expect_false (nv == (float)nv))
528 { 567 {
568 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
569
529 uint32_t fp = ecb_float_to_binary32 (nv); 570 uint32_t fp = ecb_float_to_binary32 (nv);
530
531 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
532 571
533 if (!ecb_big_endian ()) 572 if (!ecb_big_endian ())
534 fp = ecb_bswap32 (fp); 573 fp = ecb_bswap32 (fp);
535 574
536 memcpy (enc->cur, &fp, 4); 575 memcpy (enc->cur, &fp, 4);
537 enc->cur += 4; 576 enc->cur += 4;
538 } 577 }
539 else 578 else
540 { 579 {
580 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
581
541 uint64_t fp = ecb_double_to_binary64 (nv); 582 uint64_t fp = ecb_double_to_binary64 (nv);
542
543 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
544 583
545 if (!ecb_big_endian ()) 584 if (!ecb_big_endian ())
546 fp = ecb_bswap64 (fp); 585 fp = ecb_bswap64 (fp);
547 586
548 memcpy (enc->cur, &fp, 8); 587 memcpy (enc->cur, &fp, 8);
… …
557 596
558 if (SvPOKp (sv)) 597 if (SvPOKp (sv))
559 { 598 {
560 STRLEN len; 599 STRLEN len;
561 char *str = SvPV (sv, len); 600 char *str = SvPV (sv, len);
562 encode_strref (enc, SvUTF8 (sv), str, len); 601 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
563 } 602 }
564 else if (SvNOKp (sv)) 603 else if (SvNOKp (sv))
565 encode_nv (enc, sv); 604 encode_nv (enc, sv);
566 else if (SvIOKp (sv)) 605 else if (SvIOKp (sv))
567 { 606 {
… …
586static SV * 625static SV *
587encode_cbor (SV *scalar, CBOR *cbor) 626encode_cbor (SV *scalar, CBOR *cbor)
588{ 627{
589 enc_t enc = { 0 }; 628 enc_t enc = { 0 };
590 629
591 enc.cbor = *cbor; 630 enc.cbor = *cbor;
592 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 631 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
593 enc.cur = SvPVX (enc.sv); 632 enc.cur = SvPVX (enc.sv);
594 enc.end = SvEND (enc.sv); 633 enc.end = SvEND (enc.sv);
595 634
596 SvPOK_only (enc.sv); 635 SvPOK_only (enc.sv);
597 636
598 if (cbor->flags & F_PACK_STRINGS) 637 if (cbor->flags & F_PACK_STRINGS)
599 { 638 {
… …
626 U32 depth; // recursion depth 665 U32 depth; // recursion depth
627 U32 maxdepth; // recursion depth limit 666 U32 maxdepth; // recursion depth limit
628 AV *shareable; 667 AV *shareable;
629 AV *stringref; 668 AV *stringref;
630 SV *decode_tagged; 669 SV *decode_tagged;
670 SV *err_sv; // optional sv for error, needs to be freed
631} dec_t; 671} dec_t;
632 672
633#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE 673// set dec->err to ERRSV
674ecb_cold static void
675err_errsv (dec_t *dec)
676{
677 if (!dec->err)
678 {
679 dec->err_sv = newSVsv (ERRSV);
634 680
635#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") 681 // chop off the trailing \n
682 SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1);
683 *SvEND (dec->err_sv) = 0;
636 684
685 dec->err = SvPVutf8_nolen (dec->err_sv);
686 }
687}
688
689// the following functions are used to reduce code size and help the compiler to optimise
690ecb_cold static void
691err_set (dec_t *dec, const char *reason)
692{
693 if (!dec->err)
694 dec->err = reason;
695}
696
697ecb_cold static void
698err_unexpected_end (dec_t *dec)
699{
700 err_set (dec, "unexpected end of CBOR data");
701}
702
703#define ERR_DO(do) SB do; goto fail; SE
704#define ERR(reason) ERR_DO (err_set (dec, reason))
705#define ERR_ERRSV ERR_DO (err_errsv (dec))
706
707#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec))
708
637#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) 709#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 710#define DEC_DEC_DEPTH --dec->depth
639 711
640static UV 712static UV
641decode_uint (dec_t *dec) 713decode_uint (dec_t *dec)
642{ 714{
… …
708 780
709 for (;;) 781 for (;;)
710 { 782 {
711 WANT (1); 783 WANT (1);
712 784
713 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 785 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
714 { 786 {
715 ++dec->cur; 787 ++dec->cur;
716 break; 788 break;
717 } 789 }
718 790
719 av_push (av, decode_sv (dec)); 791 av_push (av, decode_sv (dec));
720 } 792 }
721 } 793 }
722 else 794 else
723 { 795 {
724 int i, len = decode_uint (dec); 796 UV i, len = decode_uint (dec);
725 797
726 WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays 798 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); 799 av_fill (av, len - 1);
728 800
729 for (i = 0; i < len; ++i) 801 for (i = 0; i < len; ++i)
… …
732 804
733 DEC_DEC_DEPTH; 805 DEC_DEC_DEPTH;
734 return newRV_noinc ((SV *)av); 806 return newRV_noinc ((SV *)av);
735 807
736fail: 808fail:
737 SvREFCNT_dec (av); 809 SvREFCNT_dec_NN (av);
738 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
739 return &PL_sv_undef; 811 return &PL_sv_undef;
740} 812}
741 813
742static void 814static void
… …
746 // byte or utf-8 strings as keys, but only when !stringref 818 // byte or utf-8 strings as keys, but only when !stringref
747 819
748 if (ecb_expect_true (!dec->stringref)) 820 if (ecb_expect_true (!dec->stringref))
749 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 821 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
750 { 822 {
751 I32 len = decode_uint (dec); 823 STRLEN len = decode_uint (dec);
752 char *key = (char *)dec->cur; 824 char *key = (char *)dec->cur;
753 825
826 WANT (len);
754 dec->cur += len; 827 dec->cur += len;
755 828
756 hv_store (hv, key, len, decode_sv (dec), 0); 829 hv_store (hv, key, len, decode_sv (dec), 0);
757 830
758 return; 831 return;
759 } 832 }
760 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 833 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
761 { 834 {
762 I32 len = decode_uint (dec); 835 STRLEN len = decode_uint (dec);
763 char *key = (char *)dec->cur; 836 char *key = (char *)dec->cur;
764 837
838 WANT (len);
765 dec->cur += len; 839 dec->cur += len;
766 840
767 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 841 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
768 if (!is_utf8_string (key, len)) 842 if (!is_utf8_string (key, len))
769 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 843 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
774 } 848 }
775 849
776 SV *k = decode_sv (dec); 850 SV *k = decode_sv (dec);
777 SV *v = decode_sv (dec); 851 SV *v = decode_sv (dec);
778 852
853 // we leak memory if uncaught exceptions are thrown by random magical
854 // methods, and this is hopefully the only place where it can happen,
855 // so if there is a chance of an exception, take the very slow path.
856 // since catching exceptions is "undocumented/internal/forbidden" by
857 // the new p5p powers, we need to call out to a perl function :/
858 if (ecb_expect_false (SvAMAGIC (k)))
859 {
860 dSP;
861
862 ENTER; SAVETMPS;
863 PUSHMARK (SP);
864 EXTEND (SP, 3);
865 PUSHs (sv_2mortal (newRV_inc ((SV *)hv)));
866 PUSHs (sv_2mortal (k));
867 PUSHs (sv_2mortal (v));
868
869 PUTBACK;
870 call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL);
871 SPAGAIN;
872
873 FREETMPS; LEAVE;
874
875 if (SvTRUE (ERRSV))
876 ERR_ERRSV;
877
878 return;
879 }
880
779 hv_store_ent (hv, k, v, 0); 881 hv_store_ent (hv, k, v, 0);
780 SvREFCNT_dec (k); 882 SvREFCNT_dec_NN (k);
781 883
782fail: 884fail:
783 ; 885 ;
784} 886}
785 887
… …
796 898
797 for (;;) 899 for (;;)
798 { 900 {
799 WANT (1); 901 WANT (1);
800 902
801 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 903 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
802 { 904 {
803 ++dec->cur; 905 ++dec->cur;
804 break; 906 break;
805 } 907 }
806 908
807 decode_he (dec, hv); 909 decode_he (dec, hv);
808 } 910 }
809 } 911 }
810 else 912 else
811 { 913 {
812 int pairs = decode_uint (dec); 914 UV pairs = decode_uint (dec);
915
916 WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes
813 917
814 while (pairs--) 918 while (pairs--)
815 decode_he (dec, hv); 919 decode_he (dec, hv);
816 } 920 }
817 921
818 DEC_DEC_DEPTH; 922 DEC_DEC_DEPTH;
819 return newRV_noinc ((SV *)hv); 923 return newRV_noinc ((SV *)hv);
820 924
821fail: 925fail:
822 SvREFCNT_dec (hv); 926 SvREFCNT_dec_NN (hv);
823 DEC_DEC_DEPTH; 927 DEC_DEC_DEPTH;
824 return &PL_sv_undef; 928 return &PL_sv_undef;
825} 929}
826 930
827static SV * 931static SV *
828decode_str (dec_t *dec, int utf8) 932decode_str (dec_t *dec, int utf8)
829{ 933{
830 SV *sv = 0; 934 SV *sv = 0;
831 935
832 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 936 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
833 { 937 {
834 // indefinite length strings 938 // indefinite length strings
835 ++dec->cur; 939 ++dec->cur;
836 940
837 U8 major = *dec->cur & MAJOR_MISC; 941 U8 major = *dec->cur & MAJOR_MISC;
… …
905 sv = newRV_noinc (decode_sv (dec)); 1009 sv = newRV_noinc (decode_sv (dec));
906 break; 1010 break;
907 1011
908 case CBOR_TAG_STRINGREF_NAMESPACE: 1012 case CBOR_TAG_STRINGREF_NAMESPACE:
909 { 1013 {
1014 // do not use SAVETMPS/FREETMPS, as these will
1015 // erase mortalised caches, e.g. "shareable"
910 ENTER; SAVETMPS; 1016 ENTER;
911 1017
912 SAVESPTR (dec->stringref); 1018 SAVESPTR (dec->stringref);
913 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1019 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
914 1020
915 sv = decode_sv (dec); 1021 sv = decode_sv (dec);
916 1022
917 FREETMPS; LEAVE; 1023 LEAVE;
918 } 1024 }
919 break; 1025 break;
920 1026
921 case CBOR_TAG_STRINGREF: 1027 case CBOR_TAG_STRINGREF:
922 { 1028 {
923 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1029 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
924 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1030 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
925 1031
926 UV idx = decode_uint (dec); 1032 UV idx = decode_uint (dec);
927 1033
928 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1034 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
929 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); 1035 ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)");
930 1036
931 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1037 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
932 } 1038 }
933 break; 1039 break;
… …
961 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1067 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
962 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1068 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
963 1069
964 UV idx = decode_uint (dec); 1070 UV idx = decode_uint (dec);
965 1071
966 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1072 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
967 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1073 ERR ("corrupted CBOR data (sharedref index out of bounds)");
968 1074
969 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1075 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
970 1076
971 if (sv == &PL_sv_undef) 1077 if (sv == &PL_sv_undef)
… …
973 } 1079 }
974 break; 1080 break;
975 1081
976 case CBOR_TAG_PERL_OBJECT: 1082 case CBOR_TAG_PERL_OBJECT:
977 { 1083 {
1084 if (dec->cbor.flags & F_FORBID_OBJECTS)
1085 goto filter;
1086
978 sv = decode_sv (dec); 1087 sv = decode_sv (dec);
979 1088
980 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1089 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
981 ERR ("corrupted CBOR data (non-array perl object)"); 1090 ERR ("corrupted CBOR data (non-array perl object)");
982 1091
… …
992 if (!method) 1101 if (!method)
993 ERR ("cannot decode perl-object (package does not have a THAW method)"); 1102 ERR ("cannot decode perl-object (package does not have a THAW method)");
994 1103
995 dSP; 1104 dSP;
996 1105
997 ENTER; SAVETMPS; PUSHMARK (SP); 1106 ENTER; SAVETMPS;
1107 PUSHMARK (SP);
998 EXTEND (SP, len + 1); 1108 EXTEND (SP, len + 1);
999 // we re-bless the reference to get overload and other niceties right 1109 // we re-bless the reference to get overload and other niceties right
1000 PUSHs (*av_fetch (av, 0, 1)); 1110 PUSHs (*av_fetch (av, 0, 1));
1001 PUSHs (sv_cbor); 1111 PUSHs (sv_cbor);
1002 1112
… …
1010 SPAGAIN; 1120 SPAGAIN;
1011 1121
1012 if (SvTRUE (ERRSV)) 1122 if (SvTRUE (ERRSV))
1013 { 1123 {
1014 FREETMPS; LEAVE; 1124 FREETMPS; LEAVE;
1015 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1125 ERR_ERRSV;
1016 } 1126 }
1017 1127
1018 SvREFCNT_dec (sv); 1128 SvREFCNT_dec_NN (sv);
1019 sv = SvREFCNT_inc (POPs); 1129 sv = SvREFCNT_inc (POPs);
1020 1130
1021 PUTBACK; 1131 PUTBACK;
1022 1132
1023 FREETMPS; LEAVE; 1133 FREETMPS; LEAVE;
1024 } 1134 }
1025 break; 1135 break;
1026 1136
1027 default: 1137 default:
1138 filter:
1028 { 1139 {
1140 SV *tag_sv = newSVuv (tag);
1141
1029 sv = decode_sv (dec); 1142 sv = decode_sv (dec);
1030 1143
1031 dSP; 1144 dSP;
1032 ENTER; SAVETMPS; PUSHMARK (SP); 1145 ENTER; SAVETMPS;
1146 SAVESTACK_POS ();
1147 PUSHMARK (SP);
1033 EXTEND (SP, 2); 1148 EXTEND (SP, 2);
1034 PUSHs (newSVuv (tag)); 1149 PUSHs (tag_sv);
1035 PUSHs (sv); 1150 PUSHs (sv);
1036 1151
1037 PUTBACK; 1152 PUTBACK;
1038 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL); 1153 int count = call_sv (dec->cbor.filter ? dec->cbor.filter : default_filter, G_ARRAY | G_EVAL);
1039 SPAGAIN; 1154 SPAGAIN;
1040 1155
1041 if (SvTRUE (ERRSV)) 1156 if (SvTRUE (ERRSV))
1042 { 1157 {
1158 SvREFCNT_dec_NN (tag_sv);
1043 FREETMPS; LEAVE; 1159 FREETMPS; LEAVE;
1044 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1160 ERR_ERRSV;
1045 } 1161 }
1046 1162
1047 if (count) 1163 if (count)
1048 { 1164 {
1165 SvREFCNT_dec_NN (tag_sv);
1049 SvREFCNT_dec (sv); 1166 SvREFCNT_dec_NN (sv);
1050 sv = SvREFCNT_inc (POPs); 1167 sv = SvREFCNT_inc_NN (POPs);
1051 } 1168 }
1052 else 1169 else
1053 { 1170 {
1054 AV *av = newAV (); 1171 AV *av = newAV ();
1055 av_push (av, newSVuv (tag)); 1172 av_push (av, tag_sv);
1056 av_push (av, sv); 1173 av_push (av, sv);
1057 1174
1058 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1175 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1059 ? cbor_tagged_stash 1176 ? cbor_tagged_stash
1060 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1177 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1189 1306
1190 if (dec.err) 1307 if (dec.err)
1191 { 1308 {
1192 if (dec.shareable) 1309 if (dec.shareable)
1193 { 1310 {
1194 // need to break cyclic links, which whould all be in shareable 1311 // need to break cyclic links, which would all be in shareable
1195 int i; 1312 int i;
1196 SV **svp; 1313 SV **svp;
1197 1314
1198 for (i = av_len (dec.shareable) + 1; i--; ) 1315 for (i = av_len (dec.shareable) + 1; i--; )
1199 if ((svp = av_fetch (dec.shareable, i, 0))) 1316 if ((svp = av_fetch (dec.shareable, i, 0)))
1200 sv_setsv (*svp, &PL_sv_undef); 1317 sv_setsv (*svp, &PL_sv_undef);
1201 } 1318 }
1202 1319
1203 SvREFCNT_dec (sv); 1320 SvREFCNT_dec_NN (sv);
1321
1322 if (dec.err_sv)
1323 sv_2mortal (dec.err_sv);
1324
1204 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); 1325 croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur);
1205 } 1326 }
1206 1327
1207 sv = sv_2mortal (sv); 1328 sv = sv_2mortal (sv);
1208 1329
… …
1352 1473
1353 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1474 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1354 1475
1355 sv_cbor = newSVpv ("CBOR", 0); 1476 sv_cbor = newSVpv ("CBOR", 0);
1356 SvREADONLY_on (sv_cbor); 1477 SvREADONLY_on (sv_cbor);
1478
1479 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1357} 1480}
1358 1481
1359PROTOTYPES: DISABLE 1482PROTOTYPES: DISABLE
1360 1483
1361void CLONE (...) 1484void CLONE (...)
… …
1381 ALIAS: 1504 ALIAS:
1382 shrink = F_SHRINK 1505 shrink = F_SHRINK
1383 allow_unknown = F_ALLOW_UNKNOWN 1506 allow_unknown = F_ALLOW_UNKNOWN
1384 allow_sharing = F_ALLOW_SHARING 1507 allow_sharing = F_ALLOW_SHARING
1385 allow_cycles = F_ALLOW_CYCLES 1508 allow_cycles = F_ALLOW_CYCLES
1509 forbid_objects = F_FORBID_OBJECTS
1386 pack_strings = F_PACK_STRINGS 1510 pack_strings = F_PACK_STRINGS
1511 text_keys = F_TEXT_KEYS
1512 text_strings = F_TEXT_STRINGS
1387 validate_utf8 = F_VALIDATE_UTF8 1513 validate_utf8 = F_VALIDATE_UTF8
1388 PPCODE: 1514 PPCODE:
1389{ 1515{
1390 if (enable) 1516 if (enable)
1391 self->flags |= ix; 1517 self->flags |= ix;
… …
1399 ALIAS: 1525 ALIAS:
1400 get_shrink = F_SHRINK 1526 get_shrink = F_SHRINK
1401 get_allow_unknown = F_ALLOW_UNKNOWN 1527 get_allow_unknown = F_ALLOW_UNKNOWN
1402 get_allow_sharing = F_ALLOW_SHARING 1528 get_allow_sharing = F_ALLOW_SHARING
1403 get_allow_cycles = F_ALLOW_CYCLES 1529 get_allow_cycles = F_ALLOW_CYCLES
1530 get_forbid_objects = F_FORBID_OBJECTS
1404 get_pack_strings = F_PACK_STRINGS 1531 get_pack_strings = F_PACK_STRINGS
1532 get_text_keys = F_TEXT_KEYS
1533 get_text_strings = F_TEXT_STRINGS
1405 get_validate_utf8 = F_VALIDATE_UTF8 1534 get_validate_utf8 = F_VALIDATE_UTF8
1406 PPCODE: 1535 PPCODE:
1407 XPUSHs (boolSV (self->flags & ix)); 1536 XPUSHs (boolSV (self->flags & ix));
1408 1537
1409void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1538void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1539 cbor_init (&cbor); 1668 cbor_init (&cbor);
1540 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1669 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1541 XPUSHs (cborstr); 1670 XPUSHs (cborstr);
1542} 1671}
1543 1672
1673#ifdef __AFL_COMPILER
1674
1675void
1676afl_init ()
1677 CODE:
1678 __AFL_INIT ();
1679
1680int
1681afl_loop (unsigned int count = 10000)
1682 CODE:
1683 RETVAL = __AFL_LOOP (count);
1684 OUTPUT:
1685 RETVAL
1686
1687#endif
1688

Diff Legend

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