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 1.45 by root, Sun Dec 14 05:48:39 2014 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 }
335 else 368 else
336 for (i = 0; i <= len; ++i) 369 for (i = 0; i <= len; ++i)
337 encode_sv (enc, AvARRAY (av)[i]); 370 {
371 SV *sv = AvARRAY (av)[i];
372 encode_sv (enc, sv ? sv : &PL_sv_undef);
373 }
338 374
339 --enc->depth; 375 --enc->depth;
340} 376}
341 377
342static void 378static void
… …
350 ++enc->depth; 386 ++enc->depth;
351 387
352 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
353 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
354 390
355 if (mg) 391 if (ecb_expect_false (mg))
356 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
357 else 393 else
358 encode_uint (enc, MAJOR_MAP, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
359 395
360 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
361 { 397 {
362 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
363 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
364 else 400 else
365 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));
366 402
367 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));
368 } 404 }
369 405
370 if (mg) 406 if (ecb_expect_false (mg))
371 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
372 408
373 --enc->depth; 409 --enc->depth;
374} 410}
375 411
… …
418 } 454 }
419 455
420 if (ecb_expect_false (SvREFCNT (sv) > 1) 456 if (ecb_expect_false (SvREFCNT (sv) > 1)
421 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
422 { 458 {
423 if (!enc->shareable) 459 if (ecb_expect_false (!enc->shareable))
424 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
425 461
426 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
427 463
428 if (SvOK (*svp)) 464 if (SvOK (*svp))
… …
442 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
443 { 479 {
444 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
445 GV *method; 481 GV *method;
446 482
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))));
447 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
448 { 487 {
449 dSP; 488 dSP;
450 489
451 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
452 // 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
453 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
454 494
455 PUTBACK; 495 PUTBACK;
456 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
… …
469 } 509 }
470 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
471 { 511 {
472 dSP; 512 dSP;
473 513
474 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 PUSHMARK (SP);
475 EXTEND (SP, 2); 516 EXTEND (SP, 2);
476 // 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
477 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 518 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
478 PUSHs (sv_cbor); 519 PUSHs (sv_cbor);
479 520
… …
485 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 526 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
486 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));
487 528
488 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 529 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
489 encode_uint (enc, MAJOR_ARRAY, count + 1); 530 encode_uint (enc, MAJOR_ARRAY, count + 1);
490 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 531 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
491 532
492 while (count) 533 {
534 int i;
535
536 for (i = 0; i < count; ++i)
493 encode_sv (enc, SP[1 - count--]); 537 encode_sv (enc, SP[i + 1 - count]);
538
539 SP -= count;
540 }
494 541
495 PUTBACK; 542 PUTBACK;
496 543
497 FREETMPS; LEAVE; 544 FREETMPS; LEAVE;
498 } 545 }
… …
521 if (ecb_expect_false (nv == (NV)(U32)nv)) 568 if (ecb_expect_false (nv == (NV)(U32)nv))
522 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 569 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
523 //TODO: maybe I32? 570 //TODO: maybe I32?
524 else if (ecb_expect_false (nv == (float)nv)) 571 else if (ecb_expect_false (nv == (float)nv))
525 { 572 {
573 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
574
526 uint32_t fp = ecb_float_to_binary32 (nv); 575 uint32_t fp = ecb_float_to_binary32 (nv);
527
528 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
529 576
530 if (!ecb_big_endian ()) 577 if (!ecb_big_endian ())
531 fp = ecb_bswap32 (fp); 578 fp = ecb_bswap32 (fp);
532 579
533 memcpy (enc->cur, &fp, 4); 580 memcpy (enc->cur, &fp, 4);
534 enc->cur += 4; 581 enc->cur += 4;
535 } 582 }
536 else 583 else
537 { 584 {
585 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
586
538 uint64_t fp = ecb_double_to_binary64 (nv); 587 uint64_t fp = ecb_double_to_binary64 (nv);
539
540 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
541 588
542 if (!ecb_big_endian ()) 589 if (!ecb_big_endian ())
543 fp = ecb_bswap64 (fp); 590 fp = ecb_bswap64 (fp);
544 591
545 memcpy (enc->cur, &fp, 8); 592 memcpy (enc->cur, &fp, 8);
… …
554 601
555 if (SvPOKp (sv)) 602 if (SvPOKp (sv))
556 { 603 {
557 STRLEN len; 604 STRLEN len;
558 char *str = SvPV (sv, len); 605 char *str = SvPV (sv, len);
559 encode_strref (enc, SvUTF8 (sv), str, len); 606 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
560 } 607 }
561 else if (SvNOKp (sv)) 608 else if (SvNOKp (sv))
562 encode_nv (enc, sv); 609 encode_nv (enc, sv);
563 else if (SvIOKp (sv)) 610 else if (SvIOKp (sv))
564 { 611 {
… …
581} 628}
582 629
583static SV * 630static SV *
584encode_cbor (SV *scalar, CBOR *cbor) 631encode_cbor (SV *scalar, CBOR *cbor)
585{ 632{
586 enc_t enc = { }; 633 enc_t enc = { 0 };
587 634
588 enc.cbor = *cbor; 635 enc.cbor = *cbor;
589 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 636 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
590 enc.cur = SvPVX (enc.sv); 637 enc.cur = SvPVX (enc.sv);
591 enc.end = SvEND (enc.sv); 638 enc.end = SvEND (enc.sv);
592 639
593 SvPOK_only (enc.sv); 640 SvPOK_only (enc.sv);
594 641
595 if (cbor->flags & F_PACK_STRINGS) 642 if (cbor->flags & F_PACK_STRINGS)
596 { 643 {
… …
623 U32 depth; // recursion depth 670 U32 depth; // recursion depth
624 U32 maxdepth; // recursion depth limit 671 U32 maxdepth; // recursion depth limit
625 AV *shareable; 672 AV *shareable;
626 AV *stringref; 673 AV *stringref;
627 SV *decode_tagged; 674 SV *decode_tagged;
675 SV *err_sv; // optional sv for error, needs to be freed
628} dec_t; 676} dec_t;
629 677
630#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);
631 685
632#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;
633 689
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
634#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)
635#define DEC_DEC_DEPTH --dec->depth 715#define DEC_DEC_DEPTH --dec->depth
636 716
637static UV 717static UV
638decode_uint (dec_t *dec) 718decode_uint (dec_t *dec)
639{ 719{
… …
705 785
706 for (;;) 786 for (;;)
707 { 787 {
708 WANT (1); 788 WANT (1);
709 789
710 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 790 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
711 { 791 {
712 ++dec->cur; 792 ++dec->cur;
713 break; 793 break;
714 } 794 }
715 795
716 av_push (av, decode_sv (dec)); 796 av_push (av, decode_sv (dec));
717 } 797 }
718 } 798 }
719 else 799 else
720 { 800 {
721 int i, len = decode_uint (dec); 801 UV i, len = decode_uint (dec);
722 802
723 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
724 av_fill (av, len - 1); 804 av_fill (av, len - 1);
725 805
726 for (i = 0; i < len; ++i) 806 for (i = 0; i < len; ++i)
… …
729 809
730 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
731 return newRV_noinc ((SV *)av); 811 return newRV_noinc ((SV *)av);
732 812
733fail: 813fail:
734 SvREFCNT_dec (av); 814 SvREFCNT_dec_NN (av);
735 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
736 return &PL_sv_undef; 816 return &PL_sv_undef;
737} 817}
738 818
739static void 819static void
… …
743 // byte or utf-8 strings as keys, but only when !stringref 823 // byte or utf-8 strings as keys, but only when !stringref
744 824
745 if (ecb_expect_true (!dec->stringref)) 825 if (ecb_expect_true (!dec->stringref))
746 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 826 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
747 { 827 {
748 I32 len = decode_uint (dec); 828 STRLEN len = decode_uint (dec);
749 char *key = (char *)dec->cur; 829 char *key = (char *)dec->cur;
750 830
831 WANT (len);
751 dec->cur += len; 832 dec->cur += len;
752 833
753 hv_store (hv, key, len, decode_sv (dec), 0); 834 hv_store (hv, key, len, decode_sv (dec), 0);
754 835
755 return; 836 return;
756 } 837 }
757 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))
758 { 839 {
759 I32 len = decode_uint (dec); 840 STRLEN len = decode_uint (dec);
760 char *key = (char *)dec->cur; 841 char *key = (char *)dec->cur;
761 842
843 WANT (len);
762 dec->cur += len; 844 dec->cur += len;
763 845
764 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 846 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
765 if (!is_utf8_string (key, len)) 847 if (!is_utf8_string (key, len))
766 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 848 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
771 } 853 }
772 854
773 SV *k = decode_sv (dec); 855 SV *k = decode_sv (dec);
774 SV *v = decode_sv (dec); 856 SV *v = decode_sv (dec);
775 857
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
776 hv_store_ent (hv, k, v, 0); 886 hv_store_ent (hv, k, v, 0);
777 SvREFCNT_dec (k); 887 SvREFCNT_dec_NN (k);
778 888
779fail: 889fail:
780 ; 890 ;
781} 891}
782 892
… …
793 903
794 for (;;) 904 for (;;)
795 { 905 {
796 WANT (1); 906 WANT (1);
797 907
798 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 908 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
799 { 909 {
800 ++dec->cur; 910 ++dec->cur;
801 break; 911 break;
802 } 912 }
803 913
804 decode_he (dec, hv); 914 decode_he (dec, hv);
805 } 915 }
806 } 916 }
807 else 917 else
808 { 918 {
809 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
810 922
811 while (pairs--) 923 while (pairs--)
812 decode_he (dec, hv); 924 decode_he (dec, hv);
813 } 925 }
814 926
815 DEC_DEC_DEPTH; 927 DEC_DEC_DEPTH;
816 return newRV_noinc ((SV *)hv); 928 return newRV_noinc ((SV *)hv);
817 929
818fail: 930fail:
819 SvREFCNT_dec (hv); 931 SvREFCNT_dec_NN (hv);
820 DEC_DEC_DEPTH; 932 DEC_DEC_DEPTH;
821 return &PL_sv_undef; 933 return &PL_sv_undef;
822} 934}
823 935
824static SV * 936static SV *
825decode_str (dec_t *dec, int utf8) 937decode_str (dec_t *dec, int utf8)
826{ 938{
827 SV *sv = 0; 939 SV *sv = 0;
828 940
829 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 941 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
830 { 942 {
831 // indefinite length strings 943 // indefinite length strings
832 ++dec->cur; 944 ++dec->cur;
833 945
834 U8 major = *dec->cur & MAJOR_MISC; 946 U8 major = *dec->cur & MAJOR_MISC;
… …
902 sv = newRV_noinc (decode_sv (dec)); 1014 sv = newRV_noinc (decode_sv (dec));
903 break; 1015 break;
904 1016
905 case CBOR_TAG_STRINGREF_NAMESPACE: 1017 case CBOR_TAG_STRINGREF_NAMESPACE:
906 { 1018 {
1019 // do not use SAVETMPS/FREETMPS, as these will
1020 // erase mortalised caches, e.g. "shareable"
907 ENTER; SAVETMPS; 1021 ENTER;
908 1022
909 SAVESPTR (dec->stringref); 1023 SAVESPTR (dec->stringref);
910 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1024 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
911 1025
912 sv = decode_sv (dec); 1026 sv = decode_sv (dec);
913 1027
914 FREETMPS; LEAVE; 1028 LEAVE;
915 } 1029 }
916 break; 1030 break;
917 1031
918 case CBOR_TAG_STRINGREF: 1032 case CBOR_TAG_STRINGREF:
919 { 1033 {
920 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1034 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
921 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1035 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
922 1036
923 UV idx = decode_uint (dec); 1037 UV idx = decode_uint (dec);
924 1038
925 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1039 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
926 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)");
927 1041
928 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1042 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
929 } 1043 }
930 break; 1044 break;
… …
958 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1072 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
959 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1073 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
960 1074
961 UV idx = decode_uint (dec); 1075 UV idx = decode_uint (dec);
962 1076
963 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1077 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
964 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1078 ERR ("corrupted CBOR data (sharedref index out of bounds)");
965 1079
966 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1080 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
967 1081
968 if (sv == &PL_sv_undef) 1082 if (sv == &PL_sv_undef)
… …
970 } 1084 }
971 break; 1085 break;
972 1086
973 case CBOR_TAG_PERL_OBJECT: 1087 case CBOR_TAG_PERL_OBJECT:
974 { 1088 {
1089 if (dec->cbor.flags & F_FORBID_OBJECTS)
1090 goto filter;
1091
975 sv = decode_sv (dec); 1092 sv = decode_sv (dec);
976 1093
977 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1094 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
978 ERR ("corrupted CBOR data (non-array perl object)"); 1095 ERR ("corrupted CBOR data (non-array perl object)");
979 1096
… …
989 if (!method) 1106 if (!method)
990 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)");
991 1108
992 dSP; 1109 dSP;
993 1110
994 ENTER; SAVETMPS; PUSHMARK (SP); 1111 ENTER; SAVETMPS;
1112 PUSHMARK (SP);
995 EXTEND (SP, len + 1); 1113 EXTEND (SP, len + 1);
996 // 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
997 PUSHs (*av_fetch (av, 0, 1)); 1115 PUSHs (*av_fetch (av, 0, 1));
998 PUSHs (sv_cbor); 1116 PUSHs (sv_cbor);
999 1117
… …
1007 SPAGAIN; 1125 SPAGAIN;
1008 1126
1009 if (SvTRUE (ERRSV)) 1127 if (SvTRUE (ERRSV))
1010 { 1128 {
1011 FREETMPS; LEAVE; 1129 FREETMPS; LEAVE;
1012 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1130 ERR_ERRSV;
1013 } 1131 }
1014 1132
1015 SvREFCNT_dec (sv); 1133 SvREFCNT_dec_NN (sv);
1016 sv = SvREFCNT_inc (POPs); 1134 sv = SvREFCNT_inc (POPs);
1017 1135
1018 PUTBACK; 1136 PUTBACK;
1019 1137
1020 FREETMPS; LEAVE; 1138 FREETMPS; LEAVE;
1021 } 1139 }
1022 break; 1140 break;
1023 1141
1024 default: 1142 default:
1143 filter:
1025 { 1144 {
1145 SV *tag_sv = newSVuv (tag);
1146
1026 sv = decode_sv (dec); 1147 sv = decode_sv (dec);
1027 1148
1028 dSP; 1149 dSP;
1029 ENTER; SAVETMPS; PUSHMARK (SP); 1150 ENTER; SAVETMPS;
1151 PUSHMARK (SP);
1030 EXTEND (SP, 2); 1152 EXTEND (SP, 2);
1031 PUSHs (newSVuv (tag)); 1153 PUSHs (tag_sv);
1032 PUSHs (sv); 1154 PUSHs (sv);
1033 1155
1034 PUTBACK; 1156 PUTBACK;
1035 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);
1036 SPAGAIN; 1158 SPAGAIN;
1037 1159
1038 if (SvTRUE (ERRSV)) 1160 if (SvTRUE (ERRSV))
1039 { 1161 {
1162 SvREFCNT_dec_NN (tag_sv);
1040 FREETMPS; LEAVE; 1163 FREETMPS; LEAVE;
1041 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1164 ERR_ERRSV;
1042 } 1165 }
1043 1166
1044 if (count) 1167 if (count)
1045 { 1168 {
1169 SvREFCNT_dec_NN (tag_sv);
1046 SvREFCNT_dec (sv); 1170 SvREFCNT_dec_NN (sv);
1047 sv = SvREFCNT_inc (POPs); 1171 sv = SvREFCNT_inc_NN (TOPs);
1172 SP -= count;
1048 } 1173 }
1049 else 1174 else
1050 { 1175 {
1051 AV *av = newAV (); 1176 AV *av = newAV ();
1052 av_push (av, newSVuv (tag)); 1177 av_push (av, tag_sv);
1053 av_push (av, sv); 1178 av_push (av, sv);
1054 1179
1055 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1180 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1056 ? cbor_tagged_stash 1181 ? cbor_tagged_stash
1057 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1182 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1160} 1285}
1161 1286
1162static SV * 1287static SV *
1163decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1288decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1164{ 1289{
1165 dec_t dec = { }; 1290 dec_t dec = { 0 };
1166 SV *sv; 1291 SV *sv;
1167 STRLEN len; 1292 STRLEN len;
1168 char *data = SvPVbyte (string, len); 1293 char *data = SvPVbyte (string, len);
1169 1294
1170 if (len > cbor->max_size && cbor->max_size) 1295 if (len > cbor->max_size && cbor->max_size)
… …
1186 1311
1187 if (dec.err) 1312 if (dec.err)
1188 { 1313 {
1189 if (dec.shareable) 1314 if (dec.shareable)
1190 { 1315 {
1191 // need to break cyclic links, which whould all be in shareable 1316 // need to break cyclic links, which would all be in shareable
1192 int i; 1317 int i;
1193 SV **svp; 1318 SV **svp;
1194 1319
1195 for (i = av_len (dec.shareable) + 1; i--; ) 1320 for (i = av_len (dec.shareable) + 1; i--; )
1196 if ((svp = av_fetch (dec.shareable, i, 0))) 1321 if ((svp = av_fetch (dec.shareable, i, 0)))
1197 sv_setsv (*svp, &PL_sv_undef); 1322 sv_setsv (*svp, &PL_sv_undef);
1198 } 1323 }
1199 1324
1200 SvREFCNT_dec (sv); 1325 SvREFCNT_dec_NN (sv);
1326
1327 if (dec.err_sv)
1328 sv_2mortal (dec.err_sv);
1329
1201 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);
1202 } 1331 }
1203 1332
1204 sv = sv_2mortal (sv); 1333 sv = sv_2mortal (sv);
1205 1334
… …
1261 1390
1262 int major = *p >> MAJOR_SHIFT; 1391 int major = *p >> MAJOR_SHIFT;
1263 1392
1264 switch (major) 1393 switch (major)
1265 { 1394 {
1395 case MAJOR_TAG >> MAJOR_SHIFT:
1396 ++count; // tags merely prefix another value
1397 break;
1398
1266 case MAJOR_BYTES >> MAJOR_SHIFT: 1399 case MAJOR_BYTES >> MAJOR_SHIFT:
1267 case MAJOR_TEXT >> MAJOR_SHIFT: 1400 case MAJOR_TEXT >> MAJOR_SHIFT:
1268 case MAJOR_ARRAY >> MAJOR_SHIFT: 1401 case MAJOR_ARRAY >> MAJOR_SHIFT:
1269 case MAJOR_MAP >> MAJOR_SHIFT: 1402 case MAJOR_MAP >> MAJOR_SHIFT:
1270 { 1403 {
… …
1345 1478
1346 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1479 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1347 1480
1348 sv_cbor = newSVpv ("CBOR", 0); 1481 sv_cbor = newSVpv ("CBOR", 0);
1349 SvREADONLY_on (sv_cbor); 1482 SvREADONLY_on (sv_cbor);
1483
1484 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1350} 1485}
1351 1486
1352PROTOTYPES: DISABLE 1487PROTOTYPES: DISABLE
1353 1488
1354void CLONE (...) 1489void CLONE (...)
… …
1374 ALIAS: 1509 ALIAS:
1375 shrink = F_SHRINK 1510 shrink = F_SHRINK
1376 allow_unknown = F_ALLOW_UNKNOWN 1511 allow_unknown = F_ALLOW_UNKNOWN
1377 allow_sharing = F_ALLOW_SHARING 1512 allow_sharing = F_ALLOW_SHARING
1378 allow_cycles = F_ALLOW_CYCLES 1513 allow_cycles = F_ALLOW_CYCLES
1514 forbid_objects = F_FORBID_OBJECTS
1379 pack_strings = F_PACK_STRINGS 1515 pack_strings = F_PACK_STRINGS
1516 text_keys = F_TEXT_KEYS
1517 text_strings = F_TEXT_STRINGS
1380 validate_utf8 = F_VALIDATE_UTF8 1518 validate_utf8 = F_VALIDATE_UTF8
1381 PPCODE: 1519 PPCODE:
1382{ 1520{
1383 if (enable) 1521 if (enable)
1384 self->flags |= ix; 1522 self->flags |= ix;
… …
1392 ALIAS: 1530 ALIAS:
1393 get_shrink = F_SHRINK 1531 get_shrink = F_SHRINK
1394 get_allow_unknown = F_ALLOW_UNKNOWN 1532 get_allow_unknown = F_ALLOW_UNKNOWN
1395 get_allow_sharing = F_ALLOW_SHARING 1533 get_allow_sharing = F_ALLOW_SHARING
1396 get_allow_cycles = F_ALLOW_CYCLES 1534 get_allow_cycles = F_ALLOW_CYCLES
1535 get_forbid_objects = F_FORBID_OBJECTS
1397 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
1398 get_validate_utf8 = F_VALIDATE_UTF8 1539 get_validate_utf8 = F_VALIDATE_UTF8
1399 PPCODE: 1540 PPCODE:
1400 XPUSHs (boolSV (self->flags & ix)); 1541 XPUSHs (boolSV (self->flags & ix));
1401 1542
1402void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1543void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1532 cbor_init (&cbor); 1673 cbor_init (&cbor);
1533 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1674 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1534 XPUSHs (cborstr); 1675 XPUSHs (cborstr);
1535} 1676}
1536 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