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 1.41 by root, Tue Dec 10 14:01:52 2013 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
13#define ECB_NO_THREADS 1
12#include "ecb.h" 14#include "ecb.h"
13 15
14// compatibility with perl <5.18 16// compatibility with perl <5.18
15#ifndef HvNAMELEN_get 17#ifndef HvNAMELEN_get
16# define HvNAMELEN_get(hv) strlen (HvNAME (hv)) 18# define HvNAMELEN_get(hv) strlen (HvNAME (hv))
… …
19# define HvNAMELEN(hv) HvNAMELEN_get (hv) 21# define HvNAMELEN(hv) HvNAMELEN_get (hv)
20#endif 22#endif
21#ifndef HvNAMEUTF8 23#ifndef HvNAMEUTF8
22# define HvNAMEUTF8(hv) 0 24# define HvNAMEUTF8(hv) 0
23#endif 25#endif
26#ifndef SvREFCNT_inc_NN
27# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv)
28#endif
24#ifndef SvREFCNT_dec_NN 29#ifndef SvREFCNT_dec_NN
25# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) 30# define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv)
26#endif 31#endif
27 32
28// known major and minor types 33// known major and minor types
… …
98 103
99#define F_SHRINK 0x00000001UL 104#define F_SHRINK 0x00000001UL
100#define F_ALLOW_UNKNOWN 0x00000002UL 105#define F_ALLOW_UNKNOWN 0x00000002UL
101#define F_ALLOW_SHARING 0x00000004UL 106#define F_ALLOW_SHARING 0x00000004UL
102#define F_ALLOW_CYCLES 0x00000008UL 107#define F_ALLOW_CYCLES 0x00000008UL
108#define F_FORBID_OBJECTS 0x00000010UL
103#define F_PACK_STRINGS 0x00000010UL 109#define F_PACK_STRINGS 0x00000020UL
110#define F_TEXT_KEYS 0x00000040UL
111#define F_TEXT_STRINGS 0x00000080UL
104#define F_VALIDATE_UTF8 0x00000020UL 112#define F_VALIDATE_UTF8 0x00000100UL
105 113
106#define INIT_SIZE 32 // initial scalar size to be allocated 114#define INIT_SIZE 32 // initial scalar size to be allocated
107 115
108#define SB do { 116#define SB do {
109#define SE } while (0) 117#define SE } while (0)
… …
182 190
183// minimum length of a string to be registered for stringref 191// minimum length of a string to be registered for stringref
184ecb_inline int 192ecb_inline int
185minimum_string_length (UV idx) 193minimum_string_length (UV idx)
186{ 194{
187 return idx > 23 195 return idx <= 23 ? 3
188 ? idx > 0xffU 196 : idx <= 0xffU ? 4
189 ? idx > 0xffffU 197 : idx <= 0xffffU ? 5
190 ? idx > 0xffffffffU 198 : idx <= 0xffffffffU ? 7
191 ? 11 199 : 11;
192 : 7
193 : 5
194 : 4
195 : 3;
196} 200}
197 201
198///////////////////////////////////////////////////////////////////////////// 202/////////////////////////////////////////////////////////////////////////////
199// encoder 203// encoder
200 204
… …
213} enc_t; 217} enc_t;
214 218
215ecb_inline void 219ecb_inline void
216need (enc_t *enc, STRLEN len) 220need (enc_t *enc, STRLEN len)
217{ 221{
218 if (ecb_expect_false (enc->cur + len >= enc->end)) 222 if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len))
219 { 223 {
220 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); 224 STRLEN cur = enc->cur - (char *)SvPVX (enc->sv);
221 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); 225 SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1);
222 enc->cur = SvPVX (enc->sv) + cur; 226 enc->cur = SvPVX (enc->sv) + cur;
223 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1; 227 enc->end = SvPVX (enc->sv) + SvLEN (enc->sv) - 1;
… …
275encode_tag (enc_t *enc, UV tag) 279encode_tag (enc_t *enc, UV tag)
276{ 280{
277 encode_uint (enc, MAJOR_TAG, tag); 281 encode_uint (enc, MAJOR_TAG, tag);
278} 282}
279 283
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
280ecb_inline void 307ecb_inline void
281encode_str (enc_t *enc, int utf8, char *str, STRLEN len) 308encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
282{ 309{
310 if (ecb_expect_false (upgrade_utf8))
311 if (!utf8)
312 {
313 encode_str_utf8 (enc, utf8, str, len);
314 return;
315 }
316
283 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len); 317 encode_uint (enc, utf8 ? MAJOR_TEXT : MAJOR_BYTES, len);
284 need (enc, len); 318 need (enc, len);
285 memcpy (enc->cur, str, len); 319 memcpy (enc->cur, str, len);
286 enc->cur += len; 320 enc->cur += len;
287} 321}
288 322
289static void 323ecb_inline void
290encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) 324encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len)
291{ 325{
292 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) 326 if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS))
293 { 327 {
294 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1); 328 SV **svp = hv_fetch (enc->stringref[!!utf8], str, len, 1);
295 329
… …
306 sv_setuv (*svp, enc->stringref_idx); 340 sv_setuv (*svp, enc->stringref_idx);
307 ++enc->stringref_idx; 341 ++enc->stringref_idx;
308 } 342 }
309 } 343 }
310 344
311 encode_str (enc, utf8, str, len); 345 encode_str (enc, upgrade_utf8, utf8, str, len);
312} 346}
313 347
314static void encode_sv (enc_t *enc, SV *sv); 348static void encode_sv (enc_t *enc, SV *sv);
315 349
316static void 350static void
… …
323 357
324 ++enc->depth; 358 ++enc->depth;
325 359
326 encode_uint (enc, MAJOR_ARRAY, len + 1); 360 encode_uint (enc, MAJOR_ARRAY, len + 1);
327 361
362 if (ecb_expect_false (SvMAGICAL (av)))
328 for (i = 0; i <= len; ++i) 363 for (i = 0; i <= len; ++i)
329 { 364 {
330 SV **svp = av_fetch (av, i, 0); 365 SV **svp = av_fetch (av, i, 0);
331 encode_sv (enc, svp ? *svp : &PL_sv_undef); 366 encode_sv (enc, svp ? *svp : &PL_sv_undef);
332 } 367 }
368 else
369 for (i = 0; i <= len; ++i)
370 {
371 SV *sv = AvARRAY (av)[i];
372 encode_sv (enc, sv ? sv : &PL_sv_undef);
373 }
333 374
334 --enc->depth; 375 --enc->depth;
335} 376}
336 377
337static void 378static void
… …
345 ++enc->depth; 386 ++enc->depth;
346 387
347 int pairs = hv_iterinit (hv); 388 int pairs = hv_iterinit (hv);
348 int mg = SvMAGICAL (hv); 389 int mg = SvMAGICAL (hv);
349 390
350 if (mg) 391 if (ecb_expect_false (mg))
351 encode_ch (enc, MAJOR_MAP | MINOR_INDEF); 392 encode_ch (enc, MAJOR_MAP | MINOR_INDEF);
352 else 393 else
353 encode_uint (enc, MAJOR_MAP, pairs); 394 encode_uint (enc, MAJOR_MAP, pairs);
354 395
355 while ((he = hv_iternext (hv))) 396 while ((he = hv_iternext (hv)))
356 { 397 {
357 if (HeKLEN (he) == HEf_SVKEY) 398 if (HeKLEN (he) == HEf_SVKEY)
358 encode_sv (enc, HeSVKEY (he)); 399 encode_sv (enc, HeSVKEY (he));
359 else 400 else
360 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));
361 402
362 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));
363 } 404 }
364 405
365 if (mg) 406 if (ecb_expect_false (mg))
366 encode_ch (enc, MAJOR_MISC | MINOR_INDEF); 407 encode_ch (enc, MAJOR_MISC | MINOR_INDEF);
367 408
368 --enc->depth; 409 --enc->depth;
369} 410}
370 411
… …
413 } 454 }
414 455
415 if (ecb_expect_false (SvREFCNT (sv) > 1) 456 if (ecb_expect_false (SvREFCNT (sv) > 1)
416 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) 457 && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING))
417 { 458 {
418 if (!enc->shareable) 459 if (ecb_expect_false (!enc->shareable))
419 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); 460 enc->shareable = (HV *)sv_2mortal ((SV *)newHV ());
420 461
421 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); 462 SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1);
422 463
423 if (SvOK (*svp)) 464 if (SvOK (*svp))
… …
437 if (ecb_expect_false (SvOBJECT (sv))) 478 if (ecb_expect_false (SvOBJECT (sv)))
438 { 479 {
439 HV *stash = SvSTASH (sv); 480 HV *stash = SvSTASH (sv);
440 GV *method; 481 GV *method;
441 482
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))));
442 if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) 486 else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0)))
443 { 487 {
444 dSP; 488 dSP;
445 489
446 ENTER; SAVETMPS; PUSHMARK (SP); 490 ENTER; SAVETMPS;
491 PUSHMARK (SP);
447 // 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
448 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 493 XPUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
449 494
450 PUTBACK; 495 PUTBACK;
451 // G_SCALAR ensures that return value is 1 496 // G_SCALAR ensures that return value is 1
… …
464 } 509 }
465 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0) 510 else if ((method = gv_fetchmethod_autoload (stash, "FREEZE", 0)) != 0)
466 { 511 {
467 dSP; 512 dSP;
468 513
469 ENTER; SAVETMPS; PUSHMARK (SP); 514 ENTER; SAVETMPS;
515 PUSHMARK (SP);
470 EXTEND (SP, 2); 516 EXTEND (SP, 2);
471 // 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
472 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash)); 518 PUSHs (sv_bless (sv_2mortal (newRV_inc (sv)), stash));
473 PUSHs (sv_cbor); 519 PUSHs (sv_cbor);
474 520
… …
480 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv) 526 if (count == 1 && SvROK (TOPs) && SvRV (TOPs) == sv)
481 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));
482 528
483 encode_tag (enc, CBOR_TAG_PERL_OBJECT); 529 encode_tag (enc, CBOR_TAG_PERL_OBJECT);
484 encode_uint (enc, MAJOR_ARRAY, count + 1); 530 encode_uint (enc, MAJOR_ARRAY, count + 1);
485 encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); 531 encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash));
486 532
487 while (count) 533 {
534 int i;
535
536 for (i = 0; i < count; ++i)
488 encode_sv (enc, SP[1 - count--]); 537 encode_sv (enc, SP[i + 1 - count]);
538
539 SP -= count;
540 }
489 541
490 PUTBACK; 542 PUTBACK;
491 543
492 FREETMPS; LEAVE; 544 FREETMPS; LEAVE;
493 } 545 }
… …
516 if (ecb_expect_false (nv == (NV)(U32)nv)) 568 if (ecb_expect_false (nv == (NV)(U32)nv))
517 encode_uint (enc, MAJOR_POS_INT, (U32)nv); 569 encode_uint (enc, MAJOR_POS_INT, (U32)nv);
518 //TODO: maybe I32? 570 //TODO: maybe I32?
519 else if (ecb_expect_false (nv == (float)nv)) 571 else if (ecb_expect_false (nv == (float)nv))
520 { 572 {
573 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
574
521 uint32_t fp = ecb_float_to_binary32 (nv); 575 uint32_t fp = ecb_float_to_binary32 (nv);
522
523 *enc->cur++ = MAJOR_MISC | MISC_FLOAT32;
524 576
525 if (!ecb_big_endian ()) 577 if (!ecb_big_endian ())
526 fp = ecb_bswap32 (fp); 578 fp = ecb_bswap32 (fp);
527 579
528 memcpy (enc->cur, &fp, 4); 580 memcpy (enc->cur, &fp, 4);
529 enc->cur += 4; 581 enc->cur += 4;
530 } 582 }
531 else 583 else
532 { 584 {
585 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
586
533 uint64_t fp = ecb_double_to_binary64 (nv); 587 uint64_t fp = ecb_double_to_binary64 (nv);
534
535 *enc->cur++ = MAJOR_MISC | MISC_FLOAT64;
536 588
537 if (!ecb_big_endian ()) 589 if (!ecb_big_endian ())
538 fp = ecb_bswap64 (fp); 590 fp = ecb_bswap64 (fp);
539 591
540 memcpy (enc->cur, &fp, 8); 592 memcpy (enc->cur, &fp, 8);
… …
549 601
550 if (SvPOKp (sv)) 602 if (SvPOKp (sv))
551 { 603 {
552 STRLEN len; 604 STRLEN len;
553 char *str = SvPV (sv, len); 605 char *str = SvPV (sv, len);
554 encode_strref (enc, SvUTF8 (sv), str, len); 606 encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len);
555 } 607 }
556 else if (SvNOKp (sv)) 608 else if (SvNOKp (sv))
557 encode_nv (enc, sv); 609 encode_nv (enc, sv);
558 else if (SvIOKp (sv)) 610 else if (SvIOKp (sv))
559 { 611 {
… …
576} 628}
577 629
578static SV * 630static SV *
579encode_cbor (SV *scalar, CBOR *cbor) 631encode_cbor (SV *scalar, CBOR *cbor)
580{ 632{
581 enc_t enc = { }; 633 enc_t enc = { 0 };
582 634
583 enc.cbor = *cbor; 635 enc.cbor = *cbor;
584 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); 636 enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE));
585 enc.cur = SvPVX (enc.sv); 637 enc.cur = SvPVX (enc.sv);
586 enc.end = SvEND (enc.sv); 638 enc.end = SvEND (enc.sv);
587 639
588 SvPOK_only (enc.sv); 640 SvPOK_only (enc.sv);
589 641
590 if (cbor->flags & F_PACK_STRINGS) 642 if (cbor->flags & F_PACK_STRINGS)
591 { 643 {
… …
618 U32 depth; // recursion depth 670 U32 depth; // recursion depth
619 U32 maxdepth; // recursion depth limit 671 U32 maxdepth; // recursion depth limit
620 AV *shareable; 672 AV *shareable;
621 AV *stringref; 673 AV *stringref;
622 SV *decode_tagged; 674 SV *decode_tagged;
675 SV *err_sv; // optional sv for error, needs to be freed
623} dec_t; 676} dec_t;
624 677
625#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);
626 685
627#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;
628 689
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
629#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)
630#define DEC_DEC_DEPTH --dec->depth 715#define DEC_DEC_DEPTH --dec->depth
631 716
632static UV 717static UV
633decode_uint (dec_t *dec) 718decode_uint (dec_t *dec)
634{ 719{
… …
700 785
701 for (;;) 786 for (;;)
702 { 787 {
703 WANT (1); 788 WANT (1);
704 789
705 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 790 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
706 { 791 {
707 ++dec->cur; 792 ++dec->cur;
708 break; 793 break;
709 } 794 }
710 795
711 av_push (av, decode_sv (dec)); 796 av_push (av, decode_sv (dec));
712 } 797 }
713 } 798 }
714 else 799 else
715 { 800 {
716 int i, len = decode_uint (dec); 801 UV i, len = decode_uint (dec);
717 802
718 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
719 av_fill (av, len - 1); 804 av_fill (av, len - 1);
720 805
721 for (i = 0; i < len; ++i) 806 for (i = 0; i < len; ++i)
… …
724 809
725 DEC_DEC_DEPTH; 810 DEC_DEC_DEPTH;
726 return newRV_noinc ((SV *)av); 811 return newRV_noinc ((SV *)av);
727 812
728fail: 813fail:
729 SvREFCNT_dec (av); 814 SvREFCNT_dec_NN (av);
730 DEC_DEC_DEPTH; 815 DEC_DEC_DEPTH;
731 return &PL_sv_undef; 816 return &PL_sv_undef;
732} 817}
733 818
734static void 819static void
… …
736{ 821{
737 // for speed reasons, we specialcase single-string 822 // for speed reasons, we specialcase single-string
738 // byte or utf-8 strings as keys, but only when !stringref 823 // byte or utf-8 strings as keys, but only when !stringref
739 824
740 if (ecb_expect_true (!dec->stringref)) 825 if (ecb_expect_true (!dec->stringref))
741 if (ecb_expect_true ((*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) 826 if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8))
742 { 827 {
743 I32 len = decode_uint (dec); 828 STRLEN len = decode_uint (dec);
744 char *key = (char *)dec->cur; 829 char *key = (char *)dec->cur;
745 830
831 WANT (len);
746 dec->cur += len; 832 dec->cur += len;
747 833
748 hv_store (hv, key, len, decode_sv (dec), 0); 834 hv_store (hv, key, len, decode_sv (dec), 0);
749 835
750 return; 836 return;
751 } 837 }
752 else if (ecb_expect_true ((*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) 838 else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8))
753 { 839 {
754 I32 len = decode_uint (dec); 840 STRLEN len = decode_uint (dec);
755 char *key = (char *)dec->cur; 841 char *key = (char *)dec->cur;
756 842
843 WANT (len);
757 dec->cur += len; 844 dec->cur += len;
758 845
759 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) 846 if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8))
760 if (!is_utf8_string (key, len)) 847 if (!is_utf8_string (key, len))
761 ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); 848 ERR ("corrupted CBOR data (invalid UTF-8 in map key)");
… …
766 } 853 }
767 854
768 SV *k = decode_sv (dec); 855 SV *k = decode_sv (dec);
769 SV *v = decode_sv (dec); 856 SV *v = decode_sv (dec);
770 857
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
771 hv_store_ent (hv, k, v, 0); 886 hv_store_ent (hv, k, v, 0);
772 SvREFCNT_dec (k); 887 SvREFCNT_dec_NN (k);
773 888
774fail: 889fail:
775 ; 890 ;
776} 891}
777 892
… …
788 903
789 for (;;) 904 for (;;)
790 { 905 {
791 WANT (1); 906 WANT (1);
792 907
793 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) 908 if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err)
794 { 909 {
795 ++dec->cur; 910 ++dec->cur;
796 break; 911 break;
797 } 912 }
798 913
799 decode_he (dec, hv); 914 decode_he (dec, hv);
800 } 915 }
801 } 916 }
802 else 917 else
803 { 918 {
804 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
805 922
806 while (pairs--) 923 while (pairs--)
807 decode_he (dec, hv); 924 decode_he (dec, hv);
808 } 925 }
809 926
810 DEC_DEC_DEPTH; 927 DEC_DEC_DEPTH;
811 return newRV_noinc ((SV *)hv); 928 return newRV_noinc ((SV *)hv);
812 929
813fail: 930fail:
814 SvREFCNT_dec (hv); 931 SvREFCNT_dec_NN (hv);
815 DEC_DEC_DEPTH; 932 DEC_DEC_DEPTH;
816 return &PL_sv_undef; 933 return &PL_sv_undef;
817} 934}
818 935
819static SV * 936static SV *
820decode_str (dec_t *dec, int utf8) 937decode_str (dec_t *dec, int utf8)
821{ 938{
822 SV *sv = 0; 939 SV *sv = 0;
823 940
824 if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) 941 if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF))
825 { 942 {
826 // indefinite length strings 943 // indefinite length strings
827 ++dec->cur; 944 ++dec->cur;
828 945
829 U8 major = *dec->cur & MAJOR_MISC; 946 U8 major = *dec->cur & MAJOR_MISC;
… …
897 sv = newRV_noinc (decode_sv (dec)); 1014 sv = newRV_noinc (decode_sv (dec));
898 break; 1015 break;
899 1016
900 case CBOR_TAG_STRINGREF_NAMESPACE: 1017 case CBOR_TAG_STRINGREF_NAMESPACE:
901 { 1018 {
1019 // do not use SAVETMPS/FREETMPS, as these will
1020 // erase mortalised caches, e.g. "shareable"
902 ENTER; SAVETMPS; 1021 ENTER;
903 1022
904 SAVESPTR (dec->stringref); 1023 SAVESPTR (dec->stringref);
905 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); 1024 dec->stringref = (AV *)sv_2mortal ((SV *)newAV ());
906 1025
907 sv = decode_sv (dec); 1026 sv = decode_sv (dec);
908 1027
909 FREETMPS; LEAVE; 1028 LEAVE;
910 } 1029 }
911 break; 1030 break;
912 1031
913 case CBOR_TAG_STRINGREF: 1032 case CBOR_TAG_STRINGREF:
914 { 1033 {
915 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1034 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
916 ERR ("corrupted CBOR data (stringref index not an unsigned integer)"); 1035 ERR ("corrupted CBOR data (stringref index not an unsigned integer)");
917 1036
918 UV idx = decode_uint (dec); 1037 UV idx = decode_uint (dec);
919 1038
920 if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) 1039 if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref)))
921 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)");
922 1041
923 sv = newSVsv (AvARRAY (dec->stringref)[idx]); 1042 sv = newSVsv (AvARRAY (dec->stringref)[idx]);
924 } 1043 }
925 break; 1044 break;
… …
953 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT)) 1072 if ((*dec->cur >> MAJOR_SHIFT) != (MAJOR_POS_INT >> MAJOR_SHIFT))
954 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)"); 1073 ERR ("corrupted CBOR data (sharedref index not an unsigned integer)");
955 1074
956 UV idx = decode_uint (dec); 1075 UV idx = decode_uint (dec);
957 1076
958 if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) 1077 if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable)))
959 ERR ("corrupted CBOR data (sharedref index out of bounds)"); 1078 ERR ("corrupted CBOR data (sharedref index out of bounds)");
960 1079
961 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); 1080 sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]);
962 1081
963 if (sv == &PL_sv_undef) 1082 if (sv == &PL_sv_undef)
… …
965 } 1084 }
966 break; 1085 break;
967 1086
968 case CBOR_TAG_PERL_OBJECT: 1087 case CBOR_TAG_PERL_OBJECT:
969 { 1088 {
1089 if (dec->cbor.flags & F_FORBID_OBJECTS)
1090 goto filter;
1091
970 sv = decode_sv (dec); 1092 sv = decode_sv (dec);
971 1093
972 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) 1094 if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV)
973 ERR ("corrupted CBOR data (non-array perl object)"); 1095 ERR ("corrupted CBOR data (non-array perl object)");
974 1096
… …
984 if (!method) 1106 if (!method)
985 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)");
986 1108
987 dSP; 1109 dSP;
988 1110
989 ENTER; SAVETMPS; PUSHMARK (SP); 1111 ENTER; SAVETMPS;
1112 PUSHMARK (SP);
990 EXTEND (SP, len + 1); 1113 EXTEND (SP, len + 1);
991 // 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
992 PUSHs (*av_fetch (av, 0, 1)); 1115 PUSHs (*av_fetch (av, 0, 1));
993 PUSHs (sv_cbor); 1116 PUSHs (sv_cbor);
994 1117
… …
1002 SPAGAIN; 1125 SPAGAIN;
1003 1126
1004 if (SvTRUE (ERRSV)) 1127 if (SvTRUE (ERRSV))
1005 { 1128 {
1006 FREETMPS; LEAVE; 1129 FREETMPS; LEAVE;
1007 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1130 ERR_ERRSV;
1008 } 1131 }
1009 1132
1010 SvREFCNT_dec (sv); 1133 SvREFCNT_dec_NN (sv);
1011 sv = SvREFCNT_inc (POPs); 1134 sv = SvREFCNT_inc (POPs);
1012 1135
1013 PUTBACK; 1136 PUTBACK;
1014 1137
1015 FREETMPS; LEAVE; 1138 FREETMPS; LEAVE;
1016 } 1139 }
1017 break; 1140 break;
1018 1141
1019 default: 1142 default:
1143 filter:
1020 { 1144 {
1145 SV *tag_sv = newSVuv (tag);
1146
1021 sv = decode_sv (dec); 1147 sv = decode_sv (dec);
1022 1148
1023 dSP; 1149 dSP;
1024 ENTER; SAVETMPS; PUSHMARK (SP); 1150 ENTER; SAVETMPS;
1151 PUSHMARK (SP);
1025 EXTEND (SP, 2); 1152 EXTEND (SP, 2);
1026 PUSHs (newSVuv (tag)); 1153 PUSHs (tag_sv);
1027 PUSHs (sv); 1154 PUSHs (sv);
1028 1155
1029 PUTBACK; 1156 PUTBACK;
1030 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);
1031 SPAGAIN; 1158 SPAGAIN;
1032 1159
1033 if (SvTRUE (ERRSV)) 1160 if (SvTRUE (ERRSV))
1034 { 1161 {
1162 SvREFCNT_dec_NN (tag_sv);
1035 FREETMPS; LEAVE; 1163 FREETMPS; LEAVE;
1036 ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); 1164 ERR_ERRSV;
1037 } 1165 }
1038 1166
1039 if (count) 1167 if (count)
1040 { 1168 {
1169 SvREFCNT_dec_NN (tag_sv);
1041 SvREFCNT_dec (sv); 1170 SvREFCNT_dec_NN (sv);
1042 sv = SvREFCNT_inc (POPs); 1171 sv = SvREFCNT_inc_NN (TOPs);
1172 SP -= count;
1043 } 1173 }
1044 else 1174 else
1045 { 1175 {
1046 AV *av = newAV (); 1176 AV *av = newAV ();
1047 av_push (av, newSVuv (tag)); 1177 av_push (av, tag_sv);
1048 av_push (av, sv); 1178 av_push (av, sv);
1049 1179
1050 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash 1180 HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash
1051 ? cbor_tagged_stash 1181 ? cbor_tagged_stash
1052 : gv_stashpv ("CBOR::XS::Tagged" , 1); 1182 : gv_stashpv ("CBOR::XS::Tagged" , 1);
… …
1155} 1285}
1156 1286
1157static SV * 1287static SV *
1158decode_cbor (SV *string, CBOR *cbor, char **offset_return) 1288decode_cbor (SV *string, CBOR *cbor, char **offset_return)
1159{ 1289{
1160 dec_t dec = { }; 1290 dec_t dec = { 0 };
1161 SV *sv; 1291 SV *sv;
1162 STRLEN len; 1292 STRLEN len;
1163 char *data = SvPVbyte (string, len); 1293 char *data = SvPVbyte (string, len);
1164 1294
1165 if (len > cbor->max_size && cbor->max_size) 1295 if (len > cbor->max_size && cbor->max_size)
… …
1181 1311
1182 if (dec.err) 1312 if (dec.err)
1183 { 1313 {
1184 if (dec.shareable) 1314 if (dec.shareable)
1185 { 1315 {
1186 // need to break cyclic links, which whould all be in shareable 1316 // need to break cyclic links, which would all be in shareable
1187 int i; 1317 int i;
1188 SV **svp; 1318 SV **svp;
1189 1319
1190 for (i = av_len (dec.shareable) + 1; i--; ) 1320 for (i = av_len (dec.shareable) + 1; i--; )
1191 if ((svp = av_fetch (dec.shareable, i, 0))) 1321 if ((svp = av_fetch (dec.shareable, i, 0)))
1192 sv_setsv (*svp, &PL_sv_undef); 1322 sv_setsv (*svp, &PL_sv_undef);
1193 } 1323 }
1194 1324
1195 SvREFCNT_dec (sv); 1325 SvREFCNT_dec_NN (sv);
1326
1327 if (dec.err_sv)
1328 sv_2mortal (dec.err_sv);
1329
1196 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);
1197 } 1331 }
1198 1332
1199 sv = sv_2mortal (sv); 1333 sv = sv_2mortal (sv);
1200 1334
… …
1256 1390
1257 int major = *p >> MAJOR_SHIFT; 1391 int major = *p >> MAJOR_SHIFT;
1258 1392
1259 switch (major) 1393 switch (major)
1260 { 1394 {
1395 case MAJOR_TAG >> MAJOR_SHIFT:
1396 ++count; // tags merely prefix another value
1397 break;
1398
1261 case MAJOR_BYTES >> MAJOR_SHIFT: 1399 case MAJOR_BYTES >> MAJOR_SHIFT:
1262 case MAJOR_TEXT >> MAJOR_SHIFT: 1400 case MAJOR_TEXT >> MAJOR_SHIFT:
1263 case MAJOR_ARRAY >> MAJOR_SHIFT: 1401 case MAJOR_ARRAY >> MAJOR_SHIFT:
1264 case MAJOR_MAP >> MAJOR_SHIFT: 1402 case MAJOR_MAP >> MAJOR_SHIFT:
1265 { 1403 {
… …
1340 1478
1341 default_filter = newSVpv ("CBOR::XS::default_filter", 0); 1479 default_filter = newSVpv ("CBOR::XS::default_filter", 0);
1342 1480
1343 sv_cbor = newSVpv ("CBOR", 0); 1481 sv_cbor = newSVpv ("CBOR", 0);
1344 SvREADONLY_on (sv_cbor); 1482 SvREADONLY_on (sv_cbor);
1483
1484 assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1));
1345} 1485}
1346 1486
1347PROTOTYPES: DISABLE 1487PROTOTYPES: DISABLE
1348 1488
1349void CLONE (...) 1489void CLONE (...)
… …
1369 ALIAS: 1509 ALIAS:
1370 shrink = F_SHRINK 1510 shrink = F_SHRINK
1371 allow_unknown = F_ALLOW_UNKNOWN 1511 allow_unknown = F_ALLOW_UNKNOWN
1372 allow_sharing = F_ALLOW_SHARING 1512 allow_sharing = F_ALLOW_SHARING
1373 allow_cycles = F_ALLOW_CYCLES 1513 allow_cycles = F_ALLOW_CYCLES
1514 forbid_objects = F_FORBID_OBJECTS
1374 pack_strings = F_PACK_STRINGS 1515 pack_strings = F_PACK_STRINGS
1516 text_keys = F_TEXT_KEYS
1517 text_strings = F_TEXT_STRINGS
1375 validate_utf8 = F_VALIDATE_UTF8 1518 validate_utf8 = F_VALIDATE_UTF8
1376 PPCODE: 1519 PPCODE:
1377{ 1520{
1378 if (enable) 1521 if (enable)
1379 self->flags |= ix; 1522 self->flags |= ix;
… …
1387 ALIAS: 1530 ALIAS:
1388 get_shrink = F_SHRINK 1531 get_shrink = F_SHRINK
1389 get_allow_unknown = F_ALLOW_UNKNOWN 1532 get_allow_unknown = F_ALLOW_UNKNOWN
1390 get_allow_sharing = F_ALLOW_SHARING 1533 get_allow_sharing = F_ALLOW_SHARING
1391 get_allow_cycles = F_ALLOW_CYCLES 1534 get_allow_cycles = F_ALLOW_CYCLES
1535 get_forbid_objects = F_FORBID_OBJECTS
1392 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
1393 get_validate_utf8 = F_VALIDATE_UTF8 1539 get_validate_utf8 = F_VALIDATE_UTF8
1394 PPCODE: 1540 PPCODE:
1395 XPUSHs (boolSV (self->flags & ix)); 1541 XPUSHs (boolSV (self->flags & ix));
1396 1542
1397void max_depth (CBOR *self, U32 max_depth = 0x80000000UL) 1543void max_depth (CBOR *self, U32 max_depth = 0x80000000UL)
… …
1448 PUSHs (sv); 1594 PUSHs (sv);
1449 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr)))); 1595 PUSHs (sv_2mortal (newSVuv (offset - SvPVX (cborstr))));
1450} 1596}
1451 1597
1452void incr_parse (CBOR *self, SV *cborstr) 1598void incr_parse (CBOR *self, SV *cborstr)
1599 ALIAS:
1600 incr_parse_multiple = 1
1453 PPCODE: 1601 PPCODE:
1454{ 1602{
1455 if (SvUTF8 (cborstr)) 1603 if (SvUTF8 (cborstr))
1456 sv_utf8_downgrade (cborstr, 0); 1604 sv_utf8_downgrade (cborstr, 0);
1457 1605
… …
1487 av_push (self->incr_count, newSViv (1)); 1635 av_push (self->incr_count, newSViv (1));
1488 1636
1489 self->incr_pos = 0; 1637 self->incr_pos = 0;
1490 self->incr_need = self->incr_pos + 1; 1638 self->incr_need = self->incr_pos + 1;
1491 } 1639 }
1492 while (GIMME_V == G_ARRAY); 1640 while (ix);
1493} 1641}
1494 1642
1495void incr_reset (CBOR *self) 1643void incr_reset (CBOR *self)
1496 CODE: 1644 CODE:
1497{ 1645{
… …
1525 cbor_init (&cbor); 1673 cbor_init (&cbor);
1526 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN; 1674 PUTBACK; cborstr = decode_cbor (cborstr, &cbor, 0); SPAGAIN;
1527 XPUSHs (cborstr); 1675 XPUSHs (cborstr);
1528} 1676}
1529 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