--- CBOR-XS/XS.xs 2016/04/24 13:15:19 1.51 +++ CBOR-XS/XS.xs 2020/11/30 18:30:29 1.68 @@ -8,6 +8,7 @@ #include #include #include +#include #define ECB_NO_THREADS 1 #include "ecb.h" @@ -22,6 +23,9 @@ #ifndef HvNAMEUTF8 # define HvNAMEUTF8(hv) 0 #endif +#ifndef SvREFCNT_inc_NN +# define SvREFCNT_inc_NN(sv) SvREFCNT_inc (sv) +#endif #ifndef SvREFCNT_dec_NN # define SvREFCNT_dec_NN(sv) SvREFCNT_dec (sv) #endif @@ -97,13 +101,28 @@ CBOR_TAG_MAGIC = 55799, // self-describe cbor }; +// known forced types, also hardcoded in CBOR.pm +enum +{ + AS_CBOR = 0, + AS_INT = 1, + AS_BYTES = 2, + AS_TEXT = 3, + AS_FLOAT16 = 4, + AS_FLOAT32 = 5, + AS_FLOAT64 = 6, + // possibly future enhancements: (generic) float, (generic) string +}; + #define F_SHRINK 0x00000001UL #define F_ALLOW_UNKNOWN 0x00000002UL #define F_ALLOW_SHARING 0x00000004UL #define F_ALLOW_CYCLES 0x00000008UL -#define F_PACK_STRINGS 0x00000010UL -#define F_UTF8_STRINGS 0x00000020UL -#define F_VALIDATE_UTF8 0x00000040UL +#define F_FORBID_OBJECTS 0x00000010UL +#define F_PACK_STRINGS 0x00000020UL +#define F_TEXT_KEYS 0x00000040UL +#define F_TEXT_STRINGS 0x00000080UL +#define F_VALIDATE_UTF8 0x00000100UL #define INIT_SIZE 32 // initial scalar size to be allocated @@ -186,15 +205,11 @@ ecb_inline int minimum_string_length (UV idx) { - return idx > 23 - ? idx > 0xffU - ? idx > 0xffffU - ? idx > 0xffffffffU - ? 11 - : 7 - : 5 - : 4 - : 3; + return idx <= 23 ? 3 + : idx <= 0xffU ? 4 + : idx <= 0xffffU ? 5 + : idx <= 0xffffffffU ? 7 + : 11; } ///////////////////////////////////////////////////////////////////////////// @@ -217,7 +232,7 @@ ecb_inline void need (enc_t *enc, STRLEN len) { - if (ecb_expect_false (enc->cur + len >= enc->end)) + if (ecb_expect_false ((uintptr_t)(enc->end - enc->cur) < len)) { STRLEN cur = enc->cur - (char *)SvPVX (enc->sv); SvGROW (enc->sv, cur + (len < (cur >> 2) ? cur >> 2 : len) + 1); @@ -233,6 +248,7 @@ *enc->cur++ = ch; } +// used for tags, intregers, element counts and so on static void encode_uint (enc_t *enc, int major, UV len) { @@ -273,25 +289,54 @@ } } +// encodes a perl value into a CBOR integer +ecb_inline void +encode_int (enc_t *enc, SV *sv) +{ + if (SvIsUV (sv)) + encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); + else if (SvIVX (sv) >= 0) + encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); + else + encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); +} + ecb_inline void encode_tag (enc_t *enc, UV tag) { encode_uint (enc, MAJOR_TAG, tag); } +// exceptional (hopefully) slow path for byte strings that need to be utf8-encoded +ecb_noinline static void +encode_str_utf8 (enc_t *enc, int utf8, char *str, STRLEN len) +{ + STRLEN ulen = len; + U8 *p, *pend = (U8 *)str + len; + + for (p = (U8 *)str; p < pend; ++p) + ulen += *p >> 7; // count set high bits + + encode_uint (enc, MAJOR_TEXT, ulen); + + need (enc, ulen); + for (p = (U8 *)str; p < pend; ++p) + if (*p < 0x80) + *enc->cur++ = *p; + else + { + *enc->cur++ = 0xc0 + (*p >> 6); + *enc->cur++ = 0x80 + (*p & 63); + } +} + ecb_inline void -encode_str (enc_t *enc, int utf8, char *str, STRLEN len) +encode_str (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) { - if (ecb_expect_false (enc->cbor.flags & F_UTF8_STRINGS)) + if (ecb_expect_false (upgrade_utf8)) if (!utf8) { - SV *sv = sv_newmortal (); - char *s; STRLEN l; - - sv_setpvn (sv, str, len); - - s = SvPVutf8 (sv, l); - encode_str (enc, 1, s, l); + encode_str_utf8 (enc, utf8, str, len); return; } @@ -301,8 +346,8 @@ enc->cur += len; } -static void -encode_strref (enc_t *enc, int utf8, char *str, STRLEN len) +ecb_inline void +encode_strref (enc_t *enc, int upgrade_utf8, int utf8, char *str, STRLEN len) { if (ecb_expect_false (enc->cbor.flags & F_PACK_STRINGS)) { @@ -323,7 +368,97 @@ } } - encode_str (enc, utf8, str, len); + encode_str (enc, upgrade_utf8, utf8, str, len); +} + +ecb_inline void +encode_float16 (enc_t *enc, NV nv) +{ + need (enc, 1+2); + + *enc->cur++ = MAJOR_MISC | MISC_FLOAT16; + + uint16_t fp = ecb_float_to_binary16 (nv); + + if (!ecb_big_endian ()) + fp = ecb_bswap16 (fp); + + memcpy (enc->cur, &fp, 2); + enc->cur += 2; +} + +ecb_inline void +encode_float32 (enc_t *enc, NV nv) +{ + need (enc, 1+4); + + *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; + + uint32_t fp = ecb_float_to_binary32 (nv); + + if (!ecb_big_endian ()) + fp = ecb_bswap32 (fp); + + memcpy (enc->cur, &fp, 4); + enc->cur += 4; +} + +ecb_inline void +encode_float64 (enc_t *enc, NV nv) +{ + need (enc, 1+8); + + *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; + + uint64_t fp = ecb_double_to_binary64 (nv); + + if (!ecb_big_endian ()) + fp = ecb_bswap64 (fp); + + memcpy (enc->cur, &fp, 8); + enc->cur += 8; +} + +ecb_inline void +encode_forced (enc_t *enc, UV type, SV *sv) +{ + switch (type) + { + case AS_CBOR: + { + STRLEN len; + char *str = SvPVbyte (sv, len); + + need (enc, len); + memcpy (enc->cur, str, len); + enc->cur += len; + } + break; + + case AS_BYTES: + { + STRLEN len; + char *str = SvPVbyte (sv, len); + encode_strref (enc, 0, 0, str, len); + } + break; + + case AS_TEXT: + { + STRLEN len; + char *str = SvPVutf8 (sv, len); + encode_strref (enc, 1, 1, str, len); + } + break; + + case AS_INT: encode_int (enc, sv); break; + case AS_FLOAT16: encode_float16 (enc, SvNV (sv)); break; + case AS_FLOAT32: encode_float32 (enc, SvNV (sv)); break; + case AS_FLOAT64: encode_float64 (enc, SvNV (sv)); break; + + default: + croak ("encountered malformed CBOR::XS::Tagged object"); + } } static void encode_sv (enc_t *enc, SV *sv); @@ -340,7 +475,7 @@ encode_uint (enc, MAJOR_ARRAY, len + 1); - if (SvMAGICAL (av)) + if (ecb_expect_false (SvMAGICAL (av))) for (i = 0; i <= len; ++i) { SV **svp = av_fetch (av, i, 0); @@ -369,7 +504,7 @@ int pairs = hv_iterinit (hv); int mg = SvMAGICAL (hv); - if (mg) + if (ecb_expect_false (mg)) encode_ch (enc, MAJOR_MAP | MINOR_INDEF); else encode_uint (enc, MAJOR_MAP, pairs); @@ -379,12 +514,12 @@ if (HeKLEN (he) == HEf_SVKEY) encode_sv (enc, HeSVKEY (he)); else - encode_strref (enc, HeKUTF8 (he), HeKEY (he), HeKLEN (he)); + encode_strref (enc, enc->cbor.flags & (F_TEXT_KEYS | F_TEXT_STRINGS), HeKUTF8 (he), HeKEY (he), HeKLEN (he)); encode_sv (enc, ecb_expect_false (mg) ? hv_iterval (hv, he) : HeVAL (he)); } - if (mg) + if (ecb_expect_false (mg)) encode_ch (enc, MAJOR_MISC | MINOR_INDEF); --enc->depth; @@ -427,8 +562,22 @@ if (svt != SVt_PVAV) croak ("encountered CBOR::XS::Tagged object that isn't an array"); - encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); - encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); + switch (av_len ((AV *)sv)) + { + case 2-1: + // actually a tagged value + encode_uint (enc, MAJOR_TAG, SvUV (*av_fetch ((AV *)sv, 0, 1))); + encode_sv (enc, *av_fetch ((AV *)sv, 1, 1)); + break; + + case 3-1: + // a forced type [value, type, undef] + encode_forced (enc, SvUV (*av_fetch ((AV *)sv, 1, 1)), *av_fetch ((AV *)sv, 0, 1)); + break; + + default: + croak ("encountered malformed CBOR::XS::Tagged object"); + } return; } @@ -437,7 +586,7 @@ if (ecb_expect_false (SvREFCNT (sv) > 1) && ecb_expect_false (enc->cbor.flags & F_ALLOW_SHARING)) { - if (!enc->shareable) + if (ecb_expect_false (!enc->shareable)) enc->shareable = (HV *)sv_2mortal ((SV *)newHV ()); SV **svp = hv_fetch (enc->shareable, (char *)&sv, sizeof (sv), 1); @@ -461,7 +610,10 @@ HV *stash = SvSTASH (sv); GV *method; - if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) + if (enc->cbor.flags & F_FORBID_OBJECTS) + croak ("encountered object '%s', but forbid_objects is enabled", + SvPV_nolen (sv_2mortal (newRV_inc (sv)))); + else if ((method = gv_fetchmethod_autoload (stash, "TO_CBOR", 0))) { dSP; @@ -490,7 +642,6 @@ dSP; ENTER; SAVETMPS; - SAVESTACK_POS (); PUSHMARK (SP); EXTEND (SP, 2); // we re-bless the reference to get overload and other niceties right @@ -507,10 +658,16 @@ encode_tag (enc, CBOR_TAG_PERL_OBJECT); encode_uint (enc, MAJOR_ARRAY, count + 1); - encode_strref (enc, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); + encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); + + { + int i; - while (count) - encode_sv (enc, SP[1 - count--]); + for (i = 0; i < count; ++i) + encode_sv (enc, SP[i + 1 - count]); + + SP -= count; + } PUTBACK; @@ -542,29 +699,9 @@ encode_uint (enc, MAJOR_POS_INT, (U32)nv); //TODO: maybe I32? else if (ecb_expect_false (nv == (float)nv)) - { - uint32_t fp = ecb_float_to_binary32 (nv); - - *enc->cur++ = MAJOR_MISC | MISC_FLOAT32; - - if (!ecb_big_endian ()) - fp = ecb_bswap32 (fp); - - memcpy (enc->cur, &fp, 4); - enc->cur += 4; - } + encode_float32 (enc, nv); else - { - uint64_t fp = ecb_double_to_binary64 (nv); - - *enc->cur++ = MAJOR_MISC | MISC_FLOAT64; - - if (!ecb_big_endian ()) - fp = ecb_bswap64 (fp); - - memcpy (enc->cur, &fp, 8); - enc->cur += 8; - } + encode_float64 (enc, nv); } static void @@ -576,19 +713,12 @@ { STRLEN len; char *str = SvPV (sv, len); - encode_strref (enc, SvUTF8 (sv), str, len); + encode_strref (enc, enc->cbor.flags & F_TEXT_STRINGS, SvUTF8 (sv), str, len); } else if (SvNOKp (sv)) encode_nv (enc, sv); else if (SvIOKp (sv)) - { - if (SvIsUV (sv)) - encode_uint (enc, MAJOR_POS_INT, SvUVX (sv)); - else if (SvIVX (sv) >= 0) - encode_uint (enc, MAJOR_POS_INT, SvIVX (sv)); - else - encode_uint (enc, MAJOR_NEG_INT, -(SvIVX (sv) + 1)); - } + encode_int (enc, sv); else if (SvROK (sv)) encode_rv (enc, SvRV (sv)); else if (!SvOK (sv)) @@ -605,10 +735,10 @@ { enc_t enc = { 0 }; - enc.cbor = *cbor; - enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); - enc.cur = SvPVX (enc.sv); - enc.end = SvEND (enc.sv); + enc.cbor = *cbor; + enc.sv = sv_2mortal (NEWSV (0, INIT_SIZE)); + enc.cur = SvPVX (enc.sv); + enc.end = SvEND (enc.sv); SvPOK_only (enc.sv); @@ -645,13 +775,46 @@ AV *shareable; AV *stringref; SV *decode_tagged; + SV *err_sv; // optional sv for error, needs to be freed } dec_t; -#define ERR(reason) SB if (!dec->err) dec->err = reason; goto fail; SE +// set dec->err to ERRSV +ecb_cold static void +err_errsv (dec_t *dec) +{ + if (!dec->err) + { + dec->err_sv = newSVsv (ERRSV); + + // chop off the trailing \n + SvCUR_set (dec->err_sv, SvCUR (dec->err_sv) - 1); + *SvEND (dec->err_sv) = 0; + + dec->err = SvPVutf8_nolen (dec->err_sv); + } +} + +// the following functions are used to reduce code size and help the compiler to optimise +ecb_cold static void +err_set (dec_t *dec, const char *reason) +{ + if (!dec->err) + dec->err = reason; +} + +ecb_cold static void +err_unexpected_end (dec_t *dec) +{ + err_set (dec, "unexpected end of CBOR data"); +} -#define WANT(len) if (ecb_expect_false (dec->cur + len > dec->end)) ERR ("unexpected end of CBOR data") +#define ERR_DO(do) SB do; goto fail; SE +#define ERR(reason) ERR_DO (err_set (dec, reason)) +#define ERR_ERRSV ERR_DO (err_errsv (dec)) -#define DEC_INC_DEPTH if (++dec->depth > dec->cbor.max_depth) ERR (ERR_NESTING_EXCEEDED) +#define WANT(len) if (ecb_expect_false ((uintptr_t)(dec->end - dec->cur) < (STRLEN)len)) ERR_DO (err_unexpected_end (dec)) + +#define DEC_INC_DEPTH if (ecb_expect_false (++dec->depth > dec->cbor.max_depth)) ERR (ERR_NESTING_EXCEEDED) #define DEC_DEC_DEPTH --dec->depth static UV @@ -727,7 +890,7 @@ { WANT (1); - if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) + if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) { ++dec->cur; break; @@ -738,7 +901,7 @@ } else { - int i, len = decode_uint (dec); + UV i, len = decode_uint (dec); WANT (len); // complexity check for av_fill - need at least one byte per value, do not allow supersize arrays av_fill (av, len - 1); @@ -751,7 +914,7 @@ return newRV_noinc ((SV *)av); fail: - SvREFCNT_dec (av); + SvREFCNT_dec_NN (av); DEC_DEC_DEPTH; return &PL_sv_undef; } @@ -765,7 +928,7 @@ if (ecb_expect_true (!dec->stringref)) if (ecb_expect_true ((U8)(*dec->cur - MAJOR_BYTES) <= LENGTH_EXT8)) { - I32 len = decode_uint (dec); + STRLEN len = decode_uint (dec); char *key = (char *)dec->cur; WANT (len); @@ -777,7 +940,7 @@ } else if (ecb_expect_true ((U8)(*dec->cur - MAJOR_TEXT) <= LENGTH_EXT8)) { - I32 len = decode_uint (dec); + STRLEN len = decode_uint (dec); char *key = (char *)dec->cur; WANT (len); @@ -795,8 +958,36 @@ SV *k = decode_sv (dec); SV *v = decode_sv (dec); + // we leak memory if uncaught exceptions are thrown by random magical + // methods, and this is hopefully the only place where it can happen, + // so if there is a chance of an exception, take the very slow path. + // since catching exceptions is "undocumented/internal/forbidden" by + // the new p5p powers, we need to call out to a perl function :/ + if (ecb_expect_false (SvAMAGIC (k))) + { + dSP; + + ENTER; SAVETMPS; + PUSHMARK (SP); + EXTEND (SP, 3); + PUSHs (sv_2mortal (newRV_inc ((SV *)hv))); + PUSHs (sv_2mortal (k)); + PUSHs (sv_2mortal (v)); + + PUTBACK; + call_pv ("CBOR::XS::_hv_store", G_VOID | G_DISCARD | G_EVAL); + SPAGAIN; + + FREETMPS; LEAVE; + + if (SvTRUE (ERRSV)) + ERR_ERRSV; + + return; + } + hv_store_ent (hv, k, v, 0); - SvREFCNT_dec (k); + SvREFCNT_dec_NN (k); fail: ; @@ -817,7 +1008,7 @@ { WANT (1); - if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) + if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) { ++dec->cur; break; @@ -828,7 +1019,9 @@ } else { - int pairs = decode_uint (dec); + UV pairs = decode_uint (dec); + + WANT (pairs); // complexity check - need at least one byte per value, do not allow supersize hashes while (pairs--) decode_he (dec, hv); @@ -838,7 +1031,7 @@ return newRV_noinc ((SV *)hv); fail: - SvREFCNT_dec (hv); + SvREFCNT_dec_NN (hv); DEC_DEC_DEPTH; return &PL_sv_undef; } @@ -848,7 +1041,7 @@ { SV *sv = 0; - if ((*dec->cur & MINOR_MASK) == MINOR_INDEF) + if (ecb_expect_false ((*dec->cur & MINOR_MASK) == MINOR_INDEF)) { // indefinite length strings ++dec->cur; @@ -926,14 +1119,16 @@ case CBOR_TAG_STRINGREF_NAMESPACE: { - ENTER; SAVETMPS; + // do not use SAVETMPS/FREETMPS, as these will + // erase mortalised caches, e.g. "shareable" + ENTER; SAVESPTR (dec->stringref); dec->stringref = (AV *)sv_2mortal ((SV *)newAV ()); sv = decode_sv (dec); - FREETMPS; LEAVE; + LEAVE; } break; @@ -944,7 +1139,7 @@ UV idx = decode_uint (dec); - if (!dec->stringref || (int)idx > AvFILLp (dec->stringref)) + if (!dec->stringref || idx >= (UV)(1 + AvFILLp (dec->stringref))) ERR ("corrupted CBOR data (stringref index out of bounds or outside namespace)"); sv = newSVsv (AvARRAY (dec->stringref)[idx]); @@ -982,7 +1177,7 @@ UV idx = decode_uint (dec); - if (!dec->shareable || (int)idx > AvFILLp (dec->shareable)) + if (!dec->shareable || idx >= (UV)(1 + AvFILLp (dec->shareable))) ERR ("corrupted CBOR data (sharedref index out of bounds)"); sv = SvREFCNT_inc_NN (AvARRAY (dec->shareable)[idx]); @@ -994,6 +1189,9 @@ case CBOR_TAG_PERL_OBJECT: { + if (dec->cbor.flags & F_FORBID_OBJECTS) + goto filter; + sv = decode_sv (dec); if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) @@ -1032,10 +1230,10 @@ if (SvTRUE (ERRSV)) { FREETMPS; LEAVE; - ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); + ERR_ERRSV; } - SvREFCNT_dec (sv); + SvREFCNT_dec_NN (sv); sv = SvREFCNT_inc (POPs); PUTBACK; @@ -1045,15 +1243,17 @@ break; default: + filter: { + SV *tag_sv = newSVuv (tag); + sv = decode_sv (dec); dSP; ENTER; SAVETMPS; - SAVESTACK_POS (); PUSHMARK (SP); EXTEND (SP, 2); - PUSHs (newSVuv (tag)); + PUSHs (tag_sv); PUSHs (sv); PUTBACK; @@ -1062,19 +1262,22 @@ if (SvTRUE (ERRSV)) { + SvREFCNT_dec_NN (tag_sv); FREETMPS; LEAVE; - ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); + ERR_ERRSV; } if (count) { - SvREFCNT_dec (sv); - sv = SvREFCNT_inc (POPs); + SvREFCNT_dec_NN (tag_sv); + SvREFCNT_dec_NN (sv); + sv = SvREFCNT_inc_NN (TOPs); + SP -= count; } else { AV *av = newAV (); - av_push (av, newSVuv (tag)); + av_push (av, tag_sv); av_push (av, sv); HV *tagged_stash = !CBOR_SLOW || cbor_tagged_stash @@ -1213,7 +1416,7 @@ { if (dec.shareable) { - // need to break cyclic links, which whould all be in shareable + // need to break cyclic links, which would all be in shareable int i; SV **svp; @@ -1222,7 +1425,11 @@ sv_setsv (*svp, &PL_sv_undef); } - SvREFCNT_dec (sv); + SvREFCNT_dec_NN (sv); + + if (dec.err_sv) + sv_2mortal (dec.err_sv); + croak ("%s, at offset %d (octet 0x%02x)", dec.err, dec.cur - (U8 *)data, (int)(uint8_t)*dec.cur); } @@ -1376,6 +1583,8 @@ sv_cbor = newSVpv ("CBOR", 0); SvREADONLY_on (sv_cbor); + + assert (("STRLEN must be an unsigned type", 0 <= (STRLEN)-1)); } PROTOTYPES: DISABLE @@ -1405,8 +1614,10 @@ allow_unknown = F_ALLOW_UNKNOWN allow_sharing = F_ALLOW_SHARING allow_cycles = F_ALLOW_CYCLES + forbid_objects = F_FORBID_OBJECTS pack_strings = F_PACK_STRINGS - utf8_strings = F_UTF8_STRINGS + text_keys = F_TEXT_KEYS + text_strings = F_TEXT_STRINGS validate_utf8 = F_VALIDATE_UTF8 PPCODE: { @@ -1424,7 +1635,10 @@ get_allow_unknown = F_ALLOW_UNKNOWN get_allow_sharing = F_ALLOW_SHARING get_allow_cycles = F_ALLOW_CYCLES + get_forbid_objects = F_FORBID_OBJECTS get_pack_strings = F_PACK_STRINGS + get_text_keys = F_TEXT_KEYS + get_text_strings = F_TEXT_STRINGS get_validate_utf8 = F_VALIDATE_UTF8 PPCODE: XPUSHs (boolSV (self->flags & ix)); @@ -1564,3 +1778,19 @@ XPUSHs (cborstr); } +#ifdef __AFL_COMPILER + +void +afl_init () + CODE: + __AFL_INIT (); + +int +afl_loop (unsigned int count = 10000) + CODE: + RETVAL = __AFL_LOOP (count); + OUTPUT: + RETVAL + +#endif +