--- CBOR-XS/XS.xs 2016/11/25 13:27:29 1.58 +++ CBOR-XS/XS.xs 2021/10/23 03:00:31 1.73 @@ -8,6 +8,7 @@ #include #include #include +#include #define ECB_NO_THREADS 1 #include "ecb.h" @@ -22,10 +23,16 @@ #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 +// perl's is_utf8_string interprets len=0 as "calculate len", but we want it to mean 0 +#define cbor_is_utf8_string(str,len) (!(len) || is_utf8_string ((str), (len))) + // known major and minor types enum cbor_type { @@ -97,14 +104,29 @@ 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, + AS_MAP = 7, + // 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_TEXT_KEYS 0x00000020UL -#define F_TEXT_STRINGS 0x00000040UL -#define F_VALIDATE_UTF8 0x00000080UL +#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 @@ -184,18 +206,14 @@ } // minimum length of a string to be registered for stringref -ecb_inline int +ecb_inline STRLEN 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; } ///////////////////////////////////////////////////////////////////////////// @@ -218,7 +236,7 @@ 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); @@ -227,6 +245,8 @@ } } +static void encode_sv (enc_t *enc, SV *sv); + ecb_inline void encode_ch (enc_t *enc, char ch) { @@ -234,6 +254,7 @@ *enc->cur++ = ch; } +// used for tags, intregers, element counts and so on static void encode_uint (enc_t *enc, int major, UV len) { @@ -274,6 +295,18 @@ } } +// 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) { @@ -344,7 +377,136 @@ encode_str (enc, upgrade_utf8, utf8, str, len); } -static void encode_sv (enc_t *enc, SV *sv); +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_bool (enc_t *enc, int istrue) +{ + encode_ch (enc, istrue ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); +} + +// encodes an arrayref containing key-value pairs as CBOR map +ecb_inline void +encode_array_as_map (enc_t *enc, SV *sv) +{ + if (enc->depth >= enc->cbor.max_depth) + croak (ERR_NESTING_EXCEEDED); + + ++enc->depth; + + // as_map does error checking for us, but we re-check in case + // things have changed. + + if (!SvROK (sv) || SvTYPE (SvRV (sv)) != SVt_PVAV) + croak ("CBOR::XS::as_map requires an array reference (did you change the array after calling as_map?)"); + + AV *av = (AV *)SvRV (sv); + int i, len = av_len (av); + + if (!(len & 1)) + croak ("CBOR::XS::as_map requires an even number of elements (did you change the array after calling as_map?)"); + + encode_uint (enc, MAJOR_MAP, (len + 1) >> 1); + + for (i = 0; i <= len; ++i) + { + SV **svp = av_fetch (av, i, 0); + encode_sv (enc, svp ? *svp : &PL_sv_undef); + } + + --enc->depth; +} + +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; + + case AS_MAP: encode_array_as_map (enc, sv); break; + + default: + croak ("encountered malformed CBOR::XS::Tagged object"); + } +} static void encode_av (enc_t *enc, AV *av) @@ -358,7 +520,7 @@ 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); @@ -387,7 +549,7 @@ 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); @@ -402,7 +564,7 @@ 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; @@ -432,7 +594,7 @@ if (stash == boolean_stash) { - encode_ch (enc, SvIV (sv) ? MAJOR_MISC | SIMPLE_TRUE : MAJOR_MISC | SIMPLE_FALSE); + encode_bool (enc, SvIV (sv)); return; } else if (stash == error_stash) @@ -445,8 +607,22 @@ 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; } @@ -455,7 +631,7 @@ 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); @@ -479,7 +655,10 @@ 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; @@ -508,7 +687,6 @@ dSP; ENTER; SAVETMPS; - SAVESTACK_POS (); PUSHMARK (SP); EXTEND (SP, 2); // we re-bless the reference to get overload and other niceties right @@ -527,8 +705,14 @@ encode_uint (enc, MAJOR_ARRAY, count + 1); encode_strref (enc, 0, HvNAMEUTF8 (stash), HvNAME (stash), HvNAMELEN (stash)); - while (count) - encode_sv (enc, SP[1 - count--]); + { + int i; + + for (i = 0; i < count; ++i) + encode_sv (enc, SP[i + 1 - count]); + + SP -= count; + } PUTBACK; @@ -560,29 +744,9 @@ 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 @@ -599,14 +763,7 @@ 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)) @@ -623,10 +780,10 @@ { 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); @@ -663,11 +820,44 @@ 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 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 WANT(len) if (ecb_expect_false ((UV)(dec->end - dec->cur) < (UV)len)) ERR ("unexpected end of CBOR data") +#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 @@ -745,7 +935,7 @@ { WANT (1); - if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) + if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) { ++dec->cur; break; @@ -769,7 +959,7 @@ return newRV_noinc ((SV *)av); fail: - SvREFCNT_dec (av); + SvREFCNT_dec_NN (av); DEC_DEC_DEPTH; return &PL_sv_undef; } @@ -802,7 +992,7 @@ dec->cur += len; if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) - if (!is_utf8_string (key, len)) + if (!cbor_is_utf8_string ((U8 *)key, len)) ERR ("corrupted CBOR data (invalid UTF-8 in map key)"); hv_store (hv, key, -len, decode_sv (dec), 0); @@ -813,8 +1003,36 @@ 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: ; @@ -835,7 +1053,7 @@ { WANT (1); - if (*dec->cur == (MAJOR_MISC | MINOR_INDEF)) + if (*dec->cur == (MAJOR_MISC | MINOR_INDEF) || dec->err) { ++dec->cur; break; @@ -858,7 +1076,7 @@ return newRV_noinc ((SV *)hv); fail: - SvREFCNT_dec (hv); + SvREFCNT_dec_NN (hv); DEC_DEC_DEPTH; return &PL_sv_undef; } @@ -868,7 +1086,7 @@ { 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; @@ -913,7 +1131,7 @@ if (utf8) { if (ecb_expect_false (dec->cbor.flags & F_VALIDATE_UTF8)) - if (!is_utf8_string (SvPVX (sv), SvCUR (sv))) + if (!cbor_is_utf8_string (SvPVX (sv), SvCUR (sv))) ERR ("corrupted CBOR data (invalid UTF-8 in text string)"); SvUTF8_on (sv); @@ -946,7 +1164,7 @@ case CBOR_TAG_STRINGREF_NAMESPACE: { - // do nmot use SAVETMPS/FREETMPS, as these will + // do not use SAVETMPS/FREETMPS, as these will // erase mortalised caches, e.g. "shareable" ENTER; @@ -966,7 +1184,7 @@ 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]); @@ -1004,7 +1222,7 @@ 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]); @@ -1016,6 +1234,9 @@ 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) @@ -1054,10 +1275,10 @@ 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; @@ -1067,6 +1288,7 @@ break; default: + filter: { SV *tag_sv = newSVuv (tag); @@ -1074,7 +1296,6 @@ dSP; ENTER; SAVETMPS; - SAVESTACK_POS (); PUSHMARK (SP); EXTEND (SP, 2); PUSHs (tag_sv); @@ -1086,16 +1307,17 @@ if (SvTRUE (ERRSV)) { - SvREFCNT_dec (tag_sv); + SvREFCNT_dec_NN (tag_sv); FREETMPS; LEAVE; - ERR (SvPVutf8_nolen (sv_2mortal (SvREFCNT_inc (ERRSV)))); + ERR_ERRSV; } if (count) { - SvREFCNT_dec (tag_sv); - SvREFCNT_dec (sv); - sv = SvREFCNT_inc (POPs); + SvREFCNT_dec_NN (tag_sv); + SvREFCNT_dec_NN (sv); + sv = SvREFCNT_inc_NN (TOPs); + SP -= count; } else { @@ -1248,7 +1470,11 @@ 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); } @@ -1348,6 +1574,7 @@ case MAJOR_MAP >> MAJOR_SHIFT: len <<= 1; + /* FALLTHROUGH */ case MAJOR_ARRAY >> MAJOR_SHIFT: if (len) { @@ -1433,6 +1660,7 @@ 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 text_keys = F_TEXT_KEYS text_strings = F_TEXT_STRINGS @@ -1453,6 +1681,7 @@ 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 @@ -1595,3 +1824,19 @@ 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 +