X-Git-Url: https://perl5.git.perl.org/perl5.git/blobdiff_plain/2e0b8fbeab3502bee36f25825c3cdd0d075c4fd3..c7c8bf5547526831b6168a1abd107399d80c0991:/sv.h diff --git a/sv.h b/sv.h index 475da61..abcfbb6 100644 --- a/sv.h +++ b/sv.h @@ -57,7 +57,8 @@ typeglob has been assigned. Assigning to it again will stop it from being a typeglob. SVt_PVLV represents a scalar that delegates to another scalar behind the scenes. It is used, e.g., for the return value of C and for tied hash and array elements. It can hold any scalar value, including -a typeglob. SVt_REGEXP is for regular expressions. SVt_INVLIST is for Perl +a typeglob. SVt_REGEXP is for regular +expressions. SVt_INVLIST is for Perl core internal use only. SVt_PVMG represents a "normal" scalar (not a typeglob, regular expression, @@ -120,8 +121,14 @@ Type flag for formats. See L. Type flag for I/O objects. See L. =cut + + These are ordered so that the simpler types have a lower value; SvUPGRADE + doesn't allow you to upgrade from a higher numbered type to a lower numbered + one; also there is code that assumes that anything that has as a PV component + has a type numbered >= SVt_PV. */ + typedef enum { SVt_NULL, /* 0 */ /* BIND was here, before INVLIST replaced it. */ @@ -154,7 +161,7 @@ typedef enum { #ifndef PERL_CORE /* Although Fast Boyer Moore tables are now being stored in PVGVs, for most - purposes eternal code wanting to consider PVBM probably needs to think of + purposes external code wanting to consider PVBM probably needs to think of PVMG instead. */ # define SVt_PVBM SVt_PVMG /* Anything wanting to create a reference from clean should ensure that it has @@ -291,7 +298,7 @@ to return a meaningful value, or check for NULLness, so it's smaller and faster. =for apidoc Am|void|SvREFCNT_dec|SV* sv -Decrements the reference count of the given SV. I may be NULL. +Decrements the reference count of the given SV. I may be NULL. =for apidoc Am|void|SvREFCNT_dec_NN|SV* sv Same as SvREFCNT_dec, but can only be used if you know I @@ -354,6 +361,7 @@ perform the upgrade if necessary. See C. subroutine in another package. Set the GvIMPORTED_CV_on() if it needs to be expanded to a real GV */ +#define SVpad_NAMELIST SVp_SCREAM /* AV is a padnamelist */ #define SVf_IsCOW 0x00010000 /* copy on write (shared hash key if SvLEN == 0) */ #define SVs_PADTMP 0x00020000 /* in use as tmp; only if ! SVs_PADMY */ @@ -397,21 +405,24 @@ perform the upgrade if necessary. See C. #define SVf_AMAGIC 0x10000000 /* has magical overloaded methods */ +/* note that SVf_AMAGIC is now only set on stashes, so this bit is free + * for non-HV SVs */ + /* Ensure this value does not clash with the GV_ADD* flags in gv.h: */ #define SVf_UTF8 0x20000000 /* SvPV is UTF-8 encoded This is also set on RVs whose overloaded stringification is UTF-8. This might only happen as a side effect of SvPV() */ - - -/* Some private flags. */ +/* PVHV */ +#define SVphv_SHAREKEYS 0x20000000 /* PVHV keys live on shared string table */ /* PVAV could probably use 0x2000000 without conflict. I assume that PVFM can be UTF-8 encoded, and PVCVs could well have UTF-8 prototypes. PVIOs haven't been restructured, so sometimes get used as string buffers. */ -/* PVHV */ -#define SVphv_SHAREKEYS 0x20000000 /* PVHV keys live on shared string table */ + +/* Some private flags. */ + /* PVNV, PVMG only, and only used in pads. Should be safe to test on any scalar SV, as the core is careful to avoid setting both. @@ -470,7 +481,6 @@ union _xnvu { U32 xlow; U32 xhigh; } xpad_cop_seq; /* used by pad.c for cop_sequence */ - I32 xbm_useful; }; union _xivu { @@ -483,7 +493,7 @@ union _xmgu { MAGIC* xmg_magic; /* linked list of magicalness */ HV* xmg_ourstash; /* Stash for our (when SvPAD_OUR is true) */ STRLEN xmg_hash_index; /* used while freeing hash entries */ -}; +}; /* also used by PadnamelistMAXNAMED */ struct xpv { _XPV_HEAD; @@ -520,7 +530,10 @@ struct xpvlv { _XPV_HEAD; union _xivu xiv_u; union _xnvu xnv_u; - STRLEN xlv_targoff; + union { + STRLEN xlvu_targoff; + SSize_t xlvu_stargoff; + } xlv_targoff_u; STRLEN xlv_targlen; SV* xlv_targ; char xlv_type; /* k=keys .=pos x=substr v=vec /=join/re @@ -528,6 +541,15 @@ struct xpvlv { char xlv_flags; /* 1 = negative offset 2 = negative len */ }; +#define xlv_targoff xlv_targoff_u.xlvu_targoff + +struct xpvinvlist { + _XPV_HEAD; + IV prev_index; + STRLEN iterator; + bool is_offset; /* */ +}; + /* This structure works in 3 ways - regular scalar, GV with GP, or fast Boyer-Moore. */ struct xpvgv { @@ -628,7 +650,7 @@ double. Checks the B setting. Use C instead. Unsets the NV/IV status of an SV. =for apidoc Am|U32|SvOK|SV* sv -Returns a U32 value indicating whether the value is defined. This is +Returns a U32 value indicating whether the value is defined. This is only meaningful for scalars. =for apidoc Am|U32|SvIOKp|SV* sv @@ -734,7 +756,8 @@ Only use when you are sure SvNOK is true. See also C. =for apidoc Am|char*|SvPVX|SV* sv Returns a pointer to the physical string in the SV. The SV must contain a -string. Prior to 5.9.3 it is not safe to execute this macro unless the SV's +string. Prior to 5.9.3 it is not safe +to execute this macro unless the SV's type >= SVt_PV. This is also used to store the name of an autoloaded subroutine in an XS @@ -750,7 +773,7 @@ attributable to C. See C. =for apidoc Am|char*|SvEND|SV* sv Returns a pointer to the spot just after the last character in the string which is in the SV, where there is usually a trailing -null (even though Perl scalars do not strictly require it). +C byte (even though Perl scalars do not strictly require it). See C. Access the character as *(SvEND(sv)). Warning: If C is equal to C, then C points to @@ -936,6 +959,7 @@ in gv.h: */ #define HvAMAGIC_off(hv) (SvFLAGS(hv) &=~ SVf_AMAGIC) +/* "nog" means "doesn't have get magic" */ #define SvPOK_nog(sv) ((SvFLAGS(sv) & (SVf_POK|SVs_GMG)) == SVf_POK) #define SvIOK_nog(sv) ((SvFLAGS(sv) & (SVf_IOK|SVs_GMG)) == SVf_IOK) #define SvUOK_nog(sv) ((SvFLAGS(sv) & (SVf_IOK|SVf_IVisUV|SVs_GMG)) == (SVf_IOK|SVf_IVisUV)) @@ -1060,27 +1084,30 @@ sv_force_normal does nothing. #if defined (DEBUGGING) && defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) # define SvVALID(sv) ({ const SV *const _svvalid = (const SV*)(sv); \ - if (SvFLAGS(_svvalid) & SVpbm_VALID) \ + if (SvFLAGS(_svvalid) & SVpbm_VALID && !SvSCREAM(_svvalid)) \ assert(!isGV_with_GP(_svvalid)); \ (SvFLAGS(_svvalid) & SVpbm_VALID); \ }) # define SvVALID_on(sv) ({ SV *const _svvalid = MUTABLE_SV(sv); \ assert(!isGV_with_GP(_svvalid)); \ + assert(!SvSCREAM(_svvalid)); \ (SvFLAGS(_svvalid) |= SVpbm_VALID); \ }) # define SvVALID_off(sv) ({ SV *const _svvalid = MUTABLE_SV(sv); \ assert(!isGV_with_GP(_svvalid)); \ + assert(!SvSCREAM(_svvalid)); \ (SvFLAGS(_svvalid) &= ~SVpbm_VALID); \ }) # define SvTAIL(sv) ({ const SV *const _svtail = (const SV *)(sv); \ - assert(SvTYPE(_svtail) != SVt_PVAV); \ - assert(SvTYPE(_svtail) != SVt_PVHV); \ + assert(SvTYPE(_svtail) != SVt_PVAV); \ + assert(SvTYPE(_svtail) != SVt_PVHV); \ + assert(!SvSCREAM(_svtail)); \ (SvFLAGS(sv) & (SVpbm_TAIL|SVpbm_VALID)) \ == (SVpbm_TAIL|SVpbm_VALID); \ }) #else -# define SvVALID(sv) (SvFLAGS(sv) & SVpbm_VALID) +# define SvVALID(sv) ((SvFLAGS(sv) & SVpbm_VALID) && !SvSCREAM(sv)) # define SvVALID_on(sv) (SvFLAGS(sv) |= SVpbm_VALID) # define SvVALID_off(sv) (SvFLAGS(sv) &= ~SVpbm_VALID) # define SvTAIL(sv) ((SvFLAGS(sv) & (SVpbm_TAIL|SVpbm_VALID)) \ @@ -1335,7 +1362,7 @@ sv_force_normal does nothing. assert(SvTYPE(sv) >= SVt_PV); \ if (SvLEN(sv)) { \ assert(!SvROK(sv)); \ - if(SvOOK(sv)) { \ + if(UNLIKELY(SvOOK(sv))) { \ STRLEN zok; \ SvOOK_offset(sv, zok); \ SvPV_set(sv, SvPVX_mutable(sv) - zok); \ @@ -1367,13 +1394,13 @@ sv_force_normal does nothing. #if defined (DEBUGGING) && defined(__GNUC__) && !defined(PERL_GCC_BRACE_GROUPS_FORBIDDEN) # define BmUSEFUL(sv) \ (*({ SV *const _bmuseful = MUTABLE_SV(sv); \ - assert(SvTYPE(_bmuseful) == SVt_PVMG); \ + assert(SvTYPE(_bmuseful) >= SVt_PVIV); \ assert(SvVALID(_bmuseful)); \ - assert(!SvNOK(_bmuseful)); \ - &(((XPVMG*) SvANY(_bmuseful))->xnv_u.xbm_useful); \ + assert(!SvIOK(_bmuseful)); \ + &(((XPVIV*) SvANY(_bmuseful))->xiv_u.xivu_iv); \ })) #else -# define BmUSEFUL(sv) ((XPVMG*) SvANY(sv))->xnv_u.xbm_useful +# define BmUSEFUL(sv) ((XPVIV*) SvANY(sv))->xiv_u.xivu_iv #endif @@ -1387,6 +1414,7 @@ sv_force_normal does nothing. #define LvTYPE(sv) ((XPVLV*) SvANY(sv))->xlv_type #define LvTARG(sv) ((XPVLV*) SvANY(sv))->xlv_targ #define LvTARGOFF(sv) ((XPVLV*) SvANY(sv))->xlv_targoff +#define LvSTARGOFF(sv) ((XPVLV*) SvANY(sv))->xlv_targoff_u.xlvu_stargoff #define LvTARGLEN(sv) ((XPVLV*) SvANY(sv))->xlv_targlen #define LvFLAGS(sv) ((XPVLV*) SvANY(sv))->xlv_flags @@ -1431,7 +1459,7 @@ Marks an SV as tainted if tainting is enabled. Untaints an SV. Be I careful with this routine, as it short-circuits some of Perl's fundamental security features. XS module authors should not use this function unless they fully understand all the implications of -unconditionally untainting the value. Untainting should be done in the +unconditionally untainting the value. Untainting should be done in the standard perl fashion, via a carefully crafted regexp, rather than directly untainting variables. @@ -1447,18 +1475,18 @@ attention to precisely which outputs are influenced by which inputs. #define sv_taint(sv) sv_magic((sv), NULL, PERL_MAGIC_taint, NULL, 0) -#if NO_TAINT_SUPPORT +#ifdef NO_TAINT_SUPPORT # define SvTAINTED(sv) 0 #else # define SvTAINTED(sv) (SvMAGICAL(sv) && sv_tainted(sv)) #endif -#define SvTAINTED_on(sv) STMT_START{ if(TAINTING_get){sv_taint(sv);} }STMT_END -#define SvTAINTED_off(sv) STMT_START{ if(TAINTING_get){sv_untaint(sv);} }STMT_END +#define SvTAINTED_on(sv) STMT_START{ if(UNLIKELY(TAINTING_get)){sv_taint(sv);} }STMT_END +#define SvTAINTED_off(sv) STMT_START{ if(UNLIKELY(TAINTING_get)){sv_untaint(sv);} }STMT_END #define SvTAINT(sv) \ STMT_START { \ - if (TAINTING_get) { \ - if (TAINT_get) \ + if (UNLIKELY(TAINTING_get)) { \ + if (UNLIKELY(TAINT_get)) \ SvTAINTED_on(sv); \ } \ } STMT_END @@ -1470,7 +1498,7 @@ only a string (C), by hook or by crook. You need force if you are going to update the C directly. Processes get magic. Note that coercing an arbitrary scalar into a plain PV will potentially -strip useful data from it. For example if the SV was C, then the +strip useful data from it. For example if the SV was C, then the referent will have its reference count decremented, and the SV itself may be converted to an C scalar with a string buffer containing a value such as C<"ARRAY(0x1234)">. @@ -1481,13 +1509,15 @@ Like C, but doesn't process get magic. =for apidoc Am|char*|SvPV|SV* sv|STRLEN len Returns a pointer to the string in the SV, or a stringified form of the SV if the SV does not contain a string. The SV may cache the -stringified version becoming C. Handles 'get' magic. See also -C for a version which guarantees to evaluate sv only once. +stringified version becoming C. Handles 'get' magic. The +C variable will be set to the length of the string (this is a macro, so +don't use C<&len>). See also C for a version which guarantees to +evaluate sv only once. Note that there is no guarantee that the return value of C is equal to C, or that C contains valid data, or that -successive calls to C will return the same pointer value each +time. This is due to the way that things like overloading and Copy-On-Write are handled. In these cases, the return value may point to a temporary buffer or similar. If you absolutely need the SvPVX field to be valid (for example, if you intend to write to it), then see @@ -1594,8 +1624,8 @@ Like C, but converts sv to byte representation first if necessary. Guarantees to evaluate sv only once; use the more efficient C otherwise. -=for apidoc Am|bool|SvIsCOW|SV* sv -Returns a boolean indicating whether the SV is Copy-On-Write (either shared +=for apidoc Am|U32|SvIsCOW|SV* sv +Returns a U32 value indicating whether the SV is Copy-On-Write (either shared hash key scalars, or full Copy On Write scalars if 5.9.0 is configured for COW). @@ -1677,6 +1707,7 @@ Like sv_utf8_upgrade, but doesn't do magic on C. (SvPOK_nog(sv) \ ? SvPVX(sv) : sv_2pv_flags(sv, 0, SV_GMAGIC)) +/* "_nomg" in these defines means no mg_get() */ #define SvPV_nomg_nolen(sv) \ (SvPOK_nog(sv) \ ? SvPVX(sv) : sv_2pv_flags(sv, 0, 0)) @@ -1727,9 +1758,9 @@ Like sv_utf8_upgrade, but doesn't do magic on C. #define SvPVutf8x_force(sv, lp) sv_pvutf8n_force(sv, &lp) #define SvPVbytex_force(sv, lp) sv_pvbyten_force(sv, &lp) -#define SvTRUE(sv) ((sv) && (SvGMAGICAL(sv) ? sv_2bool(sv) : SvTRUE_common(sv, sv_2bool_nomg(sv)))) -#define SvTRUE_nomg(sv) ((sv) && ( SvTRUE_common(sv, sv_2bool_nomg(sv)))) -#define SvTRUE_NN(sv) (SvGMAGICAL(sv) ? sv_2bool(sv) : SvTRUE_common(sv, sv_2bool_nomg(sv))) +#define SvTRUE(sv) (LIKELY(sv) && (UNLIKELY(SvGMAGICAL(sv)) ? sv_2bool(sv) : SvTRUE_common(sv, sv_2bool_nomg(sv)))) +#define SvTRUE_nomg(sv) (LIKELY(sv) && ( SvTRUE_common(sv, sv_2bool_nomg(sv)))) +#define SvTRUE_NN(sv) (UNLIKELY(SvGMAGICAL(sv)) ? sv_2bool(sv) : SvTRUE_common(sv, sv_2bool_nomg(sv))) #define SvTRUE_nomg_NN(sv) ( SvTRUE_common(sv, sv_2bool_nomg(sv))) #define SvTRUE_common(sv,fallback) ( \ !SvOK(sv) \ @@ -1880,12 +1911,6 @@ mg.c:1024: warning: left-hand operand of comma expression has no effect /* Note: To allow 256 COW "copies", a refcnt of 0 means 1. */ # define CowREFCNT(sv) (*(U8 *)(SvPVX(sv)+SvLEN(sv)-1)) # define SV_COW_REFCNT_MAX ((1 << sizeof(U8)*8) - 1) -# ifndef SV_COW_THRESHOLD -# define SV_COW_THRESHOLD 0 /* min string length for cow */ -# endif -# ifndef SV_COWBUF_THRESHOLD -# define SV_COWBUF_THRESHOLD 1250 /* min string length for cow */ -# endif /* over existing buffer */ # define CAN_COW_MASK (SVf_POK|SVf_ROK|SVp_POK|SVf_FAKE| \ SVf_OOK|SVf_BREAK|SVf_READONLY) # endif @@ -1954,12 +1979,11 @@ mg.c:1024: warning: left-hand operand of comma expression has no effect #define sv_catpvn_nomg_maybeutf8(dsv, sstr, slen, is_utf8) \ sv_catpvn_flags(dsv, sstr, slen, (is_utf8)?SV_CATUTF8:SV_CATBYTES) -#ifdef PERL_CORE +#if defined(PERL_CORE) || defined(PERL_EXT) # define sv_or_pv_len_utf8(sv, pv, bytelen) \ (SvGAMAGIC(sv) \ ? utf8_length((U8 *)(pv), (U8 *)(pv)+(bytelen)) \ : sv_len_utf8(sv)) -# define sv_or_pv_pos_u2b(sv,s,p,lp) S_sv_or_pv_pos_u2b(aTHX_ sv,s,p,lp) #endif /* @@ -1989,7 +2013,7 @@ after modifying a scalar, in case it is a magical variable like C<$|> or a tied variable (it calls C). This macro evaluates its argument more than once. -=for apidoc Am|void|SvSetSV|SV* dsb|SV* ssv +=for apidoc Am|void|SvSetSV|SV* dsv|SV* ssv Calls C if dsv is not the same as ssv. May evaluate arguments more than once. @@ -1997,7 +2021,7 @@ more than once. Calls a non-destructive version of C if dsv is not the same as ssv. May evaluate arguments more than once. -=for apidoc Am|void|SvSetMagicSV|SV* dsb|SV* ssv +=for apidoc Am|void|SvSetMagicSV|SV* dsv|SV* ssv Like C, but does any set magic required afterwards. =for apidoc Am|void|SvSetMagicSV_nosteal|SV* dsv|SV* ssv @@ -2020,8 +2044,9 @@ has been loaded. =for apidoc Am|char *|SvGROW|SV* sv|STRLEN len Expands the character buffer in the SV so that it has room for the indicated number of bytes (remember to reserve space for an extra trailing -NUL character). Calls C to perform the expansion if necessary. -Returns a pointer to the character buffer. SV must be of type >= SVt_PV. One +C character). Calls C to perform the expansion if necessary. +Returns a pointer to the character +buffer. SV must be of type >= SVt_PV. One alternative is to call C if you are not sure of the type of SV. =cut @@ -2032,19 +2057,19 @@ alternative is to call C if you are not sure of the type of SV. #define SvUNLOCK(sv) PL_unlockhook(aTHX_ sv) #define SvDESTROYABLE(sv) PL_destroyhook(aTHX_ sv) -#define SvGETMAGIC(x) ((void)(SvGMAGICAL(x) && mg_get(x))) -#define SvSETMAGIC(x) STMT_START { if (SvSMAGICAL(x)) mg_set(x); } STMT_END +#define SvGETMAGIC(x) ((void)(UNLIKELY(SvGMAGICAL(x)) && mg_get(x))) +#define SvSETMAGIC(x) STMT_START { if (UNLIKELY(SvSMAGICAL(x))) mg_set(x); } STMT_END #define SvSetSV_and(dst,src,finally) \ STMT_START { \ - if ((dst) != (src)) { \ + if (LIKELY((dst) != (src))) { \ sv_setsv(dst, src); \ finally; \ } \ } STMT_END #define SvSetSV_nosteal_and(dst,src,finally) \ STMT_START { \ - if ((dst) != (src)) { \ + if (LIKELY((dst) != (src))) { \ sv_setsv_flags(dst, src, SV_GMAGIC | SV_NOSTEAL | SV_DO_COW_SVSETSV); \ finally; \ } \ @@ -2112,7 +2137,7 @@ See also C and C. == (SVt_PVLV|SVf_FAKE)) -#ifdef PERL_NEW_COPY_ON_WRITE +#ifdef PERL_ANY_COW # define SvGROW(sv,len) \ (SvIsCOW(sv) || SvLEN(sv) < (len) ? sv_grow(sv,len) : SvPVX(sv)) #else @@ -2138,7 +2163,8 @@ struct clone_params { /* =for apidoc Am|SV*|newSVpvn_utf8|NULLOK const char* s|STRLEN len|U32 utf8 -Creates a new SV and copies a string into it. If utf8 is true, calls +Creates a new SV and copies a string (which may contain C (C<\0>) +characters) into it. If utf8 is true, calls C on the new SV. Implemented as a wrapper around C. =cut