MAN page from Old RedHat 5.X perl-5.004-4.i386.rpm

PERLGUTS

Section: Perl Programmers Reference Guide (1)
Updated: perl 5.004, patch 04
Index 

NAME

perlguts - Perl's Internal Functions 

DESCRIPTION

This document attempts to describe some of the internal functions of thePerl executable. It is far from complete and probably contains many errors.Please refer any questions or comments to the author below. 

Variables

 

Datatypes

Perl has three typedefs that handle Perl's three main data types:

    SV  Scalar Value    AV  Array Value    HV  Hash Value

Each typedef has specific routines that manipulate the various data types. 

What is an IV?

Perl uses a special typedef IV which is a simple integer type that isguaranteed to be large enough to hold a pointer (as well as an integer).

Perl also uses two special typedefs, I32 and I16, which will always be atleast 32-bits and 16-bits long, respectively. 

Working with SVs

An SV can be created and loaded with one command. There are four types ofvalues that can be loaded: an integer value (IV), a double (NV), a string,(PV), and another scalar (SV).

The five routines are:

    SV*  newSViv(IV);    SV*  newSVnv(double);    SV*  newSVpv(char*, int);    SV*  newSVpvf(const char*, ...);    SV*  newSVsv(SV*);

To change the value of an *already-existing* SV, there are six routines:

    void  sv_setiv(SV*, IV);    void  sv_setnv(SV*, double);    void  sv_setpv(SV*, char*);    void  sv_setpvn(SV*, char*, int)    void  sv_setpvf(SV*, const char*, ...);    void  sv_setsv(SV*, SV*);

Notice that you can choose to specify the length of the string to beassigned by using sv_setpvn or newSVpv, or you may allow Perl tocalculate the length by using sv_setpv or by specifying 0 as the secondargument to newSVpv. Be warned, though, that Perl will determine thestring's length by using strlen, which depends on the string terminatingwith a NUL character. The arguments of sv_setpvf are processed likesprintf, and the formatted output becomes the value.

All SVs that will contain strings should, but need not, be terminatedwith a NUL character. If it is not NUL-terminated there is a risk ofcore dumps and corruptions from code which passes the string to Cfunctions or system calls which expect a NUL-terminated string.Perl's own functions typically add a trailing NUL for this reason.Nevertheless, you should be very careful when you pass a string storedin an SV to a C function or system call.

To access the actual value that an SV points to, you can use the macros:

    SvIV(SV*)    SvNV(SV*)    SvPV(SV*, STRLEN len)

which will automatically coerce the actual scalar type into an IV, double,or string.

In the SvPV macro, the length of the string returned is placed into thevariable len (this is a macro, so you do not use &len). If you do notcare what the length of the data is, use the global variable na. Remember,however, that Perl allows arbitrary strings of data that may both containNULs and might not be terminated by a NUL.

If you want to know if the scalar value is TRUE, you can use:

    SvTRUE(SV*)

Although Perl will automatically grow strings for you, if you need to forcePerl to allocate more memory for your SV, you can use the macro

    SvGROW(SV*, STRLEN newlen)

which will determine if more memory needs to be allocated. If so, it willcall the function sv_grow. Note that SvGROW can only increase, notdecrease, the allocated memory of an SV and that it does not automaticallyadd a byte for the a trailing NUL (perl's own string functions typically doSvGROW(sv, len + 1)).

If you have an SV and want to know what kind of data Perl thinks is storedin it, you can use the following macros to check the type of SV you have.

    SvIOK(SV*)    SvNOK(SV*)    SvPOK(SV*)

You can get and set the current length of the string stored in an SV withthe following macros:

    SvCUR(SV*)    SvCUR_set(SV*, I32 val)

You can also get a pointer to the end of the string stored in the SVwith the macro:

    SvEND(SV*)

But note that these last three macros are valid only if SvPOK() is true.

If you want to append something to the end of string stored in an SV*,you can use the following functions:

    void  sv_catpv(SV*, char*);    void  sv_catpvn(SV*, char*, int);    void  sv_catpvf(SV*, const char*, ...);    void  sv_catsv(SV*, SV*);

The first function calculates the length of the string to be appended byusing strlen. In the second, you specify the length of the stringyourself. The third function processes its arguments like sprintf andappends the formatted output. The fourth function extends the stringstored in the first SV with the string stored in the second SV. It alsoforces the second SV to be interpreted as a string.

If you know the name of a scalar variable, you can get a pointer to its SVby using the following:

    SV*  perl_get_sv("package::varname", FALSE);

This returns NULL if the variable does not exist.

If you want to know if this variable (or any other SV) is actually defined,you can call:

    SvOK(SV*)

The scalar undef value is stored in an SV instance called sv_undef. Itsaddress can be used whenever an SV* is needed.

There are also the two values sv_yes and sv_no, which contain BooleanTRUE and FALSE values, respectively. Like sv_undef, their addresses canbe used whenever an SV* is needed.

Do not be fooled into thinking that (SV *) 0 is the same as &sv_undef.Take this code:

    SV* sv = (SV*) 0;    if (I-am-to-return-a-real-value) {            sv = sv_2mortal(newSViv(42));    }    sv_setsv(ST(0), sv);

This code tries to return a new SV (which contains the value 42) if it shouldreturn a real value, or undef otherwise. Instead it has returned a NULLpointer which, somewhere down the line, will cause a segmentation violation,bus error, or just weird results. Change the zero to &sv_undef in the firstline and all will be well.

To free an SV that you've created, call SvREFCNT_dec(SV*). Normally thiscall is not necessary (see the section on Reference Counts and Mortality). 

What's Really Stored in an SV?

Recall that the usual method of determining the type of scalar you have isto use Sv*OK macros. Because a scalar can be both a number and a string,usually these macros will always return TRUE and calling the Sv*Vmacros will do the appropriate conversion of string to integer/double orinteger/double to string.

If you really need to know if you have an integer, double, or stringpointer in an SV, you can use the following three macros instead:

    SvIOKp(SV*)    SvNOKp(SV*)    SvPOKp(SV*)

These will tell you if you truly have an integer, double, or string pointerstored in your SV. The ``p'' stands for private.

In general, though, it's best to use the Sv*V macros. 

Working with AVs

There are two ways to create and load an AV. The first method creates anempty AV:

    AV*  newAV();

The second method both creates the AV and initially populates it with SVs:

    AV*  av_make(I32 num, SV **ptr);

The second argument points to an array containing num SV*'s. Once theAV has been created, the SVs can be destroyed, if so desired.

Once the AV has been created, the following operations are possible on AVs:

    void  av_push(AV*, SV*);    SV*   av_pop(AV*);    SV*   av_shift(AV*);    void  av_unshift(AV*, I32 num);

These should be familiar operations, with the exception of av_unshift.This routine adds num elements at the front of the array with the undefvalue. You must then use av_store (described below) to assign valuesto these new elements.

Here are some other functions:

    I32   av_len(AV*);    SV**  av_fetch(AV*, I32 key, I32 lval);    SV**  av_store(AV*, I32 key, SV* val);

The av_len function returns the highest index value in array (justlike $#array in Perl). If the array is empty, -1 is returned. Theav_fetch function returns the value at index key, but if lvalis non-zero, then av_fetch will store an undef value at that index.The av_store function stores the value val at index key, and doesnot increment the reference count of val. Thus the caller is responsiblefor taking care of that, and if av_store returns NULL, the caller willhave to decrement the reference count to avoid a memory leak. Note thatav_fetch and av_store both return SV**'s, not SV*'s as theirreturn value.

    void  av_clear(AV*);    void  av_undef(AV*);    void  av_extend(AV*, I32 key);

The av_clear function deletes all the elements in the AV* array, butdoes not actually delete the array itself. The av_undef function willdelete all the elements in the array plus the array itself. Theav_extend function extends the array so that it contains keyelements. If key is less than the current length of the array, thennothing is done.

If you know the name of an array variable, you can get a pointer to its AVby using the following:

    AV*  perl_get_av("package::varname", FALSE);

This returns NULL if the variable does not exist.

See the section on Understanding the Magic of Tied Hashes and Arrays for moreinformation on how to use the array access functions on tied arrays. 

Working with HVs

To create an HV, you use the following routine:

    HV*  newHV();

Once the HV has been created, the following operations are possible on HVs:

    SV**  hv_store(HV*, char* key, U32 klen, SV* val, U32 hash);    SV**  hv_fetch(HV*, char* key, U32 klen, I32 lval);

The klen parameter is the length of the key being passed in (Note thatyou cannot pass 0 in as a value of klen to tell Perl to measure thelength of the key). The val argument contains the SV pointer to thescalar being stored, and hash is the precomputed hash value (zero ifyou want hv_store to calculate it for you). The lval parameterindicates whether this fetch is actually a part of a store operation, inwhich case a new undefined value will be added to the HV with the suppliedkey and hv_fetch will return as if the value had already existed.

Remember that hv_store and hv_fetch return SV**'s and not justSV*. To access the scalar value, you must first dereference the returnvalue. However, you should check to make sure that the return value isnot NULL before dereferencing it.

These two functions check if a hash table entry exists, and deletes it.

    bool  hv_exists(HV*, char* key, U32 klen);    SV*   hv_delete(HV*, char* key, U32 klen, I32 flags);

If flags does not include the G_DISCARD flag then hv_delete willcreate and return a mortal copy of the deleted value.

And more miscellaneous functions:

    void   hv_clear(HV*);    void   hv_undef(HV*);

Like their AV counterparts, hv_clear deletes all the entries in the hashtable but does not actually delete the hash table. The hv_undef deletesboth the entries and the hash table itself.

Perl keeps the actual data in linked list of structures with a typedef of HE.These contain the actual key and value pointers (plus extra administrativeoverhead). The key is a string pointer; the value is an SV*. However,once you have an HE*, to get the actual key and value, use the routinesspecified below.

    I32    hv_iterinit(HV*);            /* Prepares starting point to traverse hash table */    HE*    hv_iternext(HV*);            /* Get the next entry, and return a pointer to a               structure that has both the key and value */    char*  hv_iterkey(HE* entry, I32* retlen);            /* Get the key from an HE structure and also return               the length of the key string */    SV*    hv_iterval(HV*, HE* entry);            /* Return a SV pointer to the value of the HE               structure */    SV*    hv_iternextsv(HV*, char** key, I32* retlen);            /* This convenience routine combines hv_iternext,               hv_iterkey, and hv_iterval.  The key and retlen               arguments are return values for the key and its               length.  The value is returned in the SV* argument */

If you know the name of a hash variable, you can get a pointer to its HVby using the following:

    HV*  perl_get_hv("package::varname", FALSE);

This returns NULL if the variable does not exist.

The hash algorithm is defined in the PERL_HASH(hash, key, klen) macro:

    i = klen;    hash = 0;    s = key;    while (i--)        hash = hash * 33 + *s++;

See the section on Understanding the Magic of Tied Hashes and Arrays for moreinformation on how to use the hash access functions on tied hashes. 

Hash API Extensions

Beginning with version 5.004, the following functions are also supported:

    HE*     hv_fetch_ent  (HV* tb, SV* key, I32 lval, U32 hash);    HE*     hv_store_ent  (HV* tb, SV* key, SV* val, U32 hash);        bool    hv_exists_ent (HV* tb, SV* key, U32 hash);    SV*     hv_delete_ent (HV* tb, SV* key, I32 flags, U32 hash);        SV*     hv_iterkeysv  (HE* entry);

Note that these functions take SV* keys, which simplifies writingof extension code that deals with hash structures. These functionsalso allow passing of SV* keys to tie functions without forcingyou to stringify the keys (unlike the previous set of functions).

They also return and accept whole hash entries (HE*), making theiruse more efficient (since the hash number for a particular stringdoesn't have to be recomputed every time). See the section on API LISTING later inthis document for detailed descriptions.

The following macros must always be used to access the contents of hashentries. Note that the arguments to these macros must be simplevariables, since they may get evaluated more than once. Seethe section on API LISTING later in this document for detailed descriptions of thesemacros.

    HePV(HE* he, STRLEN len)    HeVAL(HE* he)    HeHASH(HE* he)    HeSVKEY(HE* he)    HeSVKEY_force(HE* he)    HeSVKEY_set(HE* he, SV* sv)

These two lower level macros are defined, but must only be used whendealing with keys that are not SV*s:

    HeKEY(HE* he)    HeKLEN(HE* he)

Note that both hv_store and hv_store_ent do not increment thereference count of the stored val, which is the caller's responsibility.If these functions return a NULL value, the caller will usually have todecrement the reference count of val to avoid a memory leak. 

References

References are a special type of scalar that point to other data types(including references).

To create a reference, use either of the following functions:

    SV* newRV_inc((SV*) thing);    SV* newRV_noinc((SV*) thing);

The thing argument can be any of an SV*, AV*, or HV*. Thefunctions are identical except that newRV_inc increments the referencecount of the thing, while newRV_noinc does not. For historicalreasons, newRV is a synonym for newRV_inc.

Once you have a reference, you can use the following macro to dereferencethe reference:

    SvRV(SV*)

then call the appropriate routines, casting the returned SV* to either anAV* or HV*, if required.

To determine if an SV is a reference, you can use the following macro:

    SvROK(SV*)

To discover what type of value the reference refers to, use the followingmacro and then check the return value.

    SvTYPE(SvRV(SV*))

The most useful types that will be returned are:

    SVt_IV    Scalar    SVt_NV    Scalar    SVt_PV    Scalar    SVt_RV    Scalar    SVt_PVAV  Array    SVt_PVHV  Hash    SVt_PVCV  Code    SVt_PVGV  Glob (possible a file handle)    SVt_PVMG  Blessed or Magical Scalar

    See the sv.h header file for more details.

 

Blessed References and Class Objects

References are also used to support object-oriented programming. In theOO lexicon, an object is simply a reference that has been blessed into apackage (or class). Once blessed, the programmer may now use the referenceto access the various methods in the class.

A reference can be blessed into a package with the following function:

    SV* sv_bless(SV* sv, HV* stash);

The sv argument must be a reference. The stash argument specifieswhich class the reference will belong to. Seethe section on Stashes and Globs for information on converting class names into stashes.

/* Still under construction */

Upgrades rv to reference if not already one. Creates new SV for rv topoint to. If classname is non-null, the SV is blessed into the specifiedclass. SV is returned.

        SV* newSVrv(SV* rv, char* classname);

Copies integer or double into an SV whose reference is rv. SV is blessedif classname is non-null.

        SV* sv_setref_iv(SV* rv, char* classname, IV iv);        SV* sv_setref_nv(SV* rv, char* classname, NV iv);

Copies the pointer value (the address, not the string!) into an SV whosereference is rv. SV is blessed if classname is non-null.

        SV* sv_setref_pv(SV* rv, char* classname, PV iv);

Copies string into an SV whose reference is rv. Set length to 0 to letPerl calculate the string length. SV is blessed if classname is non-null.

        SV* sv_setref_pvn(SV* rv, char* classname, PV iv, int length);

        int sv_isa(SV* sv, char* name);        int sv_isobject(SV* sv);

 

Creating New Variables

To create a new Perl variable with an undef value which can be accessed fromyour Perl script, use the following routines, depending on the variable type.

    SV*  perl_get_sv("package::varname", TRUE);    AV*  perl_get_av("package::varname", TRUE);    HV*  perl_get_hv("package::varname", TRUE);

Notice the use of TRUE as the second parameter. The new variable can nowbe set, using the routines appropriate to the data type.

There are additional macros whose values may be bitwise OR'ed with theTRUE argument to enable certain extra features. Those bits are:

    GV_ADDMULTI Marks the variable as multiply defined, thus preventing the                "Name <varname> used only once: possible typo" warning.    GV_ADDWARN  Issues the warning "Had to create <varname> unexpectedly" if                the variable did not exist before the function was called.

If you do not specify a package name, the variable is created in the currentpackage. 

Reference Counts and Mortality

Perl uses an reference count-driven garbage collection mechanism. SVs,AVs, or HVs (xV for short in the following) start their life with areference count of 1. If the reference count of an xV ever drops to 0,then it will be destroyed and its memory made available for reuse.

This normally doesn't happen at the Perl level unless a variable isundef'ed or the last variable holding a reference to it is changed oroverwritten. At the internal level, however, reference counts can bemanipulated with the following macros:

    int SvREFCNT(SV* sv);    SV* SvREFCNT_inc(SV* sv);    void SvREFCNT_dec(SV* sv);

However, there is one other function which manipulates the referencecount of its argument. The newRV_inc function, you will recall,creates a reference to the specified argument. As a side effect,it increments the argument's reference count. If this is not whatyou want, use newRV_noinc instead.

For example, imagine you want to return a reference from an XSUB function.Inside the XSUB routine, you create an SV which initially has a referencecount of one. Then you call newRV_inc, passing it the just-created SV.This returns the reference as a new SV, but the reference count of theSV you passed to newRV_inc has been incremented to two. Now youreturn the reference from the XSUB routine and forget about the SV.But Perl hasn't! Whenever the returned reference is destroyed, thereference count of the original SV is decreased to one and nothing happens.The SV will hang around without any way to access it until Perl itselfterminates. This is a memory leak.

The correct procedure, then, is to use newRV_noinc instead ofnewRV_inc. Then, if and when the last reference is destroyed,the reference count of the SV will go to zero and it will be destroyed,stopping any memory leak.

There are some convenience functions available that can help with thedestruction of xVs. These functions introduce the concept of ``mortality''.An xV that is mortal has had its reference count marked to be decremented,but not actually decremented, until ``a short time later''. Generally theterm ``short time later'' means a single Perl statement, such as a call toan XSUB function. The actual determinant for when mortal xVs have theirreference count decremented depends on two macros, SAVETMPS and FREETMPS.See the perlcall manpage and the perlxs manpage for more details on these macros.

``Mortalization'' then is at its simplest a deferred SvREFCNT_dec.However, if you mortalize a variable twice, the reference count willlater be decremented twice.

You should be careful about creating mortal variables. Strange thingscan happen if you make the same value mortal within multiple contexts,or if you make a variable mortal multiple times.

To create a mortal variable, use the functions:

    SV*  sv_newmortal()    SV*  sv_2mortal(SV*)    SV*  sv_mortalcopy(SV*)

The first call creates a mortal SV, the second converts an existingSV to a mortal SV (and thus defers a call to SvREFCNT_dec), and thethird creates a mortal copy of an existing SV.

The mortal routines are not just for SVs -- AVs and HVs can bemade mortal by passing their address (type-casted to SV*) to thesv_2mortal or sv_mortalcopy routines. 

Stashes and Globs

A ``stash'' is a hash that contains all of the different objects thatare contained within a package. Each key of the stash is a symbolname (shared by all the different types of objects that have the samename), and each value in the hash table is a GV (Glob Value). This GVin turn contains references to the various objects of that name,including (but not limited to) the following:

    Scalar Value    Array Value    Hash Value    File Handle    Directory Handle    Format    Subroutine

There is a single stash called ``defstash'' that holds the items that existin the ``main'' package. To get at the items in other packages, append thestring ``::'' to the package name. The items in the ``Foo'' package are inthe stash ``Foo::'' in defstash. The items in the ``Bar::Baz'' package arein the stash ``Baz::'' in ``Bar::"'s stash.

To get the stash pointer for a particular package, use the function:

    HV*  gv_stashpv(char* name, I32 create)    HV*  gv_stashsv(SV*, I32 create)

The first function takes a literal string, the second uses the string storedin the SV. Remember that a stash is just a hash table, so you get back anHV*. The create flag will create a new package if it is set.

The name that gv_stash*v wants is the name of the package whose symbol tableyou want. The default package is called main. If you have multiply nestedpackages, pass their names to gv_stash*v, separated by :: as in the Perllanguage itself.

Alternately, if you have an SV that is a blessed reference, you can findout the stash pointer by using:

    HV*  SvSTASH(SvRV(SV*));

then use the following to get the package name itself:

    char*  HvNAME(HV* stash);

If you need to bless or re-bless an object you can use the followingfunction:

    SV*  sv_bless(SV*, HV* stash)

where the first argument, an SV*, must be a reference, and the secondargument is a stash. The returned SV* can now be used in the same wayas any other SV.

For more information on references and blessings, consult the perlref manpage. 

Double-Typed SVs

Scalar variables normally contain only one type of value, an integer,double, pointer, or reference. Perl will automatically convert theactual scalar data from the stored type into the requested type.

Some scalar variables contain more than one type of scalar data. Forexample, the variable $! contains either the numeric value of errnoor its string equivalent from either strerror or sys_errlist[].

To force multiple data values into an SV, you must do two things: use thesv_set*v routines to add the additional scalar type, then set a flagso that Perl will believe it contains more than one type of data. Thefour macros to set the flags are:

        SvIOK_on        SvNOK_on        SvPOK_on        SvROK_on

The particular macro you must use depends on which sv_set*v routineyou called first. This is because every sv_set*v routine turns ononly the bit for the particular type of data being set, and turns offall the rest.

For example, to create a new Perl variable called ``dberror'' that containsboth the numeric and descriptive string error values, you could use thefollowing code:

    extern int  dberror;    extern char *dberror_list;

    SV* sv = perl_get_sv("dberror", TRUE);    sv_setiv(sv, (IV) dberror);    sv_setpv(sv, dberror_list[dberror]);    SvIOK_on(sv);

If the order of sv_setiv and sv_setpv had been reversed, then themacro SvPOK_on would need to be called instead of SvIOK_on. 

Magic Variables

[This section still under construction. Ignore everything here. Post nobills. Everything not permitted is forbidden.]

Any SV may be magical, that is, it has special features that a normalSV does not have. These features are stored in the SV structure in alinked list of struct magic's, typedef'ed to MAGIC.

    struct magic {        MAGIC*      mg_moremagic;        MGVTBL*     mg_virtual;        U16         mg_private;        char        mg_type;        U8          mg_flags;        SV*         mg_obj;        char*       mg_ptr;        I32         mg_len;    };

Note this is current as of patchlevel 0, and could change at any time. 

Assigning Magic

Perl adds magic to an SV using the sv_magic function:

    void sv_magic(SV* sv, SV* obj, int how, char* name, I32 namlen);

The sv argument is a pointer to the SV that is to acquire a new magicalfeature.

If sv is not already magical, Perl uses the SvUPGRADE macro toset the SVt_PVMG flag for the sv. Perl then continues by addingit to the beginning of the linked list of magical features. Any priorentry of the same type of magic is deleted. Note that this can beoverridden, and multiple instances of the same type of magic can beassociated with an SV.

The name and namlen arguments are used to associate a string withthe magic, typically the name of a variable. namlen is stored in themg_len field and if name is non-null and namlen >= 0 a malloc'dcopy of the name is stored in mg_ptr field.

The sv_magic function uses how to determine which, if any, predefined``Magic Virtual Table'' should be assigned to the mg_virtual field.See the ``Magic Virtual Table'' section below. The how argument is alsostored in the mg_type field.

The obj argument is stored in the mg_obj field of the MAGICstructure. If it is not the same as the sv argument, the referencecount of the obj object is incremented. If it is the same, or ifthe how argument is ``#'', or if it is a NULL pointer, then obj ismerely stored, without the reference count being incremented.

There is also a function to add magic to an HV:

    void hv_magic(HV *hv, GV *gv, int how);

This simply calls sv_magic and coerces the gv argument into an SV.

To remove the magic from an SV, call the function sv_unmagic:

    void sv_unmagic(SV *sv, int type);

The type argument should be equal to the how value when the SVwas initially made magical. 

Magic Virtual Tables

The mg_virtual field in the MAGIC structure is a pointer to aMGVTBL, which is a structure of function pointers and stands for``Magic Virtual Table'' to handle the various operations that might beapplied to that variable.

The MGVTBL has five pointers to the following routine types:

    int  (*svt_get)(SV* sv, MAGIC* mg);    int  (*svt_set)(SV* sv, MAGIC* mg);    U32  (*svt_len)(SV* sv, MAGIC* mg);    int  (*svt_clear)(SV* sv, MAGIC* mg);    int  (*svt_free)(SV* sv, MAGIC* mg);

This MGVTBL structure is set at compile-time in perl.h and there arecurrently 19 types (or 21 with overloading turned on). These differentstructures contain pointers to various routines that perform additionalactions depending on which function is being called.

    Function pointer    Action taken    ----------------    ------------    svt_get             Do something after the value of the SV is retrieved.    svt_set             Do something after the SV is assigned a value.    svt_len             Report on the SV's length.    svt_clear           Clear something the SV represents.    svt_free            Free any extra storage associated with the SV.

For instance, the MGVTBL structure called vtbl_sv (which correspondsto an mg_type of `\0') contains:

    { magic_get, magic_set, magic_len, 0, 0 }

Thus, when an SV is determined to be magical and of type `\0', if a getoperation is being performed, the routine magic_get is called. Allthe various routines for the various magical types begin with magic_.

The current kinds of Magic Virtual Tables are:

    mg_type  MGVTBL              Type of magic    -------  ------              ----------------------------    \0       vtbl_sv             Special scalar variable    A        vtbl_amagic         %OVERLOAD hash    a        vtbl_amagicelem     %OVERLOAD hash element    c        (none)              Holds overload table (AMT) on stash    B        vtbl_bm             Boyer-Moore (fast string search)    E        vtbl_env            %ENV hash    e        vtbl_envelem        %ENV hash element    f        vtbl_fm             Formline ('compiled' format)    g        vtbl_mglob          m//g target / study()ed string    I        vtbl_isa            @ISA array    i        vtbl_isaelem        @ISA array element    k        vtbl_nkeys          scalar(keys()) lvalue    L        (none)              Debugger %_<filename     l        vtbl_dbline         Debugger %_<filename element    o        vtbl_collxfrm       Locale transformation    P        vtbl_pack           Tied array or hash    p        vtbl_packelem       Tied array or hash element    q        vtbl_packelem       Tied scalar or handle    S        vtbl_sig            %SIG hash    s        vtbl_sigelem        %SIG hash element    t        vtbl_taint          Taintedness    U        vtbl_uvar           Available for use by extensions    v        vtbl_vec            vec() lvalue    x        vtbl_substr         substr() lvalue    y        vtbl_defelem        Shadow "foreach" iterator variable /                                  smart parameter vivification    *        vtbl_glob           GV (typeglob)    #        vtbl_arylen         Array length ($#ary)    .        vtbl_pos            pos() lvalue    ~        (none)              Available for use by extensions

When an uppercase and lowercase letter both exist in the table, then theuppercase letter is used to represent some kind of composite type (a listor a hash), and the lowercase letter is used to represent an element ofthat composite type.

The `~' and `U' magic types are defined specifically for use byextensions and will not be used by perl itself. Extensions can use'~' magic to `attach' private information to variables (typicallyobjects). This is especially useful because there is no way fornormal perl code to corrupt this private information (unlike usingextra elements of a hash object).

Similarly, `U' magic can be used much like tie() to call a C functionany time a scalar's value is used or changed. The MAGIC'smg_ptr field points to a ufuncs structure:

    struct ufuncs {        I32 (*uf_val)(IV, SV*);        I32 (*uf_set)(IV, SV*);        IV uf_index;    };

When the SV is read from or written to, the uf_val or uf_setfunction will be called with uf_index as the first arg and apointer to the SV as the second.

Note that because multiple extensions may be using `~' or `U' magic,it is important for extensions to take extra care to avoid conflict.Typically only using the magic on objects blessed into the same classas the extension is sufficient. For `~' magic, it may also beappropriate to add an I32 `signature' at the top of the private dataarea and check that. 

Finding Magic

    MAGIC* mg_find(SV*, int type); /* Finds the magic pointer of that type */

This routine returns a pointer to the MAGIC structure stored in the SV.If the SV does not have that magical feature, NULL is returned. Also,if the SV is not of type SVt_PVMG, Perl may core dump.

    int mg_copy(SV* sv, SV* nsv, char* key, STRLEN klen);

This routine checks to see what types of magic sv has. If the mg_typefield is an uppercase letter, then the mg_obj is copied to nsv, butthe mg_type field is changed to be the lowercase letter. 

Understanding the Magic of Tied Hashes and Arrays

Tied hashes and arrays are magical beasts of the `P' magic type.

WARNING: As of the 5.004 release, proper usage of the array and hashaccess functions requires understanding a few caveats. Someof these caveats are actually considered bugs in the API, to be fixedin later releases, and are bracketed with [MAYCHANGE] below. Ifyou find yourself actually applying such information in this section, beaware that the behavior may change in the future, umm, without warning.

The av_store function, when given a tied array argument, merelycopies the magic of the array onto the value to be ``stored'', usingmg_copy. It may also return NULL, indicating that the value did notactually need to be stored in the array. [MAYCHANGE] After a call toav_store on a tied array, the caller will usually need to callmg_set(val) to actually invoke the perl level ``STORE'' method on theTIEARRAY object. If av_store did return NULL, a call toSvREFCNT_dec(val) will also be usually necessary to avoid a memoryleak. [/MAYCHANGE]

The previous paragraph is applicable verbatim to tied hash access using thehv_store and hv_store_ent functions as well.

av_fetch and the corresponding hash functions hv_fetch andhv_fetch_ent actually return an undefined mortal value whose magichas been initialized using mg_copy. Note the value so returned does notneed to be deallocated, as it is already mortal. [MAYCHANGE] But you willneed to call mg_get() on the returned value in order to actually invokethe perl level ``FETCH'' method on the underlying TIE object. Similarly,you may also call mg_set() on the return value after possibly assigninga suitable value to it using sv_setsv, which will invoke the ``STORE''method on the TIE object. [/MAYCHANGE]

[MAYCHANGE]In other words, the array or hash fetch/store functions don't reallyfetch and store actual values in the case of tied arrays and hashes. Theymerely call mg_copy to attach magic to the values that were meant to be``stored'' or ``fetched''. Later calls to mg_get and mg_set actuallydo the job of invoking the TIE methods on the underlying objects. Thusthe magic mechanism currently implements a kind of lazy access to arraysand hashes.

Currently (as of perl version 5.004), use of the hash and array accessfunctions requires the user to be aware of whether they are operating on``normal'' hashes and arrays, or on their tied variants. The API may bechanged to provide more transparent access to both tied and normal datatypes in future versions.[/MAYCHANGE]

You would do well to understand that the TIEARRAY and TIEHASH interfacesare mere sugar to invoke some perl method calls while using the uniform hashand array syntax. The use of this sugar imposes some overhead (typicallyabout two to four extra opcodes per FETCH/STORE operation, in addition tothe creation of all the mortal variables required to invoke the methods).This overhead will be comparatively small if the TIE methods are themselvessubstantial, but if they are only a few statements long, the overheadwill not be insignificant. 

Localizing changes

Perl has a very handy construction

  {    local $var = 2;    ...  }

This construction is approximately equivalent to

  {    my $oldvar = $var;    $var = 2;    ...    $var = $oldvar;  }

The biggest difference is that the first construction wouldreinstate the initial value of $var, irrespective of how control exitsthe block: goto, return, die/eval etc. It is a little bitmore efficient as well.

There is a way to achieve a similar task from C via Perl API: create apseudo-block, and arrange for some changes to be automaticallyundone at the end of it, either explicit, or via a non-local exit (viadie()). A block-like construct is created by a pair ofENTER/LEAVE macros (see the section on EXAMPLE/"Returning aScalar in the perlcall manpage). Such a construct may be created specially for someimportant localized task, or an existing one (like boundaries ofenclosing Perl subroutine/block, or an existing pair for freeing TMPs)may be used. (In the second case the overhead of additionallocalization must be almost negligible.) Note that any XSUB isautomatically enclosed in an ENTER/LEAVE pair.

Inside such a pseudo-block the following service is available:

SAVEINT(int i)

SAVEIV(IV i)

SAVEI32(I32 i)

SAVELONG(long i)

These macros arrange things to restore the value of integer variablei at the end of enclosing pseudo-block.

SAVESPTR(s)

SAVEPPTR(p)

These macros arrange things to restore the value of pointers s andp. s must be a pointer of a type which survives conversion toSV* and back, p should be able to survive conversion to char*and back.

SAVEFREESV(SV *sv)

The refcount of sv would be decremented at the end ofpseudo-block. This is similar to sv_2mortal, which should (?) beused instead.

SAVEFREEOP(OP *op)

The OP * is op_free()ed at the end of pseudo-block.

SAVEFREEPV(p)

The chunk of memory which is pointed to by p is Safefree()ed at theend of pseudo-block.

SAVECLEARSV(SV *sv)

Clears a slot in the current scratchpad which corresponds to sv atthe end of pseudo-block.

SAVEDELETE(HV *hv, char *key, I32 length)

The key key of hv is deleted at the end of pseudo-block. Thestring pointed to by key is Safefree()ed. If one has a key inshort-lived storage, the corresponding string may be reallocated likethis:

  SAVEDELETE(defstash, savepv(tmpbuf), strlen(tmpbuf));

SAVEDESTRUCTOR(f,p)

At the end of pseudo-block the function f is called with theonly argument (of type void*) p.

SAVESTACK_POS()

The current offset on the Perl internal stack (cf. SP) is restoredat the end of pseudo-block.

The following API list contains functions, thus one needs toprovide pointers to the modifiable data explicitly (either C pointers,or Perlish GV *s). Where the above macros take int, a similar function takes int *.

SV* save_scalar(GV *gv)

Equivalent to Perl code local $gv.

AV* save_ary(GV *gv)

HV* save_hash(GV *gv)

Similar to save_scalar, but localize @gv and %gv.

void save_item(SV *item)

Duplicates the current value of SV, on the exit from the currentENTER/LEAVE pseudo-block will restore the value of SVusing the stored value.

void save_list(SV **sarg, I32 maxsarg)

A variant of save_item which takes multiple arguments via an arraysarg of SV* of length maxsarg.

SV* save_svref(SV **sptr)

Similar to save_scalar, but will reinstate a SV *.

void save_aptr(AV **aptr)

void save_hptr(HV **hptr)

Similar to save_svref, but localize AV * and HV *.

The Alias module implements localization of the basic types within thecaller's scope. People who are interested in how to localize things inthe containing scope should take a look there too. 

Subroutines

 

XSUBs and the Argument Stack

The XSUB mechanism is a simple way for Perl programs to access C subroutines.An XSUB routine will have a stack that contains the arguments from the Perlprogram, and a way to map from the Perl data structures to a C equivalent.

The stack arguments are accessible through the ST(n) macro, which returnsthe n'th stack argument. Argument 0 is the first argument passed in thePerl subroutine call. These arguments are SV*, and can be used anywherean SV* is used.

Most of the time, output from the C routine can be handled through use ofthe RETVAL and OUTPUT directives. However, there are some cases where theargument stack is not already long enough to handle all the return values.An example is the POSIX tzname() call, which takes no arguments, but returnstwo, the local time zone's standard and summer time abbreviations.

To handle this situation, the PPCODE directive is used and the stack isextended using the macro:

    EXTEND(sp, num);

where sp is the stack pointer, and num is the number of elements thestack should be extended by.

Now that there is room on the stack, values can be pushed on it using themacros to push IVs, doubles, strings, and SV pointers respectively:

    PUSHi(IV)    PUSHn(double)    PUSHp(char*, I32)    PUSHs(SV*)

And now the Perl program calling tzname, the two values will be assignedas in:

    ($standard_abbrev, $summer_abbrev) = POSIX::tzname;

An alternate (and possibly simpler) method to pushing values on the stack isto use the macros:

    XPUSHi(IV)    XPUSHn(double)    XPUSHp(char*, I32)    XPUSHs(SV*)

These macros automatically adjust the stack for you, if needed. Thus, youdo not need to call EXTEND to extend the stack.

For more information, consult the perlxs manpage and the perlxstut manpage. 

Calling Perl Routines from within C Programs

There are four routines that can be used to call a Perl subroutine fromwithin a C program. These four are:

    I32  perl_call_sv(SV*, I32);    I32  perl_call_pv(char*, I32);    I32  perl_call_method(char*, I32);    I32  perl_call_argv(char*, I32, register char**);

The routine most often used is perl_call_sv. The SV* argumentcontains either the name of the Perl subroutine to be called, or areference to the subroutine. The second argument consists of flagsthat control the context in which the subroutine is called, whetheror not the subroutine is being passed arguments, how errors should betrapped, and how to treat return values.

All four routines return the number of arguments that the subroutine returnedon the Perl stack.

When using any of these routines (except perl_call_argv), the programmermust manipulate the Perl stack. These include the following macros andfunctions:

    dSP    PUSHMARK()    PUTBACK    SPAGAIN    ENTER    SAVETMPS    FREETMPS    LEAVE    XPUSH*()    POP*()

For a detailed description of calling conventions from C to Perl,consult the perlcall manpage. 

Memory Allocation

It is suggested that you use the version of malloc that is distributedwith Perl. It keeps pools of various sizes of unallocated memory inorder to satisfy allocation requests more quickly. However, on someplatforms, it may cause spurious malloc or free errors.

    New(x, pointer, number, type);    Newc(x, pointer, number, type, cast);    Newz(x, pointer, number, type);

These three macros are used to initially allocate memory.

The first argument x was a ``magic cookie'' that was used to keep trackof who called the macro, to help when debugging memory problems. However,the current code makes no use of this feature (most Perl developers nowuse run-time memory checkers), so this argument can be any number.

The second argument pointer should be the name of a variable that willpoint to the newly allocated memory.

The third and fourth arguments number and type specify how many ofthe specified type of data structure should be allocated. The argumenttype is passed to sizeof. The final argument to Newc, cast,should be used if the pointer argument is different from the typeargument.

Unlike the New and Newc macros, the Newz macro calls memzeroto zero out all the newly allocated memory.

    Renew(pointer, number, type);    Renewc(pointer, number, type, cast);    Safefree(pointer)

These three macros are used to change a memory buffer size or to free apiece of memory no longer needed. The arguments to Renew and Renewcmatch those of New and Newc with the exception of not needing the``magic cookie'' argument.

    Move(source, dest, number, type);    Copy(source, dest, number, type);    Zero(dest, number, type);

These three macros are used to move, copy, or zero out previously allocatedmemory. The source and dest arguments point to the source anddestination starting points. Perl will move, copy, or zero out numberinstances of the size of the type data structure (using the sizeoffunction). 

PerlIO

The most recent development releases of Perl has been experimenting withremoving Perl's dependency on the ``normal'' standard I/O suite and allowingother stdio implementations to be used. This involves creating a newabstraction layer that then calls whichever implementation of stdio Perlwas compiled with. All XSUBs should now use the functions in the PerlIOabstraction layer and not make any assumptions about what kind of stdiois being used.

For a complete description of the PerlIO abstraction, consult the perlapio manpage. 

Putting a C value on Perl stack

A lot of opcodes (this is an elementary operation in the internal perlstack machine) put an SV* on the stack. However, as an optimizationthe corresponding SV is (usually) not recreated each time. The opcodesreuse specially assigned SVs (targets) which are (as a corollary)not constantly freed/created.

Each of the targets is created only once (but seethe section on Scratchpads and recursion below), and when an opcode needs to putan integer, a double, or a string on stack, it just sets thecorresponding parts of its target and puts the target on stack.

The macro to put this target on stack is PUSHTARG, and it isdirectly used in some opcodes, as well as indirectly in zillions ofothers, which use it via (X)PUSH[pni]. 

Scratchpads

The question remains on when the SVs which are targets for opcodesare created. The answer is that they are created when the current unit --a subroutine or a file (for opcodes for statements outside ofsubroutines) -- is compiled. During this time a special anonymous Perlarray is created, which is called a scratchpad for the currentunit.

A scratchpad keeps SVs which are lexicals for the current unit and aretargets for opcodes. One can deduce that an SV lives on a scratchpadby looking on its flags: lexicals have SVs_PADMY set, andtargets have SVs_PADTMP set.

The correspondence between OPs and targets is not 1-to-1. DifferentOPs in the compile tree of the unit can use the same target, if thiswould not conflict with the expected life of the temporary. 

Scratchpads and recursion

In fact it is not 100% true that a compiled unit contains a pointer tothe scratchpad AV. In fact it contains a pointer to an AV of(initially) one element, and this element is the scratchpad AV. Why dowe need an extra level of indirection?

The answer is recursion, and maybe (sometime soon) threads. Boththese can create several execution pointers going into the samesubroutine. For the subroutine-child not write over the temporariesfor the subroutine-parent (lifespan of which covers the call to thechild), the parent and the child should have differentscratchpads. (And the lexicals should be separate anyway!)

So each subroutine is born with an array of scratchpads (of length 1).On each entry to the subroutine it is checked that the currentdepth of the recursion is not more than the length of this array, andif it is, new scratchpad is created and pushed into the array.

The targets on this scratchpad are undefs, but they are alreadymarked with correct flags. 

Compiled code

 

Code tree

Here we describe the internal form your code is converted to byPerl. Start with a simple example:

  $a = $b + $c;

This is converted to a tree similar to this one:

             assign-to           /           \          +             $a        /   \      $b     $c

(but slightly more complicated). This tree reflect the way Perlparsed your code, but has nothing to do with the execution order.There is an additional ``thread'' going through the nodes of the treewhich shows the order of execution of the nodes. In our simplifiedexample above it looks like:

     $b ---> $c ---> + ---> $a ---> assign-to

But with the actual compile tree for $a = $b + $c it is different:some nodes optimized away. As a corollary, though the actual treecontains more nodes than our simplified example, the execution orderis the same as in our example. 

Examining the tree

If you have your perl compiled for debugging (usually done with -Doptimize=-g on Configure command line), you may examine thecompiled tree by specifying -Dx on the Perl command line. Theoutput takes several lines per node, and for $b+$c it looks likethis:

    5           TYPE = add  ===> 6                TARG = 1                FLAGS = (SCALAR,KIDS)                {                    TYPE = null  ===> (4)                      (was rv2sv)                    FLAGS = (SCALAR,KIDS)                    {    3                   TYPE = gvsv  ===> 4                        FLAGS = (SCALAR)                        GV = main::b                    }                }                {                    TYPE = null  ===> (5)                      (was rv2sv)                    FLAGS = (SCALAR,KIDS)                    {    4                   TYPE = gvsv  ===> 5                        FLAGS = (SCALAR)                        GV = main::c                    }                }

This tree has 5 nodes (one per TYPE specifier), only 3 of them arenot optimized away (one per number in the left column). The immediatechildren of the given node correspond to {} pairs on the same levelof indentation, thus this listing corresponds to the tree:

                   add                 /     \               null    null                |       |               gvsv    gvsv

The execution order is indicated by ===> marks, thus it is 34 5 6 (node 6 is not included into above listing), i.e.,gvsv gvsv add whatever. 

Compile pass 1: check routines

The tree is created by the pseudo-compiler while yacc code feeds itthe constructions it recognizes. Since yacc works bottom-up, so doesthe first pass of perl compilation.

What makes this pass interesting for perl developers is that someoptimization may be performed on this pass. This is optimization byso-called check routines. The correspondence between node namesand corresponding check routines is described in opcode.pl (do notforget to run make regen_headers if you modify this file).

A check routine is called when the node is fully constructed exceptfor the execution-order thread. Since at this time there is noback-links to the currently constructed node, one can do most anyoperation to the top-level node, including freeing it and/or creatingnew nodes above/below it.

The check routine returns the node which should be inserted into thetree (if the top-level node was not modified, check routine returnsits argument).

By convention, check routines have names ck_*. They are usuallycalled from new*OP subroutines (or convert) (which in turn arecalled from perly.y). 

Compile pass 1a: constant folding

Immediately after the check routine is called the returned node ischecked for being compile-time executable. If it is (the value isjudged to be constant) it is immediately executed, and a constantnode with the ``return value'' of the corresponding subtree issubstituted instead. The subtree is deleted.

If constant folding was not performed, the execution-order thread iscreated. 

Compile pass 2: context propagation

When a context for a part of compile tree is known, it is propagateddown through the tree. Aat this time the context can have 5 values(instead of 2 for runtime context): void, boolean, scalar, list, andlvalue. In contrast with the pass 1 this pass is processed from topto bottom: a node's context determines the context for its children.

Additional context-dependent optimizations are performed at this time.Since at this moment the compile tree contains back-references (via``thread'' pointers), nodes cannot be free()d now. To allowoptimized-away nodes at this stage, such nodes are null()ified insteadof free()ing (i.e. their type is changed to OP_NULL). 

Compile pass 3: peephole optimization

After the compile tree for a subroutine (or for an eval or a file)is created, an additional pass over the code is performed. This passis neither top-down or bottom-up, but in the execution order (withadditional compilications for conditionals). These optimizations aredone in the subroutine peep(). Optimizations performed at this stageare subject to the same restrictions as in the pass 2. 

API LISTING

This is a listing of functions, macros, flags, and variables that may beuseful to extension writers or that may be found while reading otherextensions.

AvFILL

Same as av_len.

av_clear

Clears an array, making it empty. Does not free the memory used by thearray itself.

        void    av_clear _((AV* ar));

av_extend

Pre-extend an array. The key is the index to which the array should beextended.

        void    av_extend _((AV* ar, I32 key));

av_fetch

Returns the SV at the specified index in the array. The key is theindex. If lval is set then the fetch will be part of a store. Checkthat the return value is non-null before dereferencing it to a SV*.

See the section on Understanding the Magic of Tied Hashes and Arrays for moreinformation on how to use this function on tied arrays.

        SV**    av_fetch _((AV* ar, I32 key, I32 lval));

av_len

Returns the highest index in the array. Returns -1 if the array is empty.

        I32     av_len _((AV* ar));

av_make

Creates a new AV and populates it with a list of SVs. The SVs are copiedinto the array, so they may be freed after the call to av_make. The new AVwill have a reference count of 1.

        AV*     av_make _((I32 size, SV** svp));

av_pop

Pops an SV off the end of the array.