mikepaul-LuaJIT/doc/ext_ffi_api.html
2011-02-11 01:21:46 +01:00

431 lines
16 KiB
HTML

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
<title>ffi.* API Functions</title>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<meta name="Author" content="Mike Pall">
<meta name="Copyright" content="Copyright (C) 2005-2011, Mike Pall">
<meta name="Language" content="en">
<link rel="stylesheet" type="text/css" href="bluequad.css" media="screen">
<link rel="stylesheet" type="text/css" href="bluequad-print.css" media="print">
<style type="text/css">
table.abitable { width: 30em; line-height: 1.2; }
tr.abihead td { font-weight: bold; }
td.abiparam { font-weight: bold; width: 6em; }
</style>
</head>
<body>
<div id="site">
<a href="http://luajit.org"><span>Lua<span id="logo">JIT</span></span></a>
</div>
<div id="head">
<h1><tt>ffi.*</tt> API Functions</h1>
</div>
<div id="nav">
<ul><li>
<a href="luajit.html">LuaJIT</a>
<ul><li>
<a href="install.html">Installation</a>
</li><li>
<a href="running.html">Running</a>
</li></ul>
</li><li>
<a href="extensions.html">Extensions</a>
<ul><li>
<a href="ext_ffi.html">FFI Library</a>
<ul><li>
<a href="ext_ffi_tutorial.html">FFI Tutorial</a>
</li><li>
<a class="current" href="ext_ffi_api.html">ffi.* API</a>
</li><li>
<a href="ext_ffi_semantics.html">FFI Semantics</a>
</li></ul>
</li><li>
<a href="ext_jit.html">jit.* Library</a>
</li><li>
<a href="ext_c_api.html">Lua/C API</a>
</li></ul>
</li><li>
<a href="status.html">Status</a>
<ul><li>
<a href="changes.html">Changes</a>
</li></ul>
</li><li>
<a href="faq.html">FAQ</a>
</li><li>
<a href="http://luajit.org/performance.html">Performance <span class="ext">&raquo;</span></a>
</li><li>
<a href="http://luajit.org/download.html">Download <span class="ext">&raquo;</span></a>
</li></ul>
</div>
<div id="main">
<p>
This page describes the API functions provided by the FFI library in
detail. It's recommended to read through the
<a href="ext_ffi.html">introduction</a> and the
<a href="ext_ffi_tutorial.html">FFI tutorial</a> first.
</p>
<h2 id="glossary">Glossary</h2>
<ul>
<li><b>cdecl</b> &mdash; An abstract C&nbsp;type declaration (a Lua
string).</li>
<li><b>ctype</b> &mdash; A C&nbsp;type object. This is a special kind of
<b>cdata</b> returned by <tt>ffi.typeof()</tt>. It serves as a
<b>cdata</b> <a href="#ffi_new">constructor</a> when called.</li>
<li><b>cdata</b> &mdash; A C&nbsp;data object. It holds a value of the
corresponding <b>ctype</b>.</li>
<li><b>ct</b> &mdash; A C&nbsp;type specification which can be used for
most of the API functions. Either a <b>cdecl</b>, a <b>ctype</b> or a
<b>cdata</b> serving as a template type.</li>
<li><b>VLA</b> &mdash; A variable-length array is declared with a
<tt>?</tt> instead of the number of elements, e.g. <tt>"int[?]"</tt>.
The number of elements (<tt>nelem</tt>) must be given when it's
<a href="#ffi_new">created</a>.</li>
<li><b>VLS</b> &mdash; A variable-length struct is a <tt>struct</tt> C
type where the last element is a <b>VLA</b>. The same rules for
declaration and creation apply.</li>
</ul>
<h2 id="decl">Declaring and Accessing External Symbols</h2>
<p>
External symbols must be declared first and can then be accessed by
indexing a <a href="ext_ffi_semantics.html#clib">C&nbsp;library
namespace</a>, which automatically binds the symbol to a specific
library.
</p>
<h3 id="ffi_cdef"><tt>ffi.cdef(def)</tt></h3>
<p>
Adds multiple C&nbsp;declarations for types or external symbols (named
variables or functions). <tt>def</tt> must be a Lua string. It's
recommended to use the syntactic sugar for string arguments as
follows:
</p>
<pre class="code">
ffi.cdef[[
<span style="color:#00a000;">typedef struct foo { int a, b; } foo_t; // Declare a struct and typedef.
int dofoo(foo_t *f, int n); /* Declare an external C function. */</span>
]]
</pre>
<p>
The contents of the string (the part in green above) must be a
sequence of
<a href="ext_ffi_semantics.html#clang">C&nbsp;declarations</a>,
separated by semicolons. The trailing semicolon for a single
declaration may be omitted.
</p>
<p>
Please note that external symbols are only <em>declared</em>, but they
are <em>not bound</em> to any specific address, yet. Binding is
achieved with C&nbsp;library namespaces (see below).
</p>
<p style="color: #c00000;">
C&nbsp;declarations are not passed through a C&nbsp;pre-processor,
yet. No pre-processor tokens are allowed, except for
<tt>#pragma&nbsp;pack</tt>. Replace <tt>#define</tt> in existing
C&nbsp;header files with <tt>enum</tt>, <tt>static&nbsp;const</tt>
or <tt>typedef</tt> and/or pass the files through an external
C&nbsp;pre-processor (once). Be careful not to include unneeded or
redundant declarations from unrelated header files.
</p>
<h3 id="ffi_C"><tt>ffi.C</tt></h3>
<p>
This is the default C&nbsp;library namespace &mdash; note the
uppercase <tt>'C'</tt>. It binds to the default set of symbols or
libraries on the target system. These are more or less the same as a
C&nbsp;compiler would offer by default, without specifying extra link
libraries.
</p>
<p>
On POSIX systems, this binds to symbols in the default or global
namespace. This includes all exported symbols from the executable and
any libraries loaded into the global namespace. This includes at least
<tt>libc</tt>, <tt>libm</tt>, <tt>libdl</tt> (on Linux),
<tt>libgcc</tt> (if compiled with GCC), as well as any exported
symbols from the Lua/C&nbsp;API provided by LuaJIT itself.
</p>
<p>
On Windows systems, this binds to symbols exported from the
<tt>*.exe</tt>, the <tt>lua51.dll</tt> (i.e. the Lua/C&nbsp;API
provided by LuaJIT itself), the C&nbsp;runtime library LuaJIT was linked
with (<tt>msvcrt*.dll</tt>), <tt>kernel32.dll</tt>,
<tt>user32.dll</tt> and <tt>gdi32.dll</tt>.
</p>
<h3 id="ffi_load"><tt>clib = ffi.load(name [,global])</tt></h3>
<p>
This loads the dynamic library given by <tt>name</tt> and returns
a new C&nbsp;library namespace which binds to its symbols. On POSIX
systems, if <tt>global</tt> is <tt>true</tt>, the library symbols are
loaded into the global namespace, too.
</p>
<p>
If <tt>name</tt> is a path, the library is loaded from this path.
Otherwise <tt>name</tt> is canonicalized in a system-dependent way and
searched in the default search path for dynamic libraries:
</p>
<p>
On POSIX systems, if the name contains no dot, the extension
<tt>.so</tt> is appended. Also, the <tt>lib</tt> prefix is prepended
if necessary. So <tt>ffi.load("z")</tt> looks for <tt>"libz.so"</tt>
in the default shared library search path.
</p>
<p>
On Windows systems, if the name contains no dot, the extension
<tt>.dll</tt> is appended. So <tt>ffi.load("ws2_32")</tt> looks for
<tt>"ws2_32.dll"</tt> in the default DLL search path.
</p>
<h2 id="create">Creating cdata Objects</h2>
<p>
The following API functions create cdata objects (<tt>type()</tt>
returns <tt>"cdata"</tt>). All created cdata objects are
<a href="ext_ffi_semantics.html#gc">garbage collected</a>.
</p>
<h3 id="ffi_new"><tt>cdata = ffi.new(ct [,nelem] [,init...])<br>
cdata = <em>ctype</em>([nelem,] [init...])</tt></h3>
<p>
Creates a cdata object for the given <tt>ct</tt>. VLA/VLS types
require the <tt>nelem</tt> argument. The second syntax uses a ctype as
a constructor and is otherwise fully equivalent.
</p>
<p>
The cdata object is initialized according to the
<a href="ext_ffi_semantics.html#init">rules for initializers</a>,
using the optional <tt>init</tt> arguments. Excess initializers cause
an error.
</p>
<p>
Performance notice: if you want to create many objects of one kind,
parse the cdecl only once and get its ctype with
<tt>ffi.typeof()</tt>. Then use the ctype as a constructor repeatedly.
</p>
<p style="font-size: 8pt;">
Please note that an anonymous <tt>struct</tt> declaration implicitly
creates a new and distinguished ctype every time you use it for
<tt>ffi.new()</tt>. This is probably <b>not</b> what you want,
especially if you create more than one cdata object. Different anonymous
<tt>structs</tt> are not considered assignment-compatible by the
C&nbsp;standard, even though they may have the same fields! Also, they
are considered different types by the JIT-compiler, which may cause an
excessive number of traces. It's strongly suggested to either declare
a named <tt>struct</tt> or <tt>typedef</tt> with <tt>ffi.cdef()</tt>
or to create a single ctype object for an anonymous <tt>struct</tt>
with <tt>ffi.typeof()</tt>.
</p>
<h3 id="ffi_typeof"><tt>ctype = ffi.typeof(ct)</tt></h3>
<p>
Creates a ctype object for the given <tt>ct</tt>.
</p>
<p>
This function is especially useful to parse a cdecl only once and then
use the resulting ctype object as a <a href="#ffi_new">constructor</a>.
</p>
<h3 id="ffi_cast"><tt>cdata = ffi.cast(ct, init)</tt></h3>
<p>
Creates a scalar cdata object for the given <tt>ct</tt>. The cdata
object is initialized with <tt>init</tt> using the "cast" variant of
the <a href="ext_ffi_semantics.html#convert">C&nbsp;type conversion
rules</a>.
</p>
<p>
This functions is mainly useful to override the pointer compatibility
checks or to convert pointers to addresses or vice versa.
</p>
<h2 id="info">C&nbsp;Type Information</h2>
<p>
The following API functions return information about C&nbsp;types.
They are most useful for inspecting cdata objects.
</p>
<h3 id="ffi_sizeof"><tt>size = ffi.sizeof(ct [,nelem])</tt></h3>
<p>
Returns the size of <tt>ct</tt> in bytes. Returns <tt>nil</tt> if
the size is not known (e.g. for <tt>"void"</tt> or function types).
Requires <tt>nelem</tt> for VLA/VLS types, except for cdata objects.
</p>
<h3 id="ffi_alignof"><tt>align = ffi.alignof(ct)</tt></h3>
<p>
Returns the minimum required alignment for <tt>ct</tt> in bytes.
</p>
<h3 id="ffi_offsetof"><tt>ofs [,bpos,bsize] = ffi.offsetof(ct, field)</tt></h3>
<p>
Returns the offset (in bytes) of <tt>field</tt> relative to the start
of <tt>ct</tt>, which must be a <tt>struct</tt>. Additionally returns
the position and the field size (in bits) for bit fields.
</p>
<h2 id="util">Utility Functions</h2>
<h3 id="ffi_string"><tt>str = ffi.string(ptr [,len])</tt></h3>
<p>
Creates an interned Lua string from the data pointed to by
<tt>ptr</tt>.
</p>
<p>
If the optional argument <tt>len</tt> is missing, <tt>ptr</tt> is
converted to a <tt>"char&nbsp;*"</tt> and the data is assumed to be
zero-terminated. The length of the string is computed with
<tt>strlen()</tt>.
</p>
<p>
Otherwise <tt>ptr</tt> is converted to a <tt>"void&nbsp;*"</tt> and
<tt>len</tt> gives the length of the data. The data may contain
embedded zeros and need not be byte-oriented (though this may cause
endianess issues).
</p>
<p>
This function is mainly useful to convert (temporary)
<tt>"const&nbsp;char&nbsp;*"</tt> pointers returned by
C&nbsp;functions to Lua strings and store them or pass them to other
functions expecting a Lua string. The Lua string is an (interned) copy
of the data and bears no relation to the original data area anymore.
Lua strings are 8&nbsp;bit clean and may be used to hold arbitrary,
non-character data.
</p>
<p>
Performance notice: it's faster to pass the length of the string, if
it's known. E.g. when the length is returned by a C&nbsp;call like
<tt>sprintf()</tt>.
</p>
<h3 id="ffi_copy"><tt>ffi.copy(dst, src, len)<br>
ffi.copy(dst, str)</tt></h3>
<p>
Copies the data pointed to by <tt>src</tt> to <tt>dst</tt>.
<tt>dst</tt> is converted to a <tt>"void&nbsp;*"</tt> and <tt>src</tt>
is converted to a <tt>"const void&nbsp;*"</tt>.
</p>
<p>
In the first syntax, <tt>len</tt> gives the number of bytes to copy.
Caveat: if <tt>src</tt> is a Lua string, then <tt>len</tt> must not
exceed <tt>#src+1</tt>.
</p>
<p>
In the second syntax, the source of the copy must be a Lua string. All
bytes of the string <em>plus a zero-terminator</em> are copied to
<tt>dst</tt> (i.e. <tt>#src+1</tt> bytes).
</p>
<p>
Performance notice: <tt>ffi.copy()</tt> may be used as a faster
(inlineable) replacement for the C&nbsp;library functions
<tt>memcpy()</tt>, <tt>strcpy()</tt> and <tt>strncpy()</tt>.
</p>
<h3 id="ffi_fill"><tt>ffi.fill(dst, len [,c])</tt></h3>
<p>
Fills the data pointed to by <tt>dst</tt> with <tt>len</tt> constant
bytes, given by <tt>c</tt>. If <tt>c</tt> is omitted, the data is
zero-filled.
</p>
<p>
Performance notice: <tt>ffi.fill()</tt> may be used as a faster
(inlineable) replacement for the C&nbsp;library function
<tt>memset(dst,&nbsp;c,&nbsp;len)</tt>. Please note the different
order of arguments!
</p>
<h2 id="target">Target-specific Information</h2>
<h3 id="ffi_abi"><tt>status = ffi.abi(param)</tt></h3>
<p>
Returns <tt>true</tt> if <tt>param</tt> (a Lua string) applies for the
target ABI (Application Binary Interface). Returns <tt>false</tt>
otherwise. The following parameters are currently defined:
</p>
<table class="abitable">
<tr class="abihead">
<td class="abiparam">Parameter</td>
<td class="abidesc">Description</td>
</tr>
<tr class="odd separate">
<td class="abiparam">32bit</td><td class="abidesc">32 bit architecture</td></tr>
<tr class="even">
<td class="abiparam">64bit</td><td class="abidesc">64 bit architecture</td></tr>
<tr class="odd separate">
<td class="abiparam">le</td><td class="abidesc">Little-endian architecture</td></tr>
<tr class="even">
<td class="abiparam">be</td><td class="abidesc">Big-endian architecture</td></tr>
<tr class="odd separate">
<td class="abiparam">fpu</td><td class="abidesc">Target has a hardware FPU</td></tr>
<tr class="even">
<td class="abiparam">softfp</td><td class="abidesc">softfp calling conventions</td></tr>
<tr class="odd">
<td class="abiparam">hardfp</td><td class="abidesc">hardfp calling conventions</td></tr>
<tr class="even separate">
<td class="abiparam">eabi</td><td class="abidesc">EABI variant of the standard ABI</td></tr>
<tr class="odd">
<td class="abiparam">win</td><td class="abidesc">Windows variant of the standard ABI</td></tr>
</table>
<h3 id="ffi_os"><tt>ffi.os</tt></h3>
<p>
Contains the target OS name. Same contents as
<a href="ext_jit.html#jit_os"><tt>jit.os</tt></a>.
</p>
<h3 id="ffi_arch"><tt>ffi.arch</tt></h3>
<p>
Contains the target architecture name. Same contents as
<a href="ext_jit.html#jit_arch"><tt>jit.arch</tt></a>.
</p>
<h2 id="extended">Extended Standard Library Functions</h2>
<p>
The following standard library functions have been extended to work
with cdata objects:
</p>
<h3 id="tonumber"><tt>n = tonumber(cdata)</tt></h3>
<p>
Converts a number cdata object to a <tt>double</tt> and returns it as
a Lua number. This is particularly useful for boxed 64&nbsp;bit
integer values. Caveat: this conversion may incur a precision loss.
</p>
<h3 id="tostring"><tt>s = tostring(cdata)</tt></h3>
<p>
Returns a string representation of the value of 64&nbsp;bit integers
(<tt><b>"</b>nnn<b>LL"</b></tt> or <tt><b>"</b>nnn<b>ULL"</b></tt>) or
complex numbers (<tt><b>"</b>re&plusmn;im<b>i"</b></tt>). Otherwise
returns a string representation of the C&nbsp;type of a ctype object
(<tt><b>"ctype&lt;</b>type<b>&gt;"</b></tt>) or a cdata object
(<tt><b>"cdata&lt;</b>type<b>&gt;:&nbsp;</b>address"</tt>).
</p>
<h2 id="literals">Extensions to the Lua Parser</h2>
<p>
The parser for Lua source code treats numeric literals with the
suffixes <tt>LL</tt> or <tt>ULL</tt> as signed or unsigned 64&nbsp;bit
integers. Case doesn't matter, but uppercase is recommended for
readability. It handles both decimal (<tt>42LL</tt>) and hexadecimal
(<tt>0x2aLL</tt>) literals.
</p>
<p>
The imaginary part of complex numbers can be specified by suffixing
number literals with <tt>i</tt> or <tt>I</tt>, e.g. <tt>12.5i</tt>.
Caveat: you'll need to use <tt>1i</tt> to get an imaginary part with
the value one, since <tt>i</tt> itself still refers to a variable
named <tt>i</tt>.
</p>
<br class="flush">
</div>
<div id="foot">
<hr class="hide">
Copyright &copy; 2005-2011 Mike Pall
<span class="noprint">
&middot;
<a href="contact.html">Contact</a>
</span>
</div>
</body>
</html>