don't copy win64 struct args

This commit is contained in:
Anthony Green
2010-08-05 08:24:27 -04:00
parent d14178be4c
commit 10ea848900
10 changed files with 5571 additions and 22 deletions

View File

@@ -9,3 +9,4 @@ os2
spelling
selinux
ffi_last_abi
win64-struct-args

View File

File diff suppressed because it is too large Load Diff

View File

@@ -0,0 +1,598 @@
\input texinfo @c -*-texinfo-*-
@c %**start of header
@setfilename libffi.info
@settitle libffi
@setchapternewpage off
@c %**end of header
@c Merge the standard indexes into a single one.
@syncodeindex fn cp
@syncodeindex vr cp
@syncodeindex ky cp
@syncodeindex pg cp
@syncodeindex tp cp
@include version.texi
@copying
This manual is for Libffi, a portable foreign-function interface
library.
Copyright @copyright{} 2008, 2010 Red Hat, Inc.
@quotation
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 2, or (at your option) any
later version. A copy of the license is included in the
section entitled ``GNU General Public License''.
@end quotation
@end copying
@dircategory Development
@direntry
* libffi: (libffi). Portable foreign-function interface library.
@end direntry
@titlepage
@title Libffi
@page
@vskip 0pt plus 1filll
@insertcopying
@end titlepage
@ifnottex
@node Top
@top libffi
@insertcopying
@menu
* Introduction:: What is libffi?
* Using libffi:: How to use libffi.
* Missing Features:: Things libffi can't do.
* Index:: Index.
@end menu
@end ifnottex
@node Introduction
@chapter What is libffi?
Compilers for high level languages generate code that follow certain
conventions. These conventions are necessary, in part, for separate
compilation to work. One such convention is the @dfn{calling
convention}. The calling convention is a set of assumptions made by
the compiler about where function arguments will be found on entry to
a function. A calling convention also specifies where the return
value for a function is found. The calling convention is also
sometimes called the @dfn{ABI} or @dfn{Application Binary Interface}.
@cindex calling convention
@cindex ABI
@cindex Application Binary Interface
Some programs may not know at the time of compilation what arguments
are to be passed to a function. For instance, an interpreter may be
told at run-time about the number and types of arguments used to call
a given function. @samp{Libffi} can be used in such programs to
provide a bridge from the interpreter program to compiled code.
The @samp{libffi} library provides a portable, high level programming
interface to various calling conventions. This allows a programmer to
call any function specified by a call interface description at run
time.
@acronym{FFI} stands for Foreign Function Interface. A foreign
function interface is the popular name for the interface that allows
code written in one language to call code written in another language.
The @samp{libffi} library really only provides the lowest, machine
dependent layer of a fully featured foreign function interface. A
layer must exist above @samp{libffi} that handles type conversions for
values passed between the two languages.
@cindex FFI
@cindex Foreign Function Interface
@node Using libffi
@chapter Using libffi
@menu
* The Basics:: The basic libffi API.
* Simple Example:: A simple example.
* Types:: libffi type descriptions.
* Multiple ABIs:: Different passing styles on one platform.
* The Closure API:: Writing a generic function.
* Closure Example:: A closure example.
@end menu
@node The Basics
@section The Basics
@samp{Libffi} assumes that you have a pointer to the function you wish
to call and that you know the number and types of arguments to pass
it, as well as the return type of the function.
The first thing you must do is create an @code{ffi_cif} object that
matches the signature of the function you wish to call. This is a
separate step because it is common to make multiple calls using a
single @code{ffi_cif}. The @dfn{cif} in @code{ffi_cif} stands for
Call InterFace. To prepare a call interface object, use the function
@code{ffi_prep_cif}.
@cindex cif
@findex ffi_prep_cif
@defun ffi_status ffi_prep_cif (ffi_cif *@var{cif}, ffi_abi @var{abi}, unsigned int @var{nargs}, ffi_type *@var{rtype}, ffi_type **@var{argtypes})
This initializes @var{cif} according to the given parameters.
@var{abi} is the ABI to use; normally @code{FFI_DEFAULT_ABI} is what
you want. @ref{Multiple ABIs} for more information.
@var{nargs} is the number of arguments that this function accepts.
@samp{libffi} does not yet handle varargs functions; see @ref{Missing
Features} for more information.
@var{rtype} is a pointer to an @code{ffi_type} structure that
describes the return type of the function. @xref{Types}.
@var{argtypes} is a vector of @code{ffi_type} pointers.
@var{argtypes} must have @var{nargs} elements. If @var{nargs} is 0,
this argument is ignored.
@code{ffi_prep_cif} returns a @code{libffi} status code, of type
@code{ffi_status}. This will be either @code{FFI_OK} if everything
worked properly; @code{FFI_BAD_TYPEDEF} if one of the @code{ffi_type}
objects is incorrect; or @code{FFI_BAD_ABI} if the @var{abi} parameter
is invalid.
@end defun
To call a function using an initialized @code{ffi_cif}, use the
@code{ffi_call} function:
@findex ffi_call
@defun void ffi_call (ffi_cif *@var{cif}, void *@var{fn}, void *@var{rvalue}, void **@var{avalues})
This calls the function @var{fn} according to the description given in
@var{cif}. @var{cif} must have already been prepared using
@code{ffi_prep_cif}.
@var{rvalue} is a pointer to a chunk of memory that will hold the
result of the function call. This must be large enough to hold the
result and must be suitably aligned; it is the caller's responsibility
to ensure this. If @var{cif} declares that the function returns
@code{void} (using @code{ffi_type_void}), then @var{rvalue} is
ignored. If @var{rvalue} is @samp{NULL}, then the return value is
discarded.
@var{avalues} is a vector of @code{void *} pointers that point to the
memory locations holding the argument values for a call. If @var{cif}
declares that the function has no arguments (i.e., @var{nargs} was 0),
then @var{avalues} is ignored.
@end defun
@node Simple Example
@section Simple Example
Here is a trivial example that calls @code{puts} a few times.
@example
#include <stdio.h>
#include <ffi.h>
int main()
@{
ffi_cif cif;
ffi_type *args[1];
void *values[1];
char *s;
int rc;
/* Initialize the argument info vectors */
args[0] = &ffi_type_pointer;
values[0] = &s;
/* Initialize the cif */
if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, 1,
&ffi_type_uint, args) == FFI_OK)
@{
s = "Hello World!";
ffi_call(&cif, puts, &rc, values);
/* rc now holds the result of the call to puts */
/* values holds a pointer to the function's arg, so to
call puts() again all we need to do is change the
value of s */
s = "This is cool!";
ffi_call(&cif, puts, &rc, values);
@}
return 0;
@}
@end example
@node Types
@section Types
@menu
* Primitive Types:: Built-in types.
* Structures:: Structure types.
* Type Example:: Structure type example.
@end menu
@node Primitive Types
@subsection Primitive Types
@code{Libffi} provides a number of built-in type descriptors that can
be used to describe argument and return types:
@table @code
@item ffi_type_void
@tindex ffi_type_void
The type @code{void}. This cannot be used for argument types, only
for return values.
@item ffi_type_uint8
@tindex ffi_type_uint8
An unsigned, 8-bit integer type.
@item ffi_type_sint8
@tindex ffi_type_sint8
A signed, 8-bit integer type.
@item ffi_type_uint16
@tindex ffi_type_uint16
An unsigned, 16-bit integer type.
@item ffi_type_sint16
@tindex ffi_type_sint16
A signed, 16-bit integer type.
@item ffi_type_uint32
@tindex ffi_type_uint32
An unsigned, 32-bit integer type.
@item ffi_type_sint32
@tindex ffi_type_sint32
A signed, 32-bit integer type.
@item ffi_type_uint64
@tindex ffi_type_uint64
An unsigned, 64-bit integer type.
@item ffi_type_sint64
@tindex ffi_type_sint64
A signed, 64-bit integer type.
@item ffi_type_float
@tindex ffi_type_float
The C @code{float} type.
@item ffi_type_double
@tindex ffi_type_double
The C @code{double} type.
@item ffi_type_uchar
@tindex ffi_type_uchar
The C @code{unsigned char} type.
@item ffi_type_schar
@tindex ffi_type_schar
The C @code{signed char} type. (Note that there is not an exact
equivalent to the C @code{char} type in @code{libffi}; ordinarily you
should either use @code{ffi_type_schar} or @code{ffi_type_uchar}
depending on whether @code{char} is signed.)
@item ffi_type_ushort
@tindex ffi_type_ushort
The C @code{unsigned short} type.
@item ffi_type_sshort
@tindex ffi_type_sshort
The C @code{short} type.
@item ffi_type_uint
@tindex ffi_type_uint
The C @code{unsigned int} type.
@item ffi_type_sint
@tindex ffi_type_sint
The C @code{int} type.
@item ffi_type_ulong
@tindex ffi_type_ulong
The C @code{unsigned long} type.
@item ffi_type_slong
@tindex ffi_type_slong
The C @code{long} type.
@item ffi_type_longdouble
@tindex ffi_type_longdouble
On platforms that have a C @code{long double} type, this is defined.
On other platforms, it is not.
@item ffi_type_pointer
@tindex ffi_type_pointer
A generic @code{void *} pointer. You should use this for all
pointers, regardless of their real type.
@end table
Each of these is of type @code{ffi_type}, so you must take the address
when passing to @code{ffi_prep_cif}.
@node Structures
@subsection Structures
Although @samp{libffi} has no special support for unions or
bit-fields, it is perfectly happy passing structures back and forth.
You must first describe the structure to @samp{libffi} by creating a
new @code{ffi_type} object for it.
@tindex ffi_type
@deftp ffi_type
The @code{ffi_type} has the following members:
@table @code
@item size_t size
This is set by @code{libffi}; you should initialize it to zero.
@item unsigned short alignment
This is set by @code{libffi}; you should initialize it to zero.
@item unsigned short type
For a structure, this should be set to @code{FFI_TYPE_STRUCT}.
@item ffi_type **elements
This is a @samp{NULL}-terminated array of pointers to @code{ffi_type}
objects. There is one element per field of the struct.
@end table
@end deftp
@node Type Example
@subsection Type Example
The following example initializes a @code{ffi_type} object
representing the @code{tm} struct from Linux's @file{time.h}.
Here is how the struct is defined:
@example
struct tm @{
int tm_sec;
int tm_min;
int tm_hour;
int tm_mday;
int tm_mon;
int tm_year;
int tm_wday;
int tm_yday;
int tm_isdst;
/* Those are for future use. */
long int __tm_gmtoff__;
__const char *__tm_zone__;
@};
@end example
Here is the corresponding code to describe this struct to
@code{libffi}:
@example
@{
ffi_type tm_type;
ffi_type *tm_type_elements[12];
int i;
tm_type.size = tm_type.alignment = 0;
tm_type.elements = &tm_type_elements;
for (i = 0; i < 9; i++)
tm_type_elements[i] = &ffi_type_sint;
tm_type_elements[9] = &ffi_type_slong;
tm_type_elements[10] = &ffi_type_pointer;
tm_type_elements[11] = NULL;
/* tm_type can now be used to represent tm argument types and
return types for ffi_prep_cif() */
@}
@end example
@node Multiple ABIs
@section Multiple ABIs
A given platform may provide multiple different ABIs at once. For
instance, the x86 platform has both @samp{stdcall} and @samp{fastcall}
functions.
@code{libffi} provides some support for this. However, this is
necessarily platform-specific.
@c FIXME: document the platforms
@node The Closure API
@section The Closure API
@code{libffi} also provides a way to write a generic function -- a
function that can accept and decode any combination of arguments.
This can be useful when writing an interpreter, or to provide wrappers
for arbitrary functions.
This facility is called the @dfn{closure API}. Closures are not
supported on all platforms; you can check the @code{FFI_CLOSURES}
define to determine whether they are supported on the current
platform.
@cindex closures
@cindex closure API
@findex FFI_CLOSURES
Because closures work by assembling a tiny function at runtime, they
require special allocation on platforms that have a non-executable
heap. Memory management for closures is handled by a pair of
functions:
@findex ffi_closure_alloc
@defun void *ffi_closure_alloc (size_t @var{size}, void **@var{code})
Allocate a chunk of memory holding @var{size} bytes. This returns a
pointer to the writable address, and sets *@var{code} to the
corresponding executable address.
@var{size} should be sufficient to hold a @code{ffi_closure} object.
@end defun
@findex ffi_closure_free
@defun void ffi_closure_free (void *@var{writable})
Free memory allocated using @code{ffi_closure_alloc}. The argument is
the writable address that was returned.
@end defun
Once you have allocated the memory for a closure, you must construct a
@code{ffi_cif} describing the function call. Finally you can prepare
the closure function:
@findex ffi_prep_closure_loc
@defun ffi_status ffi_prep_closure_loc (ffi_closure *@var{closure}, ffi_cif *@var{cif}, void (*@var{fun}) (ffi_cif *@var{cif}, void *@var{ret}, void **@var{args}, void *@var{user_data}), void *@var{user_data}, void *@var{codeloc})
Prepare a closure function.
@var{closure} is the address of a @code{ffi_closure} object; this is
the writable address returned by @code{ffi_closure_alloc}.
@var{cif} is the @code{ffi_cif} describing the function parameters.
@var{user_data} is an arbitrary datum that is passed, uninterpreted,
to your closure function.
@var{codeloc} is the executable address returned by
@code{ffi_closure_alloc}.
@var{fun} is the function which will be called when the closure is
invoked. It is called with the arguments:
@table @var
@item cif
The @code{ffi_cif} passed to @code{ffi_prep_closure_loc}.
@item ret
A pointer to the memory used for the function's return value.
@var{fun} must fill this, unless the function is declared as returning
@code{void}.
@c FIXME: is this NULL for void-returning functions?
@item args
A vector of pointers to memory holding the arguments to the function.
@item user_data
The same @var{user_data} that was passed to
@code{ffi_prep_closure_loc}.
@end table
@code{ffi_prep_closure_loc} will return @code{FFI_OK} if everything
went ok, and something else on error.
@c FIXME: what?
After calling @code{ffi_prep_closure_loc}, you can cast @var{codeloc}
to the appropriate pointer-to-function type.
@end defun
You may see old code referring to @code{ffi_prep_closure}. This
function is deprecated, as it cannot handle the need for separate
writable and executable addresses.
@node Closure Example
@section Closure Example
A trivial example that creates a new @code{puts} by binding
@code{fputs} with @code{stdin}.
@example
#include <stdio.h>
#include <ffi.h>
/* Acts like puts with the file given at time of enclosure. */
void puts_binding(ffi_cif *cif, unsigned int *ret, void* args[],
FILE *stream)
@{
*ret = fputs(*(char **)args[0], stream);
@}
int main()
@{
ffi_cif cif;
ffi_type *args[1];
ffi_closure *closure;
int (*bound_puts)(char *);
int rc;
/* Allocate closure and bound_puts */
closure = ffi_closure_alloc(sizeof(ffi_closure), &bound_puts);
if (closure)
@{
/* Initialize the argument info vectors */
args[0] = &ffi_type_pointer;
/* Initialize the cif */
if (ffi_prep_cif(&cif, FFI_DEFAULT_ABI, 1,
&ffi_type_uint, args) == FFI_OK)
@{
/* Initialize the closure, setting stream to stdout */
if (ffi_prep_closure_loc(closure, &cif, puts_binding,
stdout, bound_puts) == FFI_OK)
@{
rc = bound_puts("Hello World!");
/* rc now holds the result of the call to fputs */
@}
@}
@}
/* Deallocate both closure, and bound_puts */
ffi_closure_free(closure);
return 0;
@}
@end example
@node Missing Features
@chapter Missing Features
@code{libffi} is missing a few features. We welcome patches to add
support for these.
@itemize @bullet
@item
There is no support for calling varargs functions. This may work on
some platforms, depending on how the ABI is defined, but it is not
reliable.
@item
There is no support for bit fields in structures.
@item
The closure API is
@c FIXME: ...
@item
The ``raw'' API is undocumented.
@c argument promotion?
@c unions?
@c anything else?
@end itemize
@node Index
@unnumbered Index
@printindex cp
@bye

View File

@@ -0,0 +1,665 @@
/* -----------------------------------------------------------------------
ffi.c - Copyright (c) 1996, 1998, 1999, 2001, 2007, 2008 Red Hat, Inc.
Copyright (c) 2002 Ranjit Mathew
Copyright (c) 2002 Bo Thorsen
Copyright (c) 2002 Roger Sayle
Copyright (C) 2008 Free Software Foundation, Inc.
x86 Foreign Function Interface
Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
``Software''), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to
permit persons to whom the Software is furnished to do so, subject to
the following conditions:
The above copyright notice and this permission notice shall be included
in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
DEALINGS IN THE SOFTWARE.
----------------------------------------------------------------------- */
#if !defined(__x86_64__) || defined(_WIN64)
#ifdef _WIN64
#include <windows.h>
#endif
#include <ffi.h>
#include <ffi_common.h>
#include <stdlib.h>
/* ffi_prep_args is called by the assembly routine once stack space
has been allocated for the function's arguments */
void ffi_prep_args(char *stack, extended_cif *ecif)
{
register unsigned int i;
register void **p_argv;
register char *argp;
register ffi_type **p_arg;
argp = stack;
if (ecif->cif->flags == FFI_TYPE_STRUCT
#ifdef X86_WIN64
&& (ecif->cif->rtype->size != 1 && ecif->cif->rtype->size != 2
&& ecif->cif->rtype->size != 4 && ecif->cif->rtype->size != 8)
#endif
)
{
*(void **) argp = ecif->rvalue;
argp += sizeof(void*);
}
p_argv = ecif->avalue;
for (i = ecif->cif->nargs, p_arg = ecif->cif->arg_types;
i != 0;
i--, p_arg++)
{
size_t z;
/* Align if necessary */
if ((sizeof(void*) - 1) & (size_t) argp)
argp = (char *) ALIGN(argp, sizeof(void*));
z = (*p_arg)->size;
#ifdef X86_WIN64
if (z > sizeof(ffi_arg)
|| ((*p_arg)->type == FFI_TYPE_STRUCT
&& (z != 1 && z != 2 && z != 4 && z != 8))
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
|| ((*p_arg)->type == FFI_TYPE_LONGDOUBLE)
#endif
)
{
z = sizeof(ffi_arg);
*(void **)argp = *p_argv;
}
else if ((*p_arg)->type == FFI_TYPE_FLOAT)
{
memcpy(argp, *p_argv, z);
}
else
#endif
if (z < sizeof(ffi_arg))
{
z = sizeof(ffi_arg);
switch ((*p_arg)->type)
{
case FFI_TYPE_SINT8:
*(ffi_sarg *) argp = (ffi_sarg)*(SINT8 *)(* p_argv);
break;
case FFI_TYPE_UINT8:
*(ffi_arg *) argp = (ffi_arg)*(UINT8 *)(* p_argv);
break;
case FFI_TYPE_SINT16:
*(ffi_sarg *) argp = (ffi_sarg)*(SINT16 *)(* p_argv);
break;
case FFI_TYPE_UINT16:
*(ffi_arg *) argp = (ffi_arg)*(UINT16 *)(* p_argv);
break;
case FFI_TYPE_SINT32:
*(ffi_sarg *) argp = (ffi_sarg)*(SINT32 *)(* p_argv);
break;
case FFI_TYPE_UINT32:
*(ffi_arg *) argp = (ffi_arg)*(UINT32 *)(* p_argv);
break;
case FFI_TYPE_STRUCT:
*(ffi_arg *) argp = *(ffi_arg *)(* p_argv);
break;
default:
FFI_ASSERT(0);
}
}
else
{
memcpy(argp, *p_argv, z);
}
p_argv++;
#ifdef X86_WIN64
argp += (z + sizeof(void*) - 1) & ~(sizeof(void*) - 1);
#else
argp += z;
#endif
}
return;
}
/* Perform machine dependent cif processing */
ffi_status ffi_prep_cif_machdep(ffi_cif *cif)
{
unsigned int i;
ffi_type **ptr;
/* Set the return type flag */
switch (cif->rtype->type)
{
case FFI_TYPE_VOID:
#if defined(X86) || defined (X86_WIN32) || defined(X86_FREEBSD) || defined(X86_DARWIN) || defined(X86_WIN64)
case FFI_TYPE_UINT8:
case FFI_TYPE_UINT16:
case FFI_TYPE_SINT8:
case FFI_TYPE_SINT16:
#endif
#ifdef X86_WIN64
case FFI_TYPE_UINT32:
case FFI_TYPE_SINT32:
#endif
case FFI_TYPE_SINT64:
case FFI_TYPE_FLOAT:
case FFI_TYPE_DOUBLE:
#ifndef X86_WIN64
#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE
case FFI_TYPE_LONGDOUBLE:
#endif
#endif
cif->flags = (unsigned) cif->rtype->type;
break;
case FFI_TYPE_UINT64:
#ifdef X86_WIN64
case FFI_TYPE_POINTER:
#endif
cif->flags = FFI_TYPE_SINT64;
break;
case FFI_TYPE_STRUCT:
#ifndef X86
if (cif->rtype->size == 1)
{
cif->flags = FFI_TYPE_SMALL_STRUCT_1B; /* same as char size */
}
else if (cif->rtype->size == 2)
{
cif->flags = FFI_TYPE_SMALL_STRUCT_2B; /* same as short size */
}
else if (cif->rtype->size == 4)
{
#ifdef X86_WIN64
cif->flags = FFI_TYPE_SMALL_STRUCT_4B;
#else
cif->flags = FFI_TYPE_INT; /* same as int type */
#endif
}
else if (cif->rtype->size == 8)
{
cif->flags = FFI_TYPE_SINT64; /* same as int64 type */
}
else
#endif
{
cif->flags = FFI_TYPE_STRUCT;
/* allocate space for return value pointer */
cif->bytes += ALIGN(sizeof(void*), FFI_SIZEOF_ARG);
}
break;
default:
#ifdef X86_WIN64
cif->flags = FFI_TYPE_SINT64;
break;
case FFI_TYPE_INT:
cif->flags = FFI_TYPE_SINT32;
#else
cif->flags = FFI_TYPE_INT;
#endif
break;
}
for (ptr = cif->arg_types, i = cif->nargs; i > 0; i--, ptr++)
{
if (((*ptr)->alignment - 1) & cif->bytes)
cif->bytes = ALIGN(cif->bytes, (*ptr)->alignment);
cif->bytes += ALIGN((*ptr)->size, FFI_SIZEOF_ARG);
}
#ifdef X86_WIN64
/* ensure space for storing four registers */
cif->bytes += 4 * sizeof(ffi_arg);
#endif
#ifdef X86_DARWIN
cif->bytes = (cif->bytes + 15) & ~0xF;
#endif
return FFI_OK;
}
#ifdef X86_WIN64
extern int
ffi_call_win64(void (*)(char *, extended_cif *), extended_cif *,
unsigned, unsigned, unsigned *, void (*fn)(void));
#elif defined(X86_WIN32)
extern void
ffi_call_win32(void (*)(char *, extended_cif *), extended_cif *,
unsigned, unsigned, unsigned *, void (*fn)(void));
#else
extern void ffi_call_SYSV(void (*)(char *, extended_cif *), extended_cif *,
unsigned, unsigned, unsigned *, void (*fn)(void));
#endif
void ffi_call(ffi_cif *cif, void (*fn)(void), void *rvalue, void **avalue)
{
extended_cif ecif;
ecif.cif = cif;
ecif.avalue = avalue;
/* If the return value is a struct and we don't have a return */
/* value address then we need to make one */
#ifdef X86_WIN64
if (rvalue == NULL
&& cif->flags == FFI_TYPE_STRUCT
&& cif->rtype->size != 1 && cif->rtype->size != 2
&& cif->rtype->size != 4 && cif->rtype->size != 8)
{
ecif.rvalue = alloca((cif->rtype->size + 0xF) & ~0xF);
}
#else
if (rvalue == NULL
&& cif->flags == FFI_TYPE_STRUCT)
{
ecif.rvalue = alloca(cif->rtype->size);
}
#endif
else
ecif.rvalue = rvalue;
switch (cif->abi)
{
#ifdef X86_WIN64
case FFI_WIN64:
{
/* Make copies of all struct arguments
NOTE: not sure if responsibility should be here or in caller */
unsigned int i;
for (i=0; i < cif->nargs;i++) {
size_t size = cif->arg_types[i]->size;
if ((cif->arg_types[i]->type == FFI_TYPE_STRUCT
&& (size != 1 && size != 2 && size != 4 && size != 8))
#if FFI_TYPE_LONGDOUBLE != FFI_TYPE_DOUBLE
|| cif->arg_types[i]->type == FFI_TYPE_LONGDOUBLE
#endif
)
{
void *local = alloca(size);
memcpy(local, avalue[i], size);
avalue[i] = local;
}
}
ffi_call_win64(ffi_prep_args, &ecif, cif->bytes,
cif->flags, ecif.rvalue, fn);
}
break;
#elif defined(X86_WIN32)
case FFI_SYSV:
case FFI_STDCALL:
ffi_call_win32(ffi_prep_args, &ecif, cif->bytes, cif->flags,
ecif.rvalue, fn);
break;
#else
case FFI_SYSV:
ffi_call_SYSV(ffi_prep_args, &ecif, cif->bytes, cif->flags, ecif.rvalue,
fn);
break;
#endif
default:
FFI_ASSERT(0);
break;
}
}
/** private members **/
/* The following __attribute__((regparm(1))) decorations will have no effect
on MSVC - standard cdecl convention applies. */
static void ffi_prep_incoming_args_SYSV (char *stack, void **ret,
void** args, ffi_cif* cif);
void FFI_HIDDEN ffi_closure_SYSV (ffi_closure *)
__attribute__ ((regparm(1)));
unsigned int FFI_HIDDEN ffi_closure_SYSV_inner (ffi_closure *, void **, void *)
__attribute__ ((regparm(1)));
void FFI_HIDDEN ffi_closure_raw_SYSV (ffi_raw_closure *)
__attribute__ ((regparm(1)));
#ifdef X86_WIN32
void FFI_HIDDEN ffi_closure_STDCALL (ffi_closure *)
__attribute__ ((regparm(1)));
#endif
#ifdef X86_WIN64
void FFI_HIDDEN ffi_closure_win64 (ffi_closure *);
#endif
/* This function is jumped to by the trampoline */
#ifdef X86_WIN64
void * FFI_HIDDEN
ffi_closure_win64_inner (ffi_closure *closure, void *args) {
ffi_cif *cif;
void **arg_area;
void *result;
void *resp = &result;
cif = closure->cif;
arg_area = (void**) alloca (cif->nargs * sizeof (void*));
/* this call will initialize ARG_AREA, such that each
* element in that array points to the corresponding
* value on the stack; and if the function returns
* a structure, it will change RESP to point to the
* structure return address. */
ffi_prep_incoming_args_SYSV(args, &resp, arg_area, cif);
(closure->fun) (cif, resp, arg_area, closure->user_data);
/* The result is returned in rax. This does the right thing for
result types except for floats; we have to 'mov xmm0, rax' in the
caller to correct this.
TODO: structure sizes of 3 5 6 7 are returned by reference, too!!!
*/
return cif->rtype->size > sizeof(void *) ? resp : *(void **)resp;
}
#else
unsigned int FFI_HIDDEN __attribute__ ((regparm(1)))
ffi_closure_SYSV_inner (ffi_closure *closure, void **respp, void *args)
{
/* our various things... */
ffi_cif *cif;
void **arg_area;
cif = closure->cif;
arg_area = (void**) alloca (cif->nargs * sizeof (void*));
/* this call will initialize ARG_AREA, such that each
* element in that array points to the corresponding
* value on the stack; and if the function returns
* a structure, it will change RESP to point to the
* structure return address. */
ffi_prep_incoming_args_SYSV(args, respp, arg_area, cif);
(closure->fun) (cif, *respp, arg_area, closure->user_data);
return cif->flags;
}
#endif /* !X86_WIN64 */
static void
ffi_prep_incoming_args_SYSV(char *stack, void **rvalue, void **avalue,
ffi_cif *cif)
{
register unsigned int i;
register void **p_argv;
register char *argp;
register ffi_type **p_arg;
argp = stack;
#ifdef X86_WIN64
if (cif->rtype->size > sizeof(ffi_arg)
|| (cif->flags == FFI_TYPE_STRUCT
&& (cif->rtype->size != 1 && cif->rtype->size != 2
&& cif->rtype->size != 4 && cif->rtype->size != 8))) {
*rvalue = *(void **) argp;
argp += sizeof(void *);
}
#else
if ( cif->flags == FFI_TYPE_STRUCT ) {
*rvalue = *(void **) argp;
argp += sizeof(void *);
}
#endif
p_argv = avalue;
for (i = cif->nargs, p_arg = cif->arg_types; (i != 0); i--, p_arg++)
{
size_t z;
/* Align if necessary */
if ((sizeof(void*) - 1) & (size_t) argp) {
argp = (char *) ALIGN(argp, sizeof(void*));
}
#ifdef X86_WIN64
if ((*p_arg)->size > sizeof(ffi_arg)
|| ((*p_arg)->type == FFI_TYPE_STRUCT
&& ((*p_arg)->size != 1 && (*p_arg)->size != 2
&& (*p_arg)->size != 4 && (*p_arg)->size != 8)))
{
z = sizeof(void *);
*p_argv = *(void **)argp;
}
else
#endif
{
z = (*p_arg)->size;
/* because we're little endian, this is what it turns into. */
*p_argv = (void*) argp;
}
p_argv++;
#ifdef X86_WIN64
argp += (z + sizeof(void*) - 1) & ~(sizeof(void*) - 1);
#else
argp += z;
#endif
}
return;
}
#define FFI_INIT_TRAMPOLINE_WIN64(TRAMP,FUN,CTX,MASK) \
{ unsigned char *__tramp = (unsigned char*)(TRAMP); \
void* __fun = (void*)(FUN); \
void* __ctx = (void*)(CTX); \
*(unsigned char*) &__tramp[0] = 0x41; \
*(unsigned char*) &__tramp[1] = 0xbb; \
*(unsigned int*) &__tramp[2] = MASK; /* mov $mask, %r11 */ \
*(unsigned char*) &__tramp[6] = 0x48; \
*(unsigned char*) &__tramp[7] = 0xb8; \
*(void**) &__tramp[8] = __ctx; /* mov __ctx, %rax */ \
*(unsigned char *) &__tramp[16] = 0x49; \
*(unsigned char *) &__tramp[17] = 0xba; \
*(void**) &__tramp[18] = __fun; /* mov __fun, %r10 */ \
*(unsigned char *) &__tramp[26] = 0x41; \
*(unsigned char *) &__tramp[27] = 0xff; \
*(unsigned char *) &__tramp[28] = 0xe2; /* jmp %r10 */ \
}
/* How to make a trampoline. Derived from gcc/config/i386/i386.c. */
#define FFI_INIT_TRAMPOLINE(TRAMP,FUN,CTX) \
{ unsigned char *__tramp = (unsigned char*)(TRAMP); \
unsigned int __fun = (unsigned int)(FUN); \
unsigned int __ctx = (unsigned int)(CTX); \
unsigned int __dis = __fun - (__ctx + 10); \
*(unsigned char*) &__tramp[0] = 0xb8; \
*(unsigned int*) &__tramp[1] = __ctx; /* movl __ctx, %eax */ \
*(unsigned char *) &__tramp[5] = 0xe9; \
*(unsigned int*) &__tramp[6] = __dis; /* jmp __fun */ \
}
#define FFI_INIT_TRAMPOLINE_STDCALL(TRAMP,FUN,CTX,SIZE) \
{ unsigned char *__tramp = (unsigned char*)(TRAMP); \
unsigned int __fun = (unsigned int)(FUN); \
unsigned int __ctx = (unsigned int)(CTX); \
unsigned int __dis = __fun - (__ctx + 10); \
unsigned short __size = (unsigned short)(SIZE); \
*(unsigned char*) &__tramp[0] = 0xb8; \
*(unsigned int*) &__tramp[1] = __ctx; /* movl __ctx, %eax */ \
*(unsigned char *) &__tramp[5] = 0xe8; \
*(unsigned int*) &__tramp[6] = __dis; /* call __fun */ \
*(unsigned char *) &__tramp[10] = 0xc2; \
*(unsigned short*) &__tramp[11] = __size; /* ret __size */ \
}
/* the cif must already be prep'ed */
ffi_status
ffi_prep_closure_loc (ffi_closure* closure,
ffi_cif* cif,
void (*fun)(ffi_cif*,void*,void**,void*),
void *user_data,
void *codeloc)
{
#ifdef X86_WIN64
#define ISFLOAT(IDX) (cif->arg_types[IDX]->type == FFI_TYPE_FLOAT || cif->arg_types[IDX]->type == FFI_TYPE_DOUBLE)
#define FLAG(IDX) (cif->nargs>(IDX)&&ISFLOAT(IDX)?(1<<(IDX)):0)
if (cif->abi == FFI_WIN64)
{
int mask = FLAG(0)|FLAG(1)|FLAG(2)|FLAG(3);
FFI_INIT_TRAMPOLINE_WIN64 (&closure->tramp[0],
&ffi_closure_win64,
codeloc, mask);
/* make sure we can execute here */
}
#else
if (cif->abi == FFI_SYSV)
{
FFI_INIT_TRAMPOLINE (&closure->tramp[0],
&ffi_closure_SYSV,
(void*)codeloc);
}
#ifdef X86_WIN32
else if (cif->abi == FFI_STDCALL)
{
FFI_INIT_TRAMPOLINE_STDCALL (&closure->tramp[0],
&ffi_closure_STDCALL,
(void*)codeloc, cif->bytes);
}
#endif /* X86_WIN32 */
#endif /* !X86_WIN64 */
else
{
return FFI_BAD_ABI;
}
closure->cif = cif;
closure->user_data = user_data;
closure->fun = fun;
return FFI_OK;
}
/* ------- Native raw API support -------------------------------- */
#if !FFI_NO_RAW_API
ffi_status
ffi_prep_raw_closure_loc (ffi_raw_closure* closure,
ffi_cif* cif,
void (*fun)(ffi_cif*,void*,ffi_raw*,void*),
void *user_data,
void *codeloc)
{
int i;
if (cif->abi != FFI_SYSV) {
return FFI_BAD_ABI;
}
/* we currently don't support certain kinds of arguments for raw
closures. This should be implemented by a separate assembly
language routine, since it would require argument processing,
something we don't do now for performance. */
for (i = cif->nargs-1; i >= 0; i--)
{
FFI_ASSERT (cif->arg_types[i]->type != FFI_TYPE_STRUCT);
FFI_ASSERT (cif->arg_types[i]->type != FFI_TYPE_LONGDOUBLE);
}
FFI_INIT_TRAMPOLINE (&closure->tramp[0], &ffi_closure_raw_SYSV,
codeloc);
closure->cif = cif;
closure->user_data = user_data;
closure->fun = fun;
return FFI_OK;
}
static void
ffi_prep_args_raw(char *stack, extended_cif *ecif)
{
memcpy (stack, ecif->avalue, ecif->cif->bytes);
}
/* we borrow this routine from libffi (it must be changed, though, to
* actually call the function passed in the first argument. as of
* libffi-1.20, this is not the case.)
*/
void
ffi_raw_call(ffi_cif *cif, void (*fn)(void), void *rvalue, ffi_raw *fake_avalue)
{
extended_cif ecif;
void **avalue = (void **)fake_avalue;
ecif.cif = cif;
ecif.avalue = avalue;
/* If the return value is a struct and we don't have a return */
/* value address then we need to make one */
if ((rvalue == NULL) &&
(cif->rtype->type == FFI_TYPE_STRUCT))
{
ecif.rvalue = alloca(cif->rtype->size);
}
else
ecif.rvalue = rvalue;
switch (cif->abi)
{
#ifdef X86_WIN32
case FFI_SYSV:
case FFI_STDCALL:
ffi_call_win32(ffi_prep_args_raw, &ecif, cif->bytes, cif->flags,
ecif.rvalue, fn);
break;
#else
case FFI_SYSV:
ffi_call_SYSV(ffi_prep_args_raw, &ecif, cif->bytes, cif->flags,
ecif.rvalue, fn);
break;
#endif
default:
FFI_ASSERT(0);
break;
}
}
#endif
#endif /* !__x86_64__ || X86_WIN64 */

View File

@@ -8,6 +8,14 @@
* src/closures.c (selinux_enabled_check): Fix strncmp usage bug.
2010-05-11 Dan Witte <dwitte@mozilla.com>
* doc/libffi.tex: Document previous change.
2010-05-11 Makoto Kato <m_kato@ga2.so-net.ne.jp>
* src/x86/ffi.c (ffi_call): Don't copy structs passed by value.
2010-05-05 Michael Kohler <michaelkohler@live.com>
* src/dlmalloc.c (dlfree): Fix spelling.

View File

@@ -171,7 +171,9 @@ discarded.
@var{avalues} is a vector of @code{void *} pointers that point to the
memory locations holding the argument values for a call. If @var{cif}
declares that the function has no arguments (i.e., @var{nargs} was 0),
then @var{avalues} is ignored.
then @var{avalues} is ignored. Note that argument values may be
modified by the callee (for instance, structs passed by value); the
burden of copying pass-by-value arguments is placed on the caller.
@end defun

View File

@@ -9,3 +9,4 @@ os2
spelling
selinux
ffi_last_abi
win64-struct-args

68
patches/win64-struct-args Normal file
View File

@@ -0,0 +1,68 @@
Index: libffi/doc/libffi.texi
===================================================================
--- libffi.orig/doc/libffi.texi
+++ libffi/doc/libffi.texi
@@ -171,7 +171,9 @@ discarded.
@var{avalues} is a vector of @code{void *} pointers that point to the
memory locations holding the argument values for a call. If @var{cif}
declares that the function has no arguments (i.e., @var{nargs} was 0),
-then @var{avalues} is ignored.
+then @var{avalues} is ignored. Note that argument values may be
+modified by the callee (for instance, structs passed by value); the
+burden of copying pass-by-value arguments is placed on the caller.
@end defun
Index: libffi/src/x86/ffi.c
===================================================================
--- libffi.orig/src/x86/ffi.c
+++ libffi/src/x86/ffi.c
@@ -291,27 +291,8 @@ void ffi_call(ffi_cif *cif, void (*fn)(v
{
#ifdef X86_WIN64
case FFI_WIN64:
- {
- /* Make copies of all struct arguments
- NOTE: not sure if responsibility should be here or in caller */
- unsigned int i;
- for (i=0; i < cif->nargs;i++) {
- size_t size = cif->arg_types[i]->size;
- if ((cif->arg_types[i]->type == FFI_TYPE_STRUCT
- && (size != 1 && size != 2 && size != 4 && size != 8))
-#if FFI_TYPE_LONGDOUBLE != FFI_TYPE_DOUBLE
- || cif->arg_types[i]->type == FFI_TYPE_LONGDOUBLE
-#endif
- )
- {
- void *local = alloca(size);
- memcpy(local, avalue[i], size);
- avalue[i] = local;
- }
- }
- ffi_call_win64(ffi_prep_args, &ecif, cif->bytes,
- cif->flags, ecif.rvalue, fn);
- }
+ ffi_call_win64(ffi_prep_args, &ecif, cif->bytes,
+ cif->flags, ecif.rvalue, fn);
break;
#elif defined(X86_WIN32)
case FFI_SYSV:
Index: libffi/ChangeLog
===================================================================
--- libffi.orig/ChangeLog
+++ libffi/ChangeLog
@@ -8,6 +8,14 @@
* src/closures.c (selinux_enabled_check): Fix strncmp usage bug.
+2010-05-11 Dan Witte <dwitte@mozilla.com>
+
+ * doc/libffi.tex: Document previous change.
+
+2010-05-11 Makoto Kato <m_kato@ga2.so-net.ne.jp>
+
+ * src/x86/ffi.c (ffi_call): Don't copy structs passed by value.
+
2010-05-05 Michael Kohler <michaelkohler@live.com>
* src/dlmalloc.c (dlfree): Fix spelling.

View File

@@ -291,27 +291,8 @@ void ffi_call(ffi_cif *cif, void (*fn)(void), void *rvalue, void **avalue)
{
#ifdef X86_WIN64
case FFI_WIN64:
{
/* Make copies of all struct arguments
NOTE: not sure if responsibility should be here or in caller */
unsigned int i;
for (i=0; i < cif->nargs;i++) {
size_t size = cif->arg_types[i]->size;
if ((cif->arg_types[i]->type == FFI_TYPE_STRUCT
&& (size != 1 && size != 2 && size != 4 && size != 8))
#if FFI_TYPE_LONGDOUBLE != FFI_TYPE_DOUBLE
|| cif->arg_types[i]->type == FFI_TYPE_LONGDOUBLE
#endif
)
{
void *local = alloca(size);
memcpy(local, avalue[i], size);
avalue[i] = local;
}
}
ffi_call_win64(ffi_prep_args, &ecif, cif->bytes,
cif->flags, ecif.rvalue, fn);
}
ffi_call_win64(ffi_prep_args, &ecif, cif->bytes,
cif->flags, ecif.rvalue, fn);
break;
#elif defined(X86_WIN32)
case FFI_SYSV: