ag/cpython-source-deps
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

abi check fixes and Linux/x32 support

Browse Source
This commit is contained in:
Anthony Green
2012-03-03 14:46:20 -05:00
parent 6c194233a5
commit 964c5b93f8
17 changed files with 11020 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
.pc/applied-patches
View File

@@ -43,3 +43,5 @@ alpha
autoconf-archive-update autoconf-archive-update
ffitarget-include-fix ffitarget-include-fix
no-undefined-fix no-undefined-fix
x86_abi_check_fix
x32-abi

0
.pc/x32-abi/.timestamp Normal file
View File

4768
.pc/x32-abi/ChangeLog Normal file
View File

File diff suppressed because it is too large Load Diff

349
.pc/x32-abi/README Normal file
View File

@@ -0,0 +1,349 @@
Status
======
libffi-3.0.11 was released on *****************. Check the libffi web
page for updates: <URL:http://sourceware.org/libffi/>.
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 "calling
convention". The "calling convention" is essentially 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.
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. Libffi can be used in such programs to provide a
bridge from the interpreter program to compiled code.
The 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.
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
libffi library really only provides the lowest, machine dependent
layer of a fully featured foreign function interface. A layer must
exist above libffi that handles type conversions for values passed
between the two languages.
Supported Platforms
===================
Libffi has been ported to many different platforms.
For specific configuration details and testing status, please
refer to the wiki page here:
http://www.moxielogic.org/wiki/index.php?title=Libffi_3.0.11
At the time of release, the following basic configurations have been
tested:
|--------------+------------------|
| Architecture | Operating System |
|--------------+------------------|
| Alpha | Linux |
| Alpha | Tru64 |
| ARM | Linux |
| ARM | iOS |
| AVR32 | Linux |
| HPPA | HPUX |
| IA-64 | Linux |
| M68K | RTEMS |
| MIPS | IRIX |
| MIPS | Linux |
| MIPS | RTEMS |
| MIPS64 | Linux |
| PowerPC | AMIGA |
| PowerPC | Linux |
| PowerPC | Mac OSX |
| PowerPC | FreeBSD |
| PowerPC64 | Linux |
| S390 | Linux |
| S390X | Linux |
| SPARC | Linux |
| SPARC | Solaris |
| SPARC64 | Linux |
| SPARC64 | FreeBSD |
| X86 | FreeBSD |
| X86 | Interix |
| X86 | kFreeBSD |
| X86 | Linux |
| X86 | Mac OSX |
| X86 | OpenBSD |
| X86 | OS/2 |
| X86 | Solaris |
| X86 | Windows/Cygwin |
| X86 | Windows/MingW |
| X86-64 | FreeBSD |
| X86-64 | Linux |
| X86-64 | OpenBSD |
| X86-64 | Windows/MingW |
|--------------+------------------|
Please send additional platform test results to
libffi-discuss@sourceware.org and feel free to update the wiki page
above.
Installing libffi
=================
First you must configure the distribution for your particular
system. Go to the directory you wish to build libffi in and run the
"configure" program found in the root directory of the libffi source
distribution.
You may want to tell configure where to install the libffi library and
header files. To do that, use the --prefix configure switch. Libffi
will install under /usr/local by default.
If you want to enable extra run-time debugging checks use the the
--enable-debug configure switch. This is useful when your program dies
mysteriously while using libffi.
Another useful configure switch is --enable-purify-safety. Using this
will add some extra code which will suppress certain warnings when you
are using Purify with libffi. Only use this switch when using
Purify, as it will slow down the library.
It's also possible to build libffi on Windows platforms with
Microsoft's Visual C++ compiler. In this case, use the msvcc.sh
wrapper script during configuration like so:
path/to/configure CC=path/to/msvcc.sh LD=link CPP=\"cl -nologo -EP\"
For 64-bit Windows builds, use CC="path/to/msvcc.sh -m64".
You may also need to specify --build appropriately. When building with MSVC
under a MingW environment, you may need to remove the line in configure
that sets 'fix_srcfile_path' to a 'cygpath' command. ('cygpath' is not
present in MingW, and is not required when using MingW-style paths.)
For iOS builds, refer to the build-ios.sh script for guidance.
Configure has many other options. Use "configure --help" to see them all.
Once configure has finished, type "make". Note that you must be using
GNU make. You can ftp GNU make from prep.ai.mit.edu:/pub/gnu.
To ensure that libffi is working as advertised, type "make check".
This will require that you have DejaGNU installed.
To install the library and header files, type "make install".
History
=======
See the ChangeLog files for details.
3.0.11 MMM-DD-YY
Lots of build fixes.
Add Amiga newer MacOS support.
Fix Octeon and MC68881 support.
Fix code pessimizations.
3.0.10 Aug-23-11
Add support for Apple's iOS.
Add support for ARM VFP ABI.
Add RTEMS support for MIPS and M68K.
Fix instruction cache clearing problems on
ARM and SPARC.
Fix the N64 build on mips-sgi-irix6.5.
Enable builds with Microsoft's compiler.
Enable x86 builds with Oracle's Solaris compiler.
Fix support for calling code compiled with Oracle's Sparc
Solaris compiler.
Testsuite fixes for Tru64 Unix.
Additional platform support.
3.0.9 Dec-31-09
Add AVR32 and win64 ports. Add ARM softfp support.
Many fixes for AIX, Solaris, HP-UX, *BSD.
Several PowerPC and x86-64 bug fixes.
Build DLL for windows.
3.0.8 Dec-19-08
Add *BSD, BeOS, and PA-Linux support.
3.0.7 Nov-11-08
Fix for ppc FreeBSD.
(thanks to Andreas Tobler)
3.0.6 Jul-17-08
Fix for closures on sh.
Mark the sh/sh64 stack as non-executable.
(both thanks to Kaz Kojima)
3.0.5 Apr-3-08
Fix libffi.pc file.
Fix #define ARM for IcedTea users.
Fix x86 closure bug.
3.0.4 Feb-24-08
Fix x86 OpenBSD configury.
3.0.3 Feb-22-08
Enable x86 OpenBSD thanks to Thomas Heller, and
x86-64 FreeBSD thanks to Björn König and Andreas Tobler.
Clean up test instruction in README.
3.0.2 Feb-21-08
Improved x86 FreeBSD support.
Thanks to Björn König.
3.0.1 Feb-15-08
Fix instruction cache flushing bug on MIPS.
Thanks to David Daney.
3.0.0 Feb-15-08
Many changes, mostly thanks to the GCC project.
Cygnus Solutions is now Red Hat.
[10 years go by...]
1.20 Oct-5-98
Raffaele Sena produces ARM port.
1.19 Oct-5-98
Fixed x86 long double and long long return support.
m68k bug fixes from Andreas Schwab.
Patch for DU assembler compatibility for the Alpha from Richard
Henderson.
1.18 Apr-17-98
Bug fixes and MIPS configuration changes.
1.17 Feb-24-98
Bug fixes and m68k port from Andreas Schwab. PowerPC port from
Geoffrey Keating. Various bug x86, Sparc and MIPS bug fixes.
1.16 Feb-11-98
Richard Henderson produces Alpha port.
1.15 Dec-4-97
Fixed an n32 ABI bug. New libtool, auto* support.
1.14 May-13-97
libtool is now used to generate shared and static libraries.
Fixed a minor portability problem reported by Russ McManus
<mcmanr@eq.gs.com>.
1.13 Dec-2-96
Added --enable-purify-safety to keep Purify from complaining
about certain low level code.
Sparc fix for calling functions with < 6 args.
Linux x86 a.out fix.
1.12 Nov-22-96
Added missing ffi_type_void, needed for supporting void return
types. Fixed test case for non MIPS machines. Cygnus Support
is now Cygnus Solutions.
1.11 Oct-30-96
Added notes about GNU make.
1.10 Oct-29-96
Added configuration fix for non GNU compilers.
1.09 Oct-29-96
Added --enable-debug configure switch. Clean-ups based on LCLint
feedback. ffi_mips.h is always installed. Many configuration
fixes. Fixed ffitest.c for sparc builds.
1.08 Oct-15-96
Fixed n32 problem. Many clean-ups.
1.07 Oct-14-96
Gordon Irlam rewrites v8.S again. Bug fixes.
1.06 Oct-14-96
Gordon Irlam improved the sparc port.
1.05 Oct-14-96
Interface changes based on feedback.
1.04 Oct-11-96
Sparc port complete (modulo struct passing bug).
1.03 Oct-10-96
Passing struct args, and returning struct values works for
all architectures/calling conventions. Expanded tests.
1.02 Oct-9-96
Added SGI n32 support. Fixed bugs in both o32 and Linux support.
Added "make test".
1.01 Oct-8-96
Fixed float passing bug in mips version. Restructured some
of the code. Builds cleanly with SGI tools.
1.00 Oct-7-96
First release. No public announcement.
Authors & Credits
=================
libffi was originally written by Anthony Green <green@redhat.com>.
The developers of the GNU Compiler Collection project have made
innumerable valuable contributions. See the ChangeLog file for
details.
Some of the ideas behind libffi were inspired by Gianni Mariani's free
gencall library for Silicon Graphics machines.
The closure mechanism was designed and implemented by Kresten Krab
Thorup.
Major processor architecture ports were contributed by the following
developers:
alpha Richard Henderson
arm Raffaele Sena
cris Simon Posnjak, Hans-Peter Nilsson
frv Anthony Green
ia64 Hans Boehm
m32r Kazuhiro Inaoka
m68k Andreas Schwab
mips Anthony Green, Casey Marshall
mips64 David Daney
pa Randolph Chung, Dave Anglin, Andreas Tobler
powerpc Geoffrey Keating, Andreas Tobler,
David Edelsohn, John Hornkvist
powerpc64 Jakub Jelinek
s390 Gerhard Tonn, Ulrich Weigand
sh Kaz Kojima
sh64 Kaz Kojima
sparc Anthony Green, Gordon Irlam
x86 Anthony Green, Jon Beniston
x86-64 Bo Thorsen
Jesper Skov and Andrew Haley both did more than their fair share of
stepping through the code and tracking down bugs.
Thanks also to Tom Tromey for bug fixes, documentation and
configuration help.
Thanks to Jim Blandy, who provided some useful feedback on the libffi
interface.
Andreas Tobler has done a tremendous amount of work on the testsuite.
Alex Oliva solved the executable page problem for SElinux.
The list above is almost certainly incomplete and inaccurate. I'm
happy to make corrections or additions upon request.
If you have a problem, or have found a bug, please send a note to the
author at green@moxielogic.com, or the project mailing list at
libffi-discuss@sourceware.org.

635
.pc/x32-abi/src/x86/ffi64.c Normal file
View File

@@ -0,0 +1,635 @@
/* -----------------------------------------------------------------------
ffi64.c - Copyright (c) 20011 Anthony Green
Copyright (c) 2008, 2010 Red Hat, Inc.
Copyright (c) 2002, 2007 Bo Thorsen <bo@suse.de>
x86-64 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.
----------------------------------------------------------------------- */
#include <ffi.h>
#include <ffi_common.h>
#include <stdlib.h>
#include <stdarg.h>
#ifdef __x86_64__
#define MAX_GPR_REGS 6
#define MAX_SSE_REGS 8
struct register_args
{
/* Registers for argument passing. */
UINT64 gpr[MAX_GPR_REGS];
__int128_t sse[MAX_SSE_REGS];
};
extern void ffi_call_unix64 (void *args, unsigned long bytes, unsigned flags,
void *raddr, void (*fnaddr)(void), unsigned ssecount);
/* All reference to register classes here is identical to the code in
gcc/config/i386/i386.c. Do *not* change one without the other. */
/* Register class used for passing given 64bit part of the argument.
These represent classes as documented by the PS ABI, with the
exception of SSESF, SSEDF classes, that are basically SSE class,
just gcc will use SF or DFmode move instead of DImode to avoid
reformatting penalties.
Similary we play games with INTEGERSI_CLASS to use cheaper SImode moves
whenever possible (upper half does contain padding). */
enum x86_64_reg_class
{
X86_64_NO_CLASS,
X86_64_INTEGER_CLASS,
X86_64_INTEGERSI_CLASS,
X86_64_SSE_CLASS,
X86_64_SSESF_CLASS,
X86_64_SSEDF_CLASS,
X86_64_SSEUP_CLASS,
X86_64_X87_CLASS,
X86_64_X87UP_CLASS,
X86_64_COMPLEX_X87_CLASS,
X86_64_MEMORY_CLASS
};
#define MAX_CLASSES 4
#define SSE_CLASS_P(X) ((X) >= X86_64_SSE_CLASS && X <= X86_64_SSEUP_CLASS)
/* x86-64 register passing implementation. See x86-64 ABI for details. Goal
of this code is to classify each 8bytes of incoming argument by the register
class and assign registers accordingly. */
/* Return the union class of CLASS1 and CLASS2.
See the x86-64 PS ABI for details. */
static enum x86_64_reg_class
merge_classes (enum x86_64_reg_class class1, enum x86_64_reg_class class2)
{
/* Rule #1: If both classes are equal, this is the resulting class. */
if (class1 == class2)
return class1;
/* Rule #2: If one of the classes is NO_CLASS, the resulting class is
the other class. */
if (class1 == X86_64_NO_CLASS)
return class2;
if (class2 == X86_64_NO_CLASS)
return class1;
/* Rule #3: If one of the classes is MEMORY, the result is MEMORY. */
if (class1 == X86_64_MEMORY_CLASS || class2 == X86_64_MEMORY_CLASS)
return X86_64_MEMORY_CLASS;
/* Rule #4: If one of the classes is INTEGER, the result is INTEGER. */
if ((class1 == X86_64_INTEGERSI_CLASS && class2 == X86_64_SSESF_CLASS)
|| (class2 == X86_64_INTEGERSI_CLASS && class1 == X86_64_SSESF_CLASS))
return X86_64_INTEGERSI_CLASS;
if (class1 == X86_64_INTEGER_CLASS || class1 == X86_64_INTEGERSI_CLASS
|| class2 == X86_64_INTEGER_CLASS || class2 == X86_64_INTEGERSI_CLASS)
return X86_64_INTEGER_CLASS;
/* Rule #5: If one of the classes is X87, X87UP, or COMPLEX_X87 class,
MEMORY is used. */
if (class1 == X86_64_X87_CLASS
|| class1 == X86_64_X87UP_CLASS
|| class1 == X86_64_COMPLEX_X87_CLASS
|| class2 == X86_64_X87_CLASS
|| class2 == X86_64_X87UP_CLASS
|| class2 == X86_64_COMPLEX_X87_CLASS)
return X86_64_MEMORY_CLASS;
/* Rule #6: Otherwise class SSE is used. */
return X86_64_SSE_CLASS;
}
/* Classify the argument of type TYPE and mode MODE.
CLASSES will be filled by the register class used to pass each word
of the operand. The number of words is returned. In case the parameter
should be passed in memory, 0 is returned. As a special case for zero
sized containers, classes[0] will be NO_CLASS and 1 is returned.
See the x86-64 PS ABI for details.
*/
static int
classify_argument (ffi_type *type, enum x86_64_reg_class classes[],
size_t byte_offset)
{
switch (type->type)
{
case FFI_TYPE_UINT8:
case FFI_TYPE_SINT8:
case FFI_TYPE_UINT16:
case FFI_TYPE_SINT16:
case FFI_TYPE_UINT32:
case FFI_TYPE_SINT32:
case FFI_TYPE_UINT64:
case FFI_TYPE_SINT64:
case FFI_TYPE_POINTER:
{
int size = byte_offset + type->size;
if (size <= 4)
{
classes[0] = X86_64_INTEGERSI_CLASS;
return 1;
}
else if (size <= 8)
{
classes[0] = X86_64_INTEGER_CLASS;
return 1;
}
else if (size <= 12)
{
classes[0] = X86_64_INTEGER_CLASS;
classes[1] = X86_64_INTEGERSI_CLASS;
return 2;
}
else if (size <= 16)
{
classes[0] = classes[1] = X86_64_INTEGERSI_CLASS;
return 2;
}
else
FFI_ASSERT (0);
}
case FFI_TYPE_FLOAT:
if (!(byte_offset % 8))
classes[0] = X86_64_SSESF_CLASS;
else
classes[0] = X86_64_SSE_CLASS;
return 1;
case FFI_TYPE_DOUBLE:
classes[0] = X86_64_SSEDF_CLASS;
return 1;
case FFI_TYPE_LONGDOUBLE:
classes[0] = X86_64_X87_CLASS;
classes[1] = X86_64_X87UP_CLASS;
return 2;
case FFI_TYPE_STRUCT:
{
const int UNITS_PER_WORD = 8;
int words = (type->size + UNITS_PER_WORD - 1) / UNITS_PER_WORD;
ffi_type **ptr;
int i;
enum x86_64_reg_class subclasses[MAX_CLASSES];
/* If the struct is larger than 32 bytes, pass it on the stack. */
if (type->size > 32)
return 0;
for (i = 0; i < words; i++)
classes[i] = X86_64_NO_CLASS;
/* Zero sized arrays or structures are NO_CLASS. We return 0 to
signalize memory class, so handle it as special case. */
if (!words)
{
classes[0] = X86_64_NO_CLASS;
return 1;
}
/* Merge the fields of structure. */
for (ptr = type->elements; *ptr != NULL; ptr++)
{
int num;
byte_offset = ALIGN (byte_offset, (*ptr)->alignment);
num = classify_argument (*ptr, subclasses, byte_offset % 8);
if (num == 0)
return 0;
for (i = 0; i < num; i++)
{
int pos = byte_offset / 8;
classes[i + pos] =
merge_classes (subclasses[i], classes[i + pos]);
}
byte_offset += (*ptr)->size;
}
if (words > 2)
{
/* When size > 16 bytes, if the first one isn't
X86_64_SSE_CLASS or any other ones aren't
X86_64_SSEUP_CLASS, everything should be passed in
memory. */
if (classes[0] != X86_64_SSE_CLASS)
return 0;
for (i = 1; i < words; i++)
if (classes[i] != X86_64_SSEUP_CLASS)
return 0;
}
/* Final merger cleanup. */
for (i = 0; i < words; i++)
{
/* If one class is MEMORY, everything should be passed in
memory. */
if (classes[i] == X86_64_MEMORY_CLASS)
return 0;
/* The X86_64_SSEUP_CLASS should be always preceded by
X86_64_SSE_CLASS or X86_64_SSEUP_CLASS. */
if (classes[i] == X86_64_SSEUP_CLASS
&& classes[i - 1] != X86_64_SSE_CLASS
&& classes[i - 1] != X86_64_SSEUP_CLASS)
{
/* The first one should never be X86_64_SSEUP_CLASS. */
FFI_ASSERT (i != 0);
classes[i] = X86_64_SSE_CLASS;
}
/* If X86_64_X87UP_CLASS isn't preceded by X86_64_X87_CLASS,
everything should be passed in memory. */
if (classes[i] == X86_64_X87UP_CLASS
&& (classes[i - 1] != X86_64_X87_CLASS))
{
/* The first one should never be X86_64_X87UP_CLASS. */
FFI_ASSERT (i != 0);
return 0;
}
}
return words;
}
default:
FFI_ASSERT(0);
}
return 0; /* Never reached. */
}
/* Examine the argument and return set number of register required in each
class. Return zero iff parameter should be passed in memory, otherwise
the number of registers. */
static int
examine_argument (ffi_type *type, enum x86_64_reg_class classes[MAX_CLASSES],
_Bool in_return, int *pngpr, int *pnsse)
{
int i, n, ngpr, nsse;
n = classify_argument (type, classes, 0);
if (n == 0)
return 0;
ngpr = nsse = 0;
for (i = 0; i < n; ++i)
switch (classes[i])
{
case X86_64_INTEGER_CLASS:
case X86_64_INTEGERSI_CLASS:
ngpr++;
break;
case X86_64_SSE_CLASS:
case X86_64_SSESF_CLASS:
case X86_64_SSEDF_CLASS:
nsse++;
break;
case X86_64_NO_CLASS:
case X86_64_SSEUP_CLASS:
break;
case X86_64_X87_CLASS:
case X86_64_X87UP_CLASS:
case X86_64_COMPLEX_X87_CLASS:
return in_return != 0;
default:
abort ();
}
*pngpr = ngpr;
*pnsse = nsse;
return n;
}
/* Perform machine dependent cif processing. */
ffi_status
ffi_prep_cif_machdep (ffi_cif *cif)
{
int gprcount, ssecount, i, avn, n, ngpr, nsse, flags;
enum x86_64_reg_class classes[MAX_CLASSES];
size_t bytes;
gprcount = ssecount = 0;
flags = cif->rtype->type;
if (flags != FFI_TYPE_VOID)
{
n = examine_argument (cif->rtype, classes, 1, &ngpr, &nsse);
if (n == 0)
{
/* The return value is passed in memory. A pointer to that
memory is the first argument. Allocate a register for it. */
gprcount++;
/* We don't have to do anything in asm for the return. */
flags = FFI_TYPE_VOID;
}
else if (flags == FFI_TYPE_STRUCT)
{
/* Mark which registers the result appears in. */
_Bool sse0 = SSE_CLASS_P (classes[0]);
_Bool sse1 = n == 2 && SSE_CLASS_P (classes[1]);
if (sse0 && !sse1)
flags |= 1 << 8;
else if (!sse0 && sse1)
flags |= 1 << 9;
else if (sse0 && sse1)
flags |= 1 << 10;
/* Mark the true size of the structure. */
flags |= cif->rtype->size << 12;
}
}
/* Go over all arguments and determine the way they should be passed.
If it's in a register and there is space for it, let that be so. If
not, add it's size to the stack byte count. */
for (bytes = 0, i = 0, avn = cif->nargs; i < avn; i++)
{
if (examine_argument (cif->arg_types[i], classes, 0, &ngpr, &nsse) == 0
|| gprcount + ngpr > MAX_GPR_REGS
|| ssecount + nsse > MAX_SSE_REGS)
{
long align = cif->arg_types[i]->alignment;
if (align < 8)
align = 8;
bytes = ALIGN (bytes, align);
bytes += cif->arg_types[i]->size;
}
else
{
gprcount += ngpr;
ssecount += nsse;
}
}
if (ssecount)
flags |= 1 << 11;
cif->flags = flags;
cif->bytes = ALIGN (bytes, 8);
return FFI_OK;
}
void
ffi_call (ffi_cif *cif, void (*fn)(void), void *rvalue, void **avalue)
{
enum x86_64_reg_class classes[MAX_CLASSES];
char *stack, *argp;
ffi_type **arg_types;
int gprcount, ssecount, ngpr, nsse, i, avn;
_Bool ret_in_memory;
struct register_args *reg_args;
/* Can't call 32-bit mode from 64-bit mode. */
FFI_ASSERT (cif->abi == FFI_UNIX64);
/* If the return value is a struct and we don't have a return value
address then we need to make one. Note the setting of flags to
VOID above in ffi_prep_cif_machdep. */
ret_in_memory = (cif->rtype->type == FFI_TYPE_STRUCT
&& (cif->flags & 0xff) == FFI_TYPE_VOID);
if (rvalue == NULL && ret_in_memory)
rvalue = alloca (cif->rtype->size);
/* Allocate the space for the arguments, plus 4 words of temp space. */
stack = alloca (sizeof (struct register_args) + cif->bytes + 4*8);
reg_args = (struct register_args *) stack;
argp = stack + sizeof (struct register_args);
gprcount = ssecount = 0;
/* If the return value is passed in memory, add the pointer as the
first integer argument. */
if (ret_in_memory)
reg_args->gpr[gprcount++] = (long) rvalue;
avn = cif->nargs;
arg_types = cif->arg_types;
for (i = 0; i < avn; ++i)
{
size_t size = arg_types[i]->size;
int n;
n = examine_argument (arg_types[i], classes, 0, &ngpr, &nsse);
if (n == 0
|| gprcount + ngpr > MAX_GPR_REGS
|| ssecount + nsse > MAX_SSE_REGS)
{
long align = arg_types[i]->alignment;
/* Stack arguments are *always* at least 8 byte aligned. */
if (align < 8)
align = 8;
/* Pass this argument in memory. */
argp = (void *) ALIGN (argp, align);
memcpy (argp, avalue[i], size);
argp += size;
}
else
{
/* The argument is passed entirely in registers. */
char *a = (char *) avalue[i];
int j;
for (j = 0; j < n; j++, a += 8, size -= 8)
{
switch (classes[j])
{
case X86_64_INTEGER_CLASS:
case X86_64_INTEGERSI_CLASS:
reg_args->gpr[gprcount] = 0;
memcpy (&reg_args->gpr[gprcount], a, size < 8 ? size : 8);
gprcount++;
break;
case X86_64_SSE_CLASS:
case X86_64_SSEDF_CLASS:
reg_args->sse[ssecount++] = *(UINT64 *) a;
break;
case X86_64_SSESF_CLASS:
reg_args->sse[ssecount++] = *(UINT32 *) a;
break;
default:
abort();
}
}
}
}
ffi_call_unix64 (stack, cif->bytes + sizeof (struct register_args),
cif->flags, rvalue, fn, ssecount);
}
extern void ffi_closure_unix64(void);
ffi_status
ffi_prep_closure_loc (ffi_closure* closure,
ffi_cif* cif,
void (*fun)(ffi_cif*, void*, void**, void*),
void *user_data,
void *codeloc)
{
volatile unsigned short *tramp;
/* Sanity check on the cif ABI. */
{
int abi = cif->abi;
if (UNLIKELY (! (abi > FFI_FIRST_ABI && abi < FFI_LAST_ABI)))
return FFI_BAD_ABI;
}
tramp = (volatile unsigned short *) &closure->tramp[0];
tramp[0] = 0xbb49; /* mov <code>, %r11 */
*(void * volatile *) &tramp[1] = ffi_closure_unix64;
tramp[5] = 0xba49; /* mov <data>, %r10 */
*(void * volatile *) &tramp[6] = codeloc;
/* Set the carry bit iff the function uses any sse registers.
This is clc or stc, together with the first byte of the jmp. */
tramp[10] = cif->flags & (1 << 11) ? 0x49f9 : 0x49f8;
tramp[11] = 0xe3ff; /* jmp *%r11 */
closure->cif = cif;
closure->fun = fun;
closure->user_data = user_data;
return FFI_OK;
}
int
ffi_closure_unix64_inner(ffi_closure *closure, void *rvalue,
struct register_args *reg_args, char *argp)
{
ffi_cif *cif;
void **avalue;
ffi_type **arg_types;
long i, avn;
int gprcount, ssecount, ngpr, nsse;
int ret;
cif = closure->cif;
avalue = alloca(cif->nargs * sizeof(void *));
gprcount = ssecount = 0;
ret = cif->rtype->type;
if (ret != FFI_TYPE_VOID)
{
enum x86_64_reg_class classes[MAX_CLASSES];
int n = examine_argument (cif->rtype, classes, 1, &ngpr, &nsse);
if (n == 0)
{
/* The return value goes in memory. Arrange for the closure
return value to go directly back to the original caller. */
rvalue = (void *) reg_args->gpr[gprcount++];
/* We don't have to do anything in asm for the return. */
ret = FFI_TYPE_VOID;
}
else if (ret == FFI_TYPE_STRUCT && n == 2)
{
/* Mark which register the second word of the structure goes in. */
_Bool sse0 = SSE_CLASS_P (classes[0]);
_Bool sse1 = SSE_CLASS_P (classes[1]);
if (!sse0 && sse1)
ret |= 1 << 8;
else if (sse0 && !sse1)
ret |= 1 << 9;
}
}
avn = cif->nargs;
arg_types = cif->arg_types;
for (i = 0; i < avn; ++i)
{
enum x86_64_reg_class classes[MAX_CLASSES];
int n;
n = examine_argument (arg_types[i], classes, 0, &ngpr, &nsse);
if (n == 0
|| gprcount + ngpr > MAX_GPR_REGS
|| ssecount + nsse > MAX_SSE_REGS)
{
long align = arg_types[i]->alignment;
/* Stack arguments are *always* at least 8 byte aligned. */
if (align < 8)
align = 8;
/* Pass this argument in memory. */
argp = (void *) ALIGN (argp, align);
avalue[i] = argp;
argp += arg_types[i]->size;
}
/* If the argument is in a single register, or two consecutive
integer registers, then we can use that address directly. */
else if (n == 1
|| (n == 2 && !(SSE_CLASS_P (classes[0])
|| SSE_CLASS_P (classes[1]))))
{
/* The argument is in a single register. */
if (SSE_CLASS_P (classes[0]))
{
avalue[i] = &reg_args->sse[ssecount];
ssecount += n;
}
else
{
avalue[i] = &reg_args->gpr[gprcount];
gprcount += n;
}
}
/* Otherwise, allocate space to make them consecutive. */
else
{
char *a = alloca (16);
int j;
avalue[i] = a;
for (j = 0; j < n; j++, a += 8)
{
if (SSE_CLASS_P (classes[j]))
memcpy (a, &reg_args->sse[ssecount++], 8);
else
memcpy (a, &reg_args->gpr[gprcount++], 8);
}
}
}
/* Invoke the closure. */
closure->fun (cif, rvalue, avalue, closure->user_data);
/* Tell assembly how to perform return type promotions. */
return ret;
}
#endif /* __x86_64__ */

128
.pc/x32-abi/src/x86/ffitarget.h Normal file
View File

@@ -0,0 +1,128 @@
/* -----------------------------------------------------------------*-C-*-
ffitarget.h - Copyright (c) 2012 Anthony Green
Copyright (c) 1996-2003, 2010 Red Hat, Inc.
Copyright (C) 2008 Free Software Foundation, Inc.
Target configuration macros for x86 and x86-64.
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.
----------------------------------------------------------------------- */
#ifndef LIBFFI_TARGET_H
#define LIBFFI_TARGET_H
#ifndef LIBFFI_H
#error "Please do not include ffitarget.h directly into your source. Use ffi.h instead."
#endif
/* ---- System specific configurations ----------------------------------- */
/* For code common to all platforms on x86 and x86_64. */
#define X86_ANY
#if defined (X86_64) && defined (__i386__)
#undef X86_64
#define X86
#endif
#ifdef X86_WIN64
#define FFI_SIZEOF_ARG 8
#define USE_BUILTIN_FFS 0 /* not yet implemented in mingw-64 */
#endif
/* ---- Generic type definitions ----------------------------------------- */
#ifndef LIBFFI_ASM
#ifdef X86_WIN64
#ifdef _MSC_VER
typedef unsigned __int64 ffi_arg;
typedef __int64 ffi_sarg;
#else
typedef unsigned long long ffi_arg;
typedef long long ffi_sarg;
#endif
#else
typedef unsigned long ffi_arg;
typedef signed long ffi_sarg;
#endif
typedef enum ffi_abi {
FFI_FIRST_ABI = 0,
/* ---- Intel x86 Win32 ---------- */
#ifdef X86_WIN32
FFI_SYSV,
FFI_STDCALL,
FFI_THISCALL,
FFI_FASTCALL,
FFI_LAST_ABI,
/* TODO: Add fastcall support for the sake of completeness */
FFI_DEFAULT_ABI = FFI_SYSV
#elif defined(X86_WIN64)
FFI_WIN64,
FFI_LAST_ABI,
FFI_DEFAULT_ABI = FFI_WIN64
#else
/* ---- Intel x86 and AMD x86-64 - */
FFI_SYSV,
FFI_UNIX64, /* Unix variants all use the same ABI for x86-64 */
FFI_LAST_ABI,
#if defined(__i386__) || defined(__i386)
FFI_DEFAULT_ABI = FFI_SYSV
#else
FFI_DEFAULT_ABI = FFI_UNIX64
#endif
#endif
} ffi_abi;
#endif
/* ---- Definitions for closures ----------------------------------------- */
#define FFI_CLOSURES 1
#define FFI_TYPE_SMALL_STRUCT_1B (FFI_TYPE_LAST + 1)
#define FFI_TYPE_SMALL_STRUCT_2B (FFI_TYPE_LAST + 2)
#define FFI_TYPE_SMALL_STRUCT_4B (FFI_TYPE_LAST + 3)
#if defined (X86_64) || (defined (__x86_64__) && defined (X86_DARWIN))
#define FFI_TRAMPOLINE_SIZE 24
#define FFI_NATIVE_RAW_API 0
#else
#ifdef X86_WIN32
#define FFI_TRAMPOLINE_SIZE 52
#else
#ifdef X86_WIN64
#define FFI_TRAMPOLINE_SIZE 29
#define FFI_NATIVE_RAW_API 0
#define FFI_NO_RAW_API 1
#else
#define FFI_TRAMPOLINE_SIZE 10
#endif
#endif
#ifndef X86_WIN64
#define FFI_NATIVE_RAW_API 1 /* x86 has native raw api support */
#endif
#endif
#endif

0
.pc/x86_abi_check_fix/.timestamp Normal file
View File

4764
.pc/x86_abi_check_fix/ChangeLog Normal file
View File

File diff suppressed because it is too large Load Diff

216
.pc/x86_abi_check_fix/src/prep_cif.c Normal file
View File

@@ -0,0 +1,216 @@
/* -----------------------------------------------------------------------
prep_cif.c - Copyright (c) 2011, 2012 Anthony Green
Copyright (c) 1996, 1998, 2007 Red Hat, Inc.
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.
----------------------------------------------------------------------- */
#include <ffi.h>
#include <ffi_common.h>
#include <stdlib.h>
/* Round up to FFI_SIZEOF_ARG. */
#define STACK_ARG_SIZE(x) ALIGN(x, FFI_SIZEOF_ARG)
/* Perform machine independent initialization of aggregate type
specifications. */
static ffi_status initialize_aggregate(ffi_type *arg)
{
ffi_type **ptr;
if (UNLIKELY(arg == NULL || arg->elements == NULL))
return FFI_BAD_TYPEDEF;
arg->size = 0;
arg->alignment = 0;
ptr = &(arg->elements[0]);
if (UNLIKELY(ptr == 0))
return FFI_BAD_TYPEDEF;
while ((*ptr) != NULL)
{
if (UNLIKELY(((*ptr)->size == 0)
&& (initialize_aggregate((*ptr)) != FFI_OK)))
return FFI_BAD_TYPEDEF;
/* Perform a sanity check on the argument type */
FFI_ASSERT_VALID_TYPE(*ptr);
arg->size = ALIGN(arg->size, (*ptr)->alignment);
arg->size += (*ptr)->size;
arg->alignment = (arg->alignment > (*ptr)->alignment) ?
arg->alignment : (*ptr)->alignment;
ptr++;
}
/* Structure size includes tail padding. This is important for
structures that fit in one register on ABIs like the PowerPC64
Linux ABI that right justify small structs in a register.
It's also needed for nested structure layout, for example
struct A { long a; char b; }; struct B { struct A x; char y; };
should find y at an offset of 2*sizeof(long) and result in a
total size of 3*sizeof(long). */
arg->size = ALIGN (arg->size, arg->alignment);
if (arg->size == 0)
return FFI_BAD_TYPEDEF;
else
return FFI_OK;
}
#ifndef __CRIS__
/* The CRIS ABI specifies structure elements to have byte
alignment only, so it completely overrides this functions,
which assumes "natural" alignment and padding. */
/* Perform machine independent ffi_cif preparation, then call
machine dependent routine. */
/* For non variadic functions isvariadic should be 0 and
nfixedargs==ntotalargs.
For variadic calls, isvariadic should be 1 and nfixedargs
and ntotalargs set as appropriate. nfixedargs must always be >=1 */
ffi_status FFI_HIDDEN ffi_prep_cif_core(ffi_cif *cif, ffi_abi abi,
unsigned int isvariadic,
unsigned int nfixedargs,
unsigned int ntotalargs,
ffi_type *rtype, ffi_type **atypes)
{
unsigned bytes = 0;
unsigned int i;
ffi_type **ptr;
FFI_ASSERT(cif != NULL);
FFI_ASSERT((!isvariadic) || (nfixedargs >= 1));
FFI_ASSERT(nfixedargs <= ntotalargs);
#ifndef X86_WIN32
if (! (abi > FFI_FIRST_ABI) && (abi <= FFI_LAST_ABI))
return FFI_BAD_ABI;
#else
if (! (abi > FFI_FIRST_ABI && abi < FFI_LAST_ABI || abi == FFI_THISCALL))
return FFI_BAD_ABI;
#endif
cif->abi = abi;
cif->arg_types = atypes;
cif->nargs = ntotalargs;
cif->rtype = rtype;
cif->flags = 0;
/* Initialize the return type if necessary */
if ((cif->rtype->size == 0) && (initialize_aggregate(cif->rtype) != FFI_OK))
return FFI_BAD_TYPEDEF;
/* Perform a sanity check on the return type */
FFI_ASSERT_VALID_TYPE(cif->rtype);
/* x86, x86-64 and s390 stack space allocation is handled in prep_machdep. */
#if !defined M68K && !defined X86_ANY && !defined S390 && !defined PA
/* Make space for the return structure pointer */
if (cif->rtype->type == FFI_TYPE_STRUCT
#ifdef SPARC
&& (cif->abi != FFI_V9 || cif->rtype->size > 32)
#endif
)
bytes = STACK_ARG_SIZE(sizeof(void*));
#endif
for (ptr = cif->arg_types, i = cif->nargs; i > 0; i--, ptr++)
{
/* Initialize any uninitialized aggregate type definitions */
if (((*ptr)->size == 0) && (initialize_aggregate((*ptr)) != FFI_OK))
return FFI_BAD_TYPEDEF;
/* Perform a sanity check on the argument type, do this
check after the initialization. */
FFI_ASSERT_VALID_TYPE(*ptr);
#if !defined X86_ANY && !defined S390 && !defined PA
#ifdef SPARC
if (((*ptr)->type == FFI_TYPE_STRUCT
&& ((*ptr)->size > 16 || cif->abi != FFI_V9))
|| ((*ptr)->type == FFI_TYPE_LONGDOUBLE
&& cif->abi != FFI_V9))
bytes += sizeof(void*);
else
#endif
{
/* Add any padding if necessary */
if (((*ptr)->alignment - 1) & bytes)
bytes = ALIGN(bytes, (*ptr)->alignment);
bytes += STACK_ARG_SIZE((*ptr)->size);
}
#endif
}
cif->bytes = bytes;
/* Perform machine dependent cif processing */
#ifdef FFI_TARGET_SPECIFIC_VARIADIC
if (isvariadic)
return ffi_prep_cif_machdep_var(cif, nfixedargs, ntotalargs);
#endif
return ffi_prep_cif_machdep(cif);
}
#endif /* not __CRIS__ */
ffi_status ffi_prep_cif(ffi_cif *cif, ffi_abi abi, unsigned int nargs,
ffi_type *rtype, ffi_type **atypes)
{
return ffi_prep_cif_core(cif, abi, 0, nargs, nargs, rtype, atypes);
}
ffi_status ffi_prep_cif_var(ffi_cif *cif,
ffi_abi abi,
unsigned int nfixedargs,
unsigned int ntotalargs,
ffi_type *rtype,
ffi_type **atypes)
{
return ffi_prep_cif_core(cif, abi, 1, nfixedargs, ntotalargs, rtype, atypes);
}
#if FFI_CLOSURES
ffi_status
ffi_prep_closure (ffi_closure* closure,
ffi_cif* cif,
void (*fun)(ffi_cif*,void*,void**,void*),
void *user_data)
{
return ffi_prep_closure_loc (closure, cif, fun, user_data, closure);
}
#endif

16
ChangeLog
View File

@@ -1,3 +1,19 @@
2012-03-03 H.J. Lu <hongjiu.lu@intel.com>
* src/x86/ffi64.c (ffi_call): Cast the return value to unsigned
long.
(ffi_prep_closure_loc): Cast to 64bit address in trampoline.
(ffi_closure_unix64_inner): Cast return pointer to unsigned long
first.
* src/x86/ffitarget.h (FFI_SIZEOF_ARG): Defined to 8 for x32.
(ffi_arg): Set to unsigned long long for x32.
(ffi_sarg): Set to long long for x32.
2012-03-03 H.J. Lu <hongjiu.lu@intel.com>
* src/prep_cif.c (ffi_prep_cif_core): Properly check bad ABI.
2012-03-03 Andoni Morales Alastruey <ylatuya@gmail.com> 2012-03-03 Andoni Morales Alastruey <ylatuya@gmail.com>
* configure.ac: Add -no-undefined for both 32- and 64-bit x86 * configure.ac: Add -no-undefined for both 32- and 64-bit x86

3
README
View File

@@ -78,6 +78,7 @@ tested:
| X86 | Interix | | X86 | Interix |
| X86 | kFreeBSD | | X86 | kFreeBSD |
| X86 | Linux | | X86 | Linux |
| X86 | Linux/x32 |
| X86 | Mac OSX | | X86 | Mac OSX |
| X86 | OpenBSD | | X86 | OpenBSD |
| X86 | OS/2 | | X86 | OS/2 |
@@ -148,6 +149,8 @@ See the ChangeLog files for details.
3.0.11 MMM-DD-YY 3.0.11 MMM-DD-YY
Lots of build fixes. Lots of build fixes.
Add Amiga newer MacOS support. Add Amiga newer MacOS support.
Add Linux/x32 support.
Add thiscall and fastcall support on Windows.
Fix Octeon and MC68881 support. Fix Octeon and MC68881 support.
Fix code pessimizations. Fix code pessimizations.

2
patches/series
View File

@@ -44,3 +44,5 @@ alpha
autoconf-archive-update autoconf-archive-update
ffitarget-include-fix ffitarget-include-fix
no-undefined-fix no-undefined-fix
x86_abi_check_fix
x32-abi

99
patches/x32-abi Normal file
View File

@@ -0,0 +1,99 @@
Index: libffi/ChangeLog
===================================================================
--- libffi.orig/ChangeLog
+++ libffi/ChangeLog
@@ -1,5 +1,17 @@
2012-03-03 H.J. Lu <hongjiu.lu@intel.com>
+ * src/x86/ffi64.c (ffi_call): Cast the return value to unsigned
+ long.
+ (ffi_prep_closure_loc): Cast to 64bit address in trampoline.
+ (ffi_closure_unix64_inner): Cast return pointer to unsigned long
+ first.
+
+ * src/x86/ffitarget.h (FFI_SIZEOF_ARG): Defined to 8 for x32.
+ (ffi_arg): Set to unsigned long long for x32.
+ (ffi_sarg): Set to long long for x32.
+
+2012-03-03 H.J. Lu <hongjiu.lu@intel.com>
+
* src/prep_cif.c (ffi_prep_cif_core): Properly check bad ABI.
2012-03-03 Andoni Morales Alastruey <ylatuya@gmail.com>
Index: libffi/src/x86/ffi64.c
===================================================================
--- libffi.orig/src/x86/ffi64.c
+++ libffi/src/x86/ffi64.c
@@ -427,7 +427,7 @@ ffi_call (ffi_cif *cif, void (*fn)(void)
/* If the return value is passed in memory, add the pointer as the
first integer argument. */
if (ret_in_memory)
- reg_args->gpr[gprcount++] = (long) rvalue;
+ reg_args->gpr[gprcount++] = (unsigned long) rvalue;
avn = cif->nargs;
arg_types = cif->arg_types;
@@ -509,9 +509,11 @@ ffi_prep_closure_loc (ffi_closure* closu
tramp = (volatile unsigned short *) &closure->tramp[0];
tramp[0] = 0xbb49; /* mov <code>, %r11 */
- *(void * volatile *) &tramp[1] = ffi_closure_unix64;
+ *((unsigned long long * volatile) &tramp[1])
+ = (unsigned long) ffi_closure_unix64;
tramp[5] = 0xba49; /* mov <data>, %r10 */
- *(void * volatile *) &tramp[6] = codeloc;
+ *((unsigned long long * volatile) &tramp[6])
+ = (unsigned long) codeloc;
/* Set the carry bit iff the function uses any sse registers.
This is clc or stc, together with the first byte of the jmp. */
@@ -550,7 +552,7 @@ ffi_closure_unix64_inner(ffi_closure *cl
{
/* The return value goes in memory. Arrange for the closure
return value to go directly back to the original caller. */
- rvalue = (void *) reg_args->gpr[gprcount++];
+ rvalue = (void *) (unsigned long) reg_args->gpr[gprcount++];
/* We don't have to do anything in asm for the return. */
ret = FFI_TYPE_VOID;
}
Index: libffi/src/x86/ffitarget.h
===================================================================
--- libffi.orig/src/x86/ffitarget.h
+++ libffi/src/x86/ffitarget.h
@@ -61,9 +61,15 @@ typedef unsigned long long ffi_arg;
typedef long long ffi_sarg;
#endif
#else
+#if defined __x86_64__ && !defined __LP64__
+#define FFI_SIZEOF_ARG 8
+typedef unsigned long long ffi_arg;
+typedef long long ffi_sarg;
+#else
typedef unsigned long ffi_arg;
typedef signed long ffi_sarg;
#endif
+#endif
typedef enum ffi_abi {
FFI_FIRST_ABI = 0,
Index: libffi/README
===================================================================
--- libffi.orig/README
+++ libffi/README
@@ -78,6 +78,7 @@ tested:
| X86 | Interix |
| X86 | kFreeBSD |
| X86 | Linux |
+| X86 | Linux/x32 |
| X86 | Mac OSX |
| X86 | OpenBSD |
| X86 | OS/2 |
@@ -148,6 +149,8 @@ See the ChangeLog files for details.
3.0.11 MMM-DD-YY
Lots of build fixes.
Add Amiga newer MacOS support.
+ Add Linux/x32 support.
+ Add thiscall and fastcall support on Windows.
Fix Octeon and MC68881 support.
Fix code pessimizations.

25
patches/x86_abi_check_fix Normal file
View File

@@ -0,0 +1,25 @@
Index: libffi/ChangeLog
===================================================================
--- libffi.orig/ChangeLog
+++ libffi/ChangeLog
@@ -1,3 +1,7 @@
+2012-03-03 H.J. Lu <hongjiu.lu@intel.com>
+
+ * src/prep_cif.c (ffi_prep_cif_core): Properly check bad ABI.
+
2012-03-03 Andoni Morales Alastruey <ylatuya@gmail.com>
* configure.ac: Add -no-undefined for both 32- and 64-bit x86
Index: libffi/src/prep_cif.c
===================================================================
--- libffi.orig/src/prep_cif.c
+++ libffi/src/prep_cif.c
@@ -112,7 +112,7 @@ ffi_status FFI_HIDDEN ffi_prep_cif_core(
FFI_ASSERT(nfixedargs <= ntotalargs);
#ifndef X86_WIN32
- if (! (abi > FFI_FIRST_ABI) && (abi <= FFI_LAST_ABI))
+ if (! (abi > FFI_FIRST_ABI && abi < FFI_LAST_ABI))
return FFI_BAD_ABI;
#else
if (! (abi > FFI_FIRST_ABI && abi < FFI_LAST_ABI || abi == FFI_THISCALL))

2
src/prep_cif.c
View File

@@ -112,7 +112,7 @@ ffi_status FFI_HIDDEN ffi_prep_cif_core(ffi_cif *cif, ffi_abi abi,
FFI_ASSERT(nfixedargs <= ntotalargs); FFI_ASSERT(nfixedargs <= ntotalargs);
#ifndef X86_WIN32 #ifndef X86_WIN32
if (! (abi > FFI_FIRST_ABI) && (abi <= FFI_LAST_ABI)) if (! (abi > FFI_FIRST_ABI && abi < FFI_LAST_ABI))
return FFI_BAD_ABI; return FFI_BAD_ABI;
#else #else
if (! (abi > FFI_FIRST_ABI && abi < FFI_LAST_ABI || abi == FFI_THISCALL)) if (! (abi > FFI_FIRST_ABI && abi < FFI_LAST_ABI || abi == FFI_THISCALL))

10
src/x86/ffi64.c
View File

@@ -427,7 +427,7 @@ ffi_call (ffi_cif *cif, void (*fn)(void), void *rvalue, void **avalue)
/* If the return value is passed in memory, add the pointer as the /* If the return value is passed in memory, add the pointer as the
first integer argument. */ first integer argument. */
if (ret_in_memory) if (ret_in_memory)
reg_args->gpr[gprcount++] = (long) rvalue; reg_args->gpr[gprcount++] = (unsigned long) rvalue;
avn = cif->nargs; avn = cif->nargs;
arg_types = cif->arg_types; arg_types = cif->arg_types;
@@ -509,9 +509,11 @@ ffi_prep_closure_loc (ffi_closure* closure,
tramp = (volatile unsigned short *) &closure->tramp[0]; tramp = (volatile unsigned short *) &closure->tramp[0];
tramp[0] = 0xbb49; /* mov <code>, %r11 */ tramp[0] = 0xbb49; /* mov <code>, %r11 */
*(void * volatile *) &tramp[1] = ffi_closure_unix64; *((unsigned long long * volatile) &tramp[1])
= (unsigned long) ffi_closure_unix64;
tramp[5] = 0xba49; /* mov <data>, %r10 */ tramp[5] = 0xba49; /* mov <data>, %r10 */
*(void * volatile *) &tramp[6] = codeloc; *((unsigned long long * volatile) &tramp[6])
= (unsigned long) codeloc;
/* Set the carry bit iff the function uses any sse registers. /* Set the carry bit iff the function uses any sse registers.
This is clc or stc, together with the first byte of the jmp. */ This is clc or stc, together with the first byte of the jmp. */
@@ -550,7 +552,7 @@ ffi_closure_unix64_inner(ffi_closure *closure, void *rvalue,
{ {
/* The return value goes in memory. Arrange for the closure /* The return value goes in memory. Arrange for the closure
return value to go directly back to the original caller. */ return value to go directly back to the original caller. */
rvalue = (void *) reg_args->gpr[gprcount++]; rvalue = (void *) (unsigned long) reg_args->gpr[gprcount++];
/* We don't have to do anything in asm for the return. */ /* We don't have to do anything in asm for the return. */
ret = FFI_TYPE_VOID; ret = FFI_TYPE_VOID;
} }

6
src/x86/ffitarget.h
View File

@@ -61,9 +61,15 @@ typedef unsigned long long ffi_arg;
typedef long long ffi_sarg; typedef long long ffi_sarg;
#endif #endif
#else #else
#if defined __x86_64__ && !defined __LP64__
#define FFI_SIZEOF_ARG 8
typedef unsigned long long ffi_arg;
typedef long long ffi_sarg;
#else
typedef unsigned long ffi_arg; typedef unsigned long ffi_arg;
typedef signed long ffi_sarg; typedef signed long ffi_sarg;
#endif #endif
#endif
typedef enum ffi_abi { typedef enum ffi_abi {
FFI_FIRST_ABI = 0, FFI_FIRST_ABI = 0,