Add OpenBSD/hppa and OpenBSD/powerpc support.
Rebuilt.
Clear icache.
Move PR number to comment.
Likewise.
xfail dg-output on mips-sgi-irix6*.
Likewise.
Use PRIu8, PRId8 instead of %hhu, %hhd.
Define. [__sgi__] (PRId8, PRIu8): Define.
Define. Use them to handle ELF vs. ECOFF differences. [__osf__] (_GLOBAL__F_ffi_call_osf): Define.
Fix unknown FDE encoding.
ditto.
Allow for more 32-bit ABIs.
Don't -malign-double or use -ffast-math.
Rebuilt.
Regenerate.
Define.
Remove definition.
Remove definition.
Convert assertion into FFI_BAD_TYPEDEF return. Initialize arg size and alignment to 0.
Don't ASSERT ABI test, just return FFI_BAD_ABI when things are wrong.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Don't ASSERT ABI test, just return FFI_BAD_ABI when things are wrong.
Support for xlc toolchain on AIX. Declare .ffi_prep_args. Insert nops after branch instructions so that the AIX linker can insert TOC reload instructions.
Declare .ffi_closure_helper_DARWIN.
Clean up error messages.
Cast raw arg to ffi_raw*.
Add pragma for SGI compiler.
Add powerpc64--darwin support.
Mention Interix.
Add Interix to win32/cygwin/mingw case.
Ditto.
Treat Interix like Cygwin, instead of as a generic win32.
Remove xfail.
Remove xfail.
Define.
Check for bad ABI.
Define.
Check for bad types.
Add build-ios.sh, src/arm/gentramp.sh, src/arm/trampoline.S.
Add src/arc/trampoline.S.
Define.
Define for FFI_EXEC_TRAMPOLINE_TABLE case (iOS).
Handl FFI_EXEC_TRAMPOLINE_TABLE case separately.
Handle Apple iOS host.
Handle FFI_EXEC_TRAMPOLINE_TABLE case.
New file.
Rebuilt.
Mention ARM iOS.
Cast void * to uintptr_t first.
Likewise.
Define in order to avoid redefinition of mallinfo on HP-UX.
Make compatible with Solaris Studio aggregate return ABI. Flush cache.
Flush cache.
Fix grammar nit in comment.
Define and use.
Fix x86 test for pc related relocs.
Rebuilt.
Add RTEMS support for cache flushing. Handle case when CPU variant does not have long double support.
Add support for mc68000, Coldfire, and cores with soft floating point.
Add mips*--rtems support.
Regenerate.
Ensure needed constants are available for targets which do not have sgidefs.h.
Add -bindir option for windows DLLs.
Regenerate.
Provide Darwin64 implementation, update comments.
New,
Update for Darwin64.
Provide Darwin64 implementation, update comments.
Likewise.
Use double backslashes.
Likewise.
Regenerate.
Add UNWIND to .pad directive.
Same.
Move down to before ffi_closure_VFP. Add '.fpu vfp' directive.
Define.
Define.
Define.
Use FDE_ENCODING.
Simplify with FDE_ENCODE.
Check for symbol underscores on mingw-w64.
Rebuilt.
Correctly access extern symbols in respect to underscores.
Rename …
… to this.
Don't load libffi-dg.exp.
Likewise.
Add VFP register argument handling code, new parameter, and return value. Update comments.
Add case for VFP struct return values. Add call to layout_vfp_args().
Update declaration.
New declaration.
Add VFP struct return conditions. Call ffi_call_VFP() when ABI is FFI_VFP.
New declaration.
Add new vfp_args parameter, update call to ffi_prep_incoming_args_SYSV().
Update parameters. Add VFP argument case handling.
Pass ffi_closure_VFP to trampoline construction under VFP hard-float.
New function.
Same.
Same.
Same.
Add FFI_VFP. Define FFI_DEFAULT_ABI based on __ARM_PCS_VFP.
Define for adding VFP hard-float specific fields.
Define internally used type code.
Same.
Change call of ffi_prep_args() to direct call. Move function pointer load upwards.
New function.
Same.
New function.
New function.
Skip if target is arm*-- and compiler options include -mfloat-abi=hard.
Same.
Ensure cif->bytes is a multiple of 8.
New test.
Add target powerpc64--freebsd.
Regenerate.
Pass -mlong-double-128 only to linux targets.
Likewise.
Likewise.
Likewise.
Pass FFI_DEBUG define to msvcc.sh for linking to the debug CRT when –enable-debug is given.
Define it.
Translate -g and -DFFI_DEBUG appropriately.
Add X86_ANY define for all x86/x86_64 platforms.
Remove redundant ifdef checks.
Push stack space computation into src/x86/ffi.c for X86_ANY so return value space doesn't get added twice.
Don't pass -safeseh to ml64 because behavior is buggy.
Make FFI_LAST_ABI one past the last valid ABI.
Fix ABI assertion.
Ditto.
Fix strncmp usage bug.
Protect define with ifndef.
Ditto.
Ditto.
Ditto.
Align the stack pointer to 16-bytes.
Pass also mandir to submakes.
Regenerated.
Check for illegal in as output, too.
Check for .ascii.
Check for .string.
Regenerate.
Regenerate.
Use .ascii, .string or error.
Document previous change.
Don't copy structs passed by value.
Fix spelling.
Ditto.
Ditto.
Rebuilt.
Build with -W3 instead of -Wall.
Remove build warnings.
Ditto.
Ditto.
Add OS/2 support.
Rebuilt.
Ditto.
Ditto.
Ditto.
Remove unused args variable.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Disable build warnings.
Clarify windows build procedure.
New test.
Regenerate.
Regenerate.
Use @unwind section type.
Fix typo in comment.
Use /* … */ comment style.
Fix typo.
Remove.
Define for processor ARM_ARCH_7EM.
Add notes on building with Microsoft Visual C++.
New file.
Port assembly routines to MSVC and ifdef.
Tweak function declaration and remove excess parens.
Add __declspec(align(8)) to typedef struct ffi_closure.
Merge ffi_call_SYSV and ffi_call_STDCALL into new function ffi_call_win32 on X86_WIN32.
Rename to ffi_call_win32.
Remove.
Move stack space allocation code to ffi_prep_cif_machdep for x86.
To here.
Check for __i386 and __amd64 for Sun Studio compiler compatibility.
Add closure example.
Define.
Add -Wno-format on alpha*-dec-osf*.
Likewise.
Likewise.
Likewise.
Likewise.
Cast MAP_FAILED to char *.
Use .abicalls and .eh_frame with __GNUC__.
Update for libffi 3.0.9.
Fix definitions.
Define for windows hosts.
Add AM_LTLDFLAGS.
Rebuilt.
Rebuilt.
New test case from Blake Chaffin @ Apple.
Copy abi and nargs to local variables.
New function.
Call it.
Define for mips when appropriate.
Rebuilt.
Mark as xfail for avr32*--.
Ditto.
Conditionally include stdint.h and inttypes.h.
Ditto.
Add amd64--openbsd.
Rebuilt.
Link openbsd programs with -lpthread.
Remove xfail for mips*-- and arm*--.
Remove xfail for arm*--.
Define FFI_MMAP_EXEC_WRIT for the given targets.
Regenerate.
Likewise.
Remove the FFI_MMAP_EXEC_WRIT definition for Solaris/x86.
Advance intarg_count when a float arguments is passed in memory.
Mark general registers as used up when a 64bit or soft-float long double argument is passed in memory.
Fix include in examples.
Add dircategory.
Placed '__GNUC__' ifdef around '__attribute__((aligned(8)))' in ffi_closure, fixes compile for IRIX MIPSPro c99.
Added '__sgi' define to non '__attribute__((__mode__()))' integer typedefs.
Added 'defined(_MIPSEB)' to BE check.
Added FFI_LONGDOUBLE support and alignment(N32 only).
Corrected '#include <sgidefs.h>' for IRIX and fixed non '__attribute__((__mode__()))' integer typedefs.
Put '#ifdef linux' around '.abicalls' and '.eh_frame' since they are Linux/GNU Assembler specific.
Add AVR32 port.
Rebuilt.
Make i?86 build on FreeBSD and OpenBSD.
Regenerate.
Define PRIuPTR on PA HP-UX.
Handle FFI_TYPE_LONGDOUBLE type on HP-UX.
Properly align 'long double' arguments.
Define PRIuPTR on Solaris < 10.
Define.
Remove xfail for mips*--
Same.
Same.
Same.
Same.
Same.
Fix tablejump typo.
Update AIX32 code to be consistent with AIX64 code.
Same.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Reorganize 64-bit code to match linux64_closure.S.
Update from gcc/config/i386/i386.c.
Do not use the address of two consecutive SSE registers directly.
Remove xfail for x86_64 linux targets.
Increment pfr for long double split between fpr13 and stack.
Increment next_arg and fparg_count twice for long double.
Remove extra parentheses.
Fix format string. Remove xfails for x86 linux targets.
Fix typo in INT64 case.
Convert to more standard register usage. Call ffi_prep_args directly. Add long double return value support.
Double arg increment applies to FFI_TYPE_DOUBLE. Correct fpr_base increment typo. Separate FFI_TYPE_SINT32 and FFI_TYPE_UINT32 cases.
Only 16 byte stack alignment in 64 bit mode.
Remove nf and ng counters. Move temp into case.
Maintain 16 byte stack alignment. Allocate result area between params and FPRs.
Define for PPC64 Linux and AIX64.
Implement AIX64 version.
Implement AIX64 version.
Use extsb, lha and displament addresses.
Implement AIX64 support.
Same.
Same.
Same.
New test.
Remove for gnu assembly version use of ___chkstk.
Define for processors ARM_ARCH_6T2, ARM_ARCH_6M, ARM_ARCH_7, ARM_ARCH_7A, ARM_ARCH_7R, ARM_ARCH_7M. Change the conditionals to __SOFTFP__ || ARM_EABI for -mfloat-abi=softfp to work.
Add X86_FREEBSD to enable proper extension on char and short.
Remove special handling for FFI_TYPE_POINTER.
New defines.
New ffi_abi enumerations.
Set FFI_DEFAULT_ABI for soft-float.
Add handling for soft-float structure and pointer returns.
Add handling for pointer returns.
Handle soft-float.
Handle soft-float, fix pointer handling.
Declare proper argument types.
Handle soft-float.
Bump to 2.64.
Remove.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Use m4_rename_force.
Also use sign/zero-extending return types for X86_WIN32.
Handle omitted return types.
Likewise.
Define to const 0 for Cygwin.
Also use these functions on Cygwin.
Revert 2009-07-02, 2009-07-01 and 2009-06-30 commits.
Set ld_library_path to “” before adding paths. (This reinstates an assignment that was removed by my 2009-06-30 commit, but changes the initial value from “.” to “”.)
Don't add “.” to ld_library_path. Use add_path. Add just find_libgcc_s to ld_library_path, not every libgcc multilib directory.
Avoid clobbering cr3 and cr4, which are supposed to be callee-saved.
Fix overrun of return buffer for odd-size structs.
Add allow_stack_execute for Darwin.
Add missing blank lines.
Regenerate.
Fix printf format specifiers.
Define.
xfail everywhere.
Likewise.
Remove info_TEXINFOS.
testsuite/libffi.call/cls_align_sint64.c, testsuite/libffi.call/cls_align_uint64.c, testsuite/libffi.call/cls_ulonglong.c, testsuite/libffi.call/return_ll1.c, testsuite/libffi.call/stret_medium2.c: Fix printf format specifiers. testsuite/libffi.special/unwindtest.cc: include stdint.h.
Added win64 support (mingw or MSVC)
Regenerated
properly escape cygwin/w32 path
Clarify size requirements for return value.
Fix filename in comment.
Remove unused extern.
use ffi_closure_alloc instead of checking for MMAP. Use intptr_t instead of long casts.
Add xfail sh*--linux-.
Add xfail sh*--.
Likewise.
Add missing file.
New file.
Likewise.
Likewise.
New file.
New file.
Add src/x86/darwin64.S, src/dlmalloc.c.
Add X86_FREEBSD.
Bump version to 3.0.8. parisc*--linux: Add. i386--freebsd | i386--openbsd: Add. powerpc--beos: Add. AM_CONDITIONAL X86_FREEBSD: Add. AC_CONFIG_FILES: Add man/Makefile.
Change void (*)() to void (*)(void).
Import from libffi 3.0.8.
Add xfails.
Add xfails.
Add xfail x86_64--linux-.
Add xfails.
Add __UNUSED__ to args.
Likewise.
Likewise.
Likewise.
Merge stdcall changes from libffi.
Back out stdcall changes.
Change void (*)() to void (*)(void).
Likewise.
Insert licence header.
Likewise.
Likewise.
Change void (*)() to void (*)(void).
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
New function.
Add FDE for it.
Also check if assembler supports pc-relative relocs on X86_WIN32 targets.
Regenerate.
Declare extern, not global.
Add missing function type symbol .def and add EH markup labels.
Likewise.
Likewise.
Likewise.
Add hand-crafted EH data.
Change copyright header to refer to version 3 of the GNU General Public License and to point readers at the COPYING3 file and the FSF's license web page.
Likewise.
Likewise.
Regenerate.
mark expected failures on x86_64 cygwin/mingw.
use portable cast from pointer to integer (intptr_t).
disable for win64.
New test.
Regenerate.
Regenerate.
Precalculate RAW_CLOSURE_CIF_OFFSET, RAW_CLOSURE_FUN_OFFSET, RAW_CLOSURE_USER_DATA_OFFSET for the Solaris 10/x86 assembler.
Only use SYMBOL-. iff HAVE_AS_X86_PCREL.
Move to .text section.
Likewise.
Only use SYMBOL-. iff HAVE_AS_X86_PCREL.
Regenerate.
Add support for signed/unsigned int8/16 return values.
Likewise.
Likewise.
Regenerate for new libtool.
Ditto.
Ditto.
Ditto.
Add FFI_LINUX and FFI_LINUX_SOFT_FLOAT to the POWERPC_FREEBSD enum. Add note about flag bits used for FFI_SYSV_TYPE_SMALL_STRUCT. Adjust copyright notice.
Add two new flags to indicate if we have one register or two register to use for FFI_SYSV structs.
Pass the right register flag introduced above.
Fix the return type for FFI_SYSV_TYPE_SMALL_STRUCT. Comment. Adjust copyright notice.
Clean up failures from compiler warnings.
Turn INSN into an unsigned int.
Regenerate.
Regenerate.
Regenerate.
Remove.
Regenerate.
Regenerate.
Regenerate.
Add .note.GNU-stack on Linux.
Likewise.
Fix ARM comment marker.
Add .note.GNU-stack on Linux.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Regenerate.
Likewise.
Likewise.
Likewise.
Likewise.
New tests from Apple.
Change C++ comment to C comment.
Define (conditionally), and use it to include cachectl.h.
Fix cache flushing.
Define.
Add amd64--freebsd target.
Regenerate.
Calculate jump return offset based on code pointer, not data pointer.
Add test for stdcall closures.
Increase size of trampoline for stdcall closures.
Add assembly for stdcall closure.
Initialize stdcall closure trampoline.
Pop 4 byte from stack when returning struct.
Add “-O2 -fomit-frame-pointer” tests.
Regenerate.
Fix thinko.
Add code for signed/unsigned int8/16 for X86_DARWIN. Updated copyright info. Handle one and two byte structs with special cif->flags.
Add special types for one and two byte structs. Updated copyright info.
Rewrite to use a jump table like sysv.S Remove code to pop args from the stack after call. Special-case signed/unsigned for int8/16, one and two byte structs.
Handle FFI_TYPE_UINT8, FFI_TYPE_SINT8, FFI_TYPE_UINT16, FFI_TYPE_SINT16, FFI_TYPE_UINT32, FFI_TYPE_SINT32. Updated copyright info.
Replace dadd with ADDU, dsub with SUBU, add with ADDU and use smaller code sequences.
Handle long double return type.
Define if not already defined.
New typedef.
Change parameter types from ffi_raw to ffi_java_raw.
Same.
Change parameter types.
Replace FFI_SIZEOF_ARG with FFI_SIZEOF_JAVA_RAW.
Change type of raw to ffi_java_raw. Replace FFI_SIZEOF_ARG with FFI_SIZEOF_JAVA_RAW. Use sizeof(ffi_java_raw) for alignment calculations.
Same.
Add special handling for FFI_TYPE_POINTER if FFI_SIZEOF_JAVA_RAW == 4.
Same.
Change type of raw to ffi_java_raw.
Same.
Change parameter types.
Define for N32 ABI.
Use 64-bit add instruction on pointer values.
Introduce new ABI FFI_LINUX_SOFT_FLOAT. Add local FFI_TYPE_UINT128 to handle soft-float long-double-128.
Distinguish between NO_FPRS and not and set the NUM_FPR_ARG_REGISTERS according to. Add support for potential soft-float support under hard-float architecture.
Set NUM_FPR_ARG_REGISTERS to 0 in case of FFI_LINUX_SOFT_FLOAT, handle float, doubles and long-doubles according to the FFI_LINUX_SOFT_FLOAT ABI.
Likewise.
Likewise.
Make sure not to store float/double on archs where NO_FPRS is true. Add FFI_TYPE_UINT128 support.
Add support for soft-float long-double-128. Adjust copyright notice.
Move defintion of MAYBE_UNUSED from here to …
… here. Update copyright.
Load correct cr to compare if we have long double.
Likewise.
Add a comment to show which part goes into cr6.
New test.
New.
Conditionally compile unwinder directives.
Add unwinder directives.
Align structs by at least 4 bytes. Only treat r0 as a struct address if we're actually returning a struct by address. Only copy the bytes that are actually within a struct.
A Composite Type not larger than 4 bytes is returned in r0, not passed by address.
Allocate a word-sized temporary for the case where a composite is returned in r0.
Align as necessary.
Unify MIPS_IRIX and MIPS_LINUX into MIPS.
Likewise.
Regenerate.
Likewise.
Likewise.
Likewise.
New test.
Remove xfail for mips64*--.
Don't treat mips64 as a special case.
Add n32.S.
Regenerate
Ditto.
Ditto.
Indent.
New preprocessor macros.
Set for n64 case.
Define for n32 and n64 cases.
Add debug macros and labels for FDE.
New function.
New section
Clean up comments.
Pass ffi_closure parameter in $12.
Use FFI_MIPS_N32 instead of _MIPS_SIM == _ABIN32 throughout.
New, use in place of ffi_stop_here.
Use unsigned long to hold pointer values. Rewrite to support n32/n64 ABIs.
Rewrite.
Remove unused variable. Reverse position of flag bits.
Rewrite n32 portion.
Enable for n64. Add special handling for small structure return values.
Add n32 and n64 support.
Add cast to silence warning.
New functions.
Remove definition.
Use correct type specifiers.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Ditto.
Fix typo in previous checkin.
Handle FFI_TYPE_UINT8, FFI_TYPE_SINT8, FFI_TYPE_UINT16, FFI_TYPE_SINT16, FFI_TYPE_UINT32, FFI_TYPE_SINT32.
Use __clear_cache instead of directly using the sys_cacheflush syscall.
Define return type as ffi_arg. Use proper printf conversion specifier.
in x86 case, add code for signed/unsigned int8/16.
Rewrite to: Use a jump table. Remove code to pop args from the stack after call. Special-case signed/unsigned int8/16.
Revert.
Add soft-float.
Verify call result as signed char, not ffi_arg.
Set TARGET to X86_64.
Regenerate.
Don't include sys/cachectl.h.
Use __builtin___clear_cache() instead of cacheflush().
Enable for ARM.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Renamed and ajusted from (ffi_prep_closure): … this.
Adjust.
New, add closure support.
Likewise.
Likewise.
Enable closure support.
New test.
Regenerate.
Regenerate.
New macros.
Add callframe annotation.
Fix numerous test suite failures.
Likewise.
Bring up to date.
Regenerate.
Remove RCS keyword.
Tidy target case.
Allow the target to override.
Regenerate.
Don't define ffi_type_foo if LIBFFI_HIDE_BASIC_TYPES is defined.
If not HAVE_LONG_DOUBLE, define to ffi_type_double.
Define.
Mark the data const.
Special case for Alpha. Don't define if long double == double.
Assert unique value.
Handle it as the 128-bit type.
Likewise.
Likewise. Mark hidden.
Mark hidden.
Tidy definition.
Mark hidden.
Handle 128-bit long double.
Add -mieee for alpha.
Fixed bug in example.
Include sys/statfs.h.
Define on Linux.
Define.
New variable.
New function.
Define.
Use it.
Mark as static. Use 'volatile float sum' to create sum of floats to avoid false negative due to excess precision on ix86 targets.
Ditto.
Fix left-over from previous patch.
Remove unneeded casts. Add needed ones.
New.
New.
New.
New.
New file.
Replace sflags with exec_offset. [FFI_MMAP_EXEC_WRIT] (mmap_exec_offset, add_segment_exec_offset, sub_segment_exec_offset): New macros.
New macros.
Use get_segment_flags.
Use new macros.
Silence warning.
Add src/closures.c.
Rebuilt.
Implement in terms of ffi_prep_closure_loc.
Renamed and adjusted from…
… this. Re-implement in terms of the renamed version.
Renamed and adjusted from…
… this. Re-implement in terms of the renamed version.
Renamed from
… this.
Likewise.
Likewise. Adjust.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Adjust.
Renamed and adjusted from…
… this.
Renamed from
… this.
Adjust.
New file, imported version 2.8.3 of Doug Lea's malloc.
Add dummy install-pdf target.
Regenerate
Add long double handling.
Move restore of r2 immediately after bctrl instruction.
Add missing targets.
Rebuilt.
Add TARGET for x86_64--darwin.
Add rules for 64-bit sources for X86_DARWIN.
Set trampoline size for x86_64--darwin.
New file for x86_64--darwin support.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
New test case for ffi_call only.
Regenerate with aclocal -I .. as written in the Makefile.am.
New.
Call darwin_adjust_aggregate_sizes for Darwin.
Remove Darwin XFAIL.
Remove Darwin XFAIL.
Set X86_DARWIN symbol and conditional.
Regenerated.
New case.
Add src/x86/darwin.S.
Regenerated.
Regenerated.
Regenerated.
Treat like X86_WIN32, and additionally align stack to 16 bytes.
New, based on sysv.S.
Align > 8-byte structs.
Install both ffi.h and ffitarget.h in $(libdir)/gcc/$(target_alias)/$(gcc_version)/include.
Regenerated for automake 1.9.6.
Regenerated.
Regenerated.
Regenerated.
Revert accidental commit.
Remove lint directives.
Likewise.
Define correctly for 32-bit architectures.
New test case.
Remove xfail for mips, xfail remains for mips64.
Add install-html target. Add install-html to .PHONY
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Load floating point arguments from stack slot.
Remove notice about 'Crazy Comments'.
Remove lint directives. Cleanup white spaces.
Likewise.
Likewise.
Likewise.
Delete. No longer needed, all test cases migrated to the testsuite.
Remove lint directives.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Correct unwind offset calculation for ffi_closure_pa32.
Likewise.
Rewrite as macros.
Call stf_spill.
Call ldf_fill.
Adjust calls to above routines. Add local temps for macro result.
Recognize multilib directory names containing underscores.
New testcase.
Add PA_HPUX port.
Regenerate.
Likewise.
Likewise.
Add PA_HPUX rules.
Regenerate.
Rename linux target to PA_LINUX. Add PA_HPUX and PA64_HPUX. Rename FFI_LINUX ABI to FFI_PA32 ABI.
Define for 32-bit HP-UX targets.
Define.
Likewise.
Likewise.
Redefine.
Likewise.
Likewise.
Likewise.
Delete.
Use '__asm__'.
Add support for FFI_TYPE_SMALL_STRUCT2, FFI_TYPE_SMALL_STRUCT4 and FFI_TYPE_SMALL_STRUCT8.
Rename to ffi_prep_args_pa32. Update comment. Simplify incrementing of stack slot variable. Change type of local 'n' to unsigned int.
Rename to ffi_size_stack_pa32. Handle long double on PA_HPUX.
Likewise.
Likewise.
Rename to ffi_closure_inner_pa32. Change return type to ffi_status. Simplify incrementing of stack slot variable. Only copy floating point argument registers when PA_LINUX is true. Reformat debug statement. Add support for FFI_TYPE_SMALL_STRUCT2, FFI_TYPE_SMALL_STRUCT4 and FFI_TYPE_SMALL_STRUCT8.
Rename to ffi_closure_pa32. Add 'extern' to declaration.
Make linux trampoline conditional on PA_LINUX. Add nops to cache flush. Add trampoline for PA_HPUX.
New file.
Rename to ffi_call_pa32. Rename ffi_prep_args_LINUX to ffi_prep_args_pa32. Localize labels. Add support for 2, 4 and 8-byte small structs. Handle unaligned destinations in 3, 5, 6 and 7-byte small structs. Order argument type checks so that common argument types appear first.
Rename to ffi_closure_pa32. Rename ffi_closure_inner_LINUX to ffi_closure_inner_pa32.
Add FFI_LINUX. Default for 32-bit using IBM extended double format. Fix FFI_LAST_ABI.
Handle linux variant of FFI_TYPE_LONGDOUBLE.
Assert using IBM extended double.
Don't munge FFI_TYPE_LONGDOUBLE type. Handle FFI_LINUX FFI_TYPE_LONGDOUBLE return and args.
Handle FFI_LINUX.
Non FFI_LINUX long double return needs gpr3 return pointer as for struct return. Handle FFI_LINUX FFI_TYPE_LONGDOUBLE return and args. Don't increment “nf” unnecessarily.
Load both f1 and f2 for FFI_TYPE_LONGDOUBLE. Move epilogue insns into case table. Don't use r6 as pointer to results, instead use sp offset. Don't make a special call to load lr with case table address, instead use offset from previous call.
Save long double return.
Simplify long double return.
Handle float arguments passed with FP registers correctly.
Likewise.
Likewise.
Mark cif, args and userdata unused.
Mark cif and userdata unused.
Remove unused res.
Adjust FSF address. Add test runs for -O2, -O3, -Os and the warning flags -W -Wall.
Likewise.
Add an __UNUSED__ macro to mark unused parameter unused for gcc or else do nothing.
Likewise.
Mark cif and userdata unused.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Cast void* to avoid compiler warning.
Likewise.
Mark cif and userdata unused.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Mark cif and data unused.
Cast res_call to silence gcc.
Mark cif and data unused.
Cast res_call to silence gcc.
Mark cif and data unused.
Cast res_call to silence gcc.
Mark cif and data unused.
Cast res_call to silence gcc.
Mark cif and data unused.
Cast res_call to silence gcc.
Mark cif and data unused.
Cast res_call to silence gcc.
Mark cif and userdata unused.
Cast printf parameter to silence gcc.
Mark cif and userdata unused.
Cast printf parameter to silence gcc.
Mark cif and userdata unused.
Cast printf parameter to silence gcc.
Mark cif and userdata unused.
Cast printf parameter to silence gcc.
Mark cif and userdata unused.
Cast printf parameter to silence gcc.
Mark cif and userdata unused.
Mark cif and userdata unused.
Cast printf parameter to silence gcc.
Remove unused parameter e.
Remove unused variable i. Cleanup white spaces.
Remove unused variable i.
Mark cif and userdata unused.
Likewise.
Likewise.
Adjust printf formatters to silence gcc.
Mark cif and userdata unused.
Mark cif and userdata unused.
Mention related PR.
Mark cif and userdata unused.
Mark cif and userdata unused.
Mention related PR.
Mark cif and userdata unused.
Mark cif and userdata unused.
Likewise.
Likewise.
Likewise.
Cast the result to silence gcc.
Add the note mentioned in the last commit for this test case in the test case itself.
Mark cif as unused.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Fix register numbers in the FDE for ffi_closure_SYSV.
Remove static declaration to avoid a false negative on ix86. See PR323.
Remove unused variable and cast integer to void * if needed. Update the pointer to the FP register saved area correctly.
XFAIL this test until PR25630 is fixed.
Likewise.
New test case.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Remove ffi_type_mylong definition.
Add ffi_type_mylong definition here to be used by other test cases too.
New test case.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Enable libffi for sparc64--freebsd.
Rebuilt.
Don't call __ashldi3, instead do the shifting inline.
Don't compute %r5 shift count unconditionally. Simplify load sequences for 1, 2, 3, 4 and 8 byte structs, for the remaining struct sizes don't call __lshrdi3, instead do the shifting inline.
New, add closure support.
Likewise.
Likewise.
Enable closure support.
Remove obsolete sgidefs.h include. Add missing parentheses.
Code formatting. Define and use A3_OFF, FP_OFF, RA_OFF. Micro-optimizations.
Likewise, but with newly defined A3_OFF2, A2_OFF2, A1_OFF2, A0_OFF2, RA_OFF2, FP_OFF2, S0_OFF2, GP_OFF2, V1_OFF2, V0_OFF2, FA_1_1_OFF2, FA_1_0_OFF2, FA_0_1_OFF2, FA_0_0_OFF2.
Code formatting. Fix endianness bugs.
Improve trampoline instruction scheduling.
Fix endianness bugs.
Formatting.
Avoid possible aliasing problems by using unions.
Likewise.
For darwin, use -shared-libgcc not -lgcc_s, and explain why.
New typedef.
New define.
Check for result buffer overflow.
Handle linux64 long double returns.
New constant.
Handle linux64 long double returns.
Correct placement of stack homed fp args.
Correct stack sizing for same.
New test.
Add FFI_HIDDEN definition.
Rebuilt.
Rebuilt.
Remove.
Use FFI_HIDDEN instead of hidden.
Use FFI_HIDDEN instead of .hidden.
Remove, add FFI_HIDDEN to its prototype.
New.
New.
New.
Rebuilt.
Handle i*86--gnu.
Use DW_CFA_offset_extended_sf rather than DW_CFA_GNU_negative_offset_extended.
Likewise.
Stop argument popping correctly on sh3.
Change the stack layout for sh3 struct argument.
Fix sh3 argument copy, when it is partially on register.
Likewise.
Don't set too many cif->flags.
Handle small structures correctly. Remove empty line.
Remove.
Handle small structures correctly.
Likewise.
Likewise.
Likewise.
Handle 1, 2 and 4-byte return. Emit position independent code if PIC and remove wrong datalabel prefixes from EH data.
Add POWERPC_FREEBSD.
Regenerate.
Likewise.
Likewise.
Add POWERPC_FREEBSD rules.
Regenerate.
Add POWERPC_FREEBSD rules.
Define.
Add flags to handle small structure returns in ffi_call_SYSV.
Handle small structures for SYSV 4 ABI. Aka FFI_SYSV.
Likewise.
Add return types for small structures.
Add bits to handle small structures for final SYSV 4 ABI.
New test file.
Likewise.
Likewise.
Rename FFI_TYPE_SMALL_STRUCT1 as FFI_TYPE_SMALL_STRUCT3. Break out handling for 5-7 byte structures. Kill compilation warnings.
Print return values as hex in debug message. Rename FFI_TYPE_SMALL_STRUCT1 as FFI_TYPE_SMALL_STRUCT3. Properly handle 5-7 byte structure returns.
Remove.
Define.
Mark source file as using PA1.1 assembly.
Remove.
Optimize handling of 3-byte struct returns.
Properly handle 5-7 byte struct returns.
Enforce PIC code.
Likewise.
Treat i*86-*-solaris2.10 and up as X86_64.
Regenerate.
Don't use JUMPTARGET to call ffi_closure_helper_SYSV. Append @local instead.
Likewise for ffi_prep_args_SYSV.
Use AC_C_BIGENDIAN instead of AC_C_BIGENDIAN_CROSS. Use AC_CHECK_SIZEOF instead of AC_COMPILE_CHECK_SIZEOF.
Remove -I ../config.
Regenerate.
Regenerate.
Update unwind into to match code.
Save sse-used flag in bit 11 of flags.
Mask return type field. Pass ssecount to ffi_call_unix64.
Set carry bit if sse-used flag set.
Add ssecount argument. Only load sse registers if ssecount non-zero.
Only save sse registers if carry set on entry.
Add i*86--rtems, sparc*--rtems, powerpc-rtems, arm*--rtems, sh--rtems.
Regenerate.
In regsub use, have Tcl8.3-compatible intermediate variable.
Add CRIS support.
Likewise.
Regenerate.
New directory.
New files.
Wrap in ifndef __CRIS__.
Replace n with r?n in output tests.
Regenerate.
Use long long instead of DI.
Extend gp to 64 bits in ILP32 mode. Load 64 bits even for short data.
Update for -m64 multilib.
Likewise.
Do not invoke TL_AC_GCC_VERSION. Do not set tool_include_dir.
Regenerate.
Set gcc_version and toollibffidir.
Regenerate.
New test case.
Bump alignment to odd-numbered register pairs for 64-bit integer types.
Remove PPC stuff which does not belong to frv.
Add -shared-libgcc.
Remove.
Rename from FFI_INTEGRAL_TYPEDEF. Replace size and offset parameters with a type parameter; deduce size and structure alignment. Update all users.
Define with sizeof.
Fix for ia64.
Move into ffi_prep_closure.
Rewrite from scratch.
Fix typo in unwind info.
Rename from stackLayout.
Add X86_64_COMPLEX_X87_CLASS.
Check for it.
New.
Pass byte_offset by value; perform all updates inside struct case.
Add classes argument; handle X86_64_COMPLEX_X87_CLASS.
Merge into …
… here. Share stack frame with ffi_call_unix64.
Setup cif->flags for proper structure return.
Remove.
Remove dead assert.
Rename from ffi_closure_UNIX64_inner. Rewrite to use struct register_args instead of va_list. Create flags for handling structure returns.
Remove dead strings.
Rename from ffi_call_UNIX64. Rewrite to share stack frame with ffi_call. Handle structure returns properly.
Remove.
Remove.
Rename from ffi_closure_UNIX64. Rewrite to handle structure returns properly.
Remove duplicate LIBCFLAGS and PICFLAG.
Regenerated.
Use TL_AC_GCC_VERSION to set gcc_version.
Regenerate.
Regenerate.
Regenerate for libtool change.
Regenerate for libtool reversion.
Regenerate for libtool change.
Use new procs in target-libpath.exp.
Use jalr instead of jal. Use an absolute encoding for the frame information.
Remove no-dependencies. Add ACLOCAL_AMFLAGS.
Delete logic for sincludes.
Regenerate.
Likewise.
Likewise.
Align doubles and 64-bit integers on a 8-byte boundary.
Reserve frame space for arguments.
Handle functions that return long long values. Round stack allocation to a multiple of 8 bytes for ATPCS compatibility.
Rework to avoid use of APCS register names. Handle returning long long types. Add Thumb and interworking support. Improve soft-float code.
New function.
New function.
Build the testglue wrapper if needed.
Accept more than one multilib libgcc.
Set to 0.
Don't align for double data.
New test case.
Set T bit in trampoline for the function returning a structure pointed with R2.
Use R2 as the pointer to the structure return value if T bit set. Emit position independent code and EH data if PIC.
Add m32r support.
Likewise.
Regenerate.
Regenerate.
Add m32r port to FFI_INTERNAL_TYPEDEF
New directory.
New file.
Likewise.
Likewise.
New test case.
Remove whitespaces.
Likewise.
Remove unused var. Cleanup whitespaces.
Likewise.
Likewise.
Fix comments and identation.
Likewise.
Add flag for longdouble return values.
Handle longdouble arguments.
Set flags for longdouble. Calculate space for longdouble.
Add closure handling for longdouble.
Add handling of longdouble values.
Likewise.
Defined longdouble size and alignment for darwin.
Remove whitespaces.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Add frv support.
Rebuilt.
Read configure.host.
Read configure.host.
New file. frv-elf needs libgloss.
Force ffi_closure to have a nice big (8) alignment. This is needed to frv and shouldn't harm the others.
New macro.
New files.
Xfail mips64* instead of mips*.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise and set return value to zero.
Likewise.
Likewise.
New ABI.
Fix alignment calculation.
Handle FFI_O32_SOFT_FLOAT floating point parameters and return types.
Handle FFI_O32_SOFT_FLOAT ABI.
Ditto.
Handle FFI_O32_SOFT_FLOAT ABI, fix alignment calculations.
Don't use floating point instructions if FFI_O32_SOFT_FLOAT, make stack frame ABI compliant.
set `cif->flags' to contain `FFI_TYPE_UINT64' as return type for any 64-bit integer (O32 ABI only).
new function.
new function.
Define `FFI_CLOSURES' and `FFI_TRAMPOLINE_SIZE' appropriately if the ABI is o32.
add labels for .eh_frame. Return 64 bit integers correctly.
new function. Added DWARF-2 unwind info for both functions.
8-align all stack arguments.
Detect knetbsd-gnu and kfreebsd-gnu.
Regenerate.
Check for <sys/mman.h> and mmap() explicitly instead of relying on preset autoconf cache variables.
Regenerate.
Regenerate.
Fix C aliasing violation.
Remove unused prototype.
';' is a comment character on Darwin, use 'nt' instead.
Fix typo in revision/age.
New.
Use -version-info for soname.
Regenerate.
Remove useless multilib rules.
Regenerate.
Regenerate with automake 1.8.5.
Remove useless multilib configury.
Regenerate.
New file.
Insert group barrier break fp_done.
Fix f14/f15 adjustment if FLOAT_SZ is ever changed from 8.
Add x86_64--freebsd support.
Regenerate.
Test for existence of mmap and sys/mman.h before checking blacklist. Fix suggested by Jim Wilson.
Regenerate.
Go through a non-lazy pointer for initial FDE location.
Likewise.
Fix initialization error. Reported by Thomas Heller <theller@python.net>.
Likewise.
Likewise.
Fix typo.
Update.
Regenerate.
Remove.
New file.
Rename file to …
… this.
Regenerate.
Regenerate.
Regenerate.
Regenerate.
Fix EH information so it corresponds to changes in EH format resulting from addition of linkonce support.
Likewise.
Set them. Remove VPATH. Remove rules for object files. Remove multilib support.
Add.
Relace AM_CONFIG_HEADER.
Bump version to 2.59.
Fill with version info and bug address.
Remove.
Use this instead of AC_ARG_ENABLE. De-precious CC so that the right flags are passed down to multilibs.
Replace obsolete macro AC_ERROR.
Replace obsolete macro AC_LINK_FILES.
Reorganize the output with AC_CONFIG_COMMANDS.
Rebuilt.
Likewise.
Likewise.
Likewise.
Get floating point arguments from fp registers only for the first 8 parameter slots. Don't convert a float parameter when passed in memory.
Change ecif->cif->bytes to bytes.
Add braces around nested if statement.
Add PA support.
Regenerate.
Regenerate.
Add PA target.
Regenerate.
New file.
Add PA support.
New file.
Add PA support.
POWERPC64 has 8 byte pointers.
Correct long double handling.
Fix typo.
Pass -mlong-double-128 for powerpc64--.
Likewise.
Likewise.
Correct long double function return and long double arg handling.
Formatting. Delete unused “ng” var. Use “end_pfr” instead of “nf”. Correct long double handling. Localise “temp”.
Save f2 long double return value.
Allocate space for long double return value. Adjust stack frame and offsets. Load f2 long double return.
Use 16 byte long double for POWERPC64.
Shift the parameter array when the structure return address is passed in %o0.
Rename into ffi_v9_layout_struct.
Align the field following a nested structure on a word boundary. Use memmove instead of memcpy.
Update call to ffi_V9_return_struct.
Define 'ctx' only for V8.
Clone into ffi_closure_sparc_inner_v8 and ffi_closure_sparc_inner_v9.
Return long doubles by reference. Always skip the structure return address. For structures and long doubles, copy the argument directly.
Skip the structure return address only if required. Shift the maximum floating-point slot accordingly. For big structures, copy the argument directly; otherwise, left-justify the argument and call ffi_v9_layout_struct to lay out the structure on the stack.
Undef STACKFRAME before defining it.
Pass the structure return address. Update call to ffi_closure_sparc_inner_v8. Short-circuit FFI_TYPE_INT handling. Skip the 'unimp' insn when returning long doubles and structures.
Undef STACKFRAME before defining it.
Increase the frame size by 2 words. Short-circuit FFI_TYPE_INT handling. Load structures both in integers and floating-point registers on return.
Update status of the SPARC port.
Treat result value as of type ffi_arg.
Fix CHECK.
Treat result value as of type ffi_arg, not unsigned int.
Cast the RHS of an assignment instead of the LHS.
Set LD_LIBRARY_PATH_32 for Solaris.
Cast MAP_FAILED to void *.
Cast pointers to size_t instead of int.
New test case.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Cosmetics.
Include <fcntl.h>.
Likewise.
Rebuilt.
Likewise.
Convert the mmap to the right type.
Add AC_FUNC_MMAP_BLACKLIST.
Call AC_FUNC_MMAP_BLACKLIST.
Rebuilt.
Likewise.
Likewise.
Likewise.
Add include dir.
Add MMAP definitions.
Likewise.
Use MMAP functionality for ffi_closure if available.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Make the -lgcc_s conditional.
Add DYLD_LIBRARY_PATH for darwin. Add -lgcc_s to additional flags.
PR libgcj/11147, install the ffitarget.h header file in a gcc versioned and target dependent place.
Regenerated.
Likewise.
Likewise.
Print result and check with dg-output to make debugging easier.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Make ffi_closure static.
New test case.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Do a check on the result.
Likewise.
Likewise.
Likewise.
Likewise.
Cleanup whitespaces.
Move the validity check after the initialization.
Replace FFI_ASSERT(FALSE) with FFI_ASSERT(0).
Replace undefined UINT32 and friends with __attribute__((__mode__(__SI__))) and friends.
Replace FALSE/TRUE with false/true.
AC_LINK_FILES(ffitarget.h).
Regenerate.
Likewise.
Likewise.
Likewise.
Likewise.
Include ffitarget.h files
Renamed to TARGET_SRC_MIPS_IRIX.
Removed.
Renamed to TARGET_SRC_MIPS_IRIX.
Removed.
Removed.
New targets.
Removed.
Compute sizeofs only for double and long double. Use them to define and subst HAVE_LONG_DOUBLE. Include comments into AC_DEFINE instead of using acconfig.h. Create include/ffitarget.h instead of include/fficonfig.h. Rename MIPS_GCC to MIPS_IRIX, drop MIPS_SGI since we are in gcc's tree. AC_DEFINE EH_FRAME_FLAGS.
New automake macro.
Add ffitarget.h.
Remove all system specific definitions. Declare raw API even if it is not installed, why bother? Use limits.h instead of SIZEOF_* to define ffi_type_*. Do not define EH_FRAME_FLAGS, it is in fficonfig.h now. Include ffitarget.h instead of fficonfig.h. Remove ALIGN macro.
Removed, use ffi_arg and ffi_sarg instead.
Do not define.
Accept failed assertion.
Return void and accept file/line.
Pass stringized failed assertion.
New macro.
New macro.
Define here with gcc's __attribute__ macro instead of in ffi.h
Define here instead of in ffi.h
Removed. Its content moved to src/mips/ffitarget.h after separating assembly and C sections.
SIZEOF_ARG -> FFI_SIZEOF_ARG.
Include stdbool.h (provided by GCC 2.95+).
Accept stringized failed assertion.
Rewritten.
Call FFI_ASSERT_VALID_TYPE.
New files.
include fficonfig.h
Use _ABIN32, _ABIO32 instead of external _MIPS_SIM_NABI32, _MIPS_SIM_ABI32.
Declare bytes again. Used when FFI_DEBUG = 1.
Default POWERPC64 to 8 byte size and align.
Define FFI_MIPS_N32 for N32/N64 ABIs, FFI_MIPS_O32 for O32 ABI.
Set LD_LIBRARY_PATH_64 for SPARC64. Cleanup whitespaces.
Xfail mips, arm, strongarm, xscale. Cleanup whitespaces.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Cleanup whitespaces.
Cleanup whitespaces.
Likewise.
Cleanup whitespaces, comment formatting.
Likewise.
Likewise.
Add AIX and Darwin to the right TYPEDEF.
Remove the pointer to the outgoing parameter stack.
Likewise.
Handle structures according to the Darwin/AIX ABI.
Likewise.
Likewise. Remove the outgoing parameter stack logic. Simplify the evaluation of the different CASE types.
Avoid the casts on lvalues. Change the branch statement in the trampoline code.
Take account into the alignement for the register size.
Handle the structure return value address correctly.
Return the appropriate type when the registers are used for the structure return value.
Fix the stack layout for the 64-bit return value. Update copyright years.
Search in srcdir for ffi_mips.h.
Include tail padding in structure size.
Correct placement of float result.
Correct cast of “resp” for big-endian 64 bit machines.
Merge identical SH and ARM typedefs, and add POWERPC64.
Correct next_arg calc for struct split over gpr and rest.
Correct intarg_count for structures.
Fix gpr offsets.
Regenerate.
Remove build rules for ffitest.
Rebuilt.
Include <stdlib.h> to fix compiler warning about implicit declaration of abort().
Add dejagnu test framework. Fixes PR other/11411.
Rebuilt.
Add dejagnu test framework.
Rebuilt.
New file.
Built
New file.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
Likewise.
New file.
Likewise.
Likewise.
Set 0 for little endian case. Update copyright years.
Modify for changed gcc structure passing.
Likewise.
Remove code writing to parm save area.
Use return address in lr from ffi_closure_helper_LINUX64 call to calculate table address. Optimize function tail.
Handle all floating point registers.
Likewise. Fixes second part of PR target/11410.
Note that libffi is not part of GCC. Update the project URL and status.
Include ffi.h.
Avoid gas-only .uleb128/.sleb128 directives. Use C style comments.
Add SHmedia support. Fix a typo of SH support.
Regenerate.
Add target.
Regenerate.
Add SHmedia support.
New file.
New file.
Check whether .eh_frame section should be read-only.
Rebuilt.
Rebuilt.
Define.
Use EH_FRAME_FLAGS.
Likewise.
Likewise. Include ffi.h.
Use EH_FRAME_FLAGS. Use pcrel encoding if -fpic/-fPIC/-mrelocatable.
Likewise.
If HAVE_RO_EH_FRAME is defined, don't include write in .eh_frame flags.
Likewise.
Use EH_FRAME_FLAGS.
Likewise. Use pcrel encoding if -fpic/-fPIC.
Use EH_FRAME_FLAGS. Include ffi.h.
Test assembler support for .register pseudo-op.
Use it.
Rebuilt.
Rebuilt.
Define if 64-bit.
Add FFI_LINUX64 on POWERPC. Make it the default on POWERPC64.
Define to 24 on POWERPC64.
Change powerpc--linux into powerpc*--linux.
Rebuilt.
Define.
Renamed from ffi_prep_args. Cast pointers to unsigned long to shut up warnings.
New.
New function.
Handle FFI_LINUX64 ABI.
Likewise.
Likewise.
Surround by ifndef POWERPC64.
New union type.
Use it to avoid aliasing problems.
New function.
Surround whole file by ifndef powerpc64.
Likewise.
Rename ffi_prep_args to ffi_prep_args_SYSV.
New file.
New file.
Add src/powerpc/linux64.S and src/powerpc/linux64_closure.S.
Likewise.
Fix result printing on big-endian 64-bit machines.
Print tst2_arg instead of uninitialized tst2_result.
Hide what closure pointer really points to from the compiler.
Add configuration.
Regenerated.
Regenerate.
Define X86 instead of X86_64 in 32 bit mode.
Receive closure pointer through parameter, read args using __builtin_dwarf_cfa.
Send closure reference through eax.
Avoid trailing /. in toolexeclibdir.
Rebuilt.
Recode to fit dynamic libraries.
Fix alignement bug, allocate 8 bytes for the result.
Likewise.
Update stackframe description for aix/darwin_closure.S.
Add hidden visibility attribute.
Adjust command to source config-ml.in to account for changes to the libffi_basedir definition.
Remove ${srcdir} from value and include trailing slash if nonempty.
Regenerate.
Recode to fit shared libs.
Enable FFI_CLOSURES for x86_64.
New.
New.
New.
Set and AC_SUBST. Remove USE_LIBDIR conditional.
Don't override.
Rebuilt.
Add Augmentation size to unwind info.
Add unwind info.
Likewise.
Check for void retval.
Likewise.
Add unwind info.
Only use ffi_closures if those are supported.
Remove RCS version string.
Remove RCS version string.
Add closure defines for SPARC, SPARC64.
Use static storage for closure.
New.
New.
New.
Add DWARF2 unwind info.
Make section read-only.
Has size 8 on IA64.
Add dummy AM_PROG_LIBTOOL. Include ../config/accross.m4.
Likewise.
Adapt to pcrel FDE encoding.
Look for common files in the right place.
Added FFI_STDCALL ffi_type enumeration for X86_WIN32.
Added ffi_call_STDCALL function definition.
Added switch cases for recognising FFI_STDCALL and calling ffi_call_STDCALL if target is X86_WIN32.
stdcall versions of the “my_strlen” and “many” test functions (for X86_WIN32). Added test cases to test stdcall invocation using these functions.
Interpret raw data as _Jv_word values, not ffi_raw.
Likewise.
New function.
Call it.
New function.
Call it.
Interpret return value as ffi_arg, not int.
Add missing FFI_TYPE_POINTER case.
Likewise. Also, assume return values extended to word size.
Remove debug output.
Fix i386 win32 compilation.
Add s390x--linux- target.
Regenerate.
Define S390X for s390x targets.
Define for s390/s390x.
Likewise.
Likewise.
Do not compute stack space for s390.
Use 8-byte pointers on s390x.
Major rework of existing code. Add support for s390x targets. Add closure support.
Likewise.
Fix typo.
If we don't have machine/asm.h and the pre-processor has defined USER_LABEL_PREFIX, then use it in CNAME.
Handle soft-float.
Fix multilib x86-64 support.
Fix multilib parallel build.
Add brackets.
Regenerate.
Add SH support.
Regenerate.
Add target.
Regenerate.
Add SH support.
New file.
New file.
Add SH support.
New.
Support MIPS_LINUX.
Likewise.
Regenerated.
Skip.
New.
Regenerated.
Include <sgidefs.h>.
New file that adds x86-64 support.
New file that handles argument setup for x86-64.
Don't use this on x86-64.
Don't use this on x86-64. Remove unused vars.
Don't do stack size calculation for x86-64.
New test that tests a special case in the x86-64 ABI.
Likewise.
Likewise.
Likewise.
Silence warning about this when it's not used.
Add the new tests.
Fix a couple of wrong casts and silence some compiler warnings.
Add x86-64 ABI definition.
Regenerate.
Add x86-64 support.
Likewise.
Regenerate.
Likewise.
Merge settings for similar architectures. Add x86-64 sizes and alignments.
Remove unused vars.
Likewise.
Likewise.
Likewise.
Save/restore %r6. Add DWARF-2 unwind info.
Remove reference to avn.
Remove unused variable and fix formatting.
Declare fd at beginning of function (for older apple cc).
Preserve LD at script entry, and set LD to it when configuring multilibs.
Rebuilt.
Add target.
Likewise.
Regenerate.
Add Linux for S/390 support.
Regenerate.
Add Linux for S/390 support.
Regenerate.
Add Linux for S/390 support.
New file from libffi CVS tree.
New file from libffi CVS tree.
Fix SPARC test in previous change.
Check for working %r_disp32().
Use it.
Likewise.
Rebuilt.
Rebuilt.
Make .eh_frame dependent on target word size.
Add target.
Regenerate.
Add unwind info.
Likewise.
Enable i*86--netbsdelf.
Rebuilt.
Delete
Delete
Enable alpha*--freebsd.
Rebuilt.
libfficonvenience -> libffi_convenience.
Rebuilt.
Define ffitest_OBJECTS.
Rebuilt.
Add Darwin and AIX closure files.
Add aix_closure.S.
Add darwin_closure.S.
Regenerate.
Add AIX and Darwin closure definitions.
New function.
New functions.
New function.
New file.
New file.
Add typedef for ffi_arg.
Declare rint with ffi_arg.
Skip appropriate number of GPRs for floating-point arguments.
Rebuilt.
Replace CHECK_SIZEOF and endian tests with cross-compiler friendly macros.
New macros.
New.
New.
Add Darwin and AIX files.
New.
New stanza.
Regenerate.
Add AIX case.
Regenerate.
Add FFI_AIX.
Use “long” to scale frame size. Fix “long double” support.
Add FFI_AIX case.
New.
Handle powerpc--darwin.
Set source files for POWERPC_DARWIN.
Rebuilt.
Rebuilt.
Define FFI_DARWIN and FFI_DEFAULT_ABI for POWERPC_DARWIN.
New file.
New file.
Fix spelling error of “separate” as “seperate”.
Avoid gas-only .balign directive. Use C style comments.
Avoid gas-only mnemonic. Fixes PR bootstrap/3563.
Use .rdata for ECOFF.
Recognize sparc*-sun-* host.
Regenerate.
Conditionalize for ELF.
Add unwind info.
Add unwind info.
Likewise.
Remove warning about beta code.
Rebuilt.
Restore stack pointer when returning from ffi_closure_UNIX.
Fix typo in comment.
Delete unnecessary increment and decrement of loc2 to eliminate RAW DV.
Make a libtool convenience library.
Rebuilt.
Use different syntax for subdirectory creation.
Rebuilt.
Added X86_WIN32 target (Win32, CygWin, MingW).
Rebuilt.
Added X86_WIN32 target support.
Rebuilt.
Added X86_WIN32 target support.
Doesn't run structure tests for X86_WIN32 targets.
Added X86_WIN32 target support.
New file. Based on sysv.S, but with EH stuff removed and made to work with CygWin's gas.
Make target subdirectory in build dir.
Override suffix based rules to specify correct output subdirectory.
Rebuilt.
Rebuilt.
New file.
Fixed ABI compatibility bug involving long long and register pairs.
New function.
Likewise.
Likewise.
Define on PPC.
Likewise.
Likewise.
Rebuilt.
Added src/powerpc/ppc_closure.S.
Likewise.
Rebuilt.
New macro.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Remove RCS ident string.
Change sourceware.cygnus.com references to gcc.gnu.org.
Simplify struct return test.
Index rather than increment avalue and arg_types. Give ffi_closure_osf the raw return value type.
Handle return value type promotion.
Fix typo.
Likewise.
Define FFI_CLOSURES and FFI_TRAMPOLINE_SIZE.
Adjust minimal cif->bytes for new ffi_call_osf implementation.
Absorb into …
… here. Do all stack allocation here and avoid a callback function.
New.
Reimplement with no callback.
New.
Removed.
Likewise.
Rebuilt.
Include libtool macros from the top level.
Rebuilt.
Set.
Rebuilt.
Fix AC_ARG_ENABLE usage.
Rebuilt.
Doubles are not saved to memory correctly. Use conditional instructions, not branches where possible.
Rebuilt.
Match `arm*--linux-'. From Chris Dornan <cdornan@arm.com>.
Define.
Likewise.
Add section.
Rebuilt.
New variables. Set for gcc using -print-multi-lib. Export them to libtool.
Use libsuff 64 for search paths.
Don't throw away gcc's -B, -b and -V options for -shared links.
Check these for extra compiler options which need to be passed down in compiler_flags.
Rebuilt.
Change i*86-pc-linux* to i*86--linux.
Define for 64bit SPARC builds. Set SPARC FFI_DEFAULT_ABI based on SPARC64 define.
Renamed from ffi_prep_args. Replace all void * sizeofs with sizeof(int). Only compare type with FFI_TYPE_LONGDOUBLE if LONGDOUBLE is different than DOUBLE. Remove FFI_TYPE_SINT32 and FFI_TYPE_UINT32 cases (handled elsewhere).
New function.
Handle V9 ABI and long long on V8.
New function.
Handle FFI_V9 ABI from 64bit code and FFI_V8 ABI from 32bit code (not yet cross-arch calls).
Add struct return delay nop. Handle long long.
New file.
Return structure pointer is used on sparc64 only for structures larger than 32 bytes. Pass by reference for structures is done for structure arguments larger than 16 bytes.
Use 64bit rint on sparc64. Run long long tests on sparc.
Pointer is 64bit on alpha and sparc64.
long double is 128 bit aligned to 128 bits on sparc64.
New supported target.
Likewise.
Rebuilt.
Add v9.S to SPARC files.
Likewise.
Surround $(CCLD) into double quotes, so that multilib compiles work correctly.
Rebuilt.
(i*86--solaris): New libffi target. Patch proposed by Bryce McKinlay.
Hand edit for java_raw_api.lo.
Update from the gcc tree. Fix for PR libgcj/168.
Fixed ia64 by hand.
Rebuilt.
New option.
New subst.
Added multilib code.
Rebuilt.
Use `ia64', not `alpha', as directory name.
New files.
Fixed typo in argument list.
Use ffi_translate_args, not ffi_closure_translate.
New file.
New function.
Define `rint' as long long on IA64. Added new test when FFI_CLOSURES is defined.
Use size_t, not unsigned.
Recognize IA64.
Added `flt' field. Added “Java raw API” code.
Recognize ia64.
New macro.
Added java_raw_api.c.
Define in IA64 case.
Rebuilt with newer automake.
Added -I$(top_srcdir)/src.
Removed PACKAGE and VERSION defines and undefs.
Rebuilt.
Rebuilt.
Pass 3rd argument to AM_INIT_AUTOMAKE. Use AM_PROG_LIBTOOL (automake 1.4 compatibility).
Don't undef PACKAGE or VERSION.
Try to work around messy header problem with PACKAGE and VERSION.
Rebuilt.
Change version to 2.00-beta.
Rebuilt.
Define.
Rename.
New function.
Ditto.
Ditto.
Ditto.
More ditto.
Final ditto.
Add definitions for closure and raw API.
Added case for FFI_TYPE_UINT64.
Added raw_api.c
New file.
New type.
New defines.
New types.
New declarations.
Add check for endianness and sizeof void*.
Call fixup routine via argument, instead of directly.
Rebuilt.
Add x86 and powerpc BeOS configurations. From Makoto Kato <m_kato@ga2.so-net.ne.jp>.
Add warning about this being beta code. Remove src/Makefile.am from the picture.
Rebuilt.
Move logic from src/Makefile.am. Add changes to support libffi as a target library.
Rebuilt.
Upgraded to new autoconf, automake, libtool.
Tweaks.
Update copyright date.
Removed.
Removed.
Removed.
Removed.
Remboved.
Removed.
Created.