|
95df3791
|
2015-02-11T08:31:48
|
|
aarch64: Handle ILP32 ABI
|
|
2104b2a4
|
2015-01-26T12:43:57
|
|
sparc: Re-introduce hand-written unwind info
Fixes the build with the Solaris assembler.
|
|
31a61853
|
2015-01-21T05:55:47
|
|
Merge pull request #170 from fealebenpae/aarch64-trampoline-table
Support closures on ARM64 iOS
|
|
|
3ac1610a
|
2015-01-19T20:48:40
|
|
x86: Fix cygwin32 build
The section syntax is just that little bit different.
|
|
|
1ad0b171
|
2015-01-16T13:30:05
|
|
sparc: Also mark the return address in unwind info
|
|
|
d68c8aed
|
2015-01-16T11:40:33
|
|
sparc: Solaris fixes, part 2
/bin/as seems to only understand single-digit labels
/bin/as knows nothing about .rept/.endr
|
|
b740ab7c
|
2015-01-16T11:32:23
|
|
sparc: Solaris fixes
* /bin/as requires .type fn,#function instead of @function.
* /bin/as doesn't support .macro/.endm. I'm using preprocessor macros
instead to implement E in src/sparc/v[89].S.
|
|
|
f1560b7b
|
2015-01-16T11:31:37
|
|
x86: Solaris fixes
* Solaris/x86 /bin/as doesn't support .org, so I've just disabled the
uses in src/x86/{sysv, unix64}.S, as on Darwin.
* Solaris/x86 needs to use EH_FRAME_FLAGS so manually and compiler
generated .eh_frame sections match, otherwise libffi.so fails to link:
* Solaris/x86 /bin/as has different COMDAT syntax; I've disabled it for
the moment.
|
|
53636634
|
2015-01-16T15:19:38
|
|
aarch64: implement the trampoline table workaround for ffi closures on Apple systems
This is a direct copy/paste port of the ARM code, with changes because of Aarch64 pc-relative addressing restrictions.
|
|
|
05e65b4e
|
2015-01-16T15:18:04
|
|
aarch64: rewrite range syntax into list to appease Clang
Clang's assembler in Xcode 6 appears to choke when the operand of st4 is a range, but is happy with a list.
|
|
|
f27c4e46
|
2015-01-13T07:22:07
|
|
x86: Fix thinko in ffi_raw_call
Missed structure initialization for raw path.
Apparently there are no tests for this outside gcc.
|
|
|
1c61e73a
|
2015-01-10T09:23:30
|
|
Merge pull request #165 from rth7680/pcc
Support PCC as producer and consumer
|
|
|
dd0b59a5
|
2015-01-10T09:22:55
|
|
Merge pull request #164 from rth7680/darwin
Fix build on darwin
|
|
|
9131039c
|
2015-01-10T09:22:42
|
|
Merge pull request #160 from nobu/msvc-no-complex
x86: MSVC does not support Complex type
|
|
|
4ca2262a
|
2015-01-10T09:21:37
|
|
Merge pull request #159 from nobu/fix-void-arith
x86: Fix void pointer arithmetic
|
|
|
3fa5d70c
|
2015-01-05T13:03:06
|
|
x86: Avoid fastcall when building with pcc
Apparently, PCC doesn't support the fastcall calling convention.
Nor does it issue a warning or error for the attribute that it
does not understand.
|
|
|
a03d2310
|
2014-12-24T16:03:34
|
|
x86: Load structure return address into eax
|
|
|
b7f6d7aa
|
2014-12-10T13:37:36
|
|
x86: Reinstate hand-written unwind info for sysv.S
|
|
|
6cedf81c
|
2014-12-10T09:43:58
|
|
x86: Expand FFI_GO_CLOSURE
If we're going to have to hand-write unwind info for darwin,
these macros make the job harder.
|
|
ae842a51
|
2014-11-25T11:43:40
|
|
x86: More Darwin unwind fixups
EHFrame{N} IIRC is a special cue to ld64 that it should treat the unwind
in the object as "special/legacy" .. [these days everything is .cfi_xxxx
(except, cctools-as, as you noted)] .. without that much confusion arises
with ld64's atom-isation of the eh_frame section.
xxxx.eh labels are not needed for darwin ld64 >= 85.2.1 (i.e. darwin9,
xcode 3.1.4) to all intents and purposes, that's all that matters now,
since I think that anyone trying to build on 10.4/darwin8/xcode2.5 would
have to use a later ld64 (from odcctools) for other reasons.
|
|
|
8fa3c9f2
|
2014-11-25T09:27:54
|
|
x86: Reinstate hand-written unwind info for unix64.S
One more try to get default Darwin to work.
|
|
|
5f35e0ff
|
2014-11-24T16:26:50
|
|
x86: Avoid using gas local labels
Which are unsupported by Darwin cctools as.
Thankfully this doesn't uglify the source too much.
|
|
|
ed1ca277
|
2014-11-24T13:02:03
|
|
x86: Remove use of .cfi_escape
The unwind info isn't 100% correct at all points during the epilogue,
and not annotating is just as incorrect as the annotation. This works
better on systems that do not support DW_OP_call_frame_cfa.
|
|
|
1b12593d
|
2014-11-24T12:55:43
|
|
x86: Honor alignment of arguments
Darwin aligns long-double to 16, and thus all of the long double
tests were failing due to not honoring that. We ought to be able
to devise a test case for GCC using __attribute__((aligned)) that
would have failed too.
|
|
|
042b8daf
|
2014-11-24T11:24:02
|
|
x86: Use .balign not .align
The Apple assembler defaults to power of two alignment, rather than
byte alignment like everyone else. Force byte alignment by using
the proper directive.
|
|
|
0172bc02
|
2014-11-24T10:42:02
|
|
x86: Disable .org for Darwin
|
|
|
9f112619
|
2014-11-22T20:02:43
|
|
x86: Best guess at update for Darwin
|
|
7282d328
|
2014-12-22T17:14:40
|
|
x86: MSVC does not support Complex type
|
|
|
5f8881a5
|
2014-12-22T17:08:08
|
|
x86: Fix void pointer arithmetic
|
|
7ba30b19
|
2014-12-19T11:38:17
|
|
s390: Inline and tidy ffi_prep_args
As per discussion with Ulrich Weigand, document the restrictions
on the code within ffi_call_int as we simultaneously prepare
stack frames for ffi_call_SYSV and the target function.
|
|
|
f69ec6f3
|
2014-12-18T16:21:07
|
|
s390: Use pc-relative insns in 31-bit mode
It's silly to stick to esa/390 features when the compiler won't.
Detect when brasl and larl are used by the compiler and then use
them in the assembly.
|
|
|
2f530de1
|
2014-12-18T16:01:59
|
|
s390: Reorganize assembly
Avoid using ffi_prep_args as a callback; do all the work setting
up the frame within ffi_call_int directly. Save fewer registers
in ffi_closure_SYSV.
|
|
|
97512ded
|
2014-12-18T16:01:15
|
|
s390: Avoid aliasing warnings
|
|
|
c860ca9a
|
2014-12-17T13:24:03
|
|
s390: Kill trailing whitespace
|
|
02b7c899
|
2014-12-17T13:20:51
|
|
s390: Go closure support
|
|
|
75b2199f
|
2014-12-12T10:13:56
|
|
Merge pull request #157 from rth7680/x86
Two fixes for x86
|
|
|
8a3a2723
|
2014-12-12T10:13:27
|
|
Merge pull request #156 from rth7680/sparc
sparc: Define FFI_TARGET_SPECIFIC_VARIADIC for v9
|
|
|
2f652469
|
2014-12-11T14:16:00
|
|
x86: Handle void arguments as if an empty structure
Since libffi currently doesn't allow empty structures, libgo
currently maps them to ffi_type_void. Given that we'll abort
on this case, handle it gracefully.
|
|
|
097ccfd6
|
2014-12-10T13:25:14
|
|
x86: Fix some unwind errors
|
|
9ae3bc87
|
2014-12-06T23:58:41
|
|
Fix for https://github.com/atgreen/libffi/issues/141
|
|
|
590663b3
|
2014-11-18T12:56:58
|
|
powerpc: Fix ffi_go_closure_linux64
Unlike ffi_closure_LINUX64, this entry point is called normally,
so we already have the TOC in R2 and the closure in R11.
* powerpc/linux64_closure.S (ffi_closure_LINUX64): Remove a
register dependency chain.
(ffi_go_closure_linux64): Don't load r11 or r2.
|
|
fa1040c1
|
2014-11-10T09:42:31
|
|
GO closures for powerpc linux
Plus .cfi async unwind info, rearrangement of ffi_call_linux64 and
ffi_call_SYSV function params to avoid register copies, tweaks to
trampolines.
* src/powerpc/ffitarget.h (FFI_GO_CLOSURES): Define.
* src/powerpc/ffi.c (ffi_call_int): New function with extra
closure param, and args rearranged on ffi_call_linux64 and
ffi_call_SYSV calls, extracted from ..
(ffi_call): ..here.
(ffi_call_go, ffi_prep_go_closure): New functions.
* src/powerpc/ffi_linux64.c (ffi_prep_closure_loc_linux64): Make
hidden. Only flush insn part of ELFv2 trampoline. Don't shuffle
ELFv1 trampoline.
(ffi_closure_helper_LINUX64): Replace closure param with cif, fun,
user_data params.
* src/powerpc/ffi_powerpc.h (ffi_go_closure_sysv): Declare.
(ffi_go_closure_linux64): Declare.
(ffi_call_SYSV, fi_call_LINUX64): Update.
(ffi_prep_closure_loc_sysv, ffi_prep_closure_loc_linux64): Declare.
(ffi_closure_helper_SYSV, ffi_closure_helper_LINUX64): Update.
* src/powerpc/ffi_sysv.c (ASM_NEEDS_REGISTERS): Increase to 6.
(ffi_prep_closure_loc_sysv): Use bcl in trampoline, put data words
last, flush just the insn part.
(ffi_closure_helper_SYSV): Replace closure param with cif, fun and
user_data params.
* src/powerpc/linux64.S (ffi_call_LINUX64): Replace hand-written
.eh_frame with .cfi directives. Adjust for changed param order.
Pass extra "closure" param to user function in static chain. Add
.cfi directives to describe epilogue. Don't provide traceback
table for ELFv2 or _CALL_LINUX.
* src/powerpc/linux64_closure.S (ffi_closure_LINUX64): Replace
hand-written .eh_frame with .cfi directives. Adjust for changed
ffi_closure_helper_LINUX64 params. Add .cfi directives to
describe epilogue. Don't provide traceback table for ELFv2 or
_CALL_LINUX.
(ffi_go_closure_linux64): New function.
* src/powerpc/sysv.S: Remove redundant .globl ffi_prep_args_SYSV.
(ffi_call_SYSV): Make hidden. Replace hand-written .eh_frame with
.cfi directives. Adjust for changed params. Pass extra "closure"
param to user function in static chain. Add .cfi directives to
describe epilogue.
* src/powerpc/ppc_closure.S (ffi_closure_SYSV): Make hidden.
Replace hand-written .eh_frame with .cfi directives. Adjust for
changed ffi_closure_helper_SYSV params. Add .cfi directives to
describe epilogue. Don't just use nops in the dead __NO_FPRS__
epilogues.
(ffi_go_closure_sysv): New function.
|
|
|
d3d06f4c
|
2014-11-10T13:50:05
|
|
Fix powerpc breakage from 6e8a4460
* src/powerpc/ffitarget.h: #error on unexpected FFI_TYPE_LAST.
(FFI_PPC_TYPE_LAST): Define.
(FFI_TYPE_UINT128): Define in terms of FFI_PPC_TYPE_LAST.
(FFI_SYSV_TYPE_SMALL_STRUCT, FFI_V2_TYPE_FLOAT_HOMOG): Likewise.
(FFI_V2_TYPE_DOUBLE_HOMOG, FFI_V2_TYPE_SMALL_STRUCT): Likewise.
|
|
|
542e0047
|
2014-11-18T05:07:00
|
|
sparc: Define FFI_TARGET_SPECIFIC_VARIADIC for v9
This is a port of
http://gcc.gnu.org/viewcvs?rev=207763&root=gcc&view=rev
aka GCC PR libffi/60073, to the rewritten Sparc codebase.
Supposedly, we should have seen failures with the existing
libffi.call/cls_double_va.c testcase, but I hadn't.
Perhaps a gcc newer than 4.6.3 is required to see that...
|
|
a9ed0c3a
|
2014-11-18T09:37:07
|
|
dlmalloc: change defined(i386) to defined(__i386__)
When compiling with --std==c99, GCC and clang don't define i386 but __i386__
|
|
|
dea49e20
|
2014-11-14T13:05:14
|
|
x86: Fix typo in ffi_prep_go_closure
Used the wrong register for THISCALL and FASTCALL.
|
|
|
f8632815
|
2014-11-13T12:32:35
|
|
powerpc: Delete patch output
Clearly added by mistake.
|
|
|
20562ac0
|
2014-11-12T07:00:59
|
|
Fix for AArch64. Release as 3.2.1.
|
|
|
0e303c06
|
2014-11-12T03:58:58
|
|
x86: Work around clang bugs
http://llvm.org/bugs/show_bug.cgi?21500
http://llvm.org/bugs/show_bug.cgi?21501
http://llvm.org/bugs/show_bug.cgi?21515
|
|
|
6eec410f
|
2014-10-26T15:29:04
|
|
sparc: Re-add abi compliant structure support
The original code, removed in the "rewrite" patch, was incorrect for
large structures, and required dynamic allocation of a trampoline on
every ffi_call.
Instead, allocate a 4k entry table of all possible structure returns.
The table is 80k, but is read-only and dynamically paged, which ought
to be better than allocating the trampoline.
This is difficult to test with gcc. One can only use -O0 at present.
See https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63668.
|
|
|
92022496
|
2014-10-26T14:48:28
|
|
sparc: Add support for Go closures
|
|
|
ad89c2d9
|
2014-10-26T13:16:03
|
|
sparc: Add support for complex types
|
|
|
20da5b41
|
2014-10-25T16:24:41
|
|
sparc: Handle more cases of structure return directly
|
|
|
0686c2e7
|
2014-10-25T14:52:45
|
|
sparc: Preprocess float point struct return
We can eliminate recursion and speed structure return
by flattening a nested structure tree into a bitmask.
|
|
|
2b27890b
|
2014-10-24T16:10:48
|
|
sparc: Rewrite everything
It's impossible to call between v8 and v9 ABIs, because of the stack bias
in the v9 ABI. So let's not pretend it's just not implemented yet. Split
the v9 code out to a separate file.
The register windows prevent ffi_call from setting up the entire stack
frame the assembly, but we needn't make an indirect call back to prep_args.
|
|
|
5d7b5445
|
2014-10-24T11:45:53
|
|
sparc: Tidy up symbols
Assembly to use local labels, .type annotation, hidden annotation.
I do retain the _prefix for the symbols, but given that it wasn't
done consistently across all symbols, I doubt it's actually needed.
|
|
|
fd4f7cc0
|
2014-10-24T11:11:53
|
|
sparc: Eliminate long double ifdefs
|
|
|
ab83cbb9
|
2014-10-29T14:38:42
|
|
arm: Add support for Go closures
|
|
|
6fa617da
|
2014-10-21T11:27:11
|
|
arm: Add argument space for the hidden struct return pointer
This should have been failing all along, but it's only
exposed by the complex_int test case.
|
|
|
a529bec2
|
2014-10-21T11:26:59
|
|
arm: Add support for complex types
|
|
|
5e88ebe6
|
2014-10-20T15:10:43
|
|
arm: Remove internal FFI_TYPE constants
These have been replaced by the contents of internal.h.
|
|
|
a4b785ea
|
2014-10-17T02:07:32
|
|
arm: Rewrite ffi_closure
Move the push of the argument registers into ffi_closure_SYSV,
reducing the size of the trampoline.
|
|
|
e7f15f60
|
2014-10-17T01:27:16
|
|
arm: Rewrite ffi_call
Use the trick to allocate the stack frame for ffi_call_SYSV
within ffi_call itself.
|
|
|
a74a3aad
|
2014-10-17T01:21:22
|
|
arm: Rewrite vfp_type_p
Do not modify the ffi_type. Rearrange the tests so that we
quickly eliminate structures that cannot match. Return an
encoded value of element count and base type.
|
|
|
0d39b4bb
|
2014-10-17T01:02:52
|
|
arm: Deref ffi_put_arg arguments
|
|
|
57b24fb3
|
2014-10-17T00:53:21
|
|
arm: Deref ffi_align argument
|
|
|
c129bea8
|
2014-10-15T17:28:53
|
|
arm: Reindent arm/ffi.c
|
|
|
9761b7bb
|
2014-10-17T21:26:52
|
|
alpha: Add support for Go closures
|
|
|
f41bec3b
|
2014-10-17T20:46:48
|
|
alpha: Add support for complex types
|
|
|
5f917371
|
2014-10-17T13:47:26
|
|
alpha: Clean up conversion of float values
Don't use "real" conversion to double, lest we raise
exceptions when passing signalling nans.
|
|
|
32a26b75
|
2014-10-17T12:33:53
|
|
alpha: Reorganize cif flags
Unties the backend from changes to FFI_TYPE_* constants, and allows
compilation to succeed after the addition of FFI_TYPE_COMPLEX.
Delete the hand-written unwind info.
|
|
|
c6352b66
|
2014-10-23T00:26:14
|
|
aarch64: Add support for Go closures
|
|
|
0e41c73b
|
2014-10-22T23:48:12
|
|
aarch64: Move x8 out of call_context
Reduces stack size. It was only used by the closure, and there
are available argument registers.
|
|
|
a992f878
|
2014-10-22T22:58:09
|
|
aarch64: Add support for complex types
|
|
|
b55e0366
|
2014-10-22T12:33:59
|
|
aarch64: Treat void return as not passed in registers
This lets us do less post-processing when there's no return value.
|
|
|
2e32f9bf
|
2014-10-21T22:49:05
|
|
aarch64: Use correct return registers
There are fewer return registers than argument registers.
|
|
|
95a04af1
|
2014-10-21T22:41:07
|
|
aarch64: Reduce the size of register_context
We don't need to store 32 general and vector registers.
Only 8 of each are used for parameter passing.
|
|
|
77c4cddc
|
2014-10-21T13:30:40
|
|
aarch64: Simplify AARCH64_STACK_ALIGN
The iOS abi doesn't require padding between arguments, but
that's not what AARCH64_STACK_ALIGN meant. The hardware will
in fact trap if the SP register is not 16 byte aligned.
|
|
|
b5f147d8
|
2014-10-21T13:27:57
|
|
aarch64: Always distinguish LONGDOUBLE
Avoid if-deffery by forcing FFI_TYPE_LONGDOUBLE different
from FFI_TYPE_DOUBLE. This will simply be unused on hosts
that define them identically.
|
|
|
38b54b9c
|
2014-10-21T13:17:39
|
|
aarch64: Improve is_hfa
The set of functions get_homogeneous_type, element_count, and is_hfa
are all intertwined and recompute data. Return a compound quantity
from is_hfa that contains all the data and avoids the recomputation.
|
|
|
18b74ce5
|
2014-10-21T13:00:34
|
|
aarch64: Fix non-apple compilation
|
|
|
658b2b56
|
2014-10-22T22:36:07
|
|
aarch64: Remove aarch64_flags
This field was useless from the start, since the normal flags
field is available for backend use.
|
|
|
4a3cbcaa
|
2014-10-22T22:32:13
|
|
aarch64: Unify scalar fp and hfa handling
Since an HFA of a single element is exactly the same as scalar,
this tidies things up a bit.
|
|
|
12cf89ee
|
2014-10-22T21:53:30
|
|
aarch64: Move return value handling into ffi_closure_SYSV
As with the change to ffi_call_SYSV, this avoids copying data
into a temporary buffer.
|
|
|
4fe1aea1
|
2014-10-22T17:06:19
|
|
aarch64: Move return value handling into ffi_call_SYSV
This lets us pass return data directly to the caller of ffi_call
in most cases, rather than storing it into temporary storage first.
|
|
|
325471ea
|
2014-10-22T13:58:59
|
|
aarch64: Merge prep_args with ffi_call
Use the trick to allocate the stack frame for ffi_call_SYSV
within ffi_call itself.
|
|
|
8c8161cb
|
2014-10-22T12:52:07
|
|
aarch64: Tidy up abi manipulation
Avoid false abstraction, like get_x_addr. Avoid recomputing data
about the type being manipulated. Use NEON insns for HFA manipulation.
Note that some of the inline assembly will go away in a subsequent patch.
|
|
|
2650f47f
|
2014-11-06T10:57:04
|
|
x86: Use win32 name mangling for fastcall functions
|
|
|
f8c64e24
|
2014-11-05T17:04:29
|
|
x86: Add support for Go closures
|
|
|
198f469e
|
2014-11-05T16:34:41
|
|
x86: Add support for Complex
|
|
|
b21ec1ce
|
2014-11-05T10:15:25
|
|
x86: Rewrite closures
Move everything into sysv.S, removing win32.S and freebsd.S.
Handle all abis with a single ffi_closure_inner function.
Move complexity of the raw THISCALL trampoline into assembly
instead of the trampoline itself.
Only push the context for the REGISTER abi; let the rest
receive it in a register.
|
|
|
b9ac94f3
|
2014-11-01T15:10:34
|
|
x86: Rewrite ffi_call
Decouple the assembly from FFI_TYPE_*. Merge prep_args with ffi_call,
passing the frame and the stack to the assembly.
Note that this patch isn't really standalone, as this breaks closures.
|
|
|
159d3788
|
2014-10-31T12:07:02
|
|
x86: Convert to gas generated unwind info
|
|
|
e7b0056d
|
2014-10-30T13:57:39
|
|
x86: Force FFI_TYPE_LONGDOUBLE different from FFI_TYPE_DOUBLE
There are few abis that set double = long double. Eliminate the
conditional compilation and let this code simply be unused there.
|
|
|
4b2fad8f
|
2014-10-30T12:41:31
|
|
x86: Remove some conditional compilation
Removal of ifdefs made possible to due to ffi_abi unification.
|
|
|
ef762056
|
2014-10-30T12:13:31
|
|
x86: Tidy ffi_abi
The x86_64 unix port only handles one ABI; don't define all of the
other symbols. The UNIX64 symbol retains the same value.
The i386 ports ought to have the same symbols, even if we can't yet
unify the values without incrementing the libffi soname.
|
|
|
610c90bf
|
2014-10-28T11:21:50
|
|
x86_64: Add support for complex types
|
|
|
32c56831
|
2014-10-28T11:17:35
|
|
x86_64: Decouple return types from FFI_TYPE constants
We can better support structure returns, and as prep for
complex types.
|
|
|
2e9dc165
|
2014-10-27T13:41:39
|
|
x86_64: Fixups for x32
|
|
|
ebd82769
|
2014-10-23T23:57:06
|
|
win64: Remove support from ffi.c
|