From fcbf63e62c627deae76c1b8cb8c0876c536ed811 Mon Sep 17 00:00:00 2001 From: Jari Vetoniemi Date: Mon, 16 Mar 2020 18:49:26 +0900 Subject: Fresh start --- jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin.S | 444 +++++++ .../ext/fiddle/libffi-3.2.1/src/x86/darwin64.S | 416 ++++++ jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi.c | 935 ++++++++++++++ jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi64.c | 676 ++++++++++ .../ext/fiddle/libffi-3.2.1/src/x86/ffitarget.h | 152 +++ jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/freebsd.S | 463 +++++++ jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/sysv.S | 483 +++++++ jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/unix64.S | 432 +++++++ jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win32.S | 1351 ++++++++++++++++++++ jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win64.S | 519 ++++++++ 10 files changed, 5871 insertions(+) create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin.S create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin64.S create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi.c create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi64.c create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffitarget.h create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/freebsd.S create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/sysv.S create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/unix64.S create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win32.S create mode 100644 jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win64.S (limited to 'jni/ruby/ext/fiddle/libffi-3.2.1/src/x86') diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin.S b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin.S new file mode 100644 index 0000000..8f0f070 --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin.S @@ -0,0 +1,444 @@ +/* ----------------------------------------------------------------------- + darwin.S - Copyright (c) 1996, 1998, 2001, 2002, 2003, 2005 Red Hat, Inc. + Copyright (C) 2008 Free Software Foundation, Inc. + + X86 Foreign Function Interface + + Permission is hereby granted, free of charge, to any person obtaining + a copy of this software and associated documentation files (the + ``Software''), to deal in the Software without restriction, including + without limitation the rights to use, copy, modify, merge, publish, + distribute, sublicense, and/or sell copies of the Software, and to + permit persons to whom the Software is furnished to do so, subject to + the following conditions: + + The above copyright notice and this permission notice shall be included + in all copies or substantial portions of the Software. + + THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, + EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + DEALINGS IN THE SOFTWARE. + ----------------------------------------------------------------------- + */ + +#ifndef __x86_64__ + +#define LIBFFI_ASM +#include +#include + +.text + +.globl _ffi_prep_args + + .align 4 +.globl _ffi_call_SYSV + +_ffi_call_SYSV: +.LFB1: + pushl %ebp +.LCFI0: + movl %esp,%ebp +.LCFI1: + subl $8,%esp + /* Make room for all of the new args. */ + movl 16(%ebp),%ecx + subl %ecx,%esp + + movl %esp,%eax + + /* Place all of the ffi_prep_args in position */ + subl $8,%esp + pushl 12(%ebp) + pushl %eax + call *8(%ebp) + + /* Return stack to previous state and call the function */ + addl $16,%esp + + call *28(%ebp) + + /* Load %ecx with the return type code */ + movl 20(%ebp),%ecx + + /* Protect %esi. We're going to pop it in the epilogue. */ + pushl %esi + + /* If the return value pointer is NULL, assume no return value. */ + cmpl $0,24(%ebp) + jne 0f + + /* Even if there is no space for the return value, we are + obliged to handle floating-point values. */ + cmpl $FFI_TYPE_FLOAT,%ecx + jne noretval + fstp %st(0) + + jmp epilogue +0: + .align 4 + call 1f +.Lstore_table: + .long noretval-.Lstore_table /* FFI_TYPE_VOID */ + .long retint-.Lstore_table /* FFI_TYPE_INT */ + .long retfloat-.Lstore_table /* FFI_TYPE_FLOAT */ + .long retdouble-.Lstore_table /* FFI_TYPE_DOUBLE */ + .long retlongdouble-.Lstore_table /* FFI_TYPE_LONGDOUBLE */ + .long retuint8-.Lstore_table /* FFI_TYPE_UINT8 */ + .long retsint8-.Lstore_table /* FFI_TYPE_SINT8 */ + .long retuint16-.Lstore_table /* FFI_TYPE_UINT16 */ + .long retsint16-.Lstore_table /* FFI_TYPE_SINT16 */ + .long retint-.Lstore_table /* FFI_TYPE_UINT32 */ + .long retint-.Lstore_table /* FFI_TYPE_SINT32 */ + .long retint64-.Lstore_table /* FFI_TYPE_UINT64 */ + .long retint64-.Lstore_table /* FFI_TYPE_SINT64 */ + .long retstruct-.Lstore_table /* FFI_TYPE_STRUCT */ + .long retint-.Lstore_table /* FFI_TYPE_POINTER */ + .long retstruct1b-.Lstore_table /* FFI_TYPE_SMALL_STRUCT_1B */ + .long retstruct2b-.Lstore_table /* FFI_TYPE_SMALL_STRUCT_2B */ +1: + pop %esi + add (%esi, %ecx, 4), %esi + jmp *%esi + + /* Sign/zero extend as appropriate. */ +retsint8: + movsbl %al, %eax + jmp retint + +retsint16: + movswl %ax, %eax + jmp retint + +retuint8: + movzbl %al, %eax + jmp retint + +retuint16: + movzwl %ax, %eax + jmp retint + +retfloat: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstps (%ecx) + jmp epilogue + +retdouble: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstpl (%ecx) + jmp epilogue + +retlongdouble: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstpt (%ecx) + jmp epilogue + +retint64: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + movl %edx,4(%ecx) + jmp epilogue + +retstruct1b: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movb %al,0(%ecx) + jmp epilogue + +retstruct2b: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movw %ax,0(%ecx) + jmp epilogue + +retint: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + +retstruct: + /* Nothing to do! */ + +noretval: +epilogue: + popl %esi + movl %ebp,%esp + popl %ebp + ret + +.LFE1: +.ffi_call_SYSV_end: + + .align 4 +FFI_HIDDEN (ffi_closure_SYSV) +.globl _ffi_closure_SYSV + +_ffi_closure_SYSV: +.LFB2: + pushl %ebp +.LCFI2: + movl %esp, %ebp +.LCFI3: + subl $40, %esp + leal -24(%ebp), %edx + movl %edx, -12(%ebp) /* resp */ + leal 8(%ebp), %edx + movl %edx, 4(%esp) /* args = __builtin_dwarf_cfa () */ + leal -12(%ebp), %edx + movl %edx, (%esp) /* &resp */ + movl %ebx, 8(%esp) +.LCFI7: + call L_ffi_closure_SYSV_inner$stub + movl 8(%esp), %ebx + movl -12(%ebp), %ecx + cmpl $FFI_TYPE_INT, %eax + je .Lcls_retint + + /* Handle FFI_TYPE_UINT8, FFI_TYPE_SINT8, FFI_TYPE_UINT16, + FFI_TYPE_SINT16, FFI_TYPE_UINT32, FFI_TYPE_SINT32. */ + cmpl $FFI_TYPE_UINT64, %eax + jge 0f + cmpl $FFI_TYPE_UINT8, %eax + jge .Lcls_retint + +0: cmpl $FFI_TYPE_FLOAT, %eax + je .Lcls_retfloat + cmpl $FFI_TYPE_DOUBLE, %eax + je .Lcls_retdouble + cmpl $FFI_TYPE_LONGDOUBLE, %eax + je .Lcls_retldouble + cmpl $FFI_TYPE_SINT64, %eax + je .Lcls_retllong + cmpl $FFI_TYPE_SMALL_STRUCT_1B, %eax + je .Lcls_retstruct1b + cmpl $FFI_TYPE_SMALL_STRUCT_2B, %eax + je .Lcls_retstruct2b + cmpl $FFI_TYPE_STRUCT, %eax + je .Lcls_retstruct +.Lcls_epilogue: + movl %ebp, %esp + popl %ebp + ret +.Lcls_retint: + movl (%ecx), %eax + jmp .Lcls_epilogue +.Lcls_retfloat: + flds (%ecx) + jmp .Lcls_epilogue +.Lcls_retdouble: + fldl (%ecx) + jmp .Lcls_epilogue +.Lcls_retldouble: + fldt (%ecx) + jmp .Lcls_epilogue +.Lcls_retllong: + movl (%ecx), %eax + movl 4(%ecx), %edx + jmp .Lcls_epilogue +.Lcls_retstruct1b: + movsbl (%ecx), %eax + jmp .Lcls_epilogue +.Lcls_retstruct2b: + movswl (%ecx), %eax + jmp .Lcls_epilogue +.Lcls_retstruct: + lea -8(%ebp),%esp + movl %ebp, %esp + popl %ebp + ret $4 +.LFE2: + +#if !FFI_NO_RAW_API + +#define RAW_CLOSURE_CIF_OFFSET ((FFI_TRAMPOLINE_SIZE + 3) & ~3) +#define RAW_CLOSURE_FUN_OFFSET (RAW_CLOSURE_CIF_OFFSET + 4) +#define RAW_CLOSURE_USER_DATA_OFFSET (RAW_CLOSURE_FUN_OFFSET + 4) +#define CIF_FLAGS_OFFSET 20 + + .align 4 +FFI_HIDDEN (ffi_closure_raw_SYSV) +.globl _ffi_closure_raw_SYSV + +_ffi_closure_raw_SYSV: +.LFB3: + pushl %ebp +.LCFI4: + movl %esp, %ebp +.LCFI5: + pushl %esi +.LCFI6: + subl $36, %esp + movl RAW_CLOSURE_CIF_OFFSET(%eax), %esi /* closure->cif */ + movl RAW_CLOSURE_USER_DATA_OFFSET(%eax), %edx /* closure->user_data */ + movl %edx, 12(%esp) /* user_data */ + leal 8(%ebp), %edx /* __builtin_dwarf_cfa () */ + movl %edx, 8(%esp) /* raw_args */ + leal -24(%ebp), %edx + movl %edx, 4(%esp) /* &res */ + movl %esi, (%esp) /* cif */ + call *RAW_CLOSURE_FUN_OFFSET(%eax) /* closure->fun */ + movl CIF_FLAGS_OFFSET(%esi), %eax /* rtype */ + cmpl $FFI_TYPE_INT, %eax + je .Lrcls_retint + + /* Handle FFI_TYPE_UINT8, FFI_TYPE_SINT8, FFI_TYPE_UINT16, + FFI_TYPE_SINT16, FFI_TYPE_UINT32, FFI_TYPE_SINT32. */ + cmpl $FFI_TYPE_UINT64, %eax + jge 0f + cmpl $FFI_TYPE_UINT8, %eax + jge .Lrcls_retint +0: + cmpl $FFI_TYPE_FLOAT, %eax + je .Lrcls_retfloat + cmpl $FFI_TYPE_DOUBLE, %eax + je .Lrcls_retdouble + cmpl $FFI_TYPE_LONGDOUBLE, %eax + je .Lrcls_retldouble + cmpl $FFI_TYPE_SINT64, %eax + je .Lrcls_retllong +.Lrcls_epilogue: + addl $36, %esp + popl %esi + popl %ebp + ret +.Lrcls_retint: + movl -24(%ebp), %eax + jmp .Lrcls_epilogue +.Lrcls_retfloat: + flds -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retdouble: + fldl -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retldouble: + fldt -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retllong: + movl -24(%ebp), %eax + movl -20(%ebp), %edx + jmp .Lrcls_epilogue +.LFE3: +#endif + +.section __IMPORT,__jump_table,symbol_stubs,self_modifying_code+pure_instructions,5 +L_ffi_closure_SYSV_inner$stub: + .indirect_symbol _ffi_closure_SYSV_inner + hlt ; hlt ; hlt ; hlt ; hlt + + +.section __TEXT,__eh_frame,coalesced,no_toc+strip_static_syms+live_support +EH_frame1: + .set L$set$0,LECIE1-LSCIE1 + .long L$set$0 +LSCIE1: + .long 0x0 + .byte 0x1 + .ascii "zR\0" + .byte 0x1 + .byte 0x7c + .byte 0x8 + .byte 0x1 + .byte 0x10 + .byte 0xc + .byte 0x5 + .byte 0x4 + .byte 0x88 + .byte 0x1 + .align 2 +LECIE1: +.globl _ffi_call_SYSV.eh +_ffi_call_SYSV.eh: +LSFDE1: + .set L$set$1,LEFDE1-LASFDE1 + .long L$set$1 +LASFDE1: + .long LASFDE1-EH_frame1 + .long .LFB1-. + .set L$set$2,.LFE1-.LFB1 + .long L$set$2 + .byte 0x0 + .byte 0x4 + .set L$set$3,.LCFI0-.LFB1 + .long L$set$3 + .byte 0xe + .byte 0x8 + .byte 0x84 + .byte 0x2 + .byte 0x4 + .set L$set$4,.LCFI1-.LCFI0 + .long L$set$4 + .byte 0xd + .byte 0x4 + .align 2 +LEFDE1: +.globl _ffi_closure_SYSV.eh +_ffi_closure_SYSV.eh: +LSFDE2: + .set L$set$5,LEFDE2-LASFDE2 + .long L$set$5 +LASFDE2: + .long LASFDE2-EH_frame1 + .long .LFB2-. + .set L$set$6,.LFE2-.LFB2 + .long L$set$6 + .byte 0x0 + .byte 0x4 + .set L$set$7,.LCFI2-.LFB2 + .long L$set$7 + .byte 0xe + .byte 0x8 + .byte 0x84 + .byte 0x2 + .byte 0x4 + .set L$set$8,.LCFI3-.LCFI2 + .long L$set$8 + .byte 0xd + .byte 0x4 + .align 2 +LEFDE2: + +#if !FFI_NO_RAW_API + +.globl _ffi_closure_raw_SYSV.eh +_ffi_closure_raw_SYSV.eh: +LSFDE3: + .set L$set$10,LEFDE3-LASFDE3 + .long L$set$10 +LASFDE3: + .long LASFDE3-EH_frame1 + .long .LFB3-. + .set L$set$11,.LFE3-.LFB3 + .long L$set$11 + .byte 0x0 + .byte 0x4 + .set L$set$12,.LCFI4-.LFB3 + .long L$set$12 + .byte 0xe + .byte 0x8 + .byte 0x84 + .byte 0x2 + .byte 0x4 + .set L$set$13,.LCFI5-.LCFI4 + .long L$set$13 + .byte 0xd + .byte 0x4 + .byte 0x4 + .set L$set$14,.LCFI6-.LCFI5 + .long L$set$14 + .byte 0x85 + .byte 0x3 + .align 2 +LEFDE3: + +#endif + +#endif /* ifndef __x86_64__ */ diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin64.S b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin64.S new file mode 100644 index 0000000..2f7394e --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/darwin64.S @@ -0,0 +1,416 @@ +/* ----------------------------------------------------------------------- + darwin64.S - Copyright (c) 2006 Free Software Foundation, Inc. + Copyright (c) 2008 Red Hat, Inc. + derived from unix64.S + + x86-64 Foreign Function Interface for Darwin. + + 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 AUTHOR 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. + ----------------------------------------------------------------------- */ + +#ifdef __x86_64__ +#define LIBFFI_ASM +#include +#include + + .file "darwin64.S" +.text + +/* ffi_call_unix64 (void *args, unsigned long bytes, unsigned flags, + void *raddr, void (*fnaddr)(void)); + + Bit o trickiness here -- ARGS+BYTES is the base of the stack frame + for this function. This has been allocated by ffi_call. We also + deallocate some of the stack that has been alloca'd. */ + + .align 3 + .globl _ffi_call_unix64 + +_ffi_call_unix64: +LUW0: + movq (%rsp), %r10 /* Load return address. */ + leaq (%rdi, %rsi), %rax /* Find local stack base. */ + movq %rdx, (%rax) /* Save flags. */ + movq %rcx, 8(%rax) /* Save raddr. */ + movq %rbp, 16(%rax) /* Save old frame pointer. */ + movq %r10, 24(%rax) /* Relocate return address. */ + movq %rax, %rbp /* Finalize local stack frame. */ +LUW1: + movq %rdi, %r10 /* Save a copy of the register area. */ + movq %r8, %r11 /* Save a copy of the target fn. */ + movl %r9d, %eax /* Set number of SSE registers. */ + + /* Load up all argument registers. */ + movq (%r10), %rdi + movq 8(%r10), %rsi + movq 16(%r10), %rdx + movq 24(%r10), %rcx + movq 32(%r10), %r8 + movq 40(%r10), %r9 + testl %eax, %eax + jnz Lload_sse +Lret_from_load_sse: + + /* Deallocate the reg arg area. */ + leaq 176(%r10), %rsp + + /* Call the user function. */ + call *%r11 + + /* Deallocate stack arg area; local stack frame in redzone. */ + leaq 24(%rbp), %rsp + + movq 0(%rbp), %rcx /* Reload flags. */ + movq 8(%rbp), %rdi /* Reload raddr. */ + movq 16(%rbp), %rbp /* Reload old frame pointer. */ +LUW2: + + /* The first byte of the flags contains the FFI_TYPE. */ + movzbl %cl, %r10d + leaq Lstore_table(%rip), %r11 + movslq (%r11, %r10, 4), %r10 + addq %r11, %r10 + jmp *%r10 + +Lstore_table: + .long Lst_void-Lstore_table /* FFI_TYPE_VOID */ + .long Lst_sint32-Lstore_table /* FFI_TYPE_INT */ + .long Lst_float-Lstore_table /* FFI_TYPE_FLOAT */ + .long Lst_double-Lstore_table /* FFI_TYPE_DOUBLE */ + .long Lst_ldouble-Lstore_table /* FFI_TYPE_LONGDOUBLE */ + .long Lst_uint8-Lstore_table /* FFI_TYPE_UINT8 */ + .long Lst_sint8-Lstore_table /* FFI_TYPE_SINT8 */ + .long Lst_uint16-Lstore_table /* FFI_TYPE_UINT16 */ + .long Lst_sint16-Lstore_table /* FFI_TYPE_SINT16 */ + .long Lst_uint32-Lstore_table /* FFI_TYPE_UINT32 */ + .long Lst_sint32-Lstore_table /* FFI_TYPE_SINT32 */ + .long Lst_int64-Lstore_table /* FFI_TYPE_UINT64 */ + .long Lst_int64-Lstore_table /* FFI_TYPE_SINT64 */ + .long Lst_struct-Lstore_table /* FFI_TYPE_STRUCT */ + .long Lst_int64-Lstore_table /* FFI_TYPE_POINTER */ + + .text + .align 3 +Lst_void: + ret + .align 3 +Lst_uint8: + movzbq %al, %rax + movq %rax, (%rdi) + ret + .align 3 +Lst_sint8: + movsbq %al, %rax + movq %rax, (%rdi) + ret + .align 3 +Lst_uint16: + movzwq %ax, %rax + movq %rax, (%rdi) + .align 3 +Lst_sint16: + movswq %ax, %rax + movq %rax, (%rdi) + ret + .align 3 +Lst_uint32: + movl %eax, %eax + movq %rax, (%rdi) + .align 3 +Lst_sint32: + cltq + movq %rax, (%rdi) + ret + .align 3 +Lst_int64: + movq %rax, (%rdi) + ret + .align 3 +Lst_float: + movss %xmm0, (%rdi) + ret + .align 3 +Lst_double: + movsd %xmm0, (%rdi) + ret +Lst_ldouble: + fstpt (%rdi) + ret + .align 3 +Lst_struct: + leaq -20(%rsp), %rsi /* Scratch area in redzone. */ + + /* We have to locate the values now, and since we don't want to + write too much data into the user's return value, we spill the + value to a 16 byte scratch area first. Bits 8, 9, and 10 + control where the values are located. Only one of the three + bits will be set; see ffi_prep_cif_machdep for the pattern. */ + movd %xmm0, %r10 + movd %xmm1, %r11 + testl $0x100, %ecx + cmovnz %rax, %rdx + cmovnz %r10, %rax + testl $0x200, %ecx + cmovnz %r10, %rdx + testl $0x400, %ecx + cmovnz %r10, %rax + cmovnz %r11, %rdx + movq %rax, (%rsi) + movq %rdx, 8(%rsi) + + /* Bits 12-31 contain the true size of the structure. Copy from + the scratch area to the true destination. */ + shrl $12, %ecx + rep movsb + ret + + /* Many times we can avoid loading any SSE registers at all. + It's not worth an indirect jump to load the exact set of + SSE registers needed; zero or all is a good compromise. */ + .align 3 +LUW3: +Lload_sse: + movdqa 48(%r10), %xmm0 + movdqa 64(%r10), %xmm1 + movdqa 80(%r10), %xmm2 + movdqa 96(%r10), %xmm3 + movdqa 112(%r10), %xmm4 + movdqa 128(%r10), %xmm5 + movdqa 144(%r10), %xmm6 + movdqa 160(%r10), %xmm7 + jmp Lret_from_load_sse + +LUW4: + .align 3 + .globl _ffi_closure_unix64 + +_ffi_closure_unix64: +LUW5: + /* The carry flag is set by the trampoline iff SSE registers + are used. Don't clobber it before the branch instruction. */ + leaq -200(%rsp), %rsp +LUW6: + movq %rdi, (%rsp) + movq %rsi, 8(%rsp) + movq %rdx, 16(%rsp) + movq %rcx, 24(%rsp) + movq %r8, 32(%rsp) + movq %r9, 40(%rsp) + jc Lsave_sse +Lret_from_save_sse: + + movq %r10, %rdi + leaq 176(%rsp), %rsi + movq %rsp, %rdx + leaq 208(%rsp), %rcx + call _ffi_closure_unix64_inner + + /* Deallocate stack frame early; return value is now in redzone. */ + addq $200, %rsp +LUW7: + + /* The first byte of the return value contains the FFI_TYPE. */ + movzbl %al, %r10d + leaq Lload_table(%rip), %r11 + movslq (%r11, %r10, 4), %r10 + addq %r11, %r10 + jmp *%r10 + +Lload_table: + .long Lld_void-Lload_table /* FFI_TYPE_VOID */ + .long Lld_int32-Lload_table /* FFI_TYPE_INT */ + .long Lld_float-Lload_table /* FFI_TYPE_FLOAT */ + .long Lld_double-Lload_table /* FFI_TYPE_DOUBLE */ + .long Lld_ldouble-Lload_table /* FFI_TYPE_LONGDOUBLE */ + .long Lld_int8-Lload_table /* FFI_TYPE_UINT8 */ + .long Lld_int8-Lload_table /* FFI_TYPE_SINT8 */ + .long Lld_int16-Lload_table /* FFI_TYPE_UINT16 */ + .long Lld_int16-Lload_table /* FFI_TYPE_SINT16 */ + .long Lld_int32-Lload_table /* FFI_TYPE_UINT32 */ + .long Lld_int32-Lload_table /* FFI_TYPE_SINT32 */ + .long Lld_int64-Lload_table /* FFI_TYPE_UINT64 */ + .long Lld_int64-Lload_table /* FFI_TYPE_SINT64 */ + .long Lld_struct-Lload_table /* FFI_TYPE_STRUCT */ + .long Lld_int64-Lload_table /* FFI_TYPE_POINTER */ + + .text + .align 3 +Lld_void: + ret + .align 3 +Lld_int8: + movzbl -24(%rsp), %eax + ret + .align 3 +Lld_int16: + movzwl -24(%rsp), %eax + ret + .align 3 +Lld_int32: + movl -24(%rsp), %eax + ret + .align 3 +Lld_int64: + movq -24(%rsp), %rax + ret + .align 3 +Lld_float: + movss -24(%rsp), %xmm0 + ret + .align 3 +Lld_double: + movsd -24(%rsp), %xmm0 + ret + .align 3 +Lld_ldouble: + fldt -24(%rsp) + ret + .align 3 +Lld_struct: + /* There are four possibilities here, %rax/%rdx, %xmm0/%rax, + %rax/%xmm0, %xmm0/%xmm1. We collapse two by always loading + both rdx and xmm1 with the second word. For the remaining, + bit 8 set means xmm0 gets the second word, and bit 9 means + that rax gets the second word. */ + movq -24(%rsp), %rcx + movq -16(%rsp), %rdx + movq -16(%rsp), %xmm1 + testl $0x100, %eax + cmovnz %rdx, %rcx + movd %rcx, %xmm0 + testl $0x200, %eax + movq -24(%rsp), %rax + cmovnz %rdx, %rax + ret + + /* See the comment above Lload_sse; the same logic applies here. */ + .align 3 +LUW8: +Lsave_sse: + movdqa %xmm0, 48(%rsp) + movdqa %xmm1, 64(%rsp) + movdqa %xmm2, 80(%rsp) + movdqa %xmm3, 96(%rsp) + movdqa %xmm4, 112(%rsp) + movdqa %xmm5, 128(%rsp) + movdqa %xmm6, 144(%rsp) + movdqa %xmm7, 160(%rsp) + jmp Lret_from_save_sse + +LUW9: +.section __TEXT,__eh_frame,coalesced,no_toc+strip_static_syms+live_support +EH_frame1: + .set L$set$0,LECIE1-LSCIE1 /* CIE Length */ + .long L$set$0 +LSCIE1: + .long 0x0 /* CIE Identifier Tag */ + .byte 0x1 /* CIE Version */ + .ascii "zR\0" /* CIE Augmentation */ + .byte 0x1 /* uleb128 0x1; CIE Code Alignment Factor */ + .byte 0x78 /* sleb128 -8; CIE Data Alignment Factor */ + .byte 0x10 /* CIE RA Column */ + .byte 0x1 /* uleb128 0x1; Augmentation size */ + .byte 0x10 /* FDE Encoding (pcrel sdata4) */ + .byte 0xc /* DW_CFA_def_cfa, %rsp offset 8 */ + .byte 0x7 /* uleb128 0x7 */ + .byte 0x8 /* uleb128 0x8 */ + .byte 0x90 /* DW_CFA_offset, column 0x10 */ + .byte 0x1 + .align 3 +LECIE1: + .globl _ffi_call_unix64.eh +_ffi_call_unix64.eh: +LSFDE1: + .set L$set$1,LEFDE1-LASFDE1 /* FDE Length */ + .long L$set$1 +LASFDE1: + .long LASFDE1-EH_frame1 /* FDE CIE offset */ + .quad LUW0-. /* FDE initial location */ + .set L$set$2,LUW4-LUW0 /* FDE address range */ + .quad L$set$2 + .byte 0x0 /* Augmentation size */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$3,LUW1-LUW0 + .long L$set$3 + + /* New stack frame based off rbp. This is a itty bit of unwind + trickery in that the CFA *has* changed. There is no easy way + to describe it correctly on entry to the function. Fortunately, + it doesn't matter too much since at all points we can correctly + unwind back to ffi_call. Note that the location to which we + moved the return address is (the new) CFA-8, so from the + perspective of the unwind info, it hasn't moved. */ + .byte 0xc /* DW_CFA_def_cfa, %rbp offset 32 */ + .byte 0x6 + .byte 0x20 + .byte 0x80+6 /* DW_CFA_offset, %rbp offset 2*-8 */ + .byte 0x2 + .byte 0xa /* DW_CFA_remember_state */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$4,LUW2-LUW1 + .long L$set$4 + .byte 0xc /* DW_CFA_def_cfa, %rsp offset 8 */ + .byte 0x7 + .byte 0x8 + .byte 0xc0+6 /* DW_CFA_restore, %rbp */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$5,LUW3-LUW2 + .long L$set$5 + .byte 0xb /* DW_CFA_restore_state */ + + .align 3 +LEFDE1: + .globl _ffi_closure_unix64.eh +_ffi_closure_unix64.eh: +LSFDE3: + .set L$set$6,LEFDE3-LASFDE3 /* FDE Length */ + .long L$set$6 +LASFDE3: + .long LASFDE3-EH_frame1 /* FDE CIE offset */ + .quad LUW5-. /* FDE initial location */ + .set L$set$7,LUW9-LUW5 /* FDE address range */ + .quad L$set$7 + .byte 0x0 /* Augmentation size */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$8,LUW6-LUW5 + .long L$set$8 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 208,1 /* uleb128 208 */ + .byte 0xa /* DW_CFA_remember_state */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$9,LUW7-LUW6 + .long L$set$9 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 0x8 + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$10,LUW8-LUW7 + .long L$set$10 + .byte 0xb /* DW_CFA_restore_state */ + + .align 3 +LEFDE3: + .subsections_via_symbols + +#endif /* __x86_64__ */ diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi.c b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi.c new file mode 100644 index 0000000..4c68dd5 --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi.c @@ -0,0 +1,935 @@ +/* ----------------------------------------------------------------------- + ffi.c - Copyright (c) 1996, 1998, 1999, 2001, 2007, 2008 Red Hat, Inc. + Copyright (c) 2002 Ranjit Mathew + Copyright (c) 2002 Bo Thorsen + Copyright (c) 2002 Roger Sayle + Copyright (C) 2008, 2010 Free Software Foundation, Inc. + + x86 Foreign Function Interface + + Permission is hereby granted, free of charge, to any person obtaining + a copy of this software and associated documentation files (the + ``Software''), to deal in the Software without restriction, including + without limitation the rights to use, copy, modify, merge, publish, + distribute, sublicense, and/or sell copies of the Software, and to + permit persons to whom the Software is furnished to do so, subject to + the following conditions: + + The above copyright notice and this permission notice shall be included + in all copies or substantial portions of the Software. + + THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, + EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + DEALINGS IN THE SOFTWARE. + ----------------------------------------------------------------------- */ + +#if !defined(__x86_64__) || defined(_WIN64) || defined(__CYGWIN__) + +#ifdef _WIN64 +#include +#endif + +#include +#include + +#include + + +/* ffi_prep_args is called by the assembly routine once stack space + has been allocated for the function's arguments */ + +unsigned int ffi_prep_args(char *stack, extended_cif *ecif); +unsigned int ffi_prep_args(char *stack, extended_cif *ecif) +{ + register unsigned int i; + register void **p_argv; + register char *argp; + register ffi_type **p_arg; +#ifndef X86_WIN64 + const int cabi = ecif->cif->abi; + const int dir = (cabi == FFI_PASCAL || cabi == FFI_REGISTER) ? -1 : +1; + unsigned int stack_args_count = 0; + void *p_stack_data[3]; + char *argp2 = stack; +#else + #define dir 1 +#endif + + argp = stack; + + if ((ecif->cif->flags == FFI_TYPE_STRUCT + || ecif->cif->flags == FFI_TYPE_MS_STRUCT) +#ifdef X86_WIN64 + && ((ecif->cif->rtype->size & (1 | 2 | 4 | 8)) == 0) +#endif + ) + { +#ifndef X86_WIN64 + /* For fastcall/thiscall/register this is first register-passed + argument. */ + if (cabi == FFI_THISCALL || cabi == FFI_FASTCALL || cabi == FFI_REGISTER) + { + p_stack_data[stack_args_count] = argp; + ++stack_args_count; + } +#endif + + *(void **) argp = ecif->rvalue; + argp += sizeof(void*); + } + + p_arg = ecif->cif->arg_types; + p_argv = ecif->avalue; + if (dir < 0) + { + const int nargs = ecif->cif->nargs - 1; + if (nargs > 0) + { + p_arg += nargs; + p_argv += nargs; + } + } + + for (i = ecif->cif->nargs; + i != 0; + i--, p_arg += dir, p_argv += dir) + { + size_t z; + + /* Align if necessary */ + if ((sizeof(void*) - 1) & (size_t) argp) + argp = (char *) ALIGN(argp, sizeof(void*)); + + z = (*p_arg)->size; + +#ifdef X86_WIN64 + if (z > FFI_SIZEOF_ARG + || ((*p_arg)->type == FFI_TYPE_STRUCT + && (z & (1 | 2 | 4 | 8)) == 0) +#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE + || ((*p_arg)->type == FFI_TYPE_LONGDOUBLE) +#endif + ) + { + z = FFI_SIZEOF_ARG; + *(void **)argp = *p_argv; + } + else if ((*p_arg)->type == FFI_TYPE_FLOAT) + { + memcpy(argp, *p_argv, z); + } + else +#endif + if (z < FFI_SIZEOF_ARG) + { + z = FFI_SIZEOF_ARG; + switch ((*p_arg)->type) + { + case FFI_TYPE_SINT8: + *(ffi_sarg *) argp = (ffi_sarg)*(SINT8 *)(* p_argv); + break; + + case FFI_TYPE_UINT8: + *(ffi_arg *) argp = (ffi_arg)*(UINT8 *)(* p_argv); + break; + + case FFI_TYPE_SINT16: + *(ffi_sarg *) argp = (ffi_sarg)*(SINT16 *)(* p_argv); + break; + + case FFI_TYPE_UINT16: + *(ffi_arg *) argp = (ffi_arg)*(UINT16 *)(* p_argv); + break; + + case FFI_TYPE_SINT32: + *(ffi_sarg *) argp = (ffi_sarg)*(SINT32 *)(* p_argv); + break; + + case FFI_TYPE_UINT32: + *(ffi_arg *) argp = (ffi_arg)*(UINT32 *)(* p_argv); + break; + + case FFI_TYPE_STRUCT: + *(ffi_arg *) argp = *(ffi_arg *)(* p_argv); + break; + + default: + FFI_ASSERT(0); + } + } + else + { + memcpy(argp, *p_argv, z); + } + +#ifndef X86_WIN64 + /* For thiscall/fastcall/register convention register-passed arguments + are the first two none-floating-point arguments with a size + smaller or equal to sizeof (void*). */ + if ((z == FFI_SIZEOF_ARG) + && ((cabi == FFI_REGISTER) + || (cabi == FFI_THISCALL && stack_args_count < 1) + || (cabi == FFI_FASTCALL && stack_args_count < 2)) + && ((*p_arg)->type != FFI_TYPE_FLOAT && (*p_arg)->type != FFI_TYPE_STRUCT) + ) + { + if (dir < 0 && stack_args_count > 2) + { + /* Iterating arguments backwards, so first register-passed argument + will be passed last. Shift temporary values to make place. */ + p_stack_data[0] = p_stack_data[1]; + p_stack_data[1] = p_stack_data[2]; + stack_args_count = 2; + } + + p_stack_data[stack_args_count] = argp; + ++stack_args_count; + } +#endif + +#ifdef X86_WIN64 + argp += (z + sizeof(void*) - 1) & ~(sizeof(void*) - 1); +#else + argp += z; +#endif + } + +#ifndef X86_WIN64 + /* We need to move the register-passed arguments for thiscall/fastcall/register + on top of stack, so that those can be moved to registers by call-handler. */ + if (stack_args_count > 0) + { + if (dir < 0 && stack_args_count > 1) + { + /* Reverse order if iterating arguments backwards */ + ffi_arg tmp = *(ffi_arg*) p_stack_data[0]; + *(ffi_arg*) p_stack_data[0] = *(ffi_arg*) p_stack_data[stack_args_count - 1]; + *(ffi_arg*) p_stack_data[stack_args_count - 1] = tmp; + } + + int i; + for (i = 0; i < stack_args_count; i++) + { + if (p_stack_data[i] != argp2) + { + ffi_arg tmp = *(ffi_arg*) p_stack_data[i]; + memmove (argp2 + FFI_SIZEOF_ARG, argp2, (size_t) ((char*) p_stack_data[i] - (char*)argp2)); + *(ffi_arg *) argp2 = tmp; + } + + argp2 += FFI_SIZEOF_ARG; + } + } + + return stack_args_count; +#endif + return 0; +} + +/* Perform machine dependent cif processing */ +ffi_status ffi_prep_cif_machdep(ffi_cif *cif) +{ + unsigned int i; + ffi_type **ptr; + + /* Set the return type flag */ + switch (cif->rtype->type) + { + case FFI_TYPE_VOID: + case FFI_TYPE_UINT8: + case FFI_TYPE_UINT16: + case FFI_TYPE_SINT8: + case FFI_TYPE_SINT16: +#ifdef X86_WIN64 + case FFI_TYPE_UINT32: + case FFI_TYPE_SINT32: +#endif + case FFI_TYPE_SINT64: + case FFI_TYPE_FLOAT: + case FFI_TYPE_DOUBLE: +#ifndef X86_WIN64 +#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE + case FFI_TYPE_LONGDOUBLE: +#endif +#endif + cif->flags = (unsigned) cif->rtype->type; + break; + + case FFI_TYPE_UINT64: +#ifdef X86_WIN64 + case FFI_TYPE_POINTER: +#endif + cif->flags = FFI_TYPE_SINT64; + break; + + case FFI_TYPE_STRUCT: +#ifndef X86 + if (cif->rtype->size == 1) + { + cif->flags = FFI_TYPE_SMALL_STRUCT_1B; /* same as char size */ + } + else if (cif->rtype->size == 2) + { + cif->flags = FFI_TYPE_SMALL_STRUCT_2B; /* same as short size */ + } + else if (cif->rtype->size == 4) + { +#ifdef X86_WIN64 + cif->flags = FFI_TYPE_SMALL_STRUCT_4B; +#else + cif->flags = FFI_TYPE_INT; /* same as int type */ +#endif + } + else if (cif->rtype->size == 8) + { + cif->flags = FFI_TYPE_SINT64; /* same as int64 type */ + } + else +#endif + { +#ifdef X86_WIN32 + if (cif->abi == FFI_MS_CDECL) + cif->flags = FFI_TYPE_MS_STRUCT; + else +#endif + cif->flags = FFI_TYPE_STRUCT; + /* allocate space for return value pointer */ + cif->bytes += ALIGN(sizeof(void*), FFI_SIZEOF_ARG); + } + break; + + default: +#ifdef X86_WIN64 + cif->flags = FFI_TYPE_SINT64; + break; + case FFI_TYPE_INT: + cif->flags = FFI_TYPE_SINT32; +#else + cif->flags = FFI_TYPE_INT; +#endif + break; + } + + for (ptr = cif->arg_types, i = cif->nargs; i > 0; i--, ptr++) + { + if (((*ptr)->alignment - 1) & cif->bytes) + cif->bytes = ALIGN(cif->bytes, (*ptr)->alignment); + cif->bytes += (unsigned)ALIGN((*ptr)->size, FFI_SIZEOF_ARG); + } + +#ifdef X86_WIN64 + /* ensure space for storing four registers */ + cif->bytes += 4 * FFI_SIZEOF_ARG; +#endif + +#ifndef X86_WIN32 +#ifndef X86_WIN64 + if (cif->abi == FFI_SYSV || cif->abi == FFI_UNIX64) +#endif + cif->bytes = (cif->bytes + 15) & ~0xF; +#endif + + return FFI_OK; +} + +#ifdef X86_WIN64 +extern int +ffi_call_win64(unsigned int (*)(char *, extended_cif *), extended_cif *, + unsigned, unsigned, unsigned *, void (*fn)(void)); +#else +extern void +ffi_call_win32(unsigned int (*)(char *, extended_cif *), extended_cif *, + unsigned, unsigned, unsigned, unsigned *, void (*fn)(void)); +extern void ffi_call_SYSV(void (*)(char *, extended_cif *), extended_cif *, + unsigned, unsigned, unsigned *, void (*fn)(void)); +#endif + +void ffi_call(ffi_cif *cif, void (*fn)(void), void *rvalue, void **avalue) +{ + extended_cif ecif; + + ecif.cif = cif; + ecif.avalue = avalue; + + /* If the return value is a struct and we don't have a return */ + /* value address then we need to make one */ + +#ifdef X86_WIN64 + if (rvalue == NULL + && cif->flags == FFI_TYPE_STRUCT + && ((cif->rtype->size & (1 | 2 | 4 | 8)) == 0)) + { + ecif.rvalue = alloca((cif->rtype->size + 0xF) & ~0xF); + } +#else + if (rvalue == NULL + && (cif->flags == FFI_TYPE_STRUCT + || cif->flags == FFI_TYPE_MS_STRUCT)) + { + ecif.rvalue = alloca(cif->rtype->size); + } +#endif + else + ecif.rvalue = rvalue; + + + switch (cif->abi) + { +#ifdef X86_WIN64 + case FFI_WIN64: + ffi_call_win64(ffi_prep_args, &ecif, cif->bytes, + cif->flags, ecif.rvalue, fn); + break; +#else +#ifndef X86_WIN32 + case FFI_SYSV: + ffi_call_SYSV(ffi_prep_args, &ecif, cif->bytes, cif->flags, ecif.rvalue, + fn); + break; +#else + case FFI_SYSV: + case FFI_MS_CDECL: +#endif + case FFI_STDCALL: + case FFI_THISCALL: + case FFI_FASTCALL: + case FFI_PASCAL: + case FFI_REGISTER: + ffi_call_win32(ffi_prep_args, &ecif, cif->abi, cif->bytes, cif->flags, + ecif.rvalue, fn); + break; +#endif + default: + FFI_ASSERT(0); + break; + } +} + + +/** private members **/ + +/* The following __attribute__((regparm(1))) decorations will have no effect + on MSVC or SUNPRO_C -- standard conventions apply. */ +static unsigned int ffi_prep_incoming_args (char *stack, void **ret, + void** args, ffi_cif* cif); +void FFI_HIDDEN ffi_closure_SYSV (ffi_closure *) + __attribute__ ((regparm(1))); +unsigned int FFI_HIDDEN ffi_closure_SYSV_inner (ffi_closure *, void **, void *) + __attribute__ ((regparm(1))); +unsigned int FFI_HIDDEN ffi_closure_WIN32_inner (ffi_closure *, void **, void *) + __attribute__ ((regparm(1))); +void FFI_HIDDEN ffi_closure_raw_SYSV (ffi_raw_closure *) + __attribute__ ((regparm(1))); +#ifdef X86_WIN32 +void FFI_HIDDEN ffi_closure_raw_THISCALL (ffi_raw_closure *) + __attribute__ ((regparm(1))); +#endif +#ifndef X86_WIN64 +void FFI_HIDDEN ffi_closure_STDCALL (ffi_closure *); +void FFI_HIDDEN ffi_closure_THISCALL (ffi_closure *); +void FFI_HIDDEN ffi_closure_FASTCALL (ffi_closure *); +void FFI_HIDDEN ffi_closure_REGISTER (ffi_closure *); +#else +void FFI_HIDDEN ffi_closure_win64 (ffi_closure *); +#endif + +/* This function is jumped to by the trampoline */ + +#ifdef X86_WIN64 +void * FFI_HIDDEN +ffi_closure_win64_inner (ffi_closure *closure, void *args) { + ffi_cif *cif; + void **arg_area; + void *result; + void *resp = &result; + + cif = closure->cif; + arg_area = (void**) alloca (cif->nargs * sizeof (void*)); + + /* this call will initialize ARG_AREA, such that each + * element in that array points to the corresponding + * value on the stack; and if the function returns + * a structure, it will change RESP to point to the + * structure return address. */ + + ffi_prep_incoming_args(args, &resp, arg_area, cif); + + (closure->fun) (cif, resp, arg_area, closure->user_data); + + /* The result is returned in rax. This does the right thing for + result types except for floats; we have to 'mov xmm0, rax' in the + caller to correct this. + TODO: structure sizes of 3 5 6 7 are returned by reference, too!!! + */ + return cif->rtype->size > sizeof(void *) ? resp : *(void **)resp; +} + +#else +unsigned int FFI_HIDDEN __attribute__ ((regparm(1))) +ffi_closure_SYSV_inner (ffi_closure *closure, void **respp, void *args) +{ + /* our various things... */ + ffi_cif *cif; + void **arg_area; + + cif = closure->cif; + arg_area = (void**) alloca (cif->nargs * sizeof (void*)); + + /* this call will initialize ARG_AREA, such that each + * element in that array points to the corresponding + * value on the stack; and if the function returns + * a structure, it will change RESP to point to the + * structure return address. */ + + ffi_prep_incoming_args(args, respp, arg_area, cif); + + (closure->fun) (cif, *respp, arg_area, closure->user_data); + + return cif->flags; +} + +unsigned int FFI_HIDDEN __attribute__ ((regparm(1))) +ffi_closure_WIN32_inner (ffi_closure *closure, void **respp, void *args) +{ + /* our various things... */ + ffi_cif *cif; + void **arg_area; + unsigned int ret; + + cif = closure->cif; + arg_area = (void**) alloca (cif->nargs * sizeof (void*)); + + /* this call will initialize ARG_AREA, such that each + * element in that array points to the corresponding + * value on the stack; and if the function returns + * a structure, it will change RESP to point to the + * structure return address. */ + + ret = ffi_prep_incoming_args(args, respp, arg_area, cif); + + (closure->fun) (cif, *respp, arg_area, closure->user_data); + + return ret; +} +#endif /* !X86_WIN64 */ + +static unsigned int +ffi_prep_incoming_args(char *stack, void **rvalue, void **avalue, + ffi_cif *cif) +{ + register unsigned int i; + register void **p_argv; + register char *argp; + register ffi_type **p_arg; +#ifndef X86_WIN64 + const int cabi = cif->abi; + const int dir = (cabi == FFI_PASCAL || cabi == FFI_REGISTER) ? -1 : +1; + const unsigned int max_stack_count = (cabi == FFI_THISCALL) ? 1 + : (cabi == FFI_FASTCALL) ? 2 + : (cabi == FFI_REGISTER) ? 3 + : 0; + unsigned int passed_regs = 0; + void *p_stack_data[3] = { stack - 1 }; +#else + #define dir 1 +#endif + + argp = stack; +#ifndef X86_WIN64 + argp += max_stack_count * FFI_SIZEOF_ARG; +#endif + + if ((cif->flags == FFI_TYPE_STRUCT + || cif->flags == FFI_TYPE_MS_STRUCT) +#ifdef X86_WIN64 + && ((cif->rtype->size & (1 | 2 | 4 | 8)) == 0) +#endif + ) + { +#ifndef X86_WIN64 + if (passed_regs < max_stack_count) + { + *rvalue = *(void**) (stack + (passed_regs*FFI_SIZEOF_ARG)); + ++passed_regs; + } + else +#endif + { + *rvalue = *(void **) argp; + argp += sizeof(void *); + } + } + +#ifndef X86_WIN64 + /* Do register arguments first */ + for (i = 0, p_arg = cif->arg_types; + i < cif->nargs && passed_regs < max_stack_count; + i++, p_arg++) + { + if ((*p_arg)->type == FFI_TYPE_FLOAT + || (*p_arg)->type == FFI_TYPE_STRUCT) + continue; + + size_t sz = (*p_arg)->size; + if(sz == 0 || sz > FFI_SIZEOF_ARG) + continue; + + p_stack_data[passed_regs] = avalue + i; + avalue[i] = stack + (passed_regs*FFI_SIZEOF_ARG); + ++passed_regs; + } +#endif + + p_arg = cif->arg_types; + p_argv = avalue; + if (dir < 0) + { + const int nargs = cif->nargs - 1; + if (nargs > 0) + { + p_arg += nargs; + p_argv += nargs; + } + } + + for (i = cif->nargs; + i != 0; + i--, p_arg += dir, p_argv += dir) + { + size_t z; + + /* Align if necessary */ + if ((sizeof(void*) - 1) & (size_t) argp) + argp = (char *) ALIGN(argp, sizeof(void*)); + + z = (*p_arg)->size; + +#ifdef X86_WIN64 + if (z > FFI_SIZEOF_ARG + || ((*p_arg)->type == FFI_TYPE_STRUCT + && (z & (1 | 2 | 4 | 8)) == 0) +#if FFI_TYPE_DOUBLE != FFI_TYPE_LONGDOUBLE + || ((*p_arg)->type == FFI_TYPE_LONGDOUBLE) +#endif + ) + { + z = FFI_SIZEOF_ARG; + *p_argv = *(void **)argp; + } + else +#else + if (passed_regs > 0 + && z <= FFI_SIZEOF_ARG + && (p_argv == p_stack_data[0] + || p_argv == p_stack_data[1] + || p_argv == p_stack_data[2])) + { + /* Already assigned a register value */ + continue; + } + else +#endif + { + /* because we're little endian, this is what it turns into. */ + *p_argv = (void*) argp; + } + +#ifdef X86_WIN64 + argp += (z + sizeof(void*) - 1) & ~(sizeof(void*) - 1); +#else + argp += z; +#endif + } + + return (int)((size_t)argp - (size_t)stack); +} + +#define FFI_INIT_TRAMPOLINE_WIN64(TRAMP,FUN,CTX,MASK) \ +{ unsigned char *__tramp = (unsigned char*)(TRAMP); \ + void* __fun = (void*)(FUN); \ + void* __ctx = (void*)(CTX); \ + *(unsigned char*) &__tramp[0] = 0x41; \ + *(unsigned char*) &__tramp[1] = 0xbb; \ + *(unsigned int*) &__tramp[2] = MASK; /* mov $mask, %r11 */ \ + *(unsigned char*) &__tramp[6] = 0x48; \ + *(unsigned char*) &__tramp[7] = 0xb8; \ + *(void**) &__tramp[8] = __ctx; /* mov __ctx, %rax */ \ + *(unsigned char *) &__tramp[16] = 0x49; \ + *(unsigned char *) &__tramp[17] = 0xba; \ + *(void**) &__tramp[18] = __fun; /* mov __fun, %r10 */ \ + *(unsigned char *) &__tramp[26] = 0x41; \ + *(unsigned char *) &__tramp[27] = 0xff; \ + *(unsigned char *) &__tramp[28] = 0xe2; /* jmp %r10 */ \ + } + +/* How to make a trampoline. Derived from gcc/config/i386/i386.c. */ + +#define FFI_INIT_TRAMPOLINE(TRAMP,FUN,CTX) \ +{ unsigned char *__tramp = (unsigned char*)(TRAMP); \ + unsigned int __fun = (unsigned int)(FUN); \ + unsigned int __ctx = (unsigned int)(CTX); \ + unsigned int __dis = __fun - (__ctx + 10); \ + *(unsigned char*) &__tramp[0] = 0xb8; \ + *(unsigned int*) &__tramp[1] = __ctx; /* movl __ctx, %eax */ \ + *(unsigned char*) &__tramp[5] = 0xe9; \ + *(unsigned int*) &__tramp[6] = __dis; /* jmp __fun */ \ + } + +#define FFI_INIT_TRAMPOLINE_RAW_THISCALL(TRAMP,FUN,CTX,SIZE) \ +{ unsigned char *__tramp = (unsigned char*)(TRAMP); \ + unsigned int __fun = (unsigned int)(FUN); \ + unsigned int __ctx = (unsigned int)(CTX); \ + unsigned int __dis = __fun - (__ctx + 49); \ + unsigned short __size = (unsigned short)(SIZE); \ + *(unsigned int *) &__tramp[0] = 0x8324048b; /* mov (%esp), %eax */ \ + *(unsigned int *) &__tramp[4] = 0x4c890cec; /* sub $12, %esp */ \ + *(unsigned int *) &__tramp[8] = 0x04890424; /* mov %ecx, 4(%esp) */ \ + *(unsigned char*) &__tramp[12] = 0x24; /* mov %eax, (%esp) */ \ + *(unsigned char*) &__tramp[13] = 0xb8; \ + *(unsigned int *) &__tramp[14] = __size; /* mov __size, %eax */ \ + *(unsigned int *) &__tramp[18] = 0x08244c8d; /* lea 8(%esp), %ecx */ \ + *(unsigned int *) &__tramp[22] = 0x4802e8c1; /* shr $2, %eax ; dec %eax */ \ + *(unsigned short*) &__tramp[26] = 0x0b74; /* jz 1f */ \ + *(unsigned int *) &__tramp[28] = 0x8908518b; /* 2b: mov 8(%ecx), %edx */ \ + *(unsigned int *) &__tramp[32] = 0x04c18311; /* mov %edx, (%ecx) ; add $4, %ecx */ \ + *(unsigned char*) &__tramp[36] = 0x48; /* dec %eax */ \ + *(unsigned short*) &__tramp[37] = 0xf575; /* jnz 2b ; 1f: */ \ + *(unsigned char*) &__tramp[39] = 0xb8; \ + *(unsigned int*) &__tramp[40] = __ctx; /* movl __ctx, %eax */ \ + *(unsigned char *) &__tramp[44] = 0xe8; \ + *(unsigned int*) &__tramp[45] = __dis; /* call __fun */ \ + *(unsigned char*) &__tramp[49] = 0xc2; /* ret */ \ + *(unsigned short*) &__tramp[50] = (__size + 8); /* ret (__size + 8) */ \ + } + +#define FFI_INIT_TRAMPOLINE_WIN32(TRAMP,FUN,CTX) \ +{ unsigned char *__tramp = (unsigned char*)(TRAMP); \ + unsigned int __fun = (unsigned int)(FUN); \ + unsigned int __ctx = (unsigned int)(CTX); \ + unsigned int __dis = __fun - (__ctx + 10); \ + *(unsigned char*) &__tramp[0] = 0x68; \ + *(unsigned int*) &__tramp[1] = __ctx; /* push __ctx */ \ + *(unsigned char*) &__tramp[5] = 0xe9; \ + *(unsigned int*) &__tramp[6] = __dis; /* jmp __fun */ \ + } + +/* the cif must already be prep'ed */ + +ffi_status +ffi_prep_closure_loc (ffi_closure* closure, + ffi_cif* cif, + void (*fun)(ffi_cif*,void*,void**,void*), + void *user_data, + void *codeloc) +{ +#ifdef X86_WIN64 +#define ISFLOAT(IDX) (cif->arg_types[IDX]->type == FFI_TYPE_FLOAT || cif->arg_types[IDX]->type == FFI_TYPE_DOUBLE) +#define FLAG(IDX) (cif->nargs>(IDX)&&ISFLOAT(IDX)?(1<<(IDX)):0) + if (cif->abi == FFI_WIN64) + { + int mask = FLAG(0)|FLAG(1)|FLAG(2)|FLAG(3); + FFI_INIT_TRAMPOLINE_WIN64 (&closure->tramp[0], + &ffi_closure_win64, + codeloc, mask); + /* make sure we can execute here */ + } +#else + if (cif->abi == FFI_SYSV) + { + FFI_INIT_TRAMPOLINE (&closure->tramp[0], + &ffi_closure_SYSV, + (void*)codeloc); + } + else if (cif->abi == FFI_REGISTER) + { + FFI_INIT_TRAMPOLINE_WIN32 (&closure->tramp[0], + &ffi_closure_REGISTER, + (void*)codeloc); + } + else if (cif->abi == FFI_FASTCALL) + { + FFI_INIT_TRAMPOLINE_WIN32 (&closure->tramp[0], + &ffi_closure_FASTCALL, + (void*)codeloc); + } + else if (cif->abi == FFI_THISCALL) + { + FFI_INIT_TRAMPOLINE_WIN32 (&closure->tramp[0], + &ffi_closure_THISCALL, + (void*)codeloc); + } + else if (cif->abi == FFI_STDCALL || cif->abi == FFI_PASCAL) + { + FFI_INIT_TRAMPOLINE_WIN32 (&closure->tramp[0], + &ffi_closure_STDCALL, + (void*)codeloc); + } +#ifdef X86_WIN32 + else if (cif->abi == FFI_MS_CDECL) + { + FFI_INIT_TRAMPOLINE (&closure->tramp[0], + &ffi_closure_SYSV, + (void*)codeloc); + } +#endif /* X86_WIN32 */ +#endif /* !X86_WIN64 */ + else + { + return FFI_BAD_ABI; + } + + closure->cif = cif; + closure->user_data = user_data; + closure->fun = fun; + + return FFI_OK; +} + +/* ------- Native raw API support -------------------------------- */ + +#if !FFI_NO_RAW_API + +ffi_status +ffi_prep_raw_closure_loc (ffi_raw_closure* closure, + ffi_cif* cif, + void (*fun)(ffi_cif*,void*,ffi_raw*,void*), + void *user_data, + void *codeloc) +{ + int i; + + if (cif->abi != FFI_SYSV +#ifdef X86_WIN32 + && cif->abi != FFI_THISCALL +#endif + ) + return FFI_BAD_ABI; + + /* we currently don't support certain kinds of arguments for raw + closures. This should be implemented by a separate assembly + language routine, since it would require argument processing, + something we don't do now for performance. */ + + for (i = cif->nargs-1; i >= 0; i--) + { + FFI_ASSERT (cif->arg_types[i]->type != FFI_TYPE_STRUCT); + FFI_ASSERT (cif->arg_types[i]->type != FFI_TYPE_LONGDOUBLE); + } + +#ifdef X86_WIN32 + if (cif->abi == FFI_SYSV) + { +#endif + FFI_INIT_TRAMPOLINE (&closure->tramp[0], &ffi_closure_raw_SYSV, + codeloc); +#ifdef X86_WIN32 + } + else if (cif->abi == FFI_THISCALL) + { + FFI_INIT_TRAMPOLINE_RAW_THISCALL (&closure->tramp[0], &ffi_closure_raw_THISCALL, codeloc, cif->bytes); + } +#endif + closure->cif = cif; + closure->user_data = user_data; + closure->fun = fun; + + return FFI_OK; +} + +static unsigned int +ffi_prep_args_raw(char *stack, extended_cif *ecif) +{ + const ffi_cif *cif = ecif->cif; + unsigned int i, passed_regs = 0; + +#ifndef X86_WIN64 + const unsigned int abi = cif->abi; + const unsigned int max_regs = (abi == FFI_THISCALL) ? 1 + : (abi == FFI_FASTCALL) ? 2 + : (abi == FFI_REGISTER) ? 3 + : 0; + + if (cif->flags == FFI_TYPE_STRUCT) + ++passed_regs; + + for (i = 0; i < cif->nargs && passed_regs <= max_regs; i++) + { + if (cif->arg_types[i]->type == FFI_TYPE_FLOAT + || cif->arg_types[i]->type == FFI_TYPE_STRUCT) + continue; + + size_t sz = cif->arg_types[i]->size; + if (sz == 0 || sz > FFI_SIZEOF_ARG) + continue; + + ++passed_regs; + } +#endif + + memcpy (stack, ecif->avalue, cif->bytes); + return passed_regs; +} + +/* we borrow this routine from libffi (it must be changed, though, to + * actually call the function passed in the first argument. as of + * libffi-1.20, this is not the case.) + */ + +void +ffi_raw_call(ffi_cif *cif, void (*fn)(void), void *rvalue, ffi_raw *fake_avalue) +{ + extended_cif ecif; + void **avalue = (void **)fake_avalue; + + ecif.cif = cif; + ecif.avalue = avalue; + + /* If the return value is a struct and we don't have a return */ + /* value address then we need to make one */ + + if (rvalue == NULL + && (cif->flags == FFI_TYPE_STRUCT + || cif->flags == FFI_TYPE_MS_STRUCT)) + { + ecif.rvalue = alloca(cif->rtype->size); + } + else + ecif.rvalue = rvalue; + + + switch (cif->abi) + { +#ifndef X86_WIN32 + case FFI_SYSV: + ffi_call_SYSV(ffi_prep_args_raw, &ecif, cif->bytes, cif->flags, + ecif.rvalue, fn); + break; +#else + case FFI_SYSV: + case FFI_MS_CDECL: +#endif +#ifndef X86_WIN64 + case FFI_STDCALL: + case FFI_THISCALL: + case FFI_FASTCALL: + case FFI_PASCAL: + case FFI_REGISTER: + ffi_call_win32(ffi_prep_args_raw, &ecif, cif->abi, cif->bytes, cif->flags, + ecif.rvalue, fn); + break; +#endif + default: + FFI_ASSERT(0); + break; + } +} + +#endif + +#endif /* !__x86_64__ || X86_WIN64 */ + diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi64.c b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi64.c new file mode 100644 index 0000000..5a5e043 --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffi64.c @@ -0,0 +1,676 @@ +/* ----------------------------------------------------------------------- + ffi64.c - Copyright (c) 2013 The Written Word, Inc. + Copyright (c) 2011 Anthony Green + Copyright (c) 2008, 2010 Red Hat, Inc. + Copyright (c) 2002, 2007 Bo Thorsen + + 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 +#include + +#include +#include + +#ifdef __x86_64__ + +#define MAX_GPR_REGS 6 +#define MAX_SSE_REGS 8 + +#if defined(__INTEL_COMPILER) +#include "xmmintrin.h" +#define UINT128 __m128 +#else +#if defined(__SUNPRO_C) +#include +#define UINT128 __m128i +#else +#define UINT128 __int128_t +#endif +#endif + +union big_int_union +{ + UINT32 i32; + UINT64 i64; + UINT128 i128; +}; + +struct register_args +{ + /* Registers for argument passing. */ + UINT64 gpr[MAX_GPR_REGS]; + union big_int_union 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 size_t +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: + { + size_t 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; +#if FFI_TYPE_LONGDOUBLE != FFI_TYPE_DOUBLE + case FFI_TYPE_LONGDOUBLE: + classes[0] = X86_64_X87_CLASS; + classes[1] = X86_64_X87UP_CLASS; + return 2; +#endif + case FFI_TYPE_STRUCT: + { + const size_t UNITS_PER_WORD = 8; + size_t 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++) + { + size_t 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++) + { + size_t 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 size_t +examine_argument (ffi_type *type, enum x86_64_reg_class classes[MAX_CLASSES], + _Bool in_return, int *pngpr, int *pnsse) +{ + size_t n; + int i, 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, ngpr, nsse, flags; + enum x86_64_reg_class classes[MAX_CLASSES]; + size_t bytes, n; + + 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 = (unsigned)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++] = (unsigned long) rvalue; + + avn = cif->nargs; + arg_types = cif->arg_types; + + for (i = 0; i < avn; ++i) + { + size_t n, size = arg_types[i]->size; + + 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: + /* Sign-extend integer arguments passed in general + purpose registers, to cope with the fact that + LLVM incorrectly assumes that this will be done + (the x86-64 PS ABI does not specify this). */ + switch (arg_types[i]->type) + { + case FFI_TYPE_SINT8: + *(SINT64 *)®_args->gpr[gprcount] = (SINT64) *((SINT8 *) a); + break; + case FFI_TYPE_SINT16: + *(SINT64 *)®_args->gpr[gprcount] = (SINT64) *((SINT16 *) a); + break; + case FFI_TYPE_SINT32: + *(SINT64 *)®_args->gpr[gprcount] = (SINT64) *((SINT32 *) a); + break; + default: + reg_args->gpr[gprcount] = 0; + memcpy (®_args->gpr[gprcount], a, size < 8 ? size : 8); + } + gprcount++; + break; + case X86_64_SSE_CLASS: + case X86_64_SSEDF_CLASS: + reg_args->sse[ssecount++].i64 = *(UINT64 *) a; + break; + case X86_64_SSESF_CLASS: + reg_args->sse[ssecount++].i32 = *(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 , %r11 */ + *((unsigned long long * volatile) &tramp[1]) + = (unsigned long) ffi_closure_unix64; + tramp[5] = 0xba49; /* mov , %r10 */ + *((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. */ + 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]; + size_t 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 *) (unsigned long) 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]; + size_t 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] = ®_args->sse[ssecount]; + ssecount += n; + } + else + { + avalue[i] = ®_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, ®_args->sse[ssecount++], 8); + else + memcpy (a, ®_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__ */ diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffitarget.h b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffitarget.h new file mode 100644 index 0000000..ff0f718 --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/ffitarget.h @@ -0,0 +1,152 @@ +/* -----------------------------------------------------------------*-C-*- + ffitarget.h - Copyright (c) 2012, 2014 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 + +#define FFI_TARGET_SPECIFIC_STACK_SPACE_ALLOCATION +#ifndef _MSC_VER +#define FFI_TARGET_HAS_COMPLEX_TYPE +#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 +#if defined __x86_64__ && defined __ILP32__ +#define FFI_SIZEOF_ARG 8 +#define FFI_SIZEOF_JAVA_RAW 4 +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, + + /* ---- Intel x86 Win32 ---------- */ +#ifdef X86_WIN32 + FFI_SYSV, + FFI_STDCALL, + FFI_THISCALL, + FFI_FASTCALL, + FFI_MS_CDECL, + FFI_PASCAL, + FFI_REGISTER, + FFI_LAST_ABI, +#ifdef _MSC_VER + FFI_DEFAULT_ABI = FFI_MS_CDECL +#else + FFI_DEFAULT_ABI = FFI_SYSV +#endif + +#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_THISCALL, + FFI_FASTCALL, + FFI_STDCALL, + FFI_PASCAL, + FFI_REGISTER, + 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) +#define FFI_TYPE_MS_STRUCT (FFI_TYPE_LAST + 4) + +#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 + diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/freebsd.S b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/freebsd.S new file mode 100644 index 0000000..97e0b4e --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/freebsd.S @@ -0,0 +1,463 @@ +/* ----------------------------------------------------------------------- + freebsd.S - Copyright (c) 1996, 1998, 2001, 2002, 2003, 2005 Red Hat, Inc. + Copyright (c) 2008 Björn König + + X86 Foreign Function Interface for FreeBSD + + 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 __x86_64__ + +#define LIBFFI_ASM +#include +#include + +.text + +.globl ffi_prep_args + + .align 4 +.globl ffi_call_SYSV + .type ffi_call_SYSV,@function + +ffi_call_SYSV: +.LFB1: + pushl %ebp +.LCFI0: + movl %esp,%ebp +.LCFI1: + /* Make room for all of the new args. */ + movl 16(%ebp),%ecx + subl %ecx,%esp + + /* Align the stack pointer to 16-bytes */ + andl $0xfffffff0, %esp + + movl %esp,%eax + + /* Place all of the ffi_prep_args in position */ + pushl 12(%ebp) + pushl %eax + call *8(%ebp) + + /* Return stack to previous state and call the function */ + addl $8,%esp + + call *28(%ebp) + + /* Load %ecx with the return type code */ + movl 20(%ebp),%ecx + + /* Protect %esi. We're going to pop it in the epilogue. */ + pushl %esi + + /* If the return value pointer is NULL, assume no return value. */ + cmpl $0,24(%ebp) + jne 0f + + /* Even if there is no space for the return value, we are + obliged to handle floating-point values. */ + cmpl $FFI_TYPE_FLOAT,%ecx + jne noretval + fstp %st(0) + + jmp epilogue + +0: + call 1f + +.Lstore_table: + .long noretval-.Lstore_table /* FFI_TYPE_VOID */ + .long retint-.Lstore_table /* FFI_TYPE_INT */ + .long retfloat-.Lstore_table /* FFI_TYPE_FLOAT */ + .long retdouble-.Lstore_table /* FFI_TYPE_DOUBLE */ + .long retlongdouble-.Lstore_table /* FFI_TYPE_LONGDOUBLE */ + .long retuint8-.Lstore_table /* FFI_TYPE_UINT8 */ + .long retsint8-.Lstore_table /* FFI_TYPE_SINT8 */ + .long retuint16-.Lstore_table /* FFI_TYPE_UINT16 */ + .long retsint16-.Lstore_table /* FFI_TYPE_SINT16 */ + .long retint-.Lstore_table /* FFI_TYPE_UINT32 */ + .long retint-.Lstore_table /* FFI_TYPE_SINT32 */ + .long retint64-.Lstore_table /* FFI_TYPE_UINT64 */ + .long retint64-.Lstore_table /* FFI_TYPE_SINT64 */ + .long retstruct-.Lstore_table /* FFI_TYPE_STRUCT */ + .long retint-.Lstore_table /* FFI_TYPE_POINTER */ + .long retstruct1b-.Lstore_table /* FFI_TYPE_SMALL_STRUCT_1B */ + .long retstruct2b-.Lstore_table /* FFI_TYPE_SMALL_STRUCT_2B */ + +1: + pop %esi + add (%esi, %ecx, 4), %esi + jmp *%esi + + /* Sign/zero extend as appropriate. */ +retsint8: + movsbl %al, %eax + jmp retint + +retsint16: + movswl %ax, %eax + jmp retint + +retuint8: + movzbl %al, %eax + jmp retint + +retuint16: + movzwl %ax, %eax + jmp retint + +retfloat: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstps (%ecx) + jmp epilogue + +retdouble: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstpl (%ecx) + jmp epilogue + +retlongdouble: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstpt (%ecx) + jmp epilogue + +retint64: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + movl %edx,4(%ecx) + jmp epilogue + +retstruct1b: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movb %al,0(%ecx) + jmp epilogue + +retstruct2b: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movw %ax,0(%ecx) + jmp epilogue + +retint: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + +retstruct: + /* Nothing to do! */ + +noretval: +epilogue: + popl %esi + movl %ebp,%esp + popl %ebp + ret +.LFE1: +.ffi_call_SYSV_end: + .size ffi_call_SYSV,.ffi_call_SYSV_end-ffi_call_SYSV + + .align 4 +FFI_HIDDEN (ffi_closure_SYSV) +.globl ffi_closure_SYSV + .type ffi_closure_SYSV, @function + +ffi_closure_SYSV: +.LFB2: + pushl %ebp +.LCFI2: + movl %esp, %ebp +.LCFI3: + subl $40, %esp + leal -24(%ebp), %edx + movl %edx, -12(%ebp) /* resp */ + leal 8(%ebp), %edx + movl %edx, 4(%esp) /* args = __builtin_dwarf_cfa () */ + leal -12(%ebp), %edx + movl %edx, (%esp) /* &resp */ +#if defined HAVE_HIDDEN_VISIBILITY_ATTRIBUTE || !defined __PIC__ + call ffi_closure_SYSV_inner +#else + movl %ebx, 8(%esp) +.LCFI7: + call 1f +1: popl %ebx + addl $_GLOBAL_OFFSET_TABLE_+[.-1b], %ebx + call ffi_closure_SYSV_inner@PLT + movl 8(%esp), %ebx +#endif + movl -12(%ebp), %ecx + cmpl $FFI_TYPE_INT, %eax + je .Lcls_retint + + /* Handle FFI_TYPE_UINT8, FFI_TYPE_SINT8, FFI_TYPE_UINT16, + FFI_TYPE_SINT16, FFI_TYPE_UINT32, FFI_TYPE_SINT32. */ + cmpl $FFI_TYPE_UINT64, %eax + jge 0f + cmpl $FFI_TYPE_UINT8, %eax + jge .Lcls_retint + +0: cmpl $FFI_TYPE_FLOAT, %eax + je .Lcls_retfloat + cmpl $FFI_TYPE_DOUBLE, %eax + je .Lcls_retdouble + cmpl $FFI_TYPE_LONGDOUBLE, %eax + je .Lcls_retldouble + cmpl $FFI_TYPE_SINT64, %eax + je .Lcls_retllong + cmpl $FFI_TYPE_SMALL_STRUCT_1B, %eax + je .Lcls_retstruct1b + cmpl $FFI_TYPE_SMALL_STRUCT_2B, %eax + je .Lcls_retstruct2b + cmpl $FFI_TYPE_STRUCT, %eax + je .Lcls_retstruct +.Lcls_epilogue: + movl %ebp, %esp + popl %ebp + ret +.Lcls_retint: + movl (%ecx), %eax + jmp .Lcls_epilogue +.Lcls_retfloat: + flds (%ecx) + jmp .Lcls_epilogue +.Lcls_retdouble: + fldl (%ecx) + jmp .Lcls_epilogue +.Lcls_retldouble: + fldt (%ecx) + jmp .Lcls_epilogue +.Lcls_retllong: + movl (%ecx), %eax + movl 4(%ecx), %edx + jmp .Lcls_epilogue +.Lcls_retstruct1b: + movsbl (%ecx), %eax + jmp .Lcls_epilogue +.Lcls_retstruct2b: + movswl (%ecx), %eax + jmp .Lcls_epilogue +.Lcls_retstruct: + movl %ebp, %esp + popl %ebp + ret $4 +.LFE2: + .size ffi_closure_SYSV, .-ffi_closure_SYSV + +#if !FFI_NO_RAW_API + +#define RAW_CLOSURE_CIF_OFFSET ((FFI_TRAMPOLINE_SIZE + 3) & ~3) +#define RAW_CLOSURE_FUN_OFFSET (RAW_CLOSURE_CIF_OFFSET + 4) +#define RAW_CLOSURE_USER_DATA_OFFSET (RAW_CLOSURE_FUN_OFFSET + 4) +#define CIF_FLAGS_OFFSET 20 + + .align 4 +FFI_HIDDEN (ffi_closure_raw_SYSV) +.globl ffi_closure_raw_SYSV + .type ffi_closure_raw_SYSV, @function + +ffi_closure_raw_SYSV: +.LFB3: + pushl %ebp +.LCFI4: + movl %esp, %ebp +.LCFI5: + pushl %esi +.LCFI6: + subl $36, %esp + movl RAW_CLOSURE_CIF_OFFSET(%eax), %esi /* closure->cif */ + movl RAW_CLOSURE_USER_DATA_OFFSET(%eax), %edx /* closure->user_data */ + movl %edx, 12(%esp) /* user_data */ + leal 8(%ebp), %edx /* __builtin_dwarf_cfa () */ + movl %edx, 8(%esp) /* raw_args */ + leal -24(%ebp), %edx + movl %edx, 4(%esp) /* &res */ + movl %esi, (%esp) /* cif */ + call *RAW_CLOSURE_FUN_OFFSET(%eax) /* closure->fun */ + movl CIF_FLAGS_OFFSET(%esi), %eax /* rtype */ + cmpl $FFI_TYPE_INT, %eax + je .Lrcls_retint + + /* Handle FFI_TYPE_UINT8, FFI_TYPE_SINT8, FFI_TYPE_UINT16, + FFI_TYPE_SINT16, FFI_TYPE_UINT32, FFI_TYPE_SINT32. */ + cmpl $FFI_TYPE_UINT64, %eax + jge 0f + cmpl $FFI_TYPE_UINT8, %eax + jge .Lrcls_retint +0: + cmpl $FFI_TYPE_FLOAT, %eax + je .Lrcls_retfloat + cmpl $FFI_TYPE_DOUBLE, %eax + je .Lrcls_retdouble + cmpl $FFI_TYPE_LONGDOUBLE, %eax + je .Lrcls_retldouble + cmpl $FFI_TYPE_SINT64, %eax + je .Lrcls_retllong +.Lrcls_epilogue: + addl $36, %esp + popl %esi + popl %ebp + ret +.Lrcls_retint: + movl -24(%ebp), %eax + jmp .Lrcls_epilogue +.Lrcls_retfloat: + flds -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retdouble: + fldl -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retldouble: + fldt -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retllong: + movl -24(%ebp), %eax + movl -20(%ebp), %edx + jmp .Lrcls_epilogue +.LFE3: + .size ffi_closure_raw_SYSV, .-ffi_closure_raw_SYSV +#endif + + .section .eh_frame,EH_FRAME_FLAGS,@progbits +.Lframe1: + .long .LECIE1-.LSCIE1 /* Length of Common Information Entry */ +.LSCIE1: + .long 0x0 /* CIE Identifier Tag */ + .byte 0x1 /* CIE Version */ +#ifdef __PIC__ + .ascii "zR\0" /* CIE Augmentation */ +#else + .ascii "\0" /* CIE Augmentation */ +#endif + .byte 0x1 /* .uleb128 0x1; CIE Code Alignment Factor */ + .byte 0x7c /* .sleb128 -4; CIE Data Alignment Factor */ + .byte 0x8 /* CIE RA Column */ +#ifdef __PIC__ + .byte 0x1 /* .uleb128 0x1; Augmentation size */ + .byte 0x1b /* FDE Encoding (pcrel sdata4) */ +#endif + .byte 0xc /* DW_CFA_def_cfa */ + .byte 0x4 /* .uleb128 0x4 */ + .byte 0x4 /* .uleb128 0x4 */ + .byte 0x88 /* DW_CFA_offset, column 0x8 */ + .byte 0x1 /* .uleb128 0x1 */ + .align 4 +.LECIE1: +.LSFDE1: + .long .LEFDE1-.LASFDE1 /* FDE Length */ +.LASFDE1: + .long .LASFDE1-.Lframe1 /* FDE CIE offset */ +#ifdef __PIC__ + .long .LFB1-. /* FDE initial location */ +#else + .long .LFB1 /* FDE initial location */ +#endif + .long .LFE1-.LFB1 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI0-.LFB1 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 */ + .byte 0x2 /* .uleb128 0x2 */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI1-.LCFI0 + .byte 0xd /* DW_CFA_def_cfa_register */ + .byte 0x5 /* .uleb128 0x5 */ + .align 4 +.LEFDE1: +.LSFDE2: + .long .LEFDE2-.LASFDE2 /* FDE Length */ +.LASFDE2: + .long .LASFDE2-.Lframe1 /* FDE CIE offset */ +#ifdef __PIC__ + .long .LFB2-. /* FDE initial location */ +#else + .long .LFB2 +#endif + .long .LFE2-.LFB2 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI2-.LFB2 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 */ + .byte 0x2 /* .uleb128 0x2 */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI3-.LCFI2 + .byte 0xd /* DW_CFA_def_cfa_register */ + .byte 0x5 /* .uleb128 0x5 */ +#if !defined HAVE_HIDDEN_VISIBILITY_ATTRIBUTE && defined __PIC__ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI7-.LCFI3 + .byte 0x83 /* DW_CFA_offset, column 0x3 */ + .byte 0xa /* .uleb128 0xa */ +#endif + .align 4 +.LEFDE2: + +#if !FFI_NO_RAW_API + +.LSFDE3: + .long .LEFDE3-.LASFDE3 /* FDE Length */ +.LASFDE3: + .long .LASFDE3-.Lframe1 /* FDE CIE offset */ +#ifdef __PIC__ + .long .LFB3-. /* FDE initial location */ +#else + .long .LFB3 +#endif + .long .LFE3-.LFB3 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI4-.LFB3 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 */ + .byte 0x2 /* .uleb128 0x2 */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI5-.LCFI4 + .byte 0xd /* DW_CFA_def_cfa_register */ + .byte 0x5 /* .uleb128 0x5 */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI6-.LCFI5 + .byte 0x86 /* DW_CFA_offset, column 0x6 */ + .byte 0x3 /* .uleb128 0x3 */ + .align 4 +.LEFDE3: + +#endif + +#endif /* ifndef __x86_64__ */ + + .section .note.GNU-stack,"",%progbits diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/sysv.S b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/sysv.S new file mode 100644 index 0000000..3bd5477 --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/sysv.S @@ -0,0 +1,483 @@ +/* ----------------------------------------------------------------------- + sysv.S - Copyright (c) 2013 The Written Word, Inc. + - Copyright (c) 1996,1998,2001-2003,2005,2008,2010 Red Hat, Inc. + + X86 Foreign Function Interface + + Permission is hereby granted, free of charge, to any person obtaining + a copy of this software and associated documentation files (the + ``Software''), to deal in the Software without restriction, including + without limitation the rights to use, copy, modify, merge, publish, + distribute, sublicense, and/or sell copies of the Software, and to + permit persons to whom the Software is furnished to do so, subject to + the following conditions: + + The above copyright notice and this permission notice shall be included + in all copies or substantial portions of the Software. + + THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, + EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + DEALINGS IN THE SOFTWARE. + ----------------------------------------------------------------------- */ + +#ifndef __x86_64__ + +#define LIBFFI_ASM +#include +#include + +.text + +.globl ffi_prep_args + + .align 4 +.globl ffi_call_SYSV + .type ffi_call_SYSV,@function + +ffi_call_SYSV: +.LFB1: + pushl %ebp +.LCFI0: + movl %esp,%ebp +.LCFI1: + /* Make room for all of the new args. */ + movl 16(%ebp),%ecx + subl %ecx,%esp + + /* Align the stack pointer to 16-bytes */ + andl $0xfffffff0, %esp + + movl %esp,%eax + + /* Place all of the ffi_prep_args in position */ + pushl 12(%ebp) + pushl %eax + call *8(%ebp) + + /* Return stack to previous state and call the function */ + addl $8,%esp + + call *28(%ebp) + + /* Load %ecx with the return type code */ + movl 20(%ebp),%ecx + + /* Protect %esi. We're going to pop it in the epilogue. */ + pushl %esi + + /* If the return value pointer is NULL, assume no return value. */ + cmpl $0,24(%ebp) + jne 0f + + /* Even if there is no space for the return value, we are + obliged to handle floating-point values. */ + cmpl $FFI_TYPE_FLOAT,%ecx + jne noretval + fstp %st(0) + + jmp epilogue + +0: + call 1f + +.Lstore_table: + .long noretval-.Lstore_table /* FFI_TYPE_VOID */ + .long retint-.Lstore_table /* FFI_TYPE_INT */ + .long retfloat-.Lstore_table /* FFI_TYPE_FLOAT */ + .long retdouble-.Lstore_table /* FFI_TYPE_DOUBLE */ + .long retlongdouble-.Lstore_table /* FFI_TYPE_LONGDOUBLE */ + .long retuint8-.Lstore_table /* FFI_TYPE_UINT8 */ + .long retsint8-.Lstore_table /* FFI_TYPE_SINT8 */ + .long retuint16-.Lstore_table /* FFI_TYPE_UINT16 */ + .long retsint16-.Lstore_table /* FFI_TYPE_SINT16 */ + .long retint-.Lstore_table /* FFI_TYPE_UINT32 */ + .long retint-.Lstore_table /* FFI_TYPE_SINT32 */ + .long retint64-.Lstore_table /* FFI_TYPE_UINT64 */ + .long retint64-.Lstore_table /* FFI_TYPE_SINT64 */ + .long retstruct-.Lstore_table /* FFI_TYPE_STRUCT */ + .long retint-.Lstore_table /* FFI_TYPE_POINTER */ + +1: + pop %esi + add (%esi, %ecx, 4), %esi + jmp *%esi + + /* Sign/zero extend as appropriate. */ +retsint8: + movsbl %al, %eax + jmp retint + +retsint16: + movswl %ax, %eax + jmp retint + +retuint8: + movzbl %al, %eax + jmp retint + +retuint16: + movzwl %ax, %eax + jmp retint + +retfloat: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstps (%ecx) + jmp epilogue + +retdouble: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstpl (%ecx) + jmp epilogue + +retlongdouble: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + fstpt (%ecx) + jmp epilogue + +retint64: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + movl %edx,4(%ecx) + jmp epilogue + +retint: + /* Load %ecx with the pointer to storage for the return value */ + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + +retstruct: + /* Nothing to do! */ + +noretval: +epilogue: + popl %esi + movl %ebp,%esp + popl %ebp + ret +.LFE1: +.ffi_call_SYSV_end: + .size ffi_call_SYSV,.ffi_call_SYSV_end-ffi_call_SYSV + + .align 4 +FFI_HIDDEN (ffi_closure_SYSV) +.globl ffi_closure_SYSV + .type ffi_closure_SYSV, @function + +ffi_closure_SYSV: +.LFB2: + pushl %ebp +.LCFI2: + movl %esp, %ebp +.LCFI3: + subl $40, %esp + leal -24(%ebp), %edx + movl %edx, -12(%ebp) /* resp */ + leal 8(%ebp), %edx +#ifdef __SUNPRO_C + /* The SUNPRO compiler doesn't support GCC's regparm function + attribute, so we have to pass all three arguments to + ffi_closure_SYSV_inner on the stack. */ + movl %edx, 8(%esp) /* args = __builtin_dwarf_cfa () */ + leal -12(%ebp), %edx + movl %edx, 4(%esp) /* &resp */ + movl %eax, (%esp) /* closure */ +#else + movl %edx, 4(%esp) /* args = __builtin_dwarf_cfa () */ + leal -12(%ebp), %edx + movl %edx, (%esp) /* &resp */ +#endif +#if defined HAVE_HIDDEN_VISIBILITY_ATTRIBUTE || !defined __PIC__ + call ffi_closure_SYSV_inner +#else + movl %ebx, 8(%esp) +.LCFI7: + call 1f +1: popl %ebx + addl $_GLOBAL_OFFSET_TABLE_+[.-1b], %ebx + call ffi_closure_SYSV_inner@PLT + movl 8(%esp), %ebx +#endif + movl -12(%ebp), %ecx + cmpl $FFI_TYPE_INT, %eax + je .Lcls_retint + + /* Handle FFI_TYPE_UINT8, FFI_TYPE_SINT8, FFI_TYPE_UINT16, + FFI_TYPE_SINT16, FFI_TYPE_UINT32, FFI_TYPE_SINT32. */ + cmpl $FFI_TYPE_UINT64, %eax + jge 0f + cmpl $FFI_TYPE_UINT8, %eax + jge .Lcls_retint + +0: cmpl $FFI_TYPE_FLOAT, %eax + je .Lcls_retfloat + cmpl $FFI_TYPE_DOUBLE, %eax + je .Lcls_retdouble + cmpl $FFI_TYPE_LONGDOUBLE, %eax + je .Lcls_retldouble + cmpl $FFI_TYPE_SINT64, %eax + je .Lcls_retllong + cmpl $FFI_TYPE_STRUCT, %eax + je .Lcls_retstruct +.Lcls_epilogue: + movl %ebp, %esp + popl %ebp + ret +.Lcls_retint: + movl (%ecx), %eax + jmp .Lcls_epilogue +.Lcls_retfloat: + flds (%ecx) + jmp .Lcls_epilogue +.Lcls_retdouble: + fldl (%ecx) + jmp .Lcls_epilogue +.Lcls_retldouble: + fldt (%ecx) + jmp .Lcls_epilogue +.Lcls_retllong: + movl (%ecx), %eax + movl 4(%ecx), %edx + jmp .Lcls_epilogue +.Lcls_retstruct: + movl %ebp, %esp + popl %ebp + ret $4 +.LFE2: + .size ffi_closure_SYSV, .-ffi_closure_SYSV + +#if !FFI_NO_RAW_API + +/* Precalculate for e.g. the Solaris 10/x86 assembler. */ +#if FFI_TRAMPOLINE_SIZE == 10 +#define RAW_CLOSURE_CIF_OFFSET 12 +#define RAW_CLOSURE_FUN_OFFSET 16 +#define RAW_CLOSURE_USER_DATA_OFFSET 20 +#elif FFI_TRAMPOLINE_SIZE == 24 +#define RAW_CLOSURE_CIF_OFFSET 24 +#define RAW_CLOSURE_FUN_OFFSET 28 +#define RAW_CLOSURE_USER_DATA_OFFSET 32 +#else +#define RAW_CLOSURE_CIF_OFFSET ((FFI_TRAMPOLINE_SIZE + 3) & ~3) +#define RAW_CLOSURE_FUN_OFFSET (RAW_CLOSURE_CIF_OFFSET + 4) +#define RAW_CLOSURE_USER_DATA_OFFSET (RAW_CLOSURE_FUN_OFFSET + 4) +#endif +#define CIF_FLAGS_OFFSET 20 + + .align 4 +FFI_HIDDEN (ffi_closure_raw_SYSV) +.globl ffi_closure_raw_SYSV + .type ffi_closure_raw_SYSV, @function + +ffi_closure_raw_SYSV: +.LFB3: + pushl %ebp +.LCFI4: + movl %esp, %ebp +.LCFI5: + pushl %esi +.LCFI6: + subl $36, %esp + movl RAW_CLOSURE_CIF_OFFSET(%eax), %esi /* closure->cif */ + movl RAW_CLOSURE_USER_DATA_OFFSET(%eax), %edx /* closure->user_data */ + movl %edx, 12(%esp) /* user_data */ + leal 8(%ebp), %edx /* __builtin_dwarf_cfa () */ + movl %edx, 8(%esp) /* raw_args */ + leal -24(%ebp), %edx + movl %edx, 4(%esp) /* &res */ + movl %esi, (%esp) /* cif */ + call *RAW_CLOSURE_FUN_OFFSET(%eax) /* closure->fun */ + movl CIF_FLAGS_OFFSET(%esi), %eax /* rtype */ + cmpl $FFI_TYPE_INT, %eax + je .Lrcls_retint + + /* Handle FFI_TYPE_UINT8, FFI_TYPE_SINT8, FFI_TYPE_UINT16, + FFI_TYPE_SINT16, FFI_TYPE_UINT32, FFI_TYPE_SINT32. */ + cmpl $FFI_TYPE_UINT64, %eax + jge 0f + cmpl $FFI_TYPE_UINT8, %eax + jge .Lrcls_retint +0: + cmpl $FFI_TYPE_FLOAT, %eax + je .Lrcls_retfloat + cmpl $FFI_TYPE_DOUBLE, %eax + je .Lrcls_retdouble + cmpl $FFI_TYPE_LONGDOUBLE, %eax + je .Lrcls_retldouble + cmpl $FFI_TYPE_SINT64, %eax + je .Lrcls_retllong +.Lrcls_epilogue: + addl $36, %esp + popl %esi + popl %ebp + ret +.Lrcls_retint: + movl -24(%ebp), %eax + jmp .Lrcls_epilogue +.Lrcls_retfloat: + flds -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retdouble: + fldl -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retldouble: + fldt -24(%ebp) + jmp .Lrcls_epilogue +.Lrcls_retllong: + movl -24(%ebp), %eax + movl -20(%ebp), %edx + jmp .Lrcls_epilogue +.LFE3: + .size ffi_closure_raw_SYSV, .-ffi_closure_raw_SYSV +#endif + +#if defined __GNUC__ +/* Only emit dwarf unwind info when building with GNU toolchain. */ + +#if defined __PIC__ +# if defined __sun__ && defined __svr4__ +/* 32-bit Solaris 2/x86 uses datarel encoding for PIC. GNU ld before 2.22 + doesn't correctly sort .eh_frame_hdr with mixed encodings, so match this. */ +# define FDE_ENCODING 0x30 /* datarel */ +# define FDE_ENCODE(X) X@GOTOFF +# else +# define FDE_ENCODING 0x1b /* pcrel sdata4 */ +# if defined HAVE_AS_X86_PCREL +# define FDE_ENCODE(X) X-. +# else +# define FDE_ENCODE(X) X@rel +# endif +# endif +#else +# define FDE_ENCODING 0 /* absolute */ +# define FDE_ENCODE(X) X +#endif + + .section .eh_frame,EH_FRAME_FLAGS,@progbits +.Lframe1: + .long .LECIE1-.LSCIE1 /* Length of Common Information Entry */ +.LSCIE1: + .long 0x0 /* CIE Identifier Tag */ + .byte 0x1 /* CIE Version */ +#ifdef HAVE_AS_ASCII_PSEUDO_OP +#ifdef __PIC__ + .ascii "zR\0" /* CIE Augmentation */ +#else + .ascii "\0" /* CIE Augmentation */ +#endif +#elif defined HAVE_AS_STRING_PSEUDO_OP +#ifdef __PIC__ + .string "zR" /* CIE Augmentation */ +#else + .string "" /* CIE Augmentation */ +#endif +#else +#error missing .ascii/.string +#endif + .byte 0x1 /* .uleb128 0x1; CIE Code Alignment Factor */ + .byte 0x7c /* .sleb128 -4; CIE Data Alignment Factor */ + .byte 0x8 /* CIE RA Column */ +#ifdef __PIC__ + .byte 0x1 /* .uleb128 0x1; Augmentation size */ + .byte FDE_ENCODING +#endif + .byte 0xc /* DW_CFA_def_cfa */ + .byte 0x4 /* .uleb128 0x4 */ + .byte 0x4 /* .uleb128 0x4 */ + .byte 0x88 /* DW_CFA_offset, column 0x8 */ + .byte 0x1 /* .uleb128 0x1 */ + .align 4 +.LECIE1: +.LSFDE1: + .long .LEFDE1-.LASFDE1 /* FDE Length */ +.LASFDE1: + .long .LASFDE1-.Lframe1 /* FDE CIE offset */ + .long FDE_ENCODE(.LFB1) /* FDE initial location */ + .long .LFE1-.LFB1 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI0-.LFB1 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 */ + .byte 0x2 /* .uleb128 0x2 */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI1-.LCFI0 + .byte 0xd /* DW_CFA_def_cfa_register */ + .byte 0x5 /* .uleb128 0x5 */ + .align 4 +.LEFDE1: +.LSFDE2: + .long .LEFDE2-.LASFDE2 /* FDE Length */ +.LASFDE2: + .long .LASFDE2-.Lframe1 /* FDE CIE offset */ + .long FDE_ENCODE(.LFB2) /* FDE initial location */ + .long .LFE2-.LFB2 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI2-.LFB2 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 */ + .byte 0x2 /* .uleb128 0x2 */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI3-.LCFI2 + .byte 0xd /* DW_CFA_def_cfa_register */ + .byte 0x5 /* .uleb128 0x5 */ +#if !defined HAVE_HIDDEN_VISIBILITY_ATTRIBUTE && defined __PIC__ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI7-.LCFI3 + .byte 0x83 /* DW_CFA_offset, column 0x3 */ + .byte 0xa /* .uleb128 0xa */ +#endif + .align 4 +.LEFDE2: + +#if !FFI_NO_RAW_API + +.LSFDE3: + .long .LEFDE3-.LASFDE3 /* FDE Length */ +.LASFDE3: + .long .LASFDE3-.Lframe1 /* FDE CIE offset */ + .long FDE_ENCODE(.LFB3) /* FDE initial location */ + .long .LFE3-.LFB3 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI4-.LFB3 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 */ + .byte 0x2 /* .uleb128 0x2 */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI5-.LCFI4 + .byte 0xd /* DW_CFA_def_cfa_register */ + .byte 0x5 /* .uleb128 0x5 */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI6-.LCFI5 + .byte 0x86 /* DW_CFA_offset, column 0x6 */ + .byte 0x3 /* .uleb128 0x3 */ + .align 4 +.LEFDE3: + +#endif +#endif + +#endif /* ifndef __x86_64__ */ + +#if defined __ELF__ && defined __linux__ + .section .note.GNU-stack,"",@progbits +#endif diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/unix64.S b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/unix64.S new file mode 100644 index 0000000..dcd6bc7 --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/unix64.S @@ -0,0 +1,432 @@ +/* ----------------------------------------------------------------------- + unix64.S - Copyright (c) 2013 The Written Word, Inc. + - Copyright (c) 2008 Red Hat, Inc + - Copyright (c) 2002 Bo Thorsen + + 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. + ----------------------------------------------------------------------- */ + +#ifdef __x86_64__ +#define LIBFFI_ASM +#include +#include + +.text + +/* ffi_call_unix64 (void *args, unsigned long bytes, unsigned flags, + void *raddr, void (*fnaddr)(void)); + + Bit o trickiness here -- ARGS+BYTES is the base of the stack frame + for this function. This has been allocated by ffi_call. We also + deallocate some of the stack that has been alloca'd. */ + + .align 2 + .globl ffi_call_unix64 + .type ffi_call_unix64,@function + +ffi_call_unix64: +.LUW0: + movq (%rsp), %r10 /* Load return address. */ + leaq (%rdi, %rsi), %rax /* Find local stack base. */ + movq %rdx, (%rax) /* Save flags. */ + movq %rcx, 8(%rax) /* Save raddr. */ + movq %rbp, 16(%rax) /* Save old frame pointer. */ + movq %r10, 24(%rax) /* Relocate return address. */ + movq %rax, %rbp /* Finalize local stack frame. */ +.LUW1: + movq %rdi, %r10 /* Save a copy of the register area. */ + movq %r8, %r11 /* Save a copy of the target fn. */ + movl %r9d, %eax /* Set number of SSE registers. */ + + /* Load up all argument registers. */ + movq (%r10), %rdi + movq 8(%r10), %rsi + movq 16(%r10), %rdx + movq 24(%r10), %rcx + movq 32(%r10), %r8 + movq 40(%r10), %r9 + testl %eax, %eax + jnz .Lload_sse +.Lret_from_load_sse: + + /* Deallocate the reg arg area. */ + leaq 176(%r10), %rsp + + /* Call the user function. */ + call *%r11 + + /* Deallocate stack arg area; local stack frame in redzone. */ + leaq 24(%rbp), %rsp + + movq 0(%rbp), %rcx /* Reload flags. */ + movq 8(%rbp), %rdi /* Reload raddr. */ + movq 16(%rbp), %rbp /* Reload old frame pointer. */ +.LUW2: + + /* The first byte of the flags contains the FFI_TYPE. */ + movzbl %cl, %r10d + leaq .Lstore_table(%rip), %r11 + movslq (%r11, %r10, 4), %r10 + addq %r11, %r10 + jmp *%r10 + +.Lstore_table: + .long .Lst_void-.Lstore_table /* FFI_TYPE_VOID */ + .long .Lst_sint32-.Lstore_table /* FFI_TYPE_INT */ + .long .Lst_float-.Lstore_table /* FFI_TYPE_FLOAT */ + .long .Lst_double-.Lstore_table /* FFI_TYPE_DOUBLE */ + .long .Lst_ldouble-.Lstore_table /* FFI_TYPE_LONGDOUBLE */ + .long .Lst_uint8-.Lstore_table /* FFI_TYPE_UINT8 */ + .long .Lst_sint8-.Lstore_table /* FFI_TYPE_SINT8 */ + .long .Lst_uint16-.Lstore_table /* FFI_TYPE_UINT16 */ + .long .Lst_sint16-.Lstore_table /* FFI_TYPE_SINT16 */ + .long .Lst_uint32-.Lstore_table /* FFI_TYPE_UINT32 */ + .long .Lst_sint32-.Lstore_table /* FFI_TYPE_SINT32 */ + .long .Lst_int64-.Lstore_table /* FFI_TYPE_UINT64 */ + .long .Lst_int64-.Lstore_table /* FFI_TYPE_SINT64 */ + .long .Lst_struct-.Lstore_table /* FFI_TYPE_STRUCT */ + .long .Lst_int64-.Lstore_table /* FFI_TYPE_POINTER */ + + .align 2 +.Lst_void: + ret + .align 2 + +.Lst_uint8: + movzbq %al, %rax + movq %rax, (%rdi) + ret + .align 2 +.Lst_sint8: + movsbq %al, %rax + movq %rax, (%rdi) + ret + .align 2 +.Lst_uint16: + movzwq %ax, %rax + movq %rax, (%rdi) + .align 2 +.Lst_sint16: + movswq %ax, %rax + movq %rax, (%rdi) + ret + .align 2 +.Lst_uint32: + movl %eax, %eax + movq %rax, (%rdi) + .align 2 +.Lst_sint32: + cltq + movq %rax, (%rdi) + ret + .align 2 +.Lst_int64: + movq %rax, (%rdi) + ret + + .align 2 +.Lst_float: + movss %xmm0, (%rdi) + ret + .align 2 +.Lst_double: + movsd %xmm0, (%rdi) + ret +.Lst_ldouble: + fstpt (%rdi) + ret + + .align 2 +.Lst_struct: + leaq -20(%rsp), %rsi /* Scratch area in redzone. */ + + /* We have to locate the values now, and since we don't want to + write too much data into the user's return value, we spill the + value to a 16 byte scratch area first. Bits 8, 9, and 10 + control where the values are located. Only one of the three + bits will be set; see ffi_prep_cif_machdep for the pattern. */ + movd %xmm0, %r10 + movd %xmm1, %r11 + testl $0x100, %ecx + cmovnz %rax, %rdx + cmovnz %r10, %rax + testl $0x200, %ecx + cmovnz %r10, %rdx + testl $0x400, %ecx + cmovnz %r10, %rax + cmovnz %r11, %rdx + movq %rax, (%rsi) + movq %rdx, 8(%rsi) + + /* Bits 12-31 contain the true size of the structure. Copy from + the scratch area to the true destination. */ + shrl $12, %ecx + rep movsb + ret + + /* Many times we can avoid loading any SSE registers at all. + It's not worth an indirect jump to load the exact set of + SSE registers needed; zero or all is a good compromise. */ + .align 2 +.LUW3: +.Lload_sse: + movdqa 48(%r10), %xmm0 + movdqa 64(%r10), %xmm1 + movdqa 80(%r10), %xmm2 + movdqa 96(%r10), %xmm3 + movdqa 112(%r10), %xmm4 + movdqa 128(%r10), %xmm5 + movdqa 144(%r10), %xmm6 + movdqa 160(%r10), %xmm7 + jmp .Lret_from_load_sse + +.LUW4: + .size ffi_call_unix64,.-ffi_call_unix64 + + .align 2 + .globl ffi_closure_unix64 + .type ffi_closure_unix64,@function + +ffi_closure_unix64: +.LUW5: + /* The carry flag is set by the trampoline iff SSE registers + are used. Don't clobber it before the branch instruction. */ + leaq -200(%rsp), %rsp +.LUW6: + movq %rdi, (%rsp) + movq %rsi, 8(%rsp) + movq %rdx, 16(%rsp) + movq %rcx, 24(%rsp) + movq %r8, 32(%rsp) + movq %r9, 40(%rsp) + jc .Lsave_sse +.Lret_from_save_sse: + + movq %r10, %rdi + leaq 176(%rsp), %rsi + movq %rsp, %rdx + leaq 208(%rsp), %rcx + call ffi_closure_unix64_inner@PLT + + /* Deallocate stack frame early; return value is now in redzone. */ + addq $200, %rsp +.LUW7: + + /* The first byte of the return value contains the FFI_TYPE. */ + movzbl %al, %r10d + leaq .Lload_table(%rip), %r11 + movslq (%r11, %r10, 4), %r10 + addq %r11, %r10 + jmp *%r10 + +.Lload_table: + .long .Lld_void-.Lload_table /* FFI_TYPE_VOID */ + .long .Lld_int32-.Lload_table /* FFI_TYPE_INT */ + .long .Lld_float-.Lload_table /* FFI_TYPE_FLOAT */ + .long .Lld_double-.Lload_table /* FFI_TYPE_DOUBLE */ + .long .Lld_ldouble-.Lload_table /* FFI_TYPE_LONGDOUBLE */ + .long .Lld_int8-.Lload_table /* FFI_TYPE_UINT8 */ + .long .Lld_int8-.Lload_table /* FFI_TYPE_SINT8 */ + .long .Lld_int16-.Lload_table /* FFI_TYPE_UINT16 */ + .long .Lld_int16-.Lload_table /* FFI_TYPE_SINT16 */ + .long .Lld_int32-.Lload_table /* FFI_TYPE_UINT32 */ + .long .Lld_int32-.Lload_table /* FFI_TYPE_SINT32 */ + .long .Lld_int64-.Lload_table /* FFI_TYPE_UINT64 */ + .long .Lld_int64-.Lload_table /* FFI_TYPE_SINT64 */ + .long .Lld_struct-.Lload_table /* FFI_TYPE_STRUCT */ + .long .Lld_int64-.Lload_table /* FFI_TYPE_POINTER */ + + .align 2 +.Lld_void: + ret + + .align 2 +.Lld_int8: + movzbl -24(%rsp), %eax + ret + .align 2 +.Lld_int16: + movzwl -24(%rsp), %eax + ret + .align 2 +.Lld_int32: + movl -24(%rsp), %eax + ret + .align 2 +.Lld_int64: + movq -24(%rsp), %rax + ret + + .align 2 +.Lld_float: + movss -24(%rsp), %xmm0 + ret + .align 2 +.Lld_double: + movsd -24(%rsp), %xmm0 + ret + .align 2 +.Lld_ldouble: + fldt -24(%rsp) + ret + + .align 2 +.Lld_struct: + /* There are four possibilities here, %rax/%rdx, %xmm0/%rax, + %rax/%xmm0, %xmm0/%xmm1. We collapse two by always loading + both rdx and xmm1 with the second word. For the remaining, + bit 8 set means xmm0 gets the second word, and bit 9 means + that rax gets the second word. */ + movq -24(%rsp), %rcx + movq -16(%rsp), %rdx + movq -16(%rsp), %xmm1 + testl $0x100, %eax + cmovnz %rdx, %rcx + movd %rcx, %xmm0 + testl $0x200, %eax + movq -24(%rsp), %rax + cmovnz %rdx, %rax + ret + + /* See the comment above .Lload_sse; the same logic applies here. */ + .align 2 +.LUW8: +.Lsave_sse: + movdqa %xmm0, 48(%rsp) + movdqa %xmm1, 64(%rsp) + movdqa %xmm2, 80(%rsp) + movdqa %xmm3, 96(%rsp) + movdqa %xmm4, 112(%rsp) + movdqa %xmm5, 128(%rsp) + movdqa %xmm6, 144(%rsp) + movdqa %xmm7, 160(%rsp) + jmp .Lret_from_save_sse + +.LUW9: + .size ffi_closure_unix64,.-ffi_closure_unix64 + +#ifdef __GNUC__ +/* Only emit DWARF unwind info when building with the GNU toolchain. */ + +#ifdef HAVE_AS_X86_64_UNWIND_SECTION_TYPE + .section .eh_frame,"a",@unwind +#else + .section .eh_frame,"a",@progbits +#endif +.Lframe1: + .long .LECIE1-.LSCIE1 /* CIE Length */ +.LSCIE1: + .long 0 /* CIE Identifier Tag */ + .byte 1 /* CIE Version */ + .ascii "zR\0" /* CIE Augmentation */ + .uleb128 1 /* CIE Code Alignment Factor */ + .sleb128 -8 /* CIE Data Alignment Factor */ + .byte 0x10 /* CIE RA Column */ + .uleb128 1 /* Augmentation size */ + .byte 0x1b /* FDE Encoding (pcrel sdata4) */ + .byte 0xc /* DW_CFA_def_cfa, %rsp offset 8 */ + .uleb128 7 + .uleb128 8 + .byte 0x80+16 /* DW_CFA_offset, %rip offset 1*-8 */ + .uleb128 1 + .align 8 +.LECIE1: +.LSFDE1: + .long .LEFDE1-.LASFDE1 /* FDE Length */ +.LASFDE1: + .long .LASFDE1-.Lframe1 /* FDE CIE offset */ +#if HAVE_AS_X86_PCREL + .long .LUW0-. /* FDE initial location */ +#else + .long .LUW0@rel +#endif + .long .LUW4-.LUW0 /* FDE address range */ + .uleb128 0x0 /* Augmentation size */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LUW1-.LUW0 + + /* New stack frame based off rbp. This is a itty bit of unwind + trickery in that the CFA *has* changed. There is no easy way + to describe it correctly on entry to the function. Fortunately, + it doesn't matter too much since at all points we can correctly + unwind back to ffi_call. Note that the location to which we + moved the return address is (the new) CFA-8, so from the + perspective of the unwind info, it hasn't moved. */ + .byte 0xc /* DW_CFA_def_cfa, %rbp offset 32 */ + .uleb128 6 + .uleb128 32 + .byte 0x80+6 /* DW_CFA_offset, %rbp offset 2*-8 */ + .uleb128 2 + .byte 0xa /* DW_CFA_remember_state */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LUW2-.LUW1 + .byte 0xc /* DW_CFA_def_cfa, %rsp offset 8 */ + .uleb128 7 + .uleb128 8 + .byte 0xc0+6 /* DW_CFA_restore, %rbp */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LUW3-.LUW2 + .byte 0xb /* DW_CFA_restore_state */ + + .align 8 +.LEFDE1: +.LSFDE3: + .long .LEFDE3-.LASFDE3 /* FDE Length */ +.LASFDE3: + .long .LASFDE3-.Lframe1 /* FDE CIE offset */ +#if HAVE_AS_X86_PCREL + .long .LUW5-. /* FDE initial location */ +#else + .long .LUW5@rel +#endif + .long .LUW9-.LUW5 /* FDE address range */ + .uleb128 0x0 /* Augmentation size */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LUW6-.LUW5 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .uleb128 208 + .byte 0xa /* DW_CFA_remember_state */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LUW7-.LUW6 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .uleb128 8 + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LUW8-.LUW7 + .byte 0xb /* DW_CFA_restore_state */ + + .align 8 +.LEFDE3: + +#endif /* __GNUC__ */ + +#endif /* __x86_64__ */ + +#if defined __ELF__ && defined __linux__ + .section .note.GNU-stack,"",@progbits +#endif diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win32.S b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win32.S new file mode 100644 index 0000000..3680bf5 --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win32.S @@ -0,0 +1,1351 @@ +/* ----------------------------------------------------------------------- + win32.S - Copyright (c) 2014 Anthony Green + Copyright (c) 1996, 1998, 2001, 2002, 2009 Red Hat, Inc. + Copyright (c) 2001 John Beniton + Copyright (c) 2002 Ranjit Mathew + Copyright (c) 2009 Daniel Witte + + + X86 Foreign Function Interface + + Permission is hereby granted, free of charge, to any person obtaining + a copy of this software and associated documentation files (the + ``Software''), to deal in the Software without restriction, including + without limitation the rights to use, copy, modify, merge, publish, + distribute, sublicense, and/or sell copies of the Software, and to + permit persons to whom the Software is furnished to do so, subject to + the following conditions: + + The above copyright notice and this permission notice shall be included + in all copies or substantial portions of the Software. + + THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, + EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + DEALINGS IN THE SOFTWARE. + ----------------------------------------------------------------------- + */ + +#define LIBFFI_ASM +#include +#include + +#define CIF_BYTES_OFFSET 16 +#define CIF_FLAGS_OFFSET 20 + +#ifdef _MSC_VER + +#define CLOSURE_CIF_OFFSET ((FFI_TRAMPOLINE_SIZE + 3) AND NOT 3) + +.386 +.MODEL FLAT, C + +EXTRN ffi_closure_SYSV_inner:NEAR +EXTRN ffi_closure_WIN32_inner:NEAR + +_TEXT SEGMENT + +ffi_call_win32 PROC NEAR, + ffi_prep_args : NEAR PTR DWORD, + ecif : NEAR PTR DWORD, + cif_abi : DWORD, + cif_bytes : DWORD, + cif_flags : DWORD, + rvalue : NEAR PTR DWORD, + fn : NEAR PTR DWORD + + ;; Make room for all of the new args. + mov ecx, cif_bytes + sub esp, ecx + + mov eax, esp + + ;; Call ffi_prep_args + push ecif + push eax + call ffi_prep_args + add esp, 8 + + ;; Prepare registers + ;; EAX stores the number of register arguments + cmp eax, 0 + je fun + cmp eax, 3 + jl prepr_two_cmp + + mov ecx, esp + add esp, 12 + mov eax, DWORD PTR [ecx+8] + jmp prepr_two +prepr_two_cmp: + cmp eax, 2 + jl prepr_one_prep + mov ecx, esp + add esp, 8 +prepr_two: + mov edx, DWORD PTR [ecx+4] + jmp prepr_one +prepr_one_prep: + mov ecx, esp + add esp, 4 +prepr_one: + mov ecx, DWORD PTR [ecx] + cmp cif_abi, 7 ;; FFI_REGISTER + jne fun + + xchg ecx, eax + +fun: + ;; Call function + call fn + + ;; Load ecx with the return type code + mov ecx, cif_flags + + ;; If the return value pointer is NULL, assume no return value. + cmp rvalue, 0 + jne ca_jumptable + + ;; Even if there is no space for the return value, we are + ;; obliged to handle floating-point values. + cmp ecx, FFI_TYPE_FLOAT + jne ca_epilogue + fstp st(0) + + jmp ca_epilogue + +ca_jumptable: + jmp [ca_jumpdata + 4 * ecx] +ca_jumpdata: + ;; Do not insert anything here between label and jump table. + dd offset ca_epilogue ;; FFI_TYPE_VOID + dd offset ca_retint ;; FFI_TYPE_INT + dd offset ca_retfloat ;; FFI_TYPE_FLOAT + dd offset ca_retdouble ;; FFI_TYPE_DOUBLE + dd offset ca_retlongdouble ;; FFI_TYPE_LONGDOUBLE + dd offset ca_retuint8 ;; FFI_TYPE_UINT8 + dd offset ca_retsint8 ;; FFI_TYPE_SINT8 + dd offset ca_retuint16 ;; FFI_TYPE_UINT16 + dd offset ca_retsint16 ;; FFI_TYPE_SINT16 + dd offset ca_retint ;; FFI_TYPE_UINT32 + dd offset ca_retint ;; FFI_TYPE_SINT32 + dd offset ca_retint64 ;; FFI_TYPE_UINT64 + dd offset ca_retint64 ;; FFI_TYPE_SINT64 + dd offset ca_epilogue ;; FFI_TYPE_STRUCT + dd offset ca_retint ;; FFI_TYPE_POINTER + dd offset ca_retstruct1b ;; FFI_TYPE_SMALL_STRUCT_1B + dd offset ca_retstruct2b ;; FFI_TYPE_SMALL_STRUCT_2B + dd offset ca_retint ;; FFI_TYPE_SMALL_STRUCT_4B + dd offset ca_epilogue ;; FFI_TYPE_MS_STRUCT + + /* Sign/zero extend as appropriate. */ +ca_retuint8: + movzx eax, al + jmp ca_retint + +ca_retsint8: + movsx eax, al + jmp ca_retint + +ca_retuint16: + movzx eax, ax + jmp ca_retint + +ca_retsint16: + movsx eax, ax + jmp ca_retint + +ca_retint: + ;; Load %ecx with the pointer to storage for the return value + mov ecx, rvalue + mov [ecx + 0], eax + jmp ca_epilogue + +ca_retint64: + ;; Load %ecx with the pointer to storage for the return value + mov ecx, rvalue + mov [ecx + 0], eax + mov [ecx + 4], edx + jmp ca_epilogue + +ca_retfloat: + ;; Load %ecx with the pointer to storage for the return value + mov ecx, rvalue + fstp DWORD PTR [ecx] + jmp ca_epilogue + +ca_retdouble: + ;; Load %ecx with the pointer to storage for the return value + mov ecx, rvalue + fstp QWORD PTR [ecx] + jmp ca_epilogue + +ca_retlongdouble: + ;; Load %ecx with the pointer to storage for the return value + mov ecx, rvalue + fstp TBYTE PTR [ecx] + jmp ca_epilogue + +ca_retstruct1b: + ;; Load %ecx with the pointer to storage for the return value + mov ecx, rvalue + mov [ecx + 0], al + jmp ca_epilogue + +ca_retstruct2b: + ;; Load %ecx with the pointer to storage for the return value + mov ecx, rvalue + mov [ecx + 0], ax + jmp ca_epilogue + +ca_epilogue: + ;; Epilogue code is autogenerated. + ret +ffi_call_win32 ENDP + +ffi_closure_THISCALL PROC NEAR + ;; Insert the register argument on the stack as the first argument + xchg DWORD PTR [esp+4], ecx + xchg DWORD PTR [esp], ecx + push ecx + jmp ffi_closure_STDCALL +ffi_closure_THISCALL ENDP + +ffi_closure_FASTCALL PROC NEAR + ;; Insert the 2 register arguments on the stack as the first argument + xchg DWORD PTR [esp+4], edx + xchg DWORD PTR [esp], ecx + push edx + push ecx + jmp ffi_closure_STDCALL +ffi_closure_FASTCALL ENDP + +ffi_closure_REGISTER PROC NEAR + ;; Insert the 3 register arguments on the stack as the first argument + push eax + xchg DWORD PTR [esp+8], ecx + xchg DWORD PTR [esp+4], edx + push ecx + push edx + jmp ffi_closure_STDCALL +ffi_closure_REGISTER ENDP + +ffi_closure_SYSV PROC NEAR FORCEFRAME + ;; the ffi_closure ctx is passed in eax by the trampoline. + + sub esp, 40 + lea edx, [ebp - 24] + mov [ebp - 12], edx ;; resp + lea edx, [ebp + 8] +stub:: + mov [esp + 8], edx ;; args + lea edx, [ebp - 12] + mov [esp + 4], edx ;; &resp + mov [esp], eax ;; closure + call ffi_closure_SYSV_inner + mov ecx, [ebp - 12] + +cs_jumptable: + jmp [cs_jumpdata + 4 * eax] +cs_jumpdata: + ;; Do not insert anything here between the label and jump table. + dd offset cs_epilogue ;; FFI_TYPE_VOID + dd offset cs_retint ;; FFI_TYPE_INT + dd offset cs_retfloat ;; FFI_TYPE_FLOAT + dd offset cs_retdouble ;; FFI_TYPE_DOUBLE + dd offset cs_retlongdouble ;; FFI_TYPE_LONGDOUBLE + dd offset cs_retuint8 ;; FFI_TYPE_UINT8 + dd offset cs_retsint8 ;; FFI_TYPE_SINT8 + dd offset cs_retuint16 ;; FFI_TYPE_UINT16 + dd offset cs_retsint16 ;; FFI_TYPE_SINT16 + dd offset cs_retint ;; FFI_TYPE_UINT32 + dd offset cs_retint ;; FFI_TYPE_SINT32 + dd offset cs_retint64 ;; FFI_TYPE_UINT64 + dd offset cs_retint64 ;; FFI_TYPE_SINT64 + dd offset cs_retstruct ;; FFI_TYPE_STRUCT + dd offset cs_retint ;; FFI_TYPE_POINTER + dd offset cs_retsint8 ;; FFI_TYPE_SMALL_STRUCT_1B + dd offset cs_retsint16 ;; FFI_TYPE_SMALL_STRUCT_2B + dd offset cs_retint ;; FFI_TYPE_SMALL_STRUCT_4B + dd offset cs_retmsstruct ;; FFI_TYPE_MS_STRUCT + +cs_retuint8: + movzx eax, BYTE PTR [ecx] + jmp cs_epilogue + +cs_retsint8: + movsx eax, BYTE PTR [ecx] + jmp cs_epilogue + +cs_retuint16: + movzx eax, WORD PTR [ecx] + jmp cs_epilogue + +cs_retsint16: + movsx eax, WORD PTR [ecx] + jmp cs_epilogue + +cs_retint: + mov eax, [ecx] + jmp cs_epilogue + +cs_retint64: + mov eax, [ecx + 0] + mov edx, [ecx + 4] + jmp cs_epilogue + +cs_retfloat: + fld DWORD PTR [ecx] + jmp cs_epilogue + +cs_retdouble: + fld QWORD PTR [ecx] + jmp cs_epilogue + +cs_retlongdouble: + fld TBYTE PTR [ecx] + jmp cs_epilogue + +cs_retstruct: + ;; Caller expects us to pop struct return value pointer hidden arg. + ;; Epilogue code is autogenerated. + ret 4 + +cs_retmsstruct: + ;; Caller expects us to return a pointer to the real return value. + mov eax, ecx + ;; Caller doesn't expects us to pop struct return value pointer hidden arg. + jmp cs_epilogue + +cs_epilogue: + ;; Epilogue code is autogenerated. + ret +ffi_closure_SYSV ENDP + +#if !FFI_NO_RAW_API + +#define RAW_CLOSURE_CIF_OFFSET ((FFI_TRAMPOLINE_SIZE + 3) AND NOT 3) +#define RAW_CLOSURE_FUN_OFFSET (RAW_CLOSURE_CIF_OFFSET + 4) +#define RAW_CLOSURE_USER_DATA_OFFSET (RAW_CLOSURE_FUN_OFFSET + 4) + +ffi_closure_raw_THISCALL PROC NEAR USES esi FORCEFRAME + sub esp, 36 + mov esi, [eax + RAW_CLOSURE_CIF_OFFSET] ;; closure->cif + mov edx, [eax + RAW_CLOSURE_USER_DATA_OFFSET] ;; closure->user_data + mov [esp + 12], edx + lea edx, [ebp + 12] + jmp stubraw +ffi_closure_raw_THISCALL ENDP + +ffi_closure_raw_SYSV PROC NEAR USES esi FORCEFRAME + ;; the ffi_closure ctx is passed in eax by the trampoline. + + sub esp, 40 + mov esi, [eax + RAW_CLOSURE_CIF_OFFSET] ;; closure->cif + mov edx, [eax + RAW_CLOSURE_USER_DATA_OFFSET] ;; closure->user_data + mov [esp + 12], edx ;; user_data + lea edx, [ebp + 8] +stubraw:: + mov [esp + 8], edx ;; raw_args + lea edx, [ebp - 24] + mov [esp + 4], edx ;; &res + mov [esp], esi ;; cif + call DWORD PTR [eax + RAW_CLOSURE_FUN_OFFSET] ;; closure->fun + mov eax, [esi + CIF_FLAGS_OFFSET] ;; cif->flags + lea ecx, [ebp - 24] + +cr_jumptable: + jmp [cr_jumpdata + 4 * eax] +cr_jumpdata: + ;; Do not insert anything here between the label and jump table. + dd offset cr_epilogue ;; FFI_TYPE_VOID + dd offset cr_retint ;; FFI_TYPE_INT + dd offset cr_retfloat ;; FFI_TYPE_FLOAT + dd offset cr_retdouble ;; FFI_TYPE_DOUBLE + dd offset cr_retlongdouble ;; FFI_TYPE_LONGDOUBLE + dd offset cr_retuint8 ;; FFI_TYPE_UINT8 + dd offset cr_retsint8 ;; FFI_TYPE_SINT8 + dd offset cr_retuint16 ;; FFI_TYPE_UINT16 + dd offset cr_retsint16 ;; FFI_TYPE_SINT16 + dd offset cr_retint ;; FFI_TYPE_UINT32 + dd offset cr_retint ;; FFI_TYPE_SINT32 + dd offset cr_retint64 ;; FFI_TYPE_UINT64 + dd offset cr_retint64 ;; FFI_TYPE_SINT64 + dd offset cr_epilogue ;; FFI_TYPE_STRUCT + dd offset cr_retint ;; FFI_TYPE_POINTER + dd offset cr_retsint8 ;; FFI_TYPE_SMALL_STRUCT_1B + dd offset cr_retsint16 ;; FFI_TYPE_SMALL_STRUCT_2B + dd offset cr_retint ;; FFI_TYPE_SMALL_STRUCT_4B + dd offset cr_epilogue ;; FFI_TYPE_MS_STRUCT + +cr_retuint8: + movzx eax, BYTE PTR [ecx] + jmp cr_epilogue + +cr_retsint8: + movsx eax, BYTE PTR [ecx] + jmp cr_epilogue + +cr_retuint16: + movzx eax, WORD PTR [ecx] + jmp cr_epilogue + +cr_retsint16: + movsx eax, WORD PTR [ecx] + jmp cr_epilogue + +cr_retint: + mov eax, [ecx] + jmp cr_epilogue + +cr_retint64: + mov eax, [ecx + 0] + mov edx, [ecx + 4] + jmp cr_epilogue + +cr_retfloat: + fld DWORD PTR [ecx] + jmp cr_epilogue + +cr_retdouble: + fld QWORD PTR [ecx] + jmp cr_epilogue + +cr_retlongdouble: + fld TBYTE PTR [ecx] + jmp cr_epilogue + +cr_epilogue: + ;; Epilogue code is autogenerated. + ret +ffi_closure_raw_SYSV ENDP + +#endif /* !FFI_NO_RAW_API */ + +ffi_closure_STDCALL PROC NEAR FORCEFRAME + mov eax, [esp] ;; the ffi_closure ctx passed by the trampoline. + + sub esp, 40 + lea edx, [ebp - 24] + mov [ebp - 12], edx ;; resp + lea edx, [ebp + 12] ;; account for stub return address on stack + mov [esp + 8], edx ;; args + lea edx, [ebp - 12] + mov [esp + 4], edx ;; &resp + mov [esp], eax ;; closure + call ffi_closure_WIN32_inner + mov ecx, [ebp - 12] + + xchg [ebp + 4], eax ;;xchg size of stack parameters and ffi_closure ctx + mov eax, DWORD PTR [eax + CLOSURE_CIF_OFFSET] + mov eax, DWORD PTR [eax + CIF_FLAGS_OFFSET] + +cd_jumptable: + jmp [cd_jumpdata + 4 * eax] +cd_jumpdata: + ;; Do not insert anything here between the label and jump table. + dd offset cd_epilogue ;; FFI_TYPE_VOID + dd offset cd_retint ;; FFI_TYPE_INT + dd offset cd_retfloat ;; FFI_TYPE_FLOAT + dd offset cd_retdouble ;; FFI_TYPE_DOUBLE + dd offset cd_retlongdouble ;; FFI_TYPE_LONGDOUBLE + dd offset cd_retuint8 ;; FFI_TYPE_UINT8 + dd offset cd_retsint8 ;; FFI_TYPE_SINT8 + dd offset cd_retuint16 ;; FFI_TYPE_UINT16 + dd offset cd_retsint16 ;; FFI_TYPE_SINT16 + dd offset cd_retint ;; FFI_TYPE_UINT32 + dd offset cd_retint ;; FFI_TYPE_SINT32 + dd offset cd_retint64 ;; FFI_TYPE_UINT64 + dd offset cd_retint64 ;; FFI_TYPE_SINT64 + dd offset cd_epilogue ;; FFI_TYPE_STRUCT + dd offset cd_retint ;; FFI_TYPE_POINTER + dd offset cd_retsint8 ;; FFI_TYPE_SMALL_STRUCT_1B + dd offset cd_retsint16 ;; FFI_TYPE_SMALL_STRUCT_2B + dd offset cd_retint ;; FFI_TYPE_SMALL_STRUCT_4B + +cd_retuint8: + movzx eax, BYTE PTR [ecx] + jmp cd_epilogue + +cd_retsint8: + movsx eax, BYTE PTR [ecx] + jmp cd_epilogue + +cd_retuint16: + movzx eax, WORD PTR [ecx] + jmp cd_epilogue + +cd_retsint16: + movsx eax, WORD PTR [ecx] + jmp cd_epilogue + +cd_retint: + mov eax, [ecx] + jmp cd_epilogue + +cd_retint64: + mov eax, [ecx + 0] + mov edx, [ecx + 4] + jmp cd_epilogue + +cd_retfloat: + fld DWORD PTR [ecx] + jmp cd_epilogue + +cd_retdouble: + fld QWORD PTR [ecx] + jmp cd_epilogue + +cd_retlongdouble: + fld TBYTE PTR [ecx] + jmp cd_epilogue + +cd_epilogue: + mov esp, ebp + pop ebp + mov ecx, [esp + 4] ;; Return address + add esp, [esp] ;; Parameters stack size + add esp, 8 + jmp ecx +ffi_closure_STDCALL ENDP + +_TEXT ENDS +END + +#else + +#define CLOSURE_CIF_OFFSET ((FFI_TRAMPOLINE_SIZE + 3) & ~3) + +#if defined(SYMBOL_UNDERSCORE) +#define USCORE_SYMBOL(x) _##x +#else +#define USCORE_SYMBOL(x) x +#endif + .text + + # This assumes we are using gas. + .balign 16 +FFI_HIDDEN(ffi_call_win32) + .globl USCORE_SYMBOL(ffi_call_win32) +#if defined(X86_WIN32) && !defined(__OS2__) + .def _ffi_call_win32; .scl 2; .type 32; .endef +#endif +USCORE_SYMBOL(ffi_call_win32): +.LFB1: + pushl %ebp +.LCFI0: + movl %esp,%ebp +.LCFI1: + # Make room for all of the new args. + movl 20(%ebp),%ecx + subl %ecx,%esp + + movl %esp,%eax + + # Call ffi_prep_args + pushl 12(%ebp) + pushl %eax + call *8(%ebp) + addl $8,%esp + + # Prepare registers + # EAX stores the number of register arguments + cmpl $0, %eax + je .fun + cmpl $3, %eax + jl .prepr_two_cmp + + movl %esp, %ecx + addl $12, %esp + movl 8(%ecx), %eax + jmp .prepr_two +.prepr_two_cmp: + cmpl $2, %eax + jl .prepr_one_prep + movl %esp, %ecx + addl $8, %esp +.prepr_two: + movl 4(%ecx), %edx + jmp .prepr_one +.prepr_one_prep: + movl %esp, %ecx + addl $4, %esp +.prepr_one: + movl (%ecx), %ecx + cmpl $7, 16(%ebp) # FFI_REGISTER + jne .fun + + xchgl %eax, %ecx + +.fun: + # FIXME: Align the stack to a 128-bit boundary to avoid + # potential performance hits. + + # Call function + call *32(%ebp) + + # stdcall functions pop arguments off the stack themselves + + # Load %ecx with the return type code + movl 24(%ebp),%ecx + + # If the return value pointer is NULL, assume no return value. + cmpl $0,28(%ebp) + jne 0f + + # Even if there is no space for the return value, we are + # obliged to handle floating-point values. + cmpl $FFI_TYPE_FLOAT,%ecx + jne .Lnoretval + fstp %st(0) + + jmp .Lepilogue + +0: + call 1f + # Do not insert anything here between the call and the jump table. +.Lstore_table: + .long .Lnoretval-.Lstore_table /* FFI_TYPE_VOID */ + .long .Lretint-.Lstore_table /* FFI_TYPE_INT */ + .long .Lretfloat-.Lstore_table /* FFI_TYPE_FLOAT */ + .long .Lretdouble-.Lstore_table /* FFI_TYPE_DOUBLE */ + .long .Lretlongdouble-.Lstore_table /* FFI_TYPE_LONGDOUBLE */ + .long .Lretuint8-.Lstore_table /* FFI_TYPE_UINT8 */ + .long .Lretsint8-.Lstore_table /* FFI_TYPE_SINT8 */ + .long .Lretuint16-.Lstore_table /* FFI_TYPE_UINT16 */ + .long .Lretsint16-.Lstore_table /* FFI_TYPE_SINT16 */ + .long .Lretint-.Lstore_table /* FFI_TYPE_UINT32 */ + .long .Lretint-.Lstore_table /* FFI_TYPE_SINT32 */ + .long .Lretint64-.Lstore_table /* FFI_TYPE_UINT64 */ + .long .Lretint64-.Lstore_table /* FFI_TYPE_SINT64 */ + .long .Lretstruct-.Lstore_table /* FFI_TYPE_STRUCT */ + .long .Lretint-.Lstore_table /* FFI_TYPE_POINTER */ + .long .Lretstruct1b-.Lstore_table /* FFI_TYPE_SMALL_STRUCT_1B */ + .long .Lretstruct2b-.Lstore_table /* FFI_TYPE_SMALL_STRUCT_2B */ + .long .Lretstruct4b-.Lstore_table /* FFI_TYPE_SMALL_STRUCT_4B */ + .long .Lretstruct-.Lstore_table /* FFI_TYPE_MS_STRUCT */ +1: + shl $2, %ecx + add (%esp),%ecx + mov (%ecx),%ecx + add (%esp),%ecx + add $4, %esp + jmp *%ecx + + /* Sign/zero extend as appropriate. */ +.Lretsint8: + movsbl %al, %eax + jmp .Lretint + +.Lretsint16: + movswl %ax, %eax + jmp .Lretint + +.Lretuint8: + movzbl %al, %eax + jmp .Lretint + +.Lretuint16: + movzwl %ax, %eax + jmp .Lretint + +.Lretint: + # Load %ecx with the pointer to storage for the return value + movl 28(%ebp),%ecx + movl %eax,0(%ecx) + jmp .Lepilogue + +.Lretfloat: + # Load %ecx with the pointer to storage for the return value + movl 28(%ebp),%ecx + fstps (%ecx) + jmp .Lepilogue + +.Lretdouble: + # Load %ecx with the pointer to storage for the return value + movl 28(%ebp),%ecx + fstpl (%ecx) + jmp .Lepilogue + +.Lretlongdouble: + # Load %ecx with the pointer to storage for the return value + movl 28(%ebp),%ecx + fstpt (%ecx) + jmp .Lepilogue + +.Lretint64: + # Load %ecx with the pointer to storage for the return value + movl 28(%ebp),%ecx + movl %eax,0(%ecx) + movl %edx,4(%ecx) + jmp .Lepilogue + +.Lretstruct1b: + # Load %ecx with the pointer to storage for the return value + movl 28(%ebp),%ecx + movb %al,0(%ecx) + jmp .Lepilogue + +.Lretstruct2b: + # Load %ecx with the pointer to storage for the return value + movl 28(%ebp),%ecx + movw %ax,0(%ecx) + jmp .Lepilogue + +.Lretstruct4b: + # Load %ecx with the pointer to storage for the return value + movl 28(%ebp),%ecx + movl %eax,0(%ecx) + jmp .Lepilogue + +.Lretstruct: + # Nothing to do! + +.Lnoretval: +.Lepilogue: + movl %ebp,%esp + popl %ebp + ret +.ffi_call_win32_end: + .balign 16 +FFI_HIDDEN(ffi_closure_THISCALL) + .globl USCORE_SYMBOL(ffi_closure_THISCALL) +#if defined(X86_WIN32) && !defined(__OS2__) + .def _ffi_closure_THISCALL; .scl 2; .type 32; .endef +#endif +USCORE_SYMBOL(ffi_closure_THISCALL): + /* Insert the register argument on the stack as the first argument */ + xchg %ecx, 4(%esp) + xchg %ecx, (%esp) + push %ecx + jmp .ffi_closure_STDCALL_internal + + .balign 16 +FFI_HIDDEN(ffi_closure_FASTCALL) + .globl USCORE_SYMBOL(ffi_closure_FASTCALL) +#if defined(X86_WIN32) && !defined(__OS2__) + .def _ffi_closure_FASTCALL; .scl 2; .type 32; .endef +#endif +USCORE_SYMBOL(ffi_closure_FASTCALL): + /* Insert the 2 register arguments on the stack as the first two arguments */ + xchg %edx, 4(%esp) + xchg %ecx, (%esp) + push %edx + push %ecx + jmp .ffi_closure_STDCALL_internal +FFI_HIDDEN(ffi_closure_REGISTER) + .globl USCORE_SYMBOL(ffi_closure_REGISTER) +#if defined(X86_WIN32) && !defined(__OS2__) + .def _ffi_closure_REGISTER; .scl 2; .type 32; .endef +#endif +USCORE_SYMBOL(ffi_closure_REGISTER): + /* Insert the 3 register arguments on the stack as the first two arguments */ + push %eax + xchg %ecx, 8(%esp) + xchg %edx, 4(%esp) + push %ecx + push %edx + jmp .ffi_closure_STDCALL_internal + +.LFE1: + # This assumes we are using gas. + .balign 16 +FFI_HIDDEN(ffi_closure_SYSV) +#if defined(X86_WIN32) + .globl USCORE_SYMBOL(ffi_closure_SYSV) +#if defined(X86_WIN32) && !defined(__OS2__) + .def _ffi_closure_SYSV; .scl 2; .type 32; .endef +#endif +USCORE_SYMBOL(ffi_closure_SYSV): +#endif +.LFB3: + pushl %ebp +.LCFI4: + movl %esp, %ebp +.LCFI5: + subl $40, %esp + leal -24(%ebp), %edx + movl %edx, -12(%ebp) /* resp */ + leal 8(%ebp), %edx + movl %edx, 4(%esp) /* args = __builtin_dwarf_cfa () */ + leal -12(%ebp), %edx + movl %edx, (%esp) /* &resp */ +#if defined(HAVE_HIDDEN_VISIBILITY_ATTRIBUTE) || !defined(__PIC__) + call USCORE_SYMBOL(ffi_closure_SYSV_inner) +#elif defined(X86_DARWIN) + calll L_ffi_closure_SYSV_inner$stub +#else + movl %ebx, 8(%esp) + call 1f +1: popl %ebx + addl $_GLOBAL_OFFSET_TABLE_+[.-1b], %ebx + call ffi_closure_SYSV_inner@PLT + movl 8(%esp), %ebx +#endif + movl -12(%ebp), %ecx + +0: + call 1f + # Do not insert anything here between the call and the jump table. +.Lcls_store_table: + .long .Lcls_noretval-.Lcls_store_table /* FFI_TYPE_VOID */ + .long .Lcls_retint-.Lcls_store_table /* FFI_TYPE_INT */ + .long .Lcls_retfloat-.Lcls_store_table /* FFI_TYPE_FLOAT */ + .long .Lcls_retdouble-.Lcls_store_table /* FFI_TYPE_DOUBLE */ + .long .Lcls_retldouble-.Lcls_store_table /* FFI_TYPE_LONGDOUBLE */ + .long .Lcls_retuint8-.Lcls_store_table /* FFI_TYPE_UINT8 */ + .long .Lcls_retsint8-.Lcls_store_table /* FFI_TYPE_SINT8 */ + .long .Lcls_retuint16-.Lcls_store_table /* FFI_TYPE_UINT16 */ + .long .Lcls_retsint16-.Lcls_store_table /* FFI_TYPE_SINT16 */ + .long .Lcls_retint-.Lcls_store_table /* FFI_TYPE_UINT32 */ + .long .Lcls_retint-.Lcls_store_table /* FFI_TYPE_SINT32 */ + .long .Lcls_retllong-.Lcls_store_table /* FFI_TYPE_UINT64 */ + .long .Lcls_retllong-.Lcls_store_table /* FFI_TYPE_SINT64 */ + .long .Lcls_retstruct-.Lcls_store_table /* FFI_TYPE_STRUCT */ + .long .Lcls_retint-.Lcls_store_table /* FFI_TYPE_POINTER */ + .long .Lcls_retstruct1-.Lcls_store_table /* FFI_TYPE_SMALL_STRUCT_1B */ + .long .Lcls_retstruct2-.Lcls_store_table /* FFI_TYPE_SMALL_STRUCT_2B */ + .long .Lcls_retstruct4-.Lcls_store_table /* FFI_TYPE_SMALL_STRUCT_4B */ + .long .Lcls_retmsstruct-.Lcls_store_table /* FFI_TYPE_MS_STRUCT */ + +1: + shl $2, %eax + add (%esp),%eax + mov (%eax),%eax + add (%esp),%eax + add $4, %esp + jmp *%eax + + /* Sign/zero extend as appropriate. */ +.Lcls_retsint8: + movsbl (%ecx), %eax + jmp .Lcls_epilogue + +.Lcls_retsint16: + movswl (%ecx), %eax + jmp .Lcls_epilogue + +.Lcls_retuint8: + movzbl (%ecx), %eax + jmp .Lcls_epilogue + +.Lcls_retuint16: + movzwl (%ecx), %eax + jmp .Lcls_epilogue + +.Lcls_retint: + movl (%ecx), %eax + jmp .Lcls_epilogue + +.Lcls_retfloat: + flds (%ecx) + jmp .Lcls_epilogue + +.Lcls_retdouble: + fldl (%ecx) + jmp .Lcls_epilogue + +.Lcls_retldouble: + fldt (%ecx) + jmp .Lcls_epilogue + +.Lcls_retllong: + movl (%ecx), %eax + movl 4(%ecx), %edx + jmp .Lcls_epilogue + +.Lcls_retstruct1: + movsbl (%ecx), %eax + jmp .Lcls_epilogue + +.Lcls_retstruct2: + movswl (%ecx), %eax + jmp .Lcls_epilogue + +.Lcls_retstruct4: + movl (%ecx), %eax + jmp .Lcls_epilogue + +.Lcls_retstruct: + # Caller expects us to pop struct return value pointer hidden arg. + movl %ebp, %esp + popl %ebp + ret $0x4 + +.Lcls_retmsstruct: + # Caller expects us to return a pointer to the real return value. + mov %ecx, %eax + # Caller doesn't expects us to pop struct return value pointer hidden arg. + jmp .Lcls_epilogue + +.Lcls_noretval: +.Lcls_epilogue: + movl %ebp, %esp + popl %ebp + ret +.ffi_closure_SYSV_end: +.LFE3: + +#if !FFI_NO_RAW_API + +#define RAW_CLOSURE_CIF_OFFSET ((FFI_TRAMPOLINE_SIZE + 3) & ~3) +#define RAW_CLOSURE_FUN_OFFSET (RAW_CLOSURE_CIF_OFFSET + 4) +#define RAW_CLOSURE_USER_DATA_OFFSET (RAW_CLOSURE_FUN_OFFSET + 4) + +#ifdef X86_WIN32 + .balign 16 +FFI_HIDDEN(ffi_closure_raw_THISCALL) + .globl USCORE_SYMBOL(ffi_closure_raw_THISCALL) +#if defined(X86_WIN32) && !defined(__OS2__) + .def _ffi_closure_raw_THISCALL; .scl 2; .type 32; .endef +#endif +USCORE_SYMBOL(ffi_closure_raw_THISCALL): + pushl %ebp + movl %esp, %ebp + pushl %esi + subl $36, %esp + movl RAW_CLOSURE_CIF_OFFSET(%eax), %esi /* closure->cif */ + movl RAW_CLOSURE_USER_DATA_OFFSET(%eax), %edx /* closure->user_data */ + movl %edx, 12(%esp) /* user_data */ + leal 12(%ebp), %edx /* __builtin_dwarf_cfa () */ + jmp .stubraw +#endif /* X86_WIN32 */ + + # This assumes we are using gas. + .balign 16 +#if defined(X86_WIN32) + .globl USCORE_SYMBOL(ffi_closure_raw_SYSV) +#if defined(X86_WIN32) && !defined(__OS2__) + .def _ffi_closure_raw_SYSV; .scl 2; .type 32; .endef +#endif +USCORE_SYMBOL(ffi_closure_raw_SYSV): +#endif /* defined(X86_WIN32) */ +.LFB4: + pushl %ebp +.LCFI6: + movl %esp, %ebp +.LCFI7: + pushl %esi +.LCFI8: + subl $36, %esp + movl RAW_CLOSURE_CIF_OFFSET(%eax), %esi /* closure->cif */ + movl RAW_CLOSURE_USER_DATA_OFFSET(%eax), %edx /* closure->user_data */ + movl %edx, 12(%esp) /* user_data */ + leal 8(%ebp), %edx /* __builtin_dwarf_cfa () */ +.stubraw: + movl %edx, 8(%esp) /* raw_args */ + leal -24(%ebp), %edx + movl %edx, 4(%esp) /* &res */ + movl %esi, (%esp) /* cif */ + call *RAW_CLOSURE_FUN_OFFSET(%eax) /* closure->fun */ + movl CIF_FLAGS_OFFSET(%esi), %eax /* rtype */ +0: + call 1f + # Do not insert anything here between the call and the jump table. +.Lrcls_store_table: + .long .Lrcls_noretval-.Lrcls_store_table /* FFI_TYPE_VOID */ + .long .Lrcls_retint-.Lrcls_store_table /* FFI_TYPE_INT */ + .long .Lrcls_retfloat-.Lrcls_store_table /* FFI_TYPE_FLOAT */ + .long .Lrcls_retdouble-.Lrcls_store_table /* FFI_TYPE_DOUBLE */ + .long .Lrcls_retldouble-.Lrcls_store_table /* FFI_TYPE_LONGDOUBLE */ + .long .Lrcls_retuint8-.Lrcls_store_table /* FFI_TYPE_UINT8 */ + .long .Lrcls_retsint8-.Lrcls_store_table /* FFI_TYPE_SINT8 */ + .long .Lrcls_retuint16-.Lrcls_store_table /* FFI_TYPE_UINT16 */ + .long .Lrcls_retsint16-.Lrcls_store_table /* FFI_TYPE_SINT16 */ + .long .Lrcls_retint-.Lrcls_store_table /* FFI_TYPE_UINT32 */ + .long .Lrcls_retint-.Lrcls_store_table /* FFI_TYPE_SINT32 */ + .long .Lrcls_retllong-.Lrcls_store_table /* FFI_TYPE_UINT64 */ + .long .Lrcls_retllong-.Lrcls_store_table /* FFI_TYPE_SINT64 */ + .long .Lrcls_retstruct-.Lrcls_store_table /* FFI_TYPE_STRUCT */ + .long .Lrcls_retint-.Lrcls_store_table /* FFI_TYPE_POINTER */ + .long .Lrcls_retstruct1-.Lrcls_store_table /* FFI_TYPE_SMALL_STRUCT_1B */ + .long .Lrcls_retstruct2-.Lrcls_store_table /* FFI_TYPE_SMALL_STRUCT_2B */ + .long .Lrcls_retstruct4-.Lrcls_store_table /* FFI_TYPE_SMALL_STRUCT_4B */ + .long .Lrcls_retstruct-.Lrcls_store_table /* FFI_TYPE_MS_STRUCT */ +1: + shl $2, %eax + add (%esp),%eax + mov (%eax),%eax + add (%esp),%eax + add $4, %esp + jmp *%eax + + /* Sign/zero extend as appropriate. */ +.Lrcls_retsint8: + movsbl -24(%ebp), %eax + jmp .Lrcls_epilogue + +.Lrcls_retsint16: + movswl -24(%ebp), %eax + jmp .Lrcls_epilogue + +.Lrcls_retuint8: + movzbl -24(%ebp), %eax + jmp .Lrcls_epilogue + +.Lrcls_retuint16: + movzwl -24(%ebp), %eax + jmp .Lrcls_epilogue + +.Lrcls_retint: + movl -24(%ebp), %eax + jmp .Lrcls_epilogue + +.Lrcls_retfloat: + flds -24(%ebp) + jmp .Lrcls_epilogue + +.Lrcls_retdouble: + fldl -24(%ebp) + jmp .Lrcls_epilogue + +.Lrcls_retldouble: + fldt -24(%ebp) + jmp .Lrcls_epilogue + +.Lrcls_retllong: + movl -24(%ebp), %eax + movl -20(%ebp), %edx + jmp .Lrcls_epilogue + +.Lrcls_retstruct1: + movsbl -24(%ebp), %eax + jmp .Lrcls_epilogue + +.Lrcls_retstruct2: + movswl -24(%ebp), %eax + jmp .Lrcls_epilogue + +.Lrcls_retstruct4: + movl -24(%ebp), %eax + jmp .Lrcls_epilogue + +.Lrcls_retstruct: + # Nothing to do! + +.Lrcls_noretval: +.Lrcls_epilogue: + addl $36, %esp + popl %esi + popl %ebp + ret +.ffi_closure_raw_SYSV_end: +.LFE4: + +#endif /* !FFI_NO_RAW_API */ + + # This assumes we are using gas. + .balign 16 +FFI_HIDDEN(ffi_closure_STDCALL) + .globl USCORE_SYMBOL(ffi_closure_STDCALL) +#if defined(X86_WIN32) && !defined(__OS2__) + .def _ffi_closure_STDCALL; .scl 2; .type 32; .endef +#endif +USCORE_SYMBOL(ffi_closure_STDCALL): +.ffi_closure_STDCALL_internal: + /* ffi_closure ctx is at top of the stack */ + movl (%esp), %eax +.LFB5: + pushl %ebp +.LCFI9: + movl %esp, %ebp +.LCFI10: + subl $40, %esp + leal -24(%ebp), %edx + movl %edx, -12(%ebp) /* resp */ + leal 12(%ebp), %edx /* account for stub return address on stack */ + movl %edx, 4(%esp) /* args */ + leal -12(%ebp), %edx + movl %edx, (%esp) /* &resp */ +#if defined(HAVE_HIDDEN_VISIBILITY_ATTRIBUTE) || !defined(__PIC__) + call USCORE_SYMBOL(ffi_closure_WIN32_inner) +#elif defined(X86_DARWIN) + calll L_ffi_closure_WIN32_inner$stub +#else + movl %ebx, 8(%esp) + call 1f +1: popl %ebx + addl $_GLOBAL_OFFSET_TABLE_+[.-1b], %ebx + call ffi_closure_WIN32_inner@PLT + movl 8(%esp), %ebx +#endif + movl -12(%ebp), %ecx +0: + xchgl 4(%ebp), %eax /* xchg size of stack parameters and ffi_closure ctx */ + movl CLOSURE_CIF_OFFSET(%eax), %eax + movl CIF_FLAGS_OFFSET(%eax), %eax + + call 1f + # Do not insert anything here between the call and the jump table. +.Lscls_store_table: + .long .Lscls_noretval-.Lscls_store_table /* FFI_TYPE_VOID */ + .long .Lscls_retint-.Lscls_store_table /* FFI_TYPE_INT */ + .long .Lscls_retfloat-.Lscls_store_table /* FFI_TYPE_FLOAT */ + .long .Lscls_retdouble-.Lscls_store_table /* FFI_TYPE_DOUBLE */ + .long .Lscls_retldouble-.Lscls_store_table /* FFI_TYPE_LONGDOUBLE */ + .long .Lscls_retuint8-.Lscls_store_table /* FFI_TYPE_UINT8 */ + .long .Lscls_retsint8-.Lscls_store_table /* FFI_TYPE_SINT8 */ + .long .Lscls_retuint16-.Lscls_store_table /* FFI_TYPE_UINT16 */ + .long .Lscls_retsint16-.Lscls_store_table /* FFI_TYPE_SINT16 */ + .long .Lscls_retint-.Lscls_store_table /* FFI_TYPE_UINT32 */ + .long .Lscls_retint-.Lscls_store_table /* FFI_TYPE_SINT32 */ + .long .Lscls_retllong-.Lscls_store_table /* FFI_TYPE_UINT64 */ + .long .Lscls_retllong-.Lscls_store_table /* FFI_TYPE_SINT64 */ + .long .Lscls_retstruct-.Lscls_store_table /* FFI_TYPE_STRUCT */ + .long .Lscls_retint-.Lscls_store_table /* FFI_TYPE_POINTER */ + .long .Lscls_retstruct1-.Lscls_store_table /* FFI_TYPE_SMALL_STRUCT_1B */ + .long .Lscls_retstruct2-.Lscls_store_table /* FFI_TYPE_SMALL_STRUCT_2B */ + .long .Lscls_retstruct4-.Lscls_store_table /* FFI_TYPE_SMALL_STRUCT_4B */ +1: + shl $2, %eax + add (%esp),%eax + mov (%eax),%eax + add (%esp),%eax + add $4, %esp + jmp *%eax + + /* Sign/zero extend as appropriate. */ +.Lscls_retsint8: + movsbl (%ecx), %eax + jmp .Lscls_epilogue + +.Lscls_retsint16: + movswl (%ecx), %eax + jmp .Lscls_epilogue + +.Lscls_retuint8: + movzbl (%ecx), %eax + jmp .Lscls_epilogue + +.Lscls_retuint16: + movzwl (%ecx), %eax + jmp .Lscls_epilogue + +.Lscls_retint: + movl (%ecx), %eax + jmp .Lscls_epilogue + +.Lscls_retfloat: + flds (%ecx) + jmp .Lscls_epilogue + +.Lscls_retdouble: + fldl (%ecx) + jmp .Lscls_epilogue + +.Lscls_retldouble: + fldt (%ecx) + jmp .Lscls_epilogue + +.Lscls_retllong: + movl (%ecx), %eax + movl 4(%ecx), %edx + jmp .Lscls_epilogue + +.Lscls_retstruct1: + movsbl (%ecx), %eax + jmp .Lscls_epilogue + +.Lscls_retstruct2: + movswl (%ecx), %eax + jmp .Lscls_epilogue + +.Lscls_retstruct4: + movl (%ecx), %eax + jmp .Lscls_epilogue + +.Lscls_retstruct: + # Nothing to do! + +.Lscls_noretval: +.Lscls_epilogue: + movl %ebp, %esp + popl %ebp + movl 4(%esp), %ecx /* Return address */ + addl (%esp), %esp /* Parameters stack size */ + addl $8, %esp + jmp *%ecx +.ffi_closure_STDCALL_end: +.LFE5: + +#if defined(X86_DARWIN) +.section __IMPORT,__jump_table,symbol_stubs,self_modifying_code+pure_instructions,5 +L_ffi_closure_SYSV_inner$stub: + .indirect_symbol _ffi_closure_SYSV_inner + hlt ; hlt ; hlt ; hlt ; hlt +L_ffi_closure_WIN32_inner$stub: + .indirect_symbol _ffi_closure_WIN32_inner + hlt ; hlt ; hlt ; hlt ; hlt +#endif + +#if defined(X86_WIN32) && !defined(__OS2__) + .section .eh_frame,"w" +#endif +.Lframe1: +.LSCIE1: + .long .LECIE1-.LASCIE1 /* Length of Common Information Entry */ +.LASCIE1: + .long 0x0 /* CIE Identifier Tag */ + .byte 0x1 /* CIE Version */ +#ifdef __PIC__ + .ascii "zR\0" /* CIE Augmentation */ +#else + .ascii "\0" /* CIE Augmentation */ +#endif + .byte 0x1 /* .uleb128 0x1; CIE Code Alignment Factor */ + .byte 0x7c /* .sleb128 -4; CIE Data Alignment Factor */ + .byte 0x8 /* CIE RA Column */ +#ifdef __PIC__ + .byte 0x1 /* .uleb128 0x1; Augmentation size */ + .byte 0x1b /* FDE Encoding (pcrel sdata4) */ +#endif + .byte 0xc /* DW_CFA_def_cfa CFA = r4 + 4 = 4(%esp) */ + .byte 0x4 /* .uleb128 0x4 */ + .byte 0x4 /* .uleb128 0x4 */ + .byte 0x88 /* DW_CFA_offset, column 0x8 %eip at CFA + 1 * -4 */ + .byte 0x1 /* .uleb128 0x1 */ + .align 4 +.LECIE1: + +.LSFDE1: + .long .LEFDE1-.LASFDE1 /* FDE Length */ +.LASFDE1: + .long .LASFDE1-.Lframe1 /* FDE CIE offset */ +#if defined __PIC__ && defined HAVE_AS_X86_PCREL + .long .LFB1-. /* FDE initial location */ +#else + .long .LFB1 +#endif + .long .LFE1-.LFB1 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + /* DW_CFA_xxx CFI instructions go here. */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI0-.LFB1 + .byte 0xe /* DW_CFA_def_cfa_offset CFA = r4 + 8 = 8(%esp) */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 %ebp at CFA + 2 * -4 */ + .byte 0x2 /* .uleb128 0x2 */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI1-.LCFI0 + .byte 0xd /* DW_CFA_def_cfa_register CFA = r5 = %ebp */ + .byte 0x5 /* .uleb128 0x5 */ + + /* End of DW_CFA_xxx CFI instructions. */ + .align 4 +.LEFDE1: + +.LSFDE3: + .long .LEFDE3-.LASFDE3 /* FDE Length */ +.LASFDE3: + .long .LASFDE3-.Lframe1 /* FDE CIE offset */ +#if defined __PIC__ && defined HAVE_AS_X86_PCREL + .long .LFB3-. /* FDE initial location */ +#else + .long .LFB3 +#endif + .long .LFE3-.LFB3 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + /* DW_CFA_xxx CFI instructions go here. */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI4-.LFB3 + .byte 0xe /* DW_CFA_def_cfa_offset CFA = r4 + 8 = 8(%esp) */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 %ebp at CFA + 2 * -4 */ + .byte 0x2 /* .uleb128 0x2 */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI5-.LCFI4 + .byte 0xd /* DW_CFA_def_cfa_register CFA = r5 = %ebp */ + .byte 0x5 /* .uleb128 0x5 */ + + /* End of DW_CFA_xxx CFI instructions. */ + .align 4 +.LEFDE3: + +#if !FFI_NO_RAW_API + +.LSFDE4: + .long .LEFDE4-.LASFDE4 /* FDE Length */ +.LASFDE4: + .long .LASFDE4-.Lframe1 /* FDE CIE offset */ +#if defined __PIC__ && defined HAVE_AS_X86_PCREL + .long .LFB4-. /* FDE initial location */ +#else + .long .LFB4 +#endif + .long .LFE4-.LFB4 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + /* DW_CFA_xxx CFI instructions go here. */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI6-.LFB4 + .byte 0xe /* DW_CFA_def_cfa_offset CFA = r4 + 8 = 8(%esp) */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 %ebp at CFA + 2 * -4 */ + .byte 0x2 /* .uleb128 0x2 */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI7-.LCFI6 + .byte 0xd /* DW_CFA_def_cfa_register CFA = r5 = %ebp */ + .byte 0x5 /* .uleb128 0x5 */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI8-.LCFI7 + .byte 0x86 /* DW_CFA_offset, column 0x6 %esi at CFA + 3 * -4 */ + .byte 0x3 /* .uleb128 0x3 */ + + /* End of DW_CFA_xxx CFI instructions. */ + .align 4 +.LEFDE4: + +#endif /* !FFI_NO_RAW_API */ + +.LSFDE5: + .long .LEFDE5-.LASFDE5 /* FDE Length */ +.LASFDE5: + .long .LASFDE5-.Lframe1 /* FDE CIE offset */ +#if defined __PIC__ && defined HAVE_AS_X86_PCREL + .long .LFB5-. /* FDE initial location */ +#else + .long .LFB5 +#endif + .long .LFE5-.LFB5 /* FDE address range */ +#ifdef __PIC__ + .byte 0x0 /* .uleb128 0x0; Augmentation size */ +#endif + /* DW_CFA_xxx CFI instructions go here. */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI9-.LFB5 + .byte 0xe /* DW_CFA_def_cfa_offset CFA = r4 + 8 = 8(%esp) */ + .byte 0x8 /* .uleb128 0x8 */ + .byte 0x85 /* DW_CFA_offset, column 0x5 %ebp at CFA + 2 * -4 */ + .byte 0x2 /* .uleb128 0x2 */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .long .LCFI10-.LCFI9 + .byte 0xd /* DW_CFA_def_cfa_register CFA = r5 = %ebp */ + .byte 0x5 /* .uleb128 0x5 */ + + /* End of DW_CFA_xxx CFI instructions. */ + .align 4 +.LEFDE5: + +#endif /* !_MSC_VER */ + +#if defined __ELF__ && defined __linux__ + .section .note.GNU-stack,"",@progbits +#endif diff --git a/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win64.S b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win64.S new file mode 100644 index 0000000..347a26a --- /dev/null +++ b/jni/ruby/ext/fiddle/libffi-3.2.1/src/x86/win64.S @@ -0,0 +1,519 @@ +#define LIBFFI_ASM +#include +#include + +/* Constants for ffi_call_win64 */ +#define STACK 0 +#define PREP_ARGS_FN 32 +#define ECIF 40 +#define CIF_BYTES 48 +#define CIF_FLAGS 56 +#define RVALUE 64 +#define FN 72 + +/* ffi_call_win64 (void (*prep_args_fn)(char *, extended_cif *), + extended_cif *ecif, unsigned bytes, unsigned flags, + unsigned *rvalue, void (*fn)()); + */ + +#ifdef _MSC_VER +PUBLIC ffi_call_win64 + +EXTRN __chkstk:NEAR +EXTRN ffi_closure_win64_inner:NEAR + +_TEXT SEGMENT + +;;; ffi_closure_win64 will be called with these registers set: +;;; rax points to 'closure' +;;; r11 contains a bit mask that specifies which of the +;;; first four parameters are float or double +;;; +;;; It must move the parameters passed in registers to their stack location, +;;; call ffi_closure_win64_inner for the actual work, then return the result. +;;; +ffi_closure_win64 PROC FRAME + ;; copy register arguments onto stack + test r11, 1 + jne first_is_float + mov QWORD PTR [rsp+8], rcx + jmp second +first_is_float: + movlpd QWORD PTR [rsp+8], xmm0 + +second: + test r11, 2 + jne second_is_float + mov QWORD PTR [rsp+16], rdx + jmp third +second_is_float: + movlpd QWORD PTR [rsp+16], xmm1 + +third: + test r11, 4 + jne third_is_float + mov QWORD PTR [rsp+24], r8 + jmp fourth +third_is_float: + movlpd QWORD PTR [rsp+24], xmm2 + +fourth: + test r11, 8 + jne fourth_is_float + mov QWORD PTR [rsp+32], r9 + jmp done +fourth_is_float: + movlpd QWORD PTR [rsp+32], xmm3 + +done: + .ALLOCSTACK 40 + sub rsp, 40 + .ENDPROLOG + mov rcx, rax ; context is first parameter + mov rdx, rsp ; stack is second parameter + add rdx, 48 ; point to start of arguments + mov rax, ffi_closure_win64_inner + call rax ; call the real closure function + add rsp, 40 + movd xmm0, rax ; If the closure returned a float, + ; ffi_closure_win64_inner wrote it to rax + ret 0 +ffi_closure_win64 ENDP + +ffi_call_win64 PROC FRAME + ;; copy registers onto stack + mov QWORD PTR [rsp+32], r9 + mov QWORD PTR [rsp+24], r8 + mov QWORD PTR [rsp+16], rdx + mov QWORD PTR [rsp+8], rcx + .PUSHREG rbp + push rbp + .ALLOCSTACK 48 + sub rsp, 48 ; 00000030H + .SETFRAME rbp, 32 + lea rbp, QWORD PTR [rsp+32] + .ENDPROLOG + + mov eax, DWORD PTR CIF_BYTES[rbp] + add rax, 15 + and rax, -16 + call __chkstk + sub rsp, rax + lea rax, QWORD PTR [rsp+32] + mov QWORD PTR STACK[rbp], rax + + mov rdx, QWORD PTR ECIF[rbp] + mov rcx, QWORD PTR STACK[rbp] + call QWORD PTR PREP_ARGS_FN[rbp] + + mov rsp, QWORD PTR STACK[rbp] + + movlpd xmm3, QWORD PTR [rsp+24] + movd r9, xmm3 + + movlpd xmm2, QWORD PTR [rsp+16] + movd r8, xmm2 + + movlpd xmm1, QWORD PTR [rsp+8] + movd rdx, xmm1 + + movlpd xmm0, QWORD PTR [rsp] + movd rcx, xmm0 + + call QWORD PTR FN[rbp] +ret_struct4b$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_SMALL_STRUCT_4B + jne ret_struct2b$ + + mov rcx, QWORD PTR RVALUE[rbp] + mov DWORD PTR [rcx], eax + jmp SHORT ret_void$ + +ret_struct2b$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_SMALL_STRUCT_2B + jne ret_struct1b$ + + mov rcx, QWORD PTR RVALUE[rbp] + mov WORD PTR [rcx], ax + jmp SHORT ret_void$ + +ret_struct1b$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_SMALL_STRUCT_1B + jne ret_uint8$ + + mov rcx, QWORD PTR RVALUE[rbp] + mov BYTE PTR [rcx], al + jmp SHORT ret_void$ + +ret_uint8$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_UINT8 + jne ret_sint8$ + + mov rcx, QWORD PTR RVALUE[rbp] + movzx rax, al + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ + +ret_sint8$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_SINT8 + jne ret_uint16$ + + mov rcx, QWORD PTR RVALUE[rbp] + movsx rax, al + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ + +ret_uint16$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_UINT16 + jne ret_sint16$ + + mov rcx, QWORD PTR RVALUE[rbp] + movzx rax, ax + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ + +ret_sint16$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_SINT16 + jne ret_uint32$ + + mov rcx, QWORD PTR RVALUE[rbp] + movsx rax, ax + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ + +ret_uint32$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_UINT32 + jne ret_sint32$ + + mov rcx, QWORD PTR RVALUE[rbp] + mov eax, eax + mov QWORD PTR [rcx], rax + +ret_void$: + xor rax, rax + + lea rsp, QWORD PTR [rbp+16] + pop rbp + ret 0 + +ret_sint32$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_SINT32 + jne ret_float$ + + mov rcx, QWORD PTR RVALUE[rbp] + cdqe + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ + +ret_float$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_FLOAT + jne SHORT ret_double$ + + mov rax, QWORD PTR RVALUE[rbp] + movss DWORD PTR [rax], xmm0 + jmp SHORT ret_void$ + +ret_double$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_DOUBLE + jne SHORT ret_uint64$ + + mov rax, QWORD PTR RVALUE[rbp] + movlpd QWORD PTR [rax], xmm0 + jmp SHORT ret_void$ + +ret_uint64$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_UINT64 + jne SHORT ret_sint64$ + + mov rcx, QWORD PTR RVALUE[rbp] + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ + +ret_sint64$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_SINT64 + jne SHORT ret_pointer$ + + mov rcx, QWORD PTR RVALUE[rbp] + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ + +ret_pointer$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_POINTER + jne SHORT ret_int$ + + mov rcx, QWORD PTR RVALUE[rbp] + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ + +ret_int$: + cmp DWORD PTR CIF_FLAGS[rbp], FFI_TYPE_INT + jne SHORT ret_void$ + + mov rcx, QWORD PTR RVALUE[rbp] + cdqe + mov QWORD PTR [rcx], rax + jmp SHORT ret_void$ +ffi_call_win64 ENDP +_TEXT ENDS +END + +#else + +#ifdef SYMBOL_UNDERSCORE +#define SYMBOL_NAME(name) _##name +#else +#define SYMBOL_NAME(name) name +#endif + +.text + +.extern SYMBOL_NAME(ffi_closure_win64_inner) + +# ffi_closure_win64 will be called with these registers set: +# rax points to 'closure' +# r11 contains a bit mask that specifies which of the +# first four parameters are float or double +# +# It must move the parameters passed in registers to their stack location, +# call ffi_closure_win64_inner for the actual work, then return the result. +# + .balign 16 + .globl SYMBOL_NAME(ffi_closure_win64) + .seh_proc SYMBOL_NAME(ffi_closure_win64) +SYMBOL_NAME(ffi_closure_win64): + # copy register arguments onto stack + test $1,%r11 + jne .Lfirst_is_float + mov %rcx, 8(%rsp) + jmp .Lsecond +.Lfirst_is_float: + movlpd %xmm0, 8(%rsp) + +.Lsecond: + test $2, %r11 + jne .Lsecond_is_float + mov %rdx, 16(%rsp) + jmp .Lthird +.Lsecond_is_float: + movlpd %xmm1, 16(%rsp) + +.Lthird: + test $4, %r11 + jne .Lthird_is_float + mov %r8,24(%rsp) + jmp .Lfourth +.Lthird_is_float: + movlpd %xmm2, 24(%rsp) + +.Lfourth: + test $8, %r11 + jne .Lfourth_is_float + mov %r9, 32(%rsp) + jmp .Ldone +.Lfourth_is_float: + movlpd %xmm3, 32(%rsp) + +.Ldone: + .seh_stackalloc 40 + sub $40, %rsp + .seh_endprologue + mov %rax, %rcx # context is first parameter + mov %rsp, %rdx # stack is second parameter + add $48, %rdx # point to start of arguments + leaq SYMBOL_NAME(ffi_closure_win64_inner)(%rip), %rax + callq *%rax # call the real closure function + add $40, %rsp + movq %rax, %xmm0 # If the closure returned a float, + # ffi_closure_win64_inner wrote it to rax + retq + .seh_endproc + + .balign 16 + .globl SYMBOL_NAME(ffi_call_win64) + .seh_proc SYMBOL_NAME(ffi_call_win64) +SYMBOL_NAME(ffi_call_win64): + # copy registers onto stack + mov %r9,32(%rsp) + mov %r8,24(%rsp) + mov %rdx,16(%rsp) + mov %rcx,8(%rsp) + .seh_pushreg rbp + push %rbp + .seh_stackalloc 48 + sub $48,%rsp + .seh_setframe rbp, 32 + lea 32(%rsp),%rbp + .seh_endprologue + + mov CIF_BYTES(%rbp),%eax + add $15, %rax + and $-16, %rax + cmpq $0x1000, %rax + jb Lch_done +Lch_probe: + subq $0x1000,%rsp + orl $0x0, (%rsp) + subq $0x1000,%rax + cmpq $0x1000,%rax + ja Lch_probe +Lch_done: + subq %rax, %rsp + orl $0x0, (%rsp) + lea 32(%rsp), %rax + mov %rax, STACK(%rbp) + + mov ECIF(%rbp), %rdx + mov STACK(%rbp), %rcx + callq *PREP_ARGS_FN(%rbp) + + mov STACK(%rbp), %rsp + + movlpd 24(%rsp), %xmm3 + movd %xmm3, %r9 + + movlpd 16(%rsp), %xmm2 + movd %xmm2, %r8 + + movlpd 8(%rsp), %xmm1 + movd %xmm1, %rdx + + movlpd (%rsp), %xmm0 + movd %xmm0, %rcx + + callq *FN(%rbp) +.Lret_struct4b: + cmpl $FFI_TYPE_SMALL_STRUCT_4B, CIF_FLAGS(%rbp) + jne .Lret_struct2b + + mov RVALUE(%rbp), %rcx + mov %eax, (%rcx) + jmp .Lret_void + +.Lret_struct2b: + cmpl $FFI_TYPE_SMALL_STRUCT_2B, CIF_FLAGS(%rbp) + jne .Lret_struct1b + + mov RVALUE(%rbp), %rcx + mov %ax, (%rcx) + jmp .Lret_void + +.Lret_struct1b: + cmpl $FFI_TYPE_SMALL_STRUCT_1B, CIF_FLAGS(%rbp) + jne .Lret_uint8 + + mov RVALUE(%rbp), %rcx + mov %al, (%rcx) + jmp .Lret_void + +.Lret_uint8: + cmpl $FFI_TYPE_UINT8, CIF_FLAGS(%rbp) + jne .Lret_sint8 + + mov RVALUE(%rbp), %rcx + movzbq %al, %rax + movq %rax, (%rcx) + jmp .Lret_void + +.Lret_sint8: + cmpl $FFI_TYPE_SINT8, CIF_FLAGS(%rbp) + jne .Lret_uint16 + + mov RVALUE(%rbp), %rcx + movsbq %al, %rax + movq %rax, (%rcx) + jmp .Lret_void + +.Lret_uint16: + cmpl $FFI_TYPE_UINT16, CIF_FLAGS(%rbp) + jne .Lret_sint16 + + mov RVALUE(%rbp), %rcx + movzwq %ax, %rax + movq %rax, (%rcx) + jmp .Lret_void + +.Lret_sint16: + cmpl $FFI_TYPE_SINT16, CIF_FLAGS(%rbp) + jne .Lret_uint32 + + mov RVALUE(%rbp), %rcx + movswq %ax, %rax + movq %rax, (%rcx) + jmp .Lret_void + +.Lret_uint32: + cmpl $FFI_TYPE_UINT32, CIF_FLAGS(%rbp) + jne .Lret_sint32 + + mov RVALUE(%rbp), %rcx + movl %eax, %eax + movq %rax, (%rcx) + jmp .Lret_void + +.Lret_sint32: + cmpl $FFI_TYPE_SINT32, CIF_FLAGS(%rbp) + jne .Lret_float + + mov RVALUE(%rbp), %rcx + cltq + movq %rax, (%rcx) + jmp .Lret_void + +.Lret_float: + cmpl $FFI_TYPE_FLOAT, CIF_FLAGS(%rbp) + jne .Lret_double + + mov RVALUE(%rbp), %rax + movss %xmm0, (%rax) + jmp .Lret_void + +.Lret_double: + cmpl $FFI_TYPE_DOUBLE, CIF_FLAGS(%rbp) + jne .Lret_uint64 + + mov RVALUE(%rbp), %rax + movlpd %xmm0, (%rax) + jmp .Lret_void + +.Lret_uint64: + cmpl $FFI_TYPE_UINT64, CIF_FLAGS(%rbp) + jne .Lret_sint64 + + mov RVALUE(%rbp), %rcx + mov %rax, (%rcx) + jmp .Lret_void + +.Lret_sint64: + cmpl $FFI_TYPE_SINT64, CIF_FLAGS(%rbp) + jne .Lret_pointer + + mov RVALUE(%rbp), %rcx + mov %rax, (%rcx) + jmp .Lret_void + +.Lret_pointer: + cmpl $FFI_TYPE_POINTER, CIF_FLAGS(%rbp) + jne .Lret_int + + mov RVALUE(%rbp), %rcx + mov %rax, (%rcx) + jmp .Lret_void + +.Lret_int: + cmpl $FFI_TYPE_INT, CIF_FLAGS(%rbp) + jne .Lret_void + + mov RVALUE(%rbp), %rcx + cltq + movq %rax, (%rcx) + jmp .Lret_void + +.Lret_void: + xor %rax, %rax + + lea 16(%rbp), %rsp + pop %rbp + retq + .seh_endproc +#endif /* !_MSC_VER */ + -- cgit v1.2.3