diff options
Diffstat (limited to 'libffi-3.0.4/src/x86')
| -rw-r--r-- | libffi-3.0.4/src/x86/darwin.S | 443 | ||||
| -rw-r--r-- | libffi-3.0.4/src/x86/darwin64.S | 415 | ||||
| -rw-r--r-- | libffi-3.0.4/src/x86/ffi.c | 475 | ||||
| -rw-r--r-- | libffi-3.0.4/src/x86/ffi64.c | 571 | ||||
| -rw-r--r-- | libffi-3.0.4/src/x86/ffitarget.h | 90 | ||||
| -rw-r--r-- | libffi-3.0.4/src/x86/freebsd.S | 458 | ||||
| -rw-r--r-- | libffi-3.0.4/src/x86/sysv.S | 433 | ||||
| -rw-r--r-- | libffi-3.0.4/src/x86/unix64.S | 413 | ||||
| -rw-r--r-- | libffi-3.0.4/src/x86/win32.S | 391 |
9 files changed, 3689 insertions, 0 deletions
diff --git a/libffi-3.0.4/src/x86/darwin.S b/libffi-3.0.4/src/x86/darwin.S new file mode 100644 index 0000000..b0b8cdf --- /dev/null +++ b/libffi-3.0.4/src/x86/darwin.S @@ -0,0 +1,443 @@ +/* ----------------------------------------------------------------------- + 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 <fficonfig.h> +#include <ffi.h> + +.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/libffi-3.0.4/src/x86/darwin64.S b/libffi-3.0.4/src/x86/darwin64.S new file mode 100644 index 0000000..5ba0a5f --- /dev/null +++ b/libffi-3.0.4/src/x86/darwin64.S @@ -0,0 +1,415 @@ +/* ----------------------------------------------------------------------- + darwin64.S - Copyright (c) 2006 Free Software Foundation, 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 <fficonfig.h> +#include <ffi.h> + + .file "darwin64.S" +.text + +/* ffi_call_unix64 (void *args, unsigned long bytes, unsigned flags, + void *raddr, void (*fnaddr)()); + + 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/libffi-3.0.4/src/x86/ffi.c b/libffi-3.0.4/src/x86/ffi.c new file mode 100644 index 0000000..85059d0 --- /dev/null +++ b/libffi-3.0.4/src/x86/ffi.c @@ -0,0 +1,475 @@ +/* ----------------------------------------------------------------------- + ffi.c - Copyright (c) 1996, 1998, 1999, 2001, 2007 Red Hat, Inc. + Copyright (c) 2002 Ranjit Mathew + Copyright (c) 2002 Bo Thorsen + Copyright (c) 2002 Roger Sayle + Copyright (C) 2008 Free Software Foundation, Inc. + + x86 Foreign Function Interface + + Permission is hereby granted, free of charge, to any person obtaining + a copy of this software and associated documentation files (the + ``Software''), to deal in the Software without restriction, including + without limitation the rights to use, copy, modify, merge, publish, + distribute, sublicense, and/or sell copies of the Software, and to + permit persons to whom the Software is furnished to do so, subject to + the following conditions: + + The above copyright notice and this permission notice shall be included + in all copies or substantial portions of the Software. + + THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, + EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + DEALINGS IN THE SOFTWARE. + ----------------------------------------------------------------------- */ + +#ifndef __x86_64__ + +#include <ffi.h> +#include <ffi_common.h> + +#include <stdlib.h> + +/* ffi_prep_args is called by the assembly routine once stack space + has been allocated for the function's arguments */ + +void ffi_prep_args(char *stack, extended_cif *ecif) +{ + register unsigned int i; + register void **p_argv; + register char *argp; + register ffi_type **p_arg; + + argp = stack; + + if (ecif->cif->flags == FFI_TYPE_STRUCT) + { + *(void **) argp = ecif->rvalue; + argp += 4; + } + + p_argv = ecif->avalue; + + for (i = ecif->cif->nargs, p_arg = ecif->cif->arg_types; + i != 0; + i--, p_arg++) + { + size_t z; + + /* Align if necessary */ + if ((sizeof(int) - 1) & (unsigned) argp) + argp = (char *) ALIGN(argp, sizeof(int)); + + z = (*p_arg)->size; + if (z < sizeof(int)) + { + z = sizeof(int); + switch ((*p_arg)->type) + { + case FFI_TYPE_SINT8: + *(signed int *) argp = (signed int)*(SINT8 *)(* p_argv); + break; + + case FFI_TYPE_UINT8: + *(unsigned int *) argp = (unsigned int)*(UINT8 *)(* p_argv); + break; + + case FFI_TYPE_SINT16: + *(signed int *) argp = (signed int)*(SINT16 *)(* p_argv); + break; + + case FFI_TYPE_UINT16: + *(unsigned int *) argp = (unsigned int)*(UINT16 *)(* p_argv); + break; + + case FFI_TYPE_SINT32: + *(signed int *) argp = (signed int)*(SINT32 *)(* p_argv); + break; + + case FFI_TYPE_UINT32: + *(unsigned int *) argp = (unsigned int)*(UINT32 *)(* p_argv); + break; + + case FFI_TYPE_STRUCT: + *(unsigned int *) argp = (unsigned int)*(UINT32 *)(* p_argv); + break; + + default: + FFI_ASSERT(0); + } + } + else + { + memcpy(argp, *p_argv, z); + } + p_argv++; + argp += z; + } + + return; +} + +/* Perform machine dependent cif processing */ +ffi_status ffi_prep_cif_machdep(ffi_cif *cif) +{ + /* Set the return type flag */ + switch (cif->rtype->type) + { + case FFI_TYPE_VOID: +#ifdef X86 + case FFI_TYPE_STRUCT: +#endif +#if defined(X86) || defined(X86_DARWIN) + case FFI_TYPE_UINT8: + case FFI_TYPE_UINT16: + case FFI_TYPE_SINT8: + case FFI_TYPE_SINT16: +#endif + + case FFI_TYPE_SINT64: + case FFI_TYPE_FLOAT: + case FFI_TYPE_DOUBLE: + case FFI_TYPE_LONGDOUBLE: + cif->flags = (unsigned) cif->rtype->type; + break; + + case FFI_TYPE_UINT64: + cif->flags = FFI_TYPE_SINT64; + break; + +#ifndef X86 + case FFI_TYPE_STRUCT: + 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) + { + cif->flags = FFI_TYPE_INT; /* same as int type */ + } + else if (cif->rtype->size == 8) + { + cif->flags = FFI_TYPE_SINT64; /* same as int64 type */ + } + else + { + cif->flags = FFI_TYPE_STRUCT; + } + break; +#endif + + default: + cif->flags = FFI_TYPE_INT; + break; + } + +#ifdef X86_DARWIN + cif->bytes = (cif->bytes + 15) & ~0xF; +#endif + + return FFI_OK; +} + +extern void ffi_call_SYSV(void (*)(char *, extended_cif *), extended_cif *, + unsigned, unsigned, unsigned *, void (*fn)()); + +#ifdef X86_WIN32 +extern void ffi_call_STDCALL(void (*)(char *, extended_cif *), extended_cif *, + unsigned, unsigned, unsigned *, void (*fn)()); + +#endif /* X86_WIN32 */ + +void ffi_call(ffi_cif *cif, void (*fn)(), 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 */ + + if ((rvalue == NULL) && + (cif->flags == FFI_TYPE_STRUCT)) + { + ecif.rvalue = alloca(cif->rtype->size); + } + else + ecif.rvalue = rvalue; + + + switch (cif->abi) + { + case FFI_SYSV: + ffi_call_SYSV(ffi_prep_args, &ecif, cif->bytes, cif->flags, ecif.rvalue, + fn); + break; +#ifdef X86_WIN32 + case FFI_STDCALL: + ffi_call_STDCALL(ffi_prep_args, &ecif, cif->bytes, cif->flags, + ecif.rvalue, fn); + break; +#endif /* X86_WIN32 */ + default: + FFI_ASSERT(0); + break; + } +} + + +/** private members **/ + +static void ffi_prep_incoming_args_SYSV (char *stack, void **ret, + void** args, ffi_cif* cif); +void FFI_HIDDEN ffi_closure_SYSV (ffi_closure *) + __attribute__ ((regparm(1))); +unsigned int FFI_HIDDEN ffi_closure_SYSV_inner (ffi_closure *, void **, void *) + __attribute__ ((regparm(1))); +void FFI_HIDDEN ffi_closure_raw_SYSV (ffi_raw_closure *) + __attribute__ ((regparm(1))); +#ifdef X86_WIN32 +void FFI_HIDDEN ffi_closure_STDCALL (ffi_closure *) + __attribute__ ((regparm(1))); +#endif + +/* This function is jumped to by the trampoline */ + +unsigned int FFI_HIDDEN +ffi_closure_SYSV_inner (closure, respp, args) + 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 re-set RESP to point to the + * structure return address. */ + + ffi_prep_incoming_args_SYSV(args, respp, arg_area, cif); + + (closure->fun) (cif, *respp, arg_area, closure->user_data); + + return cif->flags; +} + +static void +ffi_prep_incoming_args_SYSV(char *stack, void **rvalue, void **avalue, + ffi_cif *cif) +{ + register unsigned int i; + register void **p_argv; + register char *argp; + register ffi_type **p_arg; + + argp = stack; + + if ( cif->flags == FFI_TYPE_STRUCT ) { + *rvalue = *(void **) argp; + argp += 4; + } + + p_argv = avalue; + + for (i = cif->nargs, p_arg = cif->arg_types; (i != 0); i--, p_arg++) + { + size_t z; + + /* Align if necessary */ + if ((sizeof(int) - 1) & (unsigned) argp) { + argp = (char *) ALIGN(argp, sizeof(int)); + } + + z = (*p_arg)->size; + + /* because we're little endian, this is what it turns into. */ + + *p_argv = (void*) argp; + + p_argv++; + argp += z; + } + + return; +} + +/* How to make a trampoline. Derived from gcc/config/i386/i386.c. */ + +#define FFI_INIT_TRAMPOLINE(TRAMP,FUN,CTX) \ +({ unsigned char *__tramp = (unsigned char*)(TRAMP); \ + unsigned int __fun = (unsigned int)(FUN); \ + unsigned int __ctx = (unsigned int)(CTX); \ + unsigned int __dis = __fun - (__ctx + 10); \ + *(unsigned char*) &__tramp[0] = 0xb8; \ + *(unsigned int*) &__tramp[1] = __ctx; /* movl __ctx, %eax */ \ + *(unsigned char *) &__tramp[5] = 0xe9; \ + *(unsigned int*) &__tramp[6] = __dis; /* jmp __fun */ \ + }) + +#define FFI_INIT_TRAMPOLINE_STDCALL(TRAMP,FUN,CTX,SIZE) \ +({ unsigned char *__tramp = (unsigned char*)(TRAMP); \ + unsigned int __fun = (unsigned int)(FUN); \ + unsigned int __ctx = (unsigned int)(CTX); \ + unsigned int __dis = __fun - (__ctx + 10); \ + unsigned short __size = (unsigned short)(SIZE); \ + *(unsigned char*) &__tramp[0] = 0xb8; \ + *(unsigned int*) &__tramp[1] = __ctx; /* movl __ctx, %eax */ \ + *(unsigned char *) &__tramp[5] = 0xe8; \ + *(unsigned int*) &__tramp[6] = __dis; /* call __fun */ \ + *(unsigned char *) &__tramp[10] = 0xc2; \ + *(unsigned short*) &__tramp[11] = __size; /* ret __size */ \ + }) + +/* the cif must already be prep'ed */ + +ffi_status +ffi_prep_closure_loc (ffi_closure* closure, + ffi_cif* cif, + void (*fun)(ffi_cif*,void*,void**,void*), + void *user_data, + void *codeloc) +{ + if (cif->abi == FFI_SYSV) + { + FFI_INIT_TRAMPOLINE (&closure->tramp[0], + &ffi_closure_SYSV, + (void*)closure); + } +#ifdef X86_WIN32 + else if (cif->abi == FFI_STDCALL) + { + FFI_INIT_TRAMPOLINE_STDCALL (&closure->tramp[0], + &ffi_closure_STDCALL, + (void*)closure, cif->bytes); + } +#endif + else + { + return FFI_BAD_ABI; + } + + closure->cif = cif; + closure->user_data = user_data; + closure->fun = fun; + + return FFI_OK; +} + +/* ------- Native raw API support -------------------------------- */ + +#if !FFI_NO_RAW_API + +ffi_status +ffi_prep_raw_closure_loc (ffi_raw_closure* closure, + ffi_cif* cif, + void (*fun)(ffi_cif*,void*,ffi_raw*,void*), + void *user_data, + void *codeloc) +{ + int i; + + if (cif->abi != FFI_SYSV) { + return FFI_BAD_ABI; + } + + // we currently don't support certain kinds of arguments for raw + // closures. This should be implemented by a separate assembly language + // routine, since it would require argument processing, something we + // don't do now for performance. + + for (i = cif->nargs-1; i >= 0; i--) + { + FFI_ASSERT (cif->arg_types[i]->type != FFI_TYPE_STRUCT); + FFI_ASSERT (cif->arg_types[i]->type != FFI_TYPE_LONGDOUBLE); + } + + + FFI_INIT_TRAMPOLINE (&closure->tramp[0], &ffi_closure_raw_SYSV, + codeloc); + + closure->cif = cif; + closure->user_data = user_data; + closure->fun = fun; + + return FFI_OK; +} + +static void +ffi_prep_args_raw(char *stack, extended_cif *ecif) +{ + memcpy (stack, ecif->avalue, ecif->cif->bytes); +} + +/* we borrow this routine from libffi (it must be changed, though, to + * actually call the function passed in the first argument. as of + * libffi-1.20, this is not the case.) + */ + +extern void +ffi_call_SYSV(void (*)(char *, extended_cif *), extended_cif *, unsigned, + unsigned, unsigned *, void (*fn)()); + +#ifdef X86_WIN32 +extern void +ffi_call_STDCALL(void (*)(char *, extended_cif *), extended_cif *, unsigned, + unsigned, unsigned *, void (*fn)()); +#endif /* X86_WIN32 */ + +void +ffi_raw_call(ffi_cif *cif, void (*fn)(), void *rvalue, ffi_raw *fake_avalue) +{ + extended_cif ecif; + void **avalue = (void **)fake_avalue; + + ecif.cif = cif; + ecif.avalue = avalue; + + /* If the return value is a struct and we don't have a return */ + /* value address then we need to make one */ + + if ((rvalue == NULL) && + (cif->rtype->type == FFI_TYPE_STRUCT)) + { + ecif.rvalue = alloca(cif->rtype->size); + } + else + ecif.rvalue = rvalue; + + + switch (cif->abi) + { + case FFI_SYSV: + ffi_call_SYSV(ffi_prep_args_raw, &ecif, cif->bytes, cif->flags, + ecif.rvalue, fn); + break; +#ifdef X86_WIN32 + case FFI_STDCALL: + ffi_call_STDCALL(ffi_prep_args_raw, &ecif, cif->bytes, cif->flags, + ecif.rvalue, fn); + break; +#endif /* X86_WIN32 */ + default: + FFI_ASSERT(0); + break; + } +} + +#endif + +#endif /* __x86_64__ */ diff --git a/libffi-3.0.4/src/x86/ffi64.c b/libffi-3.0.4/src/x86/ffi64.c new file mode 100644 index 0000000..5162f69 --- /dev/null +++ b/libffi-3.0.4/src/x86/ffi64.c @@ -0,0 +1,571 @@ +/* ----------------------------------------------------------------------- + ffi.c - Copyright (c) 2002, 2007 Bo Thorsen <bo@suse.de> + + x86-64 Foreign Function Interface + + Permission is hereby granted, free of charge, to any person obtaining + a copy of this software and associated documentation files (the + ``Software''), to deal in the Software without restriction, including + without limitation the rights to use, copy, modify, merge, publish, + distribute, sublicense, and/or sell copies of the Software, and to + permit persons to whom the Software is furnished to do so, subject to + the following conditions: + + The above copyright notice and this permission notice shall be included + in all copies or substantial portions of the Software. + + THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, + EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + DEALINGS IN THE SOFTWARE. + ----------------------------------------------------------------------- */ + +#include <ffi.h> +#include <ffi_common.h> + +#include <stdlib.h> +#include <stdarg.h> + +#ifdef __x86_64__ + +#define MAX_GPR_REGS 6 +#define MAX_SSE_REGS 8 + +struct register_args +{ + /* Registers for argument passing. */ + UINT64 gpr[MAX_GPR_REGS]; + __int128_t sse[MAX_SSE_REGS]; +}; + +extern void ffi_call_unix64 (void *args, unsigned long bytes, unsigned flags, + void *raddr, void (*fnaddr)(), 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 reformating penalties. + + Similary we play games with INTEGERSI_CLASS to use cheaper SImode moves + whenever possible (upper half does contain padding). */ +enum x86_64_reg_class + { + X86_64_NO_CLASS, + X86_64_INTEGER_CLASS, + X86_64_INTEGERSI_CLASS, + X86_64_SSE_CLASS, + X86_64_SSESF_CLASS, + X86_64_SSEDF_CLASS, + X86_64_SSEUP_CLASS, + X86_64_X87_CLASS, + X86_64_X87UP_CLASS, + X86_64_COMPLEX_X87_CLASS, + X86_64_MEMORY_CLASS + }; + +#define MAX_CLASSES 4 + +#define SSE_CLASS_P(X) ((X) >= X86_64_SSE_CLASS && X <= X86_64_SSEUP_CLASS) + +/* x86-64 register passing implementation. See x86-64 ABI for details. Goal + of this code is to classify each 8bytes of incoming argument by the register + class and assign registers accordingly. */ + +/* Return the union class of CLASS1 and CLASS2. + See the x86-64 PS ABI for details. */ + +static enum x86_64_reg_class +merge_classes (enum x86_64_reg_class class1, enum x86_64_reg_class class2) +{ + /* Rule #1: If both classes are equal, this is the resulting class. */ + if (class1 == class2) + return class1; + + /* Rule #2: If one of the classes is NO_CLASS, the resulting class is + the other class. */ + if (class1 == X86_64_NO_CLASS) + return class2; + if (class2 == X86_64_NO_CLASS) + return class1; + + /* Rule #3: If one of the classes is MEMORY, the result is MEMORY. */ + if (class1 == X86_64_MEMORY_CLASS || class2 == X86_64_MEMORY_CLASS) + return X86_64_MEMORY_CLASS; + + /* Rule #4: If one of the classes is INTEGER, the result is INTEGER. */ + if ((class1 == X86_64_INTEGERSI_CLASS && class2 == X86_64_SSESF_CLASS) + || (class2 == X86_64_INTEGERSI_CLASS && class1 == X86_64_SSESF_CLASS)) + return X86_64_INTEGERSI_CLASS; + if (class1 == X86_64_INTEGER_CLASS || class1 == X86_64_INTEGERSI_CLASS + || class2 == X86_64_INTEGER_CLASS || class2 == X86_64_INTEGERSI_CLASS) + return X86_64_INTEGER_CLASS; + + /* Rule #5: If one of the classes is X87, X87UP, or COMPLEX_X87 class, + MEMORY is used. */ + if (class1 == X86_64_X87_CLASS + || class1 == X86_64_X87UP_CLASS + || class1 == X86_64_COMPLEX_X87_CLASS + || class2 == X86_64_X87_CLASS + || class2 == X86_64_X87UP_CLASS + || class2 == X86_64_COMPLEX_X87_CLASS) + return X86_64_MEMORY_CLASS; + + /* Rule #6: Otherwise class SSE is used. */ + return X86_64_SSE_CLASS; +} + +/* Classify the argument of type TYPE and mode MODE. + CLASSES will be filled by the register class used to pass each word + of the operand. The number of words is returned. In case the parameter + should be passed in memory, 0 is returned. As a special case for zero + sized containers, classes[0] will be NO_CLASS and 1 is returned. + + See the x86-64 PS ABI for details. +*/ +static int +classify_argument (ffi_type *type, enum x86_64_reg_class classes[], + size_t byte_offset) +{ + switch (type->type) + { + case FFI_TYPE_UINT8: + case FFI_TYPE_SINT8: + case FFI_TYPE_UINT16: + case FFI_TYPE_SINT16: + case FFI_TYPE_UINT32: + case FFI_TYPE_SINT32: + case FFI_TYPE_UINT64: + case FFI_TYPE_SINT64: + case FFI_TYPE_POINTER: + if (byte_offset + type->size <= 4) + classes[0] = X86_64_INTEGERSI_CLASS; + else + classes[0] = X86_64_INTEGER_CLASS; + return 1; + case FFI_TYPE_FLOAT: + if (byte_offset == 0) + classes[0] = X86_64_SSESF_CLASS; + else + classes[0] = X86_64_SSE_CLASS; + return 1; + case FFI_TYPE_DOUBLE: + classes[0] = X86_64_SSEDF_CLASS; + return 1; + case FFI_TYPE_LONGDOUBLE: + classes[0] = X86_64_X87_CLASS; + classes[1] = X86_64_X87UP_CLASS; + return 2; + case FFI_TYPE_STRUCT: + { + const int UNITS_PER_WORD = 8; + int words = (type->size + UNITS_PER_WORD - 1) / UNITS_PER_WORD; + ffi_type **ptr; + int i; + enum x86_64_reg_class subclasses[MAX_CLASSES]; + + /* If the struct is larger than 16 bytes, pass it on the stack. */ + if (type->size > 16) + return 0; + + for (i = 0; i < words; i++) + classes[i] = X86_64_NO_CLASS; + + /* Merge the fields of structure. */ + for (ptr = type->elements; *ptr != NULL; ptr++) + { + int num; + + byte_offset = ALIGN (byte_offset, (*ptr)->alignment); + + num = classify_argument (*ptr, subclasses, byte_offset % 8); + if (num == 0) + return 0; + for (i = 0; i < num; i++) + { + int pos = byte_offset / 8; + classes[i + pos] = + merge_classes (subclasses[i], classes[i + pos]); + } + + byte_offset += (*ptr)->size; + } + + /* 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. */ + if (classes[i] == X86_64_SSEUP_CLASS + && (i == 0 || classes[i - 1] != X86_64_SSE_CLASS)) + classes[i] = X86_64_SSE_CLASS; + + /* X86_64_X87UP_CLASS should be preceded by X86_64_X87_CLASS. */ + if (classes[i] == X86_64_X87UP_CLASS + && (i == 0 || classes[i - 1] != X86_64_X87_CLASS)) + classes[i] = X86_64_SSE_CLASS; + } + return words; + } + + default: + FFI_ASSERT(0); + } + return 0; /* Never reached. */ +} + +/* Examine the argument and return set number of register required in each + class. Return zero iff parameter should be passed in memory, otherwise + the number of registers. */ + +static int +examine_argument (ffi_type *type, enum x86_64_reg_class classes[MAX_CLASSES], + _Bool in_return, int *pngpr, int *pnsse) +{ + int i, n, ngpr, nsse; + + n = classify_argument (type, classes, 0); + if (n == 0) + return 0; + + ngpr = nsse = 0; + for (i = 0; i < n; ++i) + switch (classes[i]) + { + case X86_64_INTEGER_CLASS: + case X86_64_INTEGERSI_CLASS: + ngpr++; + break; + case X86_64_SSE_CLASS: + case X86_64_SSESF_CLASS: + case X86_64_SSEDF_CLASS: + nsse++; + break; + case X86_64_NO_CLASS: + case X86_64_SSEUP_CLASS: + break; + case X86_64_X87_CLASS: + case X86_64_X87UP_CLASS: + case X86_64_COMPLEX_X87_CLASS: + return in_return != 0; + default: + abort (); + } + + *pngpr = ngpr; + *pnsse = nsse; + + return n; +} + +/* Perform machine dependent cif processing. */ + +ffi_status +ffi_prep_cif_machdep (ffi_cif *cif) +{ + int gprcount, ssecount, i, avn, n, ngpr, nsse, flags; + enum x86_64_reg_class classes[MAX_CLASSES]; + size_t bytes; + + gprcount = ssecount = 0; + + flags = cif->rtype->type; + if (flags != FFI_TYPE_VOID) + { + n = examine_argument (cif->rtype, classes, 1, &ngpr, &nsse); + if (n == 0) + { + /* The return value is passed in memory. A pointer to that + memory is the first argument. Allocate a register for it. */ + gprcount++; + /* We don't have to do anything in asm for the return. */ + flags = FFI_TYPE_VOID; + } + else if (flags == FFI_TYPE_STRUCT) + { + /* Mark which registers the result appears in. */ + _Bool sse0 = SSE_CLASS_P (classes[0]); + _Bool sse1 = n == 2 && SSE_CLASS_P (classes[1]); + if (sse0 && !sse1) + flags |= 1 << 8; + else if (!sse0 && sse1) + flags |= 1 << 9; + else if (sse0 && sse1) + flags |= 1 << 10; + /* Mark the true size of the structure. */ + flags |= cif->rtype->size << 12; + } + } + + /* Go over all arguments and determine the way they should be passed. + If it's in a register and there is space for it, let that be so. If + not, add it's size to the stack byte count. */ + for (bytes = 0, i = 0, avn = cif->nargs; i < avn; i++) + { + if (examine_argument (cif->arg_types[i], classes, 0, &ngpr, &nsse) == 0 + || gprcount + ngpr > MAX_GPR_REGS + || ssecount + nsse > MAX_SSE_REGS) + { + long align = cif->arg_types[i]->alignment; + + if (align < 8) + align = 8; + + bytes = ALIGN(bytes, align); + bytes += cif->arg_types[i]->size; + } + else + { + gprcount += ngpr; + ssecount += nsse; + } + } + if (ssecount) + flags |= 1 << 11; + cif->flags = flags; + cif->bytes = bytes; + + return FFI_OK; +} + +void +ffi_call (ffi_cif *cif, void (*fn)(), void *rvalue, void **avalue) +{ + enum x86_64_reg_class classes[MAX_CLASSES]; + char *stack, *argp; + ffi_type **arg_types; + int gprcount, ssecount, ngpr, nsse, i, avn; + _Bool ret_in_memory; + struct register_args *reg_args; + + /* Can't call 32-bit mode from 64-bit mode. */ + FFI_ASSERT (cif->abi == FFI_UNIX64); + + /* If the return value is a struct and we don't have a return value + address then we need to make one. Note the setting of flags to + VOID above in ffi_prep_cif_machdep. */ + ret_in_memory = (cif->rtype->type == FFI_TYPE_STRUCT + && (cif->flags & 0xff) == FFI_TYPE_VOID); + if (rvalue == NULL && ret_in_memory) + rvalue = alloca (cif->rtype->size); + + /* Allocate the space for the arguments, plus 4 words of temp space. */ + stack = alloca (sizeof (struct register_args) + cif->bytes + 4*8); + reg_args = (struct register_args *) stack; + argp = stack + sizeof (struct register_args); + + gprcount = ssecount = 0; + + /* If the return value is passed in memory, add the pointer as the + first integer argument. */ + if (ret_in_memory) + reg_args->gpr[gprcount++] = (long) rvalue; + + avn = cif->nargs; + arg_types = cif->arg_types; + + for (i = 0; i < avn; ++i) + { + size_t size = arg_types[i]->size; + int n; + + n = examine_argument (arg_types[i], classes, 0, &ngpr, &nsse); + if (n == 0 + || gprcount + ngpr > MAX_GPR_REGS + || ssecount + nsse > MAX_SSE_REGS) + { + long align = arg_types[i]->alignment; + + /* Stack arguments are *always* at least 8 byte aligned. */ + if (align < 8) + align = 8; + + /* Pass this argument in memory. */ + argp = (void *) ALIGN (argp, align); + memcpy (argp, avalue[i], size); + argp += size; + } + else + { + /* The argument is passed entirely in registers. */ + char *a = (char *) avalue[i]; + int j; + + for (j = 0; j < n; j++, a += 8, size -= 8) + { + switch (classes[j]) + { + case X86_64_INTEGER_CLASS: + case X86_64_INTEGERSI_CLASS: + reg_args->gpr[gprcount] = 0; + memcpy (®_args->gpr[gprcount], a, size < 8 ? size : 8); + gprcount++; + break; + case X86_64_SSE_CLASS: + case X86_64_SSEDF_CLASS: + reg_args->sse[ssecount++] = *(UINT64 *) a; + break; + case X86_64_SSESF_CLASS: + reg_args->sse[ssecount++] = *(UINT32 *) a; + break; + default: + abort(); + } + } + } + } + + ffi_call_unix64 (stack, cif->bytes + sizeof (struct register_args), + cif->flags, rvalue, fn, ssecount); +} + + +extern void ffi_closure_unix64(void); + +ffi_status +ffi_prep_closure_loc (ffi_closure* closure, + ffi_cif* cif, + void (*fun)(ffi_cif*, void*, void**, void*), + void *user_data, + void *codeloc) +{ + volatile unsigned short *tramp; + + tramp = (volatile unsigned short *) &closure->tramp[0]; + + tramp[0] = 0xbb49; /* mov <code>, %r11 */ + *(void * volatile *) &tramp[1] = ffi_closure_unix64; + tramp[5] = 0xba49; /* mov <data>, %r10 */ + *(void * volatile *) &tramp[6] = codeloc; + + /* Set the carry bit iff the function uses any sse registers. + This is clc or stc, together with the first byte of the jmp. */ + tramp[10] = cif->flags & (1 << 11) ? 0x49f9 : 0x49f8; + + tramp[11] = 0xe3ff; /* jmp *%r11 */ + + closure->cif = cif; + closure->fun = fun; + closure->user_data = user_data; + + return FFI_OK; +} + +int +ffi_closure_unix64_inner(ffi_closure *closure, void *rvalue, + struct register_args *reg_args, char *argp) +{ + ffi_cif *cif; + void **avalue; + ffi_type **arg_types; + long i, avn; + int gprcount, ssecount, ngpr, nsse; + int ret; + + cif = closure->cif; + avalue = alloca(cif->nargs * sizeof(void *)); + gprcount = ssecount = 0; + + ret = cif->rtype->type; + if (ret != FFI_TYPE_VOID) + { + enum x86_64_reg_class classes[MAX_CLASSES]; + int n = examine_argument (cif->rtype, classes, 1, &ngpr, &nsse); + if (n == 0) + { + /* The return value goes in memory. Arrange for the closure + return value to go directly back to the original caller. */ + rvalue = (void *) reg_args->gpr[gprcount++]; + /* We don't have to do anything in asm for the return. */ + ret = FFI_TYPE_VOID; + } + else if (ret == FFI_TYPE_STRUCT && n == 2) + { + /* Mark which register the second word of the structure goes in. */ + _Bool sse0 = SSE_CLASS_P (classes[0]); + _Bool sse1 = SSE_CLASS_P (classes[1]); + if (!sse0 && sse1) + ret |= 1 << 8; + else if (sse0 && !sse1) + ret |= 1 << 9; + } + } + + avn = cif->nargs; + arg_types = cif->arg_types; + + for (i = 0; i < avn; ++i) + { + enum x86_64_reg_class classes[MAX_CLASSES]; + int n; + + n = examine_argument (arg_types[i], classes, 0, &ngpr, &nsse); + if (n == 0 + || gprcount + ngpr > MAX_GPR_REGS + || ssecount + nsse > MAX_SSE_REGS) + { + long align = arg_types[i]->alignment; + + /* Stack arguments are *always* at least 8 byte aligned. */ + if (align < 8) + align = 8; + + /* Pass this argument in memory. */ + argp = (void *) ALIGN (argp, align); + avalue[i] = argp; + argp += arg_types[i]->size; + } + /* If the argument is in a single register, or two consecutive + 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/libffi-3.0.4/src/x86/ffitarget.h b/libffi-3.0.4/src/x86/ffitarget.h new file mode 100644 index 0000000..8178d06 --- /dev/null +++ b/libffi-3.0.4/src/x86/ffitarget.h @@ -0,0 +1,90 @@ +/* -----------------------------------------------------------------*-C-*- + ffitarget.h - Copyright (c) 1996-2003 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 + +/* ---- System specific configurations ----------------------------------- */ + +#if defined (X86_64) && defined (__i386__) +#undef X86_64 +#define X86 +#endif + +/* ---- Generic type definitions ----------------------------------------- */ + +#ifndef LIBFFI_ASM +typedef unsigned long ffi_arg; +typedef signed long ffi_sarg; + +typedef enum ffi_abi { + FFI_FIRST_ABI = 0, + + /* ---- Intel x86 Win32 ---------- */ +#ifdef X86_WIN32 + FFI_SYSV, + FFI_STDCALL, + /* TODO: Add fastcall support for the sake of completeness */ + FFI_DEFAULT_ABI = FFI_SYSV, +#endif + + /* ---- Intel x86 and AMD x86-64 - */ +#if !defined(X86_WIN32) && (defined(__i386__) || defined(__x86_64__)) + FFI_SYSV, + FFI_UNIX64, /* Unix variants all use the same ABI for x86-64 */ +#ifdef __i386__ + FFI_DEFAULT_ABI = FFI_SYSV, +#else + FFI_DEFAULT_ABI = FFI_UNIX64, +#endif +#endif + + FFI_LAST_ABI = FFI_DEFAULT_ABI + 1 +} 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) + +#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 13 +#else +#define FFI_TRAMPOLINE_SIZE 10 +#endif +#define FFI_NATIVE_RAW_API 1 /* x86 has native raw api support */ +#endif + +#endif + diff --git a/libffi-3.0.4/src/x86/freebsd.S b/libffi-3.0.4/src/x86/freebsd.S new file mode 100644 index 0000000..afde513 --- /dev/null +++ b/libffi-3.0.4/src/x86/freebsd.S @@ -0,0 +1,458 @@ +/* ----------------------------------------------------------------------- + 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 <fficonfig.h> +#include <ffi.h> + +.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 + + 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__ */ diff --git a/libffi-3.0.4/src/x86/sysv.S b/libffi-3.0.4/src/x86/sysv.S new file mode 100644 index 0000000..de08209 --- /dev/null +++ b/libffi-3.0.4/src/x86/sysv.S @@ -0,0 +1,433 @@ +/* ----------------------------------------------------------------------- + sysv.S - Copyright (c) 1996, 1998, 2001, 2002, 2003, 2005 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 <fficonfig.h> +#include <ffi.h> + +.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 + + 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 + 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_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 + +#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__ */ diff --git a/libffi-3.0.4/src/x86/unix64.S b/libffi-3.0.4/src/x86/unix64.S new file mode 100644 index 0000000..cdc065f --- /dev/null +++ b/libffi-3.0.4/src/x86/unix64.S @@ -0,0 +1,413 @@ +/* ----------------------------------------------------------------------- + unix64.S - Copyright (c) 2002 Bo Thorsen <bo@suse.de> + + x86-64 Foreign Function Interface + + Permission is hereby granted, free of charge, to any person obtaining + a copy of this software and associated documentation files (the + ``Software''), to deal in the Software without restriction, including + without limitation the rights to use, copy, modify, merge, publish, + distribute, sublicense, and/or sell copies of the Software, and to + permit persons to whom the Software is furnished to do so, subject to + the following conditions: + + The above copyright notice and this permission notice shall be included + in all copies or substantial portions of the Software. + + THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, EXPRESS + OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR + ANY CLAIM, DAMAGES OR + OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + OTHER DEALINGS IN THE SOFTWARE. + ----------------------------------------------------------------------- */ + +#ifdef __x86_64__ +#define LIBFFI_ASM +#include <fficonfig.h> +#include <ffi.h> + +.text + +/* ffi_call_unix64 (void *args, unsigned long bytes, unsigned flags, + void *raddr, void (*fnaddr)()); + + 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 + + .section .rodata +.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 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 + + .section .rodata +.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 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 + + .section .eh_frame,"a",@progbits +.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 */ + .long .LUW0-. /* FDE initial location */ + .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 */ + .long .LUW5-. /* FDE initial location */ + .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 /* __x86_64__ */ diff --git a/libffi-3.0.4/src/x86/win32.S b/libffi-3.0.4/src/x86/win32.S new file mode 100644 index 0000000..eba11c1 --- /dev/null +++ b/libffi-3.0.4/src/x86/win32.S @@ -0,0 +1,391 @@ +/* ----------------------------------------------------------------------- + win32.S - Copyright (c) 1996, 1998, 2001, 2002 Red Hat, Inc. + Copyright (c) 2001 John Beniton + Copyright (c) 2002 Ranjit Mathew + + + 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 <fficonfig.h> +#include <ffi.h> + +.text + +.globl ffi_prep_args + + # This assumes we are using gas. + .balign 16 +.globl _ffi_call_SYSV + +_ffi_call_SYSV: + pushl %ebp + movl %esp,%ebp + + # 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 + pushl 12(%ebp) + pushl %eax + call *8(%ebp) + + # Return stack to previous state and call the function + addl $8,%esp + + # FIXME: Align the stack to a 128-bit boundary to avoid + # potential performance hits. + + call *28(%ebp) + + # Remove the space we pushed for the args + movl 16(%ebp),%ecx + addl %ecx,%esp + + # Load %ecx with the return type code + movl 20(%ebp),%ecx + + # If the return value pointer is NULL, assume no return value. + cmpl $0,24(%ebp) + jne retint + + # 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 + +retint: + cmpl $FFI_TYPE_INT,%ecx + jne retfloat + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + jmp epilogue + +retfloat: + cmpl $FFI_TYPE_FLOAT,%ecx + jne retdouble + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + fstps (%ecx) + jmp epilogue + +retdouble: + cmpl $FFI_TYPE_DOUBLE,%ecx + jne retlongdouble + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + fstpl (%ecx) + jmp epilogue + +retlongdouble: + cmpl $FFI_TYPE_LONGDOUBLE,%ecx + jne retint64 + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + fstpt (%ecx) + jmp epilogue + +retint64: + cmpl $FFI_TYPE_SINT64,%ecx + jne retstruct1b + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + movl %edx,4(%ecx) + +retstruct1b: + cmpl $FFI_TYPE_SINT8,%ecx + jne retstruct2b + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + movb %al,0(%ecx) + jmp epilogue + +retstruct2b: + cmpl $FFI_TYPE_SINT16,%ecx + jne retstruct + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + movw %ax,0(%ecx) + jmp epilogue + +retstruct: + # Nothing to do! + +noretval: +epilogue: + movl %ebp,%esp + popl %ebp + ret + +.ffi_call_SYSV_end: + + # This assumes we are using gas. + .balign 16 +.globl _ffi_call_STDCALL + +_ffi_call_STDCALL: + pushl %ebp + movl %esp,%ebp + + # 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 + pushl 12(%ebp) + pushl %eax + call *8(%ebp) + + # Return stack to previous state and call the function + addl $8,%esp + + # FIXME: Align the stack to a 128-bit boundary to avoid + # potential performance hits. + + call *28(%ebp) + + # stdcall functions pop arguments off the stack themselves + + # Load %ecx with the return type code + movl 20(%ebp),%ecx + + # If the return value pointer is NULL, assume no return value. + cmpl $0,24(%ebp) + jne sc_retint + + # Even if there is no space for the return value, we are + # obliged to handle floating-point values. + cmpl $FFI_TYPE_FLOAT,%ecx + jne sc_noretval + fstp %st(0) + + jmp sc_epilogue + +sc_retint: + cmpl $FFI_TYPE_INT,%ecx + jne sc_retfloat + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + jmp sc_epilogue + +sc_retfloat: + cmpl $FFI_TYPE_FLOAT,%ecx + jne sc_retdouble + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + fstps (%ecx) + jmp sc_epilogue + +sc_retdouble: + cmpl $FFI_TYPE_DOUBLE,%ecx + jne sc_retlongdouble + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + fstpl (%ecx) + jmp sc_epilogue + +sc_retlongdouble: + cmpl $FFI_TYPE_LONGDOUBLE,%ecx + jne sc_retint64 + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + fstpt (%ecx) + jmp sc_epilogue + +sc_retint64: + cmpl $FFI_TYPE_SINT64,%ecx + jne sc_retstruct1b + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + movl %eax,0(%ecx) + movl %edx,4(%ecx) + +sc_retstruct1b: + cmpl $FFI_TYPE_SINT8,%ecx + jne sc_retstruct2b + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + movb %al,0(%ecx) + jmp sc_epilogue + +sc_retstruct2b: + cmpl $FFI_TYPE_SINT16,%ecx + jne sc_retstruct + # Load %ecx with the pointer to storage for the return value + movl 24(%ebp),%ecx + movw %ax,0(%ecx) + jmp sc_epilogue + +sc_retstruct: + # Nothing to do! + +sc_noretval: +sc_epilogue: + movl %ebp,%esp + popl %ebp + ret + +.ffi_call_STDCALL_end: + + .globl _ffi_closure_STDCALL +_ffi_closure_STDCALL: + pushl %ebp + movl %esp, %ebp + 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 */ + call _ffi_closure_SYSV_inner + movl -12(%ebp), %ecx + jmp .Lcls_return_result +.ffi_closure_STDCALL_end: + + .globl _ffi_closure_SYSV +_ffi_closure_SYSV: + pushl %ebp + movl %esp, %ebp + 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 */ + call _ffi_closure_SYSV_inner + movl -12(%ebp), %ecx +.Lcls_return_result: + cmpl $FFI_TYPE_INT, %eax + je .Lcls_retint + 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_SINT8, %eax /* 1-byte struct */ + je .Lcls_retstruct1 + cmpl $FFI_TYPE_SINT16, %eax /* 2-bytes struct */ + je .Lcls_retstruct2 +.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_retstruct1: + movsbl (%ecx), %eax + jmp .Lcls_epilogue +.Lcls_retstruct2: + movswl (%ecx), %eax + jmp .Lcls_epilogue +.ffi_closure_SYSV_end: + +#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 + + .balign 16 + .globl _ffi_closure_raw_SYSV +_ffi_closure_raw_SYSV: + 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 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 + 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 +.ffi_closure_raw_SYSV_end: + +#endif |