diff --git a/Modules/_ctypes/libffi/include/ffi.h.in b/Modules/_ctypes/libffi/include/ffi.h.in index 1a015ae50f1..7df3b06e15d 100644 --- a/Modules/_ctypes/libffi/include/ffi.h.in +++ b/Modules/_ctypes/libffi/include/ffi.h.in @@ -57,7 +57,9 @@ extern "C" { #endif /* Specify which architecture libffi is configured for. */ +#ifndef @TARGET@ #define @TARGET@ +#endif /* ---- System configuration information --------------------------------- */ diff --git a/Modules/_ctypes/libffi/src/pa/hpux32.S b/Modules/_ctypes/libffi/src/pa/hpux32.S new file mode 100644 index 00000000000..40528bad751 --- /dev/null +++ b/Modules/_ctypes/libffi/src/pa/hpux32.S @@ -0,0 +1,368 @@ +/* ----------------------------------------------------------------------- + hpux32.S - Copyright (c) 2006 Free Software Foundation, Inc. + (c) 2008 Red Hat, Inc. + based on src/pa/linux.S + + HP-UX PA 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 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. + ----------------------------------------------------------------------- */ + +#define LIBFFI_ASM +#include +#include + + .LEVEL 1.1 + .SPACE $PRIVATE$ + .IMPORT $global$,DATA + .IMPORT $$dyncall,MILLICODE + .SUBSPA $DATA$ + .align 4 + + /* void ffi_call_pa32(void (*)(char *, extended_cif *), + extended_cif *ecif, + unsigned bytes, + unsigned flags, + unsigned *rvalue, + void (*fn)(void)); + */ + + .export ffi_call_pa32,ENTRY,PRIV_LEV=3 + .import ffi_prep_args_pa32,CODE + + .SPACE $TEXT$ + .SUBSPA $CODE$ + .align 4 + +L$FB1 +ffi_call_pa32 + .proc + .callinfo FRAME=64,CALLS,SAVE_RP,SAVE_SP,ENTRY_GR=4 + .entry + stw %rp, -20(%sp) + copy %r3, %r1 +L$CFI11 + copy %sp, %r3 +L$CFI12 + + /* Setup the stack for calling prep_args... + We want the stack to look like this: + + [ Previous stack ] <- %r3 + + [ 64-bytes register save area ] <- %r4 + + [ Stack space for actual call, passed as ] <- %arg0 + [ arg0 to ffi_prep_args_pa32 ] + + [ Stack for calling prep_args ] <- %sp + */ + + stwm %r1, 64(%sp) + stw %r4, 12(%r3) +L$CFI13 + copy %sp, %r4 + + addl %arg2, %r4, %arg0 ; arg stack + stw %arg3, -48(%r3) ; save flags we need it later + + /* Call prep_args: + %arg0(stack) -- set up above + %arg1(ecif) -- same as incoming param + %arg2(bytes) -- same as incoming param */ + bl ffi_prep_args_pa32,%r2 + ldo 64(%arg0), %sp + ldo -64(%sp), %sp + + /* now %sp should point where %arg0 was pointing. */ + + /* Load the arguments that should be passed in registers + The fp args are loaded by the prep_args function. */ + ldw -36(%sp), %arg0 + ldw -40(%sp), %arg1 + ldw -44(%sp), %arg2 + ldw -48(%sp), %arg3 + + /* in case the function is going to return a structure + we need to give it a place to put the result. */ + ldw -52(%r3), %ret0 ; %ret0 <- rvalue + ldw -56(%r3), %r22 ; %r22 <- function to call + bl $$dyncall, %r31 ; Call the user function + copy %r31, %rp + + /* Prepare to store the result; we need to recover flags and rvalue. */ + ldw -48(%r3), %r21 ; r21 <- flags + ldw -52(%r3), %r20 ; r20 <- rvalue + + /* Store the result according to the return type. The most + likely types should come first. */ + +L$checkint + comib,<>,n FFI_TYPE_INT, %r21, L$checkint8 + b L$done + stw %ret0, 0(%r20) + +L$checkint8 + comib,<>,n FFI_TYPE_UINT8, %r21, L$checkint16 + b L$done + stb %ret0, 0(%r20) + +L$checkint16 + comib,<>,n FFI_TYPE_UINT16, %r21, L$checkdbl + b L$done + sth %ret0, 0(%r20) + +L$checkdbl + comib,<>,n FFI_TYPE_DOUBLE, %r21, L$checkfloat + b L$done + fstd %fr4,0(%r20) + +L$checkfloat + comib,<>,n FFI_TYPE_FLOAT, %r21, L$checkll + b L$done + fstw %fr4L,0(%r20) + +L$checkll + comib,<>,n FFI_TYPE_UINT64, %r21, L$checksmst2 + stw %ret0, 0(%r20) + b L$done + stw %ret1, 4(%r20) + +L$checksmst2 + comib,<>,n FFI_TYPE_SMALL_STRUCT2, %r21, L$checksmst3 + /* 2-byte structs are returned in ret0 as ????xxyy. */ + extru %ret0, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + b L$done + stb %ret0, 0(%r20) + +L$checksmst3 + comib,<>,n FFI_TYPE_SMALL_STRUCT3, %r21, L$checksmst4 + /* 3-byte structs are returned in ret0 as ??xxyyzz. */ + extru %ret0, 15, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret0, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + b L$done + stb %ret0, 0(%r20) + +L$checksmst4 + comib,<>,n FFI_TYPE_SMALL_STRUCT4, %r21, L$checksmst5 + /* 4-byte structs are returned in ret0 as wwxxyyzz. */ + extru %ret0, 7, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret0, 15, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret0, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + b L$done + stb %ret0, 0(%r20) + +L$checksmst5 + comib,<>,n FFI_TYPE_SMALL_STRUCT5, %r21, L$checksmst6 + /* 5 byte values are returned right justified: + ret0 ret1 + 5: ??????aa bbccddee */ + stbs,ma %ret0, 1(%r20) + extru %ret1, 7, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret1, 15, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret1, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + b L$done + stb %ret1, 0(%r20) + +L$checksmst6 + comib,<>,n FFI_TYPE_SMALL_STRUCT6, %r21, L$checksmst7 + /* 6 byte values are returned right justified: + ret0 ret1 + 6: ????aabb ccddeeff */ + extru %ret0, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + stbs,ma %ret0, 1(%r20) + extru %ret1, 7, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret1, 15, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret1, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + b L$done + stb %ret1, 0(%r20) + +L$checksmst7 + comib,<>,n FFI_TYPE_SMALL_STRUCT7, %r21, L$checksmst8 + /* 7 byte values are returned right justified: + ret0 ret1 + 7: ??aabbcc ddeeffgg */ + extru %ret0, 15, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret0, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + stbs,ma %ret0, 1(%r20) + extru %ret1, 7, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret1, 15, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret1, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + b L$done + stb %ret1, 0(%r20) + +L$checksmst8 + comib,<>,n FFI_TYPE_SMALL_STRUCT8, %r21, L$done + /* 8 byte values are returned right justified: + ret0 ret1 + 8: aabbccdd eeffgghh */ + extru %ret0, 7, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret0, 15, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret0, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + stbs,ma %ret0, 1(%r20) + extru %ret1, 7, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret1, 15, 8, %r22 + stbs,ma %r22, 1(%r20) + extru %ret1, 23, 8, %r22 + stbs,ma %r22, 1(%r20) + stb %ret1, 0(%r20) + +L$done + /* all done, return */ + copy %r4, %sp ; pop arg stack + ldw 12(%r3), %r4 + ldwm -64(%sp), %r3 ; .. and pop stack + ldw -20(%sp), %rp + bv %r0(%rp) + nop + .exit + .procend +L$FE1 + + /* void ffi_closure_pa32(void); + Called with closure argument in %r21 */ + + .SPACE $TEXT$ + .SUBSPA $CODE$ + .export ffi_closure_pa32,ENTRY,PRIV_LEV=3,RTNVAL=GR + .import ffi_closure_inner_pa32,CODE + .align 4 +L$FB2 +ffi_closure_pa32 + .proc + .callinfo FRAME=64,CALLS,SAVE_RP,SAVE_SP,ENTRY_GR=3 + .entry + + stw %rp, -20(%sp) + copy %r3, %r1 +L$CFI21 + copy %sp, %r3 +L$CFI22 + stwm %r1, 64(%sp) + + /* Put arguments onto the stack and call ffi_closure_inner. */ + stw %arg0, -36(%r3) + stw %arg1, -40(%r3) + stw %arg2, -44(%r3) + stw %arg3, -48(%r3) + + copy %r21, %arg0 + bl ffi_closure_inner_pa32, %r2 + copy %r3, %arg1 + ldwm -64(%sp), %r3 + ldw -20(%sp), %rp + ldw -36(%sp), %ret0 + bv %r0(%rp) + ldw -40(%sp), %ret1 + .exit + .procend +L$FE2: + + .SPACE $PRIVATE$ + .SUBSPA $DATA$ + + .align 4 + .EXPORT _GLOBAL__F_ffi_call_pa32,DATA +_GLOBAL__F_ffi_call_pa32 +L$frame1: + .word L$ECIE1-L$SCIE1 ;# Length of Common Information Entry +L$SCIE1: + .word 0x0 ;# CIE Identifier Tag + .byte 0x1 ;# CIE Version + .ascii "\0" ;# CIE Augmentation + .uleb128 0x1 ;# CIE Code Alignment Factor + .sleb128 4 ;# CIE Data Alignment Factor + .byte 0x2 ;# CIE RA Column + .byte 0xc ;# DW_CFA_def_cfa + .uleb128 0x1e + .uleb128 0x0 + .align 4 +L$ECIE1: +L$SFDE1: + .word L$EFDE1-L$ASFDE1 ;# FDE Length +L$ASFDE1: + .word L$ASFDE1-L$frame1 ;# FDE CIE offset + .word L$FB1 ;# FDE initial location + .word L$FE1-L$FB1 ;# FDE address range + + .byte 0x4 ;# DW_CFA_advance_loc4 + .word L$CFI11-L$FB1 + .byte 0x83 ;# DW_CFA_offset, column 0x3 + .uleb128 0x0 + .byte 0x11 ;# DW_CFA_offset_extended_sf; save r2 at [r30-20] + .uleb128 0x2 + .sleb128 -5 + + .byte 0x4 ;# DW_CFA_advance_loc4 + .word L$CFI12-L$CFI11 + .byte 0xd ;# DW_CFA_def_cfa_register = r3 + .uleb128 0x3 + + .byte 0x4 ;# DW_CFA_advance_loc4 + .word L$CFI13-L$CFI12 + .byte 0x84 ;# DW_CFA_offset, column 0x4 + .uleb128 0x3 + + .align 4 +L$EFDE1: + +L$SFDE2: + .word L$EFDE2-L$ASFDE2 ;# FDE Length +L$ASFDE2: + .word L$ASFDE2-L$frame1 ;# FDE CIE offset + .word L$FB2 ;# FDE initial location + .word L$FE2-L$FB2 ;# FDE address range + .byte 0x4 ;# DW_CFA_advance_loc4 + .word L$CFI21-L$FB2 + .byte 0x83 ;# DW_CFA_offset, column 0x3 + .uleb128 0x0 + .byte 0x11 ;# DW_CFA_offset_extended_sf + .uleb128 0x2 + .sleb128 -5 + + .byte 0x4 ;# DW_CFA_advance_loc4 + .word L$CFI22-L$CFI21 + .byte 0xd ;# DW_CFA_def_cfa_register = r3 + .uleb128 0x3 + + .align 4 +L$EFDE2: diff --git a/Modules/_ctypes/libffi/src/x86/darwin64.S b/Modules/_ctypes/libffi/src/x86/darwin64.S new file mode 100644 index 00000000000..2f7394ef4bf --- /dev/null +++ b/Modules/_ctypes/libffi/src/x86/darwin64.S @@ -0,0 +1,416 @@ +/* ----------------------------------------------------------------------- + darwin64.S - Copyright (c) 2006 Free Software Foundation, Inc. + Copyright (c) 2008 Red Hat, Inc. + derived from unix64.S + + x86-64 Foreign Function Interface for Darwin. + + Permission is hereby granted, free of charge, to any person obtaining + a copy of this software and associated documentation files (the + ``Software''), to deal in the Software without restriction, including + without limitation the rights to use, copy, modify, merge, publish, + distribute, sublicense, and/or sell copies of the Software, and to + permit persons to whom the Software is furnished to do so, subject to + the following conditions: + + The above copyright notice and this permission notice shall be included + in all copies or substantial portions of the Software. + + THE SOFTWARE IS PROVIDED ``AS IS'', WITHOUT WARRANTY OF ANY KIND, EXPRESS + OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF + MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. + IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY CLAIM, DAMAGES OR + OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + OTHER DEALINGS IN THE SOFTWARE. + ----------------------------------------------------------------------- */ + +#ifdef __x86_64__ +#define LIBFFI_ASM +#include +#include + + .file "darwin64.S" +.text + +/* ffi_call_unix64 (void *args, unsigned long bytes, unsigned flags, + void *raddr, void (*fnaddr)(void)); + + Bit o trickiness here -- ARGS+BYTES is the base of the stack frame + for this function. This has been allocated by ffi_call. We also + deallocate some of the stack that has been alloca'd. */ + + .align 3 + .globl _ffi_call_unix64 + +_ffi_call_unix64: +LUW0: + movq (%rsp), %r10 /* Load return address. */ + leaq (%rdi, %rsi), %rax /* Find local stack base. */ + movq %rdx, (%rax) /* Save flags. */ + movq %rcx, 8(%rax) /* Save raddr. */ + movq %rbp, 16(%rax) /* Save old frame pointer. */ + movq %r10, 24(%rax) /* Relocate return address. */ + movq %rax, %rbp /* Finalize local stack frame. */ +LUW1: + movq %rdi, %r10 /* Save a copy of the register area. */ + movq %r8, %r11 /* Save a copy of the target fn. */ + movl %r9d, %eax /* Set number of SSE registers. */ + + /* Load up all argument registers. */ + movq (%r10), %rdi + movq 8(%r10), %rsi + movq 16(%r10), %rdx + movq 24(%r10), %rcx + movq 32(%r10), %r8 + movq 40(%r10), %r9 + testl %eax, %eax + jnz Lload_sse +Lret_from_load_sse: + + /* Deallocate the reg arg area. */ + leaq 176(%r10), %rsp + + /* Call the user function. */ + call *%r11 + + /* Deallocate stack arg area; local stack frame in redzone. */ + leaq 24(%rbp), %rsp + + movq 0(%rbp), %rcx /* Reload flags. */ + movq 8(%rbp), %rdi /* Reload raddr. */ + movq 16(%rbp), %rbp /* Reload old frame pointer. */ +LUW2: + + /* The first byte of the flags contains the FFI_TYPE. */ + movzbl %cl, %r10d + leaq Lstore_table(%rip), %r11 + movslq (%r11, %r10, 4), %r10 + addq %r11, %r10 + jmp *%r10 + +Lstore_table: + .long Lst_void-Lstore_table /* FFI_TYPE_VOID */ + .long Lst_sint32-Lstore_table /* FFI_TYPE_INT */ + .long Lst_float-Lstore_table /* FFI_TYPE_FLOAT */ + .long Lst_double-Lstore_table /* FFI_TYPE_DOUBLE */ + .long Lst_ldouble-Lstore_table /* FFI_TYPE_LONGDOUBLE */ + .long Lst_uint8-Lstore_table /* FFI_TYPE_UINT8 */ + .long Lst_sint8-Lstore_table /* FFI_TYPE_SINT8 */ + .long Lst_uint16-Lstore_table /* FFI_TYPE_UINT16 */ + .long Lst_sint16-Lstore_table /* FFI_TYPE_SINT16 */ + .long Lst_uint32-Lstore_table /* FFI_TYPE_UINT32 */ + .long Lst_sint32-Lstore_table /* FFI_TYPE_SINT32 */ + .long Lst_int64-Lstore_table /* FFI_TYPE_UINT64 */ + .long Lst_int64-Lstore_table /* FFI_TYPE_SINT64 */ + .long Lst_struct-Lstore_table /* FFI_TYPE_STRUCT */ + .long Lst_int64-Lstore_table /* FFI_TYPE_POINTER */ + + .text + .align 3 +Lst_void: + ret + .align 3 +Lst_uint8: + movzbq %al, %rax + movq %rax, (%rdi) + ret + .align 3 +Lst_sint8: + movsbq %al, %rax + movq %rax, (%rdi) + ret + .align 3 +Lst_uint16: + movzwq %ax, %rax + movq %rax, (%rdi) + .align 3 +Lst_sint16: + movswq %ax, %rax + movq %rax, (%rdi) + ret + .align 3 +Lst_uint32: + movl %eax, %eax + movq %rax, (%rdi) + .align 3 +Lst_sint32: + cltq + movq %rax, (%rdi) + ret + .align 3 +Lst_int64: + movq %rax, (%rdi) + ret + .align 3 +Lst_float: + movss %xmm0, (%rdi) + ret + .align 3 +Lst_double: + movsd %xmm0, (%rdi) + ret +Lst_ldouble: + fstpt (%rdi) + ret + .align 3 +Lst_struct: + leaq -20(%rsp), %rsi /* Scratch area in redzone. */ + + /* We have to locate the values now, and since we don't want to + write too much data into the user's return value, we spill the + value to a 16 byte scratch area first. Bits 8, 9, and 10 + control where the values are located. Only one of the three + bits will be set; see ffi_prep_cif_machdep for the pattern. */ + movd %xmm0, %r10 + movd %xmm1, %r11 + testl $0x100, %ecx + cmovnz %rax, %rdx + cmovnz %r10, %rax + testl $0x200, %ecx + cmovnz %r10, %rdx + testl $0x400, %ecx + cmovnz %r10, %rax + cmovnz %r11, %rdx + movq %rax, (%rsi) + movq %rdx, 8(%rsi) + + /* Bits 12-31 contain the true size of the structure. Copy from + the scratch area to the true destination. */ + shrl $12, %ecx + rep movsb + ret + + /* Many times we can avoid loading any SSE registers at all. + It's not worth an indirect jump to load the exact set of + SSE registers needed; zero or all is a good compromise. */ + .align 3 +LUW3: +Lload_sse: + movdqa 48(%r10), %xmm0 + movdqa 64(%r10), %xmm1 + movdqa 80(%r10), %xmm2 + movdqa 96(%r10), %xmm3 + movdqa 112(%r10), %xmm4 + movdqa 128(%r10), %xmm5 + movdqa 144(%r10), %xmm6 + movdqa 160(%r10), %xmm7 + jmp Lret_from_load_sse + +LUW4: + .align 3 + .globl _ffi_closure_unix64 + +_ffi_closure_unix64: +LUW5: + /* The carry flag is set by the trampoline iff SSE registers + are used. Don't clobber it before the branch instruction. */ + leaq -200(%rsp), %rsp +LUW6: + movq %rdi, (%rsp) + movq %rsi, 8(%rsp) + movq %rdx, 16(%rsp) + movq %rcx, 24(%rsp) + movq %r8, 32(%rsp) + movq %r9, 40(%rsp) + jc Lsave_sse +Lret_from_save_sse: + + movq %r10, %rdi + leaq 176(%rsp), %rsi + movq %rsp, %rdx + leaq 208(%rsp), %rcx + call _ffi_closure_unix64_inner + + /* Deallocate stack frame early; return value is now in redzone. */ + addq $200, %rsp +LUW7: + + /* The first byte of the return value contains the FFI_TYPE. */ + movzbl %al, %r10d + leaq Lload_table(%rip), %r11 + movslq (%r11, %r10, 4), %r10 + addq %r11, %r10 + jmp *%r10 + +Lload_table: + .long Lld_void-Lload_table /* FFI_TYPE_VOID */ + .long Lld_int32-Lload_table /* FFI_TYPE_INT */ + .long Lld_float-Lload_table /* FFI_TYPE_FLOAT */ + .long Lld_double-Lload_table /* FFI_TYPE_DOUBLE */ + .long Lld_ldouble-Lload_table /* FFI_TYPE_LONGDOUBLE */ + .long Lld_int8-Lload_table /* FFI_TYPE_UINT8 */ + .long Lld_int8-Lload_table /* FFI_TYPE_SINT8 */ + .long Lld_int16-Lload_table /* FFI_TYPE_UINT16 */ + .long Lld_int16-Lload_table /* FFI_TYPE_SINT16 */ + .long Lld_int32-Lload_table /* FFI_TYPE_UINT32 */ + .long Lld_int32-Lload_table /* FFI_TYPE_SINT32 */ + .long Lld_int64-Lload_table /* FFI_TYPE_UINT64 */ + .long Lld_int64-Lload_table /* FFI_TYPE_SINT64 */ + .long Lld_struct-Lload_table /* FFI_TYPE_STRUCT */ + .long Lld_int64-Lload_table /* FFI_TYPE_POINTER */ + + .text + .align 3 +Lld_void: + ret + .align 3 +Lld_int8: + movzbl -24(%rsp), %eax + ret + .align 3 +Lld_int16: + movzwl -24(%rsp), %eax + ret + .align 3 +Lld_int32: + movl -24(%rsp), %eax + ret + .align 3 +Lld_int64: + movq -24(%rsp), %rax + ret + .align 3 +Lld_float: + movss -24(%rsp), %xmm0 + ret + .align 3 +Lld_double: + movsd -24(%rsp), %xmm0 + ret + .align 3 +Lld_ldouble: + fldt -24(%rsp) + ret + .align 3 +Lld_struct: + /* There are four possibilities here, %rax/%rdx, %xmm0/%rax, + %rax/%xmm0, %xmm0/%xmm1. We collapse two by always loading + both rdx and xmm1 with the second word. For the remaining, + bit 8 set means xmm0 gets the second word, and bit 9 means + that rax gets the second word. */ + movq -24(%rsp), %rcx + movq -16(%rsp), %rdx + movq -16(%rsp), %xmm1 + testl $0x100, %eax + cmovnz %rdx, %rcx + movd %rcx, %xmm0 + testl $0x200, %eax + movq -24(%rsp), %rax + cmovnz %rdx, %rax + ret + + /* See the comment above Lload_sse; the same logic applies here. */ + .align 3 +LUW8: +Lsave_sse: + movdqa %xmm0, 48(%rsp) + movdqa %xmm1, 64(%rsp) + movdqa %xmm2, 80(%rsp) + movdqa %xmm3, 96(%rsp) + movdqa %xmm4, 112(%rsp) + movdqa %xmm5, 128(%rsp) + movdqa %xmm6, 144(%rsp) + movdqa %xmm7, 160(%rsp) + jmp Lret_from_save_sse + +LUW9: +.section __TEXT,__eh_frame,coalesced,no_toc+strip_static_syms+live_support +EH_frame1: + .set L$set$0,LECIE1-LSCIE1 /* CIE Length */ + .long L$set$0 +LSCIE1: + .long 0x0 /* CIE Identifier Tag */ + .byte 0x1 /* CIE Version */ + .ascii "zR\0" /* CIE Augmentation */ + .byte 0x1 /* uleb128 0x1; CIE Code Alignment Factor */ + .byte 0x78 /* sleb128 -8; CIE Data Alignment Factor */ + .byte 0x10 /* CIE RA Column */ + .byte 0x1 /* uleb128 0x1; Augmentation size */ + .byte 0x10 /* FDE Encoding (pcrel sdata4) */ + .byte 0xc /* DW_CFA_def_cfa, %rsp offset 8 */ + .byte 0x7 /* uleb128 0x7 */ + .byte 0x8 /* uleb128 0x8 */ + .byte 0x90 /* DW_CFA_offset, column 0x10 */ + .byte 0x1 + .align 3 +LECIE1: + .globl _ffi_call_unix64.eh +_ffi_call_unix64.eh: +LSFDE1: + .set L$set$1,LEFDE1-LASFDE1 /* FDE Length */ + .long L$set$1 +LASFDE1: + .long LASFDE1-EH_frame1 /* FDE CIE offset */ + .quad LUW0-. /* FDE initial location */ + .set L$set$2,LUW4-LUW0 /* FDE address range */ + .quad L$set$2 + .byte 0x0 /* Augmentation size */ + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$3,LUW1-LUW0 + .long L$set$3 + + /* New stack frame based off rbp. This is a itty bit of unwind + trickery in that the CFA *has* changed. There is no easy way + to describe it correctly on entry to the function. Fortunately, + it doesn't matter too much since at all points we can correctly + unwind back to ffi_call. Note that the location to which we + moved the return address is (the new) CFA-8, so from the + perspective of the unwind info, it hasn't moved. */ + .byte 0xc /* DW_CFA_def_cfa, %rbp offset 32 */ + .byte 0x6 + .byte 0x20 + .byte 0x80+6 /* DW_CFA_offset, %rbp offset 2*-8 */ + .byte 0x2 + .byte 0xa /* DW_CFA_remember_state */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$4,LUW2-LUW1 + .long L$set$4 + .byte 0xc /* DW_CFA_def_cfa, %rsp offset 8 */ + .byte 0x7 + .byte 0x8 + .byte 0xc0+6 /* DW_CFA_restore, %rbp */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$5,LUW3-LUW2 + .long L$set$5 + .byte 0xb /* DW_CFA_restore_state */ + + .align 3 +LEFDE1: + .globl _ffi_closure_unix64.eh +_ffi_closure_unix64.eh: +LSFDE3: + .set L$set$6,LEFDE3-LASFDE3 /* FDE Length */ + .long L$set$6 +LASFDE3: + .long LASFDE3-EH_frame1 /* FDE CIE offset */ + .quad LUW5-. /* FDE initial location */ + .set L$set$7,LUW9-LUW5 /* FDE address range */ + .quad L$set$7 + .byte 0x0 /* Augmentation size */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$8,LUW6-LUW5 + .long L$set$8 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 208,1 /* uleb128 208 */ + .byte 0xa /* DW_CFA_remember_state */ + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$9,LUW7-LUW6 + .long L$set$9 + .byte 0xe /* DW_CFA_def_cfa_offset */ + .byte 0x8 + + .byte 0x4 /* DW_CFA_advance_loc4 */ + .set L$set$10,LUW8-LUW7 + .long L$set$10 + .byte 0xb /* DW_CFA_restore_state */ + + .align 3 +LEFDE3: + .subsections_via_symbols + +#endif /* __x86_64__ */ diff --git a/Modules/_ctypes/libffi/src/x86/ffi.c b/Modules/_ctypes/libffi/src/x86/ffi.c index 63c99990c39..767effb8380 100644 --- a/Modules/_ctypes/libffi/src/x86/ffi.c +++ b/Modules/_ctypes/libffi/src/x86/ffi.c @@ -349,14 +349,14 @@ ffi_prep_closure_loc (ffi_closure* closure, { FFI_INIT_TRAMPOLINE (&closure->tramp[0], &ffi_closure_SYSV, - (void*)closure); + (void*)codeloc); } #ifdef X86_WIN32 else if (cif->abi == FFI_STDCALL) { FFI_INIT_TRAMPOLINE_STDCALL (&closure->tramp[0], &ffi_closure_STDCALL, - (void*)closure, cif->bytes); + (void*)codeloc, cif->bytes); } #endif else diff --git a/Modules/_ctypes/libffi/src/x86/freebsd.S b/Modules/_ctypes/libffi/src/x86/freebsd.S new file mode 100644 index 00000000000..afde513164e --- /dev/null +++ b/Modules/_ctypes/libffi/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 +#include + +.text + +.globl ffi_prep_args + + .align 4 +.globl ffi_call_SYSV + .type ffi_call_SYSV,@function + +ffi_call_SYSV: +.LFB1: + pushl %ebp +.LCFI0: + movl %esp,%ebp +.LCFI1: + /* Make room for all of the new args. */ + movl 16(%ebp),%ecx + subl %ecx,%esp + + 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__ */