cb91ac4b68
copy op to shift arguments on the stack instead of transfering each argument one by one through a register. Probably doesn't affect overall operation, but makes the code a little less grotty and easier to update later if I choose to make the wrapper handle more args. Also add comments.
178 lines
6.0 KiB
ArmAsm
178 lines
6.0 KiB
ArmAsm
/*-
|
|
* Copyright (c) 2005
|
|
* Bill Paul <wpaul@windriver.com>. All rights reserved.
|
|
*
|
|
* Redistribution and use in source and binary forms, with or without
|
|
* modification, are permitted provided that the following conditions
|
|
* are met:
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
* notice, this list of conditions and the following disclaimer.
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
* documentation and/or other materials provided with the distribution.
|
|
* 3. All advertising materials mentioning features or use of this software
|
|
* must display the following acknowledgement:
|
|
* This product includes software developed by Bill Paul.
|
|
* 4. Neither the name of the author nor the names of any co-contributors
|
|
* may be used to endorse or promote products derived from this software
|
|
* without specific prior written permission.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY Bill Paul AND CONTRIBUTORS ``AS IS'' AND
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL Bill Paul OR THE VOICES IN HIS HEAD
|
|
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
|
|
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
|
|
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
|
|
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
|
|
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
|
|
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
|
|
* THE POSSIBILITY OF SUCH DAMAGE.
|
|
*
|
|
* The x86_64 callback routines were written and graciously submitted
|
|
* by Ville-Pertti Keinonen <will@exomi.com>.
|
|
*
|
|
* $FreeBSD$
|
|
*/
|
|
|
|
#include <machine/asmacros.h>
|
|
|
|
/*
|
|
* Wrapper for handling up to 16 arguments. We can't really
|
|
* know how many arguments the caller will pass us. I'm taking an
|
|
* educated guess that we'll never get over 16. Handling too
|
|
* few arguments is bad. Handling too many is inefficient, but
|
|
* not fatal. If someone can think of a way to handle an arbitrary
|
|
* number of arguments with more elegant code, freel free to let
|
|
* me know.
|
|
*
|
|
* Standard amd64 calling conventions specify the following registers
|
|
* to be used for passing the first 6 arguments:
|
|
*
|
|
* %rdi, %rsi, %rdx, %rcx, %r8, %r9
|
|
*
|
|
* Further arguments are passed on the stack (the 7th argument is
|
|
* located immediately after the return address).
|
|
*
|
|
* Windows x86_64 calling conventions only pass the first 4
|
|
* arguments in registers:
|
|
*
|
|
* %rcx, %rdx, %r8, %r9
|
|
*
|
|
* Even when arguments are passed in registers, the stack must have
|
|
* space reserved for those arguments. Thus the 5th argument (the
|
|
* first non-register argument) is placed 32 bytes after the return
|
|
* address. Additionally, %rdi and %rsi must be preserved. (These
|
|
* two registers are not scratch registers in the standard convention.)
|
|
*
|
|
* Note that in this template, we load a contrived 64 bit address into
|
|
* %r11 to represent our jump address. This is to guarantee that the
|
|
* assembler leaves enough room to patch in an absolute 64-bit address
|
|
* later. The idea behind this code is that we want to avoid having to
|
|
* manually create all the wrapper functions at compile time with
|
|
* a bunch of macros. This is doable, but a) messy and b) requires
|
|
* us to maintain two separate tables (one for the UNIX function
|
|
* pointers and another with the wrappers). This means I'd have to
|
|
* update two different tables each time I added a function.
|
|
*
|
|
* To avoid this, we create the wrappers at runtime instead. The
|
|
* image patch tables now contain two pointers: one two the normal
|
|
* routine, and a blank one for the wrapper. To construct a wrapper,
|
|
* we allocate some memory and copy the template function into it,
|
|
* then patch the function pointer for the routine we want to wrap
|
|
* into the newly created wrapper. The subr_pe module can then
|
|
* simply patch the wrapper routine into the jump table into the
|
|
* windows image. As a bonus, the wrapper pointer not only serves
|
|
* as the wrapper entry point address, it's also a data pointer
|
|
* that we can pass to free() later when we unload the module.
|
|
*/
|
|
|
|
.globl x86_64_wrap_call
|
|
.globl x86_64_wrap_end
|
|
|
|
ENTRY(x86_64_wrap)
|
|
subq $96,%rsp # allocate space on stack
|
|
mov %rsi,96-8(%rsp) # save %rsi
|
|
mov %rdi,96-16(%rsp)# save %rdi
|
|
mov %rcx,%r10 # temporarily save %rcx in scratch
|
|
mov %rsp,%rsi
|
|
add $96+56,%rsi # source == old stack top (stack+56)
|
|
mov %rsp,%rdi # destination == new stack top
|
|
mov $10,%rcx # count == 10 quadwords
|
|
rep
|
|
movsq # copy old stack contents to new location
|
|
mov %r10,%rdi # set up arg0 (%rcx -> %rdi)
|
|
mov %rdx,%rsi # set up arg1 (%rdx -> %rsi)
|
|
mov %r8,%rdx # set up arg2 (%r8 -> %rdx)
|
|
mov %r9,%rcx # set up arg3 (%r9 -> %rcx)
|
|
mov 96+40(%rsp),%r8 # set up arg4 (stack+40 -> %r8)
|
|
mov 96+48(%rsp),%r9 # set up arg5 (stack+48 -> %r9)
|
|
xor %rax,%rax # clear return value
|
|
x86_64_wrap_call:
|
|
mov $0xFF00FF00FF00FF00,%r11
|
|
callq *%r11 # call routine
|
|
mov 96-16(%rsp),%rdi# restore %rdi
|
|
mov 96-8(%rsp),%rsi # restore %rsi
|
|
addq $96,%rsp # delete space on stack
|
|
ret
|
|
x86_64_wrap_end:
|
|
|
|
/*
|
|
* Functions for invoking x86_64 callbacks. In each case, the first
|
|
* argument is a pointer to the function.
|
|
*/
|
|
|
|
ENTRY(x86_64_call1)
|
|
subq $8,%rsp
|
|
mov %rsi,%rcx
|
|
call *%rdi
|
|
addq $8,%rsp
|
|
ret
|
|
|
|
ENTRY(x86_64_call2)
|
|
subq $24,%rsp
|
|
mov %rsi,%rcx
|
|
/* %rdx is already correct */
|
|
call *%rdi
|
|
addq $24,%rsp
|
|
ret
|
|
|
|
ENTRY(x86_64_call3)
|
|
subq $24,%rsp
|
|
mov %rcx,%r8
|
|
mov %rsi,%rcx
|
|
call *%rdi
|
|
addq $24,%rsp
|
|
ret
|
|
|
|
ENTRY(x86_64_call4)
|
|
subq $40,%rsp
|
|
mov %r8,%r9
|
|
mov %rcx,%r8
|
|
mov %rsi,%rcx
|
|
call *%rdi
|
|
addq $40,%rsp
|
|
ret
|
|
|
|
ENTRY(x86_64_call5)
|
|
subq $40,%rsp
|
|
mov %r9,32(%rsp)
|
|
mov %r8,%r9
|
|
mov %rcx,%r8
|
|
mov %rsi,%rcx
|
|
call *%rdi
|
|
addq $40,%rsp
|
|
ret
|
|
|
|
ENTRY(x86_64_call6)
|
|
subq $56,%rsp
|
|
mov 56+8(%rsp),%rax
|
|
mov %r9,32(%rsp)
|
|
mov %rax,40(%rsp)
|
|
mov %r8,%r9
|
|
mov %rcx,%r8
|
|
mov %rsi,%rcx
|
|
call *%rdi
|
|
addq $56,%rsp
|
|
ret
|