src/x86/win64.S: Support compiling on non-WIN64 platforms

Non-WIN64 versions of the GNU assembler don't support the .seh_*
directives for structured exception handling, so wrap them in a macro
that compiles to nothing.

Handle the registers used for the non-Windows x86-64 calling convention
when on a non-Windows platform.  Distinguish between cases that should
refer to the native argument registers (defined as arg0, arg1, arg2, and
arg3) and cases that should always refer to the Windows argument
registers.

src/x86/win64.S

@@ -7,10 +7,19 @@
         .cfi_sections   .debug_frame
 #endif
 
+#ifdef X86_WIN64
+#define SEH(...) __VA_ARGS__
 #define arg0	%rcx
 #define arg1	%rdx
 #define arg2	%r8
 #define arg3	%r9
+#else
+#define SEH(...)
+#define arg0	%rdi
+#define arg1	%rsi
+#define arg2	%rdx
+#define arg3	%rcx
+#endif
 
 #ifdef SYMBOL_UNDERSCORE
 #define SYMBOL_NAME(name) _##name
@@ -34,7 +43,7 @@
 	.align	8
 	.globl	ffi_call_win64
 
-	.seh_proc ffi_call_win64
+	SEH(.seh_proc ffi_call_win64)
 ffi_call_win64:
 	cfi_startproc
 	/* Set up the local stack frame and install it in rbp/rsp.  */
@@ -44,9 +53,9 @@ ffi_call_win64:
 	movq	arg1, %rbp
 	cfi_def_cfa(%rbp, 16)
 	cfi_rel_offset(%rbp, 0)
-	.seh_pushreg %rbp
+	SEH(.seh_pushreg %rbp)
-	.seh_setframe %rbp, 0
+	SEH(.seh_setframe %rbp, 0)
-	.seh_endprologue
+	SEH(.seh_endprologue)
 	movq	arg0, %rsp
 
 	movq	arg2, %r10
@@ -149,7 +158,7 @@ E FFI_TYPE_SMALL_STRUCT_4B
 .purgem epilogue
 
 	cfi_endproc
-	.seh_endproc
+	SEH(.seh_endproc)
 
 
 /* 32 bytes of outgoing register stack space, 8 bytes of alignment,
@@ -161,33 +170,33 @@ E FFI_TYPE_SMALL_STRUCT_4B
 	.align	8
 	.globl	ffi_go_closure_win64
 
-	.seh_proc ffi_go_closure_win64
+	SEH(.seh_proc ffi_go_closure_win64)
 ffi_go_closure_win64:
 	cfi_startproc
 	/* Save all integer arguments into the incoming reg stack space.  */
-	movq	arg0, 8(%rsp)
+	movq	%rcx, 8(%rsp)
-	movq	arg1, 16(%rsp)
+	movq	%rdx, 16(%rsp)
-	movq	arg2, 24(%rsp)
+	movq	%r8, 24(%rsp)
-	movq	arg3, 32(%rsp)
+	movq	%r9, 32(%rsp)
 
 	movq	8(%r10), arg0			/* load cif */
 	movq	16(%r10), arg1			/* load fun */
 	movq	%r10, arg2			/* closure is user_data */
 	jmp	0f
 	cfi_endproc
-	.seh_endproc
+	SEH(.seh_endproc)
 
 	.align	8
 	.globl	ffi_closure_win64
 
-	.seh_proc ffi_closure_win64
+	SEH(.seh_proc ffi_closure_win64)
 ffi_closure_win64:
 	cfi_startproc
 	/* Save all integer arguments into the incoming reg stack space.  */
-	movq	arg0, 8(%rsp)
+	movq	%rcx, 8(%rsp)
-	movq	arg1, 16(%rsp)
+	movq	%rdx, 16(%rsp)
-	movq	arg2, 24(%rsp)
+	movq	%r8, 24(%rsp)
-	movq	arg3, 32(%rsp)
+	movq	%r9, 32(%rsp)
 
 	movq	FFI_TRAMPOLINE_SIZE(%r10), arg0		/* load cif */
 	movq	FFI_TRAMPOLINE_SIZE+8(%r10), arg1	/* load fun */
@@ -195,8 +204,8 @@ ffi_closure_win64:
 0:
 	subq	$ffi_clo_FS, %rsp
 	cfi_adjust_cfa_offset(ffi_clo_FS)
-	.seh_stackalloc ffi_clo_FS
+	SEH(.seh_stackalloc ffi_clo_FS)
-	.seh_endprologue
+	SEH(.seh_endprologue)
 
 	/* Save all sse arguments into the stack frame.  */
 	movsd	%xmm0, ffi_clo_OFF_X(%rsp)
@@ -216,4 +225,4 @@ ffi_closure_win64:
 	ret
 
 	cfi_endproc
-	.seh_endproc
+	SEH(.seh_endproc)