* The standard x86-64 ABI passes the three arguments m, g, fn
 * in %rdi, %rsi, %rdx.
 */
.globl EXT(crosscall1)
EXT(crosscall1):
	pushq %rbx
	pushq %rbp
	pushq %r12
	pushq %r13
	pushq %r14
	pushq %r15

#if defined(_WIN64)
	movq %r8, %rdi	/* arg of setg_gcc */
	call *%rdx	/* setg_gcc */
	call *%rcx	/* fn */
#else
	movq %rdi, %rbx
	movq %rdx, %rdi	/* arg of setg_gcc */
	call *%rsi	/* setg_gcc */

	XORQ rBe, rDe;         \
	MOVQ rDe, _se(oState); \
	S_RDE_RCE;             \
	                       \
	MOVQ rBo, rDi;         \
	MOVQ rBu, rDo;         \
	ANDQ rBu, rDi;         \
	ORQ  rBa, rDo;         \
	XORQ rBi, rDi;         \
	XORQ rBo, rDo;         \
	MOVQ rDi, _si(oState); \
	MOVQ rDo, _so(oState)  \

// func keccakF1600(a *[25]uint64)
TEXT ·keccakF1600(SB), 0, $200-8

	c.r9 = c.rip
}

func dumpregs(r *context) {
	print("rax     ", hex(r.rax), "\n")
	print("rbx     ", hex(r.rbx), "\n")
	print("rcx     ", hex(r.rcx), "\n")
	print("rdx     ", hex(r.rdx), "\n")
	print("rdi     ", hex(r.rdi), "\n")
	print("rsi     ", hex(r.rsi), "\n")
	print("rbp     ", hex(r.rbp), "\n")
	print("rsp     ", hex(r.rsp), "\n")
	print("r8      ", hex(r.r8), "\n")
	print("r9      ", hex(r.r9), "\n")

	R12      uint64
	Rbp      uint64
	Rbx      uint64
	R11      uint64
	R10      uint64
	R9       uint64
	R8       uint64
	Rax      uint64
	Rcx      uint64
	Rdx      uint64
	Rsi      uint64
	Rdi      uint64
	Orig_rax uint64
	Rip      uint64
	Cs       uint64
	Eflags   uint64
	Rsp      uint64
	Ss       uint64
	Fs_base  uint64
	Gs_base  uint64
	Ds       uint64
	Es       uint64
	Fs       uint64

    addl    4(%esp), %eax
    ret
'''

    private static String x64GnuAsmSource = '''
    .text
    .p2align 4,,15
.globl sumx
    .type   sumx, @function
sumx:
    leal    (%rsi,%rdi), %eax
    ret
'''

	R11L: "R11L",
	R12L: "R12L",
	R13L: "R13L",
	R14L: "R14L",
	R15L: "R15L",
	RAX:  "RAX",
	RCX:  "RCX",
	RDX:  "RDX",
	RBX:  "RBX",
	RSP:  "RSP",
	RBP:  "RBP",
	RSI:  "RSI",
	RDI:  "RDI",
	R8:   "R8",
	R9:   "R9",
	R10:  "R10",
	R11:  "R11",
	R12:  "R12",
	R13:  "R13",
	R14:  "R14",
	R15:  "R15",
	IP:   "IP",
	EIP:  "EIP",
	RIP:  "RIP",
	F0:   "F0",
	F1:   "F1",

}

type sigcontext struct {
	r8          uint64
	r9          uint64
	r10         uint64
	r11         uint64
	r12         uint64
	r13         uint64
	r14         uint64
	r15         uint64
	rdi         uint64
	rsi         uint64
	rbp         uint64
	rbx         uint64
	rdx         uint64
	rax         uint64
	rcx         uint64
	rsp         uint64
	rip         uint64
	eflags      uint64

	mmst_reg  [10]int8
	mmst_rsrv [6]int8
}

type regxmm struct {
	xmm_reg [16]int8
}

type regs64 struct {
	rax    uint64
	rbx    uint64
	rcx    uint64
	rdx    uint64
	rdi    uint64
	rsi    uint64
	rbp    uint64
	rsp    uint64
	r8     uint64
	r9     uint64
	r10    uint64
	r11    uint64
	r12    uint64
	r13    uint64
	r14    uint64
	r15    uint64

	# quotient fits in 64 bits because top 64 bits have been reduced < divisor.
	# (even though we only care about the remainder, divq also computes
	# the quotient, and it will trap if the quotient is too large.)
	movq %rdi, %rax
	divq %r8

	# expand remainder to 128 for return
	movq %rdx, %rax
	xorl %edx, %edx
	ret

1:
	# crash - only want 64-bit divisor
	xorl %ecx, %ecx
	movl %ecx, 0(%ecx)
	jmp 1b

type Sigset_t struct {
	Val [4]uint32
}

type Reg struct {
	R15    int64
	R14    int64
	R13    int64
	R12    int64
	R11    int64
	R10    int64
	R9     int64
	R8     int64
	Rdi    int64
	Rsi    int64
	Rbp    int64
	Rbx    int64
	Rdx    int64
	Rcx    int64
	Rax    int64
	Trapno uint32
	Fs     uint16
	Gs     uint16
	Err    uint32
	Es     uint16
	Ds     uint16