tinycc/i386-gen.c

721 lines
19 KiB
C

/*
* X86 code generator for TCC
*
* Copyright (c) 2001 Fabrice Bellard
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
/* number of available registers */
#define NB_REGS 4
/* a register can belong to several classes */
#define RC_INT 0x0001 /* generic integer register */
#define RC_FLOAT 0x0002 /* generic float register */
#define RC_EAX 0x0004
#define RC_FRET 0x0008 /* function return: float register */
#define RC_ECX 0x0010
#define RC_EDX 0x0020
#define RC_IRET RC_EAX /* function return: integer register */
#define RC_LRET RC_EDX /* function return: second integer register */
/* pretty names for the registers */
enum {
REG_EAX = 0,
REG_ECX,
REG_EDX,
REG_ST0,
};
int reg_classes[NB_REGS] = {
/* eax */ RC_INT | RC_IRET,
/* ecx */ RC_INT | RC_ECX,
/* edx */ RC_INT | RC_EDX,
/* st0 */ RC_FLOAT | RC_FRET,
};
/* return registers for function */
#define REG_IRET REG_EAX /* single word int return register */
#define REG_LRET REG_EDX /* second word return register (for long long) */
#define REG_FRET REG_ST0 /* float return register */
/* defined if function parameters must be evaluated in reverse order */
#define INVERT_FUNC_PARAMS
/* defined if structures are passed as pointers. Otherwise structures
are directly pushed on stack. */
//#define FUNC_STRUCT_PARAM_AS_PTR
/* pointer size, in bytes */
#define PTR_SIZE 4
/* long double size and alignment, in bytes */
#define LDOUBLE_SIZE 12
#define LDOUBLE_ALIGN 4
/* function call context */
typedef struct GFuncContext {
int args_size;
} GFuncContext;
/******************************************************/
void g(int c)
{
*(char *)ind++ = c;
}
void o(int c)
{
while (c) {
g(c);
c = c / 256;
}
}
void gen_le32(int c)
{
g(c);
g(c >> 8);
g(c >> 16);
g(c >> 24);
}
/* patch relocation entry with value 'val' */
void greloc_patch1(Reloc *p, int val)
{
switch(p->type) {
case RELOC_ADDR32:
*(int *)p->addr = val;
break;
case RELOC_REL32:
*(int *)p->addr = val - p->addr - 4;
break;
}
}
/* output a symbol and patch all calls to it */
void gsym_addr(t, a)
{
int n;
while (t) {
n = *(int *)t; /* next value */
*(int *)t = a - t - 4;
t = n;
}
}
void gsym(t)
{
gsym_addr(t, ind);
}
/* psym is used to put an instruction with a data field which is a
reference to a symbol. It is in fact the same as oad ! */
#define psym oad
/* instruction + 4 bytes data. Return the address of the data */
int oad(int c, int s)
{
o(c);
*(int *)ind = s;
s = ind;
ind = ind + 4;
return s;
}
/* output constant with relocation if 'r & VT_FORWARD' is true */
void gen_addr32(int r, int c)
{
if (!(r & VT_FORWARD)) {
gen_le32(c);
} else {
greloc((Sym *)c, ind, RELOC_ADDR32);
gen_le32(0);
}
}
/* generate a modrm reference. 'op_reg' contains the addtionnal 3
opcode bits */
void gen_modrm(int op_reg, int r, int c)
{
op_reg = op_reg << 3;
if ((r & VT_VALMASK) == VT_CONST) {
/* constant memory reference */
o(0x05 | op_reg);
gen_addr32(r, c);
} else if ((r & VT_VALMASK) == VT_LOCAL) {
/* currently, we use only ebp as base */
if (c == (char)c) {
/* short reference */
o(0x45 | op_reg);
g(c);
} else {
oad(0x85 | op_reg, c);
}
} else {
g(0x00 | op_reg | (r & VT_VALMASK));
}
}
/* load 'r' from value 'sv' */
void load(int r, SValue *sv)
{
int v, t, ft, fc, fr;
SValue v1;
fr = sv->r;
ft = sv->t;
fc = sv->c.ul;
v = fr & VT_VALMASK;
if (fr & VT_LVAL) {
if (v == VT_LLOCAL) {
v1.t = VT_INT;
v1.r = VT_LOCAL | VT_LVAL;
v1.c.ul = fc;
load(r, &v1);
fr = r;
}
if ((ft & VT_BTYPE) == VT_FLOAT) {
o(0xd9); /* flds */
r = 0;
} else if ((ft & VT_BTYPE) == VT_DOUBLE) {
o(0xdd); /* fldl */
r = 0;
} else if ((ft & VT_BTYPE) == VT_LDOUBLE) {
o(0xdb); /* fldt */
r = 5;
} else if ((ft & VT_TYPE) == VT_BYTE)
o(0xbe0f); /* movsbl */
else if ((ft & VT_TYPE) == (VT_BYTE | VT_UNSIGNED))
o(0xb60f); /* movzbl */
else if ((ft & VT_TYPE) == VT_SHORT)
o(0xbf0f); /* movswl */
else if ((ft & VT_TYPE) == (VT_SHORT | VT_UNSIGNED))
o(0xb70f); /* movzwl */
else
o(0x8b); /* movl */
gen_modrm(r, fr, fc);
} else {
if (v == VT_CONST) {
o(0xb8 + r); /* mov $xx, r */
gen_addr32(fr, fc);
} else if (v == VT_LOCAL) {
o(0x8d); /* lea xxx(%ebp), r */
gen_modrm(r, VT_LOCAL, fc);
} else if (v == VT_CMP) {
oad(0xb8 + r, 0); /* mov $0, r */
o(0x0f); /* setxx %br */
o(fc);
o(0xc0 + r);
} else if (v == VT_JMP || v == VT_JMPI) {
t = v & 1;
oad(0xb8 + r, t); /* mov $1, r */
oad(0xe9, 5); /* jmp after */
gsym(fc);
oad(0xb8 + r, t ^ 1); /* mov $0, r */
} else if (v != r) {
o(0x89);
o(0xc0 + r + v * 8); /* mov v, r */
}
}
}
/* store register 'r' in lvalue 'v' */
void store(int r, SValue *v)
{
int fr, bt, ft, fc;
ft = v->t;
fc = v->c.ul;
fr = v->r & VT_VALMASK;
bt = ft & VT_BTYPE;
/* XXX: incorrect if float reg to reg */
if (bt == VT_FLOAT) {
o(0xd9); /* fsts */
r = 2;
} else if (bt == VT_DOUBLE) {
o(0xdd); /* fstpl */
r = 2;
} else if (bt == VT_LDOUBLE) {
o(0xc0d9); /* fld %st(0) */
o(0xdb); /* fstpt */
r = 7;
} else {
if (bt == VT_SHORT)
o(0x66);
if (bt == VT_BYTE)
o(0x88);
else
o(0x89);
}
if (fr == VT_CONST ||
fr == VT_LOCAL ||
(v->r & VT_LVAL)) {
gen_modrm(r, v->r, fc);
} else if (fr != r) {
o(0xc0 + fr + r * 8); /* mov r, fr */
}
}
/* start function call and return function call context */
void gfunc_start(GFuncContext *c)
{
c->args_size = 0;
}
/* push function parameter which is in (vtop->t, vtop->c). Stack entry
is then popped. */
void gfunc_param(GFuncContext *c)
{
int size, align, r;
if ((vtop->t & VT_BTYPE) == VT_STRUCT) {
size = type_size(vtop->t, &align);
/* align to stack align size */
size = (size + 3) & ~3;
/* allocate the necessary size on stack */
oad(0xec81, size); /* sub $xxx, %esp */
/* generate structure store */
r = get_reg(RC_INT);
o(0x89); /* mov %esp, r */
o(0xe0 + r);
vset(VT_INT, r, 0);
vswap();
vstore();
c->args_size += size;
} else if (is_float(vtop->t)) {
gv(RC_FLOAT); /* only one float register */
if ((vtop->t & VT_BTYPE) == VT_FLOAT)
size = 4;
else if ((vtop->t & VT_BTYPE) == VT_DOUBLE)
size = 8;
else
size = 12;
oad(0xec81, size); /* sub $xxx, %esp */
if (size == 12)
o(0x7cdb);
else
o(0x5cd9 + size - 4); /* fstp[s|l] 0(%esp) */
g(0x24);
g(0x00);
c->args_size += size;
} else {
/* simple type (currently always same size) */
/* XXX: implicit cast ? */
r = gv(RC_INT);
if ((vtop->t & VT_BTYPE) == VT_LLONG) {
size = 8;
o(0x50 + vtop->r2); /* push r */
} else {
size = 4;
}
o(0x50 + r); /* push r */
c->args_size += size;
}
vtop--;
}
/* generate function call with address in (vtop->t, vtop->c) and free function
context. Stack entry is popped */
void gfunc_call(GFuncContext *c)
{
int r;
if ((vtop->r & (VT_VALMASK | VT_LVAL)) == VT_CONST) {
/* constant case */
/* forward reference */
if (vtop->r & VT_FORWARD) {
greloc(vtop->c.sym, ind + 1, RELOC_REL32);
oad(0xe8, 0);
} else {
oad(0xe8, vtop->c.ul - ind - 5);
}
} else {
/* otherwise, indirect call */
r = gv(RC_INT);
o(0xff); /* call *r */
o(0xd0 + r);
}
if (c->args_size)
oad(0xc481, c->args_size); /* add $xxx, %esp */
vtop--;
}
int gjmp(int t)
{
return psym(0xe9, t);
}
/* generate a test. set 'inv' to invert test. Stack entry is popped */
int gtst(int inv, int t)
{
int v, *p;
v = vtop->r & VT_VALMASK;
if (v == VT_CMP) {
/* fast case : can jump directly since flags are set */
g(0x0f);
t = psym((vtop->c.i - 16) ^ inv, t);
} else if (v == VT_JMP || v == VT_JMPI) {
/* && or || optimization */
if ((v & 1) == inv) {
/* insert vtop->c jump list in t */
p = &vtop->c.i;
while (*p != 0)
p = (int *)*p;
*p = t;
t = vtop->c.i;
} else {
t = gjmp(t);
gsym(vtop->c.i);
}
} else {
if (is_float(vtop->t)) {
vpushi(0);
gen_op(TOK_NE);
}
if ((vtop->r & (VT_VALMASK | VT_LVAL | VT_FORWARD)) == VT_CONST) {
/* constant jmp optimization */
if ((vtop->c.i != 0) != inv)
t = gjmp(t);
} else {
v = gv(RC_INT);
o(0x85);
o(0xc0 + v * 9);
g(0x0f);
t = psym(0x85 ^ inv, t);
}
}
vtop--;
return t;
}
/* generate an integer binary operation */
void gen_opi(int op)
{
int r, fr, opc, c;
switch(op) {
case '+':
case TOK_ADDC1: /* add with carry generation */
opc = 0;
gen_op8:
vswap();
r = gv(RC_INT);
vswap();
if ((vtop->r & (VT_VALMASK | VT_LVAL | VT_FORWARD)) == VT_CONST) {
/* constant case */
c = vtop->c.i;
if (c == (char)c) {
/* XXX: generate inc and dec for smaller code ? */
o(0x83);
o(0xc0 | (opc << 3) | r);
g(c);
} else {
o(0x81);
oad(0xc0 | (opc << 3) | r, c);
}
} else {
fr = gv(RC_INT);
o((opc << 3) | 0x01);
o(0xc0 + r + fr * 8);
}
vtop--;
if (op >= TOK_ULT && op <= TOK_GT) {
vtop--;
vset(VT_INT, VT_CMP, op);
}
break;
case '-':
case TOK_SUBC1: /* sub with carry generation */
opc = 5;
goto gen_op8;
case TOK_ADDC2: /* add with carry use */
opc = 2;
goto gen_op8;
case TOK_SUBC2: /* sub with carry use */
opc = 3;
goto gen_op8;
case '&':
opc = 4;
goto gen_op8;
case '^':
opc = 6;
goto gen_op8;
case '|':
opc = 1;
goto gen_op8;
case '*':
vswap();
r = gv(RC_INT);
vswap();
fr = gv(RC_INT);
vtop--;
o(0xaf0f); /* imul fr, r */
o(0xc0 + fr + r * 8);
break;
case TOK_SHL:
opc = 4;
goto gen_shift;
case TOK_SHR:
opc = 5;
goto gen_shift;
case TOK_SAR:
opc = 7;
gen_shift:
vswap();
r = gv(RC_INT);
vswap();
opc = 0xc0 | (opc << 3);
if ((vtop->r & (VT_VALMASK | VT_LVAL | VT_FORWARD)) == VT_CONST) {
/* constant case */
c = vtop->c.i & 0x1f;
o(0xc1); /* shl/shr/sar $xxx, r */
o(opc | r);
g(c);
} else {
/* we generate the shift in ecx */
gv(RC_ECX);
/* the first op may have been spilled, so we reload it if
needed */
vswap();
r = gv(RC_INT);
vswap();
o(0xd3); /* shl/shr/sar %cl, r */
o(opc | r);
}
vtop--;
vtop->r = r;
break;
case '/':
case TOK_UDIV:
case TOK_PDIV:
case '%':
case TOK_UMOD:
case TOK_UMULL:
vswap();
r = gv(RC_EAX); /* first operand must be in eax */
vswap();
/* XXX: need better constraint */
fr = gv(RC_ECX); /* second operand in ecx */
vswap();
r = gv(RC_EAX); /* reload first operand if flushed */
vswap();
vtop--;
save_reg(REG_EDX);
if (op == TOK_UMULL) {
o(0xf7); /* mul fr */
o(0xe0 + fr);
vtop->r2 = REG_EDX;
r = REG_EAX;
} else {
if (op == TOK_UDIV || op == TOK_UMOD) {
o(0xf7d231); /* xor %edx, %edx, div fr, %eax */
o(0xf0 + fr);
} else {
o(0xf799); /* cltd, idiv fr, %eax */
o(0xf8 + fr);
}
if (op == '%' || op == TOK_UMOD)
r = REG_EDX;
else
r = REG_EAX;
}
vtop->r = r;
break;
default:
opc = 7;
goto gen_op8;
}
}
/* generate a floating point operation 'v = t1 op t2' instruction. The
two operands are guaranted to have the same floating point type */
/* NOTE: currently floats can only be lvalues */
void gen_opf(int op)
{
int a, ft, fc, swapped;
/* convert constants to memory references */
if ((vtop[-1].r & (VT_VALMASK | VT_LVAL)) == VT_CONST) {
vswap();
gv(RC_FLOAT);
vswap();
}
if ((vtop[0].r & (VT_VALMASK | VT_LVAL)) == VT_CONST)
gv(RC_FLOAT);
/* must put at least one value in the floating point register */
if ((vtop[-1].r & VT_LVAL) &&
(vtop[0].r & VT_LVAL)) {
vswap();
gv(RC_FLOAT);
vswap();
}
if (op >= TOK_ULT && op <= TOK_GT) {
/* load on stack second operand */
load(REG_ST0, vtop);
if (op == TOK_GE || op == TOK_GT)
o(0xc9d9); /* fxch %st(1) */
o(0xe9da); /* fucompp */
o(0xe0df); /* fnstsw %ax */
if (op == TOK_EQ) {
o(0x45e480); /* and $0x45, %ah */
o(0x40fC80); /* cmp $0x40, %ah */
} else if (op == TOK_NE) {
o(0x45e480); /* and $0x45, %ah */
o(0x40f480); /* xor $0x40, %ah */
op = TOK_NE;
} else if (op == TOK_GE || op == TOK_LE) {
o(0x05c4f6); /* test $0x05, %ah */
op = TOK_EQ;
} else {
o(0x45c4f6); /* test $0x45, %ah */
op = TOK_EQ;
}
vtop--;
vtop->r = VT_CMP;
vtop->c.i = op;
} else {
swapped = 0;
/* swap the stack if needed so that t1 is the register and t2 is
the memory reference */
if (vtop[-1].r & VT_LVAL) {
vswap();
swapped = 1;
}
/* no memory reference possible for long double operations */
if ((vtop->t & VT_BTYPE) == VT_LDOUBLE) {
load(REG_ST0, vtop);
swapped = !swapped;
}
switch(op) {
default:
case '+':
a = 0;
break;
case '-':
a = 4;
if (swapped)
a++;
break;
case '*':
a = 1;
break;
case '/':
a = 6;
if (swapped)
a++;
break;
}
ft = vtop->t;
fc = vtop->c.ul;
if ((ft & VT_BTYPE) == VT_LDOUBLE) {
o(0xde); /* fxxxp %st, %st(1) */
o(0xc1 + (a << 3));
} else {
if ((ft & VT_BTYPE) == VT_DOUBLE)
o(0xdc);
else
o(0xd8);
gen_modrm(a, vtop->r, fc);
}
vtop--;
}
}
/* FPU control word for rounding to nearest mode */
/* XXX: should move that into tcc lib support code ! */
static unsigned short __tcc_fpu_control = 0x137f;
/* FPU control word for round to zero mode for int convertion */
static unsigned short __tcc_int_fpu_control = 0x137f | 0x0c00;
/* convert integers to fp 't' type. Must handle 'int', 'unsigned int'
and 'long long' cases. */
void gen_cvt_itof(int t)
{
gv(RC_INT);
if ((vtop->t & VT_BTYPE) == VT_LLONG) {
/* signed long long to float/double/long double (unsigned case
is handled generically) */
o(0x50 + vtop->r2); /* push r2 */
o(0x50 + (vtop->r & VT_VALMASK)); /* push r */
o(0x242cdf); /* fildll (%esp) */
o(0x08c483); /* add $8, %esp */
} else if ((vtop->t & (VT_BTYPE | VT_UNSIGNED)) ==
(VT_INT | VT_UNSIGNED)) {
/* unsigned int to float/double/long double */
o(0x6a); /* push $0 */
g(0x00);
o(0x50 + (vtop->r & VT_VALMASK)); /* push r */
o(0x242cdf); /* fildll (%esp) */
o(0x08c483); /* add $8, %esp */
} else {
/* int to float/double/long double */
o(0x50 + (vtop->r & VT_VALMASK)); /* push r */
o(0x2404db); /* fildl (%esp) */
o(0x04c483); /* add $4, %esp */
}
vtop->r = REG_ST0;
}
/* convert fp to int 't' type */
/* XXX: handle long long case */
void gen_cvt_ftoi(int t)
{
int r, r2, size;
gv(RC_FLOAT);
if (t != VT_INT)
size = 8;
else
size = 4;
oad(0x2dd9, (int)&__tcc_int_fpu_control); /* ldcw xxx */
oad(0xec81, size); /* sub $xxx, %esp */
if (size == 4)
o(0x1cdb); /* fistpl */
else
o(0x3cdf); /* fistpll */
o(0x24);
oad(0x2dd9, (int)&__tcc_fpu_control); /* ldcw xxx */
r = get_reg(RC_INT);
o(0x58 + r); /* pop r */
if (size == 8) {
if (t == VT_LLONG) {
vtop->r = r; /* mark reg as used */
r2 = get_reg(RC_INT);
o(0x58 + r2); /* pop r2 */
vtop->r2 = r2;
} else {
o(0x04c483); /* add $4, %esp */
}
}
vtop->r = r;
}
/* convert from one floating point type to another */
void gen_cvt_ftof(int t)
{
/* all we have to do on i386 is to put the float in a register */
gv(RC_FLOAT);
}
/* end of X86 code generator */
/*************************************************************/