view lwasm/insn_indexed.c @ 418:3832a68d83ef

Fix internal compiler error on "var2 = var1 + 1" patterns This appears to be the correct fix. It was provided by Tormod Volden (debian.tormod@gmail.com). The final commit is substantially different from Tormod's submission mostly due to housecleaning (removing the old patches and updating the README). Tormod's comments follow. The original addhi_mem_1 "insn" instruction pattern /matches/ two memory operands, just with the /constraint/ that these are the same location. A pattern match tells the compiler "you should be able to use this, but you might have to work on it to meet the constraints". For typical constraints on registers the compiler can add "reloads", moving stuff between registers or from memory, until the constraints are met and the instruction can be used. However, in this case, no amount of reloads can make two memory locations the same if they already weren't, so the compiler breaks down and cries "unable to generate reloads". It seems this issue only appears if optimization is enabled. The proof is in gcc's reload.c and is left as an exercise to the reader. Limiting the matching pattern to identical memory operands avoids these situations, while allowing the common "var++" cases. References: The pattern/constraints difference is explained in https://gcc.gnu.org/onlinedocs/gccint/Simple-Constraints.html#index-other-register-constraints-3335
author William Astle <lost@l-w.ca>
date Tue, 29 Mar 2016 21:21:49 -0600
parents b20f14edda5a
children cad5937314cb
line wrap: on
line source

/*
insn_indexed.c
Copyright © 2009 William Astle

This file is part of LWASM.

LWASM 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 3 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, see <http://www.gnu.org/licenses/>.
*/

/*
for handling indexed mode instructions
*/

#include <ctype.h>
#include <string.h>

#include <lw_expr.h>

#include "lwasm.h"
#include "instab.h"

/*
l -> lint: size of operand (0, 1, 2, -1 if not determined)
l -> pb: actual post byte (from "resolve" stage) or info passed
	forward to the resolve stage (if l -> line is -1); 0x80 is indir
	bits 0-2 are register number
*/
void insn_parse_indexed_aux(asmstate_t *as, line_t *l, char **p)
{
	static const char *regs9 = "X  Y  U  S     PCRPC ";
	static const char *regs  = "X  Y  U  S  W  PCRPC ";
	int i, rn;
	int indir = 0;
	int f0 = 0;
	const char *reglist;
	lw_expr_t e;
	char *tstr;
	

	if (CURPRAGMA(l, PRAGMA_6809))
	{
		reglist = regs9;
	}
	else
	{
		reglist = regs;
	}
	// is it indirect?
	if (**p == '[')
	{
		indir = 1;
		(*p)++;
	}
	lwasm_skip_to_next_token(l, p);
	if (**p == ',')
	{
		int incdec = 0;
		/* we have a pre-dec, post-inc, or no offset mode here */
		(*p)++;
		lwasm_skip_to_next_token(l, p);
		if (**p == '-')
		{
			incdec = -1;
			(*p)++;
			if (**p == '-')
			{
				incdec = -2;
				(*p)++;
			}
			lwasm_skip_to_next_token(l, p);
		}
		/* allowed registers: X, Y, U, S, or W (6309) */
		switch (**p)
		{
		case 'x':
		case 'X':
			rn = 0;
			break;
		
		case 'y':
		case 'Y':
			rn = 1;
			break;
			
		case 'u':
		case 'U':
			rn = 2;
			break;
			
		case 's':
		case 'S':
			rn = 3;
			break;
			
		case 'w':
		case 'W':
			if (CURPRAGMA(l, PRAGMA_6809))
			{
				lwasm_register_error(as, l, E_OPERAND_BAD);
				return;
			}
			rn = 4;
			break;
			
		default:
			lwasm_register_error(as, l, E_OPERAND_BAD);
			return;
		}
		(*p)++;
		lwasm_skip_to_next_token(l, p);
		if (**p == '+')
		{
			if (incdec != 0)
			{
				lwasm_register_error(as, l, E_OPERAND_BAD);
				return;
			}
			incdec = 1;
			(*p)++;
			if (**p == '+')
			{
				incdec = 2;
				(*p)++;
			}
			lwasm_skip_to_next_token(l, p);
		}
		if (indir)
		{
			if (**p != ']')
			{
				lwasm_register_error(as, l, E_OPERAND_BAD);
				return;
			}
			(*p)++;
		}
		if (indir || rn == 4)
		{
			if (incdec == 1 || incdec == -1)
			{
				lwasm_register_error(as, l, E_OPERAND_BAD);
				return;
			}
		}
		if (rn == 4)
		{
			if (indir)
			{
				if (incdec == 0)
					i = 0x90;
				else if (incdec == -2)
					i = 0xF0;
				else
					i = 0xD0;
			}
			else
			{
				if (incdec == 0)
					i = 0x8F;
				else if (incdec == -2)
					i = 0xEF;
				else
					i = 0xCF;
			}
		}
		else
		{
			switch (incdec)
			{
			case 0:
				i = 0x84;
				break;
			case 1:
				i = 0x80;
				break;
			case 2:
				i = 0x81;
				break;
			case -1:
				i = 0x82;
				break;
			case -2:
				i = 0x83;
				break;
			}
			i = (rn << 5) | i | (indir << 4);
		}
		l -> pb = i;
		l -> lint = 0;
		return;
	}
	i = toupper(**p);
	if (
			(i == 'A' || i == 'B' || i == 'D') ||
			(!CURPRAGMA(l, PRAGMA_6809) && (i == 'E' || i == 'F' || i == 'W'))
	   )
	{
		tstr = *p + 1;
		lwasm_skip_to_next_token(l, &tstr);
		if (*tstr == ',')
		{
			*p = tstr + 1;
			lwasm_skip_to_next_token(l, p);
			switch (**p)
			{
			case 'x':
			case 'X':
				rn = 0;
				break;
		
			case 'y':
			case 'Y':
				rn = 1;
				break;
			
			case 'u':
			case 'U':
				rn = 2;
				break;
			
			case 's':
			case 'S':
				rn = 3;
				break;
			
			default:
				lwasm_register_error(as, l, E_OPERAND_BAD);
				return;
			}
			(*p)++;
			lwasm_skip_to_next_token(l, p);
			if (indir)
			{
				if (**p != ']')
				{
					lwasm_register_error(as, l, E_OPERAND_BAD);
					return;
				}
				(*p)++;
			}
			
			switch (i)
			{
			case 'A':
				i = 0x86;
				break;
			
			case 'B':
				i = 0x85;
				break;
			
			case 'D':
				i = 0x8B;
				break;
			
			case 'E':
				i = 0x87;
				break;
			
			case 'F':
				i = 0x8A;
				break;
			
			case 'W':
				i = 0x8E;
				break;
			}
			l -> pb = i | (indir << 4) | (rn << 5);
			l -> lint = 0;
			return;
		}
	}
	
	/* we have the "expression" types now */
	if (**p == '<')
	{
		l -> lint = 1;
		(*p)++;
	}
	else if (**p == '>')
	{
		l -> lint = 2;
		(*p)++;
	}
	lwasm_skip_to_next_token(l, p);
	if (**p == '0')
	{
		tstr = *p + 1;
		lwasm_skip_to_next_token(l, &tstr);
		if (*tstr == ',')
		{
			f0 = 1;
		}
	}

	// now we have to evaluate the expression
	e = lwasm_parse_expr(as, p);
	if (!e)
	{
		lwasm_register_error(as, l, E_OPERAND_BAD);
		return;
	}
	lwasm_save_expr(l, 0, e);
	
	if (**p != ',')
	{
		/* if no comma, we have extended indirect */
		if (l -> lint == 1 || **p != ']')
		{
			lwasm_register_error(as, l, E_OPERAND_BAD);
			return;
		}
		(*p)++;
		l -> lint = 2;
		l -> pb = 0x9F;
		return;
	}
	(*p)++;
	lwasm_skip_to_next_token(l, p);
	// now get the register
	rn = lwasm_lookupreg3(reglist, p);
	if (rn < 0)
	{
		lwasm_register_error(as, l, E_REGISTER_BAD);
		return;
	}
	
	if (indir)
	{
		if (**p != ']')
		{
			lwasm_register_error(as, l, E_OPERAND_BAD);
			return;
		}
		else
			(*p)++;
	}

	if (rn <= 3)
	{
		// X,Y,U,S
		if (l -> lint == 1)
		{
			l -> pb = 0x88 | (rn << 5) | (indir ? 0x10 : 0);
			return;
		}
		else if (l -> lint == 2)
		{
			l -> pb = 0x89 | (rn << 5) | (indir ? 0x10 : 0);
			return;
		}
	}

	// nnnn,W is only 16 bit (or 0 bit)
	if (rn == 4)
	{
		if (l -> lint == 1)
		{
			lwasm_register_error(as, l, E_NW_8);
			return;
		}

		if (l -> lint == 2)
		{
			l -> pb = indir ? 0xb0 : 0xaf;
			l -> lint = 2;
			return;
		}
		
		l -> pb = (0x80 * indir) | rn;

/* [,w] and ,w
			if (indir)
				*b1 = 0x90;
			else
				*b1 = 0x8f;
*/
		return;
	}
	
	// PCR? then we have PC relative addressing (like B??, LB??)
	if (rn == 5 || (rn == 6 && CURPRAGMA(l, PRAGMA_PCASPCR)))
	{
		lw_expr_t e1, e2;
		// external references are handled exactly the same as for
		// relative addressing modes
		// on pass 1, adjust the expression for a subtraction of the
		// current address
		// e - (addr + linelen) => e - addr - linelen
		
		e2 = lw_expr_build(lw_expr_type_special, lwasm_expr_linelen, l);
		e1 = lw_expr_build(lw_expr_type_oper, lw_expr_oper_minus, e, e2);
		lw_expr_destroy(e2);
		e2 = lw_expr_build(lw_expr_type_oper, lw_expr_oper_minus, e1, l -> addr);
		lw_expr_destroy(e1);
		lwasm_save_expr(l, 0, e2);
		if (l -> lint == 1)
		{
			l -> pb = indir ? 0x9C : 0x8C;
			return;
		}
		if (l -> lint == 2)
		{
			l -> pb = indir ? 0x9D : 0x8D;
			return;
		}
	}
	
	if (rn == 6)
	{
		if (l -> lint == 1)
		{
			l -> pb = indir ? 0x9C : 0x8C;
			return;
		}
		if (l -> lint == 2)
		{
			l -> pb = indir ? 0x9D : 0x8D;
			return;
		}
	}

	l -> pb = (indir * 0x80) | rn | (f0 * 0x40);
}

PARSEFUNC(insn_parse_indexed)
{
	l -> lint = -1;
	insn_parse_indexed_aux(as, l, p);

	if (l -> lint != -1)
	{
		l -> len = OPLEN(instab[l -> insn].ops[0]) + l -> lint + 1;
	}
}

void insn_resolve_indexed_aux(asmstate_t *as, line_t *l, int force, int elen)
{
	// here, we have an expression which needs to be
	// resolved; the post byte is determined here as well
	lw_expr_t e, e2;
	int pb = -1;
	int v;
	
	if (l -> len != -1)
		return;

	e = lwasm_fetch_expr(l, 0);
	if (!lw_expr_istype(e, lw_expr_type_int))
	{
		// temporarily set the instruction length to see if we get a
		// constant for our expression; if so, we can select an instruction
		// size
		e2 = lw_expr_copy(e);
		// magic 2 for 8 bit (post byte + offset)
		l -> len = OPLEN(instab[l -> insn].ops[0]) + elen + 2;
		lwasm_reduce_expr(as, e2);
//		l -> len += 1;
//		e3 = lw_expr_copy(e);
//		lwasm_reduce_expr(as, e3);
		l -> len = -1;
		if (lw_expr_istype(e2, lw_expr_type_int))
		{
			v = lw_expr_intval(e2);
			// we have a reducible expression here which depends on
			// the size of this instruction
			if (v == 0 && !CURPRAGMA(l, PRAGMA_NOINDEX0TONONE) && (l -> pb & 0x07) <= 4)
			{
				if ((l -> pb & 0x07) < 4)
				{
					pb = 0x84 | ((l -> pb & 0x03) << 5) | ((l -> pb & 0x80) ? 0x10 : 0);
				}
				else
				{
					pb = (l -> pb & 0x80) ? 0x90 : 0x8F;
				}
				l -> pb = pb;
				lw_expr_destroy(e2);
				l -> lint = 0;
				return;
			}
			else if (v < -128 || v > 127)
			{
				l -> lint = 2;
				switch (l -> pb & 0x07)
				{
				case 0:
				case 1:
				case 2:
				case 3:
					pb = 0x89 | ((l -> pb & 0x03) << 5) | ((l -> pb & 0x80) ? 0x10 : 0);
					break;
			
				case 4: // W
					pb = (l -> pb & 0x80) ? 0xB0 : 0xAF;
					break;
				
				case 5: // PCR
				case 6: // PC
					pb = (l -> pb & 0x80) ? 0x9D : 0x8D;
					break;
				}
				
				l -> pb = pb;
				lw_expr_destroy(e2);
//				lw_expr_destroy(e3);
				return;
			}
			else if ((l -> pb & 0x80) || ((l -> pb & 0x07) > 3) || v < -16 || v > 15)
			{
				// if not a 5 bit value, is indirect, or is not X,Y,U,S
				l -> lint = 1;
				switch (l -> pb & 0x07)
				{
				case 0:
				case 1:
				case 2:
				case 3:
					pb = 0x88 | ((l -> pb & 0x03) << 5) | ((l -> pb & 0x80) ? 0x10 : 0);
					break;
			
				case 4: // W
					// use 16 bit because W doesn't have 8 bit, unless 0
					if (v == 0 && !(CURPRAGMA(l, PRAGMA_NOINDEX0TONONE) || l -> pb & 0x40))
					{
						pb = (l -> pb & 0x80) ? 0x90 : 0x8F;
						l -> lint = 0;
					}
					else
					{
						pb = (l -> pb & 0x80) ? 0xB0 : 0xAF;
						l -> lint = 2;
					}
					break;
				
				case 5: // PCR
				case 6: // PC
					pb = (l -> pb & 0x80) ? 0x9C : 0x8C;
					break;
				}
			
				l -> pb = pb;
				lw_expr_destroy(e2);
				return;
			}
			else
			{
				// we have X,Y,U,S and a possible 5 bit here
				l -> lint = 0;
				
				if (v == 0 && !(CURPRAGMA(l, PRAGMA_NOINDEX0TONONE) || l -> pb & 0x40))
				{
					pb = (l -> pb & 0x03) << 5 | 0x84;
				}	
				else
				{
					pb = ((l -> pb & 0x03) << 5) | (v & 0x1F);
				}
				l -> pb = pb;
				lw_expr_destroy(e2);
				return;
			}
		}
		lw_expr_destroy(e2);
	}
		
	if (lw_expr_istype(e, lw_expr_type_int))
	{
		// we know how big it is
		v = lw_expr_intval(e);
			
		if (v == 0 && !CURPRAGMA(l, PRAGMA_NOINDEX0TONONE) && (l -> pb & 0x07) <= 4 && ((l -> pb & 0x40) == 0))
		{
			if ((l -> pb & 0x07) < 4)
			{
				pb = 0x84 | ((l -> pb & 0x03) << 5) | ((l -> pb & 0x80) ? 0x10 : 0);
			}
			else
			{
				pb = (l -> pb & 0x80) ? 0x90 : 0x8F;
			}
			l -> pb = pb;
			l -> lint = 0;
			return;
		}
		else if (v < -128 || v > 127)
		{
		do16bit:
			l -> lint = 2;
			switch (l -> pb & 0x07)
			{
			case 0:
			case 1:
			case 2:
			case 3:
				pb = 0x89 | (l -> pb & 0x03) << 5 | ((l -> pb & 0x80) ? 0x10 : 0);
				break;
			
			case 4: // W
				pb = (l -> pb & 0x80) ? 0xB0 : 0xAF;
				break;
				
			case 5: // PCR
			case 6: // PC
				pb = (l -> pb & 0x80) ? 0x9D : 0x8D;
				break;
			}
			
			l -> pb = pb;
			return;
		}
		else if ((l -> pb & 0x80) || ((l -> pb & 0x07) > 3) || v < -16 || v > 15)
		{
			// if not a 5 bit value, is indirect, or is not X,Y,U,S
			l -> lint = 1;
			switch (l -> pb & 0x07)
			{
			case 0:
			case 1:
			case 2:
			case 3:
				pb = 0x88 | (l -> pb & 0x03) << 5 | ((l -> pb & 0x80) ? 0x10 : 0);
				break;
			
			case 4: // W
				// use 16 bit because W doesn't have 8 bit, unless 0
				if (v == 0 && !(CURPRAGMA(l, PRAGMA_NOINDEX0TONONE) || l -> pb & 0x40))
				{
					pb = (l -> pb & 0x80) ? 0x90 : 0x8F;
					l -> lint = 0;
				}
				else
				{
					pb = (l -> pb & 0x80) ? 0xB0 : 0xAF;
					l -> lint = 2;
				}
				break;
				
			case 5: // PCR
			case 6: // PC
				pb = (l -> pb & 0x80) ? 0x9C : 0x8C;
				break;
			}
			
			l -> pb = pb;
			return;
		}
		else
		{
			// we have X,Y,U,S and a possible 5 bit here
			l -> lint = 0;
			
			if (v == 0 && !(CURPRAGMA(l, PRAGMA_NOINDEX0TONONE) || l -> pb & 0x40))
			{
				pb = (l -> pb & 0x03) << 5 | 0x84;
			}
			else
			{
				pb = ((l -> pb & 0x03) << 5) | (v & 0x1F);
			}
			l -> pb = pb;
			return;
		}
	}
	else
	{
		// we don't know how big it is
		if (!force)
			return;
		// force 16 bit if we don't know
		l -> lint = 2;
		goto do16bit;
	}
}

RESOLVEFUNC(insn_resolve_indexed)
{
	if (l -> lint == -1)
		insn_resolve_indexed_aux(as, l, force, 0);
	
	if (l -> lint != -1 && l -> pb != -1)
	{
		l -> len = OPLEN(instab[l -> insn].ops[0]) + l -> lint + 1;
	}
}

void insn_emit_indexed_aux(asmstate_t *as, line_t *l)
{
	lw_expr_t e;
	
	if (l -> lint == 1)
	{
		int i;
		e = lwasm_fetch_expr(l, 0);
		i = lw_expr_intval(e);
		if (i < -128 || i > 127)
		{
			lwasm_register_error(as, l, E_BYTE_OVERFLOW);
		}
	}
	
	lwasm_emitop(l, instab[l -> insn].ops[0]);
	lwasm_emitop(l, l -> pb);

	l -> cycle_adj = lwasm_cycle_calc_ind(l);

	if (l -> lint > 0)
	{
		e = lwasm_fetch_expr(l, 0);
		lwasm_emitexpr(l, e, l -> lint);
	}
}

EMITFUNC(insn_emit_indexed)
{
	insn_emit_indexed_aux(as, l);
}