1 /* Assembler interface for targets using CGEN. -*- C -*-
2    CGEN: Cpu tools GENerator
3 
4    THIS FILE IS MACHINE GENERATED WITH CGEN.
5    - the resultant file is machine generated, cgen-asm.in isn't
6 
7    Copyright 1996, 1997, 1998, 1999, 2000, 2001, 2005, 2007
8    Free Software Foundation, Inc.
9 
10    This file is part of libopcodes.
11 
12    This library is free software; you can redistribute it and/or modify
13    it under the terms of the GNU General Public License as published by
14    the Free Software Foundation; either version 3, or (at your option)
15    any later version.
16 
17    It is distributed in the hope that it will be useful, but WITHOUT
18    ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
19    or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
20    License for more details.
21 
22    You should have received a copy of the GNU General Public License
23    along with this program; if not, write to the Free Software Foundation, Inc.,
24    51 Franklin Street - Fifth Floor, Boston, MA 02110-1301, USA.  */
25 
26 
27 /* ??? Eventually more and more of this stuff can go to cpu-independent files.
28    Keep that in mind.  */
29 
30 #include "sysdep.h"
31 #include <stdio.h>
32 #include "ansidecl.h"
33 #include "bfd.h"
34 #include "symcat.h"
35 #include "iq2000-desc.h"
36 #include "iq2000-opc.h"
37 #include "opintl.h"
38 #include "xregex.h"
39 #include "libiberty.h"
40 #include "safe-ctype.h"
41 
42 #undef  min
43 #define min(a,b) ((a) < (b) ? (a) : (b))
44 #undef  max
45 #define max(a,b) ((a) > (b) ? (a) : (b))
46 
47 static const char * parse_insn_normal
48   (CGEN_CPU_DESC, const CGEN_INSN *, const char **, CGEN_FIELDS *);
49 
50 /* -- assembler routines inserted here.  */
51 
52 /* -- asm.c */
53 
54 #include "safe-ctype.h"
55 
56 static const char * MISSING_CLOSING_PARENTHESIS = N_("missing `)'");
57 
58 /* Special check to ensure that instruction exists for given machine.  */
59 
60 int
61 iq2000_cgen_insn_supported (CGEN_CPU_DESC cd, const CGEN_INSN *insn)
62 {
63   int machs = cd->machs;
64 
65   return (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_MACH) & machs) != 0;
66 }
67 
68 static int
69 iq2000_cgen_isa_register (const char **strp)
70 {
71   int len;
72   int ch1, ch2;
73 
74   if (**strp == 'r' || **strp == 'R')
75     {
76       len = strlen (*strp);
77       if (len == 2)
78         {
79           ch1 = (*strp)[1];
80           if ('0' <= ch1 && ch1 <= '9')
81             return 1;
82         }
83       else if (len == 3)
84         {
85 	  ch1 = (*strp)[1];
86           ch2 = (*strp)[2];
87           if (('1' <= ch1 && ch1 <= '2') && ('0' <= ch2 && ch2 <= '9'))
88             return 1;
89           if ('3' == ch1 && (ch2 == '0' || ch2 == '1'))
90             return 1;
91         }
92     }
93   if (**strp == '%'
94       && TOLOWER ((*strp)[1]) != 'l'
95       && TOLOWER ((*strp)[1]) != 'h')
96     return 1;
97   return 0;
98 }
99 
100 /* Handle negated literal.  */
101 
102 static const char *
103 parse_mimm (CGEN_CPU_DESC cd,
104 	    const char **strp,
105 	    int opindex,
106 	    unsigned long *valuep)
107 {
108   const char *errmsg;
109 
110   /* Verify this isn't a register.  */
111   if (iq2000_cgen_isa_register (strp))
112     errmsg = _("immediate value cannot be register");
113   else
114     {
115       long value;
116 
117       errmsg = cgen_parse_signed_integer (cd, strp, opindex, & value);
118       if (errmsg == NULL)
119 	{
120 	  long x = (-value) & 0xFFFF0000;
121 
122 	  if (x != 0 && x != (long) 0xFFFF0000)
123 	    errmsg = _("immediate value out of range");
124 	  else
125 	    *valuep = (-value & 0xFFFF);
126 	}
127     }
128   return errmsg;
129 }
130 
131 /* Handle signed/unsigned literal.  */
132 
133 static const char *
134 parse_imm (CGEN_CPU_DESC cd,
135 	   const char **strp,
136 	   int opindex,
137 	   unsigned long *valuep)
138 {
139   const char *errmsg;
140 
141   if (iq2000_cgen_isa_register (strp))
142     errmsg = _("immediate value cannot be register");
143   else
144     {
145       long value;
146 
147       errmsg = cgen_parse_signed_integer (cd, strp, opindex, & value);
148       if (errmsg == NULL)
149 	{
150 	  long x = value & 0xFFFF0000;
151 
152 	  if (x != 0 && x != (long) 0xFFFF0000)
153 	    errmsg = _("immediate value out of range");
154 	  else
155 	    *valuep = (value & 0xFFFF);
156 	}
157     }
158   return errmsg;
159 }
160 
161 /* Handle iq10 21-bit jmp offset.  */
162 
163 static const char *
164 parse_jtargq10 (CGEN_CPU_DESC cd,
165 		const char **strp,
166 		int opindex,
167 		int reloc ATTRIBUTE_UNUSED,
168 		enum cgen_parse_operand_result *type_addr ATTRIBUTE_UNUSED,
169 		bfd_vma *valuep)
170 {
171   const char *errmsg;
172   bfd_vma value;
173   enum cgen_parse_operand_result result_type = CGEN_PARSE_OPERAND_RESULT_NUMBER;
174 
175   errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_IQ2000_OFFSET_21,
176 			       & result_type, & value);
177   if (errmsg == NULL && result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
178     {
179       /* Check value is within 23-bits
180 	 (remembering that 2-bit shift right will occur).  */
181       if (value > 0x7fffff)
182         return _("21-bit offset out of range");
183     }
184   *valuep = (value & 0x7FFFFF);
185   return errmsg;
186 }
187 
188 /* Handle high().  */
189 
190 static const char *
191 parse_hi16 (CGEN_CPU_DESC cd,
192 	    const char **strp,
193 	    int opindex,
194 	    unsigned long *valuep)
195 {
196   if (strncasecmp (*strp, "%hi(", 4) == 0)
197     {
198       enum cgen_parse_operand_result result_type;
199       bfd_vma value;
200       const char *errmsg;
201 
202       *strp += 4;
203       errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_HI16,
204 				   & result_type, & value);
205       if (**strp != ')')
206 	return MISSING_CLOSING_PARENTHESIS;
207 
208       ++*strp;
209       if (errmsg == NULL
210   	  && result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
211 	{
212 	  /* If value has top-bit of %lo on, then it will
213 	     sign-propagate and so we compensate by adding
214 	     1 to the resultant %hi value.  */
215 	  if (value & 0x8000)
216 	    value += 0x10000;
217 	  value >>= 16;
218 	  value &= 0xffff;
219 	}
220       *valuep = value;
221 
222       return errmsg;
223     }
224 
225   /* We add %uhi in case a user just wants the high 16-bits or is using
226      an insn like ori for %lo which does not sign-propagate.  */
227   if (strncasecmp (*strp, "%uhi(", 5) == 0)
228     {
229       enum cgen_parse_operand_result result_type;
230       bfd_vma value;
231       const char *errmsg;
232 
233       *strp += 5;
234       errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_IQ2000_UHI16,
235 				   & result_type, & value);
236       if (**strp != ')')
237 	return MISSING_CLOSING_PARENTHESIS;
238 
239       ++*strp;
240       if (errmsg == NULL
241   	  && result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
242 	value >>= 16;
243 
244       value &= 0xffff;
245       *valuep = value;
246 
247       return errmsg;
248     }
249 
250   return parse_imm (cd, strp, opindex, valuep);
251 }
252 
253 /* Handle %lo in a signed context.
254    The signedness of the value doesn't matter to %lo(), but this also
255    handles the case where %lo() isn't present.  */
256 
257 static const char *
258 parse_lo16 (CGEN_CPU_DESC cd,
259 	    const char **strp,
260 	    int opindex,
261 	    unsigned long *valuep)
262 {
263   if (strncasecmp (*strp, "%lo(", 4) == 0)
264     {
265       const char *errmsg;
266       enum cgen_parse_operand_result result_type;
267       bfd_vma value;
268 
269       *strp += 4;
270       errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_LO16,
271 				   & result_type, & value);
272       if (**strp != ')')
273 	return MISSING_CLOSING_PARENTHESIS;
274       ++*strp;
275       if (errmsg == NULL
276 	  && result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
277 	value &= 0xffff;
278       *valuep = value;
279       return errmsg;
280     }
281 
282   return parse_imm (cd, strp, opindex, valuep);
283 }
284 
285 /* Handle %lo in a negated signed context.
286    The signedness of the value doesn't matter to %lo(), but this also
287    handles the case where %lo() isn't present.  */
288 
289 static const char *
290 parse_mlo16 (CGEN_CPU_DESC cd,
291 	     const char **strp,
292 	     int opindex,
293 	     unsigned long *valuep)
294 {
295   if (strncasecmp (*strp, "%lo(", 4) == 0)
296     {
297       const char *errmsg;
298       enum cgen_parse_operand_result result_type;
299       bfd_vma value;
300 
301       *strp += 4;
302       errmsg = cgen_parse_address (cd, strp, opindex, BFD_RELOC_LO16,
303 				   & result_type, & value);
304       if (**strp != ')')
305 	return MISSING_CLOSING_PARENTHESIS;
306       ++*strp;
307       if (errmsg == NULL
308 	  && result_type == CGEN_PARSE_OPERAND_RESULT_NUMBER)
309 	value = (-value) & 0xffff;
310       *valuep = value;
311       return errmsg;
312     }
313 
314   return parse_mimm (cd, strp, opindex, valuep);
315 }
316 
317 /* -- */
318 
319 const char * iq2000_cgen_parse_operand
320   (CGEN_CPU_DESC, int, const char **, CGEN_FIELDS *);
321 
322 /* Main entry point for operand parsing.
323 
324    This function is basically just a big switch statement.  Earlier versions
325    used tables to look up the function to use, but
326    - if the table contains both assembler and disassembler functions then
327      the disassembler contains much of the assembler and vice-versa,
328    - there's a lot of inlining possibilities as things grow,
329    - using a switch statement avoids the function call overhead.
330 
331    This function could be moved into `parse_insn_normal', but keeping it
332    separate makes clear the interface between `parse_insn_normal' and each of
333    the handlers.  */
334 
335 const char *
336 iq2000_cgen_parse_operand (CGEN_CPU_DESC cd,
337 			   int opindex,
338 			   const char ** strp,
339 			   CGEN_FIELDS * fields)
340 {
341   const char * errmsg = NULL;
342   /* Used by scalar operands that still need to be parsed.  */
343   long junk ATTRIBUTE_UNUSED;
344 
345   switch (opindex)
346     {
347     case IQ2000_OPERAND__INDEX :
348       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND__INDEX, (unsigned long *) (& fields->f_index));
349       break;
350     case IQ2000_OPERAND_BASE :
351       errmsg = cgen_parse_keyword (cd, strp, & iq2000_cgen_opval_gr_names, & fields->f_rs);
352       break;
353     case IQ2000_OPERAND_BASEOFF :
354       {
355         bfd_vma value = 0;
356         errmsg = cgen_parse_address (cd, strp, IQ2000_OPERAND_BASEOFF, 0, NULL,  & value);
357         fields->f_imm = value;
358       }
359       break;
360     case IQ2000_OPERAND_BITNUM :
361       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_BITNUM, (unsigned long *) (& fields->f_rt));
362       break;
363     case IQ2000_OPERAND_BYTECOUNT :
364       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_BYTECOUNT, (unsigned long *) (& fields->f_bytecount));
365       break;
366     case IQ2000_OPERAND_CAM_Y :
367       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_CAM_Y, (unsigned long *) (& fields->f_cam_y));
368       break;
369     case IQ2000_OPERAND_CAM_Z :
370       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_CAM_Z, (unsigned long *) (& fields->f_cam_z));
371       break;
372     case IQ2000_OPERAND_CM_3FUNC :
373       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_CM_3FUNC, (unsigned long *) (& fields->f_cm_3func));
374       break;
375     case IQ2000_OPERAND_CM_3Z :
376       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_CM_3Z, (unsigned long *) (& fields->f_cm_3z));
377       break;
378     case IQ2000_OPERAND_CM_4FUNC :
379       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_CM_4FUNC, (unsigned long *) (& fields->f_cm_4func));
380       break;
381     case IQ2000_OPERAND_CM_4Z :
382       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_CM_4Z, (unsigned long *) (& fields->f_cm_4z));
383       break;
384     case IQ2000_OPERAND_COUNT :
385       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_COUNT, (unsigned long *) (& fields->f_count));
386       break;
387     case IQ2000_OPERAND_EXECODE :
388       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_EXECODE, (unsigned long *) (& fields->f_excode));
389       break;
390     case IQ2000_OPERAND_HI16 :
391       errmsg = parse_hi16 (cd, strp, IQ2000_OPERAND_HI16, (unsigned long *) (& fields->f_imm));
392       break;
393     case IQ2000_OPERAND_IMM :
394       errmsg = parse_imm (cd, strp, IQ2000_OPERAND_IMM, (unsigned long *) (& fields->f_imm));
395       break;
396     case IQ2000_OPERAND_JMPTARG :
397       {
398         bfd_vma value = 0;
399         errmsg = cgen_parse_address (cd, strp, IQ2000_OPERAND_JMPTARG, 0, NULL,  & value);
400         fields->f_jtarg = value;
401       }
402       break;
403     case IQ2000_OPERAND_JMPTARGQ10 :
404       {
405         bfd_vma value = 0;
406         errmsg = parse_jtargq10 (cd, strp, IQ2000_OPERAND_JMPTARGQ10, 0, NULL,  & value);
407         fields->f_jtargq10 = value;
408       }
409       break;
410     case IQ2000_OPERAND_LO16 :
411       errmsg = parse_lo16 (cd, strp, IQ2000_OPERAND_LO16, (unsigned long *) (& fields->f_imm));
412       break;
413     case IQ2000_OPERAND_MASK :
414       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_MASK, (unsigned long *) (& fields->f_mask));
415       break;
416     case IQ2000_OPERAND_MASKL :
417       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_MASKL, (unsigned long *) (& fields->f_maskl));
418       break;
419     case IQ2000_OPERAND_MASKQ10 :
420       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_MASKQ10, (unsigned long *) (& fields->f_maskq10));
421       break;
422     case IQ2000_OPERAND_MASKR :
423       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_MASKR, (unsigned long *) (& fields->f_rs));
424       break;
425     case IQ2000_OPERAND_MLO16 :
426       errmsg = parse_mlo16 (cd, strp, IQ2000_OPERAND_MLO16, (unsigned long *) (& fields->f_imm));
427       break;
428     case IQ2000_OPERAND_OFFSET :
429       {
430         bfd_vma value = 0;
431         errmsg = cgen_parse_address (cd, strp, IQ2000_OPERAND_OFFSET, 0, NULL,  & value);
432         fields->f_offset = value;
433       }
434       break;
435     case IQ2000_OPERAND_RD :
436       errmsg = cgen_parse_keyword (cd, strp, & iq2000_cgen_opval_gr_names, & fields->f_rd);
437       break;
438     case IQ2000_OPERAND_RD_RS :
439       errmsg = cgen_parse_keyword (cd, strp, & iq2000_cgen_opval_gr_names, & fields->f_rd_rs);
440       break;
441     case IQ2000_OPERAND_RD_RT :
442       errmsg = cgen_parse_keyword (cd, strp, & iq2000_cgen_opval_gr_names, & fields->f_rd_rt);
443       break;
444     case IQ2000_OPERAND_RS :
445       errmsg = cgen_parse_keyword (cd, strp, & iq2000_cgen_opval_gr_names, & fields->f_rs);
446       break;
447     case IQ2000_OPERAND_RT :
448       errmsg = cgen_parse_keyword (cd, strp, & iq2000_cgen_opval_gr_names, & fields->f_rt);
449       break;
450     case IQ2000_OPERAND_RT_RS :
451       errmsg = cgen_parse_keyword (cd, strp, & iq2000_cgen_opval_gr_names, & fields->f_rt_rs);
452       break;
453     case IQ2000_OPERAND_SHAMT :
454       errmsg = cgen_parse_unsigned_integer (cd, strp, IQ2000_OPERAND_SHAMT, (unsigned long *) (& fields->f_shamt));
455       break;
456 
457     default :
458       /* xgettext:c-format */
459       fprintf (stderr, _("Unrecognized field %d while parsing.\n"), opindex);
460       abort ();
461   }
462 
463   return errmsg;
464 }
465 
466 cgen_parse_fn * const iq2000_cgen_parse_handlers[] =
467 {
468   parse_insn_normal,
469 };
470 
471 void
472 iq2000_cgen_init_asm (CGEN_CPU_DESC cd)
473 {
474   iq2000_cgen_init_opcode_table (cd);
475   iq2000_cgen_init_ibld_table (cd);
476   cd->parse_handlers = & iq2000_cgen_parse_handlers[0];
477   cd->parse_operand = iq2000_cgen_parse_operand;
478 #ifdef CGEN_ASM_INIT_HOOK
479 CGEN_ASM_INIT_HOOK
480 #endif
481 }
482 
483 
484 
485 /* Regex construction routine.
486 
487    This translates an opcode syntax string into a regex string,
488    by replacing any non-character syntax element (such as an
489    opcode) with the pattern '.*'
490 
491    It then compiles the regex and stores it in the opcode, for
492    later use by iq2000_cgen_assemble_insn
493 
494    Returns NULL for success, an error message for failure.  */
495 
496 char *
497 iq2000_cgen_build_insn_regex (CGEN_INSN *insn)
498 {
499   CGEN_OPCODE *opc = (CGEN_OPCODE *) CGEN_INSN_OPCODE (insn);
500   const char *mnem = CGEN_INSN_MNEMONIC (insn);
501   char rxbuf[CGEN_MAX_RX_ELEMENTS];
502   char *rx = rxbuf;
503   const CGEN_SYNTAX_CHAR_TYPE *syn;
504   int reg_err;
505 
506   syn = CGEN_SYNTAX_STRING (CGEN_OPCODE_SYNTAX (opc));
507 
508   /* Mnemonics come first in the syntax string.  */
509   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
510     return _("missing mnemonic in syntax string");
511   ++syn;
512 
513   /* Generate a case sensitive regular expression that emulates case
514      insensitive matching in the "C" locale.  We cannot generate a case
515      insensitive regular expression because in Turkish locales, 'i' and 'I'
516      are not equal modulo case conversion.  */
517 
518   /* Copy the literal mnemonic out of the insn.  */
519   for (; *mnem; mnem++)
520     {
521       char c = *mnem;
522 
523       if (ISALPHA (c))
524 	{
525 	  *rx++ = '[';
526 	  *rx++ = TOLOWER (c);
527 	  *rx++ = TOUPPER (c);
528 	  *rx++ = ']';
529 	}
530       else
531 	*rx++ = c;
532     }
533 
534   /* Copy any remaining literals from the syntax string into the rx.  */
535   for(; * syn != 0 && rx <= rxbuf + (CGEN_MAX_RX_ELEMENTS - 7 - 4); ++syn)
536     {
537       if (CGEN_SYNTAX_CHAR_P (* syn))
538 	{
539 	  char c = CGEN_SYNTAX_CHAR (* syn);
540 
541 	  switch (c)
542 	    {
543 	      /* Escape any regex metacharacters in the syntax.  */
544 	    case '.': case '[': case '\\':
545 	    case '*': case '^': case '$':
546 
547 #ifdef CGEN_ESCAPE_EXTENDED_REGEX
548 	    case '?': case '{': case '}':
549 	    case '(': case ')': case '*':
550 	    case '|': case '+': case ']':
551 #endif
552 	      *rx++ = '\\';
553 	      *rx++ = c;
554 	      break;
555 
556 	    default:
557 	      if (ISALPHA (c))
558 		{
559 		  *rx++ = '[';
560 		  *rx++ = TOLOWER (c);
561 		  *rx++ = TOUPPER (c);
562 		  *rx++ = ']';
563 		}
564 	      else
565 		*rx++ = c;
566 	      break;
567 	    }
568 	}
569       else
570 	{
571 	  /* Replace non-syntax fields with globs.  */
572 	  *rx++ = '.';
573 	  *rx++ = '*';
574 	}
575     }
576 
577   /* Trailing whitespace ok.  */
578   * rx++ = '[';
579   * rx++ = ' ';
580   * rx++ = '\t';
581   * rx++ = ']';
582   * rx++ = '*';
583 
584   /* But anchor it after that.  */
585   * rx++ = '$';
586   * rx = '\0';
587 
588   CGEN_INSN_RX (insn) = xmalloc (sizeof (regex_t));
589   reg_err = regcomp ((regex_t *) CGEN_INSN_RX (insn), rxbuf, REG_NOSUB);
590 
591   if (reg_err == 0)
592     return NULL;
593   else
594     {
595       static char msg[80];
596 
597       regerror (reg_err, (regex_t *) CGEN_INSN_RX (insn), msg, 80);
598       regfree ((regex_t *) CGEN_INSN_RX (insn));
599       free (CGEN_INSN_RX (insn));
600       (CGEN_INSN_RX (insn)) = NULL;
601       return msg;
602     }
603 }
604 
605 
606 /* Default insn parser.
607 
608    The syntax string is scanned and operands are parsed and stored in FIELDS.
609    Relocs are queued as we go via other callbacks.
610 
611    ??? Note that this is currently an all-or-nothing parser.  If we fail to
612    parse the instruction, we return 0 and the caller will start over from
613    the beginning.  Backtracking will be necessary in parsing subexpressions,
614    but that can be handled there.  Not handling backtracking here may get
615    expensive in the case of the m68k.  Deal with later.
616 
617    Returns NULL for success, an error message for failure.  */
618 
619 static const char *
620 parse_insn_normal (CGEN_CPU_DESC cd,
621 		   const CGEN_INSN *insn,
622 		   const char **strp,
623 		   CGEN_FIELDS *fields)
624 {
625   /* ??? Runtime added insns not handled yet.  */
626   const CGEN_SYNTAX *syntax = CGEN_INSN_SYNTAX (insn);
627   const char *str = *strp;
628   const char *errmsg;
629   const char *p;
630   const CGEN_SYNTAX_CHAR_TYPE * syn;
631 #ifdef CGEN_MNEMONIC_OPERANDS
632   /* FIXME: wip */
633   int past_opcode_p;
634 #endif
635 
636   /* For now we assume the mnemonic is first (there are no leading operands).
637      We can parse it without needing to set up operand parsing.
638      GAS's input scrubber will ensure mnemonics are lowercase, but we may
639      not be called from GAS.  */
640   p = CGEN_INSN_MNEMONIC (insn);
641   while (*p && TOLOWER (*p) == TOLOWER (*str))
642     ++p, ++str;
643 
644   if (* p)
645     return _("unrecognized instruction");
646 
647 #ifndef CGEN_MNEMONIC_OPERANDS
648   if (* str && ! ISSPACE (* str))
649     return _("unrecognized instruction");
650 #endif
651 
652   CGEN_INIT_PARSE (cd);
653   cgen_init_parse_operand (cd);
654 #ifdef CGEN_MNEMONIC_OPERANDS
655   past_opcode_p = 0;
656 #endif
657 
658   /* We don't check for (*str != '\0') here because we want to parse
659      any trailing fake arguments in the syntax string.  */
660   syn = CGEN_SYNTAX_STRING (syntax);
661 
662   /* Mnemonics come first for now, ensure valid string.  */
663   if (! CGEN_SYNTAX_MNEMONIC_P (* syn))
664     abort ();
665 
666   ++syn;
667 
668   while (* syn != 0)
669     {
670       /* Non operand chars must match exactly.  */
671       if (CGEN_SYNTAX_CHAR_P (* syn))
672 	{
673 	  /* FIXME: While we allow for non-GAS callers above, we assume the
674 	     first char after the mnemonic part is a space.  */
675 	  /* FIXME: We also take inappropriate advantage of the fact that
676 	     GAS's input scrubber will remove extraneous blanks.  */
677 	  if (TOLOWER (*str) == TOLOWER (CGEN_SYNTAX_CHAR (* syn)))
678 	    {
679 #ifdef CGEN_MNEMONIC_OPERANDS
680 	      if (CGEN_SYNTAX_CHAR(* syn) == ' ')
681 		past_opcode_p = 1;
682 #endif
683 	      ++ syn;
684 	      ++ str;
685 	    }
686 	  else if (*str)
687 	    {
688 	      /* Syntax char didn't match.  Can't be this insn.  */
689 	      static char msg [80];
690 
691 	      /* xgettext:c-format */
692 	      sprintf (msg, _("syntax error (expected char `%c', found `%c')"),
693 		       CGEN_SYNTAX_CHAR(*syn), *str);
694 	      return msg;
695 	    }
696 	  else
697 	    {
698 	      /* Ran out of input.  */
699 	      static char msg [80];
700 
701 	      /* xgettext:c-format */
702 	      sprintf (msg, _("syntax error (expected char `%c', found end of instruction)"),
703 		       CGEN_SYNTAX_CHAR(*syn));
704 	      return msg;
705 	    }
706 	  continue;
707 	}
708 
709       /* We have an operand of some sort.  */
710       errmsg = cd->parse_operand (cd, CGEN_SYNTAX_FIELD (*syn),
711 					  &str, fields);
712       if (errmsg)
713 	return errmsg;
714 
715       /* Done with this operand, continue with next one.  */
716       ++ syn;
717     }
718 
719   /* If we're at the end of the syntax string, we're done.  */
720   if (* syn == 0)
721     {
722       /* FIXME: For the moment we assume a valid `str' can only contain
723 	 blanks now.  IE: We needn't try again with a longer version of
724 	 the insn and it is assumed that longer versions of insns appear
725 	 before shorter ones (eg: lsr r2,r3,1 vs lsr r2,r3).  */
726       while (ISSPACE (* str))
727 	++ str;
728 
729       if (* str != '\0')
730 	return _("junk at end of line"); /* FIXME: would like to include `str' */
731 
732       return NULL;
733     }
734 
735   /* We couldn't parse it.  */
736   return _("unrecognized instruction");
737 }
738 
739 /* Main entry point.
740    This routine is called for each instruction to be assembled.
741    STR points to the insn to be assembled.
742    We assume all necessary tables have been initialized.
743    The assembled instruction, less any fixups, is stored in BUF.
744    Remember that if CGEN_INT_INSN_P then BUF is an int and thus the value
745    still needs to be converted to target byte order, otherwise BUF is an array
746    of bytes in target byte order.
747    The result is a pointer to the insn's entry in the opcode table,
748    or NULL if an error occured (an error message will have already been
749    printed).
750 
751    Note that when processing (non-alias) macro-insns,
752    this function recurses.
753 
754    ??? It's possible to make this cpu-independent.
755    One would have to deal with a few minor things.
756    At this point in time doing so would be more of a curiosity than useful
757    [for example this file isn't _that_ big], but keeping the possibility in
758    mind helps keep the design clean.  */
759 
760 const CGEN_INSN *
761 iq2000_cgen_assemble_insn (CGEN_CPU_DESC cd,
762 			   const char *str,
763 			   CGEN_FIELDS *fields,
764 			   CGEN_INSN_BYTES_PTR buf,
765 			   char **errmsg)
766 {
767   const char *start;
768   CGEN_INSN_LIST *ilist;
769   const char *parse_errmsg = NULL;
770   const char *insert_errmsg = NULL;
771   int recognized_mnemonic = 0;
772 
773   /* Skip leading white space.  */
774   while (ISSPACE (* str))
775     ++ str;
776 
777   /* The instructions are stored in hashed lists.
778      Get the first in the list.  */
779   ilist = CGEN_ASM_LOOKUP_INSN (cd, str);
780 
781   /* Keep looking until we find a match.  */
782   start = str;
783   for ( ; ilist != NULL ; ilist = CGEN_ASM_NEXT_INSN (ilist))
784     {
785       const CGEN_INSN *insn = ilist->insn;
786       recognized_mnemonic = 1;
787 
788 #ifdef CGEN_VALIDATE_INSN_SUPPORTED
789       /* Not usually needed as unsupported opcodes
790 	 shouldn't be in the hash lists.  */
791       /* Is this insn supported by the selected cpu?  */
792       if (! iq2000_cgen_insn_supported (cd, insn))
793 	continue;
794 #endif
795       /* If the RELAXED attribute is set, this is an insn that shouldn't be
796 	 chosen immediately.  Instead, it is used during assembler/linker
797 	 relaxation if possible.  */
798       if (CGEN_INSN_ATTR_VALUE (insn, CGEN_INSN_RELAXED) != 0)
799 	continue;
800 
801       str = start;
802 
803       /* Skip this insn if str doesn't look right lexically.  */
804       if (CGEN_INSN_RX (insn) != NULL &&
805 	  regexec ((regex_t *) CGEN_INSN_RX (insn), str, 0, NULL, 0) == REG_NOMATCH)
806 	continue;
807 
808       /* Allow parse/insert handlers to obtain length of insn.  */
809       CGEN_FIELDS_BITSIZE (fields) = CGEN_INSN_BITSIZE (insn);
810 
811       parse_errmsg = CGEN_PARSE_FN (cd, insn) (cd, insn, & str, fields);
812       if (parse_errmsg != NULL)
813 	continue;
814 
815       /* ??? 0 is passed for `pc'.  */
816       insert_errmsg = CGEN_INSERT_FN (cd, insn) (cd, insn, fields, buf,
817 						 (bfd_vma) 0);
818       if (insert_errmsg != NULL)
819         continue;
820 
821       /* It is up to the caller to actually output the insn and any
822          queued relocs.  */
823       return insn;
824     }
825 
826   {
827     static char errbuf[150];
828 #ifdef CGEN_VERBOSE_ASSEMBLER_ERRORS
829     const char *tmp_errmsg;
830 
831     /* If requesting verbose error messages, use insert_errmsg.
832        Failing that, use parse_errmsg.  */
833     tmp_errmsg = (insert_errmsg ? insert_errmsg :
834 		  parse_errmsg ? parse_errmsg :
835 		  recognized_mnemonic ?
836 		  _("unrecognized form of instruction") :
837 		  _("unrecognized instruction"));
838 
839     if (strlen (start) > 50)
840       /* xgettext:c-format */
841       sprintf (errbuf, "%s `%.50s...'", tmp_errmsg, start);
842     else
843       /* xgettext:c-format */
844       sprintf (errbuf, "%s `%.50s'", tmp_errmsg, start);
845 #else
846     if (strlen (start) > 50)
847       /* xgettext:c-format */
848       sprintf (errbuf, _("bad instruction `%.50s...'"), start);
849     else
850       /* xgettext:c-format */
851       sprintf (errbuf, _("bad instruction `%.50s'"), start);
852 #endif
853 
854     *errmsg = errbuf;
855     return NULL;
856   }
857 }
858