/* tc-cris.c -- Assembler code for the CRIS CPU core. Copyright (C) 2000 Free Software Foundation, Inc. Contributed by Axis Communications AB, Lund, Sweden. Originally written for GAS 1.38.1 by Mikael Asker. Updated, BFDized and GNUified by Hans-Peter Nilsson. This file is part of GAS, the GNU Assembler. GAS 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, or (at your option) any later version. GAS 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 GAS; see the file COPYING. If not, write to the Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ #include #include #include "as.h" #include "subsegs.h" #include "opcode/cris.h" /* Conventions used here: Generally speaking, pointers to binutils types such as "fragS" and "expressionS" get parameter and variable names ending in "P", such as "fragP", to harmonize with the rest of the binutils code. Other pointers get a "p" suffix, such as "bufp". Any function or type-name that could clash with a current or future binutils or GAS function get a "cris_" prefix. */ /* This might be CRIS_INSN_NONE if we're assembling a prefix-insn only. Note that some prefix-insns might be assembled as CRIS_INSN_NORMAL. */ enum cris_insn_kind { CRIS_INSN_NORMAL, CRIS_INSN_NONE, CRIS_INSN_BRANCH }; /* An instruction will have one of these prefixes. Although the same bit-pattern, we handle BDAP with an immediate expression (eventually quick or [pc+]) different from when we only have register expressions. */ enum prefix_kind { PREFIX_NONE, PREFIX_BDAP_IMM, PREFIX_BDAP, PREFIX_BIAP, PREFIX_DIP, PREFIX_PUSH }; /* The prefix for an instruction. */ struct cris_prefix { enum prefix_kind kind; int base_reg_number; unsigned int opcode; /* There might be an expression to be evaluated, like I in [rN+I]. */ expressionS expr; /* If there's an expression, we might need a relocation. Here's the type of what relocation to start relaxaton with. The relocation is assumed to start immediately after the prefix insn, so we don't provide an offset. */ enum bfd_reloc_code_real reloc; }; /* The description of the instruction being assembled. */ struct cris_instruction { /* If CRIS_INSN_NONE, then this insn is of zero length. */ enum cris_insn_kind insn_type; /* If a special register was mentioned, this is its description, else it is NULL. */ const struct cris_spec_reg *spec_reg; unsigned int opcode; /* An insn may have at most one expression; theoretically there could be another in its prefix (but I don't see how that could happen). */ expressionS expr; /* The expression might need a relocation. Here's one to start relaxation with. */ enum bfd_reloc_code_real reloc; /* The size in bytes of an immediate expression, or zero in nonapplicable. */ int imm_oprnd_size; }; static void cris_process_instruction PARAMS ((char *, struct cris_instruction *, struct cris_prefix *)); static int get_bwd_size_modifier PARAMS ((char **, int *)); static int get_bw_size_modifier PARAMS ((char **, int *)); static int get_gen_reg PARAMS ((char **, int *)); static int get_spec_reg PARAMS ((char **, const struct cris_spec_reg **)); static int get_autoinc_prefix_or_indir_op PARAMS ((char **, struct cris_prefix *, int *, int *, int *, expressionS *)); static int get_3op_or_dip_prefix_op PARAMS ((char **, struct cris_prefix *)); static int cris_get_expression PARAMS ((char **, expressionS *)); static int get_flags PARAMS ((char **, int *)); static void gen_bdap PARAMS ((int, expressionS *)); static int branch_disp PARAMS ((int)); static void gen_cond_branch_32 PARAMS ((char *, char *, fragS *, symbolS *, symbolS *, long int)); static void cris_number_to_imm PARAMS ((char *, long, int, fixS *)); static void cris_create_short_jump PARAMS ((char *, addressT, addressT, fragS *, symbolS *)); /* Handle to the opcode hash table. */ static struct hash_control *op_hash = NULL; const pseudo_typeS md_pseudo_table[] = { {"dword", cons, 4}, {NULL, 0, 0} }; static int warn_for_branch_expansion = 0; const char cris_comment_chars[] = ";"; /* This array holds the chars that only start a comment at the beginning of a line. If the line seems to have the form '# 123 filename' .line and .file directives will appear in the pre-processed output. */ /* Note that input_file.c hand-checks for '#' at the beginning of the first line of the input file. This is because the compiler outputs #NO_APP at the beginning of its output. */ /* Also note that slash-star will always start a comment. */ const char line_comment_chars[] = "#"; const char line_separator_chars[] = "@"; /* Now all floating point support is shut off. See md_atof. */ const char EXP_CHARS[] = ""; const char FLT_CHARS[] = ""; /* For CRIS, we encode the relax_substateTs (in e.g. fr_substate) as: 2 1 0 ---/ /--+-----------------+-----------------+-----------------+ | what state ? | how long ? | ---/ /--+-----------------+-----------------+-----------------+ The "how long" bits are 00 = byte, 01 = word, 10 = dword (long). This is a Un*x convention. Not all lengths are legit for a given value of (what state). Groups for CRIS address relaxing: 1. Bcc length: byte, word, 10-byte expansion 2. BDAP length: byte, word, dword */ #define STATE_CONDITIONAL_BRANCH (1) #define STATE_BASE_PLUS_DISP_PREFIX (2) #define STATE_LENGTH_MASK (3) #define STATE_BYTE (0) #define STATE_WORD (1) #define STATE_DWORD (2) /* Symbol undefined. */ #define STATE_UNDF (3) #define STATE_MAX_LENGTH (3) /* These displacements are relative to the adress following the opcode word of the instruction. The first letter is Byte, Word. The 2nd letter is Forward, Backward. */ #define BRANCH_BF ( 254) #define BRANCH_BB (-256) #define BRANCH_WF (2 + 32767) #define BRANCH_WB (2 + -32768) #define BDAP_BF ( 127) #define BDAP_BB (-128) #define BDAP_WF ( 32767) #define BDAP_WB (-32768) #define ENCODE_RELAX(what, length) (((what) << 2) + (length)) const relax_typeS md_cris_relax_table[] = { /* Error sentinel (0, 0). */ {1, 1, 0, 0}, /* Unused (0, 1). */ {1, 1, 0, 0}, /* Unused (0, 2). */ {1, 1, 0, 0}, /* Unused (0, 3). */ {1, 1, 0, 0}, /* Bcc o (1, 0). */ {BRANCH_BF, BRANCH_BB, 0, ENCODE_RELAX (1, 1)}, /* Bcc [PC+] (1, 1). */ {BRANCH_WF, BRANCH_WB, 2, ENCODE_RELAX (1, 2)}, /* BEXT/BWF, BA, JUMP (external), JUMP (always), Bnot_cc, JUMP (default) (1, 2). */ {0, 0, 10, 0}, /* Unused (1, 3). */ {1, 1, 0, 0}, /* BDAP o (2, 0). */ {BDAP_BF, BDAP_BB, 0, ENCODE_RELAX (2, 1)}, /* BDAP.[bw] [PC+] (2, 1). */ {BDAP_WF, BDAP_WB, 2, ENCODE_RELAX (2, 2)}, /* BDAP.d [PC+] (2, 2). */ {0, 0, 4, 0} }; #undef BRANCH_BF #undef BRANCH_BB #undef BRANCH_WF #undef BRANCH_WB #undef BDAP_BF #undef BDAP_BB #undef BDAP_WF #undef BDAP_WB /* Target-specific multicharacter options, not const-declared at usage in 2.9.1 and CVS of 2000-02-16. */ struct option md_longopts[] = { {NULL, no_argument, NULL, 0} }; /* Not const-declared at usage in 2.9.1. */ size_t md_longopts_size = sizeof (md_longopts); const char *md_shortopts = "hHN"; /* At first glance, this may seems wrong and should be 4 (ba + nop); but since a short_jump must skip a *number* of long jumps, it must also be a long jump. Here, we hope to make it a "ba [16bit_offs]" and a "nop" for the delay slot and hope that the jump table at most needs 32767/4=8191 long-jumps. A branch is better than a jump, since it is relative; we will not have a reloc to fix up somewhere. Note that we can't add relocs, because relaxation uses these fixed numbers, and md_create_short_jump is called after relaxation. */ const int md_short_jump_size = 6; const int md_long_jump_size = 6; /* Report output format. */ const char * cris_target_format () { switch (OUTPUT_FLAVOR) { case bfd_target_aout_flavour: return "a.out-cris"; case bfd_target_elf_flavour: return "elf32-cris"; default: abort (); return NULL; } } /* Prepare machine-dependent frags for relaxation. Called just before relaxation starts. Any symbol that is now undefined will not become defined. Return the correct fr_subtype in the frag. Return the initial "guess for fr_var" to caller. The guess for fr_var is *actually* the growth beyond fr_fix. Whatever we do to grow fr_fix or fr_var contributes to our returned value. Although it may not be explicit in the frag, pretend fr_var starts with a value. */ int md_estimate_size_before_relax (fragP, segment_type) fragS *fragP; /* The segment is either N_DATA or N_TEXT. */ segT segment_type; { int old_fr_fix; old_fr_fix = fragP->fr_fix; switch (fragP->fr_subtype) { case ENCODE_RELAX (STATE_CONDITIONAL_BRANCH, STATE_UNDF): if (S_GET_SEGMENT (fragP->fr_symbol) == segment_type) { /* The symbol lies in the same segment - a relaxable case. */ fragP->fr_subtype = ENCODE_RELAX (STATE_CONDITIONAL_BRANCH, STATE_BYTE); } else { /* Unknown or not the same segment, so not relaxable. */ char *writep; /* A small branch-always (2 bytes) to the "real" branch instruction, plus a delay-slot nop (2 bytes), plus a jump (2 plus 4 bytes). See gen_cond_branch_32. */ fragP->fr_fix += 2 + 2 + 2 + 4; writep = fragP->fr_literal + old_fr_fix; gen_cond_branch_32 (fragP->fr_opcode, writep, fragP, fragP->fr_symbol, (symbolS *) NULL, fragP->fr_offset); frag_wane (fragP); } break; case ENCODE_RELAX (STATE_CONDITIONAL_BRANCH, STATE_BYTE): case ENCODE_RELAX (STATE_CONDITIONAL_BRANCH, STATE_WORD): /* We *might* give a better initial guess if we peek at offsets now, but the caller will relax correctly and without this, so don't bother. */ break; case ENCODE_RELAX (STATE_BASE_PLUS_DISP_PREFIX, STATE_UNDF): /* Note that we can not do anything sane with relaxing [rX + a_known_symbol_in_text], it will have to be a 32-bit value. We could play tricks with managing a constant pool and make a_known_symbol_in_text a "bdap [pc + offset]" pointing there, but that's pointless, it can only be longer and slower. Off-topic: If PIC becomes *really* important, and has to be done in the assembler and linker only (which would be weird or clueless), we can so something. Imagine: move.x [r + 32_bit_symbol],r move.x [32_bit_symbol],r move.x 32_bit_symbol,r can be shortened by a word (8-bit offset) if we are close to the symbol or keep its length (16-bit offset) or be a word longer (32-bit offset). Then change the 32_bit_symbol into a "bdap [pc + offset]", and put the offset to the 32_bit_symbol in "offset". Weird, to say the least, and we still have to add support for a PC-relative relocation in the loader (shared libraries). But it's an interesting thought. */ if (S_GET_SEGMENT (fragP->fr_symbol) != absolute_section) { /* Go for dword if not absolute or same segment. */ fragP->fr_subtype = ENCODE_RELAX (STATE_BASE_PLUS_DISP_PREFIX, STATE_DWORD); fragP->fr_var += 4; } else { /* Absolute expression. */ long int value; value = S_GET_VALUE (fragP->fr_symbol) + fragP->fr_offset; if (value >= -128 && value <= 127) { /* Byte displacement. */ (fragP->fr_opcode)[0] = value; } else { /* Word or dword displacement. */ int pow2_of_size = 1; char *writep; if (value < -32768 || value > 32767) { /* Outside word range, make it a dword. */ pow2_of_size = 2; } /* Modify the byte-offset BDAP into a word or dword offset BDAP. Or really, a BDAP rX,8bit into a BDAP.[wd] rX,[PC+] followed by a word or dword. */ (fragP->fr_opcode)[0] = BDAP_PC_LOW + pow2_of_size * 16; /* Keep the register number in the highest four bits. */ (fragP->fr_opcode)[1] &= 0xF0; (fragP->fr_opcode)[1] |= BDAP_INCR_HIGH; /* It grew by two or four bytes. */ fragP->fr_fix += 1 << pow2_of_size; writep = fragP->fr_literal + old_fr_fix; md_number_to_chars (writep, value, 1 << pow2_of_size); } frag_wane (fragP); } break; default: BAD_CASE (fragP->fr_subtype); } return fragP->fr_var + (fragP->fr_fix - old_fr_fix); } /* Perform post-processing of machine-dependent frags after relaxation. Called after relaxation is finished. In: Address of frag. fr_type == rs_machine_dependent. fr_subtype is what the address relaxed to. Out: Any fixS:s and constants are set up. The caller will turn the frag into a ".space 0". */ void md_convert_frag (abfd, sec, fragP) bfd *abfd ATTRIBUTE_UNUSED; segT sec ATTRIBUTE_UNUSED; fragS *fragP; { /* Pointer to first byte in variable-sized part of the frag. */ char *var_partp; /* Pointer to first opcode byte in frag. */ char *opcodep; /* Used to check integrity of the relaxation. One of 2 = long, 1 = word, or 0 = byte. */ int length_code; /* Size in bytes of variable-sized part of frag. */ int var_part_size = 0; /* This is part of *fragP. It contains all information about addresses and offsets to varying parts. */ symbolS *symbolP; unsigned long var_part_offset; /* Where, in file space, is _var of *fragP? */ unsigned long address_of_var_part = 0; /* Where, in file space, does addr point? */ unsigned long target_address; know (fragP->fr_type == rs_machine_dependent); length_code = fragP->fr_subtype & STATE_LENGTH_MASK; know (length_code >= 0 && length_code < STATE_MAX_LENGTH); var_part_offset = fragP->fr_fix; var_partp = fragP->fr_literal + var_part_offset; opcodep = fragP->fr_opcode; symbolP = fragP->fr_symbol; target_address = (symbolP ? S_GET_VALUE (symbolP) + symbol_get_frag(fragP->fr_symbol)->fr_address : 0 ) + fragP->fr_offset; address_of_var_part = fragP->fr_address + var_part_offset; switch (fragP->fr_subtype) { case ENCODE_RELAX (STATE_CONDITIONAL_BRANCH, STATE_BYTE): opcodep[0] = branch_disp ((target_address - address_of_var_part)); var_part_size = 0; break; case ENCODE_RELAX (STATE_CONDITIONAL_BRANCH, STATE_WORD): /* We had a quick immediate branch, now turn it into a word one i.e. a PC autoincrement. */ opcodep[0] = BRANCH_PC_LOW; opcodep[1] &= 0xF0; opcodep[1] |= BRANCH_INCR_HIGH; md_number_to_chars (var_partp, (long) (target_address - (address_of_var_part + 2)), 2); var_part_size = 2; break; case ENCODE_RELAX (STATE_CONDITIONAL_BRANCH, STATE_DWORD): gen_cond_branch_32 (fragP->fr_opcode, var_partp, fragP, fragP->fr_symbol, (symbolS *) NULL, fragP->fr_offset); /* Ten bytes added: a branch, nop and a jump. */ var_part_size = 2 + 2 + 4 + 2; break; case ENCODE_RELAX (STATE_BASE_PLUS_DISP_PREFIX, STATE_BYTE): var_partp[0] = target_address - (address_of_var_part + 1); var_part_size = 0; break; case ENCODE_RELAX (STATE_BASE_PLUS_DISP_PREFIX, STATE_WORD): /* We had a BDAP 8-bit "quick immediate", now turn it into a 16-bit one that uses PC autoincrement. */ opcodep[0] = BDAP_PC_LOW + (1 << 4); opcodep[1] &= 0xF0; opcodep[1] |= BDAP_INCR_HIGH; md_number_to_chars (var_partp, (long) (target_address), 2); var_part_size = 2; break; case ENCODE_RELAX (STATE_BASE_PLUS_DISP_PREFIX, STATE_DWORD): /* We had a BDAP 16-bit "word", change the offset to a dword. */ opcodep[0] = BDAP_PC_LOW + (2 << 4); opcodep[1] &= 0xF0; opcodep[1] |= BDAP_INCR_HIGH; if (fragP->fr_symbol == NULL) md_number_to_chars (var_partp, fragP->fr_offset, 4); else fix_new (fragP, var_partp - fragP->fr_literal, 4, fragP->fr_symbol, fragP->fr_offset, 0, BFD_RELOC_32); var_part_size = 4; break; default: BAD_CASE (fragP->fr_subtype); break; } fragP->fr_fix += var_part_size; } /* Generate a short jump around a secondary jump table. Used by md_create_long_jump. This used to be md_create_short_jump, but is now called from md_create_long_jump instead, when sufficient. since the sizes of the jumps are the same. It used to be brittle, making possibilities for creating bad code. */ static void cris_create_short_jump (storep, from_addr, to_addr, fragP, to_symbol) char *storep; addressT from_addr; addressT to_addr; fragS *fragP ATTRIBUTE_UNUSED; symbolS *to_symbol ATTRIBUTE_UNUSED; { long int distance; distance = to_addr - from_addr; if (-254 <= distance && distance <= 256) { /* Create a "short" short jump: "BA distance - 2". */ storep[0] = branch_disp (distance - 2); storep[1] = BA_QUICK_HIGH; /* A nop for the delay slot. */ md_number_to_chars (storep + 2, NOP_OPCODE, 2); /* The extra word should be filled with something sane too. Make it a nop to keep disassembly sane. */ md_number_to_chars (storep + 4, NOP_OPCODE, 2); } else { /* Make it a "long" short jump: "BA (PC+)". */ md_number_to_chars (storep, BA_PC_INCR_OPCODE, 2); /* ".WORD distance - 4". */ md_number_to_chars (storep + 2, (long) (distance - 4), 2); /* A nop for the delay slot. */ md_number_to_chars (storep + 4, NOP_OPCODE, 2); } } /* Generate a long jump in a secondary jump table. storep Where to store the jump instruction. from_addr Address of the jump instruction. to_addr Destination address of the jump. fragP Which frag the destination address operand lies in. to_symbol Destination symbol. */ void md_create_long_jump (storep, from_addr, to_addr, fragP, to_symbol) char *storep; addressT from_addr; addressT to_addr; fragS *fragP; symbolS *to_symbol; { long int distance; distance = to_addr - from_addr; if (-32763 <= distance && distance <= 32772) { /* Then make it a "short" long jump. */ cris_create_short_jump (storep, from_addr, to_addr, fragP, to_symbol); } else { /* We have a "long" long jump: "JUMP (PC+)". */ md_number_to_chars (storep, JUMP_PC_INCR_OPCODE, 2); /* Follow with a ".DWORD to_addr". */ fix_new (fragP, storep + 2 - fragP->fr_literal, 4, to_symbol, 0, 0, BFD_RELOC_32); } } /* Port-specific assembler initialization. */ void md_begin () { const char *hashret = NULL; int i = 0; /* Set up a hash table for the instructions. */ op_hash = hash_new (); if (op_hash == NULL) as_fatal (_("Virtual memory exhausted")); while (cris_opcodes[i].name != NULL) { const char *name = cris_opcodes[i].name; hashret = hash_insert (op_hash, name, (PTR) &cris_opcodes[i]); if (hashret != NULL && *hashret != '\0') as_fatal (_("Can't hash `%s': %s\n"), cris_opcodes[i].name, *hashret == 0 ? _("(unknown reason)") : hashret); do { if (cris_opcodes[i].match & cris_opcodes[i].lose) as_fatal (_("Buggy opcode: `%s' \"%s\"\n"), cris_opcodes[i].name, cris_opcodes[i].args); ++i; } while (cris_opcodes[i].name != NULL && strcmp (cris_opcodes[i].name, name) == 0); } } /* Assemble a source line. */ void md_assemble (str) char *str; { struct cris_instruction output_instruction; struct cris_prefix prefix; char *opcodep; char *p; know (str); /* Do the low-level grunt - assemble to bits and split up into a prefix and ordinary insn. */ cris_process_instruction (str, &output_instruction, &prefix); /* Handle any prefixes to the instruction. */ switch (prefix.kind) { case PREFIX_NONE: break; /* When the expression is unknown for a BDAP, it can need 0, 2 or 4 extra bytes, so we handle it separately. */ case PREFIX_BDAP_IMM: gen_bdap (prefix.base_reg_number, &prefix.expr); break; case PREFIX_BDAP: case PREFIX_BIAP: case PREFIX_DIP: opcodep = frag_more (2); /* Output the prefix opcode. */ md_number_to_chars (opcodep, (long) prefix.opcode, 2); /* This only happens for DIP, but is ok for the others as they have no reloc. */ if (prefix.reloc != BFD_RELOC_NONE) { /* Output an absolute mode address. */ p = frag_more (4); fix_new_exp (frag_now, (p - frag_now->fr_literal), 4, &prefix.expr, 0, prefix.reloc); } break; case PREFIX_PUSH: opcodep = frag_more (2); /* Output the prefix opcode. Being a "push", we add the negative size of the register to "sp". */ if (output_instruction.spec_reg != NULL) { /* Special register. */ opcodep[0] = -output_instruction.spec_reg->reg_size; } else { /* General register. */ opcodep[0] = -4; } opcodep[1] = (REG_SP << 4) + (BDAP_QUICK_OPCODE >> 8); break; default: BAD_CASE (prefix.kind); } /* If we only had a prefix insn, we're done. */ if (output_instruction.insn_type == CRIS_INSN_NONE) return; /* Done with the prefix. Continue with the main instruction. */ opcodep = frag_more (2); /* Output the instruction opcode. */ md_number_to_chars (opcodep, (long) (output_instruction.opcode), 2); /* Output the symbol-dependent instruction stuff. */ if (output_instruction.insn_type == CRIS_INSN_BRANCH) { segT to_seg = absolute_section; int is_undefined = 0; int length_code; if (output_instruction.expr.X_op != O_constant) { to_seg = S_GET_SEGMENT (output_instruction.expr.X_add_symbol); if (to_seg == undefined_section) is_undefined = 1; } if (output_instruction.expr.X_op == O_constant || to_seg == now_seg || is_undefined) { /* If is_undefined, then the expression may BECOME now_seg. */ length_code = is_undefined ? STATE_UNDF : STATE_BYTE; /* Make room for max ten bytes of variable length. */ frag_var (rs_machine_dependent, 10, 0, ENCODE_RELAX (STATE_CONDITIONAL_BRANCH, length_code), output_instruction.expr.X_add_symbol, output_instruction.expr.X_add_number, opcodep); } else { /* We have: to_seg != now_seg && to_seg != undefined_section. This means it is a branch to a known symbol in another section. Code in data? Weird but valid. Emit a 32-bit branch. */ gen_cond_branch_32 (opcodep, frag_more (10), frag_now, output_instruction.expr.X_add_symbol, (symbolS *) NULL, output_instruction.expr.X_add_number); } } else { if (output_instruction.imm_oprnd_size > 0) { /* The intruction has an immediate operand. */ enum bfd_reloc_code_real reloc = 0; switch (output_instruction.imm_oprnd_size) { /* Any byte-size immediate constants are treated as word-size. FIXME: Thus overflow check does not work correctly. */ case 2: reloc = BFD_RELOC_16; break; case 4: reloc = BFD_RELOC_32; break; default: BAD_CASE (output_instruction.imm_oprnd_size); } p = frag_more (output_instruction.imm_oprnd_size); fix_new_exp (frag_now, (p - frag_now->fr_literal), output_instruction.imm_oprnd_size, &output_instruction.expr, 0, reloc); } else if (output_instruction.reloc != BFD_RELOC_NONE) { /* An immediate operand that has a relocation and needs to be processed further. */ /* It is important to use fix_new_exp here and everywhere else (and not fix_new), as fix_new_exp can handle "difference expressions" - where the expression contains a difference of two symbols in the same segment. */ fix_new_exp (frag_now, (opcodep - frag_now->fr_literal), 2, &output_instruction.expr, 0, output_instruction.reloc); } } } /* Low level text-to-bits assembly. */ static void cris_process_instruction (insn_text, out_insnp, prefixp) char *insn_text; struct cris_instruction *out_insnp; struct cris_prefix *prefixp; { char *s; char modified_char = 0; const char *args; struct cris_opcode *instruction; char *operands; int match = 0; int mode; int regno; int size_bits; /* Reset these fields to a harmless state in case we need to return in error. */ prefixp->kind = PREFIX_NONE; prefixp->reloc = BFD_RELOC_NONE; out_insnp->insn_type = CRIS_INSN_NORMAL; out_insnp->imm_oprnd_size = 0; /* Find the end of the opcode mnemonic. We assume (true in 2.9.1) that the caller has translated the opcode to lower-case, up to the first non-letter. */ for (operands = insn_text; islower (*operands); ++operands) ; /* Terminate the opcode after letters, but save the character there if it was of significance. */ switch (*operands) { case '\0': break; case '.': /* Put back the modified character later. */ modified_char = *operands; /* Fall through. */ case ' ': /* Consume the character after the mnemonic and replace it with '\0'. */ *operands++ = '\0'; break; default: as_bad (_("Unknown opcode: `%s'"), insn_text); return; } /* Find the instruction. */ instruction = (struct cris_opcode *) hash_find (op_hash, insn_text); if (instruction == NULL) { as_bad (_("Unknown opcode: `%s'"), insn_text); return; } /* Put back the modified character. */ switch (modified_char) { case 0: break; default: *--operands = modified_char; } /* Try to match an opcode table slot. */ for (s = operands;;) { int imm_expr_found; /* Initialize *prefixp, perhaps after being modified for a "near match". */ prefixp->kind = PREFIX_NONE; prefixp->reloc = BFD_RELOC_NONE; /* Initialize *out_insnp. */ memset (out_insnp, 0, sizeof (*out_insnp)); out_insnp->opcode = instruction->match; out_insnp->reloc = BFD_RELOC_NONE; out_insnp->insn_type = CRIS_INSN_NORMAL; out_insnp->imm_oprnd_size = 0; imm_expr_found = 0; /* Build the opcode, checking as we go to make sure that the operands match. */ for (args = instruction->args;; ++args) { switch (*args) { case '\0': /* If we've come to the end of arguments, we're done. */ if (*s == '\0') match = 1; break; case '!': /* Non-matcher character for disassembly. Ignore it here. */ continue; case ',': case ' ': /* These must match exactly. */ if (*s++ == *args) continue; break; case 'B': /* This is not really an operand, but causes a "BDAP -size,SP" prefix to be output, for PUSH instructions. */ prefixp->kind = PREFIX_PUSH; continue; case 'b': /* This letter marks an operand that should not be matched in the assembler. It is a branch with 16-bit displacement. The assembler will create them from the 8-bit flavor when necessary. The assembler does not support the [rN+] operand, as the [r15+] that is generated for 16-bit displacements. */ break; case 'c': /* A 5-bit unsigned immediate in bits <4:0>. */ if (! cris_get_expression (&s, &out_insnp->expr)) break; else { if (out_insnp->expr.X_op == O_constant && (out_insnp->expr.X_add_number < 0 || out_insnp->expr.X_add_number > 31)) as_bad (_("Immediate value not in 5 bit unsigned range: %ld"), out_insnp->expr.X_add_number); out_insnp->reloc = BFD_RELOC_CRIS_UNSIGNED_5; continue; } case 'C': /* A 4-bit unsigned immediate in bits <3:0>. */ if (! cris_get_expression (&s, &out_insnp->expr)) break; else { if (out_insnp->expr.X_op == O_constant && (out_insnp->expr.X_add_number < 0 || out_insnp->expr.X_add_number > 15)) as_bad (_("Immediate value not in 4 bit unsigned range: %ld"), out_insnp->expr.X_add_number); out_insnp->reloc = BFD_RELOC_CRIS_UNSIGNED_4; continue; } case 'D': /* General register in bits <15:12> and <3:0>. */ if (! get_gen_reg (&s, ®no)) break; else { out_insnp->opcode |= regno /* << 0 */; out_insnp->opcode |= regno << 12; continue; } case 'f': /* Flags from the condition code register. */ { int flags = 0; if (! get_flags (&s, &flags)) break; out_insnp->opcode |= ((flags & 0xf0) << 8) | (flags & 0xf); continue; } case 'i': /* A 6-bit signed immediate in bits <5:0>. */ if (! cris_get_expression (&s, &out_insnp->expr)) break; else { if (out_insnp->expr.X_op == O_constant && (out_insnp->expr.X_add_number < -32 || out_insnp->expr.X_add_number > 31)) as_bad (_("Immediate value not in 6 bit range: %ld"), out_insnp->expr.X_add_number); out_insnp->reloc = BFD_RELOC_CRIS_SIGNED_6; continue; } case 'I': /* A 6-bit unsigned immediate in bits <5:0>. */ if (! cris_get_expression (&s, &out_insnp->expr)) break; else { if (out_insnp->expr.X_op == O_constant && (out_insnp->expr.X_add_number < 0 || out_insnp->expr.X_add_number > 63)) as_bad (_("Immediate value not in 6 bit unsigned range: %ld"), out_insnp->expr.X_add_number); out_insnp->reloc = BFD_RELOC_CRIS_UNSIGNED_6; continue; } case 'M': /* A size modifier, B, W or D, to be put in a bit position suitable for CLEAR instructions (i.e. reflecting a zero register). */ if (! get_bwd_size_modifier (&s, &size_bits)) break; else { switch (size_bits) { case 0: out_insnp->opcode |= 0 << 12; break; case 1: out_insnp->opcode |= 4 << 12; break; case 2: out_insnp->opcode |= 8 << 12; break; } continue; } case 'm': /* A size modifier, B, W or D, to be put in bits <5:4>. */ if (! get_bwd_size_modifier (&s, &size_bits)) break; else { out_insnp->opcode |= size_bits << 4; continue; } case 'o': /* A branch expression. */ if (! cris_get_expression (&s, &out_insnp->expr)) break; else { out_insnp->insn_type = CRIS_INSN_BRANCH; continue; } case 'O': /* A BDAP expression for any size, "expr,r". */ if (! cris_get_expression (&s, &prefixp->expr)) break; else { if (*s != ',') break; s++; if (!get_gen_reg (&s, &prefixp->base_reg_number)) break; /* Since 'O' is used with an explicit bdap, we have no "real" instruction. */ prefixp->kind = PREFIX_BDAP_IMM; out_insnp->insn_type = CRIS_INSN_NONE; continue; } case 'P': /* Special register in bits <15:12>. */ if (! get_spec_reg (&s, &out_insnp->spec_reg)) break; else { /* Use of some special register names come with a specific warning. Note that we have no ".cpu type" pseudo yet, so some of this is just unused framework. */ if (out_insnp->spec_reg->warning) as_warn (out_insnp->spec_reg->warning); else if (out_insnp->spec_reg->applicable_version == cris_ver_warning) /* Others have a generic warning. */ as_warn (_("Unimplemented register `%s' specified"), out_insnp->spec_reg->name); out_insnp->opcode |= out_insnp->spec_reg->number << 12; continue; } case 'p': /* This character is used in the disassembler to recognize a prefix instruction to fold into the addressing mode for the next instruction. It is ignored here. */ continue; case 'R': /* General register in bits <15:12>. */ if (! get_gen_reg (&s, ®no)) break; else { out_insnp->opcode |= regno << 12; continue; } case 'r': /* General register in bits <3:0>. */ if (! get_gen_reg (&s, ®no)) break; else { out_insnp->opcode |= regno /* << 0 */; continue; } case 'S': /* Source operand in bit <10> and a prefix; a 3-operand prefix. */ if (! get_3op_or_dip_prefix_op (&s, prefixp)) break; else continue; case 's': /* Source operand in bits <10>, <3:0> and optionally a prefix; i.e. an indirect operand or an side-effect prefix. */ if (! get_autoinc_prefix_or_indir_op (&s, prefixp, &mode, ®no, &imm_expr_found, &out_insnp->expr)) break; else { if (prefixp->kind != PREFIX_NONE) { /* A prefix, so it has the autoincrement bit set. */ out_insnp->opcode |= (AUTOINCR_BIT << 8); } else /* No prefix. The "mode" variable contains bits like whether or not this is autoincrement mode. */ out_insnp->opcode |= (mode << 10); out_insnp->opcode |= regno /* << 0 */ ; continue; } case 'x': /* Rs.m in bits <15:12> and <5:4>. */ if (! get_gen_reg (&s, ®no) || ! get_bwd_size_modifier (&s, &size_bits)) break; else { out_insnp->opcode |= (regno << 12) | (size_bits << 4); continue; } case 'y': /* Source operand in bits <10>, <3:0> and optionally a prefix; i.e. an indirect operand or an side-effect prefix. The difference to 's' is that this does not allow an "immediate" expression. */ if (! get_autoinc_prefix_or_indir_op (&s, prefixp, &mode, ®no, &imm_expr_found, &out_insnp->expr) || imm_expr_found) break; else { if (prefixp->kind != PREFIX_NONE) { /* A prefix, and those matched here always have side-effects (see 's' case). */ out_insnp->opcode |= (AUTOINCR_BIT << 8); } else { /* No prefix. The "mode" variable contains bits like whether or not this is autoincrement mode. */ out_insnp->opcode |= (mode << 10); } out_insnp->opcode |= regno /* << 0 */; continue; } case 'z': /* Size modifier (B or W) in bit <4>. */ if (! get_bw_size_modifier (&s, &size_bits)) break; else { out_insnp->opcode |= size_bits << 4; continue; } default: BAD_CASE (*args); } /* We get here when we fail a match above or we found a complete match. Break out of this loop. */ break; } /* Was it a match or a miss? */ if (match == 0) { /* If it's just that the args don't match, maybe the next item in the table is the same opcode but with matching operands. */ if (instruction[1].name != NULL && ! strcmp (instruction->name, instruction[1].name)) { /* Yep. Restart and try that one instead. */ ++instruction; s = operands; continue; } else { /* We've come to the end of instructions with this opcode, so it must be an error. */ as_bad (_("Illegal operands")); return; } } else { /* We have a match. Check if there's anything more to do. */ if (imm_expr_found) { /* There was an immediate mode operand, so we must check that it has an appropriate size. */ switch (instruction->imm_oprnd_size) { default: case SIZE_NONE: /* Shouldn't happen; this one does not have immediate operands with different sizes. */ BAD_CASE (instruction->imm_oprnd_size); break; case SIZE_FIX_32: out_insnp->imm_oprnd_size = 4; break; case SIZE_SPEC_REG: switch (out_insnp->spec_reg->reg_size) { case 1: if (out_insnp->expr.X_op == O_constant && (out_insnp->expr.X_add_number < -128 || out_insnp->expr.X_add_number > 255)) as_bad (_("Immediate value not in 8 bit range: %ld"), out_insnp->expr.X_add_number); /* Fall through. */ case 2: /* FIXME: We need an indicator in the instruction table to pass on, to indicate if we need to check overflow for a signed or unsigned number. */ if (out_insnp->expr.X_op == O_constant && (out_insnp->expr.X_add_number < -32768 || out_insnp->expr.X_add_number > 65535)) as_bad (_("Immediate value not in 16 bit range: %ld"), out_insnp->expr.X_add_number); out_insnp->imm_oprnd_size = 2; break; case 4: out_insnp->imm_oprnd_size = 4; break; default: BAD_CASE (out_insnp->spec_reg->reg_size); } break; case SIZE_FIELD: switch (size_bits) { case 0: if (out_insnp->expr.X_op == O_constant && (out_insnp->expr.X_add_number < -128 || out_insnp->expr.X_add_number > 255)) as_bad (_("Immediate value not in 8 bit range: %ld"), out_insnp->expr.X_add_number); /* Fall through. */ case 1: if (out_insnp->expr.X_op == O_constant && (out_insnp->expr.X_add_number < -32768 || out_insnp->expr.X_add_number > 65535)) as_bad (_("Immediate value not in 16 bit range: %ld"), out_insnp->expr.X_add_number); out_insnp->imm_oprnd_size = 2; break; case 2: out_insnp->imm_oprnd_size = 4; break; default: BAD_CASE (out_insnp->spec_reg->reg_size); } } } } break; } } /* Get a B, W, or D size modifier from the string pointed out by *cPP, which must point to a '.' in front of the modifier. On successful return, *cPP is advanced to the character following the size modifier, and is undefined otherwise. cPP Pointer to pointer to string starting with the size modifier. size_bitsp Pointer to variable to contain the size bits on successful return. Return 1 iff a correct size modifier is found, else 0. */ static int get_bwd_size_modifier (cPP, size_bitsp) char **cPP; int *size_bitsp; { if (**cPP != '.') return 0; else { /* Consume the '.'. */ (*cPP)++; switch (**cPP) { case 'B': case 'b': *size_bitsp = 0; break; case 'W': case 'w': *size_bitsp = 1; break; case 'D': case 'd': *size_bitsp = 2; break; default: return 0; } /* Consume the size letter. */ (*cPP)++; return 1; } } /* Get a B or W size modifier from the string pointed out by *cPP, which must point to a '.' in front of the modifier. On successful return, *cPP is advanced to the character following the size modifier, and is undefined otherwise. cPP Pointer to pointer to string starting with the size modifier. size_bitsp Pointer to variable to contain the size bits on successful return. Return 1 iff a correct size modifier is found, else 0. */ static int get_bw_size_modifier (cPP, size_bitsp) char **cPP; int *size_bitsp; { if (**cPP != '.') return 0; else { /* Consume the '.'. */ (*cPP)++; switch (**cPP) { case 'B': case 'b': *size_bitsp = 0; break; case 'W': case 'w': *size_bitsp = 1; break; default: return 0; } /* Consume the size letter. */ (*cPP)++; return 1; } } /* Get a general register from the string pointed out by *cPP. The variable *cPP is advanced to the character following the general register name on a successful return, and has its initial position otherwise. cPP Pointer to pointer to string, beginning with a general register name. regnop Pointer to int containing the register number. Return 1 iff a correct general register designator is found, else 0. */ static int get_gen_reg (cPP, regnop) char **cPP; int *regnop; { char *oldp; oldp = *cPP; switch (**cPP) { case 'P': case 'p': /* "P" as in "PC"? Consume the "P". */ (*cPP)++; if ((**cPP == 'C' || **cPP == 'c') && ! isalnum ((*cPP)[1])) { /* It's "PC": consume the "c" and we're done. */ (*cPP)++; *regnop = REG_PC; return 1; } break; case 'R': case 'r': /* Hopefully r[0-9] or r1[0-5]. Consume 'R' or 'r'. */ (*cPP)++; if (isdigit (**cPP)) { /* It's r[0-9]. Consume and check the next digit. */ *regnop = **cPP - '0'; (*cPP)++; if (! isalnum (**cPP)) { /* No more digits, we're done. */ return 1; } else { /* One more digit. Consume and add. */ *regnop = *regnop * 10 + (**cPP - '0'); /* We need to check for a valid register number; Rn, 0 <= n <= MAX_REG. */ if (*regnop <= MAX_REG) { /* Consume second digit. */ (*cPP)++; return 1; } } } break; case 'S': case 's': /* "S" as in "SP"? Consume the "S". */ (*cPP)++; if (**cPP == 'P' || **cPP == 'p') { /* It's "SP": consume the "p" and we're done. */ (*cPP)++; *regnop = REG_SP; return 1; } break; default: /* Just here to silence compilation warnings. */ ; } /* We get here if we fail. Restore the pointer. */ *cPP = oldp; return 0; } /* Get a special register from the string pointed out by *cPP. The variable *cPP is advanced to the character following the special register name if one is found, and retains its original position otherwise. cPP Pointer to pointer to string starting with a special register name. sregpp Pointer to Pointer to struct spec_reg, where a pointer to the register description will be stored. Return 1 iff a correct special register name is found. */ static int get_spec_reg (cPP, sregpp) char **cPP; const struct cris_spec_reg **sregpp; { char *s1; const char *s2; const struct cris_spec_reg *sregp; /* Loop over all special registers. */ for (sregp = cris_spec_regs; sregp->name != NULL; sregp++) { /* Start over from beginning of the supposed name. */ s1 = *cPP; s2 = sregp->name; while (*s2 != '\0' && (isupper (*s1) ? tolower (*s1) == *s2 : *s1 == *s2)) { s1++; s2++; } /* For a match, we must have consumed the name in the table, and we must be outside what could be part of a name. Assume here that a test for alphanumerics is sufficient for a name test. */ if (*s2 == 0 && ! isalnum (*s1)) { /* We have a match. Update the pointer and be done. */ *cPP = s1; *sregpp = sregp; return 1; } } /* If we got here, we did not find any name. */ return 0; } /* Get an unprefixed or side-effect-prefix operand from the string pointed out by *cPP. The pointer *cPP is advanced to the character following the indirect operand if we have success, else it contains an undefined value. cPP Pointer to pointer to string beginning with the first character of the supposed operand. prefixp Pointer to structure containing an optional instruction prefix. is_autoincp Pointer to int indicating the indirect or autoincrement bits. src_regnop Pointer to int containing the source register number in the instruction. imm_foundp Pointer to an int indicating if an immediate expression is found. imm_exprP Pointer to a structure containing an immediate expression, if success and if *imm_foundp is nonzero. Return 1 iff a correct indirect operand is found. */ static int get_autoinc_prefix_or_indir_op (cPP, prefixp, is_autoincp, src_regnop, imm_foundp, imm_exprP) char **cPP; struct cris_prefix *prefixp; int *is_autoincp; int *src_regnop; int *imm_foundp; expressionS *imm_exprP; { /* Assume there was no immediate mode expression. */ *imm_foundp = 0; if (**cPP == '[') { /* So this operand is one of: Indirect: [rN] Autoincrement: [rN+] Indexed with assign: [rN=rM+rO.S] Offset with assign: [rN=rM+I], [rN=rM+[rO].s], [rN=rM+[rO+].s] Either way, consume the '['. */ (*cPP)++; /* Get the rN register. */ if (! get_gen_reg (cPP, src_regnop)) /* If there was no register, then this cannot match. */ return 0; else { /* We got the register, now check the next character. */ switch (**cPP) { case ']': /* Indirect mode. We're done here. */ prefixp->kind = PREFIX_NONE; *is_autoincp = 0; break; case '+': /* This must be an auto-increment mode, if there's a match. */ prefixp->kind = PREFIX_NONE; *is_autoincp = 1; /* We consume this character and break out to check the closing ']'. */ (*cPP)++; break; case '=': /* This must be indexed with assign, or offset with assign to match. */ (*cPP)++; /* Either way, the next thing must be a register. */ if (! get_gen_reg (cPP, &prefixp->base_reg_number)) /* No register, no match. */ return 0; else { /* We've consumed "[rN=rM", so we must be looking at "+rO.s]" or "+I]", or "-I]", or "+[rO].s]" or "+[rO+].s]". */ if (**cPP == '+') { int index_reg_number; (*cPP)++; if (**cPP == '[') { int size_bits; /* This must be [rx=ry+[rz].s] or [rx=ry+[rz+].s] or no match. We must be looking at rz after consuming the '['. */ (*cPP)++; if (!get_gen_reg (cPP, &index_reg_number)) return 0; prefixp->kind = PREFIX_BDAP; prefixp->opcode = (BDAP_INDIR_OPCODE + (prefixp->base_reg_number << 12) + index_reg_number); if (**cPP == '+') { /* We've seen "[rx=ry+[rz+" here, so now we know that there must be "].s]" left to check. */ (*cPP)++; prefixp->opcode |= AUTOINCR_BIT << 8; } /* If it wasn't autoincrement, we don't need to add anything. */ /* Check the next-to-last ']'. */ if (**cPP != ']') return 0; (*cPP)++; /* Check the ".s" modifier. */ if (! get_bwd_size_modifier (cPP, &size_bits)) return 0; prefixp->opcode |= size_bits << 4; /* Now we got [rx=ry+[rz+].s or [rx=ry+[rz].s. We break out to check the final ']'. */ break; } /* It wasn't an indirection. Check if it's a register. */ else if (get_gen_reg (cPP, &index_reg_number)) { int size_bits; /* Indexed with assign mode: "[rN+rM.S]". */ prefixp->kind = PREFIX_BIAP; prefixp->opcode = (BIAP_OPCODE + (index_reg_number << 12) + prefixp->base_reg_number /* << 0 */); if (! get_bwd_size_modifier (cPP, &size_bits)) /* Size missing, this isn't a match. */ return 0; else { /* Size found, break out to check the final ']'. */ prefixp->opcode |= size_bits << 4; break; } } /* Not a register. Then this must be "[rN+I]". */ else if (cris_get_expression (cPP, &prefixp->expr)) { /* We've got offset with assign mode. Fill in the blanks and break out to match the final ']'. */ prefixp->kind = PREFIX_BDAP_IMM; break; } else /* Neither register nor expression found, so this can't be a match. */ return 0; } /* Not "[rN+" but perhaps "[rN-"? */ else if (**cPP == '-') { /* We must have an offset with assign mode. */ if (! cris_get_expression (cPP, &prefixp->expr)) /* No expression, no match. */ return 0; else { /* We've got offset with assign mode. Fill in the blanks and break out to match the final ']'. */ prefixp->kind = PREFIX_BDAP_IMM; break; } } else /* Neither '+' nor '-' after "[rN=rM". Lose. */ return 0; } default: /* Neither ']' nor '+' nor '=' after "[rN". Lose. */ return 0; } } /* When we get here, we have a match and will just check the closing ']'. We can still fail though. */ if (**cPP != ']') return 0; else { /* Don't forget to consume the final ']'. Then return in glory. */ (*cPP)++; return 1; } } /* No indirection. Perhaps a constant? */ else if (cris_get_expression (cPP, imm_exprP)) { /* Expression found, this is immediate mode. */ prefixp->kind = PREFIX_NONE; *is_autoincp = 1; *src_regnop = REG_PC; *imm_foundp = 1; return 1; } /* No luck today. */ return 0; } /* This function gets an indirect operand in a three-address operand combination from the string pointed out by *cPP. The pointer *cPP is advanced to the character following the indirect operand on success, or has an unspecified value on failure. cPP Pointer to pointer to string begining with the operand prefixp Pointer to structure containing an instruction prefix Returns 1 iff a correct indirect operand is found. */ static int get_3op_or_dip_prefix_op (cPP, prefixp) char **cPP; struct cris_prefix *prefixp; { int reg_number; if (**cPP != '[') /* We must have a '[' or it's a clean failure. */ return 0; /* Eat the first '['. */ (*cPP)++; if (**cPP == '[') { /* A second '[', so this must be double-indirect mode. */ (*cPP)++; prefixp->kind = PREFIX_DIP; prefixp->opcode = DIP_OPCODE; /* Get the register or fail entirely. */ if (! get_gen_reg (cPP, ®_number)) return 0; else { prefixp->opcode |= reg_number /* << 0 */ ; if (**cPP == '+') { /* Since we found a '+', this must be double-indirect autoincrement mode. */ (*cPP)++; prefixp->opcode |= AUTOINCR_BIT << 8; } /* There's nothing particular to do, if this was a double-indirect *without* autoincrement. */ } /* Check the first ']'. The second one is checked at the end. */ if (**cPP != ']') return 0; /* Eat the first ']', so we'll be looking at a second ']'. */ (*cPP)++; } /* No second '['. Then we should have a register here, making it "[rN". */ else if (get_gen_reg (cPP, &prefixp->base_reg_number)) { /* This must be indexed or offset mode: "[rN+I]" or "[rN+rM.S]" or "[rN+[rM].S]" or "[rN+[rM+].S]". */ if (**cPP == '+') { /* Not the first alternative, must be one of the last three. */ int index_reg_number; (*cPP)++; if (**cPP == '[') { /* This is "[rx+["... Expect a register next. */ int size_bits; (*cPP)++; if (!get_gen_reg (cPP, &index_reg_number)) return 0; prefixp->kind = PREFIX_BDAP; prefixp->opcode = (BDAP_INDIR_OPCODE + (prefixp->base_reg_number << 12) + index_reg_number); /* We've seen "[rx+[ry", so check if this is autoincrement. */ if (**cPP == '+') { /* Yep, now at "[rx+[ry+". */ (*cPP)++; prefixp->opcode |= AUTOINCR_BIT << 8; } /* If it wasn't autoincrement, we don't need to add anything. */ /* Check a first closing ']': "[rx+[ry]" or "[rx+[ry+]". */ if (**cPP != ']') return 0; (*cPP)++; /* Now expect a size modifier ".S". */ if (! get_bwd_size_modifier (cPP, &size_bits)) return 0; prefixp->opcode |= size_bits << 4; /* Ok, all interesting stuff has been seen: "[rx+[ry+].S" or "[rx+[ry].S". We only need to expect a final ']', which we'll do in a common closing session. */ } /* Seen "[rN+", but not a '[', so check if we have a register. */ else if (get_gen_reg (cPP, &index_reg_number)) { /* This is indexed mode: "[rN+rM.S]" or "[rN+rM.S+]". */ int size_bits; prefixp->kind = PREFIX_BIAP; prefixp->opcode = (BIAP_OPCODE | prefixp->base_reg_number /* << 0 */ | (index_reg_number << 12)); /* Consume the ".S". */ if (! get_bwd_size_modifier (cPP, &size_bits)) /* Missing size, so fail. */ return 0; else /* Size found. Add that piece and drop down to the common checking of the closing ']'. */ prefixp->opcode |= size_bits << 4; } /* Seen "[rN+", but not a '[' or a register, so then it must be a constant "I". */ else if (cris_get_expression (cPP, &prefixp->expr)) { /* Expression found, so fill in the bits of offset mode and drop down to check the closing ']'. */ prefixp->kind = PREFIX_BDAP_IMM; } else /* Nothing valid here: lose. */ return 0; } /* Seen "[rN" but no '+', so check if it's a '-'. */ else if (**cPP == '-') { /* Yep, we must have offset mode. */ if (! cris_get_expression (cPP, &prefixp->expr)) /* No expression, so we lose. */ return 0; else { /* Expression found to make this offset mode, so fill those bits and drop down to check the closing ']'. */ prefixp->kind = PREFIX_BDAP_IMM; } } else { /* We've seen "[rN", but not '+' or '-'; rather a ']'. Hmm. Normally this is a simple indirect mode that we shouldn't match, but if we expect ']', then we have a zero offset, so it can be a three-address-operand, like "[rN],rO,rP", thus offset mode. Don't eat the ']', that will be done in the closing ceremony. */ prefixp->expr.X_op = O_constant; prefixp->expr.X_add_number = 0; prefixp->expr.X_add_symbol = NULL; prefixp->expr.X_op_symbol = NULL; prefixp->kind = PREFIX_BDAP_IMM; } } /* A '[', but no second '[', and no register. Check if we have an expression, making this "[I]" for a double-indirect prefix. */ else if (cris_get_expression (cPP, &prefixp->expr)) { /* Expression found, the so called absolute mode for a double-indirect prefix on PC. */ prefixp->kind = PREFIX_DIP; prefixp->opcode = DIP_OPCODE | (AUTOINCR_BIT << 8) | REG_PC; prefixp->reloc = BFD_RELOC_32; } else /* Neither '[' nor register nor expression. We lose. */ return 0; /* We get here as a closing ceremony to a successful match. We just need to check the closing ']'. */ if (**cPP != ']') /* Oops. Close but no air-polluter. */ return 0; /* Don't forget to consume that ']', before returning in glory. */ (*cPP)++; return 1; } /* Get an expression from the string pointed out by *cPP. The pointer *cPP is advanced to the character following the expression on a success, or retains its original value otherwise. cPP Pointer to pointer to string beginning with the expression. exprP Pointer to structure containing the expression. Return 1 iff a correct expression is found. */ static int cris_get_expression (cPP, exprP) char **cPP; expressionS *exprP; { char *saved_input_line_pointer; segT exp; /* The "expression" function expects to find an expression at the global variable input_line_pointer, so we have to save it to give the impression that we don't fiddle with global variables. */ saved_input_line_pointer = input_line_pointer; input_line_pointer = *cPP; exp = expression (exprP); if (exprP->X_op == O_illegal || exprP->X_op == O_absent) { input_line_pointer = saved_input_line_pointer; return 0; } /* Everything seems to be fine, just restore the global input_line_pointer and say we're successful. */ *cPP = input_line_pointer; input_line_pointer = saved_input_line_pointer; return 1; } /* Get a sequence of flag characters from *spp. The pointer *cPP is advanced to the character following the expression. The flag characters are consecutive, no commas or spaces. cPP Pointer to pointer to string beginning with the expression. flagp Pointer to int to return the flags expression. Return 1 iff a correct flags expression is found. */ static int get_flags (cPP, flagsp) char **cPP; int *flagsp; { for (;;) { switch (**cPP) { case 'd': case 'D': case 'm': case 'M': *flagsp |= 0x80; break; case 'e': case 'E': case 'b': case 'B': *flagsp |= 0x40; break; case 'i': case 'I': *flagsp |= 0x20; break; case 'x': case 'X': *flagsp |= 0x10; break; case 'n': case 'N': *flagsp |= 0x8; break; case 'z': case 'Z': *flagsp |= 0x4; break; case 'v': case 'V': *flagsp |= 0x2; break; case 'c': case 'C': *flagsp |= 1; break; default: /* We consider this successful if we stop at a comma or whitespace. Anything else, and we consider it a failure. */ if (**cPP != ',' && **cPP != 0 && ! isspace (**cPP)) return 0; else return 1; } /* Don't forget to consume each flag character. */ (*cPP)++; } } /* Generate code and fixes for a BDAP prefix. base_regno Int containing the base register number. exprP Pointer to structure containing the offset expression. */ static void gen_bdap (base_regno, exprP) int base_regno; expressionS *exprP; { unsigned int opcode; char *opcodep; /* Put out the prefix opcode; assume quick immediate mode at first. */ opcode = BDAP_QUICK_OPCODE | (base_regno << 12); opcodep = frag_more (2); md_number_to_chars (opcodep, opcode, 2); if (exprP->X_op == O_constant) { /* We have an absolute expression that we know the size of right now. */ long int value; int size; value = exprP->X_add_number; if (value < -32768 || value > 32767) /* Outside range for a "word", make it a dword. */ size = 2; else /* Assume "word" size. */ size = 1; /* If this is a signed-byte value, we can fit it into the prefix insn itself. */ if (value >= -128 && value <= 127) opcodep[0] = value; else { /* This is a word or dword displacement, which will be put in a word or dword after the prefix. */ char *p; opcodep[0] = BDAP_PC_LOW + (size << 4); opcodep[1] &= 0xF0; opcodep[1] |= BDAP_INCR_HIGH; p = frag_more (1 << size); md_number_to_chars (p, value, 1 << size); } } else /* The expression is not defined yet but may become absolute. We make it a relocation to be relaxed. */ frag_var (rs_machine_dependent, 4, 0, ENCODE_RELAX (STATE_BASE_PLUS_DISP_PREFIX, STATE_UNDF), exprP->X_add_symbol, exprP->X_add_number, opcodep); } /* Encode a branch displacement in the range -256..254 into the form used by CRIS conditional branch instructions. offset The displacement value in bytes. */ static int branch_disp (offset) int offset; { int disp; disp = offset & 0xFE; if (offset < 0) disp |= 1; return disp; } /* Generate code and fixes for a 32-bit conditional branch instruction created by "extending" an existing 8-bit branch instruction. opcodep Pointer to the word containing the original 8-bit branch instruction. writep Pointer to "extension area" following the first instruction word. fragP Pointer to the frag containing the instruction. add_symP, Parts of the destination address expression. sub_symP, add_num. */ static void gen_cond_branch_32 (opcodep, writep, fragP, add_symP, sub_symP, add_num) char *opcodep; char *writep; fragS *fragP; symbolS *add_symP; symbolS *sub_symP; long int add_num; { if (warn_for_branch_expansion) { /* FIXME: Find out and change to as_warn_where. Add testcase. */ as_warn (_("32-bit conditional branch generated")); } /* Here, writep points to what will be opcodep + 2. First, we change the actual branch in opcodep[0] and opcodep[1], so that in the final insn, it will look like: opcodep+10: Bcc .-6 This means we don't have to worry about changing the opcode or messing with te delay-slot instruction. So, we move it to last in the "extended" branch, and just change the displacement. Admittedly, it's not the optimal extended construct, but we should get this rarely enough that it shouldn't matter. */ writep[8] = branch_disp (-2 - 6); writep[9] = opcodep[1]; /* Then, we change the branch to an unconditional branch over the extended part, to the new location of the Bcc: opcodep: BA .+10 opcodep+2: NOP Note that these two writes are to currently different locations, merged later. */ md_number_to_chars (opcodep, BA_QUICK_OPCODE + 8, 2); md_number_to_chars (writep, NOP_OPCODE, 2); /* Then the extended thing, the 32-bit jump insn. opcodep+4: JUMP [PC+] */ md_number_to_chars (writep + 2, JUMP_PC_INCR_OPCODE, 2); /* We have to fill in the actual value too. opcodep+6: .DWORD This is most probably an expression, but we can cope with an absolute value too. FIXME: Testcase needed. */ if (add_symP == NULL && sub_symP == NULL) /* An absolute address. */ md_number_to_chars (writep + 4, add_num, 4); else { /* Not absolute, we have to make it a frag for later evaluation. */ know (sub_symP == 0); fix_new (fragP, writep + 4 - fragP->fr_literal, 4, add_symP, add_num, 0, BFD_RELOC_32); } } /* This *could* be: Turn a string in input_line_pointer into a floating point constant of type TYPE, and store the appropriate bytes in *LITP. The number of LITTLENUMS emitted is stored in *SIZEP. type A character from FLTCHARS that describes what kind of floating-point number is wanted. litp A pointer to an array that the result should be stored in. sizep A pointer to an integer where the size of the result is stored. But we don't support floating point constants in assembly code *at all*, since it's suboptimal and just opens up bug opportunities. GCC emits the bit patterns as hex. All we could do here is to emit what GCC would have done in the first place. *Nobody* writes floating-point code as assembly code, but if they do, they should be able enough to find out the correct bit patterns and use them. */ char * md_atof (type, litp, sizep) char type ATTRIBUTE_UNUSED; char *litp ATTRIBUTE_UNUSED; int *sizep ATTRIBUTE_UNUSED; { /* FIXME: Is this function mentioned in the internals.texi manual? If not, add it. */ return _("Bad call to md_atof () - floating point formats are not supported"); } /* Turn a number as a fixS * into a series of bytes that represents the number on the target machine. The purpose of this procedure is the same as that of md_number_to_chars but this procedure is supposed to handle general bit field fixes and machine-dependent fixups. bufp Pointer to an array where the result should be stored. val The value to store. n The number of bytes in "val" that should be stored. fixP The fix to be applied to the bit field starting at bufp. */ static void cris_number_to_imm (bufp, val, n, fixP) char *bufp; long val; int n; fixS *fixP; { segT sym_seg; know (n <= 4); know (fixP); /* We put the relative "vma" for the other segment for inter-segment relocations in the object data to stay binary "compatible" (with an uninteresting old version) for the relocation. Maybe delete some day. */ if (fixP->fx_addsy && (sym_seg = S_GET_SEGMENT (fixP->fx_addsy)) != now_seg) val += sym_seg->vma; switch (fixP->fx_r_type) { /* Ditto here, we put the addend into the object code as well as the reloc addend. Keep it that way for now, to simplify regression tests on the object file contents. FIXME: Seems uninteresting now that we have a test suite. */ case BFD_RELOC_32: /* No use having warnings here, since most hosts have a 32-bit type for "long" (which will probably change soon, now that I wrote this). */ bufp[3] = (val >> 24) & 0xFF; bufp[2] = (val >> 16) & 0xFF; bufp[1] = (val >> 8) & 0xFF; bufp[0] = val & 0xFF; break; /* FIXME: The 16 and 8-bit cases should have a way to check whether a signed or unsigned (or any signedness) number is accepted. FIXME: Does the as_bad calls find the line number by themselves, or should we change them into as_bad_where? */ case BFD_RELOC_16: if (val > 0xffff || val < -32768) as_bad (_("Value not in 16 bit range: %ld"), val); if (! fixP->fx_addsy) { bufp[1] = (val >> 8) & 0xFF; bufp[0] = val & 0xFF; } break; case BFD_RELOC_8: if (val > 255 || val < -128) as_bad (_("Value not in 8 bit range: %ld"), val); if (! fixP->fx_addsy) bufp[0] = val & 0xFF; break; case BFD_RELOC_CRIS_UNSIGNED_4: if (val > 15 || val < 0) as_bad (_("Value not in 4 bit unsigned range: %ld"), val); if (! fixP->fx_addsy) bufp[0] |= val & 0x0F; break; case BFD_RELOC_CRIS_UNSIGNED_5: if (val > 31 || val < 0) as_bad (_("Value not in 5 bit unsigned range: %ld"), val); if (! fixP->fx_addsy) bufp[0] |= val & 0x1F; break; case BFD_RELOC_CRIS_SIGNED_6: if (val > 31 || val < -32) as_bad (_("Value not in 6 bit range: %ld"), val); if (! fixP->fx_addsy) bufp[0] |= val & 0x3F; break; case BFD_RELOC_CRIS_UNSIGNED_6: if (val > 63 || val < 0) as_bad (_("Value not in 6 bit unsigned range: %ld"), val); if (! fixP->fx_addsy) bufp[0] |= val & 0x3F; break; case BFD_RELOC_CRIS_BDISP8: if (! fixP->fx_addsy) bufp[0] = branch_disp (val); break; case BFD_RELOC_NONE: /* May actually happen automatically. For example at broken words, if the word turns out not to be broken. FIXME: When? Which testcase? */ if (! fixP->fx_addsy) md_number_to_chars (bufp, val, n); break; case BFD_RELOC_VTABLE_INHERIT: /* This borrowed from tc-ppc.c on a whim. */ if (fixP->fx_addsy && !S_IS_DEFINED (fixP->fx_addsy) && !S_IS_WEAK (fixP->fx_addsy)) S_SET_WEAK (fixP->fx_addsy); /* FALLTHROUGH. */ case BFD_RELOC_VTABLE_ENTRY: /* FIXME: I'm not sure why we do this (as does other ports), but it might be that this relocation can only be finished at link time. */ fixP->fx_done = 0; break; default: BAD_CASE (fixP->fx_r_type); } } /* Processes machine-dependent command line options. Called once for each option on the command line that the machine-independent part of GAS does not understand. */ int md_parse_option (arg, argp) int arg; char *argp ATTRIBUTE_UNUSED; { switch (arg) { case 'H': case 'h': md_show_usage (stdout); /* Don't continue. */ exit (0); case 'N': warn_for_branch_expansion = 1; return 1; default: return 0; } } /* Round up a section size to the appropriate boundary. */ valueT md_section_align (segment, size) segT segment; valueT size; { /* Round all sects to multiple of 4, except the bss section, which we'll round to word-size. FIXME: Check if this really matters. All sections should be rounded up, and all sections should (optionally) be assumed to be dword-aligned, it's just that there is actual usage of linking to a multiple of two. */ if (OUTPUT_FLAVOR == bfd_target_aout_flavour) { if (segment == bss_section) return (size + 1) & ~1; return (size + 3) & ~3; } else { /* FIXME: Is this wanted? It matches the testsuite, but that's not really a valid reason. */ if (segment == text_section) return (size + 3) & ~3; } return size; } /* Generate a machine-dependent relocation. */ arelent * tc_gen_reloc (section, fixP) asection *section ATTRIBUTE_UNUSED; fixS *fixP; { arelent *relP; bfd_reloc_code_real_type code; switch (fixP->fx_r_type) { case BFD_RELOC_32: case BFD_RELOC_16: case BFD_RELOC_8: case BFD_RELOC_VTABLE_INHERIT: case BFD_RELOC_VTABLE_ENTRY: code = fixP->fx_r_type; break; default: as_bad_where (fixP->fx_file, fixP->fx_line, _("Semantics error. This type of operand can not be relocated, it must be an assembly-time constant")); return 0; } relP = (arelent *) xmalloc (sizeof (arelent)); assert (relP != 0); relP->sym_ptr_ptr = (asymbol **) xmalloc (sizeof (asymbol *)); *relP->sym_ptr_ptr = symbol_get_bfdsym (fixP->fx_addsy); relP->address = fixP->fx_frag->fr_address + fixP->fx_where; if (fixP->fx_pcrel) /* FIXME: Is this correct? */ relP->addend = fixP->fx_addnumber; else /* At least *this one* is correct. */ relP->addend = fixP->fx_offset; /* This is the standard place for KLUDGEs to work around bugs in bfd_install_relocation (first such note in the documentation appears with binutils-2.8). That function bfd_install_relocation does the wrong thing with putting stuff into the addend of a reloc (it should stay out) for a weak symbol. The really bad thing is that it adds the "segment-relative offset" of the symbol into the reloc. In this case, the reloc should instead be relative to the symbol with no other offset than the assembly code shows; and since the symbol is weak, any local definition should be ignored until link time (or thereafter). To wit: weaksym+42 should be weaksym+42 in the reloc, not weaksym+(offset_from_segment_of_local_weaksym_definition) To "work around" this, we subtract the segment-relative offset of "known" weak symbols. This evens out the extra offset. That happens for a.out but not for ELF, since for ELF, bfd_install_relocation uses the "special function" field of the howto, and does not execute the code that needs to be undone. */ if (OUTPUT_FLAVOR == bfd_target_aout_flavour && fixP->fx_addsy && S_IS_WEAK (fixP->fx_addsy) && ! bfd_is_und_section (S_GET_SEGMENT (fixP->fx_addsy))) { relP->addend -= S_GET_VALUE (fixP->fx_addsy); } relP->howto = bfd_reloc_type_lookup (stdoutput, code); if (! relP->howto) { const char *name; name = S_GET_NAME (fixP->fx_addsy); if (name == NULL) name = _(""); as_fatal (_("Cannot generate relocation type for symbol %s, code %s"), name, bfd_get_reloc_code_name (code)); } return relP; } /* Machine-dependent usage-output. */ void md_show_usage (stream) FILE *stream; { fprintf (stream, _("\n-- GNU as for CRIS\n")); fprintf (stream, "\n"); fprintf (stream, _("*** Usage: as-cris [switches] [-o objectfile] [files...]\n")); fprintf (stream, _("Target-specific switches:\n")); fprintf (stream, _(" -h, -H : Don't execute, print this help text.\n")); fprintf (stream, _(" -N : Warn when branches are expanded to jumps.\n\n")); fprintf (stream, _("Use as-cris --help to see more options.\n")); fprintf (stream, _("objectfile : Output file in the a.out format %s"), _("described in the users manual.\n")); fprintf (stream, _(" Default output file is \"a.out\".\n")); fprintf (stream, _("files ... : Input files in the source format %s"), _(" described in the users manual.\n")); fprintf (stream, _(" Default input file is standard input.\n")); fprintf (stream, _("Description : Assembler for the CRIS processor.\n")); } /* Apply a fixS (fixup of an instruction or data that we didn't have enough info to complete immediately) to the data in a frag. */ int md_apply_fix (fixP, valP) fixS *fixP; valueT *valP; { long val = *valP; char *buf = fixP->fx_where + fixP->fx_frag->fr_literal; if (fixP->fx_addsy == 0 && !fixP->fx_pcrel) fixP->fx_done = 1; if (fixP->fx_bit_fixP || fixP->fx_im_disp != 0) { as_bad_where (fixP->fx_file, fixP->fx_line, _("Invalid relocation")); fixP->fx_done = 1; } else { /* I took this from tc-arc.c, since we used to not support fx_subsy != NULL. I'm not totally sure it's TRT. */ if (fixP->fx_subsy != (symbolS *) NULL) { if (S_GET_SEGMENT (fixP->fx_subsy) == absolute_section) val -= S_GET_VALUE (fixP->fx_subsy); else { /* We can't actually support subtracting a symbol. */ as_bad_where (fixP->fx_file, fixP->fx_line, _("expression too complex")); } } cris_number_to_imm (buf, val, fixP->fx_size, fixP); } return 1; } /* All relocations are relative to the location just after the fixup; the address of the fixup plus its size. */ long md_pcrel_from (fixP) fixS *fixP; { valueT addr = fixP->fx_where + fixP->fx_frag->fr_address; /* FIXME: We get here only at the end of assembly, when X in ".-X" is still unknown. Since we don't have pc-relative relocations, this is invalid. What to do if anything for a.out, is to add pc-relative relocations everywhere including the elinux program loader. */ as_bad_where (fixP->fx_file, fixP->fx_line, _("Invalid pc-relative relocation")); return fixP->fx_size + addr; } /* We have no need to give defaults for symbol-values. */ symbolS * md_undefined_symbol (name) char *name ATTRIBUTE_UNUSED; { return 0; } /* Definition of TC_FORCE_RELOCATION. FIXME: Unsure of this. Can we omit it? Just copied from tc-i386.c when doing multi-object format with ELF, since it's the only other multi-object-format target with a.out and ELF. */ int md_cris_force_relocation (fixp) struct fix *fixp; { if (fixp->fx_r_type == BFD_RELOC_VTABLE_INHERIT || fixp->fx_r_type == BFD_RELOC_VTABLE_ENTRY) return 1; return 0; } /* Check and emit error if broken-word handling has failed to fix up a case-table. This is called from write.c, after doing everything it knows about how to handle broken words. */ void tc_cris_check_adjusted_broken_word (new_offset, brokwP) offsetT new_offset; struct broken_word *brokwP; { if (new_offset > 32767 || new_offset < -32768) /* We really want a genuine error, not a warning, so make it one. */ as_bad_where (brokwP->frag->fr_file, brokwP->frag->fr_line, _("Adjusted signed .word (%ld) overflows: `switch'-statement too large."), (long) new_offset); } /* * Local variables: * eval: (c-set-style "gnu") * indent-tabs-mode: t * End: */