X-Git-Url: http://drtracing.org/?a=blobdiff_plain;f=gas%2Fapp.c;h=9666ce58aca850a01b948b266d941b0866e1a7c1;hb=61cc02671150a81ea68f25b8409b8ace18bda9ae;hp=0494aec23b6816fdc6280862fcd64e765cdd18ea;hpb=ed288bb597072176e84fc8279707a3f2f475779b;p=deliverable%2Fbinutils-gdb.git diff --git a/gas/app.c b/gas/app.c index 0494aec23b..9666ce58ac 100644 --- a/gas/app.c +++ b/gas/app.c @@ -1,5 +1,6 @@ /* This is the Assembler Pre-Processor - Copyright (C) 1987, 90, 91, 92, 93, 94, 95, 96, 97, 98, 1999 + Copyright 1987, 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, + 1999, 2000, 2001, 2002, 2003 Free Software Foundation, Inc. This file is part of GAS, the GNU Assembler. @@ -16,17 +17,17 @@ 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. */ + Software Foundation, 51 Franklin Street - Fifth Floor, Boston, MA + 02110-1301, USA. */ -/* Modified by Allen Wirfs-Brock, Instantiations Inc 2/90 */ +/* Modified by Allen Wirfs-Brock, Instantiations Inc 2/90. */ /* App, the assembler pre-processor. This pre-processor strips out excess spaces, turns single-quoted characters into a decimal constant, and turns # into a .line \n.file pair. This needs better error-handling. */ #include -#include "as.h" /* For BAD_CASE() only */ +#include "as.h" /* For BAD_CASE() only. */ #if (__STDC__ != 1) #ifndef const @@ -34,6 +35,7 @@ #endif #endif +#ifdef TC_M68K /* Whether we are scrubbing in m68k MRI mode. This is different from flag_m68k_mri, because the two flags will be affected by the .mri pseudo-op at different times. */ @@ -42,9 +44,12 @@ static int scrub_m68k_mri; /* The pseudo-op which switches in and out of MRI mode. See the comment in do_scrub_chars. */ static const char mri_pseudo[] = ".mri 0"; +#else +#define scrub_m68k_mri 0 +#endif #if defined TC_ARM && defined OBJ_ELF -/* The pseudo-op for which we need to special-case `@' characters. +/* The pseudo-op for which we need to special-case `@' characters. See the comment in do_scrub_chars. */ static const char symver_pseudo[] = ".symver"; static const char * symver_state; @@ -68,42 +73,48 @@ static const char symbol_chars[] = #define LEX_IS_DOUBLEDASH_1ST 12 #endif #ifdef TC_M32R +#define DOUBLEBAR_PARALLEL +#endif +#ifdef DOUBLEBAR_PARALLEL #define LEX_IS_DOUBLEBAR_1ST 13 #endif +#define LEX_IS_PARALLEL_SEPARATOR 14 #define IS_SYMBOL_COMPONENT(c) (lex[c] == LEX_IS_SYMBOL_COMPONENT) #define IS_WHITESPACE(c) (lex[c] == LEX_IS_WHITESPACE) #define IS_LINE_SEPARATOR(c) (lex[c] == LEX_IS_LINE_SEPARATOR) +#define IS_PARALLEL_SEPARATOR(c) (lex[c] == LEX_IS_PARALLEL_SEPARATOR) #define IS_COMMENT(c) (lex[c] == LEX_IS_COMMENT_START) #define IS_LINE_COMMENT(c) (lex[c] == LEX_IS_LINE_COMMENT_START) #define IS_NEWLINE(c) (lex[c] == LEX_IS_NEWLINE) -static int process_escape PARAMS ((int)); +static int process_escape (int); /* FIXME-soon: The entire lexer/parser thingy should be built statically at compile time rather than dynamically - each and every time the assembler is run. xoxorich. */ + each and every time the assembler is run. xoxorich. */ -void -do_scrub_begin (m68k_mri) - int m68k_mri; +void +do_scrub_begin (int m68k_mri ATTRIBUTE_UNUSED) { const char *p; int c; - scrub_m68k_mri = m68k_mri; - lex[' '] = LEX_IS_WHITESPACE; lex['\t'] = LEX_IS_WHITESPACE; lex['\r'] = LEX_IS_WHITESPACE; lex['\n'] = LEX_IS_NEWLINE; - lex[';'] = LEX_IS_LINE_SEPARATOR; lex[':'] = LEX_IS_COLON; +#ifdef TC_M68K + scrub_m68k_mri = m68k_mri; + if (! m68k_mri) +#endif { lex['"'] = LEX_IS_STRINGQUOTE; -#ifndef TC_HPPA +#if ! defined (TC_HPPA) && ! defined (TC_I370) + /* I370 uses single-quotes to delimit integer, float constants. */ lex['\''] = LEX_IS_ONECHAR_QUOTE; #endif @@ -118,9 +129,7 @@ do_scrub_begin (m68k_mri) /* Note that these override the previous defaults, e.g. if ';' is a comment char, then it isn't a line separator. */ for (p = symbol_chars; *p; ++p) - { - lex[(unsigned char) *p] = LEX_IS_SYMBOL_COMPONENT; - } /* declare symbol characters */ + lex[(unsigned char) *p] = LEX_IS_SYMBOL_COMPONENT; for (c = 128; c < 256; ++c) lex[c] = LEX_IS_SYMBOL_COMPONENT; @@ -140,50 +149,51 @@ do_scrub_begin (m68k_mri) #define tc_comment_chars comment_chars #endif for (p = tc_comment_chars; *p; p++) - { - lex[(unsigned char) *p] = LEX_IS_COMMENT_START; - } /* declare comment chars */ + lex[(unsigned char) *p] = LEX_IS_COMMENT_START; for (p = line_comment_chars; *p; p++) - { - lex[(unsigned char) *p] = LEX_IS_LINE_COMMENT_START; - } /* declare line comment chars */ + lex[(unsigned char) *p] = LEX_IS_LINE_COMMENT_START; for (p = line_separator_chars; *p; p++) - { - lex[(unsigned char) *p] = LEX_IS_LINE_SEPARATOR; - } /* declare line separators */ + lex[(unsigned char) *p] = LEX_IS_LINE_SEPARATOR; + +#ifdef tc_parallel_separator_chars + /* This macro permits the processor to specify all characters which + separate parallel insns on the same line. */ + for (p = tc_parallel_separator_chars; *p; p++) + lex[(unsigned char) *p] = LEX_IS_PARALLEL_SEPARATOR; +#endif /* Only allow slash-star comments if slash is not in use. FIXME: This isn't right. We should always permit them. */ if (lex['/'] == 0) - { - lex['/'] = LEX_IS_TWOCHAR_COMMENT_1ST; - } + lex['/'] = LEX_IS_TWOCHAR_COMMENT_1ST; +#ifdef TC_M68K if (m68k_mri) { lex['\''] = LEX_IS_STRINGQUOTE; lex[';'] = LEX_IS_COMMENT_START; lex['*'] = LEX_IS_LINE_COMMENT_START; /* The MRI documentation says '!' is LEX_IS_COMMENT_START, but - then it can't be used in an expression. */ + then it can't be used in an expression. */ lex['!'] = LEX_IS_LINE_COMMENT_START; } +#endif #ifdef TC_V850 lex['-'] = LEX_IS_DOUBLEDASH_1ST; #endif -#ifdef TC_M32R +#ifdef DOUBLEBAR_PARALLEL lex['|'] = LEX_IS_DOUBLEBAR_1ST; #endif #ifdef TC_D30V - /* must do this is we want VLIW instruction with "->" or "<-" */ + /* Must do this is we want VLIW instruction with "->" or "<-". */ lex['-'] = LEX_IS_SYMBOL_COMPONENT; #endif -} /* do_scrub_begin() */ +} -/* Saved state of the scrubber */ +/* Saved state of the scrubber. */ static int state; static int old_state; static char *out_string; @@ -201,24 +211,26 @@ static char mri_last_ch; That's why we have to save it all. */ struct app_save - { - int state; - int old_state; - char * out_string; - char out_buf[sizeof (out_buf)]; - int add_newlines; - char * saved_input; - int saved_input_len; - int scrub_m68k_mri; - const char * mri_state; - char mri_last_ch; +{ + int state; + int old_state; + char * out_string; + char out_buf[sizeof (out_buf)]; + int add_newlines; + char * saved_input; + int saved_input_len; +#ifdef TC_M68K + int scrub_m68k_mri; +#endif + const char * mri_state; + char mri_last_ch; #if defined TC_ARM && defined OBJ_ELF - const char * symver_state; + const char * symver_state; #endif - }; +}; char * -app_push () +app_push (void) { register struct app_save *saved; @@ -236,14 +248,16 @@ app_push () memcpy (saved->saved_input, saved_input, saved_input_len); saved->saved_input_len = saved_input_len; } +#ifdef TC_M68K saved->scrub_m68k_mri = scrub_m68k_mri; +#endif saved->mri_state = mri_state; saved->mri_last_ch = mri_last_ch; #if defined TC_ARM && defined OBJ_ELF saved->symver_state = symver_state; #endif - /* do_scrub_begin() is not useful, just wastes time. */ + /* do_scrub_begin() is not useful, just wastes time. */ state = 0; saved_input = NULL; @@ -251,13 +265,12 @@ app_push () return (char *) saved; } -void -app_pop (arg) - char *arg; +void +app_pop (char *arg) { register struct app_save *saved = (struct app_save *) arg; - /* There is no do_scrub_end (). */ + /* There is no do_scrub_end (). */ state = saved->state; old_state = saved->old_state; out_string = saved->out_string; @@ -273,7 +286,9 @@ app_pop (arg) saved_input_len = saved->saved_input_len; free (saved->saved_input); } +#ifdef TC_M68K scrub_m68k_mri = saved->scrub_m68k_mri; +#endif mri_state = saved->mri_state; mri_last_ch = saved->mri_last_ch; #if defined TC_ARM && defined OBJ_ELF @@ -281,13 +296,13 @@ app_pop (arg) #endif free (arg); -} /* app_pop() */ +} /* @@ This assumes that \n &c are the same on host and target. This is not necessarily true. */ -static int -process_escape (ch) - int ch; + +static int +process_escape (int ch) { switch (ch) { @@ -322,10 +337,7 @@ process_escape (ch) This is the way the old code used to work. */ int -do_scrub_chars (get, tostart, tolen) - int (*get) PARAMS ((char *, int)); - char *tostart; - int tolen; +do_scrub_chars (int (*get) (char *, int), char *tostart, int tolen) { char *to = tostart; char *toend = tostart + tolen; @@ -349,17 +361,25 @@ do_scrub_chars (get, tostart, tolen) -1: output string in out_string and go to the state in old_state -2: flush text until a '*' '/' is seen, then go to state old_state #ifdef TC_V850 - 12: After seeing a dash, looking for a second dash as a start of comment. + 12: After seeing a dash, looking for a second dash as a start + of comment. #endif -#ifdef TC_M32R - 13: After seeing a vertical bar, looking for a second vertical bar as a parallel expression seperator. +#ifdef DOUBLEBAR_PARALLEL + 13: After seeing a vertical bar, looking for a second + vertical bar as a parallel expression separator. +#endif +#ifdef TC_IA64 + 14: After seeing a `(' at state 0, looking for a `)' as + predicate. + 15: After seeing a `(' at state 1, looking for a `)' as + predicate. #endif */ /* I added states 9 and 10 because the MIPS ECOFF assembler uses constructs like ``.loc 1 20''. This was turning into ``.loc 120''. States 9 and 10 ensure that a space is never dropped in - between characters which could appear in a identifier. Ian + between characters which could appear in an identifier. Ian Taylor, ian@cygnus.com. I added state 11 so that something like "Lfoo add %r25,%r26,%r27" works @@ -396,13 +416,13 @@ do_scrub_chars (get, tostart, tolen) I don't want to make such a significant change to the assembler's memory usage. */ -#define PUT(pch) \ - do \ - { \ - *to++ = (pch); \ - if (to >= toend) \ - goto tofull; \ - } \ +#define PUT(pch) \ + do \ + { \ + *to++ = (pch); \ + if (to >= toend) \ + goto tofull; \ + } \ while (0) if (saved_input != NULL) @@ -422,8 +442,8 @@ do_scrub_chars (get, tostart, tolen) while (1) { /* The cases in this switch end with continue, in order to - branch back to the top of this while loop and generate the - next output character in the appropriate state. */ + branch back to the top of this while loop and generate the + next output character in the appropriate state. */ switch (state) { case -1: @@ -506,9 +526,9 @@ do_scrub_chars (get, tostart, tolen) case 5: /* We are going to copy everything up to a quote character, - with special handling for a backslash. We try to - optimize the copying in the simple case without using the - GET and PUT macros. */ + with special handling for a backslash. We try to + optimize the copying in the simple case without using the + GET and PUT macros. */ { char *s; int len; @@ -517,7 +537,7 @@ do_scrub_chars (get, tostart, tolen) { ch = *s; /* This condition must be changed if the type of any - other character can be LEX_IS_STRINGQUOTE. */ + other character can be LEX_IS_STRINGQUOTE. */ if (ch == '\\' || ch == '"' || ch == '\'' @@ -538,7 +558,7 @@ do_scrub_chars (get, tostart, tolen) ch = GET (); if (ch == EOF) { - as_warn (_("end of file in string: inserted '\"'")); + as_warn (_("end of file in string; inserted '\"'")); state = old_state; UNGET ('\n'); PUT ('"'); @@ -558,8 +578,7 @@ do_scrub_chars (get, tostart, tolen) else if (scrub_m68k_mri && ch == '\n') { /* Just quietly terminate the string. This permits lines like - bne label loop if we haven't reach end yet - */ + bne label loop if we haven't reach end yet. */ state = old_state; UNGET (ch); PUT ('\''); @@ -583,6 +602,11 @@ do_scrub_chars (get, tostart, tolen) PUT ('\\'); continue; + case EOF: + as_warn (_("end of file in string; '\"' inserted")); + PUT ('"'); + continue; + case '"': case '\\': case 'b': @@ -602,20 +626,12 @@ do_scrub_chars (get, tostart, tolen) case '6': case '7': break; -#if defined(IGNORE_NONSTANDARD_ESCAPES) | defined(ONLY_STANDARD_ESCAPES) - default: - as_warn (_("Unknown escape '\\%c' in string: Ignored"), ch); - break; -#else /* ONLY_STANDARD_ESCAPES */ + default: - /* Accept \x as x for any x */ +#ifdef ONLY_STANDARD_ESCAPES + as_warn (_("unknown escape '\\%c' in string; ignored"), ch); +#endif break; -#endif /* ONLY_STANDARD_ESCAPES */ - - case EOF: - as_warn (_("End of file in string: '\"' inserted")); - PUT ('"'); - continue; } PUT (ch); continue; @@ -638,13 +654,49 @@ do_scrub_chars (get, tostart, tolen) state = 0; PUT (ch); continue; + +#ifdef DOUBLEBAR_PARALLEL + case 13: + ch = GET (); + if (ch != '|') + abort (); + + /* Reset back to state 1 and pretend that we are parsing a + line from just after the first white space. */ + state = 1; + PUT ('|'); + continue; +#endif } - /* OK, we are somewhere in states 0 through 4 or 9 through 11 */ + /* OK, we are somewhere in states 0 through 4 or 9 through 11. */ /* flushchar: */ ch = GET (); +#ifdef TC_IA64 + if (ch == '(' && (state == 0 || state == 1)) + { + state += 14; + PUT (ch); + continue; + } + else if (state == 14 || state == 15) + { + if (ch == ')') + { + state -= 14; + PUT (ch); + ch = GET (); + } + else + { + PUT (ch); + continue; + } + } +#endif + recycle: #if defined TC_ARM && defined OBJ_ELF @@ -678,9 +730,9 @@ do_scrub_chars (get, tostart, tolen) #ifdef TC_M68K /* We want to have pseudo-ops which control whether we are in - MRI mode or not. Unfortunately, since m68k MRI mode affects - the scrubber, that means that we need a special purpose - recognizer here. */ + MRI mode or not. Unfortunately, since m68k MRI mode affects + the scrubber, that means that we need a special purpose + recognizer here. */ if (mri_state == NULL) { if ((state == 0 || state == 1) @@ -717,14 +769,14 @@ do_scrub_chars (get, tostart, tolen) else { /* We've read the entire pseudo-op. mips_last_ch is - either '0' or '1' indicating whether to enter or - leave MRI mode. */ + either '0' or '1' indicating whether to enter or + leave MRI mode. */ do_scrub_begin (mri_last_ch == '1'); mri_state = NULL; /* We continue handling the character as usual. The - main gas reader must also handle the .mri pseudo-op - to control expression parsing and the like. */ + main gas reader must also handle the .mri pseudo-op + to control expression parsing and the like. */ } } #endif @@ -761,9 +813,26 @@ do_scrub_chars (get, tostart, tolen) break; } +#ifdef KEEP_WHITE_AROUND_COLON + if (lex[ch] == LEX_IS_COLON) + { + /* Only keep this white if there's no white *after* the + colon. */ + ch2 = GET (); + UNGET (ch2); + if (!IS_WHITESPACE (ch2)) + { + state = 9; + UNGET (ch); + PUT (' '); + break; + } + } +#endif if (IS_COMMENT (ch) || ch == '/' - || IS_LINE_SEPARATOR (ch)) + || IS_LINE_SEPARATOR (ch) + || IS_PARALLEL_SEPARATOR (ch)) { if (scrub_m68k_mri) { @@ -831,18 +900,14 @@ do_scrub_chars (get, tostart, tolen) state = 10; /* Sp after symbol char */ goto recycle; case 11: - if (flag_m68k_mri -#ifdef LABELS_WITHOUT_COLONS - || 1 -#endif - ) + if (LABELS_WITHOUT_COLONS || flag_m68k_mri) state = 1; else { /* We know that ch is not ':', since we tested that - case above. Therefore this is not a label, so it - must be the opcode, and we've just seen the - whitespace after it. */ + case above. Therefore this is not a label, so it + must be the opcode, and we've just seen the + whitespace after it. */ state = 3; } UNGET (ch); @@ -874,7 +939,7 @@ do_scrub_chars (get, tostart, tolen) break; /* This UNGET will ensure that we count newlines - correctly. */ + correctly. */ UNGET (ch2); } @@ -884,6 +949,21 @@ do_scrub_chars (get, tostart, tolen) ch = ' '; goto recycle; } +#ifdef DOUBLESLASH_LINE_COMMENTS + else if (ch2 == '/') + { + do + { + ch = GET (); + } + while (ch != EOF && !IS_NEWLINE (ch)); + if (ch == EOF) + as_warn ("end of file in comment; newline inserted"); + state = 0; + PUT ('\n'); + break; + } +#endif else { if (ch2 != EOF) @@ -897,13 +977,13 @@ do_scrub_chars (get, tostart, tolen) case LEX_IS_STRINGQUOTE: if (state == 10) { - /* Preserve the whitespace in foo "bar" */ + /* Preserve the whitespace in foo "bar". */ UNGET (ch); state = 3; PUT (' '); /* PUT didn't jump out. We could just break, but we - know what will happen, so optimize a bit. */ + know what will happen, so optimize a bit. */ ch = GET (); old_state = 3; } @@ -919,7 +999,7 @@ do_scrub_chars (get, tostart, tolen) case LEX_IS_ONECHAR_QUOTE: if (state == 10) { - /* Preserve the whitespace in foo 'b' */ + /* Preserve the whitespace in foo 'b'. */ UNGET (ch); state = 3; PUT (' '); @@ -948,7 +1028,7 @@ do_scrub_chars (get, tostart, tolen) if ((ch = GET ()) != '\'') { #ifdef REQUIRE_CHAR_CLOSE_QUOTE - as_warn (_("Missing close quote: (assumed)")); + as_warn (_("missing close quote; (assumed)")); #else if (ch != EOF) UNGET (ch); @@ -970,10 +1050,14 @@ do_scrub_chars (get, tostart, tolen) #endif case LEX_IS_COLON: +#ifdef KEEP_WHITE_AROUND_COLON + state = 9; +#else if (state == 9 || state == 10) state = 3; else if (state != 3) state = 1; +#endif PUT (ch); break; @@ -984,50 +1068,54 @@ do_scrub_chars (get, tostart, tolen) --add_newlines; UNGET (ch); } - /* fall thru into... */ + /* Fall through. */ case LEX_IS_LINE_SEPARATOR: state = 0; PUT (ch); break; + case LEX_IS_PARALLEL_SEPARATOR: + state = 1; + PUT (ch); + break; + #ifdef TC_V850 case LEX_IS_DOUBLEDASH_1ST: - ch2 = GET(); + ch2 = GET (); if (ch2 != '-') { UNGET (ch2); goto de_fault; } - /* read and skip to end of line */ + /* Read and skip to end of line. */ do { ch = GET (); } while (ch != EOF && ch != '\n'); + if (ch == EOF) - { - as_warn (_("end of file in comment; newline inserted")); - } + as_warn (_("end of file in comment; newline inserted")); + state = 0; PUT ('\n'); break; -#endif -#ifdef TC_M32R +#endif +#ifdef DOUBLEBAR_PARALLEL case LEX_IS_DOUBLEBAR_1ST: - ch2 = GET(); + ch2 = GET (); + UNGET (ch2); if (ch2 != '|') - { - UNGET (ch2); - goto de_fault; - } - /* Reset back to state 1 and pretend that we are parsing a line from - just after the first white space. */ - state = 1; - PUT ('|'); + goto de_fault; + + /* Handle '||' in two states as invoking PUT twice might + result in the first one jumping out of this loop. We'd + then lose track of the state and one '|' char. */ + state = 13; PUT ('|'); break; -#endif +#endif case LEX_IS_LINE_COMMENT_START: /* FIXME-someday: The two character comment stuff was badly thought out. On i386, we want '/' as line comment start @@ -1046,7 +1134,7 @@ do_scrub_chars (get, tostart, tolen) { UNGET (ch2); } - } /* bad hack */ + } if (state == 0 || state == 1) /* Only comment at start of line. */ { @@ -1059,24 +1147,26 @@ do_scrub_chars (get, tostart, tolen) ch = GET (); } while (ch != EOF && IS_WHITESPACE (ch)); + if (ch == EOF) { as_warn (_("end of file in comment; newline inserted")); PUT ('\n'); break; } + if (ch < '0' || ch > '9' || state != 0 || startch != '#') { /* Not a cpp line. */ while (ch != EOF && !IS_NEWLINE (ch)) ch = GET (); if (ch == EOF) - as_warn (_("EOF in Comment: Newline inserted")); + as_warn (_("end of file in comment; newline inserted")); state = 0; PUT ('\n'); break; } - /* Loks like `# 123 "filename"' from cpp. */ + /* Looks like `# 123 "filename"' from cpp. */ UNGET (ch); old_state = 4; state = -1; @@ -1093,12 +1183,14 @@ do_scrub_chars (get, tostart, tolen) Trap is the only short insn that has a first operand that is neither register nor label. We must prevent exef0f ||trap #1 to degenerate to exef0f ||trap#1 . - We can't make '#' LEX_IS_SYMBOL_COMPONENT because it is already - LEX_IS_LINE_COMMENT_START. However, it is the only character in - line_comment_chars for d10v, hence we can recognize it as such. */ + We can't make '#' LEX_IS_SYMBOL_COMPONENT because it is + already LEX_IS_LINE_COMMENT_START. However, it is the + only character in line_comment_chars for d10v, hence we + can recognize it as such. */ /* An alternative approach would be to reset the state to 1 when we see '||', '<'- or '->', but that seems to be overkill. */ - if (state == 10) PUT (' '); + if (state == 10) + PUT (' '); #endif /* We have a line comment character which is not at the start of a line. If this is also a normal comment @@ -1118,11 +1210,15 @@ do_scrub_chars (get, tostart, tolen) #if defined TC_ARM && defined OBJ_ELF /* On the ARM, `@' is the comment character. Unfortunately this is also a special character in ELF .symver - directives (and .type, though we deal with those another way). So - we check if this line is such a directive, and treat the character - as default if so. This is a hack. */ + directives (and .type, though we deal with those another way). + So we check if this line is such a directive, and treat + the character as default if so. This is a hack. */ if ((symver_state != NULL) && (*symver_state == 0)) goto de_fault; +#endif +#ifdef WARN_COMMENTS + if (!found_comment) + as_where (&found_comment_file, &found_comment); #endif do { @@ -1151,7 +1247,7 @@ do_scrub_chars (get, tostart, tolen) state = 9; /* This is a common case. Quickly copy CH and all the - following symbol component or normal characters. */ + following symbol component or normal characters. */ if (to + 1 < toend && mri_state == NULL #if defined TC_ARM && defined OBJ_ELF @@ -1166,21 +1262,23 @@ do_scrub_chars (get, tostart, tolen) { int type; - ch2 = * (unsigned char *) s; + ch2 = *(unsigned char *) s; type = lex[ch2]; if (type != 0 && type != LEX_IS_SYMBOL_COMPONENT) break; } + if (s > from) - { - /* Handle the last character normally, for - simplicity. */ - --s; - } + /* Handle the last character normally, for + simplicity. */ + --s; + len = s - from; + if (len > (toend - to) - 1) len = (toend - to) - 1; + if (len > 0) { PUT (ch); @@ -1203,7 +1301,7 @@ do_scrub_chars (get, tostart, tolen) case 2: *to++ = *from++; case 1: *to++ = *from++; } - } + } ch = GET (); } } @@ -1214,19 +1312,36 @@ do_scrub_chars (get, tostart, tolen) /* Some relatively `normal' character. */ if (state == 0) { - state = 11; /* Now seeing label definition */ + state = 11; /* Now seeing label definition. */ } else if (state == 1) { - state = 2; /* Ditto */ + state = 2; /* Ditto. */ } else if (state == 9) { - if (lex[ch] != LEX_IS_SYMBOL_COMPONENT) + if (!IS_SYMBOL_COMPONENT (ch)) state = 3; } else if (state == 10) { + if (ch == '\\') + { + /* Special handling for backslash: a backslash may + be the beginning of a formal parameter (of a + macro) following another symbol character, with + whitespace in between. If that is the case, we + output a space before the parameter. Strictly + speaking, correct handling depends upon what the + macro parameter expands into; if the parameter + expands into something which does not start with + an operand character, then we don't want to keep + the space. We don't have enough information to + make the right choice, so here we are making the + choice which is more likely to be correct. */ + PUT (' '); + } + state = 3; } PUT (ch); @@ -1254,4 +1369,3 @@ do_scrub_chars (get, tostart, tolen) return to - tostart; } -/* end of app.c */