/* This is the Assembler Pre-Processor
- Copyright (C) 1987, 90, 91, 92, 93, 94, 95, 1996
- Free Software Foundation, Inc.
+ Copyright (C) 1987-2020 Free Software Foundation, Inc.
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)
+ the Free Software Foundation; either version 3, 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.
+ 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. */
+ along with GAS; see the file COPYING. If not, write to the Free
+ Software Foundation, 51 Franklin Street - Fifth Floor, Boston, MA
+ 02110-1301, USA. */
-/* 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
- # <number> <filename> <garbage> into a .line <number>\n.file <filename>
- pair. This needs better error-handling. */
+/* 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 the # in # <number> <filename> <garbage> into a
+ .linefile. This needs better error-handling. */
-#include <stdio.h>
-#include "as.h" /* For BAD_CASE() only */
+#include "as.h"
#if (__STDC__ != 1)
#ifndef const
#endif
#endif
+#ifdef H_TICK_HEX
+int enable_h_tick_hex = 0;
+#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. */
+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.
+ See the comment in do_scrub_chars. */
+static const char symver_pseudo[] = ".symver";
+static const char * symver_state;
+#endif
+#ifdef TC_ARM
+static char last_char;
+#endif
+
static char lex[256];
static const char symbol_chars[] =
"$._ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
#define LEX_IS_COMMENT_START 4
#define LEX_IS_LINE_COMMENT_START 5
#define LEX_IS_TWOCHAR_COMMENT_1ST 6
-#define LEX_IS_TWOCHAR_COMMENT_2ND 7
#define LEX_IS_STRINGQUOTE 8
#define LEX_IS_COLON 9
#define LEX_IS_NEWLINE 10
#define LEX_IS_ONECHAR_QUOTE 11
+#ifdef TC_V850
+#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
+#ifdef H_TICK_HEX
+#define LEX_IS_H 15
+#endif
#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 ()
+void
+do_scrub_begin (int m68k_mri ATTRIBUTE_UNUSED)
{
const char *p;
+ int c;
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;
- if (! flag_m68k_mri)
+#ifdef TC_M68K
+ scrub_m68k_mri = m68k_mri;
+
+ if (! m68k_mri)
+#endif
{
lex['"'] = LEX_IS_STRINGQUOTE;
-#ifndef TC_HPPA
+#if ! defined (TC_HPPA)
lex['\''] = LEX_IS_ONECHAR_QUOTE;
#endif
/* 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;
+
+#ifdef tc_symbol_chars
+ /* This macro permits the processor to specify all characters which
+ may appears in an operand. This will prevent the scrubber from
+ discarding meaningful whitespace in certain cases. The i386
+ backend uses this to support prefixes, which can confuse the
+ scrubber as to whether it is parsing operands or opcodes. */
+ for (p = tc_symbol_chars; *p; ++p)
+ lex[(unsigned char) *p] = LEX_IS_SYMBOL_COMPONENT;
+#endif
- for (p = comment_chars; *p; p++)
- {
- lex[(unsigned char) *p] = LEX_IS_COMMENT_START;
- } /* declare comment chars */
+ /* The m68k backend wants to be able to change comment_chars. */
+#ifndef tc_comment_chars
+#define tc_comment_chars comment_chars
+#endif
+ for (p = tc_comment_chars; *p; p++)
+ 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 */
+#ifndef tc_line_separator_chars
+#define tc_line_separator_chars line_separator_chars
+#endif
+ for (p = tc_line_separator_chars; *p; p++)
+ 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 */
+ /* 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;
- }
- /* FIXME-soon. This is a bad hack but otherwise, we can't do
- c-style comments when '/' is a line comment char. xoxorich. */
- if (lex['*'] == 0)
- {
- lex['*'] = LEX_IS_TWOCHAR_COMMENT_2ND;
- }
+ lex['/'] = LEX_IS_TWOCHAR_COMMENT_1ST;
- if (flag_m68k_mri)
+#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;
}
-} /* do_scrub_begin() */
+#endif
+
+#ifdef TC_V850
+ lex['-'] = LEX_IS_DOUBLEDASH_1ST;
+#endif
+#ifdef DOUBLEBAR_PARALLEL
+ lex['|'] = LEX_IS_DOUBLEBAR_1ST;
+#endif
+#ifdef TC_D30V
+ /* Must do this is we want VLIW instruction with "->" or "<-". */
+ lex['-'] = LEX_IS_SYMBOL_COMPONENT;
+#endif
-/* Saved state of the scrubber */
+#ifdef H_TICK_HEX
+ if (enable_h_tick_hex)
+ {
+ lex['h'] = LEX_IS_H;
+ lex['H'] = LEX_IS_H;
+ }
+#endif
+}
+
+/* Saved state of the scrubber. */
static int state;
static int old_state;
-static char *out_string;
+static const char *out_string;
static char out_buf[20];
static int add_newlines;
static char *saved_input;
-static int saved_input_len;
+static size_t saved_input_len;
+static char input_buffer[32 * 1024];
+static const char *mri_state;
+static char mri_last_ch;
/* Data structure for saving the state of app across #include's. Note that
app is called asynchronously to the parsing of the .include's, so our
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 state;
+ int old_state;
+ const char * out_string;
+ char out_buf[sizeof (out_buf)];
+ int add_newlines;
+ char * saved_input;
+ size_t 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;
+#endif
+#ifdef TC_ARM
+ char last_char;
+#endif
+};
char *
-app_push ()
+app_push (void)
{
- register struct app_save *saved;
+ struct app_save *saved;
- saved = (struct app_save *) xmalloc (sizeof (*saved));
+ saved = XNEW (struct app_save);
saved->state = state;
saved->old_state = old_state;
saved->out_string = out_string;
memcpy (saved->out_buf, out_buf, sizeof (out_buf));
saved->add_newlines = add_newlines;
- saved->saved_input = saved_input;
- saved->saved_input_len = saved_input_len;
+ if (saved_input == NULL)
+ saved->saved_input = NULL;
+ else
+ {
+ saved->saved_input = XNEWVEC (char, saved_input_len);
+ 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
+#ifdef TC_ARM
+ saved->last_char = last_char;
+#endif
- /* do_scrub_begin() is not useful, just wastes time. */
+ /* do_scrub_begin() is not useful, just wastes time. */
state = 0;
saved_input = NULL;
+ add_newlines = 0;
return (char *) saved;
}
-void
-app_pop (arg)
- char *arg;
+void
+app_pop (char *arg)
{
- register struct app_save *saved = (struct app_save *) arg;
+ 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;
memcpy (out_buf, saved->out_buf, sizeof (out_buf));
add_newlines = saved->add_newlines;
- saved_input = saved->saved_input;
- saved_input_len = saved->saved_input_len;
+ if (saved->saved_input == NULL)
+ saved_input = NULL;
+ else
+ {
+ gas_assert (saved->saved_input_len <= sizeof (input_buffer));
+ memcpy (input_buffer, saved->saved_input, saved->saved_input_len);
+ saved_input = input_buffer;
+ 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
+ symver_state = saved->symver_state;
+#endif
+#ifdef TC_ARM
+ last_char = saved->last_char;
+#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)
{
machine, and saves its state so that it may return at any point.
This is the way the old code used to work. */
-int
-do_scrub_chars (get, tostart, tolen)
- int (*get) PARAMS ((char **));
- char *tostart;
- int tolen;
+size_t
+do_scrub_chars (size_t (*get) (char *, size_t), char *tostart, size_t tolen)
{
char *to = tostart;
char *toend = tostart + tolen;
char *from;
char *fromend;
- int fromlen;
- register int ch, ch2 = 0;
- int not_cpp_line = 0;
+ size_t fromlen;
+ int ch, ch2 = 0;
+ /* Character that started the string we're working on. */
+ static char quotechar;
/*State 0: beginning of normal line
1: After first whitespace on line (flush more white)
2: After first non-white (opcode) on line (keep 1white)
3: after second white on line (into operands) (flush white)
- 4: after putting out a .line, put out digits
+ 4: after putting out a .linefile, put out digits
5: parsing a string, then go to old-state
6: putting out \ escape in a "d string.
- 7: After putting out a .appfile, put out string.
- 8: After putting out a .appfile string, flush until newline.
+ 7: no longer used
+ 8: no longer used
9: After seeing symbol char in state 3 (keep 1white after symchar)
10: After seeing whitespace in state 9 (keep white before symchar)
11: After seeing a symbol character in state 0 (eg a label definition)
-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.
+#endif
+#ifdef DOUBLEBAR_PARALLEL
+ 13: After seeing a vertical bar, looking for a second
+ vertical bar as a parallel expression separator.
+#endif
+#ifdef TC_PREDICATE_START_CHAR
+ 14: After seeing a predicate start character at state 0, looking
+ for a predicate end character as predicate.
+ 15: After seeing a predicate start character at state 1, looking
+ for a predicate end character as predicate.
+#endif
+#ifdef TC_Z80
+ 16: After seeing an 'a' or an 'A' at the start of a symbol
+ 17: After seeing an 'f' or an 'F' in state 16
+#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
correctly on the PA (and any other target where colons are optional).
- Jeff Law, law@cs.utah.edu. */
+ Jeff Law, law@cs.utah.edu.
+
+ I added state 13 so that something like "cmp r1, r2 || trap #1" does not
+ get squashed into "cmp r1,r2||trap#1", with the all important space
+ between the 'trap' and the '#1' being eliminated. nickc@cygnus.com */
/* This macro gets the next input character. */
-#define GET() \
- (from < fromend \
- ? *from++ \
- : ((saved_input != NULL \
- ? (free (saved_input), \
- saved_input = NULL, \
- 0) \
- : 0), \
- fromlen = (*get) (&from), \
- fromend = from + fromlen, \
- (fromlen == 0 \
- ? EOF \
- : *from++)))
+#define GET() \
+ (from < fromend \
+ ? * (unsigned char *) (from++) \
+ : (saved_input = NULL, \
+ fromlen = (*get) (input_buffer, sizeof input_buffer), \
+ from = input_buffer, \
+ fromend = from + fromlen, \
+ (fromlen == 0 \
+ ? EOF \
+ : * (unsigned char *) (from++))))
/* This macro pushes a character back on the input stream. */
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)
}
else
{
- fromlen = (*get) (&from);
+ fromlen = (*get) (input_buffer, sizeof input_buffer);
if (fromlen == 0)
return 0;
+ from = input_buffer;
fromend = from + fromlen;
}
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:
if (ch == EOF)
{
- as_warn ("end of file in comment");
+ as_warn (_("end of file in comment"));
goto fromeof;
}
if (ch == EOF)
{
- as_warn ("end of file in comment");
+ as_warn (_("end of file in comment"));
goto fromeof;
}
}
state = old_state;
- PUT (' ');
+ UNGET (' ');
continue;
case 4:
ch = GET ();
if (ch == '"')
{
- UNGET (ch);
- if (flag_m68k_mri)
- out_string = "\n\tappfile ";
- else
- out_string = "\n\t.appfile ";
- old_state = 7;
- state = -1;
- PUT (*out_string++);
+ quotechar = ch;
+ state = 5;
+ old_state = 3;
+ PUT (ch);
}
else
{
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;
+ ptrdiff_t len;
for (s = from; s < fromend; s++)
{
ch = *s;
- /* This condition must be changed if the type of any
- other character can be LEX_IS_STRINGQUOTE. */
if (ch == '\\'
- || ch == '"'
- || ch == '\''
+ || ch == quotechar
|| ch == '\n')
break;
}
memcpy (to, from, len);
to += len;
from += len;
+ if (to >= toend)
+ goto tofull;
}
}
ch = GET ();
if (ch == EOF)
{
- as_warn ("end of file in string: inserted '\"'");
+ /* This buffer is here specifically so
+ that the UNGET below will work. */
+ static char one_char_buf[1];
+
+ as_warn (_("end of file in string; '%c' inserted"), quotechar);
state = old_state;
+ from = fromend = one_char_buf + 1;
+ fromlen = 1;
UNGET ('\n');
- PUT ('"');
+ PUT (quotechar);
}
- else if (lex[ch] == LEX_IS_STRINGQUOTE)
+ else if (ch == quotechar)
{
state = old_state;
PUT (ch);
}
-#ifndef NO_STRING_ESCAPES
- else if (ch == '\\')
+ else if (TC_STRING_ESCAPES && ch == '\\')
{
state = 6;
PUT (ch);
}
-#endif
- else if (flag_m68k_mri && ch == '\n')
+ 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 ('\'');
PUT ('\\');
continue;
+ case EOF:
+ as_warn (_("end of file in string; '%c' inserted"), quotechar);
+ PUT (quotechar);
+ continue;
+
case '"':
case '\\':
case 'b':
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;
- case 7:
+#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 ('|');
+#ifdef TC_TIC6X
+ /* "||^" is used for SPMASKed instructions. */
ch = GET ();
- state = 5;
- old_state = 8;
- if (ch == EOF)
- goto fromeof;
- PUT (ch);
- continue;
-
- case 8:
- do
- ch = GET ();
- while (ch != '\n' && ch != EOF);
if (ch == EOF)
goto fromeof;
- state = 0;
- PUT (ch);
+ else if (ch == '^')
+ PUT ('^');
+ else
+ UNGET (ch);
+#endif
continue;
+#endif
+#ifdef TC_Z80
+ case 16:
+ /* We have seen an 'a' at the start of a symbol, look for an 'f'. */
+ ch = GET ();
+ if (ch == 'f' || ch == 'F')
+ {
+ state = 17;
+ PUT (ch);
+ }
+ else
+ {
+ state = 9;
+ break;
+ }
+ /* Fall through. */
+ case 17:
+ /* We have seen "af" at the start of a symbol,
+ a ' here is a part of that symbol. */
+ ch = GET ();
+ state = 9;
+ if (ch == '\'')
+ /* Change to avoid warning about unclosed string. */
+ PUT ('`');
+ else if (ch != EOF)
+ UNGET (ch);
+ break;
+#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_PREDICATE_START_CHAR
+ if (ch == TC_PREDICATE_START_CHAR && (state == 0 || state == 1))
+ {
+ state += 14;
+ PUT (ch);
+ continue;
+ }
+ else if (state == 14 || state == 15)
+ {
+ if (ch == TC_PREDICATE_END_CHAR)
+ {
+ state -= 14;
+ PUT (ch);
+ ch = GET ();
+ }
+ else
+ {
+ PUT (ch);
+ continue;
+ }
+ }
+#endif
+
recycle:
+
+#if defined TC_ARM && defined OBJ_ELF
+ /* We need to watch out for .symver directives. See the comment later
+ in this function. */
+ if (symver_state == NULL)
+ {
+ if ((state == 0 || state == 1) && ch == symver_pseudo[0])
+ symver_state = symver_pseudo + 1;
+ }
+ else
+ {
+ /* We advance to the next state if we find the right
+ character. */
+ if (ch != '\0' && (*symver_state == ch))
+ ++symver_state;
+ else if (*symver_state != '\0')
+ /* We did not get the expected character, or we didn't
+ get a valid terminating character after seeing the
+ entire pseudo-op, so we must go back to the beginning. */
+ symver_state = NULL;
+ else
+ {
+ /* We've read the entire pseudo-op. If this is the end
+ of the line, go back to the beginning. */
+ if (IS_NEWLINE (ch))
+ symver_state = NULL;
+ }
+ }
+#endif /* TC_ARM && OBJ_ELF */
+
+#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. */
+ if (mri_state == NULL)
+ {
+ if ((state == 0 || state == 1)
+ && ch == mri_pseudo[0])
+ mri_state = mri_pseudo + 1;
+ }
+ else
+ {
+ /* We advance to the next state if we find the right
+ character, or if we need a space character and we get any
+ whitespace character, or if we need a '0' and we get a
+ '1' (this is so that we only need one state to handle
+ ``.mri 0'' and ``.mri 1''). */
+ if (ch != '\0'
+ && (*mri_state == ch
+ || (*mri_state == ' '
+ && lex[ch] == LEX_IS_WHITESPACE)
+ || (*mri_state == '0'
+ && ch == '1')))
+ {
+ mri_last_ch = ch;
+ ++mri_state;
+ }
+ else if (*mri_state != '\0'
+ || (lex[ch] != LEX_IS_WHITESPACE
+ && lex[ch] != LEX_IS_NEWLINE))
+ {
+ /* We did not get the expected character, or we didn't
+ get a valid terminating character after seeing the
+ entire pseudo-op, so we must go back to the
+ beginning. */
+ mri_state = NULL;
+ }
+ else
+ {
+ /* We've read the entire pseudo-op. mips_last_ch is
+ 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. */
+ }
+ }
+#endif
+
if (ch == EOF)
{
if (state != 0)
{
- as_warn ("end of file not at end of a line; newline inserted");
+ as_warn (_("end of file not at end of a line; newline inserted"));
state = 0;
PUT ('\n');
}
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 ();
+ if (ch2 != EOF)
+ 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))
{
- /* cpp never outputs a leading space before the #, so
- try to avoid being confused. */
- not_cpp_line = 1;
- if (flag_m68k_mri)
+ if (scrub_m68k_mri)
{
/* In MRI mode, we keep these spaces. */
UNGET (ch);
not permitted between the label and the colon. */
if ((state == 2 || state == 11)
&& lex[ch] == LEX_IS_COLON
- && ! flag_m68k_mri)
+ && ! scrub_m68k_mri)
{
state = 1;
PUT (ch);
switch (state)
{
- case 0:
- state++;
- goto recycle; /* Punted leading sp */
case 1:
/* We can arrive here if we leave a leading whitespace
character at the beginning of a line. */
PUT (' ');
break;
case 3:
- if (flag_m68k_mri)
+#ifndef TC_KEEP_OPERAND_SPACES
+ /* For TI C6X, we keep these spaces as they may separate
+ functional unit specifiers from operands. */
+ if (scrub_m68k_mri)
+#endif
{
/* In MRI mode, we keep these spaces. */
UNGET (ch);
goto recycle; /* Sp in operands */
case 9:
case 10:
- if (flag_m68k_mri)
+#ifndef TC_KEEP_OPERAND_SPACES
+ if (scrub_m68k_mri)
+#endif
{
/* In MRI mode, we keep these spaces. */
state = 3;
state = 10; /* Sp after symbol char */
goto recycle;
case 11:
- state = 1;
+ 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. */
+ state = 3;
+ }
UNGET (ch);
PUT (' '); /* Sp after label definition. */
break;
case LEX_IS_TWOCHAR_COMMENT_1ST:
ch2 = GET ();
- if (ch2 != EOF && lex[ch2] == LEX_IS_TWOCHAR_COMMENT_2ND)
+ if (ch2 == '*')
{
for (;;)
{
if (ch2 != EOF && IS_NEWLINE (ch2))
add_newlines++;
}
- while (ch2 != EOF &&
- (lex[ch2] != LEX_IS_TWOCHAR_COMMENT_2ND));
+ while (ch2 != EOF && ch2 != '*');
- while (ch2 != EOF &&
- (lex[ch2] == LEX_IS_TWOCHAR_COMMENT_2ND))
- {
- ch2 = GET ();
- }
+ while (ch2 == '*')
+ ch2 = GET ();
- if (ch2 == EOF
- || lex[ch2] == LEX_IS_TWOCHAR_COMMENT_1ST)
+ if (ch2 == EOF || ch2 == '/')
break;
- UNGET (ch);
+
+ /* This UNGET will ensure that we count newlines
+ correctly. */
+ UNGET (ch2);
}
+
if (ch2 == EOF)
- as_warn ("end of file in multiline comment");
+ as_warn (_("end of file in multiline comment"));
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)
break;
case LEX_IS_STRINGQUOTE:
+ quotechar = ch;
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;
}
PUT (ch);
break;
-#ifndef IEEE_STYLE
case LEX_IS_ONECHAR_QUOTE:
+#ifdef H_TICK_HEX
+ if (state == 9 && enable_h_tick_hex)
+ {
+ char c;
+
+ c = GET ();
+ as_warn ("'%c found after symbol", c);
+ UNGET (c);
+ }
+#endif
if (state == 10)
{
- /* Preserve the whitespace in foo 'b' */
+ /* Preserve the whitespace in foo 'b'. */
UNGET (ch);
state = 3;
PUT (' ');
ch = GET ();
if (ch == EOF)
{
- as_warn ("end of file after a one-character quote; \\0 inserted");
+ as_warn (_("end of file after a one-character quote; \\0 inserted"));
ch = 0;
}
if (ch == '\\')
ch = GET ();
if (ch == EOF)
{
- as_warn ("end of file in escape character");
+ as_warn (_("end of file in escape character"));
ch = '\\';
}
else
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);
out_string = out_buf;
PUT (*out_string++);
break;
-#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;
--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 ();
+ if (ch2 != '-')
+ {
+ if (ch2 != EOF)
+ UNGET (ch2);
+ goto de_fault;
+ }
+ /* 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"));
+
+ state = 0;
+ PUT ('\n');
+ break;
+#endif
+#ifdef DOUBLEBAR_PARALLEL
+ case LEX_IS_DOUBLEBAR_1ST:
+ ch2 = GET ();
+ if (ch2 != EOF)
+ UNGET (ch2);
+ if (ch2 != '|')
+ 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
case LEX_IS_LINE_COMMENT_START:
- if (state == 0) /* Only comment at start of line. */
+ /* FIXME-someday: The two character comment stuff was badly
+ thought out. On i386, we want '/' as line comment start
+ AND we want C style comments. hence this hack. The
+ whole lexical process should be reworked. xoxorich. */
+ if (ch == '/')
{
- /* FIXME-someday: The two character comment stuff was
- badly thought out. On i386, we want '/' as line
- comment start AND we want C style comments. hence
- this hack. The whole lexical process should be
- reworked. xoxorich. */
- if (ch == '/')
+ ch2 = GET ();
+ if (ch2 == '*')
{
- ch2 = GET ();
- if (ch2 == '*')
- {
- state = -2;
- break;
- }
- else
- {
- UNGET (ch2);
- }
- } /* bad hack */
+ old_state = 3;
+ state = -2;
+ break;
+ }
+ else if (ch2 != EOF)
+ {
+ UNGET (ch2);
+ }
+ }
- if (ch != '#')
- not_cpp_line = 1;
+ if (state == 0 || state == 1) /* Only comment at start of line. */
+ {
+ int startch;
+
+ startch = ch;
do
{
ch = GET ();
}
while (ch != EOF && IS_WHITESPACE (ch));
+
if (ch == EOF)
{
- as_warn ("end of file in comment; newline inserted");
+ as_warn (_("end of file in comment; newline inserted"));
PUT ('\n');
break;
}
- if (ch < '0' || ch > '9' || not_cpp_line)
+
+ if (ch < '0' || ch > '9' || state != 0 || startch != '#')
{
- /* Non-numerics: Eat whole comment line */
+ /* 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"));
+ PUT ('\n');
+ }
+ else /* IS_NEWLINE (ch) */
+ {
+ /* To process non-zero add_newlines. */
+ UNGET (ch);
+ }
state = 0;
- PUT ('\n');
break;
}
- /* Numerics begin comment. Perhaps CPP `# 123 "filename"' */
+ /* Looks like `# 123 "filename"' from cpp. */
UNGET (ch);
old_state = 4;
state = -1;
- if (flag_m68k_mri)
- out_string = "\tappline ";
+ if (scrub_m68k_mri)
+ out_string = "\tlinefile ";
else
- out_string = "\t.appline ";
+ out_string = "\t.linefile ";
PUT (*out_string++);
break;
}
+#ifdef TC_D10V
+ /* All insns end in a char for which LEX_IS_SYMBOL_COMPONENT is true.
+ 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. */
+ /* 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 (' ');
+#endif
/* We have a line comment character which is not at the
start of a line. If this is also a normal comment
character, fall through. Otherwise treat it as a default
character. */
- if (strchr (comment_chars, ch) == NULL
- && (! flag_m68k_mri
+ if (strchr (tc_comment_chars, ch) == NULL
+ && (! scrub_m68k_mri
|| (ch != '!' && ch != '*')))
goto de_fault;
- if (flag_m68k_mri
+ if (scrub_m68k_mri
&& (ch == '!' || ch == '*' || ch == '#')
&& state != 1
&& state != 10)
goto de_fault;
/* Fall through. */
case LEX_IS_COMMENT_START:
+#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. */
+ if ((symver_state != NULL) && (*symver_state == 0))
+ goto de_fault;
+#endif
+
+#ifdef TC_ARM
+ /* For the ARM, care is needed not to damage occurrences of \@
+ by stripping the @ onwards. Yuck. */
+ if ((to > tostart ? to[-1] : last_char) == '\\')
+ /* Do not treat the @ as a start-of-comment. */
+ goto de_fault;
+#endif
+
+#ifdef WARN_COMMENTS
+ if (!found_comment)
+ found_comment_file = as_where (&found_comment);
+#endif
do
{
ch = GET ();
}
while (ch != EOF && !IS_NEWLINE (ch));
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;
+#ifdef H_TICK_HEX
+ case LEX_IS_H:
+ /* Look for strings like H'[0-9A-Fa-f] and if found, replace
+ the H' with 0x to make them gas-style hex characters. */
+ if (enable_h_tick_hex)
+ {
+ char quot;
+
+ quot = GET ();
+ if (quot == '\'')
+ {
+ UNGET ('x');
+ ch = '0';
+ }
+ else
+ UNGET (quot);
+ }
+#endif
+ /* Fall through. */
+
case LEX_IS_SYMBOL_COMPONENT:
if (state == 10)
{
break;
}
+#ifdef TC_Z80
+ /* "af'" is a symbol containing '\''. */
+ if (state == 3 && (ch == 'a' || ch == 'A'))
+ {
+ state = 16;
+ PUT (ch);
+ ch = GET ();
+ if (ch == 'f' || ch == 'F')
+ {
+ state = 17;
+ PUT (ch);
+ break;
+ }
+ else
+ {
+ state = 9;
+ if (ch == EOF || !IS_SYMBOL_COMPONENT (ch))
+ {
+ if (ch != EOF)
+ UNGET (ch);
+ break;
+ }
+ }
+ }
+#endif
if (state == 3)
state = 9;
/* This is a common case. Quickly copy CH and all the
- following symbol component or normal characters. */
- if (to + 1 < toend)
+ following symbol component or normal characters. */
+ if (to + 1 < toend
+ && mri_state == NULL
+#if defined TC_ARM && defined OBJ_ELF
+ && symver_state == NULL
+#endif
+ )
{
char *s;
- int len;
+ ptrdiff_t len;
for (s = from; s < fromend; s++)
{
int type;
- ch2 = *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);
- if (len > 8)
- {
- memcpy (to, from, len);
- to += len;
- from += len;
- }
- else
- {
- switch (len)
- {
- case 8: *to++ = *from++;
- case 7: *to++ = *from++;
- case 6: *to++ = *from++;
- case 5: *to++ = *from++;
- case 4: *to++ = *from++;
- case 3: *to++ = *from++;
- case 2: *to++ = *from++;
- case 1: *to++ = *from++;
- }
- }
+ memcpy (to, from, len);
+ to += len;
+ from += len;
+ if (to >= toend)
+ goto tofull;
ch = GET ();
}
}
/* 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. */
+ if (to + 1 >= toend)
+ {
+ /* If we're near the end of the buffer, save the
+ character for the next time round. Otherwise
+ we'll lose our state. */
+ UNGET (ch);
+ goto tofull;
+ }
+ *to++ = ' ';
+ }
+
state = 3;
}
PUT (ch);
fromeof:
/* We have reached the end of the input. */
+#ifdef TC_ARM
+ if (to > tostart)
+ last_char = to[-1];
+#endif
return to - tostart;
tofull:
processed. */
if (fromend > from)
{
- char *save;
-
- save = (char *) xmalloc (fromend - from);
- memcpy (save, from, fromend - from);
- if (saved_input != NULL)
- free (saved_input);
- saved_input = save;
+ saved_input = from;
saved_input_len = fromend - from;
}
else
- {
- if (saved_input != NULL)
- {
- free (saved_input);
- saved_input = NULL;
- }
- }
+ saved_input = NULL;
+
+#ifdef TC_ARM
+ if (to > tostart)
+ last_char = to[-1];
+#endif
return to - tostart;
}
-
-/* end of app.c */