gas: fold is_end_of_line[] into lex_type[]
... by way of introducing LEX_EOL and LEX_EOS. As a prereq convert the
remaining open-coded accesses.
The Alpha change is actually a functional one: The array slot for '!'
having been set to 1 is very unlikely to have been correct. 1 means "end
of line", when surely "end of statement" was always meant.
diff --git a/gas/cond.c b/gas/cond.c
index 9213f91..c14399d 100644
--- a/gas/cond.c
+++ b/gas/cond.c
@@ -141,7 +141,7 @@
if (current_cframe != NULL && current_cframe->ignoring)
{
operand.X_add_number = 0;
- while (! is_end_of_line[(unsigned char) *input_line_pointer])
+ while (! is_end_of_stmt (*input_line_pointer))
++input_line_pointer;
}
else
@@ -201,7 +201,7 @@
int is_eol;
SKIP_WHITESPACE ();
- is_eol = is_end_of_line[(unsigned char) *input_line_pointer];
+ is_eol = is_end_of_stmt (*input_line_pointer);
cframe.ignoring = (test_blank == !is_eol);
}
@@ -232,7 +232,7 @@
{
++s;
++input_line_pointer;
- while (! is_end_of_line[(unsigned char) *input_line_pointer])
+ while (! is_end_of_stmt (*input_line_pointer))
{
*s++ = *input_line_pointer++;
if (s[-1] == '\'')
@@ -247,7 +247,7 @@
else
{
while (*input_line_pointer != terminator
- && ! is_end_of_line[(unsigned char) *input_line_pointer])
+ && ! is_end_of_stmt (*input_line_pointer))
++input_line_pointer;
s = input_line_pointer;
while (s > ret && is_whitespace (s[-1]))
@@ -330,7 +330,7 @@
if (current_cframe == NULL || current_cframe->ignoring)
{
- while (! is_end_of_line[(unsigned char) *input_line_pointer])
+ while (! is_end_of_stmt (*input_line_pointer))
++input_line_pointer;
if (current_cframe == NULL)
@@ -401,7 +401,7 @@
if (flag_mri)
{
- while (! is_end_of_line[(unsigned char) *input_line_pointer])
+ while (! is_end_of_stmt (*input_line_pointer))
++input_line_pointer;
}
@@ -448,7 +448,7 @@
if (flag_mri)
{
- while (! is_end_of_line[(unsigned char) *input_line_pointer])
+ while (! is_end_of_stmt (*input_line_pointer))
++input_line_pointer;
}
diff --git a/gas/config/tc-alpha.c b/gas/config/tc-alpha.c
index b39a7c8..d6245e5 100644
--- a/gas/config/tc-alpha.c
+++ b/gas/config/tc-alpha.c
@@ -888,7 +888,7 @@
#ifdef RELOC_OP_P
/* ??? Wrest control of ! away from the regular expression parser. */
- is_end_of_line[(unsigned char) '!'] = 1;
+ lex_type[(unsigned char) '!'] |= LEX_EOS;
#endif
while (tok < end_tok && *input_line_pointer)
@@ -1029,21 +1029,21 @@
debug_exp (orig_tok, ntok - (end_tok - tok));
#endif
#ifdef RELOC_OP_P
- is_end_of_line[(unsigned char) '!'] = 0;
+ lex_type[(unsigned char) '!'] &= ~LEX_EOS;
#endif
return ntok - (end_tok - tok);
err:
#ifdef RELOC_OP_P
- is_end_of_line[(unsigned char) '!'] = 0;
+ lex_type[(unsigned char) '!'] &= ~LEX_EOS;
#endif
input_line_pointer = old_input_line_pointer;
return TOKENIZE_ERROR;
#ifdef RELOC_OP_P
err_report:
- is_end_of_line[(unsigned char) '!'] = 0;
+ lex_type[(unsigned char) '!'] &= ~LEX_EOS;
#endif
input_line_pointer = old_input_line_pointer;
return TOKENIZE_ERROR_REPORT;
diff --git a/gas/expr.c b/gas/expr.c
index b22346e..099be43 100644
--- a/gas/expr.c
+++ b/gas/expr.c
@@ -298,7 +298,7 @@
#define valuesize 32
#endif
- if (is_end_of_line[(unsigned char) *input_line_pointer])
+ if (is_end_of_stmt (*input_line_pointer))
{
expressionP->X_op = O_absent;
return;
@@ -803,7 +803,7 @@
SKIP_WHITESPACE (); /* Leading whitespace is part of operand. */
c = *input_line_pointer++; /* input_line_pointer -> past char in c. */
- if (is_end_of_line[(unsigned char) c])
+ if (is_end_of_stmt (c))
goto eol;
switch (c)
@@ -946,7 +946,7 @@
/* If it says "0f" and it could possibly be a floating point
number, make it one. Otherwise, make it a local label,
and try to deal with parsing the rest later. */
- if (!is_end_of_line[(unsigned char) input_line_pointer[1]]
+ if (!is_end_of_stmt (input_line_pointer[1])
&& strchr (FLT_CHARS, 'f') != NULL)
{
char *cp = input_line_pointer + 1;
@@ -1668,7 +1668,7 @@
c = *input_line_pointer & 0xff;
*num_chars = 1;
- if (is_end_of_line[c])
+ if (is_end_of_stmt (c))
return O_illegal;
#ifdef md_operator
diff --git a/gas/listing.c b/gas/listing.c
index be71597..a84541e 100644
--- a/gas/listing.c
+++ b/gas/listing.c
@@ -362,8 +362,7 @@
int seen_slash = 0;
for (copy = input_line_pointer;
- *copy && (seen_quote
- || is_end_of_line [(unsigned char) *copy] != 1);
+ seen_quote ? *copy : !is_end_of_line (*copy);
copy++)
{
if (seen_slash)
@@ -1580,7 +1579,7 @@
{
if (quoted
? *input_line_pointer == '\"'
- : is_end_of_line[(unsigned char) *input_line_pointer])
+ : is_end_of_stmt (*input_line_pointer))
{
if (listing)
{
diff --git a/gas/read.c b/gas/read.c
index 2939c68..e451721 100644
--- a/gas/read.c
+++ b/gas/read.c
@@ -79,7 +79,7 @@
#ifndef CR_EOL
#define LEX_CR LEX_WHITE
#else
-#define LEX_CR 0
+#define LEX_CR LEX_EOL
#endif
#ifndef LEX_AT
@@ -118,7 +118,7 @@
/* Used by is_... macros. our ctype[]. */
char lex_type[256] = {
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 0, 0, 0, LEX_CR, 0, 0, /* @ABCDEFGHIJKLMNO */
+ 0x20, 0, 0, 0, 0, 0, 0, 0, 0, 8, 0x20, 0, 0, LEX_CR, 0, 0, /* @ABCDEFGHIJKLMNO */
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* PQRSTUVWXYZ[\]^_ */
8, 0, 0, LEX_HASH, LEX_DOLLAR, LEX_PCT, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0, /* _!"#$%&'()*+,-./ */
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, LEX_QM, /* 0123456789:;<=>? */
@@ -136,32 +136,6 @@
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3
};
-/* In: a character.
- Out: 1 if this character ends a line.
- 2 if this character is a line separator. */
-char is_end_of_line[256] = {
-#ifdef CR_EOL
- 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 0, /* @abcdefghijklmno */
-#else
- 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, /* @abcdefghijklmno */
-#endif
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* _!"#$%&'()*+,-./ */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* 0123456789:;<=>? */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, /* */
- 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 /* */
-};
-
#ifndef TC_CASE_SENSITIVE
char original_case_string[128];
#endif
@@ -290,8 +264,7 @@
#endif
/* Use machine dependent syntax. */
for (p = tc_line_separator_chars; *p; p++)
- is_end_of_line[(unsigned char) *p] = 2;
- /* Use more. FIXME-SOMEDAY. */
+ lex_type[(unsigned char) *p] = LEX_EOS;
if (flag_mri)
lex_type['?'] = LEX_BEGIN_NAME | LEX_NAME;
@@ -944,7 +917,8 @@
/* We now have input_line_pointer->1st char of next line.
If input_line_pointer [-1] == '\n' then we just
scanned another line: so bump line counters. */
- was_new_line = is_end_of_line[(unsigned char) input_line_pointer[-1]];
+ was_new_line = lex_type[(unsigned char) input_line_pointer[-1]]
+ & (LEX_EOL | LEX_EOS);
if (was_new_line)
{
symbol_set_value_now (&dot_symbol);
@@ -993,7 +967,7 @@
#endif
next_char = *input_line_pointer;
- if (was_new_line == 1
+ if ((was_new_line & LEX_EOL)
&& (strchr (line_comment_chars, '#')
? next_char == '#'
: next_char && strchr (line_comment_chars, next_char)))
@@ -1087,7 +1061,7 @@
}
if (strncasecmp (rest, "MACRO", 5) == 0
&& (is_whitespace (rest[5])
- || is_end_of_line[(unsigned char) rest[5]]))
+ || is_end_of_stmt (rest[5])))
mri_line_macro = 1;
}
@@ -1343,7 +1317,7 @@
}
/* Empty statement? */
- if (is_end_of_line[(unsigned char) next_char])
+ if (is_end_of_stmt (next_char))
continue;
if ((LOCAL_LABELS_DOLLAR || LOCAL_LABELS_FB) && ISDIGIT (next_char))
@@ -1502,7 +1476,7 @@
know (flag_m68k_mri);
for (s = input_line_pointer;
- ((!is_end_of_line[(unsigned char) *s] && !is_whitespace (*s))
+ ((!is_end_of_stmt (*s) && !is_whitespace (*s))
|| inquote);
s++)
{
@@ -1511,7 +1485,7 @@
}
#else
for (s = input_line_pointer;
- !is_end_of_line[(unsigned char) *s];
+ !is_end_of_stmt (*s);
s++)
;
#endif
@@ -1530,7 +1504,7 @@
input_line_pointer = stop;
*stop = stopc;
- while (!is_end_of_line[(unsigned char) *input_line_pointer])
+ while (!is_end_of_stmt (*input_line_pointer))
++input_line_pointer;
}
@@ -1562,7 +1536,7 @@
if (flag_mri)
stop = mri_comment_field (&stopc);
- if (is_end_of_line[(unsigned char) *input_line_pointer])
+ if (is_end_of_stmt (*input_line_pointer))
{
if (arg < 0)
align = 0;
@@ -2038,7 +2012,7 @@
backquote. */
if (flag_m68k_mri
&& *input_line_pointer == '\''
- && is_end_of_line[(unsigned char) input_line_pointer[1]])
+ && is_end_of_stmt (input_line_pointer[1]))
++input_line_pointer;
demand_empty_rest_of_line ();
@@ -2161,7 +2135,7 @@
break;
}
- if (!is_end_of_line[(unsigned char)*input_line_pointer])
+ if (!is_end_of_stmt (*input_line_pointer))
file = NULL;
}
@@ -2196,7 +2170,7 @@
/* The MRI assembler permits the start symbol to follow .end,
but we don't support that. */
SKIP_WHITESPACE ();
- if (!is_end_of_line[(unsigned char) *input_line_pointer]
+ if (!is_end_of_stmt (*input_line_pointer)
&& *input_line_pointer != '*'
&& *input_line_pointer != '!')
as_warn (_("start address not supported"));
@@ -2425,7 +2399,7 @@
{
input_line_pointer++;
SKIP_WHITESPACE ();
- if (is_end_of_line[(unsigned char) *input_line_pointer])
+ if (is_end_of_stmt (*input_line_pointer))
c = '\n';
}
@@ -2485,7 +2459,7 @@
type = LINKONCE_DISCARD;
- if (!is_end_of_line[(unsigned char) *input_line_pointer])
+ if (!is_end_of_stmt (*input_line_pointer))
{
char *s;
char c;
@@ -4016,7 +3990,7 @@
SKIP_WHITESPACE ();
if (input_line_pointer > buffer_limit)
return;
- if (is_end_of_line[(unsigned char) *input_line_pointer])
+ if (is_end_of_stmt (*input_line_pointer))
input_line_pointer++;
else
{
@@ -5784,7 +5758,7 @@
{
c = * input_line_pointer ++;
- if (c >= 256 || is_end_of_line [c])
+ if (c >= 256 || is_end_of_stmt (c))
{
as_bad (_("end of line encountered inside .base64 string"));
ignore_rest_of_line ();
@@ -6311,7 +6285,7 @@
is_it_end_of_statement (void)
{
SKIP_WHITESPACE ();
- return (is_end_of_line[(unsigned char) *input_line_pointer]);
+ return is_end_of_stmt (*input_line_pointer);
}
void
@@ -6499,7 +6473,7 @@
{
SKIP_WHITESPACE ();
i = 0;
- while (!is_end_of_line[(unsigned char) *input_line_pointer]
+ while (!is_end_of_stmt (*input_line_pointer)
&& !is_whitespace (*input_line_pointer))
{
obstack_1grow (¬es, *input_line_pointer);
@@ -6509,7 +6483,7 @@
obstack_1grow (¬es, '\0');
filename = (char *) obstack_finish (¬es);
- while (!is_end_of_line[(unsigned char) *input_line_pointer])
+ while (!is_end_of_stmt (*input_line_pointer))
++input_line_pointer;
}
@@ -6795,7 +6769,7 @@
char inquote = '\0';
int inescape = 0;
- while (!is_end_of_line[(unsigned char) *s]
+ while (!is_end_of_stmt (*s)
|| (inquote && !ISCNTRL (*s))
|| (inquote == '\'' && flag_mri)
#ifdef TC_EOL_IN_INSN
@@ -6804,7 +6778,7 @@
/* PR 6926: When we are parsing the body of a macro the sequence
\@ is special - it refers to the invocation count. If the @
character happens to be registered as a line-separator character
- by the target, then the is_end_of_line[] test above will have
+ by the target, then the is_end_of_stmt() test above will have
returned true, but we need to ignore the line separating
semantics in this particular case. */
|| (in_macro && inescape && *s == '@')
diff --git a/gas/read.h b/gas/read.h
index 535f028..f6d0167 100644
--- a/gas/read.h
+++ b/gas/read.h
@@ -41,6 +41,8 @@
#define LEX_BEGIN_NAME (2) /* may begin a name */
#define LEX_END_NAME (4) /* ends a name */
#define LEX_WHITE (8) /* whitespace */
+#define LEX_EOS (0x10) /* end of statement */
+#define LEX_EOL (0x20) /* end of line */
#define is_name_beginner(c) \
( lex_type[(unsigned char) (c)] & LEX_BEGIN_NAME )
@@ -56,12 +58,11 @@
#undef ISBLANK
/* The distinction of "line" and "statement" sadly is blurred by unhelpful
- naming of e.g. the underlying array. Most users really mean "end of
- statement". Going forward only these wrappers are supposed to be used. */
+ naming in a few places. Most users really mean "end of statement". */
#define is_end_of_stmt(c) \
- (is_end_of_line[(unsigned char) (c)])
+ ( lex_type[(unsigned char) (c)] & (LEX_EOS | LEX_EOL) )
#define is_end_of_line(c) \
- (is_end_of_line[(unsigned char) (c)] == 1)
+ ( lex_type[(unsigned char) (c)] & LEX_EOL )
#ifndef is_a_char
#define CHAR_MASK (0xff)
@@ -70,7 +71,6 @@
#endif /* is_a_char() */
extern char lex_type[];
-extern char is_end_of_line[];
extern int is_it_end_of_statement (void);
extern char *find_end_of_line (char *, int);
