old-cross-binutils/gdb/parse.c

1408 lines
36 KiB
C
Raw Normal View History

/* Parse expressions for GDB.
1999-06-07 19:19:32 +00:00
Copyright (C) 1986, 89, 90, 91, 94, 98, 1999 Free Software Foundation, Inc.
Modified from expread.y by the Department of Computer Science at the
State University of New York at Buffalo, 1991.
1999-07-07 20:19:36 +00:00
This file is part of GDB.
1999-07-07 20:19:36 +00:00
This program 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 of the License, or
(at your option) any later version.
1999-07-07 20:19:36 +00:00
This program 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.
1999-07-07 20:19:36 +00:00
You should have received a copy of the GNU General Public License
along with this program; if not, write to the Free Software
Foundation, Inc., 59 Temple Place - Suite 330,
Boston, MA 02111-1307, USA. */
/* Parse an expression from text in a string,
and return the result as a struct expression pointer.
That structure contains arithmetic operations in reverse polish,
with constants represented by operations that are followed by special data.
See expression.h for the details of the format.
What is important here is that it can be built up sequentially
during the process of parsing; the lower levels of the tree always
come first in the result. */
1999-07-07 20:19:36 +00:00
1999-06-07 19:19:32 +00:00
#include <ctype.h>
#include "defs.h"
#include "gdb_string.h"
#include "symtab.h"
#include "gdbtypes.h"
#include "frame.h"
#include "expression.h"
#include "value.h"
#include "command.h"
#include "language.h"
#include "parser-defs.h"
#include "gdbcmd.h"
1999-07-07 20:19:36 +00:00
#include "symfile.h" /* for overlay functions */
1999-10-12 04:37:53 +00:00
/* Symbols which architectures can redefine. */
/* Some systems have routines whose names start with `$'. Giving this
macro a non-zero value tells GDB's expression parser to check for
such routines when parsing tokens that begin with `$'.
On HP-UX, certain system routines (millicode) have names beginning
with `$' or `$$'. For example, `$$dyncall' is a millicode routine
that handles inter-space procedure calls on PA-RISC. */
#ifndef SYMBOLS_CAN_START_WITH_DOLLAR
#define SYMBOLS_CAN_START_WITH_DOLLAR (0)
#endif
/* Global variables declared in parser-defs.h (and commented there). */
struct expression *expout;
int expout_size;
int expout_ptr;
struct block *expression_context_block;
struct block *innermost_block;
int arglist_len;
union type_stack_elt *type_stack;
int type_stack_depth, type_stack_size;
char *lexptr;
char *namecopy;
int paren_depth;
int comma_terminates;
static int expressiondebug = 0;
extern int hp_som_som_object_present;
2000-05-22 09:02:23 +00:00
static void free_funcalls (void *ignore);
2000-05-28 01:12:42 +00:00
static void prefixify_expression (struct expression *);
static void
2000-05-28 01:12:42 +00:00
prefixify_subexp (struct expression *, struct expression *, int, int);
2000-05-28 01:12:42 +00:00
void _initialize_parse (void);
1999-05-25 18:09:09 +00:00
/* Data structure for saving values of arglist_len for function calls whose
arguments contain other function calls. */
struct funcall
{
struct funcall *next;
int arglist_len;
};
static struct funcall *funcall_chain;
/* Assign machine-independent names to certain registers
(unless overridden by the REGISTER_NAMES table) */
unsigned num_std_regs = 0;
1999-06-07 19:19:32 +00:00
struct std_regs *std_regs;
/* The generic method for targets to specify how their registers are
named. The mapping can be derived from three sources:
REGISTER_NAME; std_regs; or a target specific alias hook. */
int
target_map_name_to_register (str, len)
char *str;
int len;
{
int i;
/* First try target specific aliases. We try these first because on some
systems standard names can be context dependent (eg. $pc on a
multiprocessor can be could be any of several PCs). */
#ifdef REGISTER_NAME_ALIAS_HOOK
1999-07-07 20:19:36 +00:00
i = REGISTER_NAME_ALIAS_HOOK (str, len);
if (i >= 0)
return i;
#endif
/* Search architectural register name space. */
for (i = 0; i < NUM_REGS; i++)
if (REGISTER_NAME (i) && len == strlen (REGISTER_NAME (i))
&& STREQN (str, REGISTER_NAME (i), len))
{
return i;
}
/* Try standard aliases */
for (i = 0; i < num_std_regs; i++)
if (std_regs[i].name && len == strlen (std_regs[i].name)
&& STREQN (str, std_regs[i].name, len))
{
return std_regs[i].regnum;
}
return -1;
}
/* Begin counting arguments for a function call,
saving the data about any containing call. */
void
start_arglist ()
{
register struct funcall *new;
new = (struct funcall *) xmalloc (sizeof (struct funcall));
new->next = funcall_chain;
new->arglist_len = arglist_len;
arglist_len = 0;
funcall_chain = new;
}
/* Return the number of arguments in a function call just terminated,
and restore the data for the containing function call. */
int
end_arglist ()
{
register int val = arglist_len;
register struct funcall *call = funcall_chain;
funcall_chain = call->next;
arglist_len = call->arglist_len;
1999-07-07 20:19:36 +00:00
free ((PTR) call);
return val;
}
/* Free everything in the funcall chain.
Used when there is an error inside parsing. */
static void
2000-05-22 09:02:23 +00:00
free_funcalls (void *ignore)
{
register struct funcall *call, *next;
for (call = funcall_chain; call; call = next)
{
next = call->next;
1999-07-07 20:19:36 +00:00
free ((PTR) call);
}
}
/* This page contains the functions for adding data to the struct expression
being constructed. */
/* Add one element to the end of the expression. */
/* To avoid a bug in the Sun 4 compiler, we pass things that can fit into
a register through here */
void
write_exp_elt (expelt)
union exp_element expelt;
{
if (expout_ptr >= expout_size)
{
expout_size *= 2;
expout = (struct expression *)
xrealloc ((char *) expout, sizeof (struct expression)
+ EXP_ELEM_TO_BYTES (expout_size));
}
expout->elts[expout_ptr++] = expelt;
}
void
write_exp_elt_opcode (expelt)
enum exp_opcode expelt;
{
union exp_element tmp;
tmp.opcode = expelt;
write_exp_elt (tmp);
}
void
write_exp_elt_sym (expelt)
struct symbol *expelt;
{
union exp_element tmp;
tmp.symbol = expelt;
write_exp_elt (tmp);
}
void
write_exp_elt_block (b)
struct block *b;
{
union exp_element tmp;
tmp.block = b;
write_exp_elt (tmp);
}
void
write_exp_elt_longcst (expelt)
LONGEST expelt;
{
union exp_element tmp;
tmp.longconst = expelt;
write_exp_elt (tmp);
}
void
write_exp_elt_dblcst (expelt)
DOUBLEST expelt;
{
union exp_element tmp;
tmp.doubleconst = expelt;
write_exp_elt (tmp);
}
void
write_exp_elt_type (expelt)
struct type *expelt;
{
union exp_element tmp;
tmp.type = expelt;
write_exp_elt (tmp);
}
void
write_exp_elt_intern (expelt)
struct internalvar *expelt;
{
union exp_element tmp;
tmp.internalvar = expelt;
write_exp_elt (tmp);
}
/* Add a string constant to the end of the expression.
String constants are stored by first writing an expression element
that contains the length of the string, then stuffing the string
constant itself into however many expression elements are needed
to hold it, and then writing another expression element that contains
the length of the string. I.E. an expression element at each end of
the string records the string length, so you can skip over the
expression elements containing the actual string bytes from either
end of the string. Note that this also allows gdb to handle
strings with embedded null bytes, as is required for some languages.
Don't be fooled by the fact that the string is null byte terminated,
this is strictly for the convenience of debugging gdb itself. Gdb
Gdb does not depend up the string being null terminated, since the
actual length is recorded in expression elements at each end of the
string. The null byte is taken into consideration when computing how
many expression elements are required to hold the string constant, of
course. */
void
write_exp_string (str)
struct stoken str;
{
register int len = str.length;
register int lenelt;
register char *strdata;
/* Compute the number of expression elements required to hold the string
(including a null byte terminator), along with one expression element
at each end to record the actual string length (not including the
null byte terminator). */
lenelt = 2 + BYTES_TO_EXP_ELEM (len + 1);
/* Ensure that we have enough available expression elements to store
everything. */
if ((expout_ptr + lenelt) >= expout_size)
{
expout_size = max (expout_size * 2, expout_ptr + lenelt + 10);
expout = (struct expression *)
xrealloc ((char *) expout, (sizeof (struct expression)
+ EXP_ELEM_TO_BYTES (expout_size)));
}
/* Write the leading length expression element (which advances the current
expression element index), then write the string constant followed by a
terminating null byte, and then write the trailing length expression
element. */
write_exp_elt_longcst ((LONGEST) len);
strdata = (char *) &expout->elts[expout_ptr];
memcpy (strdata, str.ptr, len);
*(strdata + len) = '\0';
expout_ptr += lenelt - 2;
write_exp_elt_longcst ((LONGEST) len);
}
/* Add a bitstring constant to the end of the expression.
Bitstring constants are stored by first writing an expression element
that contains the length of the bitstring (in bits), then stuffing the
bitstring constant itself into however many expression elements are
needed to hold it, and then writing another expression element that
contains the length of the bitstring. I.E. an expression element at
each end of the bitstring records the bitstring length, so you can skip
over the expression elements containing the actual bitstring bytes from
either end of the bitstring. */
void
write_exp_bitstring (str)
struct stoken str;
{
register int bits = str.length; /* length in bits */
register int len = (bits + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
register int lenelt;
register char *strdata;
/* Compute the number of expression elements required to hold the bitstring,
along with one expression element at each end to record the actual
bitstring length in bits. */
lenelt = 2 + BYTES_TO_EXP_ELEM (len);
/* Ensure that we have enough available expression elements to store
everything. */
if ((expout_ptr + lenelt) >= expout_size)
{
expout_size = max (expout_size * 2, expout_ptr + lenelt + 10);
expout = (struct expression *)
xrealloc ((char *) expout, (sizeof (struct expression)
+ EXP_ELEM_TO_BYTES (expout_size)));
}
/* Write the leading length expression element (which advances the current
expression element index), then write the bitstring constant, and then
write the trailing length expression element. */
write_exp_elt_longcst ((LONGEST) bits);
strdata = (char *) &expout->elts[expout_ptr];
memcpy (strdata, str.ptr, len);
expout_ptr += lenelt - 2;
write_exp_elt_longcst ((LONGEST) bits);
}
/* Add the appropriate elements for a minimal symbol to the end of
the expression. The rationale behind passing in text_symbol_type and
data_symbol_type was so that Modula-2 could pass in WORD for
data_symbol_type. Perhaps it still is useful to have those types vary
based on the language, but they no longer have names like "int", so
the initial rationale is gone. */
static struct type *msym_text_symbol_type;
static struct type *msym_data_symbol_type;
static struct type *msym_unknown_symbol_type;
void
write_exp_msymbol (msymbol, text_symbol_type, data_symbol_type)
struct minimal_symbol *msymbol;
struct type *text_symbol_type;
struct type *data_symbol_type;
{
CORE_ADDR addr;
write_exp_elt_opcode (OP_LONG);
write_exp_elt_type (lookup_pointer_type (builtin_type_void));
addr = SYMBOL_VALUE_ADDRESS (msymbol);
if (overlay_debugging)
addr = symbol_overlayed_address (addr, SYMBOL_BFD_SECTION (msymbol));
write_exp_elt_longcst ((LONGEST) addr);
1999-07-07 20:19:36 +00:00
write_exp_elt_opcode (OP_LONG);
write_exp_elt_opcode (UNOP_MEMVAL);
1999-07-07 20:19:36 +00:00
switch (msymbol->type)
{
case mst_text:
case mst_file_text:
case mst_solib_trampoline:
write_exp_elt_type (msym_text_symbol_type);
break;
case mst_data:
case mst_file_data:
case mst_bss:
case mst_file_bss:
write_exp_elt_type (msym_data_symbol_type);
break;
default:
write_exp_elt_type (msym_unknown_symbol_type);
break;
}
write_exp_elt_opcode (UNOP_MEMVAL);
}
/* Recognize tokens that start with '$'. These include:
1999-07-07 20:19:36 +00:00
$regname A native register name or a "standard
register name".
1999-07-07 20:19:36 +00:00
$variable A convenience variable with a name chosen
by the user.
1999-07-07 20:19:36 +00:00
$digits Value history with index <digits>, starting
from the first value which has index 1.
1999-07-07 20:19:36 +00:00
$$digits Value history with index <digits> relative
to the last value. I.E. $$0 is the last
value, $$1 is the one previous to that, $$2
is the one previous to $$1, etc.
1999-07-07 20:19:36 +00:00
$ | $0 | $$0 The last value in the value history.
1999-07-07 20:19:36 +00:00
$$ An abbreviation for the second to the last
value in the value history, I.E. $$1
1999-07-07 20:19:36 +00:00
*/
void
write_dollar_variable (str)
struct stoken str;
{
/* Handle the tokens $digits; also $ (short for $0) and $$ (short for $$1)
and $$digits (equivalent to $<-digits> if you could type that). */
int negate = 0;
int i = 1;
/* Double dollar means negate the number and add -1 as well.
Thus $$ alone means -1. */
if (str.length >= 2 && str.ptr[1] == '$')
{
negate = 1;
i = 2;
}
if (i == str.length)
{
/* Just dollars (one or two) */
1999-07-07 20:19:36 +00:00
i = -negate;
goto handle_last;
}
/* Is the rest of the token digits? */
for (; i < str.length; i++)
if (!(str.ptr[i] >= '0' && str.ptr[i] <= '9'))
break;
if (i == str.length)
{
i = atoi (str.ptr + 1 + negate);
if (negate)
1999-07-07 20:19:36 +00:00
i = -i;
goto handle_last;
}
1999-07-07 20:19:36 +00:00
/* Handle tokens that refer to machine registers:
$ followed by a register name. */
1999-07-07 20:19:36 +00:00
i = target_map_name_to_register (str.ptr + 1, str.length - 1);
if (i >= 0)
goto handle_register;
1999-10-12 04:37:53 +00:00
if (SYMBOLS_CAN_START_WITH_DOLLAR)
{
1999-10-12 04:37:53 +00:00
struct symbol *sym = NULL;
struct minimal_symbol *msym = NULL;
/* On HP-UX, certain system routines (millicode) have names beginning
with $ or $$, e.g. $$dyncall, which handles inter-space procedure
calls on PA-RISC. Check for those, first. */
/* This code is not enabled on non HP-UX systems, since worst case
symbol table lookup performance is awful, to put it mildly. */
sym = lookup_symbol (copy_name (str), (struct block *) NULL,
VAR_NAMESPACE, (int *) NULL, (struct symtab **) NULL);
if (sym)
{
write_exp_elt_opcode (OP_VAR_VALUE);
write_exp_elt_block (block_found); /* set by lookup_symbol */
write_exp_elt_sym (sym);
write_exp_elt_opcode (OP_VAR_VALUE);
return;
}
msym = lookup_minimal_symbol (copy_name (str), NULL, NULL);
if (msym)
{
write_exp_msymbol (msym,
lookup_function_type (builtin_type_int),
builtin_type_int);
return;
}
}
1999-07-07 20:19:36 +00:00
/* Any other names starting in $ are debugger internal variables. */
write_exp_elt_opcode (OP_INTERNALVAR);
write_exp_elt_intern (lookup_internalvar (copy_name (str) + 1));
1999-07-07 20:19:36 +00:00
write_exp_elt_opcode (OP_INTERNALVAR);
return;
1999-07-07 20:19:36 +00:00
handle_last:
write_exp_elt_opcode (OP_LAST);
write_exp_elt_longcst ((LONGEST) i);
write_exp_elt_opcode (OP_LAST);
return;
1999-07-07 20:19:36 +00:00
handle_register:
write_exp_elt_opcode (OP_REGISTER);
write_exp_elt_longcst (i);
1999-07-07 20:19:36 +00:00
write_exp_elt_opcode (OP_REGISTER);
return;
}
/* Parse a string that is possibly a namespace / nested class
specification, i.e., something of the form A::B::C::x. Input
(NAME) is the entire string; LEN is the current valid length; the
output is a string, TOKEN, which points to the largest recognized
prefix which is a series of namespaces or classes. CLASS_PREFIX is
another output, which records whether a nested class spec was
recognized (= 1) or a fully qualified variable name was found (=
0). ARGPTR is side-effected (if non-NULL) to point to beyond the
string recognized and consumed by this routine.
The return value is a pointer to the symbol for the base class or
variable if found, or NULL if not found. Callers must check this
first -- if NULL, the outputs may not be correct.
This function is used c-exp.y. This is used specifically to get
around HP aCC (and possibly other compilers), which insists on
generating names with embedded colons for namespace or nested class
members.
(Argument LEN is currently unused. 1997-08-27)
Callers must free memory allocated for the output string TOKEN. */
1999-07-07 20:19:36 +00:00
static const char coloncolon[2] =
{':', ':'};
struct symbol *
parse_nested_classes_for_hpacc (name, len, token, class_prefix, argptr)
1999-07-07 20:19:36 +00:00
char *name;
int len;
char **token;
int *class_prefix;
char **argptr;
{
1999-07-07 20:19:36 +00:00
/* Comment below comes from decode_line_1 which has very similar
code, which is called for "break" command parsing. */
/* We have what looks like a class or namespace
scope specification (A::B), possibly with many
levels of namespaces or classes (A::B::C::D).
Some versions of the HP ANSI C++ compiler (as also possibly
other compilers) generate class/function/member names with
embedded double-colons if they are inside namespaces. To
handle this, we loop a few times, considering larger and
larger prefixes of the string as though they were single
symbols. So, if the initially supplied string is
A::B::C::D::foo, we have to look up "A", then "A::B",
then "A::B::C", then "A::B::C::D", and finally
"A::B::C::D::foo" as single, monolithic symbols, because
A, B, C or D may be namespaces.
Note that namespaces can nest only inside other
namespaces, and not inside classes. So we need only
consider *prefixes* of the string; there is no need to look up
"B::C" separately as a symbol in the previous example. */
1999-07-07 20:19:36 +00:00
register char *p;
char *start, *end;
char *prefix = NULL;
char *tmp;
struct symbol *sym_class = NULL;
struct symbol *sym_var = NULL;
struct type *t;
int prefix_len = 0;
int done = 0;
1999-07-07 20:19:36 +00:00
char *q;
/* Check for HP-compiled executable -- in other cases
return NULL, and caller must default to standard GDB
behaviour. */
if (!hp_som_som_object_present)
return (struct symbol *) NULL;
p = name;
1999-07-07 20:19:36 +00:00
/* Skip over whitespace and possible global "::" */
while (*p && (*p == ' ' || *p == '\t'))
p++;
if (p[0] == ':' && p[1] == ':')
p += 2;
1999-07-07 20:19:36 +00:00
while (*p && (*p == ' ' || *p == '\t'))
p++;
while (1)
{
/* Get to the end of the next namespace or class spec. */
/* If we're looking at some non-token, fail immediately */
start = p;
if (!(isalpha (*p) || *p == '$' || *p == '_'))
1999-07-07 20:19:36 +00:00
return (struct symbol *) NULL;
p++;
1999-07-07 20:19:36 +00:00
while (*p && (isalnum (*p) || *p == '$' || *p == '_'))
p++;
if (*p == '<')
{
/* If we have the start of a template specification,
scan right ahead to its end */
q = find_template_name_end (p);
if (q)
p = q;
}
end = p;
1999-07-07 20:19:36 +00:00
/* Skip over "::" and whitespace for next time around */
while (*p && (*p == ' ' || *p == '\t'))
p++;
if (p[0] == ':' && p[1] == ':')
1999-07-07 20:19:36 +00:00
p += 2;
while (*p && (*p == ' ' || *p == '\t'))
p++;
1999-07-07 20:19:36 +00:00
/* Done with tokens? */
if (!*p || !(isalpha (*p) || *p == '$' || *p == '_'))
1999-07-07 20:19:36 +00:00
done = 1;
tmp = (char *) alloca (prefix_len + end - start + 3);
if (prefix)
1999-07-07 20:19:36 +00:00
{
memcpy (tmp, prefix, prefix_len);
memcpy (tmp + prefix_len, coloncolon, 2);
memcpy (tmp + prefix_len + 2, start, end - start);
tmp[prefix_len + 2 + end - start] = '\000';
}
else
1999-07-07 20:19:36 +00:00
{
memcpy (tmp, start, end - start);
tmp[end - start] = '\000';
}
prefix = tmp;
prefix_len = strlen (prefix);
1999-07-07 20:19:36 +00:00
/* See if the prefix we have now is something we know about */
1999-07-07 20:19:36 +00:00
if (!done)
{
/* More tokens to process, so this must be a class/namespace */
sym_class = lookup_symbol (prefix, 0, STRUCT_NAMESPACE,
0, (struct symtab **) NULL);
}
else
1999-07-07 20:19:36 +00:00
{
/* No more tokens, so try as a variable first */
sym_var = lookup_symbol (prefix, 0, VAR_NAMESPACE,
0, (struct symtab **) NULL);
/* If failed, try as class/namespace */
if (!sym_var)
sym_class = lookup_symbol (prefix, 0, STRUCT_NAMESPACE,
0, (struct symtab **) NULL);
}
if (sym_var ||
1999-07-07 20:19:36 +00:00
(sym_class &&
(t = check_typedef (SYMBOL_TYPE (sym_class)),
(TYPE_CODE (t) == TYPE_CODE_STRUCT
|| TYPE_CODE (t) == TYPE_CODE_UNION))))
{
/* We found a valid token */
*token = (char *) xmalloc (prefix_len + 1);
memcpy (*token, prefix, prefix_len);
(*token)[prefix_len] = '\000';
break;
}
/* No variable or class/namespace found, no more tokens */
if (done)
1999-07-07 20:19:36 +00:00
return (struct symbol *) NULL;
}
/* Out of loop, so we must have found a valid token */
if (sym_var)
*class_prefix = 0;
else
*class_prefix = 1;
if (argptr)
*argptr = done ? p : end;
1999-07-07 20:19:36 +00:00
return sym_var ? sym_var : sym_class; /* found */
}
char *
find_template_name_end (p)
1999-07-07 20:19:36 +00:00
char *p;
{
int depth = 1;
int just_seen_right = 0;
int just_seen_colon = 0;
int just_seen_space = 0;
1999-07-07 20:19:36 +00:00
if (!p || (*p != '<'))
return 0;
while (*++p)
{
switch (*p)
1999-07-07 20:19:36 +00:00
{
case '\'':
case '\"':
case '{':
case '}':
/* In future, may want to allow these?? */
return 0;
case '<':
depth++; /* start nested template */
if (just_seen_colon || just_seen_right || just_seen_space)
return 0; /* but not after : or :: or > or space */
break;
case '>':
if (just_seen_colon || just_seen_right)
return 0; /* end a (nested?) template */
just_seen_right = 1; /* but not after : or :: */
if (--depth == 0) /* also disallow >>, insist on > > */
return ++p; /* if outermost ended, return */
break;
case ':':
if (just_seen_space || (just_seen_colon > 1))
return 0; /* nested class spec coming up */
just_seen_colon++; /* we allow :: but not :::: */
break;
case ' ':
break;
default:
if (!((*p >= 'a' && *p <= 'z') || /* allow token chars */
(*p >= 'A' && *p <= 'Z') ||
(*p >= '0' && *p <= '9') ||
(*p == '_') || (*p == ',') || /* commas for template args */
(*p == '&') || (*p == '*') || /* pointer and ref types */
(*p == '(') || (*p == ')') || /* function types */
(*p == '[') || (*p == ']'))) /* array types */
return 0;
}
if (*p != ' ')
1999-07-07 20:19:36 +00:00
just_seen_space = 0;
if (*p != ':')
1999-07-07 20:19:36 +00:00
just_seen_colon = 0;
if (*p != '>')
1999-07-07 20:19:36 +00:00
just_seen_right = 0;
}
return 0;
}
1999-07-07 20:19:36 +00:00
/* Return a null-terminated temporary copy of the name
of a string token. */
char *
copy_name (token)
struct stoken token;
{
memcpy (namecopy, token.ptr, token.length);
namecopy[token.length] = 0;
return namecopy;
}
/* Reverse an expression from suffix form (in which it is constructed)
to prefix form (in which we can conveniently print or execute it). */
static void
prefixify_expression (expr)
register struct expression *expr;
{
register int len =
1999-07-07 20:19:36 +00:00
sizeof (struct expression) + EXP_ELEM_TO_BYTES (expr->nelts);
register struct expression *temp;
register int inpos = expr->nelts, outpos = 0;
temp = (struct expression *) alloca (len);
/* Copy the original expression into temp. */
memcpy (temp, expr, len);
prefixify_subexp (temp, expr, inpos, outpos);
}
/* Return the number of exp_elements in the subexpression of EXPR
whose last exp_element is at index ENDPOS - 1 in EXPR. */
int
length_of_subexp (expr, endpos)
register struct expression *expr;
register int endpos;
{
register int oplen = 1;
register int args = 0;
register int i;
if (endpos < 1)
error ("?error in length_of_subexp");
i = (int) expr->elts[endpos - 1].opcode;
switch (i)
{
/* C++ */
case OP_SCOPE:
oplen = longest_to_int (expr->elts[endpos - 2].longconst);
oplen = 5 + BYTES_TO_EXP_ELEM (oplen + 1);
break;
case OP_LONG:
case OP_DOUBLE:
case OP_VAR_VALUE:
oplen = 4;
break;
case OP_TYPE:
case OP_BOOL:
case OP_LAST:
case OP_REGISTER:
case OP_INTERNALVAR:
oplen = 3;
break;
case OP_COMPLEX:
1999-07-07 20:19:36 +00:00
oplen = 1;
args = 2;
1999-07-07 20:19:36 +00:00
break;
case OP_FUNCALL:
case OP_F77_UNDETERMINED_ARGLIST:
oplen = 3;
args = 1 + longest_to_int (expr->elts[endpos - 2].longconst);
break;
case UNOP_MAX:
case UNOP_MIN:
oplen = 3;
break;
1999-07-07 20:19:36 +00:00
case BINOP_VAL:
case UNOP_CAST:
case UNOP_MEMVAL:
oplen = 3;
args = 1;
break;
case UNOP_ABS:
case UNOP_CAP:
case UNOP_CHR:
case UNOP_FLOAT:
case UNOP_HIGH:
case UNOP_ODD:
case UNOP_ORD:
case UNOP_TRUNC:
oplen = 1;
args = 1;
break;
case OP_LABELED:
case STRUCTOP_STRUCT:
case STRUCTOP_PTR:
args = 1;
/* fall through */
case OP_M2_STRING:
case OP_STRING:
case OP_NAME:
case OP_EXPRSTRING:
oplen = longest_to_int (expr->elts[endpos - 2].longconst);
oplen = 4 + BYTES_TO_EXP_ELEM (oplen + 1);
break;
case OP_BITSTRING:
oplen = longest_to_int (expr->elts[endpos - 2].longconst);
oplen = (oplen + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
oplen = 4 + BYTES_TO_EXP_ELEM (oplen);
break;
case OP_ARRAY:
oplen = 4;
args = longest_to_int (expr->elts[endpos - 2].longconst);
args -= longest_to_int (expr->elts[endpos - 3].longconst);
args += 1;
break;
case TERNOP_COND:
case TERNOP_SLICE:
case TERNOP_SLICE_COUNT:
args = 3;
break;
/* Modula-2 */
1999-07-07 20:19:36 +00:00
case MULTI_SUBSCRIPT:
oplen = 3;
1999-07-07 20:19:36 +00:00
args = 1 + longest_to_int (expr->elts[endpos - 2].longconst);
break;
case BINOP_ASSIGN_MODIFY:
oplen = 3;
args = 2;
break;
/* C++ */
case OP_THIS:
oplen = 2;
break;
default:
args = 1 + (i < (int) BINOP_END);
}
while (args > 0)
{
oplen += length_of_subexp (expr, endpos - oplen);
args--;
}
return oplen;
}
/* Copy the subexpression ending just before index INEND in INEXPR
into OUTEXPR, starting at index OUTBEG.
In the process, convert it from suffix to prefix form. */
static void
prefixify_subexp (inexpr, outexpr, inend, outbeg)
register struct expression *inexpr;
struct expression *outexpr;
register int inend;
int outbeg;
{
register int oplen = 1;
register int args = 0;
register int i;
int *arglens;
enum exp_opcode opcode;
/* Compute how long the last operation is (in OPLEN),
and also how many preceding subexpressions serve as
arguments for it (in ARGS). */
opcode = inexpr->elts[inend - 1].opcode;
switch (opcode)
{
/* C++ */
case OP_SCOPE:
oplen = longest_to_int (inexpr->elts[inend - 2].longconst);
oplen = 5 + BYTES_TO_EXP_ELEM (oplen + 1);
break;
case OP_LONG:
case OP_DOUBLE:
case OP_VAR_VALUE:
oplen = 4;
break;
case OP_TYPE:
case OP_BOOL:
case OP_LAST:
case OP_REGISTER:
case OP_INTERNALVAR:
oplen = 3;
break;
case OP_COMPLEX:
1999-07-07 20:19:36 +00:00
oplen = 1;
args = 2;
break;
case OP_FUNCALL:
case OP_F77_UNDETERMINED_ARGLIST:
oplen = 3;
args = 1 + longest_to_int (inexpr->elts[inend - 2].longconst);
break;
case UNOP_MIN:
case UNOP_MAX:
oplen = 3;
break;
case UNOP_CAST:
case UNOP_MEMVAL:
oplen = 3;
args = 1;
break;
case UNOP_ABS:
case UNOP_CAP:
case UNOP_CHR:
case UNOP_FLOAT:
case UNOP_HIGH:
case UNOP_ODD:
case UNOP_ORD:
case UNOP_TRUNC:
1999-07-07 20:19:36 +00:00
oplen = 1;
args = 1;
break;
case STRUCTOP_STRUCT:
case STRUCTOP_PTR:
case OP_LABELED:
args = 1;
/* fall through */
case OP_M2_STRING:
case OP_STRING:
case OP_NAME:
case OP_EXPRSTRING:
oplen = longest_to_int (inexpr->elts[inend - 2].longconst);
oplen = 4 + BYTES_TO_EXP_ELEM (oplen + 1);
break;
case OP_BITSTRING:
oplen = longest_to_int (inexpr->elts[inend - 2].longconst);
oplen = (oplen + HOST_CHAR_BIT - 1) / HOST_CHAR_BIT;
oplen = 4 + BYTES_TO_EXP_ELEM (oplen);
break;
case OP_ARRAY:
oplen = 4;
args = longest_to_int (inexpr->elts[inend - 2].longconst);
args -= longest_to_int (inexpr->elts[inend - 3].longconst);
args += 1;
break;
case TERNOP_COND:
case TERNOP_SLICE:
case TERNOP_SLICE_COUNT:
args = 3;
break;
case BINOP_ASSIGN_MODIFY:
oplen = 3;
args = 2;
break;
/* Modula-2 */
1999-07-07 20:19:36 +00:00
case MULTI_SUBSCRIPT:
oplen = 3;
args = 1 + longest_to_int (inexpr->elts[inend - 2].longconst);
break;
/* C++ */
case OP_THIS:
oplen = 2;
break;
default:
args = 1 + ((int) opcode < (int) BINOP_END);
}
/* Copy the final operator itself, from the end of the input
to the beginning of the output. */
inend -= oplen;
memcpy (&outexpr->elts[outbeg], &inexpr->elts[inend],
EXP_ELEM_TO_BYTES (oplen));
outbeg += oplen;
/* Find the lengths of the arg subexpressions. */
arglens = (int *) alloca (args * sizeof (int));
for (i = args - 1; i >= 0; i--)
{
oplen = length_of_subexp (inexpr, inend);
arglens[i] = oplen;
inend -= oplen;
}
/* Now copy each subexpression, preserving the order of
the subexpressions, but prefixifying each one.
In this loop, inend starts at the beginning of
the expression this level is working on
and marches forward over the arguments.
outbeg does similarly in the output. */
for (i = 0; i < args; i++)
{
oplen = arglens[i];
inend += oplen;
prefixify_subexp (inexpr, outexpr, inend, outbeg);
outbeg += oplen;
}
}
/* This page contains the two entry points to this file. */
/* Read an expression from the string *STRINGPTR points to,
parse it, and return a pointer to a struct expression that we malloc.
Use block BLOCK as the lexical context for variable names;
if BLOCK is zero, use the block of the selected stack frame.
Meanwhile, advance *STRINGPTR to point after the expression,
at the first nonwhite character that is not part of the expression
(possibly a null character).
If COMMA is nonzero, stop if a comma is reached. */
struct expression *
parse_exp_1 (stringptr, block, comma)
char **stringptr;
struct block *block;
int comma;
{
struct cleanup *old_chain;
lexptr = *stringptr;
paren_depth = 0;
type_stack_depth = 0;
comma_terminates = comma;
if (lexptr == 0 || *lexptr == 0)
error_no_arg ("expression to compute");
2000-05-22 09:02:23 +00:00
old_chain = make_cleanup (free_funcalls, 0 /*ignore*/);
funcall_chain = 0;
expression_context_block = block ? block : get_selected_block ();
namecopy = (char *) alloca (strlen (lexptr) + 1);
expout_size = 10;
expout_ptr = 0;
expout = (struct expression *)
xmalloc (sizeof (struct expression) + EXP_ELEM_TO_BYTES (expout_size));
expout->language_defn = current_language;
make_cleanup (free_current_contents, &expout);
if (current_language->la_parser ())
current_language->la_error (NULL);
discard_cleanups (old_chain);
/* Record the actual number of expression elements, and then
reallocate the expression memory so that we free up any
excess elements. */
expout->nelts = expout_ptr;
expout = (struct expression *)
xrealloc ((char *) expout,
sizeof (struct expression) + EXP_ELEM_TO_BYTES (expout_ptr));;
/* Convert expression from postfix form as generated by yacc
parser, to a prefix form. */
if (expressiondebug)
1999-07-07 17:31:57 +00:00
dump_prefix_expression (expout, gdb_stdlog,
"before conversion to prefix form");
prefixify_expression (expout);
if (expressiondebug)
1999-07-07 17:31:57 +00:00
dump_postfix_expression (expout, gdb_stdlog,
"after conversion to prefix form");
*stringptr = lexptr;
return expout;
}
/* Parse STRING as an expression, and complain if this fails
to use up all of the contents of STRING. */
struct expression *
parse_expression (string)
char *string;
{
register struct expression *exp;
exp = parse_exp_1 (&string, 0, 0);
if (*string)
error ("Junk after end of expression.");
return exp;
}
/* Stuff for maintaining a stack of types. Currently just used by C, but
probably useful for any language which declares its types "backwards". */
1999-07-07 20:19:36 +00:00
void
push_type (tp)
enum type_pieces tp;
{
if (type_stack_depth == type_stack_size)
{
type_stack_size *= 2;
type_stack = (union type_stack_elt *)
xrealloc ((char *) type_stack, type_stack_size * sizeof (*type_stack));
}
type_stack[type_stack_depth++].piece = tp;
}
void
push_type_int (n)
int n;
{
if (type_stack_depth == type_stack_size)
{
type_stack_size *= 2;
type_stack = (union type_stack_elt *)
xrealloc ((char *) type_stack, type_stack_size * sizeof (*type_stack));
}
type_stack[type_stack_depth++].int_val = n;
}
1999-07-07 20:19:36 +00:00
enum type_pieces
pop_type ()
{
if (type_stack_depth)
return type_stack[--type_stack_depth].piece;
return tp_end;
}
int
pop_type_int ()
{
if (type_stack_depth)
return type_stack[--type_stack_depth].int_val;
/* "Can't happen". */
return 0;
}
/* Pop the type stack and return the type which corresponds to FOLLOW_TYPE
as modified by all the stuff on the stack. */
struct type *
follow_types (follow_type)
struct type *follow_type;
{
int done = 0;
int array_size;
struct type *range_type;
while (!done)
switch (pop_type ())
{
case tp_end:
done = 1;
break;
case tp_pointer:
follow_type = lookup_pointer_type (follow_type);
break;
case tp_reference:
follow_type = lookup_reference_type (follow_type);
break;
case tp_array:
array_size = pop_type_int ();
/* FIXME-type-allocation: need a way to free this type when we are
done with it. */
range_type =
create_range_type ((struct type *) NULL,
builtin_type_int, 0,
array_size >= 0 ? array_size - 1 : 0);
follow_type =
create_array_type ((struct type *) NULL,
follow_type, range_type);
if (array_size < 0)
1999-07-07 20:19:36 +00:00
TYPE_ARRAY_UPPER_BOUND_TYPE (follow_type)
= BOUND_CANNOT_BE_DETERMINED;
break;
case tp_function:
/* FIXME-type-allocation: need a way to free this type when we are
done with it. */
follow_type = lookup_function_type (follow_type);
break;
}
return follow_type;
}
2000-05-28 01:12:42 +00:00
static void build_parse (void);
1999-06-01 15:44:41 +00:00
static void
build_parse ()
{
1999-06-07 19:19:32 +00:00
int i;
msym_text_symbol_type =
init_type (TYPE_CODE_FUNC, 1, 0, "<text variable, no debug info>", NULL);
TYPE_TARGET_TYPE (msym_text_symbol_type) = builtin_type_int;
msym_data_symbol_type =
init_type (TYPE_CODE_INT, TARGET_INT_BIT / HOST_CHAR_BIT, 0,
"<data variable, no debug info>", NULL);
msym_unknown_symbol_type =
init_type (TYPE_CODE_INT, 1, 0,
"<variable (not text or data), no debug info>",
NULL);
1999-06-07 19:19:32 +00:00
/* create the std_regs table */
num_std_regs = 0;
#ifdef PC_REGNUM
if (PC_REGNUM >= 0)
num_std_regs++;
#endif
#ifdef FP_REGNUM
if (FP_REGNUM >= 0)
num_std_regs++;
#endif
1999-07-19 23:30:11 +00:00
#ifdef SP_REGNUM
1999-06-07 19:19:32 +00:00
if (SP_REGNUM >= 0)
num_std_regs++;
#endif
#ifdef PS_REGNUM
if (PS_REGNUM >= 0)
num_std_regs++;
#endif
/* create an empty table */
std_regs = xmalloc ((num_std_regs + 1) * sizeof *std_regs);
i = 0;
/* fill it in */
#ifdef PC_REGNUM
std_regs[i].name = "pc";
std_regs[i].regnum = PC_REGNUM;
i++;
#endif
#ifdef FP_REGNUM
std_regs[i].name = "fp";
std_regs[i].regnum = FP_REGNUM;
i++;
#endif
#ifdef SP_REGNUM
std_regs[i].name = "sp";
std_regs[i].regnum = SP_REGNUM;
i++;
#endif
#ifdef PS_REGNUM
std_regs[i].name = "ps";
std_regs[i].regnum = PS_REGNUM;
i++;
#endif
memset (&std_regs[i], 0, sizeof (std_regs[i]));
1999-06-01 15:44:41 +00:00
}
void
_initialize_parse ()
{
type_stack_size = 80;
type_stack_depth = 0;
type_stack = (union type_stack_elt *)
xmalloc (type_stack_size * sizeof (*type_stack));
build_parse ();
1999-06-14 18:08:47 +00:00
/* FIXME - For the moment, handle types by swapping them in and out.
Should be using the per-architecture data-pointer and a large
struct. */
register_gdbarch_swap (&msym_text_symbol_type, sizeof (msym_text_symbol_type), NULL);
register_gdbarch_swap (&msym_data_symbol_type, sizeof (msym_data_symbol_type), NULL);
register_gdbarch_swap (&msym_unknown_symbol_type, sizeof (msym_unknown_symbol_type), NULL);
register_gdbarch_swap (&num_std_regs, sizeof (std_regs), NULL);
register_gdbarch_swap (&std_regs, sizeof (std_regs), NULL);
register_gdbarch_swap (NULL, 0, build_parse);
add_show_from_set (
2000-03-28 02:25:14 +00:00
add_set_cmd ("expression", class_maintenance, var_zinteger,
1999-07-07 20:19:36 +00:00
(char *) &expressiondebug,
"Set expression debugging.\n\
When non-zero, the internal representation of expressions will be printed.",
2000-03-28 02:25:14 +00:00
&setdebuglist),
&showdebuglist);
}