binutils-gdb/gdb/rust-exp.y

2813 lines
68 KiB
Plaintext
Raw Normal View History

/* Bison parser for Rust expressions, for GDB.
Copyright (C) 2016-2018 Free Software Foundation, Inc.
This file is part of GDB.
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 3 of the License, or
(at your option) any later version.
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.
You should have received a copy of the GNU General Public License
along with this program. If not, see <http://www.gnu.org/licenses/>. */
/* Removing the last conflict seems difficult. */
%expect 1
%{
#include "defs.h"
#include "block.h"
#include "charset.h"
#include "cp-support.h"
#include "gdb_obstack.h"
#include "gdb_regex.h"
#include "rust-lang.h"
#include "parser-defs.h"
#include "selftest.h"
#include "value.h"
#include "vec.h"
#define GDB_YY_REMAP_PREFIX rust
#include "yy-remap.h"
#define RUSTSTYPE YYSTYPE
struct rust_op;
typedef std::vector<const struct rust_op *> rust_op_vector;
/* A typed integer constant. */
struct typed_val_int
{
LONGEST val;
struct type *type;
};
/* A typed floating point constant. */
struct typed_val_float
{
Target FP: Use target format throughout expression parsing When parsing floating-point literals, the language parsers currently use parse_float or some equivalent routine to parse the input string into a DOUBLEST, which is then stored within a OP_DOUBLE expression node. When evaluating the expression, the OP_DOUBLE is finally converted into a value in target format. On the other hand, *decimal* floating-point literals are parsed directly into target format and stored that way in a OP_DECFLOAT expression node. In order to eliminate the DOUBLEST, this patch therefore unifies the handling of binary and decimal floating- point literals and stores them both in target format within a new OP_FLOAT expression node, replacing both OP_DOUBLE and OP_DECFLOAT. In order to store literals in target format, the parse_float routine needs to know the type of the literal. All parsers therefore need to be changed to determine the appropriate type (e.g. by detecting suffixes) *before* calling parse_float, instead of after it as today. However, this change is mostly straightforward -- again, this is already done for decimal FP today. The core of the literal parsing is moved into a new routine floatformat_from_string, mirroring floatformat_to_string. The parse_float routine now calls either floatformat_from_string or decimal_from_sting, allowing it to handle any type of FP literal. All language parsers need to be updated. Some notes on specific changes to the various languages: - C: Decimal FP is now handled in parse_float, and no longer needs to be handled specially. - D: Straightforward. - Fortran: Still used a hard-coded "atof", also replaced by parse_float now. Continues to always use builtin_real_s8 as the type of literal, even though this is probably wrong. - Go: This used to handle "f" and "l" suffixes, even though the Go language actually doesn't support those. I kept this support for now -- maybe revisit later. Note the the GDB test suite for some reason actually *verifies* that GDB supports those unsupported suffixes ... - Pascal: Likewise -- this handles suffixes that are not supported in the language standard. - Modula-2: Like Fortran, used to use "atof". - Rust: Mostly straightforward, except for a unit-testing hitch. The code use to set a special "unit_testing" flag which would cause "rust_type" to always return NULL. This makes it not possible to encode a literal into target format (which type?). The reason for this flag appears to have been that during unit testing, there is no "rust_parser" context set up, which means no "gdbarch" is available to use its types. To fix this, I removed the unit_testing flag, and instead simply just set up a dummy rust_parser context during unit testing. - Ada: This used to check sizeof (DOUBLEST) to determine which type to use for floating-point literal. This seems questionable to begin with (since DOUBLEST is quite unrelated to target formats), and in any case we need to get rid of DOUBLEST. I'm now simply always using the largest type (builtin_long_double). gdb/ChangeLog: 2017-10-25 Ulrich Weigand <uweigand@de.ibm.com> * doublest.c (floatformat_from_string): New function. * doublest.h (floatformat_from_string): Add prototype. * std-operator.def (OP_DOUBLE, OP_DECFLOAT): Remove, replace by ... (OP_FLOAT): ... this. * expression.h: Do not include "doublest.h". (union exp_element): Replace doubleconst and decfloatconst by new element floatconst. * ada-lang.c (resolve_subexp): Handle OP_FLOAT instead of OP_DOUBLE. (ada_evaluate_subexp): Likewise. * eval.c (evaluate_subexp_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. * expprint.c (print_subexp_standard): Likewise. (dump_subexp_body_standard): Likewise. * breakpoint.c (watchpoint_exp_is_const): Likewise. * parse.c: Include "dfp.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): New function. (operator_length_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. (operator_check_standard): Likewise. (parse_float): Do not accept suffix. Take type as input. Return bool. Return target format buffer instead of host DOUBLEST. Use floatformat_from_string and decimal_from_string to parse either binary or decimal floating-point types. (parse_c_float): Remove. * parser-defs.h: Do not include "doublest.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): Add prototype. (parse_float): Update prototype. (parse_c_float): Remove. * c-exp.y: Do not include "dfp.h". (typed_val_float): Use byte buffer instead of DOUBLEST. (typed_val_decfloat): Remove. (DECFLOAT): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. Handle decimal and binary FP types the same way. * d-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT_LITERAL): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * f-exp.y: Replace dval by typed_val_float. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Use parse_float instead of atof. * go-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (parse_go_float): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of parse_go_float. Parse suffixes and determine type before calling parse_float. * p-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * m2-exp.y: Replace dval by byte buffer val. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of atof. * rust-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (lex_number): Call parse_float instead of strtod. (ast_dliteral): Use OP_FLOAT instead of OP_DOUBLE. (convert_ast_to_expression): Handle OP_FLOAT instead of OP_DOUBLE. Use write_exp_elt_floatcst. (unit_testing): Remove static variable. (rust_type): Do not check unit_testing. (rust_lex_tests): Do not set uint_testing. Set up dummy rust_parser. * ada-exp.y (type_float, type_double): Remove. (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. * ada-lex.l (processReal): Use parse_float instead of sscanf.
2017-10-25 15:32:23 +02:00
gdb_byte val[16];
struct type *type;
};
/* An identifier and an expression. This is used to represent one
element of a struct initializer. */
struct set_field
{
struct stoken name;
const struct rust_op *init;
};
typedef std::vector<set_field> rust_set_vector;
static int rustyylex (void);
static void rust_push_back (char c);
static const char *rust_copy_name (const char *, int);
static struct stoken rust_concat3 (const char *, const char *, const char *);
static struct stoken make_stoken (const char *);
static struct block_symbol rust_lookup_symbol (const char *name,
const struct block *block,
const domain_enum domain);
static struct type *rust_lookup_type (const char *name,
const struct block *block);
static struct type *rust_type (const char *name);
static const struct rust_op *crate_name (const struct rust_op *name);
static const struct rust_op *super_name (const struct rust_op *name,
unsigned int n_supers);
static const struct rust_op *ast_operation (enum exp_opcode opcode,
const struct rust_op *left,
const struct rust_op *right);
static const struct rust_op *ast_compound_assignment
(enum exp_opcode opcode, const struct rust_op *left,
const struct rust_op *rust_op);
static const struct rust_op *ast_literal (struct typed_val_int val);
static const struct rust_op *ast_dliteral (struct typed_val_float val);
static const struct rust_op *ast_structop (const struct rust_op *left,
const char *name,
int completing);
static const struct rust_op *ast_structop_anonymous
(const struct rust_op *left, struct typed_val_int number);
static const struct rust_op *ast_unary (enum exp_opcode opcode,
const struct rust_op *expr);
static const struct rust_op *ast_cast (const struct rust_op *expr,
const struct rust_op *type);
static const struct rust_op *ast_call_ish (enum exp_opcode opcode,
const struct rust_op *expr,
rust_op_vector *params);
static const struct rust_op *ast_path (struct stoken name,
rust_op_vector *params);
static const struct rust_op *ast_string (struct stoken str);
static const struct rust_op *ast_struct (const struct rust_op *name,
rust_set_vector *fields);
static const struct rust_op *ast_range (const struct rust_op *lhs,
const struct rust_op *rhs,
bool inclusive);
static const struct rust_op *ast_array_type (const struct rust_op *lhs,
struct typed_val_int val);
static const struct rust_op *ast_slice_type (const struct rust_op *type);
static const struct rust_op *ast_reference_type (const struct rust_op *type);
static const struct rust_op *ast_pointer_type (const struct rust_op *type,
int is_mut);
static const struct rust_op *ast_function_type (const struct rust_op *result,
rust_op_vector *params);
static const struct rust_op *ast_tuple_type (rust_op_vector *params);
/* The current rust parser. */
struct rust_parser;
static rust_parser *current_parser;
/* A regular expression for matching Rust numbers. This is split up
since it is very long and this gives us a way to comment the
sections. */
static const char *number_regex_text =
/* subexpression 1: allows use of alternation, otherwise uninteresting */
"^("
/* First comes floating point. */
/* Recognize number after the decimal point, with optional
exponent and optional type suffix.
subexpression 2: allows "?", otherwise uninteresting
subexpression 3: if present, type suffix
*/
"[0-9][0-9_]*\\.[0-9][0-9_]*([eE][-+]?[0-9][0-9_]*)?(f32|f64)?"
#define FLOAT_TYPE1 3
"|"
/* Recognize exponent without decimal point, with optional type
suffix.
subexpression 4: if present, type suffix
*/
#define FLOAT_TYPE2 4
"[0-9][0-9_]*[eE][-+]?[0-9][0-9_]*(f32|f64)?"
"|"
/* "23." is a valid floating point number, but "23.e5" and
"23.f32" are not. So, handle the trailing-. case
separately. */
"[0-9][0-9_]*\\."
"|"
/* Finally come integers.
subexpression 5: text of integer
subexpression 6: if present, type suffix
subexpression 7: allows use of alternation, otherwise uninteresting
*/
#define INT_TEXT 5
#define INT_TYPE 6
"(0x[a-fA-F0-9_]+|0o[0-7_]+|0b[01_]+|[0-9][0-9_]*)"
"([iu](size|8|16|32|64))?"
")";
/* The number of subexpressions to allocate space for, including the
"0th" whole match subexpression. */
#define NUM_SUBEXPRESSIONS 8
/* The compiled number-matching regex. */
static regex_t number_regex;
/* Obstack for data temporarily allocated during parsing. Points to
the obstack in the rust_parser, or to a temporary obstack during
unit testing. */
static auto_obstack *work_obstack;
/* An instance of this is created before parsing, and destroyed when
parsing is finished. */
struct rust_parser
{
rust_parser (struct parser_state *state)
: rust_ast (nullptr),
pstate (state)
{
gdb_assert (current_parser == nullptr);
current_parser = this;
work_obstack = &obstack;
}
~rust_parser ()
{
/* Clean up the globals we set. */
current_parser = nullptr;
work_obstack = nullptr;
}
/* Create a new rust_set_vector. The storage for the new vector is
managed by this class. */
rust_set_vector *new_set_vector ()
{
rust_set_vector *result = new rust_set_vector;
set_vectors.push_back (std::unique_ptr<rust_set_vector> (result));
return result;
}
/* Create a new rust_ops_vector. The storage for the new vector is
managed by this class. */
rust_op_vector *new_op_vector ()
{
rust_op_vector *result = new rust_op_vector;
op_vectors.push_back (std::unique_ptr<rust_op_vector> (result));
return result;
}
/* Return the parser's language. */
const struct language_defn *language () const
{
return parse_language (pstate);
}
/* Return the parser's gdbarch. */
struct gdbarch *arch () const
{
return parse_gdbarch (pstate);
}
/* A pointer to this is installed globally. */
auto_obstack obstack;
/* Result of parsing. Points into obstack. */
const struct rust_op *rust_ast;
/* This keeps track of the various vectors we allocate. */
std::vector<std::unique_ptr<rust_set_vector>> set_vectors;
std::vector<std::unique_ptr<rust_op_vector>> op_vectors;
/* The parser state gdb gave us. */
struct parser_state *pstate;
};
%}
%union
{
/* A typed integer constant. */
struct typed_val_int typed_val_int;
/* A typed floating point constant. */
struct typed_val_float typed_val_float;
/* An identifier or string. */
struct stoken sval;
/* A token representing an opcode, like "==". */
enum exp_opcode opcode;
/* A list of expressions; for example, the arguments to a function
call. */
rust_op_vector *params;
/* A list of field initializers. */
rust_set_vector *field_inits;
/* A single field initializer. */
struct set_field one_field_init;
/* An expression. */
const struct rust_op *op;
/* A plain integer, for example used to count the number of
"super::" prefixes on a path. */
unsigned int depth;
}
%{
/* Rust AST operations. We build a tree of these; then lower them
to gdb expressions when parsing has completed. */
struct rust_op
{
/* The opcode. */
enum exp_opcode opcode;
/* If OPCODE is OP_TYPE, then this holds information about what type
is described by this node. */
enum type_code typecode;
/* Indicates whether OPCODE actually represents a compound
assignment. For example, if OPCODE is GTGT and this is false,
then this rust_op represents an ordinary ">>"; but if this is
true, then this rust_op represents ">>=". Unused in other
cases. */
unsigned int compound_assignment : 1;
/* Only used by a field expression; if set, indicates that the field
name occurred at the end of the expression and is eligible for
completion. */
unsigned int completing : 1;
/* For OP_RANGE, indicates whether the range is inclusive or
exclusive. */
unsigned int inclusive : 1;
/* Operands of expression. Which one is used and how depends on the
particular opcode. */
RUSTSTYPE left;
RUSTSTYPE right;
};
%}
%token <sval> GDBVAR
%token <sval> IDENT
%token <sval> COMPLETE
%token <typed_val_int> INTEGER
%token <typed_val_int> DECIMAL_INTEGER
%token <sval> STRING
%token <sval> BYTESTRING
%token <typed_val_float> FLOAT
%token <opcode> COMPOUND_ASSIGN
/* Keyword tokens. */
%token <voidval> KW_AS
%token <voidval> KW_IF
%token <voidval> KW_TRUE
%token <voidval> KW_FALSE
%token <voidval> KW_SUPER
%token <voidval> KW_SELF
%token <voidval> KW_MUT
%token <voidval> KW_EXTERN
%token <voidval> KW_CONST
%token <voidval> KW_FN
%token <voidval> KW_SIZEOF
/* Operator tokens. */
%token <voidval> DOTDOT
%token <voidval> DOTDOTEQ
%token <voidval> OROR
%token <voidval> ANDAND
%token <voidval> EQEQ
%token <voidval> NOTEQ
%token <voidval> LTEQ
%token <voidval> GTEQ
%token <voidval> LSH RSH
%token <voidval> COLONCOLON
%token <voidval> ARROW
%type <op> type
%type <op> path_for_expr
%type <op> identifier_path_for_expr
%type <op> path_for_type
%type <op> identifier_path_for_type
%type <op> just_identifiers_for_type
%type <params> maybe_type_list
%type <params> type_list
%type <depth> super_path
%type <op> literal
%type <op> expr
%type <op> field_expr
%type <op> idx_expr
%type <op> unop_expr
%type <op> binop_expr
%type <op> binop_expr_expr
%type <op> type_cast_expr
%type <op> assignment_expr
%type <op> compound_assignment_expr
%type <op> paren_expr
%type <op> call_expr
%type <op> path_expr
%type <op> tuple_expr
%type <op> unit_expr
%type <op> struct_expr
%type <op> array_expr
%type <op> range_expr
%type <params> expr_list
%type <params> maybe_expr_list
%type <params> paren_expr_list
%type <field_inits> struct_expr_list
%type <one_field_init> struct_expr_tail
/* Precedence. */
%nonassoc DOTDOT DOTDOTEQ
%right '=' COMPOUND_ASSIGN
%left OROR
%left ANDAND
%nonassoc EQEQ NOTEQ '<' '>' LTEQ GTEQ
%left '|'
%left '^'
%left '&'
%left LSH RSH
%left '@'
%left '+' '-'
%left '*' '/' '%'
/* These could be %precedence in Bison, but that isn't a yacc
feature. */
%left KW_AS
%left UNARY
%left '[' '.' '('
%%
start:
expr
{
/* If we are completing and see a valid parse,
rust_ast will already have been set. */
if (current_parser->rust_ast == NULL)
current_parser->rust_ast = $1;
}
;
/* Note that the Rust grammar includes a method_call_expr, but we
handle this differently, to avoid a shift/reduce conflict with
call_expr. */
expr:
literal
| path_expr
| tuple_expr
| unit_expr
| struct_expr
| field_expr
| array_expr
| idx_expr
| range_expr
| unop_expr /* Must precede call_expr because of ambiguity with
sizeof. */
| binop_expr
| paren_expr
| call_expr
;
tuple_expr:
'(' expr ',' maybe_expr_list ')'
{
$4->push_back ($2);
error (_("Tuple expressions not supported yet"));
}
;
unit_expr:
'(' ')'
{
struct typed_val_int val;
val.type
= (language_lookup_primitive_type
(current_parser->language (), current_parser->arch (),
"()"));
val.val = 0;
$$ = ast_literal (val);
}
;
/* To avoid a shift/reduce conflict with call_expr, we don't handle
tuple struct expressions here, but instead when examining the
AST. */
struct_expr:
path_for_expr '{' struct_expr_list '}'
{ $$ = ast_struct ($1, $3); }
;
struct_expr_tail:
DOTDOT expr
{
struct set_field sf;
sf.name.ptr = NULL;
sf.name.length = 0;
sf.init = $2;
$$ = sf;
}
| IDENT ':' expr
{
struct set_field sf;
sf.name = $1;
sf.init = $3;
$$ = sf;
}
| IDENT
{
struct set_field sf;
sf.name = $1;
sf.init = ast_path ($1, NULL);
$$ = sf;
}
;
struct_expr_list:
/* %empty */
{
$$ = current_parser->new_set_vector ();
}
| struct_expr_tail
{
rust_set_vector *result = current_parser->new_set_vector ();
result->push_back ($1);
$$ = result;
}
| IDENT ':' expr ',' struct_expr_list
{
struct set_field sf;
sf.name = $1;
sf.init = $3;
$5->push_back (sf);
$$ = $5;
}
| IDENT ',' struct_expr_list
{
struct set_field sf;
sf.name = $1;
sf.init = ast_path ($1, NULL);
$3->push_back (sf);
$$ = $3;
}
;
array_expr:
'[' KW_MUT expr_list ']'
{ $$ = ast_call_ish (OP_ARRAY, NULL, $3); }
| '[' expr_list ']'
{ $$ = ast_call_ish (OP_ARRAY, NULL, $2); }
| '[' KW_MUT expr ';' expr ']'
{ $$ = ast_operation (OP_RUST_ARRAY, $3, $5); }
| '[' expr ';' expr ']'
{ $$ = ast_operation (OP_RUST_ARRAY, $2, $4); }
;
range_expr:
expr DOTDOT
{ $$ = ast_range ($1, NULL, false); }
| expr DOTDOT expr
{ $$ = ast_range ($1, $3, false); }
| expr DOTDOTEQ expr
{ $$ = ast_range ($1, $3, true); }
| DOTDOT expr
{ $$ = ast_range (NULL, $2, false); }
| DOTDOTEQ expr
{ $$ = ast_range (NULL, $2, true); }
| DOTDOT
{ $$ = ast_range (NULL, NULL, false); }
;
literal:
INTEGER
{ $$ = ast_literal ($1); }
| DECIMAL_INTEGER
{ $$ = ast_literal ($1); }
| FLOAT
{ $$ = ast_dliteral ($1); }
| STRING
{
const struct rust_op *str = ast_string ($1);
struct set_field field;
struct typed_val_int val;
struct stoken token;
rust_set_vector *fields = current_parser->new_set_vector ();
/* Wrap the raw string in the &str struct. */
field.name.ptr = "data_ptr";
field.name.length = strlen (field.name.ptr);
field.init = ast_unary (UNOP_ADDR, ast_string ($1));
fields->push_back (field);
val.type = rust_type ("usize");
val.val = $1.length;
field.name.ptr = "length";
field.name.length = strlen (field.name.ptr);
field.init = ast_literal (val);
fields->push_back (field);
token.ptr = "&str";
token.length = strlen (token.ptr);
$$ = ast_struct (ast_path (token, NULL), fields);
}
| BYTESTRING
{ $$ = ast_string ($1); }
| KW_TRUE
{
struct typed_val_int val;
val.type = language_bool_type (current_parser->language (),
current_parser->arch ());
val.val = 1;
$$ = ast_literal (val);
}
| KW_FALSE
{
struct typed_val_int val;
val.type = language_bool_type (current_parser->language (),
current_parser->arch ());
val.val = 0;
$$ = ast_literal (val);
}
;
field_expr:
expr '.' IDENT
{ $$ = ast_structop ($1, $3.ptr, 0); }
| expr '.' COMPLETE
{
$$ = ast_structop ($1, $3.ptr, 1);
current_parser->rust_ast = $$;
}
| expr '.' DECIMAL_INTEGER
{ $$ = ast_structop_anonymous ($1, $3); }
;
idx_expr:
expr '[' expr ']'
{ $$ = ast_operation (BINOP_SUBSCRIPT, $1, $3); }
;
unop_expr:
'+' expr %prec UNARY
{ $$ = ast_unary (UNOP_PLUS, $2); }
| '-' expr %prec UNARY
{ $$ = ast_unary (UNOP_NEG, $2); }
| '!' expr %prec UNARY
{
/* Note that we provide a Rust-specific evaluator
override for UNOP_COMPLEMENT, so it can do the
right thing for both bool and integral
values. */
$$ = ast_unary (UNOP_COMPLEMENT, $2);
}
| '*' expr %prec UNARY
{ $$ = ast_unary (UNOP_IND, $2); }
| '&' expr %prec UNARY
{ $$ = ast_unary (UNOP_ADDR, $2); }
| '&' KW_MUT expr %prec UNARY
{ $$ = ast_unary (UNOP_ADDR, $3); }
| KW_SIZEOF '(' expr ')' %prec UNARY
{ $$ = ast_unary (UNOP_SIZEOF, $3); }
;
binop_expr:
binop_expr_expr
| type_cast_expr
| assignment_expr
| compound_assignment_expr
;
binop_expr_expr:
expr '*' expr
{ $$ = ast_operation (BINOP_MUL, $1, $3); }
| expr '@' expr
{ $$ = ast_operation (BINOP_REPEAT, $1, $3); }
| expr '/' expr
{ $$ = ast_operation (BINOP_DIV, $1, $3); }
| expr '%' expr
{ $$ = ast_operation (BINOP_REM, $1, $3); }
| expr '<' expr
{ $$ = ast_operation (BINOP_LESS, $1, $3); }
| expr '>' expr
{ $$ = ast_operation (BINOP_GTR, $1, $3); }
| expr '&' expr
{ $$ = ast_operation (BINOP_BITWISE_AND, $1, $3); }
| expr '|' expr
{ $$ = ast_operation (BINOP_BITWISE_IOR, $1, $3); }
| expr '^' expr
{ $$ = ast_operation (BINOP_BITWISE_XOR, $1, $3); }
| expr '+' expr
{ $$ = ast_operation (BINOP_ADD, $1, $3); }
| expr '-' expr
{ $$ = ast_operation (BINOP_SUB, $1, $3); }
| expr OROR expr
{ $$ = ast_operation (BINOP_LOGICAL_OR, $1, $3); }
| expr ANDAND expr
{ $$ = ast_operation (BINOP_LOGICAL_AND, $1, $3); }
| expr EQEQ expr
{ $$ = ast_operation (BINOP_EQUAL, $1, $3); }
| expr NOTEQ expr
{ $$ = ast_operation (BINOP_NOTEQUAL, $1, $3); }
| expr LTEQ expr
{ $$ = ast_operation (BINOP_LEQ, $1, $3); }
| expr GTEQ expr
{ $$ = ast_operation (BINOP_GEQ, $1, $3); }
| expr LSH expr
{ $$ = ast_operation (BINOP_LSH, $1, $3); }
| expr RSH expr
{ $$ = ast_operation (BINOP_RSH, $1, $3); }
;
type_cast_expr:
expr KW_AS type
{ $$ = ast_cast ($1, $3); }
;
assignment_expr:
expr '=' expr
{ $$ = ast_operation (BINOP_ASSIGN, $1, $3); }
;
compound_assignment_expr:
expr COMPOUND_ASSIGN expr
{ $$ = ast_compound_assignment ($2, $1, $3); }
;
paren_expr:
'(' expr ')'
{ $$ = $2; }
;
expr_list:
expr
{
$$ = current_parser->new_op_vector ();
$$->push_back ($1);
}
| expr_list ',' expr
{
$1->push_back ($3);
$$ = $1;
}
;
maybe_expr_list:
/* %empty */
{
/* The result can't be NULL. */
$$ = current_parser->new_op_vector ();
}
| expr_list
{ $$ = $1; }
;
paren_expr_list:
'(' maybe_expr_list ')'
{ $$ = $2; }
;
call_expr:
expr paren_expr_list
{ $$ = ast_call_ish (OP_FUNCALL, $1, $2); }
;
maybe_self_path:
/* %empty */
| KW_SELF COLONCOLON
;
super_path:
KW_SUPER COLONCOLON
{ $$ = 1; }
| super_path KW_SUPER COLONCOLON
{ $$ = $1 + 1; }
;
path_expr:
path_for_expr
{ $$ = $1; }
| GDBVAR
{ $$ = ast_path ($1, NULL); }
| KW_SELF
{ $$ = ast_path (make_stoken ("self"), NULL); }
;
path_for_expr:
identifier_path_for_expr
| KW_SELF COLONCOLON identifier_path_for_expr
{ $$ = super_name ($3, 0); }
| maybe_self_path super_path identifier_path_for_expr
{ $$ = super_name ($3, $2); }
| COLONCOLON identifier_path_for_expr
{ $$ = crate_name ($2); }
| KW_EXTERN identifier_path_for_expr
{
/* This is a gdb extension to make it possible to
refer to items in other crates. It just bypasses
adding the current crate to the front of the
name. */
$$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL),
$2->right.params);
}
;
identifier_path_for_expr:
IDENT
{ $$ = ast_path ($1, NULL); }
| identifier_path_for_expr COLONCOLON IDENT
{
$$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::",
$3.ptr),
NULL);
}
| identifier_path_for_expr COLONCOLON '<' type_list '>'
{ $$ = ast_path ($1->left.sval, $4); }
| identifier_path_for_expr COLONCOLON '<' type_list RSH
{
$$ = ast_path ($1->left.sval, $4);
rust_push_back ('>');
}
;
path_for_type:
identifier_path_for_type
| KW_SELF COLONCOLON identifier_path_for_type
{ $$ = super_name ($3, 0); }
| maybe_self_path super_path identifier_path_for_type
{ $$ = super_name ($3, $2); }
| COLONCOLON identifier_path_for_type
{ $$ = crate_name ($2); }
| KW_EXTERN identifier_path_for_type
{
/* This is a gdb extension to make it possible to
refer to items in other crates. It just bypasses
adding the current crate to the front of the
name. */
$$ = ast_path (rust_concat3 ("::", $2->left.sval.ptr, NULL),
$2->right.params);
}
;
just_identifiers_for_type:
IDENT
{ $$ = ast_path ($1, NULL); }
| just_identifiers_for_type COLONCOLON IDENT
{
$$ = ast_path (rust_concat3 ($1->left.sval.ptr, "::",
$3.ptr),
NULL);
}
;
identifier_path_for_type:
just_identifiers_for_type
| just_identifiers_for_type '<' type_list '>'
{ $$ = ast_path ($1->left.sval, $3); }
| just_identifiers_for_type '<' type_list RSH
{
$$ = ast_path ($1->left.sval, $3);
rust_push_back ('>');
}
;
type:
path_for_type
| '[' type ';' INTEGER ']'
{ $$ = ast_array_type ($2, $4); }
| '[' type ';' DECIMAL_INTEGER ']'
{ $$ = ast_array_type ($2, $4); }
| '&' '[' type ']'
{ $$ = ast_slice_type ($3); }
| '&' type
{ $$ = ast_reference_type ($2); }
| '*' KW_MUT type
{ $$ = ast_pointer_type ($3, 1); }
| '*' KW_CONST type
{ $$ = ast_pointer_type ($3, 0); }
| KW_FN '(' maybe_type_list ')' ARROW type
{ $$ = ast_function_type ($6, $3); }
| '(' maybe_type_list ')'
{ $$ = ast_tuple_type ($2); }
;
maybe_type_list:
/* %empty */
{ $$ = NULL; }
| type_list
{ $$ = $1; }
;
type_list:
type
{
rust_op_vector *result = current_parser->new_op_vector ();
result->push_back ($1);
$$ = result;
}
| type_list ',' type
{
$1->push_back ($3);
$$ = $1;
}
;
%%
/* A struct of this type is used to describe a token. */
struct token_info
{
const char *name;
int value;
enum exp_opcode opcode;
};
/* Identifier tokens. */
static const struct token_info identifier_tokens[] =
{
{ "as", KW_AS, OP_NULL },
{ "false", KW_FALSE, OP_NULL },
{ "if", 0, OP_NULL },
{ "mut", KW_MUT, OP_NULL },
{ "const", KW_CONST, OP_NULL },
{ "self", KW_SELF, OP_NULL },
{ "super", KW_SUPER, OP_NULL },
{ "true", KW_TRUE, OP_NULL },
{ "extern", KW_EXTERN, OP_NULL },
{ "fn", KW_FN, OP_NULL },
{ "sizeof", KW_SIZEOF, OP_NULL },
};
/* Operator tokens, sorted longest first. */
static const struct token_info operator_tokens[] =
{
{ ">>=", COMPOUND_ASSIGN, BINOP_RSH },
{ "<<=", COMPOUND_ASSIGN, BINOP_LSH },
{ "<<", LSH, OP_NULL },
{ ">>", RSH, OP_NULL },
{ "&&", ANDAND, OP_NULL },
{ "||", OROR, OP_NULL },
{ "==", EQEQ, OP_NULL },
{ "!=", NOTEQ, OP_NULL },
{ "<=", LTEQ, OP_NULL },
{ ">=", GTEQ, OP_NULL },
{ "+=", COMPOUND_ASSIGN, BINOP_ADD },
{ "-=", COMPOUND_ASSIGN, BINOP_SUB },
{ "*=", COMPOUND_ASSIGN, BINOP_MUL },
{ "/=", COMPOUND_ASSIGN, BINOP_DIV },
{ "%=", COMPOUND_ASSIGN, BINOP_REM },
{ "&=", COMPOUND_ASSIGN, BINOP_BITWISE_AND },
{ "|=", COMPOUND_ASSIGN, BINOP_BITWISE_IOR },
{ "^=", COMPOUND_ASSIGN, BINOP_BITWISE_XOR },
{ "..=", DOTDOTEQ, OP_NULL },
{ "::", COLONCOLON, OP_NULL },
{ "..", DOTDOT, OP_NULL },
{ "->", ARROW, OP_NULL }
};
/* Helper function to copy to the name obstack. */
static const char *
rust_copy_name (const char *name, int len)
{
return (const char *) obstack_copy0 (work_obstack, name, len);
}
/* Helper function to make an stoken from a C string. */
static struct stoken
make_stoken (const char *p)
{
struct stoken result;
result.ptr = p;
result.length = strlen (result.ptr);
return result;
}
/* Helper function to concatenate three strings on the name
obstack. */
static struct stoken
rust_concat3 (const char *s1, const char *s2, const char *s3)
{
return make_stoken (obconcat (work_obstack, s1, s2, s3, (char *) NULL));
}
/* Return an AST node referring to NAME, but relative to the crate's
name. */
static const struct rust_op *
crate_name (const struct rust_op *name)
{
std::string crate = rust_crate_for_block (expression_context_block);
struct stoken result;
gdb_assert (name->opcode == OP_VAR_VALUE);
if (crate.empty ())
error (_("Could not find crate for current location"));
result = make_stoken (obconcat (work_obstack, "::", crate.c_str (), "::",
name->left.sval.ptr, (char *) NULL));
return ast_path (result, name->right.params);
}
/* Create an AST node referring to a "super::" qualified name. IDENT
is the base name and N_SUPERS is how many "super::"s were
provided. N_SUPERS can be zero. */
static const struct rust_op *
super_name (const struct rust_op *ident, unsigned int n_supers)
{
const char *scope = block_scope (expression_context_block);
int offset;
gdb_assert (ident->opcode == OP_VAR_VALUE);
if (scope[0] == '\0')
error (_("Couldn't find namespace scope for self::"));
if (n_supers > 0)
{
int len;
std::vector<int> offsets;
unsigned int current_len;
current_len = cp_find_first_component (scope);
while (scope[current_len] != '\0')
{
offsets.push_back (current_len);
gdb_assert (scope[current_len] == ':');
/* The "::". */
current_len += 2;
current_len += cp_find_first_component (scope
+ current_len);
}
len = offsets.size ();
if (n_supers >= len)
error (_("Too many super:: uses from '%s'"), scope);
offset = offsets[len - n_supers];
}
else
offset = strlen (scope);
obstack_grow (work_obstack, "::", 2);
obstack_grow (work_obstack, scope, offset);
obstack_grow (work_obstack, "::", 2);
obstack_grow0 (work_obstack, ident->left.sval.ptr, ident->left.sval.length);
return ast_path (make_stoken ((const char *) obstack_finish (work_obstack)),
ident->right.params);
}
/* A helper that updates the innermost block as appropriate. */
static void
update_innermost_block (struct block_symbol sym)
{
if (symbol_read_needs_frame (sym.symbol))
innermost_block.update (sym);
}
/* A helper to look up a Rust type, or fail. This only works for
types defined by rust_language_arch_info. */
static struct type *
rust_type (const char *name)
{
struct type *type;
type = language_lookup_primitive_type (current_parser->language (),
current_parser->arch (),
name);
if (type == NULL)
error (_("Could not find Rust type %s"), name);
return type;
}
/* Lex a hex number with at least MIN digits and at most MAX
digits. */
static uint32_t
lex_hex (int min, int max)
{
uint32_t result = 0;
int len = 0;
/* We only want to stop at MAX if we're lexing a byte escape. */
int check_max = min == max;
while ((check_max ? len <= max : 1)
&& ((lexptr[0] >= 'a' && lexptr[0] <= 'f')
|| (lexptr[0] >= 'A' && lexptr[0] <= 'F')
|| (lexptr[0] >= '0' && lexptr[0] <= '9')))
{
result *= 16;
if (lexptr[0] >= 'a' && lexptr[0] <= 'f')
result = result + 10 + lexptr[0] - 'a';
else if (lexptr[0] >= 'A' && lexptr[0] <= 'F')
result = result + 10 + lexptr[0] - 'A';
else
result = result + lexptr[0] - '0';
++lexptr;
++len;
}
if (len < min)
error (_("Not enough hex digits seen"));
if (len > max)
{
gdb_assert (min != max);
error (_("Overlong hex escape"));
}
return result;
}
/* Lex an escape. IS_BYTE is true if we're lexing a byte escape;
otherwise we're lexing a character escape. */
static uint32_t
lex_escape (int is_byte)
{
uint32_t result;
gdb_assert (lexptr[0] == '\\');
++lexptr;
switch (lexptr[0])
{
case 'x':
++lexptr;
result = lex_hex (2, 2);
break;
case 'u':
if (is_byte)
error (_("Unicode escape in byte literal"));
++lexptr;
if (lexptr[0] != '{')
error (_("Missing '{' in Unicode escape"));
++lexptr;
result = lex_hex (1, 6);
/* Could do range checks here. */
if (lexptr[0] != '}')
error (_("Missing '}' in Unicode escape"));
++lexptr;
break;
case 'n':
result = '\n';
++lexptr;
break;
case 'r':
result = '\r';
++lexptr;
break;
case 't':
result = '\t';
++lexptr;
break;
case '\\':
result = '\\';
++lexptr;
break;
case '0':
result = '\0';
++lexptr;
break;
case '\'':
result = '\'';
++lexptr;
break;
case '"':
result = '"';
++lexptr;
break;
default:
error (_("Invalid escape \\%c in literal"), lexptr[0]);
}
return result;
}
/* Lex a character constant. */
static int
lex_character (void)
{
int is_byte = 0;
uint32_t value;
if (lexptr[0] == 'b')
{
is_byte = 1;
++lexptr;
}
gdb_assert (lexptr[0] == '\'');
++lexptr;
/* This should handle UTF-8 here. */
if (lexptr[0] == '\\')
value = lex_escape (is_byte);
else
{
value = lexptr[0] & 0xff;
++lexptr;
}
if (lexptr[0] != '\'')
error (_("Unterminated character literal"));
++lexptr;
rustyylval.typed_val_int.val = value;
rustyylval.typed_val_int.type = rust_type (is_byte ? "u8" : "char");
return INTEGER;
}
/* Return the offset of the double quote if STR looks like the start
of a raw string, or 0 if STR does not start a raw string. */
static int
starts_raw_string (const char *str)
{
const char *save = str;
if (str[0] != 'r')
return 0;
++str;
while (str[0] == '#')
++str;
if (str[0] == '"')
return str - save;
return 0;
}
/* Return true if STR looks like the end of a raw string that had N
hashes at the start. */
static bool
ends_raw_string (const char *str, int n)
{
int i;
gdb_assert (str[0] == '"');
for (i = 0; i < n; ++i)
if (str[i + 1] != '#')
return false;
return true;
}
/* Lex a string constant. */
static int
lex_string (void)
{
int is_byte = lexptr[0] == 'b';
int raw_length;
if (is_byte)
++lexptr;
raw_length = starts_raw_string (lexptr);
lexptr += raw_length;
gdb_assert (lexptr[0] == '"');
++lexptr;
while (1)
{
uint32_t value;
if (raw_length > 0)
{
if (lexptr[0] == '"' && ends_raw_string (lexptr, raw_length - 1))
{
/* Exit with lexptr pointing after the final "#". */
lexptr += raw_length;
break;
}
else if (lexptr[0] == '\0')
error (_("Unexpected EOF in string"));
value = lexptr[0] & 0xff;
if (is_byte && value > 127)
error (_("Non-ASCII value in raw byte string"));
obstack_1grow (work_obstack, value);
++lexptr;
}
else if (lexptr[0] == '"')
{
/* Make sure to skip the quote. */
++lexptr;
break;
}
else if (lexptr[0] == '\\')
{
value = lex_escape (is_byte);
if (is_byte)
obstack_1grow (work_obstack, value);
else
convert_between_encodings ("UTF-32", "UTF-8", (gdb_byte *) &value,
sizeof (value), sizeof (value),
work_obstack, translit_none);
}
else if (lexptr[0] == '\0')
error (_("Unexpected EOF in string"));
else
{
value = lexptr[0] & 0xff;
if (is_byte && value > 127)
error (_("Non-ASCII value in byte string"));
obstack_1grow (work_obstack, value);
++lexptr;
}
}
rustyylval.sval.length = obstack_object_size (work_obstack);
rustyylval.sval.ptr = (const char *) obstack_finish (work_obstack);
return is_byte ? BYTESTRING : STRING;
}
/* Return true if STRING starts with whitespace followed by a digit. */
static bool
space_then_number (const char *string)
{
const char *p = string;
while (p[0] == ' ' || p[0] == '\t')
++p;
if (p == string)
return false;
return *p >= '0' && *p <= '9';
}
/* Return true if C can start an identifier. */
static bool
rust_identifier_start_p (char c)
{
return ((c >= 'a' && c <= 'z')
|| (c >= 'A' && c <= 'Z')
|| c == '_'
|| c == '$');
}
/* Lex an identifier. */
static int
lex_identifier (void)
{
const char *start = lexptr;
unsigned int length;
const struct token_info *token;
int i;
int is_gdb_var = lexptr[0] == '$';
gdb_assert (rust_identifier_start_p (lexptr[0]));
++lexptr;
/* For the time being this doesn't handle Unicode rules. Non-ASCII
identifiers are gated anyway. */
while ((lexptr[0] >= 'a' && lexptr[0] <= 'z')
|| (lexptr[0] >= 'A' && lexptr[0] <= 'Z')
|| lexptr[0] == '_'
|| (is_gdb_var && lexptr[0] == '$')
|| (lexptr[0] >= '0' && lexptr[0] <= '9'))
++lexptr;
length = lexptr - start;
token = NULL;
for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
{
if (length == strlen (identifier_tokens[i].name)
&& strncmp (identifier_tokens[i].name, start, length) == 0)
{
token = &identifier_tokens[i];
break;
}
}
if (token != NULL)
{
if (token->value == 0)
{
/* Leave the terminating token alone. */
lexptr = start;
return 0;
}
}
else if (token == NULL
&& (strncmp (start, "thread", length) == 0
|| strncmp (start, "task", length) == 0)
&& space_then_number (lexptr))
{
/* "task" or "thread" followed by a number terminates the
parse, per gdb rules. */
lexptr = start;
return 0;
}
if (token == NULL || (parse_completion && lexptr[0] == '\0'))
rustyylval.sval = make_stoken (rust_copy_name (start, length));
if (parse_completion && lexptr[0] == '\0')
{
/* Prevent rustyylex from returning two COMPLETE tokens. */
prev_lexptr = lexptr;
return COMPLETE;
}
if (token != NULL)
return token->value;
if (is_gdb_var)
return GDBVAR;
return IDENT;
}
/* Lex an operator. */
static int
lex_operator (void)
{
const struct token_info *token = NULL;
int i;
for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
{
if (strncmp (operator_tokens[i].name, lexptr,
strlen (operator_tokens[i].name)) == 0)
{
lexptr += strlen (operator_tokens[i].name);
token = &operator_tokens[i];
break;
}
}
if (token != NULL)
{
rustyylval.opcode = token->opcode;
return token->value;
}
return *lexptr++;
}
/* Lex a number. */
static int
lex_number (void)
{
regmatch_t subexps[NUM_SUBEXPRESSIONS];
int match;
int is_integer = 0;
int could_be_decimal = 1;
int implicit_i32 = 0;
const char *type_name = NULL;
struct type *type;
int end_index;
int type_index = -1;
int i;
match = regexec (&number_regex, lexptr, ARRAY_SIZE (subexps), subexps, 0);
/* Failure means the regexp is broken. */
gdb_assert (match == 0);
if (subexps[INT_TEXT].rm_so != -1)
{
/* Integer part matched. */
is_integer = 1;
end_index = subexps[INT_TEXT].rm_eo;
if (subexps[INT_TYPE].rm_so == -1)
{
type_name = "i32";
implicit_i32 = 1;
}
else
{
type_index = INT_TYPE;
could_be_decimal = 0;
}
}
else if (subexps[FLOAT_TYPE1].rm_so != -1)
{
/* Found floating point type suffix. */
end_index = subexps[FLOAT_TYPE1].rm_so;
type_index = FLOAT_TYPE1;
}
else if (subexps[FLOAT_TYPE2].rm_so != -1)
{
/* Found floating point type suffix. */
end_index = subexps[FLOAT_TYPE2].rm_so;
type_index = FLOAT_TYPE2;
}
else
{
/* Any other floating point match. */
end_index = subexps[0].rm_eo;
type_name = "f64";
}
/* We need a special case if the final character is ".". In this
case we might need to parse an integer. For example, "23.f()" is
a request for a trait method call, not a syntax error involving
the floating point number "23.". */
gdb_assert (subexps[0].rm_eo > 0);
if (lexptr[subexps[0].rm_eo - 1] == '.')
{
Rename _const functions to use overloading instead This renames a few functions -- skip_spaces_const, skip_to_space_const, get_number_const, extract_arg_const -- to drop the "_const" suffix and instead rely on overloading. This makes future const fixes simpler by reducing the number of lines that must be changed. I think it is also not any less clear, as all these functions have the same interface as their non-const versions by design. Furthermore there's an example of using an overload in-tree already, namely check_for_argument. This patch was largely created using some perl one-liners; then a few fixes were applied by hand. ChangeLog 2017-09-11 Tom Tromey <tom@tromey.com> * common/common-utils.h (skip_to_space): Remove macro, redeclare as function. (skip_to_space): Rename from skip_to_space_const. * common/common-utils.c (skip_to_space): New function. (skip_to_space): Rename from skip_to_space_const. * cli/cli-utils.h (get_number): Rename from get_number_const. (extract_arg): Rename from extract_arg_const. * cli/cli-utils.c (get_number): Rename from get_number_const. (extract_arg): Rename from extract_arg_const. (number_or_range_parser::get_number): Use ::get_number. * aarch64-linux-tdep.c, ada-lang.c, arm-linux-tdep.c, ax-gdb.c, break-catch-throw.c, breakpoint.c, cli/cli-cmds.c, cli/cli-dump.c, cli/cli-script.c, cli/cli-setshow.c, compile/compile.c, completer.c, demangle.c, disasm.c, findcmd.c, linespec.c, linux-tdep.c, linux-thread-db.c, location.c, mi/mi-parse.c, minsyms.c, nat/linux-procfs.c, printcmd.c, probe.c, python/py-breakpoint.c, record.c, rust-exp.y, serial.c, stack.c, stap-probe.c, tid-parse.c, tracepoint.c: Update all callers.
2017-09-10 22:19:19 +02:00
const char *next = skip_spaces (&lexptr[subexps[0].rm_eo]);
if (rust_identifier_start_p (*next) || *next == '.')
{
--subexps[0].rm_eo;
is_integer = 1;
end_index = subexps[0].rm_eo;
type_name = "i32";
could_be_decimal = 1;
implicit_i32 = 1;
}
}
/* Compute the type name if we haven't already. */
std::string type_name_holder;
if (type_name == NULL)
{
gdb_assert (type_index != -1);
type_name_holder = std::string (lexptr + subexps[type_index].rm_so,
(subexps[type_index].rm_eo
- subexps[type_index].rm_so));
type_name = type_name_holder.c_str ();
}
/* Look up the type. */
type = rust_type (type_name);
/* Copy the text of the number and remove the "_"s. */
std::string number;
for (i = 0; i < end_index && lexptr[i]; ++i)
{
if (lexptr[i] == '_')
could_be_decimal = 0;
else
number.push_back (lexptr[i]);
}
/* Advance past the match. */
lexptr += subexps[0].rm_eo;
/* Parse the number. */
if (is_integer)
{
uint64_t value;
int radix = 10;
int offset = 0;
if (number[0] == '0')
{
if (number[1] == 'x')
radix = 16;
else if (number[1] == 'o')
radix = 8;
else if (number[1] == 'b')
radix = 2;
if (radix != 10)
{
offset = 2;
could_be_decimal = 0;
}
}
value = strtoul (number.c_str () + offset, NULL, radix);
if (implicit_i32 && value >= ((uint64_t) 1) << 31)
type = rust_type ("i64");
rustyylval.typed_val_int.val = value;
rustyylval.typed_val_int.type = type;
}
else
{
rustyylval.typed_val_float.type = type;
Target FP: Use target format throughout expression parsing When parsing floating-point literals, the language parsers currently use parse_float or some equivalent routine to parse the input string into a DOUBLEST, which is then stored within a OP_DOUBLE expression node. When evaluating the expression, the OP_DOUBLE is finally converted into a value in target format. On the other hand, *decimal* floating-point literals are parsed directly into target format and stored that way in a OP_DECFLOAT expression node. In order to eliminate the DOUBLEST, this patch therefore unifies the handling of binary and decimal floating- point literals and stores them both in target format within a new OP_FLOAT expression node, replacing both OP_DOUBLE and OP_DECFLOAT. In order to store literals in target format, the parse_float routine needs to know the type of the literal. All parsers therefore need to be changed to determine the appropriate type (e.g. by detecting suffixes) *before* calling parse_float, instead of after it as today. However, this change is mostly straightforward -- again, this is already done for decimal FP today. The core of the literal parsing is moved into a new routine floatformat_from_string, mirroring floatformat_to_string. The parse_float routine now calls either floatformat_from_string or decimal_from_sting, allowing it to handle any type of FP literal. All language parsers need to be updated. Some notes on specific changes to the various languages: - C: Decimal FP is now handled in parse_float, and no longer needs to be handled specially. - D: Straightforward. - Fortran: Still used a hard-coded "atof", also replaced by parse_float now. Continues to always use builtin_real_s8 as the type of literal, even though this is probably wrong. - Go: This used to handle "f" and "l" suffixes, even though the Go language actually doesn't support those. I kept this support for now -- maybe revisit later. Note the the GDB test suite for some reason actually *verifies* that GDB supports those unsupported suffixes ... - Pascal: Likewise -- this handles suffixes that are not supported in the language standard. - Modula-2: Like Fortran, used to use "atof". - Rust: Mostly straightforward, except for a unit-testing hitch. The code use to set a special "unit_testing" flag which would cause "rust_type" to always return NULL. This makes it not possible to encode a literal into target format (which type?). The reason for this flag appears to have been that during unit testing, there is no "rust_parser" context set up, which means no "gdbarch" is available to use its types. To fix this, I removed the unit_testing flag, and instead simply just set up a dummy rust_parser context during unit testing. - Ada: This used to check sizeof (DOUBLEST) to determine which type to use for floating-point literal. This seems questionable to begin with (since DOUBLEST is quite unrelated to target formats), and in any case we need to get rid of DOUBLEST. I'm now simply always using the largest type (builtin_long_double). gdb/ChangeLog: 2017-10-25 Ulrich Weigand <uweigand@de.ibm.com> * doublest.c (floatformat_from_string): New function. * doublest.h (floatformat_from_string): Add prototype. * std-operator.def (OP_DOUBLE, OP_DECFLOAT): Remove, replace by ... (OP_FLOAT): ... this. * expression.h: Do not include "doublest.h". (union exp_element): Replace doubleconst and decfloatconst by new element floatconst. * ada-lang.c (resolve_subexp): Handle OP_FLOAT instead of OP_DOUBLE. (ada_evaluate_subexp): Likewise. * eval.c (evaluate_subexp_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. * expprint.c (print_subexp_standard): Likewise. (dump_subexp_body_standard): Likewise. * breakpoint.c (watchpoint_exp_is_const): Likewise. * parse.c: Include "dfp.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): New function. (operator_length_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. (operator_check_standard): Likewise. (parse_float): Do not accept suffix. Take type as input. Return bool. Return target format buffer instead of host DOUBLEST. Use floatformat_from_string and decimal_from_string to parse either binary or decimal floating-point types. (parse_c_float): Remove. * parser-defs.h: Do not include "doublest.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): Add prototype. (parse_float): Update prototype. (parse_c_float): Remove. * c-exp.y: Do not include "dfp.h". (typed_val_float): Use byte buffer instead of DOUBLEST. (typed_val_decfloat): Remove. (DECFLOAT): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. Handle decimal and binary FP types the same way. * d-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT_LITERAL): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * f-exp.y: Replace dval by typed_val_float. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Use parse_float instead of atof. * go-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (parse_go_float): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of parse_go_float. Parse suffixes and determine type before calling parse_float. * p-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * m2-exp.y: Replace dval by byte buffer val. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of atof. * rust-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (lex_number): Call parse_float instead of strtod. (ast_dliteral): Use OP_FLOAT instead of OP_DOUBLE. (convert_ast_to_expression): Handle OP_FLOAT instead of OP_DOUBLE. Use write_exp_elt_floatcst. (unit_testing): Remove static variable. (rust_type): Do not check unit_testing. (rust_lex_tests): Do not set uint_testing. Set up dummy rust_parser. * ada-exp.y (type_float, type_double): Remove. (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. * ada-lex.l (processReal): Use parse_float instead of sscanf.
2017-10-25 15:32:23 +02:00
bool parsed = parse_float (number.c_str (), number.length (),
rustyylval.typed_val_float.type,
rustyylval.typed_val_float.val);
gdb_assert (parsed);
}
return is_integer ? (could_be_decimal ? DECIMAL_INTEGER : INTEGER) : FLOAT;
}
/* The lexer. */
static int
rustyylex (void)
{
/* Skip all leading whitespace. */
while (lexptr[0] == ' ' || lexptr[0] == '\t' || lexptr[0] == '\r'
|| lexptr[0] == '\n')
++lexptr;
/* If we hit EOF and we're completing, then return COMPLETE -- maybe
we're completing an empty string at the end of a field_expr.
But, we don't want to return two COMPLETE tokens in a row. */
if (lexptr[0] == '\0' && lexptr == prev_lexptr)
return 0;
prev_lexptr = lexptr;
if (lexptr[0] == '\0')
{
if (parse_completion)
{
rustyylval.sval = make_stoken ("");
return COMPLETE;
}
return 0;
}
if (lexptr[0] >= '0' && lexptr[0] <= '9')
return lex_number ();
else if (lexptr[0] == 'b' && lexptr[1] == '\'')
return lex_character ();
else if (lexptr[0] == 'b' && lexptr[1] == '"')
return lex_string ();
else if (lexptr[0] == 'b' && starts_raw_string (lexptr + 1))
return lex_string ();
else if (starts_raw_string (lexptr))
return lex_string ();
else if (rust_identifier_start_p (lexptr[0]))
return lex_identifier ();
else if (lexptr[0] == '"')
return lex_string ();
else if (lexptr[0] == '\'')
return lex_character ();
else if (lexptr[0] == '}' || lexptr[0] == ']')
{
/* Falls through to lex_operator. */
--paren_depth;
}
else if (lexptr[0] == '(' || lexptr[0] == '{')
{
/* Falls through to lex_operator. */
++paren_depth;
}
else if (lexptr[0] == ',' && comma_terminates && paren_depth == 0)
return 0;
return lex_operator ();
}
/* Push back a single character to be re-lexed. */
static void
rust_push_back (char c)
{
/* Can't be called before any lexing. */
gdb_assert (prev_lexptr != NULL);
--lexptr;
gdb_assert (*lexptr == c);
}
/* Make an arbitrary operation and fill in the fields. */
static const struct rust_op *
ast_operation (enum exp_opcode opcode, const struct rust_op *left,
const struct rust_op *right)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = opcode;
result->left.op = left;
result->right.op = right;
return result;
}
/* Make a compound assignment operation. */
static const struct rust_op *
ast_compound_assignment (enum exp_opcode opcode, const struct rust_op *left,
const struct rust_op *right)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = opcode;
result->compound_assignment = 1;
result->left.op = left;
result->right.op = right;
return result;
}
/* Make a typed integer literal operation. */
static const struct rust_op *
ast_literal (struct typed_val_int val)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = OP_LONG;
result->left.typed_val_int = val;
return result;
}
/* Make a typed floating point literal operation. */
static const struct rust_op *
ast_dliteral (struct typed_val_float val)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
Target FP: Use target format throughout expression parsing When parsing floating-point literals, the language parsers currently use parse_float or some equivalent routine to parse the input string into a DOUBLEST, which is then stored within a OP_DOUBLE expression node. When evaluating the expression, the OP_DOUBLE is finally converted into a value in target format. On the other hand, *decimal* floating-point literals are parsed directly into target format and stored that way in a OP_DECFLOAT expression node. In order to eliminate the DOUBLEST, this patch therefore unifies the handling of binary and decimal floating- point literals and stores them both in target format within a new OP_FLOAT expression node, replacing both OP_DOUBLE and OP_DECFLOAT. In order to store literals in target format, the parse_float routine needs to know the type of the literal. All parsers therefore need to be changed to determine the appropriate type (e.g. by detecting suffixes) *before* calling parse_float, instead of after it as today. However, this change is mostly straightforward -- again, this is already done for decimal FP today. The core of the literal parsing is moved into a new routine floatformat_from_string, mirroring floatformat_to_string. The parse_float routine now calls either floatformat_from_string or decimal_from_sting, allowing it to handle any type of FP literal. All language parsers need to be updated. Some notes on specific changes to the various languages: - C: Decimal FP is now handled in parse_float, and no longer needs to be handled specially. - D: Straightforward. - Fortran: Still used a hard-coded "atof", also replaced by parse_float now. Continues to always use builtin_real_s8 as the type of literal, even though this is probably wrong. - Go: This used to handle "f" and "l" suffixes, even though the Go language actually doesn't support those. I kept this support for now -- maybe revisit later. Note the the GDB test suite for some reason actually *verifies* that GDB supports those unsupported suffixes ... - Pascal: Likewise -- this handles suffixes that are not supported in the language standard. - Modula-2: Like Fortran, used to use "atof". - Rust: Mostly straightforward, except for a unit-testing hitch. The code use to set a special "unit_testing" flag which would cause "rust_type" to always return NULL. This makes it not possible to encode a literal into target format (which type?). The reason for this flag appears to have been that during unit testing, there is no "rust_parser" context set up, which means no "gdbarch" is available to use its types. To fix this, I removed the unit_testing flag, and instead simply just set up a dummy rust_parser context during unit testing. - Ada: This used to check sizeof (DOUBLEST) to determine which type to use for floating-point literal. This seems questionable to begin with (since DOUBLEST is quite unrelated to target formats), and in any case we need to get rid of DOUBLEST. I'm now simply always using the largest type (builtin_long_double). gdb/ChangeLog: 2017-10-25 Ulrich Weigand <uweigand@de.ibm.com> * doublest.c (floatformat_from_string): New function. * doublest.h (floatformat_from_string): Add prototype. * std-operator.def (OP_DOUBLE, OP_DECFLOAT): Remove, replace by ... (OP_FLOAT): ... this. * expression.h: Do not include "doublest.h". (union exp_element): Replace doubleconst and decfloatconst by new element floatconst. * ada-lang.c (resolve_subexp): Handle OP_FLOAT instead of OP_DOUBLE. (ada_evaluate_subexp): Likewise. * eval.c (evaluate_subexp_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. * expprint.c (print_subexp_standard): Likewise. (dump_subexp_body_standard): Likewise. * breakpoint.c (watchpoint_exp_is_const): Likewise. * parse.c: Include "dfp.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): New function. (operator_length_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. (operator_check_standard): Likewise. (parse_float): Do not accept suffix. Take type as input. Return bool. Return target format buffer instead of host DOUBLEST. Use floatformat_from_string and decimal_from_string to parse either binary or decimal floating-point types. (parse_c_float): Remove. * parser-defs.h: Do not include "doublest.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): Add prototype. (parse_float): Update prototype. (parse_c_float): Remove. * c-exp.y: Do not include "dfp.h". (typed_val_float): Use byte buffer instead of DOUBLEST. (typed_val_decfloat): Remove. (DECFLOAT): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. Handle decimal and binary FP types the same way. * d-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT_LITERAL): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * f-exp.y: Replace dval by typed_val_float. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Use parse_float instead of atof. * go-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (parse_go_float): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of parse_go_float. Parse suffixes and determine type before calling parse_float. * p-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * m2-exp.y: Replace dval by byte buffer val. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of atof. * rust-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (lex_number): Call parse_float instead of strtod. (ast_dliteral): Use OP_FLOAT instead of OP_DOUBLE. (convert_ast_to_expression): Handle OP_FLOAT instead of OP_DOUBLE. Use write_exp_elt_floatcst. (unit_testing): Remove static variable. (rust_type): Do not check unit_testing. (rust_lex_tests): Do not set uint_testing. Set up dummy rust_parser. * ada-exp.y (type_float, type_double): Remove. (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. * ada-lex.l (processReal): Use parse_float instead of sscanf.
2017-10-25 15:32:23 +02:00
result->opcode = OP_FLOAT;
result->left.typed_val_float = val;
return result;
}
/* Make a unary operation. */
static const struct rust_op *
ast_unary (enum exp_opcode opcode, const struct rust_op *expr)
{
return ast_operation (opcode, expr, NULL);
}
/* Make a cast operation. */
static const struct rust_op *
ast_cast (const struct rust_op *expr, const struct rust_op *type)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = UNOP_CAST;
result->left.op = expr;
result->right.op = type;
return result;
}
/* Make a call-like operation. This is nominally a function call, but
when lowering we may discover that it actually represents the
creation of a tuple struct. */
static const struct rust_op *
ast_call_ish (enum exp_opcode opcode, const struct rust_op *expr,
rust_op_vector *params)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = opcode;
result->left.op = expr;
result->right.params = params;
return result;
}
/* Make a structure creation operation. */
static const struct rust_op *
ast_struct (const struct rust_op *name, rust_set_vector *fields)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = OP_AGGREGATE;
result->left.op = name;
result->right.field_inits = fields;
return result;
}
/* Make an identifier path. */
static const struct rust_op *
ast_path (struct stoken path, rust_op_vector *params)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = OP_VAR_VALUE;
result->left.sval = path;
result->right.params = params;
return result;
}
/* Make a string constant operation. */
static const struct rust_op *
ast_string (struct stoken str)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = OP_STRING;
result->left.sval = str;
return result;
}
/* Make a field expression. */
static const struct rust_op *
ast_structop (const struct rust_op *left, const char *name, int completing)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = STRUCTOP_STRUCT;
result->completing = completing;
result->left.op = left;
result->right.sval = make_stoken (name);
return result;
}
/* Make an anonymous struct operation, like 'x.0'. */
static const struct rust_op *
ast_structop_anonymous (const struct rust_op *left,
struct typed_val_int number)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = STRUCTOP_ANONYMOUS;
result->left.op = left;
result->right.typed_val_int = number;
return result;
}
/* Make a range operation. */
static const struct rust_op *
ast_range (const struct rust_op *lhs, const struct rust_op *rhs,
bool inclusive)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = OP_RANGE;
result->inclusive = inclusive;
result->left.op = lhs;
result->right.op = rhs;
return result;
}
/* A helper function to make a type-related AST node. */
static struct rust_op *
ast_basic_type (enum type_code typecode)
{
struct rust_op *result = OBSTACK_ZALLOC (work_obstack, struct rust_op);
result->opcode = OP_TYPE;
result->typecode = typecode;
return result;
}
/* Create an AST node describing an array type. */
static const struct rust_op *
ast_array_type (const struct rust_op *lhs, struct typed_val_int val)
{
struct rust_op *result = ast_basic_type (TYPE_CODE_ARRAY);
result->left.op = lhs;
result->right.typed_val_int = val;
return result;
}
/* Create an AST node describing a reference type. */
static const struct rust_op *
ast_slice_type (const struct rust_op *type)
{
/* Use TYPE_CODE_COMPLEX just because it is handy. */
struct rust_op *result = ast_basic_type (TYPE_CODE_COMPLEX);
result->left.op = type;
return result;
}
/* Create an AST node describing a reference type. */
static const struct rust_op *
ast_reference_type (const struct rust_op *type)
{
struct rust_op *result = ast_basic_type (TYPE_CODE_REF);
result->left.op = type;
return result;
}
/* Create an AST node describing a pointer type. */
static const struct rust_op *
ast_pointer_type (const struct rust_op *type, int is_mut)
{
struct rust_op *result = ast_basic_type (TYPE_CODE_PTR);
result->left.op = type;
/* For the time being we ignore is_mut. */
return result;
}
/* Create an AST node describing a function type. */
static const struct rust_op *
ast_function_type (const struct rust_op *rtype, rust_op_vector *params)
{
struct rust_op *result = ast_basic_type (TYPE_CODE_FUNC);
result->left.op = rtype;
result->right.params = params;
return result;
}
/* Create an AST node describing a tuple type. */
static const struct rust_op *
ast_tuple_type (rust_op_vector *params)
{
struct rust_op *result = ast_basic_type (TYPE_CODE_STRUCT);
result->left.params = params;
return result;
}
/* A helper to appropriately munge NAME and BLOCK depending on the
presence of a leading "::". */
static void
munge_name_and_block (const char **name, const struct block **block)
{
/* If it is a global reference, skip the current block in favor of
the static block. */
if (strncmp (*name, "::", 2) == 0)
{
*name += 2;
*block = block_static_block (*block);
}
}
/* Like lookup_symbol, but handles Rust namespace conventions, and
doesn't require field_of_this_result. */
static struct block_symbol
rust_lookup_symbol (const char *name, const struct block *block,
const domain_enum domain)
{
struct block_symbol result;
munge_name_and_block (&name, &block);
result = lookup_symbol (name, block, domain, NULL);
if (result.symbol != NULL)
update_innermost_block (result);
return result;
}
/* Look up a type, following Rust namespace conventions. */
static struct type *
rust_lookup_type (const char *name, const struct block *block)
{
struct block_symbol result;
struct type *type;
munge_name_and_block (&name, &block);
result = lookup_symbol (name, block, STRUCT_DOMAIN, NULL);
if (result.symbol != NULL)
{
update_innermost_block (result);
return SYMBOL_TYPE (result.symbol);
}
type = lookup_typename (current_parser->language (), current_parser->arch (),
name, NULL, 1);
if (type != NULL)
return type;
/* Last chance, try a built-in type. */
return language_lookup_primitive_type (current_parser->language (),
current_parser->arch (),
name);
}
static struct type *convert_ast_to_type (struct parser_state *state,
const struct rust_op *operation);
static const char *convert_name (struct parser_state *state,
const struct rust_op *operation);
/* Convert a vector of rust_ops representing types to a vector of
types. */
static std::vector<struct type *>
convert_params_to_types (struct parser_state *state, rust_op_vector *params)
{
std::vector<struct type *> result;
if (params != nullptr)
{
for (const rust_op *op : *params)
result.push_back (convert_ast_to_type (state, op));
}
return result;
}
/* Convert a rust_op representing a type to a struct type *. */
static struct type *
convert_ast_to_type (struct parser_state *state,
const struct rust_op *operation)
{
struct type *type, *result = NULL;
if (operation->opcode == OP_VAR_VALUE)
{
const char *varname = convert_name (state, operation);
result = rust_lookup_type (varname, expression_context_block);
if (result == NULL)
error (_("No typed name '%s' in current context"), varname);
return result;
}
gdb_assert (operation->opcode == OP_TYPE);
switch (operation->typecode)
{
case TYPE_CODE_ARRAY:
type = convert_ast_to_type (state, operation->left.op);
if (operation->right.typed_val_int.val < 0)
error (_("Negative array length"));
result = lookup_array_range_type (type, 0,
operation->right.typed_val_int.val - 1);
break;
case TYPE_CODE_COMPLEX:
{
struct type *usize = rust_type ("usize");
type = convert_ast_to_type (state, operation->left.op);
result = rust_slice_type ("&[*gdb*]", type, usize);
}
break;
case TYPE_CODE_REF:
case TYPE_CODE_PTR:
/* For now we treat &x and *x identically. */
type = convert_ast_to_type (state, operation->left.op);
result = lookup_pointer_type (type);
break;
case TYPE_CODE_FUNC:
{
std::vector<struct type *> args
(convert_params_to_types (state, operation->right.params));
struct type **argtypes = NULL;
type = convert_ast_to_type (state, operation->left.op);
if (!args.empty ())
argtypes = args.data ();
result
= lookup_function_type_with_arguments (type, args.size (),
argtypes);
result = lookup_pointer_type (result);
}
break;
case TYPE_CODE_STRUCT:
{
std::vector<struct type *> args
(convert_params_to_types (state, operation->left.params));
int i;
const char *name;
obstack_1grow (work_obstack, '(');
for (i = 0; i < args.size (); ++i)
{
std::string type_name = type_to_string (args[i]);
if (i > 0)
obstack_1grow (work_obstack, ',');
obstack_grow_str (work_obstack, type_name.c_str ());
}
obstack_grow_str0 (work_obstack, ")");
name = (const char *) obstack_finish (work_obstack);
/* We don't allow creating new tuple types (yet), but we do
allow looking up existing tuple types. */
result = rust_lookup_type (name, expression_context_block);
if (result == NULL)
error (_("could not find tuple type '%s'"), name);
}
break;
default:
gdb_assert_not_reached ("unhandled opcode in convert_ast_to_type");
}
gdb_assert (result != NULL);
return result;
}
/* A helper function to turn a rust_op representing a name into a full
name. This applies generic arguments as needed. The returned name
is allocated on the work obstack. */
static const char *
convert_name (struct parser_state *state, const struct rust_op *operation)
{
int i;
gdb_assert (operation->opcode == OP_VAR_VALUE);
if (operation->right.params == NULL)
return operation->left.sval.ptr;
std::vector<struct type *> types
(convert_params_to_types (state, operation->right.params));
obstack_grow_str (work_obstack, operation->left.sval.ptr);
obstack_1grow (work_obstack, '<');
for (i = 0; i < types.size (); ++i)
{
std::string type_name = type_to_string (types[i]);
if (i > 0)
obstack_1grow (work_obstack, ',');
obstack_grow_str (work_obstack, type_name.c_str ());
}
obstack_grow_str0 (work_obstack, ">");
return (const char *) obstack_finish (work_obstack);
}
static void convert_ast_to_expression (struct parser_state *state,
const struct rust_op *operation,
const struct rust_op *top,
bool want_type = false);
/* A helper function that converts a vec of rust_ops to a gdb
expression. */
static void
convert_params_to_expression (struct parser_state *state,
rust_op_vector *params,
const struct rust_op *top)
{
for (const rust_op *elem : *params)
convert_ast_to_expression (state, elem, top);
}
/* Lower a rust_op to a gdb expression. STATE is the parser state.
OPERATION is the operation to lower. TOP is a pointer to the
top-most operation; it is used to handle the special case where the
top-most expression is an identifier and can be optionally lowered
to OP_TYPE. WANT_TYPE is a flag indicating that, if the expression
is the name of a type, then emit an OP_TYPE for it (rather than
erroring). If WANT_TYPE is set, then the similar TOP handling is
not done. */
static void
convert_ast_to_expression (struct parser_state *state,
const struct rust_op *operation,
const struct rust_op *top,
bool want_type)
{
switch (operation->opcode)
{
case OP_LONG:
write_exp_elt_opcode (state, OP_LONG);
write_exp_elt_type (state, operation->left.typed_val_int.type);
write_exp_elt_longcst (state, operation->left.typed_val_int.val);
write_exp_elt_opcode (state, OP_LONG);
break;
Target FP: Use target format throughout expression parsing When parsing floating-point literals, the language parsers currently use parse_float or some equivalent routine to parse the input string into a DOUBLEST, which is then stored within a OP_DOUBLE expression node. When evaluating the expression, the OP_DOUBLE is finally converted into a value in target format. On the other hand, *decimal* floating-point literals are parsed directly into target format and stored that way in a OP_DECFLOAT expression node. In order to eliminate the DOUBLEST, this patch therefore unifies the handling of binary and decimal floating- point literals and stores them both in target format within a new OP_FLOAT expression node, replacing both OP_DOUBLE and OP_DECFLOAT. In order to store literals in target format, the parse_float routine needs to know the type of the literal. All parsers therefore need to be changed to determine the appropriate type (e.g. by detecting suffixes) *before* calling parse_float, instead of after it as today. However, this change is mostly straightforward -- again, this is already done for decimal FP today. The core of the literal parsing is moved into a new routine floatformat_from_string, mirroring floatformat_to_string. The parse_float routine now calls either floatformat_from_string or decimal_from_sting, allowing it to handle any type of FP literal. All language parsers need to be updated. Some notes on specific changes to the various languages: - C: Decimal FP is now handled in parse_float, and no longer needs to be handled specially. - D: Straightforward. - Fortran: Still used a hard-coded "atof", also replaced by parse_float now. Continues to always use builtin_real_s8 as the type of literal, even though this is probably wrong. - Go: This used to handle "f" and "l" suffixes, even though the Go language actually doesn't support those. I kept this support for now -- maybe revisit later. Note the the GDB test suite for some reason actually *verifies* that GDB supports those unsupported suffixes ... - Pascal: Likewise -- this handles suffixes that are not supported in the language standard. - Modula-2: Like Fortran, used to use "atof". - Rust: Mostly straightforward, except for a unit-testing hitch. The code use to set a special "unit_testing" flag which would cause "rust_type" to always return NULL. This makes it not possible to encode a literal into target format (which type?). The reason for this flag appears to have been that during unit testing, there is no "rust_parser" context set up, which means no "gdbarch" is available to use its types. To fix this, I removed the unit_testing flag, and instead simply just set up a dummy rust_parser context during unit testing. - Ada: This used to check sizeof (DOUBLEST) to determine which type to use for floating-point literal. This seems questionable to begin with (since DOUBLEST is quite unrelated to target formats), and in any case we need to get rid of DOUBLEST. I'm now simply always using the largest type (builtin_long_double). gdb/ChangeLog: 2017-10-25 Ulrich Weigand <uweigand@de.ibm.com> * doublest.c (floatformat_from_string): New function. * doublest.h (floatformat_from_string): Add prototype. * std-operator.def (OP_DOUBLE, OP_DECFLOAT): Remove, replace by ... (OP_FLOAT): ... this. * expression.h: Do not include "doublest.h". (union exp_element): Replace doubleconst and decfloatconst by new element floatconst. * ada-lang.c (resolve_subexp): Handle OP_FLOAT instead of OP_DOUBLE. (ada_evaluate_subexp): Likewise. * eval.c (evaluate_subexp_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. * expprint.c (print_subexp_standard): Likewise. (dump_subexp_body_standard): Likewise. * breakpoint.c (watchpoint_exp_is_const): Likewise. * parse.c: Include "dfp.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): New function. (operator_length_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. (operator_check_standard): Likewise. (parse_float): Do not accept suffix. Take type as input. Return bool. Return target format buffer instead of host DOUBLEST. Use floatformat_from_string and decimal_from_string to parse either binary or decimal floating-point types. (parse_c_float): Remove. * parser-defs.h: Do not include "doublest.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): Add prototype. (parse_float): Update prototype. (parse_c_float): Remove. * c-exp.y: Do not include "dfp.h". (typed_val_float): Use byte buffer instead of DOUBLEST. (typed_val_decfloat): Remove. (DECFLOAT): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. Handle decimal and binary FP types the same way. * d-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT_LITERAL): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * f-exp.y: Replace dval by typed_val_float. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Use parse_float instead of atof. * go-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (parse_go_float): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of parse_go_float. Parse suffixes and determine type before calling parse_float. * p-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * m2-exp.y: Replace dval by byte buffer val. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of atof. * rust-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (lex_number): Call parse_float instead of strtod. (ast_dliteral): Use OP_FLOAT instead of OP_DOUBLE. (convert_ast_to_expression): Handle OP_FLOAT instead of OP_DOUBLE. Use write_exp_elt_floatcst. (unit_testing): Remove static variable. (rust_type): Do not check unit_testing. (rust_lex_tests): Do not set uint_testing. Set up dummy rust_parser. * ada-exp.y (type_float, type_double): Remove. (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. * ada-lex.l (processReal): Use parse_float instead of sscanf.
2017-10-25 15:32:23 +02:00
case OP_FLOAT:
write_exp_elt_opcode (state, OP_FLOAT);
write_exp_elt_type (state, operation->left.typed_val_float.type);
Target FP: Use target format throughout expression parsing When parsing floating-point literals, the language parsers currently use parse_float or some equivalent routine to parse the input string into a DOUBLEST, which is then stored within a OP_DOUBLE expression node. When evaluating the expression, the OP_DOUBLE is finally converted into a value in target format. On the other hand, *decimal* floating-point literals are parsed directly into target format and stored that way in a OP_DECFLOAT expression node. In order to eliminate the DOUBLEST, this patch therefore unifies the handling of binary and decimal floating- point literals and stores them both in target format within a new OP_FLOAT expression node, replacing both OP_DOUBLE and OP_DECFLOAT. In order to store literals in target format, the parse_float routine needs to know the type of the literal. All parsers therefore need to be changed to determine the appropriate type (e.g. by detecting suffixes) *before* calling parse_float, instead of after it as today. However, this change is mostly straightforward -- again, this is already done for decimal FP today. The core of the literal parsing is moved into a new routine floatformat_from_string, mirroring floatformat_to_string. The parse_float routine now calls either floatformat_from_string or decimal_from_sting, allowing it to handle any type of FP literal. All language parsers need to be updated. Some notes on specific changes to the various languages: - C: Decimal FP is now handled in parse_float, and no longer needs to be handled specially. - D: Straightforward. - Fortran: Still used a hard-coded "atof", also replaced by parse_float now. Continues to always use builtin_real_s8 as the type of literal, even though this is probably wrong. - Go: This used to handle "f" and "l" suffixes, even though the Go language actually doesn't support those. I kept this support for now -- maybe revisit later. Note the the GDB test suite for some reason actually *verifies* that GDB supports those unsupported suffixes ... - Pascal: Likewise -- this handles suffixes that are not supported in the language standard. - Modula-2: Like Fortran, used to use "atof". - Rust: Mostly straightforward, except for a unit-testing hitch. The code use to set a special "unit_testing" flag which would cause "rust_type" to always return NULL. This makes it not possible to encode a literal into target format (which type?). The reason for this flag appears to have been that during unit testing, there is no "rust_parser" context set up, which means no "gdbarch" is available to use its types. To fix this, I removed the unit_testing flag, and instead simply just set up a dummy rust_parser context during unit testing. - Ada: This used to check sizeof (DOUBLEST) to determine which type to use for floating-point literal. This seems questionable to begin with (since DOUBLEST is quite unrelated to target formats), and in any case we need to get rid of DOUBLEST. I'm now simply always using the largest type (builtin_long_double). gdb/ChangeLog: 2017-10-25 Ulrich Weigand <uweigand@de.ibm.com> * doublest.c (floatformat_from_string): New function. * doublest.h (floatformat_from_string): Add prototype. * std-operator.def (OP_DOUBLE, OP_DECFLOAT): Remove, replace by ... (OP_FLOAT): ... this. * expression.h: Do not include "doublest.h". (union exp_element): Replace doubleconst and decfloatconst by new element floatconst. * ada-lang.c (resolve_subexp): Handle OP_FLOAT instead of OP_DOUBLE. (ada_evaluate_subexp): Likewise. * eval.c (evaluate_subexp_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. * expprint.c (print_subexp_standard): Likewise. (dump_subexp_body_standard): Likewise. * breakpoint.c (watchpoint_exp_is_const): Likewise. * parse.c: Include "dfp.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): New function. (operator_length_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. (operator_check_standard): Likewise. (parse_float): Do not accept suffix. Take type as input. Return bool. Return target format buffer instead of host DOUBLEST. Use floatformat_from_string and decimal_from_string to parse either binary or decimal floating-point types. (parse_c_float): Remove. * parser-defs.h: Do not include "doublest.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): Add prototype. (parse_float): Update prototype. (parse_c_float): Remove. * c-exp.y: Do not include "dfp.h". (typed_val_float): Use byte buffer instead of DOUBLEST. (typed_val_decfloat): Remove. (DECFLOAT): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. Handle decimal and binary FP types the same way. * d-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT_LITERAL): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * f-exp.y: Replace dval by typed_val_float. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Use parse_float instead of atof. * go-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (parse_go_float): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of parse_go_float. Parse suffixes and determine type before calling parse_float. * p-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * m2-exp.y: Replace dval by byte buffer val. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of atof. * rust-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (lex_number): Call parse_float instead of strtod. (ast_dliteral): Use OP_FLOAT instead of OP_DOUBLE. (convert_ast_to_expression): Handle OP_FLOAT instead of OP_DOUBLE. Use write_exp_elt_floatcst. (unit_testing): Remove static variable. (rust_type): Do not check unit_testing. (rust_lex_tests): Do not set uint_testing. Set up dummy rust_parser. * ada-exp.y (type_float, type_double): Remove. (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. * ada-lex.l (processReal): Use parse_float instead of sscanf.
2017-10-25 15:32:23 +02:00
write_exp_elt_floatcst (state, operation->left.typed_val_float.val);
write_exp_elt_opcode (state, OP_FLOAT);
break;
case STRUCTOP_STRUCT:
{
convert_ast_to_expression (state, operation->left.op, top);
if (operation->completing)
mark_struct_expression (state);
write_exp_elt_opcode (state, STRUCTOP_STRUCT);
write_exp_string (state, operation->right.sval);
write_exp_elt_opcode (state, STRUCTOP_STRUCT);
}
break;
case STRUCTOP_ANONYMOUS:
{
convert_ast_to_expression (state, operation->left.op, top);
write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
write_exp_elt_longcst (state, operation->right.typed_val_int.val);
write_exp_elt_opcode (state, STRUCTOP_ANONYMOUS);
}
break;
case UNOP_SIZEOF:
convert_ast_to_expression (state, operation->left.op, top, true);
write_exp_elt_opcode (state, UNOP_SIZEOF);
break;
case UNOP_PLUS:
case UNOP_NEG:
case UNOP_COMPLEMENT:
case UNOP_IND:
case UNOP_ADDR:
convert_ast_to_expression (state, operation->left.op, top);
write_exp_elt_opcode (state, operation->opcode);
break;
case BINOP_SUBSCRIPT:
case BINOP_MUL:
case BINOP_REPEAT:
case BINOP_DIV:
case BINOP_REM:
case BINOP_LESS:
case BINOP_GTR:
case BINOP_BITWISE_AND:
case BINOP_BITWISE_IOR:
case BINOP_BITWISE_XOR:
case BINOP_ADD:
case BINOP_SUB:
case BINOP_LOGICAL_OR:
case BINOP_LOGICAL_AND:
case BINOP_EQUAL:
case BINOP_NOTEQUAL:
case BINOP_LEQ:
case BINOP_GEQ:
case BINOP_LSH:
case BINOP_RSH:
case BINOP_ASSIGN:
case OP_RUST_ARRAY:
convert_ast_to_expression (state, operation->left.op, top);
convert_ast_to_expression (state, operation->right.op, top);
if (operation->compound_assignment)
{
write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
write_exp_elt_opcode (state, operation->opcode);
write_exp_elt_opcode (state, BINOP_ASSIGN_MODIFY);
}
else
write_exp_elt_opcode (state, operation->opcode);
if (operation->compound_assignment
|| operation->opcode == BINOP_ASSIGN)
{
struct type *type;
type = language_lookup_primitive_type (parse_language (state),
parse_gdbarch (state),
"()");
write_exp_elt_opcode (state, OP_LONG);
write_exp_elt_type (state, type);
write_exp_elt_longcst (state, 0);
write_exp_elt_opcode (state, OP_LONG);
write_exp_elt_opcode (state, BINOP_COMMA);
}
break;
case UNOP_CAST:
{
struct type *type = convert_ast_to_type (state, operation->right.op);
convert_ast_to_expression (state, operation->left.op, top);
write_exp_elt_opcode (state, UNOP_CAST);
write_exp_elt_type (state, type);
write_exp_elt_opcode (state, UNOP_CAST);
}
break;
case OP_FUNCALL:
{
if (operation->left.op->opcode == OP_VAR_VALUE)
{
struct type *type;
const char *varname = convert_name (state, operation->left.op);
type = rust_lookup_type (varname, expression_context_block);
if (type != NULL)
{
/* This is actually a tuple struct expression, not a
call expression. */
rust_op_vector *params = operation->right.params;
if (TYPE_CODE (type) != TYPE_CODE_NAMESPACE)
{
if (!rust_tuple_struct_type_p (type))
error (_("Type %s is not a tuple struct"), varname);
for (int i = 0; i < params->size (); ++i)
{
char *cell = get_print_cell ();
xsnprintf (cell, PRINT_CELL_SIZE, "__%d", i);
write_exp_elt_opcode (state, OP_NAME);
write_exp_string (state, make_stoken (cell));
write_exp_elt_opcode (state, OP_NAME);
convert_ast_to_expression (state, (*params)[i], top);
}
write_exp_elt_opcode (state, OP_AGGREGATE);
write_exp_elt_type (state, type);
write_exp_elt_longcst (state, 2 * params->size ());
write_exp_elt_opcode (state, OP_AGGREGATE);
break;
}
}
}
convert_ast_to_expression (state, operation->left.op, top);
convert_params_to_expression (state, operation->right.params, top);
write_exp_elt_opcode (state, OP_FUNCALL);
write_exp_elt_longcst (state, operation->right.params->size ());
write_exp_elt_longcst (state, OP_FUNCALL);
}
break;
case OP_ARRAY:
gdb_assert (operation->left.op == NULL);
convert_params_to_expression (state, operation->right.params, top);
write_exp_elt_opcode (state, OP_ARRAY);
write_exp_elt_longcst (state, 0);
write_exp_elt_longcst (state, operation->right.params->size () - 1);
write_exp_elt_longcst (state, OP_ARRAY);
break;
case OP_VAR_VALUE:
{
struct block_symbol sym;
const char *varname;
if (operation->left.sval.ptr[0] == '$')
{
write_dollar_variable (state, operation->left.sval);
break;
}
varname = convert_name (state, operation);
sym = rust_lookup_symbol (varname, expression_context_block,
VAR_DOMAIN);
if (sym.symbol != NULL && SYMBOL_CLASS (sym.symbol) != LOC_TYPEDEF)
{
write_exp_elt_opcode (state, OP_VAR_VALUE);
write_exp_elt_block (state, sym.block);
write_exp_elt_sym (state, sym.symbol);
write_exp_elt_opcode (state, OP_VAR_VALUE);
}
else
{
struct type *type = NULL;
if (sym.symbol != NULL)
{
gdb_assert (SYMBOL_CLASS (sym.symbol) == LOC_TYPEDEF);
type = SYMBOL_TYPE (sym.symbol);
}
if (type == NULL)
type = rust_lookup_type (varname, expression_context_block);
if (type == NULL)
error (_("No symbol '%s' in current context"), varname);
if (!want_type
&& TYPE_CODE (type) == TYPE_CODE_STRUCT
&& TYPE_NFIELDS (type) == 0)
{
/* A unit-like struct. */
write_exp_elt_opcode (state, OP_AGGREGATE);
write_exp_elt_type (state, type);
write_exp_elt_longcst (state, 0);
write_exp_elt_opcode (state, OP_AGGREGATE);
}
else if (want_type || operation == top)
{
write_exp_elt_opcode (state, OP_TYPE);
write_exp_elt_type (state, type);
write_exp_elt_opcode (state, OP_TYPE);
}
else
error (_("Found type '%s', which can't be "
"evaluated in this context"),
varname);
}
}
break;
case OP_AGGREGATE:
{
int length;
rust_set_vector *fields = operation->right.field_inits;
struct type *type;
const char *name;
length = 0;
for (const set_field &init : *fields)
{
if (init.name.ptr != NULL)
{
write_exp_elt_opcode (state, OP_NAME);
write_exp_string (state, init.name);
write_exp_elt_opcode (state, OP_NAME);
++length;
}
convert_ast_to_expression (state, init.init, top);
++length;
if (init.name.ptr == NULL)
{
/* This is handled differently from Ada in our
evaluator. */
write_exp_elt_opcode (state, OP_OTHERS);
}
}
name = convert_name (state, operation->left.op);
type = rust_lookup_type (name, expression_context_block);
if (type == NULL)
error (_("Could not find type '%s'"), operation->left.sval.ptr);
if (TYPE_CODE (type) != TYPE_CODE_STRUCT
|| rust_tuple_type_p (type)
|| rust_tuple_struct_type_p (type))
error (_("Struct expression applied to non-struct type"));
write_exp_elt_opcode (state, OP_AGGREGATE);
write_exp_elt_type (state, type);
write_exp_elt_longcst (state, length);
write_exp_elt_opcode (state, OP_AGGREGATE);
}
break;
case OP_STRING:
{
write_exp_elt_opcode (state, OP_STRING);
write_exp_string (state, operation->left.sval);
write_exp_elt_opcode (state, OP_STRING);
}
break;
case OP_RANGE:
{
enum range_type kind = BOTH_BOUND_DEFAULT;
if (operation->left.op != NULL)
{
convert_ast_to_expression (state, operation->left.op, top);
kind = HIGH_BOUND_DEFAULT;
}
if (operation->right.op != NULL)
{
convert_ast_to_expression (state, operation->right.op, top);
if (kind == BOTH_BOUND_DEFAULT)
kind = (operation->inclusive
? LOW_BOUND_DEFAULT : LOW_BOUND_DEFAULT_EXCLUSIVE);
else
{
gdb_assert (kind == HIGH_BOUND_DEFAULT);
kind = (operation->inclusive
? NONE_BOUND_DEFAULT : NONE_BOUND_DEFAULT_EXCLUSIVE);
}
}
else
{
/* Nothing should make an inclusive range without an upper
bound. */
gdb_assert (!operation->inclusive);
}
write_exp_elt_opcode (state, OP_RANGE);
write_exp_elt_longcst (state, kind);
write_exp_elt_opcode (state, OP_RANGE);
}
break;
default:
gdb_assert_not_reached ("unhandled opcode in convert_ast_to_expression");
}
}
/* The parser as exposed to gdb. */
int
rust_parse (struct parser_state *state)
{
int result;
/* This sets various globals and also clears them on
destruction. */
rust_parser parser (state);
Eliminate make_cleanup_obstack_free, introduce auto_obstack This commit eliminates make_cleanup_obstack_free, replacing it with a new auto_obstack type that inherits obstack to add cdtors. These changes in the parsers may not be obvious: - obstack_init (&name_obstack); - make_cleanup_obstack_free (&name_obstack); + name_obstack.clear (); Here, the 'name_obstack' variable is a global. The change means that the obstack's contents from a previous parse will stay around until the next parsing starts. I.e., memory won't be reclaimed until then. I don't think that's a problem, these objects don't really grow much at all. The other option I tried was to add a separate type that is like auto_obstack but manages an external obstack, just for those cases. I like the current approach better as that other approach adds more boilerplate and yet another type to learn. gdb/ChangeLog: 2017-06-27 Pedro Alves <palves@redhat.com> * c-exp.y (name_obstack): Now an auto_obstack. (yylex): Use auto_obstack::clear. (c_parse): Use auto_obstack::clear instead of reinitializing and freeing the obstack. * c-lang.c (evaluate_subexp_c): Use auto_obstack. * d-exp.y (name_obstack): Now an auto_obstack. (yylex): Use auto_obstack::clear. (d_parse): Use auto_obstack::clear instead of reinitializing and freeing the obstack. * dwarf2loc.c (fetch_const_value_from_synthetic_pointer): Use auto_obstack. * dwarf2read.c (create_addrmap_from_index) (dwarf2_build_psymtabs_hard) (update_enumeration_type_from_children): Likewise. * gdb_obstack.h (auto_obstack): New type. * go-exp.y (name_obstack): Now an auto_obstack. (build_packaged_name): Use auto_obstack::clear. (go_parse): Use auto_obstack::clear instead of reinitializing and freeing the obstack. * linux-tdep.c (linux_make_mappings_corefile_notes): Use auto_obstack. * printcmd.c (printf_wide_c_string, ui_printf): Use auto_obstack. * rust-exp.y (work_obstack): Now an auto_obstack. (rust_parse, rust_lex_tests): Use auto_obstack::clear instead of reinitializing and freeing the obstack. * utils.c (do_obstack_free, make_cleanup_obstack_free): Delete. (host_char_to_target): Use auto_obstack. * utils.h (make_cleanup_obstack_free): Delete declaration. * valprint.c (generic_emit_char, generic_printstr): Use auto_obstack.
2017-06-27 12:07:14 +02:00
result = rustyyparse ();
if (!result || (parse_completion && parser.rust_ast != NULL))
convert_ast_to_expression (state, parser.rust_ast, parser.rust_ast);
return result;
}
/* The parser error handler. */
void
-Wwrite-strings: The Rest This is the remainder boring constification that all looks more of less borderline obvious IMO. gdb/ChangeLog: 2017-04-05 Pedro Alves <palves@redhat.com> * ada-exp.y (yyerror): Constify. * ada-lang.c (bound_name, get_selections) (ada_variant_discrim_type) (ada_variant_discrim_name, ada_value_struct_elt) (ada_lookup_struct_elt_type, is_unchecked_variant) (ada_which_variant_applies, standard_exc, ada_get_next_arg) (catch_ada_exception_command_split) (catch_ada_assert_command_split, catch_assert_command) (ada_op_name): Constify. * ada-lang.h (ada_yyerror, get_selections) (ada_variant_discrim_name, ada_value_struct_elt): Constify. * arc-tdep.c (arc_print_frame_cache): Constify. * arm-tdep.c (arm_skip_stub): Constify. * ax-gdb.c (gen_binop, gen_struct_ref_recursive, gen_struct_ref) (gen_aggregate_elt_ref): Constify. * bcache.c (print_bcache_statistics): Constify. * bcache.h (print_bcache_statistics): Constify. * break-catch-throw.c (catch_exception_command_1): * breakpoint.c (struct ep_type_description::description): Constify. (add_solib_catchpoint): Constify. (catch_fork_command_1): Add cast. (add_catch_command): Constify. * breakpoint.h (add_catch_command, add_solib_catchpoint): Constify. * bsd-uthread.c (bsd_uthread_state): Constify. * buildsym.c (patch_subfile_names): Constify. * buildsym.h (next_symbol_text_func, patch_subfile_names): Constify. * c-exp.y (yyerror): Constify. (token::oper): Constify. * c-lang.h (c_yyerror, cp_print_class_member): Constify. * c-varobj.c (cplus_describe_child): Constify. * charset.c (find_charset_names): Add cast. (find_charset_names): Constify array and add const_cast. * cli/cli-cmds.c (complete_command, cd_command): Constify. (edit_command): Constify. * cli/cli-decode.c (lookup_cmd): Constify. * cli/cli-dump.c (dump_memory_command, dump_value_command): Constify. (struct dump_context): Constify. (add_dump_command, restore_command): Constify. * cli/cli-script.c (get_command_line): Constify. * cli/cli-script.h (get_command_line): Constify. * cli/cli-utils.c (check_for_argument): Constify. * cli/cli-utils.h (check_for_argument): Constify. * coff-pe-read.c (struct read_pe_section_data): Constify. * command.h (lookup_cmd): Constify. * common/print-utils.c (decimal2str): Constify. * completer.c (gdb_print_filename): Constify. * corefile.c (set_gnutarget): Constify. * cp-name-parser.y (yyerror): Constify. * cp-valprint.c (cp_print_class_member): Constify. * cris-tdep.c (cris_register_name, crisv32_register_name): Constify. * d-exp.y (yyerror): Constify. (struct token::oper): Constify. * d-lang.h (d_yyerror): Constify. * dbxread.c (struct header_file_location::name): Constify. (add_old_header_file, add_new_header_file, last_function_name) (dbx_next_symbol_text, add_bincl_to_list) (find_corresponding_bincl_psymtab, set_namestring) (find_stab_function_addr, read_dbx_symtab, start_psymtab) (dbx_end_psymtab, read_ofile_symtab, process_one_symbol): * defs.h (command_line_input, print_address_symbolic) (deprecated_readline_begin_hook): Constify. * dwarf2read.c (anonymous_struct_prefix, dwarf_bool_name): Constify. * event-top.c (handle_line_of_input): Constify and add cast. * exceptions.c (catch_errors): Constify. * exceptions.h (catch_errors): Constify. * expprint.c (print_subexp_standard, op_string, op_name) (op_name_standard, dump_raw_expression, dump_raw_expression): * expression.h (op_name, op_string, dump_raw_expression): Constify. * f-exp.y (yyerror): Constify. (struct token::oper): Constify. (struct f77_boolean_val::name): Constify. * f-lang.c (f_word_break_characters): Constify. * f-lang.h (f_yyerror): Constify. * fork-child.c (fork_inferior): Add cast. * frv-tdep.c (struct gdbarch_tdep::register_names): Constify. (new_variant): Constify. * gdbarch.sh (pstring_ptr, pstring_list): Constify. * gdbarch.c: Regenerate. * gdbcore.h (set_gnutarget): Constify. * go-exp.y (yyerror): Constify. (token::oper): Constify. * go-lang.h (go_yyerror): Constify. * go32-nat.c (go32_sysinfo): Constify. * guile/scm-breakpoint.c (gdbscm_breakpoint_expression): Constify. * guile/scm-cmd.c (cmdscm_function): Constify. * guile/scm-param.c (pascm_param_value): Constify. * h8300-tdep.c (h8300_register_name, h8300s_register_name) (h8300sx_register_name): Constify. * hppa-tdep.c (hppa32_register_name, hppa64_register_name): Constify. * ia64-tdep.c (ia64_register_names): Constify. * infcmd.c (construct_inferior_arguments): Constify. (path_command, attach_post_wait): Constify. * language.c (show_range_command, show_case_command) (unk_lang_error): Constify. * language.h (language_defn::la_error) (language_defn::la_name_of_this): Constify. * linespec.c (decode_line_2): Constify. * linux-thread-db.c (thread_db_err_str): Constify. * lm32-tdep.c (lm32_register_name): Constify. * m2-exp.y (yyerror): Constify. * m2-lang.h (m2_yyerror): Constify. * m32r-tdep.c (m32r_register_names): Constify and make static. * m68hc11-tdep.c (m68hc11_register_names): Constify. * m88k-tdep.c (m88k_register_name): Constify. * macroexp.c (appendmem): Constify. * mdebugread.c (fdr_name, add_data_symbol, parse_type) (upgrade_type, parse_external, parse_partial_symbols) (mdebug_next_symbol_text, cross_ref, mylookup_symbol, new_psymtab) (new_symbol): Constify. * memattr.c (mem_info_command): Constify. * mep-tdep.c (register_name_from_keyword): Constify. * mi/mi-cmd-env.c (mi_cmd_env_path, _initialize_mi_cmd_env): Constify. * mi/mi-cmd-stack.c (list_args_or_locals): Constify. * mi/mi-cmd-var.c (mi_cmd_var_show_attributes): Constify. * mi/mi-main.c (captured_mi_execute_command): Constify and add cast. (mi_execute_async_cli_command): Constify. * mips-tdep.c (mips_register_name): Constify. * mn10300-tdep.c (register_name, mn10300_generic_register_name) (am33_register_name, am33_2_register_name) * moxie-tdep.c (moxie_register_names): Constify. * nat/linux-osdata.c (osdata_type): Constify fields. * nto-tdep.c (nto_parse_redirection): Constify. * objc-lang.c (lookup_struct_typedef, lookup_objc_class) (lookup_child_selector): Constify. (objc_methcall::name): Constify. * objc-lang.h (lookup_objc_class, lookup_child_selector) (lookup_struct_typedef): Constify. * objfiles.c (pc_in_section): Constify. * objfiles.h (pc_in_section): Constify. * p-exp.y (struct token::oper): Constify. (yyerror): Constify. * p-lang.h (pascal_yyerror): Constify. * parser-defs.h (op_name_standard): Constify. (op_print::string): Constify. (exp_descriptor::op_name): Constify. * printcmd.c (print_address_symbolic): Constify. * psymtab.c (print_partial_symbols): Constify. * python/py-breakpoint.c (stop_func): Constify. (bppy_get_expression): Constify. * python/py-cmd.c (cmdpy_completer::name): Constify. (cmdpy_function): Constify. * python/py-event.c (evpy_add_attribute) (gdbpy_initialize_event_generic): Constify. * python/py-event.h (evpy_add_attribute) (gdbpy_initialize_event_generic): Constify. * python/py-evts.c (add_new_registry): Constify. * python/py-finishbreakpoint.c (outofscope_func): Constify. * python/py-framefilter.c (get_py_iter_from_func): Constify. * python/py-inferior.c (get_buffer): Add cast. * python/py-param.c (parm_constant::name): Constify. * python/py-unwind.c (fprint_frame_id): Constify. * python/python.c (gdbpy_parameter_value): Constify. * remote-fileio.c (remote_fio_func_map): Make 'name' const. * remote.c (memory_packet_config::name): Constify. (show_packet_config_cmd, remote_write_bytes) (remote_buffer_add_string): * reverse.c (exec_reverse_once): Constify. * rs6000-tdep.c (variant::name, variant::description): Constify. * rust-exp.y (rustyyerror): Constify. * rust-lang.c (rust_op_name): Constify. * rust-lang.h (rustyyerror): Constify. * serial.h (serial_ops::name): Constify. * sh-tdep.c (sh_sh_register_name, sh_sh3_register_name) (sh_sh3e_register_name, sh_sh2e_register_name) (sh_sh2a_register_name, sh_sh2a_nofpu_register_name) (sh_sh_dsp_register_name, sh_sh3_dsp_register_name) (sh_sh4_register_name, sh_sh4_nofpu_register_name) (sh_sh4al_dsp_register_name): Constify. * sh64-tdep.c (sh64_register_name): Constify. * solib-darwin.c (lookup_symbol_from_bfd): Constify. * spu-tdep.c (spu_register_name, info_spu_dma_cmdlist): Constify. * stabsread.c (patch_block_stabs, read_type_number) (ref_map::stabs, ref_add, process_reference) (symbol_reference_defined, define_symbol, define_symbol) (error_type, read_type, read_member_functions, read_cpp_abbrev) (read_one_struct_field, read_struct_fields, read_baseclasses) (read_tilde_fields, read_struct_type, read_array_type) (read_enum_type, read_sun_builtin_type, read_sun_floating_type) (read_huge_number, read_range_type, read_args, common_block_start) (find_name_end): Constify. * stabsread.h (common_block_start, define_symbol) (process_one_symbol, symbol_reference_defined, ref_add): * symfile.c (get_section_index, add_symbol_file_command): * symfile.h (get_section_index): Constify. * target-descriptions.c (tdesc_type::name): Constify. (tdesc_free_type): Add cast. * target.c (find_default_run_target): (add_deprecated_target_alias, find_default_run_target) (target_announce_detach): Constify. (do_option): Constify. * target.h (add_deprecated_target_alias): Constify. * thread.c (print_thread_info_1): Constify. * top.c (deprecated_readline_begin_hook, command_line_input): Constify. (init_main): Add casts. * top.h (handle_line_of_input): Constify. * tracefile-tfile.c (tfile_write_uploaded_tsv): Constify. * tracepoint.c (tvariables_info_1, trace_status_mi): Constify. (tfind_command): Rename to ... (tfind_command_1): ... this and constify. (tfind_command): New function. (tfind_end_command, tfind_start_command): Adjust. (encode_source_string): Constify. * tracepoint.h (encode_source_string): Constify. * tui/tui-data.c (tui_partial_win_by_name): Constify. * tui/tui-data.h (tui_partial_win_by_name): Constify. * tui/tui-source.c (tui_set_source_content_nil): Constify. * tui/tui-source.h (tui_set_source_content_nil): Constify. * tui/tui-win.c (parse_scrolling_args): Constify. * tui/tui-windata.c (tui_erase_data_content): Constify. * tui/tui-windata.h (tui_erase_data_content): Constify. * tui/tui-winsource.c (tui_erase_source_content): Constify. * tui/tui.c (tui_enable): Add cast. * utils.c (defaulted_query): Constify. (init_page_info): Add cast. (puts_debug, subset_compare): Constify. * utils.h (subset_compare): Constify. * varobj.c (varobj_format_string): Constify. * varobj.h (varobj_format_string): Constify. * vax-tdep.c (vax_register_name): Constify. * windows-nat.c (windows_detach): Constify. * xcoffread.c (process_linenos, xcoff_next_symbol_text): Constify. * xml-support.c (gdb_xml_end_element): Constify. * xml-tdesc.c (tdesc_start_reg): Constify. * xstormy16-tdep.c (xstormy16_register_name): Constify. * xtensa-tdep.c (xtensa_find_register_by_name): Constify. * xtensa-tdep.h (xtensa_register_t::name): Constify. gdb/gdbserver/ChangeLog: 2017-04-05 Pedro Alves <palves@redhat.com> * gdbreplay.c (sync_error): Constify. * linux-x86-low.c (push_opcode): Constify.
2017-04-05 20:21:37 +02:00
rustyyerror (const char *msg)
{
const char *where = prev_lexptr ? prev_lexptr : lexptr;
error (_("%s in expression, near `%s'."), (msg ? msg : "Error"), where);
}
#if GDB_SELF_TEST
/* Initialize the lexer for testing. */
static void
rust_lex_test_init (const char *input)
{
prev_lexptr = NULL;
lexptr = input;
paren_depth = 0;
}
/* A test helper that lexes a string, expecting a single token. It
returns the lexer data for this token. */
static RUSTSTYPE
rust_lex_test_one (const char *input, int expected)
{
int token;
RUSTSTYPE result;
rust_lex_test_init (input);
token = rustyylex ();
SELF_CHECK (token == expected);
result = rustyylval;
if (token)
{
token = rustyylex ();
SELF_CHECK (token == 0);
}
return result;
}
/* Test that INPUT lexes as the integer VALUE. */
static void
rust_lex_int_test (const char *input, int value, int kind)
{
RUSTSTYPE result = rust_lex_test_one (input, kind);
SELF_CHECK (result.typed_val_int.val == value);
}
/* Test that INPUT throws an exception with text ERR. */
static void
rust_lex_exception_test (const char *input, const char *err)
{
TRY
{
/* The "kind" doesn't matter. */
rust_lex_test_one (input, DECIMAL_INTEGER);
SELF_CHECK (0);
}
CATCH (except, RETURN_MASK_ERROR)
{
SELF_CHECK (strcmp (except.message, err) == 0);
}
END_CATCH
}
/* Test that INPUT lexes as the identifier, string, or byte-string
VALUE. KIND holds the expected token kind. */
static void
rust_lex_stringish_test (const char *input, const char *value, int kind)
{
RUSTSTYPE result = rust_lex_test_one (input, kind);
SELF_CHECK (result.sval.length == strlen (value));
SELF_CHECK (strncmp (result.sval.ptr, value, result.sval.length) == 0);
}
/* Helper to test that a string parses as a given token sequence. */
static void
rust_lex_test_sequence (const char *input, int len, const int expected[])
{
int i;
lexptr = input;
paren_depth = 0;
for (i = 0; i < len; ++i)
{
int token = rustyylex ();
SELF_CHECK (token == expected[i]);
}
}
/* Tests for an integer-parsing corner case. */
static void
rust_lex_test_trailing_dot (void)
{
const int expected1[] = { DECIMAL_INTEGER, '.', IDENT, '(', ')', 0 };
const int expected2[] = { INTEGER, '.', IDENT, '(', ')', 0 };
const int expected3[] = { FLOAT, EQEQ, '(', ')', 0 };
const int expected4[] = { DECIMAL_INTEGER, DOTDOT, DECIMAL_INTEGER, 0 };
rust_lex_test_sequence ("23.g()", ARRAY_SIZE (expected1), expected1);
rust_lex_test_sequence ("23_0.g()", ARRAY_SIZE (expected2), expected2);
rust_lex_test_sequence ("23.==()", ARRAY_SIZE (expected3), expected3);
rust_lex_test_sequence ("23..25", ARRAY_SIZE (expected4), expected4);
}
/* Tests of completion. */
static void
rust_lex_test_completion (void)
{
const int expected[] = { IDENT, '.', COMPLETE, 0 };
parse_completion = 1;
rust_lex_test_sequence ("something.wha", ARRAY_SIZE (expected), expected);
rust_lex_test_sequence ("something.", ARRAY_SIZE (expected), expected);
parse_completion = 0;
}
/* Test pushback. */
static void
rust_lex_test_push_back (void)
{
int token;
rust_lex_test_init (">>=");
token = rustyylex ();
SELF_CHECK (token == COMPOUND_ASSIGN);
SELF_CHECK (rustyylval.opcode == BINOP_RSH);
rust_push_back ('=');
token = rustyylex ();
SELF_CHECK (token == '=');
token = rustyylex ();
SELF_CHECK (token == 0);
}
/* Unit test the lexer. */
static void
rust_lex_tests (void)
{
int i;
auto_obstack test_obstack;
scoped_restore obstack_holder = make_scoped_restore (&work_obstack,
&test_obstack);
Target FP: Use target format throughout expression parsing When parsing floating-point literals, the language parsers currently use parse_float or some equivalent routine to parse the input string into a DOUBLEST, which is then stored within a OP_DOUBLE expression node. When evaluating the expression, the OP_DOUBLE is finally converted into a value in target format. On the other hand, *decimal* floating-point literals are parsed directly into target format and stored that way in a OP_DECFLOAT expression node. In order to eliminate the DOUBLEST, this patch therefore unifies the handling of binary and decimal floating- point literals and stores them both in target format within a new OP_FLOAT expression node, replacing both OP_DOUBLE and OP_DECFLOAT. In order to store literals in target format, the parse_float routine needs to know the type of the literal. All parsers therefore need to be changed to determine the appropriate type (e.g. by detecting suffixes) *before* calling parse_float, instead of after it as today. However, this change is mostly straightforward -- again, this is already done for decimal FP today. The core of the literal parsing is moved into a new routine floatformat_from_string, mirroring floatformat_to_string. The parse_float routine now calls either floatformat_from_string or decimal_from_sting, allowing it to handle any type of FP literal. All language parsers need to be updated. Some notes on specific changes to the various languages: - C: Decimal FP is now handled in parse_float, and no longer needs to be handled specially. - D: Straightforward. - Fortran: Still used a hard-coded "atof", also replaced by parse_float now. Continues to always use builtin_real_s8 as the type of literal, even though this is probably wrong. - Go: This used to handle "f" and "l" suffixes, even though the Go language actually doesn't support those. I kept this support for now -- maybe revisit later. Note the the GDB test suite for some reason actually *verifies* that GDB supports those unsupported suffixes ... - Pascal: Likewise -- this handles suffixes that are not supported in the language standard. - Modula-2: Like Fortran, used to use "atof". - Rust: Mostly straightforward, except for a unit-testing hitch. The code use to set a special "unit_testing" flag which would cause "rust_type" to always return NULL. This makes it not possible to encode a literal into target format (which type?). The reason for this flag appears to have been that during unit testing, there is no "rust_parser" context set up, which means no "gdbarch" is available to use its types. To fix this, I removed the unit_testing flag, and instead simply just set up a dummy rust_parser context during unit testing. - Ada: This used to check sizeof (DOUBLEST) to determine which type to use for floating-point literal. This seems questionable to begin with (since DOUBLEST is quite unrelated to target formats), and in any case we need to get rid of DOUBLEST. I'm now simply always using the largest type (builtin_long_double). gdb/ChangeLog: 2017-10-25 Ulrich Weigand <uweigand@de.ibm.com> * doublest.c (floatformat_from_string): New function. * doublest.h (floatformat_from_string): Add prototype. * std-operator.def (OP_DOUBLE, OP_DECFLOAT): Remove, replace by ... (OP_FLOAT): ... this. * expression.h: Do not include "doublest.h". (union exp_element): Replace doubleconst and decfloatconst by new element floatconst. * ada-lang.c (resolve_subexp): Handle OP_FLOAT instead of OP_DOUBLE. (ada_evaluate_subexp): Likewise. * eval.c (evaluate_subexp_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. * expprint.c (print_subexp_standard): Likewise. (dump_subexp_body_standard): Likewise. * breakpoint.c (watchpoint_exp_is_const): Likewise. * parse.c: Include "dfp.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): New function. (operator_length_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. (operator_check_standard): Likewise. (parse_float): Do not accept suffix. Take type as input. Return bool. Return target format buffer instead of host DOUBLEST. Use floatformat_from_string and decimal_from_string to parse either binary or decimal floating-point types. (parse_c_float): Remove. * parser-defs.h: Do not include "doublest.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): Add prototype. (parse_float): Update prototype. (parse_c_float): Remove. * c-exp.y: Do not include "dfp.h". (typed_val_float): Use byte buffer instead of DOUBLEST. (typed_val_decfloat): Remove. (DECFLOAT): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. Handle decimal and binary FP types the same way. * d-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT_LITERAL): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * f-exp.y: Replace dval by typed_val_float. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Use parse_float instead of atof. * go-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (parse_go_float): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of parse_go_float. Parse suffixes and determine type before calling parse_float. * p-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * m2-exp.y: Replace dval by byte buffer val. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of atof. * rust-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (lex_number): Call parse_float instead of strtod. (ast_dliteral): Use OP_FLOAT instead of OP_DOUBLE. (convert_ast_to_expression): Handle OP_FLOAT instead of OP_DOUBLE. Use write_exp_elt_floatcst. (unit_testing): Remove static variable. (rust_type): Do not check unit_testing. (rust_lex_tests): Do not set uint_testing. Set up dummy rust_parser. * ada-exp.y (type_float, type_double): Remove. (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. * ada-lex.l (processReal): Use parse_float instead of sscanf.
2017-10-25 15:32:23 +02:00
// Set up dummy "parser", so that rust_type works.
struct parser_state ps (0, &rust_language_defn, target_gdbarch ());
Target FP: Use target format throughout expression parsing When parsing floating-point literals, the language parsers currently use parse_float or some equivalent routine to parse the input string into a DOUBLEST, which is then stored within a OP_DOUBLE expression node. When evaluating the expression, the OP_DOUBLE is finally converted into a value in target format. On the other hand, *decimal* floating-point literals are parsed directly into target format and stored that way in a OP_DECFLOAT expression node. In order to eliminate the DOUBLEST, this patch therefore unifies the handling of binary and decimal floating- point literals and stores them both in target format within a new OP_FLOAT expression node, replacing both OP_DOUBLE and OP_DECFLOAT. In order to store literals in target format, the parse_float routine needs to know the type of the literal. All parsers therefore need to be changed to determine the appropriate type (e.g. by detecting suffixes) *before* calling parse_float, instead of after it as today. However, this change is mostly straightforward -- again, this is already done for decimal FP today. The core of the literal parsing is moved into a new routine floatformat_from_string, mirroring floatformat_to_string. The parse_float routine now calls either floatformat_from_string or decimal_from_sting, allowing it to handle any type of FP literal. All language parsers need to be updated. Some notes on specific changes to the various languages: - C: Decimal FP is now handled in parse_float, and no longer needs to be handled specially. - D: Straightforward. - Fortran: Still used a hard-coded "atof", also replaced by parse_float now. Continues to always use builtin_real_s8 as the type of literal, even though this is probably wrong. - Go: This used to handle "f" and "l" suffixes, even though the Go language actually doesn't support those. I kept this support for now -- maybe revisit later. Note the the GDB test suite for some reason actually *verifies* that GDB supports those unsupported suffixes ... - Pascal: Likewise -- this handles suffixes that are not supported in the language standard. - Modula-2: Like Fortran, used to use "atof". - Rust: Mostly straightforward, except for a unit-testing hitch. The code use to set a special "unit_testing" flag which would cause "rust_type" to always return NULL. This makes it not possible to encode a literal into target format (which type?). The reason for this flag appears to have been that during unit testing, there is no "rust_parser" context set up, which means no "gdbarch" is available to use its types. To fix this, I removed the unit_testing flag, and instead simply just set up a dummy rust_parser context during unit testing. - Ada: This used to check sizeof (DOUBLEST) to determine which type to use for floating-point literal. This seems questionable to begin with (since DOUBLEST is quite unrelated to target formats), and in any case we need to get rid of DOUBLEST. I'm now simply always using the largest type (builtin_long_double). gdb/ChangeLog: 2017-10-25 Ulrich Weigand <uweigand@de.ibm.com> * doublest.c (floatformat_from_string): New function. * doublest.h (floatformat_from_string): Add prototype. * std-operator.def (OP_DOUBLE, OP_DECFLOAT): Remove, replace by ... (OP_FLOAT): ... this. * expression.h: Do not include "doublest.h". (union exp_element): Replace doubleconst and decfloatconst by new element floatconst. * ada-lang.c (resolve_subexp): Handle OP_FLOAT instead of OP_DOUBLE. (ada_evaluate_subexp): Likewise. * eval.c (evaluate_subexp_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. * expprint.c (print_subexp_standard): Likewise. (dump_subexp_body_standard): Likewise. * breakpoint.c (watchpoint_exp_is_const): Likewise. * parse.c: Include "dfp.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): New function. (operator_length_standard): Handle OP_FLOAT instead of OP_DOUBLE and OP_DECFLOAT. (operator_check_standard): Likewise. (parse_float): Do not accept suffix. Take type as input. Return bool. Return target format buffer instead of host DOUBLEST. Use floatformat_from_string and decimal_from_string to parse either binary or decimal floating-point types. (parse_c_float): Remove. * parser-defs.h: Do not include "doublest.h". (write_exp_elt_dblcst, write_exp_elt_decfloatcst): Remove. (write_exp_elt_floatcst): Add prototype. (parse_float): Update prototype. (parse_c_float): Remove. * c-exp.y: Do not include "dfp.h". (typed_val_float): Use byte buffer instead of DOUBLEST. (typed_val_decfloat): Remove. (DECFLOAT): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. Handle decimal and binary FP types the same way. * d-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT_LITERAL): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * f-exp.y: Replace dval by typed_val_float. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Use parse_float instead of atof. * go-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (parse_go_float): Remove. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of parse_go_float. Parse suffixes and determine type before calling parse_float. * p-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Update to new parse_float interface. Parse suffixes and determine type before calling parse_float. * m2-exp.y: Replace dval by byte buffer val. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. (parse_number): Call parse_float instead of atof. * rust-exp.y (typed_val_float): Use byte buffer instead of DOUBLEST. (lex_number): Call parse_float instead of strtod. (ast_dliteral): Use OP_FLOAT instead of OP_DOUBLE. (convert_ast_to_expression): Handle OP_FLOAT instead of OP_DOUBLE. Use write_exp_elt_floatcst. (unit_testing): Remove static variable. (rust_type): Do not check unit_testing. (rust_lex_tests): Do not set uint_testing. Set up dummy rust_parser. * ada-exp.y (type_float, type_double): Remove. (typed_val_float): Use byte buffer instead of DOUBLEST. (FLOAT): Use OP_FLOAT and write_exp_elt_floatcst. * ada-lex.l (processReal): Use parse_float instead of sscanf.
2017-10-25 15:32:23 +02:00
rust_parser parser (&ps);
rust_lex_test_one ("", 0);
rust_lex_test_one (" \t \n \r ", 0);
rust_lex_test_one ("thread 23", 0);
rust_lex_test_one ("task 23", 0);
rust_lex_test_one ("th 104", 0);
rust_lex_test_one ("ta 97", 0);
rust_lex_int_test ("'z'", 'z', INTEGER);
rust_lex_int_test ("'\\xff'", 0xff, INTEGER);
rust_lex_int_test ("'\\u{1016f}'", 0x1016f, INTEGER);
rust_lex_int_test ("b'z'", 'z', INTEGER);
rust_lex_int_test ("b'\\xfe'", 0xfe, INTEGER);
rust_lex_int_test ("b'\\xFE'", 0xfe, INTEGER);
rust_lex_int_test ("b'\\xfE'", 0xfe, INTEGER);
/* Test all escapes in both modes. */
rust_lex_int_test ("'\\n'", '\n', INTEGER);
rust_lex_int_test ("'\\r'", '\r', INTEGER);
rust_lex_int_test ("'\\t'", '\t', INTEGER);
rust_lex_int_test ("'\\\\'", '\\', INTEGER);
rust_lex_int_test ("'\\0'", '\0', INTEGER);
rust_lex_int_test ("'\\''", '\'', INTEGER);
rust_lex_int_test ("'\\\"'", '"', INTEGER);
rust_lex_int_test ("b'\\n'", '\n', INTEGER);
rust_lex_int_test ("b'\\r'", '\r', INTEGER);
rust_lex_int_test ("b'\\t'", '\t', INTEGER);
rust_lex_int_test ("b'\\\\'", '\\', INTEGER);
rust_lex_int_test ("b'\\0'", '\0', INTEGER);
rust_lex_int_test ("b'\\''", '\'', INTEGER);
rust_lex_int_test ("b'\\\"'", '"', INTEGER);
rust_lex_exception_test ("'z", "Unterminated character literal");
rust_lex_exception_test ("b'\\x0'", "Not enough hex digits seen");
rust_lex_exception_test ("b'\\u{0}'", "Unicode escape in byte literal");
rust_lex_exception_test ("'\\x0'", "Not enough hex digits seen");
rust_lex_exception_test ("'\\u0'", "Missing '{' in Unicode escape");
rust_lex_exception_test ("'\\u{0", "Missing '}' in Unicode escape");
rust_lex_exception_test ("'\\u{0000007}", "Overlong hex escape");
rust_lex_exception_test ("'\\u{}", "Not enough hex digits seen");
rust_lex_exception_test ("'\\Q'", "Invalid escape \\Q in literal");
rust_lex_exception_test ("b'\\Q'", "Invalid escape \\Q in literal");
rust_lex_int_test ("23", 23, DECIMAL_INTEGER);
rust_lex_int_test ("2_344__29", 234429, INTEGER);
rust_lex_int_test ("0x1f", 0x1f, INTEGER);
rust_lex_int_test ("23usize", 23, INTEGER);
rust_lex_int_test ("23i32", 23, INTEGER);
rust_lex_int_test ("0x1_f", 0x1f, INTEGER);
rust_lex_int_test ("0b1_101011__", 0x6b, INTEGER);
rust_lex_int_test ("0o001177i64", 639, INTEGER);
rust_lex_test_trailing_dot ();
rust_lex_test_one ("23.", FLOAT);
rust_lex_test_one ("23.99f32", FLOAT);
rust_lex_test_one ("23e7", FLOAT);
rust_lex_test_one ("23E-7", FLOAT);
rust_lex_test_one ("23e+7", FLOAT);
rust_lex_test_one ("23.99e+7f64", FLOAT);
rust_lex_test_one ("23.82f32", FLOAT);
rust_lex_stringish_test ("hibob", "hibob", IDENT);
rust_lex_stringish_test ("hibob__93", "hibob__93", IDENT);
rust_lex_stringish_test ("thread", "thread", IDENT);
rust_lex_stringish_test ("\"string\"", "string", STRING);
rust_lex_stringish_test ("\"str\\ting\"", "str\ting", STRING);
rust_lex_stringish_test ("\"str\\\"ing\"", "str\"ing", STRING);
rust_lex_stringish_test ("r\"str\\ing\"", "str\\ing", STRING);
rust_lex_stringish_test ("r#\"str\\ting\"#", "str\\ting", STRING);
rust_lex_stringish_test ("r###\"str\\\"ing\"###", "str\\\"ing", STRING);
rust_lex_stringish_test ("b\"string\"", "string", BYTESTRING);
rust_lex_stringish_test ("b\"\x73tring\"", "string", BYTESTRING);
rust_lex_stringish_test ("b\"str\\\"ing\"", "str\"ing", BYTESTRING);
rust_lex_stringish_test ("br####\"\\x73tring\"####", "\\x73tring",
BYTESTRING);
for (i = 0; i < ARRAY_SIZE (identifier_tokens); ++i)
rust_lex_test_one (identifier_tokens[i].name, identifier_tokens[i].value);
for (i = 0; i < ARRAY_SIZE (operator_tokens); ++i)
rust_lex_test_one (operator_tokens[i].name, operator_tokens[i].value);
rust_lex_test_completion ();
rust_lex_test_push_back ();
}
#endif /* GDB_SELF_TEST */
void
_initialize_rust_exp (void)
{
int code = regcomp (&number_regex, number_regex_text, REG_EXTENDED);
/* If the regular expression was incorrect, it was a programming
error. */
gdb_assert (code == 0);
#if GDB_SELF_TEST
Add selftests run filtering With the growing number of selftests, I think it would be useful to be able to run only a subset of the tests. This patch associates a name to each registered selftest. It then allows doing something like: (gdb) maintenance selftest aarch64 Running self-tests. Running selftest aarch64-analyze-prologue. Running selftest aarch64-process-record. Ran 2 unit tests, 0 failed or with gdbserver: ./gdbserver --selftest=aarch64 In both cases, only the tests that contain "aarch64" in their name are ran. To help validate that the tests you want to run were actually ran, it also prints a message with the test name before running each test. Right now, all the arch-dependent tests are registered as a single test of the selftests. To be able to filter those too, I made them "first-class citizen" selftests. The selftest type is an interface, with different implementations for "simple selftests" and "arch selftests". The run_tests function simply iterates on that an invokes operator() on each test. I changed the tests data structure from a vector to a map, because - it allows iterating in a stable (alphabetical) order - it allows to easily verify if a test with a given name has been registered, to avoid duplicates There's also a new command "maintenance info selftests" that lists the registered selftests. gdb/ChangeLog: * common/selftest.h (selftest): New struct/interface. (register_test): Add name parameter, add new overload. (run_tests): Add filter parameter. (for_each_selftest_ftype): New typedef. (for_each_selftest): New declaration. * common/selftest.c (tests): Change type to map<string, unique_ptr<selftest>>. (simple_selftest): New struct. (register_test): New function. (register_test): Add name parameter and use it. (run_tests): Add filter parameter and use it. Add prints. Adjust to vector -> map change. * aarch64-tdep.c (_initialize_aarch64_tdep): Add names when registering selftests. * arm-tdep.c (_initialize_arm_tdep): Likewise. * disasm-selftests.c (_initialize_disasm_selftests): Likewise. * dwarf2-frame.c (_initialize_dwarf2_frame): Likewise. * dwarf2loc.c (_initialize_dwarf2loc): Likewise. * findvar.c (_initialize_findvar): Likewise. * gdbarch-selftests.c (_initialize_gdbarch_selftests): Likewise. * maint.c (maintenance_selftest): Update call to run_tests. (maintenance_info_selftests): New function. (_initialize_maint_cmds): Register "maintenance info selftests" command. Update "maintenance selftest" doc. * regcache.c (_initialize_regcache): Add names when registering selftests. * rust-exp.y (_initialize_rust_exp): Likewise. * selftest-arch.c (gdbarch_selftest): New struct. (gdbarch_tests): Remove. (register_test_foreach_arch): Add name parameter. Call register_test. (tests_with_arch): Remove, move most content to gdbarch_selftest::operator(). (_initialize_selftests_foreach_arch): Remove. * selftest-arch.h (register_test_foreach_arch): Add name parameter. (run_tests_with_arch): New declaration. * utils-selftests.c (_initialize_utils_selftests): Add names when registering selftests. * utils.c (_initialize_utils): Likewise. * unittests/array-view-selftests.c (_initialize_array_view_selftests): Likewise. * unittests/environ-selftests.c (_initialize_environ_selftests): Likewise. * unittests/function-view-selftests.c (_initialize_function_view_selftests): Likewise. * unittests/offset-type-selftests.c (_initialize_offset_type_selftests): Likewise. * unittests/optional-selftests.c (_initialize_optional_selftests): Likewise. * unittests/scoped_restore-selftests.c (_initialize_scoped_restore_selftests): Likewise. * NEWS: Document "maintenance selftest" and "maint info selftests". gdb/gdbserver/ChangeLog: * server.c (captured_main): Accept argument for --selftest. Update run_tests call. * linux-x86-tdesc-selftest.c (initialize_low_tdesc): Add names when registering selftests. gdb/doc/ChangeLog: * gdb.texinfo (Maintenance Commands): Document filter parameter of "maint selftest". Document "maint info selftests" command.
2017-09-16 14:06:03 +02:00
selftests::register_test ("rust-lex", rust_lex_tests);
#endif
}