* [RFA 1/2] Linespec rewrite (update 2)
@ 2012-03-26 23:18 Keith Seitz
2012-03-27 1:45 ` asmwarrior
` (3 more replies)
0 siblings, 4 replies; 26+ messages in thread
From: Keith Seitz @ 2012-03-26 23:18 UTC (permalink / raw)
To: gdb-patches@sourceware.org ml
[-- Attachment #1: Type: text/plain, Size: 4667 bytes --]
Hi,
Tom requested a repost of this patch, so here it is. Here is a list of
differences between this version and the previously posted patch:
* skip_quote_char rewritten (Doug found a host of bugs in it)
* added backward compatibility for quote-enclosed linespecs (so no NEWS
entry needed)
* copy_token_string now uses replace_trailing_whitespace function from
cli/cli-utils.[ch]
* miscellaneous test fiddling (from other posts by Tom and Pedro)
* some new error checking/conditions added (bugs reported by Doug)
This patch is now essentially standalone, as Pedro requested.
[Previously I posted patches to be committed all together when finally
approved.]
Keith
ChangeLog
2012-03-26 Keith Seitz <keiths@redhat.com>
* linespec.c (decode_compound): Remove.
(enum offset_relative_sign): New enum.
(struct line_offset): New struct.
(struct linespec): New struct.
(struct linespec_state): Move file_symtabs,
user_filename, and user_function into struct linespec.
Make result an anonymous struct holding vectors of
symbolp and minsym_and_objfile_d.
(enum ls_token_type): New enum.
(linespec_keywords): New array.
(struct ls_token): New struct.
(struct ls_parser): New struct.
(linespec_lexer_lex_number): New function.
(linespec_lexer_lex_keyword): New function.
(is_ada_operator): New function.
(skip_quote_char): New function.
(copy_token_string): New function.
(linespec_lexer_lex_string): New function.
(linespec_lexer_lex_one): New function.
(linespec_lexer_consume_token): New function.
(linespec_lexer_peek_token): New function.
(cplusplus_error): Remove unused function.
(find_methods): Update comment.
(find_toplevel_char): Return const.
(is_objc_method_format): Remove unused function.
(find_toplevel_string): New function.
(is_linespec_boundary): Remove.
(symbol_not_found_error): New function.
(find_method_overload_end): Remove function.
(unexpected_linespec_error): New function.
(keep_name_info): Remove.
(linespec_parse_line_offset): New function.
(linespec_parse_basic): New function.
(canonicalize_linespec): New function.
(decode_line_internal): Remove.
(create_sals_line_offset): New function adapted from
decode_all_digits.
(convert_linespec_to_sals): New function.
(parse_linespec): New function.
(linespec_parser_new): New function.
(linespec_state_destructor): Change parameter type to
struct linespec_state *.
Remove freeing of moved members.
(linespec_parser_delete): New function.
(decode_line_full): Use parse_linespec and linespec_parser_new.
(decode_line_1): Likewise.
(decode_indirect): Rename to ...
(linespec_expression_to_pc): ... this and rewrite
to simply find CORE_ADDR, storing this result for later
conversion to SALs.
(locate_first_half): Remove.
(deocde_objc): Add parameter LS.
Initialize new struct collect_info members.
Handle minimal symbols, too.
(decode_compound): Delete.
(lookup_prefix_sym): Rewrite.
(compare_msymbols): New function.
(find_method): Rewrite.
Do not call cplusplus_error.
(symtabs_from_filename): Rewrite.
(collect_function_symbols): Delete.
(find_function_symbols): Rewrite without ARGPTR-style
processing.
(decode_all_digits): Delete. (Rewritten as create_sals_line_offset.)
(decode_dollar): Adapted and renamed to ...
(linespec_parse_variable): ... this.
(find_linespec_symbols): New function.
(decode_label): Adapted and renamed to ...
(find_label_symbols): ... this.
(decode_digits_list_mode): Add and use LS argument.
(decode_digits_ordinary): Likewise.
(collect_symbols): Do not collect SALs, just symbols and msymbols.
If in list mode, allow any symbol class. Otherwise, only
permit LOC_BLOCK symbols.
(minsym_found): Update comments.
(search_minsyms_for_name): Do not convert the matching symbol
into a SAL. Simply push the symbol and objfile into the
result vector.
(decode_variable): Delete. Contents adapted into
find_linespec_symbols.
testsuite/ChangeLog
2012-03-26 Keith Seitz <keiths@redhat.com>
* gdb.base/advance.exp: Update error message for
"advance malformed" test.
* gdb.base/break.exp: Likewise for "breakpoint with
trailing garbage" test.
* gdb.base/hbreak2.exp: Likewise for "hardware breakpoint
with trailing garbage" test.
* gdb.base/sepdebug.exp: Likewise for "breakpoint with
trailng garbage" test.
* gdb.base/until.exp: Likewise for "malformed until" test.
* gdb.cp/ovldbreak.exp: Create the breakpoint table
for "breakpoint info (after setting on all)".
* gdb.cp/userdef.exp: Remove quoting for "break A2::operator+"
tests.
* gdb.cp/cplabel.cc: New file.
* gdb.cp/cplabel.exp: New test.
* gdb.linespec/ls-errs.c: New file.
* gdb.linespec/ls-errs.exp: New test.
[-- Attachment #2: linespec-rewrite-parser-2.patch --]
[-- Type: text/x-patch, Size: 137481 bytes --]
diff --git a/gdb/linespec.c b/gdb/linespec.c
index 1e9770e..50cf2e0 100644
--- a/gdb/linespec.c
+++ b/gdb/linespec.c
@@ -63,6 +63,74 @@ struct address_entry
CORE_ADDR addr;
};
+/* A helper struct which just holds a minimal symbol and the object
+ file from which it came. */
+
+typedef struct minsym_and_objfile
+{
+ struct minimal_symbol *minsym;
+ struct objfile *objfile;
+} minsym_and_objfile_d;
+
+DEF_VEC_O (minsym_and_objfile_d);
+
+/* An enumeration of possible signs for a line offset. */
+enum offset_relative_sign
+{
+ /* No sign */
+ none,
+
+ /* A plus sign ("+") */
+ plus,
+
+ /* A minus sign ("-") */
+ minus,
+
+ /* A special "sign" for unspecified offset. */
+ unknown
+};
+
+/* A line offset in a linespec. */
+
+struct line_offset
+{
+ /* Line offset and any specified sign. */
+ int offset;
+ enum offset_relative_sign sign;
+};
+
+/* A linespec. Elements of this structure are filled in by a parser
+ (either parse_linespec or some other function). The structure is
+ then converted into SALs by convert_linespec_to_sals. */
+
+struct linespec
+{
+ /* An expression and the resulting PC. */
+ char *expression;
+ CORE_ADDR expr_pc;
+
+ /* Any specified file symtabs. */
+ char *source_filename;
+ VEC (symtab_p) *file_symtabs;
+
+ /* The name of a function or method and any matching symbols. */
+ char *function_name;
+ VEC (symbolp) *function_symbols;
+ VEC (minsym_and_objfile_d) *minimal_symbols;
+
+ /* The name of a label and matching symbols. */
+ char *label_name;
+ struct
+ {
+ VEC (symbolp) *label_symbols;
+ VEC (symbolp) *function_symbols;
+ } labels;
+
+ /* Line offset */
+ struct line_offset line_offset;
+};
+typedef struct linespec *linespec_t;
+
/* An instance of this is used to keep all state while linespec
operates. This instance is passed around as a 'this' pointer to
the various implementation methods. */
@@ -78,19 +146,6 @@ struct linespec_state
/* The default line to use. */
int default_line;
- /* If the linespec started with "FILE:", this holds all the matching
- symtabs. Otherwise, it will hold a single NULL entry, meaning
- that the default symtab should be used. */
- VEC (symtab_p) *file_symtabs;
-
- /* If the linespec started with "FILE:", this holds an xmalloc'd
- copy of "FILE". */
- char *user_filename;
-
- /* If the linespec is "FUNCTION:LABEL", this holds an xmalloc'd copy
- of "FUNCTION". */
- char *user_function;
-
/* The 'funfirstline' value that was passed in to decode_line_1 or
decode_line_full. */
int funfirstline;
@@ -117,67 +172,125 @@ struct collect_info
/* The linespec object in use. */
struct linespec_state *state;
+ /* A list of symtabs to which to restrict matches. */
+ VEC (symtab_p) *file_symtabs;
+
/* The result being accumulated. */
- struct symtabs_and_lines result;
+ struct
+ {
+ VEC (symbolp) *symbols;
+ VEC (minsym_and_objfile_d) *minimal_symbols;
+ } result;
};
-/* Prototypes for local functions. */
+/* Token types */
-static void initialize_defaults (struct symtab **default_symtab,
- int *default_line);
+enum ls_token_type
+{
+ /* A keyword */
+ LSTOKEN_KEYWORD = 0,
-static struct symtabs_and_lines decode_indirect (struct linespec_state *self,
- char **argptr);
+ /* A colon "separator" */
+ LSTOKEN_COLON,
-static char *locate_first_half (char **argptr, int *is_quote_enclosed);
+ /* A string */
+ LSTOKEN_STRING,
-static struct symtabs_and_lines decode_objc (struct linespec_state *self,
- char **argptr);
+ /* A number */
+ LSTOKEN_NUMBER,
+
+ /* EOI (end of input) */
+ LSTOKEN_EOI,
+
+ /* Consumed token */
+ LSTOKEN_CONSUMED
+};
+typedef enum ls_token_type linespec_token_type;
-static struct symtabs_and_lines decode_compound (struct linespec_state *self,
- char **argptr,
- char *saved_arg,
- char *p);
+/* List of keywords */
-static VEC (symbolp) *lookup_prefix_sym (char **argptr, char *p,
- VEC (symtab_p) *,
- char **);
+const char * const linespec_keywords[] = { "if", "thread", "task" };
+
+/* A token of the linespec lexer */
+
+struct ls_token
+{
+ /* The type of the token */
+ linespec_token_type type;
+
+ /* Data for the token */
+ union
+ {
+ /* A string, given as a stoken */
+ struct stoken string;
+
+ /* A keyword */
+ const char *keyword;
+ } data;
+};
+typedef struct ls_token linespec_token;
+
+#define LS_TOKEN_STOKEN(TOK) (TOK).data.string
+#define LS_TOKEN_KEYWORD(TOK) (TOK).data.keyword
+
+/* An instance of the linespec parser. */
+
+struct ls_parser
+{
+ /* Lexer internal data */
+ struct
+ {
+ /* Save head of input stream. */
+ char *saved_arg;
+
+ /* Head of the input stream. */
+ char **stream;
+#define PARSER_STREAM(P) (*(P)->lexer.stream)
+
+ /* The current token. */
+ linespec_token current;
+ } lexer;
+
+ /* Is the entire linespec quote-enclosed? */
+ int is_quote_enclosed;
-static struct symtabs_and_lines find_method (struct linespec_state *self,
- char *saved_arg,
- char *copy,
- const char *class_name,
- VEC (symbolp) *sym_classes);
+ /* The state of the parse. */
+ struct linespec_state state;
+#define PARSER_STATE(PPTR) (&(PPTR)->state)
-static void cplusplus_error (const char *name, const char *fmt, ...)
- ATTRIBUTE_NORETURN ATTRIBUTE_PRINTF (2, 3);
+ /* The result of the parse. */
+ struct linespec result;
+#define PARSER_RESULT(PPTR) (&(PPTR)->result)
+};
+typedef struct ls_parser linespec_parser;
-static char *find_toplevel_char (char *s, char c);
+/* Prototypes for local functions. */
-static int is_objc_method_format (const char *s);
+static void initialize_defaults (struct symtab **default_symtab,
+ int *default_line);
-static VEC (symtab_p) *symtabs_from_filename (char **argptr,
- char *p, int is_quote_enclosed,
- char **user_filename);
+static CORE_ADDR linespec_expression_to_pc (char **exp_ptr);
-static VEC (symbolp) *find_function_symbols (char **argptr, char *p,
- int is_quote_enclosed,
- char **user_function);
+static struct symtabs_and_lines decode_objc (struct linespec_state *self,
+ linespec_t ls,
+ char **argptr);
-static struct symtabs_and_lines decode_all_digits (struct linespec_state *self,
- char **argptr,
- char *q);
+static VEC (symtab_p) *symtabs_from_filename (const char *);
-static struct symtabs_and_lines decode_dollar (struct linespec_state *self,
- char *copy);
+static VEC (symbolp) *find_label_symbols (struct linespec_state *self,
+ VEC (symbolp) *function_symbols,
+ VEC (symbolp) **label_funcs_ret,
+ const char *name);
-static int decode_label (struct linespec_state *self,
- VEC (symbolp) *function_symbols,
- char *copy,
- struct symtabs_and_lines *result);
+void find_linespec_symbols (struct linespec_state *self,
+ VEC (symtab_p) *file_symtabs,
+ const char *name,
+ VEC (symbolp) **symbols,
+ VEC (minsym_and_objfile_d) **minsyms);
-static struct symtabs_and_lines decode_variable (struct linespec_state *self,
- char *copy);
+static struct line_offset
+ linespec_parse_variable (struct linespec_state *self,
+ const char *variable);
static int symbol_to_sal (struct symtab_and_line *result,
int funfirstline, struct symbol *sym);
@@ -190,6 +303,367 @@ static void add_all_symbol_names_from_pspace (struct collect_info *info,
struct program_space *pspace,
VEC (const_char_ptr) *names);
+static VEC (symtab_p) *collect_symtabs_from_filename (const char *file);
+
+static void decode_digits_ordinary (struct linespec_state *self,
+ linespec_t ls,
+ int line,
+ struct symtabs_and_lines *sals,
+ struct linetable_entry **best_entry);
+
+static void decode_digits_list_mode (struct linespec_state *self,
+ linespec_t ls,
+ struct symtabs_and_lines *values,
+ struct symtab_and_line val);
+
+static void minsym_found (struct linespec_state *self, struct objfile *objfile,
+ struct minimal_symbol *msymbol,
+ struct symtabs_and_lines *result);
+
+static int compare_symbols (const void *a, const void *b);
+
+static int compare_msymbols (const void *a, const void *b);
+
+static const char *find_toplevel_char (const char *s, char c);
+
+/* Permitted quote characters for the parser. This is different from the
+ completer's quote characters to allow backward compatibility with the
+ previous parser. */
+static const char *const linespec_quote_characters = "\"\'";
+
+/* Lexer functions. */
+
+/* Lex a number from the input in PARSER. This only supports
+ decimal numbers. */
+
+static linespec_token
+linespec_lexer_lex_number (linespec_parser *parser)
+{
+ linespec_token token;
+
+ token.type = LSTOKEN_NUMBER;
+ LS_TOKEN_STOKEN (token).length = 0;
+ LS_TOKEN_STOKEN (token).ptr = PARSER_STREAM (parser);
+
+ /* Keep any sign at the start of the stream. */
+ if (*PARSER_STREAM (parser) == '+' || *PARSER_STREAM (parser) == '-')
+ {
+ ++LS_TOKEN_STOKEN (token).length;
+ ++(PARSER_STREAM (parser));
+ }
+
+ while (isdigit (*PARSER_STREAM (parser)))
+ {
+ ++LS_TOKEN_STOKEN (token).length;
+ ++(PARSER_STREAM (parser));
+ }
+
+ return token;
+}
+
+/* Does P represent one of the keywords? If so, return
+ the keyword. If not, return NULL. */
+
+static const char *
+linespec_lexer_lex_keyword (const char *p)
+{
+ int i;
+
+ if (p != NULL)
+ {
+ for (i = 0; i < ARRAY_SIZE (linespec_keywords); ++i)
+ {
+ int len = strlen (linespec_keywords[i]);
+
+ /* If P begins with one of the keywords and the next
+ character is not a valid identifier character,
+ we have found a keyword. */
+ if (strncmp (p, linespec_keywords[i], len) == 0
+ && !(isalnum (p[len]) || p[len] == '_'))
+ return linespec_keywords[i];
+ }
+ }
+
+ return NULL;
+}
+
+/* Does STRING represent an Ada operator? If so, return the length
+ of the decoded operator name. If not, return 0. */
+
+static int
+is_ada_operator (const char *string)
+{
+ const struct ada_opname_map *mapping;
+
+ for (mapping = ada_opname_table;
+ mapping->encoded != NULL
+ && strncmp (mapping->decoded, string,
+ strlen (mapping->decoded)) != 0; ++mapping)
+ ;
+
+ return mapping->decoded == NULL ? 0 : strlen (mapping->decoded);
+}
+
+/* Find QUOTE_CHAR in STRING, accounting for the ':' terminal. Return
+ the location of QUOTE_CHAR, or NULL if not found. */
+
+static char *
+skip_quote_char (const char *string, char quote_char)
+{
+ const char *p, *last;
+
+ p = last = find_toplevel_char (string, quote_char);
+ while (p && *p != '\0' && *p != ':')
+ {
+ p = find_toplevel_char (p, quote_char);
+ if (p != NULL)
+ last = p++;
+ }
+
+ return (char *) last;
+}
+
+/* Make a writable copy of the string given in TOKEN, trimming
+ any trailing whitespace. */
+
+static char *
+copy_token_string (linespec_token token)
+{
+ char *str, *s;
+
+ if (token.type == LSTOKEN_KEYWORD)
+ return xstrdup (LS_TOKEN_KEYWORD (token));
+
+ str = savestring (LS_TOKEN_STOKEN (token).ptr,
+ LS_TOKEN_STOKEN (token).length);
+ s = remove_trailing_whitespace (str, str + LS_TOKEN_STOKEN (token).length);
+ *s = '\0';
+
+ return str;
+}
+
+/* Lex a string from the input in PARSER. */
+
+static linespec_token
+linespec_lexer_lex_string (linespec_parser *parser)
+{
+ linespec_token token;
+ char *start = PARSER_STREAM (parser);
+
+ token.type = LSTOKEN_STRING;
+
+ /* If the input stream starts with a quote character, skip to the next
+ quote character, regardless of the content. */
+ if (strchr (linespec_quote_characters, *PARSER_STREAM (parser)))
+ {
+ char *end;
+ char quote_char = *PARSER_STREAM (parser);
+
+ /* Special case: Ada operators. */
+ if (current_language->la_language == language_ada
+ && quote_char == '\"')
+ {
+ int len = is_ada_operator (PARSER_STREAM (parser));
+
+ if (len != 0)
+ {
+ /* The input is an Ada operator. Return the quoted string
+ as-is. */
+ LS_TOKEN_STOKEN (token).ptr = PARSER_STREAM (parser);
+ LS_TOKEN_STOKEN (token).length = len;
+ PARSER_STREAM (parser) += len;
+ return token;
+ }
+
+ /* The input does not represent an Ada operator -- fall through
+ to normal quoted string handling. */
+ }
+
+ /* Skip past the beginning quote. */
+ ++(PARSER_STREAM (parser));
+
+ /* Mark the start of the string. */
+ LS_TOKEN_STOKEN (token).ptr = PARSER_STREAM (parser);
+
+ /* Skip to the ending quote. */
+ end = skip_quote_char (PARSER_STREAM (parser), quote_char);
+
+ /* Error if the input did not terminate properly. */
+ if (end == NULL)
+ error (_("unmatched quote"));
+
+ /* Skip over the ending quote and mark the length of the string. */
+ PARSER_STREAM (parser) = ++end;
+ LS_TOKEN_STOKEN (token).length = PARSER_STREAM (parser) - 2 - start;
+ }
+ else
+ {
+ char *p;
+
+ /* Otherwise, only identifier characters are permitted.
+ Spaces are the exception. In general, we keep spaces,
+ but only if the next characters in the input do not resolve
+ to one of the keywords.
+
+ This allows users to forgo quoting CV-qualifiers, template arguments,
+ and similar common language constructs. */
+
+ while (1)
+ {
+ if (isspace (*PARSER_STREAM (parser)))
+ {
+ p = skip_spaces (PARSER_STREAM (parser));
+ if (linespec_lexer_lex_keyword (p) != NULL)
+ {
+ LS_TOKEN_STOKEN (token).ptr = start;
+ LS_TOKEN_STOKEN (token).length
+ = PARSER_STREAM (parser) - start;
+ return token;
+ }
+
+ /* Advance past the whitespace. */
+ PARSER_STREAM (parser) = p;
+ }
+
+ /* If the next character is EOI or (single) ':', the
+ string is complete; return the token. */
+ if (*PARSER_STREAM (parser) == 0)
+ {
+ LS_TOKEN_STOKEN (token).ptr = start;
+ LS_TOKEN_STOKEN (token).length = PARSER_STREAM (parser) - start;
+ return token;
+ }
+ else if (PARSER_STREAM (parser)[0] == ':')
+ {
+ /* Do not tokenize the C++ scope operator. */
+ if (PARSER_STREAM (parser)[1] == ':')
+ ++(PARSER_STREAM (parser));
+
+ /* Do not tokenify if the input length so far is one
+ (i.e, a single-letter drive name) and the next character
+ is a directory separator. This allows Windows-style
+ paths to be recognized as filenames without quoting it. */
+ else if ((PARSER_STREAM (parser) - start) != 1
+ || !IS_DIR_SEPARATOR (PARSER_STREAM (parser)[1]))
+ {
+ LS_TOKEN_STOKEN (token).ptr = start;
+ LS_TOKEN_STOKEN (token).length
+ = PARSER_STREAM (parser) - start;
+ return token;
+ }
+ }
+ /* Special case: permit quote-enclosed linespecs. */
+ else if (parser->is_quote_enclosed
+ && strchr (linespec_quote_characters,
+ PARSER_STREAM (parser)[0])
+ && PARSER_STREAM (parser)[1] == '\0')
+ {
+ LS_TOKEN_STOKEN (token).ptr = start;
+ LS_TOKEN_STOKEN (token).length = PARSER_STREAM (parser) - start;
+ return token;
+ }
+
+ /* Advance the stream. */
+ ++(PARSER_STREAM (parser));
+ }
+ }
+
+ return token;
+}
+
+/* Lex a single linespec token from PARSER. */
+
+static linespec_token
+linespec_lexer_lex_one (linespec_parser *parser)
+{
+ const char *keyword;
+
+ if (parser->lexer.current.type == LSTOKEN_CONSUMED)
+ {
+ /* Skip any whitespace. */
+ PARSER_STREAM (parser) = skip_spaces (PARSER_STREAM (parser));
+
+ /* Check for a keyword. */
+ keyword = linespec_lexer_lex_keyword (PARSER_STREAM (parser));
+ if (keyword != NULL)
+ {
+ parser->lexer.current.type = LSTOKEN_KEYWORD;
+ LS_TOKEN_KEYWORD (parser->lexer.current) = keyword;
+ return parser->lexer.current;
+ }
+
+ /* Handle other tokens. */
+ switch (*PARSER_STREAM (parser))
+ {
+ case 0:
+ parser->lexer.current.type = LSTOKEN_EOI;
+ break;
+
+ case '+': case '-':
+ case '0': case '1': case '2': case '3': case '4':
+ case '5': case '6': case '7': case '8': case '9':
+ parser->lexer.current = linespec_lexer_lex_number (parser);
+ break;
+
+ case ':':
+ /* If we have a scope operator, lex the input as a string.
+ Otherwise, return LSTOKEN_COLON. */
+ if (PARSER_STREAM (parser)[1] == ':')
+ parser->lexer.current = linespec_lexer_lex_string (parser);
+ else
+ {
+ parser->lexer.current.type = LSTOKEN_COLON;
+ ++(PARSER_STREAM (parser));
+ }
+ break;
+
+ case '\'': case '\"':
+ /* Special case: permit quote-enclosed linespecs. */
+ if (parser->is_quote_enclosed && PARSER_STREAM (parser)[1] == '\0')
+ {
+ ++(PARSER_STREAM (parser));
+ parser->lexer.current.type = LSTOKEN_EOI;
+ }
+ else
+ parser->lexer.current = linespec_lexer_lex_string (parser);
+ break;
+
+ default:
+ /* If the input is not a number, it must be a string.
+ [Keywords were already considered above.] */
+ parser->lexer.current = linespec_lexer_lex_string (parser);
+ break;
+ }
+ }
+
+ return parser->lexer.current;
+}
+
+/* Consume the current token and return the next token in PARSER's
+ input stream. */
+
+static linespec_token
+linespec_lexer_consume_token (linespec_parser *parser)
+{
+ parser->lexer.current.type = LSTOKEN_CONSUMED;
+ return linespec_lexer_lex_one (parser);
+}
+
+/* Return the next token without consuming the current token. */
+
+static linespec_token
+linespec_lexer_peek_token (linespec_parser *parser)
+{
+ linespec_token next;
+ char *saved_stream = PARSER_STREAM (parser);
+ linespec_token saved_token = parser->lexer.current;
+
+ next = linespec_lexer_consume_token (parser);
+ PARSER_STREAM (parser) = saved_stream;
+ parser->lexer.current = saved_token;
+ return next;
+}
+
/* Helper functions. */
/* Add SAL to SALS. */
@@ -288,39 +762,6 @@ maybe_add_address (htab_t set, struct program_space *pspace, CORE_ADDR addr)
return 1;
}
-/* Issue a helpful hint on using the command completion feature on
- single quoted demangled C++ symbols as part of the completion
- error. */
-
-static void
-cplusplus_error (const char *name, const char *fmt, ...)
-{
- struct ui_file *tmp_stream;
- char *message;
-
- tmp_stream = mem_fileopen ();
- make_cleanup_ui_file_delete (tmp_stream);
-
- {
- va_list args;
-
- va_start (args, fmt);
- vfprintf_unfiltered (tmp_stream, fmt, args);
- va_end (args);
- }
-
- while (*name == '\'')
- name++;
- fprintf_unfiltered (tmp_stream,
- ("Hint: try '%s<TAB> or '%s<ESC-?>\n"
- "(Note leading single quote.)"),
- name, name);
-
- message = ui_file_xstrdup (tmp_stream, NULL);
- make_cleanup (xfree, message);
- throw_error (NOT_FOUND_ERROR, "%s", message);
-}
-
/* A callback function and the additional data to call it with. */
struct symbol_and_data_callback
@@ -468,7 +909,7 @@ get_search_block (struct symtab *symtab)
}
/* A helper for find_method. This finds all methods in type T which
- match NAME. It adds resulting symbol names to RESULT_NAMES, and
+ match NAME. It adds matching symbol names to RESULT_NAMES, and
adds T's direct superclasses to SUPERCLASSES. */
static void
@@ -542,14 +983,14 @@ find_methods (struct type *t, const char *name,
strings. Also, ignore the char within a template name, like a ','
within foo<int, int>. */
-static char *
-find_toplevel_char (char *s, char c)
+static const char *
+find_toplevel_char (const char *s, char c)
{
int quoted = 0; /* zero if we're not in quotes;
'"' if we're in a double-quoted string;
'\'' if we're in a single-quoted string. */
int depth = 0; /* Number of unclosed parens we've seen. */
- char *scan;
+ const char *scan;
for (scan = s; *scan; scan++)
{
@@ -573,23 +1014,34 @@ find_toplevel_char (char *s, char c)
return 0;
}
-/* Determines if the gives string corresponds to an Objective-C method
- representation, such as -[Foo bar:] or +[Foo bar]. Objective-C symbols
- are allowed to have spaces and parentheses in them. */
+/* The string equivalent of find_toplevel_char. Returns a pointer
+ to the location of NEEDLE in HAYSTACK, ignoring any occurrences
+ inside "()" and "<>". Returns NULL if NEEDLE was not found. */
-static int
-is_objc_method_format (const char *s)
+static const char *
+find_toplevel_string (const char *haystack, const char *needle)
{
- if (s == NULL || *s == '\0')
- return 0;
- /* Handle arguments with the format FILENAME:SYMBOL. */
- if ((s[0] == ':') && (strchr ("+-", s[1]) != NULL)
- && (s[2] == '[') && strchr(s, ']'))
- return 1;
- /* Handle arguments that are just SYMBOL. */
- else if ((strchr ("+-", s[0]) != NULL) && (s[1] == '[') && strchr(s, ']'))
- return 1;
- return 0;
+ const char *s = haystack;
+
+ do
+ {
+ s = find_toplevel_char (s, *needle);
+
+ if (s != NULL)
+ {
+ /* Found first char in HAYSTACK; check rest of string. */
+ if (strncmp (s, needle, strlen (needle)) == 0)
+ return s;
+
+ /* Didn't find it; loop over HAYSTACK, looking for the next
+ instance of the first character of NEEDLE. */
+ ++s;
+ }
+ }
+ while (s != NULL && *s != '\0');
+
+ /* NEEDLE was not found in HAYSTACK. */
+ return NULL;
}
/* Given FILTERS, a list of canonical names, filter the sals in RESULT
@@ -758,118 +1210,640 @@ decode_line_2 (struct linespec_state *self,
do_cleanups (old_chain);
}
-/* Valid delimiters for linespec keywords "if", "thread" or "task". */
+\f
-static int
-is_linespec_boundary (char c)
+/* The parser of linespec itself. */
+
+/* Throw an appropriate error when SYMBOL is not found (optionally in
+ FILENAME). */
+
+static void ATTRIBUTE_NORETURN
+symbol_not_found_error (char *symbol, char *filename)
{
- return c == ' ' || c == '\t' || c == '\0' || c == ',';
+ if (symbol == NULL)
+ symbol = "";
+
+ if (!have_full_symbols ()
+ && !have_partial_symbols ()
+ && !have_minimal_symbols ())
+ throw_error (NOT_FOUND_ERROR,
+ _("No symbol table is loaded. Use the \"file\" command."));
+
+ /* If SYMBOL starts with '$', the user attempted to either lookup
+ a function/variable in his code starting with '$' or an internal
+ variable of that name. Since we do not know which, be concise and
+ explain both possibilities. */
+ if (*symbol == '$')
+ {
+ if (filename)
+ throw_error (NOT_FOUND_ERROR,
+ _("Undefined convenience variable or function \"%s\" "
+ "not defined in \"%s\"."), symbol, filename);
+ else
+ throw_error (NOT_FOUND_ERROR,
+ _("Undefined convenience variable or function \"%s\" "
+ "not defined."), symbol);
+ }
+ else
+ {
+ if (filename)
+ throw_error (NOT_FOUND_ERROR,
+ _("Function \"%s\" not defined in \"%s\"."),
+ symbol, filename);
+ else
+ throw_error (NOT_FOUND_ERROR,
+ _("Function \"%s\" not defined."), symbol);
+ }
}
-/* A helper function for decode_line_1 and friends which skips P
- past any method overload information at the beginning of P, e.g.,
- "(const struct foo *)".
+/* Throw an appropriate error when an unexpected token is encountered
+ in the input. */
- This function assumes that P has already been validated to contain
- overload information, and it will assert if *P != '('. */
-static char *
-find_method_overload_end (char *p)
+static void ATTRIBUTE_NORETURN
+unexpected_linespec_error (linespec_parser *parser)
{
- int depth = 0;
+ linespec_token token;
+ static const char * token_type_strings[]
+ = {"keyword", "colon", "string", "number", "end of input"};
- gdb_assert (*p == '(');
+ /* Get the token that generated the error. */
+ token = linespec_lexer_lex_one (parser);
- while (*p)
+ /* Finally, throw the error. */
+ if (token.type == LSTOKEN_STRING || token.type == LSTOKEN_NUMBER
+ || token.type == LSTOKEN_KEYWORD)
{
- if (*p == '(')
- ++depth;
- else if (*p == ')')
+ char *string;
+ struct cleanup *cleanup;
+
+ string = copy_token_string (token);
+ cleanup = make_cleanup (xfree, string);
+ throw_error (GENERIC_ERROR,
+ _("malformed linespec error: unexpected %s, \"%s\""),
+ token_type_strings[token.type], string);
+ }
+ else
+ throw_error (GENERIC_ERROR,
+ _("malformed linespec error: unexpected %s"),
+ token_type_strings[token.type]);
+}
+
+/* Parse and return a line offset in STRING. */
+
+static struct line_offset
+linespec_parse_line_offset (char *string)
+{
+ struct line_offset line_offset = {0, none};
+
+ if (*string == '+')
+ {
+ line_offset.sign = plus;
+ ++string;
+ }
+ else if (*string == '-')
+ {
+ line_offset.sign = minus;
+ ++string;
+ }
+
+ /* Right now, we only allow base 10 for offsets. */
+ line_offset.offset = atoi (string);
+ return line_offset;
+}
+
+/* Parse the basic_spec in PARSER's input. */
+
+static void
+linespec_parse_basic (linespec_parser *parser)
+{
+ char *name;
+ linespec_token token;
+ VEC (symbolp) *symbols, *labels;
+ VEC (minsym_and_objfile_d) *minimal_symbols;
+ struct cleanup *cleanup;
+
+ /* Get the next token. */
+ token = linespec_lexer_lex_one (parser);
+
+ /* If it is EOI or KEYWORD, issue an error. */
+ if (token.type == LSTOKEN_KEYWORD || token.type == LSTOKEN_EOI)
+ unexpected_linespec_error (parser);
+ /* If it is a LSTOKEN_NUMBER, we have an offset. */
+ else if (token.type == LSTOKEN_NUMBER)
+ {
+ /* Record the line offset and get the next token. */
+ name = copy_token_string (token);
+ cleanup = make_cleanup (xfree, name);
+ PARSER_RESULT (parser)->line_offset = linespec_parse_line_offset (name);
+ do_cleanups (cleanup);
+
+ /* Get the next token. */
+ token = linespec_lexer_consume_token (parser);
+
+ /* If we're in list mode, and the next token is a string beginning
+ with ",", we're dealing with a ranged listing. Stop parsing
+ and return. */
+ if (PARSER_STATE (parser)->list_mode
+ && token.type == LSTOKEN_STRING
+ && *LS_TOKEN_STOKEN (token).ptr == ',')
+ return;
+
+ /* If the next token is anything but EOI or KEYWORD, issue
+ an error. */
+ if (token.type != LSTOKEN_KEYWORD && token.type != LSTOKEN_EOI)
+ unexpected_linespec_error (parser);
+ }
+
+ if (token.type == LSTOKEN_KEYWORD || token.type == LSTOKEN_EOI)
+ return;
+
+ /* Next token must be LSTOKEN_STRING. */
+ if (token.type != LSTOKEN_STRING)
+ unexpected_linespec_error (parser);
+
+ /* The current token will contain the name of a function, method,
+ or label. */
+ name = copy_token_string (token);
+ cleanup = make_cleanup (xfree, name);
+
+ /* Try looking it up as a function/method. */
+ find_linespec_symbols (PARSER_STATE (parser),
+ PARSER_RESULT (parser)->file_symtabs, name,
+ &symbols, &minimal_symbols);
+
+ if (symbols != NULL || minimal_symbols != NULL)
+ {
+ PARSER_RESULT (parser)->function_symbols = symbols;
+ PARSER_RESULT (parser)->minimal_symbols = minimal_symbols;
+ PARSER_RESULT (parser)->function_name = name;
+ symbols = NULL;
+ discard_cleanups (cleanup);
+ }
+ else
+ {
+ /* NAME was not a function or a method. So it must be a label
+ name. */
+ labels = find_label_symbols (PARSER_STATE (parser), NULL,
+ &symbols, name);
+ if (labels != NULL)
{
- if (--depth == 0)
+ PARSER_RESULT (parser)->labels.label_symbols = labels;
+ PARSER_RESULT (parser)->labels.function_symbols = symbols;
+ PARSER_RESULT (parser)->label_name = name;
+ symbols = NULL;
+ discard_cleanups (cleanup);
+ }
+ else
+ {
+ /* The name is also not a label. Abort parsing. Do not throw
+ an error here. parse_linespec will do it for us. */
+
+ /* Save a copy of the name we were trying to lookup. */
+ PARSER_RESULT (parser)->function_name = name;
+ discard_cleanups (cleanup);
+ return;
+ }
+ }
+
+ /* Get the next token. */
+ token = linespec_lexer_consume_token (parser);
+
+ if (token.type == LSTOKEN_COLON)
+ {
+ /* User specified a label or a lineno. */
+ token = linespec_lexer_consume_token (parser);
+
+ if (token.type == LSTOKEN_NUMBER)
+ {
+ if (PARSER_RESULT (parser)->function_name != NULL)
{
- ++p;
- break;
+ /* We could do this, but it is a new feature. */
+ throw_error (UNSUPPORTED_ERROR,
+ _("FUNCTION:OFFSET is unimplemented"));
+ }
+ else
+ {
+ /* User specified FILE:LINE. Record the line offset and
+ get the next token. */
+ name = copy_token_string (token);
+ cleanup = make_cleanup (xfree, name);
+ PARSER_RESULT (parser)->line_offset
+ = linespec_parse_line_offset (name);
+ do_cleanups (cleanup);
+
+ /* Ge the next token. */
+ token = linespec_lexer_consume_token (parser);
+ }
+ }
+ else if (token.type == LSTOKEN_STRING)
+ {
+ /* Grab a copy of the label's name and look it up. */
+ name = copy_token_string (token);
+ cleanup = make_cleanup (xfree, name);
+ labels = find_label_symbols (PARSER_STATE (parser),
+ PARSER_RESULT (parser)->function_symbols,
+ &symbols, name);
+
+ if (labels != NULL)
+ {
+ PARSER_RESULT (parser)->labels.label_symbols = labels;
+ PARSER_RESULT (parser)->labels.function_symbols = symbols;
+ PARSER_RESULT (parser)->label_name = name;
+ symbols = NULL;
+ discard_cleanups (cleanup);
+ }
+ else
+ {
+ /* We don't know what it was, but it isn't a label. */
+ throw_error (NOT_FOUND_ERROR,
+ _("No label \"%s\" defined in function \"%s\"."),
+ name, PARSER_RESULT (parser)->function_name);
+ }
+
+ /* Check for a line offset. */
+ token = linespec_lexer_consume_token (parser);
+ if (token.type == LSTOKEN_COLON)
+ {
+ /* Get the next token. */
+ token = linespec_lexer_consume_token (parser);
+
+ /* It must be a line offset. */
+ if (token.type != LSTOKEN_NUMBER)
+ unexpected_linespec_error (parser);
+
+ /* Record the lione offset and get the next token. */
+ name = copy_token_string (token);
+ cleanup = make_cleanup (xfree, name);
+
+ PARSER_RESULT (parser)->line_offset
+ = linespec_parse_line_offset (name);
+ do_cleanups (cleanup);
+
+ /* Get the next token. */
+ token = linespec_lexer_consume_token (parser);
}
}
- ++p;
+ else
+ {
+ /* Trailing ':' in the input. Issue an error. */
+ unexpected_linespec_error (parser);
+ }
}
+}
+
+/* Canonicalize the linespec contained in LS. The result is saved into
+ STATE->canonical. */
+
+static void
+canonicalize_linespec (struct linespec_state *state, linespec_t ls)
+{
+ /* If canonicalization was not requested, no need to do anything. */
+ if (!state->canonical)
+ return;
+
+ /* Shortcut expressions, which can only appear by themselves. */
+ if (ls->expression != NULL)
+ state->canonical->addr_string = xstrdup (ls->expression);
+ else
+ {
+ struct ui_file *buf;
+ int need_colon = 0;
+
+ buf = mem_fileopen ();
+ if (ls->source_filename)
+ {
+ fputs_unfiltered (ls->source_filename, buf);
+ need_colon = 1;
+ }
+
+ if (ls->function_name)
+ {
+ if (need_colon)
+ fputc_unfiltered (':', buf);
+ fputs_unfiltered (ls->function_name, buf);
+ need_colon = 1;
+ }
- return p;
+ if (ls->label_name)
+ {
+ if (need_colon)
+ fputc_unfiltered (':', buf);
+
+ if (ls->function_name == NULL)
+ {
+ struct symbol *s;
+
+ /* No function was specified, so add the symbol name. */
+ gdb_assert (ls->labels.function_symbols != NULL
+ && (VEC_length (symbolp, ls->labels.function_symbols)
+ == 1));
+ s = VEC_index (symbolp, ls->labels.function_symbols, 0);
+ fputs_unfiltered (SYMBOL_NATURAL_NAME (s), buf);
+ fputc_unfiltered (':', buf);
+ }
+
+ fputs_unfiltered (ls->label_name, buf);
+ need_colon = 1;
+ state->canonical->special_display = 1;
+ }
+
+ if (ls->line_offset.sign != unknown)
+ {
+ if (need_colon)
+ fputc_unfiltered (':', buf);
+ fprintf_filtered (buf, "%s%d",
+ (ls->line_offset.sign == none ? ""
+ : ls->line_offset.sign == plus ? "+" : "-"),
+ ls->line_offset.offset);
+ }
+
+ state->canonical->addr_string = ui_file_xstrdup (buf, NULL);
+ ui_file_delete (buf);
+ }
}
-/* Keep important information used when looking up a name. This includes
- template parameters, overload information, and important keywords, including
- the possible Java trailing type. */
+/* Given a line offset in LS, construct the relevant SALs. */
-static char *
-keep_name_info (char *p, int on_boundary)
+static struct symtabs_and_lines
+create_sals_line_offset (struct linespec_state *self,
+ linespec_t ls)
{
- const char *quotes = get_gdb_completer_quote_characters ();
- char *saved_p = p;
- int nest = 0;
+ struct symtabs_and_lines values;
+ struct symtab_and_line val;
+ int use_default = 0;
- while (*p)
+ init_sal (&val);
+ values.sals = NULL;
+ values.nelts = 0;
+
+ /* This is where we need to make sure we have good defaults.
+ We must guarantee that this section of code is never executed
+ when we are called with just a function anme, since
+ set_default_source_symtab_and_line uses
+ select_source_symtab that calls us with such an argument. */
+
+ if (VEC_length (symtab_p, ls->file_symtabs) == 1
+ && VEC_index (symtab_p, ls->file_symtabs, 0) == NULL)
{
- if (strchr (quotes, *p))
- break;
+ set_current_program_space (self->program_space);
- if (*p == ',' && !nest)
- break;
+ /* Make sure we have at least a default source line. */
+ set_default_source_symtab_and_line ();
+ initialize_defaults (&self->default_symtab, &self->default_line);
+ VEC_pop (symtab_p, ls->file_symtabs);
+ VEC_free (symtab_p, ls->file_symtabs);
+ ls->file_symtabs
+ = collect_symtabs_from_filename (self->default_symtab->filename);
+ use_default = 1;
+ }
+
+ val.line = ls->line_offset.offset;
+ switch (ls->line_offset.sign)
+ {
+ case plus:
+ if (ls->line_offset.offset == 0)
+ val.line = 5;
+ if (use_default)
+ val.line = self->default_line + val.line;
+ break;
+
+ case minus:
+ if (ls->line_offset.offset == 0)
+ val.line = 15;
+ if (use_default)
+ val.line = self->default_line - val.line;
+ else
+ val.line = -val.line;
+ break;
+
+ case none:
+ break; /* No need to adjust val.line. */
+ }
+
+ if (self->list_mode)
+ decode_digits_list_mode (self, ls, &values, val);
+ else
+ {
+ struct linetable_entry *best_entry = NULL;
+ int *filter;
+ struct block **blocks;
+ struct cleanup *cleanup;
+ struct symtabs_and_lines intermediate_results;
+ int i, j;
+
+ intermediate_results.sals = NULL;
+ intermediate_results.nelts = 0;
+
+ decode_digits_ordinary (self, ls, val.line, &intermediate_results,
+ &best_entry);
+ if (intermediate_results.nelts == 0 && best_entry != NULL)
+ decode_digits_ordinary (self, ls, best_entry->line,
+ &intermediate_results, &best_entry);
+
+ cleanup = make_cleanup (xfree, intermediate_results.sals);
- if (on_boundary && !nest)
+ /* For optimized code, the compiler can scatter one source line
+ across disjoint ranges of PC values, even when no duplicate
+ functions or inline functions are involved. For example,
+ 'for (;;)' inside a non-template, non-inline, and non-ctor-or-dtor
+ function can result in two PC ranges. In this case, we don't
+ want to set a breakpoint on the first PC of each range. To filter
+ such cases, we use containing blocks -- for each PC found
+ above, we see if there are other PCs that are in the same
+ block. If yes, the other PCs are filtered out. */
+
+ filter = XNEWVEC (int, intermediate_results.nelts);
+ make_cleanup (xfree, filter);
+ blocks = XNEWVEC (struct block *, intermediate_results.nelts);
+ make_cleanup (xfree, blocks);
+
+ for (i = 0; i < intermediate_results.nelts; ++i)
{
- const char *const words[] = { "if", "thread", "task" };
- int wordi;
+ set_current_program_space (intermediate_results.sals[i].pspace);
- for (wordi = 0; wordi < ARRAY_SIZE (words); wordi++)
- if (strncmp (p, words[wordi], strlen (words[wordi])) == 0
- && is_linespec_boundary (p[strlen (words[wordi])]))
- break;
- if (wordi < ARRAY_SIZE (words))
- break;
+ filter[i] = 1;
+ blocks[i] = block_for_pc_sect (intermediate_results.sals[i].pc,
+ intermediate_results.sals[i].section);
}
- if (*p == '(' || *p == '<' || *p == '[')
- nest++;
- else if ((*p == ')' || *p == '>' || *p == ']') && nest > 0)
- nest--;
+ for (i = 0; i < intermediate_results.nelts; ++i)
+ {
+ if (blocks[i] != NULL)
+ for (j = i + 1; j < intermediate_results.nelts; ++j)
+ {
+ if (blocks[j] == blocks[i])
+ {
+ filter[j] = 0;
+ break;
+ }
+ }
+ }
- p++;
+ for (i = 0; i < intermediate_results.nelts; ++i)
+ if (filter[i])
+ {
+ struct symbol *sym = (blocks[i]
+ ? block_containing_function (blocks[i])
+ : NULL);
- /* The ',' check could fail on "operator ,". */
- p += cp_validate_operator (p);
+ if (self->funfirstline)
+ skip_prologue_sal (&intermediate_results.sals[i]);
+ /* Make sure the line matches the request, not what was
+ found. */
+ intermediate_results.sals[i].line = val.line;
+ add_sal_to_sals (self, &values, &intermediate_results.sals[i],
+ sym ? SYMBOL_NATURAL_NAME (sym) : NULL);
+ }
- on_boundary = is_linespec_boundary (p[-1]);
+ do_cleanups (cleanup);
}
- while (p > saved_p && is_linespec_boundary (p[-1]))
- p--;
+ if (values.nelts == 0)
+ {
+ if (ls->source_filename)
+ throw_error (NOT_FOUND_ERROR, _("No line %d in file \"%s\"."),
+ val.line, ls->source_filename);
+ else
+ throw_error (NOT_FOUND_ERROR, _("No line %d in the current file."),
+ val.line);
+ }
- return p;
+ return values;
}
-\f
-/* The parser of linespec itself. */
+/* Create and return SALs from the linespec LS. */
+
+static struct symtabs_and_lines
+convert_linespec_to_sals (struct linespec_state *state, linespec_t ls)
+{
+ struct symtabs_and_lines sals = {NULL, 0};
+
+ if (ls->expression != NULL)
+ {
+ /* We have an expression. No other attribute is allowed. */
+ sals.sals = XMALLOC (struct symtab_and_line);
+ sals.nelts = 1;
+ sals.sals[0] = find_pc_line (ls->expr_pc, 0);
+ sals.sals[0].pc = ls->expr_pc;
+ sals.sals[0].section = find_pc_overlay (ls->expr_pc);
+ sals.sals[0].explicit_pc = 1;
+ }
+ else if (ls->labels.label_symbols != NULL)
+ {
+ /* We have just a bunch of functions/methods or labels. */
+ int i;
+ struct symtab_and_line sal;
+ struct symbol *sym;
+
+ for (i = 0; VEC_iterate (symbolp, ls->labels.label_symbols, i, sym); ++i)
+ {
+ symbol_to_sal (&sal, state->funfirstline, sym);
+ add_sal_to_sals (state, &sals, &sal,
+ SYMBOL_NATURAL_NAME (sym));
+ }
+ }
+ else if (ls->function_symbols != NULL)
+ {
+ /* We have just a bunch of functions and/or methods. */
+ int i;
+ struct symtab_and_line sal;
+ struct symbol *sym;
+ minsym_and_objfile_d *elem;
+ struct program_space *pspace;
+
+ /* Sort symbols so that symbols with the same program space are next
+ to each other. */
+ qsort (VEC_address (symbolp, ls->function_symbols),
+ VEC_length (symbolp, ls->function_symbols),
+ sizeof (symbolp), compare_symbols);
+
+ for (i = 0; VEC_iterate (symbolp, ls->function_symbols, i, sym); ++i)
+ {
+ pspace = SYMTAB_PSPACE (SYMBOL_SYMTAB (sym));
+ set_current_program_space (pspace);
+ symbol_to_sal (&sal, state->funfirstline, sym);
+ if (maybe_add_address (state->addr_set, pspace, sal.pc))
+ add_sal_to_sals (state, &sals, &sal, SYMBOL_NATURAL_NAME (sym));
+ }
+
+ /* Sort minimal symbols by program space, too. */
+ qsort (VEC_address (minsym_and_objfile_d, ls->minimal_symbols),
+ VEC_length (minsym_and_objfile_d, ls->minimal_symbols),
+ sizeof (minsym_and_objfile_d), compare_msymbols);
+
+ for (i = 0;
+ VEC_iterate (minsym_and_objfile_d, ls->minimal_symbols, i, elem);
+ ++i)
+ {
+ pspace = elem->minsym->ginfo.obj_section->objfile->pspace;
+ set_current_program_space (pspace);
+ minsym_found (state, elem->objfile, elem->minsym, &sals);
+ }
+ }
+ else if (ls->minimal_symbols != NULL)
+ {
+ /* We found minimal symbols, but no normal symbols. */
+ int i;
+ minsym_and_objfile_d *elem;
+
+ for (i = 0;
+ VEC_iterate (minsym_and_objfile_d, ls->minimal_symbols, i, elem);
+ ++i)
+ minsym_found (state, elem->objfile, elem->minsym, &sals);
+ }
+ else if (ls->line_offset.sign != unknown)
+ {
+ /* Only an offset was specified. */
+ sals = create_sals_line_offset (state, ls);
-/* Parse a string that specifies a line number.
+ /* Make sure we have a filename for canonicalization. */
+ if (ls->source_filename == NULL)
+ ls->source_filename = xstrdup (state->default_symtab->filename);
+ }
+ else
+ {
+ /* We haven't found any results... */
+ return sals;
+ }
+
+ canonicalize_linespec (state, ls);
+
+ if (sals.nelts > 0 && state->canonical != NULL)
+ state->canonical->pre_expanded = 1;
+
+ return sals;
+}
+
+/* Parse a string that specifies a linespec.
Pass the address of a char * variable; that variable will be
advanced over the characters actually parsed.
- The string can be:
+ The basic grammar of linespecs:
- LINENUM -- that line number in current file. PC returned is 0.
- FILE:LINENUM -- that line in that file. PC returned is 0.
- FUNCTION -- line number of openbrace of that function.
- PC returned is the start of the function.
- LABEL -- a label in the current scope
- VARIABLE -- line number of definition of that variable.
- PC returned is 0.
- FILE:FUNCTION -- likewise, but prefer functions in that file.
- *EXPR -- line in which address EXPR appears.
+ linespec -> expr_spec | var_spec | basic_spec
+ expr_spec -> '*' STRING
+ var_spec -> '$' (STRING | NUMBER)
- This may all be followed by an "if EXPR", which we ignore.
+ basic_spec -> file_offset_spec | function_spec | label_spec
+ file_offset_spec -> opt_file_spec offset_spec
+ function_spec -> opt_file_spec function_name_spec opt_label_spec
+ label_spec -> label_name_spec
- FUNCTION may be an undebuggable function found in minimal symbol table.
+ opt_file_spec -> "" | file_name_spec ':'
+ opt_label_spec -> "" | ':' label_name_spec
+
+ file_name_spec -> STRING
+ function_name_spec -> STRING
+ label_name_spec -> STRING
+ function_name_spec -> STRING
+ offset_spec -> NUMBER
+ -> '+' NUMBER
+ -> '-' NUMBER
+
+ This may all be followed by several keywords such as "if EXPR",
+ which we ignore.
+
+ The function may be an undebuggable function found in minimal symbol table.
If the argument FUNFIRSTLINE is nonzero, we want the first line
of real code inside a function when a function is specified, and it is
@@ -889,334 +1863,205 @@ keep_name_info (char *p, int on_boundary)
if no file is validly specified. Callers must check that.
Also, the line number returned may be invalid. */
-/* We allow single quotes in various places. This is a hideous
- kludge, which exists because the completer can't yet deal with the
- lack of single quotes. FIXME: write a linespec_completer which we
- can use as appropriate instead of make_symbol_completion_list. */
+/* Parse the linespec in ARGPTR. */
static struct symtabs_and_lines
-decode_line_internal (struct linespec_state *self, char **argptr)
+parse_linespec (linespec_parser *parser, char **argptr)
{
- char *p;
- char *q;
-
- char *copy;
- /* This says whether or not something in *ARGPTR is quoted with
- completer_quotes (i.e. with single quotes). */
- int is_quoted;
- /* Is *ARGPTR enclosed in double quotes? */
- int is_quote_enclosed;
- int is_objc_method = 0;
- char *saved_arg = *argptr;
- /* If IS_QUOTED, the end of the quoted bit. */
- char *end_quote = NULL;
- /* Is *ARGPTR enclosed in single quotes? */
- int is_squote_enclosed = 0;
- /* The "first half" of the linespec. */
- char *first_half;
-
- /* If we are parsing `function:label', this holds the symbols
- matching the function name. */
- VEC (symbolp) *function_symbols = NULL;
- /* If FUNCTION_SYMBOLS is not NULL, then this is the exception that
- was thrown when trying to parse a filename. */
+ linespec_token token;
+ struct symtabs_and_lines values;
volatile struct gdb_exception file_exception;
+ struct cleanup *cleanup;
- struct cleanup *cleanup = make_cleanup (null_cleanup, NULL);
-
- /* Defaults have defaults. */
-
- initialize_defaults (&self->default_symtab, &self->default_line);
-
- /* See if arg is *PC. */
-
- if (**argptr == '*')
- {
- do_cleanups (cleanup);
- return decode_indirect (self, argptr);
- }
-
- is_quoted = (strchr (get_gdb_completer_quote_characters (),
- **argptr) != NULL);
-
- if (is_quoted)
- {
- end_quote = skip_quoted (*argptr);
- if (*end_quote == '\0')
- is_squote_enclosed = 1;
- }
-
- /* Check to see if it's a multipart linespec (with colons or
- periods). */
-
- /* Locate the end of the first half of the linespec.
- After the call, for instance, if the argptr string is "foo.c:123"
- p will point at ":123". If there is only one part, like "foo", p
- will point to "". If this is a C++ name, like "A::B::foo", p will
- point to "::B::foo". Argptr is not changed by this call. */
-
- first_half = p = locate_first_half (argptr, &is_quote_enclosed);
-
- /* First things first: if ARGPTR starts with a filename, get its
- symtab and strip the filename from ARGPTR.
- Avoid calling symtab_from_filename if we know can,
- it can be expensive. We know we can avoid the call if we see a
- single word (e.g., "break NAME") or if we see a qualified C++
- name ("break QUAL::NAME"). */
-
- if (*p != '\0' && !(p[0] == ':' && p[1] == ':'))
+ /* A special case to start. It has become quite popular for
+ IDEs to work around bugs in the previous parser by quoting
+ the entire linespec, so we attempt to deal with this nicely. */
+ parser->is_quote_enclosed = 0;
+ if (!is_ada_operator (*argptr)
+ && strchr (linespec_quote_characters, **argptr) != NULL)
{
- TRY_CATCH (file_exception, RETURN_MASK_ERROR)
- {
- self->file_symtabs = symtabs_from_filename (argptr, p,
- is_quote_enclosed,
- &self->user_filename);
- }
-
- if (file_exception.reason >= 0)
- {
- /* Check for single quotes on the non-filename part. */
- is_quoted = (**argptr
- && strchr (get_gdb_completer_quote_characters (),
- **argptr) != NULL);
- if (is_quoted)
- end_quote = skip_quoted (*argptr);
-
- /* Locate the next "half" of the linespec. */
- first_half = p = locate_first_half (argptr, &is_quote_enclosed);
- }
+ char *end;
- if (VEC_empty (symtab_p, self->file_symtabs))
+ end = skip_quote_char (*argptr + 1, **argptr);
+ if (end != NULL && end[1] == '\0')
{
- /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */
- VEC_safe_push (symtab_p, self->file_symtabs, NULL);
+ /* Here's the special case. Skip ARGPTR past the initial
+ quote. */
+ ++(*argptr);
+ parser->is_quote_enclosed = 1;
}
}
- else
- {
- /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */
- VEC_safe_push (symtab_p, self->file_symtabs, NULL);
- }
- /* Check if this is an Objective-C method (anything that starts with
- a '+' or '-' and a '['). */
- if (is_objc_method_format (p))
- is_objc_method = 1;
+ parser->lexer.saved_arg = *argptr;
+ parser->lexer.stream = argptr;
+ file_exception.reason = 0;
- /* Check if the symbol could be an Objective-C selector. */
+ /* Initialize the default symtab and line offset. */
+ initialize_defaults (&PARSER_STATE (parser)->default_symtab,
+ &PARSER_STATE (parser)->default_line);
- {
- struct symtabs_and_lines values;
+ /* Objective-C shortcut. */
+ values = decode_objc (PARSER_STATE (parser), PARSER_RESULT (parser), argptr);
+ if (values.sals != NULL)
+ return values;
- values = decode_objc (self, argptr);
- if (values.sals != NULL)
- {
- do_cleanups (cleanup);
- return values;
- }
- }
+ /* Start parsing. */
- /* Does it look like there actually were two parts? */
+ /* Get the first token. */
+ token = linespec_lexer_lex_one (parser);
- if (p[0] == ':' || p[0] == '.')
+ /* It must be either LSTOKEN_STRING or LSTOKEN_NUMBER. */
+ if (token.type == LSTOKEN_STRING && *LS_TOKEN_STOKEN (token).ptr == '*')
{
- /* Is it a C++ or Java compound data structure?
- The check on p[1] == ':' is capturing the case of "::",
- since p[0]==':' was checked above.
- Note that the call to decode_compound does everything
- for us, including the lookup on the symbol table, so we
- can return now. */
-
- if (p[0] == '.' || p[1] == ':')
- {
- /* We only perform this check for the languages where it might
- make sense. For instance, Ada does not use this type of
- syntax, and trying to apply this logic on an Ada linespec
- may trigger a spurious error (for instance, decode_compound
- does not like expressions such as `ops."<"', which is a
- valid function name in Ada). */
- if (current_language->la_language == language_c
- || current_language->la_language == language_cplus
- || current_language->la_language == language_java)
- {
- struct symtabs_and_lines values;
- volatile struct gdb_exception ex;
- char *saved_argptr = *argptr;
+ char *expr, *copy;
- if (is_quote_enclosed)
- ++saved_arg;
+ /* User specified an expression, *EXPR. */
+ copy = expr = copy_token_string (token);
+ cleanup = make_cleanup (xfree, expr);
+ PARSER_RESULT (parser)->expr_pc = linespec_expression_to_pc (©);
+ discard_cleanups (cleanup);
+ PARSER_RESULT (parser)->expression = expr;
- /* Initialize it just to avoid a GCC false warning. */
- memset (&values, 0, sizeof (values));
+ /* This is a little hacky/tricky. If linespec_expression_to_pc
+ did not evaluate the entire token, then we must find the
+ string COPY inside the original token buffer. */
+ if (*copy != '\0')
+ {
+ PARSER_STREAM (parser) = strstr (parser->lexer.saved_arg, copy);
+ gdb_assert (PARSER_STREAM (parser) != NULL);
+ }
- TRY_CATCH (ex, RETURN_MASK_ERROR)
- {
- values = decode_compound (self, argptr, saved_arg, p);
- }
- if ((is_quoted || is_squote_enclosed) && **argptr == '\'')
- *argptr = *argptr + 1;
+ /* Consume the token. */
+ linespec_lexer_consume_token (parser);
- if (ex.reason >= 0)
- {
- do_cleanups (cleanup);
- return values;
- }
+ goto convert_to_sals;
+ }
+ else if (token.type == LSTOKEN_STRING && *LS_TOKEN_STOKEN (token).ptr == '$')
+ {
+ char *var;
- if (ex.error != NOT_FOUND_ERROR)
- throw_exception (ex);
+ /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */
+ VEC_safe_push (symtab_p, PARSER_RESULT (parser)->file_symtabs, NULL);
- *argptr = saved_argptr;
- }
- }
- else
- {
- /* If there was an exception looking up a specified filename earlier,
- then check whether we were really given `function:label'. */
- if (file_exception.reason < 0)
- {
- function_symbols = find_function_symbols (argptr, p,
- is_quote_enclosed,
- &self->user_function);
+ /* User specified a convenience variable or history value. */
+ var = copy_token_string (token);
+ cleanup = make_cleanup (xfree, var);
+ PARSER_RESULT (parser)->line_offset
+ = linespec_parse_variable (PARSER_STATE (parser), var);
- /* If we did not find a function, re-throw the original
- exception. */
- if (!function_symbols)
- throw_exception (file_exception);
+ /* If a line_offset wasn't found (VAR is the name of a user
+ variable/function), then skip to normal symbol processing. */
+ if (PARSER_RESULT (parser)->line_offset.sign != unknown)
+ {
+ discard_cleanups (cleanup);
- make_cleanup (VEC_cleanup (symbolp), &function_symbols);
- }
+ /* Consume this token. */
+ linespec_lexer_consume_token (parser);
- /* Check for single quotes on the non-filename part. */
- if (!is_quoted)
- {
- is_quoted = (**argptr
- && strchr (get_gdb_completer_quote_characters (),
- **argptr) != NULL);
- if (is_quoted)
- end_quote = skip_quoted (*argptr);
- }
+ goto convert_to_sals;
}
- }
- /* self->file_symtabs holds the specified file symtabs, or 0 if no file
- specified.
- If we are parsing `function:symbol', then FUNCTION_SYMBOLS holds the
- functions before the `:'.
- arg no longer contains the file name. */
+ do_cleanups (cleanup);
+ }
+ else if (token.type != LSTOKEN_STRING && token.type != LSTOKEN_NUMBER)
+ unexpected_linespec_error (parser);
- /* If the filename was quoted, we must re-check the quotation. */
+ /* Shortcut: If the next token is not LSTOKEN_COLON, we know that
+ this token cannot represent a filename. */
+ token = linespec_lexer_peek_token (parser);
- if (end_quote == first_half && *end_quote!= '\0')
+ if (token.type == LSTOKEN_COLON)
{
- is_quoted = (**argptr
- && strchr (get_gdb_completer_quote_characters (),
- **argptr) != NULL);
- if (is_quoted)
- end_quote = skip_quoted (*argptr);
- }
+ char *user_filename;
- /* Check whether arg is all digits (and sign). */
+ /* Get the current token again and extract the filename. */
+ token = linespec_lexer_lex_one (parser);
+ user_filename = copy_token_string (token);
- q = *argptr;
- if (*q == '-' || *q == '+')
- q++;
- while (*q >= '0' && *q <= '9')
- q++;
+ /* Check if the input is a filename. */
+ TRY_CATCH (file_exception, RETURN_MASK_ERROR)
+ {
+ PARSER_RESULT (parser)->file_symtabs
+ = symtabs_from_filename (user_filename);
+ }
- if (q != *argptr && (*q == 0 || *q == ' ' || *q == '\t' || *q == ',')
- && function_symbols == NULL)
- {
- struct symtabs_and_lines values;
+ if (file_exception.reason >= 0)
+ {
+ /* Symtabs were found for the file. Record the filename. */
+ PARSER_RESULT (parser)->source_filename = user_filename;
- /* We found a token consisting of all digits -- at least one digit. */
- values = decode_all_digits (self, argptr, q);
- do_cleanups (cleanup);
- return values;
- }
+ /* Get the next token. */
+ token = linespec_lexer_consume_token (parser);
- /* Arg token is not digits => try it as a variable name
- Find the next token (everything up to end or next whitespace). */
+ /* This token must be LSTOKEN_COLON. */
+ if (token.type != LSTOKEN_COLON)
+ return values;
- if (**argptr == '$') /* May be a convenience variable. */
- /* One or two $ chars possible. */
- p = skip_quoted (*argptr + (((*argptr)[1] == '$') ? 2 : 1));
- else if (is_quoted || is_squote_enclosed)
- {
- p = end_quote;
- if (p[-1] != '\'')
- error (_("Unmatched single quote."));
+ /* Consume the LSTOKEN_COLON. */
+ linespec_lexer_consume_token (parser);
+ }
+ else
+ {
+ /* No symtabs found -- discard user_filename. */
+ xfree (user_filename);
+
+ /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */
+ VEC_safe_push (symtab_p, PARSER_RESULT (parser)->file_symtabs, NULL);
+ }
}
- else if (is_objc_method)
+ /* If the next token is not EOI or KEYWORD, issue an error.
+ Exception: In list mode, we allow the next token to be STRING,
+ as long as it starts with ',' (to accomodate ranges). */
+ else if (token.type != LSTOKEN_EOI && token.type != LSTOKEN_KEYWORD
+ && (!PARSER_STATE (parser)->list_mode
+ || (token.type == LSTOKEN_STRING
+ && *LS_TOKEN_STOKEN (token).ptr != ',')))
{
- /* allow word separators in method names for Obj-C. */
- p = skip_quoted_chars (*argptr, NULL, "");
+ /* TOKEN is the _next_ token, not the one currently in the parser.
+ Consuming the token will give the correct error message. */
+ linespec_lexer_consume_token (parser);
+ unexpected_linespec_error (parser);
}
else
{
- p = skip_quoted (*argptr);
- }
-
- /* Keep any important naming information. */
- p = keep_name_info (p, p == saved_arg || is_linespec_boundary (p[-1]));
-
- copy = (char *) alloca (p - *argptr + 1);
- memcpy (copy, *argptr, p - *argptr);
- copy[p - *argptr] = '\0';
- if (p != *argptr
- && copy[0]
- && copy[0] == copy[p - *argptr - 1]
- && strchr (get_gdb_completer_quote_characters (), copy[0]) != NULL)
- {
- copy[p - *argptr - 1] = '\0';
- copy++;
+ /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */
+ VEC_safe_push (symtab_p, PARSER_RESULT (parser)->file_symtabs, NULL);
}
- else if (is_quoted || is_squote_enclosed)
- copy[p - *argptr - 1] = '\0';
-
- *argptr = skip_spaces (p);
- /* If it starts with $: may be a legitimate variable or routine name
- (e.g. HP-UX millicode routines such as $$dyncall), or it may
- be history value, or it may be a convenience variable. */
+ /* Parse the rest of the linespec. */
+ linespec_parse_basic (parser);
- if (*copy == '$' && function_symbols == NULL)
+ if (PARSER_RESULT (parser)->function_symbols == NULL
+ && PARSER_RESULT (parser)->labels.label_symbols == NULL
+ && PARSER_RESULT (parser)->line_offset.sign == unknown
+ && PARSER_RESULT (parser)->minimal_symbols == NULL)
{
- struct symtabs_and_lines values;
+ /* The linespec didn't parse. Re-throw the file exception if
+ there was one. */
+ if (file_exception.reason < 0)
+ throw_exception (file_exception);
- values = decode_dollar (self, copy);
- do_cleanups (cleanup);
- return values;
- }
-
- /* Try the token as a label, but only if no file was specified,
- because we can only really find labels in the current scope. */
-
- if (VEC_length (symtab_p, self->file_symtabs) == 1
- && VEC_index (symtab_p, self->file_symtabs, 0) == NULL)
- {
- struct symtabs_and_lines label_result;
- if (decode_label (self, function_symbols, copy, &label_result))
- {
- do_cleanups (cleanup);
- return label_result;
- }
+ /* Otherwise, the symbol is not found. */
+ symbol_not_found_error (PARSER_RESULT (parser)->function_name,
+ PARSER_RESULT (parser)->source_filename);
}
- if (function_symbols)
- throw_exception (file_exception);
+ convert_to_sals:
- /* Look up that token as a variable.
- If file specified, use that file's per-file block to start with. */
+ /* Get the last token and record how much of the input was parsed,
+ if necessary. */
+ token = linespec_lexer_lex_one (parser);
+ if (token.type != LSTOKEN_EOI && token.type != LSTOKEN_KEYWORD)
+ PARSER_STREAM (parser) = LS_TOKEN_STOKEN (token).ptr;
- {
- struct symtabs_and_lines values;
+ /* Convert the data in PARSER_RESULT to SALs. */
+ values = convert_linespec_to_sals (PARSER_STATE (parser),
+ PARSER_RESULT (parser));
- values = decode_variable (self, copy);
- do_cleanups (cleanup);
- return values;
- }
+ return values;
}
+
/* A constructor for linespec_state. */
static void
@@ -1237,19 +2082,63 @@ linespec_state_constructor (struct linespec_state *self,
xfree, xcalloc, xfree);
}
+/* Initialize a new linespec parser. */
+static void
+linespec_parser_new (linespec_parser *parser,
+ int flags,
+ struct symtab *default_symtab,
+ int default_line,
+ struct linespec_result *canonical)
+{
+ parser->lexer.current.type = LSTOKEN_CONSUMED;
+ memset (PARSER_RESULT (parser), 0, sizeof (struct linespec));
+ PARSER_RESULT (parser)->line_offset.sign = unknown;
+ linespec_state_constructor (PARSER_STATE (parser), flags,
+ default_symtab, default_line, canonical);
+}
+
/* A destructor for linespec_state. */
static void
-linespec_state_destructor (void *arg)
+linespec_state_destructor (struct linespec_state *self)
{
- struct linespec_state *self = arg;
-
- xfree (self->user_filename);
- xfree (self->user_function);
- VEC_free (symtab_p, self->file_symtabs);
htab_delete (self->addr_set);
}
+/* Delete a linespec parser. */
+
+static void
+linespec_parser_delete (void *arg)
+{
+ linespec_parser *parser = (linespec_parser *) arg;
+
+ if (PARSER_RESULT (parser)->expression)
+ xfree (PARSER_RESULT (parser)->expression);
+ if (PARSER_RESULT (parser)->source_filename)
+ xfree (PARSER_RESULT (parser)->source_filename);
+ if (PARSER_RESULT (parser)->label_name)
+ xfree (PARSER_RESULT (parser)->label_name);
+ if (PARSER_RESULT (parser)->function_name)
+ xfree (PARSER_RESULT (parser)->function_name);
+
+ if (PARSER_RESULT (parser)->file_symtabs != NULL)
+ VEC_free (symtab_p, PARSER_RESULT (parser)->file_symtabs);
+
+ if (PARSER_RESULT (parser)->function_symbols != NULL)
+ VEC_free (symbolp, PARSER_RESULT (parser)->function_symbols);
+
+ if (PARSER_RESULT (parser)->minimal_symbols != NULL)
+ VEC_free (minsym_and_objfile_d, PARSER_RESULT (parser)->minimal_symbols);
+
+ if (PARSER_RESULT (parser)->labels.label_symbols != NULL)
+ VEC_free (symbolp, PARSER_RESULT (parser)->labels.label_symbols);
+
+ if (PARSER_RESULT (parser)->labels.function_symbols != NULL)
+ VEC_free (symbolp, PARSER_RESULT (parser)->labels.function_symbols);
+
+ linespec_state_destructor (PARSER_STATE (parser));
+}
+
/* See linespec.h. */
void
@@ -1260,10 +2149,11 @@ decode_line_full (char **argptr, int flags,
const char *filter)
{
struct symtabs_and_lines result;
- struct linespec_state state;
struct cleanup *cleanups;
char *arg_start = *argptr;
VEC (const_char_ptr) *filters = NULL;
+ linespec_parser parser;
+ struct linespec_state *state;
gdb_assert (canonical != NULL);
/* The filter only makes sense for 'all'. */
@@ -1274,12 +2164,12 @@ decode_line_full (char **argptr, int flags,
|| select_mode == multiple_symbols_cancel);
gdb_assert ((flags & DECODE_LINE_LIST_MODE) == 0);
- linespec_state_constructor (&state, flags,
- default_symtab, default_line, canonical);
- cleanups = make_cleanup (linespec_state_destructor, &state);
+ linespec_parser_new (&parser, flags, default_symtab, default_line, canonical);
+ cleanups = make_cleanup (linespec_parser_delete, &parser);
save_current_program_space ();
- result = decode_line_internal (&state, argptr);
+ result = parse_linespec (&parser, argptr);
+ state = PARSER_STATE (&parser);
gdb_assert (result.nelts == 1 || canonical->pre_expanded);
gdb_assert (canonical->addr_string != NULL);
@@ -1290,15 +2180,15 @@ decode_line_full (char **argptr, int flags,
{
int i;
- if (state.canonical_names == NULL)
- state.canonical_names = xcalloc (result.nelts, sizeof (char *));
- make_cleanup (xfree, state.canonical_names);
+ if (state->canonical_names == NULL)
+ state->canonical_names = xcalloc (result.nelts, sizeof (char *));
+ make_cleanup (xfree, state->canonical_names);
for (i = 0; i < result.nelts; ++i)
{
- if (state.canonical_names[i] == NULL)
- state.canonical_names[i] = savestring (arg_start,
- *argptr - arg_start);
- make_cleanup (xfree, state.canonical_names[i]);
+ if (state->canonical_names[i] == NULL)
+ state->canonical_names[i] = savestring (arg_start,
+ *argptr - arg_start);
+ make_cleanup (xfree, state->canonical_names[i]);
}
}
@@ -1316,13 +2206,13 @@ decode_line_full (char **argptr, int flags,
{
make_cleanup (VEC_cleanup (const_char_ptr), &filters);
VEC_safe_push (const_char_ptr, filters, filter);
- filter_results (&state, &result, filters);
+ filter_results (state, &result, filters);
}
else
- convert_results_to_lsals (&state, &result);
+ convert_results_to_lsals (state, &result);
}
else
- decode_line_2 (&state, &result, select_mode);
+ decode_line_2 (state, &result, select_mode);
do_cleanups (cleanups);
}
@@ -1333,15 +2223,15 @@ decode_line_1 (char **argptr, int flags,
int default_line)
{
struct symtabs_and_lines result;
- struct linespec_state state;
+ linespec_parser parser;
struct cleanup *cleanups;
- linespec_state_constructor (&state, flags,
- default_symtab, default_line, NULL);
- cleanups = make_cleanup (linespec_state_destructor, &state);
+ linespec_parser_new (&parser, flags, default_symtab, default_line, NULL);
+ cleanups = make_cleanup (linespec_parser_delete, &parser);
save_current_program_space ();
- result = decode_line_internal (&state, argptr);
+ result = parse_linespec (&parser, argptr);
+
do_cleanups (cleanups);
return result;
}
@@ -1369,178 +2259,20 @@ initialize_defaults (struct symtab **default_symtab, int *default_line)
\f
-/* Decode arg of the form *PC. */
+/* Evaluate the expression pointed to by EXP_PTR into a CORE_ADDR,
+ advancing EXP_PTR past any parsed text. */
-static struct symtabs_and_lines
-decode_indirect (struct linespec_state *self, char **argptr)
+static CORE_ADDR
+linespec_expression_to_pc (char **exp_ptr)
{
- struct symtabs_and_lines values;
- CORE_ADDR pc;
- char *initial = *argptr;
-
if (current_program_space->executing_startup)
/* The error message doesn't really matter, because this case
should only hit during breakpoint reset. */
throw_error (NOT_FOUND_ERROR, _("cannot evaluate expressions while "
"program space is in startup"));
- (*argptr)++;
- pc = value_as_address (parse_to_comma_and_eval (argptr));
-
- values.sals = (struct symtab_and_line *)
- xmalloc (sizeof (struct symtab_and_line));
-
- values.nelts = 1;
- values.sals[0] = find_pc_line (pc, 0);
- values.sals[0].pc = pc;
- values.sals[0].section = find_pc_overlay (pc);
- values.sals[0].explicit_pc = 1;
-
- if (self->canonical)
- self->canonical->addr_string = savestring (initial, *argptr - initial);
-
- return values;
-}
-
-\f
-
-/* Locate the first half of the linespec, ending in a colon, period,
- or whitespace. (More or less.) Also, check to see if *ARGPTR is
- enclosed in double quotes; if so, set is_quote_enclosed, advance
- ARGPTR past that and zero out the trailing double quote.
- If ARGPTR is just a simple name like "main", p will point to ""
- at the end. */
-
-static char *
-locate_first_half (char **argptr, int *is_quote_enclosed)
-{
- char *ii;
- char *p, *p1;
- int has_comma;
-
- /* Check if the linespec starts with an Ada operator (such as "+",
- or ">", for instance). */
- p = *argptr;
- if (p[0] == '"'
- && current_language->la_language == language_ada)
- {
- const struct ada_opname_map *op;
-
- for (op = ada_opname_table; op->encoded != NULL; op++)
- if (strncmp (op->decoded, p, strlen (op->decoded)) == 0)
- break;
- if (op->encoded != NULL)
- {
- *is_quote_enclosed = 0;
- return p + strlen (op->decoded);
- }
- }
-
- /* Maybe we were called with a line range FILENAME:LINENUM,FILENAME:LINENUM
- and we must isolate the first half. Outer layers will call again later
- for the second half.
-
- Don't count commas that appear in argument lists of overloaded
- functions, or in quoted strings. It's stupid to go to this much
- trouble when the rest of the function is such an obvious roach hotel. */
- ii = find_toplevel_char (*argptr, ',');
- has_comma = (ii != 0);
-
- /* Temporarily zap out second half to not confuse the code below.
- This is undone below. Do not change ii!! */
- if (has_comma)
- {
- *ii = '\0';
- }
-
- /* Maybe arg is FILE : LINENUM or FILE : FUNCTION. May also be
- CLASS::MEMBER, or NAMESPACE::NAME. Look for ':', but ignore
- inside of <>. */
-
- p = *argptr;
- if (p[0] == '"')
- {
- *is_quote_enclosed = 1;
- (*argptr)++;
- p++;
- }
- else
- {
- *is_quote_enclosed = 0;
- if (strchr (get_gdb_completer_quote_characters (), *p))
- {
- ++(*argptr);
- ++p;
- }
- }
-
-
- /* Check for a drive letter in the filename. This is done on all hosts
- to capture cross-compilation environments. On Unixen, directory
- separators are illegal in filenames, so if the user enters "e:/foo.c",
- he is referring to a directory named "e:" and a source file named
- "foo.c", and we still want to keep these two pieces together. */
- if (isalpha (p[0]) && p[1] == ':' && IS_DIR_SEPARATOR (p[2]))
- p += 3;
-
- for (; *p; p++)
- {
- if (p[0] == '<')
- {
- char *temp_end = find_template_name_end (p);
-
- if (!temp_end)
- error (_("malformed template specification in command"));
- p = temp_end;
- }
-
- if (p[0] == '(')
- p = find_method_overload_end (p);
-
- /* Check for a colon and a plus or minus and a [ (which
- indicates an Objective-C method). */
- if (is_objc_method_format (p))
- {
- break;
- }
- /* Check for the end of the first half of the linespec. End of
- line, a tab, a colon or a space. But if enclosed in double
- quotes we do not break on enclosed spaces. */
- if (!*p
- || p[0] == '\t'
- || (p[0] == ':')
- || ((p[0] == ' ') && !*is_quote_enclosed))
- break;
- if (p[0] == '.' && strchr (p, ':') == NULL)
- {
- /* Java qualified method. Find the *last* '.', since the
- others are package qualifiers. Stop at any open parenthesis
- which might provide overload information. */
- for (p1 = p; *p1 && *p1 != '('; p1++)
- {
- if (*p1 == '.')
- p = p1;
- }
- break;
- }
- }
- p = skip_spaces (p);
-
- /* If the closing double quote was left at the end, remove it. */
- if (*is_quote_enclosed)
- {
- char *closing_quote = strchr (p - 1, '"');
-
- if (closing_quote && closing_quote[1] == '\0')
- *closing_quote = '\0';
- }
-
- /* Now that we've safely parsed the first half, put back ',' so
- outer layers can see it. */
- if (has_comma)
- *ii = ',';
-
- return p;
+ (*exp_ptr)++;
+ return value_as_address (parse_to_comma_and_eval (exp_ptr));
}
\f
@@ -1553,28 +2285,34 @@ locate_first_half (char **argptr, int *is_quote_enclosed)
the existing C++ code to let the user choose one. */
static struct symtabs_and_lines
-decode_objc (struct linespec_state *self, char **argptr)
+decode_objc (struct linespec_state *self, linespec_t ls, char **argptr)
{
struct collect_info info;
VEC (const_char_ptr) *symbol_names = NULL;
+ struct symtabs_and_lines values;
char *new_argptr;
struct cleanup *cleanup = make_cleanup (VEC_cleanup (const_char_ptr),
&symbol_names);
info.state = self;
- info.result.sals = NULL;
- info.result.nelts = 0;
+ info.file_symtabs = NULL;
+ VEC_safe_push (symtab_p, info.file_symtabs, NULL);
+ info.result.symbols = NULL;
+ info.result.minimal_symbols = NULL;
+ values.nelts = 0;
+ values.sals = NULL;
new_argptr = find_imps (*argptr, &symbol_names);
if (VEC_empty (const_char_ptr, symbol_names))
{
do_cleanups (cleanup);
- return info.result;
+ return values;
}
add_all_symbol_names_from_pspace (&info, NULL, symbol_names);
- if (info.result.nelts > 0)
+ if (!VEC_empty (symbolp, info.result.symbols)
+ || !VEC_empty (minsym_and_objfile_d, info.result.minimal_symbols))
{
char *saved_arg;
@@ -1582,12 +2320,16 @@ decode_objc (struct linespec_state *self, char **argptr)
memcpy (saved_arg, *argptr, new_argptr - *argptr);
saved_arg[new_argptr - *argptr] = '\0';
+ ls->function_symbols = info.result.symbols;
+ ls->minimal_symbols = info.result.minimal_symbols;
+ values = convert_linespec_to_sals (self, ls);
+
if (self->canonical)
{
self->canonical->pre_expanded = 1;
- if (self->user_filename)
+ if (ls->source_filename)
self->canonical->addr_string
- = xstrprintf ("%s:%s", self->user_filename, saved_arg);
+ = xstrprintf ("%s:%s", ls->source_filename, saved_arg);
else
self->canonical->addr_string = xstrdup (saved_arg);
}
@@ -1596,241 +2338,8 @@ decode_objc (struct linespec_state *self, char **argptr)
*argptr = new_argptr;
do_cleanups (cleanup);
- return info.result;
-}
-
-/* This handles C++ and Java compound data structures. P should point
- at the first component separator, i.e. double-colon or period. As
- an example, on entrance to this function we could have ARGPTR
- pointing to "AAA::inA::fun" and P pointing to "::inA::fun". */
-
-static struct symtabs_and_lines
-decode_compound (struct linespec_state *self,
- char **argptr, char *the_real_saved_arg, char *p)
-{
- struct symtabs_and_lines values;
- char *p2;
- char *saved_arg2 = *argptr;
- char *temp_end;
- struct symbol *sym;
- char *copy;
- VEC (symbolp) *sym_classes;
- char *saved_arg, *class_name;
- struct cleanup *cleanup = make_cleanup (null_cleanup, NULL);
-
- /* If the user specified any completer quote characters in the input,
- strip them. They are superfluous. */
- saved_arg = alloca (strlen (the_real_saved_arg) + 1);
- {
- char *dst = saved_arg;
- char *src = the_real_saved_arg;
- char *quotes = get_gdb_completer_quote_characters ();
- while (*src != '\0')
- {
- if (strchr (quotes, *src) == NULL)
- *dst++ = *src;
- ++src;
- }
- *dst = '\0';
- }
-
- /* First check for "global" namespace specification, of the form
- "::foo". If found, skip over the colons and jump to normal
- symbol processing. I.e. the whole line specification starts with
- "::" (note the condition that *argptr == p). */
- if (p[0] == ':'
- && ((*argptr == p) || (p[-1] == ' ') || (p[-1] == '\t')))
- saved_arg2 += 2;
-
- /* Given our example "AAA::inA::fun", we have two cases to consider:
-
- 1) AAA::inA is the name of a class. In that case, presumably it
- has a method called "fun"; we then look up that method using
- find_method.
-
- 2) AAA::inA isn't the name of a class. In that case, either the
- user made a typo, AAA::inA is the name of a namespace, or it is
- the name of a minimal symbol.
- In this case we just delegate to decode_variable.
-
- Thus, our first task is to find everything before the last set of
- double-colons and figure out if it's the name of a class. So we
- first loop through all of the double-colons. */
-
- p2 = p; /* Save for restart. */
-
- /* This is very messy. Following the example above we have now the
- following pointers:
- p -> "::inA::fun"
- argptr -> "AAA::inA::fun
- saved_arg -> "AAA::inA::fun
- saved_arg2 -> "AAA::inA::fun
- p2 -> "::inA::fun". */
-
- /* In the loop below, with these strings, we'll make 2 passes, each
- is marked in comments. */
-
- while (1)
- {
- static char *break_characters = " \t(";
-
- /* Move pointer up to next possible class/namespace token. */
-
- p = p2 + 1; /* Restart with old value +1. */
-
- /* PASS1: at this point p2->"::inA::fun", so p->":inA::fun",
- i.e. if there is a double-colon, p will now point to the
- second colon. */
- /* PASS2: p2->"::fun", p->":fun" */
-
- /* Move pointer ahead to next double-colon. */
- while (*p
- && strchr (break_characters, *p) == NULL
- && strchr (get_gdb_completer_quote_characters (), *p) == NULL)
- {
- if (current_language->la_language == language_cplus)
- p += cp_validate_operator (p);
-
- if (p[0] == '<')
- {
- temp_end = find_template_name_end (p);
- if (!temp_end)
- error (_("malformed template specification in command"));
- p = temp_end;
- }
- /* Note that, since, at the start of this loop, p would be
- pointing to the second colon in a double-colon, we only
- satisfy the condition below if there is another
- double-colon to the right (after). I.e. there is another
- component that can be a class or a namespace. I.e, if at
- the beginning of this loop (PASS1), we had
- p->":inA::fun", we'll trigger this when p has been
- advanced to point to "::fun". */
- /* PASS2: we will not trigger this. */
- else if ((p[0] == ':') && (p[1] == ':'))
- break; /* Found double-colon. */
- else
- {
- /* PASS2: We'll keep getting here, until P points to one of the
- break characters, at which point we exit this loop. */
- if (*p)
- {
- if (p[1] == '('
- && strncmp (&p[1], CP_ANONYMOUS_NAMESPACE_STR,
- CP_ANONYMOUS_NAMESPACE_LEN) == 0)
- p += CP_ANONYMOUS_NAMESPACE_LEN;
- else if (strchr (break_characters, *p) == NULL)
- ++p;
- }
- }
- }
-
- if (*p != ':')
- break; /* Out of the while (1). This would happen
- for instance if we have looked up
- unsuccessfully all the components of the
- string, and p->""(PASS2). */
-
- /* We get here if p points to one of the break characters or "" (i.e.,
- string ended). */
- /* Save restart for next time around. */
- p2 = p;
- /* Restore argptr as it was on entry to this function. */
- *argptr = saved_arg2;
- /* PASS1: at this point p->"::fun" argptr->"AAA::inA::fun",
- p2->"::fun". */
-
- /* All ready for next pass through the loop. */
- } /* while (1) */
-
-
- /* Start of lookup in the symbol tables. */
-
- /* Lookup in the symbol table the substring between argptr and
- p. Note, this call changes the value of argptr. */
- /* Before the call, argptr->"AAA::inA::fun",
- p->"", p2->"::fun". After the call: argptr->"fun", p, p2
- unchanged. */
- sym_classes = lookup_prefix_sym (argptr, p2, self->file_symtabs,
- &class_name);
- make_cleanup (VEC_cleanup (symbolp), &sym_classes);
- make_cleanup (xfree, class_name);
-
- /* If a class has been found, then we're in case 1 above. So we
- look up "fun" as a method of those classes. */
- if (!VEC_empty (symbolp, sym_classes))
- {
- /* Arg token is not digits => try it as a function name.
- Find the next token (everything up to end or next
- blank). */
- if (**argptr
- && strchr (get_gdb_completer_quote_characters (),
- **argptr) != NULL)
- {
- p = skip_quoted (*argptr);
- *argptr = *argptr + 1;
- }
- else
- {
- /* At this point argptr->"fun". */
- char *a;
-
- p = *argptr;
- while (*p && *p != ' ' && *p != '\t' && *p != ',' && *p != ':'
- && *p != '(')
- p++;
- /* At this point p->"". String ended. */
- /* Nope, C++ operators could have spaces in them
- ("foo::operator <" or "foo::operator delete []").
- I apologize, this is a bit hacky... */
- if (current_language->la_language == language_cplus
- && *p == ' ' && p - 8 - *argptr + 1 > 0)
- {
- /* The above loop has already swallowed "operator". */
- p += cp_validate_operator (p - 8) - 8;
- }
-
- /* Keep any important naming information. */
- p = keep_name_info (p, 1);
- }
-
- /* Allocate our own copy of the substring between argptr and
- p. */
- copy = (char *) alloca (p - *argptr + 1);
- memcpy (copy, *argptr, p - *argptr);
- copy[p - *argptr] = '\0';
- if (p != *argptr
- && copy[p - *argptr - 1]
- && strchr (get_gdb_completer_quote_characters (),
- copy[p - *argptr - 1]) != NULL)
- copy[p - *argptr - 1] = '\0';
-
- /* At this point copy->"fun", p->"". */
-
- /* No line number may be specified. */
- *argptr = skip_spaces (p);
- /* At this point arptr->"". */
-
- /* Look for copy as a method of sym_class. */
- /* At this point copy->"fun", sym_class is "AAA:inA",
- saved_arg->"AAA::inA::fun". This concludes the scanning of
- the string for possible components matches. If we find it
- here, we return. If not, and we are at the and of the string,
- we'll lookup the whole string in the symbol tables. */
-
- values = find_method (self, saved_arg, copy, class_name, sym_classes);
-
- do_cleanups (cleanup);
- return values;
- } /* End if symbol found. */
-
- /* We couldn't find a class, so we're in case 2 above. We check the
- entire name as a symbol instead. The simplest way to do this is
- to just throw an exception and let our caller fall through to
- decode_variable. */
-
- throw_error (NOT_FOUND_ERROR, _("see caller, this text doesn't matter"));
+ return values;
}
/* An instance of this type is used when collecting prefix symbols for
@@ -1876,46 +2385,20 @@ collect_one_symbol (struct symbol *sym, void *d)
return 1; /* Continue iterating. */
}
-/* Return the symbol corresponding to the substring of *ARGPTR ending
- at P, allowing whitespace. Also, advance *ARGPTR past the symbol
- name in question, the compound object separator ("::" or "."), and
- whitespace. Note that *ARGPTR is changed whether or not the
- this call finds anything (i.e we return NULL). As an
- example, say ARGPTR is "AAA::inA::fun" and P is "::inA::fun". */
+/* Return any symbols corresponding to CLASS_NAME in FILE_SYMTABS. */
static VEC (symbolp) *
-lookup_prefix_sym (char **argptr, char *p, VEC (symtab_p) *file_symtabs,
- char **class_name)
+lookup_prefix_sym (struct linespec_state *state, VEC (symtab_p) *file_symtabs,
+ const char *class_name)
{
- char *p1;
- char *copy;
int ix;
struct symtab *elt;
struct decode_compound_collector collector;
struct cleanup *outer;
struct cleanup *cleanup;
- struct block *search_block;
-
- /* Extract the class name. */
- p1 = p;
- while (p != *argptr && p[-1] == ' ')
- --p;
- copy = (char *) xmalloc (p - *argptr + 1);
- memcpy (copy, *argptr, p - *argptr);
- copy[p - *argptr] = 0;
- *class_name = copy;
- outer = make_cleanup (xfree, copy);
-
- /* Discard the class name from the argptr. */
- p = p1 + (p1[0] == ':' ? 2 : 1);
- p = skip_spaces (p);
- *argptr = p;
-
- /* At this point p1->"::inA::fun", p->"inA::fun" copy->"AAA",
- argptr->"inA::fun". */
collector.symbols = NULL;
- make_cleanup (VEC_cleanup (symbolp), &collector.symbols);
+ outer = make_cleanup (VEC_cleanup (symbolp), &collector.symbols);
collector.unique_syms = htab_create_alloc (1, htab_hash_pointer,
htab_eq_pointer, NULL,
@@ -1926,10 +2409,10 @@ lookup_prefix_sym (char **argptr, char *p, VEC (symtab_p) *file_symtabs,
{
if (elt == NULL)
{
- iterate_over_all_matching_symtabs (copy, STRUCT_DOMAIN,
+ iterate_over_all_matching_symtabs (class_name, STRUCT_DOMAIN,
collect_one_symbol, &collector,
NULL, 0);
- iterate_over_all_matching_symtabs (copy, VAR_DOMAIN,
+ iterate_over_all_matching_symtabs (class_name, VAR_DOMAIN,
collect_one_symbol, &collector,
NULL, 0);
}
@@ -1942,9 +2425,9 @@ lookup_prefix_sym (char **argptr, char *p, VEC (symtab_p) *file_symtabs,
gdb_assert (!SYMTAB_PSPACE (elt)->executing_startup);
set_current_program_space (SYMTAB_PSPACE (elt));
search_block = get_search_block (elt);
- LA_ITERATE_OVER_SYMBOLS (search_block, copy, STRUCT_DOMAIN,
+ LA_ITERATE_OVER_SYMBOLS (search_block, class_name, STRUCT_DOMAIN,
collect_one_symbol, &collector);
- LA_ITERATE_OVER_SYMBOLS (search_block, copy, VAR_DOMAIN,
+ LA_ITERATE_OVER_SYMBOLS (search_block, class_name, VAR_DOMAIN,
collect_one_symbol, &collector);
}
}
@@ -1959,14 +2442,42 @@ lookup_prefix_sym (char **argptr, char *p, VEC (symtab_p) *file_symtabs,
symbols with the same program space end up next to each other. */
static int
-compare_symbols (const void *a, const void *b)
+compare_symbols (const void *a, const void *b)
+{
+ struct symbol * const *sa = a;
+ struct symbol * const *sb = b;
+ uintptr_t uia, uib;
+
+ uia = (uintptr_t) SYMTAB_PSPACE (SYMBOL_SYMTAB (*sa));
+ uib = (uintptr_t) SYMTAB_PSPACE (SYMBOL_SYMTAB (*sb));
+
+ if (uia < uib)
+ return -1;
+ if (uia > uib)
+ return 1;
+
+ uia = (uintptr_t) *sa;
+ uib = (uintptr_t) *sb;
+
+ if (uia < uib)
+ return -1;
+ if (uia > uib)
+ return 1;
+
+ return 0;
+}
+
+/* Like compare_symbols but for minimal symbols. */
+
+static int
+compare_msymbols (const void *a, const void *b)
{
- struct symbol * const *sa = a;
- struct symbol * const *sb = b;
+ struct minimal_symbol * const *sa = a;
+ struct minimal_symbol * const *sb = b;
uintptr_t uia, uib;
- uia = (uintptr_t) SYMTAB_PSPACE (SYMBOL_SYMTAB (*sa));
- uib = (uintptr_t) SYMTAB_PSPACE (SYMBOL_SYMTAB (*sb));
+ uia = (uintptr_t) SYMBOL_OBJ_SECTION (*sa)->objfile->pspace;
+ uib = (uintptr_t) SYMBOL_OBJ_SECTION (*sb)->objfile->pspace;
if (uia < uib)
return -1;
@@ -2031,14 +2542,16 @@ find_superclass_methods (VEC (typep) *superclasses,
do_cleanups (cleanup);
}
-/* This finds the method COPY in the class whose type is given by one
- of the symbols in SYM_CLASSES. */
+/* This finds the method METHOD_NAME in the class CLASS_NAME whose type is
+ given by one of the symbols in SYM_CLASSES. Matches are returned
+ in SYMBOLS (for debug symbols) and MINSYMS (for minimal symbols). */
-static struct symtabs_and_lines
-find_method (struct linespec_state *self, char *saved_arg,
- char *copy, const char *class_name, VEC (symbolp) *sym_classes)
+static void
+find_method (struct linespec_state *self, VEC (symtab_p) *file_symtabs,
+ const char *class_name, const char *method_name,
+ VEC (symbolp) *sym_classes, VEC (symbolp) **symbols,
+ VEC (minsym_and_objfile_d) **minsyms)
{
- char *canon;
struct symbol *sym;
struct cleanup *cleanup = make_cleanup (null_cleanup, NULL);
int ix;
@@ -2048,15 +2561,6 @@ find_method (struct linespec_state *self, char *saved_arg,
struct collect_info info;
char *name_iter;
- /* NAME is typed by the user: it needs to be canonicalized before
- searching the symbol tables. */
- canon = cp_canonicalize_string_no_typedefs (copy);
- if (canon != NULL)
- {
- copy = canon;
- make_cleanup (xfree, copy);
- }
-
/* Sort symbols so that symbols with the same program space are next
to each other. */
qsort (VEC_address (symbolp, sym_classes),
@@ -2065,11 +2569,12 @@ find_method (struct linespec_state *self, char *saved_arg,
compare_symbols);
info.state = self;
- info.result.sals = NULL;
- info.result.nelts = 0;
+ info.file_symtabs = file_symtabs;
+ info.result.symbols = NULL;
+ info.result.minimal_symbols = NULL;
/* Iterate over all the types, looking for the names of existing
- methods matching COPY. If we cannot find a direct method in a
+ methods matching METHOD_NAME. If we cannot find a direct method in a
given program space, then we consider inherited methods; this is
not ideal (ideal would be to respect C++ hiding rules), but it
seems good enough and is what GDB has historically done. We only
@@ -2093,7 +2598,7 @@ find_method (struct linespec_state *self, char *saved_arg,
pspace = SYMTAB_PSPACE (SYMBOL_SYMTAB (sym));
set_current_program_space (pspace);
t = check_typedef (SYMBOL_TYPE (sym));
- find_methods (t, copy, &result_names, &superclass_vec);
+ find_methods (t, method_name, &result_names, &superclass_vec);
/* Handle all items from a single program space at once; and be
sure not to miss the last batch. */
@@ -2105,7 +2610,8 @@ find_method (struct linespec_state *self, char *saved_arg,
/* If we did not find a direct implementation anywhere in
this program space, consider superclasses. */
if (VEC_length (const_char_ptr, result_names) == last_result_len)
- find_superclass_methods (superclass_vec, copy, &result_names);
+ find_superclass_methods (superclass_vec, method_name,
+ &result_names);
/* We have a list of candidate symbol names, so now we
iterate over the symbol tables looking for all
@@ -2117,31 +2623,18 @@ find_method (struct linespec_state *self, char *saved_arg,
}
}
- if (info.result.nelts > 0)
+ if (!VEC_empty (symbolp, info.result.symbols)
+ || !VEC_empty (minsym_and_objfile_d, info.result.minimal_symbols))
{
- if (self->canonical)
- {
- self->canonical->pre_expanded = 1;
- if (self->user_filename)
- self->canonical->addr_string
- = xstrprintf ("%s:%s", self->user_filename, saved_arg);
- else
- self->canonical->addr_string = xstrdup (saved_arg);
- }
-
+ *symbols = info.result.symbols;
+ *minsyms = info.result.minimal_symbols;
do_cleanups (cleanup);
-
- return info.result;
+ return;
}
- if (copy[0] == '~')
- cplusplus_error (saved_arg,
- "the class `%s' does not have destructor defined\n",
- class_name);
- else
- cplusplus_error (saved_arg,
- "the class %s does not have any method named %s\n",
- class_name, copy);
+ /* Throw an NOT_FOUND_ERROR. This will be caught by the caller
+ and other attempts to locate the symbol will be made. */
+ throw_error (NOT_FOUND_ERROR, _("see caller, this text doesn't matter"));
}
\f
@@ -2203,488 +2696,192 @@ collect_symtabs_from_filename (const char *file)
return collector.symtabs;
}
-/* Return all the symtabs associated to the filename given by the
- substring of *ARGPTR ending at P, and advance ARGPTR past that
- filename. */
-
-static VEC (symtab_p) *
-symtabs_from_filename (char **argptr, char *p, int is_quote_enclosed,
- char **user_filename)
-{
- char *p1;
- char *copy;
- struct cleanup *outer;
- VEC (symtab_p) *result;
-
- p1 = p;
- while (p != *argptr && p[-1] == ' ')
- --p;
- if ((*p == '"') && is_quote_enclosed)
- --p;
- copy = xmalloc (p - *argptr + 1);
- outer = make_cleanup (xfree, copy);
- memcpy (copy, *argptr, p - *argptr);
- /* It may have the ending quote right after the file name. */
- if ((is_quote_enclosed && copy[p - *argptr - 1] == '"')
- || copy[p - *argptr - 1] == '\'')
- copy[p - *argptr - 1] = 0;
- else
- copy[p - *argptr] = 0;
-
- result = collect_symtabs_from_filename (copy);
-
- if (VEC_empty (symtab_p, result))
- {
- if (!have_full_symbols () && !have_partial_symbols ())
- throw_error (NOT_FOUND_ERROR,
- _("No symbol table is loaded. "
- "Use the \"file\" command."));
- throw_error (NOT_FOUND_ERROR, _("No source file named %s."), copy);
- }
-
- /* Discard the file name from the arg. */
- if (*p1 == '\0')
- *argptr = p1;
- else
- *argptr = skip_spaces (p1 + 1);
-
- discard_cleanups (outer);
- *user_filename = copy;
- return result;
-}
-
-/* A callback used by iterate_over_all_matching_symtabs that collects
- symbols for find_function_symbols. */
-
-static int
-collect_function_symbols (struct symbol *sym, void *arg)
-{
- VEC (symbolp) **syms = arg;
-
- if (SYMBOL_CLASS (sym) == LOC_BLOCK)
- VEC_safe_push (symbolp, *syms, sym);
-
- return 1; /* Continue iterating. */
-}
-
-/* Look up a function symbol in *ARGPTR. If found, advance *ARGPTR
- and return the symbol. If not found, return NULL. */
-
-static VEC (symbolp) *
-find_function_symbols (char **argptr, char *p, int is_quote_enclosed,
- char **user_function)
-{
- char *p1;
- char *copy;
- VEC (symbolp) *result = NULL;
-
- p1 = p;
- while (p != *argptr && p[-1] == ' ')
- --p;
- if ((*p == '"') && is_quote_enclosed)
- --p;
- copy = (char *) xmalloc (p - *argptr + 1);
- *user_function = copy;
- memcpy (copy, *argptr, p - *argptr);
- /* It may have the ending quote right after the file name. */
- if ((is_quote_enclosed && copy[p - *argptr - 1] == '"')
- || copy[p - *argptr - 1] == '\'')
- copy[p - *argptr - 1] = 0;
- else
- copy[p - *argptr] = 0;
-
- iterate_over_all_matching_symtabs (copy, VAR_DOMAIN,
- collect_function_symbols, &result, NULL,
- 0);
-
- if (VEC_empty (symbolp, result))
- VEC_free (symbolp, result);
- else
- {
- /* Discard the file name from the arg. */
- *argptr = skip_spaces (p1 + 1);
- }
-
- return result;
-}
-
-\f
-
-/* A helper for decode_all_digits that handles the 'list_mode' case. */
-
-static void
-decode_digits_list_mode (struct linespec_state *self,
- struct symtabs_and_lines *values,
- struct symtab_and_line val)
-{
- int ix;
- struct symtab *elt;
-
- gdb_assert (self->list_mode);
-
- for (ix = 0; VEC_iterate (symtab_p, self->file_symtabs, ix, elt); ++ix)
- {
- /* The logic above should ensure this. */
- gdb_assert (elt != NULL);
-
- set_current_program_space (SYMTAB_PSPACE (elt));
-
- /* Simplistic search just for the list command. */
- val.symtab = find_line_symtab (elt, val.line, NULL, NULL);
- if (val.symtab == NULL)
- val.symtab = elt;
- val.pspace = SYMTAB_PSPACE (elt);
- val.pc = 0;
- val.explicit_line = 1;
-
- add_sal_to_sals (self, values, &val, NULL);
- }
-}
-
-/* A helper for decode_all_digits that iterates over the symtabs,
- adding lines to the VEC. */
-
-static void
-decode_digits_ordinary (struct linespec_state *self,
- int line,
- struct symtabs_and_lines *sals,
- struct linetable_entry **best_entry)
-{
- int ix;
- struct symtab *elt;
-
- for (ix = 0; VEC_iterate (symtab_p, self->file_symtabs, ix, elt); ++ix)
- {
- int i;
- VEC (CORE_ADDR) *pcs;
- CORE_ADDR pc;
-
- /* The logic above should ensure this. */
- gdb_assert (elt != NULL);
-
- set_current_program_space (SYMTAB_PSPACE (elt));
-
- pcs = find_pcs_for_symtab_line (elt, line, best_entry);
- for (i = 0; VEC_iterate (CORE_ADDR, pcs, i, pc); ++i)
- {
- struct symtab_and_line sal;
-
- init_sal (&sal);
- sal.pspace = SYMTAB_PSPACE (elt);
- sal.symtab = elt;
- sal.line = line;
- sal.pc = pc;
- add_sal_to_sals_basic (sals, &sal);
- }
-
- VEC_free (CORE_ADDR, pcs);
- }
-}
-
-/* This decodes a line where the argument is all digits (possibly
- preceded by a sign). Q should point to the end of those digits;
- the other arguments are as usual. */
-
-static struct symtabs_and_lines
-decode_all_digits (struct linespec_state *self,
- char **argptr,
- char *q)
-{
- struct symtabs_and_lines values;
- struct symtab_and_line val;
- int use_default = 0;
- char *saved_arg = *argptr;
-
- enum sign
- {
- none, plus, minus
- }
- sign = none;
-
- init_sal (&val);
- values.sals = NULL;
- values.nelts = 0;
-
- /* This is where we need to make sure that we have good defaults.
- We must guarantee that this section of code is never executed
- when we are called with just a function name, since
- set_default_source_symtab_and_line uses
- select_source_symtab that calls us with such an argument. */
-
- if (VEC_length (symtab_p, self->file_symtabs) == 1
- && VEC_index (symtab_p, self->file_symtabs, 0) == NULL)
- {
- set_current_program_space (self->program_space);
-
- /* Make sure we have at least a default source file. */
- set_default_source_symtab_and_line ();
- initialize_defaults (&self->default_symtab, &self->default_line);
- VEC_pop (symtab_p, self->file_symtabs);
- VEC_free (symtab_p, self->file_symtabs);
- self->file_symtabs
- = collect_symtabs_from_filename (self->default_symtab->filename);
- use_default = 1;
- }
-
- if (**argptr == '+')
- sign = plus, (*argptr)++;
- else if (**argptr == '-')
- sign = minus, (*argptr)++;
- val.line = atoi (*argptr);
- switch (sign)
- {
- case plus:
- if (q == *argptr)
- val.line = 5;
- if (use_default)
- val.line = self->default_line + val.line;
- break;
- case minus:
- if (q == *argptr)
- val.line = 15;
- if (use_default)
- val.line = self->default_line - val.line;
- else
- val.line = 1;
- break;
- case none:
- break; /* No need to adjust val.line. */
- }
-
- *argptr = skip_spaces (q);
-
- if (self->list_mode)
- decode_digits_list_mode (self, &values, val);
- else
- {
- struct linetable_entry *best_entry = NULL;
- int *filter;
- struct block **blocks;
- struct cleanup *cleanup;
- struct symtabs_and_lines intermediate_results;
- int i, j;
-
- intermediate_results.sals = NULL;
- intermediate_results.nelts = 0;
-
- decode_digits_ordinary (self, val.line, &intermediate_results,
- &best_entry);
- if (intermediate_results.nelts == 0 && best_entry != NULL)
- decode_digits_ordinary (self, best_entry->line, &intermediate_results,
- &best_entry);
-
- cleanup = make_cleanup (xfree, intermediate_results.sals);
-
- /* For optimized code, compiler can scatter one source line
- accross disjoint ranges of PC values, even when no duplicate
- functions or inline functions are involved. For example,
- 'for (;;)' inside non-template non-inline non-ctor-or-dtor
- function can result in two PC ranges. In this case, we don't
- want to set breakpoint on first PC of each range. To filter
- such cases, we use containing blocks -- for each PC found
- above we see if there are other PCs that are in the same
- block. If yes, the other PCs are filtered out. */
-
- filter = xmalloc (intermediate_results.nelts * sizeof (int));
- make_cleanup (xfree, filter);
- blocks = xmalloc (intermediate_results.nelts * sizeof (struct block *));
- make_cleanup (xfree, blocks);
-
- for (i = 0; i < intermediate_results.nelts; ++i)
- {
- set_current_program_space (intermediate_results.sals[i].pspace);
-
- filter[i] = 1;
- blocks[i] = block_for_pc_sect (intermediate_results.sals[i].pc,
- intermediate_results.sals[i].section);
- }
-
- for (i = 0; i < intermediate_results.nelts; ++i)
- {
- if (blocks[i] != NULL)
- for (j = i + 1; j < intermediate_results.nelts; ++j)
- {
- if (blocks[j] == blocks[i])
- {
- filter[j] = 0;
- break;
- }
- }
- }
-
- for (i = 0; i < intermediate_results.nelts; ++i)
- if (filter[i])
- {
- struct symbol *sym = (blocks[i]
- ? block_containing_function (blocks[i])
- : NULL);
-
- if (self->funfirstline)
- skip_prologue_sal (&intermediate_results.sals[i]);
- /* Make sure the line matches the request, not what was
- found. */
- intermediate_results.sals[i].line = val.line;
- add_sal_to_sals (self, &values, &intermediate_results.sals[i],
- sym ? SYMBOL_NATURAL_NAME (sym) : NULL);
- }
-
- do_cleanups (cleanup);
- }
-
- if (values.nelts == 0)
- {
- if (self->user_filename)
- throw_error (NOT_FOUND_ERROR, _("No line %d in file \"%s\"."),
- val.line, self->user_filename);
- else
- throw_error (NOT_FOUND_ERROR, _("No line %d in the current file."),
- val.line);
- }
+/* Return all the symtabs associated to the FILENAME. */
- if (self->canonical)
- {
- char *copy = savestring (saved_arg, q - saved_arg);
+static VEC (symtab_p) *
+symtabs_from_filename (const char *filename)
+{
+ VEC (symtab_p) *result;
+
+ result = collect_symtabs_from_filename (filename);
- self->canonical->pre_expanded = 1;
- gdb_assert (self->user_filename || use_default);
- self->canonical->addr_string
- = xstrprintf ("%s:%s", (self->user_filename
- ? self->user_filename
- : self->default_symtab->filename),
- copy);
- xfree (copy);
+ if (VEC_empty (symtab_p, result))
+ {
+ if (!have_full_symbols () && !have_partial_symbols ())
+ throw_error (NOT_FOUND_ERROR,
+ _("No symbol table is loaded. "
+ "Use the \"file\" command."));
+ throw_error (NOT_FOUND_ERROR, _("No source file named %s."), filename);
}
- return values;
+ return result;
}
-\f
-
-/* Decode a linespec starting with a dollar sign. */
+/* Look up a function symbol named NAME in symtabs FILE_SYMTABS. Matching
+ debug symbols are returned in SYMBOLS. Matching minimal symbols are
+ returned in MINSYMS. */
-static struct symtabs_and_lines
-decode_dollar (struct linespec_state *self, char *copy)
+static void
+find_function_symbols (struct linespec_state *state,
+ VEC (symtab_p) *file_symtabs, const char *name,
+ VEC (symbolp) **symbols,
+ VEC (minsym_and_objfile_d) **minsyms)
{
- LONGEST valx;
- int index = 0;
- struct symtabs_and_lines values;
- struct symtab_and_line val;
- char *p;
- struct symbol *sym;
- struct minimal_symbol *msymbol;
- int ix;
- struct symtab *elt;
+ struct collect_info info;
+ VEC (const_char_ptr) *symbol_names = NULL;
+ struct cleanup *cleanup = make_cleanup (VEC_cleanup (const_char_ptr),
+ &symbol_names);
- p = (copy[1] == '$') ? copy + 2 : copy + 1;
- while (*p >= '0' && *p <= '9')
- p++;
- if (!*p) /* Reached end of token without hitting non-digit. */
- {
- /* We have a value history reference. */
- struct value *val_history;
+ info.state = state;
+ info.result.symbols = NULL;
+ info.result.minimal_symbols = NULL;
+ info.file_symtabs = file_symtabs;
- sscanf ((copy[1] == '$') ? copy + 2 : copy + 1, "%d", &index);
- val_history = access_value_history ((copy[1] == '$') ? -index : index);
- if (TYPE_CODE (value_type (val_history)) != TYPE_CODE_INT)
- error (_("History values used in line "
- "specs must have integer values."));
- valx = value_as_long (val_history);
+ /* Try NAME as an Objective-C selector. */
+ find_imps ((char *) name, &symbol_names);
+ if (!VEC_empty (const_char_ptr, symbol_names))
+ add_all_symbol_names_from_pspace (&info, NULL, symbol_names);
+ else
+ add_matching_symbols_to_info (name, &info, NULL);
+
+ do_cleanups (cleanup);
+
+ if (VEC_empty (symbolp, info.result.symbols))
+ {
+ VEC_free (symbolp, info.result.symbols);
+ *symbols = NULL;
}
else
+ *symbols = info.result.symbols;
+
+ if (VEC_empty (minsym_and_objfile_d, info.result.minimal_symbols))
{
- /* Not all digits -- may be user variable/function or a
- convenience variable. */
+ VEC_free (minsym_and_objfile_d, info.result.minimal_symbols);
+ *minsyms = NULL;
+ }
+ else
+ *minsyms = info.result.minimal_symbols;
+}
+
+/* Find all symbols named NAME in FILE_SYMTABS, returning debug symbols
+ in SYMBOLS and minimal symbols in MINSYMS. */
- volatile struct gdb_exception exc;
+void
+find_linespec_symbols (struct linespec_state *state,
+ VEC (symtab_p) *file_symtabs,
+ const char *name,
+ VEC (symbolp) **symbols,
+ VEC (minsym_and_objfile_d) **minsyms)
+{
+ char *klass, *method, *canon;
+ const char *lookup_name, *last, *p, *scope_op;
+ struct cleanup *cleanup;
+ VEC (symbolp) *classes;
+ volatile struct gdb_exception except;
- /* Avoid "may be used uninitialized" warning. */
- values.sals = NULL;
- values.nelts = 0;
+ cleanup = demangle_for_lookup (name, current_language->la_language,
+ &lookup_name);
+ if (current_language->la_language == language_ada)
+ {
+ /* In Ada, the symbol lookups are performed using the encoded
+ name rather than the demangled name. */
+ lookup_name = ada_name_for_lookup (name);
+ make_cleanup (xfree, (void *) lookup_name);
+ }
- TRY_CATCH (exc, RETURN_MASK_ERROR)
- {
- values = decode_variable (self, copy);
- }
+ canon = cp_canonicalize_string_no_typedefs (lookup_name);
+ if (canon != NULL)
+ {
+ lookup_name = canon;
+ cleanup = make_cleanup (xfree, canon);
+ }
- if (exc.reason == 0)
- return values;
+ /* See if we can find a scope operator and break this symbol
+ name into namespaces${SCOPE_OPERATOR}class_name and method_name. */
+ scope_op = "::";
+ p = find_toplevel_string (lookup_name, scope_op);
+ if (p == NULL)
+ {
+ /* No C++ scope operator. Try Java. */
+ scope_op = ".";
+ p = find_toplevel_string (lookup_name, scope_op);
+ }
- if (exc.error != NOT_FOUND_ERROR)
- throw_exception (exc);
+ last = NULL;
+ while (p != NULL)
+ {
+ last = p;
+ p = find_toplevel_string (p + strlen (scope_op), scope_op);
+ }
- /* Not a user variable or function -- must be convenience variable. */
- if (!get_internalvar_integer (lookup_internalvar (copy + 1), &valx))
- error (_("Convenience variables used in line "
- "specs must have integer values."));
+ /* If no scope operator was found, lookup the name as a symbol. */
+ if (last == NULL)
+ {
+ find_function_symbols (state, file_symtabs, lookup_name,
+ symbols, minsyms);
+ do_cleanups (cleanup);
+ return;
}
- init_sal (&val);
+ /* NAME points to the class name.
+ LAST points to the method name. */
+ klass = xmalloc ((last - lookup_name + 1) * sizeof (char));
+ make_cleanup (xfree, klass);
+ strncpy (klass, lookup_name, last - lookup_name);
+ klass[last - lookup_name] = '\0';
- values.sals = NULL;
- values.nelts = 0;
+ /* Skip past the scope operator. */
+ last += strlen (scope_op);
+ method = xmalloc ((strlen (last) + 1) * sizeof (char));
+ make_cleanup (xfree, method);
+ strcpy (method, last);
- for (ix = 0; VEC_iterate (symtab_p, self->file_symtabs, ix, elt); ++ix)
+ /* Find a list of classes named KLASS. */
+ classes = lookup_prefix_sym (state, file_symtabs, klass);
+ if (!VEC_empty (symbolp, classes))
{
- if (elt == NULL)
+ /* Now locate a list of suitable methods named METHOD. */
+ TRY_CATCH (except, RETURN_MASK_ERROR)
{
- elt = self->default_symtab;
- set_current_program_space (self->program_space);
+ find_method (state, file_symtabs, klass, method, classes,
+ symbols, minsyms);
}
- else
- set_current_program_space (SYMTAB_PSPACE (elt));
-
- /* Either history value or convenience value from above, in valx. */
- val.symtab = elt;
- val.line = valx;
- val.pc = 0;
- val.pspace = elt ? SYMTAB_PSPACE (elt) : current_program_space;
-
- add_sal_to_sals (self, &values, &val, NULL);
- }
- if (self->canonical)
- {
- self->canonical->pre_expanded = 1;
- if (self->user_filename)
- self->canonical->addr_string = xstrprintf ("%s:%s",
- self->user_filename, copy);
- else
- self->canonical->addr_string = xstrdup (copy);
+ /* If successful, we're done. If NOT_FOUND_ERROR
+ was not thrown, rethrow the exception that we did get.
+ Otherwise, fall back to looking up the entire name as a symbol.
+ This can happen with namespace::function. */
+ if (except.reason >= 0)
+ {
+ do_cleanups (cleanup);
+ return;
+ }
+ else if (except.error != NOT_FOUND_ERROR)
+ throw_exception (except);
}
- return values;
+ /* We couldn't find a class, so we check the entire name as a symbol
+ instead. */
+ find_function_symbols (state, file_symtabs, lookup_name, symbols, minsyms);
+ do_cleanups (cleanup);
}
-\f
-
-/* A helper for decode_line_1 that tries to find a label. The label
- is searched for in the current block.
- FUNCTION_SYMBOLS is a list of the enclosing functions; or NULL if none
- specified.
- COPY is the name of the label to find.
- CANONICAL is the same as the "canonical" argument to decode_line_1.
- RESULT is a pointer to a symtabs_and_lines structure which will be
- filled in on success.
- This function returns 1 if a label was found, 0 otherwise. */
+/* Return all labels named NAME in FUNCTION_SYMBOLS. Return the
+ actual function symbol in which the label was found in LABEL_FUNC_RET. */
-static int
-decode_label (struct linespec_state *self,
- VEC (symbolp) *function_symbols, char *copy,
- struct symtabs_and_lines *result)
+static VEC (symbolp) *
+find_label_symbols (struct linespec_state *self,
+ VEC (symbolp) *function_symbols,
+ VEC (symbolp) **label_funcs_ret, const char *name)
{
- struct symbol *fn_sym;
int ix;
+ struct block *block;
+ struct symbol *sym;
+ struct symbol *fn_sym;
+ VEC (symbolp) *result = NULL;
if (function_symbols == NULL)
{
- struct block *block;
- struct symbol *sym;
- struct symtab_and_line sal;
- struct symtabs_and_lines values;
-
- values.nelts = 0;
- values.sals = NULL;
-
set_current_program_space (self->program_space);
block = get_search_block (NULL);
@@ -2693,69 +2890,174 @@ decode_label (struct linespec_state *self,
block = BLOCK_SUPERBLOCK (block))
;
if (!block)
- return 0;
+ return NULL;
fn_sym = BLOCK_FUNCTION (block);
- sym = lookup_symbol (copy, block, LABEL_DOMAIN, 0);
-
- if (sym == NULL)
- return 0;
+ sym = lookup_symbol (name, block, LABEL_DOMAIN, 0);
- symbol_to_sal (&sal, self->funfirstline, sym);
- add_sal_to_sals (self, &values, &sal,
- SYMBOL_NATURAL_NAME (fn_sym));
+ if (sym != NULL)
+ {
+ VEC_safe_push (symbolp, result, sym);
- if (self->canonical)
+ if (label_funcs_ret != NULL)
+ VEC_safe_push (symbolp, *label_funcs_ret, fn_sym);
+ }
+ }
+ else
+ {
+ for (ix = 0;
+ VEC_iterate (symbolp, function_symbols, ix, fn_sym); ++ix)
{
- self->canonical->special_display = 1;
- self->canonical->addr_string
- = xstrprintf ("%s:%s", SYMBOL_NATURAL_NAME (fn_sym),
- copy);
+ set_current_program_space (SYMTAB_PSPACE (SYMBOL_SYMTAB (fn_sym)));
+ block = SYMBOL_BLOCK_VALUE (fn_sym);
+ sym = lookup_symbol (name, block, LABEL_DOMAIN, 0);
+
+ if (sym != NULL)
+ {
+ VEC_safe_push (symbolp, result, sym);
+ if (label_funcs_ret != NULL)
+ VEC_safe_push (symbolp, *label_funcs_ret, fn_sym);
+ }
}
+ }
+
+ return result;
+}
- *result = values;
+\f
- return 1;
+/* A helper for create_sals_line_offset that handles the 'list_mode' case. */
+
+static void
+decode_digits_list_mode (struct linespec_state *self,
+ linespec_t ls,
+ struct symtabs_and_lines *values,
+ struct symtab_and_line val)
+{
+ int ix;
+ struct symtab *elt;
+
+ gdb_assert (self->list_mode);
+
+ for (ix = 0; VEC_iterate (symtab_p, ls->file_symtabs, ix, elt);
+ ++ix)
+ {
+ /* The logic above should ensure this. */
+ gdb_assert (elt != NULL);
+
+ set_current_program_space (SYMTAB_PSPACE (elt));
+
+ /* Simplistic search just for the list command. */
+ val.symtab = find_line_symtab (elt, val.line, NULL, NULL);
+ if (val.symtab == NULL)
+ val.symtab = elt;
+ val.pspace = SYMTAB_PSPACE (elt);
+ val.pc = 0;
+ val.explicit_line = 1;
+
+ add_sal_to_sals (self, values, &val, NULL);
}
+}
+
+/* A helper for create_sals_line_offset that iterates over the symtabs,
+ adding lines to the VEC. */
- result->sals = NULL;
- result->nelts = 0;
+static void
+decode_digits_ordinary (struct linespec_state *self,
+ linespec_t ls,
+ int line,
+ struct symtabs_and_lines *sals,
+ struct linetable_entry **best_entry)
+{
+ int ix;
+ struct symtab *elt;
- for (ix = 0; VEC_iterate (symbolp, function_symbols, ix, fn_sym); ++ix)
+ for (ix = 0; VEC_iterate (symtab_p, ls->file_symtabs, ix, elt); ++ix)
{
- struct block *block;
- struct symbol *sym;
+ int i;
+ VEC (CORE_ADDR) *pcs;
+ CORE_ADDR pc;
- set_current_program_space (SYMTAB_PSPACE (SYMBOL_SYMTAB (fn_sym)));
- block = SYMBOL_BLOCK_VALUE (fn_sym);
- sym = lookup_symbol (copy, block, LABEL_DOMAIN, 0);
+ /* The logic above should ensure this. */
+ gdb_assert (elt != NULL);
- if (sym != NULL)
+ set_current_program_space (SYMTAB_PSPACE (elt));
+
+ pcs = find_pcs_for_symtab_line (elt, line, best_entry);
+ for (i = 0; VEC_iterate (CORE_ADDR, pcs, i, pc); ++i)
{
struct symtab_and_line sal;
- char *symname;
-
- symbol_to_sal (&sal, self->funfirstline, sym);
- symname = xstrprintf ("%s:%s",
- SYMBOL_NATURAL_NAME (fn_sym),
- SYMBOL_NATURAL_NAME (sym));
- add_sal_to_sals (self, result, &sal, symname);
- xfree (symname);
+
+ init_sal (&sal);
+ sal.pspace = SYMTAB_PSPACE (elt);
+ sal.symtab = elt;
+ sal.line = line;
+ sal.pc = pc;
+ add_sal_to_sals_basic (sals, &sal);
}
+
+ VEC_free (CORE_ADDR, pcs);
}
+}
+
+\f
+
+/* Return the line offset represented by VARIABLE. */
+
+static struct line_offset
+linespec_parse_variable (struct linespec_state *self, const char *variable)
+{
+ int index = 0;
+ const char *p;
+ struct line_offset offset = {0, none};
- if (self->canonical && result->nelts > 0)
+ p = (variable[1] == '$') ? variable + 2 : variable + 1;
+ if (*p == '$')
+ ++p;
+ while (*p >= '0' && *p <= '9')
+ ++p;
+ if (!*p) /* Reached end of token without hitting non-digit. */
{
- self->canonical->pre_expanded = 1;
- self->canonical->special_display = 1;
+ /* We have a value history reference. */
+ struct value *val_history;
- gdb_assert (self->user_function);
- self->canonical->addr_string
- = xstrprintf ("%s:%s", self->user_function, copy);
+ sscanf ((variable[1] == '$') ? variable + 2 : variable + 1, "%d", &index);
+ val_history
+ = access_value_history ((variable[1] == '$') ? -index : index);
+ if (TYPE_CODE (value_type (val_history)) != TYPE_CODE_INT)
+ error (_("History values used in line "
+ "specs must have integer values."));
+ offset.offset = value_as_long (val_history);
+ }
+ else
+ {
+ /* Not all digits -- may be user variable/function or a
+ convenience variable. */
+ LONGEST valx;
+ struct internalvar *ivar;
+
+ /* Try it as a convenience variable. If it is not a convenience
+ variable, return and allow normal symbol lookup to occur. */
+ ivar = lookup_only_internalvar (variable + 1);
+ if (ivar == NULL)
+ /* No internal variable with that name. Mark the offset
+ as unknown to allow the name to be looked up as a symbol. */
+ offset.sign = unknown;
+ else
+ {
+ /* We found a valid variable name. If it is not an integer,
+ throw an error. */
+ if (!get_internalvar_integer (ivar, &valx))
+ error (_("Convenience variables used in line "
+ "specs must have integer values."));
+ else
+ offset.offset = valx;
+ }
}
- return result->nelts > 0;
+ return offset;
}
+\f
/* A callback used to possibly add a symbol to the results. */
@@ -2763,20 +3065,16 @@ static int
collect_symbols (struct symbol *sym, void *data)
{
struct collect_info *info = data;
- struct symtab_and_line sal;
-
- if (symbol_to_sal (&sal, info->state->funfirstline, sym)
- && maybe_add_address (info->state->addr_set,
- SYMTAB_PSPACE (SYMBOL_SYMTAB (sym)),
- sal.pc))
- add_sal_to_sals (info->state, &info->result, &sal,
- SYMBOL_NATURAL_NAME (sym));
+ /* In list mode, add all matching symbols, regardless of class.
+ This allows the user to type "list a_global_variable". */
+ if (SYMBOL_CLASS (sym) == LOC_BLOCK || info->state->list_mode)
+ VEC_safe_push (symbolp, info->result.symbols, sym);
return 1; /* Continue iterating. */
}
-/* We've found a minimal symbol MSYMBOL to associate with our
- linespec; add it to the result symtabs_and_lines. */
+/* We've found a minimal symbol MSYMBOL in OBJFILE to associate with our
+ linespec; return the SAL in RESULT. */
static void
minsym_found (struct linespec_state *self, struct objfile *objfile,
@@ -2804,17 +3102,6 @@ minsym_found (struct linespec_state *self, struct objfile *objfile,
add_sal_to_sals (self, result, &sal, SYMBOL_NATURAL_NAME (msymbol));
}
-/* A helper struct which just holds a minimal symbol and the object
- file from which it came. */
-
-typedef struct minsym_and_objfile
-{
- struct minimal_symbol *minsym;
- struct objfile *objfile;
-} minsym_and_objfile_d;
-
-DEF_VEC_O (minsym_and_objfile_d);
-
/* A helper struct to pass some data through
iterate_over_minimal_symbols. */
@@ -2967,8 +3254,8 @@ search_minsyms_for_name (struct collect_info *info, const char *name,
if (classify_mtype (MSYMBOL_TYPE (item->minsym)) != classification)
break;
- minsym_found (info->state, item->objfile, item->minsym,
- &info->result);
+ VEC_safe_push (minsym_and_objfile_d,
+ info->result.minimal_symbols, item);
}
}
@@ -2988,7 +3275,7 @@ add_matching_symbols_to_info (const char *name,
int ix;
struct symtab *elt;
- for (ix = 0; VEC_iterate (symtab_p, info->state->file_symtabs, ix, elt); ++ix)
+ for (ix = 0; VEC_iterate (symtab_p, info->file_symtabs, ix, elt); ++ix)
{
struct symbol *sym;
@@ -3012,67 +3299,6 @@ add_matching_symbols_to_info (const char *name,
}
}
-/* Decode a linespec that's a variable. If FILE_SYMTAB is non-NULL,
- look in that symtab's static variables first. */
-
-static struct symtabs_and_lines
-decode_variable (struct linespec_state *self, char *copy)
-{
- struct collect_info info;
- const char *lookup_name;
- char *canon;
- struct cleanup *cleanup;
-
- info.state = self;
- info.result.sals = NULL;
- info.result.nelts = 0;
-
- cleanup = demangle_for_lookup (copy, current_language->la_language,
- &lookup_name);
- if (current_language->la_language == language_ada)
- {
- /* In Ada, the symbol lookups are performed using the encoded
- name rather than the demangled name. */
- lookup_name = ada_name_for_lookup (copy);
- make_cleanup (xfree, (void *) lookup_name);
- }
-
- canon = cp_canonicalize_string_no_typedefs (lookup_name);
- if (canon != NULL)
- {
- make_cleanup (xfree, canon);
- lookup_name = canon;
- }
-
- add_matching_symbols_to_info (lookup_name, &info, NULL);
-
- if (info.result.nelts > 0)
- {
- if (self->canonical)
- {
- self->canonical->pre_expanded = 1;
- if (self->user_filename)
- self->canonical->addr_string
- = xstrprintf ("%s:%s", self->user_filename, copy);
- else
- self->canonical->addr_string = xstrdup (copy);
- }
- return info.result;
- }
-
- if (!have_full_symbols ()
- && !have_partial_symbols ()
- && !have_minimal_symbols ())
- throw_error (NOT_FOUND_ERROR,
- _("No symbol table is loaded. Use the \"file\" command."));
- if (self->user_filename)
- throw_error (NOT_FOUND_ERROR, _("Function \"%s\" not defined in \"%s\"."),
- copy, self->user_filename);
- else
- throw_error (NOT_FOUND_ERROR, _("Function \"%s\" not defined."), copy);
-}
-
-
\f
/* Now come some functions that are called from multiple places within
diff --git a/gdb/testsuite/gdb.base/advance.exp b/gdb/testsuite/gdb.base/advance.exp
index 456268d..617a7fb 100644
--- a/gdb/testsuite/gdb.base/advance.exp
+++ b/gdb/testsuite/gdb.base/advance.exp
@@ -45,7 +45,8 @@ gdb_test "advance [gdb_get_line_number "advance this location"]" \
# Verify that a malformed "advance" is gracefully caught.
#
gdb_test "advance [gdb_get_line_number "advance malformed"] then stop" \
- "Junk at end of arguments." "malformed advance"
+ "malformed linespec error: unexpected string, \"then stop\"" \
+ "malformed advance"
# Verify that "advance <funcname>" works.
#
diff --git a/gdb/testsuite/gdb.base/break.exp b/gdb/testsuite/gdb.base/break.exp
index 71eb3a6..a203364 100644
--- a/gdb/testsuite/gdb.base/break.exp
+++ b/gdb/testsuite/gdb.base/break.exp
@@ -600,7 +600,7 @@ gdb_test "break $bp_location12 thread foo" \
# trailing garbage.
#
gdb_test "break $bp_location12 foo" \
- "Junk at end of arguments.*" \
+ "malformed linespec error: unexpected string, \"foo\".*" \
"breakpoint with trailing garbage disallowed"
# Verify that GDB responds gracefully to a "clear" command that has
diff --git a/gdb/testsuite/gdb.base/hbreak2.exp b/gdb/testsuite/gdb.base/hbreak2.exp
index 7f52cd3..a11d8b6 100644
--- a/gdb/testsuite/gdb.base/hbreak2.exp
+++ b/gdb/testsuite/gdb.base/hbreak2.exp
@@ -356,7 +356,7 @@ gdb_test "hbreak $bp_location12 thread foo" \
# trailing garbage.
#
gdb_test "hbreak $bp_location12 foo" \
- "Junk at end of arguments.*" \
+ "malformed linespec error: unexpected string, \"foo\".*" \
"hardware breakpoint with trailing garbage disallowed"
# Verify that GDB responds gracefully to a "clear" command that has
diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp
index a5577e2..a7b2177 100644
--- a/gdb/testsuite/gdb.base/jump.exp
+++ b/gdb/testsuite/gdb.base/jump.exp
@@ -92,7 +92,7 @@ gdb_test "jump" "Argument required .starting address.*" \
# trailing junk.
#
gdb_test "jump 21 100" \
- "Junk at end of line specification: 100.*" \
+ "malformed linespec error: unexpected number, \"100\"" \
"jump with trailing argument junk"
diff --git a/gdb/testsuite/gdb.base/sepdebug.exp b/gdb/testsuite/gdb.base/sepdebug.exp
index bd60c70..e6973fb 100644
--- a/gdb/testsuite/gdb.base/sepdebug.exp
+++ b/gdb/testsuite/gdb.base/sepdebug.exp
@@ -408,7 +408,7 @@ gdb_test "break $bp_location12 thread foo" \
#
gdb_test "break $bp_location12 foo" \
- "Junk at end of arguments.*" \
+ "malformed linespec error: unexpected string, \"foo\".*" \
"breakpoint with trailing garbage disallowed"
# Verify that GDB responds gracefully to a "clear" command that has
diff --git a/gdb/testsuite/gdb.base/until.exp b/gdb/testsuite/gdb.base/until.exp
index 6b647cd..bdee21d 100644
--- a/gdb/testsuite/gdb.base/until.exp
+++ b/gdb/testsuite/gdb.base/until.exp
@@ -40,7 +40,8 @@ gdb_test "until $bp_location1" \
# Verify that a malformed "advance" is gracefully caught.
#
gdb_test "until 80 then stop" \
- "Junk at end of arguments." "malformed until"
+ "malformed linespec error: unexpected string, \"then stop\"." \
+ "malformed until"
# Rerun up to factorial, outer invocation
if { ![runto factorial] } then { gdb_suppress_tests; }
diff --git a/gdb/testsuite/gdb.cp/ovldbreak.exp b/gdb/testsuite/gdb.cp/ovldbreak.exp
index 3e35b79..2c27f2d 100644
--- a/gdb/testsuite/gdb.cp/ovldbreak.exp
+++ b/gdb/testsuite/gdb.cp/ovldbreak.exp
@@ -320,23 +320,17 @@ gdb_expect {
}
}
-gdb_test "info break" \
- "Num Type\[\t \]+Disp Enb Address\[\t \]+What.*
-\[0-9\]+\[\t \]+breakpoint keep y\[\t \]+<MULTIPLE>\[\t \]*\r
-\[0-9\]+.1\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(double\\) at.*$srcfile:140\r
-\[0-9\]+.2\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(float\\) at.*$srcfile:137\r
-\[0-9\]+.3\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\((unsigned long|long unsigned)( int)?\\) at.*$srcfile:134\r
-\[0-9\]+.4\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(long( int)?\\) at.*$srcfile:131\r
-\[0-9\]+.5\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\((unsigned|unsigned int)\\) at.*$srcfile:128\r
-\[0-9\]+.6\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(int\\) at.*$srcfile:125\r
-\[0-9\]+.7\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\((unsigned short|short unsigned)( int)?\\) at.*$srcfile:122\r
-\[0-9\]+.8\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(short( int)?\\) at.*$srcfile:119\r
-\[0-9\]+.9\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(unsigned char\\) at.*$srcfile:116\r
-\[0-9\]+.10\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(signed char\\) at.*$srcfile:113\r
-\[0-9\]+.11\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(char\\) at.*$srcfile:110\r
-\[0-9\]+.12\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\((void|)\\) at.*$srcfile:107" \
- "breakpoint info (after setting on all)"
+# Create the breakpoint table for "info breakpoint".
+set bptable "Num Type\[\t \]+Disp Enb Address\[\t \]+What.*\[\r\n]+"
+append bptable "\[0-9\]+\[\t \]+breakpoint\[\t \]+keep\[\t \]y\[\t \]+<MULTIPLE>.*\[\r\n\]+"
+foreach ovld {void char signed_char unsigned_char short_int \
+ unsigned_short_int int unsigned_int long_int \
+ unsigned_long_int float double} {
+ append bptable [format "\[0-9\]+.\[0-9\]+\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(%s\\) at.*$srcfile:%d\[\r\n\]+" \
+ $types($ovld) $line($ovld)]
+}
+gdb_test "info break" $bptable "breakpoint info (after setting on all)"
# Run through each breakpoint.
proc continue_to_bp_overloaded {bpnumber might_fail line argtype argument} {
diff --git a/gdb/testsuite/gdb.linespec/ls-errs.c b/gdb/testsuite/gdb.linespec/ls-errs.c
new file mode 100644
index 0000000..bb4c096
--- /dev/null
+++ b/gdb/testsuite/gdb.linespec/ls-errs.c
@@ -0,0 +1,29 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2012 Free Software Foundation, Inc.
+
+ 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/>. */
+
+int myfunction (void) { return 0; }
+
+int
+main (void)
+{
+ int a;
+
+ a = myfunction ();
+
+ here:
+ return a;
+}
diff --git a/gdb/testsuite/gdb.linespec/ls-errs.exp b/gdb/testsuite/gdb.linespec/ls-errs.exp
new file mode 100644
index 0000000..ce83807
--- /dev/null
+++ b/gdb/testsuite/gdb.linespec/ls-errs.exp
@@ -0,0 +1,189 @@
+# Copyright 2012 Free Software Foundation, Inc.
+
+# 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/>.
+
+# Tests for linespec error conditions
+
+set base ls-errs
+set srcfile "$base.c"
+set testfile "$base.exp"
+set exefile $base
+
+if {[prepare_for_testing $testfile $exefile $srcfile \
+ {debug nowarnings}]} {
+ return -1
+}
+
+# Turn off the pending breakpoint queries.
+gdb_test_no_output "set breakpoint pending off"
+
+# We intentionally do not use gdb_breakpoint for these tests.
+
+# Add the (invalid) LINESPEC to the test array named in ARRAY_NAME.
+# Use the index into ::error_messages MSG_ID and ARGS to create
+# an error message which is the expect result of attempting to
+# break on the given LINESPEC.
+proc add {array_name linespec msg_id args} {
+ global error_messages
+ upvar $array_name tests
+
+ lappend tests(linespecs) $linespec
+ set tests("$linespec") [string_to_regexp \
+ [eval format \$error_messages($msg_id) $args]]
+}
+
+# Common error message format strings.
+array set error_messages {
+ invalid_file "No source file named %s."
+ invalid_function "Function \"%s\" not defined."
+ invalid_var_or_func "Undefined convenience variable or function \"%s\" not defined."
+ invalid_function_f "Function \"%s\" not defined in \"%s\"."
+ invalid_var_or_func_f \
+ "Undefined convenience variable or function \"%s\" not defined in \"%s\"."
+ invalid_label "No label \"%s\" defined in function \"%s\"."
+ invalid_offset "No line %d in the current file."
+ invalid_offset_f "No line %d in file \"%s\"."
+ unexpected "malformed linespec error: unexpected %s"
+ unexpected_opt "malformed linespec error: unexpected %s, \"%s\""
+ unmatched_quote "unmatched quote"
+}
+
+# Some commonly used whitespace tests around ':'.
+set spaces [list ":" ": " " :" " : " "\t: " " :\t" "\t:\t" " \t:\t " \
+ "\t \t:\t \t \t"]
+
+# A list of invalid offsets.
+set invalid_offsets [list -100 +500 1000]
+
+# THE_TESTS will hold all of our test information. Array index
+# "linespecs" will contain the complete list of all linespecs
+# to be tested. An array index of \"$linespec\" will contain
+# the expected result.
+set the_tests(linespecs) {}
+
+# Try some simple, invalid linespecs involving spaces.
+foreach x $spaces {
+ add the_tests $x unexpected "colon"
+}
+
+# Test invalid filespecs starting with offset. This is done
+# first so that default offsets are tested.
+foreach x $invalid_offsets {
+ set offset $x
+
+ # Relative offsets are relative to line 16. Adjust
+ # expected offset from error message accordingly.
+ if {[string index $x 0] == "+" ||
+ [string index $x 0] == "-"} {
+ incr offset 16
+ }
+ add the_tests $x invalid_offset $offset
+}
+
+# Test offsets with trailing tokens w/ and w/o spaces.
+foreach x $spaces {
+ add the_tests "3$x" unexpected "colon"
+ add the_tests "+10$x" unexpected "colon"
+ add the_tests "-10$x" unexpected "colon"
+}
+
+foreach x {1 +1 +100 -10} {
+ add the_tests "3 $x" unexpected_opt "number" $x
+ add the_tests "+10 $x" unexpected_opt "number" $x
+ add the_tests "-10 $x" unexpected_opt "number" $x
+}
+
+add the_tests "3 foo" unexpected_opt "string" "foo"
+add the_tests "+10 foo" unexpected_opt "string" "foo"
+add the_tests "-10 foo" unexpected_opt "string" "foo"
+
+# Test invalid linespecs starting with filename.
+foreach x [list "this_file_doesn't_exist.c" \
+ "this file has spaces.c" \
+ "\"file::colons.c\"" \
+ "'file::colons.c'" \
+ "\"this \"file\" has quotes.c\"" \
+ "'this \"file\" has quotes.c'" \
+ "'this 'file' has quotes.c'" \
+ "\"this 'file' has quotes.c\""] {
+ # Remove any quoting from FILENAME for the error message.
+ add the_tests "$x:3" invalid_file [string trim $x \"']
+}
+
+# Test unmatched quotes.
+foreach x {"\"src-file.c'" "'src-file.c"} {
+ add the_tests "$x:3" unmatched_quote
+}
+
+add the_tests $srcfile invalid_function $srcfile
+foreach x {"foo" " foo" " foo "} {
+ # Trim any leading/trailing whitespace for error messages.
+ add the_tests "$srcfile:$x" invalid_function_f [string trim $x] $srcfile
+ add the_tests "$srcfile:main:$x" invalid_label [string trim $x] "main"
+}
+
+foreach x $spaces {
+ add the_tests "$srcfile$x" unexpected "end of input"
+ add the_tests "$srcfile:main$x" unexpected "end of input"
+}
+
+add the_tests "${srcfile}::" invalid_function "${srcfile}::"
+add the_tests "$srcfile:3 1" unexpected_opt "number" "1"
+add the_tests "$srcfile:3 +100" unexpected_opt "number" "+100"
+add the_tests "$srcfile:3 -100" unexpected_opt "number" "-100"
+add the_tests "$srcfile:3 foo" unexpected_opt "string" "foo"
+
+foreach x $invalid_offsets {
+ add the_tests "$srcfile:$x" invalid_offset_f $x $srcfile
+}
+
+# Test invalid filespecs starting with function.
+foreach x {"foobar" "foo::bar" "foo.bar" "foo ." "foo bar" "foo 1" \
+ "foo 0" "foo +10" "foo -10" "foo +100" "foo -100"} {
+ add the_tests $x invalid_function $x
+}
+
+foreach x $spaces {
+ add the_tests "main${x}there" invalid_label "there" "main"
+ add the_tests "main:here${x}" unexpected "end of input"
+}
+
+add the_tests "main 3" invalid_function "main 3"
+add the_tests "main +100" invalid_function "main +100"
+add the_tests "main -100" invalid_function "main -100"
+add the_tests "main foo" invalid_function "main foo"
+
+foreach x {"3" "+100" "-100" "foo"} {
+ add the_tests "main:here $x" invalid_label "here $x" "main"
+}
+
+foreach x {"if" "task" "thread"} {
+ add the_tests $x unexpected_opt "keyword" $x
+}
+
+add the_tests "'main.c'flubber" unexpected_opt "string" "flubber"
+add the_tests "'main.c',21" unexpected_opt "string" ",21"
+add the_tests "'main.c' " invalid_function "main.c"
+add the_tests "'main.c'3" unexpected_opt "number" "3"
+add the_tests "'main.c'+3" unexpected_opt "number" "+3"
+
+# Test undefined convenience variables.
+set x {$zippo}
+add the_tests $x invalid_var_or_func $x
+add the_tests "$srcfile:$x" invalid_var_or_func_f $x $srcfile
+
+# Run the tests
+foreach linespec $the_tests(linespecs) {
+ gdb_test "break $linespec" $the_tests("$linespec")
+}
diff --git a/gdb/testsuite/gdb.cp/cplabel.cc b/gdb/testsuite/gdb.cp/cplabel.cc
new file mode 100644
index 0000000..876f802
--- /dev/null
+++ b/gdb/testsuite/gdb.cp/cplabel.cc
@@ -0,0 +1,80 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2012 Free Software Foundation, Inc.
+
+ 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/>. */
+
+class foo
+{
+public:
+ static int bar (void)
+ {
+ int i = 5;
+ bool first = true;
+
+ to_the_top: /* bar:to_the_top */
+ while (1)
+ {
+ if (i == 1)
+ {
+ if (first)
+ {
+ first = false;
+ goto to_the_top;
+ }
+ else
+ goto get_out_of_here;
+ }
+
+ --i;
+ }
+
+ get_out_of_here: /* bar:get_out_of_here */
+ return i;
+ }
+
+ int baz (int a)
+ {
+ int i = a;
+ bool first = true;
+
+ to_the_top: /* baz:to_the_top */
+ while (1)
+ {
+ if (i == 1)
+ {
+ if (first)
+ {
+ first = false;
+ goto to_the_top;
+ }
+ else
+ goto get_out_of_here;
+ }
+
+ --i;
+ }
+
+ get_out_of_here: /* baz:get_out_of_here */
+ return i;
+ }
+};
+
+int
+main (void)
+{
+ foo f;
+ return f.baz (foo::bar () + 3);
+}
+
diff --git a/gdb/testsuite/gdb.cp/cplabel.exp b/gdb/testsuite/gdb.cp/cplabel.exp
new file mode 100644
index 0000000..4f99b55
--- /dev/null
+++ b/gdb/testsuite/gdb.cp/cplabel.exp
@@ -0,0 +1,38 @@
+# Copyright 2012 Free Software Foundation, Inc.
+
+# 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/>.
+
+# Tests for breakpoint on labels in methods.
+
+set testfile cplabel
+set srcfile "$testfile.cc"
+if {[prepare_for_testing "$testfile.exp" $testfile $srcfile]} {
+ return -1
+}
+
+if {![runto_main]} {
+ untested "could not run to main"
+ return -1
+}
+
+set methods {"bar" "baz"}
+set labels {"to_the_top" "get_out_of_here"}
+
+foreach m $methods {
+ foreach l $labels {
+ set line [gdb_get_line_number "$m:$l"]
+ gdb_test "break foo::$m:$l" \
+ "Breakpoint $decimal at $hex: file .*/$srcfile, line $line\."
+ }
+}
diff --git a/gdb/testsuite/gdb.cp/userdef.exp b/gdb/testsuite/gdb.cp/userdef.exp
index 9c7cb57..90eb99d 100644
--- a/gdb/testsuite/gdb.cp/userdef.exp
+++ b/gdb/testsuite/gdb.cp/userdef.exp
@@ -136,8 +136,8 @@ gdb_test "print one += 7" "\\\$\[0-9\]* = {x = 9, y = 10}"
gdb_test "print two = one" "\\\$\[0-9\]* = {x = 9, y = 10}"
# Check that GDB tolerates whitespace in operator names.
-gdb_test "break A2::'operator+'" ".*Breakpoint $decimal at.*"
-gdb_test "break A2::'operator +'" ".*Breakpoint $decimal at.*"
+gdb_test "break A2::operator+" ".*Breakpoint $decimal at.*"
+gdb_test "break A2::operator +" ".*Breakpoint $decimal at.*"
# Check that GDB handles operator* correctly.
gdb_test "print c" "\\\$\[0-9\]* = {m = {z = .*}}"
diff --git a/gdb/testsuite/gdb.linespec/ls-dollar.cc b/gdb/testsuite/gdb.linespec/ls-dollar.cc
^ permalink raw reply [flat|nested] 26+ messages in thread* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-26 23:18 [RFA 1/2] Linespec rewrite (update 2) Keith Seitz @ 2012-03-27 1:45 ` asmwarrior 2012-03-28 18:55 ` Tom Tromey 2012-03-27 13:56 ` Joel Brobecker ` (2 subsequent siblings) 3 siblings, 1 reply; 26+ messages in thread From: asmwarrior @ 2012-03-27 1:45 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml Hi, Keith. Thanks for the improvement. I'm testing your line spec rewrite branch, it works great. When I review the log in gdb CVS HEAD: http://sourceware.org/ml/gdb-patches/2012-03/msg00819.html [debug]> break "E:/code/cb/cb_trunk/src/plugins/codecompletion/parser/token.cpp:64" [debug]No source file named E:/code/cb/cb_trunk/src/plugins/codecompletion/parser/token.cpp. [debug]Breakpoint 2 ("E:/code/cb/cb_trunk/src/plugins/codecompletion/parser/token.cpp:64) pending. See the third line, the response from GDB does not have quote balanced. I mean, the tail quote is lost. So, to catch such string, we have currently use the regular expression in Code::blocks' source code: static wxRegEx rePendingBreakpoint(_T("Breakpoint ([0-9]+)[ \t]\\(\\\"(.+):([0-9]+)\\)[ \t]pending\\.")); Now, I'm using your branch, and the result is: [debug]> break "E:/code/cb/test_code/DebugDLLTest/TestDLL/dllmain.cpp:29" [debug]No source file named E:/code/cb/test_code/DebugDLLTest/TestDLL/dllmain.cpp. [debug]Breakpoint 4 ("E:/code/cb/test_code/DebugDLLTest/TestDLL/dllmain.cpp:29") pending. You can see, the third line, it does have a tail quote. I believe adding the tail quote is the right direction. Also, I can easily adjust the regex to: static wxRegEx rePendingBreakpoint(_T("Breakpoint ([0-9]+)[ \t]\\(\\\"(.+):([0-9]+)\\\"\\)[ \t]pending\\.")); Yuanhui Zhang Code::blocks developer ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-27 1:45 ` asmwarrior @ 2012-03-28 18:55 ` Tom Tromey 0 siblings, 0 replies; 26+ messages in thread From: Tom Tromey @ 2012-03-28 18:55 UTC (permalink / raw) To: asmwarrior; +Cc: Keith Seitz, gdb-patches@sourceware.org ml >>>>> ">" == ext asmwarrior <asmwarrior@gmail.com> writes: >> I'm testing your line spec rewrite branch, it works great. >> Now, I'm using your branch, and the result is: >> [debug]> break "E:/code/cb/test_code/DebugDLLTest/TestDLL/dllmain.cpp:29" >> [debug]No source file named >> E:/code/cb/test_code/DebugDLLTest/TestDLL/dllmain.cpp. >> [debug]Breakpoint 4 >> ("E:/code/cb/test_code/DebugDLLTest/TestDLL/dllmain.cpp:29") pending. >> You can see, the third line, it does have a tail quote. >> I believe adding the tail quote is the right direction. Also, I can >> easily adjust the regex to: I think so too; and also I think that we should not support reliance on the contents or formatting of error messages in a front end. This sort of thing is why MI exists. Tom ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-26 23:18 [RFA 1/2] Linespec rewrite (update 2) Keith Seitz 2012-03-27 1:45 ` asmwarrior @ 2012-03-27 13:56 ` Joel Brobecker 2012-03-27 14:31 ` Keith Seitz 2012-03-28 20:46 ` Tom Tromey 2012-04-03 21:19 ` Doug Evans 3 siblings, 1 reply; 26+ messages in thread From: Joel Brobecker @ 2012-03-27 13:56 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml Hi Keith, I tested your patch locally, and unfortunately, it regresses in the following situation: (gdb) break ops.adb:"+":10 FUNCTION:OFFSET is unimplemented It seems to be doing it for all operators, for instance: (gdb) break ops.adb:"/":25 FUNCTION:OFFSET is unimplemented It's a series of tests I meant to add after you check in your patch, because I believe the output changes (that's something we were discussing on Archer). Perhaps it would be better if I wrote a patch to be applied after we apply yours. That way, you can include it in your tree and immediately notice this regression... I'll probably have some time to look at this this afternoon, if you're tied up. Git also notices the following little details when applying your patch: > /home/brobecke/linespec-rewrite-parser-2.patch:810: trailing whitespace. > /* Throw an appropriate error when an unexpected token is encountered > /home/brobecke/linespec-rewrite-parser-2.patch:3180: trailing whitespace. > /home/brobecke/linespec-rewrite-parser-2.patch:3960: trailing whitespace. > { > /home/brobecke/linespec-rewrite-parser-2.patch:4106: trailing whitespace. > add the_tests "$srcfile:main:$x" invalid_label [string trim $x] "main" > /home/brobecke/linespec-rewrite-parser-2.patch:4246: trailing whitespace. > return f.baz (foo::bar () + 3); > warning: squelched 1 whitespace error > warning: 6 lines add whitespace errors. Not terribly important, but easy to fix... -- Joel ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-27 13:56 ` Joel Brobecker @ 2012-03-27 14:31 ` Keith Seitz 2012-03-27 14:54 ` Joel Brobecker 0 siblings, 1 reply; 26+ messages in thread From: Keith Seitz @ 2012-03-27 14:31 UTC (permalink / raw) To: Joel Brobecker; +Cc: gdb-patches@sourceware.org ml On 03/27/2012 06:56 AM, Joel Brobecker wrote: > I tested your patch locally, and unfortunately, it regresses in > the following situation: > > (gdb) break ops.adb:"+":10 > FUNCTION:OFFSET is unimplemented > > It seems to be doing it for all operators, for instance: > > (gdb) break ops.adb:"/":25 > FUNCTION:OFFSET is unimplemented Yes, this is a new feature AFAICT. As far as I can tell, CVS HEAD is also not properly dealing with this linespec: cvs$ ./gdb -nx -q testsuite/gdb.ada/operator_bp/ops_test -ex 'break ops.adb:"+":10' -ex 'break ops.adb:"/":25' Reading symbols from testsuite/gdb.ada/operator_bp/ops_test...done. Breakpoint 1 at 0x401176: ops.adb:"+":10. (2 locations) Breakpoint 2 at 0x4011c8: file testsuite/gdb.ada/operator_bp/ops.adb, line 40. (gdb) inf br Num Type Disp Enb Address What 1 breakpoint keep y <MULTIPLE> 1.1 y 0x0000000000401176 in ops."+" at testsuite/gdb.ada/operator_bp/ops.adb:25 1.2 y 0x000000000040136d in ops."+" at testsuite/gdb.ada/operator_bp/ops.adb:119 2 breakpoint keep y 0x00000000004011c8 in ops."/" at testsuite/gdb.ada/operator_bp/ops.adb:40 As you can see, the ":10" and ":25" were simply ignored. I've stepped through the code, and decode_variable will see "\"+\":10\n", but ada_name_for_lookup will return "+". From there on out, the ":10" is lost. Unless I'm missing something, this appears to be another special case of maintaining bug-for-bug compatibility. > I'll probably have some time to look at this this afternoon, if you're > tied up. How would you like me to fix this? > Git also notices the following little details when applying your > patch: > >> /home/brobecke/linespec-rewrite-parser-2.patch:810: trailing whitespace. >> /* Throw an appropriate error when an unexpected token is encountered >> /home/brobecke/linespec-rewrite-parser-2.patch:3180: trailing whitespace. >> /home/brobecke/linespec-rewrite-parser-2.patch:3960: trailing whitespace. >> { >> /home/brobecke/linespec-rewrite-parser-2.patch:4106: trailing whitespace. >> add the_tests "$srcfile:main:$x" invalid_label [string trim $x] "main" >> /home/brobecke/linespec-rewrite-parser-2.patch:4246: trailing whitespace. >> return f.baz (foo::bar () + 3); >> warning: squelched 1 whitespace error >> warning: 6 lines add whitespace errors. I'll fix those. Thank you for giving this a whirl (again)! Keith ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-27 14:31 ` Keith Seitz @ 2012-03-27 14:54 ` Joel Brobecker 2012-03-27 15:05 ` Joel Brobecker 0 siblings, 1 reply; 26+ messages in thread From: Joel Brobecker @ 2012-03-27 14:54 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml > As you can see, the ":10" and ":25" were simply ignored. I've > stepped through the code, and decode_variable will see "\"+\":10\n", > but ada_name_for_lookup will return "+". From there on out, the > ":10" is lost. Hah! I thought that the line numbers were because GDB tries to find the first line that matches, starting from the given one... > Unless I'm missing something, this appears to be another special > case of maintaining bug-for-bug compatibility. I think that the syntax itself should be accepted, and that it's not a bug (operators are functions). But we probably do not want to perpetuate the bug where the line number is ignored :-). > How would you like me to fix this? I was hoping that we'd be able to handle this in the parser? I am not completely familiar with the new code yet, so it's just a wild guess. I am almost done with the extension of operator_bp.exp; I'll send it when I've removed all the typos and thinkos... Thanks! -- Joel ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-27 14:54 ` Joel Brobecker @ 2012-03-27 15:05 ` Joel Brobecker 2012-03-27 18:08 ` Keith Seitz 0 siblings, 1 reply; 26+ messages in thread From: Joel Brobecker @ 2012-03-27 15:05 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml [-- Attachment #1: Type: text/plain, Size: 296 bytes --] > I am almost done with the extension of operator_bp.exp; I'll send it > when I've removed all the typos and thinkos... Here it is... As expected, it fails on HEAD, so cannot commit just yet. But PASSes with your previous iteration of the patch series, and now fails... Thanks, Keith. -- Joel [-- Attachment #2: 0001-Add-more-tests-to-gdb.ada-operator_bp.exp.patch --] [-- Type: text/x-diff, Size: 6745 bytes --] From 88358dd0af5a781493eb5e98666f28b8bd486391 Mon Sep 17 00:00:00 2001 From: Joel Brobecker <brobecker@adacore.com> Date: Tue, 27 Mar 2012 07:37:31 -0700 Subject: [PATCH] Add more tests to gdb.ada/operator_bp.exp This patch adds testing of the FILE:OP:LINENO linespec (where OP is the name of an Ada operator) as well as testing of the FILE:OP linespec. gdb/testsuite/ChangeLog: * gdb.ada/operator_bp/ops.adb: Add source line markers. * gdb.ada/operator_bp.exp: Add testing of FILE:OP:LINENO as well as FILE:OP linespecs where OP is the name of an Ada operator. --- gdb/testsuite/gdb.ada/operator_bp.exp | 57 +++++++++++++++++++++++++++++ gdb/testsuite/gdb.ada/operator_bp/ops.adb | 36 +++++++++--------- 2 files changed, 75 insertions(+), 18 deletions(-) diff --git a/gdb/testsuite/gdb.ada/operator_bp.exp b/gdb/testsuite/gdb.ada/operator_bp.exp index 18cf46d..3c20f39 100644 --- a/gdb/testsuite/gdb.ada/operator_bp.exp +++ b/gdb/testsuite/gdb.ada/operator_bp.exp @@ -88,4 +88,61 @@ foreach op { "+" "-" "*" "/" "mod" "rem" "**" "<" "<=" ">" ">=" "=" "and" "or" " "continue to ops.\"$op\"" } +# Repeat the test, but qualify the operators with a filename and +# line number. + +clean_restart ${testfile} + +runto "ops_test.adb:$bp_location" + +foreach op { "+" "-" } { + set lineno [gdb_get_line_number "STOP \"$op\"" ${testdir}/ops.adb] + set op_re [string_to_regexp $op] + gdb_test "break ops.adb:\"$op\":$lineno" \ + "Breakpoint $decimal at $hex: ops\\.adb:\"$op_re\":$lineno\. \\(2 locations\\)" +} + +foreach op { "*" "/" "mod" "rem" "**" "<" "<=" ">" ">=" "=" "and" "or" "xor" "&" "abs" "not"} { + set lineno [gdb_get_line_number "STOP \"$op\"" ${testdir}/ops.adb] + set op_re [string_to_regexp $op] + gdb_test "break ops.adb:\"$op\":$lineno" \ + "Breakpoint $decimal at $hex: file .*ops.adb, line $decimal." +} + +# Make sure we stop correctly in each operator function. + +foreach op { "+" "-" "*" "/" "mod" "rem" "**" "<" "<=" ">" ">=" "=" "and" "or" "xor" "&" "abs" "not"} { + set op_re [string_to_regexp $op] + gdb_test "continue" \ + "Breakpoint $decimal, ops\\.\"$op_re\" .*"\ + "continue to ops.adb:\"$op\":LINE" +} + +# Repeat the test, but qualify the operators with a filename alone. + +clean_restart ${testfile} + +runto "ops_test.adb:$bp_location" + +foreach op { "+" "-" } { + set op_re [string_to_regexp $op] + gdb_test "break ops.adb:\"$op\"" \ + "Breakpoint $decimal at $hex: ops\\.adb:\"$op_re\". \\(2 locations\\)" +} + +foreach op { "*" "/" "mod" "rem" "**" "<" "<=" ">" ">=" "=" "and" "or" "xor" "&" "abs" "not"} { + set op_re [string_to_regexp $op] + gdb_test "break ops.adb:\"$op\"" \ + "Breakpoint $decimal at $hex: file .*ops.adb, line $decimal." +} + +# Make sure we stop correctly in each operator function. + +foreach op { "+" "-" "*" "/" "mod" "rem" "**" "<" "<=" ">" ">=" "=" "and" "or" "xor" "&" "abs" "not"} { + set op_re [string_to_regexp $op] + gdb_test "continue" \ + "Breakpoint $decimal, ops\\.\"$op_re\" .*"\ + "continue to ops.adb:\"$op\"" +} + diff --git a/gdb/testsuite/gdb.ada/operator_bp/ops.adb b/gdb/testsuite/gdb.ada/operator_bp/ops.adb index e7d94dc..fe9168c 100644 --- a/gdb/testsuite/gdb.ada/operator_bp/ops.adb +++ b/gdb/testsuite/gdb.ada/operator_bp/ops.adb @@ -22,38 +22,38 @@ package body Ops is function "+" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) + IntRep (I2)); + return Int (IntRep (I1) + IntRep (I2)); -- STOP "+" end; function "-" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) - IntRep (I2)); + return Int (IntRep (I1) - IntRep (I2)); -- STOP "-" end; function "*" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) * IntRep (I2)); + return Int (IntRep (I1) * IntRep (I2)); -- STOP "*" end; function "/" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) / IntRep (I2)); + return Int (IntRep (I1) / IntRep (I2)); -- STOP "/" end; function "mod" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) mod IntRep (I2)); + return Int (IntRep (I1) mod IntRep (I2)); -- STOP "mod" end; function "rem" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) rem IntRep (I2)); + return Int (IntRep (I1) rem IntRep (I2)); -- STOP "rem" end; function "**" (I1, I2 : Int) return Int is Result : IntRep := 1; begin - for J in 1 .. IntRep (I2) loop + for J in 1 .. IntRep (I2) loop -- STOP "**" Result := IntRep (I1) * Result; end loop; return Int (Result); @@ -61,57 +61,57 @@ package body Ops is function "<" (I1, I2 : Int) return Boolean is begin - return IntRep (I1) < IntRep (I2); + return IntRep (I1) < IntRep (I2); -- STOP "<" end; function "<=" (I1, I2 : Int) return Boolean is begin - return IntRep (I1) <= IntRep (I2); + return IntRep (I1) <= IntRep (I2); -- STOP "<=" end; function ">" (I1, I2 : Int) return Boolean is begin - return IntRep (I1) > IntRep (I2); + return IntRep (I1) > IntRep (I2); -- STOP ">" end; function ">=" (I1, I2 : Int) return Boolean is begin - return IntRep (I1) >= IntRep (I2); + return IntRep (I1) >= IntRep (I2); -- STOP ">=" end; function "=" (I1, I2 : Int) return Boolean is begin - return IntRep (I1) = IntRep (I2); + return IntRep (I1) = IntRep (I2); -- STOP "=" end; function "and" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) and IntRep (I2)); + return Int (IntRep (I1) and IntRep (I2)); -- STOP "and" end; function "or" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) or IntRep (I2)); + return Int (IntRep (I1) or IntRep (I2)); -- STOP "or" end; function "xor" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) xor IntRep (I2)); + return Int (IntRep (I1) xor IntRep (I2)); -- STOP "xor" end; function "&" (I1, I2 : Int) return Int is begin - return Int (IntRep (I1) and IntRep (I2)); + return Int (IntRep (I1) and IntRep (I2)); -- STOP "&" end; function "abs" (I1 : Int) return Int is begin - return Int (abs IntRep (I1)); + return Int (abs IntRep (I1)); -- STOP "abs" end; function "not" (I1 : Int) return Int is begin - return Int (not IntRep (I1)); + return Int (not IntRep (I1)); -- STOP "not" end; function "+" (I1 : Int) return Int is -- 1.7.1 ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-27 15:05 ` Joel Brobecker @ 2012-03-27 18:08 ` Keith Seitz 0 siblings, 0 replies; 26+ messages in thread From: Keith Seitz @ 2012-03-27 18:08 UTC (permalink / raw) To: Joel Brobecker; +Cc: gdb-patches@sourceware.org ml On 03/27/2012 08:05 AM, Joel Brobecker wrote: >> I am almost done with the extension of operator_bp.exp; I'll send it >> when I've removed all the typos and thinkos... > > Here it is... As expected, it fails on HEAD, so cannot commit just > yet. But PASSes with your previous iteration of the patch series, > and now fails... If it is failing on the "unimplemented" message, we can certainly whack that from the patch -- it doesn't affect test results at all. In fact, I've pushed this. The decision to issue an unimplemented error should really come from create_linespec_from_sals, since that is the function that actually does the work, and if mi/13139 ever gets implemented, this will affect that as well. I am worried that it appears to give a false positive on your tests with no indication that anything has gone awry. [I worry because I intended to actually implement symbol-relative offsets sometime in the future.] Keith ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-26 23:18 [RFA 1/2] Linespec rewrite (update 2) Keith Seitz 2012-03-27 1:45 ` asmwarrior 2012-03-27 13:56 ` Joel Brobecker @ 2012-03-28 20:46 ` Tom Tromey 2012-03-29 19:18 ` Keith Seitz 2012-04-03 21:19 ` Doug Evans 3 siblings, 1 reply; 26+ messages in thread From: Tom Tromey @ 2012-03-28 20:46 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml >>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes: Keith> Tom requested a repost of this patch, so here it is. Thanks, Keith. This patch is pretty hard to review. I've been reading the branch instead, but even there I find it hard to wrap my head around it all. That said, I like what I see. Congratulations on conquering linespec. I'm glad to see others testing this patch on their various scenarios, I think that raises the confidence level in the patch. Keith> +const char * const linespec_keywords[] = { "if", "thread", "task" }; Should also be static. Keith> + /* If we're in list mode, and the next token is a string beginning Keith> + with ",", we're dealing with a ranged listing. Stop parsing Keith> + and return. */ Keith> + if (PARSER_STATE (parser)->list_mode Keith> + && token.type == LSTOKEN_STRING Keith> + && *LS_TOKEN_STOKEN (token).ptr == ',') Keith> + return; I thought that historically a top-level comma always terminated a linespec -- not just in list mode. It should be possible to write a test case for this in Python pretty easily. Keith> +static void Keith> +canonicalize_linespec (struct linespec_state *state, linespec_t ls) [...] Keith> + if (ls->line_offset.sign != unknown) Keith> + { Keith> + if (need_colon) Keith> + fputc_unfiltered (':', buf); Keith> + fprintf_filtered (buf, "%s%d", Keith> + (ls->line_offset.sign == none ? "" Keith> + : ls->line_offset.sign == plus ? "+" : "-"), Keith> + ls->line_offset.offset); I am curious when this code can trigger. Can we end up with a canonical form like "function:+5"? I was hoping to reserve that syntax for a later addition; and anyway in general I think relative linespecs need to be made absolute by the canonicalization process, since otherwise re-setting won't do the right thing. Keith> + /* We have an expression. No other attribute is allowed. */ It would be helpful if the constraints on the fields of 'struct linespec' were documented there. Keith> + pspace = elem->minsym->ginfo.obj_section->objfile->pspace; Should use SYMBOL_OBJ_SECTION. I didn't audit for other instances. Keith> + else if (ls->minimal_symbols != NULL) Keith> + { Keith> + /* We found minimal symbols, but no normal symbols. */ Keith> + int i; Keith> + minsym_and_objfile_d *elem; Keith> + Keith> + for (i = 0; Keith> + VEC_iterate (minsym_and_objfile_d, ls->minimal_symbols, i, elem); Keith> + ++i) Keith> + minsym_found (state, elem->objfile, elem->minsym, &sals); Why are minsyms sorted by pspace in one branch but not another? Keith> +++ b/gdb/testsuite/gdb.cp/cplabel.exp [...] Keith> +if {[prepare_for_testing "$testfile.exp" $testfile $srcfile]} { Keith> + return -1 I suspect this needs a skip_cplus_tests check. Tom ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-28 20:46 ` Tom Tromey @ 2012-03-29 19:18 ` Keith Seitz 2012-03-30 15:34 ` Tom Tromey 0 siblings, 1 reply; 26+ messages in thread From: Keith Seitz @ 2012-03-29 19:18 UTC (permalink / raw) To: Tom Tromey; +Cc: gdb-patches@sourceware.org ml On 03/28/2012 01:46 PM, Tom Tromey wrote: > This patch is pretty hard to review. I've been reading the branch > instead, but even there I find it hard to wrap my head around it all. Yeah, it is pretty massive. I apologize. I simply don't know any other way to submit this. [I'm all eyes/ears if you would like to see it in any other form/format.] > Keith> +const char * const linespec_keywords[] = { "if", "thread", "task" }; > > Should also be static. Yup. Fixed. > Keith> + /* If we're in list mode, and the next token is a string beginning > Keith> + with ",", we're dealing with a ranged listing. Stop parsing > Keith> + and return. */ > Keith> + if (PARSER_STATE (parser)->list_mode > Keith> + && token.type == LSTOKEN_STRING > Keith> + && *LS_TOKEN_STOKEN (token).ptr == ',') > Keith> + return; > > I thought that historically a top-level comma always terminated a > linespec -- not just in list mode. It should be possible to write a > test case for this in Python pretty easily. I don't know. The whole comma thing is undocumented. The test suite does contain list ranges. That's how I originally discovered this. I've removed the list mode restriction, though, and it doesn't affect test results at all. One cannot "break 30,31,32" in either branch, so I can see no differences in functionality. > Keith> +static void > Keith> +canonicalize_linespec (struct linespec_state *state, linespec_t ls) > [...] > Keith> + if (ls->line_offset.sign != unknown) > Keith> + { > Keith> + if (need_colon) > Keith> + fputc_unfiltered (':', buf); > Keith> + fprintf_filtered (buf, "%s%d", > Keith> + (ls->line_offset.sign == none ? "" > Keith> + : ls->line_offset.sign == plus ? "+" : "-"), > Keith> + ls->line_offset.offset); > > I am curious when this code can trigger. > Can we end up with a canonical form like "function:+5"? > I was hoping to reserve that syntax for a later addition; and anyway in > general I think relative linespecs need to be made absolute by the > canonicalization process, since otherwise re-setting won't do the right > thing. Yes, we can end up with a canonical form like "function:+5" or "file:+5". The former is permitted (per recent maintainer request) because we currently ignore the offset. [It is unprocessed in convert_linespec_to_sals.] I'm not a fan of this, FYI CVS HEAD creates canonical linespecs like "file:+5" today, without converting to absolute offset. It also rejects "function:+5". I haven't done anything with this. AFAICT, the linespecs are identical in both CVS HEAD and the branch. [Or would be if either CVS HEAD ignored function-relative offsets or we issued an error again in my branch.] What would you like me to do? > Keith> + /* We have an expression. No other attribute is allowed. */ > > It would be helpful if the constraints on the fields of 'struct > linespec' were documented there. I'll add something. The reason this is here because this is no inherent limitation on expressions (or other members of struct linespec). We could permit file:*foo, if we wanted to, but we simply don't. > Keith> + pspace = elem->minsym->ginfo.obj_section->objfile->pspace; > > Should use SYMBOL_OBJ_SECTION. I didn't audit for other instances. Fixed. It was the only offender. > Keith> + else if (ls->minimal_symbols != NULL) > Keith> + { > Keith> + /* We found minimal symbols, but no normal symbols. */ > Keith> + int i; > Keith> + minsym_and_objfile_d *elem; > Keith> + > Keith> + for (i = 0; > Keith> + VEC_iterate (minsym_and_objfile_d, ls->minimal_symbols, i, elem); > Keith> + ++i) > Keith> + minsym_found (state, elem->objfile, elem->minsym,&sals); > > Why are minsyms sorted by pspace in one branch but not another? No real good reason, other than that is the way it is done today. I tried to keep the codepaths as similar as possible. I've merged the two branches together. No need for minsyms to be singled out like this. > Keith> +++ b/gdb/testsuite/gdb.cp/cplabel.exp > [...] > Keith> +if {[prepare_for_testing "$testfile.exp" $testfile $srcfile]} { > Keith> + return -1 > > I suspect this needs a skip_cplus_tests check. Indeed. It was also missing the c++ and debug flags for prepare_for_testing, which I've also added. I've pushed the requested changes to my archer branch. Thank you for the feedback. Keith ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-29 19:18 ` Keith Seitz @ 2012-03-30 15:34 ` Tom Tromey 2012-03-30 15:59 ` Keith Seitz 0 siblings, 1 reply; 26+ messages in thread From: Tom Tromey @ 2012-03-30 15:34 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml >>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes: Keith> I don't know. The whole comma thing is undocumented. The test suite Keith> does contain list ranges. That's how I originally discovered Keith> this. I've removed the list mode restriction, though, and it doesn't Keith> affect test results at all. You can write a test case using python that calls gdb.decode_line and examines the remainder of the line. Comma-termination isn't documented but I think it has to be preserved anyway. Keith> Yes, we can end up with a canonical form like "function:+5" or Keith> "file:+5". The former is permitted (per recent maintainer request) Keith> because we currently ignore the offset. [It is unprocessed in Keith> convert_linespec_to_sals.] I'm not a fan of this, What is the rationale for having a linespec where parts are ignored? I couldn't think of a use for it. And, if current cvs rejects it, then it seems like it is interfering with a useful future feature as well. Tom> Why are minsyms sorted by pspace in one branch but not another? Keith> No real good reason, other than that is the way it is done today. I Keith> tried to keep the codepaths as similar as possible. I've merged the Keith> two branches together. No need for minsyms to be singled out like Keith> this. Thanks. Keeping things the same is sufficient rationale, but now that you've merged it, that is fine too. Tom ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-30 15:34 ` Tom Tromey @ 2012-03-30 15:59 ` Keith Seitz 2012-03-30 16:37 ` Tom Tromey 0 siblings, 1 reply; 26+ messages in thread From: Keith Seitz @ 2012-03-30 15:59 UTC (permalink / raw) To: Tom Tromey; +Cc: gdb-patches@sourceware.org ml On 03/30/2012 08:33 AM, Tom Tromey wrote: >>>>>> "Keith" == Keith Seitz<keiths@redhat.com> writes: > > Keith> I don't know. The whole comma thing is undocumented. The test suite > Keith> does contain list ranges. That's how I originally discovered > Keith> this. I've removed the list mode restriction, though, and it doesn't > Keith> affect test results at all. > > You can write a test case using python that calls gdb.decode_line and > examines the remainder of the line. I'll see about writing a test that does this. > Comma-termination isn't documented but I think it has to be preserved > anyway. Already done. :-) > Keith> Yes, we can end up with a canonical form like "function:+5" or > Keith> "file:+5". The former is permitted (per recent maintainer request) > Keith> because we currently ignore the offset. [It is unprocessed in > Keith> convert_linespec_to_sals.] I'm not a fan of this, > > What is the rationale for having a linespec where parts are ignored? > I couldn't think of a use for it. And, if current cvs rejects it, then > it seems like it is interfering with a useful future feature as well. I agree, but perhaps I simply misunderstood Joel? http://sourceware.org/ml/gdb-patches/2012-03/msg00904.html Keith ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-30 15:59 ` Keith Seitz @ 2012-03-30 16:37 ` Tom Tromey 2012-03-30 17:09 ` Joel Brobecker 0 siblings, 1 reply; 26+ messages in thread From: Tom Tromey @ 2012-03-30 16:37 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml >>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes: Tom> What is the rationale for having a linespec where parts are ignored? Tom> I couldn't think of a use for it. And, if current cvs rejects it, then Tom> it seems like it is interfering with a useful future feature as well. Keith> I agree, but perhaps I simply misunderstood Joel? Keith> http://sourceware.org/ml/gdb-patches/2012-03/msg00904.html I read the followup more as a request for the feature to be added: http://sourceware.org/ml/gdb-patches/2012-03/msg00907.html However, I think that it is better to have this patch be as behavior-unchanging as reasonably possible and leave feature additions for follow-up patches. Tom ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-30 16:37 ` Tom Tromey @ 2012-03-30 17:09 ` Joel Brobecker 2012-03-30 17:56 ` Tom Tromey 0 siblings, 1 reply; 26+ messages in thread From: Joel Brobecker @ 2012-03-30 17:09 UTC (permalink / raw) To: Tom Tromey; +Cc: Keith Seitz, gdb-patches@sourceware.org ml > I read the followup more as a request for the feature to be added: > > http://sourceware.org/ml/gdb-patches/2012-03/msg00907.html > > However, I think that it is better to have this patch be as > behavior-unchanging as reasonably possible and leave feature additions > for follow-up patches. My take on this is that the linespec was accepted before, and now it is going to generate an error. What Keith highlighted was the fact that the line number portion is currently ignored, which is definitely a bug. I don't mind if we preserve the bug of ignoring the line number, but I am a little less comfortable with error-ing out. I don't mind if we create a temporary regression in that linespec if that makes things easier. But it would be nice to be able to not let it linger too long. I am hoping to have some time to look at this maybe mid-April. -- Joel ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-30 17:09 ` Joel Brobecker @ 2012-03-30 17:56 ` Tom Tromey 2012-03-30 18:05 ` Keith Seitz 0 siblings, 1 reply; 26+ messages in thread From: Tom Tromey @ 2012-03-30 17:56 UTC (permalink / raw) To: Joel Brobecker; +Cc: Keith Seitz, gdb-patches@sourceware.org ml Joel> My take on this is that the linespec was accepted before, and now Joel> it is going to generate an error. What Keith highlighted was the fact Joel> that the line number portion is currently ignored, which is definitely Joel> a bug. I don't mind if we preserve the bug of ignoring the line number, Joel> but I am a little less comfortable with error-ing out. Ok, I understand now. Keith, are there other issues outstanding? Maybe it is time for the final patch. Tom ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-30 17:56 ` Tom Tromey @ 2012-03-30 18:05 ` Keith Seitz 2012-04-03 23:22 ` Keith Seitz 0 siblings, 1 reply; 26+ messages in thread From: Keith Seitz @ 2012-03-30 18:05 UTC (permalink / raw) To: Tom Tromey; +Cc: gdb-patches@sourceware.org ml On 03/30/2012 10:55 AM, Tom Tromey wrote: > Joel> My take on this is that the linespec was accepted before, and now > Joel> it is going to generate an error. What Keith highlighted was the fact > Joel> that the line number portion is currently ignored, which is definitely > Joel> a bug. I don't mind if we preserve the bug of ignoring the line number, > Joel> but I am a little less comfortable with error-ing out. > > Ok, I understand now. > > Keith, are there other issues outstanding? The only thing I can think of is the comma issue. I am about to submit a patch to test for commas in linespecs. This isn't working properly on the archer branch. Keith ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-30 18:05 ` Keith Seitz @ 2012-04-03 23:22 ` Keith Seitz 2012-04-05 15:22 ` Tom Tromey 2012-07-22 19:33 ` Andreas Schwab 0 siblings, 2 replies; 26+ messages in thread From: Keith Seitz @ 2012-04-03 23:22 UTC (permalink / raw) To: Tom Tromey; +Cc: gdb-patches@sourceware.org ml [-- Attachment #1: Type: text/plain, Size: 4819 bytes --] On 03/30/2012 11:05 AM, Keith Seitz wrote: > On 03/30/2012 10:55 AM, Tom Tromey wrote: >> Keith, are there other issues outstanding? > > The only thing I can think of is the comma issue. I am about to submit a > patch to test for commas in linespecs. This isn't working properly on > the archer branch. I fixed this on Friday and pushed them to my archer branch. [I think I mentioned this on IRC.] Given Doug's comments earlier today, I'm reposting anything that hasn't already been approved or which changed. The easiest way to see what's happened is, of course, by inspecting the commit history of the branch. I'll follow-up separately with the other patch that was updated. Keith ChangeLog 2012-04-03 Keith Seitz <keiths@redhat.com> * linespec.c (decode_compound): Remove. (enum offset_relative_sign): New enum. (struct line_offset): New struct. (struct linespec): New struct. (struct linespec_state): Move file_symtabs, user_filename, and user_function into struct linespec. Make result an anonymous struct holding vectors of symbolp and minsym_and_objfile_d. Add language member. (enum ls_token_type): New enum. (linespec_keywords): New array. (struct ls_token): New struct. (struct ls_parser): New struct. (linespec_lexer_lex_number): New function. (linespec_lexer_lex_keyword): New function. (is_ada_operator): New function. (skip_quote_char): New function. (copy_token_string): New function. (is_closing_quote_enclosed): New function. (find_parameter_list_end): New function. (linespec_lexer_lex_string): New function. (linespec_lexer_lex_one): New function. (linespec_lexer_consume_token): New function. (linespec_lexer_peek_token): New function. (cplusplus_error): Remove unused function. (find_methods): Update comment. (find_toplevel_char): Return const. (is_objc_method_format): Remove unused function. (find_toplevel_string): New function. (is_linespec_boundary): Remove. (symbol_not_found_error): New function. (find_method_overload_end): Remove function. (unexpected_linespec_error): New function. (keep_name_info): Remove. (linespec_parse_line_offset): New function. (linespec_parse_basic): New function. (canonicalize_linespec): New function. (decode_line_internal): Remove. (create_sals_line_offset): New function adapted from decode_all_digits. (convert_linespec_to_sals): New function. (parse_linespec): New function. (linespec_parser_new): New function. (linespec_state_destructor): Change parameter type to struct linespec_state *. Add language parameter. Remove freeing of moved members. (linespec_parser_delete): New function. (decode_line_full): Use parse_linespec and linespec_parser_new. (decode_line_1): Likewise. (decode_indirect): Rename to ... (linespec_expression_to_pc): ... this and rewrite to simply find CORE_ADDR, storing this result for later conversion to SALs. (locate_first_half): Remove. (deocde_objc): Add parameter LS. Initialize new struct collect_info members. Handle minimal symbols, too. (decode_compound): Delete. (lookup_prefix_sym): Rewrite. (compare_msymbols): New function. (find_method): Rewrite. Do not call cplusplus_error. (symtabs_from_filename): Rewrite. (collect_function_symbols): Delete. (find_function_symbols): Rewrite without ARGPTR-style processing. (decode_all_digits): Delete. (Rewritten as create_sals_line_offset.) (decode_dollar): Adapted and renamed to ... (linespec_parse_variable): ... this. (find_linespec_symbols): New function. (decode_label): Adapted and renamed to ... (find_label_symbols): ... this. (decode_digits_list_mode): Add and use LS argument. (decode_digits_ordinary): Likewise. (collect_symbols): Do not collect SALs, just symbols and msymbols. If in list mode, allow any symbol class. Otherwise, only permit LOC_BLOCK symbols. (minsym_found): Update comments. (search_minsyms_for_name): Do not convert the matching symbol into a SAL. Simply push the symbol and objfile into the result vector. (decode_variable): Delete. Contents adapted into find_linespec_symbols. testsuite/ChangeLog 2012-04-03 Keith Seitz <keiths@redhat.com> * gdb.base/advance.exp: Update error message for "advance malformed" test. * gdb.base/break.exp: Likewise for "breakpoint with trailing garbage" test. * gdb.base/hbreak2.exp: Likewise for "hardware breakpoint with trailing garbage" test. * gdb.base/sepdebug.exp: Likewise for "breakpoint with trailng garbage" test. * gdb.base/until.exp: Likewise for "malformed until" test. * gdb.cp/ovldbreak.exp: Create the breakpoint table for "breakpoint info (after setting on all)". * gdb.cp/userdef.exp: Remove quoting for "break A2::operator+" tests. * gdb.cp/cplabel.cc: New file. * gdb.cp/cplabel.exp: New test. * gdb.linespec/ls-errs.c: New file. * gdb.linespec/ls-errs.exp: New test. [-- Attachment #2: linespec-rewrite-parser-3.patch --] [-- Type: text/x-patch, Size: 151510 bytes --] diff --git a/gdb/linespec.c b/gdb/linespec.c index 1e9770e..228214b 100644 --- a/gdb/linespec.c +++ b/gdb/linespec.c @@ -63,12 +63,108 @@ struct address_entry CORE_ADDR addr; }; +/* A helper struct which just holds a minimal symbol and the object + file from which it came. */ + +typedef struct minsym_and_objfile +{ + struct minimal_symbol *minsym; + struct objfile *objfile; +} minsym_and_objfile_d; + +DEF_VEC_O (minsym_and_objfile_d); + +/* An enumeration of possible signs for a line offset. */ +enum offset_relative_sign +{ + /* No sign */ + LINE_OFFSET_NONE, + + /* A plus sign ("+") */ + LINE_OFFSET_PLUS, + + /* A minus sign ("-") */ + LINE_OFFSET_MINUS, + + /* A special "sign" for unspecified offset. */ + LINE_OFFSET_UNKNOWN +}; + +/* A line offset in a linespec. */ + +struct line_offset +{ + /* Line offset and any specified sign. */ + int offset; + enum offset_relative_sign sign; +}; + +/* A linespec. Elements of this structure are filled in by a parser + (either parse_linespec or some other function). The structure is + then converted into SALs by convert_linespec_to_sals. */ + +struct linespec +{ + /* An expression and the resulting PC. Specifying an expression + currently precludes the use of other members. */ + + /* The expression entered by the user. */ + char *expression; + + /* The resulting PC expression derived from evaluating EXPRESSION. */ + CORE_ADDR expr_pc; + + /* Any specified file symtabs. */ + + /* The user-supplied source filename or NULL if none was specified. */ + char *source_filename; + + /* The list of symtabs to search to which to limit the search. May not + be NULL. If SOURCE_FILENAME is NULL (no user-specified filename), + FILE_SYMTABS should contain one single NULL member. This will + cause the code to use the default symtab. */ + VEC (symtab_p) *file_symtabs; + + /* The name of a function or method and any matching symbols. */ + + /* The user-specified function name. If no function name was + supplied, this may be NULL. */ + char *function_name; + + /* A list of matching function symbols and minimal symbols. Both lists + may be NULL if no matching symbols were found. */ + VEC (symbolp) *function_symbols; + VEC (minsym_and_objfile_d) *minimal_symbols; + + /* The name of a label and matching symbols. */ + + /* The user-specified label name. */ + char *label_name; + + /* A structure of matching label symbols and the corresponding + function symbol in which the label was found. Both may be NULL + or both must be non-NULL. */ + struct + { + VEC (symbolp) *label_symbols; + VEC (symbolp) *function_symbols; + } labels; + + /* Line offset. It may be LINE_OFFSET_UNKNOWN, meaning that no + offset was specified. */ + struct line_offset line_offset; +}; +typedef struct linespec *linespec_p; + /* An instance of this is used to keep all state while linespec operates. This instance is passed around as a 'this' pointer to the various implementation methods. */ struct linespec_state { + /* The language in use during linespec processing. */ + const struct language_defn *language; + /* The program space as seen when the module was entered. */ struct program_space *program_space; @@ -78,19 +174,6 @@ struct linespec_state /* The default line to use. */ int default_line; - /* If the linespec started with "FILE:", this holds all the matching - symtabs. Otherwise, it will hold a single NULL entry, meaning - that the default symtab should be used. */ - VEC (symtab_p) *file_symtabs; - - /* If the linespec started with "FILE:", this holds an xmalloc'd - copy of "FILE". */ - char *user_filename; - - /* If the linespec is "FUNCTION:LABEL", this holds an xmalloc'd copy - of "FUNCTION". */ - char *user_function; - /* The 'funfirstline' value that was passed in to decode_line_1 or decode_line_full. */ int funfirstline; @@ -117,67 +200,128 @@ struct collect_info /* The linespec object in use. */ struct linespec_state *state; + /* A list of symtabs to which to restrict matches. */ + VEC (symtab_p) *file_symtabs; + /* The result being accumulated. */ - struct symtabs_and_lines result; + struct + { + VEC (symbolp) *symbols; + VEC (minsym_and_objfile_d) *minimal_symbols; + } result; }; -/* Prototypes for local functions. */ +/* Token types */ -static void initialize_defaults (struct symtab **default_symtab, - int *default_line); +enum ls_token_type +{ + /* A keyword */ + LSTOKEN_KEYWORD = 0, -static struct symtabs_and_lines decode_indirect (struct linespec_state *self, - char **argptr); + /* A colon "separator" */ + LSTOKEN_COLON, -static char *locate_first_half (char **argptr, int *is_quote_enclosed); + /* A string */ + LSTOKEN_STRING, -static struct symtabs_and_lines decode_objc (struct linespec_state *self, - char **argptr); + /* A number */ + LSTOKEN_NUMBER, + + /* A comma */ + LSTOKEN_COMMA, + + /* EOI (end of input) */ + LSTOKEN_EOI, + + /* Consumed token */ + LSTOKEN_CONSUMED +}; +typedef enum ls_token_type linespec_token_type; + +/* List of keywords */ + +static const char * const linespec_keywords[] = { "if", "thread", "task" }; + +/* A token of the linespec lexer */ + +struct ls_token +{ + /* The type of the token */ + linespec_token_type type; + + /* Data for the token */ + union + { + /* A string, given as a stoken */ + struct stoken string; + + /* A keyword */ + const char *keyword; + } data; +}; +typedef struct ls_token linespec_token; + +#define LS_TOKEN_STOKEN(TOK) (TOK).data.string +#define LS_TOKEN_KEYWORD(TOK) (TOK).data.keyword + +/* An instance of the linespec parser. */ + +struct ls_parser +{ + /* Lexer internal data */ + struct + { + /* Save head of input stream. */ + char *saved_arg; -static struct symtabs_and_lines decode_compound (struct linespec_state *self, - char **argptr, - char *saved_arg, - char *p); + /* Head of the input stream. */ + char **stream; +#define PARSER_STREAM(P) (*(P)->lexer.stream) -static VEC (symbolp) *lookup_prefix_sym (char **argptr, char *p, - VEC (symtab_p) *, - char **); + /* The current token. */ + linespec_token current; + } lexer; -static struct symtabs_and_lines find_method (struct linespec_state *self, - char *saved_arg, - char *copy, - const char *class_name, - VEC (symbolp) *sym_classes); + /* Is the entire linespec quote-enclosed? */ + int is_quote_enclosed; + + /* The state of the parse. */ + struct linespec_state state; +#define PARSER_STATE(PPTR) (&(PPTR)->state) -static void cplusplus_error (const char *name, const char *fmt, ...) - ATTRIBUTE_NORETURN ATTRIBUTE_PRINTF (2, 3); + /* The result of the parse. */ + struct linespec result; +#define PARSER_RESULT(PPTR) (&(PPTR)->result) +}; +typedef struct ls_parser linespec_parser; -static char *find_toplevel_char (char *s, char c); +/* Prototypes for local functions. */ -static int is_objc_method_format (const char *s); +static void initialize_defaults (struct symtab **default_symtab, + int *default_line); -static VEC (symtab_p) *symtabs_from_filename (char **argptr, - char *p, int is_quote_enclosed, - char **user_filename); +static CORE_ADDR linespec_expression_to_pc (char **exp_ptr); -static VEC (symbolp) *find_function_symbols (char **argptr, char *p, - int is_quote_enclosed, - char **user_function); +static struct symtabs_and_lines decode_objc (struct linespec_state *self, + linespec_p ls, + char **argptr); -static struct symtabs_and_lines decode_all_digits (struct linespec_state *self, - char **argptr, - char *q); +static VEC (symtab_p) *symtabs_from_filename (const char *); -static struct symtabs_and_lines decode_dollar (struct linespec_state *self, - char *copy); +static VEC (symbolp) *find_label_symbols (struct linespec_state *self, + VEC (symbolp) *function_symbols, + VEC (symbolp) **label_funcs_ret, + const char *name); -static int decode_label (struct linespec_state *self, - VEC (symbolp) *function_symbols, - char *copy, - struct symtabs_and_lines *result); +void find_linespec_symbols (struct linespec_state *self, + VEC (symtab_p) *file_symtabs, + const char *name, + VEC (symbolp) **symbols, + VEC (minsym_and_objfile_d) **minsyms); -static struct symtabs_and_lines decode_variable (struct linespec_state *self, - char *copy); +static struct line_offset + linespec_parse_variable (struct linespec_state *self, + const char *variable); static int symbol_to_sal (struct symtab_and_line *result, int funfirstline, struct symbol *sym); @@ -190,217 +334,645 @@ static void add_all_symbol_names_from_pspace (struct collect_info *info, struct program_space *pspace, VEC (const_char_ptr) *names); -/* Helper functions. */ +static VEC (symtab_p) *collect_symtabs_from_filename (const char *file); -/* Add SAL to SALS. */ +static void decode_digits_ordinary (struct linespec_state *self, + linespec_p ls, + int line, + struct symtabs_and_lines *sals, + struct linetable_entry **best_entry); -static void -add_sal_to_sals_basic (struct symtabs_and_lines *sals, - struct symtab_and_line *sal) -{ - ++sals->nelts; - sals->sals = xrealloc (sals->sals, sals->nelts * sizeof (sals->sals[0])); - sals->sals[sals->nelts - 1] = *sal; -} +static void decode_digits_list_mode (struct linespec_state *self, + linespec_p ls, + struct symtabs_and_lines *values, + struct symtab_and_line val); -/* Add SAL to SALS, and also update SELF->CANONICAL_NAMES to reflect - the new sal, if needed. If not NULL, SYMNAME is the name of the - symbol to use when constructing the new canonical name. */ +static void minsym_found (struct linespec_state *self, struct objfile *objfile, + struct minimal_symbol *msymbol, + struct symtabs_and_lines *result); -static void -add_sal_to_sals (struct linespec_state *self, - struct symtabs_and_lines *sals, - struct symtab_and_line *sal, - const char *symname) -{ - add_sal_to_sals_basic (sals, sal); +static int compare_symbols (const void *a, const void *b); - if (self->canonical) - { - char *canonical_name = NULL; +static int compare_msymbols (const void *a, const void *b); - self->canonical_names = xrealloc (self->canonical_names, - sals->nelts * sizeof (char *)); - if (sal->symtab && sal->symtab->filename) - { - char *filename = sal->symtab->filename; +static const char *find_toplevel_char (const char *s, char c); - /* Note that the filter doesn't have to be a valid linespec - input. We only apply the ":LINE" treatment to Ada for - the time being. */ - if (symname != NULL && sal->line != 0 - && current_language->la_language == language_ada) - canonical_name = xstrprintf ("%s:%s:%d", filename, symname, - sal->line); - else if (symname != NULL) - canonical_name = xstrprintf ("%s:%s", filename, symname); - else - canonical_name = xstrprintf ("%s:%d", filename, sal->line); - } +/* Permitted quote characters for the parser. This is different from the + completer's quote characters to allow backward compatibility with the + previous parser. */ +static const char *const linespec_quote_characters = "\"\'"; - self->canonical_names[sals->nelts - 1] = canonical_name; +/* Lexer functions. */ + +/* Lex a number from the input in PARSER. This only supports + decimal numbers. */ + +static linespec_token +linespec_lexer_lex_number (linespec_parser *parser) +{ + linespec_token token; + + token.type = LSTOKEN_NUMBER; + LS_TOKEN_STOKEN (token).length = 0; + LS_TOKEN_STOKEN (token).ptr = PARSER_STREAM (parser); + + /* Keep any sign at the start of the stream. */ + if (*PARSER_STREAM (parser) == '+' || *PARSER_STREAM (parser) == '-') + { + ++LS_TOKEN_STOKEN (token).length; + ++(PARSER_STREAM (parser)); + } + + while (isdigit (*PARSER_STREAM (parser))) + { + ++LS_TOKEN_STOKEN (token).length; + ++(PARSER_STREAM (parser)); } + + return token; } -/* A hash function for address_entry. */ +/* Does P represent one of the keywords? If so, return + the keyword. If not, return NULL. */ -static hashval_t -hash_address_entry (const void *p) +static const char * +linespec_lexer_lex_keyword (const char *p) { - const struct address_entry *aep = p; - hashval_t hash; + int i; - hash = iterative_hash_object (aep->pspace, 0); - return iterative_hash_object (aep->addr, hash); + if (p != NULL) + { + for (i = 0; i < ARRAY_SIZE (linespec_keywords); ++i) + { + int len = strlen (linespec_keywords[i]); + + /* If P begins with one of the keywords and the next + character is not a valid identifier character, + we have found a keyword. */ + if (strncmp (p, linespec_keywords[i], len) == 0 + && !(isalnum (p[len]) || p[len] == '_')) + return linespec_keywords[i]; + } + } + + return NULL; } -/* An equality function for address_entry. */ +/* Does STRING represent an Ada operator? If so, return the length + of the decoded operator name. If not, return 0. */ static int -eq_address_entry (const void *a, const void *b) +is_ada_operator (const char *string) { - const struct address_entry *aea = a; - const struct address_entry *aeb = b; + const struct ada_opname_map *mapping; - return aea->pspace == aeb->pspace && aea->addr == aeb->addr; + for (mapping = ada_opname_table; + mapping->encoded != NULL + && strncmp (mapping->decoded, string, + strlen (mapping->decoded)) != 0; ++mapping) + ; + + return mapping->decoded == NULL ? 0 : strlen (mapping->decoded); } -/* Check whether the address, represented by PSPACE and ADDR, is - already in the set. If so, return 0. Otherwise, add it and return - 1. */ +/* Find QUOTE_CHAR in STRING, accounting for the ':' terminal. Return + the location of QUOTE_CHAR, or NULL if not found. */ -static int -maybe_add_address (htab_t set, struct program_space *pspace, CORE_ADDR addr) +static const char * +skip_quote_char (const char *string, char quote_char) { - struct address_entry e, *p; - void **slot; - - e.pspace = pspace; - e.addr = addr; - slot = htab_find_slot (set, &e, INSERT); - if (*slot) - return 0; + const char *p, *last; - p = XNEW (struct address_entry); - memcpy (p, &e, sizeof (struct address_entry)); - *slot = p; + p = last = find_toplevel_char (string, quote_char); + while (p && *p != '\0' && *p != ':') + { + p = find_toplevel_char (p, quote_char); + if (p != NULL) + last = p++; + } - return 1; + return last; } -/* Issue a helpful hint on using the command completion feature on - single quoted demangled C++ symbols as part of the completion - error. */ +/* Make a writable copy of the string given in TOKEN, trimming + any trailing whitespace. */ -static void -cplusplus_error (const char *name, const char *fmt, ...) +static char * +copy_token_string (linespec_token token) { - struct ui_file *tmp_stream; - char *message; + char *str, *s; - tmp_stream = mem_fileopen (); - make_cleanup_ui_file_delete (tmp_stream); + if (token.type == LSTOKEN_KEYWORD) + return xstrdup (LS_TOKEN_KEYWORD (token)); - { - va_list args; + str = savestring (LS_TOKEN_STOKEN (token).ptr, + LS_TOKEN_STOKEN (token).length); + s = remove_trailing_whitespace (str, str + LS_TOKEN_STOKEN (token).length); + *s = '\0'; - va_start (args, fmt); - vfprintf_unfiltered (tmp_stream, fmt, args); - va_end (args); - } + return str; +} - while (*name == '\'') - name++; - fprintf_unfiltered (tmp_stream, - ("Hint: try '%s<TAB> or '%s<ESC-?>\n" - "(Note leading single quote.)"), - name, name); +/* Does P represent the end of a quote-enclosed linespec? */ - message = ui_file_xstrdup (tmp_stream, NULL); - make_cleanup (xfree, message); - throw_error (NOT_FOUND_ERROR, "%s", message); +static int +is_closing_quote_enclosed (const char *p) +{ + if (strchr (linespec_quote_characters, *p)) + ++p; + p = skip_spaces ((char *) p); + return (*p == '\0' || linespec_lexer_lex_keyword (p)); } -/* A callback function and the additional data to call it with. */ +/* Find the end of the parameter list that starts with *INPUT. + This helper function assists with lexing string segments + which might contain valid (non-terminating) commas. */ -struct symbol_and_data_callback +static char * +find_parameter_list_end (char *input) { - /* The callback to use. */ - symbol_found_callback_ftype *callback; - - /* Data to be passed to the callback. */ - void *data; -}; + char end_char, start_char; + int depth; + char *p; -/* A helper for iterate_over_all_matching_symtabs that is used to - restrict calls to another callback to symbols representing inline - symbols only. */ + start_char = *input; + if (start_char == '(') + end_char = ')'; + else if (start_char == '<') + end_char = '>'; + else + return NULL; -static int -iterate_inline_only (struct symbol *sym, void *d) -{ - if (SYMBOL_INLINED (sym)) + p = input; + depth = 0; + while (*p) { - struct symbol_and_data_callback *cad = d; - - return cad->callback (sym, cad->data); + if (*p == start_char) + ++depth; + else if (*p == end_char) + { + if (--depth == 0) + { + ++p; + break; + } + } + ++p; } - return 1; /* Continue iterating. */ + + return p; } -/* Some data for the expand_symtabs_matching callback. */ -struct symbol_matcher_data +/* Lex a string from the input in PARSER. */ + +static linespec_token +linespec_lexer_lex_string (linespec_parser *parser) { - /* The lookup name against which symbol name should be compared. */ - const char *lookup_name; + linespec_token token; + char *start = PARSER_STREAM (parser); - /* The routine to be used for comparison. */ - symbol_name_cmp_ftype symbol_name_cmp; -}; + token.type = LSTOKEN_STRING; -/* A helper for iterate_over_all_matching_symtabs that is passed as a - callback to the expand_symtabs_matching method. */ + /* If the input stream starts with a quote character, skip to the next + quote character, regardless of the content. */ + if (strchr (linespec_quote_characters, *PARSER_STREAM (parser))) + { + const char *end; + char quote_char = *PARSER_STREAM (parser); -static int -iterate_name_matcher (const char *name, void *d) -{ - const struct symbol_matcher_data *data = d; + /* Special case: Ada operators. */ + if (PARSER_STATE (parser)->language->la_language == language_ada + && quote_char == '\"') + { + int len = is_ada_operator (PARSER_STREAM (parser)); - if (data->symbol_name_cmp (name, data->lookup_name) == 0) - return 1; /* Expand this symbol's symbol table. */ - return 0; /* Skip this symbol. */ -} + if (len != 0) + { + /* The input is an Ada operator. Return the quoted string + as-is. */ + LS_TOKEN_STOKEN (token).ptr = PARSER_STREAM (parser); + LS_TOKEN_STOKEN (token).length = len; + PARSER_STREAM (parser) += len; + return token; + } -/* A helper that walks over all matching symtabs in all objfiles and - calls CALLBACK for each symbol matching NAME. If SEARCH_PSPACE is - not NULL, then the search is restricted to just that program - space. If INCLUDE_INLINE is nonzero then symbols representing - inlined instances of functions will be included in the result. */ + /* The input does not represent an Ada operator -- fall through + to normal quoted string handling. */ + } -static void -iterate_over_all_matching_symtabs (const char *name, - const domain_enum domain, - symbol_found_callback_ftype *callback, - void *data, - struct program_space *search_pspace, - int include_inline) -{ - struct objfile *objfile; - struct program_space *pspace; - struct symbol_matcher_data matcher_data; + /* Skip past the beginning quote. */ + ++(PARSER_STREAM (parser)); - matcher_data.lookup_name = name; - matcher_data.symbol_name_cmp = - current_language->la_get_symbol_name_cmp != NULL - ? current_language->la_get_symbol_name_cmp (name) - : strcmp_iw; + /* Mark the start of the string. */ + LS_TOKEN_STOKEN (token).ptr = PARSER_STREAM (parser); - ALL_PSPACES (pspace) - { - if (search_pspace != NULL && search_pspace != pspace) - continue; - if (pspace->executing_startup) + /* Skip to the ending quote. */ + end = skip_quote_char (PARSER_STREAM (parser), quote_char); + + /* Error if the input did not terminate properly. */ + if (end == NULL) + error (_("unmatched quote")); + + /* Skip over the ending quote and mark the length of the string. */ + PARSER_STREAM (parser) = (char *) ++end; + LS_TOKEN_STOKEN (token).length = PARSER_STREAM (parser) - 2 - start; + } + else + { + char *p; + + /* Otherwise, only identifier characters are permitted. + Spaces are the exception. In general, we keep spaces, + but only if the next characters in the input do not resolve + to one of the keywords. + + This allows users to forgo quoting CV-qualifiers, template arguments, + and similar common language constructs. */ + + while (1) + { + if (isspace (*PARSER_STREAM (parser))) + { + p = skip_spaces (PARSER_STREAM (parser)); + if (linespec_lexer_lex_keyword (p) != NULL) + { + LS_TOKEN_STOKEN (token).ptr = start; + LS_TOKEN_STOKEN (token).length + = PARSER_STREAM (parser) - start; + return token; + } + + /* Advance past the whitespace. */ + PARSER_STREAM (parser) = p; + } + + /* If the next character is EOI or (single) ':', the + string is complete; return the token. */ + if (*PARSER_STREAM (parser) == 0) + { + LS_TOKEN_STOKEN (token).ptr = start; + LS_TOKEN_STOKEN (token).length = PARSER_STREAM (parser) - start; + return token; + } + else if (PARSER_STREAM (parser)[0] == ':') + { + /* Do not tokenize the C++ scope operator. */ + if (PARSER_STREAM (parser)[1] == ':') + ++(PARSER_STREAM (parser)); + + /* Do not tokenify if the input length so far is one + (i.e, a single-letter drive name) and the next character + is a directory separator. This allows Windows-style + paths to be recognized as filenames without quoting it. */ + else if ((PARSER_STREAM (parser) - start) != 1 + || !IS_DIR_SEPARATOR (PARSER_STREAM (parser)[1])) + { + LS_TOKEN_STOKEN (token).ptr = start; + LS_TOKEN_STOKEN (token).length + = PARSER_STREAM (parser) - start; + return token; + } + } + /* Special case: permit quote-enclosed linespecs. */ + else if (parser->is_quote_enclosed + && strchr (linespec_quote_characters, + *PARSER_STREAM (parser)) + && is_closing_quote_enclosed (PARSER_STREAM (parser))) + { + LS_TOKEN_STOKEN (token).ptr = start; + LS_TOKEN_STOKEN (token).length = PARSER_STREAM (parser) - start; + return token; + } + /* Because commas may terminate a linespec and appear in + the middle of valid string input, special cases for + '<' and '(' are necessary. */ + else if (*PARSER_STREAM (parser) == '<' + || *PARSER_STREAM (parser) == '(') + { + char *p; + + p = find_parameter_list_end (PARSER_STREAM (parser)); + if (p != NULL) + { + PARSER_STREAM (parser) = p; + continue; + } + } + /* Commas are terminators, but not if they are part of an + operator name. */ + else if (*PARSER_STREAM (parser) == ',') + { + if ((PARSER_STATE (parser)->language->la_language + == language_cplus) + && (PARSER_STREAM (parser) - start) > 8 + /* strlen ("operator") */) + { + char *p = strstr (start, "operator"); + + if (p != NULL && is_operator_name (p)) + { + /* This is an operator name. Keep going. */ + ++(PARSER_STREAM (parser)); + continue; + } + } + + /* Comma terminates the string. */ + LS_TOKEN_STOKEN (token).ptr = start; + LS_TOKEN_STOKEN (token).length = PARSER_STREAM (parser) - start; + return token; + } + + /* Advance the stream. */ + ++(PARSER_STREAM (parser)); + } + } + + return token; +} + +/* Lex a single linespec token from PARSER. */ + +static linespec_token +linespec_lexer_lex_one (linespec_parser *parser) +{ + const char *keyword; + + if (parser->lexer.current.type == LSTOKEN_CONSUMED) + { + /* Skip any whitespace. */ + PARSER_STREAM (parser) = skip_spaces (PARSER_STREAM (parser)); + + /* Check for a keyword. */ + keyword = linespec_lexer_lex_keyword (PARSER_STREAM (parser)); + if (keyword != NULL) + { + parser->lexer.current.type = LSTOKEN_KEYWORD; + LS_TOKEN_KEYWORD (parser->lexer.current) = keyword; + return parser->lexer.current; + } + + /* Handle other tokens. */ + switch (*PARSER_STREAM (parser)) + { + case 0: + parser->lexer.current.type = LSTOKEN_EOI; + break; + + case '+': case '-': + case '0': case '1': case '2': case '3': case '4': + case '5': case '6': case '7': case '8': case '9': + parser->lexer.current = linespec_lexer_lex_number (parser); + break; + + case ':': + /* If we have a scope operator, lex the input as a string. + Otherwise, return LSTOKEN_COLON. */ + if (PARSER_STREAM (parser)[1] == ':') + parser->lexer.current = linespec_lexer_lex_string (parser); + else + { + parser->lexer.current.type = LSTOKEN_COLON; + ++(PARSER_STREAM (parser)); + } + break; + + case '\'': case '\"': + /* Special case: permit quote-enclosed linespecs. */ + if (parser->is_quote_enclosed + && is_closing_quote_enclosed (PARSER_STREAM (parser))) + { + ++(PARSER_STREAM (parser)); + parser->lexer.current.type = LSTOKEN_EOI; + } + else + parser->lexer.current = linespec_lexer_lex_string (parser); + break; + + case ',': + parser->lexer.current.type = LSTOKEN_COMMA; + LS_TOKEN_STOKEN (parser->lexer.current).ptr + = PARSER_STREAM (parser); + LS_TOKEN_STOKEN (parser->lexer.current).length = 1; + ++(PARSER_STREAM (parser)); + break; + + default: + /* If the input is not a number, it must be a string. + [Keywords were already considered above.] */ + parser->lexer.current = linespec_lexer_lex_string (parser); + break; + } + } + + return parser->lexer.current; +} + +/* Consume the current token and return the next token in PARSER's + input stream. */ + +static linespec_token +linespec_lexer_consume_token (linespec_parser *parser) +{ + parser->lexer.current.type = LSTOKEN_CONSUMED; + return linespec_lexer_lex_one (parser); +} + +/* Return the next token without consuming the current token. */ + +static linespec_token +linespec_lexer_peek_token (linespec_parser *parser) +{ + linespec_token next; + char *saved_stream = PARSER_STREAM (parser); + linespec_token saved_token = parser->lexer.current; + + next = linespec_lexer_consume_token (parser); + PARSER_STREAM (parser) = saved_stream; + parser->lexer.current = saved_token; + return next; +} + +/* Helper functions. */ + +/* Add SAL to SALS. */ + +static void +add_sal_to_sals_basic (struct symtabs_and_lines *sals, + struct symtab_and_line *sal) +{ + ++sals->nelts; + sals->sals = xrealloc (sals->sals, sals->nelts * sizeof (sals->sals[0])); + sals->sals[sals->nelts - 1] = *sal; +} + +/* Add SAL to SALS, and also update SELF->CANONICAL_NAMES to reflect + the new sal, if needed. If not NULL, SYMNAME is the name of the + symbol to use when constructing the new canonical name. */ + +static void +add_sal_to_sals (struct linespec_state *self, + struct symtabs_and_lines *sals, + struct symtab_and_line *sal, + const char *symname) +{ + add_sal_to_sals_basic (sals, sal); + + if (self->canonical) + { + char *canonical_name = NULL; + + self->canonical_names = xrealloc (self->canonical_names, + sals->nelts * sizeof (char *)); + if (sal->symtab && sal->symtab->filename) + { + char *filename = sal->symtab->filename; + + /* Note that the filter doesn't have to be a valid linespec + input. We only apply the ":LINE" treatment to Ada for + the time being. */ + if (symname != NULL && sal->line != 0 + && self->language->la_language == language_ada) + canonical_name = xstrprintf ("%s:%s:%d", filename, symname, + sal->line); + else if (symname != NULL) + canonical_name = xstrprintf ("%s:%s", filename, symname); + else + canonical_name = xstrprintf ("%s:%d", filename, sal->line); + } + + self->canonical_names[sals->nelts - 1] = canonical_name; + } +} + +/* A hash function for address_entry. */ + +static hashval_t +hash_address_entry (const void *p) +{ + const struct address_entry *aep = p; + hashval_t hash; + + hash = iterative_hash_object (aep->pspace, 0); + return iterative_hash_object (aep->addr, hash); +} + +/* An equality function for address_entry. */ + +static int +eq_address_entry (const void *a, const void *b) +{ + const struct address_entry *aea = a; + const struct address_entry *aeb = b; + + return aea->pspace == aeb->pspace && aea->addr == aeb->addr; +} + +/* Check whether the address, represented by PSPACE and ADDR, is + already in the set. If so, return 0. Otherwise, add it and return + 1. */ + +static int +maybe_add_address (htab_t set, struct program_space *pspace, CORE_ADDR addr) +{ + struct address_entry e, *p; + void **slot; + + e.pspace = pspace; + e.addr = addr; + slot = htab_find_slot (set, &e, INSERT); + if (*slot) + return 0; + + p = XNEW (struct address_entry); + memcpy (p, &e, sizeof (struct address_entry)); + *slot = p; + + return 1; +} + +/* A callback function and the additional data to call it with. */ + +struct symbol_and_data_callback +{ + /* The callback to use. */ + symbol_found_callback_ftype *callback; + + /* Data to be passed to the callback. */ + void *data; +}; + +/* A helper for iterate_over_all_matching_symtabs that is used to + restrict calls to another callback to symbols representing inline + symbols only. */ + +static int +iterate_inline_only (struct symbol *sym, void *d) +{ + if (SYMBOL_INLINED (sym)) + { + struct symbol_and_data_callback *cad = d; + + return cad->callback (sym, cad->data); + } + return 1; /* Continue iterating. */ +} + +/* Some data for the expand_symtabs_matching callback. */ + +struct symbol_matcher_data +{ + /* The lookup name against which symbol name should be compared. */ + const char *lookup_name; + + /* The routine to be used for comparison. */ + symbol_name_cmp_ftype symbol_name_cmp; +}; + +/* A helper for iterate_over_all_matching_symtabs that is passed as a + callback to the expand_symtabs_matching method. */ + +static int +iterate_name_matcher (const char *name, void *d) +{ + const struct symbol_matcher_data *data = d; + + if (data->symbol_name_cmp (name, data->lookup_name) == 0) + return 1; /* Expand this symbol's symbol table. */ + return 0; /* Skip this symbol. */ +} + +/* A helper that walks over all matching symtabs in all objfiles and + calls CALLBACK for each symbol matching NAME. If SEARCH_PSPACE is + not NULL, then the search is restricted to just that program + space. If INCLUDE_INLINE is nonzero then symbols representing + inlined instances of functions will be included in the result. */ + +static void +iterate_over_all_matching_symtabs (struct linespec_state *state, + const char *name, + const domain_enum domain, + symbol_found_callback_ftype *callback, + void *data, + struct program_space *search_pspace, + int include_inline) +{ + struct objfile *objfile; + struct program_space *pspace; + struct symbol_matcher_data matcher_data; + + matcher_data.lookup_name = name; + matcher_data.symbol_name_cmp = + state->language->la_get_symbol_name_cmp != NULL + ? state->language->la_get_symbol_name_cmp (name) + : strcmp_iw; + + ALL_PSPACES (pspace) + { + if (search_pspace != NULL && search_pspace != pspace) + continue; + if (pspace->executing_startup) continue; set_current_program_space (pspace); @@ -468,7 +1040,7 @@ get_search_block (struct symtab *symtab) } /* A helper for find_method. This finds all methods in type T which - match NAME. It adds resulting symbol names to RESULT_NAMES, and + match NAME. It adds matching symbol names to RESULT_NAMES, and adds T's direct superclasses to SUPERCLASSES. */ static void @@ -542,14 +1114,14 @@ find_methods (struct type *t, const char *name, strings. Also, ignore the char within a template name, like a ',' within foo<int, int>. */ -static char * -find_toplevel_char (char *s, char c) +static const char * +find_toplevel_char (const char *s, char c) { int quoted = 0; /* zero if we're not in quotes; '"' if we're in a double-quoted string; '\'' if we're in a single-quoted string. */ int depth = 0; /* Number of unclosed parens we've seen. */ - char *scan; + const char *scan; for (scan = s; *scan; scan++) { @@ -573,23 +1145,34 @@ find_toplevel_char (char *s, char c) return 0; } -/* Determines if the gives string corresponds to an Objective-C method - representation, such as -[Foo bar:] or +[Foo bar]. Objective-C symbols - are allowed to have spaces and parentheses in them. */ +/* The string equivalent of find_toplevel_char. Returns a pointer + to the location of NEEDLE in HAYSTACK, ignoring any occurrences + inside "()" and "<>". Returns NULL if NEEDLE was not found. */ -static int -is_objc_method_format (const char *s) +static const char * +find_toplevel_string (const char *haystack, const char *needle) { - if (s == NULL || *s == '\0') - return 0; - /* Handle arguments with the format FILENAME:SYMBOL. */ - if ((s[0] == ':') && (strchr ("+-", s[1]) != NULL) - && (s[2] == '[') && strchr(s, ']')) - return 1; - /* Handle arguments that are just SYMBOL. */ - else if ((strchr ("+-", s[0]) != NULL) && (s[1] == '[') && strchr(s, ']')) - return 1; - return 0; + const char *s = haystack; + + do + { + s = find_toplevel_char (s, *needle); + + if (s != NULL) + { + /* Found first char in HAYSTACK; check rest of string. */ + if (strncmp (s, needle, strlen (needle)) == 0) + return s; + + /* Didn't find it; loop over HAYSTACK, looking for the next + instance of the first character of NEEDLE. */ + ++s; + } + } + while (s != NULL && *s != '\0'); + + /* NEEDLE was not found in HAYSTACK. */ + return NULL; } /* Given FILTERS, a list of canonical names, filter the sals in RESULT @@ -758,118 +1341,625 @@ decode_line_2 (struct linespec_state *self, do_cleanups (old_chain); } -/* Valid delimiters for linespec keywords "if", "thread" or "task". */ +\f -static int -is_linespec_boundary (char c) +/* The parser of linespec itself. */ + +/* Throw an appropriate error when SYMBOL is not found (optionally in + FILENAME). */ + +static void ATTRIBUTE_NORETURN +symbol_not_found_error (char *symbol, char *filename) { - return c == ' ' || c == '\t' || c == '\0' || c == ','; + if (symbol == NULL) + symbol = ""; + + if (!have_full_symbols () + && !have_partial_symbols () + && !have_minimal_symbols ()) + throw_error (NOT_FOUND_ERROR, + _("No symbol table is loaded. Use the \"file\" command.")); + + /* If SYMBOL starts with '$', the user attempted to either lookup + a function/variable in his code starting with '$' or an internal + variable of that name. Since we do not know which, be concise and + explain both possibilities. */ + if (*symbol == '$') + { + if (filename) + throw_error (NOT_FOUND_ERROR, + _("Undefined convenience variable or function \"%s\" " + "not defined in \"%s\"."), symbol, filename); + else + throw_error (NOT_FOUND_ERROR, + _("Undefined convenience variable or function \"%s\" " + "not defined."), symbol); + } + else + { + if (filename) + throw_error (NOT_FOUND_ERROR, + _("Function \"%s\" not defined in \"%s\"."), + symbol, filename); + else + throw_error (NOT_FOUND_ERROR, + _("Function \"%s\" not defined."), symbol); + } } -/* A helper function for decode_line_1 and friends which skips P - past any method overload information at the beginning of P, e.g., - "(const struct foo *)". +/* Throw an appropriate error when an unexpected token is encountered + in the input. */ - This function assumes that P has already been validated to contain - overload information, and it will assert if *P != '('. */ -static char * -find_method_overload_end (char *p) +static void ATTRIBUTE_NORETURN +unexpected_linespec_error (linespec_parser *parser) { - int depth = 0; + linespec_token token; + static const char * token_type_strings[] + = {"keyword", "colon", "string", "number", "comma", "end of input"}; - gdb_assert (*p == '('); + /* Get the token that generated the error. */ + token = linespec_lexer_lex_one (parser); - while (*p) + /* Finally, throw the error. */ + if (token.type == LSTOKEN_STRING || token.type == LSTOKEN_NUMBER + || token.type == LSTOKEN_KEYWORD) { - if (*p == '(') - ++depth; - else if (*p == ')') + char *string; + struct cleanup *cleanup; + + string = copy_token_string (token); + cleanup = make_cleanup (xfree, string); + throw_error (GENERIC_ERROR, + _("malformed linespec error: unexpected %s, \"%s\""), + token_type_strings[token.type], string); + } + else + throw_error (GENERIC_ERROR, + _("malformed linespec error: unexpected %s"), + token_type_strings[token.type]); +} + +/* Parse and return a line offset in STRING. */ + +static struct line_offset +linespec_parse_line_offset (char *string) +{ + struct line_offset line_offset = {0, LINE_OFFSET_NONE}; + + if (*string == '+') + { + line_offset.sign = LINE_OFFSET_PLUS; + ++string; + } + else if (*string == '-') + { + line_offset.sign = LINE_OFFSET_MINUS; + ++string; + } + + /* Right now, we only allow base 10 for offsets. */ + line_offset.offset = atoi (string); + return line_offset; +} + +/* Parse the basic_spec in PARSER's input. */ + +static void +linespec_parse_basic (linespec_parser *parser) +{ + char *name; + linespec_token token; + VEC (symbolp) *symbols, *labels; + VEC (minsym_and_objfile_d) *minimal_symbols; + struct cleanup *cleanup; + + /* Get the next token. */ + token = linespec_lexer_lex_one (parser); + + /* If it is EOI or KEYWORD, issue an error. */ + if (token.type == LSTOKEN_KEYWORD || token.type == LSTOKEN_EOI) + unexpected_linespec_error (parser); + /* If it is a LSTOKEN_NUMBER, we have an offset. */ + else if (token.type == LSTOKEN_NUMBER) + { + /* Record the line offset and get the next token. */ + name = copy_token_string (token); + cleanup = make_cleanup (xfree, name); + PARSER_RESULT (parser)->line_offset = linespec_parse_line_offset (name); + do_cleanups (cleanup); + + /* Get the next token. */ + token = linespec_lexer_consume_token (parser); + + /* If the next token is a comma, stop parsing and return. */ + if (token.type == LSTOKEN_COMMA) + return; + + /* If the next token is anything but EOI or KEYWORD, issue + an error. */ + if (token.type != LSTOKEN_KEYWORD && token.type != LSTOKEN_EOI) + unexpected_linespec_error (parser); + } + + if (token.type == LSTOKEN_KEYWORD || token.type == LSTOKEN_EOI) + return; + + /* Next token must be LSTOKEN_STRING. */ + if (token.type != LSTOKEN_STRING) + unexpected_linespec_error (parser); + + /* The current token will contain the name of a function, method, + or label. */ + name = copy_token_string (token); + cleanup = make_cleanup (xfree, name); + + /* Try looking it up as a function/method. */ + find_linespec_symbols (PARSER_STATE (parser), + PARSER_RESULT (parser)->file_symtabs, name, + &symbols, &minimal_symbols); + + if (symbols != NULL || minimal_symbols != NULL) + { + PARSER_RESULT (parser)->function_symbols = symbols; + PARSER_RESULT (parser)->minimal_symbols = minimal_symbols; + PARSER_RESULT (parser)->function_name = name; + symbols = NULL; + discard_cleanups (cleanup); + } + else + { + /* NAME was not a function or a method. So it must be a label + name. */ + labels = find_label_symbols (PARSER_STATE (parser), NULL, + &symbols, name); + if (labels != NULL) { - if (--depth == 0) + PARSER_RESULT (parser)->labels.label_symbols = labels; + PARSER_RESULT (parser)->labels.function_symbols = symbols; + PARSER_RESULT (parser)->label_name = name; + symbols = NULL; + discard_cleanups (cleanup); + } + else + { + /* The name is also not a label. Abort parsing. Do not throw + an error here. parse_linespec will do it for us. */ + + /* Save a copy of the name we were trying to lookup. */ + PARSER_RESULT (parser)->function_name = name; + discard_cleanups (cleanup); + return; + } + } + + /* Get the next token. */ + token = linespec_lexer_consume_token (parser); + + if (token.type == LSTOKEN_COLON) + { + /* User specified a label or a lineno. */ + token = linespec_lexer_consume_token (parser); + + if (token.type == LSTOKEN_NUMBER) + { + /* User specified an offset. Record the line offset and + get the next token. */ + name = copy_token_string (token); + cleanup = make_cleanup (xfree, name); + PARSER_RESULT (parser)->line_offset + = linespec_parse_line_offset (name); + do_cleanups (cleanup); + + /* Ge the next token. */ + token = linespec_lexer_consume_token (parser); + } + else if (token.type == LSTOKEN_STRING) + { + /* Grab a copy of the label's name and look it up. */ + name = copy_token_string (token); + cleanup = make_cleanup (xfree, name); + labels = find_label_symbols (PARSER_STATE (parser), + PARSER_RESULT (parser)->function_symbols, + &symbols, name); + + if (labels != NULL) { - ++p; - break; + PARSER_RESULT (parser)->labels.label_symbols = labels; + PARSER_RESULT (parser)->labels.function_symbols = symbols; + PARSER_RESULT (parser)->label_name = name; + symbols = NULL; + discard_cleanups (cleanup); + } + else + { + /* We don't know what it was, but it isn't a label. */ + throw_error (NOT_FOUND_ERROR, + _("No label \"%s\" defined in function \"%s\"."), + name, PARSER_RESULT (parser)->function_name); + } + + /* Check for a line offset. */ + token = linespec_lexer_consume_token (parser); + if (token.type == LSTOKEN_COLON) + { + /* Get the next token. */ + token = linespec_lexer_consume_token (parser); + + /* It must be a line offset. */ + if (token.type != LSTOKEN_NUMBER) + unexpected_linespec_error (parser); + + /* Record the lione offset and get the next token. */ + name = copy_token_string (token); + cleanup = make_cleanup (xfree, name); + + PARSER_RESULT (parser)->line_offset + = linespec_parse_line_offset (name); + do_cleanups (cleanup); + + /* Get the next token. */ + token = linespec_lexer_consume_token (parser); } } - ++p; + else + { + /* Trailing ':' in the input. Issue an error. */ + unexpected_linespec_error (parser); + } } +} - return p; +/* Canonicalize the linespec contained in LS. The result is saved into + STATE->canonical. */ + +static void +canonicalize_linespec (struct linespec_state *state, linespec_p ls) +{ + /* If canonicalization was not requested, no need to do anything. */ + if (!state->canonical) + return; + + /* Shortcut expressions, which can only appear by themselves. */ + if (ls->expression != NULL) + state->canonical->addr_string = xstrdup (ls->expression); + else + { + struct ui_file *buf; + int need_colon = 0; + + buf = mem_fileopen (); + if (ls->source_filename) + { + fputs_unfiltered (ls->source_filename, buf); + need_colon = 1; + } + + if (ls->function_name) + { + if (need_colon) + fputc_unfiltered (':', buf); + fputs_unfiltered (ls->function_name, buf); + need_colon = 1; + } + + if (ls->label_name) + { + if (need_colon) + fputc_unfiltered (':', buf); + + if (ls->function_name == NULL) + { + struct symbol *s; + + /* No function was specified, so add the symbol name. */ + gdb_assert (ls->labels.function_symbols != NULL + && (VEC_length (symbolp, ls->labels.function_symbols) + == 1)); + s = VEC_index (symbolp, ls->labels.function_symbols, 0); + fputs_unfiltered (SYMBOL_NATURAL_NAME (s), buf); + fputc_unfiltered (':', buf); + } + + fputs_unfiltered (ls->label_name, buf); + need_colon = 1; + state->canonical->special_display = 1; + } + + if (ls->line_offset.sign != LINE_OFFSET_UNKNOWN) + { + if (need_colon) + fputc_unfiltered (':', buf); + fprintf_filtered (buf, "%s%d", + (ls->line_offset.sign == LINE_OFFSET_NONE ? "" + : (ls->line_offset.sign + == LINE_OFFSET_PLUS ? "+" : "-")), + ls->line_offset.offset); + } + + state->canonical->addr_string = ui_file_xstrdup (buf, NULL); + ui_file_delete (buf); + } } -/* Keep important information used when looking up a name. This includes - template parameters, overload information, and important keywords, including - the possible Java trailing type. */ +/* Given a line offset in LS, construct the relevant SALs. */ -static char * -keep_name_info (char *p, int on_boundary) +static struct symtabs_and_lines +create_sals_line_offset (struct linespec_state *self, + linespec_p ls) { - const char *quotes = get_gdb_completer_quote_characters (); - char *saved_p = p; - int nest = 0; + struct symtabs_and_lines values; + struct symtab_and_line val; + int use_default = 0; - while (*p) + init_sal (&val); + values.sals = NULL; + values.nelts = 0; + + /* This is where we need to make sure we have good defaults. + We must guarantee that this section of code is never executed + when we are called with just a function anme, since + set_default_source_symtab_and_line uses + select_source_symtab that calls us with such an argument. */ + + if (VEC_length (symtab_p, ls->file_symtabs) == 1 + && VEC_index (symtab_p, ls->file_symtabs, 0) == NULL) { - if (strchr (quotes, *p)) - break; + set_current_program_space (self->program_space); - if (*p == ',' && !nest) - break; + /* Make sure we have at least a default source line. */ + set_default_source_symtab_and_line (); + initialize_defaults (&self->default_symtab, &self->default_line); + VEC_pop (symtab_p, ls->file_symtabs); + VEC_free (symtab_p, ls->file_symtabs); + ls->file_symtabs + = collect_symtabs_from_filename (self->default_symtab->filename); + use_default = 1; + } + + val.line = ls->line_offset.offset; + switch (ls->line_offset.sign) + { + case LINE_OFFSET_PLUS: + if (ls->line_offset.offset == 0) + val.line = 5; + if (use_default) + val.line = self->default_line + val.line; + break; + + case LINE_OFFSET_MINUS: + if (ls->line_offset.offset == 0) + val.line = 15; + if (use_default) + val.line = self->default_line - val.line; + else + val.line = -val.line; + break; + + case LINE_OFFSET_NONE: + break; /* No need to adjust val.line. */ + } + + if (self->list_mode) + decode_digits_list_mode (self, ls, &values, val); + else + { + struct linetable_entry *best_entry = NULL; + int *filter; + struct block **blocks; + struct cleanup *cleanup; + struct symtabs_and_lines intermediate_results; + int i, j; + + intermediate_results.sals = NULL; + intermediate_results.nelts = 0; + + decode_digits_ordinary (self, ls, val.line, &intermediate_results, + &best_entry); + if (intermediate_results.nelts == 0 && best_entry != NULL) + decode_digits_ordinary (self, ls, best_entry->line, + &intermediate_results, &best_entry); + + cleanup = make_cleanup (xfree, intermediate_results.sals); - if (on_boundary && !nest) + /* For optimized code, the compiler can scatter one source line + across disjoint ranges of PC values, even when no duplicate + functions or inline functions are involved. For example, + 'for (;;)' inside a non-template, non-inline, and non-ctor-or-dtor + function can result in two PC ranges. In this case, we don't + want to set a breakpoint on the first PC of each range. To filter + such cases, we use containing blocks -- for each PC found + above, we see if there are other PCs that are in the same + block. If yes, the other PCs are filtered out. */ + + filter = XNEWVEC (int, intermediate_results.nelts); + make_cleanup (xfree, filter); + blocks = XNEWVEC (struct block *, intermediate_results.nelts); + make_cleanup (xfree, blocks); + + for (i = 0; i < intermediate_results.nelts; ++i) { - const char *const words[] = { "if", "thread", "task" }; - int wordi; + set_current_program_space (intermediate_results.sals[i].pspace); - for (wordi = 0; wordi < ARRAY_SIZE (words); wordi++) - if (strncmp (p, words[wordi], strlen (words[wordi])) == 0 - && is_linespec_boundary (p[strlen (words[wordi])])) - break; - if (wordi < ARRAY_SIZE (words)) - break; + filter[i] = 1; + blocks[i] = block_for_pc_sect (intermediate_results.sals[i].pc, + intermediate_results.sals[i].section); } - if (*p == '(' || *p == '<' || *p == '[') - nest++; - else if ((*p == ')' || *p == '>' || *p == ']') && nest > 0) - nest--; + for (i = 0; i < intermediate_results.nelts; ++i) + { + if (blocks[i] != NULL) + for (j = i + 1; j < intermediate_results.nelts; ++j) + { + if (blocks[j] == blocks[i]) + { + filter[j] = 0; + break; + } + } + } - p++; + for (i = 0; i < intermediate_results.nelts; ++i) + if (filter[i]) + { + struct symbol *sym = (blocks[i] + ? block_containing_function (blocks[i]) + : NULL); - /* The ',' check could fail on "operator ,". */ - p += cp_validate_operator (p); + if (self->funfirstline) + skip_prologue_sal (&intermediate_results.sals[i]); + /* Make sure the line matches the request, not what was + found. */ + intermediate_results.sals[i].line = val.line; + add_sal_to_sals (self, &values, &intermediate_results.sals[i], + sym ? SYMBOL_NATURAL_NAME (sym) : NULL); + } - on_boundary = is_linespec_boundary (p[-1]); + do_cleanups (cleanup); } - while (p > saved_p && is_linespec_boundary (p[-1])) - p--; + if (values.nelts == 0) + { + if (ls->source_filename) + throw_error (NOT_FOUND_ERROR, _("No line %d in file \"%s\"."), + val.line, ls->source_filename); + else + throw_error (NOT_FOUND_ERROR, _("No line %d in the current file."), + val.line); + } - return p; + return values; } -\f -/* The parser of linespec itself. */ +/* Create and return SALs from the linespec LS. */ + +static struct symtabs_and_lines +convert_linespec_to_sals (struct linespec_state *state, linespec_p ls) +{ + struct symtabs_and_lines sals = {NULL, 0}; + + if (ls->expression != NULL) + { + /* We have an expression. No other attribute is allowed. */ + sals.sals = XMALLOC (struct symtab_and_line); + sals.nelts = 1; + sals.sals[0] = find_pc_line (ls->expr_pc, 0); + sals.sals[0].pc = ls->expr_pc; + sals.sals[0].section = find_pc_overlay (ls->expr_pc); + sals.sals[0].explicit_pc = 1; + } + else if (ls->labels.label_symbols != NULL) + { + /* We have just a bunch of functions/methods or labels. */ + int i; + struct symtab_and_line sal; + struct symbol *sym; + + for (i = 0; VEC_iterate (symbolp, ls->labels.label_symbols, i, sym); ++i) + { + symbol_to_sal (&sal, state->funfirstline, sym); + add_sal_to_sals (state, &sals, &sal, + SYMBOL_NATURAL_NAME (sym)); + } + } + else if (ls->function_symbols != NULL || ls->minimal_symbols != NULL) + { + /* We have just a bunch of functions and/or methods. */ + int i; + struct symtab_and_line sal; + struct symbol *sym; + minsym_and_objfile_d *elem; + struct program_space *pspace; + + if (ls->function_symbols != NULL) + { + /* Sort symbols so that symbols with the same program space are next + to each other. */ + qsort (VEC_address (symbolp, ls->function_symbols), + VEC_length (symbolp, ls->function_symbols), + sizeof (symbolp), compare_symbols); + + for (i = 0; VEC_iterate (symbolp, ls->function_symbols, i, sym); ++i) + { + pspace = SYMTAB_PSPACE (SYMBOL_SYMTAB (sym)); + set_current_program_space (pspace); + symbol_to_sal (&sal, state->funfirstline, sym); + if (maybe_add_address (state->addr_set, pspace, sal.pc)) + add_sal_to_sals (state, &sals, &sal, SYMBOL_NATURAL_NAME (sym)); + } + } + + if (ls->minimal_symbols != NULL) + { + /* Sort minimal symbols by program space, too. */ + qsort (VEC_address (minsym_and_objfile_d, ls->minimal_symbols), + VEC_length (minsym_and_objfile_d, ls->minimal_symbols), + sizeof (minsym_and_objfile_d), compare_msymbols); + + for (i = 0; + VEC_iterate (minsym_and_objfile_d, ls->minimal_symbols, i, elem); + ++i) + { + pspace = SYMBOL_OBJ_SECTION (elem->minsym)->objfile->pspace; + set_current_program_space (pspace); + minsym_found (state, elem->objfile, elem->minsym, &sals); + } + } + } + else if (ls->line_offset.sign != LINE_OFFSET_UNKNOWN) + { + /* Only an offset was specified. */ + sals = create_sals_line_offset (state, ls); + + /* Make sure we have a filename for canonicalization. */ + if (ls->source_filename == NULL) + ls->source_filename = xstrdup (state->default_symtab->filename); + } + else + { + /* We haven't found any results... */ + return sals; + } + + canonicalize_linespec (state, ls); + + if (sals.nelts > 0 && state->canonical != NULL) + state->canonical->pre_expanded = 1; + + return sals; +} -/* Parse a string that specifies a line number. +/* Parse a string that specifies a linespec. Pass the address of a char * variable; that variable will be advanced over the characters actually parsed. - The string can be: + The basic grammar of linespecs: - LINENUM -- that line number in current file. PC returned is 0. - FILE:LINENUM -- that line in that file. PC returned is 0. - FUNCTION -- line number of openbrace of that function. - PC returned is the start of the function. - LABEL -- a label in the current scope - VARIABLE -- line number of definition of that variable. - PC returned is 0. - FILE:FUNCTION -- likewise, but prefer functions in that file. - *EXPR -- line in which address EXPR appears. + linespec -> expr_spec | var_spec | basic_spec + expr_spec -> '*' STRING + var_spec -> '$' (STRING | NUMBER) - This may all be followed by an "if EXPR", which we ignore. + basic_spec -> file_offset_spec | function_spec | label_spec + file_offset_spec -> opt_file_spec offset_spec + function_spec -> opt_file_spec function_name_spec opt_label_spec + label_spec -> label_name_spec - FUNCTION may be an undebuggable function found in minimal symbol table. + opt_file_spec -> "" | file_name_spec ':' + opt_label_spec -> "" | ':' label_name_spec + + file_name_spec -> STRING + function_name_spec -> STRING + label_name_spec -> STRING + function_name_spec -> STRING + offset_spec -> NUMBER + -> '+' NUMBER + -> '-' NUMBER + + This may all be followed by several keywords such as "if EXPR", + which we ignore. + + A comma will terminate parsing. + + The function may be an undebuggable function found in minimal symbol table. If the argument FUNFIRSTLINE is nonzero, we want the first line of real code inside a function when a function is specified, and it is @@ -889,344 +1979,208 @@ keep_name_info (char *p, int on_boundary) if no file is validly specified. Callers must check that. Also, the line number returned may be invalid. */ -/* We allow single quotes in various places. This is a hideous - kludge, which exists because the completer can't yet deal with the - lack of single quotes. FIXME: write a linespec_completer which we - can use as appropriate instead of make_symbol_completion_list. */ +/* Parse the linespec in ARGPTR. */ static struct symtabs_and_lines -decode_line_internal (struct linespec_state *self, char **argptr) +parse_linespec (linespec_parser *parser, char **argptr) { - char *p; - char *q; - - char *copy; - /* This says whether or not something in *ARGPTR is quoted with - completer_quotes (i.e. with single quotes). */ - int is_quoted; - /* Is *ARGPTR enclosed in double quotes? */ - int is_quote_enclosed; - int is_objc_method = 0; - char *saved_arg = *argptr; - /* If IS_QUOTED, the end of the quoted bit. */ - char *end_quote = NULL; - /* Is *ARGPTR enclosed in single quotes? */ - int is_squote_enclosed = 0; - /* The "first half" of the linespec. */ - char *first_half; - - /* If we are parsing `function:label', this holds the symbols - matching the function name. */ - VEC (symbolp) *function_symbols = NULL; - /* If FUNCTION_SYMBOLS is not NULL, then this is the exception that - was thrown when trying to parse a filename. */ + linespec_token token; + struct symtabs_and_lines values; volatile struct gdb_exception file_exception; + struct cleanup *cleanup; + + /* A special case to start. It has become quite popular for + IDEs to work around bugs in the previous parser by quoting + the entire linespec, so we attempt to deal with this nicely. */ + parser->is_quote_enclosed = 0; + if (!is_ada_operator (*argptr) + && strchr (linespec_quote_characters, **argptr) != NULL) + { + const char *end; - struct cleanup *cleanup = make_cleanup (null_cleanup, NULL); - - /* Defaults have defaults. */ - - initialize_defaults (&self->default_symtab, &self->default_line); - - /* See if arg is *PC. */ - - if (**argptr == '*') - { - do_cleanups (cleanup); - return decode_indirect (self, argptr); - } - - is_quoted = (strchr (get_gdb_completer_quote_characters (), - **argptr) != NULL); - - if (is_quoted) - { - end_quote = skip_quoted (*argptr); - if (*end_quote == '\0') - is_squote_enclosed = 1; - } - - /* Check to see if it's a multipart linespec (with colons or - periods). */ - - /* Locate the end of the first half of the linespec. - After the call, for instance, if the argptr string is "foo.c:123" - p will point at ":123". If there is only one part, like "foo", p - will point to "". If this is a C++ name, like "A::B::foo", p will - point to "::B::foo". Argptr is not changed by this call. */ - - first_half = p = locate_first_half (argptr, &is_quote_enclosed); - - /* First things first: if ARGPTR starts with a filename, get its - symtab and strip the filename from ARGPTR. - Avoid calling symtab_from_filename if we know can, - it can be expensive. We know we can avoid the call if we see a - single word (e.g., "break NAME") or if we see a qualified C++ - name ("break QUAL::NAME"). */ - - if (*p != '\0' && !(p[0] == ':' && p[1] == ':')) - { - TRY_CATCH (file_exception, RETURN_MASK_ERROR) - { - self->file_symtabs = symtabs_from_filename (argptr, p, - is_quote_enclosed, - &self->user_filename); - } - - if (file_exception.reason >= 0) - { - /* Check for single quotes on the non-filename part. */ - is_quoted = (**argptr - && strchr (get_gdb_completer_quote_characters (), - **argptr) != NULL); - if (is_quoted) - end_quote = skip_quoted (*argptr); - - /* Locate the next "half" of the linespec. */ - first_half = p = locate_first_half (argptr, &is_quote_enclosed); - } - - if (VEC_empty (symtab_p, self->file_symtabs)) + end = skip_quote_char (*argptr + 1, **argptr); + if (end != NULL && is_closing_quote_enclosed (end)) { - /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */ - VEC_safe_push (symtab_p, self->file_symtabs, NULL); + /* Here's the special case. Skip ARGPTR past the initial + quote. */ + ++(*argptr); + parser->is_quote_enclosed = 1; } } - else - { - /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */ - VEC_safe_push (symtab_p, self->file_symtabs, NULL); - } - /* Check if this is an Objective-C method (anything that starts with - a '+' or '-' and a '['). */ - if (is_objc_method_format (p)) - is_objc_method = 1; + parser->lexer.saved_arg = *argptr; + parser->lexer.stream = argptr; + file_exception.reason = 0; - /* Check if the symbol could be an Objective-C selector. */ + /* Initialize the default symtab and line offset. */ + initialize_defaults (&PARSER_STATE (parser)->default_symtab, + &PARSER_STATE (parser)->default_line); - { - struct symtabs_and_lines values; + /* Objective-C shortcut. */ + values = decode_objc (PARSER_STATE (parser), PARSER_RESULT (parser), argptr); + if (values.sals != NULL) + return values; - values = decode_objc (self, argptr); - if (values.sals != NULL) - { - do_cleanups (cleanup); - return values; - } - } + /* Start parsing. */ - /* Does it look like there actually were two parts? */ + /* Get the first token. */ + token = linespec_lexer_lex_one (parser); - if (p[0] == ':' || p[0] == '.') + /* It must be either LSTOKEN_STRING or LSTOKEN_NUMBER. */ + if (token.type == LSTOKEN_STRING && *LS_TOKEN_STOKEN (token).ptr == '*') { - /* Is it a C++ or Java compound data structure? - The check on p[1] == ':' is capturing the case of "::", - since p[0]==':' was checked above. - Note that the call to decode_compound does everything - for us, including the lookup on the symbol table, so we - can return now. */ - - if (p[0] == '.' || p[1] == ':') - { - /* We only perform this check for the languages where it might - make sense. For instance, Ada does not use this type of - syntax, and trying to apply this logic on an Ada linespec - may trigger a spurious error (for instance, decode_compound - does not like expressions such as `ops."<"', which is a - valid function name in Ada). */ - if (current_language->la_language == language_c - || current_language->la_language == language_cplus - || current_language->la_language == language_java) - { - struct symtabs_and_lines values; - volatile struct gdb_exception ex; - char *saved_argptr = *argptr; + char *expr, *copy; - if (is_quote_enclosed) - ++saved_arg; + /* User specified an expression, *EXPR. */ + copy = expr = copy_token_string (token); + cleanup = make_cleanup (xfree, expr); + PARSER_RESULT (parser)->expr_pc = linespec_expression_to_pc (©); + discard_cleanups (cleanup); + PARSER_RESULT (parser)->expression = expr; - /* Initialize it just to avoid a GCC false warning. */ - memset (&values, 0, sizeof (values)); + /* This is a little hacky/tricky. If linespec_expression_to_pc + did not evaluate the entire token, then we must find the + string COPY inside the original token buffer. */ + if (*copy != '\0') + { + PARSER_STREAM (parser) = strstr (parser->lexer.saved_arg, copy); + gdb_assert (PARSER_STREAM (parser) != NULL); + } - TRY_CATCH (ex, RETURN_MASK_ERROR) - { - values = decode_compound (self, argptr, saved_arg, p); - } - if ((is_quoted || is_squote_enclosed) && **argptr == '\'') - *argptr = *argptr + 1; + /* Consume the token. */ + linespec_lexer_consume_token (parser); - if (ex.reason >= 0) - { - do_cleanups (cleanup); - return values; - } + goto convert_to_sals; + } + else if (token.type == LSTOKEN_STRING && *LS_TOKEN_STOKEN (token).ptr == '$') + { + char *var; - if (ex.error != NOT_FOUND_ERROR) - throw_exception (ex); + /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */ + VEC_safe_push (symtab_p, PARSER_RESULT (parser)->file_symtabs, NULL); - *argptr = saved_argptr; - } - } - else - { - /* If there was an exception looking up a specified filename earlier, - then check whether we were really given `function:label'. */ - if (file_exception.reason < 0) - { - function_symbols = find_function_symbols (argptr, p, - is_quote_enclosed, - &self->user_function); + /* User specified a convenience variable or history value. */ + var = copy_token_string (token); + cleanup = make_cleanup (xfree, var); + PARSER_RESULT (parser)->line_offset + = linespec_parse_variable (PARSER_STATE (parser), var); - /* If we did not find a function, re-throw the original - exception. */ - if (!function_symbols) - throw_exception (file_exception); + /* If a line_offset wasn't found (VAR is the name of a user + variable/function), then skip to normal symbol processing. */ + if (PARSER_RESULT (parser)->line_offset.sign != LINE_OFFSET_UNKNOWN) + { + discard_cleanups (cleanup); - make_cleanup (VEC_cleanup (symbolp), &function_symbols); - } + /* Consume this token. */ + linespec_lexer_consume_token (parser); - /* Check for single quotes on the non-filename part. */ - if (!is_quoted) - { - is_quoted = (**argptr - && strchr (get_gdb_completer_quote_characters (), - **argptr) != NULL); - if (is_quoted) - end_quote = skip_quoted (*argptr); - } + goto convert_to_sals; } - } - /* self->file_symtabs holds the specified file symtabs, or 0 if no file - specified. - If we are parsing `function:symbol', then FUNCTION_SYMBOLS holds the - functions before the `:'. - arg no longer contains the file name. */ + do_cleanups (cleanup); + } + else if (token.type != LSTOKEN_STRING && token.type != LSTOKEN_NUMBER) + unexpected_linespec_error (parser); - /* If the filename was quoted, we must re-check the quotation. */ + /* Shortcut: If the next token is not LSTOKEN_COLON, we know that + this token cannot represent a filename. */ + token = linespec_lexer_peek_token (parser); - if (end_quote == first_half && *end_quote!= '\0') + if (token.type == LSTOKEN_COLON) { - is_quoted = (**argptr - && strchr (get_gdb_completer_quote_characters (), - **argptr) != NULL); - if (is_quoted) - end_quote = skip_quoted (*argptr); - } + char *user_filename; - /* Check whether arg is all digits (and sign). */ + /* Get the current token again and extract the filename. */ + token = linespec_lexer_lex_one (parser); + user_filename = copy_token_string (token); - q = *argptr; - if (*q == '-' || *q == '+') - q++; - while (*q >= '0' && *q <= '9') - q++; + /* Check if the input is a filename. */ + TRY_CATCH (file_exception, RETURN_MASK_ERROR) + { + PARSER_RESULT (parser)->file_symtabs + = symtabs_from_filename (user_filename); + } - if (q != *argptr && (*q == 0 || *q == ' ' || *q == '\t' || *q == ',') - && function_symbols == NULL) - { - struct symtabs_and_lines values; + if (file_exception.reason >= 0) + { + /* Symtabs were found for the file. Record the filename. */ + PARSER_RESULT (parser)->source_filename = user_filename; - /* We found a token consisting of all digits -- at least one digit. */ - values = decode_all_digits (self, argptr, q); - do_cleanups (cleanup); - return values; - } + /* Get the next token. */ + token = linespec_lexer_consume_token (parser); - /* Arg token is not digits => try it as a variable name - Find the next token (everything up to end or next whitespace). */ + /* This is LSTOKEN_COLON; consume it. */ + linespec_lexer_consume_token (parser); + } + else + { + /* No symtabs found -- discard user_filename. */ + xfree (user_filename); - if (**argptr == '$') /* May be a convenience variable. */ - /* One or two $ chars possible. */ - p = skip_quoted (*argptr + (((*argptr)[1] == '$') ? 2 : 1)); - else if (is_quoted || is_squote_enclosed) - { - p = end_quote; - if (p[-1] != '\'') - error (_("Unmatched single quote.")); + /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */ + VEC_safe_push (symtab_p, PARSER_RESULT (parser)->file_symtabs, NULL); + } } - else if (is_objc_method) + /* If the next token is not EOI, KEYWORD, or COMMA, issue an error. */ + else if (token.type != LSTOKEN_EOI && token.type != LSTOKEN_KEYWORD + && token.type != LSTOKEN_COMMA) { - /* allow word separators in method names for Obj-C. */ - p = skip_quoted_chars (*argptr, NULL, ""); + /* TOKEN is the _next_ token, not the one currently in the parser. + Consuming the token will give the correct error message. */ + linespec_lexer_consume_token (parser); + unexpected_linespec_error (parser); } else { - p = skip_quoted (*argptr); - } - - /* Keep any important naming information. */ - p = keep_name_info (p, p == saved_arg || is_linespec_boundary (p[-1])); - - copy = (char *) alloca (p - *argptr + 1); - memcpy (copy, *argptr, p - *argptr); - copy[p - *argptr] = '\0'; - if (p != *argptr - && copy[0] - && copy[0] == copy[p - *argptr - 1] - && strchr (get_gdb_completer_quote_characters (), copy[0]) != NULL) - { - copy[p - *argptr - 1] = '\0'; - copy++; + /* A NULL entry means to use GLOBAL_DEFAULT_SYMTAB. */ + VEC_safe_push (symtab_p, PARSER_RESULT (parser)->file_symtabs, NULL); } - else if (is_quoted || is_squote_enclosed) - copy[p - *argptr - 1] = '\0'; - - *argptr = skip_spaces (p); - /* If it starts with $: may be a legitimate variable or routine name - (e.g. HP-UX millicode routines such as $$dyncall), or it may - be history value, or it may be a convenience variable. */ + /* Parse the rest of the linespec. */ + linespec_parse_basic (parser); - if (*copy == '$' && function_symbols == NULL) + if (PARSER_RESULT (parser)->function_symbols == NULL + && PARSER_RESULT (parser)->labels.label_symbols == NULL + && PARSER_RESULT (parser)->line_offset.sign == LINE_OFFSET_UNKNOWN + && PARSER_RESULT (parser)->minimal_symbols == NULL) { - struct symtabs_and_lines values; - - values = decode_dollar (self, copy); - do_cleanups (cleanup); - return values; - } - - /* Try the token as a label, but only if no file was specified, - because we can only really find labels in the current scope. */ + /* The linespec didn't parse. Re-throw the file exception if + there was one. */ + if (file_exception.reason < 0) + throw_exception (file_exception); - if (VEC_length (symtab_p, self->file_symtabs) == 1 - && VEC_index (symtab_p, self->file_symtabs, 0) == NULL) - { - struct symtabs_and_lines label_result; - if (decode_label (self, function_symbols, copy, &label_result)) - { - do_cleanups (cleanup); - return label_result; - } + /* Otherwise, the symbol is not found. */ + symbol_not_found_error (PARSER_RESULT (parser)->function_name, + PARSER_RESULT (parser)->source_filename); } - if (function_symbols) - throw_exception (file_exception); + convert_to_sals: - /* Look up that token as a variable. - If file specified, use that file's per-file block to start with. */ + /* Get the last token and record how much of the input was parsed, + if necessary. */ + token = linespec_lexer_lex_one (parser); + if (token.type != LSTOKEN_EOI && token.type != LSTOKEN_KEYWORD) + PARSER_STREAM (parser) = LS_TOKEN_STOKEN (token).ptr; - { - struct symtabs_and_lines values; + /* Convert the data in PARSER_RESULT to SALs. */ + values = convert_linespec_to_sals (PARSER_STATE (parser), + PARSER_RESULT (parser)); - values = decode_variable (self, copy); - do_cleanups (cleanup); - return values; - } + return values; } + /* A constructor for linespec_state. */ static void linespec_state_constructor (struct linespec_state *self, - int flags, + int flags, const struct language_defn *language, struct symtab *default_symtab, int default_line, struct linespec_result *canonical) { memset (self, 0, sizeof (*self)); + self->language = language; self->funfirstline = (flags & DECODE_LINE_FUNFIRSTLINE) ? 1 : 0; self->list_mode = (flags & DECODE_LINE_LIST_MODE) ? 1 : 0; self->default_symtab = default_symtab; @@ -1237,19 +2191,64 @@ linespec_state_constructor (struct linespec_state *self, xfree, xcalloc, xfree); } +/* Initialize a new linespec parser. */ + +static void +linespec_parser_new (linespec_parser *parser, + int flags, const struct language_defn *language, + struct symtab *default_symtab, + int default_line, + struct linespec_result *canonical) +{ + parser->lexer.current.type = LSTOKEN_CONSUMED; + memset (PARSER_RESULT (parser), 0, sizeof (struct linespec)); + PARSER_RESULT (parser)->line_offset.sign = LINE_OFFSET_UNKNOWN; + linespec_state_constructor (PARSER_STATE (parser), flags, language, + default_symtab, default_line, canonical); +} + /* A destructor for linespec_state. */ static void -linespec_state_destructor (void *arg) +linespec_state_destructor (struct linespec_state *self) { - struct linespec_state *self = arg; - - xfree (self->user_filename); - xfree (self->user_function); - VEC_free (symtab_p, self->file_symtabs); htab_delete (self->addr_set); } +/* Delete a linespec parser. */ + +static void +linespec_parser_delete (void *arg) +{ + linespec_parser *parser = (linespec_parser *) arg; + + if (PARSER_RESULT (parser)->expression) + xfree (PARSER_RESULT (parser)->expression); + if (PARSER_RESULT (parser)->source_filename) + xfree (PARSER_RESULT (parser)->source_filename); + if (PARSER_RESULT (parser)->label_name) + xfree (PARSER_RESULT (parser)->label_name); + if (PARSER_RESULT (parser)->function_name) + xfree (PARSER_RESULT (parser)->function_name); + + if (PARSER_RESULT (parser)->file_symtabs != NULL) + VEC_free (symtab_p, PARSER_RESULT (parser)->file_symtabs); + + if (PARSER_RESULT (parser)->function_symbols != NULL) + VEC_free (symbolp, PARSER_RESULT (parser)->function_symbols); + + if (PARSER_RESULT (parser)->minimal_symbols != NULL) + VEC_free (minsym_and_objfile_d, PARSER_RESULT (parser)->minimal_symbols); + + if (PARSER_RESULT (parser)->labels.label_symbols != NULL) + VEC_free (symbolp, PARSER_RESULT (parser)->labels.label_symbols); + + if (PARSER_RESULT (parser)->labels.function_symbols != NULL) + VEC_free (symbolp, PARSER_RESULT (parser)->labels.function_symbols); + + linespec_state_destructor (PARSER_STATE (parser)); +} + /* See linespec.h. */ void @@ -1260,10 +2259,11 @@ decode_line_full (char **argptr, int flags, const char *filter) { struct symtabs_and_lines result; - struct linespec_state state; struct cleanup *cleanups; char *arg_start = *argptr; VEC (const_char_ptr) *filters = NULL; + linespec_parser parser; + struct linespec_state *state; gdb_assert (canonical != NULL); /* The filter only makes sense for 'all'. */ @@ -1274,12 +2274,13 @@ decode_line_full (char **argptr, int flags, || select_mode == multiple_symbols_cancel); gdb_assert ((flags & DECODE_LINE_LIST_MODE) == 0); - linespec_state_constructor (&state, flags, - default_symtab, default_line, canonical); - cleanups = make_cleanup (linespec_state_destructor, &state); + linespec_parser_new (&parser, flags, current_language, default_symtab, + default_line, canonical); + cleanups = make_cleanup (linespec_parser_delete, &parser); save_current_program_space (); - result = decode_line_internal (&state, argptr); + result = parse_linespec (&parser, argptr); + state = PARSER_STATE (&parser); gdb_assert (result.nelts == 1 || canonical->pre_expanded); gdb_assert (canonical->addr_string != NULL); @@ -1290,15 +2291,15 @@ decode_line_full (char **argptr, int flags, { int i; - if (state.canonical_names == NULL) - state.canonical_names = xcalloc (result.nelts, sizeof (char *)); - make_cleanup (xfree, state.canonical_names); + if (state->canonical_names == NULL) + state->canonical_names = xcalloc (result.nelts, sizeof (char *)); + make_cleanup (xfree, state->canonical_names); for (i = 0; i < result.nelts; ++i) { - if (state.canonical_names[i] == NULL) - state.canonical_names[i] = savestring (arg_start, - *argptr - arg_start); - make_cleanup (xfree, state.canonical_names[i]); + if (state->canonical_names[i] == NULL) + state->canonical_names[i] = savestring (arg_start, + *argptr - arg_start); + make_cleanup (xfree, state->canonical_names[i]); } } @@ -1316,13 +2317,13 @@ decode_line_full (char **argptr, int flags, { make_cleanup (VEC_cleanup (const_char_ptr), &filters); VEC_safe_push (const_char_ptr, filters, filter); - filter_results (&state, &result, filters); + filter_results (state, &result, filters); } else - convert_results_to_lsals (&state, &result); + convert_results_to_lsals (state, &result); } else - decode_line_2 (&state, &result, select_mode); + decode_line_2 (state, &result, select_mode); do_cleanups (cleanups); } @@ -1333,15 +2334,16 @@ decode_line_1 (char **argptr, int flags, int default_line) { struct symtabs_and_lines result; - struct linespec_state state; + linespec_parser parser; struct cleanup *cleanups; - linespec_state_constructor (&state, flags, - default_symtab, default_line, NULL); - cleanups = make_cleanup (linespec_state_destructor, &state); + linespec_parser_new (&parser, flags, current_language, default_symtab, + default_line, NULL); + cleanups = make_cleanup (linespec_parser_delete, &parser); save_current_program_space (); - result = decode_line_internal (&state, argptr); + result = parse_linespec (&parser, argptr); + do_cleanups (cleanups); return result; } @@ -1369,178 +2371,20 @@ initialize_defaults (struct symtab **default_symtab, int *default_line) \f -/* Decode arg of the form *PC. */ +/* Evaluate the expression pointed to by EXP_PTR into a CORE_ADDR, + advancing EXP_PTR past any parsed text. */ -static struct symtabs_and_lines -decode_indirect (struct linespec_state *self, char **argptr) +static CORE_ADDR +linespec_expression_to_pc (char **exp_ptr) { - struct symtabs_and_lines values; - CORE_ADDR pc; - char *initial = *argptr; - if (current_program_space->executing_startup) /* The error message doesn't really matter, because this case should only hit during breakpoint reset. */ throw_error (NOT_FOUND_ERROR, _("cannot evaluate expressions while " "program space is in startup")); - (*argptr)++; - pc = value_as_address (parse_to_comma_and_eval (argptr)); - - values.sals = (struct symtab_and_line *) - xmalloc (sizeof (struct symtab_and_line)); - - values.nelts = 1; - values.sals[0] = find_pc_line (pc, 0); - values.sals[0].pc = pc; - values.sals[0].section = find_pc_overlay (pc); - values.sals[0].explicit_pc = 1; - - if (self->canonical) - self->canonical->addr_string = savestring (initial, *argptr - initial); - - return values; -} - -\f - -/* Locate the first half of the linespec, ending in a colon, period, - or whitespace. (More or less.) Also, check to see if *ARGPTR is - enclosed in double quotes; if so, set is_quote_enclosed, advance - ARGPTR past that and zero out the trailing double quote. - If ARGPTR is just a simple name like "main", p will point to "" - at the end. */ - -static char * -locate_first_half (char **argptr, int *is_quote_enclosed) -{ - char *ii; - char *p, *p1; - int has_comma; - - /* Check if the linespec starts with an Ada operator (such as "+", - or ">", for instance). */ - p = *argptr; - if (p[0] == '"' - && current_language->la_language == language_ada) - { - const struct ada_opname_map *op; - - for (op = ada_opname_table; op->encoded != NULL; op++) - if (strncmp (op->decoded, p, strlen (op->decoded)) == 0) - break; - if (op->encoded != NULL) - { - *is_quote_enclosed = 0; - return p + strlen (op->decoded); - } - } - - /* Maybe we were called with a line range FILENAME:LINENUM,FILENAME:LINENUM - and we must isolate the first half. Outer layers will call again later - for the second half. - - Don't count commas that appear in argument lists of overloaded - functions, or in quoted strings. It's stupid to go to this much - trouble when the rest of the function is such an obvious roach hotel. */ - ii = find_toplevel_char (*argptr, ','); - has_comma = (ii != 0); - - /* Temporarily zap out second half to not confuse the code below. - This is undone below. Do not change ii!! */ - if (has_comma) - { - *ii = '\0'; - } - - /* Maybe arg is FILE : LINENUM or FILE : FUNCTION. May also be - CLASS::MEMBER, or NAMESPACE::NAME. Look for ':', but ignore - inside of <>. */ - - p = *argptr; - if (p[0] == '"') - { - *is_quote_enclosed = 1; - (*argptr)++; - p++; - } - else - { - *is_quote_enclosed = 0; - if (strchr (get_gdb_completer_quote_characters (), *p)) - { - ++(*argptr); - ++p; - } - } - - - /* Check for a drive letter in the filename. This is done on all hosts - to capture cross-compilation environments. On Unixen, directory - separators are illegal in filenames, so if the user enters "e:/foo.c", - he is referring to a directory named "e:" and a source file named - "foo.c", and we still want to keep these two pieces together. */ - if (isalpha (p[0]) && p[1] == ':' && IS_DIR_SEPARATOR (p[2])) - p += 3; - - for (; *p; p++) - { - if (p[0] == '<') - { - char *temp_end = find_template_name_end (p); - - if (!temp_end) - error (_("malformed template specification in command")); - p = temp_end; - } - - if (p[0] == '(') - p = find_method_overload_end (p); - - /* Check for a colon and a plus or minus and a [ (which - indicates an Objective-C method). */ - if (is_objc_method_format (p)) - { - break; - } - /* Check for the end of the first half of the linespec. End of - line, a tab, a colon or a space. But if enclosed in double - quotes we do not break on enclosed spaces. */ - if (!*p - || p[0] == '\t' - || (p[0] == ':') - || ((p[0] == ' ') && !*is_quote_enclosed)) - break; - if (p[0] == '.' && strchr (p, ':') == NULL) - { - /* Java qualified method. Find the *last* '.', since the - others are package qualifiers. Stop at any open parenthesis - which might provide overload information. */ - for (p1 = p; *p1 && *p1 != '('; p1++) - { - if (*p1 == '.') - p = p1; - } - break; - } - } - p = skip_spaces (p); - - /* If the closing double quote was left at the end, remove it. */ - if (*is_quote_enclosed) - { - char *closing_quote = strchr (p - 1, '"'); - - if (closing_quote && closing_quote[1] == '\0') - *closing_quote = '\0'; - } - - /* Now that we've safely parsed the first half, put back ',' so - outer layers can see it. */ - if (has_comma) - *ii = ','; - - return p; + (*exp_ptr)++; + return value_as_address (parse_to_comma_and_eval (exp_ptr)); } \f @@ -1553,28 +2397,34 @@ locate_first_half (char **argptr, int *is_quote_enclosed) the existing C++ code to let the user choose one. */ static struct symtabs_and_lines -decode_objc (struct linespec_state *self, char **argptr) +decode_objc (struct linespec_state *self, linespec_p ls, char **argptr) { struct collect_info info; VEC (const_char_ptr) *symbol_names = NULL; + struct symtabs_and_lines values; char *new_argptr; struct cleanup *cleanup = make_cleanup (VEC_cleanup (const_char_ptr), &symbol_names); info.state = self; - info.result.sals = NULL; - info.result.nelts = 0; + info.file_symtabs = NULL; + VEC_safe_push (symtab_p, info.file_symtabs, NULL); + info.result.symbols = NULL; + info.result.minimal_symbols = NULL; + values.nelts = 0; + values.sals = NULL; new_argptr = find_imps (*argptr, &symbol_names); if (VEC_empty (const_char_ptr, symbol_names)) { do_cleanups (cleanup); - return info.result; + return values; } add_all_symbol_names_from_pspace (&info, NULL, symbol_names); - if (info.result.nelts > 0) + if (!VEC_empty (symbolp, info.result.symbols) + || !VEC_empty (minsym_and_objfile_d, info.result.minimal_symbols)) { char *saved_arg; @@ -1582,12 +2432,16 @@ decode_objc (struct linespec_state *self, char **argptr) memcpy (saved_arg, *argptr, new_argptr - *argptr); saved_arg[new_argptr - *argptr] = '\0'; + ls->function_symbols = info.result.symbols; + ls->minimal_symbols = info.result.minimal_symbols; + values = convert_linespec_to_sals (self, ls); + if (self->canonical) { self->canonical->pre_expanded = 1; - if (self->user_filename) + if (ls->source_filename) self->canonical->addr_string - = xstrprintf ("%s:%s", self->user_filename, saved_arg); + = xstrprintf ("%s:%s", ls->source_filename, saved_arg); else self->canonical->addr_string = xstrdup (saved_arg); } @@ -1596,241 +2450,8 @@ decode_objc (struct linespec_state *self, char **argptr) *argptr = new_argptr; do_cleanups (cleanup); - return info.result; -} - -/* This handles C++ and Java compound data structures. P should point - at the first component separator, i.e. double-colon or period. As - an example, on entrance to this function we could have ARGPTR - pointing to "AAA::inA::fun" and P pointing to "::inA::fun". */ - -static struct symtabs_and_lines -decode_compound (struct linespec_state *self, - char **argptr, char *the_real_saved_arg, char *p) -{ - struct symtabs_and_lines values; - char *p2; - char *saved_arg2 = *argptr; - char *temp_end; - struct symbol *sym; - char *copy; - VEC (symbolp) *sym_classes; - char *saved_arg, *class_name; - struct cleanup *cleanup = make_cleanup (null_cleanup, NULL); - - /* If the user specified any completer quote characters in the input, - strip them. They are superfluous. */ - saved_arg = alloca (strlen (the_real_saved_arg) + 1); - { - char *dst = saved_arg; - char *src = the_real_saved_arg; - char *quotes = get_gdb_completer_quote_characters (); - while (*src != '\0') - { - if (strchr (quotes, *src) == NULL) - *dst++ = *src; - ++src; - } - *dst = '\0'; - } - - /* First check for "global" namespace specification, of the form - "::foo". If found, skip over the colons and jump to normal - symbol processing. I.e. the whole line specification starts with - "::" (note the condition that *argptr == p). */ - if (p[0] == ':' - && ((*argptr == p) || (p[-1] == ' ') || (p[-1] == '\t'))) - saved_arg2 += 2; - - /* Given our example "AAA::inA::fun", we have two cases to consider: - - 1) AAA::inA is the name of a class. In that case, presumably it - has a method called "fun"; we then look up that method using - find_method. - - 2) AAA::inA isn't the name of a class. In that case, either the - user made a typo, AAA::inA is the name of a namespace, or it is - the name of a minimal symbol. - In this case we just delegate to decode_variable. - - Thus, our first task is to find everything before the last set of - double-colons and figure out if it's the name of a class. So we - first loop through all of the double-colons. */ - - p2 = p; /* Save for restart. */ - - /* This is very messy. Following the example above we have now the - following pointers: - p -> "::inA::fun" - argptr -> "AAA::inA::fun - saved_arg -> "AAA::inA::fun - saved_arg2 -> "AAA::inA::fun - p2 -> "::inA::fun". */ - - /* In the loop below, with these strings, we'll make 2 passes, each - is marked in comments. */ - - while (1) - { - static char *break_characters = " \t("; - - /* Move pointer up to next possible class/namespace token. */ - p = p2 + 1; /* Restart with old value +1. */ - - /* PASS1: at this point p2->"::inA::fun", so p->":inA::fun", - i.e. if there is a double-colon, p will now point to the - second colon. */ - /* PASS2: p2->"::fun", p->":fun" */ - - /* Move pointer ahead to next double-colon. */ - while (*p - && strchr (break_characters, *p) == NULL - && strchr (get_gdb_completer_quote_characters (), *p) == NULL) - { - if (current_language->la_language == language_cplus) - p += cp_validate_operator (p); - - if (p[0] == '<') - { - temp_end = find_template_name_end (p); - if (!temp_end) - error (_("malformed template specification in command")); - p = temp_end; - } - /* Note that, since, at the start of this loop, p would be - pointing to the second colon in a double-colon, we only - satisfy the condition below if there is another - double-colon to the right (after). I.e. there is another - component that can be a class or a namespace. I.e, if at - the beginning of this loop (PASS1), we had - p->":inA::fun", we'll trigger this when p has been - advanced to point to "::fun". */ - /* PASS2: we will not trigger this. */ - else if ((p[0] == ':') && (p[1] == ':')) - break; /* Found double-colon. */ - else - { - /* PASS2: We'll keep getting here, until P points to one of the - break characters, at which point we exit this loop. */ - if (*p) - { - if (p[1] == '(' - && strncmp (&p[1], CP_ANONYMOUS_NAMESPACE_STR, - CP_ANONYMOUS_NAMESPACE_LEN) == 0) - p += CP_ANONYMOUS_NAMESPACE_LEN; - else if (strchr (break_characters, *p) == NULL) - ++p; - } - } - } - - if (*p != ':') - break; /* Out of the while (1). This would happen - for instance if we have looked up - unsuccessfully all the components of the - string, and p->""(PASS2). */ - - /* We get here if p points to one of the break characters or "" (i.e., - string ended). */ - /* Save restart for next time around. */ - p2 = p; - /* Restore argptr as it was on entry to this function. */ - *argptr = saved_arg2; - /* PASS1: at this point p->"::fun" argptr->"AAA::inA::fun", - p2->"::fun". */ - - /* All ready for next pass through the loop. */ - } /* while (1) */ - - - /* Start of lookup in the symbol tables. */ - - /* Lookup in the symbol table the substring between argptr and - p. Note, this call changes the value of argptr. */ - /* Before the call, argptr->"AAA::inA::fun", - p->"", p2->"::fun". After the call: argptr->"fun", p, p2 - unchanged. */ - sym_classes = lookup_prefix_sym (argptr, p2, self->file_symtabs, - &class_name); - make_cleanup (VEC_cleanup (symbolp), &sym_classes); - make_cleanup (xfree, class_name); - - /* If a class has been found, then we're in case 1 above. So we - look up "fun" as a method of those classes. */ - if (!VEC_empty (symbolp, sym_classes)) - { - /* Arg token is not digits => try it as a function name. - Find the next token (everything up to end or next - blank). */ - if (**argptr - && strchr (get_gdb_completer_quote_characters (), - **argptr) != NULL) - { - p = skip_quoted (*argptr); - *argptr = *argptr + 1; - } - else - { - /* At this point argptr->"fun". */ - char *a; - - p = *argptr; - while (*p && *p != ' ' && *p != '\t' && *p != ',' && *p != ':' - && *p != '(') - p++; - /* At this point p->"". String ended. */ - /* Nope, C++ operators could have spaces in them - ("foo::operator <" or "foo::operator delete []"). - I apologize, this is a bit hacky... */ - if (current_language->la_language == language_cplus - && *p == ' ' && p - 8 - *argptr + 1 > 0) - { - /* The above loop has already swallowed "operator". */ - p += cp_validate_operator (p - 8) - 8; - } - - /* Keep any important naming information. */ - p = keep_name_info (p, 1); - } - - /* Allocate our own copy of the substring between argptr and - p. */ - copy = (char *) alloca (p - *argptr + 1); - memcpy (copy, *argptr, p - *argptr); - copy[p - *argptr] = '\0'; - if (p != *argptr - && copy[p - *argptr - 1] - && strchr (get_gdb_completer_quote_characters (), - copy[p - *argptr - 1]) != NULL) - copy[p - *argptr - 1] = '\0'; - - /* At this point copy->"fun", p->"". */ - - /* No line number may be specified. */ - *argptr = skip_spaces (p); - /* At this point arptr->"". */ - - /* Look for copy as a method of sym_class. */ - /* At this point copy->"fun", sym_class is "AAA:inA", - saved_arg->"AAA::inA::fun". This concludes the scanning of - the string for possible components matches. If we find it - here, we return. If not, and we are at the and of the string, - we'll lookup the whole string in the symbol tables. */ - - values = find_method (self, saved_arg, copy, class_name, sym_classes); - - do_cleanups (cleanup); - return values; - } /* End if symbol found. */ - - - /* We couldn't find a class, so we're in case 2 above. We check the - entire name as a symbol instead. The simplest way to do this is - to just throw an exception and let our caller fall through to - decode_variable. */ - - throw_error (NOT_FOUND_ERROR, _("see caller, this text doesn't matter")); + return values; } /* An instance of this type is used when collecting prefix symbols for @@ -1876,46 +2497,20 @@ collect_one_symbol (struct symbol *sym, void *d) return 1; /* Continue iterating. */ } -/* Return the symbol corresponding to the substring of *ARGPTR ending - at P, allowing whitespace. Also, advance *ARGPTR past the symbol - name in question, the compound object separator ("::" or "."), and - whitespace. Note that *ARGPTR is changed whether or not the - this call finds anything (i.e we return NULL). As an - example, say ARGPTR is "AAA::inA::fun" and P is "::inA::fun". */ +/* Return any symbols corresponding to CLASS_NAME in FILE_SYMTABS. */ static VEC (symbolp) * -lookup_prefix_sym (char **argptr, char *p, VEC (symtab_p) *file_symtabs, - char **class_name) +lookup_prefix_sym (struct linespec_state *state, VEC (symtab_p) *file_symtabs, + const char *class_name) { - char *p1; - char *copy; int ix; struct symtab *elt; struct decode_compound_collector collector; struct cleanup *outer; struct cleanup *cleanup; - struct block *search_block; - - /* Extract the class name. */ - p1 = p; - while (p != *argptr && p[-1] == ' ') - --p; - copy = (char *) xmalloc (p - *argptr + 1); - memcpy (copy, *argptr, p - *argptr); - copy[p - *argptr] = 0; - *class_name = copy; - outer = make_cleanup (xfree, copy); - - /* Discard the class name from the argptr. */ - p = p1 + (p1[0] == ':' ? 2 : 1); - p = skip_spaces (p); - *argptr = p; - - /* At this point p1->"::inA::fun", p->"inA::fun" copy->"AAA", - argptr->"inA::fun". */ collector.symbols = NULL; - make_cleanup (VEC_cleanup (symbolp), &collector.symbols); + outer = make_cleanup (VEC_cleanup (symbolp), &collector.symbols); collector.unique_syms = htab_create_alloc (1, htab_hash_pointer, htab_eq_pointer, NULL, @@ -1926,10 +2521,10 @@ lookup_prefix_sym (char **argptr, char *p, VEC (symtab_p) *file_symtabs, { if (elt == NULL) { - iterate_over_all_matching_symtabs (copy, STRUCT_DOMAIN, + iterate_over_all_matching_symtabs (state, class_name, STRUCT_DOMAIN, collect_one_symbol, &collector, NULL, 0); - iterate_over_all_matching_symtabs (copy, VAR_DOMAIN, + iterate_over_all_matching_symtabs (state, class_name, VAR_DOMAIN, collect_one_symbol, &collector, NULL, 0); } @@ -1942,9 +2537,9 @@ lookup_prefix_sym (char **argptr, char *p, VEC (symtab_p) *file_symtabs, gdb_assert (!SYMTAB_PSPACE (elt)->executing_startup); set_current_program_space (SYMTAB_PSPACE (elt)); search_block = get_search_block (elt); - LA_ITERATE_OVER_SYMBOLS (search_block, copy, STRUCT_DOMAIN, + LA_ITERATE_OVER_SYMBOLS (search_block, class_name, STRUCT_DOMAIN, collect_one_symbol, &collector); - LA_ITERATE_OVER_SYMBOLS (search_block, copy, VAR_DOMAIN, + LA_ITERATE_OVER_SYMBOLS (search_block, class_name, VAR_DOMAIN, collect_one_symbol, &collector); } } @@ -1954,19 +2549,47 @@ lookup_prefix_sym (char **argptr, char *p, VEC (symtab_p) *file_symtabs, return collector.symbols; } -/* A qsort comparison function for symbols. The resulting order does - not actually matter; we just need to be able to sort them so that - symbols with the same program space end up next to each other. */ +/* A qsort comparison function for symbols. The resulting order does + not actually matter; we just need to be able to sort them so that + symbols with the same program space end up next to each other. */ + +static int +compare_symbols (const void *a, const void *b) +{ + struct symbol * const *sa = a; + struct symbol * const *sb = b; + uintptr_t uia, uib; + + uia = (uintptr_t) SYMTAB_PSPACE (SYMBOL_SYMTAB (*sa)); + uib = (uintptr_t) SYMTAB_PSPACE (SYMBOL_SYMTAB (*sb)); + + if (uia < uib) + return -1; + if (uia > uib) + return 1; + + uia = (uintptr_t) *sa; + uib = (uintptr_t) *sb; + + if (uia < uib) + return -1; + if (uia > uib) + return 1; + + return 0; +} + +/* Like compare_symbols but for minimal symbols. */ static int -compare_symbols (const void *a, const void *b) +compare_msymbols (const void *a, const void *b) { - struct symbol * const *sa = a; - struct symbol * const *sb = b; + struct minimal_symbol * const *sa = a; + struct minimal_symbol * const *sb = b; uintptr_t uia, uib; - uia = (uintptr_t) SYMTAB_PSPACE (SYMBOL_SYMTAB (*sa)); - uib = (uintptr_t) SYMTAB_PSPACE (SYMBOL_SYMTAB (*sb)); + uia = (uintptr_t) SYMBOL_OBJ_SECTION (*sa)->objfile->pspace; + uib = (uintptr_t) SYMBOL_OBJ_SECTION (*sb)->objfile->pspace; if (uia < uib) return -1; @@ -2031,14 +2654,16 @@ find_superclass_methods (VEC (typep) *superclasses, do_cleanups (cleanup); } -/* This finds the method COPY in the class whose type is given by one - of the symbols in SYM_CLASSES. */ +/* This finds the method METHOD_NAME in the class CLASS_NAME whose type is + given by one of the symbols in SYM_CLASSES. Matches are returned + in SYMBOLS (for debug symbols) and MINSYMS (for minimal symbols). */ -static struct symtabs_and_lines -find_method (struct linespec_state *self, char *saved_arg, - char *copy, const char *class_name, VEC (symbolp) *sym_classes) +static void +find_method (struct linespec_state *self, VEC (symtab_p) *file_symtabs, + const char *class_name, const char *method_name, + VEC (symbolp) *sym_classes, VEC (symbolp) **symbols, + VEC (minsym_and_objfile_d) **minsyms) { - char *canon; struct symbol *sym; struct cleanup *cleanup = make_cleanup (null_cleanup, NULL); int ix; @@ -2048,15 +2673,6 @@ find_method (struct linespec_state *self, char *saved_arg, struct collect_info info; char *name_iter; - /* NAME is typed by the user: it needs to be canonicalized before - searching the symbol tables. */ - canon = cp_canonicalize_string_no_typedefs (copy); - if (canon != NULL) - { - copy = canon; - make_cleanup (xfree, copy); - } - /* Sort symbols so that symbols with the same program space are next to each other. */ qsort (VEC_address (symbolp, sym_classes), @@ -2065,11 +2681,12 @@ find_method (struct linespec_state *self, char *saved_arg, compare_symbols); info.state = self; - info.result.sals = NULL; - info.result.nelts = 0; + info.file_symtabs = file_symtabs; + info.result.symbols = NULL; + info.result.minimal_symbols = NULL; /* Iterate over all the types, looking for the names of existing - methods matching COPY. If we cannot find a direct method in a + methods matching METHOD_NAME. If we cannot find a direct method in a given program space, then we consider inherited methods; this is not ideal (ideal would be to respect C++ hiding rules), but it seems good enough and is what GDB has historically done. We only @@ -2093,7 +2710,7 @@ find_method (struct linespec_state *self, char *saved_arg, pspace = SYMTAB_PSPACE (SYMBOL_SYMTAB (sym)); set_current_program_space (pspace); t = check_typedef (SYMBOL_TYPE (sym)); - find_methods (t, copy, &result_names, &superclass_vec); + find_methods (t, method_name, &result_names, &superclass_vec); /* Handle all items from a single program space at once; and be sure not to miss the last batch. */ @@ -2105,7 +2722,8 @@ find_method (struct linespec_state *self, char *saved_arg, /* If we did not find a direct implementation anywhere in this program space, consider superclasses. */ if (VEC_length (const_char_ptr, result_names) == last_result_len) - find_superclass_methods (superclass_vec, copy, &result_names); + find_superclass_methods (superclass_vec, method_name, + &result_names); /* We have a list of candidate symbol names, so now we iterate over the symbol tables looking for all @@ -2117,31 +2735,18 @@ find_method (struct linespec_state *self, char *saved_arg, } } - if (info.result.nelts > 0) + if (!VEC_empty (symbolp, info.result.symbols) + || !VEC_empty (minsym_and_objfile_d, info.result.minimal_symbols)) { - if (self->canonical) - { - self->canonical->pre_expanded = 1; - if (self->user_filename) - self->canonical->addr_string - = xstrprintf ("%s:%s", self->user_filename, saved_arg); - else - self->canonical->addr_string = xstrdup (saved_arg); - } - + *symbols = info.result.symbols; + *minsyms = info.result.minimal_symbols; do_cleanups (cleanup); - - return info.result; + return; } - if (copy[0] == '~') - cplusplus_error (saved_arg, - "the class `%s' does not have destructor defined\n", - class_name); - else - cplusplus_error (saved_arg, - "the class %s does not have any method named %s\n", - class_name, copy); + /* Throw an NOT_FOUND_ERROR. This will be caught by the caller + and other attempts to locate the symbol will be made. */ + throw_error (NOT_FOUND_ERROR, _("see caller, this text doesn't matter")); } \f @@ -2203,488 +2808,192 @@ collect_symtabs_from_filename (const char *file) return collector.symtabs; } -/* Return all the symtabs associated to the filename given by the - substring of *ARGPTR ending at P, and advance ARGPTR past that - filename. */ - -static VEC (symtab_p) * -symtabs_from_filename (char **argptr, char *p, int is_quote_enclosed, - char **user_filename) -{ - char *p1; - char *copy; - struct cleanup *outer; - VEC (symtab_p) *result; - - p1 = p; - while (p != *argptr && p[-1] == ' ') - --p; - if ((*p == '"') && is_quote_enclosed) - --p; - copy = xmalloc (p - *argptr + 1); - outer = make_cleanup (xfree, copy); - memcpy (copy, *argptr, p - *argptr); - /* It may have the ending quote right after the file name. */ - if ((is_quote_enclosed && copy[p - *argptr - 1] == '"') - || copy[p - *argptr - 1] == '\'') - copy[p - *argptr - 1] = 0; - else - copy[p - *argptr] = 0; - - result = collect_symtabs_from_filename (copy); - - if (VEC_empty (symtab_p, result)) - { - if (!have_full_symbols () && !have_partial_symbols ()) - throw_error (NOT_FOUND_ERROR, - _("No symbol table is loaded. " - "Use the \"file\" command.")); - throw_error (NOT_FOUND_ERROR, _("No source file named %s."), copy); - } - - /* Discard the file name from the arg. */ - if (*p1 == '\0') - *argptr = p1; - else - *argptr = skip_spaces (p1 + 1); - - discard_cleanups (outer); - *user_filename = copy; - return result; -} - -/* A callback used by iterate_over_all_matching_symtabs that collects - symbols for find_function_symbols. */ - -static int -collect_function_symbols (struct symbol *sym, void *arg) -{ - VEC (symbolp) **syms = arg; - - if (SYMBOL_CLASS (sym) == LOC_BLOCK) - VEC_safe_push (symbolp, *syms, sym); - - return 1; /* Continue iterating. */ -} - -/* Look up a function symbol in *ARGPTR. If found, advance *ARGPTR - and return the symbol. If not found, return NULL. */ - -static VEC (symbolp) * -find_function_symbols (char **argptr, char *p, int is_quote_enclosed, - char **user_function) -{ - char *p1; - char *copy; - VEC (symbolp) *result = NULL; - - p1 = p; - while (p != *argptr && p[-1] == ' ') - --p; - if ((*p == '"') && is_quote_enclosed) - --p; - copy = (char *) xmalloc (p - *argptr + 1); - *user_function = copy; - memcpy (copy, *argptr, p - *argptr); - /* It may have the ending quote right after the file name. */ - if ((is_quote_enclosed && copy[p - *argptr - 1] == '"') - || copy[p - *argptr - 1] == '\'') - copy[p - *argptr - 1] = 0; - else - copy[p - *argptr] = 0; - - iterate_over_all_matching_symtabs (copy, VAR_DOMAIN, - collect_function_symbols, &result, NULL, - 0); - - if (VEC_empty (symbolp, result)) - VEC_free (symbolp, result); - else - { - /* Discard the file name from the arg. */ - *argptr = skip_spaces (p1 + 1); - } - - return result; -} - -\f - -/* A helper for decode_all_digits that handles the 'list_mode' case. */ - -static void -decode_digits_list_mode (struct linespec_state *self, - struct symtabs_and_lines *values, - struct symtab_and_line val) -{ - int ix; - struct symtab *elt; - - gdb_assert (self->list_mode); - - for (ix = 0; VEC_iterate (symtab_p, self->file_symtabs, ix, elt); ++ix) - { - /* The logic above should ensure this. */ - gdb_assert (elt != NULL); - - set_current_program_space (SYMTAB_PSPACE (elt)); - - /* Simplistic search just for the list command. */ - val.symtab = find_line_symtab (elt, val.line, NULL, NULL); - if (val.symtab == NULL) - val.symtab = elt; - val.pspace = SYMTAB_PSPACE (elt); - val.pc = 0; - val.explicit_line = 1; - - add_sal_to_sals (self, values, &val, NULL); - } -} - -/* A helper for decode_all_digits that iterates over the symtabs, - adding lines to the VEC. */ - -static void -decode_digits_ordinary (struct linespec_state *self, - int line, - struct symtabs_and_lines *sals, - struct linetable_entry **best_entry) -{ - int ix; - struct symtab *elt; - - for (ix = 0; VEC_iterate (symtab_p, self->file_symtabs, ix, elt); ++ix) - { - int i; - VEC (CORE_ADDR) *pcs; - CORE_ADDR pc; - - /* The logic above should ensure this. */ - gdb_assert (elt != NULL); - - set_current_program_space (SYMTAB_PSPACE (elt)); - - pcs = find_pcs_for_symtab_line (elt, line, best_entry); - for (i = 0; VEC_iterate (CORE_ADDR, pcs, i, pc); ++i) - { - struct symtab_and_line sal; - - init_sal (&sal); - sal.pspace = SYMTAB_PSPACE (elt); - sal.symtab = elt; - sal.line = line; - sal.pc = pc; - add_sal_to_sals_basic (sals, &sal); - } - - VEC_free (CORE_ADDR, pcs); - } -} - -/* This decodes a line where the argument is all digits (possibly - preceded by a sign). Q should point to the end of those digits; - the other arguments are as usual. */ - -static struct symtabs_and_lines -decode_all_digits (struct linespec_state *self, - char **argptr, - char *q) -{ - struct symtabs_and_lines values; - struct symtab_and_line val; - int use_default = 0; - char *saved_arg = *argptr; - - enum sign - { - none, plus, minus - } - sign = none; - - init_sal (&val); - values.sals = NULL; - values.nelts = 0; - - /* This is where we need to make sure that we have good defaults. - We must guarantee that this section of code is never executed - when we are called with just a function name, since - set_default_source_symtab_and_line uses - select_source_symtab that calls us with such an argument. */ - - if (VEC_length (symtab_p, self->file_symtabs) == 1 - && VEC_index (symtab_p, self->file_symtabs, 0) == NULL) - { - set_current_program_space (self->program_space); - - /* Make sure we have at least a default source file. */ - set_default_source_symtab_and_line (); - initialize_defaults (&self->default_symtab, &self->default_line); - VEC_pop (symtab_p, self->file_symtabs); - VEC_free (symtab_p, self->file_symtabs); - self->file_symtabs - = collect_symtabs_from_filename (self->default_symtab->filename); - use_default = 1; - } - - if (**argptr == '+') - sign = plus, (*argptr)++; - else if (**argptr == '-') - sign = minus, (*argptr)++; - val.line = atoi (*argptr); - switch (sign) - { - case plus: - if (q == *argptr) - val.line = 5; - if (use_default) - val.line = self->default_line + val.line; - break; - case minus: - if (q == *argptr) - val.line = 15; - if (use_default) - val.line = self->default_line - val.line; - else - val.line = 1; - break; - case none: - break; /* No need to adjust val.line. */ - } - - *argptr = skip_spaces (q); - - if (self->list_mode) - decode_digits_list_mode (self, &values, val); - else - { - struct linetable_entry *best_entry = NULL; - int *filter; - struct block **blocks; - struct cleanup *cleanup; - struct symtabs_and_lines intermediate_results; - int i, j; - - intermediate_results.sals = NULL; - intermediate_results.nelts = 0; - - decode_digits_ordinary (self, val.line, &intermediate_results, - &best_entry); - if (intermediate_results.nelts == 0 && best_entry != NULL) - decode_digits_ordinary (self, best_entry->line, &intermediate_results, - &best_entry); - - cleanup = make_cleanup (xfree, intermediate_results.sals); - - /* For optimized code, compiler can scatter one source line - accross disjoint ranges of PC values, even when no duplicate - functions or inline functions are involved. For example, - 'for (;;)' inside non-template non-inline non-ctor-or-dtor - function can result in two PC ranges. In this case, we don't - want to set breakpoint on first PC of each range. To filter - such cases, we use containing blocks -- for each PC found - above we see if there are other PCs that are in the same - block. If yes, the other PCs are filtered out. */ - - filter = xmalloc (intermediate_results.nelts * sizeof (int)); - make_cleanup (xfree, filter); - blocks = xmalloc (intermediate_results.nelts * sizeof (struct block *)); - make_cleanup (xfree, blocks); - - for (i = 0; i < intermediate_results.nelts; ++i) - { - set_current_program_space (intermediate_results.sals[i].pspace); - - filter[i] = 1; - blocks[i] = block_for_pc_sect (intermediate_results.sals[i].pc, - intermediate_results.sals[i].section); - } - - for (i = 0; i < intermediate_results.nelts; ++i) - { - if (blocks[i] != NULL) - for (j = i + 1; j < intermediate_results.nelts; ++j) - { - if (blocks[j] == blocks[i]) - { - filter[j] = 0; - break; - } - } - } - - for (i = 0; i < intermediate_results.nelts; ++i) - if (filter[i]) - { - struct symbol *sym = (blocks[i] - ? block_containing_function (blocks[i]) - : NULL); - - if (self->funfirstline) - skip_prologue_sal (&intermediate_results.sals[i]); - /* Make sure the line matches the request, not what was - found. */ - intermediate_results.sals[i].line = val.line; - add_sal_to_sals (self, &values, &intermediate_results.sals[i], - sym ? SYMBOL_NATURAL_NAME (sym) : NULL); - } - - do_cleanups (cleanup); - } - - if (values.nelts == 0) - { - if (self->user_filename) - throw_error (NOT_FOUND_ERROR, _("No line %d in file \"%s\"."), - val.line, self->user_filename); - else - throw_error (NOT_FOUND_ERROR, _("No line %d in the current file."), - val.line); - } +/* Return all the symtabs associated to the FILENAME. */ - if (self->canonical) - { - char *copy = savestring (saved_arg, q - saved_arg); +static VEC (symtab_p) * +symtabs_from_filename (const char *filename) +{ + VEC (symtab_p) *result; + + result = collect_symtabs_from_filename (filename); - self->canonical->pre_expanded = 1; - gdb_assert (self->user_filename || use_default); - self->canonical->addr_string - = xstrprintf ("%s:%s", (self->user_filename - ? self->user_filename - : self->default_symtab->filename), - copy); - xfree (copy); + if (VEC_empty (symtab_p, result)) + { + if (!have_full_symbols () && !have_partial_symbols ()) + throw_error (NOT_FOUND_ERROR, + _("No symbol table is loaded. " + "Use the \"file\" command.")); + throw_error (NOT_FOUND_ERROR, _("No source file named %s."), filename); } - return values; + return result; } -\f - -/* Decode a linespec starting with a dollar sign. */ +/* Look up a function symbol named NAME in symtabs FILE_SYMTABS. Matching + debug symbols are returned in SYMBOLS. Matching minimal symbols are + returned in MINSYMS. */ -static struct symtabs_and_lines -decode_dollar (struct linespec_state *self, char *copy) +static void +find_function_symbols (struct linespec_state *state, + VEC (symtab_p) *file_symtabs, const char *name, + VEC (symbolp) **symbols, + VEC (minsym_and_objfile_d) **minsyms) { - LONGEST valx; - int index = 0; - struct symtabs_and_lines values; - struct symtab_and_line val; - char *p; - struct symbol *sym; - struct minimal_symbol *msymbol; - int ix; - struct symtab *elt; + struct collect_info info; + VEC (const_char_ptr) *symbol_names = NULL; + struct cleanup *cleanup = make_cleanup (VEC_cleanup (const_char_ptr), + &symbol_names); - p = (copy[1] == '$') ? copy + 2 : copy + 1; - while (*p >= '0' && *p <= '9') - p++; - if (!*p) /* Reached end of token without hitting non-digit. */ - { - /* We have a value history reference. */ - struct value *val_history; + info.state = state; + info.result.symbols = NULL; + info.result.minimal_symbols = NULL; + info.file_symtabs = file_symtabs; - sscanf ((copy[1] == '$') ? copy + 2 : copy + 1, "%d", &index); - val_history = access_value_history ((copy[1] == '$') ? -index : index); - if (TYPE_CODE (value_type (val_history)) != TYPE_CODE_INT) - error (_("History values used in line " - "specs must have integer values.")); - valx = value_as_long (val_history); + /* Try NAME as an Objective-C selector. */ + find_imps ((char *) name, &symbol_names); + if (!VEC_empty (const_char_ptr, symbol_names)) + add_all_symbol_names_from_pspace (&info, NULL, symbol_names); + else + add_matching_symbols_to_info (name, &info, NULL); + + do_cleanups (cleanup); + + if (VEC_empty (symbolp, info.result.symbols)) + { + VEC_free (symbolp, info.result.symbols); + *symbols = NULL; } else + *symbols = info.result.symbols; + + if (VEC_empty (minsym_and_objfile_d, info.result.minimal_symbols)) { - /* Not all digits -- may be user variable/function or a - convenience variable. */ + VEC_free (minsym_and_objfile_d, info.result.minimal_symbols); + *minsyms = NULL; + } + else + *minsyms = info.result.minimal_symbols; +} - volatile struct gdb_exception exc; +/* Find all symbols named NAME in FILE_SYMTABS, returning debug symbols + in SYMBOLS and minimal symbols in MINSYMS. */ - /* Avoid "may be used uninitialized" warning. */ - values.sals = NULL; - values.nelts = 0; +void +find_linespec_symbols (struct linespec_state *state, + VEC (symtab_p) *file_symtabs, + const char *name, + VEC (symbolp) **symbols, + VEC (minsym_and_objfile_d) **minsyms) +{ + char *klass, *method, *canon; + const char *lookup_name, *last, *p, *scope_op; + struct cleanup *cleanup; + VEC (symbolp) *classes; + volatile struct gdb_exception except; - TRY_CATCH (exc, RETURN_MASK_ERROR) - { - values = decode_variable (self, copy); - } + cleanup = demangle_for_lookup (name, state->language->la_language, + &lookup_name); + if (state->language->la_language == language_ada) + { + /* In Ada, the symbol lookups are performed using the encoded + name rather than the demangled name. */ + lookup_name = ada_name_for_lookup (name); + make_cleanup (xfree, (void *) lookup_name); + } + + canon = cp_canonicalize_string_no_typedefs (lookup_name); + if (canon != NULL) + { + lookup_name = canon; + cleanup = make_cleanup (xfree, canon); + } - if (exc.reason == 0) - return values; + /* See if we can find a scope operator and break this symbol + name into namespaces${SCOPE_OPERATOR}class_name and method_name. */ + scope_op = "::"; + p = find_toplevel_string (lookup_name, scope_op); + if (p == NULL) + { + /* No C++ scope operator. Try Java. */ + scope_op = "."; + p = find_toplevel_string (lookup_name, scope_op); + } - if (exc.error != NOT_FOUND_ERROR) - throw_exception (exc); + last = NULL; + while (p != NULL) + { + last = p; + p = find_toplevel_string (p + strlen (scope_op), scope_op); + } - /* Not a user variable or function -- must be convenience variable. */ - if (!get_internalvar_integer (lookup_internalvar (copy + 1), &valx)) - error (_("Convenience variables used in line " - "specs must have integer values.")); + /* If no scope operator was found, lookup the name as a symbol. */ + if (last == NULL) + { + find_function_symbols (state, file_symtabs, lookup_name, + symbols, minsyms); + do_cleanups (cleanup); + return; } - init_sal (&val); + /* NAME points to the class name. + LAST points to the method name. */ + klass = xmalloc ((last - lookup_name + 1) * sizeof (char)); + make_cleanup (xfree, klass); + strncpy (klass, lookup_name, last - lookup_name); + klass[last - lookup_name] = '\0'; - values.sals = NULL; - values.nelts = 0; + /* Skip past the scope operator. */ + last += strlen (scope_op); + method = xmalloc ((strlen (last) + 1) * sizeof (char)); + make_cleanup (xfree, method); + strcpy (method, last); - for (ix = 0; VEC_iterate (symtab_p, self->file_symtabs, ix, elt); ++ix) + /* Find a list of classes named KLASS. */ + classes = lookup_prefix_sym (state, file_symtabs, klass); + if (!VEC_empty (symbolp, classes)) { - if (elt == NULL) + /* Now locate a list of suitable methods named METHOD. */ + TRY_CATCH (except, RETURN_MASK_ERROR) { - elt = self->default_symtab; - set_current_program_space (self->program_space); + find_method (state, file_symtabs, klass, method, classes, + symbols, minsyms); } - else - set_current_program_space (SYMTAB_PSPACE (elt)); - - /* Either history value or convenience value from above, in valx. */ - val.symtab = elt; - val.line = valx; - val.pc = 0; - val.pspace = elt ? SYMTAB_PSPACE (elt) : current_program_space; - - add_sal_to_sals (self, &values, &val, NULL); - } - if (self->canonical) - { - self->canonical->pre_expanded = 1; - if (self->user_filename) - self->canonical->addr_string = xstrprintf ("%s:%s", - self->user_filename, copy); - else - self->canonical->addr_string = xstrdup (copy); + /* If successful, we're done. If NOT_FOUND_ERROR + was not thrown, rethrow the exception that we did get. + Otherwise, fall back to looking up the entire name as a symbol. + This can happen with namespace::function. */ + if (except.reason >= 0) + { + do_cleanups (cleanup); + return; + } + else if (except.error != NOT_FOUND_ERROR) + throw_exception (except); } - return values; + /* We couldn't find a class, so we check the entire name as a symbol + instead. */ + find_function_symbols (state, file_symtabs, lookup_name, symbols, minsyms); + do_cleanups (cleanup); } -\f - -/* A helper for decode_line_1 that tries to find a label. The label - is searched for in the current block. - FUNCTION_SYMBOLS is a list of the enclosing functions; or NULL if none - specified. - COPY is the name of the label to find. - CANONICAL is the same as the "canonical" argument to decode_line_1. - RESULT is a pointer to a symtabs_and_lines structure which will be - filled in on success. - This function returns 1 if a label was found, 0 otherwise. */ +/* Return all labels named NAME in FUNCTION_SYMBOLS. Return the + actual function symbol in which the label was found in LABEL_FUNC_RET. */ -static int -decode_label (struct linespec_state *self, - VEC (symbolp) *function_symbols, char *copy, - struct symtabs_and_lines *result) +static VEC (symbolp) * +find_label_symbols (struct linespec_state *self, + VEC (symbolp) *function_symbols, + VEC (symbolp) **label_funcs_ret, const char *name) { - struct symbol *fn_sym; int ix; + struct block *block; + struct symbol *sym; + struct symbol *fn_sym; + VEC (symbolp) *result = NULL; if (function_symbols == NULL) { - struct block *block; - struct symbol *sym; - struct symtab_and_line sal; - struct symtabs_and_lines values; - - values.nelts = 0; - values.sals = NULL; - set_current_program_space (self->program_space); block = get_search_block (NULL); @@ -2693,69 +3002,171 @@ decode_label (struct linespec_state *self, block = BLOCK_SUPERBLOCK (block)) ; if (!block) - return 0; + return NULL; fn_sym = BLOCK_FUNCTION (block); - sym = lookup_symbol (copy, block, LABEL_DOMAIN, 0); - - if (sym == NULL) - return 0; + sym = lookup_symbol (name, block, LABEL_DOMAIN, 0); - symbol_to_sal (&sal, self->funfirstline, sym); - add_sal_to_sals (self, &values, &sal, - SYMBOL_NATURAL_NAME (fn_sym)); - - if (self->canonical) + if (sym != NULL) + { + VEC_safe_push (symbolp, result, sym); + VEC_safe_push (symbolp, *label_funcs_ret, fn_sym); + } + } + else + { + for (ix = 0; + VEC_iterate (symbolp, function_symbols, ix, fn_sym); ++ix) { - self->canonical->special_display = 1; - self->canonical->addr_string - = xstrprintf ("%s:%s", SYMBOL_NATURAL_NAME (fn_sym), - copy); + set_current_program_space (SYMTAB_PSPACE (SYMBOL_SYMTAB (fn_sym))); + block = SYMBOL_BLOCK_VALUE (fn_sym); + sym = lookup_symbol (name, block, LABEL_DOMAIN, 0); + + if (sym != NULL) + { + VEC_safe_push (symbolp, result, sym); + VEC_safe_push (symbolp, *label_funcs_ret, fn_sym); + } } + } - *result = values; + return result; +} - return 1; +\f + +/* A helper for create_sals_line_offset that handles the 'list_mode' case. */ + +static void +decode_digits_list_mode (struct linespec_state *self, + linespec_p ls, + struct symtabs_and_lines *values, + struct symtab_and_line val) +{ + int ix; + struct symtab *elt; + + gdb_assert (self->list_mode); + + for (ix = 0; VEC_iterate (symtab_p, ls->file_symtabs, ix, elt); + ++ix) + { + /* The logic above should ensure this. */ + gdb_assert (elt != NULL); + + set_current_program_space (SYMTAB_PSPACE (elt)); + + /* Simplistic search just for the list command. */ + val.symtab = find_line_symtab (elt, val.line, NULL, NULL); + if (val.symtab == NULL) + val.symtab = elt; + val.pspace = SYMTAB_PSPACE (elt); + val.pc = 0; + val.explicit_line = 1; + + add_sal_to_sals (self, values, &val, NULL); } +} - result->sals = NULL; - result->nelts = 0; +/* A helper for create_sals_line_offset that iterates over the symtabs, + adding lines to the VEC. */ + +static void +decode_digits_ordinary (struct linespec_state *self, + linespec_p ls, + int line, + struct symtabs_and_lines *sals, + struct linetable_entry **best_entry) +{ + int ix; + struct symtab *elt; - for (ix = 0; VEC_iterate (symbolp, function_symbols, ix, fn_sym); ++ix) + for (ix = 0; VEC_iterate (symtab_p, ls->file_symtabs, ix, elt); ++ix) { - struct block *block; - struct symbol *sym; + int i; + VEC (CORE_ADDR) *pcs; + CORE_ADDR pc; + + /* The logic above should ensure this. */ + gdb_assert (elt != NULL); - set_current_program_space (SYMTAB_PSPACE (SYMBOL_SYMTAB (fn_sym))); - block = SYMBOL_BLOCK_VALUE (fn_sym); - sym = lookup_symbol (copy, block, LABEL_DOMAIN, 0); + set_current_program_space (SYMTAB_PSPACE (elt)); - if (sym != NULL) + pcs = find_pcs_for_symtab_line (elt, line, best_entry); + for (i = 0; VEC_iterate (CORE_ADDR, pcs, i, pc); ++i) { struct symtab_and_line sal; - char *symname; - - symbol_to_sal (&sal, self->funfirstline, sym); - symname = xstrprintf ("%s:%s", - SYMBOL_NATURAL_NAME (fn_sym), - SYMBOL_NATURAL_NAME (sym)); - add_sal_to_sals (self, result, &sal, symname); - xfree (symname); + + init_sal (&sal); + sal.pspace = SYMTAB_PSPACE (elt); + sal.symtab = elt; + sal.line = line; + sal.pc = pc; + add_sal_to_sals_basic (sals, &sal); } + + VEC_free (CORE_ADDR, pcs); } +} + +\f + +/* Return the line offset represented by VARIABLE. */ + +static struct line_offset +linespec_parse_variable (struct linespec_state *self, const char *variable) +{ + int index = 0; + const char *p; + struct line_offset offset = {0, LINE_OFFSET_NONE}; - if (self->canonical && result->nelts > 0) + p = (variable[1] == '$') ? variable + 2 : variable + 1; + if (*p == '$') + ++p; + while (*p >= '0' && *p <= '9') + ++p; + if (!*p) /* Reached end of token without hitting non-digit. */ { - self->canonical->pre_expanded = 1; - self->canonical->special_display = 1; + /* We have a value history reference. */ + struct value *val_history; - gdb_assert (self->user_function); - self->canonical->addr_string - = xstrprintf ("%s:%s", self->user_function, copy); + sscanf ((variable[1] == '$') ? variable + 2 : variable + 1, "%d", &index); + val_history + = access_value_history ((variable[1] == '$') ? -index : index); + if (TYPE_CODE (value_type (val_history)) != TYPE_CODE_INT) + error (_("History values used in line " + "specs must have integer values.")); + offset.offset = value_as_long (val_history); + } + else + { + /* Not all digits -- may be user variable/function or a + convenience variable. */ + LONGEST valx; + struct internalvar *ivar; + + /* Try it as a convenience variable. If it is not a convenience + variable, return and allow normal symbol lookup to occur. */ + ivar = lookup_only_internalvar (variable + 1); + if (ivar == NULL) + /* No internal variable with that name. Mark the offset + as unknown to allow the name to be looked up as a symbol. */ + offset.sign = LINE_OFFSET_UNKNOWN; + else + { + /* We found a valid variable name. If it is not an integer, + throw an error. */ + if (!get_internalvar_integer (ivar, &valx)) + error (_("Convenience variables used in line " + "specs must have integer values.")); + else + offset.offset = valx; + } } - return result->nelts > 0; + return offset; } +\f /* A callback used to possibly add a symbol to the results. */ @@ -2763,20 +3174,16 @@ static int collect_symbols (struct symbol *sym, void *data) { struct collect_info *info = data; - struct symtab_and_line sal; - - if (symbol_to_sal (&sal, info->state->funfirstline, sym) - && maybe_add_address (info->state->addr_set, - SYMTAB_PSPACE (SYMBOL_SYMTAB (sym)), - sal.pc)) - add_sal_to_sals (info->state, &info->result, &sal, - SYMBOL_NATURAL_NAME (sym)); + /* In list mode, add all matching symbols, regardless of class. + This allows the user to type "list a_global_variable". */ + if (SYMBOL_CLASS (sym) == LOC_BLOCK || info->state->list_mode) + VEC_safe_push (symbolp, info->result.symbols, sym); return 1; /* Continue iterating. */ } -/* We've found a minimal symbol MSYMBOL to associate with our - linespec; add it to the result symtabs_and_lines. */ +/* We've found a minimal symbol MSYMBOL in OBJFILE to associate with our + linespec; return the SAL in RESULT. */ static void minsym_found (struct linespec_state *self, struct objfile *objfile, @@ -2804,17 +3211,6 @@ minsym_found (struct linespec_state *self, struct objfile *objfile, add_sal_to_sals (self, result, &sal, SYMBOL_NATURAL_NAME (msymbol)); } -/* A helper struct which just holds a minimal symbol and the object - file from which it came. */ - -typedef struct minsym_and_objfile -{ - struct minimal_symbol *minsym; - struct objfile *objfile; -} minsym_and_objfile_d; - -DEF_VEC_O (minsym_and_objfile_d); - /* A helper struct to pass some data through iterate_over_minimal_symbols. */ @@ -2967,8 +3363,8 @@ search_minsyms_for_name (struct collect_info *info, const char *name, if (classify_mtype (MSYMBOL_TYPE (item->minsym)) != classification) break; - minsym_found (info->state, item->objfile, item->minsym, - &info->result); + VEC_safe_push (minsym_and_objfile_d, + info->result.minimal_symbols, item); } } @@ -2988,13 +3384,13 @@ add_matching_symbols_to_info (const char *name, int ix; struct symtab *elt; - for (ix = 0; VEC_iterate (symtab_p, info->state->file_symtabs, ix, elt); ++ix) + for (ix = 0; VEC_iterate (symtab_p, info->file_symtabs, ix, elt); ++ix) { struct symbol *sym; if (elt == NULL) { - iterate_over_all_matching_symtabs (name, VAR_DOMAIN, + iterate_over_all_matching_symtabs (info->state, name, VAR_DOMAIN, collect_symbols, info, pspace, 1); search_minsyms_for_name (info, name, pspace); @@ -3012,67 +3408,6 @@ add_matching_symbols_to_info (const char *name, } } -/* Decode a linespec that's a variable. If FILE_SYMTAB is non-NULL, - look in that symtab's static variables first. */ - -static struct symtabs_and_lines -decode_variable (struct linespec_state *self, char *copy) -{ - struct collect_info info; - const char *lookup_name; - char *canon; - struct cleanup *cleanup; - - info.state = self; - info.result.sals = NULL; - info.result.nelts = 0; - - cleanup = demangle_for_lookup (copy, current_language->la_language, - &lookup_name); - if (current_language->la_language == language_ada) - { - /* In Ada, the symbol lookups are performed using the encoded - name rather than the demangled name. */ - lookup_name = ada_name_for_lookup (copy); - make_cleanup (xfree, (void *) lookup_name); - } - - canon = cp_canonicalize_string_no_typedefs (lookup_name); - if (canon != NULL) - { - make_cleanup (xfree, canon); - lookup_name = canon; - } - - add_matching_symbols_to_info (lookup_name, &info, NULL); - - if (info.result.nelts > 0) - { - if (self->canonical) - { - self->canonical->pre_expanded = 1; - if (self->user_filename) - self->canonical->addr_string - = xstrprintf ("%s:%s", self->user_filename, copy); - else - self->canonical->addr_string = xstrdup (copy); - } - return info.result; - } - - if (!have_full_symbols () - && !have_partial_symbols () - && !have_minimal_symbols ()) - throw_error (NOT_FOUND_ERROR, - _("No symbol table is loaded. Use the \"file\" command.")); - if (self->user_filename) - throw_error (NOT_FOUND_ERROR, _("Function \"%s\" not defined in \"%s\"."), - copy, self->user_filename); - else - throw_error (NOT_FOUND_ERROR, _("Function \"%s\" not defined."), copy); -} - - \f /* Now come some functions that are called from multiple places within diff --git a/gdb/testsuite/gdb.base/advance.exp b/gdb/testsuite/gdb.base/advance.exp index 456268d..617a7fb 100644 --- a/gdb/testsuite/gdb.base/advance.exp +++ b/gdb/testsuite/gdb.base/advance.exp @@ -45,7 +45,8 @@ gdb_test "advance [gdb_get_line_number "advance this location"]" \ # Verify that a malformed "advance" is gracefully caught. # gdb_test "advance [gdb_get_line_number "advance malformed"] then stop" \ - "Junk at end of arguments." "malformed advance" + "malformed linespec error: unexpected string, \"then stop\"" \ + "malformed advance" # Verify that "advance <funcname>" works. # diff --git a/gdb/testsuite/gdb.base/break.exp b/gdb/testsuite/gdb.base/break.exp index 71eb3a6..a203364 100644 --- a/gdb/testsuite/gdb.base/break.exp +++ b/gdb/testsuite/gdb.base/break.exp @@ -600,7 +600,7 @@ gdb_test "break $bp_location12 thread foo" \ # trailing garbage. # gdb_test "break $bp_location12 foo" \ - "Junk at end of arguments.*" \ + "malformed linespec error: unexpected string, \"foo\".*" \ "breakpoint with trailing garbage disallowed" # Verify that GDB responds gracefully to a "clear" command that has diff --git a/gdb/testsuite/gdb.base/hbreak2.exp b/gdb/testsuite/gdb.base/hbreak2.exp index 7f52cd3..a11d8b6 100644 --- a/gdb/testsuite/gdb.base/hbreak2.exp +++ b/gdb/testsuite/gdb.base/hbreak2.exp @@ -356,7 +356,7 @@ gdb_test "hbreak $bp_location12 thread foo" \ # trailing garbage. # gdb_test "hbreak $bp_location12 foo" \ - "Junk at end of arguments.*" \ + "malformed linespec error: unexpected string, \"foo\".*" \ "hardware breakpoint with trailing garbage disallowed" # Verify that GDB responds gracefully to a "clear" command that has diff --git a/gdb/testsuite/gdb.base/jump.exp b/gdb/testsuite/gdb.base/jump.exp index a5577e2..a7b2177 100644 --- a/gdb/testsuite/gdb.base/jump.exp +++ b/gdb/testsuite/gdb.base/jump.exp @@ -92,7 +92,7 @@ gdb_test "jump" "Argument required .starting address.*" \ # trailing junk. # gdb_test "jump 21 100" \ - "Junk at end of line specification: 100.*" \ + "malformed linespec error: unexpected number, \"100\"" \ "jump with trailing argument junk" diff --git a/gdb/testsuite/gdb.base/sepdebug.exp b/gdb/testsuite/gdb.base/sepdebug.exp index bd60c70..e6973fb 100644 --- a/gdb/testsuite/gdb.base/sepdebug.exp +++ b/gdb/testsuite/gdb.base/sepdebug.exp @@ -408,7 +408,7 @@ gdb_test "break $bp_location12 thread foo" \ # gdb_test "break $bp_location12 foo" \ - "Junk at end of arguments.*" \ + "malformed linespec error: unexpected string, \"foo\".*" \ "breakpoint with trailing garbage disallowed" # Verify that GDB responds gracefully to a "clear" command that has diff --git a/gdb/testsuite/gdb.base/until.exp b/gdb/testsuite/gdb.base/until.exp index 6b647cd..bdee21d 100644 --- a/gdb/testsuite/gdb.base/until.exp +++ b/gdb/testsuite/gdb.base/until.exp @@ -40,7 +40,8 @@ gdb_test "until $bp_location1" \ # Verify that a malformed "advance" is gracefully caught. # gdb_test "until 80 then stop" \ - "Junk at end of arguments." "malformed until" + "malformed linespec error: unexpected string, \"then stop\"." \ + "malformed until" # Rerun up to factorial, outer invocation if { ![runto factorial] } then { gdb_suppress_tests; } diff --git a/gdb/testsuite/gdb.cp/cplabel.cc b/gdb/testsuite/gdb.cp/cplabel.cc new file mode 100644 index 0000000..876f802 --- /dev/null +++ b/gdb/testsuite/gdb.cp/cplabel.cc @@ -0,0 +1,80 @@ +/* This testcase is part of GDB, the GNU debugger. + + Copyright 2012 Free Software Foundation, Inc. + + 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/>. */ + +class foo +{ +public: + static int bar (void) + { + int i = 5; + bool first = true; + + to_the_top: /* bar:to_the_top */ + while (1) + { + if (i == 1) + { + if (first) + { + first = false; + goto to_the_top; + } + else + goto get_out_of_here; + } + + --i; + } + + get_out_of_here: /* bar:get_out_of_here */ + return i; + } + + int baz (int a) + { + int i = a; + bool first = true; + + to_the_top: /* baz:to_the_top */ + while (1) + { + if (i == 1) + { + if (first) + { + first = false; + goto to_the_top; + } + else + goto get_out_of_here; + } + + --i; + } + + get_out_of_here: /* baz:get_out_of_here */ + return i; + } +}; + +int +main (void) +{ + foo f; + return f.baz (foo::bar () + 3); +} + diff --git a/gdb/testsuite/gdb.cp/cplabel.exp b/gdb/testsuite/gdb.cp/cplabel.exp new file mode 100644 index 0000000..30dfa39 --- /dev/null +++ b/gdb/testsuite/gdb.cp/cplabel.exp @@ -0,0 +1,40 @@ +# Copyright 2012 Free Software Foundation, Inc. + +# 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/>. + +# Tests for breakpoint on labels in methods. + +if {[skip_cplus_tests]} { continue } + +set testfile cplabel +set srcfile "$testfile.cc" +if {[prepare_for_testing "$testfile.exp" $testfile $srcfile {c++ debug}]} { + return -1 +} + +if {![runto_main]} { + untested "could not run to main" + return -1 +} + +set methods {"bar" "baz"} +set labels {"to_the_top" "get_out_of_here"} + +foreach m $methods { + foreach l $labels { + set line [gdb_get_line_number "$m:$l"] + gdb_test "break foo::$m:$l" \ + "Breakpoint $decimal at $hex: file .*/$srcfile, line $line\." + } +} diff --git a/gdb/testsuite/gdb.cp/ovldbreak.exp b/gdb/testsuite/gdb.cp/ovldbreak.exp index 3e35b79..2c27f2d 100644 --- a/gdb/testsuite/gdb.cp/ovldbreak.exp +++ b/gdb/testsuite/gdb.cp/ovldbreak.exp @@ -320,23 +320,17 @@ gdb_expect { } } -gdb_test "info break" \ - "Num Type\[\t \]+Disp Enb Address\[\t \]+What.* -\[0-9\]+\[\t \]+breakpoint keep y\[\t \]+<MULTIPLE>\[\t \]*\r -\[0-9\]+.1\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(double\\) at.*$srcfile:140\r -\[0-9\]+.2\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(float\\) at.*$srcfile:137\r -\[0-9\]+.3\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\((unsigned long|long unsigned)( int)?\\) at.*$srcfile:134\r -\[0-9\]+.4\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(long( int)?\\) at.*$srcfile:131\r -\[0-9\]+.5\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\((unsigned|unsigned int)\\) at.*$srcfile:128\r -\[0-9\]+.6\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(int\\) at.*$srcfile:125\r -\[0-9\]+.7\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\((unsigned short|short unsigned)( int)?\\) at.*$srcfile:122\r -\[0-9\]+.8\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(short( int)?\\) at.*$srcfile:119\r -\[0-9\]+.9\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(unsigned char\\) at.*$srcfile:116\r -\[0-9\]+.10\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(signed char\\) at.*$srcfile:113\r -\[0-9\]+.11\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(char\\) at.*$srcfile:110\r -\[0-9\]+.12\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\((void|)\\) at.*$srcfile:107" \ - "breakpoint info (after setting on all)" +# Create the breakpoint table for "info breakpoint". +set bptable "Num Type\[\t \]+Disp Enb Address\[\t \]+What.*\[\r\n]+" +append bptable "\[0-9\]+\[\t \]+breakpoint\[\t \]+keep\[\t \]y\[\t \]+<MULTIPLE>.*\[\r\n\]+" +foreach ovld {void char signed_char unsigned_char short_int \ + unsigned_short_int int unsigned_int long_int \ + unsigned_long_int float double} { + append bptable [format "\[0-9\]+.\[0-9\]+\[\t \]+y\[\t \]+$hex\[\t \]+in foo::overload1arg\\(%s\\) at.*$srcfile:%d\[\r\n\]+" \ + $types($ovld) $line($ovld)] +} +gdb_test "info break" $bptable "breakpoint info (after setting on all)" # Run through each breakpoint. proc continue_to_bp_overloaded {bpnumber might_fail line argtype argument} { diff --git a/gdb/testsuite/gdb.cp/userdef.exp b/gdb/testsuite/gdb.cp/userdef.exp index 9c7cb57..90eb99d 100644 --- a/gdb/testsuite/gdb.cp/userdef.exp +++ b/gdb/testsuite/gdb.cp/userdef.exp @@ -136,8 +136,8 @@ gdb_test "print one += 7" "\\\$\[0-9\]* = {x = 9, y = 10}" gdb_test "print two = one" "\\\$\[0-9\]* = {x = 9, y = 10}" # Check that GDB tolerates whitespace in operator names. -gdb_test "break A2::'operator+'" ".*Breakpoint $decimal at.*" -gdb_test "break A2::'operator +'" ".*Breakpoint $decimal at.*" +gdb_test "break A2::operator+" ".*Breakpoint $decimal at.*" +gdb_test "break A2::operator +" ".*Breakpoint $decimal at.*" # Check that GDB handles operator* correctly. gdb_test "print c" "\\\$\[0-9\]* = {m = {z = .*}}" diff --git a/gdb/testsuite/gdb.linespec/ls-errs.c b/gdb/testsuite/gdb.linespec/ls-errs.c new file mode 100644 index 0000000..bb4c096 --- /dev/null +++ b/gdb/testsuite/gdb.linespec/ls-errs.c @@ -0,0 +1,29 @@ +/* This testcase is part of GDB, the GNU debugger. + + Copyright 2012 Free Software Foundation, Inc. + + 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/>. */ + +int myfunction (void) { return 0; } + +int +main (void) +{ + int a; + + a = myfunction (); + + here: + return a; +} diff --git a/gdb/testsuite/gdb.linespec/ls-errs.exp b/gdb/testsuite/gdb.linespec/ls-errs.exp new file mode 100644 index 0000000..21e01bf --- /dev/null +++ b/gdb/testsuite/gdb.linespec/ls-errs.exp @@ -0,0 +1,189 @@ +# Copyright 2012 Free Software Foundation, Inc. + +# 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/>. + +# Tests for linespec error conditions + +set base ls-errs +set srcfile "$base.c" +set testfile "$base.exp" +set exefile $base + +if {[prepare_for_testing $testfile $exefile $srcfile \ + {debug nowarnings}]} { + return -1 +} + +# Turn off the pending breakpoint queries. +gdb_test_no_output "set breakpoint pending off" + +# We intentionally do not use gdb_breakpoint for these tests. + +# Add the (invalid) LINESPEC to the test array named in ARRAY_NAME. +# Use the index into ::error_messages MSG_ID and ARGS to create +# an error message which is the expect result of attempting to +# break on the given LINESPEC. +proc add {array_name linespec msg_id args} { + global error_messages + upvar $array_name tests + + lappend tests(linespecs) $linespec + set tests("$linespec") [string_to_regexp \ + [eval format \$error_messages($msg_id) $args]] +} + +# Common error message format strings. +array set error_messages { + invalid_file "No source file named %s." + invalid_function "Function \"%s\" not defined." + invalid_var_or_func "Undefined convenience variable or function \"%s\" not defined." + invalid_function_f "Function \"%s\" not defined in \"%s\"." + invalid_var_or_func_f \ + "Undefined convenience variable or function \"%s\" not defined in \"%s\"." + invalid_label "No label \"%s\" defined in function \"%s\"." + invalid_offset "No line %d in the current file." + invalid_offset_f "No line %d in file \"%s\"." + unexpected "malformed linespec error: unexpected %s" + unexpected_opt "malformed linespec error: unexpected %s, \"%s\"" + unmatched_quote "unmatched quote" +} + +# Some commonly used whitespace tests around ':'. +set spaces [list ":" ": " " :" " : " "\t: " " :\t" "\t:\t" " \t:\t " \ + "\t \t:\t \t \t"] + +# A list of invalid offsets. +set invalid_offsets [list -100 +500 1000] + +# THE_TESTS will hold all of our test information. Array index +# "linespecs" will contain the complete list of all linespecs +# to be tested. An array index of \"$linespec\" will contain +# the expected result. +set the_tests(linespecs) {} + +# Try some simple, invalid linespecs involving spaces. +foreach x $spaces { + add the_tests $x unexpected "colon" +} + +# Test invalid filespecs starting with offset. This is done +# first so that default offsets are tested. +foreach x $invalid_offsets { + set offset $x + + # Relative offsets are relative to line 16. Adjust + # expected offset from error message accordingly. + if {[string index $x 0] == "+" || + [string index $x 0] == "-"} { + incr offset 16 + } + add the_tests $x invalid_offset $offset +} + +# Test offsets with trailing tokens w/ and w/o spaces. +foreach x $spaces { + add the_tests "3$x" unexpected "colon" + add the_tests "+10$x" unexpected "colon" + add the_tests "-10$x" unexpected "colon" +} + +foreach x {1 +1 +100 -10} { + add the_tests "3 $x" unexpected_opt "number" $x + add the_tests "+10 $x" unexpected_opt "number" $x + add the_tests "-10 $x" unexpected_opt "number" $x +} + +add the_tests "3 foo" unexpected_opt "string" "foo" +add the_tests "+10 foo" unexpected_opt "string" "foo" +add the_tests "-10 foo" unexpected_opt "string" "foo" + +# Test invalid linespecs starting with filename. +foreach x [list "this_file_doesn't_exist.c" \ + "this file has spaces.c" \ + "\"file::colons.c\"" \ + "'file::colons.c'" \ + "\"this \"file\" has quotes.c\"" \ + "'this \"file\" has quotes.c'" \ + "'this 'file' has quotes.c'" \ + "\"this 'file' has quotes.c\""] { + # Remove any quoting from FILENAME for the error message. + add the_tests "$x:3" invalid_file [string trim $x \"'] +} + +# Test unmatched quotes. +foreach x {"\"src-file.c'" "'src-file.c"} { + add the_tests "$x:3" unmatched_quote +} + +add the_tests $srcfile invalid_function $srcfile +foreach x {"foo" " foo" " foo "} { + # Trim any leading/trailing whitespace for error messages. + add the_tests "$srcfile:$x" invalid_function_f [string trim $x] $srcfile + add the_tests "$srcfile:main:$x" invalid_label [string trim $x] "main" +} + +foreach x $spaces { + add the_tests "$srcfile$x" unexpected "end of input" + add the_tests "$srcfile:main$x" unexpected "end of input" +} + +add the_tests "${srcfile}::" invalid_function "${srcfile}::" +add the_tests "$srcfile:3 1" unexpected_opt "number" "1" +add the_tests "$srcfile:3 +100" unexpected_opt "number" "+100" +add the_tests "$srcfile:3 -100" unexpected_opt "number" "-100" +add the_tests "$srcfile:3 foo" unexpected_opt "string" "foo" + +foreach x $invalid_offsets { + add the_tests "$srcfile:$x" invalid_offset_f $x $srcfile +} + +# Test invalid filespecs starting with function. +foreach x {"foobar" "foo::bar" "foo.bar" "foo ." "foo bar" "foo 1" \ + "foo 0" "foo +10" "foo -10" "foo +100" "foo -100"} { + add the_tests $x invalid_function $x +} + +foreach x $spaces { + add the_tests "main${x}there" invalid_label "there" "main" + add the_tests "main:here${x}" unexpected "end of input" +} + +add the_tests "main 3" invalid_function "main 3" +add the_tests "main +100" invalid_function "main +100" +add the_tests "main -100" invalid_function "main -100" +add the_tests "main foo" invalid_function "main foo" + +foreach x {"3" "+100" "-100" "foo"} { + add the_tests "main:here $x" invalid_label "here $x" "main" +} + +foreach x {"if" "task" "thread"} { + add the_tests $x unexpected_opt "keyword" $x +} + +add the_tests "'main.c'flubber" unexpected_opt "string" "flubber" +add the_tests "'main.c',21" invalid_function "main.c" +add the_tests "'main.c' " invalid_function "main.c" +add the_tests "'main.c'3" unexpected_opt "number" "3" +add the_tests "'main.c'+3" unexpected_opt "number" "+3" + +# Test undefined convenience variables. +set x {$zippo} +add the_tests $x invalid_var_or_func $x +add the_tests "$srcfile:$x" invalid_var_or_func_f $x $srcfile + +# Run the tests +foreach linespec $the_tests(linespecs) { + gdb_test "break $linespec" $the_tests("$linespec") +} ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-04-03 23:22 ` Keith Seitz @ 2012-04-05 15:22 ` Tom Tromey 2012-04-05 15:55 ` Doug Evans 2012-07-22 19:33 ` Andreas Schwab 1 sibling, 1 reply; 26+ messages in thread From: Tom Tromey @ 2012-04-05 15:22 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml >>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes: Keith> Given Doug's comments earlier today, I'm reposting anything that Keith> hasn't already been approved or which changed. The easiest way to see Keith> what's happened is, of course, by inspecting the commit history of the Keith> branch. Keith> I'll follow-up separately with the other patch that was updated. I have no further comments. I think it looks great. I think at this point you just need Doug's ok. Tom ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-04-05 15:22 ` Tom Tromey @ 2012-04-05 15:55 ` Doug Evans 2012-04-05 19:01 ` Keith Seitz 0 siblings, 1 reply; 26+ messages in thread From: Doug Evans @ 2012-04-05 15:55 UTC (permalink / raw) To: Tom Tromey; +Cc: Keith Seitz, gdb-patches@sourceware.org ml On Thu, Apr 5, 2012 at 8:21 AM, Tom Tromey <tromey@redhat.com> wrote: >>>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes: > > Keith> Given Doug's comments earlier today, I'm reposting anything that > Keith> hasn't already been approved or which changed. The easiest way to see > Keith> what's happened is, of course, by inspecting the commit history of the > Keith> branch. > > Keith> I'll follow-up separately with the other patch that was updated. > > I have no further comments. I think it looks great. > > I think at this point you just need Doug's ok. All I had were nits, so 'tis ok with me. ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-04-05 15:55 ` Doug Evans @ 2012-04-05 19:01 ` Keith Seitz 0 siblings, 0 replies; 26+ messages in thread From: Keith Seitz @ 2012-04-05 19:01 UTC (permalink / raw) To: gdb-patches@sourceware.org ml On 04/05/2012 08:54 AM, Doug Evans wrote: > On Thu, Apr 5, 2012 at 8:21 AM, Tom Tromey<tromey@redhat.com> wrote: >>>>>>> "Keith" == Keith Seitz<keiths@redhat.com> writes: >> >> Keith> Given Doug's comments earlier today, I'm reposting anything that >> Keith> hasn't already been approved or which changed. The easiest way to see >> Keith> what's happened is, of course, by inspecting the commit history of the >> Keith> branch. >> >> Keith> I'll follow-up separately with the other patch that was updated. >> >> I have no further comments. I think it looks great. >> >> I think at this point you just need Doug's ok. > > All I had were nits, so 'tis ok with me. And with that, I've committed the lot. Thank you to everyone who reviewed and played with the patchset! Keith ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-04-03 23:22 ` Keith Seitz 2012-04-05 15:22 ` Tom Tromey @ 2012-07-22 19:33 ` Andreas Schwab 2012-07-23 17:57 ` Keith Seitz 1 sibling, 1 reply; 26+ messages in thread From: Andreas Schwab @ 2012-07-22 19:33 UTC (permalink / raw) To: Keith Seitz; +Cc: Tom Tromey, gdb-patches@sourceware.org ml Keith Seitz <keiths@redhat.com> writes: > + for (i = 0; VEC_iterate (symbolp, ls->labels.label_symbols, i, sym); ++i) > + { > + symbol_to_sal (&sal, state->funfirstline, sym); > + add_sal_to_sals (state, &sals, &sal, > + SYMBOL_NATURAL_NAME (sym)); If symbol_to_sal returns 0 then sal is uninitialized. > + } > + } > + else if (ls->function_symbols != NULL || ls->minimal_symbols != NULL) > + { > + /* We have just a bunch of functions and/or methods. */ > + int i; > + struct symtab_and_line sal; > + struct symbol *sym; > + minsym_and_objfile_d *elem; > + struct program_space *pspace; > + > + if (ls->function_symbols != NULL) > + { > + /* Sort symbols so that symbols with the same program space are next > + to each other. */ > + qsort (VEC_address (symbolp, ls->function_symbols), > + VEC_length (symbolp, ls->function_symbols), > + sizeof (symbolp), compare_symbols); > + > + for (i = 0; VEC_iterate (symbolp, ls->function_symbols, i, sym); ++i) > + { > + pspace = SYMTAB_PSPACE (SYMBOL_SYMTAB (sym)); > + set_current_program_space (pspace); > + symbol_to_sal (&sal, state->funfirstline, sym); > + if (maybe_add_address (state->addr_set, pspace, sal.pc)) > + add_sal_to_sals (state, &sals, &sal, SYMBOL_NATURAL_NAME (sym)); Likewise. Andreas. -- Andreas Schwab, schwab@linux-m68k.org GPG Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5 "And now for something completely different." ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-07-22 19:33 ` Andreas Schwab @ 2012-07-23 17:57 ` Keith Seitz 2012-07-23 18:55 ` Tom Tromey 0 siblings, 1 reply; 26+ messages in thread From: Keith Seitz @ 2012-07-23 17:57 UTC (permalink / raw) To: gdb-patches@sourceware.org ml [-- Attachment #1: Type: text/plain, Size: 684 bytes --] On 07/22/2012 12:32 PM, Andreas Schwab wrote: > Keith Seitz <keiths@redhat.com> writes: > >> + for (i = 0; VEC_iterate (symbolp, ls->labels.label_symbols, i, sym); ++i) >> + { >> + symbol_to_sal (&sal, state->funfirstline, sym); >> + add_sal_to_sals (state, &sals, &sal, >> + SYMBOL_NATURAL_NAME (sym)); > > If symbol_to_sal returns 0 then sal is uninitialized. > Nice catch, Andreas. I propose the attached patch. I guess this should be considered for inclusion on 7.5. [What say ye?] Keith ChangeLog 2012-07-23 Keith Seitz <keiths@redhat.com> * linespec.c (convert_linespec_to_sal): Don't add any symbols to the result vector if symbol_to_sal returns zero. [-- Attachment #2: uninitialized_sal-from-symbol_to_sal.patch --] [-- Type: text/x-patch, Size: 1271 bytes --] Index: linespec.c =================================================================== RCS file: /cvs/src/src/gdb/linespec.c,v retrieving revision 1.162 diff -u -p -r1.162 linespec.c --- linespec.c 18 Jul 2012 20:38:18 -0000 1.162 +++ linespec.c 23 Jul 2012 17:54:48 -0000 @@ -1860,9 +1860,9 @@ convert_linespec_to_sals (struct linespe for (i = 0; VEC_iterate (symbolp, ls->labels.label_symbols, i, sym); ++i) { - symbol_to_sal (&sal, state->funfirstline, sym); - add_sal_to_sals (state, &sals, &sal, - SYMBOL_NATURAL_NAME (sym), 0); + if (symbol_to_sal (&sal, state->funfirstline, sym)) + add_sal_to_sals (state, &sals, &sal, + SYMBOL_NATURAL_NAME (sym), 0); } } else if (ls->function_symbols != NULL || ls->minimal_symbols != NULL) @@ -1886,8 +1886,8 @@ convert_linespec_to_sals (struct linespe { pspace = SYMTAB_PSPACE (SYMBOL_SYMTAB (sym)); set_current_program_space (pspace); - symbol_to_sal (&sal, state->funfirstline, sym); - if (maybe_add_address (state->addr_set, pspace, sal.pc)) + if (symbol_to_sal (&sal, state->funfirstline, sym) + && maybe_add_address (state->addr_set, pspace, sal.pc)) add_sal_to_sals (state, &sals, &sal, SYMBOL_NATURAL_NAME (sym), 0); } ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-07-23 17:57 ` Keith Seitz @ 2012-07-23 18:55 ` Tom Tromey 2012-07-23 21:13 ` Keith Seitz 0 siblings, 1 reply; 26+ messages in thread From: Tom Tromey @ 2012-07-23 18:55 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml >>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes: Keith> I guess this should be considered for inclusion on 7.5. [What say ye?] Sounds good Keith> 2012-07-23 Keith Seitz <keiths@redhat.com> Keith> * linespec.c (convert_linespec_to_sal): Don't add Keith> any symbols to the result vector if symbol_to_sal Keith> returns zero. Ok. Tom ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-07-23 18:55 ` Tom Tromey @ 2012-07-23 21:13 ` Keith Seitz 0 siblings, 0 replies; 26+ messages in thread From: Keith Seitz @ 2012-07-23 21:13 UTC (permalink / raw) To: gdb-patches@sourceware.org ml On 07/23/2012 11:55 AM, Tom Tromey wrote: >>>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes: > > Keith> I guess this should be considered for inclusion on 7.5. [What say ye?] > > Sounds good > > Keith> 2012-07-23 Keith Seitz <keiths@redhat.com> > Keith> * linespec.c (convert_linespec_to_sal): Don't add > Keith> any symbols to the result vector if symbol_to_sal > Keith> returns zero. > > Ok. Committed on both HEAD and gdb_7_5-branch. Thank you, Tom. Keith ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-03-26 23:18 [RFA 1/2] Linespec rewrite (update 2) Keith Seitz ` (2 preceding siblings ...) 2012-03-28 20:46 ` Tom Tromey @ 2012-04-03 21:19 ` Doug Evans 2012-04-03 23:14 ` Keith Seitz 3 siblings, 1 reply; 26+ messages in thread From: Doug Evans @ 2012-04-03 21:19 UTC (permalink / raw) To: Keith Seitz; +Cc: gdb-patches@sourceware.org ml On Mon, Mar 26, 2012 at 4:17 PM, Keith Seitz <keiths@redhat.com> wrote: > Hi, > > Tom requested a repost of this patch, so here it is. Here is a list of > differences between this version and the previously posted patch: > > * skip_quote_char rewritten (Doug found a host of bugs in it) > * added backward compatibility for quote-enclosed linespecs (so no NEWS > entry needed) > * copy_token_string now uses replace_trailing_whitespace function from > cli/cli-utils.[ch] > * miscellaneous test fiddling (from other posts by Tom and Pedro) > * some new error checking/conditions added (bugs reported by Doug) > > This patch is now essentially standalone, as Pedro requested. [Previously I > posted patches to be committed all together when finally approved.] > > Keith > > ChangeLog > 2012-03-26 Keith Seitz <keiths@redhat.com> Hi. I don't know if any of these have been addressed, but here are my new comments. All nits. 1) +typedef struct linespec *linespec_t; Replace "_t" suffix with "p" or "_p" or some such. E.g., typedef struct symtab *symtab_p; typedef struct symbol *symbolp; typedef struct type *typep; 2) /* An enumeration of possible signs for a line offset. */ enum offset_relative_sign { /* No sign */ none, /* A plus sign ("+") */ plus, /* A minus sign ("-") */ minus, /* A special "sign" for unspecified offset. */ unknown }; --> /* An enumeration of possible signs for a line offset. */ enum line_offset_sign { /* No sign */ line_offset_none, /* A plus sign ("+") */ line_offset_plus, /* A minus sign ("-") */ line_offset_minus, /* A special "sign" for unspecified offset. */ line_offset_unknown }; 3) In: /* A linespec. Elements of this structure are filled in by a parser (either parse_linespec or some other function). The structure is then converted into SALs by convert_linespec_to_sals. */ struct linespec { /* An expression and the resulting PC. */ char *expression; CORE_ADDR expr_pc; /* Any specified file symtabs. */ char *source_filename; VEC (symtab_p) *file_symtabs; /* The name of a function or method and any matching symbols. */ char *function_name; VEC (symbolp) *function_symbols; VEC (minsym_and_objfile_d) *minimal_symbols; /* The name of a label and matching symbols. */ char *label_name; struct { VEC (symbolp) *label_symbols; VEC (symbolp) *function_symbols; } labels; /* Line offset */ struct line_offset line_offset; }; The members here need more documentation. Something like explicitly stating that if the value was absent in the linespec, then the pointer is NULL and the corresponding values (e.g., expr_pc) are invalid. 4) /* Token types */ enum ls_token_type { /* A keyword */ LSTOKEN_KEYWORD = 0, Yay, uppercase enum values. Can we uppercase all enum values in this file? 5) static char * skip_quote_char (const char *string, char quote_char) Return const char *, and if the cast to char * is necessary, put it in the caller. 6) /* Special case: Ada operators. */ if (current_language->la_language == language_ada && quote_char == '\"') So some aspect of ' vs " remains? Plus, IWBN if the language to use was not taken from a global. It can start out as a global, but the innards should take it as a parameter (e.g. record in linespec_parser). 7) In parse_linespec: /* This token must be LSTOKEN_COLON. */ if (token.type != LSTOKEN_COLON) return values; It's not clear to me what case this is handling. Is "must" too strong a word? When will the "then" clause be taken? 8) /* Initialize a new linespec parser. */ static void linespec_parser_new (linespec_parser *parser, Blank line after comment. ^ permalink raw reply [flat|nested] 26+ messages in thread
* Re: [RFA 1/2] Linespec rewrite (update 2) 2012-04-03 21:19 ` Doug Evans @ 2012-04-03 23:14 ` Keith Seitz 0 siblings, 0 replies; 26+ messages in thread From: Keith Seitz @ 2012-04-03 23:14 UTC (permalink / raw) To: Doug Evans; +Cc: gdb-patches@sourceware.org ml On 04/03/2012 02:18 PM, Doug Evans wrote: > 1) Replace "_t" suffix with "p" or "_p" or some such. Done. > 2) [enums UPPERCASE & prefixed] > /* No sign */ > line_offset_none, > > /* A plus sign ("+") */ > line_offset_plus, > > /* A minus sign ("-") */ > line_offset_minus, > > /* A special "sign" for unspecified offset. */ > line_offset_unknown > }; > 4) > > /* Token types */ > > enum ls_token_type > { > /* A keyword */ > LSTOKEN_KEYWORD = 0, > > Yay, uppercase enum values. > Can we uppercase all enum values in this file? Done. > 3) struct linespec > The members here need more documentation. > Something like explicitly stating that if the value was absent in the > linespec, then the pointer is NULL and the corresponding values (e.g., expr_pc) > are invalid. I've added some comments. Please review and let me know if you think more is needed. This also elided the fact that find_label_symbols requires its third parameter and it may not be NULL. I've removed the check for non-NULL LABEL_FUNCS_RET in that function (and checked all callers). > 5) > > static char * > skip_quote_char (const char *string, char quote_char) > > Return const char *, and if the cast to char * is necessary, put it in > the caller. Done. > 6) > > /* Special case: Ada operators. */ > if (current_language->la_language == language_ada > && quote_char == '\"') > > So some aspect of ' vs " remains? Ada allows operators named (literally) "+". The quote is part of the actual method name. That's what the check is attempting to ascertain. > Plus, IWBN if the language to use was not taken from a global. > It can start out as a global, but the innards should take it > as a parameter (e.g. record in linespec_parser). Done. I had this in an earlier version. I don't know why I changed it. > 7) > > In parse_linespec: > > /* This token must be LSTOKEN_COLON. */ > if (token.type != LSTOKEN_COLON) > return values; > > It's not clear to me what case this is handling. > Is "must" too strong a word? When will the "then" clause be taken? If the linespec contains a filename, it *must* also have LSTOKEN_COLON. That is the only permitted usage of a filename. Since this case is already dealt with by the optimization earlier (where it checks if the next token is LSTOKEN_COLON before looking for the symtab), this isn't really needed at all. I've removed it. > 8) > > /* Initialize a new linespec parser. */ > static void > linespec_parser_new (linespec_parser *parser, > > Blank line after comment. Added. I've pushed the requested changes to my archer branch. [Tom's asked me to repost it as well.] Thank you very much for taking a look. Keith ^ permalink raw reply [flat|nested] 26+ messages in thread
end of thread, other threads:[~2012-07-23 21:13 UTC | newest] Thread overview: 26+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2012-03-26 23:18 [RFA 1/2] Linespec rewrite (update 2) Keith Seitz 2012-03-27 1:45 ` asmwarrior 2012-03-28 18:55 ` Tom Tromey 2012-03-27 13:56 ` Joel Brobecker 2012-03-27 14:31 ` Keith Seitz 2012-03-27 14:54 ` Joel Brobecker 2012-03-27 15:05 ` Joel Brobecker 2012-03-27 18:08 ` Keith Seitz 2012-03-28 20:46 ` Tom Tromey 2012-03-29 19:18 ` Keith Seitz 2012-03-30 15:34 ` Tom Tromey 2012-03-30 15:59 ` Keith Seitz 2012-03-30 16:37 ` Tom Tromey 2012-03-30 17:09 ` Joel Brobecker 2012-03-30 17:56 ` Tom Tromey 2012-03-30 18:05 ` Keith Seitz 2012-04-03 23:22 ` Keith Seitz 2012-04-05 15:22 ` Tom Tromey 2012-04-05 15:55 ` Doug Evans 2012-04-05 19:01 ` Keith Seitz 2012-07-22 19:33 ` Andreas Schwab 2012-07-23 17:57 ` Keith Seitz 2012-07-23 18:55 ` Tom Tromey 2012-07-23 21:13 ` Keith Seitz 2012-04-03 21:19 ` Doug Evans 2012-04-03 23:14 ` Keith Seitz
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox