URL
https://opencores.org/ocsvn/openrisc/openrisc/trunk
Subversion Repositories openrisc
[/] [openrisc/] [trunk/] [gnu-old/] [gdb-7.1/] [gdb/] [f-valprint.c] - Rev 227
Go to most recent revision | Compare with Previous | Blame | View Log
/* Support for printing Fortran values for GDB, the GNU debugger. Copyright (C) 1993, 1994, 1995, 1996, 1998, 1999, 2000, 2003, 2005, 2006, 2007, 2008, 2009, 2010 Free Software Foundation, Inc. Contributed by Motorola. Adapted from the C definitions by Farooq Butt (fmbutt@engage.sps.mot.com), additionally worked over by Stan Shebs. This file is part of GDB. This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program. If not, see <http://www.gnu.org/licenses/>. */ #include "defs.h" #include "gdb_string.h" #include "symtab.h" #include "gdbtypes.h" #include "expression.h" #include "value.h" #include "valprint.h" #include "language.h" #include "f-lang.h" #include "frame.h" #include "gdbcore.h" #include "command.h" #include "block.h" #if 0 static int there_is_a_visible_common_named (char *); #endif extern void _initialize_f_valprint (void); static void info_common_command (char *, int); static void list_all_visible_commons (char *); static void f77_create_arrayprint_offset_tbl (struct type *, struct ui_file *); static void f77_get_dynamic_length_of_aggregate (struct type *); int f77_array_offset_tbl[MAX_FORTRAN_DIMS + 1][2]; /* Array which holds offsets to be applied to get a row's elements for a given array. Array also holds the size of each subarray. */ /* The following macro gives us the size of the nth dimension, Where n is 1 based. */ #define F77_DIM_SIZE(n) (f77_array_offset_tbl[n][1]) /* The following gives us the offset for row n where n is 1-based. */ #define F77_DIM_OFFSET(n) (f77_array_offset_tbl[n][0]) int f77_get_lowerbound (struct type *type) { if (TYPE_ARRAY_LOWER_BOUND_IS_UNDEFINED (type)) error (_("Lower bound may not be '*' in F77")); return TYPE_ARRAY_LOWER_BOUND_VALUE (type); } int f77_get_upperbound (struct type *type) { if (TYPE_ARRAY_UPPER_BOUND_IS_UNDEFINED (type)) { /* We have an assumed size array on our hands. Assume that upper_bound == lower_bound so that we show at least 1 element. If the user wants to see more elements, let him manually ask for 'em and we'll subscript the array and show him. */ return f77_get_lowerbound (type); } return TYPE_ARRAY_UPPER_BOUND_VALUE (type); } /* Obtain F77 adjustable array dimensions */ static void f77_get_dynamic_length_of_aggregate (struct type *type) { int upper_bound = -1; int lower_bound = 1; int retcode; /* Recursively go all the way down into a possibly multi-dimensional F77 array and get the bounds. For simple arrays, this is pretty easy but when the bounds are dynamic, we must be very careful to add up all the lengths correctly. Not doing this right will lead to horrendous-looking arrays in parameter lists. This function also works for strings which behave very similarly to arrays. */ if (TYPE_CODE (TYPE_TARGET_TYPE (type)) == TYPE_CODE_ARRAY || TYPE_CODE (TYPE_TARGET_TYPE (type)) == TYPE_CODE_STRING) f77_get_dynamic_length_of_aggregate (TYPE_TARGET_TYPE (type)); /* Recursion ends here, start setting up lengths. */ lower_bound = f77_get_lowerbound (type); upper_bound = f77_get_upperbound (type); /* Patch in a valid length value. */ TYPE_LENGTH (type) = (upper_bound - lower_bound + 1) * TYPE_LENGTH (check_typedef (TYPE_TARGET_TYPE (type))); } /* Function that sets up the array offset,size table for the array type "type". */ static void f77_create_arrayprint_offset_tbl (struct type *type, struct ui_file *stream) { struct type *tmp_type; int eltlen; int ndimen = 1; int upper, lower, retcode; tmp_type = type; while ((TYPE_CODE (tmp_type) == TYPE_CODE_ARRAY)) { upper = f77_get_upperbound (tmp_type); lower = f77_get_lowerbound (tmp_type); F77_DIM_SIZE (ndimen) = upper - lower + 1; tmp_type = TYPE_TARGET_TYPE (tmp_type); ndimen++; } /* Now we multiply eltlen by all the offsets, so that later we can print out array elements correctly. Up till now we know an offset to apply to get the item but we also have to know how much to add to get to the next item */ ndimen--; eltlen = TYPE_LENGTH (tmp_type); F77_DIM_OFFSET (ndimen) = eltlen; while (--ndimen > 0) { eltlen *= F77_DIM_SIZE (ndimen + 1); F77_DIM_OFFSET (ndimen) = eltlen; } } /* Actual function which prints out F77 arrays, Valaddr == address in the superior. Address == the address in the inferior. */ static void f77_print_array_1 (int nss, int ndimensions, struct type *type, const gdb_byte *valaddr, CORE_ADDR address, struct ui_file *stream, int recurse, const struct value_print_options *options, int *elts) { int i; if (nss != ndimensions) { for (i = 0; (i < F77_DIM_SIZE (nss) && (*elts) < options->print_max); i++) { fprintf_filtered (stream, "( "); f77_print_array_1 (nss + 1, ndimensions, TYPE_TARGET_TYPE (type), valaddr + i * F77_DIM_OFFSET (nss), address + i * F77_DIM_OFFSET (nss), stream, recurse, options, elts); fprintf_filtered (stream, ") "); } if (*elts >= options->print_max && i < F77_DIM_SIZE (nss)) fprintf_filtered (stream, "..."); } else { for (i = 0; i < F77_DIM_SIZE (nss) && (*elts) < options->print_max; i++, (*elts)++) { val_print (TYPE_TARGET_TYPE (type), valaddr + i * F77_DIM_OFFSET (ndimensions), 0, address + i * F77_DIM_OFFSET (ndimensions), stream, recurse, options, current_language); if (i != (F77_DIM_SIZE (nss) - 1)) fprintf_filtered (stream, ", "); if ((*elts == options->print_max - 1) && (i != (F77_DIM_SIZE (nss) - 1))) fprintf_filtered (stream, "..."); } } } /* This function gets called to print an F77 array, we set up some stuff and then immediately call f77_print_array_1() */ static void f77_print_array (struct type *type, const gdb_byte *valaddr, CORE_ADDR address, struct ui_file *stream, int recurse, const struct value_print_options *options) { int ndimensions; int elts = 0; ndimensions = calc_f77_array_dims (type); if (ndimensions > MAX_FORTRAN_DIMS || ndimensions < 0) error (_("Type node corrupt! F77 arrays cannot have %d subscripts (%d Max)"), ndimensions, MAX_FORTRAN_DIMS); /* Since F77 arrays are stored column-major, we set up an offset table to get at the various row's elements. The offset table contains entries for both offset and subarray size. */ f77_create_arrayprint_offset_tbl (type, stream); f77_print_array_1 (1, ndimensions, type, valaddr, address, stream, recurse, options, &elts); } /* Print data of type TYPE located at VALADDR (within GDB), which came from the inferior at address ADDRESS, onto stdio stream STREAM according to OPTIONS. The data at VALADDR is in target byte order. If the data are a string pointer, returns the number of string characters printed. */ int f_val_print (struct type *type, const gdb_byte *valaddr, int embedded_offset, CORE_ADDR address, struct ui_file *stream, int recurse, const struct value_print_options *options) { struct gdbarch *gdbarch = get_type_arch (type); enum bfd_endian byte_order = gdbarch_byte_order (gdbarch); unsigned int i = 0; /* Number of characters printed */ struct type *elttype; LONGEST val; CORE_ADDR addr; int index; CHECK_TYPEDEF (type); switch (TYPE_CODE (type)) { case TYPE_CODE_STRING: f77_get_dynamic_length_of_aggregate (type); LA_PRINT_STRING (stream, builtin_type (gdbarch)->builtin_char, valaddr, TYPE_LENGTH (type), NULL, 0, options); break; case TYPE_CODE_ARRAY: fprintf_filtered (stream, "("); f77_print_array (type, valaddr, address, stream, recurse, options); fprintf_filtered (stream, ")"); break; case TYPE_CODE_PTR: if (options->format && options->format != 's') { print_scalar_formatted (valaddr, type, options, 0, stream); break; } else { addr = unpack_pointer (type, valaddr); elttype = check_typedef (TYPE_TARGET_TYPE (type)); if (TYPE_CODE (elttype) == TYPE_CODE_FUNC) { /* Try to print what function it points to. */ print_address_demangle (gdbarch, addr, stream, demangle); /* Return value is irrelevant except for string pointers. */ return 0; } if (options->addressprint && options->format != 's') fputs_filtered (paddress (gdbarch, addr), stream); /* For a pointer to char or unsigned char, also print the string pointed to, unless pointer is null. */ if (TYPE_LENGTH (elttype) == 1 && TYPE_CODE (elttype) == TYPE_CODE_INT && (options->format == 0 || options->format == 's') && addr != 0) i = val_print_string (TYPE_TARGET_TYPE (type), addr, -1, stream, options); /* Return number of characters printed, including the terminating '\0' if we reached the end. val_print_string takes care including the terminating '\0' if necessary. */ return i; } break; case TYPE_CODE_REF: elttype = check_typedef (TYPE_TARGET_TYPE (type)); if (options->addressprint) { CORE_ADDR addr = extract_typed_address (valaddr + embedded_offset, type); fprintf_filtered (stream, "@"); fputs_filtered (paddress (gdbarch, addr), stream); if (options->deref_ref) fputs_filtered (": ", stream); } /* De-reference the reference. */ if (options->deref_ref) { if (TYPE_CODE (elttype) != TYPE_CODE_UNDEF) { struct value *deref_val = value_at (TYPE_TARGET_TYPE (type), unpack_pointer (type, valaddr + embedded_offset)); common_val_print (deref_val, stream, recurse, options, current_language); } else fputs_filtered ("???", stream); } break; case TYPE_CODE_FUNC: if (options->format) { print_scalar_formatted (valaddr, type, options, 0, stream); break; } /* FIXME, we should consider, at least for ANSI C language, eliminating the distinction made between FUNCs and POINTERs to FUNCs. */ fprintf_filtered (stream, "{"); type_print (type, "", stream, -1); fprintf_filtered (stream, "} "); /* Try to print what function it points to, and its address. */ print_address_demangle (gdbarch, address, stream, demangle); break; case TYPE_CODE_INT: if (options->format || options->output_format) { struct value_print_options opts = *options; opts.format = (options->format ? options->format : options->output_format); print_scalar_formatted (valaddr, type, &opts, 0, stream); } else { val_print_type_code_int (type, valaddr, stream); /* C and C++ has no single byte int type, char is used instead. Since we don't know whether the value is really intended to be used as an integer or a character, print the character equivalent as well. */ if (TYPE_LENGTH (type) == 1) { fputs_filtered (" ", stream); LA_PRINT_CHAR ((unsigned char) unpack_long (type, valaddr), type, stream); } } break; case TYPE_CODE_FLAGS: if (options->format) print_scalar_formatted (valaddr, type, options, 0, stream); else val_print_type_code_flags (type, valaddr, stream); break; case TYPE_CODE_FLT: if (options->format) print_scalar_formatted (valaddr, type, options, 0, stream); else print_floating (valaddr, type, stream); break; case TYPE_CODE_VOID: fprintf_filtered (stream, "VOID"); break; case TYPE_CODE_ERROR: fprintf_filtered (stream, "<error type>"); break; case TYPE_CODE_RANGE: /* FIXME, we should not ever have to print one of these yet. */ fprintf_filtered (stream, "<range type>"); break; case TYPE_CODE_BOOL: if (options->format || options->output_format) { struct value_print_options opts = *options; opts.format = (options->format ? options->format : options->output_format); print_scalar_formatted (valaddr, type, &opts, 0, stream); } else { val = extract_unsigned_integer (valaddr, TYPE_LENGTH (type), byte_order); if (val == 0) fprintf_filtered (stream, ".FALSE."); else if (val == 1) fprintf_filtered (stream, ".TRUE."); else /* Not a legitimate logical type, print as an integer. */ { /* Bash the type code temporarily. */ TYPE_CODE (type) = TYPE_CODE_INT; f_val_print (type, valaddr, 0, address, stream, recurse, options); /* Restore the type code so later uses work as intended. */ TYPE_CODE (type) = TYPE_CODE_BOOL; } } break; case TYPE_CODE_COMPLEX: type = TYPE_TARGET_TYPE (type); fputs_filtered ("(", stream); print_floating (valaddr, type, stream); fputs_filtered (",", stream); print_floating (valaddr + TYPE_LENGTH (type), type, stream); fputs_filtered (")", stream); break; case TYPE_CODE_UNDEF: /* This happens (without TYPE_FLAG_STUB set) on systems which don't use dbx xrefs (NO_DBX_XREFS in gcc) if a file has a "struct foo *bar" and no complete type for struct foo in that file. */ fprintf_filtered (stream, "<incomplete type>"); break; case TYPE_CODE_STRUCT: case TYPE_CODE_UNION: /* Starting from the Fortran 90 standard, Fortran supports derived types. */ fprintf_filtered (stream, "( "); for (index = 0; index < TYPE_NFIELDS (type); index++) { int offset = TYPE_FIELD_BITPOS (type, index) / 8; f_val_print (TYPE_FIELD_TYPE (type, index), valaddr + offset, embedded_offset, address, stream, recurse, options); if (index != TYPE_NFIELDS (type) - 1) fputs_filtered (", ", stream); } fprintf_filtered (stream, " )"); break; default: error (_("Invalid F77 type code %d in symbol table."), TYPE_CODE (type)); } gdb_flush (stream); return 0; } static void list_all_visible_commons (char *funname) { SAVED_F77_COMMON_PTR tmp; tmp = head_common_list; printf_filtered (_("All COMMON blocks visible at this level:\n\n")); while (tmp != NULL) { if (strcmp (tmp->owning_function, funname) == 0) printf_filtered ("%s\n", tmp->name); tmp = tmp->next; } } /* This function is used to print out the values in a given COMMON block. It will always use the most local common block of the given name */ static void info_common_command (char *comname, int from_tty) { SAVED_F77_COMMON_PTR the_common; COMMON_ENTRY_PTR entry; struct frame_info *fi; char *funname = 0; struct symbol *func; /* We have been told to display the contents of F77 COMMON block supposedly visible in this function. Let us first make sure that it is visible and if so, let us display its contents */ fi = get_selected_frame (_("No frame selected")); /* The following is generally ripped off from stack.c's routine print_frame_info() */ func = find_pc_function (get_frame_pc (fi)); if (func) { /* In certain pathological cases, the symtabs give the wrong function (when we are in the first function in a file which is compiled without debugging symbols, the previous function is compiled with debugging symbols, and the "foo.o" symbol that is supposed to tell us where the file with debugging symbols ends has been truncated by ar because it is longer than 15 characters). So look in the minimal symbol tables as well, and if it comes up with a larger address for the function use that instead. I don't think this can ever cause any problems; there shouldn't be any minimal symbols in the middle of a function. FIXME: (Not necessarily true. What about text labels) */ struct minimal_symbol *msymbol = lookup_minimal_symbol_by_pc (get_frame_pc (fi)); if (msymbol != NULL && (SYMBOL_VALUE_ADDRESS (msymbol) > BLOCK_START (SYMBOL_BLOCK_VALUE (func)))) funname = SYMBOL_LINKAGE_NAME (msymbol); else funname = SYMBOL_LINKAGE_NAME (func); } else { struct minimal_symbol *msymbol = lookup_minimal_symbol_by_pc (get_frame_pc (fi)); if (msymbol != NULL) funname = SYMBOL_LINKAGE_NAME (msymbol); else /* Got no 'funname', code below will fail. */ error (_("No function found for frame.")); } /* If comname is NULL, we assume the user wishes to see the which COMMON blocks are visible here and then return */ if (comname == 0) { list_all_visible_commons (funname); return; } the_common = find_common_for_function (comname, funname); if (the_common) { if (strcmp (comname, BLANK_COMMON_NAME_LOCAL) == 0) printf_filtered (_("Contents of blank COMMON block:\n")); else printf_filtered (_("Contents of F77 COMMON block '%s':\n"), comname); printf_filtered ("\n"); entry = the_common->entries; while (entry != NULL) { print_variable_and_value (NULL, entry->symbol, fi, gdb_stdout, 0); entry = entry->next; } } else printf_filtered (_("Cannot locate the common block %s in function '%s'\n"), comname, funname); } /* This function is used to determine whether there is a F77 common block visible at the current scope called 'comname'. */ #if 0 static int there_is_a_visible_common_named (char *comname) { SAVED_F77_COMMON_PTR the_common; struct frame_info *fi; char *funname = 0; struct symbol *func; if (comname == NULL) error (_("Cannot deal with NULL common name!")); fi = get_selected_frame (_("No frame selected")); /* The following is generally ripped off from stack.c's routine print_frame_info() */ func = find_pc_function (fi->pc); if (func) { /* In certain pathological cases, the symtabs give the wrong function (when we are in the first function in a file which is compiled without debugging symbols, the previous function is compiled with debugging symbols, and the "foo.o" symbol that is supposed to tell us where the file with debugging symbols ends has been truncated by ar because it is longer than 15 characters). So look in the minimal symbol tables as well, and if it comes up with a larger address for the function use that instead. I don't think this can ever cause any problems; there shouldn't be any minimal symbols in the middle of a function. FIXME: (Not necessarily true. What about text labels) */ struct minimal_symbol *msymbol = lookup_minimal_symbol_by_pc (fi->pc); if (msymbol != NULL && (SYMBOL_VALUE_ADDRESS (msymbol) > BLOCK_START (SYMBOL_BLOCK_VALUE (func)))) funname = SYMBOL_LINKAGE_NAME (msymbol); else funname = SYMBOL_LINKAGE_NAME (func); } else { struct minimal_symbol *msymbol = lookup_minimal_symbol_by_pc (fi->pc); if (msymbol != NULL) funname = SYMBOL_LINKAGE_NAME (msymbol); } the_common = find_common_for_function (comname, funname); return (the_common ? 1 : 0); } #endif void _initialize_f_valprint (void) { add_info ("common", info_common_command, _("Print out the values contained in a Fortran COMMON block.")); if (xdb_commands) add_com ("lc", class_info, info_common_command, _("Print out the values contained in a Fortran COMMON block.")); }
Go to most recent revision | Compare with Previous | Blame | View Log