* c-exp.y (%union): Change dval to typed_val_float. Use DOUBLEST
to store actual data. Change types of INT and FLOAT tokens to
typed_val_int and typed_val_float respectively. Create new token
DOUBLE_KEYWORD to specify the string `double'. Make production
for FLOAT use type determined by parse_number. Add production for
"long double" data type.
* (parse_number): Use sscanf to parse numbers as float, double or
long double depending upon the type of typed_val_float.dval. Also
allow user to specify `f' or `l' suffix to explicitly specify
float or long double constants. Change typed_val to
typed_val_int.
* (yylex): Change typed_val to typed_val_int. Also, scan for
"double" keyword.
* coffread.c (decode_base_type): Add support for T_LNGDBL basic
type.
* configure, configure.in: Add check for long double support in
the host compiler.
* defs.h: Define DOUBLEST appropriatly depending on whether
HAVE_LONG_DOUBLE (from autoconf) is defined. Also, fix prototypes
for functions that handle this type.
* expression.h (union exp_element): doubleconst is now type
DOUBLEST.
* m2-exp.y f-exp.y (%union): dval becomes type DOUBLEST.
* findvar.c (extract_floating): Make return value be DOUBLEST.
Also, add support for numbers with size of long double.
* (store_floating): Arg `val' is now type DOUBLEST. Handle all
floating types.
* parser-defs.h parse.c (write_exp_elt_dblcst): Arg expelt is now
DOUBLEST.
* valarith.c (value_binop): Change temp variables v1, v2 and v to
type DOUBLEST. Coerce type of result to long double if either op
was of that type.
* valops.c (value_arg_coerce): If argument type is bigger than
double, coerce to long double.
* (call_function_by_hand): If REG_STRUCT_HAS_ADDR is defined, and
arg type is float and > 8 bytes, then use pointer-to-object
calling conventions.
* valprint.c (print_floating): Arg doub is now type DOUBLEST.
Use appropriate format and precision to print out floating point
values.
* value.h: Fixup prototypes for value_as_double,
value_from_double, and unpack_double to use DOUBLEST.
* values.c (record_latest_value): Remove check for invalid
floats. Allow history to store them so that people may examine
them in hex if they want.
* (value_as_double unpack_double): Change return value to DOUBLEST.
* (value_from_double): Arg `num' is now DOUBLEST.
* (using_struct_return): Use RETURN_VALUE_ON_STACK macro (target
specific) to expect certain types to always be returned on the stack.
* Makefile.in (SFILES): Add bcache.c.
(symtab_h): Add bcache.h.
(HFILES_NO_SRCDIR): add bcache.h
(COMMON_OBJS): Add bcache.o
(bcache.o): New target.
* dbxread.c (start_psymtab): Make global_syms & static_syms
type "partial_symbol **".
* hpread.c (hpread_start_symtab): Ditto.
* os9kread.c (os9k_start_psymtab): Ditto.
* stabsread.h (start_psymtab): Ditto.
* {symfile.c, symfile.h} (start_psymtab_common): Ditto.
* maint.c (maintenance_print_statistics): Call
print_symbol_bcache_statistics.
* objfiles.c (allocate_objfile): Initialize psymbol bcache malloc
and free pointers.
* solib.c (allocate_rt_common_objfile): Ditto.
* symfile.c (reread_symbols): Ditto.
(free_objfile): Free psymbol bcache when objfile is freed.
(objfile_relocate): Use new indirect psymbol pointers.
* objfiles.h (struct objfile): Add psymbol cache.
* symfile.c (compare_psymbols): Now passed pointers to pointers to
psymbols.
(reread_symbols): Free psymbol bcache when freeing other objfile
resources.
(add_psymbol_to_list, add_psymbol_addr_to_list): Initialize new
psymbol using the psymbol bcache.
(init_psymbol_list): Psymbol lists now contain pointers rather than
the actual psymbols.
* symfile.h (psymbol_allocation_list): Psymbol lists now dynamically
grown arrays of pointers.
(ADD_PSYMBOL_VT_TO_LIST): Initialize new symbol using the psymbol
bcache.
* symmisc.c (print_partial_symbols): Now takes pointer to pointer
to partial symbol.
(print_symbol_bcache_statistics): New function to print per objfile
bcache statistics.
(print_partial_symbol, print_partial_symbols,
maintenance_check_symtabs, extend_psymbol_list):
Account for change to pointer to pointer to partial symbol.
* symtab.c (find_pc_psymbol, lookup_partial_symbol, decode_line_2,
make_symbol_completion_list):
Account for change to pointer to pointer to partial symbol.
* symtab.h (bcache.h): Include.
* xcoffread.c (xcoff_start_psymtab): Make global_syms & static_syms
type "partial_symbol **".
character is one of the commonly used C++ marker characters.
* defs.h (is_cplus_marker): Add prototype.
* c-typeprint.c (c_type_print_base), ch-lang.c (chill_demangle),
cp-valprint.c (cp_print_class_method), mdebugread.c (parse_symbol),
stabsread.c (define_symbol, read_member_functions, read_struct_fields),
symtab.h (OPNAME_PREFIX_P, VTBL_PREFIX_P, DESTRUCTOR_PREFIX_P),
values.c (vb_match): Use is_cplus_marker instead of comparison
with CPLUS_MARKER.
allocate_saved_f77_common_node, allocate_common_entry_node,
add_common_block): Use xmalloc rather than malloc, some of which
were unchecked.
* gnu-regex.c: At same point as other gdb specific changes
#undef malloc and then #define it to xmalloc.
* ch-exp.c (growbuf_by_size): Use xmalloc/xrealloc rather than
bare unchecked calls to malloc/realloc.
* stabsread.c (dbx_lookup_type): Use xmalloc rather than bare
unchecked call to malloc.
double-clicking will work again.
(create_asm_win): Put "break" at end of all B1 bindings.
(create_file_win): Lower "sel" tag, don't raise it.
(ensure_line_visible): New proc.
(update_listing, update_assembly): Use it.
(create_copyright_window): Destroy window on Leave event.
(create_command_window): Put "break" at end of all B2 bindings.
AT_name tag is missing.
* procfs.c (procfs_stopped_by_watchpoint): Fix logic when
FLTWATCH and FLTKWATCH are defined.
* remote.c (remote_read_bytes): Advance memaddr for transfers,
return number of bytes transferred for partial reads.
* top.c (init_signals): Reset SIGTRAP to SIG_DFL.
here from config/i386/xm-i386mach.h, fix name.
* config/i386/nm-sun386.h: Ditto, from config/i386/xm-sun386.h.
* config/i386/nm-ptx4.h (CHILD_PREPARE_TO_STORE): Move to
here from config/i386/xm-ptx4.h.
* config/i386/nm-ptx4.h: Ditto, from config/i386/xm-ptx.h.
* config/i386/nm-symmetry.h: Ditto, from config/i386/xm-symmetry.h.
* config/m68k/nm-sun3.h: Ditto, from config/m68k/xm-sun3.h.
more housekeeping
* config/sparc/nm-nbsd.h (CHILD_PREPARE_TO_STORE): Move to here
from config/sparc/xm-nbsd.h.
* config/sparc/nm-sun4os4 (CHILD_PREPARE_TO_STORE): Ditto, from
config/sparc/xm-sparc.h.
* config/sparc/nm-sun4sol2.h: New file, renamed from nm-sysv4.h.
(PRSVADDR_BROKEN): Move here from xm-sun4sol2.h.
* config/sparc/sun4sol2.mh (NAT_FILE): Update.
* config/sparc/xm-sun4os4.h (HAVE_TERMIOS): Remove.
housekeeping
the length of an array type (see comment).
(chill_val_print (case TYPE_CODE_ARRAY)): If the length of an
array type is zero, call calculate_array_length.
* gdbtypes.c (get_discrete_bounds (case TYPE_CODE_ENUM)): They values
may not be sorted. Scan all entries and set the real lower and
Also check for gregset_t and fpregset_t types.
* configure: Regenerate.
* core-regset.c (sys/procfs.h): Only include if HAVE_SYS_PROCFS_H
is defined.
(fetch_core_registers): Turn into stub unless both HAVE_GREGSET_T
and HAVE_FPREGSET_T are defined. These changes allow systems
like linux that are migrating to /proc support to use a single
configuration for both new and old versions.
* config/i386/linux.mt: Note that this is now for both a.out and
ELF systems.
* config/i386/linux.mh (NATDEPFILES): Add solib.o, core-regset.o,
i386v4-nat.o
* config/i386/tm-linux.h (tm-sysv4.h): Include.
* config/i386/xm-linux.h (solib.h): Include
(SVR4_SHARED_LIBS): Define.
* i386v4-nat.c: Only compile if HAVE_SYS_PROCFS_H is defined.
(supply_gregset, fill_gregset): Compile if HAVE_GREGSET_T defined.
(supply_fpregset, fill_fpregset): Compile if HAVE_FPREGSET_T defined.
(COERCE_ENUM): Don't COERCE_REF.
(COERCE_NUMBER): New macro (same as COERCE_ARRAY then COERCE_ENUM).
* valops.c (value_assign): Only do COERCE_ARRAY if internalvar (let
value_cast handle it otherwise); do *not* COERCE_ENUM either way.
* valarith.c: Use COERCE_NUMBER instead od COEREC_ARRAY.
Add COERCE_REF before COERCE_ENUM.
* values.c (value_as_long): Simplify.
This fixes PR chill/8869.
* valops.c (value_array): Create internalvar if !c_style_arrays.
All references changed.
* breakpoint.c (bpstat_what): Add shlib_event to the class types.
Update state table. Reformat so that it's still readable.
When we hit the shlib_event breakpoint, set the calss of shlib_event.
(breakpoint_1): Add "shlib events" as a breakpoint type.
Print the shlib_event breakpoint like other breakpoints.
(create_solib_event_breakpoint): New function.
(breakpoint_re_set_one): Handle solib_event breakpoints.
* breakpoint.h (enum bytype): Add bp_shlib_event breakpoint type.
(enum bpstat_what_main_action): Add BPSTAT_WHAT_CHECK_SHLIBS
action.
(create_solib_event_breakpoint): Declare.
* infrun.c (wait_for_inferior): Handle CHECK_SHLIBS bpstat.
(normal_stop): Inform the user when the inferior stoped due
to a shared library event.
(_initialize_infrun): Add new set/show variable "stop_on-solib-events"
to control whether or not gdb continues the inferior or stops it when
a shared library event occurs.
* minsyms.c (lookup_minimal_symbol_solib_trampoline): New function.
* somsolib.c (TODO list): Update.
(som_solib_create_inferior_hook): Arrange for gdb to be notified
when significant shared library events occur.
* hppa-tdep.c (find_unwind_entry): No longer static.
First cut at the machine independent changes for 7363. Also includes
code to automatically track shl_load/shl_unload calls on hpux.
fetch_core_registers.
(hp300ux_core_fns): Remove, is unused.
(_initialize_core_hp300ux): Remove, is unused.
(gdbcore.h): Remove #include, no longer needed.
* remote-mips.c (pmon_opn, pmon_wait, pmon_makeb64, pmon_zeroset,
pmon_checkset, pmon_make_fastrec, pmon_check_ack,
pmon_load_fast): New functions. Support for the PMON monitor world.
(common_open): New function to merge support for different monitors.
(mips_open): Use common_open().
(mips_send_command): New function.
(mips_send_packet): Scan out-of-sequence packets.
(mips_enter_debug, mips_exit_debug): New functions.
(pmon_ops): New target definition structure.
These changes provide the initial support for the PMON (3.0.83)
monitor that appears on the Cogent (CMA101) Vr4300 development
board. The changes do not affect the IDT monitor support, other than
in the area of dealing with out-of-sequence packets.
for TYPE_CODE_SET and TYPE_CODE_BITSTRING too.
* eval.c (evaluate_subexp_standard): When known, use the formal
parameter type as the expected type when evaluating arg expressions.
* ch-lang.c (evaluate_subexp_chill): Likewise (for MULTI_SUBSCRIPT).
This (with a fix to gcc/config/sparc/sparc.h) fixes PR chill/8742.
* configure.in, configure: Recognize rs6000-*-aix4*.
* config/powerpc/xm-aix.h: Reduce to include "xm-aix4.h".
* config/rs6000/aix4.mh (XM_FILE): Point to xm-aix4.h.
* config/rs6000/xm-aix4.h: New file.
* config/xm-aix4.h: New file.
From the Mentor work...enable configure & build on rs6000-ibm-aix4.1.*.
Note this is an unbelievably buggy debugger on this host.
* gdbserver/low-linux.c: New file.
* remote.c (remote_read_bytes): Fix aborts on larger packets.
* config/i386/linux.mh (GDBSERVER_DEPFILES, GDBSERVER_LIBS):
Define.
* stabsread.c (define_symbol): If register value is too large,
tell what it is and what max is.
arguments. Sorting is now dependent on OBJF_REORDERED. All
callers/references changed.
* dbxread.c (read_ofile_symtab): Correctly determine value for
last_source_start_addr for reordered executables.
(process_one_symbol): Handle N_FUN with no name as an end of
function marker.
* partial-stab.h (case N_FN, N_TEXT): Don't assume CUR_SYMBOL_VALUE
is the high text address for a psymtab.
(case N_SO): Likewise.
(case N_FUN): Handle N_FUN with no name as an end of function
marker.
* minsyms.c (lookup_minimal_symbol_by_pc): Examine all symbols
at the same address rather than a random subset of them.
* coffread.c (coff_symfile_init): Set OBJF_REORDERED.
* elfread.c (elf_symfile_init): Similarly.
* somread.c (som_symfile_init): Similarly.
* xcoffread.c (xcoff_symfile_init): Similarly.
Support for debugging reordered executables. Remaining mentor vm
changes.
(dump_symtab): Change RETURN_MASK_ERROR to RETURN_MASK_ALL so
that catch_errors doesn't get blindsided by QUIT and lose the
cleanup chain. This fixes a problem where ^C while in a
user-defined command sometimes leaves instream NULL and causes a
segfault in command_loop.
that hardware watchpoint removal problems won't leave breakpoint
traps in the target.
* configure configure.in: Make --enable-gdbtk be the default.
* remote-e7000.c (e7000_insert_breakpoint,
e7000_remove_breakpoint): Use e7000 based breakpoints, not memory
breakpoints.
* (e7000_wait): Adjust PC back by two when we see a breakpoint to
compensate for e7000 maladjustment.
* sparcl-tdep.c (sparclite_check_watch_resources): Fix logic bug
which prevented hardware watchpoints from working.
that use ptrace(). This is not as useful as doing a full dump in a
human readable format, but is better than nothing when you are trying
to verify what is actually in the user struct.
* infptrace.c (udot_info): New function.
(PT_*): Define each individually if that one is not defined.
Update copyright to 1996.
* rs6000-nat.c (kernel_u_size): New function
Include <sys/user.h> for "struct user"
Update copyright to 1996.
* alpha-nat.c (kernel_u_size): New function.
Include <sys/user.h> for "struct user"
Update copyright to 1996.
* sparc-nat.c (kernel_u_size): New function.
Include <sys/user.h> for "struct user"
Update copyright to 1996.
* i386b-nat.c (kernel_u_size): New function.
Update copyright to 1996.
* i386v-nat.c (kernel_u_size): New function.
Update copyright to 1996.
* config/i386/nm-fbsd.h (KERNEL_U_SIZE): Define.
(kernel_u_size): Declare.
Update copyright to 1996.
* config/i386/nm-linux.h (KERNEL_U_SIZE): Define.
(kernel_u_size): Declare.
Update copyright to 1996.
* config/sparc/nm-sun4os4.h (KERNEL_U_SIZE): Define.
(kernel_u_size): Declare.
Update copyright to 1996.
* config/alpha/nm-osf2.h (KERNEL_U_SIZE): Define.
(kernel_u_size): Declare.
Update copyright to 1996.
* config/rs6000/nm-rs6000.h (KERNEL_U_SIZE): Define.
(kernel_u_size): Declare.
Update copyright to 1996.
from NUM_REGS to get number of general registers that we care about.
Update copyright to 1996.
* config/i386/tm-i386.h (REGISTER_BYTES): Define in terms
of number of general regs and number of floating point regs.
Update copyright to 1996.
collision in C++.
* infcmd.c (path_command): Use empty string if PATH name not set.
* i386-tdep.c (skip_trampoline_code): New function.
* srec.c: Renamed dsrec.c to avoid filename collision.
* Makefile.in: Cope with renaming.
Update copyrights to 1996.
* configure: Update.
* config/powerpc/macos.mh, config/powerpc/macos.mt,
config/powerpc/nm-macos.h, config/powerpc/tm-macos.h, mac-nat.c:
New files, native PowerMac debugging support.
* Makefile.in (mac-nat.o): Add build rule.
* mpw-config.in (enable_cflags): Add support.
(m68k-apple-macos, powerpc-apple-macos): Fix natdepfiles to
list object file instead of source file.
* mpw-make.sed (@ENABLE_CFLAGS@): Don't edit out, replace with
value of variable.
(install, install-only): Edit MPW-specific installation into
place of Unix shell code.
* mac-gdb.r: Fix version resources to use symbolic version strings.
(cfrg): New resource, code fragment for PowerMac.
* ch-exp.c (match_integer_literal): Fix long long support.
* gdbtypes.c (get_discrete_bounds): Make TYPE_LENGTH (type) ==
sizeof (LONGEST) case work OK.
* language.h (LA_BOOL_TYPE): New macro.
* eval.c (evaluate_subexp_standard) Use LA_BOOL_TYPE instead
of builtin_type_int where appropriate,
* valarith.c (value_subscript): Likewise.
* ch-exp.c: New file. Use recursive-descent.
Recognize labelled array tuples and powerset ranges.
* Makefile.in: Update for no longer using yacc for ch-exp.
* c-lang.c: Make various functions non-static.
* c-lang.h: Add bunches of prototypes.
* cp-valprint.c (cp_print_value_fields): Also take address.
(cp_print_value): Likewise. Use baselcass_offset.
* stabsread.c (current_symbol): New static variable.
(type_synonym_name): Remove.
(read_type): If copying, make copy be a TYPE_CODE_TYPEDEF.
(read_array_type): Don't need to handle undefined element type here.
(cleanup_undefined_types): Ditto.
(read_range_type): Look for Chill ranges.
* valops.c (value_assign): Fix case lval_internalvar - don't try
to assign into old value (which might be too small!).
(value_coerce_array): No longer need special VALUE_REPEATED handling.
(value_arg_coerce): Cleaner array->pointer decay mechanism.
(search_struct_field): Use baseclass_offset rather than
baseclass_addr.
(value_slice): Use get_discrete_bounds.
* value.h (COERCE_VARYING_ARRAY): Take type argumnt as well.
* values.c (baseclass_offset): Change parameter interface.
(baseclass_addr): Removed.
* c-typeprint.c, c-valprint.c, ch-valprint.c, values.c, valops.c:
Add check_typedef/CHECK_TYPEDEF as needed.
Add check_typedef/CHECK_TYPEDEF as needed.
* f-typeprint.c: Various cleaning up.
* valarith.c (value_subscript): Also subscript bitstrings (for Chill).
* typeprint.c (print_type_scalar): Also support TYPE_CODE_RANGE.
* eval.c (evaluate_subexp_standard case OP_ARRAY): Implement
support for labelled array tuples and ranges in powerset tuples.
(init_array_element): New function.
(check_typedef): New prototype.
(CHECK_TYPEDEF): New macro.
(TYPE_DUMMY_RANGE): Removed.
* gdbtypes.c (get_discrete_bounds): Fix paren error; make more robust.
(create_array_type): Don't force_to_range_type; users of the
array are responsible for handling non-range index types.
(create_set_type): Likewise.
(force_to_range_type): Removed.
(check_typedef): New function handles stub types and typedefs.
(check_stub_type): Just call check_typedef. (To be removed.)
(recursive_dump_type): Handle TYPE_CODE_TYPEDEF.
* ch-lang.c (type_lower_upper): Use get_discrete_bounds.
(evaluate_subexp_chill): Handle string repetition.
Re-arrange to handle EVAL_AVOID_SIDE_EFFECTS better.
* ch-typeprint.c (chill_type_print_base): Handle TYPE_CODE_TYPEDEF.
Pass show=0 in recursive calls various places.
(case TYPE_CODE_ARRAY): Don't require index type to have
TYPE_CODE_RANGE.
(case TYPE_CODE_RANGE): Don't need to support TYPE_DUMMY_RANGE.
* gdbtypes.c, ch-lang.c, ch-typeprint.c (numerous places):
Add check_typedef/CHECK_TYPEDEF as needed.
(CC_FOR_TARGET, CXX_FOR_TARGET): Use it to find target libraries.
* configure.in (X_CFLAGS): Fix typo.
(target_subdir): Set to "${target_alias}/" if cross.
* configure.in: Regenerated.
Recognize enums from alpha cc -migrate.
(upgrade_type): Pass correct fd to parse_symbol when parsing
the index type of an array.
(parse_procedure, parse_lines, psymtab_to_symtab_1): Handle
unsorted procedure descriptors from Irix 5.x and Alpha OSF/1-3.x
shared libraries. Use CORE_ADDR instead of `unsigned long' in
procedure descriptor address computations.
* symtab.c (decode_line_1): Prevent accidental strchr match
of a null character with the terminating null character of
gdb_completer_quote_characters.
(cplusplus_hint): Make sure that only a single quote is printed
in the hint message.
Also force HAVE_MMAP to be defined since autoconf is currently broken
for detecting a working mmap under hpux.
* config/pa/xm-hppah.h (MMAP_BASE_ADDRESS): Tweak MMAP_BASE_ADDRESS
to a better value suggested by Jeffrey A Law (law@cygnus.com).
trampolines. These are pieces of code between the caller and the
callee that figure out the address of the callee's code at run
time. Upon entry, we can't figure out the callee's address, so we
set a breakpoint within the trampoline where the address will be
known, and continue the target. Once we hit the breakpoint, we
break at the callee's address and proceed as usual.
start-sanitize-gm
* config/tm-magic.h: Define DYNAMIC_TRAMPOLINE_NEXTPC. Turn off
IN_SOLIB_CALL_TRAMPOLINE and IN_SOLIB_RETURN_TRAMPOLINE.
end-sanitize-gm
symbol tables so that it is obvious that they are not supported on
this particular machine rather than implying they are not supported
at all in this version of gdb.
* solib.h (PC_SOLIB): New macro; define using solib_address.
* stack.c (print_frame_info) [PC_SOLIB]: If no function name, try
PC_SOLIB on the PC value.
S-record download code into srec.c.
* srec.c srec.h: New files. Contain S-record loading routines
formerly in monitor.c.
* serial.c serial.h: New routine just like fprintf, but uses
serial_t instead of FILE *.
* sh-tdep.c (frame_find_saved_regs init_extra_frame_info):
Don't add four to saved pc (makes things match manual). Also, fix
bug where we didn't get pc from stack frame correctly.
* config/sh/tm-sh.h (SAVED_PC_AFTER_CALL): Don't add four to
saved pc. Real hardware does this for you.
* sh3-rom.c (sh3_load): New routine. Sets up for download then
calls generic S-record loader.
* config/h8300/h8300.mt, config/h8500/h8500.mt,
config/m68k/monitor.mt, config/pa/hppapro.mt, config/sh/sh.mt:
Add srec.o to TDEPFILES.
NOTE: Only the BigEndian build has the simulator included, because a
LittleEndian simulator is not constructed at the moment (it should
work, but has not been tested fully).
Jamie. Basically had case statement in the wrong place...
* (mips_load): Remove unnecessary `db tty0' command. It's all
handled by mips_initialize now.
monitor board.
The mips_initialize() code seemed to have a "funny" switch
statement. I also removed an unnecessary switch into debug monitor
mode just before the call to mips_initialize() in mips_load().
Stu Grossman
Fred Fish
Stan Shebs
Tom Tromey
Per Bothner
Steve Chamberlain
Peter Schauer
Jeff Law
Wilfried Moser
David Edelsohn
Jackie Smith Cashion
Brendan Kehoe
Michael Meissner
Rob Savoye
Raymond Jou
Jason Molenda