* sparclet-stub.c: a few clean-ups, removing dead code
* config/sparc/tm-sparclet.h: make registers ASR15, ASR19 invisible
(they're not useful, you can't change, write or even read them)
* sparclet-rom.c: Remove includes of Unix system files.
Add function "sparclet_supply_register" so that parse_register_dump
will not seg-fault by calling a null function pointer.
Remove XMODEM support (unfinished work?).
Remove flag "MO_HANDLE_NL", so monitor's output can be read by humans.
Add fill command.
Remove colon from getreg.resp_delim so PSR register will work.
Remove pointer to sparclet_load (downloading SREC's doesn't work).
Null out local register names for %g0, all %cc and all %asr regs,
since the monitor can't report them. Will return zero instead.
* sparclet-stub.c: New -- remote protocol support for sparclet CPU.
* config/sparc/tm-sparclet.h: Re-arrange REGISTER_NAMES:
Add back %g0 and %psr, add %cc coprocessor regs, add %asr regs.
Adjust NUM_REGS and REGISTER_BYTES accordingly
from want and want_nopc.
(want_h8300s, want_nopc_h8300s): New variables for H8/S register
lists.
(e7000_fetch_registers): Use H8/300H or H8/S register list string
as needed.
(e7000_wait): Likewise.
Bringing over a change from hmse-h8s-beta branch.
interrupted by a signal, figure out whether the registers that
the proc_desc claims are saved have been saved yet.
(mips_push_dummy_frame): Write dummy frame register after all
registers have been saved in the dummy frame. Update comments
to reflect the fact that we are now using an AT_ENTRY_POINT
call dummy.
definition in inferior.h is sufficient.
* mips-tdep.c (mips_pc_in_call_dummy): Ditto.
(mips_push_arguments): Make sure that the stack is aligned to a
multiple of 8 after the arguments are pushed.
Structures are always passed by value in the old ABI.
Adjust argument register value on big endian targets when passing
a value whose length is less than the register size.
Write stack arguments with a single write_memory call.
(mips_pop_frame): Use frame_saved_regs instead of proc_desc to
decide which registers have to be restored.
destructor name with call to destructor_name_p.
* c-lang.h, c-typeprint.c (cp_type_print_method_args): Removed,
no longer needed.
* c-typeprint.c (c_type_print_varspec_prefix, c_type_print_base):
Replace remaining fprintf_unfiltered calls with their filtered variant.
(c_type_print_base): Do not print return type for destructors from
template classes.
Replace cp_type_print_method_args with cplus_demangle call to get
consistent type output for stubbed and unstubbed methods.
* cp-valprint.c (cp_print_class_method): Replace
cp_type_print_method_args with cplus_demangle call to get consistent
type output for stubbed and unstubbed methods.
* gdbtypes.c, gdbtypes.h (get_destructor_fn_field): New function
to find the destructor field indices for a type.
* gdbtypes.h (struct type): Clarify comments for vptr_basetype
and arg_types fields.
(struct fn_field): Remove args field, no longer used.
* symtab.c (decode_line_1), valops.c (value_struct_elt,
check_field_in): Use get_destructor_fn_field to find the destructor
field indices instead of assuming that the compiler passes the member
function fields in a specific order.
* symtab.c (find_methods): Pass NULL instead of SYMBOL_BLOCK_VALUE
to lookup_symbol.
(list_symbol): Replace cp_type_print_method_args with cplus_demangle
call in zapped out code and explain why this code is zapped out.
Minor changes to make pre-ANSI compilers happy.
* mips-tdep.c: Minor changes to make pre-ANSI compilers happy.
(mips_push_arguments): Rewrite to partially support EABI.
(mips_pc_in_call_dummy): New function.
* infcmd.c: Include symfile.h to get prototype of entry_point_address,
which fixes 64-bit sign extension bug on MIPS.
* remote-sim.c (gdb_os_error): New function.
(init_callbacks): Fix initializing of gdb_callback. Add gdb_os_error.
(gdb_os_printf_filtered): Use gdb_stdout, not stdout.
* config.in, configure.in, configure: Check for <objlist.h>.
* configure.tgt: Handle mips*-sgi-irix6* like irix5 for now.
* cp-valprint.c (cp_print_value_fields): Use SYMBOL_VALUE_ADDRESS
instead of SYMBOL_BLOCK_VALUE to get the address of a static member.
* dwarf2read.c: Turn warnings and recoverable errors into complaints,
add new complaints where appropriate.
gcc -Wall cleanup.
(struct line_head): Change line_base from char to int to avoid
problems with compilers whose plain char is represented by an
unsigned char.
(struct partial_die_info): Add is_declaration field.
(dwarf2_tmp_obstack): New obstack for allocating temporary storage
used during symbol reading.
(cu_header_offset): New variable for resolving relative reference
dies.
(optimized_out, basereg, islocal, frame_base_reg, frame_base_offset):
New interface variables for decode_locdesc.
(struct dwarf2_pinfo): New structure for communication between
psymtab and symtab reading, passed via pst->read_symtab_private.
(dwarf2_has_info, dwarf2_build_psymtabs): Accept objects files
without line number sections.
(dwarf2_build_psymtabs_hard): Initialize temporary obstack
for symbol reading.
Allocate and initialize pst->read_symtab_private.
Relocate pst->textlow and pst->texthigh with baseaddr.
(scan_partial_symbols): Do not add DW_AT_declaration symbols
to the partial symbol table.
Add file scope enumerator symbols to the partial symbol table.
Fix typo in highpc computation.
If we didn't find a lowpc, set it to highpc to avoid complaints
from `maint check.
(add_partial_symbol): Relocate symbol values with baseaddr.
Add static DW_TAG_subprogram and DW_TAG_variable symbols to the
minimal symbol table.
Obtain symbol values for DW_TAG_variable symbols from the location
descriptor, skip symbols with missing location desciptors.
Skip symbols for aggregate types without children.
Handle enumerator symbols.
(dwarf2_psymtab_to_symtab): Issue symbol reading message if verbose.
(psymtab_to_symtab_1): Set local variables from
pst->read_symtab_private, set cu_header_offset and baseaddr.
Initialize temporary obstack for symbol reading, initialize
buildsym and add a cleanup to really_free_pendings.
Relocate highpc with baseaddr when calling end_symtab.
If the compilation is from a C file generated by language
preprocessors, do not set the symtab language if it was already
deduced by start_subfile.
Removed verbose sorting symbol table message.
(process_die): Handle DW_TAG_ptr_to_member_type and
DW_TAG_reference_type.
Use read_subroutine_type to get the function type for
DW_TAG_subprogram before calling read_func_scope.
(read_file_scope): Initialize file name to <unknown>, start_subfile
expects a non-NULL name.
If we didn't find a lowpc, set it to highpc to avoid complaints
from finish_symbol.
Relocate lowpc and highpc with baseaddr.
Get rid of Irix6.2 native cc compile machine prefix in comp_dir.
Zero out ftypes for each new compilation unit (may be different
language or different objfile).
Accept compilation units without line number information, pass
comp_dir to decode_lines.
(read_func_scope): Initialize function name to <unknown> to avoid
core dumps when DW_AT_name is missing.
Relocate lowpc and highpc with baseaddr.
Handle DW_AT_frame_base, keep result for DW_OP_fbreg operations.
Pass function type to new_symbol.
(read_lexical_block_scope): Relocate lowpc and highpc with baseaddr.
(read_structure_scope): Set TYPE_TAG_NAME, not TYPE_NAME.
Handle DW_TAG_class_type.
Copy fields to type_obstack, release temporary storage for fields.
Don't add symbol if die is a stub die and has no children.
Handle C++ static member fields.
(read_enumeration): Set TYPE_TAG_NAME, not TYPE_NAME.
Copy fields to type_obstack, release temporary storage for fields.
Let new_symbol handle the symbol creation for enumerators
instead of handcrafting a symbol.
Determine signedness of enum type from enumerators.
(dwarf_read_array_type): Handle variable length arrays.
Use lookup_pointer_type instead of handcrafting a type.
Create array type only if a DW_TAG_subrange_type was found.
(read_tag_pointer_type, read_tag_reference_type):
Use lookup_pointer_type and lookup_reference_type instead
of handcrafting a type.
(read_tag_ptr_to_member_type): New function to handle
DW_TAG_ptr_to_member_type.
(read_subroutine_type): Handle parameter dies.
Use lookup_function_type instead of handcrafting a type.
(read_typedef): Allocate a TYPE_CODE_TYPEDEF type for the typedef.
(read_base_type): If the type has a name, use init_type to create
a new type instead of second guessing a fundamental type.
(read_comp_unit): Reset die reference table before building
a new one.
(dwarf2_read_section): Read section contents into psymbol_obstack.
(dwarf2_read_abbrevs): Handle unterminated abbreviations
for a compile unit gracefully.
(read_partial_die): Zero partial die before reading its info.
Handle DW_AT_declaration.
Fix typo in handling of DW_FORM_block4.
(read_full_die): Fix typo in handling of DW_FORM_block4.
(read_1_signed_byte, read_2_signed_bytes, read_4_signed_bytes):
New routines to get signed values from a buffer.
(read_n_bytes, read_string): Allocate storage from the temporary
obstack. If the host char size permits it, return pointer
to buffer instead of allocating storage.
(set_cu_language): Handle DW_LANG_Mips_Assembler.
(dwarf_attr): Return NULL if reference die for DW_AT_specification
or DW_AT_abstract_origin die is not found.
(record_minimal_symbol): Removed, replaced with a direct call to
prim_record_minimal_symbol, it now handles saving the string itself.
(convert_locdesc): Removed, partial symtab reading now uses
decode_locdesc.
(dwarf_attr): Use dwarf2_get_ref_die_offset to get the absolute
offset for the die reference.
(dwarf_decode_lines): Complain if the line section info is missing.
Use read_1_signed_byte to extract lh.line_base to avoid
problems with compilers whose plain char is represented by an
unsigned char.
Add cleanups for allocated temporary storage.
Start a subfile for the first file in the state machine.
Fix off by one problem with dirs.dirs access.
Use comp_dir when directory index is 0.
Support multiple sequences (from Jason Merrill <jason@cygnus.com>).
(dwarf2_start_subfile): Try to keep line numbers from identical
absolute and relative file names in a common subfile.
(new_symbol): Allocate symbol and symbol name on the symbol_obstack.
Set SYMBOL_LINE from DW_AT_decl_line if present.
Set SYMBOL_TYPE from passed type if not NULL.
Change DW_TAG_variable symbol types with missing type entries
to a sensible type.
Handle optimized_out, offreg and islocal storage classes.
Add external symbols with type information whose address isn't
known as LOC_UNRESOLVED symbols.
Synthesize typedefs for C++ classes, structs, unions and enumerations.
Handle DW_TAG_enumerator symbols, complain for unrecognized
symbol tags.
(die_type): A missing DW_AT_type represents a void type.
Use dwarf2_get_ref_die_offset to get the absolute offset for
the die reference.
(die_containing_type): New function to build type from
DW_AT_containing_type attribut.
(read_type_die): Handle DW_TAG_ptr_to_member_type.
Treat DW_TAG_subprogram like DW_TAG_subroutine_type.
(dwarf_base_type): Fix typo with creation of FT_UNSIGNED_SHORT
fundamental type.
(create_name): Removed, symbol name allocation is now done
in new_symbol.
(dump_die): Use print_address_numeric to print a CORE_ADDR.
(dwarf2_empty_die_ref_table): New function to clear the die
reference table.
(dwarf2_get_ref_die_offset): New function to get the absolute
die offset from a die reference attribute.
(decode_locdesc): Complete rewrite using a stack, code mostly
borrowed from dwarfread.c:locval.
(dwarf_alloc_type): Removed, replaced by direct calls to alloc_type.
(dwarf_alloc_block): Allocate block on temporary obstack.
* elfread.c (elf_symtab_read): When handling Irix dynamic symbols,
skip section name symbols and relocate all others.
(elf_symfile_read): Build dwarf2 psymtab even if offset is non-zero.
* irix5-nat.c (fetch_core_registers): Handle core_reg_sect
from N32 executables. Call registers_fetched after extracting
the registers.
(obj_list_variant, struct link_map, LM_OFFSET, LM_ADDR): New
definitions to enable support of O32 and N32 format objlists.
(struct so_list): New members offset, so_name and lmstart to
eliminate dependencies from the objlist format used.
(solib_map_sections, symbol_add_stub, solib_add,
info_sharedlibrary_command, solib_address, clear_solib): Use
so_name and LM_OFFSET.
(first_link_map_member): Rewrite to enable support of O32 and N32
format objlists.
(next_link_map_member, xfer_link_map_member): New functions to
support O32 and N32 format objlists.
(find_solib): Use first_link_map_member, next_link_map_member and
xfer_link_map_member.
(solib_create_inferior_hook): Use TARGET_SIGNAL_* instead of
host signal numbers.
* mdebugread.c (parse_partial_symbols, handle_psymbol_enumerators):
Pass CORE_ADDR variant to add_psymbol_to_list.
* mips-tdep.c (heuristic_proc_desc): Stop examining the prologue
if we encounter a positive stack adjustment. Handle `move $30,$sp'.
Handle `sd reg,offset($sp)' for 32 bit ABIs.
* symmisc.c (dump_msymbols, print_partial_symbols): Use
print_address_numeric to print a SYMBOL_VALUE_ADDRESS.
(dump_symtab): Print compilation directory if it is not NULL.
* valops.c (search_struct_field, value_struct_elt_for_reference):
Use SYMBOL_VALUE_ADDRESS instead of SYMBOL_BLOCK_VALUE to get the
address of a static member.
Added some generic target-independant code for managing call-dummy frames.
Wed Nov 27 10:32:14 1996 Michael Snyder <msnyder@cleaver.cygnus.com>
* breakpoint.c: DELETE command will not delete CALL_DUMMY breakpoint.
* blockframe.c: Add target-independant support for managing
CALL_DUMMY frames on the host side.
* frame.h: Declarations for generic CALL_DUMMY frame support.
* h8300-tdep.c: Add target function calls using generic frame support.
* config/h8300/tm-h8300.h: config for generic target function calls.
start-sanitize-m32r
* m32r-tdep.c: Add target function calls using generic frame support.
* config/m32r/tm-m32r.h: config for generic target function calls.
end-sanitize-m32r
* sh-tdep.c: Add target function calls using generic frame support.
* config/sh/tm-sh.h: config for generic target function calls.
start-sanitize-v850
* v850-tdep.c: Add target function calls using generic frame support.
* config/v850/tm-v850.h: config for generic target function calls.
end-sanitize-v850
* valops.c: ADD PUSH_RETURN_ADDRESS so that it doesn't have to be
done by PUSH_ARGUMENTS when there's no CALL_DUMMY.
(mips_addr_bits_remove): Declare.
* mips-tdep.c (mips_push_dummy_frame): Fix heuristic-fence-post
errors when hitting breakpoints during inferior function calls
in 64-bit programs.
(fix_sign_extension): Make public, rename to mips_addr_bits_remove.
* utils.c (paddr_nz, preg_nz): New functions, similar to
paddr and preg but don't print leading zeroes.
* defs.h (paddr_nz, preg_nz): Declare.
* remote-mips.c: Use paddr_nz instead of paddr throughout
to reduce packet size.
(pmon_end_download): Improve timeout error handling.
* infcmd.c (do_registers_info): Call val_print with the
virtual buf instead of the raw buf. Needed for REGISTER_CONVERT
to work with non-floating point regs.
start-sanitize-d10v
* d10v-tdep.c (d10v_skip_prologue): If we have line debugging
information, then the end of the prologue should the first
assembly instruction of the first source line.
* values.c (value_from_longest): Put in D10V call to
fix up address pointers.
* config/d10v/tm-d10v.h (REGISTER_VIRTUAL_SIZE): Modified.
(REGISTER_VIRTUAL_TYPE): Modified for PC_REGNUM and SP_REGNUM.
(REGISTER_CONVERTIBLE): Make PC and SP convertible.
(REGISTER_CONVERT_TO_VIRTUAL): Define.
(REGISTER_CONVERT_TO_RAW): Define.
(D10V_MAKE_DADDR): Define.
(D10V_MAKE_IADDR): Define.
end-sanitize-d10v
* procfs.c (wait_fd): Handle EINTR error return from poll
by restarting the poll.
* defs.h (PIDGET): Define a default version that just
returns its argument unchanged.
* inflow.c (terminal_init_inferior): Eliminate #ifdef
of PIDGET and fold both alternatives into common code.
(pass_signal): Use PIDGET for pid passed to kill().
* sh-tdep.c: Add functionality for target function calls.
* config/sh/tm-sh.h: Add support for target function calls.
This is a safety check-in: everything works, and there'll be another
clean-up round shortly.
Tue Nov 12 12:06:58 1996 Michael Snyder <msnyder@cleaver.cygnus.com>
* m32r-tdep.c: Add functionality for target function calls.
* valops.c: Small change to support target function calls.
* config/m32r/tm-m32r.h: Add support for target function calls.
This is a safety checkin -- everything works, but there'll be another
clean-up round later.
* defs.h: modify Nov 11 12:59:00 change so _MSC_VER is checked
instead of _WIN32
* win32-nat.c: fix Nov 11 12:59:00 change (windows.h should
be included instead of windefs.h for compilers other than
VC++).
* mswin/windefs.h: remove ^Ms and change C++ style comments
to C style comments
* mips-tdep.c (get_frame_pointer, fix_sign_extension):
New functions to consolidate common code.
(mips_frame_chain, init_extra_frame_info): Use new functions
to fix problems with backtrace and finish commands on ddb board.
config/mips/tm-mips.h: Add/fix bugs for 64-bit mips support.
* defs.h: Cleanup; add prototypess
* corefile.c: Change FIXME #ifdef
* win32-nat.c: Include windefs instead of windows.h.
* utils.c: Add routines for printing addresses and registers
based on type size.
if we encounter a positive stack adjustment.
(find_proc_desc): If heuristic_fence_post is non-zero, use
heuristic_proc_start to determine the start of a function before
calling heuristic_proc_desc.
* coffread.c (coff_symtab_read): Change minimal symbol types
for C_LABEL symbols from mst_* to mst_file_*.
* config/m68k/sun3os4.mh (MMALLOC_CFLAGS): Define MMCHECK_FORCE to 1.
* configure.in: Handle error message from sun3 native ld when
configuring HLDFLAGS.
* configure: Regenerated with autoconf.
* c-valprint.c (c_value_print): Adjust value address by VALUE_OFFSET.
* cp-valprint.c (cp_print_value): Prevent gdb crashes by making sure
that the virtual base pointer from an user object still points to
accessible memory.
* dbxread.c (dbx_symfile_init): Initialize sym_stab_info to
clear the recently added header_files fields.
(dbx_symfile_finish): Free hfiles[i].vector to avoid storage leak.
* d10v-tdep.c: Fix some problems with inferior function calls.
* config/d10v/tm-d10v.h (EXTRA_FRAME_INFO): Change dummy to be
a pointer to the dummy's stack instead of just a flag.
missing or zero-sized .text sections properly.
* mdebugread.c: Handle scRConst and scSUndefined storage classes.
* stabsread.c (scan_file_globals): Try to resolve symbols
for shared libraries from the minimal symbol table of the main
executable first.
* m32r-tdep.c: Improved frame_chain and fn prologue analysis.
* configure.tgt: Add entry for m32r target.
* monitor.h: Add a flag to tell monitor_store_register to use
(val, regno) instead of (regno, val).
* monitor.c: Make monitor_store_register honor the above flag.
Make monitor_exp ignore DC1/DC3 for m32r.
Increase buf size in monitor_dump_regs.
the entry to the exit system call to detect termination of the
inferior stopped working. Trace termination of the inferior via
PRFS_STOPTERM instead.
(procfs_init_inferior): Do not trace entry to exit system call
if PIOCSSPCACT is defined.
(procfs_wait): Handle PR_DEAD event, which signals the termination
of the inferior if PRFS_STOPTERM is set.
* mdebugread.c (parse_partial_symbols): Ignore stNil section
start address symbols.
* sparc-tdep.c (get_saved_register): Get saved PC from the
frame info if not in innermost frame.
external forms may have different sizes. Don't assume that an
internal_nlist has the same layout as an external_nlist. Create
symbol for n_strx element so to hide specifics of nlist from
partial-stab.h.
* partial-stab.h: Don't reference dbxread symbols directly. Use
CUR_SYMBOL_STRX instead.
* config/i386/xm-windows.h: Define SIGQUIT and SIGTRAP.
* config/v850/tm-v850.h: Define PS_REGNUM and TARGET_V850 for
MSVC builds.
* mswin/gdbwin.c (reg_order): Define register order for V850.
* mswin/gui.cpp (CGuiApp::InitInstance): Define target name for
V850.
* mswin/regdoc.h: Define MAXREGS for V850.
prologues correctly. First, prologue end is now defined by
presence of a branch, jump or call insn. Second, can no longer
fix frame offsets because we may not know the offset until after a
register has been saved.
* (v850_init_extra_frame_info): Fixup frame offsets here because
we have all the info at this time.
* (v850_frame_chain): Use new calling convention for scan_prologue.
Tue Oct 22 10:25:29 1996 Martin M. Hunt <hunt@pizza.cygnus.com>
* d10v-tdep.c, config/d10v/tm-d10v.h: Changes to allow stack
backtraces and inferior function calls.
end-sanitize-d10v
an open core_bfd.
* cp-valprint.c (cp_print_value_fields): Pass correct address
to val_print, not 0.
From Andreas Schwab (schwab@issan.informatik.uni-dortmund.de):
* eval.c (evaluate_subexp_standard) [case BINOP_REPEAT]: Chase
typedefs before checking for integral type of right operand.
errors unique.
* testsuite/gdb.base/nodebug.exp: Whack out all -g options
explicitly.
* v850-tdep.c (v850_init_extra_frame_info v850_frame_chain): Fix
sign bugs with scanning prologues. Get a little smarter about
calculating the length of uninteresting instructions.
LOC_LOCAL_ARG as arguments so that GDB will know about function
args declared this way. Mostly affects dwarf.
* dwarfread.c (decode_die_type): Change default type from int to
void. This allows GDB to recognize void functions.
* (new_symbol): If AT_PROTOTYPED is present, set a flag in the
type structure.
* findvar.c (extract_floating store_floating): Clean up comments
to reflect reality.
* gdbtypes.h: Add TYPE_FLAG_PROTOTYPED so that we can tell if a
function has a prototype. Currently, only dwarf supports this.
* utils.c (floatformat_from_doublest): Fix logic error with
converting from double to float. (It wasn't shifting mant_long if
it had a hidden bit.)
* v850-tdep.c: Add support for function calling. Fix some
problems with debugging code w/o debug symbols.
* config/v850/tm-v850.h: Ditto.
struct frame to make it possible to use frame_saved_args in
EXTRA_FRAME_INFO macro.
* v850-tdep.c config/v850/tm-v850.h: Lotsa new functions and
macros to make frame operations (such as backtrace) work.
