Jim Kingdon [Wed, 10 Nov 1993 21:05:38 +0000 (21:05 +0000)]
* dbxread.c (process_one_symbol, N_RBRAC): Don't clear
within_function just because local_symbols is NULL. It appears
that this bug has existed since the 10 Apr 89 change which started
clearing within_function here.
Jim Kingdon [Wed, 10 Nov 1993 16:31:17 +0000 (16:31 +0000)]
* dstread.c (dst_sym_fns): Update for flavours.
* symfile.c (find_sym_fns): Add kludge for apollo like for rs/6000.
* dstread.c (dst_symfile_offsets): Set objfile->num_sections.
Jim Kingdon [Wed, 10 Nov 1993 15:58:48 +0000 (15:58 +0000)]
gcc -Wall lint:
* thread.c: Include "gdbcmd.h" and <ctype.h>.
* Makefile.in: Update dependency.
* thread.c (thread_command): Remove unused variable p.
* values.c (unpack_double): Use len instead of TYPE_LENGTH (type).
* valprint.c (print_floating): Correctly check sign bit now that
we are using unsigned arithmetic.
* symtab.c (find_pc_line_range): Remove unused variables exact_match,
ind, and l.
Jim Kingdon [Wed, 10 Nov 1993 01:34:39 +0000 (01:34 +0000)]
* Makefile.in (init.c): Add udi2go32.o to list of files that we
should not try to search for _initialize_* functions.
* remote-udi.c (udi_wait): Change UDIGdb_StdoutReady back to
UDIStdoutReady. It accidentally got changed on 24 Oct 93 when
stdout was changed to gdb_stdout. Likewise for UDIGetStdout,
UDIStderrReady, and UDIGetStderr.
Mark Eichin [Tue, 9 Nov 1993 22:07:50 +0000 (22:07 +0000)]
add patsubst %-lynxos to subst-from-hell in stamp-[34]stage-compared.
Maybe someday I'll learn enough of gnumake to turn this into something
reasonable...
* elfcode.h (elf_object_p): Rather than looking through an array
of architectures, get the ELF EM_xxx code from the backend
information. Let the generic ELF target match any EM_xxx code not
matched by another ELF target. Call elf_backend_object_p to let
the backend do more checks and set global information.
* libelf.h (struct elf_backend_data): Added elf_machine_code and
elf_backend_object_p fields.
(struct bfd_elf_arch_map): Removed.
(bfd_elf_arch_map, bfd_elf_arch_map_size): Don't declare.
* elf32-target.h, elf64-target.h: Initialize elf_machine_code
field with ELF_MACHINE_CODE. Initialize elf_backend_object_p
field with elf_backend_object_p (if it is defined).
* elf32-gen.c, elf32-hppa.c, elf32-i386.c, elf32-i860.c,
elf32-m68k.c, elf32-m88k.c, elf32-mips.c, elf32-sparc.c,
elf64-gen.c (ELF_MACHINE_CODE): Defined.
* elf32-mips.c: Include ecoffswap.h to get ECOFF swapping
routines.
(mips_elf_object_p): Set the right machine number.
(mips_elf_ecoff_debug_swap): Defined.
(elf_backend_object_p): Defined to be mips_elf_object_p.
(elf_backend_ecoff_debug_swap): Defined to be
mips_elf_ecoff_debug_swap.
* elf.c (bfd_elf_arch_map, bfd_elf_arch_map_size): Removed.
* elf32-mips.c (elf_mips_howto_table): Don't complain on overflow
for R_MIPS_26. Correct overflow detection requires matching the
upper four bits of the destination against the PC. From Ted Lemon
<mellon@pepper.ncd.com>.
* elf32-hppa.c (elf_hppa_reloc_type_lookup): Return type
should point to const data.
Jim Kingdon [Tue, 9 Nov 1993 18:28:04 +0000 (18:28 +0000)]
* .gdbinit: Remove "source /.gdbinit". It causes a spurious error
if /.gdbinit doesn't exist (and I know of no convention of putting
something in /.gdbinit).
* libelf.h: Added some comments.
(struct elf_backend_data): Added elf_backend_ecoff_debug_swap
field. Removed unused write_relocs field.
* elf32-target.h: Adjusted elf_backend_data initialization
accordingly.
* elf64-target.h: Corrected elf_backend_data initialization to
fill in all fields and to set elf_64_p to 1.
Jeff Law [Tue, 9 Nov 1993 08:51:02 +0000 (08:51 +0000)]
* config/tc-hppa.c (pa_type_args): Renamed from pa_export_args.
Accept new argument "is_export". All callers changed. When
processing a .export directive for a function, do not allow
the user to set the type to "CODE", instead warn and set the
type to "ENTRY".
* elfcode.h (bfd_section_from_shdr): Remove duplicate assignment
to filepos in SHT_STRTAB case.
(assign_file_position_for_section): Set BFD section filepos as
well as ELF section sh_offset.
Jim Kingdon [Mon, 8 Nov 1993 17:27:48 +0000 (17:27 +0000)]
* stabsread.c (read_one_struct_field): Don't give up on unknown
visibility character, just shove it in fip->list->visibility.
(read_baseclasses): Don't give up on unknown virtual or visibility
characters, just assume a reasonable default, complain, and keep
going.
(attach_fields_to_type): Complain on unrecognized visibility.
One result of all this is that '9' (VISIBILITY_IGNORE) can be used
in a stab as well as being something which GDB uses internally.
Jim Kingdon [Mon, 8 Nov 1993 17:02:08 +0000 (17:02 +0000)]
* stabsread.c: Remove long_kludge_name code (it was already #if 0).
* stabsread.c (read_one_struct_field): Clean up comments to reflect
Kung's change of 5 Nov 93.
Peter Schauer [Mon, 8 Nov 1993 07:53:45 +0000 (07:53 +0000)]
* alpha-opc.h, alpha-dis.c (print_insn_alpha): Add
FLOAT_MEMORY_FORMAT_CODE, FLOAT_BRANCH_FORMAT_CODE, correct
FLOAT_FORMAT_CODE to put out floating point register names.
Jeff Law [Sun, 7 Nov 1993 20:33:18 +0000 (20:33 +0000)]
* First cut at cleaning up PA instruction parsing.
* config/tc-hppa.c (pa_get_absolute_expression): Accept pointer to
insn structure as an argument, and a pointer to a string. All
callers changed. Always read any field selector here. Call
evaluate absolute to get a return value.
(evaluate_absolute): Addept pointer to insn structure as its
argument. All callers changed.
(INSERT_FIELD_AND_CONTINUE): New macro for inserting a bitfield
into an instruction and continuing the main pa_ip loop.
(CHECK_FIELD): New macro for simple range checking of fields.
(pa_ip): Delete unused variables. Use INSERT_FIELD_AND_CONTINUE
and CHECK_FIELD. All immediate fields now pass through
pa_get_absolute_expression which will also handle field selectors.
Delete dead code. Simplify.
(md_apply_fix_1): Use CHECK_FIELD to verify any fixes that are
applied are in range. Use bfd_put_32 rather than inserting each
byte of the fixed instrution into the buffer ourselves.
Jeff Law [Sun, 7 Nov 1993 20:08:19 +0000 (20:08 +0000)]
* write.c (fixup_segment): Delete {SEG,GLOBAL}_DIFF_ALLOWED code,
it was PA specific and is no longer needed (it's now handled
within the PA backend).
* config/tc-hppa.h (SEG_DIFF_ALLOWED): Delete definition.
* config/tc-hppa.c (fix_new_hppa): If the subtract symbol for
a fixup is $global$ change it to NULL as $global$ is really only
needed long enough to determine the base type of relocation to use.
Jeff Law [Sun, 7 Nov 1993 20:07:51 +0000 (20:07 +0000)]
* config/tc-hppa.h (SEG_DIFF_ALLOWED): Delete definition.
* config/tc-hppa.c (fix_new_hppa): If the subtract symbol for
a fixup is $global$ change it to NULL as $global$ is really only
needed long enough to determine the base type of relocation to use.
* gdbtypes.h : add a field 'ignore_field_bits in cplus_specific,
and macros to handle the bits.
* stabsread.c (read_one_struct_field): add VISIBILITY_IGNORE, and
for field of length 0, set this bit on.
* cp-valprint.c (cp_print_value_fields): for VISIBILITY_IGNORE
field, print <no value>.
Stu Grossman [Fri, 5 Nov 1993 19:27:49 +0000 (19:27 +0000)]
* inflow.c (terminal_init_inferior): Temporarily use Lynx PIDGET
macro to set process groups.
* infptrace.c (child_resume): Temporarily use Lynx PIDGET to
specify resumption of all threads.
* infrun.c (wait_for_inferior): Fix handling of thread-specific
breakpoints for systems where DECR_PC_AFTER_BREAK > 0 (ie: backup
PC by the right amount when continuing the thread).
* thread.c (thread_apply_command): Add the `thread apply'
command to apply a given GDB command to a list of threads.
Jim Kingdon [Fri, 5 Nov 1993 04:57:29 +0000 (04:57 +0000)]
* dbxread.c (unknown_symchar_complaint): Make message refer to
"symbol descriptor" not "symbol type character" for consistency
with stabs.texinfo terminology.
Jim Kingdon [Fri, 5 Nov 1993 01:02:31 +0000 (01:02 +0000)]
* Makefile.in: Change RUNTEST_FLAGS back to RUNTESTFLAGS per
etc/make-stds.texi. The underscore came from gcc, and dje now
agrees that RUNTESTFLAGS is the correct name.
Jeff Law [Thu, 4 Nov 1993 16:58:27 +0000 (16:58 +0000)]
* From Pete Hoogenboom (hoogen@cs.utah.edu)
* elf32-hppa.c (hppa_elf_get_section_contents): Fix logic error
in last change. Always rebuild symbol extension section the first
time though if output sections exist (fixes ld -r problems).