daily update

ld/testsuite/
* ld-mips-elf/reloc-estimate-1.d: Make endian-neutral.

Provide virtual target "cfi" for targets supporting CFI.
* ld-elf/eh-frame-hdr.d: Replace target and xfail statements by single
`target: cfi'.
* ld-elf/eh-group.exp: Call check_as_cfi instead of is_elf_format.
* ld-elf/eh-group1.s, elf/eh-group2.s: Use more compatible section
flags prefix '%'.
* ld-elf/eh5.d: Replace target statement by `target: cfi' with an Alpha
exception.  Relax the `Code alignment factor' matching.
* lib/ld-lib.exp: Rename istarget as istarget_ld.
(istarget, check_as_cfi): New procedure.
(run_dump_test): New comment for the virtual target `cfi'.

daily update

*** empty log message ***

Add plugin functionality for link-time optimization (LTO).

include/:
* plugin-api.h: New file.

gold/:
* configure.ac (plugins): Add --enable-plugins option.
* configure: Regenerate.
* config.in: Regenerate.
* Makefile.am (LIBDL): New variable.
(CCFILES): Add plugin.cc.
(HFILES): Add plugin.h.
(ldadd_var): Add LIBDL.
* Makefile.in: Regenerate.

* archive.cc: Include "plugin.h".
(Archive::setup): Don't preread archive symbols when using a plugin.
(Archive::get_file_and_offset): Add memsize parameter.  Change callers.
(Archive::get_elf_object_for_member): Call plugin hooks for claiming
files.
(Archive::include_member): Add symbols from plugin objects.
* archive.h (Archive::get_file_and_offset): Add memsize parameter.
* descriptors.cc (Descriptors::open): Check for file descriptors
abandoned by plugins.
(Descriptors::claim_for_plugin): New function.
* descriptors.h (Descriptors::claim_for_plugin): New function.
(Open_descriptor::is_claimed): New field.
(claim_descriptor_for_plugin): New function.
* fileread.cc (File_read::claim_for_plugin): New function.
* fileread.h (File_read::claim_for_plugin): New function.
(File_read::descriptor): New function.
* gold.cc: Include "plugin.h".
(queue_initial_tasks): Add task to call plugin hooks for generating
new object files.
* main.cc: Include "plugin.h".
(main): Load plugin libraries.
* object.h (Pluginobj): Declare.
(Object::pluginobj): New function.
(Object::do_pluginobj): New function.
(Object::set_target): New function.
* options.cc: Include "plugin.h".
(General_options::parse_plugin): New function.
(General_options::General_options): Initialize plugins_ field.
(General_options::add_plugin): New function.
* options.h (Plugin_manager): Declare.
(General_options): Add --plugin option.
(General_options::has_plugins): New function.
(General_options::plugins): New function.
(General_options::add_plugin): New function.
(General_options::plugins_): New field.
* plugin.cc: New file.
* plugin.h: New file.
* readsyms.cc: Include "plugin.h".
(Read_symbols::do_read_symbols): Check for archive before checking
for ELF file.  Call plugin hooks to claim files.
* resolve.cc (Symbol_table::resolve): Record when symbol is referenced
from a real object file; force override when processing replacement
files.
* symtab.cc (Symbol::init_fields): Initialize in_real_elf_ field.
(Symbol::init_base_object): Likewise.
(Symbol::init_base_output_data): Likewise.
(Symbol::init_base_output_segment): Likewise.
(Symbol::init_base_constant): Likewise.
(Symbol::init_base_undefined): Likewise.
(Symbol::output_section): Assert that object is not a plugin.
(Symbol_table::add_from_pluginobj): New function.
(Symbol_table::sized_finalize_symbol): Treat symbols from plugins as
undefined.
(Symbol_table::sized_write_globals): Likewise.
(Symbol_table::add_from_pluginobj): Instantiate template.
* symtab.h (Sized_pluginobj): Declare.
(Symbol::in_real_elf): New function.
(Symbol::set_in_real_elf): New function.
(Symbol::in_real_elf_): New field.
(Symbol_table::add_from_pluginobj): New function.

* testsuite/Makefile.am (AM_CFLAGS): New variable.
(LIBDL): New variable.
(LDADD): Add LIBDL.
(check_PROGRAMS): Add plugin_test_1 and plugin_test_2.
(check_SCRIPTS): Add plugin_test_1.sh and plugin_test_2.sh.
(check_DATA): Add plugin_test_1.err and plugin_test_2.err.
(MOSTLYCLEANFILES): Likewise.
* testsuite/Makefile.in: Regenerate.
* testsuite/plugin_test.c: New file.
* testsuite/plugin_test_1.sh: New file.
* testsuite/plugin_test_2.sh: New file.

2008-09-19  Andrew Stubbs  <ams@codesourcery.com>

* MAINTAINERS: Update my email address.

2008-09-19  Andrew Stubbs  <ams@codesourcery.com>

* frame.c (get_frame_register_bytes): Detect bad debug info.

* write.c (TC_FORCE_RELOCATION_SUB_LOCAL): Heed md_register_arithmetic.
(TC_VALIDATE_FIX_SUB): Likewise.
* config/tc-frv.h (TC_FORCE_RELOCATION_SUB_LOCAL): Likewise.
* config/tc-hppa.h (TC_FORCE_RELOCATION_SUB_LOCAL): Likewise.
* config/tc-mn10300.h (TC_VALIDATE_FIX_SUB): Likewise.
* config/tc-sh.h (TC_VALIDATE_FIX_SUB): Likewise.
(TC_FORCE_RELOCATION_SUB_LOCAL): Likewise.
* config/tc-sh64.h (TC_VALIDATE_FIX_SUB): Likewise.
* config/tc-xtensa.h (TC_VALIDATE_FIX_SUB): Likewise.
* doc/internals.texi (TC_FORCE_RELOCATION_SUB_ABS,
TC_FORCE_RELOCATION_SUB_LOCAL, TC_VALIDATE_FIX_SUB): Show new param.

* elf32-ppc.c (ppc_elf_finish_dynamic_sections): Handle vxworks
_GLOBAL_OFFSET_TABLE_ in .got.plt section.  Add BFD_ASSERTs.

* write.c (md_register_arithmetic): Define.
(fixup_segment): Adjust TC_FORCE_RELOCATION_SUB_ABS invocation.
Modify error message when registers involved.
(TC_FORCE_RELOCATION_SUB_ABS): Heed md_register_arithmetic.
* config/tc-sh.h (TC_FORCE_RELOCATION_SUB_ABS): Likewise.

daily update

*** empty log message ***

    * elf32-xtensa.c (elf_xtensa_check_relocs): Check for negative refcount.

daily update

*** empty log message ***

Fix a crash on uninitialized ECS->EVENT_THREAD for a newly found thread.
* infrun.c (wait_for_inferior): Move this ECS->EVENT_THREAD
initialization ...
(fetch_inferior_event): ... and this ECS->EVENT_THREAD initialization
...
(handle_inferior_event): ... here after the add_thread call together
with the local adjust_pc_after_break and reinit_frame_cache calls.

bfd/
PR 6893 - Do not consider FDEs for discarded sections as invalid.
* elf-eh-frame.c (_bfd_elf_parse_eh_frame): New REQUIRE_CLEARED_RELOCS.
Consider FDEs with cleared relocations as valid and ignorable.

ld/testsuite/
* ld-elf/eh-group.exp, ld-elf/eh-group1.s, ld-elf/eh-group2.s: New test.

binutils/
Suppress warnings on NONE relocations to discarded sections.
* readelf.c (is_none_reloc): New function.
(debug_apply_relocations): Ignore is_none_reloc() relocations.

Fix testcases compilation failures due to unused variables.
* ld-elfcomm/common1b.c (dummy1): Mark the variable for GCC as USED.
* ld-shared/main.c (main): Move the P variable only to the places where
it is used.

daily update

*** empty log message ***

* breakpoint.c (bpstat_stop_status): Clear breakpoint_at for
all hardware bpstats.

    * gstdint.h: Delete.

* target-reloc.h (relocate_section): Check whether a symbol is
defined by the ABI before reporting an undefined symbol error.
* target.h (Target::is_defined_by_abi): Make parameter const.
(Target::do_is_defined_by_abi): Likewise.
* i386.cc (Target_i386::do_is_defined_by_abi): Likewise.
* powerpc.cc (Target_powerpc::do_is_defined_by_abi): Likewise.
* sparc.cc (Target_sparc::do_is_defined_by_abi): Likewise.
* x86_64.cc (Target_x86_64::do_is_defined_by_abi): Likewise.
* testsuite/Makefile.am (tls_test_shared.so): Add -Wl,-z,defs.
* testsuite/Makefile.in: Rebuild.

* fileread.cc (make_view): Add casts to avoid warning.

bfd/

2008-09-16  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/6877
* elflink.c (_bfd_elf_merge_symbol): Allow a common symbol to
override the function in a shared library.

ld/testsuite/

2008-09-16  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/6877
* ld-elf/comm1.c: New.
* ld-elf/func1.c: Likewise.

* ld-elf/shared.exp: Add tests for libfunc1.so and comm1.

PR 6844
* elf32-ppc.c (SYM_VAL): Define.  Use throughout to find symbol vma.
(ppc_elf_relocate_section): Correct GOT offset calculation.
(ppc_elf_finish_dynamic_symbol): Use PPC_HA and PPC_LO.
(ppc_elf_finish_dynamic_sections): Likewise.  Error if
htab->elf.hgot symbol is not defined in htab->got section.

* i386.cc (Target_i386::define_tls_base_symbol): Update comments.
* x86_64.cc (Target_x86_64::define_tls_base_symbol): Likewise.

* options.h (General_options::output_is_executable): New.
(General_options::output_is_pie): New.
* i386.cc (Target_i386::define_tls_base_symbol): Use SEGMENT_START
for shared libraries.
* x86_64.cc (Target_x86_64::define_tls_base_symbol): Likewise.

*** empty log message ***

daily update

PR 6888
* elf32-ppc.c (ppc_elf_relocate_section): Handle NULL symbol section
on R_PPC_TOC16, R_PPC_SDAREL16, R_PPC_EMB_SDA2REL, R_PPC_EMB_SDA21,
R_PPC_EMB_RELSDA, R_PPC_SECTOFF* relocs.

* infcall.c (generic_push_dummy_code): Remove.
(push_dummy_code): Unconditionally call gdbarch_push_dummy_code.

* dwarf2read.c (struct abbrev_info): Make members name, form 16 bits.
(struct attribute): Ditto.

* gas/all/gas.exp: Don't run redef tests on a bunch of targets.
* gas/elf/elf.exp: Likewise.

* write.c (install_reloc): Correct EMIT_SECTION_SYMBOLS test.

* config/tc-frv.c (md_apply_fix): Use abs_section_sym for
relocs with no symbol.
* config/tc-mmix.c (md_assemble): Mark fake symbol on
BFD_RELOC_MMIX_BASE_PLUS_OFFSET as OK for use by relocs.
(mmix_md_end): Likewise mark mmix reg contents section symbol.

* hppa-linux-tdep.c (hppa_linux_sigtramp_frame_unwind_cache): Record
HPPA_IPSW_REGNUM and HPPA_SAR_REGNUM values.

* hppa-linux-tdep.c (hppa_dwarf_reg_to_regnum): Remove surrounding
"#if 0" "#endif".  Fix mapping of DWARF DBX registers to GDB registers.
Correct arguments and improve comments.
(hppa_linux_init_abi): Call set_gdbarch_dwarf2_reg_to_regnum.  Delete
disabled code.
* hppa-tdep.c (hppa64_dwarf_reg_to_regnum): Fix check for floating
point DBX register, change error to warning, and improve comments.

* dwarf2read.c (struct die_info): Make members tag, num_attrs 16 bits.

*** empty log message ***

daily update

* hppa-linux-tdep.c (hppa_linux_supply_fpregset): Correct iteration.

Add new tests for z80

Fix bugs in the disassembly of some ld-instructions

    * defs.h (GCC_GENERATED_STDINT_H): Define.

Fix Opcode generation of ld a,(bc) and ld a,(de) on target z80

*** empty log message ***

daily update

* varobj.c (varobj_set_display_format): Use xfree.
* tracepoint.c (stringify_collection_list): Use xfree.
* remote-fileio.c (remote_fileio_reset): Use xfree.
* mipsread.c (read_alphacoff_dynamic_symtab): Use xfree.
* dfp.c (decimal_from_floating): Use xfree, xstrprintf.  Don't use
asprintf.
* cp-support.c (mangled_name_to_comp): Use xfree.

    * ada-lang.c (remove_extra_symbols): Remove stub symbols if
        the associated complete symbol is also in the list.
        (ada_add_local_symbols, ada_add_non_local_symbols): New functions,
        extracted out from ada_lookup_symbol_list.
        (ada_lookup_symbol_list): Use them.  Remove the search through
        the minimal symbols.

    * dwarf2read.c (add_partial_subprogram): New procedure.
        (scan_partial_symbols): Use it.
        (load_partial_dies): Read in children of subprogram and lexical
        blocks for Ada compilation units.

* gdb.mi/mi-nonstop.exp: Do not check thread state while a
stop is pending.  Avoid ".*" when two stops are pending.
* lib/gdb.exp (fullname_syntax_POSIX, fullname_syntax_UNC)
(fullname_syntax_DOS_CASE, fullname_syntax_DOS): Do not match
newlines in fullnames.
* lib/mi-support.exp (mi_run_cmd): Do not require an anchor.
(mi_expect_stop): Update comments.  Only anchor in sync mode.
Do not match newlines.
(mi_send_resuming_command_raw): Always return status.
(mi_get_stop_line): Do not match more than one line by accident.
Only anchor in sync mode.
(mi_run_inline_test): If -exec-next fails, give up.

* symfile.c (build_id_verify): Free 'found'.
(find_separate_debug_file): Use xfree, not free.

*** empty log message ***

daily update

* corefile.c (write_memory): Remove unnecessary copying.

* sol-thread.c (_initialize_sol_thread): Add FIXME regarding
order of _initialize_* fns.

* dwarf2read.c (comp_unit_head): Rename first_die_ptr to
first_die_offset.  All uses updated.
Delete unused members cu_head_ptr, next.
Move members base_known, base_address to ...
(dwarf2_cu) ... here.  All uses updated.

2008-09-12  Sterling Augustine  <sterling@tensilica.com>
* config/tc-xtensa.c (init_op_placement_info_table): Allow number of
operands equal to MAX_INSN_ARGS.

gdb/
* Makefile.in (generated_files): Add $(NAT_GENERATED_FILES).
* config/i386/i386gnu.mh (NAT_GENERATED_FILES): New.

gdb/doc/
* gdbint.texinfo (Native Debugging): Mention NAT_GENERATED_FILES.

2008-09-11  Chris Demetriou  <cgd@google.com>

        * options.h (origin): New -z option.
        * layout.cc (Layout:finish_dynamic_section): If "-z origin"
        is specified, set DF_ORIGIN in DT_FLAGS and set DF_1_ORIGIN
        in DT_FLAGS_1.

daily update

*** empty log message ***

gas/testsuite/

2008-09-11  H.J. Lu  <hongjiu.lu@intel.com>

* gas/i386/sse2avx.s: Remove pclmulXXX tests.  Add tests for
Intel syntax.
* gas/i386/x86-64-sse2avx.s: Likewise.

* gas/i386/sse2avx.d: Updated.
* gas/i386/x86-64-sse2avx.d: Likewise.

opcodes/

2008-09-11  H.J. Lu  <hongjiu.lu@intel.com>

* i386-opc.tbl: Fix memory operand size for cmpXXXs[sd].
* i386-tbl.h: Regenerated.

bfd/
Fix loading large elf64 binaries on 32bit hosts.
* configure.in: Call AC_SYS_LARGEFILE.
* config.in: Regenerate.
* configure: Regenerate.

binutils/
Fix loading large elf64 binaries on 32bit hosts.
* configure.in: Call AC_SYS_LARGEFILE.
* config.in: Regenerate.
* configure: Regenerate.

gas/
* configure.in: Call AC_SYS_LARGEFILE.
* config.in: Regenerate.
* configure: Regenerate.

gprof/
* configure.in: Call AC_SYS_LARGEFILE.
* gconfig.in: Regenerate.
* configure: Regenerate.

ld/
* configure.in: Call AC_SYS_LARGEFILE.
* config.in: Regenerate.
* configure: Regenerate.

libiberty/
* configure.in: Call AC_SYS_LARGEFILE.
* config.in: Regenerate.
* configure: Regenerate.

* gdb.base/stack-checking.c (big_frame): Reduce stack consumption
on SPU.

* fork-child.c (startup_inferior): Use target_wait and target_resume
directly instead of calling wait_for_inferior / resume.

* infcmd.c (kill_if_already_running): Do not call no_shared_libraries
or init_wait_for_inferior.
(run_command_1): Call init_wait_for_inferior.

Add missing ChangeLog entry.

* gdbtypes.h (builtin_type_void_data_ptr, builtin_type_void_func_ptr,
builtin_type_CORE_ADDR, builtin_type_char, builtin_type_short,
builtin_type_int, builtin_type_long, builtin_type_signed_char,
builtin_type_unsigned_char, builtin_type_unsigned_short,
builtin_type_unsigned_int, builtin_type_unsigned_long,
builtin_type_float, builtin_type_double, builtin_type_long_double,
builtin_type_complex, builtin_type_double_complex, builtin_type_string,
builtin_type_bool, builtin_type_long_long,
builtin_type_unsigned_long_long): Remove macros.

(builtin_type_f_character, builtin_type_f_integer,
builtin_type_f_integer_s2, builtin_type_f_logical,
builtin_type_f_logical_s1, builtin_type_f_logical_s2,
builtin_type_f_real, builtin_type_f_real_s8, builtin_type_f_real_s16,
builtin_type_f_complex_s8, builtin_type_f_complex_s16,
builtin_type_f_complex_s32): Likewise.

(builtin_type_m2_char, builtin_type_m2_int, builtin_type_m2_card,
builtin_type_m2_real, builtin_type_m2_bool): Likewise.

(struct builtin_f_type, builtin_f_type): Move to f-lang.h.
(struct builtin_m2_type, builtin_m2_type): Move to m2-lang.h.

* f-lang.h (struct builtin_f_type, builtin_f_type): Move here.
* m2-lang.h (struct builtin_m2_type, builtin_m2_type): Move here.

* solib-svr4.c (LM_ADDR_FROM_LINK_MAP): Use builtin types of
target_gdbarch instead of builtin_type_void_data_ptr.
(LM_DYNAMIC_FROM_LINK_MAP, LM_NEXT, LM_NAME,
IGNORE_FIRST_LINK_MAP_ENTRY, scan_dyntag, elf_locate_base,
solib_svr4_r_map, solib_svr4_r_brk, solib_svr4_r_ldsomap,
open_symbol_file_object): Likewise.
* nto-tdep.c (LM_ADDR): Likewise.

* bsd-uthread.c (bsd_uthread_read_memory_address): New function.
(bsd_uthread_fetch_registers, bsd_uthread_store_registers,
bsd_uthread_wait, bsd_uthread_find_new_threads): Use it.

* procfs.c (procfs_address_to_host_pointer): Use target_gdbarch
and its associated types to perform pointer conversion.
(procfs_can_use_hw_breakpoint): Likewise.
(procfs_auxv_parse): Remove unused variable.

* auxv.c (default_auxv_parse): Use gdbarch_ptr_bit (target_gdbarch)
instead of builtin_type_void_data_ptr.
* target.c (default_region_ok_for_hw_watchpoint): Likewise.

* expprint.c (print_subexp_standard): Compare against builtin type
associated with exp->gdbarch instead of builtin_type_char.

* f-valprint.c (f_val_print): Use extract_unsigned_integer to
extract values of arbitrary logical type.  Handle arbitrary
complex types.

* printcmd.c (float_type_from_length): New function.
(print_scalar_formatted, printf_command): Use it.

* valops.c: Include "objfiles.h" and "symtab.h".
(find_function_in_inferior): New argument OBJF_P.  Use it to return
objfile where function is defined.  Use per-objfile arch types
instead of builtin_type_ to define default return type.

* linux-fork.c (checkpoint_command): Update calls.  Use per-objfile
architecture to define inferior call argument types.
* gcore.c (derive_heap_segment): Likewise.
* objc-lang.c (value_nsstring): Likewise.
* scm-lang.c (scm_lookup_name): Likewise.
* scm-valprint.c (scm_inferior_print): Likewise.
* valops.c (value_allocate_space_in_inferior): Likewise.

* eval.c (evaluate_subexp_standard): Update calls.
* objc-lang.c (lookup_objc_class, print_object_command): Likewise.

* linux-fork.c: Include "objfiles.h".
* scm-lang.c: Include "objfiles.h".
* scm-valprint.c: Include "objfiles.h".

ChangeLog:

* gdbarch.sh (name_of_malloc): Remove.
* gdbarch.c, gdbarch.h: Re-generate.
* valops.c (value_allocate_space_in_inferior): Do not call
gdbarch_name_of_malloc.

doc/ChangeLog:

* gdbint.texinfo (Target Conditionals): Remove documentation
for gdbarch_name_of_malloc.

* valarith.c (value_x_unop): Use builtin_type_int8 as type for
UNOP_POSTINCREMENT/UNOP_POSTDECREMENT constant 0 argument.
(value_bit_index): Use extract_unsigned_integer
instead of unpack_long to read single byte.

* infcall.c (value_arg_coerce): Add GDBARCH parameter.  Use its
associates types instead of builtin_type_ macros.
(find_function_addr): Leave output VALUE_TYPE NULL if unknown.
(call_function_by_hand): Use per-architecture "int" type as
fall-back if find_function_addr returns NULL VALUE_TYPE.
Update call to value_arg_coerce.

* cp-abi.h (cplus_method_ptr_size): Add TO_TYPE parameter.
(cplus_make_method_ptr): Add TYPE parameter.
* cp-abi.c (cplus_method_ptr_size): Add TO_TYPE parameter.  Pass it
on to current_cp_abi.method_ptr_size callback.
(cplus_make_method_ptr): Add TYPE parameter.  Pass it on to
current_cp_abi.make_method_ptr callback.

* gdbtypes.c (lookup_methodptr_type): Pass target type
argument to cplus_method_ptr_size.
* valops.c (value_cast): Pass type argument to cplus_make_method_ptr.
(value_struct_elt_for_reference): Likewise.

* gnu-v3-abi.c (get_class_arch): New function.
(vtable_address_point_offset): Add GDBARCH parameter.  Use it
instead of current_gdbarch.  Update all callers.
(gnuv3_get_vtable): Likewise.
(gnuv3_get_virtual_fn): Likewise.
(gnuv3_decode_method_ptr): Likewise.
(gnuv3_rtti_type): Call get_class_arch to determine architecture.
Use it instead of current_gdbarch.
(gnuv3_virtual_fn_field): Likewise.
(gnuv3_baseclass_offset): Likewise.
(gnuv3_print_method_ptr): Likewise.
(gnuv3_method_ptr_to_value): Likewise.
(gnuv3_method_ptr_size): Add TYPE parameter.  Use it to determine
class architecture.  Use architecture types instead of builtin types.
(gnuv3_make_method_ptr): Likewise.

* cp-valprint.c (cp_print_class_member): Expect pointer type
instead of class type.  Use its length when extracting value.
* c-valprint.c (c_val_print): Update call to cp_print_class_member.

* stack.c (return_command): Use frame architecture to determine
default integer return type.

* f-valprint.c (f77_get_dynamic_lowerbound): Use frame architecture
to determine pointer types.
(f77_get_dynamic_upperbound): Likewise.

* objc-lang.c (OBJC_FETCH_POINTER_ARGUMENT): Remove.
(resolve_msgsend): Use architecture of current frame to determine
pointer types.  Inline OBJC_FETCH_POINTER_ARGUMENT.
(resolve_msgsend_stret, resolve_msgsend_super,
resolve_msgsend_super_stret): Likewise.

* alpha-tdep.c (alpha_register_type): Use builtin_type (gdbarch)
instead of builtin_type_ macros.
* amd64-tdep.c (amd64_register_type): Likewise.
(amd64_get_longjmp_target): Likewise.
* arm-tdep.c (arm_register_type): Likewise.
* avr-tdep.c (avr_register_type): Likewise.
* cris-tdep.c (cris_register_type, crisv32_register_type): Likewise.
* frv-tdep.c (frv_register_type): Likewise.
* h8300-tdep.c (h8300_register_type): Likewise.
* hppa-tdep.c (hppa32_convert_from_func_ptr_addr,
hppa_skip_trampoline_code): Likewise.
* i386-tdep.c (i386_register_type): Likewise.
(i386_unwind_pc, i386_sse_type): Likewise.
* ia64-tdep.c (ia64_register_type): Likewise.
* m32r-tdep.c (m32r_register_type): Likewise.
* m68k-tdep.c (m68k_register_type, m68k_unwind_pc): Likewise.
* m88k-tdep.c (m88k_register_type): Likewise.
* mep-tdep.c (mep_register_type): Likewise.
* mips-tdep.c (mips_pseudo_register_type): Likewise.
* mn10300-tdep.c (mn10300_register_type): Likewise.
* mt-tdep.c (mt_copro_register_type): Likewise.
* rs6000-tdep.c (rs6000_builtin_type_vec64): Likewise.
(rs6000_convert_register_p, rs6000_register_to_value,
rs6000_value_to_register): Likewise.
* s390-tdep.c (s390_register_type): Likewise.
* sh64-tdep.c (sh64_register_type): Likewise.
(sh64_build_float_register_type, sh64_do_fp_register): Likewise.
* sh-tdep.c (sh_sh2a_register_type, sh_sh3e_register_type,
sh_sh4_build_float_register_type, sh_sh4_register_type,
sh_default_register_type): Likewise.
* sparc64-tdep.c (sparc64_register_type): Likewise.
* sparc-tdep.c (sparc32_register_type): Likewise.
* spu-tdep.c (spu_builtin_type_vec128, spu_register_type): Likewise.
* v850-tdep.c (v850_register_type): Likewise.
* vax-tdep.c (vax_register_type): Likewise.
* xtensa-tdep.c (xtensa_register_type, xtensa_unwind_pc,
xtensa_push_dummy_call): Likewise.

* std-regs.c (value_of_builtin_frame_fp_reg,
value_of_builtin_frame_pc_reg): Likewise.
* target-descriptions.c (tdesc_register_type): Likewise.

* ada-lang.c (ada_coerce_to_simple_array_type): Use builtin_type_int32
instead of builtin_type_int as default unspecified integral type.
(ada_index_type, ada_array_bound_from_type, ada_variant_discrim_type,
assign_component, to_fixed_range_type): Likewise.
* ada-typeprint.c (print_range, print_range_bound,
print_range_type_named): Likewise.
* ada-valprint.c (print_optional_low_bound, ada_val_print_1): Likewise.
* eval.c (evaluate_subexp_standard): Likewise.
* gnu-v2-abi.c (gnuv2_virtual_fn_field): Likewise.
* gnu-v3-abi.c (gnuv3_get_virtual_fn, gnuv3_baseclass_offset,
build_gdb_vtable_type): Likewise.
* jv-lang.c (java_array_type): Likewise.
* m2-typeprint.c (m2_print_bounds, m2_is_long_set_of_type): Likewise.
* m2-valprint.c (m2_print_long_set): Likewise.
* parse.c (follow_types): Likewise.
* p-typeprint.c (pascal_type_print_base): Likewise.
* valops.c (value_one, value_array, value_string,
value_bitstring): Likewise.
* value.c (allocate_repeat_value, value_from_string): Likewise.
* varobj.c (c_describe_child): Likewise.
* mt-tdep.c (mt_register_type): Likewise.
* sh-tdep.c (sh_sh4_build_float_register_type): Likewise.
* sh64-tdep.c (sh64_build_float_register_type): Likewise.

* defs.h (struct gdbarch): Add forward declaration.
(set_next_address): Add GDBARCH argument.
* printcmd.c (set_next_address): Use it to find pointer type.
* breakpoint.c (breakpoint_1): Update call.
* source.c (line_info): Likewise.
* findcmd.c (find_command): Use current_gdbarch to find pointer type.

* breakpoint.c (set_breakpoint_count): Use platform-neutral
types for internal variable values.
* infrun.c (handle_inferior_event): Likewise.
* source.c (forward_search_command, reverse_search_command): Likewise.
* tracepoint.c (set_tracepoint_count, set_traceframe_num,
set_tracepoint_num, set_traceframe_context): Likewise.

* gdbtypes.h (struct builtin_type): Remove builtin_true_char
and builtin_true_unsigned_char.
(builtin_type_true_char): Remove macro, add extern declaration.
(builtin_type_true_unsigned_char): Add extern declaration.
* gdbtypes.c (builtin_type_true_char): New global variable.
(builtin_type_true_unsigned_char): Likewise.
(_initialize_gdbtypes): Initialize them.
(gdbtypes_post_init): Do not initialize builtin_true_char
and builtin_true_unsigned_char members of struct builtin_type.

* printcmd.c (print_scalar_formatted): Do not use builtin_type;
use builtin_type_true_unsigned_char instead.

* ada-valprint.c (ada_val_print_1): Use builtin_type_true_char
instead of builtin_type_char for internal string.

* gdbtypes.h (builtin_type_void): Remove macro, add declaration.
(builtin_type_f_void): Remove macro.
* gdbtypes.c (builtin_type_void): New global variable.
(_initialize_gdbtypes): Initialize it.

* gnu-v3-abi.c (build_gdb_vtable_type): Do not call
lookup_pointer_type or lookup_function_type on builtin_type_void.
* printcmd.c (set_next_address): Likewise.
* objc-lang.c (value_nsstring): Likewise.
* mt-tdep.c (mt_copro_register_type): Likewise.
* xtensa-tdep.c (xtensa_register_type): Likewise.

* symfile.c (syms_from_objfile): Remove special handling
of builtin_type_void and builtin_type_char.

* eval.c (evaluate_subexp_standard): Use exp->gdbarch types instead
of builtin_type_ macros when handling OP_OBJC_ operations.
* objc-lang.c (print_object_command): Likewise.

* ada-valprint.c: Include "objfiles.h".
(ada_val_print_1): Use the gdbarch associated with the objfile whether
a System.Address type is defined to retrieve the proper pointer type
to use to print it.

* ada-lang.c (value_pos_atr): Add TYPE argument.  Use it as
result type instead of builtin_type_int.
(value_subscript_packed): Use pos_atr instead of value_pos_atr.
(ada_value_subscript): Update call to value_pos_atr.
(ada_value_ptr_subscript): Likewise.
(ada_evaluate_subexp): Likewise.

* ada-lang.c (cast_to_fixed): Do not cast to builtin_type_double.
(cast_from_fixed_to_double): Rename to ...
(cast_from_fixed): ... this.  Add TYPE parameter.  Use it instead
of builtin_type_double.
(ada_value_cast): Use cast_from_fixed instead of casting result
of cast_from_fixed_to_double.
(ada_evaluate_subexp): Update calls to cast_from_fixed_to_double.

* valops.c (value_ind): No longer allow dereferencing an
integer type.
* eval.c (evaluate_subexp_standard): Handle deferencing an
integer type here.
* ada-lang.c (ada_evaluate_subexp): Likewise.

* ada-valprint.c (ada_val_print_1): When implicitly dereferencing
a reference type, pass the reference type directly to unpack_pointer.
* c-valprint.c (c_val_print): Likewise.
* f-valprint.c (f_val_print): Likewise.
* m2-valprint.c (print_variable_at_address, m2_val_print): Likewise.
* p-valprint.c (pascal_val_print): Likewise.

* eval.c (evaluate_subexp_standard): Use builtin_type_int8
to construct the EVAL_SKIP dummy return value.
* ada-lang.c (ada_evaluate_subexp): Likewise.
* jv-lang.c (evaluate_subexp_java): Likewise.
* m2-lang.c (evaluate_subexp_modula2): Likewise.
* scm-lang.c (evaluate_exp): Likewise.

* value.h (coerce_enum, coerce_number): Remove prototypes.
* value.c (coerce_enum, coerce_number): Remove.
* valarith.c (value_x_binop): Do not call coerce_enum.
(value_x_unop): Likewise.
(value_logical_not): Call coerce_array instead of coerce_number.

* ax-gdb.c: Include "language.h".
(gen_frame_args_address): Add GDBARCH parameter; use it
instead of current_gdbarch.
(gen_frame_locals_address): Likewise.
(gen_var_ref): Add GDBARCH parameter.  Update calls to
gen_frame_args_address and gen_frame_locals_address.  Use
pointer type from gdbarch.
(gen_usual_unary): Add EXP parameter.  Use integer type
from exp->gdbarch.
(gen_usual_arithmetic): Likewise.
(gen_integral_promotions): Likewise.
(gen_add, gen_sub): Remove.
(gen_ptradd, gen_ptrsub, gen_ptrdiff): New functions.
(gen_logical_not): Use passed-in boolean result type
instead of builtin_type_int.
(gen_complement): Do not call gen_usual_unary or
gen_integral_promotions.
(gen_struct_ref): Call require_rvalue instead of gen_usual_unary.
(gen_repeat): Add EXP parameter.  Update call to gen_expr.
Use builtin_type_int32 as internal range type.
(gen_sizeof): Add EXP and SIZE_TYPE parameters.  Use SIZE_TYPE
as result type.  Update call to gen_expr.
(gen_expr): Add EXP parameter.  Update calls to gen_expr,
gen_repeat, gen_var_ref, gen_usual_unary, gen_usual_arithmetic,
and gen_integral_promotions.  Call gen_ptradd, gen_ptrsub,
gen_ptrdiff, or gen_binop instead of gen_add or gen_sub.
Use exp->gdbarch instead of current_gdbarch.
Call language_bool_type to determine result type of UNOP_LOGICAL_NOT.

* eval.c (evaluate_subexp_standard): Add calls to binop_promote
and unop_promote before calling value_binop et. al.
* ada-lang.c (ada_evaluate_subexp): Add calls to binop_promote
and unop_promote before calling value_binop et. al.

* valarith.c (value_binop): Do not call binop_promote or unop_promote.
(value_pos): Do not call unop_promote.
(value_neg, value_complement): Likewise.

* value.h (unop_promote, binop_promote): Add prototypes.
* eval.c (unop_promote, binop_promote): New functions.
* valarith.c (unop_result_type, binop_result_type): Remove.
(value_binop): Call binop_promote or unop_promote.
Inline remaining parts of binop_result_type.  Remove special
code to truncate integer values for unsigned operations.
(value_pos): Call unop_promote.  Inline remaining parts of
unop_result_type.
(value_neg, value_complement): Likewise.

* value.h (value_add, value_sub): Remove.
(value_ptradd, value_ptrsub, value_ptrdiff): Add prototypes.
* valarith.c (value_add, value_sub): Remove.
(value_ptradd, value_ptrsub, value_ptrdiff): New functions.
(find_size_for_pointer_math): Add assertion.  Update comment.
(value_binop): Update comment.

* eval.c (ptrmath_type_p): New function.
(evaluate_subexp_standard): Replace value_add and value_sub
by value_ptradd, value_ptrsub, value_ptrdiff or value_binop.
Use builtin_type_uint8 instead of builtin_type_char to hold
the increment for BINOP_{PRE,POST}{IN,DE}CREMENT operations.
* valarith.c (value_subscript): Replace value_add by
value_ptradd.  Replace value_sub by value_binop.
* ada-lang.c (ada_value_ptr_subscript): Likewise.
(ada_tag_name_2): Replace value_add by value_ptradd.
(ada_evaluate_subexp): Replace value_add and value_sub by
value_binop.
* m2-lang.c (evaluate_subexp_modula2): Replace value_add
by value_ptradd.
* gnu-v2-abi.c (gnuv2_virtual_fn_field): Likewise.
* gnu-v3-abi.c (gnuv3_method_ptr_to_value): Likewise.

* eval.c (evaluate_subexp_for_sizeof): Use builtin_int type of
the expression architecture instead of builtin_type_int as the
sizeof return type.

* expression.h (enum exp_opcode): Document OP_COMPLEX to take
a type parameter as expression element.
* eval.c (evaluate_subexp_standard) [OP_COMPLEX]: Retrieve result
type as expression element.
* f-exp.y: Pass in type when buildin OP_COMPLEX expression.
* parse.c (operator_length_standard): Update length of OP_COMPLEX.