merge from gcc

Correctly handle phex(*,0) and phex_nz(*,0).

* utils.c (phex_nz): For default case, set str to phex_nz return
value.

* config/tc-i386.h (md_end): Define.
(i386_elf_emit_arch_note): Declare.
(CpuUnknown): Delete.
* config/tc-i386.c (default_arch): Constify.
(smallest_imm_type): Remove CpuUnknown test.
(md_assemble): Don't bother checking cpu_arch_flags non-zero.
(i386_elf_emit_arch_note): New function.

* common.h (NT_ARCH): Define.  Remove incorrect comment.

* readelf.c: Fix warnings without terminating newline.

* binutils-all/objcopy.exp: Allow '? main', to suit powerpc64.

* emultempl/elf32.em (gld${EMULATION_NAME}_finish): Only emit this
function when LDEMUL_FINISH isn't set to the same name.  Don't
call ${LDEMUL_FINISH}.
(ld_${EMULATION_NAME}_emulation): Call $LDEMUL_FINISH if defined.
* emultempl/armelf.em (arm_elf_finish): Call
gld${EMULATION_NAME}_finish.
* emultempl/hppaelf.em (hppaelf_finish): Rename to
gld${EMULATION_NAME}_finish.  Call bfd_elf32_discard_info and
hppaelf_layout_sections_again if necessary.
(need_laying_out): New var.
(hppaelf_layaout_sections_again): Rename to
hppaelf_layout_sections_again.  Clear need_laying_out.
(PARSE_AND_LIST_OPTIONS): Format text.

* po/POTFILES.in: Regenerate.

* po/SRC-POTFILES.in: Regenerate.

* config/tc-mips.c (mips_ip): Re-allow %hi() op for non-ELF assembler.

* elflink.h (elf_link_input_bfd): Complain about all relocs
against local syms from discarded sections.
(elf_reloc_symbol_deleted_p): Don't try to swap in external syms.

revert last change.

2001-11-14  H.J. Lu  <hjl@gnu.org>

* emultempl/armelf.em (arm_elf_finish): Renamed from
gld${EMULATION_NAME}_finish.
(LDEMUL_FINISH): Set to arm_elf_finish.

* TODO (register_buffer): Delete.
* regcache.c (register_buffer): Make static.
(regcache_collect): New function.
* regcache.h (register_buffer): Delete declaration.
(regcache_collect): Declare.
* remote.c (store_register_using_P): Rewrite using
regcache_collect.
(remote_store_registers): Ditto.
* go32-nat.c (store_register): Ditto.

gas/ChangeLog
* dwarf2dbg.c (get_frag_fix): Align last frag size.

Move global state variables to ``struct remote_state''.

* elf64-sparc.c (sparc64_elf_relocate_section): Disregard
overflows in the .stab section.

* remote.c: Include "gdb_assert.h".
(tty_input): Wire buffer to 400 bytes.
(readsocket): Check tty_input doesn't overflow.
(build_remote_gdbarch_data, _initialize_remote): Don't multi-arch
tty_input.

2001-11-14  Michael Snyder  <msnyder@redhat.com>

        * d10v-tdep.c (d10v_pointer_to_address): Use new type flag
TYPE_FLAG_CODE_SPACE to recognize a pointer that has been cast
into the instruction address space.
* Makefile.in (doublest.o): Add dependency on gdbtypes.h.

2001-11-13  Michael Snyder  <msnyder@redhat.com>

* gdb.base/cvexpr.c, gdb.base/cvexpr.exp: New files.
Tests for expressions using 'const' and 'volatile'.

2001-11-14  Michael Snyder  <msnyder@redhat.com>
Add address space identifiers to expression language for types.
* c-exp.y (space_identifier, cv_with_space_id,
const_or_volatile_or_space_identifier_noopt,
const_or_volatile_or_space_identifier): New terminals.
(ptype): Accept const_or_volatile_or_space_identifier.
(typebase): Accept const_or_volatile_or_space_identifier.
* c-typeprint.c (c_type_print_cv_qualifier): Rename to
c_type_print_modifier.  Handle address space modified types.
* gdbtypes.h (TYPE_FLAG_CODE_SPACE, TYPE_FLAG_DATA_SPACE):
New type flags.
(struct type): Add new field as_type for addr-space qualified types.
(TYPE_AS_TYPE): New macro, retrieves the chain of types that are
identical to this one except for address-space qualification.
* gdbtypes.c (alloc_type): Initialize new field 'as_type'.
(address_space_name_to_int): New function.
(address_space_int_to_name): New function.
(make_type_with_address_space): New function.
(make_cv_type): Handle as_type field of new struct type object.
* parse.c (check_type_stack_depth): New function.
(push_type_address_space): New function.
(follow_types): Handle types with address-space qualifier.
* parser-defs.h (enum type_pieces): Add enum tp_space_identifier.

2001-11-11  Daniel Jacobowitz  <drow@mvista.com>

* bfd-in.h (bfd_elf32_discard_info): Add prototype.
(bfd_elf64_discard_info): Likewise.
* bfd-in2.h: Regenerate.
* elf-bfd.h (struct elf_reloc_cookie): New.
(struct elf_backend_data): Add elf_backend_discard_info,
elf_backend_ignore_discarded_relocs, and elf_backend_write_section.
(_bfd_elf32_reloc_symbol_deleted_p): Add prototype.
(_bfd_elf64_reloc_symbol_deleted_p): Likewise.
* elf32-mips.c (_bfd_elf32_mips_discard_info): New.
(_bfd_elf32_mips_ignore_discarded_relocs): New.
(_bfd_elf32_mips_write_section): New.
(elf_backend_discard_info): Define.
(elf_backend_ignore_discarded_relocs): Define.
(elf_backend_write_section): Define.
* elfcode.h (elf_bfd_discard_info): Define.
(elf_reloc_symbol_deleted_p): Define.
* elflink.h (elf_link_input_bfd): Check
elf_section_ignore_discarded_relocs.  Call
bed->elf_backend_write_section if available.
(elf_reloc_symbol_deleted_p): New.
(elf_bfd_discard_info): New.
(elf_section_ignore_discarded_relocs): New.
* elfxx-target.h (elf_backend_discard_info): Define.
(elf_backend_ignore_discarded_relocs): Define.
(elf_backend_write_section): Define.
(elfNN_bed): Add elf_backend_discard_info,
elf_backend_ignore_discarded_relocs, and
elf_backend_write_section.
* libbfd-in.h (_bfd_discard_section_stabs): Add prototype.
* libbfd.h: Regenerate.
* stabs.c (_bfd_discard_section_stabs): New.

2001-11-11  Daniel Jacobowitz  <drow@mvista.com>

* emultempl/elf32.em (gld${EMULATION_NAME}_finish): New.
(struct ld_emulation_xfer_struct): Use it.

* binutils-all/readelf.exp: powerpc-*-eabi* is an ELF target.

binutils/ChangeLog
* doc/binutils.texi (objdump): Document ppc -M options.
gas/ChangeLog
* config/tc-ppc.c (ppc_insert_operand): Pass (ppc_cpu | ppc_size)
to operand->insert.
(md_assemble): Likewise.
gas/testsuite/ChangeLog
* gas/ppc/booke.d: Modify reloc and target matches for powerpc64.
include/opcode/ChangeLog
* ppc.h (struct powerpc_operand <insert, extract>): Add dialect param.
opcodes/ChangeLog
* ppc-opc.c (PPC64): Revert 2001-10-12. Do include PPC_OPCODE_PPC.
(insert_bat, extract_bat, insert_bba, extract_bba,
insert_bd, extract_bd, insert_bdm, extract_bdm,
insert_bdp, extract_bdp, valid_bo,
insert_bo, extract_bo, insert_boe, extract_boe,
insert_ds, extract_ds, insert_de, extract_de,
insert_des, extract_des, insert_li, extract_li,
insert_mbe, extract_mbe, insert_mb6, extract_mb6,
insert_nb, extract_nb, insert_nsi, extract_nsi,
insert_ral, insert_ram, insert_ras,
insert_rbs, extract_rbs, insert_sh6, extract_sh6,
insert_spr, extract_spr, insert_tbr, extract_tbr): Add dialect param.
(extract_bd, extract_bdm, extract_bdp,
extract_ds, extract_des,
extract_li, extract_nsi): Implement sign extension without conditional.
(insert_bdm, extract_bdm,
insert_bdp, extract_bdp, valid_bo): Handle 64 bit branch hints.
(extract_bdm, extract_bdp): Correct 32 bit validation.
(AT1_MASK, AT2_MASK): Define.
(BBOAT_MASK): Define.
(BBOATCB_MASK, BBOAT2CB_MASK, BBOATBI_MASK): Define.
(BOFM64, BOFP64, BOTM64, BOTP64): Define.
(BODNZM64, BODNZP64, BODZM64, BODZP64): Define.
(PPCCOM32, PPCCOM64): Define.
(powerpc_opcodes): Modify existing 32 bit insns with branch hints
and add new patterns to implement 64 bit branches with hints.  Move
booke instructions so they match before ppc64.
* ppc-dis.c (powerpc_dialect): Set PPC_OPCODE_64 in dialect for
64 bit default targets, and parse "32" and "64" in options.
Formatting fixes.
(print_insn_powerpc): Pass dialect to operand->extract.

*** empty log message ***

daily update

* s390-tdep.c (s390_pop_frame_regular): On the S/390, the frame
pointer and the SP are often the same, so we can't pop the frame
by setting the SP to the FP; we need to get the old SP from
saved_regs.

* s390-tdep.c (s390_extract_return_value): Returned `float' values
can simply be copied bitwise from the registers into the value
object's buffer.

* s390-tdep.c (s390_get_frame_info): Initialize SP's element of
the frame's saved_regs array correctly.

* symfile.c (simple_read_overlay_table): Make sure we can find
both `_novlys' and `_ovly_table' before we try anything else;
print a helpful error message.
(simple_overlay_update): No need to print error message here.

2001-11-14  Dave Brolley  <brolley@redhat.com>

* cgen-dis.c (count_decodable_bits): New function.
(add_insn_to_hash_chain): New function.
(hash_insn_array): Call add_insn_to_hash_chain.
(hash_insn_list): Call add_insn_to_hash_chain.

when #size-cells is zero, don't expect a size.

2001-11-14  Dave Brolley  <brolley@redhat.com>

* arch.c: Regenerate.
* arch.h: Regenerate.
* cpu.c: Regenerate.
* cpu.h: Regenerate.
* cpuall.h: Regenerate.
* cpux.c: Regenerate.
* cpux.h: Regenerate.
* decode.c: Regenerate.
* decode.h: Regenerate.
* decodex.c: Regenerate.
* decodex.h: Regenerate.
* model.c: Regenerate.
* modelx.c: Regenerate.
* sem-switch.c: Regenerate.
* sem.c: Regenerate.
* semx-switch.c: Regenerate.

2001-11-14  Dave Brolley  <brolley@redhat.com>

* arch.c: Regenerate.
* arch.h: Regenerate.
* cpu.c: Regenerate.
* cpu.h: Regenerate.
* cpuall.h: Regenerate.
* decode.c: Regenerate.
* decode.h: Regenerate.
* model.c: Regenerate.
* sem-switch.c: Regenerate.
* sem.c: Regenerate.

2001-11-14  Dave Brolley  <brolley@redhat.com>

* m32r-dis.c: Regenerated.
* fr30-dis.c: Regenerated.

* elf32-s390.c (elf_s390_relocate_section): Use the "unresolved_reloc"
scheme to get rid of an ugly complicated test.
* elf64-s390.c (elf_s390_relocate_section): Likewise.

2001-11-14  Michael Snyder  <msnyder@redhat.com>

* Makefile.in (doublest.o): Add dependency on gdbtypes.h.

Use frag_make_room() to grow the obstack.

* gdb.asm/asm-sources.exp: Allow defining linker flags.

Do not remove the symbol's value from the addend for fixups against local
symbols in SEC_MERGE sections - it was not added in, in the first place.

* cpu-i386.c (bfd_x86_64_arch_intel_syntax,bfd_x86_64_arch): Use
x86-64 for display.

* i386-dis.c (print_insn): Use x86-64 as option.

* doc/binutils.texi (objdump): Fix description to use x86-64.

Replace use of mkstemp with mkstemps, provided by libiberty.

Only scan tail of executable for PE targets.

* dwarf2.c (struct line_head) [total_length]: Change type to
bfd_vma.
(struct dwarf2_debug): Add dwarf_str_buffer and dwarf_str_size.
(struct comp_unit): Add stash and offset_size.
(read_8_bytes): Return bfd_vma.
(read_indirect_string): New.
(read_attribute): Handle DW_FORM_strp.
(decode_line_info): Support 64-bit DWARF format.
(_bfd_dwarf2_find_nearest_line): Likewise.
(parse_comp_unit): Set offset_size and stash.

* arm-tdep.c (arm_register_type): New function.
(arm_register_convertible): Delete.
(arm_register_convert_to_virtual): Delete.
(arm_register_convert_to_raw): Delete.
* config/arm/tm-arm.h (REGISTER_CONVERTIBLE): Delete.
(REGISTER_CONVERT_TO_VIRTUAL): Delete.
(REGISTER_CONVERT_TO_RAW): Delete.
(REGISTER_VIRTUAL_TYPE): Redefine as call to arm_register_type.
(arm_register_type): Declare.

2001-11-13  Jeff Holcomb  <jeffh@redhat.com>

Merged from net gcc:
2001-07-30  Jeff Sturm  <jsturm@one-pont.com>
* ltcf-c.sh: Use $objext, not $ac_objext.
2001-07-27  Mark Kettenis  <kettenis@gnu.org>
* ltcf-cxx.sh: Add support for GNU.
2001-07-22  Timothy Wall  <twall@redhat.com>
* ltcf-c.sh: Don't disable shared libraries for AIX5/IA64.  Preserve
default settings if using GNU tools with that configuration.
* ltcf-cxx.sh: Ditto.
* ltcf-gcj.sh: Ditto.
2001-07-21  Michael Chastain  <chastain@redhat.com>
* ltconfig: Set max_cmd_len to a maximum of 512Kb, as it seems some
HPUX 11.0 systems have trouble with 1MB.  Mark as gcc-local.
* ltmain.sh: Mark as gcc-local.

2001-11-13  Jeff Holcomb  <jeffh@redhat.com>

* Makefile.in (all-bison): Revert 2001-10-24.
Don't depend on texinfo.

binutils/ChangeLog
* doc/binutils.texi (objdump): Document x86 -M options.
include/ChangeLog
  * dis-asm.h (print_insn_i386): Declare.
opcodes/ChangeLog
* disassemble.c (disassembler): Call print_insn_i386.
* i386-dis.c (SUFFIX_ALWAYS): Define.
(struct dis_private): Add orig_sizeflag.
(print_insn_i386): Make it a wrapper, calling..
(print_insn): ..The old body of print_insn_i386.  Avoid longjmp
warning without using volatile by moving orig_sizeflag to priv,
and removing inbuf.  Parse disassembler_options.
(print_insn_i386_att, print_insn_i386_intel): Move initialisation
code to print_insn.
(putop): Remove #ifdef SUFFIX_ALWAYS.

*** empty log message ***

    * dwarf2.c (read_attribute_value): New function to handle
        DW_FORM_indirect.
        (read_attribute): Use it.

    * readelf.c (read_and_display_attr_value): New function to
        handle DW_FORM_indirect.
        (read_and_display_attr): Use it.

daily update

Index: bfd/ChangeLog
2001-11-11  Geoffrey Keating  <geoffk@redhat.com>

* dwarf2.c (decode_line_info): Properly deal with unknown standard
opcodes.

Index: binutils/ChangeLog
2001-11-11  Geoffrey Keating  <geoffk@redhat.com>

* readelf.c (display_debug_lines): Deal with unknown standard
opcodes.  Handle DW_LNS_set_prologue_end, DW_LNS_set_epilogue_begin,
DW_LNS_set_isa.

Index: include/elf/ChangeLog
2001-11-11  Geoffrey Keating  <geoffk@redhat.com>

* dwarf2.h (dwarf_line_number_ops): Add DWARF 3 opcodes.

2001-11-13  Elena Zannoni  <ezannoni@redhat.com>

From Andrew Cagney <cagney@redhat.com>:
* gdbtypes.h (builtin_type_int128, builtin_type_uint128): Declare.
* gdbtypes.c (_initialize_gdbtypes, build_gdbtypes): Initialize
new builtin types.

2001-11-07  Michael Snyder  <msnyder@redhat.com>

        * gdb.c++/templates.exp (test_template_breakpoints):
If we get an overload menu, but it does not match what
we expect, we still need to issue the "cancel" command.

* gdbint.texinfo (COERCE_FLOAT_TO_DOUBLE): Clarify.

2001-11-07  Michael Snyder  <msnyder@redhat.com>

        * gdb.c++/templates.exp: Replace "void \\*" with "void ?\\*",
making the whitespace optional.  Argument for "new" may be
"unsigned" as well as "unsigned int/long".

    * emulparams/elf64hppa.sh (OTHER_BSS_END_SECTIONS): Add
        additional symbols referenced by newer crt0.o files from HP.

* s390-tdep.c: Rewrite inferior function call code.  This may
break zSeries support; that should be fixed soon.
#include "gdb_assert.h".
(is_integer_like, is_pointer_like, is_simple_arg,
pass_by_copy_ref, extend_simple_arg, is_double_arg, round_up,
round_down, alignment_of): New functions.
(s390_push_arguments): Rewritten to handle passing large arguments
by value, and to make more readable.

* s390-tdep.c (s390_pop_frame): Call generic_pop_current_frame, to
interact correctly with generic dummy frames.
(s390_pop_frame_regular): Move the guts of the frame-popping code
to here, to be called by generic_pop_current_frame.  Use the
frame's saved_regs array; this works for `return' as well as
inferior function calls.

* s390-tdep.c (s390_gdbarch_init): Register the function
`standard_coerce_float_to_double', since GCC for the S/390 follows
the standard rules for passing floats.

* valops.c (default_coerce_float_to_double,
standard_coerce_float_to_double): Doc fixes.

Patch from Peter Schauer:

* symtab.c (find_pc_sect_line): If we can't find the function
containing PC, we certainly won't have line number information for
that location, so return zero immediately.

remove diff indicators

Fix tic54x testsuite failures and Lmem disassembly bugs.

merge from gcc

* section.c (bfd_make_section_anyway): Don't increment section_id
and the BFD's section_count if adding the section failed.
* vms.c (vms_new_section_hook): Adjust for bfd_make_section_anyway
change.

* config/obj-elf.c (obj_elf_version): Ensure terminating NUL is
put in note section.  Use sizeof instead of hard-coded constants.

Eliminate REGISTER_CONVERTIBLE().

Zap REGISTER_CONVERTIBLE from m88k.

* i386.h (i386_optab): Add entries for "sldr", "smsw" and "str" to
accept WordReg.
* i386-dis.c (grps): Change "sldt", "str", and "smsw" entries
to "sldtQ", "strQ", "smswQ" respectively; all with Ev operand
category instead of Ew.

*** empty log message ***

* elf32-cris.c (cris_elf_grok_prstatus): Adjust values to match new
core dump format.

daily update

* i386-tdep.h (FPU_REG_RAW_SIZE): Define unconditionally.
* i387-nat.c, i387-tdep.c: Unconditionally include "i386-tdep.h".
* config/i386/tm-i386.h (FPU_REG_RAW_SIZE): Removed.
(SIZEOF_FPU_REGS): Don't use FPU_REG_RAW_SIZE here.

* config/tc-mips.c (support_64bit_objects): Define for OBJ_ELF only.
(md_longopts): Allow OPTION_MABI for ELF compilation only. RE-allow
OPTION_GP32, OPTION_GP64, OPTION_FP32 for non-ELF compilation.
Sort options a bit more logical.
(md_parse_option): Allow OPTION_32, OPTION_N32, OPTION_N64,
OPTION_MABI only for elf targets.
* gas/mips/mips.exp: Change naming of some conditionals to reflect
the object format they actually mean. Don't try mips-abi32 and
mips-abi32-pic tests for ecoff.

* config/tc-mips.c (set_at): Add cast needed for varargs.
(load_register): Likewise.
(macro): Likewise. Some code reformatting.
(macro2): Add cast needed for varargs.
(mips16_macro): Likewise.

* config/tc-mips.c (md_assemble): Remove superflous casts.
(append_insn): Likewise.
(mips16_macro_build): Likewise.
(macro): Likewise.
(mips16_ip): Likewise.
(s_cpload): Likewise.
(mips_relax_frag): Likewise.

Patch from Andreas Schwab <schwab@suse.de>:
* eval.c (evaluate_subexp_standard): Fix memory leak: use alloca
instead of xmalloc.

* corefile.c (write_memory_unsigned_integer,
write_memory_signed_integer): New functions.
(write_memory): Move to be with other write_memory_* functions.
* gdbcore.h (write_memory_unsigned_integer,
write_memory_signed_integer): New declarations.

2001-11-12  Hans-Peter Nilsson  <hp@bitrange.com>

        * COPYING.NEWLIB: Add BSD-style license/copyright blurb for my work.

* dwarf2read.c (dwarf_decode_lines): Properly deal with
unknown standard opcodes.

    * lib/mi-support.exp (mi_run_to_helper): Move comments
        outside of gdb_expect.

* config/tc-mips.c (mips_abi_level): Move in front of
mips_set_options.
(mips_set_options): Add members gp32, fp32, abi.
(file_mips_gp32): New flag.
(file_mips_fp32): New flag.
(mips_opts): Initialize the new members.
(mips_gp32): Remove.
(mips_fp32): Remove.
(HAVE_32BIT_GPRS): Use the new values from mips_opts.
(HAVE_32BIT_FPRS): Likewise.
(HAVE_NEWABI): Likewise.
(HAVE_64BIT_OBJECTS): Likewise.
(md_begin): Likewise. Save default (file) values.
(md_parse_option): Use the new values from mips_opts.
(s_mipsset): Likewise. Fix logic to keep the ABI selection if
possible. Let .set mipsN work together with .set push/pop.
Enhance error messages.
(mips_elf_final_processing): Use file_mips_* for header processing.

Fix definitions of wddata[bwl].

Only compare the tail end of the two binary files produced in order to avoid
date stamp present in PE binaries.

Use mkstemp in place of mktemp.

New symbol: DATA_START_SYMBOLS

* Makefile.am (GENSCRIPTS): Quote ${exec_prefix}.
* Makefile.in: Regenerate.

*** empty log message ***

2001-11-11  H.J. Lu <hjl@gnu.org>

* elf32-mips.c (_bfd_mips_elf_finish_dynamic_sections): Call
_bfd_elf_strtab_size instead of _bfd_stringtab_size.

* elf64-sparc.c (sparc64_elf_size_dynamic_sections): Change
dynstr type to `struct elf_strtab_hash *'.

daily update

* elf-strtab.c (_bfd_elf_strtab_finalize): Make first variable i
size_t.  Rename second i to j.

    * lib/mi-support.exp: (mi_run_to_helper, mi_run_to,
        mi_step_to, mi_next_to, mi_continue_to, mi_finish_to,
        mi0_step_to, mi0_next_to, mi0_continue_to, mi0_finish_to,
        mi0_run_to): New functions.
        * gdb.mi/mi-simplerun.exp: Use them.
        * gdb.mi/mi0-simplerun.exp: Likewise.
        * gdb.mi/mi-var-cmd.exp: Likewise.
        * gdb.mi/mi0-var-cmd.exp: Likewise.

Change target name from stormy16 to xstormy16.

* README (alpha-dec-osf5.1): Mention -DUSE_LDR_ROUTINES.
(sparcv9-sun-solars2.8): Mention problem with 64 bit GCC 3.0.x.
(i586-sco-sysv5uw7.1.1): Mention problem with threads.

From Mark Kettenis  <kettenis@gnu.org>:
* breakpoint.c (breakpoint_re_set_one): Don't discard SHLIB
breakpoints when they fail.