daily update

Rework Stringpool to not compute the hash code twice when adding a new
string.

* elf32-spu.c (spu_elf_size_stubs): Do consider branches to
non-function symbols for overlay stubs.

* config.sub, config.guess: Update from upstream sources.

From Craig Silverstein: Support -o -.

From Craig Silverstein: Test some flags.

Stringpool stats.  Also make Symbol_table support functions inline.

*** empty log message ***

Support lexical blocks and function bodies that occupy
non-contiguous address ranges.
* addrmap.c, addrmap.h: New files.
* block.h (struct addrmap): New forward declaration.
(struct blockvector): New member, 'map'.
(BLOCKVECTOR_MAP): New accessor macro.
* block.c: #include "addrmap.h"
(blockvector_for_pc_sect): If the blockvector we've found has
an address map, use it instead of searching the blocks.
* buildsym.c: #include "addrmap.h"
(pending_addrmap_obstack, pending_addrmap_interesting): New static
variables.
(really_free_pendings): If we have a pending addrmap, free it too.
(record_block_range): New function.
(make_blockvector): If we have an interesting pending addrmap,
record it in the new blockvector.
(start_symtab, buildsym_init): Assert that there is no pending
addrmap now; we should have cleaned up any addrmaps we'd built
previously.
(end_symtab): If there is a pending addrmap left over that didn't
get included in the blockvector, free it.
* buildsym.h (struct addrmap): New forward declaration.
(record_block_range): New prototype.
* objfiles.c: #include "addrmap.h".
(objfile_relocate): Relocate the blockvector's address map, if
present.
* dwarf2read.c (dwarf2_record_block_ranges): New function.
(read_func_scope, read_lexical_block_scope): Call it.
* Makefile.in (SFILES): Add addrmap.c.
(addrmap_h): New header dependency variable.
(COMMON_OBS): Add addrmap.o.
(addrmap.o): New rule.l
(block.o, objfiles.o, buildsym.o): Depend on $(addrmap_h).

* block.c (blockvector_for_pc, blockvector_for_pc_sect): Return a
pointer to the block, not its index in the blockvector.
(block_for_pc_sect): Use the returned block, instead of looking it
up ourselves.
* block.h (blockvector_for_pc, blockvector_for_pc_sect): Update
declarations.
* breakpoint.c (resolve_sal_pc): Use returned block, instead of
looking it up ourselves.
* stack.c (print_frame_label_vars): Disable function, which
depends on the block's index.

* buildsym.c (finish_block): Return the block we've built.
* buildsym.h (finish_block): Update prototype.

* defs.h (CORE_ADDR_MAX): New constant.

From Craig Silverstein: Support irregular output files.

* coffread.c (decode_type): Use builtin_type_int32 instead
of FT_INTEGER fundamental type for array range index type.
(decode_base_type): Use builtin types of current_gdbarch
instead of fundamental types.

* dwarf2read.c (struct dwarf2_cu): Remove ftypes member.
(read_file_scope): Do not initialize ftypes member.
(dwarf_base_type, dwarf2_fundamental_types): Remove functions.
(read_array_type): Use builtin_type_int32 instead of FT_INTEGER
fundamental type for array range index type.
(read_tag_string_type): Likewise for string range index type.
Also, do not overwrite FT_CHAR type with new string type.
(read_base_type): If DW_AT_name is missing, create unnamed type
with given properties instead of looking for a fundamental type.
Create new types as TYPE_TARGET_TYPE for DW_ATE_address and
DW_ATE_complex_float types.
(read_subrange_type): Create new type to represent missing
DW_AT_type instead of looking for a fundamental type.
(die_type): Use builtin type to represent "void" instead of
looking for a fundamental type.

* stabsread.c (define_symbol): Use builtin types to represent
'r' and 'i' floating-point and integer constants.

* gdbtypes.c (lookup_fundamental_type): Remove.
* gdbtypes.h (lookup_fundamental_type): Remove prototype.
(FT_VOID, FT_BOOLEAN, FT_CHAR, FT_SIGNED_CHAR, FT_UNSIGNED_CHAR,
FT_SHORT, FT_SIGNED_SHORT, FT_UNSIGNED_SHORT, FT_INTEGER,
FT_SIGNED_INTEGER, FT_UNSIGNED_INTEGER, FT_LONG, FT_SIGNED_LONG,
FT_UNSIGNED_LONG, FT_LONG_LONG, FT_SIGNED_LONG_LONG,
FT_UNSIGNED_LONG_LONG, FT_FLOAT, FT_DBL_PREC_FLOAT, FT_EXT_PREC_FLOAT,
FT_COMPLEX, FT_DBL_PREC_COMPLEX, FT_EXT_PREC_COMPLEX, FT_STRING,
FT_FIXED_DECIMAL, FT_FLOAT_DECIMAL, FT_BYTE, FT_UNSIGNED_BYTE,
FT_TEMPLATE_ARG, FT_DECFLOAT, FT_DBL_PREC_DECFLOAT,
FT_EXT_PREC_DECFLOAT, FT_NUM_MEMBERS): Remove macros.
* objfiles.c (struct objfile): Remove fundamental_types member.
* symfile.c (reread_symbols): Do not clear fundamental_types.

* language.h (struct language_defn): Remove la_fund_type member.
(create_fundamental_type): Remove.
* language.c (unk_lang_create_fundamental_type): Remove.
(unknown_language_defn, auto_language_defn,
local_language_defn): Adapt initializer.
* ada-lang.c (ada_create_fundamental_type): Remove.
(ada_language_defn): Adapt initializer.
* c-lang.h (c_create_fundamental_type): Remove prototype.
* c-lang.c (c_create_fundamental_type): Remove.
(c_language_defn, cplus_language_defn, asm_language_defn,
minimal_language_defn): Adapt initializer.
* f-lang.c (f_create_fundamental_type): Remove.
(f_language_defn): Adapt initializer.
* jv-lang.c (java_create_fundamental_type): Remove.
(java_language_defn): Adapt initializer.
* m2-lang.c (m2_create_fundamental_type): Remove.
(m2_language_defn): Adapt initializer.
* objc-lang.c (objc_create_fundamental_type): Remove.
(objc_language_defn): Adapt initializer.
* p-lang.h (pascal_create_fundamental_type): Remove prototype.
* p-lang.c (pascal_create_fundamental_type): Remove.
(pascal_language_defn): Adapt initializer.
* scm-lang.c (scm_language_defn): Adapt initializer.

Print symbol table statistics with --stats.

daily update

2007-11-26  Thiago Jung Bauermann  <bauerman@br.ibm.com>

* Makefile.in (decimal32.o): Prepend $(srcdir) to dependencies
and substitute $< for the source file in compilation command.
(decimal64.o): Likewise.
(decimal128.o): Likewise.
(bid2dpd_dpd2bid.o): Likewise.
(host-ieee32.o): Likewise.
(host-ieee64.o): Likewise.
(host-ieee128.o): Likewise.

* elf32-xtensa.c (elf_xtensa_size_dynamic_sections): Create DT_PLTGOT
entry for all dynamic objects.

* features/mips-cpu.xml: Specify "regnum" for "r0" explicitly.
* features/mips64-cpu.xml: Likewise.
* features/mips-fpu.xml: Specify "regnum" for "f0" explicitly.
* features/mips64-fpu.xml: Likewise.
* features/mips-linux.c: Regenerate.
* features/mips64-linux.c: Regenerate.

* target.h (target_find_new_threads): Fix definition.

* elf32-spu.c (spu_elf_size_stubs): Correct section alignment.

From Craig Silverstein: Use gold_fatal in options.h, beef up value checks.

*** empty log message ***

* readelf.c (is_32bit_abs_reloc, is_32bit_pcrel_reloc): Add SPU.

* config/tc-ppc.c (ppc_tc): Allow a space between toc symbol
name and bracket.

daily update

* win32-low.c (attaching): New global.
(win32_create_inferior): Clear the `attaching' global.
(win32_attach): Set the `attaching' global.
(get_child_debug_event) [_WIN32_WCE]: Stop the inferior when
attaching.  Only set a breakpoint at the entry point if not
attaching.

* server.c (main): Don't report dll events on the initial
connection on attaches.

* server.c (main): Relax numerical bases supported for the pid of
the --attach command line argument.

* win32-low.c (win32_attach): Call OpenProcess before
DebugActiveProcess, not after.  Add last error output to error
call.

* win32-low.c (win32_get_thread_context)
(win32_set_thread_context): New functions.
(thread_rec): Use win32_get_thread_context.
(continue_one_thread, win32_resume): Use win32_set_thread_context.
* win32-low.h (win32_thread_info) [_WIN32_WCE]: Add `base_context'
field.

* win32-low.c (soft_interrupt_requested, faked_breakpoint): New
global variables.
(child_add_thread): Minor cleanup.
(child_continue): Resume artificially suspended threads before
calling ContinueDebugEvent.
(suspend_one_thread): New.
(fake_breakpoint_event): New.
(get_child_debug_event): Change return type to int.  Check here if
gdb sent an interrupt request.  If a soft interrupt was requested,
fake a breakpoint event.  Return 0 if there is no event to handle,
and 1 otherwise.
(win32_wait): Don't check here if gdb sent an interrupt request.
Ensure there is a valid event to handle.
(win32_request_interrupt): Add soft interruption method as last
resort.

* win32-low.h (win32_thread_info): Add descriptions to the
structure members.  Replace `suspend_count' counter by a
`suspended' flag.
* win32-low.c (thread_rec): Update condition of when to get the
context from the inferior.  Rely on ContextFlags being set if it
has already been retrieved.  Only suspend the inferior thread if
we haven't already.  Warn if that fails.
(continue_one_thread): s/suspend_count/suspended/.  Only call
ResumeThread once.  Warn if that fails.

*** empty log message ***

* win32-low.c (win32_wait): Don't read from the inferior when it
has already exited.

* Makefile.in (win32_low_h): New variable.
(win32-low.o): Add dependency on $(win32_low_h).
(win32-arm-low.o, win32-i386-low.o): New rules.

daily update

2007-12-02  Pierre Muller  <muller@ics.u-strasbg.fr>

* win32-nat.c (win32_resume): Set the trace bit
in the thread identified by inferior_ptid.

*win32-nat.c (open_process_used): New static variable.
(win32_init_thread_list): Remove call to CloseHandle for thread.
(win32_delete_thread): Ditto.
(fake_create_process): Set open_process_used if OpenProcess call is successful.
(get_win32_debug_event): Do not close process handle.
(do_initial_win32_stuff): Set open_process_used to zero.
(win32_mourn_inferior): Call CloseHandle for current_process_handle if
open_process_used is set.
(win32_kill_inferior): Do not close process and main_thread handles.

*** empty log message ***

daily update

More -fPIC options.

Add missing space.

Compile shared library code with -fPIC.  Also, alphabetize.

Add casts to avoid warnings.

Compress all debug sections.

* hostio.c: Correct copyright year.

*** empty log message ***

gas/
* config/tc-xtensa.h (md_allow_eh_opt): Define.
gas/testsuite/
* gas/elf/elf.exp: Disable ehopt test for Xtensa.

daily update

From Cary Coutant: Avoid compilation warnings if no zlib.h.

* remote.c (remote_cmdlist): New variable.
(PACKET_vFile_open, PACKET_vFile_pread, PACKET_vFile_pwrite)
(PACKET_vFile_close, PACKET_vFile_unlink): New constants.
(remote_buffer_add_string, remote_buffer_add_bytes)
(remote_buffer_add_int, remote_hostio_parse_result)
(remote_hostio_send_command, remote_hostio_open, remote_hostio_pwrite)
(remote_hostio_pread, remote_hostio_close, remote_hostio_unlink)
(remote_fileio_errno_to_host, remote_hostio_error, fclose_cleanup)
(remote_hostio_close_cleanup, remote_file_put, remote_file_get)
(remote_file_delete, remote_put_command, remote_get_command)
(remote_delete_command, remote_command): New functions.
(_initialize_remote): Register new packets and commands.
* Makefile.in (gdb_fileio_h): New variable.
(remote.o): Update.
(SUBDIR_MI_OBS): Add mi-cmd-target.o.
(SUBDIR_MI_SRCS): Add mi/mi-cmd-target.c.
(mi-cmd-target.o): New rule.
* mi/mi-cmd-target.c: New file.
* mi/mi-cmds.c (mi_cmds): Add target-file-delete, target-file-get,
and target-file-put.
* mi/mi-cmds.h (mi_cmd_target_file_get, mi_cmd_target_file_put)
(mi_cmd_target_file_delete): Declare.
* remote.h (remote_file_put, remote_file_get, remote_file_delete):
Declare.
* NEWS: Describe new file transfer support.

* gdb.texinfo (Debugging Programs with Multiple Processes): Correct
formatting.
(Remote Debugging): Add File Transfer section.
(Remote Configuration): Document Host I/O packets.
(GDB/MI): Add GDB/MI File Transfer Commands section.
(Remote Protocol): Add Host I/O Packets section.
(Packets): Add vFile.

* Makefile.in (OBS): Add hostio.o.
(hostio.o): New rule.
* server.h (handle_vFile): Declare.
* hostio.c: New file.
* server.c (handle_v_requests): Take packet_len and new_packet_len
for binary packets.  Call handle_vFile.
(main): Update call to handle_v_requests.

* gdb.server/file-transfer.exp, gdb.server/transfer.txt,
gdb.mi/mi-file-transfer.exp: New.

* infrun.c (handle_inferior_event): Don't
        ignore beakpoints if trap_expected is set.

Make insert_breakpoints return void.
* breakpoint.h (insert_breakpoints): Change
return type to void.
* breakpoint.c (insert_breakpoints): Change
return type to void.  Rename local return_val
variable to error.
* infrun.c (keep_going): Instead of checking
return value from insert_breakpoints, catch exception.

bfd/
* elf-eh-frame.c (_bfd_elf_discard_section_eh_frame): Use the
ELF header to choose between 32-bit and 64-bit relocation structures.

From Craig Silverstein: Add support for compressing .debug_str section.

*** empty log message ***

daily update

Clean up setting address and section offset.

2007-11-28  Ralf Wildenhues  <Ralf.Wildenhues@gmx.de>

        * config-ml.in: Robustify against white space in absolute file
        names.

bfd/
* archures.c (bfd_mach_mips_loongson_2e): New.
(bfd_mach_mips_loongson_2f): New.
* bfd-in2.h (bfd_mach_mips_loongson_2e): New.
(bfd_mach_mips_loongson_2f): New.
* cpu-mips.c: Add I_loongson_2e and I_loongson_2f to
anonymous enum.
(arch_info_struct): Add Loongson-2E and Loongson-2F entries.
* elfxx-mips.c (_bfd_elf_mips_mach): Handle Loongson-2E
and Loongson-2F flags.
(mips_set_isa_flags): Likewise.
(mips_mach_extensions): Add Loongson-2E and Loongson-2F
entries.

binutils/
* readelf.c (get_machine_flags): Handle Loongson-2E and -2F
flags.

gas/
* config/tc-mips.c (mips_cpu_info_table): Add loongson2e
and loongson2f entries.
* doc/c-mips.texi: Document -march=loongson{2e,2f} options.

gas/testsuite/
* gas/mips/mips.exp: Add loongson-2e and -2f tests.
* gas/mips/loongson-2e.d: New.
* gas/mips/loongson-2e.s: New.
* gas/mips/loongson-2f.d: New.
* gas/mips/loongson-2f.s: New.

include/elf/
* mips.h (E_MIPS_MACH_LS2E): New.
(E_MIPS_MACH_LS2F): New.

include/opcode/
* mips.h (INSN_LOONGSON_2E): New.
(INSN_LOONGSON_2F): New.
(CPU_LOONGSON_2E): New.
(CPU_LOONGSON_2F): New.
(OPCODE_IS_MEMBER): Update for Loongson-2E and -2F flags.

opcodes/
* mips-dis.c (mips_arch_choices): Add Loongson-2E and -2F
entries.
* mips-opc.c (IL2E): New.
(IL2F): New.
(mips_builtin_opcodes): Add Loongson-2E and -2F instructions.
Allow movz and movn for Loongson-2E and -2F.  Add movnz entry.
Move coprocessor encodings to the end of the table.  Allow
certain MIPS V .ps instructions on the Loongson-2E and -2F.

include/opcode/
* mips.h (INSN_ISA*): Redefine certain values as an
enumeration.  Update comments.
(mips_isa_table): New.
(ISA_MIPS*): Redefine to match enumeration.
(OPCODE_IS_MEMBER): Modify to correctly test new INSN_ISA*
values.

opcodes/
* mips-opc.c (I3_32, I3_33, I4_32, I4_33, I5_33): New.
(mips_builtin_opcodes): Use these new I* values.

2007-11-29  Martin Schwidefsky  <schwidefsky@de.ibm.com>

* config/tc-s390.c (md_begin): If the -mesa option is specified
add zarch opcodes to the hash table only if there is no variant
that is available for the esa mode as well.

2007-11-29  Martin Schwidefsky  <schwidefsky@de.ibm.com>

* gas/s390/esa-z9-109.d: Add check for old version of sske.
* gas/s390/esa-z9-109.s: Likewise.

PR ld/5398
* elf32-arm.c (bfd_elf32_arm_process_before_allocation): Do not complain if
  there is no glue bfd, just return.

Stop infrun from tracking breakpoint insertion status.

The checks of breakpoints_inserted before calling
remove_breakpoints are removed, as remove_breakpoint
won't touch uninserted breakpoints. In a number of places,
we're interested if a breakpoint is inserted at particular
PC, and we now use breakpoint_inserted_here_p.  In a few
places, insert_breakpoints can be called unconditionally,
since it won't try to insert already inserted breakpoint.

* breakpoint.h (regular_breakpoint_inserted_here_p): New
declaration.
* breakpoint.c (regular_breakpoint_inserted_here_p): New.
(breakpoint_inserted_here_p): Use
regular_breakpoint_inserted_here_p.
* infrun.c (breakpoints_inserted): Remove.
(resume): Don't check for breakpoints_inserted before
remove_hw_watchpoints. Use breakpoint_inserted_here_p.
(proceed, init_wait_for_inferior): Don't set breakpoints_inserted.
(handle_inferior_event): Don't use breakpoints_inserted.
Use breakpoints_meant_to_be_inserted and
breakpoints_inserted_here_p.
(insert_step_resume_breakpoint_at_sal, keep_going): Use
breakpoints_meant_to_be_inserted. Don't set breakpoints_inserted.
(normal_stop): Don't check for breakpoints_inserted.  Don't
set breakpoints_inserted.
(keep_going): Don't check for breakpoints_inserted.
(insert_step_resume_breakpoint_at_sal): Don't insert
breakpoints

*** empty log message ***

daily update

Don't try to delete the output file if we don't know its name yet.

From Craig Silverstein: Fix up comments.

* breakpoint.c (watch_command_1): When the watchpoint isn't local
to any frame, initialize watchpoint_frame using null_frame_id, not
a memset.

From Craig Silverstein: Better handling of ODR violations which are
not in a function.

* internal.h (ELF_IS_SECTION_IN_SEGMENT): Adjust to cope with
segments at the end of memory.

* elf-m10300.c (mn10300_elf_relax_section): Extend previous fix to cover forward jumps.
* ld-mn10300/i135409-5.s: New test case.  Check for relaxation to a 16-bit backward jump instruction.
* ld-mn10300/i135409-5.t: Linker script for the new test.
* ld-mn10300/i135409-5.d: Expected disassembly of new test.
* ld-mn10300/mn10300.exp: Run the new test.

* infrun.c (resume): Set right thread even if
stepping over breakpoint using software single step.

*** empty log message ***

(print_varobj): Revert change from 2007-08-31.
(mi_print_value_p): Guard against type = NULL.

*** empty log message ***

daily update

* dfp.c (decimal_from_string): Remove superfluous newline from
error string.
(decimal_to_string): Likewise.
* printcmd.c (printf_command): Change string buffer to use
MAX_DECIMAL_STRING constant.
* value.c (value_from_decfloat): Likewise.

* Makefile.in (ALL_TARGET_OBS): Remove object files that require
64-bit CORE_ADDR and BFD support, move them to ...
(ALL_64_TARGET_OBS): ... this new variable.
* configure.ac: Check for --enable-64-bit-bfd option.  Only add
64-bit targets with --enable-targets=all if BFD supports 64-bit.
* configure: Regenerate.

* i386-cygwin-tdep.c (i386_cygwin_osabi_sniffer): Do not claim
all elf32-i386 executables, only cygwin core files.

2007-11-27  Andreas Krebbel  <krebbel1@de.ibm.com>

* s390-opc.txt ("tcet", "tcdt", "tcxt", "tget", "tgdt",
"tgxt"): Removed.
("tdcet", "tdcdt", "tdcxt", "tdget", "tdgdt", "tdgxt"): Added.

2007-11-27  Andreas Krebbel  <krebbel1@de.ibm.com>

* s390-opc.txt ("tcet", "tcdt", "tcxt", "tget", "tgdt",
"tgxt"): Removed.
("tdcet", "tdcdt", "tdcxt", "tdget", "tdgdt", "tdgxt"): Added.

2007-11-27  Andreas Krebbel  <krebbel1@de.ibm.com>

* gas/s390/zarch-z9-ec.d: ("tcet", "tcdt", "tcxt", "tget",
"tgdt", "tgxt"): Removed.
        ("tdcet", "tdcdt", "tdcxt", "tdget", "tdgdt", "tdgxt"): Added.
* gas/s390/zarch-z9-ec.s: Likewise.

Rework merge_map for speed.

*** empty log message ***

daily update

* cxxfilt.c (demangle_it): Don't call printf without format string.

* config/tc-alpha.c (assemble_insn): Don't segv on NULL reloc_operand.

*** empty log message ***

daily update

gdb/
* dwarf2read.c (dwarf2_get_pc_bounds): Moved the `DW_AT_ranges' parsing
code with its variables OBJFILE, CU_HEADER and OBFD into ...
(dwarf2_ranges_read): ... a new function.
(read_partial_die): Implemented the parsing of `DW_AT_ranges'.

gdb/testsuite/
* gdb.dwarf2/dw2-ranges.S, gdb.dwarf2/dw2-ranges.exp: New files.

*** empty log message ***

daily update

* win32-nat.c (DR6_CLEAR_VALUE): New define.
(thread_info_struct): Rename suspend_count to suspended, to be
used as a flag.
(thread_rec): Only suspend the thread if it wasn't suspended by
gdb before.  Warn if suspending failed.
(win32_add_thread): Set Dr6 to DR6_CLEAR_VALUE.
(win32_continue): Set Dr6 to DR6_CLEAR_VALUE.  Update usage of the
`suspended' flag.  Do ContinueDebugEvent after resuming the
suspended threads, not before.  Set threads' contexts before
resuming them, not after.
(win32_resume): Set Dr6 to DR6_CLEAR_VALUE.

*** empty log message ***

daily update

* breakpoint.c (insert_breakpoints)
(insert_bp_location): Remove stale comments.

* mips-linux-tdep.h: Fix some formatting.

*** empty log message ***

daily update

* gdb.trace/backtrace.exp: Fix a typo.
* gdb.trace/circ.exp: Likewise.
* gdb.trace/collection.exp: Likewise.
* gdb.trace/limits.exp: Likewise.
* gdb.trace/report.exp: Likewise.
* gdb.trace/tfind.exp: Likewise.
* gdb.trace/while-dyn.exp: Likewise.

Make Merge_key_less operator() inline.

Add threading support.

*** empty log message ***