2013-04-30  Jan-Benedict Glaw  <jbglaw@lug-owl.de>

config/
* picflag.m4: Merge from GCC.
* dfp.m4: Ditto.
* bootstrap-asan.mk: Ditto.
* ChangeLog: Ditto.

Fix display of structures/bitfields in register description.

Add support for displaying structures and bitfields for registers when
executing "maint print c-tdesc". This command is also used when
converting the xml target description file into c file.

Example of the behaviour is given below reporting a snipet of the xml file
and a snippet of the c code generated.

XML file contains:
...
    <union id="vecint">
      <field name="v4" type="v4int8"/>
      <field name="v2" type="v2int16"/>
    </union>

    <struct id="struct1">
      <field name="v4" type="v4int8"/>
      <field name="v2" type="v2int16"/>
    </struct>

    <struct id="struct2" size="8">
      <field name="f1" start="0" end="34"/>
      <field name="f2" start="63" end="63"/>
    </struct>
...

Setting this xml file as target description file and
issuing the maintenance print c-tdesc the following output
is obtained:

  feature = tdesc_create_feature (result, "extra");
  field_type = tdesc_named_type (feature, "int8");
  tdesc_create_vector (feature, "v4int8", field_type, 4);

  field_type = tdesc_named_type (feature, "int16");
  tdesc_create_vector (feature, "v2int16", field_type, 2);

  type = tdesc_create_union (feature, "vecint");
  field_type = tdesc_named_type (feature, "v4int8");
  tdesc_add_field (type, "v4", field_type);
  field_type = tdesc_named_type (feature, "v2int16");
  tdesc_add_field (type, "v2", field_type);

C output is not supported type "struct1".

This is finally the issue.

2013-03-27  Walfred Tedeschi  <walfred.tedeschi@intel.com>

        * target-descriptions.c (maint_print_c_tdesc_cmd):
        Add case to parse structures as register types and
        bitfields.

testsuite/

* gdb.xml/maint_print_struct.exp: New file.
* gdb.xml/maint_print_struct.xml: New file.

Change-Id: I2e20b095d508319c80275e724a9452c7e2834067
Signed-off-by: Walfred Tedeschi <walfred.tedeschi@intel.com>

PR binutils/15417
* elflink.c (elf_link_add_object_symbols): Initialise 'idx' to
zero.

Fix date in some of my recent ChangeLog entries...

Fix -Wpointer-sign warning in sol-thread.c

This fixes a couple of compiler warnings in rw_common when calling
target_read_memory/target_write_memory due to the type of parameter
"buf" (char *) not matching what these function expect (gdb_byte *).

gdb/ChangeLog:

        * sol-thread.c (rw_common): Change type of parameter "buf"
        to "gdb_byte *".
        (ps_pdwrite, ps_ptwrite): Cast parameter "buf" in call to
        rw_common to "gdb_byte *" instead of "char *".

Fix build error in sol-thread.c:info_cb

gdb/ChangeLog:

        * sol-thread.c (info_cb) [ti.ti_startfunc != 0]: Change type
        of local variable msym to const struct bound_minimal_symbol.
        Adjust use accordingly.
        [ti.ti_state == TD_THR_SLEEP]: Likewise.

Add myself as a maintainer.

Change-Id: Ie1b0cb082a384144eef327e9294949e9bacc4415

2013-04-30  Samuel Thibault  <samuel.thibault@gnu.org>

* i386gnu-nat.c (CREG_OFFSET): New macro.
(creg_offset): New array.
(CREG_ADDR): Use creg_offset instead of reg_offset.

mep: define gdbarch_pc_regnum instead of gdbarch_write_pc.

gdb/ChangeLog:

    * mep-tdep.c (mep_write_pc): Delete.
    (mep_gdbarch_init): Remove call to set_gdbarch_write_pc.
    Add call to set_gdbarch_pc_regnum.

Delete ChangeLog entry with no corresponding change in sources.

This ChangeLog entry was the result of a result of botched commit
which resulted in the ChangeLog update being checked in, while
the actual code code itself did not make it to CVS.

This patch deletes the entry in gdb/ChangeLog titled:

    mep: define gdbarch_pc_regnum instead of gdbarch_write_pc.

filestuff.c: Include "gdb_dirent.h" instead of <dirent.h>

gdb/ChangeLog:

        * common/filestuff.c: Replace #include <dirent.h> by
        #include "gdb_dirent.h".

filestuff.c: Use gdb_stat.h instead of <sys/stat.h>

gdb/ChangeLog:

        * common/filestuff.c: Replace #include <sys/stat.h> by
        #include "gdb_stat.h".

* elflink.c (bfd_elf_record_link_assignment): Dont make
STV_INTERNAL symbols STV_HIDDEN.

daily update

*** empty log message ***

PR python/14204:
* gdb.texinfo (Python API): Fix menu entry.
(Blocks In Python): Fix subsubsection text.  Rewrite intro.
Define global and static block.  Add example.  Clarify
block relationship for inline functions.

* gdb.texinfo (Python API): Mention Python help and keyword
arguments.  Move pagination text to Basic Python.
(Basic Python): Put pagination text here.  Document
close-on-exec, SIGCHLD, and SIGINT.

2013-04-29  Alexander Ivchenko  <alexander.ivchenko@intel.com>

gold/
* output.cc (Output_section::add_merge_input_section): Allow
to merge sections if the alignment is more than character size.
* merge.h (Output_merge_string::Output_merge_string): Remove
assert.
* merge.cc (Output_merge_string<Char_type>::do_add_input_section): Count
only not-null strings. Check the alignment of strings.
* stringpool.h
(Stringpool_template<Stringpool_char>::Stringpool_template): Add
alignment as the argument.
(Stringpool_template<Stringpool_char>::addralign_): New class member.
* stringpool.cc (Stringpool_template<Stringpool_char>::new_key_offset):
Align non-zero length strings according to the addralign_.
(Stringpool_template<Stringpool_char>::set_string_offsets):
Updating offsets according to the given alignment.
* testsuite/Makefile.am (text_section_grouping): Test if string
literals are getting merged.
* testsuite/Makefile.in: Regenerate.
* testsuite/merge_string_literals_1.c: New file.
* testsuite/merge_string_literals_2.c: Ditto.
* testsuite/merge_string_literals.sh: Ditto.

* config.guess: Update from config repo.
* config.sub: Ditto.

* elflink.c (_bfd_elf_gc_mark_extra_sections): Remove mark from
fragmented .debug_line sections associated with unmarked code
sections.

* dwarf.c (read_debug_line_header): New function.  Reads in a
header in a .debug_line section.
(display_debug_lines_raw): Use new function.  Handle fragmentary
.debug_line sections.
(display_debug_lines_decoded): Likewise.
* readelf.c (process_section_headers): Handle fragmenatry
.debug_line sections.
(display_debug_section): Likewise.

* as.c (Options): Add -gdwarf-sections.
(parse_args): Likewise.
* as.h (flag_dwarf_sections): Declare.
* dwarf2dbg.c (emit_fixed_inc_line_addr): Skip section changes.
(process_entries): When -gdwarf-sections is enabled generate
fragmentary .debug_line sections.
(out_debug_line): Set the section for the .debug_line section end
symbol.
* doc/as.texinfo: Document -gdwarf-sections.
* NEWS: Mention -gdwarf-sections.

* gas/elf/dwarf2-3.d: Fix expected readelf output.

* scripttempl/DWARF.sc: Add support for .debug_line.* and
.debug_line_end.

mep: define gdbarch_pc_regnum instead of gdbarch_write_pc.

gdb/ChangeLog:

* mep-tdep.c (mep_write_pc): Delete.
(mep_gdbarch_init): Remove call to set_gdbarch_write_pc.
Add call to set_gdbarch_pc_regnum.

Use gdbarch_pc_regnum instead of gdbarch_write_pc.

gdb/ChangeLog:

* m32r-tdep.c (m32r_write_pc): Delete.
(m32r_gdbarch_init): Remove call to set_gdbarch_write_pc.
Add call to set_gdbarch_pc_regnum.

* dwarf2read.c (handle_DW_AT_stmt_list): Avoid ARI warning for
editCase function rule.
(get_DW_AT_signature_type): Likewise.

bfd/efl32-arm.c: Fix handling of IRELATIVE relocs.

bfd/ChangeLog:

2013-04-24  Will Newton  <will.newton@linaro.org>

* elf32-arm.c (elf32_arm_populate_plt_entry): Call
elf32_arm_add_dynreloc when emitting R_ARM_IRELATIVE relocs.

ld/testsuite/ChangeLog:

2013-04-24  Will Newton  <will.newton@linaro.org>

* ld-arm/arm-elf.exp: Add IFUNC test 17.
* ld-arm/ifunc-17.dd: New file.
* ld-arm/ifunc-17.gd: Likewise.
* ld-arm/ifunc-17.rd: Likewise.
* ld-arm/ifunc-17.s: Likweise.
* ld-arm/ifunc-1.rd: Reorder relocs to match linker output.
* ld-arm/ifunc-2.rd: Likewise.
* ld-arm/ifunc-5.rd: Likewise.
* ld-arm/ifunc-6.rd: Likewise.

ld-plugin/lto.exp: Disable ld/12942 test for gcc < 4.7.0.

The test for ld/12942 fails with gcc versions before 4.7.0. This patch
disables this test with these versions of gcc.

ld/testsuite/ChangeLog:

2013-04-29  Will Newton  <will.newton@linaro.org>

* ld-plugin/lto.exp: Disable ld/12942 test for gcc < 4.7.0.

* elf64-aarch64.c (elf64_aarch64_check_relocs): Move relocation
error check up and add error message.

* ./contrib/ari/gdb_ari.sh (editCase rule): Fix spelling error.

* emultempl/pe.em [cygwin]: Do not merge rdata with v2
psuedo-relocs.

Add ARI marker to get_DW_AT_signature_type.

gdb/ChangeLog:

        * dwarf2read.c (get_DW_AT_signature_type): Add ARI marker.

daily update

*** empty log message ***

ld/
* scripttempl/armbpabi.sc: Replace "source" usage with ".".
* scripttempl/avr.sc: Likewise.
* scripttempl/elf.sc: Likewise.
* scripttempl/elf32cr16.sc: Likewise.
* scripttempl/elf32crx.sc: Likewise.
* scripttempl/elf32msp430.sc: Likewise.
* scripttempl/elf32msp430_3.sc: Likewise.
* scripttempl/elf32sh-symbian.sc: Likewise.
* scripttempl/elf64hppa.sc: Likewise.
* scripttempl/elf_chaos.sc: Likewise.
* scripttempl/elfd10v.sc: Likewise.
* scripttempl/elfd30v.sc: Likewise.
* scripttempl/elfi370.sc: Likewise.
* scripttempl/elfm68hc11.sc: Likewise.
* scripttempl/elfm68hc12.sc: Likewise.
* scripttempl/elfxgate.sc: Likewise.
* scripttempl/elfxtensa.sc: Likewise.
* scripttempl/epiphany_4x4.sc: Likewise.
* scripttempl/i386beos.sc: Likewise.
* scripttempl/i386go32.sc: Likewise.
* scripttempl/ia64vms.sc: Likewise.
* scripttempl/ip2k.sc: Likewise.
* scripttempl/iq2000.sc: Likewise.
* scripttempl/mep.sc: Likewise.
* scripttempl/mmo.sc: Likewise.
* scripttempl/v850.sc: Likewise.
* scripttempl/v850_rh850.sc: Likewise.
* scripttempl/xstormy16.sc: Likewise.

gdb/

* solib-dsbt.c (fetch_loadmap): Re-indent.
(displacement_from_map, enable_break2): Likewise.
(dsbt_relocate_section_addresses): Likewise.

daily update

*** empty log message ***

* target-reloc.h (relocate_section): If the reloc offset is out of
range, pass VIEW as NULL to relocate.relocate.
* arm.cc (Target_arm:Relocate::relocate): Check for a NULL view.
* i386.cc (Target_i386::Relocate::relocate): Likewise.
* powerpc.cc (Target_powerpc::Relocate::relocate): Likewise.
* sparc.cc (Target_sparc::Relocate::relocate): Likewise.
* tilegx.cc (Target_tilegx::Relocate::relocate): Likewise.
* x86_64.cc (Target_x86_64::Relocate::relocate): Likewise.

*** empty log message ***

daily update

2013-04-26  Geoff Pike  <gpike@chromium.org>

gold/
* gold.cc (queue_final_tasks): invoke layout->queue_build_id_tasks().
* layout.cc (Hash_task): New class.
(Layout::queue_build_id_tasks): New function.
(Layout::write_build_id): Handle single-thread portion of build ID
computation.  (In some cases, all of it is single-threaded.)  Replace
{sha1,md5}_process_bytes with {sha1,md5}_buffer to get the same
functionality in fewer lines of code.
* layout.h (Layout::queue_build_id_tasks): New function declaration.
* options.h (General_options): make "--build-id" default to tree
rather than sha1.  Add two new options related to --build-id=tree:
--build-id-chunk-size-for-treehash and
--build-id-min-file-size-for-treehash.
* Makefile.am: add testing of --build-id=tree and related new options
(these tests will be invoked by "make check").
* Makefile.in: Regenerate.

* MAINTAINERS: Add myself and Cary as gold maintainers.

fix changelog entry: we still have 2013, not 2014

* config/tc-z8k.c (md_parse_option): Set z8k_target_from_cmdline
according to the target parameter.  Don't call s_segm since s_segm
calls bfd_set_arch_mach using stdoutput, but stdoutput isn't
initialized yet.
(md_begin): Call s_segm according to target parameter from command
line.

* scripttempl/avr.sc: Add ALIGN directive after *(.progmem*).

GDB 7.6 released.

Add missing test files from 2013-03-21 commit.

Remove some dead code from elf64_aarch64_check_relocs.

bfd/ChangeLog:

2013-04-25  Will Newton  <will.newton@linaro.org>

* elf64-aarch64.c (elf64_aarch64_check_relocs): Remove dead code.

* emultempl/elf32.em (gld${EMULATION_NAME}_after_open): Test
unresolved_syms_in_shared_libs rather than !executable to
determine whether to load DT_NEEDED libraries.

daily update

*** empty log message ***

PR corefiles/14983:
* dwarf2read.c (process_full_comp_unit): Always create a static
block.
gdb/testsuite
* gdb.dwarf2/nostaticblock.exp: New file.

bfd/
* config.bfd: Add powerpc64le-linux.
gas/
* configure.in: Allow little-endian linux.
* configure: Regenerate.
gold/
* configure.tgt: Add powerpcle and powerpc64le.
ld/
* Makefile.am (ALL_EMULATION_SOURCES): Add eelf32lppclinux.c.
(eelf32lppclinux.c): New rule.
* Makefile.in: Regenerate.
* configure.tgt: Merge powerpc-linux and other powerpc-elf targets
with corresponding little-endian targets.
* emulparams/elf32lppc.sh: Update comment.
* emulparams/elf32lppclinux.sh: New.

Fix format error of ChangeLog.

2013-04-25  Hui Zhu  <hui@codesourcery.com>

PR gdb/15186

* ax.c (ax_printf): Add fflush.

*** empty log message ***

daily update

2013-04-25  Hui Zhu  <hui@codesourcery.com>

* breakpoint.c (build_target_command_list): Change loc->cond_bytecode
to loc->cmd_bytecode.

2013-04-24  Sandra Loosemore  <sandra@codesourcery.com>

gas/
* config/tc-nios2.c (nios2_control_register_arg_p): Rename
"fstatus" control register to "eccinj".

opcodes/
* nios2-opc.c (nios2_builtin_reg): Rename "fstatus" control register
to "eccinj".

* dwarf2read.c (setup_type_unit_groups): Fix comment.

Enable x32 for x86_64-*-elf*

This patch enables x32 for x86_64-*-elf* for embedded target and disables
rex tests since it uses '/' as prefix separator which is `\' for
x86_64-*-elf*.

bfd/

* config.bfd (targ_selvecs): Add bfd_elf32_x86_64_vec for
x86_64-*-elf*.

gas/testsuite/

* gas/i386/rex.d: Skip x86_64-*-elf*.
* gas/i386/ilp32/rex.d: Likewise.

ld/

* configure.tgt (targ_extra_emuls): Adds elf32_x86_64 for
x86_64-*-elf*.
(targ_extra_libpath): Likewise.
(tdir_elf_i386): Replace x86_64 with i386 for x86_64-*-elf*.

* tracepoint.c (trace_save):  Call the writer's start method.

ld/

* emulparams/aarch64elf.sh: Define ELFSIZE=64.
* emulparams/aarch64linux.sh: Ditto.
* emultempl/aarch64elf.em (aarch64_elf_before_allocation):
Replace elf64 with elf${ELFSIZE}.
(elf64_aarch64_add_stub_section): Likewise.
(build_section_lists): Likewise.
(gld${EMULATION_NAME}_after_allocation): Likewise.
(gld${EMULATION_NAME}_finish): Likewise.
(aarch64_elf_create_output_section_statements): Likewise.

bfd/
* elf32-arm.c (elf32_arm_allocate_plt_entry): If HTAB->nacl_p,
allocate space for PLT header even if IS_IPLT_ENTRY.
(arm_nacl_put_plt0): New function, broken out of ...
(elf32_arm_finish_dynamic_sections): ... here.  Call it.
If HTAB->nacl_p, set up the PLT header in .iplt too.
(elf32_arm_output_arch_local_syms): If HTAB->nacl_p, write
a mapping symbol for the start of .iplt too.

PR ld/15389
* scripttempl/avr.sc: Add .note.gnu.build-id section.

2013-04-24  Muhammad Bilal  <mbilal@codesourcery.com>

PR gdb/10462
* cli/cli-decode.c (lookup_command): Show an error if there is no space
before argument.

2013-04-24  Muhammad Bilal  <mbilal@codesourcery.com>

PR gdb/10462
* gdb.base/setshow.exp: Add test case.

oops - revert accidental commit

*** empty log message ***

* ecoff.c: casting the return value of BFD_ALIGN to int
* elf32-tic6x.c: removing unused variables
* elf32-v850.c: redefine the type of a variable (int -> bfd_vma)
* vms-alpha.c: initialize a variable

*** empty log message ***

daily update

* common/filestuff.c: Check USE_WIN32API before including
sys/socket.h.
(HAVE_F_GETFD): New define.
(mark_cloexec): Check HAVE_F_GETFD.
(gdb_open_cloexec): Change 'mode' to unsigned long.
(gdb_socketpair_cloexec): Check HAVE_SOCKETPAIR.
(gdb_pipe_cloexec): Check HAVE_PIPE.
* common/filestuff.h (gdb_open_cloexec): Change 'mode' to unsigned
long.

merge from gcc

2013-04-23  Hui Zhu  <hui@codesourcery.com>

PR gdb/15293

* breakpoint.c (bpstat_what): Add BPSTAT_WHAT_SINGLE to bp_dprintf.

2013-04-23  Hui Zhu  <hui@codesourcery.com>

PR gdb/15293

* gdb.base/dprintf.exp: Add ignore command.

2013-04-23  Hui Zhu  <hui@codesourcery.com>

PR gdb/15165

* breakpoint.c (dprintf_print_recreate): New.
(save_breakpoints): Let it not save dprintf commands.
(initialize_breakpoint_ops): Set dprintf_print_recreate.

2013-04-23  Hui Zhu  <hui@codesourcery.com>

PR gdb/15165

* gdb.base/save-bp.exp: Add test for dprintf.

daily update

*** empty log message ***

* gdb.gdb/selftest.exp (do_steps_and_nexts): Check for
notice_open_fds.

2013-04-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>

* Makefile.def: Sync with GCC.
* Makefile.in: Regenerate.

PR gdb/7912:
* Makefile.in (SFILES): Add filestuff.c
(COMMON_OBS): Add filestuff.o.
(filestuff.o): New target.
* auto-load.c (auto_load_objfile_script_1): Use
gdb_fopen_cloexec.
* auxv.c (procfs_xfer_auxv): Use gdb_open_cloexec.
* cli/cli-cmds.c (shell_escape): Call close_most_fds.
* cli/cli-dump.c (fopen_with_cleanup): Use gdb_fopen_cloexec.
* common/agent.c (gdb_connect_sync_socket): Use
gdb_socket_cloexec.
* common/filestuff.c: New file.
* common/filestuff.h: New file.
* common/linux-osdata.c (linux_common_core_of_thread)
(command_from_pid, commandline_from_pid, print_source_lines)
(linux_xfer_osdata_shm, linux_xfer_osdata_sem)
(linux_xfer_osdata_msg, linux_xfer_osdata_modules): Use
gdb_fopen_cloexec.
* common/linux-procfs.c (linux_proc_get_int)
(linux_proc_pid_has_state): Use gdb_fopen_cloexec.
* config.in, configure: Rebuild.
* configure.ac: Don't check for sys/socket.h.  Check for
fdwalk, pipe2.
* corelow.c (core_open): Use gdb_open_cloexec.
* dwarf2read.c (write_psymtabs_to_index): Use gdb_fopen_cloexec.
* fork-child.c (fork_inferior): Call close_most_fds.
* gdb_bfd.c (gdb_bfd_open): Use gdb_open_cloexec.
* inf-child.c (inf_child_fileio_readlink): Use gdb_open_cloexec.
* linux-nat.c (linux_nat_thread_name, linux_proc_pending_signals):
Use gdb_fopen_cloexec.
(linux_proc_xfer_partial, linux_proc_xfer_spu): Use
gdb_open_cloexec.
(linux_async_pipe): Use gdb_pipe_cloexec.
* remote-fileio.c (remote_fileio_func_open): Use
gdb_open_cloexec.
* remote.c (remote_file_put, remote_file_get): Use
gdb_fopen_cloexec.
* ser-pipe.c (pipe_open): Use gdb_socketpair_cloexec,
close_most_fds.
* ser-tcp.c (net_open): Use gdb_socket_cloexec.
* ser-unix.c (hardwire_open): Use gdb_open_cloexec.
* solib.c (solib_find): Use gdb_open_cloexec.
* source.c (openp, find_and_open_source): Use gdb_open_cloexec.
* tracepoint.c (tfile_start): Use gdb_fopen_cloexec.
(tfile_open): Use gdb_open_cloexec.
* tui/tui-io.c (tui_initialize_io): Use gdb_pipe_cloexec.
* ui-file.c (gdb_fopen): Use gdb_fopen_cloexec.
* xml-support.c (xml_fetch_content_from_file): Use
gdb_fopen_cloexec.
* main.c (captured_main): Call notice_open_fds.
gdbserver
* Makefile.in (SFILES): Add filestuff.c.
(OBS): Add filestuff.o.
(filestuff.o): New target.
* config.in, configure: Rebuild.
* configure.ac: Check for fdwalk, pipe2.

2013-04-22  Edjunior Machado  <emachado@linux.vnet.ibm.com>

* remote-sim.c (dump_mem): Change the type of 'buf' parameter from
'char *' to 'gdb_byte *'.
(gdbsim_store_register): Change the type of 'tmp' from 'char' to
'gdb_byte'.

2013-04-22  Jan-Benedict Glaw  <jbglaw@lug-owl.de>

* configure.ac: Sync with GCC.
* configure: Regenerated.

PR ld/15382
* elf-bfd.h (RELOC_AGAINST_DISCARDED_SECTION): Don't multiply
sh_size or reloc_count adjustment by count.

gdb/

* infrun.c: Fix typo in comment.

2013-04-22  Sergio Durigan Junior  <sergiodj@redhat.com>

* gdb.arch/arm-bl-branch-dest.exp: Replace additional_flags by
ldflags.

Andrew Haley found a bug on GDB running on ARM when using
--enable-64-bit-bfd.  Basically the issue happens when dealing with "bl"
instructions: GDB does branch destination calculation and (wrongly)
sign-extends the PC.  Here is a piece of his original message explaining
the problem:

>      next_pc = arm_get_next_pc (frame, get_frame_pc (frame));
>
>      /* The Linux kernel offers some user-mode helpers in a high page.  We can
>         not read this page (as of 2.6.23), and even if we could then we couldn't
>         set breakpoints in it, and even if we could then the atomic operations
>         would fail when interrupted.  They are all called as functions and return
>         to the address in LR, so step to there instead.  */
>      if (next_pc > 0xffff0000)
>        next_pc = get_frame_register_unsigned (frame, ARM_LR_REGNUM);
>
>      arm_insert_single_step_breakpoint (gdbarch, aspace, next_pc);
>
>    Unfortunately, branch destination addresses are SIGN EXTENDED to 64
>    bits.  So,
>
>    (top-gdb) p/x next_pc
>    $14 = 0xffffffffb6df2864
>
>    Which triggers the next_pc = get_frame_register_unsigned(), and we
>    cannot step into any branches because the destination PC is wrong.

Anyway, the fix is simple and Andrew himself provided it for us.  It
took a while for me to figure out how to trigger the bug (in order to
write a testcase for it), but I finally made it.

The attached patch fixes the problem (by casting to `unsigned long'
instead of just `long'), and also includes a testcase to reproduce the
issue.

gdb/ChangeLog:

2013-04-22  Andrew Haley  <aph@redhat.com>

* arm-tdep.c (BranchDest): Cast result as "unsigned long",
instead of "long".

gdb/testsuite/ChangeLog:

2013-04-22  Sergio Durigan Junior  <sergiodj@redhat.com>

* gdb.arch/arm-bl-branch-dest.c: New file.
* gdb.arch/arm-bl-branch-dest.exp: Likewise.

bfd/
* elf64-ppc.c (ppc64_elf_check_relocs): Don't call
create_linkage_sections here..
(ppc64_elf_init_stub_bfd): ..do so here.  Return status.
(create_linkage_sections): Move earlier in file.
(ppc64_elf_setup_section_lists): Remove now useless htab->brlt test.
* elf64-ppc.h (ppc64_elf_init_stub_bfd): Update proto.
ld/
* emultempl/ppc64elf.em (ppc_create_output_section_statements):
Check return from ppc64_elf_init_stub_bfd.
ld/testsuite/
* ld-powerpc/tlsexe.d: Adjust.
* ld-powerpc/tlsexe.r: Adjust.
* ld-powerpc/tlsexetoc.d: Adjust.
* ld-powerpc/tlsexetoc.r: Adjust.
* ld-powerpc/tlsso.d: Adjust.
* ld-powerpc/tlsso.r: Adjust.
* ld-powerpc/tlstocso.d: Adjust.
* ld-powerpc/tlstocso.r: Adjust.

*** empty log message ***

daily update

PR gold/15355
* layout.cc (Layout::segment_precedes): Allow more than one
segment with the same type and flags.

*** empty log message ***

daily update

gdb/

* ctf.c (ctf_fetch_registers): Change the type of 'regs' from
'char *' to 'gdb_byte *'.  Cast the return value of
'bt_ctf_get_char_array' to 'gdb_byte *'.

*** empty log message ***

daily update

s/signness/signedness in gdbint.texinfo.

"signedness" is more typical.

gdb/doc/
2013-04-19  Pedro Alves  <palves@redhat.com>

* gdbint.texinfo (Misc Guidelines) <Compiler Warnings>: Write
"signedness" instead of "signness".

Enable -Wpointer-sign by default.

This enables -Wpointer-sign by default.

I've checked that --enable-targets=all builds fine with the following
as --host, on x86_64 Fedora 17 --build:

x86_64 GNU/Linux
i386 GNU/Linux
i386 MinGW-w64
i386 msdos/djgpp

OK?

gdb/
2013-04-19  Pedro Alves  <palves@redhat.com>

* configure.ac (build_warnings): Replace -Wno-pointer-sign with
-Wpointer-sign.
* configure: Regenerate.

gdb/doc
2013-04-19  Pedro Alves  <palves@redhat.com>

* gdbint.texinfo (Misc Guidelines) <Compiler Warnings>: Replace
-Wno-pointer-sign text with text on -Wpointer-sign.

2013-04-19  Nathan Froyd  <froydnj@codesourcery.com>

* ppc-instructions (isel): New instruction.

Make GDB -Wpointer-sign clean on MinGW too.

This is the remaining issue impeding GDB to build with "-Wpointer-sign
-Werror" with Fedora 17's i686-w64-mingw32 cross toolchain.

../../src/gdb/ser-tcp.c: In function 'net_read_prim':
../../src/gdb/ser-tcp.c:341:3: error: pointer targets in passing argument 2 of 'recv' differ in signedness [-Werror=pointer-sign]
In file included from ../../src/gdb/serial.h:23:0,
                 from ../../src/gdb/ser-tcp.c:21:
/usr/i686-w64-mingw32/sys-root/mingw/include/winsock2.h:983:34: note: expected 'char *' but argument is of type 'unsigned char *'

gdb/
2013-04-19  Pedro Alves  <palves@redhat.com>

* ser-tcp.c (net_read_prim): Cast second argument to recv to
'void *'.

-Wpointer-sign: monitor.c.

This fixes -Wpointer-sign issues in monitor.c.

Tested by building on x86_64 Fedora 17 w/ --enable-targets=all.

gdb/
2013-04-19  Pedro Alves  <palves@redhat.com>

* monitor.c (monitor_write_memory, monitor_write_memory_bytes):
Change type of 'myaddr' parameter to gdb_byte pointer.
(monitor_write_memory_longlongs): Likewise.  Cast 'myaddr' pointer
to 'long long' pointer instead of to 'unsigned long long'.
(monitor_write_memory_block, monitor_read_memory_single)
(monitor_read_memory): Change type of 'myaddr' parameter to
gdb_byte pointer.

-Wpointer-sign: record.c.

../../src/gdb/record.c: In function ‘set_record_insn_history_size’:
../../src/gdb/record.c:670:5: error: pointer targets in passing argument 2 of ‘validate_history_size’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/record.c:646:1: note: expected ‘int *’ but argument is of type ‘unsigned int *’
../../src/gdb/record.c: In function ‘set_record_call_history_size’:
../../src/gdb/record.c:682:5: error: pointer targets in passing argument 2 of ‘validate_history_size’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/record.c:646:1: note: expected ‘int *’ but argument is of type ‘unsigned int *’

This fixes it in the obvious way.

gdb/
2013-04-19  Pedro Alves  <palves@redhat.com>

* record.c (validate_history_size): Make parameter 'setting'
unsigned.

-Wpointer-sign: ctf.c.

ctf_save_write's second parameter is gdb_byte *, and all these
arguments are 'char *'.  Since this function is ultimately just
writing host bytes to a local file with fwrite, an alternative would
be to change ctf_save_write to take a 'void *' instead of 'gdb_byte
*', thus removing the need for any cast (we have more calls with casts
than without).

gdb/
2013-04-19  Pedro Alves  <palves@redhat.com>

* ctf.c (ctf_write_uploaded_tsv, ctf_write_uploaded_tp): Add casts
to 'gdb_byte *'.