serial_write: change prototype to take a void-pointer buffer.

While remote.c works with "char *" buffers most of the time, other
remote targets have binary-ish-er protocols, and choose to use
"unsigned char" throughout, like e.g., remote-mips.c or
remote-m32r-sdi.c.  That results in -Wpointer-sign warnings in those
targets, unless we add casts in calls to serial_write.  Since
serial_write is only concerned about sending raw host bytes out, and
serial_ops->write_prim already works with "void *"/"size_t", a similar
interface to the "write" or "send" system calls, I find it natural to
change serial_write's prototype accordingly, avoiding the need for
casts.

Tested on x86_64 Fedora 17, and also by building x86_64-mingw32
and DJGPP/go32 -hosted gdbs.

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

* ser-base.c (ser_base_write): Change prototype -- take 'void *'
buffer and size_t size.  Adjust.
* ser-base.h (ser_base_write): Adjust.
* ser-go32.c (cnts): Change type to size_t.
(dos_write): Change prototype -- take 'void *'
buffer and size_t size.  Adjust.
(dos_info): Print elements of 'cnts' as unsigned long.
* serial.c (serial_write): Likewise.
* serial.h (serial_write): Adjust.
(struct serial_ops) <write>: Change prototype -- take 'void *'
buffer and size_t size.  Adjust.

Cast result of obstack_base to gdb_byte * in a couple spots.

obstack_base returns char *.  Need to cast to gdb_byte * in a couple spots.

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

* c-lang.c (evaluate_subexp_c): Cast result of obstack_base to
gdb_byte *.
* linux-tdep.c (linux_make_mappings_corefile_notes): Likewise.

-Wpointer-sign: alpha-tdep.c.

../../src/gdb/alpha-tdep.c: In function ‘alpha_extract_return_value’:
../../src/gdb/alpha-tdep.c:520:4: error: pointer targets in passing argument 3 of ‘regcache_cooked_read_signed’ differ in signedness [-Werror=pointer-sign]
In file included from ../../src/gdb/alpha-tdep.c:36:0:

We use regcache_cooked_read_unsigned everywhere else too.

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

* alpha-tdep.c (alpha_extract_return_value): Use
regcache_cooked_read_unsigned to read 'v0'.

-Wpointer-sign: xtensa-tdep.c.

../../src/gdb/xtensa-tdep.c:2914:4: error: pointer targets in passing argument 7 of ‘xtensa_operand_get_field’ differ in signedness [-Werror=pointer-sign]
In file included from ../../src/gdb/xtensa-tdep.c:53:0:
../../src/gdb/../include/xtensa-isa.h:487:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/xtensa-tdep.c:2916:4: error: pointer targets in passing argument 4 of ‘xtensa_operand_decode’ differ in signedness [-Werror=pointer-sign]
In file included from ../../src/gdb/xtensa-tdep.c:53:0:
../../src/gdb/../include/xtensa-isa.h:507:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/xtensa-tdep.c:2918:4: error: pointer targets in passing argument 7 of ‘xtensa_operand_get_field’ differ in signedness [-Werror=pointer-sign]
In file included from ../../src/gdb/xtensa-tdep.c:53:0:
../../src/gdb/../include/xtensa-isa.h:487:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/xtensa-tdep.c:2920:4: error: pointer targets in passing argument 4 of ‘xtensa_operand_decode’ differ in signedness [-Werror=pointer-sign]
In file included from ../../src/gdb/xtensa-tdep.c:53:0:
../../src/gdb/../include/xtensa-isa.h:507:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/xtensa-tdep.c:2922:4: error: pointer targets in passing argument 7 of ‘xtensa_operand_get_field’ differ in signedness [-Werror=pointer-sign]
In file included from ../../src/gdb/xtensa-tdep.c:53:0:
../../src/gdb/../include/xtensa-isa.h:487:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/xtensa-tdep.c:2924:4: error: pointer targets in passing argument 4 of ‘xtensa_operand_decode’ differ in signedness [-Werror=pointer-sign]
In file included from ../../src/gdb/xtensa-tdep.c:53:0:
../../src/gdb/../include/xtensa-isa.h:507:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’

Those bfd functions that decode instructions output uint32_t values.
Hence this fix:

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

* xtensa-tdep.c (execute_l32e, execute_s32e): Change type of
parameters 'at', 'as' and 'offset' to uint32_t.

-Wpointer-sign: aarch64-tdep.c.

../../src/gdb/aarch64-tdep.c: In function ‘aarch64_analyze_prologue’:
../../src/gdb/aarch64-tdep.c:713:7: error: pointer targets in passing argument 3 of ‘decode_cb’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/aarch64-tdep.c:386:1: note: expected ‘int *’ but argument is of type ‘unsigned int *’
../../src/gdb/aarch64-tdep.c:747:7: error: pointer targets in passing argument 3 of ‘decode_stur’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/aarch64-tdep.c:597:1: note: expected ‘int *’ but argument is of type ‘unsigned int *’

'is64' is just used as a boolean; signed/unsigned doesn't really matter.

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

* aarch64-tdep.c (aarch64_analyze_prologue): Change type of local
'is64' to signed 'int'.

-Wpointer-sign: s390-tdep.c.

-Wpointer-sign reveals a bunch of:

../../src/gdb/s390-tdep.c:1342:7: error: pointer targets in passing argument 4 of ‘is_rx’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/s390-tdep.c:1038:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/s390-tdep.c:1343:9: error: pointer targets in passing argument 5 of ‘is_rxy’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/s390-tdep.c:1055:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/s390-tdep.c:1344:9: error: pointer targets in passing argument 5 of ‘is_rxy’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/s390-tdep.c:1055:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
...
../../src/gdb/s390-tdep.c:1363:7: error: pointer targets in passing argument 5 of ‘is_rs’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/s390-tdep.c:966:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/s390-tdep.c:1364:9: error: pointer targets in passing argument 6 of ‘is_rsy’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/s390-tdep.c:983:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
../../src/gdb/s390-tdep.c:1365:9: error: pointer targets in passing argument 6 of ‘is_rsy’ differ in signedness [-Werror=pointer-sign]
../../src/gdb/s390-tdep.c:983:1: note: expected ‘unsigned int *’ but argument is of type ‘int *’
...

I don't know much about s390, but from reading the code I believe the
right fix is to treat d2 as signed.

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

* s390-tdep.c (is_rs, is_rsy, is_rx, is_rxy): Change type of 'd2'
parameter to int *.

ppc-linux-tdep.c: Wrong signness for buffer holding instructions.

There seems to be no reason for this buffer to be signed.  We pass it
around to functions expecting it to be unsigned (which triggers
-Wpointer-sign warnings).

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

* ppc-linux-tdep.c (ppc_skip_trampoline_code): Change local
'insnbuf' buffer type to unsigned int[].

mips-tdep.c: Wrong signness for local holding PC register.

Addresses on MIPS are signed, and we're indeed using
regcache_cooked_read_signed to read the PC, but, we're passing it the
address of an unsigned variable, which triggers a -Wpointer-sign
warning.  I've chosen to change the variable's type.  I believe this
will end up being the same (though I can't test it).

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

* mips-tdep.c (mips_read_pc): Change local 'pc' type to LONGEST.

mep-tdep.c: Wrong signness for instruction buffer.

There seems to be no reason for this to be signed.

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

* mep-tdep.c (mep_get_insn): Change 'insn' parameter type to
unsigned long *.

alpha-tdep.c/mips-tdep.c: "set heuristic-fence-post" is signed/zinteger.

These commands are currently var_zinteger, hence their control
variable is signed.

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

* alpha-tdep.c (heuristic_fence_post): Change type to int.
(alpha_heuristic_proc_start): Adjust to check -1 instead of
UINT_MAX.
* mips-tdep.c (heuristic_fence_post): Change type to int.
(heuristic_proc_start): Adjust to check -1 instead of UINT_MAX.

cris-tdep.c: Use unsigned variable for unsigned command.

"set cris-version" is an unsigned command:

    /* CRIS-specific user-commands.  */
    add_setshow_uinteger_cmd ("cris-version", class_support,
       &usr_cmd_cris_version,
       _("Set the current CRIS version."),
       _("Show the current CRIS version."),
       _("\
  Set to 10 for CRISv10 or 32 for CRISv32 if autodetection fails.\n\
  Defaults to 10. "),

Tested by building on x86_64 Fedora 17.

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

* cris-tdep.c (usr_cmd_cris_version): Make unsigned.
(struct gdbarch_tdep) <cris_version>: Make unsigned.
(cris_spec_reg_applicable, cris_gdbarch_init): Adjust locals.

-Wpointer-sign: gdb_byte -> char.

This is sort of the opposite of the previous patch.  Places that
manipulate strings or interfaces that return strings are changed to
use char* instead of gdb_byte*.

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

* avr-tdep.c (avr_io_reg_read_command): New local 'bufstr'.  Use
it to get a string view of the byte buffer.
* i386-cygwin-tdep.c (core_process_module_section): Change local 'buf'
type to gdb_byte *.  Adjust.
* linux-tdep.c (linux_info_proc, linux_find_memory_regions_full):
Change local to char *.
* solib-darwin.c (find_program_interpreter): Change return type to
char *.  Adjust.
(darwin_solib_get_all_image_info_addr_at_init): Adjust.
* solib-dsbt.c (enable_break2): Change local 'buf' to char *.
* solib-frv.c (enable_break2): Change local 'buf' to char *.
* solib-spu.c (spu_current_sos): Add gdb_byte * cast.
* solib-svr4.c (find_program_interpreter): Change return type to
char *.  Adjust.
(enable_break): Change local 'interp_name' to char *.
* spu-multiarch.c (spu_xfer_partial): Add cast to 'char *'.
* spu-tdep.c (spu_pseudo_register_read_spu): Add cast to 'char *'.
(spu_pseudo_register_write_spu): Use char for string buffer.
Adjust.
(info_spu_event_command, info_spu_signal_command): Add casts to
'char *'.

-Wpointer-sign: char -> gdb_byte.

-Wpointer-sign catches all these cases across the codebase that should
be using gdb_byte for raw target bytes.  I think these are all
obvious, hence I've collapsed into a single patch.

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

* aarch64-tdep.c (aarch64_default_breakpoint): Change type to
gdb_byte[].
(aarch64_breakpoint_from_pc): Change return type to gdb_byte *.
* ada-lang.c (ada_value_assign): Use gdb_byte.
* alphanbsd-tdep.c (sigtramp_retcode): Change type to gdb_byte[].
(alphanbsd_sigtramp_offset): Use gdb_byte.
* arm-linux-tdep.c (arm_linux_arm_le_breakpoint)
(arm_linux_arm_be_breakpoint, eabi_linux_arm_le_breakpoint)
(eabi_linux_arm_be_breakpoint, arm_linux_thumb_be_breakpoint)
(arm_linux_thumb_le_breakpoint, arm_linux_thumb2_be_breakpoint)
(arm_linux_thumb2_le_breakpoint): Change type to gdb_byte[].
* arm-tdep.c (arm_stub_unwind_sniffer)
(arm_displaced_init_closure): Use gdb_byte.
(arm_default_arm_le_breakpoint, arm_default_arm_be_breakpoint)
(arm_default_thumb_le_breakpoint)
(arm_default_thumb_be_breakpoint): Change type to gdb_byte[].
* arm-tdep.h (struct gdbarch_tdep) <arm_breakpoint,
thumb_breakpoint, thumb2_breakpoint>: Change type to gdb_byte *.
* arm-wince-tdep.c (arm_wince_le_breakpoint)
(arm_wince_thumb_le_breakpoint): Change type to gdb_byte[].
* armnbsd-tdep.c (arm_nbsd_arm_le_breakpoint)
(arm_nbsd_arm_be_breakpoint, arm_nbsd_thumb_le_breakpoint)
(arm_nbsd_thumb_be_breakpoint): Change type to gdb_byte[].
* armobsd-tdep.c (arm_obsd_thumb_le_breakpoint)
(arm_obsd_thumb_be_breakpoint): Change type to gdb_byte[].
* cris-tdep.c (push_stack_item, cris_push_dummy_call)
(cris_store_return_value, cris_extract_return_value): Use
gdb_byte.
(constraint): Change type of parameter to char * from signed
char*.  Use gdb_byte.
* dwarf2loc.c (read_pieced_value, write_pieced_value): Change type
of local buffer to gdb_byte *.
* dwarf2read.c (read_index_from_section): Use gdb_byte.
(create_dwp_hash_table): Change type of locals to gdb_byte *.
(add_address_entry): Change type of local buffer to gdb_byte[].
* frv-tdep.c (frv_adjust_breakpoint_address, find_func_descr)
(frv_push_dummy_call): Use gdb_byte.
* hppa-hpux-tdep.c (hppa_hpux_push_dummy_code)
(hppa_hpux_supply_ss_fpblock, hppa_hpux_supply_ss_wide)
(hppa_hpux_supply_save_state): Use gdb_byte.
* hppa-tdep.c (hppa32_push_dummy_call)
(hppa64_convert_code_addr_to_fptr): Use gdb_byte.
* ia64-tdep.c (extract_bit_field, replace_bit_field)
(slotN_contents, replace_slotN_contents): Change type of parameter
to gdb_byte *.
(fetch_instruction, ia64_pseudo_register_write)
(ia64_register_to_value, ia64_value_to_register)
(ia64_extract_return_value, ia64_store_return_value)
(ia64_push_dummy_call): Use gdb_byte.
* m32c-tdep.c (m32c_return_value): Remove cast.
* m68hc11-tdep.c (m68hc11_pseudo_register_write)
(m68hc11_push_dummy_call, m68hc11_store_return_value): Use
gdb_byte.
* mipsnbsd-tdep.c (mipsnbsd_get_longjmp_target): Use gdb_byte.
* mn10300-tdep.c (mn10300_store_return_value)
(mn10300_breakpoint_from_pc, mn10300_push_dummy_call): Use
gdb_byte.
* moxie-tdep.c (moxie_process_readu): Use gdb_byte.
(moxie_process_record): Remove casts.
* ppc-ravenscar-thread.c (supply_register_at_address)
(ppc_ravenscar_generic_store_registers): Use gdb_byte.
* ravenscar-thread.c (get_running_thread_id): Use gdb_byte.
* remote-m32r-sdi.c (m32r_fetch_register): Use gdb_byte.
* remote-mips.c (mips_xfer_memory): Use gdb_byte.
* remote.c (compare_sections_command): Use gdb_byte.
* score-tdep.c (score7_free_memblock): Change type of parameter to
gdb_byte *.
* sh-tdep.c (sh_justify_value_in_reg): Change return type to
gdb_byte *.  Use gdb_byte.
(sh_push_dummy_call_fpu): Use gdb_byte.
(sh_extract_return_value_nofpu, sh_extract_return_value_fpu)
(sh_store_return_value_nofpu, sh_store_return_value_fpu)
(sh_register_convert_to_virtual, sh_register_convert_to_raw):
Change parameter type to 'gdb_byte *'.  Use gdb_byte.
(sh_pseudo_register_read, sh_pseudo_register_write): Use gdb_byte.
* sh64-tdep.c (sh64_push_dummy_call): Use gdb_byte.
(sh64_store_return_value, sh64_register_convert_to_virtual):
Change parameter type to 'gdb_byte *'.  Use gdb_byte.
(sh64_pseudo_register_write): Use gdb_byte.
* solib-darwin.c (darwin_current_sos): Add casts to 'gdb_byte *'.
* solib-irix.c (fetch_lm_info): Likewise.  Use gdb_byte for byte
buffer.
(irix_current_sos): Use gdb_byte.
* solib-som.c (som_current_sos): Use gdb_byte.
* sparc-ravenscar-thread.c (supply_register_at_address)
(sparc_ravenscar_generic_store_registers): Use gdb_byte.
* spu-multiarch.c (spu_xfer_partial): Add cast to 'char *'.
* spu-tdep.c (spu_get_overlay_table): Use gdb_byte.
* tic6x-tdep.c (tic6x_breakpoint_from_pc): Change return type to
'gdb_byte *'.
* tic6x-tdep.h (struct gdbarch_tdep) <breakpoint>: Change type to
'gdb_byte *'.
* tracepoint.c (tfile_fetch_registers): Use gdb_byte.
* xstormy16-tdep.c (xstormy16_extract_return_value)
(xstormy16_store_return_value): Change parameter type to
'gdb_byte *'.  Adjust.
(xstormy16_push_dummy_call): Use gdb_byte.
* xtensa-tdep.c (xtensa_scan_prologue, call0_ret)
(call0_analyze_prologue, execute_code): Use gdb_byte.

Fix the x87 FP register printout when issuing the “info float” command.

Consider the following simple program:

.globl  _start
.text
_start:
      fldt    val
.data
      val: .byte 0x00,0x00,0x45,0x07,0x11,0x19,0x22,0xe9,0xfe,0xbf

With current GDB on x86-64 GNU/Linux hosts, after the moment the fldt
command has been executed the register st(0) looks like this,
according to the “info regs” output (TOP=7):

  R7: Valid   0xffffffbffffffffeffffffe922191107450000 -0.910676542908976927

which is clearly wrong (just count its length).  The problem is due to
the printf statement (see patch) printing a promoted integer value of
a char argument "raw[i]", and, since char is signed on x86-64
GNU/Linux, the erroneous “ffffff” are printed for the first three
bytes which turn out to be "negative".  The fix is to use gdb_byte
instead which is unsigned (and is the type of value_contents(), the
type to be used for raw target bytes anyway).  After the fix the value
will be printed correctly:

  R7: Valid   0xbffee922191107450000 -0.910676542908976927

gdb/
2013-04-19  Vladimir Kargov <kargov@gmail.com>
    Pedro Alves  <palves@redhat.com>

* i387-tdep.c (i387_print_float_info): Use gdb_byte for pointer to
value contents.

gdb/testsuite/
2013-04-19  Vladimir Kargov  <kargov@gmail.com>
    Pedro Alves  <palves@redhat.com>

* gdb.arch/i386-float.S: New file.
* gdb.arch/i386-float.exp: New file.

    * configure.tgt (i386-*-cygwin): Handle x86_64 cygwin.

PR binutils/15356
* compress.c (decompress_contents): Always call inflateEnd, even
when another inflation operation fails.

[Note - the actual patch to compress.c was accidentally committed previously].

daily update

*** empty log message ***

* gdb.mi/mi-var-create-rtti.exp: Create a variable of
type void *.

*** empty log message ***

daily update

* lib/dwarf.exp (Dwarf): New proc "tu".
* gdb.dwarf2/missing-sig-type.exp: New file.

* dwarf2read.c (struct signatured_type): New member type.
(struct attribute): Replace member signatured_type with signature.
(DW_SIGNATURE): Replaces DW_SIGNATURE_TYPE.
(read_call_site_scope): Call follow_die_ref instead of
follow_die_ref_or_sig.
(read_structure_type): Rewrite handling of signatured types.
(read_enumeration_type): Ditto.
(read_attribute_value): Update.
(build_error_marker_type): New function.
(lookup_die_type): Add assert.  Rewrite handling of signatured types.
Don't call error for bad types, just build an error marker type.
(dump_die_shallow): Update.
(follow_die_sig_1): Renamed from follow_die_sig.
Don't call error for bad types, instead return NULL.
(follow_die_sig): New function.
(get_signatured_type, get_DW_AT_signature_type): New functions.

Update copyright year on elf-ifunc.c

* elf-ifunc.c: Update copyright year.

Fix line length problem in last entry.

gdb/

* aarch64-tdep.c (aarch64_write_pc): Removed.
(aarch64_gdbarch_init): Remove set_gdbarch_write_pc of the above function.

* coffcode.h: Added a cast to void when a bfd_set_section_*()
macro's return value is ignored.
* elf32-hppa.c: Likewise.
* elf32-tic6x.c: Likewise.
* mach-o.c: Likewise.
* mmo.c: Likewise.
* opncls.c: Likewise.
* peicode.h: Likewise.
* elf32-m32r.c: Check return value of bfd_set_section_*().
* elfnn-ia64.c: Likewise.
* elfxx-mips.c: Likewise.
* vms-alpha.c: Likewise.

PR binutils/15369
* cgen-dis.c (hash_insn_array): Use CGEN_CPU_INSN_ENDIAN instead
of CGEN_CPU_ENDIAN.
(hash_insns_list): Likewise.

Use AC_DEFINE for USE_THREAD_DB

Use AC_DEFINE for USE_THREAD_DB instead of manually passing it down
through -D flags.

gdb/gdbserver/
2013-04-17  Pedro Alves  <palves@redhat.com>

* configure.ac (USE_THREAD_DB): Delete variable.
(if test "$srv_linux_thread_db" = "yes"): AC_DEFINE USE_THREAD_DB.
Don't AC_SUBST USE_THREAD_DB.
* Makefile.in (INTERNAL_CFLAGS): Remove @USE_THREAD_DB@.
* config.in, configure: Regenerate.

gdb/
* top.c (print_gdb_configuration): Print configure-time
parameter on using libbabeltrace or not.

daily update

*** empty log message ***

Only define 'struct lwp_info'::thread_known if using libthread-db.

A small cleanup.  'struct lwp_info'::thread_known is only useful for
thread-db.c.

gdbserver/
2013-04-16  Pedro Alves  <palves@redhat.com>

* linux-low.h (struct lwp_info) <thread_known>: Move under
the USE_THREAD_DB #ifdef.

Fix remaining GDBserver issues with !HAVE_THREAD_DB_H.

The previous patches are still not sufficient to build gdbserver with
our copy of thread_db.h.

../../../src/gdb/gdbserver/thread-db.c: In function ‘find_one_thread’:
../../../src/gdb/gdbserver/thread-db.c:316:6: error: ‘struct lwp_info’ has no member named ‘th’
../../../src/gdb/gdbserver/thread-db.c: In function ‘attach_thread’:
../../../src/gdb/gdbserver/thread-db.c:341:6: error: ‘struct lwp_info’ has no member named ‘th’
../../../src/gdb/gdbserver/thread-db.c: In function ‘thread_db_get_tls_address’:
../../../src/gdb/gdbserver/thread-db.c:514:47: error: ‘struct lwp_info’ has no member named ‘th’
make: *** [thread-db.o] Error 1

First, linux-low.h is including <thread_db.h> directly instead of our
gdb_thread_db.h, although thread-db.c includes the latter.  Then the
'th' field of struct lwp_info is only defined if HAVE_THREAD_DB_H is
defined, which is not true if we're using our replacement copy of
thread_db.h.  We have a USE_THREAD_DB symbol defined if we're building
thread-db.c that's ideal for this, however, it's currently only
defined when compiling linux-low.c (through a Makefile rule).  The
patch makes it defined when compiling any file.

gdb/gdbserver/
2013-04-16  Pedro Alves  <palves@redhat.com>

* Makefile.in (INTERNAL_CFLAGS): Add @USE_THREAD_DB@.
(linux-low.o): Delete rule.
* linux-low.h: Always include "gdb_thread_db.h" instead of
conditionally including thread_db.h.
(struct lwp_info) <th>: Guard with #ifdef USE_THREAD_DB instead of
HAVE_THREAD_DB_H.

copyright.py: Don't update glibc_thread_db.h.

As glibc_thread_db.h is a 3rd party imported file, we should not
update its copyright year range in the new year process.

gdb/
2013-04-16  Pedro Alves  <palves@redhat.com>

* copyright.py (EXCLUDE_LIST): Add gdb/common/glibc_thread_db.h.

Fix previous entry.

Update glibc_thread_db.h from upstream.

Attempting to build gdbserver with our copy of thread_db.h yields:

In file included from ../../../src/gdb/gdbserver/../common/gdb_thread_db.h:4:0,
                 from ../../../src/gdb/gdbserver/thread-db.c:30:
../../../src/gdb/gdbserver/../common/glibc_thread_db.h:108:3: error: unknown type name ‘uint32_t’
In file included from ../../../src/gdb/gdbserver/../common/gdb_thread_db.h:4:0,
                 from ../../../src/gdb/gdbserver/thread-db.c:30:
../../../src/gdb/gdbserver/../common/glibc_thread_db.h:199:5: error: unknown type name ‘uintptr_t’
../../../src/gdb/gdbserver/../common/glibc_thread_db.h:269:3: error: unknown type name ‘intptr_t’
../../../src/gdb/gdbserver/../common/glibc_thread_db.h:270:3: error: unknown type name ‘intptr_t’

We used to have a workaround for this, but the patch to import
gnulib's stdint.h removed it:

  http://www.sourceware.org/ml/gdb-patches/2008-06/msg00050.html

and defs.h made to always include stdint.h.  However, gdbserver
doesn't include stdint.h in its equivalent server.h.

Rather than working around the issue, I've imported a more recent
version from glibc, which itself includes <stdint.h>.  Other than
copyright years and FSF snail mail address, the file hasn't been
touched since 2003 in glibc.  AFAICS, our version was updated last in
2000-09-03.

A note on the apparent license change: before the previous patch, this
file's contents were part of gdb_thread_db.h, and we can see that its
license's text was changed in this patch
<http://sourceware.org/ml/gdb-patches/2009-03/msg00251.html>.  That
was certainly just an easy to overlook grep/sed mistake that fell
through the cracks.

gdb/common/
2013-04-16  Pedro Alves  <palves@redhat.com>

* glibc_thread_db.h: Update from upstream glibc
(git 568035b7874a099087b77f7bba3e36a1173787b0).

Move fallback thread_db.h to a separate file.

Having this on a separate file makes it easier to import a new version
-- one can just copy over instead of having to care about preserving
the GDB-specific bits.

2013-04-16  Pedro Alves  <palves@redhat.com>

* common/gdb_thread_db.h [!HAVE_THREAD_DB_H]: Factor out to ...
* common/glibc_thread_db.h: ... this new file ...
* common/gdb_thread_db.h [!HAVE_THREAD_DB_H]: ... and include it.

PR build/11881: LIBTHREAD_DB_SO can be undefined.

This patch:

 http://sourceware.org/ml/gdb-patches/2009-04/msg00115.html

Changed behaviour by only defining LIBTHREAD_DB_SO if thread_db.h exists. The
definition of LIBTHREAD_DB_SO and LIBTHREAD_DB_SEARCH_PATH should be moved
outside of #ifdef HAVE_THREAD_DB_H.

This is based on a patch attached to the PR, however, it needed a
tweak, as it was it broke the HAVE_THREAD_DB_H path.

2013-04-16  Will Newton  <will.newton@gmail.com>
    Pedro Alves  <palves@redhat.com>

PR build/11881

* common/gdb_thread_db.h (LIBTHREAD_DB_SO)
(LIBTHREAD_DB_SEARCH_PATH): Move outside of #ifdef
HAVE_THREAD_DB_H.

* gdb.texinfo (Set Catchpoints): Mention earliest version of
GCC that has the SDT probe points.

* Makefile.am (emmo.c): Add a dependency upon scripttempl/DWARF.sc.

NEWS: Mention "set foo unlimited".

Mention "set foo unlimited" in NEWS, right below the "New options"
section.

2013-04-16  Pedro Alves  <palves@redhat.com>
    Eli Zaretskii  <eliz@gnu.org>

* NEWS: Mention "set foo unlimited".

*** empty log message ***

daily update

* dwarf2read.c (struct dwo_file): Replace member "cus" with "cu".
(struct create_dwo_cu_data): Renamed from create_dwo_info_table_data.
(create_dwo_cu_reader): Renamed from
create_dwo_debug_info_hash_table_reader.
(create_dwo_cu): Renamed from create_dwo_debug_info_hash_table.
Remove support for multiple CUs in a DWO file.
(open_and_init_dwo_file, lookup_dwo_cutu): Update.

* dwarf2read.c (create_debug_types_hash_table): Use hex_string
instead of phex.
(lookup_dwo_unit, create_dwo_debug_info_hash_table_reader): Ditto.
(create_dwo_in_dwp): Ditto.

Check regular reference without non-GOT reference

non_got_ref may not be set when building shared library. We need to set
non_got_ref if there are any non-PIC relocations.  But we only did this
when there were no PLT/GOT relocations.  It failed when there is a PLT
relocation.  This checkin moves the non_got_ref check out.

bfd/

2013-04-15  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/15371
* elf-ifunc.c (_bfd_elf_allocate_ifunc_dyn_relocs): Check
regular reference without non-GOT reference when building
shared library.

ld/testsuite/

2013-04-15  H.J. Lu  <hongjiu.lu@intel.com>

PR ld/15371
* ld-ifunc/ifunc-20-i386.d: New file.
* ld-ifunc/ifunc-20-x86-64.d: Likewise.
* ld-ifunc/ifunc-20.s: Likewise.
diff --git a/bfd/elf-ifunc.c b/bfd/elf-ifunc.c
index e56427d..7e7ec36 100644
--- a/bfd/elf-ifunc.c
+++ b/bfd/elf-ifunc.c
@@ -187,23 +187,20 @@ _bfd_elf_allocate_ifunc_dyn_relocs (struct bfd_link_info *info,

   htab = elf_hash_table (info);

+  /* When building shared library, we need to handle the case where it is
+     marked with regular reference, but not non-GOT reference since the
+     non-GOT reference bit may not be set here.  */
+  if (info->shared && !h->non_got_ref && h->ref_regular)
+    for (p = *head; p != NULL; p = p->next)
+      if (p->count)
+ {
+   h->non_got_ref = 1;
+   goto keep;
+ }
+
   /* Support garbage collection against STT_GNU_IFUNC symbols.  */
   if (h->plt.refcount <= 0 && h->got.refcount <= 0)
     {
-      /* When building shared library, we need to handle the case
-         where it is marked with regular reference, but not non-GOT
-  reference.  It may happen if we didn't see STT_GNU_IFUNC
-  symbol at the time when checking relocations.  */
-      if (info->shared
-   && !h->non_got_ref
-   && h->ref_regular)
- for (p = *head; p != NULL; p = p->next)
-   if (p->count)
-     {
-       h->non_got_ref = 1;
-       goto keep;
-     }
-
       h->got = htab->init_got_offset;
       h->plt = htab->init_plt_offset;
       *head = NULL;
diff --git a/ld/testsuite/ld-ifunc/ifunc-20-i386.d b/ld/testsuite/ld-ifunc/ifunc-20-i386.d
new file mode 100644
index 0000000..9373fcf
--- /dev/null
+++ b/ld/testsuite/ld-ifunc/ifunc-20-i386.d
@@ -0,0 +1,13 @@
+#source: ifunc-20.s
+#ld: -shared -m elf_i386 -z nocombreloc
+#as: --32
+#readelf: -r --wide
+#target: x86_64-*-* i?86-*-*
+
+Relocation section '.rel.ifunc' at offset 0x[0-9a-f]+ contains 1 entries:
+[ ]+Offset[ ]+Info[ ]+Type[ ]+.*
+[0-9a-f]+[ ]+[0-9a-f]+[ ]+R_386_32[ ]+ifunc\(\)[ ]+ifunc
+
+Relocation section '.rel.plt' at offset 0x[0-9a-f]+ contains 1 entries:
+[ ]+Offset[ ]+Info[ ]+Type[ ]+.*
+[0-9a-f]+[ ]+[0-9a-f]+[ ]+R_386_JUMP_SLOT[ ]+ifunc\(\)[ ]+ifunc
diff --git a/ld/testsuite/ld-ifunc/ifunc-20-x86-64.d b/ld/testsuite/ld-ifunc/ifunc-20-x86-64.d
new file mode 100644
index 0000000..39492d4
--- /dev/null
+++ b/ld/testsuite/ld-ifunc/ifunc-20-x86-64.d
@@ -0,0 +1,13 @@
+#source: ifunc-20.s
+#ld: -shared -m elf_x86_64 -z nocombreloc
+#as: --64
+#readelf: -r --wide
+#target: x86_64-*-*
+
+Relocation section '.rela.ifunc' at offset 0x[0-9a-f]+ contains 1 entries:
+[ ]+Offset[ ]+Info[ ]+Type[ ]+.*
+[0-9a-f]+[ ]+[0-9a-f]+[ ]+R_X86_64_64[ ]+ifunc\(\)[ ]+ifunc \+ 0
+
+Relocation section '.rela.plt' at offset 0x[0-9a-f]+ contains 1 entries:
+[ ]+Offset[ ]+Info[ ]+Type[ ]+.*
+[0-9a-f]+[ ]+[0-9a-f]+[ ]+R_X86_64_JUMP_SLOT[ ]+ifunc\(\)[ ]+ifunc \+ 0
diff --git a/ld/testsuite/ld-ifunc/ifunc-20.s b/ld/testsuite/ld-ifunc/ifunc-20.s
new file mode 100644
index 0000000..9d45455
--- /dev/null
+++ b/ld/testsuite/ld-ifunc/ifunc-20.s
@@ -0,0 +1,16 @@
+ .section .data.rel,"aw",@progbits
+ .globl ifunc_ptrt
+ .type ifunc_ptr, @object
+ifunc_ptr:
+ .dc.a ifunc
+ .text
+ .type ifunc, @gnu_indirect_function
+ .globl ifunc
+ifunc:
+ ret
+ .size ifunc, .-ifunc
+ .type bar, @function
+ .globl bar
+bar:
+ call ifunc@PLT
+ .size bar, .-bar

2013-04-15  Siva Chandra Reddy  <sivachandra@google.com>

Add option to link testcases with Pthreads library when
using 'prepare_for_testing' in tests.

testsuite/
* lib/gdb.exp (build_executable_from_specs): Use
gdb_compile_pthreads to compile if option "pthreads" is
specified.

* NEWS: Move recent entries into "since 7.6" section.

PR c++/13588:
* NEWS: Update.
* break-catch-throw.c (struct exception_catchpoint)
<exception_rx, pattern>: New fields.
(fetch_probe_arguments, dtor_exception_catchpoint)
(check_status_exception_catchpoint)
(print_one_detail_exception_catchpoint): New functions.
(handle_gnu_v3_exceptions): Add "except_rx" argument.
Compile regular expression if needed.
(extract_exception_regexp): New function.
(catch_exception_command_1): Use extract_exception_regexp.
(compute_exception): Use fetch_probe_arguments.
(initialize_throw_catchpoint_ops): Set dtor, print_one_detail,
and check_status fields.
* cp-abi.c (cplus_typename_from_type_info): New function.
* cp-abi.h (cplus_typename_from_type_info): Declare.
(struct cp_abi_ops) <get_typename_from_type_info>: New field.
* gdb_regex.h (compile_rx_or_error): Declare.
* gnu-v3-abi.c (gnuv3_get_typename_from_type_info): Update
comment.
(init_gnuv3_ops): Set get_type_from_type_info field.
* probe.c (compile_rx_or_error): Move...
* utils.c (compile_rx_or_error): ... here.
gdb/doc
* gdb.texinfo (Set Catchpoints): Document regexp syntax for
exception catchpoints.
gdb/testsuite
* gdb.cp/exceptprint.exp: Add regexp catchpoint tests.

PR c++/15176:
* NEWS: Update.
* break-catch-throw.c (compute_exception): New function.
(exception_funcs): New global.
(_initialize_break_catch_throw): Create $_exception.
* cp-abi.c (cplus_type_from_type_info): New function.
* cp-abi.h (cplus_type_from_type_info): Declare.
(struct cp_abi_ops) <get_type_from_type_info>: New field.
* gnu-v3-abi.c (gnuv3_get_typename_from_type_info)
(gnuv3_get_type_from_type_info): New functions.
(init_gnuv3_ops): Set get_type_from_type_info ABI field.
gdb/doc
* gdb.texinfo (Set Catchpoints): Document $_exception.
(Convenience Vars): Mention $_exception.
gdb/testsuite
* gdb.base/default.exp: Update for $_exception.
* gdb.cp/exceptprint.cc: New file.
* gdb.cp/exceptprint.exp: New file.
* lib/gdb.exp (skip_libstdcxx_probe_tests): New proc.

* break-catch-throw.c (struct exception_names): New.
(exception_functions): Change type.
(re_set_exception_catchpoint): Look for SDT probes.

PR c++/10119:
* break-catch-throw.c (exception_functions): New global.
(gnu_v3_exception_catchpoint_ops): Move earlier.
(struct exception_catchpoint): New.
(classify_exception_breakpoint): Rewrite.
(re_set_exception_catchpoint): New function.
(handle_gnu_v3_exceptions): Return void.  Use init_catchpoint.
Allocate a struct exception_catchpoint.
(catch_exception_command_1): Update.
(initialize_throw_catchpoint_ops): Set 're_set' method.

* Makefile.in (SFILES): Add break-catch-throw.c
(COMMON_OBS): Add break-catch-throw.o.
* break-catch-throw.c: New file.
* breakpoint.c: Move exception-catching code to new file.
(ep_parse_optional_if_clause): No longer static.
* breakpoint.h (ep_parse_optional_if_clause): Declare.

PR c++/9065:
* NEWS: Update.
* breakpoint.c (watchpoint_exp_is_const): Add OP_TYPEID.
* c-exp.y (TYPEID): New token.
(exp): Add new TYPEID productions.
(ident_tokens): Add "typeid".
* cp-abi.c (cplus_typeid, cplus_typeid_type): New functions.
* cp-abi.h (cplus_typeid, cplus_typeid_type): Declare.
(struct cp_abi_ops) <get_typeid, get_typeid_type>: New fields.
* eval.c (evaluate_subexp_standard) <OP_TYPEID>: New case.
* expprint.c (dump_subexp_body_standard) <OP_TYPEID>: New
case.
* gnu-v3-abi.c (std_type_info_gdbarch_data): New global.
(build_std_type_info_type, gnuv3_get_typeid_type)
(gnuv3_get_typeid): New functions.
(init_gnuv3_ops): Initialize std_type_info_gdbarch_data.  Set
new fields on ABI object.
* parse.c (operator_length_standard) <OP_TYPEID>: New case.
* std-operator.def (OP_TYPEID): New.
gdb/testsuite
* gdb.cp/typeid.cc: New file.
* gdb.cp/typeid.exp: New file.

* elfread.c (elf_symtab_read): Install versioned symbol under
unversioned name as well.

PR c++/11990:
* c-lang.c (cplus_language_defn): Use gdb_demangle.
* c-typeprint.c (c_type_print_base): Use gdb_demangle.
* cp-support.c (mangled_name_to_comp): Use gdb_demangle.
(gdb_demangle): New function.
* cp-support.h (gdb_demangle): Declare.
* dwarf2read.c (dwarf2_physname, fixup_partial_die)
(dwarf2_name): Use gdb_demangle.
* gdbtypes.c (check_stub_method): Use gdb_demangle.
* gnu-v3-abi.c (gnuv3_rtti_type): Strip @plt and version
suffixes from name.
(gnuv3_print_method_ptr): Use gdb_demangle.
* jv-lang.c (java_demangle): Use gdb_demangle.
* jv-typeprint.c (java_type_print_base): Use gdb_demangle.
* language.c (unk_lang_demangle): Use gdb_demangle.
* symtab.c (symbol_find_demangled_name)
(demangle_for_lookup): Use gdb_demangle.

PR c++/12824:
* NEWS: Update.
* breakpoint.c (enum exception_event_kind) <EX_EVENT_RETHROW>:
New constant.
(classify_exception_breakpoint): New function.
(print_it_exception_catchpoint, print_one_exception_catchpoint)
(print_mention_exception_catchpoint)
(print_recreate_exception_catchpoint, handle_gnu_v3_exceptions)
(catch_exception_command_1): Handle "rethrow" catchpoint.
(catch_rethrow_command): New function.
(_initialize_breakpoint): Add "catch rethrow" command.
gdb/doc
* gdb.texinfo (Set Catchpoints): Reorganize exception
catchpoints.  Document "catch rethrow".
(Debugging C Plus Plus): Mention "catch rethrow".
gdb/testsuite
* gdb.cp/exception.exp: Add "catch rethrow" tests.

* gdb.texinfo (Set Catchpoints): Remove obsolete text.

gold/
* layout.cc (Layout::set_relocatable_section_offsets): Don't
allocate space in file for BSS sections.

gold/
* script-sections.cc (Orphan_output_section): Reset address
to zero after each orphaned section for relocatable links.

gold/
     * symtab.cc (Symbol_table::sized_write_globals): Subtract
     section starting address for relocatable link.
     * testsuite/Makefile.am (script_test_11): New test.
     * testsuite/Makefile.in: Regenerate.
     * testsuite/script_test_11.c: New source file.
     * testsuite/script_test_11.t: New linker script.

* contrib/ari/gdb_ari.sh (write_pc rule): Do not consider
set_gdbarch_write_pc as deprecated anymore.

Add missing empty line in spu_write_pc.

gdb/ChangeLog:

        * spu-tdep.c (spu_write_pc): Add empty line after local variable
        declarations.

* Makefile.am (ELF_DEPS): Add a dependency upon
scripttempl/DWARF.sc.
(ELF_GEN_DEPS): Likewise.
* Makefile.in: Regenerate.

gas/
    * expr.c (add_to_result, subtract_from_result): Make global.
    * expr.h (add_to_result, subtract_from_result): Add prototypes.
    * config/tc-sh.c (sh_optimize_expr): Use add_to_result,
    subtract_from_result to handle extra bit of precision for .sleb128
    directive operands.

    gas/testsuite/
    * gas/all/gas.exp (sleb128-7): Don't run for tic4x, tic54x.
    * gas/all/sleb128-2.s: Reformat, use _ at start of labels, remove
    cruft.
    * gas/all/sleb128-3.s: Likewise.
    * gas/all/sleb128-4.s: Likewise.
    * gas/all/sleb128-5.s: Likewise.
    * gas/all/sleb128-7.s: Likewise.
    * gas/all/sleb128-2.d: Handle data sections named $DATA$.
    * gas/all/sleb128-3.d: Likewise.
    * gas/all/sleb128-4.d: Likewise.
    * gas/all/sleb128-5.d: Likewise.
    * gas/all/sleb128-7.d: Likewise.

* scripttempl/armbpabi.sc: Replace DWARF sections with an
inclusion of DWARF.sc.
* 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.
* scripttempl/DWARF.sc: New.

* Makefile.am ($(MKDOC)): Append $(EXEEXT_FOR_BUILD) to temp file.
* Makefile.in: Regenerate.

* archive.c (_bfd_archive_close_and_cleanup): Clear parent
cache slot for archives.

daily update

*** empty log message ***

* mmo.c (mmo_write_chunk): Break out abfd->tdata.mmo_data to new
local variable mmop.

*** empty log message ***

daily update

gdb/

* ctf.c (_initialize_ctf): Include "completer.h".
Call add_target_with_completer instead of add_target.

gdb/testsuite/

* gdb.base/completion.exp: Test completion of command
'target ctf' if target ctf is supported.

* powerpc.cc (Stub_control::can_add_to_stub_group): Don't set
owner when sections are not adjacent and exceed group size.
(Target_powerpc::group_sections): Handle corner case.
(Target_powerpc::Branch_info::make_stub): Handle case where
stub table doesn't exist due to branches in non-exec sections.
(Target_powerpc::Relocate::relocate): Likewise.

daily update

*** empty log message ***

gdb/doc/
* gdb.texinfo (Auto-loading safe path): Add quick -iex using command
line below the sample output.

Re: "show configuration"/--configuration patch: mention --configuration too.

Clean up "show configuration" entry.

gdb/
Fix GDB regression related to PR binutils/14813.
* jit.c (mem_bfd_iovec_close): Return 0 for success.
* minidebug.c (lzma_close): Add return value comment.
* remote.c (remote_bfd_iovec_close): Return 0 for success.
* solib-spu.c (spu_bfd_iovec_close): Likewise.
* spu-linux-nat.c (spu_bfd_iovec_close): Likewise.

gdb/
* config.in: Regenerate.

* dwarf2-frame.c (struct comp_unit) <dwarf_frame_buffer>: Now
const.
* dwarf2read.c (struct dwarf2_section_info) <buffer>: Now const.
(struct die_reader_specs) <buffer>: Likewise.
(die_reader_func_ftype): Make 'info_ptr' const.
(struct line_header) <include_dirs, statement_program_start,
statement_program_end>: Now const.
(struct file_entry) <name>: Likewise.
(struct partial_die_info) <sibling>: Likewise.
(struct dwarf_block) <data>: Likewise.
(dwarf2_read_section): Remove cast.
(dwarf2_get_section_info): Make 'bufp' const.
(read_index_from_section): Constify.
(dw2_get_file_names_reader): Make 'info_ptr' const.
(dw2_get_primary_filename_reader): Likewise.
(read_comp_unit_head): Make 'info_ptr' and return type const.
(read_and_check_comp_unit_head, read_and_check_type_unit_head):
Likewise.
(read_abbrev_offset): Constify.
(dwarf2_create_include_psymtab): Make 'name' const.
(create_debug_types_hash_table): Update.
(read_cutu_die_from_dwo): Make 'result_info_ptr' const.
(init_cutu_and_read_dies, init_cutu_and_read_dies_no_follow):
Constify.
(process_psymtab_comp_unit_reader, build_type_psymtabs_reader)
(load_partial_comp_unit_reader): Make 'info_ptr' const.
(read_comp_units_from_section): Constify.
(peek_abbrev_code, peek_die_abbrev, skip_children, skip_one_die)
(locate_pdi_sibling, load_full_comp_unit_reader): Make 'info_ptr'
const.
(dwarf2_compute_name, setup_type_unit_groups): Constify.
(create_dwo_debug_info_hash_table): Make 'info_ptr' const.
(create_dwp_hash_table, dwarf2_ranges_read)
(dwarf2_record_block_ranges): Constify.
(read_die_and_children, read_die_and_siblings_1)
(read_die_and_siblings): Make 'info_ptr' and 'new_info_ptr'
const.
(read_full_die_1, read_full_die): Make 'info_ptr' const.
(abbrev_table_read_table): Constify.
(load_partial_dies): Make 'info_ptr' const.
(read_partial_die, read_attribute_value, read_attribute): Make
'info_ptr' and return type const.
(read_address, read_initial_length)
(read_checked_initial_length_and_offset, read_offset)
(read_offset_1, read_n_bytes, read_direct_string): Make 'buf'
const.
(read_direct_string): Make 'buf' and return type const.
(read_indirect_string_at_offset, read_indirect_string_from_dwz)
(read_indirect_string): Make return type const.
(read_unsigned_leb128, read_signed_leb128): Make 'buf' const.
(read_addr_index_from_leb128, dwarf2_read_addr_index_reader): Make
'info_ptr' const.
(read_str_index): Make return type const.
(add_include_dir): Make 'include_dir' const.
(add_file_name): Make 'name' const.
(dwarf_decode_line_header): Constify.
(psymtab_include_file_name): Make return type const.
(dwarf_decode_lines_1, dwarf_decode_lines): Constify.
(dwarf2_start_subfile): Make 'filename' const.
(dwarf2_const_value_attr): Make 'bytes' const.
(read_signatured_type_reader): Make 'info_ptr' const.
(decode_locdesc): Constify.
(skip_form_bytes): Make 'bytes', 'buffer_end', and return type
const.
(skip_unknown_opcode): Make 'opcode_definitions', 'mac_ptr',
'mac_end', and return type const.
(dwarf_parse_macro_header): Make 'mac_ptr' and return type const.
(dwarf_decode_macro_bytes): Make 'mac_ptr', 'mac_end', and return
type const.
(per_cu_header_read_in): Constify.
* symfile.h (dwarf2_get_section_info): Update.

* symtab.h (struct general_symbol_info) <value.bytes>: Make const.

NEWS: Document 'show configuration'.

Implement a new command to show GDB build-time configuration parameters.

* top.c (print_gdb_configuration): New function, displays the
details about GDB configure-time parameters.
(print_gdb_version): Mention "show configuration".

* cli/cli-cmds.c (show_configuration): New function.
(_initialize_cli_cmds): Add the "show configuration" command.

* main.c (captured_main) <print_configuration>: New static var.
<long_options>: Use it.
If --configuration was given, call print_gdb_configuration.

* doc/gdb.texinfo (Mode Options): Document '-configuration'.
(Help): Document "show configuration".
(Bug Reporting): Add requirements to include the configuration
details in bug reports.

*** empty log message ***

daily update

gdb/
Fix parsing tabs in ${gdb_target_obs}.
* configure.tgt (gdb_have_gcore): Replace case with for and if.

gdb/
* Makefile.in (HAVE_NATIVE_GCORE_TARGET): New.
(generated_files): Add gcore.
(install-only, uninstall): Add gcore if HAVE_NATIVE_GCORE_TARGET or
HAVE_NATIVE_GCORE_HOST.
(gcore): New.
* NEWS (Changes since GDB 7.6): Mention newly installed gcore.
* config/alpha/alpha-osf3.mh, config/i386/fbsd.mh,
config/i386/fbsd64.mh, config/i386/i386gnu.mh, config/i386/i386sol2.mh,
config/i386/sol2-64.mh, config/mips/irix5.mh, config/mips/irix6.mh,
config/powerpc/fbsd.mh, config/sparc/fbsd.mh, config/sparc/sol2.mh:
Add HAVE_NATIVE_GCORE_HOST.
* configure: Regenerate.
* configure.ac (HAVE_NATIVE_GCORE_TARGET): New, set it, AC_SUBST it.
New AC_SUBST fir GDB_TRANSFORM_NAME and GCORE_TRANSFORM_NAME.  New
AC_CONFIG_FILES for gcore.
* configure.tgt: Add gdb_have_gcore to the initial comment.  Set
gdb_have_gcore.
* gdb_gcore.sh: Rename to ...
* gcore.in: ... here.  Remove gcore.sh comment.  Use GDB_TRANSFORM_NAME
and GCORE_TRANSFORM_NAME substitutions.

gdb/doc/
* Makefile.in (MAN1S): Add gcore.1.
Remove "Host, target, and site specific Makefile fragments" comment.
(@host_makefile_frag@, HAVE_NATIVE_GCORE_TARGET): New.
(install-man1, uninstall-man1): Conditionalize gcore.1.
(gcore.1): New.
* gdb.texinfo (Man Pages): Add gcore man.
(gcore man): New node.

gdb/
* remote.c (unpush_and_perror): Add output message final dot.

gas/
    * read.c (convert_to_bignum): Add sign parameter. Use it
    instead of X_unsigned to determine sign of resulting bignum.
    (emit_expr): Pass extra argument to convert_to_bignum.
    (emit_leb128_expr): Use X_extrabit instead of X_unsigned. Pass
    X_extrabit to convert_to_bignum.
    (parse_bitfield_cons): Set X_extrabit.
    * expr.c (make_expr_symbol, expr_build_uconstant, operand):
    Initialise X_extrabit field as appropriate.
    (add_to_result): New.
    (subtract_from_result): New.
    (expr): Use above.
    * expr.h (expressionS): Add X_extrabit field.

    gas/testsuite/
    * gas/all/sleb128-2.s: New test.
    * gas/all/sleb128-3.s: Likewise.
    * gas/all/sleb128-4.s: Likewise.
    * gas/all/sleb128-5.s: Likewise.
    * gas/all/sleb128-7.s: Likewise.
    * gas/all/sleb128-2.d: New.
    * gas/all/sleb128-3.d: New.
    * gas/all/sleb123-4.d: New.
    * gas/all/sleb123-5.d: New.
    * gas/all/sleb123-7.d: New.
    * gas/all/gas.exp (sleb128-2, sleb128-3, sleb128-4, sleb128-5)
    (sleb128-7): Run new tests.

gdb/

* tracepoint.c (tfile_interp_line): Fit parameters line and
utpp in one line.

PR gold/15354
* powerpc.cc (Target_powerpc::make_brlt_section): Name section
.branch_lt to match bfd ld.  Adjust comments throughout file.

Do not overwrite so_list's so_name in solib_map_sections

gdb/ChangeLog:

        * solib.c (solib_map_sections): Remove code overwriting
        SO->SO_NAME with the bfd's filename.

*** empty log message ***

daily update

opcodes/
* rl78-dis.c (print_insn_rl78): Use alternative form as a GCC false
warning workaround.

Accept "set foo unlimited" in integer/uinteger/zuinteger_unlimited commands.

Currently, several commands take "0" or "-1" to mean "unlimited".

"show" knows when to print "unlimited":

 (gdb) show height
 Number of lines gdb thinks are in a page is 45.
 (gdb) set height 0
 (gdb) show height
 Number of lines gdb thinks are in a page is unlimited.

However, the user can't herself specify "unlimited" directly:

 (gdb) set height unlimited
 No symbol table is loaded.  Use the "file" command.
 (gdb)

This patch addresses that, by adjusting the set handler for all
integer/uinteger/zuinteger_unlimited commands to accept literal
"unlimited".  It also installs a completer.  Presently, we complete on
symbols by default, and at
<http://sourceware.org/ml/gdb-patches/2013-03/msg00864.html> I've
shown a WIP prototype that tried to keep that half working in these
commands.  In the end, it turned out to be more complicated than
justifiable, IMO.  It's super rare to want to pass the value of a
variable/symbol in the program to a GDB set/show knob.  That'll still
work, it's just that we won't assist with completion anymore.  This
patch just sticks with the simple, and completes on "unlimited", and
nothing else.  This simplification means that

  "set he<tab><tab>"

is all it takes to get to:

  "set height unlimited"

The patch then goes through all integer/uinteger/zuinteger_unlimited
commands in the tree, and updates both the online help and the manual
to mention that "unlimited" is accepted in addition to 0/-1.  In the
cases where the command had no online help text at all, this adds it.
I've tried to make the texts read in a way that "unlimited" is
suggested before "0" or "-1" is.

Tested on x86_64 Fedora 17.

gdb/
2013-04-10  Pedro Alves  <palves@redhat.com>

* cli/cli-decode.c (integer_unlimited_completer): New function.
(add_setshow_integer_cmd, add_setshow_uinteger_cmd)
(add_setshow_zuinteger_unlimited_cmd): Install the "unlimited"
completer.
* cli/cli-setshow.c: Include "cli/cli-utils.h".
(is_unlimited_literal): New function.
(do_set_command): Handle literal "unlimited" arguments.
* frame.c (_initialize_frame) <set backtrace limit>: Document
"unlimited".
* printcmd.c (_initialize_printcmd) <set print
max-symbolic-offset>: Add help text.
* record-full.c (_initialize_record_full) <set record full
insn-number-max>: Likewise.
* record.c (_initialize_record) <set record
instruction-history-size, set record function-call-history-size>:
Add help text.
* ser-tcp.c (_initialize_ser_tcp) <set tcp connect-timeout>: Add
help text.
* tracepoint.c (_initialize_tracepoint) <set trace-buffer-size>:
Likewise.
* source.c (_initialize_source) <set listsize>: Add help text.
* utils.c (initialize_utils) <set height, set width>: Likewise.
<set pagination>: Mention "set height unlimited".
* valprint.c (_initialize_valprint) <set print elements, set print
repeats>: Document "unlimited".

gdb/doc/
2013-04-10  Pedro Alves  <palves@redhat.com>

* gdb.texinfo (Process Record and Replay): Document that "set
record full insn-number-max", "set record
instruction-history-size" and "set record
function-call-history-size" accept "unlimited".
(Backtrace): Document that "set backtrace limit" accepts
"unlimited".
(List): Document that "set listsize" accepts "unlimited".
(Print Settings)" Document that "set print max-symbolic-offset",
"set print elements" and "set print repeats" accept "unlimited".
(Starting and Stopping Trace Experiments): Document that "set
trace-buffer-size" accepts "unlimited".
(Remote Configuration): Document that "set tcp connect-timeout"
accepts "unlimited".
(Command History): Document that "set history size" accepts
"unlimited".
(Screen Size): Document that "set height" and "set width" accepts
"unlimited".  Adjust "set pagination"'s description to suggest
"set height unlimited" instead of "set height 0".

gdb/testsuite/
2013-04-10  Pedro Alves  <palves@redhat.com>

* gdb.base/completion.exp: Test "set height", "set listsize" and
"set trace-buffer-size" completion.
* gdb.base/setshow.exp: Test "set height unlimited".
* gdb.trace/trace-buffer-size.exp: Test "set trace-buffer-size
unlimited".