Properly handle multiple operands for x32 quad.

gas/

2011-03-29  H.J. Lu  <hongjiu.lu@intel.com>

* config/tc-i386.c (handle_quad): Properly handle multiple
operands.

gas/testsuite/

2011-03-29  H.J. Lu  <hongjiu.lu@intel.com>

* gas/i386/ilp32/quad.d: Add tests for multiple operands.
* gas/i386/ilp32/quad.s: Likewise.

gas: blackfin: gas: blackfin: reject invalid BYTEUNPACK insns

The destination registers must be different with BYTEUNPACK insns,
otherwise the hardware throws up an exception.  So reject them.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

gas: blackfin: gas: blackfin: reject invalid BYTEOP16M insns

The destination registers must be different with BYTEOP16M insns,
otherwise the hardware throws up an exception.  So reject them.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

gas: blackfin: gas: blackfin: reject invalid BYTEOP16P insns

The destination registers must be different with BYTEOP16P insns,
otherwise the hardware throws up an exception.  So reject them.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

PR ld/12613
* ldlex.l (lex_warn_invalid): Don't assume char is unsigned.

* coffdump.c: Include bfd_stdint.h

libdecnumber: start a gitignore

sim: bfin: fix sign extension with 16bit acc add insns

The current implementation attempts to handle the 16bit sign extension
itself.  Unfortunately, it gets it right in some cases.  So rather than
fix that logic, just drop it in favor of using 16bit signed casts.  Now
gcc will take care of getting the logic right.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

gas: blackfin: reject invalid 16bit acc add insns

The 16bit acc add insn cannot assign the two results to the same dreg,
so make sure gas rejects attempts to use this insn variant.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

*** empty log message ***

daily update

Support .quad for x32.

gas/

2011-03-28  H.J. Lu  <hongjiu.lu@intel.com>

* config/tc-i386.c (handle_quad): New.
(md_pseudo_table): Add "quad".

gas/testsuite/

2011-03-28  H.J. Lu  <hongjiu.lu@intel.com>

* gas/i386/ilp32/inval.s: Remove .quad.
* gas/i386/ilp32/inval.l: Updated.

* gas/i386/ilp32/quad.d: New.
* gas/i386/ilp32/quad.s: Likewise.

gdb/testsuite/
Test STT_GNU_IFUNC support.
* gdb.base/gnu-ifunc-lib.c: New file.
* gdb.base/gnu-ifunc.c: New file.
* gdb.base/gnu-ifunc.exp: New file.

gdb/
Support resolution of STT_GNU_IFUNC via breakpoints.
* breakpoint.c (print_it_typical): Support bp_gnu_ifunc_resolver and
bp_gnu_ifunc_resolver_return.
(bpstat_what): Rename parameter to bs_head, new variable bs, adjust
the loop.  Support bp_gnu_ifunc_resolver and
bp_gnu_ifunc_resolver_return.  New comment after the loop.  New loop
for bp_gnu_ifunc_resolver and bp_gnu_ifunc_resolver_return
breakpoints.
(bptype_string, print_one_breakpoint_location): Support
bp_gnu_ifunc_resolver and bp_gnu_ifunc_resolver_return.
(user_settable_breakpoint): Return true also for
bp_gnu_ifunc_resolver.
(allocate_bp_location): Support bp_gnu_ifunc_resolver and
bp_gnu_ifunc_resolver_return.
(set_breakpoint_location_function): New parameter explicit_loc,
describe it.  Call find_pc_partial_function_gnu_ifunc with new
variable IS_GNU_IFUNC and adjust the address for STT_GNU_IFUNC if
EXPLICIT_LOC is not set.
(set_raw_breakpoint): Set EXPLICIT_LOC for
set_breakpoint_location_function.
(clone_momentary_breakpoint): Use true for EXPLICIT_LOC of
set_breakpoint_location_function.
(mention): Support bp_gnu_ifunc_resolver and
bp_gnu_ifunc_resolver_return.
(add_location_to_breakpoint): Set EXPLICIT_LOC for
set_breakpoint_location_function.
(update_breakpoint_locations): Remove static.
(breakpoint_re_set_one): Support bp_gnu_ifunc_resolver and
bp_gnu_ifunc_resolver_return.
* breakpoint.h (enum bptype): New fields bp_gnu_ifunc_resolver and
bp_gnu_ifunc_resolver_return.
(update_breakpoint_locations): New declaration.
* elfread.c: Include gdbthread.h and regcache.h.
(elf_gnu_ifunc_resolver_stop, elf_gnu_ifunc_resolver_return_stop): New
functions.
(elf_gnu_ifunc_fns): Install them.
* minsyms.c (stub_gnu_ifunc_resolver_stop)
(stub_gnu_ifunc_resolver_return_stop): New functions.
(stub_gnu_ifunc_fns): Install them.
* symtab.h (struct gnu_ifunc_fns): New fields gnu_ifunc_resolver_stop
and gnu_ifunc_resolver_return_stop.
(gnu_ifunc_resolver_stop, gnu_ifunc_resolver_return_stop): New.

gdb/
STT_GNU_IFUNC reader implementation.
* elfread.c: Include gdbtypes.h, value.h and infcall.h.
(SYMBOL_GOT_PLT_SUFFIX, elf_rel_plt_read)
(elf_objfile_gnu_ifunc_cache_data, struct elf_gnu_ifunc_cache)
(elf_gnu_ifunc_cache_hash, elf_gnu_ifunc_cache_eq)
(elf_gnu_ifunc_record_cache, elf_gnu_ifunc_resolve_by_cache)
(elf_gnu_ifunc_resolve_by_got, elf_gnu_ifunc_resolve_name)
(elf_gnu_ifunc_resolve_addr): New.
(elf_symfile_read): Call elf_rel_plt_read.
(elf_gnu_ifunc_fns): New.
(_initialize_elfread): Initialize elf_objfile_gnu_ifunc_cache_data.
Install elf_gnu_ifunc_fns.
* infcall.c (find_function_return_type): New function.
(find_function_addr): Resolve TYPE_GNU_IFUNC functions, if possible.
* minsyms.c (stub_gnu_ifunc_resolve_addr)
(stub_gnu_ifunc_resolve_name): New functions.
(stub_gnu_ifunc_fns, gnu_ifunc_fns_p): New variables.
* symtab.h (struct gnu_ifunc_fns, gnu_ifunc_resolve_addr)
(gnu_ifunc_resolve_name, gnu_ifunc_fns_p): New.

gdb/
Code cleanup for later STT_GNU_IFUNC support.
* infcall.c (find_function_addr): Remove variable code, use explicit
dereferences for it.  Move VALUE_TYPE initialization later.

gdb/
GDB find_pc_partial_function support for STT_GNU_IFUNC.
* blockframe.c (cache_pc_function_is_gnu_ifunc): New variable.
(clear_pc_function_cache): Clear it.
(find_pc_partial_function): Rename to ...
(find_pc_partial_function_gnu_ifunc): ... this function.  New
parameter is_gnu_ifunc_p, describe it.  Set *IS_GNU_IFUNC_P.
(find_pc_partial_function): New wrapper for this function.
* symtab.h (find_pc_partial_function_gnu_ifunc): New declaration.

gdb/
GDB internal type support for STT_GNU_IFUNC.
* elfread.c (record_minimal_symbol): Support mst_text_gnu_ifunc.
(elf_symtab_read): Set mst_text_gnu_ifunc for
BSF_GNU_INDIRECT_FUNCTION.
* eval.c (evaluate_subexp_standard): Support TYPE_GNU_IFUNC.
* gdbtypes.c (init_type): Support TYPE_FLAG_GNU_IFUNC,
builtin_func_func, nodebug_text_gnu_ifunc_symbol and
nodebug_got_plt_symbol.
* gdbtypes.h (enum type_flag_value): New entry TYPE_FLAG_GNU_IFUNC.
(TYPE_GNU_IFUNC): New.
(struct main_type): New field flag_gnu_ifunc.
(struct builtin_type): New field builtin_func_func.
(struct objfile_type): New fields nodebug_text_gnu_ifunc_symbol and
nodebug_got_plt_symbol.
* minsyms.c (lookup_minimal_symbol_text): Support mst_text_gnu_ifunc.
(in_gnu_ifunc_stub): New.
(prim_record_minimal_symbol, find_solib_trampoline_target): Support
mst_text_gnu_ifunc.
* parse.c (write_exp_msymbol): New variable ifunc_msym.  Detect and
support mst_text_gnu_ifunc.  Support mst_slot_got_plt.
* solib-svr4.c (svr4_in_dynsym_resolve_code): Return true also for
in_gnu_ifunc_stub.
* symmisc.c (dump_msymbols): Support mst_text_gnu_ifunc.
* symtab.c (search_symbols): Likewise.
* symtab.h (enum minimal_symbol_type): New fields mst_text_gnu_ifunc
and mst_slot_got_plt.
(in_gnu_ifunc_stub): New declaration.

gdb/
Support a ring of related breakpoints.
* breakpoint.c (watchpoint_del_at_next_stop): New, move here code from
other functions, add gdb_assert.
(update_watchpoint, watchpoint_check): Add gdb_assert.  Use
watchpoint_del_at_next_stop.
(bpstat_check_watchpoint): Use watchpoint_del_at_next_stop.
(bpstat_stop_status): Handle ring in related_breakpoint.
(set_raw_breakpoint_without_location): Initialize ring in
related_breakpoint.
(delete_breakpoint): Handle ring in related_breakpoint, use
watchpoint_del_at_next_stop.
(map_breakpoint_numbers): Handle ring in related_breakpoint.

gdb/testsuite/
Support a ring of related breakpoints.
* gdb.base/watchpoint-delete.c: New file.
* gdb.base/watchpoint-delete.exp: New file.

gdb/testsuite/
* gdb.ada/arrayparam.exp (print first after function call): Use
explicit package name.  Add a comment
(print lasta after function call): Rename ...
(print last after function call): ... it and use explicit package
name.
(print length after function call): Use explicit package name.
* gdb.ada/str_ref_cmp.exp (operator = works for strings): New test.
* gdb.ada/sym_print_name.exp: Change `i' to `integervar'.
(multiple matches for symbol i): Rename ...
(multiple matches for symbol integervar): ... it.
* gdb.ada/sym_print_name/foo.adb (Foo): Change `I' to `IntegerVar'.
* gdb.ada/sym_print_name/pck.ads (Pck): Likewise.

PR symtab/12441:
* dwarf2read.c (prepare_one_comp_unit): Don't call set_cu_language
with `language_minimal'.

* gdb.texinfo (Set Tracepoints): Fix typo.

* coffdump.c (coff_dump): Correct spelling error.
(show_usage): Replace SYSROFF by COFF.

* configure.ac (i[[3456789]]86-*-msdosdjgpp*): Don't disable
libffi on host.
(x86_64-*-mingw*, i[[3456789]]86-*-mingw32*): Don't disable newlib
on host.
(c54x*-*-* | tic54x-*-*): Don't disable newlib on host.
* configure: Regenerate.

* coffdump.c (dump_coff_scope): Use double typecast for pointer P
to allow compilation for all targets.

Add support for DragonFlyBSD target.

* configure.in: Check for windows.h, not Windows.h.
* configure: Regenerate.
* plugin.c: Include windows.h, not Windows.h.

*** empty log message ***

daily update

* archive.cc (Archive::interpret_header): Return -1 if something
goes wrong.  Change callers accordingly.

sim: bfin: handle saturation with RND12 sub insns

The current handling of the subtraction insn with the RND12 modifier
works when saturation isn't involved.  So add handling for this edge
case to match the hardware.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

*** empty log message ***

daily update

sim: bfin: add missing VS set with add/sub insns

The 16bit add/sub insns missed setting the VS bit in ASTAT whenever the
V bit was also set.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

*** empty log message ***

daily update

* testsuite/Makefile.am (final_layout.stdout): Use -n option with nm.
* testsuite/Makefile.in: Regenerate.

* arm-tdep.c (arm_elf_make_msymbol_special): Use ARM_SYM_BRANCH_TYPE
instead of checking for STT_ARM_TFUNC symbol type.

Regenerate Makefile.in.

oops - fix typo

* dwarf.c (decode_location_expression): Use dwarf_vmatoa
function to display DW_OP_addr OP.
(process_debug_info): Use dwarf_vma type for local variables
length and type_offset.

* dwarf.c (process_debug_info): Use offset_size to determine
the bit-size of the computation unit's offset.

* bucomm.c (make_tempname): Stop memory leak.
ICVS: ----------------------------------------------------------------------

* objdump.c (dump_reloc_set): Free malloced memory.

* stabs.c (parse_stab_enum_type): Fix memory leaks.
(parse_stab_struct_type): Ditto.
(parse_stab_struct_fields): Ditto.
(parse_stab_one_struct_field): Ditto.
(parse_stab_members): Ditto.
(stab_demangle_qualified): Ditto.

* prdbg.c (pr_function_type): Close memory leaks.
(pr_method_type): Likewise.
(tg_class_static_member): Likewise.
(tg_class_method_variant): Likewise.
(tg_class_static_method_variant): Likewise.

* ieee.c (ieee_add_bb11): Plug memory leak.
(ieee_function_type): Likewise.
(ieee_class_baseclass): Likewise.

* ar.c (write_archive): Plug memory leak.

* ar.c (delete_members): Plug memory leak.

* ar.c (move_members): Plug memory leak.

* linespec.c (symbol_found): Restore line-based result for
non-LOC_LABEL symbols.

* strings.c (print_strings): Plug memory leak.

* coffcode.h (coff_set_alignment_hook): Check return of bfd_seek.

merge from gcc

2011-03-24  Mark Wielaard  <mjw@redhat.com>

* dwarf2.h (dwarf_form): Remove deprecated DW_FORM_sig8 define.

2011-03-25  Tristan Gingold  <gingold@adacore.com>

* vms-alpha.c (_bfd_vms_find_nearest_dst_line): Allow the use
of find_nearest_line on object files.

2011-03-25  Tristan Gingold  <gingold@adacore.com>

* vms-alpha.c (evax_section_flags): Make $CODE$ section read-only.
Minor reordering.
(alpha_vms_create_eisd_for_section): Make code sections read-only.

Replace bfd_vma type and analog types by dwarf_vma and analogs.
Use dwarf specific print functions to display these type values.
* dwarf.h (dwarf_signed_vma): New type;
(DWARF2_External_LineInfo): Replace bfd_vma by dwarf_vma.
(DWARF2_External_PubNames): Likewise.
(DWARF2_External_CompUnit): Likewise.
(DWARF2_External_ARange): Likewise.
(read_leb128): Change return type to dwarf_vma.
* dwarf.c (print_dwarf_vma): Use __MINGW32__ conditional and
check byte_size values.
(dwarf_vmatoa): Change parameter type to dwarf_vma.
(dwarf_svmatoa): New static function.
(read_leb128): Change return type to dwarf_vma.
(read_sleb128): New static function.
(struct State_Machine_Registers): Change address field type to
dwarf_vma.
(process_extended_line_op): Adapt to type changes.
(fetch_indirect_string): Likewise.
(idisplay_block): Likewise.
(decode_location_expression): Likewise.
(read_and_display_attr_value): Likewise.
(process_debug_info): Likewise.
(display_debug_lines_raw): Likewise.
(display_debug_lines_decoded): Likewise.
(SLEB macro): Use new read_sleb128 function.

* configure.ac (native_only): Remove.
(i[[3456789]]86-*-msdosdjgpp*): Don't disable expect dejagnu
send-pr uudecode guile gnuserv on host.
(x86_64-*-mingw*): Don't disable expect dejagnu autoconf automake
send-pr rcs guile perl texinfo libtool on host.
(i[[3456789]]86-*-mingw32*): Don't disable expect dejagnu autoconf
automake send-pr rcs guile perl texinfo libtool on host.
(*-*-cygwin*, *-*-netbsd*): Remove host cases.
(*-*-kaos*): Don't disable target-examples target-gperf on target.
(alpha*-dec-osf*): Don't disable fileutils on target.
(sh*-*-pe|mips*-*-pe|*arm-wince-pe): Don't disable target-examples
texinfo send-pr expect dejagnu on target.
(arm-*-elf* | arm*-*-eabi*, arm*-*-linux-gnueabi): Don't disable
target-qthreads on target.
(hppa*-hp-hpux11*, hppa*-*-*): Don't disable shellutils on target.
(ia64*-*-elf*, ia64*-*-*vms*): Don't disable mmalloc on target.
(i[[3456789]]86-w64-mingw*, i[[3456789]]86-*-mingw*,
x86_64-*-mingw*): Don't disable expect on target.
(*-*-cygwin*): Don't disable target-gperf on target.
(powerpc*-*-winnt* | powerpc*-*-pe*): Don't disable make expect
gnuserv on target.
(powerpcle-*-solaris*): Don't disable make expect gnuserv on
target.
* configure: Regenerate.

Move environment modifications into gdb_init so it's done just once per
test, and clear the GREP_OPTIONS environment variable. See,
  http://sourceware.org/ml/gdb-patches/2011-03/msg00886.html

* configure.ac (target_tools): Remove target-groff.
(native_only): Remove target-groff.
(hppa*64*-*-*): Don't disable byacc.
(i[[3456789]]86-*-mingw32*): Remove commented-out noconfigdirs
setting.
(*-*-kaos*): Don't skip target-librx and target-groff.
(*-*-netware*): Don't skip target-libmudflap.
(*-*-tpf*): Don't skip target-libmudflap.
(sh*-*-pe|mips*-*-pe|*arm-wince-pe): Don't condition configured
directories on the host.
(ia64*-*-*vms*): Don't skip tix.
(sh-*-* | sh64-*-*): Don't condition skipped directories on the
host.
* configure: Regenerate.

2011-03-25  Kai Tietz  <ktietz@redhat.com>

        * tui/tui-source.c (tui_set_source_content): Use filename_cmp
        instead of strcmp for comparison.
        (tui_source_is_displayed): Likewise.
        * tui/tui-winsource.c (tui_update_breakpoint_info): Likewise.

sim: bfin: add hw tracing to gpio/sic port events

Makes it a lot easier to find out what's going on with interrupt lines
if the ports have tracing output.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: fix GPIO logic bugs when processing events

We need the DIR bit cleared, not set, in order for the pin to be treated
as an input.

When looking up the data value, we need to shift the "level" value over by
"my_port" rather than "bit" as the latter has already been shifted over.
We also should normalize the "level" coming in from the outside worlds to
the set of {0,1} since those are the only values that matter to GPIOs.

We need the BOTH bit set, not cleared, in order for the pin to trigger
on both edges.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: fix clear/set/toggle GPIO handling

The clear/set/toggle MMRs aren't backed by "real" data; they implicitly
perform bit operations on the associated data register.  So when we go
to process writes to them, we need to adjust the pointer accordingly so
that the actual backing data is modified.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

*** empty log message ***

daily update

2011-03-24  Mark Wielaard  <mjw@redhat.com>

        * dwarf2read.c (lookup_signatured_type): Use DW_FORM_ref_sig8 in
        complaint.
        (skip_one_die): Use DW_FORM_ref_sig8, not DW_FORM_sig8.
        (find_partial_die_in_comp_unit): Likewise in comment.
        (read_attribute_value): Likewise.
        (lookup_die_type): Likewise.
        (dwarf_form_name): Likewise.
        (dump_die_shallow): Likewise.
        (follow_die_ref_or_sig): Likewise.

gdb
PR breakpoints/11816:
* linespec.c (decode_line_1): Parse `function:label' linespecs.
(decode_compound): Update.
(find_function_symbol): New function.
(decode_dollar): Update.
(decode_label): Add 'function_symbol' parameter.  Handle
function-relative labels.
(decode_variable): Update.
(symbol_found): Add 'function_symbol' parameter.  Use label's PC,
not its line.  Set `special_display' and canonical name for
labels.
gdb/doc
* gdb.texinfo (Specify Location): Document `function:label'
linespec.
gdb/testsuite
* gdb.base/label.exp: Use `main:there' linespec.  Add re-running
test.

* linespec.h (struct linespec_result) <special_display>: New
field.
* breakpoint.h (struct breakpoint) <display_canonical>: New
field.
* breakpoint.c (print_breakpoint_location): Respect
display_canonical.
(create_breakpoint_sal): Add 'display_canonical' parameter.
(create_breakpoints_sal): Update.
(create_breakpoint): Update.

* symtab.c (decode_line_spec): Update.
* linespec.c (build_canonical_line_spec): Change type of
'canonical'.
(decode_line_2, decode_line_1, decode_objc, decode_compound)
(find_method, decode_all_digits, decode_dollar, decode_label)
(symbol_found): Likewise.
(init_linespec_result): New function.
* breakpoint.c (struct captured_parse_breakpoint_args)
<canonical_p>: New field, replaces addr_string_p.
(create_breakpoints_sal): Add 'canonical' parameter, replacing
'addr_string'.
(parse_breakpoint_sals): Likewise.
(do_captured_parse_breakpoint): Update.
(create_breakpoint): Use struct linespec_result.
(until_break_command): Update.
(breakpoint_re_set_one): Update.
(decode_line_spec_1): Update.
* linespec.h (struct linespec_result): New.
(init_linespec_result): Declare.

2011-03-24  Paolo Bonzini  <pbonzini@redhat.com>

        * configure.ac: Remove references to mt-mep, mt-netware,
        mt-wince.
        * Makefile.def: Add all-utils soft dependencies.
        * Makefile.tpl: Remove GDB_NLM_DEPS.
        * configure: Regenerate.
        * Makefile.in: Regenerate.

config:
2011-03-24  Paolo Bonzini  <pbonzini@redhat.com>

        * mt-mep: Remove, obsolete.
        * mt-netware: Remove, obsolete.
        * mt-wince: Remove, obsolete.
        * mt-v810: Remove, unused.

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

Sync from GCC:

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

* configure.ac: Remove empty cases.
* configure: Regenerate.

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

* Makefile.def: Add dependency from termcap to gdb.
* Makefile.in: Regenerate.

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

* configure.ac: Remove all mentions of mh-sysv4 and mh-solaris.
* configure: Regenerate.
* Makefile.def: Remove all mentions of X11_FLAGS_TO_PASS.
* Makefile.tpl: Likewise.
* Makefile.in: Regenerate.

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

* configure.ac: Remove all mentions of tentative_cc.
* configure: Regenerate.

2011-03-16  Jack Howarth  <howarth@bromo.med.uc.edu>

PR lto/48086
* configure.ac: Re-enable LTO on *-apple-darwin9.
* configure: Regenerate.

config:
2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

Sync from GCC:

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

* mh-cygwin: Remove obsolete variables and dependencies.

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

* mh-sysv4: Remove.
* mh-solaris: Remove.

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

* mh-sysv4: Remove AR_CFLAGS.

2011-03-24  Eric B. Weddington  <eric.weddington@atmel.com>

* config/tc-avr.c (mcu_types): Add new devices: atmega325pa,
atmega3250pa, atmega3290pa, atmega16hvbrevb, atmega32hvbrevb,
at90pwm161.
* doc/c-avr.texi: Document new device names.

2011-03-24  Paolo Bonzini  <bonzini@gnu.org>

* mh-sysv4: Remove AR_CFLAGS.

* configure.ac (i[[3456789]]86-*-vsta, i[[3456789]]86-*-go32*,
i[[3456789]]86-*-beos*, powerpc-*-beos*, m68k-hp-hpux*,
m68k-apollo-sysv*, m68k-apollo-bsd*, m88k-dg-dgux*,
m88k-harris-cxux*, m88k-motorola-sysv*, mips*-dec-ultrix*,
mips*-nec-sysv4*, mips*-sgi-irix4*, mips*-*-sysv4*, mips*-*-sysv*,
i370-ibm-opened*, i[[3456789]]86-*-sysv5*, i[[3456789]]86-*-dgux*,
i[[3456789]]86-ncr-sysv4.3*, i[[3456789]]86-ncr-sysv4*,
i[[3456789]]86-*-sco3.2v5*, i[[3456789]]86-*-sco*,
i[[3456789]]86-*-udk*, vax-*-ultrix2*, m68k-sun-sunos*,
hppa*-*-hiux*, *-*-hiux*, rs6000-*-lynxos*, *-*-sysv4*,
*-*-rhapsody*): Remove host cases.
* configure: Regenerate.

config:
* mh-cxux, mh-decstation, mh-dgux386, mh-lynxrs6k, mh-ncr3000,
mh-necv4, mh-sco, mh-sysv5: Remove.

* configure.ac (ppc*-*-pe): Remove host case.
(strongarm-*-coff | xscale-*-coff, strongarm-*-elf* |
xscale-*-elf*, thumb-*-coff, thumb-*-elf, thumb-*-pe, ep9312-*-elf
| ep9312-*-coff, parisc*64*-*-linux*, ppc*-*-pe): Remove target
cases.
* configure: Regenerate.

* regcache.c (regcache_raw_read): If the target didn't supply a
given raw register, mark it as unavailable.

* config.sub: Update to version 2011-03-23.

gas: blackfin: reject invalid register destinations for vector add/sub

The destination registers with vector add/sub insns must be different,
so make sure gas rejects attempt to write these.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

opcodes: blackfin: ignore (M) on MAC0-only dsp mac funcs

If the MAC1 part of the insn is disabled, then the (M) flag is ignored.
Rather than include it in the decode, move the MM clearing to the MAC0
portion of the code.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

gas: blackfin: test all 16bit insns

The current 16bit insn test doesn't actually cover all illegal insns
since it stops at 0xa000 instead of 0xc000.  But rather than address
that, replace it with a test that covers all 16bit insns.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

gas: blackfin: catch invalid dest dregs in dsp mult insns

While we were catching a few mismatches in vectorized dsp mult insns,
the error we displayed was misleading.  Once we fix that up, we can
convert previously dead code into proper checking for destination
dreg matching.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

gas: blackfin: catch invalid register combinations with SEARCH/BITMUX

The destination registers for SEARCH cannot be the same.  Same rule
for the source registers for BITMUX.

Signed-off-by: Mike Frsyinger <vapier@gentoo.org>

sim: bfin: document SIC limitation

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: fix inverted W1C logic

When I originally wrote the w1c helper funcs, I used it in a few places.
Then I forgot how it worked and when I later documented it, I described
the 3rd arg in the exact opposite way it is actually used.  This error
propagated to a bunch of devices registers that were not explicitly
tested (a bunch of the devices are stubs which merely exist to say "no
device is connected" to make device drivers happy).

So once the documentation is unscrewed, fix all of the broken call sites.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: define more UART LSR bits

We'll need these bits in an upcoming patch, so map out the whole
LSR MMR now.

Fix up indentation style while we're here.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: fix typo in TWI stat reg

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: update VIT_MAX behavior to match hardware when Acc.X bits are set

The Blackfin PRM says that the top 8 bits of the accumulator must be
cleared when using the VIT_MAX insn, so the sim has followed this spec.
Matching the hardware behavior though when the high bits are not cleared
is easy to do and doesn't break existing behavior, so go for it.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: always do 16bit sign extension with the SEARCH insn

The Blackfin PRM does not cover this case, but the hardware is clear: even
if the search criteria is not met (and thus a new 16bit value is loaded up
into the accumulator), the accumulator undergoes 16bit sign extension.  So
simply reload the low signed 16bits in that case.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: update AV and AC ASTAT bits with acc negation

The Acc=-Acc insn can overflow or carry with edge values, so make sure
we update the ASTAT bits accordingly to match the hardware.  Also fix
a thinko where we always updated AC0 even when working with A1 regs.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: fix thinko in SIC pin encoding

When encoding the SIC/pin info into unique input port ids, I used bases
of 100 when I meant to use 0x100.  Rather than simply fix the decoding
math in the different functions, create a few helper macros to simplify
the SIC/pin encoding and decoding steps.  This makes the resulting tables
nice & clear.

And now that pins are clear, the 533 and 537 port_event handlers may
easily be merged into one.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: allow byteop[123]p src regs to be the same

The hardware allows the byteop[123]p insns to use the same src reg pair,
so remove the combination check in the sim.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

sim: bfin: fix thinko in bfin_gpio bus addresses

The bus addresses have to be valid numbers, so 'g' and 'h' won't work.
Oddly, the common code silently ignored this which is why I didn't notice
in the first place.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>

*** empty log message ***

daily update

* lexsup.c (parse_args): Only set report_plugin_symbols if plugins
are enabled.  Mark level as possibly unused.

2011-03-23  Kai Tietz  <ktietz@redhat.com>

* breakpoint.c (clear_command): Use filename_cmp
instead of strcmp for comparison.
* buildsym.c (watch_main_source_file_lossage): Likewise.
(patch_subfile_names): Use IS_DIR_SEPARATOR instead of
checking just for slash.
* dbxread.c (read_dbx_symtab): Use lbasename instead of
strrchr and filename_cmp instead of strcmp for filenames.
(add_old_header_file): Use filename_cmp
instead of strcmp for comparison.
* exec.c (exec_set_section_address): Likewise.
* macrotab.c (macro_lookup_inclusion): Likewise.
(macro_lookup_inclusion): Likewise.
* elfread.c (_initialize_elfread): Likewise.
(elfstab_offset_sections): Likewise.
(elfstab_offset_sections): Use lbasename instead of
strrchr.
* mdebugread.c (parse_partial_symbols): Likewise.
(arse_partial_symbols): Use filename_(n)cmp instead of
str(n)cmp for comparison.
* minsyms.c (lookup_minimal_symbol): Likewise.
* psymtab.c (read_psymtabs_with_filename): Likewise.
* solib.c (solib_read_symbols): Likewise.
(reload_shared_libraries_1): Likewise.
* symmisc.c (maintenance_print_symbols): Likewise.
* symfile.c (separate_debug_file_exists): Likewise.
(reread_symbols): Likewise.
(find_separate_debug_file_by_debuglink): Likewise.
* remote-fileio.c (remote_fileio_func_rename): Likewise.
* source.c (add_path): Likewise.
* symtab.c (filename_seen): Likewise.
(file_matches): Likewise.
(print_symbol_info): Likewise.
(maybe_add_partial_symtab_filename): Likewise.
(make_source_files_completion_list): Likewise.
* xml-syscall.c (init_sysinfo): Likewise.
* windows-nat.c (_initialize_check_for_gdb_ini): Use
IS_DIR_SEPARATOR for checking for trailing path separator.

bfd/
* elf32-ppc.c (ppc_elf_tls_optimize): Catch more cases where
old-style __tls_get_addr calls without marker relocs don't match
their arg setup insn one for one.  If such mismatches are found
report the reloc and don't do any tls optimization.
* elf64-ppc.c (ppc64_elf_tls_optimize): Likewise.
ld/testsuite/
* ld-powerpc/tlsmark.s: Delete non-optimizable section.
* ld-powerpc/tlsmark32.s: Likewise.
* ld-powerpc/tlsmark.d: Adjust to suit.
* ld-powerpc/tlsmark32.d: Likewise.
* ld-powerpc/tlsopt1.d, * ld-powerpc/tlsopt1.s: New.
* ld-powerpc/tlsopt2.d, * ld-powerpc/tlsopt2.s: New.
* ld-powerpc/tlsopt3.d, * ld-powerpc/tlsopt3.s: New.
* ld-powerpc/tlsopt4.d, * ld-powerpc/tlsopt4.s: New.
* ld-powerpc/tlsopt1_32.d, * ld-powerpc/tlsopt1_32.s: New.
* ld-powerpc/tlsopt2_32.d, * ld-powerpc/tlsopt2_32.s: New.
* ld-powerpc/tlsopt3_32.d, * ld-powerpc/tlsopt3_32.s: New.
* ld-powerpc/tlsopt4_32.d, * ld-powerpc/tlsopt4_32.s: New.
* ld-powerpc/powerpc.exp: Run new tests.

2011-03-23  Eric B. Weddington  <eric.weddington@atmel.com>

* config/tc-avr.c (mcu_types): Add new xmega devices: atxmega64a1u,
atxmega128a1u, atxmega16x1, atxmega32x1, atxmega128b1, atxmega256a3bu.
* doc/c-avr.texi: Document new device names.