Joseph Myers [Fri, 12 Sep 2014 12:05:18 +0000 (13:05 +0100)]
Remove LIBGCC2_HAS_?F_MODE target macros.
This patch removes the LIBGCC2_HAS_{SF,DF,XF,TF}_MODE target macros,
replacing them by predefines with -fbuilding-libgcc, together with a
target hook that can influence those predefines when needed.
The new default is that a floating-point mode is supported in libgcc
if (a) it passes the scalar_mode_supported_p hook (otherwise it's not
plausible for it to be supported in libgcc) and (b) it's one of those
four modes (since those are the modes for which libgcc hardcodes the
possibility of support). The target hook can override the default
choice (in either direction) for modes that pass
scalar_mode_supported_p (although overriding in the direction of
returning true when the default would return false only makes sense if
all relevant functions are specially defined in libgcc for that
particular target).
The previous default settings depended on various settings such as
LIBGCC2_LONG_DOUBLE_TYPE_SIZE, as well as targets defining the above
target macros if the default wasn't correct.
The default scalar_mode_supported_p only declares a floating-point
mode to be supported if it matches one of float / double / long
double. This means that in most cases where a mode is only supported
conditionally in libgcc (TFmode only supported if it's the mode of
long double, most commonly), the default gets things right. Overrides
were needed in the following cases:
* SFmode would always have been supported in libgcc (the condition was
BITS_PER_UNIT == 8, true for all current targets), but pdp11
defaults to 64-bit float, and in that case SFmode would fail
scalar_mode_supported_p. I don't know if libgcc actually built for
pdp11 (and the port may well no longer be being used), but this
patch adds a scalar_mode_supported_p hook to it to ensure SFmode is
treated as supported.
* Certain i386 and ia64 targets need the new hook to match the
existing cases for when XFmode or TFmode support is present in
libgcc. For i386, the hook can always declare XFmode to be
supported - the cases where it's not are the cases where long double
is TFmode, in which case XFmode fails scalar_mode_supported_p[*] -
but TFmode support needs to be conditional. (And of the targets not
defining LIBGCC2_HAS_TF_MODE before this patch, some defined
LONG_DOUBLE_TYPE_SIZE to 64, so ensuring LIBGCC2_HAS_TF_MODE would
always be false, while others did not define it, so allowing it to
be true in the -mlong-double-128 case. This patch matches that
logic, although I suspect all the latter targets would have been
broken if you tried to enable -mlong-double-128 by default, for lack
of the soft-fp TFmode support in libgcc, which is separately
configured.)
[*] I don't know if it's deliberate not to support __float80 at all
with -mlong-double-128.
In order to implement the default version of the new hook,
insn-modes.h was made to contain macros such as HAVE_TFmode for each
machine mode, so the default hook can contain conditionals on whether
XFmode and TFmode exist (to match the hardcoding of a list of modes in
libgcc). This is also used in fortran/trans-types.c; previously it
had a conditional on defined(LIBGCC2_HAS_TF_MODE) (a bit dubious,
since it ignored the value of the macro), which is replaced by testing
defined(HAVE_TFmode), in conjunction with requiring
targetm.libgcc_floating_mode_supported_p.
(Fortran is testing something stronger than that hook: not only is
libgcc support required, but also libm or equivalent. Thus, it has a
test for ENABLE_LIBQUADMATH_SUPPORT in the case that the mode is
TFmode and that's not the same as any of the three standard types.
The old and new tests are intended to accept exactly the same set of
modes for all targets.)
Apart from the four target macros eliminated by this patch, it gets us
closer to eliminating LIBGCC2_LONG_DOUBLE_TYPE_SIZE as well, though a
few more places using that macro need changing first.
Bootstrapped with no regressions on x86_64-unknown-linux-gnu; also
built cc1 for crosses to ia64-elf and pdp11-none as a minimal test of
changes for those targets.
gcc/c-family:
* c-cppbuiltin.c (c_cpp_builtins): Define __LIBGCC_HAS_%s_MODE__
macros for floating-point modes.
gcc/fortran:
* trans-types.c (gfc_init_kinds): Check
targetm.libgcc_floating_mode_supported_p for floating-point
modes. Check HAVE_TFmode instead of LIBGCC2_HAS_TF_MODE.
libgcc:
* libgcc2.h (LIBGCC2_HAS_SF_MODE): Define using
__LIBGCC_HAS_SF_MODE__.
(LIBGCC2_HAS_DF_MODE): Define using __LIBGCC_HAS_DF_MODE__.
(LIBGCC2_HAS_XF_MODE): Define using __LIBGCC_HAS_XF_MODE__.
(LIBGCC2_HAS_TF_MODE): Define using __LIBGCC_HAS_TF_MODE__.
* config/libbid/bid_gcc_intrinsics.h
(LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Do not define.
(LIBGCC2_HAS_XF_MODE): Define using __LIBGCC_HAS_XF_MODE__.
(LIBGCC2_HAS_TF_MODE): Define using __LIBGCC_HAS_TF_MODE__.
* fixed-bit.h (LIBGCC2_LONG_DOUBLE_TYPE_SIZE): Do not define.
(LIBGCC2_HAS_SF_MODE): Define using __LIBGCC_HAS_SF_MODE__.
(LIBGCC2_HAS_DF_MODE): Define using __LIBGCC_HAS_DF_MODE__.
Marc Glisse [Fri, 12 Sep 2014 10:42:47 +0000 (12:42 +0200)]
tree.c (integer_each_onep): New function.
2014-09-12 Marc Glisse <marc.glisse@inria.fr>
gcc/
* tree.c (integer_each_onep): New function.
* tree.h (integer_each_onep): Declare it.
* fold-const.c (fold_binary_loc): Use it for ~A + 1 to -A and
-A - 1 to ~A. Disable (X & 1) ^ 1, (X ^ 1) & 1 and ~X & 1 to
(X & 1) == 0 for vector and complex.
gcc/testsuite/
* gcc.dg/vec-andxor1.c: New file.
[AArch64] Add regmove_costs for Cortex-A57 and A53
2014-09-12 Wilco Dijkstra <wdijkstr@arm.com>
* gcc/config/aarch64/aarch64.c: (cortexa57_regmove_cost): New cost table for
A57.
(cortexa53_regmove_cost): New cost table for A53. Increase GP2FP/FP2GP cost
to spilling from integer to FP registers.
gcc/
* ira.h (ira_finish_once): Delete.
* ira-int.h (target_ira_int::~target_ira_int): Declare.
(target_ira_int::free_ira_costs): Likewise.
(target_ira_int::free_register_move_costs): Likewise.
(ira_finish_costs_once): Delete.
* ira.c (free_register_move_costs): Replace with...
(target_ira_int::free_register_move_costs): ...this new function.
(target_ira_int::~target_ira_int): Define.
(ira_init): Call free_register_move_costs as a member function rather
than a global function.
(ira_finish_once): Delete.
* ira-costs.c (free_ira_costs): Replace with...
(target_ira_int::free_ira_costs): ...this new function.
(ira_init_costs): Call free_ira_costs as a member function rather
than a global function.
(ira_finish_costs_once): Delete.
* target-globals.c (target_globals::~target_globals): Call the
target_ira_int destructor.
* toplev.c: Include lra.h.
(finalize): Call lra_finish_once rather than ira_finish_once.
Jan Hubicka [Thu, 11 Sep 2014 23:16:42 +0000 (01:16 +0200)]
common.opt (flto-odr-type-merging): New flag.
* common.opt (flto-odr-type-merging): New flag.
* ipa-deivrt.c (hash_type_name): Use ODR names for hasing if availale.
(types_same_for_odr): Likewise.
(odr_subtypes_equivalent_p): Likewise.
(add_type_duplicate): Do not walk type variants.
(register_odr_type): New function.
* ipa-utils.h (register_odr_type): Declare.
(odr_type_p): New function.
* langhooks.c (lhd_set_decl_assembler_name): Do not compute
TYPE_DECLs
* doc/invoke.texi (-flto-odr-type-merging): Document.
* tree.c (need_assembler_name_p): Compute ODR names when asked
for it.
* tree.h (DECL_ASSEMBLER_NAME): Update comment.
re PR target/63223 ([avr] Make jumptables work with -Wl,--section-start,.text=)
gcc/
PR target/63223
* config/avr/avr.md (*tablejump.3byte-pc): New insn.
(*tablejump): Restrict to !AVR_HAVE_EIJMP_EICALL. Add void clobber.
(casesi): Expand to *tablejump.3byte-pc if AVR_HAVE_EIJMP_EICALL.
libgcc/
PR target/63223
* config/avr/libgcc.S (__tablejump2__): Rewrite to use RAMPZ, ELPM
and R24 as needed. Make work for all devices and .text locations.
(__do_global_ctors, __do_global_dtors): Use word addresses.
(__tablejump__, __tablejump_elpm__): Remove functions.
* t-avr (LIB1ASMFUNCS): Remove _tablejump, _tablejump_elpm.
Add _tablejump2.
(XICALL, XIJMP): New macros.
Michael Meissner [Wed, 10 Sep 2014 21:13:37 +0000 (21:13 +0000)]
vsx.md (vsx_fmav4sf4): Use correct constraints for V2DF, V4SF, DF, and DI modes.
2014-09-10 Michael Meissner <meissner@linux.vnet.ibm.com>
* config/rs6000/vsx.md (vsx_fmav4sf4): Use correct constraints for
V2DF, V4SF, DF, and DI modes.
(vsx_fmav2df2): Likewise.
(vsx_float_fix_<mode>2): Likewise.
(vsx_reduc_<VEC_reduc_name>_v2df_scalar): Likewise.
David Malcolm [Wed, 10 Sep 2014 14:35:16 +0000 (14:35 +0000)]
insn_extract takes an rtx_insn
gcc/ChangeLog:
2014-09-10 David Malcolm <dmalcolm@redhat.com>
* genextract.c (print_header): When writing out insn_extract to
insn-extract.c, strengthen the param "insn" from rtx to rtx_insn *.
* recog.h (insn_extract): Strengthen the param from rtx to
rtx_insn *.
Martin Jambor [Wed, 10 Sep 2014 11:36:29 +0000 (13:36 +0200)]
cgraphunit.c (expand_thunk): If not expanding, set analyzed flag.
2014-09-10 Martin Jambor <mjambor@suse.cz>
* cgraphunit.c (expand_thunk): If not expanding, set analyzed flag.
(analyze): Do not set analyze flag if expand_thunk returns false;.
(create_wrapper): Likewise.
* cgraphclones.c (duplicate_thunk_for_node): Likewise.
Jakub Jelinek [Wed, 10 Sep 2014 09:23:16 +0000 (11:23 +0200)]
flag-types.h (enum sanitize_code): Add SANITIZE_NONNULL_ATTRIBUTE and SANITIZE_RETURNS_NONNULL_ATTRIBUTE...
gcc/
* flag-types.h (enum sanitize_code): Add SANITIZE_NONNULL_ATTRIBUTE
and SANITIZE_RETURNS_NONNULL_ATTRIBUTE, or them into SANITIZE_UNDEFINED.
* opts.c (common_handle_option): Handle SANITIZE_NONNULL_ATTRIBUTE and
SANITIZE_RETURNS_NONNULL_ATTRIBUTE and disable
flag_delete_null_pointer_checks for them.
* sanitizer.def (BUILT_IN_UBSAN_HANDLE_NONNULL_ARG,
BUILT_IN_UBSAN_HANDLE_NONNULL_ARG_ABORT,
BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): New.
* ubsan.c (instrument_bool_enum_load): Set *gsi back to
stmt's iterator.
(instrument_nonnull_arg, instrument_nonnull_return): New functions.
(pass_ubsan::gate): Return true even for SANITIZE_NONNULL_ATTRIBUTE
or SANITIZE_RETURNS_NONNULL_ATTRIBUTE.
(pass_ubsan::execute): Call instrument_nonnull_{arg,return}.
* doc/invoke.texi (-fsanitize=nonnull-attribute,
-fsanitize=returns-nonnull-attribute): Document.
gcc/testsuite/
* c-c++-common/ubsan/attrib-3.c: New test.
* c-c++-common/ubsan/nonnull-1.c: New test.
* c-c++-common/ubsan/nonnull-2.c: New test.
* c-c++-common/ubsan/nonnull-3.c: New test.
* c-c++-common/ubsan/nonnull-4.c: New test.
* c-c++-common/ubsan/nonnull-5.c: New test.
libsanitizer/
* ubsan/ubsan_handlers.cc, ubsan/ubsan_handlers.h: Cherry pick
upstream r215485, r217389, r217391 and r217400.
Jakub Jelinek [Wed, 10 Sep 2014 09:21:25 +0000 (11:21 +0200)]
ubsan.h (struct ubsan_mismatch_data): Removed.
* ubsan.h (struct ubsan_mismatch_data): Removed.
(ubsan_create_data): Remove MISMATCH argument, add LOCCNT argument.
* ubsan.c (ubsan_source_location): For unknown locations,
pass { NULL, 0, 0 } instead of { "<unknown>", x, y }.
(ubsan_create_data): Remove MISMATCH argument, add LOCCNT argument.
Allow more than one location and arbitrary extra arguments passed
in ... instead of through MISMATCH pointer.
(ubsan_instrument_unreachable, ubsan_expand_bounds_ifn,
ubsan_expand_null_ifn, ubsan_build_overflow_builtin,
instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
callers.
c-family/
* c-ubsan.c (ubsan_instrument_division, ubsan_instrument_shift,
ubsan_instrument_vla, ubsan_instrument_return): Adjust
ubsan_create_data callers.
(ubsan_instrument_bounds): Don't emit UBSAN_BOUNDS at all if
index is constant or BIT_AND_EXPR with constant mask and is
small enough for the bound.
* c-gimplify.c (ubsan_walk_array_refs_r): For ADDR_EXPR of
ARRAY_REF, make sure the inner ARRAY_REF is not walked again.
gcc/
* config/i386/sse.md (define_mode_iterator VF_AVX512VL): New.
(define_mode_iterator FMAMODEM): Allow 128/256bit EVEX version.
(define_mode_iterator FMAMODE_AVX512): New.
(define_mode_iterator FMAMODE): Remove conditions.
(define_expand "fma4i_fmadd_<mode>"): Use FMAMODE_AVX512 mode iterator.
(define_expand "<avx512>_fmadd_<mode>_maskz<round_expand_name>"): Rename
from "<avx512>_fmadd_<mode>_maskz<round_expand_name>" and use VF_AVX512VL
mode iterator.
(define_mode_iterator FMAMODE_NOVF512): Remove.
(define_insn "*fma_fmadd_<mode>"): Rename from
"<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>" and use
FMAMODE mode iterator.
(define_mode_iterator VF_SF_AVX512VL): New.
(define_insn "<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name><round_name>"):
Use VF_SF_AVX512VL mode iterator.
(define_insn "<avx512>_fmadd_<mode>_mask<round_name>"): Rename from
"avx512f_fmadd_<mode>_mask<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "<avx512>_fmadd_<mode>_mask3<round_name>"): Rename from
"avx512f_fmadd_<mode>_mask3<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "*fma_fmsub_<mode>"): Rename from
"<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>" and use
FMAMODE mode iterator.
(define_insn "<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name><round_name>"):
Use VF_SF_AVX512VL mode iterator.
(define_insn "<avx512>_fmsub_<mode>_mask<round_name>"): Rename from
"avx512f_fmsub_<mode>_mask<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "<avx512>_fmsub_<mode>_mask3<round_name>"): Rename from
"avx512f_fmsub_<mode>_mask3<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "*fma_fnmadd_<mode>"): Rename from
"<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>" and
use FMAMODE mode iterator.
(define_insn "<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name><round_name>"):
Use VF_SF_AVX512VL mode iterator.
(define_insn "<avx512>_fnmadd_<mode>_mask<round_name>"): Rename from
"avx512f_fnmadd_<mode>_mask<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "<avx512>_fnmadd_<mode>_mask3<round_name>"): Rename from
"avx512f_fnmadd_<mode>_mask3<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "*fma_fnmsub_<mode>"): Rename from
"<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>" and use
FMAMODE mode iterator.
(define_insn "<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name><round_name>"):
Use VF_SF_AVX512VL mode iterator.
(define_insn "<avx512>_fnmsub_<mode>_mask<round_name>"): Rename from
"avx512f_fnmsub_<mode>_mask<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "<avx512>_fnmsub_<mode>_mask3<round_name>"): Rename from
"avx512f_fnmsub_<mode>_mask3<round_name>" and use VF_AVX512VL mode
iterator.
(define_expand "<avx512>_fmaddsub_<mode>_maskz<round_expand_name>"):
Rename from "avx512f_fmaddsub_<mode>_maskz<round_expand_name>" and
use VF_AVX512VL mode iterator.
(define_insn "*fma_fmaddsub_<mode>"): Rename from
"<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>" and
remove subst usage.
(define_insn "<sd_mask_codefor>fma_fmaddsub_<mode><sd_maskz_name><round_name>"):
Use VF_SF_AVX512VL mode iterator.
(define_insn "<avx512>_fmaddsub_<mode>_mask<round_name>"): Rename from
"avx512f_fmaddsub_<mode>_mask<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "<avx512>_fmaddsub_<mode>_mask3<round_name>"): Rename from
"avx512f_fmaddsub_<mode>_mask3<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "*fma_fmsubadd_<mode>"): Rename from
"<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>" and
remove usage of subst.
(define_insn "<sd_mask_codefor>fma_fmsubadd_<mode><sd_maskz_name><round_name>"):
Use VF_SF_AVX512VL mode iterator.
(define_insn "<avx512>_fmsubadd_<mode>_mask<round_name>"): Rename from
"avx512f_fmsubadd_<mode>_mask<round_name>" and use VF_AVX512VL mode
iterator.
(define_insn "<avx512>_fmsubadd_<mode>_mask3<round_name>"): Rename from
"avx512f_fmsubadd_<mode>_mask3<round_name>" and use VF_AVX512VL mode
iterator.
Tony Wang [Wed, 10 Sep 2014 04:45:32 +0000 (04:45 +0000)]
re PR libgcc/56846 (_Unwind_Backtrace on ARM and noexcept)
2014-09-10 Tony Wang <tony.wang@arm.com>
libstdc++-v3/
PR target/56846
* libsupc++/eh_personality.cc (PERSONALITY_FUNCTION):
Return with CONTINUE_UNWINDING when the state pattern
contains: _US_VIRTUAL_UNWIND_FRAME | _US_FORCE_UNWIND
Revert r213751:
* calls.c (precompute_arguments): Check
promoted_for_signed_and_unsigned_p and set the promoted mode.
(promoted_for_signed_and_unsigned_p): New function.
(expand_expr_real_1): Check promoted_for_signed_and_unsigned_p
and set the promoted mode.
* expr.h (promoted_for_signed_and_unsigned_p): New function definition.
* cfgexpand.c (expand_gimple_stmt_1): Call emit_move_insn if
SUBREG is promoted with SRP_SIGNED_AND_UNSIGNED.
opth-gen.awk: Generate mapping from cpp message reasons to the options that enable them.
gcc/ChangeLog:
2014-09-09 Manuel López-Ibáñez <manu@gcc.gnu.org>
* opth-gen.awk: Generate mapping from cpp message reasons to the
options that enable them.
* doc/options.texi (CppReason): Document.
gcc/c-family/ChangeLog:
2014-09-09 Manuel López-Ibáñez <manu@gcc.gnu.org>
* c.opt: Add CppReason to various flags.
(Wdate-time): Re-sort.
* c-common.c: Include c-common.h earlier.
(struct reason_option_codes_t): Delete.
(c_option_controlling_cpp_error): Prefix global type and struct
with cpp_.
PR target/63195
* config/rs6000/rs6000.md (*bool<mode>3): Allow only register
operands. Split off the constant operand alternative to ...
(*bool<mode>3_imm): New.
David Malcolm [Tue, 9 Sep 2014 17:02:34 +0000 (17:02 +0000)]
single_set takes an insn
gcc/ChangeLog:
2014-09-09 David Malcolm <dmalcolm@redhat.com>
* rtl.h (single_set_2): Strengthen first param from const_rtx to
const rtx_insn *, and move prototype to above...
(single_set): ...this. Convert this from a macro to an inline
function, enforcing the requirement that the param is a const
rtx_insn *.
(find_args_size_adjust): Strengthen param from rtx to rtx_insn *.
* config/arm/aarch-common-protos.h (aarch_crypto_can_dual_issue):
Strengthen both params from rtx to rtx_insn *.
* config/arm/aarch-common.c (aarch_crypto_can_dual_issue):
Likewise; introduce locals "producer_set", "consumer_set", using
them in place of "producer" and "consumer" when dealing with SET
rather than insn.
* config/avr/avr.c (avr_out_plus): Add checked cast to rtx_insn *
when invoking single_set in region guarded by INSN_P.
(avr_out_bitop): Likewise.
(_reg_unused_after): Introduce local rtx_sequence * "seq" in
region guarded by GET_CODE check, using methods to strengthen
local "this_insn" from rtx to rtx_insn *, and for clarity.
* config/avr/avr.md (define_insn_and_split "xload8<mode>_A"):
Strengthen local "insn" from rtx to rtx_insn *.
(define_insn_and_split "xload<mode>_A"): Likewise.
* config/bfin/bfin.c (trapping_loads_p): Likewise for param
"insn".
(find_load): Likewise for return type.
(workaround_speculation): Likewise for both locals named
"load_insn".
* config/cris/cris.c (cris_cc0_user_requires_cmp): Likewise for
local "cc0_user".
* config/cris/cris.md (define_peephole2 ; moversideqi): Likewise
for local "prev".
* config/h8300/h8300-protos.h (notice_update_cc): Likewise for
param 2.
* config/h8300/h8300.c (notice_update_cc): Likewise.
* config/i386/i386.c (ix86_flags_dependent): Likewise for params
"insn" and "dep_insn".
(exact_store_load_dependency): Likewise for both params.
(ix86_macro_fusion_pair_p): Eliminate local named "single_set"
since this now clashes with inline function. Instead, delay
calling single_set until the point where its needed, and then
assign the result to "compare_set" and rework the conditional that
follows.
* config/ia64/ia64.md (define_expand "tablejump"): Strengthen
local "last" from rtx to rtx_insn *.
* config/mips/mips-protos.h (mips_load_store_insns): Likewise for
second param.
(mips_store_data_bypass_p): Likewise for both params.
* config/mips/mips.c (mips_load_store_insns): Likewise for second
param.
(mips_store_data_bypass_p): Likewise for both params.
(mips_orphaned_high_part_p): Likewise for param "insn".
* config/mn10300/mn10300.c (extract_bundle): Likewise.
(mn10300_bundle_liw): Likewise for locals "r", "insn1", "insn2".
Introduce local rtx "insn2_pat".
* config/rl78/rl78.c (move_elim_pass): Likewise for locals "insn",
"ninsn".
(rl78_remove_unused_sets): Likewise for locals "insn", "ninsn".
Introduce local rtx "set", using it in place of "insn" for the
result of single_set. This appears to fix a bug, since the call
to find_regno_note on a SET does nothing.
* config/rs6000/rs6000.c (set_to_load_agen): Strengthen both
params from rtx to rtx_insn *.
(set_to_load_agen): Likewise.
* config/s390/s390.c (s390_label_align): Likewise for local
"prev_insn". Introduce new rtx locals "set" and "src", using
them in place of "prev_insn" for the results of single_set
and SET_SRC respectively.
(s390_swap_cmp): Strengthen local "jump" from rtx to rtx_insn *.
Introduce new rtx local "set" using in place of "jump" for the
result of single_set. Use SET_SRC (set) rather than plain
XEXP (set, 1).
* config/sh/sh.c (noncall_uses_reg): Strengthen param 2from
rtx to rtx_insn *.
(noncall_uses_reg): Likewise.
(reg_unused_after): Introduce local rtx_sequence * "seq" in region
guarded by GET_CODE check, using its methods for clarity, and to
enable strengthening local "this_insn" from rtx to rtx_insn *.
* config/sh/sh.md (define_expand "mulhisi3"): Strengthen local
"insn" from rtx to rtx_insn *.
(define_expand "umulhisi3"): Likewise.
(define_expand "smulsi3_highpart"): Likewise.
(define_expand "umulsi3_highpart"): Likewise.
* config/sparc/sparc.c (sparc_do_work_around_errata): Likewise for
local "after". Replace GET_CODE check with a dyn_cast,
introducing new local rtx_sequence * "seq", using insn method for
typesafety.
* dwarf2cfi.c (dwarf2out_frame_debug): Strengthen param "insn"
from rtx to rtx_insn *. Introduce local rtx "pat", using it in
place of "insn" once we're dealing with patterns rather than the
input insn.
(scan_insn_after): Strengthen param "insn" from rtx to rtx_insn *.
(scan_trace): Likewise for local "elt", updating lookups within
sequence to use insn method rather than element method.
* expr.c (find_args_size_adjust): Strengthen param "insn" from rtx
to rtx_insn *.
* gcse.c (gcse_emit_move_after): Likewise for local "new_rtx".
* ifcvt.c (noce_try_abs): Likewise for local "insn".
* ira.c (fix_reg_equiv_init): Add checked cast to rtx_insn * when
invoking single_set.
* lra-constraints.c (insn_rhs_dead_pseudo_p): Strengthen param
"insn" from rtx to rtx_insn *.
(skip_usage_debug_insns): Likewise for return type, adding a
checked cast.
(check_secondary_memory_needed_p): Likewise for local "insn".
(inherit_reload_reg): Likewise.
* modulo-sched.c (sms_schedule): Likewise for local "count_init".
* recog.c (peep2_attempt): Likewise for local "old_insn", adding
checked casts.
(store_data_bypass_p): Likewise for both params.
(if_test_bypass_p): Likewise.
* recog.h (store_data_bypass_p): Likewise for both params.
(if_test_bypass_p): Likewise.
* reload.c (find_equiv_reg): Likewise for local "where".
* reorg.c (delete_jump): Likewise for param "insn".
* rtlanal.c (single_set_2): Strenghen param "insn" from const_rtx
to const rtx_insn *.
* store-motion.c (replace_store_insn): Likewise for param "del".
(delete_store): Strengthen local "i" from rtx to rtx_insn_list *,
and use its methods for clarity, and to strengthen local "del"
from rtx to rtx_insn *.
(build_store_vectors): Use insn method of "st" when calling
replace_store_insn for typesafety and clarity.
David Malcolm [Tue, 9 Sep 2014 16:34:56 +0000 (16:34 +0000)]
recog_memoized works on an rtx_insn *
gcc/ChangeLog:
2014-09-09 David Malcolm <dmalcolm@redhat.com>
* caller-save.c (rtx saveinsn): Strengthen this variable from rtx
to rtx_insn *.
(restinsn): Likewise.
* config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_move):
Likewise for param.
* config/aarch64/aarch64.c (aarch64_simd_attr_length_move):
Likewise.
* config/arc/arc-protos.h (arc_adjust_insn_length): Likewise for
first param.
(arc_hazard): Likewise for both params.
* config/arc/arc.c (arc600_corereg_hazard): Likewise, adding
checked casts to rtx_sequence * and uses of the insn method for
type-safety.
(arc_hazard): Strengthen both params from rtx to rtx_insn *.
(arc_adjust_insn_length): Likewise for param "insn".
(struct insn_length_parameters_s): Likewise for first param of
"get_variants" callback field.
(arc_get_insn_variants): Likewise for first param and local
"inner". Replace a check of GET_CODE with a dyn_cast to
rtx_sequence *, using methods for type-safety and clarity.
* config/arc/arc.h (ADJUST_INSN_LENGTH): Use casts to
rtx_sequence * and uses of the insn method for type-safety when
invoking arc_adjust_insn_length.
* config/arm/arm-protos.h (arm_attr_length_move_neon): Likewise
for param.
(arm_address_offset_is_imm): Likewise.
(struct tune_params): Likewise for params 1 and 3 of the
"sched_adjust_cost" callback field.
* config/arm/arm.c (cortex_a9_sched_adjust_cost): Likewise for
params 1 and 3 ("insn" and "dep").
(xscale_sched_adjust_cost): Likewise.
(fa726te_sched_adjust_cost): Likewise.
(cortexa7_older_only): Likewise for param "insn".
(cortexa7_younger): Likewise.
(arm_attr_length_move_neon): Likewise.
(arm_address_offset_is_imm): Likewise.
* config/avr/avr-protos.h (avr_notice_update_cc): Likewise.
* config/avr/avr.c (avr_notice_update_cc): Likewise.
* config/bfin/bfin.c (hwloop_pattern_reg): Likewise.
(workaround_speculation): Likewise for local "last_condjump".
* config/c6x/c6x.c (shadow_p): Likewise for param "insn".
(shadow_or_blockage_p): Likewise.
(get_unit_reqs): Likewise.
(get_unit_operand_masks): Likewise.
(c6x_registers_update): Likewise.
(returning_call_p): Likewise.
(can_use_callp): Likewise.
(convert_to_callp): Likewise.
(find_last_same_clock): Likwise for local "t".
(reorg_split_calls): Likewise for local "shadow".
(hwloop_pattern_reg): Likewise for param "insn".
* config/frv/frv-protos.h (frv_final_prescan_insn): Likewise.
* config/frv/frv.c (frv_final_prescan_insn): Likewise.
(frv_extract_membar): Likewise.
(frv_optimize_membar_local): Strengthen param "last_membar" from
rtx * to rtx_insn **.
(frv_optimize_membar_global): Strengthen param "membar" from rtx
to rtx_insn *.
(frv_optimize_membar): Strengthen local "last_membar" from rtx *
to rtx_insn **.
* config/ia64/ia64-protos.h (ia64_st_address_bypass_p): Strengthen
both params from rtx to rtx_insn *.
(ia64_ld_address_bypass_p): Likewise.
* config/ia64/ia64.c (ia64_safe_itanium_class): Likewise for param
"insn".
(ia64_safe_type): Likewise.
(group_barrier_needed): Likewise.
(safe_group_barrier_needed): Likewise.
(ia64_single_set): Likewise.
(is_load_p): Likewise.
(record_memory_reference): Likewise.
(get_mode_no_for_insn): Likewise.
(important_for_bundling_p): Likewise.
(unknown_for_bundling_p): Likewise.
(ia64_st_address_bypass_p): Likewise for both params.
(ia64_ld_address_bypass_p): Likewise.
(expand_vselect): Introduce new local rtx_insn * "insn", using it
in place of rtx "x" after the emit_insn call.
* config/i386/i386-protos.h (x86_extended_QIreg_mentioned_p):
Strengthen param from rtx to rtx_insn *.
(ix86_agi_dependent): Likewise for both params.
(ix86_attr_length_immediate_default): Likewise for param 1.
(ix86_attr_length_address_default): Likewise for param.
(ix86_attr_length_vex_default): Likewise for param 1.
* config/i386/i386.c (ix86_attr_length_immediate_default):
Likewise for param "insn".
(ix86_attr_length_address_default): Likewise.
(ix86_attr_length_vex_default): Likewise.
(ix86_agi_dependent): Likewise for both params.
(x86_extended_QIreg_mentioned_p): Likewise for param "insn".
(vselect_insn): Likewise for this variable.
* config/m68k/m68k-protos.h (m68k_sched_attr_opx_type): Likewise
for param 1.
(m68k_sched_attr_opy_type): Likewise.
* config/m68k/m68k.c (sched_get_operand): Likewise.
(sched_attr_op_type): Likewise.
(m68k_sched_attr_opx_type): Likewise.
(m68k_sched_attr_opy_type): Likewise.
(sched_get_reg_operand): Likewise.
(sched_get_mem_operand): Likewise.
(m68k_sched_address_bypass_p): Likewise for both params.
(sched_get_indexed_address_scale): Likewise.
(m68k_sched_indexed_address_bypass_p): Likewise.
* config/m68k/m68k.h (m68k_sched_address_bypass_p): Likewise.
(m68k_sched_indexed_address_bypass_p): Likewise.
* config/mep/mep.c (mep_jmp_return_reorg): Strengthen locals
"label", "ret" from rtx to rtx_insn *, adding a checked cast and
removing another.
* config/mips/mips-protos.h (mips_linked_madd_p): Strengthen both
params from rtx to rtx_insn *.
(mips_fmadd_bypass): Likewise.
* config/mips/mips.c (mips_fmadd_bypass): Likewise.
(mips_linked_madd_p): Likewise.
(mips_macc_chains_last_hilo): Likewise for this variable.
(mips_macc_chains_record): Likewise for param.
(vr4130_last_insn): Likewise for this variable.
(vr4130_swap_insns_p): Likewise for both params.
(mips_ls2_variable_issue): Likewise for param.
(mips_need_noat_wrapper_p): Likewise for param "insn".
(mips_expand_vselect): Add a new local rtx_insn * "insn", using it
in place of "x" after the emit_insn.
* config/pa/pa-protos.h (pa_fpstore_bypass_p): Strengthen both
params from rtx to rtx_insn *.
* config/pa/pa.c (pa_fpstore_bypass_p): Likewise.
(pa_combine_instructions): Introduce local "par" for result of
gen_rtx_PARALLEL, moving decl and usage of new_rtx for after call
to make_insn_raw.
(pa_can_combine_p): Strengthen param "new_rtx" from rtx to rtx_insn *.
* config/rl78/rl78.c (insn_ok_now): Likewise for param "insn".
(rl78_alloc_physical_registers_op1): Likewise.
(rl78_alloc_physical_registers_op2): Likewise.
(rl78_alloc_physical_registers_ro1): Likewise.
(rl78_alloc_physical_registers_cmp): Likewise.
(rl78_alloc_physical_registers_umul): Likewise.
(rl78_alloc_address_registers_macax): Likewise.
(rl78_alloc_physical_registers): Likewise for locals "insn", "curr".
* config/s390/predicates.md (execute_operation): Likewise for
local "insn".
* config/s390/s390-protos.h (s390_agen_dep_p): Likewise for both
params.
* config/s390/s390.c (s390_safe_attr_type): Likewise for param.
(addr_generation_dependency_p): Likewise for param "insn".
(s390_agen_dep_p): Likewise for both params.
(s390_fpload_toreg): Likewise for param "insn".
* config/sh/sh-protos.h (sh_loop_align): Likewise for param.
* config/sh/sh.c (sh_loop_align): Likewise for param and local
"next".
* config/sh/sh.md (define_peephole2): Likewise for local "insn2".
* config/sh/sh_treg_combine.cc
(sh_treg_combine::make_inv_ccreg_insn): Likewise for return type
and local "i".
(sh_treg_combine::try_eliminate_cstores): Likewise for local "i".
* config/stormy16/stormy16.c (combine_bnp): Likewise for locals
"and_insn", "load", "shift".
* config/tilegx/tilegx.c (match_pcrel_step2): Likewise for param
"insn".
* final.c (final_scan_insn): Introduce local rtx_insn * "other"
for XEXP (note, 0) of the REG_CC_SETTER note.
(cleanup_subreg_operands): Strengthen param "insn" from rtx to
rtx_insn *, eliminating a checked cast made redundant by this.
* gcse.c (process_insert_insn): Strengthen local "insn" from rtx
to rtx_insn *.
* genattr.c (main): When writing out the prototype to
const_num_delay_slots, strengthen the param from rtx to
rtx_insn *.
* genattrtab.c (write_const_num_delay_slots): Likewise when
writing out the implementation of const_num_delay_slots.
* hw-doloop.h (struct hw_doloop_hooks): Strengthen the param
"insn" of callback field "end_pattern_reg" from rtx to rtx_insn *.
* ifcvt.c (noce_emit_store_flag): Eliminate local rtx "tmp" in
favor of new rtx locals "src" and "set" and new local rtx_insn *
"insn" and "seq".
(noce_emit_move_insn): Strengthen locals "seq" and "insn" from rtx
to rtx_insn *.
(noce_emit_cmove): Eliminate local rtx "tmp" in favor of new rtx
locals "cond", "if_then_else", "set" and new rtx_insn * locals
"insn" and "seq".
(noce_try_cmove_arith): Strengthen locals "insn_a" and "insn_b",
"last" from rtx to rtx_insn *. Likewise for a local "tmp",
renaming to "tmp_insn". Eliminate the other local rtx "tmp" from
the top-level scope, replacing with new more tightly-scoped rtx
locals "reg", "pat", "mem" and rtx_insn * "insn", "copy_of_a",
"new_insn", "copy_of_insn_b", and make local rtx "set" more
tightly-scoped.
* ira-int.h (ira_setup_alts): Strengthen param "insn" from rtx to
rtx_insn *.
* ira.c (setup_prohibited_mode_move_regs): Likewise for local
"move_insn".
(ira_setup_alts): Likewise for param "insn".
* lra-constraints.c (emit_inc): Likewise for local "add_insn".
* lra.c (emit_add3_insn): Split local rtx "insn" in two, an rtx
and an rtx_insn *.
(lra_emit_add): Eliminate top-level local rtx "insn" in favor of
new more-tightly scoped rtx locals "add3_insn", "insn",
"add2_insn" and rtx_insn * "move_insn".
* postreload-gcse.c (eliminate_partially_redundant_load): Add
checked cast on result of gen_move_insn when invoking
extract_insn.
* recog.c (insn_invalid_p): Strengthen param "insn" from rtx to
rtx_insn *.
(verify_changes): Add a checked cast on "object" when invoking
insn_invalid_p.
(extract_insn_cached): Strengthen param "insn" from rtx to
rtx_insn *.
(extract_constrain_insn_cached): Likewise.
(extract_insn): Likewise.
* recog.h (insn_invalid_p): Likewise for param 1.
(recog_memoized): Likewise for param.
(extract_insn): Likewise.
(extract_constrain_insn_cached): Likewise.
(extract_insn_cached): Likewise.
* reload.c (can_reload_into): Likewise for local "test_insn".
* reload.h (cleanup_subreg_operands): Likewise for param.
* reload1.c (emit_insn_if_valid_for_reload): Rename param from
"insn" to "pat", reintroducing "insn" as an rtx_insn * on the
result of emit_insn. Remove a checked cast made redundant by this
change.
* sel-sched-ir.c (sel_insn_rtx_cost): Strengthen param "insn" from
rtx to rtx_insn *.
* sel-sched.c (get_reg_class): Likewise.
David Malcolm [Tue, 9 Sep 2014 15:20:07 +0000 (15:20 +0000)]
INSN_LOCATION takes an rtx_insn
gcc/ChangeLog:
2014-09-09 David Malcolm <dmalcolm@redhat.com>
* rtl.h (INSN_LOCATION): Strengthen param from const_rtx to
const rtx_insn *, and from rtx to rtx_insn * for the other
overloaded variant.
(RTL_LOCATION): Add a checked cast to rtx_insn * when invoking
INSN_LOCATION, since we know INSN_P holds.
(insn_line): Strengthen param from const_rtx to const rtx_insn *.
(insn_file): Likewise.
(insn_scope): Likewise.
(insn_location): Likewise.
* config/mips/mips.c (mips16_gp_pseudo_reg): Strengthen local
"insn" from rtx to rtx_insn *, introducing a new local rtx "set"
for the result of gen_load_const_gp.
* config/rs6000/rs6000-protos.h (output_call): Strengthen first
param from rtx to rtx_insn *.
* config/rs6000/rs6000.c (output_call): Likewise.
* dwarf2out.c (dwarf2out_var_location): Likewise for local "prev",
introducing a checked cast to rtx_sequence * and use of the insn
method.
* emit-rtl.c (emit_copy_of_insn_after): Strengthen both params
from rtx to rtx_insn *.
(insn_scope): Strengthen param from const_rtx to const rtx_insn *.
(insn_line): Likewise.
(insn_file): Likewise.
(insn_location): Likewise.
* emit-rtl.h (emit_copy_of_insn_after): Strengthen both params
from rtx to rtx_insn *.
* print-rtl.c (print_rtx): Introduce local "in_insn" via a checked
cast, using it for calls to INSN_HAS_LOCATION and insn_location.
* recog.c (peep2_attempt): Introduce local rtx_insn * "peepinsn"
via a checked cast.
* reorg.c (relax_delay_slots): Strengthen locals named "after"
from rtx to rtx_insn *; use methods of "pat" for type-safety.
David Malcolm [Tue, 9 Sep 2014 15:14:03 +0000 (15:14 +0000)]
Handcode gen_rtx_INSN
gcc/ChangeLog:
2014-09-09 David Malcolm <dmalcolm@redhat.com>
* combine.c (try_combine): Eliminate checked cast on result of
gen_rtx_INSN.
* emit-rtl.c (gen_rtx_INSN): New function, improving over the prior
autogenerated one by strengthening the return type and params 2 and 3
from rtx to rtx_insn *, and by naming the params.
* gengenrtl.c (special_rtx): Add INSN to those that are
special-cased.
* rtl.h (gen_rtx_INSN): New prototype.
David Malcolm [Tue, 9 Sep 2014 15:10:39 +0000 (15:10 +0000)]
struct ira_reg_equiv_s's "init_insns" is an rtx_insn_list
gcc/ChangeLog:
2014-09-09 David Malcolm <dmalcolm@redhat.com>
* ira.c (ira_update_equiv_info_by_shuffle_insn): Use NULL rather
than NULL_RTX.
(no_equiv): Likewise.
(update_equiv_regs): Likewise.
(setup_reg_equiv): Likewise. Strengthen locals "elem",
"prev_elem", "next_elem" from rtx to rtx_insn_list *, and "insn"
from rtx to rtx_insn *. Use methods of "elem" for typesafety and
clarity.
* ira.h (struct ira_reg_equiv_s): Strengthen field "init_insns"
from rtx to rtx_insn_list *.
* lra-assigns.c (spill_for): Strengthen local "x" from rtx to
rtx_insn_list * and use methods for clarity and typesafety.
* lra-constraints.c (contains_deleted_insn_p): Likewise for param
"list".
(init_insn_rhs_dead_pseudo_p): Likewise for local "insns". Remove
redundant check on INSN_P (insns): this cannot hold, as "insns" is
an INSN_LIST, not an insn.
(reverse_equiv_p): Strengthen local "insns" from rtx to
rtx_insn_list * and use methods for clarity and typesafety.
(contains_reloaded_insn_p): Likewise for local "list".
Jiong Wang [Tue, 9 Sep 2014 12:29:36 +0000 (12:29 +0000)]
[ARM] Enable auto-vectorization for copysignf
gcc/
* config/arm/arm.c (NEON_COPYSIGNF): New enum.
(arm_init_neon_builtins): Support NEON_COPYSIGNF.
(arm_builtin_vectorized_function): Likewise.
* config/arm/arm_neon_builtins.def: New macro for copysignf.
* config/arm/neon.md (neon_copysignf<mode>): New pattern for vector
copysignf.
gcc/testsuite/
* gcc.target/arm/vect-copysignf.c: New testcase.
gcc/
* bb-reorder.h (default_target_bb_reorder): Remove redundant GTY.
* builtins.h (default_target_builtins): Likewise.
* gcse.h (default_target_gcse): Likewise.
* target-globals.h (target_globals): Add a destructor. Convert
void-pointer fields back to their real type and change from
GTY((atomic)) to GTY((skip)).
(restore_target_globals): Remove casts accordingly.
* target-globals.c (save_target_globals): Use XCNEW rather than
ggc_internal_cleared_alloc to allocate non-GC structures.
Use ggc_cleared_alloc to allocate the target_globals structure
itself.
(target_globals::~target_globals): Define.
Nick Clifton [Tue, 9 Sep 2014 08:27:39 +0000 (08:27 +0000)]
invoke.texi (Optimization Options): Add missing @gol to the end of a line.
* doc/invoke.texi (Optimization Options): Add missing @gol to the
end of a line.
(S/390 and zSeries Options): Remove superfluous word from the
description of the -mhotpatch option.
David Malcolm [Mon, 8 Sep 2014 20:45:32 +0000 (20:45 +0000)]
gcse.c's modify_mem_list is a list of insns
gcc/ChangeLog:
2014-09-08 David Malcolm <dmalcolm@redhat.com>
* gcse.c (modify_mem_list): Strengthen this variable from
vec<rtx> * to vec<rtx_insn *> *.
(vec_rtx_heap): Strengthen this typedef from vec<rtx> to
vec<rtx_insn *>.
(load_killed_in_block_p): Strengthen local "list" from vec<rtx> to
vec<rtx_insn *>, and local "setter" from rtx to rtx_insn *.
(record_last_mem_set_info): Strengthen param "insn" from rtx to
rtx_insn *.
(record_last_set_info): Likewise for local "last_set_insn".