PR ipa/79224
* ipa-inline-analysis.c (get_minimal_bb): New function.
(record_modified): Use it.
(remap_edge_change_prob): Handle also ancestor functions.
From-SVN: r245357
PR middle-end/79454
* internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
result computation whenever lhs doesn't have vector mode, not
just when it has BLKmode.
From-SVN: r245354
* doc/install.texi (Specific): Use https for blackfin.uclinux.org.
(Specific): Update mingw-w64 reference.
(Binaries): Ditto.
(Specific): Remove broken link to Renesas RX processor.
From-SVN: r245332
2017-02-10 Richard Biener <rguenther@suse.de>
* toplev.c (process_options): Do not mention obsolete graphite
options when printing sorry message about missing graphite support.
Mention -floop-nest-optimize.
From-SVN: r245331
PR tree-optimization/79411
* tree-ssa-reassoc.c (is_reassociable_op): Return false if
stmt operands are SSA_NAMEs used in abnormal phis.
(can_reassociate_p): Return false if op is SSA_NAME used in abnormal
phis.
* gcc.c-torture/compile/pr79411.c: New test.
From-SVN: r245324
PR c/79413
* gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
not arbitrary TREE_CONSTANT.
* gcc.c-torture/compile/pr79413.c: New test.
From-SVN: r245304
2017-02-09 Richard Biener <rguenther@suse.de>
PR tree-optimization/69823
* graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
* gcc.dg/graphite/pr69823.c: New testcase.
From-SVN: r245295
The __NPS400__ define is currently created in CPP_SPEC unlike the other
target defines, which are created in arc-c.def. Further, the current
__NPS400__ define is (currently) only created when -mcpu=nps400 is
passed, which is fine, except that if GCC is configured using
--with-cpu=nps400 then the -mcpu option is not required and the
__NPS400__ define will not be created.
This commit moves the __NPS400__ define into arc-c.def inline with all
of the other target defines, and removes the code in CPP_SPEC that used
to create the define.
In order to support the creation of the define in arc-c.def, a new
TARGET_NPS400 macro is created in arc.h.
gcc/ChangeLog:
* config/arc/arc-c.def: Add __NPS400__ definition.
* config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
(TARGET_NPS400): Define.
From-SVN: r245294
Currently we only make the base_architecture globally available, this
means we can tell if we have selected arc700/archs/etc but it's not
possible to tell if the user has selected a specific cpu variant, for
example nps400.
One problem this causes is, for example, in arc-c.def, if we want to add
an __NPS400__ define then we need a flag we can check to determine if
this is the right thing to do.
In this commit the arc_selected_cpu variable (previously local within
arc.c) has been made global. Two other variables arc_base_cpu and
arc_selected_arch have been deleted, all of this information can be
found within (or through) arc_selected_cpu.
All uses of arc_base_cpu and arc_selected_arch have been updated. This
commit does not introduce any new defines (like __NPS400__), this is
just a restructuring commit.
The declaration of arc_selected_cpu has moved into arc-arch.h, in
contrast to the declaration of arc_base_cpu which was previously in
arc.h. This avoids a compilation issue when building libgcc, as the
structure and enums declared in arc-arch.h are not included for libgcc
then declaring an arc_selected_cpu (a struct type) in arc.h would result
in an unknown struct error. We got away with this for arc_base_cpu as
that was an enum type. The declaration of arc_selected_cpu in
arc.h could have been wrapped in a '#ifndef IN_LIBGCC2 ... #endif', but
it felt neater to simply move the declaration into arc-arch.h.
gcc/ChangeLog:
* config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
file.
(arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
pointer, arch_info.
(arc_cpu_types): Fill the arch_info field with a pointer into the
arc_arch_types table.
(arc_selected_cpu): Declare.
* config/arc/arc.c (arc_selected_cpu): Make global.
(arc_selected_arch): Delete.
(arc_base_cpu): Delete.
(arc_override_options): Remove references to deleted variables,
update access to arch information.
(ARC_OPT): Update access to arch information.
(ARC_OPTX): Likewise.
* config/arc/arc.h (arc_base_cpu): Remove declaration.
(TARGET_ARC600): Update access to arch information.
(TARGET_ARC601): Likewise.
(TARGET_ARC700): Likewise.
(TARGET_EM): Likewise.
(TARGET_HS): Likewise.
* config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
information.
From-SVN: r245293
It was spelled Altivec in two user-facing messages in rs6000.opt; let's
be consistent.
PR translation/79397
* config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
of AltiVec.
From-SVN: r245276
2017-02-08 Martin Jambor <mjambor@suse.cz>
PR ipa/79375
* ipa-prop.c (ipa_alloc_node_params): Make static, return bool
whether allocation happened.
(ipa_initialize_node_params): Do not call ipa_alloc_node_params if
nothing was allocated.
From-SVN: r245275
PR tree-optimization/79408
* tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
constant, but SSA_NAME with a known integer range, use the minimum
of that range instead of op1 to determine if modulo can be replaced
with its first operand.
* gcc.dg/tree-ssa/pr79408.c: New test.
From-SVN: r245273
2017-02-08 Richard Biener <rguenther@suse.de>
PR tree-optimization/71824
* graphite-scop-detection.c (scop_detection::build_scop_breadth):
Check all loops contained in the merged region.
* gcc.dg/graphite/pr71824-2.c: New testcase.
From-SVN: r245270
PR middle-end/79399
* ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
type from int to size_t.
* ira-costs.c (struct_costs_size): Change type from int to size_t.
Co-Authored-By: Richard Biener <rguenther@suse.de>
From-SVN: r245256
PR rtl-optimization/79386
* cprop.c (bypass_conditional_jumps): Initialize
bypass_last_basic_block already before splitting bbs after
unconditional traps...
(bypass_conditional_jumps): ... rather than here.
* gcc.c-torture/compile/pr79386.c: New test.
From-SVN: r245251
operands.
gcc/
* config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
builtins to SImode and emit a zero-extend, if necessary.
From-SVN: r245243
gcc/ChangeLog:
2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
* doc/invoke.texi (RISC-V Options): Use two spaces to separate
options.
From-SVN: r245230
[gcc]
2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
PR target/66144
* config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
false values to be constant vectors with all 0 or all 1 bits set.
(vcondu<mode><mode>): Likewise.
* config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
predicate.
(fpmask_comparison_operator): Update comment.
(vecint_comparison_operator): New predicate.
* config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
vector conditionals when the true and false values are constant
vectors with all 0 bits or all 1 bits set.
[gcc/testsuite]
2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
PR target/66144
* gcc.target/powerpc/pr66144-1.c: New test.
* gcc.target/powerpc/pr66144-2.c: Likewise.
* gcc.target/powerpc/pr66144-3.c: Likewise.
From-SVN: r245222
Scheduling should never move very expensive instructions to places they
are executed more frequently. This patch fixes that, reducing the
execution time of c-ray by over 40% (I tested on a BE Power7 system).
This introduces a new target hook sched.can_speculate_insn which returns
whether the scheduler is allowed to speculate a given instruction. The
rs6000 implementation disallows all divide and square root instructions.
PR rtl-optimization/68664
* target.def (can_speculate_insn): New hook.
* doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
* doc/tm.texi: Regenerate.
* sched-rgn.c (can_schedule_ready_p): Use the new hook.
* config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
(rs6000_sched_can_speculate_insn): New function.
From-SVN: r245215
PR tree-optimization/79284
* tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
* tree-vect-stmts.c (vect_get_vec_def_for_operand,
vectorizable_mask_load_store, vectorizable_operation,
vect_is_simple_cond, get_same_sized_vectype): Use it instead
of comparing TREE_CODE of a type against BOOLEAN_TYPE.
* tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
* tree-vect-slp.c (vect_get_constant_vectors): Likewise.
* tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
is_gimple_assign (stmt). Replace another such test with
is_gimple_assign (stmt).
testsuite/
* gcc.c-torture/compile/pr79284.c: New test.
From-SVN: r245214