This patch changes the element type of (auto_)vec_perm_indices from
unsigned char to unsigned short.  This is needed for fixed-length
2048-bit SVE.  (SVE is variable-length by default, but it's possible
to ask for specific vector lengths if you want to.)

2017-09-25  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
	* target.h (vec_perm_indices): Use unsigned short rather than
	unsigned char.
	(auto_vec_perm_indices): Likewise.
	* config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
	Use unsigned int rather than unsigned char.
	* config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.

From-SVN: r253150

2017-09-25  Richard Biener  <rguenther@suse.de>

	* cfgloop.h (sort_sibling_loops): Declare.
	* cfgloop.c (sort_sibling_loops_cmp): New helper.
	(sort_sibling_loops): New function sorting the sibling loop list
	in RPO order.
	* graphite.c (graphite_transform_loops): Sort sibling loops.

From-SVN: r253149

This patch makes TARGET_VECTORIZE_VEC_PERM_CONST_OK take the permute
vector in the form of a vec_perm_indices instead of an unsigned char *.
It follows on from the recent patch that did the same in target-independent
code.

It was easy to make ARM and AArch64 use vec_perm_indices internally
as well, and converting AArch64 helps with SVE.  I did try doing the same
for the other ports, but the surgery needed was much more invasive and
much less obviously correct.

2017-09-22  Richard Sandiford  <richard.sandifird@linaro.org>

gcc/
	* target.def (vec_perm_const_ok): Change sel parameter to
	vec_perm_indices.
	* optabs-query.c (can_vec_perm_p): Update accordingly.
	* doc/tm.texi: Regenerate.
	* config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
	auto_vec_perm_indices and remove separate nelt field.
	(aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
	(aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
	(aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
	(aarch64_expand_vec_perm_const): Update accordingly.
	(aarch64_vectorize_vec_perm_const_ok): Likewise.  Change sel
	to vec_perm_indices.
	* config/arm/arm.c (expand_vec_perm_d): Change perm to
	auto_vec_perm_indices and remove separate nelt field.
	(arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
	(arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
	(arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
	accordingly.
	(arm_vectorize_vec_perm_const_ok): Likewise.  Change sel
	to vec_perm_indices.
	* config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
	sel to vec_perm_indices.
	* config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
	* config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
	* config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
	Likewise.
	* config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
	Likewise.

From-SVN: r253148

This patch is an attempt to fix the crash reported in PR82155.

When generating a C++ class method for a class that is itself nested in
a class method, dwarf2out_early_global_decl currently leaves the
existing context DIE as it is if it already exists.  However, it is
possible that this call happens at a point where this context DIE is
just a declaration that is itself not located in its own context.

From there, if dwarf2out_early_global_decl is not called on any of the
FUNCTION_DECL in the context chain, DIEs will be left badly scoped and
some (such as the nested method) will be removed by the type pruning
machinery.  As a consequence, dwarf2out_abstract_function will will
crash when called on the corresponding DECL because it asserts that the
DECL has a DIE.

This patch fixes this crash making dwarf2out_early_global_decl process
context DIEs the same way we process abstract origins for FUNCTION_DECL:
if the corresponding DIE exists but is only a declaration, call
dwarf2out_decl anyway on it so that it is turned into a more complete
DIE and so that it is relocated in the proper context.

Bootstrapped and regtested on x86_64-linux.

gcc/

	PR debug/82155
	* dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
	on the FUNCTION_DECL function context if it has a DIE that is a
	declaration.

gcc/testsuite/

	* g++.dg/pr82155.C: New testcase.

From-SVN: r253147

2017-09-25  Richard Biener  <rguenther@suse.de>

	PR tree-optimization/82285
	* tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
	enumeral types.

	* gcc.dg/torture/pr82285.c: New testcase.

From-SVN: r253146

2017-09-25  Tom de Vries  <tom@codesourcery.com>

	PR target/80035
	PR target/81069
	* config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
	noreturn function.

From-SVN: r253145

2017-09-25  Justin Squirek  <squirek@adacore.com>

	* aspects.adb, bindgen.adb, clean.adb, erroutc.adb, exp_ch13.adb,
	exp_dbug.adb, exp_unst.adb, exp_util.adb, frontend.adb, gnat1drv.adb,
	gnatdll.adb, gnatlink.adb, gnatls.adb, gnatname.adb, gnatxref.adb,
	gnatfind.adb, libgnat/a-cfhama.ads, libgnat/a-exetim__mingw.adb,
	libgnat/a-strmap.adb, libgnat/a-teioed.adb, libgnat/g-alvety.ads,
	libgnat/g-expect.adb, libgnat/g-regist.adb, libgnat/g-socket.adb,
	libgnat/g-socthi__mingw.ads, libgnat/s-stausa.adb,
	libgnat/s-tsmona__linux.adb, libgnat/s-tsmona__mingw.adb,
	libgnarl/s-taenca.adb, libgnarl/s-tassta.adb, libgnarl/s-tarest.adb,
	libgnarl/s-tpobop.adb, make.adb, makeusg.adb, namet.adb, output.ads,
	put_scos.adb, repinfo.adb, rtsfind.adb, scn.ads, sem_attr.adb,
	sem_aux.ads, sem_warn.ads, targparm.adb, xr_tabls.adb, xref_lib.adb:
	Removal of ineffective use-clauses.
	* exp_ch9.adb (Is_Simple_Barrier_Name): Check for false positives with
	constant folded barriers.
	* ghost.adb, sprint.adb, sem_ch10.adb, sem_warn.adb: Change access to
	Subtype_Marks and Names list in use-clause nodes to their new singular
	counterparts (e.g. Subtype_Mark, Name).
	* par.adb, par-ch8.adb (Append_Use_Clause): Created to set
	Prev_Ids and More_Ids in use-clause nodes.
	(P_Use_Clause): Modify to take a list as a parameter.
	(P_Use_Package_Clause, P_Use_Type_Clause): Divide names and
	subtype_marks within an aggregate use-clauses into individual clauses.
	* par-ch3.adb, par-ch10.adb, par-ch12.adb: Trivally modify call to
	P_Use_Clause to match its new behavior.
	* sem.adb (Analyze): Mark use clauses for non-overloaded entities.
	* sem_ch4.adb (Try_One_Interp): Add sanity check to handle previous
	errors.
	* sem_ch6.adb (Analyze_Generic_Subprogram_Body,
	Analyze_Subprogram_Body_Helper): Update use clause chain at the end of
	the declarative region.
	* sem_ch7.adb (Analyze_Package_Body_Helper): Update use clause chain
	after analysis (Analyze_Package_Specification): Update use clause chain
	when there is no body.
	* sem_ch8.ads, sem_ch8.adb (Analyze_Use_Package, Analyze_Use_Type): Add
	parameter to determine weither the installation of scopes should also
	propagate on the use-clause "chain".
	(Mark_Use_Clauses): Created to traverse use-clause chains and determine
	what constitutes a valid "use" of a clause.
	(Update_Use_Clause_Chain): Created to aggregate common machinary used
	to clean up use-clause chains (and warn on ineffectiveness) at the end
	of declaritive regions.
	* sem_ch8.adb (Analyze_Package_Name): Created to perform analysis on a
	package name from a use-package clause.
	(Analyze_Package_Name_List): Created to perform analysis on a list of
	package names (similar to Analyze_Package_Name).
	(Find_Most_Prev): Created to traverse to the beginning of a given
	use-clause chain.
	(Most_Decendant_Use_Clause): Create to identify which clause from a
	given set is highest in scope (not always the most prev).
	(Use_One_Package, Use_One_Type): Major cleanup and reorganization to
	handle the new chaining algorithm, also many changes related to
	redundant clauses. A new parameter has also been added to force
	installation to handle certain cases.
	* sem_ch9.adb (Analyze_Entry_Body, Analyze_Protected_Body,
	Analyze_Task_Body): Mark use clauses on relevant entities.
	* sem_ch10.adb, sem_ch10.ads (Install_Context_Clauses,
	Install_Parents): Add parameter to determine weither the installation
	of scopes should also propagate on the use-clause "chain".
	* sem_ch12.adb (Inline_Instance_Body): Add flag in call to
	Install_Context to avoid redundant chaining of use-clauses.
	* sem_ch13.adb: Minor reformatting.
	* sem_res.adb (Resolve): Mark use clauses on operators.
	(Resolve_Call, Resolve_Entity_Name): Mark use clauses on relevant
	entities.
	* sinfo.adb, sinfo.ads (Is_Effective_Use_Clause,
	Set_Is_Effective_Use_Clause): Add new flag to N_Use_Clause nodes to
	represent any given clause's usage/reference/necessity.
	(Prev_Use_Clause, Set_Prev_Use_Clause): Add new field to N_Use_Clause
	nodes to allow loose chaining of redundant clauses.
	(Set_Used_Operations, Set_Subtype_Mark, Set_Prev_Ids, Set_Names,
	Set_More_Ids, Set_Name): Modify set procedure calls to reflect
	reorganization in node fields.
	* types.ads (Source_File_Index): Adjust index bounds.
	(No_Access_To_Source_File): New constant.

2017-09-25  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch13.adb (Analyze_One_Aspect): In ASIS mode make a full copy of
	the expression to be used in the generated attribute specification
	(rather than relocating it) to avoid resolving a potentially malformed
	tree when the expression is resolved through an ASIS-specific call to
	Resolve_Aspect_Expressions.  This manifests itself as a crash on a
	function with parameter associations.

From-SVN: r253144

2017-09-25  Yannick Moy  <moy@adacore.com>

	* exp_spark.adb (Expand_SPARK_Indexed_Component,
	Expand_SPARK_Selected_Component): New procedures to insert explicit
	dereference if required.
	(Expand_SPARK): Call the new procedures.

2017-09-25  Patrick Bernardi  <bernardi@adacore.com>

	* libgnat/a-stwiun.adb, libgnat/s-stchop__vxworks.adb,
	libgnat/g-socthi__vxworks.ads, libgnat/a-stzunb.adb,
	libgnat/a-strunb.adb, libgnarl/s-osinte__lynxos178.adb,
	libgnarl/s-intman__vxworks.adb, libgnarl/s-osinte__darwin.adb,
	libgnarl/a-exetim__darwin.adb: Removed ineffective use-clauses.

2017-09-25  Vasiliy Fofanov  <fofanov@adacore.com>

	* adaint.c (win32_wait): Properly handle error and take into account
	the WIN32 limitation on the number of simultaneous wait objects.

2017-09-25  Yannick Moy  <moy@adacore.com>

	* sem_ch3.adb (Constant_Redeclaration): Do not insert a call to the
	invariant procedure in GNATprove mode.
	* sem_ch5.adb (Analyze_Assignment): Likewise.

From-SVN: r253143

graphite-optimize-isl.c (optimize_isl): Fail and dump if ISL errors other than isl_error_quota happen.

2017-09-25  Richard Biener  <rguenther@suse.de>

	* graphite-optimize-isl.c (optimize_isl): Fail and dump if
	ISL errors other than isl_error_quota happen.  Dump if the
	schedule is the same.
	* graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
	errors instead of aborting inside ISL.

From-SVN: r253142

2017-09-25  Piotr Trojanek  <trojanek@adacore.com>

	* adabkend.adb (Call_Back_End): Fix wording of "front-end" and
	"back-end" in comments.

2017-09-25  Ed Schonberg  <schonberg@adacore.com>

	* exp_ch6.adb (Expand_Call_Helper): The extra accessibility check in a
	call that appears in a classwide precondition and that mentions an
	access formal of the subprogram, must use the accessibility level of
	the actual in the call. This is one case in which a reference to a
	formal parameter appears outside of the body of the subprogram.

From-SVN: r253141

2017-09-25  Hristian Kirtchev  <kirtchev@adacore.com>

	* sem_res.adb (Replace_Actual_Discriminants): Replace a discriminant
	for GNATprove.
	(Resolve_Entry): Clean up predicate

2017-09-25  Hristian Kirtchev  <kirtchev@adacore.com>

	* sem_prag.adb (Analyze_Constituent): Raise Unrecoverable_Error rather
	than Program_Error because U_E is more in line with respect to the
	intended behavior.

2017-09-25  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch13.adb (Resolve_Aspect_Expressions): The expression for aspect
	Storage_Size does not freeze, and thus can include references to
	deferred constants.

2017-09-25  Hristian Kirtchev  <kirtchev@adacore.com>

	* exp_spark.adb (Expand_SPARK_Potential_Renaming): Do not process a
	reference when it appears within a pragma of no significance to SPARK.
	(In_Insignificant_Pragma): New routine.
	* sem_prag.ads: Add new table Pragma_Significant_In_SPARK.

2017-09-25  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch12.adb (Analyze_Associations, case N_Formal_Package): If the
	actual is a renaming, indicate that it is the renamed package that must
	be frozen before the instantiation.

2017-09-25  Yannick Moy  <moy@adacore.com>

	* doc/gnat_ugn/gnat_and_program_execution.rst: Fix typo in description
	of dimensionality system in GNAT UG.
	* gnat_ugn.texi: Regenerate.

2017-09-25  Yannick Moy  <moy@adacore.com>

	* gnat1drv.adb: Call Check_Safe_Pointers from the frontend in
	GNATprove_Mode when switch -gnatdF used.

2017-09-25  Piotr Trojanek  <trojanek@adacore.com>

	* adabkend.adb (Call_Back_End): Reset Current_Error_Node when starting
	the backend.

From-SVN: r253140

exp_imgv.adb (Expand_Image_Attribute): Disable the optimized expansion of user-defined enumeration types when...

gcc/ada/

2017-09-25  Javier Miranda  <miranda@adacore.com>

	* exp_imgv.adb (Expand_Image_Attribute): Disable the optimized
	expansion of user-defined enumeration types when the generation of
	names for enumeration literals is suppressed.

2017-09-25  Gary Dismukes  <dismukes@adacore.com>

	* libgnarl/s-taprop__linux.adb: Minor reformatting.

2017-09-25  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch13.adb (Resolve_Aspect_Expressions): Do not resolve identifiers
	that appear as selector names of parameter associations, as these are
	never resolved by visibility.

2017-09-25  Justin Squirek  <squirek@adacore.com>

	* sem_res.adb (Resolve_Entry): Generate reference for index entities.

gcc/testsuite/

2017-09-25  Justin Squirek  <squirek@adacore.com>

	* gnat.dg/entry_family.adb: New testcase

From-SVN: r253139

2017-09-25  Doug Rupp  <rupp@adacore.com>

	* libgnarl/s-taprop__linux.adb (Compute_Base_Monotonic_Clock): Refine.

2017-09-25  Javier Miranda  <miranda@adacore.com>

	* exp_imgv.adb (Is_User_Defined_Enumeration_Type): New subprogram.
	(Expand_User_Defined_Enumeration_Image): New subprogram.
	(Expand_Image_Attribute): Enable speed-optimized expansion of
	user-defined enumeration types when we are compiling with optimizations
	enabled.

2017-09-25  Piotr Trojanek  <trojanek@adacore.com>

	* sem_util.adb (Has_Null_Abstract_State): Remove, as an exactly same
	routine is already provided by Einfo.
	* einfo.adb (Has_Null_Abstract_State): Replace with the body from
	Sem_Util, which had better comments and avoided double calls to
	Abstract_State.

From-SVN: r253138

	PR target/80556
	* config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
	of libgcc_eh for m64.
	* config/i386/darwin64.h: Likewise.

/* WORKAROUND pr80556:
   For x86_64 Darwin10 and later, the unwinder is in libunwind (redirected
   from libSystem).  This doesn't use the keymgr (see keymgr.c) and therefore
   the calls that libgcc makes to obtain the KEYMGR_GCC3_DW2_OBJ_LIST are not
   updated to include new images, and might not even be valid for a single
   image.
   Therefore, for 64b exes at least, we must use the libunwind implementation,
   even when static-libgcc is specified.  We put libSystem first so that
   unwinder symbols are satisfied from there.

From-SVN: r253137

2017-09-25  Bob Duff  <duff@adacore.com>

	* exp_ch3.adb: Rename Comp_Type_Simple to be Comp_Simple_Init.

2017-09-25  Doug Rupp  <rupp@adacore.com>

	* libgnarl/s-taprop__linux.adb (Base_Monotonic_Clock): New variable.
	(Compute_Base_Monotonic_Clock): New function.
	(Timed_Sleep): Adjust to use Base_Monotonic_Clock.
	(Timed_Delay): Likewise.
	(Monotonic_Clock): Likewise.
	* s-oscons-tmplt.c (CLOCK_MONOTONIC): Use on Linux.

From-SVN: r253136

2017-09-25  Ed Schonberg  <schonberg@adacore.com>

	* sem_ch12.adb (Save_References_In_Aggregate): Small correction to
	previous change.

2017-09-25  Hristian Kirtchev  <kirtchev@adacore.com>

	* exp_ch5.adb, sem_ch4.adb, sem_ch13.adb, sem_attr.adb, exp_ch3.adb:
	Minor reformatting.

From-SVN: r253135

2017-09-25  Richard Biener  <rguenther@suse.de>

	PR middle-end/82144
	* dwarf2out.c (gen_enumeration_type_die): Do not add alignment
	attribute for incomplete types nor twice for complete ones.

From-SVN: r253134

2017-09-25  Tom de Vries  <tom@codesourcery.com>

	* MAINTAINERS (CPU Port Maintainers): Add myself as nvptx maintainer.

From-SVN: r253133

This patch removes an unnecessary fold_convert to boolean_type_node at
the end of gfc_likely and gfc_unlikely.  It makes no difference to the
generated code, but makes tree dumps a little bit cleaner.

2017-09-25  Janne Blomqvist  <jb@gcc.gnu.org>
        * trans.c (gfc_unlikely): Remove unnecessary fold_convert.
        (gfc_likely): Likewise.

Regtested on x86_64-pc-linux-gnu.

From-SVN: r253132

From-SVN: r253131

Since the upper 32 bits of stack register are always zero for x32, we
can encode %esp as %rsp to avoid 0x67 prefix in address if there is no
index or base register.

gcc/

	PR target/82267
	* config/i386/i386.c (ix86_print_operand_address_as): Encode
	%esp as %rsp to avoid 0x67 prefix if there is no index or base
	register.

gcc/testsuite/

	PR target/82267
	* gcc.target/i386/pr82267.c: New test.

From-SVN: r253127

2017-09-24  Thomas Koenig  <tkoenig@gcc.gnu.org>
	PR fortran/66328

	* gfortran.dg/data_derived_1.f90: New test.

From-SVN: r253126

re PR fortran/54633 (ICEs and reject valid with MINLOC/MINVAL (MAXLOC/MAXVAL) due to lacking compile-time simplification)

2017-09-24  Thomas Koenig  <tkoenig@gcc.gnu.org>

	PR fortran/54633
	* gfortran.dg/intrinsic_bounds_1.f90: New test.
	* gfortran.dg/intrinsic_param_1.f90: New test.

From-SVN: r253125

	PR libgfortran/79612
	* runtime/bounds.c: Use GFC_ASSERT.

From-SVN: r253124

2017-09-24  Thomas Koenig  <tkoenig@gcc.gnu.org>
	    Steven G. Kargl  <kargl@gcc.gnu.org>

	PR fortran/80118
	* expr.c (gfc_get_full_arrayspec_from_expr): If there is
	no symtree, set array spec to NULL.

2017-09-24  Thomas Koenig  <tkoenig@gcc.gnu.org>
	    Steven G. Kargl  <kargl@gcc.gnu.org>

	PR fortran/80118
	* gfortran.dg/zero_sized_7.f90: New test.


Co-Authored-By: Steven G. Kargl <kargl@gcc.gnu.org>

From-SVN: r253123

From-SVN: r253122

	PR bootstrap/82306
	* config/i386/i386.opt (mprefer-avx256): Use
	ix86_target_flags variable.
	* config/i386/i386.c (ix86_target_string): Move
	-mprefer-avx256 to flag2_opts.

From-SVN: r253118

2017-09-23  Janus Weil  <janus@gcc.gnu.org>

	PR fortran/82143
	* lang.opt: Add the options -fdefault-real-10 and -fdefault-real-16.
	Rename flag_default_real to flag_default_real_8.
	* invoke.texi: Add documentation.
	* module.c (use_iso_fortran_env_module): flag_default_real is renamed.
	* trans-types.c (gfc_init_kinds): Implement the flags
	-fdefault-real-10 and -fdefault-real-16. Make -fdefault-double-8 work
	without -fdefault-real-8.

2017-09-23  Janus Weil  <janus@gcc.gnu.org>

	PR fortran/82143
	* gfortran.dg/promotion_3.f90: New test case.
	* gfortran.dg/promotion_4.f90: New test case.

From-SVN: r253117

gcc/testsuite:
	* gcc.target/i386/pr82196-1.c: (b): Remove volatile asm.
	* gcc.target/i386/pr82196-2.c: (b): Likewise.

libgcc:
	* configure.ac: Add Check for HAVE_AS_AVX.
	* config.in: Regenerate.
	* configure: Likewise.
	* config/i386/i386-asm.h: Include auto-target.h from libgcc.
	(SSE_SAVE, SSE_RESTORE): Emit .byte sequence for !HAVE_AS_AVX.
	Correct out-of-date comments.

From-SVN: r253116

From-SVN: r253114

	PR bootstrap/81926
	* cp-objcp-common.c (cp_get_debug_type): Do only one lookup.

From-SVN: r253109

re PR middle-end/35691 (Missed (a == 0) && (b == 0) into (a|(typeof(a)(b)) == 0 when the types don't match)

	PR middle-end/35691
	* match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
	and x != -1 | y != -1 into (x & y) != -1.

	* gcc.dg/pr35691-1.c: Use -fdump-tree-forwprop1-details
	instead of -fdump-tree-forwprop-details in dg-options.
	* gcc.dg/pr35691-2.c: Likewise.
	* gcc.dg/pr35691-3.c: New test.
	* gcc.dg/pr35691-4.c: New test.

From-SVN: r253107

	PR sanitizer/81929
	* tree.c (struct replace_placeholders_t): Add pset field.
	(replace_placeholders_r): Call cp_walk_tree with d->pset as
	last argument instead of NULL.  Formatting fix.
	(replace_placeholders): Add pset variable, add its address
	into data.  Pass &pset instead of NULL to cp_walk_tree.

	* g++.dg/ubsan/pr81929.C: New test.

From-SVN: r253106

    
    Reviewed-on: https://go-review.googlesource.com/64592

From-SVN: r253105

2017-09-22  Steve Ellcey  <sellcey@cavium.com>

	* config.gcc: Add new case statement to set
	default_gnu_indirect_function.  Remove it from x86_64-*-linux*,
	i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
	s390x-*-linux* case statements.   Added aarch64 to the list of
	supported architectures.

From-SVN: r253104

This PR shows that we weren't filtering out irrelevant stmts in
vect_get_peeling_costs_all_drs (unlike related loops in which
we iterate over all datarefs).

2017-09-22  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
	PR tree-optimization/82289
	* tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
	STMT_VINFO_RELEVANT_P.

gcc/testsuite/
	PR tree-optimization/82289
	* gcc.dg/vect/pr82289.c: New test.

From-SVN: r253103

extract_range_from_binary_expr_1 had:

      if (range_int_cst_p (&vr0)
          && range_int_cst_p (&vr1)
          && TYPE_OVERFLOW_WRAPS (expr_type))
        ...
      ...
      extract_range_from_multiplicative_op_1 (vr, code, &vr0, &vr1);

but extract_range_from_multiplicative_op_1 also requires range_int_cst_p.
I think we should bail out if either range isn't a constant.

This might only be theoretical with current sources, but it's needed
once polynomial constants are added.

2017-09-22  Richard Sandiford  <richard.sandiford@linaro.org>
	    Alan Hayward  <alan.hayward@arm.com>
	    David Sherwood  <david.sherwood@arm.com>

gcc/
	* tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
	for VR_RANGE only; don't allow VR_ANTI_RANGE.
	(extract_range_from_binary_expr_1): Don't call
	extract_range_from_multiplicative_op_1 if !range_int_cst_p.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>

From-SVN: r253102

The vectoriser aligned vectors to TYPE_ALIGN unconditionally, although
there was also a hard-coded assumption that this was equal to the type
size.  This was inconvenient for SVE for two reasons:

- When compiling for a specific power-of-2 SVE vector length, we might
  want to align to a full vector.  However, the TYPE_ALIGN is governed
  by the ABI alignment, which is 128 bits regardless of size.

- For vector-length-agnostic code it doesn't usually make sense to align,
  since the runtime vector length might not be a power of two.  Even for
  power of two sizes, there's no guarantee that aligning to the previous
  16 bytes will be an improveent.

This patch therefore adds a target hook to control the preferred
vectoriser (as opposed to ABI) alignment.

2017-09-22  Richard Sandiford  <richard.sandiford@linaro.org>
	    Alan Hayward  <alan.hayward@arm.com>
	    David Sherwood  <david.sherwood@arm.com>

gcc/
	* target.def (preferred_vector_alignment): New hook.
	* doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
	hook.
	* doc/tm.texi: Regenerate.
	* targhooks.h (default_preferred_vector_alignment): Declare.
	* targhooks.c (default_preferred_vector_alignment): New function.
	* tree-vectorizer.h (dataref_aux): Add a target_alignment field.
	Expand commentary.
	(DR_TARGET_ALIGNMENT): New macro.
	(aligned_access_p): Update commentary.
	(vect_known_alignment_in_bytes): New function.
	* tree-vect-data-refs.c (vect_calculate_required_alignment): New
	function.
	(vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
	Calculate the misalignment based on the target alignment rather than
	the vector size.
	(vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
	rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
	(vect_enhance_data_refs_alignment): Mask the byte misalignment with
	the target alignment, rather than masking the element misalignment
	with the number of elements in a vector.  Also use the target
	alignment when calculating the maximum number of peels.
	(vect_find_same_alignment_drs): Use vect_calculate_required_alignment
	instead of TYPE_ALIGN_UNIT.
	(vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
	Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
	(vect_create_addr_base_for_vector_ref): Update call accordingly.
	(vect_create_data_ref_ptr): Likewise.
	(vect_setup_realignment): Realign by ANDing with
	-DR_TARGET_MISALIGNMENT.
	* tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
	the number of peels based on DR_TARGET_ALIGNMENT.
	* tree-vect-stmts.c (get_group_load_store_type): Compare the gap
	with the guaranteed alignment boundary when deciding whether
	overrun is OK.
	(vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
	relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
	(ensure_base_align): Remove stmt_info parameter.  Get the
	target base alignment from DR_TARGET_ALIGNMENT.
	(vectorizable_store): Update call accordingly.   Interpret
	DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
	TYPE_ALIGN_UNIT.
	(vectorizable_load): Likewise.

gcc/testsuite/
	* gcc.dg/vect/vect-outer-3a.c: Adjust dump scan for new wording
	of alignment message.
	* gcc.dg/vect/vect-outer-3a-big-array.c: Likewise.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>

From-SVN: r253101

gcc/testsuite/ChangeLog:

	PR c/81854
	* gcc.target/i386/pr80732.c: Correct a type error.

From-SVN: r253100

This patch adds a helper function for getting the number of bytes
accessed by an unvectorised data reference, which helps when general
modes have a variable size.

2017-09-22  Richard Sandiford  <richard.sandiford@linaro.org>
	    Alan Hayward  <alan.hayward@arm.com>
	    David Sherwood  <david.sherwood@arm.com>

gcc/
	* tree-vectorizer.h (vect_get_scalar_dr_size): New function.
	* tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
	(vect_enhance_data_refs_alignment): Likewise.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>

From-SVN: r253099