| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
vectorizable_reduction has old code to cope with cases in which the
given statement belongs to a reduction group but isn't the first statement.
That can no longer happen, since all statements in the group go into the
same SLP node, and we only check the first statement in each node.
The point is to remove the only path through vectorizable_reduction
in which stmt and stmt_info refer to different statements.
2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
gcc/
* tree-vect-loop.c (vectorizable_reduction): Assert that the
function is not called for second and subsequent members of
a reduction group.
From-SVN: r263117
|
|
This minor clean-up avoids repeating the test for double reductions
and also moves the vect_get_vec_def_for_operand call to the same
function as the corresponding vect_get_vec_def_for_stmt_copy.
2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
gcc/
* tree-vect-loop.c (get_initial_def_for_reduction): Move special
cases for nested loops from here to ...
(vect_create_epilog_for_reduction): ...here. Only call
vect_is_simple_use for inner-loop reductions.
From-SVN: r263116
|
|
P1008R1 - prohibit aggregates with user-declared constructors
* class.c (check_bases_and_members): For C++2a set
CLASSTYPE_NON_AGGREGATE based on TYPE_HAS_USER_CONSTRUCTOR rather than
type_has_user_provided_or_explicit_constructor.
* g++.dg/ext/is_aggregate.C: Add tests with deleted or defaulted ctor.
* g++.dg/cpp0x/defaulted1.C (main): Ifdef out for C++2a B b = {1};.
* g++.dg/cpp0x/deleted2.C: Expect error for C++2a.
* g++.dg/cpp2a/aggr1.C: New test.
* g++.dg/cpp2a/aggr2.C: New test.
From-SVN: r263115
|
|
gcc/testsuite/
PR target/86640
* gcc.target/arm/pr86640.c: New testcase.
From-SVN: r263114
|
|
2018-07-31 Martin Liska <mliska@suse.cz>
PR gcov-profile/85338
PR gcov-profile/85350
PR gcov-profile/85372
* profile.c (struct location_triplet): New.
(struct location_triplet_hash): Likewise.
(output_location): Do not output a BB that
is already recorded for a line.
(branch_prob): Use streamed_locations.
2018-07-31 Martin Liska <mliska@suse.cz>
PR gcov-profile/85338
PR gcov-profile/85350
PR gcov-profile/85372
* gcc.misc-tests/gcov-pr85338.c: New test.
* gcc.misc-tests/gcov-pr85350.c: New test.
* gcc.misc-tests/gcov-pr85372.c: New test.
From-SVN: r263113
|
|
2018-07-31 Martin Liska <mliska@suse.cz>
PR gcov-profile/85370
* coverage.c (coverage_begin_function): Do not mark target
clones as artificial functions.
From-SVN: r263112
|
|
2018-07-31 Martin Liska <mliska@suse.cz>
PR gcov-profile/83813
PR gcov-profile/84758
PR gcov-profile/85217
PR gcov-profile/85332
* profile.c (branch_prob): Do not record GOTO expressions
for GIMPLE statements which locations are already streamed.
2018-07-31 Martin Liska <mliska@suse.cz>
PR gcov-profile/83813
PR gcov-profile/84758
PR gcov-profile/85217
PR gcov-profile/85332
* gcc.misc-tests/gcov-pr83813.c: New test.
* gcc.misc-tests/gcov-pr84758.c: New test.
* gcc.misc-tests/gcov-pr85217.c: New test.
* gcc.misc-tests/gcov-pr85332.c: New test.
From-SVN: r263111
|
|
2018-07-31 Martin Liska <mliska@suse.cz>
PR c++/86653
* parser.c (cp_parser_condition): Initialize non_constant_p
to false.
From-SVN: r263110
|
|
2018-07-31 Arnaud Charlet <charlet@adacore.com>
gcc/ada/
* clean.adb, gnatchop.adb, gnatfind.adb, gnatls.adb,
gnatmake.ads, gnatxref.adb, make.adb, make.ads, make_util.ads,
sfn_scan.adb, vxaddr2line.adb, xeinfo.adb, xoscons.adb,
xr_tabls.adb, xref_lib.adb: Address CodePeer messages.
From-SVN: r263108
|
|
2018-07-31 Arnaud Charlet <charlet@adacore.com>
gcc/ada/
* gnatlink.adb: Fix potential Constraint_Error if
Library_Version is too long.
From-SVN: r263107
|
|
2018-07-31 Arnaud Charlet <charlet@adacore.com>
gcc/ada/
* sem_elab.adb: Remove duplicate condition detected by CodePeer.
From-SVN: r263106
|
|
2018-07-31 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* exp_unst.adb (Subp_Index): In the case of a protected
operation, the relevant entry is the generated
protected_subprogram_body into which the original body is
rewritten. Assorted cleanup and optimizations.
From-SVN: r263105
|
|
2018-07-31 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* exp_attr.adb (Expand_Attribute, case Fixed_Value): Set the
base type of the result to ensure that proper overflow and range
checks are generated. If the target is a fixed-point tyoe,
generate the required overflow and range checks explicitly,
rather than relying on Apply_Type_Conversion_Checks, which might
simply set the Do_Range_Check flag and rely on the backend to
add the check.
From-SVN: r263104
|
|
This patch modifies call resolution to recognize when the designated type of
an access-to-subprogram requires secondary stack management, and establish
the proper transient block.
------------
-- Source --
------------
-- leak7.adb
procedure Leak7 is
Max_Iterations : constant := 10_000;
function Func return String is
begin
return "Will this leak? Or will it dry?";
end Func;
type Func_Ptr is access function return String;
procedure Anonymous_Leak (Func : access function return String) is
begin
for Iteration in 1 .. Max_Iterations loop
declare
Val : constant String := Func.all;
begin null; end;
end loop;
end Anonymous_Leak;
procedure Named_Leak (Func : Func_Ptr) is
begin
for Iteration in 1 .. Max_Iterations loop
declare
Val : constant String := Func.all;
begin null; end;
end loop;
end Named_Leak;
begin
Anonymous_Leak (Func'Access);
Named_Leak (Func'Access);
end Leak7;
----------------------------
-- Compilation and output --
----------------------------
$ gnatmake -q leak7.adb
$ valgrind ./leak7 >& leak7.txt
$ grep -c "still reachable" leak7.txt
0
2018-07-31 Hristian Kirtchev <kirtchev@adacore.com>
gcc/ada/
* sem_res.adb (Resolve_Call): Establish a transient scope to
manage the secondary stack when the designated type of an
access-to-subprogram requires it.
From-SVN: r263103
|
|
2018-07-31 Ed Schonberg <schonberg@adacore.com>
gcc/ada
* exp_ch7.adb (Check_Unnesting_Elaboration_Code): To find local
subprograms in the elaboration code for a package body, recurse
through nested statement sequences because a compiler-generated
procedure may appear within a condition statement.
From-SVN: r263102
|
|
This patch fixes a spurious compiler error on a call to a protected
operation whose profile includes a defaulted in-parameter that is a call
to another protected function of the same object.
2018-07-31 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* exp_ch6.adb (Expand_Protected_Subprogram_Call): Handle
properly a protected call that includes a default parameter that
is a call to a protected function of the same type.
gcc/testsuite/
* gnat.dg/prot5.adb, gnat.dg/prot5_pkg.adb,
gnat.dg/prot5_pkg.ads: New testcase.
From-SVN: r263101
|
|
This patch corrects an issue whereby building a multi-unit compilation with
missing sources resulted in a cryptic "code generation" error instead of the
appropriate file not found error.
------------
-- Source --
------------
-- main.adb
with Types;
procedure Main is
begin
null;
end;
-- types.ads
package Types is
procedure Force;
end;
----------------------------
-- Compilation and output --
----------------------------
& gnatmake -q main.adb
gnatmake: "types.adb" not found
2018-07-31 Justin Squirek <squirek@adacore.com>
gcc/ada/
* lib-writ.adb (Write_With_Lines): Modfiy the generation of
dependencies within ali files so that source unit bodies are
properly listed even if said bodies are missing. Perform legacy
behavior in GNATprove mode.
* lib-writ.ads: Modify documentation to reflect current behavior.
From-SVN: r263100
|
|
The alignment of the couple of types from System.OS_Interface was wrongly
set to 4 (32-bit) instead of 8 (64-bit) in 32-bit mode.
2018-07-31 Eric Botcazou <ebotcazou@adacore.com>
gcc/ada/
* libgnarl/s-osinte__solaris.ads (upad64_t): New private type.
(mutex_t): Use it for 'lock' and 'data' components.
(cond_t): Likewise for 'data' and use single 'flags' component.
From-SVN: r263099
|
|
This patch corrects an issue whereby objects of a record type with a
representation clause which are overlain by address would fail to get
assigned values properly when one or both of said objects were marked
volatile.
2018-07-31 Justin Squirek <squirek@adacore.com>
gcc/ada/
* exp_ch5.adb (Make_Field_Assign): Force temporarily generated
objects for assignment of overlaid user objects to be renamings
instead of constant declarations.
gcc/testsuite/
* gnat.dg/addr11.adb: New testcase.
From-SVN: r263098
|
|
This patch modifies the expansion of stand-alone subprogram bodies that appear
in the body of a protected type to properly associate aspects and pragmas to
the newly created spec for the subprogram body. As a result, the annotations
are properly associated with the initial declaration of the subprogram.
2018-07-31 Hristian Kirtchev <kirtchev@adacore.com>
gcc/ada/
* exp_ch9.adb (Analyze_Pragmas): New routine.
(Build_Private_Protected_Declaration): Code clean up. Relocate
relevant aspects and pragmas from the stand-alone body to the
newly created spec. Explicitly analyze any pragmas that have
been either relocated or produced by the analysis of the
aspects.
(Move_Pragmas): New routine.
* sem_prag.adb (Find_Related_Declaration_Or_Body): Recognize the
case where a pragma applies to the internally created spec for a
stand-along subprogram body declared in a protected body.
gcc/testsuite/
* gnat.dg/global.adb, gnat.dg/global.ads: New testcase.
From-SVN: r263097
|
|
When compiling with an assertion-enabled compiler, Assert_Failure can be
raised when expanded an extended_return_statement whose enclosing scope
is not a function (such as when it's a block_statement). The simple fix
is to change the Assert to test Current_Subprogram rather than Current_Scope.
Three such Assert pragmas are corrected in this way.
2018-07-31 Gary Dismukes <dismukes@adacore.com>
gcc/ada/
* exp_ch6.adb (Expand_N_Extended_Return_Statement): Replace
calls to Current_Scope in three assertions with calls to
Current_Subprogram.
gcc/testsuite/
* gnat.dg/block_ext_return_assert_failure.adb: New testcase.
From-SVN: r263096
|
|
This patch suppresses a spurious warning on the use of a 64-bit modular type
in a quantified expression, where the range of iteration will include a bound
that appears larger than the run-time representation of Universal_Integer'last.
2018-07-31 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* sem_eval.adb (Check_Non_Static_Context): Do not warn on an
integer literal greater than the upper bound of
Universal_Integer'Last when expansion is disabled, to avoid a
spurious warning over ranges involving 64-bit modular types.
gcc/testsuite/
* gnat.dg/iter3.adb: New testcase.
From-SVN: r263095
|
|
2018-07-31 Arnaud Charlet <charlet@adacore.com>
gcc/ada/
* einfo.adb (Write_Entity_Flags): Also print
Is_Activation_Record flag.
From-SVN: r263094
|
|
High-level wrappers are easier to read. This change came up while reading
some code related to GNATprove, but then uniformly applied to the entire
frontend. For the few remaining membership tests that could be replaced
by Is_Formal it is not obvious whether the high-level routine makes the
code better.
2018-07-31 Piotr Trojanek <trojanek@adacore.com>
gcc/ada/
* exp_aggr.adb, exp_ch4.adb, exp_ch6.adb, lib-xref.adb,
repinfo.adb, sem_ch9.adb: Minor replace Ekind membership tests
with a wrapper routine.
From-SVN: r263093
|
|
In GNATprove we used to store a variant of cross-reference information in
the ALI file in lines that started with an 'F' letter. This is no longer
the case, so the letter can be returned to the pool of unused prefixes.
2018-07-31 Piotr Trojanek <trojanek@adacore.com>
gcc/ada/
* ali.adb (Known_ALI_Lines): Remove 'F' as a prefix for lines
related to the FORMAL analysis done by GNATprove.
From-SVN: r263092
|
|
2018-07-31 Javier Miranda <miranda@adacore.com>
gcc/ada/
* sem.ads (Inside_Preanalysis_Without_Freezing): New global
counter.
* sem.adb (Semantics): This subprogram has now the
responsibility of resetting the counter before analyzing a unit,
and restoring its previous value before returning.
* freeze.adb (Freeze_Entity): Do not freeze if we are
preanalyzing without freezing.
* sem_res.adb (Preanalyze_And_Resolve): Set & restore
In_Preanalysis_Without_Freezing.
From-SVN: r263091
|
|
2018-07-31 Ed Schonberg <schonberg@adacore.com>
gcc/ada/
* sem_ch4.adb (Traverse_Homonyms): Consider generic actuals that
may rename a matching class-wide operation only if the renaming
declaration for that actual is in an enclosing scope (i.e.
within the corrresponding instance).
From-SVN: r263090
|
|
2018-07-31 Hristian Kirtchev <kirtchev@adacore.com>
gcc/ada
* checks.adb, contracts.adb, exp_aggr.adb, exp_ch5.adb,
exp_disp.adb, make.adb, sem_ch4.adb, sem_eval.adb, sem_res.adb,
usage.adb: Minor reformatting.
From-SVN: r263089
|
|
This patch fixes a bug in which if "pragma Default_Storage_Pool (null);"
is given, then a build-in-place function will get an incorrect error
message "allocation from empty storage pool" even though there is no
such allocation in the source program.
2018-07-31 Bob Duff <duff@adacore.com>
gcc/ada/
* sem_res.adb (Resolve_Allocator): Do not complain about the
implicit allocator that occurs in the expansion of a return
statement for a build-in-place function.
From-SVN: r263088
|
|
2018-07-31 Olivier Hainque <hainque@adacore.com>
* gcc.c (handle_spec_function): Accept a soft_matched_part
argument, as do_spec_1. Pass it down to ...
(eval_spec_function): Accept a soft_matched_part argument,
and pass it down to ...
(do_spec_2): Accept a soft_matched_part argument, and pass
it down to do_spec_1.
(do_spec_1): Pass soft_matched_part to handle_spec_function.
(handle_braces): Update call to handle_spec_function.
(driver::set_up_specs): Update calls to do_spec_2.
(compare_debug_dump_opt_spec_function): Likewise.
(compare_debug_self_opt_spec_function): Likewise.
From-SVN: r263087
|
|
2018-06-07 Olivier Hainque <hainque@adacore.com>
* common.opt (nolibc): New option.
* doc/invoke.texi (Link Options): Document it.
* gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
* config/alpha/linux.h: Likewise.
* config/arc/elf.h: Likewise.
* config/arm/uclinux-elf.h: Likewise.
* config/arm/unknown-elf.h: Likewise.
* config/avr/avrlibc.h: Likewise.
* config/bfin/bfin.h: Likewise.
* config/bfin/linux.h: Likewise.
* config/bfin/uclinux.h: Likewise.
* config/darwin.h: Likewise.
* config/darwin10.h: Likewise.
* config/darwin12.h: Likewise.
* config/gnu-user.h: Likewise.
* config/lm32/uclinux-elf.h: Likewise.
* config/pa/pa-hpux11.h: Likewise.
* config/pa/pa64-hpux.h: Likewise.
* config/sparc/sparc.h: Likewise.
From-SVN: r263083
|
|
2018-07-31 Andre Vieira <andre.simoesdiasvieira@arm.com>
Revert 'AsyncI/O patch committed'
2018-07-25 Nicolas Koenig <koenigni@gcc.gnu.org>
Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/25829
* gfortran.texi: Add description of asynchronous I/O.
* trans-decl.c (gfc_finish_var_decl): Treat asynchronous variables
as volatile.
* trans-io.c (gfc_build_io_library_fndecls): Rename st_wait to
st_wait_async and change argument spec from ".X" to ".w".
(gfc_trans_wait): Pass ID argument via reference.
2018-07-31 Andre Vieira <andre.simoesdiasvieira@arm.com>
Revert 'AsyncI/O patch committed'
2018-07-25 Nicolas Koenig <koenigni@gcc.gnu.org>
Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/25829
* gfortran.dg/f2003_inquire_1.f03: Add write statement.
* gfortran.dg/f2003_io_1.f03: Add wait statement.
2018-07-31 Andre Vieira <andre.simoesdiasvieira@arm.com>
Revert 'AsyncI/O patch committed'
2018-07-25 Nicolas Koenig <koenigni@gcc.gnu.org>
Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/25829
* Makefile.am: Add async.c to gfor_io_src.
Add async.h to gfor_io_headers.
* Makefile.in: Regenerated.
* gfortran.map: Add _gfortran_st_wait_async.
* io/async.c: New file.
* io/async.h: New file.
* io/close.c: Include async.h.
(st_close): Call async_wait for an asynchronous unit.
* io/file_pos.c (st_backspace): Likewise.
(st_endfile): Likewise.
(st_rewind): Likewise.
(st_flush): Likewise.
* io/inquire.c: Add handling for asynchronous PENDING
and ID arguments.
* io/io.h (st_parameter_dt): Add async bit.
(st_parameter_wait): Correct.
(gfc_unit): Add au pointer.
(st_wait_async): Add prototype.
(transfer_array_inner): Likewise.
(st_write_done_worker): Likewise.
* io/open.c: Include async.h.
(new_unit): Initialize asynchronous unit.
* io/transfer.c (async_opt): New struct.
(wrap_scalar_transfer): New function.
(transfer_integer): Call wrap_scalar_transfer to do the work.
(transfer_real): Likewise.
(transfer_real_write): Likewise.
(transfer_character): Likewise.
(transfer_character_wide): Likewise.
(transfer_complex): Likewise.
(transfer_array_inner): New function.
(transfer_array): Call transfer_array_inner.
(transfer_derived): Call wrap_scalar_transfer.
(data_transfer_init): Check for asynchronous I/O.
Perform a wait operation on any pending asynchronous I/O
if the data transfer is synchronous. Copy PDT and enqueue
thread for data transfer.
(st_read_done_worker): New function.
(st_read_done): Enqueue transfer or call st_read_done_worker.
(st_write_done_worker): New function.
(st_write_done): Enqueue transfer or call st_read_done_worker.
(st_wait): Document as no-op for compatibility reasons.
(st_wait_async): New function.
* io/unit.c (insert_unit): Use macros LOCK, UNLOCK and TRYLOCK;
add NOTE where necessary.
(get_gfc_unit): Likewise.
(init_units): Likewise.
(close_unit_1): Likewise. Call async_close if asynchronous.
(close_unit): Use macros LOCK and UNLOCK.
(finish_last_advance_record): Likewise.
(newunit_alloc): Likewise.
* io/unix.c (find_file): Likewise.
(flush_all_units_1): Likewise.
(flush_all_units): Likewise.
* libgfortran.h (generate_error_common): Add prototype.
* runtime/error.c: Include io.h and async.h.
(generate_error_common): New function.
2018-07-31 Andre Vieira <andre.simoesdiasvieira@arm.com>
Revert 'AsyncI/O patch committed'.
2018-07-25 Nicolas Koenig <koenigni@gcc.gnu.org>
Thomas Koenig <tkoenig@gcc.gnu.org>
PR fortran/25829
* testsuite/libgomp.fortran/async_io_1.f90: New test.
* testsuite/libgomp.fortran/async_io_2.f90: New test.
* testsuite/libgomp.fortran/async_io_3.f90: New test.
* testsuite/libgomp.fortran/async_io_4.f90: New test.
* testsuite/libgomp.fortran/async_io_5.f90: New test.
* testsuite/libgomp.fortran/async_io_6.f90: New test.
* testsuite/libgomp.fortran/async_io_7.f90: New test.
From-SVN: r263082
|
|
So the value can be used in places where an absolute path
is expected.
2018-07-31 Olivier Hainque <hainque@adacore.com>
* gcc.c (getenv_spec_function): Prepend '/' to value for
allowed undefined variables.
From-SVN: r263081
|
|
From-SVN: r263080
|
|
In arm_block_set_aligned_vect 8-bit constants are generated as zero-
extended const_ints, not sign-extended as required. Fix that.
PR target/86640
* config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
instead of GEN_INT.
From-SVN: r263075
|
|
Implement the proposed resolution from LWG 1052, which also resolves
DR 2118 by avoiding taking the address in the first place.
PR libstdc++/86734
* include/bits/stl_iterator.h (reverse_iterator::operator->): Call
_S_to_pointer (LWG 1052, LWG 2118).
(reverse_iterator::_S_to_pointer): Define overloaded helper functions.
* testsuite/24_iterators/reverse_iterator/dr1052.cc: New test.
* testsuite/24_iterators/reverse_iterator/dr2188.cc: New test.
From-SVN: r263074
|
|
20_util/memory_resource/2.cc FAILs on AIX 7.2.0.0, because aligned_alloc
incorrectly requires the alignment to be a multiple of sizeof(void*).
This adds a workaround to the operator new overload taking an alignment
value, to increase the alignment (and size) if needed.
* libsupc++/new_opa.cc (operator new(size_t, align_val_t)): Add
workaround for aligned_alloc bug on AIX.
* testsuite/18_support/new_aligned.cc: New test.
From-SVN: r263073
|
|
gcc/testsuite/
PR rtl-optimization/85160
* gcc.target/powerpc/combine-2-2.c: New testcase.
From-SVN: r263072
|
|
From-SVN: r263070
|
|
2018-07-30 Tom Tromey <tom@tromey.com>
* cplus-dem.c (remember_Btype): Don't call memcpy with LEN==0.
From-SVN: r263069
|
|
terminated string literal.
2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
* tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
terminated string literal.
From-SVN: r263068
|
|
This patch allows combine to combine two insns into two. This helps
in many cases, by reducing instruction path length, and also allowing
further combinations to happen. PR85160 is a typical example of code
that it can improve.
This patch does not allow such combinations if either of the original
instructions was a simple move instruction. In those cases combining
the two instructions increases register pressure without improving the
code. With this move test register pressure does no longer increase
noticably as far as I can tell.
(At first I also didn't allow either of the resulting insns to be a
move instruction. But that is actually a very good thing to have, as
should have been obvious).
PR rtl-optimization/85160
* combine.c (is_just_move): New function.
(try_combine): Allow combining two instructions into two if neither of
the original instructions was a move.
From-SVN: r263067
|
|
2018-07-30 Christophe Lyon <christophe.lyon@linaro.org>
* config/arm/ieee754-df.S: Fix comment for code working on
architectures >= 4.
* config/arm/ieee754-sf.S: Likewise.
From-SVN: r263066
|
|
PR target/86673
* doc/extend.texi (Global Register Variables): Discourage use of type
qualifiers.
(Local Register Variables): Likewise.
From-SVN: r263065
|
|
In r262961 I only updated the out-of-line copy of ceil_log2. This patch
applies the same change to the other (inline) one.
2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
gcc/
PR tree-optimization/86506
* hwint.h (ceil_log2): Resync with hwint.c implementation.
From-SVN: r263064
|
|
The idea behind the rclass loop in spill_hard_reg_in_range() seems to
be: find a hard_regno, which in general conflicts with reload regno,
but does not do so between `from` and `to`, and then do the live range
splitting based on this information. To check the absence of conflicts,
we make use of insn_bitmap, which does not contain insns which clobber
the hard_regno.
gcc/ChangeLog:
2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
PR target/86547
* lra-constraints.c (spill_hard_reg_in_range): When selecting the
hard_regno, make sure no insn between `from` and `to` clobbers it.
From-SVN: r263063
|
|
Currently parallel-loop-1.c fails at -O0 on a Quadro M1200, because one of the
kernel launch configurations exceeds the resources available in the device, due
to the default dimensions chosen by the runtime.
This patch fixes that by taking the per-function max_threads_per_block into
account when using the default dimensions.
2018-07-30 Tom de Vries <tdevries@suse.de>
* plugin/plugin-nvptx.c (MIN, MAX): Redefine.
(nvptx_exec): Ensure worker and vector default dims don't exceed
targ_fn->max_threads_per_block.
From-SVN: r263062
|
|
The default dimensions are calculated using per-device properties, but
initialized once and used on all devices.
This patch fixes this problem by introducing per-device default dimensions.
2018-07-30 Tom de Vries <tdevries@suse.de>
* plugin/plugin-nvptx.c (struct ptx_device): Add default_dims field.
(nvptx_open_device): Init default_dims for device.
(nvptx_exec): Use default_dims from device.
From-SVN: r263061
|
|
Currently, if the user doesn't specify the number of workers for an openacc
region, the compiler hardcodes it to a default value.
This patch removes this functionality, such that the libgomp runtime can decide
on a default value.
2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
Tom de Vries <tdevries@suse.de>
* config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
(PTX_DEFAULT_RUNTIME_DIM): ... this.
(nvptx_goacc_validate_dims): Set default worker and gang dims to
PTX_DEFAULT_RUNTIME_DIM.
(nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
Co-Authored-By: Tom de Vries <tdevries@suse.de>
From-SVN: r263060
|
|
From-SVN: r263059
|
