2019-05-02 Richard Biener <rguenther@suse.de>
PR tree-optimization/89653
* tree-ssa-loop.c (pass_data_tree_loop_init): Execute
update-address-taken before the pass.
* passes.def (pass_tree_loop_init): Put comment before it.
* g++.dg/vect/pr89653.cc: New testcase.
From-SVN: r270800
2019-05-02 Richard Biener <rguenther@suse.de>
PR tree-optimization/89509
* tree-ssa-structalias.c (compute_dependence_clique): Look
at the first subvar when determining whether it is restrict.
* gcc.dg/torture/restrict-8.c: New testcase.
From-SVN: r270799
This patch does two things. For the general vectoriser, it adds support to
perform fully masked reductions over expressions that don't support masking.
This is achieved by using VEC_COND_EXPR where possible. At the moment this is
implemented for DOT_PROD_EXPR only, but the framework is there to extend it to
other expressions.
Related to that, this patch adds support to vectorize dot product using SVE. It
also uses the new functionality to ensure that the resulting loop is masked.
Given this input code:
uint32_t
dotprod (uint8_t *restrict x, uint8_t *restrict y, int n)
{
uint32_t sum = 0;
for (int i = 0; i < n; i++)
{
sum += x[i] * y[i];
}
return sum;
}
The resulting SVE code is:
0000000000000000 <dotprod>:
0: 7100005f cmp w2, #0x0
4: 5400024d b.le 4c <dotprod+0x4c>
8: d2800003 mov x3, #0x0 // #0
c: 93407c42 sxtw x2, w2
10: 2538c001 mov z1.b, #0
14: 25221fe0 whilelo p0.b, xzr, x2
18: 2538c003 mov z3.b, #0
1c: d503201f nop
20: a4034002 ld1b {z2.b}, p0/z, [x0, x3]
24: a4034020 ld1b {z0.b}, p0/z, [x1, x3]
28: 0430e3e3 incb x3
2c: 0523c000 sel z0.b, p0, z0.b, z3.b
30: 25221c60 whilelo p0.b, x3, x2
34: 44820401 udot z1.s, z0.b, z2.b
38: 54ffff41 b.ne 20 <dotprod+0x20> // b.any
3c: 2598e3e0 ptrue p0.s
40: 04812021 uaddv d1, p0, z1.s
44: 1e260020 fmov w0, s1
48: d65f03c0 ret
4c: 1e2703e1 fmov s1, wzr
50: 1e260020 fmov w0, s1
54: d65f03c0 ret
Notice how udot is used inside a fully masked loop.
I tested this patch in an aarch64 machine bootstrapping the compiler and
running the checks.
gcc/Changelog:
2019-05-02 Alejandro Martinez <alejandro.martinezvicente@arm.com>
* config/aarch64/aarch64-sve.md (<sur>dot_prod<vsi2qi>): Taken from SVE
ACLE branch.
* config/aarch64/iterators.md: Copied Vetype_fourth, VSI2QI and vsi2qi from
SVE ACLE branch.
* tree-vect-loop.c (use_mask_by_cond_expr_p): New function to check if a
VEC_COND_EXPR be inserted to emulate a conditional internal function.
(build_vect_cond_expr): Emit the VEC_COND_EXPR.
(vectorizable_reduction): Use the functions above to vectorize in a
fully masked loop codes that don't have a conditional internal
function.
gcc/testsuite/Changelog:
2019-05-02 Alejandro Martinez <alejandro.martinezvicente@arm.com>
* gcc.target/aarch64/sve/dot_1.c: New test for dot product.
From-SVN: r270790
2019-05-02 Martin Liska <mliska@suse.cz>
* gcc.c (process_command): Add dummy file only
if n_infiles == 0.
* opts-global.c (decode_options): Pass lang_mask.
* opts.c (print_help): New function.
(finish_options): Print --help if help_option_argument
is set.
(common_handle_option): Factor out content of OPT__help_
into print_help.
* opts.h (finish_options): Add new argument.
From-SVN: r270788
2019-05-02 Martin Liska <mliska@suse.cz>
PR target/88809
* config/i386/i386.c (ix86_expand_strlen): Use strlen call.
With -minline-all-stringops use inline expansion using 4B loop.
* doc/invoke.texi: Document the change of
-minline-all-stringops.
2019-05-02 Martin Liska <mliska@suse.cz>
PR target/88809
* gcc.target/i386/pr88809.c: New test.
* gcc.target/i386/pr88809-2.c: New test.
From-SVN: r270787
A direct interface is an interface whose data word contains the
actual data value, instead of a pointer to it. The gc toolchain
creates a direct interface if the value is pointer shaped, that
includes pointers (including unsafe.Pointer), functions, channels,
maps, and structs and arrays containing a single pointer-shaped
field. In gccgo, we only do this for pointers. This CL unifies
direct interface types with gc. This reduces allocations when
converting such types to interfaces.
Our method functions used to always take pointer receivers, to
make interface calls easy. Now for direct interface types, their
value methods will take value receivers. For a pointer to those
types, when converted to interface, the interface data contains
the pointer. For that interface to call a value method, it will
need a wrapper method that dereference the pointer and invokes
the value method. The wrapper method, instead of the actual one,
is put into the itable of the pointer type.
In the runtime, adjust funcPC for the new layout of interfaces of
functions.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/168409
From-SVN: r270779
Previously, each time we do an interface conversion for which the
method table is not known at compile time, we allocate a new
method table.
This CL ports the mechanism of itab caching from the gc runtime,
adapted to our itab representation and method finding mechanism.
With the cache, we reuse the same itab for the same (interface,
concrete) type pair. This reduces allocations in interface
conversions.
Unlike the gc runtime, we don't prepopulate the cache with
statically allocated itabs, as currently we don't have a way to
find them. This means we don't deduplicate run-time allocated
itabs with compile-time allocated ones. But that is not too bad
-- it is just a cache anyway.
As now itabs are never freed, it is also possible to drop the
write barrier for writing the first word of an interface header.
I'll leave this optimization for the future.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/171617
From-SVN: r270778
2019-05-01 Dominique d'Humieres <dominiq@gcc.gnu.org>
PR fortran/60144
* match.c (gfc_match_parens): Change the location for missing ')'.
(gfc_match_if): Detect a missing '('. Remove the spurious named
constant error. Change the wording of some errors.
(gfc_match_else): Change the wording of an error.
(gfc_match_elseif): Detect a missing '('. Improve the matching
process to get a better syntax analysis.
PR fortran/60144
* gfortran.dg/block_name_2.f90: Adjust dg-error.
* gfortran.dg/dec_type_print_3.f90.f90: Likewise
* gfortran.dg/pr60144.f90: New test.
From-SVN: r270776
PR tree-optimization/88797
* gimple-ssa-split-paths (is_feasible_trace): Reject cases where the
PHI feeds a conditional on the RHS of an assignment.
PR tree-optimization/88797
* g++.dg/tree-ssa/pr88797.C: New test.
From-SVN: r270775
The current generic implementation of __complex_proj used when cproj is
not available calculates the wrong projection, giving a different result
than given by C99's cproj.
When C99 cproj is not available but isinf and copysign are, use those to
give correct results for float, double and long double. Otherwise, and
for other specializations of std::complex, just use a generic version
that returns its argument, and so doesn't support infinities.
We might want to consider adding additional overloads of __complex_proj
to support extended types such as _Float64x, _Float128 etc.
PR libstdc++/61761
* include/std/complex (__complex_proj): Return parameter unchanged.
[_GLIBCXX_USE_C99_COMPLEX] (__complex_proj): Change overloads for
floating-point types to take std::complex arguments.
[_GLIBCXX_USE_C99_MATH_TR1] (__complex_proj): Add overloads for
floating-point types.
* testsuite/26_numerics/complex/proj.cc: New test.
From-SVN: r270759
2019-04-30 Martin Liska <mliska@suse.cz>
* opts.c (enable_warning_as_error): Provide hints
for unknown options.
2019-04-30 Martin Liska <mliska@suse.cz>
* c-pragma.c (handle_pragma_diagnostic): Provide hints
for unknown options.
2019-04-30 Martin Liska <mliska@suse.cz>
* gcc.dg/Werror-13.c: Add new tests for it.
* gcc.dg/pragma-diag-6.c: Likewise.
From-SVN: r270693
PR target/89093
* config/aarch64/aarch64.c (aarch64_process_one_target_attr): Don't skip
whitespace at the start of target attribute string.
* gcc.target/aarch64/pr89093.c: New test.
* gcc.target/aarch64/pr63304_1.c: Remove space from target string.
From-SVN: r270690
This keeps coming up repeatedly and the ACLE has finally added
__ARM_FEATURE_ATOMICS for the LSE feature in GCC. This is now part of
the latest ACLE release
(https://developer.arm.com/docs/101028/latest/5-feature-test-macros)
I know it's late for GCC-9 but this is a simple macro which need not
wait for another year.
Ok for trunk and to backport to all release branches ?
Tested with a simple build and a smoke test.
PR target/86538
* config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Define
__ARM_FEATURE_ATOMICS
From-SVN: r270686
The root_path.cc test had some debugging macros left in accidentally, so
didn't FAIL correctly if an assertion failed.
The string-char8_t.cc tests didn't compile on Windows.
* testsuite/27_io/filesystem/path/decompose/root_path.cc: Remove
macros accidentally left in.
* testsuite/27_io/filesystem/path/native/string-char8_t.cc: Remove
unnecessary -lstdc++fs option. Fix test for mingw.
* testsuite/experimental/filesystem/path/native/string-char8_t.cc:
Fix test for mingw.
From-SVN: r270685
2019-04-30 Martin Liska <mliska@suse.cz>
* gimple-fold.c (gimple_fold_builtin_memory_op): Change endp
into built_in_function enum. Remove code for endp == 2 and
use BUILT_IN_* constants.
(gimple_fold_builtin): Call the function with fcode.
From-SVN: r270682
2019-04-30 Martin Liska <mliska@suse.cz>
* config/i386/i386.c (ix86_builtin_reciprocal): Cast
DECL_FUNCTION_CODE into ix86_builtins enum before
the switch statement.
From-SVN: r270681
