Skip to content

R
riscv-gcc-1
Switch branch/tag
  1. 12 Jun, 2019 5 commits
    • re PR target/90811 ([nvptx] ptxas error on OpenMP offloaded code) · 26d7a5e6 
      	PR target/90811
	* cfgexpand.c (align_local_variable): Add really_expand argument,
	don't SET_DECL_ALIGN if it is false.
	(add_stack_var): Add really_expand argument, pass it through to
	align_local_variable.
	(expand_one_stack_var_1): Pass true as really_expand to
	align_local_variable.
	(expand_one_ssa_partition): Pass true as really_expand to
	add_stack_var.
	(expand_one_var): Pass really_expand through to add_stack_var.

From-SVN: r272181
      Jakub Jelinek committed
    • [arm] Implement usadv16qi and ssadv16qi standard names · 84ae7213 
      
This patch implements the usadv16qi and ssadv16qi standard names for arm.

The V16QImode variant is important as it is the most commonly used pattern:
reducing vectors of bytes into an int.
The midend expects the optab to compute the absolute differences of operands 1
and 2 and reduce them while widening along the way up to SImode. So the inputs
are V16QImode and the output is V4SImode.

I've based my solution on Aarch64 usadv16qi and ssadv16qi standard names
current implementation (r260437). This solution emits below sequence of
instructions:

        VABDL.u8        tmp, op1, op2   # op1, op2 lowpart
        VABAL.u8        tmp, op1, op2   # op1, op2 highpart
        VPADAL.u16      op3, tmp

So, for the code:

$ arm-none-linux-gnueabihf-gcc -S -O3 -march=armv8-a+simd -mfpu=auto -mfloat-abi=hard usadv16qi.c -dp

#define N 1024
unsigned char pix1[N];
unsigned char pix2[N];

int
foo (void)
{
  int i_sum = 0;
  int i;
  for (i = 0; i < N; i++)
    i_sum += __builtin_abs (pix1[i] - pix2[i]);
  return i_sum;
}

we now generate on arm:
foo:
        movw    r3, #:lower16:pix2      @ 57    [c=4 l=4]  *arm_movsi_vfp/3
        movt    r3, #:upper16:pix2      @ 58    [c=4 l=4]  *arm_movt/0
        vmov.i32        q9, #0  @ v4si  @ 3     [c=4 l=4]  *neon_movv4si/2
        movw    r2, #:lower16:pix1      @ 59    [c=4 l=4]  *arm_movsi_vfp/3
        movt    r2, #:upper16:pix1      @ 60    [c=4 l=4]  *arm_movt/0
        add     r1, r3, #1024   @ 8     [c=4 l=4]  *arm_addsi3/4
.L2:
        vld1.8  {q11}, [r3]!    @ 11    [c=8 l=4]  *movmisalignv16qi_neon_load
        vld1.8  {q10}, [r2]!    @ 10    [c=8 l=4]  *movmisalignv16qi_neon_load
        cmp     r1, r3  @ 21    [c=4 l=4]  *arm_cmpsi_insn/2
        vabdl.u8        q8, d20, d22    @ 12    [c=8 l=4]  neon_vabdluv8qi
        vabal.u8        q8, d21, d23    @ 15    [c=88 l=4]  neon_vabaluv8qi
        vpadal.u16      q9, q8  @ 16    [c=8 l=4]  neon_vpadaluv8hi
        bne     .L2             @ 22    [c=16 l=4]  arm_cond_branch
        vadd.i32        d18, d18, d19   @ 24    [c=120 l=4]  quad_halves_plusv4si
        vpadd.i32       d18, d18, d18   @ 25    [c=8 l=4]  neon_vpadd_internalv2si
        vmov.32 r0, d18[0]      @ 30    [c=12 l=4]  vec_extractv2sisi/1

instead of:
foo:
        @ args = 0, pretend = 0, frame = 0
        @ frame_needed = 0, uses_anonymous_args = 0
        @ link register save eliminated.
        movw    r3, #:lower16:pix1
        movt    r3, #:upper16:pix1
        vmov.i32        q9, #0  @ v4si
        movw    r2, #:lower16:pix2
        movt    r2, #:upper16:pix2
        add     r1, r3, #1024
.L2:
        vld1.8  {q8}, [r3]!
        vld1.8  {q11}, [r2]!
        vmovl.u8 q10, d16
        cmp     r1, r3
        vmovl.u8 q8, d17
        vmovl.u8 q12, d22
        vmovl.u8 q11, d23
        vsub.i16        q10, q10, q12
        vsub.i16        q8, q8, q11
        vabs.s16        q10, q10
        vabs.s16        q8, q8
        vaddw.s16       q9, q9, d20
        vaddw.s16       q9, q9, d21
        vaddw.s16       q9, q9, d16
        vaddw.s16       q9, q9, d17
        bne     .L2
        vadd.i32        d18, d18, d19
        vpadd.i32       d18, d18, d18
        vmov.32 r0, d18[0]

2019-06-12  Przemyslaw Wirkus  <przemyslaw.wirkus@arm.com>

        * config/arm/iterators.md (VABAL): New int iterator.
        * config/arm/neon.md (<sup>sadv16qi): New define_expand.
        * config/arm/unspecs.md ("unspec"): Define UNSPEC_VABAL_S, UNSPEC_VABAL_U
        values.

        * gcc.target/arm/ssadv16qi.c: New test.
        * gcc.target/arm/usadv16qi.c: Likewise.

From-SVN: r272180
      Przemyslaw Wirkus committed
    • Remove wrong assert about single value profiler. · d134323b 
      2019-06-12  Martin Liska  <mliska@suse.cz>

	* value-prof.c (stream_out_histogram_value): Only first value
	can't be negative.

From-SVN: r272179
      Martin Liska committed
    • re PR c/90760 (ICE on attributes section and alias in set_section, at symtab.c:1573) · f3139680 
      	PR c/90760
	* symtab.c (symtab_node::set_section): Allow being called on aliases
	as long as they aren't analyzed yet.

	* gcc.dg/pr90760.c: New test.

From-SVN: r272178
      Jakub Jelinek committed
    • Daily bump. · bfde1e21 
      From-SVN: r272177
      GCC Administrator committed
  3. 11 Jun, 2019 19 commits
  5. 10 Jun, 2019 16 commits
    • * config/i386/i386.md (absneg fp_reg splitter): Fix operator mode. · 4f59794b 
      From-SVN: r272134
      Uros Bizjak committed
    • compiler: support inlining functions that use index expressions · e898243c 
          
    Also move the determine_types pass on an inlined function body to one
    place, rather than doing it ad hoc as needed.
    
    This adds 79 new inlinable functions in the standard library, such as
    bytes.HasPrefix and bytes.LastIndexByte.
    
    Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/181261

From-SVN: r272133
      Ian Lance Taylor committed
    • compiler: make heap expression's write barrier conditional · 35c19de6 
          
    Heap_expression::do_get_backend emits an unconditional write
    barrier if the type has pointers and it is not a stack allocation.
    This CL changes it to use a write barrier for the assignment only
    when write barriers are enabled. While here, also change it to
    call gcWriteBarrier instead of typedmemmove for pointer-shaped
    types.
    
    For this to work, Function::build needs to be adjusted so that
    Heap_expression::do_get_backend is called when there is a parent
    block.
    
    Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/181540

From-SVN: r272132
      Ian Lance Taylor committed
    • compiler: permit inlining functions with labels and goto statements · fc917b42 
          
    This permits inlining functions with for loops and some switches, as
    they are lowered to if and goto statements before exporting them.
    
    This by itself only adds three new inlinable functions in the standard
    library: sort.Search, context.(*emptyCtx).String, and
    cmd/go/internal/work.(*Builder).disableBuildID.
    
    Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/181197

From-SVN: r272131
      Ian Lance Taylor committed
    • compiler: use gcWriteBarrier for pointer-shaped struct/array · d480455f 
          
    If a struct/array is pointer-shaped (i.e. having a single field
    that is pointer-shaped), we can use gcWriteBarrier instead of
    typedmemmove for the write barrier.
    
    Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/181539

From-SVN: r272130
      Ian Lance Taylor committed
    • re PR c++/87250 (Internal compiler error with -Os -fsyntax-only) · fa237d91 
      	PR c++/87250
	* g++.dg/cpp0x/pr87250.C: New test.

From-SVN: r272129
      Matthew Beliveau committed
    • constexpr.c (free_constructor): New. · 620adbec 
      	* constexpr.c (free_constructor): New.

	(cxx_eval_call_expression): Free parameter value CONSTRUCTORs.

From-SVN: r272127
      Jason Merrill committed
    • Reduce unsharing in constexpr call evaluation. · 9b9eb42a 
      	* constexpr.c (unshare_constructor): Only unshare if T is itself a
	CONSTRUCTOR.
	(cxx_eval_call_expression): Don't call it on the result here.

From-SVN: r272126
      Jason Merrill committed
    • Reduce constexpr_call memory consumption. · 3c961dc7 
      	* constexpr.c (cxx_bind_parameters_in_call): Use TREE_VEC rather
	than TREE_LIST.
	(constexpr_call_hasher::equal, cxx_bind_parameters_in_call)
	(cxx_eval_call_expression): Adjust.

From-SVN: r272125
      Jason Merrill committed
    • compiler: make escape analysis work with imported inlineable functions · c2f879e1 
          
    The escape analysis was written before we import inlineable
    function bodies, and in some places it skipped functions that are
    not in the local package. Now that there are imported function
    bodies, make the escape analysis work with them.
    
    Note that it is necessary for the escape analysis to run on
    imported function bodies, even if they are already tagged. The
    tags only have the information of the parameters (receiver,
    results), but not the internal nodes, e.g. local variables. We
    still need to do the analysis to get all the information. (In the
    future maybe we could export/import escape info for internal
    nodes also, then we don't need to redo the analysis.)
    
    Also add assertions to ensure that if we analyze the same
    function in multiple places, they'd better agree with each other.
    
    Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/181537

From-SVN: r272124
      Ian Lance Taylor committed
    • i386-protos.h (ix86_split_fp_absneg_operator): New prototype. · f359611b 
      	* config/i386/i386-protos.h (ix86_split_fp_absneg_operator):
	New prototype.
	* config/i386/i386-expand.c (ix86_expand_fp_absneg_operator):
	Emit clobber also for non-sse operations.
	(ix86_split_fp_absneg_operator): New function.
	* config/i386/i386.md (SSEMODEF): New mode iterator.
	(ssevecmodef): New mode attribute.
	(<code>tf2): Use absneg code iterator.
	(*<code>tf2_1): Rename from *absnegtf3_sse. Use absneg code iterator.
	Add three-operand AVX alternatives.
	(*<code><mode>2_i387_1): Rename from *absnegxf2_i387.
	Use absneg code iterator and X87MODEF mode iterator.
	(absneg fp_reg non-sse splitter): Call absneg code iterator
	and X87MODEF mode iterator.
	(absneg general_reg non-sse splitter): Use absneg code iterator
	and X87MODEF mode iterator.  Use ix86_split_fp_absneg_operator.
	(*<code><mode>2_1): Rename from *absneg<mode>2.  Use absneg
	code iterator.  Add three-operand AVX alternative.
	(absneg sse_reg splitter): Use absneg code iterator
	and SSEMODEF mode iterator.  Handle AVX operands.
	(absneg fp_reg splitter): Use absneg code iterator
	and MODEF mode iterator.
	(absneg general_reg splitter): Merge splitters using MODEF mode
	iterator.  Use absneg code iterator.  Call
	ix86_split_fp_absneg_operator.
	(*<code><mode>2_i387): Rename from *<code><mode>2_1.
	Do not enable for non-sse modes before reload.
	(CSGNMODE): Remove.
	(CSGNVMODE): Ditto.
	(copysing<mode>3): Use SSEMODEF instead of CSGNMODE and
	ssevecmodef mode attribute instaed of CSGNVMODE.
	(copysign<mode>3_const): Ditto.
	(copysign<mode>3_var): Ditto.
	* config/i386/i386.md (*<code><mode>2): Rename from *absneg<mode>2.
	Use absneg code iterator.  Simplify code using std::swap.
	* config/i386/predicates.md (absneg_operator): Remove.

From-SVN: r272123
      Uros Bizjak committed
    • re PR testsuite/90772 (g++.dg/cpp1y/feat-cxx14.C and several others fails starting with r272011) · f2ee751d 
      	PR testsuite/90772
	* g++.dg/cpp1y/feat-cxx14.C: Use std::size_t instead of size_t.
	* g++.dg/cpp1z/feat-cxx1z.C: Likewise.
	* g++.dg/cpp2a/feat-cxx2a.C: Likewise.
	* g++.dg/cpp1z/pr85569.C: Include <functional>.
	* g++.dg/tree-ssa/pr80293.C: Include <cstdint>.
	* g++.dg/tree-ssa/pr69336.C: Include <stdexcept>.

From-SVN: r272122
      Jakub Jelinek committed
    • PR other/90695 reduce testcase to remove library dependency · 39d1184b 
      This reproduces the original ICE fixed by r178857 (tested at r178852 and
r178860), without depending on a libstdc++ header that keeps changing.

The number of errors differs between C++14 and C++17 modes, so the fixed
test uses dg-excess-errors to match any number of them. The precise
errors aren't what's being tested for here anyway, the point of the test
is to verify the ICE in PR 50391 is fixed.

	PR other/90695
	* g++.dg/cpp0x/noexcept15.C: Remove dependency on library header.

From-SVN: r272121
      Jonathan Wakely committed
    • gimple-fold.c (get_range_strlen): Update comment that didn't make it into… · 97623b52 
      gimple-fold.c (get_range_strlen): Update comment that didn't make it into r267503 or related commits.


gcc/ChangeLog:

	* gimple-fold.c (get_range_strlen): Update comment that didn't
	make it into r267503 or related commits.

From-SVN: r272120
      Martin Sebor committed
    • gcov-tool: Mark {merge,rewrite}_usage with noreturn attribute · c31783fd 
      2019-06-10  Vladislav Ivanishin  <vlad@ispras.ru>

	* gcov-tool.c (merge_usage, rewrite_usage): Mark with
	ATTRIBUTE_NORETURN thus making consistent with overlap_usage.

From-SVN: r272119
      Vladislav Ivanishin committed
    • tree.def (OMP_SCAN): New tree code. · bf38f7e9 
      	* tree.def (OMP_SCAN): New tree code.
	* tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_INCLUSIVE and
	OMP_CLAUSE_EXCLUSIVE.
	* tree.h (OMP_CLAUSES): Use OMP_SCAN instead of OMP_TASKGROUP.
	(OMP_SCAN_BODY, OMP_SCAN_CLAUSES): Define.
	* tree.c (omp_clause_num_ops, omp_clause_code_name): Add entries for
	OMP_CLAUSE_{IN,EX}CLUSIVE.
	(walk_tree_1): Handle OMP_CLAUSE_{IN,EX}CLUSIVE.
	* tree-nested.c (convert_nonlocal_reference_stmt,
	convert_local_reference_stmt, convert_gimple_call): Handle
	GIMPLE_OMP_SCAN.
	* tree-pretty-print.c (dump_omp_clause): Handle
	OMP_CLAUSE_{IN,EX}CLUSIVE.
	(dump_generic_node): Handle OMP_SCAN.
	* gimple.def (GIMPLE_OMP_SCAN): New gimple code.
	* gimple.h (gomp_scan): New type.
	(is_a_helper <gomp_scan *>::test,
	is_a_helper <const gomp_scan *>::test): New templates.
	(gimple_build_omp_scan): Declare.
	(gimple_omp_scan_clauses, gimple_omp_scan_clauses_ptr,
	gimple_omp_scan_set_clauses): New inline functions.
	(CASE_GIMPLE_OMP): Add case GIMPLE_OMP_SCAN:.
	* gimple.c (gimple_build_omp_scan): New function.
	(gimple_copy): Handle GIMPLE_OMP_SCAN.
	* gimple-walk.c (walk_gimple_op, walk_gimple_stmt): Likewise.
	* gimple-pretty-print.c (dump_gimple_omp_block): Don't handle
	GIMPLE_OMP_TASKGROUP.
	(dump_gimple_omp_scan): New function.
	(pp_gimple_stmt_1): Handle GIMPLE_OMP_SCAN.
	* gimple-low.c (lower_stmt): Handle GIMPLE_OMP_SCAN.
	* tree-inline.c (remap_gimple_stmt, estimate_num_insns): Likewise.
	* gimplify.c (enum gimplify_omp_var_data): Add GOVD_REDUCTION_INSCAN.
	(is_gimple_stmt): Handle OMP_SCAN.
	(gimplify_scan_omp_clauses): Reject inscan reductions on constructs
	other than OMP_FOR or OMP_SIMD.  Handle OMP_CLAUSE_{IN,EX}CLUSIVE.
	(gimplify_adjust_omp_clauses): Diagnose inscan reductions not
	mentioned in nested #pragma omp scan.  Handle
	OMP_CLAUSE_{IN,EX}CLUSIVE.
	(gimplify_expr): Handle OMP_SCAN.
	* omp-low.c (check_omp_nesting_restrictions): For parent context,
	look through GIMPLE_OMP_SCAN context.  Allow #pragma omp scan in
	simd constructs.
	(scan_omp_1_stmt, lower_omp_1, diagnose_sb_1, diagnose_sb_2): Handle
	GIMPLE_OMP_SCAN.
c-family/
	* c-pragma.h (enum pragma_kind): Add PRAGMA_OMP_SCAN.
	* c-pragma.c (omp_pragmas_simd): Add #pragma omp scan.
	* c-omp.c (c_omp_split_clauses): Diagnose inscan reductions on
	combined/composite constructs where it is not allowed.  Copy over
	OMP_CLAUSE_REDUCTION_INSCAN.
c/
	* c-parser.c (c_parser_pragma): Reject PRAGMA_OMP_SCAN.
	(c_parser_omp_clause_reduction): Don't sorry_at on inscan reductions.
	(c_parser_omp_scan_loop_body): New function.
	(c_parser_omp_for_loop): Call c_parser_omp_scan_loop_body if there are
	inscan reduction clauses.
	* c-typeck.c (c_finish_omp_clauses): Reject mixing inscan with
	non-inscan reductions on the same construct, or inscan reductions with
	ordered or schedule clauses, or inscan array reductions.
cp/
	* parser.c (cp_parser_omp_clause_reduction): Don't sorry_at on inscan
	reductions.
	(cp_parser_omp_scan_loop_body): New function.
	(cp_parser_omp_for_loop): Call cp_parser_omp_scan_loop_body if there
	are inscan reduction clauses.
	(cp_parser_pragma): Reject PRAGMA_OMP_SCAN.
	* semantics.c (finish_omp_clauses): Reject mixing inscan with
	non-inscan reductions on the same construct, or inscan reductions with
	ordered or schedule clauses, or inscan array reductions.
	* pt.c (tsubst_omp_clauses): Handle OMP_CLAUSE_{IN,EX}CLUSIVE.
	(tsubst_expr): Handle OMP_SCAN.
testsuite/
	* c-c++-common/gomp/scan-1.c: New test.
	* c-c++-common/gomp/scan-2.c: New test.
	* c-c++-common/gomp/scan-3.c: New test.
	* c-c++-common/gomp/scan-4.c: New test.

From-SVN: r272117
      Jakub Jelinek committed