platform/upstream/gcc.git
6 years agoFix ICE in find_taken_edge_computed_goto (PR 84136)
David Malcolm [Fri, 9 Feb 2018 01:07:11 +0000 (01:07 +0000)]
Fix ICE in find_taken_edge_computed_goto (PR 84136)

PR 84136 reports an ICE within sccvn_dom_walker when handling a
C/C++ source file that overuses the labels-as-values extension.
The code in question stores a jump label into a global, and then
jumps to it from another function, which ICEs after inlining:

void* a;

void foo() {
  if ((a = &&l))
      return;

  l:;
}

int main() {
  foo();
  goto *a;

  return 0;
}

This appears to be far beyond what we claim to support in this
extension - but we shouldn't ICE.

What's happening is that, after inlining, we have usage of a *copy*
of the label, which optimizes away the if-return logic, turning it
into an infinite loop.

On entry to the sccvn_dom_walker we have this gimple:

main ()
{
  void * a.0_1;

  <bb 2> [count: 0]:
  a = &l;

  <bb 3> [count: 0]:
l:
  a.0_1 = a;
  goto a.0_1;
}

and:
  edge taken = find_taken_edge (bb, vn_valueize (val));
reasonably valueizes the:
  goto a.0_1;
after the:
  a = &l;
  a.0_1 = a;
as if it were:
  goto *&l;

find_taken_edge_computed_goto then has:

2380   dest = label_to_block (val);
2381   if (dest)
2382     {
2383       e = find_edge (bb, dest);
2384       gcc_assert (e != NULL);
2385     }

which locates dest as a self-jump from block 3 back to itself.

However, the find_edge call returns NULL - it has a predecessor edge
from block 2, but no successor edges.

Hence the assertion fails and we ICE.

A successor edge from the computed goto could have been created by
make_edges if the label stmt had been in the function, but make_edges
only looks in the current function when handling computed gotos, and
the label only appeared after inlining.

The following patch removes the assertion, fixing the ICE.

gcc/testsuite/ChangeLog:
PR tree-optimization/84136
* gcc.c-torture/compile/pr84136.c: New test.

gcc/ChangeLog:
PR tree-optimization/84136
* tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
that the result of find_edge is non-NULL.

From-SVN: r257509

6 years agoDaily bump.
GCC Administrator [Fri, 9 Feb 2018 00:16:14 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r257508

6 years agore PR target/83008 ([performance] Is it better to avoid extra instructions in data...
Sergey Shalnov [Thu, 8 Feb 2018 22:31:15 +0000 (23:31 +0100)]
re PR target/83008 ([performance] Is it better to avoid extra instructions in data passing between loops?)

PR target/83008
* config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
storing integer register in SImode.  Fix cost of 256 and 512
byte aligned SSE register store.

* config/i386/i386.c (ix86_multiplication_cost): Fix
multiplication cost for TARGET_AVX512DQ.

testsuite/ChangeLog:

PR target/83008
* gcc.target/i386/pr83008.c: New test.

From-SVN: r257505

6 years agore PR target/81143 (New test case gcc.target/powerpc/pr79799-2.c fails on powerpc BE)
Peter Bergner [Thu, 8 Feb 2018 20:40:32 +0000 (14:40 -0600)]
re PR target/81143 (New test case gcc.target/powerpc/pr79799-2.c fails on powerpc BE)

PR target/81143
* gcc.target/powerpc/pr79799-2.c: Use __LITTLE_ENDIAN__.

From-SVN: r257504

6 years agoconstexpr.c (cxx_eval_component_reference): Use INDIRECT_REF_P.
Paolo Carlini [Thu, 8 Feb 2018 18:56:17 +0000 (18:56 +0000)]
constexpr.c (cxx_eval_component_reference): Use INDIRECT_REF_P.

2018-02-08  Paolo Carlini  <paolo.carlini@oracle.com>

* constexpr.c (cxx_eval_component_reference): Use INDIRECT_REF_P.
* lambda.c (build_capture_proxy): Likewise.
* search.c (field_access_p): Likewise.
* semantics.c (omp_clause_decl, omp_privatize_field,
finish_omp_clauses): Likewise.

From-SVN: r257503

6 years agore PR c++/83806 (Spurious -Wunused-but-set-parameter with nullptr)
Paolo Carlini [Thu, 8 Feb 2018 18:54:39 +0000 (18:54 +0000)]
re PR c++/83806 (Spurious -Wunused-but-set-parameter with nullptr)

/cp
2018-02-08  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/83806
* typeck.c (decay_conversion): Use mark_rvalue_use for the special
case of nullptr too.

/testsuite
2018-02-08  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/83806
* g++.dg/warn/Wunused-parm-11.C: New.

From-SVN: r257502

6 years agoMark previous change with:
Mike Stump [Thu, 8 Feb 2018 18:39:43 +0000 (18:39 +0000)]
Mark previous change with:

PR target/84113

From-SVN: r257500

6 years agore PR tree-optimization/84238 (ICE tree check: expected integer_cst, have plus_expr...
Marek Polacek [Thu, 8 Feb 2018 16:18:04 +0000 (16:18 +0000)]
re PR tree-optimization/84238 (ICE tree check: expected integer_cst, have plus_expr in to_wide, at tree.h:5527)

PR tree-optimization/84238
* tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
get_range_strlen.

* gcc.dg/Wstringop-overflow-3.c: New test.

From-SVN: r257497

6 years ago[C++ PATCH] initializer_list diagnostic
Nathan Sidwell [Thu, 8 Feb 2018 16:11:39 +0000 (16:11 +0000)]
[C++ PATCH] initializer_list diagnostic

https://gcc.gnu.org/ml/gcc-patches/2018-02/msg00434.html
* class.c (finish_struct): Fix std:initializer_list diagnostic
formatting.

* g++.dg/cpp0x/initlist93.C: Adjust diagnostic.

From-SVN: r257496

6 years agoruntime: get missing function name from symbol table
Ian Lance Taylor [Thu, 8 Feb 2018 15:37:43 +0000 (15:37 +0000)]
runtime: get missing function name from symbol table

    If we trace back through code that has no debug info, as when calling
    through C code compiled with -g0, we won't have a function name.
    Try to fetch the function name using the symbol table.

    Adding the test case revealed that gotest failed to use the gccgo tag
    when matching files, so add that.

    Reviewed-on: https://go-review.googlesource.com/92756

From-SVN: r257495

6 years agoMakefile.am (check-gccgo, check-gcc): Add options to pick up target libstdc++, to...
Ian Lance Taylor [Thu, 8 Feb 2018 15:34:42 +0000 (15:34 +0000)]
Makefile.am (check-gccgo, check-gcc): Add options to pick up target libstdc++, to permit tests that use C++.

* Makefile.am (check-gccgo, check-gcc): Add options to pick up
target libstdc++, to permit tests that use C++.
* Makefile.in: Rebuild.

From-SVN: r257494

6 years agolibgo: update to Go1.10rc2
Ian Lance Taylor [Thu, 8 Feb 2018 15:30:22 +0000 (15:30 +0000)]
libgo: update to Go1.10rc2

    Reviewed-on: https://go-review.googlesource.com/92736

From-SVN: r257493

6 years agoAnother fix for single-element permutes (PR 84265)
Richard Sandiford [Thu, 8 Feb 2018 15:17:20 +0000 (15:17 +0000)]
Another fix for single-element permutes (PR 84265)

PR83753 was about a case in which we ended up trying to "vectorise"
a group of loads ore stores using single-element vectors.  The problem
was that we were classifying the load or store as VMAT_CONTIGUOUS_PERMUTE
rather than VMAT_CONTIGUOUS, even though it doesn't make sense to permute
a single-element vector.

In that PR it was enough to change get_group_load_store_type,
because vectorisation ended up being unprofitable and so we didn't
take things further.  But when vectorisation is profitable, the same
fix is needed in vectorizable_load and vectorizable_store.

2018-02-08  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
PR tree-optimization/84265
* tree-vect-stmts.c (vectorizable_store): Don't treat
VMAT_CONTIGUOUS accesses as grouped.
(vectorizable_load): Likewise.

gcc/testsuite/
PR tree-optimization/84265
* gcc.dg/vect/pr84265.c: New test.

From-SVN: r257492

6 years agoUse nonzero bits to refine range in split_constant_offset (PR 81635)
Richard Sandiford [Thu, 8 Feb 2018 15:16:29 +0000 (15:16 +0000)]
Use nonzero bits to refine range in split_constant_offset (PR 81635)

This patch is part 2 of the fix for PR 81635.  It means that
split_constant_offset can handle loops like:

  for (unsigned int i = 0; i < n; i += 4)
    {
      a[i] = ...;
      a[i + 1] = ...;
    }

CCP records that "i" must have its low 2 bits clear, but we don't
include this information in the range of "i", which remains [0, +INF].
I tried making set_nonzero_bits update the range info in the same
way that set_range_info updates the nonzero bits, but it regressed
cases like vrp117.c and made some other tests worse.

vrp117.c has a multiplication by 10, so CCP can infer that the low bit
of the result is clear.  If we included that in the range, the range
would go from [-INF, +INF] to [-INF, not-quite-+INF].  However,
the multiplication is also known to overflow in all cases, so VRP
saturates the result to [INT_MAX, INT_MAX].  This obviously creates a
contradiction with the nonzero bits, and intersecting the new saturated
range with an existing not-quite-+INF range would make us drop to
VR_UNDEFINED.  We're prepared to fold a comparison with an [INT_MAX,
INT_MAX] value but not with a VR_UNDEFINED value.

The other problems were created when intersecting [-INF, not-quite-+INF]
with a useful VR_ANTI_RANGE like ~[-1, 1].  The intersection would
keep the former range rather than the latter.

The patch therefore keeps the adjustment local to split_constant_offset
for now, but adds a helper routine so that it's easy to move this later.

2018-02-08  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
PR tree-optimization/81635
* wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
* wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
(test_round_for_mask): New functions.
(wide_int_cc_tests): Call test_round_for_mask.
* tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
* tree-vrp.c (intersect_range_with_nonzero_bits): New function.
* tree-data-ref.c (split_constant_offset_1): Use it to refine the
range returned by get_range_info.

gcc/testsuite/
PR tree-optimization/81635
* gcc.dg/vect/bb-slp-pr81635-3.c: New test.
* gcc.dg/vect/bb-slp-pr81635-4.c: Likewise.

From-SVN: r257491

6 years agore PR ipa/81360 (ice in estimate_edge_growth, at ipa-inline.h:86)
Jan Hubicka [Thu, 8 Feb 2018 14:51:51 +0000 (15:51 +0100)]
re PR ipa/81360 (ice in estimate_edge_growth, at ipa-inline.h:86)

PR ipa/81360
* cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
* symtab.c: Include builtins.h
(symtab_node::output_to_lto_symbol_table_p): Move here
from lto-streamer-out.c:output_symbol_p.
* lto-streamer-out.c (write_symbol): Turn early exit to assert.
(output_symbol_p): Move all logic to symtab.c
(produce_symtab): Update.

* lto.c (unify_scc): Register prevailing trees, not trees to be freed.
(read_cgraph_and_symbols): Use
symtab_node::output_to_lto_symbol_table_p.

From-SVN: r257490

6 years agoS/390: Disable prediction of indirect branches
Andreas Krebbel [Thu, 8 Feb 2018 14:45:53 +0000 (14:45 +0000)]
S/390: Disable prediction of indirect branches

This patch implements GCC support for mitigating vulnerability
CVE-2017-5715 known as Spectre #2 on IBM Z.

In order to disable prediction of indirect branches the implementation
makes use of an IBM Z specific feature - the execute instruction.
Performing an indirect branch via execute prevents the branch from
being subject to dynamic branch prediction.

The implementation tries to stay close to the x86 solution regarding
user interface.

x86 style options supported (without thunk-inline):

-mindirect-branch=(keep|thunk|thunk-extern)
-mfunction-return=(keep|thunk|thunk-extern)

IBM Z specific options:

-mindirect-branch-jump=(keep|thunk|thunk-extern|thunk-inline)
-mindirect-branch-call=(keep|thunk|thunk-extern)
-mfunction-return-reg=(keep|thunk|thunk-extern)
-mfunction-return-mem=(keep|thunk|thunk-extern)

These options allow us to enable/disable the branch conversion at a
finer granularity.

-mindirect-branch sets the value of -mindirect-branch-jump and
 -mindirect-branch-call.

-mfunction-return sets the value of -mfunction-return-reg and
 -mfunction-return-mem.

All these options are supported on GCC command line as well as
function attributes.

'thunk' triggers the generation of out of line thunks (expolines) and
replaces the formerly indirect branch with a direct branch to the
thunk.  Depending on the -march= setting two different types of thunks
are generated.  With -march=z10 or higher exrl (execute relative long)
is being used while targeting older machines makes use of larl/ex
instead.  From a security perspective the exrl variant is preferable.

'thunk-extern' does the branch replacement like 'thunk' but does not
emit the thunks.

'thunk-inline' is only available for indirect jumps.  It should be used
in environments where correct CFI is important - known as user space.

Additionally the patch introduces the -mindirect-branch-table option
which generates tables pointing to the locations which have been
modified.  This is supposed to allow reverting the changes without
re-compilation in situations where it isn't required. The sections are
split up into one section per option.

gcc/ChangeLog:

2018-02-08  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>

* config/s390/s390-opts.h (enum indirect_branch): Define.
* config/s390/s390-protos.h (s390_return_addr_from_memory)
(s390_indirect_branch_via_thunk)
(s390_indirect_branch_via_inline_thunk): Add function prototypes.
(enum s390_indirect_branch_type): Define.
* config/s390/s390.c (struct s390_frame_layout, struct
machine_function): Remove.
(indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
(indirect_branch_table_label_no, indirect_branch_table_name):
Define variables.
(INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
(enum s390_indirect_branch_option): Define.
(s390_return_addr_from_memory): New function.
(s390_handle_string_attribute): New function.
(s390_attribute_table): Add new attribute handler.
(s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
(s390_indirect_branch_via_thunk): New function.
(s390_indirect_branch_via_inline_thunk): New function.
(s390_function_ok_for_sibcall): When jumping via thunk disallow
sibling call optimization for non z10 compiles.
(s390_emit_call): Force indirect branch target to be a single
register.  Add r1 clobber for non-z10 compiles.
(s390_emit_epilogue): Emit return jump via return_use expander.
(s390_reorg): Handle JUMP_INSNs as execute targets.
(s390_option_override_internal): Perform validity checks for the
new command line options.
(s390_indirect_branch_attrvalue): New function.
(s390_indirect_branch_settings): New function.
(s390_set_current_function): Invoke s390_indirect_branch_settings.
(s390_output_indirect_thunk_function):  New function.
(s390_code_end): Implement target hook.
(s390_case_values_threshold): Implement target hook.
(TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
macros.
* config/s390/s390.h (struct s390_frame_layout)
(struct machine_function): Move here from s390.c.
(TARGET_INDIRECT_BRANCH_NOBP_RET)
(TARGET_INDIRECT_BRANCH_NOBP_JUMP)
(TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
(TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
(TARGET_INDIRECT_BRANCH_NOBP_CALL)
(TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
(TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
(TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
(TARGET_INDIRECT_BRANCH_TABLE): Define macros.
* config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
(INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
(mnemonic attribute): Add values which aren't recognized
automatically.
("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
pattern for branch conversion.  Fix mnemonic attribute.
("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
indirect branch via thunk if requested.
("indirect_jump", "<code>"): Expand patterns for branch conversion.
("*indirect_jump"): Disable for branch conversion using out of
line thunks.
("indirect_jump_via_thunk<mode>_z10")
("indirect_jump_via_thunk<mode>")
("indirect_jump_via_inlinethunk<mode>_z10")
("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
("casesi_jump_via_inlinethunk<mode>_z10")
("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
("*indirect2_jump"): Disable for branch conversion.
("casesi_jump"): Turn into expander and expand patterns for branch
conversion.
("return_use"): New expander.
("*return"): Emit return via thunk and rename it to ...
("*return<mode>"): ... this one.
* config/s390/s390.opt: Add new options and and enum for the
option values.

gcc/testsuite/ChangeLog:

2018-02-08  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>

* gcc.target/s390/nobp-function-pointer-attr.c: New test.
* gcc.target/s390/nobp-function-pointer-nothunk.c: New test.
* gcc.target/s390/nobp-function-pointer-z10.c: New test.
* gcc.target/s390/nobp-function-pointer-z900.c: New test.
* gcc.target/s390/nobp-indirect-jump-attr.c: New test.
* gcc.target/s390/nobp-indirect-jump-inline-attr.c: New test.
* gcc.target/s390/nobp-indirect-jump-inline-z10.c: New test.
* gcc.target/s390/nobp-indirect-jump-inline-z900.c: New test.
* gcc.target/s390/nobp-indirect-jump-nothunk.c: New test.
* gcc.target/s390/nobp-indirect-jump-z10.c: New test.
* gcc.target/s390/nobp-indirect-jump-z900.c: New test.
* gcc.target/s390/nobp-return-attr-all.c: New test.
* gcc.target/s390/nobp-return-attr-neg.c: New test.
* gcc.target/s390/nobp-return-mem-attr.c: New test.
* gcc.target/s390/nobp-return-mem-nothunk.c: New test.
* gcc.target/s390/nobp-return-mem-z10.c: New test.
* gcc.target/s390/nobp-return-mem-z900.c: New test.
* gcc.target/s390/nobp-return-reg-attr.c: New test.
* gcc.target/s390/nobp-return-reg-mixed.c: New test.
* gcc.target/s390/nobp-return-reg-nothunk.c: New test.
* gcc.target/s390/nobp-return-reg-z10.c: New test.
* gcc.target/s390/nobp-return-reg-z900.c: New test.
* gcc.target/s390/nobp-table-jump-inline-z10.c: New test.
* gcc.target/s390/nobp-table-jump-inline-z900.c: New test.
* gcc.target/s390/nobp-table-jump-z10.c: New test.
* gcc.target/s390/nobp-table-jump-z900.c: New test.

From-SVN: r257489

6 years agoSimplify LRA lowpart subreg fix
Richard Sandiford [Thu, 8 Feb 2018 13:50:16 +0000 (13:50 +0000)]
Simplify LRA lowpart subreg fix

r257177 made the else arms equivalent to the if arms.

2018-02-08  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* lra-constraints.c (match_reload): Unconditionally use
gen_lowpart_SUBREG, rather than selecting between that
and equivalent gen_rtx_SUBREG code.

From-SVN: r257488

6 years agore PR middle-end/84233 (ICE (segfault) in gimple_assign_rhs_code)
Richard Biener [Thu, 8 Feb 2018 13:29:15 +0000 (13:29 +0000)]
re PR middle-end/84233 (ICE (segfault) in gimple_assign_rhs_code)

2018-02-08  Richard Biener  <rguenther@suse.de>

PR tree-optimization/84233
* tree-ssa-phiprop.c (propagate_with_phi): Use separate
changed flag instead of boguously re-using phi_inserted.

* g++.dg/torture/pr84233.C: New testcase.

From-SVN: r257486

6 years ago[hsa] Fix PR82416 testcase
Martin Jambor [Thu, 8 Feb 2018 13:06:26 +0000 (14:06 +0100)]
[hsa] Fix PR82416 testcase

2018-02-08  Martin Jambor  <mjambor@suse.cz>

* testsuite/libgomp.hsa.c/pr82416.c: Make the function with target
clonable.

From-SVN: r257485

6 years ago[hsa] Set program allocation for static local variables
Martin Jambor [Thu, 8 Feb 2018 13:03:52 +0000 (14:03 +0100)]
[hsa] Set program allocation for static local variables

2018-02-08  Martin Jambor  <mjambor@suse.cz>

* hsa-gen.c (get_symbol_for_decl): Set program allocation for
static local variables.

libgomp/
* testsuite/libgomp.hsa.c/staticvar.c: New test.

From-SVN: r257484

6 years agore PR target/84278 (claims initv4sfv2sf is available but inits through stack)
Richard Biener [Thu, 8 Feb 2018 12:53:19 +0000 (12:53 +0000)]
re PR target/84278 (claims initv4sfv2sf is available but inits through stack)

2018-02-08  Richard Biener  <rguenther@suse.de>

PR tree-optimization/84278
* tree-vect-stmts.c (vectorizable_store): When looking for
smaller vector types to perform grouped strided loads/stores
make sure the mode is supported by the target.
(vectorizable_load): Likewise.

* gcc.target/i386/pr84278.c: New testcase.

From-SVN: r257483

6 years ago[AArch64] Use more LDP/STP in shrinkwrapping
Wilco Dijkstra [Thu, 8 Feb 2018 12:32:51 +0000 (12:32 +0000)]
[AArch64] Use more LDP/STP in shrinkwrapping

The shrinkwrap optimization added in GCC 7 allows each callee-save to
be delayed and done only across blocks which need a particular callee-save.
Although this reduces unnecessary memory traffic on code paths that need
few callee-saves, it typically uses LDR/STR rather than LDP/STP.  This
means more memory accesses and increased codesize, ~1.0% on average.

To improve this, if a particular callee-save must be saved/restored, also
add the adjacent callee-save to allow use of LDP/STP.  This significantly
reduces codesize (for example gcc_r, povray_r, parest_r, xalancbmk_r are
1% smaller).  This is a simple fix which can be backported.  A more advanced
approach would scan blocks for pairs of callee-saves, but that requires a
full rewrite of all the callee-save code which is too late at this stage.

An example epilog in a shrinkwrapped function before:

ldp    x21, x22, [sp,#16]
ldr    x23, [sp,#32]
ldr    x24, [sp,#40]
ldp    x25, x26, [sp,#48]
ldr    x27, [sp,#64]
ldr    x28, [sp,#72]
ldr    x30, [sp,#80]
ldr    d8, [sp,#88]
ldp    x19, x20, [sp],#96
ret

And after this patch:

ldr    d8, [sp,#88]
ldp    x21, x22, [sp,#16]
ldp    x23, x24, [sp,#32]
ldp    x25, x26, [sp,#48]
ldp    x27, x28, [sp,#64]
ldr    x30, [sp,#80]
ldp    x19, x20, [sp],#96
ret

    gcc/
* config/aarch64/aarch64.c (aarch64_components_for_bb):
Increase LDP/STP opportunities by adding adjacent callee-saves.

From-SVN: r257482

6 years agoPR84068, PR83459: Fix sort order of SCHED_PRESSURE_MODEL
Wilco Dijkstra [Thu, 8 Feb 2018 12:29:28 +0000 (12:29 +0000)]
PR84068, PR83459: Fix sort order of SCHED_PRESSURE_MODEL

The comparison function for SCHED_PRESSURE_MODEL is incorrect.  If either
instruction is not in target_bb, the ordering is not well defined.
Since all instructions outside the target_bb get the highest model_index,
all we need to do is sort on model_index.  If the model_index is the same
we defer to RFS_DEP_COUNT and/or RFS_TIE.

    gcc/
PR rtl-optimization/84068
PR rtl-optimization/83459
* haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.

    gcc/testsuite
PR rtl-optimization/84068
PR rtl-optimization/83459
* gcc.dg/pr84068.c: New test.

From-SVN: r257481

6 years agore PR tree-optimization/84224 (ICE in execute, at gimple-ssa-warn-alloca.c:448)
Aldy Hernandez [Thu, 8 Feb 2018 11:16:25 +0000 (11:16 +0000)]
re PR tree-optimization/84224 (ICE in execute, at gimple-ssa-warn-alloca.c:448)

PR tree-optimization/84224
* gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
* calls.c (gimple_alloca_call_p): Only return TRUE when we have
non-zero arguments.

From-SVN: r257480

6 years agoslp-pr56812.cc: Allow either basic-block or loop vectorization to happen.
Richard Biener [Thu, 8 Feb 2018 10:52:00 +0000 (10:52 +0000)]
slp-pr56812.cc: Allow either basic-block or loop vectorization to happen.

2018-02-08  Richard Biener  <rguenther@suse.de>

* g++.dg/vect/slp-pr56812.cc: Allow either basic-block or
loop vectorization to happen.

From-SVN: r257479

6 years agore PR c++/83204 (c++ -std=c++14 ICE in maybe_undo_parenthesized_ref, at cp/semantics...
Paolo Carlini [Thu, 8 Feb 2018 09:06:33 +0000 (09:06 +0000)]
re PR c++/83204 (c++ -std=c++14 ICE in maybe_undo_parenthesized_ref, at cp/semantics.c:1694)

/cp
2018-02-08  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/83204
* pt.c (tsubst_copy_and_build): Use force_paren_expr for INDIRECT_REF.

/testsuite
2018-02-08  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/83204
* g++.dg/cpp0x/lambda/lambda-ice25.C: New.

From-SVN: r257478

6 years agoDaily bump.
GCC Administrator [Thu, 8 Feb 2018 00:16:20 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r257477

6 years agoinstall.texi (Configuration): Document the --with-long-double-format={ibm,ieee} Power...
Michael Meissner [Wed, 7 Feb 2018 23:58:28 +0000 (23:58 +0000)]
install.texi (Configuration): Document the --with-long-double-format={ibm,ieee} PowerPC configuration options.

2018-02-07  Michael Meissner  <meissner@linux.vnet.ibm.com>

* doc/install.texi (Configuration): Document the
--with-long-double-format={ibm,ieee} PowerPC configuration
options.

From-SVN: r257473

6 years agoaltivec.md (*restore_world): Remove LR use.
Iain Sandoe [Wed, 7 Feb 2018 23:42:51 +0000 (23:42 +0000)]
altivec.md (*restore_world): Remove LR use.

2018-02-07  Iain Sandoe  <iain@codesourcery.com>

* config/rs6000/altivec.md (*restore_world): Remove LR use.
* config/rs6000/predicates.md (restore_world_operation): Adjust op
count, remove one USE.

From-SVN: r257472

6 years agovsxcopy.c: Update scan-assembler stanzas.
Will Schmidt [Wed, 7 Feb 2018 23:02:00 +0000 (23:02 +0000)]
vsxcopy.c: Update scan-assembler stanzas.

[testsuite]

2018-02-07  Will Schmidt  <will_schmidt@vnet.ibm.com>

* gcc.target/powerpc/vsxcopy.c: Update scan-assembler stanzas.

From-SVN: r257471

6 years agore PR target/84154 (PowerPC GCC 7 and 8 have regression in converting fp to short...
Michael Meissner [Wed, 7 Feb 2018 22:54:59 +0000 (22:54 +0000)]
re PR target/84154 (PowerPC GCC 7 and 8 have regression in converting fp to short/char and returning it)

[gcc]
2018-02-07  Michael Meissner  <meissner@linux.vnet.ibm.com>

PR target/84154
* config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
Convert from define_expand to be define_insn_and_split.  Rework
float/double/_Float128 conversions to QI/HI/SImode to work with
both ISA 2.07 (power8) or ISA 3.0 (power9).  Fix regression where
conversions to QI/HImode types did a store and then a load to
truncate the value.  For conversions to VSX registers, don't split
the insn, instead emit the code directly.  Use the code iterator
any_fix to combine signed and unsigned conversions.
(fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
(fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
(fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
(fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
(fix_<mode>di2_hw): Likewise.
(fixuns_<mode>di2_hw): Likewise.
(fix_<mode>si2_hw): Likewise.
(fixuns_<mode>si2_hw): Likewise.
(fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
(fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
(fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
fix<uns>_trunc<SFDF:mode>si2_p8.
(fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer
used.
(fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
(fix<uns>_<mode>_mem): Likewise.
(fctiw<u>z_<mode>_mem): Likewise.
(fix<uns>_<mode>_mem): Likewise.
(fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
the register allocator from doing a direct move to the GPRs to do
a store, and instead use the ISA 3.0 store byte/half-word from
vector register instruction.  For IEEE 128-bit floating point,
also optimize stores of 32-bit ints.
(fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.

[gcc/testsuite]
2018-02-07  Michael Meissner  <meissner@linux.vnet.ibm.com>

PR target/84154
* gcc.target/powerpc/pr84154-1.c: New tests.
* gcc.target/powerpc/pr84154-2.c: Likewise.
* gcc.target/powerpc/pr84154-3.c: Likewise.

From-SVN: r257470

6 years agobuiltins-mergew-mergeow.c: Update dg-requires.
Will Schmidt [Wed, 7 Feb 2018 22:46:16 +0000 (22:46 +0000)]
builtins-mergew-mergeow.c: Update dg-requires.

[testsuite]

2018-02-07  Will Schmidt  <will_schmidt@vnet.ibm.com>

* gcc.target/powerpc/builtins-mergew-mergeow.c:  Update dg-requires.

From-SVN: r257469

6 years agore PR c++/84082 (ICE with broken template function definition)
Jakub Jelinek [Wed, 7 Feb 2018 22:30:51 +0000 (23:30 +0100)]
re PR c++/84082 (ICE with broken template function definition)

PR c++/84082
* parser.c (cp_parser_dot_deref_incomplete): New function.
(cp_parser_postfix_dot_deref_expression): Use it.

* g++.dg/template/incomplete11.C: New test.
* g++.dg/parse/crash67.C: Expect an incomplete type diagnostics too.

From-SVN: r257466

6 years agore PR fortran/82994 (ICE in gfc_match_deallocate, at fortran/match.c:4478)
Steven G. Kargl [Wed, 7 Feb 2018 22:29:22 +0000 (22:29 +0000)]
re PR fortran/82994 (ICE in gfc_match_deallocate, at fortran/match.c:4478)

2018-02-07  Steven G. Kargl  <kargl@gcc.gnu.org>

PR fortran/82994
* match.c (gfc_match_deallocate): Check for NULL pointer.

2018-02-07  Steven G. Kargl  <kargl@gcc.gnu.org>

PR fortran/82994
* gfortran.dg/deallocate_error_3.f90: New test.
* gfortran.dg/deallocate_error_4.f90: New test.

From-SVN: r257465

6 years ago* es.po: Update.
Joseph Myers [Wed, 7 Feb 2018 22:20:39 +0000 (22:20 +0000)]
* es.po: Update.

From-SVN: r257464

6 years agoruntime: don't call funcPC from a function
Ian Lance Taylor [Wed, 7 Feb 2018 22:04:55 +0000 (22:04 +0000)]
runtime: don't call funcPC from a function

    The escape analysis support is not yet good enough to avoid escaping
    the argument to funcPC.  This causes unnecessary and often harmful
    memory allocation.  E.g., (*cpuProfile).addExtra can be called from a
    signal handler, and it must not allocate memory.

    Move the calls to funcPC to use variables instead.  This was done in
    the original migration to using funcPC, but was not done for newer code.

    In one case, in signal handling code, use getSigtramp.

    Reviewed-on: https://go-review.googlesource.com/92735

From-SVN: r257463

6 years agore PR fortran/68560 (The test gfortran.dg/shape_8.f90 now fails when compiled with...
Thomas Koenig [Wed, 7 Feb 2018 21:08:51 +0000 (21:08 +0000)]
re PR fortran/68560 (The test gfortran.dg/shape_8.f90 now fails when compiled with -flto)

2018-02-07  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/68560
* trans-intrinsic.c (gfc_conv_intrinsic_shape): New function.
(gfc_conv_intrinsic_function): Call it.

2018-02-07  Thomas Koenig  <tkoenig@gcc.gnu.org>

PR fortran/68560
* gfortran.dg/shape_9.f90: New test.

From-SVN: r257462

6 years agore PR fortran/82049 (ICE with character(*),parameter array constructor)
Steven G. Kargl [Wed, 7 Feb 2018 20:43:33 +0000 (20:43 +0000)]
re PR fortran/82049 (ICE with character(*),parameter array constructor)

2018-02-06  Steven G. Kargl  <kargl@gcc.gnu.org>

PR fortran/82049
* match.c (gfc_match_type_spec): If the charlen is non-NULL, then
try to resolve it.  While here return early if possible.

2018-02-06  Steven G. Kargl  <kargl@gcc.gnu.org>

PR fortran/82049
* gfortran.dg/assumed_charlen_parameter.f90: New test.

From-SVN: r257459

6 years agoFix libgomp.oacc-c-c++-common/pr84217.c for C++
Rainer Orth [Wed, 7 Feb 2018 19:32:21 +0000 (19:32 +0000)]
Fix libgomp.oacc-c-c++-common/pr84217.c for C++

* testsuite/libgomp.oacc-c-c++-common/pr84217.c (abort)
[__cplusplus]: Declare extern "C".

From-SVN: r257457

6 years agoC++: avoid most reserved words as misspelling suggestions (PR c++/81610 and PR c...
David Malcolm [Wed, 7 Feb 2018 17:55:54 +0000 (17:55 +0000)]
C++: avoid most reserved words as misspelling suggestions (PR c++/81610 and PR c++/80567)

lookup_name_fuzzy can offer some reserved words as suggestions for
misspelled words, helping with "singed"/"signed" typos.

PR c++/81610 and PR c++/80567 report problems where the C++ frontend
suggested "if", "for" and "else" as corrections for misspelled variable
names.

The root cause is that in r247233
  ("Fix spelling suggestions for reserved words (PR c++/80177)")
I loosened the conditions on these reserved words, adding this condition:
   if (kind == FUZZY_LOOKUP_TYPENAME)
to the logic for rejecting words that don't start decl-specifiers, to
allow for "static_assert" to be offered.

This is too loose a condition: we don't want to suggest *any* reserved word
when we're in a context where we don't know we expect a typename.

For the kinds of error-recover situations where we're suggesting
spelling corrections we don't have much contextual information, so it
seems prudent to be stricter about which reserved words we offer
as spelling suggestions; I don't think it makes sense for us to
suggest e.g. "for".

This patch implements that by effectively reinstating the old logic,
but special-casing RID_STATIC_ASSERT, moving the logic to a new
subroutine (in case we want to allow for other special-cases).

I attempted to add suggestions for the various RID_*CAST, to cope
with e.g. "reinterptet_cast" (I can never type that correctly on the
first try), but the following '<' token confuses the error-recovery
enough that the suggestion code isn't triggered.

gcc/cp/ChangeLog:
PR c++/81610
PR c++/80567
* name-lookup.c (suggest_rid_p): New function.
(lookup_name_fuzzy): Replace enum-rid-filtering logic with call to
suggest_rid_p.

gcc/testsuite/ChangeLog:
PR c++/81610
PR c++/80567
* g++.dg/spellcheck-reswords.C: New test case.
* g++.dg/spellcheck-stdlib.C: Remove xfail from dg-bogus
suggestion of "if".

From-SVN: r257456

6 years agoSupport >26 operands in generation code.
Alan Hayward [Wed, 7 Feb 2018 16:25:28 +0000 (16:25 +0000)]
Support >26 operands in generation code.

2018-02-07  Alan Hayward  <alan.hayward@arm.com>

        * genextract.c (push_pathstr_operand): New function to support [a-zA-Z].
        (walk_rtx): Call push_pathstr_operand.
        (print_path): Support [a-zA-Z].

From-SVN: r257455

6 years agoPR c++/84182 - ICE with captured lambda
Jason Merrill [Wed, 7 Feb 2018 16:02:50 +0000 (11:02 -0500)]
PR c++/84182 - ICE with captured lambda

PR c++/84181
* pt.c (extract_locals_r, extract_local_specs): New.
(tsubst_pack_expansion): Use them.

From-SVN: r257454

6 years agore PR tree-optimization/84037 (Speed regression of polyhedron benchmark since r256644)
Richard Biener [Wed, 7 Feb 2018 15:46:17 +0000 (15:46 +0000)]
re PR tree-optimization/84037 (Speed regression of polyhedron benchmark since r256644)

2018-02-07  Richard Biener  <rguenther@suse.de>

PR tree-optimization/84037
* tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
(cse_and_gimplify_to_preheader): Declare.
(vect_get_place_in_interleaving_chain): Likewise.
* tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
ivexpr_map.
(_loop_vec_info::~_loop_vec_info): Delete it.
(cse_and_gimplify_to_preheader): New function.
* tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
* tree-vect-stmts.c (vectorizable_store): CSE base and steps.
(vectorizable_load): Likewise.  For grouped stores always base
the IV on the first element.
* tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
condition before gimplifying.

From-SVN: r257453

6 years agoRevert behavior to r251316.
Martin Liska [Wed, 7 Feb 2018 13:16:04 +0000 (14:16 +0100)]
Revert behavior to r251316.

2018-02-07  Martin Liska  <mliska@suse.cz>

PR c++/84059.
* class.c (add_method): Append argument value.
* cp-tree.h (maybe_version_functions): Add new argument.
* decl.c (decls_match): Call it if a declaration does not
have DECL_FUNCTION_VERSIONED.
(maybe_version_functions): record argument is added.
2018-02-07  Martin Liska  <mliska@suse.cz>

PR c++/84059.
* g++.dg/ext/mv26.C: New test.

From-SVN: r257451

6 years agotree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for *DIV_EXPR and ...
Jakub Jelinek [Wed, 7 Feb 2018 12:53:31 +0000 (13:53 +0100)]
tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for *DIV_EXPR and *MOD_EXPR.

* tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
*DIV_EXPR and *MOD_EXPR.

From-SVN: r257450

6 years ago[testsuite] Require alloca in gcc.dg/pr83844.c
Tom de Vries [Wed, 7 Feb 2018 11:35:30 +0000 (11:35 +0000)]
[testsuite] Require alloca in gcc.dg/pr83844.c

2018-02-07  Tom de Vries  <tom@codesourcery.com>

* gcc.dg/pr83844.c: Require effective target alloca.

From-SVN: r257447

6 years ago[testsuite] Require global_constructor in gcc.dg/torture/pr83055.c
Tom de Vries [Wed, 7 Feb 2018 11:35:18 +0000 (11:35 +0000)]
[testsuite] Require global_constructor in gcc.dg/torture/pr83055.c

2018-02-07  Tom de Vries  <tom@codesourcery.com>

* gcc.dg/torture/pr83055.c: Require effective target global_constructor.

From-SVN: r257446

6 years agoUse -fcf-protection=return in cet-intrin-4.c
H.J. Lu [Wed, 7 Feb 2018 10:49:53 +0000 (10:49 +0000)]
Use -fcf-protection=return in cet-intrin-4.c

Since -fcf-protection requires both -mshstk and -mibt, use
-fcf-protection=return with -mshstk in cet-intrin-4.c.

PR target/84243
* gcc.target/i386/cet-intrin-4.c (dg-options): Use
-fcf-protection=return.

From-SVN: r257445

6 years agoi386: Mask out the CF_SET bit for -fcf-protection check
H.J. Lu [Wed, 7 Feb 2018 10:48:39 +0000 (10:48 +0000)]
i386: Mask out the CF_SET bit for -fcf-protection check

Since ix86_option_override_internal sets the CF_SET bit in
flag_cf_protection and it can be called more than once via pragma,
we need to mask out the CF_SET bit when checking flag_cf_protection.

PR target/84248
* config/i386/i386.c (ix86_option_override_internal): Mask out
the CF_SET bit when checking -fcf-protection.

From-SVN: r257444

6 years ago[openacc] Fix diff_type in expand_oacc_collapse_init
Tom de Vries [Wed, 7 Feb 2018 10:37:55 +0000 (10:37 +0000)]
[openacc] Fix diff_type in expand_oacc_collapse_init

2018-02-07  Tom de Vries  <tom@codesourcery.com>

PR libgomp/84217
* omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
enough.

* c-c++-common/goacc/pr84217.c: New test.
* gfortran.dg/goacc/pr84217.f90: New test.

* testsuite/libgomp.oacc-c-c++-common/pr84217.c: New test.

From-SVN: r257443

6 years ago* lto.c (register_resolution): Remove forgotten sanity check.
Jan Hubicka [Wed, 7 Feb 2018 10:20:03 +0000 (11:20 +0100)]
* lto.c (register_resolution): Remove forgotten sanity check.

From-SVN: r257442

6 years agore PR tree-optimization/84204 ([graphite] ICE in set_codegen_error, at graphite-isl...
Richard Biener [Wed, 7 Feb 2018 10:14:25 +0000 (10:14 +0000)]
re PR tree-optimization/84204 ([graphite] ICE in set_codegen_error, at graphite-isl-ast-to-gimple.c:206)

2018-02-07  Richard Biener  <rguenther@suse.de>

PR tree-optimization/84204
* tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
this place.

* gcc.dg/graphite/pr84204.c: New testcase.

PR tree-optimization/84205
* graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
special-case isl_ast_op_zdiv_r.

* gcc.dg/graphite/pr84205.c: New testcase.

PR tree-optimization/84223
* graphite-scop-detection.c (gather_bbs::before_dom_children):
Only add conditions from within the region.
(gather_bbs::after_dom_children): Adjust.

* gfortran.dg/graphite/pr84223.f90: New testcase.

From-SVN: r257441

6 years agore PR target/84209 ([avr] Don't split SP in split2)
Georg-Johann Lay [Wed, 7 Feb 2018 09:59:52 +0000 (09:59 +0000)]
re PR target/84209 ([avr] Don't split SP in split2)

PR target/84209
* config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
* config/avr/avr.md: Only post-reload split REG-REG moves if
either register is REGERAL_REG_P.

From-SVN: r257440

6 years agore PR c++/71662 ([DR 1485] ICE on invalid C++11 code with unqualified name look up...
Paolo Carlini [Wed, 7 Feb 2018 09:35:10 +0000 (09:35 +0000)]
re PR c++/71662 ([DR 1485] ICE on invalid C++11 code with unqualified name look up: in tsubst_copy, at cp/pt.c:14010)

2018-02-07  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/71662
* g++.dg/cpp0x/scoped_enum7.C: New.

From-SVN: r257439

6 years ago[testsuite] Fix gcc.dg/cse_recip.c for AArch64 after r257181.
Christophe Lyon [Wed, 7 Feb 2018 09:12:48 +0000 (09:12 +0000)]
[testsuite] Fix gcc.dg/cse_recip.c for AArch64 after r257181.

2018-02-07  Christophe Lyon <christophe.lyon@linaro.org>

PR tree-optimization/83008
* gcc.dg/cse_recip.c: Add -fno-tree-slp-vectorize.

From-SVN: r257438

6 years agore PR tree-optimization/84235 (Miscompilation of floating point code by dom2)
Jakub Jelinek [Wed, 7 Feb 2018 08:29:58 +0000 (09:29 +0100)]
re PR tree-optimization/84235 (Miscompilation of floating point code by dom2)

PR tree-optimization/84235
* tree-ssa-scopedtables.c
(avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
if the subtraction is performed in floating point type where NaNs are
honored.  For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
build 1.  Formatting fix.

* gcc.c-torture/execute/ieee/pr84235.c: New test.

From-SVN: r257437

6 years agocompiler: make single Btype for methods table of identical interface type
Ian Lance Taylor [Wed, 7 Feb 2018 01:52:48 +0000 (01:52 +0000)]
compiler: make single Btype for methods table of identical interface type

    Normally we ensure to build a single Btype for identical types.
    We did not do this for methods table of identical interface
    types, however. If there are two identical interface type I, I2,
    they have the same Btype BI, but different Btypes for their
    methods tables, BM and BM2. From the backend's point of view
    only one of them is linked to BI. This can cause inconsitency
    in the backend's type system, like unresolved placeholder. This
    CL ensures we create a single Btype for methods table of
    identical interface type.

    Reviewed-on: https://go-review.googlesource.com/92436

From-SVN: r257436

6 years agoDaily bump.
GCC Administrator [Wed, 7 Feb 2018 00:16:28 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r257435

6 years agoDon't pass x86-only options on non-x86 targets in c-c++-common/fcf-protection-[67...
Rainer Orth [Tue, 6 Feb 2018 23:31:09 +0000 (23:31 +0000)]
Don't pass x86-only options on non-x86 targets in c-c++-common/fcf-protection-[67].c (PR testsuite/84243)

PR testsuite/84243
* c-c++-common/fcf-protection-6.c: Only pass -mshstk on x86
targets.
* c-c++-common/fcf-protection-7.c: Likewise for -mibt.

From-SVN: r257432

6 years agore PR target/84146 (ICE with -mcet in dwarf2out_var_location, involving sigsetjmp)
Jakub Jelinek [Tue, 6 Feb 2018 20:32:45 +0000 (21:32 +0100)]
re PR target/84146 (ICE with -mcet in dwarf2out_var_location, involving sigsetjmp)

PR target/84146
* config/i386/i386.c (rest_of_insert_endbranch): Only skip
NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
and skip it regardless of bb boundaries.  Use CALL_P macro,
don't test INSN_P (insn) together with CALL_P or JUMP_P check
unnecessarily, formatting fix.

* gcc.target/i386/pr84146.c: New test.

From-SVN: r257431

6 years ago2018-02-06 Michael Collison <michael.collison@arm.com>
Michael Collison [Tue, 6 Feb 2018 20:27:08 +0000 (20:27 +0000)]
2018-02-06  Michael Collison  <michael.collison@arm.com>

* config/arm/thumb2.md:
(*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
(*thumb_mov_notscc): Ditto.
* gcc.target/arm/pr7676.c: New testcase.

From-SVN: r257430

6 years agore PR target/84154 (PowerPC GCC 7 and 8 have regression in converting fp to short...
Michael Meissner [Tue, 6 Feb 2018 20:15:40 +0000 (20:15 +0000)]
re PR target/84154 (PowerPC GCC 7 and 8 have regression in converting fp to short/char and returning it)

2018-02-06  Michael Meissner  <meissner@linux.vnet.ibm.com>

PR target/84154
* config/rs6000/rs6000.md (su code attribute): Use "u" for
unsigned_fix, not "s".

From-SVN: r257429

6 years agoFix HAVE_GAS_CFI_DIRECTIVE for x86_64-pc-solaris2.*
Rainer Orth [Tue, 6 Feb 2018 18:33:19 +0000 (18:33 +0000)]
Fix HAVE_GAS_CFI_DIRECTIVE for x86_64-pc-solaris2.*

* configure.ac (gcc_fn_eh_frame_ro): New function.
(gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
correct .eh_frame permissions.
* configure: Regenerate.

From-SVN: r257424

6 years agoinvoke.texi: Add section for the PowerPC SPE backend.
Andrew Jenner [Tue, 6 Feb 2018 17:37:46 +0000 (17:37 +0000)]
invoke.texi: Add section for the PowerPC SPE backend.

       * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove irrelevant options.

From-SVN: r257422

6 years agore PR tree-optimization/84225 (ICE in operation_no_trapping_overflow, at tree.c:7206)
Aldy Hernandez [Tue, 6 Feb 2018 17:11:01 +0000 (17:11 +0000)]
re PR tree-optimization/84225 (ICE in operation_no_trapping_overflow, at tree.c:7206)

PR tree-optimization/84225
Add test for previous commit for PR84225.

From-SVN: r257420

6 years agors6000.c (rs6000_option_override_internal): Display warning message for -mno-speculat...
Bill Schmidt [Tue, 6 Feb 2018 16:55:01 +0000 (16:55 +0000)]
rs6000.c (rs6000_option_override_internal): Display warning message for -mno-speculate-indirect-jumps.

[gcc]

2018-02-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>

* config/rs6000/rs6000.c (rs6000_option_override_internal):
Display warning message for -mno-speculate-indirect-jumps.

[gcc/testsuite]

2018-02-06  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>

* gcc.target/powerpc/safe-indirect-jump-1.c: Detect deprecation
warning for -mno-speculate-indirect-jumps.
* gcc.target/powerpc/safe-indirect-jump-2.c: Likewise.
* gcc.target/powerpc/safe-indirect-jump-3.c: Likewise.
* gcc.target/powerpc/safe-indirect-jump-4.c: Likewise.
* gcc.target/powerpc/safe-indirect-jump-5.c: Likewise.
* gcc.target/powerpc/safe-indirect-jump-6.c: Likewise.
* gcc.target/powerpc/safe-indirect-jump-7.c: Likewise.

From-SVN: r257419

6 years agopowerpcspe.opt: (msimple-fpu, mfpu) Add Undocumented.
Andrew Jenner [Tue, 6 Feb 2018 16:10:43 +0000 (16:10 +0000)]
powerpcspe.opt: (msimple-fpu, mfpu) Add Undocumented.

       * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
       Undocumented.
       * config/powerpcspe/sysv4.opt (mbit-align): Likewise.

From-SVN: r257417

6 years agore PR tree-optimization/84225 (ICE in operation_no_trapping_overflow, at tree.c:7206)
Aldy Hernandez [Tue, 6 Feb 2018 15:44:51 +0000 (15:44 +0000)]
re PR tree-optimization/84225 (ICE in operation_no_trapping_overflow, at tree.c:7206)

PR tree-optimization/84225
* tree-eh.c (find_trapping_overflow): Only call
operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.

From-SVN: r257416

6 years agocompiler: avoid negative zero in float constants
Ian Lance Taylor [Tue, 6 Feb 2018 15:30:06 +0000 (15:30 +0000)]
compiler: avoid negative zero in float constants

    Check for negative numbers with very small magnitudes that will round
    to negative zero, and force them to positive zero instead.

    This implements the spec clarification in https://golang.org/cl/14727.
    The test is in https://golang.org/cl/91895.

    Fixes golang/go#12621

    Reviewed-on: https://go-review.googlesource.com/92175

From-SVN: r257415

6 years agoFix checking -mibt and -mshstk options for control flow protection
Igor Tsimbalist [Tue, 6 Feb 2018 15:25:31 +0000 (16:25 +0100)]
Fix checking -mibt and -mshstk options for control flow protection

PR target/84145
* config/i386/i386.c: Reimplement the check of possible options
-mibt/-mshstk conbination. Change error messages.
* doc/invoke.texi: Fix a typo: remove extra '='.
* c-c++-common/fcf-protection-1.c: Change a compared message.
* c-c++-common/fcf-protection-2.c: Likewise.
* c-c++-common/fcf-protection-3.c: Likewise.
* c-c++-common/fcf-protection-5.c: Likewise.
* c-c++-common/fcf-protection-6.c: New test.
* c-c++-common/fcf-protection-7.c: Likewise.

From-SVN: r257414

6 years agoruntime: correct runtime structfield type to match reflect
Ian Lance Taylor [Tue, 6 Feb 2018 15:18:50 +0000 (15:18 +0000)]
runtime: correct runtime structfield type to match reflect

    The offset field in structfield has changed to offsetAnon, and now
    requires a shift to get the actual offset value.

    Fixes golang/go#23391

    Reviewed-on: https://go-review.googlesource.com/92275

From-SVN: r257413

6 years agore PR lto/81004 (linking failed with -flto and static libboost_program_options)
Jan Hubicka [Tue, 6 Feb 2018 13:27:04 +0000 (14:27 +0100)]
re PR lto/81004 (linking failed with -flto and static libboost_program_options)

PR lto/81004
* lto.c: Include builtins.h
(register_resolution): Merge resolutions in case trees was
merged across units.
(lto_maybe_register_decl): Break out from ...
(lto_read_decls): ... here.
(unify_scc): Also register decls here.
(read_cgraph_and_symbols): Sanity check that all resolutions was
read.

From-SVN: r257412

6 years agore PR tree-optimization/84228 (Bogus -Wstringop-truncation warning with -g)
Marek Polacek [Tue, 6 Feb 2018 13:25:54 +0000 (13:25 +0000)]
re PR tree-optimization/84228 (Bogus -Wstringop-truncation warning with -g)

PR tree-optimization/84228
* tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.

* c-c++-common/Wstringop-truncation-3.c: New test.

From-SVN: r257411

6 years agore PR target/82641 (Unable to enable crc32 for a certain function with target attribu...
Tamar Christina [Tue, 6 Feb 2018 11:20:55 +0000 (11:20 +0000)]
re PR target/82641 (Unable to enable crc32 for a certain function with target attribute on ARM (aarch32))

2018-02-06  Tamar Christina  <tamar.christina@arm.com>

PR target/82641
* config/arm/arm.c (arm_print_asm_arch_directives): Record already
emitted arch directives.
* config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
__ARM_FEATURE_COPROC before changing architectures.

gcc/testsuite
2018-02-06  Tamar Christina  <tamar.christina@arm.com>

PR target/82641
* gcc.target/arm/pragma_arch_switch_2.c: New.

From-SVN: r257410

6 years agoAvoid cc1 SEGV in gcc.dg/rtl/x86_64/final.c (PR target/79975)
Rainer Orth [Tue, 6 Feb 2018 10:28:47 +0000 (10:28 +0000)]
Avoid cc1 SEGV in gcc.dg/rtl/x86_64/final.c (PR target/79975)

PR target/79975
* gcc.dg/rtl/x86_64/final.c: Add -fdwarf2-cfi-asm to dg-options.

From-SVN: r257408

6 years agoi386.c (print_reg): Fix typo.
Richard Biener [Tue, 6 Feb 2018 09:26:48 +0000 (09:26 +0000)]
i386.c (print_reg): Fix typo.

2018-02-06  Richard Biener  <rguenther@suse.de>

* config/i386/i386.c (print_reg): Fix typo.
(ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.

From-SVN: r257407

6 years agogcc-plugin.m4 (GCC_ENABLE_PLUGINS): Remove -q option passed to grep.
Eric Botcazou [Tue, 6 Feb 2018 08:30:53 +0000 (08:30 +0000)]
gcc-plugin.m4 (GCC_ENABLE_PLUGINS): Remove -q option passed to grep.

config/
* gcc-plugin.m4 (GCC_ENABLE_PLUGINS): Remove -q option passed to grep.
gcc/
* configure: Regenerate.

From-SVN: r257406

6 years agoAdd upstream svn rev for PR sanitizer/82825.
Rainer Orth [Tue, 6 Feb 2018 08:16:22 +0000 (08:16 +0000)]
Add upstream svn rev for PR sanitizer/82825.

From-SVN: r257405

6 years agoDaily bump.
GCC Administrator [Tue, 6 Feb 2018 00:16:13 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r257404

6 years ago* sv.po: Update.
Joseph Myers [Mon, 5 Feb 2018 23:47:14 +0000 (23:47 +0000)]
* sv.po: Update.

From-SVN: r257401

6 years agoPR tree-optimization/83369 - Missing diagnostics during inlining
Martin Sebor [Mon, 5 Feb 2018 22:45:04 +0000 (22:45 +0000)]
PR tree-optimization/83369 - Missing diagnostics during inlining

gcc/ChangeLog:

PR tree-optimization/83369
* tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
inlining context.

gcc/testsuite/ChangeLog:

PR tree-optimization/83369
gcc.dg/Wnonnull.c: New test.

From-SVN: r257400

6 years agoMake lto.exp work with Tcl 8.4
Richard Sandiford [Mon, 5 Feb 2018 21:34:46 +0000 (21:34 +0000)]
Make lto.exp work with Tcl 8.4

"dict" was added in Tcl 8.5, but until a couple of weeks ago the
testsuite had worked with 8.4.

This patch uses arrays instead, like we do for the caching in
target-supports.exp.  It is a bit uglier than using dicts was,
but hopefully not too bad...

2018-02-05  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/testsuite/
* lib/lto.exp (lto_handle_diagnostics): Remove messages_by_file
argument and use dg-messages-by-file instead.  Expect it to be
an array rather than a dict.
(lto-link-and-maybe-run): Remove messages_by_file argument and
use an upvar for dg-messages-by-file.  Update call to
lto_handle_diagnostics.
(lt-get-options): Treat dg-messages-by-file as an array
rather than a dict.
(lto-get-options-main): Likewise.  Set the entry rather than appending.
(lto-execute): Treat dg-messages-by-file as an array rather than
a dict.  Update call to lto-link-and-maybe-run.

From-SVN: r257397

6 years agocompiler: rollback "avoid negative zero in float constants"
Ian Lance Taylor [Mon, 5 Feb 2018 19:26:29 +0000 (19:26 +0000)]
compiler: rollback "avoid negative zero in float constants"

    It uses functions that are not available in MPFR 2.4.2, which is the
    current version supported by GCC.

    Original change description:

        compiler: avoid negative zero in float constants

        Check for negative numbers with very small magnitudes that will round
        to negative zero, and force them to positive zero instead.

        This implements the spec clarification in https://golang.org/cl/14727.
        The test is in https://golang.org/cl/91895.

        Fixes golang/go#12621

    Updates golang/go#12621

    Reviewed-on: https://go-review.googlesource.com/92055

From-SVN: r257393

6 years agocompiler: avoid negative zero in float constants
Ian Lance Taylor [Mon, 5 Feb 2018 15:28:59 +0000 (15:28 +0000)]
compiler: avoid negative zero in float constants

    Check for negative numbers with very small magnitudes that will round
    to negative zero, and force them to positive zero instead.

    This implements the spec clarification in https://golang.org/cl/14727.
    The test is in https://golang.org/cl/91895.

    Fixes golang/go#12621

    Reviewed-on: https://go-review.googlesource.com/91915

From-SVN: r257390

6 years agoclass.c: Remove unused global variables.
Marek Polacek [Mon, 5 Feb 2018 11:47:48 +0000 (11:47 +0000)]
class.c: Remove unused global variables.

* class.c: Remove unused global variables.
        (build_primary_vtable): Don't gather statistics.
(print_class_statistics): Remove.
* cp-tree.h (print_class_statistics): Remove.
* tree.c (cxx_print_statistics): Don't call print_class_statistics.

From-SVN: r257389

6 years agore PR c++/82782 (ICE: nested template alias and specialized template with auto templa...
Paolo Carlini [Mon, 5 Feb 2018 11:15:55 +0000 (11:15 +0000)]
re PR c++/82782 (ICE: nested template alias and specialized template with auto template parameter)

2018-02-05  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/82782
* g++.dg/cpp1z/inline-var4.C: New.

From-SVN: r257388

6 years agoCherry-pick libsanitizer pointer-pair tristate option.
Martin Liska [Mon, 5 Feb 2018 11:01:50 +0000 (12:01 +0100)]
Cherry-pick libsanitizer pointer-pair tristate option.

2018-02-05  Martin Liska  <mliska@suse.cz>

* doc/invoke.texi: Cherry-pick upstream r323995.
2018-02-05  Martin Liska  <mliska@suse.cz>

* c-c++-common/asan/pointer-compare-1.c: Adjust ASAN_OPTIONS
options.
* c-c++-common/asan/pointer-compare-2.c: Likewise.
* c-c++-common/asan/pointer-subtract-1.c: Likewise.
* c-c++-common/asan/pointer-subtract-2.c: Likewise.
* c-c++-common/asan/pointer-subtract-3.c: Likewise.
* c-c++-common/asan/pointer-subtract-4.c: Likewise.
* c-c++-common/asan/pointer-compare-3.c: New test.
2018-02-05  Martin Liska  <mliska@suse.cz>

* asan/asan_flags.inc: Cherry-pick upstream r323995.
* asan/asan_report.cc (CheckForInvalidPointerPair):
Cherry-pick upstream r323995.

From-SVN: r257387

6 years ago[AArch64] Remove SVE XFAILs
Richard Sandiford [Mon, 5 Feb 2018 10:48:49 +0000 (10:48 +0000)]
[AArch64] Remove SVE XFAILs

These tests started passing after r257293, which had the side-effect
of renumbering the SSA names and leaving the COND_EXPRs in their
natural order.

This does show a deeper underlying issue that code generation is too
sensitive to internal things like SSA_NAME versions, but it no longer
affects these particular tests (for now).

2018-02-05  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/testsuite/
* gcc.target/aarch64/sve/vcond_4.c: Remove XFAILs.
* gcc.target/aarch64/sve/vcond_5.c: Likewise.

From-SVN: r257386

6 years agoAdjust ira_init_register_move_cost comment
Richard Sandiford [Mon, 5 Feb 2018 10:47:56 +0000 (10:47 +0000)]
Adjust ira_init_register_move_cost comment

2018-02-05  Richard Sandiford  <richard.sandiford@linaro.org>

gcc/
* ira.c (ira_init_register_move_cost): Adjust comment.

From-SVN: r257385

6 years agoFix GCOV documentation (PR gcov-profile/84137).
Martin Liska [Mon, 5 Feb 2018 09:59:16 +0000 (10:59 +0100)]
Fix GCOV documentation (PR gcov-profile/84137).

2018-02-05  Martin Liska  <mliska@suse.cz>

PR gcov-profile/84137
* doc/gcov.texi: Fix typo in documentation.

From-SVN: r257384

6 years agoDocument --dynamic-list-data option for --coverage usage.
Martin Liska [Mon, 5 Feb 2018 09:19:18 +0000 (10:19 +0100)]
Document --dynamic-list-data option for --coverage usage.

2018-02-05  Martin Liska  <mliska@suse.cz>

PR gcov-profile/83879
* doc/gcov.texi: Document necessity of --dynamic-list-data when
using dlopen functionality.

From-SVN: r257383

6 years agoAdd missing mask[z]_range[_round]_s[d,s] intrinsics
Olga Makhotina [Mon, 5 Feb 2018 07:08:24 +0000 (07:08 +0000)]
Add missing mask[z]_range[_round]_s[d,s] intrinsics

gcc/
* config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
_mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
_mm_maskz_range_ss, _mm_mask_range_round_ss,
_mm_maskz_range_round_ss): New intrinsics.
(__builtin_ia32_rangesd128_round)
(__builtin_ia32_rangess128_round): Remove.
(__builtin_ia32_rangesd128_mask_round,
__builtin_ia32_rangess128_mask_round): New builtins.
* config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
__builtin_ia32_rangess128_round): Remove.
(__builtin_ia32_rangesd128_mask_round,
__builtin_ia32_rangess128_mask_round): New builtins.
* config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
(ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
"<round_saeonly_constraint>")): Changed to ...
((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
"<round_saeonly_scalar_constraint>")): ... this.
("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
%0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
%1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
%2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
gcc/testsuite
* gcc.target/i386/avx512dq-vrangesd-1.c (_mm_mask_range_sd,
_mm_maskz_range_sd, _mm_mask_range_round_sd,
_mm_maskz_range_round_sd): Test new intrinsics.
* gcc.target/i386/avx512dq-vrangesd-2.c (_mm_range_sd,
_mm_mask_range_sd, _mm_maskz_range_sd, _mm_range_round_sd,
_mm_mask_range_round_sd, _mm_maskz_range_round_sd): Test new intrinsics.
* gcc.target/i386/avx512dq-vrangess-1.c (_mm_mask_range_ss,
_mm_maskz_range_ss, _mm_mask_range_round_ss,
_mm_maskz_range_round_ss): Test new intrinsics.
* gcc.target/i386/avx512dq-vrangess-2.c (_mm_range_ss,
_mm_mask_range_ss, _mm_maskz_range_ss, _mm_range_round_ss,
_mm_mask_range_round_ss, _mm_maskz_range_round_ss): Test new intrinsics.
* gcc.target/i386/avx-1.c (__builtin_ia32_rangesd128_round,
__builtin_ia32_rangess128_round): Remove builtins.
(__builtin_ia32_rangesd128_mask_round,
__builtin_ia32_rangess128_mask_round): Test new builtins.
* gcc.target/i386/sse-13.c: Ditto.
* gcc.target/i386/sse-23.c: Ditto.

From-SVN: r257382

6 years agocompiler: update iota handling, fix using iota in array length
Ian Lance Taylor [Mon, 5 Feb 2018 01:57:42 +0000 (01:57 +0000)]
compiler: update iota handling, fix using iota in array length

    CL 71750 changed the definition of how iota works.  This patch updates
    gccgo for the new definition.

    We've been mishandling iota appearing in a type that appears in a
    const expression, as in `c = len([iota]int{})`.  Correct that by copying
    type expressions when we copy an expression.  For simplicity only copy
    when it can change the size of a type, as that is the only case where
    iota in a type can affect the value of a constant (I think).  This is
    still a bunch of changes, but almost all boilerplate.

    Fixes golang/go#22341

    Reviewed-on: https://go-review.googlesource.com/91475

From-SVN: r257379

6 years agocompiler: permit empty statements after fallthrough
Ian Lance Taylor [Mon, 5 Feb 2018 01:53:23 +0000 (01:53 +0000)]
compiler: permit empty statements after fallthrough

    The language spec permits empty statements after a fallthrough
    statement, so implement that.  Also give a better error message when a
    fallthrough statement is in the wrong place.  The test case for this
    is in the master repository, test/fixedbugs/issue14540.go, just not
    yet in the gccgo repository.

    Fixes golang/go#14538

    Reviewed-on: https://go-review.googlesource.com/91855

From-SVN: r257378

6 years agocompiler: in range, evaluate array if it has receives or calls
Ian Lance Taylor [Mon, 5 Feb 2018 01:50:22 +0000 (01:50 +0000)]
compiler: in range, evaluate array if it has receives or calls

    The last change was incomplete, in that it did not evaluate the array
    argument in some cases where it had to be evaluated.  This reuses the
    existing code for checking whether len/cap is constant.

    Also clean up the use of _ as the second variable in a for/range,
    which was previously inconsistent depending on whether the statement
    used = or :=.

    Updates golang/go#22313

    Reviewed-on: https://go-review.googlesource.com/91715

From-SVN: r257377

6 years agocompiler: give error for non-int arguments to make
Ian Lance Taylor [Mon, 5 Feb 2018 01:46:07 +0000 (01:46 +0000)]
compiler: give error for non-int arguments to make

    This implements a requirement of the language spec.

    While we're here fix the value returned by the type method of a
    builtin call expression to make, although this doesn't seem to make
    any difference anywhere since we lower this to a runtime call before
    the determine_types pass anyhow.

    There is already a test for this error in the master repository:
    test/fixedbugs/issue16949.go. It just hasn't made it into the gccgo
    testsuite yet.

    Fixes golang/go#16949

    Reviewed-on: https://go-review.googlesource.com/91697

From-SVN: r257376

6 years agocompiler: don't error for goto over type or const declaration
Ian Lance Taylor [Mon, 5 Feb 2018 01:43:24 +0000 (01:43 +0000)]
compiler: don't error for goto over type or const declaration

    We should only issue an error for a goto over a var declaration.

    The test case for this is already in the master repository, at
    test/fixedbugs/issue8042.go.  It just hasn't been copied into the
    gccgo repository yet.

    Fixes golang/go#19089

    Reviewed-on: https://go-review.googlesource.com/91696

From-SVN: r257375

6 years agocompiler: correct parse of parenthesized select case
Ian Lance Taylor [Mon, 5 Feb 2018 01:40:51 +0000 (01:40 +0000)]
compiler: correct parse of parenthesized select case

    We used to mishandle `select { case (<-c): }` and friends.

    The test case for this is https://golang.org/cl/91657.

    Fixes golang/go#20923

    Reviewed-on: https://go-review.googlesource.com/91695

From-SVN: r257374

6 years agocmd/go: don't lose last flag from _cgo_flags
Ian Lance Taylor [Mon, 5 Feb 2018 01:38:34 +0000 (01:38 +0000)]
cmd/go: don't lose last flag from _cgo_flags

    The quoting code that read _cgo_flags, currently only in the gccgo
    version of cmd/go, was losing the last flag read from the file.

    Fixes golang/go#23666

    Reviewed-on: https://go-review.googlesource.com/91655

From-SVN: r257373

6 years agoDaily bump.
GCC Administrator [Mon, 5 Feb 2018 00:16:12 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r257372