2018-11-16 Martin Liska <mliska@suse.cz>
PR lto/88004
* lto-symtab.c (lto_symtab_merge_symbols_1): Do not call
lto_symtab_symbol_p as it does checking of transparent alias.
These needs to be also merged in the function.
From-SVN: r266207
Fixes rXsbg_mode_sXl test failures.
Combine used to give us
(set (reg:SI 65)
(ior:SI (lshiftrt:SI (reg:SI 3 %r3 [ bD.2238 ])
(const_int 2 [0x2]))
(reg:SI 2 %r2 [ aD.2237 ])))
but now we get
(set (reg:SI 65)
(ior:SI (subreg:SI (zero_extract:DI (reg:DI 69)
(const_int 32 [0x20])
(const_int 30 [0x1e])) 4)
(subreg:SI (reg:DI 68) 4)))
or
(set (reg:SI 65)
(ior:SI (subreg:SI (and:DI (lshiftrt:DI (reg:DI 69)
(const_int 2 [0x2]))
(const_int 4294967295 [0xffffffff])) 4)
(subreg:SI (reg:DI 68) 4)))
with an extra subreg, which appears because pseudos, unlike hard
registers, can be accessed only using their natural mode.
This patch adds a special case for that. Also, it performs r*sbg
bit index computations during gcc run, so that expectations do not
depend on which concrete pattern was matched.
gcc/ChangeLog:
2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
* config/s390/s390.md
(*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
assembler.
(*r<noxa>sbg_<mode>_sll): Likewise.
(*r<noxa>sbg_<mode>_srl): Likewise.
(*r<noxa>sbg_sidi_srl): New pattern.
gcc/testsuite/ChangeLog:
2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
* gcc.target/s390/md/rXsbg_mode_sXl.c: Do not use arithmetic in
r{o,x}sbg expectations.
* gcc.target/s390/risbg-ll-2.c: Likewise.
From-SVN: r266203
2018-11-16 Jerome Lambourg <lambourg@adacore.com>
* config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
Default to TARGET_VXWORKS_RTP.
(SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
of TARGET_VXWORKS_RTP.
* config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
* config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
From-SVN: r266201
2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
PR c++/25759
gcc/
* doc/extend.texi (Common Type Attributes): Make it explicit
that attribute "packed" can apply to C++ classes.
From-SVN: r266199
GCN vector sizes range between 64 and 512 bytes, none of which have
correspondingly sized integer modes. This breaks a number of assumptions
throughout the compiler, but I don't really want to create modes just for this
purpose.
Instead, this patch fixes up the cases that I've found, so far, such that the
compiler tries something else, or fails to optimize, rather than just ICE.
2018-11-15 Andrew Stubbs <ams@codesourcery.com>
Kwok Cheung Yeung <kcy@codesourcery.com>
gcc/
* tree-vect-stmts.c (vectorizable_store): Don't ICE when
int_mode_for_size fails.
(vectorizable_load): Likewise.
Co-Authored-By: Kwok Cheung Yeung <kcy@codesourcery.com>
From-SVN: r266190
As of r266186, auto_diagnostic_group is now user-visible
(via -fdiagnostics-format=json), so update the diagnostic
guidelines accordingly.
gcc/ChangeLog:
* doc/ux.texi (Group logically-related diagnostics): Move
discussion of auto_diagnostic_group into this new subsection.
Give an example of where this grouping is used.
From-SVN: r266187
2018-11-15 Richard Biener <rguenther@suse.de>
PR middle-end/88029
* gimple.c (gimple_call_flags): Union flags from decl, type
and call fntype.
* trans-mem.c (is_tm_pure_call): Simplify.
* gcc.dg/tree-ssa/pr88029.c: New testcase.
From-SVN: r266183
2018-11-15 Richard Biener <rguenther@suse.de>
PR tree-optimization/88031
* tree-vect-loop.c (vectorizable_reduction): Move check
for multiple types earlier so we get the expected dump.
Simplify calls to vectorizable_condition.
* tree-vect-stmts.h (vectorizable_condition): Update prototype.
* tree-vect-stmts.c (vectorizable_condition): Instead of
reduc_def and reduc_index take just a flag. Simplify
code-generation now that we can rely on the defs being set up.
(vectorizable_comparison): Remove unused argument.
* gcc.dg/pr88031.c: New testcase.
From-SVN: r266182
https://gcc.gnu.org/ml/gcc-patches/2018-11/msg01376.html
* cp-tree.h (enum special_function_kind): Reorder and comment.
* method.c (SFK_CTOR_P, SFK_DTOR_P, SFK_ASSIGN_P, SFK_COPY_P)
(SFK_MOVE_P): New predicates.
(walk_field_subobs, synthesized_method_base_walk): Drop
copy_arg_p, move_p, assign_p args. Use new SFK predicates. Order
parameters consistently.
(synthesized_method_walk): Drop ctor_p, copy_arg_p, move_p,
assign_p calculations. Use new SFK predicates. Adjust calls to
worker functions.
From-SVN: r266180
* gcc-interface/misc.c (gnat_init_gcc_eh): Do not override the switch
-fnon-call-exceptions passed on the command line in -gnatp mode.
From-SVN: r266176
2018-11-15 Richard Biener <rguenther@suse.de>
PR tree-optimization/88030
* tree-complex.c (need_eh_cleanup): New global.
(update_complex_assignment): Mark blocks that need EH update.
(expand_complex_comparison): Likewise.
(tree_lower_complex): Allocate and deallocate need_eh_cleanup,
perform EH cleanup and schedule CFG cleanup if that did anything.
* gcc.dg/tsan/pr88030.c: New testcase.
From-SVN: r266175
PR rtl-optimization/88018
* cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
on the presence of fallthru edge, rather than if it is a USE or not.
* g++.dg/tsan/pr88018.C: New test.
From-SVN: r266174
2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
PR lto/55102
PR lto/56700
gcc/
* doc/invoke.texi (Optimize Options): Remove bad example about
interaction between -flto and -O. Replace it with a note that
you need to compile with -O and not just link. Copy-edit -flto
discussion to reduce verbiage and improve flow.
From-SVN: r266168
A recent change caused a performance regression. This restores the
previous performance and adds a performance test.
* scripts/check_performance: Allow tests to choose a -std flag.
* src/c++17/memory_resource.cc (bitset::get_first_unset()): Use local
variables of the right types. Call update_next_word() unconditionally.
* testsuite/20_util/unsynchronized_pool_resource/cons.cc: New test.
* testsuite/performance/20_util/memory_resource/pools.cc: New test.
* testsuite/util/testsuite_performance.h (time_counter): Allow
timer to be restarted.
From-SVN: r266164
2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
PR middle-end/59658
gcc/
* doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
also suppress many optimizations. Alphabetize option lists for
-O1, -O2, and -Os. Add list of options disabled with -Og, and
correct documentation for those options to say that.
* opts.c (default_options_table): Sort table by level and option
name, to make it easier to correlate to the manual.
From-SVN: r266162
P1236R1 - Signed integers are two's complement
gcc/cp/
* constexpr.c (cxx_eval_check_shift_p): Disable the signed LSHIFT_EXPR
checks for c++2a.
gcc/c-family/
* c-warn.c (maybe_warn_shift_overflow): Don't warn for c++2a.
* c-ubsan.c (ubsan_instrument_shift): Make signed shifts
with in-range second operand well defined for -std=c++2a.
gcc/
* doc/invoke.texi (Wshift-overflow): Adjust documentation for
c++2a.
gcc/testsuite/
* g++.dg/cpp2a/constexpr-shift1.C: New test.
* g++.dg/warn/permissive-1.C (enum A, enum D): Don't expect
diagnostics here for c++2a.
* g++.dg/cpp0x/constexpr-shift1.C (fn3, i3, fn4, i4): Don't expect
diagnostics here for c++2a.
* g++.dg/cpp0x/constexpr-60049.C (f3, x3, y3): Likewise.
* g++.dg/ubsan/cxx11-shift-1.C (main): Add some further tests.
* g++.dg/ubsan/cxx11-shift-2.C (main): Likewise.
* g++.dg/ubsan/cxx2a-shift-1.C: New test.
* g++.dg/ubsan/cxx2a-shift-2.C: New test.
From-SVN: r266153
PR bootstrap/86739
* hash-map.h (hash_map::iterator::reference_pair): New class.
(hash_map::iterator::operator*): Return it rather than std::pair.
From-SVN: r266152
* optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
iff the operand is a constant.
* gcc.c-torture/compile/20181114.c: New test.
2018-11-14 Ilya Leoshkevich <iii@linux.ibm.com>
From-SVN: r266151
* scripts/create_testsuite_files: Add special_functions to the list
of directories to search. Add comment referring to conformance.exp.
* testsuite/libstdc++-dg/conformance.exp: Add comment referring
to create_testsuite_files.
From-SVN: r266146
mode
The emitted address is .long, not .quad, in that case.
gcc/testsuite/ChangeLog:
2018-11-14 Ilya Leoshkevich <iii@linux.ibm.com>
* gcc.target/s390/mrecord-mcount.c (profileme): Expect .long in
31-bit mode.
From-SVN: r266145
These tests rely on larl->movdi merge, which is not implemented for
31-bit mode.
gcc/testsuite/ChangeLog:
2018-11-14 Ilya Leoshkevich <iii@linux.ibm.com>
* gcc.target/s390/global-array-almost-huge-element.c: Run only
in 64-bit mode.
* gcc.target/s390/global-array-almost-negative-huge-element.c:
Likewise.
* gcc.target/s390/global-array-even-element.c: Likewise.
From-SVN: r266144
