* adaint.c: Test for __linux__ instead of linux and __sun__ instead
of sun. Add missing leading underscore to AIX. Remove #elif 0.
* adaint.h: Likewise.
* cio.c: Likewise.
* cstreams.c: Likewise.
* env.c: Likewise.
* gsocket.h: Likewise.
* init.c: Likewise. Test for __i386__ instead of i386.
* link.c: Likewise.
* s-oscons-tmplt.c: Likewise.
* sysdep.c: Likewise.
* terminals.c: Likewise. Use BSD symbol instead of FREEBSD.
* tracebak.c: Likewise. Test for __sparc__ instead of sparc.
From-SVN: r223890
If you create a toolchain with the target alphaev68-unknown-linux-gnu, gcc
will use the -mcpu=ev67 by default when compiling. Some software packages
(like gmp) will use this target info to decide that it may freely use
assembly code that targets ev67+. The trouble comes in when trying to
compile that pure assembly code.
While gcc will use -mcpu=ev67 just fine, it will invoke gas without an
-mev67 option, so the assembler will default to the lowest common
denominator -- ev4 in this case. Inline assembly in C code is normally
just peachy because gcc's assembler output will start with ".arch <cpu>"
and the assembler will accept that. But if the hand coded assembly code
lacks that .arch, you easily end up with errors like so:
opcode `cttz' not supported for target <all>
While the assembly code could/should be fixed to explicitly output the
.arch directive, I think it's reasonable to expect this to work:
echo 'cttz $20,$21' | gcc -x assembler -c - -o /dev/null -mcpu=ev67
This simple patch implements that, although I guess it is a bit redundant
in the default case where gcc outputs .arch. Perhaps that should all be
punted in favor of a specs-only approach. Considering gas respects .arch
in the code over the command line, it should also make things more natural.
The command line is processed in the standard/expected way -- gcc defaults
the -m option while user's custom -mcpu/-Wa,-m options come after, and the
guy writing the assembly code is free to use .arch to override everything
else.
From-SVN: r223888
* alias.c (alias_set_entry_d): Add is_pointer and has_pointer.
(alias_stats): Add num_universal.
(alias_set_subset_of): Special case pointers; be ready for NULL
children.
(alias_sets_conflict_p): Special case pointers; be ready for NULL
children.
(init_alias_set_entry): Break out from ...
(record_alias_subset): ... here; propagate new fields;
allocate children only when really needed.
(get_alias_set): Do less generous pointer globbing.
(dump_alias_stats_in_alias_c): Update statistics.
* gcc.dg/alias-8.c: Do not xfail.
* gcc.dg/pr62167.c: Prevent FRE.
* gcc.dg/alias-14.c: New testcase.
From-SVN: r223883
* config/rs6000/rs6000.c (split_stack_arg_pointer_used_p): Scan
correct block for use of r12.
(rs6000_expand_split_stack_prologue): Error on r29 asm global reg.
From-SVN: r223878
PR libstdc++/66327
* include/bits/stl_algobase.h (__equal<true>::equal): Do not call
memcmp with null pointers.
(__lexicographical_compare<true>::__lc): Do not call memcmp for empty
ranges.
From-SVN: r223865
PR tree-optimization/66142
* tree-if-conv.c (if_convertible_phi_p): Don't give up on
virtual phis that feed themselves.
* gcc.dg/vect/pr66142.c: New test.
From-SVN: r223863
2015-05-29 Richard Biener <rguenther@suse.de>
PR tree-optimization/66314
* tree-ssa-threadupdate.c (create_block_for_threading): Add
parameter that says which loop the new block belongs to.
(ssa_create_duplicates): Blocks duplicated for the threaded
path belong to the loop of the thread destination.
* gcc.dg/asan/pr66314.c: New testcase.
From-SVN: r223861
* config/msp430/t-msp430 (MULTILIB_OPTIONS): Add multilibs for the
different types of hardware multiply available on the MSP430.
(MULTILIB_DIRNAMES): Likewise.
(MULTILIB_EXCEPTIONS): Add exceptions for incompatible cpus and
hardware multiply types.
(MULTILIB_MATCHES): Add matches for known MCU names and hardware
multiply types.
From-SVN: r223860
2015-05-28 Andreas Tobler <andreast@gcc.gnu.org>
* configure.ac: Move the atoll check from AC_CHECK_FUNCS to
gcc_AC_CHECK_DECLS.
* configure: Regenerate.
From-SVN: r223847
2015-05-28 Andreas Tobler <andreast@gcc.gnu.org>
* configure.host: Add bits for FreeBSD amd64 and i386.
* configure.ac: Add signal handler for FreeBSD (amd64/i386)
* configure: Regenerate.
* include/freebsd-signal.h: New file.
From-SVN: r223845
* gcc-interface/utils.c (max_size) <tcc_binary>: Add special code to
deal with the subtraction of a "negative" value in an unsigned type.
From-SVN: r223837
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Subprogram_Type>: Do
not error out on a return type which has a size that overflows if the
return is done by invisible reference.
From-SVN: r223836
* gcc-interface/utils.c (gnat_pushdecl): Reunify the handling of array
and pointer types wrt DECL_ORIGINAL_TYPE and adjust left and right.
From-SVN: r223835
* gcc-interface/gigi.h (gnat_stabilize_reference): Adjust.
(rewrite_fn): Remove third parameter.
(type_is_padding_self_referential): New inline predicate.
(return_type_with_variable_size_p): Likewise.
* gcc-interface/decl.c (allocatable_size_p): More around.
(cannot_be_superflat_p): Rename into...
(cannot_be_superflat ): ...this.
(initial_value_needs_conversion): New predicate.
(gnat_to_gnu_entity): Invoke type_is_padding_self_referential,
initial_value_needs_conversion and adjust to above renaming.
For a renaming, force the materialization if the inner expression
is compound. Adjust calls to elaborate_reference and build a
compound expression if needed.
(struct er_dat): Add N field.
(elaborate_reference_1): Remove N parameter and adjust.
(elaborate_reference): Add INIT parameter and pass it in the call to
gnat_rewrite_reference. Adjust initial expression.
* gcc-interface/trans.c (Call_to_gnu): Treat renamings the same way as
regular object declarations when it comes to creating a temporary.
Adjust call to gnat_stabilize_reference and build a compound expression
if needed. Invoke return_type_with_variable_size_p.
(gnat_to_gnu): Invoke type_is_padding_self_referential. In case #4,
return a call to a function unmodified if it returns with variable size
and is also the initial expression in an object declaration.
* gcc-interface/utils2.c (build_binary_op) <INIT_EXPR>: Use the RHS'
type if it is a call to a function that returns with variable size.
(build_unary_op): Invoke type_is_padding_self_referential.
(gnat_stabilize_reference_1): Remove N parameter and adjust.
(gnat_stabilize_reference): Add INIT parameter and pass it in the call
to gnat_rewrite_reference.
(gnat_rewrite_reference): Remove N, add INIT parameter and adjust.
<COMPOUND_EXPR>: New case.
From-SVN: r223834
Define CPP_SPEC for nios2 linux targets so that -posix & -pthread work
like on all other linux targets.
Discovered via Gentoo bug https://bugs.gentoo.org/543114
From-SVN: r223823
Define CPP_SPEC for microblaze linux targets so that -posix & -pthread
work like on all other linux targets.
Discovered via Gentoo bug https://bugs.gentoo.org/543114
From-SVN: r223822
Define CPP_SPEC for parisc linux targets so that -posix & -pthread work
like on all other linux targets.
Discovered via Gentoo bug https://bugs.gentoo.org/543114
From-SVN: r223820
2015-05-28 Richard Biener <rguenther@suse.de>
* tree-vect-loop.c (vect_fixup_reduc_chain): New function.
(vect_fixup_scalar_cycles_with_patterns): Likewise.
(vect_analyze_loop_2): Call vect_fixup_scalar_cycles_with_patterns
after pattern recog.
(vect_create_epilog_for_reduction): Properly handle reductions
with patterns.
(vectorizable_reduction): Likewise.
* tree-vect-slp.c (vect_analyze_slp_instance): Properly mark
reduction chains.
(vect_get_constant_vectors): Create the correct number of
initial values for reductions.
(vect_schedule_slp_instance): Handle reduction chains that are
type changing properly.
* tree-vect-stmts.c (vect_analyze_stmt): Adjust.
* gcc.dg/vect/slp-reduc-sad.c: New testcase.
From-SVN: r223818
2015-05-28 Ed Schonberg <schonberg@adacore.com>
* sem_ch3.adb (Is_Visible_Component): Component is visible
in a derived type if inherited through an ancestor that has a
partial view of the original type holding the component, if the
full view of that original type is in scope.
* sem_util.ads (Get_Body_From_Stub): Works for all kinds of stubs.
2015-05-28 Bob Duff <duff@adacore.com>
* sem_util.adb (Requires_Transient_Scope): For definite untagged
subtypes, we should never have to use the secondary stack. This moves
toward that goal. But there are still cases that don't work.
Here, we move the check for Is_Definite first, but add a
special-purpose check for Has_Discrim_Dep_Array.
From-SVN: r223817
2015-05-28 Bob Duff <duff@adacore.com>
* sem_util.adb (Requires_Transient_Scope): Avoid returning
function results on the secondary stack in so many cases.
From-SVN: r223814
2015-05-28 Ed Schonberg <schonberg@adacore.com>
* sem_util.adb (Wrong_Type): In any instance, do not emit error
if type of expression is the partial view of the expected type.
2015-05-28 Ed Schonberg <schonberg@adacore.com>
* sem_res.adb (Resolve_Actuals): a) The replacement of formal
names in named associations only needs to be done within an
instance, on a call to a primitive of a formal derived type,
where the actual subprogram may have different formal names than
those of the primitive operation of the formal type.
b) Defaulted parameters must be taken into account when obtaining
the names of the formals of the actual subprogram being called.
From-SVN: r223804
* oacc-init.c (resolve_device): Add FAIL_IS_ERROR argument. Update
function comment. Only call gomp_fatal if new argument is true.
(acc_dev_num_out_of_range): New function.
(acc_init_1, acc_shutdown_1): Update call to resolve_device. Call
acc_dev_num_out_of_range as appropriate.
(acc_get_num_devices, acc_set_device_type, acc_get_device_type)
(acc_get_device_num, acc_set_device_num): Update calls to resolve_device.
* testsuite/libgomp.oacc-c-c++-common/lib-4.c: Update expected test
output.
From-SVN: r223803
PR libgomp/65742
gcc/
* builtins.c (expand_builtin_acc_on_device): Don't use open-coded
sequence for !ACCEL_COMPILER.
libgomp/
* oacc-init.c (plugin/plugin-host.h): Include.
(acc_on_device): Check whether we're in an offloaded region for
host_nonshm
plugin. Don't use __builtin_acc_on_device.
* plugin/plugin-host.c (GOMP_OFFLOAD_openacc_parallel): Set
nonshm_exec flag in thread-local data.
(GOMP_OFFLOAD_openacc_create_thread_data): Allocate thread-local
data for host_nonshm plugin.
(GOMP_OFFLOAD_openacc_destroy_thread_data): Free thread-local data
for host_nonshm plugin.
* plugin/plugin-host.h: New.
From-SVN: r223801
2015-05-28 Robert Dewar <dewar@adacore.com>
* sem_ch13.adb, sem_disp.ads: Minor reformatting.
2015-05-28 Ed Schonberg <schonberg@adacore.com>
* sem_util.adb (Set_Debug_Info_Needed): For a private type
whose full view is itself a derived private type, set flag on
underlying full view as well, for proper gdb display.
2015-05-28 Bob Duff <duff@adacore.com>
* exp_tss.ads: Minor comment fix.
* exp_ch3.adb (Build_Array_Init_Proc, Build_Record_Init_Proc):
Inline init_procs when the type has controlled parts. Remove
obsolete comments about those init_procs -- init_procs for
such types are no longer complex. A typical init_proc just
initializes the 'Tag field, and calls the parent init_proc
(e.g. for Limited_Controlled), which calls the grandparent
(for Root_Controlled), which does nothing. This all boils down
to one instruction when inlined.
* exp_ch7.adb (Create_Finalizer): Inline the finalizer.
2015-05-28 Ed Schonberg <schonberg@adacore.com>
* sem_ch4.adb (Analyze_Selected_Component): If the type to use
is a derived type and is a generic actual, the selected component
appears within an instance body, and the check over the type
has failed, examine ancestor types for the desired component.
(Find_Component_In_Instance): If record type is a derived type,
examine all ancestors in order to locate desired component.
From-SVN: r223800
* config/rx/rx.c (push_regs): New function. Extracts code from...
(rx_expand_prologue): ... here. Use push_regs to push even small
spans of registers.
(pop_regs): New function.
(rx_expand_epilogue): Use pop_regs to pop even small spans of
registers.
From-SVN: r223799
