This avoids HWI -> unsigned truncation to end up with zero alignment
which set_ptr_info_alignment ICEs on.

2020-03-13  Richard Biener  <rguenther@suse.de>

	PR tree-optimization/94163
	* tree-ssa-pre.c (create_expression_by_pieces): Check
	whether alignment would be zero.

	PR lto/94157
	* lto-wrapper.c (run_gcc): Use concat for appending
	to collect_gcc_options.
	PR lto/94157
	* gcc.dg/lto/pr94157_0.c: New test.

This patch is to apply the same fix as r267528 to another similar case
bb-slp-over-widen-2.c which requires misaligned vector access.

gcc/testsuite/ChangeLog

    PR testsuite/93935
    * gcc.dg/vect/bb-slp-over-widen-2.c: Expect basic block vectorized
    messages only on vect_hw_misalign targets.

> I'm getting this ICE with -mabi=ilp32:
>
> during RTL pass: fwprop1
> /opt/gcc/gcc-20200312/gcc/testsuite/gcc.dg/pr94121.c: In function 'bar':
> /opt/gcc/gcc-20200312/gcc/testsuite/gcc.dg/pr94121.c:16:1: internal compiler error: in decompose, at rtl.h:2279

That is a preexisting issue, caused by another bug in the same function.
When mode is SImode and moffset is 0x80000000 (or anything else with the
bit 31 set), we need to sign-extend it.

2020-03-13  Jakub Jelinek  <jakub@redhat.com>

	PR target/94121
	* config/aarch64/aarch64.c (aarch64_add_offset_1): Use gen_int_mode
	instead of GEN_INT.

There is no need to set mode attribute to XImode nor V8DFmode since
ix86_output_ssemov can properly encode xmm16-xmm31 registers with and
without AVX512VL.

gcc/

	PR target/89229
	* config/i386/i386.c (ix86_output_ssemov): Handle MODE_DF.
	* config/i386/i386.md (*movdf_internal): Call ix86_output_ssemov
	for TYPE_SSEMOV.  Remove TARGET_AVX512F, TARGET_PREFER_AVX256,
	TARGET_AVX512VL and ext_sse_reg_operand check.

gcc/testsuite/

	PR target/89229
	* gcc.target/i386/pr89229-4a.c: New test.
	* gcc.target/i386/pr89229-4b.c: Likewise.
	* gcc.target/i386/pr89229-4c.c: Likewise.

-mlow-precision-div hard-coded the number of iterations to 2 for double
and 1 for float.  This patch adds a --param to control the number.

2020-03-13  Bu Le  <bule1@huawei.com>

gcc/
	PR target/94154
	* config/aarch64/aarch64.opt (-param=aarch64-float-recp-precision=)
	(-param=aarch64-double-recp-precision=): New options.
	* doc/invoke.texi: Document them.
	* config/aarch64/aarch64.c (aarch64_emit_approx_div): Use them
	instead of hard-coding the choice of 1 for float and 2 for double.

The issue is that relax_delay_slots can streamline the CFG in some cases,
in particular remove BARRIERs, but removing BARRIERs changes the way the
instructions are associated with (basic) blocks by the liveness analysis
code in resource.c (find_basic_block) and thus can cause entries in the
cache maintained by resource.c to become outdated, thus producing wrong
answers downstream.

The fix is to invalidate the cache entries affected by the removal of
BARRIERs in relax_delay_slots, i.e. for the instructions down to the
next BARRIER.

	PR rtl-optimization/94119
	* resource.h (clear_hashed_info_until_next_barrier): Declare.
	* resource.c (clear_hashed_info_until_next_barrier): New function.
	* reorg.c (add_to_delay_list): Fix formatting.
	(relax_delay_slots): Call clear_hashed_info_until_next_barrier on
	the next instruction after removing a BARRIER.

store_integral_bit_field is ready to handle BLKmode fields, there is
even a subtlety with their handling on big-endian targets, see e.g.
PR middle-end/50325, but not if they are unaligned, so the fix is
simply to call extract_bit_field for them in order to generate an
unaligned load.  As a bonus, this subsumes  the big-endian specific
path that was added under PR middle-end/50325.

	PR middle-end/92071
	* expmed.c (store_integral_bit_field): For fields larger than a
	word, call extract_bit_field on the value if the mode is BLKmode.
	Remove specific path for big-endian targets and tidy things up a
	little bit.

        PR rtl-optimization/90275
        * cse.c (cse_insn): Delete no-op register moves too.

        PR rtl-optimization/90275
        * gcc.c-torture/compile/pr90275.c: New test.

Support for the CPEN control register was removed in rev .50 of the RXv1 Instruction Set Architecture manual in Feb 2009.  This patch removes it from GCC.

	* config/rx/rx.md (CTRLREG_CPEN): Remove.
	* config/rx/rx.c (rx_print_operand): Remove CTRLREG_CPEN support.

2020-03-12  Jakub Jelinek  <jakub@redhat.com>

	* gcc_release (upload_files): Without -l, pass -m 755 to the mkdir
	command invoked through ssh.

scripts/update_web_docs_git -r 9.3.0 -d gcc-9.3.0
failed after the sourceware upgrade, there is no python-sphinx10 package and
python3-sphinx is new enough that the docs build succeeded.

2020-03-12  Jakub Jelinek  <jakub@redhat.com>

	* update_web_docs_git: Drop SPHINXBUILD=/usr/bin/sphinx-1.0-build.

VN currently replaces a load of a 16 byte entity 128 bits of precision
(TImode) with the result of a load of a 16 byte entity with 80 bits of
mode precision (XFmode).  That will go downhill since if the padding
bits are not actually filled with memory contents those bits are
missing.

2020-03-12  Richard Biener  <rguenther@suse.de>

	PR tree-optimization/94103
	* tree-ssa-sccvn.c (visit_reference_op_load): Avoid type
	punning when the mode precision is not sufficient.

	* gcc.target/i386/pr94103.c: New testcase.

This test fails in the Fedora RPM build (but not elsewhere, for unknown
reasons). The warning is correct, we're passing a null pointer.

	* testsuite/tr1/8_c_compatibility/cstdlib/functions.cc: Do not pass
	a null pointer to functions with nonnull(1) attribute.

There is no need to set mode attribute to XImode since ix86_output_ssemov
can properly encode xmm16-xmm31 registers with and without AVX512VL.

	PR target/89229
	* config/i386/i386.c (ix86_output_ssemov): Handle MODE_DI,
	MODE_V1DF and MODE_V2SF.
	* config/i386/mmx.md (MMXMODE:*mov<mode>_internal): Call
	ix86_output_ssemov for TYPE_SSEMOV.  Remove ext_sse_reg_operand
	check.

2020-03-12  Tobias Burnus  <tobias@codesourcery.com>

        PR middle-end/94120
        * openmp.c (gfc_match_oacc_declare): Accept function-result
        variables; reject variables declared in a different scoping unit.

2020-03-12  Tobias Burnus  <tobias@codesourcery.com>

        PR middle-end/94120
        * gfortran.dg/goacc/pr78260-2.f90: Correct scan-tree-dump-times.
        Extend test case to result variables.
        * gfortran.dg/goacc/declare-2.f95: Actually check module-declaration
        restriction of OpenACC.
        * gfortran.dg/goacc/declare-3.f95: Remove case where this
        restriction is violated.
        * gfortran.dg/goacc/pr94120-1.f90: New.
        * gfortran.dg/goacc/pr94120-2.f90: New.
        * gfortran.dg/goacc/pr94120-3.f90: New.

When looking into PR94134, I've noticed bugs in the
ASM_OUTPUT_ALIGNED_DECL_LOCAL documentation.  varasm.c has:
  #if defined ASM_OUTPUT_ALIGNED_DECL_LOCAL
    unsigned int align = symtab_node::get (decl)->definition_alignment ();
    ASM_OUTPUT_ALIGNED_DECL_LOCAL (asm_out_file, decl, name,
                                   size, align);
    return true;
  #elif defined ASM_OUTPUT_ALIGNED_LOCAL
    unsigned int align = symtab_node::get (decl)->definition_alignment ();
    ASM_OUTPUT_ALIGNED_LOCAL (asm_out_file, name, size, align);
    return true;
  #else
    ASM_OUTPUT_LOCAL (asm_out_file, name, size, rounded);
    return false;
  #endif
and the ASM_OUTPUT_ALIGNED_LOCAL documentation properly mentions:
Like @code{ASM_OUTPUT_LOCAL} and mentions the same macro in another place.
The ASM_OUTPUT_ALIGNED_DECL_LOCAL description mentions non-existing macros
ASM_OUTPUT_ALIGNED_DECL and ASM_OUTPUT_DECL instead of the right ones
ASM_OUTPUT_ALIGNED_LOCAL and ASM_OUTPUT_LOCAL.

2020-03-12  Jakub Jelinek  <jakub@redhat.com>

	* doc/tm.texi.in (ASM_OUTPUT_ALIGNED_DECL_LOCAL): Change
	ASM_OUTPUT_ALIGNED_DECL in description to ASM_OUTPUT_ALIGNED_LOCAL
	and ASM_OUTPUT_DECL to ASM_OUTPUT_LOCAL.
	* doc/tm.texi: Regenerated.

As the testcase shows, if DSE decides to head trim {mem{set,cpy,move},strncpy}
and the call has lhs, it is incorrect to leave the lhs as is, because it
will then point to the adjusted address (base + head_trim) instead of the
original base.
The following patch fixes that by dropping the lhs of the call and assigning
lhs the original base in a following statement.

2020-03-12  Jakub Jelinek  <jakub@redhat.com>

	PR tree-optimization/94130
	* tree-ssa-dse.c: Include gimplify.h.
	(increment_start_addr): If stmt has lhs, drop the lhs from call and
	set it after the call to the original value of the first argument.
	Formatting fixes.
	(decrement_count): Formatting fix.

	* gcc.c-torture/execute/pr94130.c: New test.

Isn't it wasteful to first copy perhaps a large constructor (recursively)
and then truncate it to very few elts (zero in this case)?

> We should certainly avoid copying if they're the same.  The code above for
> only copying the bits that aren't going to be thrown away seems pretty
> straightforward, might as well use it even if the savings aren't likely to
> be large.

Calling vec_safe_truncate with the same number of elts the vector already
has is a nop, so IMHO we just should make sure we only unshare if it
changed.

2020-03-12  Jakub Jelinek  <jakub@redhat.com>

	PR c++/94124
	* decl.c (reshape_init_array_1): Don't unshare constructor if there
	aren't any trailing zero elts, otherwise only unshare the first
	nelts.

This updates myself to the right place in MAINTAINERS.

gcc/ChangeLog

2020-03-11  Bin Bin Lv  <shlb@linux.ibm.com>

	* MAINTAINERS (Write After Approval): Update myself.

The source file rs6000.c was split up into several smaller source files
through commit 1acf0246.  However, variable "altivec_builtin_mask_for_load" and
"builtin_mode_to_type[MAX_MACHINE_MODE][2]" were marked with the wrong syntax
"GTY(([options])) type name", which led these two variables were not marked as
roots correctly and wrongly GCed.  And when "altivec_builtin_mask_for_load"
was wrongly GCed, the compiling for openJDK is failed with ICEs enabling
precompiled header under mcpu=power7.  So roots must be declared using one of
the following syntaxes: "extern GTY(([options])) type name;" and "static
GTY(([options])) type name;".

And the following patch adds variable "altivec_builtin_mask_for_load" and
"builtin_mode_to_type[MAX_MACHINE_MODE][2]" into the roots array.

Bootstrap and regression tests were done on powerpc64le-linux-gnu (LE) with no
regressions.

gcc/ChangeLog

2020-03-11  Bin Bin Lv  <shlb@linux.ibm.com>

	* config/rs6000/rs6000-internal.h (altivec_builtin_mask_for_load,
	builtin_mode_to_type): Remove the declaration.
	* config/rs6000/rs6000.h (altivec_builtin_mask_for_load,
	builtin_mode_to_type): Add an extern GTY(()) declaration.
	* config/rs6000/rs6000.c (altivec_builtin_mask_for_load,
	builtin_mode_to_type): Remove the GTY(()) declaration.

This adds myself to MAINTAINERS in the Write After Approval section.

gcc/ChangeLog

2020-03-11  Bin Bin Lv  <shlb@linux.ibm.com>

	* MAINTAINERS (Write After Approval): Add myself.

The test FAILs on 32-bit targets that don't have __int128 type.

2020-03-12  Jakub Jelinek  <jakub@redhat.com>

	PR c++/93907
	* g++.dg/cpp2a/concepts-using2.C (cc): Use long long instead of
	__int128 if __SIZEOF_INT128__ isn't defined.

The problem here was that we were checking satisfaction once with 'e', a
typedef of 'void', and another time with 'void' directly, and treated them
as different for hashing based on the assumption that
canonicalize_type_argument would have already removed a typedef that wasn't
a complex dependent alias.  But that wasn't happening here, so let's add a
call.

gcc/cp/ChangeLog
2020-03-11  Jason Merrill  <jason@redhat.com>

	PR c++/93907
	* constraint.cc (tsubst_parameter_mapping): Canonicalize type
	argument.

I got a report that building Chromium fails with the "modifying a const
object" error.  After some poking I realized it's a bug in GCC, not in
their codebase.

Much like with ARRAY_REFs, which can be const even though the array
itself isn't, COMPONENT_REFs can be const although neither the object
nor the field were declared const.  So let's dial down the checking.
Here the COMPONENT_REF was const because of the "const_cast<const U &>(m)"
thing -- cxx_eval_component_reference then builds a COMPONENT_REF with
TREE_TYPE (t).

While looking into this I noticed that we don't detect modifying a const
object in certain cases like in
<https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94074#c2>.  That's because
we never evaluate an X::X() CALL_EXPR -- there's none.  Fixed as per
Jason's suggestion by setting TREE_READONLY on a CONSTRUCTOR after
initialization in cxx_eval_store_expression.

2020-03-11  Marek Polacek  <polacek@redhat.com>
	    Jason Merrill  <jason@redhat.com>

	PR c++/94074 - wrong modifying const object error for COMPONENT_REF.
	* constexpr.c (cref_has_const_field): New function.
	(modifying_const_object_p): Consider a COMPONENT_REF
	const only if any of its fields are const.
	(cxx_eval_store_expression): Mark a CONSTRUCTOR of a const type
	as readonly after its initialization has been done.

	* g++.dg/cpp1y/constexpr-tracking-const17.C: New test.
	* g++.dg/cpp1y/constexpr-tracking-const18.C: New test.
	* g++.dg/cpp1y/constexpr-tracking-const19.C: New test.
	* g++.dg/cpp1y/constexpr-tracking-const20.C: New test.
	* g++.dg/cpp1y/constexpr-tracking-const21.C: New test.
	* g++.dg/cpp1y/constexpr-tracking-const22.C: New test.

This adds a tests that verifies taking the split_view of a non-forward range
works correctly.  Doing so revealed a typo in one of _OuterIter's constructors.

It also revealed that the default constructor of
__gnu_test::test_range::iterator misbehaves, because by delegating to
Iter<T>(nullptr, nullptr) we perform a null-pointer deref at runtime in
input_iterator_wrapper's constructor due to the ITERATOR_VERIFY check therein.
Instead of delegating to this constructor it seems we can just inherit the
protected default constructor, which does not contain this ITERATOR_VERIFY
check.

libstdc++-v3/ChangeLog:

	* include/std/ranges (split_view::_OuterIter::_OuterIter): Typo fix,
	'address' -> 'std::__addressof'.
	* testsuite/std/ranges/adaptors/split.cc: Test taking the split_view of
	a non-forward input_range.
	* testsuite/util/testsuite_iterators.h (output_iterator_wrapper): Make
	default constructor protected instead of deleted, like with
	input_iterator_wrapper.
	(test_range::iterator): Add comment explaining that this type is used
	only when the underlying wrapper is input_iterator_wrapper or
	output_iterator_wrapper.  Remove delegating defaulted constructor so
	that the inherited default constructor is used instead.

This patch fixes a bug introduced by my earlier patch (
https://gcc.gnu.org/pipermail/gcc-patches/2020-March/541680.html ).
It introduces a new scalar builtin type that was missing in the original
patch.

Bootstrapped cleanly on arm-none-linux-gnueabihf.
Tested for regression on arm-none-linux-gnueabihf. No regression from
before the original patch.
Tests that failed or became unsupported because of the original tests
now work as they did before it.

	* config/arm/arm-builtins.c
	(arm_init_simd_builtin_scalar_types): New.
	* config/arm/arm_neon.h (vld2_bf16): Used new builtin type.
	(vld2q_bf16): Used new builtin type.
	(vld3_bf16): Used new builtin type.
	(vld3q_bf16): Used new builtin type.
	(vld4_bf16): Used new builtin type.
	(vld4q_bf16): Used new builtin type.
	(vld2_dup_bf16): Used new builtin type.
	(vld2q_dup_bf16): Used new builtin type.
	(vld3_dup_bf16): Used new builtin type.
	(vld3q_dup_bf16): Used new builtin type.
	(vld4_dup_bf16): Used new builtin type.
	(vld4q_dup_bf16): Used new builtin type.

As mentioned in the PR, the generic code decides to put the a variable into
lcomm_section, which is a NOSWITCH section and thus the generic code doesn't
switch into a particular section before using
ASM_OUTPUT{_ALIGNED{,_DECL}_}_LOCAL, on many targets that results just in
.lcomm (or for non-local .comm) directives which don't need a switch to some
section, other targets put switch_to_section (bss_section) at the start of
that macro.
pdp11 doesn't do that (and doesn't have bss_section), and so emits the
lcomm/comm variables in whatever section is current (it has only .text/.data
and for DEC assembler rodata).

The following patch fixes that by putting it always into data section, and
additionally avoids emitting an empty line in the assembly for the lcomm
vars.

2020-03-11  Jakub Jelinek  <jakub@redhat.com>

	PR target/94134
	* config/pdp11/pdp11.c (pdp11_asm_output_var): Call switch_to_section
	at the start to switch to data section.  Don't print extra newline if
	.globl directive has not been emitted.

	* gcc.c-torture/execute/pr94134.c: New test.

After IRA changes, atomic version will use one more register, but
non-atomic still use 2 registers, however this testcase isn't testing for
atomic feature, so I decide change the testcase to always use COUNT++
to test.

ChangeLog

gcc/testsuite/

Kito Cheng  <kito.cheng@sifive.com>

	* gcc.target/riscv/interrupt-2.c: Update testcase and expected output.

This avoids breaking the old broken pointer offsetting via
(T)(ptr - ((T)0)->x) which should have used offsetof.  Breakage
was exposed by the introduction of POINTER_DIFF_EXPR and making
PTA not considering that producing a pointer.  The mitigation
for simple cases is to canonicalize

  _2 = _1 - 8B;
  o_9 = (struct obj *) _2;

to

  o_9 = &MEM[_1 + -8B];

eliding one statement and the offending pointer subtraction.

2020-03-11  Richard Biener  <rguenther@suse.de>

	* match.pd ((T *)(ptr - ptr-cst) -> &MEM[ptr + -ptr-cst]):
	New pattern.

	* gcc.dg/torture/20200311-1.c: New testcase.

When using `check-function-bodies`, the subroutine `parse_function_bodies` uses
the `fluff` regexp to remove uninteresting assembly lines.

Arm targets generate assembly with some lines prefixed by `@`, these lines are
left by this process.

As an example of some lines prefixed by `@': the assembly output from the
`stacktest1` function in "bfloat16_simd_3_1.c" is:

        .align  2
        .global stacktest1
        .arch armv8.2-a
        .syntax unified
        .arm
        .fpu neon-fp-armv8
        .type   stacktest1, %function
stacktest1:
        @ args = 0, pretend = 0, frame = 8
        @ frame_needed = 0, uses_anonymous_args = 0
        @ link register save eliminated.
        sub     sp, sp, #8
        add     r3, sp, #6
        vst1.16 {d0[0]}, [r3]
        vld1.16 {d0[0]}, [r3]
        add     sp, sp, #8
        @ sp needed
        bx      lr
        .size   stacktest1, .-stacktest1

It seems that previous uses of `check-function-bodies` in the arm backend have
avoided problems with such lines since they use the `...` regexp in each place
such fluff occurs.

I'm currently writing a patch that I'd like to match the entire function body,
so I'd like to remove such `@` lines automatically.

gcc/testsuite/ChangeLog:

2020-03-11  Matthew Malcomson  <matthew.malcomson@arm.com>

	* lib/scanasm.exp (parse_function_bodies): Lines starting with '@' also
	counted as fluff.

The issue is that tree_is_indexable doesn't return the same result for
a FIELD_DECL with QUAL_UNION_TYPE and the QUAL_UNION_TYPE, resulting
in two instances of the QUAL_UNION_TYPE in the bytecode.  The result
for the type is the correct one (false, since it is variably modified)
while the result for the field is falsely true because:

  else if (TREE_CODE (t) == FIELD_DECL
           && lto_variably_modified_type_p (DECL_CONTEXT (t)))
    return false;

is not satisfied.  The reason for this is that the DECL_QUALIFIER of
fields of a QUAL_UNION_TYPE depends on a discriminant in Ada, which
means that the size of the type does too (CONTAINS_PLACEHOLDER_P),
which in turn means that it is reset to a mere PLACEHOLDER_EXPR by
free_lang_data, which finally means that the size of DECL_CONTEXT is
too, so RETURN_TRUE_IF_VAR is false.

In other words, the CONTAINS_PLACEHOLDER_P property of the DECL_QUALIFIER
of fields of a QUAL_UNION_TYPE hides the variably_modified_type_p property
of  these fields, if you look from the outside.

	PR middle-end/93961
	* tree.c (variably_modified_type_p) <RECORD_TYPE>: Recurse into
	fields whose type is a qualified union.

If the type is derived in the current compilation unit, and Allocate
is not overridden on derivation (as is typically the case with
Root_Storage_Pool_With_Subpools), the entity for Allocate of the
derived type is an alias for System.Storage_Pools.Subpools.Allocate.

The main assertion in gnat_to_gnu_entity fails in this case, since
this is not a definition and Is_Public is false (since the entity
is nested in the same compilation unit).

2020-03-11  Richard Wai  <richard@annexi-strayline.com>

	* gcc-interface/decl.c (gnat_to_gnu_entity): Also test Is_Public on
	the Alias of the entitiy, if is present, in the main assertion.

abs_hwi asserts that the argument is not HOST_WIDE_INT_MIN and as the
(invalid) testcase shows, the function can be called with such an offset.
The following patch is IMHO minimal fix, absu_hwi unlike abs_hwi allows even
that value and will return (unsigned HOST_WIDE_INT) HOST_WIDE_INT_MIN
in that case.  The function then uses moffset in two spots which wouldn't
care if the value is (unsigned HOST_WIDE_INT) HOST_WIDE_INT_MIN or
HOST_WIDE_INT_MIN and wouldn't accept it (!moffset and
aarch64_uimm12_shift (moffset)), then in one spot where the signedness of
moffset does matter and using unsigned is the right thing -
moffset < 0x1000000 - and finally has code which will handle even this
value right; the assembler doesn't really care for DImode immediates if
        mov     x1, -9223372036854775808
or
        mov     x1, 9223372036854775808
is used and similarly it doesn't matter if we add or sub it in DImode.

2020-03-11  Jakub Jelinek  <jakub@redhat.com>

	PR target/94121
	* config/aarch64/aarch64.c (aarch64_add_offset_1): Use absu_hwi
	instead of abs_hwi, change moffset type to unsigned HOST_WIDE_INT.

	* gcc.dg/pr94121.c: New test.

Jeff has recently fixed dump_histogram_value to use std::abs instead of abs,
because on FreeBSD apparently the ::abs isn't overloaded and only has
int abs (int);
Seems on Solaris /usr/include/iso/stdlib_iso.h abs has:
int abs (int);
long abs (long);
overloads but already not
long long abs (long long);
and there is another abs use in get_nth_most_common_value, also on int64_t.
The long long std::abs (long long); overload is there only in C++11 and we
in GCC10 still support C++98.

Martin has said that a counter should never be INT64_MIN, so IMHO it is
better to use abs_hwi which will assert that.

2020-03-11  Jakub Jelinek  <jakub@redhat.com>

	PR bootstrap/93962
	* value-prof.c (dump_histogram_value): Use abs_hwi instead of
	std::abs.
	(get_nth_most_common_value): Use abs_hwi instead of abs.

As e.g. decimal_from_decnumber shows, the REAL_VALUE_TYPE representation
contains a decimal128 embedded in ->sig only if it is rvc_normal, for
other kinds like rvc_inf or rvc_nan, ->sig is ignored and everything is
contained in the REAL_VALUE_TYPE flags (cl, sign, signalling and decimal).
decimal_to_binary which is used when folding a decimal{32,64,128} constant
to a binary floating point type ignores this and thus folds infinities and
NaNs into +0.0.
The following patch fixes that by only doing that for rvc_normal.
Similarly to the binary to decimal folding, it goes through a string, in
order to e.g. deal with canonical NaN mantissas, or binary float formats
that don't support infinities and/or NaNs.

2020-03-11  Jakub Jelinek  <jakub@redhat.com>

	PR middle-end/94111
	* dfp.c (decimal_to_binary): Only use decimal128ToString if from->cl
	is rvc_normal, otherwise use real_to_decimal to print the number to
	string.

	* gcc.dg/dfp/pr94111.c: New test.

As the testcase shows, arithmetics that for -ftrapv would need multiple
basic blocks can show up not just in nb_bytes expressions where we
are calling rewrite_to_non_trapping_overflow for a while already,
but also in the pointer expression to the start of the region.
While the testcase covers just the first hunk and I've failed to create
a testcase for the latter, it is at least in theory possible too, so I've
adjusted that hunk too.

2020-03-11  Jakub Jelinek  <jakub@redhat.com>

	PR tree-optimization/94114
	* tree-loop-distribution.c (generate_memset_builtin): Call
	rewrite_to_non_trapping_overflow even on mem.
	(generate_memcpy_builtin): Call rewrite_to_non_trapping_overflow even
	on dest and src.

	* gcc.dg/pr94114.c: New test.

... a call to ranges::begin on an input range.

This implements LWG 3286.  The new wording for the single-argument constructor
for subrange is implemented by splitting the constructor into two delegating
constructors, one constrained by _S_store_size and the other by !_S_store_size.

Tested on x86_64-pc-linux-gnu, both added tests fail before the patch and pass
with the patch.

libstdc++-v3/ChangeLog:

	LWG 3286 ranges::size is not required to be valid after a call to
	ranges::begin on an input range
	* include/std/ranges (subrange::subrange): Split single-argument
	constructor into two, one constrained by _S_store_size and another by
	!_S_store_size.
	(take_view::begin): Call size() before calling ranges::begin(_M_base).
	* testsuite/std/ranges/adaptors/lwg3286.cc: New test.
	* testsuite/std/ranges/subrange/lwg3286.cc: New test.