glibc mallinfo is now deprecated resulting in make check-performance
failure. When glibc => 2.33 prefer mallinfo2.
libstdc++-v3/ChangeLog:
* testsuite/util/testsuite_performance.h (__gnu_test::MallocInfo): New.
(__gnu_test::malloc_info): New, replace mallinfo on current platform
supporting it and use mallinfo2 when glibc >= 2.33.
omp/ompx sentinels are for vendor extensions; as they might be required for
the correctness of the program, a warning should be printable. This patch
documents in the OpenMP 5.2 table the existing warnings, including the new
warning for for fixed source form Fortran.
PR fortran/106670
gcc/fortran/ChangeLog:
* scanner.cc (skip_fixed_omp_sentinel): Add -Wsurprising warning
for 'omx' sentinels with -fopenmp.
* invoke.texi (-Wsurprising): Document additional warning case.
libgomp/ChangeLog:
* libgomp.texi (OpenMP 5.2): Add comment to ompx/omx entry.
gcc/testsuite/ChangeLog:
* c-c++-common/gomp/ompx-1.c: New test.
* c-c++-common/gomp/ompx-2.c: New test.
* g++.dg/gomp/ompx-attrs-1.C: New test.
* gfortran.dg/gomp/ompx-1.f90: New test.
* gfortran.dg/gomp/omx-1.f: New test.
* gfortran.dg/gomp/omx-2.f: New test.
The check_trait_type function is used for a number of different type
traits that have different requirements on their arguments. For example,
__is_constructible allows arrays of unknown bound even if the array
element is an incomplete type, but __is_aggregate does not, it always
requires the array element type to be complete. Other traits have
different requirements again, e.g. __is_empty allows incomplete unions,
and arrays (of known or unknown bound) of incomplete types.
This alters the check_trait_type function to take an additional KIND
parameter which indicates which set of type trait requirements to check.
As noted in a comment, the requirements for __is_aggregate deviate from
the ones for std::is_aggregate in the standard. It's not necessary for
the elements of an array to be complete types, because arrays are always
aggregates.
The type_has_virtual_destructor change is needed to avoid an ICE.
Previously it could never be called for incomplete union types as they
were (incorrectly) rejected by check_trait_type.
This change causes some additional diagnostics in some libstdc++ tests,
where the front end was not previously complaining about invalid types
that the library assertions diagnosed. We should consider removing the
library assertions from traits where the front end implements the
correct checks now.
PR c++/106838
gcc/cp/ChangeLog:
* class.cc (type_has_virtual_destructor): Return false for
union types.
* semantics.cc (check_trait_type): Add KIND parameter to support
different sets of requirements.
(finish_trait_expr): Pass KIND argument for relevant traits.
gcc/ChangeLog:
* doc/extend.texi (Type Traits): Fix requirements. Document
__is_aggregate and __is_final.
gcc/testsuite/ChangeLog:
* g++.dg/ext/array4.C: Fix invalid use of __is_constructible.
* g++.dg/ext/unary_trait_incomplete.C: Fix tests for traits with
different requirements.
libstdc++-v3/ChangeLog:
* testsuite/20_util/is_complete_or_unbounded/memoization_neg.cc:
Prune additional errors from front-end.
* testsuite/20_util/is_move_constructible/incomplete_neg.cc:
Likewise.
* testsuite/20_util/is_nothrow_swappable/incomplete_neg.cc:
Likewise.
* testsuite/20_util/is_nothrow_swappable_with/incomplete_neg.cc:
Likewise.
* testsuite/20_util/is_swappable_with/incomplete_neg.cc:
Likewise.
This change implements P0528 which requires that padding bits not
participate in atomic compare exchange operations. All arguments to the
generic template are 'sanitized' by the __builtin_clear_padding intrinsic
before they are used in comparisons. This requires that any stores
also sanitize the incoming value.
Co-authored-by: Jakub Jelinek <jakub@redhat.com>
Co-authored-by: Jonathan Wakely <jwakely@redhat.com>
Signed-off-by: Thomas Rodgers <trodgers@redhat.com>
libstdc++-v3/ChangeLog:
* include/bits/atomic_base.h (__atomic_impl::__maybe_has_padding):
New function.
(__atomic_impl::clear_padding): Likewise.
(__atomic_impl::__compare_exchange): Likewise.
(__atomic_impl::compare_exchange_weak): Delegate to
__compare_exchange.
(__atomic_impl::compare_exchange_strong): Likewise.
* include/std/atomic (atomic<T>::atomic(T)): Clear padding when
possible in a constexpr function.
(atomic::store): Clear padding.
(atomic::exchange): Likewise.
(atomic::compare_exchange_weak): Use __compare_exchange.
(atomic::compare_exchange_strong): Likewise.
* testsuite/29_atomics/atomic/compare_exchange_padding.cc: New
test.
* testsuite/29_atomics/atomic_ref/compare_exchange_padding.cc:
New test.
The new proposed resolution for LWG 3629 says that std::error_code and
std::error_condition should only use ADL to find their customization
points. This means we need to use a poison pill to prevent lookup from
finding overloads in the enclosing namespaces.
We can also remove the forward declarations of std::make_error_code and
std::make_error_condition, because they aren't needed now. ADL can find
them anyway (when std is an associated namespace), and unqualified name
lookup will not (and should not) find them.
libstdc++-v3/ChangeLog:
* include/std/system_error (__adl_only::make_error_code): Add
deleted function.
(__adl_only::make_error_condition): Likewise.
(error_code::error_code(ErrorCodeEnum)): Add using-declaration
for deleted function.
(error_condition::error_condition(ErrorConditionEnum)):
Likewise.
* testsuite/19_diagnostics/error_code/cons/lwg3629.cc: New test.
* testsuite/19_diagnostics/error_condition/cons/lwg3629.cc: New test.
This patch adds support for reasoning about the inequality of two symbolic
values in the special case specifically suited for reasoning about
out-of-bounds past the end of the buffer. With this patch, the analyzer
catches off-by-one errors and more even when the offset and capacity is
symbolic.
Regrtested on Linux x86_64 and tested on coreutils, curl, httpd and
openssh as usual.
2022-09-07 Tim Lange <mail@tim-lange.me>
gcc/analyzer/ChangeLog:
PR analyzer/106625
* analyzer.h (region_offset): Eliminate m_is_symbolic member.
* region-model-impl-calls.cc (region_model::impl_call_realloc):
Refine implementation to be more precise.
* region-model.cc (class symbolic_past_the_end):
Abstract diagnostic class to complain about accesses past the end
with symbolic values.
(class symbolic_buffer_overflow):
Concrete diagnostic class to complain about buffer overflows with
symbolic values.
(class symbolic_buffer_overread):
Concrete diagnostic class to complain about buffer overreads with
symbolic values.
(region_model::check_symbolic_bounds): New function.
(maybe_get_integer_cst_tree): New helper function.
(region_model::check_region_bounds):
Add call to check_symbolic_bounds if offset is not concrete.
(region_model::eval_condition_without_cm):
Add support for EQ_EXPR and GT_EXPR with binaryop_svalues.
(is_positive_svalue): New hleper function.
(region_model::symbolic_greater_than):
New function to handle GT_EXPR comparisons with symbolic values.
(region_model::structural_equality): New function to compare
whether two svalues are structured the same, i.e. evaluate to
the same value.
(test_struct): Reflect changes to region::calc_offset.
(test_var): Likewise.
(test_array_2): Likewise and add selftest with symbolic i.
* region-model.h (class region_model): Add check_symbolic_bounds,
symbolic_greater_than and structural_equality.
* region.cc (region::get_offset):
Reflect changes to region::calc_offset.
(region::calc_offset):
Compute the symbolic offset if the offset is not concrete.
(region::get_relative_symbolic_offset): New function to return the
symbolic offset in bytes relative to its parent.
(field_region::get_relative_symbolic_offset): Likewise.
(element_region::get_relative_symbolic_offset): Likewise.
(offset_region::get_relative_symbolic_offset): Likewise.
(bit_range_region::get_relative_symbolic_offset): Likewise.
* region.h: Add get_relative_symbolic_offset.
* store.cc (binding_key::make):
Reflect changes to region::calc_offset.
(binding_map::apply_ctor_val_to_range): Likewise.
(binding_map::apply_ctor_pair_to_child_region): Likewise.
(binding_cluster::bind_compound_sval): Likewise.
(binding_cluster::get_any_binding): Likewise.
(binding_cluster::maybe_get_compound_binding): Likewise.
gcc/ChangeLog:
PR analyzer/106625
* doc/invoke.texi:
State that the checker also reasons about symbolic values.
gcc/testsuite/ChangeLog:
PR analyzer/106625
* gcc.dg/analyzer/data-model-1.c: Change expected result.
* gcc.dg/analyzer/out-of-bounds-5.c: New test.
* gcc.dg/analyzer/out-of-bounds-realloc-grow.c: New test.
* gcc.dg/analyzer/symbolic-gt-1.c: New test.
This patch considers the environment variable syntax extension for
device-specific variants of environment variables from OpenMP 5.1 (see
OpenMP 5.1 specification, p. 75 and p. 639). An environment variable (e.g.
OMP_NUM_TEAMS) can have different suffixes:
_DEV (e.g. OMP_NUM_TEAMS_DEV): affects all devices but not the host.
_DEV_<device> (e.g. OMP_NUM_TEAMS_DEV_42): affects only device with
number <device>.
no suffix (e.g. OMP_NUM_TEAMS): affects only the host.
In future OpenMP versions also suffix _ALL will be introduced (see discussion
https://github.com/OpenMP/spec/issues/3179). This is also considered in this
patch:
_ALL (e.g. OMP_NUM_TEAMS_ALL): affects all devices and the host.
The precedence is as follows (descending). For the host:
1. no suffix
2. _ALL
For devices:
1. _DEV_<device>
2. _DEV
3. _ALL
That means, _DEV_<device> is used whenever available. Otherwise _DEV is used if
available, and at last _ALL. If there is no value for any of the variable
variants, default values are used as already implemented before.
This patch concerns parsing (a), storing (b), output (c) and transmission to the
device (d):
(a) The actual number of devices and the numbering are not known when parsing
the environment variables. Thus all environment variables are iterated and
searched for device-specific ones.
(b) Only configured device-specific variables are stored. Thus, a linked list
is used.
(c) The output is done in omp_display_env (see specification p. 468f). Global
ICVs are tagged with [all], see https://github.com/OpenMP/spec/issues/3179.
ICVs which are not global but aren't handled device-specific yet are tagged
with [host]. omp_display_env outputs the initial values of the ICVs. That is
why a dedicated data structure is introduced for the inital values only
(gomp_initial_icv_list).
(d) Device-specific ICVs are transmitted to the device via GOMP_ADDITIONAL_ICVS.
libgomp/ChangeLog:
* config/gcn/icv-device.c (omp_get_default_device): Return device-
specific ICV.
(omp_get_max_teams): Added for GCN devices.
(omp_set_num_teams): Likewise.
(ialias): Likewise.
* config/nvptx/icv-device.c (omp_get_default_device): Return device-
specific ICV.
(omp_get_max_teams): Added for NVPTX devices.
(omp_set_num_teams): Likewise.
(ialias): Likewise.
* env.c (struct gomp_icv_list): New struct to store entries of initial
ICV values.
(struct gomp_offload_icv_list): New struct to store entries of device-
specific ICV values that are copied to the device and back.
(struct gomp_default_icv_values): New struct to store default values of
ICVs according to the OpenMP standard.
(parse_schedule): Generalized for different variants of OMP_SCHEDULE.
(print_env_var_error): Function that prints an error for invalid values
for ICVs.
(parse_unsigned_long_1): Removed getenv. Generalized.
(parse_unsigned_long): Likewise.
(parse_int_1): Likewise.
(parse_int): Likewise.
(parse_int_secure): Likewise.
(parse_unsigned_long_list): Likewise.
(parse_target_offload): Likewise.
(parse_bind_var): Likewise.
(parse_stacksize): Likewise.
(parse_boolean): Likewise.
(parse_wait_policy): Likewise.
(parse_allocator): Likewise.
(omp_display_env): Extended to output different variants of environment
variables.
(print_schedule): New helper function for omp_display_env which prints
the values of run_sched_var.
(print_proc_bind): New helper function for omp_display_env which prints
the values of proc_bind_var.
(enum gomp_parse_type): Collection of types used for parsing environment
variables.
(ENTRY): Preprocess string lengths of environment variables.
(OMP_VAR_CNT): Preprocess table size.
(OMP_HOST_VAR_CNT): Likewise.
(INT_MAX_STR_LEN): Constant for the maximal number of digits of a device
number.
(gomp_get_icv_flag): Returns if a flag for a particular ICV is set.
(gomp_set_icv_flag): Sets a flag for a particular ICV.
(print_device_specific_icvs): New helper function for omp_display_env to
print device specific ICV values.
(get_device_num): New helper function for parse_device_specific.
Extracts the device number from an environment variable name.
(get_icv_member_addr): Gets the memory address for a particular member
of an ICV struct.
(gomp_get_initial_icv_item): Get a list item of gomp_initial_icv_list.
(initialize_icvs): New function to initialize a gomp_initial_icvs
struct.
(add_initial_icv_to_list): Adds an ICV struct to gomp_initial_icv_list.
(startswith): Checks if a string starts with a given prefix.
(initialize_env): Extended to parse the new syntax of environment
variables.
* icv-device.c (omp_get_max_teams): Added.
(ialias): Likewise.
(omp_set_num_teams): Likewise.
* icv.c (omp_set_num_teams): Moved to icv-device.c.
(omp_get_max_teams): Likewise.
(ialias): Likewise.
* libgomp-plugin.h (GOMP_DEVICE_NUM_VAR): Removed.
(GOMP_ADDITIONAL_ICVS): New target-side struct that
holds the designated ICVs of the target device.
* libgomp.h (enum gomp_icvs): Collection of ICVs.
(enum gomp_device_num): Definition of device numbers for _ALL, _DEV, and
no suffix.
(enum gomp_env_suffix): Collection of possible suffixes of environment
variables.
(struct gomp_initial_icvs): Contains all ICVs for which we need to store
initial values.
(struct gomp_default_icv):New struct to hold ICVs for which we need
to store initial values.
(struct gomp_icv_list): Definition of a linked list that is used for
storing ICVs for the devices and also for _DEV, _ALL, and without
suffix.
(struct gomp_offload_icvs): New struct to hold ICVs that are copied to
a device.
(struct gomp_offload_icv_list): Definition of a linked list that holds
device-specific ICVs that are copied to devices.
(gomp_get_initial_icv_item): Get a list item of gomp_initial_icv_list.
(gomp_get_icv_flag): Returns if a flag for a particular ICV is set.
* libgomp.texi: Updated.
* plugin/plugin-gcn.c (GOMP_OFFLOAD_load_image): Extended to read
further ICVs from the offload image.
* plugin/plugin-nvptx.c (GOMP_OFFLOAD_load_image): Likewise.
* target.c (gomp_get_offload_icv_item): Get a list item of
gomp_offload_icv_list.
(get_gomp_offload_icvs): New. Returns the ICV values
depending on the device num and the variable hierarchy.
(gomp_load_image_to_device): Extended to copy further ICVs to a device.
* testsuite/libgomp.c-c++-common/icv-5.c: New test.
* testsuite/libgomp.c-c++-common/icv-6.c: New test.
* testsuite/libgomp.c-c++-common/icv-7.c: New test.
* testsuite/libgomp.c-c++-common/icv-8.c: New test.
* testsuite/libgomp.c-c++-common/omp-display-env-1.c: New test.
* testsuite/libgomp.c-c++-common/omp-display-env-2.c: New test.
Add #undef of MUSL_DYNAMIC_LINKER before #define, to satisfy build checks
when configured with --enable-werror-always.
gcc/ChangeLog:
* config/nios2/linux.h (MUSL_DYNAMIC_LINKER): Add #undef before #define.
CFG cleanup resets the control altering flag for noreturn functions
when they are ECF_LEAF (like __builtin_unreachable ()). The
.ABNORMAL_DISPATCHER call built during CFG construction is not
marked as control altering. Several passes inserting traps or
unreachables fail to set the flag. And more.
PR middle-end/106870
* gimple-harden-conditionals.cc (insert_check_and_trap):
Set the control-altering flag on the built IFN_TRAP.
* gimple.cc (gimple_build_builtin_unreachable): Likewise.
* tree-cfg.cc (handle_abnormal_edges): Set the control-altering
flag on the .ABNORMAL_DISPATCHER call.
* tree-cfgcleanup.cc (cleanup_call_ctrl_altering_flag): Avoid
resetting the control altering flag for ECF_NORETURN calls.
(cleanup_control_flow_bb): Set the control altering flag on
discovered noreturn calls.
* symtab-thunks.cc (expand_thunk): Set the control altering
flag for the noreturn tailcall case.
* tree-eh.cc (lower_resx): Likewisw for trap and unwind_resume
calls.
Here the use of 'value' from within an unevaluated context causes us
to overeagerly instantiate it, via maybe_instantiate_decl called from
mark_used, despite the use occurring in a context that doesn't require
a definition.
This seems to only affect constexpr variable specializations, though
we used to have the same issue for constexpr function specializations
until r6-1309-g81371eff9bc7ef made us delay their instantiation until
necessary during constexpr evaluation.
This patch expands upon the r6-1309 fix to make mark_used avoid
unnecessarily instantiating constexpr variable specializations too,
by pulling out from maybe_instantiate_decl the condition
(decl_maybe_constant_var_p (decl)
|| (TREE_CODE (decl) == FUNCTION_DECL
&& DECL_OMP_DECLARE_REDUCTION_P (decl))
|| undeduced_auto_decl (decl))
into each of its three callers (including mark_used), removing the
problematic first test from mark_used, and simplifying accordingly.
The net result is that only mark_used is changed because the other two
callers, resolve_address_of_overloaded_function and decl_constant_var_p,
already guard the call appropriately. (This relaxation of mark_used
seems to be safe because during constexpr evaluation we already take
care to instantiate a constexpr variable as necessary via
decl_constant_value etc).
PR c++/99130
gcc/cp/ChangeLog:
* decl2.cc (maybe_instantiate_decl): Adjust function comment.
Check VAR_OR_FUNCTION_DECL_P. Pull out the disjunction into ...
(mark_used): ... here, removing the decl_maybe_constant_var_p
part of it.
gcc/testsuite/ChangeLog:
* g++.dg/cpp0x/constexpr-decltype5.C: New test.
The function rebuild_location_adhoc_htab() was meant to reconstruct the
adhoc location hash map after restoring a line_maps instance from a
PCH. However, the function has never performed as intended because it
missed the last step of adding the data into the newly reconstructed hash
map. This patch fixes that.
It does not seem possible to construct a test case such that the current
incorrect behavior is observable as a compiler issue. It would be
observable, if it were possible for a precompiled header to contain an
adhoc location with a non-zero custom data pointer. But currently, such
data pointers are used only by the middle end to track inlining
information, and this happens later, too late to show up in a PCH.
I also noted that location_adhoc_data_update, which updates the hash map
pointers in a different scenario, was relying on undefined pointer
arithmetic behavior. I'm not aware of this having caused any issue in
practice, but in this patch I have also changed it to use defined pointer
operations instead.
libcpp/ChangeLog:
* line-map.cc (location_adhoc_data_update): Remove reliance on
undefined behavior.
(get_combined_adhoc_loc): Likewise.
(rebuild_location_adhoc_htab): Fix issue where the htab was not
properly updated.
This patch implements doacross(sink: omp_cur_iteration - 1) that the
previous patchset emitted a sorry on during omp expansion.
It can be implemented with existing library functions.
To recap, depend(source)/doacross(source:)/doacross(source:omp_cur_iteration)
is implemented calling GOMP_doacross_post or GOMP_doacross_ull_post,
called with an array of long or unsigned long long elements, one for
all collapsed loops together and one for each further ordered loop if any.
We initialize that array in each thread when grabbing further set of iterations
and update it at the end of loops, so that it represents the current iteration
(as 0 based counters). When the worksharing loop is created, we tell the
library through another similar array the counts (the loop needs to be
rectangular) in each dimension, first element is count of all logical iterations
in the collapsed loops.
depend(sink:v1 op N1, v2 op N2, ...) is then implemented by conditionally calling
GOMP_doacross_wait/GOMP_doacross_ull_wait. For N? of 0 there is no check,
otherwise if it wants to wait in a particular dimension for a previous iteration,
we check that the corresponding iterator isn't the first one (or first few),
where the previous iterator in that dimension would be out of range, and similarly
for checking of next iteration in a dimension that it isn't the last one (or last few)
where it would be similarly out of bounds. Then the collapsed loop counters are
folded into a single 0 based counter (first argument) and then other 0 based
iterations counters on what iteration it should wait for.
Now, doacross(sink: omp_cur_iteration - 1) is supposed to wait for the previous
logical iteration in the combined iteration space of all ordered loops.
For the very first iteration in that combined iteration space it does nothing,
there is no previous iteration. And similarly it does nothing if there
are more ordered loops than collapsed loop and it isn't the first logical
iteration of the combined loops inside of the collapsed loops, because as implemented
we know the previous iteration in that case is always executed by the same thread
as the current one.
In the implementation, we use the same value as is stored in the first element
of the array for GOMP_doacross_post/GOMP_doacross_ull_post, if that value is 0,
we do nothing. The rest is different based on if ordered argument is equal to
collapse or not. If it is, then we otherwise call
GOMP_doacross_wait/GOMP_doacross_ull_wait with a single argument, one less than
that counter we compare against 0.
If ordered argument is bigger than collapse, we add a per-thread boolean variable
.first.N, which we set to true at the start of the outermost ordered loop inside
of the collapsed set of loops and set to false at the end of the innermost
ordered loop. If .first.N is false, we don't do anything (we know the previous
iteration was handled by the current thread and by my reading of the spec we don't
need to emit even a memory barrier in that case, because it is just synchronization
with the same thread), otherwise we call GOMP_doacross_wait/GOMP_doacross_ull_wait
with the first argument one less than the counter we compare against 0, and then
one less than 2nd and following counts if iterations we pass to the workshare
initialization. If say .counts.N passed to the workshare initialization is
{ 256, 13, 5, 2 } for collapse(3) ordered(6) loop, then
GOMP_doacross_post/GOMP_doacross_ull_post is called with arguments equal to
.ordereda.N[0] - 1, 12, 4, 1.
2022-09-08 Jakub Jelinek <jakub@redhat.com>
gcc/
* omp-expand.cc (expand_omp_ordered_sink): Add CONT_BB argument.
Add doacross(sink:omp_cur_iteration-1) support.
(expand_omp_ordered_source_sink): Clear counts[fd->ordered + 1].
Adjust expand_omp_ordered_sink caller.
(expand_omp_for_ordered_loops): If counts[fd->ordered + 1] is
non-NULL, set that variable to true at the start of outermost
non-collapsed loop and set it to false at the end of innermost
ordered loop.
(expand_omp_for_generic): If fd->ordered, allocate
1 + (fd->ordered - fd->collapse) further elements in counts array.
Copy to counts + 2 + fd->ordered the counts of fd->collapse ..
fd->ordered - 1 loop if any.
gcc/testsuite/
* c-c++-common/gomp/doacross-7.c: New test.
libgomp/
* libgomp.texi (OpenMP 5.2): Mention that omp_cur_iteration is now
fully supported.
* testsuite/libgomp.c/doacross-4.c: New test.
* testsuite/libgomp.c/doacross-5.c: New test.
* testsuite/libgomp.c/doacross-6.c: New test.
* testsuite/libgomp.c/doacross-7.c: New test.
The following avoids adding fallthru edges to the control chain from
the post-dominator walk.
PR tree-optimization/106881
* gimple-predicate-analysis.cc (compute_control_dep_chain_pdom):
Add only non-fallthru edges and avoid the same set of edges
as the caller does.
* gcc.dg/uninit-pr106881.c: New testcase.
The tm.h header would pull in config/elfos.h, which defines
TARGET_D_MINFO_SECTION needed for the D module support in the front-end
to emit data to the correct section for the run-time library to pick up.
The removal of it in r13-2385 caused a stage2 bootstrap failure on all
Solaris targets.
The memmodel header has also been removed as it is no longer required
now tm_p.h is no longer used by these sources.
gcc/ChangeLog:
* config/darwin-d.cc: Include tm.h.
* config/dragonfly-d.cc: Likewise.
* config/freebsd-d.cc: Remove memmodel.h.
* config/glibc-d.cc: Likewise.
* config/netbsd-d.cc: Include tm.h.
* config/openbsd-d.cc: Likewise.
* config/sol2-d.cc: Likewise.
Several MVE builtins incorrectly use the same predicate/constraint
pair for several modes, which does not match the specification.
This patch uses the appropriate iterator instead.
2022-09-06 Christophe Lyon <christophe.lyon@arm.com>
gcc/
* config/arm/mve.md (mve_vqshluq_n_s<mode>): Use
MVE_pred/MVE_constraint instead of mve_imm_7/Ra.
(mve_vqshluq_m_n_s<mode>): Likewise.
(mve_vqrshrnbq_n_<supf><mode>): Use MVE_pred3/MVE_constraint3
instead of mve_imm_8/Rb.
(mve_vqrshrunbq_n_s<mode>): Likewise.
(mve_vqrshrntq_n_<supf><mode>): Likewise.
(mve_vqrshruntq_n_s<mode>): Likewise.
(mve_vrshrnbq_n_<supf><mode>): Likewise.
(mve_vrshrntq_n_<supf><mode>): Likewise.
(mve_vqrshrnbq_m_n_<supf><mode>): Likewise.
(mve_vqrshrntq_m_n_<supf><mode>): Likewise.
(mve_vrshrnbq_m_n_<supf><mode>): Likewise.
(mve_vrshrntq_m_n_<supf><mode>): Likewise.
(mve_vqrshrunbq_m_n_s<mode>): Likewise.
(mve_vsriq_n_<supf><mode): Use MVE_pred2/MVE_constraint2 instead
of mve_imm_selective_upto_8/Rg.
(mve_vsriq_m_n_<supf><mode>): Likewise.
Currently, these two splitters (touched in this patch) are using predicate
`int_reg_operand_not_pseudo`, then they work in split2 pass after RA in
most times, and can not run before RA.
It would not be a bad idea to allow these splitters before RA. Then more
passes (e.g. combine, sched...) could optimize the emitted instructions.
And if splitting before RA, for current constant splitter, we may have more
freedom to create pseduo to help to generate more parallel instructions.
For the example in the leading patch [PATCH 1/2]: pli+plit+rldimi would be
better than pli+sldi+paddi.
Test this patch with spec, we could see performance gain some times; while
the improvement is not stable and woud caused by the patch indirectly.
gcc/ChangeLog:
* config/rs6000/rs6000.md (splitter for set to and_mask constants):
Use int_reg_operand (instead of int_reg_operand_not_pseudo).
(splitter for multi-insn constant loads): Ditto.
This file is missing the GCC Runtime Library Exception text in the
licence header. That is unintentional, and it should have been present.
libstdc++-v3/ChangeLog:
* include/std/barrier: Add missing runtime exception.
Talking about the declarator form doesn't help when fixing that would get
you a different error about placeholders not being valid in a parameter.
This also adds a <> fixit, which isn't enough for most templates, but is a
start.
PR c++/106793
gcc/cp/ChangeLog:
* decl.cc (grokdeclarator): Improve placeholder diagnostics.
* parser.cc (cp_parser_type_id_1): Add fixit.
gcc/testsuite/ChangeLog:
* g++.dg/cpp23/auto-array2.C: Adjust.
* g++.dg/cpp1z/class-deduction113.C: New test.
In schedule_region(), a basic block that does not contain any real insns
is not scheduled and the dfa state at the entry of the bb is not copied
to the fallthru basic block. However a DEBUG insn is treated as a real
insn, and if a bb contains non-real insns and a DEBUG insn, it's dfa
state is copied to the fallthru bb. This was resulting in
-fcompare-debug failure as the incoming dfa state of the fallthru block
is different with -g. We should always copy the dfa state of a bb to
it's fallthru bb even if the bb does not contain real insns.
2022-08-22 Surya Kumari Jangala <jskumari@linux.ibm.com>
gcc/
PR rtl-optimization/105586
* sched-rgn.cc (save_state_for_fallthru_edge): New function.
(schedule_region): Use it for all blocks.
gcc/testsuite/
PR rtl-optimization/105586
* gcc.target/powerpc/pr105586.c: New test.
In the edge case of a coroutine not containing any locals, the ifcd/switch
temporaries would get added to the coroutine frame, corrupting its
layout. To prevent this, we can make sure there is always a BIND_EXPR at
the top of the function body, and thus, always a place for our new
temporaries to go without interfering with the coroutine frame.
PR c++/106188 - Incorrect frame layout after transforming conditional statement without top-level bind expression
PR c++/106713 - if (co_await ...) crashes with a jump to ud2
PR c++/106188
PR c++/106713
gcc/cp/ChangeLog:
* coroutines.cc (coro_rewrite_function_body): Ensure we have a
BIND_EXPR wrapping the function body.
gcc/testsuite/ChangeLog:
* g++.dg/coroutines/pr106188.C: New test.
Signed-off-by: Arsen Arsenović <arsen@aarsen.me>
Instead of defining is_reference in terms of is_[lr]value_reference,
just define it directly.
libstdc++-v3/ChangeLog:
* include/std/type_traits (is_reference): Make the primary
template derive from false_type. Define two partial
specializations that derive from true_type.
Instead of defining these in terms of a helper class template
and the relatively expensive __remove_cv_t, just declare four
explicit specializations of the main template, one for each choice
of cv-quals.
libstdc++-v3/ChangeLog:
* include/std/type_traits (__is_void_helper): Remove.
(is_void): Make the primary template derive from false_type,
and define four explicit specializations that derive from
true_type.
(__is_null_pointer_helper, is_null_pointer): Likewise.
This change speeds up the simple benchmark below by about 40%.
struct A { virtual ~A() {} };
struct B: A { } b;
A* ap = &b;
void *sink;
int main()
{
for (long i = 0; i < 4000000000L; ++i)
sink = dynamic_cast<B*>(ap);
}
libstdc++-v3/ChangeLog:
* libsupc++/dyncast.cc (__dynamic_cast): Avoid virtual function
call in simple success case.
C2x follows C++ in making alignas, alignof, bool, false,
static_assert, thread_local and true keywords; implement this
accordingly. This implementation makes them normal keywords in C2x
mode just like any other keyword (C2x leaves open the possibility of
implementation using predefined macros instead - thus, there aren't
any testcases asserting that they aren't macros). As in C++ and
previous versions of C, true and false are handled like signed 1 and 0
in #if (there was an intermediate state in some C2x drafts where they
had different macro expansions that were unsigned in #if).
Bootstrapped with no regressions for x86_64-pc-linux-gnu.
As with the removal of unprototyped functions, this change has a high
risk of breaking some old code and people doing GNU/Linux distribution
builds may wish to see how much is broken in a build with a -std=gnu2x
default.
gcc/
* ginclude/stdalign.h [defined __STDC_VERSION__ &&
__STDC_VERSION__ > 201710L]: Disable all content.
* ginclude/stdbool.h [defined __STDC_VERSION__ && __STDC_VERSION__
> 201710L] (bool, true, false): Do not define.
gcc/c-family/
* c-common.cc (c_common_reswords): Use D_C2X instead of D_CXXONLY
for alignas, alignof, bool, false, static_assert, thread_local and
true.
gcc/c/
* c-parser.cc (c_parser_static_assert_declaration_no_semi)
(c_parser_alignas_specifier, c_parser_alignof_expression): Allow
for C2x spellings of keywords.
(c_parser_postfix_expression): Handle RID_TRUE and RID_FALSE.
gcc/testsuite/
* gcc.dg/c11-keywords-1.c, gcc.dg/c2x-align-1.c,
gcc.dg/c2x-align-6.c, gcc.dg/c2x-bool-2.c,
gcc.dg/c2x-static-assert-3.c, gcc.dg/c2x-static-assert-4.c,
gcc.dg/c2x-thread-local-1.c: New tests.
* gcc.dg/c2x-bool-1.c: Update expectations.
libcpp/
* include/cpplib.h (struct cpp_options): Add true_false.
* expr.cc (eval_token): Check true_false not cplusplus to
determine whether to handle true and false keywords.
* init.cc (struct lang_flags): Add true_false.
(lang_defaults): Update.
(cpp_set_lang): Set true_false.
When DCE clears cfun->calls_setjmp then suddenly we don't need
any abnormal call edges anymore. The following makes sure to
prune them which otherwise can confuse other passes.
PR tree-optimization/106866
* tree-ssa-dce.cc (eliminate_unnecessary_stmts): When
we changed cfun->calls_setjmp make sure to purge all
abnormal call edges.
* gcc.dg/uninit-pr106866.c: New testcase.
When building gdb with --enable-shared, I run into:
...
ld: build/zlib/libz.a(libz_a-inffast.o): relocation R_X86_64_32S against \
`.rodata' can not be used when making a shared object; recompile with -fPIC
ld: build/zlib/libz.a(libz_a-inflate.o): warning: relocation against \
`inflateResetKeep' in read-only section `.text'
collect2: error: ld returned 1 exit status
make[3]: *** [libbfd.la] Error 1
...
This is a regression since commit a08bdb159bb ("[gdb/build] Fix gdbserver
build with -fsanitize=thread").
The problem is that a single case statement in configure is shared to handle
special requirements for both the host libiberty and host zlib, which has the
effect that only one is handled.
Fix this by handling libiberty and zlib each in its own case statement.
Build on x86_64-linux, with and without --enable-shared.
ChangeLog:
2022-09-07 Tom de Vries <tdevries@suse.de>
* configure.ac: Set extra_host_libiberty_configure_flags and
extra_host_zlib_configure_flags in separate case statements.
* configure: Regenerate.
The original code will cause the warning:
/usr/lib/gcc-snapshot/bin/g++ -fno-PIE -c -g -O2 -DIN_GCC -DCROSS_DIRECTORY_STRUCTURE -fno-exceptions -fno-rtti -fasynchronous-unwind-tables -W -Wall -Wno-narrowing -Wwrite-strings -Wcast-qual -Wmissing-format-attribute -Woverloaded-virtual -pedantic -Wno-long-long -Wno-variadic-macros -Wno-overlength-strings -Werror -fno-common -DHAVE_CONFIG_H -I. -I. -I../../gcc/gcc -I../../gcc/gcc/. -I../../gcc/gcc/../include -I../../gcc/gcc/../libcpp/include -I../../gcc/gcc/../libcody -I../../gcc/gcc/../libdecnumber -I../../gcc/gcc/../libdecnumber/dpd -I../libdecnumber -I../../gcc/gcc/../libbacktrace -o csky.o -MT csky.o -MMD -MP -MF ./.deps/csky.TPo ../../gcc/gcc/config/csky/csky.cc
In file included from ../../gcc/gcc/config/csky/csky.h:183,
from ./tm.h:20,
from ../../gcc/gcc/backend.h:28,
from ../../gcc/gcc/config/csky/csky.cc:27:
../../gcc/gcc/config/csky/csky.cc: In function 'bool csky_emit_compare_float(rtx_code, rtx, rtx)':
../../gcc/gcc/config/csky/csky_isa.h:29:37: error: enum constant in boolean context [-Werror=int-in-bool-context]
29 | #define CSKY_ISA_FEATURE_DEFINE(x) isa_bit_ ## x
| ^~~~~~~~
../../gcc/gcc/config/csky/csky_isa.h:30:37: note: in expansion of macro 'CSKY_ISA_FEATURE_DEFINE'
30 | #define CSKY_ISA_FEATURE_GET(x) CSKY_ISA_FEATURE_DEFINE (x)
| ^~~~~~~~~~~~~~~~~~~~~~~
../../gcc/gcc/config/csky/csky.cc:6346:43: note: in expansion of macro 'CSKY_ISA_FEATURE_GET'
6346 | || CSKY_ISA_FEATURE_GET(fpv2_df)
| ^~~~~~~~~~~~~~~~~~~~
../../gcc/gcc/config/csky/csky_isa.h:29:37: error: enum constant in boolean context [-Werror=int-in-bool-context]
29 | #define CSKY_ISA_FEATURE_DEFINE(x) isa_bit_ ## x
| ^~~~~~~~
../../gcc/gcc/config/csky/csky_isa.h:30:37: note: in expansion of macro 'CSKY_ISA_FEATURE_DEFINE'
30 | #define CSKY_ISA_FEATURE_GET(x) CSKY_ISA_FEATURE_DEFINE (x)
| ^~~~~~~~~~~~~~~~~~~~~~~
../../gcc/gcc/config/csky/csky.cc:6346:43: note: in expansion of macro 'CSKY_ISA_FEATURE_GET'
6346 | || CSKY_ISA_FEATURE_GET(fpv2_df)
| ^~~~~~~~~~~~~~~~~~~~
../../gcc/gcc/config/csky/csky_isa.h:29:37: error: enum constant in boolean context [-Werror=int-in-bool-context]
29 | #define CSKY_ISA_FEATURE_DEFINE(x) isa_bit_ ## x
| ^~~~~~~~
../../gcc/gcc/config/csky/csky_isa.h:30:37: note: in expansion of macro 'CSKY_ISA_FEATURE_DEFINE'
30 | #define CSKY_ISA_FEATURE_GET(x) CSKY_ISA_FEATURE_DEFINE (x)
| ^~~~~~~~~~~~~~~~~~~~~~~
../../gcc/gcc/config/csky/csky.cc:6347:43: note: in expansion of macro 'CSKY_ISA_FEATURE_GET'
6347 | || CSKY_ISA_FEATURE_GET(fpv2_divd)))
| ^~~~~~~~~~~~~~~~~~~~
The 'CSKY_ISA_FEATURE_DEFINE (x)' is an enum contant, it will cause the condition always being true.
In addition to warning, it will let FPUV3 to generate a move instruction, which is unnecessary.
In a simple test case, the move instruction can be combined, so it was not found in the testsuite.
But in some complex scenarios, this may generate some redundant instructions.
The correct way is to use 'CSKY_ISA_FEATURE' instead of 'CSKY_ISA_FEATURE_DEFINE'.
gcc/
* config/csky/csky.cc (csky_emit_compare_float): Fix the expanding of
float LE comparing with zero for fpuv3.
* config/csky/csky.h (TARGET_SUPPORT_FPV2): New, true if any fpuv2
features are enabled.
The following fixes a mistake in loop splitting which assumes loop
latches have a single predecessor and that edge is from the exit
test. Instead work from the single exit edge we have to find the
edge towards the latch.
PR tree-optimization/106860
* tree-ssa-loop-split.cc (split_loop): Find the exit to
latch edge from the loop exit edge instead of from the
latch. Verify we're going to find it.
* g++.dg/opt/pr106860.C: New testcase.
The following makes sure to mark the dominating region also for
USE predicate discovery, avoiding compute_control_dep_chain to
walk to unrelated areas, eating up walking budget.
* gimple-predicate-analysis.cc (dfs_mark_dominating_region):
Adjust to take the region exit source as argument.
(uninit_analysis::init_from_phi_def): Adjust.
(uninit_analysis::init_use_preds): Mark the dominating region
before computing control dependences.
8-bit and 16-bit FPR moves would ICE for +nosimd+fp, and some other
moves would handle FPR<-zero inefficiently. This is very much a
niche case at the moment, but something like it becomes more
important with SME streaming mode.
The si, di and vector tests already passed, they're just included for
completeness.
We're a bit inconsistent about whether alternatives involving FPRs
are marked with arch==fp or arch=* (i.e. default). E.g. FPR loads
and stores are sometimes * and sometimes fp.
IMO * makes more sense. FPRs should not be used at all without
TARGET_FLOAT, so TARGET_FLOAT represents the base architecture
when FPRs are enabled. I think it's more useful if non-default
arches represent a genuine restriction.
gcc/
* config/aarch64/aarch64.md (*mov<SHORT:mode>_aarch64): Extend
w<-w, r<-w and w<-r alternatives to !simd, using 32-bit moves
in that case. Extend w<-r to w<-Z.
(*mov<HFBF:mode>_aarch64): Likewise, but with Y instead of Z.
(*movti_aarch64): Use an FMOV from XZR for w<-Z if MOVI is not
available.
(define_split): Do not apply the floating-point immediate-to-register
split to zeros, even if MOVI is not available.
gcc/testsuite/
* gcc.target/aarch64/movqi_1.c: New test.
* gcc.target/aarch64/movhi_1.c: Likewise.
* gcc.target/aarch64/movsi_1.c: Likewise.
* gcc.target/aarch64/movdi_2.c: Likewise.
* gcc.target/aarch64/movti_2.c: Likewise.
* gcc.target/aarch64/movhf_1.c: Likewise.
* gcc.target/aarch64/movsf_1.c: Likewise.
* gcc.target/aarch64/movdf_1.c: Likewise.
* gcc.target/aarch64/movtf_2.c: Likewise.
* gcc.target/aarch64/movv8qi_1.c: Likewise.
* gcc.target/aarch64/movv16qi_1.c: Likewise.
+nofp disabled the automatic allocation of FPRs, but it didn't stop
users from explicitly putting register variables in FPRs. We'd then
either report an ICE or generate unsupported instructions.
It's still possible (and deliberately redundant) to specify FPRs in
clobber lists.
gcc/
* config/aarch64/aarch64.cc (aarch64_conditional_register_usage):
Disallow use of FPRs in register asms for !TARGET_FLOAT.
gcc/testsuite/
* gcc.target/aarch64/nofp_2.c: New test.
Just calling current_nonlambda_class_type in static member functions returns
non-NULL, but something that isn't *this and if unlucky can match part of the
IL and can be added to target clauses.
if (DECL_NONSTATIC_MEMBER_P (decl)
&& current_class_ptr)
is a guard used elsewhere (in check_accessibility_of_qualified_id).
2022-09-07 Jakub Jelinek <jakub@redhat.com>
PR c++/106829
* semantics.cc (finish_omp_target_clauses): If current_function_decl
isn't a nonstatic member function, don't set data.current_object to
non-NULL.
* g++.dg/gomp/pr106829.C: New test.
On Tue, Aug 30, 2022 at 09:10:37PM +0000, Joseph Myers wrote:
> I'm seeing build failures of glibc for powerpc64, as illustrated by the
> following C code:
>
> #if 0
> \NARG
> #endif
>
> (the actual sysdeps/powerpc/powerpc64/sysdep.h code is inside #ifdef
> __ASSEMBLER__).
>
> This shows some problems with this feature - and with delimited escape
> sequences - as it affects C. It's fine to accept it as an extension
> inside string and character literals, because \N or \u{...} would be
> invalid in the absence of the feature (i.e. the syntax for such literals
> fails to match, meaning that the rule about undefined behavior for a
> single ' or " as a pp-token applies). But outside string and character
> literals, the usual lexing rules apply, the \ is a pp-token on its own and
> the code is valid at the preprocessing level, and with expansion of macros
> appearing before or after the \ (e.g. u defined as a macro in the \u{...}
> case) it may be valid code at the language level as well. I don't know
> what older C++ versions say about this, but for C this means e.g.
>
> #define z(x) 0
> #define a z(
> int x = a\NARG);
>
> needs to be accepted as expanding to "int x = 0;", not interpreted as
> using the \N feature in an identifier and produce an error.
The following patch changes this, so that:
1) outside of string/character literals, \N without following { is never
treated as an error nor warning, it is silently treated as \ separate
token followed by whatever is after it
2) \u{123} and \N{LATIN SMALL LETTER A WITH ACUTE} are not handled as
extension at all outside of string/character literals in the strict
standard modes (-std=c*) except for -std=c++{23,2b}, only in the
-std=gnu* modes, because it changes behavior on valid sources, e.g.
#define z(x) 0
#define a z(
int x = a\u{123});
int y = a\N{LATIN SMALL LETTER A WITH ACUTE});
3) introduces -Wunicode warning (on by default) and warns for cases
of what looks like invalid delimited escape sequence or named
universal character escape outside of string/character literals
and is treated as separate tokens
2022-09-07 Jakub Jelinek <jakub@redhat.com>
libcpp/
* include/cpplib.h (struct cpp_options): Add cpp_warn_unicode member.
(enum cpp_warning_reason): Add CPP_W_UNICODE.
* init.cc (cpp_create_reader): Initialize cpp_warn_unicode.
* charset.cc (_cpp_valid_ucn): In possible identifier contexts, don't
handle \u{ or \N{ specially in -std=c* modes except -std=c++2{3,b}.
In possible identifier contexts, don't emit an error and punt
if \N isn't followed by {, or if \N{} surrounds some lower case
letters or _. In possible identifier contexts when not C++23, don't
emit an error but warning about unknown character names and treat as
separate tokens. When treating as separate tokens \u{ or \N{, emit
warnings.
gcc/
* doc/invoke.texi (-Wno-unicode): Document.
gcc/c-family/
* c.opt (Winvalid-utf8): Use ObjC instead of objC. Remove
" in comments" from description.
(Wunicode): New option.
gcc/testsuite/
* c-c++-common/cpp/delimited-escape-seq-4.c: New test.
* c-c++-common/cpp/delimited-escape-seq-5.c: New test.
* c-c++-common/cpp/delimited-escape-seq-6.c: New test.
* c-c++-common/cpp/delimited-escape-seq-7.c: New test.
* c-c++-common/cpp/named-universal-char-escape-5.c: New test.
* c-c++-common/cpp/named-universal-char-escape-6.c: New test.
* c-c++-common/cpp/named-universal-char-escape-7.c: New test.
* g++.dg/cpp23/named-universal-char-escape1.C: New test.
* g++.dg/cpp23/named-universal-char-escape2.C: New test.
There are two splitters, both are calling rs6000_emit_set_const to emit
instructions for constant building.
One splitter checks `const_int_operand`, this splitter is always used.
Another spitter checks `const_scalar_int_operand`, this one is never
used now.
Checking the history, that splitter is introduced around 1999 for
HOST_BITS_PER_WIDE_INT=32, as below:
(define_split
[(set (match_operand:DI 0 "gpc_reg_operand" "")
(match_operand:DI 1 "const_double_operand" ""))]
"HOST_BITS_PER_WIDE_INT == 32 && TARGET_POWERPC64
&& num_insns_constant (operands[1], DImode) > 1"
And `HOST_BITS_PER_WIDE_INT == 32` is eliminated later, since it is
`const_double_operand` can cover it.
When wide_int is introduced, `const_double_operand` is replaced by
`const_scalar_int_operand`.
Now, HOST_BITS_PER_WIDE_INT is forced to 64, this splitter is safe
to remove.
gcc/ChangeLog:
* config/rs6000/rs6000.md (const_scalar_int splitter): Remove.
As a comment in
https://gcc.gnu.org/pipermail/gcc-patches/2022-August/599556.html
Those splitters call rs6000_emit_set_const directly, and the replacements
are never used. Using (pc) would be less misleading.
gcc/ChangeLog:
* config/rs6000/rs6000.md: (constant splitters): Use "(pc)" as the
replacements.
We were ignoring all functions starting with "__morestack_", but not
the function "__morestack" itself. Without this change, some tests
such as recover.go started failing recently, though I'm not sure
exactly what changed.
Reviewed-on: https://go-review.googlesource.com/c/gofrontend/+/427935
As the failure of test case gcc.target/powerpc/pr92398.p9-.c in
PR106345 shows, some test sources for some powerpc effective
targets use empty translation unit wrongly. The test sources
could go with options like "-ansi -pedantic-errors", then those
effective target checkings will fail unexpectedly with the
error messages like:
error: ISO C forbids an empty translation unit [-Wpedantic]
This patch is to fix empty TUs with one dummy function definition
accordingly.
PR testsuite/106345
gcc/testsuite/ChangeLog:
* lib/target-supports.exp (check_effective_target_powerpc_sqrt): Add
a function definition to avoid pedwarn about empty translation unit.
(check_effective_target_has_arch_pwr5): Likewise.
(check_effective_target_has_arch_pwr6): Likewise.
(check_effective_target_has_arch_pwr7): Likewise.
(check_effective_target_has_arch_pwr8): Likewise.
(check_effective_target_has_arch_pwr9): Likewise.
(check_effective_target_has_arch_pwr10): Likewise.
(check_effective_target_has_arch_ppc64): Likewise.
(check_effective_target_ppc_float128): Likewise.
(check_effective_target_ppc_float128_insns): Likewise.
(check_effective_target_powerpc_vsx): Likewise.
