Jonathan Wakely [Mon, 22 Aug 2022 14:16:16 +0000 (15:16 +0100)]
libstdc++: Check for overflow in regex back-reference [PR106607]
Currently we fail to notice integer overflow when parsing a
back-reference expression, or when converting the parsed result from
long to int. This changes the result to be int, so no conversion is
needed, and uses the overflow-checking built-ins to detect an
out-of-range back-reference.
libstdc++-v3/ChangeLog:
PR libstdc++/106607
* include/bits/regex_compiler.tcc (_Compiler::_M_cur_int_value):
Use built-ins to check for integer overflow in back-reference
number.
* testsuite/28_regex/basic_regex/106607.cc: New test.
Dimitar Dimitrov [Sat, 20 Aug 2022 20:58:27 +0000 (23:58 +0300)]
pru: Optimize 64-bit logical operations
The earlyclobber in the pattern yields inefficient code due to
unnecessarily generated moves. Optimize by removing the earlyclobber
for two special alternatives:
- If OP2 is a small constant integer.
- If the logical bit operation has only two operands.
gcc/ChangeLog:
* config/pru/pru.md (pru_<code>di3): New alternative for
two operands but without earlyclobber.
Dimitar Dimitrov [Sun, 14 Aug 2022 15:50:18 +0000 (18:50 +0300)]
PR target/106564: pru: Optimize 64-bit sign- and zero-extend
Add new patterns to optimize 64-bit sign- and zero-extend operations for
the PRU target.
The new 64-bit zero-extend patterns are straightforward define_insns.
The old 16/32-bit sign-extend pattern has been rewritten from scratch
in order to add 64-bit support. The new pattern expands into several
optimized insns for filling bytes with zeros or ones, and for
conditional branching on bit-test. The bulk of this patch is to
implement the patterns for those new optimized insns.
PR target/106564
gcc/ChangeLog:
* config/pru/constraints.md (Um): New constraint for -1.
(Uf): New constraint for IOR fill-bytes constants.
(Uz): New constraint for AND zero-bytes constants.
* config/pru/predicates.md (const_fillbytes_operand): New
predicate for IOR fill-bytes constants.
(const_zerobytes_operand): New predicate for AND zero-bytes
constants.
* config/pru/pru-protos.h (pru_output_sign_extend): Remove.
(struct pru_byterange): New struct to describe a byte range.
(pru_calc_byterange): New declaration.
* config/pru/pru.cc (pru_rtx_costs): Add penalty for
64-bit zero-extend.
(pru_output_sign_extend): Remove.
(pru_calc_byterange): New helper function to extract byte
range info from a constant.
(pru_print_operand): Remove 'y' and 'z' print modifiers.
* config/pru/pru.md (zero_extendqidi2): New pattern.
(zero_extendhidi2): New pattern.
(zero_extendsidi2): New pattern.
(extend<EQS0:mode><EQD:mode>2): Rewrite as an expand.
(@pru_ior_fillbytes<mode>): New pattern.
(@pru_and_zerobytes<mode>): New pattern.
(<code>di3): Rewrite as an expand and handle ZERO and FILL
special cases.
(pru_<code>di3): New name for <code>di3.
(@cbranch_qbbx_const_<BIT_TEST:code><HIDI:mode>): New pattern to
handle bit-test for 64-bit registers.
gcc/testsuite/ChangeLog:
* gcc.target/pru/pr106564-1.c: New test.
* gcc.target/pru/pr106564-2.c: New test.
* gcc.target/pru/pr106564-3.c: New test.
* gcc.target/pru/pr106564-4.c: New test.
Harald Anlauf [Sat, 20 Aug 2022 18:36:28 +0000 (20:36 +0200)]
Fortran: fix simplification of intrinsics IBCLR and IBSET [PR106557]
gcc/fortran/ChangeLog:
PR fortran/106557
* simplify.cc (gfc_simplify_ibclr): Ensure consistent results of
the simplification by dropping a redundant memory representation
of argument x.
(gfc_simplify_ibset): Likewise.
gcc/testsuite/ChangeLog:
PR fortran/106557
* gfortran.dg/pr106557.f90: New test.
Richard Biener [Mon, 22 Aug 2022 12:45:38 +0000 (14:45 +0200)]
Remove dead predicate analysis GENERIC expr building code
The following removes the unused def_expr, use_expr and expr APIs
from the predicate class including the unconditional build of the
GENERIC use_expr on each uninit analysis run.
Martin Liska [Thu, 18 Aug 2022 11:03:42 +0000 (13:03 +0200)]
jobserver: detect properly O_NONBLOCK
PR lto/106700
gcc/ChangeLog:
* configure.ac: Detect O_NONBLOCK flag for open.
* config.in: Regenerate.
* configure: Regenerate.
* opts-common.cc (jobserver_info::connect): Set is_connected
properly based on O_NONBLOCK.
* opts-jobserver.h (struct jobserver_info): Add is_connected
member variable.
gcc/lto/ChangeLog:
* lto.cc (wait_for_child): Ask if we are connected to jobserver.
(stream_out_partitions): Likewise.
Tobias Burnus [Mon, 22 Aug 2022 07:49:52 +0000 (09:49 +0200)]
lto-wrapper.cc: Delete offload_names temp files in case of error [PR106686]
Usually, the caller takes care of the .o files for the offload compilers
(suffix: ".target.o"). However, if an error occurs during processing
(e.g. fatal error by lto1), they were not deleted.
gcc/ChangeLog:
PR lto/106686
* lto-wrapper.cc (free_array_of_ptrs): Move before tool_cleanup.
(tool_cleanup): Unlink offload_names.
(compile_offload_image): Take filename argument to set it early.
(compile_images_for_offload_targets): Update call; set
offload_names to NULL after freeing the array.
The following avoids adding PHIs to the worklist for uninit processing
if we reach them following backedges. That confuses predicate analysis
because it assumes the use is happening in the same iteration as the the
definition. For the testcase in the PR the situation is like
void foo (int val)
{
int uninit;
# val = PHI <..> (B)
for (..)
{
if (..)
{
.. = val; (C)
val = uninit;
}
# val = PHI <..> (A)
}
}
and starting from (A) with 'uninit' as argument we arrive at (B)
and from there at (C). Predicate analysis then tries to prove
the predicate of (B) (not the backedge) can prove that the
path from (B) to (C) is unreachable which isn't really what it
necessary - that's what we'd need to do when the preheader
edge of the loop were the edge with the uninitialized def.
So the following makes those cases intentionally false negatives.
PR tree-optimization/105937
* tree-ssa-uninit.cc (find_uninit_use): Do not queue PHIs
on backedges.
(execute_late_warn_uninitialized): Mark backedges.
Richard Biener [Thu, 18 Aug 2022 13:49:58 +0000 (15:49 +0200)]
Improve uninit analysis
The following reduces the number of false positives in uninit analysis
by providing fallback for situations the current analysis gives up
and thus warns because it cannot prove initialization.
The first situation is when compute_control_dep_chain gives up walking
because it runs into either param_uninit_control_dep_attempts or
MAX_CHAIN_LEN. If in the process it did not collect a single path
from function entry to the interesting PHI edge then we'll give up
and diagnose. The following patch insteads provides a sparse path
including only those predicates that always hold when the PHI edge
is reached in that case. That's cheap to produce but may in some
odd cases prove less precise than what the code tries now (enumerating
all possible paths from function entry to the PHI edge, but only
use the first N of those and only require unreachability of those N).
The second situation is when the set of predicates computed to hold
on the use stmt was formed from multiple paths (there's a similar
enumeration of all paths and their predicates from the PHI def to the
use). In that case use_preds.use_cannot_happen gives up because
it doesn't know which of the predicates from which path from PHI to
the use it can use to prove unreachability of the PHI edge that has
the uninitialized def. The patch for this case simply computes
the intersection of the predicates and uses that for further analysis,
but in a crude way since the predicate vectors are not sorted.
Fortunately the total size is limited - we have max MAX_NUM_CHAINS
number of predicates each of length MAX_CHAIN_LEN so the brute
force intersection code should behave quite reasonable in practice.
* gimple-predicate-analysis.cc (predicate::use_cannot_happen):
If the use is guarded with multiple predicate paths compute
the predicates intersection before going forward. When
compute_control_dep_chain wasn't able to come up with at
least one path from function entry to the PHI edge compute
a conservative sparse path instead.
Martin Liska [Mon, 22 Aug 2022 03:13:18 +0000 (05:13 +0200)]
analyzer: add missing final keyword
Fixes the following clang warning:
gcc/analyzer/region-model.cc:5096:8: warning: 'subclass_equal_p' overrides a member function but is not marked 'override' [-Winconsistent-missing-override]
Jakub Jelinek [Sat, 20 Aug 2022 19:20:04 +0000 (21:20 +0200)]
fortran: Drop -static-lib{gfortran,quadmath} from f951 [PR46539]
As discussed earlier, all other -static-lib* options are Driver only,
these 2 are Driver in common.opt and Fortran in lang.opt.
The spec files never pass the -static-lib* options down to any compiler
(f951 etc.), so the 2 errors below are reported only when one
runs ./f951 -static-libgfortran by hand.
The following patch just removes f951 support of these options, the
gfortran driver behavior remains as before. For other -static-lib*
option (and even these because it is never passed to f951) we never
error if we can't support those options, and e.g. Darwin is actually
able to handle those options through other means.
2022-08-20 Jakub Jelinek <jakub@redhat.com>
PR fortran/46539
* lang.opt (static-libgfortran, static-libquadmath): Change Fortran
to Driver.
* options.cc (gfc_handle_option): Don't handle OPT_static_libgfortran
nor OPT_static_libquadmath here.
Lulu Cheng [Thu, 18 Aug 2022 09:26:13 +0000 (17:26 +0800)]
LoongArch: Add support code model extreme.
Use five instructions to calculate a signed 64-bit offset relative to the pc.
gcc/ChangeLog:
* config/loongarch/loongarch-opts.cc: Allow cmodel to be extreme.
* config/loongarch/loongarch.cc (loongarch_call_tls_get_addr):
Add extreme support for TLS GD and LD types.
(loongarch_legitimize_tls_address): Add extreme support for TLS LE
and IE.
(loongarch_split_symbol): When compiling with -mcmodel=extreme,
the symbol address will be obtained through five instructions.
(loongarch_print_operand_reloc): Add support.
(loongarch_print_operand): Add support.
(loongarch_print_operand_address): Add support.
(loongarch_option_override_internal): Set '-mcmodel=extreme' option
incompatible with '-mno-explicit-relocs'.
* config/loongarch/loongarch.md (@lui_l_hi20<mode>):
Loads bits 12-31 of data into registers.
(lui_h_lo20): Load bits 32-51 of the data and spell bits 0-31 of
the source register.
(lui_h_hi12): Load bits 52-63 of the data and spell bits 0-51 of
the source register.
* config/loongarch/predicates.md: Symbols need to be decomposed
when defining the macro TARGET_CMODEL_EXTREME
* doc/invoke.texi: Modify the description information of cmodel in the document.
Document -W[no-]extreme-plt.
Tobias Burnus [Fri, 19 Aug 2022 14:09:28 +0000 (16:09 +0200)]
mkoffload: Cleanup temporary omp_requires_file
The file (suffix ".mkoffload.omp_requires") used to save the 'omp requires'
data has to be passed to maybe_unlink for cleanup or -v -save-temps stderr
diagnostic. That was missed before. - For GCN, the same has to be done for
the files with suffix ".mkoffload.dbg.o".
gcc/ChangeLog:
* config/gcn/mkoffload.cc (main): Add omp_requires_file and dbgobj to
files_to_cleanup.
* config/i386/intelmic-mkoffload.cc (prepare_target_image): Add
omp_requires_file to temp_files.
* config/nvptx/mkoffload.cc (omp_requires_file): New global static var.
(main): Remove local omp_requires_file var.
(tool_cleanup): Handle omp_requires_file.
Aldy Hernandez [Thu, 18 Aug 2022 15:55:19 +0000 (17:55 +0200)]
Remove path_range_query constructor that takes an edge.
The path_range_query constructor that takes an edge is really a
convenience function for the loop-ch pass. It feels wrong to pollute
the API with such a specialized function that could be done with
a small inline function closer to its user.
As an added benefit, we remove one use of reset_path. The last
remaining one is the forward threader one.
Tested, thread-counted, and benchmarked on x86-64 Linux.
Joseph Myers [Thu, 18 Aug 2022 23:01:01 +0000 (23:01 +0000)]
preprocessor: Support #warning for standard C2x
ISO C2x standardizes the existing #warning extension. Arrange
accordingly for it not to be diagnosed with -std=c2x -pedantic, but to
be diagnosed with -Wc11-c2x-compat.
Bootstrapped with no regressions for x86_64-pc-linux-gnu.
libcpp/
* include/cpplib.h (struct cpp_options): Add warning_directive.
* init.cc (struct lang_flags, lang_defaults): Add
warning_directive.
* directives.cc (DIRECTIVE_TABLE): Mark #warning as STDC2X not
EXTENSION.
(directive_diagnostics): Diagnose #warning with -Wc11-c2x-compat,
or with -pedantic for a standard not supporting #warning.
No longer needs the dedicated hard register (A11) for the address of the
call and the split patterns for fixups, due to the introduction of appropriate
register class and constraint.
(Note: "ISC_REGS" contains a hard register A8 used as a "static chain"
pointer for nested functions, but no problem; Pointer to nested function
actually points to "trampoline", and trampoline itself doesn't receive
"static chain" pointer to its parent's stack frame from the caller.)
gcc/ChangeLog:
* config/xtensa/xtensa.h
(enum reg_class, REG_CLASS_NAMES, REG_CLASS_CONTENTS):
Add new register class "ISC_REGS".
* config/xtensa/constraints.md (c): Add new register constraint.
* config/xtensa/xtensa.md (define_constants): Remove "A11_REG".
(sibcall_internal, sibcall_value_internal):
Change to use the new register constraint, and remove two split
patterns for fixups that are no longer needed.
gcc/testsuite/ChangeLog:
* gcc.target/xtensa/sibcalls.c: Add a new test function to ensure
that registers for arguments (occupy from A2 to A7) and for indirect
sibcall (should be assigned to A8) neither conflict nor spill out.
Tim Lange [Thu, 18 Aug 2022 09:36:08 +0000 (11:36 +0200)]
analyzer: warn on the use of floating-points operands in the size argument [PR106181]
This patch fixes the ICE reported in PR106181 and adds a new warning to
the analyzer complaining about the use of floating-point operands.
Regrtested on Linux x86_64.
2022-08-17 Tim Lange <mail@tim-lange.me>
gcc/analyzer/ChangeLog:
PR analyzer/106181
* analyzer.opt: Add Wanalyzer-imprecise-floating-point-arithmetic.
* region-model.cc (is_any_cast_p): Formatting.
(region_model::check_region_size): Ensure precondition.
(class imprecise_floating_point_arithmetic): New abstract
diagnostic class for all floating-point related warnings.
(class float_as_size_arg): Concrete diagnostic class to complain
about floating-point operands inside the size argument.
(class contains_floating_point_visitor):
New visitor to find floating-point operands inside svalues.
(region_model::check_dynamic_size_for_floats): New function.
(region_model::set_dynamic_extents):
Call to check_dynamic_size_for_floats.
* region-model.h (class region_model):
Add region_model::check_dynamic_size_for_floats.
PR analyzer/106181
* gcc.dg/analyzer/allocation-size-1.c: New test.
* gcc.dg/analyzer/imprecise-floating-point-1.c: New test.
* gcc.dg/analyzer/pr106181.c: New test.
Aldy Hernandez [Wed, 17 Aug 2022 11:18:01 +0000 (13:18 +0200)]
Make path_range_query standalone and add reset_path.
These are a bunch of cleanups inspired by Richi's suggestion of making
path_range_query standalone, instead of having to call
compute_ranges() for each path.
I've made the ranger need explicit, and moved the responsibility for
its creation to the caller.
I've also investigated and documented why the forward threader needs its
own compute exit dependencies variant. I can't wait for it to go away
:-/.
I've also added constructors that take a path and dependencies, and
made compute_ranges() private. Unfortunately, reinstantiating
path_range_query in the forward threader caused a 14% performance
regression in DOM, because the old threader calls it over and over on
the same path to simplify statements (some of which not even in the
IL, but that's old news).
In the meantime, I've left the ability to reset a path, but this time
appropriately called reset_path().
Tested, benchmarked, and thread counted on x86-64 Linux.
gcc/ChangeLog:
* gimple-range-path.cc (path_range_query::path_range_query): Add
various constructors to take a path.
(path_range_query::~path_range_query): Remove m_alloced_ranger.
(path_range_query::range_on_path_entry): Adjust for m_ranger being
a reference.
(path_range_query::set_path): Rename to...
(path_range_query::reset_path): ...this and call compute_ranges.
(path_range_query::ssa_range_in_phi): Adjust for m_ranger
reference.
(path_range_query::range_defined_in_block): Same.
(path_range_query::compute_ranges_in_block): Same.
(path_range_query::adjust_for_non_null_uses): Same.
(path_range_query::compute_exit_dependencies): Use m_path instead
of argument.
(path_range_query::compute_ranges): Remove path argument.
(path_range_query::range_of_stmt): Adjust for m_ranger reference.
(path_range_query::compute_outgoing_relations): Same.
* gimple-range-path.h (class path_range_query): Add various
constructors.
Make compute_ranges and compute_exit_dependencies private.
Rename set_path to reset_path.
Make m_ranger a reference.
Remove m_alloced_ranger.
* tree-ssa-dom.cc (pass_dominator::execute): Adjust constructor to
path_range_query.
* tree-ssa-loop-ch.cc (entry_loop_condition_is_static): Take a
ranger and instantiate a new path_range_query every time.
(ch_base::copy_headers): Pass ranger instead of path_range_query.
* tree-ssa-threadbackward.cc (class back_threader): Remove m_solver.
(back_threader::~back_threader): Remove m_solver.
(back_threader::find_taken_edge_switch): Adjust for m_ranger
reference.
(back_threader::find_taken_edge_cond): Same.
(back_threader::dump): Remove m_solver.
(back_threader::back_threader): Move verify_marked_backedges
here from the path_range_query constructor.
* tree-ssa-threadedge.cc (hybrid_jt_simplifier::simplify): Move
some code from compute_ranges_from_state here.
(hybrid_jt_simplifier::compute_ranges_from_state): Rename...
(hybrid_jt_simplifier::compute_exit_dependencies): ...to this.
* tree-ssa-threadedge.h (class hybrid_jt_simplifier): Rename
compute_ranges_from_state to compute_exit_dependencies.
Remove m_path.
Now that we have parts of fold_binary_op_with_conditional_arg duplicated
in match.pd and are using ! to take or throw away the result we have to
be careful to not have both implementations play games which each other,
causing quadratic behavior. In particular the match.pd implementation
requires both arms to simplify while the fold-const.cc is happy with
just one arm simplifying (something we cannot express in match.pd).
The fix is to simply not enable the match.pd pattern for GENERIC.
PR middle-end/106617
* match.pd ((a ? b : c) > d -> a ? (b > d) : (c > d)): Fix
guard, disable on GENERIC to not cause quadratic behavior
with the fold-const.cc implementation and the use of !
* g++.target/i386/vect-bfloat16-1.C: New test.
* gcc.target/i386/vect-bfloat16-1.c: New test.
* gcc.target/i386/vect-bfloat16-2a.c: New test.
* gcc.target/i386/vect-bfloat16-2b.c: New test.
* gcc.target/i386/vect-bfloat16-typecheck_1.c: New test.
* gcc.target/i386/vect-bfloat16-typecheck_2.c: New test.
Use of libelf was removed from gcc in r0-104274-g48215350c24d52 ("re PR
lto/46273 (Failed to bootstrap)") around 2010, before gcc-4.6.0.
This change removes unused references to libelf from top-level configure
and Makefile.
/
* Makefile.def: Drop libelf module and gcc-configure dependency
on it.
* Makefile.in: Regenerate with 'autogen Makefile.def'.
* Makefile.tpl (HOST_EXPORTS): Drop unused LIBELFLIBS and
LIBELFINC.
* configure: Regenrate.
* configure.ac (host_libs): Drop unused libelf.
This patch is preparing patch for RVV support.
Because we didn't introduce vector machine_mode yet, it safe to just change HOST_WIDE_INT into poly_int.
Also it safe to use "to_constant()" function for scalar operation.
This patch has been tested by full dejagnu regression.
xtensa: Optimize stack pointer updates in function pro/epilogue under certain conditions
This patch enforces the use of "addmi" machine instruction instead of
addition/subtraction with two source registers for adjusting the stack
pointer, if the adjustment fits into a signed 16-bit and is also a multiple
of 256.
/* example */
void test(void) {
char buffer[4096];
__asm__(""::"m"(buffer));
}
* config/xtensa/xtensa.cc (xtensa_expand_prologue):
Use an "addmi" machine instruction for updating the stack pointer
rather than addition/subtraction via hard register A9, if the amount
of change satisfies the literal value conditions of that instruction
when the CALL0 ABI is used.
(xtensa_expand_epilogue): Ditto.
And also inhibit the stack pointer addition of constant zero.
RISC-V/testsuite: Restrict remaining `fmin'/`fmax' tests to hard float
Complement commit 7915f6551343 ("RISC-V/testsuite: constraint some of
tests to hard_float") and also restrict the remaining `fmin'/`fmax'
tests to hard-float test configurations.
Roger Sayle [Wed, 17 Aug 2022 21:23:33 +0000 (22:23 +0100)]
[Committed] PR target/106640: Fix use of XINT in TImode compute_convert_gain.
Thanks to Zdenek Sojka for reporting PR target/106640 where an RTL checking
build reveals a thinko in my recent patch to support TImode shifts/rotates
in STV. My "senior moment" was to inappropriately use XINT where I should
be using INTVAL of XEXP.
2022-08-17 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog
PR target/106640
* config/i386/i386-features.cc
(timde_scalar_chain::compute_convert_gain): Replace incorrect use
of XINT with INTVAL (XEXP (src, 1)).
Aldy Hernandez [Wed, 17 Aug 2022 15:47:21 +0000 (17:47 +0200)]
Reset root oracle from path_oracle::reset_path.
When we cross a backedge in the path solver, we reset the path
relations and nuke the root oracle. However, we forget to reset it
for the next path. This is causing us to miss threads because
subsequent paths will have no root oracle to use.
With this patch we get 201 more threads in the threadfull passes in my
.ii files and 118 more overall (DOM gets less because threadfull runs
before).
Normally, I'd recommend this for the GCC 12 branch, but considering
how sensitive other passes are to jump threading, and that there is no
PR associated with this, perhaps we should leave this out. Up to the
release maintainers of course.
Marek Polacek [Wed, 17 Aug 2022 17:36:52 +0000 (13:36 -0400)]
c++: Extend -Wredundant-move for const-qual objects [PR90428]
In this PR, Jon suggested extending the -Wredundant-move warning
to warn when the user is moving a const object as in:
struct T { };
T f(const T& t)
{
return std::move(t);
}
where the std::move is redundant, because T does not have
a T(const T&&) constructor (which is very unlikely). Even with
the std::move, T(T&&) would not be used because it would mean
losing the const. Instead, T(const T&) will be called.
I had to restructure the function a bit, but it's better now. This patch
depends on my other recent patches to maybe_warn_pessimizing_move.
PR c++/90428
gcc/cp/ChangeLog:
* typeck.cc (can_do_rvo_p): Rename to ...
(can_elide_copy_prvalue_p): ... this.
(maybe_warn_pessimizing_move): Extend the
-Wredundant-move warning to warn about std::move on a
const-qualified object.
Marek Polacek [Wed, 3 Aug 2022 23:47:30 +0000 (19:47 -0400)]
c++: Tweak for -Wpessimizing-move in templates [PR89780]
In my previous patches I've been extending our std::move warnings,
but this tweak actually dials it down a little bit. As reported in
bug 89780, it's questionable to warn about expressions in templates
that were type-dependent, but aren't anymore because we're instantiating
the template. As in,
template <typename T>
Dest withMove() {
T x;
return std::move(x);
}
template Dest withMove<Dest>(); // #1
template Dest withMove<Source>(); // #2
Saying that the std::move is pessimizing for #1 is not incorrect, but
it's not useful, because removing the std::move would then pessimize #2.
So the user can't really win. At the same time, disabling the warning
just because we're in a template would be going too far, I still want to
warn for
template <typename>
Dest withMove() {
Dest x;
return std::move(x);
}
because the std::move therein will be pessimizing for any instantiation.
So I'm using the suppress_warning machinery to that effect.
Problem: I had to add a new group to nowarn_spec_t, otherwise
suppressing the -Wpessimizing-move warning would disable a whole bunch
of other warnings, which we really don't want.
PR c++/89780
gcc/cp/ChangeLog:
* pt.cc (tsubst_copy_and_build) <case CALL_EXPR>: Maybe suppress
-Wpessimizing-move.
* typeck.cc (maybe_warn_pessimizing_move): Don't issue warnings
if they are suppressed.
(check_return_expr): Disable -Wpessimizing-move when returning
a dependent expression.
Marek Polacek [Mon, 1 Aug 2022 21:02:23 +0000 (17:02 -0400)]
c++: Extend -Wpessimizing-move to other contexts
In my recent patch which enhanced -Wpessimizing-move so that it warns
about class prvalues too I said that I'd like to extend it so that it
warns in more contexts where a std::move can prevent copy elision, such
as:
T t = std::move(T());
T t(std::move(T()));
T t{std::move(T())};
T t = {std::move(T())};
void foo (T);
foo (std::move(T()));
This patch does that by adding two maybe_warn_pessimizing_move calls.
These must happen before we've converted the initializers otherwise the
std::move will be buried in a TARGET_EXPR.
PR c++/106276
gcc/cp/ChangeLog:
* call.cc (build_over_call): Call maybe_warn_pessimizing_move.
* cp-tree.h (maybe_warn_pessimizing_move): Declare.
* decl.cc (build_aggr_init_full_exprs): Call
maybe_warn_pessimizing_move.
* typeck.cc (maybe_warn_pessimizing_move): Handle TREE_LIST and
CONSTRUCTOR. Add a bool parameter and use it. Adjust a diagnostic
message.
(check_return_expr): Adjust the call to maybe_warn_pessimizing_move.
gcc/testsuite/ChangeLog:
* g++.dg/cpp0x/Wpessimizing-move7.C: Add dg-warning.
* g++.dg/cpp0x/Wpessimizing-move8.C: New test.
Jakub Jelinek [Wed, 17 Aug 2022 15:00:33 +0000 (17:00 +0200)]
fortran: Add -static-libquadmath support [PR46539]
The following patch is a revival of the
https://gcc.gnu.org/legacy-ml/gcc-patches/2014-10/msg00771.html
patch. While trunk configured against recent glibc and with linker
--as-needed support doesn't really need to link against -lquadmath
anymore, there are still other targets where libquadmath is still in
use.
As has been discussed, making -static-libgfortran imply statically
linking both libgfortran and libquadmath is undesirable because of
the significant licensing differences between the 2 libraries.
Compared to the 2014 patch, this one doesn't handle -lquadmath
addition in the driver, which to me looks incorrect, libgfortran
configure determines where in libgfortran.spec -lquadmath should
be present if at all and with what it should be wrapped, but
analyzes gfortran -### -static-libgfortran stderr and based on
that figures out what gcc/configure.ac determined.
2022-08-17 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
Jakub Jelinek <jakub@redhat.com>
PR fortran/46539
gcc/
* common.opt (static-libquadmath): New option.
* gcc.cc (driver_handle_option): Always accept -static-libquadmath.
* config/darwin.h (LINK_SPEC): Handle -static-libquadmath.
gcc/fortran/
* lang.opt (static-libquadmath): New option.
* invoke.texi (-static-libquadmath): Document it.
* options.cc (gfc_handle_option): Error out if -static-libquadmath
is passed but we do not support it.
libgfortran/
* acinclude.m4 (LIBQUADSPEC): From $FC -static-libgfortran -###
output determine -Bstatic/-Bdynamic, -bstatic/-bdynamic,
-aarchive_shared/-adefault linker support or Darwin remapping
of -lgfortran to libgfortran.a%s and use that around or instead
of -lquadmath in LIBQUADSPEC.
* configure: Regenerated.
Tobias Burnus [Wed, 17 Aug 2022 14:19:58 +0000 (16:19 +0200)]
Fortran: OpenMP fix declare simd inside modules and absent linear step [PR106566]
gcc/fortran/ChangeLog:
PR fortran/106566
* openmp.cc (gfc_match_omp_clauses): Fix setting linear-step value
to 1 when not specified.
(gfc_match_omp_declare_simd): Accept module procedures.
gcc/testsuite/ChangeLog:
PR fortran/106566
* gfortran.dg/gomp/declare-simd-4.f90: New test.
* gfortran.dg/gomp/declare-simd-5.f90: New test.
* gfortran.dg/gomp/declare-simd-6.f90: New test.
Tobias Burnus [Wed, 17 Aug 2022 13:52:23 +0000 (15:52 +0200)]
OpenMP requires: Fix diagnostic filename corner case
The issue occurs when there is, e.g., main._omp_fn.0 in two files with
different OpenMP requires clauses. The function entries in the offload
table ends up having the same decl tree and, hence, the diagnostic showed
the same filename for both. Solution: Use the .o filename in this case.
Note that the issue does not occur with same-named 'static' functions and
without the fatal error from the requires diagnostic, there would be
later a linker error due to having two 'main'.
gcc/
* lto-cgraph.cc (input_offload_tables): Improve requires diagnostic
when filenames come out identically.
When splay_tree_prefix is defined, the .h file
defines splay_* macros to add the prefix. However,
before those were only unset when additionally
splay_tree_c was defined.
Additionally, for consistency undefine splay_tree_c
also when no splay_tree_prefix is defined - there
is no interdependence either.
libgomp/ChangeLog:
* splay-tree.h: Fix splay_* macro unsetting if
splay_tree_prefix is defined.
Tobias Burnus [Wed, 17 Aug 2022 12:17:08 +0000 (14:17 +0200)]
OpenMP/C++: Allow classes with static members to be mappable [PR104493]
As this is the last lang-specific user of the omp_mappable_type hook,
the hook is removed, keeping only a generic omp_mappable_type for
incomplete types (or error_node).
* cp-objcp-common.h (LANG_HOOKS_OMP_MAPPABLE_TYPE): Remove.
* cp-tree.h (cp_omp_mappable_type, cp_omp_emit_unmappable_type_notes):
Remove.
* decl2.cc (cp_omp_mappable_type_1, cp_omp_mappable_type,
cp_omp_emit_unmappable_type_notes): Remove.
(cplus_decl_attributes): Call omp_mappable_type instead of
removed langhook.
* decl.cc (cp_finish_decl): Likewise; call cxx_incomplete_type_inform
in lieu of cp_omp_emit_unmappable_type_notes.
* semantics.cc (finish_omp_clauses): Likewise.
gcc/ChangeLog:
* gimplify.cc (omp_notice_variable): Call omp_mappable_type
instead of removed langhook.
* omp-general.h (omp_mappable_type): New prototype.
* omp-general.cc (omp_mappable_type): New; moved from ...
* langhooks.cc (lhd_omp_mappable_type): ... here.
* langhooks-def.h (lhd_omp_mappable_type,
LANG_HOOKS_OMP_MAPPABLE_TYPE): Remove.
(LANG_HOOKS_FOR_TYPES_INITIALIZER): Remote the latter.
* langhooks.h (struct lang_hooks_for_types): Remove
omp_mappable_type.
gcc/testsuite/ChangeLog:
* g++.dg/gomp/unmappable-1.C: Remove dg-error; remove dg-note no
longer shown as TYPE_MAIN_DECL is NULL.
* c-c++-common/gomp/map-incomplete-type.c: New test.
Christophe Lyon [Wed, 17 Aug 2022 08:06:24 +0000 (10:06 +0200)]
arm: Define with_float to hard when target name ends with hf
On arm, the --with-float= configure option is used to define include
files search path (among other things). However, when targeting
arm-linux-gnueabihf, one would expect to automatically default to the
hard-float ABI, but this is not the case. As a consequence, GCC
bootstrap fails on an arm-linux-gnueabihf target if --with-float=hard
is not used.
This patch checks if the target name ends with 'hf' and defines
with_float to hard if not already defined. This is achieved in
gcc/config.gcc, just before selecting the default CPU depending on the
$with_float value.
Richard Biener [Mon, 15 Aug 2022 13:38:09 +0000 (15:38 +0200)]
Refactor back_threader_profitability
The following refactors profitable_path_p in the backward threader,
splitting out parts that can be computed once the exit block is known,
parts that contiguously update and that can be checked allowing
for the path to be later identified as FSM with larger limits,
possibly_profitable_path_p, and final checks done when the whole
path is known, profitable_path_p.
I've removed the back_threader_profitability instance from the
back_threader class and instead instantiate it once per path
discovery. I've kept the size compute non-incremental to simplify
the patch and not worry about unwinding.
There's key changes to previous behavior - namely we apply
the param_max_jump_thread_duplication_stmts early only when
we know the path cannot become an FSM one (multiway + thread through
latch) but make sure to elide the path query when we we didn't
yet discover that but are over this limit. Similarly the
speed limit is now used even when we did not yet discover a
hot BB on the path. Basically the idea is to only stop path
discovery when we know the path will never become profitable
but avoid the expensive path range query when we know it's
currently not.
I've done a few cleanups, merging functions, on the way.
* tree-ssa-threadbackward.cc
(back_threader_profitability): Split profitable_path_p
into possibly_profitable_path_p and itself, keep state
as new members.
(back_threader::m_profit): Remove.
(back_threader::find_paths): Likewise.
(back_threader::maybe_register_path): Take profitability
instance as parameter.
(back_threader::find_paths_to_names): Likewise. Use
possibly_profitable_path_p and avoid the path range query
when the path is currently too large.
(back_threader::find_paths): Fold into ...
(back_threader::maybe_thread_block): ... this.
(get_gimple_control_stmt): Remove.
(back_threader_profitability::possibly_profitable_path_p):
Split out from profitable_path_p, do early profitability
checks.
(back_threader_profitability::profitable_path_p): Do final
profitability path after the taken edge has been determined.
Keef Aragon [Wed, 17 Aug 2022 06:45:15 +0000 (08:45 +0200)]
Fix bug in emergency cxa pool free
This probably has never actually affected anyone in practice. The normal
ABI implementation just uses malloc and only falls back to the pool on
malloc failure. But if that happens a bunch of times the freelist gets out
of order which violates some of the invariants of the freelist (as well as
the comments that follow the bug). The bug is just a comparison reversal
when traversing the freelist in the case where the pointer being returned
to the pool is after the existing freelist.
Xi Ruoyao [Tue, 16 Aug 2022 07:34:36 +0000 (15:34 +0800)]
LoongArch: Provide fmin/fmax RTL pattern
We already had smin/smax RTL pattern using fmin/fmax instruction. But
for smin/smax, it's unspecified what will happen if either operand is
NaN. So we would generate calls to libc fmin/fmax functions with
-fno-finite-math-only (the default for all optimization levels expect
-Ofast).
But, LoongArch fmin/fmax instruction is IEEE-754-2008 conformant so we
can also use the instruction for fmin/fmax pattern and avoid the library
function call.
gcc/ChangeLog:
* config/loongarch/loongarch.md (fmax<mode>3): New RTL pattern.
(fmin<mode>3): Likewise.
Tom Honermann [Mon, 1 Aug 2022 18:49:00 +0000 (14:49 -0400)]
c++: Fix pragma suppression of -Wc++20-compat diagnostics [PR106423]
Gcc's '#pragma GCC diagnostic' directives are processed in "early mode"
(see handle_pragma_diagnostic_early) for the C++ frontend and, as such,
require that the target diagnostic option be enabled for the preprocessor
(see c_option_is_from_cpp_diagnostics). This change modifies the
-Wc++20-compat option definition to register it as a preprocessor option
so that its associated diagnostics can be suppressed. The changes also
implicitly disable the option in C++20 and later modes. These changes
are consistent with the definition of the -Wc++11-compat option.
This support is motivated by the need to suppress the following diagnostic
otherwise issued in C++17 and earlier modes due to the char8_t typedef
present in the uchar.h header file in glibc 2.36.
warning: identifier ‘char8_t’ is a keyword in C++20 [-Wc++20-compat]
Tests are added to validate suppression of both -Wc++11-compat and
-Wc++20-compat related diagnostics (fixes were only needed for the C++20
case).
PR c++/106423
gcc/c-family/ChangeLog:
* c-opts.cc (c_common_post_options): Disable -Wc++20-compat
diagnostics in C++20 and later.
* c.opt (Wc++20-compat): Enable hooks for the preprocessor.
Kito Cheng [Mon, 4 Jul 2022 15:09:57 +0000 (23:09 +0800)]
RISC-V: Support zfh and zfhmin extension
Zfh and Zfhmin are extensions for IEEE half precision, both are ratified
in Jan. 2022[1]:
- Zfh has full set of operation like F or D for single or double precision.
- Zfhmin has only provide minimal support for half precision operation,
like conversion, load, store and move instructions.
Kito Cheng [Mon, 4 Jul 2022 12:17:47 +0000 (20:17 +0800)]
RISC-V: Support _Float16 type.
RISC-V decide use _Float16 as primary IEEE half precision type, and this
already become part of psABI, this patch has added folloing support for
_Float16:
- Soft-float support for _Float16.
- Make sure _Float16 available on C++ mode.
- Name mangling for _Float16 on C++ mode.
Kito Cheng [Wed, 10 Aug 2022 13:54:24 +0000 (21:54 +0800)]
soft-fp: Update soft-fp from glibc
This patch is updating all soft-fp from glibc, most changes are
copyright years update, removing "Contributed by" lines and update URL for
license, and changes other than those update are adding conversion
function between IEEE half and 32-bit/64-bit integer, those functions are
required by RISC-V _Float16 support.
Richard Biener [Tue, 16 Aug 2022 12:14:55 +0000 (14:14 +0200)]
Stop backwards thread discovery when leaving a loop
The backward threader copier cannot deal with the situation of
copying blocks belonging to different loops and will reject those
paths late. The following uses this to prune path discovery,
saving on compile-time. Note the off-loop block is still considered
as entry edge origin.
* tree-ssa-threadbackward.cc (back_threader::find_paths_to_names):
Do not walk further if we are leaving the current loop.
`xputenv()` (and `putenv()`) don't copy strings and only store the
pointer in the `environ` global table. As a result `environ` got
corrupted as soon as `jinfo.skipped_makeflags` store got deallocated.
This started causing bootstrap crashes in `execv()` calls:
xgcc: fatal error: cannot execute '/build/build/./prev-gcc/collect2': execv: Bad address
The change restores memory allocation for `xputenv()` argument.
From what I can see, this has been voted in as a DR and as it means
we warn less often than before in -std={gnu,c}++2{0,3} modes or with
-Wvolatile, I wonder if it shouldn't be backported to affected release
branches as well.
2022-08-16 Jakub Jelinek <jakub@redhat.com>
* typeck.cc (cp_build_modify_expr): Implement
P2327R1 - De-deprecating volatile compound operations. Don't warn
for |=, &= or ^= with volatile lhs.
* expr.cc (mark_use) <case MODIFY_EXPR>: Adjust warning wording,
leave out simple.
* g++.dg/cpp2a/volatile1.C: Adjust for de-deprecation of volatile
compound |=, &= and ^= operations.
* g++.dg/cpp2a/volatile3.C: Likewise.
* g++.dg/cpp2a/volatile5.C: Likewise.
Iain Buclaw [Tue, 16 Aug 2022 10:22:10 +0000 (12:22 +0200)]
d: Update DIP links in gdc documentation to point at upstream repository
The wiki links probably worked at some point in the distant past, but
now the official location of tracking all D Improvement Proposals is on
the upstream dlang/DIPs GitHub repository.
PR d/106638
gcc/d/ChangeLog:
* gdc.texi: Update DIP links to point at upstream dlang/DIPs
repository.
gcc/value-range-storage.h:129:40: warning: class with destructor marked 'final' cannot be inherited from [-Wfinal-dtor-non-final-class]
gcc/value-range-storage.h:146:36: warning: class with destructor marked 'final' cannot be inherited from [-Wfinal-dtor-non-final-class]
gcc/ChangeLog:
* value-range-storage.h (class obstack_vrange_allocator): Mark
the class as final.
(class ggc_vrange_allocator): Likewise.
Martin Liska [Tue, 16 Aug 2022 09:15:30 +0000 (11:15 +0200)]
VR: add more virtual dtors
Add 2 virtual destructors in order to address:
gcc/alloc-pool.h:522:5: warning: destructor called on non-final 'value_range_equiv' that has virtual functions but non-virtual destructor [-Wdelete-non-abstract-non-virtual-dtor]
gcc/ggc.h:166:3: warning: destructor called on non-final 'int_range<1>' that has virtual functions but non-virtual destructor [-Wdelete-non-abstract-non-virtual-dtor]
Richard Biener [Tue, 16 Aug 2022 07:43:24 +0000 (09:43 +0200)]
middle-end/106630 - avoid ping-pong between extract_muldiv and match.pd
The following avoids ping-pong between the match.pd pattern changing
(sizetype) ((a_9 + 1) * 48) to (sizetype)(a_9 + 1) * 48 and
extract_muldiv performing the reverse transform by restricting the
match.pd pattern to narrowing conversions as the comment indicates.
Martin Liska [Tue, 16 Aug 2022 09:06:04 +0000 (11:06 +0200)]
VR: add missing override keyworks
Address:
gcc/value-range-equiv.h:57:8: warning: 'set_undefined' overrides a member function but is not marked 'override' [-Winconsistent-missing-override]
gcc/value-range-equiv.h:58:8: warning: 'set_varying' overrides a member function but is not marked 'override' [-Winconsistent-missing-override]
gcc/config/i386/i386-features.h:186:8: warning: 'convert_op' overrides a member function but is not marked 'override' [-Winconsistent-missing-override]
gcc/config/i386/i386-features.h:186:8: warning: 'convert_op' overrides a member function but is not marked 'override' [-Winconsistent-missing-override]
gcc/config/i386/i386-features.h:199:8: warning: 'convert_op' overrides a member function but is not marked 'override' [-Winconsistent-missing-override]
gcc/config/i386/i386-features.h:199:8: warning: 'convert_op' overrides a member function but is not marked 'override' [-Winconsistent-missing-override]
gcc/ChangeLog:
* config/i386/i386-features.h (class general_scalar_chain): Add
final override for a method.
(class timode_scalar_chain): Likewise.
Kewen.Lin [Tue, 16 Aug 2022 05:24:07 +0000 (00:24 -0500)]
rs6000: Adjust mov optabs for opaque modes [PR103353]
As PR103353 shows, we may want to continue to expand built-in
function __builtin_vsx_lxvp, even if we have already emitted
error messages about some missing required conditions. As
shown in that PR, without one explicit mov optab on OOmode
provided, it would call emit_move_insn recursively.
So this patch is to allow the mov pattern to be generated during
expanding phase if compiler has already seen errors.
PR target/103353
gcc/ChangeLog:
* config/rs6000/mma.md (define_expand movoo): Move TARGET_MMA condition
check to preparation statements and add handlings for !TARGET_MMA.
(define_expand movxo): Likewise.
Kewen Lin [Tue, 16 Aug 2022 05:18:51 +0000 (00:18 -0500)]
vect: Don't allow vect_emulated_vector_p type in vectorizable_call [PR106322]
As PR106322 shows, in some cases for some vector type whose
TYPE_MODE is a scalar integral mode instead of a vector mode,
it's possible to obtain wrong target support information when
querying with the scalar integral mode. For example, for the
test case in PR106322, on ppc64 32bit vectorizer gets vector
type "vector(2) short unsigned int" for scalar type "short
unsigned int", its mode is SImode instead of V2HImode. The
target support querying checks umul_highpart optab with SImode
and considers it's supported, then vectorizer further generates
.MULH IFN call for that vector type. Unfortunately it's wrong
to use SImode support for that vector type multiply highpart
here.
This patch is to teach vectorizable_call analysis not to allow
vect_emulated_vector_p type for both vectype_in and vectype_out
as Richi suggested.
PR tree-optimization/106322
gcc/ChangeLog:
* tree-vect-stmts.cc (vectorizable_call): Don't allow
vect_emulated_vector_p type for both vectype_in and vectype_out.
gcc/testsuite/ChangeLog:
* gcc.target/i386/pr106322.c: New test.
* gcc.target/powerpc/pr106322.c: New test.
xtensa: Turn on -fsplit-wide-types-early by default
Since GCC10, the "subreg2" optimization pass was no longer tied to enabling
"subreg1" unless -fsplit-wide-types-early was turned on (PR88233). However
on the Xtensa port, the lack of "subreg2" can degrade the quality of the
output code, especially for those that produce many D[FC]mode pseudos.
This patch turns on -fsplit-wide-types-early by default in order to restore
the previous behavior.
gcc/ChangeLog:
* common/config/xtensa/xtensa-common.cc
(xtensa_option_optimization_table): Add OPT_fsplit_wide_types_early
for OPT_LEVELS_ALL in order to restore pre-GCC10 behavior.
Iain Buclaw [Mon, 15 Aug 2022 17:00:43 +0000 (19:00 +0200)]
d: Defer compiling inline definitions until after the module has finished.
This is to prevent the case of when generating the methods of a struct
type, we don't accidentally emit an inline function that references it,
as the outer struct itself would still be incomplete.
gcc/d/ChangeLog:
* d-tree.h (d_defer_declaration): Declare.
* decl.cc (function_needs_inline_definition_p): Defer checking
DECL_UNINLINABLE and DECL_DECLARED_INLINE_P.
(maybe_build_decl_tree): Call d_defer_declaration instead of
build_decl_tree.
* modules.cc (deferred_inline_declarations): New variable.
(build_module_tree): Set deferred_inline_declarations and a handle
declarations pushed to it.
(d_defer_declaration): New function.
Iain Buclaw [Mon, 15 Aug 2022 15:51:03 +0000 (17:51 +0200)]
d: Fix internal compiler error: Segmentation fault at gimple-expr.cc:88
Because complex types are deprecated in the language, the new way to
expose native complex types is by defining an enum with a basetype of a
library-defined struct that is implicitly treated as-if it is native.
As casts are not implicitly added by the front-end when downcasting from
enum to its underlying type, we must insert an explicit cast during the
code generation pass.
PR d/106623
gcc/d/ChangeLog:
* d-codegen.cc (underlying_complex_expr): New function.
(d_build_call): Handle passing native complex objects as the
library-defined equivalent.
* d-tree.h (underlying_complex_expr): Declare.
* expr.cc (ExprVisitor::visit (DotVarExp *)): Call
underlying_complex_expr instead of build_vconvert.
Iain Buclaw [Mon, 15 Aug 2022 11:43:29 +0000 (13:43 +0200)]
d: Build internal TypeInfo types when module name is "object"
If for whatever reason the module declaration doesn't exist in the
object file, ensure that the internal definitions for TypeInfo and
TypeInfo_Class are still created, otherwise an ICE could occur later if
they are required for a run-time helper call.
gcc/d/ChangeLog:
* d-compiler.cc (Compiler::onParseModule): Call create_tinfo_types
when module name is object.
* typeinfo.cc (create_tinfo_types): Add guard for multiple
invocations.
Iain Buclaw [Mon, 15 Aug 2022 07:54:15 +0000 (09:54 +0200)]
d: Field names of anonymous delegates should be same as regular delegate types.
Doesn't change anything in the code generation or ABI, but makes it
consistent with regular delegates as names would match up when
inspecting tree dumps.
gcc/d/ChangeLog:
* d-codegen.cc (build_delegate_cst): Give anonymous delegate field
names same as per ABI spec.
David Malcolm [Mon, 15 Aug 2022 18:47:03 +0000 (14:47 -0400)]
analyzer: fix direction of -Wanalyzer-out-of-bounds note [PR106626]
Fix a read/write typo.
Also, add more test coverage of -Wanalyzer-out-of-bounds to help
establish a baseline for experiments on tweaking the wording of
the warning (PR analyzer/106626).
gcc/analyzer/ChangeLog:
PR analyzer/106626
* region-model.cc (buffer_overread::emit): Fix copy&paste error in
direction of the access in the note.
gcc/testsuite/ChangeLog:
PR analyzer/106626
* gcc.dg/analyzer/out-of-bounds-read-char-arr.c: New test.
* gcc.dg/analyzer/out-of-bounds-read-int-arr.c: New test.
* gcc.dg/analyzer/out-of-bounds-write-char-arr.c: New test.
* gcc.dg/analyzer/out-of-bounds-write-int-arr.c: New test.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
David Malcolm [Mon, 15 Aug 2022 18:47:02 +0000 (14:47 -0400)]
analyzer: better fix for -Wanalyzer-use-of-uninitialized-value [PR106573]
gcc/analyzer/ChangeLog:
PR analyzer/106573
* region-model.cc (region_model::on_call_pre): Use check_call_args
when ensuring that we call get_arg_svalue on all args. Remove
redundant call from handling for stdio builtins.
Signed-off-by: David Malcolm <dmalcolm@redhat.com>
Roger Sayle [Mon, 15 Aug 2022 16:46:38 +0000 (17:46 +0100)]
Support shifts and rotates by integer constants in TImode STV on x86_64.
This patch adds support for converting 128-bit TImode shifts and rotates
to SSE equivalents using V1TImode during the TImode STV pass.
Previously, only logical shifts by multiples of 8 were handled
(from my patch earlier this month).
As an example of the benefits, the following rotate by 32-bits:
unsigned __int128 a, b;
void rot32() { a = (b >> 32) | (b << 96); }
when compiled on x86_64 with -O2 previously generated:
[which uses a V4SI permutation for those that don't read SSE].
This should help 128-bit cryptography codes, that interleave XORs
with rotations (but that don't use additions or subtractions).
2022-08-15 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog
* config/i386/i386-features.cc
(timode_scalar_chain::compute_convert_gain): Provide costs for
shifts and rotates.
(timode_scalar_chain::convert_insn): Handle ASHIFTRT, ROTATERT
and ROTATE just like existing ASHIFT and LSHIFTRT cases.
(timode_scalar_to_vector_candidate_p): Handle all shifts and
rotates by integer constants between 0 and 127.
gcc/testsuite/ChangeLog
* gcc.target/i386/sse4_1-stv-9.c: New test case.
Roger Sayle [Mon, 15 Aug 2022 16:43:02 +0000 (17:43 +0100)]
Improved gain calculation for COMPARE to 0 or -1 in TImode STV on x86_64.
This patch tweaks timode_scalar_chain::compute_convert_gain to provide
more accurate costs for converting TImode comparisons against zero or
minus 1 to V1TImode equivalents.
2022-08-15 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog
* config/i386/i386-features.cc
(timode_scalar_chain::compute_convert_gain): Provide gains for
comparisons against 0/-1, including "*testti" patterns.
Roger Sayle [Mon, 15 Aug 2022 16:39:47 +0000 (17:39 +0100)]
PR tree-optimization/64992: (B << 2) != 0 is B when B is Boolean.
This patch resolves both PR tree-optimization/64992 and PR
tree-optimization/98956 which are missed optimization enhancement
request, for which Andrew Pinski already has a proposed solution
(related to a fix for PR tree-optimization/98954). Yesterday,
I proposed an alternate improved patch for PR98954, which although
superior in most respects, alas didn't address this case [which
doesn't include a BIT_AND_EXPR], hence this follow-up fix.
For many functions, F(B), of a (zero-one) Boolean value B, the
expression F(B) != 0 can often be simplified to just B. Hence
"(B * 5) != 0" is B, "-B != 0" is B, "bswap(B) != 0" is B,
"(B >>r 3) != 0" is B. These are all currently optimized by GCC,
with the strange exception of left shifts by a constant (possibly
due to the undefined/implementation defined behaviour when the
shift constant is larger than the first operand's precision).
This patch adds support for this particular case, when the shift
constant is valid.
2022-08-15 Roger Sayle <roger@nextmovesoftware.com>
gcc/ChangeLog
PR tree-optimization/64992
PR tree-optimization/98956
* match.pd (ne (lshift @0 @1) 0): Simplify (X << C) != 0 to X
when X is zero_one_valued_p and the shift constant C is valid.
(eq (lshift @0 @1) 0): Likewise, simplify (X << C) == 0 to !X
when X is zero_one_valued_p and the shift constant C is valid.
gcc/testsuite/ChangeLog
PR tree-optimization/64992
* gcc.dg/pr64992.c: New test case.
Roger Sayle [Mon, 15 Aug 2022 16:32:26 +0000 (17:32 +0100)]
PR tree-optimization/71343: Optimize (X<<C)&(Y<<C) as (X&Y)<<C.
This patch is the first part of a solution to PR tree-optimization/71343,
a missed-optimization enhancement request where GCC fails to see that
(a<<2)+(b<<2) == a*4+b*4.
This piece is that (X<<C) op (Y<<C) can be simplified to (X op Y) << C,
for many binary operators, including AND, IOR, XOR, and (if overflow
isn't an issue) PLUS and MINUS. Likewise, the right shifts (both logical
and arithmetic) and bit-wise logical operators can be simplified in a
similar fashion. These all reduce the number of GIMPLE binary operations
from 3 to 2, by combining/eliminating a shift operation.
2022-08-15 Roger Sayle <roger@nextmovesoftware.com>
Richard Biener <rguenther@suse.de>
gcc/ChangeLog
PR tree-optimization/71343
* match.pd (op (lshift @0 @1) (lshift @2 @1)): Optimize the
expression (X<<C) + (Y<<C) to (X+Y)<<C for multiple operators.
(op (rshift @0 @1) (rshift @2 @1)): Likewise, simplify (X>>C)^(Y>>C)
to (X^Y)>>C for binary logical operators, AND, IOR and XOR.
gcc/testsuite/ChangeLog
PR tree-optimization/71343
* gcc.dg/pr71343-1.c: New test case.
Nathan Sidwell [Mon, 15 Aug 2022 14:19:36 +0000 (07:19 -0700)]
c++: Fix module line no testcase
Not all systems have the same injected headers, leading to line
location table differences that are immaterial to the test. Fix the
regexp more robustly.
Richard Biener [Mon, 15 Aug 2022 11:03:54 +0000 (13:03 +0200)]
Simplify range_on_path_entry
I've noticed that range_on_path_entry does mightly complicated things
that don't make sense to me and the commentary might just be
out of date. For the sake of it I replaced it with range_on_entry
and statistics show we thread _more_ jumps with that, so better
not do magic there.
* gimple-range-path.cc (range_on_path_entry): Just
call range_on_entry.
projects / thirdparty / gcc.git / log
