Fix a rejects-(potentially)-valid code for ALLOCATE of CHARACTER with
type-spec, and implement a string-length check for -fcheck=bounds.
Implement more detailed errors or warnings when character function
declarations and references do not match.
PR fortran/53357
gcc/fortran/ChangeLog:
* dependency.cc (gfc_dep_compare_expr): Return correct result if
relationship of expressions could not be determined.
* interface.cc (gfc_check_result_characteristics): Implement error
messages if character function declations and references do not
agree, else emit warning in cases where a mismatch is suspected.
* trans-stmt.cc (gfc_trans_allocate): Implement a string length
check for -fcheck=bounds.
gcc/testsuite/ChangeLog:
* gfortran.dg/auto_char_len_4.f90: Adjust patterns.
* gfortran.dg/typebound_override_1.f90: Likewise.
* gfortran.dg/bounds_check_strlen_10.f90: New test.
The code was passing factor == 0 to vect_get_loop_len which always
returns an unmodified length, even if the number of scalar elements
doesn't agree. It also failed to insert the eventually generated
code.
PR tree-optimization/117594
* tree-vect-loop.cc (vectorizable_live_operation_1): Pass
factor == 1 to vect_get_loop_len, insert generated stmts.
* gcc.dg/vect/pr117594.c: New testcase.
And stage3 begins...
Zdenek's fuzzer caught this one. Essentially using simplify_gen_subreg
directly with an offset of 0 when we just needed a lowpart.
The offset of 0 works for little endian, but for big endian it's simply wrong.
simplify_gen_subreg will return NULL_RTX because the case isn't representable.
We then embed that NULL_RTX into an insn that's later scanned during
mark_jump_label.
Scanning the port I see a couple more instances of this incorrect idiom. One
is pretty obvious to fix. The others look a bit goofy and I'll probably need
to sync with Patrick on them.
Anyway tested on riscv64-elf and riscv32-elf with no regressions. Pushing to
the trunk.
PR target/117595
gcc/
* config/riscv/sync.md (atomic_compare_and_swap<mode>): Use gen_lowpart
rather than simplify_gen_subreg.
* config/riscv/riscv.cc (riscv_legitimize_move): Similarly.
gcc/testsuite/
* gcc.target/riscv/pr117595.c: New test.
This patch ensures that the tokens defining the full declaration of an
ARRAY type is stored in the symbol table and used during production of
error messages.
gcc/m2/ChangeLog:
PR modula2/117660
* gm2-compiler/P2Build.bnf (ArrayType): Update tok with the
composite token produced during array type declaration.
* gm2-compiler/P2SymBuild.mod (EndBuildArray): Create the
combinedtok and store it into the symbol table.
Also ensure combinedtok is pushed to the quad stack.
(BuildFieldArray): Preserve typetok.
* gm2-compiler/SymbolTable.def (PutArray): Rename parameters.
* gm2-compiler/SymbolTable.mod (PutArray): Rename parameters.
gcc/testsuite/ChangeLog:
PR modula2/117660
* gm2/iso/fail/arraymismatch.mod: New test.
Signed-off-by: Gaius Mulley <gaiusmod2@gmail.com>
While these haven't shown up in my tester (not configs I test) and I think
we're likely going to be deprecating the nds32 target. we might as well go
ahead and fix them.
I'm going to include this under the pr117628 umbrella.
PR target/117628
libgcc/
* config/arm/freebsd-atomic.c (bool): Remove unnecessary typedef.
* config/arm/linux-atomic-64bit.c: Likewise.
* config/arm/linux-atomic.c: Likewise.
* config/nds32/linux-atomic.c: Likewise.
* config/nios2/linux-atomic.c: Likewise.
csky fails to build libgcc after the c23 changes because it has a typedef for
bool. AFAICT it's internal to the file, so removing the typedef isn't an ABI
change.
Similiarly for c6x which includes unwind-arm-common.inc. I suspect most, if
not all of the arm-v7 and older targets are failing to build right now.
I've built and regression tested both csky-linux-gnu and c6x-elf with this
change. OK for the trunk?
PR target/117628
libgcc/
* config/csky/linux-atomic.c (bool): Remove unnecessary typedef.
* unwind-arm-common.inc (bool): Similarly.
This patch is a follow on from PR modula2/117371 which could include
a check to enforce the ISO restriction on a zero for loop step.
gcc/m2/ChangeLog:
PR modula2/117371
* gm2-compiler/M2GenGCC.mod (PerformLastForIterator):
Add check for zero step value and issue an error message.
gcc/testsuite/ChangeLog:
PR modula2/117371
* gm2/iso/fail/forloopbyzero.mod: New test.
Signed-off-by: Gaius Mulley <gaiusmod2@gmail.com>
The semantics of the GNAT-specific Predicate aspect should be equivalent
to those of the Static_Predicate aspect when the predicate expression is
static, but that is not correctly implemented for static case expressions.
gcc/ada/ChangeLog:
* exp_ch4.adb (Expand_N_Case_Expression): Remove the test on
enclosing predicate function for the return optimization.
Rewrite it in the general case to catch all nondynamic predicates.
(Expand_N_If_Expression): Remove the test on enclosing predicate
function for the return optimization.
Remove all user-level documentation of the check name
"Atomic_Synchronization". The documentation was confusing because
this check should never be used in source code, and because it
raises the question of whether All_Checks applies to it (it does
not).
Change the name Atomic_Synchronization to be _Atomic_Synchronization
(with a leading underscore) so that it cannot be used in source code.
This "check" is not really a check at all; it is used only internally in
the implementation of Disable/Enable_Atomic_Synchronization, because the
placement and scope of these pragmas match pragma Suppress.
gcc/ada/ChangeLog:
* doc/gnat_rm/implementation_defined_characteristics.rst:
Remove Atomic_Synchronization.
* doc/gnat_ugn/building_executable_programs_with_gnat.rst:
Likewise.
* doc/gnat_rm/implementation_defined_pragmas.rst: DRY.
Consolidate documentation of Disable/Enable_Atomic_Synchronization.
* checks.adb: Comment fix.
* exp_util.ads: Likewise.
* targparm.ads: Likewise.
* types.ads: Likewise.
* gnat1drv.adb: Likewise. DRY.
* sem_prag.adb (Process_Disable_Enable_Atomic_Sync):
Change name of Atomic_Synchronization to start with
underscore.
(Process_Suppress_Unsuppress): No need to check Comes_From_Source for
Atomic_Synchronization anymore; _Atomic_Synchronization can never
come from source. (Anyway, it shouldn't be ignored; it should be
an error.)
* snames.ads-tmpl (Atomic_Synchronization):
Change name to start with underscore.
* switch-c.adb (Scan_Front_End_Switches):
Minor cleanup: Use 'in'.
* gnat_rm.texi: Regenerate.
* gnat_ugn.texi: Regenerate.
In addition to Resolve_Indexed_Component, Eval_Indexed_Component can also
set the Do_Range_Check flag on the expressions of an N_Indexed_Component
node through the call on Check_Non_Static_Context, so this also needs to
be blocked by the Kill_Range_Check flag.
gcc/ada/ChangeLog:
* sem_eval.adb (Eval_Indexed_Component): Clear Do_Range_Check on
the expressions if Kill_Range_Check is set on the node.
This is another glitch associated with Initialization_Statements.
gcc/ada/ChangeLog:
* exp_util.adb (Remove_Init_Call): Rewrite a compound statement in
the Initialization_Statements of the variable as a null statement
instead of removing it.
* freeze.adb (Explode_Initialization_Compound_Statement): Small
comment tweaks.
The processing of static array aggregates in Exp_Aggr requires that their
bounds be representable as Int(eger) values for practical purposes, and
the previous changes have exposed another path where this is not checked.
This introduces a UI_Are_In_Int_Range local predicate for convenience.
gcc/ada/ChangeLog:
* exp_aggr.adb (UI_Are_In_Int_Range): New predicate.
(Aggr_Size_OK): Use it.
(Flatten): Likewise.
(Packed_Array_Aggregate_Handled): Likewise.
(Static_Array_Aggregate): Likewise.
The problem occurs for an anonymous array object declared with an aspect and
when pragma {Initialize,Normalize}_Scalars is in effect: in this case, the
synthesized aggregate is attached to the Initialization_Statements field by
Convert_Aggr_In_Object_Decl, but Explode_Initialization_Compound_Statement
puts it back at the point of declaration instead of the freeze point, thus
voiding the effects of the mechanism.
This was previously hidden because of a bypass in Freeze_Entity which drops
the freeze node on the floor in this case, so the change fixes the issue and
removes the bypass in the process.
gcc/ada/ChangeLog:
* freeze.ads (Explode_Initialization_Compound_Statement): Adjust the
description.
* freeze.adb (Explode_Initialization_Compound_Statement): If the
entity has its freezing delayed, append the initialization actions
to its freeze actions.
(Freeze_Object_Declaration): Remove commented out code.
(Freeze_Entity): Remove bypass for object of anonymous array type.
The exception handler that catches Abort_Signal does nothing nowadays.
This refactors the code to use Build_Abort_Block more consistently and
also makes it simpler by dropping the identifier on the abort block.
No functional changes.
gcc/ada/ChangeLog:
* exp_sel.ads (Build_Abort_Block): Remove second parameter and
rename the third.
(Build_Abort_Block_Handler): Fix description.
* exp_sel.adb (Build_Abort_Block): Remove second parameter, rename
the third and adjust accordingly.
* exp_ch9.adb (Expand_N_Asynchronous_Select): Fix the description
of the exception handler throughout. Remove Abort_Block_Ent and
Hdle local variables. Call Build_Abort_Block consistently to build
the abort block and adjust existing calls.
In some cases an array aggregate with statically known bounds and at least
one bound outside of the range of a 32-bit signed integer causes
a bugbox.
gcc/ada/ChangeLog:
* exp_aggr.adb (Convert_To_Positional.Flatten): Avoid raising
Constraint_Error in UI_To_Int by testing UI_Is_In_Int_Range first.
This mainly decouples the handling of the declaration case from that of the
assignment case in Expand_Array_Aggregate, as well as moves the expansion
in the case of an aggregate that can be processed by the back end to the
Build_Array_Aggr_Code routine.
gcc/ada/ChangeLog:
* exp_aggr.adb (Build_Array_Aggr_Code): Build the simple assignment
for the case of an aggregate that can be handled by the back end.
(Expand_Array_Aggregate): Adjust description of the processing.
Move handling of declaration case to STEP 4 and remove handling of
the case of an aggregate that can be processed by the back end.
(Late_Expansion): Likewise for the second part.
* exp_ch3.adb (Expand_N_Object_Declaration): Deal with a delayed
aggregate synthesized for the default initialization, if any.
* sem_eval.adb (Eval_Indexed_Component): Bail out for the name of
an assignment statement.
This mainly decouples the handling of the allocator case from that of the
assignment case in Expand_Array_Aggregate and also makes Must_Slide a bit
more forgiving.
gcc/ada/ChangeLog:
* exp_aggr.adb (In_Place_Assign_OK): Remove handling of allocators
and call Must_Slide instead of implementing the check manually.
(Convert_To_Assignments): Adjust outdated comment.
(Expand_Array_Aggregate): Move handling of allocator case to STEP 3
and call Must_Slide directly for it.
(Must_Slide): Replace tests based on Is_OK_Static_Expression with
tests based on Compile_Time_Known_Value.
Convert_Array_Aggr_In_Allocator does nothing that Late_Expansion cannot do,
so this deletes the former and moves its support code for Storage_Model to
the latter. No functional changes.
gcc/ada/ChangeLog:
* exp_aggr.adb (Convert_Array_Aggr_In_Allocator): Delete.
(Convert_Aggr_In_Allocator): Do not call above procedure.
(Late_Expansion): Deal with a target that is the dereference of a
prefix with a Storage_Model. Remove an useless actual parameter
in the call to Build_Array_Aggr_Code.
Reverse the meaning of switch -gnatd_P; that is, enable by default
the generating of a runtime check when the prefix of the call is
an access-to-subprogram type with a null value.
gcc/ada/ChangeLog:
* sem_res.adb (Resolve_Actuals): Add by default a null-exclusion
check on the prefix of the call when it is an access-type; it can
be disabled using -gnatd_P.
* debug.adb (gnatd_P): Update documentation.
This patch fixes a crash in the compiler when the actual for an anonymous
access type formal is an 'Access of a Sream_Element_Array object during
the calculation of said actual's accessibility level.
gcc/ada/ChangeLog:
* accessibility.adb (Accessibility_Level): Handle the Input attribute
case
This change is part of an effort to reduce usage of
Is_Predefined_Filename.
gcc/ada/ChangeLog:
* frontend.adb (Frontend): tweak test for predefined main unit.
libgomp/ChangeLog:
* plugin/plugin-gcn.c (GOMP_OFFLOAD_openacc_async_construct): In
case of an error, call GOMP_PLUGIN_fatal not ..._error; use NULL
not false in return.
The test file pr117093.c failed on platforms other than aarch64, because
it uses arm_neon.h. We moved it into gcc.target/aarch64.
The patch was bootstrapped and tested on aarch64-linux-gnu and
x86_64-linux-gnu, no regression.
Committed as obvious.
Signed-off-by: Jennifer Schmitz <jschmitz@nvidia.com>
gcc/testsuite/
PR tree-optimization/117093
* gcc.dg/tree-ssa/pr117093.c: Move to gcc.target/aarch64.
* gcc.target/aarch64/pr117093.c: New test.
This patch adds an undefined else operand to the masked loads.
gcc/ChangeLog:
* config/gcn/predicates.md (maskload_else_operand): New
predicate.
* config/gcn/gcn-valu.md: Use new predicate.
This adds zero else operands to masked loads and their intrinsics.
I needed to adjust more than initially thought because we rely on
combine for several instructions and a change in a "base" pattern
needs to propagate to all those.
gcc/ChangeLog:
* config/aarch64/aarch64-sve-builtins-base.cc: Add else
handling.
* config/aarch64/aarch64-sve-builtins.cc (function_expander::use_contiguous_load_insn):
Ditto.
* config/aarch64/aarch64-sve-builtins.h: Add else operand to
contiguous load.
* config/aarch64/aarch64-sve.md (@aarch64_load<SVE_PRED_LOAD:pred_load>
_<ANY_EXTEND:optab><SVE_HSDI:mode><SVE_PARTIAL_I:mode>):
Split and add else operand.
(@aarch64_load_<ANY_EXTEND:optab><SVE_HSDI:mode><SVE_PARTIAL_I:mode>):
Ditto.
(*aarch64_load_<ANY_EXTEND:optab>_mov<SVE_HSDI:mode><SVE_PARTIAL_I:mode>):
Ditto.
* config/aarch64/aarch64-sve2.md: Ditto.
* config/aarch64/iterators.md: Remove unused iterators.
* config/aarch64/predicates.md (aarch64_maskload_else_operand):
Add zero else operand.
This patch adds an else operand to vectorized masked load calls.
The current implementation adds else-value arguments to the respective
target-querying functions that is used to supply the vectorizer with the
proper else value.
We query the target for its supported else operand and uses that for the
maskload call. If necessary, i.e. if the mode has padding bits and if
the else operand is nonzero, a VEC_COND enforcing a zero else value is
emitted.
gcc/ChangeLog:
* optabs-query.cc (supports_vec_convert_optab_p): Return icode.
(get_supported_else_val): Return supported else value for
optab's operand at index.
(supports_vec_gather_load_p): Add else argument.
(supports_vec_scatter_store_p): Ditto.
* optabs-query.h (supports_vec_gather_load_p): Ditto.
(get_supported_else_val): Ditto.
* optabs-tree.cc (target_supports_mask_load_store_p): Ditto.
(can_vec_mask_load_store_p): Ditto.
(target_supports_len_load_store_p): Ditto.
(get_len_load_store_mode): Ditto.
* optabs-tree.h (target_supports_mask_load_store_p): Ditto.
(can_vec_mask_load_store_p): Ditto.
* tree-vect-data-refs.cc (vect_lanes_optab_supported_p): Ditto.
(vect_gather_scatter_fn_p): Ditto.
(vect_check_gather_scatter): Ditto.
(vect_load_lanes_supported): Ditto.
* tree-vect-patterns.cc (vect_recog_gather_scatter_pattern):
Ditto.
* tree-vect-slp.cc (vect_get_operand_map): Adjust indices for
else operand.
(vect_slp_analyze_node_operations): Skip undefined else operand.
* tree-vect-stmts.cc (exist_non_indexing_operands_for_use_p):
Add else operand handling.
(vect_get_vec_defs_for_operand): Handle undefined else operand.
(check_load_store_for_partial_vectors): Add else argument.
(vect_truncate_gather_scatter_offset): Ditto.
(vect_use_strided_gather_scatters_p): Ditto.
(get_group_load_store_type): Ditto.
(get_load_store_type): Ditto.
(vect_get_mask_load_else): Ditto.
(vect_get_else_val_from_tree): Ditto.
(vect_build_one_gather_load_call): Add zero else operand.
(vectorizable_load): Use else operand.
* tree-vectorizer.h (vect_gather_scatter_fn_p): Add else
argument.
(vect_load_lanes_supported): Ditto.
(vect_get_mask_load_else): Ditto.
(vect_get_else_val_from_tree): Ditto.
When predicating a load we implicitly assume that the else value is
zero. This matters in case the loaded value is padded (like e.g.
a Bool) and we must ensure that the padding bytes are zero on targets
that don't implicitly zero inactive elements.
A former version of this patch still had this handling in ifcvt but
the latest version defers it to the vectorizer.
gcc/ChangeLog:
* tree-if-conv.cc (predicate_load_or_store): Add zero else
operand and comment.
This patch amends the documentation for masked loads (maskload,
vec_mask_load_lanes, and mask_gather_load as well as their len
counterparts) with an else operand.
gcc/ChangeLog:
* doc/md.texi: Document masked load else operand.
libgomp/ChangeLog:
PR libgomp/117626
* plugin/plugin-nvptx.c (nvptx_open_device): Use 'CUDA_CALL_ERET'
with 'NULL' as error return instead of 'CUDA_CALL' that returns false.
For pointers I forgot that BIT_IOR_EXPR is not valid so when
I added the pattern to convert `max<a,b> != 0` (r15-5356), GCC
would start to ICEing saying pointer types were not valid for
BIT_IOR_EXPR.
This fixes the problem by casting to the unsigned type of the
inner type. There was another way of fixing this to handling it
as `a == 0 & b == 0` but both match and reassoication (for pointers)
will then convert it back into the form I am creating here so
let's just use that form instead.
Bootstrapped and tested on x86_64-linux-gnu.
PR tree-optimization/117646
gcc/ChangeLog:
* match.pd (`max<a,b>==0`): Add casts to `unsigned type`.
gcc/testsuite/ChangeLog:
* gcc.dg/torture/minmaxneeqptr-1.c: New test.
Signed-off-by: Andrew Pinski <quic_apinski@quicinc.com>
François pointed out that static_cast<__node_ptr>(&_M_before_begin) is
invalid, because _M_before_begin is only a node-base not a node.
Refactor the new merge overloads to only cast when we know we have a
valid node.
He also pointed out some optimizations to allow reusing hash codes that
might be cached in the node. The _M_src_hash_code function already has
the right logic to decide when a cached hash code can be reused by a
different _Hashtable object.
libstdc++-v3/ChangeLog:
* include/bits/hashtable.h (_Hashtable::_M_src_hash_code):
Improve comments.
(_Hashtable::_M_merge_unique(_Hashtable&)): Use pointer_traits
to get before-begin pointer. Only use static_cast on valid
nodes, not the before-begin pointer. Reuse a hash code cached in
the node when possible.
(_Hashtable::_M_merge_multi(_Hashtable&)): Likewise.
Reviewed-by: François Dumont <fdumont@gcc.gnu.org>
The dependency output for header unit modules is based on the absolute
pathname of the header file, but that's not something that a makefile can
portably refer to. This patch adds a .c++-header-unit target based on the
header name relative to an element of the include path.
libcpp/ChangeLog:
* internal.h (_cpp_get_file_dir): Declare.
* files.cc (_cpp_get_file_dir): New fn.
* mkdeps.cc (make_write): Use it.
gcc/testsuite/ChangeLog:
* g++.dg/modules/dep-4.H: New test.
Instead of doing a dg-run with a specific target check for linux.
Use signal as the effective-target since this requires the use
of ALARM signal to do the testing.
Also use check_vect in the main and renames main to main1 to make sure
we don't use the registers.
Tested on x86_64-linux-gnu.
PR testsuite/117494
gcc/testsuite/ChangeLog:
* gcc.dg/vect/pr101145inf.c: Remove dg-do and replace
with dg-require-effective-target of signal.
* gcc.dg/vect/pr101145inf_1.c: Likewise.
* gcc.dg/vect/pr101145inf.inc: Rename main to main1
and mark as noinline.
Include tree-vect.h. Have main call check_vect and main1.
Signed-off-by: Andrew Pinski <quic_apinski@quicinc.com>
I was looking at a regression in ext-dce's behavior just before Cauldron.
Essentially a bugfix in ext-dce ended up causing us to fail to eliminate some
useless extensions.
When we have a SUBREG object with SUBREG_PROMOTED_VAR* flags set, we generally
have to be more conservative in how we process bit group liveness, making bits
live that wouldn't obviously be live otherwise.
That's not always necessary though. For example, if we're storing a promoted
subreg into memory, we may not care about those extra live bits on this
instance of the subreg object (remember subregs are not shared!). Essentially
if the mode of the memory reference is not wider than the mode of the inner
REG, then we can clear the promoted state which in turn may allow more
extension elimination.
So at the start of ext-dce we do a simple pass over the IL and remove promoted
subreg state when it's obviously safe to do so (memory stores when the modes
allow it). That prevents extra bits from being live and ultimately allows us
to remove more useless extensions.
The testcase is in theory generic, but many targets won't have an opportunity
to optimize this case. So rather then build out a large inclusion/exclusion
list, I've just made the test risc-v specific.
Bootstrapped and regression tested on aarch64, riscv64, s390x, etc in my tester.
gcc/
* ext-dce.cc (maybe_clear_subreg_promoted_p): New function.
(ext_dce_execute): Call it.
gcc/testsuite
* gcc.target/riscv/ext-dce-1.c: New test.
Remove stray `;;' from the middle of the introductory comment for the
"unaligned_store<mode>" expander, clearly a leftover from a previous
edition.
gcc/
* config/alpha/alpha.md (unaligned_store<mode>): Remove stray
`;;'.
2024-11-17 John David Anglin <danglin@gcc.gnu.org>
gcc/ChangeLog:
PR target/69374
* doc/install.texi (Specific) <hppa*-hp-hpux11>: Update anchor
and heading to reflect removal of 32-bit hppa support on HP-UX.
Trim 32-bit related text.
This should have been part of r15-5367. One day I'll remember to do this
before buildbot sends me hate mail.
gcc/c-family/ChangeLog:
* c.opt.urls: Regenerate.
In C23, bool is now a keyword. So, doing a typedef for it is invalid.
2024-11-17 John David Anglin <danglin@gcc.gnu.org>
libgcc/ChangeLog:
PR target/117627
* config/pa/linux-atomic.c: Remove typedef for bool type.
This warning covers the C23 incompibilities resulting from using
() as parameter lists in function declarations.
The warning name comes from Clang. The implementation is not
perfect because GCC treats these two declarations as equivalent:
void f ();
void f (not_a_type);
This is a bit confusing because they are clearly visually distinct.
However, as of GCC 14, the second form is an error by default, so
treating both the same as far as -Wdeprecated-non-prototype does
not seem so bad from a user experience view.
gcc/c-family/
PR c/95445
* c-opts.cc (c_common_post_options): Initialize
warn_deprecated_non_prototype.
* c.opt (Wdeprecated-non-prototype): New option.
* c.opt.urls: Regenerate.
gcc/c/
PR c/95445
* c-decl.cc (start_function): Warn about parameters
after parameter-less declaration.
* c-typeck.cc (build_function_call_vec): Pass fntype
to convert_arguments.
(convert_arguments): Change argument to fntype and
compute typelist. Warn about parameter list mismatches
on first parameter.
gcc/
PR c/95445
* doc/invoke.texi: Document -Wdeprecated-non-prototype.
gcc/testsuite/
PR c/95445
* gcc.dg/Wdeprecated-non-prototype-1.c: New test.
* gcc.dg/Wdeprecated-non-prototype-2.c: New test.
* gcc.dg/Wdeprecated-non-prototype-3.c: New test.
* gcc.dg/Wdeprecated-non-prototype-4.c: New test.
While experimenting with testing module std I noticed that gcc -M broke on
it; it seems I need to set directives_only even sooner than I did in
r15-4219.
gcc/c-family/ChangeLog:
* c-ppoutput.cc (preprocess_file): Don't set directives_only here.
gcc/cp/ChangeLog:
* module.cc (module_preprocess_options): Set directives_only here.
The C++ modules code has a -fmodule-header (or -x c++-{user,system}-header)
option to specify looking up headers to compile to header units on the usual
include paths. I'd like to have the same functionality for full C++20
modules such as module std, which I proposed to live on the include path at
bits/std.cc. But this behavior doesn't seem necessarily connected to
modules, so I'm proposing a general C/C++ option to specify the behavior of
looking in the include path for the input files specified on the command
line.
Other ideas for the name of the option are very welcome.
The libcpp change is to allow -fsearch-include-path{,=user} to find files in
the current working directory, like -include. This can be handy for a quick
compile of both std.cc and a file that imports it, e.g.
g++ -std=c++20 -fmodules -fsearch-include-path bits/std.cc importer.cc
gcc/ChangeLog:
* doc/cppopts.texi: Document -fsearch-include-path.
* doc/invoke.texi: Mention it for modules.
gcc/c-family/ChangeLog:
* c.opt: Add -fsearch-include-path.
* c-opts.cc (c_common_post_options): Handle it.
gcc/cp/ChangeLog:
* module.cc (module_preprocess_options): Don't override it.
libcpp/ChangeLog:
* internal.h (search_path_head): Declare.
* files.cc (search_path_head): No longer static.
* init.cc (cpp_read_main_file): Use it.
This patch introduces an installed source form of module std and std.compat.
To help a build system find them, we install a libstdc++.modules.json file
alongside libstdc++.so, which tells the build system where the files are and
any special flags it should use when compiling them (none, in this case).
The format is from a proposal in SG15. The build system can find this file
with 'gcc -print-file-name=libstdc++.modules.json'.
It seems preferable to use a relative path from this file to the sources so
that moving the installation doesn't break the reference, but I didn't see
any obvious way to compute that without relying on coreutils, perl, or
python, so I wrote a POSIX shell script for it. The .. canonicalization
bits aren't necessary since I discovered $(abspath), but I guess I might as
well leave them in.
Currently this installs the sources under $(gxx_include_dir)/bits/,
i.e. /usr/include/c++/15/bits. So with my -fsearch-include-path change,
std.cc can be compiled with g++ -fsearch-include-path bits/std.cc. Note
that if someone actually tries to #include <bits/std.cc> it will fail with
"error: module control-line cannot be in included file".
Any ideas about a more user-friendly way to express "compile module std" are
welcome.
The sources currently have the extension .cc, like other source files.
std.cc started with m.cencora's implementation in PR114600. I've made some
adjustments, but more is probably desirable, e.g. of the <algorithm>
handling of namespace ranges, and to remove exports of templates that are
only specialized in a particular header. I've filled in a bunch of missing
exports, and added some FIXMEs where I noticed bits that are not implemented
yet.
Since bits/stdc++.h also intends to include the whole standard library, I
include it rather than duplicate it. But stdc++.h comments out <execution>,
due to TBB issues; I include it separately and suppress TBB usage, so module
std won't currently provide parallel execution.
It seemed most convenient for the two files to be monolithic so we don't
need to worry about include paths. So the C library names that module
std.compat exports in both namespace std and :: are a block of code that is
appended to both files, adjusted based on whether the macro STD_COMPAT is
defined before the block.
In this implementation std.compat imports std; it would also be valid for it
to duplicate everything in std. I see the libc++ std.compat also imports
std.
As discussed in the PR, module std is supported in C++20 mode even though it
was added in C++23.
Changes to test module std will follow in a separate patch. In my testing
I've noticed a few compiler bugs that break various testcases, so I don't
expect to enable module std testing by default at first.
PR libstdc++/106852
libstdc++-v3/ChangeLog:
* include/bits/version.def: Add __cpp_lib_modules.
* include/bits/version.h: Regenerate.
* src/c++23/Makefile.am: Add modules std and std.compat.
* src/c++23/Makefile.in: Regenerate.
* src/c++23/std-clib.cc.in: New file.
* src/c++23/std.cc.in: New file.
* src/c++23/std.compat.cc.in: New file.
* src/c++23/libstdc++.modules.json.in: New file.
contrib/ChangeLog:
* relpath.sh: New file.
