| blob | 0e1ff28bdfd43584ed28d7141e09413955852a3e |
1 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
3 PR fortran/90608
4 * invoke.texi(finline-intrinsics): Document new flag.
5 * lang.opt (finline-intrinsics, finline-intrinsics=,
6 fno-inline-intrinsics): New flags.
7 * options.cc (gfc_post_options): If the option variable controlling
8 the inlining of MAXLOC (respectively MINLOC) has not been set, set
9 it or clear it depending on the optimization option variables.
10 * trans-intrinsic.cc (gfc_inline_intrinsic_function_p): Return false
11 if inlining for the intrinsic is disabled according to the option
12 variable.
14 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
16 PR fortran/90608
17 * trans-intrinsic.cc (gfc_conv_intrinsic_minmaxloc): Generate a set
18 of index variables. Set them using the loop indexes before leaving
19 the first set of loops. Generate a new loop entry predicate.
20 Initialize it. Set it before leaving the first set of loops. Clear
21 it in the body of the second set of loops. For the second set of
22 loops, update each loop lower bound to use the corresponding index
23 variable if the predicate variable is set.
25 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
27 PR fortran/90608
28 * trans-intrinsic.cc (gfc_conv_intrinsic_minmaxloc): Initialize
29 and update all the variables. Put the label and goto in the
30 outermost scalarizer loop. Don't start the second loop where the
31 first stopped.
32 (gfc_inline_intrinsic_function_p): Also return TRUE for array MASK
33 or for any REAL type.
35 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
37 PR fortran/90608
38 * trans-intrinsic.cc (gfc_conv_intrinsic_minmaxloc): Generate
39 variable initialization for each dimension in the else branch of
40 the toplevel condition.
41 (gfc_inline_intrinsic_function_p): Return TRUE for scalar MASK.
43 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
45 PR fortran/90608
46 * trans-array.cc (gfc_conv_ss_startstride): Set the MINLOC/MAXLOC
47 result upper bound using the rank of the ARRAY argument. Ajdust
48 the error message for intrinsic result arrays. Only check array
49 bounds for array references. Move bound check decision code...
50 (bounds_check_needed): ... here as a new predicate. Allow bound
51 check for MINLOC/MAXLOC intrinsic results.
52 * trans-intrinsic.cc (gfc_conv_intrinsic_minmaxloc): Change the
53 result array upper bound to the rank of ARRAY. Update the NONEMPTY
54 variable to depend on the non-empty extent of every dimension. Use
55 one variable per dimension instead of a single variable for the
56 position and the offset. Update their declaration, initialization,
57 and update to affect the variable of each dimension. Use the first
58 variable only in areas only accessed with rank 1 ARRAY argument.
59 Set every element of the result using its corresponding variable.
60 (gfc_inline_intrinsic_function_p): Return true for integral ARRAY
61 and absent DIM and MASK.
63 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
65 * trans-array.cc (gfc_conv_ss_startstride): Move array bound check
66 generation code...
67 (add_check_section_in_array_bounds): ... here as a new function.
69 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
71 * frontend-passes.cc (optimize_minmaxloc): Remove.
72 (optimize_expr): Remove dispatch to optimize_minmaxloc.
74 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
76 PR fortran/90608
77 * trans-array.cc (gfc_conv_ss_startstride): Set the scalarization
78 rank based on the MINLOC/MAXLOC rank if needed. Call the inline
79 code generation and setup the scalarizer array descriptor info
80 in the MINLOC and MAXLOC cases.
81 * trans-intrinsic.cc (gfc_conv_intrinsic_minmaxloc): Return the
82 result array element if the scalarizer is setup and we are inside
83 the loops. Restrict library function call dispatch to the case
84 where inline expansion is not supported. Declare an array result
85 if the expression isn't scalar. Initialize the array result single
86 element and return the result variable if the expression isn't
87 scalar.
88 (walk_inline_intrinsic_minmaxloc): New function.
89 (walk_inline_intrinsic_function): Add MINLOC and MAXLOC cases,
90 dispatching to walk_inline_intrinsic_minmaxloc.
91 (gfc_add_intrinsic_ss_code): Add MINLOC and MAXLOC cases.
92 (gfc_inline_intrinsic_function_p): Return true if ARRAY has rank 1,
93 regardless of DIM.
95 2024-09-21 Mikael Morin <mikael@gcc.gnu.org>
97 PR fortran/90608
98 * frontend-passes.cc (optimize_minmaxloc): Skip if we can generate
99 inline code for the unmodified expression.
100 * trans-intrinsic.cc (gfc_inline_intrinsic_function_p): Add
101 MINLOC and MAXLOC cases.
103 2024-09-20 David Malcolm <dmalcolm@redhat.com>
105 PR other/116613
106 * error.cc: Include "diagnostic-format-text.h".
107 (gfc_diagnostic_starter): Rename to...
108 (gfc_diagnostic_text_starter): ...this. Convert first
109 param from diagnostic_context * to diagnostic_text_output_format &
110 and update accordingly.
111 (gfc_diagnostic_finalizer, gfc_diagnostic_text_finalizer):
112 Likewise.
113 (gfc_diagnostics_init): Update for renamings.
114 (gfc_diagnostics_finish): Likewise.
116 2024-09-19 Andre Vehreschild <vehre@gcc.gnu.org>
118 PR fortran/106606
119 * trans-types.cc (gfc_get_derived_type): Only build non-pointer
120 derived types as component's types when they are not yet built.
122 2024-09-19 Andre Vehreschild <vehre@gcc.gnu.org>
124 PR fortran/85002
125 * trans-array.cc (duplicate_allocatable_coarray): Allow adding
126 of deep copy code in the when-allocated case. Add bounds
127 computation before condition, because coarrays need the bounds
128 also when not allocated.
129 (structure_alloc_comps): Duplication in the coarray case is done
130 already, omit it. Add the deep-code when duplication a coarray.
131 * trans-expr.cc (gfc_trans_structure_assign): Fix comment.
133 2024-09-15 Mikael Morin <mikael@gcc.gnu.org>
135 * trans-expr.cc (gfc_advance_se_ss_chain): Don't use an element
136 from the parent scalarization chain when the current chain reaches
137 its end.
139 2024-09-13 Tobias Burnus <tburnus@baylibre.com>
141 PR fortran/116661
142 * gfortran.h (gfc_omp_namelist): Rename 'init' members for clarity.
143 * match.cc (gfc_free_omp_namelist): Handle renaming.
144 * dump-parse-tree.cc (show_omp_namelist): Update for new format
145 and features.
146 * openmp.cc (gfc_match_omp_prefer_type): Parse list to 'fr' and 'attr';
147 store 'fr' values as integer.
148 (gfc_match_omp_init): Rename variable names.
150 2024-09-12 Steven G. Kargl <kargl@gcc.gnu.org>
152 * module.cc (bt_types): Add BT_UNSIGNED.
154 2024-09-11 Tobias Burnus <tburnus@baylibre.com>
156 PR fortran/116661
157 * openmp.cc (gfc_match_omp_prefer_type): NULL init a gfc_expr
158 variable and use right locus in gfc_error.
160 2024-09-09 David Malcolm <dmalcolm@redhat.com>
162 * cpp.cc (cb_cpp_diagnostic_cpp_option): Convert return type from
163 "int" to "diagnostic_option_id".
164 (cb_cpp_diagnostic): Update for renaming of
165 diagnostic_override_option_index to diagnostic_set_option_id.
166 * error.cc (gfc_warning): Update for renaming of diagnostic_info
167 field.
168 (gfc_warning_now_at): Likewise.
169 (gfc_warning_now): Likewise.
170 (gfc_warning_internal): Likewise.
172 2024-09-09 David Malcolm <dmalcolm@redhat.com>
174 PR other/116613
175 * error.cc (gfc_clear_pp_buffer): Rename diagnostic_context's
176 "printer" field to "m_printer".
177 (gfc_warning): Likewise.
178 (gfc_diagnostic_build_kind_prefix): Likewise.
179 (gfc_diagnostic_build_locus_prefix): Likewise.
180 (gfc_diagnostic_starter): Likewise.
181 (gfc_diagnostic_starter): Likewise.
182 (gfc_diagnostic_start_span): Likewise.
183 (gfc_diagnostic_finalizer): Likewise.
184 (gfc_warning_check): Likewise.
185 (gfc_error_opt): Likewise.
186 (gfc_error_check): Likewise.
188 2024-09-07 Thomas Koenig <tkoenig@gcc.gnu.org>
190 * arith.cc (gfc_reduce_unsigned): New function.
191 (gfc_arith_error): Add ARITH_UNSIGNED_TRUNCATED and
192 ARITH_UNSIGNED_NEGATIVE.
193 (gfc_arith_init_1): Initialize unsigned types.
194 (gfc_check_unsigned_range): New function.
195 (gfc_range_check): Handle unsigned types.
196 (gfc_arith_uminus): Likewise.
197 (gfc_arith_plus): Likewise.
198 (gfc_arith_minus): Likewise.
199 (gfc_arith_times): Likewise.
200 (gfc_arith_divide): Likewise.
201 (gfc_compare_expr): Likewise.
202 (eval_intrinsic): Likewise.
203 (gfc_int2int): Also convert unsigned.
204 (gfc_uint2uint): New function.
205 (gfc_int2uint): New function.
206 (gfc_uint2int): New function.
207 (gfc_uint2real): New function.
208 (gfc_uint2complex): New function.
209 (gfc_real2uint): New function.
210 (gfc_complex2uint): New function.
211 (gfc_log2uint): New function.
212 (gfc_uint2log): New function.
213 * arith.h (gfc_int2uint, gfc_uint2uint, gfc_uint2int, gfc_uint2real):
214 Add prototypes.
215 (gfc_uint2complex, gfc_real2uint, gfc_complex2uint, gfc_log2uint):
216 Likewise.
217 (gfc_uint2log): Likewise.
218 * check.cc (gfc_boz2uint): New function
219 (type_check2): New function.
220 (int_or_real_or_unsigned_check): New function.
221 (less_than_bitsizekind): Adjust for unsingeds.
222 (less_than_bitsize2): Likewise.
223 (gfc_check_allocated): Likewise.
224 (gfc_check_mod): Likewise.
225 (gfc_check_bge_bgt_ble_blt): Likewise.
226 (gfc_check_bitfcn): Likewise.
227 (gfc_check_digits): Likewise.
228 (gfc_check_dshift): Likewise.
229 (gfc_check_huge): Likewise.
230 (gfc_check_iu): New function.
231 (gfc_check_iand_ieor_ior): Adjust for unsigneds.
232 (gfc_check_ibits): Likewise.
233 (gfc_check_uint): New function.
234 (gfc_check_ishft): Adjust for unsigneds.
235 (gfc_check_ishftc): Likewise.
236 (gfc_check_min_max): Likewise.
237 (gfc_check_merge_bits): Likewise.
238 (gfc_check_selected_int_kind): Likewise.
239 (gfc_check_shift): Likewise.
240 (gfc_check_mvbits): Likewise.
241 (gfc_invalid_unsigned_ops): Likewise.
242 * decl.cc (gfc_match_decl_type_spec): Likewise.
243 * dump-parse-tree.cc (show_expr): Likewise.
244 * expr.cc (gfc_get_constant_expr): Likewise.
245 (gfc_copy_expr): Likewise.
246 (gfc_extract_int): Likewise.
247 (numeric_type): Likewise.
248 * gfortran.h (enum arith): Extend with ARITH_UNSIGNED_TRUNCATED
249 and ARITH_UNSIGNED_NEGATIVE.
250 (enum gfc_isym_id): Extend with GFC_ISYM_SU_KIND and GFC_ISYM_UINT.
251 (gfc_check_unsigned_range): New prototype-
252 (gfc_arith_error): Likewise.
253 (gfc_reduce_unsigned): Likewise.
254 (gfc_boz2uint): Likewise.
255 (gfc_invalid_unsigned_ops): Likewise.
256 (gfc_convert_mpz_to_unsigned): Likewise.
257 * gfortran.texi: Add some rudimentary documentation.
258 * intrinsic.cc (gfc_type_letter): Adjust for unsigneds.
259 (add_functions): Add uint and adjust functions to be called.
260 (add_conversions): Add unsigned conversions.
261 (gfc_convert_type_warn): Adjust for unsigned.
262 * intrinsic.h (gfc_check_iu, gfc_check_uint, gfc_check_mod, gfc_simplify_uint,
263 gfc_simplify_selected_unsigned_kind, gfc_resolve_uint): New prototypes.
264 * invoke.texi: Add -funsigned.
265 * iresolve.cc (gfc_resolve_dshift): Handle unsigneds.
266 (gfc_resolve_iand): Handle unsigneds.
267 (gfc_resolve_ibclr): Handle unsigneds.
268 (gfc_resolve_ibits): Handle unsigneds.
269 (gfc_resolve_ibset): Handle unsigneds.
270 (gfc_resolve_ieor): Handle unsigneds.
271 (gfc_resolve_ior): Handle unsigneds.
272 (gfc_resolve_uint): Handle unsigneds.
273 (gfc_resolve_merge_bits): Handle unsigneds.
274 (gfc_resolve_not): Handle unsigneds.
275 * lang.opt: Add -funsigned.
276 * libgfortran.h: Add BT_UNSIGNED.
277 * match.cc (gfc_match_type_spec): Match UNSIGNED.
278 * misc.cc (gfc_basic_typename): Add UNSIGNED.
279 (gfc_typename): Likewise.
280 * primary.cc (convert_unsigned): New function.
281 (match_unsigned_constant): New function.
282 (gfc_match_literal_constant): Handle unsigned.
283 * resolve.cc (resolve_operator): Handle unsigned.
284 (resolve_ordinary_assign): Likewise.
285 * simplify.cc (convert_mpz_to_unsigned): Renamed to...
286 (gfc_convert_mpz_to_unsigned): and adjusted.
287 (gfc_simplify_bit_size): Adjusted for unsigned.
288 (compare_bitwise): Likewise.
289 (gfc_simplify_bge): Likewise.
290 (gfc_simplify_bgt): Likewise.
291 (gfc_simplify_ble): Likewise.
292 (gfc_simplify_blt): Likewise.
293 (simplify_cmplx): Likewise.
294 (gfc_simplify_digits): Likewise.
295 (simplify_dshift): Likewise.
296 (gfc_simplify_huge): Likewise.
297 (gfc_simplify_iand): Likewise.
298 (gfc_simplify_ibclr): Likewise.
299 (gfc_simplify_ibits): Likewise.
300 (gfc_simplify_ibset): Likewise.
301 (gfc_simplify_ieor): Likewise.
302 (gfc_simplify_uint): Likewise.
303 (gfc_simplify_ior): Likewise.
304 (simplify_shift): Likewise.
305 (gfc_simplify_ishftc): Likewise.
306 (gfc_simplify_merge_bits): Likewise.
307 (min_max_choose): Likewise.
308 (gfc_simplify_mod): Likewise.
309 (gfc_simplify_modulo): Likewise.
310 (gfc_simplify_popcnt): Likewise.
311 (gfc_simplify_range): Likewise.
312 (gfc_simplify_selected_unsigned_kind): Likewise.
313 (gfc_convert_constant): Likewise.
314 * target-memory.cc (size_unsigned): New function.
315 (gfc_element_size): Adjust for unsigned.
316 * trans-const.h (gfc_conv_mpz_unsigned_to_tree): Add prototype.
317 * trans-const.cc (gfc_conv_mpz_unsigned_to_tree): Handle unsigneds.
318 (gfc_conv_constant_to_tree): Likewise.
319 * trans-decl.cc (gfc_conv_cfi_to_gfc): Put in "not yet implemented".
320 * trans-expr.cc (gfc_conv_gfc_desc_to_cfi_desc): Likewise.
321 * trans-stmt.cc (gfc_trans_integer_select): Handle unsigned.
322 (gfc_trans_select): Likewise.
323 * trans-intrinsic.cc (gfc_conv_intrinsic_mod): Handle unsigned.
324 (gfc_conv_intrinsic_shift): Likewise.
325 (gfc_conv_intrinsic_function): Add GFC_ISYM_UINT.
326 * trans-io.cc (enum iocall): Add IOCALL_X_UNSIGNED and IOCALL_X_UNSIGNED_WRITE.
327 (gfc_build_io_library_fndecls): Add transfer_unsigned and transfer_unsigned_write.
328 (transfer_expr): Handle unsigneds.
329 * trans-types.cc (gfc_unsinged_kinds): New array.
330 (gfc_unsigned_types): Likewise.
331 (gfc_init_kinds): Handle them.
332 (validate_unsigned): New function.
333 (gfc_validate_kind): Use it.
334 (gfc_build_unsigned_type): New function.
335 (gfc_init_types): Use it.
336 (gfc_get_unsigned_type): New function.
337 (gfc_typenode_for_spec): Handle unsigned.
338 * trans-types.h (gfc_get_unsigned_type): New prototype.
340 2024-09-06 Tobias Burnus <tburnus@baylibre.com>
342 * dump-parse-tree.cc (show_omp_namelist): Handle OMP_LIST_INIT.
343 (show_omp_clauses): Handle OMP_LIST_{INIT,USE,DESTORY}; move 'nowait'
344 from end-directive to the directive dump.
345 (show_omp_node, show_code_node): Handle EXEC_OMP_INTEROP.
346 * gfortran.h (enum gfc_statement): Add ST_OMP_INTEROP.
347 (OMP_LIST_INIT, OMP_LIST_USE, OMP_LIST_DESTROY): Add.
348 (enum gfc_exec_op): Add EXEC_OMP_INTEROP.
349 (struct gfc_omp_namelist): Add interop items to union.
350 (gfc_free_omp_namelist): Add boolean arg.
351 * match.cc (gfc_free_omp_namelist): Update to free
352 interop union members.
353 * match.h (gfc_match_omp_interop): New.
354 * openmp.cc (gfc_omp_directives): Uncomment 'interop' entry.
355 (gfc_free_omp_clauses, gfc_match_omp_allocate,
356 gfc_match_omp_flush, gfc_match_omp_clause_reduction): Update
357 call.
358 (enum omp_mask2): Add OMP_CLAUSE_{INIT,USE,DESTROY}.
359 (OMP_INTEROP_CLAUSES): Use it.
360 (gfc_match_omp_clauses): Match those clauses.
361 (gfc_match_omp_prefer_type, gfc_match_omp_init,
362 gfc_match_omp_interop): New.
363 (resolve_omp_clauses): Handle interop clauses.
364 (omp_code_to_statement): Add ST_OMP_INTEROP.
365 (gfc_resolve_omp_directive): Add EXEC_OMP_INTEROP.
366 * parse.cc (decode_omp_directive): Parse 'interop' directive.
367 (next_statement, gfc_ascii_statement): Handle ST_OMP_INTEROP.
368 * st.cc (gfc_free_statement): Likewise
369 * resolve.cc (gfc_resolve_code): Handle EXEC_OMP_INTEROP.
370 * trans.cc (trans_code): Likewise.
371 * trans-openmp.cc (gfc_trans_omp_directive): Print 'sorry'
372 for EXEC_OMP_INTEROP.
374 2024-09-05 Harald Anlauf <anlauf@gmx.de>
376 PR fortran/100273
377 * trans-decl.cc (gfc_create_module_variable): Handle module
378 variable also when it is needed for the result specification
379 of a contained function.
381 2024-09-03 David Malcolm <dmalcolm@redhat.com>
383 * error.cc (gfc_move_error_buffer_from_to): Prefix all
384 output_buffer fields with "m_".
385 (gfc_diagnostics_init): Likewise.
387 2024-08-31 Harald Anlauf <anlauf@gmx.de>
389 * match.cc (gfc_match_namelist): Downgrade feature from GNU to
390 legacy extension.
392 2024-08-30 Harald Anlauf <anlauf@gmx.de>
394 PR fortran/98454
395 * resolve.cc (resolve_symbol): Add default-initialization of
396 non-allocatable, non-pointer derived-type function results.
398 2024-08-29 David Malcolm <dmalcolm@redhat.com>
400 * error.cc (gfc_format_decoder): Convert final param from
401 const char **buffer_ptr to pp_token_list &formatted_token_list,
402 and update call to default_tree_printer accordingly.
404 2024-08-29 Harald Anlauf <anlauf@gmx.de>
406 PR fortran/116530
407 * trans-io.cc (transfer_namelist_element): Prevent NULL pointer
408 dereference.
410 2024-08-23 Paul Thomas <pault@gcc.gnu.org>
412 Revert:
413 2024-08-06 Paul Thomas <pault@gcc.gnu.org>
415 PR fortran/102689
416 * trans-array.cc (get_array_ref_dim_for_loop_dim): Use the arg1
417 class container carried in ss->info as the seed for a lhs in
418 class valued transformational intrinsic calls that are not the
419 rhs of an assignment. Otherwise, the lhs variable expression is
420 taken from the loop chain. For this latter case, the _vptr and
421 _len fields are set.
422 (gfc_trans_create_temp_array): Use either the lhs expression
423 seeds to build a class variable that will take the returned
424 descriptor as its _data field. In the case that the arg1 expr.
425 is used, a class typespec must be built with the correct rank
426 and the _vptr and _len fields set. The element size is provided
427 for the temporary allocation and to set the descriptor span.
428 (gfc_array_init_size): When an intrinsic type scalar expr3 is
429 used in allocation of a class array, use its element size in
430 the descriptor dtype.
431 * trans-expr.cc (gfc_conv_class_to_class): Class valued
432 transformational intrinsics return the pointer to the array
433 descriptor as the _data field of a class temporary. Extract
434 directly and return the address of the class temporary.
435 (gfc_conv_procedure_call): store the expression for the first
436 argument of a class valued transformational intrinsic function
437 in the ss info class_container field. Later, use its type as
438 the element type in the call to gfc_trans_create_temp_array.
439 (fcncall_realloc_result): Add a dtype argument and use it in
440 the descriptor, when available.
441 (gfc_trans_arrayfunc_assign): For class lhs, build a dtype with
442 the lhs rank and the rhs element size and use it in the call to
443 fcncall_realloc_result.
445 2024-08-23 Gerald Pfeifer <gerald@pfeifer.com>
447 * invoke.texi (Code Gen Options): Add a missing word.
449 2024-08-23 Andre Vehreschild <vehre@gcc.gnu.org>
451 PR fortran/86468
452 * trans-stmt.cc (trans_associate_var): Remove superfluous
453 view_convert.
455 2024-08-21 Andre Vehreschild <vehre@gcc.gnu.org>
457 PR fortran/86468
458 * trans-intrinsic.cc (conv_intrinsic_move_alloc): Correct
459 comment.
460 * trans-types.cc (gfc_sym_type): Pass coarray rank, not false.
461 (gfc_get_derived_type): Only propagate codimension for coarrays
462 and pointers to array components in derived typed coarrays.
464 2024-08-21 Andre Vehreschild <vehre@gcc.gnu.org>
466 PR fortran/77518
467 * trans-intrinsic.cc (gfc_conv_intrinsic_sizeof): Use
468 class_container of se when set.
470 2024-08-20 Andre Vehreschild <vehre@gcc.gnu.org>
472 PR fortran/84246
473 * resolve.cc (caf_possible_reallocate): Detect arrays that may
474 be reallocated by caf_send.
475 (resolve_ordinary_assign): More reliably detect assignments
476 where a rewrite to caf_send is needed.
477 * trans-expr.cc (gfc_trans_assignment_1): Remove rewrite to
478 caf_send, because this is done by resolve now.
479 * trans-intrinsic.cc (conv_caf_send): Prevent unneeded temporary
480 arrays.
482 2024-08-19 Andre Vehreschild <vehre@gcc.gnu.org>
484 PR fortran/46371
485 PR fortran/56496
486 * expr.cc (gfc_is_coindexed): Detect is coindexed also when
487 rewritten to caf_get.
488 * trans-stmt.cc (trans_associate_var): Always accept a
489 descriptor for coarrays.
491 2024-08-19 Andre Vehreschild <vehre@gcc.gnu.org>
493 PR fortran/84244
494 * trans-types.cc (gfc_get_derived_type): When a caf_sub_token is
495 generated for a component, link it to the component it is
496 generated for (the previous one).
498 2024-08-16 Harald Anlauf <anlauf@gmx.de>
500 PR fortran/114146
501 * intrinsic.texi: Fix documentation of arguments of RANDOM_INIT,
502 which is conforming to the F2018 standard.
504 2024-08-15 Jakub Jelinek <jakub@redhat.com>
506 PR bootstrap/116387
507 * resolve.cc (resolve_operator): Use %L rather than %%L in format
508 string.
510 2024-08-15 Andre Vehreschild <vehre@gcc.gnu.org>
512 PR fortran/110033
513 * class.cc (gfc_is_class_scalar_expr): Coarray refs that ref
514 only self, aka this image, are regarded as scalar, too.
515 * resolve.cc (resolve_assoc_var): Ignore this image coarray refs
516 and do not build a new class type.
517 * trans-expr.cc (gfc_get_caf_token_offset): Get the caf token
518 from the descriptor for associated variables.
519 (gfc_conv_variable): Same.
520 (gfc_trans_pointer_assignment): Assign token to temporary
521 associate variable, too.
522 (gfc_trans_scalar_assign): Add flag that assign is for associate
523 and use it to assign the token.
524 (is_assoc_assign): Detect that expressions are for associate
525 assign.
526 (gfc_trans_assignment_1): Treat associate assigns like pointer
527 assignments where possible.
528 * trans-stmt.cc (trans_associate_var): Set same_class only for
529 class-targets.
530 * trans.h (gfc_trans_scalar_assign): Add flag to
531 trans_scalar_assign for marking associate assignments.
533 2024-08-15 Andre Vehreschild <vehre@gcc.gnu.org>
535 * arith.cc (reduce_unary): Use expr.corank.
536 (reduce_binary_ac): Same.
537 (reduce_binary_ca): Same.
538 (reduce_binary_aa): Same.
539 * array.cc (gfc_match_array_ref): Same.
540 * check.cc (dim_corank_check): Same.
541 (gfc_check_move_alloc): Same.
542 (gfc_check_image_index): Same.
543 * class.cc (gfc_add_class_array_ref): Same.
544 (finalize_component): Same.
545 * data.cc (gfc_assign_data_value): Same.
546 * decl.cc (match_clist_expr): Same.
547 (add_init_expr_to_sym): Same.
548 * expr.cc (simplify_intrinsic_op): Same.
549 (simplify_parameter_variable): Same.
550 (gfc_check_assign_symbol): Same.
551 (gfc_get_variable_expr): Same.
552 (gfc_add_full_array_ref): Same.
553 (gfc_lval_expr_from_sym): Same.
554 (gfc_get_corank): Removed.
555 * frontend-passes.cc (callback_reduction): Use expr.corank.
556 (create_var): Same.
557 (combine_array_constructor): Same.
558 (optimize_minmaxloc): Same.
559 * gfortran.h (gfc_get_corank): Add corank to gfc_expr.
560 * intrinsic.cc (gfc_get_intrinsic_function_symbol): Use
561 expr.corank.
562 (gfc_convert_type_warn): Same.
563 (gfc_convert_chartype): Same.
564 * iresolve.cc (resolve_bound): Same.
565 (gfc_resolve_cshift): Same.
566 (gfc_resolve_eoshift): Same.
567 (gfc_resolve_logical): Same.
568 (gfc_resolve_matmul): Same.
569 * match.cc (copy_ts_from_selector_to_associate): Same.
570 * matchexp.cc (gfc_get_parentheses): Same.
571 * parse.cc (parse_associate): Same.
572 * primary.cc (gfc_match_rvalue): Same.
573 * resolve.cc (resolve_structure_cons): Same.
574 (resolve_actual_arglist): Same.
575 (resolve_elemental_actual): Same.
576 (resolve_generic_f0): Same.
577 (resolve_unknown_f): Same.
578 (resolve_operator): Same.
579 (gfc_expression_rank): Same and set dimen_type for coarray to
580 default.
581 (gfc_op_rank_conformable): Use expr.corank.
582 (add_caf_get_intrinsic): Same.
583 (resolve_variable): Same.
584 (gfc_fixup_inferred_type_refs): Same.
585 (check_host_association): Same.
586 (resolve_compcall): Same.
587 (resolve_expr_ppc): Same.
588 (resolve_assoc_var): Same.
589 (fixup_array_ref): Same.
590 (resolve_select_type): Same.
591 (add_comp_ref): Same.
592 (get_temp_from_expr): Same.
593 (resolve_fl_var_and_proc): Same.
594 (resolve_symbol): Same.
595 * symbol.cc (gfc_is_associate_pointer): Same.
596 * trans-array.cc (walk_coarray): Same.
597 (gfc_conv_expr_descriptor): Same.
598 (gfc_walk_array_ref): Same.
599 * trans-array.h (gfc_walk_array_ref): Same.
600 * trans-expr.cc (gfc_get_ultimate_alloc_ptr_comps_caf_token):
601 Same.
602 * trans-intrinsic.cc (trans_this_image): Same.
603 (trans_image_index): Same.
604 (conv_intrinsic_cobound): Same.
605 (gfc_walk_intrinsic_function): Same.
606 (conv_intrinsic_move_alloc): Same.
607 * trans-stmt.cc (gfc_trans_lock_unlock): Same.
608 (trans_associate_var): Same and adapt to slightly different
609 behaviour of expr.corank and gfc_get_corank.
610 (gfc_trans_allocate): Same.
611 * trans.cc (gfc_add_finalizer_call): Same.
613 2024-08-14 Harald Anlauf <anlauf@gmx.de>
615 * simplify.cc (gfc_simplify_sizeof): Clear used gmp variable.
616 * target-memory.cc (gfc_target_expr_size): Likewise.
618 2024-08-14 Andre Vehreschild <vehre@gcc.gnu.org>
620 PR fortran/102973
621 * match.cc (gfc_match_associate): Reset proc_ptr parsing flag on
622 error.
624 2024-08-14 Andre Vehreschild <vehre@gcc.gnu.org>
626 PR fortran/116292
627 * trans-intrinsic.cc (conv_intrinsic_move_alloc): Get the vtab
628 only for derived types and classes and adjust _len for class
629 types.
631 2024-08-13 Harald Anlauf <anlauf@gmx.de>
632 Steven G. Kargl <kargl@gcc.gnu.org>
634 PR fortran/114308
635 * array.cc (resolve_array_list): Reject array constructor value if
636 its declared type is abstract (F2018:C7114).
638 2024-08-12 Andre Vehreschild <vehre@gcc.gnu.org>
640 PR fortran/85510
641 * resolve.cc (resolve_variable): Mark the variable as host
642 associated only, when it is not in an associate block.
643 * trans-decl.cc (generate_coarray_init): Remove incorrect unused
644 flag on parameter.
646 2024-08-11 Harald Anlauf <anlauf@gmx.de>
648 PR fortran/116221
649 * intrinsic.cc (gfc_get_intrinsic_sub_symbol): Initialize variable.
650 * symbol.cc (gfc_get_ha_symbol): Likewise.
652 2024-08-07 Julian Brown <julian@codesourcery.com>
653 Tobias Burnus <tobias@baylibre.com>
655 * gfortran.h (gfc_option_t): Add disable_omp_is_initial_device.
656 * lang.opt (fbuiltin-): Add.
657 * options.cc (gfc_handle_option): Handle
658 -fno-builtin-omp_is_initial_device.
659 * f95-lang.cc (gfc_init_builtin_functions): Handle
660 DEF_GOMP_BUILTIN_COMPILER.
661 * trans-decl.cc (gfc_get_extern_function_decl): Add code to use
662 DEF_GOMP_BUILTIN_COMPILER for 'omp_is_initial_device'.
664 2024-08-06 Jerry DeLisle <jvdelisle@gcc.gnu.org>
666 PR fortran/109105
667 * resolve.cc (CHECK_INTERFACES): New helper macro.
668 (resolve_operator): Replace use of snprintf with
669 gfc_error.
671 2024-08-06 Paul Thomas <pault@gcc.gnu.org>
673 PR fortran/102689
674 * trans-array.cc (get_array_ref_dim_for_loop_dim): Use the arg1
675 class container carried in ss->info as the seed for a lhs in
676 class valued transformational intrinsic calls that are not the
677 rhs of an assignment. Otherwise, the lhs variable expression is
678 taken from the loop chain. For this latter case, the _vptr and
679 _len fields are set.
680 (gfc_trans_create_temp_array): Use either the lhs expression
681 seeds to build a class variable that will take the returned
682 descriptor as its _data field. In the case that the arg1 expr.
683 is used, a class typespec must be built with the correct rank
684 and the _vptr and _len fields set. The element size is provided
685 for the temporary allocation and to set the descriptor span.
686 (gfc_array_init_size): When an intrinsic type scalar expr3 is
687 used in allocation of a class array, use its element size in
688 the descriptor dtype.
689 * trans-expr.cc (gfc_conv_class_to_class): Class valued
690 transformational intrinsics return the pointer to the array
691 descriptor as the _data field of a class temporary. Extract
692 directly and return the address of the class temporary.
693 (gfc_conv_procedure_call): store the expression for the first
694 argument of a class valued transformational intrinsic function
695 in the ss info class_container field. Later, use its type as
696 the element type in the call to gfc_trans_create_temp_array.
697 (fcncall_realloc_result): Add a dtype argument and use it in
698 the descriptor, when available.
699 (gfc_trans_arrayfunc_assign): For class lhs, build a dtype with
700 the lhs rank and the rhs element size and use it in the call to
701 fcncall_realloc_result.
703 2024-08-03 Steve Kargl <kargls@comcast.net>
705 PR fortran/104626
706 * symbol.cc (gfc_add_save): Add checks for SAVE attribute
707 conflicts and duplicate SAVE attribute.
709 2024-08-02 Mikael Morin <mikael@gcc.gnu.org>
711 * trans-intrinsic.cc (maybe_absent_optional_variable): New function.
712 (gfc_conv_intrinsic_minmaxloc): Remove BACK from scalarization and
713 evaluate it before. Add a check that BACK is not null if the
714 expression is an optional dummy. Save the resulting expression to a
715 variable. Use the variable in the generated inline code.
717 2024-08-01 Jakub Jelinek <jakub@redhat.com>
719 * trans-types.cc (gfc_get_array_descr_info): Don't test if
720 !integer_zerop (dtype_off), use fold_build_pointer_plus
721 unconditionally.
723 2024-07-29 Tobias Burnus <tburnus@baylibre.com>
725 PR fortran/115559
726 * trans-common.cc (build_common_decl): Add 'omp declare target' and
727 'omp declare target link' variables to offload_vars.
728 * trans-decl.cc (add_attributes_to_decl): Likewise; update args and
729 call decl_attributes.
730 (get_proc_pointer_decl, gfc_get_extern_function_decl,
731 build_function_decl): Update calls.
732 (gfc_get_symbol_decl): Likewise; move after 'DECL_STATIC (t)=1'
733 to avoid errors with symtab_node::get_create.
735 2024-07-22 Andre Vehreschild <vehre@gcc.gnu.org>
737 PR fortran/88624
738 * trans-expr.cc (gfc_conv_procedure_call): Treat
739 pointers/references (e.g. from parameters) correctly by derefing
740 them.
742 2024-07-21 Paul Thomas <pault@gcc.gnu.org>
744 PR fortran/59104
745 * gfortran.h : Add decl_order to gfc_symbol.
746 * symbol.cc : Add static next_decl_order..
747 (gfc_set_sym_referenced): Set symbol decl_order.
748 * trans-decl.cc : Include dependency.h.
749 (decl_order): Replace symbol declared_at.lb->location with
750 decl_order.
752 2024-07-19 Harald Anlauf <anlauf@gmx.de>
754 PR fortran/103115
755 * trans-array.cc (gfc_trans_array_constructor_value): If the first
756 element of an array constructor is deferred-length character and
757 therefore does not have an element size known at compile time, do
758 not try to collect subsequent constant elements into a constructor
759 for optimization.
761 2024-07-18 Andre Vehreschild <vehre@gcc.gnu.org>
763 PR fortran/78466
764 PR fortran/80774
765 * array.cc (gfc_compare_array_spec): Take cotype into account.
766 * class.cc (gfc_build_class_symbol): Coarrays are also arrays.
767 * gfortran.h (IS_CLASS_COARRAY_OR_ARRAY): New macro to detect
768 regular and coarray class arrays.
769 * interface.cc (compare_components): Take codimension into
770 account.
771 * resolve.cc (resolve_symbol): Improve error message.
772 * simplify.cc (simplify_bound_dim): Remove duplicate.
773 * trans-array.cc (gfc_trans_array_cobounds): Coarrays are also
774 arrays.
775 (gfc_trans_array_bounds): Same.
776 (gfc_trans_dummy_array_bias): Same.
777 (get_coarray_as): Get the as having a non-zero codim.
778 (is_explicit_coarray): Detect explicit coarrays.
779 (gfc_conv_expr_descriptor): Create a new descriptor for explicit
780 coarrays.
781 * trans-decl.cc (gfc_build_qualified_array): Coarrays are also
782 arrays.
783 (gfc_build_dummy_array_decl): Same.
784 (gfc_get_symbol_decl): Same.
785 (gfc_trans_deferred_vars): Same.
786 * trans-expr.cc (class_scalar_coarray_to_class): Get the
787 descriptor from the correct location.
788 (gfc_conv_variable): Pick up the descriptor when needed.
789 * trans-types.cc (gfc_is_nodesc_array): Coarrays are also
790 arrays.
791 (gfc_get_nodesc_array_type): Indentation fix only.
792 (cobounds_match_decl): Match a tree's bounds to the expr's
793 bounds and return true, when they match.
794 (gfc_get_derived_type): Create a new type tree/descriptor, when
795 the cobounds of the existing declaration and expr to not
796 match. This happends for class arrays in parameter list, when
797 there are different cobound declarations.
799 2024-07-18 Paul Thomas <pault@gcc.gnu.org>
801 PR fortran/108889
802 * gfortran.h: Add bit field 'allocated_in_scope' to gfc_symbol.
803 * trans-array.cc (gfc_array_allocate): Set 'allocated_in_scope'
804 after allocation if not a component reference.
805 (gfc_alloc_allocatable_for_assignment): If 'allocated_in_scope'
806 not set, not a component ref and not allocated, set the array
807 bounds and offset to give zero length in all dimensions. Then
808 set allocated_in_scope.
810 2024-07-18 Andre Vehreschild <vehre@gcc.gnu.org>
812 PR fortran/82904
813 * trans-types.cc (gfc_sym_type): Use type `char*` for saved
814 deferred length char arrays.
815 * trans.cc (get_array_span): Get `.span` also for `char*` typed
816 arrays, i.e. for those that have INTEGER_TYPE instead of
817 ARRAY_TYPE.
819 2024-07-17 Kewen Lin <linkw@linux.ibm.com>
821 PR target/112993
822 * trans-types.cc (get_real_kind_from_node): Consider the case where
823 more than one modes have the same precision.
825 2024-07-16 Paul Thomas <pault@gcc.gnu.org>
827 PR fortran/84868
828 * simplify.cc (gfc_simplify_len_trim): If the argument is an
829 element of a parameter array, simplify all the elements and
830 build a new parameter array to hold the result, after checking
831 that it doesn't already exist.
832 * trans-expr.cc (gfc_get_interface_mapping_array) if a string
833 length is available, use it for the typespec.
834 (gfc_add_interface_mapping): Supply the se string length.
836 2024-07-16 Prathamesh Kulkarni <prathameshk@nvidia.com>
838 * trans-expr.cc (gfc_trans_zero_assign): Handle allocatable arrays.
840 2024-07-13 Mikael Morin <mikael@gcc.gnu.org>
842 * trans-intrinsic.cc (gfc_conv_intrinsic_minmaxloc): Add the
843 preliminary code generated for MASK to the preliminary code of
844 MINLOC/MAXLOC.
846 2024-07-12 Mikael Morin <morin-mikael@orange.fr>
848 * trans-intrinsic.cc (gfc_conv_intrinsic_minmaxloc): Create a new
849 scalar scalarization chain element if BACK is present. Add it to
850 the loop. Set the scalarization chain before evaluating the
851 argument.
853 2024-07-11 Andre Vehreschild <vehre@gcc.gnu.org>
855 * trans-array.cc (gfc_conv_array_parameter): Init variable to
856 NULL_TREE to fix bootstrap.
858 2024-07-11 Andre Vehreschild <vehre@gcc.gnu.org>
860 PR fortran/96992
861 * trans-array.cc (gfc_trans_array_bounds): Set a starting
862 stride, when descriptor expects a variable for the stride.
863 (gfc_trans_dummy_array_bias): Allow storage association for
864 dummy class arrays, when they are not elemental.
865 (gfc_conv_array_parameter): Add more general class support
866 and packing for classes, too.
867 * trans-array.h (gfc_conv_array_parameter): Add lbound shift
868 for class arrays.
869 * trans-decl.cc (gfc_build_builtin_function_decls): Add decls
870 for internal_un-/pack_class.
871 * trans-expr.cc (gfc_reset_vptr): Allow supplying a type-tree
872 to generate the vtab from.
873 (gfc_class_set_vptr): Allow supplying a class-tree to take the
874 vptr from.
875 (class_array_data_assign): Rename to gfc_class_array_data_assign
876 and make usable from other compile units.
877 (gfc_class_array_data_assign): Renamed from class_array_data_
878 assign.
879 (gfc_conv_derived_to_class): Remove assert to
880 allow converting derived to class type arrays with assumed
881 rank. Reduce code base and use gfc_conv_array_parameter also
882 for classes.
883 (gfc_conv_class_to_class): Use gfc_class_data_assign.
884 (gfc_conv_procedure_call): Adapt to new signature of
885 gfc_conv_derived_to_class.
886 * trans-io.cc (transfer_expr): Same.
887 * trans-stmt.cc (trans_associate_var): Same.
888 * trans.h (gfc_conv_derived_to_class): Signature changed.
889 (gfc_class_array_data_assign): Made public.
890 (gfor_fndecl_in_pack_class): Added declaration.
891 (gfor_fndecl_in_unpack_class): Same.
893 2024-07-08 Mikael Morin <mikael@gcc.gnu.org>
895 * trans-array.cc (gfc_trans_preloop_setup): Use a separate variable
896 for iteration. Use directly the value of variable I if it is known.
897 Move the definition of the variable to the branch where the
898 remaining uses are.
900 2024-07-03 Harald Anlauf <anlauf@gmx.de>
902 PR fortran/115700
903 * trans-stmt.cc (trans_associate_var): When the associate target
904 is an array-valued character variable, the length is known at entry
905 of the associate block. Move setting of string length of the
906 selector to the initialization part of the block.
908 2024-07-01 Andrew Stubbs <ams@baylibre.com>
909 Thomas Schwinge <thomas@codesourcery.com>
911 * openmp.cc (is_predefined_allocator): Update valid ranges to
912 incorporate ompx_gnu_pinned_mem_alloc.
914 2024-06-29 Harald Anlauf <anlauf@gmx.de>
916 PR fortran/114019
917 * trans-stmt.cc (gfc_trans_allocate): Fix handling of case of
918 scalar character expression being used for SOURCE.
920 2024-06-28 Andre Vehreschild <vehre@gcc.gnu.org>
922 * trans-expr.cc (gfc_class_set_vptr): Add setting of _vptr
923 member.
924 * trans-intrinsic.cc (conv_intrinsic_move_alloc): First use
925 of gfc_class_set_vptr and refactor very similar code.
926 * trans.h (gfc_class_set_vptr): Declare the new function.
928 2024-06-28 Andre Vehreschild <vehre@gcc.gnu.org>
930 * trans-array.cc (structure_alloc_comps): Use reset_vptr.
931 * trans-decl.cc (gfc_trans_deferred_vars): Same.
932 (gfc_generate_function_code): Same.
933 * trans-expr.cc (gfc_reset_vptr): Allow supplying the class
934 type.
935 (gfc_conv_procedure_call): Use reset_vptr.
936 * trans-intrinsic.cc (gfc_conv_intrinsic_transfer): Same.
938 2024-06-24 Harald Anlauf <anlauf@gmx.de>
940 PR fortran/55978
941 * trans-array.cc (gfc_conv_array_parameter): Do not dereference
942 data component of a missing allocatable dummy array argument for
943 passing as actual to optional dummy. Harden logic of presence
944 check for optional pointer dummy by using TRUTH_ANDIF_EXPR instead
945 of TRUTH_AND_EXPR.
947 2024-06-20 Paul Thomas <pault@gcc.gnu.org>
949 PR fortran/59104
950 * dependency.cc (dependency_fcn, gfc_function_dependency): New
951 functions to detect dependency in array bounds and character
952 lengths on old style function results.
953 * dependency.h : Add prototype for gfc_function_dependency.
954 * error.cc (error_print): Remove trailing space.
955 * gfortran.h : Remove dummy_order and add fn_result_spec.
956 * symbol.cc : Remove declaration of next_dummy_order..
957 (gfc_set_sym_referenced): remove setting of symbol dummy order.
958 * trans-array.cc (gfc_trans_auto_array_allocation): Detect
959 non-dummy symbols with function dependencies and put the
960 allocation at the end of the initialization code.
961 * trans-decl.cc : Include dependency.h.
962 (decl_order): New function that determines uses the location
963 field of the symbol 'declared_at' to determine the order of two
964 declarations.
965 (gfc_defer_symbol_init): Call gfc_function_dependency to put
966 dependent symbols in the right part of the tlink chain. Use
967 the location field of the symbol declared_at to determine the
968 order of declarations.
969 (gfc_trans_auto_character_variable): Put character length
970 initialization of dependent symbols at the end of the chain.
971 * trans.cc (gfc_add_init_cleanup): Add boolean argument with
972 default false that determines whther an expression is placed at
973 the back or the front of the initialization chain.
974 * trans.h : Update the prototype for gfc_add_init_cleanup.
976 2024-06-19 Harald Anlauf <anlauf@gmx.de>
978 PR fortran/115390
979 * trans-decl.cc (gfc_conv_cfi_to_gfc): Move derivation of type sizes
980 for character via gfc_trans_vla_type_sizes to after character length
981 has been set.
983 2024-06-19 Andre Vehreschild <vehre@gcc.gnu.org>
985 PR fortran/90076
986 * trans-decl.cc (gfc_generate_function_code): Set vptr for
987 results to declared class type.
988 * trans-expr.cc (gfc_reset_vptr): Allow to provide the typespec
989 instead of the expression.
990 * trans.h (gfc_reset_vptr): Same.
992 2024-06-17 Andre Vehreschild <vehre@gcc.gnu.org>
994 * trans.cc (gfc_deallocate_with_status): Check that object to deref
995 is an array, before applying array deref.
997 2024-06-12 David Malcolm <dmalcolm@redhat.com>
999 * error.cc (gfc_clear_pp_buffer): Likewise.
1000 (gfc_warning): Likewise.
1001 (gfc_warning_check): Likewise.
1002 (gfc_error_opt): Likewise.
1003 (gfc_error_check): Likewise.
1005 2024-06-07 Andre Vehreschild <vehre@gcc.gnu.org>
1007 PR fortran/90068
1008 * trans-array.cc (gfc_trans_array_ctor_element): Eval non-
1009 variable expressions once only.
1010 (gfc_trans_array_constructor_value): Add statements of
1011 final block.
1012 (trans_array_constructor): Detect when final block is required.
1014 2024-06-07 Andre Vehreschild <vehre@gcc.gnu.org>
1016 PR fortran/90072
1017 * expr.cc (gfc_is_alloc_class_scalar_function): Detect
1018 allocatable class return types also for user-defined
1019 functions.
1020 * trans-expr.cc (gfc_conv_procedure_call): Same.
1021 (trans_class_vptr_len_assignment): Compute vptr len
1022 assignment correctly for user-defined functions.
1024 2024-06-05 Jakub Jelinek <jakub@redhat.com>
1025 Frederik Harwath <frederik@codesourcery.com>
1026 Sandra Loosemore <sandra@codesourcery.com>
1028 * gfortran.h (enum gfc_statement): Add ST_OMP_UNROLL,
1029 ST_OMP_END_UNROLL, ST_OMP_TILE and ST_OMP_END_TILE.
1030 (struct gfc_omp_clauses): Add sizes_list, partial, full and erroneous
1031 members.
1032 (enum gfc_exec_op): Add EXEC_OMP_UNROLL and EXEC_OMP_TILE.
1033 (gfc_expr_list_len): Declare.
1034 * match.h (gfc_match_omp_tile, gfc_match_omp_unroll): Declare.
1035 * openmp.cc (gfc_get_location): Declare.
1036 (gfc_free_omp_clauses): Free sizes_list.
1037 (match_oacc_expr_list): Rename to ...
1038 (match_omp_oacc_expr_list): ... this. Add is_omp argument and
1039 change diagnostic wording if it is true.
1040 (enum omp_mask2): Add OMP_CLAUSE_{FULL,PARTIAL,SIZES}.
1041 (gfc_match_omp_clauses): Parse full, partial and sizes clauses.
1042 (gfc_match_oacc_wait): Use match_omp_oacc_expr_list instead of
1043 match_oacc_expr_list.
1044 (OMP_UNROLL_CLAUSES, OMP_TILE_CLAUSES): Define.
1045 (gfc_match_omp_tile, gfc_match_omp_unroll): New functions.
1046 (resolve_omp_clauses): Diagnose full vs. partial clause conflict.
1047 Resolve sizes clause arguments.
1048 (find_nested_loop_in_chain): Use switch instead of series of ifs.
1049 Handle EXEC_OMP_TILE and EXEC_OMP_UNROLL.
1050 (gfc_resolve_omp_do_blocks): Set omp_current_do_collapse to
1051 list length of sizes_list if present.
1052 (gfc_resolve_do_iterator): Return for EXEC_OMP_TILE or
1053 EXEC_OMP_UNROLL.
1054 (restructure_intervening_code): Remove spurious ()s around & operands.
1055 (is_outer_iteration_variable): Handle EXEC_OMP_TILE and
1056 EXEC_OMP_UNROLL.
1057 (check_nested_loop_in_chain): Likewise.
1058 (expr_is_invariant): Likewise.
1059 (resolve_omp_do): Handle EXEC_OMP_TILE and EXEC_OMP_UNROLL. Diagnose
1060 tile without sizes clause. Use sizes_list length for count if
1061 non-NULL. Set code->ext.omp_clauses->erroneous on loops where we've
1062 reported diagnostics. Sorry for mixing non-rectangular loops with
1063 generated loops.
1064 (omp_code_to_statement): Handle EXEC_OMP_TILE and EXEC_OMP_UNROLL.
1065 (gfc_resolve_omp_directive): Likewise.
1066 * parse.cc (decode_omp_directive): Parse end tile, end unroll, tile
1067 and unroll. Move nothing entry alphabetically.
1068 (case_exec_markers): Add ST_OMP_TILE and ST_OMP_UNROLL.
1069 (gfc_ascii_statement): Handle ST_OMP_END_TILE, ST_OMP_END_UNROLL,
1070 ST_OMP_TILE and ST_OMP_UNROLL.
1071 (parse_omp_do): Add nested argument. Handle ST_OMP_TILE and
1072 ST_OMP_UNROLL.
1073 (parse_omp_structured_block): Adjust parse_omp_do caller.
1074 (parse_executable): Likewise. Handle ST_OMP_TILE and ST_OMP_UNROLL.
1075 * resolve.cc (gfc_resolve_blocks): Handle EXEC_OMP_TILE and
1076 EXEC_OMP_UNROLL.
1077 (gfc_resolve_code): Likewise.
1078 * st.cc (gfc_free_statement): Likewise.
1079 * trans.cc (trans_code): Likewise.
1080 * trans-openmp.cc (gfc_trans_omp_clauses): Handle full, partial and
1081 sizes clauses. Use tree_cons + nreverse instead of
1082 temporary vector and build_tree_list_vec for tile_list handling.
1083 (gfc_expr_list_len): New function.
1084 (gfc_trans_omp_do): Rename tile to oacc_tile. Handle sizes clause.
1085 Don't assert code->op is EXEC_DO. Handle EXEC_OMP_TILE and
1086 EXEC_OMP_UNROLL.
1087 (gfc_trans_omp_directive): Handle EXEC_OMP_TILE and EXEC_OMP_UNROLL.
1088 * dump-parse-tree.cc (show_omp_clauses): Dump full, partial and
1089 sizes clauses.
1090 (show_omp_node): Handle EXEC_OMP_TILE and EXEC_OMP_UNROLL.
1091 (show_code_node): Likewise.
1093 2024-06-05 Kewen Lin <linkw@linux.ibm.com>
1095 * trans-intrinsic.cc (build_round_expr): Use TYPE_PRECISION of
1096 long_double_type_node to replace LONG_DOUBLE_TYPE_SIZE.
1097 * trans-types.cc (gfc_build_real_type): Use TYPE_PRECISION of
1098 {float,double,long_double}_type_node to replace
1099 {FLOAT,DOUBLE,LONG_DOUBLE}_TYPE_SIZE.
1101 2024-06-04 Harald Anlauf <anlauf@gmx.de>
1103 PR fortran/83865
1104 * trans-stmt.cc (gfc_trans_allocate): Restrict special case for
1105 source-expression with zero-length character to rank 0, so that
1106 the array shape is not discarded.
1108 2024-05-29 Andre Vehreschild <vehre@gcc.gnu.org>
1110 PR fortran/90069
1111 * trans-expr.cc (gfc_conv_procedure_call): Evaluate
1112 expressions with side-effects only ones and ensure
1113 old is freeed.
1115 2024-05-24 Harald Anlauf <anlauf@gmx.de>
1116 Steven G. Kargl <kargl@gcc.gnu.org>
1118 PR fortran/93635
1119 * symbol.cc (conflict_std): Helper function for reporting attribute
1120 conflicts depending on the Fortran standard version.
1121 (conf_std): Helper macro for checking standard-dependent conflicts.
1122 (gfc_check_conflict): Use it.
1124 2024-05-24 Harald Anlauf <anlauf@gmx.de>
1126 PR fortran/86100
1127 * trans-array.cc (gfc_conv_ss_startstride): Use abridged_ref_name
1128 to generate a more user-friendly name for bounds-check messages.
1129 * trans-expr.cc (gfc_copy_class_to_class): Fix bounds check for
1130 rank>1 by looping over the dimensions.
1132 2024-05-23 Paul Thomas <pault@gcc.gnu.org>
1134 PR fortran/103312
1135 * dependency.cc (gfc_dep_compare_expr): Handle component call
1136 expressions. Return -2 as default and return 0 if compared with
1137 a function expression that is from an interface body and has
1138 the same name.
1139 * expr.cc (gfc_reduce_init_expr): If the expression is a comp
1140 call do not attempt to reduce, defer to resolution and return
1141 false.
1142 * trans-types.cc (gfc_get_dtype_rank_type,
1143 gfc_get_nodesc_array_type): Fix whitespace.
1145 2024-05-20 Mikael Morin <mikael@gcc.gnu.org>
1147 PR fortran/99798
1148 * symbol.cc (gfc_release_symbol): Move the condition guarding
1149 the handling cyclic references...
1150 (cyclic_reference_break_needed): ... here as a new predicate.
1151 Remove superfluous parts. Add a condition preventing any premature
1152 release with submodule symbols.
1154 2024-05-20 Tobias Burnus <tburnus@baylibre.com>
1156 PR fortran/115150
1157 * trans-intrinsic.cc (gfc_conv_intrinsic_bound): Fix SHAPE
1158 for zero-size arrays
1160 2024-05-20 Tobias Burnus <tburnus@baylibre.com>
1162 * invoke.texi (fcoarray): Link to OpenCoarrays.org;
1163 mention libcaf_single.
1165 2024-05-17 Paul Thomas <pault@gcc.gnu.org>
1167 PR fortran/114874
1168 * gfortran.h: Add 'assoc_name_inferred' to gfc_namespace.
1169 * match.cc (gfc_match_select_type): Set 'assoc_name_inferred'
1170 in select type namespace if the selector has inferred type.
1171 * primary.cc (gfc_match_varspec): If a select type temporary
1172 is apparently scalar and a left parenthesis has been detected,
1173 check the current namespace has 'assoc_name_inferred' set. If
1174 so, set inferred_type.
1175 * resolve.cc (resolve_variable): If the namespace of a select
1176 type temporary is marked with 'assoc_name_inferred' call
1177 gfc_fixup_inferred_type_refs to ensure references are OK.
1178 (gfc_fixup_inferred_type_refs): Catch invalid array refs..
1180 2024-05-15 Jakub Jelinek <jakub@redhat.com>
1182 PR c/115103
1183 * openmp.cc (resolve_omp_clauses): Diagnose grainsize
1184 used together with num_tasks or nogroup used together with
1185 reduction.
1187 2024-05-13 Paul Thomas <pault@gcc.gnu.org>
1189 PR fortran/113363
1190 * trans-array.cc (gfc_array_init_size): Use the expr3 dtype so
1191 that the correct element size is used.
1192 * trans-expr.cc (gfc_conv_procedure_call): Remove restriction
1193 that ss and ss->loop be present for the finalization of class
1194 array function results.
1195 (trans_class_assignment): Use free and malloc, rather than
1196 realloc, for character expressions assigned to unlimited poly
1197 entities.
1198 * trans-stmt.cc (gfc_trans_allocate): Build a correct rhs for
1199 the assignment of an unlimited polymorphic 'source'.
1201 2024-05-12 Harald Anlauf <anlauf@gmx.de>
1203 * primary.cc (gfc_match_varspec): Replace 'ref' argument to
1204 is_inquiry_ref() by NULL when the result is not needed to avoid
1205 a memleak.
1207 2024-05-12 Paul Thomas <pault@gcc.gnu.org>
1209 PR fortran/84006
1210 PR fortran/100027
1211 PR fortran/98534
1212 * iresolve.cc (gfc_resolve_transfer): Emit a TODO error for
1213 unlimited polymorphic mold.
1214 * trans-expr.cc (gfc_resize_class_size_with_len): Use the fold
1215 even if a block is not available in which to fix the result.
1216 (trans_class_assignment): Enable correct assignment of
1217 character expressions to unlimited polymorphic variables using
1218 lhs _len field and rse string_length.
1219 * trans-intrinsic.cc (gfc_conv_intrinsic_storage_size): Extract
1220 the class expression so that the unlimited polymorphic class
1221 expression can be used in gfc_resize_class_size_with_len to
1222 obtain the storage size for character payloads. Guard the use
1223 of GFC_DECL_SAVED_DESCRIPTOR by testing for DECL_LANG_SPECIFIC
1224 to prevent the ICE. Also, invert the order to use the class
1225 expression extracted from the argument.
1226 (gfc_conv_intrinsic_transfer): In same way as 'storage_size',
1227 use the _len field to obtaining the correct length for arg 1.
1228 Add a branch for the element size in bytes of class expressions
1229 with provision to make use of the unlimited polymorphic _len
1230 field. Again, the class references are explicitly identified.
1231 'mold_expr' was already declared. Use it instead of 'arg'. Do
1232 not fix 'dest_word_len' for deferred character sources because
1233 reallocation on assign makes use of it before it is assigned.
1235 2024-05-11 Harald Anlauf <anlauf@gmx.de>
1237 PR fortran/115039
1238 * expr.cc (gfc_traverse_expr): An inquiry ref does not constitute
1239 a dependency and cannot collide with a symbol.
1241 2024-05-05 Harald Anlauf <anlauf@gmx.de>
1243 PR fortran/114827
1244 * trans-array.cc (gfc_alloc_allocatable_for_assignment): Take into
1245 account _len of unlimited polymorphic entities when calculating
1246 the effective element size for allocation size and array span.
1247 Set _len of lhs to _len of rhs.
1248 * trans-expr.cc (trans_class_assignment): Take into account _len
1249 of unlimited polymorphic entities for allocation size.
1251 2024-04-29 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
1253 * expr.cc (check_transformational): Add SELECTED_LOGICAL_KIND
1254 to allowed functions for Fortran 2023.
1255 * gfortran.h (GFC_ISYM_SL_KIND): New.
1256 * gfortran.texi: Mention SELECTED_LOGICAL_KIND.
1257 * intrinsic.cc (add_functions): Add SELECTED_LOGICAL_KIND.
1258 (gfc_intrinsic_func_interface): Allow it in initialization
1259 expressions.
1260 * intrinsic.h (gfc_simplify_selected_logical_kind): New proto.
1261 * intrinsic.texi: Add SELECTED_LOGICAL_KIND.
1262 * simplify.cc (gfc_simplify_selected_logical_kind): New
1263 function.
1264 * trans-decl.cc (gfc_build_intrinsic_function_decls): Initialize
1265 gfor_fndecl_sl_kind.
1266 * trans-intrinsic.cc (gfc_conv_intrinsic_sl_kind): New function.
1267 (gfc_conv_intrinsic_function): Call it for GFC_ISYM_SL_KIND.
1268 * trans.h (gfor_fndecl_sl_kind): New symbol.
1270 2024-04-29 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
1272 * iso-fortran-env.def: Add logical{8,16,32,64} and
1273 real16 named constants.
1275 2024-04-29 Paul Thomas <pault@gcc.gnu.org>
1277 PR fortran/114959
1278 * trans-expr.cc (gfc_trans_class_init_assign): Return NULL_TREE
1279 if the default initializer has all NULL fields. Guard this
1280 by a requirement that the code not be EXEC_INIT_ASSIGN and that
1281 the object be an INTENT_OUT dummy.
1282 * trans-stmt.cc (gfc_trans_allocate): Change the initializer
1283 code for allocate with mold to EXEC_ALLOCATE to allow an
1284 initializer with all NULL fields.
1286 2024-04-25 Paul Thomas <pault@gcc.gnu.org>
1288 PR fortran/93678
1289 * trans-expr.cc (gfc_conv_procedure_call): Use the interface,
1290 where possible, to obtain the type of character procedure
1291 pointers of class entities.
1293 2024-04-25 Paul Thomas <pault@gcc.gnu.org>
1294 Jakub Jelinek <jakub@gcc.gnu.org>
1296 PR fortran/89462
1297 * decl.cc (build_sym): Add an extra argument 'elem'. If 'elem'
1298 is greater than 1, gfc_new_charlen is called to generate a new
1299 charlen, registered in the symbol namespace.
1300 (variable_decl, enumerator_decl): Set the new argument in the
1301 calls to build_sym.
1303 2024-04-21 Paul Thomas <pault@gcc.gnu.org>
1305 PR fortran/103471
1306 * resolve.cc (resolve_actual_arglist): Catch variables silently
1307 set as untyped, resetting the flag so that gfc_resolve_expr can
1308 generate the no implicit type error.
1309 (gfc_resolve_index_1): Block index expressions of unknown type
1310 from being converted to default integer, avoiding the fatal
1311 error in trans-decl.cc.
1312 * symbol.cc (gfc_set_default_type): Remove '(symbol)' from the
1313 'no IMPLICIT type' error message.
1314 * trans-decl.cc (gfc_get_symbol_decl): Change fatal error locus
1315 to that of the symbol declaration.
1316 (gfc_trans_deferred_vars): Remove two trailing tabs.
1318 2024-04-18 Paul Thomas <pault@gcc.gnu.org>
1320 PR fortran/114739
1321 * primary.cc (gfc_match_varspec): Check for default type before
1322 checking for derived types with the right component name.
1324 2024-04-16 Harald Anlauf <anlauf@gmx.de>
1326 PR fortran/113793
1327 * resolve.cc (resolve_allocate_expr): Reject ALLOCATE with SOURCE=
1328 or MOLD= specifier for unequal length.
1329 * trans-stmt.cc (gfc_trans_allocate): If an allocatable character
1330 variable has fixed length, use it and do not use the source length.
1331 With bounds-checking enabled, add a runtime check for same length.
1333 2024-04-10 Harald Anlauf <anlauf@gmx.de>
1335 PR fortran/106500
1336 * check.cc (is_c_interoperable): Fix checks for C_SIZEOF.
1337 (gfc_check_c_f_pointer): Reject function returning a pointer as FPTR,
1338 and improve an error message.
1340 2024-04-09 Paul Thomas <pault@gcc.gnu.org>
1342 PR fortran/114535
1343 * resolve.cc (resolve_symbol): Remove last chunk that checked
1344 for finalization of unreferenced symbols.
1346 2024-04-09 Paul Thomas <pault@gcc.gnu.org>
1348 PR fortran/113956
1349 * trans-expr.cc (gfc_trans_pointer_assignment): Remove assert
1350 causing the ICE since it was unnecesary.
1352 2024-04-02 Paul Thomas <pault@gcc.gnu.org>
1354 PR fortran/106999
1355 * interface.cc (gfc_compare_interfaces): Add error for a
1356 subroutine proc pointer passed to a variable formal.
1357 (compare_parameter): If a procedure pointer is being passed to
1358 a non-procedure formal arg, and there is an an interface, use
1359 gfc_compare_interfaces to check and provide a more useful error
1360 message.
1362 2024-04-02 Paul Thomas <pault@gcc.gnu.org>
1364 PR fortran/112407
1365 * resolve.cc (resolve_procedure_expression): Change the test for
1366 for recursion in the case of hidden procedures from modules.
1367 (resolve_typebound_static): Add warning for possible recursive
1368 calls to typebound procedures.
1369 * trans-expr.cc (gfc_trans_class_init_assign): Do not apply
1370 default initializer to class dummy where component initializers
1371 are all null.
1373 2024-03-29 Harald Anlauf <anlauf@gmx.de>
1375 PR fortran/50410
1376 * trans-expr.cc (gfc_conv_structure): Check for NULL pointer.
1378 2024-03-29 Paul Thomas <pault@gcc.gnu.org>
1380 PR fortran/36337
1381 PR fortran/110987
1382 PR fortran/113885
1383 * trans-expr.cc (gfc_trans_assignment_1): Place finalization
1384 block before rhs post block for elemental rhs.
1385 * trans.cc (gfc_finalize_tree_expr): Check directly if a type
1386 has no components, rather than the zero components attribute.
1387 Treat elemental zero component expressions in the same way as
1388 scalars.
1390 2024-03-28 Harald Anlauf <anlauf@gmx.de>
1392 PR fortran/114474
1393 * primary.cc (gfc_variable_attr): Catch variables used in structure
1394 constructors within DATA statements that are still tagged with a
1395 temporary type BT_PROCEDURE from match_actual_arg and which have the
1396 target attribute, and fix their typespec.
1398 2024-03-28 Mikael Morin <mikael@gcc.gnu.org>
1399 Harald Anlauf <anlauf@gmx.de>
1401 PR fortran/111781
1402 PR fortran/114475
1403 * expr.cc (check_restricted): In submodules, allow variables host-
1404 associated from the parent module.
1406 2024-03-22 Harald Anlauf <anlauf@gmx.de>
1408 PR fortran/55978
1409 * interface.cc (gfc_compare_actual_formal): Skip size check for
1410 NULL() actual without MOLD argument.
1412 2024-03-22 Mikael Morin <mikael@gcc.gnu.org>
1414 PR fortran/107426
1415 * gfortran.h (gfc_save_module_list, gfc_restore_old_module_list):
1416 New declarations.
1417 * module.cc (old_module_list_tail): New global variable.
1418 (gfc_save_module_list, gfc_restore_old_module_list): New functions.
1419 (gfc_use_modules): Set module_list and old_module_list_tail.
1420 * parse.cc (next_statement): Save module_list before doing any work.
1421 (reject_statement): Restore module_list to its saved value.
1423 2024-03-22 Mikael Morin <mikael@gcc.gnu.org>
1425 PR fortran/111781
1426 * symbol.cc (gfc_get_procedure_ns, gfc_get_spec_ns): New functions.
1427 * gfortran.h (gfc_get_procedure_ns, gfc_get_spec ns): Declare them.
1428 (gfc_is_formal_arg): Remove.
1429 * expr.cc (check_restricted): Remove special case allowing local
1430 variable in dummy argument bound expressions. Use gfc_get_spec_ns
1431 to get the right namespace.
1432 * resolve.cc (gfc_is_formal_arg, formal_arg_flag): Remove.
1433 (gfc_resolve_formal_arglist): Set gfc_current_ns. Quit loop and
1434 restore gfc_current_ns instead of early returning.
1435 (resolve_symbol): Factor common array spec resolution code to...
1436 (resolve_symbol_array_spec): ... this new function. Additionnally
1437 set and restore gfc_current_ns.
1439 2024-03-21 Harald Anlauf <anlauf@gmx.de>
1441 PR fortran/30802
1442 * trans-array.cc (abridged_ref_name): New helper function.
1443 (trans_array_bound_check): Use it.
1444 (array_bound_check_elemental): Likewise.
1445 (gfc_conv_array_ref): Likewise.
1447 2024-03-20 Jakub Jelinek <jakub@redhat.com>
1449 * trans-array.cc (structure_alloc_comps): Use integer_zero_node
1450 instead of build_zero_cst (integer_type_node) or
1451 build_int_cst (integer_type_node, 0) and integer_one_node instead of
1452 build_int_cst (integer_type_node, 1).
1453 * trans-expr.cc (conv_scalar_char_value): Likewise.
1454 * trans-stmt.cc (gfc_trans_form_team, gfc_trans_change_team,
1455 gfc_trans_sync_team, gfc_trans_sync): Likewise.
1456 * trans-decl.cc (create_main_function): Likewise.
1457 * trans-intrinsic.cc (trans_this_image, conv_intrinsic_move_alloc):
1458 Likewise.
1459 * trans.cc (gfc_allocate_using_caf_lib, gfc_deallocate_with_status):
1460 Likewise.
1462 2024-03-18 Harald Anlauf <anlauf@gmx.de>
1464 PR fortran/103715
1465 * frontend-passes.cc (check_externals_expr): Prevent invalid read
1466 in case of mismatch of external subroutine with function.
1468 2024-03-17 Harald Anlauf <anlauf@gmx.de>
1470 PR fortran/101135
1471 * trans-array.cc (gfc_get_dataptr_offset): Check for optional
1472 arguments being present before dereferencing data pointer.
1474 2024-03-15 Paul Thomas <pault@gcc.gnu.org>
1476 PR fortran/87477
1477 PR fortran/89645
1478 PR fortran/99065
1479 PR fortran/114141
1480 PR fortran/114280
1481 * class.cc (gfc_change_class): New function needed for
1482 associate names, when rank changes or a derived type is
1483 produced by resolution
1484 * dump-parse-tree.cc (show_code_node): Make output for SELECT
1485 TYPE more comprehensible.
1486 * expr.cc (find_inquiry_ref): Do not simplify expressions of
1487 an inferred type.
1488 * gfortran.h : Add 'gfc_association_list' to structure
1489 'gfc_association_list'. Add prototypes for
1490 'gfc_find_derived_types', 'gfc_fixup_inferred_type_refs' and
1491 'gfc_change_class'. Add macro IS_INFERRED_TYPE.
1492 * match.cc (copy_ts_from_selector_to_associate): Add bolean arg
1493 'select_type' with default false. If this is a select type name
1494 and the selector is a inferred type, build the class type and
1495 apply it to the associate name.
1496 (build_associate_name): Pass true to 'select_type' in call to
1497 previous.
1498 * parse.cc (parse_associate): If the selector is inferred type
1499 the associate name is too. Make sure that function selector
1500 class and rank, if known, are passed to the associate name. If
1501 a function result exists, pass its typespec to the associate
1502 name.
1503 * primary.cc (resolvable_fcns): New function to check that all
1504 the function references are resolvable.
1505 (gfc_match_varspec): If a scalar derived type select type
1506 temporary has an array reference, match the array reference,
1507 treating this in the same way as an equivalence member. Do not
1508 set 'inquiry' if applied to an unknown type the inquiry name
1509 is ambiguous with the component of an accessible derived type.
1510 Check that resolution of the target expression is OK by testing
1511 if the symbol is declared or is an operator expression, then
1512 using 'resolvable_fcns' recursively. If all is well, resolve
1513 the expression. If this is an inferred type with a component
1514 reference, call 'gfc_find_derived_types' to find a suitable
1515 derived type. If there is an inquiry ref and the symbol either
1516 is of unknown type or is inferred to be a derived type, set the
1517 primary and symbol TKR appropriately.
1518 * resolve.cc (resolve_variable): Call new function below.
1519 (gfc_fixup_inferred_type_refs): New function to ensure that the
1520 expression references for a inferred type are consistent with
1521 the now fixed up selector.
1522 (resolve_assoc_var): Ensure that derived type or class function
1523 selectors transmit the correct arrayspec to the associate name.
1524 (resolve_select_type): If the selector is an associate name of
1525 inferred type and has no component references, the associate
1526 name should have its typespec. Simplify the conversion of a
1527 class array to class scalar by calling 'gfc_change_class'.
1528 Make sure that a class, inferred type selector with an array
1529 ref transfers the typespec from the symbol to the expression.
1530 * symbol.cc (gfc_set_default_type): If an associate name with
1531 unknown type has a selector expression, try resolving the expr.
1532 (find_derived_types, gfc_find_derived_types): New functions
1533 that search for a derived type with a given name.
1534 * trans-expr.cc (gfc_conv_variable): Some inferred type exprs
1535 escape resolution so call 'gfc_fixup_inferred_type_refs'.
1536 * trans-stmt.cc (trans_associate_var): Tidy up expression for
1537 'class_target'. Finalize and free class function results.
1538 Correctly handle selectors that are class functions and class
1539 array references, passed as derived types.
1541 2024-03-14 Thomas Schwinge <thomas@codesourcery.com>
1543 * dump-parse-tree.cc (show_omp_clauses): Handle 'self_expr'.
1544 * openmp.cc (gfc_free_omp_clauses): Likewise.
1545 * trans-openmp.cc (gfc_split_omp_clauses): Don't handle 'self_expr'.
1547 2024-03-14 Chung-Lin Tang <cltang@baylibre.com>
1549 * dump-parse-tree.cc (show_omp_namelist): Print "readonly," for
1550 OMP_LIST_MAP and OMP_LIST_CACHE if n->u.map.readonly is set.
1551 Adjust 'n->u.map_op' to 'n->u.map.op'.
1552 * gfortran.h (typedef struct gfc_omp_namelist): Adjust map_op as
1553 'ENUM_BITFIELD (gfc_omp_map_op) op:8', add 'bool readonly' field,
1554 change to named struct field 'map'.
1555 * openmp.cc (gfc_match_omp_map_clause): Adjust 'n->u.map_op' to
1556 'n->u.map.op'.
1557 (gfc_match_omp_clause_reduction): Likewise.
1558 (gfc_match_omp_clauses): Add readonly modifier parsing for OpenACC
1559 copyin clause, set 'n->u.map.op' and 'n->u.map.readonly' for parsed
1560 clause. Adjust 'n->u.map_op' to 'n->u.map.op'.
1561 (gfc_match_oacc_declare): Adjust 'n->u.map_op' to 'n->u.map.op'.
1562 (gfc_match_oacc_cache): Add readonly modifier parsing for OpenACC
1563 cache directive.
1564 (resolve_omp_clauses): Adjust 'n->u.map_op' to 'n->u.map.op'.
1565 * trans-decl.cc (add_clause): Adjust 'n->u.map_op' to 'n->u.map.op'.
1566 (finish_oacc_declare): Likewise.
1567 * trans-openmp.cc (gfc_trans_omp_clauses): Set OMP_CLAUSE_MAP_READONLY,
1568 OMP_CLAUSE__CACHE__READONLY to 1 when readonly is set. Adjust
1569 'n->u.map_op' to 'n->u.map.op'.
1570 (gfc_add_clause_implicitly): Adjust 'n->u.map_op' to 'n->u.map.op'.
1572 2024-03-13 Harald Anlauf <anlauf@gmx.de>
1574 PR fortran/114001
1575 * expr.cc (gfc_is_simply_contiguous): Adjust logic so that CLASS
1576 symbols are also handled.
1578 2024-03-13 Tobias Burnus <tburnus@baylibre.com>
1580 PR fortran/114283
1581 * trans-openmp.cc (gfc_omp_predetermined_mapping): Map dummy
1582 procedures as firstprivate.
1584 2024-03-12 Harald Anlauf <anlauf@gmx.de>
1586 PR fortran/110826
1587 * array.cc (gfc_array_dimen_size): When walking the ref chain of an
1588 array and the ultimate component is a procedure pointer, do not try
1589 to figure out its dimension even if it is a array-valued function.
1591 2024-03-06 Harald Anlauf <anlauf@gmx.de>
1593 PR fortran/103707
1594 PR fortran/106987
1595 * arith.cc (is_hard_arith_error): New helper function to determine
1596 whether an arithmetic error is "hard" or not.
1597 (check_result): Use it.
1598 (gfc_arith_divide): Set "Division by zero" only for regular
1599 numerators of real and complex divisions.
1600 (reduce_unary): Use is_hard_arith_error to determine whether a hard
1601 or (recoverable) soft error was encountered. Terminate immediately
1602 on hard error, otherwise remember code of first soft error.
1603 (reduce_binary_ac): Likewise.
1604 (reduce_binary_ca): Likewise.
1605 (reduce_binary_aa): Likewise.
1607 2024-03-01 Harald Anlauf <anlauf@gmx.de>
1609 PR fortran/104819
1610 * check.cc (gfc_check_null): Handle nested NULL()s.
1611 (is_c_interoperable): Check for MOLD argument of NULL() as part of
1612 the interoperability check.
1613 * interface.cc (gfc_compare_actual_formal): Extend checks for NULL()
1614 actual arguments for presence of MOLD argument when required by
1615 Interp J3/22-146.
1617 2024-02-29 Alexander Westbrooks <alexanderw@gcc.gnu.org>
1619 PR fortran/82943
1620 PR fortran/86148
1621 PR fortran/86268
1622 * decl.cc (gfc_get_pdt_instance): Set the PDT instance field
1623 'f2k_derived', if not set already, to point to the given
1624 PDT template 'f2k_derived' namespace in order to give the
1625 PDT instance referential access to the typebound procedures
1626 of the template.
1627 * gfortran.h (gfc_pdt_is_instance_of): Add prototype.
1628 * resolve.cc (resolve_typebound_procedure): If the derived type
1629 does not have the attribute 'pdt_template' set, compare the
1630 dummy argument to the 'resolve_bindings_derived' type like usual.
1631 If the derived type is a 'pdt_template', then check if the
1632 dummy argument is an instance of the PDT template. If the derived
1633 type is a PDT template, and the dummy argument is an instance of
1634 that template, but the dummy argument 'param_list' is not
1635 SPEC_ASSUMED, check if there are any LEN parameters in the
1636 dummy argument. If there are no LEN parameters, then this implies
1637 that there are only KIND parameters in the dummy argument.
1638 If there are LEN parameters, this would be an error, for all
1639 LEN parameters for the dummy argument MUST be assumed for
1640 typebound procedures of PDTs.
1641 (resolve_pdt): Add a check for ALLOCATABLE and POINTER attributes for
1642 SPEC_DEFERRED parameters of PDT class symbols. ALLOCATABLE and
1643 POINTER attributes for a PDT class symbol are stored in the
1644 'class_pointer' and 'allocatable' attributes of the '_data'
1645 component respectively.
1646 * symbol.cc (gfc_pdt_is_instance_of): New function.
1648 2024-02-26 Harald Anlauf <anlauf@gmx.de>
1650 PR fortran/114012
1651 * trans-expr.cc (gfc_conv_procedure_call): Evaluate non-trivial
1652 arguments just once before assigning to an unlimited polymorphic
1653 dummy variable.
1655 2024-02-23 Steve Kargl <kargl@gcc.gnu.org>
1656 Harald Anlauf <anlauf@gmx.de>
1658 PR fortran/114024
1659 * trans-stmt.cc (gfc_trans_allocate): When a source expression has
1660 substring references, part-refs, or %re/%im inquiries, wrap the
1661 entity in parentheses to force evaluation of the expression.
1663 2024-02-23 Tobias Burnus <tburnus@baylibre.com>
1665 * openmp.cc (gfc_match_omp_depobj): Use OPT_Wopenmp
1666 as warning category in gfc_warning.
1668 2024-02-20 Peter Hill <peter.hill@york.ac.uk>
1670 PR fortran/105658
1671 * trans-expr.cc (gfc_conv_intrinsic_to_class): When passing an
1672 array component reference of intrinsic type to a procedure
1673 with an unlimited polymorphic dummy argument, a temporary
1674 should be created.
1676 2024-02-17 Jakub Jelinek <jakub@redhat.com>
1678 PR fortran/113503
1679 * trans-expr.cc (alloc_scalar_allocatable_subcomponent): Don't
1680 overwrite expr2->ts.u.cl->backend_decl, instead set size to
1681 expr2->ts.u.cl->backend_decl first and use size instead of
1682 expr2->ts.u.cl->backend_decl.
1683 (gfc_trans_subcomponent_assign): Emit se.pre into block
1684 before calling alloc_scalar_allocatable_subcomponent instead of
1685 after it.
1687 2024-02-17 Harald Anlauf <anlauf@gmx.de>
1689 PR fortran/113911
1690 * trans-array.cc (gfc_trans_deferred_array): Do not clobber
1691 deferred length for a character variable passed as dummy argument.
1693 2024-02-15 Kwok Cheung Yeung <kcyeung@baylibre.com>
1695 * dump-parse-tree.cc (show_attr): Handle omp_declare_target_indirect
1696 attribute.
1697 * f95-lang.cc (gfc_gnu_attributes): Add entry for 'omp declare
1698 target indirect'.
1699 * gfortran.h (symbol_attribute): Add omp_declare_target_indirect
1700 field.
1701 (struct gfc_omp_clauses): Add indirect field.
1702 * openmp.cc (omp_mask2): Add OMP_CLAUSE_INDIRECT.
1703 (gfc_match_omp_clauses): Match indirect clause.
1704 (OMP_DECLARE_TARGET_CLAUSES): Add OMP_CLAUSE_INDIRECT.
1705 (gfc_match_omp_declare_target): Check omp_device_type and apply
1706 omp_declare_target_indirect attribute to symbol if indirect clause
1707 active. Show warning if there are only device_type and/or indirect
1708 clauses on the directive.
1709 * trans-decl.cc (add_attributes_to_decl): Add 'omp declare target
1710 indirect' attribute if symbol has indirect attribute set.
1712 2024-02-14 Steve Kargl <kargl@gcc.gnu.org>
1714 PR fortran/105847
1715 * trans-io.cc (transfer_namelist_element): When building the
1716 namelist object name, if the use rename attribute is set, use
1717 the local name specified in the use statement.
1719 2024-02-14 Jakub Jelinek <jakub@redhat.com>
1721 * error.cc (error_print): For u printing of ptrdiff_t,
1722 print ptrdiff_t argument converted to unsigned long long and
1723 masked with 2ULL * PTRDIFF_MAX + 1.
1725 2024-02-13 Tobias Burnus <tburnus@baylibre.com>
1727 PR middle-end/113904
1728 * trans-openmp.cc (gfc_trans_omp_declare_variant): Handle splitting of
1729 OMP_TRAIT_PROPERTY_EXPR into OMP_TRAIT_PROPERTY_{DEV_NUM,BOOL}_EXPR.
1730 * openmp.cc (gfc_match_omp_context_selector): Likewise; rejects
1731 non-const device_num/condition; improve diagnostic.
1733 2024-02-13 Harald Anlauf <anlauf@gmx.de>
1735 PR fortran/113866
1736 * trans-expr.cc (gfc_conv_procedure_call): When passing an optional
1737 dummy argument to an optional dummy argument of a bind(c) procedure
1738 and the dummy argument is passed via a CFI descriptor, no special
1739 presence check and passing of a default NULL pointer is needed.
1741 2024-02-13 Steve Kargl <kargl@gcc.gnu.org>
1743 PR fortran/113883
1744 * trans-array.cc (gfc_trans_deferred_array): Set length to zero,
1745 avoiding extraneous diagnostics.
1747 2024-02-10 Jakub Jelinek <jakub@redhat.com>
1749 * error.cc (error_print): Handle z and t modifiers on d, i and u.
1750 * check.cc (gfc_check_transfer): Use %zd instead of %ld and casts to
1751 long.
1752 * primary.cc (gfc_convert_to_structure_constructor): Use %td instead
1753 of %ld and casts to long.
1755 2024-02-10 Jakub Jelinek <jakub@redhat.com>
1757 * trans-common.cc (build_common_decl): Use %wu instead of %lu and
1758 casts to unsigned long.
1759 * resolve.cc (resolve_ordinary_assign): Use %wd instead of %ld and
1760 casts to long.
1761 * array.cc (gfc_resolve_character_array_constructor): Likewise.
1762 * data.cc (create_character_initializer): Likewise.
1764 2024-02-09 Harald Anlauf <anlauf@gmx.de>
1766 PR fortran/113799
1767 * arith.cc (reduce_unary): Remember any overflow encountered during
1768 reduction of unary arithmetic operations on array constructors and
1769 continue, and return error status, but terminate on serious errors.
1771 2024-01-28 Harald Anlauf <anlauf@gmx.de>
1773 PR fortran/113377
1774 * trans-expr.cc (conv_dummy_value): Treat NULL actual argument to
1775 optional dummy with the VALUE attribute as not present.
1776 (gfc_conv_procedure_call): Likewise.
1778 2024-01-27 Harald Anlauf <anlauf@gmx.de>
1780 PR fortran/104908
1781 * trans-array.cc (gfc_conv_array_ref): Restrict use of transformed
1782 descriptor (sym->backend_decl) to the unlimited polymorphic case.
1784 2024-01-24 Harald Anlauf <anlauf@gmx.de>
1786 PR fortran/113377
1787 * trans-expr.cc (conv_dummy_value): New.
1788 (gfc_conv_procedure_call): Factor code for handling dummy arguments
1789 with the VALUE attribute in the scalar case into conv_dummy_value().
1790 Reuse and adjust for calling elemental procedures.
1792 2024-01-21 Harald Anlauf <anlauf@gmx.de>
1794 PR fortran/113377
1795 * trans-expr.cc (gfc_conv_procedure_call): Fix handling of optional
1796 scalar arguments of intrinsic type with the VALUE attribute.
1798 2024-01-20 Mikael Morin <mikael@gcc.gnu.org>
1800 PR fortran/48776
1801 PR fortran/111291
1802 * parse.cc: Restore current interface to its previous value on error.
1804 2024-01-19 Harald Anlauf <anlauf@gmx.de>
1806 PR fortran/113471
1807 * trans-array.cc (array_bound_check_elemental): Array bounds check
1808 shall apply here to elemental dimensions of an array section only.
1810 2024-01-13 Harald Anlauf <anlauf@gmx.de>
1812 PR fortran/67277
1813 * trans-intrinsic.cc (gfc_conv_intrinsic_ishftc): Handle optional
1814 dummy argument for SIZE passed to ISHFTC. Set default value to
1815 BIT_SIZE(I) when missing.
1817 2024-01-13 Harald Anlauf <anlauf@gmx.de>
1819 PR fortran/113305
1820 * gfortran.h (gfc_loop_annot): New.
1821 (gfc_iterator, gfc_forall_iterator): Use for annotation control.
1822 * array.cc (gfc_copy_iterator): Adjust.
1823 * gfortran.texi: Document annotations IVDEP, UNROLL n, VECTOR,
1824 NOVECTOR as applied to DO CONCURRENT.
1825 * parse.cc (parse_do_block): Parse annotations IVDEP, UNROLL n,
1826 VECTOR, NOVECTOR as applied to DO CONCURRENT. Apply UNROLL only to
1827 first loop control variable.
1828 * trans-stmt.cc (iter_info): Use gfc_loop_annot.
1829 (gfc_trans_simple_do): Adjust.
1830 (gfc_trans_forall_loop): Annotate loops with IVDEP, UNROLL n,
1831 VECTOR, NOVECTOR as needed for DO CONCURRENT.
1832 (gfc_trans_forall_1): Handle loop annotations.
1834 2024-01-08 Harald Anlauf <anlauf@gmx.de>
1836 PR fortran/113245
1837 * trans-intrinsic.cc (gfc_conv_intrinsic_size): Use
1838 gfc_conv_expr_present() for proper check of optional DIM argument.
1840 2024-01-06 Harald Anlauf <anlauf@gmx.de>
1841 José Rui Faustino de Sousa <jrfsousa@gmail.com>
1843 PR fortran/96724
1844 * iresolve.cc (gfc_resolve_repeat): Force conversion to
1845 gfc_charlen_int_kind before call to gfc_multiply.
1847 2024-01-04 David Malcolm <dmalcolm@redhat.com>
1849 * lang.opt.urls: New file, autogenerated by
1850 regenerate-opt-urls.py.
1852 2024-01-03 Harald Anlauf <anlauf@gmx.de>
1854 * trans-types.cc (gfc_get_nodesc_array_type): Clear used gmp
1855 variables.
1857 2024-01-03 Jakub Jelinek <jakub@redhat.com>
1859 * gfortranspec.cc (lang_specific_driver): Update copyright notice
1860 dates.
1861 * gfc-internals.texi: Bump @copying's copyright year.
1862 * gfortran.texi: Ditto.
1863 * intrinsic.texi: Ditto.
1864 * invoke.texi: Ditto.
1866 \f
1867 Copyright (C) 2024 Free Software Foundation, Inc.
1869 Copying and distribution of this file, with or without modification,
1870 are permitted in any medium without royalty provided the copyright
1871 notice and this notice are preserved.