PS Quite right about the allocation in PR93813 - consider it to be included.
Cheers and thanks
Paul
On Fri, 7 Apr 2023 at 22:35, Paul Richard Thomas <
paul.richard.tho...@gmail.com> wrote:
> Hi Harald,
>
> Well done on noticing the memory leak :-) I have a fix for it that I was
> going to post separately. Actually, it is a trivial one liner, which I
> could include with the patch.
> @@ -2554,23 +2559,25 @@ gfc_conv_string_length (gfc_charlen * cl, gfc_expr
> * expr, stmtblock_t * pblock)
> expr_flat = gfc_copy_expr (expr);
> flatten_array_ctors_without_strlen (expr_flat);
> gfc_resolve_expr (expr_flat);
> -
> - gfc_conv_expr (&se, expr_flat);
> - gfc_add_block_to_block (pblock, &se.pre);
> - cl->backend_decl = convert (gfc_charlen_type_node,
> se.string_length);
> -
> + if (expr_flat->rank)
> + gfc_conv_expr_descriptor (&se, expr_flat);
> + else
> + gfc_conv_expr (&se, expr_flat);
> + if (expr_flat->expr_type != EXPR_VARIABLE)
> + gfc_add_block_to_block (pblock, &se.pre);
> + se.expr = convert (gfc_charlen_type_node, se.string_length);
> + gfc_add_block_to_block (pblock, &se.post);
> // <<<right here>>>
> gfc_free_expr (expr_flat);
> - return;
> }
> -
> - /* Convert cl->length. */
> -
> - gcc_assert (cl->length);
> -
> - gfc_conv_expr_type (&se, cl->length, gfc_charlen_type_node);
> - se.expr = fold_build2_loc (input_location, MAX_EXPR,
> gfc_charlen_type_node,
> - se.expr, build_zero_cst (TREE_TYPE
> (se.expr)));
> - gfc_add_block_to_block (pblock, &se.pre);
> + else
> + {
> + /* Convert cl->length. */
> + gfc_conv_expr_type (&se, cl->length, gfc_charlen_type_node);
> + se.expr = fold_build2_loc (input_location, MAX_EXPR,
> + gfc_charlen_type_node, se.expr,
> + build_zero_cst (TREE_TYPE (se.expr)));
> + gfc_add_block_to_block (pblock, &se.pre);
> + }
>
> if (cl->backend_decl && VAR_P (cl->backend_decl))
> gfc_add_modify (pblock, cl->backend_decl, se.expr);
>
> Cheers
>
> Paul
>
>
> On Fri, 7 Apr 2023 at 20:28, Harald Anlauf <anl...@gmx.de> wrote:
>
>> Hi Paul,
>>
>> On 4/7/23 15:53, Paul Richard Thomas via Gcc-patches wrote:
>> > duuuh! Please find them attached.
>>
>> the patch LGTM. Thanks!
>>
>> However, I have comments on the new testcase associate_60.f90:
>> subroutine pr93813 is missing an allocation of x, e.g.:
>>
>> allocate (t :: x)
>>
>> otherwise it would be invalid. Please check and fix.
>>
>> Interestingly, subroutine pr92779 exhibits a small memory leak
>> with memory allocated by the spread intrinsic. I played a little
>> and found that the leak depends on the presence of trim(): omitting
>> trim() removes the leak. But looking at the related pr, it seems
>> that trim() was essential, so omitting it is likely not an option.
>>
>> I think the best way is to proceed and to open a PR on the memory
>> leak rather than leaving pr92779 open. What do you think?
>>
>> Cheers,
>> Harald
>>
>>
>> > Thanks
>> >
>> > Paul
>> >
>> >
>> > On Fri, 7 Apr 2023 at 10:41, Harald Anlauf <anl...@gmx.de> wrote:
>> >
>> >> Hi Paul,
>> >>
>> >> I don't see the new testcases. Is this an issue on my side,
>> >> or did you forget to attach them?
>> >>
>> >> Thanks,
>> >> Harald
>> >>
>> >> On 4/7/23 09:07, Paul Richard Thomas via Gcc-patches wrote:
>> >>> Dear All,
>> >>>
>> >>> Please find attached a slightly updated version of the patch with a
>> >>> consolidated testcase. The three additional testcases are nothing to
>> do
>> >>> with associate and test fixes of character related bugs.
>> >>>
>> >>> OK for mainline?
>> >>>
>> >>> Cheers
>> >>>
>> >>> Paul
>> >>> Fortran: Fix some of the bugs in associate [PR87477]
>> >>>
>> >>> 2023-04-07 Paul Thomas <pa...@gcc.gnu.org>
>> >>>
>> >>> gcc/fortran
>> >>> PR fortran/87477
>> >>> * resolve.cc (resolve_assoc_var): Handle parentheses around the
>> >>> target expression.
>> >>> (resolve_block_construct): Remove unnecessary static decls.
>> >>> * trans-array.cc (gfc_conv_expr_descriptor): Guard string len
>> >>> expression in condition. Improve handling of string length and
>> >>> span, especially for substrings of the descriptor.
>> >>> (duplicate_allocatable): Make element type more explicit with
>> >>> 'eltype'.
>> >>> * trans_decl.cc (gfc_get_symbol_decl): Emit a fatal error with
>> >>> appropriate message instead of ICE if symbol type is unknown.
>> >>> * trans-expr.cc (gfc_get_expr_charlen): Retain last charlen in
>> >>> 'previous' and use if end expression in substring reference is
>> >>> null.
>> >>> (gfc_conv_string_length): Use gfc_conv_expr_descriptor if
>> >>> 'expr_flat' is an array.
>> >>> (gfc_trans_alloc_subarray_assign): If this is a deferred string
>> >>> length component, store the string length in the hidden comp.
>> >>> Update the typespec length accordingly. Generate a new type
>> >>> spec for the call to gfc_duplicate-allocatable in this case.
>> >>> * trans-io.cc (gfc_trans_transfer): Scalarize transfer of
>> >>> deferred character array components.
>> >>>
>> >>>
>> >>> gcc/testsuite/
>> >>> PR fortran/87477
>> >>> * gfortran.dg/finalize_47.f90 : Enable substring test.
>> >>> * gfortran.dg/finalize_51.f90 : Update an error message.
>> >>>
>> >>> PR fortran/85686
>> >>> PR fortran/88247
>> >>> PR fortran/91941
>> >>> PR fortran/92779
>> >>> PR fortran/93339
>> >>> PR fortran/93813
>> >>> PR fortran/100948
>> >>> PR fortran/102106
>> >>> * gfortran.dg/associate_60.f90 : New test
>> >>>
>> >>> PR fortran/98408
>> >>> * gfortran.dg/pr98408.f90 : New test
>> >>>
>> >>> PR fortran/105205
>> >>> * gfortran.dg/pr105205.f90 : New test
>> >>>
>> >>> PR fortran/106918
>> >>> * gfortran.dg/pr106918.f90 : New test
>> >>
>> >>
>> >
>>
>>
>
> --
> "If you can't explain it simply, you don't understand it well enough" -
> Albert Einstein
>
--
"If you can't explain it simply, you don't understand it well enough" -
Albert Einstein
