This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
The Fortran part of your patch is OK - hopefully, all 7 patches with pending Dwarf2out approval get reviewed/approved before stage3. Thanks for spotting and fixing this bug.This patch fixes debuginfo for Fortran subref array pointers (stride has to be multiplied by the content of its span variable instead of element size) and also fixes a couple of small issues (e.g. TREE_STATIC has been always set on the span variable, which is undesirable at least for OpenMP or recursive procedures).
That said, subref array pointers don't appear to work for dummy arguments (f951 ICEs if a dummy array pointer is later pointed to is_subref_array) and I don't see how it can possibly work with array pointers in modules, or in common blocks, or threadprivate.
Is there a plan to do it right with the next major ABI change (i.e.
represent strides in bytes rather than elements and kill this span stuff)?
I think there are comments about intending to go from 7 to 15 maximum rank
and that will need ABI changes to the array descriptors anyway.
2008-08-28 Jakub Jelinek <jakub@redhat.com>
* trans.h (struct lang_type): Add span.
(GFC_TYPE_ARRAY_SPAN): Define.
* trans-decl.c (gfc_get_symbol_decl): For subref array pointers,
copy TREE_STATIC from decl to span instead of setting it
unconditionally, set DECL_ARTIFICIAL, fix type of initializer
and set GFC_TYPE_ARRAY_SPAN on decl's type.
* trans-types.c (gfc_get_array_descr_info): If
GFC_TYPE_ARRAY_SPAN is non-NULL, use it as element size.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |