===================================================================
@@ -851,8 +851,8 @@ gfc_trans_create_temp_array (stmtblock_t * pre, st
of the descriptor fields. */
tmp = fold_build2_loc (input_location,
MINUS_EXPR, gfc_array_index_type,
- gfc_conv_descriptor_ubound_get (desc, gfc_rank_cst[n]),
- gfc_conv_descriptor_lbound_get (desc, gfc_rank_cst[n]));
+ gfc_conv_descriptor_ubound_get (desc, gfc_rank_cst[dim]),
+ gfc_conv_descriptor_lbound_get (desc, gfc_rank_cst[dim]));
loop->to[n] = tmp;
continue;
}
Hello, I'm going to commit the patch below as obvious, once the regression test has finished. Historical note: this has changed several times already in the past: - Revision 162276: n -> dim At that time n and dim were guaranteed to be equal so it didn't matter, but I suppose transpose was already in sight then. - Revision 164112: dim -> n This revision introduced a linear to[] array with transpose already taken into account, so n had to be used for that part. One part was still using loop->to[], so it shouldn't have been changed (it is the cause of the regression). Yet again, it didn't matter at the time. - Revision 164205: This is the regression Starting from here, the dimensions can be transposed, so n and dim can be different. Mikael 2010-12-22 Mikael Morin <mikael.morin@gcc.gnu.org> PR fortran/46978 Revert part of revision 164112 * trans-array.c (gfc_trans_create_temp_array): Set loop n'th upper bound from (possibly transposed) array's dim bounds. 2010-12-22 Mikael Morin <mikael@gcc.gnu.org PR fortran/46978 * gfortran.dg/transpose_intrinsic_func_call_1.f90: New test.