Hi! For ADDR_EXPR gimple_debug_bind_get_value fold_stmt_1 uses maybe_canonicalize_mem_ref_addr earlier and I think that should resolve the concerns raised in PR52329. But folding ADDR_EXPR operand using maybe_fold_reference and then taking address of that looks like an invalid transformation, it can transform # DEBUG D.4293 => &a[0] into # DEBUG D.4293 => &2.0e+0 etc., all we want to allow are the lhs folding of the operand which maybe_fold_reference no longer does since r12-21-g0bf8cd9d5e8ac.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk? 2022-01-05 Jakub Jelinek <ja...@redhat.com> PR fortran/103691 * gimple-fold.c (fold_stmt_1): Don't call maybe_fold_reference for DEBUG stmts with ADDR_EXPR gimple_debug_bind_get_value, it can do unwanted rhs folding like &a[0] into &2.0 etc. * gfortran.dg/pr103691.f90: New test. --- gcc/gimple-fold.c.jj 2022-01-03 10:40:42.872140874 +0100 +++ gcc/gimple-fold.c 2022-01-04 13:29:17.571321684 +0100 @@ -6283,8 +6283,7 @@ fold_stmt_1 (gimple_stmt_iterator *gsi, if (gimple_debug_bind_p (stmt)) { tree val = gimple_debug_bind_get_value (stmt); - if (val - && REFERENCE_CLASS_P (val)) + if (val && REFERENCE_CLASS_P (val)) { tree tem = maybe_fold_reference (val); if (tem) @@ -6292,18 +6291,6 @@ fold_stmt_1 (gimple_stmt_iterator *gsi, gimple_debug_bind_set_value (stmt, tem); changed = true; } - } - else if (val - && TREE_CODE (val) == ADDR_EXPR) - { - tree ref = TREE_OPERAND (val, 0); - tree tem = maybe_fold_reference (ref); - if (tem) - { - tem = build_fold_addr_expr_with_type (tem, TREE_TYPE (val)); - gimple_debug_bind_set_value (stmt, tem); - changed = true; - } } } break; --- gcc/testsuite/gfortran.dg/pr103691.f90.jj 2022-01-04 13:31:23.564557726 +0100 +++ gcc/testsuite/gfortran.dg/pr103691.f90 2022-01-04 13:31:08.053774883 +0100 @@ -0,0 +1,9 @@ +! PR fortran/103691 +! { dg-do compile } +! { dg-options "-O2 -g" } + +program pr103691 + real, parameter :: a(0) = 2.0 + real, allocatable :: b(:) + allocate (b, mold=a) +end Jakub