https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94970
--- Comment #2 from Iain Buclaw <ibuclaw at gdcproject dot org> ---
Because RegexMatch needs destruction, a temporary is created that requires
scope destruction. The temporary is wrapped in a TARGET_EXPR, and dtor call
set in TARGET_EXPR_CLEANUP.
TARGET_EXPR <D.4046, tmp = m () [return slot optimization], __dtor(&tmp)>
A clean-up point is then created around the initialization of 'a'.
<<cleanup_point a = ... >>
The problem is that for the array concat expression on the RHS, a BIND_EXPR is
set-up for a temporary created for the character literal (its address needs to
be taken).
The routine gimplify_cleanup_point_expr does not look any lower than the first
level of sequences in the body, so it completely misses the
GIMPLE_WITH_CLEANUP_EXPRs in the following:
{
const char D.4043;
try
{
D.4043 = 32;
D.4110.length = 1;
D.4110.ptr = &D.4043;
__tmpfordtor57 = m (); [return slot optimization]
<<< Unknown GIMPLE statement: gimple_with_cleanup_expr >>>
<<< Unknown GIMPLE statement: gimple_with_cleanup_expr >>>
D.4111 = index (&__tmpfordtor57);
D.4112 = _d_arraycatT (&_D12TypeInfo_Aya6__initZ, D.4111, D.4110);
retval.0 = VIEW_CONVERT_EXPR<struct >(D.4112);
}
finally
{
D.4043 = {CLOBBER};
}
}
retval.1 = retval.0;
a = retval.1;
There is a note above gimplify_cleanup_point_expr:
FIXME should we complexify the prequeue handling instead? Or use flags
for all the cleanups and let the optimizer tighten them up? The current
code seems pretty fragile; it will break on a cleanup within any
non-conditional nesting. But any such nesting would be broken, anyway;
we can't write a TRY_FINALLY_EXPR that starts inside a nesting construct
and continues out of it. We can do that at the RTL level, though, so
having an optimizer to tighten up try/finally regions would be a Good
Thing.
