[Bug c++/66786] [5/6 Regression] ICE: Segmentation fault

[mpolacek at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66786

Marek Polacek <mpolacek at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |NEW
           Assignee|mpolacek at gcc dot gnu.org        |unassigned at gcc dot gnu.org

--- Comment #3 from Marek Polacek <mpolacek at gcc dot gnu.org> ---
Unfortunately, I'll have to give up on this one.

What happens here is that we ICE because we have a lambda expr, whose
LAMBDA_EXPR_CAPTURE_LIST has null TREE_PURPOSE.

10272           /* We need to instantiate the capture list from the template
10273              after we've instantiated the closure members, but before we
10274              consider adding the conversion op.  Also keep any captures
10275              that may have been added during instantiation of the op(). 
*/
10276           tree tmpl_expr = CLASSTYPE_LAMBDA_EXPR (pattern);
10277           tree tmpl_cap
10278             = tsubst_copy_and_build (LAMBDA_EXPR_CAPTURE_LIST
(tmpl_expr),
10279                                      args, tf_warning_or_error,
NULL_TREE,
10280                                      false, false);
10281 
10282           LAMBDA_EXPR_CAPTURE_LIST (expr)
10283             = chainon (tmpl_cap, nreverse (LAMBDA_EXPR_CAPTURE_LIST
(expr)));

but tsubst_copy_and_build returns tmpl_cap without TREE_PURPOSE, because

16621             /* Expand the argument expressions.  */
16622             if (TREE_PURPOSE (t))
16623               purposevec = tsubst_pack_expansion (TREE_PURPOSE (t), args,
16624                                                  complain, in_decl);

returns null.  That is probably because we end up calling
13898           /* Otherwise return the full NONTYPE_ARGUMENT_PACK that
13899              tsubst_decl put in the hash table.  */
13900           return retrieve_specialization (t, args, 0);

which is where the null originates.  Alas, this pt.c maze is something I can't
figure out, so unassigning :(.