https://gcc.gnu.org/bugzilla/show_bug.cgi?id=83990
--- Comment #7 from Martin Jambor <jamborm at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #4)
>
> ...
>
> The statement on which on the original testcase it warns indeed has no
> location, that was created during IPA optimizations:
> ipa_modify_call_arguments does:
> loc = DECL_P (base) ? DECL_SOURCE_LOCATION (base)
> : EXPR_LOCATION (base);
> This itself is weird in this case, base is a SSA_NAME this_4(D), so has no
> location and loc is UNKNOWN_LOCATION.
Interesting, I see base being an ADDR_EXPR on trunk. But anyway...
> Then we:
> /* If expr is not a valid gimple call argument emit
> a load into a temporary. */
> if (is_gimple_reg_type (TREE_TYPE (expr)))
> {
> gimple *tem = gimple_build_assign (NULL_TREE, expr);
> if (gimple_in_ssa_p (cfun))
> {
> gimple_set_vuse (tem, gimple_vuse (stmt));
> expr = make_ssa_name (TREE_TYPE (expr), tem);
> }
> else
> expr = create_tmp_reg (TREE_TYPE (expr));
> gimple_assign_set_lhs (tem, expr);
> gsi_insert_before (&gsi, tem, GSI_SAME_STMT);
> and don't gimple_set_location for it at all. Wonder if loc shouldn't be set
> to e.g. gimple_location (stmt)
That might indeed make more sense, I will look into it.
But please note that I get exactly the same warning when I add
-fno-ipa-sra on the command line (and I have verified in gdb that
ipa_modify_call_arguments is not invoked at all with the option just
to double check) so ipa_modify_call_arguments cannot be the whole
story.