https://gcc.gnu.org/bugzilla/show_bug.cgi?id=85047
Richard Biener <rguenth at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |ASSIGNED
Last reconfirmed| |2018-03-23
CC| |marxin at gcc dot gnu.org
Component|ada |lto
Ever confirmed|0 |1
--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
I think it's a fundamental issue with the streamer and how tree_is_indexable
is a "flat" predicate but we test it from the toplevel like in lto_output_tree.
The fix is probably to handle refs from within the DFS walk which also means
to not simply stop at "indexable" refs (but still do a lookup whether they
already exist in the index!) and handle them when they appear as leafs.
A C testcase that should run into similar issues as the Ada one is
void foo(int n)
{
struct X { int a[n]; } a;
struct Y { struct X *p; } b;
b.p = &a;
}
which isn't ICEing but at least is streaming two different copies of X.
