[Bug gdb/31281] [gdb] intrusive_list.h:329: internal-error: push_back: Assertion elem_node->prev == INTRUSIVE_LIST_UNLINKED_VALUE failed.

["simark at simark dot ca" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=31281

--- Comment #9 from Simon Marchi <simark at simark dot ca> ---
(In reply to Tom de Vries from comment #8)
> (In reply to Simon Marchi from comment #6)
> > I bisected gcc, the failure appeared with this commit:
> > 
> > c++: non-dependent .* operand folding [PR112427]
> > https://gitlab.com/gnutools/gcc/-/commit/d3f48f68227
> > 
> > I know nothing about compilers, so I don't know what the code changes.  But
> > the included test case uses a pointer-to-member, which is somewhat relevant
> > in our case.
> 
> At the parser level (original tree dump file), I see:
> ...
> ;; Function static intrusive_list_node<T>* intrusive_member_node<T,
> MemberNode>::as_node(T*) [with T = thread_info; intrusive_list_node<T> T::*
> MemberNode = &thread_info::resumed_with_pending_wait_status_node] (null)
> ;; enabled by -tree-original
> 
> 
> return <retval> = NON_LVALUE_EXPR <elem> != 0B ? &NON_LVALUE_EXPR
> <elem>->D.191606 + (sizetype) 520 : (struct intrusive_list_node *) 520;
> ...
> 
> The 520 is the offset of resumed_with_pending_wait_status_node:
> ...
> (gdb) p /x (void *)&thread->resumed_with_pending_wait_status_node - (void
> *)thread
> $46 = 0x208
> (gdb) p 0x208
> $12 = 520
> ...
> 
> The D.191606 bit adds an offset of 8 (this becomes apparent at the expand
> rtl dump file).  That also happens to be the offset of the inherited
> intrusive list node, see:
> ...
> class thread_info : public refcounted_object,
>                     public intrusive_list_node<thread_info>
> ...
> and:
> ...
> (gdb) p (void *) &thread.next - (void *) thread
> $13 = 8
> ...

I'm not sure I understand, but aren't you conflating thread_info::next with
thread_info::resumed_with_pending_wait_status_node::next?

-- 
You are receiving this mail because:
You are on the CC list for the bug.