Re: list corruption on removal of snd_seq_dummy

[Takashi Iwai]

At Tue, 27 Jun 2006 08:05:01 -0400,
Chuck Ebbert wrote:
> 
> In-Reply-To: <s5hejxb7xrb.wl%tiwai@xxxxxxx>
> 
> On Tue, 27 Jun 2006 11:28:56 +0200, Takashi Iwai wrote:
> 
> > >  > > The code in question is doing..
> > >  > > 
> > >  > >         __list_add(&deleted_list,
> > >  > >                client->ports_list_head.prev,
> > >  > >                client->ports_list_head.next);
> > >  > > 
> > >  > > which looks fishy, as those two elements aren't going to be consecutive,
> > >  > > as __list_add expects.
> > >  > 
> > >  > I think the code behaves correctly but probably misusing __list_add().
> > >  > It movies the whole entries from an existing list_head A
> > >  > (clients->ports_list_head) to a new list_head B (deleted_list).
> > >  > The above is exapnded:
> > >  > 
> > >  >  A->next->prev = B;
> > >  >  B->next = A->next;
> > >  >  B->prev = A->prev;
> > >  >  A->prev->next = B;
> > >  > 
> > >  > Any better way to achieve it using standard macros?
> > > 
> > > Why can't you just list_move() the elements ?
> > 
> > No, list_move() can't move the whole elements without loop.
> > 
> > A solution is
> > 
> >       list_add(B, A);
> >       list_del_init(A);
> > 
> > (although this introduces a bit more code :)
> 
> Shouldn't it be like this?
> 
>         ports_list_first = client->ports_list_head.next;
>         list_del_init(client->ports_list_head);
>         list_splice(ports_list_first, &deleted_list);

This requires INIT_LIST_HEAD(&deleted_list) first, so obviously
a longer code :)


Takashi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/