On Mon, Feb 12, 2018 at 07:39:02PM +0200, Nikolay Borisov wrote: > > > On 12.02.2018 16:17, Liu Bo wrote: > > On Tue, Feb 06, 2018 at 11:11:55AM +0200, Nikolay Borisov wrote: [...] > >> > >> Use list_for_each_entry_safe to make it more apparent you are going to > >> be removing from the list. The code as-is works since you are doing a > >> break after deleting element from the list but this is somewhat subtle. > > > > To be honest, I don't see much difference. > > > > I think the _safe version is to protect us from some race when others > > remove objects from list, and write lock is held so we're safe. > > No, the _safe version uses the second argument (n) as the list iterator. > The non-safe version just uses 'pos', and in case you remove 'pos' from > the list AND continue iterating you will deref an invalid pointer. So > _safe is actually really necessary for correctness when you intend to > remove an entry from a list you are iterating, irrespective of any locks > you might have. You're right, I know it takes an extra pointer to store the next entry, but miss the point when deleting by itself. Thank you for the clarification. thanks, -liubo -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
