On Thu, Oct 9, 2014 at 9:38 AM, Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote: > On Fri, 2014-09-12 at 08:58 +0200, Henning Rogge wrote: > [...] > > In general, this looks fine. However, > >> @@ -691,6 +719,9 @@ int mpp_path_add(struct ieee80211_sub_if_data *sdata, >> >> spin_unlock(&tbl->hashwlock[hash_idx]); >> read_unlock_bh(&pathtbl_resize_lock); >> + >> + mpp_paths_generation++; > > This doesn't seem to be concurrency safe? > > Neither is the existing mesh_paths_generation++, I think, so maybe it's > not something we really need to worry about? Uhh... I must admit I am not completely sure... this is just my third kernel patch, so I am still learning about all the arcane locking/RCU systems of the kernel. I think it should be reasonable safe. The variable is only increased (never decreased) and is used as a "information version number" for the dumps via netlink to the userspace. I think the current userspace tools ignore the variable completely. Henning Rogge -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
