Messages in this thread | | | Date | Fri, 20 Feb 2009 17:04:09 +0100 | From | Ingo Molnar <> | Subject | Re: [PATCH] mm: fix lazy vmap purging (use-after-free error) |
| |
* Paul E. McKenney <paulmck@linux.vnet.ibm.com> wrote:
> On Fri, Feb 20, 2009 at 03:51:28PM +0100, Vegard Nossum wrote: > > 2009/2/20 Ingo Molnar <mingo@elte.hu>: > > > > > > * Ingo Molnar <mingo@elte.hu> wrote: > > > > > >> ah, indeed: > > >> > > >> list_del_rcu(&va->list); > > >> > > >> i suspect it could be hit big time in a workload that opens > > >> more than 512 files, as expand_files() uses a > > >> vmalloc()+vfree() pair in that case. > > > > > > hm, perhaps it's not a problem after all. The freeing is done > > > via rcu, and list_del_rcu() leaves the forward pointer intact. > > > > Well, it's not the particular line that you posted, in any case. > > That's &va->list, but the traversed list is &va->purge_list. > > > > I thought it would be the line: > > > > call_rcu(&va->rcu_head, rcu_free_va); > > > > (which does kfree() in the callback) that was the problem. > > > > > So how did it happen that the entry got kfree()d before the loop > > > was done? We are in a spinlocked section so the CPU should not > > > have entered rcu processing. > > > > I added some printks to __free_vmap_area() and rcu_free_va(), and it > > shows that the kfree() is being called immediately (inside the list > > traversal). So the call_rcu() is happening immediately (or almost > > immediately). > > > > If I've understood correctly, the RCU processing can happen inside a > > spinlock, as long as interrupts are enabled. (Won't the timer IRQ > > trigger softirq processing, which triggers RCU callback processing, > > for example?) > > > > And interrupts are enabled when this happens: EFLAGS: 00000292 > > > > Please correct me if I am wrong! > > If you are using preemptable RCU, and if the read side > accesses are not protected by rcu_read_lock(), this can > happen. At least for values of "immediately" in the > millisecond range. > > If you were using classic or hierarchical RCU, the fact that > the call_rcu() is within a spinlock (as opposed to mutex) > critical section should prevent the grace period from ending. > > So, what flavor of RCU were you using?
well, even in preemptible RCU the grace period should be extended as long as we are non-preempt (which we are here), correct?
Preemptible RCU does make an rcu_read_lock() critical section preemptible, so if this were protected by rcu_read_lock() it would be a bug. But it does not make spin_lock() section preemptible, and this is a spin_lock(&vmap_area_lock) section:
spin_lock(&vmap_area_lock); - list_for_each_entry(va, &valist, purge_list) + list_for_each_entry_safe(va, n_va, &valist, purge_list) __free_vmap_area(va); spin_unlock(&vmap_area_lock);
Ingo
| |